T.C. Pearce, S.S. Schiffman, H.T. Nagle, J.W. Gardner ...read.pudn.com/downloads120/ebook/510257/Handbook of machine … · 8.2.6 Scanning Light-Pulse Technique 191 8.3 The Tufts

T. C. Pearce, S. S. Schiffman, H.T. Nagle, J.W. Gardner

Handbook of Machine Olfaction

Handbook of Machine Olfaction: Electronic Nose Technology.Edited by T.C. Pearce, S.S. Schiffman, H.T. Nagle, J.W. GardnerCopyright ª 2003 WILEY-VCH Verlag GmbH & Co. KGaA, WeinheimISBN: 3-527-30358-8

Related titles:

J.W. Gardner, V.K. Varadan

Microsensors, MEMS and Smart Devices

ISBN 0-471-86109-X

H.K. Tonshoff, I. Inasaki

Sensors in Manufacturing

ISBN 3-527-29558-5

O. Gassmann, H. Meixner

Sensors in Intelligent Buildings

ISBN 3-527-29557-7

H. Baltes, G.K. Fedder, J.G. Korvink

Sensors Update

ISSN 1432-2404


Handbook of Machine Olfaction

Electronic Nose Technology

Tim C. Pearce, PhD

Department of Engineering

University of Leicester

Leicester LE1 7RH

U.K.

Prof. Susan S. Schiffman

Department of Psychiatry

Duke University Medical School

54212 Woodhall Building

P.O. Box 3259

Durham, NC 27710

USA

Prof. H. Troy Nagle

Department of Electrical

and Computer Engineering

North Carolina State University

432 Daniels Hall

Raleigh, NC 27695-7911

USA

Prof. Julian W. Gardner

Division of Electrical

& Electronic Engineering

The University of Warwick

Coventry CV4 7AL

U.K.

This book was carefully produced. Nevertheless,

authors, editors and publisher do not warrant the

information contained therein to be free of errors.

Readers are advised to keep inmind that statements,

data, illustrations, procedural details or other items

may inadvertently be inaccurate.

Library of Congress Card No. applied for.

British Library Cataloguing-in-Publication

Data:

A catalogue record for this book is available

from the British Library.

Bibliographic information published by Die Deutsche

Bibliothek

Die Deutsche Bibliothek lists this publication in

the Deutsche Nationalbibliografie; detailed

bibliographic data is available in the Internet at

<http://dnb.ddb.de>.

ª 2003 WILEY-VCH Verlag GmbH & Co. KGaA,

Weinheim

All rights reserved (including those of translation

into other languages). No part of this book may be

reproduced in any form – by photoprinting,

microfilm, or any other means – nor transmitted

or translated into a machine language without

written permission from the publishers.

Registered names, trademarks, etc. used in this

book, even when not specifically marked as such,

are not to be considered unprotected by law.

Printed in the Federal Republic of Germany

Printed on acid-free paper

Typesetting Mitterweger & Partner,

Kommunikationsgesellschaft mbH, Plankstadt

Printing and Bookbinding Druckhaus Darmstadt

GmbH, Darmstadt

ISBN 3-527-30358-8

Contents

1 Introduction to Olfaction: Perception, Anatomy, Physiology,

and Molecular Biology 1

1.1 Introduction to Olfaction 1

1.2 Odor Classification Schemes Based on Adjective Descriptors 4

1.3 Odor Classification Based on Chemical Properties 7

1.3.1 History of Structure-activity Studies of Olfaction 8

1.3.2 Odor Structures Associated with Specific Odor Classes Based on QualitativeDescriptors 8

1.3.3 Relationship of Physicochemical Parameters to Classifications of OdorBased on Similarity Measures 11

1.3.3.1 Study 1: Broad Range of Unrelated Odorants 12

1.3.3.2 Study 2: Pyrazines 14

1.3.4 Molecular Parameters and Odor Thresholds 16

1.3.5 Conclusions Regarding Physicochemical Parameters and Odor Quality 16

1.4 Physiology and Anatomy of Olfaction 17

1.4.1 Basic Anatomy 17

1.4.2 Transduction and Adaptation of Olfactory Signals 20

1.5 Molecular Biology Of Olfaction 21

1.6 Taste 23

1.6.1 Taste Classification Schemes Based on Sensory Properties 23

1.6.2 Physiology and Anatomy of Taste 23

1.6.3 Transduction of Taste Signals 25

1.6.4 Molecular Biology of Taste 25

1.7 Final Comment 26

2 Chemical Sensing in Humans and Machines 33

2.1 Human Chemosensory Perception of Airborne Chemicals 33

2.2 Nasal Chemosensory Detection 34

2.2.1 Thresholds for Odor and Nasal Pungency 35

2.2.2 Stimulus-Response (Psychometric) Functions for Odor and NasalPungency 37

2.3 Olfactory and Nasal Chemesthetic Detection of Mixtures of Chemicals 38

2.4 Physicochemical Determinants of Odor and Nasal Pungency 39


2.4.1 The Linear Solvation Model 39

2.4.2 Application of the Solvation Equation to Odor and Nasal PungencyThresholds 40

2.5 Human Chemical Sensing: Olfactometry 42

2.5.1 Static Olfactometry 42

2.5.2 Dynamic Olfactometry 44

2.5.3 Environmental Chambers 45

2.6 Instruments for Chemical Sensing:GasChromatography-Olfactometry 47

2.6.1 Charm Analysis 48

2.6.2 Aroma Extract Dilution Analysis (AEDA) 49

2.6.3 Osme Method 50

3 Odor Handling and Delivery Systems 55

3.1 Introduction 55

3.2 Physics of Evaporation 56

3.3 Sample Flow System 57

3.3.1 Headspace Sampling 57

3.3.2 Diffusion Method 60

3.3.3 Permeation Method 61

3.3.4 Bubbler 61

3.3.5 Method using a Sampling Bag 62

3.4 Static System 64

3.5 Preconcentrator 65

3.5.2 Sensitivity Enhancement 65

3.5.2 Removal of Humidity 66

3.5.3 Selectivity Enhancement by Varying Temperature 66

3.5.3.1 Selectivity Enhancement using a Preconcentrator 66

3.5.3.2 Autonomous System with Plasticity 67

3.5.3.3 Experiment on Plasticity 69

3.6 Measurement of Sensor Directly Exposed to Ambient Vapor 70

3.6.1 Analysis of Transient Sensor Response using an Optical Tracer 70

3.6.2 Homogenous Sensor Array for Visualizing Gas/Odor Flow 72

3.6.3 Response of Sensor Mounted on an Odor-Source Localization System 74

3.7 Summary 74

4 Introduction to Chemosensors 79

4.1 Introduction 79

4.2 Survey and Classification of Chemosensors 79

4.3 Chemoresistors 81

4.3.1 MOS 81

4.3.2 Organic CPs 84

4.4 Chemocapacitors (CAP) 87

4.5 Potentiometric Odor Sensors 88

4.5.1 MOSFET 88

4.6 Gravimetric Odor Sensors 89

ContentsVI

4.6.1 QCM 90

4.6.2 SAW 92

4.7 Optical Odor Sensors 93

4.7.1 SPR 93

4.7.2 Fluorescent Odor Sensors 94

4.7.3 Other Optical Approaches 95

4.8 Thermal (Calorimetric) Sensors 96

4.9 Amperometric Sensors 96

4.10 Summary of Chemical Sensors 98

5 Signal Conditioning and Preprocessing 105

5.1 Introduction 105

5.2 Interface Circuits 106

5.2.1 Chemoresistors 106

5.2.1.1 Voltage Dividers 106

5.2.1.2 The Wheatstone Bridge 108

5.2.1.3 AC Impedance Spectroscopy 109

5.2.2 Acoustic Wave Sensors 110

5.2.3 Field-Effect Gas Sensors 112

5.2.4 Temperature Control 113

5.3 Signal Conditioning 114

5.3.1 Operational Amplifiers 114

5.3.2 Buffering 116

5.3.3 Amplification 116

5.3.4 Filtering 116

5.3.5 Compensation 118

5.3.5.1 Linearization of Resistance Measurements 118

5.3.5.2 Miscellaneous Functions 119

5.4 Signal Preprocessing 120

5.4.1 Baseline Manipulation 120

5.4.2 Compression 122

5.4.3 Normalization 123

5.4.3.1 Local Methods 123

5.4.3.2 Global Methods 125

5.5 Noise in Sensors and Circuits 125

5.6 Outlook 128

5.6.1 Temperature Modulation 128

5.7 Conclusions 129

5.8 Acknowledgements 130

6 Pattern Analysis for Electronic Noses 133


6.1.1 Nature of Sensor Array Data 135

6.1.2 Classification of Analysis Techniques 136

6.1.3 Overview 137

Contents VII

6.2 Statistical Pattern Analysis Techniques 138

6.2.1 Linear Calibration Methods 139

6.2.2 Linear Discriminant Analysis (LDA) 140

6.2.3 Principal Components Analysis (PCA) 141

6.2.4 Cluster Analysis (CA) 143

6.3 ‘Intelligent’ Pattern Analysis Techniques 145

6.3.1 Multilayer Feedforward Networks 146

6.3.2 Competitive and Feature Mapping Networks 150

6.3.3 ‘Fuzzy’ Based Pattern Analysis 152

6.3.4 Neuro-Fuzzy Systems (NFS) 154

6.4 Outlook and Conclusions 155

6.4.1 Criteria for Comparison 155

6.4.2 Intelligent Sensor Systems 157

6.4.3 Conclusions 158

7 Commercial Electronic Nose Instruments 161


7.1.1 Geographical Expansion 162

7.1.2 Scientific and Technological Broadening 162

7.1.3 Conceptual Expansion 163

7.2 Commercial Availability 164

7.2.1 Global Market Players 164

7.2.1.1 Alpha M.O.S. 165

7.2.1.2 AppliedSensor Group 165

7.2.1.3 Lennartz Electronic 167

7.2.1.4 Marconi Applied Technologies (now ELV Technologies) 167

7.2.1.5 Osmetech plc 168

7.2.2 Handheld Devices 170

7.2.2.1 AppliedSensor Group 170

7.2.2.2 Cyrano Sciences, Inc. 170

7.2.2.3 Microsensor Systems, Inc. 171

7.2.3 Enthusiastic Sensor Developers 171

7.2.3.1 Bloodhound Sensors Ltd. 171

7.2.3.2 HKR Sensorsysteme GmbH 171

7.2.3.3 OligoSense n.v. 172

7.2.3.4 Quality Sensor Systems Ltd. 172

7.2.3.5 Quartz Technology Ltd. 172

7.2.3.6 Technobiochip 173

7.2.4 Non-Electronic Noses 173

7.2.4.1 Laboratory of Dr. Zesiger 173

7.2.4.2 Agilent Technologies, Inc. 174

7.2.4.3 Illumina, Inc. 174

7.2.4.4 Electronic Sensor Technology, Inc. 174

7.2.5 Specific Driven Applications 175

7.2.5.1 Astrium 175

ContentsVIII

7.2.5.2 Element Ltd. 175

7.2.5.3 Environics Industry Oy 175

7.2.5.4 WMA Airsense Analysentechnik GmbH 175

7.3 Some Market Considerations 176

8 Optical Electronic Noses 181


8.1.1 Optical Sensors 181

8.1.2 Advantages and Disadvantages of Optical Transduction 182

8.2 Optical Vapor Sensing 183

8.2.1 Waveguides 183

8.2.2 Luminescent Methods 183

8.2.3 Colorimetric Methods 185

8.2.4 Surface Plasmon Resonance (SPR) 187

8.2.5 Interference and Reflection-Based Methods 189

8.2.6 Scanning Light-Pulse Technique 191

8.3 The Tufts Artificial Nose 191

8.4 Conclusion 198

9 Hand-held and Palm-Top Chemical Microsensor Systems for Gas Analysis 201


9.2 Conventional Hand-held Systems 203

9.2.1 Hardware Setup 203

9.2.2 Fundamentals of the Sensing Process 205

9.2.3. Commercially Available Instruments Based on ConventionalTechnology 206

9.2.3.1 Hand-held Units Based on Mass-Sensitive Sensors 207

9.2.3.2 Hand-held Units Based on Chemoresistors 210

9.3 Silicon-Based Microsensors 211

9.3.1 Micromachining Techniques 212

9.3.1.1 Bulk Micromachining 212

9.3.1.2 Surface Micromachining 213

9.3.2 Microstructured Chemocapacitors 213

9.3.3 Micromachined Resonating Cantilevers 216

9.3.4 Micromachined Calorimetric Sensors 219

9.3.5 Single-Chip Multisensor System 221

9.3.6 Operation Modes for CMOS Microsystems 223

9.3.6.1 Reverse Mode of Operation (RMO) 224

9.4 Summary and Outlook 226

10 Integrated Electronic Noses and Microsystems for Chemical Analysis 231


10.2 Microcomponents for Fluid Handling 233

10.2.1 Microchannels and Mixing Chambers 233

10.2.2 Microvalves 238

Contents IX

10.2.2.1 Active Microvalves 238

10.2.2.2 Passive Microvalves (Check Valves) 240

10.2.3 Micropumps 241

10.2.3.1 Mechanical Micropumps 241

10.2.3.2 Nonmechanical Micropumps 245

10.3 Integrated E-Nose Systems 245

10.3.1 Monotype Sensor Arrays 245

10.3.2 Multi-type Sensor Arrays 250

10.4 Microsystems for Chemical Analysis 251

10.4.1 Gas Chromatographs 251

10.4.2 Mass Spectrometers 255

10.4.3 Optical Spectrometers 258

10.5 Future Outlook 260

11 Electronic Tongues and Combinations of Artificial Senses 267


11.2 Electronic Tongues 269

11.2.1 Measurement Principles 269

11.2.2 Potentiometric Devices 270

11.2.2.1 The Taste Sensor 271

11.2.2.2 Ion-Selective Electrodes 273

11.2.2.3 Surface Potential Mapping Methods 274

11.2.3 Voltammetric Devices 275

11.2.3.1 The Voltammetric Electronic Tongue 277

11.2.3.2 Feature Extraction 279

11.2.3.3 Industrial Applications using the Voltammetric Electronic Tongue 280

11.2.3 Piezoelectric Devices 283

11.3 The Combination or Fusion of Artificial Senses 284

11.3.1 The Combination of an Electronic Nose and an Electronic Tongue 285

11.3.2 The Artificial Mouth and Sensor Head 286

11.4 Conclusions 287

12 Dynamic Pattern Recognition Methods and System Identification 293


12.2 Dynamic Models and System Identification 294

12.2.1 Linear Models 295

12.2.2 Multi-exponential Models 297

12.2.3 Non-linear Models 300

12.3 Identifying a Model 304

12.3.1 Non-Parametric Approach 304

12.3.1.1 Time-Domain Methods 305

12.3.1.2 Frequency-Domain Methods 307

12.3.2 Parametric Approach 308

12.4 Dynamic Models and Intelligent Sensor Systems 309

12.4.1 Dynamic Pattern Recognition for Selectivity Enhancement 311

ContentsX

12.4.2 Calibration Time Reduction 314

12.4.3 Building of Response Models 315

12.4.4 Drift Counteraction 317

12.5 Outlook 319

13 Drift Compensation, Standards, and Calibration Methods 325

13.1 Physical Reasons for Drift and Sensor Poisoning 325

13.2 Examples of Sensor Drift 329

13.3 Comparison of Drift and Noise 331

13.4 Model Building Strategies 332

13.5 Calibration Transfer 332

13.6 Drift Compensation 333

13.6.1 Reference Gas Methods 335

13.6.2 Modeling of Sensor Behavior 339

13.6.3 Pattern-Oriented Techniques for Classification 340

13.6.4 Drift-Free Parameters 343

13.6.5 Self-Adapting Models 343


14 Chemical Sensor Array Optimization: Geometric and Information Theoretic

Approaches 347

14.1 The Need for Array Performance Definition and Optimization 347

14.2 Historical Perspective 349

14.3 Geometric Interpretation 351

14.3.1 Linear Transformations 352

14.4 Noise Considerations 355

14.4.1 Number of Discriminable Features 355

14.4.2 Measurement Accuracy 357

14.4.3 2-Sensor 2-Odor Example 360

14.5 Non-linear Transformations 363

14.6 Array Performance as a Statistical Estimation Problem 366

14.7 Fisher Information Matrix and the Best Unbiased Estimator 367

14.8 FIM Calculations for Chemosensors 369

14.8.1 2-Sensor 2-Odor Example 370

14.9 Performance Optimization 370

14.9.1 Optimization Example 371


14.A Overdetermined Case 375

14.B General Case with Gaussian Input Statistics 375

14.C Equivalence Between the Geometric Approach and the Fisher InformationMaximization 375

15 Correlating Electronic Nose and Sensory Panel Data 377

15.2 Sensory Panel Methods 378

15.2.1 Odor Perception 378

Contents XI

15.2.2 Measurement of Detectability 379

15.2.3 Transforming the Measurement of the Subject to the Subject’sMeasurement of an Odor 379

15.2.4 Assessor Selection 380

15.2.5 Types of Dynamic Dilution Olfactometry 380

15.2.5.1 Choice Modes 380

15.2.5.2 Yes/No Mode 381

15.2.5.3 The Forced Choice Mode 381

15.2.5.4 Laboratory Conditions 382

15.2.5.5 Laboratory Performance Quality Criteria 382

15.2.5.6 Compliance with the Quality Criteria 383

15.2.6 Assessment of Odor Intensity 384

15.2.7 Assessment of Odor Quality 386

15.2.8 Judgment of Hedonic Tone 387

15.3 Applications of Electronic Noses for Correlating Sensory Data 387

15.4 Algorithms for Correlating Sensor Array Data with Sensory Panels 388

15.4.1 Multidimensional Scaling 389

15.4.2 Regression Methods 390

15.4.3 Principal Components Regression 391

15.4.4 Partial Least Squares Regression 391

15.4.5 Neural Networks 392

15.4.6 Fuzzy-Based Data Analysis 392

15.5 Correlations of Electronic Nose Data with Sensory Panel Data 393

15.5.1 Data from Mouldy Grain 394


16 Machine Olfaction for Mobile Robots 399


16.2 Olfactory-Guided Behavior of Animals 400

16.2.1 Basic Behaviors Found in Small Organisms 400

16.2.2 Plume Tracking 400

16.2.3 Trail Following by Ant 402

16.3 Sensors and Signal Processing in Mobile Robots 403

16.3.1 Chemical Sensors 403

16.3.2 Robot Platforms 404

16.4. Trail Following Robots 404

16.4.1 Odor Trails to Guide Robots 404

16.4.2 Robot Implementations 406

16.4.3 Engineering Technologies for Trail-Following Robots 406

16.5 Plume Tracking Robots 407

16.5.1 Chemotactic Robots 408

16.5.2 Olfactory Triggered Anemotaxis 410

16.5.3 Multiphase Search Algorithm 411

16.6 Other Technologies in Developing Plume Tracking Systems 413

16.6.1 Olfactory Video Camera 413

ContentsXII

16.6.2 Odor Compass 414

16.7 Concluding Remarks 416

17 Environmental Monitoring 419


17.1.1 Water 419

17.1.2 Land 421

17.1.3 Air 421

17.2 Special Considerations for Environmental Monitoring 425

17.2.1 Sample Handling Problems 425

17.2.1.1 Sample Lifetime 425

17.2.1.2 Humidity 425

17.2.1.3 Extraction of volatiles 425

17.2.1.4 Tubing system 425

17.2.1.5 Temperature 425

17.2.2 Signal Processing Challenges 426

17.3 Case Study 1: Livestock Odor Classification 426

17.3.1 Background 426

17.3.2. Description of the problem 427

17.3.3. Methods 427

17.3.4 Signal Processing Algorithms 428

17.3.4.1 Bias Removal 428

17.3.4.2 Humidity 428

17.3.4.3 Concentration 428

17.3.4.4 Dimensionality Reduction 428

17.3.5. Results 429

17.3.6. Discussion 429

17.4 Case Study 2: Swine Odor Detection Thresholds 430

17.4.1. Description of the Problem 430

17.4.2 Methods 431

17.4.3 Results 431

17.4.4 Discussion 431

17.5 Case Study 3: Biofilter Evaluation 432

17.5.1 Description of the Problem 432

17.5.2 Methods 432

17.5.3. Results 434


17.6 Case Study 4: Mold Detection 437


17.6.2 Description of the Problem 437

17.6.3 The NC State E-Nose 437

17.6.4 Methods 440

17.6.5 Results 440


17.7 Future Directions 441

Contents XIII

18 Medical Diagnostics and Health Monitoring 445


18.2 Special Considerations in Medical/Healthcare Applications 449

18.3 Monitoring Metabolic Defects in Humans Using a Conducting PolymerSensor Array to Measure Odor 450


18.3.2 Methodology 451

18.3.3 Results 452

18.4 The Use of an Electronic Nose for the Detection of BacterialVaginosis 454


18.4.2 Methodology 456

18.4.3 Results 456


18.4.5 Conclusion 458

18.5 Conclusion 458

19 Recognition of Natural Products 461


19.2 Recent Literature Review 462

19.3 Sampling Techniques 462

19.3.1 Sample Containment 462

19.3.2 Sample Treatments 468

19.3.2.1 Heating 468

19.3.2.2 Cooling 468

19.3.2.3 Removal of Base Component 468

19.3.2.4 Preconcentration 469

19.3.2.5 Grinding 469

19.3.3 Instrument and Sample Conditioning 469

19.3.3.1 Modifying Baseline 470

19.3.3.2 Purge Technique 470

19.3.3.3 Temperature Control 470

19.3.4 Sample Storage 470

19.3.5 Seasonal Variations 471

19.3.6 Inherent Variability of Natural Products 471

19.4 Case Study: The Rapid Detection of Natural Products as a Means ofIdentifying Plant Species 471

19.4.1 Wood Chip Sorting 472

19.4.2 Experimental Procedure 472

19.4.3 SPME-GC Analysis of the Sapwood of the Conifers Used in Pulp and PaperIndustries 473

19.4.4 Conclusion: Wood Chip Sorting 475

19.5 Case Study: Differentiation of Essential Oil-Bearing Plants 475

19.5.1 Golden Rod Essential Oils 475

19.5.2 Essential Oils of Tansy 477

ContentsXIV

19.5.3 Conclusion: Essential Oils 478

19.6 Conclusion and Future Outlook 478

20 Process Monitoring 481


20.1.1 On-line Bioprocess Monitoring 482

20.1.2 At-line Food Process Monitoring 483

20.2 Previous Work 483

20.2.1 Quantitative Bioprocess Monitoring 483

20.2.2 Qualitative Bioprocess Monitoring 485

20.2.3 At-line Food Process Monitoring 486

20.3 Special Considerations 487

20.4 Selected Process Monitoring Examples 487

20.4.1 On-line Monitoring of Bioprocesses 487

20.4.2 At-line Monitoring of a Feed Raw Material Production Process 488

20.4.3 Monitoring Setup 489

20.4.4 Signal Processing 489

20.4.5 Chemometrics 491

20.4.5.1 Study 1: Estimation of Cell growth in Escherichia coli Fermentations 491

20.4.5.2 Study 2: Physiologically Motivated Monitoring of Escherichia coliFermentations 493

20.4.5.3 Study 3: Quality Control of a Slaughter Waste Process 496

20.4.5.4 Discussion 500

20.5 Future Prospects 501

21 Food and Beverage Quality Assurance 505


21.2 Literature Survey 507

21.3 Methodological Issues in Food Measurement with Electronic Nose 510

21.4 Selected Case 511

21.4.1 LibraNose 511

21.4.2 Case Study: Fish Quality 515



22 Automotive and Aerospace Applications 525


22.2 Automotive Applications 525

22.3 Aerospace Applications 526

22.4 Polymer Composite Films 529

22.5 Electronic Nose Operation in Spacecraft 530

22.5.1 The JPL Enose Flight Experiment 532

22.5.2 Data Analysis 533

22.5.2.1 Data Pre-Processing 534

22.5.3 Pattern Recognition Method 536

Contents XV

22.6 Method Development 536

22.6.1 Levenberg-Marquart Nonlinear Least Squares Method 537

22.6.2 Single gases 539

22.6.3 Mixed Gases 541

22.6.4 STS-95 Flight Data Analysis Results 541


22.7.1 Sensors 543

22.7.2 Data Acquisition 543

22.7.3 Data Analysis 544

22.8 Conclusion 545

23 Detection of Explosives 547



23.3 State-of-the-art of Various Explosive Vapor Sensors 549

23.4 Case Study 557



24 Cosmetics and Fragrances 561


24.2 The Case for an Electronic Nose in Perfumery 562

24.3 Current Challenges and Limitations of Electronic Noses 563

24.4 Literature Review of Electronic Noses in Perfumery and Cosmetics 564

24.5 Special Considerations for using Electronic Noses to Classify and JudgeQuality of Perfumes, PRMs, and Products 566

24.6 Case Study 1: Use in Classification of PRMs with Different Odor Characterbut of Similar Composition 567

24.6.1 The Problem 567

24.6.2 Methods 568

24.6.3 Results 568

24.6.4 Conclusions for Case Study 1 570

24.7 Case Study 2: Use in Judging the Odor Quality of a Sunscreen Product 570


24.7.2 The Problem 572

24.7.3 Equipment and Methods 573

24.7.3.1 Equipment 573

24.7.4 Results 574

24.7.4.1 Sensory Correlation and Long Term Repeatability 574

24.7.4.2 Database transfer from Dubendorf to Vernier 574

24.7.5 Conclusions for Case Study 2 575



Index 579

ContentsXVI

Preface

In the past decade, electronic nose instrumentation has generated much interestinternationally for its potential to solve a wide variety of problems in fragrance andcosmetics production, food and beverages manufacturing, chemical engineering,environmental monitoring, and more recently, medical diagnostics and biopro-cesses. Several dozen companies are now designing and selling electronic nose unitsglobally for a wide variety of expandingmarkets. An electronic nose is amachine that isdesigned to detect and discriminate among complex odors using a sensor array. Thesensor array consists of broadly tuned (non-specific) sensors that are treated with avariety of odor-sensitive biological or chemical materials. An odor stimulus generatesa characteristic fingerprint (or smellprint) from the sensor array. Patterns or finger-prints from known odors are used to construct a database and train a pattern recogni-tion system so that unknown odors can subsequently be classified and identified.Thus, electronic nose instruments are comprised of hardware components to collectand transport odors to the sensor array – as well as electronic circuitry to digitize andstore the sensor responses for signal processing.This book provides a comprehensive and timely overview of our current state of

knowledge of the use of electronic sensors for detection and identification of odorouscompounds and mixtures. The handbook covers the scientific principles and technol-ogies that are necessary to implement the use of an electronic nose. A comprehensiveand definitive coverage of this emerging field is provided for both academic and prac-ticing scientists. The handbook is intended to enable readers with a specific back-ground, e.g. sensor technology, to become acquainted with other specialist aspectsof this very multidisciplinary field.Following this Preface, Part A covers the fundamentals of the key aspects related to

electronic nose technology, from the biological olfactory system that has inspired thedevelopment of electronic nose technology, through to sensor materials and patternanalysis methods for use with chemical sensor arrays. This section provides a valuabletutorial for those readers who are new to the field before delving into the more spe-cialist material in later chapters.More advanced aspects of the technology are dealt with in Parts B and C, which

provide an up-to-date survey of current research directions in the areas of instrumen-tation (Part B) and pattern analysis (Part C). Advanced instrumentation issues include


XVII

novel sensingmaterials through to handheld chemical sensing devices and distributedchemosensory systems.Recent topics in pattern analysis include on-line learning methods to extend cali-

bration life-span, dynamic signal processing methods that exploit sensor transientbehavior and optimization strategies for chemical sensor arrays.An important element of the handbook is the inclusion of case studies of various

applications of the electronic nose (Part D). Leading manufacturers of electronic noseequipment and key end-users have provided most of the chapters covering severalinteresting application areas.

Part A Overview: Fundamentals of Odor Sensing

Part A of the book is an overview of the fundamental key aspects of biological andmachine olfaction. The section begins with two chapters that review the field of bio-logical olfaction. The next four chapters cover the basic functional components ofelectronic noses including the sample handling system, gas sensor arrays andtypes, and signal processing systems for classification and identification of odorouscompounds. The first chapter by Schiffman and Pearce describes how the biologicalsense of smell utilizes a remarkable sensor array of neurons that detects and discri-minates among a vast number of volatile compounds (and mixtures of compounds)present in minute concentrations. This exquisite sensitivity is the reason why scien-tists and engineers have developed and begun to market machines that mimic thisbiological apparatus to detect and discriminate among volatile chemicals. The initialchapter provides an overview of the physicochemical andmolecular properties of odor-ous molecules (called odorants) along with a description of odor classification and itslimitations. It also provides an introduction to the biological olfactory pathway includ-ing descriptions of the olfactory epithelium, olfactory sensory neurons, seven-mem-brane-spanning receptors, the olfactory bulb, and the olfactory cortex. The chapteremphasizes that as few as 40 molecules of some compounds (e.g. mercaptans) aresufficient for humans to perceive an odor. Second, the range of distinctive odor sensa-tions is vast, and a skilled perfume chemist can recognize and distinguish 8000 to10 000 different substances on the basis of their odor quality. The remarkable discri-minability is achieved by a coding scheme in which different odor stimuli are recog-nized by different combinations of olfactory receptors. That is, the biological olfactorysystem uses a combinatorial receptor coding scheme such that the specific patterns ofactivation across many neurons induced by an odor stimulus makes it possible todiscriminate among the vast number of distinct smells.The second chapter of Part A by Cometto-Muniz expands on the first chapter with

additional details of human olfactory perception and an overview of the topic of che-mesthesis (the common chemical sense). Olfactory perception is achieved by stimula-tion of the olfactory nerve (cranial nerve I), which allows us to discriminate betweenodor stimuli such as chocolate and coffee. Chemesthetic sensations, on the otherhand, include piquancy, prickling, stinging, burning, freshness, tingling, and irrita-tion, which are grouped under the term pungency and are mediated by a different

PrefaceXVIII

nerve called the trigeminal nerve (cranial nerve V). Airborne compounds elicit odorsensations at concentrations below those that induce pungency. Methods for quantify-ing odor and pungency in humans are described including the determination ofthresholds, the relationship between concentration and perceived intensity, and thesensory consequences of adding multiple compounds together in a mixture. Ap-proaches for quantifying odor with static olfactometry, dynamic olfactometry, and en-vironmental chambers are explained. In static olfactometry the vapor stimulus isdrawn from an enclosed container in which the liquid and odorous vapor of the che-mical(s) are in equilibrium with one another. In dynamic olfactometry, the vapor flowscontinuously in a carrier-gas stream, typically odorless air or nitrogen. Amathematicalmodel is presented that can be used to predict odor and pungency threshold concen-trations from physicochemical determinants. Instrumentation currently used by theflavor industry to analyze odorous mixtures including gas chromatography and massspectrometry (GC/MS) is described. Overall, the sense of smell outperforms conven-tional analytic instruments (specifically GC/MS) in detecting and identifying odoroussubstances.The third chapter by Nakamoto covers basic principals of odor handling and delivery

of samples to electronic noses with two main types of systems (flow and static) de-scribed. In flow systems, the sensors are placed in the vapor flow of the samplingsystem so that the vapor around the sensors is constantly exchanged. Several flowsystems are described, including headspace sampling, diffusion and permeationmethods, a bubbler, and sampling bags. In static systems there is no vapor flowaround the sensors but rather the sensors are exposed to vapor with a constant con-centration. For static systems, the steady-state response of the sensors is measured. Anopen system is also illustrated in which a sensor is directly exposed to a vapor without asensor chamber. Because different types of sensors vary widely in their sensitivity,methods for increasing the sensitivity are described using a preconcentrator tube.The physics of evaporation are also covered because most samples submitted to elec-tronic noses are liquids from which odorants are evaporated. Issues of removal ofhumidity from samples are also described.The fourth chapter by Nanto and Stetter is an overview of chemosensors that can be

used in electronic nose systems to convert chemical information into an electricalsignal. The chapter describes conductometric chemosensors (metal-oxide semicon-ductors (MOS) and conducting polymers (CPs)), chemocapacitors, potentiometric che-mosensors (e.g. MOS field-effect transistors (MOSFETs)), gravimetric chemosensors(quartz crystal microbalance (QCM), surface acoustic wave (SAW)), optical chemosen-sors (surface plasmon resonance (SPR), fluorescent sensors), calorimetric sensors,and amperometric sensors. The underlying principle of conductometric sensors(also called chemoresistors) is the conductivity change that occurs when gaseous mo-lecules react chemically withMOS or organic CPs. These are the simplest of type of gassensors and are widely used to make arrays for gas and odor measurements. In che-mocapacitor (CAP) devices, a polymer adsorbs the gaseous analyte, which alters theelectrical (e.g. dielectric constant e) and physical properties (e.g. volume V) of thepolymer relative to the baseline capacitance of the polymer when no gaseous analytemolecules are present. Potentiometric chemosensors of the MOSFET type utilize a

Preface XIX

gate that is made of a gas sensitive metal as a catalyst for gas sensing. Gravimetric odorsensors detect the effect of sorbed molecules on propagation of acoustic waves. Thetwo main types of gravimetric sensors include QCM and SAW devices that are con-figured as mass-change sensing devices in the electronic nose. Optical chemosensorshave several principals of operation. SPR is a physical process that can occur whenplane-polarized light hits a metal film under total internal reflection conditions. Inorder to utilize this system as a gas sensor, a very thin film of methylmethacrylate,polyester resin or propylene ether as a sensing membrane can be deposited ongold metal thin film, and the angle of the reflected light is measured. Anothertype of chemosensor consists of optical fibers deposited with a fluorescent indicatordye in polymer matrices of varying polarity, hydrophobicity, pore size, elasticity, andswelling tendency to create unique sensing regions that interact differently with vapormolecules. Thermal sensors record the heat of solution of an analyte in the coating,with greater heat generated by larger amounts of absorbed analyte. The principle ofamperometric gas sensors is the electrochemical oxidation or reduction of the analytegas at a catalytic electrode surface that generates electrical current proportional to theconcentration of the analyte.The next chapter by Gutierrez-Osuna, Nagle, Kermani, and Schiffman covers inter-

face circuits, signal conditioning electronics, and pre-processing algorithms; topicsthat serve as a bridge between the previous chapter on odor sensors (see Nantoand Stetter Chapter 4) and the following chapter on pattern analysis techniques(Hines and colleagues Chapter 6). The chapter presents a review of interface circuitsfor the most widely used odor sensors (chemoresistive, acoustic wave, and field effect),as well as an introduction to analog conditioning circuits for signal amplification,filtering, and compensation. Signal preprocessing algorithms commonly used priorto pattern analysis, including baseline manipulation, compression, and normaliza-tion, are also reviewed.The final chapter in Section A by Hines, Boilot, Gardner, Gongora, Llobet deals with

pattern analysis for electronic noses. There is an introduction into the nature of sensorarray data and classification of analysis techniques including conventional statisticalmethods as well as biologically motivated technologies. This is followed by a moredetailed discussion of statistical techniques such as principal components analysis(PCA), discriminant function analysis (DFA), partial least squares (PLS), multiple lin-ear regression (MLR), and cluster analysis (CA) including nearest neighbor (NN). Thediscussion of biologically motivated technologies covers artificial neural networks(ANN), fuzzy inference systems (FIS), self-organizing map (SOM), radial basis func-tion (RBF), genetic algorithms (GA), wavelets, neuro-fuzzy systems (NFS), and adap-tive resonance theory (ART). Biologically motivated technologies for pattern analysisare especially attractive for use with electronic nose technology because they have thepotential to perform incremental learning and offer self-organizing and self-stabilizingpotential.

PrefaceXX

Part B Overview: Advanced Instrumentation

Part B of the book describes in some detail sensor technologies and instrumentationfor electronic nose systems. The section begins with a chapter that reviews the field ofelectronic nose instruments that are currently available. These commercial instru-ments are predominantly large desktop-based systems that require an automatedheadspace sampler and a personal computer to operate the whole system. More recentinstruments may be described as handheld but tend to have a limited battery lifecaused by either the need for the sensors to be held at a constant (elevated) tempera-ture or high computing power.The next chapter considers the development of optical rather than solid-state elec-

tronic noses. In this type of instrument, chemically sensitive materials are used as thesensing elements. For example, Dickinson et al. describe the operation of an optical‘smell camera’ based upon the 2D raster scanning of the surface of a distributed ca-pacitor, in order to read out the charge generated by a local catalytic reaction with theodor molecule. The composition and temperature of the catalyst, making up one elec-trode of the capacitor, is varied to generate a 2D image of the smell. In a differentapproach, Walt et al. coat a large number of small glass beads with a variety of fluor-escent indicator dyes and these are used to create pixels in a composite image of anodor. This involves the fixing of the beads on to the end of optical fibers to complete thetransducer. The process has been simplified more recently by Suslick et al. who havecreated a small rectangular array of porphyrin based sensing elements that changetheir chromatic properties when exposed to reactive gases. This colorimetric electro-nic nose can work from an ordinary light source and CCD array, and so is quite similarin technology to a commercial color flatbed scanner. The concept of an opto-electronicelectronic nose is an attractive one and it remains to be seen how this technologystands against the alternatives.The chapter by Baltes et al. explores the current research being undertaken in the

development of small palm-top electronic noses. The approach focuses on the use ofCMOS technology to fabricate a low-cost, low-power and miniature electronic nose.This necessitates the use of room-temperature gas-sensitive materials that can be de-posited at a low temperature (compared with CMOS processes). Consequently, thechapter describes the development of capacitors, resistors, calorimeters, and cantile-ver beams predominantly coated with compounds used as the stationary phase in gaschromatography, i.e. rubbers and polymers. The fabrication of CMOS sensors permitthe integration of CMOS or even BiCMOS circuitry next to the sensing elements andthus produce simple voltage read out. It is thus an attractive technology for the pro-duction of electronic noses at high volume, e.g. millions of units per year.Gardner et al. expands upon the concept of a micro nose and investigates the pos-

sible development of an electronic nose that has integrated mechanical as well as elec-trical components. There has been rapid progress in the field of micro electro mechan-ical systems in recent years and this chapter considers related advances in the fabrica-tion of micro valves, micro pumps and other micro-fluidic components. The chal-lenges associated with making an analytical instrument on a chip are also presentedwith a description of work being carried out to make micro gas chromatographs and

Preface XXI

micro mass spectrometers. This approach poses a number of technological challengesbecause it has to deal with the transportation of the odor through the nose as well as thesensing elements. However these analytical micro noses may well compete with solid-state noses in terms of discriminating power.The final chapter describes the advances taking place to create another sensory in-

strument, namely, the ‘electronic tongue’. Clearly, an instrument that can mimic boththe sense of smell and taste would provide valuable information on the nature of theflavor of a compound. In some ways the electronic tongue, as described, here behavesas an electronic nose under water – in other words the chemical sensors work in theliquid rather than gaseous phase. Thus the sensors are not specific to detecting thefour primary tastes, salty, bitter, sweet, and sour (or putative additional taste primariessuch as metallic and monosodium glutamate) but will provide signals that can becorrelated with them. For example, the bitterness of a compound can be related tothe acidity (i.e. pH value) while the sweetness will relate to the conductivity. The spe-cificity of electrochemical sensors may be enhanced through the use of biological coat-ings of, for example, shear-horizontalmode SAW (SH-SAW) devices. Unfortunately,this type of biosensor tends to suffer (like all biosensors) from a short life when ex-posed to the environment. Nevertheless the development of electronic tongue technol-ogy could well lead to further advances in electronic nose technology.

Part C Overview: Advanced Signal Processing and Pattern Analysis

The foundations of signal processing strategies for chemical sensor array systemswere provided in Chapter 6, which outlined the fundamentals of applying signal pro-cessing (predominantly pattern recognition based) techniques to chemical sensor ar-rays, for recognizing and discriminating specific ‘fingerprints’ of sensor array re-sponse that correspond to distinct categories of odor stimuli. This section of the hand-book continues this theme by consideringmore advanced or, perhapsmore accurately,specialized aspects of signal processing related to chemical sensor arrays – each chap-ter exploring fertile areas for future research in machine olfaction.A key theme here is the technological advantage that can be achieved in these sys-

tems through the development of their integral signal/information processing system.The chapters in this section are representative of current trends in research in this areathat appear to emphasize two distinct aspects. First, the improvement in system per-formance through advances in information processing strategies applied to chemicalsensor arrays, for example by considering transient sensor response (as opposed to thesingle-valued steady-state response) to enhance discrimination or the detection thresh-old of these instruments. Second, widening the scope of applications of such systemsand solving novel chemosensory detection problems, for example by correlating quan-titative electronic nose data with qualitative human sensory panel information in anattempt to achieve automated sensory panel analysis through technological means.The first of these themes looks more to the past, in terms of refining and improving

on what has gone before, whereas the second theme is firmly looking to the future ofthis technology, in terms of opening up new domains in which the technology may be

PrefaceXXII

applied. For this reason this section of the handbook provides a taste (!) of some ex-citing prospects for the future of electronic nose technology as we move further intothe 21st century, which will be driven by parallel developments in sensor technologyand information processing capability.The performance of electronic nose systems depends greatly on each of its compo-

nents: from the odor delivery system; through to the choice and diversity of chemo-sensor materials; the interface circuitry; as well as the computational subsystem fordiscriminating between array responses.The first three chapters relate to the first theme – that is, how to improve system

performance by developing signal-processing strategies that may be applied to ma-chine olfaction. Although perhaps at first sight not quite as groundbreaking in itsambition as the second theme, the topics covered in these chapters are vital to thefuture welfare of this field as a commercial, scientific, and technological endeavor.Key issues are covered here that are important for overcoming existing technologicalbarriers to the take-up and deployment of the technology.The first chapter in this section, by Llobet, covers aspects of dynamical model ap-

proaches for interpreting chemical sensor response information. Shifting the empha-sis from steady-state sensor response information to transient sensor response pro-mises less sensitivity to drift, the possibility of yielding additional discrimination ofstimuli, and becomes essential when environmental conditions vary on a similar timescale as sensor response. An overview of a number of dynamical models and systemidentification techniques are provided alongside an example of how these might beapplied to a specific sensing problem.In many cases the practical performance of chemical sensor array systems is limited

by changes in characteristics of sensor response over time or with chemical exposure.Commercial systems require frequent calibration against known standardized sam-ples in order to minimize these effects and assure some minimum measurementaccuracy. In many cases, recalibration may be required on a daily basis in order tomaintain acceptable performance in the field. Therefore, the development of sig-nal-processing strategies that counteract the affect of these shifts in sensor character-istics to repeated and identical stimuli are of considerable importance to the practi-tioner and researcher. A true understanding of temporal drift in sensor characteristicswill only ultimately be found through a detailed physical understanding of interactionof chemicals with sensing materials. Even then, only if the mechanisms involved arepurely deterministic will it be possible to eliminate their effects entirely. In the mean-time, empirical methods for compensation can be developed and these are consideredby Artursson and Holmberg in Chapter 13 as practical strategies for coping with thisphenomenon in working instruments.Due to the distributed nature of chemical sensor arrays it is not simple to define

their sensing performance in terms of the properties of the underlying chemical sen-sors. However, this is vital if a rigorous approach to specification of sensor perfor-mance and future optimization of sensor arrays is ever going to be achieved. Pearceand Sanchez-Montanes (Chapter 14) describe recent work on quantifying sensor arrayperformance for multidimensional stimuli such as odors that allows the system detec-tion performance to be predicted given the tuning and noise properties of the under-

Preface XXIII

lying chemosensors. This allows the selection of chemosensors for specific detectiontasks to be made, which until recently has been achieved by ad hoc means. In thischapter the theory of performance definition is applied to consider the practical issueof optimizing detection thresholds in artificial olfactory systems.The final two chapters of this section describe new domains where artificial olfactory

systems find application. New areas of application open up to this technology all thetime but future challenges will also require new and refined signal-processing stra-tegies. Here we consider two areas where the signal-processing subsystems play akey part in this development.The first of these considers signal-processing strategies for correlating human-de-

fined sensory panel information with chemical sensor-array responses. This has im-portant consequences, particularly in the food and beverage industry wheremillions ofdollars are spent each year on both instrumental analyses (mostly GC and MS-basedmethods) and sensory panel investigations. Neither of these approaches in isolationoffers a complete picture of odor or flavor quality. By applying multivariate statisticalanalysis techniques to chemical sensor array data there is the possibility for artificialolfactory systems to provide the missing link between instrumental and sensory-basedinvestigations. Some of these methods and an example of an environmental monitor-ing problem is provided by Sneath and Persaud in Chapter 15.Finally a promising new area of research in machine olfaction is presented – apply-

ing chemical sensor systems to mobile robotic systems. Ishida and Moriizumi con-sider the possibilities for mobile chemosensory systems. Two possible modes of op-eration are considered here: relatively straightforward chemical trail following and thefar more complex problem of chemical source localization in turbulent odor plumes.Insect models are used as the inspiration for the approach – the ant for trail followingbehavior and the moth for chemotaxis within airborne odor plumes. Although theirexperiments are preliminary and work in this area is at an early stage, there are manyexciting research challenges that will need to be considered in the future.

Part D Overview: Applications and Case Studies

This final section of the Handbook presents a variety of areas in which electronic nosetechnology has been applied. In each application, the tools and techniques of Parts A,B, and C are selectively employed to achieve specific performance goals.In the first chapter, Nagle, Gutierrez-Osuna, Kermani, and Schiffman examine en-

vironmental applications. Examples of water, land, and air monitoring experimentsreported in the open literature are examined, followed by four case studies of workdone by the authors. The first three demonstrate the ability of the AromaScanA32S electronic nose to classify odors from animal confinement facilities. In thefirst, the A32S was employed to classify the source of an odor emission as beingfrom the lagoon, the confinement building exhaust fan, or a downwind ambientair. In the second, the A32S was used to determine the detection threshold concen-tration for acetic acid, a major individual constituent in swine slurry odor. In the thirdcase study, the A32S was used to evaluate the performance of a biofilter of earth, wood

PrefaceXXIV

chips, small twigs, and straw on the confinement building exhaust as an odor reme-diation measure. In the fourth case study, the NS State Electronic nose, a prototypeunit with fifteen commercially available MOSs, demonstrated that an electronic nosecan differentiate between five types of fungi that commonly lower indoor air quality inoffice buildings and industrial plants. These four case studies demonstrate that theelectronic nose can perform well in various environmental monitoring applications.The next chapter by Persaud, Pisanelli, and Evans gives a summary of medical di-

agnostics and health-monitoring applications. Many diseases and intoxications areaccompanied by characteristic odors, and their recognition can provide diagnosticclues, guide the laboratory evaluation, and affect the choice of immediate therapy.After reviewing the history of electronic nose uses in this area, two case studiesare introduced. In the first, metabolic changes due to myopathies are detected by ur-ine odor. The electronic nose was able to differentiate the normal population from thatwith myopathies. In the second case study, an electronic nose was employed to detectbacterial vaginosis. Success in this area led Osmetech to seek federal drugs adminis-tration (FDA) approval of one of their instruments for this application.Next, Deffenderfer, Feast, and Garneau provide a comprehensive overview of the

electronic nose as an analytical tool for applications in natural products rangingfrom identifying solvents and the discrimination of spirits, to beverage and grain qual-ity. Following this overview, they then illustrate two specific case studies. In the first,the Cyranose 320 is used to identify trees of different species for the pulp and paperindustries in eastern Canada. In the second case study, the Cyranose 320 is employedto differentiate essential oil-bearing plants. Their results indicate that the electronicnose has great potential in these industries.Process monitoring is the subject of the fourth applications chapter. Haugen and

Bachinger give an overview of the fundamentals of non-invasive on-line monitoring ofbiological processes, followed by two case studies. The electronic nose in their studiesused a set of 10MOSFETs sensors, up to 19MOS sensors and 1 CO2-monitor based oninfrared adsorption. The MOSFET sensors were produced in-house at Linkoping Uni-versity (Sweden) with different catalytic metal gates of Pd, Pt, and Ir. TheMOS sensorsused were commercially available sensors of Taguchi (TGS) or fuzzy inference systems(FIS) type. The electronic nose was used to monitor the aroma of cell cultures to gaininsight into cell and process state changes as well as to identify process faults. In theirfirst case study, ANN technology was used successfully to relate the gas sensor signalpattern to the cell biomass from Escherichia coli fermentations. The second case studyfocused on using an electronic nose to monitor the composition of the bioreactorheadspace gas, and thus to track physiological state changes. Fast cell transition stateswere monitored in a semiquantitative approach appropriate for on-line and non-inva-sive control of industrial bioprocesses.The next applications chapter focuses on food and beverage quality assurance. In

this chapter, DiNatale states that ‘the analysis of foodstuff is one of themost promisingand also the most traveled road towards industrial applications for this technology.’After a review of the literature in this field, a case study in fish freshness is de-tailed. The study uses a prototype instrument called the LibraNose from the Univer-sity of Rome ‘Tor Vergata’. The LibraNose is based on an array of QCM sensors whose

Preface XXV

chemical sensitivity is given bymolecular films of metalloporphyrins and similar com-pounds. Spoilage in fish can be detected through the measure of the amount ofamines, such as trimethylamine, in the headspace of storage containers. In thestudy, the LibraNose was able to track two important parameters indicating thatthe electronic nose is a good candidate for future use in food freshness applications.The next chapter focuses on automotive and aerospace electronic nose applications.

Automotive applications include monitoring the exhaust for combustion efficiency,monitoring the engine compartment for leaking oil or other fluids, and monitoringthe cabin air for passenger safety (offgassing of fabrics and materials, leaks of coolantfrom the air-conditioning system, and intake of air from the roadway and the enginecompartment). Aerospace applications vary from the addition of an electronic nose tostudy the variations in atmosphere over days or seasons on other planets, to monitor-ing air quality in human habitats. The electronic nose developed at the Jet PropulsionLaboratory (JPL) was designed to detect a suite of compounds in the crew habitat ofa spacecraft, an enclosed space where air is recycled and it is unlikely that unknownand unexpected vapors will be released. In this chapter, Ryan and Zhou present a casestudy in which the JPL ENose in a flight experiment on the Space Shuttle flight STS-95(October–November 1998) was tested as a continuous air quality monitor to distin-guish among, identify and quantify 10 common contaminants which may be presentas a spill or leak in the recirculated breathing air of the space shuttle or space station.The JPL ENose has an array of 32 sensors, coated with 16 polymers/carbon compositesensing films developed at Caltech. In the study, the JPL ENose was trained to 12compounds, the 10 compounds most likely to leak or spill and the other two beinghumidity change and vapor from a medical swab (2-propanol and water) used dailyto confirm that the device was operating properly. For all cases except one (formal-dehyde), the JPL ENose was able to detect the compound at or below the expectedlevels.Pamula investigates the use of the electronic nose for the detection of explosives.

After reviewing the literature in this important application of electronic nose technol-ogy, the author reviews progress of the defense advanced research projects agency(DARPA) program to detect explosive mines by their chemical signatures. The chap-ter concludes with a case study of the Nomadics’ Fido (Fluorescence ImpersonatingDog Olfaction) device. The device uses fluorescent polymer beads to detect traceamounts of TNT emanating from landmines. This technology shows great promisefor future deployment in demining applications.In the final applications chapter, Rodriguez, Tan, and Gygax survey electronic nose

applications in cosmetics and fragrances. Even though the use of electronic noses inthe cosmetic and fragrance industry has been more limited than in many other areas,the published literature shows that, with optimization, many cosmetic and fragrancerelated analytical tasks can be solved. After the literature review, this chapter presentstwo case studies. In the first, eight fragrant samples with distinct odor characters butsimilar bulk composition were tested. Samples were analyzed by anHP4440 ChemicalSensor and by capillary GC/FID. Both approaches were successful in classifying anddifferentiating the odorous samples. In the second study, an Alpha MOS Fox4000electronic nose with 18 chemical sensors and a human panel were used to judge

PrefaceXXVI

the odor quality of a sunscreen product. The product samples had already passed ana-lytical tests prior to undergoing sensory evaluation. Expert panel evaluations weremade on � 150 samples judged to fall in three categories: meets sensory standard,does not meet sensory standard but can be used as a ‘diluent’ when adjusting bulkquality, and does not meet sensory standard and is rejected. Over a six-month evalua-tion period, the Fox4000 demonstrated its ability to carry out sensory analyses by ac-curately classifying ‘good’ and ‘bad’ batches of the tested product.We believe that the material presented in the Handbook of Electronic Noses should

not only help readers to find out more about this new and challenging subject, but alsoact as a useful reference in the future.

November 2002


Preface XXVII

List of Contributors

Thomas BachingerIndependent ConsultantSt. Larsgatan 8B/6.3S-582 23 LinkopingSweden

Henry BaltesInstitut fur QuantenelektronikDept. Physik (D-PHYS)HPT H 7ETH HonggerbergCH-8093 ZurichSwitzerland

Marina ColeDivision of Electrical & ElectronicEngineeringSchool of EngineeringCoventry CV4 7ALUK

J. Enrique Cometto-MunizChemosensory Perception LaboratoryDept. of Surgery (Otolaryngology)University of California, San DiegoMail Code 0957La Jolla, CA 92093-0957USA

Todd DickinsonIllumina, Inc.9390 Towne Centre DriveSuite 200San DiegoCA 92121USA

Corrado Di NataleDepartment of Electronic EngineeringUniversity of Rome Tor Vergatavia di Tor Vergata 11000133 RomaItaly

Philip EvansOsmetech PLCElectra HouseElectra WayCreweCW1 6WZUK

Julian W. GardnerDivision of Electrical & ElectronicEngineeringSchool of EngineeringCoventry CV4 7ALUK


XXIX

Ricardo Gutierrez-Osuna401 Russ Engineering CenterComputer Science and EngineeringWright State UniversityDayton, OH 45435USA

John-Erik HaugenMATFORSKOsloveien 1N-1430 AsNorway

Andreas HierlemannPhysical Electronics LaboratoryETH Hoenggerberg, HPT-H 4.2, IQECH-8093 ZurichSwitzerland

Evor L. HinesElectrical & Electronic EngineeringDivisionSchool of EngineeringUniversity of WarwickCoventryCV4 7ALUK

Martin HolmbergS-SENCE and Applied PhysicsIFMLinkoping UniversityS-581 83 LinkopingSweden

Bahram G. KermaniIllumina, Inc.9390 Towne Centre Drive, Suite 200San DiegoCA 92121-3015USA

Eduard LlobetDept. of Electronic EngineeringUniversitat Rovira i VirgiliAutovia de Salou s/n43006, TarragonaCataloniaSpain

Toysaka MoriizumiFaculty of EngineeringTokyo Insititute of TechnologyOokayama, Meguro–KuTokyo 152Japan

H. Troy NagleDepartment of Electrical and ComputerEngineering432 Daniels HallNorth Carolina State UniversityBox 79 11RaleighNC 27695-7911USA

Takamichi NakamotoDepartment of Physical ElectronicsGraduaute school of Science andEngineeringTokyo Institute of Technology2-12-1, Ookayama, Meguro-kuTokyo 152-8552Japan

Hidehito NantoChair Division of Materials ScienceAdvanced Materials Science Research &Development CentreKanazawa Institute of Technology3-1 YatsukahoMattoIshikawa 924-0838Japan

List of ContributorsXXX

Vamsee K. PamulaElectrical EngineeringDuke UniversityDurham, NCUSA

Tim C. PearceDepartment of EngineeringUniversity of LeicesterUniversity RoadLeicester LE1 7RHUK

Krishna C. PersaudDIASUMISTPO BOX 88Sackville StreetManchester M60 1QDUK

M. A. RyanMail Stop 303–308Jet Propulsion Laboratory4800 Oak Grove DrivePasadena CA 91109USA

Manuel A. Sanchez-MontanesETS de InformaticaUniversidad Autonoma de MadridMadrid 28049Spain

Susan S. SchiffmanDepartment of Psychiatry54212 Woodhall BuildingBox 32 59Duke University Medical SchoolDurham, NC 27710USA

Robert SneathSilsoe Research InstituteWrest ParkSilsoeBedford MK45 4HSUK

Joseph StetterDepartment of Biological, Chemical &Physical SciencesLife Sciences Building, room 1823101 South Dearborn St.Chicago, IL 60616USA

Tsung TanAlpha MOS Add.20 avenue Didier Daurat31400 ToulouseFrance

Emmanuel VannesteUniversity of AntwerpenUniversiteitsplein 1 C2.28B-2610 WilrijkBelgium

David WaltDepartment of ChemistryTufts UniversityPearson LabMedford, MA 02155USA

Udo WeimarInstitute of Physical ChemistryUniversity of TubingenAuf der Morgenstelle 8D-72076 TubingenGermany

List of Contributors XXXI

Fredrik WinquistDivision of Applied Physics andthe Swedish Sensor CenterDepartment of Physics andMeasurement TechnologyLinkoping UniversityS-581 83 LinkopingSweden

Otto WolfbeisUniversity of RegensburgInstitute of AnalyticalChemistryDE-93040 RegensburgGermany

Francois-Xavier GarneauDepartement des SciencesFondamentalesUniversite du Quebec a Chicoutimi555 Boulevard de l’UniversiteChicoutimi (Quebec)G7H 2B1Canada

Hanying ZhouMS 303–300Jet Propulsion Laboratory4800 Oak Grove DrivePasadena CA 91109USA

List of ContributorsXXXII

Part A

Fundamentals of Odor Sensing

Handbook of Machine Olfaction: Electronic Nose Technology.Edited by T.C. Pearce, S.S. Schiffman, H.T. Nagle, J.W. GardnerCopyright ª 2003 WILEY-VCH Verlag GmbH Co. KGaA, WeinheimISBN: 3-527-30358-8

1

Introduction to Olfaction: Perception, Anatomy, Physiology,

and Molecular Biology

Susan S. Schiffman, Tim C. Pearce

Abstract

Odors are sensations that occur when compounds (called odorants) stimulate recep-tors located in the olfactory epithelium at the roof of the nasal cavity. Odorants arehydrophobic, volatile compounds with a molecular weight of less than 300 daltons.Humans can recognize and distinguish up to 10 000 different substances on the basisof their odor quality. Odorant receptors (ORs) in the nasal cavity detect and discrimi-nate among these thousands of diverse chemical ligands. An individual odorant canbind to multiple receptor types, and structurally different odorants can bind to a singlereceptor. Specific patterns of activation generate signals that allow us to discriminatebetween the vast number of distinct smells. The physicochemical attributes of odor-ants that induce specific odor sensations are not well understood. The genes that codefor ORs have been cloned, and results from cloning studies indicate that ORs aremembers of a superfamily of hundreds of different G-protein-coupled receptorsthat possess seven transmembrane domains. A complete knowledge of structure-odor relationships in olfaction awaits the three-dimensional analysis of this large fa-mily of ORs. Ultimately, simultaneous knowledge of the three-dimensional structureof ORs as well as odorants will allow us to develop a pattern recognition paradigm thatcan predict odor quality.

1.1

Introduction to Olfaction

All living organisms from simple bacteria to complex mammals including humansrespond to chemicals in their environment. Chemical signals play a major role infeeding (e.g. nutrients), territorial recognition, sexual behavior, and detection of po-tentially harmful conditions such as fire, gas, and rancid food. In higher organisms,special chemical sensing systems (smell and taste) have developed that are distin-guished anatomically by the location of their receptors in the nasal and oral cav-ities, respectively. This chapter will focus on the nature of odors (sensations) and odor-ants (odorous molecules) that are relevant to human smell perception. The physiology


11

and biochemistry of olfaction will be addressed as well. Taste will also be reviewedbriefly.Odor sensations are induced by the interaction of odorants with specialized recep-

tors in the olfactory epithelium in the top of the nasal cavity. In air-breathing animalsincluding humans, odorants are volatile, hydrophobic compounds that have molecularweights of less than 300 daltons. The largest known odorant to date is labdane that hasa molecular weight of 296 [1]. Chemical reactivity has little to do with odor potentialsince odorant molecules are uncharged. Odorants vary widely in structure and includemany chemical classes including organic acids, alcohols, aldehydes, amides, amines,aromatics, esters, ethers, fixed gases, halogenated hydrocarbons, hydrocarbons, ke-tones, nitriles, other nitrogen-containing compounds, phenols, and sulfur-containingcompounds. The signals induced by the interaction of odorants with olfactory recep-tors (ORs) in the olfactory epithelium are transmitted to the olfactory bulb and ulti-mately to the brain (see Fig. 1.1 and Section 1.4).The sense of smell is a remarkably sensitive system that responds to very low con-

centrations of chemicals. It is estimated that only 2% of the volatile compounds avail-

Fig. 1.1 Cross-section of the skull, showing the location of the

olfactory epithelium, olfactory sensory neurons, cribriform plate,

olfactory bulb, and some central connections

1 Introduction to Olfaction: Perception, Anatomy, Physiology, and Molecular Biology2

able in a single sniff will reach the olfactory receptors, and as few as 40 molecules ofsome mercaptans are sufficient to perceive an odor [2, 3]. The exquisite sensitivity ofthe smell system is illustrated by the human detection thresholds given in Table 1.1(data from ref. [4]). It can be seen that these compounds can be detected at concentra-tions in the low parts-per-billion (ppb) and even low parts-per-trillion (ppt) range as inthe case of thiophenol, thiocresol, and propyl mercaptan. Over the course of a day,odorants have enormous opportunities to reach olfactory receptors during the processof inhalation and exhalation. An average person breathes 15 times per minute (or21 600 times per day) moving an average of 0.5 liters of air per breath (or 10 800 litersof air per day).Most odor sensations are produced by mixtures of hundreds of odorants rather than

by a single compound. Individual components tend to harmonize or blend together inmixtures leading to perceptual fusion. Humans have limited capacity to identify singleodorants in mixtures with three to four components being maximum [5]. This limita-tion in the ability to identify the individual components of mixtures appears to be aninherent property of olfaction since it is unrelated to the experience of the subjects orthe type of odorants.Odor sensations are characterized by general descriptors, such as sulfurous, fruity,

floral, and earthy, or by their source such as banana or orange. The range of distinctiveodor sensations is enormous, and a skilled perfume chemist can recognize and dis-tinguish 8000 to 10 000 different substances on the basis of their odor quality [6, 7] andeven respond to chemicals never before encountered in our environment. The olfac-tory system detects and discriminates among this immense number of odorant typesdue to the broad repertoire of olfactory receptor proteins that are encoded by a largeolfactory gene family [8–10] (see Section 1.5). Humans have several hundred distinctgenes that encode olfactory receptor proteins and rodents have upwards of 500 to 1000separate genes, that is, as much as 1% of the genome [9, 10]. This extremely broadrange of receptor types permits the detection of odor sources comprised of unpredict-able mixtures of molecular species, and even allows detection of newly synthesizedcompounds with no known function.

Table 1.1 Odor thresholds of representative sulfur compounds [44].

Compound Characteristic odor Odor Threshold

Allyl mercaptan Garlic-coffee 0.05 ppb

Amyl mercaptan Unpleasant strong 0.3 ppb

Benzyl mercaptan Unpleasant strong 0.19 ppb

Crotyl mercaptan Skunk-like 0.029 ppb

Dimethyl sulfide Decayed vegetables 0.1 ppb

Ethyl mercaptan Decayed cabbage 0.19 ppb

Hydrogen sulfide Rotten eggs 1.1 ppb

Methyl mercaptan Decayed cabbage 1.1 ppb

Propyl mercaptan Unpleasant 0.075 ppb

t-butyl mercaptan Skunk, unpleasant 0.08 ppb

Thiocresol Skunk, rancid 0.062 ppb

Thiophenol Putrid, garlic-like 0.062 ppb

1.1 Introduction to Olfaction 33

1.2

Odor Classification Schemes Based on Adjective Descriptors

Classification systems based on adjective descriptors have been used historically toorganize the many thousands of different odor sensations into a limited numberof groups. Table 1.2 shows some of the early schemes for classification of odor sensa-tions. Modern olfactory specialists including perfumers who work with fragrances,however, find the small number of categories of early classification schemes to beinadequate for describing the odors that they encounter in their work. Over thelast half century there has been a movement away from trying to classify odors ina few limited classes but rather to develop an extensive vocabulary that is relevantfor use with the odor stimuli being examined. Hence, modern odor classificationmethods are based on an extensive array of adjective descriptors selected for theirrelevance for specific applications.Modern descriptive classificationmethods can be general (e.g. for the broad range of

odors encountered in everyday life) or specific (e.g. relevant to particular applicationsin the food or fragrance industry). In the food industry, the odors of chemical com-pounds are often categorized by the identity of the edible material of which they arereminiscent. Sample odor classes for foods include caramel, honey, vanilla, citrus, andbutter. Fragrance odors are often classified by floral and herbal groupings, such asjasmine, rose, balsam, or pine. Table 1.3 presents a series of 146 adjective descriptorsdeveloped by the American Society for Testing and Materials [20] for general classi-fication of odors commonly encountered in everyday life. Table 1.4 gives a more spe-cific list of descriptors used by the fragrance industry [21]. Other odor descriptors canbe found in flavor and fragrance catalogs (Aldrich [22], for example) as well as ontechnical web sites (for example, ref. [23]).

Table 1.2 Descriptive categories proposed historically for smell

sensations.

Number of Categories Category Classification Reference

6 Sweet, acid (sour), harsh, rich/fat, astringent, fetid 11

7 Aromatic, fragrant, ambrosial (musk-like), alliaceous

(garlic-like), hircine (goat-like), foul, nauseating

12

9 Aromatic, ethereal, fragrant, ambrosial, empyreumatic

(burnt), alliaceous, hircine, repulsive, nauseous

13

6 Flowery, fruity, spicy, resinous, burnt, putrid 14

8 Flowery, fruity, herbaceous (green), animal/ambrosial/

human flesh aura, spicy/minty/camphoric, earthy/

fungoid, woody/balsamic/nut-like, Disagreeable: acrid/

phenolic/burnt/nauseating

15

7 Ethereal, floral, pepperminty, camphoraceous, musky,

pungent, putrid

16–18

9 Etherish, fragrant, sweet, spicy, oily, burnt, sulfurous,

rancid, metallic

19


Table 1.3 ASTM descriptive categories used for general odor quality

characterizations.a)

(01) Fragrant (50) Vanilla-like (99) Alcohol-like

(02) Sweaty (51) Fecal (like manure) (100) Dill-like

(03) Almond-like (52) Floral (101) Chemical

(04) Burnt, smoky (53) Yeasty (102) Creosote

(05) Herbal, green, cut grass (54) Cheesy (103) Green pepper

(06) Etherish, anesthetic (55) Honey-like (104) Household gas

(07) Sour, acid, vinegar (56) Anise (licorice) (105) Peanut butter

(08) Like blood, raw meat (57) Turpentine (pine oil) (106) Violets

(09) Dry, powdery (58) Fresh green vegetables (107) Tea-leaves-like

(10) Like ammonia (59) Medicinal (108) Strawberry-like

(11) Disinfectant, carbolic (60) Orange (fruit) (109) Stale

(12) Aromatic (61) Buttery (110) Cork-like

(13) Meaty (cooked) (62) Like burnt paper (111) Lavender

(14) Sickening (63) Cologne (112) Cat-urine-like

(15) Mushy, earthy, moldy (64) Caraway (113) Pineapple (fruit)

(16) Sharp, pungent, acid (65) Bark-like, birch bark (114) Fresh tobacco smoke

(17) Camphor-like (66) Rose-like (115) Nutty

(18) Light (67) Celery (116) Fried fat

(19) Heavy (68) Burnt candle (117) Wet paper-like

(20) Cool, cooling (69) Mushroom-like (118) Coffee-like

(21) Warm (70) Wet wool, wet dog (119) Peach (fruit)

(22) Metallic (71) Chalky (120) Laurel leaves

(23) Perfumery (72) Leather-like (121) Scorched milk

(24) Malty (73) Pear (fruit) (122) Sewer odor

(25) Cinnamon (74) Stale tobacco smoke (123) Sooty

(26) Popcorn (75) Raw cucumber-like (124) Crushed weeds

(27) Incense (76) Raw potato-like (125) Rubbery (new rubber)

(28) Cantalope, honey dew melon (77) Mouse-like (126) Bakery, fresh bread

(29) Tar-like (78) Black pepper-like (127) Oak wood, cognac-like

(30) Eucalyptus (79) Bean-like (128) Grapefruit

(31) Oily, fatty (80) Banana-like (129) Grape-juice-like

(32) Like mothballs (81) Burnt rubber-like (130) Eggy (fresh eggs)

(33) Like gasoline, solvent (82) Geranium leaves (131) Bitter

(34) Cooked vegetables (83) Urine-like (132) Cadaverous, dead animal

(35) Sweet (84) Beery (beer-like) (133) Maple (syrup)

(36) Fishy (85) Cedar wood-like (134) Seasoning (for meat)

(37) Spicy (86) Coconut-like (135) Apple (fruit)

(38) Paint-like (87) Rope-like (136) Soupy

(39) Rancid (88) Seminal, sperm-like (137) Grainy (as grain)

(40) Minty, peppermint (89) Like cleaning fluid (138) Clove-like

(41) Sulphidic (90) Cardboard-like (139) Raisins

(42) Fruity (citrus) (91) Lemon (fruit) (140) Hay

(43) Fruity (other) (92) Dirty linen-like (141) Kerosene

(44) Putrid, foul, decayed (93) Kippery (smoked fish) (142) Nail polish remover

(45) Woody, resinous (94) Caramel (143) Fermented fruit

(46) Musk-like (95) Sauerkraut-like (144) Cherry (berry)

(47) Soapy (96) Crushed grass (145) Varnish

(48) Garlic, onion (97) Chocolate (146) Sour milk

(49) Animal (98) Molasses

a) Odor Quality characterizations. Each sample israted on 146 adjectives using a five-point scale

where 0 indicates no odor and 5 indicates extre-mely strong odor.

1.2 Odor Classification Schemes Based on Adjective Descriptors 55

Table 1.4 Odor descriptors in Allured’s Perfumer’s Compendium [21]

used by perfumers and flavorists.

(1) Agrumen (50) Gardenia (99) Ozone (fresh air, marine)

(2) Aldehydic (51) Geranium (100) Patchouli

(3) Almond (52) Ginger (101) Peach

(4) Amber (53) Grape (102) Pear

(5) Ambergris (54) Grapefruit (103) Pepper

(6) Animal (55) Grass (104) Peppermint

(7) Anisic (56) Green (105) Petal

(8) Apple Blossom (57) Hawthorne (106) Petitgrain

(9) Apple Fruity (58) Hay (107) Pimento

(10) Armoise (59) Herbal (108) Pine

(11) Balsamic (60) Honey (109) Pineapple

(12) Banana (61) Honeysuckle (110) Plum

(13) Basil (62) Hyacinth (111) Powdery

(14) Bay (63) Incense (112) Raspberry

(15) Bergamot (64) Jasmin (113) Rooty

(16) Camphoraceous (65) Juicy (114) Rose

(17) Cardamom (66) Juniper (115) Sage

(18) Carnation (67) Kiwi (116) Sandalwood

(19) Cassie (68) Labdanum (117) Sappy-green wood

(20) Cassis (69) Lactonic (118) Smokey

(21) Castoreum (70) Lavender (119) Spicy

(22) Cedar (71) Leafy (120) Strawberry

(23) Celery (72) Leather (121) Styrax

(24) Chamomile (73) Lemon (122) Sweet

(25) Cherry (74) Lilac (123) Sweet pea

(26) Chocolate (75) Lime (124) Tagette

(27) Chrysanthemum (76) Mandarin (125) Tangerine

(28) Cinnamon (77) Medicated (126) Tea

(29) Citrus (78) Melon (127) Thyme

(30) Civet (79) Metallic (128) Tobacco

(31) Clary sage (80) Mimosa (129) Tolu

(32) Clove (81) Minty (130) Tonka

(33) Coconut (82) Moss (131) Tuberose

(34) Cognac (83) Muguet (132) Vanilla

(35) Coriander (84) Mushroom (133) Verbena

(36) Costus (85) Musk (134) Vetivert

(37) Cumin (86) Myrrh (135) Violet

(38) Dry (87) Narcisse (136) Waxy

(39) Earthy (88) Nasturtium (137) Wintergreen

(40) Eucalyptus (89) Neroli (138) Woody

(41) Fatty (90) Nutmeg (139) Ylang

(42) Fecal-animal (91) Nutty (140) Zesty, peely (citrus)

(43) Fig (92) Oily

(44) Floral (93) Olibanum

(45) Floral bouquet (94) Opoponax

(46) Fougere (95) Orange flower

(47) Freesia (96) Orange fruit

(48) Fruity (97) Oriental

(49) Galbanum (98) Orris


Classification schemes for odor quality are beset, however, by a variety of limitations.First there are inherent interindividual differences in the emotional and hedonic prop-erties of odors. Labels such as pleasant, delightful, disgusting, and revolting are com-mon associations with odors, and these subjective evaluations can influence the choiceof descriptors of odor quality. Emotional responses to odors probably derive from thefact that olfaction is a primal sense that is used in the animal kingdom to identify food,mates, predators, and warnings of danger. Second, there are individual differences inthe actual perception of odor based on genetic differences [24–26]. Third, there areindividual differences in the use of odor descriptors even among trained panelists.Fourth, the vocabulary of most languages lacks words that describe the full rangeof odor sensations. For this last reason, measures of similarity rather than adjectivedescriptors have been used to quantify odor quality by arranging odor sensations inmultidimensional spaces (to be described in the next section).

1.3

Odor Classification Based on Chemical Properties

Although much progress has been made in our knowledge of olfactory physiology andbiochemistry, the fundamental relationship between odor quality and molecular prop-erties is still poorly understood. Even slight alterations in the chemical structure of anodorant can induce profound changes in odor quality. Current structure-activity mod-els in olfaction are, for the most part, simply collections of disparate facts with nounifying theme; furthermore, they have inadequate predictive accuracy [27]. As a con-sequence, the basic logic necessary to develop a comprehensive odor classificationscheme based on particular features of molecules remains elusive.There are several reasons for the lack of progress in classifying odors on the basis of

chemical properties. First, it is not yet possible to model odorant-receptor interactionsbecause the three-dimensional (3D) protein structures of the receptor sites are notknown. Second, unlike structure-activity counterparts in pharmacology, there arevast numbers of agonist types (thousands of odorant structures) as well as thousandsof different odor sensations. Third, identical molecules may activate different receptortypes depending on the orientation of the molecule at the receptor. Beets [6] empha-sized that identical molecules arrive near receptor sites at different orientations andwith different conformations. Thus, a given odorant would be expected to interact witha variety of receptor types, and odor quality must be encoded by a pattern of informa-tion frommultiple receptors (rather than activation of a single receptor type). A fourthproblem is that there are no standard methods for quantifying odor quality for use instructure-activity studies.


1.3.1

History of Structure-activity Studies of Olfaction

In spite of these limitations, a variety of attempts have been made to relate odors tophysicochemical parameters. Amoore [16] proposed that the shape and size of a mo-lecule are the physicochemical parameters that determine odor quality, i.e. odorants fitinto receptor sites in a ‘lock and key’ fashion with molecules of similar size and shapeexpected to have similar odor quality. In support of this theory, Amoore and Venstrom[28] reported significant correlations between odor quality of 107 odorants and a hand-calculated index of molecular size and shape for five classes of odors (ethereal, cam-phoraceous, musky, floral, and minty). Amoore [29] also reported a correlation of 0.90between odor quality and a computer-generated molecular shape index when 25 sub-stances were compared with benzaldehyde (almond odor).Wright [30] challenged Amoore’s results, indicating that it is inappropriate to repre-

sent a complex 3Dmolecular shape by an index consisting of a single number becausemany different 3D profiles could share the same molecular shape index. Wright [31]suggested that the mechanism for stimulation of olfactory receptors is low-energymolecular vibrations, and that molecules with similar vibrational frequency patternsshould have similar odor quality. Wright and Robson [32] supported their hypothesiswith the finding of similarity between the pattern of frequencies in the far infraredspectra for odorants with a bitter almond odor.Dravnieks and Laffort [33] suggested that four factors related to intermolecular

interaction forces (an apolar factor, a proton receptor factor, an electron factor, and aproton donor factor) could predict both quantitative and qualitative odor discrimina-tion in human beings. In spite of many attempts in addition to those just described, nogeneral structure-activity model or theory has yet been proposed that accurately pre-dicts odor quality of molecules a priori from physicochemical parameters[1, 6, 27, 34, 35].

1.3.2

Odor Structures Associated with Specific Odor Classes Based on Qualitative Descriptors

Figures 1.2 to 1.6 provide examples of chemical structures for compounds classified byexperienced odor specialists as having musk, ambergris, muguet, green, and bitteralmond odors. Each figure gives the structure of representative chemicals withineach specific odor quality. These figures illustrate that compounds with widely vary-ing chemical structures can have similar odor qualities. Musk is an odor category thatis used in fragrance with its original source being the glandular secretions of the malemusk deer. Molecules with this odor quality are very diverse in structure as shown inFig. 1.2; they include steroidal, linear, macrocyclic, nitro, as well as bi- and tricycliccompounds. Ambergris is an odor quality used in fragrance that originally derivedfrom the sperm or cachalot whale. Muguet is a lily-of-valley odor. Green is theodor of natural green vegetable products. Bitter almond is an odor quality of an es-sential oil obtained by stem distillation of kernals from bitter almond (P. amygda-lus). The types ofmolecules within each odor quality can vary considerably in structure.


Studies of enantiomers have also been used to gain insight into the relationshipbetween physicochemical properties and odor quality. These studies reveal that enan-tiomers of chiral odorants may or may not show differences in odor quality[1, 6, 27, 34, 35]. There are significant differences in the two enantiomers of carvonewith R-(�)-carvone having an odor of spearmint oil and S-(þ)-carvone having an odorof caraway oil (see Fig. 1.7). Significant differences in the odor quality of enantiomersof nootkatone have also been reported. However, enantiomers of 2-octanol and carbi-naol were not found to differ in odor quality.Overall, Figs. 1.2–1.7 demonstrate that structurally unrelated chemicals can yield

similar odor qualities. Furthermore, differences in the odor quality of certain enan-tiomers indicate that very subtle differences in structure are capable of producing verydifferent and distinct odors. A better understanding of the physicochemical para-

Fig. 1.2 Compounds with musk odor: a) androst-16-en-3b-ol,b) ethyl citronellyl oxalate, c) cyclopentadecanolide, d) musk ketone,

e) Traesolide�, f) Galaxolide


Fig. 1.3 Compounds with ambergris odor:

a) oxalactone, b) cyclohexyltetrahydrofuran,

c) Karanal, d) timberol, e) cedramber

Fig. 1.4 Compounds with muguet odor: a) lilial, b) mugetanol

Fig. 1.5 Compounds with green odor: a) cis-3-he-

xen-1-ol (leaf alcohol), b) Ligustral�, c) nonadienal


meters responsible for specific odor qualities requires more knowledge about the 3Dstructure of ORs.

1.3.3

Relationship of Physicochemical Parameters to Classifications of Odor Based on SimilarityMeasures

Themethodology of multidimensional scaling has also been used to better understandthe relationship between odor quality and physicochemical variables [36, 37]. Multi-dimensional scaling (MDS) procedures represent odor sensations in spatial maps.The input for multidimensional scaling procedures consists of quantitative measuresof similarity between pairs of odors. For example, if two odors are judged by humansubjects to have similar odor quality, they will be positioned near each other in themultidimensional quality space. Stimuli judged to be dissimilar are located distantfrom one another. Two examples of studies that relate physicochemical propertiesto odor quality as defined by multidimensional maps are given below. The mathema-

Fig. 1.6 Compounds with bitter almond odor: a) benzaldehyde,

b) hydrogen cyanide

Fig. 1.7 Enantiomers of carvone. a) R-(�) carvone which has a

spearmint-like odor, b) S-(þ) carvone which has a caraway-like odor


tical procedures used to relate the physicochemical properties to themultidimensionalmaps are given in the Appendix.

1.3.3.1 Study 1: Broad Range of Unrelated Odorants

A group of 50 odorants (5 of which were duplications) that ranged widely in quality andstructure were arranged in a two-dimensional (2D) space by MDS on the basis of odorsimilarity [38]. The 2D space that is shown in Fig. 1.8 accounts for 91% of the humansimilarity data. The odor stimuli were roughly positioned by MDS into two groups; thelarger subset on the left is affectively more pleasant than the one on the right.Because the spatial arrangement could not be accounted for by a single physico-

chemical variable (such as chemical structure, molecular weight, number of doublebonds, or dipole moment), a series of physicochemical variables were weighted in anattempt to regenerate the space. A mathematical technique generated weights for aseries of physicochemical variables such that the distances and thus the spatial ar-rangements among the stimuli in Fig. 1.8 were regenerated.The mathematical procedure used to maximize the configurational similarity of the

psychologically determined space in Fig. 1.8, with a space generated by weighted phy-sicochemical parameters was based on a least-squares method (see Appendix). Thephysicochemical parameters that were weighted to reconstruct the 2D space inFig. 1.8 as well as the means for these physicochemical variables are shown in Ta-ble 1.5. Functional groups were coded according to their number in a particular mo-lecule; for example, benzaldehyde has one aldehyde group and the mean number of

Fig. 1.8 Two-dimensional solution for a broad range of odor stimuli.

Compounds with similar odor qualities are located near each other in

space. The more pleasant stimuli are located in the subset to the left,

and the more unpleasant stimuli are in the subset on the right

(modified after Schiffman [38])


aldehyde groups for all themolecules in the space in Fig. 1.8 is 0.10. Cyclic compoundswere coded ‘1’ and noncyclic compounds ‘0.’ Raman spectra from 100 to 1000 cm�1

were included because they contain much information that could be correlated withthe pleasantness or unpleasantness of the molecules (i.e. that they fell to the left orright in the space). A large weight will expand the difference between these two stimulimore than a small weight, such that physicochemical variables with large weights areof greater importance in discrimination among the odor stimuli.Although thismethodology was successful in relating strict quantitative measures of

olfactory quality with quantitative physicochemical measures (i.e. 84% of the variancewas accounted for), the number of physicochemical variables needed to account forodor quality were too large to be of practical value. That is, the success in correlatingphysicochemical properties to odor quality did not improve the ability to predict ordesign of molecules with specific odor qualities.

Table 1.5 Weights that were applied to standard scores for physico-

chemical variables to regenerate the space in Fig 1.8. Functional groups

were coded by their number in a molecule, thus, benzaldehyde was

coded ‘1’. Cyclic compounds were coded ‘1’ while noncyclic com-

pounds were coded ‘0.’

Physicochemical variable Weight Mean

Molecular weight 6.24 116.57

Number of double bonds 0.51 0.74

Phenol 2.33 0.13

Aldehyde 3.21 0.10

Ester 0.24 0.05

Alcohol 2.54 0.26

Carboxylic acid 5.50 0.13

Sulfur 3.44 0.08

Nitrogen 3.15 0.08

Benzene � 0.14 0.33

Halogen � 0.34 0.03

Ketone � 0.19 0.03

Cyclic 4.56 0.31

Mean Raman intensity

Below 175 cm�1 0.01 0.51

176–250 cm�1 3.57 2.36

251–325 cm�1 � 0.75 1.65

326–400 cm�1 3.81 1.56

401–475 cm�1 1.65 2.10

476–550 cm�1 � 3.63 1.54

551–625 cm�1 � 0.69 2.07

626–700 cm�1 � 1.16 1.07

701–775 cm�1 0.07 2.36

776–850 cm�1 3.04 4.36

851–925 cm�1 0.24 3.44

926–1000 cm�1 0.36 2.06


1.3.3.2 Study 2: Pyrazines

Multiple physicochemical parameters were also necessary to account for an odor spacerepresenting the similarity among related compounds (pyrazines). The pyrazines wereordered in the 3D space in Fig. 1.9 on the basis of similarity of odor quality [39]. Next, aset of descriptors (see Table 1.6) was generated by the automated data analysis andpattern recognition toolkit (ADAPT), a computer system for automated data analysesby pattern recognition techniques [40, 41]. The substructures used to generate theenvironment descriptors are given in Fig. 1.10.Canonical regression, a common statistical technique [42], was used to relate the

descriptors in Table 1.6 to the 3D arrangement in Fig. 1.9. Canonical analysis extendsmultiple regression analysis from one criterion variable to a set of criterion variables.For simple multiple regression, the relationship of a set of predictors to a single cri-terion variable is analyzed. In the current application, canonical regression was used todetermine the relationships between two sets of variables, that is, the stimulus coor-

Table 1.6 Descriptors generated by ADAPT [40] for analysis of

pyrazines.

1 Number of atoms except hydrogen

2 Number of carbon atoms

3 Number of oxygen atoms

4 Number of bonds

5 Number of single bonds

6 Number of double bonds

7 Molecular weight

8 Path 1 molecular connectivity for all bonds in the structure

9 Path 1 molecular connectivity corrected for rings

10 Path 1molecular connectivity calculated using the valences of heteroatoms and corrected for rings

11 Path 2 molecular connectivity



14 Molecular volume

15 Number of substructure 1 (see Fig. 1.11)

16 Environment-substructure 1 (calculates connectivity for substructure 1 and nearest neighbors)

17 Number of substructure 2

18 Environment-substructure 2














Fig. 1.9a and 1.9b Two-dimensional cross-sections through the three-dimensional space for pyrazines

[39]. Duplicate samples of the same stimulus are represented by two datapoints.


dinates of the 3D MDS space and the physicochemical parameters in Table 1.6. Theequations for canonical correlation used are given in the Appendix.Small subsets of the physicochemical parameters were used in the tests because in

canonical correlation analysis, the number of stimuli should be greater than the num-ber of dimensions and physicochemical parameters combined. The analysis revealedthat a linear combination of two ADAPT parameters in Table 1.6 (number of oxygenatoms and chemical environment of substructure 7) in addition to a concentrationvariable accounted for 63% of the arrangement of the pyrazine odor space inFig. 1.9. This study, along with Study 1, again illustrates the difficulty in relating quan-titative physicochemical parameters with odor quality.

1.3.4

Molecular Parameters and Odor Thresholds

In addition to odor quality, attempts have been made to determine the relationshipbetween odor thresholds (or suprathreshold intensity) and molecular parameters.Variables that have been related to thresholds and intensities include molecularweight, cross-sectional area, adsorption constants at an oil-water interface, hydropho-bicity, molar volume, pKa, saturated vapor pressure, polarizability, hydrogen bondingability, air/water partition coefficients, log P (octanol-water partition coefficient), para-meters derived from gas chromatograpy, Taft polar constants, and various steric para-meters [34]). Like structure-activity studies of odor quality, there appear to be no rulesthat can be generalized for the entire range of odorous compounds.

1.3.5

Conclusions Regarding Physicochemical Parameters and Odor Quality

Although it is possible to develop techniques that weight a series of parameters topredict odor quality, this is of little practical use in understanding the physiological

Fig. 1.10 The substructures utilized by ADAPT for generating

environment descriptors for analysis of pyrazines


basis of odor quality. A more complete understanding of structure-activity relation-ships in olfaction will occur when the molecular structure of the odorant receptor(including the stereoelectronic arrangements of binding sites) is brought into theequation along with the structure of the odorant.

1.4

Physiology and Anatomy of Olfaction

1.4.1

Basic Anatomy

The functional organization of the olfactory system is similar to other sensory systems(e.g. vision) but, in this case, the sensory input is provided bymolecules (i.e. odorants).Odorants are recognized by specific receptor proteins situated on the ciliary mem-branes of olfactory sensory neurons located in the olfactory epithelium at the topof the nasal cavity (see Fig. 1.1). The olfactory epithelium is comprised of three celltypes as shown in Fig. 1.1: the bipolar olfactory sensory neurons (primary sensoryneurons) with dendritic cilia projecting from their terminal ends in a thin mucuslayer (10–100 lM thick); supporting or sustentacular cells (a type of glial cell) thatterminate in microvilli; and basal cells (like stem cells) which make new olfactoryreceptor cells.The olfactory epithelium is a thin tissue in the nasal cavity that is easily distinguish-

able bilaterally in rats and dogs due to its yellowish color. In humans, however, the twosmall patches (about 2 square inches or 6.5 square centimeters) are more difficult tovisualize because their pinkish hue blends with the respiratory epithelium that lines

Fig. 1.11 Olfactory epithelium showing three cell

types: olfactory sensory neurons (also called receptor

cells), supporting (or sustenacular cells), and basal

cells


the rest of the nasal cavity. Odorants can reach the olfactory receptors via orthonasaltransport through the nares (e.g. when sniffing) or via retronasal transport from theoral cavity (e.g. when chewing food). In orthonasal transport, the turbinates (bones inthe nose) create turbulent airflow patterns that direct volatile compounds to the olfac-tory receptor cells in the top of the nasal cavity. Inhaled odorants have been shown toreach the blood and brain after three hours of exposure [43], and as a consequence,olfactory receptors can also respond to blood-borne odorants [44].There are vast numbers of olfactory sensory neurons with estimates between 106 and

108 in man [45, 46]. These olfactory neurons turn over continuously with an averagetime for replacement of approximately 30 days. This neurogenesis is active throughoutthe lifespan, and arises from basal cells deep in the epithelium [47]. Olfactory sensoryneurons consist structurally of a soma (cell body), a peripheral dendritic knob withfine, long cilia that project into the watery mucus that protects the nasal cavity,and an unmyelinated axon that projects centrally from the soma and propagates actionpotentials to the olfactory bulb. Specific receptor subtypes are expressed in subsets ofolfactory sensory neurons spatially distributed in distinct zones of the olfactory epithe-lium, and only one odorant receptor type is expressed on the vast majority of individualolfactory sensory neurons [9, 10]. Yet, single olfactory cells respond to a range of com-pounds with a variety of olfactory qualities because individual olfactory receptors haverelatively broad molecular receptive ranges [48].Axons of the bipolar olfactory sensory neurons fasiculate together and coarse

through tiny holes in the cribriform plate of the ethmoid bone to the olfactorybulb where they make their first synapses with second-order neurons in intricate sphe-rical masses of neuropil called glomeruli (see Fig. 1.12). The axons of the bipolar cellsconstitute the fibers of the olfactory nerve. The neuropil of glomeruli consists of theaxons of incoming olfactory sensory neurons and the dendrites of the mitral cells onwhich they synapse. Olfactory sensory neurons that express a specific odor receptortype converge upon a common glomerulus in the olfactory bulb [9, 10, 49]. In humans,axons from thousands of olfactory sensory neurons expressing a single odorant recep-tor type are thought to converge onto two or three glomeruli in the olfactory bulb, witheach glomerulus receiving input from a single type of olfactory receptor. Local neu-ronal circuits in the bulb provide the first tier of central processing of odors with ol-factory signals sharpened via lateral inhibition among glomerular modules [50]. As aresult of this neural processing, mitral cells have narrower molecular receptive rangesthan olfactory receptor neurons [48]. Because individual olfactory sensory neurons canrespond to multiple odorants, it follows that the pattern across multiple glomeruliprovides the basis for discrimination of olfactory quality. The distinct spatial patternsof glomerular activation by specific odorants can be visualized using optical imagingtechniques [51, 52].Olfactory information from the olfactory bulb is next transmitted by the olfactory

tract to the anterior olfactory nucleus, the olfactory tubercle, the prepyriform cor-tex, and the amygdala, and ultimately to higher brain centers that process the olfactorysignals. The prepyriform cortex and the amygdala are brain structures that are part ofthe limbic system, which processes emotions and memories in addition to olfactorysignals. Olfactory information is ultimately transmitted to the hypothalamus (which


mediates food intake) and to the neocortex. Non-invasive imaging techniques such aselectroencephalography, positron emission tomography, and functional magnetic re-sonance spectrometry have found that the degree of activation of the pyriform cortex,orbitofrontal areas, and parts of the parietal and temporal cortices is dependent on theodor quality and pleasantness of the stimuli (for example see refs. [53] and [54]). Age-related losses occur in the olfactory epithelium, olfactory bulb and nerves, hippocam-pus and amygdaloid complex, and hypothalamus, and these changes parallel percep-tual losses in the olfactory system during the aging process.At elevated concentrations, odorants can also stimulate free nerve endings of the

trigeminal nerve in the nose. Trigeminal stimulation by odorous chemicals inducessensations such as irritation, tickling, burning, stinging, scratching, prickling, anditching [55, 56]. Sensory information transmitted by the trigeminal nerve is not con-sidered an ‘odor’ because the trigeminal nerve is not directly stimulated by electricalsignals from olfactory receptor neurons; rather trigeminal stimulation involves a dif-ferent sense called chemesthesis which is related to nociception (e.g. pain).

Fig. 1.12 Cross-section of the

olfactory bulb. A.C. indicates

anterior commissure.


1.4.2

Transduction and Adaptation of Olfactory Signals

Odorants first traverse the aqueous interphase that lines the surface of the olfactoryepithelium in order to interact with the olfactory receptors in the ciliary membranes.This process is facilitated by soluble odorant binding proteins that ‘shuttle’ the hydro-phobic odorants through the aqueous mucus layer towards specific odorant receptors.Odorant receptors are members of a superfamily of up to 1000 different G-protein-coupled receptors that possess seven transmembrane (7TM) domains. The location ofodorant binding is thought to be a hydrophobic pocket in transmembrane regions 3, 4,and 5 of the seven-membrane-spanning receptor. Olfactory signaling is initiated by theinteraction of an odorant molecule with a protein receptor on the ciliary surface. Thisligand binding triggersmultistep intracellular reaction cascades that open an ion chan-nel in the cell membrane leading to depolarization by a few tens of millivolts [8, 57].Figure 1.13 illustrates the binding of an odorant to a G-protein-coupled receptor inolfactory neurons that results in activation of cAMP. Odorant binding involves a sig-naling pathway that includes a Gs-like protein (Golf) that activates a specific adenylylcyclase leading to generation of cyclic AMP (cAMP). cAMP binds directly to a cyclicnucleotide-gated (CNG) ion channel in the cell membrane that increases the probabil-ity of positive ions flowing into the cell, leading to depolarization and action potentials.

Figure 1.13 a. and b. In most olfactory neurons, an odorant binds to

an odorant receptor (OR) leading to an exchange of GTP (guanosine

triphosphate) for GDP (guanosine diphosphate) on the heterotrimeric

G-protein (Golf ). c. The a subunit of Golf activates adenylyl cyclase

leading to generation of cAMP. d. Cyclic AMP binds directly to a cyclic

nucleotide-gated (CNG) ion channel in the cell membrane that in-

creases the probability of positive ions flowing into the cell. This causes

depolarization of the cell membrane and transmission of a signal along

the axon to the bulb. CNG channels are nonselective and permeable to

cations including Naþ and Ca2þ


Another intracellular second messenger, inositol triphosphate may also mediatechanges in the conductance in some olfactory neurons, leading to depolarizationof olfactory cells in response to odorant-receptor binding. Olfactory signaling is ter-minated when receptors are phosphorylated via a negative feedback reaction catalyzedby two types of kinases [57].The large family of G-protein-coupled 7TM receptors just described may not be the

only odorant receptors. An alternate signaling pathway for olfactory transduction hasrecently been proposed by Gibson and Garbers [58]. They have found a large family ofolfactory neuron-specific guanylyl cyclases that are membrane-bound and contain ex-tracellular domains that may constitute a second family of odorant receptors. Activa-tion of guanylyl cyclase elevates cyclic GMP (cGMP) that converges on the same CNGchannels as cAMP to generate action potentials.Repeated stimulation of olfactory receptor neurons leads to decrements in the neur-

al responses, i.e. adaptation. Three forms of olfactory adaptation can take place inolfactory receptor neurons: two rapid forms and one persistent form. These threedifferent adaptation phenomena are controlled, at least in part, by separate molecularmechanisms. These mechanisms involve Ca2þ entry through CNG channels, Ca2þ-dependent CNG channel modulation, Ca2þ/calmodulin kinase II-dependent attenua-tion of adenylyl cyclase, and the activity of the carbon monoxide/cyclic GMP secondmessenger system [59].

1.5

Molecular Biology Of Olfaction

The molecular era of olfaction began in 1991 with the discovery by Buck and Axel of amultigene family of G-protein-coupled ORs with a 7TM-spanning typology. Buck andAxel [8] obtained complementary DNA (cDNA) utilizing olfactory epithelial RNA fromrat in conjunction with an amplification process called the polymerase chain reaction(PCR). (Complementary DNA is a copy of a messenger RNA). They found a PCRproduct (PCR 13) that contained multiple species of DNA that are representativeof a multiple gene family that encodes transmembrane domain proteins that are re-stricted to the olfactory epithelium. Further work has shown that there is a conserva-tion of certain amino acidmotifs within OR gene sequences that distinguish ORs fromother 7TM proteins [8, 60, 61]. There are also hypervariable regions within certainmembrane regions of ORs (i.e. TMs 3, 4, and 5) that provide a diversity of ligand-binding pockets [61]. A single amino acid substitution in the hypervariable regioncan change ligand-binding specificity [62]. This diversity in ligand-binding domainsis necessary to accommodate the enormous number of structurally diverse volatilechemicals that can activate the olfactory sensory neurons.Early estimates suggested that there are approximately 500 to 750 genes that encode

ORs in humans with an estimated 1000 genes in mouse and rat [7, 9, 10, 63]. However,there appears to be a high frequency of pseudogenes (genes with defects that are in-compatible with receptor function) in the human but not rat OR repertoire; between38% to 76% of the human sequences do not encode full-length polypeptides

1.5 Molecular Biology Of Olfaction 2121

[9, 10, 64, 65]. One recent report claims to have identified and physically cloned 347human OR genes that they believe represent the complete repertoire of functionalhuman ORs [66]. This reduction in the functional fraction of olfactory receptor genesin humans compared to rats implies that olfaction may have become less importantduring the course of evolution, perhaps due to relaxed selective constraints [65]. Thedecrease in viable odorant genesmay also be the cause of specific anosmias in humans(inabilities to smell a particular odorant).OR genes are typically organized in clusters of ten ormore and are distributed across

numerous chromosomes [9, 10, 66–68]. The 347 olfactory genes identified by Zozulyaet al. [66] were located on all human chromosomes, except for 2, 4, 18, 20, 21, and Y,with themajority (155 ORs) on chromosome 11 followed in frequency by chromosome1 (42ORs), 9 (26ORs) and 6 (24ORs). The average human OR is approximately 315amino acids long. In general, only one OR gene is expressed in a single olfactorysensory neuron [69], and olfactory sensory neurons (OSNs) that express a singleOR converge on the same glomerulus in the olfactory bulb. Thus for the adult mousewhich has � 1000OR types and � 1800 glomeruli [70], each OR may be associatedwith only two specific glomeruli. However, it should be noted that one recent studyreported that there may be a subset of OSNs that expresses two distinct receptor types[71].Knowledge of the physiological functioning of specific ORs is still in its infancy.

That is, we know very little about the range of ligands that interact with each ofthe particular odorant receptors. This is due in part to the large number of odorantreceptors and the enormous repertoire (many thousands) of odorous compounds.Experimental approaches in which ORs are functionally expressed in olfactory sen-sory neurons are necessary to determine the tuning of a specific OR. Functional ex-pression of a specific ORs is achieved experimentally when a given receptor type isinserted into the plasma membrane, couples with the second messenger system,and produces a measurable response to an odorant ligand.Direct functional proof that the 7TM receptors cloned by Buck and Axel [8] were

actually odor receptors was obtained by Zhao et al. [72] who inserted a gene discoveredby Buck and Axel into the rat olfactory system, producing electrical activity in olfactoryneurons to specific odorant chemicals. Zhao et al. functionally expressed an OR inolfactory sensory neurons of rat in vivo using an adenovirus-mediated gene transferof a cloned OR, I7 (see ref. 8 for nomenclature). They inserted the I7 genes into anadenovirus vector linked to a gene for green fluorescent protein (GFP) that is used tomark genetically altered cells. (Disabled adenovirus vectors are used as a tool to trans-fer genes into mammalian cells. A viral gene can be replaced with another gene thatencodes an OR protein.) Cells that carried the rat I7 gene also carried the GFP gene,and thus could be visualized because they glowed bright green when exposed to bluelight. Extracellular transepithelial potential recordings from summed activity of manyolfactory neurons (called an electro-olfactogram) in the infected epithelium were ele-vated to heptaldehyde (C7), octyl aldehyde (C8), nonyl aldehyde (C9), and decyl aldehyde(C10) when compared with uninfected epithelium [72]. However, electro-olfactogramamplitudes were not elevated for hexaldehyde (C6) or undecylic aldehyde (C11). Thesefindings suggested that the response profile of the 17 receptor is relatively specific for


C7 to C10 saturated aliphatic aldehydes at least within the limited set of 74 odorants thatwas tested. Heptaldehyde (C7), octyl aldehyde (C8), nonyl aldehyde (C9), and decyl al-dehyde (C10) can be differentiated on the basis of odor quality so that a single receptortype does not code for a specific odor quality.Malnic et al. [69] used a combination of calcium imaging and single-cell RT-PCR

(PCR with reverse transcription) to identify ORs for odorants with related structuresbut varied odors. Their results indicate that one OR recognizes multiple odorants, oneodorant is recognized by multiple ORs, but that different odorants are recognized bydifferent combinations of ORs. They concluded that the olfactory system uses a com-binatorial receptor coding scheme to encode odor identities.

1.6

Taste

A brief overview of taste will also be given here because some of the sensors describedin this book are ‘taste sensors.’

1.6.1

Taste Classification Schemes Based on Sensory Properties

Historically, the taste literature often suggests that there are only four (or possibly five)basic taste qualities (sweet, sour, salty, and bitter, and possibly ‘umami’ which is thetaste of glutamate salts). All other tastes have been presumed to be combinations ofthese basic tastes. However, data are now accumulating that the range of taste sensa-tions is much broader and includes qualities such as astringency, metallic, fatty, andcalcium-like (e.g. chalky) [73–78].

1.6.2

Physiology and Anatomy of Taste

The receptor cells for taste are neuroepithelial cells that are clustered into buds anddistributed on the dorsal surface of the tongue, tongue cheek margin, base of thetongue near ducts of the sublingual glands, the soft palate, pharynx, larynx, epiglot-tis, uvula, and first third of the esophagus (see Schiffman and Warwick [79] for anoverview of anatomy). Taste sensations are induced by the interaction of chemicals(e.g. from food) with taste-buds during ingestion, chewing, and swallowing. Indivi-dual taste cells generally respond to more than one type of taste. Taste buds consistof approximately 50–100 cells that arranged in an onion-like structure (see Fig. 1.14).Individual cells in a taste-bud undergo continuous renewal every 10 to 10 1/2 days.Taste-buds on the tongue are positioned on specialized epithelial projections termedpapillae. There are three different kinds of lingual papillae that contain taste-buds:fungiform papillae (which are shaped somewhat like mushrooms), foliate papillae

1.6 Taste 2323

(which consist of linear depressions or vertical folds), and circumvallate papillae(which are surrounded by deep moats). The entire tongue is sensitive to all taste qua-lities but there are regional differences in sensitivity; for example, buds on fungiformpapillae are more sensitive to sodium salts, foliate papillae to acids, and circumvallateto bitter compounds.

Fig. 1.14 Taste bud

Fig. 1.15 Anatomy of taste showing the cranial nerves and nucleus of the solitary tract


Taste bud cells form direct neural connections called synapses with three cranialnerves: the facial nerve (VII), glossopharyngeal nerve (IX), and vagus nerve (X). Thesethree cranial nerves relay signals from taste receptor cells to the rostral portion of thenucleus of the solitary tract located in the medulla in the brain stem (see Fig. 1.15).Signals are ultimately transmitted to the thalamus and gustatory cortex. Electrophy-siological studies indicate that individual taste neurons have broad, overlapping re-sponse patterns (i.e. they are broadly tuned) so that an individual fiber is non-specificbut collectively the pattern of activity across multiple neurons is unique for a givenstimulus [77, 80].

1.6.3

Transduction of Taste Signals

Taste stimuli interact with taste proteins (e.g. taste receptors) or with ion channels onthe surface of taste cells, which induces electrical signals that ultimately reach thebrain to register a taste. The salty taste of sodium salts is produced when Naþ ionstraverse sodium channels in the membranes of taste cells [81]. The taste of potassiumsalts, like sodium salts, involves conductance of Kþ cations through taste cell mem-branes [82] Most studies indicate that the detection of bitter and sweet by tastantsreceptor cells involves G-protein-coupled receptors. Some but not all sweet com-pounds appear to bind to 7TM-spanning cell-surface receptors that activate the ade-nylate cyclase second messenger cascade [83]. At least two pathways are involved inbitter taste transduction: 1) the phosphatidylinositol secondmessenger cascade, and 2)the alpha-gustducin/phosphodiesterase pathway [86].

1.6.4

Molecular Biology of Taste

At current writing, two families of G-protein-coupled receptors designated as T1R(taste receptor family 1) and T2R (taste receptor family 2) are known to be selectivelyexpressed in subsets of taste receptor cells. In 1999, Hoon et al. [84] cloned and char-acterized two novel 7TM domain proteins T1R1 and T1R2 (taste receptor family 1,members 1 and 2) that are expressed in topographically distinct subpopulations oftaste receptor cells and taste buds. The receptors were localized to the taste pore.The following year, a novel family of receptors T2R were identified [85–87], andlike T1R1 and T1R2, the T2R genes were selectively expressed in taste receptorcells. The T2R family consists of 40-80 proteins that appear to code specifically forbitter tastants. A candidate sweet receptor gene, called T1R3 (taste receptor family1, member 3) was also been identified [88–91]. Further research has shown that re-ceptors T1R2 and T1R3 combine by dimerization producing heterodimers (T1R2 þ 3)to recognize sweet-tasting molecules with different structures such as sucrose andsaccharin [92]. Receptors T1R1 and T1R3 combine by dimerization producing hetero-dimers (T1R1 þ 3) that are broadly tuned to recognize L-amino acids [93]. A receptor

1.6 Taste 2525

that binds the amino acid L-glutamate called mGluR4 has also been cloned and char-acterized [94].Cells expressing T1R2 þ 3 are found predominantly on the posterior tongue, which

is innervated by the glossopharyngeal nerve [92]. Cells expressing T1R2 þ 3 are alsolocated on the palate. Cells expressing T1R1 þ 3 are found predominantly on the frontof the tongue, which is innervated by the chorda tympani nerve. Even though thesedifferent taste receptor types appear to be segregated anatomically, electrophysiologi-cal experiments indicate that individual taste cells and nerve fibers respond to stimulihaving multiple taste qualities [77, 80]. Thus, further research is needed clarify the fullrange of taste receptors as well as elucidate how this taste information is coded by thenervous system.

1.7

Final Comment

The biological chemosensory systems just described share many analogies to exam-ples of machine olfaction described in this book. For example, both the human olfac-tory system andmachines havemechanisms for sample handing. In humans, a sniff isinitiated when the diaphragm creates a relative negative pressure in the lungs andforces an air sample to be drawn through the nostrils and directed by the curved tur-binates onto the sensory layer of the olfactory epithelium. In a typical electronic nose, avacuum pump produces a negative relative pressure to draw the air sample through atube (plastic or stainless steel) in a small chamber housing the electronic sensor array.Both biological systems and machines have far fewer sensors than the thousands ofknown odorants. Humans have several hundred different receptor types while theelectronic nose typically has only 5 to 32 sensors. Both biological and machinessend their responses into multilevel neural networks that identify and characterizethe odor being produced by the odorant sample. Future advances in the molecularbiology of smell and taste will undoubtedly impact the development of new electronicnose and electronic tongue devices.

References

1 G. Ohloff. Chemistry of odor stimuli. Ex-perientia 1986, 42, 271–279.

2 M. Stuiver. Biophysics of the sense of smell.PhD Thesis. Groningen. 1958.

3 H. DeVries, M. Stuiver. The absolute sen-sitivity of the human sense of smell. inSensory Communication (Ed.: W. A. Rosen-blith), JohnWiley and Sons, New York, 1961,Chapter 9, pp. 159–167.

4 USEPA (United States Environmental Pro-tection Agency). Odor and corrosion control insanitary sewerage systems and treatment plants.USEPA, Cincinnati, OH EPA/625/1-85/018,1985.

5 A. Jinks, D. G. Laing. The analysis of odormixtures: evidence for a configurationalprocess. Physiol. Behav. 2001, 72, 51–63.

6 M. G. J. Beets. Structure-activity relationshipsin human chemoreception. Applied SciencePublishers Ltd., London, 1978.

7 R. Axel. The molecular logic of smell.Sci. Am. 1995, 273, 154–159.

8 L. Buck, R. Axel. A novel multigene familymay encode odorant receptors: a molecularbasis for odor recognition. Cell 1991, 65,175–87.


9 P. Mombaerts. Molecular biology of odorantreceptors in vertebrates. Annu. Rev. Neurosci.1999, 22, 487–509.

10 P. Mombaerts. Seven-transmembraneproteins as odorant and chemosensoryreceptors. Science 1999, 286, 707–711.

11 Aristotle, De Anima, translated byW. S. Hett, Heinemann, London, Revisedand reprinted, 1957.

12 C. Linnaeus (C. Von Linne) Odores medica-mentorum in Amoenitates Academicae, vol. 3,Lars Salvius: Stockholm, 1752, p. 183.

13 H. Zwaardemaker. Die Physiologie desGeruchs, translated from Dutch by A. J.von Langegg, W. Engelmann: Leipzig, 1895,p. 324.

14 H. Henning. Der Geruch I, Z. Psychol.Physiol. Sinnesorgane 1915, 73, 161–257.

15 S. Klein. Primary odour element classifica-tion. Amer. Perfum. Essent. Oil Rev. 1947, 50,453–454.

16 J. E. Amoore. The stereochemical specifici-ties of human olfactory receptors. Perfum.Essent. Oil Rec. 1952, 43, 321–323, and 330.

17 J. E. Amoore. The stereochemical theory ofolfaction. 1. Identification of seven primaryodours. Proc. Sci. Sect. Toilet Goods Assoc.New York 1962, 37, S1–S12.

18 J. E. Amoore. The stereochemical theory ofolfaction. 2. Elucidation of the stereoche-mical properties of the olfactory receptorsites. Proc. Sci. Sect. Toilet Goods Assoc.New York 1962, 37, S13–S23.

19 H. G. Schutz. A matching standards methodfor characterising odour qualities. Ann.N. Y. Acad. Sci. 1964, 116, 517–526.

20 American Society for Testing and Materials(ASTM), Atlas Of Odor Character Profiles. DS61. ASTM, Philadelphia, 1992.

21 www.AlluredCompendium.com. AlluredPublishing Corp., 362 S. Schmale Road,Carol Stream, IL USA 60188-2787, 2001.

22 Aldrich Flavors and Fragrances http://www.sigma-aldrich.com. Aldrich,1001 St. Paul Avenue, Milwaukee,WI 53233 USA.

23 http://www.leffingwell.com.24 D. Whissell-Buechy, J. E. Amoore. Odour-

blindness to musk: simple recessive in-heritance. Nature 1973, 242, 271–273.

25 C. J. Wysocki, G. K. Beauchamp. Abilityto smell androstenone is genetically deter-mined. Proc. Natl. Acad. Sci. USA 1984, 81,4899–4902.

26 J. E. Amoore, Specific anosmia and theconcept of primary odors. Chem. SensesFlavor 1977, 2, 267–281.

27 L. Turin, F. Yoshii. Structure-odor relations:A modern perspective. in Handbook of Ol-faction and Gustation (Ed.: R. L. Doty),Marcel Dekker, New York, 2002.

28 J. E. Amoore, D. Venstrom. Correlationsbetween sterochemical assessment and or-ganoleptic analysis of odorous compounds.In Olfaction and Taste, vol. 2 (Ed.: T.Hayashi), Pergamon, Oxford, 1967, p. 3–17.

29 J. E. Amoore. Stereochemical and vibrationaltheories of odour. Nature 1971, 233,270–271.

30 R. H. Wright. Stereochemical and vibratio-nal theories of odour. Nature 1972, 239, 226.

31 R. H. Wright. Odour and molecular vibra-tion. Nature 1966, 209, 571–572.

32 R. H. Wright, A. Robson. Basis of odourspecificity: homologues of benzaldehyde andnitrobenzene. Nature 1969, 222, 290–292.

33 A. Dravnieks, P. Laffort. Physico-chemicalbasis of quantitative and qualitative odordiscrimination in humans. in Olfaction andTaste, vol. 4, (Ed.: D. Schneider), Wissen-schaftliche Verlagsgesellschaft, Stuttgart,1972, pp. 142–148.

34 M. Chastrette. Trends in structure-odorrelationships. SARQSAR Environ. Res. 1997,6, 215–254.

35 K. J. Rossiter. Structure-Odor Relationships.Chem. Rev. 1996, 96, 3201–3240.

36 S. S. Schiffman, M. L. Reynolds, F. W.Young. Introduction to multidimensional sca-ling: Theory, methods, and applications. NewYork: Academic Press, New York, 1981.

37 S. S. Schiffman. Future design of flavourmolecules by computer. Chem Ind. 1983, 3,39–42.

38 S. S. Schiffman. Physicochemical correlatesof olfactory quality. Science 1974, 185,112–117.

39 S. S. Schiffman, J. C. Leffingwell. Perceptionof odors of simple pyrazines by youngand elderly subjects: A multidimensionalanalysis. Pharmacol. Biochem. Behav. 1981,14, 787–798.

40 A. J. Stuper, P. C. Jurs. ADAPT: A computersystem for automated data analysis usingpattern recognition techniques. J. Chem. Inf.Comp. Sci. 1976, 16, 99–105.

41 http://zeus.chem.psu.edu/ADAPT.html


42 J. Cohen, P. Cohen. Applied multiple regres-sion/correlation analysis for the behavioral sci-ences. Lawrence Erlbaum, Hillsdale NJ, 1983.

43 V. A. Benignus, K. E. Muller, J. A. Graham,C. N. Barton. Toluene levels in blood andbrain of rats as a function of toluene level ininspired air. Environ. Res. 1984, 33, 39–46.

44 J. A. Maruniak, W. L. Silver, D. G. Moulton.Olfactory receptors respond to blood-borneodorants. Brain Res. 1983, 265, 312-316.

45 D. G.Moulton. Dynamics of cell populationsin the olfactory epithelium. Ann. N Y Acad.Sci. 1974, 237, 52–61.

46 D. Lancet. Olfaction. The strong scent ofsuccess. Nature 1991, 351, 275–276.

47 E. E. Morrison. Morphology and plasticiy ofthe vertebrate olfactory epithelium. inScience of olfaction (Ed.: M. J. Serby,K. L. Chobor) Springer-Verlag, New York,1992, p. 31–50.

48 K. Mori, G. M. Shepherd. Emergingprinciples of molecular signal processingby mitral/tufted cells in the olfactory bulb.Semin. Cell Biol. 1994, 5, 65–74.

49 K. J. Ressler, S. L. Sullivan, L. B. Buck.A zonal organization of odorant receptorgene expression in the olfactory epithelium.Cell 1993, 73, 597–609.

50 K. Mori , H. Nagao, Y. Yoshihara. Theolfactory bulb: coding and processing ofodor molecule information. Science 1999,286, 711–715.

51 B. D. Rubin, L. C. Katz. Optical imaging ofodorant representations in the mammalianolfactory bulb. Neuron 1999, 23, 499–511.

52 L. Belluscio, L. C. Katz. Symmetry, stereo-typy, and topography of odorant represen-tations in mouse olfactory bulbs. J. Neurosci.2001, 21, 2113–2122.

53 B. Kettenmann, C. Hummel, H. Stefan,G. Kobal. Multichannel magnetoencephalo-graphical recordings: separation of corticalresponses to different chemical stimulationin man. Electroencephalogr. Clin. Neurophy-siol. Suppl. 1996, 46, 271–274.

54 D. H. Zald, J. V. Pardo. Emotion, olfactionand the human amygdala: Amygdalaactivation during aversive olfactorystimulation. Proc. Natl. Acad. Sci. USA 1997,94, 4119–4124.

55 J. E. Cometto-Muniz, W. S. Cain. Sensoryirritation. Relation to indoor air pollution.Ann. N. Y. Acad. Sci. 1992, 641, 137–151.

56 J. E. Cometto-Muniz, W. S. Cain. Sensoryreactions of nasal pungency and odor tovolatile organic compounds: the alkylben-zenes. AIHAJ 1994, 55, 811–817.

57 H. Breer. Odor recognition and secondmessenger signaling in olfactory receptorneurons. Semin. Cell Biol. 1994, 5, 25–32.

58 A. D. Gibson, D. L. Garbers. Guanylylcyclases as a family of putative odorantreceptors. Ann. Rev. Neurosci. 2000, 23,417–439.

59 F. Zufall, T. Leinders-Zufall. The cellularand molecular basis of odor adaptation.Chem. Senses 2000, 25, 473–481.

60 W. C. Probst, L. A. Snyder, D. I. Schuster,J. Brosius, S. C. Sealfon. Sequencealignment of the G-protein coupled receptorsuperfamily. DNA Cell Biol. 1992, 11, 1–20.

61 Y. Pilpel, D. Lancet. The variable andconserved interfaces of modeled olfactoryreceptor proteins. Protein Sci. 1999, 8,969–977.

62 D. Krautwurst, K. W. Yau, R. R. Reed.Identification of ligands for olfactory recep-tors by functional expression of a receptorlibrary. Cell 1998, 95, 917–926.

63 L. B. Buck. Information coding in the ver-tebrate olfactory system.Annu. Rev. Neurosci.1996, 19, 517–544.

64 S. Rouqier, C. Friedman, C. Delettre,G. van den Engh, A. Blancher,B. Crouau-Roy, B. J. Trask, D. Giorgi.A gene recently inactivated in humandefines a new olfactory receptor familyin mammals. Hum. Mol. Gen. 1998, 7,1337–1345.

65 S. Rouquier, A. Blancher, D. Giorgi.The olfactory receptor gene repertoire inprimates and mouse: evidence for reductionof the functional fraction in primates. Proc.Natl. Acad. Sci. USA 2000, 97, 2870–2874.

66 S. Zozulya, F. Echeverri, T. Nguyen. Thehuman olfactory receptor repertoire.GenomeBiol. 2001, 2(6), research0018.1-0018.12,http://genomebiology.com/2001/2/6/research/0018.

67 N. Ben-Arie N, D. Lancet, C. Taylor,M. Khen, N. Walker, D. H. Ledbetter,R. Carrozzo, K. Patel, D. Sheer, H. Lehrach.Olfactory receptor gene cluster on humanchromosome 17: Possible duplication of anancestral receptor repertoire. Hum. Mol.Genet. 1994, 3, 229–235.


68 B. J. Trask, C. Friedman, A.Martin-Gallardo,L. Rowen, C. Akinbami, J. Blankenship,C. Collins, D. Giorgi, S. Iadonato,F. Johnson, W. L. Kuo, H. Massa,T. Morrish, S. Naylor, O. T. Nguyen,S. Rouquier, T. Smith, D. J. Wong, J.Youngblom, G. van den Engh. Members ofthe olfactory receptor gene family are con-tained in large blocks of DNA duplicatedpolymorphically near the ends of humanchromosomes. Hum. Mol. Genet. 1998, 7,13–26.

69 B. Malnic, J. Hirono, T. Sato, L. B. Buck.Combinatorial receptor codes for odors. Cell1999, 96, 713–723.

70 J. P. Royet, C. Souchier, F. Jourdan,H. Ploye. Morphometric study of the glo-merular population in the mouse olfactorybulb: numerical density and size distributionalong the rostrocaudal axis. J. Comp Neurol.1988, 270, 559–568.

71 N. E. Rawson, J. Eberwine, R. Dotson,J. Jackson, P. Ulrich, D. Restrepo. Expres-sion of mRNAs encoding for two differentolfactory receptors in a subset of olfactoryreceptor neurons. J. Neurochem. 2000, 75,185–195.

72 H. Zhao, L. Ivic, J. M. Otaki , M. Hashimoto,K. Mikoshiba, S. Firestein. Functionalexpression of a mammalian odorant recep-tor. Science 1998, 279, 237–242.

73 R. D. Mattes. The taste of fat elevates post-prandial triacylglycerol. Physiol. Behav. 2001,74, 343–348.

74 S. S. Schiffman, M. S. Suggs, A. L. Sostman,S. A. Simon. Chorda tympani and lingualnerve responses to astringent compounds inrodents. Physiol. Behav. 1992, 51, 55–63.

75 S. S. Schiffman, B. G. Graham, E. A. Sattely-Miller, Z. S. Warwick. Orosensory percep-tion of dietary fat. Curr. Dir. Psychol. Sci.1998, 7, 137–143.

76 M. G. Tordoff. Calcium: Taste, intake, andappetite. Physiol. Rev. 2001, 81, 1567–1597.

77 S. S. Schiffman. Taste quality and neuralcoding: implications from psychophysicsand neurophysiology. Physiol. Behav. 2000,69, 147–159.

78 T. A. Gilbertson. Gustatory mechanismsfor the detection of fat.Curr. Opin. Neurobiol.1998, 8, 447–452.

79 S. S. Schiffman, Z. S. Warwick. The biologyof taste and food intake. in The Science OfFood Regulation: Food Intake, Taste, Nutrient

Partitioning, And Energy Expenditure,Pennington Center Nutrition Series, Volume 2(Ed.: G. A. Bray, D. H. Ryan), LouisianaState University Press, Baton Rouge, 1992,p. 293–312.

80 T. A.Gilbertson, J. D. Boughter Jr,H. Zhang,D. V. Smith. Distribution of gustatorysensitivities in rat taste cells: whole-cellresponses to apical chemical stimulation.J. Neurosci. 2001, 21, 4931–4941.

81 S. S. Schiffman, E. Lockhead, F. W. Maes.Amiloride reduces the taste intensity ofNaþ and Liþ salts and sweeteners. Proc. Natl.Acad. Sci. 1983, 80, 6136–6140.

82 M. Kim, C. M. Mistretta. 4-aminopyridinereduces chorda tympani nerve tasteresponses to potassium and alkali saltsin rat. Brain Res. 1993, 612, 96–103.

83 B. Lindemann. Chemoreception: tastingthe sweet and the bitter. Curr. Biol. 1996, 6,1234–1237.

84 M. A. Hoon, E. Adler, J. Lindemeier,J. F. Battey, N. J. Ryba, C. S. Zuker. Putativemammalian taste receptors: a class of taste-specific GPCRs with distinct topographicselectivity. Cell 1999, 96, 541–551.

85 E. Adler, M. A. Hoon, K. L. Mueller,J. Chandrashekar, N. J. Ryba, C. S. Zuker.A novel family of mammalian taste recep-tors. Cell 2000, 100, 693–702.

86 H. Matsunami, J. P. Montmayeur,L. B. Buck. A family of candidate tastereceptors in human and mouse. Nature2000, 404, 601–604.

87 J. Chandrashekar, K. L. Mueller,M. A. Hoon, E. Adler, L. Feng, W. Guo,C. S. Zuker, N. J. Ryba. T2Rs functionas bitter taste receptors. Cell 2000, 100,703–711.

88 J. P. Montmayeur, S. D. Liberles,H. Matsunami, L. B. Buck. A candidatetaste receptor gene near a sweet taste locus.Nat. Neurosci. 2001, 4, 492–498.

89 M. Max, Y. G. Shanker, L. Huang, M. Rong,Z. Liu, F. Campagne, H. Weinstein,S. Damak , R. F. Margolskee. Tas1r3,encoding a new candidate taste receptor,is allelic to the sweet responsiveness locusSac. Nat. Genet. 2001, 28, 58–63.

90 E. Sainz, J. N. Korley, J. F. Battey, S. L.Sullivan. Identification of a novel memberof the T1R family of putative taste receptors.J. Neurochem. 2001, 77, 896–903.


91 X. Li, M. Inoue, D. R. Reed, T. Huque,R. B. Puchalski, M. G. Tordoff, Y. Ninomiya ,G. K. Beauchamp, A. A Bachmanov. High-resolution genetic mapping of the saccharinpreference locus (Sac) and the putative sweettaste receptor (T1R1) gene (Gpr70) to mousedistal Chromosome 4.Mamm.Genome 2001,12, 13–16.

92 G. Nelson, M. A. Hoon, J. Chandrashekar,Y. Zhang, N. J. Ryba, C. S. Zuker.Mammalian sweet taste receptors. Cell 2001,106, 381–390.

93 G. Nelson, J. Chandrashekar, M. A. Hoon„L. Feng, G Zhao, N. J. Ryba, C. S. Zuker.An amino-acid receptor. Nature onlinepublication, 24 February 2002 (DOI10.1038/nature726).

94 N. Chaudhari, A. M. Landin, S. D. Roper.A metabotropic glutamate receptor variantfunctions as a taste receptor. Nat. Neurosci.2000, 3, 113–119.

Appendix

The basic matrix equations used by Schiffman [38] to maximize the configurationalsimilarity of the psychologically determined space in Fig. 1.8 with a space generated byweighted physicochemical parameters are:

P ¼ P þ EP ¼ DQP ¼ DQ þ E

where P is an (n)(n�1)/2 column vector whose elements pij represent all the inter-stimulus distances between stimulus i and stimulus j and where n is the total numberof stimuli; P is an (n)(n�1)/2 column vector representing the proximity measuresbased on weighted physicochemical parameters; D is an [(n)(n�2)/2] by k scalar di-stance matrix whose elements d2ðijÞk are the squared differences between stimulus i andstimulus j for each physicochemical parameter k; Q is a k element column vector ofweights for the k physicochemical parameters; and E is an (n)(n�1)/2 column vectorrepresenting the error between the subjective proximities and the proximities based onphysicochemical measures.

The error to be minimized is

@E 0E=@Qk ¼ 0

leading to the least squares solution

Q ¼ ðD 0DÞ�1D 0P

The equations for canonical correlation used to relate the descriptors in Table 1.6 to thethree dimensional arrangement in Fig. 1.9 are given below.

yyki ¼ ako þ ak1ðyi1Þ þ ::::: þ akrðyirÞ

xxki ¼ bko þ bk1ðxi1Þ þ :::::þ bkrðxirÞ


where xil, xi2, etc. are the values of stimulus i on dimensions 1 and 2 of the MDS spacejust as in multiple regression equations, and yi1, yi2, etc. are ratings of stimulus i onseveral physicochemical parameters. The intercepts and weights are solved to maxi-mize the correlation between yykiand xxki.


2

Chemical Sensing in Humans and Machines

J. Enrique Cometto-Muniz

Abstract

Chemosensory detection of airborne chemicals by humans is accomplished princi-pally through olfaction and mucosal chemesthesis. Odors are perceived via stimula-tion of the olfactory nerve (CN I) whereas nasal chemesthetic sensations (i.e., prick-ling, irritation, stinging, burning, freshness, piquancy, etc), grouped under the termnasal pungency, are mediated by the trigeminal nerve (CN V). Airborne compoundselicit odor sensations at concentrations orders of magnitude below those producingpungency. The physicochemical basis for odor and pungency potency of chemicals,either singly or in mixtures, is far from understood. The sensitivity of the sense ofsmell often outperforms that of the most sophisticated chemico-analytical methodslike gas chromatography and mass spectrometry. The combined used of these tech-niques with human odor detection (olfactometry), however, has proved an invaluabletool for understanding the chemosensory properties of complex mixtures such asfoods, flavors, and fragrances.

2.1

Human Chemosensory Perception of Airborne Chemicals

Humans detect the presence of volatile organic compounds (VOCs) in their surround-ings principally through their senses of olfaction and “chemesthesis” [1, 2], the latter isalso known as the “common chemical sense” [3, 4]. Activation of the olfactory nerve(CN I) produces odor sensations; Chapter 1 describes the biological basis of this che-mosensory pathway. Activation of chemoreceptors on the trigeminal nerve (CNV)innervating the face mucosae produces chemesthetic responses (see, for example,[5]). These responses evoked in the nose include stinging, piquancy, burning, fresh-ness, tingling, irritation, prickling, and the like. All these nasal sensations can begrouped under the term nasal pungency [6].Chemesthetic responses to airborne VOCs can also be produced in the ocular, oral,

and upper airway mucosae, where they are referred to as eye, mouth, and throat irrita-tion. In the back of the mouth and the throat, other nerves, such as the glossophar-


3333

yngeal (CN IX) and vagus (CNX), are also stimulated by airborne VOCs and contributeto perceived irritation.In this chapter we will focus on the human sense of smell and nasal chemesthesis.

We will review psychophysical studies performed on both sensory modalities addres-sing the possible basis for the odor and irritation potency of VOCs. We will also sum-marize various techniques that combine the power of the human nose with that ofchemical-analytical instruments, such as gas chromatography and mass spectrome-try, to quantify the chemosensory activity of volatile chemicals and to help understandbetter the characteristics of human chemosensory perception.

2.2

Nasal Chemosensory Detection

Odor thresholds represent an important biological characteristic of airborne chemi-cals. Nevertheless, compilation of such values [7–9] show an extreme variabilityfor any particular substance, even after attempting to standardize the values reportedin different sources [10]. This scatter severely limits the practical application of theinformation available. An important block in our understanding of smell and nasalchemesthesis is the lack of information regarding what particular characteristics ofchemicals govern the potency (i.e., threshold and suprathreshold) and type (i.e., qual-ity) of olfactory and trigeminal sensations that they evoke. The situation stands insharp contrast with the senses of vision and hearing where we have a precise knowl-edge of the range of electromagnetic and vibrational energy, respectively, to which oureyes and ears are tuned. From a few known, well-defined parameters of light andsound it is relatively straightforward to predict its visual and auditory perceptual prop-erties. It is not easy to predict the odor or chemesthetic perceptual properties from thestructural and physicochemical properties of a compound.Attempts to correlate odor with structural and physicochemical properties of odor-

ants have focused, typically, on one or a small number of odor qualities (see reviews in[11, 12]), probably because broader generalizations have failed to lead to a productiveoutcome. As has been pointed out [13], an important drawback of many structure-activity relationships in olfaction [14–19] is the difficulty in interpreting the chemicalfeatures that are shown to correlate with odor activity.Regarding chemesthesis in the upper airways, a pioneer review paper [20] described

the possible chemical mechanisms of sensory irritation. This study focused principallyon “reactive” chemicals, that is, substances producing chemesthetic responses prin-cipally via direct chemical reaction with mucosal tissues. A more recent review of thetopic [21] also addressed the mechanism by which relatively nonreactive compoundscould produce pungency. In fact, relatively nonreactive VOCs are the prime candidatesfor the production of adverse chemosensory symptoms in cases of indoor air pollutionsuch as the sick-building syndrome (cf. [22]).Among the various factors accounting for the large variability of measured odor

thresholds, apart from true biological variability, are: method of vapor-stimulus con-trol and/or delivery, psychophysical methodology, criteria to arrive at a threshold re-

2 Chemical Sensing in Humans and Machines34

sponse, number of subjects, and number of trials per subject [23, 24]. In the case ofnasal pungency thresholds, a crucial additional factor is the use of a procedure thatavoids odor biases, because almost all chemicals have both odor and pungency and theodor could be quite strong at the concentrations needed to produce barely perceptiblenasal pungency. Additionally, in order to build a chemosensory structure-activity re-lationship, a chemical stimulus continuum of some sort can be very helpful.In a wide-ranging research program started more than 10 years ago [25], odor and

nasal pungency thresholds were measured using a standardized procedure aimed atminimizing many of the variability sources mentioned above, and to produce a dataset with robust internal consistency. Some of the procedural features employed in-cluded:

1) Delivery of vapors monorhinally (i.e. one nostril at a time) via “static” olfactometry[26] from plastic squeeze bottles [27].

2) Short-term exposures (1–2 seconds).3) Rigorous measurement and follow-up of presented vapor-phase concentrations by

gas chromatography.4) Use of a two-alternative, forced-choice procedure against a blank to minimize bias;

presentation of chemicals in an ascending concentration series to minimize sen-sory adaptation; and the use of, at least, duplicate bottles containing identical con-centrations to alternate sniff sampling and avoid depletion of stimulus in the head-space.

5) Use of a constant and fixed criterion for threshold, such as five correct choices in arow, across subjects, repetitions, chemosensory modality (i.e. odor and nasal pun-gency), and different studies.

6) Selection of subjects with no sense of smell (called anosmics) to measure nasalpungency thresholds thus avoiding odor biases, and of subjects with normal senseof smell (normosmics) to measure odor thresholds. Normosmics were matched tothe anosmics by age, gender and smoking status, all demographic variables knownto influence chemosensory perception (see review in [28]).

7) Selection of stimuli from homologous chemical series, where physicochemicalproperties change systematically and where carbon chain length provides a conve-nient “unit of change” (i.e., a continuum) against which to relate the sensory results.

2.2.1

Thresholds for Odor and Nasal Pungency

The systematic studies of odor and nasal pungency thresholds along homologous che-mical series included testing of n-aliphatic alcohols [25], n-acetate esters [29], sec- andtert-alcohols and acetate esters [30], ketones [30], alkylbenzenes [31], and aliphatic al-dehydes and carboxylic acids [6]. Figure 2.1 summarizes the results obtained with allthese series.The outcome clearly shows how both chemosensory thresholds decline as carbon

chain length increases; this means that sensory potency (both olfactory and trigeminal)increases along each homologous series. The rate at which odor thresholds decline, at


least for the first few members of each series, tends to be higher than that for nasalpungency thresholds. In various instances, such as for acetate esters, ketones, andalkylbenzenes, odor thresholds seem to reach a plateau. Nasal pungency thresh-olds, in contrast, reach a “cut-off” effect [6]: beginning with a certain homolog mem-ber, nasal pungency fails to be consistently evoked, and this effect deepens for allensuing members. In other words, the ability of that particular homolog and of allfollowing homologs to produce nasal pungency fades away. The reduced biologicalresponse due to the cut-off effect, seen at some point in a chemical series, is awell-known pharmacological phenomenon in the field of anesthesia [32, 33]. At leasttwomechanisms can account for such cut-offs [33]: a physical mechanismwhereby themaximum vapor-phase concentration of the stimulus molecule, at a certain tempera-ture and pressure, falls below the threshold; and a biological mechanism whereby the

Fig. 2.1 Thresholds for odor (empty squares) and nasal pungency

(filled squares) along homologous chemical series of alcohols, acetate

esters, ketones, alkylbenzenes, aliphatic aldehydes, and carboxylic

acids. Only primary and unbranched homologs are joined by a line. The

segment of dotted lines on nasal pungency shows those homologs for

which pungency begins to “cut-off” (see text). Bars (sometimes hidden

by the symbol) indicate standard deviation


stimulus molecule lacks a crucial property to trigger transduction. For example, themolecule could be too large to fit into the binding pocket of a receptive macromoleculeor to interact effectively with a target site.

2.2.2

Stimulus-Response (Psychometric) Functions for Odor and Nasal Pungency

Studies that aimed at measuring thresholds for olfaction and nasal chemesthesis witha uniform methodology, particularly in the context of testing homologous chemicalseries, proved to be useful tools in understanding how physicochemical propertiesgovern sensory potency. The use of a standard testing procedure was instrumentalin developing robust quantitative structure-activity relationships (QSARs) (see be-low). Nevertheless, measurement of a punctate chemosensory threshold accordingto a fixed criterion of performance has limitations [34]. A more comprehensive knowl-edge of the chemosensory processes involved can be gained by measurement of com-plete stimulus-response (called psychometric) functions (e.g., [23, 24]). These func-tions span the range from chance detection to virtually perfect detection and, thuscross the boundaries between perithreshold and suprathreshold sensations. Givena certain set of testing conditions, psychometric functions depict a continuous trackof how the detectability of the chemical(s) grow with increasing concentration, render-ing a dynamic picture of the process.Figure 2.2 presents psychometric functions for the odor and nasal pungency evoked

by 1-butanol, 2-heptanone, butyl acetate, and toluene. All functions in Figure 2.2 show

Fig. 2.2 Psychometric function for the odor (empty symbols) and nasal

pungency (filled symbols) detection of butyl acetate (diamonds), 2-heptanone

(circles), toluene (triangles), and 1-butanol (squares)


an ogival shape with a close-to-linear section in the middle of the range. As expectedfrom previous studies on thresholds (see review in [5]), odor detection occurred ordersof magnitude below nasal pungency detection. The gap between olfactory and chemes-thetic detection (at halfway between chance and perfect detection) ranged between 3.4and 6.4 orders of magnitude. The two chemosensory modalities also differed in theslope along the linear portion of the function. Odor functions for these four chemicalshave slopes between 0.35 and 0.5 [34, 35] whereas nasal pungency functions haveslopes between 0.7 and 1.0, except toluene which showed an even steeper slope inthe range 1.7–2.9 [34, 36].

2.3

Olfactory and Nasal Chemesthetic Detection of Mixtures of Chemicals

In typical, everyday experiences, olfactory and chemesthetic sensations arise from ex-posures to mixtures of substances. Rarely are they the result of exposure to a singlechemical. In addition, the study of the chemosensory detection of mixtures comparedto the detection of the individual components has the potential to uncover basic prin-ciples of functioning of the senses of smell and chemesthesis.Studies on the olfactory detection of mixtures of airborne chemicals have relied, for

the most part, on measurement of thresholds according to a fixed criterion of perfor-mance, and have typically expressed the results in terms of the stimulus (that is, con-centration of the chemical). Their outcome suggests partial and simple stimulus agon-ism [37–39] with some indications of synergistic stimulus agonism as number ofcomponents increases [39–42]. To illustrate the meaning of these terms, let ustake the example of a 3-component mixture whose constituents are present at sen-sory-equivalent concentrations, i.e. at the samemultiple or submultiple of their respec-tive individual thresholds. The terms simple, synergistic, and partial agonism indicate,respectively, that the mixture achieves threshold when each component is present atone third, less than one third, and more than one third (but less than one time) itsindividual threshold concentration. The term independence indicates that the mixtureachieves threshold only when at least one of the components is present at its individualthreshold. The term antagonism indicates that the mixture achieves threshold onlywhen the components are present at a concentration even higher than their respectiveindividual thresholds. A recent study looking at the olfactory (and trigeminal) detect-ability of binary mixtures of 1-butanol and 2-heptanone via measurement of psycho-metric functions lent support, as a first approximation, to an outcome of simple agon-ism [35].Not surprisingly, studies on the trigeminal detection of mixtures are much fewer

than those on olfaction. A comprehensive study, measuring trigeminal thresholds forsingle chemicals and for mixtures of up to nine components, revealed a trend for thedegree of agonism to increase with the number of components and with the lipophi-licity of such components [39]. A couple of recent investigations used psychometricfunction measurements to look in detail at the trigeminal detectability of binary mix-tures compared to the detectability of the single components [35, 36]. The general out-


come again supported simple agonism with the suggested possibility, open to furtherscrutiny, that for chemesthetic responses, simple agonism might weaken to partialagonism as the detectability of the mixtures approach perfect detection [36], that is,as the mixtures leave the perithreshold region and enter into the suprathreshold re-gion. If such weakening of agonism is confirmed and extended to olfactory responses,it would correlate well with the finding of partial agonism (called hypoadditivity) verycommonly reported for mixtures of odorants at the suprathreshold range (e.g. [43])even when the analysis considers “addition” of concentration (mass) and not simplyaddition of sensation [44].It has been suggested that, within each chemosensory modality, compounds with

similar slopes in psychometric functions will tend to show simple agonism in mix-tures, whereas compounds with different slopes will tend to show a lesser degreeof agonism, e.g. partial agonism [36]. At this stage, psychometric functions for addi-tional substances tested in binary and higher order mixtures need to be measured toconfirm the trend.

2.4

Physicochemical Determinants of Odor and Nasal Pungency

As mentioned above, the senses of olfaction and chemesthesis allow the detection ofairborne chemicals. To gain a better understanding of how these sensory channelsfunction it is important to know what particular features of chemicals govern theirpotency as odorants and irritants, including threshold and suprathreshold intensi-ties. Regarding olfaction, a large number of such features have been suggested, in-cluding Wiswesser notation formulas [14], structural parameters directly derivedfrom the chemical formula [45] or derived from gas chromatographic measurements[17, 19], steric and electronic descriptors [46], molecular vibration [47–49], partitioncoefficients (specifically, water-air and octanol-water) [50] and an electron-topologicalmethod [51]. Some of these investigations focused on one or just a few odor qualities(e.g. musk) whereas others studied a broader spectrum.Regarding chemesthesis, there have also been a number of chemical features re-

ported to correlate with sensory irritation. Among them, normal boiling point [52],adjusted boiling point [53], saturated vapor pressure [54], Ostwald solubility coeffi-cient (i.e., log L where L ¼ concentration in solvent/concentration in gas phase)[55], and partition coefficients, specifically water-air and octanol-water [56]. Interest-ingly, all these descriptors are physicochemical parameters and do not involve theprecise chemical structure of the irritant.

2.4.1

The Linear Solvation Model

Many of the QSARs cited above for olfaction and chemesthesis are difficult to interpreteither chemically or mechanistically [13]. A recently developed model has the advan-


tage of not only providing a strong statistical fit to human psychophysical data, but alsoof conveying chemically and mechanistically meaningful information on both the sti-mulus (i.e. odorant or irritant) and the biophase where sensory reception initially takesplace, e.g. for olfaction, the membrane covering the cilia of the olfactory receptor neu-ron, and, for nasal chemesthesis, the membrane of the free nerve endings of the tri-geminal nerve. Thismodel is based on a general solvation equation developed by Abra-ham [57, 58]:

log SP ¼ c þ r � R2 þ s � pH2 þ a �X

aH2 þ b �X

bH2 þ l � log L16 ð1Þ

where SP is the dependent variable that, in the present context, represents a sensoryproperty defined as the reciprocal of the odor detection threshold (1/ODT) or the re-ciprocal of the nasal pungency threshold (1/NPT). The reciprocals are used simplybecause the larger the quantity, the more potent is the odorant or irritant. Thereare five independent variables: excess molar refraction (R2), dipolarity/polarizability(pH2 ), overall or effective hydrogen-bond acidity (

PaH2 ), overall or effective hydro-

gen-bond basicity (P

bH2 ), and gas-liquid partition coefficient on hexadecane at298 K (L16). The L16 descriptor is a combination of two properties of the odorantor irritant: 1) a general measure of size, and 2) the ability of the odorant or irritantto interact with a biophase through dispersion forces. The term c and the coefficientfor each of the independent variables (r, s, a, b, and l) are obtained by multiple linearregression analysis. However, these are not simply fitted coefficients. They have che-mical and mechanistic meaning since they reflect the complementary properties thatthe biophase must show in order to be receptive to the odorant or irritant. In otherwords, the independent variables provide a physicochemical characterization of thestimulus whereas the corresponding coefficients provide a characterization of the re-ceptive biophase likely to interact with that stimulus. The r-coefficient measures thetendency of the biophase to interact with the odorant or irritant via polarizability-typeinteractions, mostly via p- and n-electron pairs. The s-coefficient reflects the biophasedipolarity/polarizability, since a dipolar odorant or irritant will interact with a dipolarbiophase, and a polarizable odorant or irritant will interact with a polarizable biophase.The a-coefficient represents the complementary property to the odorant or irritanthydrogen-bond acidity, and thus is a measure of the biophase hydrogen-bond basici-ty, since an acidic odorant or irritant will interact with a basic biophase. Similarly, the b-coefficient is a measure of the biophase hydrogen-bond acidity, since a basic odorantor irritant will interact with an acidic biophase. Finally, the l-coefficient is a measure ofthe biophase lipophilicity [13].

2.4.2

Application of the Solvation Equation to Odor and Nasal Pungency Thresholds

The standardized procedure employed to measure the odor and nasal pungencythresholds depicted in Fig. 2.1 provided a firm basis to develop QSARs based onthe solvation model described above. Under this model, the odorant or irritant is


seen as a solute that travels through a series of solvent phases (air, nasal mucus, nasaltissue) until it exerts its (sensory) action upon a receptive biophase. Thus, the modelonly applies to what can be called “transport” processes. These are processes where thefundamental step is either the distribution of the stimulus between biophases or therate of transfer of the stimulus from one biophase to another. The model does notapply to stimuli acting through exact conformational or geometrical states since thesesorts of molecular changes would barely affect the above mentioned physicochemicaldescriptors but, when relevant, could affect potency dramatically. In addition, themod-el does not apply to “reactive” compounds, that is, substances that produce nasal pun-gency via direct chemical reaction with nasal tissue [21]. The solvation equation wouldunderestimate the potency of such chemically reactive stimuli [59, 60].The original equation for odor thresholds [13] was recently improved [61] with the

addition of two additional terms:

1) A parabolic term (D�D2) whereD is themaximum length of the odorant moleculeobtained by computer-assisted molecular modeling and geometry optimization.

2) An indicator variable,H, chosen as 2.0 for carboxylic acids and aldehydes, and zerofor all other odorants. The need to introduceH arises because carboxylic acids andaldehydes are more potent than predicted [61]. The odor equation looks as follows:

log ð1=ODTÞ ¼ �7:445þ 0:304R2 þ 1:652 pH2 þ 2:104X

aH2

ß þ 1:500X

bH2 þ 0:822 log L16þ 0:369D�0:016D2þ1:000H ð2Þ

with n ¼ 60, r2 ¼ 0:84, SD ¼ 0:601, where n is the number of odorants included, r isthe correlation coefficient, and SD is the standard deviation in the dependent variable.All symbols are as described for Eq. (1).The solvation equation model has performed even better for the description and

prediction of nasal pungency thresholds [6, 62–65] than for odor thresholds. Its suc-cess indicates that transport processes indeed constitute a key step in the production ofnasal pungency by nonreactive airborne chemicals. The latest version of the nasalpungency equation looks as follows:

log ð1=NPTÞ ¼ �8:519þ 2:154pH2 þ 3:522X

aH2 þ 1:397X

bH2

þ0:860 log L16 ð3Þ

with n ¼ 43, r2 ¼ 0:955, SD ¼ 0:27, where all letters and symbols are as defined abo-ve. In this case, the term r:R2 from the general Eq. (1) did not achieve significance andwas omitted.It must be pointed out that Eq. (3) does not account for the observed cut-off effect on

nasal pungency that we have mentioned in Section 2.2.1. Future research should aimat optimizing the range of applicability of Eq. (3) by including a “size” factor capable ofaccounting for such molecular cut-offs in chemesthesis. This line of work is likely togather critical knowledge not only on the molecular boundaries of airborne pungentstimuli but also on those of the putative nasal chemesthetic receptor as well.


2.5

Human Chemical Sensing: Olfactometry

All studies exploring how humans detect and perceive airborne chemicals need todevise a strategy to generate and deliver the stimuli at predetermined concentrationlevels. Generation, delivery, and control of chemical stimuli entail more complexitythan the equivalent processes for physical stimuli such as lights and sounds. In addi-tion, there are practically no well-established, accepted, and widely used commercialdevices to perform such tasks. Inmany cases, a one-of-a-kind olfactometer is built withmuch effort and time for one or a few studies, only to be left in disuse, replaced, orsubstantially modified for other studies. As a rule, no steps are taken in order to under-stand how results obtained with the “old” device compare with those obtained with the“new” one.In this section we will discuss three broad olfactometric techniques that, with var-

iations, have been and are still being used in the study of human chemosensory per-ception [26].

2.5.1

Static Olfactometry

In general, olfactometric techniques can be classified into “static” or “dynamic” de-pending on whether the vapor stimulus is drawn from an enclosed container wherethe liquid and vapor phases of the tested chemical are in equilibrium, or the vaporflows continually in a carrier-gas stream, typically odorless air or nitrogen. Importantaspects in the static approach include the type of container, the way in which the vaporis drawn to the nose, and the type of connection between the headspace of the contain-er and the nose of the subject.Containers in static olfactometry are typically glass or (almost) odorless plastic. As a

rule, a series of dilutions of the substance(s) of interest are prepared in individualvessels using an odorless solvent. Choice of the solvent is not straightforward. Dis-tilled and deionized water could serve in some cases but some chemicals are unstablein water. For example, esters tend to hydrolyze producing the alcohol and the car-boxylic acid. Also, most odorants have little or extremely low water solubility. Alter-native solvents are lipophilic substances where odorants are more stable and soluble.These include, for example, mineral oil and propylene glycol. Nevertheless, these arenot always completely odorless and might present a low odor background. Many of theolfactory and nasal chemesthetic studies mentioned above resorted to the use ofsqueeze bottles [66] (Fig. 2.3(a)). Their caps have pop-up spouts that fit into one orthe other nostril allowing monorhinic testing, which in addition to their easy availabil-ity and simplicity of use has made them useful not only in research but also in theclinic [27]. A recent study, using three members each of homologous alcohols, acet-ates, and ketones, has shown that a newly developed glass vessel system possessesadvantages over the plastic squeeze bottles, producing nasal pungency thresholds sys-tematically lower by an average factor of 4.6 compared to those obtained via squeezebottles [67] (Fig. 2.3(b).


Subjects can sample the vapors in the headspace of the container actively by sniffingor they can receive them passively, for example, when the experimenter activates avalve that sends a fixed volume of headspace into the participant’s nostrils. The secondmethod [68] makes stimulation independent of the sniffing pattern of the subjects butit can cause progressive drying of the nasal mucosa, leading to irritation with repetitivestimulation, and can also lead to confusion between air pressure and odor sensations[69]. In addition, more recent studies have shown that natural sniffing maximizesolfactory performance in humans [70].The type of connection between the vapor container and the subject’s nostrils de-

termines the effective concentration reaching the nose. The squeeze bottles, with theirpop-up spouts that fit inside one nostril, represented an improvement over other con-tainers that are simply open and placed under the subject’s nose, but still left room fordilution of the stimulus from surrounding air. The above mentioned glass vesselsinclude Teflon made nosepieces that fit snugly into both nostrils of the subject, max-imizing the efficiency of the stimulus delivery [67].It is important to stress that in all these techniques of static olfactometry, the actual

stimulus is the vapor above the solution in the container. In principle, the vapor con-centration is proportional to the liquid concentration, but such proportionality varieswith odorants, solvents, and, sometimes, among concentrations of the same odorant-solvent pair. For these reasons, actual measurement of the vapor-phase concentrationin each container, and periodic follow-ups to ensure stability, become the only safe-guard against incorrect assumptions. Unfortunately, all too often olfactory investiga-tions do not include such vapor measurements. Gas chromatography provides a re-latively simple way to measure and calibrate vapor concentrations for use in staticolfactometry.

Fig. 2.3 (a) Olfactory testing of a subject via plastic squeeze bottles

and caps with pop-up spouts. (b) Olfactory testing of a subject via glass

vessels with Teflon nosepieces


2.5.2

Dynamic Olfactometry

Under the principles of dynamic olfactometry, the chemical stimulus flows continu-ously in a carrier gas stream of either purified air or nitrogen. The various concentra-tions of the substance(s) tested are typically achieved by mixing, in different propor-tions, the carrier-gas line with the odorant line. A number of elements including tub-ing, capillaries, flowmeters, mass flow controllers, valves, saturating and mixing ves-sels, deodorizing and air conditioning (i.e., temperature and humidity) devices con-stitute the necessary equipment for the generation and control of odorants. As in thecase of static olfactometry, the interface between the exit of the stimulus and the noseis an important feature regarding possible unwanted dilution of the targeted concen-tration. The complete assembly is referred to as an “olfactometer”.In a very detailed analysis of various olfactometers and of the many principles guid-

ing their design, Dravnieks [71] has described devices used in both animal and humanstudies. Dravnieks himself proposed a Binary Dilution Olfactometer [71] (Fig. 2.4).This instrument combines portability and stability of concentrations with ease ofuse and maintenance. Its simplicity arises from the fact that it uses saturated vaporas the source of undiluted stimulus and employs a series of capillaries of variouswidths and lengths to achieve 7 fixed increasing dilutions of the odorant, all presentedat a final flow rate of 160 mL/min. One of the suggested applications of this device wasto use it with 1-butanol so as to express the odor intensity of any source in terms of anodor-equivalent concentration of 1-butanol (in ppm by volume) [72]. The techniquebecame an ASTM (American Society for Testing and Materials) recommended pro-cedure [73]. Dravnieks also developed a Dynamic Forced-Choice Triangle Olfact-ometer for measurement of thresholds [74, 75]. Both types of olfactometers foundan important application in the measurement of environmental odors [76]).Chemical stimulation of the olfactory and trigeminal chemosensory systems in the

nose gives rise to both peripheral electrical potentials [77, 78] and central evoked po-tentials [79]. In order to study such electrophysiological events, an olfactometer wasneeded that 1) delivered the stimulus without altering the mechanical or thermal con-ditions at the stimulated mucosa, and 2) produced a sharp, square-wave type, stimulusonset and offset. Such an instrument was pioneered by Kobal and collaborators[77, 79]. Their instrument achieved these goals by embedding pulses of odorant orirritant in a constantly flowing air stream under controlled temperature (36.5 8C)and humidity (80% RH).An interesting development in the area of dynamic olfactometry emerged from the

description of an olfactometer that also served to measure respiratory parameters [80–83] (Fig. 2.5). The instrument evolved through the years and in its latest version pre-sents the odorants and irritants to subjects through amask, with a good seal monitoredby pressure, covering the nose and mouth in a room-temperature warmed (25 8C) andhumidified (35% RH) airflow. The concentration of the stimulus on the line feedingthe mask is continuously monitored by a photo-ionization detector (PID).


2.5.3

Environmental Chambers

Use of whole-body environmental chambers to explore human chemosensory re-sponses provides a close approximation to a “natural” setting. In static and dynamicolfactometry, two crucial issues that must be controlled include the actual concentra-tion of the stimulus, typically measured via detectors used in gas chromatography suchas PID or flame ionization detector (FID), and the nosepiece/nose interface. A looseinterface between the nostrils and the stimulus exit, whether under a static approach(e.g., squeeze bottles) or a dynamic approach (e.g., Dravnieks olfactometer) probably

Fig. 2.4 (a) Drawing illustrating some of the principles in the Dravnieks

Binary Dilution Olfactometer (from [71]). (b) A perspective drawing of the same

olfactometer (from [14])


results in a dilution of the effective stimulus. Different sniffing styles among subjectsmay also contribute to variability. Investigation of the “typical” characteristics of hu-man sniffing provide some interesting values: the “average” human sniff draws avolume of 200 mL, lasts a minimum of 0.4 sec and reaches an instantaneous flowrate of 30 L/min [70, 84, 85]. These studies also concluded that:

1) individuals vary in their sniffing techniques but are consistent with their patternsacross odorants and tasks;

2) most of the odor information is obtained in the first sniff;3) natural sniffing provides optimum odor perception.

Many of the above mentioned characteristics can not be easily achieved by static ordynamic olfactometry, hence the appeal of using environmental chambers. Neverthe-less, in a room-sized exposure chamber, build-up, control, and rapid change of stimu-lus concentration become complex and problematic as the large surfaces in the cham-ber (including the bodies and clothing of subjects) adsorb and desorb airborne che-micals. For these reasons, even when whole-body exposures constitute the gold stan-dard, the pace of testing under this approach is much slower. This highlights theimportance of understanding the rules of interconvertibility among sensory resultsobtained with the different approaches and, given the enormous number of odorantsand irritants, the need to develop robust quantitative structure-activity relationships for

Fig. 2.5 Schematic representation of the test station for measure-

ment of sensory responses and breathing parameters. (from [83].)


the prediction of chemosensory responses. Examples of these relationships have beenprovided above in Section 2.4.Chamber studies have been particularly useful when applied to the understanding of

issues of indoor air quality and associated topics. Since exposures in chambers can lastfor hours, they possess a clear advantage over other strategies when studying the effectof time of stimulation on chemosensory perception. Studies performed in environ-mental chambers have explored, among others, sensory responses to environmentaltobacco smoke [76, 86–89], body odor [90], volatile organic compounds [91–96], fra-grance materials in air fresheners [97], and formaldehyde (a substance off-gassingfrom certain home-insulation materials) [98].

2.6

Instruments for Chemical Sensing: Gas Chromatography-Olfactometry

Gas chromatography (GC), one of the most widely used techniques in analytical chem-istry, was first formalized in 1952 [99]. As described in a couple of recent reviews[100, 101], researchers interested in odors and aromas quickly took advantage ofthis powerful separation technique to identify the principal odorants of specific pro-ducts, for example foods, beverages, fragrances, and perfumes [102]. This particularapplication of GC is now known as gas chromatography-olfactometry (GC-O). In brief,the method uses GC to separate the individual components of a mixture (e.g., a foodproduct) which are then presented, as they elute, to a subject (called a sniffer) forsensory detection and/or characterization.Soon researchers found that direct sniffing from the GC effluent, at the exit of a non-

destructive detector, had important drawbacks. Among them, the hot and dry gasesdried the nasal mucosa, producing serious discomfort, and the odorous backgroundemitted by hot plastic components interfered with the detection of the eluting odorants[100]. This prompted the design of substantial improvements in the system that even-tually led to present day GC-O. An important step along the way was the addition ofhumidified air to the GC effluent, resulting in the delivery of a pulsed wave of odorant,similar to that eluting from the GC, but minimizing nasal dehydration and discomfortfor the human sniffer [103]. Further improvements included a venturi system thateliminated background odors, was able to handle narrow-bore GC columns with mini-mum loss of resolution, and provided additional comfort to the subject [104].As the techniques of GC and GC-mass spectroscopy (GC-MS) became widespread

and more sophisticated, it was possible to separate and chemically identify the dozensor hundreds of individual substances present in food, flavor, and fragrance products. Ithas been argued [105] that this knowledge created the illusion that the flavor chemistryof these products was well understood. These powerful analytical techniques by them-selves have no way of identifying and weighting which compounds are contributingsignificantly to flavor, and to what extent, hence the crucial importance of the GC-Oapproach that incorporates human sensory detection. In fact, there are indications thatthe performance of GC-O rivals, and can even outperform the most sensitive andselective chemico-analytical methods like GC-MS, particularly towards the most

2.6 Instruments for Chemical Sensing: Gas Chromatography-Olfactometry 4747

powerful odorants [106]. In addition, GC-O requires comparatively little sample pre-paration and no need for synthesis of labeled compounds. The usefulness of GC-Ocontinues to grow and expand as it combines with the latest analytical tools suchas solid-phase microextraction (SPME) [107, 108].Many GC-O systems are designed to split the GC effluent, sending part to a chemical

detector and part to the sniffing port. Typically, humans are more sensitive than mostchemical detectors so it is common that less than 10% of the effluent is directed to thesniffing port while more than 90% is directed to the detectors [109]. However, the useof non-destructive detectors, such as a thermal conductivity detector, TCD, allows allthe GC effluent to be sent to the sniffing port, maximizing sensitivity [101].We have discussed issues that deal with the optimization of GC effluents for che-

mosensory evaluation by human subjects. There are also issues that deal with theoverall strategy for presenting the stimulus (typically a complex mixture of odorantsand non-odorants) to the subjects and, very importantly, the procedure used to gatherand quantify sensory information from the subjects [109]. The application field weremany of these methods were developed and investigated relates to food and flavorresearch. At present, there are at least three techniques commonly used in the studyof the sensory properties of the chemical components of foods and flavors by GC-O.These are “charm analysis”, aroma extract dilution analysis (AEDA), and “osme” (fromthe Greek word meaning smell). We will briefly describe each of these methods.

2.6.1

Charm Analysis

This dilution technique was introduced in the middle 1980s [105]. On each run, thesubject is exposed to the GC effluents from one of a series of increasing dilutions of theparticular stimulus investigated, typically a complex mixture of chemicals. The parti-cipant strikes a key from a computer keyboard each time an odor begins to be detectedand, again, when the odor is no longer detectable. During this interval, the subject isalso required to report, for example with another key stroke, the quality of the per-ceived odor. The procedure renders a record of the time on the GC run where theodor occurred, its duration, and its quality. As the authors point out, a crucial partof the method calls for the use of chromatographic standards (e.g., n-paraffins) totransform the retention times at which odors appear into retention indexes, thus as-sociating the sensory response with a reproducible chemical property. A run as de-scribed above is made for each of the successive serial dilutions until no odor is de-tected.The responses are summarized as the “charm” value c, that is a simple function of

the dilution factor d and the number of coincident responses n. The term “coincidentresponses” refers to the number of times that an odor is detected across successivedilutions for a particular retention index. In this way, the relationship is expressed as:c ¼ dn�1. A charm response chromatogram is defined as a plot of c vs. retention index.Figure 2.6 illustrates how the charm plot is obtained. Results obtained by charm ana-lysis compare well with those obtained by using traditional psychophysical proceduressuch as line-length (a visual analogue scale) and finger-span [110].


Charm analysis has been applied to study, among other products, apples [111],grapes [112, 113], orange juices [114] and the off-flavors form plastic packaging offood products [115].

2.6.2

Aroma Extract Dilution Analysis (AEDA)

AEDA is another dilution technique [116]. As in charm analysis, an extract from theproduct of interest is diluted in series and each dilution is analyzed by GC-O. In AEDA,results are expressed as flavor dilution (FD) factors. This factor is simply the ratio of theconcentration of the odorant in the initial extract to its concentration at the highestdilution at which an odor is detected by GC-O [117, 118].AEDA chromatograms plot the flavor dilution factor vs. retention index. Graphs

obtained by charm analysis and by AEDA of the same flavor product are very similar[101] only that charm analysis produces areas for each relevant retention index (seeFig. 2.6) whereas AEDA produces heights, that is, a single number on the y-axis(equal to the FD) for each relevant retention index. In this way, AEDA focuses onthe highest dilution at which a compound is detected whereas charm analysis alsotakes into account the time for which the odor is perceived [110].AEDA has also been applied to the study of numerous food products, including olive

oil, butter, Swiss cheese, meat, bread, beer, green tea, dill herb, and off-flavors [118],and wines [119].

Fig. 2.6 Example of a “charm”

response chromatogram produ-

ced from the relationship

c ¼ dn�1, where d is the dilution

constant and n is the number of

coincident responses at any gi-

ven retention index (from [105])


2.6.3

Osme Method

The word “osme” given to this method [120] derives from Greek and means smell,hence the terms used above: “anosmia”, lack of sense of smell, and “normosmia”,normal sense of smell. In contrast to the two techniques described in Sec-tions 2.6.1 and 2.6.2, osme measures perceived odor intensity and is not based ondilutions to odor detection thresholds. The subject uses a time-intensity tracking pro-cedure to rate the intensity of each eluting odorant from the GC and, at the same time,provides verbal descriptions of the odor-active regions of the chromatogram [121].Similar to charm analysis and AEDA, retention times for the odor peaks are convertedinto standardized retention indices to confirm the chemical identity of the odorants. Insome cases, further confirmation is achieved by GC-MS [121].Variations on the specific procedure of time-intensity odor tracking, for example a

PCmouse moved on a 60 cm scale vs. a rheostat apparatus that measured finger span,were shown to make no significant difference to the odor peaks obtained [110]. Osmehas been applied to the analysis of wines [121] and hop oils and beers [122].

Acknowledgments

Preparation of this article was supported by research grant number R01DC02741from the National Institute on Deafness and Other Communication Disorders, Na-tional Institutes of Health, and by the Center for Indoor Air Research.

References

1 B. G. Green, J. R. Mason, M. R. Kare.In Chemical Senses. Vol. 2: Irritation (Ed.:B. G. Green, J. R. Mason, M. R. Kare),Marcel Dekker, Inc., New York, 1990,v–vii.

2 B. G. Green, H. T. Lawless. In Smelland Taste in Health and Disease (Ed.:T. V. Getchell, R. L. Doty, L. M. Bartoshuk,J. B. Snow Jr.), Raven Press, New York,1991, 235–253.

3 G. H. Parker. J. Acad. Nat. Sci. Phila. 1912,15, 221–234.

4 C. A. Keele. Arch. Int. Pharmacodyn. Ther.1962, 139, 547–557.

5 J. E. Cometto-Muniz. In Indoor Air QualityHandbook (Ed.: J. D. Spengler, J. Samet,J. F. McCarthy), McGraw-Hill, New York,2001, 20.1–20.21.

6 J. E. Cometto-Muniz, W. S. Cain,M. H. Abraham. Exp. Brain Res. 1998, 118,180–188.

7 F. A. Fazzalari, ed. Compilation of odor andtaste threshold value data. Baltimore:

American Society for Testing andMaterials; 1978.

8 Odor thresholds for chemicals with establishedoccupational health standards. AmericanIndustrial Hygiene Association. 1989.

9 L. J. van Gemert. Compilations of odourthreshold values in air and water. TNONutrition and Food Research Institute.1999.

10 M. Devos, F. Patte, J. Rouault, P. Laffort,L. J. van Gemert, eds. Standardized HumanOlfactory Thresholds. Oxford:IRL Press;1990.

11 K. J. Rossiter. Chem. Rev. 1996, 96,3201–3240.

12 M. Chastrette. SAR QSAR Environ. Res.1997, 6, 215–254.

13 M. H. Abraham. In Indoor Air and HumanHealth. 2nd Edition (Ed.: R. B. Gammage,B. A. Berven), CRC Lewis Publishers, BocaRaton, 1996, 67–91.

14 A. Dravnieks. In Flavor Quality: ObjectiveMeasurement (ACS Symposium Series,


No. 51) (Ed.: R. A. Scanlan), AmericanChemical Society, 1977, 11–28.

15 P. Laffort, F. Patte. J Chromatogr. 1987, 406,51–74.

16 P. A. Edwards, P. C. Jurs. Chem. Senses1989, 14, 281–291.

17 L. S. Anker, P. C. Jurs, P. A. Edwards. Anal.Chem. 1990, 62, 2676–2684.

18 P. A. Edwards, L. S. Anker, P. C. Jurs. Chem.Senses 1991, 16, 447–465.

19 L. M. Egolf, P. C. Jurs. Anal. Chem. 1993, 65,3119–3126.

20 Y. Alarie. CRC Crit Rev Toxicol 1973, 2,299–366.

21 G. D. Nielsen. CRC Crit. Rev. Toxicol. 1991,21, 183–208.

22 J. E. Cometto-Muniz, W. S. Cain. Ann.N. Y. Acad. Sci. 1992, 641, 137–151.

23 J. C. Stevens, W. S. Cain, R. J. Burke. Chem.Senses 1988, 13, 643–653.

24 W. S. Cain, J. F. Gent. J. Exp. Psychol.: Hum.Percep. & Perform. 1991, 17, 382–391.

25 J. E. Cometto-Muniz, W. S. Cain. Physiol.Behav. 1990, 48, 719–725.

26 W. S. Cain, J. E. Cometto-Muniz,R. A. de Wijk. In Science of Olfaction (Ed.:M. J. Serby, K. L. Chobor), Springer-Verlag,New York, 1992, 279–308.

27 W. S. Cain. Ear Nose Throat J. 1989, 68,316–328.

28 J. E. Cometto-Muniz, W. S. Cain. In Smelland Taste in Health and Disease (Ed.:T. V. Getchell, R. L. Doty, L. M. Bartoshuk,J. B. Snow Jr.), Raven Press, New York,1991, 765–785.

29 J. E. Cometto-Muniz, W. S. Cain. Phar-macol. Biochem. Behav. 1991, 39, 983–989.

30 J. E. Cometto-Muniz, W. S. Cain. Arch.Environ. Health 1993, 48, 309–314.

31 J. E. Cometto-Muniz, W. S. Cain. Am. Ind.Hyg. Assoc. J. 1994, 55, 811–817.

32 N. P. Franks, W. R. Lieb. Nature 1985,316, 349–351.

33 N. P. Franks, W. R. Lieb. Environ. HealthPerspect. 1990, 87, 199–205.

34 J. E. Cometto-Muniz, W. S. Cain,M. H. Abraham, J. M. R. Gola. J. Appl.Toxicol. 2002, 22, 25–30.

35 J. E. Cometto-Muniz, W. S. Cain,M. H. Abraham, J. M. R. Gola. Physiol.Behav. 1999, 67, 269–276.

36 J. E. Cometto-Muniz, W. S. Cain,M. H. Abraham, J. M. R. Gola. Toxicol. Sci.2001, 63, 233–244.

37 D. G. Guadagni, R. G. Buttery,S. Okano, H. K. Burr. Nature 1963, 200,1288–1289.

38 M. Q. Patterson, J. C. Stevens, W. S. Cain,J. E. Cometto-Muniz.Chem. Senses 1993, 18,723–734.

39 J. E. Cometto-Muniz, W. S. Cain,H. K. Hudnell. Percept. Psychophys. 1997,59, 665–674.

40 A. A. Rosen, J. B. Peter, F. M. Middleton.J. Water Pollut. Control Fed. 1962, 34, 7–14.

41 R. A. Baker. J. Water Pollut. Control Fed.1963, 35, 728–741.

42 M. Laska, R. Hudson. Chem. Senses 1991,16, 651–662.

43 B. Berglund, M. J. Olsson. Percept.Psychophys. 1993, 53, 475–482.

44 W. S. Cain, F. T. Schiet, M. J. Olsson,R. A. de Wijk. Chem. Senses 1995, 20,625–637.

45 K.-O. Schnabel, H.-D. Belitz,C. von Ranson. Z Lebensm Unters Forsch1988, 187, 215–223.

46 M. Chastrette, J. Y. d. Saint Laumer. Eur.J. Med. Chem. 1991, 26, 829–833.

47 G. M. Dyson. Chem. Ind. 1938, 57,647–651.

48 R. H. Wright. J. Theor. Biol. 1977, 64,473–502.

49 L. Turin. Chem. Senses 1996, 21, 773–791.50 K. M. Hau, D. W. Connell. Indoor Air 1998,

8, 23–33.51 N. M. Shvets, A. S. Dimoglo.Nahrung 1998,

42, 364–370.52 J. Muller, G. Greff. Food Chem. Toxicol.

1984, 22, 661–664.53 D. W. Roberts. Chem. Biol. Interactions

1986, 57, 325–345.54 G. D. Nielsen, E. S. Thomsen, Y. Alarie.

Acta Pharmacol. Nord. 1990, 1, 31–44.55 G. D. Nielsen, L. F. Hansen, Y. Alarie.

In Chemical, microbiological, health andcomfort aspects of indoor air quality –State of the art in SBS (Ed.: H. Knoppel,P. Wolkoff), Kluwer Academic Publishers,Dordrecht, 1992, 99–114.

56 K. M. Hau, D. W. Connell, B. J. Richardson.Toxicol. Sci. 1999, 47, 93–98.

57 M. H. Abraham. Pure Appl. Chem. 1993, 65,2503–2512.

58 M. H. Abraham. Chem. Soc. Rev. 1993, 22,73–83.

59 M. H. Abraham, G. S. Whiting, Y. Alarie,J. J. Morris, P. J. Taylor, R. M. Doherty,


R. W. Taft, G. D. Nielsen. Quant. Struct.-Act. Relat. 1990, 9, 6–10.

60 M. H. Abraham, G. D. Nielsen, Y. Alarie.J. Pharm. Sci. 1994, 83, 680–688.

61 M. H. Abraham, J. M. R. Gola, J. E.Cometto-Muniz, W. S. Cain. Chem. Senses2002, 27, 95–104.

62 M. H. Abraham, J. Andonian-Haftvan,J. E. Cometto-Muniz, W. S. Cain. Fundam.Appl. Toxicol. 1996, 31, 71–76.

63 M. H. Abraham, R. Kumarsingh, J. E.Cometto-Muniz, W. S. Cain. Arch. Toxicol.1998, 72, 227–232.

64 M. H. Abraham, R. Kumarsingh, J. E.Cometto-Muniz, W. S. Cain, M. Roses,E. Bosch, M. L. Dıaz. J. Chem. Soc. PerkinTrans. 2 1998, 2405–2411.

65 J. E. Cometto-Muniz, W. S. Cain,M. H. Abraham, R. Kumarsingh. Phar-macol. Biochem. Behav. 1998, 60, 765–770.

66 J. E. Amoore, B. G. Ollman.Rhinology 1983,21, 49–54.

67 J. E. Cometto-Muniz, W. S. Cain,T. Hiraishi, M. H. Abraham, J. M. R. Jola.Chem. Senses 2000, 25, 285–291.

68 C. A. Elsberg, I. Levy. Bull. Neurol. Inst.N. Y. 1935, 4, 5–19.

69 B. M. Wenzel. Psychol. Bull. 1948, 45,231–247.

70 D. G. Laing. Perception 1983, 12, 99–117.71 A. Dravnieks. In Methods in Olfactory

Research (Ed.: D. G. Moulton, A. Turk,J. W. J. Johnston), Academic Press,New York, 1975, 1–58.

72 H. R. Moskowitz, A. Dravnieks, W. S. Cain,A. Turk. Chem. Senses Flavor 1974, 1,235–237.

73 ASTM, E 544.Recommended practice for odorsuprathreshold intensity referencing, Am. Soc.Test. Materials, Philadelphia: 1975.

74 A. Dravnieks, W. H. Prokop. J. Air Pollut.Control Assoc. 1975, 25, 28–35.

75 A. Dravnieks, W. H. Prokop, W. R.Boehme. J. Air Pollut. Control Assoc. 1978,28, 1124–1130.

76 W. S. Cain, B. P. Leaderer. Environ. Int.1982, 8, 505–514.

77 G. Kobal. Pain 1985, 22, 151–163.78 T. Hummel, M. Knecht, G. Kobal. Brain

Res. 1996, 717, 160–164.79 G. Kobal, C. Hummel. Electroenceph. clin.

Neurophysiol. 1988, 71, 241–250.

80 J. C. Walker, D. B. Kurtz, F. M. Shore,M. W. Ogden, J. H. I. Reynolds. Chem.Senses 1990, 15, 165–177.

81 D. Warren, J. C. Walker, A. F. Drake,R. Lutz. Physiol. Behav. 1992, 51, 425–430.

82 D. Warren, J. C. Walker, A. F. Drake,R. Lutz. Laryngoscope 1994, 104, 623–626.

83 M. Kendal-Reed, J. C. Walker,W. T. Morgan, M. LaMacchio, R. W. Lutz.Chem. Senses 1998, 23, 71–82.

84 D. G. Laing. Perception 1982, 11, 221–230.85 D. G. Laing. Physiol. Behav. 1985, 34,

569–574.86 W. S. Cain, B. P. Leaderer, R. Isseroff,

L. G. Berglund, R. J. Huey, E. D. Lipsitt,D. Perlman. Atmos. Environ. 1983, 17,1183–1197.

87 W. S. Cain, T. Tosun, L.-C. See, B. Leade-rer. Atmos. Environ. 1987, 21, 347–353.

88 G. H. Clausen, P. O. Fanger, W. S. Cain,B. P. Leaderer. In Indoor Air. Volume 3.Sensory and Hyperreactivity Reactions to SickBuildings (Ed.: B. Berglund, T. Lindvall,J. Sundell), Swedish Council for BuildingResearch, Stockholm, 1984, 437–441.

89 J. C. Walker, P. R. Nelson, W. S. Cain,M. J. Utell, M. B. Joyce, W. T. Morgan,T. J. Steichen, W. S. Pritchard,M. W. Stancill. Indoor Air 1997, 7,173–188.

90 G. H. Clausen, P. O. Fanger, W. S. Cain,B. P. Leaderer. Environ. Int. 1986, 12,201–205.

91 H. K. Hudnell, D. A. Otto, D. E. House,L. Mølhave. Arch. Environ. Health 1992, 47,31–38.

92 S. K. Kjærgaard, L. Mølhave, O. F. Peder-sen. Atmos. Environ. 1991, 25A,1417–1426.

93 L. Mølhave, B. Bach, O. F. Pedersen.Environ. Int. 1986, 12, 167–175.

94 L. Mølhave, J. Grønkjær Jensen, S. Larsen.Atmos. Environ. 1991, 25A, 1283–1293.

95 L. Mølhave. Ann. N. Y. Acad. Sci. 1992, 641,46–55.

96 D. Otto, L. Mølhave, G. Rose,H. K. Hudnell, D. House. Neurotoxicol.Teratol. 1990, 12, 649-652.

97 F. T. Schiet, W. S. Cain. Perception 1990, 19,123–132.

98 W. S. Cain, L. C. See, T. Tosun. In IAQ’86.Managing Indoor Air for Health and EnergyConservation (Ed.: American Society ofHeating, Refrigerating and Air-Conditio-


ning Engineers, Inc., Atlanta, Georgia,USA, 1986, 126–137.

99 A. T. James, A. J. P. Martin. Biochem.J. 1952, 50, 679–690.

100 T. E. Acree. Anal. Chem. News & Features1997, 170A–175A.

101 Y.-W. Feng, T. E. Acree. Foods Food Ingre-dients J. Jpn. 1999, 179, 57–66.

102 G. H. Fuller, R. Steltenkamp, G. A.Tisserand. Ann. N.Y. Acad. Sci. 1964,116, 711–724.

103 A. Dravnieks, A. J. O’Donnell. J. Agric. FoodChem. 1971, 19, 1049–1056.

104 T. E. Acree, R. M. Butts, R. R. Nelson,C. Y. Lee. Anal. Chem. 1976, 48, 1821–1822.

105 T. E. Acree, J. Barnard, D. G. Cunningham.Food Chem. 1984, 14, 273–286.

106 P. Pollien, L. B. Fay, M. Baumgartner,A. Chaintreau. Anal. Chem. 1999, 71,5391–5397.

107 K. D. Deibler, T. E. Acree, E. H. Lavin.J. Agric. Food Chem. 1999, 47, 1616–1618.

108 R. T. Marsili, N. Miller. J. Chromatogr. Sci.2000, 38, 307–314.

109 T. E. Acree, J. Barnard. In Trends in FlavourResearch (Ed.: H. Maarse, D. G. vander Heij), Elsevier, Amsterdam, 1994,211–220.

110 H. Guichard, E. Guichard, D. Langlois,S. Issanchou, N. Abbott.Z. Lebensm. Unters.Forsch. 1995, 201, 344–350.

111 D. G. Cunningham, T. E. Acree, J. Barnard,R. M. Butts, P. A. Braell. Food Chem. 1986,19, 137–147.

112 P. A. Braell, T. E. Acree, R. M. Butts,P. G. Zhou in Biogeneration of Aromas (Ed.:T. H. Parliment, R. Croteau), AmericanChemical Society, Washington, DC, 1986,75–84.

113 T. E. Acree, E. H. Lavin, R. Nishida,S. Watanabe in Flavour Science and Tech-nology (Ed.: Y. Bessiere, A. F. Thomas),Wiley & Sons, Geneva, 1990, 49–52.

114 A. B. Marin, T. E. Acree, J. H. Hotchkiss,S. Nagy. J. Agric. Food Chem. 1992, 40,650–654.

115 A. Bravo, J. H. Hotchkiss, T. E. Acree.J. Agric. Food Chem. 1992, 40, 1881–1885.

116 F. Ulrich, W. Grosch. Z. Lebensm. Unters.Forsch. 1987, 184, 277–282.

117 W. Grosch. Trends Food Sci. Technol. 1993,4, 68–73.

118 W. Grosch. Flavour Fragr. J. 1994, 9,147–158.

119 Y. Kotseridis, A. Razungles, A. Bertrand,R. Baumes. J. Agric. Food Chem. 2000, 48,5383–5388.

120 M. R. McDaniel, R. Miranda-Lopez,B. T. Watson, N. J. Micheals, L. M. Libbey.In Flavor and Off-flavors (Proceedings of the6th International Flavor Conference) (Ed.:G. Charalambous), Elsevier Science,Amsterdam, 1990, 23.

121 R. Miranda-Lopez, L. M. Libbey,B. T. Watson, M. R. McDaniel. J. Food Sci.1992, 57, 985–993, 1019.

122 N. Sanchez, C. L. Lederer, G. Nickerson,L. M. Libbey, M. R. McDaniel. In FoodScience and Human Nutrition (Ed.:G. Charalambous), Elsevier Science,Amsterdam, 1992.


3

Odor Handling and Delivery Systems

Takamichi Nakamoto

Abstract

A handling and delivery system significantly contributes to the capability and reliabil-ity in an odor sensing system. Various techniques of the sample flow, static, and pre-concentrator systems are described in the present chapter. The sample flow system isconvenient because the measurement cycle is short and easy to handle. The staticsystem is the basic one used to measure the steady-state sensor response. A precon-centrator is often used to enhance the sensitivity, and can be also used to autono-mously enhance the selectivity of a sensor array. Direct exposure of the sensor tothe vapor is sometimes used in field measurement. The analysis of the transient sen-sor response, a homogeneous sensor array for an olfactory video camera, and thesensor responses in the plume-tracing robot are briefly introduced. Due to the varietyof methods available, the most appropriate odor handling and delivery system shouldbe selected for the project.

3.1

Introduction

There are two main types of odor handling and delivery, the sample flow system andthe static system. In the sample flow system the sensors are placed in the vapor flow,which allows the rapid exchange of vapor and hence many samples can be measuredwithin a short time. In the static system there is no vapor flow around the sensor, andmeasurements are usually made on the steady-state responses of the sensors exposedto vapor at a constant concentration. The sample flow and static systems are closedunits. In a third method the direct exposure to the vapor is described, which is anopen system having no sensor chamber, hence rapid concentration change aroundthe sensors is measured. Three examples are given.


5555

3.2

Physics of Evaporation

Most of the samples tested in an odor sensing system are liquids from which odorantsare evaporated. It is therefore important to know the physicochemical behavior ofevaporation when you design an odor handling and delivery system. One of themost important points is that saturated vapor pressure is dependent on tempera-ture. The vapor concentration should be kept below the maximum correspondingto the saturated vapor pressure, otherwise the excess of the vapor pressure abovethe saturated point leads to its condensation into liquid drops. The relationship be-tween saturated vapor pressure P and temperature T is:

lnðPÞ ¼ c=T þ d ð1Þ

where c and d are constants. Vapor pressures at several temperatures are summarizedin the literature [1]. Let P3 and P4 be the saturated vapor pressures at T3 and T4, re-spectively. The constants are different for different vapors and can be determined fromEqs. (2) and (3).

c ¼ lnðP3Þ � lnðP4Þ1

T3

� 1

T4

ð2Þ

and

d ¼ T4 lnðP4Þ � T3 lnðP3ÞT4 � T3

: ð3Þ

The saturated vapor can therefore be obtained at arbitrary temperatures. The pressureof a compound with high odor intensity is typically small whereas highly volatile com-pounds have high saturated vapor pressures.When there is a mixture of compounds, the phenomenon of the vapor-liquid equi-

librium state becomes a little complicated. In ideal solutions, Raoult’s law expressedas:

PA ¼ NAP0A ð4Þ

is valid. PA is partial pressure of compound A, NA the molar ratio of that compound inthe solution, P0

A the vapor pressure of the pure compound. Equation (4) indicates thatthe partial pressure of the ideal solution is equal to the product of its molar ratio andthe vapor pressure of the pure compound. In the ideal solution, the superpositiontheorem for the plural compounds is valid.Most compounds, however, are non-ideal solutions. In the non-ideal solution,

Eq. (4) is replaced by:

PA ¼ cANAP0A ð5Þ

3 Odor Handling and Delivery Systems56

where cA is an activity coefficient dependent upon NA. Since interaction between thecomponents occurs in the non-ideal solution, the superposition theorem for the com-pound mixture is not valid. In that case, the equation derived by Wilson is more sui-table [2].

3.3

Sample Flow System

The sample flow system is the most popular odor handling and delivery system. Sev-eral sample flow systems exist such as headspace sampling, diffusion, permeation,and bubbler, and sampling bag methods are described.

3.3.1

Headspace Sampling

Figure 3.1 shows a schematic diagram of a headspace sampling method. The head-space is the space just above the liquid sample in a bottle. The carrier gas such asdry air is supplied at the inlet and the vapor evaporated at the liquid surface carriedby the carrier gas is supplied to the sensors. Solenoid valves alternately switch the purecarrier gas and the headspace sample vapor, and the difference in the sensor output isrecorded. The frequency shift from that in air to that in the sample vapor is regarded asthe sensor response in the case of a quartz crystal microbalance (QCM) gas sensor. Asemiconductor gas sensor response of a ratio of the resistance in the sample vapor tothat in the air. The distance between the liquid surface and the tips of the syringeneedles should be kept constant since the vapor in the headspace is often unsaturatedand its concentration varies according to its distance.The headspace sampling method is an easy method to use as described in the lit-

erature [3–7]. Although many samples can be measured within a short time, the sup-plied vapor concentration is not known and varies during the vapor supply. The vaporconcentration at the outlet of the bottle gradually changes until it reaches the liquid-vapor equilibrium as is illustrated in Fig. 3.2(a). The vapor-concentration profile some-times influences the waveform of the sensor response, which is a convolution of itsprofile and the sensor impulse response. When a sufficiently narrow vapor pulse, asshown in Fig. 3.2(b), is supplied to the sensor, the sensor response is not influenced by

Fig. 3.1 Headspace sampling


the concentration profile. The pulse width of the vapor should be sufficiently smallerthan the sensor response/recovery time. The pulse vapor supply method can thereforebe used to ignore the influence of the concentration variation during vapor supply [7].The example of the headspace samplingmethod is shown in detail in Fig. 3.3 [8]. Dry

air is supplied to the sample bottle through a mass flow controller (MFC). MFC is usedto precisely control the flow rate independently of its pressure load. The sample bottleis a vial of volume 22 mL. Liquid samples, such as perfumes and flavors, with thetypically constant volume of 4 mL are poured into the vials using a micropippete.The vials are placed in a temperature-controlled bath to avoid temperature inducedvariations of vapor pressure.The syringe needles are driven by an autosampling stage that moves from vial to vial

allowing many samples to be measured automatically. Care should be taken that thearm of the stage is not deformed when the syringe needles pierce the rubber seal(septum) of the vial, because the distance between the liquid surface and the tipsof the needles must be kept constant. The dry air and the sample vapor are alternatelyswitched by miniature solenoid valves controlled by a computer and supplied to asensor cell a sensor housing. A solenoid valve with a small internal volume is recom-mended, and is often driven by DC voltage of 12 V or 24 V. An internal solenoid wallmade of Teflon prevents odorant adsorption; additionally it is preferable to repeatedlyand quickly switch the solenoid valve for a few minutes after each measurement toensure it is kept clean. The drive circuit simply consists of a discrete transistor. Zenordiodes or surge absorbers are often used to suppress the surge generated when thevalve is switched. The input of the circuit can be connected to the printer port of acomputer.The sensor cell is one of the most important parts in an odor-handling system, its

structure determining the response time. The sensor response is sometimes influ-enced by its position within the cell, especially when dense vapor with a high boiling

Fig. 3.2 Concentration at outlet of headspace sampler,

(a) long duration of flowing carrier gas and (b) vapor pulse


point is supplied. The cell should therefore have an internal volume as small as pos-sible to minimize any effect due to the sensor location. The sensor response however,is sometimes slow enough to ignore that effect. There are several types of flow cells asshown in Figs. 3.4(a) and (b).The structure illustrated in Fig. 3.4 (a) is easy to fabricate because the flow rate at

each sensor is the same. The sensor response is independent of location when placedin parallel as is illustrated in Fig. 3.4(b) [9]. Additionally the same flow rate at eachsensor can be guaranteed by having a symmetrical structure. A more sophisticatedstructure with a tiny internal volume is shown in reference [7]. The material of the

Fig. 3.3 Headspace measurement system using autosampling stage

Fig. 3.4 Structures of sensor cells for a sensor array.

(a) series type and (b) parallel type


sensor cell is typically stainless. A type of rubber having low odor adsorption should beused to prevent gas leaking when the sensor is attached to the sensor cell.If the sensor responses are likely to be influenced by temperature, the sensor cell can

be kept at constant temperature by use of a peltier device, or immersion in a thermobath. In the example in Fig. 3.3, the water from the temperature-controlled bath usedfor the sample bottles is circulated around the sensor cell.

3.3.2

Diffusion Method

In this method diffusion of vapor from a tube of accurately known dimensions ismeasured [10]. Low concentrations are usually measured, because it is difficult toobtain vapor with concentration more than a few percent above saturation usingthis method. An example apparatus is shown in Fig. 3.5.The liquid in the reservoir is allowed to evaporate and the vapor slowly diffuses from

a reservoir through the diffusion tube into a flowing gas stream at a constant rate. Theresultant mixture concentration is determined by the ratio of the diffusion rate to thatof the flowing gas stream. The reservoir filled with liquid is kept at constant tempera-ture since the diffusion coefficient of the vapor depends upon the temperature.The diffusion rate is given by:

S ¼ D M P A

R T Lln

P

P � pð6Þ

where S is the rate of diffusion of vapor out of the capillary tube (g/ml),M is the relativemolecular mass of the vapor (g/mol), P the pressure in the diffusion cell at the openend of the capillary (atm), A the cross-sectional area of the tube (cm2), D the diffusioncoefficient (cm2/s), R the molar gas constant (mL atmmol�1 K�1), T temperature (K), L

Fig. 3.5 Apparatus for diffusion method


the length of the capillary tube (cm), and p is the partial pressure of the sample vapor(atm). The actual concentration sometimes deviates from Eq. (6) when the vapor abovethe liquid is not saturated. An alternative method is to precisely measure the masschange of the liquid reservoir during the constant period using a balance. The reduc-tion in the amount of liquid over a certain time indicates the diffusion rate. It is areliable method in spite of the time taken. Several researchers in the EN field usethe standard gas generator based upon the diffusion method because it is commer-cially available. Examples of sensor systems including diffusionmethods are describedin the references [11, 12].

3.3.3

Permeation Method

The permeation method is similar to the diffusion method, using similar equipmentexcept that a permeation tube is used (Fig. 3.6). Liquefied gas, when enclosed in aninert plastic tube, may escape by dissolving in and permeating though the walls of thetube. The permeation rate is proportional to the length of the tube and varies logarith-mically with 1/T, hence temperature should be kept constant. Permeation tubes ofseveral kinds of vapors are commercially available.

3.3.4

Bubbler

A bubbler is a bottle in which a vapor is generated by bubbling, as illustrated in Fig. 3.7.A carrier gas such as air is passed through the liquid in the bottle, and takes away thegenerated vapor. Although it is easy to obtain the vapor by this method, several pointsshould be taken into account. The headspace over the liquid sample sometimes doesnot saturate. Glass particles are sometimes put in the liquid so that the area of contactbetween the liquid and the carrier gas can be increased. Moreover, tiny liquid particles,not vaporized ones, are sometimes carried to the sensors due to heavy bubbling at afast flow rate. Examples of the bubblers are given in references [13, 14].

Fig. 3.6 Permeation tube


3.3.5

Method using a Sampling Bag

Vapor is generated after a liquid sample is injected to the large-air-volume samplingbag by syringe, and then evaporated. The vapor in the bag is then sucked out using apump and introduced into a sensor cell, as illustrated in Fig. 3.8(a). The vapor con-centration is determined by the combined volume of the injected liquid and thatof the sampling bag. MFCs are used if a vapor blender is constructed. The concentra-tion of each vapor is determined by the ratio of the corresponding flow rate to the totalflow rate.The sample flow system in Fig. 3.8(a) is a simplified one, the actual system has

solenoid valves to switch the vapor abruptly [15]. It takes a little time for the MFCflow rate to settle to the set point value. The baseline of the sensor response dependson the flow rate, which should therefore be kept constant. Amore sophisticated systemis illustrated in Fig. 3.8(b). The valves V1 and �VV1, V2 and �VV2 are complementarilyswitched. The flow rates of both the vapor from the blender and the air are alwaysthe same and constant. Each flow path to either the sensor cell or the bypass isabruptly switched by the solenoid valves without changing the flow rate at the sensorcell. The air in the vapor blender is used to keep the flow rate constant at the outlet ofthe vapor blender.The material of the sampling bag should be carefully selected to avoid water and

other molecules permeating through. The generation of vapors can also occur in cer-tain types of plastic bag. Adsorption of the sampled vapor inside the bag cannot beignored in case of low concentration. A fluorine-containing resin bag is the bestone due to low permeability and low adsorption capability. A glass vessel is oftenused to sample the atmosphere in environmental testing, and so requires careful hand-ling so that it cannot be broken – it is also expensive. Systems using several MFCs arereported because vapor with an arbitrary concentration is automatically and rapidlygenerated [16]. They are convenient in spite of the fact that they are expensive.

Fig. 3.7 Bubbler


Fig. 3.8 Vapor supply method using a sampling bag

(a) simplified method and (b) actual method


3.4

Static System

The fundamental static system measures the steady-state response of a sensor to avapor at constant concentration and at a constant temperature. In the case of aQCM sensor, the most basic characteristic such as a partition coefficient of a sensingfilm, defined as the ratio of the concentration in the film to that in the vapor, can beobtained in this system.The principle is illustrated in Fig. 3.9. The tiny volume, typically a fewmicroliters, of

liquid sample is injected into a chamber having a volume of a few liters, and is eva-porated. The sensor response is measured after equilibrium is reached [17–24]. Thechamber is typically made of Teflon or glass to avoid vapor adsorption onto the internalwall. The whole chamber can be immersed in a temperature-controlled bath, thus thewhole system can be kept at the same temperature; in the sample flow system thetemperature at the sensor sometimes does not agree with that of the vapor.Manual injection of the sample liquid by the syringe is the basic method, however it

is possible to automate this procedure [25]. Because the volume of the plumbing tubecannot be ignored, a technique similar to FIA (Flow Injection Analysis) is used tosample a few microliters of the liquid precisely. The automated system consists ofa sample selector, a sample injector, and the measurement system. It selects the sam-ples among several candidates, injects the sample liquid and measures the sensorresponses after equilibrium. Since it takes time to measure the steady-state responsedue to the slow evaporation of the sample liquid, the automation is quite indispensableif many data need to be systematically measured.

Fig. 3.9 Principle of the static measurement system


3.5

Preconcentrator

3.5.2

Sensitivity Enhancement

A preconcentrator tube is often used to enhance the sensitivity of the sensor [26, 27].After it accumulates the vapor, a heat pulse is applied to the tube to desorb the con-centrated vapor, and the limit of detection is thus improved. Although it originatesfrom the technique of gas chromatography, it is often used for sensors.A simplified preconcentrator system is illustrated in Fig. 3.10. A preconcentrator

stainless tube with a length of a few cm is packed with adsorbent such as Tenax-TA. The adsorbent of a few tens of milligrams is held in place with glass wool.The tube is heated using a coil of insulated nichrome wire around it in case of thermaldesorption. A temperature controller is used to adjust the power supplied to the heaterso that the temperature given by a computer can be maintained.The typical temperature during heating is around 200 8C, hence the use of metal

connectors for plumbing the preconcentrator is recommended. However, a certainheatproof flexible tube is available for plumbing it if the heat capacity of the connec-tors cannot be ignored.The temperature characteristic varies in different preconcentrators since it is diffi-

cult to reproducibly wind up the heating coil. The gap between the coil and the stain-less tube is critical. A flexible heater stuck to the preconcentrator tube is preferable.

Fig. 3.10 Simplified system of preconcentrator


Another way is to heat the preconcentrator tube directly by flowing the current throughthe surface of the tube. The tube diameter should be minimized to increase the heaterresistance for direct heating. There is a tradeoff between the heater resistance and theflow rate.There are several adsorbents for the preconcentrator. They are Tenax TA, Tenax GR,

Carbopack B, Carbotrap, Carboxen 569, Carbosieve SIII etc [28]. Some of them arepolar and some are nonpolar. The selection of the adsorbent should be accordingto the purpose. Some kinds of adsorbents can be used as coating films of QCMgas sensors because they are dissolved into an organic solvent. However, its charac-teristic seems to be different from that of particles. The sensor response is slow butcannot accumulate the vapor in the film.

3.5.2

Removal of Humidity

Hydrophobic adsorbents do not capture water. Water just goes though the preconcen-trator tube whereas other vapor molecules are accumulated. It is possible to desorb thevapors without water at the heating stage after passing the sample to the tube. Sincemany sensor responses are affected by humidity change, it is convenient to remove thehumidity before the actual measurement. Many samples such as juice, soup, and cof-fee include water. The removal of the water influence is indispensable for reliablemeasurement from the practical point of view. The removal of alcohol is sometimesrequired for alcoholic beverages such as beer, whiskey, liquor, wine etc. The influenceof alcohol is critical especially for semiconductor gas sensors since the contributions ofother components are masked by the alcohol.It is better to keep the preconcentrator temperature a little higher than room tem-

perature even during the adsorption process so that the removal of water can be com-pletely performed [29]. Moreover, a slightly complicated sequence is sometimes usedto avoid the exposure of the sensor to the unwanted vapor (water and/or ethanol) incases when it would take much time to recover from the response to that vapor. How-ever, the most simple and basic system of the preconcentrator is that shown inFig. 3.10.

3.5.3

Selectivity Enhancement by Varying Temperature

3.5.3.1 Selectivity Enhancement using a Preconcentrator

In addition to sensitivity increase, it is possible to enhance the selectivity of samples byusing a preconcentrator. There are two ways to enhance the selectivity. First is to utilizechromatographic behavior when the gases pass through the preconcentrator tube. Thesecond is to separate samples by varying the desorption temperature since that tem-perature changes from compound to compound.In the first method, the chromatographic behavior is observed [30]. The samples

interact with the adsorbent and the degree of the interaction depends upon the sample


type. When the interaction is strong, it takes time for the sample to elute at the exit ofthe preconcentrator tube, in the same manner as in gas chromatography, whereas itdoes not takemuch time for sample elution in the case of low interaction. Although theseparation of the samples at the exit of the preconcentrator is not sufficient, the re-tention time is just within a few tens of seconds and sample discrimination can beachieved by a sensor array and pattern recognition techniques. The eluted samplesfrom the preconcentrator tube is detected by a sensor array made up of multiple sen-sors with different characteristics, and with its output pattern recognized by a neuralnetwork or multivariate analysis. It can be regarded as a kind of higher-order sensingtechnique [31] since the information is included in both transient waveform of eachsensor and output pattern from the sensor array.The second method of selectivity enhancement using the preconcentrator tube is to

vary the temperature for the vapor desorption [32, 33]. When the vapor accumulated inthe preconentrator tube is desorbed by raising its temperature, each vapor will desorbat its own particular desorption temperature. Although it depends upon its boilingpoint, other factors such as polarity seem to influence it. When the temperature ofthe preconentrator tube is changed several times, the vapor with low desorption tem-perature appears due to the small temperature increase, whereas the one with highdesorption temperature comes to the sensors due to the large temperature increase.Thus, the adequate sequence of the preconcentrator heating improves the selectivity.The abrupt temperature change is preferable to a ramp shape of the temperature pro-file because a sharp sensor response is obtained at the point of the abrupt temperaturechange.

3.5.3.2 Autonomous System with Plasticity

It is possible to autonomously obtain the heating sequence according to the samplesdesorption temperatures. The pattern separation among the samples is improved afterthe optimization of the heating sequence. Since the number of odor types is huge andthere are too many parameters to be optimized, such autonomous behavior is helpfulto achieve good capability of discrimination for a short time. It is a kind of activesensing system [34] since the sensing system itself enhances its capability autono-mously through interaction with the targets. The flexible and accurate system canbe realized based upon that concept, compared with the conventional passive sensingsystem. This concept was first realized in the semiconductor gas sensing system and iscalled a characteristic of plasticity [11]. Plasticity is the biological capability, e.g. synap-tic modification, to organize in such a way as to adapt to an environment. The char-acteristic of plasticity can be realized in an odor sensing system only when three fun-damental technologies are available. They are the gas sensor device with its character-istic easily changed by a controllable parameter, the evaluation index of the adaptation,and the algorithm for changing the parameters. In the preconcentrator system, the gassensor device mentioned above is the preconcentrator tube with variable desorptiontemperature in addition to a sensor. Its characteristic is easily controlled by the voltageapplied to the preconcentrator heater.


The second fundamental technology in the preconcenetrator system is the index ofthe pattern separation among the target samples. The sensor-array output pattern canbe regarded as a vector with each component corresponding to each sensor response.Several pattern separation indices for the output pattern from a sensor array such asthe Euclidean distance in that vector space, the vector angles, Maharanobis distance,Wilks Lambda [35] are available. The simple indices such as the Euclidean distance andthe vector angle are preferable when the number of the data is insufficient.The third fundamental technology in the preconcentrator system is the algorithm

often called an optimization algorithm [36]. This is the algorithm used to obtain theappropriate parameter values by evaluating the index. The parameters are repeatedlymodified in the optimization process until a good evaluation index is obtained. Thereare several optimization algorithms such as the simplex method and method of stee-pest descent [15, 37]. In the preconcentrator system, the optimization process is illu-strated in Fig. 3.11. Since several heat pulses with different temperatures are repeat-edly applied to the preconcentrator tube at every measurement, the pattern separationindex is expressed as a function of those temperatures. The shape of the curved surfaceof the pattern separation index is not a priori known. The temperature profile is itera-tively modified to find the point with the maximum index. The exploration task isachieved using the optimization algorithm.In the case of the gradient method, the jth peak temperature at the iþ 1 step T ðiþ1Þ

j isdetermined by:

T ðiþ1Þj ¼ T ð0Þ

j þ e@I

Tj

ð7Þ

Fig. 3.11 Principle of realizing plasticity


where I is the pattern separation index, e the empirically determined constant. Thepoint 0 is that with the maximum index before the iþ 1 measurement. @I

@Tjaround the

point 0 can be approximately obtained using the data around that point [38]. The de-rivation of @I

@Tjis shown in the appendix.

3.5.3.3 Experiment on Plasticity

An example of autonomous enhancement of the selectivity is described here. Thepreconcentrator tube was packed with 30 mg Tenax-TA. Two samples were pure pro-pyl acetate and the mixture of propyl acetate and hexyl acetate (ratio 1:1 v/v). The head-space vapors were supplied to seven QCM (20 MHz, AT-CUT) sensors at the flow rateof 200 ml/min and the time for the vapor accumulation was 13 s. The sensor coatingswere Tenax-TA, UCON-90000, DEGS (Diethylene Glycol Succinate), squalane, sphin-gomyelin, ethylcellulose and PolyEthyleneGlycol (PEG) 1000. The heat pulses wereapplied three times and the final temperature peak was fixed to 230 8C. The tempera-ture profile can be expressed by the two parameters of the first and second peak tem-peratures. Since three heat pulses were applied during one cycle and the number of thesensors was seven, the dimension of the sensor response pattern was 21. The patternseparation index used was a vector angle between two samples. It was found that theindex was successfully improved after the temperature profile wasmodified five times.Two typical sensor responses extracted among seven sensors before and after tem-

perature-profile modification are shown in Fig. 3.12 (a) to (d). The sensors were QCMscoated with polar film DEGS and nonpolar film squalane. The first peaks were theresponses to the vapors not accumulated at the preconcentrator tube during the vaporsupply. The peaks at 110 s and 160 s were the responses caused by the second andthird heat pulses. Although the first heat pulse was applied at 50 s, the responsesat that point were small. It was found from the figures that the difference of the re-sponse pattern between the two samples became larger after the exploration of thetemperature profile. The difference appeared at the peak due to the third heatpulse. The two samples were easily discriminated after the optimization when theresponses at the third peak were taken into account. The plastic characteristic canbe realized using preconcentrator tube with variable temperature and optimizationalgorithm.


3.6

Measurement of Sensor Directly Exposed to Ambient Vapor

3.6.1

Analysis of Transient Sensor Response using an Optical Tracer

The flow-type and the static systems mentioned above are closed systems. It is alsopossible to directly expose a sensor to a gas. Direct exposure is often performedwhen the rapid concentration change in an open system should be captured. How-ever, the sensor response does not correspond to instantaneous gas concentrationdue to its response delay, even if it is open to the ambient atmosphere. The sensordynamics can be analyzed when both sensor response and gas concentration changeare simultaneously obtained [39, 40]. Thereafter, it is possible to model the sensordynamics.

Fig. 3.12 Sensor responses before and after modification (a) mixture

of propyl acetate and hexyl acetate before optimization, (b) propyl

acetate before modification, (c) mixture of propyl acetate and hexyl

acetate after modification, (d) propyl acetate after modification


The instantaneous gas concentration is approximately obtained using an opticaltracer because a response speed of an optical sensor is much faster than that ofthe gas sensor. One of the smart methods is to utilize the optical tracer accompaniedwith smell. The use of white smoke from joss sticks is a good candidate for that pur-pose. The gas concentration is measured as the brightness of the CCD camera image,and the transient response of the sensor to the smoke measured simultaneously. Thelight sheet is illuminated by a xenon lamp through a slit just above the gas sensor.When the smoke of the joss sticks flows, the gas sensor responds to the smoke. Si-multaneously, the light scattered by the smoke particles is captured by the camera. Theoptical data is sampled at the video rate and the brightness integration over the area ofthe gas sensor can be regarded as optical sensor response.The transient response of a semiconductor gas sensor was observed in a wind tunnel

[40]. In the case of the semiconductor gas sensor, the rise time of the response isdifferent from the recovery time. Thus, the transient response of the gas sensor can-not be modeled by a simple linear time-invariant system, and hence the two-phasemodel was proposed. In this model, the time-series data is divided into the responsephase and the recovery one.It is assumed that the gas concentration in each phase can be expressed by a second-

order differential equation:

d2sðtÞdt2

þ aidsðtÞdt

þ bisðtÞ ¼ gilsðtÞ ð8Þ

where s(t) is gas sensor response Rgas=Rair at time t, ai, bi, gi are constants (i ¼ 1:response phase, i ¼ 2: recovery phase). lsðtÞ is the steady-state sensor response calcu-lated from a calibration curve. Equation (8) is transformed into the discrete-time equa-tion.

sðkþ 1Þ ¼ pisðkÞ þ qisðk� 1Þ þ rilsðkÞ ð9Þ

where s(k) and lsðkÞ are gas sensor response and transformed steady-state sensor re-sponse corresponding to the brightness at time kDt. pi, qi, ri are the constants; i ¼ 1:response phase, i ¼ 2: recovery phase. Moreover, the following constraint is requiredsince sðkþ lÞ ¼ sðkÞ ¼ sðk� 1Þ ¼ lsðkÞ in the steady state.

pi þ qi þ ri ¼ 1 ði ¼ 1; 2Þ: ð10Þ

The scheme for dividing a time-series data into the two phases is as follows. If the gasconcentration increases rapidly and lsðkÞ becomes less than s(k), the gas sensor re-sponse begins to decrease toward lsðkÞ. Thus, the data at that moment can be regardedas the response-phase data. If lðkÞ > sðkÞ, the data at that moment can be regarded asthe recovery phase data in the same way. The parameters, pi, qi and ri are estimated forthe response phase and the recovery phase respectively using the least-squares me-thod.The response of the semiconductor gas sensor (TGS800, Figaro) is compared with

the calculated value based upon Eqs. (8) to (10), as shown in Fig. 3.13. The gas sensor


response calculated from the optical data was in good agreement with the experimentaldata. Since the accuracy depends upon the gap between the sensor and the light sheet,that gap should be minimized to obtain the high accuracy of the data.This is one of the methods used to model gas sensor behaviors. If the flow-type

system is used, the actual concentration profile at the sensor is sometimes differentfrom that at the vapor source. The exact waveform of the gas concentration is requiredto model the behavior of the gas sensor response. Since the gas concentration justabove the sensor is obtained in this method, it is beneficial for sensor-behavior mod-eling. When the gas sensor response is very fast, the modulation of the gas concen-tration by moving gas outlets was effective to estimate its time constant [41].

3.6.2

Homogenous Sensor Array for Visualizing Gas/Odor Flow

Another example of direct exposure of the sensor to ambient vapor is the visualizationsystem of the gas-concentration distribution. A two-dimensional homogeneous sensorarray can capture the dynamic scene of the gas flow as illustrated in Fig. 3.14(a). It iscalled an olfactory video camera since the dynamic distribution of the gas concentra-tion can be stored in a computer and can be played back in the same way as that of aconventional video. Knowing the direction of the gas flow from the dynamic image andsimultaneous measurement of the gas concentration at many points enhances thereliability of the gas-flow direction estimation, because the influence of the wind tur-bulence is large on the measurement data.After the initial experiment on the pulse drive semiconductor gas sensor array [42],

the 5� 5 QCM gas sensor array was fabricated [43]. The recovery time of a QCMgas sensor, which is quite essential in the gas flow visualization, is typically less

Fig. 3.13 Comparison of gas sensor response calculated from the

optical data with measured data


than 1 second. A miniaturized QCM (27.8 MHz, AT-CUT, SMD type) with internallyinstalled oscillator (8� 4 mm) was used. It was coated with phosphatidylcholine,which is relatively sensitive to triethyl amine used as a target vapor. It is one of thetypical bad smells and the system is expected to be applied to environmental moni-toring. The compact 25-channel frequency counter implemented into a FPGA(Field Programmable Gate Array) was used to measure their sensor responses everysecond. The 25-channel frequency data were transferred to a computer via RS232Cinterface.The experiment was conducted in a wind tunnel with wind speed less than 5 cm/s.

The headspace vapor of triethyl amine was spouted from a nozzle at the rate of 75 ml/min. The dynamic scene of triethyl amine behavior was captured by the olfactory videocamera as is shown in Fig. 3.14 (b). The images are for three successive seconds andare displayed as binary images to enhance the contrast. The flow direction was almostalways grasped using this system sincemany fragments of the odor cloud generated bythe turbulence come to the array and the vapor distribution is not uniform. The clearimage was obtained here because of quick response/recovery time of QCM gas sen-sors. Furthermore, the direction estimation was successfully performed using theimage processing algorithm [44] even if the instantaneous wind direction is not con-stant. Some gas sensors can be directly exposed to the vapor to obtain the instanta-neous concentration in the field with wind turbulence and the example of QCMgas sensor has been described in this subsection.

Fig. 3.14 Olfactory video camera (a) concept and

(b) binary image from olfactory video camera


3.6.3

Response of Sensor Mounted on an Odor-Source Localization System

The final example of the direct exposure of the sensor to the vapor is an odor-sourcelocalization system [45–47]. It is a plume-tracing robot that can locate the source of theodor using the gas sensors and the anemometric sensors. The gas sensorsmounted onthat robot were directly exposed to the vapor. Tin-oxide gas sensors (TGS822, Figaro)are mounted on the robot to determine the gas concentration gradient. The robotmoves to find the plume if it is situated outside the plume, whereas it moves alongthe wind direction if it is inside the plume. The fast response/recovery time as well assensitivity is required.The sensor responses to the vapor during the process of the target-source explora-

tion are shown in Fig. 3.15. The ethanol vapor was spouted in the clean roomwhere thewind field was relatively constant. The responses of the four gas sensors mounted onthe same robot are shown here. The four sensor responses were calibrated in advance.The starting point of the robot was 1.3 m away from the target source and the speed ofthe robot was 3 cm s�1. Since the sensor response is expressed as Rgas=Rair, the re-sponse value is small when the concentration is high. It is seen in the figure thatthe robot was approaching the target source because the ethanol concentration wasincreasing. However, the speed of the robot was limited by the response/recoverytime of the gas sensor. When the robot moved too fast, it wandered around thesame place and could not escape from it. Especially, the recovery time is importantsince the sensor with fast response time does not always have fast recovery time. Asensor with faster recovery time is required for the robot application.

3.7

Summary

The most convenient odor handling and delivery system is a sample-flow system,because it is easy to handle and the measurement cycle is short. On the otherhand, the static system is suitable for studying the fundamental behavior of the sen-

Fig. 3.15 Gas sensor responses mounted on a mobile robot

searching for target vapor source


sor. The direct exposure of the sensor to the vapor is sometimes performed in case ofthe field measurement. The appropriate method should be selected according to thepurpose.

Appendix: Optimization Algorithm for Realizing Plasticity

Let us briefly see how @I@T j

in Eq. (7) is determined. The curved surface mentioned aboveis defined as

I ¼ f ðT1;T2; :::;TmÞ ðA:1Þ

wherem is the number of the temperature peaks. The function above can be expandedaround point 0 in the following manner.

I � I0 ¼@f

@T1T1¼T ð0Þ

1ðT1 � T ð0Þ

1 Þ þ @f

@T2

��

��T2¼T ð0Þ

2

ðT2 � T ð0Þ2 Þ

þ:::þ @f

@Tm

��Tm¼T ð0Þ

m

ðTm � T ð0Þm Þ ðA:2Þ

where I0 is the index at point 0. Using n point data around point 0, the followingequations are obtained.

I1 � I0 ¼@f

@T1T1¼T ð0Þ

1ðT11 � T ð0Þ

1 Þ þ @f

@T2

��

��T2¼T ð0Þ

2


þ:::þ @f

@Tm

��Tm¼T ð0Þ

m

ðTm1 � T ð0Þm Þ

I2 � I0 ¼@f

@T1T1¼T ð0Þ

1ðT12 � T ð0Þ

1 Þ þ @f

@T2

��

��T2¼T ð0Þ

2


þ:::þ @f

@Tm

��Tm¼T ð0Þ

m

ðTm2 � T ð0Þm Þ

..

.

In � I0 ¼@f

@T1T1¼T ð0Þ

1ðT1n � T ð0Þ

1 Þ þ @f

@T2

��

��T2¼T ð0Þ

2

ðT2n � T ð0Þ2 Þ

þ:::þ @f

@Tm

��Tm¼T ð0Þ

m

ðTmn � T ð0Þm Þ

Tlk is the lth peak temperature of the kth point around the point 0. Note that n shouldbe larger than m. If D~II, DT , @f

@~TT, i.e., the approximate gradient vector at the point 0, are

defined as

D~II ¼

I1 � I0I2 � I0

..

.

In � I0

26664

37775; ðA:4Þ

3.7 Summary 7575

DT ¼

T11 � T ð0Þ1 T21 � T ð0Þ

2 T31 � T ð0Þ3 ::: Tm1 � T ð0Þ

1

T12 � T ð0Þ1 T22 � T ð0Þ

2 T32 � T ð0Þ3 ::: Tm2 � T ð0Þ

1

..

. ... ..

. ... ..

.

..

. ... ..

. ... ..

.

T1n � T ð0Þn T2n � T ð0Þ

2 T3n � T ð0Þ3 ::: Tmn � T ð0Þ

1

26666664

37777775

ðA:5Þ

and

@f

@~TT¼

@f

@T1

@f

@T2

..

.

@f

@Tm

266666666664

377777777775

; ðA:6Þ

then, Eq. (A.7) is obtained.

D~II ¼ ½DT � @f

@~TT

� �: ðA:7Þ

Note that DT is generally non-square matrix. e2, the sum of the squares of errors in nmeasurement points is

e2 ¼ D~II � ½DT � @f

@~TT

� �� T

DI � ½DT � @f

@~TT

� �� : ðA:8Þ

@e2

@Ti

is replaced by the variable ai.@e2

@~aais

@e2

@~aa¼

@e2

@a1@e2

@a2...

@e2

@am

2666666666664

3777777777775

: ðA:9Þ

It becomes

@e2

@~aa¼ 2ð½DT �T ½DT �~aa� ½DT �T D~IIÞ: ðA:10Þ


Since @e2

@~aa is zero at the point with the least squares of errors, the gradient direction~aa isgiven by

~aa ¼ ð½DT �T ½DT �Þ�1ð½DT �TD~IIÞ: ðA:11Þ

ð½DT �T ½DT �Þ is a symmetrical matrix. When the determinant of that matrix is close tozero, the inverse matrix is unstable and is not reliable. In that case, the eigenvalueanalysis technique called the singular value decomposition technique is used to sup-press the contribution of the negligibly small eigenvalues. The pseudo-inverse matrixis obtained using SVD technique [38].

References

1 D. R. Lide (Ed.). Handbook of Chemistry andPhysics, 76th Edition, CRC Press, 6–77(1995).

2 G. M. Wislon. J. Am. Chem. Soc., 86 (1964)127.

3 T. Nakamoto, A. Fukuda, T. Moriizumi,Y. Asakura. Sensors and Actuators B, 3(1991) 221.

4 T. Nakamoto, A. Fukuda, T. Moriizumi.Sensors and Actuators B 10 (1993) 85.

5 J. W.Gardner, T. C.Pearce, S. Friel,P. N. Bartlett, N. Blair. Sensors andActuators B, 18-19 (1994) 240.

6 M. A. Craven, J. W. Gardner. Trans InstMC 20 (1998) 67.

7 A. Iguchi, T. Nakamoto, T. Moriizumi.Sensors and Actuators B (2000) 155.

8 J. Ide, T. Nakamoto, T. Moriizumi. Sensorsand Actuators B, 13–14 (1993) 351–354.

9 H. Sundgren, F. Winquist, I. Lundstrom.Technical digest of Transducers 91 (1991)574.

10 R. S. Barratt. The analyst, 106 (1981) 817.11 T. Nakamoto, T. Fukuda, T. Moriizumi.

Sensors and Actuators B, 3 (1991) 1.12 M. Nakamura, I. Sugimoto, H. Kuwano,

R. Lemos. Technical Digest ofTransducers 93, 1993, p.434.

13 M. Ohnishi, T. Ishibashi, Y. Kijima,C. Ishimoto, J. Seto. Sensors and Materials,1 (1992) 53.

14 S. J. Martin, A. J. Ricco, D. S. Ginley,T. E. Zipperian. IEEE Trans. On UFFC, 2,UFFC-34 (1987) 142.

15 T. Nakamoto, S. Utsumi, N. Yamashita,T. Moriizumi. Sensors and Actuators B, 20(1994) 131.

16 J. W. Grate, D. S. Ballantine, H. Wohltjen.Sensors and Actuators B, 11 (1987) 173.

17 W. P. Carey, K. R. Beebe, B. R. Kowalski.Anal. Chem. 59 (1987) 1529.

18 J. V. Hatfield, P. Neaves, P. J. Hicks, K.Persaud, P. Travers. Sensors and ActuatorsB, 18–19 (1994) 221.

19 J. Ide, T. Nakamoto, T. Moriizumi. Sensorsand Actuators A, 49 (1995) 73.

20 H. Abe, T. Yoshimura, S. Kanaya,Y. Takahashi, Y. Miyashita, S. Sasaki.Analytica, Chimca Acta, 194 (1987) 1.

21 K. Yokoyama, F. Ebisawa. Anal. Chem. 55(1993) 677.

22 T. C. Pearce, J. W. Gardner, S. Friel,P. N. Bartlett, N. Blair, Analyst, 118 (1993)371.

23 H. Muramatsu, E. Tamiya, I. Karube. Anal.Chim. Acta, 225 (1989) 399.

24 Y. Okahata, O. Shimizu, H. Ebato. Bull.Chem. Soc. Jpn., 63 (1990) 3082.

25 K. Nakamura, T. Nakamoto, T. Moriizumi.Sensors and Actuators B, 61 (1999) 6.

26 J. W. Grate, S. L. Rose-Pehrsson,D. L. Venezky, M. Klusty, H. Wohltjen.Anal. Chem. 65 (1993) 123.

27 Q. Cai, J. Park D. Heldsinger, M. Hsieh,E. T. Zeller. Sensors and Actuators B 62(2000) 121.

28 W. A. Groves, E. T. Zellers, G. C. Frye.Analytica Chimca Acta 371 (1998) 131.

29 J. Kita, Y. Aoyama,M. Kinoshita, H. Nakano,H. Akamatsu. Technical Digest of SensorSymposium, IEEJ 2000, p.301.

30 R. E. Shaffer, S. L. Rose-Pehrsson,R. A. McGill. Field Analytical Chemistryand Technology, 2 (1998) 179.

3.7 Summary 7777

31 K. S. Booksh, B. R. Kowalski. Anal. Chem. 66(1994) 782A.

32 Y. Isaka, T. Nakamoto, T. Moriizumi.Technical Digest of Transducers 99 (1999)3P3.3.

33 T. Nakamoto, Y. Isaka, T. Ishige,T. Moriizumi. Sensors and Actuators B,69 (2000) 58.

34 T. Nakamoto, H. Ishida, T. Moriizumi. Proc.IEEE International Symposium on Indust-rial Electronics (1997) SS128.

35 W. R. Dillon, M. Goldstein. MultivariateAnalysis, Wiley, 1984, p. 163–422.

36 B. S. Gottfried, J. Weisman. Introduction tooptimization theory, Prentice-Hall, Engle-wood Cliffs, 1973, p. 84.

37 M. A. Sharaf, D. L. Illman, B. R. Kowalski.Chemometrics, Wiley, 1986, 164–310.

38 T. Nakamoto, H. Matsushita, N. Okazaki.Sensors and Actuators A, 50 (1995) 191.

39 J. W. Gardner, E. Llobet, E. L. Hines. Proc.IEE Circuits, Devices and Systems, 146(1999) 101.

40 T. Yamanaka, H. Ishida, T. Nakamoto,T. Moriizumi. Sensors and Actuators A,69 (1998) 77.

41 P. Tobias, P. Martensson, A. Goras,I. Lundstrom. Sensors and Actuators B,58 (1999) 389.

42 H. Ishida, T. Nakamoto, T. Moriizumi.Meeting abstract of electrochemical society,1999, p. 1078.

43 T. Nakamoto, T. Tokuhiro, H. Ishida,T. Moriizumi. Latenews of Transducers 99,1999, LN9.

44 H. Ishida, T. Yamanka, N. Kushida,T. Nakamoto, T. Moriizumi. Sensors andActuators B, 65 (2000) 14.

45 T. Nakamoto, H. Ishida, T. Moriizumi. Anal.Chem. 4 (1999) 531A.

46 R. A. Russell, R. A. Thiel, D. Deveza,A. Mackay-Sim. IEEE Int. Conf. Roboticsand Automation, (1995) 556.

47 Y. Kuwana, I. Shimoyama, H. Miura. Int.Conf. Intelligent Robots and Systems, 1995,530.


4

Introduction to Chemosensors

H. Nanto, J. R. Stetter

4.1

Introduction

We believe that the 21st century can be the aroma age. The culture of aroma developedwith human civilization and the good smell of foodstuffs gives great comfort to thehuman heart. The sense of smell is, therefore, one of the most interesting of the fivehuman senses and yet is understood the least. The human nose is widely used as ananalytical sensing tool to assess the quality of such as drinks, foodstuffs, perfumes, andmany other household products in our daytime activities, and of many products in thefood, cosmetic, and chemical industries. However, practical use of the human nose isseverely limited by the fact that the human sense of smell is subjective, often affectedby physical and mental conditions, and tires easily. Consequently, there is consider-able need for a device that couldmimic the human sense of smell and could provide anobjective, quantitative estimation of smell or odor.Recently, there has been increasing interest in the development of such a device, the

so-called ‘electronic nose (e-nose)’. This is an electronic instrument that is capable ofdetecting and recognizing many gases and odors, and comprises a sensor array usingseveral chemosensors and a computer. The different types of chemosensors, especiallyodor sensors, that have been employed within an e-nose are described in this chapter.

4.2

Survey and Classification of Chemosensors

A chemosensor is a device that is capable of converting a chemical quantity into anelectrical signal and respondate the concentration of specific particles such as atoms,molecules, or ions in gases or liquids by providing an electrical signal. Chemosensorsare very different from physical sensors. Although approximately 100 physical mea-surands can be detected using physical sensors, in the case of chemosensors this num-ber is higher by several orders of magnitude. The types of chemosensors that can be


7979

used in an e-nose need to respond to odorous molecules in the gas phase, which aretypically volatile organic molecules with different relative molar masses.

Tab.

4.1

Classification

ofchem

osen

sors

that

have

been

exploited

sofar.Metal

oxidesemicon

ductor,MOS;

MOSfield

effect

tran

sistor,MOSF

ET;qu

artz

crystalmicrobalance,QCM;surfaceacou

stic

wave,

SAW;surfaceplasm

onresonan

ce,SP

R.

Principle

Measurand

Sen

sortype

Fabrication

method

sAvailability/sensitivity

Advantages

Disadvantages

Con

ductom

etricCon

ductan

ceChem

oresistor

MOS

Microfabricated,

Sputtering

Com

mercial,man

y

types,5–500ppm

Inexpen

sive,

microfabricated

Operates

athigh

temperature

Con

ducting

polym

er

Microfabricated,

Electroplating,

Plasm

aCVD,Screen

printing,

Spin

coating

Com

mercial,man

y

types,0.1–100ppm

Operates

atroom

temperature,

microfabricated

Verysensitive

to

humidity

Capacitive

Capacitan

ceChem

ocapacitor

Polym

erMicrofabricated,

Spin

coating

Research

Applicableto

CMOS-based

chem

osen

sor

Verysensitive

to

humidity

Poten

tiom

etric

Voltage/e.m

.f.

Chem

diode

SchottkyDiode

Microfabricated

Research

Integrated,

Applicableto

CMOS-based

chem

osen

sor

NeedsPd,Pt,

Au,Ir

(expen

si-

ve)

I-V/C

-VChem

otransistor

MOSFET

Microfabricated

Com

mercial,

specialorder

only,ppm

Integrated,

Applicableto

CMOS-based

chem

osen

sor

Odoran

t

reaction

product

must

pen

etrate

gate

Calorim

etric

Tem

perature

Them

al

chem

osen

sor

Thermister

(pyroelectric)

Microfabricated,

Ceram

icfab.

Research

Low

cost

Slow

respon

se

Pellistor

Microfabricated

Research

Low

cost

Slow

respon

se

Thermocou

ple

Microfabricated

Research

Low

cost

Slow

respon

se

Gravimetric

Piezoelectricity

Mass-sensitive

chem

osen

sor

QCM

Microfabricated,

Screenprinting,

Dip-coating,

Spin

coating

Com

mercial,

severaltypes,

1.0ngmass

chan

ge

Wellunderstood

technology

MEMsfabrica-

tion

,interface

electron

ics?

SAW

Microfabricated,

Screenprinting,

Dip-

coating,

Spin

coating

Com

mercial,

severaltypes,

1.0ngmasschan

ge

Differential

devices

canbe

quitesensitive

Interface

electron

ics?

Optical

Refractiveindex

Reson

ant-type

chem

osen

sor

SPR

Microfabricated,

Screenprinting,

Dip-coating,

Spin

coating

Research

Highelectrical

noise

immunity

Expen

sive

Intensity/spec-

trum

Fiber-optic

chem

osen

sor

Fluorescence,

chem

oluminescence

Dip-coating

Research

Highelectrical

noise

immunity

Restrictedavai-

lability

ofligh

t

sources

Amperom

etry

curren

tToxic

Gas

Sen

sor

Electrocatalyst

Com

posite

Electrodes

Com

mercial

ppb-ppm

Low

cost

noR

h

interferen

ce

Size

4 Introduction to Chemosensors80

Chemosensors as listed in Table 4.1 have been exploited and some already manu-factured. Principles such as electrical, thermal, optical, and mass can be used to or-ganize these chemosensors according to their device class. The chemosensors usingmetal oxide semiconductors (MOS), organic conducting polymers (CP), chemocapa-citors, MOS field-effect transistors (MOSFET), quartz crystal microbalance (QCM),surface acoustic wave (SAW), surface plasmon resonance (SPR), fluorescence, andothers that can be easily used as the sensor for an e-nose are included in the followingdiscussion. Details about the types of chemosensors discussed here and others can befound in the literature [1–6].

4.3

Chemoresistors

Chemoresistors based on the conductivity change of MOS or organic CPs by chemicalreaction with gaseousmolecules are the simplest of gas sensors, and are widely used tomake arrays for gas and odor measurements.

4.3.1

MOS

Metal oxides such as SnO2, ZnO, Fe2O3, and WO3 are intrisically n-type semiconduc-tors. At temperatures of 200–500 8C, these respond to reducible gases such as H2,CH4, CO, C2H5, or H2S and increase their conductivity. The conductivity r andthe resistivity q is given by

r ¼ 1=q ¼ enl ð1Þ

where e is the charge on the electron (1:6022� 10�19 C), n the carrier (electron or hole)concentration (cm�3) and l the carrier mobility (cm2 V�1s�1). In the atmosphere, someoxygenatomsareadsorbedon thesurfaceofn-type semiconductors to trap freeelectronsfrom the semiconductor, and consequently a highly resistive layer is produced in thevicinity of the semiconductor surface. The adsorption of oxygen atoms on the semi-conductor surface and at grain boundaries of polycrystalline semiconductors creates anelectrical-double layer that acts as the scattering center for conducting electrons. Theconsequent increase in free electrons and decrease in scattering centers results in anincrease in conductivity. The mechanism is similar for p-type semiconductors but is ofopposite sign [101].The mechanism of the increase in carrier concentration by reacting with the redu-

cible gases as described above can be understood from the following reactions:

eþ 1

2O2 ! OðsÞ� ð2Þ

RðgÞ þOðsÞ� ! ROðgÞ þ e ð3Þ


where e is an electron from the conduction band of the oxide semiconductors, R(g) isthe reducible gas, and s and g imply surface and gas, respectively. Equation (2) impliesthat oxygen is physico-chemically adsorbed onto lattice vacancies in the oxide semi-conductor, and consequently the conductivity of the oxide semiconductor becomeslower than that in the case of no adsorbed oxygen. An electron is, however, generatedby the reaction with reducible gases R(g) through Eq. (3). Consequently, the conduc-tivity is increased following Eq. (3) as a result of the increase in carrier concentration.In contrast, p-type semiconductors such as CuO, NiO, and CoO respond to oxidizablegases such as O2, NO2, and Cl2 [101].The schematic diagram in Fig. 4.1 explains the conductivity increase due to the car-

rier mobility for SnO2 gas sensors. In clean air, oxygen atoms that trap free electrons inthe bulk SnO2, is adsorbed onto the SnO2 particle surface, forming a potential barrierin the grain boundaries as shown in Fig. 4.1a. This potential barrier restricts the flow ofelectrons, causing the electrical conductivity to decrease, because the potential barrieracts as the scattering center for electron conduction. When the sensor is exposed to anatmosphere containing reducible gases, e.g. combustible gases, CO, and other similarvapors, the SnO2 surface adsorbs these gas molecules and causes oxidation. This low-ers the potential barrier, allowing electrons to flow more easily, thereby increasing theelectrical conductivity as shown in Fig. 4.1b.The reaction between gases and surface oxygen will vary depending on the operating

temperature of the sensor and the activity of sensor materials. The increasing sensi-tivity and selectivity of the sensors for exposure to gases can be realized by incorpora-tion of a small amount of impurities and catalytic metal additives such as palladium(Pd) or platinum (Pt). The impurities act as extrinsic donors (or acceptors) and, con-sequently, controlling the doped amount of impurities can change the conductivity ofthe sensors. Doping of the catalytic metal to the sensor or coating with thin catalytic

Fig. 4.1 Schematic diagram explaining the conductivity increases

caused by the carrier mobility increase in SnO2 gas sensors. (a) Oxygen

is adsorbed onto the SnO2 particle surface, forming a potential barrier

in the grain boundaries. (b) The potential barrier is lowered bymeans of

reaction of the oxygen atoms with reducing gas, allowing electrons to

flow more easily, thereby increasing the electrical conductivity


Tab. 4.2 Commercially available metal oxide semiconductor chemo-

sensors.

Manufacturer Applications Model Typical detection range and features

FGARO ENG Combustible gas TGS813 For detection of various combustible gases

TGS816 500–10 000 (ppm)

TGS842 Improved sensitivity to CH4

500–10 000 (ppm)

TGS821 High selectivity and sensitivity to H2

500–10 000 (ppm)

Toxic gas TGS203 High selectivity and sensitivity to CO

50–1000 (ppm)

TGS825 High sensitivity to H2S

5–100 (ppm)

TGS826 High sensitivity to NH3 and amine compounds

30–300 (ppm)

Solvent vapor TGS822 High sensitivity to alcohol and organic com-

pounds such as toluene and xylene

TGS823

Halocarbon gas TGS830 High sensitivity to various CFCs, HCFCs

TGS831 100–3000 (ppm)

TGS832

Air quality control TGS800 High sensitivity to gaseous air contaminants

(such as cigarette smoke and gasoline exhaust)

1–10 (ppm)

Cooking control TGS880 Vaporized gases and water vapor form food in

the cooking process

10–1000 (ppm)

TGS882 Alcohol vapor from food in the cooking process

50–5000 (ppm)

TGS883 Water vapor from food in cooking process

1–150 (g m�3)

NEW

COSMOS ELEC.CO,

LTD.

Combustible gas CH-H High sensitivity and selectivity to H2,

50–1000 (ppm)

CH-M High sensitivity to VOCs such as CH4 and i-

C4H10

1000–10 000 (ppm)

CH-CO High sensitivity to CO

100–1000 (ppm)

CH-E2 High sensitivity to alcohol

CH-E3 1–1000 (ppm)

CH-L High sensitivity to LPgas

Toxic gas CH-N High sensitivity and selectivity to NH3

AET-S High sensitivity and selectivity to H2S

Thin film type


metal film of the sensor surface changes the selectivity of the sensor. As describedabove, the crystallographic structure of the semiconductors used as the sensor mate-rial is commonly polycrystalline, and thus includes many grain boundaries. Thesegrain boundaries act as the scattering centers for conducting electrons to producethe change of carrier mobility, and therefore consequently the extent of crystallinityaffects the sensitivity of the sensors.The most widely used semiconducting material as a gas sensor is SnO2 doped with

small amounts of impurities and catalytic metal additives. By changing the choice ofimpurity and catalyst and operating conditions such as temperature, many types of gassensors using SnO2 have been developed. The gas sensors using metal oxide semi-conductors exhibit relatively poor selectivity for gases and remain responsive to amany kinds of combustible gases. Table 4.2 lists some of the commercially availablegas sensors of SnO2 and ZnO that are manufactured by New Cosmos Electric Co., Ltdand Figaro Engineering Inc. (Japan).Figure 4.2 shows schematically the basic construction of the sintering-type and thin-

film-type of gas sensors. The type of sensor materials and operating temperatures oftypical gas sensors using MOSs that have been reported so far are listed in Table 4.3.

4.3.2

Organic CPs

Chemoresistors made from organic CPs also exhibit a change in conductance whenthey are exposed to reducible or oxidizable gases. Organic CPs show reversible

Tab. 4.3 Type of conduction and operating temperatures of typical gas

sensors using metal oxide semiconductors.

Materials (Dopants) n-type or p-type Top ( 8C) Detecting gases Ref.

ZnO(Al) n 200 H2 30

ZnO(Al) n 350 NH3 31

ZnO(Al,In,Ga) n 400 TMA 32

ZnO n 280–470 CO 33

ZnO n 450 CCl2F4, CHClF2, 34

WO3(Pt) n 250–400 N2H4, NH3, H2S, 35

WO3 n 500 CO, CH4, SO2 36

TiO2(Ru) n 560 TMA 37

a-Fe2O3 n 400 H2, CH4 38

c-Fe2O3 n 420 H2, CH4, C3H8C4H10, C2H5OH 39

CdIn2O3 n 300 CO 40

CuTa2O6 n 400 H2, CO 41

CuO/ZnO p/n 250 H2, CO 42

Co3O4 p 200–500 CO, H2, NOX 101

Cr2O3(Ti) n 420 TMA 43

In2O3 (Mg or Zn) n 420 TMA 43

BaSnO3 n 300–500 H2, CO, CH4, H2S, SO2 44

Bi2Sn2O7 p 500 H2, CO, C2H4, NH3 44

Bi6Fe2Nb6O30 n/p 500 C3H8, Cl2, NO2, SO2, H2S 45


changes in conductivity when chemical substances (e.g. methanol, ethanol, and ethylacetate) adsorb and desorb from the polymer. The mechanism by which the conduc-tivity is changed by this adsorption is not clear at present.There are a large number of different electronically conducting polymers. Polypyr-

role was first prepared electrochemically in 1968 [23] and has been most extensivelystudied so far. Electroconducting conjugated polymers (ECP) can exhibit intrinsic elec-tronic conductivity. Their structure contains a one-dimensional organic backbone withalternating single and double bonds, which enables a super-orbital to be formed forelectronic conduction. The most commonly applied polymers for gas-sensing applica-tions have been polypyrrole, polyaniline, polythiophene, and polyacetylene, which are

Fig. 4.2 The basic construction of the sintering-type (a) and thin-film-

type (b) of the gas sensors that are commercially available


based on pyrrole, aniline or thiophenemonomers [24]. Because of their properties theyhave remarkable transductionmatrices that are sensitive to gases and vapors, resultingin a straightforward conductance change via the modulation of their doping level. Theearly studies [25, 26] of the gas-sensing application of organic CPs concentrated on theresponse to reactive gases such as ammonia and hydrogen sulfide. Gustafsson et al.[27] have reported that gas sensors using polypyrrole films exhibit a high sensitivity forammonia gas. Subsequent work [28–30] also showed that gas sensors using organicCPs such as polypyrrole respond to a wide range of organic vapors such as methanol.More recently, studies have been carried out on preparation of thin-film CPs for gas

sensing applications [25, 31]. Thin films of heteroaromatic monomers such as pyr-roles, thiophenes, indoles, and furans were grown electrochemically on interdigitatedelectrodes to produce gas-sensitive chemoresistors [25].Chemoresistors using organic CPs respond to a wide range of polar molecules at

temperatures as low as room temperature (RT) and more recent reports suggest that ahigh sensitivity down to 0.1 ppm is possible. This result indicates that organic CP is apotentially useful material for applications in odor-sensing and e-nose applications.The use of organic CPs as odor sensor materials is very attractive for the following

reasons:

1) a wide range of materials can be simply prepared;2) they are relatively low cost materials;3) they have a high sensitivity to many kinds of organic vapors;4) gas sensors using organic CPs operate at low temperatures.

Comparison between the properties of the organic CP odor sensor and the MOS odorsensor is shown in Table 4.4.

Tab. 4.4 Comparison of the properties of the conducting polymer

odor sensor and the metal oxide odor sensor (thick-film and thin-film

types).

Properties Conducting polymer SnO2 (thick film) SnO2 (thin film)

Key measurand Conductance Conductance Conductance

Fabrication Electrochemical

growth, plasma CVD

paste Sputtering, Sol-gel

Choice of materials Wide Limited Limited

Operating temperature 10–110 8C 250–600 8C 250–600 8CMolecular Receptive range Wide range Combustible vapors Combustible vapors

Detection Range less than 20 ppm 10–1000 ppm 1–100 ppm

Response time 60 s 20 s 20 s

Size Less than 1 mm2 1 � 3 mm Less than 1 mm2

Power Consumption Less than 10 mW 800 mW 80 mW

Integrated array Yes No Yes

Stability Moderate Relatively poor Poor

Interferences Acidic gases, water SO2, Cl2, H2O SO2, Cl2, H2O


Another way to use CPs is to make non-conducting materials, e.g. silicone [32] andpolystyrene [33], conductive by inclusion of carbon-black metal powder. These sensorsare used in e-noses and can exhibit high sensitivity [34].

4.4

Chemocapacitors (CAP)

The principle of chemocapacitors using polymers is schematically shown in Fig. 4.3.There are two steady states for the sensitive layer during operation. In the first state asshown in Fig. 4.3a, no gaseous analyte molecules are present in the sampling envir-onment and consequently only air is incorporated into the polymer. As a result, acertain capacitance C of the sensitive polymer layer is measured and constitutesthe baseline. In the second state, gaseous analyte molecules are present in the sam-pling environment as shown in Fig. 4.3b. When the polymer absorbs the gaseous ana-lyte, the sensitive polymer layer changes its electrical (e.g. dielectric constant e) andphysical properties (e.g. volume V) to produce deviations (De, DV) from the first state(reference state). The changes in electrical and physical properties of polymers are theresult of reversible incorporation of gaseous analyte molecules into the polymer ma-trix.The CMOS-based chemical sensor using chemocapacitive microsensors for detect-

ing volatile organic compounds (VOCs) was built with two interdigitated electrodesspin-coated or spray-coated with polymers such as (poly)etherurethane (PEUT) byKoll et al. [35].

Fig. 4.3 Chemocapacitor based on capacitance measurement of

sensitive layers. There are two steady states for the sensitive layer

during operation; (a) no analyte molecules are present in the sampling

environment, and (b) analyte molecules are present in the sampling

environment


4.5

Potentiometric Odor Sensors

Gas sensors utilizing the electrical characteristics of Schottky diodes and the MOSFEThave also been investigated. Those using the Schottky diode are based on a change inthe work function because of the presence of chemical species on their surface. Ex-amples are catalytic metals (inorganic Schottky diodes) such as Pd and Pt, and organicCPs (organic Schottky diodes) such as polypyrrole. Gas sensors using a MOSFET arebased onmetal-insulator-semiconductor structures in which themetal gate is a catalystfor gas sensing. In this section, mainly potentiometric odor sensors using MOSFETsare included and discussed.

4.5.1

MOSFET

Themicrochemosensor using the structure of a MOSFET in which the gate is made ofa gas-sensitive metal such as Pd was first proposed by Lundstrom in 1975 [36]. Thissensor exhibited a threshold voltage shift depending upon the gas concentration andwas particularly sensitive to hydrogen down to the ppm level with maximum threshold

Fig. 4.4 Basic structures of n-channel MISFET and MISCAP, which

operate on the same basic principle but differ in measurands


voltage shift of about 0.5 V. The use of other metal gate materials such as Pt and Ir andoperating the sensors at different temperatures has led to reasonable selectivity togases such as NH3, H2S, and ethanol [37]. There are two basic structures such asMISFET (metal-insulator-semiconductor FET) and MISCAP (MIS CAPacitor ). Thebasic structures of n-channel MISFET and MISCAP that operate on the same basicprinciple but differ in measurands are shown in Fig. 4.4. In the MISFET, the draincurrent iD flowing through the semiconductor is controlled by the surface potentialdue to the applied gate voltage VG, and in the MISCAP the capacitance of the MISstructure is determined by the surface potential. These devices can respond to expo-sure to any gas that changes the surface potential or the work function of the gatemetal. The materials used in MOSFET-type odor sensors as well as the Schottky-type odor sensors are listed in Table 4.5 in comparison to those of MOS-type andCP-type odor sensors.

4.6

Gravimetric Odor Sensors

Recently, gravimetric odor sensors using acoustic wave devices that operate by detect-ing the effect of sorbed molecules on the propagation of acoustic waves have beeninvestigated for application to an e-nose. Two main types utilizing QCM (or bulkacoustic wave, BAW) and SAW devices have been used as the odor sensors, althoughother types of device have been investigated. In both types, the basic device consists ofa piezoelectric substrate, such as quartz, lithium niobate and ZnO, coated with a sui-table sorbent membrane [38]. Sorption of vapor molecules into the sorbent membranecoated on the substrate can then be detected by their effect on the propagation of the

Tab. 4.5 Materials used in the different odor sensors. MOSFET –

metal oxide semiconductor field effect transistor.

Chemosensor

type

Structure Examples of sensor

materials used

Examples of

detecting gases

MOSFET type Metal-gate MOSFET Pd(Pt)-gate FET

(SiO2, SnO2-Si, SiC)

H2, CO, H2S, NH3

Schottky type Metal/Semiconductor Pd-TiO2 (ZnO) H2, CO, CH3SH

p/n Nb2O3-Bi2O3

p/n ZnO-CuO

Metal/polymer Al/poly(3-octythiophene) NH3, NOx

Chemoresistors n-type semiconductors SnO2, ZnO, a-Fe2O3, TiO2, In2O3,

V2O3, SnO2 þ Pd, ZnO þ Pt,

SnO2þ ThO2 þ Pd,

H2, CO, alcohols,

hydrocarbons,

O2, NO2, Cl2p-type semiconductors CoO, Co3O4, CuO,

Sm0.5Sr0.5CoO3, Co0.3Mg0.7O,

La0.35Sr0.65Co0.7Fe0.3O3-x

H2, O2, CO, alcohols

Conducting polymers Anthracene, phthalocyanine,

polypyrrol, polyacrylonitorile,

polyphenylacetylene

NO, NO2, O2, SO2,

CO, NH3, alcohols


acoustic wave causing changes in the resonant frequency and the wave velocity. Theacoustic waves used are at ultrasonic frequencies ranging typically from 1 to 500 MHz.Both types are discussed in this section.

4.6.1

QCM

QCM or thickness shear mode (TSM) devices using BAWs in piezoelectric materialsare probably the simplest type of odor sensor using a piezoelectric device. Rock crystalsuch as single crystal quartz has an interesting property in that it is distorted by appliedelectric voltage and conversely an electric field is generated by applied pressure. Thisphenomenon is called the piezoelectric effect. Because of this effect, upon excitation byapplication of a suitable a.c. voltage across the quartz crystal, the crystal can bemade tooscillate at a characteristic resonant frequency. A QCM odor sensor comprises of aslice of a single crystal of quartz, typically around 1 cm in diameter, with thin-filmgold electrodes that are evaporated onto both surfaces of the sliced crystal. The quartzcrystal oscillates in such manner that particle displacements on the QCM sensor sur-face are normal to the direction of wave propagation.The thickness of the quartz crystal determines the wavelength of the fundamental

harmonics of oscillation. The resonant frequency of the QCM sensor is related to thechange of the mass of QCM loading by the Sauerbrey equation [39]:

Df ¼ �2f 20 mf=AðqqlqÞ1=2 ð4Þ

where Df is the change in resonant frequency, f0 is the resonant frequency, mf is themass change due to adsorption of gases, A is the electrode area, qq is the density ofquartz and lq is the shear modulus. For typical AT-cut quartz crystal operating at10 MHz, a mass change of the order of 1 ng produces a frequency change of about1 Hz. Thus small changes in mass can be measured using a QCM coated with a mo-lecular recognition membrane on which odorant molecules are adsorbed, as shown inFig. 4.5. The selectivity of the QCM sensor is determined by the coating membranedeposited on the surface of the crystal.The functional design of the polymer-film-coated QCM odor sensor, based on the

solubility parameter of the sensing membrane and detecting gases, was carried out in

Fig. 4.5 Schematic diagram of the structure of a QCM chemosensor.

The sensor consists of polymer membrane that recognizes analyte

molecules and odors, and a QCM as a transducer


order to develop a sensor with excellent selectivity and high sensitivity for harmfulgases such as toluene, xylene, ammonia, and acetaldehyde by Nanto et al [40, 41].The polymer films such as propylene-butyl, polycarbonate, and acrylic resin of whichthe solubility parameters almost coincide with those of toluene, acetaldehyde, andammonia gas, respectively, are chosen as the sensing membrane material coatedon the QCM surface. They found that propylene-butyl-coated sensor exhibited ahigh sensitivity and excellent selectivity for toluene and xylene gases, as expectedfrom the functional design based on solubility parameters. They also found thatthe polycarbonate-coated and acrylic-resin-coated sensors exhibited high sensitivityand excellent selectivity for acetaldehyde and ammonia gases, respectively, also asexpected. The result strongly suggests that the solubility parameter is effective in

Tab. 4.6 Research on e-noses using different types of chemosensors,

including: quartz crystal microbalance, QCM; surface acoustic wave,

SAW; metal oxide semiconductor, MOS; MOS field effect transistor,

MOSFET. Pattern recognition types: multi-layer perception, MLP;

principal component analysis, PCA; fuzzy learning vector quantization,

FLVQ; cluster analysis, CA; Kohonen network, KOH; linear regression,

LR; feature weighting, FW; least square, LS; discriminant function

analysis, DFA; and fuzzy reasoning, FUZ.

Chemosensor type Number of sensors Applications Pattern recognition Ref.

QCM 8 Spirits, perfumes, odors MLP, PCA, FLVQ 46–51

4 Odors 52

8 Odors PCA, CA 53

6 Odors 54, 55

3 Harmful gases PCA 18, 19

SAW 6 Perfumes 56

4 Odors 57

12 58–60

MOSFET 10 Meat MLP, KOH 61

324 Odors 62

MOS 3 Odors 63

3 Odors, tobacco LR, FW 64

12 Odors, coffee LS 65, 66

12 Odors, beverages PCA, CA 67

12 Odors, beers MLP 68, 69

3 Odors CA, LS 70

12 Wines MLP, KOH 71

6 Odors LS 72

8 Odors 73

8 Odors MLP, LS 74

6 Odors LS 75, 76

7–8 Odors LR, PCA, CA 77, 78

6 Spirits, coffee CA, PCA, DFA 79, 80

3 Odors MLP 81

6 Odors KOH 82

3 Fish 83

3 Odors FUZ 84

AGS 4 Grain KNN, NN 106


the functional design of the sensingmembrane of QCM odor sensors. The research one-nose applications using QCM odor sensors as well as those using other type of che-mosensors such as SAW, MOSFET, and MOS are listed in Table 4.6.Recently, studies on QCM odor sensors with plasma-polymerized organic film as

the molecular recognition membrane [42–45] and odor sensors using fundamentaland overtone modes of QCM with high frequency [46, 47] have been reported.

4.6.2

SAW

The SAW device is made of a relatively thick plate of piezoelectric materials (ZnO andlithium niobate) with interdigitated electrodes to excite the oscillation of the surfacewave [87–89]. The SAW is stimulated by applying an a.c. voltage to the fingers of aninterdigitated electrode to lead to a deformation of the piezoelectric crystal surface. TheSAW devices are usually operated in one of two configurations such as a delay line anda resonator. In both cases, the propagation of the SAW is affected by changes in theproperties of the piezoelectric crystal surface. In common gas sensors using a SAWdevice with a dual delay line structure, one arm of the delay line is coated with thesorbent membrane, the other acts as a reference to reduce the change of environmen-tal conditions such as temperature drift and other effects. The change in frequency ofthe SAW with sorption of vapor, Df V, is given by

DfV ¼ DfpcVKp=qp ð5Þ

for a simple mass loading effect, where Dfp is the change in frequency caused by poly-mer membrane itself, cV is the vapor concentration, Kp is the partition coefficient andqp is the density of the polymer membrane used.Considerable work [87] has been reported on the measurement of inorganic gases

such as NO2, H2, H2S, and SO2, and organic gases and vapors such as CH4, C6H6, andC2H5OH. This type of sensor using polymer materials as a sensing membrane can bechemically modified to obtain a higher degree of specificity, because the choice ofchemically sensitive membrane determines the selectivity of the sensor. The SAWodor sensors generally work at much higher frequencies of the order of GHz thanthat of the BAW odor sensor (10 MHz). The main problems with SAW odor sensorare a relatively poor long-term stability and a high sensitivity to humidity. A goodreview of acoustic sensors is available [6].


4.7

Optical Odor Sensors

4.7.1

SPR

SPR is an optical phenomenon in which incident light excites a charge-density wave atthe interface between a highly conductive metal and a dielectric material. The condi-tions for excitation are determined by the permittivities of the metal and the dielectricmaterial. The SPR transduction principle is widely used as an analytical tool for mea-suring small changes in the refractive index of a thin region adjacent to the metalsurface. The optical excitation of surface plasmon on a thin metallic film has, there-fore, been recognized as a promising technique for sensitive detection of chemicalspecies such as odor, vapor and liquid [90]. Several methods have been employedto monitor the excitation of SPR by measuring the light reflected from the sensorinterface. These include analysis of angle modulation [91], wavelength modulation[92], intensity modulation [93], and phase modulation [94].Optical SPR sensors are sensitive to the change in the refractive index of a sample

surface. Recently, it has been reported that toxic gases such as ammonia, toluene,xylene, ethylacetate, 4-methyl-2-pentanone, and propionic acid can be detected bymea-suring the SPR using angle modulation [95]. The SPR was measured using theKretschmann configration, illustrated in Fig. 4.6, with a prism and a thin, highly con-ductive gold metal layer deposited on the prism base. The LED emitting 660 nm lightwas used as the light source to excite the SPR. The SPR reflection spectrum (reflectedlight intensity versus angle of incidence with respect to the normal of the metal/di-electric interface) was measured by coupling transverse magnetically polarized mono-chromatic light into the prism and measuring the reflected light intensity of the rayexiting the prism versus the angle incidence. In order to utilize this system as a gassensor, a very thin film of methyl methacrylate, polyester resin, or propylene ether asthe sensing membrane was deposited on gold metal thin film using a spin-coatingmethod. The reflected light was measured using a CCD camera attached to a personalcomputer. The angle at which the minimum reflection intensity occurs is the reso-nance angle at which coupling of energy occurs between the incident light and the

Fig. 4.6 Kretschmann configuration of SPR apparatus used

in toxic gas detection [29]


surface plasmon waves. Four channel images of reflected light were observed by usingthe CCD camera. The schematic configuration of the SPR sensor is shown in Fig. 4.7.The SPR sensor with synthetic polymer thin film on the gold metal film as a sensingmembrane exhibited high sensitivity for toxic gases such as ammonia, toluene, xylene,ethylacetate, 4-methyl-2-pentanone, and propionic acid.

4.7.2

Fluorescent Odor Sensors

Recently, a new sensing device has been developed that consists of an array of opticallybased chemosensors providing input to a pattern recognition system. This type ofchemosensor consists of optical fibers deposited with fluorescent indicator NileRed dye in polymer matrices of varying polarity, hydrophobicity, pore size, elasti-city, and swelling tendency to create unique sensing regions that interact differentlywith vapor molecules [96].Fiber-optic sensors most often consist of an analyte-sensing element deposited at

the end of an optical fiber. Individual optical fibers with a diameters as small as 2 lmand imaging bundles with a diameter of 500 lm are available, enabling easy minia-turization, and are free from electrical interference. In a fiber-optic chemosensingsystem, the optical sensing element is typically composed of a reagent phase immo-bilized at the fiber tip by either physical entrapment or chemical binding. This reagentphase usually contains a chemical indicator that experiences some change in opticalproperties, such as intensity change, spectrum change, lifetime change, and wave-

Fig. 4.7 Schematic configuration of the SPR

sensor


length shift in fluorescence, upon interaction with analyte gases or vapors. The re-sponses depend upon the nature of the organic vapor and the strength of its interac-tion with the different polymer systems used.The most common configuration of optical fiber chemosensor utilizing fluores-

cence and example of the response are shown in Fig. 4.8. The authors then analyzedthe transient responses of the sensor array to distinguish different organic vapors suchas odor samples a, b, and c.At present, the sensitivity of some types of optical chemosensor is not high (detec-

tion limits of several 1000 ppm) and there is little information about the lifetime,reproducibility or stability of the sensor system. Nevertheless, this is an interestingapproach and one worthy of future work.

4.7.3

Other Optical Approaches

The use of a colorimeter coupled to optical fibers makes an inherently simple sensor[97], can be found in many forms, and was one of the earliest of the optical chemicalsensor approaches. Color changes, or more generally, changes in absorption or emis-sion of radiation, and polymer swelling by changes in refractive index of fiber coatingscan be monitored optically. More recent approaches make e-noses from arrays of mi-crobeads on the end of a fiber [96, 98]. These systems can bemade exquisitely sensitivewith the appropriate chemistry on the fiber tip. The future of optical arrays within thee-nose are very promising.

Fig. 4.8 (a) The most common configuration of an optical fiber chemosensor utilizing fluorescence, and

(b) an example of the response


4.8

Thermal (Calorimetric) Sensors

There are two sensor classes that are based on thermal technology. Those using pyro-electric [38] or thermopile sensors with coatings that absorb the analyte of interest. Theunderlying thermal sensor records the heat of solution of the analyte in the coating.They are quantitative because the more analyte that is absorbed, the more heat isgenerated. The theory and analytical performance of these sensors is similar to thecoated SAW or chemiresistor polymer sensors, except that the underlying transduceris a heat sensor.The second class of thermal sensor is the Pellister, catalytic bead, or combustible gas

sensor [99]. The catalytic sensor is typically a tiny bead of catalyst a millimeter or less indiameter that surrounds a coil of thin, 0.025 mm, Pt wire that acts as a Pt resistancethermometer. When resistively heated to about 500 8C, any contact with a hydrocarboncauses catalytic oxidation of the hydrocarbon with commensurate liberation of the heatof combustion. This heat is at the surface of the catalyst bead and some is lost to thesurroundings while some is transferred to the tiny catalyst sensor bead. The heat trans-ferred to the bead raises the temperature of the sensor, and it is this temperaturechange that is sensed as a change in resistance by the thin Pt wire. The sensor istypically placed in a Wheatstone Bridge circuit to measure the tiny changes in resis-tance of the Pt wire. The larger the resistance change, the higher the concentration ofhydrocarbon. These sensors are typically used for combustible gases and were used invery early e-noses [100]. There are many formulations of the catalyst material and thesesensors are operated at constant temperature or at constant voltage to serve differentapplications.

4.9

Amperometric Sensors

The amperometric gas sensor, or AGS, was one of the first sensors to be used in an e-nose format [100, 101, 103] and has been included in a heterogeneous sensor array-based instrument [132]. Amperometry is an old electroanalytical technique that en-compasses coulometry, voltammetry, and constant potential techniques, and is widelyused to identify and quantify electroactive species in liquid and gas phases. For liquidphase analytes, the electrodes and analytes are immersed in a common electrolyte andthese have resulted in electronic tongues [102]. In contrast, application of amperome-try to gas-phase analytes involves a unique gas-liquid/solid interfacial transport pro-cess. The AGS is a class of electrochemical gas sensors sometimes called voltam-metric, micro-fuel cell, polarographic, amperostatic, or other names [103, 104]. Thecommon characteristic of all AGSs is that measurements are made by recordingthe current in the electrochemical cell between the working and counter electrodesas a function of the analyte concentration. Figure 4.9 illustrates an amperometric sen-sor consisting of working, counter, and reference electrodes dipped in an electrolyte.The analyte is reacted electrochemically, i.e. oxidized or reduced, and this process,


governed by Faraday’s Law, either produces or consumes electrons at the workingelectrode. The amperometric class of electrochemical sensor complements the othertwo classes of electrochemical sensors, i.e. potentiometric sensors that measure theNernst potential at zero current, and conductometric sensors that measure changes inimpedance [130].The AGS, Figure 4.9, is controlled by a potentiostatic circuit and produces its current

or signal when exposed to a gas/vapor containing an electroactive analyte. The analytediffuses into the electrochemical cell and to the working electrode surface and where itparticipates in a redox reaction. The cell current is directly related to the rate of reactiontaking place at the electrode surface and is described by application of Faraday’s Law,relating the mass, W, of a substance of molecular mass M (grams mol�1) as:

W ¼ Q M

F nð6Þ

where Q is the charge per unit electrode area, F is Faraday’s constant in coulombs/equivalent, and n is the number of electron equivalents per mole of the reacting ana-lyte. Assuming there are no other reacting species in the solution, the observed cur-rent, dQ/dt (t ¼ time) or i, is directly proportional to the amount of analyte, W, that issupplied to the working electrode and, this in turn can be related to the gaseous analyteconcentration (see Eq. 7).The potentiostat allows control of the working electrode thermodynamic potential

while the reaction occurs. The AGS is made reactive towards a variety of analytes bychoosing different potentials, working electrode catalysts, electrolytes, porous mem-branes, and different electroanalytical methods. The working electrode reaction thatproduces current in the example of a CO sensor in Fig. 4.9 is usually taken as:

CO½g� þH2O ¼ CO2 þ 2Hþ½aq� þ 2e�:

The CO diffuses or is pumped to the region of the working electrode, dissolves in theelectrolyte, diffuses to the working electrode surface where it undergoes reaction withsubsequent desorption of the CO2 product and conduction of the 2e� away through themetal electrode. The more CO that is present, the larger the current. Typical currentsare in the micro- or pico-ampere level for ppm level reactants. Response times, mea-

Fig. 4.9 An amperometric gas sensor

4.8 Thermal (Calorimetric) Sensors 9797

sured as time to 90% of signal, have ranged from milliseconds for some oxygen sen-sors to several minutes for other analytes.It is usually preferable that a sensor works in the limiting current region in which

themagnitude of the sensor signal is practically independent of the electrode potential.In theory, the limiting current region can be achieved in any case when the rate-limit-ing step is a step prior to electron transfer. The rate of electrode reactionmay be limitedby the rate of diffusion through a membrane or a capillary that is placed somewherebetween the gas stream containing the analyte and the catalyst layer of the electrode. Insuch cases, the limiting current, ilim, can be written:

ilim ¼ k½CO�gas ð7Þ

where the constant k is the proportionality constant relating the gaseous concentrationto the current in some convenient units like lA (ppmv)�1 (parts per million by vo-lume). The amperometric gas sensor is one of the most widely used sensors for toxicgas detection, i.e. CO, NO, NO2, H2S, SO2, O2, and so on. The AGS was used in the e-nose [105] for one of the earliest determinations of bacterial contamination [106] andidentification of discrete analytes [107]. The AGS has been microfabricated [99, 108]but such versions are not yet commercially available. The main advantages of the am-perometric approach are high sensitivity, a good deal of control over selectivity accom-panied by relatively low cost, small size, and long stable lifetimes.

4.10

Summary of Chemical Sensors

Commercially available-nose instruments listed in Table 4.7 are concentrated on twomain types of chemosensors, such as MOS-type and CP-type. More recent work isbeginning to exploit other sensors for application to the food and drink industriesas listed in Table 4.8. There are a number of books and references in other sectionsof this Handbook that point the user towards the myriad of e-noses that have beenconstructed as well as the various classes and types of sensors. New sensors, includingmicro instruments, will also contribute to the growing number of e-noses that willinevitably lead to an improvement in analytical capability.More and more is being demanded of sensors as time goes on. Quantitative and

qualitative analytical results are not enough and we are requested to answer morepertinent and complex questions such as: Where is the contamination? Is this hazar-dous? Is this pure or the same as something else? These questions are often complexchemically. Sensors provide critical data for the e-nose and other analytical instru-ments that can address such complicated analytical tasks. Without good performancewe have no chance for good data or good answers to these types of questions. Sensorsand sensory data must therefore continue to be improved.


Tab.

4.7

Com

merciallyavailablee-noseinstrumen

ts.Abb

reviations:

metal

oxide

semicon

ductor,M

OS;

organiccondu

ctingpolym

er,C

P;quartzcrystalm

icrobalance,

QCM;surfaceacou

stic

wave,

SAW;gaschromatog

raphy,GC;qu

adrupolemass

spectrom

etry,QMS;

infrared,IR;an

dMOSfield

effect

tran

sistor,MOSF

ET.

Pattern

recogn

ition:artificialneuralnetwork,

ANN;distan

ceclassifiers,DC;

principalcompon

entanalysis,PCA;statisticalpattern

recogn

ition,SPR;discrim

inan

t

functionan

alysis,DFA

;clusteran

alysis,CA;an

dprincipal

compon

ents

regression

,

PCR.

Man

ufacturer

Chem

osensortype

Numberof

sensors

Sizeof

Instrument(CostUS$)

Pattern

recogn

ition

Com

ments

Airsensan

alysis

GmbH

(German

y)

MOS

10Laptop(20000–43

000)

ANN,DC,PCA,SPR

Small,fast

&robu

st

AlphaMOS-M

ulti

Organ

olepticSystems(France)CP,MOS,QCM,SAW

6–24

Desktop

(2000

0–10

000

0)ANN,DFA,PCA

Autosampleran

dair

conditioningunitavailable

AromaScanPLC

(UK)

CP

32Desktop

(20000–75

000)

ANN

Autosampleran

dair

conditioningunitavailable

Array

Tech

QCM

8

Blood

hou

ndSen

sors

Ltd.(U

K)

CP

14Laptop

ANN,CA,PCA

Smallcompan

y,instrumen

t

basedon

research

atLees

University

Cyran

oScien

ceInc.

(USA)

CP

32Palmtop(5000)

PCA

EEVLtd.Chem

ical

Sen

sor

System

(UK)

CP,MOS,QCM,SAW

8–28

Desktop

ANN,DFA,PCA

Electronic

Sen

sor

TechnologyInc.

(USA)

GC,SAW

1Desktop

(19500–25

000)

SPR

Hew

lett-PakardCo.

(USA)

QMS

–Desktop

(79900)

Standardchem

ometrix

HKR-Sen

sorsystemeGmbH

(German

y)

QCM

6Desktop

ANN,CA,DFA,PCA

Smallcompan

y.Based

onre-

search

atUniversityof

Munich

Lennartz

Electronic

GmbH

(German

y)

MOS,QCM

16–40

Desktop

(5500

0)ANN,PCA,PCR

MOSESII

MastiffElectronic

SystemsLtd.

CP

16Sniffedpalmsforpersonal

iden

tification

Nordic

Sen

ser

TechnologiesAB(Sweden

)

IR,MOS,MOSFET,QCM

22Laptop(40000–60

000)

ANN,CPA

Iden

tification

ofpurity,

origin.

RSTRostock

Rau

m-fah

rt

undUmweltschatzGmbH

(German

y)

MOS,QCM,SAW

6–10

Desktop

(5000

0)ANN,PCA

Neotron

icsScien

ceLtd.(U

K)CP

12—

—Medium

size

d

compan

y.

Shim

adzu

Co.

(Japan

)MOS

6Desktop

(70000)

PCA

Largecompan

y.

Saw

tekInc.

SAW

2Palmtop(5000)

Proprietary


Tab. 4.8 Chemosensors used in recent e-nose studies for application

to food and drink industries.

Food or Drink Test Chemosensor type Number of

sensors

Ref.

Alcohols Identification MOS (SnO2) 12 67

Fish (cod, haddock) Freshness MOS (SnO2) 6 83

Fish (squid) Freshness MOS (MgO-In2O3) 9 85

Coffee Discrimination MOS (SnO2) 12 66

Fish Freshness MOS (Ru-In2O3) 1 43, 86

Soup Quality control MOS (Ru-WO3) 4 87

Sea foods (squid, oyster,

sea bream, sardine)

Freshness MOS (Al-ZnO) 1 88–90

Alcohol Freshness MOS (ZnO-SnO2) 1 91

Ground pork/Beef Discrimination and

effect of ageing

Mixed 15 61

Wine Varieties and vintages

of same wine

MOS (SnO2, WO3) 4 92

Beef Freshness MOS (WO3-ZnO) 1 93

Fish (trout) Freshness MOS 8 94

Wheats Grade quality MOS, AGS 4 � 4 95, [106]

Wheats and cheese Discrimination and

ageing

CP 20 96

Cheeses Maturity of cheddars CP 20 97

Coffees Discrimination

between varieties

CP 12 98

Beers Diacetyl taint in

synthetic beer

CP 12 99

Beers Discrimination

between lager and ales

CP 12 100

Liqors Discrimination

between brandy,

gin and whisky

CP 5 101

Boar Taints in meat MOS 14 102

Sausage meats Discrimination MOS 6 103

Water Taints in drinking

water

MOS 4 104

Colas Discrimination

between diet and

normal colas

MOS 6 103

Coffees Discriminate

C. arabica and C. robustaMOS 6 80, 105

Food flavors

(orange, strawberry,

apple, grape, peach)

Flavor identification QCM 8 46

Tomatoes Effect of irradiation

and stress

Mixed 7 106

Whiskies Discrimination

of Japanese whiskies

QCM 8 51


References

1 P. Hauptmann. Sensors-Principles and Ap-plications, (Carl Hanser Verlag & PrenticeHall), 1993, 115–153.

2 J. W. Gardner. Microsensors-Principles andApplication,s (John Wiley & Sons), 1994,224–246.

3 J. W. Gardner, P. N. Bartlett. E-noses-Principles and Applications, (Oxford), 1999,67–100.

4 K. Toko. Biomimetic Sensor Technology,(Cambridge University Press), 2000,92–111.

5 Sensors: A Comprehensive Survey, Eds.W. Gopel, J. Hesse and J. N. Zemel. (VCHVerlagsgesellschaft GmbH, Weinheim,Germany), 1991. Book series.and SensorsUpdate book volume series, baltes, et al.,eds. VCH Verlagsgesellschaft GmbH,Weinheim, Germany. 1997-present.

6 D. W. Ballantine, R. M. White, S. J. Martin,A. J. Ricco, G. C. Frye, E. T. Zellers,H.Wohltjen. Acoustic Wave Sensors, TheoryDesign and Physico-Chemical Applications,(Academic Press, NY) 1997.

7 S. Pezzini, A. Gaiambitto, A. Riva,J. L. Gurnani, C. M. Mari. In: High Tech.Ceramics, Ed. P.Y.Vinecenzini, (Elsevier,Amsterdam), 1987, 2155.

8 H. Nanto, T. Minami, S. Takata. J. Appl.Phys., 1986, 61, 482.

9 H. Nanto, S. Tsubakino, T. Kawai,M. Ikeda, S. Kitagawa, M. Habara. J. Mater.Sci., 1994, 29, 6529.

10 B. Bott, T. A. Jones, B. Mann. Sens.Actuators, 1984, 5, 65.

11 M. Shiratori, M. Katsura, H. Okuma. Proc.1st Sensor Symp. (Japan), 1981, 69.

12 P. J. Shaver. Appl. Phys. Lett., 1967, 11, 255.13 V. Lantto, P. Romppainen, S. Leppavuori.

Sens. Actuators, 1988, 15, 347.14 M. Egashira, Y. Shimizu, Y. Takao. Sens.

Actuators B, 1990, 1, 108.15 Y. Nakatani, M. Sakai, M. Matsuoka. Jpn.

J. Appl. Phys., 1983, 22, 912.16 Y. Nakatani, M. Matsuoka, Y. Iida. IEEE

Trans. Components Hybrid Manufact.Technol., 1982, CHMT-5, 522.

17 Z. Szklarki, B. Zakrzewski, M. Rekas.Thin Solid Films, 1989, 174, 269.

18 P. T. Mosely, D. E. Williams, L.O. W. Norries. Sens. Actuators, 1988, 14,79.

19 M.Miyayama. Tech. Digest of The 15th SensorSymp. (Japan), 1997, 229.

20 Y. Takao, Y. Miya, Y. Tachiyama,Y. Shimizu, M. Egashira. Denki Kagaku,1990, 58, 1162.

21 P. T. Moseley, A. M. Stoneham,D. E. Williams. Techniques and Mechanismsin Gas Sensing Eds. P. T. Mosely,J. O. W. Norries and D. E. Williams,(Adams Higer, Bristol), 1991, 248–267.

22 P. T. Moseley. Sens. Actuators B, 1991, 3,167.

23 A. Dall’Olio, G. Dascola, V. Varacca,V. Bocchi. C. R. Seances Acad. Sci. Paris Ser.C, 1968, 267, 433–435.

24 G. Bidan. Sens. Actuators B, 1992, 6, 45–56.25 J. J. Miasik, A. Hooper, B. C. Tofield.

J. Chem. Soc., Faraday Trans., 1986, I82,1117-1126.

26 G. Gustafsson, I. Lundstrom. Synth. Met.,1987, 2721, 203–208.

27 G. Gustafsson, I. Lundstrom, B. Liedberg,C. R. Wu, O. Inganas. Syth. Met., 1989, 31,163–179.

28 P. N. Batlett, P. B. M. Archer, S. K. Ling-Chung. Sens. Actuators, 1989, 19, 125–140.

29 P. N. Bartlett, S. K. Ling-Chung. Sens.Actuators, 1989, 19, 141–150.

30 P. N. Bartlett, S. K. Ling-Chung. Sens.Actuators, 1989, 20, 287–292.

31 J. W. Gardner, P. N. Bartlett. Nano-technology, 1991, 2, 19–33.

32 (a) G. J. Maclay, C. Yue, M. W. Findlay,J. R. Stetter. Appl. Occupational andEnv. Hygiene, 1991, 6(8), 677–682.(b) U.S. Patent 4,847,594; July 11, 1989.Sensor for Detecting the Exhaustionof an Adsorbent Bed.

33 J. R. Stetter, S. Zaromb, M. W. Findlay Jr.Sens. Actuators A, 1984, 6, 269–288.

34 M. C. Burl, B. J. Doleman, A. Schaffer,N. S. Lewis. Sens. Actuators B, 2001, 72(2):149–159.

35 A. Koll, S. Kawahito, F. Mayer,C. Hagleitner, D. Scheiwiller, O. Brand,H. Baltes. Proc. SPIE, 1998, 3328, 223–232.

36 I. Lundstrom, S. Shivaraman, C. Svensson,L. Lundkuist. Appl. Phys. Lett., 1975, 26,55–57.

37 I. Lundstrom, E. Hedborg, A. Spetz,H. Sundgren, F. Winquist. Sensors andSensory Systems for an E-nose, Eds.


J. W. Gardner and P. N. Bartlett, NATOASISeries (Kluwer, Dordrecht), 1992, 212,303–319.

38 J. N. Zemel. Sens. Actuators A, 1996, 56(1–2) 57–62.

39 G. Z.Sauerbrey. Z, Phys., 1959, 155,206–222.

40 H. Nanto, N. Dougami, T. Mukai,M. Habara, E. Kusano, A. Kinbara,T. Ogawa, T. Oyabu. Sens. Actuators B,2000, 66, 16–18.

41 H. Nanto, Y. Yokoi, T. Mukai, J. Fijioka,E. Kusano, A. Kinbara, Y. Douguchi.Mater.Sci. Eng., 2000, 12, 43–48.

42 S. Kurosawa, N. Kamo, D. Matsui,Y. Kobatake. Anal. Chem., 1990, 62, 353.

43 D. B.Radloff, S. Kurosawa, K. Hirayama,T. Arimura, K. Otake, A. Sekiya,N. Minoura, M. Rapp, Hans-J. Ache. Mol.Cryst. Liq. Cryst., 1997, 295, 141.

44 H. Nanto, Y. Yokoi, Y. Hamaguchi,S. Kurosawa, T. Oyabu, E. Kusano,A. Kinbara. Technical Report of IEICE, 2000,OME 2000-95, 39.

45 T. Matsumoto, K. Tanabe, S. Kurosawa,T. Mukai, H. Nanto. Chem. Software, 2000,22, 85.

46 S. Kurosawa, S. Higashi, H. Aizawa,Dae-Sang.Han. M. Yoshimoto. Chem.Sensors, 2000, 16, 37. (in Japanese)

47 S. Kurosawa, D. Tachiyuki, Das-Sang.Han,H. Aizawa, M. Yoshimoto. Chem. Sensors,2000, 16, 103. (in Japanese)

48 T. Nakamoto, A. Fukuda and T. Moriizumi.Sens. Actuators B, 1993, 10, 85–91.

49 Y. Sakuraba, T. Nakamoto, T. Moriizumi.Trans. Inst. Electron. Comm. Eng., 1990,J73D-II, 1863–1871.

50 J. Ede, T. Nakamoto, T. Moriizumi. Sens.Actuators B, 1993, 13–14, 351–354.

51 K. Ema, M. Yokoyama, T. Nakamoto,T. Moriizumi. Sens. Actuators, 1989, 18,291–296.

52 T. Nakamoto, A. Fukuda, T. Moriizumi,Y. Asakura. Sens. Actuators B, 1991, 3,221–226.

53 T. Nakamoto, A. Fukuda, T. Moriizumi.Sens. Actuators B, 1990, 1, 473–476.

54 H. Muramatsu, E. Tamiya, I. Karube. Anal.Chem., 1990, 63, 399–408.

55 K. Yokoyama, F. Ebisawa. Anal. Chem.,1993, 65, 673–677.

56 Y. Okahata, O. Shimizu. Langmuir, 1987, 3,1171–1172.

57 Y. Okahata, G. En-na, H. Ebata. Anal.Chem., 1989, 62, 1431–1438.

58 M. Ohnishi, T. Ishibashi, Y. Kijima,C. Ishimoto, J. Seto. Sens. Mater., 1992, 1,53–60.

59 S. M. Chang, E. Tamiya, I. Karube, M. Sato,Y. Masuda. Sens. Actuators B, 1991, 5,53–58.

60 D. S. Ballantine, S. L. Rose-Pehrsson,J. W. Grate, H. Wohltjen. Anal. Chem.,1986, 58, 3058–3066.

61 S. L. Rose-Pehrsson, J. W. Grate,D. S. Ballantine, P. C. Jurs. Anal. Chem.,1988, 60, 2801–2811.

62 S. L. Rose-Pehrsson, J. W. Grate. SPIEProc., 1993, 299–311.

63 F. Winquist, E. G. Hornsten, H. Sundgren,I. Lundstrom. Meas. Sci. Technol., 1993, 4,1493–1500.

64 I. Lundstrom, R. Erlandsson, U. Frykman,E. Hedborg, A .Setz, H. Sundgren. Nature,1991, 352, 47–50.

65 K. C. Persaud, G. H.Dodd. Nature, 1982,299, 352–355.

66 H. V. Shurmer, J. W. Gardner, H. T. Chan.Sens. Actuators, 1989, 18, 361–371.

67 H. V. Shurmer, J. W. Gardner, P. Corcoran.Sens. Actuators B, 1990, 1, 256–260.

68 J. W. Gardner, H. V. Shurmer, T. T. Tan.Sens. Actuators B, 1992, 6, 71.

69 J. W. Gardner. Sens. Actuators B, 1991, 4,109–115.

70 J. W. Gardner, E. L. Hines, M. Wilkinson.Meas. Sci. Tech., 1990, 1, 446–451.

71 J. W. Gardner, E. L. Hines, H. C. Tang.Sens. Actuators B, 1992, 9, 9–15.

72 A. D. Walmsley, S. J. Haswell, E. Metcalfe.Anal. Chem., 1991, 250, 257–264.

73 P. Corocoran, P. Lowery. Proc. of the 4th

Inter. Conf. On Artificial Neural Networks,1995, 415–420.

74 B. S. Hoffheins, R. J. Lauf. Sensor ExpoProceedings, 1988, 205, 1–7.

75 X.Wang, S. Yee, P. Carey. Sens. Actuators B,1993, 13-14, 458–461.

76 X. Wang, J. Fang, P. Carey, S. Yee. Sens.Actuators B, 1993, 13–14, 455–477.

77 A. Ikegami, M. Kaneyasu. Proc. of Inter.Conf. on Solid State Sensors and Actuators,1985, 136–139.

78 M. Kaneyasu, A. Ikegami, H. Arima,S. Iwanaga. IEEE Comp., 1987, CHMT-10,267–273.


79 H. Abe, T. Yoshimura, S. Kanaya,Y. Takahashi, Y. Miyashita, S. I. Sasaki.Anal. Chem., 1987, 194, 1–9.

80 H. Abe, S. Kanaya, Y. Takahashi,S. I. Sasaki. Anal. Chem., 1988, 215,155–168.

81 T. Aishima. J. Agr. Food, 1991, 39,752–758.

82 T. Aishima. Anal. Chem., 1991, 243,293–300.

83 T. Nakamoto, H. Takagi, S. Usami,T. Moriizumi. Sens. Actuators B, 1992, 8,181–186.

84 F. A. M. Davide, C. Di.Natale, A. D’Amico.Sens. Actuators B, 1994, 18-19, 244–258.

85 R. Olafsson, E. Martindotti, G. Olafsdotti,O. I. Sigfusson, J. W. Gardner. Sensors andSensory Systems for an E-nose, NATOASI Series E, Ed. J. W. Gardner andP. N. Bartlett, (Kluwer AcademicPublishers, Dordrecht), 1992, 257–272.

86 Yea, R. Konishi, T. Osaki, K. Sugahara.Sens. Actuators A, 1994, 45, 159–165.

87 C. G. Fox, J. F. Alder. Techniques andmechanisms in gas sensing, Eds. P.T.Mosely,I.O.W.Norries and D.E.Williams, (AdamHilger, Bristol), 1991, 324–346.

88 J. W. Grate, S. J. Martin, R. M. White. Anal.Chem., 1993, 65, 940–948.

89 J. W. Grate, S. J. Martin, R. M. White. Anal.Chem., 1993, 65, 987–996.

90 B. Liedberg, C. Nylander, I. Lundstrom.Sens. Actuators, 1983, 4, 299–302.

91 E. Kretschmann. Z. Phys., 1971, 241, 313.92 K. S. Johnston, S. R. Karlson. C. Jung,

S. S. Yee.Mater. Chem. Phys., 1995, 42, 242.93 B. Chadwick, M. Gal. Jpn. J. Appl. Phys.,

1993, 32, 2716.94 S. Nelson, K. S. Johnston, S. S. Yee. Sens.

Actuators B, 1996, 35/36, 187.95 H. Nanto, M. Habara, N. Dougami,

T. Mukai, H. Sugiyama, E. Kusano,A. Kinbara, Y. Douguchi. Tech. Digestof the 7th Inter. Meeting on Chemical Sensors,1998, 695–697.

96 J. White, J. S. Kauer, T. A. Dikkinson,D. R. Walt. Anal. Chem., 1996, 2191–2202.

97 D. S. Ballantine Jr., D. Callahan,G. J. Maclay, J. R. Stetter. Talanta, 1992,39(12), 1657–1667.

98 K. J. Albert, D. R. Walt, D. S. Gill,T. C. Pearce. Anal. Chem., 2001, 73(11),2501–2508.

99 W. J. Buttner, J. R. Stetter, G. J. Maclay.Sens. Mater., 1990, 2, 99–106.

100 (a) J. R. Stetter, S. Zaromb, M. W. Findlay.U.S. Patent 5055266, 1991. (b) J.R.Stetter,S.Zaromb, W.R.Penrose, U.S. Patent4670405, 1987. (c) J.R.Stetter, ChemicalSensor Arrays: Practical Insights andExamples, in Sensors and Sensory Systemsfor an E-nose, Eds. J.Gardner andP.N.Bartlett, (Kluwer Academic Publis-hers). 1992, 273–301.

101 J. R. Stetter. J. Colloid Int. Sci., 1978, 65(3),432–443.

102 F. P. Winquist, P. Wide, I. Lundstrom.Anal. Chim. Acta., 1997, 357, 21–31.

103 S. C. Chang, J. R. Stetter, C. S. Cha.Talanta, 1993, 40(4), 461–467.

104 Z. Cao, W. J. Buttner, J. R. Stetter.Electroanalysis, 1992, 4, 253–266.

105 Artificial Chemical Sensing: Proceedings of theEighth International Symposium on Olfactionand the E-nose (ISOEN 2001), March 26-28,2001, Washington DC., Eds. J. R. Stetter,W. R. Penrose, (The ElectrochemicalSociety, Pennington, NJ), 2001.

106 J. R. Stetter, M. W. Findlar, K. M.Schroeder, C. Yue, W. R. Penrose. Anal.Chim. Acta., 1993, 284, 1.

107 J. R. Stetter, P. C. Jurs, S. L. Rose. Anal.Chem., 1986, 58, 860–866.

108 Buttner, J. William, G. J. Maclay,J. R. Stetter. Sens. Actuators B, 1990, 1,303–307.

109 Y. Takao, Y. Shimizu, M. Egashira. Sens.Mater., 1992, 3, 249.

110 Y. Shimizu, Y. Takao, M. Egashira.J. Electrochem. Soc., 1988, 135, 2539.

111 M. Egashira, Y. Shimizu, Y. Takao. Sens.Actuators B, 1993, 13–14, 443.

112 H. Nanto, H. Sokooshi, T. Usuda. Sens.Actuators B, 1993, 10, 79.

113 H.Nanto, H. Sokooshi, T. Kawai, T. Usuda.J. Mater. Sci. Lett., 1992, 11, 235.

114 H. Nanto, H. Sokooshi, T. Kawai. Sens.Actuators B, 1993, 13–14, 175.

115 H. Nanto, T. Morita, M. Habara, K. Kondo,Y. Douguchi, T. Minami. Sens. Actuators B,1996, 35–35, 384.

116 C. D. Natale et al.. Sens. Actuators B, 1995,24–25, 801.

117 H. Miura et al.. IEE of Japan, 1996, E117,306.

118 M. S. Berberich, S. Vaihinger, W. Gopel.Sens. Actuators B, 1994, 18–19, 282.


119 A. Pisanelli, A. A. Qutob, P. Travers,S. Szyszko, K. C.Persaud. Life Chem.Reports, 1994, 11, 303.

120 K. C. Persaud, P. J. Travers. Handbook ofbiosensors and e-noses, Ed. E. K. Rogers,(CRC Press Inc., Ohio), 1997, 52–59.

121 J. W. Gardner. P. N. Bartlett. Proc. ofOlfaction and Taste XI, (Springer Verlag),1994.

122 J. W. Gardner, T. C. Pearce, S. Friel,P. N. Bartlett, N. Blair. Sens. Actuators B,1994, 18, 240.

123 T. C. Pearce, J. W. Gardner, S. Friel,P. N. Bartlett, N. Blair. Analyst, 1993, 118,371.

124 J. M. Slater, J. Paynter, E. J. Watt. Analyst,1993, 118, 371.

125 B. Bourrounet, T. Talou, A. Gaset. Sens.Actuators B, 1995, 26–27, 250.

126 T. Tan, Q. Lucas, L. Moy, J. W.Gardner,P. N. Bartlett. LC-GC International, 1995, 8,218.

127 A. A. Fekada, E.L. Hines, J. W. Gardner.In: Artificial Neural Networks and GeneticAlgorithms, Eds. R.A.Albrecht, C.R.Reevesand N.C.Steele, (Springer-Verlag,New York), 1993, 691–698.

128 T. Aishima. ASIC 14th Colloque(San Francisco), 1991.

129 F. Winquist et al.. Proc. of 8th Inter. Conf.on Solid State Sensors and Actuators, 1995.

130 N. Barsan, M. Schweizer-Berberich,W. Gopel. Fresenius J. Anal. Chem., 1999,365(4), 287–304.

131 J. B.Miller. IEEE Sensors, 2001, 1, 88.132 J. R. Stetter, S. Zaromb, W. R. Penrose,

M. W. Findlay Jr., T. Otagawa, A. J. Sincali.Portable device for detecting and identifyinghazardous vapors, in: Proc. 1984 HazardousMaterial Spills Conference, April 9–12,Nashville, TN, 1984, 183–194.


5

Signal Conditioning and Preprocessing

R. Gutierrez-Osuna, H. Troy Nagle, B. Kermani, Susan S. Schiffman

5.1

Introduction

The topics covered in this chapter establish the connection between gas sensors andpattern recognition, the two fundamental modules of an odor-sensing instrument thatare covered in Chapters 4 and 6, respectively. A number of electronic circuits are in-volved in integrating pattern analysis algorithms with the underlying chemical trans-ductionmechanisms, as shown in Fig. 5.1. First, the response of the odor sensors (e.g.,a resistance change) needs to be measured and converted into an electrical signal (e.g.,a voltage). This operation is performed by means of interface circuits. Second, theelectrical signal undergoes analog conditioning (e.g., filtering) to enhance its informa-tion content. Third, the analog signal is sampled, digitized and stored in computermemory (not covered in this chapter due to space constraints). Finally, the sampledsignal is digitally preprocessed (e.g., autoscaling) in order to make it suitable for pat-tern analysis.This chapter is organized in three basic parts: interface circuits, signal conditioning,

and preprocessing. Section 5.2 presents the fundamental interface circuits for thethree primary odor sensor types: resistive, piezoelectric, and field-effect. Section 5.3reviews the primary functions performed by analog signal conditioning circuits. Sec-tion 5.4 covers data preprocessing – the first stage of digital signal processing. Theissue of sensor and instrumentation noise, one of the most important factors deter-mining electronic-nose performance, is also reviewed in Section 5.5. The chapter con-

Fig. 5.1 Organization of this chapter


105105

cludes with a review of current instrumentation trends aimed at increasing the selec-tivity of odor sensor systems.

5.2

Interface Circuits

Sensor interface circuits constitute the first stage of electronic instrumentation. Thepurpose of these circuits is to generate an electrical signal that reflects changes in thesensors. Since interface circuits are tightly coupled to the underlying sensing technol-ogy, we will focus our presentation on three widely used odor sensors: conductivity(metal-oxide and conductive-polymer chemoresistors), piezo-electric (surface acousticwave and quartz crystal microbalance), and field effects (metal-oxide field-effect tran-sistors). In addition, this section reviews the issue of operating temperature control,essential for the operation of metal-oxide transducers.

5.2.1

Chemoresistors

In chemoresistive sensors the presence of volatile compounds changes the conduc-tance (or resistance) of the sensing membrane. Interface circuits for these sensorsare, therefore, relatively simple since they only involve measuring resistancechanges. Two types of resistance measurement circuits are commonly used: voltagedividers and Wheatstone bridges. These circuits are presented and analyzed in thefollowing subsections. Linear versions of these circuits that involve operational am-plifiers are presented in section 5.3.5 as a special type of analog signal condition-ing. Finally, AC impedance measurement techniques for chemoresistors are brieflyreviewed at the end of this section.

5.2.1.1 Voltage Dividers

The standard method for measuring large resistance changes is a voltage divider, asshown in Fig. 5.2a. This instrumentation circuit is very popular due to its simplicity.The resistive sensor is placed in series with a load resistor RL and connected to avoltage reference VCC. The current through the sensitive element and load resistancebecomes:

IS ¼ VCC

RS þ RL

ð1Þ

Changes in sensor resistance are then measured as voltage changes across the sensor(VS) or the load resistor (VL). For convenience, we will use the voltage across the loadresistor since it is a single-ended measurement and the subsequent derivation beco-mes simpler. Using Ohm’s Law (V ¼ IR), the resulting output voltage becomes:

5 Signal Conditioning and Preprocessing106

VL ¼ ISRL ¼VCC

RS þ RL

RL ð2Þ

The value of the load resistor should be selected to maximize the sensitivity of thecircuit, that is, the slope of the VL � RS curve, which can be calculated as:

S ¼ @VL

@RS

¼ @

@RS

RL

RS þ RL

VCC

� �¼ VCC

�RL

ðRS þ RLÞ2 ð3Þ

The maximum of the selectivity is finally determined by finding the zeros of its partialderivative with respect to RL:

dS

dRL

¼ @

@RL

�RL

ðRS þ RLÞ2 VCC

!¼ 0 ð4Þ

It can be shown that the optimal load resistor isRL ¼ RS, this is the sensor resistance atthe operating point, typically defined by a reference gas (e.g., clean air). The voltagedivider is the circuit recommended by several metal-oxide sensor manufacturers [1, 2]but it has several shortcomings. First, the relationship between the sensor resistanceRS and the output voltage VL is nonlinear since the current IS through the sensordepends not only on the load resistor but also on the sensor resistance (refer to sec-tion 5.3.5.1 for linearization circuits). Second, andmore importantly, the circuit is onlyappropriate for measuring large resistance changes, such as those typical of metal-oxide sensors. Conducting polymer chemoresistors have sensitivities one order ofmagnitude lower [3] and require the use of Wheatstone bridges.

Fig. 5.2 (a) Voltage divider and (b) Wheatstone bridge for resistive sensors. (c–d) Sensitivity improve-

ments with a gain stage


5.2.1.2 The Wheatstone Bridge

When the resistance changes to be measured are small relative to the baseline resis-tance of the sensor, the information in the output voltage will consist of small fluctua-tions superimposed on a large offset voltage. Although the sensitivity can be boostedwith a gain stage, the problem remains since a large portion of the dynamic range ofthe ADC will be ‘wasted’ in measuring the offset voltage. One solution for measuringsmall resistance changes is to subtract the offset voltage with a second voltage divider,as shown in Fig. 5.2b. The differential voltage in the bridge is:

VOut ¼ RLIS � R2I2 ¼ RL

VCC

RS þ RL

� R2

VCC

R1 þ R2

¼ VCC

RL

RS þ RL

� R2

R1 þ R2

� �ð5Þ

As in the voltage divider of Fig. 5.2a, sometimes called a half-bridge circuit, the ma-ximum sensitivity for the Wheatstone bridge is obtained by choosing resistors R1,R2and RL equal to the sensor baseline resistance. This measurement approach isknown as a deflection method, because the sensor response is measured as a diffe-rential voltage when the bridge becomes unbalanced. An alternative approach, knownas the null method, consists of adjusting the resistors R1, and R2 to cancel the diffe-rential voltage VOUT . The sensor resistance is then obtained from the balance condi-tion:

VOUT ¼ 0 $ R1

R2

¼ RS

RL

! RS ¼ RL

R1

R2

ð6Þ

By comparing Eqs. (5) and (6) it can be inferred that, unlike deflection measurements,the null method is insensitive to fluctuations in the supply voltage. The deflectionmethod, on the other hand, is easier to implement and yields faster responses, ma-king it more appealing for dynamic measurements.It must be noted that the Wheatstone bridge (deflection-method) has the same sen-

sitivity as a voltage divider. Notice that the only difference between Eqs. (2) and (5) isthe offset voltage provided by the R1 � R2 arm, which does not depend on the sensorresistance. The main advantage of the Wheatstone bridge is that it affords higher am-plification gains since the offset voltage has already been removed. To illustrate thispoint, assume a gas sensor that has a resistance that decreases in the presenceof an odor, RS ¼ R0ð1� aÞ. Figure 5.2c shows the response of both circuits forØ = � a � 1=3, R1 ¼ R2 ¼ RL ¼ R0, and VCC ¼ 10V. If this signal is to be capturedwith a data acquisition system that has a dynamic range of 0 V to 10 V, the maximumgain that can be applied to the voltage divider is only 5/3. Although the Wheatstonebridge has the same initial sensitivity (slope), removal of the baseline offset allows amaximum gain of 10, as shown in Fig. 5.2d. The figure also illustrates the nonlinearityintroduced by the deflection measurements.It is important to mention that voltage dividers and Wheatstone bridges can be used

to remove common-mode effects by replacing the load resistor RL with a referencesensor that is shielded from the variable being sensed by the primary sensor but un-shielded from environmental conditions. This approach is widely employed in straingages to compensate for temperature interference, and in pellistors for both tempera-


ture and humidity compensation [4]. The linearized voltage dividers covered in sec-tion 5.3.5.1 are also commonly used for compensation purposes. These types ofmeasurements, based on the ratio between a primary sensor and a reference sen-sor, are known as ratiometric techniques [5].

5.2.1.3 AC Impedance Spectroscopy

Impedance spectroscopic techniques are commonly used to determine the contributionof the different structures in a device (e.g., surface, bulk, grain, and contacts). Impedancespectroscopy is performed by applying a small-amplitude AC voltage to the sensor andmeasuring the resulting current. By sweeping the frequency of the AC signal and mea-suring the impedance at multiple frequencies, an equivalent electrical model can bederived that reveals the contributions of each structure for different gases [6, 7]. Im-pedance spectroscopy requires specialized (and expensive) test and measurementequipment such as impedance analyzers or frequency response analyzers.Several studies have proposed the use of impedance spectroscopy to improve the

selectivity of chemoresistors. Weimar and Gopel [8] have employed two-point mea-surements at frequencies between 1 Hz and 1 MHz to extract the complex impedanceof a custom tin-oxide sensor with interdigitated electrodes. Figure 5.3a shows the Cole-

Fig. 5.3 (a) Cole-Cole impedance plot and equi-

valent circuit for an interdigitated SnO2 sensor [8].

(b) CO and NO2 sensitivity versus frequency of a

SnO2 sensor [9]. (c) Dissipation factor versus

frequency response of a conducting polymer sensor

[10]


Cole impedance plot of a sensor exposed to pure carrier gas, before and after the addi-tion of 10 000 ppmH2. The parameters of the equivalent electrical circuit shown in theupper right corner of the figure were obtained by fitting the impedance modelR1 þ R2kC2 þ R3kQ3 (solid line) to the experimental data (dotted). The resistanceR1 models contributions from the bulk and the surface of the tin oxide. Contributionfrom the SnO2/Pt contacts are modeled by only one parallel component (R2, C2) sincethe two-point setup cannot separate the impedance of the two electrodes. These contactcontributions are responsible for the large semicircle in the figure. The third contri-bution (R3, Q3), caused by migration of surface species along the grain boundaries atlow frequencies, is responsible for the small semicircle in the impedance plot. Thiscontribution becomes inductive in the presence of H2 (notice that the small semicircleis mirrored with respect to the one for synthetic air). This study indicates that sensi-tivity to CO, NO2, and H2 can be improved by measuring the AC impedance of thesensor at DC, 3 kHz, and 20 kHz, respectively. Qualitatively similar conclusions,shown in Fig. 5.3b have been reported [9]. Amrani et al. [10] have performed impe-dance spectroscopy at higher frequencies (100–1000 MHz) to characterize conduct-ing polymer sensors. Their results, summarized in Fig. 5.3c, indicate that methanol,ethyl acetate, and acetone (with dipole moments of 1.69 lD, 1.78 lD and 2.88 lD,respectively) induce peaks in the dissipation factor (the ratio of resistance to reac-tance, R/XC) at different frequencies, with the peak amplitude being a monotonicallyincreasing function of the vapor concentration.

5.2.2

Acoustic Wave Sensors

Instrumentation electronics for acoustic wave gas sensors are more complex thanthose employed for chemoresistors, as they involve AC signals of high frequency(e.g., MHz range). According to the number of piezo-electric transducers used inthe device, acoustic wave sensors can be classified into one-port and two-port devices:

* One-port devices consist of a single transducer that is used both as an input and asan output. The port is used to generate an acoustic signal, which is combined withthe charges induced in the device to produce a measurable impedance change, or ashift in resonance frequency if using an oscillator circuit. A representative sensorfor this type of device is the QMB, also known as a thickness-shear mode sensor.

* Two-port devices, as the name indicates, have separate inputs and outputs. Aninput interdigitated transducer (IDT) is used to induce an acoustic signal, whichpropagates across the surface of the device. When the acoustic wave reaches theoutput transducer, an electrical signal is regenerated, and its phase and/or ampli-tude changes with respect to the input signal are used as measurement variables. Arepresentative two-port device is the SAW delay line sensora.

a) One-port or resonant SAW sensor configurati-ons are also employed. A single IDT is placed inthe center of the device and mechanical ‘groo-ves’ are micro-fabricated on the edges of the

substrate to reflect the acoustic waves backto the IDT, creating a ‘resonant cavity’ in thecenter of the device [12].


Three instrumentation configurations, illustrated in Fig. 5.4, are commonly employedfor acoustic wave sensors: oscillator circuits, vector voltmeters, and network analyzers.Oscillator circuits can be used for one-port (not shown in the figure) and two-portdevices (Fig. 5.4a). The sensor is used as the resonant element in the feedbackloop of an RF-amplifier circuit. Mass changes in the sensitive layer induce shiftsin the resonance frequency, which are measured with a frequency counter. Oscillatorcircuits have several advantages, including low cost, relative simplicity, and excellentfrequency stability [11]. However, these circuits generally provide information aboutwave velocity, and not amplitude, which may be necessary to monitor wave attenua-tions. A second configuration, shown in Fig. 5.4b, overcomes this limitation, providingboth wave velocity and amplitude measurements in two-port devices. A signal genera-tor is used to supply an RF voltage to the input transducer, and a vector voltmetermeasures phase and amplitude changes at the output IDT relative to the input sig-nal. Vector voltmeters are, however, relatively expensive pieces of laboratory equip-ment, and their phase measurements are 10–100 times less sensitive than frequencymeasurements with oscillator circuits. A third alternative, shown in Fig. 5.4c, is to usea network analyzer to perform a complete characterization of the device at multiplefrequencies [11, 12].To compensate for interferents (e.g., temperature, pressure, drift), SAW sensors are

typically used in the dual configuration illustrated in Fig. 5.4d. One delay line is coated

Fig. 5.4 Instrumentation configurations for acoustic wave sensors:

(a) oscillator circuit, (b) impedance meter, and (c) network analyzer.

(d) Dual delay SAW structure for temperature compensation [3, 11, 12].

(e) QMB sensor interface circuit [15]


with a sensing film that responds strongly to odors, and the second line is used as areference to capture only interferent effects. Subtraction of the two signals yields ameasurement that is, theoretically, independent of the common-mode interferents[13]. Fig. 5.4e shows a compact, low-power circuit for a QMB sensor [14, 15]. A10 MHz sensor crystal is connected to an integrated oscillator whose output frequencydecreases when odor molecules are absorbed into the crystal coating. The output of thesensor oscillator is compared to a reference oscillator with an uncoated 10 MHz crystalby means of a D flip-flop, which generates the difference frequency.

5.2.3

Field-Effect Gas Sensors

As described in Chapter 4, two configurations can be used inmetal-insulator-semicon-ductor field-effect gas sensors: capacitor (MISCAP) and transistor (MISFET). The twostructures depicted in Fig. 4.4 of Chapter 4 yield similar information, the differencesbeing in the required measurement circuitsb. In the case of MISCAP sensors, changesin the voltage-capacitance curve can be measured with a small AC-voltage (e.g.,1 MHz) superimposed on a DC-potential [16]. Changes in the ID � VG curve of MIS-FET sensors, on the other hand, may be measured with constant-voltage [17] or con-stant-current circuits [18]. Figure 5.5 shows a conventional two-terminal arrangement

for an n-channel MISFET with a common gate-drain configuration, and a possibleconstant-current interface circuit. The shift in the VGDS � ID curve upon exposureto volatile organic compunds is the change in the threshold voltage, which is inturn related to the shift in work function, surface states, and charge. A current sourceis used to inject a constant current into the drain, and the resulting voltage VGDS isbuffered (see Section 5.3.2) and sampled to create a time-resolved signal. Field-effectsensors operate at high temperatures (100–200 8C for Si substrates, up to 700 8C for

b) MISCAPs have a simpler structure and are,therefore, often used for exploratory work [16]

Fig. 5.5 MISFET gas sensors: (a) two-terminal configuration

and (b) possible constant-current interface circuit [18, 19]


SiC) and, like metal-oxide chemoresistors, require temperature control circuits. Field-effect sensors also suffer from baseline drift, which can be compensated for by usingdifferential configurations having an active gate FET and a passive reference FET [16].

5.2.4

Temperature Control

Metal-oxide gas sensors are commonly operated in the so-called isothermal mode, inwhich the temperature of the sensor is kept constant during exposure to odorsc. Thesimplest and most widely used method for pseudo-isothermal operation consists ofapplying a constant voltage across the terminals of the resistive heater RH , as shown inFig. 5.6a. Temperature stability is achieved by using heater materials with a positivetemperature coefficientd so that the thermoresistive effect serves as negative feedback[20]. This simple constant-voltage operationmay be used when temperature stability isnot critical.

Improved stability (e.g., to gas-flow cooling effects) may be achieved by controllingthe heater resistance rather than the heater voltage [21]. In constant-resistance opera-tion, the sensor heater is placed in a Wheatstone bridge and compared against a re-ference potentiometer that determines a set-point resistance, as shown in Fig. 5.6b.Deviations from the set-point resistance result in a differential voltage across thebridge, which is used to control a current or voltage source. Capteur Ltd. implementsconstant-resistance control by using a FET operating as a voltage-controlled currentsource [22]. Constant resistance, however, requires heater materials with a reasonablyhigh thermoresistive coefficient.A third alternative consists of embedding a temperature sensor in the substrate [8],

or using the heater as a temperature sensor [24, 25]. The latter method, however, also

Fig. 5.6 (a) Constant heater voltage and (b) constant heater resistance circuits [20]

c) If the sensor is normally operated at lowtemperature, it is then necessary to shift to ahigh temperature to burn off excess organiccontaminants from the sensor surface [28].

d) The heater resistance RH is a function of tem-perature T: RH ¼ R0ð1þ aTÞ, where R0 is thebaseline resistance at zero degrees and a is thetemperature coefficient. For positive a, theheater resistance increases with temperature.


requires a large positive thermoresistive coefficient, which is not the case for certaincommercial metal-oxide sensors [26]. Sensor surface temperatures can also be mea-sured with infrared thermometers, but these measurements have been shown to berather inaccurate [26]. Additional temperature control strategies may be found in theliterature [27].

5.3

Signal Conditioning

The electrical signals generated by sensor interface circuits are often not adequate foracquisition into a computer, and must be further processed by a number of analogsignal conditioning circuits. The four basic roles of these circuits: buffering, ampli-fication, filtering, and special functions, are surveyed in the following subsectionsalong with a brief review of operational amplifiers.

5.3.1

Operational Amplifiers

Operational amplifiers (op-amps) are analog integrated circuits widely used to imple-ment a variety of instrumentation circuits. Although a thorough coverage of op-ampsis beyond the scope of this chapter, we provide a brief review that will allow the readerto analyze the circuits presented in the remaining sections of this chapter. An op-amp,shown in Fig. 5.7a, is essentially a high-gain amplifier that generates an output voltageV0 ¼ GOLVd proportional to the difference voltage Vd between a noninverting (þ) andan inverting input (�). The power necessary to perform the signal amplification(GOL ffi 104 � 106) is derived from the supply voltages (�VS) and, therefore, the out-put voltage V0 is constrained by �VS � V0 � þVS. Op-amp circuits in this open-loopconfiguration are not practical since very small difference voltages Vd will drive theoutput voltage to saturation. In addition, the open-loop gain GOL has a limited band-width (GOL decays significantly with frequency), and is very sensitive to temperatureand power supply fluctuations. For these reasons, op-amps circuits typically contain afeedback loop to control the gain, as shown in Fig. 5.7b.A large number of these op-amp feedback circuits can be analyzed by assuming ideal

op-amp characteristics, primarily (1) infinite open-loop gain and bandwidthGOLðf Þ ¼ 1, (2) infinite input impedance ZIN ¼ 1, and (3) zero output impedanceZOUT ¼ 0. The latter simply implies that loading effects are negligible, that is,V0 ¼ VOUT in the equivalent op-amp circuit of Fig. 5.7a. These ideal characteristicslead to two ‘golden rules’ that are sufficient for analyzing many practical op-amp feed-back circuits [23, 29]:

* Rule 1: Inputs stick together. Since the gain is infinite and VOUT must be bounded,the feedback network will enforce an output VOUT that cancels the differential vol-tage Vd ¼ 0.

* Rule 2: Inputs draw no current. This follows from the assumption that ZIN ¼ 1.


To illustrate the use of these rules, we derive the transfer function of the circuit shownin Fig. 5.7b. From Rule 1 we can establish that the voltage at the noninverting input isequal to the input voltage VIN. This allows us to express the current i1 flowing throughresistor R1 as i1 ¼ VIN=R1. Since the noninverting input does not draw current(Rule 2), we infer that the current i2 through resistor R2 is equal to i1. As a result,the voltage at the output becomes:

VOUT ¼ VIN þ R2i2 ¼ VIN þ R2

VIN

R1

¼ VIN

R2

R1

þ 1

� �¼ VINGCL ð7Þ

This circuit is known as a noninverting amplifier since it provides an amplificationgain GCL while preserving the phase (sign) of the input voltage VIN.

Fig. 5.7 (a) Op-amp simplified internal model and (b) analysis of

feedback circuits. Amplifier circuits: (c) buffer, (d) inverting amplifier,

(e) difference amplifier, and (f) instrumentation amplifier


5.3.2

Buffering

The first and simplest application of op-amps is buffering, which is required to isolatedifferent electronic stages and avoid impedance-loading errors. An analog buffer canbe implemented with the voltage-follower circuit shown in Fig. 5.7c. This circuit pro-vides (assuming an ideal op-amp) infinite input impedance and zero output impe-dance.

5.3.3

Amplification

An amplification or gain stage is typically required to bring the signal of the interfacecircuits to a level that is suitable for the dynamic range of a subsequent analog-to-digital converter. Amplifier circuits can be broadly classified into single-ended or dif-ferential. A single-ended signal VIN, such as the one from a voltage divider, can beamplified with the noninverting amplifier described earlier in Fig. 5.7b or its invertingcounterpart shown in Fig. 5.7d, in which the feedback resistor has been replaced by apotentiometer to allow for manual adjustments of the gain.In the case of Wheatstone bridge interface circuits, a differential amplifier stage,

such as the one shown in Fig. 5.7e, may be used. This simple design, however, pre-sents two basic drawbacks. First, the input impedance is significantly reduced sincethe R1 resistors are in series with the input signals. Second, accurate matching of theresistor pairs (RA1 ¼ RB1) and (RA2 ¼ RB2) is required to ensure that the differentialgains are similar and, therefore, provide good common-mode rejection. Due to theselimitations, the so-called ‘instrumentation amplifiers’ are commonly used as differ-ence stages. Fig. 5.7f shows a classical instrumentation amplifier design with threeop-amps that can achieve high input impedance and common-mode rejection ratiowithout critical resistor matching [23]. The two op-amps at the input stage providehigh differential gain and unity common-mode gain, whereas the second stage gen-erates a single-ended output. Integrated instrumentation amplifiers are convenientlyavailable from several manufacturers, with all components internal to the chip exceptfor R2, which can be connected externally to provide a programmable gain.

5.3.4

Filtering

Analog filters are used to remove unwanted frequency components from the sensorsignals. Filters can be broadly grouped into four classes according to their frequencyresponse [30, 31]: low-pass, high-pass, band-pass, and band-reject (Fig. 5.8). Low-passfilters allow frequencies below a cutoff frequencye to pass, while blocking frequencies

e) The cutoff frequency is defined as thefrequency at which the gain is reducedby 3 dB (or a signal ratio of 0.707)


above the cutoff. High-pass filters perform the opposite function, passing only fre-quencies above a cutoff. Band-pass filters allow passage of frequencies within aband. Band-reject (or notch) filters allow passage of all frequencies except for thosewithin a, typically narrow, band.These analog filters can be implemented using passive or active circuits. Passive

filters consist of networks of resistors, capacitors, and inductors, whereas active filtersutilize active components (e.g., op-amps, transistors), in addition to passive devices,e.g. resistors and capacitors. Active filters are capable of implementing ‘virtual’ induc-tors by placing capacitors in the feedback loop, thus avoiding the bulk and nonlinearityof inductorsf. Active filters are suitable for low frequency, small signals, and are pre-ferred over passive filters because they can have gains greater than 0 dB. Conversely,active filters require a power supply and are limited by the bandwidth of the activeelement. Passive filters have the advantage of being low-noise. Fig. 5.9a shows a pas-

sive implementation of a first-order Butterworth (low-pass) filter, with a cut-off fre-quency FC ¼ ð2pR2C2Þ

�1 and a roll-off slopeg of 20 dB/decade. Figure 5.9b showsan equivalent implementation with an inductor and a resistor. The active circuitshown in Fig. 5.9c also has a similar frequency response plus a static gain ofR2=R1. Finally, integrated circuits with low-pass, high-pass, band-pass, and band-re-ject outputs are also available in a single package from several manufacturers. Thesecircuits, known as state-variable filters, are provided with extensive design formulasand tables and can be easily configured using only external resistors.

Fig. 5.8 Frequency response of analog filters

f ) Active filters could also use inductors, althoughthey usually do not.

g) Steeper roll-offs may be achieved by cascadingseveral filters in series.

Fig. 5.9 Low-pass first order filters: (a, b) passive and (c) active


5.3.5

Compensation

A number of special functions may be implemented with analog circuits to compen-sate for deficiencies, cross-sensitivities, and nonlinearities in the sensor response, andreduce the computational load of a subsequent digital signal processing stage. Thesecircuits perform various functions including linearization, integration, differentiation,logarithmic and antilogarithmic conversion, peak-to-peak and phase detection, andtemperature compensation [29]. We now introduce several interface circuits for che-moresistors that can be used to obtain linear resistance-voltage relationships. Thesecircuits are presented here, rather than in Section 5.2.1 with the remaining interfacecircuits, because they require familiarity with op-amps and they perform a compensa-tion function. Additional compensation circuits for concentration and temperature arereviewed in Section 5.3.5.2.

5.3.5.1 Linearization of Resistance Measurements

Among other shortcomings, voltage dividers have a nonlinear resistance-to-voltagetransfer function. As a result, the sensitivity of the circuit is not constant over thedynamic range of the sensor. The resistance-to-voltage relationship can be easily lin-earized, however, by driving the sensing element at constant-voltage or constant-cur-rent. Figure 5.10a illustrates a constant-voltage measurement circuit that employs avirtual ground at the inverting input of the operational amplifier to apply a constantvoltage VCC across the sensor RS [20]. Negative feedback through a load resistor gen-erates an output that changes linearly with the sensor conductance GS (the inverse ofsensor resistance RS):

VOUT ¼ �ISRL ¼ �VCC

RS

RL ¼ �VCCRLGS ð8Þ

An additional advantage of this circuit is that the load resistor RL can be chosen toprovide different amplification gains.Constant-current excitation is illustrated in Fig. 5.10b. The current IS through the

sensor is entirely determined by the load resistor since the voltage at the op-amp in-verting input is constant and equal to VCC [4]. The differential voltage across the sensoris then linearly proportional to the sensor resistance:

VOUT ¼ RSIS ¼ RS

VCC

RL

ð9Þ

A similar constant-current arrangement can be used to provide a linear resistance-voltage relationship in Wheatstone bridges, as shown in Fig. 5.10c [4]. The operationalamplifier provides a virtual ground to the midpoint of the sensor arm, generating aconstant current through the sensor:

IS ¼ VCC

R0

ð10Þ


The voltage at the output of the op-amp is then proportional to the sensor resistance:

V0 ¼ �RSIS � RS

VCC

R0

ð11Þ

and the output of the circuit becomes:

VOut ¼1

2VCC 1� RS

R0

� �¼ 1

2VCC 1� R0ð1� aÞ

R0

� �¼ 1

2VCCa ð12Þ

5.3.5.2 Miscellaneous Functions

A number of miscellaneous compensation functions may be performed with analogcircuits. Figure 5.11a shows a logarithmic amplifier that may be used to compensatefor the power-law concentration-resistance relationship R / ½C��b of metal-oxide che-moresistors [32] and provide an output voltage proportional to the log concentrationlog[C] of the analyteh. Figure 5.11b illustrates a circuit that is employed in commercial

Fig. 5.10 Linearizing a voltage divider through constant-voltage (a)

or constant-current (b) measurements. Linearization of a Wheatstone

bridge with a constant-current arrangement (c)

Fig. 5.11 Special functions: (a) logarithmic amplifier

and (b) temperature compensation [1]

h) The relationship VBE / logðICÞ may be usedto derive the logarithmic transfer function.This simple circuit, however, requires additional

compensation for oscillations and ambienttemperature [29].


gas alarm circuits to compensate for temperature [1, 2]. The circuit includes a thermis-tor RTH (temperature dependent resistor) that adapts the alarm reference voltage VREF

according to ambient temperature. The schematic in Fig. 5.11b uses a voltage regulator(7805) to provide a stable 5 V DC supply voltage to the heater and the voltage divider.Finally, the output of the comparator is current-boosted with an NPN transistor inorder to drive an alarm.

5.4

Signal Preprocessing

Following an appropriate conditioning stage, the sensor array signals are digitized andeither processed online or stored for future analysis. Due to space constraints, thereader is referred to the existing literature [30, 33] for a review of data acquisitionfor sensor systems (e.g., sample/hold, anti-aliasing, and analog-to-digital conver-sion). It is important to mention, however, that in order to avoid aliasing effects,the sampling rate during data acquisition should be at least twice the highest fre-quency in the sensor response. This is known as the Nyquist sampling theorem [34].With this in mind, we focus our attention on signal preprocessing, the first com-

putational stage after the sensor array data has been sampled and stored into computermemory. The goal of signal preprocessing is to extract relevant information from thesensor responses and prepare the data for multivariate pattern analysis (covered inChapter 6). The choice of signal preprocessing is critical and can have a significantimpact on the performance of subsequent modules in the pattern analysis system[35]. Although signal preprocessing is somewhat dependent on the underlying sensortechnology, three general stages can be identified [36]: baseline manipulation, com-pression, and normalization.

5.4.1

Baseline Manipulation

The first stage of preprocessing consists of manipulating the sensor responsewith respect to its baseline (e.g., response to a reference analyte) for the purposesof drift compensation, contrast enhancement and scaling. Considering the dynamicresponse of the sensor xSðtÞ shown in Fig. 5.12a, three techniques are commonly em-ployed [3]:

* Differential: the baseline xSð0Þ is subtracted from the sensor response. As a result,any additive noise or drift dA that may be present in the sensor signal is effectivelyremoved from the preprocessed response ySðtÞ:ySðtÞ ¼ ðxSðtÞ þ dAÞ � ðxSð0Þ þ dAÞ ¼ xSðtÞ � xSð0Þ ð13Þ


* Relative: the sensor response is divided by the baseline. Relative measurementseliminate the effect of multiplicative drift dM and provide a dimensionless responseySðtÞ:

ySðtÞ ¼xSðtÞð1þ dMÞxSð0Þð1þ dMÞ

¼ xSðtÞxSð0Þ

ð14Þ

* Fractional: the baseline is subtracted and then divided from the sensor response.Fractional measurements are not only dimensionless but also normalized since theresulting response ySðtÞ is a per-unit change with respect to the baseline, whichcompensates for sensors that have intrinsically large (or small) response levels:

ySðtÞ ¼xSðtÞ � xSð0Þ

xSð0Þð15Þ

The choice of baseline manipulation technique and response parameter xSðtÞ (e.g.,resistance, conductance, frequency) is highly dependent on the sensor technologyand the particular application, but a few general guidelines can be extracted from

Fig. 5.12 Gas sensor transient response to an

odor pulse (a). Transient analysis approaches:

(b) sub-sampling, (c) parameter-extraction, and

(d) system-identification


the literature. Gardner et al. [37, 38] have shown that the fractional change in conduc-tance ySðtÞ ¼ ðGSðtÞ �GSð0Þ=GSð0Þ provides the best pattern-recognition perfor-mance for (n-type) MOS chemoresistors, compensating for temperature cross-sensi-tivity and nonlinearities in the concentration dependence [39]. Fractional methods forMOS chemoresistors are also widely used [40, 41]. In the case of conducting polymerchemoresistors, fractional changes in resistance are commonly employed, both inresearch prototypes and in commercial instruments [42, 43]. For piezo-electricoscillators, where the response xSðtÞ being monitored is a frequency, differentialmeasurements with respect to a reference analyte (and/or an uncoated reference sen-sor) are commonly used [12, 44]. Differential measurements are also widely used forMOSFETs [45, 46], where the response xSðtÞ is a voltage shift in the I(V) curve asdescribed in Section 5.2.3. Finally, a number of variations of these three basic base-line-manipulation techniques have been proposed in the literature, including data-driven procedures to optimize the baseline-manipulation stage for specific applica-tions [35, 36, 47].

5.4.2

Compression

The second stage in preprocessing is aimed at compressing the sensor-array responsedown to a few descriptors to form a feature vector or fingerprint. In most cases this isperformed by extracting a single parameter (e.g., steady-state, final, or maximum re-sponse) from each sensor, disregarding the initial transient response, which may beaffected by the fluid dynamics of the odor delivery system (covered in Chapter 3). How-ever, with careful instrument design and sampling procedures, transient analysis cansignificantly improve the performance of gas sensor arrays:

* Improved selectivity. The dynamic response to an odor exposure (and the subse-quent odor recovery) carries a wealth of odor-discriminatory information that can-not always be captured with a single parameter. In some situations, transient para-meters have also been reported to exhibit better repeatability than static descriptors[48–50]. Therefore, sensor transients can be used as dynamic fingerprints to im-prove selectivity by pattern-recognition means.

* Reduced acquisition time. The duration of the acquisition cycles can be signifi-cantly shortened if the initial sensor transients contain sufficient discriminatoryinformation, avoiding the lengthy acquisition times required to reach steady state[51]. As a consequence, the sensors also require less time to recover their baseline, aprocess that can be particularly slow when the target odors have high concentra-tions.

* Increased sensor lifetime. By reducing the duration of the odor pulse and, thereforeminimizing irreversible binding, the lifetime of the sensors can also be increased.

For these reasons, transient analysis has received much attention in recent years. Ac-cording to the procedure employed to generate the dynamic fingerprint, transientcompression methods can be broadly grouped into three classes:


* Sub-sampling methods: As depicted in Fig. 5.12b, these methods exploit dynamicinformation by sampling the sensor transient response (and/or its derivatives) atdifferent times during the odor exposure and/or odor recovery phase[36, 49, 52, 53].

* Parameter-extraction methods: These methods compress the transient responseusing a number of descriptors, such as rise times, maximum/minimum responsesand slopes, and curve integrals. [48, 54–56].

* System-identification methods: These methods fit a theoretical model (e.g., multi-exponential, auto-regressive) to the experimental transients and use themodel para-meters as features [55, 57, 58].

Exponential curve-fitting methods can result in nearly lossless compression of thesensor transients, but are computationally intensive [57, 59]. For these reasons, sub-sampling and parameter-extraction methods are more commonly employed. A finalword of caution regarding the use of transient information: a large number of dynamicparameters will require an exponentially increasing number of training examples inorder to prevent the pattern recognition system from over-fitting the data. Alterna-tively, one may use resampling techniques (e.g., cross-validation, bootstrap) or regu-larization (e.g., shrinkage, weight decay) to control the complexity of the model.Further details on small-database considerations and dynamic pattern-recognitionmethods may be found in Chapter 12 of this Handbook.

5.4.3

Normalization

Normalization constitutes the final stage of digital preprocessing prior to multivariatepattern analysis. Normalization techniques can be broadly grouped in two classes:local and global methods. Local methods operate across the sensor array on each in-dividual “sniff” in order to compensate for sample-to-sample variations caused byanalyte concentration and sensor drift, among others. Global methods, on the otherhand, operate across the entire database for a single sensor (e.g., the complete historyof each sensor), and are generally employed to compensate for differences in sensorscaling. In what follows, we will denote by xðkS the response of sensor ‘s’ to the k-thexample in the database.

5.4.3.1 Local Methods

Themost widely used local method is vector normalization, in which the feature vectorof each individual ‘sniff’ is divided by its norm and, as a result, forced to lie on a hyper-sphere of unit radius, as shown in Fig. 5.13d,e:

yðkS ¼ xðkSffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiPSðxðkS Þ

2r ð16Þ


Vector normalization can be employed to compensate for differences in concentrationbetween samples. Assuming the power-law relationship xðks;a ¼ as;a½C

ðka �b of metal-oxi-

de chemoresistors [32], where xðks;a is the response of sensor ‘s’ to the k-th sample ofodor ‘a’, as;a is the sensitivity of sensor ‘s’ to odor ‘a’, and ½Cðk

a � is the concentration ofthe k-th sample of odor ‘a’, then:

yðks;a ¼xðks;affiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiP

sxðks;a� �2r ¼

as;a Cðka

h ib

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiPs

as;a Cðka

h ib� �2s ¼

as;affiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiPs

as;a

� �2r ð17Þ

To the extent that these simplifying assumptions hold, vector normalization can the-refore be used to compensate for sample-to-sample variations in concentration. In thiscontext, vector normalization can be applied in situations when each odor has a uniqueconcentration, but discrimination is to be performed on the basis of odor quality (e.g.,the direction of the response vector x!ðk

a ) rather than odor intensity (e.g., the magni-

tude of x!ðka ). Conversely, thismethod should not be used when the vector amplitude is

known to carry relevant information.

Fig. 5.13 Normalization procedures: (a,d) raw data, (b) sensor

autoscaling, (c) sensor normalization and (e) vector normalization


5.4.3.2 Global Methods

Two global procedures are commonly employed in electronic nose systems:

* Sensor autoscaling, in which the distribution of values for each sensor across theentire database is set to have zero mean and unit standard deviation:

yðks ¼ xðks �mean½xs�std½xs�

ð18Þ

* Sensor normalization, in which the range of values for each individual sensor is setto [0,1]. This is simply done by subtracting the minimum and dividing by the rangeof the sensor across the entire database:

yðks ¼ xðks �min8k½xðks �

max8k½xðks � �min8k½x

ðks �

ð19Þ

Global methods are typically used to ensure that sensor magnitudes are comparable,preventing subsequent pattern-recognition procedures from being overwhelmed bysensors with arbitrarily large values. For instance, nearest-neighbors proceduresare extremely sensitive to feature weighting, and multilayer perceptrons can saturatetheir sigmoidal activation functions for large inputs. Sensor normalization makes fulluse of the input dynamic range but, as illustrated in Fig. 5.13a,c, is very sensitive tooutliers since the range is determined by data outliers. Autoscaling, on the other hand,cannot provide tight bounds for the input range but is robust to outliers. However, itmust be noted that both techniques can amplify noise since all the sensors (particularlythose which may not carry information) are weighted equally.Logarithm metrics have also been used to compensate for highly nonlinear concen-

tration effects [41]. It is also worthmentioning the Box-Cox transform [60], which couldbe employed to compensate for nonlinearities, as well as compress the dynamic rangeof the sensors:

yðks ¼ðxðks Þk�1

k k 6¼ 0

ln xðks� �

k ¼ 0

8><

>:ð20Þ

5.5

Noise in Sensors and Circuits

Noise is generally considered to be any unwanted effect that obscures the detection ormeasurement of the desired signal. As shown in Fig. 5.14a, noise can arise at variousstages in the measurement process, including the quantity under measurement itself,the sensors, the analog processing system, the data acquisition stage and the digitalsignal processing system. Among these, noise in the early measurement stages isclearly most harmful as it propagates and can be potentially amplified through the


subsequent stages in the signal pathway [61]. Several noise sources, such as thermaland shot noise, are inherent to the underlying physics of the sensors or electroniccomponents and are, therefore, irreducible. Other types of noise, conversely, are ori-ginated from processes that could be avoided, and include 1/f noise, transmission andquantization noise.Thermal noise, also known as Johnson or Nyquist noise, arises in any medium that

dissipates energy, such as a conductor. This means that even a simple resistor is anoise source. The open-circuit noise voltage generated by a resistance R isVnoise ¼

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi4kTRDf

p, where k is Boltzman’s constant, T is the absolute temperature

(Kelvins), and Df is the bandwidth (Hz) over which the measurement is made[23]. Therefore, the larger the resistance, the more noise it can introduce. Thermalnoise has a flat power spectral density (PSD), and is oftentimes called white noisein analogy to white light, which has a flat distribution of all frequencies in thevisible spectrum. In addition, the amplitude distribution of thermal noise is Gaussian[23].Shot or Schottky noise arises from the random fluctuations in the number of charge

carriers (electrons and holes) that cross a potential barrier in the charge flow, and istypical of p-n junctions in diodes and transistors. The shot-noise RMS current fluctua-tion is Inoise ¼

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2qlDCDf

p, where q is the electron charge, IDC is the average current

through the barrier, andDf is the bandwidth. Shot noise is also white andGaussian [4].1/f (read ‘one-over-f’) or flicker noise is considered to arise from imperfections in the

manufacturing process of electronic components. As the name indicates, 1/f noise has

Fig. 5.14 (a) Sources of noise in sensor systems. (b) Power spectral

density of white and 1/f noise. (c) Quantization noise in A/D conversion


a PSD that is inversely proportional to frequency. For this reason it is also known aslow-frequency or pink noise (red is at the low side of the visible spectrum). It is alsoreferred to as excess noise because it appears in addition to white noise, as illustratedin Fig. 5.14b. 1/f noise is most pronounced at frequencies below 100 Hz, where manysensors operate, and becomes barely noticeable at frequencies above a few hundredKHz where white noise dominates. In contrast with thermal noise, which equallyaffects a cheap carbon resistor or the most carefully made resistor, 1/f noise canbe reduced by using good quality metal film or wire-wound resistors at the early stagesof sensor interface circuits [23].Noise can also be transmitted from interferences such as fluctuations in the DC

power supply, 50–60 Hz pickup, changes in ambient temperature, capacitive or in-ductive couplings, and ground loops. A careful layout and construction of the electro-nics, with proper shielding and grounding, must be used to reduce electromagneticinterference noise to acceptable levels [23]. In addition, differential measurements,such as the ones in Fig. 5.4d,e, can be employed to compensate for noise effectsthat are additive in nature. Multiplicative effects, on the other hand, can be reducedby means of ratiometric measurement techniques [5]. Analog filtering (Section 5.3.4)and digital signal preprocessing (Section 5.4) can also be employed to further reducenoise. For instance, differentiation can be used to reduce low-frequency noise (e.g.,drift) at the expense of amplifying high-frequency components. Conversely, integra-tion or averaging reduces high-frequency noise while amplifying low-frequency com-ponents.As mentioned earlier, noise can also arise in the latter stages of the signal pathway,

primarily during analog-to-digital conversion, when the continuous sensor signals areconverted into a discrete subset of values and stored in computer memory. This pro-cess introduces nonlinear quantization errors that can be treated as an additional noisesource, as depicted in Fig. 5.14c. Quantization noise must be controlled by selecting anappropriate gain in the signal conditioning circuits to fully utilize the dynamic range ofthe analog-to-digital converter, and by employing differential measurements to re-move uninformative baseline offsets in the sensor response [62]. Limitations in ma-chine precision and fixed-point arithmetic can also introduce digital noise in the signalpathway. For a systematic treatment of quantization and finite word-length noise, thereader is referred to the literature [34].Finally, it is important to notice that the inherent drift and poor repeatability of the

sensor responses can sometimes be significantly larger than most of the other noisesources described in this section, effectively limiting the sensitivity of electronic nosesystems. As proposed previously [61], the global effect of all these noise sources can becombined into a single parameter called the noise-equivalent concentration, whichindicates the gas concentration that results in a unit signal-to-noise ratio.


5.6

Outlook

From their original conception as arrays of homogeneous gas sensors with overlap-ping selectivities, electronic-nose systems, including those commercially available, areslowly evolving towards hybrid arrays that take advantage of several sensor technolo-gies [63]. The use of sample preconditioning such as thermal-desorption units or chro-matographic columns, is also becoming increasingly popular as the means to increasethe sensitivity and selectivity of the instrument [64–66]. An additional trend in elec-tronic-nose systems has become the measurement of multiple parameters from thesame sensing membrane [67]. We focus our attention on the latter, since the use ofhybrid systems does not introduce conceptual problems other than the integration ofthe various sensor technologies into a single package, and sample preconditioningmethods are covered in Chapter 3 of this Handbook. Multiparameter sensing ap-proaches can be broadly grouped in three categories:

* Similar sensing layer but different transduction principles: these systems extractmultiple physical parameters from the same sensing layer, such as work functionand conductance on MOS sensors, or resistance and mass changes in conductingpolymer sensors.

* Similar sensing layer and transduction principle but different operating modes: inthis case, the selectivity of the sensor is modified by modulating the operating con-ditions, such as temperature cycling in MOS sensors or AC impedance spectro-scopy in MOS or conducting polymer sensors.

* Similar sensing layer, transduction principle, and operating modes but differentfeatures: A third possibility is to extract multiple parameters from the sensor tran-sient response.

In this section, we review a multiparameter technique for metal-oxide sensors that hasreceived much attention in recent years: temperature modulation. AC impedancespectroscopy and transient analysis, which can also been used as multiparameter ap-proaches to improve the selectivity of gas sensors, were covered in Sections 5.2.1.3 and5.4.2, respectively. For additional material onmultiparameter sensor systems the read-er is referred to the authoritative review of Weimar and Gopel [67].

5.6.1

Temperature Modulation

The selectivity of metal-oxide sensors is greatly influenced by the operating tempera-ture of the device, since the reaction rates for different volatile compounds and thestability of adsorbed oxygen species are a function of surface temperature [68].This temperature-selectivity dependence can be utilized to improve the performanceof MOS sensors. Rather than maintaining a constant operating point, as described inSection 5.2.4, the temperature of the sensor may be cycled during exposure to an odor


to obtain a multivariate dynamic signature. Figure 5.15a illustrates the sensitivity pro-files of several doped tin-oxide gas sensors at different temperatures when exposed tovarious analytes. If maximum sensitivity to a particular analyte, say C3H8, were needed,a constant temperature of 250 8C for the Pd-doped sensor would then bemost suitable.For machine olfaction applications, however, where the analyte detection range isbroader, it would be advantageous to capture the response of the sensor over the entiretemperature range. Figure 5.15b shows the conductance-temperature dynamic re-sponse to various analytes when a sinusoidal voltage (2–5 V, 0.04 Hz) is applied tothe heater of a commercial SnO2 sensor (Figaro TGS813). It can be observed thatnot only the magnitude of the conductance but also the shape of the dynamic responseis unique to each analyte. An excellent survey of temperature modulation in semicon-ductor gas sensing may be found in [69].

5.7

Conclusions

This chapter has presented the hardware and software components that constitute theinterface between chemical sensor arrays and pattern analysis techniques, the twocritical building blocks in odor-sensing systems. We have surveyed a number of inter-face circuits that can be used to generate electrical signals for the most popular gassensing technologies: chemoresistive, acoustic wave, and field effect sensors. Analogsignal conditioning of the resulting electrical signals has also been outlined, includinga gentle review of operational amplifiers. Various approaches for controlling the

Fig. 5.15 Left: Sensitivity-temperature profile for Pt- and Pd-doped

tin-oxide sensors [70]. Right: conductance-temperature response of

a tin-oxide gas sensor in (a) air, (b) methane, (c) ethane, (d) propane,

(e) n-butane, (f) isobutene, (g) ethylene, (h) propylene, and (i) carbon

monoxide [71]


operating temperature of metal-oxide sensors have also been presented. Finally, pre-processing algorithms to prepare sensor-array data for multivariate pattern analysishave been described. Although often overlooked, careful selection of sensor interfacecircuits, signal conditioning, and preprocessing is critical for achieving optimal per-formance in odor-sensing systems.

5.8

Acknowledgements

This work was partially supported by the award NSF/CAREER 9984426. The authorsare grateful to J. W. Gardner and T. C. Pearce for helpful suggestions during the reviewprocess of this manuscript.

References

1 Figaro, General Information for TGS Sensors,Figaro Engineering, Inc., Osaka, Japan,1996.

2 FIS, Products Review, FIS Inc., Osaka, Japan,1998.

3 J. W. Gardner, P. N. Bartlett. Electronic Noses,Principles and Applications, Oxford Uni-versity Press, Oxford, UK, 1999.

4 R. Pallas-Areny, J. G. Webster. Sensorsand Signal Conditioning, 2nd Edition, Wiley,New York, 2001.

5 J. Fraden. Handbook of Modern Sensors.Physics, Designs and Applications, 2ndEdition, American Institute of Physics,Woodbury, New York, 1997.

6 W. Gopel, K. D. Schierbaum. Sens.Actuators B, 1995, 26–27, 1–12.

7 U. Hoefer, K. Steiner, E. Wagner. Sens.Actuators B, 1995, 26–27, 59–63.

8 U. Weimar, W. Gopel. Sens. Actuators B,1995, 26–27, 13–18.

9 G. Sberveglieri. Sens. Actuators B, 1995, 23,103–109.

10 M. E. H. Amrani, K.C. Persaud, P. A. Payne.Meas. Sci. Technol., 1995, 6(10), 1500–1507.

11 J. W. Grate, G. C. Frye. In Sensors UpdateVol. 2, (Eds.: H. Baltes, W. Gopel andJ. Hesse), VCH,Weinheim, 1996, Chapter 2.

12 J. W. Grate, S. J., Martin, R. M. White. Anal.Chem., 1993, 65(21), 940–948.

13 D. S. Ballantine, R. M. White, S. I. Martin,A. J. Ricco, E. T. Zellers, G.C. Frye,H. Wohltjen. Acoustic Wave Sensors. Theory,

Design and Physico-Chemical Applications,Academic Press, San Diego, CA, 1997.

14 A. Russell. Odour detection by mobile robots,World Scientific, Singapore, 1999.

15 R. A. Russell, L. Kleeman, S. Kennedy.Proceedings of the 2000 Australian Conferenceon Robotics and Automation, Melbourne,Aug. 30-Sept. 1, 2000, 87–92.

16 A. Spetz, F. Winquist, H. Sundgren,I. Lundstrom. 1992, in Gas Sensors(Ed.: G. Sverveglieri), Kluwer AcademicPublishers, 1992, 219–279.

17 J. V. Hatfield, J. A. Covington, J.W. Gardner.Sens. Actuators B, 2000, 65(1–3), 253–256.

18 I. Lundstrom, E. Hedborg, A. Spetz, H.Sundgren, F. Winquist. In Sensors andSensory Systems for an Electronic Nose,(Eds.: J. W. Gardner, P. N. Bartlett), KluwerAcademic Publishers, Dordrecht, 1992,303–319.

19 R. C. C. Li, P. C. H. Chan, P. W. Cheung.Sens. Actuators B, 1995, 28(3), 233–242.

20 K. Ikohura, J. Watson. The Stannic Oxide GasSensor, Principles and Applications, CRCPress, Boca Raton, FL., 1994.

21 M. Benammar, W. C. Maskell. J. Phys. E:Sci. Instrum., 1989, 22, 933–936.

22 P. McGeeghin, P. T. Moseley,D. E. Williams. Sensor Review, 1994, 14(1),13–19.

23 P. Horowitz, W. Hill. The art of electronics,Cambridge University Press, Cambridge,UK, 1989.


24 A. Heilig, N. Barsan, U. Weimar, W. Gopel.Sens. Actuators B, 1999, 58(1–3), 302–309.

25 S. Jonda, M. Fleischer, H. Meixner. Sens.Actuators B, 1996, 34(1–3), 396–400.

26 A. P. Lee, B. J. Reedy. Sens. Actuators B, 2000,69(1–2), pp. 37–45.

27 P. Mielle. Sens. Actuators B, 1996, 34(1–3),533–538.

28 W. M. Sears, K. Colbow, F. Consadori. Sens.Actuators, 1989, 19, 333–349.

29 J. J. Carr. Designer’s Handbook of Instru-mentation and Control Circuits, AcademicPress, San Diego, CA, 1991.

30 H. R. Taylor. Data Acquisition for SensorSystems, Chapman and Hall, London, UK,1997.

31 D. C. Ramsay. Principles of EngineeringInstrumentation, Arnold, London, UK, 1996.

32 P. K. Clifford, D. T. Tuma. Sens. Actuators,1982, 3, 233–254.

33 J. Brignell, N. White. Intelligent sensorsystems, Institute of Physics Publishing,Bristol, UK, 1996.

34 C. L. Phillips, H. T. Nagle. Digital ControlSystem Analysis and Design, Prentice Hall,Englewood Cliffs, New Jersey, 1995.

35 J. W. Gardner, M. Craven, C. Dow,E. L. Hines. Meas. Sci. Technol., 1998, 9,120–127.

36 R. Gutierrez-Osuna, H. T. Nagle. IEEETrans. Sys. Man Cyber. B, 1999, 29(5),626–632.

37 J. W. Gardner. Sens. Actuators B, 1991, 4,109–115.

38 J. W. Gardner, E. L. Hines, H. C. Tang.Sens. Actuators B, 1992, 9, 9–15.

39 J. W. Gardner, P. N. Bartlett. Sens. ActuatorsB, 1994, 18–19, 211–220.

40 G. Horner, C. Hierold. Sens. Actuators B,1990, 2, 173–184.

41 H. Abe, T. Yoshimura, S. Kanaya, Y. Taka-hashi, Y. Miyashita, S.-I. Sasaki. Anal. Chim.Acta, 1987, 194, 1–9.

42 K.C. Persaud, S.M. Khaffaf, J.S. Payne,A.M. Pisanelli, D.-H. Lee, H.-G. Byun. Sens.Actuators B, 1996, 36(1–3), 267–273.

43 E. J. Severin, B. J. Doleman, N. S. Lewis.Anal. Chem., 2000, 72(4), 658–668.

44 A. Hierlemann, U. Weimar, G. Kraus,M. Schweizer-Berberich, W. Gopel. Sens.Actuators B, 1995, 26(1–3), 126–134.

45 H. Sundgren, F. Winquist, I. Lukkari, I.Lundstrom. Meas. Sci. Technol., 1991, 2(5),464–469.

46 N. Paulsson, F. Winquist. Forensic Sci. Int.,1999, 105(2), 95–114.

47 T.C. Pearce, J. W. Gardner. The Analyst,1998, 123, 2057–2066.

48 E. Llobet, J. Brezmes, X. Vilanova,X. Correig, J.E. Sueiras. Sens. Actuators B,1997, 41(1–3), 13–21.

49 S. Roussel, G. Forsberg, V. Steinmetz,P. Grenier, V. Bellon-aurel. J. Food Eng.,1998, 37, 207–22.

50 B.W. Saunders, D.V. Thiel, A. Mackay-Sim.The Analyst, 1995, 120, 1013–1018.

51 F. Sarry, M. Lumbreras. Sens. Actuators B,2000, 67, 258–264.

52 J. White, J. S. Kauer, T. A. Dickinson,D. R. Walt. Anal. Chem., 1996, 68(13),2191–2202.

53 B. G. Kermani, S. S. Schiffman, H. T. Nagle.IEEE Trans. Instrum. Meas., 1998, 47(3),728–741.

54 D. M. Wilson, S. P. DeWeerth. Sens.Actuators B, 1995, 28, 123–128.

55 T. Eklov, P. Martensson, I. Lundstrom. Anal.Chim. Acta, 1997, 353, 291–300.

56 T. D. Gibson, O. Prosser, J. N. Hulbert,R. W. Marshall, P. Corcoran, P. Lowery,E. A. Ruck-Keene, S. Heron. Sens. ActuatorsB, 1997, 44(1–3), 413–422.

57 R. Gutierrez-Osuna, H. T. Nagle,S. S. Schiffman. Sens. Actuators B, 1999,61(1–3), 170–182.

58 T. Nakamoto, A. Iguchi, T. Moriizumi. Sens.Actuators B, 2000, 71, 155–160.

59 E. Llobet, X. Villanova, J. Brezmes,R. Alcubilla, J. Calderer, J. E. Sueiras,J. Correig. Meas. Sci. Technol., 1997, 8,1133–1138.

60 G. E. P. Box, D. R. Cox. J. Roy. Statist. Soc.Ser. B, 1964, 26, 211–243.

61 F. Bordoni, A. D’Amico. Sens. Actuators A,1990, 21–23, 17–24.

62 P. Corcoran. Sens. Actuators B, 1994, 18–19,649–653.

63 H. Ulmer, J. Mitrovics, U. Weimar, W.Gopel. Sens. Actuators B, 2000, 65(1–3),79–81.

64 J. W. Grate, S. L. Rose-Pehrsson,D. L. Venezky, M. Klusty, H. Wohltjen.Anal. Chem., 1993, 65, 1868–1881.

65 B. Hivert, M. Hoummady, D. Hauden,P. Mielle, G. Mauvais, J. M. Henrioud. Sens.Actuators B, 1995, 27(1–3), 242–245.

66 S. Strathmann. Sample Conditioning forMulti-Sensor Systems, Ph.D. Dissertation,

5.8 Acknowledgements 131131

Institute for Physical and TheoreticalChemistry, University of Tubingen,Germany, 2001.

67 U. Weimar, W. Gopel. Sens. Actuators B,1998, 52, 143–161.

68 M. J. Madou, S. R. Morrison. ChemicalSensing with Solid State Devices, AcademicPress, Boston, MA, 1989.

69 A. P. Lee, B. J. Reedy. Sens. Actuators B, 1999,60, 35–42.

70 N. Yamazoe, N. Miura. In Chemical SensorTechnology Vol. 4 (Ed.: S. Yamauchi),Chemical Sensor Technology, 1992, 19–42.

71 S. Nakata, S. Akakabe, M. Nakasuji,K. Yoshikawa. Anal. Chem., 1996, 68,2067–2072.


6

Pattern Analysis for Electronic Noses

Evor L. Hines, Pascal Boilot, Julian W. Gardner and Mario A. Gongora

Abstract

This chapter provides a detailed description of a comprehensive set of pattern recogni-tion (PARC) techniques that have been employed to analyze electronic nose (EN) data;i.e. well-known and commonly used techniques, research algorithms and futuretrends in pattern analysis. The problem of pattern analysis of EN data is closely linkedto that of multivariate data analysis. Both statistical and non-parametric multivariateanalysis techniques are discussed here. The chapter focuses on basic chemometrictechniques and so those based on the principles of engineering, mathematics andstatistics. We first describe methods that are common conventional statistical meth-ods, such as principal components analysis (PCA), partial least squares (PLS), multiplelinear regression (MLR), principal component regression (PCR), discriminant func-tion analysis (DFA) including linear discriminant analysis (LDA), cluster analysis (CA)including nearest neighbor (NN). We then briefly explore the development of biolo-gically motivated non-parametric methodologies, such as artificial neural networks(ANNS) including multi-layer perceptron (MLP), fuzzy inference systems (FIS),self-organizing map (SOM), radial basis function (RBF), genetic algorithms (GAS),neuro-fuzzy systems (NFS) and adaptive resonance theory (ART). There has alwaysbeen an appeal when working on EN architectures that mimic the human olfactorysystem, namely to build physiologically inspired PARC systems that imitate the hu-man brain. The classification scheme presented here is made on three levels: first adistinction is made between statistical and biological approaches, then between quan-titative and qualitative pattern analysis algorithms, and finally supervised and unsu-pervised techniques. Together these provide the reader with a comprehensive reviewof pattern analysis techniques for ENS.


133133

6.1

Introduction

The electronic nose (EN) is an instrument that has been developed to mimic the hu-man organ for smell; i.e. the biological olfactory system presented in Chapter 1. TheEN is both a chemical sensing and a data analysis system that can, to some extent,discriminate between different simple or complex odors. The nature of odors andaromatic volatile compounds has been previously discussed in Chapter 1. Gener-ally, the EN is developed as a match-model for the natural nose comprising the variousstages between a volatile odorant and its recognition, namely: interaction, signal gen-eration, processing, and identification, as outlined by the parallel between biologicaland artificial noses in Fig. 6.1. The system comprises a chemical sensor array, togetherwith an interfacing electronic circuitry and a pattern-recognition unit that acts as asignal processing system [1]. However, a simpler model based on an array of sensorsand a pattern recognition system was later introduced, which helps to better under-stand and represent how the nose functions [2]. A discussion of chemosensors andsignal pre-processing is given in Chapters 4 and 5, respectively. Both models men-tioned above incorporate a pattern recognition system, yet much effort in EN devel-opment work has focused on the sensor and instrumentation design while data ex-ploration has perhaps been relatively neglected for long periods. In this chapter,we review the pattern analysis techniques, classification systems, identification meth-ods and recognition algorithms that have been applied to solve olfactory problems.Data analysis, machine learning or chemometrics are being widely used today in

physical, chemical, and engineering sciences, so that currently there are a large num-ber of pattern recognition (PARC) techniques available. In order to select appropriatePARC algorithms for EN applications, it is important to understand the fundamental

Fig. 6.1 Basic diagram showing the analogy between biological

and artificial noses

6 Pattern Analysis for Electronic Noses134

nature of the data being analyzed. The problem of analyzing EN data sets is one ofdetermining the underlying relationships between one set of independent variables(e.g. outputs from an array of n sensors) and another set of dependent variables(e.g. odor classes and component concentrations) using for example multivariate ana-lysis [3]. The general multivariate problem in odor sensing is commonly referred to asPARC and is used to analyze qualitatively the odor patterns produced by these instru-ments; but it can also possibly be used quantitatively, for example to compute indi-vidual component concentrations. It is envisaged that efficient data processing andpattern analysis will provide more accurate models and better understanding ofthe data generated. Pattern recognition algorithms and data processing techniquesare a critical component in the implementation, development and successful commer-cialization of ENs.

6.1.1

Nature of Sensor Array Data

Now, let us consider an array of n discrete sensors, as illustrated in Fig. 6.2, where eachsensor i produces a time-dependent output signal XijðtÞ in response to an odor j. Theelectrical sensor signal depends on several physical parameters (e.g. flow rate of odoracross sensor, ambient pressure, temperature and humidity), but the sensor outputsare expected to reach constant asymptotic values when presented with a constant inputstimulus. It has been common practice to use only the static or steady-state values ofthe sensor signals rather than the dynamic or transient responses, the response is thensimply a time-independent parameter, XijðtÞ ! Xij. However, the choice of the re-sponse parameter is fundamental to the subsequent performance of the PARC, sothe pre-processing technique, which is applied to the response vectors, is usually de-signed to help analyze data in the context of a specific problem. Generally, in order toextract relevant key features from the data in terms of the static change in sensorparameter (e.g. resistance or conductivity), a good choice is to use a fractional differ-ence model: Xij ¼ ðXodor

ij � X0i Þ=X0

i where Xodorij is the response of the sensor i to the

sample odor j, and X0i is the baseline or reference signal, such as the value in ambient

room air. The response generated by the n-sensor array to an odor j can then be re-presented by a time-independent vector: Xj ¼ ðX1j; X2j; :::; Xij; :::; XnjÞT. When thesame array is presented to a set ofm odors, the responses can be regarded as a set ofmvectors, which are best represented by a response matrix R:

R ¼

X11 X12 ::: X1mX21 X22 ::: X2m

..

. ...

Xij...

Xn1 Xn2 ::: Xnm

0BBB@

1CCCA ð6:1Þ

Each column represents a response vector associated with a particular odor, whereasthe rows are the responses of an individual sensor to the different measurands. Asodor sensors are not entirely specific, an individual sensor will respond to a variety


of odors but with varying sensitivity (e.g. speed and intensity of the response). As aresult, the off-diagonal terms of R are usually non-zero, and thus, under these condi-tions, PARC techniques are required to process the data and solve the class predictionproblem.

6.1.2

Classification of Analysis Techniques

The responses generated by an array of odor sensors may be processed using a varietyof techniques. In Fig. 6.2, where the basic data-processing structure of an EN is pre-sented, the array formed from the sensor outputs is pre-processed and normalized sothat the modified response matrix can be fed into a PARC engine (see Chapter 5). Thenature of a PARC engine is usually classified in terms of being parametric or non-parametric, and supervised or unsupervised.

* Parametric. A parametric technique, commonly referred to as a statistical approach,is based on the assumption that the spread of the sensor data can be described by aprobability density function (PDF). In most cases, the assumption made is that thedata follow a normal distribution with a constant mean and variance. These tech-niques try to find an underlying mathematically formulated relationship betweensystem inputs, odor vectors and its outputs, classes or descriptors.

* Non-parametric. Non-parametric methods do not assume any specific PDF for thesensor data and thus apply more generally. This approach to multivariate data ana-lysis has led to the fields of artificial neural networks (ANNS) and expert systems.

Fig. 6.2 Basic architecture of a data processing system for an EN


* Supervised. In a supervised learning PARC method, a set of known odors are sys-tematically introduced to the EN, which then classifies them according to knowndescriptors or classes held in a knowledge base. Then, in a second stage for iden-tification, an unknown odor is tested against the knowledge base, now containingthe learnt relationship, and then the class membership is predicted. Unknown odorvectors are analyzed using relationships found a priori from a set of known odorvectors used in an initial calibration, learning, or training stage. The idea of testing amethod using unclassified response vectors is well established and is often referredto as cross-validation.

* Unsupervised. For unsupervised learning, PARC methods learn to separate the dif-ferent classes from the response vectors routinely, discriminating between un-known odor vectors without being presented with the corresponding descrip-tors. These methods are closer to the way that the human olfactory system worksusing intuitive associations with no, or little, prior knowledge.

6.1.3

Overview

This chapter provides a detailed description of a comprehensive list of PARC techni-ques that have been employed to analyze EN data; i.e. well-known and commonly usedtechniques, up-to-date algorithms and future trends in pattern analysis. Both statisticaland non-parametric analysis techniques are discussed. The chapter focuses on basicchemometric techniques and so those based on the principles of engineering, mathe-matics and statistics [4]. Thus we first describemethods that are common conventionalstatistical methods, such as principal components analysis (PCA), partial least square(PLS), multiple linear regression (MLR), principal component regression (PCR), dis-criminant function analysis (DFA) including linear discriminant analysis (LDA), andcluster analysis (CA) including nearest neighbor (NN). Then we briefly explore thedevelopment of biologically motivated methodologies, such as artificial neural net-works (ANNS) including multi-layer perceptron (MLP), fuzzy inference systems(FIS), self-organizing map (SOM), radial basis function (RBF), genetic algorithms(GAS), neuro-fuzzy systems (NFS) and adaptive resonance theory (ART). Therehas always been an appeal, when working on EN that mimic the human olfactorysystem, to build physiologically inspired PARC systems that imitate the human brain.As stated above, the problem of pattern analysis of EN data is closely linked to the

multivariate analysis of data sets. Figure 6.3 summarizes the main multivariate dataprocessing techniques, or PARC algorithms, that have been employed in the field ofENs and which are explored in this chapter. The classification scheme ismade on threelevels: a first distinction is made between statistical and biological approaches, thenbetween quantitative and qualitative pattern analysis algorithms, and finally betweensupervised and unsupervised techniques. Specifically, Section 6.2 describes the com-monly used conventional or classical statistical pattern analysis techniques, whereasSection 6.3 describes some biologically inspired or ‘intelligent’ PARC models, such asANNs. Key factors for a comparison of these algorithms are presented in Section 6.4


together with future trends in EN pattern analysis in terms of the use of dynamicalanalysis and intelligent sensor systems.

6.2

Statistical Pattern Analysis Techniques

Classical statistical methods, using a probability model, were first developed and usedin the field of applied mathematics, now called chemometrics. In this section somemathematical methods that may be applied to the multi-component analysis of odors,are presented. Categorization of classifiers, as presented in the previous paragraph,can be made based on certain features, such as supervised or unsupervised, model-based or model-free, qualitative or quantitative. For example, discriminant functionanalysis (DFA) is a parametric and supervised learning classifier, which can beused for both qualitative and quantitative analysis. Principal components analysis(PCA) is a non-parametric projection method and is often used to implement a linearsupervised classifier, in conjunction with discriminant analysis.

Fig. 6.3 Classification scheme of the multivariate pattern analysis

techniques applied to EN data


6.2.1

Linear Calibration Methods

Linear multivariate calibrationmethods, using linear algebra, are often used to processsensor array data and obtain the concentrations within a multi-component mixture.This is usually based on two basic assumptions: 1) that the response of each sensor isproportional to the component concentration (linear sensor model), and 2) that theresponse of a sensor mixture equals the sum of the responses to the individual com-ponents (superposition model). The multiple linear regression (MLR) method is com-monly used to analyze mixtures of gases and vapors. MLR uses sensor responses vari-ables Xij to predict component concentrations csj from a regressive equation holdingthe partial regression coefficients bij [5].

csj ¼ b1jX1j þ b2jX2j þ :::þ bijXij þ :::þ bnjXnj ð6:2Þ

The goal of MLR is to calculate the values of the regression coefficients bij for thesensors, minimizing the sum of squared deviations (gradient descent) between thepredicted component concentration values csj and the actual measured concentrationvalues. MLR has been successfully applied to analyze the response of nine odor sen-sors to certain organic vapors [6]. It is a technique widely used in chemometrics thatworks best with orthogonal variables for which sensors are component specific, how-ever it is sensitive to noise and suffers from the considerable degree of co-linearitypresent in solid-state odor sensors, for example tin-oxide resistors.When it is desirable to determine the individual gas concentrations from a multi-

variate calibration, two other methods used in preference to MLR, are PLS and PCR,which assume that a linear-inverse model can be applied to the data. In the model, theconcentration vector c is related to the response matrix R by c ¼ Rm þ e where m is aregression vector containing all the model parameters, and e is an error vector contain-ing the concentration residuals from other gases. The regression vectors are estimatedin PLS and PCR by finding the pseudo-inverse response matrix in terms of orthonor-mal and diagonal matrices [7]. PLS was first described in the mid-1960s and has sincebeen refined and specialized for chemical applications [8]. PLS is often applied to gasmixture analysis because it accepts collinear data, separates out noise from usefulsample information, and makes meaningful linear combinations for different concen-trations. It is also one of the latest regression procedures, based on the properties ofMLR, to be developed for concentration prediction. The main difference between PLSand PCR is that PLS includes information about the concentration vector in the modelbuilding while PCR does not. This is important when analyzing data to classify odorsrather than to predict chemical concentrations.Since most chemical sensors have a non-linear concentration dependence, these

techniques are only approximately valid within a small, or a low (e.g. Langmuir mod-el) concentration range. In order to handle non-linear data and improve the perfor-mance of linear PARC, the sensor response against concentration can be linearizedusing either an appropriate pre-processing technique, or by using a non-linear MLRmodel [9]. A non-linear PLS for correcting non-linearities after calculations has been


applied to evaluate signals for gas sensor array and used for quantitative multi-com-ponent analysis [10]. This type of technique shows good results when applied to binaryor tertiary gas mixtures (n = 2 or 3) using an array of sensors (n > 3), however thecalibration method becomes impossible when working with complex odor samplesthat may contain tens or even hundreds of different gases or components. Conse-quently, most research has focused on the use of qualitative types of classificationmethods for EN data, such as discriminant analysis and cluster analysis.

6.2.2

Linear Discriminant Analysis (LDA)

In DFA, a parametric pattern analysis method, it is first assumed that the data aremultinormal-distributed and then the discriminant functions Zp are determined.The set of discriminant functions Zp is calculated from the variables by separatingthe odor classes, finding the linear combination of the independent sensor responsesXij in following equation:

Zp ¼ a1pX1j þ a2pX2j þ :::þ aipXij þ :::þ anpXnj ð6:3Þ

The coefficients aip are determined so that the F-ratio on the analysis of the variance ismaximized subject to Zp being uncorrelated with Zp:::Zp�1 within groups. Once theregression coefficients aip have been computed on the known data, following super-vised learning, then they can be used to form the classification functions, which pre-dict the group membership of unknown response vectors (referred to as cross-valida-

Fig. 6.4 Results of linear DFA on

the analysis of three commercial

roasted coffees using a 12-element

tin oxide EN. (Reprinted from ref.

[12], Elsevier Science, with per-

mission.)


tion). There are many ways of performing DFA, but the classical approach is LDA, forwhich a straight-line hyperplane passing through the data is found using differentcriteria [11]. However, sometimes, overlap occurs between classes and so there isno exact or crisp cut-off value. LDA has been applied to the discrimination of com-mercial coffee flavors, as shown in Fig. 6.4, and alcohol vapors with almost 100%success rate [12]. Figure 6.4 shows the results of applying DFA to the response (frac-tional change of conductance) of 12 tin oxide gas sensors sampling the headspace ofthree different coffees. Plots of the first two discriminant functions show reasonableseparation of the three classes. The observed classification rate was 81% when half ofthe data was used for cross-validation.Other more advanced models have been developed, including quadratic or logistic

discrimination that require some assumptions about the original data, but providebetter discrimination performance. In Shaffer et al. [13], two examples of LDA arepresented. The first one, the Mahalanobis linear discriminant analysis (MLDA) clas-sifier is based on the Mahalanobis distance metric, it is trained by computing a meanvector for each class and the pooled covariance matrix in order to define the classboundaries. To classify a new pattern (Xj), the Mahalanobis distance to the mean vec-tor (�XXm) for each class is computed as: djm ¼

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiðXj � �XXmÞ

0S�1ðXj � �XXmÞq

, where d isthe distance between pattern vector j and themean pattern vector for classm and S�1 isthe inverse of the pooled covariance matrix (estimate of the common covariance of theclasses). The classification of the new pattern is assigned to the classification of theclosest mean vector, i.e. smallest d. However, the use of the pooled covariance matriximplies that the covariance matrices for each class are not significantly different. Thesecond method, the Bayes linear discriminant analysis (BLDA) is based on the Bayesstrategy for minimizing the risk associated with the classification decision. The train-ing is performed using the mean vector for each class, and the pooled covariancematrix to position a linear separating surface. The assignment of class membershipfor a new pattern is determined by the side of the discriminant in which it lies using ascalar dot product of the pattern with each linear separating surface.

6.2.3

Principal Components Analysis (PCA)

PCA is a linear unsupervised method that has been widely used by various researchersto display the response of an EN to simple and complex odors (e.g. alcohols, beers,coffees). It is a multivariate statistical method, based on Karhunen-Loeve expansion,used by classification models to produce qualitative results for ENPARC. The methodconsists of expressing the response vectors Xj in terms of linear combinations oforthogonal vectors along a new set of coordinate axes, and is sometimes referredto as vector decomposition and thus helps to display multivariate data in two or threedimensions. Along the new axes the sample variances are extremes and uncorrelatedso that an analysis in terms of principal components can show linear interdependencein data. Each orthogonal vector, principal component, accounts for a certain amount ofvariance in the data with a decreasing degree of importance. The scalar product of the


orthogonal vectors with the response vectors gives the value of the pth principal com-ponent:

Zp ¼ a1pX1j þ a2pX2j þ :::aipXij þ :::þ anpXnj ð6:4Þ

The variance of each principal component score, Zp, is maximized under the con-straint that the sum of the coefficients of the orthogonal vectors or eigenvectorsap ¼ ðaip; :::; ajp; :::; anpÞ is set to unity, and the vectors are uncorrelated. The corre-sponding eigenvalues give an indication of the amount of information the respectiveprincipal components represent. The eigenvector associated with the largest eigen-value has the same direction as the first principal component. The eigenvector asso-ciated with the second largest eigenvalue determines the direction of the second prin-cipal component. Since there is often a high degree of sensor co-linearity in EN data,the majority of the information held in response space can often be displayed using asmall number of principal components. PCA is in essence a data dimensionality re-duction technique for correlated data, such that a two- or three-dimensional plot candescribe an n-dimensional problem. It can be applied to high dimensional data-sets toexplore the nature of the classification problem in gas sensor applications and deter-mine the linear separability of the response vectors. However, if the sensor outputparameters are not linear, the results of PCA are not straightforward and the interpola-tion of features may be suspect, sometimes referred to as ‘pure created artifacts’. PCAis a linear technique that treats all sensors equally, thus the sensors may unduly in-fluence its performance [3].Figure 6.5 shows the results of applying PCA to an array of four tin-oxide sensors

when applied to aromatic headspace of bananas [14]. Since metal oxide sensors gen-

Fig. 6.5 Results of PCA analysis of the response of a four-element

tin-oxide sensor based EN to bananas aromas, showing clusters

of increasing ripeness, a–g. (Reprinted from ref. [14], IOP Publishing

Ltd, with permission.)


erally respond in a similar manner (i.e. correlated), over 99% of the variance is typi-cally described by the first two principal components [15]. Seven clusters or categoriesare apparent and are associated to seven states of ripeness; from left to right the groupsappear according to increasing ripeness. However, the occurrence of complex bound-aries suggests that a non-linear classification method is needed in order to obtain goodperformance in terms of PARC, rather than linear methods. Figure 6.6 illustrates howwell this technique works when analyzing the response of an array of 32 carbon-poly-mer composite sensors. The system is being used to identify six species of bacteria,commonly associated with eye infections [16]. Most of the variance in the data is ex-plained by considering only the first principal component, which implies that thesensor responses are again highly correlated. It can be seen in Fig. 6.6 that six groupsexist and are associated with the different bacteria species. Although PCA is useful as atool with which to assess the performance of an EN, CA presented in the followingsection, is perhaps a more appropriate tool because it is an unsupervised technique forenhancing the differences between the response vectors.

6.2.4

Cluster Analysis (CA)

Clustering is the separation of a data set into a number of groups, called clusters, basedon measures of similarity. CA is an unsupervised, non-parametric technique that iswidely used to discriminate between response vectors in n-dimensional space by en-hancing their differences. It is also used to identify clusters or groups to which un-known vectors are likely to belong. The goal is to find a set of clusters for which sam-ples within a cluster aremore similar than samples from different clusters. Commonly

Fig. 6.6 Results of PCA analysis of the response of a 32 carbon-

polymer composite sensor based EN to bacteria causing eye infections.

(Reprinted from ref. [16], IOP Publishing Ltd, with permission.)


clusters are allowed tomerge and split dynamically by the clustering algorithm. CA is amodel-free qualitative analysis that generally undergoes an unsupervised learningphase. Four basic types of clustering methods have been identified [17]: hierarchi-cal, optimization-partitioning, clumping, and density-seeking. Hierarchical and parti-tioning methods are the most popular. Hierarchical techniques calculate the multi-variate distances d of each individual to all others, and then cluster them using a pro-cess of either agglomeration (bottom up) or division (top down). The agglomerationtechniques, among which are nearest neighbor (NN), furthest neighbor, fusion andWard’smethod, assumes that all individuals start off being alone, i.e. in a group of one,the nearest groups are then merged and this process continues until all patterns formsuitable groups. The partitioning technique works on the opposite principle, it as-sumes that all the individuals start in one group and then splits them into two,and so on until all are in a group of their own. Hierarchical techniques produce astructured tree or dendrogram depending upon the definition of the distance-metricd and the way closeness and proximity of individuals are defined. The grooping isbased on the proximity of the vectors in feature space. To do so a multi-distance metricdij is calculated between data points i and j according to the expression:

dij ¼XN

k¼1

ðXik � XjkÞN

!1=N

ð6:5Þ

N is normally set to 2 and so the Euclidean (linear) metric is used, there seems to belittle advantage to be gained from using a non-linear metric when analyzing most ENdata. To classify a new pattern, the Euclidean distance between the new pattern andeach pattern in the training set is computed. The proximity of all points relative to eachother is then found by computing a so-called similarity value Sij, such that:

Sij ¼ 1� ðdij=maxfdijgÞ ð6:6Þ

This is called complete linkage because the distance metric is divided through by themaximum separation between all data points. Thus the similarity value is zero for thefurthest neighbors and close to unity for the nearest neighbors. Other definitions canbe considered for the similarity value but the choice of metric and linkage has a mar-ginal effect on the results. Many techniques exist, such as the one that links togethergroups in which the average distance, median distance, or distance between centroidsis small enough, (Ward’s method and k-nearest neighbors). The proximity can be re-presented by plotting either the multivariate distance d or the similarity index S. Itshould be mentioned that the Euclidean distance can sometimes produce unexpectedresults unless the pattern vector is normalized (or scaled), so that CA is very sensitiveto data pre-processing methods. Figure 6.7 shows the results of a CA (Euclidean me-tric, complete linkage) of the response of a metal-oxide EN to different alcohols [15].The dendrogram connects up response vectors with the nearest similarity value andthus illustrates how the odors are interrelated.CA is a method easy to use and rapidly provides the user with pertinent information,

and is widely used in the field of EN pattern analysis. PCA is used to identify groups or


clusters of points in feature space. However, the nature of EN data is such that it isoften desirable to use a more powerful pattern analysis method. Typically, a method isrequired that not only copes with non-linear, non-parametric data but also generates ametric, which can adapt locally to regions of closely-packed response vectors and sogive improved predictive performance. This has led to the rapid and widespread ap-plication of ANN to the analysis of patterns generated by EN. More ‘intelligent’ tech-niques will be considered in the following section.

6.3

‘Intelligent’ Pattern Analysis Techniques

The nature of EN data is such that it is often desirable to use a more powerful PARCmethod that is able to cope with non-linear data, and has further advantages, overmoreconventional methods, such as learning capabilities, self-organizing, generalizationand noise tolerance [18]. When the objective is to develop an EN that mimics the hu-man olfactory system there is always an intellectual appeal to work on physiologically

Fig. 6.7 Dendrogram showing

results of CA on responses of

12-element tin oxide EN to five

alcohol samples, resulting in

clusters A, B, C, D, E.

(Reprinted from ref. [15],

Elsevier Science, with per-

mission.)


inspired PARC systems that imitate the human brain by learning from patterns. Re-cent interest in learning from data has resulted in the development of biologicallymotivated methodologies such as ANN, FIS, SOM, GA, NFS, and ART. ANNS, some-times called neurocomputers, consist of parallel interconnected and usually adaptiveprocessing elements that are attractive as they, to a certain extent, mimic the neuro-biological system [1]. The processing elements represent the biological brain cells orneurones, and their interconnections, the synaptic links. The pattern recognition abil-ity of ANNS is potentially higher than the classical PARC paradigms described pre-viously, due to parallel signal processing and great tolerance to sensor drift and noise.For a historical review of ANNS, see for example Haykin [19], the following sectiongives a comprehensive review of ANN-based EN systems and applications.

6.3.1

Multilayer Feedforward Networks

ANN generally give results quickly, are efficient with information processing, andlearn by presenting examples; however it is sometimes difficult to choose the optimalnetwork parameters and training procedures. Recently, ANNS have been widely usedin odor recognition and many different ANN paradigms have been applied in thiscontext. Since three-layered networks have sufficient computational degrees of free-dom to solve any classification problem [20], most EN workers have adopted this to-pology of network for implementing MLPs. Other feedforward networks can be usedand the main ones are presented in this section, these include RBF and probabilisticneural networks (PNN).MLP, as a three-layered feedforward back-propagation (BP) trained network, is the

most popular arrangement of neurones in odor classification and was the first one tobe applied to EN [1]. In a network, the processing elements are organized in a regulararchitecture of three distinct groups of neurones: input, hidden, and output layers.Only the units in the hidden and output layers are neurones and so a MLP hastwo layers of weights. The number of input nodes is typically determined to corre-spond to the number of sensors in the array. The number of neurones in the hiddenlayer is determined experimentally and the number of odors analyzed generally de-termines the number of output neurones. When using a one-in-N coding scheme,there is one output neurone for each potential odor class. There are more efficientcoding schemes but this is the simplest. A MLP has a supervised learning phase,which employs a set of training vectors, followed by the prediction, test or recall phaseof unknown input vectors. Figure 6.8 shows the topology of a network used to identifyfive alcoholic odors using a twelve-element tin-oxide sensor EN [1]. MLP with BP learn-ing algorithm has been applied to the prediction of bacteria type and culture growthphase using an array of six different metal-oxide semiconductor gas sensors [21]. Re-sults show that the best MLP was found to classify successfully 96% of unknownsamples on the basis of 360 training vectors and 360 test vectors.Using BP to train the network, it is necessary to provide it with a number of sample

inputs (training set) with their corresponding target outputs. Each neurone computes


its weighted inputs and performs a non-linear transformation of this sum using agiven activation function, for example a sigmoid transfer function which constrainsthe output to a value between [0,þ1] or [�1,þ1]. Considering a neurone h, with ninputs, [1,…,i,…,n] and an input vector j, the summation function ajh multipliesand sums the input signals Xij with associated adaptable weights whi considering afixed weight called a bias, hh0, which is then transformed by a non-linear activationfunction f(.) (e.g. sigmoid) to produce the single output zjh; the overall computationfollows:

zjh ¼ f ðajhÞ ¼ 1=ð1þ expð�ajhÞÞ ¼ fXn

j¼1

ðwhiXij � hh0Þ !

ð6:7Þ

The calculation is carried out for each neurone and each layer feeding the valuesthrough to the output layer, forward pass. During this learning phase, the weightsare adjusted to minimize the difference between the actual output zjh and the idealor target output tjh for the considered input vector j using the expression djh ¼ zjh � tjh.The error term is often called delta, and the widely used parameter-updating scheme isknown as the delta learning rule [22], the component difference vector is calculatedusing the expression djh ¼ ðtjh � zjhÞð1� tjhÞ. In the backward pass computations; thestochastic approximation procedure updates synaptic weight values during each pre-sentation of the jth training sample on each iteration (or epoch) s using, for example,the gradient descent method with momentum:

wðsÞkh ¼ wðs�1Þ

kh þ DwðsÞkh ¼ wðs�1Þ

kh � gdjhzðsÞjh þ lDwðs�2Þ

kh ð6:8Þ

Basically, the new set of weights, wðsÞkh , is made of a combination of the old weight

values, wðs�1Þkh (from the previous epoch) and a weight update or delta, DwðsÞ

kh . The chan-ge in weights is based on two parameters for this example:

Fig. 6.8 Structure of a fully

connected three-layer (layer i

are o, 1 and 2) backpropagation

network used to process data

from a 12-element in oxide EN

for five alcoholic odors. (Reprinted

from ref. [1], Institute of Physics

Publishing, with permission.)


1) g, the learning rate, a small positive number (default is generally 0.9) that deter-mines the rate of convergence to a solution of minimum error.

2) l, the momentum term, a small positive number (default is generally 0.5) is oftenadded to improve the speed and stability of the learning.

Thus the change in adjustable synaptic weights is proportional to the error and to theactivation of the input unit. Using BP, the weights and biases associated with theneurones are modified to minimize the mapping error, when stabilized, the networkis said to be trained. The total sum squared error can be used to measure the networkperformance. The updating procedure is repeated for a number of epochs until thenetwork error has fallen to a small constant level. Once the network is trained, it can beused to predict the membership of novel, unseen and untrained samples in a valida-tion set. The classification of new patterns is accomplished by propagating the newpattern through the network and the output neurone with the highest score indicatesthe class.The success of the training process, in terms of a fast rate of convergence and good

generalization, can be affected by the choice of architecture and initial parameters (e.g.learning rate and initial weights). Various learning paradigms are available to train aBP MLP network; Boilot et al. [16] used both gradient descent with momentum andLevenberg-Marquardt variations as supervised learning algorithms. For all architec-tures of the MLP networks tested, the latter paradigm outperformed gradient des-cent. Since architecture and parameters are to be determined experimentally,much time may be spent searching for the optimal ANN. An often employed rule-of-thumb is to set the number of inputs equal to the number of sensors or the numberof extracted features considered for the sensor array, the number of output nodes notgreater than the number of species or compounds to be discriminated (in a one-in-Ncoding), and a hidden layer not larger than the largest of the two other layers. It is alsorecommend having twice as many training vectors as there are weights in the networkdeveloped in order to achieve good generalization.An alternative method to optimize the ANN design is to use a GA to determine

automatically a suitable network architecture (e.g. growing or pruning the network)and a set of parameters (e.g. learning rate, momentum term) from a restricted regionof design space [23]. GA are heuristic search algorithms based on the mechanics ofnatural selection. The structure and parameters of the neural network, learning rate,initial weights, number of layers, number of neurones per layer, and connectivity, arecoded using binary strings, which are concatenated to form chromosomes. GA are thenapplied to search populations of chromosomes using defined typical genetic operatorssuch as parent selection, crossover and mutation. The performance of the networkrepresented by each chromosome ci is evaluated using a fitness function;FðciÞ ¼ a uðciÞ þ b where F is the fitness function, u is the objective function to opti-mize, and a and b are transformation parameters that are dynamically adjusted to avoidpremature convergence. The objective function is generally a weighted sum of thevarious performance measures. In the sensor data classification problem, the perfor-mance measures used in the objective function are based on, for example, the network


prediction error, speed of convergence, size of the network and degree of generaliza-tion [24].The probabilistic neural network (PNN) operates by defining a probability density

function (PDF) for each data class based on the training data set and an optimizedkernel width [13]. A multivariate estimate of the PDF for each class can be expressedas the sum of individual training pattern Gaussian-shaped kernels. The PDF definesthe boundaries for each data class, while the kernel width determines the amount ofinterpolation between adjacent kernels. The classification of new patterns is per-formed by propagating the pattern vector through the PNN; the input layer is usedto store the new pattern while it is serially passed through the hidden layer. Thedot product distance between the new pattern and the training set pattern stored iscomputed at each neurone in the hidden layer. The summation layer consists ofone neurone for each data class and sums the outputs from all hidden neuronesof each respective data class. The products of the summation layer are forwardedto the output layer where the estimated probability of the new pattern, being a mem-ber of that data class, is computed.RBFs are attractive when other ANNmethods fail to get a good classification due to a

significant difference between classes in terms of shape, volume or density, of over-lapping classes. RBF networks are supervised learning paradigms very similar to MLPexcept that they use radial basis transfer functions for the hidden layer rather thanlinear or sigmoid ones. Hence they classify data using hyper-spheres rather thanhyper-planes [25]. The purpose of RBF is to allow the screening of the input spacewith overlapping receptive fields. The non-adaptive RBF is a fast two-stage trainingprocedure using a hybrid-learning rule:

1) Unsupervised learning in the input layer for the determination of the receptive fieldcenters and widths.

2) Supervised learning of weights in the output layer simply using the delta learningrule via linear least squares.

Hence RBF implementations differ mainly in the choice of heuristics used for select-ing basis function centers and widths. For example, taking every sample as a center(may result in over-fitting), selecting centers as representative prototypes using thegeneralized Lloyd algorithm (GLA) and Kohonen’s SOMs, or adding new basis func-tions centered on one of the training samples sequentially. Although RBF networks donot provide error estimates, they have an intrinsic ability to indicate when they areextrapolating since the activation function of the receptive fields is directly relatedto the proximity of the test pattern to the training data. RBF are becoming moreand more popular for EN pattern analysis. However, one of the main difficultieswhen using this type of system is the determination of the optimal architecture –the number of hidden nodes necessary to achieve a good classification. Boilot et al.[16] report on the use of RBF for the prediction of bacteria causing eye infections.Although RBF networks classify bounded regions of sensor space, this can makethem more sensitive to sensor drift and so less robust; this is a trade-off betweenmodel accuracy and robustness.


6.3.2

Competitive and Feature Mapping Networks

There are many different types of neural network but the ones considered in thischapter are those that have been applied to EN data. One of them is a single-layerneural network with competition, such as Kohonen’s self-organizing map. Competi-tive layers are used in ANN to improve the discrimination process and, unlike tradi-tional network layers, there are connections between the neurones within a layer. Thebasic principle is that competition enhances the difference between the level of activa-tion of the neurones, sometimes in extreme cases the ‘winner-takes-all’ and one neu-rone only is allowed to be switched on. A Hamming network is a fixed-weight com-petitive ANNwhere the lower network feeds into a competitive network called maxnet.It uses a maximum likelihood classifier, based on the measure of similarity as in sta-tistical clustering technique, defined as n minus the Hamming distance between theinput unknown odor vector and the exemplar reference vectors. An N-tuple compe-titive network was used to classify the responses of a 12-element MOS odor sensorarray to both a set of alcohols and to a set of different blends of roasted and groundcoffee beans [26]. In this case, the neural network outperformed statistical linear dis-criminant function analysis with a success rate of 87%.Another competitive network that has been applied to EN data is the self-organizing

neural network or Kohonen network [27]. The SOM algorithm was developed by Ko-honen to transform an incoming signal pattern of arbitrary dimension into a one- ortwo- dimensional discrete map. SOM is more closely related to the neural structures ofthe human olfactory cortex than other neural networks presented before because itemulates parts of the brain. SOM applied to EN systems typically contain a two-dimen-sional single layer of neurones in addition to an input layer of branched nodes. If thesystem is left for learning in the environment of interest, the learning algorithm of thenetwork processes the sensor outputs step by step, and constructs an internal repre-sentation of the environment [27]. SOM accumulate a lot of statistical information inan unsupervised fashion, using a competition layer in the form of a Kohonen organiz-ing map so that all weight vectors of the winner and adjacent neurones are updated.They are interesting for EN systems because of their inherent features such as dimen-sionality reduction and invariance to drift and transitory noise [28].We assume here that there are m neurones in the Kohonen neural layer, typically

arranged as the knots of a square lattice, and each one has a parameter weight vectorVðlÞ of dimension n, which is the same as the input feature vectors (i.e. the number ofvectors). A vector describes each neurone in this layer so that the vector componentsare the knot coordinates in the lattice. The weight vectors are randomly initialized atthe beginning. One input vector Xi is selected from the dataset and put into the net-work, so that the distances between Xi and each VðlÞ are computed using the compo-nents:

d2il ¼Xn

j¼1

ðXij � V ðlÞj Þ2 l ¼ 1; :::;m j ¼ 1; :::; n ð6:9Þ


The minimum distance dil* is then determined to obtain the neurone l that is thewinner over the others. In a winner-takes-all strategy, only the winning neuroneweights are updated using Vðl*Þ

new ¼ Vðl*Þold þ gðXi � Vðl*Þ

old Þ where g is the step gain orlearning rate, whereas all other neurones keep their old weights. In another strate-gy, all neurones s that are close to the winner are updated usingVðsÞnew ¼ VðsÞ

old þ ghsl*ðXi � VðsÞoldÞ. hsl* ¼ exp �ks� l*k2=2r2

� �is called the excitatory re-

sponse and is only appreciable for the neurone that coincides with l* and its neighbors.r is the length scale of the proximities to l and is generally fixed to a value in the rangeof 2 to 5 lattice units. It is desirable that after a number of iterations the weights nolonger change, and therefore the map is able to stabilize asymptotically in an equili-brium state, with g decreasing to zero.In a supervised learning scheme, the SOM is provided with the desired output func-

tions; it is called learning vector quantization (LVQ) and integrates supervised learningtechniques in a self-organizing feature map [29]. It combines some of the features ofnearest neighborhood and competitive learning to define a smaller set of referencevectors that span the same space as the original training set pattern. Figure 6.9 showsa schematic diagram of a LVQ network, the hidden layer in the network is a Kohonenlayer, which does the learning and classifying. The LVQ scheme has phases that con-sist of LVQ1 and LVQ2 algorithms. LVQ1, is the basic LVQ learning algorithm, whichhelps all processing elements to take an active part in the learning. LVQ2 is a fine-tuning mechanism, which refines class boundaries. Therefore the output fromLVQ2 is the final encoded version of the original input signal applied to LVQ1.The number of training patterns to ensure equal accuracy to other approaches coulddramatically decrease because the given calibration data set is not the unique source of

Fig. 6.9 Schematic diagram of

LVQ with Kohonen a layer


information collected by the system during unsupervised learning. However, an im-portant limitation of this approach is that lengthy computation is required when ap-plied to real problems.SOMhave been applied to a wide variety of applications including, with some degree

of success, classification of odors and patterns generated by an EN [30]. Hines et al. [28]used the supervised Kohonen SOM on the alcohol and coffee data sets and found goodperformance results in terms of both accuracy and generalization. Shin et al. [31] usedLVQ to classify the strain and growth phase of cyanobacteria using a 6-element ENwith excellent results. When trained on two classes of a gas mixture, after a shortperiod of time the weights of the network appear to be strictly correlated with theassigned classes. The network does not have any direct information about theclasses, except for the sensor outputs [32].

6.3.3

‘Fuzzy’ Based Pattern Analysis

Fuzzy set theory (FST) was invented by Zadeh [33] to provide a mathematical tool fordealing with the linguistic variables and imprecise language used by humans (forexample hot, cold, slow, quite slow). A fuzzy set is defined as a set whose boundaryis not sharp. Fuzzy logic has been applied to EN pattern analysis and attempts havebeen reported to use fuzzy functions in order to identify odors. FST is therefore at-tractive in the field of machine olfaction in which odor samples are described by ol-factory descriptors, such as peppery, floral, or sweet, and intensity attributes, such asquite, very, or strong. Gardner and Bartlett [3] describe three principal approacheswhen fuzzifying the EN classification problem:

1) Sensor space can be defined using fuzzy functions.2) The pattern recognition algorithm can be fuzzified.3) Classification space can be defined using fuzzy functions.

Fuzzy clustering essentially deals with the task of splitting a set of patterns into anumber of classes with respect to a suitable similarity measure of the pattern belong-ing to a given cluster. Fuzzy clustering provides partitioning results with additionalinformation supplied by the cluster membership values indicating different degrees ofbelongingness. Fuzzy clustering can be precisely formulated as an optimization pro-blem of class centers and spreads. The fuzzy c-means (FCM) algorithm, for example,provides an iterative approach for this optimization. Most of the FCM approaches toEN pattern analysis need to be given the correct number of clusters but can prove to bevery attractive for finding patterns in data sets or can even be applied to clusters ex-tracted from data with PCA. Yea et al. [34] used fuzzy logic to fuzzify sensor space byassigning the steady-state voltage of three gas sensors to one of the three odor classes,giving an excellent classification rate.Another approach is to use fuzzy logic to fuzzify the neuronal weights and weight

calculations in a multi-layer neural network. Conventional networks are trained using


randomly initiated weights, which may be a problem for the overall training process.This is because the search for the best set of weights to both classify the trainingpatterns and identify new ones usually starts from a poor point that may never reachthe desired optimal point. On the other hand, a suitable starting point, depending onthe nature of data, is desirable to speed up the process and reduce the likelihood ofsettling in local minima. A type of fuzzy neural network (FNN) can be used to makeuse of possibility distributions to determine the initial weights using membership-class restrictions imposed on a variable defining the range of values [35]. Possibilitydistributions, based on fuzzy logic theory, are often triangular and so they are similarin shape to normal distributions with the means having the highest possibility of oc-currence. In FNN, the signal conditioning that occurs during fuzzification and defuz-zification translate many properties of overlapping sensor arrays into parameters thatare better handled by the classifier. In Singh et al. [36], the use of fuzzy neuronal treecomputing is reported when used on coffee and tainted-water data from an EN. Theirversion of a FNN proved to be better than classical ANN. Ping and Jun [37] used acombined neural network (RBF) with a fuzzy clustering (FCM) algorithm andwere able to demonstrate the unusual effectiveness and the good recognition perfor-mance of their method. FNN are becoming more and more popular and represent aconsiderable potential improvement in the analysis of certain EN problems.ART was introduced as a theory of human cognition in information processing [38]

and it is based on the fact that the human brain can learn new events without neces-sarily forgetting those learnt in the past. ART networks are intelligent systems that arecapable of autonomously adapting in real time to changes in the environment, and thatare stable enough to incorporate new information without destroying the memories ofprevious learning. ART networks have been applied to metal-oxide sensor based ENwith results very similar to those obtained with BP trained MLP, but with a shortertraining time on small data sets [39].Carpenter et al. [40] introduced Fuzzy ARTMAP for incremental supervised learning

and non-stationary PARC problems. Fuzzy ARTMAP carries out supervised learning,like BP MLP, but it is self-organizing, self-stabilizing and suitable for incrementallearning. It can deal with uncertainty or fuzzy data, a key element in many measure-ment systems and generally shows superior performance in learning compared withMLP, exhibiting fast learning for rare events. Figure 6.10 shows the schematic archi-tecture of a Fuzzy ARTMAP neural network that consists of two ART modules inter-connected by an associative memory and internal control structures. The orientingsubsystem is responsible for generating a reset signal while the gain control sumsthe input signal. One of the main advantages of Fuzzy ARTMAP is that it is ableto perform real-time learning without forgetting previously learnt patterns and sothere is potentially no off-line training phase like MLP. This is very importantfrom a practical point of view because the data-set used to train the network maybe increased during the development phase by adding new measurements. Some ear-lier work by Llobet et al. [41] showed that Fuzzy ARTMAP is a promising technique forEN data analysis. Llobet et al. used it to analyze the state of ripeness of bananas andobtained results that exceeded those for MLP [14]. Shin et al. [31] used it to classify thestrain and growth phase of bacteria and once again it outperformed MLP.


6.3.4

Neuro-Fuzzy Systems (NFS)

Both FIS and ANN are branches of an emerging research area of artificial intelligencecalled soft computing. This approach can be used, with some restrictions, as non-algorithmic model-free (i.e. heuristic) estimator for data processing purposes. Fuzzysystems can be built to express knowledge in the form of fuzzy linguistic If-Thenrules and perform some fuzzy clustering analysis, while neural networks can beused to learn from data and perform pattern recognition and classification. NFSare one of the most promising approaches that have been developed to deliver thebenefits of both and overcome their limitations, combining or fusing these two com-plementary techniques into an integrated system [42]. Boilot et al. [43] report on severalsoftware-based hybrid neuro-fuzzy systems used for specific real world applicationslinked to data processing. They focus on the extraction of knowledge from a represen-tative data set of alcohol test vectors, collected using a 12-element metal-oxide EN. Thepaper also introduces a classification scheme for grouping the various software, dis-cussing their merits and demerits, drawing upon a comparison of delineated criteriafor evaluating their efficacy (i.e. performances) and interpretability (i.e. semantics).Using these techniques for data exploration, the results from NFS-based EN maybe viewed with more confidence because they provide a better representation ofthe information embedded within data-sets. Users will find it helpful to generateNFS in the context of extracting knowledge from EN data-sets, and representing itas a clear and interpretable set of fuzzy rules. The exploration of EN data and patternanalysis using ‘intelligent’ systems has so far mainly been done using ANN, yet whenthey perform a classification of various odors they give little or no insight into the truenature of the data. Using NFS for data processing and exploration does not only pro-vide an opportunity to discover unknown dependencies and relationships, but alsoallows us to present them as a set of rules that are more interpretable than the weightmatrices returned by ANN.

Fig. 6.10 Architecture of a Fuzzy

ARTMAP neural network


6.4

Outlook and Conclusions

An EN detects simple and complex odors using an array of non-specific chemicalsensors. Essentially, each odor generates a characteristic fingerprint or smell-printof responses from the sensor array and so known odors can be used to build up adatabase and used to train a pattern recognition system. It is impractical to have spe-cific sensors when an odor may contain hundreds or even thousands of compounds,and so the solution is to use a PARC system to classify smell-prints or patterns. PARCsare therefore a critical component in the successful implementation of ENs. The ob-jective of pattern analysis is to train or configure the recognition system in order toproduce unique classifications, or clusterings, of each odorant so that automated iden-tification can be implemented. Many different pattern analysis techniques have beenapplied to EN patterns in recent years. In this section we summarize the various con-siderations relating to EN pattern analysis. First, we discuss the basic criteria for thecomparison of the various PARC paradigms with respect to both quantitative andqualitative pattern analysis. Next, we discuss sensor modeling and ‘intelligent’ sensorsystems. Finally, we draw some conclusions regarding the application of pattern re-cognition to ENs.

6.4.1

Criteria for Comparison

Compared to other applications, chemical sensor array pattern recognition or EN sys-tem pattern analysis has a unique set of requirements and needs [13]. The patterndimensionality for a sensor array (typically < 40) is considerably smaller than formany other applications of PARC in science and engineering (e.g. spectroscopy orchromatography), thus the computational load on the grouping algorithm and theresources needed to learn the classification rules are greatly decreased. Therefore,many of the accepted procedures that are used in traditional pattern recognitionand chemometrics in general may not be pertinent or relevant when applied to ENpattern analysis in particular. EN are expected to be operated in various types of en-vironments and situations, and the pattern analysis paradigm should be able to copewith these conditions. For example, when a system is used in field measurements,additional challenges not seen in the laboratory or a controlled environment are likelyto occur, and the system is still expected to detect and identify the target analytes whilein the presence of large concentrations of unknown interfering species. As suggestedby Shaffer et al. [13], there are a few criteria or qualities that an ideal pattern recognitionalgorithm should have, such as accuracy, speed or ability to cope with uncertainty.

* High accuracy. For application of an EN in the field, the PARC algorithm mustaccurately classify new patterns, with a low false alarm rate (true negative) andideally no missed detection (false positive). For military applications, such as detec-tion of toxic chemical vapors, classification rates should be higher than 90% accu-racy even for low concentration of compounds.


* Fast speed. When used in real-time applications, the PARC algorithm must be ableto classify patterns quickly, so that computationally intense paradigms are not prac-tical.

* Simple to train. The classification rules and the classification itself must be learnedquickly and the training patterns database of the system will need to be updatedperiodically, therefore the paradigm should be able to ‘relearn’. This proceduremust be performed as simply and quickly as possible, keeping the learning out-come simple for the user to be able to understand it.

* Low memory requirements. In field applications, the hand-held EN requires on boardpattern analysis, so the algorithm should be able to be transferred, embedded andrun on a simple micro-controller with limited memory resources. Thus high com-putational power and large memory requirement algorithms are not appropriate forfield units.

* Robustness to outliers. When used in uncontrolled environments, the PARC algo-rithm must be able to differentiate between sensor signals it was trained onand those that it was not, recognizing all the important compounds and ignoringparasitic, noisy or ambiguous signals.

* Produce a measure of uncertainty. For most applications, the PARC paradigm mustbe able to produce a measure of the certainty of the classification results, expressedeither as a percentage, a confidence factor or a category.

Unfortunately, no PARC algorithm is able today to meet all of these requirements, butresearchers, in an attempt to determine the optimal classifier, have performed com-parative studies. The qualitative comparison performed by Derde and Massart [44] onseveral popular chemometric classifiers focused on technical aspects, such as optimaldecision boundaries, overlapping regions, degree of uncertainty and outliers, and prac-tical aspects, such as updates, variables of mixed type, irrelevant parameters and easeof use, of supervised PARC. They conclude with a confirmation of the need for anapplication specific choice of algorithm and the potential that hybrid approachescan bring. The book published by Michie et al. [45] is probably the most comprehen-sive and complete comparison study published as they present 23 types of machinelearning, statistical and neural-classification methods and conclude by presenting therelative merits and demerits, and on the choice of an appropriate algorithm for a givenapplication. The book of Cherkassky and Mulier [4] provides a treatment of the prin-ciples and methods for learning dependencies from data using statistics, neural net-works and fuzzy logic oriented around case studies and examples. It also provides adetailed description of the new learning methodology called support vector machines(SVM). To date, a comparison study published by Schaffer et al. [13] is the only one onEN data. It focuses on qualitative criteria together with one quantitative measurement,namely, the classification accuracy, and proposes the use of a combination of LVQ andPNN in order to exploit the advantages of both methods. The study of NFS for EN dataprocessing presented by Boilot et al. [43] again reinforces the potential of hybrid tech-niques and their practical implementation on micro-controllers.


6.4.2

Intelligent Sensor Systems

The modeling techniques used so far in the context of ‘intelligent sensor’ systems aimto enhance the sensor selectivity, reduce the time necessary for calibration, and coun-teract drift [18]. A careful exploration or analysis of the system is required before ap-plying any dynamic model; unimportant sensors should be discarded using, for ex-ample, PCA loadings at this preliminary data exploration stage.

* Enhancing sensor selectivity. To date, models using parameters estimated from thetransient sensor responses can enhance selectivity. These parameters may be re-lated to the physical and chemical properties of the sensing material and thus arebased on physical models giving some insights into the dynamic behavior of thesensor. However, transient signals can be influenced by previous measurements(short-term memory effect), by drift due to ageing of the system, or variations inambient temperature and humidity, so that models that do not consider these is-sues will deteriorate over time.

* Calibration time reduction. Some applications of time-varying sensor signals offer areduction in the time necessary to calibrate the sensor array to odors of interest.Results with ARMA and ad hoc multi-exponential models applied to the dynamicresponse of tin-oxide sensor arrays have been reported [46]. In this application, theprediction of the static response from the initial part of the dynamic response per-mits a reduction of the calibration time by a factor of four.

* Response models. Dynamic measurements are interesting when changes in eitherthe odors or conditions are of the same time-scale as the sensor response. Thecorrelation approach is a modeling method used to deal with noise, calculatinglinear systems impulse response and non-linear systems Weiner kernels. How-ever, models constructed using black-box models based on input-output dataonly, do not give enough insight into the inner structure of the sensor and themodel cannot be defined in terms of physical and chemical properties. On theother hand, block-structured models are more related to the intrinsic characteris-tics of the sensing mechanisms. The use of non-parametric approaches (e.g. cross-correlation) to estimate the impulse response with low errors requires long datasequences and can be rather time-consuming.

* Drift counteraction. All approaches described include memory effects and thus canassess the problem of short-term drift. Long-term drift caused by sensor poisoning(or system ageing) implies non-stationary measurements with which most of thetechniques, apart from ANN, cannot cope. SOM with residual plasticity can help tomaintain the PARC ability of a sensor system affected by drift. ARTMAP and FuzzyARTMAP contain a self-stabilizing memory that permits accumulating knowledgeof new events in a non-stationary environment; the short term memory gives thenetwork some plasticity to adapt to sensor drift, while the long term memory givesthe necessary rigidity to avoid forgetting previously learnt patterns [47].


6.4.3

Conclusions

For most scientists working in the field of EN systems, the two most commonly usedpattern analysis techniques are first PCA to display known odors, explore how the datacluster in the multi-sensor aroma space and assess their linear separability, and MLPBP trained ANNS to provide a predictive classification of unknown odor vectors. How-ever, PCA can only be used to give a linear representation of the clusters and not forclassification purposes, moreover its outcomes can sometimes be criticized. BPtrained ANNS have arguably been the most successful for many applications todate that focus upon the discrimination of quite dissimilar simple or complexodors, or the staling of a specific complex odor. MLP, even if it is a powerful non-linear classification paradigm and has proven to perform well with EN data, can some-times fail to achieve high levels of correct classification, moreover it is difficult tointerpret its results as the system appears as a black box to the user. Among the algo-rithms presented in this chapter, neural network approaches (MLP, RBF, PNN, LVQ)are the most accurate classifiers and can cope with overlapping clusters observed withlinear techniques. For these applications there is no need to use more complex orsophisticated PARC techniques, and this is why so many commercial EN availabletoday provide a standard BP network as the predictive classifier. Nowadays, research-ers have turned to more reliable and advanced techniques to perform pattern analysisfor field EN and handheld units, such as cluster analysis based on fuzzy clustering ornearest neighbors. Even in the field of neural networks, the performances of the pop-ular MLP are often outperformed by LVQ and RBF networks in terms of sensitivitiesand specificities. These two techniques together with other forms of self-organizingtechniques are being seen as the benchmark for predictive classifiers in EN applica-tions. It is our belief that the best strategy to perform pattern analysis on EN data is toemploy algorithms that can cope, up to a certain extent, with a degree of fuzziness likethe human olfactory system and that presents attractive features. In this context, FuzzyARTMAP networks, for example, are very attractive for pattern classification in thecontext of field instruments because they are able to perform incremental learningand offer self-organizing and self-stabilizing potential. We believe that NFS will beincreasingly used in more challenging EN applications because they include both fuz-zy and neural capabilities and so produce a classification based on an understandableset of rules. However it is always dangerous to try and predict future events!First generation of commercial EN have existed since the early 1990s and now there

are more than 15 manufacturers with applications covering food, cosmetic, environ-ment and medical domains [48]. More sophisticated pre-processing and PARC meth-ods are needed in more challenging applications of EN, such as detecting sub-ppmtaints of components, and in the development of hand-held units. The PARC techni-ques employed in a hand-held EN are likely to mimic more closely the signal proces-sing present in our own olfactory system. The next generation of EN are being devel-oped in university laboratories and research institutions using more biologically in-spired models of the olfactory system. They will need to be more flexible and ableto work in less controlled environments, incorporating all the sensors, signal proces-


sing and neuro-inspired models of olfaction to identify and analyze a wide variety ofodors in a constantly changing background. However, we believe that it will be severalyears before dynamical neural networks are developed with the enormous discrimi-nating power and sensitivity of our truly remarkable olfactory system. The human noseis a complex differential (i.e. adaptive) signal processor that can detect an increase ordecrease in the intensity of a smell, and thus an EN mimicking it may require the useof sophisticated adaptive filter combined with fuzzy classification functions.

Acknowledgements

Pascal Boilot gratefully acknowledges financial support from EPSRC (award number99310943) and the University of Warwick during his stay and his studies as a PhDstudent. We thank our colleagues, students, etc. who have contributed directly or in-directly to this work. Finally we would like to thank Roger Granthier for proof readingthis document.

References

1 J. W. Gardner, E. L. Hines, M. Wilkinson.Meas. Sci. Technol.. 1990, 1, 446–451.

2 P. E. Keller, SPIE proceedings series, 1999,3722, 144–152.

3 J. W. Gardner. P. N. Bartlett. Electronic Noses:Principles and Applications, Oxford Univer-sity Press, New York, 1999, Chapter 7.

4 V. Cherkassky, F. Mulier. Learning fromData: Concepts, Theory, and Methods, Wiley,New York, 1998.

5 J. W. Gardner, E. L. Hines. in Handbookof Biosensors and Electronic Noses (Ed.:E. Kress-Roger), CRC Press, Boca Raton,1997, Chapter 27.

6 W. P. Carey, K. R. Beebe, B. R. Kowalski.Anal. Chem., 1987, 59, 1529–1534.

7 B. F. T. Manly, Multivariate StatisticalMethods, Chapman & Hall, London, 1986.

8 F. Wold, Festschiftjerzyneymanv, Wiley,New York, 1966.

9 M. S. Nayak, R. Dwivedi, S. K. Srivastava.Sens. Actuators B, 1993, 12, 103–110.

10 J. Mitrovics, U. Weimar, W. Gopel.Proceedings of the 8th International Conferenceon Solid-State Sensors and Actuators andEurosensors IX, 1995, 1, 707–710.

11 R. Fisher, Annals of Eugenics, 1936, 7,179–188.

12 J. W. Gardner, H. V. Shurmer, T. T. Tan.Sens. Actuators B, 1992, 6, 71–75.

13 R. E. Shaffer, S. L. Rose-Pehrsson,A. R. McGill. Analytica Chimica Acta, 1999,384, 305–317.

14 E. Llobet, E. L. Hines, J. W. Gardner,S. Franco. Meas. Sci. Technol., 1999, 10,538–548.

15 J. W. Gardner, Sens. Actuators B, 1991, 4,71–75.

16 P. Boilot, E. L. Hines, J. Spencer, J. Mitchell,F. Lopez, J. W. Gardner, E. Llobet, M. Hero,C. Fink, M. Gongora. In Electronic Noses andOlfaction 2000 (Eds.: J. W. Gardner andK. C. Persaud), IoP Publishing, Bristol, 2000,189–196.

17 B. S. Everitt, Cluster Analysis, Heinemann,London, 1981.

18 E. L. Hines, E. Llobet, J. W. Gardner. IEEProc.-Circuits Devices Syst., 1999, 146(6).

19 S. Haykin S, Neural Networks: a Compre-hensive Foundation, MacMillan PublishingCompany, New York, 1994.

20 R. P. Lippmann, IEEE ASSP Mag., 1987,4(2), 4–22.

21 J. W. Gardner, M. Craven, C. Dow,E. L. Hines. Meas. Sci. Technol, 1998, 9,120–127.

22 B. Widrow, M. E. Hoff. IRE WESCONConvention Record, 1960, 4, 94–104.

23 A. K. Srivastava, K. K. Shukla,S. K. Srivastava. Microelectronics Journal,1998, 29, 921–931.


24 A. A. Fekadu, E. L. Hines, J. W. Gardner. InArtificial Neural Nets and Genetic Algorithms(Eds.: R. F. Albrecht and N. C. Steele),Springer-Verlag, New York, 1993, 691–698.

25 S. Chen, C. F. N. Cowan, P. N. Grant. IEEETrans. on Neural Networks, 1991&/hf, 2,302–309.

26 J. D. Mason, PhD thesis, University ofWarwick, Coventry, UK, 1994.

27 T. Kohonen, Biol. Cybern., 1982, 43, 59–69.28 E. L. Hines, J. W. Gardner, C. E. R. Potter.

Meas. and Control, 1997, 30, 262–268.29 T. Kohonen, University of Technology,

Helsinki, Finland, 1986.30 F. Davide, C. Di Natale, A. D’Amico. Sens.

Actuators B, 1994, 18–19, 244–258.31 H. W. Shin, E. Llobet, J. W. Gardner,

E. L. Hines, C. Dow. IEE Proc. – Sci. Meas.Technol., 2000, 147(4), 158–164.

32 C. Di Natale, F. Davide, A. D’Amico.Sens. Actuators B, 1995, 23, 111–118.

33 L. A. Zadeh, Information and Control, 1965,8, 338–353.

34 B. Yea, R. Konishi, T. Osaki, K. Sugahara K.Sens. Actuators A, 1994, 45, 159–165.

35 M. M. Gupta, J. Qi. In Fuzzy Logicfor the Management of Uncertainty(Ed.: L. A. Zadeh), John Wiley, New York,1992, 479–490.

36 S. Singh, E. L. Hines, J. W. Gardner. Sens.Actuators B, 1996, 30, 185–190.

37 W. Ping, X. Jun. Meas. Sci. Technol., 1996,7(2), 1707–1712.

38 G. A. Carpenter, S. Grossberg. Comput. Vis.Graph. Image Process., 1987, 37, 116–165.

39 J. W. Gardner, E. L. Hines, C. Pang. Meas.Control, 1996, 29, 172–178.

40 G. A. Carpenter, N. Grossberg, N. Marku-zon, J. Reynolds, D. Rosen. IEEE Trans. onNeural Networks, 1992, 3, 698–713.

41 E. Llobet, E. L. Hines, J. W. Gardner,P. N. Bartlett, T. T. Mottram. Sens. ActuatorsB, 1999, 61, 183–190.

42 C.-T. Lin, C. S. G. Lee. Neural Fuzzy Systems:A Neuro-Fuzzy Synergism to IntelligentSystems, Prenctice Hall P T R, Upper SaddleRiver, 1995.

43 P. Boilot, E. L. Hines, J. W. Gardner. InSensors Update (Eds.: H. Baltes, J. Hesse andW. Gopel), Wiley-VCH, Weinheim, 2000,Chapter 4.

44 M. P. Derde, D. L. Massart. AnalyticaChimica Acta, 1986, 191, 1–16.

45 D. Michie, D. J. Spielgelhalter, C. C. Taylor.Machine Learning, Neural and StatisticalClassification, Ellis Horwood, New York,1994.

46 C. Di Natale, S. Marco, F. Davide,A. D’Amico. Sens. Actuators B, 1995, 24–25,578–583.

47 G. A. Carpenter, N. Grossberg, J. Reynolds.IEEE Trans. on Neural Networks, 1995, 6(6),1330–1336.

48 J.W. Gardner, K. C. Persaud (eds.). ElectronicNoses and Olfaction 2000, Institute of PhysicsPublishing, Bristol, 2000.


Part B

Advanced Instrumentation


7

Commercial Electronic Nose Instruments

E. Vanneste, H.J. Geise

7.1

Introduction

For a long time the human nose has been an important tool in assessing the quality ofmany products, food products being good examples. Whereas all other parts of pro-duction processes, including these of the food industry, were becoming more andmore automated, there was still no ‘objective’ means for using the ‘subjective’ infor-mation confined in the smell of products.In 1982, when G. Dodd and K. Persaud [1] of theWarwick Olfaction Research Group

presented their much-celebrated scientific publication in Nature, it heralded the be-ginning of a new technology: artificial olfaction. The expression electronic nose (EN),however, appeared for the first time in 1988. Much research is being undertaken inorder to find new and more diverse sensors while also improving the pattern recogni-tion engines, and today there are several companies offering ENs. This chapter intendsto give the reader a description of the individual companies, and explain the technologyused. For a comprehensive and detailed description of the different sensor technol-ogies and data-algorithms used in the commercially available equipment we refer toelsewhere in this book. References to previous reviews can be found here [2–8].The term EN works as an advantage as well as a disadvantage for the development of

the concept towards its applications. One might even venture to refer to the EN du-alism. The advantage is that the expression immediately evokes associations to expertsand non-experts alike for a device that measures odors. It appeals to one’s imaginationand the term is easily uttered. The disadvantage, however, is that it creates great ex-pectations, perhaps too great, because the expression suggests a faithful imitation ofthe biological sense, which is utterly incorrect: the biological sense of smell is still farsuperior over today’s artificial odor recognition. This situation will most likely con-tinue for some time. In the absence of a better term, throughout this chapter wewill consider the expressions EN and sensor array system as equivalent.As the new concept grew gradually, more and bigger players entered the market.

Presently, the EN market is characterized by three trends. We note a geographical


161161

expansion with a concomitant shift of the financial center, a scientific and technolo-gical broadening, and a conceptual extension.

7.1.1

Geographical Expansion

The commercial availability of the initial benchtop systems began in the early 1990s.Initially, the EN business used to be an almost exclusive european matter. The firstcompany active in this field was the British OdourMapper Ltd. (1992), shortly after-wards transformed to Aromascan Ltd, Crewe, Cheshire, UK. Located in the Midi-Pyr-enees, at a stone’s throw from the Mediterranean, Toulouse is the sunny host forAlpha Multi Organoleptic Systems (M.O.S.) (founded in December 1992), commonlyregarded as the present market leader.Gradually more players are participating in this emerging field. Back in the UK, in

Stansted, Essex, Neotronics Scientific Ltd. (founded in 1994) developed and sold theirapparatus NOSE (Neotronics Olfactory Sensing Equipment). The two latter companieshave amutual origin in the tandemUniversity ofWarwick/University of Southamptoncollaboration (Warwick-Southampton EN Group). In fact, quite a few of the currentmanufacturers find their cradle at a particular university, relying on them for conco-mitant scientific support. Also in the year 1994, there was an expansion to the northwith Nordic Sensor Technologies AB (Sweden) as a newcomer. Slightly before the turnof the century, we note a geographical displacement to the other side of the Atlantic,where new companies such as Cyrano Sciences and Agilent Technologies (formerlyknown as Hewlett Packard) entered the market. There is little known on the commer-cial efforts on the Australian and Asian market, although some competence centersexist.

7.1.2

Scientific and Technological Broadening

The factual starting point of EN science was the NATO advanced research workshop‘Sensors and Sensory Systems for an Electronic Nose’ [9], held in August 1991, Rey-kjavik, Iceland.In the beginning, conducting polymers (CP) were the pet subject of many research-

ers and EN producers. The systems built and commercialized by Aromascan and Neo-tronics were both based on these materials. Furthermore, Alpha M.O.S. offers theircustomers a CP-module as an option alongside their metal-oxide semiconductor(MOS) sensor modules. Soon, MOS materials became widely employed, not least be-cause of their proven usefulness in more classical sensors. Other sensitive detectingsystems were devised on the basis of other measuring principles, e.g., MOS field effecttransistor (MOSFET) and mass detection with surface acoustic wave (SAW) andquartz-crystal microbalances (QMB). On the global scale, the search is on for newtypes of chemical sensors to implement in an array, shown by the development ofcalorimetric sensors [10], optical sensors [11–13], electrochemical sensors [14], com-

7 Commercial Electronic Nose Instruments162

posite polymer-carbon black polymers [15–17], conducting oligomers [18], and phta-locyanine-based sensors. Hybrid ENs are composed of a diversity of different sensortechnologies [19–21]. But progress is not only made on the sensory part: optimizedhardware and systems design, and overall increase of better data processing algo-rithms with drift counteraction features contribute to better performing ENs.The emergence of a new promising sensor technology and its strong technological

and scientific foundation motivated existing as well as new companies to enter the ENmarket. As example, we mention OligoSense n.v. (Belgium) as a starter, which pro-duces sensor materials and sensor arrays. Existing sensor producers such as QuartzTechnology Ltd. (UK), HKR Sensorsysteme GmbH (Germany), Bloodhound SensorsLtd. (U.K.), Marconi Applied Technologies (U.K.), and Microsensor Systems Inc.(U.S.), offer sensor arrays to implement in existing nose platforms as a moduleand/or gradually profile themselves as producers of complete sensor systems butwith a focus on sensors. The former approach is preferential since it avoids perfusionof complete sensor-array systems, which then have to compete on an emerging and toocrowded market. Also, it allows an optimal use of available hardware, and no precioustime and effort is lost on the repeated design of complete systems. Furthermore, thesensor designer can concentrate on the technology by which implementation of im-provements is accelerated, the supply is increased and the design of application spe-cific sensors and arrays is facilitated.Finally, some other companies are constructing systems for a dedicated application

such as Element, Iceland (e.g., quality control of fish), Environics Industry Oy, Finland(e.g., military-industrial), and WMA Airsense Analysentechnik, Germany (e.g., envir-onmental).It is of interest to see that, at least for the time being, established classical sensor

producers such as Figaro (Japan), Capteur (U.K.), FIS (Sweden), and Drager (Ger-many) do not take the risk, but that their products find their way to prominent ENconstructors.

7.1.3

Conceptual Expansion

One can acknowledge three conceptual displacements. First, the definition by Gardnerand Bartlett [5] ‘an electronic nose is an instrument which comprises an array of elec-tronic chemical sensors with partial specificity and an appropriate pattern recognitionsystem, capable of recognizing simple or complex odors’ became inaccurate whenmass spectrometric detection (SMart Nose and Agilent Technologies) or (flash-)gas chromatography-based separation adjoining SAW-sensor detection (ElectronicSensor Technology) were introduced. Secondly, handheld devices have their own ty-pical target market (e.g., leakage detection) where the low concentrations of typicalodorous molecules are not of primordial importance, but where people are interestedin detecting rather high (> 100 ppm) concentrations of (predominantly organic) vo-latile compounds. Last but not least, the array principle is conveyed to the wet phasewhere potentiometric and amperometric chemical sensors form the building blocks


for an electronic tongue (first record of the expression in 1993 [22]). The commercialavailability of such a system is currently restricted to Astree from Alpha M.O.S. Sincethis new technology falls beyond the scope of this chapter, we would like to refer thereader to the literature [23–38] and Chapter 11.

7.2

Commercial Availability

7.2.1

Global Market Players

See Table 7.1 for a summary of the main players and the basis of their instruments.

Tab. 7.1 Electronic Nose Manufacturers, Models and Sensor Cores

Company Sensor Core System

Agilent Technologies MS 4440

Alpha M.O.S. MOS, CP, SAW

MS and MS-EN

electronic tongue

Fox, Centauri

Kronos & Prometheus

Astree

Applied Sensor MOSFET, MOS, QCM

4 � MOS, 8 � QCM

QCM

3320, 3310

VOCseries1

VOCcheck1

Bloodhound Sensors CP BH114

Cyrano Sciences Inc. CP (composite) Cyranose 3201

Daimler Chrysler Aerospace QCM, SAW, MOS SAM system

Electronic Sensor Technology SAW zNose

Element MOS FreshSense

Environics Industry IMCELL MGD-1

Forschungszentrum Karlsruhe MOS, SAW Sagas

HKR Sensorsysteme QCM, MS QMB6

Lennartz Electronic QCM, MOS, electrochemical MosesII

Marconi Applied Technologies CP, MOS, QCM e-Nose 5000

Microsensor Systems SAW ProSat

Osmetech CP OMA and core sensor module

Quartz Technology QCM QTS-1

SMart Nose MS Smartnose-300

WMA Airsense Analysentechnik MOS PEN

1 handheld deviceCP conducting polymerIMCELL ion-mobilityMOS metal-oxide semiconductorMOSFET metal-oxide semiconductor field-effect transistorMS mass spectrometry-basedQCM quartz-crystal microbalanceSAW surface accoustic wave


7.2.1.1 Alpha M.O.S.

As mentioned earlier, Alpha M.O.S. is one of the pioneers on the ENmarket. Since itsestablishment in 1992, the company has seen a steady growth, which resulted in astock market quotation on 2 April 1998 in Paris on ‘Le Nouveau Marche’ (ticker:6280). It has settlements and branch offices in France, the United States, the UnitedKingdom, and Germany, while Bran&Luebbe tends to its distribution network in otherparts of the world. This makes the company a global player.Without any doubt, AlphaM.O.S. also has the largest range of different systems. We

mention the modular laboratory system FOX, which encompasses FOX2000,FOX3000, FOX4000 and FOX5000, systems which contain one, two, three andfour sensor arrays, respectively, each of which containing six sensors. A standard ar-ray board consists ofMOS sensors, of which two are available and can be extended witha QMB and/or CP board. The company engages also in the mass-spectrometric ap-proach of olfactometry with their device a-Kronos. In addition they intend to intro-duce Centauri, a new hyphenated technique that couples an EN to a mass-spectro-metric module. In the near future, they intend to introduce the first commercial elec-tronic tongue, under the name Astree Liquid and Taste Analyser.The software used to interpret the data, called a-Soft, originated as National Instru-

ments’ Labview and has now reached its seventh release. It allows techniques likeprincipal component analysis (PCA), projection to latent structure (PLS), and artificialneural networks (ANNs), as well as a transferability utility to convert data from dif-ferent systems (i.e., slightly different sensors and systems).Alpha M.O.S. took the initiative and in 1993 organized the first ‘International Sym-

posium on Olfaction and the EN’. This initiative was taken over by the academic worldin 1998.

7.2.1.2 AppliedSensor Group

On December 4th, 2000, Nordic Sensor Technology and MoTech announced the mer-ger between them. The new alliance is called AppliedSensor Group, with offices inboth Sweden and the USA.

MoTech Sensorik, Netzwerke und Consulting GmbH

Founded by a couple of researchers from the MOSES II project at the University ofTubingen, this company initially provided services, sensors, and software to the noseproducer Lennartz Electronic GmbH. Their own developments include a scale of por-table and handheld sensor array systems based on Tagushi andmass-sensitive devices.Scientific backup and cooperation is provided by the Steinbeis Transfer Center forInterface Analysis and Sensors, and the Institute for Physical and Theoretical Chem-istry at the University of Tubingen.There are four members in the VOCmeter series: VOCmeter MOS, VOCmeter

QMB, VOCmeter HYBRID (ranging from Q 11 400–Q 18 900) and the VOCmeterVARIO (priced at Q 7900 exclusive of sensors, individual sensors at Q 490) to measure8 external sensors. Signal recording and processing of the VOCmeter series is per-


formed using a RS232-linked PC with an uncomplicated user interface called Argus.The VOCcheck (Fig. 7.1) is a handheld device based on QMB sensors, allowing rapid(< 10 s) identification of volatile compounds, especially in the field of leakage detec-tion and emission control, and is a comparative method against pre-measured refer-ence samples. OEMmodules on the basis of the VOCmeter technology can be adaptedfor a large variety of different requirements.

Nordic Sensor Technologies AB

The origin of this leading company was the research made at the University of Lin-koping (Sweden) in 1994, known by the name Nordic Sensor. A financial injection inMarch 1996 lead to the formation of Nordic Sensor Technologies AB. There is still atight symbiosis with the research group ‘Laboratory of Applied Physics’ and the Swed-ish sensor center ‘S-Sence’: the home of sensing MOSFETs [39] since 1975.The successor for the first NST 3210 Emission Analyser is the NST 3220 Lab Emis-

sion Analyser. These systems are available for atline (batch) and online (continuous)quality control measurements. Improvements on this blue-and-graymachine includedan uncomplicated carousel, allowing 8 specimens in vials of 250 ml to be sampled. At asecond stage, the carousel was thoroughly re-examined. This resulted in a 12-positioncarousel, allowing heating (up to 65 8C) and cooling of the samples: the NST 3320 EN(Fig. 7.2).Although the gas sampling and system design was largely adapted, the core sensor

technology remained the same, based on two arrays of 5 MOSFET sensors (at differentoperating temperatures 140 8C and 170 8C) and one array of 5 MOS sensors. Optionalsensors include CO2 IR devices (1% or 10%) and other in-house mass sensitive de-vices. Its modular principle allows one to include other sensor technology-based ar-rays.The company is targeting quality control, process control, environmental analysis,

and medical diagnosis. One important breakthrough was reported in the field of on-

Fig. 7.1 AppliedSensor’s

handheld VOCcheck. Reprinted

with kind permission


line monitoring of fermentation and other bioprocesses [40]. In addition, AppliedSen-sor also focuses on OEM (technology platform) and component sales.The proprietary data-acquisition and data-processing software Senstool (current re-

lease 2.7.4.26) is a straightforward Windows-based graphical user interface (GUI). Itcontains PCA, PLS and ANN algorithms. The files are saved in the Microsoft Excelformat, allowing easy (re-) processing.

Resulting Merger: AppliedSensor

Themerger of Nordic Sensor Technology andMoTech leads to a powerful company bybringing together a massive amount of knowledge such as sensor technologies, dataprocessing, hardware, and software. This global player offers a variety of handheld andbenchtop sensor array systems. AppliedSensor will be a fearsome opponent of AlphaM.O.S., let the battle begin!

7.2.1.3 Lennartz Electronic

Lennartz Electronic GmbH has more than 30 years of experience in physical sensorsand high-quality data acquisition systems. Their modular EN is called MOSES (MOd-ular SEnsor System). MOSES II (Fig. 7.3) has been developed in close cooperation withSteinbeis-Transferzentrum Grenzflachenanalytik und Sensorik at the Universitat Tu-bingen (Center for Interface Analytics and Sensors). Lennartz Electronic GmbHuses abasic sensor configuration, consisting of eight commercially available Tagushi sensorsand eight quartz microbalance sensors coated with different polymers. These quartzmicrobalances are manufactured at the Steinbeis-Transferzentrum Grenzflachenana-lytik und Sensorik. Under current investigation is a calorimetric module.

7.2.1.4 Marconi Applied Technologies (now ELV Technologies)

Marconi Applied Technologies is a general designer and manufacturer of electroniccomponents. It acquired EEV Chemical Sensor Systems, which was formerly knownas Neotronics Scientific. The eNOSE 5000 range of instruments was originally de-

Fig. 7.2 AppliedSensor’s Electronic Nose Model 3310. Reprinted

with kind permission


signed for the laboratory-based profiling of samples throughmeasurement of the char-acteristic sample headspace. Marconi Applied Technologies no longer markets theeNOSE 5000 instruments for general-purpose laboratory applications. Instead ithas developed a real-time monitoring system based on chemical sensor-array technol-ogy ProSAT (for atline or online monitoring), predominantly for bioprocessing, fer-mentation monitoring, food industrial applications, water and wastewater treatment,and the chemical industry. As a consequence of its expertise in sensor development, italso has proprietary libraries of discrete sensors using conducting polymers (CP),MOS (SnO2; CrTiO2), SAW (@260 MHz), and QMB (@10 MHz) and hence canform custom-made arrays. Investigation is focused towards the development of mo-lecularly specific sensors. In the discrete sensor product range a variety of sensors areoffered such as pellistor-type catalytic gas sensors, thermal conductivity and infrared-based sensors (about 35 proprietary sensors in total). An additional benefit of all Mar-coni’s sensor designs is that individual sensors can be substituted or replaced withinan array, allowing for array optimization. Tight manufacturing control ensures thatsensor reproducibility is high and preserves training model validity when sensorsare replaced. Typical sensor arrays contain between 4 and 12 sensors, with 8 inthe standard configuration.Some of these are based on standard multivariate techniques such as PCA, multiple

discriminant analysis (MDA), canonical analysis (CA), and ANNs. Advanced calibra-tion algorithms are used to compensate for long-term sensor drift and to ensure va-lidity of data sets from module to module.

7.2.1.5 Osmetech plc

From as early as 1980, research has been conducted at the University of ManchesterInstitute of Science and Technology (UMIST) to come up with an instrumental equiva-lent of the biological nose. The great originality of the project was the use of CP sensorswith their broad sensitivity to various vapors coupled to an extensive data processingsystem. The work led to the first operational prototype in 1990. The establishment ofthe spin-off company, OdourMapper Ltd, by a group of researchers related to this

Fig. 7.3 Modular Sensor Sy-

stem II from Lennartz Electronic.

Reprinted with kind permission


project followed in 1992. As early as 1994, the spin-off went to the Alternative Invest-mentMarket (now London Stock Exchange, TechMARK, ticker OMH). It collected £11million in this stock market quotation, while converting into AromaScan plc. Thecompany became market leader in its field, and was awarded the Prince of Wales’Award for Innovation. Its competence shows in more than 25 patents and publica-tions on data processing [41] as well as on sensor design and development [42, 43].They patented a method and apparatus for detecting microorganisms, and enteredthe area of biomedical applications [44]. After some wanderings in various parts ofthe quasi-infinite number of possible areas of applications, the company focusedon biomedical uses, and changed its name to Osmetech plc. (1999) in the process.Their emphasis is now on the detection of volatile metabolites excreted in bacterialinfection of the urinary tract, bacterial vaginosis, early diagnosis of bacterial pneumo-nia, and bacterial pharyngitis. In addition to these biomedical applications the com-pany sells industrial systems applicable to the quality control of basis products used ine.g., health and body care, plastics, and polyurethane foam.The old AS32 systems, with their external humidity controller and 20 to 32 organic-

based sensors (see Fig. 7.4), have been replaced by a new and upgraded line of appa-ratus, OsmetechMicrobial Analyser� (OMA). The new apparatus houses 50 glass vials,capped with a spectrum through which the headspace can be purged (dynamic head-space sampling). The sensor section of these systems is constructed as an independentmodule, the so-called Core Sensor Module (CSM). The CSM contains up to 48 sensorssituated on a circular substrate. It also contains the essential temperature controllerand the electronics for signal processing and the data acquisition interface. Dedicatedsensor arrays specifically designed for certain clinical infections are offered, while auniversal CSM suffices for the other areas of applications.

Fig. 7.4 The heart of Osmetech’s sensor module

is this substrate equipped with 48 sensors


7.2.2

Handheld Devices

7.2.2.1 AppliedSensor Group

For a detailed description see Section 7.2.1.2.

7.2.2.2 Cyrano Sciences, Inc.

Cyrano Science was founded in 1997, and raised over $12 million to further developthe patented [16] original composite polymer technology, elaborated by Nathan Lewiset al. at the California Institute of Technology. The company holds 10 US and 1 Eur-opean patents in total, the last one in the emerging and remunerative field of medicaland biomedical applications. The Cyranose 320 (Fig. 7.5) is a handheld device compris-ing a 32-polymer composite (polymers filled with the conductive particles carbon blackor another conductive filler) sensor array (Nose Chip�). The launch of this $9,000-priced handheld device took place at the technology exhibition Pittcon2000 in NewOrleans. The detection limit of the Cyranose320 for different volatile compoundsis estimated roughly at 0.1% of the standard vapor pressure.The dedicated on-board firmware (current release 30.1) is capable of differentiating

six different classes for eachmethod stored. The instrument settings, definedmethodsand raw data can be swapped, stored and further processed on a Windows-based PCusing PCnose software (current release 6.5). To share the knowledge optimally, col-laboration agreements were signed with Agilent and Osmetech. The Osmetech agree-ment comprehends a Healthcare Collaboration Agreement, in which Osmetech poly-mer sensors will be implemented in the Cyranose 320 and used for validation on thedetection of the presence of bacteria in urine causing urinary tract infections. Agilenthas signed a collaborative research agreement with Cyrano Sciences, sharing amongother things the Infometrix Pirouette software.

Fig. 7.5 The Cyrano 320 handheld

device from Cyrano Sciences


7.2.2.3 Microsensor Systems, Inc.

Ever since 1979, Sawtek has been a dedicated SAW device developer for a countlessnumber of applications in communications, cellular wireless data transmission, andother signal-processing applications. In 1998 it merged with Microsensor Systems, acompany developing chemical sensing technology using the same SAWs. Using anadvanced, polymer-coated SAW array, a broad spectrum of chemical vapors can beaccurately identified. SAW sensors have excellent long-term stability and are effectivesensors for higher molecular weight, semi-volatile organic compounds not readilydetected by other sensor technologies. VaporLab (Fig. 7.6), a handheld, battery-pow-ered chemical vapor identification system costing $10 000, goes where you need it,providing on-the-spot information on the current status of your process, product,or environment so that immediate action can be taken as required. Typical foremostapplications include environmental, food and beverage, fragrance and cosmetics,safety exposure and personal monitoring, and medical and dental.

7.2.3

Enthusiastic Sensor Developers

7.2.3.1 Bloodhound Sensors Ltd.

CP sensor research work for Bloodhound Sensors began at the University of Leeds,where the company is currently based. The rather compact BH114 is an instrumentcomprising an array of 14CP sensors, and the data processing is performed usingMicrosoft Excel add-ins and specialized add-ins such as Neuralyst� from PalisadeCorp. The sensor technology is based on CPs and discotic liquid crystals. These de-vices are also available individually or in an array.

7.2.3.2 HKR Sensorsysteme GmbH

HKR Sensorsysteme was founded in 1993 by three researchers from the TechnicalUniversity of Munich. An array of six QMBs forms the heart of their benchtop EN

Fig. 7.6 Microsensor systems’ handheld device

VaporlabTM


consisting of an automated Perkin-Elmer HS 40XL or Dani HSS86.50 headspace-sam-pler and the proprietary QMB6 array. Optionally, a thermal desorption trap (MarkesInt.) can solve the problem of too low concentration by trapping the analytes of interestand purging the high volatile compounds, allowing analytes in the lower parts perbillion range to be detected. MS – Sensor� is a sensor system on a mass spectrometricbasis, using a quadrupole mass spectrometer, TurboMass.Qmbsoft for Windows NT controls the automated measurements and acquires and

evaluates the data using PCA, GDF, and RBF neural network pattern recognition tech-niques.

7.2.3.3 OligoSense n.v.

Research on sensors for an EN has been conducted since 1993 at Antwerp University.The original focus was on electrically conducting polymers, however it was noted thatshort fragments of these polymers, oligomers, have better sensory properties thantheir polymeric analogs [18], and the investigation was then concentrated on thisarea. This new focus of investigation will hopefully lead to a steady stream of newsensor materials and sensor modules. As a consequence, OligoSense n.v. has beenformed to produce and market the oligomeric technology.

7.2.3.4 Quality Sensor Systems Ltd.

Q-Sensor developed a chemical sensor array instrument dedicated to applications inthe food and food packaging industry. The QMBA8000, based on eight QMB sensors,has been developed on a generic platform, and it is this modular approach to designwhich allows chemical sensor systems to be developed for a diverse range of applica-tion areas by offering the appropriate sampling system and chemical sensor array.

7.2.3.5 Quartz Technology Ltd.

Started in March 1996, Quartz Technology’s main objective is to commercialize QMB-based sensor technology. Nowadays, they market their standard balanced eight-sensorarray instrument QTS-1, and in addition a range of separate QMB sensors and evenblank quartz crystals are available. Focusing on applications, the QTS-1 can beequipped with a custom array, or even dedicated systems with more sensors canbe designed. This company also provides custom solutions to specific measurementproblems. The compact system accepts at its inlet (no carousel) sample air from jars orvials or introduced from an external sampling system. The sensor signals are pro-cessed and compared to an online library for rapid identification. The software is writ-ten for a Windows98/NT platform. Although Quartz Technology would never claimthat QTS-1 is an EN, it is capable of diagnosing many aroma problems based on differ-ing chemical fingerprints.


7.2.3.6 Technobiochip

Ever since 1995, research at the Tor Vergata University in Rome has been carried onporphyrins and related compounds for coatingmass transducers for chemical sensors.Themain feature of such sensors is the dependence of the sensing properties (in termsof selectivity and sensitivity) on the nature of the central metal and on the peripheralsubstituents. Technobiochip produces this instrument, named LibraNose, and ships itwith a number of data processing algorithms based on PCA, CA and ANNs that areused for information extraction.

7.2.4

Non-Electronic Noses

This section deals with systems that don’t meet the strict definition as explained earlierthis chapter (Section 1.3). Note that Alpha M.O.S. (see Section 2.1.1) and HKR Sen-sorsysteme (see Section 2.3.2) also offer a mass spectrometry-based system.

7.2.4.1 Laboratory of Dr. Zesiger

Usingmass spectrometry should overcome the typical chemical sensor problems suchas their sensitivity towards sample and environmental moisture. The advantages ofthis technique are its sensitivity and robustness. Contrary to GC/MS, there is no pre-ceding separation of the volatile constituents allowingmeasurements every 5 minutes.However, the basic operation of this kind of equipment needs an adapted gas supply(helium) and requires high vacuum pumps inextricably bound up with a high systemprice. The price of a nose on the other hand is predominantly determined by the re-search and development contribution and could eventually go down substantiallywhen the market increases.SMart Nose (Fig. 7.7) is a fully automated combination of a Balzers Instrument Inc.

quadrupole mass spectrometer with an autosampler for 2 ml or 20 ml vials. The sys-tem is entirely software controlled: the Quadstar� from Balzer Instruments controls

Fig. 7.7 SMart Nose mass

spectrometer with autosampler


operation of the mass spec, channel selection and sample measurement. The SMartNose software processes the raw mass spectrometric data using statistical algorithmssuch as PCA or discriminant function analysis (DFA), to yield a more user friendlyrepresentation of the results.

7.2.4.2 Agilent Technologies, Inc.

Agilent Technolgies (formerly known as Hewlett-Packard) is a well-known manufac-turer of all kinds of analytical instruments, in fact a scientific instrument giant. Bycombining a recent in-house mass spectrometer 5973NMSD with an in-house mod-ified headspace autosampler 7694, the Agilent system uses quadrupole technology as amass sensor to provide qualitative information about sample attributes. The Infome-trix software allows the raw data to be processed for classification purposes usingmultivariate techniques and pattern recognition. The Agilent 4440A Chemical Sensor(Fig. 7.8) is priced at $80 000, and is currently available through Gerstel GmbH.Agilent and Cyrano signed a pact to jointly develop new versions of their ENs, con-

ceivably expanding the mass spectrometer with a classical composite polymer-basedsensorarray, and to collaborate on marketing.

7.2.4.3 Illumina, Inc.

Optical sensing technology [12] has been reported by Dickinson et al. of Tufts Uni-versity. Illumina licensed this technology, and has recently started to market anEN. Until now, their main focus was on the large-scale analysis of genetic variationand function. Illumina’s technology is also suited for chemical detection applications,because their BeadArray� fiber optic bundles can be designed to ‘house’ cross-reac-tive, nonspecific sensors capable of responding to a wide variety of solvent vapors.

7.2.4.4 Electronic Sensor Technology, Inc.

Electronic Sensor Technology produces the zNose�, which consists of only a singlepatented sensor based on SAW technology and a directly heated 1 m length of capillarychromatography column. Visualization software for making radar plots, EST System

Fig. 7.8 Agilent 4440A Chemi-

cal Sensor


Software for Windows95, is used. A benchtop and a handheld version of the zNose�,priced at $25 000, are offered commercially.

7.2.5

Specific Driven Applications

7.2.5.1 Astrium

Astrium is a subsidiary of RST Rostock Raumfahrt und Umweltschutz GmbH, whichbelongs to the DaimlerChrysler Aerospace division. The EN technology Sam is a mea-surement technique suitable for an objective, quick, and low cost analysis of odor,aromas, and volatile compounds. They offer a range of three sensor systems basedon a modular concept using MOS, QMB, and SAW technology.

7.2.5.2 Element Ltd.

Element started developing gas detectors in co-operation with the Science Institute atthe University of Iceland, in 1992, then under the name RKS Sensor Systems. Therelationship with the university is still maintained. The gas detector systems formthe main product line of the company together with Medistor, a data acquisition sys-tem.In a project in co-operation with the Icelandic Fisheries Laboratories, Element has

developed an instrument called FreshSense to detect fish freshness. FreshSense de-tects components that are produced in fish during storage and gives comparable re-sults to traditional methods to evaluate freshness such as sensory analysis. FreshSenseis built on an array of six commercially available electrochemical sensors (Drager) withPCA and PLS algorithms to classify samples.

7.2.5.3 Environics Industry Oy

The Environics company targets chemical detection applications for the military. De-tection is based on a proprietary ionmobility cell (IMCELL�), where sample moleculesare first ionized using some radioactive source (e.g., Am241) and then flow towards anarray of six detector electrodes. The MGD-1 Industrial Multi-Gas Detector can be usedas a portable or a fixed version. VisualNose for Windows is software designed to pre-sent the data that has been collected with MGD-1 in 2D format.

7.2.5.4 WMA Airsense Analysentechnik GmbH

WMA Airsense’s portable EN PEN2 consists of an array of 10MOS sensors withadapted software. It is designed for laboratory measurements as well as for onlineprocess monitoring. Focusing on air pollution and air quality control measure-ments, an optional enrichment unit (EDU2 – absorbent trapping on Tenax�) canbe valuable for this Q 14 900 unit. For operation in hostile industrial environments,an industrial process control EN, i-PEN, is offered. Different configurations of thei-PEN are available: a basic module i-PEN-MOD (based on 10MOS devices) has an


on-boardmicrocontroller; a process control nose i-PEN-PCN consists of a sensor array,gas pumps, and a patented sampling system. The i-PEN-ET has an additional enrich-ment and desorption unit with A3-technology (automatic ranging, automatic calibra-tion, and automatic enrichment). The software provided forWindows NT4 incorporatePCA and LDA algorithms for visualization of the data and DFA and ANNs for classi-fication and online evaluation. Recommended prices are Q 4900 for the i-PEN-MODmodule to Q 14 900 for the PEN-2.

7.3

Some Market Considerations

Prudence is called for when assessing the size of the EN market. The estimates rangefrom a modest Q 10 million to a dazzling Q 4.5 billion globally a year predicted by theEconomist [45] based on a worldmarket of 100 000 units sold annually in the first yearsof the 21st century.The best we can do is to give below the most recent results of a short list of market

evaluation studies:

1. David Walt of Illumina and Tufts University estimated that 200 units were sold inthe last five years (1994–1998) [46].

2. In April 1998, the Wall Street Journal published an estimate of the market at thattime to the amount of Q 10–15 million [47]. A document by Greenberg [48] esti-mated the market value at Q 15 million, and these figures seem to be acceptable.

3. According to Bartlett and Gardner [2], the market is estimated at about Q 145 mil-for the year 2000. This estimate is corroborated by a Technical Insights report [50]that states the sales of 2500 units.

Fig. 7.9 Incomplete Overview of trade volume in 1000’s Q for five leading companies (Alpha M.O.S.,

Bloodhound, Lennartz, Osmetech, and Neotronics) for the period 1994–1998 [51]


4. The German Intotech Consulting Group foresees a market potential of Q 1.2 billionby the year 2004. The British/American journal The Economist even gives a poten-tial of Q 4.5 billion annually [45].

By examining the turnover figures coming from Graydon reports of 5 leading com-panies at that time (Alpha M.O.S., Bloodhound, Lennartz, Osmetech, and Neotronics)for the period 1994–1998 [51], our conclusions are somewhat modest.The figures for 1997 can be extrapolated for all players active in those days. If one

considers a market penetration of NST (with an estimated penetration of 15% at thattime), a rough estimate of the market at Q 10 million seems to be acceptable.Unfortunately, we don’t have the most recent figures, but it would seem there was

no explosion of the market. Taking into account an annual growth rate of 7.5% (whichis commonly used for the market of ‘classical’ analytical devices for the period 1995–2000), themarket would be worth around Q 13.4 million for 2000. If we take the annualgrowth rate of the US market for ‘new generation’ analytical instruments and com-ponents [52] of 19%, the market would be worth some Q 20.0 million for 2000.The newcomers with handheld devices have added an additional newmarket, which

falls beyond the scope of this deduction. The producers express optimistic views withregard to the trade volume. The total market for electronic noses was $ 140 million for1998 and is projected to be $ 200 million by the year 2003 [49].There is of course much space for other interpretations. These figures demonstrate

the large uncertainties in evaluating the young EN market. It is clearly an emerginghigh-tech market with enormous potential as well as high risks. Therefore, it is ofinterest to look at the amount of venture capital that is invested in EN technologyup to now (May 2001). A conservative estimate says that well over Q 350 millionhas been invested in this technology throughout the years, of which Osmetech aloneaccounts for some Q 70 million. This reveals that investors have an optimistic outlookon the growth potential of this emerging technology, however a loss of Q 6.4 millionwas reported for Osmetech for the year 2001.

References

1 G. H. Dodd, K. C. Persaud. Nature, 1982,299, 352–355.

2 J. W. Gardner, P. Bartlett. –8.3 CommercialInstruments, in Electronic Noses: Principlesand Applications. 1999, Oxford UniversityPress: Oxford. p. 194.

3 D. J. Strike, M. G. H. Meijerink,M. Koudelka-Hep. Fresenius Journal ofAnalytical Chemistry, 1999, 364, 499–505.

4 M. A. Craven, J. W. Gardner, P. N. Bartlett.Trends in Analytical Chemistry, 1996, 15(9),486–493.

5 J. W. Gardner, P. N. Bartlett. Sensors andActuators B: Chemical, 1994, 18(1–3),211–220.

6 E. Vanneste. Review on the commercialavailability and research efforts on electronicnoses. http://nose.uia.ac.be/review.

7 H. T. Nagle, R. Gutierrez-Osuna,S. S. Schiffman. IEEE Spectrum, 1998, 35(9),22–34.

8 E. Zubritsky. Analytical Chemistry, 2000,72(11), 421A–426A.

9 J. W. Gardner, P. N. Bartlett, eds. Sensors andSensory Systems for an Electronic Nose. NATOASI Series: Applied Science. Vol. 212. 1992,Kluwer Academic Publishers: Dordrecht,the Netherlands. p. 327.

10 J. Lerchner, D. Caspary, G. Wolf. Sensorsand Actuators B: Chemical, 2000, 70(1–3),57–66.


11 J. White, J. S. Kauer, T. A. Dickinson, D. R.Walt. Analytical Chemistry, 1996, 68(13),2191–2202.

12 T.A. Dickinson, J. White, J. S. Kauer,D. R. Walt. Nature, 1996, 382(6593),697–700.

13 D. R. Walt, T. Dickinson, J. White, J. Kauer,et al.. Biosensors and Bioelectronics, 1998,13(6), 697–698.

14 J. Stetter, P. Jurs, S. Rose. AnalyticalChemistry, 1986, 58(4), 860–866.

15 B. J. Doleman, M. C. Lonergan, E. J. Severin,T. P. Vaid, et al.. Analytical Chemistry, 1998,70(19), 4177–4190.

16 N. S. Lewis, M. S. Freund. US5571401:Sensor arrays for detecting analytes in fluids,1996.

17 M. Lonergan, E. Severin, B. Doleman,S. Beaber, et al.. Chemistry of Materials, 1996,8(9), 2298–2312.

18 M. De Wit, E. Vanneste, F. Blockhuys,L. J. Nagels, et al.. Chemically sensitive sensorcomprising arylene alkenylene oligomers,EP0878711; JP11072474; US6042788, 1997.

19 H. Ulmer, J. Mitrovics, U. Weimar,W. Gopel. Sensors and Actuators B: Chemical,2000, 65(1–3), 79–81.

20 H. Ulmer, J. Mitrovics, G. Noetzel,U. Weimar, et al.. Sensors and Actuators B:Chemical, 1997, 43(1–3), 24–33.

21 M. Holmberg, F. Winquist, I. Lundstrom,J. Gardner, et al.. Sensors and Actuators B:Chemical, 1995, 26–27, 246–248.

22 P. Wide, F. Winquist.WO9913325 ElectronicTongue, 1993.

23 Y. G. Vlasov, A. V. Legin, A. M. Rudnitskaya,C. DiNatale, et al.. Russian Journal of AppliedChemistry, 1996, 69(6), 848–853.

24 C. Di Natale, A. Macagnano, F. Davide,A. D’Amico et al.. Sensors and Actuators B:Chemical, 1997, 44(1–3), 423–428.

25 A. Legin, A. Rudnitskaya, Y. Vlasov,C. DiNatale, et al.. Sensors and Actuators B:Chemical, 1997, 44(1–3), 291–296.

26 Y. G. Vlasov, A. V. Legin, A. M. Rudnitskaya,A. D’Amico et al.. Journal of AnalyticalChemistry, 1997, 52(11), 1087–1092.

27 Y. Vlasov, A. Legin, A. Rudnitskaya. Sensorsand Actuators B Chemical, 1997, 44,532–537.

28 F. Winquist, P. Wide, I. Lundstrom. Analy-tica Chimica Acta, 1997, 357(1–2), 21–31.

29 Y. Vlasov, A. Legin. Fresenius Journalof Analytical Chemistry, 1998, 361(3),255–260.

30 P. Wide, F. Winquist, P. Bergsten,E. M. Petriu. IEEE Transactions on Instru-mentation and Measurement, 1998, 47(5),1072–1077.

31 A. V. Legin, A. M. Rudnitskaya, Y. G. Vlasov,C. Di Natale, et al.. Sensors and Actuators B:Chemical, 1999, 58(1–3), 464–468.

32 F. Winquist, I. Lundstrom, P. Wide. Sensorsand Actuators B: Chemical, 1999, 58(1–3),512–517.

33 C. Di Natale, R. Paolesse, A. Macagnano,A. Mantini, et al.. Sensors and Actuators B:Chemical, 2000, 64(1–3), 15–21.

34 A. Legin, A. Rudnitskaya, Y. Vlasov,C. Di Natale, et al.. Sensors and Actuators B:Chemical, 2000, 65(1–3), 232–234.

35 L. Rong, W. Ping, W. L. Hu. Sensors andActuators B: Chemical, 2000, 66(1–3),246–250.

36 F. Winquist, S. Holmin, C. Krantz Rulcker,P. Wide, et al.. Analytica Chimica Acta, 2000,406(2), 147–157.

37 K. Toko. Measurement Science & Technology,1998, 9(12), 1919–1936.

38 C. Krantz-Rulcker, M. Stenberg,F. Winquist, I. Lundstrom. AnalyticaChimica Acta, 2001, 426(2), 217–226.

39 I. Lundstrom, S. Shivamaran, C. Svensson,L. Lundqvist. Applied Physics Letters, 1975,26(2), 55.

40 T. Bachinger, P. Martensson,C.F. Mandenius. Journal of Biotechnology,1998, 60(1–2), 55–66.

41 K. C. Persaud, P. J. Wells. Pattern RecognitionWith Combination Of Mappings, EP0909426;WO9801818, 1998.

42 K. C. Persaud, P. Pelosi. SemiconductingOrganic Polymers, EP0766819; WO9600384,1996.

43 K. C. Persaud, P. Pelosi. Semiconductingorganic polymers for gas sensors, EP0766818,US5882497, 1999.

44 P. A. Payne, K. C. Persaud. Method andapparatus for detecting microorganisms,EP0765399, US5807701, 1998.

45 The Economist, Artificial Noses. Now to sniffat., Sept 5 1998.

46 Electronic Noses Grow Up: Versatile Sensors ontheir Way to Market, Technical Insights, JohnWiley, 1998.


47 Wall Street Journal, Electronic-nose firm seekssweet smell of success, April 20 1998.

48 I.Greenberg.TechnologyReview,August1998.49 M. Bourne, Intelligent Sensing: Micro Noses,

Eyes and Tongue, G236, Business Communi-cations Company, 1999.

50 Electronic Noses: Detection Revolution for Food,Chemical and Healthcare Industries, Marketfor electronic noses. 1998, New York, NY, USA:Technical Insights/Frost & Sullivan.

51 OligoSense. Vooronderzoek met betrekkingtot ontwerp van een protoype oligomeersensorenmodule voor implementatiein bestaande toestellen, te omschrijven alselektronische neuzen, 1999, 12., Antwerp (inFlemish).

52 C. Wrotnowski. Business CommunicationsCompany, G171, 1998. (The New Generationof Analytical Instruments).


8

Optical Electronic Noses

Todd A. Dickinson, David R. Walt

8.1

Introduction

A tremendous amount of technical infrastructure and scientific development has ta-ken place in the area of optics, optical communications, and optical hardware over thelast several decades. These developments have led to new light sources, such as solid-state lasers, laser diodes, and light-emitting diodes (LEDs). Improved materials forconducting light, such as optical fibers and optical fiber arrays, have been devel-oped. Revolutions in detector technology have also taken place; high sensitivity detec-tors, such as avalanche photodiodes, have been developed with the ability to detectsingle photons. Array detectors, such as charge coupled device (CCD) cameras, inten-sified CCD cameras (ICCD), and CMOS (complementarymetal oxide semi-conductor)detectors are in widespread use for such applications as digital photography and as-tronomy. Color versions of these array detectors are also being introduced commer-cially. In addition to these components, significant advances in materials science haveled to new types of filters, dichroics, light-directing components such as micromirrorarrays, and infinity optics. Most of these devices and components have been developedto advance the telecommunications, entertainment, and computer industries for suchapplications as fiber-optic communications, digital music, projection devices, and op-tical information storage. With the advent of these new capabilities, a parallel devel-opment has been taking place in the field of optical sensing.

8.1.1

Optical Sensors

Optical sensors are devices that measure the modulation of a light property. Examplesinclude changes in absorbance, fluorescence, polarization, refractive index, interfer-ence, scattering, and reflectance. Optical sensors are comprised of four basic compo-nents: 1) a light source to interrogate the sensor; 2) suitable optics for directing light toand from the sensor; 3) a detector for detecting the light signal coming from the sensor;


181181

and 4) the sensor itself. In the simplest type of sensor, referred to as an intrinsic sensor,the chemical species being measured carries its own signal. For example, some or-ganic molecules absorb light at specific wavelengths, or fluoresce and thereby emitlight at particular wavelengths. These molecules can be detected directly by measuringchanges in absorbance or fluorescence at their absorption or emission wavelengths,respectively. In these systems, the ‘sensors’ are the molecules themselves. Thus, onlythe three instrument components are required as the sensor transduction mechanismis intrinsic to the molecule or molecules being detected.In the more common type of optical sensor, an indicating species is employed.

These types of sensors are referred to as extrinsic sensors. Indicators can be dyes,polymers, or other materials that interact with the chemical species of interest, theanalyte, to produce signal modulation. For example, an optical sensing materialcan be prepared by attaching a chemically sensitive dye to a substrate. When an ana-lyte interacts with the sensing material, an absorbance or fluorescence change occurs,which is monitored by the optical instrumentation. A variety of substrates can be em-ployed for optical sensors. Polymeric films can be used as supports to attach indicators.Glass slides can be used both as vehicles for attaching materials to their surface as wellas for coupling light to the detection system. Optical fibers, also called fiber optics, canbe used to carry light both to and from a sensing material attached to its surface, eitherat its tip or surrounding the fiber along its annulus.

8.1.2

Advantages and Disadvantages of Optical Transduction

Optical sensors have a number of advantages over other sensor transduction mechan-isms. As described above, most of the supporting optical instrumentation has beendeveloped for other applications and can be brought to bear on the optical sensingfield. The ready availability of inexpensive instrumentation, and the promise of im-proved performance with new developments in light sources, optics, and detectors,will continue to enable major advances in optical sensing technologies. The continuedmovement toward fully integrated optical communication and computation bodes wellfor the field. In addition to the ready availability of instrumentation, there is a largeknowledge base, as well as commercial accessibility to a multitude of indicators thatare suitable for optical sensing. Optical signals are not susceptible to electromagneticinterferences. Light is fast. Light attenuation is extremely low through modern fiberoptics, which enables remote sensing over long distances with no need for repeaters oramplifiers. Optical measurements, in particular fluorescence, are extremely sensitiveand can be used to detect single molecules. Optical sensing can be readily multiplexedbecause different optical signals can be carried and detected simultaneously. There arealso several disadvantages of optical sensing compared to other sensing methods. Ingeneral, optical instrumentation tends to be more expensive, materials intensive, andmore complex than sensors based on mass or electrical transduction. These latter twomethods employ instrumentation that can be largely designed as integrated circuits,making them simpler and less expensive. In addition, optical methods are sometimes

8 Optical Electronic Noses182

susceptible to interference by stray light. Finally, optical approaches that utilize fluor-escent indicators suffer from eventual photodegradation of the dye molecules.A variety of electronic noses have been developed using a diversity of optical trans-

duction mechanisms [1]. In most cases, these systems employ cross-reactive sensors(discussed below) combined with smart signal processing (described in other chap-ters). Optical sensor arrays have a much shorter history than electronic noses. Con-sequently, there is the hope that these systems will develop rapidly over the next fewyears.

8.2

Optical Vapor Sensing

Given the wonderfully diverse nature of optical signals, the past 25 years have bornewitness to the development of a wide range of light-based chemical vapor sensors.Although the ‘artificial nose’ approach to designing sensing systems was first con-ceived in the early 1980s [2], only in the last few years has this concept been extendedto the optical arena. An increasing number of research groups are now beginning toexplore the utility of employing optical sensors in cross-reactive arrays for improvingsensing capacity and performance. This section provides a general overview of some ofthe key approaches to building optical vapor sensors that have been developed over thepast two decades, and the transition of some of these approaches into ‘optical electro-nic noses’.

8.2.1

Waveguides

Central to many optical chemical sensors is the use of waveguides in one of severaldifferent formats. Fiber optics, capillary tubes, and planar waveguides all exploit thephenomenon of total internal reflection. Optical fibers, for example, are strands ofglass or plastic in which a central ‘core’ is surrounded by a ‘clad’ with a slightly lowerrefractive index. Light introduced into the fiber core is reflected at the clad/core inter-face and is thereby conducted via total internal reflection to the distal tip of the fiber.Hollow capillary tubes or planar substrates comprised of two or more materials withdiffering refractive indices can also be made to guide light extremely efficiently fromone end to the other. A wide range of creative ways to exploit the properties of wave-guides for chemical sensing have been explored.

8.2.2

Luminescent Methods

Fluorescence methods continue to be among the most popular optical sensing andgeneral spectroscopy approaches for a wide range of applications, usually becauseof high quantum yields, well-separated excitation and emission spectra, and intrinsic


sensitivity. For a detailed review of fluorescence spectroscopy the reader is referred toLakowicz [3]. Briefly, fluorophores are molecules that absorb light at one wavelengthand emit light at a longer wavelength. This difference in wavelength, and thus energy,is referred to as the Stoke’s shift and represents vibrational relaxation and other energylosses experienced by the molecule following light absorption. How well a fluorophoreconverts absorbed photons to emitted photons is called its quantum yield or quantumefficiency.Walt [4] and co-workers first combined fiber-optic waveguides with fluorescent dyes

for the measurement of organic vapors in 1991 using the polarity-sensitive, solvato-chromic dye, Nile Red. Following this initial work, the approach was extended to high-er-level arrays of solvatochromic sensors and, finally, to its current configuration ashigh-density microsphere arrays. This work and its evolution are described in moredetail in the final section of this chapter.A number of other groups have also begun to explore fluorescence-based methods

for vapor sensing. Fluorescent dyes can exhibit spectral changes based on several me-chanisms. One such mechanism is the twisted intramolecular charge transfer (TICT)excited state. Molecules such as the one designed and synthesized by Orellana et al. [5],shown in Fig. 8.1, can assume a number of different, highly polar configurations intheir excited state. These excited states will be stabilized when solvated in polar en-vironments such as alcohol vapors and lead to red-shifts in their emission spectra.The degree of these shifts will depend on the particular solvation environment andthus can be used to detect specific vapors. By adsorbing these dyes to silica geland immobilizing the resulting gel at the tip of an optical fiber, Orellana has beenable to demonstrate the reversible measurement of various alcohols.Reichardt’s dye, a betaine fluorophore, is another example of a solvatochromic dye

that exhibits high sensitivity to polarity changes, and has been used to create the ET(30)polarity measurement scale for solvents. An increasing number of groups have begunto incorporate betaine dyes onto the ends of optical fibers in various ways to preparechemical sensors. One group modified the dye molecule and covalently attached it to aMerrifield peptide resin via a five-step synthesis. Following immobilization to a fiber,the resulting sensor was successfully used to measure polar octane improvers in ga-solines [6]. In a similar study, Rose-Pehrrson et al. [7] entrapped Reichardt’s dye withina series of different polymer films and studied the responses resulting from the vary-ing absorption of analytes.A number of groups have begun to explore the potential for exploiting host-guest

supramolecular chemistry for sensing. For example, host compounds that form crys-talline inclusions, or clathrates, by temporarily trapping guest molecules within theirlattice structures have been utilized for detecting solvent vapors [8]. By incorporating a

Fig. 8.1 A polyaromatic-substituted 1,3-oxazole (or 1,3-thiazole)

fluorescent indicator that displays polarity-sensitive TICT excited

states [5]


fluorescent anthracene moiety as well as a few key functional groups to impart selec-tivity for vapors, the authors created a class of compounds they call ‘fluoroclathrands’.When vapors are introduced into a hydrogel layer containing these compounds, thehost molecules surround the guest vapor molecules and form inclusion complexeswith specific crystal structures and characteristic fluorescence behavior. Dependingon the guest molecule, the complexes exhibit both wavelength shifts and quantumefficiency (intensity) changes in their emission spectra. The authors speculate thatthe bathochromic shifts are due to energy losses associated with increased packingdensity in the inclusion compound, while the intensity changes are most likely a resultof self-quenching that varies as a function of the distance between the fluorophores inthe crystal.Unlike fluorophores, which require an excitation source to generate the emission

signals, chemiluminescence-based sensors employ chemically reactive species capableof directly emitting photons following oxidation. This approach offers the advantage ofsimplified instrumentation, by circumventing the need for excitation light sources, aswell as high sensitivity since signals arise from initially dark backgrounds. While che-miluminescence has frequently been employed for oxygen and metal-ion sensors, themethod has recently been extended to detecting organic vapors such as chlorinatedhydrocarbons, hydrazine, and ammonia [9]. The commonly-used reagent luminolwas used to detect oxidants while a Ru(bpy)3

3þ complex was used for reductants. Lu-minol sensing capacity was expanded to halogenated hydrocarbons by the addition ofan inline heated platinum filament used as a pre-oxidative step.

8.2.3

Colorimetric Methods

Sensors that measure changes in absorbance (i.e., color), or local refractive indexchanges resulting from indicator color changes, have also been developed for vaporsensing. Some of the earliest work in this area was done by Wohltjen and colleagues[10], who developed a reversible capillary tube-based sensor for ammonia, hydrazine,and pyridine by coating a glass capillary with an oxazine perchlorate dye film. Colorchanges experienced by the dye upon exposure to these vapors from 60 to 1000 ppmcaused proportional changes in transmission through the tube and were detected by asimple phototransistor. Similarly, Stetter, Maclay and Ballantine [11] used a coating ofbromothymol blue suspended in a Nafion polymer layer to detect and quantify H2Sand HCl acid vapors down to 10 ppb levels.Even commercially available thermal printer papers have been shown to exhibit

reversible interactions with solvent vapors and may be useful in solvent vapor sen-sing. Wolfbeis and colleagues [12] demonstrated that thermal papers could be im-mersed in an ether atmosphere to produce a dark blue or black color. The treatedpaper was found to decolorize to varying extents upon exposure to different polarsolvent vapors. By incorporating these papers into various optical devices and mon-itoring light absorption at 605 nm, sensors were prepared that were capable of mid-ppm to high-ppm detection levels for typical laboratory solvents such as alcohols and


acetates. Response times of these sensors ranged from 30 seconds to 3 minutes, withrecovery times of up to 7 minutes for certain analytes.Polymers are frequently employed in a large number of optical sensor constructs for

their differential vapor sorption or binding properties as well as their emissive proper-ties. For example, the color changes exhibited by amine-containing poly(vinylchloride)membranes when interacting with polynitroaromatics have been used to detect 2,4-dinitrotoluene (DNT), a compound commonly present in landmines [13, 14]. Absorp-tion into the polymer generates a complex with an absorbance at 430 nm that can bemonitored over time to characterize DNT levels in an area of interest.Sensor materials play a central role in all of these various optical approaches, and

their study and development has become a major field of exploration in its own right[15]. All of the above vapor-sensing techniques rely on changes in color of an organicsensing material. Inorganic compounds that exhibit environmental sensitivity in boththeir absorptive and emissive properties are another exciting class of sensing materi-als. At the University of Minnesota, Mann et al. [16] have shown substantial shifts inmaximum absorption and emission wavelengths of platinum and palladium isocya-nide complexes resulting from exposure to volatile organic compounds (VOCs). ThePt-Pt compound [Pt(p-C10H21PhNC)4][Pt(CN)4], for example, was found to exhibit ab-sorption and emission maxima shifts as large as 91 nm and 74 nm, respectively, whenexposed to vapor environments ranging from air to CHCl3. The researchers believethat the incorporation of VOCs into the lattice (which appears to be fully reversi-ble) causes a perturbation in the stacking of the anion and cation complexes that leadsto the observed color changes. In the case of polar VOCs, dipole-dipole and/or H-bond-ing interactions with the Pt(CN)4

2� anion are thought to be involved; for nonpolarcompounds, however, the ‘vapochromism’ is explained by lypophilic interactionswith the isocyanide complexes. Photostability and an insensitivity to water vapormake these materials particularly attractive for incorporation into an opto-electronicnose sensing device.Metalloporphyrins (Fig. 8.2) represent another class of inorganic materials that are

particularly good indicators for sensing as they are stable, well characterized, and easilymodified with a wide range of substituents.

Fig. 8.2 General structure of a metalloporphyrin.

Modifications can occur at each R and R’ position,

and a wide range of metals can be incorporated at the

core of the complex


These compounds can both form coordination complexes with analytes as well asadsorb them via van der Waal’s and H-bonding interactions, giving rise to broad se-lectivity particularly suitable for electronic-nose applications. As a result of their aro-matic p-systems, porphyrins exhibit unique absorption and luminescence propertiesdepending on themetal centers and peripheral substituents involved. D’Amico and co-workers [17] were able to distinguish between six different liquors by monitoring ab-sorbance changes with a simple LED and photodetector system. The researchers rea-soned that the optical changes were caused by competitive interaction of the VOCswith aggregated porphyrin complexes that lead to broadening and shifting of spectralbands.Rakow and Suslick [18] used metalloporphyrins to construct a colorimetric array

detector for vapor-phase ligands. An array was assembled by spotting a series of dif-ferentially metalated porphyrins onto silica thin-layer chromatography plates. Imagingthe array with a common office scanner before and after vapor exposure revealed aunique pattern of response for each of the various analytes (Fig. 8.3). The degreeof spectral shift is thought to be a function of the degree of polarizability of the li-gand. Thus, by incorporating a range of metal centers of varying ligand-binding affi-nity, an array can be made to discriminate between several different analytes. Theauthors report good reversibility as well as linearity of the sensors. A cobalt-basedsensor, for example, responded linearly to binary mixtures of trimethylphosphiteand 2-methylpyridine, and could therefore be used to predict the composition of thesesolvent mixtures. Typically, 15-minute exposures were used with the arrays to ensuremaximum array response, although the authors showed that these times could bereduced to 30 seconds for at least one of the sensors. The work employed hydrophobicsubstrates for the array such as reverse phase silica or Teflon films, which had theadvantage of limiting interference from water vapor (one of the most formidable chal-lenges that plague electronic noses). Colorimetric techniques, such as these porphyrinarrays, generally employ simple instrumentation. Sensor reproducibility with sensi-tivity below the ppm level are presumably among the areas targeted for furtherwork with this approach.

8.2.4

Surface Plasmon Resonance (SPR)

In other work, coordination polymers were used as sensing layers in a SPR setup todetect benzene, ethanol, toluene, acetonitrile, and water [19]. Langmuir-Blodgett filmswere created using poly(CuMBSH) (MBSH þ 5,5’-methylenebis (N-hexadecylsalicyli-deneamine), which were found to be excellent sensing materials due to their rapid andreversible interaction with vapor-phase analyte molecules. The SPR technique exploitsthe delocalized conducting electron clouds found at the surface of metal films such assilver and gold. The electron clouds maintain a collective wave vector parallel to theinterface. Light of a particular wavelength and polarization incident at the interface at aprecise, ‘resonant’ angle will couple to these electromagnetic modes, resulting in asharp decrease in the measured reflected intensity of the excitation beam. The mo-


Fig.

8.3

Scan

ned

images

ofan

arrayof

differen

tmetal-con

taining

porphyrinsbefore

andafterexposureto

variou

schem

icalvapors,as

indicated,for1

5minutes.R

eprintedby

permission

from

Nature[17],

copyright

2000,Macmillan

Magazines

Ltd


mentum matching condition, and thus the resonant angle, is dependent upon therefractive index of the dielectric medium. Therefore any changes in refractive indexat the surface, such as that caused by the sorption of vapor molecules into a polymernetwork at the surface, can be closely measured in real time by monitoring the illu-mination angle needed to give a minimum in the measured reflected light. Alterna-tively, since the resonant angle is also a function of the wavelength of the incident light,a white light source can be used in place of a laser to monitor the wavelength at whichthe surface plasmon resonance occurs [20, 21]. Although the sensitivity was relativelylow in this study, responses to high ppm levels of benzene were demonstrated. TheSPR signals are thought to be directly related to refractive index changes at the surfacedue to swelling of the polymer and/or increased density upon absorption of the analytevapor.In related work, Abdelghani et al. [22] have applied the SPR technique to optical

fibers by coating a 50 nm thick layer of silver onto the core of a silica fiber. To protectagainst oxidation, alkanethiol layers were assembled onto the silver layers. A fluori-nated siloxane was selected to serve as the final cladding layer due to its appropriaterefractive index, surface tension, and gas permeability properties. Although the result-ing sensor responses appear to have improved reproducibility and signal-to-noise ra-tios, the detection limits reported were in the high ppm level for both the aromatic andchlorinated compounds tested, and the cumulative response and recovery times wereof the order of several minutes.

8.2.5

Interference and Reflection-Based Methods

Another area of recent activity for sensor development has been the use of interferencespectroscopy. Having demonstrated that analyte-swelled polymer films experiencemuch larger changes in optical thickness than refractive index [23], Gauglitz andothers have pursued reflectometric interference spectroscopy (RIfS) methods for op-tical vapor sensing. In this approach, light incident at the interface between two planaroptical layers can be reflected from both the top and bottom of a polymer sensing film,setting up an interference pattern that is very sensitive to changes in the optical thick-ness of the polymer layer. Gauglitz suggested that the method offers two primaryadvantages over non-optical techniques: 1) the ability to use strictly inert materials(glass and siloxane polymer films) in contact with the vapor samples; and 2) abuilt-in control for checking the condition of the sensing layer. One of the challengesassociated with measuring changes in the interference spectrum has been the require-ment for relatively bulky and expensive light delivery and detection equipment. Im-provements to this approach have been pursued through simpler and less expensiveoptical components [24]. Recent work using four inexpensive LEDs and a single photo-diode demonstrated that despite the lower-resolution, four-point spectrum, the sim-plified RIfS system yields comparable sensitivity and linearity to its more costly pre-cursor [25].The RIfS technique has also been extended to enantiomer discrimination. By de-

positing polymer solutions containing chiral peptide residues from the ‘Chirasil-


Val’ chromatographic stationary phase, Gopel and colleagues [26] studied the re-sponses of their sensors to several mixtures of (R) and (S)-methyl lactate in varyingproportions. A direct correlation was found: as the concentration of the (S) enantiomerrose in the analyte mixture, the amplitude of the (S)-Octyl-Chirasil-Val sensor rosewhile the (R)-sensor fell.Interference measurements have also been applied to porous silicon chips (PSi).

Sailor and coworkers [27] have developed simple chemical etching methods for gen-erating porous silicon films that display both interferometric and photoluminescenceproperties. In the case of photoluminescence, the group proposed that quenching canbe induced via energy transfer by the adsorption of analyte molecules in the pores ofthe silicon. Thus, by monitoring emission at a specific wavelength (670 nm in thiscase), one can observe sharp decreases in intensity as the interaction with analytevapors takes place. Likewise, adsorption events give rise to refractive index changesthat lead to shifts in Fabry-Perot interference fringes, measured as changes in reflec-tivity. Both of these optical attributes were recently used to measure a range of per-fumes and solvent vapors. When compared side-by-side to a commercial electronicnose containing metal-oxide sensors, the PSi chips displayed comparable discrimina-tion ability for a few standard solvents, ethyl esters, and perfumes. At the saturatedvapor conditions used, the silicon sensors showed significantly faster recovery timesthan their metal-oxide counterparts (30 s versus 15 min). The ability to create a diversearray with high sensitivity and broad selectivity with this approach, however, remainsto be proven.Another absorbance type of vapor sensor is based on simple transmission attenua-

tion through a fiber. Microbending caused by the vapor-induced swelling of siloxanelayers adjacent to the fiber results in transmission attenuation [28].Yet another creative reflection-based approach to chemical sensing has been the use

of resonating microcantilevers such as those used in atomic force microscopy (AFM)for atomic-level imaging. Based on the mass-sensing concepts of resonating piezoelec-tric crystals (e.g., quartz crystal microbalances), the approach uses 180 lm long can-tilevers micromachined into silicon that are sensitive to changes in mass occurring attheir surfaces. Several groups have explored coating polymer films onto these canti-levers and measuring small changes in mass loading. The technique uses optical de-tection by measuring the deflection of an incident laser beam as analyte vapors areadsorbed to the surface. In one study, Thundat et al. [29] showed that such sensorscould be modified to possess desired selectivities, for example by employing hygro-scopic coatings to improve sensitivity to water vapor.A group in Switzerland recently proposed that arrays of differentially coated canti-

levers could be used as a new form of chemical nose [30]. Working in their own mi-crofabrication facility, the group constructed an eight-cantilever sensor array fromsilicon. The individual cantilever coatings included platinum thin films, alkythiolself-assembled monolayers (SAMs), zeolites, and poly(methylmethacrylate). Theauthors studied detection of water vapor, alcohols, and several natural flavors.Although the array was read out sequentially due to the use of a single laser andphoto-sensitive device, one can envision ways of multiplexing through beam-splittersand larger, higher-resolution two-dimensional detector arrays. Detection limits were


not calculated in this study, making it difficult to compare the sensitivity of the ap-proach to other methods. In addition, despite the small size of the devices, reportedcycle times were of the order of several minutes. Other challenges with the cantileverapproach include interference from pressure changes during sampling, loss of signaldue to severe bending of the cantilever, laser heating of the cantilever, and limiteddynamic range [29]. Nevertheless, as they continue to be developed and improved,cantilever arrays may prove to be a promising opto-electronic nose format capableof simple integration into silicon-based microelectronic devices.

8.2.6

Scanning Light-Pulse Technique

Lundstrom and coworkers have taken an innovative optical approach by employing amethod called the scanning light-pulse technique [31–34]. In this approach, light im-pinges on the surface of a metal-oxide semiconductor field effect transistor (MOSFET)coated with a thin metal film and penetrates the metal to induce a photocapacitivecurrent. To maintain a constant current, the applied gate voltage (V) must be variedto sustain a constant surface potential. Changes in the gate voltage are monitored andresult in a map of the change in voltage (DV ) as a function of position on the sensingsurface. In one demonstration, a MOSFET array was prepared with three continuousstrips of Pt, Pd, and Ir. The sensor surface was divided into a grid, and a temperaturegradient (110–180 8C) was established down the length of the sensor surface. Thistemperature gradient provided a different sensitivity and selectivity at each point ofthe sensor grid. The sensor grid was exposed to hydrogen, ammonia, and ethanol,and DV was determined. In this manner, image maps of the gases were created.These sensor grids can be applied to identifying gas mixtures, rapid and simultaneousscreening of new sensing materials, and mapping spatially inhomogeneous reactions.Light-pulsed sensing combines many types of information, including the catalyticactivity of the gate metals, gas flow turbulence, edge effects, etc. While not an opticaldetection technique, the method demonstrates the utility of employing light combinedwith electrochemical detection.

8.3

The Tufts Artificial Nose

Optical fibers can be used to create fluorescent-based optical sensors. In this approach,a fluorescent indicating species is attached to the fiber’s distal tip using a variety ofimmobilization techniques. Excitation light is introduced into the fiber, which carrieslight efficiently to the fiber’s distal tip. The fluorescent indicator is excited and some ofthe resulting isotropically emitted light is captured by the same fiber, directed throughsuitable optics, filtered and sent to a detector. The modulated light signal returning tothe detector corresponds to the presence and amount of an analyte.In order to design a cross-reactive optical sensing array, it is necessary to find an

appropriate array of sensing materials to respond to a wide variety of analytes. Our


laboratory has developed a series of fluorescence-based optical sensors. It was our goalto create a fluorescent-based cross-reactive array. In 1991, we published a paper inwhich we used a solvatochromic indicator, Nile Red, to create a generic optical vaporsensor [4]. The sensor was based on immobilizing Nile Red within a polymer matrixand attaching the resulting material to the distal tip of an optical fiber. As discussedabove, solvatochromic indicators report on the polarity of their local environment, alsocalled the microenvironment. When solvatochromic dyes, such as Nile Red, are em-bedded in polymers, they report on the polarity of the polymer’s microenvironmentindicated by their color, in particular, their absorption and/or emission spectra. Forexample, Nile Red has an emission spectrum that is relatively blue in nonpolar, hydro-phobic environments, and is red in polar, hydrophilic environments. When an organicvapor sensor containing Nile Red, immobilized within a polymer, is in contact with air,it has an emission spectrum that represents the polarity of the polymer. When such apolymer is exposed to an organic vapor, the organic vapor diffuses into the polymerand modifies the microenvironmental polarity, which is signaled by a change in theemission spectrum of Nile Red. The emission spectrum shift is highly predictable. Avapor that is more polar than the polymer will shift the spectrum to a higher wave-length, whereas a less polar polymer will shift the spectrum to a lower wavelength(Fig. 8.4). The extent of the shift depends both on the polarity difference as well asthe partition coefficient of the vapor into the polymer. In this manner, a generic or-ganic vapor sensor was created by simply immobilizing a single solvatochromic dyewithin a dimethylsiloxane polymer. The sensor was used to detect leaks of hydrocarbonliquids from underground storage tanks by detecting the vapors that preceded theliquid leak.The same sensing principle was used to design a cross-reactive vapor-sensing array

[35–37]. In this system, Nile Red was immobilized within a series of polymers. Hun-dreds of polymers were screened empirically. Each polymer defined the initial polarity

Fig. 8.4 The spectra of four sensors made by incorporating Nile Red

into four polymers of differing polarity. The emission max shifts to the

red with increasing polarity of the polymer matrix


of the microenvironment as reported by Nile Red. These polymers were dip-coatedonto the ends of individual optical fibers. Nineteen sensors were bundled into an arrayformat. Upon exposure to an organic vapor, each polymer sensor absorbed vapor ac-cording to its partition coefficient for that vapor. The change in each sensor’s fluor-escence spectrum depended on howmuch vapor partitioned into that sensor as well asthe difference between the vapor’s and polymer’s polarities.There are several aspects of the optical sensor array’s operating mechanisms that

require elaboration. First, we decided that, unlike most electronic noses, we would notlook at static headspace measurements but rather would mimic a sniff by observingthe kinetics of the response upon vapor exposure. To this end, we employed a vapordelivery system that was originally designed for delivering odors to animals in olfactoryresearch [38]. The vapor delivery was accomplished by presenting square-wave vaporpulses for a defined period of time to the distal face of the bundled fiber array. Fluor-escence detection was accomplished by using a two-dimensional detector, a CCD cam-era, so that we could acquire fluorescent signals from all the sensors in the array si-multaneously. To simplify signal detection, the fluorescence was collected at a singlewavelength by interposing an emission filter between the fiber and the CCD chip. Theresulting measured fluorescence signals coming from each sensor, upon exposure toorganic vapors, were simply the intensity changes relative to their starting intensity atthat particular emission wavelength. An intensity increase simplymeant that the emis-sion spectrum of the dye in a particular polymer upon exposure to a particular vaporwas shifting closer to the wavelength range defined by the emission filter. Conversely,a decrease in fluorescence intensity indicated that the emission spectrum of the dyewas moving further away from the emission filter range. A final aspect of the responsemechanism resulted from the interaction of the vapor with the polymer. Some of thepolymers exhibited a swelling effect in which the polymer volume increased as vaporpartitioned into it. Polymer swelling causes a dye molecule to increase its averagedistance relative to the fiber surface. As described above, the isotropically emitted lightis captured by the optical fiber. When amolecule moves further from the fiber surface,the capture efficiency for the light decreases because the sine of the half angle of thereturning light is reduced. Therefore, the response of each sensor is due to a combi-nation of vapor partitioning into the polymer, polarity differences between the polymerand the vapor, and polymer swelling. Because the solvatochromic and swelling effectsoperate under different kinetic regimes (i.e., swelling at the bulk polymer surfaceoccurs rapidly while the solvatochromicity requires an intimate slower redistributionof vapor molecules within the polymer matrix), nonlinear effects can be observed. Thefluorescence images are collected before, during, and after a vapor pulse to provide acharacteristic response profile for each sensor in the array.A video image of an array of 19 sensors exposed to a three second pulse of benzene is

shown in Fig. 8.5. The digitized responses of each sensor in the array are shown in thegraph in Fig. 8.6. These complex temporal responses are characteristic of a benzenepulse at a particular concentration and can be used to train a computational classifica-tion program. Both parametric (e.g., intensities, slopes) and nonparametric methodscan be used to train the responses. One of the major challenges in the field of electro-nic noses/cross-reactive arrays is array-to-array variability. This lack of reproducibility


results from the inability to prepare polymeric materials identically. When polymersare put onto optical substrates or other surfaces by dip coating, liquid dispensing,photopolymerization, or electropolymerization, slight volume differences, initiatorconditions, or minor heterogeneities can cause significant differences in materialcomposition. These differences, even if the variation is only a few percent, canlead to loss in training fidelity. To address this problem, we have switched to a dif-ferent array platform. Instead of using individual single-core optical fibers we nowemploy optical-imaging fiber arrays. These arrays are comprised of thousands of in-dividual optical fibers, each of which is surrounded by a clad material (Fig. 8.7). Thearrays are fabricated such that they are coherent in nature meaning that the position ofan individual optical fiber within the array retains its position from one end to theother. In this manner, such arrays can be used to carry images, an application thatis being pursued for medical endoscopy. These arrays are fused unitary bundlesrather than mechanically fixed strands of individual fibers. Thus, they maintain theirflexibility and can be handled similarly to single core fibers. A typical optical arraycontains between 10 000 and 50 000 individual fibers in a diameter of a few hundredmicrons with the individual fibers having diameters on the order of 3–5microns each.The difference in materials composition between clads and cores provides a method

for selectively etching the cores. When the polished distal tip of a custom optical ima-ging fiber array is placed into an acid etchant, the cores etch at a faster rate than theclads leading to an array of wells. At the bottom of each well is the distal face of anoptical fiber (Fig. 8.8A). In this manner, each well is ‘optically wired’ to its own indi-vidual optical fiber. We discovered that latex or silica beads, matched in size to thedimensions of the individual wells, would spontaneously assemble into each well

Fig. 8.5 A sequence of images depicting the fluorescence response of a 19-fiber sensor array to a pulse of

benzene vapor


in a highly efficient self-organizing fashion. This approach could be used to createsensor arrays based on polymeric microspheres.Microsphere sensors can be created by taking monodisperse polymeric micro-

spheres and swelling them in a suitable organic solvent containing dissolved NileRed [39]. Upon removal from the solvent, evaporation of residual solvent occurs re-sulting in Nile Red being trapped within the polymeric matrix. Another class of beadsensors uses surface modified silica beads to which Nile Red is adsorbed (Fig. 8.9).Many different bead types can be prepared out of a variety of polymers and surface

Fig. 8.6 Temporal plots from 19-fiber array response to benzene vapor pulse

Fig. 8.7 Components of a fiber-optic imaging bundle


functional groups. As discussed above, in each of these sensors, the Nile Red reportson the polarity of its local environment. A library of bead types is created containing adiversity of responses to vapors. To create a sensing array, the desired individual beadtypes are mixed. 100 milligrams of beads contains approximately 10 billion beads. Thebeads are randomly distributed onto the distal face of an etched imaging fiber such thatone bead occupies each well (Fig. 8.8B). In order to register the position of each bead inthe array after fabrication, the fiber is connected to the optical imaging system and avapor is pulsed onto the fiber’s sensor end. Because each different type of bead pro-duces a unique and characteristic response profile when exposed to a particular vapor,the responses to the vapor pulse enable the image-processing program to register thebead type occupying each well. We refer to this registration protocol as ‘self-encoding’;that is, the sensor is identified by its response profile to a particular vapor [36]. In thismanner, a library of beads can be used to create hundreds to thousands of individualsensing arrays with each array having the same bead types but located in differentpositions. The bead registration task involves exposing each array to a particular re-

Fig. 8.8 A) wells formed by etching an imaging bundle, and B) beads immobilized in the wells

Fig. 8.9 Silica beads can be modified in a variety of ways before being dyed in order to generate a diverse

library of sensors


gistration vapor and using an image-processing program to automatically register theposition of each bead in the array using a lookup scheme.A key advantage of the self-encoding array sensors is that the training can be trans-

ferred from one sensor array to another. All the sensor beads of a particular type givevirtually identical responses because they are all prepared at the same time. Thus,when mixed in a library, each bead type maintains its particular response profile. An-other important feature of these cross-reactive optical arrays is the built-in redundancyof each of the sensors. The small size of the fibers combined with the random dis-tribution of the different microspheres in the array dictates that there will be replicatesof each sensor in every array. The numbers of each sensor type will distribute them-selves according to Poisson statistics. Replicates provide significant advantages interms of signal-to-noise. The signal-to-noise ratio scales as 1/5 n, where n is equalto the number of sensors of each type. By summing or averaging sensor repli-cates, significant signal-to-noise enhancements can be achieved resulting in improveddetection limits due to the ability to make more precise measurements at lower con-centrations (Fig. 8.10) [36]. The microsphere arrays also have several other advantagessuch as flexibility of array types, scalability, and simple manufacturing.The major limitation with fluorescent dyes for optical sensor arrays is photobleach-

ing. Upon exposure to light, any indicating material loses its intensity because ofphotooxidation. Over long periods of exposure, the light intensity degrades consider-ably. In order to avoid this problem, we employ autoscaled response profiles so thattraining is not dependent on absolute signal intensities. Despite this autoscaling pro-cedure, photobleaching eventually degrades the signal-to-noise ratios. At this point,the array must be replaced. Since each array has the identical sensing elements,the training performed on one array is transferable to a second array. We have recentlydemonstrated training transfer of a classifier over a nine-month period with robust-ness of classification.

Fig. 8.10 Signal-to-noise

ratios can be dramatically

improved by averaging over

multiple copies of the same

bead type within an array


8.4

Conclusion

Optical electronic noses have a relatively short history relative to conducting polymer ormetal-oxide-based approaches. In the roughly five years since they were first reported,there have been a variety of advances in the types of optical signals employed as wellas the materials used to perform the recognition [18]. The area of molecular recognitionis burgeoning. Many of these receptors have built-in optical transduction. New polymers[14] and nanostructured materials [27] with recognition and optical signaling are beingdeveloped. In addition, the data richness of optical sensor arrays should make them at-tractive as analytical systems. With continued emphasis on new optical materials anddevices development for the telecommunications and computer industries, combinedwith advances in molecular recognition and advanced materials, optical approachesto sensing should continue to improve in sensitivity, selectivity, and performance.

Acknowledgments

The authors wish to thank the ONR and DARPA for research funding, and KeithAlbert and Shannon Stitzel for assistance with figures.

Tab. 8.1 Summary table of optical electronic nose approaches.

Transduction

Mechanism

Description References

Luminescence Fiber-optic sensors using polarity sensitive fluorophores such

as solvatochromic or TICT dyes.

4–7, 35–37

Randomly assembled solvatochromic bead arrays. 39, 40

Host-guest supramolecular chemistry: shifts in wavelength and

intensity of ‘fluoroclathrands’ based on packing density changes

caused by vapors.

8

Chemiluminescence-based detection, using luminol and Rubpy

dyes.

9

Colorimetric Color changes of an oxazine perchlorate dye coated on glass

capillaries.

10

Bromothymol blue in Nafion polymer layers. 11

Thermal printer paper as vapor sensors. 12

Inorganic sensing materials (e. g. Pt-Pt compounds): color

changes caused by perturbation of stacking in charged complexes.

16

Metalloporphyrins: formation of coordination complexes with

analytes, and use of different metals for changing sensing

properties.

17, 18

Surface plasmon

resonance

Method for detecting changes in refractive index at a surface. 20–22

Interference,

reflection

Reflective interferometric ipectroscopy for detecting changes

in optical thickness of polymer layers.

23–26

Interference measurements using chemically etched porous

silicon chips.

27

Mass loading Detecting mass changes on resonating atomic force microscope

microcantilevers.

29, 30


References

1 K. J. Albert, N. S. Lewis, C. L. Schauer,G. A Sotzing, S. E. Stitzel, T. P. Vaid,D. R. Walt. Chem. Rev., 2000, 100,2595–2626.

2 K. Persaud, G. Dodd. Nature, 1982, 299,352–355.

3 J. Lakowicz. Principles of Fluorescence Spec-troscopy, Plenum Press, New York, 1982.

4 S. Barnard, D. R. Walt. Environ. Sci. Technol,1991, 25, 1301–1304.

5 G. Orellana et al.. Anal. Chem., 1995, 67,2231–2238.

6 C. Hubert, D. Fichou, F. Garnier. Adv.Mater., 1995, 11, 914–917.

7 H. Krech, S. L. Rose-Pehrsson. Anal. Chim.Acta., 1997, 341, 53–62.

8 T. H. Brehmer, P. P. Korkas, E. Weber. Sens.Actuators B, 1997, 44, 595–600.

9 G. E. Collins, S. L. Rose-Pehrsson. Sens.Actuators B, 1996, 34, 317–322.

10 Giuliani et al.. Reversible Optical WaveguideVapor Sensor, US4513087. April 23, 1985.

11 J. R. Stetter, J. Maclay, D. S. Ballantine.Optical Waveguide Vapor Sensor, US5315673.May 24, 1994.

12 H. E. Posch, O. S. Wolfbeis, J. Pusterhofer.Talanta, 1988, 35, 89–94.

13 L. Gheorghiu, W. R. Seitz, D. Arbuthnot,J. L. Elkind. SPIE Conference on Environ-mental Monitoring and Remediation Techno-logies II, 1999, 3853, 296–302.

14 D. T. McQuade, A. E. Pullen, T. M. Swager.Chem. Rev., 2000, 7, 2537–2574.

15 F. L. Dickert, A. Haunschild, Adv. Mater.,1993, 12, 887–895.

16 C. A. Daws, C. L. Exstrom, J. R. Sowa, Jr.,K. R. Mann. Chem. Mater., 1997, 9,363–368.

17 A. D’Amico et al.. Sens. Actuators B, 1999, 65,209–215.

18 N. A. Rakow, K. S. Suslick. Nature, 2000,406, 710–713.

19 R. Casalini et al.. Sens. Actuators B, 1999, 57,28–34.

20 R. W. Nelson, J. R. Krone, O. Jansson. Anal.Chem., 1997, 69, 4369–4374.

21 BIAcore Probe literature, Pharmacia Bio-sensor. http://www.biacore.com.

22 A. Abdelghani et al..Anal. Chim. Acta., 1997,337, 225–232.

23 K. Spaeth, G. Kraus, G. Gauglitz. Fresenius’J.Anal. Chem., 1997, 357, 292.

24 Y. Liu et al.. Optical Sensor Apparatusfor Detecting Vapor of Organic Solvent,EU95203669.7. 1995.

25 R. Reichl, R. Krage, C. Krummel,G. Gauglitz. Appl. Spectrosc., 2000, 54,583–586.

26 K. Bodenhofer et al.. Nature, 1997, 387,577–580.

27 S. Letant, S. Content, T. Tan, F. Zenhausern,M. Sailor. Sens. Actuators B, 2000, 69,193–198.

28 A. Yasser, B. Lawrence. Sensors, 1996, April,76–77.

29 T. Thundat et al.. Anal. Chem., 1995, 67,519–521.

30 H. P. Lang, et al.. Appl. Phys. Lett, 1998, 72,383–385.

31 M. Lofdahl, M. Eriksson, I. Lundstrom. Sens.Actuators B, 2000, 70, 77–82.

32 F. Winquist, H. Sundgren, E. Hedborg,A. Spetz, I. Lundstrom. Sens. Actuators B,1992, B6, 157–168.

33 I. Lundstrom, R. Erlandsson, U. Frykman,E. Hedborg, A. Spetz, H. Sundgren,S. Welin, F. Winquist. Nature, 1991, 352, 47.

34 I. Lundstrom, H. Sundgren, F. Winquist.J. Appl. Phys., 1993, 74, 6953–6962.

35 J. White, J. S. Kauer, T. A. Dickinson,D. R. Walt. Anal. Chem., 1996, 68,2191–2202.

36 T. A. Dickinson, D. R. Walt, J. White,J. S. Kauer. Anal. Chem., 1997, 69,3413–3418.

37 T. A. Dickinson, J. S. White, J. S. Kauer,D. R. Walt. Nature, 1996, 382, 697–700.

38 J. Kauer, G. Shepherd. J. Physiol., 1977, 272,495–516.

39 T. A. Dickinson, K. L Michael, J. S. Kauer,D. R. Walt. Anal. Chem., 1999, 71,2192–2198.

40 K. J. Albert, D. R. Walt. Anal. Chem., 2000,72, 1947–1955.

8.4 Conclusion 199199

9

Hand-held and Palm-Top Chemical Microsensor Systems

for Gas Analysis

A. Hierlemann, U. Weimar, and H. Baltes

Abstract

The characteristics and fundamentals of hand-held chemical sensor units for gas ana-lysis are described, commercially available systems based on conventional sensor tech-nology are briefly portrayed, and the emerging field of microsensors and microsensorsystems based on planar integrated circuit (IC) technology and their use in hand-heldinstruments is detailed.Conventional sensor technology is at the base of most hand-held instruments in

research and on the market to date. Systems based on mass-sensitive sensors andon electrochemical sensors (chemoresistors) are presented. They are used to detectorganic volatiles and rely on changes of physical properties of polymeric layersupon volatile absorption. The same polymers can be used with microsensors basedon silicon or IC technology. These microsensors offer substantial advantages suchas low power consumption, a very crucial issue in battery-operated systems, smallsize, rapid response, and batch fabrication at industrial standards and low costs.The present state of the art in IC-based microsensors is summarized and the inclu-sion of such sensors into hand-held systems is shown.

9.1

Introduction

The first hand-held systems, which are still available on the market [1, 2], were tubes orbadges. They are lightweight, inexpensive, disposable devices based on diffusion ex-posure. They provide an immediate visual indication when a specific chemical hazardis present. They mostly include an indicator layer or impregnated paper, which pro-vides homogeneous and stable color formation or color change upon presence of thetarget compound. These devices are not continuously operating, exhibit irreversiblecharacteristics, are disposable, and, therefore, are usually referred to as dosimeters[3, 4] rather than as chemical sensors.At present, there are two different categories of sensor-based (i.e., non-disposable),

continuously operating hand-held instruments. The first category includes personal


201201

warning and safety systems as an advancement of the already mentioned tubes andbadges, which have been on the market for quite some time [5], second there arerecently developedmultisensor-systems with onboard pattern recognition and/ormul-ticomponent analysis algorithms sometimes denoted as “electronic noses” [6]. Thedistinction between those two types of devices is mainly due to differences in theinstrument architecture or complexity, and in the target applications.

Key requirements for both types of hand-held instruments include:

* Ease of use* Ruggedness* Low power consumption* Low cost and low maintenance* Short recovery and response times* Long-term stability (low drift) and reliability (self-calibration)

For the personal safety devices of the first category, it is desirable that the system alsoexhibits

* High sensitivity and low limit of detection (LOD)* High selectivity to target analyte and low cross-sensitivity to interferants

The hand-held personal safety devices include in most cases only one or two sensorsspecifically engineered to detect selected individual gaseous compounds at trace level[1, 5]. Upon reaching a threshold value, the devices issue a warning or an alarm. Thedevice calibration is univariate, i.e., the devices are calibrated using pure gases. Theirapplications include the detection of toxic or explosive gases in all branches of industry,the measurement of hazardous substances during firefighting operations, and thedetection of airborne contaminants such as carbon monoxide, hydrazine, ammo-

Fig. 9.1 Typical hand-held gas warning system (PAC III by Draeger,

L€uubeck, Germany [5]) detecting carbon monoxide (CO). By

exchanging the sensor, hydrogen sulfide or oxygen can be detected.

Reprinted with kind permission of Draeger

9 Hand-held and Palm-Top Chemical Microsensor Systems for Gas Analysis202

nia, hydrogen sulfide or hydrides. Figure 9.1 displays a carbon monoxide monitorbased on a single electrochemical cell [5]. Electrochemical cells are predominantlyused since their sensitivity to anything other than the desired compounds are inmost cases negligible.The second category of hand-held instruments includes a sensor array with different

coatings on the same type of transducer or even different types of transducers. Thetarget compounds are individual gases or a multitude of gaseous and volatile com-pounds generating a characteristic fingerprint response pattern. Pattern recognitionand multi-component analysis algorithms rely on multivariate calibration. Trainingand gas phase analysis are restricted to a defined sample set, which has to be cali-brated prior to instrument use. Such multisensor systems are the main topic ofthis chapter and book. Target applications of these hand-held instruments includequality and process control in industrial settings (food processing, packaging, rawmaterial inspection), aroma and odor identification, environmental monitoring, hazar-dous material identification, and some medical pilot applications.Conventional sensor technology is at the base of most hand-held instruments in

research and on the market to date. Recently, microsensors based on silicon or inte-grated circuit (IC) technology have been developed [7–11], which offer substantialadvantages such as low power consumption, a very crucial issue in battery-operatedsystems, small size, rapid response, and batch fabrication at industrial standards en-suring a high level of sensor-to-sensor reproducibility, quality, and inferring low costs.Additional features include the possibility of on-chip signal conditioning or data pre-processing [12, 13].In the following, we will describe the characteristics and fundamentals of hand-held

instruments, then detail the approach using conventional sensor technology by brieflyportraying commercially available systems, and finally we will describe the emergingfield of hand-held instruments relying on microsensors and microsensor systemsbased on planar IC technology.

9.2

Conventional Hand-held Systems

9.2.1

Hardware Setup

A schematic of a hand-held instrument comprising all vital components used in com-mercial and research-type instruments is shown in Fig. 9.2. The hand-held instrumentconsists of two major blocks. The upper part represents the gas intake unit withpumps, valves, filter, and the measurement chamber. The bottom “electronic” partincludes the sensors, sensor electronics, power pack, the data processing unit withdisplay, and the communication interface.In some cases it is sufficient to rely on diffusion of the analyte molecules to the

sensors. This “passive” sampling does not require any of the components in the

9.2. Conventional Hand-held Systems 203203

gas intake unit, only an opening of the sensor chamber to the ambient allowing for fastin-diffusion of the analyte.For “active” sampling, pumps, valves, and filters are required. An active sampling

stage is realized in all commercially available systems since the gas phase compositionin the sensor chamber is less subject to fluctuations and can bemuch better controlled.Different operation modes for an active sampling unit have been implemented:

* Pumping only: test gas is pumped into the measurement chamber and pumped outthrough an exhaust (flow-through), or through the inlet by reversing the pumpdirection.

* Pumping and valving: test gas is pumped to the sensors either from separate inletsfor reference gas and analyte gas or by routing a fraction of the analyte gas throughan on-board filter unit.

Pumping and valving require a more sophisticated intake unit design but offers theadvantage of re-establishing the baseline of the sensors using a filtered or pure purgeor reference gas. This leads to better recognition of drift effects and sensor malfunc-tioning.The bottom part of the schematic represents the electronic part of the instrument.

The sensor array is mounted in a measurement chamber and connected to a micro-controller. The AD/DA and I/O channels of the microcontroller can be directly used,or dedicated electronic components can be added. The sensor-microcontroller connec-tion depends on the number of channels and the targeted sensor resolution. Novelresearch-type instruments exhibit digital communication between sensor array and

Fig. 9.2 Typical setup of a research-type or commercial hand-held

instrument. The upper part represents the gas intake unit with pumps,

valves, filter, and the measurement chamber. The bottom electronic

part includes the sensors, sensor electronics, power pack, the data

processing unit with display, and the communication interface


microcontroller, which requires additional electronics (A/D conversion, bus interface)on the sensor side. The microcontroller usually hosts pattern recognition algorithms(KNN, PCA, see Chapter 6), allows for storing calibration data or analyte pattern li-braries, and enables logging a limited amount of acquired measurement data.Batteries or accumulators usually power the hand-held instrument. The capacity of a

typical battery (or accumulator) is of the order of 5 to 15 Wh. Therefore, the averagepower needed to operate the instrument should be below 1 W to ensure a decent op-eration time.Different types of displays are used to communicate the desired information to the

user. The simplest display can be realized by a red and green LED indicating a binarydecision, e.g., the membership to a class such as “sample o.k.” or “not o.k.”. Moreinformation is provided by alphanumeric displays (one or several lines) presentingqualitative (e.g. classification) and quantitative (e.g. analyte concentration) results. Gra-phic displays are very versatile and comfortable and can show, for example, completePCA plots.All currently available systems are also equipped with a computer interface to con-

nect to an external PC for downloading measurement data to the computer and trans-ferring, for example, calibration data to the hand-held instrument. The most commoninterfaces are RS-232 (serial) and, more recently, universal serial bus (USB). In thenear future infrared interfaces and BluetoothTM systems enabling wireless communi-cation will be introduced.

9.2.2

Fundamentals of the Sensing Process

All commercially available hand-held units, and the CMOS chemical microsensors(detailed in Section 9.2) rely on polymeric coatings as sensitive films for the detectionof volatile compounds in air. The predominant sensing mechanism is hence physi-sorption and bulk dissolution of the analyte molecules within the polymer volume.Upon absorption of analyte by the coating, the physical properties of the polymerfilm, such as its mass, volume or dielectric constant, change. Considering bulk dis-solution in polymers, all effects are based on thermodynamics and/or kinetics. Highsensor selectivity (strong interaction) and perfect reversibility (weak interaction) im-pose conflicting constraints on the design of the sensitive layer. For ensuring rever-sibility, polymers showing partial selectivity to some of the detected species are com-monly used. The desired identification of the compounds is then achieved by using anarray of different partially selective sensors and applying numerical methods of dataevaluation (see Chapter 6) [14–18].At “infinite dilution”, that is an analyte partial pressure below 3% of its saturation

vapor pressure at the sensor operation temperature, one usually observes a linear cor-relation between the analyte concentrations and the sensor signals. In this low-con-centration range, Henry’s law still holds [19]. Therefore, it is possible to calculate parti-tion coefficients at “infinite dilution”, K, as characteristic thermodynamic equilibriumconstants for a certain organic volatile dissolved in a polymer:


K ¼ cpoly=cA ð9:1Þ

Here, cpoly and cA denote the analyte concentration in the polymer and the gas phase.The partition coefficient, K, is hence a dimensionless “enrichment factor” relating theconcentration of a species in the sensing layer to that in the probed gas phase. K isusually of the order of 200 to 5000 for most analytes sorbed into standard polymers.Ideally, an experimentally determined partition coefficient should be constant for thechosen analyte/polymer combination, and independent of the transducer principle[20].

9.2.3.

Commercially Available Instruments Based on Conventional Technology

In this section, a brief overview will be given on commercially available hand-heldinstruments, which are all based on conventional sensor technology (see Chap-ter 4). These hand-held systems weigh between 0.5 and 1 kg and are battery or accu-mulator operated. All hand-held units feature basic pattern recognition software (PCA,KNN, etc., for details, see Chapter 6) and have some on-board data storage possibilityas well as a RS 232 serial interface to communicate with external equipment such aslaptops or computers. The devices are specified to operate in a temperature rangebetween 263 and 323 K. The devices feature a LCD display (some include even a gra-phic display) to show the result of the sensor analysis or pattern recognition. Thekeyboard is in all cases very simple and allows for using the hand-held device withonly a few commands. All instrument producers are providing additional data loggingand storage software for external PCs. The individual configuration of this softwaredepends on the user needs. The gas bus and sensor array are different for the varioussystems and will be described in more detail below. Fig. 9.3 shows three of the cur-rently commercially available systems, the VOCcheck� by AppliedSensor [21], theVaporLab� by Microsensor Systems [22], and the Cyranose 320� by Cyrano Sciences[23]. A summary of their characteristic features is given in Table 9.1.

Fig. 9.3 Currently commercial-

ly available systems: the VOC-

check� by AppliedSensor [21],

the VaporLab� by Microsensor

Systems [22], and the Cyranose

320� by Cyrano Sciences [23].

Reprinted with kind permission

of AppliedSensor, Microsensor

Systems, and Cyrano Sciences


Since there are severe constraints on overall size, weight and power consumption,the performance of hand-held devices is generally inferior to that of bench-top instru-ments. Precision and the LOD are degraded by a factor of approximately 10 as com-pared to a bench-top setup due to the less effective temperature stabilization and due tothe use of less bulky measurement and recording electronics such as no precisionvoltage sources, less-sophisticated counter modules, and off-the-shelf electronicparts. The hand-held units are less pricy and intentionally designed to suit only afew specific applications, in consequence their versatility is rather limited. Theunit configuration usually has to be optimized with regard to the target application.In the following we will briefly describe the characteristics of the commercially avail-

able systems as well as the underlying transducer principle, formore details we refer toChapter 19, where some more information on these systems can be found.

9.2.3.1 Hand-held Units Based on Mass-Sensitive Sensors

Mass-sensitive sensors are in the simplest case gravimetric sensors responding to themass of species accumulated in a sensing layer. Some of the sensor devices are ad-ditionally capable of detecting changes in a variety of other properties of solid or liquidmedia in contact with their surface such as polymer moduli, liquid density and visc-osity [24–26], which will not be discussed here. The high sensitivity of mass-sensitivesensors provides good chemical sensitivity: mass changes in the picogram range orlower can be detected and ppm (parts per million) to ppb (parts per billion) detectionlevels have been reported, for example, for gas and vapor sensors [24, 26]. Most of themass-sensitive sensors rely on piezoelectric materials such as quartz, lithium tantalateor niobate, aluminum nitride, zinc oxide and others. Piezoelectricity results in general

Tab. 9.1 Characteristic features of three commercially available hand-

held units: VOCcheck� by AppliedSensor [21], VaporLab� by Micro-

sensor Systems [22], and Cyranose 320� by Cyrano Sciences [23].

Features VOCcheck� [21] VaporLab� [22] Cyranose 320� [23]

Number of Sensors 4 4–6 32

Transducer Type Thickness Shear Mode

Resonator

Surface Acoustic Wave

Device

Chemoresistor Array

Sensitive Layer Polymer Polymer Carbon-Loaded Polymer

Target Analytes Organic Volatiles Organic Volatiles Organic Volatiles

Response Time < 15 s < 1 s 10 s

Operating Temperature 10 8C–40 8C 5 8C–40 8C 0 8C–40 8CWeight 400 g 570 g 910 g

Dimensions 180 � 82 � 53 mm 180 � 85 � 56 mm 220 � 100 � 50 mm

On-board Software Application-Specific Not Specified KNN, Kmeans, PCA, CDA

Sampling Stage Pump Pumps and Valves Pump and Valves

Display 12 � 4 Alphanumeric 119 � 73 Pixel LCD

Backlight

320 � 200 Graphic

Backlight

Battery Life 6–12 Hours 4–12 Hours 3 Hours

Data Log Capacity 100 Measurements 100 Measurements 100 Measurements

Interface RS 232 RS 232 RS 232, USB


from coupling of electrical and mechanical effects. The prerequisite is an anisotropic,noncentrosymmetric crystal lattice. Upon mechanical stress, charged particles are dis-placed thus generating a measurable electric charge in the crystal. In turn, mechanicaldeformations can be achieved by applying a voltage to such a crystal (for details, see[27]). Using an alternating current (AC), the crystals can be electrically excited into afundamental mechanical resonance mode. The resonance frequency, which is therecorded sensor output in most cases, changes in proportion to the mass loadingon the crystal or device. The more mass (analyte molecules) that is absorbed, for ex-ample in a polymer coated onto a piezoelectric substrate or transducer, the lower is theresonance frequency of the device [28, 29]:

Df ¼ �Cf 20 Dm=A ð9:2Þ

Df here denotes the frequency shift due to the addedmass (in Hz),C is a constant, f0 isthe fundamental frequency of the quartz crystal (in Hz) and Dm/A the surface massloading (in g cm�2). The two most common mass-sensitive sensors are the thicknessshear mode resonator (TSMR) and the Rayleigh surface acoustic wave (SAW) device(Fig. 9.4), which will be detailed below.

VOCcheck� of Applied Sensor [21]

This instrument relies on four discrete polymer-coated quartz TSMRs operating at afundamental frequency of 30 MHz. TSMRs typically consist of circular quartz plateswith thin metal (gold) films on both sides. By applying an AC voltage to the electrodes,bulk waves are generated that travel perpendicular to the plate surfaces, the wavelengthof which is determined by the plate thickness. Both faces of the quartz plate execute ashear motion (see Fig. 9.4a). For more details on TSMRs, see [20, 24–26].The system is equipped with an active sampling stage driven by a pump. There is no

reference gas or analyte filter on board. Different sampling probes can be attached tothe input port of the instrument. A single analyte identification cycle takes about 15 s.The battery pack is designed to enable more than 1000 measurement cycles. This

Fig. 9.4 Schematics of mass-sensitive devices. (a) Quartz plate with

electrodes on both sides excited into a shear mode by applying AC.

(b) Launching, propagation and detection of a Rayleigh-type surface

acoustic wave by applying AC to interdigitated transducers


allows operation, for example, over 14 hours performing one measurement per min-ute. Four different data evaluation methods can be stored in the system. Each methodis designed to identify up to 15 analyte classes [21].An important consequence of the simple thermodynamic bulk physisorption me-

chanism of polymer-coated TSMRs (the same holds for SAWs detailed below) is thegood sensor-to-sensor reproducibility as can be taken from Table 9.2 [30]. A signalintensity value, which is composed of the weighted signals of the four TSMRs is dis-played for three different VOCcheck� systems. The TSMRs in the three systems arecoated with the same polymers at identical layer thickness. All three systems repeat-edly recorded signals upon exposure to 1000 ppm toluene, 500 ppm anisole, and2000 ppm propan-1-ol. The intensity value, its standard deviation in three subsequentmeasurements and the similarity (based on normalized feature vectors, max. 100) inreference to system 1 are given. The standard deviation is less than 6%, the similarity93 and better in all cases, which ensures a good system-to-system transferability ofresults and calibration methods.The VOCcheck� is targeted at rapidly identifying organic volatiles in air using its

internal library, which can be customized for each application. The applications in-clude inspecting incoming chemicals or containers, identifying solvents and chemi-cals, verifying the shelf life of products, detecting leaks and monitoring waste andemissions [21].

Tab. 9.2 System-to-system-reproducibility: TSMRs belonging to three

different instruments were coated with the same polymers at identical

layer thickness. All three systems repeatedly recorded signals upon

exposure to 1000 pm toluene, 50 pm anisole, and 2000 pm propan-1-

ol. The intensity value, its standard deviation in three subsequent

measurements and the similarity (max. 100) in reference to instrument

1 are given (Table courtesy of AppliedSensor GmbH, Reutlingen,

Germany).

Analyte Instrument 1

(reference)

Instrument 2 Similarity

(vers. ref.)

Instrument 3 Similarity

(vers. ref.)

Toluene 1000 ppm 329 301 97 317 96

Toluene 1000 ppm 341 280 96 316 97

Toluene 1000 ppm 339 289 98 328 97

Standard deviation 5.2 8.6 5.4

Anisole 500 ppm 1020 978 98 917 93

Anisole 500 ppm 1027 1056 98 1011 93

Anisole 500 ppm 1031 1104 98 1008 93


Propan-1-ol 2000 ppm 172 154 96 166 95

Propan-1-ol 2000 ppm 165 139 97 158 98

Propan-1-ol 2000 ppm 171 135 97 154 98



VaporLab� of Microsensor Systems [22]

The VaporLab� instrument is based on an array of typically four (up to 6) Rayleigh-SAW sensors. The schematic of a SAWdevice is shown in Fig. 9.4b. By applying an ACvoltage to a set of interdigital transducers patterned on a piezoelectric substrate withappropriate orientation of the crystal axes, one set of the fingersmoves downwards, theother upwards, thereby creating a mechanical surface deformation. This surface de-formation generates an acoustic wave, which propagates along the surface and is con-verted back into an electrical signal by deforming the surface in the region of thereceiving transducer. The electrical signal of the receiving transducer is recordedand represents the sensor signal. The SAW devices are coated with thin polymericfilms. For more details on SAW devices, see refs [20, 24–26].The instrument includes a sophisticated sampling system with up to two pumps,

two three-way valves and two preconcentrators, which can be thermally desorbed. Themeasurement cycle time is less than 30 s with a sensor response time of less than 1 s.The battery pack allows for operation between 4 and 12 hours depending on the deviceconfiguration. Up to 200 vapor patterns can be stored in the on-board library and 100measurements can be logged on the instrument [22].Applications of VaporLab� include quality control of packaging, food and beverage

freshness, identification of hazardous chemicals, inspection of raw materials, processcontrol in the food and petrochemical industries, aroma identification of products,monitoring flavor formulation in the food industry, and environmental monitoringof VOCs [22].A variant of the VaporLab� system is theHAZMATCAD� (hazardous material che-

mical agent detector) system, which relies on both SAW devices and up to 3 electro-chemical sensors (for details on electrochemical sensors, see Chapter 4). The ruggedsystem features fast-mode (20 s response time) and sensitive-mode (120 s) operation.The mission life is 8 hours in the fast and 12 hours in the sensitive mode. The dataincluding alarm level, time and date can be logged for 8 hours. The system exhibits anadditional infrared data port and is targeted at detecting chemical warfare agents suchas nerve and blister agents, blood and choking agents, as well as toxic industrial che-micals such as hydrides (arsine, silane), halogens (chlorine, fluorine) and acidic gases(sulphur dioxide) at trace levels [22].

9.2.3.2 Hand-held Units Based on Chemoresistors

Chemoresistors rely on changes in the electric conductivity of a film or bulk materialupon interaction with an analyte. Conductance, G, is defined as the current, I (A),divided by the applied potential, U, (V). The unit of conductance is X�1 or S (Sie-mens). The reciprocal of conductance is the resistance, R, (X). The resistance of asample increases with its length and decreases with its cross-sectional area.Conductometric sensors are usually arranged in a metal-electrode-1/sensitive layer/

metal-electrode-2 configuration [27]. The conductance measurement is done either viaaWheatstone bridge arrangement or by recording the current at an applied voltage in aDC mode or in a low-amplitude, low-frequency AC mode to avoid electrode polariza-tion (for more details on chemoresistors, see Chapter 4).


Several classes of predominantly organic materials are used for application withchemoresistors at room temperature such as conducting polymers or carbon black-loaded polymers. The chemically sensitive layer is applied over interdigitated electro-des on an insulating substrate. Electrode spacing is typically 5 to 100 lm, and the totalelectrode area is a few mm2. The applied voltage ranges between 1 and 5 V.Carbon-Black-Loaded Polymers exhibit conducting carbon black particles dispersed

in non-conducting polymers. The conductivity is by particle-to-particle charge percola-tion so that if the polymer absorbs vapor molecules and swells, the particles are, onaverage, further apart and the conductivity of the film is reduced [31–33]. When thesensor is purged with clean air, the analyte desorbs from the polymer volume, the filmshrinks and the conductive pathways are re-established [34]. The conductivity of thesensors depends critically on the morphology of the sensitive layer, i.e., the averagedistance of the dispersed particles, which involves high demands on sensor-to-sensorreproducibility.

Cyranose 320� of Cyrano Sciences [23]

The Cyranose 320� relies on a 32-channel carbon-black polymer composite chemir-esistor array [31–33]. A sampling pump and an inlet probe is provided with the sys-tem. The system response time is approximately 10 s. The battery pack allows for3 hours of operation. Two classification methods with 6 classes per method can bestored on board. A maximum of 100 identifications can be saved on the instru-ment. A universal serial bus (USB) will be available in future software upgrades [23].Typical applications of the Cyranose 320� include spot testing or continuous mon-

itoring of batch-to-batch consistency and spoilage in raw food materials, solvent ver-ification, identification of organic acids in waste water streams, recognition of gaso-line, diesel and crude oil contamination in recycled containers, or enabling quick as-sessment of the chemical status of industrial processes in the food (coffee roasting andfermentation), petrochemical (plastics manufacture and gasoline blending) and con-sumer products sector (detergents and deodorants). Possible medical applications in-clude obtaining information on the identity of certain chemical compounds in exhaledair and excreted urine or body fluids related to specific metabolic conditions, certainskin diseases or bacterial infections, such as those common to leg or burn wounds [23].

9.3

Silicon-Based Microsensors

Semiconductor technology provides excellent means to meet some of the key criteriaof chemical sensors such as rapid response, low cost, batch fabrication, and offersadditional features such as small size, and on-chip signal processing. The rapid devel-opment of integrated circuit technology during the past few decades has initiatedmany initiatives to fabricate chemical sensors consisting of a chemically sensitivelayer on a signal-transducing silicon chip [35, 36]. Multi-chip solutions with electro-nics and sensors on separate chips have been proposed [37, 38].


The largely two-dimensional integrated circuit (IC) and chemical sensor structuresprocessed by combining lithographic, thin film, etching, diffusive and oxidative stepshave been recently extended into the third dimension using micromachining technol-ogies – a combination of special etchants, etch stops and sacrificial layers [7–11]. Avariety of micromechanical structures including cantilever beams, suspended mem-branes, freestanding bridges, etc. have been produced using micromachining technol-ogy (MicroElectroMechanical Systems, MEMS) [7–11]. MEMS technology thus pro-vides a number of key features, which can serve to enhance the functionality of che-mical sensor systems. In a further step, microelectronics and micromechanics(MEMS-structures) have been realized on a single chip allowing for on-chip controland monitoring of the mechanical functions as well as for data preprocessing such assignal amplification, signal conditioning, and data reduction [7–11, 39–42].In the following, a short introduction into micromachining technology (Sec-

tion 9.3.1) will be given, and three different types of CMOS-based (CMOS, comple-mentary metal-oxide semiconductor, is a standard IC fabrication process) transducerswill be described (Section 9.3.2 to 9.3.4), which serve as components of a single-chipsystem. The multisensor-chip, which forms integral part of a hand-held microsensor-based gas detection unit, will be detailed in Section 9.3.5.

9.3.1

Micromachining Techniques

9.3.1.1 Bulk Micromachining

One approach to enhance the functionality of IC-based substrates includes microma-chining the bulk substrate, which in most cases consists of silicon. Silicon can be dryor wet etched by various techniques [9–11, 43]. Some wet etchants such as nitric acid/hydrofluoric acid lead to isotropic etching – the same etch rate in all directions, otherssuch as potassium hydroxide lead to anisotropic etching, that is they preferentially etchaway the silicon along certain crystal planes while preserving it in other directions(Fig. 9.5a). Typical structures obtained by, for example, anisotropic wet etchingthrough the complete bulk silicon of a CMOS wafer include membranes consistingof the remaining dielectric CMOS layers. The thermal oxide serves as an etch-stoplayer. The resulting membrane structures can be used for sensors requiring excellentthermal insulation, such as calorimetric chemical sensors or semiconductor-oxide-cov-ered microhotplates. Polysilicon and metal structures sandwiched in-between the di-electric layers can be used to create, e.g., thermopiles and heating resistors [13].Another technique is dry etching. Again, there is isotropic etching performed by

using, for example, xenon difluoride or anisotropic etching by reactive-ion-etching(RIE). RIE is used, for example, to release cantilevers or create bridge structuresfrom preformed membranes [44].


9.3.1.2 Surface Micromachining

Surface micromachining comprises a number of techniques to produce microstruc-tures from thin films previously deposited onto a substrate and is based on a sacrificiallayermethod (Fig. 9.5b). In contrast to bulkmicromachining, surfacemicromachiningleaves the substrate intact. A sacrificial layer is deposited and patterned on a substrate.After that, a structural thin film, in most cases polysilicon, is deposited and patterned,which will perform the mechanical or electrical functions in the final device. A selec-tive etchant then removes exclusively the sacrificial layer material. The thickness of thesacrificial layer determines the distance of the structural parts from the substrate sur-face. Common sacrificial layer materials include silicon oxide etched by hydrogenfluoride and aluminum etched by a mixture of phosphoric, nitric and acetic acid.Clamped beams, microbridges, or microchannels can be fabricated this way, micro-rotors and even microgears can be realized by repeated layer deposition and etching[9–11, 43].

9.3.2

Microstructured Chemocapacitors

Chemocapacitors (dielectrometers) rely on changes in the dielectric properties of asensing material upon analyte exposure (chemical modulation of the capacitanceby changes in the dielectric constant of the sensitive layer). Interdigitated structuresrather than plate capacitors are predominantly used [45–47]. In some cases, plate-capacitor-type structures with the sensitive layer sandwiched between a porousthin metal film (permeable to the analyte) and an electrode patterned on a siliconsupport are used [48, 49]. The capacitances are usually measured at an AC frequencyof a few kHz up to 500 kHz.The size of the capacitor shown in Fig. 9.6 is 800 � 800 lm2, its electrode width and

spacing are 1.6 lm. Since the nominal capacitance of this interdigitated capacitor is inthe range of few picoFarad, and the expected capacitance changes upon analyte absorp-tion are in the attoFarad range, a dedicated measurement configuration and specificsignal conditioning circuitry had to be developed. The sensor response is read out asthe differential signal between a polymer-coated sensing and a nitride-passivated re-ference capacitor. Both sensor and reference capacitors are split into two parts to im-prove the charge transfer efficiency. The sensing capacitor, (CS), and the reference

Fig. 9.5 Micromachining

techniques: (a) bulk micro-

machining, anisotropic and

isotropic etching, (b) surface

micromachining with sacrificial

layer, structural layer and a

subsequent etch step


capacitor, (CR), are incorporated in the first stage of a fully differential second-orderSigma-Delta-modulator (Fig. 9.7) with two switched-capacitor-integrators and a subse-quent comparator [50, 51]. A second-order Sigma-Delta-modulator is used to achieveshorter analog-to-digital conversion time. Since the output bit stream of the Sigma-Delta-modulator is proportional to the ratio (CS-CR)/(Cfb), the four feedback capacitors(FB in Fig. 9.6, Cfb in Fig. 9.7) are realized as interdigital capacitors with the samematerials as the sensing and reference capacitors in order to eliminate differencesin temperature behavior and ageing. Due to the small signal bandwidth, the outputbit stream of the Sigma-Delta-modulator is decimated using a frequency counter. Formore details on circuitry see [50, 52].Two effects change the capacitance of a polymeric sensitive layer upon absorption of

an analyte: (i) swelling and (ii) change of the dielectric constant due to incorporation ofthe analyte molecules into the polymer matrix [52, 53]. For a simple interdigitated

Fig. 9.6 Micrograph of a capaci-

tive sensor system including a

polymer-coated sensing capacitor

(S), a passivated reference capa-

citor (R), four interdigitated feed-

back capacitors (FB), and the

Sigma-Delta circuitry (RD) asdetailed in the text. Reprinted from

[50] with permission

Fig. 9.7 Schematic of the fully differential second-order Sigma-Delta-

modulator exhibiting two switched-capacitor-integrators and a subse-

quent comparator. Four feedback capacitors (Cfb) are realized as

interdigital capacitors. The Sigma-Delta-modulator provides a pulse-

density-modulated digital output that is decimated using the frequency

counter


structure, the space containing 95% of the field lines includes the polymer volumewithin a layer thickness of half the periodicity of the electrodes [54]. For a layer thick-ness less than half the periodicity, swelling of the polymer upon analyte absorptionalways results in an increase of the measured capacitance regardless of the dielectricconstant of the analyte (Eq. 4). This results from the increased polymer/analyte volumewithin the field line region exhibiting a larger dielectric constant than that of the sub-stituted air. The capacitance change for a polymer layer thicker than half the periodicityof the electrodes is determined by the ratio of the dielectric constants of analyte, eA (theanalyte is assumed to be in the liquid state) and polymer, epoly (Eq. 3). If the dielectricconstant of the polymer is lower than that of the analyte, the capacitance will be in-creased. Conversely, if the polymer dielectric constant is larger, the capacitance will bedecreased (Fig. 9.8). These effects have been discussed and supported by simulationsin [53], where the following formulae have been used to describe the change of thesensor capacitance.

eeff ¼ epolyð1� VFAcAÞ þ eA � VFAcA ð9:3Þ

heff ¼ hð1þ SAcAÞ ð9:4Þ

Here eeff denotes the resulting effective dielectric constant of the polymer/analytesystem. epoly is the dielectric constant of the polymer, cA the concentration of the ana-lyte in the gas phase,VFA the volume fraction of the analyte in the polymer per unit gasphase concentration, heff the resulting effective polymer thickness after analyte absorp-tion and SA the experimental swelling coefficient of the polymer per unit gas phaseconcentration for the respective analyte. VFA and SA are constants (Henry’s law isassumed to be valid) and have to be determined experimentally for every polymer/analyte combination by mass-sensitive or optical measurements. Typical sensor si-gnals for a polymer layer (poly(etherurethane), PEUT), which is thicker (4.3 lm)than the surface-normal extension of the field lines, are shown in Fig. 9.7. The capa-citor is alternately exposed to various concentrations of toluene and ethanol at 301 Kand the pure carrier gas. The ratio of the dielectric constants of polymer (2.9) andanalytes (toluene: 2.4, ethanol: 24.5) controls the signs of the signals. Ethanol with

Fig. 9.8 Frequency responses

of a switched-capacitor device

upon exposure to different ana-

lytes at 301 K. For a thick layer of

PEUT (4.3 lm), toluene (die-

lectric constant lower than that

of PEUT) causes positive fre-

quency shifts, ethanol (dielectric

constant higher than that of

PEUT) leads to negative fre-

quency shifts. Reprinted from

[13] with permission


a dielectric constant larger than that of PEUT causes positive frequency shifts, whereastoluene with a smaller dielectric constant causes negative frequency shifts. The limit ofdetection of the capacitive microsystem at 301 K is approximately 8 ppm for tolueneand 5 ppm for ethanol [51].For conducting measurements at defined temperatures, sensor and reference capa-

citors can be placed on thermally isolatedmembrane structures [50]. The fabrication ofcapacitors integrated with CMOS circuitry components is described in [50–53, 55, 56].The main application of microstructured capacitors includes humidity sensing with

polyimide films [45–48, 55, 56], since water has a high dielectric constant of 78.5(liquid state) at 298 K and, therefore, generates large capacitance changes. Capacitivehumidity sensors are commercially available from, for example, Sensirion (SUI), Vai-sala (FIN), and Humirel (F) [57–59].

9.3.3

Micromachined Resonating Cantilevers

Micromachined cantilevers commonly employed in atomic force microscopy (AFM)constitute a promising type ofmass-sensitive transducer for chemical sensors [60–64].The sensing principle is quite simple. The cantilever is a layered structure (Fig. 9.9)composed of the dielectric layers of a standard CMOS process, silicon, metallizations,and, eventually, zinc oxide. The cantilever base is firmly attached to the silicon support.The free-standing cantilever end is coated with a sensitive layer and is deflected de-pending on the added mass. The first step to achieve cantilever structures is the fab-rication of membranes using anisotropic bulk etching with an electrochemical etch-stop technique (Fig. 9.10). The cantilevers are then released with additional front-sidereactive-ion-etching steps [12, 13, 65]. After micromachining, the cantilevers are spray-coated with the chemically sensitive polymer film.There are two fundamentally different operation methods: (a) static mode: measure-

ment of the cantilever deflection upon stress changes or mass loading by means of

Fig. 9.9 Schematic representation (a) and micrograph (b) of an in-

tegrated CMOS cantilever with on-chip circuitry. The cantilever is

thermally actuated, its vibration is detected by piezoresistors. Reprinted

from [13] with permission


laser light reflection [61, 63, 64]; (b) dynamic mode: excitation of the cantilever in itsfundamental mode and measurement of the change in resonance frequency uponmass loading [12, 13, 62, 65] in analogy to other mass-sensitive devices. The two meth-ods impose completely different constraints on the cantilever design for optimumsensitivity. Method (a) requires long and soft cantilevers to achieve large deflec-tions, whereas method (b) requires short and stiff cantilevers to achieve high opera-tion frequencies. Method (b) is preferable with regard to integration of electronics andsimplicity of the setup (feedback loop) [12, 13, 62, 65, 66]. Method (a) can be applied inliquids as well [63], which is rather difficult using the dynamic mode. The excitation ofthe cantilever in the resonant mode is usually performed by applying piezoelectricmaterials (ZnO) [66] or by making use of the bimorph effect, which is the differenttemperature coefficients or mechanical stress coefficients of the various cantilevermaterials [12, 13, 60–65]. This difference in material properties gives rise to a canti-lever deflection upon heating or applying mechanical forces. Periodic heating pulsesin the cantilever base thus can be used to thermally excite the cantilever in its reso-nance mode at 10–500 kHz [12, 13, 65]. The cantilever vibration is detected by em-bedded piezoresistors in a Wheatstone-bridge configuration (Fig. 9.11). The cantileverexhibits a quality factor of approximately 1000 in air at a resonance frequency of380 kHz [67], and acts as the frequency-determining element in an oscillation circuit(Fig. 9.11), which is entirely integrated on the chip. The oscillation frequency is mea-sured with an on-chip counter. The first stage of the feedback circuitry includes a low-noise fully differential difference amplifier (DDA) with a gain of 30 dB, which ampli-fies the output signal of the Wheatstone bridge. The feedback additionally includes ahigh-pass filter followed by another amplifier, a limiter, a programmable digital delayline, and a driving stage. The delay line is used to adjust the phase to achieve positive

Fig. 9.10 Cantilever fabrication: (a) thinned CMOS wafer with Si-nitride layer

on backside, (b) backside KOH wet etching (bulk micromachining) with

electrochemical etch stop, and (c) frontside RIE to release the cantilever [67]

Fig. 9.11 Schematic of the cantilever feedback circuitry, which includes two

cascaded amplifiers, a high-pass filter, a limiter, a programmable digital delay

line, and a driving stage


feedback at the fundamental resonance frequency. For more details on the electronics,see [65, 67].In a first approximation, the change in the resonance frequency of the cantilever, f

(Hz), is proportional to the change in the analyte gas concentration, cA (mol/m3) (linearslope in Fig. 9.12):

f ¼ G �MA � K � cA ð9:5Þ

Here, G (Hz/(kg/m3)) denotes the gravimetric sensitivity of the cantilever. G is canti-lever-specific and depends on the geometric dimensions (thickness, length) and thematerial of the cantilever [67]. Here it is on the order of 10–20 Hz/(kg/m3). A frequen-cy shift of 1 Hz hence corresponds to a change in the vibrating mass of approximately5 pg.MA [kg/mol] is the analyte molecular weight and K the thermodynamic partitioncoefficient (Eq. 1), which is dimensionless and characteristic for the particular poly-mer-analyte combination.The mass resolution of the cantilevers is in the range of a few picograms [60–67].

This high mass sensitivity does not necessarily imply an exceptionally high sensitivityto analytes since the area coated with the sensitive layer usually is very small (on theorder of 100 � 150 lm). The sensing layer is deformed upon motion of the cantilever,therefore, modulus effects are expected to contribute to the overall signal, especiallysince the coating thickness may exceed that of the cantilever. IC process-compatiblefabrication sequences for monolithic integration of the cantilevers with electronics aredetailed in [12, 13, 62, 65, 67].Typical application areas are environmental monitoring such as the detection of

different kinds of organic volatiles, such as hydrocarbons, chlorinated hydrocar-bons, alcohols in the gas phase by using polymeric layers [60–67]. Figure 9.12 showsthe measured frequency shift of a cantilever coated with 3.7 lm PEUT upon exposureto different concentrations of n-octane. The concentrations are ramped up and down totest for reproducibility [67]. The measurement chamber is purged with synthetic air

Fig. 9.12 Measured frequency

shifts of a cantilever coated with

3.7 lm PEUT upon exposure to

different concentrations of n-

octane (250–1500 ppm). The

mass-sensitivity is approxima-

tely 5 picograms/Hz [67]


after each analyte exposure. A continuous directional sensor drift of 0.15 Hz/min hasbeen subtracted from the results shown in Fig. 9.12.

9.3.4

Micromachined Calorimetric Sensors

This type of sensor relies on the thermoelectric or Seebeck-effect. If two differentsemiconductors or metals are connected at a hot junction and a temperature differ-ence is maintained between this hot junction and a colder point, then an open circuitvoltage is developed between the different leads at the cold point. This thermovoltage isproportional to the difference of the Galvani potentials (the Galvani potential is definedas the difference of the Fermi levels of the two materials) at the two temperatures andthus proportional to the temperature difference itself [68]. This effect can be used todevelop a thermal sensor by placing the hot junction on a thermally isolated structurelike a membrane, bridge etc. and the cold junction on the bulk chip with the thermallywell-conducting silicon underneath [13, 69–71]. The membrane structure (hot junc-tion) is covered with a sensitive or chemically active layer liberating or abstracting heatupon interaction with an analyte. The resulting temperature gradient between hot andcold junctions then generates a thermovoltage, which can be measured.Figure 9.13 displays the schematic of a CMOS thermopile. The overall sensor system

includes two 700 lm by 1500 lm dielectric membranes with 300 polysilicon/alumi-num thermocouples each (Seebeck coefficient: 111 lV/K) and an on-chip amplifier.One of the membranes is coated with a gas-sensitive polymer, the other one is passi-vated and serves as a reference [13, 71, 72]. The sensing and reference thermopiles areconnected in parallel to the input stage of an on-chip amplifier for monitoring the

Fig. 9.13 Schematic of a thermoelectric sensor. Polysilicon/aluminum

thermopiles (hot junctions on the membrane, cold junctions on the bulk chip)

are used to record temperature variations upon analyte sorption in the polymer.

The overall system includes a sensor membrane coated with gas-sensitive

polymer, a passivated (Si-nitride) reference membrane and an on-chip amplifier

[13, 71]


temperature differences between the two membranes (Fig. 9.14). The low-noise chop-per-stabilized instrumentation amplifier features a tunable gain of up to 8000 and abandwidth of 500 Hz with an equivalent input noise of 15 nV/Hz1/2 [73]. The anti-aliasing filter prevents downsampling of high-frequency noise into the low-frequencysignal band by the A/D converter. After Sigma-Delta A/D conversion and after passinga decimation filter (13 bit word length at 800/s), the data are read out or transferred to adigital interface [81].Calorimetric detection includes four principal steps: (i) absorption and partitioning,

or chemical reaction; (ii) generation of heat, which causes (iii) temperature changes tobe transformed in (iv) thermovoltage changes (see examples [13, 70]). Each of the foursteps contributes to the overall sensor signal. The thermovoltage change U (V) is pro-portional to the derivative of the analyte concentration as a function of time dcA/dt(mol/m3s):

U ¼ A � B � Vpoly �H � K � dcA=dt ð9:6Þ

Here A (K � s/J) and B (V/K) are device- and coating-specific constants describing thetranslation of a generatedmolar absorption enthalpyH (J/mol) via a temperature chan-ge into a thermovoltage change. Vpoly denotes the sensitive polymer volume, and K isthe partition coefficient (Eq. 1) or reaction equilibrium constant.The calorimetric sensor only detects changes in the heat budget at nonequilibrium

state (transients) upon changes in the analyte concentration. Thus, the sensors providea signal upon absorption (condensation heat) and desorption (vaporization heat) ofgaseous analytes into the polymer [70–72, 74, 75] or during chemical reaction ofan analyte with the sensing material [75–79]. Processing sequences for the integra-tion of thermoelectric sensors with circuitry in a CMOS standard process are detailedin [71, 79]. Sensors are commercially available from Xensor Integration (NL) [80].Typical applications include the detection of different kinds of organic volatiles in

the gas phase, for example, hydrocarbons, chlorinated hydrocarbons, and alcohols, byusing polymeric layers [13, 70–75]. Figures 9.15 (a) and (b) show the output voltage ofthe microsystem while switching from synthetic air (nitrogen/oxygen mixture withouthumidity) to toluene (4000 ppm) and back to air at a temperature of 301 K [13, 71].Enthalpy changes can be roughly approximated by integration over the peak areaof the sensor signals [71]. The peak maximum and signal characteristics differ for

Fig. 9.14 Schematic of the calorimeter circuitry. Sensing and refe-

rence thermopiles are connected in parallel to the input stage of a low-

noise chopper-stabilized instrumentation amplifier followed by an anti-

aliasing filter, a Sigma-Delta A/D converter and a decimation filter


switching on and switching off the analyte. This is mainly due to the different timeneeded for generating the gasmixture (longer time) and just shutting down the analytegas flow (short time). Optimizing the gas flow and considerable shortening of the gaspaths are the most urgent improvements planned for the setup used here.

9.3.5

Single-Chip Multisensor System

The CMOS fabrication approach makes it possible to realize all three different trans-ducers with their signal conditioning circuitry on the same chip. The orthogonality ofthe sensor signals is ensured by the fundamentally different transduction principlesand is system-inherent. Additionally, several of such microsystem chips can be coatedwith different sensitive coatings and arranged in an array to further enhance the ana-lyte characterization and discrimination performance.A schematic of the overall microsensor system architecture is shown in Fig. 9.16, a

micrograph is displayed in Fig. 9.17. On the left-hand side of Fig. 9.16, are the fourdifferent sensors: capacitive, mass-sensitive, calorimetric and temperature [81]. Thechip includes a temperature sensor in addition to the three different chemical sensortypes since bulk physisorption of volatiles in polymers is strongly temperature-depen-dent: A temperature increase by 10 8C decreases the fraction of analyte moleculesabsorbed in the polymer by approximately 50% and consequently leads to a drasticsensor signal reduction.

Fig. 9.15 Thermovoltage of the microsystem while (a) switching from

synthetic air (nitrogen/oxygen mixture without humidity) to toluene

(4000 ppm) and (b) back to air at a temperature of 301 K [13]. Enthalpy

changes can be roughly approximated by integration over the peak area

of the sensor signals. Reprinted from [13] with permission. For details,

see text


The sensor front end represents all the sensor-specific driving circuitry and signal-conditioning circuitry. The analog/digital conversion is done on chip as well, whichachieves a unique signal-to-noise ratio since noisy connections are avoided and a ro-bust digital signal is generated on chip and afterwards transmitted to an off-chip dataport via an I2C� interface [82]. The I2C� bus interface offers the additional advantage ofhaving only very few signal lines (essentially two) for bi-directional communicationand also allows for operating multiple chips on the same bus system. An on-chipdigital controller manages the sensor timing and the chip power budget [81].Figure 9.17 shows the processed and tested single-chip gas sensor system [81]. The

chip is 7 by 7 mm in size. The chip processing steps include an unaltered 15-maskcommercial CMOS process [83] followed by applying a backside mask and anisotropicsodium hydroxide etching of the membrane structures for the calorimetric sensorsand the cantilever. Front side reactive ion etching (RIE) is then executed to releasethe cantilever from the respective membrane. Finally, the sensors (cantilever, calori-meter, capacitor) have to be coated with the polymer using an airbrush method. Thesensors are located in the center within ametal frame, which is used to apply a flip-chippackaging technique [84].The signals of all three transducers correlate linearly with the analyte concentration

in the low-concentration range (below 3% of saturation vapour pressure at operating

Fig. 9.16 Schematic of the

overall microsystem architecture

comprising sensors, driving and

signal conditioning circuitry

(sensor front end), analog/digi-

tal converters, sensor control

and power management unit,

and a digital interface. Reprinted

from [81] with permission

Fig. 9.17 Micrograph of the

single-chip gas microsensor sy-

stem. Reprinted from [81] with

permission


temperature) [85]. Each transducer provides different information on the target ana-lytes. Alcohols, for example, provide comparably low signals on mass-sensitive trans-ducers due to their high saturation vapor pressure and low molecular mass. On theother hand, alcohols exhibit a dielectric constant of 24.5 and provide large signals oncapacitors. Drastic changes in thermovoltages on the thermopiles are measured uponexposure to chlorinated hydrocarbons (not shown here) used in cooling sprays, forexample, which in turn have a low dielectric constant and provide only small signalson capacitors. Thus, comprehensive and complementary information is acquired withthe multi-transducer system, and analyte characterization and identification are sig-nificantly improved.Six multi-sensor chips have been arranged on a ceramic board (Fig. 9.18) as sensing

unit for a hand-held instrument. Each of the chips is coated with a different polymer.The unit thus provides a total of 18 different (6 capacitive, 6 mass-sensitive and 6calorimetric) chemical sensor signals. The hand-held unit comprises in addition tothe sensors a gas intake unit with pumps, valves and filters, a signal processingunit (microcontroller) and a power pack allowing for more than 24 hours of contin-uous operation. This is the first hand-held unit benefitting from microsystem tech-nology components as sensor elements. The average power consumption of one ofthe multisensor chips is approximately 100 mW.

9.3.6

Operation Modes for CMOS Microsystems

Since hand-held systems are designed to by small and thus have severe constraintsimposed on the use of calibration gases or filter units as well as on the overall powerconsumption, one has to come up with dedicated operation modes, which still enablereliable qualitative and quantitative measurements. One possibility includes the so-called “reverse mode of operation” (RMO) [86], which has been developed for thehand-held microsensor unit and adapted to the needs of the different transducingprinciples as will be described in the following.

Fig. 9.18 Six multi-sensor chips arranged on a

ceramic board as sensing unit in a hand-held device.

Each of the chips is coated with a different polymer


9.3.6.1 Reverse Mode of Operation (RMO) [86]

In classical systems, the baseline is established by purging with pure carrier gas (ty-pically nitrogen or synthetic air) or filtered ambient air. Purging is in most cases thebasic state of the system and the sensors are exposed to the target analytes only for ashort time. The sensor signal upon analyte exposure is then related to the carrier gas orambient air signal, which serves as the “zero-signal”. Since neither carrying large re-ference gas cylinders nor using high-capacity filters is feasible with portable hand-heldunits, another solution had to be found.The new concept is based on the idea to invert operation conditions. Instead of

purging as standard state and short gas exposure times, the system is now continu-ously exposed to analyte gas, and only short pulses of filtered or reference gas are usedto re-establish the baseline. This operation mode induces higher demands on the sen-sor stability, since large drift deteriorates the sensor signal much more in using RMOthan in using “classical” operation conditions. For the microsensor system, the refer-ence baseline is established by 5 seconds of purging with filtered ambient air. Thistechnique allows performing some hundred measurement cycles using a very smallfilter element.Figure 9.19 illustrates the operation states of valves and pumps in the RMO and the

corresponding gas concentrations in the chamber. Figure 9.19 additionally shows thestrategy and timing of the signal recording for the different transducers and the re-sulting sensor signals. Line 1 is indicating the valve status. “0” represents the basicstate of the valve, when ambient analyte-loaded gas is directly transferred to the mea-surement chamber. In state “1”, the ambient gas passes a filter unit, and analyte mo-lecules are removed from the gas stream: cleaned “reference” gas is flowing over thesensors. Line 2 represents the pump status. “0” means “pump off”, “1” means that thepump is operational. In line 3, the corresponding analyte gas phase concentrations aredisplayed. In the beginning of a measurement sequence the gas composition in themeasurement chamber is not defined. The pump is then switched on for three sec-onds and analyte gas is pumped into the measurement chamber. The gas remains inthe chamber for two seconds. Equilibrium signals of the capacitive and mass sensitivetransducers are recorded, the measurement timing of which is displayed in line 4. Theresulting sensor signals are schematically shown in line 5. The pump is then switchedon again for three seconds, and the valve is set to make the analyte gas pass the filter. Aconcentration step between analyte-loaded and filtered gas is thus generated.Equilibrium state capacitive and mass-sensitive reference signals in filtered air are

recorded over two seconds, after which the pump is switched on once more with thevalve set to analyte gas. The last pump operation would not be necessary for the equi-librium-based sensors but it is necessary to get the second calorimetric transient asshown in line 7. As already described in Section 9.3.4, the calorimetric sensor relies ontransients and provides signals exclusively upon concentration changes. Therefore, thecalorimetric recording has to be performed at high time resolution (1 kHz) in twoshort intervals covering both flanks of the concentration signal (line 3), i.e., at themaximum gradient of the analyte concentration. The two transient signals of the ca-lorimetric transducer (negative upon analyte desorption, positive upon analyte absorp-tion) are displayed in line 7.


Fig. 9.19 Reverse Mode of Operation (RMO) [86], as developed for a

hand-held unit including micromachined multisensor chips: Operation

states of valves and pumps and corresponding gas concentrations in

the chamber (lines 1–3), and timing of the signal recording for the

different transducers as well as resulting sensor signals (lines 4–7). For

details see text


9.4

Summary and Outlook

The field of hand-held sensor systems has been developing rather fast during the pastfive years. Most of the commercially available systems rely on polymers (pure or car-bon-loaded) as sensitive layers, and are hence targeted at detecting organic volatiles atroom temperature. Metal-oxide-based sensors operating at higher temperatures arenot yet included into hand-held units due to their high energy consumption. The ap-plications of hand-held instruments presently include, for example, quality and pro-cess control in industry, medical and environmental monitoring or hazardousmaterialidentification in industrial and military settings. The trend in commercial hand-helddevices is for dedicated sensor solutions serving defined target applications. This trendmay trigger the application-specific development of individual hand-held systems forcertain key applications rather than following the “nose”-concept of using one univer-sal hand-held system for a wealth of applications. Extensive customer support will be acrucial issue in this context.CMOS-based chemical microsensors with polymeric layers constitute a promising

approach and offer a number of substantial advantages such as full microelectronicscompatibility, extremely small size, low power consumption, and production at indus-trial standards. Further research and the fast progress of microelectronics develop-ment will help to significantly improve the system performance and reduce itssize. Palm-size or even credit-card-size detection units based on CMOS technologyare conceivable in the near future.

Acknowledgments

The authors are greatly indebted to current and former staff of the Physical ElectronicsLaboratory at ETH Zurich and of the sensor laboratory at University of Tubingen in-volved in the chemical microsensor development, in particular Christoph Hagleitner,Dirk Lange, Andreas Krauss, and Michael Frank. The authors like to express theirgratitude to Dr. Heiko Ulmer, AppliedSensor GmbH, Reutlingen, Germany, for pro-viding scientific material. This work has been financially supported by the KorberFoundation, Hamburg, Germany.

References

1 http://www.draeger.com2 http://www.afcintl.com/badges.htm3 R. A. Durst, R. W.Murray, K. Izutsu,K. M. Kadish, L. R. Faulkner. Draft IUPACReport Commission V.5.

4 A. Hulanicki, S. Glab, F. Ingman. Pure &Appl. Chem., 1991, 63, 1247.

5 http://www.draeger.com/com/ST/Prod/Detection/Mobile_messtechnik/Gasmess-geraete/PacIII/PacIII.jsp

6 J. W. Gardner, P. N. Bartlett. Sens. ActuatorsB, 1994, 18–19, 211–220.

7 (a) S. Middelhoek, S. A. Audet. SiliconSensors, Academic Press Inc., London,1989; (b) Middelhoek, S., Sens, Actuators A,1994 41–42, 1–8.

8 S. M. Sze, (ed.). Semiconductor Sensors,Wiley, New York, 1994.

9 G. T. A. Kovacs. Micromachined Transducers,WCB McGraw-Hill, Boston, 1998.


10 M. Elwenspoek, H. Hansen. Silicon Micro-machining, Cambridge University Press,Cambridge, 1998.

11 M.Madou. Fundamentals of Microfabrication,CRC Press, Boca Raton, FL, 1997.

12 H. Baltes, D. Lange, A. Koll. IEEE Spectrum,1998, 9, 35–38.

13 A. Hierlemann, D. Lange, C. Hagleitner,N. Kerness, A. Koll, O. Brand, H. Baltes.Sens. Actuators B, 2000, 70, 2–11.

14 W. P. Carey, K. R. Beebe. B. R. Kowalski.Anal. Chem. 1986, 58, 149–153.

15 D. S. Ballantine, S. L Rose, J. W. Grate,H. Wohltjen. Anal. Chem. 1986 58,3058–3066.

16 A. Hierlemann, M. Schweizer-Berberich,U. Weimar, G. Kraus, A. Pfau, W. Gopel.Pattern Recognition and MulticomponentAnalysis, In: Sensors Update Vol. 2, Baltes ,W. Gopel H., Hesse J., Eds.; VCH: Wein-heim, Germany, 1996.

17 D. L. Massart, B. G. M.Vandeginste,S. N. Deming, Y. Michotte, L. Kaufman.Data Handling in Science and Technology 2:Chemometrics: a Textbook, Amsterdam,Elsevier 1988.

18 R. G. Brereton, (ed.). Multivariate patternrecognition in chemometrics, Data handlingin science and technology, Vol. 9, Elsevier,Amsterdam, New York, 1992.

19 P. W. Atkins. Physical Chemistry, OxfordUniversity Press, Oxford, 5th edition, 1994.

20 K. Bodenhofer, A. Hierlemann, G. Noetzel,U. Weimar, W. Gopel. Anal. Chem., 1996,68, 2210–2218.

21 http://www.appliedsensor.com22 http://www.microsensorsystems.com/

frameset_products.htm23 http://cyranosciences.com/products/

cyranose.html24 J. W. Grate, G. C. Frye. In: Sensors Update;

Vol. 2; Baltes, H.; Gopel, W.; Hesse, J., Eds.;VCH: Weinheim, FRG, 1996; pp. 37–83.

25 J. W. Grate, S. J. Martin, R. M. White.Anal. Chem., 1993, 65, 940A–948A and987A–996A.

26 D. S. Ballantine, R. M. White, S. J. Martin,A. J. Ricco, G. C. Frye, E. T. Zellers,H. Wohltjen. Acoustic Wave Sensors: Theory,Design, and Physico-Chemical Application,Academic Press, San Diego, 1997.

27 J. Janata. Principles of Chemical Sensors,Plenum, New York, 1989.

28 G. Z. Sauerbrey. Phys., 1959, 155, 206–222.

29 D. S. Ballantine, H.Wohltjen. Anal. Chem.,1989, 61, 705–712.

30 Table reprinted with permission of AppliedSensor GmbH, Reutlingen, Germany.

31 M. C. Lonergan, E. J. Severin, B. J. Doleman,S. A. Beaber, R. H. Grubbs, N. S. Lewis.Chem. Mater., 1996, 8, 2298–2312.

32 E. J. Severin, B. J. Doleman, N. S. Lewis.Anal. Chem., 2000, 72, 658–668.

33 B. J. Doleman, M. Lonergan, E. J. Severin,T. P. Vaid, N. S. Lewis. Anal. Chem., 1998,70, 4177–4190.

34 D. Stauffer, A. Aharony. Introduction intopercolation theory, Taylor and Francis, Bristol,PA, 1994.

35 A. Van den Berg, P. D. van der Waal,B. B. van der Schoot, N. F. de Rooij. Sens.Mat. 1994, 6, 23–43.

36 G. Muller, P. P. Deimel, W. Hellmich,C. Wagner. Thin Solid Films, 1997, 296,157–163.

37 J. V. Hatfield, P. I. Neaves, P. J. Hicks,K. Persaud, P. Traves. Sens. Actuators B,1994, 18–19, 221–228.

38 P. I. Neaves, J. V. Hatfield. Sens. ActuatorsB, 1995, 26–27, 223–231.

39 J. L. Rodriguez, R. C. Hughes, W. T. Corbett,P. J. McWhorter. Technical Digest IEEEInternational Electron Devices MeetingNew York, NY, USA, 1992, 521–524.

40 M. J. Vellekoop, G. W. Lubking, P. M. Sarro,A. Venema. Sens. Actuators A, 1994, 44,249–263.

41 H. Baltes, O. Brand. IEEE AES SystemsMagazine, 1999, 14, 29–34; and Proc. SPIE,1999, 3673, 2–10.

42 J. N. Zemel. Rev. Sci. Instrum. 1990, 61 (6),1579–1606.

43 W. Menz, J. Mohr, O. Paul. MicrosystemTechnology, Wiley-VCH, Weinheim,New York, 2001, ISBN 352-729-634-4.

44 O. Brand, H. Baltes.Micromachined resonantsensors, In: Sensors Update, Vol. 4, Baltes,H.; Gopel, W.; Hesse, J., (Eds.), VCH,Weinheim, Germany, 1999.

45 N. F. Sheppard, D. R. Day, H. L. Lee,S. D. Senturia. Sens. Actuators, 1982, 2,263–274.

46 M. C. Glenn, J. A. Schuetz. Technical DigestTransducers 1985, 217–219.

47 S. D. Senturia. Technical Digest Trans-ducers 1985, 198–201.


48 G. Delapierre, H. Grange, B. Chambaz,L. Destannes. Sens. Actuators, 1983, 4,97–104.

49 H. Shibata, M. Ito, M. Asakursa,K. Watanabe. IEEE Transactions onInstrumentation and Measurement, 1996,45, 564–569.

50 (a) C. Hagleitner, A. Koll, R. Vogt, O. Brand,H. Baltes. Proc. Transducers ’99, Sendai,Japan, 1999, 1012–1015, (b) S. Kawahito,A. Koll, C. Hagleitner, H. Baltes,Y. Tadokoro. Trans. IEE of Japan, 1999,119–E, 3, 138–142.

51 A. Koll, A. Kummer, O. Brand, H. Baltes.Proc. of SPIE Smart Structures andMaterials, Vol. 3673, Newport 1999.

52 C. Cornila, A. Hierlemann, R. Lenggen-hager, P. Malcovati, H. Baltes, G. Noetzel,U. Weimar, W. Gopel. Sens. Actuators B,1995, 24–25, 357–361.

53 F. P. Steiner, A .Hierlemann, C. Cornila,G. Noetzel, M. Bachtold, J. G. Korvink,W. Gopel, H. Baltes. Technical DigestTransducers 1995, Vol. 2, 814–817.

54 P. Van Gerwen, W. Laureys, G. Huybe-rechts, M. Op De Beeck, K. Baert, J. Suls,A.Varlan, W. Sansen, L. Hermans,R. Mertens. Technical Digest Transducers1997, Vol. 2, 907–910.

55 T. Boltshauser, H. Baltes. Sens. Actuators A,1991, 25–27, 509–512.

56 T. Boltshauser, L. Chandran, H. Baltes,F. Bose, D. Steiner. Sens. Actuators B, 1991,5, 161–164.

57 http://www.sensirion.com58 http://www.vaisala.com59 http://www.humirel.com60 T. Thundat, G. Y. Chen, R. J. Warmack,

D. P. Allison, E. A. Wachter. Anal. Chem.1995, 67, 519–521.

61 M. Maute, S. Raible, F. E. Prins, D. P. Kern,H. Ulmer, U. Weimar, W. Gopel. Sens.Actuators B 1999, 58, 505–511.

62 H. Jesenius, J. Thaysen, A. A. Rasmussen,L. H. Veje, O. Hansen, A. Boisen. Appl.Phys. Letters, 2000, 76, 2615–2617.

63 J. Fritz, M. K. Baller, H. P. Lang,H. Rothuizen, P. Vettiger, E. Meyer,H. J. Guntherodt, C. Gerber,K. J. Gimzewski. Science, 2000, 288,316–318.

64 R. Berger, E. Delamarche, H. P. Lang,C. Gerber, J. K. Gimzewski, E. Meyer,H. J. Guntherodt. Science, 1997, 276, 2021.

65 (a) D. Lange, C. Hagleitner, O. Brand,H. Baltes. Proc. Transducers ’99, Sendai,Japan, 1999 1020–1023, (b) C. Hagleitner,D. Lange, O. Brand, A. Hierlemann,H. Baltes. Digest of Technical Papers,IEEE International Solid-State CircuitsConference (ISSCC), San Francisco, USA,2001, 246–247, ISBN 0-7803-6608-5.

66 S. S. Lee, R. M. White. Sens. Actuators A,1996, 52, 41–45.

67 D. Lange. D. Dissertation ETHZurich, 2000.68 A. W. Van Herwaarden, P. M. Sarro. Sens.

Actuators, &hf,1986, 10, 321–346.69 P. M. Sarro, A. W. van Herwaarden,

W. van der Vlist. Sens. Actuators A, 1994,41–42, 666–671.

70 A. W. Van Herwaarden, P. M. Sarro,J. W. Gardner, P. Bataillard. Sens.Actuators A, 1994, 43, 24–30.

71 A. Koll, A. Schaufelbuhl, O. Brand,H. Baltes, C. Menolfi, H. Huang. Proc.IEEE Workshop on Micro ElectroMechanical Systems MEMS 99, Orlando,1999 547–551, ISBN 0-7803-5194-0.

72 N. Kerness, A. Koll, A.Schhaufelbuehl,C. Hagleitner, A. Hierlemann, O. Brand,H. Baltes. Proc. IEEE Workshop onMicro Electro Mechanical SystemsMEMS 2000, Myazaki, Japan, 2000, 96–101,ISBN 0-7803-5273-4.

73 C. Menolfi, Q. Huang. IEEE Journal SolidState Circuits, 1997, 32, 1–9.

74 J. Lerchner, J. Seidel, G. Wolf, E. Weber.Sens. Actuators B, 1996, 32, 71–75

75 D. Caspary, M. Schropfer, J. Lerchner,G. Wolf. Thermochimica Acta 1999, 337,19–26.

76 J. Lerchner, A. Wolf, G. Wolf.. J. Thermal.Anal., 1999, 55, 212–223.

77 P. Bataillard, E. Steffgen, S. Haemmerli,A. Manz, H. M. Widmer. Biosensors,Bioelectronics, 1993, 8, 89–98.

78 J. M. Kohler, E. Kessler, G. Steinhage,B. Grundig, K. Cammann. Mikrochim.Acta 1995 120, 309–319.

79 P. M. Sarro, H. Yashiro, A. M. van Her-waarden, S. Middelhoek. Sens. Actuators,1988 14, 191–201.

80 A.W. Van Herwaarden. Meas. Sci. Technol.,1992, 3, 935–937.

81 C. Hagleitner, A. Hierlemann D. Lange,A. Kummer, N. Kerness, O. Brand,H. Baltes. Nature, 2001, 414, 293–296.


82 I2C� is a communication standard developedby Philips, Eindhoven, The Netherlands.

83 CMOS process as provided by Austria MikroSysteme International, Unterpremstatten,Austria.

84 A. Koll, S. Kawahito, F. Mayer,C. Hagleitner, D. Scheiwiller, O. Brand,H. Baltes. Proc. of SPIE Smart Structuresand Materials 3328, 1998, 223–232.

85 A. Hierlemann, A. J. Ricco, K. Bodenhofer,A. Dominik, W. Gopel. Anal Chem. 2000,72, 3696–3708.

86 A. Krauss, D.Lange, U. Weimar,A. Hierlemann, O. Brand, H. Baltes.Proc. InternationaI Symposium onElectronic Noses, ISOEN, Brighton, 2000,35–36. 40


10

Integrated Electronic Noses and Microsystems for Chemical

Analysis

Julian W. Gardner, Marina Cole

Abstract

There is considerable interest in the miniaturization and mass production of electro-nic noses through the exploitation of recent advances in the emergent field of micro-systems technology. In this chapter we explore the future outlook for integrated elec-tronic noses by first reviewing the different types ofmicrofluidic components that havebeen reported in the literature, such as microchannels, microvalves, andmicropumps.Next, we describe recent efforts to develop microelectronic noses based upon the in-tegration of sensor arrays and smart interfaces. Finally we report upon work in therelated field of micro total analysis systems, in which, for example, a micro gas chro-matograph or a micro mass spectrometer are being fabricated; these physically-basedmicroinstruments may be regarded as a type of micronose and thus in competitionwith integrated electronic micronoses.

10.1

Introduction

The integration of gasmicrosensors and signal processing circuitry is a subject of ever-increasing importance in the chemical sensor community. It offers lower unit costthrough batch production of wafers, smaller device size, better reproducibility, super-ior signal conditioning by less noise generated in the transmission of the sensor sig-nals to the processing electronics, and an improved limit of detection for the wholesensing system. The full integration of gas microsensors and signal processing circui-try has been brought a step nearer with reports of an increase in sensor reproducibilityby the integration of arrays of sensors onto the same substrate [1–4], improvements insensor sensitivity through advances in individual microsensor technologies [5, 6] andfinally the development of novel gas-sensitive materials, for example see Gardner andBartlett [7], and Attard et al. [8]. Many commercial (non-chemical) sensors have beenrealized in recent years through the integration of both the electronic signal processingcircuitry and the sensing part on the same silicon die. Some examples include pressuresensors, ultrasound sensors and gas flow sensors, proximity and temperature sensors


231231

[9–12]. For a review of silicon sensors, readers are referred to Gardner et al. [13]. Thefield of gas sensors is covered in Chapter 9 of this book where a detailed description ofthe design of the single-chip multisensor system comprising four different sensors, aswell as driving and signal-conditioning circuitry, can be found.Recent years have also seen substantial efforts in the development of smart and

intelligent sensor technology. The main advantages of intelligent sensors are theirimproved performance and reliability – achieved through the addition of self-testingand self-diagnostic functions [14]. Emphasis has also been given to the development ofapplication-specific integrated circuits (ASICs) for intelligent sensors. Taner andBrignell [15] have studied the advantages of ASIC technology, which enables intelli-gent devices to deal both with systematic variation in sensor parameters and providesgood solutions for sensor communications.Parallel with the integration of microsensors and signal processing electronics, and

the realization of smart sensor interfaces, sampling and fluid-handling techniqueshave been rapidly developing. Micro flow sensors, micropumps and microvalvesstarted emerging in the late 1980s marking the beginning of the field of micro-fluidics. So far, life sciences and chemistry have been the main application areasof microfluidics in the liquid phase. Considering that sample handling is a criticalarea, which has an enormous influence on the performance of e-noses (see Chap-ter 3), microfluidics should have a significant impact on the future development ofsuperior, integrated electronic nose (e-nose) systems. Microfluidic technology com-bined with smart silicon sensor arrays could lead to the development of cheap,and possibly disposable devices, particularly important in medical applicationssuch as chemical and biological assays.In this chapter, a review of different solid-state sensor systems and smart sensor

interfaces for e-noses will be given together with an overview of existing microfabri-cated components for fluid handling, such as microvalves, micropumps, and micro-

Fig. 10.1 Basic components that make up an integrated e-nose or

chemical analysis system

10 Integrated Electronic Noses and Microsystems for Chemical Analysis232

channels. A block diagram of the basic components that make up an integrated e-nosesystem is shown in Fig. 10.1 and will be described below. The integration of differentcomponents into microsystems and microinstruments will also be discussed, and thefuture outlook concerning sensor arrays, biological assay devices, and neuromorphicsystems will be briefly outlined. Microsystems for chemical analysis based on gaschromatography, mass spectrometry, and optical spectrometry techniques will alsobe reviewed. Finally, a future outlook is given of e-noses and microsystems for che-mical analysis.

10.2

Microcomponents for Fluid Handling

In the early 1990s, microfluidics was established as a general term for the researchdiscipline dealing with fluid transport phenomena on the micrometer scale and fluidiccomponents, devices, and systems built with microfabrication technologies. The ma-jor applications of microfluidics are in the fields of medical diagnostics, genetic se-quencing, drug discovery, and proteomics. This section focuses on microcomponentsfor fluid handling, such as microchannels, microchambers, microvalves and micro-pumps that could be applicable to the development of integrated e-noses and micro-systems for chemical rather than biological analysis.Advancements in photolithography turned the possibility ofminiaturizing analytical

systems into reality. Initially, only simple channels and reservoirs could be made byphotolithography on glass or silicon wafers, and electro-osmosis was the only way tomove liquids. Over the last 10 years, the fabrication of new microfluidic components,such as valves, pressure systems, metering systems, reaction chambers, and detectionsystems, has led towards the development of more complex manufacturing technol-ogies, e.g. deep reactive ion etching (DRIE), andmultiplayer processes such as the five-layer polysilicon Sandia process) and hence the possibility for lab-on-a-chip prototypes[16]. Apart from their use in research, microfluidic devices also have significant com-mercial potential. In 1999, the Systems Planning Corporation, Arlington, VA, releaseda market study on microelectromechanical systems (MEMS) that projected a micro-fluidics market of 3 to 4.5 billion euros by 2003. 30% of this total is split equally be-tween sensors and lab-on-a-chip applications. Another microsystems market studycompleted in 1996 by a task force of the European Commission’s Network of Excel-lence in Multi-functional Microsystems (NEXUS) forecasted a market of Q 2.8 billionfor microstructure-based disposable assay devices alone by 2002.

10.2.1

Microchannels and Mixing Chambers

Microchannels are essential components of microfluidic systems. They provide theconnections between pumps, valves and sensors [17], and they are used as separationcolumns for different types of gas or liquid chromatographs [18, 19]. They also act as


Fig. 10.2 Scanning electron microscope (SEM)

images of several types of microchannels, fabricated

with: (a) bulk micromachining and wafer bonding,

(b) surface micromachining, and (c) buried channel

technology. From Boer et al. [21]


heat exchangers in the cooling of electronic chips [20]. Commonmethods used for thefabrication of microchannels include bulk etching with wafer-to-wafer bonding, bulketching with sealing using low-pressure chemical vapour deposition (LPCVD) materi-als, conventional surface micromachining of channels, imprinting plastic substrates,X-ray LIGA (German acronym: Lithographie, Galvanoformung, Abformung) technol-ogy and DRIE, and channel forming from UV photodefinable SU-8 photoresist usingLIGA (sometimes called UV LIGA). Figure 10.2 shows scanning electron micrograph(SEM) images of several types of microchannels fabricated with bulk micromachiningand wafer bonding (Fig. 10.2a), surface micromachining (Fig. 10.2b), and buried chan-nel technology (BCT) (Fig. 10.2c). One problem with wafer-to-wafer techniques, suchas anodic bonding or direct fusion bonding, is the possible creation of wafer-to-wafermisalignments and the formation of microvoids at the bonding interface, which mayaffect the functional performance of the device. Another difficulty is that electroniccircuitry (e.g. CMOS) cannot be incorporated on the same substrate because of thehigh process temperatures and voltages needed to perform anodic bonds. The useof surface micromachining obviates the need for accurate wafer alignment. In thisapproach, structural parts are embedded in layers of a suitable sacrificial materialon the surface of a substrate. Dissolving the sacrificial material forms a completemicrochannel. By this method microchannels can be fabricated in various differentpassive materials, e.g. silicon nitride, polysilicon, metal, polymer, and silicon diox-ide. One major disadvantage of this technique has been that the vertical dimensionof such channels is restricted by the maximum sacrificial layer thickness that can bedeposited and etched within an acceptable time period. Researchers at the Universityof Twente have proposed BCT as an alternative to conventional bulk and surface mi-cromachining [21]. BCT allows the fabrication of complete microchannels in a singlewafer with only one lithographic mask, and processing on one side of the wafer. Themicrostructures are constructed by trench etching, coating of the sidewalls of thetrench, removal of the coating at the bottom of the trench, and finally etching intothe bulk of the silicon substrate. This method for the fabrication of these deviceswas derived from the SCREAM (single-crystal reactive etching and metallization) pro-cess [22]. The structure can be sealed by the deposition of a suitable layer that closes thetrench. Using the above procedure it is possible to construct cavities, reaction cham-bers or crossing channels. A spiral-shaped channel with a length of 10 m and a dia-meter of 30 lmwas also developed by the same research group for possible applicationas a separation element in gas chromatography.The method of imprinting plastic substrates involves the low-temperature pattern-

ing of plastic substrates using either small diameter wire or a micromachined silicontemplate. Silicon templates have also been used as a negative master tool for fabrica-tion of polymer microchannels and mixers by hot embossing and microinjectionmolding [23]. Templates have also been made in metals and, more recently, dia-mond. Figure 10.3 shows a micrograph of a mixer, which has been replicated on ahot embossed polymethylmethacrylate wafer. The advantage of this method is thatthe resultant channels are robust, easy to fabricate at low cost, and compatiblewith biological fluids (unlike silicon); it also allows the integration of other microflui-dic elements and the sensors but not electronic circuitry. Another technology that


Fig. 10.3 Micrograph of

an embossed mixer (smallest

channel dimensions:

50 lm � 50 lm). From

Greschke et al. [22]

Fig. 10.4 Process steps for fabricating the MWlCs(a) etch channel, (b) form porous silicon (PS), (c)

under-etch PS. From Tjerkstra et al. [24]


could be used is LIGA technology. The principal advantage of the LIGA process is thatmicrodevices can be fabricated with a height-to-width aspect ratio of up to 200, typicallyseveral millimeters in height and 10 lm in width, but the microchannels fabricatedusing this technology are usually made from plastics, metals, and ceramics rather thansilicon. The process is also very expensive because it requires a synchrotron source.Finally, channels can be made using thick UV LIGA. Using this method only thesidewalls of the channel can be formed, so to seal them some method of bondingis required. The advantage is that the thickness of the wall can be easily controlledand a high aspect ratio achieved.Releasing electrochemically formed porous silicon from the bulk silicon substrate

by under-etching at increased current density is another technique that can be used forthe microfabrication of microchannels, in particular multi-walled microchannels(MWlCs) [24]. Figure 10.4 shows the main processing steps for the formation of aMWlC by etching channels in a p-type silicon wafer using LPCVD silicon-rich siliconnitride as the mask material. Figure 10.5 shows an SEM image of a MWlC containingtwo porous layers. To create more robust devices, i.e. to increase the strength of thestructure, microchannels can be fabricated with a porous silicon membrane sus-pended halfway across an etched cavity surrounded by silicon nitride walls.Most of the above methods allow for some integration with sensors, but external

integration with the electronic circuitry is typically used, e.g. hybrid packaging ormulti-chip modules. In order to allow direct integration of sensors, actuators, andother electronics with the microchannels, Rasmussen et al. have proposed two meth-ods for the fabrication of microchannels using the standard CMOS process and simpleand inexpensive post-processing steps [25]. In the first method, shallowmicrochannelsof the order of 0.4 lm are realized by removing surface layers incorporated in a stan-dard CMOS integrated circuit process. Larger channels with depth of 30 to 300 lm canbe fabricated through the second method that employs bulk micromachining techni-ques. Both methods offer the possibility to create a complete smart microfluidic sys-

Fig. 10.5 SEM photograph of a

MWlC containing two porous

layers. From Tjerkstra et al. [24]


tem that possesses integratedmicrofluidic elements, sensors and electronics as shownin Fig. 10.6.

10.2.2

Microvalves

Microvalves are one of the most important building blocks of microfluidic systemsused for fluid flow control. They can be classified in two categories: active valves(with an actuator) and passive check valves (without an actuator).

10.2.2.1 Active Microvalves

An active microvalve consists of a device body that contains the fluid under pressure, avalve seat to modify the fluid flow, and an actuator to control the position of the valveseat. The first reported microvalve was designed as an injection valve for use in inte-grated gas chromatography [26]. It had a silicon valve seat and a nickel diaphragmactuated by an external solenoid. Following this first design, a large number of micro-valves have been designed and reported, and they can be classified on the basis of theactuation method employed. These methods include pneumatic, thermopneumatic,thermomechanic, piezoelectric, electrostatic, electromagnetic, electrochemical, andcapillary force microvalves. Figure 10.7 shows some of these actuating devices.Pneumatic valves have a membrane structure as the valve seat. Although pneumatic

actuation is based upon a very simple principle it requires an external pressure source,which makes the pneumatic valves unsuitable for most compact applications. A lowspring constant is also an important parameter and in order to achieve it thin mem-branes or corrugated membranes have to be designed. Soft elastic materials, such as

Fig. 10.6 Microscope photo-

graphs of a fabricated chip for

surface-micromachining shallow

channels before etch. From

Rasmussen et al. [25]


silicon rubber or Parylene, can be used to realize the low spring constant [27],whereashard materials such as silicon and glass are problematic.Thermopneumatic valves utilize a sealed pressure cavity filled with a liquid. Actuation

is based on the change in volume of the sealed liquid. The phase change from liquid togas or from solid to liquid can also be used if a larger volume expansion is required.The disadvantage of these types of valves is the incompatibility of the technology, sincethe liquid has to be primed, filled, and sealed individually. Thin films of solid paraffinmaterial could be used as an alternative [28] since they could be integrated in the batchfabrication.Thermomechanic microvalves utilize the principle of the conversion of thermal energy

directly into mechanical stress. There are three types of thermomechanic actuators:solid-expansion [29], shape-memory alloy [30], and bimetallic actuators. A bimetallicvalve as shown in Fig. 10.7a, was introduced by Jerman [31] and was one of the firstcommercialized microfluidic components. The valve was designed for a gas flow con-troller, and its actuator consists of a central boss, a circular diaphragm made of bime-tallic materials, and a circular heater. The temperature, controlled by the heater, of thebimetallic structure on the diaphragm was used to vary the force applied to the boss bythe diaphragm so that the gas flow can be adjusted. The valve was designed to controlthe gas volumetric flow-rate from 0 to 90 mL min�1.Piezoelectric microvalves utilize the piezoelectric effect as an actuation principle. The

thin-film piezoelectric actuators do not deliver sufficient force/displacement for valve

Fig. 10.7 Schematics and principles of operation of (a) bimetallic

(from Jerman [31]), (b) electrostatically actuated (from Shoji and Esashi

[41]), and (c) electromagnetic active microvalves (from Yanagisawa et

al. [37])


application, so all reported piezoelectric microvalves use external stack-type piezoelec-tric actuators, bimorph piezocantilevers, or bimorph piezodisks. Bimorph actuatorsare used if larger deflections are required. They either consist of two bonded piezo-electric layers, with anti-parallel polarization [32], or of a piezoelectric layer bonded to anon-piezoelectric layer [33].Electrostatically actuated microvalves have typically been designed with a valve seat in

the form of a cantilever. Figure 10.7b shows a cantilever structure fabricated by surfacemicromachining. The valve seat could also have a form of flexible metal bridge. Such adesign of microvalve was fabricated by Shikida and Sato [34] in order to allow an ade-quate gas flow-rate under low pressure. The valve is suitable for rarefied gas controlsystems but it requires a large actuating voltage of 100 V. In order for a large deflectionto be achieved with a relatively small drive voltage, a combination of the electrostaticforce, buckling effect, and pneumatic force were used. This arrangement was utilizedfor a bistable valve in an implantable drug delivery system [35]. Vandelli et al. havedesigned a microvalve array for fluid flow control with the valve membrane andone electrode made out of a polysilicon layer, whereas the other electrode was bulksilicon fabricated using surface micromachining. The valves were arranged in anarray so that the flow rate could be controlled digitally [36].Electromagnetic microvalve with a valve cap made out of a soft magnetic Ni-Fe alloy

supported by a spring is shown in Fig. 10.7c [37]. It moves vertically in the magneticfield gradient applied by the external electromagnet. This valve was designed as a flowregulator for a high vacuum application. A microvalve activated by a combination ofelectromagnetic and electrostatic forces has also been fabricated [38]. The structureconsists of a gas flow inlet having a counter electrode, a deflectable membrane coatedwith a metal conductor and two permanent magnets. Current pulses of 200 mA and avoltage of 30 V were typically applied for electromagnetic and electrostatic actuation.The response time was below 0.4 ms.There are also some reported examples of so-called electrochemical and capillary-

force valves. Electrochemical valves are actuated gas bubbles generated by the electro-lysis reaction, where the pressure inside the bubble is proportional to the surface ten-sion and the radius of curvature. Capillary-force valves use capillary-force actuationwhere the surface tension and the capillary force can be controlled actively or passivelyby different means: electrocapillary, thermocapillary, and passive capillary. The use ofsome of these effects are described in two published papers [39, 40].

10.2.2.2 Passive Microvalves (Check Valves)

This type of microvalve is typically designed for use in micropumps where a very smallleakage under reverse applied pressure and a large reverse-to-forward flow resistanceratio is required. The dimensions of check valves are small in comparison with thevalves with integrated or external actuators. A review paper by Shoji and Esashi de-scribes a wide variety of check valve structures [41]. A typical cantilever-type structureis shown in Fig. 10.8. Oosterbroek et al. have reported on the design, simulation, andrealization of in-plane operating passive microvalves [42]. Figure 10.9 shows func-tional, art, and SEM impressions of a 2 � 5 high-density in-plane check valve array.


10.2.3

Micropumps

Micropumps can be classified into two categories: mechanical pumps with movingparts and non-mechanical without moving parts. Mechanical pumps can be furtherdivided into three major groups: reciprocating, peristaltic, and valve-less rectificationpumps. Non-mechanical pumps are mainly used to move liquids and constituents inliquids, which need to avoid any moving mechanical parts. Electrohydrodynamic,electro-osmotic, or ultrasonic effects are normally employed in the operation ofnon-mechanical micropumps.

10.2.3.1 Mechanical Micropumps

A reciprocating-type micropump consists of a pump pressure chamber with a flexiblemembrane driven by an actuator unit and passive microvalves (check valves). Twomain conditions have to be satisfied in order for reciprocating micropumps to func-tion correctly. First, the minimum compression ratio of micropumps (the ratio be-tween the stroke volume and the dead volume that causes serious constraints inmost micropump designs) has to be determined and secondly the pump pressurehas to be high. For the gas micropumps that are of interest in integrated e-nose ap-

Fig. 10.8 Schematic of cantilever-type passive (check-valve)

microvalve. From Shoji and Esashi [41]

Fig. 10.9 (a) Functional (b) art, and (c) SEM impressions of a

2 � 5 high-density in-plane check valve array. From Osterbroek

et al. [42]


plications, the criterion for the minimum compression ratio e is given by Richter et al.[43]:

e >p0

p0 � jDpcritj

� �1=c

�1

where p0 is atmospheric pressure, Dpcrit is the critical pressure required for the ope-ning of the check valve, and c is the adiabatic coefficient (for air c takes a value of 1.4).(The above equation could be simplified at small critical pressures and low pumpfrequencies where there is isothermal behaviour.)The maximum output pressure of the micropump depends directly on the available

force of the actuator used. The different types of actuators that have been employed todate are piezoelectric, pneumatic, electrostatic, and thermopneumatic. The first piezo-electric micropump (Fig. 10.10), developed and reported in 1988 by van Lintel et al.[44], was made in silicon and used a piezoelectric disk. Since then, various differenttypes of micropumps have been designed in order to satisfy the two requirementsmentioned above. Over the years the dead volume of the pump chamber has becamesmaller, and the check valves and the pump membranes have been made out of softermaterials with low spring constants. An example of a self-priming micropump [45]with a small dead volume able to pump gas is shown in Fig. 10.11. Here the deadvolume was minimised by minimising the dead volume of the valve unit down to500 nL, hence increasing the compression ratio to 0.111. In another development[46] van Lintel’s original design was improved to achieve a compression ratio of1.16. Another way to achieve improvements in the micropump design is throughthe design of flexible check-valves. The valves, in the forms of cantilever, havebeen used with integrated electrostatic [47] or bimorph piezoelectric disks [48] as ac-tuators. Materials that are more flexible than silicon, such as polyimide, polyester, andparylene, have also been used in the design of flexible check-valves. A thermo-pneu-matically driven micropump was fabricated using the LIGA process [49]. In this thepump case is made by injection moulding of polysulfone (PSU) and the pump cham-ber is covered by a polyimide membrane. A similar design was fabricated by plasticinjection moulding and uses a polyester valve [50] whereas the design reported in [51]has a pump membrane made out of silicone rubber, and the disk valve from parylene

Fig. 10.10 Piezoelectric disk reciprocating micropump. From van Lintel et al. [44]


deposited through CVD process. Maximum flow rates reported for reciprocating typemicropumps range from 4 to 13 000 lL min�1.Functionality of peristaltic micropumps is based on the peristaltic motion of the pump

chambers, and theoretically, peristaltic pumps should have three or more pump cham-bers with reciprocating membranes. These types of pumps do not require passivevalves for the flow rectification, nor do they require a high chamber pressure, sothe two main conditions that have to be satisfied are a large stroke volume and a largecompression ratio. Three types of actuation principles have been employed in reportedperistaltic micropumps: piezoelectric, electrostatic and thermopneumatic along withseveral types of fabrication processes, such as bulk micromachining, surface micro-machining, plastic moulding, and a combination of bulk micromachining and anodicbonding. Maximum reported flow rates range from 3 to 30 000 lL min�1 (for air).Examples include a surface micromachined pump with electrostatic actuators [52],a thermopneumatically driven micropump having three active pressure chamberswith flexible membranes [53], a micropump with curved pump chambers and a flex-ible plastic membrane with electrostatic actuation [54]. A new pumping principlecalled dual-diaphragm pump, which consists of two actuating membranes in thepump chamber, is reported by Cabuz et al. [55]. This type of pump is able topump up to the maximum reported flow-rate 30 mL min�1 of air.Valveless rectification micropumps are similar to check-valve pumps except that, in-

stead of check-valves, diffusers/nozzles are used for the flow rectification. In orderto optimize the valveless pump designs, the stroke volume has to be maximized whilethe dead volume has to be minimized. The first piezoelectric micropump using noz-zle/diffuser elements instead of check-valves [56] was presented in 1993. The originalvalve was fabricated in brass using precision machining. The same research group in1997 presented the first valveless diffuser pump [57] fabricated using DRIE fabricationprocess shown in Fig. 10.12. The maximum pump pressure was 74 kPa and the max-imum pump volumetric flow-rate was 2.3 mL min�1 for water. Two different thermo-plastic replication methods for the fabrication of valveless pumps: hot embossing andinjection moulding have also been tested [58]. Deep precision-milled brass mouldinserts and deep microelectroformed nickel mould inserts defined from DRIE siliconwafers were used for these designs. Figure 10.13 shows the design of the precision-

Fig. 10.11 Check-valve self-priming micropump. From Linneman et al. [45]


milled brass mould insert pump. Tested pumps had a maximum volumetric flow-rateof 1.2 mL min�1. Nguyen and Huang [59] have demonstrated the design of miniaturevalveless pumps based on a printed circuit board technique. The pump could be op-erated as a single diffuser/nozzle pump (or a peristaltic pump) and has a maximumflow-rate of 3 mL min�1.

Fig. 10.12 Fabrication process of the DRIE diffuser

pump. From Olsson et al. [57]

Fig. 10.13 (a) Top layout view of single-chamber

diffuser pump, and cross-sectional views of (b) a

single-chamber pump unit, and (c) two pump units

stacked in parallel arrangement for high pump

flow. From Olsson et al. [58]


10.2.3.2 Nonmechanical Micropumps

Nonmechanical pumps are based on non-mechanical pumping principles and candriven by capillary-force, thermal, chemical, electrical or magnetic means. Micro-pumps using electrohydrodynamic and electrokinetic effects have been reported.Electrohydrodynamic (EHD) actuation is based on electrostatic forces acting on di-

electric fluids, such as organic solutions. Two main types of electrohydrodynamicmicropumps have been reported: the EHD induction [60] and the EHD injectionpumps [61].Electrokinetic micropumps use the electrical field for pumping conductive fluid. They

can be divided into two categories: pumps based on the electrophoresis effect andpumps based on the electro-osmosis effect. Electrophoresis can be described as themotion of charged particles under an electric field in a fluid relative to the unchargedfluid molecules. The velocity of the charged species is proportional to the applied elec-trical field. Electrophoresis pumps have their application in processes such as theseparation of large molecules such as DNA or proteins [62]. The separation, per-formed in microchannels, is called capillary electrophoresis. Electro-osmosis is theeffect of pumping fluid in a channel under an applied electric field. Changing theapplied electric field or the pH of the solution that affects the potential arisingfrom the charge on the channel wall can control the electro-osmotic flow velocity.In microanalysis systems, the electro-osmosis effect is used for delivering buffer solu-tions [63] and, when combined with the electrophoretic effect, for separating out dif-ferent molecules. The drawbacks of electro-osmosis are that it cannot be used whenseveral interconnected channels are required for sample processing, and it is not com-patible with high-ionic-strength buffers.Information on various other types of micropumps, such as surface-tension driven

pumps, magnetohydrodynamic pumps, and ferrofluidic magnetic pumps can also befound in the literature.

10.3

Integrated E-Nose Systems

10.3.1

Monotype Sensor Arrays

The performance of integrated e-nose systems largely depends upon the performanceof the sensor array used. The integration of the sensor array on to the same substrateoffers a reduction in sensor variation and also improves device reliability. Other ad-vantages include reduced fabrication cost, smaller dimensions and lower power con-sumption. The majority of sensor arrays reported to date are monotype and can bedivided into several categories based upon either the sensor material or type em-ployed, such as conducting polymer, tin oxide, quartz resonator, surface acousticwave (SAW) and FET sensor arrays. Readers are referred to Chapter 4 for a full de-scription of the different types of chemical sensors.


Conducting polymers are very attractive for gas-sensing applications because of theirease of deposition, the large variety of available polymer combinations, and their abilityto operate at room temperature [64, 65]. Neaves andHatfield [66, 67] reported on one ofthe first ASIC chips for conducting polymer sensor arrays. In this, the integrated sen-sor array consists of 64 parallel gold electrodes (forming 32 resistive sensors) and aground plane fabricated on a ceramic substrate. Each conducting polymer sensor con-sists of a two-layer composite: the poly(pyrrole) as the base polymer and a secondpolymer grown electrochemically onto this base. Figure 10.14 shows a micrographof a resistance-measuring ASIC used in the final design of the integrated nose systemfor interrogation of 32 conducting polymer resistors. Another sensor array consistingof five conducting polymers in a microbridge configuration has been developed atWarwick University [68]. The microbridge arrangement was used in order to reducethe temperature dependence of the discrete polymer elements. The array, which hasbeen fabricated in a standard 0.8 lmCMOS technology, is shown in Fig. 10.15. P-typesilicon (or metal electrodes) is used to form resistors for either electrochemical deposi-tion of polymers or spray coating of carbon-black polymer compositematerials, such asthose reported by Lewis [69] (Caltech). These polymer materials can also be depositedon co-fired ceramic substrates, glass slides, and silicon. Thesematerials have also beenused for other micromachined gas sensor arrays. Zee and Judy [70] reported two typesof devices, bulk micromachined and patterned thick-film sensors. The microma-chined, so-called ‘wells’, have been designed to contain the liquid volume during de-position. This type of design permits good reproducibility in the deposition and createslarger exposure areas for sensing while minimizing the chip area. It also allows for theintegration of on-chip electronics for signal conditioning and processing. Figure 10.16shows photographs of micromachined gas sensor arrays with polymer carbon blackcomposite materials.

Fig. 10.14 Photomicrograph

of a resistance-measuring ASIC

used in the design of the inte-

grated nose system for interro-

gation of 32 conducting polymer

resistors. From Neaves and

Hatfield [67]


A novel ChemFET sensor array reported by Covington et al. [71] also uses carbon-black composite polymer materials. A linear dependence to toluene concentration andsensitivity of up to 2.8 lVppm�1 was observed. The device comprises an array of fourn-type enhanced MOSFET sensors, fabricated by the Institute of Microtechnology(IMT, University of Neuchatel, Switzerland). Briand et al. [72] first reported on thesedevices as catalytic field-effect gas sensors in 2000. In this case, three of the MOSFETshad their gate covered with thin catalytic metals and were used as gas sensors, whilethe fourth one had a standard gate covered with nitride and acted as a reference. Sen-sitivities to the gases hydrogen and ammonia were tested.Most of the reported integrated gas array sensors are based on tin oxide technology.

Gardner et al. [73] reported on an array of six MOS odour sensors on single siliconsubstrate with six separate integrated heaters in 1995. A sensor array reported byDas et al. [74] consists of four integrated thick-film tin-oxide gas sensors. The arraywas fabricated on a single substrate and the sensor responses to different concentra-tions of various alcohols and alcoholic beverages were reported. Another micro-machined tin oxide gas sensor array composed of three different devices on thesame rectangular membrane and working at different temperatures was used forthe detection of NO2, CO, and toluene [75]. A sensor array consisting of 40 monolithicsensor elements with different sensitivities achieved by gradient techniques was usedfor halitosis analysis [76]. Forty-one parallel platinum strips partitioned the surface ofthe device into 40 gas-sensitive segments (SnO2 and WO3 were used). Four heatingelements were based on the reverse side and the sensitivity of the array to malodourcomponents was tested.Flexural plate wave (FPW) sensors and SAW devices have both been used as the

elements for analytical sensor systems. In both sensors acoustic waves are generatedwithin a piezoelectric substrate that has usually been coated with a chemically sensitivefilm. A pair of interdigital transducers is normally used to generate and receive acous-tic waves. The difference between these two types of devices is that the active region ofthe FPW sensor is the membrane of thickness much smaller than the acoustic wave-

Fig. 10.15 Photograph of a

five-element CMOS gas sensor

[68]


length. SAW sensor arrays have often been used and reported in gas analysis instru-ments [77, 78] but rarely as integrated arrays. Baca et al. [79], from the Sandia NationalLaboratories (USA), have reported on the development of a GaAs monolithic surfaceacoustic wave integrated circuit (Fig. 10.17) aimed at chemical sensing applications. Aprototype instrument describing an integrated array of six polymer-coated FPWs usedtogether with an absorbent pre-concentrator is reported by Cai et al. [80]. Each FPWmembrane is a layered composite, 5-lm thick, consisting of a silicon nitride layer, apolished layer of p-doped polysilicon, and a ZnO piezoelectric layer attached periph-erally to the silicon substrate. The whole system is shown in Fig. 10.18. Responses tothermally desorbed samples of individual organic solvent vapours and binary and tern-ary vapour mixtures are reported. Another example of an integrated FPW sensor arrayhas been recently reported by Cunningham et al. [81] of the Draper Laboratory. Theyhave designed a chemical-vapour detection and biosensor array based on microfabri-cated silicon resonators (FPW sensors) coated with thin-film polymer sorption layers.The devices were fabricated on silicon-on-insulator (SOI) wafers and the work was aninitial step towards the development of a large multi-element FPW array with severalhundred devices operating within a single silicon chip.Bulk acoustic wave sensors, in particular the thickness mode quartz-crystal micro-

balances (QCM), have also been used in e-nose applications but not as integratedmicrosensor arrays. A monolithic sensor array based on six elements integrated onthe same quartz crystal designed for monitoring agricultural emissions was reported

Fig. 10.16 Photograph of gas

sensors with polymer-carbon

black films deposited: (a) on a

custom built low-temperature

co-fired ceramic substrate and

(b) on a micromachined chip

attached to the ceramic sub-

strate. From Zee and Judy [70]


by Boeker et al. [82]. The dimensions of the quartz substrate are 12 mm � 20 mm edgelength and 168 lm thickness with resonant frequency of 10 MHz.Optical sensor arrays using image processing are another attractive technique for

application in e-nose systems (see Chapter 8). A fibre-optic bead-based sensor arrayhas been designed at Tufts University and employed to discriminate between differ-ent odours [83]. The system incorporates high-density arrays of micrometer-scale op-tical fibres, with polymer beads doped with fluorescent dyes placed at the end of eachfibre. The binding of vapor molecules to the polymers changes the light emitted fromthe dyes, forming a colour signature. A similar technique has been used for the char-acterization of multicomponent monosaccharide solutions. In this, a chip-based sen-sor array composed of individually addressable polystyrene-poly(ethylene glycol) andagarose microspheres has been used. The microspheres are arranged in anisotropi-cally etched cavities that are designed to serve as miniaturized reaction vessels andanalysis chambers (Fig. 10.19). Identification of analytes takes place through colori-metric and fluorescence changes to receptor and indicator molecules, which are cova-lently attached to termination sites on the polymeric microspheres [84]. Photomechan-

Fig. 10.17 Micrograph of a

monolithic GaAs SAW integrated

circuit. From Baca et al. [79]

Fig. 10.18 Photograph showing an integrated

array of six polymers coated flexural plate wave

sensors (FPWs) with an absorbent pre-concen-

trator (PCT). From Cai et al. [80]


ical chemical microsensors reported by Datskos et al. [85] should also bementioned. Inthis work it was demonstrated that photo-induced bending of microcantilevers de-pends on the number of absorbed molecules on their surface. The authors claimthat by choosing different wavelengths tuneable chemical selectivity could beachieved. Apart from identification, a real time visualisation of gas/odour flow hasalso been studied (see Chapter 16 on odour tracing). A portable homogeneous gassensor array was used to visualize the flow of a target gas, and the direction of thegas source was estimated using a real-time image-processing algorithm [86].Finally, silicon-based microelectrode arrays for chemical analysis have been re-

ported. An array consisting of various electrode shapes and sizes designed andused for a systematic study on some aspects of electrochemical sensing (i.e. influenceof electrode geometry) was reported by Schoning et al. [87]. Sensor arrays with differ-ent electrode geometries have been studied at Warwick University for organic crystals,metal oxide, and polymer resistive devices [88–90] and offer certain functional im-provements, such as faster responses or higher common-mode rejection ratios.

10.3.2

Multi-type Sensor Arrays

A study on the advantage of hybrid modular systems over monosystems, aimed at thepossibility of achieving optimum discrimination power of an e-nose system, has beenconducted by Ulmer et al. [91] at Tuebingen University. The system used for thiscomparative study consisted of 16QCMs and MOSs. The results suggest that when-ever high reliability and a high degree of reproducibility and separation power arerequired in the analysis of a complex gas matrix hybrid modular systems shouldbe used. Another hybrid instrument, designed by Dyer and Gardner [92] at WarwickUniversity, employs both resistive and piezoelectric sensors in arrays with improved

Fig. 10.19 Microspheres ar-

ranged in anisotropically etched

cavities designed as miniaturi-

zed reaction vessels and analysis

chambers. From Goodey et al.

[84]


dynamic characteristics, and agrees with the above findings. High-precision program-mable interface circuitry was developed for this system and a resolution of 0.05% wasachieved. Recently, two examples of multi-type sensor arrays have been reported. Thefirst one consisting of four different sensors designed at ETH Zurich has been de-scribed in Chapter 9. The second system is a result of collaboration between Cam-bridge and Warwick Universities. In this, an integrated smart sensor has been devel-oped consisting of two types of devices, chemo-resistive gas sensors and microcalor-imeteric devices with active microFET heaters and temperature sensors on an SOImembrane [93]. The smart SOI sensors can operate at temperatures up to 350 8Cand offer excellent, uniform thermal distribution over the sensing area.A method for selecting an optimum sensor array has been suggested by Chaudry et

al. [94]. A step-wise elimination procedure, which ranks the inclusion of sensors in anarray according to their contribution to the overall sensitivity and selectivity values, wasadopted in this study. Various other techniques could be used to optimize sensor arrayresponse through either smart sensor interfacing [95] or smart signal processing (i.e.adaptive thresholding for improving selectivity or signal processing for improving gassensor response time using analogue VLSI) [96, 97]. A combination of microfluidictechnology, sensor arrays, smart sensor interfacing and signals processing shouldresult in the development of superior e-nose systems and they may, perhaps, be com-parable to the conventional chemical analysis microsystems currently being devel-oped. These micro, total analysis systems (lTAS) are described in the next section.

10.4

Microsystems for Chemical Analysis

10.4.1

Gas Chromatographs

Chromatography is a popular analytical tool commonly employed by chemists to ana-lyze liquid and gas mixtures. Figure 10.20 illustrates the basic components of a typicalgas chromatograph (GC) [98], namely, a carrier gas bottle, an injection port, a longseparation column through which the gas components pass down, a detector, anda data processing system. The components in the gas mixture are separated out be-cause the column is either coated or packed with a stationary-phase film that absorbsthe different components to differing degrees. Consequently, the components traveldown the tube at different rates depending on their specific sorptive properties, and

Fig. 10.20 Basic set-up of a gas chromatography system used to analyze gas mixtures


hence are partitioned out. Ideally, the components will be totally separated out in timewhen they emerge from the column and hence can be measured by a single detector.The ideal graph is illustrated in Fig. 10.21 with five major components clearly visible.This technique is widely used and a description can be found of it in most analyticalchemistry books. However, GC systems tend to be bulky, fragile and expensive itemsof equipment with limited sensitivity.Gas chromatography has been used in olfaction to help analyze complex odours with

only limited success. They can be used to separate out fairly large concentrations ofcertain organic components for which specific coatings (stationary phases) exist. GCsare also used as the front-end of an olfactometer with a person sniffing the output,instead of the sensor, and recording the specific notes as they emerge. This so-calledGC olfactometer can help organoleptic panels identify the presence of certain notes incomplex odors. In fact, there is some evidence that the human olfactory system gen-erates its own spatio-temporal sorption patterns in the olfactory mucosa and so is itselfa type of GC [99].The first attempts to make a micromachined version of a GC were initially reported

as long ago as 1975 by Terry at Stanford University (USA). The separation column wasmade from the isotropic wet etching of a silicon wafer. Figure 10.22 shows a cross-section of the device reported later in 1979 [100] with a pyrex glass lid. The systemincluded a sample injector (silicon valve) and integrated thermal conductivity sensorbut not the air supply. From 1975 to 1998 this research group further developed themicro GC and a recent review of the field has been published by Kolesar et al. [101].Figure 10.23 shows a photograph of a micromachined GC column that is 10 lm deep,300 lm wide, and 0.9 m to 1.5 m in length. In this case copper phthalocyanine has

Fig. 10.21 Ideal gas chromatograph in which all the com-

ponents of a chemical mixture are separated out and appear

as distinct peaks in the time trace

Fig. 10.22 Cross-section of a

micromachined GC unit sho-

wing an integrated thermal

conductivity sensor at the end of

the silicon column. From Terry et

al. [100]


been sputtered down to act as the stationary phase sensitive to reactive gases. Themicro column was shown to separate mixtures of ammonia and nitrogen dioxidein air.In 1997 a Japanese group led by Hannoe et al. [102] (Japanese Integrated Informa-

tion and Energy Systems Laboratories) reported on the use of an ultrasonic etchingtechnique to produce the micro channel, again with a pyrex lid, but this time a PCTFE-sputtered thin-film coating. The GCmicro column has the dimensions of 10 lmdeep,100 lm across and 2 m long. Then Wiranto et al. [103], an Australian group, isotro-pically etched a GC column again with a pyrex lid; this time the column was 20 lmdeep, 200 lm wide and only 125 cm long.The problem associated with the etching of deep channels was solved in the late

1990s with the advent of the DRIE process, and so it is now possible to make microGC columns more accurately and with superior properties. Perhaps the most sophis-ticated system is that being developed by Matzke et al. [104] of the Sandia Laboratoriesusing a plasma-etched (DRIE Bosch process) pyrex lid. The GC column is part of whatis referred to as the ChemLab and Fig. 10.24a shows the schematic arrangement of thischip. The columns are now 200 to 400 lm deep with width of 10, 40 and 80 lm andlengths of only 10, 30 and 100 cm. The group plans to microfabricate a pre-concen-trator and pump thus making the entire instrument on a chip as shown in Fig. 10.24a.It is also possible to try and simplify the integration process through the combinationof electrophoresis to pump the mobile phase, and chromatography to separate withstationary phases, a method called micro capillary electrochromatography [105]. How-ever this technique is mainly suitable for a liquid mobile phase, and requires a highvoltage supply which are incompatible with standard integrated circuit processes.

Fig. 10.23 SEM of the cross-section of a silicon micromachined GC column [101]


Fig. 10.24 (a) Schematic layout of the ChemLab (Sandia Laborato-

ries, USA). The chip is envisaged to be the size of a dime coin,

(b) commercial portable ‘micro’ GC called the Chrompack and used

widely for environmental gas analysis


GC is a useful analytical tool for chemists and there are a number of companies thatmake portable micro GC with somemicromachined parts in them – but still about thesize of a computer tower case. For example, the company ASTmake a battery-operatedunit that contains two micro GC columns with a silicon micromachined thermal con-ductivity sensor. Similar units are made by MTI (see Fig. 10.24b) and by ChrompackInternational; these so-called Chrompack units are widely used to analyze the air fororganic pollutants [106]; modifications to this basic unit by Tuan et al. [107] have alsobeen reported that seek to enhance its basic performance. However, all of these microGCs have some major drawbacks as regards analyzing complex odours. Firstly, thetime it takes for the odorant components to travel down the columns and partitioncan be tens or even hundreds of seconds and, secondly, the separation for some im-portant classes of odours is relatively poor. However, there are two other analyticaltools used by chemists alongside the GC, namely, the mass spectrometer and theoptical spectrometer that may be regarded as complementary techniques. We shallnow discuss them in turn.

10.4.2

Mass Spectrometers

The composition of a liquid (or vapour) can be analyzed using a mass spectrometer(MS). Figure 10.25 shows the general layout of an MS in which the sample is injectedin to themobile phase (normally helium gas) and themolecules ionized [108]. The ionsare first accelerated in a vacuum by applying a voltage and finally separated by a mag-netic field according to the ratio of their mass to charge. The number of ions is countedfor each particular mass (the ions are usually singly charged species) using an iongauge and this is commonly referred to as the abundance. The magnetic sectorcan be replaced by either a quadrupole electrostatic lens or a time-of-flight elementto produce a more compact unit. Indeed a quadrupole mass spectrometer is now mar-keted by Agilent Technologies Inc. (USA) as the Chemical Sensor (Agilent 4440) and

Fig. 10.25 Layout of a magnetic sector mass spectrometer.

From Gardner and Bartlett [108]


comprises a headspace autosampler connected up to a quadrupole mass spectrometerunit and a PC for data analysis (see Fig. 10.26). This unit has been used to analyzevarious odorant problems and Fig. 10.27 shows the mass spectra for a complex odourgenerated from the headspace of a bacterial sample and covers amass range from 45 to550 Daltons [109]. As can be seen, themass spectra for natural odours is complex and apattern recognition system is needed to analyze the differences. In this example, alinear technique such as discriminant function analysis was able to resolve the differ-ences between the growth phases.The MS, like the GC, is a fairly large, heavy, and expensive instrument. Recent

efforts have been made to miniaturize parts of the MS, such as the quadrupolelens and the sampling orifice, using various micromachining techniques. For exam-ple, Fig. 10.28a shows a miniature quadrupole lens system produced by Syms et al.[110] in 1996 together with a more recent version reported by Friedhoff [111] in

Fig. 10.26 Photograph of the

Chemical Sensor (a quadrupole

mass spectrometer sold by Agi-

lent Technologies)

Fig. 10.27 Mass spectra for the headspace of the bacteria E. coli

when in two of its phases of growth. From de Matos et al. [109]


Fig. 10.28 Micro mass spectrometer: (a) schematic parts of a quadrupole

electrostatic lens, (b) photograph of a micro quadrupole lens (from Syms et al.

[110]), and (c) mass spectrum from a micro mass spectrometer


1999 (Fig. 10.28b). These relatively crude microsystems are capable of separating out asmall number of different light masses as shown in Fig. 10.28c.Further advances are being made in the development of micro-injection ports for

micro-MS instruments but the challenges associated with making a miniature ion-source detector and a vacuum system are significant. Nevertheless an integratedmicrofluidic-tandem MS has been reported by Figeys et al. [112] for the analysis ofprotein and peptide masses (in solution). This instrument examines the highermasses of 500 to 1,000 Daltons and is aimed at analyzing biological systems at theprotein level rather than odours – by definition, odours have lower weights otherwisethey are not volatile.Finally, it should also be noted that there are clearly many examples in which mo-

lecules of the same mass have quite different odors. For example, the position of aketone group in undecanone causes the smell to change from fruity to rue-like. Simi-lar changes occur when comparing cis and trans isomers of unsaturated compounds.The other examples of this phenomenon may be found in Chapter 1. Consequently,there is no simple mass-activity relationship for odours. The situation is further com-plicated by the fact that the ionization of a single fragile odorantmolecule can lead to itsfragmentation and so the mass spectrum is more complicated. Consequently, themass spectra should really be considered as a chemical signature rather than an ac-curate measure of themass content in the original complex odour, and of course, theremay not be a unique mapping between smell and mass signature. The combination ofa GC followed by an MS instrument is a powerful and sensitive analytical tool but isclearly an extremely large and expensive unit. Making a micro GC-MS would be theultimate challenge!

10.4.3

Optical Spectrometers

Molecules have characteristic modes of mechanical vibration and rotation, and thesecan be detected by looking at the amount of light at different frequencies that is ab-sorbed by the molecules. The technique is called optical spectroscopy and the mole-cules are usually analyzed using light in the UV to IR range. Although the technique isgenerally much less sensitive to odorant molecules than GC or MS, micro spectro-meter integrated circuits are being developed rapidly for the telecommunications in-dustry. For example, Fig. 10.29 shows the principle in which light from a fibre-optic issplit by a deflection grating in to its various frequencies, and these are detected using a256-element CCD array [113]. The combination of these technologies with a micro-fluidic system could lead to a low-cost solution for the screening of simple odours.However, that will require improvements in both the sensitivity of the optical sensorarray and the width of the frequency spectrum.


Fig. 10.29 Optical micro spectrometer integrated circuit chip:

(a) schematic and (b) actual device. From Gardner et al. [113]


10.5

Future Outlook

A nose in its totality comprises a sampling system (i.e. sniffing mechanism), and afluid flow system as well as a distributed sensor array and complex signal processingarchitecture. In this chapter we have described some of the current efforts towardsmaking a micro nose that integrates the sampling and fluidic system with the actualsensing system. It is clear that the integration of a sampling system will improve thereliability and performance of an electronic nose, as it is also evident that the creationof multi-type sensor arrays enhances its dynamic range. The latter may also be furtherextended through the use of biological materials within the sensing elements; howeverthe lack of stability of most biological materials exposed to the environment suggeststhat such bio-electronic noses could only really be used for a very short period of time.This lack of stability creates the need for a micro cassette that holds a sequence ofbiosensors that can be employed at the appropriate time, somewhat analogous to aphotographic film cassette. Based on an analogy with the human olfactory system,the cassette would need to be wound on every 20 or so days.The production of a reliable sampling microsystem will also enable the use of the

dynamical part of the sensor signal, which has been shown to be very useful [114].However, miniaturization of the system is essential so that the time-constants asso-ciated with the physical transport of the odour around the channels and chambers aremuch smaller than the response times of the sensors themselves. This permits thedifferent rate kinetics of the chemical sensors for the different analytes to be ob-served, and thus used to help the discrimination process. The micro channels, microchambers and micro pumps will permit the delivery of odours extremely quickly andreproducibly to the sensor array (or mass filter), and so this should permit the creationof a new generation of dynamical micronoses.Nevertheless, the technological advances that permit the creation of such a physical

embodiment of e-noses will not, in our opinion, be sufficient to solve the more com-plex odour problems. It is difficult to visualize a mass/optical spectrum from a mass/optical spectrometer (miniaturized or not) resolving subtle differences in the head-space of such as cheeses and beverages. Instead, the spatio-temporal information gen-erated by GC-based and/or sensor-based micronoses will require quite different typesof signal processing algorithms from the customary the principal components analy-sis, discriminant function analysis and neural networking methods described in ear-lier chapters. The types of nonlinear dynamical filters that will be required may well beneuromorphic algorithms similar to those used in our human olfactory systems. Con-sequently, the future emphasis will turn from the construction of the miniature hard-ware towards the identification of suitable dynamical models, which could either bedata-driven or parametric. Of course this generic approach is challenging and will leadto integrated noses whose cost may be unacceptable in some application fields. Forinstance, the most likely competitor to e-noses in the medical domain may be dispo-sable biochips that seek specific proteins or protein sequences. It is unlikely that ageneric micronose can compete with such a low-cost screening method. However,there are other biomedical applications in which it is possible to use an e-nose to


screen for whole viable micro-organisms (see Chapter 18) because a protein biochiplacks such a capability.

References

1 J. W. Gardner. Intelligent gas sensingusing an integrated sensor pair, Sensorsand Actuators B, 26–27 (1995), 261–266.

2 A. C. Pike. University of Warwick, UK,1996, PhD Thesis.

3 P. Althainz, J. Goschnick, S. Ehrmann,H. J. Ache. Multisensor microsystem forContaminant in Air, Int. Conf. on SolidState Sensor and Actuators, Transducers’95, Stockholm, Sweden, 1995 pp.699–702.

4 C. L. Johnson, J. W. Schwank, K. D. Wise.Integrated ultra-thin-film gas sensors,Sensors and Actuators B, 20 (1994),55–62.

5 W.Gopel, K. D. Schierbaum. SnO2 sensors:current status and future prospects, Sen-sors and Actuators B, 26–27, (1995) 1–12.

6 E. Souteyrand, D. Nicolas. E. Queau,J. R. Martin. Influence of surface modifi-cations on semiconductor gas sensor be-haviour, Sensors and Actuators B, 26–27,(1995) 174–178.

7 J. W. Gardner, P. N. Bartlett. Device forsensing volatile materials, Patent Appl.No. W093/03355, Feb. 1993.

8 G. S. Attard, P. N. Bartlett, N. R. B. Cole-man, J. M. Elliott, J. R. Owen, J. H. Wang.Nanostructured platinum films fromlyotropic liquid crystalline phases, Science,778 (1997), 838–840.

9 F. V. Schnatz, U. Schoneberg,W. Brockherde, P. Kopystynski,T. Mehlhorn, E. Obermeier, H. Bensel.Smart CMOS capacitive pressure trans-ducer with on-chip capability, Sensorsand Actuators A, 34 (1992) 77–83.

10 C. Kuratly, Q. Huang. A fully integratedself-calibrating transmitter/receiver ICfor an ultrasound presence detectormicrosystem, IEEE Journal of Solid-StateCircuits, 33 (1998), 833–841.

11 P. A. Passeraub, P. A. Besse, A. Bayadroun,E. Bernascony, R. S. Popovic. First inte-grated inductive proximity sensor withon-chip CMOS readout circuit and elec-trodeposited 1 mm flat coil, The 12 Euro.

Conf. on Solid-State Transducers,Southampton, UK, 1998, 575–578.

12 M. Tuthill. A switched-current, switched-capacitor temperature sensor in 0.6 lmCMOS, IEEE Journal of solid-state circuits,33(7) (1998), 1117–1122.

13 J. W. Gardner, V. Varadan, O. O. Awadel-karim. Microsensors MEMS and SmartDevices, J. Wiley and Sons Ltd, Chicester,2001, 503.

14 S. Middelhoek, A. C. Hoogerwerf. Smartsensors: when and where?, Sensors andActuators B, 8, (1985) 39–48.

15 A. H. Taner, J. E. Brignell. Aspects ofintelligent sensor reconfiguration, Sensorsand Actuators A, 47, (1995) 525–529.

16 M. A. Burns, et al.. An integrated nanoliterDNA analysis device, Science, 282 (1998),484–487.

17 M. Elwenspoek, T. S. J. Lammerink,R. Miyake, H. J. Fluitman. Toward inte-grated micro liquid handling systems,Journal of Micromechanics and Micro-engineering, 4, (1994), 227–245.

18 S. T. Terry, J. J. Jerman, J. B. Angell.A gas-chromatographic air analyserfabricated on a silicon wafer, IEEETransactions of Electron Devices, ED-26,(1979), 1880–1886.

19 R. R. Reston, E. S. Kolesar. Jr.. Siliconmicromachined gas chromatography sy-stem used to separate and detect ammoniaand nitrogen dioxide – Part I: Design,fabrication, and integration of the gaschromatography system, IEEE Journal ofMicroelectromechanical Systems, 3,(1994), 134–146.

20 A. Weisberg, H. M. Bau, J. N. Zemel.Analysis of microchannels for integratedcooling, International Journal of Heat andMass Transfer, 35, (1992), 2465–2473.

21 M. J. de Boer, R. W. Tjerkstra,J. W. Berenschot, H. V. Jansen,G. J. Burger, J. G. E. Gardeniers.M. Elwenspoek , A. van den Berg.Micromachining of buried micro channelsin silicon, IEEE Journal of Micro-


electromechanical Systems, 9, (2000),94–103.

22 K. A. Shaw, Z. L. Zhang, N. C. MacDonald.SCREAM I: a single mask, single-crystalsilicon, reactive ion etching processfor micromechanical structures, Sensorsand Actuators A, 40 (1994) 63–70.

23 O. Geschke, W. Rong, P. T. Tang,J. P. Kutter, P. Telleman. Polymer Struc-tures for lTAS, The 12th Euro. Conf.on Solid-State Transducers, Copenhagen,Denmark, 2000, 291–293.

24 R. W. Tjerkstra, J. G. E. Gardeniers.J .J. Kelly, A. van den Berg. Multi-WalledMicrochannels: Free-standing poroussilicon membranes for use in lTAS,IEEE Journal of MicroelectromechanicalSystems, 9, (2000), 495–501.

25 A. Rasmussen, M. Gaitan, L .E. Locascio,M.E. Zaghloul. Fabrication techniques torealize CMOS-compatible microfluidicmicrochannels, IEEE Journal of Micro-electromechanical Systems, 10, (2001),286–297.

26 S. C. Terry, J. H. Jerman, J. B. Angell. A gaschromatographic air analyser fabricatedon a silicon wafer, IEEE Transactionsof Electron Devices ED-26, (1979),1880–1886.

27 C. A. Rich, K. D. Wise. An 8-bit microflowcontroller using pneumatically-actuatedmicrovalves, Proceedings of MEMS’99,the 12th IEEE International WorkshopMicro Electromechanical Systems,Orlando, FL, (1999), 130–134.

28 E. T. Claren, C. H. Mastrangelo. Parafinactuated surface micromachined valves,Proceedings of MEMS’00, the 13th IEEEInternational Workshop Micro Electro-mechanical Systems, Miyazaci, Japan,(2000), 381–385.

29 T. Lisec, et al.. Thermally drivenmicrovalvewith buckling behaviour for pneumaticapplications, Proceedings of MEMS’94,the 7th IEEE International Workshop MicroElectromechanical Systems, Oiso, Japan,(1994), 13–17.

30 H. Kahn, M. A. Huff, A. H. Heuer.The Ti Ni shape-memory alloy and itsapplications for MEMS, Journal of Micro-mechanics and Microengineering, 8,(1998), 213–221.

31 H. Jerman. Electrically-activated micro-machined diaphragm valves, Technical

Digest of the IEEE Solid-State Sensor andActuator Workshop, Hilton Head Island,SC, (1990) 65–69.

32 M. Stehr, et al.. The VAMP – a new devicefor handling liquids or gases, Sensors andActuators A, 57, (1996), 153–157.

33 N. T. Nguyen, et al.. Hybrid-assembledmicro dosing system using silicon-basedmicropump/valve and mass flow sensor,Sensors and Actuators A, 69, (1998),85–91.

34 M. Shikida, K. Sato. Characteristics of anelectrostatically-driven gas valve underhigh-pressure conditions, Proceedingsof MEMS’94, the 7th IEEE InternationalWorkshop Micro ElectromechanicalSystems, Oiso, Japan, (1994), 235–240.

35 B. Wagner, et al.. Micromachined bistablevalves for implantable drug deliverysystem, Proceedings of the 18hth AnnualInternational Conference of the IEEE Eng.In Med. and Boi. Soc., Amsterdam,Netherlands, (1996), 254–255.

36 N. Vandelli, D. Wroblewski, M. Velonis,T. Bifano. Development of a MEMSmicrovalve array for fluid flow control,IEEE Journal of MicroelectromechanicalSystems, 7, (1998), 395–403.

37 K. Yanagisawa, H. Kuwano, A. Tago.An electromagnetically driven microvalve,Proceedings of Transducers ’93, the 7hthInternational Conference on Solid-StateSensors and Actuators, Yokohama, Japan,(1993), 102–105.

38 D. Bosch, B. Heimhofer, G. Muck,H. Seidel, U. Thumser, W. Wesler.A silicon microvalve with combinedelectromagnetic/electrostatic actuation,Sensors and Actuators A, 37–38, (1992),684–692.

39 A. P. Papavasiliou, D. Liepmann,A. P. Pisano. Electrolysis-bubble actuatedgate valve, Technical Digest of the IEEESolid State sensor and Actuator Workshop,Hilton Head Island, SC, (2000), 48–51.

40 P. F. Man, et al.. Microfabricated capillary-driven stop valve and sample injector,Proceedings of MEMS’98, the 11th IEEEInternational Workshop Micro Electro-mechanical Systems, Heidelberg,Germany, (1998), 45–50.

41 S. Shoji, M. Esashi. Microflow devices andsystems, Journal of Micromechanics andMicroengineering, 4, (1994), 157–171.


42 R. E. Osterbroek, et al.. Designing, simu-lation and realization of in-plane operatingmicrovalves, using new etching techni-ques, Journal of Micromechanics andMicroengineering, 9, (1999), 194–198.

43 M. Richter, R. Linnemann, P. Woias.Robust design of gas and liquid micro-pumps, Sensors and Actuators A, 68,(1998), 480–486.

44 H. T. G. van Lintel, F. C .M. van den Pol,S. Bouwstra. A piezoelectric micropumpbased on micromachining in silicon,Sensors and Actuators A, 15, (1988),153–167.

45 R. Linneman, et al.. A self-priming andbubble tolerant piezoelectric siliconmicropump for liquids and gases,Proceedings of MEMS’98, the 11th IEEEInternational Workshop Micro Electro-mechanical Systems, Heidelberg,Germany, (1998), 532–537.

46 D. Maillefer, et al.. A high-performancesilicon micropump for an implantabledrug delivery system, Proceedings ofMEMS’99, the 12th IEEE InternationalWorkshop Micro ElectromechanicalSystems, Orlando, FL, (1999), 541–546.

47 R. Zengerle, et al.. A bi-directional siliconmicropump, Proceedings of MEMS’95,the 8th IEEE International Workshop MicroElectromechanical Systems, Amsterdam,The Netherlands, (1995), 19–24.

48 M. Koch, N. Harris, A. G. R. Evans,N. M. White, A. Brunnschweiler. A novelmicromachined pump based on thick-filmpiezoelectric actuation, Sensors andActuators A, 70 (1998), 98–103.

49 W. K. Schomburg, et al.. Microfluidiccomponents in LIGA technique, Journal ofMicromechanics and Microengineering, 4,(1994), 186–191.

50 S. Boehm, W. Olthuis, P. Bergveld.A plastic micropump constructed withconventional techniques and materials,Sensors and Actuators A, 77 (1999),223–228.

51 E. Meng, et al.. A check-valved siliconediaphragm pump, Proceedings ofMEMS’00, the 13th IEEE InternationalWorkshop Micro ElectromechanicalSystems, Miyazaci, Japan, (2000), 62–67.

52 J. W. Judy, T. Tamagawa, D. L. Polla.Surface-machined micromechanicalmembrane pump, Proceedings of

MEMS’91, the 3rd IEEE InternationalWorkshop Micro ElectromechanicalSystems, Nara, Japan, (1991), 182–186.

53 J. A. Folta, N. F. Raley, E. W. Hee. Designfabrication and testing of miniature peri-staltic membrane pump, Technical Digestof the IEEE Solid State Sensor and ActuatorWorkshop, Hilton Head Island, SC, (1992),186–189.

54 C. Cabuz, et al.. Mesoscopic sampler basedon 3D array of electrostatically activateddiaphragms, Proceedings of Transducers’99, the 10th International Conference onSolid-State Sensors and Actuators, Sendai,Japan, (1999), 1890–1891.

55 C. Cabuz, et al.. The Dual DiaphragmPump, Proceedings of MEMS’01, the 14th

IEEE International Workshop MicroElectromechanical Systems, Interlaken,Switzerland, (2001), 519–522.

56 E. Stemme, G. Stemme. A valveless dif-fuser/nozzle-based fluid pump, Sensorsand Actuators A, 39 (1993), 159–167.

57 A. Olsson, P. Enoksson, G. Stemme,E. Stemme. Micromachined flat-walledvalveless diffuser pump, Journal ofMicroelectromechanical Systems, 6,(1997), 161–166.

58 A. Olsson, O. Larsson, J. Holm,L. Lundbladh, O. Ohman. Valvelessdiffuser micropumps fabricated usingthermoplastic replication, Sensors andActuators A, 64 (1998), 63–68.

59 N.-T. Nguyen, X. Huang. Miniaturevalveless pumps based on printed circuitboard technique, Sensors and Actuators A,88 (2001), 104–111.

60 G. Fuhr, et al.. A micromachined electro-hydrodynamic (EHD) pumps for liquidsof higher conductivity, Journal of Micro-electromechanical Systems, 1, (1992),141–145.

61 A. Richter, et al.. A micromachined elec-trohydrodynamic (EHD) pump, Sensorsand Actuators A, 29 (1991), 159–168.

62 J. R. Webster, et al.. Electrophoresis systemwith integrated on-chip fluorescencedetection, Proceedings of MEMS’00, the13th IEEE International Workshop MicroElectromechanical Systems, Miyazaci,Japan, (2000), 306–310.

63 O. T. Guenat, et al.. Partial electro-osmoticpumping in complex capillary systems.Part 2: Fabrication and application of a


micro total analysis system (lTAS) suitedfor continuous volumetric nanotitrations,Sensors and Actuators B, 72 (2001)273–282.

64 A. C. Pike. University of Warwick, UK,1996, PhD Thesis.

65 J. W. Gardner, P. N. Bartlett. Application ofconducting polymers in microsystems,Sensors and Actuators A, 51 (1995) 57–66.

66 P. I. Neaves, J. V. Hatfield. Current-modemultiplexer for integrating resistive arraysensors, Electronics Letters, 30 (1994),942–943.

67 P. I. Neaves, J. V. Hatfield. A new gene-ration of integrated electronic noses,Sensors and Actuators B, 26–27 (1995)223–231.

68 M. Cole, et al.. Active bridge polymericresistive devices for vapor sensing, Euro-sensors XIV, the 14th European Conferenceon Solid-State Transducers, Copenhagen,Denmark, (2000), 895–898.

69 E. J. Severin, B. J. Doleman, N. S. Lewis.An investigation in the concentrationdependence and response to analyte mix-tures of carbon black/insulating organicpolymer composite vapor detectors, Ana-lytical Chemistry, 72 (2000), 658–668.

70 F. Zee, J. W. Judy. Micromachined poly-mer-based chemical gas sensor array,Sensors and Actuators A, 72 (2001)120–128.

71 J. A. Covington, J. W. Gardner, D. Briand,N.F. de Rooij. A polymer gate FET sensorarray for detecting organic vapours, Sen-sors and Actuators A, 77 (2001) 155–162.

72 D. Briand, B. van der Schoot, N. F. de Rooij,H. Sundgren, I. Lundstrom. A low-powermicromachined MOSFET gas sensor,Journal of MicroelectromechanicalSystems, 9, (2000), 303–308.

73 J. W. Gardner, A. Pike, N. F. de Rooij,M. Koudelka-Hep, P. A. Clerc, A. Hierle-mann, W. Gopel. Integrated chemicalsensor array for detecting organic solvents,Sensors and Actuators B, 26 (1995),135–139.

74 R. R. Das, K. K. Shukla, R. Dwivedy,A. R. Srivastava. Discrimination ofindividual gas/odour using responses ofintegrated thick film tin oxide sensor arrayand fuzzy-neuro concept, MicroelectronicsJournal, 30, (1999) 793–800.

75 C. Cane, I. Gracia, A. Gotz, L. Fonseca,E. Lora-Tamayo,M. C.Horrillo, I. Sayago, J.I. Robla, J. Rodrigo, J. Gutierez. Detectionof gases with arrays of micromachined tinoxide gas sensors, Sensors and Actuators B,65, (2000) 244–246.

76 S. Ehrmann, J. Jungst, J. Goschnick,D. Everhard. Application of a gas sensormicroarray to human breath analysis,Sensors and Actuators B, 65, (2000),247–249.

77 M. Rapp, J. Reibel, A. Voigt, M. Balzer,O. Bulow. New miniaturized SAW-sensorarray for organic gas detection drivenby multiplexed oscillators, Sensors andActuators B, 65, (2000), 169–172.

78 J. Reibel, U. Stahl, T. Wessa, M. Rapp. Gasanalysis with SAW sensor systems, Sensorsand Actuators B, 65, (2000), 173–175.

79 A. G. Baca, E. J. Heller, V. M. Hietala,S. A. Casalnuovo, G. C. Frye-Mason,J. F. Klem, T. J. Drummond. Developmentof a GaAs monolitic surface acoustic waveintegrated circuit, IEEE Journal of Solid-State Circuits, 34, (1999), 1254–1258.

80 Q. Y. Cai, J. Park, D. Heldsinger,M.-D. Hsieh, E. T. Zellers. Vaporrecognition with an integrated array ofpolymer-coated flexural plate wave sensors,Sensors and Actuators B, 62, (2000),121–130.

81 B. Cunningham, et al.. Design, fabricationand vapor characterisation of microfabri-cated flexural plate resonator sensor andapplication to integrated sensor arrays,Sensors and Actuators B, 73, (2001),112–123.

82 P. Boeker, G. Horner, S. Rosler. Monolithicsensor array based on a quartz micro-balance transducer with enhanced sensiti-vity for monitoring agricultural emissions,Sensors and Actuators B, 70, (2000),37–42.

83 K. J. Albert, D. R. Walt, D. S. Gill,T. C. Pearce. Optical multibead arraysfor simple and complex odour discrimi-nation, Analytical Chemistry, 73, (2001)2501–2508.

84 A. Goodey, et al.. Development of multi-analyte sensor arrays composed ofchemically derivatized polymeric micro-spheres localized in micromachinedcavities, Journal of the American ChemicalSociety, 123, (2001), 2559–2570.


85 P. G. Datskos,M. J. Sepaniak, C. A. Tripple,N. Lavrik. Photomechanical chemicalmicrosensors, Sensors and Actuators B, 76,(2001), 393–402.

86 H. Ishida, T. Yamanaka, N. Kushida,T. Nakamoto, T. Moriizumi. Study of real-time visualisation of gas/odour flow imageusing gas sensor array, Sensors andActuators B, 65, (2000), 14–16.

87 M. J. Schoning, et al.. A silicon-basedmicroelectrode array for chemical analysis,Sensors and Actuators B, 65, (2000),284–287.

88 J. W. Gardner, M. Iskandarani, B. Bott.Effect of electrode geometry on gas sensi-tivity of lead phthalocyanine thin films,Sensors and Actuators B, 9, (1992),133–142.

89 J. W. Gardner. Intelligent gas sensingusing an integrated sensor pair, Sensorsand Actuators B, 27, (1995), 261–266.

90 J. W. Gardner, P. N. Bartlett, K. F. Pratt.Modelling of gas-sensitive conductingpolymer devices, IEE Proceedings ofCircuits Devices and Systems, 142, (1995),321–333.

91 H. Ulmer, J. Mitrovics, U. Weimar,W. Gopel. Sensor arrays with only one orseveral transducer principles? The advan-tage of hybrid modular systems, Sensorsand Actuators B, 65, (2000), 79–81.

92 D. C. Dyer, J. W. Gardner. High-precisionintelligent interface for a hybrid electronicnose, Sensors and Actuators A, 62, (1997),724–728.

93 F. Udrea, D. Setiadi, J. A. Covington,T. Dogaru, C.-C. Lu, W. I. Milne. Designand simulations of a new class ofSOI CMOS micro hot-plate gas sensors,Sensors and Actuators B, 78, (2001),180–190.

94 A. N. Chaudry, T. M. Hawkins,P. J. Travers. A method for selecting anoptimum sensor array, Sensors andActuators B, 69, (2000), 236–242.

95 G. C. Cardinali, et al.. A smart sensorsystem for carbon monoxide detection,Analog Integrated Circuits and SignalProcessing, 14, (1997), 275–296.

96 D. M. Wilson, S. P. DeWeerth. Nonlinearpreprocessing for smart chemical sensingsystems, Int. Conf. On Solid-State Sensorsand Actuators, Transducers ’95, Stock-holm, Sweden, 1995, pp. 814–817.

97 D. M. Wilson, S. P. DeWeerth. Signalprocessing for improving gas sensorresponse time, Sensors and Actuators B,41, (1997), 63–70.

98 D. H. Desty (ed.). Gas Chromatography,Butterworths, London (1958).

99 D. E. Hornung, S. L. Youngentob,M. Mozell. Olfactory mucosa/air partitio-ning of odorants, Brain Research, 413,(1987), 147–154.

100 S. C. Terry, J. H. Jerman, J. B. Angell. A gaschromatographic air analyser fabricated ona silicon wafer, IEEE Transactions onElectron Devices, 26, (1979), 147–1886.

101 E. S. Kolesar, R. R. Reston. Review andsummary of a silicon micromachined gaschromatography system, IEEE Compo-nents, Packaging andMachine Technology,21, (1998), 324–28.

102 S. Hannoe, I. Sugimoto, K. Yanagisawa,H. Kuwano. Enhanced chromatographicperformance of silicon-micromachinedcapillary column with clean structure andinteractive plasma organic films, Int. Conf.on Solid-State Sensors and Actuators,Transducers ’97, Chicago, USA, 1997, pp.515–518.

103 G. Wiranto, N. D. Samaan, D. E. Mulcahy,D. E. Davey. Microfabrication of capillarycolumns on silicon, SPIE, 324, (1997),59–64.

104 M. Matzke, et al.. Microfabricated silicongas chromatographic microchannels:fabrication and performance, SPIE, 3511,(1998), 262–268.

105 S. Constantin, R. Freitag, D. Solignac,A. Sayah, M. Gijs. Capillary electrochro-matography chip integrated in cartidge,Proceedings of Eurosensors XIV, Copen-hagen, Denmark, 2000, 287–290.

106 G. Etiope. Evaluation of a micro gas chro-matographic technique for environmentalanalyses of CO2 and C1–C6 alkanes, Jour-nal of Chromatography A, 775 (1997),243–249.

107 H. P. Tuan, et al.. Novel preconcentrationtechnique for on-line coupling to high-speed narrow-bore capillary gas chroma-tography, Journal of Chromatography A,791, (1997), 187–195.

108 J. W. Gardner, P. N. Bartlett. Electronicnoses: principles and applications, OxfordUniversity Press, Oxford, 1999, p60.


109 R. Esteves de Matos, D. J. Mason,C. S. Dow, J. W. Gardner. Investigation ofthe growth characteristics of E. coli usingheadspace analysis, in Electronic Noses andOlfaction, eds. J.W. Gardner and K.C.Persaud, IOP Publishing Ltd, Bristol, 2000,181–188.

110 R. R. Syms, T. J. Tate, M. Ahmad, S. Taylor.Fabrication of a microengineered quadru-pole electrostatic lens, Electronic Letters,32, (1996), 2094–2095.

111 C. B Friedhoff, et al.. Chemical sensingusing non-optical microelectromechanicalsystems, Journal of Vacuum Science andTechnology, 17, (1999), 2300–2307.

112 D. Figeys, S. P. Gygi, G. McKinnon,R. Aebersold. An integrated microfluidic-tandem mass spectrometry systemfor automated protein analysis, AnalyticalChemistry 70 3728–3734.

113 J. W. Gardner, V. K. Varadan, O. Awadel-karim. Microsensors, MEMS and SmartDevices, J. Wiley and Sons Ltd, Chichester,2001, 436.

114 E. L. Hines, E. Llobet, J. W. Gardner.Electronic noses: a review of signal pro-cessing techniques, Proceedings IEEE:Circuits, Systems and Devices, 146, (1999),297–310.


11

Electronic Tongues and Combinations of Artificial Senses

F. Winquist, C. Krantz-Rulcker, I. Lundstrom

11.1

Introduction

The field of measurement technology is rapidly changing due to the increased use ofmultivariate data analysis, which has led to a change in the attitude of how to handleinformation. Instead of using specific sensors for measuring single parameters, it hasinmany cases becomemore desirable to get information of quality parameters, such assample condition, state of a process, or expected human perception of, for example,food. This is done by using arrays of sensors with partially overlapping selectivities andtreating the data obtained with multivariate methods. These systems are often referredto as artificial senses, since they function in a similar way as the human senses. Onesuch system, the electronic nose, has attracted much interest [1–3]. This concept isbased on the combination of a gas sensor array with different selectivity patterns withpattern recognition software. A large number of different compounds contribute to ameasured smell; the chemical sensor array of the electronic nose then provides anoutput pattern that represents a combination of all the components. Although thespecificity of each sensor may be low, the combination of several specificity classesallows a very large number of odors to be detected.Similar concepts, but for use in aqueous surroundings have also recently been de-

veloped. These systems are related to the sense of taste in a similar way as the elec-tronic nose to olfaction, thus, for these systems the terms ‘electronic tongue’ or ‘tastesensor’ have been coined [4–6].In some applications, there are advantages when measuring in the aqueous phase

compared to measurements in the gaseous phase; gas analysis is an indirect methodthat gives the final information about the aqueous phase via measurements in thegaseous phase. Many compounds, such as ions or those having a low vapor pres-sure, can only be measured in the aqueous phase, also for many online or inline ap-plications it is only possible to use systems that measure directly in the solution.Furthermore, the development of electronic tongues offers an intriguing possibilityto study their combinations with other types of artificial senses.


267

In principle, the electronic tongue or taste sensor functions in a similar way to theelectronic nose, in that the sensor array produces signals that are not necessarily spe-cific for any particular species; rather a signal pattern is generated that can be corre-lated to certain features or qualities of the sample. Electronic noses and tongues arenormally used to give qualitative answers about the sample studied, and only in specialcases to predict the concentration of individual species in the sample.Different sensing principles can also be used in electronic tongues or taste sensors,

such as electrochemical methods like potentiometry or voltammetry, optical methods,or measurements of mass changes based on, for example, quartz crystals.The sense of taste may have two meanings. One aspect denotes the five basic tastes

of the tongue; sour, salt, bitter, sweet, and ‘umami’. These originate from different,discrete regions on the tongue containing specific receptors called papillae. This aspectof taste is often referred to as the sensation of basic taste. The other aspect of taste is theimpression obtained when food enters the mouth. The basic taste is then merged withthe information from the olfactory receptor cells, when aroma from the food enters thenasal cavities via the inner passage. This merged sensory experience is referred to asthe descriptive taste by sensory panels.The approach to more specifically mimic the basic taste of the tongue is made by the

taste sensor system [4, 7, 8], in which different types of lipid membranes are used todetermine qualities of food and liquids in terms of taste variables such as sweetness,sourness, saltiness, bitterness, and ‘umami’. There is thus a difference between theuse of a sensor array as electronic tongues or as taste sensors. A taste sensor system isused to classify the different basic taste sensations mentioned above, and the resultsare compared with human test panels. An electronic tongue classifies a quality of oneor another kind in food, such as drinks, water, and process fluids, and the results arenot necessarily compared with human sensations, but with other quality properties ofthe sample.The concept of the electronic tongue and the taste sensor has developed very quickly

during the last years due to its large potential. There are already commercial versionson the market [9, 10], and a number of other applications have also been reported, andare described later.Theperformanceofanartificialsensesuchastheelectronictonguecanbeconsiderably

enhancedbythecombinationofsensorsbasedondifferenttechnologies.Thereasonis,ofcourse, that foreachnewmeasurementprincipleadded,anewdimensionofinformationis also added. A natural extension of this fundamental concept is the combination ofdifferent artificial senses. This is especially important when estimating the quality offood, since the guide is the impression of the human being using all five senses.A first attempt to measure the elusive parameter ‘mouthfeel’ for crispy products

such as potato chips or crispbread was made by the development of an artificialmouth. The intention was to collect information mimicking three human senses: ol-faction, auditory, and tactile. The samples were placed in a special ‘crush chamber’,and, while crushed, information corresponding to three senses could be obtained:‘auditory’ by a microphone, ‘tactile’ by a force sensor, and ‘olfaction’ by a gas sensorarray [11, 12]. Furthermore, combinations of electronic noses and tongues have beenused for quality estimation of different wines [13, 14].

11 Electronic Tongues and Combinations of Artificial Senses268

A new dimension for the assessment of human-based quality evaluation is thusobtained by using the artificial analogs to all the five human senses. All informationobtained from this sensor system is then fused together to form a human-like decision.Such a sensor head has been used for quality estimation of crispy products, such ascrispbread and chips [15].

11.2

Electronic Tongues

11.2.1

Measurement Principles

There are several measurement principles that have the potential to be used in elec-tronic tongues. The most important ones are based on electrochemical techniquessuch as potentiometry, voltammetry, and conductometry, and there are a numberof textbooks on the subject [16–18]. The use of electrochemical measurements foranalytical purposes has found a vast range of applications. There are two basic elec-trochemical principles: potentiometric and voltammetric. Both require at least twoelectrodes and an electrolyte solution. One electrode responds to the target moleculeand is called the working electrode, and the second one is of constant potential and iscalled the reference electrode.Potentiometry is a zero-current-based technique, in which a potential across a sur-

face region on the working electrode is measured. Different types of membrane ma-terials have been developed, having different recognition properties. These types ofdevices are widely used for the measurement of a large number of ionic species,the most important being the pH electrode, other examples are electrodes for cal-cium, potassium, sodium, and chloride.In voltammetric techniques, the electrode potential is used to drive an electron trans-

fer reaction, and the resulting current is measured. The size of the electrode potentialdetermines if the target molecules will lose or gain electrons. Voltammetric methodscan thus measure any chemical specie that is electroactive. Voltammetric methodsprovide high sensitivity, a wide linear range, and simple instrumentation. Further-more, these methods also enable measurements of conductivity and the amount ofpolar compounds in the solution.Almost all electronic tongue or taste sensors developed are based on potentiometry

or voltammetry. There are, however, also some other techniques that are interesting touse and which have special features making them useful for electronic tongues, suchas optical techniques or techniques based on mass sensitive devices.Optical techniques are based on light absorption at specific wavelengths, in the re-

gion from ultraviolet via the visible region to near infrared and infrared. Many com-pounds have distinct absorption spectra, and by scanning a certain wavelength region,a specific spectrum for the sample tested will be obtained. Optical methods offer ad-vantages of high reproducibility and good long-term stability.Mass sensitive devices, based on piezo electric crystals are also useful. A quartz

crystal resonator is operated at a given frequency, and by the absorption of certain


compounds on the surface of the crystal, its frequency will be influenced [19]. For asurface acoustic wave (SAW) based device, a surface wave is propagated along thesurface of the device [20], and due to adsorption of a compound in its way, the proper-ties of this surface wave will be changed. These types of devices are very general andprovide for the possibility to detect a large number of different compounds.

11.2.2

Potentiometric Devices

The equipment necessary for potentiometric studies includes an ion-selective elec-trode, a reference electrode, and a potential measuring device, as shown schematicallyin Fig. 11.1. A commonly used reference electrode is the silver-silver chloride electrodebased on the half-cell reaction:

AgClþ e� ! Agþ Cl� E0 ¼ þ0:22V ð1Þ

The electrode consists of a silver wire coated with silver chloride placed into a solutionof chloride ions. A porous plug will serve as a voltage bridge to the outer solution.The ion-selective electrode has a similar configuration, but instead of a voltage

bridge, an ion-selective membrane is applied. This membrane should be nonpor-ous, water insoluble and mechanically stable. It should have an affinity for the selectedion that is high and selective. Due to the binding of the ions, a membrane potential willdevelop. This potential, E, follows the well-known Nernst relation:

E ¼ E0 þ ðRT=nFÞlna ð2Þ

where E0 is a constant for the system given by the standard electrode potentials, R isthe gas constant, T the temperature, n the number of electrons involved in the reaction,F the Faraday constant and finally, a is the activity of the measured specie. The po-

Fig. 11.1 Schematic diagram

of an electrochemical cell for

potentiometric measurements


tential change is thus logarithmic in ionic activity, and ideally, a ten-fold increase of theactivity of a monovalent ion should result in a 59.2 mV change in the membrane po-tential at room temperature.In the early 1970s, ion-selective field effect transistors (ISFETs) were developed, in

which the ion-selectivematerial is directly integrated with solid-state electronics [21]. Aschematic diagram of an ISFET is shown in Fig. 11.2. The current between the drainand source (IDS) depends on the charge density at the semiconductor surface. This iscontrolled by the gate potential, which in turn is determined by ions interacting withthe ion-selective membrane. In the ISFET, the normal metal gate is replaced with thereference electrode and sample solution. An attractive feature of ISFETs is their smallsize and ability to be directly integrated with microelectronics, for example, signalprocessing, furthermore, if mass fabricated, they can be made very cheaply. Thesefeatures make them especially valuable for use in electronic tongues.Potentiometric devices offer several advantages for use in electronic tongues or taste

sensors. There are a large number of different membranes available with differentselectivity properties, such as glass membranes and lipid layers. A disadvantage isthat the technique is limited to the measurement of charged species only.

11.2.2.1 The Taste Sensor

The first concepts of a taste sensor were published in 1990 [22, 23]. It was based on ion-sensitive lipid membranes and developed to respond to the basic tastes of the tongue,that is sour, sweet, bitter, salt, and ‘umami’.The multichannel taste sensor [5, 23] was also based on ion-sensitive lipid mem-

branes, immobilized with the polymer PVC. In this taste sensor, five different lipidanalogs were used: n-decyl alcohol, oleic acid, dioctyl phosphate (bis-2-ethylhex-yl)hydrogen phosphate, trioctylmethyl ammonium chloride, and oleylamine, togetherwith mixtures of these. Altogether eight different membranes were fitted on a multi-channel electrode, where each electrode consisted of a silver wire with deposited silverchloride inside a potassium chloride solution, with the membrane facing the solutionto be tasted. A schematic of the multichannel electrode is shown in Fig. 11.3. Thevoltage between a given electrode and a Ag/AgCl reference electrode was mea-sured. The setup is shown in Fig. 11.4. This taste sensor has been used to study re-sponses from the five typical ground tastes, HCl (sour), NaCl (salt), quinine (bitter),

Fig. 11.2 Schematic diagram

of an ion-sensitive field effect

transistor


sucrose (sweet), and monosodium glutamate (umami). The largest responses wereobtained from the sour and bitter compounds, thereafter umami and salt, and forsucrose almost no response was obtained. For other sweet tasting substances, suchas the amino acids glycine and alanine, larger responses were obtained. It was furthershown that similar substances, such as sour substances like HCl, acetic acid, citricacid, or salty substances such as NaCl, KCl, and KBr showed similar response pat-terns. The system does not respond well to nonelectrolytes, which have little effecton the membrane potential [24].The multichannel system has been commercialized [9]. The detecting part is an

eight-channel multisensor, placed on a robot arm and controlled by a computer.

Fig. 11.3 Schematics of the

multichannel electrode with

eight lipid/polymer membranes

Fig. 11.4 The measurement setup for the eight-channel electrode system


The samples to be tested are placed in a sample holder together with a cleaning solu-tion as well as reference solutions. The measurements then take place in a specialorder: first the multisensor is cleaned by dipping into the cleaning solution, thereafterinto the sample solution, and the cycle repeats. At certain intervals, the multisensor isplaced in the reference solution for calibration purposes.This taste sensing system has been used in a number of different applications.

These have mainly dealt with discrimination and estimation of the taste of differentdrinks. In one investigation, 33 different brands of beers were studied [25]. The sam-ples were analyzed both by using a sensory panel and by the taste sensing system. Thesensory panel expressed the taste of the different beers in the parameters sharp-versus-mild and rich-versus-light. The output pattern from the taste sensor was analyzedusing principal components analysis (PCA). An interesting observation was thatthe first principal component corresponded well to the taste parameter rich-versus-light taste, and the second principal component corresponded well to the parametersharp-versus-mild taste.Mineral water has also been studied using the taste sensing system [24]. A good

correlation of the sensor responses to the hardness of the water could be seen inPCA plots, and also the sensor could discriminate between different brands.Other applications involve the monitoring of a fermentation process of soybean

paste [26], estimation of the taste of milk [27] or coffee [28], and the developmentof a monitoring system for water quality [29].

11.2.2.2 Ion-Selective Electrodes

The term ‘electronic tongue’ was first coined in 1996 at the Eurosensors X conference[5, 30]. The concept had been developed as a research collaboration between an Italiangroup (DiNatale, Davide and D’Amico) and a Russian group (Legin, Vlasov and Rud-nitskaya). This device has now been developed further, and a large number of applica-tions have been studied, and are described in the following.The first devices consisted of potentiometric sensor arrays of two general kinds:

conventional ones such as pH, sodium and potassium-selective electrodes, and espe-cially designed ones. The latter ones were based mainly on chalcogenide vitreous ma-terials. Altogether the array included 20 potentiometric sensors: glass, crystalline, PVCplasticized sensor, and metal electrodes. The sensor system was used for the recogni-tion of different kinds of drinks such as tea, soft drinks, juices, and beers. Each samplewas measured twice, and the information obtained from the sensor array was treatedusing PCA. The score plots showed good separation between all these samples. Thedeterioriation of orange juice during storage was also followed, and by using an arti-ficial neural network (ANN) on the data obtained, a model for storage-time predictioncould be made.The measurements of compounds of relevance for pollution monitoring in river

water using this electronic tongue have also been reported [31]. River water was takenat three locations and artificially polluted with Cu, Cd, Fe, Cr, and Zn, all in ionic form,representing a ‘common’ pollution from the industries. The sensor array consisted of22 electrodes mainly based on chalcogenide glasses and conventional electrodes. Dif-


ferent approaches of data analysis were performed such as multiple linear regression(MLR), projection to latent structure (PLS), nonlinear least square (NLLS), back-pro-pagation ANN, and a self-organizing map (SOM). Two modular models were devel-oped, the first a combination of PCA and PLS, the second a combination of ANN andSOM, and both could predict pollutant ions well.A similar setup of this electronic tongue has been used for qualitative analysis of

mineral water and wine [32], and for multicomponent analysis of biological liquids[33]. A flow-injection system based on chalcogenide glass electrodes for the determi-nation of the heavy metals Pb, Cr, Cu, and Cd was also developed [34]. The approach ofcombining flow injection analysis in combination with a multisensor system and ana-lyzing data using multivariate data analysis appears very advantageous. The flow in-jection analysis (FIA) technique offers several advantages: since relative measure-ments are performed, the system is less influenced by sensor baseline drift, calibra-tion samples can be injected within a measurement series, and the system is welladapted for automization. One should also remember that most electronic nose mea-surements are based on a gas-phase FIA technique, one reason is to compensate forthe drift of the gas sensors.

11.2.2.3 Surface Potential Mapping Methods

A very interesting technique has been developed, in which the surface potential of asemiconductor structure is measured locally [35–37]. This is a new type of a potentio-metric system that provides for a contactless sensing over a surface and is thus a con-venient way to analyze a multifunctional surface. It also opens up possibilities to usegradients of different functional groups as the sensing principle. The semiconductorsurface acts as the working electrode on to which the test solution is applied. Into thissolution a reference electrode and an auxiliary electrode are also applied. On the back-side, a light-emitting diode is applied, which can scan the surface in both x and ydirections. By illuminating a certain region on the semiconductor (via the back-side), a photocurrent will be generated, the size being a measure of the surface po-tential at that particular region.In one application [35], five lipid membranes (oleic acid, lecithin, cholesterol, phos-

phatidyl ethanolamine, and dioctyl phosphate) were deposited at different areas on thesemiconductor surface. First, one lipid was coated onto the whole area, the next on twothirds of the area, and the third on the last third of the area. The whole surface wasrotated by 908, and the procedure was repeated with the remaining lipids. The sensingarea could thus be divided into nine different regions with varying composition andthickness of lipid layer. This sensor surface was then investigated for the basic tastesubstances, HCl (sour), NaCl (salt), quinine (bitter), sucrose (sweet), andmonosodiumglutamate (umami). The responses obtained had similar responses to the taste sensorsystem described earlier, that is the largest responses were obtained from the sour andbitter compounds, thereafter umami and salt, and for sucrose almost no response wasobtained. The method has also been further developed [36–38].


11.2.3

Voltammetric Devices

In voltammetric devices, the current is measured at given potentials. This current isthen a measure of the concentration of a target analyte. The reactions taking place atthe electrode surface are:

Oþ ne� ! R ð3Þ

where O is the oxidized form and R is the reduced form of the analyte. At standardconditions, this redox reaction has the standard potential E0. The potential of the elec-trode, Ep, can be used to establish a correlation between the concentration of the oxi-dized (C0) and the reduced form (Cr ) of the analyte, according to the Nernst relation:

Ep ¼ E0 þ RT=nFðlnðC0=CrÞÞ ð4Þ

A well-known voltammetric device is the Clark oxygen electrode, which operates at�700 mV, the potential at which oxygen is reduced to hydrogen peroxide on a plati-num electrode. By reverting the potential, the electrode will be sensitive to hydrogenperoxide.The use of voltammetry as a sensing principle in an electronic tongue appears to

have several advantages: the technique is commonly used in analytical chemistry dueto features such as very high sensitivity, versatility, simplicity, and robustness. Thetechnique also offers the possibility to use and combine different analytical principlessuch as cyclic, stripping, or pulsed voltammetry. Depending on the technique, variousaspects of information can be obtained from the measured solution. Normally, redox-active species are being measured at a fixed potential, but by using, for example, pulsevoltammetry or studies of transient responses when Helmholtz layers are formed,information concerning diffusion coefficients of charged species can be obtained.Further information is also obtainable by the use of different types of metals forthe working electrode.Whenusingvoltammetryincomplexmediacontainingmanyredox-activecompounds

and different ions, the selectivity of the system is normally insufficient for specificanalysis of single components, since the single steps in the voltammogram are tooclose tobe individuallydiscriminated.Rathercomplicatedspectraare thereforeobtainedandtheinterpretationofdataisverydifficultduetoitscomplexity.Thesevoltammogramscontain a large amount of information, and to extract this there has been an increasinginterest and use of multivariate analysis methods in the field [39–42].Among the various techniques mentioned, pulse voltammetry is of special interest

due to the advantages of greater sensitivity and resolution. Two types of pulse voltam-metry are commonly used, large amplitude pulse voltammetry (LAPV) and small am-plitude pulse voltammetry (SAPV). At the onset of a voltage pulse, charged species andoriented dipoles will arrange next to the surface of the working electrode, forming aHelmholtz double layer. A charging nonfaradic current will then initially flow as thelayer builds up. The current flow, i, is equivalent to the charging of a capacitor in serieswith a resistor, and follows an equation of the form:


i ¼ E*RSexpð�t=R*SBÞ ð5Þ

where RS is the resistance of the circuit (¼ solution), E* is the applied potential, t is thetime, and B is an electrode related equivalent capacitance.The redox current from electroactive species shows a similar behavior, initially large

when compounds close to the electrode surface are oxidized or reduced, but decayswith time when the diffusion layer spreads out. The current follows the Cottrell equa-tion [16–18]:

i ¼ nFADCðð1=ðpDtÞ1=2Þ þ 1=rxÞÞ ð6Þ

where A is the area of the working electrode, D is the diffusion constant, C is theconcentration of analyte and 1/rx is an electrode constant. At constant concentrati-on, the equation can be simplified:

i ¼ K1ð1=tÞ1=2 þ K2 ð7Þ

where K1 and K2 are constants.In LAPV, the electrode is held at a base potential at which negligible electrode reac-

tions occur. After a fixed waiting period, the potential is stepped to a final potential. Acurrent will then flow to the electrode, initially sharp when the Helmholtz double-layeris formed. The current will then decay as the double-layer capacitance is charged andelectroactive compounds are consumed, until the diffusion-limited faradic currentremains, as depicted by Eqs. (5) and (7). The size and shape of the transient responsereflect the amount and diffusion coefficients of both electroactive and charged com-pounds in the solution. When the electrode potential is stepped back to its startingvalue, similar but opposite reactions occur. The excitation waveform consists of suc-cessive pulses of gradually changing amplitude between which the base potential isapplied.The instantaneous faradic current at the electrode is related to surface concentra-

tions and charge transfer rate constants, and depends exponentially on the differenceof the electrode potential between the start value and the final potential.In SAPV, a slow continuous direct current (DC) scan is applied to the electrode on to

which small amplitude voltage pulses are superimposed. This DC scan causes achange in the concentration profile of the electroactive species at the surface. Sinceonly small pulse changes in the electrode potential are considered, this will result insmall perturbations in the surface concentration from its original value prior to theapplication of the small amplitude excitation. Normally for SAPV, the current issampled twice, one just before the application of the pulse, and one at the end ofthe pulse, and the difference between these is recorded as the output. This differentialmeasurement gives a peaked output, rather than the wave-like responses that areusually obtained.


11.2.3.1 The Voltammetric Electronic Tongue

The first voltammetric electronic tongue described used both LAPV and SAPV appliedto a double working electrode, an auxiliary, and a reference electrode [6]. The doubleworking electrode consisted of one wire of platinum and the other of gold, both with alength of 5 mm and a diameter of 1 mm. Current and current transient responseswere measured by a potentiostat connected to a personal computer (PC). The PCwas also used for onset of pulses and measurement of current transient responsesand to store data. Via two relays, the PC was also used to shift the type of workingelectrode (gold or platinum) used. Current responses from both LAPV and SAPVwere collected and used as input data for PCA.In a first study, samples of different orange juices, milk, and phosphate buffer were

studied. A PCA plot performed on the data showed good separation of the samples, asshown in Fig. 11.5. This electronic tongue was also used to follow the ageing process oforange juice when stored at room temperature.The voltammetric electronic tongue has been further developed. A recent configura-

tion is shown in Fig. 11.6. It consisted of five working electrodes, a reference electrodeand an auxiliary electrode of stainless steel. Metal wires of gold, iridium, palladium,platinum, and rhodium used as working electrodes were embedded in epoxy resin andplaced around a reference electrode in such a way that only the ends of the workingelectrodes and the reference electrodes were exposed. The opposite ends of the work-ing electrodes were connected to electric wires. The arrangement was inserted in aplastic tube ending with a stainless steel tube as an auxiliary electrode. The wiresfrom the working electrode were connected to a relay box, enabling each workingelectrode to be connected separately in a standard three-electrode configuration. Dif-ferent types of pulsed voltammetry could be applied, LAPV, SAPV and staircase. InFig. 11.7, typical voltage pulses and the corresponding current responses are shown.

Fig. 11.5 PCA analysis of different samples analyzed with the

voltammetric electronic tongue


This electronic tongue has been used to follow the deterioration of milk due to micro-bial growth when stored at room temperature [43]. The data obtained were treated withPCA, and the deterioration process could clearly be followed in the diagrams. Tomakemodels for predictions, projections to latent structure and ANNs were used. Whentrained, both models could satisfactorily predict the proceedings of bacterial growthin the milk samples.A hybrid electronic tongue has also been developed, based on the combination of the

measurement techniques potentiometry, voltammetry, and conductivity [44]. The hy-brid electronic tongue was used for classification of six different types of fermentedmilk. Using ion-selective electrodes, the parameters pH, carbon dioxide, and chlorideion concentrations were measured. The voltammetric electronic tongue consisted ofsix working electrodes of different metals (gold, iridium, palladium, platinum, rhe-nium, and rhodium) and a Ag/AgCl reference electrode. The measurement principlewas based on large amplitude pulse voltammetry in which current transients weremeasured. The data obtained from the measurements were treated with multivariatedata processing based on PCA and an ANN. The hybrid tongue could separate all sixdifferent types of fermented milks. Also, the composition of the microorganisms ofthe different fermentations was reflected in the PCA results.A measurement system, based on the FIA technique applied to a voltammetric elec-

tronic tongue has also been developed [45]. A reference solution was continuouslypumped through a cell with a voltammetric electronic tongue, and test samples

Fig. 11.6 A recent configurati-

on of the voltammetric electro-

nic tongue


were injected into the flow stream. Responses were obtained by measuring the result-ing pulse height. The FIA technique offered several advantages: since relative mea-surements are performed, the system is less influenced by sensor baseline drift, cali-bration samples can be injected within a measurement series, and the system is welladapted for automization. The system was used to analyze standard solutions of H2O2,KCl, CuNO3, K4[Fe(CN)6], and NaCl, and results obtained were treated with multivari-ate data analysis. PCA showed that electrode drift was considerably decreased. Thesetup was also used for classification of different orange juices.The voltammetric electronic tongue has also been used for the monitoring of drink-

ing water quality, and a review has recently been published [46].

11.2.3.2 Feature Extraction

To be able to describe correctly the shape of the current pulses during the voltagepulses, a large amount of variables are collected. For each pulse, up to 50 variablescan be taken for the multivariate data processing. In a complete measurement seriesusing up to 100 pulses applied to four electrodes, a total number of up to 2000 discretevalues can be collected. Most of these are redundant having a low level of information.

Fig. 11.7 Three different pulsed voltammetric techniques

used by the voltammetric electronic tongue. The upper part

shows applied voltage pulses. The lower part shows the

corresponding current responses for four different elec-

trodes (gold, iridium, palladium, and platinum) due to the

onset of voltage pulses


The shape of the current responses for LAPV follows Eqs. (5) and (7) in principal,which means that constants can be calculated that express the current response. In afirst attempt, constants fitting Eq. (5) were calculated, and for a given application forclassification of different teas, PCA showed that a better separation was obtained usingthese constants compared with the original data [47].

11.2.3.3 Industrial Applications using the Voltammetric Electronic Tongue

The list of possible industrial applications for voltammetric electronic tongues can bemade very long. Electronic tongues are versatile in their applicability since they cangive general information as well as specific information, such as pH and conductivity,about a sample [48]. In addition the construction of the voltammetric electronic tonguecan be made very robust – another reason that makes it suitable for many differentareas of applications. One example where this quality is important is in the food in-dustry where the use of sensors made of glass, for example, is not always acceptable.The voltammetric electronic tongue has been studied in a number of different in-

dustrial applications. One example is in the pulp and paper industry where the increas-ing machine speed and system closure of the papermaking process have caused anincreased need to control the wet-end chemistry of the paper machine. The main chal-lenges have been to establish knowledge of its impact on product properties as well asthe most important relations between wet-end chemistry and performance of stocktrades such as paper chemicals and pulp in order to improve productivity and runability. The voltammetric electronic tongue has been evaluated on pulp samplesand the prediction ability of six reference parameters – pH, conductivity, chemicaloxygen demand, cationic demand, zeta potential, and turbidity – was evaluated usingPLS models. The results indicated that the electronic tongue studied had very promis-ing features as a tool for wet-end control. Flexibility, fast response and wide sensitivityspectra make the electronic tongue suitable for a vast number of possible applicationsin the papermaking process [49].Another example of an industrial application where the electronic tongue has been

studied is as a sensor system in household appliances such as dishwashers and wash-ing machines. The machines are today programmed to secure a good result, whichoften implies that the settings, such as temperature and washing time, are toohigh resulting in an unnecessarily large consumption of energy, water, and detergent.A sensor that can give information about the water quality, type of soil loaded, and

when the rinse water is free from detergents would increase the efficiency of thesemachines. The voltammetric electronic tongue has, for example, been able to distin-guish between different standardized soil types, even at high levels of detergents addedto the solutions [50]. Much work remains to be done before the electronic tonguemight be a conventional sensor technology in this type of machine, but these preli-minary studies show its potential.The third example of industrial applications for electronic tongues is as amonitoring

device in drinking-water production plants [46]. The quality of drinking water variesdue to the origin and quality of the raw water (untreated surface or ground water), butalso due to efficiency variations in the drinking-water production process. Problems


can be related to occurrence of, e.g., algae, bacteria, pesticides, and herbicides, andindustrial contamination, in the raw water. The character of the raw water, and thebiological activity at the production plant as well as in the distribution net may allcause quality problems such as bad odor/taste and/or unhealthiness. A method formonitoring variations in the raw water quality as well as the efficiency of separateprocess steps would therefore be of considerable value. To evaluate the voltammetricelectronic tongue for this purpose, water samples from each of eight parallel sandfilters in a drinking-water production were collected and measured, as shown inFig. 11.8. A PCA plot for the samples is shown in Fig. 11.9. The raw water samplesare well separated from the treated water samples (slow and fast filter, and clean) in theplot. One interesting observation is that the water quality after flowing through someof the slow filters cluster close to that after the fast filter, which indicates that thechemical composition of the two are similar. This result suggests that these slow fil-ters are not working properly. The water quality after flowing through three of theother slow filters cluster, however, much closer to the clean water (which has alsobeen chemically treated), which in a similar way indicates that these filters are work-ing properly. This implies of course that the quality of the clean water is acceptable.Figure 11.10 demonstrates a possible use of electronic tongues (and PCA) namely tocheck the performance of given filters of the production plant.The results for the drinking-water plant above suggest a possible use of the electro-

nic tongue in continuous monitoring of the status of a given filter or other parts of theplant. Aftermaintenance of a filter, for example, the initial position in a PCA plot of the

Fig. 11.8 Top: A water production plant. Bottom: Schematics of the production plant

showing the inlet of raw water, a fast filter, eight parallel sand filters and the final pH

adjustment and chlorination step. The sampling positions of the electronic tongue are

indicated


water coming out of the filter is determined. Through a continuous measurement onthe water after flowing through the filter, the time evaluation of the position in the PCAplot is followed. As long as the points cluster together within an area (determinedinitially by experience) the filter is performing well enough. Deviations from the ’nor-mal’ cluster indicate a malfunctioning filter (Fig. 11.10). To be able to associate a de-viation from the ‘normal’ cluster to any specific parameter the reasons for malfunc-tioning filters must be studied. For this purpose traditional analytical chemical as wellas biological methods must be used. The signals from the electronic tongue can thenbe correlated to these reference methods, and if there is a correlation, specific distur-

Fig. 11.9 PCA plot of samples obtained from the water production plant

Fig. 11.10 Schematic illustrati-

on of time-dependent PCA ana-

lysis used to detect changes in

performance of a part of a plant


bances of the properties of a filter can be tracked to certain areas of the PCA plot. Thepossibility to detect a malfunctioning filter, regardless of the parameters causing it, isvery valuable since it allows early measures to be taken against the problem.Other application areas that are under study are the use of the electronic tongue for

detection of microbial activity [48, 51]. One important industrial area for such applica-tions is in the food industry where the quality of food is very much determined by itsmicrobial status. This can be unwanted microbial occurrence like pathogenic micro-organisms as well as wanted microbial growth in, for example, fermented foodstuff.Studies have shown that it is possible to follow the growth of mould and bacteria, andalso to separate between different strains of molds with the voltammetric electronictongue [48, 51].

11.2.3

Piezoelectric Devices

Piezoelectricmaterials have an interesting property in that an electric field is generatedby the application of pressure, and that it is distorted by the application of an electricfield. The crystal will generate a stable oscillation of the electric voltage across it whenan AC voltage is applied using an external oscillatory circuit. This resonance frequencyis changed with the mass of the crystal according to the equation:

Df ¼ cf 2ðDM=AÞ ð8Þ

whereDf andDM are the changes in resonance frequency andmass, respectively, c is apositive constant, f the resonance frequency, and A the electrode area.Quartz crystals are widely used as sensors where the chemical sensitivity and selec-

tivity is obtained from an adsorbent layer on the crystal. For a quartz crystal micro-balance, analyte sorption on this layer will result in a frequency change [19]. Depend-ing on the affinity properties of the adsorbing layer, different chemical compounds canbe measured. Using an array based on these kinds of devices coated with hydrophilicmono- and dicarbon acids, organic and inorganic acids, and amines in drinking watercould be detected [52].A quartz resonator coated with a lipid/polymer membrane has also been investi-

gated. The oscillation frequency showed different responses depending on taste sub-stances and the lipid in the membrane [53].SAW devices have also been applied for sensors in the gas and aqueous phases. For

use in liquids, shear-horizontal mode SAW (SH-SAW) must be used [20, 54]. Using a368 rotated Y-cut X-propagating LiTaO3 device, a sensing system for the identificationof fruit juices was developed. The device was divided in two parts: one metallized areaas reference, the other area having a free surface that was electrically active. The sensorsensitivity was controlled by changing the excitation frequency. The phase differenceand amplitude ratio between the reference and sensing signals were measured. Asystem was developed using three SH-SAW devices operated at the frequencies 30,50 and 100 MHz, respectively, which was used to identify eleven different fruit juices


[54]. In another study, a similar system was used to classify thirteen different kinds ofwhisky samples [55]. The device has been further studied in an application for thediscrimination of four commercial brands of natural spring water [56]. Transient fre-quency responses were studied, and using pattern recognition based on ANNs, all foursamples could be easily discriminated.A review on recent efforts towards the development of both electronic tongues and

electronic noses has been published [57], in which working principles, and the con-struction and performance of these systems mainly based on SAWs are discussed.

11.3

The Combination or Fusion of Artificial Senses

Appreciation of food is based on the combination or fusion of many senses, in fact fora total estimation all five human senses are involved: vision, tactile, auditory, taste, andolfaction. The first impression is given by the look of the food, thereafter informationof weight and surface texture is gained by holding it in the hand. Thus, even before thefood has come in contact with the mouth, a first conception is already made. In themouth, additional information is given by the basic taste on the tongue, and the smell.Other quality parameters such as chewing resistance, melting properties, crisp sound,and temperature are added. This is often referred to as the mouth-feel, and is a veryimportant property of the food. Individual properties correlated to special food pro-ducts are especially important for their characterization, such as the crispness of crisp-bread or chips, the chilling properties of chocolate when melting on the tongue, or thesoftness of a banana.A challenging problem in the food processing industry is maintenance of the quality

of food products, and, consequently much time and effort are spent on methods forthis. Panels of trained experts evaluating quality parameters are often used, which,however, entails some drawbacks. Discrepancy might occur due to human fatigueor stress, sensory panels are time consuming, expensive, and cannot be used for on-line measurements. The development of replacement methods for panels for objectivemeasurement of food products in a consistent and cost-effective manner is thus highlywanted by the food industry.In this respect, the combination of artificial senses has great potential to at least in

part replace these panels, since the outcome of such a combination will resemble ahuman-based sensory experience. For these purposes, both simple and more complexcombinations of artificial senses have been investigated. Depending on the art of thequality parameters to be investigated, different types of artificial senses are important.For estimation of the crispness of potato chips, the human sense analogs of olfaction,auditory, and tactile would be satisfactory, but for total quality estimations, all fivehuman sense analogs should be represented.Applications of the combinations of artificial senses have so far only been developed

for the food and beverage industry, dealing with classification and quality issues. In thefuture, however, it is expected that this approach also will find applications in othertypes of the process industry.


An important aspect is how to fuse the sense information. How a body of algo-rithms, methods, and procedures can be used to fuse together data of different originsand nature in order to optimize the information content has been discussed [58, 59].The approach of abstraction level is introduced, namely the level at which the sensordata are fused together. A low level of abstraction means that the signals from thesensors are merely added together in a matrix. A high level of abstraction meansthat the data of each sensor system is analyzed as a stand-alone set, thus a selectionof the most important features of each system can be selected and then merged to-gether.

11.3.1

The Combination of an Electronic Nose and an Electronic Tongue

Various applications concerning the combination of an electronic nose and an elec-tronic tongue have been reported. In a first study, different types of wine were classi-fied using a taste-sensor array using lipid/polymer membranes and a smell-sensorarray using conducting polymer electrodes [14]. A clear discrimination was foundfor the different samples. Also the effect of the ageing process was studied. Later in-vestigations performed in more detail evaluate the different information obtainedfrom the different sensor systems, thus, in one study of wines, an electronic nosebased on eight QMB sensors using different metallo porphyrins as sensing layers,and an electronic tongue based on six porphyrin-based electrodes were used [59].The data obtained were correlated with analysis of chemical parameters. PCA loadingplots showed that the artificial sensory systems were orthogonal to each other, whichimplies the independence of the information obtained from them.The combination of an electronic tongue and an electronic nose for classification of

different fruit juices has also been described [60]. The ‘electronic nose’ was based onan array of gas sensors consisting of 10 metal-oxide-semiconductor field effect tran-sistors (MOSFETs) with gates of thin catalytically active metals such as Pt, Ir, and Pdand four semiconducting metal-oxide-type sensors. The electronic tongue was basedon pulse voltammetry, and consisted of six working electrodes of different metals, anauxiliary electrode, and a reference electrode. Using PCA, it was shown that the elec-tronic nose or the electronic tongue alone was able to discriminate fairly well betweendifferent samples of fruit juices (pineapple, orange, and apple). It was also shown thatthe classification properties were improved when information from both sources werecombined, both in the unsupervised PCA and the supervised PLS.An original sensor fusion method based on human expert opinions about smell and

taste andmeasurement data from artificial nose and taste sensors have been presented[12, 61]. This is achieved by a combination of ANNs and conventional signal handlingthat approximates a Bayesian decision strategy for classifying the sensor information.Further, a fusion algorithm based on the maximum-likelihood principle provides acombination of the smell and taste opinions, respectively, into an overall integratedopinion similar to human beings.

11.3 The Combination or Fusion of Artificial Senses 285285

11.3.2

The Artificial Mouth and Sensor Head

Quality estimation of crispy products such as chips or crisp bread offers an intriguingproblem. The human perception of crispy quality comes from the impressions col-lected when the product enters the mouth and is chewed. While chewing and crush-ing, impressions of chewing resistance, crushing vibrations, and crushing sound aswell as the descriptive taste of the sample, will all contribute to give an overall qualityimpression. Methods developed so far only measure the crispness in terms of thehardness and brittleness of the sample. It appears that to give a better descriptionof the crispness experienced, more subtle quality parameters referring to the ‘mouthfeel’ should be accounted for.A special ‘artificial mouth’ or ‘crush chamber’ has been designed, in which infor-

mation corresponding to three senses could be obtained: ‘auditory’ by a microphone,‘tactile’ by a force sensor, and ‘olfaction’ by a gas sensor array, thus collecting informa-tion mimicking these three human senses [11, 12]. In this artificial mouth, crispy pro-ducts could be crushed under controlled conditions. The schematic of the artificialmouth is shown in Fig. 11.11. A piston could be moved at a constant speed by theaction of a stepping motor connected to a lever. The force applied to the pistonwas recorded by a force sensor, and a dynamic microphone was placed at the bottomof the chamber. The chamber was thermostated to 37 8C. The sensor array consisted of10 MOSFET gas sensors, with gates of thin, catalytically active metals such as Pt, Ir,and Pd, and four semiconducting metal-oxide type sensors.Five types of crispbread have been investigated, one based on wheat flour, the other

four based on rye flour. The information from the three information sources was first

Fig. 11.11 Schematics of the crush chamber or ‘electronic mouth’


individually examined. Using information from the gas sensors, only the wheat flourbased crispbread could be separated from the others. Using the sound information, acorrelation to the hardness and brittleness of the samples could be obtained, and si-milar results were obtained from the force sensor. By combining all sense analogs, allfive samples could be separated [11].The quality of potato chips has also been investigated [12]. The aim of the study was

to follow the ageing process during storage. For these studies, one set of experimentswas performed on potato chips stored in an opened bag, the other set in a closed bagthat was opened only for sample taking. PCA analysis of data obtained from the arti-ficial mouth showed that the information from the single information sources was notsufficient to explain the ageing process, but with merged data, the ageing processcould be followed. A closer examination of the loading plot revealed that much ofthe data were strongly correlated, and from this plot, a smaller subset of data couldbe collected. This was used for an ANN, in which the prediction of age was modeled,and it was found that predicted values of age correlated well with true values.To make a complete sensory evaluation, all five human senses are involved. A new

approach for the assessment of human-based quality evaluation has been obtained bythe design of an electronic sensor head [15]. The investigated sample enters an arti-ficial mouth for detection of chewing resistance and recording of the chewing soundvia a microphone. A video camera is used for the identification of color, shape, andsimilar properties of the sample. In parallel, aroma liberated during the crushing pro-cess is measured by a gas sensor array. Finally, the crushed sample is mixed with asaline solution, and an electrochemical multi-electrode arrangement analyzes the mix-ture. The artificial analogs to all the five human senses are therefore used for qualityevaluation of the sample. All information obtained from the sensor system is thenfused together to form a human-based decision. The arrangement was originally de-signed for quality studies of potato chips directly atline in the factory, hence it was alsoequipped with a robot arm, which could take out samples from the line. This sensorhead has been used for quality estimation of crispy products, such as crispbread andchips.For the chips application, it was interesting to note that vision alone could predict the

quality parameters of freshness, spots, and spiciness, the olfactory analogs the amountof spiciness, and the auditory and touch analogs the freshness. The freshness of thechips can thus be determined both by change in color and by change in texture. Also,the spiciness of a chip can be determined both by the smell and by the number andcolor of the spices as seen by the camera. If all senses are fused together, all qualityparameters could of course be correctly predicted.

11.4

Conclusions

Biomimetic measurement methods, as illustrated by the electronic nose and the elec-tronic tongue, are rapidly being introduced in different applications. It is an interestingdevelopment where new achievements in both hardware and software act together to


improve the performance of the sensor arrays. Some of the techniques used, such asthe pulse voltammetric measurements on a number of different (metal) electrodes,produce an enormous amount of data, in most cases with a large redundancy. Anefficient data evaluation method is therefore necessary in order to utilize the measure-ments in an optimal way. The further development of algorithms is therefore an im-portant task especially for sensor arrays based on simple, but well investigated, indi-vidual sensors. The biomimetic concept should, however, not be exaggerated. Thehuman senses are strongly connected in the brain and give rise to associations basedon an integrated previous experience. With regard to taste, the human taste sensationcan, in general, not be described by one of the five simple ‘basic’ tastes. In olfaction, thesituation is similar. One should therefore be aware of the fact that the manmade sen-sor arrays give responses that are only related to the taste and smell, even when theycorrelate with the sensation obtained by humans. Sometimes the sensor arrays do noteven respond to the same molecules which give rise to the human sensation.With this knowledge in mind, the sensor arrays are still extremely useful for quality

control of products and processes as indicated in this contribution. In many applica-tions there is no need to compare the sensor signals with sensory results, the signalsthemselves and their variations contain enough information. In many (industrial) ap-plications the arrays will therefore not be calibrated against humans, but against tradi-tional analytical techniques.Another interesting possibility is to follow the evaluation of the data in a ‘human

dependent’ PCA plot. In this case, process or quality monitoring can be made usingreferences in the PCA plot itself, as discussed in correlation with the clean water pro-duction plant.A combination of electronic noses and tongues with mechanical sensors and cam-

eras of course increases the possibility to evaluate the properties of a given sample. Theexperiments made so far indicate that such a ‘biomimetic sensor head’ or robot has alarge potential with regard to the evaluation of food, both of raw material and finishedproducts. Such an approach will also have uses in process and product control ingeneral.

References

1 This book.2 J. W. Gardner, P. N. Bartlett. ‘A brief historyof electronic noses’, Sensors and Actuators1994 B18-19 211–220.

3 F. Winquist, H. Sundgren, I. Lundstrom.‘Electronic Noses for Food Control’,in Biosensors for Food Analysis, 1998,A.O. Scott, Ed., The Royal Society ofChemistry, Athenaeum Press Ltd, UK.

4 K. Toko. ‘Taste sensor with globalselectivity’, Materials Science and Engineering1996 C4 69–82.

5 A. Legin, A. Rudinitskaya, Y. Vlasov,C. Di Natale, F. Davide, A. D’Amico. ‘Ta-sting of beverages using an electronic tonguebased on potentiometric sensor array’,Technical digest of Eurosensors X, Leuven,Belgium 1996 427–430.

6 F. Winquist, P. Wide, I. Lundstrom.‘An electronic tongue based on voltam-metry’, Analytica Chimica Acta 1997 35721–31.

7 K. Toko. ‘Taste sensor’, Sensors and Actuators2000 B64 205–215.


8 K. Toko. ‘A taste sensor’, MeasurementScience and Technology 1998 9 1919–1936.

9 Taste Sensing System SA401, Anritsu Corp.,Japan.

10 The Astree Liquid & Taste Analyzer,Alpha MOS, Toulouse, France.

11 F. Winquist, P. Wide T. Eklov, C. Hjort,I. Lundstrom. ‘Crispbread quality evaluationbased on fusion of information from thesensor analogies to the human olfactory,auditory and tactile senses’, Journal of FoodProcess Engineering 1999 22 337–358.

12 P. Wide, F. Winquist, A. Lauber. ‘The per-ceiving sensory estimated in an artificialhuman estimation based sensor system’,Proc. IEEE Instrumentation and MeasurementTechnology Conference, Ottawa, Canada, May1997.

13 L. Rong, W. Ping, H. Wenlei. ‘A novelmethod for wine analysis based on sensorfusion technique’, Sensors and Actuators2000 B66 246–250.

14 S. Baldacci, T. Matsuno, K. Toko, R. Stella,D. De Rossi. ‘Discrimination of wine usingtaste and smell sensors’, Sensors andMaterials 1998 10(3) 185–200.

15 P. Wide, F. Winquist, I. Kalaykov. ‘Theartificial sensor head: A new approachin assessment of human based quality’,Proceedings of the Second International Con-ference on Information Fusion, FUSION ‘99.Int. Soc. Inf. Fusion, Mountain View, CA,USA 2 1999 1144–1149.

16 A. J. Bard, L. R. Faulkner. ‘ElectrochemicalMethods – Fundamentals and Applications’,John Wiley & Sons, Inc. 1980.

17 J. Wang. ‘Analytical Electrochemistry’,Wiley-VCH 1994.

18 P. T. Kissinger, W. R. Heineman. ‘Labora-tory Techniques in ElectroanalyticalChemistry’, 2nd Edition, Marcel Dekker,Inc. 1996.

19 R. Lucklum, P. Hauptmann. ‘The quartzcrystal microbalance. Mass sensitivity,viscoelasticity and acoustic amplification’,Sensors and Actuators 2000 B70 30–36.

20 T. Yamazaki, J. Kondoh, Y. Matsui,S. Shiokawa. ‘Estimation of componentsin mixture solutions of electrolytes usinga liquid flow system with SH-SAW sensor’,Sensors and Actuators B 2000 83 34–39.

21 P. Bergveld. ‘The ISFET’, IEEE Trans.Biomed. Eng. 1970 BME-19.

22 K. Toko, K. Hayashi, M. Yamanaka,K. Yamafuji. ‘Multichannel taste sensorwith lipid membranes’ Tech. Digest 9th Sens.Symp., Tokyo, Japan 1990 193–196.

23 K. Hayashi, M. Yamanaka, K. Toko,K. Yamafuji. ‘Multichannel taste sensorusing lipid membranes’, Sensors andActuators 1990 B2 205–213.

24 K. Toko. ‘Biomimetic Sensor technology’,Cambridge University Press 2000.

25 K. Toko. ‘Electronic Tongue’, Biosensors andBioelectronics 1998 13 701–709.

26 T. Imamura, K. Toko, S. Yanagisawa,T. Kume. ‘Monitoring of fermentationprocess of miso (soybean paste) usingmultichannel taste sensor’, Sensors andActuators 1996 B37 179–185.

27 H. Yamada, Y. Mizota, K. Toko, T. Doi.‘Highly sensitive discrimination of tasteof milk with homogenization treatmentusing a taste sensor’, Materials Science andEngineering 1997 C5 41–45.

28 T. Fukunaga, K. Toko, S. Mori,Y. Nakabayashi, M. Kanda. ‘Quantificationof taste of coffee using sensor with globalselectivity’, Sensors and Materials 1996 8(1)47–56.

29 A. Taniguchi, Y. Naito, N. Maeda, Y. Sato,H. Ikezaki. ‘Development of a monitoringsystem for water quality using a tastesensor’, Sensors and Materials 1999 11(7)437–446.

30 C. Di Natale, F. Davide, A. D’Amico,A. Legin, A. Rudinitskaya, B. L. Selezenev,Y. Vlasov. ‘Applications of an electronictongue to the environmental control’,Technical digest of Eurosensors X, Leuven,Belgium, 1996 1345–1348.

31 C. Di Natale, A. Macagnano, F. Davide,A. D’Amico, A. Legin, Y. Vlasov, A. Rudi-nitskaya, B. L. Selezenev. ‘Multicomponentanalysis on polluted water by means of anelectronic tongue’, Sensors and Actuators1997 B44 423–428.

32 A. Legin, A. Rudinitskaya, Y. Vlasov,C. Di Natale, E. Mazzone and A. D’Amico.‘Application of Electronic tongue for quan-titative analysis of mineral water and wine’,Electroanalysis 1999 11(10–11) 814–820.

33 A. Legin, A. Smirnova, A. Rudinitskaya,L. Lvova, E. Suglobova, Y. Vlasov. ‘Chemicalsensor array for multicomponent analysis ofbiological liquids’, Analytica Chimica Acta1999 385 131–135.


34 J. Mortensen, A. Legin, A. Ipatov, A. Rudi-nitskaya, Y. Vlasov, K. Hjuler. ‘A flowinjection system based on chalcogenideglass sensors for the determination of heavymetals’, Analytica Chimica Acta 2000 403273–277.

35 Y. Kanai, M. Shimizu, H. Uchida,H. Nakahara, C. G. Zhou, H. Maekawa,T. Katsube. ‘Integrated taste sensor usingsurface photovoltage technique’, Sensorsand Actuators 1994 B20 175–179.

36 Y. Sasaki, Y. Kanai, H. Uchida, T. Katsube.‘Highly sensitive taste sensor with a newdifferential LAPS method’, Sensors andActuators 1995 B24-25 819–822.

37 M. George, W. Parak, H. Gaub. ‘Highlyintegrated surface potential sensors’, Sensorsand Actuators 2000 B69 266–275.

38 Y. Murakami, T. Kikuchi, A. Yamamura,T. Sakaguchi, K. Yokoyama, Y. Ito,M. Takiue, H. Uchida, T. Katsube,E. Tamiya. ‘An organic pollution sensorbased on surface photovoltage’, Sensorsand Actuators 1998 B53 163–172.

39 S. Brown, R. Bear. ‘Chemometric techniquesin electrochemistry: A critical review’,Critical Reviews in Analytical Chemistry 199324(2) 99–131.

40 J. M. Diaz-Cruz, R. Tauler, B. Grabaric,M. Esteban, E. Casassas. ‘Application ofmultivariate curve resolution to voltam-metric data. Part 1. Study of Zn(II) com-plexation with some polyelectrolytes’,Journal of Electroanalytical Chemistry 1995393 7–16.

41 J. Menditeta, M. S. Diaz-Cruz, R. Tauler,M. Esteban. ‘Application of multivariatecurve resolution to voltammetric data. Part 2.Study of metal-binding properties of thepeptides’, Analytical Biochemistry 1996 240134–141.

42 J. Simons, M. Bos, W. E. van der Linden.‘Data processing for amperometric signals’,Analyst 1995 120 1009–1012.

43 F. Winquist, C. Krantz-Rulcker, P. Wide,I. Lundstrom. ‘Monitoring of milk freshnessby an electronic tongue based on volt-ammetry’Measurement Science and Technolgy1998 9 1937–1946.

44 F. Winquist, S. Holmin, C. Krantz-Rulcker,P. Wide, I. Lundstrom. ‘A hybrid electronictongue’, Analytica Chimica Acta 2000 406147–157.

45 F. Winquist, S. Holmin, C. Krantz-Rulcker,I. Lundstrom. ‘Flow injection analysisapplied to a voltammetric electronic tongue’,Int. J. Food Microbiology (at press).

46 C. Krantz-Rulcker, M. Stenberg, F. Win-quist, I. Lundstrom. ‘Electronic tongues forenvironmental monitoring based on sensorarrays and pattern recognition: a review’,Analytica Chimica Acta 2001 426 217–226.

47 T. Artursson. Licentiate Thesis no. 148:“Development of preprocessing methods formultivariate sensor data”. Linkoping Uni-versity 2000.

48 U. Koller. Licentiate Thesis no. 859, ‘Theelectronic tongue in the dairy industry’,Linkoping University 2000.

49 A. Carlsson, C. Krantz-Rulcker, F. Winquist.‘An electronic tongue as a tool for wet-endcontrol’, unpublished.

50 P. Ivarsson. Licentiate Thesis no.858,‘Artificial senses – New technologyin household appliances’, LinkopingUniversity 2000.

51 C. Soderstrom, H. Boren, F. Winquist,C. Krantz-Rulcker. ‘Analysis of mouldgrowth in liquid media with an electronictongue’, unpublished.

52 R. Borngraber, J. Hartmann, R. Lucklum,S. Rosler, P. Hauptmann. ‘Detection of ioniccompounds in water with a new polycarbonacid coated quartz crystal resonator’, Sensorsand Actuators 2000 B65 273–276.

53 S. Ezaki, S. Iiyama.‘Detection of interactionsbetween lipid/polymer membranes andtaste substances by quartz resonator’ Sensorsand Materials 2001 13(2) 119–127.

54 J. Kondoh, S. Shiokawa. ‘New applicationof shear horizontal surface acoustic wavesensors to identifying fruit juices’ JapanJournal of Applied PhysicsK 1994, 33, part I,3095–3099.

55 J. Kondoh, S. Shiokawa. ‘Liquid identifica-tion using SH-SAW sensors’, Technical digestof Transducers 95 – Eurosensors IX, Stockholm1995 716–719.

56 A. Campitelli, W. Wlodarski, M. Houm-mady. ‘Identification of natural spring waterusing shear horizontal SAW based sensors’,Sensors and Actuators 1998 B49 195–201.

57 V. Varadan, J. W. Gardner. ‘Smart tongueand nose’, Proc. SPIE International Soc. Eng.1999, 3673, 67–76.


58 C. Di Natale, R. Paolesse, A. Macagnano,A. Mantini, A. D’Amico, A. Legin, L. Lvova,A. Rudinitskaya, Y. Vlasov. ‘Electronic noseand electronic tongue integration forimproved classification of clinical and foodsamples’, Sensors and Actuators 2000 B6415–21.

59 C. Di Natale, R. Paolesse, A. Macagnano,A. Mantini, A. D’Amico,M. Ubigli, A. Legin,L. Lvova, A. Rudinitskaya, Y. Vlasov.‘Application of a combined artificialolfaction and taste system to the quanti-fication of relevant compounds in red wine’,Sensors and Actuators 2000 B69 243–347.

60 F. Winquist, P. Wide, I. Lundstrom. ‘Thecombination of an electronic tongue and anelectronic nose’, Sensors and Actuators 2000B69 243–347.

61 P. Wide, F. Winquist, P. Bergsten, E. Petru.‘The human based multisensor fusionmethod for artificial nose and tongue data’,Proc. IEEE Instrumentation and MeasurementTechnology Conference, St. Paul, Minnesota,USA May 1998.


Part C

Advanced Signal Processing and Pattern Analysis


12

Dynamic Pattern Recognition Methods and System Identification

E. Llobet

Abstract

The field of electronic noses has developed rapidly in the past few years. There aremore than 25 research groups working in this area and many companies have devel-oped commercial instruments. Most of the work found in the literature and commer-cial applications, however, relate to the use of traditional static pattern analysis meth-ods, based on either statistical or neural approaches. In this chapter, the emerging fieldof the dynamic analysis of the gas/odor sensor response is reviewed. The differentdynamic signal processing techniques used to date include well-established para-metric and non-parametric methods borrowed from the field of system identifica-tion. These include linear filters, multiexponential models, functional expansions,time series neural networks and others. The way in which all these techniquesmay solve electronic nose problems such as lack of selectivity, interference effects,and drift, is analyzed and some examples are discussed. Finally, a few guidelinesto select a suitable model for the dynamic modeling of application-specific electronicnose systems are suggested.

12.1

Introduction

It is only in the last few years that the use of dynamic signals from a multisensorsystem has received any significant attention. There are several reasons why dynamicsignal processing techniques are of importance to the field of electronic noses. Recentreports suggest that the dynamic response of solid-state gas sensors contains usefulinformation about the sensor kinetics and, these vary with both sensor and analyte.This additional information can be extracted from the transient response of a sensor toa controlled change in the analyte concentration (that is, concentration modulation) orto a change in the temperature of operation of the sensor (that is, temperature mod-ulation). In some applications the use of these techniques has resulted in an enhance-ment of the sensor array selectivity [1–4].


293293

Some sensors respond very slowly to weakly interacting odors. Non-steady statemeasurements are required when the environmental changes are on the sametime-scale as the sensor response. This may help to broaden the field of applicationof intelligent sensor systems (for example, continuous pollution monitoring).The sample delivery system and the sensor array are both parts of a dynamic system.

The time taken for the system to reach steady-state depends on parameters such asflow rate, volume of the test chamber, diffusion rate and reaction rate. When the sen-sors are modeled using steady-state values, the calibration time can be very long, espe-cially when a multicomponent calibration is performed. The calibration time is thetime needed to obtain the sensor response signal, which is a multi n-dimensionalnon-linear function of the analytical information of all detectable n components.From the calibration, important parameters such as partial sensitivities and selectivitycan be deduced. Because the dynamicmodeling allows for the estimation of the steady-state sensor responses [5, 6], it may significantly reduce the time of each calibrationexperiment.Even when sensors are exposed to identical gas mixtures, they do not give stable

responses over a long period of time. In other words, sensor signals tend to showsignificant temporal variation, typically referred to as long-term drift. This variationmay be due to unknown processes in the sensor system, like poisoning, aging orchanges in the environment, (that is, temperature and humidity). Drift may seriouslyaffect calibration. Therefore, when an intelligent sensor system is to be operated for along period of time, long-term drift should be addressed by the pattern recognitionalgorithms [7, 8].Finally, the baseline signal (in air) and response of a sensor can depend on its pre-

vious chemical history. These changes can be considered as a short-term drift. Forexample, a dynamic model that uses the knowledge of present and past inputs andoutputs of the sensor would be able to predict its baseline behavior.In the next section, a review of the different dynamic methods usually applied in

system identification is given. This is followed by a review of the techniques that areused to identify a model from measured data. Finally, the way, in which these tech-niques may solve electronic nose problems, ameliorate interference effects, the anddrift experienced is shown. Some guidelines to select a suitable model for application-specific electronic nose systems are then suggested.

12.2

Dynamic Models and System Identification

The techniques that are typically used to model the dynamic sensor response are bor-rowed from the field of system identification. System identification is the process ofdeveloping a mathematical representation of a physical/chemical dynamic systemusing experimental input-output data. The majority of methods that have been devel-oped to study engineering problems assume linearity and stationarity. In the context ofsensors, linearity implies that their calibration curve for all detectable components islinear, while stationarity implies that their dynamic response is not affected by time-

12 Dynamic Pattern Recognition Methods and System Identification294

varying trends. However, almost all real chemical transducers are characterized bynon-linear dynamics and response drift. This section reviews some models for thedynamic response of odor sensors.

12.2.1

Linear Models

Linear methods have been applied in diverse fields such as econometrics, biologicalsystems, control systems, and many others. Their application to the identification ofsensor array systems for gas analysis is recent [9]. The objective of the dynamical modelis to forecast the output of the sensor from knowledge of the input signals in dynamicconditions (forward modeling). Only the inversion of the model would allow us toidentify the input (gases/odors) given the output signals (inverse modeling). Themost common models are ARMA (Auto-Regressive Moving Average), ARX (Auto-Re-gressive with eXtra Input, also Auto-Regressive eXogenous), ARMAX, and Box-Jen-kins. These models are of interest in digital signal processing because the time seriescan be considered to be the output of a linear filter with a rational transfer function. Inthe following, their mathematical expressions are given. x[n], y[n] and e[n] are input,output and residual term or noise signals respectively. The generic relationship be-tween these variables is depicted in Fig. 12.1. Here x[n], y[n] and e[n] are discrete-time sequences, in which the time index n assumes integer values only. This is gen-erally the case in the context of chemical sensors, where the output signal is a sampledversion of the continuous-time sensor dynamic response.

ARMA ðq; pÞ : y½n� ¼Xq

i¼1

aiy½n� i� þXp

j¼0

bje½n� j� ð12:1Þ

The current value of the output is modeled using q past values of the output and thepresent and p past values of the noise. Two different sub-models of this one can beconsidered. The Auto-regressive (AR) and the Moving average (MA).

AR ðqÞ : yn� ¼Xq

i¼1

aiy½n� i� þ e½n� ð12:2Þ

MA ðpÞ : y½n� ¼Xp

j¼0


Moving average models are also known as all-zero models.

ARX ðq; kÞ : y½n� ¼Xq

i¼1

aiy½n� i� þXr

k¼0

ckx½n� k� þ e½n� ð12:4Þ

The present value of the output is modeled using a linear combination of the past qvalues of the output, and the present and past r values of the input.


ARMAX ðq; k; pÞ : y½n� ¼Xq

i¼1

aiy½n� i� þXr

k¼0

ckx½n� k� þXp

j¼0


Similar to the previous one but including a moving average term:

Box-Jenkins ðr; p : y½n� ¼Xr

k¼0

ckx½n� k� þXp

j¼0


In this model, the prediction of the output is made without the use of past values of theoutput. It uses present and past values of the input in addition to filtered noise.For the previous models, the parameter vector h is defined as

h ¼ ðal:::aq bl:::bp cl:::crÞ. Identifying the model requires the identification ofthe parameters in h. The choice of which type of model to use is highly problem-de-pendent, however, and there are different means of choosing a model for a particularproblem, which will be discussed later in this chapter.

State-Space Models

In the state-space form, the relationship between the input, noise and output signals iswritten as a system of first-order difference equations using an auxiliary state vector nn.This description of linear dynamical systems became increasingly important after Kal-man’s work on prediction and linear quadratic control [10]. Insights into the physicalmechanisms of the system can usually be more easily incorporated into space-statemodels than into the models described previously. The state-space model can be ex-pressed as:

nnþ1 ¼ AðhÞnn þ BðhÞx½n� þ ep½n� ð12:7Þ

y½n� ¼ CðhÞnn þ em½n� ð12:8Þ

where A, B and C are matrices of appropriate dimensions. h is a vector of parametersthat typically correspond to unknown values of physical coefficients, em is the measu-rement noise and ep is the process noise acting on the states. The disturbances em[n]and ep[n] are assumed to be sequences of independent random variables.

Fig. 12.1 A generic black-box model describes the relationship be-

tween the output (y), the measured signal or input (x) and disturbance

or noise (e)


12.2.2

Multi-exponential Models

The transient response of electrochemical and chemoresistive sensors when exposedto a volatile compound is of an exponential nature [11–13]. Therefore, it seems reason-able to model the response curves of these sensors by fitting a sum of exponentialfunctions:

xðtÞ ¼XN

i¼1

Gie�t=si ð12:9Þ

The task of modeling a curve with a set of exponential functions is not straightforward.Because exponential functions are not an orthogonal base of functions on the real axis,the determination of the set of coefficients (Gi, si, i ¼ 1,N) from finite-time and finite-precision samples of the response transient, will not have a unique solution. There-fore, an important issue is the determination of N, the number of exponential compo-nents that should be used to fit the response transient [14]. There are different de-convolution techniques that have been applied for data analysis. These include spec-tral methods, such as Gardner transform [15] or multiexponential transient spectros-copy (METS) [16] and non-spectral methods, such as non-linear least squares fitting[17], and Pade-Laplace [18] or Pade-Z transforms [19].Spectral methods do not need previous knowledge of the exponential terms. The

number of peaks in the spectrum gives directly the number of exponential termsused in the model. Furthermore, the shape of peaks can give information aboutthe adequacy of the model. For example, wider peaks suggest that two or more similartime constants have not been resolved. On the other hand, the non-linear least squaresfittingmethod approximates the response transient with a known number of exponen-tial terms, and thus is not suitable for component detection. Unlike spectral methods,which return a distribution that needs further analysis, non-spectral methods such asthe Pade-Laplace and Pade-Z transforms attempt to identify the finite set of coeffi-cients (Gi, si, i ¼ 1, N). Pade-Laplace and Pade-Z transforms perform data compres-sion and feature extraction simultaneously.The followingbriefly reviews someof themultiexponentialmodeling techniques. For

further details, the reader is referred to the references given. Those readers who are notinterested in themaths canskip thispart andproceed to sub-section12.2.3.Someresultson the use of such multi-exponential models are revised in sub-section 12.4.1.

Gardner Transform

This method, which is based on the Fourier transform, was introduced forty years agoby Gardner [15]. Later, the recovery of the spectrum was improved by applying a low-pass filter before the de-convolution step of the method [20].Assuming an experimental response function x(t) such that:

xðtÞ ¼XN

i¼1

Gie�ait ð12:10Þ


Eq. (10) can be rewritten using the spectrum gðaÞ:

xðtÞ ¼ð1

0gðaÞe�atda ¼

ð1

0

XN

i¼1

Gidða� aiÞ !

e�atda ð12:11Þ

Making the variable change p ¼ lnðtÞ, q ¼ �lnðaÞ, which changes the time axis fromlinear to logarithmic, Eq. (11) becomes:

xðepÞ ¼ð1

�1

gðe�qÞe�qexp �eðp�qÞh i

dq ð12:12Þ

Considering Eq. (12), the Fourier transform of epxðepÞ can be expressed as:

FðxÞ ¼ 1ffiffiffiffiffiffi2p

pð1

�1

epxðepÞejxpdp ¼ 1ffiffiffiffiffiffi2p

pð1

�1ð1

�1

gðe�qÞep�qexp �eðp�qÞh i

dq

0@

1Aejxpdq ð12:13Þ

Finally, by defining r ¼ p� q, Eq. (13) can be rewritten:

FðxÞ ¼ 1ffiffiffiffiffiffi2p

pð1

�1gðe�qÞejxqdq

ð1

�1

er � exp �er½ �ejxrdr ¼ GðxÞKðxÞ ð12:14Þ

Therefore, the Fourier transform GðxÞ of the spectrum gðe�qÞ can be found as theratio of F(x) and K(x), the Fourier transforms of the functions epxðepÞ andexp½�e�r�, respectively. The spectrum g(a) is related to the inverse Fourier transformof G(x) by:

gðe�qÞdq ¼ gðaÞa

da ð12:15Þ

The fact that g(a) and a are coupled in Eq. (15), biases the Gardner transform towardsmultiexponential curves for which the product of time constant and amplitude is si-milar for all the exponential components.

METS

METS is based on a numerical multi-differentiation of the response transient [16]. Thefirst order signal METS1 is defined as follows:

METS1ðtÞ ¼dxðtÞd lnðtÞ ¼

d

d lnðtÞ

ð1

o

GðaÞe�atda

24

35 ¼ �

ð1

0

atGðaÞe�atda ð12:16Þ


Making the variable change s ¼ 1=a, p ¼ lnðtÞ and q ¼ lnðsÞ, which is equivalent tochange the time axis from linear to logarithmic, Eq. (16) can be rewritten:

METS1ðpÞ ¼ �ð1

�1

hðp� qÞTGðqÞdq ¼ �hðpÞ � TGðqÞ ð12:17Þ

where hðpÞ ¼ exp½p� ep� and TGðqÞ ¼ eqGðeqÞ. The h(p) function has a bell shapewith a peak located at y ¼ 0. Therefore METS1 will present peaks at every time con-stant. The relative amplitude of peaks is proportional to the amplitude of the exponen-tial component. If the h(p) function were narrower, the method would give us the timeconstant distribution with improved resolution power. To reach this objective, we cansubstitute the h(p) function by hnðpÞ ¼ exp½np� ep� in Eq. (17), obtaining the nth ordersignal METSn:

METSnðpÞ ¼ �hnðpÞ � TGðqÞ ð12:18Þ

The differentiation of Eq. (18) leads to a recurrent formula for the trivial computationof METS signals from experimental data:

dMETSnðpÞdp

¼ nMETSnðpÞ �METSnþ1ðpÞ ð12:19Þ

The fact that hnðpÞ presents a peak at p ¼ lnðnÞ implies a shift towards the right of thereal axis. This results in a distortion of the spectrum. Because high orderMETS signalsare obtained by successive differentiation, the method may become very sensitive tohigh-frequency noise.

Pade-Laplace

This method is based on the theory of Pade approximants and the Laplace transform[18, 19]. The Laplace transform of the response function defined in Eq. (10) is:

XðsÞ ¼ð1

0

e�stxðtÞdt ¼XN

i¼1

Gi

s� 1=sið12:20Þ

The Pade-Laplace method proceeds in three steps to estimate the Laplace transform ofthe response transient. First, the Laplace transform is approximated at an expansionpoint s0 by using a Taylor series:

XXðsÞ ¼XK

k¼0

1

k!

dk

dskXðsÞjs¼s0

ðs� s0Þk ð12:21Þ


where

dk

dskXðsÞjs¼s0

¼ð1

0

ð�tÞkxðtÞe�s0tdt ð12:22Þ

Second, a Pade approximant is computed for the expression (21). Pade approximantsare rational expressions obtained by dividing two polynomials P(s) and Q(s). Thepower series expansion of a Pade approximant [M/N](s), agrees with the Taylor seriesup to the term sMþL.

½M=N�ðsÞ ¼ PðsÞQðsÞ ¼

p0 þ p1sþ :::þ pMsM

q0 þ q1sþ :::þ qNsN

ð12:23Þ

And third, the partial fraction expansion of the Pade approximant yields the time con-stants and amplitudes from the poles and residues of the expansion, respectively.When the order of the approximant exceeds the true number of exponentials, unstable(that is, artificial) poles will become noticeable. Therefore, the method requires thecomputation of the [i, iþ 1] approximants for i ¼ 0; :::;N.

Pade-Z

The method is similar to the Pade-Laplace, but it employs the discrete Z-transforminstead of the continuous Laplace transform. If x[k] is the sampled version of the re-sponse transient x(t):

x½k� ¼XN

i¼1

Gie�kT=si ð12:24Þ

then, the Z-transform is:

X ½k� ¼XN

i¼1

Gi

z

z� e�T=sið12:25Þ

Similarly to the Pade-Laplace method, the Z-transform is approximated by its Taylorseries expansion at a point z0 and the [i/i ] (i ¼ 1; :::;N) Pade approximants are com-puted for the Taylor expansion.

12.2.3

Non-linear Models

Chemical sensors are non-linear for high concentrations. Most of them are inherentlynon-linear even at low concentrations. Transport, adsorption and reaction processestaking place at the sensor include intrinsic non-linear dynamics. Thus, an electronicnose instrument can be represented as a non-linear system.


The analysis of non-linear systems poses many problems that do not appear in theirlinear counterparts. For instance, the law of superposition cannot been applied and theaddition of two input signals may lead to unknown results. Traditionally, the methodsused to identify non-linear systems are parametric methods that make assumptionsabout the structure of the system. If the structure is not accurate enough, the modelwill not work for all inputs. Recently, a few non-linear time series and other non-linearmodels have been proposed. Some of them will be reviewed briefly below.

Non-Linear Time Series

Some of the non-linear models are introduced in this section. The reader is referred tothe work of Tong [21] for a more comprehensive survey. One of the more importantclasses of non-linear models is the class of non-linear auto-regression. y[n] is said tofollow a non-linear auto-regressive mode of order k if there exists a non-linear functionf such that:

yn� ¼ f ðy½n� 1�; y½n� 2�; :::; y½n� k�; e½n�Þ ð12:26Þ

where e[n] is noise. As a ‘dual’ to non-linear auto-regressive models, we may have non-linear moving average models (for example, of order q):

y½n� ¼ gðe½n�; e½n� 1�; :::; e½n� q�; qÞ ð12:27Þ

q being a vector of parameters.Since the most important linear time series model is the ARMA model, it seems

natural to develop a non-linear generalization of it. For suitable k and q:

nn ¼ ð1; e½n� qþ 1�; :::; e½n�; y½n� kþ 1�; :::; y½n�ÞT ð12:28Þ

nn is called a carrier vector. Choosing suitable matrices, F, G and H, we may achievethe non-linearisation of ARMA models by introducing:

nn ¼ Fðnn�1Þnn�1 þGðnn�1nn; y½n� ¼ Hnn�1 ð12:29Þ

This is formally equivalent, under suitable choices of F, G and H to:

y½n� ¼Xq

i¼1

aiðnn�1Þy½n� i� þ v½nn�1� þXp

j¼0

bjðnn�1Þe½n� j� ð12:30Þ

The carrier vector can be regarded as a state vector and the model above as a state-dependent model (SDM) [22].

Functional Expansions

Functional expansions were studied by Volterra [23] and Wiener [24]. They are validrepresentations of non-linear systems under very weak assumptions (stationarity). The


concept of a functional was introduced to describe the input/output relationship of asystem. Assuming that x(t) is the input and y(t) the output, then:

yðtÞ ¼ F½t; xðt 0Þ; t 0 � t� ð12:31Þ

The task of modeling consists of obtaining a mathematical expression for the functio-nal F. This is to identify the input/output map of the system, determining the effect ofpast values of the input on the output. In the case of a non-linear time invariant system,F can be expressed as a Volterra functional expansion of the form:

y½t� ¼X1

n¼1

ð1

0

. . .

ð1

0

zfflfflfflfflffl}|fflfflfflfflffl{n

knðs1 . . . snÞxðt� s1Þ . . . xðt� snÞds1 . . . dsn ð12:32Þ

The kernels kn ðs1; :::; snÞ constitute the descriptors of the system dynamics. The nth

kernel attains the effect of the cross interaction of n past values of the input on theoutput. Wiener redefined the basis functionals so that they were orthogonal for whiteGaussian inputs.

Block-Structured Network Models

Block-structured network models consist of interconnections of two different classesof blocks, which implement either dynamic linear models or static non-linear models.

Fig. 12.2 Several block-structured models for bi-input systems. Ni

blocks are static non-linear models and Li blocks are dynamic linear

models. (From S. Marco et al., Sensors and Actuators B, Vol. 34, pp.

213–223 ª1996 Elsevier Science, with permission)


Thismodeling strategy is closely related to the functional expansionmethod, because aclose examination of the relationship between the Wiener kernels is necessary to de-termine the topology of the network. This method is preferred by some authors tofunctional expansion because of the difficulty involved in interpretation of the ker-nels. Furthermore, block-structured models may be related to the inner structuresof the system. Figure 12.2 shows some of the different topologies (for bi-input sys-tems) typically used in the block-structured approach. The reader is referred to thework of Chen et al. [25, 26], where a systematic structural classification procedureemploying Wiener kernels is reviewed.

Neural Networks

In recent years, multi-layer perceptrons (series-parallel identification method) andtime-delay or recurrent neural networks (parallel identification method) have beenproposed for system identification and modeling purposes [27]. It has been provedthat the output of an artificial neural network (ANN), whose inputs are delayed valuesof the input signals, can be expressed as an infinite Volterra series [27]. In this case,since the expansion is not limited to the first or second kernels, the network is able tomodel highly non-linear relations if there are enough hidden neurones. The output ofthe network is a non-linear function of q delayed outputs and p delayed inputs:

y½kþ 1� ¼ f ðy½k�; y½k� 1�; . . . ; y½k� q�; x½k�; . . . ; x½k� p�Þ ð12:33Þ

From the point of view of system identification, a multilayer neural network can beassumed to be a non-linear map. The elements on the weight matrices are parameters,

Fig. 12.3 Architecture of a recurrent network, which could be used to identify a single-input system


whose optimum values should be found by training the ANN over a training set. Fi-gure 12.3 shows the topology of a time-delay neural network and Fig. 12.4 shows thedifferences between the series-parallel and the parallel identification methods. Thefirst method is generally applied for calibration. The stability of the second me-thod, which uses a neural network with feedback, cannot be assured [28–30].

12.3

Identifying a Model

The techniques used to identify a model from measured data typically consist of para-metric or non-parametric approaches. With non-parametric techniques, very few as-sumptions about the system to be modeled are required, and therefore apply moregenerally. However, parametric techniques can sometimes lead to better results, espe-cially when the amount of data is limited (that is, short time series). This section re-views the different techniques available for model selection.

12.3.1

Non-Parametric Approach

A linear time-invariant system can be described by its transfer function or by the cor-responding impulse response. A non-linear time-invariant system can be describedusing functional expansions (Wiener kernels). Transfer functions, impulse responses

Fig. 12.4 (a) In the series-par-

allel system identification me-

thod, the neural network is

supplied with lagged inputs and

outputs of the system to be

identified. (b) The parallel sy-

stem identification method uses

a neural network with feedback


and Wiener kernels may be determined by direct techniques. Such methods are oftencalled non-parametric since they do not explicitly employ a parameter vector in thesearch for a best description.

12.3.1.1 Time-Domain Methods

Time-domain methods include impulse-response analysis, step-response analysis andcorrelation analysis. Impulse response analysis is impractical because many processesdo not allow impulse inputs of such amplitude that the error is insignificant comparedto the impulse response coefficients. Step-response analysis can furnish some basiccharacteristics to a sufficient degree of accuracy (that is, delay time, static gain, dom-inating time constants). Using correlation analysis, an estimate of the impulse re-sponse g(t) can be obtained, through the cross-correlation of input (white noise)and output signals. If the input is white noise so that RRxxðsÞ ¼ ads, then

ggðsÞ ¼RRyxðsÞa

ð12:34Þ

where

RRyxðsÞ ¼1

N

XN

t¼s

yðtÞxðt� sÞ ð12:35Þ

If the input is not white noise, then an estimate of the auto-correlation of the input canbe obtained as

RRxxðsÞ ¼1

N

XN

t¼s

xðtÞxðt� sÞ ð12:36Þ

and solve

RRyxðsÞ ¼XM

k¼1

ggðkÞRRxxðk� sÞ ð12:37Þ

to estimate g(k).To identify ARMA models, the estimated auto-correlation and partial auto-correla-

tion functions of the input signal provide valuable information. Auto-regressive pro-cesses of order 1,2,… are fitted successively and the residuals calculated. The partialauto-correlation is the correlation of these residuals and the input signal. If there is asharp cut-off in the estimated auto-correlation function after lag k, the model can beidentified as an MA(k). If the auto-correlation function tails-off but the partial auto-correlation function shows a sharp cut-off after lag q, the model can be identified as anAR(q). If both functions tail-off, an ARMA model is to be used. If the auto-correlationfunction does not tail-off nor cut-off, the process is non-stationary. If this occurs, thedata can be successively differenced until the resulting time series appears to be sta-tionary. Differencing provides a simple way of removing trends in the data. The first


difference of a time series y[k], Dy[k] is defined by the transformationDy½k� ¼ y½k� � y½k� 1�. Higher order differences are defined by successive applicationof the transformation. In this case, an ARIMA (auto-regressive integrated movingaverage) model is identified. ARIMA is an extension to the ARMA class of processesas empirical descriptors of non-stationary time series. Differencing the input signalincreases the noise level, therefore smoothing of the resulting signal may be necessary.Figure 12.5 illustrates the identification process. There are many different criteria thatcan be used to select the order of the model. In general, they do not provide the samemodel order for the analyzed series of data. The reader is referred to the works of Ljung[31] and Diggle [32] for a more detailed discussion.If the system shows a non-linear behavior, it is possible to use either a linear model

(this can be a good check for the relative importance of the non-linear component inthe system) or a non-linear model (Wiener kernels). The reader is referred to the workof Lee and Schetzen [33], where a non-parametric method based on correlation tech-niques is introduced for the estimation ofWiener kernels. This method uses Gaussianwhite noise as the input to the system. The idea of using white noise as a stimulus inorder to identify a system is based on the fact that the system is tested on all the pos-sible inputs regarding values and frequencies (depending on the length of the test).Another approach developed by Barker [34, 35] consists of using multi-level pseudo-

Fig. 12.5 Flowchart illustrating the identification of ARMA/ARIMA processes


random sequences. In a Volterra series expansion, it becomes extremely difficult toidentify kernels of order three or more. Therefore, these time-domain methods areaimed at identifying second-order kernels.

12.3.1.2 Frequency-Domain Methods

The frequency response of a system H(jx) may be determined from an estimation ofits transfer functionH(s) by setting the complex Laplace s parameter to jx. More com-monly it can be determined from the time-domain signals by taking a Fourier trans-form (continuous or discrete) of the input x(t) and output y(t) signals, namely

HðjxÞ ¼ YðjxÞXðjx ; ð12:38Þ

where YðjxÞ ¼ 1ffiffiffiffiffiffi2p

pð1

�1

yðtÞe�jxtdt; XðjxÞ ¼ 1ffiffiffiffiffiffi2p

pð1

�1

xðtÞe�jxtdt

It should be noted that the Fourier transform is a linear integral transform and x(t) andy(t) must be non-trivial (that is, non-zero) to determine the frequency response usingthis method. When the input x(t) is a periodic signal, the estimate of the frequencyresponse is only of significance at the frequencies present in the input.When the inputis not periodic (that is, a realization of a stochastic process), the quality of the estimatefalls at those previous frequencies but is a better estimate at the other frequencies. Theestimates at different frequencies are asymptotically uncorrelated. This makes theestimate of the frequency response relatively crude in practical situations [31].Spectral analysis for determining transfer functions of linear systems was developed

from statistical methods for spectral estimation. The reader is referred to the work ofBrillinger [36] for a detailed account of the method. The only way to improve the poorvariance properties of the transfer function estimate is to assume that the values of thetrue transfer function at different frequencies are related. Since the transfer functionestimates at neighboring frequencies are asymptotically uncorrelated, the variance canbe reduced by averaging over these (for example, using a window such as Bartlett,Parzen or Hamming). While a broad window leads to biased estimates and low var-iance, a narrow window leads to unbiased estimates but high variance (appearance ofspurious peaks). Another way of smoothing the transfer function estimate is to splitthe data set into different sub-sets. The estimates over different sub-sets will be un-correlated and averages over these can be formed.In the frequency domain, the relationship between the input X(jx) and the output

Y(jx) of a non-linear system is the Volterra functional series expansion of the form:

YðjxÞ ¼ HðjxÞXðjxÞ þH2ðjx1; jx2ÞXðjx1ÞXðjx2Þ þ . . .

þHnðjx1; jx2;K; jxnÞXðjx1ÞXðjx2Þ ^ XðjxnÞ þ . . . ð12:39Þ

where H(HðjxÞ) is the linear system frequency response. The identification of non-linear characteristics in the frequency domain is, in practice, restricted to the second-


order kernel transformation H2ðjx1; jx2Þ, because higher order Volterra kernel trans-formations are difficult to display and interpret [37]. Barker [38] described a method toestimate the kernel transformations, which uses signals obtained from multi-levelmaximum length pseudo-random sequences [37].

12.3.2

Parametric Approach

In this approach, a set of candidate models is selected and parameterized as a modelstructure, using a parameter vector h. The search for the best model within the setbecomes a problem of determining or estimating h. To do so, two main strategiescan be considered: minimizing prediction errors and correlating prediction errorswith past data.The first approach employs well-known procedures such as the least-squares meth-

od and the maximum likelihood method, and is closely related to the Bayesian max-imum a posteriori estimation. The second approach is based on the correlation be-tween the prediction error and past data. Ideally, the prediction error of a good modelshould be independent of past data. A pragmatic way of checking this condition is thatif the prediction error is correlated with the past data, then there wasmore informationavailable in the past data about the actual output than was picked up by the model(predicted output). Therefore, the model was not ideal. See Ljung [31] for a detailedreview of these methods.The non-parametric approach introduced by Lee and Schetzen [33] for the estima-

tion of the kernels that characterize a non-linear system, requires long data sequencesfor optimum performance. Short data sequences lead to significant errors in the es-timated kernels. Haber [39] introduced a parametric method to estimate the kernelswhich reduces their variance, leading to a better estimation when short data series areavailable. Billings [40] described a method to compute second-order kernel transfor-mations,H2ðjx1; jx2Þ, which includes estimating a non-linear auto-regressive movingaverage (NARMA)model (see Eq. 33). The frequency responses can be computed fromthe postulated model [40, 41]. To estimateHðjxÞ, x[k] is set to ejxkD and the coefficientsejxkD are equated. To estimateH2ðjx1; jx2Þ, x[k] is set to ejx1kD þ ejx2kD and the coeffi-cients ejðx1þx2ÞkD in the model are then equated. This method requires long data se-quences to be accurate.When using block-structured models, accurate kernel estimation is crucial for the

identification of the topology of interconnection. Since the estimation of high orderkernels is impractical, especially with short data series, the topology of the system isusually selected from a set of universal representations [42]. This selection can bebased on a previous knowledge (or postulation) of the inner characteristics of the sys-tem or by performing a structural testing procedure introduced by Chen [26]. If thesystem being studied does not satisfy the test criteria, the structure can be rejected andanother selection can be made. On the other hand, if the system satisfies the test itcannot be concluded that it has this specific structure. Once the topology has been


selected, the linear time-variant blocks can be identified using cross-correlation tech-niques and the static non-linear blocs are usually identified by fitting a polynomial [43].A particular case of a parametric approach is the use of multi-exponential models.

Multi-exponential models, such as Gardner transform, METS, Pade-Laplace and Pade-Z transforms are parametric because an exponential response transient is assumed. Toimplement the Gardner transform, the experimental transient must be sampled in thelogarithmic scale [20]. As the transient is normally sampled at constant time intervals,an interpolation step must be performed, which can be difficult if the experimentalcurve is noisy. Furthermore, the de-convolution of the FFT of the spectrum favorshigh-frequency components (experimental noise). Therefore, the low-pass filteringof the FFT of the spectrum prior to the de-convolution process, leads to a better sig-nal-to-noise ratio at the price of a lower spectral resolution. Similarly to Gardner trans-form, METS requires logarithmic sampling (or interpolation) of the experimental re-sponse transient. But the implementation of the method is easier compared to theGardner transform [16]. Pade-Laplace and Pade-Z transform methods require the se-lection of an expansion point to approximate the Laplace (or Z) transform by a Taylorseries. The selection of the expansion point is an important issue because both meth-ods will not work properly for all the values of this point. If the expansion point is toosmall, the numerical integration in Eq. (22) will not converge in the time range pro-vided by the samples of the experimental measurements. If the expansion point is toolarge, the numerical integration in Eq. (22) will truncate the data too early and theslowest poles will not be identified. There exist several heuristic search methods tofind an optimal value for the expansion point [18, 44].

12.4

Dynamic Models and Intelligent Sensor Systems

In this section we briefly review the modeling techniques in the context of electronicnose systems. The models and techniques used so far aim to enhance the sensor arrayselectivity, to reduce the time necessary for calibration (for example, forecasting thesteady-state response using the transient response) and to counteract drift. A summaryof the main approaches is shown in Table 12.1. The main ones will be discussed inmore detail later. Before applying any technique to dynamically model the sensor sys-tem, sensors that are not relevant for the specific application, or that do not workproperly, should be eliminated. This requires careful ‘pre-analysis’ of the system.The use of classical techniques such as PCA may be very helpful in this preliminarystage. The reader is referred to Chapter 5 of this book for a detailed account of thedifferent pre-processing techniques.


Tab. 12.1 Types of modeling approaches in intelligent sensor systems

Modeling Technique Identification Technology Application Ref.

Linear filters, andstate-space models

Parametric ARMA,

sensor oriented model

Thick-film SnO2 Calibration time

reduction

[5]

Parametric ARX,

sensor oriented

1 sensor 4 QMB Sensor response

Prediction

[45]

Parametric Box-Jen-

kins, sensor oriented

polymer coated 10

MOSFETs

2 thick-film SnO2

Identification of

2 gases. Drift rejection

[8]

Parametric AR,

sensor oriented

6 QMB Identification of

3 vapors

[9, 46]

Parametric state-

space model, system

oriented

polymer coated Quantitative analysis

of ternary mixtures

[47]

Parametric Box-

Jenkins FIR, sensor

oriented

4 QMB polymer

coated, 2 SnO2

Quantitative analysis

of 2 vapors

[48]

6 BAW polymer

coated

Multiexponentialmodels

Parametric, sensor

oriented

Resistive

(metal oxide and

conducting polymer)

Feature extraction

for odor recognition

[5, 13]

Functional expansions(non-linear)

Non-parametric, 6 QMB Sensor response [48, 49]

correlation

techniques.

polymer coated prediction [50]

sensor oriented

Parametric,

sensor oriented

4 QMB

polymer coated

Sensor response

prediction

[45]

Block-structured Parametric

combining correlation

and polynomial fitting

6 QMB

polymer coated

Structure

identification,

response prediction

[45, 50]

Neural networks– SOM

– Time-delay

– ART

– fuzzy ART

Non-parametric,

system oriented,

adaptive

Arrays of SnO2,

MOSFET and QMB

polymer coated

Gas/aroma

identification,

drift rejection

[45, 51, 52]

[6, 48, 53]

[54–60]

Other techniquesAd-hoc models

through odor or

temperature

modulation and

noise techniques

Parametric,

sensor oriented

models or FFT

techniques

Metal oxides,

conducting

polymers, QMB

polymer coated

Sensor selectivity

enhancement,

gas/aroma

identification and

quantification

[3, 61, 62]

[2, 4, 63, 64]

[1, 65–70]


12.4.1

Dynamic Pattern Recognition for Selectivity Enhancement

To date most of the attempts to use transient information in the sensor signal arebased on ad hoc models. These models allow for the estimation of parameters thatcharacterize the transient response conferring some selectivity on the sensors. Gen-erally, an advantage of these models is that they account for physical and chemicalproperties of the sensing material (e.g. diffusion, reaction). Therefore, some insightinto the sensors’ dynamic behavior can be realized [63, 64]. Their main weakness is

Fig. 12.6 Results of a PCA analysis of the response of a four-element

tin-oxide electronic nose to three organic volatile compounds using

static (a) and dynamic (b) signals. Results of a PCA of the response to

binary mixtures using static (c) and dynamic (d) signals. (From E.

Llobet et al., in Proceedings of IEEE Solid-state Sensors and Actuators

Conference, Transducers, Vol. 2, pp. 971–974, ª1997 IEEE, with per-

mission)


that transient signals are influenced by previous measurements (memory effect) andby drift (for example, aging of the sensor, variations in temperature or humidity). Sincethese aspects are not considered by the models, the pattern recognition ability of asensor system which is initially learnt can deteriorate after a period of time. Fig-ure 12.6 shows the PCA results when an array of 4 thick-film tin oxide gas sensorswere used to identify different volatile organic compounds and their binary mixtures[2]. The use of transient signals such as the rise time of the sensor conductance whenthe odor concentration varies stepwise, helps in the identification task. The identifica-tion of single components, using a feed-forward back-propagation trained neural net-work gave a 76% success rate (using static signals only) and a 100% success rate(using both static and dynamic signals). The success rate in the identification of binarymixtures increased from 75% (using static signals) to 86% (using static and dynamicsignals).In a recent study [13], different techniques to identify multiexponential models were

used to analyze the response transients of a 32-element sensor array. The sensors werebased on conducting polymers and the modeling was carried out in the context of odorrecognition. Figure 12.7 shows a typical response of the polymer sensors to fruit juice.Two spectral methods (Gardner transform and METS) and two non-spectral methods(Pade-Laplace and Pade-Z transform) were investigated. The results of applying thesemethods to the sensors’ responses are shown in Fig. 12.8. Both non-spectral methodsoutperformed the spectral ones. The slow sampling rate of the transients and the ex-perimental noise required previous smoothing of the experimental signals. The Gard-ner transform was found to be very sensitive to the smoothing process. In METS, thedifferentiation of the transient and associated decrease of the signal-to-noise ratio,prevented higher-order signals to be of use. Therefore, both spectral methods wereable to identify one exponential component. Non-spectral methods were found tobe less sensitive to experimental noise and the response transients could be modeledwith two exponential components. These methods led to very similar results. The dotsin Fig. 12.8 (bottom) are the exponential components (Gi, si) for each sensor. While theclusters with small time constants account for the initial transient of the signal, thescattered clusters with higher time constants represent the steady state. These scatter

Fig. 12.7 Typical response of a 32-element conducting polymer sensor array to fruit juice


diagrams can be though of as odor signatures, which can be of use for odor recogni-tion. However, since exponential functions are not an orthogonal basis of functions,further work is needed to check the repeatability of the extracted signatures.Other dynamic pattern recognition methods for selectivity enhancement consist of

modulating the working temperature of the sensor or using an a.c. interrogation tech-nique. The reader is referred to chapters 5 and 16 of this book for a detailed account ofthese methods.A variation of the a.c. interrogation technique is the pseudo-random binary se-

quence (PRBS) interrogation technique [69]. A PRBS voltage is applied to the gas sen-sor electrodes and the output signal is then taken from a resistive load connected inseries to the resistive (conducting polymer) sensor. PRBS are easy to obtain and have anearly uniform power spectral density (PSD) over a wide frequency band. Figure 12.9shows a PBRS generator and the signal PSD. PRBS are interesting because they aredeterministic, and thus measurements are repeatable. The output signal is processedusing the FFT to convert it from the time domain to the frequency domain. The energyspectral density (ESD) of the output signal is a characteristic feature of the gas sensorin the presence of an odor. Figure 12.10 shows the ESD of a conducting polymer gassensor in the presence of methanol and acetone [69]. The relative amplitude of peakscan be seen as a fingerprint for the tested odors.Another strategy consists of measuring the PSD of the random resistance fluctua-

tions of a d.c. biased resistive sensor. It has been shown [70] that for a conductivepolymer sensor, a significant variation in the PSD is obtained in the presence of odors.

Fig. 12.8 Results of applying different multiexponential methods to model the tran-

sients shown in Fig. 12.7. Top left: Gardner transform. Top right: METS1. Time constants

(si) and amplitudes (Gi) derived from the Pade-Laplace (bottom left) and Pade-Z

(bottom right) methods, respectively. After [13]


12.4.2

Calibration Time Reduction

Some applications of sensor response prediction aim to reduce the time necessary tocalibrate the sensor array for the gases/odors of interest. Results with ARMA andmulti-exponential models applied to the dynamic response of tin oxide sensor arrayshave been reported [5]. The dynamic models were used to predict the static response ofthe sensors to small concentrations of nitrogen dioxide (0–9 ppm). Because the auto-correlation for the transient response of the sensors tailed-off and the partial autocor-relation cut-off after lag 1, an AR(1) model was identified (see Fig. 12.5). However, thisAR model was found to underestimate the static response of the sensors. The com-putation of the first-order METS (see Eq. (17)) for the transients, which showed twopeaks, suggested that two exponentials were suitable for the modeling of the sensorresponse. Table 12.2 shows the relative errors made by the dynamic multiexponentialmodel, which performed better than the AR(1) model in the extrapolation of the gasconcentration. In this application, the prediction of the static response from the initialpart of the dynamic response permits a reduction of the calibration time by a factor offour.

Fig. 12.9 A pseudo-random binary sequence (PRBS) generator

and power spectrum density (PDS) of the generated sequence.

(Reprinted fromM.E.H. Amrani et al., Sensors and Actuators B, Vol. 47,

pp. 118–124 ª1998 Elsevier Science, with permission)


12.4.3

Building of Response Models

Dynamic measurements are interesting when the odors or the environmental condi-tions undergo changes with the same time-scale as the sensor response times. Thissituation is not uncommon because chemical sensors are often slow responding de-vices. In such cases, the inversion of the dynamical model allows for the concentra-tions input to the sensor or sensor array to be reconstructed. Another advantage ofdynamical models compared with static models is the possibility of predicting futuresensor responses from the knowledge of their past and present inputs and outputs.Methods of dealing with noise that allow for calculating the impulse response (of

linear systems) or the Wiener kernels (of non-linear systems), using the correlation

Fig. 12.10 (a) Energy spectral density of the gas sensor response to 500 ppm

methanol vapor. (b) Energy spectral density of the gas sensor response to 500 ppm

acetone vapor. (Reprinted from M.E.H. Amrani et al., Sensors and Actuators B,

Vol. 47, pp. 118–124 ª1998 Elsevier Science, with permission)


approach, appear to be useful for constructing models for the sensor response to dif-ferent odors. Linear filters that use lagged values of the input and the output (i.e.previous values of these signals) to characterize the sensor (sensor oriented mod-els) or the sensor array (system oriented models), identified using parametric ap-proaches, such as the least-squares method, are also promising. In [8] Box-Jenkinslinear filters were applied to model an array of metal oxide and MOSFET odor sensorsin the presence of four alcohols and water vapor. Five models for each sensor werecreated (one for each alcohol and one for water vapor). The classification was done inprediction error space, and the alcohol whose model gave the lowest total squaredprediction error for all sensors was identified as the unknown odor (Bayesian ap-

Tab. 12.2 Relative errors made by a multi-exponential model in the

extrapolation of the concentration value of NO2 at different calibration

times. (Reprinted from C. DiNatale et al., Sensors and Actuators B,

Vol. 24–25, pp. 578–583, ª1995 Elsevier Science, with permission)

Time (s) Error at 1 ppm (%) Error at 6 ppm (%) Error at 9 ppm (%)

100 55.2 35.8 13.1

200 17.1 4.7 7.1

400 7.8 2.5 3.7

800 1.3 0.3 0.6

Fig. 12.11 Prediction errors for all the 5 models used for each sensor

when the measured gas was 1-propanol. (Reprinted from

M. Holmberg et al., Sensors and Actuators B, Vol. 35–36,

pp. 528–535, ª1996 Elsevier Science, with permission)


proach). Figure 12.11 shows the total sum squared prediction error for all sensors andfor every model when the measured gas was 1-propanol. The 1-propanol model givesthe lowest prediction error in almost all cases, mostly leading to a correct classification.However, linear and non-linear models constructed using input-output data (black-boxmodels) do not give any insight into the inner structure of the sensors. In other words,it is not possible to discuss the identified model in terms of physical or chemical prop-erties of the system. On the other hand, block-structured models are more related tothe intrinsic characteristics of the sensing mechanisms. Figure 12.12 shows thescheme of a two-input block-structured model of a polymer-coated quartz-microba-lance sensor in the presence of n-octane and toluene [50]. The impulse response ofthe two linear blocks, which describe all the memory effects of the system, were es-timated using the cross-correlation approach. The static input-output non-linearity wasestimated by fitting a five-order polynomial. However, thismethod has not been widelyapplied because the identification of themodel is complicated. In fact, the use of a non-parametric approach, such as the cross-correlation method, to estimate the impulseresponse with low errors, requires long data sequences. This can result in time con-sumingmeasurements to identify the sensor array or even worse, can be impractical insome applications.

12.4.4

Drift Counteraction

Because all of the approaches described above include memory effects, they are gen-erally useful to address the problem of short-term drift (effects in the present responseof the system due to measurements in its recent past). Another strategy consists of

Fig. 12.12 Complete scheme of the estimated two-input Wiener

model of a polymer-coated QMB. (Reprinted from F. Davide et al.,

Sensors and Actuators B, Vol. 24–25, pp. 830–842, ª1995 Elsevier

Science, with permission)


using neural networks with residual plasticity. This allows the networks to deal effec-tively with small variations in the sensor response [51, 55].Long-term drift caused by sensor poisoning or aging implies that the system under

identification is non-stationary. All the methods, except the neural network approach,assume that the sensor system is stationary and thus, are not suitable to analyze theeffects of long-term drift. It has been shown that self-organizing maps (SOM) withresidual plasticity can help to maintain the pattern recognition ability of a sensor sys-tem affected by drift [55]. The reader is referred to Chapter 13 for a detailed discussionon SOMs. Figure 12.13 shows the identification performances of an electronic nosebased on six tin-oxide gas sensors and static and adaptive SOMs. The gases measuredwere H2, CO, CO2, CH4, and binary mixtures of H2 with CO and CH4 with CO. Itshows that if an adaptive SOM is used, the identification ability of the electronicnose remains almost unchanged when the drift in the sensor response is up to20%. However, SOMs with residual plasticity require the frequent measurementof all the patterns. If this requirement is not fulfilled, patterns that seldom occurwill be forgotten.Recently, in some preliminary work, adaptive resonance theory (ART) neural net-

works have been proposed to deal with sensor drift [56]. The short-timememory of thenetwork gives it some plasticity to adapt to sensor drift, while the long-time memorymay give the necessary rigidity to avoid forgetting previously learnt patterns. ARTMAP(adaptive resonance theory supervised predictive mapping) and fuzzy ARTMAP arenon-parametric, adaptive networks that are well suited to solve pattern classificationproblems [71, 72]. With other adaptive algorithms, the learning of new events tends towash away the memory traces of previous, but still useful, knowledge. ARTMAP andfuzzy ARTMAP contain a self-stabilizing memory that permits accumulating knowl-edge to new events in a non-stationary environment [73]. Very recently, it has been

Fig. 12.13 Comparison of the

identification performances of

non-adaptive and adaptive

SOMs of a six-element tin-oxide

gas sensor array in the presence

of simulated drift. (From S.

Marco et al., in Proceedings of

IEEE Instrumentation and Mea-

surement Technology Conference,

pp. 904–907, ª1997 IEEE, with

permission)


shown that the incremental learning capability of fuzzy ARTMAP is very promising toaddress drift in electronic nose systems. In particular, the method has been success-fully applied to the classification of alcohols and coffees [57], the non-destructive de-termination of fruit ripeness [58–60] and the classification of bacteria [74, 75]. Thereader is referred to Chapter 13 for a detailed discussion on ART networks. Duringtraining the ARTa module was supplied with the response vectors of a four-elementtin-oxide sensor arrays, while the ARTb module was supplied with the correspondingcorrect categories. Using fast node commitment and slow node re-code, this networkperformed incremental learning without forgetting previous knowledge. These resultsare shown in Table 12.3, where the performance of fuzzy ARTMAP is compared toother neural paradigms, such as multi-layer perceptron (MLP) and learning vectorquantization (LVQ). The data were split in three data-sets to perform incrementallearning.

12.5

Outlook

There is no universal sensor system that can solve all odor or gas mixture analysisproblems. Instead there is a need to employ intelligent application-specific sensorsystems that are appropriate to the application. This means building-in intelligencethrough the development of suitable sensor structures, sensor materials and patternrecognition methods [76]. New pattern recognition methods should make use of thetransient information in the sensor signal to enhance the identification ability of thesystem. This requires the use of dynamic models for the sensor system that can ac-count for the drift in sensor parameters and thus extend the calibration period.

Tab. 12.3 Incremental learning on the three data-sets with Fuzzy

ARTMAP, [LVQ] and (MLP). For Fuzzy ARTMAP, the recode rate was

fixed to b ¼ 0:1. Number of patterns correctly classified/Total number

of patterns in the category. (Reprinted from E. Llobet et al., Meas. Sci.

Technol., Vol. 10, pp. 538–548, ª1999 IOP Publishing, with permis-

sion.)

Category Performance

Learning/Test sets a b c d e f g (%)

1 / 1 – 21/21 – 7/8 20/20 – – 98.0

[20/21] [8/8] [19/20] [95.9]

(20/21) (8/8) (18/20) (93.8)

2 / 1 and 2 10/10 28/29 7/8 16/17 19/20 – – 94.6

[9/10] [22/29] [8/8] [15/17] [18/20] [85.7]

(9/10) (27/29) (8/8) (15/17) (0/20) (70.2)

3 / 1, 2 and 3 24/24 28/29 7/8 25/26 33/35 23/24 29/29 96.0

[21/24], [28/29] [8/8] [14/26] [23/35] [23/24] [23/29] [80.0]

(23/24) (0/29) (0/8) (25/26) (15/35) (21/24) (27/29) (63.4)

12.5 Outlook 319319

The importance of many problems associated with current chemical sensor tech-nology is application specific. If the system has to analyze low levels of low reactivespecies, sensors tend to perform well. If the system has to analyze high levels of re-active species, poisoning of the sensors is likely and drift effects become very signifi-cant. The baseline of sensing devices (for example, metal oxides, polymeric chemor-esistors and polymer coated QMB) is sensitive to the operating temperature, the hu-midity and type of carrier gas [77]. Very often, the sensors require a long recovery timebetween measurements to reach their baseline. In continuous monitoring or repeatedmeasurement applications, the response of the sensors is influenced by their previoushistory (short-term memory effect). Under these constraints, the choice of a suitablemodeling strategy should be considered carefully:

* Non-adaptive models can be useful when the application implies the analysis ofweakly reacting species with systems where temperature and humidity are strictlycontrolled by the sample delivery system. Drift is likely to be small in such a system.

* Adaptive models are required when analysis of strongly reacting species is to beperformed and the sensors are likely to drift due to poisoning. These modelscan also handle drift caused by slight variations in the temperature and humidityof the carrier gas.

* Of the non-adaptive models, ad hoc parametric models are interesting because theymay give some insight into sensor behavior. The measured parameters can be feddirectly into well-established pattern recognition systems. Linear filters and non-linear models can be used to compensate for the short-term drift caused by thememory effect of the array when successive measurements are performed.

* The development of non-linear, adaptive models in which competition betweencomponent gases occurs may best be solved using neural paradigms.

Fig. 12.14 Selection of a dynamical PARC method for linear or quasi-linear problems


* SOMswith residual plasticity can be a good choice when frequent measurements ofall the patterns are performed. When this condition is not fulfilled, the ART ap-proach is a promising one.

* The on-line incremental learning capability of fuzzy ART is a very promising fea-ture for drift counter-action in electronic nose systems.

These basic ideas are contained in Figs. 12.14 and 12.15, where the suitability of aspecific dynamic model to a particular type of problem is shown [78]. In Fig. 12.14the sensor responses are considered to be linear or quasi-linear in concentration.This is generally the case when the species concentration is low, for example for con-ducting polymer resistive sensors, or when the concentration range is small and so is,step-wise, approximately linear. If the sensor response is non-linear in concentrationin a well-defined manner, a pre-processing linearization algorithm can be used [79].On the other hand, in Fig. 12.15, the selection assumes that the non-linear part of thesensor response is important and must be accounted for in the models.The first attempts to use the dynamic sensor signals in electronic noses have essen-

tially consisted of the development of ad hoc sensor-oriented parametric models. Todevelop a new generation of electronic noses, there is a need to extend these modelstaking into account the effects of environmental variables such as temperature andhumidity, and to implement improved adaptive models to counter-act sensor driftand poisoning.

Fig. 12.15 Selection of a dynamical PARC method for non-linear problems

12.5 Outlook 321321

References

1 R. E. Cavicchi, J. S. Suehle, K. G. Kreider,M. Gaitan, P. Chaparala. Sensors andActuators B, 1996, 33: p. 142–146.

2 E. Llobet, J. Brezmes, X. Vilanova,L. Fondevila, X. Correig. Quantitative vaporanalysis using the transient response of non-selective thick-film tin oxide gas sensors.In IEEE Transducers’97, 1997, Chicago, USA.Vol. 2, 971–974.

3 F. J. Auerbach, Pattern recognition usinggas modulation. In Transducers’95 – Euro-sensors IX. 1995. Stockholm, Sweden.

4 D. M. Wilson, S. P. Deweerth. Sensors andActuators B, 1995, 28, 123–128.

5 C. DiNatale, S. Marco, F. Davide, A D’amico.Sensors and Actuators B, 1995, 24–25,578–583.

6 M. Schweizer, J. Goppert, A. Hierlemann,J. Mitrovics, U. Weimar, W. Rosentiel,W. Gopel. Sensors and Actuators B, 1995,26–27, 232–236.

7 M. Roth, R. Hartlinger, R. Faul,H. E. Endres. Sensors andActuators B, 1996,35–36, 358– -62.

8 M. Holmberg, F. Winquist, I. Lundstrom,F. Davide, C. DiNatale, A D’amico. Sensorsand Actuators B, 1996, 35–36, 528–535.

9 M. Nakamura, I. Sugimoto, H. Kuwano,R. Lemos. Sensors and Actuators B, 1994,20, 231–237.

10 R.E. Kalman, On the general theory of controlsystems. in First IFAC Congress. 1960,Moscow, Butterworths, London.

11 S. Vaihinger, W. Gopel, J. R. Stetter.Sensors, and Actuators B, 1991, 4, 337–343.

12 E. Llobet, X. Vilanova, X. Correig.Proceedings SPIE, 1995, 2504, 559–566.

13 R. Gutierrez-Osuna, H. Troy Nagle,S. S. Schiffman. Sensors Actuators B,1999, 61, 170–182.

14 C. Lanczos, Applied Analysis, Prentice-Hall,1956.

15 D. G. Gardner, J. C. Gardner, G. L. Laush,W. W. Meinke, J. Chem Phys.. 1959, 31,978–86.

16 S. Marco, J. Samitier, J. R. Morante. Meas.Sci. Technol., 1995, 6, 135–142.

17 M. R. Osbourne, SIAM J. Numer. Anal.,1975, 12, 571–92.

18 E. Yeremian, P. Claverie, Nature. 1987, 326,169–174.

19 S. A. Ivanov, V. N. Ivanova, V. B. Smirnov,B. Z. Tabin. Opt. Spectros., 1992, 73,150–53.

20 M. R. Smith, S. Cohn-Sfetcu. Techno-metrics, 1976, 18, 467–482.

21 H. Tong, Non-linear Times Series. A Dyna-mical System Approach. 1990, Oxford:Clarendon Press. Chapter 3, 96–120.

22 M. B. Prietsley, State dependent models:a general approach to non-linear time seriesanalysis. J. Times Ser. Anal., 1980, 1, 57–71.

23 V. Volterra, Theory of Functionals andof Integro-differential equations. 1930,New York, Dover.

24 N. Wiener, Non-linear Problems in RandomTheory. 1958, New York: Wiley.

25 H. W. Chen, N. Ishii, N. Suzumura. Int.J. Systems Sci., 1986, 17, 371–377.

26 H.W. Chen, L. D. Jacobson, J. P. Gaska. Biol.Cybernet., 1990, 63, 341–357.

27 S. A. Billings, H. B. Jamaluddin, S. Chen.Int. J. Control, 1992, 55, 193–224.

28 J. G. Kuschewski, S. Hui, S. H. Zak. IEEETrans. on Control Systems Technology,1993, 1, 37–49.

29 C. L. Giles, G. M. Kuhn, R. J. Williams.IEEE Trans. on Neural Networks, 1994, 5,153–55.

30 S. Z. Qin, H. T. Su, T. J. Mcavoy. IEEE Trans.on Neural Networks, 1992, 1, 123–130.

31 L. Ljung, System Identification: Theory for theUser, 1987, Englewood Cliffs, New Jersey:Prentice Hall. Chapters 6 and 8 to 11.

32 P. J. Diggle, Times Series: A BiostatisticalIntroduction, 1990, Oxford: Clarendon Press.Chapters 2 and 6.

33 Y. W. Lee, M. Schetzen. Int. J. Control, 1965,2, 237–254.

34 H. A. Barker, R. W. Davy. Proceedings IEE,1975, 122, 305–311.

35 H. A. Barker, R. W. Davy. Int. J. Control,1978, 27, 277–291.

36 D. R. Brillinger, Time series: Data Analysisand Theory, 1981, San Francisco, CA,Holden Day. Chapter 6.

37 K. Godfrey, Perturbation signals for systemidentification, 1993, Prentice Hall, UK,Chapter 2.

38 H. A. Barker, Nonlinear system identificationusing pseudorandom signals with partiallyorthogonal transforms, in 7th IFAC Symposium


on Identification and System ParameterEstimation, 1985, York.

39 R. Haber, IEE Proc. D, 1988, 135, 405–420.40 S. A. Billings, Springer Lecture Notes in

Control and Information Sciences, 1986, 79,263–294.

41 S.A. Billings, K. M. Tsang. MechanicalSystems and signal Processing, 1989, 2,319–339.

42 M. J. Koremberg, Identifying noisy cascadesof linear and static non-linear systems, in 7th

IFAC Symposium on Identification and SystemParameter Estimation, 1985, York.

43 D. T. Westwick, R. E. Kearney. Biol.Cybernet., 1983, 68, 75–85.

44 N. J. Clayden, J. Chem. Soc. Faraday Trans.,1992, 88, 2481–2486.

45 S. Marco, A. Pardo, F. Davide, C. DiNatale,A. D’amico, J. Hierlemann, M. Schweizer,U. Weimar, W. Gopel. Sensors and Actua-tors B, 1996, 34, 213–223.

46 M. Nakamura, I. Sugimoto, H. Kuwano.Sensors and Actuators B, 1996, 33, 122–127.

47 T. Nakamoto, N. Okazaki, T. Morizumi.Sensors and Actuators B, 1997, 41, 183–188.

48 A. Pardo, S. Marco, J. Samitier. Dynamicmeasurements with chemical sensor arraysbased on inverse modeling. In IEEE Instru-mentation and Measurement TechnologyConference, 1996, Brussels, Belgium,904–907.

49 F. Davide, C. DiNatale, A. D’amico,A., A. Hierlemann, J. Mitrovics,M. Schweizer, U. Weimar, W. Gopel,S. Marco, A. Pardo. Sensors and Actuators B,1995, 26–27, 275–285.

50 F. Davide, C. DiNatale, A. D’amico,A. Hierlemann, J. Mitrovics, M. Schweizer,U. Weimar, W. Gopel. Sensors and Actua-tors B, 1995, 24–25, 830–842.

51 F. Davide, C. DiNatale, A. D’amico. Sensorsand Actuators B, 1994, 18–19, 244–258.

52 C. DiNatale, F. Davide, F., A. D’amico.Sensors and Actuators B, 1995, 26–27,237–241.

53 C. DiNatale, A. Macagnano, A. D’amico,F. Davide. Meas. Sci. Technol., 1997,1236–43.

54 B. W. Saunders, D. V. Thiel, A. Mackay-Sim.An artificial olfactory system using tieredartificial neural networks. In Australian andNew Zealand Conferrence on IntelligentInformation Systems, 1994.

55 S. Marco, A. Pardo, A. Ortega, J. Samitier.Gas identification with tin oxide sensor arrayand self organizing maps: adaptive correction ofsensor drifts, In IEEE Instrumentation andMeasurement Technology Conference, 1997,Otawa, Canada.

56 D. S. Vlachos, D. K. Fragoulis, J. N. Avarit-siotis. Sensors and Actuators B, 1997, 45,223–228.

57 E. Llobet, E. L. Hines, J. W. Gardner,P. N. Bartlett, T. T. Mottram. Sensors andActuators B, 1999, 61, 183–190.

58 E. L. Hines, E. Llobet, J. W. Gardner.Electronics Letters, 1999, 35, 821–823.

59 E. Llobet, E. L. Hines, J. W. Gardner,S. Franco. Meas. Sci. Technol., 1999, 10,538–548.

60 J. Brezmes, E. Llobet, X. Vilanova, G. Saiz,X. Correig. Sensors and Actuators B, 2000,in press.

61 M. Ippommatsu, H. Sasaki, J. Electrochem.Soc., 1989, 136, 2123–2128.

62 H. E. Endres, W. Gottler, H. D. Jander,S. Drost, G. Sberveglieri, G. Faglia,C. Perego. Sensors and Actuators B, 1995,24–25, 785–789.

63 X. Vilanova, E. Llobet, J. E. Sueiras,R. Alcubilla, X. Correig. Sensors andActuators B, 1996, 31, 175–180.

64 E. Llobet, X. Vilanova, J. Brezmes,R. Alcubilla, X. Correig. J. Electrochem. Soc,1998, 145, 1772–1779.

65 J. W. Gardner, P. N. Bartlett, E. L. Hines,F. Molinier, T. T. Mottram. IEE Proceedings:Sci., Meas. Technol., 1999, 146, 102–106.

66 H. Nanto, K. Kondo, M. Habara,Y. Douguchi, R. I. Waite, H. Nakazumi.Sensors and Actuators B, 1996, 35–36,183–186.

67 W. M. Sears, K. Colbow, R. Slamka,F. Consadori. Sensors and Actuators B,1990, 2, 283–289.

68 B. Yea, R. Konishi, A.K. Sugahar, T. Osaki.An advanced gas discrimination method utili-zing the periodic operation of a semiconductorgas sensor. In IEEE Conference on IndustrialAutomation and Control, 1995.

69 M. E .H. Amrani, R. M. Dowdeswell,P. A. Payne, K. C. Persaud. Sensors andActuators B, 1998, 47, 118–124.

70 P. Bruschi, A. Nannini, B. Neri. Sensorsand Actuators B. 1995, 24–25, 429–432.

12.5 Outlook 323323

71 G. A. Carpenter, S. Grossberg,J. H. Reynolds. Neural Networks, 1991, 4,565–588.

72 G. A. Carpenter, S. Grossberg,J. H. Reynolds. IEEE Trans. NeuralNetworks, 1995, 6, 1330–1336.

73 G. A. Carpenter, S. Grossberg. Fuzzy sets,neural networks, and soft computing,R.R. Yager, Zadeh, L.A., Editor. 1994,Van Nostrand Reinhold: New York,126–165.

74 H. W. Shin, E. Llobet, J. W. Gardner,E. L. Hines. C. S. Dow. IEE Proceedings,Sci. Meas. Technol., 2000, 147, 158–164.

75 P Boilot, E. L. Hines, S. John, J. Mitchell,F. Lopez, J. W. Gardner, E. Llobet, M. Hero,C. Fink, M. A. Gongora. Detection of bacteriacausing eye infections using a neural networkbased electronic nose system, Proceedings of7th ISOEN, 2000.

76 W.Gopel, J. Hesse, J. N. Zemel Eds.. Sensors:a comprehensive survey, Vol. 2, 1991, VCH,Weinheim. Chapter 6.

77 M. Sriyudthsak, L. Promsong, S. Panyakeow.Sensors and Actuators B, 1993, 13–14,139–142.

78 E.L. Hines, E. Llobet, J. W. Gardner. IEEProceedings, Circuits Devices and Systems,1999, 146, 297–310.

79 S. W. Moore, J. W. Gardner, E. L. Hines,W. Gopel, U. Weimar. Sensors andActuators B, 1993, 15–16, 344–348.


13

Drift Compensation, Standards, and Calibration Methods

M. Holmberg and T. Artursson

Abstract

In Webster’s Seventh New Collegiate Dictionary, drift is defined as “a gradual changein any quantitative characteristic that is supposed to remain constant”. Thus, a driftingchemical sensor does not give exactly the same response even if it is exposed to exactlythe same environment for a long time. Drift is a common problem for all chemicalsensors, and thus needs to be considered as soon asmeasurements aremade for a longperiod of time.First in this chapter, possible reasons for drift will be discussed. A distinction is

made between drift in the sensors, and drift in the measurement system. Afterthis, typical features of drift as seen in the measurements will be shown. These fea-tures include gradual increase or decrease, and jumps in the responses. At the end,many different methods for reducing the effects of drift will be described. These driftreduction methods try to compensate for the changes in sensor performance usingmathematical models and thus maintaining the gas identification capability of theelectronic nose. Many different methods have been applied for different situations.It is impossible to compare all the methods since each one of them makes some as-sumptions of how themeasurements aremade and/or how the drift is manifested. Notall examples discussed are for measurements with electronic noses, but the conceptsmay easily be transferred also to such applications. The purpose of describing all themethods is to show some possible ways of reasoning when dealing with a data-set fromdrifting sensors.

13.1

Physical Reasons for Drift and Sensor Poisoning

In this section, some of the common causes of drift in chemical sensors will be de-scribed. Other effects giving rise to similar phenomena will also be mentioned. Theaim of this section is not to give detailed information of the chemical processes thatoccur, but only to give a brief introduction to these effects.


325325

Ideally, a chemical sensor will always give the same response when exposed to anidentical gas mixture. This will, however, not be true when the sensors are operatedover a long time period. There will be changes in the size of the sensor response for acertain amount of a given gas; the selectivity of the sensor may change, i.e. the re-sponse changes differently for different gases; the speed of response may alsochange, see Figs. 13.1 and 13.2. These changes in the sensor behavior togethergive rise to drift in the sensor responses. Drift has plagued sensor researchers fora long time, but it is not until recent years that methods for reducing its effectshave been developed [1, 2].The response of chemical sensors depends upon chemical or physical interactions

between molecules in the gas phase and the sensor surface and/or bulk material. A lotof effort has been made to find sensor materials which interact reversibly with the gas,such that themolecules that have reacted on the sensor will leave it as soon as the gas is

Fig. 13.1 Idealized sensor responses for a chemical sensor. The curve

shows the three phases of a measurement: baseline measurement

(usually made with pure air), test gas measurement, and recovery time

(during which the sensor again is exposed to pure air, the recovery time

is usually much longer, but the last part of the curve has been omitted

here). Curve a) shows an example of a typical response curve for an

arbitrary gas; curve b) shows how the sensor response for the same gas

would be if drift has caused the speed of response to decrease; curve c)

shows the sensor response to the same gas but where the sensor

response has decreased compared to curve a). For a typical measu-

rement, the x-axis shows time in seconds, while the y-axis is in arbitrary

units, depending on the sensor type used

13 Drift Compensation, Standards, and Calibration Methods326

Fig. 13.2 Examples of the sensor response shown as bar charts for

one sensor when exposed to ten different gases. a) shows the re-

sponse as it appears without drift; b) shows the response to the same

gases when the response of the sensor has decreased by the same

amount for each gas. Note that the pattern is preserved, even though

the absolute values change; c) shows the response to the same gases

when the response has changed differently for the different gases.

This is referred to as a change in the selectivity of the sensor

13.1 Physical Reasons for Drift and Sensor Poisoning 327327

no longer present at the sensor surface. In laboratory conditions with well-controlledatmospheres, this may be achieved. However, for “real” environments where a lot ofdifferent gases are present (several of them in very small amounts), little can be saidbeforehand regarding the chemical reactions and their reversibility. Therefore, somereactions will be irreversible, thereby blocking or creating reaction sites on the sensorsurface and/or bulk of the sensing material. This will lead to a change in the sensitivityof the sensor towards other gases. Another effect that might occur is the re-organiza-tion of the sensing material, for example clustering of metal particles. This may hap-pen spontaneously with large time constants for all materials, but the effect may bespeeded up by operating in reactive environments and/or at high temperatures. Thisaging of the material also changes the number of reaction sites and thus also thesensitivity of the sensor. A time-dependent change in the response to an identicalchemical environment will therefore result, and this is how we see drift in our mea-surements. For different sensor types, different causes for drift will dominate [3–5].Several papers have been published on workmade to improve the long-term stability ofgas sensors [6, 7]. However, some regeneration of the sensor may be performed by, forexample, annealing of the sensor and thereby removing some of the irreversibly boundspecies.In the electronic nose concept it is also very important to consider drift in the mea-

surement system. This may be due to temperature variations in the measured headspace or on the sensors; reactions of gas species in the sampling system; variations inthe gas flow; humidity variations in the sample; ambient pressure variations, or otherphysical/chemical processes. It is very difficult, or maybe even impossible, to distin-guish between sensor drift and drift in the measurement system. It is, however, pos-sible to optimize the system components for each application in order to remove asmuch of the system drift as possible. This may be done by careful control of the sampleand sensor temperatures or by reducing the amount of tubing that the sample gasneeds to flow through. In the remainder of this chapter, the effects of sensor driftand system drift will not be separated, but will always together be termed drift.There are also other phenomena that may give similar effects to drift. One that is

worth mentioning is memory effects, i.e. that the response of the sensor depends onwhat it has recently been exposed to. The remnants of previous gases may be presenteither in the sampling system, or on the sensor surface itself. At the exposure of a testgas, these old remnants give an additional effect to the sensor response. This phenom-enon is different from drift since it is a temporary effect that may last only for minutesor hours. For longer time constants, this effect will not be distinguishable from drift.The best way to deal with this phenomenon is not to use drift compensation algo-rithms, but to improve the measurement procedure, e.g. by limiting the size ofthe sampling system, or to introduce “cleaning cycles”, i.e. short pulses of cleanair and/or high temperature annealing between the samples. Another effect that isoften seen is that the sensors need some time before they give a stable response afterstart-up of a measurement series. This means that the response increases or decreasesfor the first minutes or hours of operation. This is sometimes called short-term drift,but the nature is different from ordinary drift and it will not be dealt with in thischapter.


13.2

Examples of Sensor Drift

In Fig. 13.3, the sensor responses as a function of time are shown for an experimentmade using a gas mixing system and 39 sensors. The responses for the three sensorsshown are all for one well-controlled gas mixture (“odor”), but other gas mixtures werealso measured in between the measurements shown. In this experiment, all the sen-sors were freshly made. Since drift influences the sensors strongest in the beginningafter their fabrication due to thermal relaxation of the device, the sensors show ratherstrong drift over the measurement period, which was about two months.There are some features in the graph that often can be seen in long-time measure-

ment series:

* The most obvious feature is an exponential or linear decrease or increase in thesensor signal. This change comes either from changes in the sensitivity of the de-vice, or from changes in the baseline.

* There are some jumps in the data set, i.e. places where the sensor signal for noapparent reason suddenly changes value. In this case, the jumps are rather small(a few percent), but when the sensors are put in a more reactive atmosphere, thejumps may be much larger.

* There is also some noise superimposed on all the sensor signals.

In real-life applications, the situation may be even more complex since also variationsin the samples and/or the sampling system come into play. It is therefore importantfor all applications to carefully control the samples and the sampling system in ordernot to make the situation more complex than necessary.

Fig. 13.3 Sensor signals from

three different sensors for one

gas mixture as a function of time

in an experiment using a gas

mixture system

13.2 Examples of Sensor Drift 329329

The data as seen in Fig. 13.3 can be said to be univariate, which means that we studyone variable (sensor) at the time. In a multi-sensor system it is, however, often con-venient to study all sensors collectively usingmultivariate techniques such as PrincipalComponent Analysis (PCA), see Chapter 6. A PCA gives you a mapping of the data,from the original multivariate space with the number of dimensions equal to the num-ber of sensors, to a low-dimensional space which is much easier to visualize. Usually,the first few principal components are a good approximation of the data set for initialstudies. Figure 13.4 shows the same experiment as in Fig. 13.3, but for nine differentgas mixtures, and now using a PCA to visualize the data from all 39 sensors. As in-dicated by the arrows, the drift tends to move the sensor responses mainly in onedirection, and the direction is similar for all gas mixtures. The reason why the drifttends to go in only one direction for each cluster is that the sensors used are exposed tothe same (but not constant) environment all the time, so they tend to drift in a similarmanner. This means that the drift may be described in only a few (in this case one)dimensions even though the process is rather complicated. The reason why all theclusters drift in a similar direction is that the gas mixtures are very similar, sowhen the sensor changes, this change affects the responses for all gas mixtures ina similar way. In a situation where many different sensor types are used, one cannotassume that the drift will occur in a few dimensions only, but rather that one dimen-sion will be needed to describe the drift for each sensor type. It is also important to notethat different gas mixtures might drift in different directions, so when choosing areference gas for compensating drift, it has to be very similar to the test gases inthe application.

Fig. 13.4 Sensor responses

changing over time due to drift

as seen in a PCA plot. The data

set consists of data from 39

sensors, measuring on 9 diffe-

rent gas mixtures. The percen-

tage shown after the PC number

on the axes show how large part

of the total variance that is

explained by that PC. (Reprinted

from J. Chemometrics, 14, T.

Artursson, T. Eklov, I. Lundstrom,

P. Martensson, M. Sjostrom, and

M. Holmberg, Drift corrections

for gas sensors using multivariate

methods, 711–724, 2000, with

permission from JohnWiley & Sons

Limited.)


13.3

Comparison of Drift and Noise

In a real measurement series, it may sometimes be necessary to attribute a smallchange in the sensor response to either a change in the sample; noise in the measure-ments; or a drift induced change in the sensor response. If only one such change isoccurring, this distinction is impossible to make. On the other hand, by analyzing along time-series of data, much can be learnt about the intrinsic noise in the system.This information may then be used in statistical models to ascertain if the change isdue to noise or to changes in the sample. However, sensor drift may change the sta-tistical limits, thereby making the models useless. Some information regarding thefrequency spectra of the noise and the drift may be obtained from such studies,but it may be difficult to use this information in practice, since the frequency spec-trum of the sensors also may change over time due to drift.Very little can be found in the literature regarding the relationship between drift and

noise. There has been one study [8] where a frequency analysis of a long time serieswas made. In the study, it was assumed that similar sensors drift in a similar manner.The signals from the sensors were passed through band-pass filters, and the correla-tion between the filtered signals was studied by seeing how well a model could predict

Fig. 13.5 (a) A model f1 is built on instrument 1 using samples re-

presentative of all possible measurements in the future. (b) In order to

be able to use the information in instrument 1 without remaking all the

measurements used to build f1, some known samples are measured on

both instrument 1 and i. Then, one searches for a transformation of

either the model f1 (initially equal to f1), Xi, or Yi;1 that renders

Yi;ref ¼ Y1;ref . (c) Depending on which transformation was chosen, the

data evaluation for the new instrument n is made according to one of the

three schemes shown. The first case with a transformation of f is not very

simple, and therefore not so common. The second case with a trans-

formation of X is referred to as direct transformation. The third case with

a transformation of Y is called bias correction

13.3 Comparison of Drift and Noise 331331

the output of a sensor using other sensor signals as inputs. A low prediction errormeant that there was a high correlation and vice versa. For the frequency ranges wherethere was a correlation between the different sensors in the array, drift was said todominate, while noise dominated in the frequency range where the variationswere not correlated between the sensors. The sensor correlation was thus used asan instrument to distinguish the different frequency ranges of drift and noise.

13.4

Model Building Strategies

In general, some model, f, is used to map the measured sensor data, often termed X-data, to some output, Y, which gives us the information we desire, e.g. the class and/orquality of the sample, so Y ¼ f ðXÞ, see Fig. 13.5a. The model could be anything from asimple linear regression to more complex model types such as Artificial Neural Net-works, as described in previous chapters. When we study drift, we need to use furtherconsiderations in the model building and the model validation. The first thing to con-sider is the choice of sensors. Do some of the sensors vary more than the others, ormaybe even stop to respond after some time? In that case, it might be wise not toinclude those sensors in the model building. Also, when a drift reduction methodis tested, it is not a good idea to use parts of the data set from the whole time periodin the data set for themodel building. This could lead to the variations in the data beingbuilt into the model rather than actually being reduced by the drift reduction method.Instead, it is wise to build a model of the first measurements, and then apply the driftreduction method to subsequent measurements, thus validating both the model andthe drift reduction method.

13.5

Calibration Transfer

The transfer of calibration methods from one instrument to another almost (but notexactly) identical instrument is a problem for many different instrument manufac-turers. [9, 10] A model, f1, which maps X to Y, is built using measurements on instru-ment 1, see Fig. 13.5a. The measurements, X, used to build the model must representall possible situations that may arise during later operation of the instrument in orderto get a representative model. Many measurements are therefore necessary to buildthis model, the exact number depends on the instrument and model complexity butranges from a few tens to several thousand. For all other presumably identical instru-ments manufactured, we want to avoid making all the measurements again sincemeasurements in general are time-consuming and expensive. So, the aim is to trans-fer the information contained in f1 (the model built using measurements made withinstrument 1) to a model for instrument i (fi ) using as few measurements as possible.If the instruments were identical, there would be no need to make any new measure-


ments, since the models f1 and fi would be equal. If the instruments were completelydifferent, there exists no common information for the two instruments, and thus allpossible situations have to be measured also on instrument i. If we assume that theinstruments are similar, but not equal, we can make a few new measurements on athird instrument n, and then assume that other measurements in similar environ-ments have changed in a similar manner for all the instruments. We can then reducethe number of measurements necessary to build the newmodel, fi . If the sensors in anelectronic nose have drifted slightly, this can be seen as having one instrument at timet1, and another slightly different instrument at a later time t. The concepts for calibra-tion transfer and for drift reduction are therefore similar, even though the problem isdifferent.Mathematically speaking, the aim is to approximate a function, fi capable of map-

ping Xi to Yi, by using a low number of measurements, Xi, and another function, f1, asa first approximation:

fi ¼ Tðf1jYi ¼ fiðXiÞÞ ð13:1Þ

where T is a transformation operator, different for different calibration transfermethods, see Fig. 13.5b. For a complete description of possible transformations,see the references mentioned above.The model f is often changed by either pre-processing of the X-data, or post-proces-

sing of the Y-data, see Fig. 13.5c. In the second case in the figure, often called directtransformation, a relationship between the X-values for instrument 1 and i is calcu-lated using some known samples. It is important that these samples are chosen so thatthey span as large part of the response space as possible in order to find a represen-tative transformation for all possible X-values. The relationship is then used to trans-form the X-values obtained for instrument i to the same situation as for instrument 1.The originally built models on instrument 1 can then be used to predict Y also forinstrument i.The third case in the figure, where the Y-values are corrected, is termed bias correc-

tion since it is assumed that the error between the measurements can be seen as a biasin the predicted Y-values. For the ith instrument, the original model for predicting Yfrom X is used, and a model to correct the predicted Y is built using some references.The Y-correction is then used for all subsequent measurements, making it possible touse the models for instrument 1 also for instrument i.

13.6

Drift Compensation

In order to get an estimation of the size of the drift, measurements are often made onone or several so-called reference gas (or gases). Measurements are made on the re-ference gas(es) in the beginning of the measurement series (time t1), and then withsome intervals (usually a few times per week) as long as the sensors are used. Thechange in the sensor responses to the reference gas is taken as a measure of the re-


sponse change for all other measurements, with different assumptions for differentmethods as described below. In order to get a good estimate of the drift for the realsamples, the reference gas has to be well-chosen, meaning that the drift in the refer-ence gas should reflect the drift for all other samples. Different researchers have cho-sen different approaches to find good reference gases; some use the same referencegas (often water) for all applications, while others choose the reference gas dependingon the application (e.g. the head-space of a given concentration of ethanol in water fordetermination of the intoxication level with breath analysis). Usually, it is a good idea tochoose a reference gas that is close to the real samples in sensor response space (as canbe seen in a PCA score plot). A good reference gas also has to be stable over time (notdegrade) and be easy to measure so that the variation in gas concentration over timebecomes minimal.It can sometimes be helpful to categorize the different methods used for drift com-

pensation. One such distinction is if the sensors are considered one at the time, or as agroup. The first case means that the sensors are considered to operate independentlyof each other, which is called a univariate approach. In this case, one drift correctionmodel is made for each of the sensors. In the second case, one drift correctionmodel ismade for a group (often all) of sensors. This is called a multivariate approach, seeFig. 13.6.Another way of distinguishing between different methods is to see where the adap-

tation due to drift is made. Basically, there are three strategies for compensating driftin the sensors: direct transformation (adaptation of X), bias correction (adaptation ofY), or the use of self-adapting models (adaptation of f). After the strategy has beenchosen, one has to decide what model type to use for calculating the compensa-tion, and what assumptions and information to use to build the compensation mod-els. Many different solutions have been tested, and in the following sections we will

Fig. 13.6 A diagram showing the difference between univariate, (a), and multivariate, (b), drift correction


give an overview of some attempts that have been made. It is important to rememberthat a method that works well in one situation does not necessarily work in all otherapplications, so it is important to study the data to find what restrictions and possi-bilities that you have in your own data set before trying out a new method.As discussed previously, drift can be manifested in several ways. If drift causes the

baseline of the sensor to change, the response will be increased or decreased by thesame amount, so the drift will be additive. By measurement on a reference gas, theamount of change can be calculated and used for all measurements on the samples. Ifthe drift causes the sensitivity of the sensor to change instead, the drift is termed asmultiplicative, that is the response is increased or decreased by some factor. A refer-ence gas can then be used, as for the additive case, to calculate the correction factor.These two corrections will be exact if the sensors are linear, but they will also work wellas a first approximation for non-linear sensors.After applying a drift reduction method, it is a good idea to also check its effective-

ness by comparing the prediction capabilities of the classification/quantification mod-el with and without drift reduction. If the change in prediction error is not statisticallysignificant, then the method should not be used since the increase in model complex-ity introduced by adding an extra algorithm might compromise the overall perfor-mance for future measurements. When the sensor array is used for quantificationof one or several gases, the relative change in the RMSEP value can be used as a per-formance measure of the drift reduction method:

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1mPm

i¼1ðytruei � ypredi Þ2

s !with drift reduction

�ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1mPm


s !without drift reduction��

��ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1mPm


s !without drift reduction

ð13:2Þ

where m is the number of measurements, and ytruei and ypredi are the true and predictedquantification value, respectively. For the case where only classification is desired, therelative change of the Mahalanobis distance between the different clusters before andafter drift reduction is a good measure of the effectiveness of the method [11].It is also possible to obtain a measure of the performance of the drift reduction

method simply by comparing the classification rate before and after drift reduc-tion. The comparison can in this case be made with a simple k-nearest-neighbor clas-sifier or other standard classification methods if desired.

13.6.1

Reference Gas Methods

It can sometimes be a problem to distinguish between sensor drift and changes in thesample. To separate drift from sample changes, a stable sample called a reference gasis often measured. Different approaches to correct gas sensor data using a reference


gas as a reference value and then correcting all subsequent readings accordingly havebeen made. Five different examples will be given where a reference gas has been usedto reduce the drift. All five examples make the assumption that there is a strong cor-relation between the drift in the response of the sensors to the reference gas and to thesamples. After this, a sixth method is presented, where replicates of the samples areused as pseudo-reference gases. The first four methods and the sixth use direct trans-formation of the data, while the fifth method uses the bias correction procedure. Re-garding the data treatment, the first two methods and the last work in the univariatemode, while the third and fourth work in themultivariate mode. The fifthmethod onlycorrects one Y-variable, so here it is not meaningful to use these terms. The univariatemethods assume that the changes in the relationship between the response to thereference gas and the response to the test gas can be compensated for one sensorat the time.The first method used by Fryder et al. [12] assumed that the drift was additive, i.e

independent of the signal level. They reduced the drift in the measurements madewith an electronic nose by subtracting the response to the reference gas from thesample responses, see Eq. (3).

x 0t;i ¼ xsample;t;i � xreference;t;i ð13:3Þ

where x 0t;i, xsample;t;i, and xreference;t;i are the drift corrected sensor response, the uncor-

rected sensor response, and the response for the measured reference gas, respectively,all measured on sensor i at time t. The additive drift was removed and all the measure-ments were studied relative to the reference gas.The second example is closely related to the first one, but instead of reducing only

the additive drift, themethod corrects for multiplicative drift effects for measurementsmade within the same day see Haugen et al [13]. This method was successfully used intheir experiments to reduce drift from fresh fish measurements with an electronicnose, measured over five days. The ratio between the responses at time t and atthe initial time for the reference gas was calculated for each sensor (see Eq. (4)),and this ratio was used to compensate the responses for the samples. In this specificwork, a linear trend line was also fitted to this ratio to find the correction factor for thesample measurements made between the reference gas measurements, see Eq. (5)

quote ¼ xreference;t1;i=xreference;t;i ð13:4Þ

x 0t;i ¼ xsample;t;i

� ft;i ¼ xsample;t;i � ða � tþ bÞi ð13:5Þ

where ft;i is the trend line, a is the slope of the trend line, b is the intercept, t1 is theinitial time, and i is the sensor number. An additive correction was used for correctionbetween the different days.The third method is a drift reduction method based on PCA and PLS, called com-

ponent correction, CC [14]. The method assumes that the drift has a preferred direc-tion in the measurement space and removes this direction from the measurements.The direction of the drift, p, is calculated frommeasurements of a reference gas. If thesensor responses to the reference gas have significant drift, the first components, p, in


a PCA analysis of this gas will describe the direction of the drift. The vector p com-prises the direction coefficients of the one dimensional principal component space,and can be used also to see which parameters contain the most drift. Projecting thesample gas measurements on this vector gives a score vector, t, which contains theamount of drift for each sample, see Fig. 13.7. The drift component, tpT, can thenbe removed from the sample gas data. The direction in the data set that is removedis a linear approximation of the drift direction. By removing this direction all the otherdirections are preserved and the important variances that separate different clustersand concentrations are maintained in the data set, unless the information is found inthe same direction as the drift. This method was applied to data sets from measure-ments using a electronic nose and a gas-mixing system with mixtures of four differentgases (hydrogen, ammonia, ethanol, and ethene) both for classification and quantifi-cation over a period of more than twomonths. The results for the data seen in Fig. 13.4are shown in Fig. 13.8. A similar method, but based on canonical correlation analysishas been proposed by Gutierrez-Osuna, who used metal-oxide sensors for measure-ments on spices over a period of three months. [15]The fourth method uses several reference gases in a transformation model, linear or

non-linear. A prediction model to predict Y from X is built at time t1, i.e. in the begin-ning of themeasurement series. In latermeasurements a drift reductionmodel is builtwith the reference gas measurements at time t, t > t1, as inputs and the reference gasmeasurements at time 1 as outputs. This model should then be able to transform X-data at time t to the value they should have had at time t1, i.e. when no drift had afflictedthe sensors. All the other data is then transformed using the drift reduction model.After this pre-processing the original identification model is then used to predict Y.

Fig. 13.7 A vector diagram showing the projec-

tion of the sample i down to the drift vector p1.

The projected value ti is the amount of drift for

sample i


Goodacre et al. [16] used artificial neural networks (ANN) both for the prediction mod-el and the drift reduction model for data from pyrolysis mass spectrometry used topredict bacteria concentration with good results.The fifth method uses the bias correction procedure, i.e. the original model to pre-

dict Y is used for all data, but the predicted value is then corrected by a factor, which iscalculated frommeasurements on reference gases, see Fig. 13.5c. In this case, the aimwas to measure the alcohol content in breath samples from intoxicated persons usingan electronic nose. [17] An ANNmodel was built to predict the alcohol content, using agas chromatograph as a reference instrument. A reference sample with 109 mol-ppmEtOH in technical air was also included in the measurements. The measurementswith the highest and lowest EtOH concentrations in the test set were also used asreference samples. A linear regression model between predicted and measuredEtOH concentration in the reference samples were calculated. From this linear regres-sion model correction factors such as slope and intercept were calculated and used forcorrection of the Y data in the test set.It may sometimes be difficult to find good reference gases for the measurements. If

that is the case, it is possible to use replicates of the samples as pseudo-reference gases.This can be done as long as the samples are stable over time or reliable standardizedsamples are available. This has been done by Salit et al. [18], who used both an additiveand a multiplicative drift correction algorithm with replicates of the samples as refer-ence values for measurements made with inductively coupled plasma-optical emissionspectroscopy. The signals, which suffered from additive and/or multiplicative drift,

Fig. 13.8 The data from Fig. 13.2 with drift reduced by the Component

Correction (CC-) method. (Reprinted from J. Chemometrics, 14,

T. Artursson, T. Eklov, I. Lundstrom, P. Martensson, M. Sjostrom, and

M. Holmberg, Drift corrections for gas sensors using multivariate methods,

711–724, 2000, with permission from John Wiley & Sons Limited.)


were defined as the sum of the true value (xtruth), the drift influence (edriftðtÞ), and thenoise (enoise), see Eqs. (6) and (7).

xmeasured ¼ xtruth þ edrift þ enoise ð13:6Þ

xmeasured ¼ xtruth � ð1þ edrift þ enoiseÞ ð13:7Þ

The aim was then to find the drift influence and remove it from the data. Instead ofspending time frequently measuring standards they measured replicates of the samp-les, which is common when precise analytical results are wanted. For each sample themean of the individual signals, xmean, is used as an estimate of xtruth. The drift andnoise contribution was calculated as the difference between the measured sample andthe estimate of the true value, xtruth. If the drift was additive a smooth curve was fittedto the deviation values. This curve was assumed to describe the drift, edriftðtÞ, and theresiduals to the curve were defined as noise, enoise. For multiplicative drift, the devia-tions emeasured=xmean were fitted to a smooth curve. By these definitions it was thenpossible to reduce the drift. For additive drift, these corrections are directly predictedfrom the function edriftðtÞ, and for multiplicative drift the correction isxtruth � ð1� edriftðtÞÞ. The use of replicates from all the samples, instead of replicatesfrom one standard, reduced the uncertainty in the drift corrections.

13.6.2

Modeling of Sensor Behavior

The most exact drift counteraction model would probably be a physical one, where allphysical changes of the sensor are modeled and accounted for. However, this type ofmodel is very hard to make general for gas sensors. In a well-controlled system withvery few gases it would be possible to know what reactions might occur, and thus todescribe the drift with a physical model. The problem comes in a real application whenthe system is not so well controlled, and there are a lot of different gases and combina-tions of these.For the pH ISFET sensor measuring in liquid, a physical model for different pH can

be made [19]. The origin of drift for these sensors is a chemical modification of theinsulator surface, which is covered by a hydrated layer. The variation in thickness ofthis hydrated layer changes the capacitance of the insulator and thereby causes drift. Byconsidering the correlation between the layer and its limiting factor for transport ofwater related species to the insulator, a model is built describing the drift. The modeldescribes the drift behavior for Si3N4-gate and Al2O3-gate pH ISFETs measured in0.1 M KCl solution.Another way of modeling the sensor behavior is not to consider the reactions that

occur, but to study how the sensors behave in their operating conditions, and thenassume that the sensors will always behave in the same manner when they are ex-posed to the same environment. This would be a mathematical model rather thana physical, but could still be useful in situations where the environment causes driftin the sensors, but does not change much over time. It can then be assumed that the


sensors follow a certain mathematical curve over time. This requires well-controlledmeasurements as in the case for modeling of sensor behavior. Pearce et al. [20] used alinear fit to compensate for base-line drift in measurements with an electronic nose onbeer over a period of 12 days. The base-line value was measured for each sensor, and alinear fit was used to determine the base-line drift and compensating for additive drift.

13.6.3

Pattern-Oriented Techniques for Classification

When a measurement is made, the responses of all sensors are measured. These re-sponses can be said to form a pattern, imagine for example plotting the responses in ahistogram that gives a pattern of bars, one for each sensor. We may then assume thateach class in a classification problem has a typical pattern, preserved over time. Therelative relationship between different sensors rather than their absolute outputs con-serves the pattern, see Fig. 13.9. If the relative relationships stay constant over time, asimple normalization (e.g. by setting one sensor to always have the value one andscaling the others accordingly) would do the trick. In reality things are not that sim-ple. Noise and different amounts of drift for different sensors make it necessary to useother tricks to see if the pattern is conserved.By studying the steady state or the transient behavior of sensors in an electronic

nose, and threshold the values into a binary output, Wilson et al. [21] managed todiscriminate between different chemicals. The output voltage from an array of ten

Fig. 13.9 A diagram showing a constant relative relationship

between different sensors, for time t ¼ 1 and t ¼ 2


tin-oxide sensors were arranged in ascending order and the output from each sensorwas set to zero if it was smaller than the median output, and set to one if it was largerthan the median. The resulting output from the threshold function was a pattern of

Fig. 13.10 Binary response pattern for (a) ace-

tone, (b) ethanol, (c) hexane, (d) isopropyl alcohol,

(e) methanol and (f) carbon monoxide. (Reprinted

from Sens. Actuators B, 28, D. M. Wilson, S.P. De-

Weerth, Odor discrimation using steady-state and

transient characteristics of tin-oxide sensors, 123–

128, 1995, with permission from Elsevier Science.)

Fig. 13.11 The VLSI circuitry for the winner-take-all signal processing.

The output voltage is fed into the winner-take-all, WTA, and loser-take-

all, LTA cells. Here, the WTA output is compared with its neighbor’s

outputs giving slope left and slope right as outputs. (Reprinted from

Sens. Actuators B, 26–27, D. Bednarczyk, S.P. DeWeerth, Smart chemical

sensing arrays using tin oxide sensors and analog winner-take-all signal

processing, 271–274, 1995, with permission from Elsevier Science.)


zeros and ones, see Fig. 13.10. This way of thresholding the signal removes much ofthe information, but the information which is left was sufficient to discriminate be-tween acetone, ethanol, hexane, isopropanol, methanol and carbon monoxide. Theresulting pattern is more robust than the use of absolute sensor values since it isrelative, and it is not very sensitive to noise. The drawback with the method is thatit also adjusts for changing concentration levels and is therefore useful only for clas-sification purposes.Bednarczyk et al. [22] worked with a sensor array of ten tin oxide sensors. From this

array they located the sensors with the highest (winner), and smallest (loser) outputsvoltage for each sample, and also the slope between the winner/loser and its two near-est neighbors was calculated, see Fig. 13.11. This gives a total of six outputs from eachsample giving a specific pattern that changed little over time. The winner and loserwere used for a rough classification, and the slopes were used for finer classification.For example, ethanol is first classified as an alcohol from its winning and losing sen-sor, and after that as being ethanol from the values of the slopes. All calculation wasprocessed in VLSI circuitry. The fact that the pattern of the responses rather than theabsolute values was used allowed a robust chemical discrimination to be made.Another approach for using the pattern of responses rather than individual sensor

responses was made by Holmberg et al. [23] In this case, four different alcohols andwater were measured over a period of two months with large drift in the responses. Itwas assumed that the pattern was preserved over time for each class (i.e. the differentalcohols or water), but only for sensors that were similar enough. For that reason, asmall subset of three sensors was chosen for the model building. Then, for each class amodel was built to predict the output of one sensor, using two other sensors as inputsto the model. These models were different for the different alcohols. When a new

Fig. 13.12 A block diagram describing the routines of training and

prediction of sensor response, where time-invariant relationships be-

tween the sensor responses are used to reduce the influence of drift


measurement was made, the sensor responses were put into all the different models,and the new sample was identified as belonging to the class whose model gave thelowest prediction error. The approach was also improved by allowing updating ofthemodels to adjust for possible changes in the relationships, see Fig. 13.12 [24]. How-ever, also in this case the models are insensitive to variations in concentration, and canthus only be used for classification.

13.6.4

Drift-Free Parameters

Another drift counteraction approach is to find parameters in the measurements thatremain constant even though the responses changes. Roth and co-workers [25] usedthis approach to measure CO2 with gas sensors with organic coatings. They used anappropriate temperature profile in order to decrease the time the sensors were heated.This improved the overall lifetime of the sensor coatings. To further reduce the in-fluence of drift they used the normalized response slope, instead of using the driftsensitive absolute values. The slope of the sensor signals was normalized with theoverall amplitude of the signal, in this way the drift sensitivity of the parametersused was reduced, see Eq. (8).

slopenorm ¼ slope

amplitude¼ slope

max �minð19:8Þ

where slopenorm is the normalized slope and max and min are the maximum and mini-mum response values. Effects like aging and poisoning, which alter the baseline, didnot affect the calculated parameters.

13.6.5

Self-Adapting Models

Models that are adjusted on-line are usually called adaptive. This kind of model isuseful if the process studied has large variations. Adaptive modeling can be usedfor both linear and non-linear models. Davide and co-workers [26] introduced an adap-tive Self-Organizing Map (SOM) to reduce the influence of sensor drift. The basic ideawas to follow the odor pattern that suffered from drift. For a SOM, different neuronsare assigned to different classes in the model building process, see Fig. 13.13. Whenthe SOM is used it learns in real-time by continuously moving the nearest neuronstowards the input data, by adjustments of the weights. As the sensor responses change,so do the neurons and classification can thus still be made. In this way discriminationbetween different odors was possible. However, if one of the patterns is not measuredfor a long time, its neurons will be influenced by measurements of other classes andmoved in an undesired way. One approach to avoid this was proposed by Distante et al.[27], who let each class be described by one SOM, thereby avoiding the confusion thatmight arise if the classes are encountered with different frequency.


Another adaptive network, which can be used to reduce the effects of drift is Adap-tive Resonance Theory, (ART) see Chapter 16 for details. This kind of network hasbeen used for classifications of odors subjected to drift in the chemical sensors[28]. ART networks have the ability to learn a new pattern in real-time and updatethe prototype vectors describing the different classes. The algorithm finds the proto-type vector closest to the sample, and if the degree of match is higher than a thresholdvalue the weights of the prototype vector are refined. If the degree of match is lowerthan the threshold value a new pattern will be created. A key to reliable results is to finda good threshold value, so the number of learned pattern becomes the right one. Vla-chos et al. [29] compared ARTwith a back-propagation neural network and showed thatthe probability to get a successful answer increased with the ART.Another way to use ANNs for drift reduction has been published by Smits et al. [30],

where they have used signals (not necessarily from electronic noses) that change overtime. They simulated drifting data and then compared the classification performancefor an ANN with uncorrected data; with data corrected for additive drift; and withuncorrected data, but with an extra input to the ANN describing the amount ofdrift. Their results indicate that the last strategy gives the best result.

13.7

Conclusions

Drift is a common problem for electronic noses due to the varying and often reactiveenvironment they are used in. The reasons for the drift vary, and stem from both thesensors and from the measurement system. Usually, the drift has a rather low fre-quency (the variations occur on the order of days), but it may be different whenthe sensors are fresh, or the environment contains aggressive gases.In order to reduce drift in the best possible way, measurements have to bemade over

a long time period so that the drift effects can be studied. It is important to establishthat there is a drift, because if there is no drift, drift counteraction methods should ofcourse not be used. If drift exists, the next step is to find trends and/or correlations inthe data set that can be used as a drift reductionmethod. No drift reductionmethod hasbeen found to be superior to the others in all different types of situations, so it may benecessary to use different algorithms for different applications. It is also important toremember that different methods put different requirements on the data set, such as acalibration gas is necessary; quantification of different gases is required; or the envir-

Fig. 13.13 Interpretation of the SOM, where each square symbolizes

a neuron. Three distinct classes are visible: a, b and c


onment stays almost constant so the drift may bemodeled. If the application allows theuse of a reference gas it should be used, since it gives the user reliable information ofthe amount of drift in the sensor system. Furthermore, the use of a reference gas givesuser the possibility to discriminate between sensor drift and changes in the sampleover the time. The pattern-oriented techniques are attractive since they give stableresults, but the drawback is that they are only useful for classification purposes.Both pattern-oriented techniques and the self-adapting approaches may fail whenboth the sample and the sensor system change over time, since they do not discrimi-nate between drift and sample changes over time. In order to model the sensor be-havior the systems need to be very well controlled, and these models are therefore hardto use in real applications. In any case, it is important to understand the method that isused. A good drift reduction method that is used in the wrong way may give confusingresults, and therefore be more harmful than helpful.

Acknowledgements

The authors would like to thank all colleagues that have contributed with valuablediscussions and comments during this work, but we owe a special gratitude to TomasEklov, David Lindgren, Fabrizio Davide, and Ingemar Lundstrom for their support andfeedback.

References

1 W. Gopel, K.-D. Schierbaum. In Chemicaland biochemical sensors, part I, Vol. 2(Ed. W. Gopel, T.A. Jones, M. Kleitz, I.Lundstrom and T. Seiyama), VCH Verlags-gesellschaft, Weinheim, Germany, 1992,pp. 1–28.

2 J. W. Gardner, P. N. Bartlett. Electronic Noses– Principles and Applications, Oxford SciencePublications, 1999, 126–128 and 178–179.

3 I. Lundstrom, A. van den Berg, B. H. van derSchoot, H. H. van den Vlekkert,M. Armgarth, C. I. Nylander. In Chemicaland biochemical sensors, part I, Vol. 2(Ed. W. Gopel, T.A. Jones, M. Kleitz, I.Lundstrom and T. Seiyama), VCH Verlags-gesellschaft, Weinheim, Germany, 1992,pp. 493–494 and 516–519 and referencestherein.

4 C. Caliendo, E. Verona, A D’Amico. InGas Sensors (Ed. G. Sberveglieri), KluwerAcademic Publishers, The Netherlands,1992, p. 281–306.

5 D. Kohl. In Gas Sensors (Ed. G. Sberveglieri),Kluwer Academic Publishers, The Nether-lands, 1992, p. 43–88

6 U. Schoneberg, H. G. Dura, B. J. Hosticka,W. Mokwa. 1991 International Conferenceon Solid-State Sensors and Actuators,San Francisco, USA, 1991.

7 K. Dobos, R. Strotman, G. Zimmer. Sensorsand Actuators, 1983, 4, 593–598.

8 F. A. M. Davide, C. Di Natale, M. Holmberg,F. Winquist. In Proceedings of 1st Italianconference on sensors and microsystems(Ed. C. Di Natale and A. D’Amico), WorldScientific, Singapore, 1996, pp. 150–154.

9 O. E. de Noord. Chemometrics and IntelligentLaboratory Systems, 1994, 25, 85–97.

10 Y. Wang, D. J. Veltkamp, B. R. Kowalski.Analytical Chemistry, 1991, 63, 2750–2756.

11 P. Spangeus, D. Lindgren. Submitted toIEEE Sensors Journal.

12 M. Fryder, M. Holmberg, F. Winquist I.Lundstrom. In Proceedings of Transducers ’95and Eurosensors IX, Stockholm, Sweden,1995, 683–686.

13 J.-E. Haugen, O. Tomic, K. Kvaal. AnalyticaChimica Acta, 2000, 407, 23–39.

14 T. Artursson, T. Eklov, I. Lundstrom,P. Martensson, M. Sjostrom, M. Holmberg.Journal of Chemometrics, 2000, 14, 711–723.


15 R. Gutierrez-Osuna. In ISOEN 2000 ab-stracts (Ed. J. W. Gardner and K.C. Persaud),European Chemoreception Research Orga-nisation, Brighton, UK, 2000, 137–138.

16 R. Goodacre, D. Kell. Analytical Chemistry,1996, 68, 271–280.

17 N. Paulsson, F. Winquist. Submitted toMeasurement Science and Technology.

18 M. L. Salit, G. C. Turk. Analytical Chemistry,1998, 70, 3184–3190.

19 S. Jamasb, S. Collins, R. L. Smith. Sensorsand Actuators B, 1998, 49, 146–155.

20 T. Pearce, J. W. Gardner. Analyst, 1998, 123,2057–2066.

21 D. M. Wilson, S. P. DeWeerth. Sensorsand Actuators B, 1995, 28, 123–128.

22 D. Bednarczyk, S. P. DeWeerth. Sensorsand Actuators B, 1995, 26–27, 271–274.

23 M. Holmberg, F. Winquist, I. Lundstrom,F. A. M. Davide, C. Di Natale, A. D’Amico.Sensors and Actuators B, 1996, 35–36,528–535.

24 M. Holmberg, F. A. M. Davide, C. Di Natale,A. D’Amico, F. Winquist, I. Lundstrom.Sensors and Actuators B, 1997, 42, 185–194.

25 M. Roth, R. Hartinger, R. Faul, H.-E. Endres.Sensors and Actuators B, 1996, 35–36,358–362.

26 F. A. M. Davide, C. Di Natale, A. D’Amico.Sensors and Actuators B, 1994, 18–19,244–258.

27 C. Distante, T. Artursson, P. Siciliano,M. Holmberg, I. Lundstrom. In Olfactionand Electronic Noses 2, 2000 (Ed. J. W.Gardner and K. C. Persaud), The Instituteof Physics, 2000.

28 J. W. Gardner, E. L. Hines, C. Pang.Measurement þ Control, 1996, 29, 172–178.

29 D. S. Vlachos, D. K. Fragoulis,J. N. Avaritsiotis. Sensors and Actuators B,1997, 45, 223–228.

30 J. R. M. Smits, W. J. Melssen,M. W. J. Derksen, G. Kateman. AnalyticaChimica Acta, 1993, 284, 91–105.


14

Chemical Sensor Array Optimization: Geometric and Information

Theoretic Approaches

Tim C. Pearce, Manuel A. Sanchez-Montanes

Electronic nose technology – which exploits arrays of broadly-tuned chemical sensors– has matured to the point where it is routinely applied to the quality control of a widerange of commercial products, such as foods, beverages, and cosmetics. Even though alarge number of companies exist that design, implement, and sell this technology, theissue of how a practical system is configured and optimized to a particular applicationdomain is, at best, carried out using heuristic methods, or more often, completelyignored. The key theme of this chapter is how the selection of different chemical sen-sors is crucial to the overall system performance of these analytical instruments. Bytaking a geometric approach combined with simple linear algebra analysis, we demon-strate how the ‘tunings’ of individual sensors affect the overall performance. Newperformance measures based on information theory are defined here that shouldbe adopted for optimizing the performance of electronic nose systems.

14.1

The Need for Array Performance Definition and Optimization

Electronic nose instruments are used today for a very wide range of detection tasksfrom quality control of various food products to medical diagnosis. Clearly, each de-tection task requires sensitivity in the instrument to a number of different chemicalcompounds, which are likely to be very different from application to application. Over10 000 odorous compounds are known to exist in nature, but only a handful of theseare likely to be important in solving any discrimination task. The concept of a universalelectronic nose instrument, able to solve all odor detection problems, is unlikely tobecome a commercial reality, particularly because creating sensor diversity withinan instrument is expensive and most instruments are dedicated to a very restrictedrange of detection tasks. In practice, the entire instrument, from sample deliveryto sensor array, signal processing and classifier stages, is usually optimized to a par-ticular problem domain in order to provide suitable sensing performance. The opti-mization of signal processing, classifier, and sample preparation are dealt with else-where in this book.


347347

In this chapter we consider exclusively the problem of tailoring a chemosensor arrayto a particular detection task. One approach might be to augment an existing array byadding sensors appropriate to the new task, but this is an expensive and wasteful solu-tion. Most systems have a limited number of channels and, as we shall see, moresensors does not guarantee improved performance due to noise considerations –for example, when combined with a practical classifier adding a sensor with closeto zero sensitivity to the compounds of interest but with significant noise will poten-tially degrade the performance of the array as a whole. In practical terms, optimizationof chemosensors within an electronic nose instrument usually means selecting be-tween a potentially large pool of different sensors (even comprising completely differ-ent sensing technologies). The optimization task is to select a combination of sensorsbest suited to the detection task, and ideally to be able to specify a detection limit foreach compound of interest.Electronic nose instruments rely on a range of broadly tuned chemosensors in order

to discriminate complex multicomponent odor stimuli. It is the pattern of responseacross the array that is used in discriminating between complex (multicomponent)odor stimuli. This sensing arrangement makes the question of detection performancedefinition and optimization non-trivial, because it is not usually possible to account forthe sensitivity of the system to any one odor component in terms of any single che-mosensor within an array. In the converse case, where a set of highly specific sensorseach responding uniquely to a single component of the stimulus, optimization wouldbe straightforward, because the signals from sensors responding weakly to the com-ponents of interest should be amplified, and those responding to interfering or un-important components should be attenuated or ignored entirely. Furthermore, thedetection performance would be simple to quantify, because the detection of the sys-tem for each compound would be uniquely defined by the signal-to-noise performanceof the underlying sensor.The need for chemical sensor array optimization becomes obvious when we observe

that one set of chemosensors used to solve a given problem may be poor at solvinganother, new detection problem. This is especially true for small array sizes wheresensor diversity is limited, and sensor choice is more critical. But what propertiesof the array make the difference between it being suited to a particular detection pro-blem or not? Clearly, before we can address the issue of performance optimization wemust develop a rigorous framework for describing the criteria affecting the ability of anarray to solve the problem.The inability of an array to solve a defined detection or discrimination task might

result from one or more of four key factors:

1. There is insufficient sensitivity in any of the sensors within the array to the keycompounds of interest at the concentration levels required to solve the new task.

2. Those sensors sensitive to the key compounds relevant to the new task are too noisyto yield sufficient information to solve the task.

3. The array response to a repeated and identical stimulus is not sufficiently repro-ducible to permit discrimination between similar stimuli.

14 Chemical Sensor Array Optimization: Geometric and Information Theoretic Approaches348

4. There is insufficient sensor diversity within the array to discriminate between keycompounds relevant to the new task.We will refer to this as the ‘tuning’ of the array.

Issues one and two are very closely related because ultimately sensitivity is limited bynoise, therefore the real parameter of interest here is the signal-to-noise ratio. Issuethree can be considered as a special case of issue two, because sensor-response repro-ducibility can be quantified probabilistically in a similar way to noise. So we see that, ingeneral, the problem reduces to two basic issues, sensor noise (where wemight chooseto include sensor response reproducibility information) and sensor array tuning. Anycomprehensive scheme for performance definition or optimization of chemosensorarrays needs to take both these aspects into account.

14.2

Historical Perspective

Zaromb and Stetter recognized very early the need to quantify sensor-array perfor-mance [1]. In 1984 they considered the case of using an array of non-specific chemicalsensors for multicomponent gas analysis: a problem closely related to describing com-plex odors. By first assuming that the response of each sensor was binary to eachstimulus (response vs. no response) they argued for a combinatorial measure ofthe number of sensors required to detect a given number of chemical species

2n � 1 �XA

i¼1

m!

ðm � iÞ! i! ; ð14:1Þ

where n is the number of sensors within the array, m is the number of different che-mical compounds to be detected, and A is the maximum number of compounds(A � m) appearing as a mixture at any one time. This inequality provided a lowerbound on the number of sensors required to solve a particular sensing task. For ex-ample, according to Eq. (14.1) more than 18 sensors (n � 18) would be required todetect a tertiary mixture (A ¼ 3) taken from 100 single chemical compounds(m ¼ 100).Because the derivation of Eq. (14.1) was made on the basis of each sensor respond-

ing in a binary fashion to the stimulus, this severely limits the information provided byeach sensor and so the inequality produces a gross overestimate of the actual numberof sensors required to solve a particular problem – in practice the bandwidth within thesystem is far higher than suggested here. This limitation can be partially overcome byconsidering each sensor to respond in an n-ary fashion by splitting the full-scale sensorrange into p discrete domains and so the left hand side of Eq. (14.1) becomes (pn � 1),yielding a more realistic estimate of the number of sensors required to solve a parti-cular task. A more severe limitation, however, is the lack of any account of noise orsensor reproducibility in their analysis. This becomes obvious when considering thatthe bound given by Eq. (14.1) becomes meaningless in the extreme case where eachsensor responds in some completely arbitrary (random) fashion to the stimulus be-

4.2 Historical Perspective 349349

cause of extremely low signal-to-noise performance. Their analysis therefore applies tonoiseless systems which cannot be obtained in practice.Not until Gardner and Barlett’s paper in 1996 on performance specification for

chemosensor arrays was there any serious further treatment of this topic [2]. Theywere careful to consider noise to be central to the performance of these systems.By considering the chemical sensor array to perform a noisy (and therefore irrever-sible) mapping of a single point in the sample space to a spread of points in sensorspace, they were able to quantity the effect of individual sensor noise on array perfor-mance. They defined an error volume, Vn, as an ellipsoid within sensor space wherethe principal axes define the noise dispersion (or random error), rxi , of each sensorresponse, xi

Vn ¼2pn=2Pn

i¼1rxinCðn=2Þ ; ð14:2Þ

where Cð�Þ is the standard Gamma function. This equation provides a useful measureof the error introduced by noisy sensors. They then went on to define an importantquantity of the total number of array response vectors that may be discriminated, Nn,in view of this noise as

Nn �Pn

i¼1FSDðxiÞVn

ð14:3Þ

this being the total volume within sensor space divided by the error volume of a singlehyperellipsoid feature, where FSDðxiÞ gives the full-scale deflection of sensor xi. Whilethis is a useful measure of the theoretical limit to the number of distinct featuresidentifiable by an array in principle, in practice it is unlikely to be attainable becausenot all of the sensor space may be accessible by the system, depending on the range ofthe stimulus and the tuning of the array. As an extreme case, consider an array ofsensors each with identical sensitivities (tunings) to the stimulus. As we will seefrom the geometrical arguments below, the response of such an array would be con-fined to a 1D sub-space (line) oriented within sensor space and would be unable todiscriminate between any two compounds. As the dimensionality (n) of the array in-creases, this effect becomes more severe and usually electronic nose systems use anextremely small portion of the available sensor space as a result of the non-orthogonalsensor tunings and dynamic range of the stimulus. Consequently, the effects of arraytuning and range of the input are as fundamental as noise in defining the systemperformance. Also note that Eq. (14.3) is an approximate bound because it assumesoptimal packing of error hyperellipsoids in sensor space.Although the factors defining the performance of chemical sensor arrays for odor

analysis have been given some consideration during the development of electronicnose technology over the past twenty years, there still exists no comprehensive theoryof performance that can be widely applied to these systems. Without such a theory it isnot possible for amanufacturer, user, or researcher to specify the likely performance ofa given sensor array for a particular problem domain and, even more importantly,


optimize its performance for a given task. The lack of a clearly defined performancespecification is a real barrier to the uptake of electronic nose systems, because themanufacturers of competing chemical sensing technologies such as gas chromato-graphic or mass spectrometric-based instrument manufacturers are able to rigorouslyspecify detection limits for particular analytes, either individually or in combination.Current methods of specification for electronic nose systems are largely empirical,requiring vast numbers of measurements to be made to a wide range of single ana-lytes. Since these individual measurements cannot predict the overall system perfor-mance to complex mixtures of analytes that are routinely encountered in the realworld, this makes a complete empirical specification impossible for all but themost constrained and artificial of cases. Furthermore, system performance cannotbe quantified in any meaningful manner. Empirically based optimization strate-gies, which rely on databases of measurements to different stimuli, may be used,but usually the number of parameters to be optimized is prohibitive.The lack of a performance theory also means that any attempts at array and system

optimization must be carried out using empirically-based heuristic methods. Thereare no guarantees of optimizing the performance for chemical sensor arrays designedusing these methods, and the user cannot be sure that they have the best array for theirtask.In this chapter we discuss the recent work on this topic by the authors, which relates

both the array tuning and noise aspects to sensor-array performance. We believe thisrepresents a unified framework within which to rigorously define system performancethat provides the means to specify, and the foundation to optimize electronic nosesystems. Optimization measures are developed to characterize different aspects ofsensor array performance including system detection limits to specific odor stimu-li, a theoretical maximum of the number of odor features that may be detected bya chemical sensor array (to a given confidence interval), and the resolution of an arrayto neighboring odor stimuli (closely related to the signal-to-noise ratio).These measures may be widely applied independently of sensor technology, sensor

preprocessing methods, pattern recognition techniques, or the odor delivery system.Finally, we consider how these measures may be used within an optimization schemeto select the best chemical sensor array for a particular problem domain.

14.3

Geometric Interpretation

In order to demonstrate the effects of noise and tuning on array performance, we needto show the mapping between odor space and sensor space as carried out by a sensorarray. We firstly assume this to be linear although we will later drop this restriction.


14.3.1

Linear Transformations

We begin by considering a linear stationary chemical sensor model

x ¼ a0 þ a1c1 þ a2c2 þ :::ajcj þ :::þ amcm; ð14:4Þ

where x is the sensor response (note here there is no noise and so x is a deterministicfunction of the stimulus – later we will consider x to be a random variable whichfluctuates around some mean response value), cj gives the concentration of analytej, and aj defines the sensitivity of the sensor to the same analyte. The term a0 givesthe sensor response when no stimulus is present, often referred to as the ‘baseline’response for the sensor. Although this linear model only applies to a subset of che-mical sensor technologies (e.g. electrochemical cells and fluorescent indicators), andonly then up to an operating limit, more general models of sensor response will beconsidered in Section 14.5 after results have been developed for the linear case.An electronic nose may be modeled as comprising n sensors within an array, each

with potentially different sensitivity terms, aij. This linear model is convenient sincewe may apply linear algebra to represent the array as

x1x2...

xn

0BBB@

1CCCA ¼

a11 a12 . . . a1ma21 a22 . . . a2m... ..

. . .. ..

.

an1 an2 . . . anm

0BBB@

1CCCA

c1c2...

cm

0BBB@

1CCCAþ

a10a20...

an0

0BBB@

1CCCA ð14:5Þ

or simply

x ¼ Acþ a0; ð14:6Þ

where A is termed the sensitivity matrix and a0 the residual baseline vector for the array.Using this simplified view we may consider the array of sensors to be carrying out alinear (affine) geometric transformation between odor space, c, and sensor space, x.We may choose any basis for representing c and x, but the simplest for the purposes ofvisualization is over Rm and Rn respectively. Within this representation we can uni-quely define any combination of odor stimuli and with it a specific sensor array re-sponse.From Eq. (14.6) it is clear that the nature of the transformation between odor and

sensor space is uniquely defined by A and a0, which are properties of the array. Interms of the capability of the array to detect changes in the stimulus, the residualbaseline vector is of no interest, because it has no effect on the response of the arrayto different odor compounds – it acts only as an offset term. Consequently, we will notconsider a0 any further in our analysis. On the other hand, the sensitivity matrix isfundamental to the system performance as it determines the array response to thestimulus in the linear case, and so this will be the main focus of our discussion.It is instructive to visualize the action of the sensor array directly, by considering the

trivialized example of a 2-odor to 2-sensor transformation for a variety of sensitivity


matrices, as shown in Fig. 14.1. It is clear that the sensitivity matrix has a profoundeffect on the nature of the transformation between the odor space (domain) and thesensor space (range). In particular, for perfectly orthogonal sensors (with unit gain) asshown in Figu. 14.1a, where the sensitivity matrix is simply the identity matrix, I, notransformation occurs from the domain onto the range and so it preserves the area ofthe original odor space, in other words the transformation is isometric. However, asthe orthogonality of the individual sensor sensitivities decreases, as shown inFig. 14.1b, c, there is a noticeable collapsing of the domain onto the range so as torestrict the possible array response. In the other extreme, where the sensors are iden-tical, as shown in Fig. 14.1d, all points within the domain aremapped onto a single linein the range. Clearly, such an array would be unable to distinguish between the twoodor compounds, but would only be able to provide an estimate of the combined ana-lyte concentrations. From these observations we can define an important performanceparameter for an array, the hypervolume of accessible sensor space, Vs, which in eachexample is equal to the area spanned by the transformation of the domain onto range.It is noticeable in Fig. 14.1 that the total transformed area, Vs, is related to the ortho-

gonality of the two sensors. A well-known result from linear algebra states that, givenan affine transformation defined by a square matrix D, then a region of unit volumewithin the domain is transformed into a region within the range, the volume of whichis equal to the absolute value of the determinant of the transformation matrix, that is,jDj [4]. Consequently, if the possible linear combinations of odor stimuli covers a

Fig. 14.1 Visualization of a 2-odor to

2-sensor transformation for different

examples of linear sensitivity matrices,

A, a) orthogonal sensors through to d)

identical sensors. V0: Hypervolume of

accessible odor space, Vs: Hypervolume

of accessible sensor space. (Reprinted

with permission from Pearce [3])


defined volume in the domain, which we term the hypervolume of accessible odor space,V0, then in the m-odor to n-sensor (m ¼ n) case we have

Vs ¼ V0absðjAjÞ; ð14:7Þ

where V0 ¼ Pic0i gives the volume in odor space (c 0 is the maximum concentration

considered for a specific odor component). The absolute value must be taken becausethe determinant gives the ‘oriented volume’ which may be negative. The form ofEq. (14.7) is very similar to the array optimization measure proposed by Zaromband Stetter as long ago as 1984 [1]. However, they never discussed how this measureapplies generally to chemical sensor arrays, because the determinant is only definedfor a square matrix, and so can only be used when the number of odors is equal to thenumber of sensors. In general, electronic noses map many more odor componentsonto fewer sensors in order to discriminate between complex odors using as simplean array as possible. Consequently, we need to generalize the measure defined byEq. (14.7) for a transformation of arbitrary dimensionality that may be carried outby a chemical sensor array.To do this we need to consider an example transformation for which the sensitivity

matrix is not square, as shown in Fig. 14.2. This visualization shows how the cube ofunit side within odor space is mapped onto the plane in sensor space. Clearly, in thisexample the area defining Vs cannot be found by a single determinant. If we considerthe three 2 � 2minors (of order 2) ofA, then each of these represents how a single faceof the cube is transformed into the range. That is, each face of the cube is transformedinto a region in sensor space defined by its correspondingminor of A. So, for example,the face of unit area {(0,0,0),(0,0,1),(0,1,1),(0,1,0)} in the domain has a transformedarea equal to the absolute value of the determinant of the 2nd-order minor,

abs det:5 22 2

� �� ¼ 3, in the range.

Fig. 14.2 Visualization of a 3-odor to 2-sensor transformation, A ¼ 2 :5 21 2 2

� �


Furthermore, it is evident from Fig. 14.2 that the total region Vs comprises of thethree transformed perpendicular faces of the cube, suggesting the general result

Vs ¼Xm

p¼1

Xm

q 6¼p;q¼1

. . .Xm

r 6¼p;r 6¼q;r¼1

c 0pc0q . . . c

0r absðjMpq...r jÞ for m � n; ð14:8Þ

whereMpq...r is the minor of order n which is obtained by taking the columns (p, q,…,r)of A. Again the absolute value is taken because the areas defined by the minors mustbe additive. This result can be shown to apply generally m � n to any affine transfor-mation betweenm-dimensional odor space and n-dimensional sensor space [5], and somay be used to calculate the allowable space that may be accessed by a given array for astimulus volume. For an array of linear chemosensors Eq. (14.8) completely specifiesthe role of the array tuning in terms of defining the total volume of accessible sensorspace, which may be considered as the range of the system as a whole. For instance,applying Eq. (14.8) to the example shown in Fig. 14.2 gives a value for Vs ¼ 8:5, whichcan be easily verified using elementary geometry.

14.4

Noise Considerations

Although the performance of perfectly specific chemical sensor array (such as onewhere the off-diagonal terms of A are zero) is simple to characterize – by simply mea-suring the detection limit of the sensors individually – the case for cross-sensitivesensors is less straightforward. In the latter case, the overall sensitivity of the arrayto an individual compound arises from the combined sensitivity of a number of de-vices. Consequently, it is necessary to understand how these individual sensitivitiescontribute to the array performance.

14.4.1

Number of Discriminable Features

So far we have considered the transformation carried out by a sensor array to be noise-less, that is, there is a perfect correspondence between points within odor space andpoints within sensor space. In the noiseless case, the magnitude of Vs is unimportantsince it is always possible to perfectly resolve neighboring points in odor space, nomatter how close in proximity.In practice, of course, all measurements are limited by noise and so chemical sen-

sors generate a non-reproducible response to the same stimulus. Instead of therebeing perfect correspondence between odor and sensor space, we must now viewthe noise process as mapping single points in the stimulus space onto a region (usual-ly small) in sensor space where the likelihood of obtaining a particular measurement isdetermined by some probability density function. When the noise process is intro-duced into the transformation, the magnitude of Vs becomes of great importance be-


cause it restricts the total number of discriminatable features for a given significancelevel.In the simplest case, where the noise in each sensor is considered to be independent

of both the stimulus and the response magnitude, wemay define a confidence intervalin sensor space as anm-dimensional hyperellipsoid, where the cross-section along theprincipal axes is given by dxi ¼ �rxi , the standard deviation of the noise (or randomerror) for sensor i. A representation of the noise process combined with the sensorarray transformation is shown in Fig. 14.3 for the 2-sensor case where the region Vs ispacked by the error ellipses. After Gardner and Bartlett [2], each ellipse corresponds toa single stimulus point in the domain, the number of ellipses that may be packed intothe region Vs gives the number of discriminable odor features,Nn, a bound for which wasgiven in Eq. (14.3). By also taking into account the accessibility of the sensor space for adefined region of the sensor space, as discussed, we can estimate the number offeatures which can be discriminated by the array on the average

Nn �Vs

Vn

: ð14:9Þ

Most importantly Eq. (14.9) provides an estimate of the number of discriminable fea-tures that can be coded by a chemosensor array, taking into account both noise and

Fig. 14.3 Sensor space representation where sensor noise has been

represented as ellipses superimposed on the region Vs. The addition

of the noise components for each sensor dxi is equal to rxi . For illustrationpurposes we assume here that the noise is independent of the stimulus

or sensor response. (Reprinted with permission from Pearce [3])


array tuning. The value of Vs limits the access to the sensor space depending on thedynamic range of the stimulus and the array tuning, through Eq. (14.8). By using theformulæ given for Vs in the linear case, Eq. (14.8), and non-linear case, Eq. (14.23) (aswill be discussed in Section 14.5), it is possible to produce an estimate for Nn for anychemical sensor array.

14.4.2

Measurement Accuracy

Of particular interest is how the noise generated in sensor space determines the mea-surement accuracy of the array to individual components of the odor stimulus. Thismay be achieved by considering the inverse mapping of noise in sensor space ontoodor space. We first define a noise matrix, gx, which comprises each of the sensorerrors as a diagonal matrix of the form

gx ¼

rx1 0 . . . 0

0 rx2...

..

. . ..

00 . . . 0 rxn

0BBBBB@

1CCCCCA: ð14:10Þ

We can now quantify the inverse transformation of the noisematrix, gx, viaA�1 so as to

generate the noise components in terms of the odor space, to give a correspondingdetection limit for each individual odor component, Dc. This corresponds to solvingthe system of equations gx ¼ ADc for Dc. Depending on the form of A there are threepossible cases to consider, as shown in Table 14.1.The most straightforward case is where there are the same number of odor com-

ponents as there are sensors, which produces a square matrix A, and is of full rank (allsensors are linearly independent but not necessarily orthogonal, if this is not the casethen we consider the system to be underdetermined). The overdetermined case occurswhen there aremore sensors than individual chemical compounds, given that the rankof A is m. Because of the typically high dimensionality of the stimulus in the case ofolfaction, the overdetermined case would not be usual. However, it is of direct interestto researchers who use arrays of broadly tuned chemical sensor arrays for single gasanalysis or sensing mixtures of gases using such systems. This case is dealt with inAppendix 14.A. More usual in electronic nose systems is the underdetermined casewhere there are more odor compounds than independent sensors within the array.This case is studied in Appendix 14.B, assuming that the distribution of the stimuliis Gaussian.In the case where n ¼ m and A is of full rank, there is no loss of dimensionality

during the forward transformation, i.e. Vs > 0. A unique two-sided inverse exists,A–1, and each point within the domain has a one-to-one mapping with points inthe range (subject to noise constraints). Now

DC ¼ A�1gx; ð14:11Þ


Table 14.1 Possible cases of transformations between odor and

sensor space, showing examples for each case

Odor space Sensor space

Uniquely determined

ðn ¼ mÞ \ ðVs > 0Þ

Overdetermined ðn > mÞ

Underdetermined

ðm > nÞ [ ðn ¼ m \ Vs ¼ 0Þ


and so the detection limit for the array is simply the noise matrix scaled by the ele-ments of the two-sided inverse of the sensitivity matrix and is therefore simple tocalculate. The solution is then of the form

DC ¼

dc11 dc12 . . . dc1ndc21 dc22 . . . dc2n... ..

. . .. ..

.

dcm1 dcm2 . . . dcmn

0BBB@

1CCCA; ð14:12Þ

where each column ðdc1i; dc2i; . . . ; dcmiÞT gives the noise vector for sensor i projected

onto odor space, and each row (dcj1; dcj2; . . . ; dcjn) gives the noise components foreach sensor projected onto the odor component j. These noise components may act inthe same or opposite directions and so the total squared error for the array is

e2 ¼Xn

i¼1

Xm

j¼1

dc2ji; ð14:13Þ

whereas the overall contribution of sensor i to error in odor space is

e2xi ¼Xm

j¼1

dc2ji; ð14:14Þ

and finally the total error produced by all the sensors for odor component j is

e2cj ¼Xn

i¼1

dc2ji: ð14:15Þ

The latter expression is particularly important because it provides measure of the de-tection limit of the noisy chemosensor array to each compound j owing to the arraytuning and noise properties.Finally, it is also useful to define a signal-to-noise ratio for neighboring points in

odor space. This tells us how easy it will be to discriminate between these points giventhe array tuning and noise performance. For two given stimuli separated by Dc we seethat this corresponds to a sensor response of magnitude

Dx ¼ ADc; ð14:16Þ

which leads to the local signal-to-noise ratio for stimulus difference Dc

SNRDc0 ¼ kDxk2

trðgxgTx Þð14:17Þ

where k � k is the Euclidean vector norm and tr(�) is the matrix trace operation.This measure is extremely useful because it allows us to predict the likelihood ofdiscrimination between two neighboring points in odor space using a particular che-mosensor array.


To apply this theory, the experimentalist or practitioner needs to be able to providesuitable values for the parameters within Eq. (14.11) and Eq. (14.17). In particular,measuring values for A and gx is the key requirement. The sensitivity matrix, A,can be measured directly by varying individual stimulus components and calculatingthe regression parameters of a linear fit between concentration and sensor response(least squares). Because at this point themodel assumes a linear behavior, it is straight-forward (although time consuming) to estimate all of the values for A, because thesensitivity of each sensor to a particular compound can be measured independentlyand then assumed to sum linearly in our model. Therefore, over some linear operatingregion (often assumed to be for low concentrations), regression parameters for theconcentration dependence of each sensor to each compound can be estimated directlyfrom the sensor response data. Of course, the number of individual compounds maybe too high to be able to realistically estimate an individual sensitivity between eachsensor and each compound. However, note that many compounds may be groupedtogether to act as a single component (dimension in our model) as long as the sensorresponds linearly to the mixture over the operating region.Estimating values for gx provides more of a challenge because it requires estimation

of noise properties in each of the sensors. The model assumes the noise for eachsensor is constant over the stimulus range and is independent of noise sources inother sensors (later we will show that we can also deal with stimulus-dependent noiseproperties). This assumption makes estimation of the standard deviation of the noisestraightforward.There may be two forms of noise that the practitioner might wish to take account of

when using the model. First, intermediate to high-frequency noise in the sensor re-sponse (arising from instantaneous noise sources in the sensor or interface electro-nics), which may be quantified from the fluctuations in the time series of data to nostimulus or constant stimulus. The second form of noise is the reproducibility of theresponse of the sensor to repeated stimulus. This would require repeating identicalstimuli many times and quantifying the dispersion of responses in each of the sensors.Because the noise is assumed here to be independent, then the noise can be charac-terized independently in each sensor. If the noise varies over the stimulus range then amean value can be assumed for the purposes of the linear model. If any of these re-strictions do not seem reasonable given the data available for the sensors being opti-mized, then a more complex, non-linear model such as those discussed below, willneed to be considered.

14.4.3

2-Sensor 2-Odor Example

Some of the concepts become clearer through a trivialized example of a 2-linear sensorarray responding to 2-odor compounds. To simplify the calculations, we assume thatboth sensors have the same noise r and that this is independent of the stimulus orsensor response. The sensitivity matrix is then simply


a11 a21a12 a22

� �; ð14:18Þ

and gx is

r 00 r

� �; ð14:19Þ

and so applying Eq. (14.11) we obtain the solution

DC ¼ ra11a22 � a12a21

a11 a21a12 a22

� �; ð14:20Þ

giving the formula for the total squared error for the array as

e2 ¼ r2a211 þ a212 þ a221 þ a222ða11a22 � a12a21Þ

2 : ð14:21Þ

As an example of performance optimization we might wish to choose a11, a12, a21, anda22 in order to minimize this error. Clearly, a unique solution is not possible, but byfixing the sensitivities of one of the sensors, say a1j, we can visualize the effect on theerror as we vary the tunings for the other sensor, say a2j (Fig. 14.4). The results areintuitive by considering the situation when one sensor possesses sensitivity terms thatare multiples of the other (i.e. the sensors are identical after normalization). In this

Fig. 14.4 The effect on the optimal squared estimation error, e2, from variations in the

tuning of one sensor within an odor sensing array of two sensors, after fixing the sen-

sitivities for the other sensor. The array is composed of 2-linear sensors with Gaussian

noise, where the tunings of one of the sensors is fixed, a11 ¼ 1; a12 ¼ 0:5. (Reproduced

with permission from Sanchez-Montanes and Pearce [6])


case, the array is unable to distinguish between the individual stimuli so the recon-struction error tends asymptotically towards infinity, reflecting the impossibility ofdiscrimination between the separate stimuli in this case. This is represented bythe ridge along the center of Fig. 14.4, (left), where the ratio between the sensitivityterms a21 : a22 is 2:1. If we constrain each of the sensitivity terms to the range [0, 1] (i.e.the sensor response can only increase from its baseline value and its sensitivity isconstrained), then the best performance is obtained when a21 ¼ 0 and a11 ¼ 1,that is, when the second sensor is as different as possible from the first sensor withinthe specified constraints.

Table 14.2 Models of concentration dependence for a variety of

chemical sensors and their behaviors. All models assume that no

competition for sites within the sensor takes place, and that chemicals

act independently on the sensor. (Reprinted with permission from

Pearce [3])

Device Model Behavior

Electrochemical fuel cell,

fluorescent indicators

Linear x ¼Pm

j¼1 ajcj þ a0

Metal oxide semiconductor Power x ¼Pm

j¼1 ajcij þ a0

Conducting polymer Langmuir x ¼Pm

j¼1½bj aj cj �1þaj cj

þ a0


14.5

Non-linear Transformations

Because only a subset of chemical sensors is considered to behave linearly up to anoperating limit, it is necessary to extend the methods developed in Sections 14.3 and14.4 so that they may be applied more generally.The concentration dependence of themost popular chemical sensor types to be used

within electronic nose systems are shown in Table 14.2. Of these, metal-oxide semi-conductor sensors are arguably the most widely used in existing systems. These havebeen modeled by a power law, where ri typically lies between 0.6 and 0.8 but may also

Fig. 14.5 (a) Visualization of 2-odor to 2-sensor transformation using

the non-linear power lawmodel for metal oxide semiconductor devices:

x1 ¼ a11cr11 þ a12c

r12 , x2 ¼ a21x

r21 þ a22c

r22 , where r1 ¼ r2 ¼ 0:8 and

a11 ¼ 0:8, a12 ¼ 0:25, a21 ¼ 0:6, and a22 ¼ 0:25. (b) Plot of the

determinant of the Jacobian for the same 2-sensor metal oxide device

array showing how the localized feature volume varies with the

stimulus. (Reprinted with permission from Pearce [3])


depend on the stimulus. Conducting polymer devices are also very popular for usewithin chemical sensor arrays and these have been described as behaving accordingto a Langmuir isotherm model.For these and other non-linear sensors, a sensitivity matrix may be formed in the

non-linear case from the Jacobian matrix, A

Fig. 14.6 (a) Visualization of 2-odor to 2-sensor transformation using

the non-linear Langmuir isotherm model for conducting polymer de-

vices: x1 ¼a11c1

1þa11c1þ a12c2

1þa12c2, x2 ¼

a21c11þa21c1

þ a22c21þa22c2

, (b) plot of the deter-

minant of the Jacobian for the same 2-sensor conducting polymer

device array showing how the localized feature volume varies with the

stimulus. (Reprinted with permission from Pearce [3])


A ¼

@x1@c1

@x1@c2

. . .@x1@cm

@x2@c1

@x2@c2

. . .@x2@cm

..

. ... . .

. ...

@xn@c1

@xn@c2

. . .@xn@cm

0BBBBB@

1CCCCCA

��c1;c2;...;cm

ð14:22Þ

for some operating point (c1, c2, …, cm) in odor space. This linearized sensitivity matrixmay then be used in place ofA as defined by Eq. (14.6) so that the analysis developed inSections 14.3 and 14.4 may then be applied in the general non-linear case. The deter-minant of the Jacobian, | ¼ jAj, may then be used to approximate the localized hyper-volume for the transformation for a particular operating point, which we call the lo-calized feature volume.Furthermore, the Jacobian may also be applied to calculate the hypervolume of

accessible sensor space in the non-linear case, because

Vs ¼ðc 0m

0. . .

ðc 020

ðc 010| dc1dc2 . . . dcm: ð14:23Þ

The fitting of experimental data for the practitioner using these non-linear models isstraightforward. Rather than finding the regression parameters that fit the concentra-tion dependence of sensor reponse in the linear case, we should now estimate theregression parameters of the model in the general non-linear case. Such non-linearregression can be achieved by most statistical software packages.Because of the nature of the models described in Table 14.2, the action of each of the

compounds still sums linearly (even though their dependence on individual com-pounds may be non-linear) and so the sensor response to each compound may beanalyzed independently. More complex models of analyte competition for sites ineach sensor could be developed and may still be applied using Eqs. (14.22) and(14.23). As with the linear models, the noise is considered to be independent ofthe stimulus. The Fisher information approach, to be described below, should beused in the case of stimulus-dependent noise.Examples of calculations for the noiseless non-linear case are shown for metal-oxide

semiconductor sensors in Fig. 14.5 and for conducting polymer sensors in Fig. 14.6,using the sensor models summarized in Table 14.2. For both examples, the non-linearmapping of 2-odor space is shown, showing how the non-linearity in each sensorcontributes to the transformation as a whole. The nature of the models impliesthat the metal oxide semiconductor devices are far more linear in their behavior,which is verified by contrasting the mappings onto sensor space for both sensor vari-eties. In particular, the localized volume of the transformation in the conducting poly-mer case is shown to tend towards zero with increasing stimulus concentration. This isalso shown by Figure 14.6 which shows the linearized Jacobian at different points inthe stimulus space. As c1; c2 ! 1 then | ! 0, verifying the observation. In contrast,the determinant of the Jacobian for the metal-oxide array never reaches zero, becausethe behavior of these sensors is more linear. Note that the same analysis on an array of


linear sensors would produce a perfectly flat feature localized volume. Hence the lo-calized feature volume map provides an intuitive visualization of the performance ofthe sensor array in detecting the stimulus.

14.6

Array Performance as a Statistical Estimation Problem

We can also consider the definition of chemosensor array performance in a differentcontext, one which we will show provides certain advantages in the calculation of thearray error. Here we consider the data produced by a chemosensor array as being partof a statistical estimation problem as outlined in Fig. 14.7. Each sensor within the arrayproduces a response dependent on its tuning to the stimulus plus some noise. A hy-pothetical statistical estimator (one produced using, for example, maximum likelihoodor Bayesian estimation methods) uses the noisy response from the array to attempt toreconstruct the stimulus.Because of this noise, if we present the same stimulus c to the system several times,

the estimator response c will not be the same on each occasion but will fluctuatearound a certain mean value. An estimator should be right on the average, that is,if we present the same stimulus c many times, the mean of the different estimationsc should be equal to c. If the estimation satisfies the property we call it unbiased.Moreover, the variance of the response of the estimator when the stimulus is fixedshould be as small as possible. If the estimator is unbiased, its squared error inthe estimation coincides with its variance.Depending on the tuning parameters of the individual sensor elements and their

noise properties, the accuracy of the overall sensor system in estimating the stimulusvaries in addition to the range of stimuli that may be tested. A typical goal in choosing

Fig. 14.7 A hypothetical statistical estimator takes the response

vector x, from a sensor array and uses this in order to estimate

(reconstruct) the stimulus. The tuning parameters for each of

the sensors are represented as parameters to the sensor array.

(Reproduced with permission from Sanchez-Montanes and

Pearce [6])


which sensors to incorporate into an artificial olfactory system is to maximize theaccuracy with which the sensory system can estimate/predict the stimulus or opti-mally discriminate between similar stimuli. By considering a hypothetical unbiasedstatistical estimator that uses the sensor array response in order to estimate the indi-vidual stimuli within a complex odor mixture, we can define and test how differenttuning parameters of the sensor array effect the accuracy of stimulus reconstruction.This arrangement is shown in Fig. 14.7 where each sensor, i, generates a response, xi,to the multicomponent stimulus c.Conveniently, our problem when placed in this context is well known to the field of

statistical estimation, and classical results exist that we can call upon here. For exam-ple, the variance of any unbiased estimator that might be constructed for this purposehas a well defined limit through the “Cramer-Rao bound”, which we will make use of[7]. Furthermore a direct relationship between the Cramer-Rao bound and Fisher in-formation exists that allows us to calculate this bound, and therefore quantify the per-formance of the array in reconstructing the stimulus.

14.7

Fisher Information Matrix and the Best Unbiased Estimator

When a multicomponent odor stimulus c is exposed to the sensor array, the array ofsensors gives a response x, of which component i denotes the response of sensor i.Because of the noise and nonreproducibity of the sensor, the array response is notdeterministic so it follows some probability density function pðxijcÞ conditioned onthe stimulus. The elements of the Fisher information matrix (FIM), Jjj 0 ðcÞ, are definedas [7]

Jjj 0 ðcÞ ¼ðdx pðxjcÞ @

@cjln pðxjcÞ

!@

@cj 0ln pðxjcÞ

!; ð14:24Þ

where j and j 0 are both individual stimulus components. Some understanding of whatis being measured by the Fisher information can be gained by considering the sim-plified case of a single sensor responding to a single odor component. For ease of theanalysis, let us assume that the sensor responds to stimulus concentration with aGaussian tuning curve (note this is not physically reasonable for a chemosensorbut is for illustrative purposes only). In this case we have the situation shown in Fi-gure 14.8a. Now Eq. (14.24) reduces to

JðcÞ ¼ 1

r2df ðcÞdc

� �2

; ð14:25Þ

where f(c) is the mean sensor output to the stimulus c, hxjci, in this example followingthe Gaussian curve (see Fig. 14.8a) and r is the standard deviation of the noise shownas error bars in the same figure. From this simplified example we see that the Fisherinformation scales inversely with the noise variance but is linearly dependent on the

14.7 Fisher Information Matrix and the Best Unbiased Estimator 367367

square of the slope of the tuning curve (here concentration dependence). We see thatthe slope of the tuning curve is greatest at the inflexion points of the Gaussian, whichfrom Eq. (14.25) is also where the Fisher information is maximum, Fig. 14.8b. At thepeak of the Gaussian where the slope is zero, the Fisher information is also zero,Figure 14.8c. This result is intuitive because if we wish to measure a small changein the stimulus it is far better to be operating on the slopes of the tuning curve, wherewe obtain a relatively large change in sensor output for a given stimulus change, com-pared to at the peak, where the change in sensor response will be close to zero. So wesee that the Fisher information concisely describes the combined role of sensor tuningand noise in defining estimation performance.Although the Fisher information may not be straightforward to interpret it directly,

we can relate it to the reconstruction error of the stimulus through the Cramer-Raobound. This states that for every unbiased estimator that uses the data x for estimatingthe stimulus c, as c, the squared error for stimulus component j satisfies

varðccjjcÞ � ðJ�1ðcÞÞjj e2cj

D E

opt; ð14:26Þ

where var means variance, �h i is the expected value or mean, and ccj is the estimation ofthe component j of c, j ¼ 1; . . .m [7]. And so this also provides a valuable link to thegeometric theory of array error considered in Appendix 14.c.This result allows us to directly calculate the minimum expected reconstruction

error for a given stimulus component j from the jth diagonal element of the inverseof the FIM. Furthermore, the total expected squared reconstruction error across theentire array is equal to the summation of the errors in each of the components. That is,

varðccjcÞ ¼Xm

j¼1

varðccjjcÞ �Xm

j¼1

ðJ�1ðcÞÞjj e2� �

opt; ð14:27Þ

Fig. 14.8 In this example the sensor is characterized by (a) a bell-

shaped tuning curve with overlapping Gaussian noise. The bars show

the standard deviation of the noise. (b) The points that maximize Fisher

information are those where the slope of the receptive field is higher.

(c) Points where the slope is zero make the Fisher information

minimum


and so the overall performance of the array in detecting all of the stimuli is defined bythe elements of the FIM, Jjj 0 . In Appendix 14.C we show that the Fisher informationand geometric approaches to sensor array optimization are equivalent when the noisein the sensors is independent of the stimulus. In the case of stimulus-dependent noise,the Fisher information approach should be used.There is another notion related to Fisher information called ‘discriminability’. This

measures the ability of the system to distinguish between two similar stimuli c1 and c2.If we call Dc ¼ c2 � c1, the ability of the system to discriminate between these is givenby

d 0 DcTFDc: ð14:28Þ

The maximization of this quantity can be shown to be equivalent to the maximizationof the local signal-to-noise ratio defined in Eq. (14.17).We now need to be able to calculate the FIM for different sensor array configurations

in order to proceed.

14.8

FIM Calculations for Chemosensors

First, the FIM for an individual sensor i is given by the elements of the matrix

Jijj 0 ðcÞ ¼ðdxi pðxijcÞ

@

@cjln pðxijcÞ

!@

@cj 0ln pðxijcÞ

!: ð14:29Þ

It can easily be shown that when the array of sensors has uncorrelated noise, the FIMof the entire array, J, is equal to the summation of the individual FIMmatrices for eachsensor i, that is

Pi J

i. This is valid in a general sense – in other words the noise andconcentration dependence of the sensors can be different across the array and cancomprise different sensor technologies, noise properties and tunings.We now calculate the FIM elements for two example cases of chemical sensor by

substituting the appropriate probability density function into Eq. (14.29) and rearran-ging.Case 1: Analog chemical sensor with Gaussian noise:

Jijj 0 ðcÞ ¼1

r2xiðcÞ@fxiðcÞ@cj

@fxiðcÞ@cj 0

þ 21

r2xiðcÞ@rxiðcÞ@cj

@rxiðcÞ@cj 0

ð14:30Þ

where fxi is the mean response for sensor xi, i.e. fxiðcÞ ¼ xijc� �

, which would be ex-pected to follow some model of concentration dependence, e.g. a simple linear modelsuch as given by Eq. (14.1). However, the sensor model for the concentration depen-dence can be a far more complex, non-linear one. Note also that, in principle, the noisedispersion can depend on the stimulus. The Gaussian noise case is most appropriatefor describing metal-oxide semiconductor and conducting polymer chemosensors

14.8 Fisher Information Matrix Calculations for Chemosensors 369369

used within electronic nose systems, where the partial derivatives can be calculated forthe sensor models given in Table 14.2.Case 1: Analog chemical sensor with Laplacian noise:

Jijj 0 ðcÞ ¼1

a2xiðcÞ@fxiðcÞ@cj

@fxiðcÞ@cj 0

þ 1

a2xiðcÞ@axiðcÞ@cj

@axiðcÞ@cj 0

ð14:31Þ

where axi (c) is the dispersion parameter of the Laplacian noise for that sensor. TheLaplacian case is most appropriate for describing fluorescence-based optical chemo-sensors used within artificial olfactory systems, where the concentration dependenceis approximately linear up to saturated vapor pressures of analyte [8].

14.8.1

2-Sensor 2-Odor Example

To illustrate these concepts we again consider two linear sensors to generate an analogresponse that is corrupted by Gaussian noise, identical to the example given in Sec-tion 14.4.1. This is a linear model and so the sensitivity of sensor i to stimulus com-

ponent j is a constant aij @fxi

ðcÞ@cj

. Using Eq. (14.30) we can calculate the FIMs for eachsensor

J1 ¼ 1

r2a211 a11a12

a11a12 a212

� �J2 ¼ 1

r2a221 a21a22

a21a22 a222

� �

Adding these to form the FIM of the array, then substituting into Eq. (14.27) and rear-ranging we obtain exactly the same form as Eq. (14.21), demonstrating equivalencebetween the geometric and information theoretic approaches in this case. In Appen-dix 14.C we show that this equivalence holds for any input dimension.

14.9

Performance Optimization

An outline of the optimization problem we will consider is shown in Fig. 14.9. A poolof k different sensor types is available, each with a unique profile of response to themdistinct molecular species relevant to the problem. Our instrument provides n chan-nels, each of which we can assume may house any of the available sensors. Further-more, we will not consider duplication of sensor types in the array since this yields noadditional information about the stimulus, but acts to reduce the noise in the system ifaveraging is employed (this case can be dealt with for independent noise by replacingthe l identical sensors in the calculations with a single sensor of the same type but withnoise variance, r 0

xi2 ¼ r2xi

l Þ.The optimization problem is then to select the single configuration that provides the

best sensing performance to the compounds of interest out of kn

� ¼ k!

n!ðk�nÞ! possible


configurations. What is best here depends on the detection task to be solved. We en-visage three possible criteria to be optimized in a practical system

1. Maximize the total number of Nn separate features that can be detected by an array.This is optimizing the range of the system and can be directly quantified from thegeometric approach (Eq. 14.9). Shannon information theoretic approaches aremore suited to calculating this value than the Fisher information [7].

2. Maximize the signal-to-noise ratio obtained from the array for some vector or set ofvectors in stimulus space. This is optimizing the resolving power or discriminationability of the array and may be quantified using either the geometric (Eq. 14.17) orFisher information (Eq. 14.28) approaches.

3. Estimate the concentrations of some of the compounds or some function of these,e.g. interfering compounds (distractors) could be present. This is optimizing thedetection threshold of sensitivity for the system to specific components, which canbe quantified using either the geometric or Fisher information approaches.

The case where we are interested in reconstructing the concentration of all the stimu-lus compounds has been extensively described in this chapter.

14.9.1

Optimization Example

We will illustrate the Fisher information maximization principle with a simple exam-ple. Consider a set of linear chemosensors each responding to combinations of threesingle odor components (m ¼ 3). The noise in all the sensors available to us is as-sumed to be Gaussian, independent of each other and the stimulus, and with equal

Fig. 14.9 A cartoon of the op-

timization problem for chemical

sensor arrays

14.9 Performance Optimization 371371

variance. For this example we assume each sensor is available with five graded levels ofsensitivity to each of the three components, that is, 0, 0.25, 0.50, 0.75 or 1.0. Therefore,there are k ¼ 53 ¼ 125 possible sensor types. We would like to select a sensor arrayconsisting of any three of these available sensors (that is, n ¼ 3). Therefore, we shouldselect those three sensors from the 125 available that optimize the performance of thesystem in terms of the overall reconstruction error (criteria 3 above). For each of the

125!ð125�3Þ!3! ¼ 317,750 possible array configurations we calculate the system Fisher infor-mation as previously described, in order to evaluate their performances.In Table 14.3 we show the three best groups of solutions. Note that the optimal

configuration is formed by sensors with non-zero sensitivities as well as zero sensi-tivities i.e. they are mixed. The non-zero sensitivities are maximum in each case, show-ing that intermediate sensitivities are disregarded. This is intuitive, because providingas much gain as possible to each of the analytes will maximize the performance underall three optimization criteria discussed above – increased gain is always advantageousas long as it is not commensurate with equal amounts of noise. Importantly, the spe-cific case (in which each sensor responds to a different component with maximumgain while its sensitivity to the others is zero) is not the best in our example (the ex-pected squared error of this configuration is exactly 3r, in units of the standard devia-tion of the noice). This demonstrates that even if it were possible to develop perfectlyspecific sensors for given compounds, this would not yield the best possible perfor-mance for electronic nose systems, because some amount of overlap in sensor re-sponse is shown to be advantageous. Interestingly, the sensors forming the optimalconfiguration tend to have the same number of zero and non-zero sensitivities as theinput dimension increases (data not shown). The number of zero sensitivities in eachsensor of the optimal configuration tends to be the same as the number of zero sen-sitivities as the input dimension increases (data not shown).Arrays formed by non-independent sensors (some linear relationship between the

sensitivities exists within the array) have infinite expected error because they are notable to discriminate between three-dimensional stimuli. For example, the singularconfiguration shown in Table 14.3 has only 2 independent sensors. Therefore, itcan only discriminate between two-dimensional stimuli.

Table 14.3 Best three groups of solutions in the optimization ex-

ample, and one singular solution (dependent sensors). The errors are

given in units of the noise variance. The best error is achieved by just

one solution (shown in the table), while the second best error and third

best error are each achieved by six solutions (corresponding to replace

a 1 sensitivity with 0.75 and 0.25 respectively). The table shows in-

stantiations of these sub-optimal solutions

Best 2nd Best 3rd Best Singular

Sensitivities a1 a2 a3 a1 a2 a3 a1 a2 a3 a1 a2 a3Sensor 1 0 1 1 0 0.75 1 0 1 1 0.5 0.25 1

Sensor 2 1 0 1 1 0 1 1 0.25 1 0.25 0 0.25

Sensor 3 1 1 0 1 1 0 1 1 0 0.25 0.25 0.75

e2 2.25 2.51 2.65 infinity


The errors of all the 317 750 possible arrays are sorted and shown in Fig. 14.10.Critically, the error of any given configuration can be orders of magnitude greaterthan the error of the optimal configuration. Therefore, if the sensory array is designedrandomly choosing three of the available sensors, we are likely to select a far-from-optimal configuration. We stress this point to indicate the importance of optimizationin chemical-sensor-array design. For example, the probability of having an expectedsquared error more than 100 times the optimal one is 22.46% (see Fig. 14.10).The technique illustrated in this example can be analogously used in general

conditions: non-linear sensor noise that depends on the stimulus, other types ofnoise (non-Gaussian), bipolar sensor sensitivities, and arrays of sensors with differentkinds of responses and noises [6]. Many more complex examples can be easily con-structed.

14.10

Conclusions

In this chapter we have described two unified theories of chemical-sensor-array per-formance, using both geometric-based linear algebra and Fisher information ap-proaches. The theories may be applied in a variety of conditions such as differentsensor noise properties and different concentration-dependence models. More gen-erally, any variety of different sensor types may be optimized within the same ar-ray. The geometric theory is particularly suited to visualization of the sensor arrayperformance and the Fisher information copes with more complex scenarios, wherefor example the sensor noise is dependent on the stimulus.The utility of the approaches for array optimization is demonstrated using a number

of simple examples that serve as the basis for more realistic applications of the theory.Manufacturers of electronic nose instruments may easily apply this theory in order tooptimize the sensing performance of the systems they sell. Furthermore, we can en-visage a catalog of parameters for each sensor used within practical systems today,which would make the optimization of sensor arrays to particular detection tasks asimple and routine operation.

Fig. 14.10 Expected squared error for all possible

sensor array configurations. The configurations are

ranked according to their squared error. The grey

zone indicates solutions where an unbiased exti-

naty is comfortable its construct (dependent sen-

sors). The error is normalized by the best expected

squared error. Dotted line indicates the configu-

ration in which the squared error starts to be

greater than 100. The percentage of configurations

whose error is greater than 100 is 22.46%


Acknowledgments

T.C.P. was supported by grant IST-2001-33066 from the European Commission andGR/R37968/01 from the United Kingdom Engineering and Physical Sciences Re-search Council. M.A.S-M. was supported by grant BFI2000-0157 from MCyT.

References

1 S. Zaromb, J. R. Stetter. Theoretical basis foridentification and measurement of air con-taminants using an array of sensors havingpartly overlapping selectivities. Sensors &Actuators, 6 (1984) 225–243.

2 J. W. Gardner, P. N. Bartlett. Performancedefinition and standardization of electronicnoses. Sensors & Actuators B, 33 (1996)60–67.

3 T. C. Pearce. Odor to sensor space trans-formations in biological and artificial noses,Neurocomputing, 32–33 (2000), 941–952.

4 J. R. Wicks. Linear algebra an interactiveapproach with Mathematica, Addison-Wesley, 1996.

5 J. R. Wicks. personal communication.6 M. A. Sanchez-Montanes, T. C. Pearce.

Fisher information and optimal odorsensors, Neurocomputing, 38–40 (2001)335– 341.

7 T. M. Cover, J. A. Thomas. Elements ofInformation Theory, John Wiley, 1991.

8 T. C. Pearce, P. F. M. J. Verschure, J. White,J. S. Kauer. Robust stimulus encodingin olfactory processing: hyperacuity andefficient signal transmission, in Neuralcomputation architectures based on neuro-science, (eds. Wermter S., Austin, J., andWillshaw D.), Spinger-Verlag 2001.


Appendices

14.A

Overdetermined Case

The Fisher information approach described in the main text operates correctly in theoverdetermined case. However, for the geometric approach described in the main text,we must find the least squares solution which leads to

DC ¼ ðATg�2x AÞ�1ATgx ð14:32Þ

and

e2 ¼Xm

j¼1

Xn

i¼1

dc2ji: ð14:33Þ

It can be easily verified that when A is square and non-singular these equations are thesame as Eqs. (14.11) and (14.13) respectively.

14.B

General Case with Gaussian Input Statistics

Here we consider the global optimal estimator (biased or unbiased) which minimizesthe global expected error. When the sensors are linear and the noise is Gaussian thisminimum error can be shown to be trðATg�2

x Aþ V�1Þ�1Þ, where V is the covariancematrix of the input stimuli, which are assumed to be Gaussian distributed. This equa-tion is valid for all the cases (square A, underdetermined and overdetermined cases) aswell as when the input statistics are not homogeneous, and so is the most generalresult.

14.C

Equivalence Between the Geometric Approach and the Fisher Information Maximization

BecauseP

i;j x2ij ¼ trðxxT Þ, using Eq. (14.32) we can rewrite Eq. (14.33) as

e2 ¼ trððATg�2x AÞ�1ATgxg

Tx AððATg�2

x AÞ�1TÞ ¼ trððATg�2x AÞ�1Þ ð14:34Þ

On the other hand, if the sensors are linear and their noise does not depend on thestimulus, Eqs. (14.30) and (14.31) can both be expressed as

Jijj 0 ðcÞ ¼ ðgxÞ�2ii aijaij 0 ð14:35Þ

Then the total Fisher information matrix is J ¼ ATg�2x A so by Eq. (14.27) the optimal

error is simply trððATg�2x AÞ�1Þ, which coincides with that derived for the geometrical

approach. A similar proof can be shown for when the sensors are non-linear.


15

Correlating Electronic Nose and Sensory Panel Data

Robert W. Sneath, Krishna C. Persaud

Analytical methods such as gas chromatography-mass spectrometry (GC-MS), or nearinfrared spectroscopy provide the mainstay for measurement of volatile componentsin food, agricultural, chemical, or environmental industries. Although data obtainedgive very precise measurements of individual components in a mixture, they give verypoor indication of the sensory quality perceived by the human nose or tongue. Thecontrol of odor quality within these industries is associated with problems that areunique, because they also rely on human perception and preference for particulartypes of odors or tastes. It is difficult to relate the output of traditional analytical instru-ments to human perception, because the chemosensory systems of smell and taste useinformation gathered from the interaction of complex chemical mixtures with thebiological sensors without separation of individual components. Many such indu-stries therefore rely on human sensory panels that are trained to discriminate subtlenuances of smell and taste in a given product or raw material, or to quantify the odorlevel in a sample. This in itself presents problems because such panels can only copewith relatively few sample assessments per day, and are very costly to run. Theymay beused for optimization of a new product, periodic sampling of problematic systems, andrandom quality control. This highlights the need for automated chemical sensing sy-stems that produce data that are easily correlated to human odor perception.The human nose contains a large array of chemical sensors, and patterns of infor-

mation are processed in the olfactory brain of an animal in order to achieve quanti-fication and discrimination of odors based on previous learning experiences. Withinstrumental means of odor measurement, the human user interface needs to beconsidered very carefully, as the results need to be presented in a form that can beeasily interpreted by the user. If an electronic nose is applied, the signals producedby an array of sensors consist of measurements of responses to odors producing dif-ferent patterns that are projected into multidimensional space. In many instances weare dealing with complex mixtures of compounds in which only relatively few com-ponents (which may be at very low concentrations relative to other components) areimportant in the determination of odor quality by a human sensory panel [1, 2]. Ol-factory data depend strongly on individual physiological differences, on measurementmethods, and on psychological factors. Classifications of odors are necessary to put


377377

some order in odor descriptions that are used in structure-odor relationships. Pub-lished classifications have been based on empirical, semi-empirical, or statistical ap-proaches. In the last category, data may be obtained using semantic descriptions orprofiles, or similarity estimations. The intensity data are perceived as the strength of astimulus. They also present a huge variability, whichmakes it difficult to relate them tophysicochemical properties.

15.2

Sensory Panel Methods

Correlation of human sensory panel responses with data from electronic nose instru-mentation demands that both sets of information have good repeatability and accu-racy, which are usually accounted for by frequent calibration against known standards.Novel methods sometimes need to be developed to calibrate a sensor array. Compli-ance of sensory panel data to accuracy and repeatability standards is often neglected.Unless this feature of data collection is attended to, correlations are likely to be poor.There are few sensory panel standards but one that is relevant for correlations withelectronic noses’ is the European standard EN13725 [3]. However, although it only setscriteria for detection threshold measurements, it has many features that can be takenon board when measurements of the other dimensions of odor require standardiza-tion.

15.2.1

Odor Perception

Sensory perception of odors has four major dimensions: detectability, intensity, qual-ity, and hedonic tone, and problems arise when we want to assign values to perception.

1) Detectability. There is no conscious subjectivity to this dimension: either the smellis detected or it is not, but every person will have their own detection threshold,which will vary in people depending on their own situation at the time.

2) Intensity, which refers to the perceived strength of the odor sensation, and the odorhas to be at a supra-threshold level.

3) Quality, i.e. what the substance smells like; assessors usually work from an agreedlist of descriptors.

4) Hedonic tone. This is a category judgement of the relative pleasantness or unplea-santness of the odor, which is a very personal description and can only have anyobjectivity assigned to it if a comparison is made with other odors.

Perceptions are qualitative, and will of course vary from person to person so we have todevise ways of standardizing the descriptions of the odor and/or standardizing thepeople who make the assessments. In all odor or taste-related industries this is com-mon practice, selected and trained staff are used when blending teas, coffees, and

15 Correlating Electronic Nose and Sensory Panel Data378

perfumes, and they use a set of agreed descriptors between themselves in an attempt tomake the descriptions objective.

15.2.2

Measurement of Detectability

Detectability is the only one of those dimensions that can be reduced to an objectiveperception. The only answers to the question “Can you detect the odor?” are “Yes” or“No”, although the value of the response depends on the assertiveness and honesty ofthe subject. The threshold of detection is different for each individual and can beaffected by factors such as where the person is, by background odors, or by familiaritywith that odor. Therefore, threshold values are not fixed physiological facts or physicalconstants, but represent the best statistically estimated value from a group of indivi-dual responses.Odor thresholds are estimated in one of two ways, by getting a yes/no response, as

above, or by a ‘forced choice’ response where the subject is forced to choose which airstream, from two or more, smells. In the former classical evaluation, yes/no answersare, amongst other factors, dependent on the subjects’ honesty and motivation. Ifodors at a range of concentrations, alternating with blanks, are presented a sufficientlylarge number of times, yes/no answers may be evaluated with the aid of signal detec-tion theory, to eliminate the effects of context.The forced-choice procedure is an attempt to measure a subject’s sensitivity, which

is not influenced by fluctuations in criteria. Two or more choices are presented to thesubject at a range of odorant concentrations, and it is the subject’s task to choose theone that is odorous from the other that is not. The assumption is made that the ob-server chooses the one that gives the largest sensory excitation, provided that there isno response bias towards one or more of the options. If the comparison stimuli(blanks) have been carefully defined and controlled, the proportion of correct re-sponses can be used as a measure of sensitivity, because it will always be measuredin comparison to blanks.

15.2.3

Transforming the Measurement of the Subject to the Subject’s Measurement of an Odor

The detection threshold value is a measure of the sensitivity of the assessor, but whatwe need to do is to measure, in a reliable way, the odor we are interested in.In all measurements, two criteria must be satisfied: accuracy and repeatability. This

usually means manufacturing a sensor that produces the correct answer and will pro-duce the same answer repeatedly. In olfactometry our sensor is the human nose. Thesesensors have been produced in a manufacturing process that has no quality control:therefore, from the production run, we must choose sensors that fit our criteria foraccuracy and repeatability. The machine that presents the odor sample to the sensorsmust also be constructed and operated to achieve the criteria of accuracy and repeat-ability.


15.2.4

Assessor Selection

The key part of accurate odor measurement is the selection of the odor assessors. Inorder to select odor assessors, n-butanol has been specified in EN13725 as the refer-ence material. Although it is recognized that a single component reference gas is notthe ideal, no representative odorant mixture has yet been formulated. Only people witha mean personal threshold for n-butanol in neutral gas of between 20 ppb and 80 ppband a log standard deviation of less than 2.3, calculated from the last 10 to 20 individualthreshold estimates (ITEs), are acceptable. These assessors are continually checked fortheir detection threshold (at a minimum after every 12 odor measurements) and haveto remain within these limits to be a panel member.This selection criteria used at the Silsoe Research Institute (UK) laboratory leads to

the rejection of about 43% of those tested because they are not sensitive enough and12% because they are too sensitive to n-butanol. The complete distribution of sensi-tivities of all 164 people tested in the Silsoe Research Institute laboratory, to date, isillustrated in Fig. 15.1. The butanol thresholds are grouped into 0.3 log intervals, i.e.less than 1.0, 1.0 to 1.3, 1.3 to 1.6, etc. plotted as a linear scale on the y-axis of Fig. 15.1.Of those who have a qualifying sensitivity, about two thirds have a threshold above theaccepted reference value of 40 ppb (log 1.6). Selection of the panel members using theabove method will lead to acceptable accuracy and precision and enable a laboratory tocomply with the criteria set in EN13725 (Section 15.2.1).

15.2.5

Types of Dynamic Dilution Olfactometry

15.2.5.1 Choice Modes

Two different choice modes can be used to obtain an individual threshold estimate.These choice modes and their requirements are described here. They all produce the

Fig. 15.1 Distribution of n-bu-

tanol olfactory thresholds for

164 subjects. The histogram

highlights subjects who would

qualify for sensory panel mea-

surements


common result of an ITE. The use of the ITE derived from either of these methods inthe calculation of an odor concentration is then identical throughout this standard.

15.2.5.2 Yes/No Mode

In the yes/no olfactometer; (Fig. 15.2) either neutral gas or diluted odor passes fromthe single port. The panel member is asked to evaluate gas presented from the singleport and to indicate if an odor is perceived (yes/no). The panel members are aware thatin some cases blanks (only neutral gas) will be presented. (A second port always pre-senting neutral gas may be made available to the assessor to provide a reference.) Thesamples may be presented to the assessors either randomly or in order of increasingconcentration. When using the yes/no mode, 20% of the presentations in a set ofdilution series must be blanks to satisfy the operator that the panel members aregiving the correct response when there is no odor present. For each panel memberthe measurement must include a dilution step at which they respond ‘No’ to a dilutedodor and for two adjacent dilutions they must respond, ‘Yes’.

15.2.5.3 The Forced Choice Mode

A forced choice olfactometer (Fig. 15.3) has two or three outlet ports, from one ofwhich the diluted odor flows, while clean odor-free air flows from the other(s). Inthis method, panel members assess the ports of the olfactometer, from one of whichthe diluted odor flows, neutral gas flows from the other port(s).Themeasurement starts with a dilution of the sample large enough tomake the odor

concentration beyond the panel members’ thresholds. The concentration is increasedby an equal factor in each successive presentation: this factor may be between 1.4 and2.4. The port carrying the odorous flow is chosen randomly by the control sequence oneach presentation. The assessors indicate from which of the ports the diluted odorsample is flowing, using a personal keyboard. They also indicate whether their choice

Fig. 15.2 Schematic diagram of a ‘Yes/No’ olfactometer. When the presentations are

sorted in order of ascending concentration, the geometric mean of the dilution factors of

the last FALSE and the first of at least two TRUE presentations determines the individual

threshold estimate (ITE) for a panel member. The odor concentration for a sample is

calculated from the geometric mean of at least two ITEs for each panel member


was a guess, whether they had an inkling, or whether they were certain they chose thecorrect port. Only when the correct port is chosen and the panel member is certain thattheir choice was correct is it taken as a TRUE response. At least two consecutive TRUEresponses must be obtained for each panel member. The geometric mean of the dilu-tion factors of the last FALSE and the first of at least two TRUE presentations deter-mines the ITE for a panel member. The odor concentration for a sample is calculatedfrom the geometric mean of at least two ITEs for each panel member.The odor concentration has units of ouE m

�3 (European odor units per cubic meter).For measurements on reference odorants, this value can be converted to a detection

threshold, expressed as a mass concentration using the known concentration of thereference gas divided by the ITE.

15.2.5.4 Laboratory Conditions

For laboratories to conform to the required standard, they must be guaranteed to befree from odor. They are usually air-conditioned with activated charcoal filtration. Theymust also have a source of odor-free air, i.e. neutral gas, with which to dilute the odorsample. The olfactometer, which is a dilution device, is made entirely from approvedmaterials, glass, tetrafluoroethylene hexafluoropropylene copolymer, or stainless steel.Samples are processed within 30 hours of collection.

15.2.5.5 Laboratory Performance Quality Criteria

The EN13725 is based on the following accepted reference value, which shall be usedwhen assessing trueness and precision:

Fig. 15.3 Schematic diagram of a forced choice olfactometer.

Panel members assess the ports of the olfactometer, from one of which

the diluted odor flows, neutral gas flows from the other port(s).

The port carrying the odorous flow is chosen randomly by the control

sequence on each presentation


1 ouE � 1 EROM (European reference odor mass) ¼ 123 lg n-butanol

When 123 lg n-butanol is evaporated in 1 m3 of neutral gas at standard conditions(20 8C) for olfactometry the concentration is 0.040 lmol mol�1 (40 ppb). Two qualitycriteria, as below, are specified to measure the performance of the laboratory in termsof the standard accuracy and precision, respectively. Accuracy reflects the trueness orcloseness to the correct value, in this case the true value for the reference material is40 ppb and the precision is the random error. The standard specifies how these twoquality criteria are calculated [3]. The criterion for accuracy Aod (accuracy of the odormeasurement) i.e. closeness to the accepted reference value is:

Aod � 0:217

In addition to the overall accuracy criterion, the precision, expressed as repeatability, r,should comply with

r � 0:477

This criterion for repeatability can also be expressed as:

10r � 3:0

This repeatability requirement implies that the factor that expresses the differencebetween two consecutive single measurements, performed on the same testing mate-rial in one laboratory will not be larger than a factor of 3 in 95% of cases.

15.2.5.6 Compliance with the Quality Criteria

The performance of an olfactmetry laboratory is monitored continuously by checkingthe accuracy and repeatability of the results of measurements of n-butanol. Fig-ures 15.1, 15.4 and 15.5 illustrate this over the first five months of the year 2000 atthe Silsoe Research Institute laboratory. Each point on the graphs is the result ofthe previous 20 panel threshold n-butanol measurements. The panel thresholdsare shown in Fig. 15.4. This shows the accuracy to be slightly biased to the highside of the accepted reference value of 1.6. This is explained by reference toFig. 15.1, the distribution of threshold values. To date, panel members are selectedrandomly from our list of qualified assessors, thus the panel is biased towards thehigher n-butanol threshold. Closer agreement with the accepted reference valuecan be achieved by selecting panel members more rigorously. In Fig. 15.5 the recordof accuracy and repeatability criteria over the same period shows that the laboratoryexceeded the quality criteria of the standard (accuracy criterion shown as &, and re-peatability criterion shown as ~).


15.2.6

Assessment of Odor Intensity

The second dimension of the sensory perception of odors, intensity, refers to the per-ceived strengths of the odor sensation. Intensity increases as a function of concentra-tion. The dependence may be described as a theoretically derived logarithmic functionaccording to Fechner [4]:

S ¼ kW � log I=I0 ; ð15:1Þ

where:S ¼ perceived intensity of sensation (theoretically determined)I ¼ physical intensity (odor concentration)

Fig. 15.4 Five-month history of average panel threshold at the Silsoe Research Institute Laboratory

Fig. 15.5 The accuracy and repeatability of the daily measurements of n-butanol with the chosen sensory

panel


I0 ¼ threshold concentrationkW ¼Weber-Fechner coefficient.Stevens [5] suggests a power relationship should be applied:

S ¼ k � In; ð1Þ

where:S ¼ perceived intensity of sensation (empirically determined)I ¼ physical intensity (odor concentration)n ¼ Stevens’ exponentk ¼ a constant.

Which one of these two descriptions applies depends on the method used. To date, notheory has been able to derive the psychophysical relationship from knowledge aboutthe absolute odor threshold of various substances [6].Odor intensity is measured using this category estimation technique. After deter-

mining the odor concentration of the samples, a range of suprathreshold dilutions ispresented in random order to panel members. They are required to indicate theirperception of intensity at each dilution according to the scale shown in Table 15.1.Intensity scores are obtained from each panel member at each of 12 presentations of

suprathreshold dilutions and the average score for each presentation plotted againstlog10concentration. A linear regression is performed on intensity vs.log10concentration and the line of best fit plotted on the graph.Examples of two such measurements are shown in Figs. 15.6 and 15.7. The fresh

landfill material has an intensity of 2.5 (faint to distinct odor) at 0.5 log10ouE·m�3,

(3.2 ouE·m�3), whereas at the same odor concentration the stale landfill gas has

an intensity of only 1.5 (very faint to faint odor). This means that at the same odorconcentration the odor from fresh landfill material will be perceived to be the strongerodor.If these data had been obtained from an odor source for which an odor-abatement

plant needs to be designed, then it could be that the intensity of a ‘faint odor’, at acomplainant’s premises, was considered as the unacceptable limit. In that case theoutlet concentration from the abatement equipment would have to be designed soas to deliver an odor with a concentration of less than 2 ouE·m

�3 (fresh landfillmaterial) or 6 ouE·m

�3 (stale landfill gas), respectively, to the nearest complainant.

Table 15.1 Scaling of odor intensity by a human sensory panel

0 No odor;

1 Very faint odor;

2 Faint odor;

3 Distinct odor;

4 Strong odor;

5 Very strong;

6 Extremely strong odor


15.2.7

Assessment of Odor Quality

Some useful information about the characteristics of an odor can be obtained if qualityassessments are made at a range of dilution ratios close to the panel detection thresh-old, although these are not included in the standard.

Fig. 15.6 Plot of odor intensity versus odor concentration for volatiles from fresh landfill material

Fig. 15.7 Plot of odor intensity versus odor concentration for volatiles from stale landfill gas


One assessment we often carry out is a description of the odor. Our odor panelmembers are asked to smell the odor at a dilution ratio of between 12 and 100and indicate, from a choice of descriptors, which comes closest to their perceptionof the odor. Typically the panel is asked if the odor sample smells like: sewage,fish, rotten cabbage, rotten eggs, bleach, earthy, compost, tarry, smoky, or other.This method is useful for diagnosing if a piece of abatement equipment is changingthe odor as well as reducing the concentration.For a food or beverage application such as wines, the requirements for the descrip-

tive terms have to be specific and analytical and not be hedonic or the result of anintegrated or judgmental response. Floral is a general but analytical descriptiveterm, whereas fragrant, elegant or harmonious are either imprecise and vague (fra-grant) or hedonic, and judgmental [7, 8], and often an ‘odor wheel’ containing a seriesof descriptive terms is used to guide the human panel. Each application presents itsown specific problems, and appropriate descriptors need to be devised and standar-dized. For Scotch whisky production for example, the key characteristics arising dur-ing production are: estery (the fruity, fragrant, pear-drops aromas that characterizecertain malts particularly), phenolic (from woodsmoke to tar, iodine to sea-weed –typified by some malts), aldehydic (leafy, grassy scents, sometimes like Parma vio-lets, often found in various types of malts ) and feinty. The aromas associated withfeints are not pleasant – they are notes of sweat, vomit, and rotten fruit – but theygive Scotch whisky its character and are essential to the overall flavor. They are presentto a greater or lesser extent in all malts [9]. Similar odor descriptor wheels are availablefor beers, coffee, tea, andmany other commodities. The data from such an assessmentis usually presented as a histogram of the panels’ response.

15.2.8

Judgment of Hedonic Tone

Hedonic tone is a judgement of the un/pleasantness of the odor. In a similar way to theassessment of the intensity, the panel members are asked to score their perception ofthe odor on a scale from 1 to 5 at a range of odor concentrations above the odor thresh-old. A graph similar to the intensity graph can be plotted.

15.3

Applications of Electronic Noses for Correlating Sensory Data

Using an array of sensors together with appropriate data processing may allow map-ping of sensory panel attributes to electronic nose data. Multivariate analysis was ap-plied to electronic nose data to correlate sensory panel data for marjoram assessment[10]. Discriminant analysis and neuro-fuzzy treatment of electronic nose and/or colormeasurement data of marjoram were applied. The aim was to investigate if the judge-ments of a sensory panel regarding taste, smell, and appearance or genetically deter-mined differences of marjoram samples can be predicted. Frank et al. [11], in co-op-

15.3 Applications of Electronic Noses for Correlating Sensory Data 387387

eration with packaging material suppliers and a food manufacturer investigated thequality of different kinds of wrapping foils for chocolate bars using a hybrid modularsensor system (MOSES II). A GC-MS unit connected to a headspace-sampler was usedas an analytical reference. A human sensory panel using a sniff-test also qualified allanalyzed samples. The different packaging material species could be distinguished ina principal component analysis (PCA). With the aid of a principal component regres-sion (PCR) a correlation between human and technical odor perception was carriedout, to determine the spoilage of fish [12], storage of chicken [13], evaluation of tomatoquality [14], dairy products [15], and correlation of malodors from sewage [16]. Othersensory attributes may be equally important. For example Benedito et al. [17] inves-tigated methods of improving Mahon cheese texture assessment, where the relation-ship between instrumental and sensory measurements was sought. For that purpose,30 pieces of Mahon cheese from different batches and 2 different manufacturers wereexamined. Textural characteristics at different curing times were evaluated by uniaxialcompression, puncture, and sensory analysis. Significant linear correlations werefound between instrumental and sensory measurements. A logarithmic model (We-ber-Fechner) fitted data better than a linear one. Pearce and Garner [18, 19] describe anovel method for predicting the organoleptic scores of complex odors using an array ofnon-specific chemosensors. The application of this method to characterizing beer fla-vor was demonstrated, which predicted a single organoleptic score as defined underthe joint European Brewing Companies/American Association of Brewing Chemists/Master Brewing Association of the Americas (EBC/ASBC/MBAA) international flavorwheel for beer.

15.4

Algorithms for Correlating Sensor Array Data with Sensory Panels

One problem that needs to be solved is to map responses from a sensor array to ana-lytes of various concentrations (or mixtures) to psychophysical measurements from ahuman sensory panel, so as to correlate parameters such as quality (in terms of adescriptor) or intensity (in terms of a nonlinear scale).In the ideal situation, we have knowledge of the physical processes underlying the

relationship between sensor responses and human panel responses. A theoretical for-mula can then be used to calculate somemeaningful number from the input variables.Usually we do not have this sort of information to hand, however, we can see that thereis a relationship there. This is where calibration becomes important. Instead of tryingto calculate the theoretical relationship between input and output variables, we makesimple assumptions as to the underlying relationship. Using some given examples ofinput and output variables, we then try to estimate the parameters of this relationship.We can take as X variables the quantities we wish to measure using an electronic nose.Typically, these aremore convenient tomeasure than the values we wish tomodel. TheY variables are the quantities we wish to predict. These will be the values estimatedfrom the X values using the model. Together with measurement there is some var-iance, which is a measure of the spread of a variable about its average value. In multi-


dimensional data from sensor arrays the covariance is also important. This is a mea-sure of the similarity of two variables. Variables having high covariance are stronglyrelated to each other. To know the strength of this relationship, we also need to knowthe variance of the individual variables. For multidimensional data, matrices becomeimportant representations of data.

15.4.1

Multidimensional Scaling

Multidimensional scaling (MDS) encompasses a collection of methods that allow us togain insight in the underlying structure of relations between entities by providing ageometrical representation of these relations [20]. MDS has its roots in two importanttraditions within psychology. The first is in psychophysics and the other in psycho-metrics. These methods belong to the more general category of methods for multi-variate data analysis. MDS can be characterized by the generality of the type of ob-served relations, which can be submitted to the data analysis on the one hand, andby the specificity of the type of geometrical representation of these relations on theother hand. Whatever kind of relation between a pair of entities that can be translatedinto a proximity measure, or conversely into a dissimilarity measure, can be consid-ered as possible input forMDS. The choice of a particular type of spatial representationcan be considered to be the most important part of the modeling which goes togetherwith the application of a specific MDS-algorithm on the set of proximities. Young andHouseholder [21] wanted to extend the methodology of unidimensional scaling ofperceptual characteristics of stimuli to the simultaneous scaling of several character-istics. Guttman [22] was interested in a less restrictive model than the factor analyticmodel to represent the relations between several assessment variables. This wouldallow for a much more systematic way to formulate hypotheses on the underlyingstructure for assessment variables. The psychophysical approach led to algorithmicdevelopments, which soon came to be known as MDS, while the psychometric ap-proach preferred to label its own production of algorithms under the heading of ‘smal-lest space analysis’.We can use the symbol pij to refer to the proximity measure between entities i and j.

If a subject has to indicate the perceived dissimilarity between two odors on a ratingscale (0 for ‘no difference’ and 10 for ‘maximal difference’), then this rating can beconsidered to be a reversed measure of the proximity between the two odor stimuli. Ora correlation coefficient between variables i and j can be considered to be a proximitymeasure for these two variables. The proximities are then represented in a geometricalspace, e.g. in a Euclidean space. The distance d between two vectors u and x in a j-dimensional space is given by the formula:

d ¼X

j

juj � xjjl

!1=l

ð3Þ

where l ¼ 2 for the Euclidean distance measure commonly used.


Threemethods of analysis are closely related toMDS: PCA, correspondence analysisand cluster analyis (CA). These are described in detail in Chapter 6.

15.4.2

Regression Methods

Univariate linear regression may be used for establishing a correlation. In its simplestform, this will be familiar as finding the line of best fit through a cloud of points. Weassume that the relationship between a single X variable and one Y variable is linear,i.e.

Y ¼ bX þ a ð2Þ

where b is the slope of the line, and a is the intercept at the Y axis.Univariate linear regression estimates the values of b and a by minimizing the sum

of squared vertical distances from points to the line. We choose a candidate slope, band intercept, a. For each recorded (X, Y) pair, we square Y – bX – a and add it to thetotal. The line having the smallest total is the best-fit line. In practice, calculus gives usa formula for estimating b directly, and thence a, bb ¼ CovðX ; YÞ=VarðXÞ. The indi-cates that the value is an estimate of b. We can ignore a if we center all our variablesbefore using them. To center each variable, we calculate its average value, and thensubtract this value from all sample values. a can be calculated after modeling using theestimated value of b and the subtracted averages.When working with centered data, wecan express the linear regression equation for b in matrix form as ðXTXÞ�1XTY . Notethat if the variance of X is zero, then we cannot estimate b. This occurs when the Xvariable has the same value for all values of Y. The matrix form of the linear regressionalso works for multiple X values, and so in using a multisensor array, the resultingestimate of b is a vector containing the weights applied to the X variables, and this istermed multiple linear regression (MLR). There are many situations when ðXTXÞ�1

cannot be calculated, and so some care has to be taken when using MLR. Note that ifthe number of recorded samples is less than the number of X variables, then collinear-ity (correlated X variables) is guaranteed to occur. In this situation, the usual solution isto discard variables. The process of selecting variables for MLR is known as stepwiseMLR.Because of difficulties in carrying out MLR without prior inspection of the data,

methods of visualizing structure in multidimensional data have to be used. PCA pro-vides a method for finding structure in such data sets (See Chapter 6). This methodrotates the data into a new set of axes, such that the first few axes reflect most of thevariations within the data. By plotting the data on these axes, we can spot major under-lying structures automatically. The value of each point, when rotated to a given axis, iscalled the principal component value.Correspondence analysis is classically used with the aim to visualize the relations

(i.e. deviations from statistical independence) between the row and column categories.The unfolding models do the same: subjects (row categories) and objects (columncategories) are visualized in a way that the order of the distances between a sub-


ject-point and the object-point reflects the preference ranking of the subject. The mea-sure of ‘proximity’ used in correspondence analysis is the chi-square distance betweenthe profiles.Cluster analysis models are equally applicable to proximity data. The main differ-

ence with the MDS models is that most models for cluster analysis lead to a hierarch-ical structure. Path distances under a number of restrictions approach the dissimila-rities. The path distances are looked for in a way that minimizes the sum of squarederrors.

15.4.3

Principal Components Regression

PCA selects a new set of axes for the data. These are selected in decreasing order ofvariance within the data. They are also perpendicular to each other so that the principalcomponents are uncorrelated. Some components may be constant, but these will beamong the last selected. The problem with MLR is that correlated variables cause in-stability. So the strategy adopted is to calculate principal components, throwing awaythe ones that only appear to contribute noise (or constants), and using MLR on these:this process is known as PCR. Rather than forming a single model, as we did withMLR, we can now form models using more than one component, and decide howmany are optimal. If the original variables contained collinearity, then some of ourcomponents will contribute only noise. So long as we drop these, we can guaranteethat our models will be stable. This method is commonly used to correlate instrumen-tal analyses with human sensory panel data.

15.4.4

Partial Least Squares Regression

The intention in using PCR was to extract the underlying effects in the X data, and touse these to predict the Y values. In this way, we could guarantee that only independenteffects were used, and that low-variance noise effects were excluded. This improvedthe quality of themodel significantly. However, PCR still has a problem. If the relevantunderlying effects are small in comparison with some irrelevant ones, then they maynot appear among the first few principal components. So, we are still left with a com-ponent selection problem – we cannot just include the first n principal components, asthese may serve to degrade the performance of the model. Instead, we have to extractall components, and determine whether adding each one of these improves the model.This is a complex problem that may be solved using partial least squares regression(PLSR). The algorithm used examines both X and Y data, and extracts components(now called factors) that are directly relevant to both sets of variables. These are ex-tracted in decreasing order of relevance. So, to form a model now, all we have todo is extract the correct number of factors to model relevant underlying effects. Acombination of MLR, PLS, factor analysis, and PCR are often used [10, 11, 14, 23–27].


In data from sensor arrays there are often underlying effects. In multivariate cali-bration, these are called latent variables. A latent variable is one that we do not observedirectly, but we can infer its existence by the properties of our observed variables. Wecan view latent variables in several ways: Assuming that all relationships betweenlatent and observed variables are linear, we can use PCA (if we assume that onlythe X variables are affected by the latent variables), or PLSR (assuming that both Xand Y are affected). If the relationships are thought to be nonlinear, then PCA andPLSR are not ideal, since these assume linearity. If we have an idea of the mathema-tical form of the nonlinearity, we can try transforming the X and Y variables to linearizethem. Failing that, we can use artificial neural networks (ANNs), which use a latentvariable model that does not assume linearity.

15.4.5

Neural Networks

A parametric regression model usually refers to the regression model where the formof the functional relationship is known (e.g. the linear regression or a specified poly-nomial regression). Nonparametric regression does not need to specify the form of theunknown functional relationship. The function is modeled using an equation contain-ing unknown parameters but in a way that allows the class of functions that the modelcan represent to be very broad. Typically the equation, in some functional form, hasmany unknown parameters, and none of the parameters have any physical meaning inrelation to the problem to be solved. Neural networks, including multilayer percep-trons and radial basis function (RBF) networks are nonparametric regression modelsand these have been described in Chapter 6.Various ANN algorithms can be used to discriminate gases and odors, but themulti-

layer perceptron network has been adapted for various industrial applications fromamong many models of ANNs. The development of a learning algorithm, calledback-propagation, by Rumelhart et al. [28] revolutionized pattern recognition metho-dology. An example of the use of neural networks for classification is given by Stetter[29] who used a sixteen-element electrochemical sensor array to identify differentgrades of wheat, and reported an excellent identification accuracy using the multilayerneural network.For mapping sensor array responses to human sensory panel responses the generic

interpolation problem must be solved. The RBF method solves the interpolation pro-blem by constructing a set of linear equations of basis functions [31]. The RBF networkmakes a linear function space that depends on the positions of pattern vectors accord-ing to an arbitrary distance measure. RBF networks can be combined with fuzzy algo-rithms for enhanced effectiveness in array sensing applications [30].

15.4.6

Fuzzy-Based Data Analysis

Fuzzy set theory was introduced in Chapter 6. There are many areas of uncertainty insensor systems, and fuzzy set theory offers opportunities in many aspects of signal


processing. These include the evaluation of noisy signals, automatic fault diagnosis,the use of indirect measured values to measure process variables, the automation ofmeasurement and evaluation procedures based on expert knowledge, and the fusion ofsensor information in a multisensor environment.The latter application is of importance in the mapping of multisensor data to a hu-

man sensory scale. The data provided by the sensors may contain information fromseveral variables, or the information from several sensors is used to provide measure-ment of a single variable. One way of approaching this is to fuzzify sensor data fromeach sensor in the array i.e. the numerical value is transformed into a linguistic vari-able. The results of this step are analyzed by a fuzzy rule base that describes the variousrelationships between the possible sensor array outputs. The possible outcomes of thefuzzy analysis are then combined and defuzzified to produce the crisp measurementvalues. This method applied to an odormeasurement scenario allows both the ‘quality’and the ‘intensity’ of the odor to be mapped to sensor responses.

15.5

Correlations of Electronic Nose Data with Sensory Panel Data

At Silsoe Research Institute we use our odor panel selected and monitored as requiredby the EN13725 when we need to correlate electronic nose responses with humansensory perception. The results of one example of this technique is discussed below.

Fig. 15.8 Sensory panel evaluation of grain from 1998 harvest. Classes are divided into good and bad, a

grade mark of 2.5 being the threshold


15.5.1

Data from Mouldy Grain

An odor panel selected according to EN13725 was trained at Silsoe Research Instituteto evaluate commercial grain samples from 1998. The samples were presented threetimes in random order and graded into very good, good, bad, and very bad classes, andthe panel were asked to mark the samples 1–4 respectively. Grain samples of onevariety from the 1998 harvest were used as the training set for the electronic nosethis data and the odor panel assessments were the input data for the RBF networkdescribed in [31]. Selections of 13 varieties, of wheat from the 1999 harvest were pre-sented as the unknowns.The odor panel classification of the 1998 grain is shown in Fig. 15.8 with classes

divided as good and bad, a grade mark of 2.5 being the threshold. The neural net-work, trained with these grain samples and classes was then used on-line with theelectronic nose to classify the 1999 harvest grain.In some instances, discrimination between good and bad grain types has merely

been as a result of different moisture content of the grain samples. In our workwe could show this is not the case, as illustrated by Fig. 15.9. Of the 50 samples ana-lyzed, the system correctly classified 38, the remainder were not graded by the system

Fig. 15.9 Classification of grain by an RBFnetwork. Of the 50 samples

analyzed, the system correctly classified 38, the remainder were not

graded by the system as bad or good. In the majority of cases this

corresponded with an intermediate rating of the grain i.e. somewhere

between good and bad. Each bar G1/8–G5/8 represents good grain

samples and B1/8 represent bad samples. The error bars are the

standard deviations of repeated presentations to the panellists.


as bad or good. In the majority of cases this corresponded with an intermediate ratingof the grain i.e. somewhere between good and bad.A PCA plot of the data, Fig. 15.10, shows that the good and bad grain fall into distinct

groups although the data could be considered as part of one elliptical cluster withsubcategories within the cluster with opposite ends representing the best and worstgrain samples. The training data for the RBF network was collected with the grainanalysis sensor prototype (GASP) three weeks prior to classifying the unknown sam-ples, indicating that sensor drift had minimal affect on the result. The PCA plot ismerely a representation of the data for visualization purposes, to give an indicationof the ‘sense’ of the data.In this instance the simplistic view of grain as being either good or bad somewhat

limits the data and forces the decisionmaking into arbitrary choices. The benefit of theRBF network is that it produces an intermediate result. Further work with an enlargeddata set should enable the grain to be re-evaluated against a more robust classificationsystem such as good, intermediate good, intermediate bad and bad, because the grain(as can be seen from the PCA plot) does not instantaneously transform from good tobad but follows a gradual transition from an ‘optimal’ good state through an intermedi-ate stage and on to bad. However, the initial panel-evaluated data set was not largeenough to give a reliable enough training set to produce sub-classified data againstwhich real grain samples could be evaluated.Further development will investigate the robustness of the system over long periods

of use and across a range of different grain samples. Grain-quality classification intomore groups would be a welcome improvement. The principal drawbacks of enlargingthe number of grain classifications are the enlarged training set required, obtainingreliably classified training examples, and the time involved in acquiring the data beforereal samples can be run.

Fig. 15.10 A PCA plot of the training and test data for the neural

network. It shows that the good and bad grain fall into distinct groups

although the data could be considered as part of one elliptical cluster

with subcategories within the cluster with opposite ends representing

the best and worst grain samples. The training data for the RBFnetwork

was collected with the GASP three weeks before classifying the un-

known samples, which indicated that sensor drift hadminimal affect on

the result


Developments of the system to enable characterization of other grain contaminantsof interest such as invertebrates are planned. A dedicated system capable of determin-ing the quality of wheat at points of transfer has been developed. The dedicated natureof the system has enabled amore robust and user-friendly system to be developed. Thedesign of the instrument has ensured that classification of large samples can be re-peatable. Important factors in the design are temperature and humidity control, con-sistent presentation of the sample, sensor cleaning and a neural network that is robustand quickly trained.

15.6

Conclusions

In many sensory panel measurements arbitrary scales are used. This makes it difficultto correlate instrumental data, unless some standard can be utilized in both sensorypanel measurements and the instrument. Odor measurements no longer need be thearbitrary assessment they have often been perceived to be. Olfactometry to the CENdraft standard, EN13725, ensures a measurable accuracy criterion for the laboratory,and ensures reproducibility of results between laboratories.Once an odor concentration measurement has been made on a sample, then the

other three dimensions of odor can be investigated systematically. Measurementsof odor intensity can give useful indications of the amount of abatement required,especially when combined with an assessment of hedonic tone.A variety of multivariate analysis methods are applicable to electronic nose data,

provided that the questions are clearly defined. The use of neural networks providesa powerful tool when the parameters defining complex relationships between sensorresponses and human responses are not well understood.As can be seen from the results, the GASP system is capable of classifying grain at a

level equivalent to a trained odor panel, with the implementation of a RBF network.The classifications are independent of grain moisture content. The RBF network isinterpolative and allows both qualitative as well as quantitative mapping of sensorarray outputs to odor descriptors, intensity, or other sensory parameters.The combination of sensor arrays with multivariate algorithms for mapping com-

plex relationships opens a new route for measuring a percept rather than individualcomponents in a mixture. Incorporation of array sensing technology, signal proces-sing, and computation to produce integrated, low-cost measurement devices is onthe horizon, and this will make them increasingly useful in quality control applica-tions in a large number of industries. Thus, industries that rely on human perceptionand preference for particular types of odors or tastes will now have access to instru-mental measurement and control of odor.


References

1 H. Guth, W. Grosch. Journal of Agricultureand Food Chemistry 1994, 42, 2852–2866.

2 P. Semmelroch, W. Grosch. Journal ofAgriculture and Food Chemistry 1996, 44,537–543.

3 ‘Air quality – Determination of odorconcentration measurement by dynamicolfactometry’; Draft prEN 13725; EuropeanCommittee for Standardization, editor,CEN: Brussels, 1999.

4 G. T. Fechner. Elemente der PsychophysikBreitkopf and Hartel: Leipsig, 1860.

5 S. S. Stevens. Psychological Review 1957, 64,153–181.

6 R. L. Doty. Perceptual and Motor Skills 1997,85(3), 1439–1449.

7 A. C. Noble. Abstracts of Papers of the Ame-rican Chemical Society 1998, 216, 130-AGFD.

8 C. D. Owens, P. Schlich, K. Wada, A. C.Noble. Olfaction and Taste Xii 1998, 855854–859.

9 M. MacLean. Pocket Whisky Book; ReedInternationalBooks Ltd.: 1995.

10 M. Hirschfelder, A. Forster, S. Kuhne,J. Langbehn, W. Junghanns, F. Pank,D. Hanrieder. Sensors and ActuatorsB-Chemical 2000, 69(3), 404–409.

11 M. Frank, H. Ulmer, J. Ruiz, P. Visani,U. Weimar. Analytica Chimica Acta 2001,431(1), 11–29.

12 G. Olafsdottir, E. Martinsdottir,E. H. Jonsson. Journal of Agricultural andFood Chemistry 1997, 45(7), 2654–2659.

13 B. Siegmund, W. Pfannhauser.Zeitschriftfur Lebensmittel-Untersuchung Und-ForschungA-Food Research and Technology 1999,208(5–6), 336–341.

14 F. Sinesio, C. Di Natale, G. B. Quaglia,F. M. Bucarelli, E. Moneta, A. Macagnano,R. Paolesse, A. D’Amico. Journal of theScience of Food and Agriculture 2000, 80(1),63–71.

15 F. R. Visser, M. Taylor. Journal of SensoryStudies 1998, 13(1), 95–120.

16 R. M. Stuetz, G. Engin, R. A. Fenner. WaterScience and Technology 1998, 38(3), 331–335.

17 J. Benedito, R. Gonzalez, C. Rossello,A. Mulet. Journal of Food Science 2000, 65(7),1170–1174.

18 T. C. Pearce, J. W. Gardner. Analyst 1998,123(10), 2047–2055.

19 T. C. Pearce, J. W. Gardner. Analyst 1998,123(10), 2057–2066.

20 J. B. Kruskal, M. Wish. MultidimensionalScaling. Beverly Hills, California: Sage, 1978.

21 G. Young, A. S. Householder. Psychometrika1938, 3, 19–22.

22 L. Guttman. Psychometrika 1968, 33, 469–506.

23 A. HenryBressolette, B. Launay, M. Danzart.Sciences des Aliments 1996, 16(1), 3–22.

24 P. J. Hobbs, T. H.Misselbrook, T. R. Cumby.Journal of Agricultural Engineering Research1999, 72(3), 291–298.

25 J. E. Parker, G. M. E. Hassell, D. S. Mottram,R. C. E. Guy. Journal of Agricultural and FoodChemistry 2000, 48(8), 3497–3506.

26 A. K. Thybo, M. Martens. Journal of TextureStudies 1998, 29(4), 453–468.

27 M. C. Zamora, A. M. Calvino. Journal ofSensory Studies 1996, 11(3), 211–226.

28 D. Rumelhart, G. E. Hinton, R. J. Williams.Nature 1986, 323 533–536.

29 J. Stetter. Chemical sensor array: practicalinsights and examples, in Sensors and sensorysystems for an electronic nose, Gardner, J.;Bartlett, P., editors; Springer-Verlag: Berlin,1992.

30 W. Ping, X. Jun. Sensors and ActuatorsB-Chemical 1996, 37(3), 169–174.

31 P. Evans, K. C. Persaud, A. S. McNeish,R. W. Sneath, N. Hobson, N. Magan. Sensorsand Actuators B 2000, 69(3), 348–358.


16

Machine Olfaction for Mobile Robots

Hiroshi Ishida and Toyosaka Moriizumi

Abstract

Olfaction often plays an important role in orienting behaviors of animals. Famousexamples are ants following pheromone trails marked on the ground and moths track-ing aerial pheromone plumes. Inspired by these olfactory-guided behaviors, roboticsystems that perform chemical trail following and plume tracking have been devel-oped. In this chapter, the achievements so far are reviewed to demonstrate the currentstatus of this new application of chemical sensor technologies.

16.1

Introduction

The development of electronic noses has seen a successful transfer of knowledge frombiological studies to engineering products. The fundamental mechanism of animals’olfaction, i.e., an array of sensors combined with a pattern recognition algorithm, hasbecome a key element in artificial odor sensing systems. There are, however, otherinteresting features of olfaction that can be used as models to build engineering sys-tems.One of those features is the close interaction between olfaction and behavior. Olfac-

tion often plays an important role in orienting behavior, and many species of animalsrely for their survival on this ability. Swimming up or down the gradient of chemicalconcentration is one of the oldest types of behavior and can be even seen in micro-organisms [1]. For some animal species, olfactory cues are far more effective thanvisual or auditory cues in search for objects such as foods and nests [1]. Olfactionis also used for various types of pheromonal communications [2].Inspired by these olfactory-guided behaviors, research has been initiated on the use

of chemical sensor technologies for navigation of robots. There have been two types ofrobotic systems developed so far. One is to follow odor trails marked on the ground; itsbiological model is ants following pheromone trails. The other type of robots trackaerial or underwater plumes of chemicals to find their sources. A wide range of ani-mals from bacteria to insects and mammals show this type of search behavior.


399399

Complete understanding of the two types of behavior has not yet been attained.However, biological studies have been gradually revealing the underlying mechan-isms, and some of them have been successfully transferred to engineering sys-tems. In this chapter, achievements made to realize mobile chemical sensing systemsare reviewed after a brief overview of animal behaviors.

16.2

Olfactory-Guided Behavior of Animals

Among a variety of olfactory-guided behaviors, here we focus on two fundamentalsearch behaviors, plume tracking and trail following. A brief overview of the behaviorsis shown below to give useful insights for designing mobile robots. More detailedinformation can be found in other reviews [1, 3].

16.2.1

Basic Behaviors Found in Small Organisms

Most fundamental forms of olfactory-guided behaviors can be found in microorgan-isms. Some unicellular eukaryotes and human neutrophil leucocytes swim up to thesources of chemical attractants [1]. They are known to detect the polarity of a concen-tration gradient by direct comparison of signal intensities at multiple chemoreceptorsites. If similar direct gradient detection is performed using symmetrically placedolfactory organs such as left and right antennae of insects, it is called tropotaxis [1].Bacteria such as Escherichia coli have a different strategy termed klinokinesis [1].

Since the variation in concentration over their small body length is too small to de-tect, they compare the stimulus intensity over time. If the detected concentration of anattractant is increasing, a bacterium swims straight. Decrease in concentration meansswimming in a wrong direction. However, it can’t tell from the temporal comparisonwhich direction leads to the source. Therefore, a bacterium performs an abrupt turn,and randomly chooses a new direction.Another important class of behavior is klinotaxis in which the concentration gradi-

ent is detected by scanning with a single receptor [1]. This yields amuch straighter pathto the source than klinokinesis [4], but the pathmay be longer than that of tropotaxis bythe length of scanning movements.

16.2.2

Plume Tracking

All three types of behaviors introduced in the previous section lead a searcher to achemical source provided that smooth and stable concentration gradients are estab-lished by molecular diffusion. Although it is true for short-range search in microor-ganisms, motion of fluidmedium (air or water) is almost always more dominant in the

16 Machine Olfaction for Mobile Robots400

scale of engineering interests than slow molecular diffusion. Therefore, we must facemore complex situations.Figure 16.1 shows chemical plumes formed in air and water flows. It is the turbu-

lence of the flows thatmainly determines the distributions of chemicals [5]. A chemicalsubstance released from its source trails in the downstream direction, and a number ofeddies in the turbulent flow stretch and twist the plume. The result is a complicated,patchy meandering plume. There is no spatially smooth gradient of concentration inthese instantaneous images that might guide a searcher to the source. When averagedover several minutes, chemical plumes have continuous concentration gradients.However, this requires too long a time in most of the engineering applications,and is unlikely to be employed by animals.When a stationary sensor is placed in such a chemical plume, a fluctuating signal is

obtained [5, 6]. Isolated sharp peaks of concentration are observed when patches of theplume pass over the sensor. As seen in Fig. 16.1, fundamental characteristics of theplumes are the same for both aerial and underwater plumes. Therefore, the discus-sions on aerial plume tracking can be directly applied to those on underwater plumetracking or vice versa.Animals are able to track the smells of food, mates, nests, etc even in this difficult

situation [1, 3]. One of the most intensively studied animals is a male moth trackingsexual pheromone released from a conspecific female [3, 7]. In contrast to the simplechemotactic behavior of bacteria, the fundamental behavioral strategy of moths is up-wind flight (anemotaxis) triggered by olfactory cues. When a male moth encounters apatch of a pheromone plume, it turns and progresses in the upwind direction as shownin Fig. 16.2. As long as the moth is flying in the plume, repeated “upwind surges”bring the moth closer to the female. When the male accidentally leaves the plumeand the pheromone signal is lost, it starts to fly from side to side across the windwith a gradually broadening scanning area. This behavior is called “casting,” andis effective in relocating the lost plume. Therefore, from the engineering point of

Fig. 16.1 Chemical plumes formed in turbulent flows. (A) Top view of an

aerial plume in a wind tunnel visualized by smoke of TiCl4. (B) Side view of an

underwater plume visualized by a dye in fully developed open channel flow of

20 cm depth (photograph courtesy of Drs. Phil Roberts and Don Webster at

the Georgia Institute of Technology)

16.2 Olfactory-Guided Behavior of Animals 401401

view, moth’s strategy to achieve a reliable search is twofold: (1) the use of the winddirection combined with the olfactory information and (2) the ability to recover fromfailures.Extensive work has been also done to reveal the mechanisms underlying the search

behaviors of marine animals [8, 9]. There is similarity to a certain extent between thebehaviors of terrestrial and marine animals. For example, blue crabs show rheotacticbehavior similar to upwind flight of moths [9]; they crawl upstreamwhen they perceivesmells of food.However, there is one distinctive difference in the number of sensorsused. While a moth uses only a pair of antennae to track a pheromone plume, marineanimals seem to make the best use of their chemical sensors, which are spatially dis-tributed over their bodies. A blue crab has chemoreceptors on its eight legs as well ason a pair of antennules, and recent studies suggest all of them are important in track-ing odor plumes [10].

16.2.3

Trail Following by Ant

Chemical substances are often used to mark trails or territories [4]. A famous exampleis an ant laying a pheromone trail on its way back home from a food source. The basic

Fig. 16.2 Male moths tracking a sexual pheromone plume released

from a female. Male 1 is flying in the plume, and thus repeatedly

encounters patches of the plume. This results in iterated upwind

surges. Each dot indicates the contact with a patch Male 2 shows

casting flight when it has accidentally left the plume. After several

scans, the contact with the plume has been regained and it has

resumed upwind surges

Fig. 16.3 An ant following a pheromone trail marked on the ground.

Concentrations perceived at the left and right antennae are compared

and used to turn back to the trail


mechanism of trail following by ants is tropotaxis [2] (see Fig. 16.3), as described inSection 16.2.1. Experiments showed that ants do not detect the polarity of the trail. It issaid, however, that some animals such as snails and snakes can distinguish one direc-tion from the other [4].

16.3

Sensors and Signal Processing in Mobile Robots

16.3.1

Chemical Sensors

While animals have keen senses for chemical stimuli, sensors for robots with capabil-ities close to those of animals are not yet available. In the case of gas sensors, a com-promise has been made on the rise and decay times. While the response time of ananimal’s chemoreceptor is in the order of 100 ms [3, 6], typical gas sensors need severaltens of seconds before their responses reach the steady state values. Therefore, thelocomotion of robots was slowed down to a few cm/s in most of the studies, suchas [11]. When appropriate filters are used to extract rapid changes in concentrationfrom slow sensor responses, the speed of the robots can be increased a few times[12, 13].Slow sensor response also poses a serious problem in employing a sensor array and

a pattern recognition algorithm for odor discrimination. Chemical sensors on mobilerobots are exposed to fluctuating concentration in plumes. Since steady-state responseis rarely established in this situation, one must use transient sensor response to na-vigate robots. However, the patterns obtained from transient responses are distortedbecause sensors with different selectivities tend to have different response times. Forthis reason, e-nose techniques have not been used for mobile robots except for an arrayof semiconductor gas sensors and a pattern classifier reported by Rozas et al. [14].The robotic systems developed so far are mostly made to test the ideas in laboratory

environments. Therefore, the combinations of target chemicals and sensors were cho-sen mainly from the ease of handling. The most commonly used combination is al-cohol and commercially available tin-oxide gas sensors [11, 13, 15]. QCMs [16, 17] andpolymer-based conductometric sensors [18, 19] have been used with camphor, alcohol,and other odorants. Live insect antennae can be also used as odor sensors since theyyield voltage differences between their tips and bases according to the intensities ofodor stimuli. The measured signal is called EAG (electroantennogram), and robotswith antennae cut off from silkworm moths were reported [20–22].For chemical detection in water, widely used potentiometric sensors, e.g., ion-selec-

tive electrodes and ISFETs, suffer from their slow responses. Amperometric micro-electrode sensors are promising since fast response comparable to chemoreceptors ofanimals can be easily achieved [6, 23]. Conductivity sensors can also be used to detectthe concentration of ionic solution in fresh water [24].

16.3 Sensors and Signal Processing in Mobile Robots 403403

16.3.2

Robot Platforms

Most of the mobile chemosensory systems reported so far have used small wheeledrobots for their platforms. Legged robots have better maneuverability if they are prop-erly controlled to achieve stable gaits. A six-legged robot mimicking trail following ofants has been reported [25]. Some robots, e.g., the Robolobster [24] and the silkwormmoth robot [22], are specifically designed after their model animals. To test the hy-potheses on olfactory-guided behaviors of animals, the robots’ sizes and speeds arematched with those of the model animals. The robots are also equipped with chemicalsensors that have spatial and temporal resolutions comparable to the chemosensoryorgans of the model animals.There are several classes of robot configurations as shown in Fig. 16.4. Simple ro-

botic algorithms can be incorporated into a combination of analog and logic circuitsshown in Fig. 16.4A [19]. To make the robot perform tropotactic behavior, for example,the logic circuit is wired to turn on the right motor when the left sensor detects anodorant and vice versa. The robot then turns towards the stimulus or move straight ifthe both sensors are equally stimulated.An on-board microprocessor shown in Fig. 16.4B can accomplish more complicated

tasks. As described in the previous section, most of the chemical sensors are slowdevices, and because their outputs do not change rapidly, sampling rates of a fewHz are usually sufficient. Therefore, a high-speed processor is not always needed.An 8-bit microprocessor, Motorola 68HC11, with a built-in A/D converter is oftenused to control a small robot [13, 25]. If more computational power is needed, fastermicrocomputer boards are available [24]. The flexibility obtained by using micropro-cessors also enables robots to have sensors of other modalities. The sensors that havebeen incorporated in robots include flow detectors [11, 26, 27] to achieve anemotaxis, agyro to control turning of the robot [24], and a bump sensor for obstacle avoidance [26].Another way to accomplish heavy computation is to use a telemetric robot, as shown

in Fig. 16.4C. Although an on-board circuitry has to be small enough to fit in a smallrobot, a fully equipped PC can be used for signal processing in this configuration.Wireless transmitters and receivers are used for the communication between thePC and the robot [27]. Custom-made ASIC chips such as [28] are also promisingfor signal processing in robotic applications since faster processing can be achievedwith smaller circuits.

16.4.

Trail Following Robots

16.4.1

Odor Trails to Guide Robots

Automated guided vehicles (AGVs) are a class of industrial mobile robots that followmetal wires buried under the floor and convey parts andmaterials [25]. The behavior ofants following odor trails implies that a chemical substance can be used as an inex-


Fig. 16.4 Block diagrams of robotic systems. (A) Simplest form of

robot. Signals from the left and right sensors are conditioned through

the amplifiers and filters. The comparators covert the analog signals

into on-off digital values, and the logic circuit is used to map these

values to motor commands. (B) Robot controlled by an on-board

microprocessor. Signals from sensors are processed in the micro-

processor to yield the motor commands. (C) Telemetric robot. The on-

board microprocessor acquires sensor signals and transmits them to

the PC. After processing the signals, it then sends back motor com-

mands to the on-board processor

16.4 Trail Following Robots 405405

pensive alternative to these wires [18]. Odor trails provide higher flexibility since theyare easier to lay on the floor. The fundamental constraint is, however, odor trails decayover time as the chemical substance gradually evaporates.Russell proposed other scenarios in which odor trails simplify the tasks to be accom-

plished by robots [16, 25, 29]. They include: (1) an area coverage task such as cleaningthe floor in which odor trails are used to mark the finished area, (2) a cooperative taskin which a pathfinder robot lays an odor trail to guide other robots, and (3) an exploringtask in which a robot lays an odor trail on its way out and tracks it back to the initialposition when the task is accomplished.

16.4.2

Robot Implementations

The most straightforward implementation of ants’ behavior is a robot performingtropotaxis with left and right odor sensors. In the early work of Russell et al., a robotwith two QCM sensors was developed [16]. An odor trail is laid by dissolving camphorin an organic solvent and applying the solution to the floor. Although the solvent im-mediately evaporates, the camphor trail can persist for several hours. The robot suc-cessfully traced an odor trail consisting of two straight sections of 50 cm and a sharpturn of 30 degrees between them [16].There are several variations of this trail following robot. Stella et al. reported a robot

equipped with two conducting polymer sensors [18]. Russell later reported a simplerobot with a single QCM sensor [25]. In this case, a klinotactic algorithm is employed tofollow the edge of a trail. Webb developed a robot with two semiconductor gas sensors(SB-AQ1, Figaro) to investigate the behavioral mechanism of ants [13]. An artificialneural network devised after the tropotactic behavior of ants was employed to controlthe robot.

16.4.3

Engineering Technologies for Trail-Following Robots

A major disturbance in trail following is external odor confusion [25]. When the leftsensor is above the trail and the right sensor is off, the right sensor should ideally showno response. In reality, convection and diffusion bring the odor molecules to the rightsensor resulting in a confusing response. This problem can be overcome by using aircurtain [25, 29]. Figure 16.5 shows the second generation of air-curtain sensor devel-oped by Russell [25].Another attempt being made to extend the ability of trail-following robots is to en-

code useful information into odor trails. Russell proposed several ways of informationencoding [25]. A pulse-coded trail as shown in Fig. 16.6 can store information such asthe direction of the trail, the identity of the robot laying the trail, and a warning aboutconditions further along the trail [25]. The information can be retrieved by scanningthe trail with a sensor array. This type of system could be also used to obtain odorimages evaporating from buried objects such as leaking gas pipes.


16.5

Plume Tracking Robots

As reviewed in Section 16.2.2, some animals have the excellent ability to locate odorsources by tracking their plumes. In this section, the robotic researches inspired bythese animal behaviors are reviewed. The potential applications for the robots thattrack aerial or underwater plumes include searches for hazardous chemicals, pollu-tant sources, fire origins, and natural resources.Difficulties in plume tracking come from the random and unstable nature of che-

mical plumes. While chemical trails marked on the ground never change their shapes,turbulence of flow meanders chemical plumes. They sometimes even change theirdirections when the direction of air or water flow shifts. Therefore, occasional failuresare almost inevitable in the tracking of plumes. As revealed from the following sec-tions, the keys for successful tracking not only lie in the algorithms to track plumes butalso in the fail-safe mechanisms to relocate the lost plumes in case of failure.

Fig. 16.5 Air curtain sensor for trail following robots (adapted from [25]).

A small fan creates airflow to repel external odor. The part of this airflow

moves inward through the QCM sensor to the exhaust, and thus brings the

odor to the sensor only from the trail just beneath the sensor


16.5.1

Chemotactic Robots

The research on plume tracking robots started with purely chemotactic robots. Sandiniet al. developed a robot with two semiconductor gas sensors (TGS800, Figaro) for gasleak detection [15]. This robot performs tropotactic search as shown in Fig. 16.7A. Asimilar tropotactic algorithm was also employed in the robot with two conductometricpolymer sensors developed by Kazadi et al. [19].As mentioned in Section 16.3.2, robots can be used as tools for biologists to inves-

tigate mechanisms of animal behaviors. Consi et al. developed an underwater robotbased on American lobsters that crawl on the bed of oceans tracking food smells[24, 30]. The robot was equipped with conductivity sensors that mimicked the sizeand the separation of lobsters’ antennules. Kuwana et al. reported a small robot mi-micking a male silkwormmoth that walks to a female releasing the sexual pheromone[22]. Two pheromone sensors made of moths’ antennae were used for the robot.All the robots introduced above employ similar tropotactic algorithms. However,

there are differences in how these robots react when no chemical signal is per-ceived. Due to the time-varying nature of a chemical plume, a robot may sometimesleave the plume by chance. Since there is no signal outside the plume, a simple tro-potactic robot continues to move straight and never returns. Therefore, it is importantto incorporate algorithms to relocate the lost plume. Several algorithms including

Fig. 16.6 Robotic system with eight QCM sensors

to detect coded trails (adapted from [25]). While the

two leftmost sensors are used to trace the conti-

nuous guide path, the others are used to detect

pulse-coded trails


backing up [11, 30], random walk [15], and zigzag walk embedded in a recurrent arti-ficial neural network [21] have been proposed as fail-safe mechanisms in chemotacticsearch (see Fig. 16.7B).The limit in applying purely chemotactic strategies lies in the structure of chemical

plumes. As seen in Fig. 16.1, local and instantaneous gradients fluctuate significantly.Those fluctuations often mislead a chemotactic robot resulting in the circuitous move-ment of the robot in Fig. 16.7A. The concentration gradient along the plume centerlineis extremely small except in the vicinity of the source. Therefore, when a robot hasstarted a search from a distant place, the success rate can be low [27, 30]. One wayto overcome this problem would be to use a swarm of cooperative robots [15, 31]. San-dini et al. proposed to use multiple tropotactic robots with a communication link toexchange information among the robots nearby [15]. It was reported that the robotssuccessfully gathered around the source location by programming each robot to beattracted to the robot signaling the higher concentration. Another way to overcomethe weakness of chemotaxis is to use flow direction in navigating a robot, which isdescribed in the next section.

Fig. 16.7 Chemotactic robots. (A) A robot tracking concentration gradients detected

by the comparison of the left and right chemical sensor outputs. (B) Various algorithms to

relocate a chemical plume when the robots accidentally lose contact with it. Robot 1

is programmed to back up when neither sensor detects chemical. Robot 2 performs

random walk. Robot 3 mimics the behavior of a male silkworm moth. When one of the

sensors is stimulated, the robot surges in that direction to track a plume. When the

chemical signal is lost, the robot performs zigzag walk and circling to relocate the lost

plume


16.5.2

Olfactory Triggered Anemotaxis

As we have seen in the behavior of moths, the direction of flow carrying odor mole-cules is a strong directional cue in searching their source. We have developed a mobilerobot equipped with both gas and airflow sensors to incorporate the keys in moths’behavior into a robotic system [11, 27]. Wind direction with an accuracy of 458 is ob-tained in this robotic system for a wind velocity of 5–30 cm/s from the response pat-tern of the four thermistor airflow sensors (F6201-1, Shibaura Electronics). The twoorthogonal components of the concentration gradient are also measured as the re-sponse differences between the two pairs of diagonally aligned semiconductor gassensors (TGS822, Figaro). This robot tracks a chemical plume as shown inFig. 16.8A. While tracking the plume, the gas sensors are used to keep the robot head-ing towards the plume centerline. Due to the random nature of chemical plumes,however, a fail-safe mechanism to relocate the lost plume is again required to yielda high success rate. Although the algorithms for chemotactic robots shown in

Fig. 16.8 Anemotactic search algorithm triggered by chemical cues.

(A) The robot proceeds obliquely upwind to the side with higher

concentration. This oblique movement keeps the robot close to the

plume centerline. When the robot accidentally leaves the plume, it

performs side-by-side scanning similar to moths’ casting. (B) Experi-

mental result of the robot tracking a chemical plume at 1 cm/s [27]. A

nozzle releasing ethanol vapor at 75 ml/min was placed in a clean room

where an air conditioner was producing wind of about 30 cm/s. The

solid lines show the track of the robot when moving upwind, and the

dotted lines show that when the robot was casting


Fig. 16.7B can be also applied here, the robot with airflow sensors can again employ thewind direction as a useful directional cue.When a moth has lost contact with an odor plume, it scans across the wind with

gradually increasing width as shown in Fig. 16.2. This casting flight is a reasonablestrategy in relocating the lost plume since the possibility of hitting into a plume elon-gated in the wind direction is maximized when a searcher travels across the winddirection [32]. As shown in Fig. 16.8B, this casting behavior was successfully incorpo-rated into the robot. It quickly recovered the plume after a single scan and resumedtracking the plume.Russell et al. reported a mobile robot equipped with a custom-made wind vane [26].

Although the robot performed a similar anemotactic search, a klinotactic strategy witha single QCM gas sensor was employed to adjust the robot position across the winddirection. This simplifies the robot structure, and eliminates the need to match thesensitivities of gas sensors used. However, there is a trade-off between the simplicityand themeasurement time. In klinotaxis, the robot needs to scan left and right tomakea comparison.

16.5.3

Multiphase Search Algorithm

Olfactory triggered anemotaxis described in the previous section shows its maximumperformance in uniform flow fields, which we encounter in wind tunnels or flumes totest robots and animals. However, plume-tracking robots may face more complicatedflow fields in real applications. In a domestic or industrial building, for example, themain source of wind is an air conditioner and a robot often encounters winds frommultiple air-supply openings simultaneously.One way to tackle a difficult task is to decompose it into easier subtasks. In order to

cope with winds from multiple directions, a multiphase algorithm shown in Fig. 16.9was devised [27]. When a wind from another direction is merging into a side of achemical plume, the wind direction in this merging area becomes unstable. There-fore, care should be taken to employ the anemotactic strategy. When detected concen-tration is low, the robot might be in this merging area where unstable winds oftendirect an anemotactic robot in wrong directions. The robot should search for higherconcentration by a chemotactic strategy. Anemotaxis can be employed only when highconcentration is detected. In this case, the robot is thought to be in the center of theplume where only the wind from the source direction exists. This change in strategycan bemade with a pre-defined threshold in concentration. In Fig. 16.9B, however, thechange was made when one strategy makes no significant progress for 60 s. Thisensures timely changes in strategies even when the pre-defined threshold is inap-propriate.To accomplish a fully autonomous search in real applications, there still remain

many questions, including how to locate a plume for the first time in the absenceof any chemical signals, and how to decide when the odor source has been locatedso as to terminate the search. In the multiphase algorithm described above, the robot


sits still until a certain level of gas is detected. This is based on the scenario that therobot is placed in a room as a replacement for conventional gas alarms and that it needsto save its energy until a leakage actually occurs. If the robot is brought to the placewhere a leakage is detected, more active strategies should be employed. Moving acrossthe wind [26] as shown in Fig. 16.10 would be a choice. This is known to be the mostefficient strategy in finding a plume whenmultiple sources are dispersed in a uniformwind field [32]. As shown in Fig. 16.10, a plume extends downstream from each sourceto a finite length until the concentration is diluted below the detection limits of thesensors. The robot crossing the flow would eventually hit into one of the plumes in thefield although it might have passed by some of them on its way.When the characteristics of the source, such as its shape and size, are known, they

can be used for identifying it. Russell et al. proposed the use of a bump sensor for bothobstacle avoidance and declaration of the source [26] (see Fig. 16.10). It may not benecessary for a robot to go all way up to the source to declare the source location.When a robot scans the chemical plume on its way to the source, the concentration

Fig. 16.9 Multiphase search algorithm to cope with winds from

multiple directions. (A) The robot first tracks concentration gradients to

escape from the area of unstable wind. When the robot reaches the

center of the plume, high concentration of the target chemical is

detected. The robot then tracks the plume in the upwind direction.

(B) Experimental result [27]. The multiphase algorithm was tested in

the same clean room as in Fig. 16.8B. The starting position of the robot

was moved to the side of the ethanol plume where the wind from

another direction was merging. Thick lines show the path of the robot

tracking the concentration gradient, and thin lines show that in tracking

the plume in the upwind direction


change tracing the plume shape is observed. When an appropriate plume model isprepared, the source location can be found by curve-fitting the model to the observedconcentration change and extrapolating the curve to the source location [33].

16.6

Other Technologies in Developing Plume Tracking Systems

16.6.1

Olfactory Video Camera

An array of chemoreceptors on the eight legs of a blue crab might be able to detectinformation that is not accessible with the pair of antennae on a flying moth. The“olfactory video camera” shown in Fig. 16.11 is an engineering realization of suchspatially distributed sensor arrays [17, 34]. As described in Section 16.2.2, the sensorresponses observed in a chemical plume are highly intermittent, and this intermit-

Fig. 16.10 Multiphase algorithm proposed for a robot equipped with a

bump sensor (adapted from [26]). The robot is first made to move across

the wind until it hits into a chemical plume. The robot then starts tracking

the plume in the upwind direction. When the robot hits an obstacle while

tracking the plume, it circles around the obstacle by using the bump sensor.

If the target gas is detected at the upstream edge of the obstacle, the robot

resumes upwind search. If not, it can be declared that the obstacle is the

source of the target gas

16.6Other Technologies in Developing Plume Tracking Systems 413413

tency enables to track patches of the plume. When a patch passes over the array, theflow direction and speed can be determined from the visualized image.When the array is moved tracking the visualized plume reversely, it eventually ap-

proaches the source. As seen in Fig. 16.10, the source location can be determined to bethe point where the target gas is detected on its downstream edge but not on its up-stream edge. This can be easily judged from the visualized image when the array isplaced over the source location [35].

16.6.2

Odor Compass

Marine crustaceans flick their antennules, and terrestrial vertebrates show sniffingbehavior. These actions modulate the reception of chemical signals at the animals’sensors [36]. An interesting example of this signal modulation is wing fanning ofa male silkworm moth tracking a pheromone plume. Mimicking this mechanism,a sensing probe consisting of two semiconductor gas sensors (TGS822, Figaro)and a small fan was devised and termed an “odor compass” [37]. Experiments showedthat the effect of the fan is significant in obtaining directional cues (see Fig. 16.12). Thedirection toward the source can be found by rotating the compass and determining thedirection where the two sensor responses match.This sensing mechanism can be extended to a three-dimensional search by adding

two vertically aligned gas sensors and rotating the compass head three dimensionally[38]. It was shown from the experiments that this system is effective in searching odorsources around obstacles where complicated three-dimensional fields are formed (seeFig. 16.13) [38, 39].

Fig. 16.11 Schematic diagram of a gas/odor flow imaging system

termed “olfactory video camera” [17, 34]. It consists of an array

of 5 � 5 gas sensors with 1 cm spacing, and presents the visualized

image of instantaneous concentration distribution over the small

array


Fig. 16.12 Mechanism of odor compass consisting of two gas

sensors and a small fan [37]. Since the gas concentration gradient

along the plume axis is small, no significant difference in the left

and right sensor responses is observed when the fan is turned off.

When it is turned on, however, the plume is deformed by the airflow

and the sensor closer to the source shows a stronger response

Fig. 16.13 Result of plume tracking using a three-dimensional odor

compass [39]. A nozzle releasing ethanol vapor at 300 ml/min was

successfully located from behind a large obstacle. The compass

was iteratively moved in the estimated source direction by 30 cm.

(A) Perspective view. (B) Side view

16.6Other Technologies in Developing Plume Tracking Systems 415415

16.7

Concluding Remarks

Various aspects of the mechanisms underlying olfactory-guided behaviors of animalshave been transferred to robotic platforms. Animals show a variety of behaviors each ofwhich is optimized for the habitat of that species, and there seems to be no singleengineering implementation that can be used in every situation. Future work isneeded to establish design strategies that can tell us which type of system is best suitedfor the problem of current interest and how we can determine its design parameters.Development of chemical sensors for mobile robots is also an important subject for

future work. Current chemical sensor technologies were originally developed for sta-tionary sensing systems, and their limitations such as long response times have beenstrong constraints in the development of mobile biomimetic robots. Chemical sensorstailor-made for mobile robots would further expand the abilities of chemosensory ro-bots and open up new directions in the sensor technologies.

Acknowledgement

We gratefully acknowledge that the ideas presented in this article came from the con-tinuing collaborative work with Dr. Takamichi Nakamoto. Enlightening discussionswith a chemist (Dr. Jiri Janata), fluid mechanical engineers (Drs. Philip Roberts andDonald Webster), and biologists (Drs. Marc Weissburg, David Dusenbery, and TroyKeller) are also acknowledged. We thank Dr. R. Andrew Russell for giving us thepermission to quote his interesting work.

References

1 W. J. Bell, T. R. Tobin. Biol. Rev. 1982, 57,219–260.

2 W. C. Agosta. Chemical Communication:The Language of Pheromones, ScientificAmerican Library, New York, 1992.

3 E. A. Arbas, M. A. Willis, R. Kanzaki.In Biological Neural Networks in InvertebrateNeuroethology and Robotics (Eds.: R. D. Beer,R. E. Ritzmann, T. McKenna), AcademicPress, San Diego, 1993, Chapter VIII.

4 D. B. Dusenbery. Sensory Ecology,W. H. Freeman and Company, New York,1992.

5 J. Murlis, J. S. Elkinton, R. T. Carde.Annu. Rev. Entomol. 1992, 37, 505–532.

6 P. A. Moore, J. Atema. Biol. Bull. 1991, 181,408–418.

7 T. D. Wyatt. Nature 1994, 369, 98–99.8 J. Atema. Biol. Bull. 1996, 191, 129–138.9 M. J. Weissburg, R. K. Zimmer-Faust.J. Exp. Biol. 1994, 197, 349–375.

10 T. A. Keller, M. J. Weissburg. Abstr. AquaticSciences Meeting Amer. Soc. Limnol. Oceanogr.2000.

11 H. Ishida, K. Suetsugu, T. Nakamoto,T. Moriizumi. Sensors and Actuators A 1994,45, 153–157.

12 T. Nakamoto, T. Yamanaka, H. Ishida,T. Moriizumi. Meeting Abstr: Electrochem.Soc. 1996, 96–2, 1163.

13 B. Webb. Neural Networks 1998, 11,1479–1496.

14 R. Rozas, J. Morales, D. Vega. Fifth Inter-national Conference on Advanced Robotics1991, 1730–1733.

15 G. Sandini, G. Lucarini, M. Varoli. Proc.1993 IEEE/RSJ Int. Conf. Intelligent Robotsand Systems 1993, 429–432.

16 R. Deveza, D. Thiel, A. Russell, A. Mackay-Sim. The International Journal of RoboticsResearch 1994, 13, 232–239.


17 T. Nakamoto, T. Tokuhiro, H. Ishida,T. Moriizumi. Technical Digest of Transducers’99 1999, 1878–1879.

18 E. Stella, F. Musio, L. Vasanelli, A. Distante.Proc. 1995 Intelligent Vehicles Symposium1995, 147–151.

19 S. Kazadi, R. Goodman, D. Tsikata, D.Green, H. Lin. Autonomous Robots 2000, 9,175–188.

20 Y. Kuwana, I. Shimoyama, H. Miura. Proc.1995 IEEE/RSJ Int. Conf. Intelligent Robotsand Systems 1995, 530–535.

21 Y. Kuwana, I. Shimoyama. The InternationalJournal of Robotics Research 1998, 17,924–933.

22 Y. Kuwana, S Nagasawa, I. Shimoyama,R. Kanzaki. Biosensors and Bioelectronics1999, 14, 195–202.

23 T. Kikas, H. Ishida, P. J. W. Roberts,D. R. Webster, J. Janata. Electroanalysis 2000,12, 974–979.

24 T. R. Consi, J. Atema, C. A. Goudey, J. Cho,C. Chryssostomidis. Proc. 1994 Symp.Autonomous Underwater Vehicle Technology1994, 450–455.

25 R. A. Russell. Odour Detection by MobileRobots, World Scientific, Singapore, 1999.

26 R. A. Russell, D. Thiel, R. Deveza,A. Mackay-Sim. Proc. 1995 IEEE Int. Conf.on Robotics and Automation 1995, 556–561.

27 H. Ishida, Y. Kagawa, T. Nakamoto,T. Moriizumi. Sensors and Actuators B 1996,33, 115–121.

28 S. Kawamura, K. Matsuyama, T. Nakamoto,T. Moriizumi. Technical Digest of the 17thSensor Symposium 2000, 321–324.

29 R. A. Russell. IEEE Robotics and AutomationMagazine 1995, 2, 3–9.

30 F. W. Grasso, T. R. Consi, D. C. Mountain,J. Atema. Robotics and Autonomous Systems2000, 30, 115–131.

31 V. Genovese, P. Dario, R. Magni, L. Odetti.Proc. 1992 IEEE/RSJ Int. Conf. IntelligentRobots and Systems 1992, 1575–1582.

32 D. B. Dusenbery. J. Chem. Ecol. 1989, 15,2511–2519.

33 H. Ishida, T. Nakamoto, T. Moriizumi.Sensors and Actuators B 1998, 49, 52–57.

34 H. Ishida, T. Yamanaka, N. Kushida,T. Nakamoto, T. Moriizumi. Sensors andActuators B 2000, 65, 14–16.

35 H. Ishida, T. Nakamoto, T. Moriizumi,T. Kikas, J. Janata. Biol. Bull. 2001, 200, 222–226.

36 M. A. R. Koehl. Mar. Fresh. Behav. Physiol.1996, 27, 127–141.

37 T. Nakamoto, H. Ishida, T. Moriizumi.Sensors and Actuators B 1996, 35, 32–36.

38 H. Ishida, A. Kobayashi, T. Nakamoto,T. Moriizumi. IEEE Trans. Robot. Autom.1999, 15, 251–257.

39 H. Ishida, T. Nakamoto, T. Moriizumi.Sensors Update 1999, 6, 397–418.

16.7 Concluding Remarks 417417

Part D

Applications and Case Studies


17

Environmental Monitoring

H. Troy Nagle, Ricardo Gutierrez-Osuna, Bahram G. Kermani, Susan S. Schiffman

Abstract

In this chapter, we review some of the previous proof-of-principle work done in thisfield. Examples of water, land, and air monitoring experiments are examined. Fourcase studies are then presented. The first three demonstrate the ability of the electro-nic nose (e-nose) to classify odors from animal confinement facilities (odor sourcedetermination, odorant threshold detection, and odor abatement evaluation). Thefourth case study demonstrates that the e-nose can differentiate between five typesof fungi that commonly diminish indoor air quality in office buildings and industrialplants. Finally, we conclude that environmental monitoring is a promising applicationarea for e-nose technology.

17.1

Introduction

The field of environmental monitoring encompasses a broad range of activities. Con-tamination of the environment can occur not only by dumping wastes in water, land,and air, but also by generating noise in the audio and communications frequencyranges. Sensing systems have been developed for all of these applications. In thischapter, we focus on efforts to employ an electronic nose (e-nose) to monitor airbornevolatile organic compounds that are released when waste products are dumped inwater, land, or air.

17.1.1

Water

Water quality is threatened when agricultural and industrial concerns allow their wasteproducts to seep into groundwater or to flow into streams or rivers. The e-nose can beused in these applications on samples of the effluent. The headspace of such samplescan be tested with an e-nose system, on-line or off-line, to establish the time-course of


419419

emission profiles. Boreholes can also be employed to collect samples to test ground-water contamination. Several research groups have studied the e-nose as an instru-ment for monitoring water quality.Some teams have utilized metal-oxide sensors for monitoring. Baby et al. [1] used

the MOSES II e-nose to measure contaminating residues of insecticides and productsfrom leather manufacture that are often offloaded into streams and rivers. Dewettincket al. [2] employed an e-nose consisting of 12 metal-oxide sensors to monitor volatilecompounds in the effluents of a domestic wastewater treatment plant over a 12-weekperiod. Correlation between the relative overall e-nose output and the parameter ‘vo-latile suspended solids’ was good, indicating adsorption of volatile organic compounds(VOCs) onto the organic particles. This study also concluded that the e-nose has pro-mise in wastewater monitoring applications. In another study by the same group, VanHege et al. [3] explored the application of evaporative technology as an alternativedesalination technique for wastewater treatment plant effluents. Evaporation comple-tely removed most inorganic and organic contaminants. An e-nose was employed tomonitor changes in odor quality and intensity due to volatilization of the VOCs presentin the effluent.Conducting polymers have also been used to analyze wastewater. Di Francesco et al.

[4] studied the use of an e-nose with conducting-polymer sensors and fuzzy-logic-based pattern recognition algorithms to test wastewater samples. In other work ane-nose with 12 polypyrrole conducting-polymer sensors was used to monitor quies-cent sewage liquors at three wastewater treatment plants over an 8-month period[5–7]. The e-nose was evaluated as a replacement for human panels in monitoringliquid wastewater samples, wastewater odor, and tainting compounds in water sup-plies. The study revealed that a strong linear relationship is expected for site/source-specific odor samples. The study also showed that low levels of organic pollu-tants can be detected by monitoring water samples with the e-nose. In addition, thestudy suggested that it might be feasible to use an e-nose to monitor and/or control thebiochemical activities of a wastewater treatment process. More recently, Bourgeois andStuetz [8] reported the use of a similar sensor array to analyze wastewater samplessparged with N2 gas in a temperature-controlled flow-cell. The headspace gas wasthen supplied through a temperature-controlled transfer line to the conducting-poly-mer sensors. They concluded that an externally generated headspace gas could be usedto monitor changes in wastewater quality, and could provide a simple non-invasivetechnique for on-line monitoring of wastewater.Continuing this avenue of research, Stuetz et al. [9] and Bourgeois et al. [10] exam-

ined the use of real-time sensors and array systems for monitoring global organicparameters such as biochemical oxygen demand and total organic carbon. Stuetzet al. [9] and Stuetz [11] compared the odor profiles of sewage liquids with correspond-ing biochemical oxygen demand and total organic carbon measurements, anddetermined that a number of different wastewater quality relationships could be for-mulated from the e-nose analysis of a sewage liquid. They concluded that the organiccontent of wastewater, as well as the potential of wastewater to produce nuisanceodors, could be predicted from a single headspace analysis of a sewage liquid usinga sensor array.

17 Environmental Monitoring420

Di Natale et al. [12] used a sensor array of ion-sensitive electrodes to analyze pollutedwater. The sensor array was processed using chemometrics, non-linear least squaresand neural networks. The devices that use sensor arrays to test liquid samples arecalled electronic tongues rather than e-noses. See Chapter 11 for more informationon electronic-tongue devices.Gardner et al. [13] and Shin et al. [14] developed a system for detecting cyanobacteria

(blue-green algae) in potable water. The e-nose system, employing an array of six com-mercial gas sensors, was able to detect 100% of the unknown toxic cyanobacteria usinga multi-layer perceptron (MLP) neural network. The results showed the potential for aneural network-based e-nose, as opposed to more traditional instruments such as li-quid chromatography or optical microscopy, to test the quality of potable water.

17.1.2

Land

Land contamination by toxic and radioactive materials is a chief concern in manycountries around the world. Garbage waste dumps are problems everywhere. Thee-nose has applications in this arena as well. Borehole samples can be placed in sam-ple containers to generate headspace VOCs. Adding specific reagents to some of thesesamples can accelerate the generation of VOCs and improve the sensitivity of the e-nose instruments. This is an emerging area for e-nose instrumentation and thereshould be considerable future growth in this segment of the e-nose market.There have been few research studies in this area. One example of note is Biey and

Verstraete [15]. They investigated the use of a 5-W UV lamp, generating ozone forseven hours per day, to reduce the odors produced by the decomposition of kitchenand vegetable waste. An Alpha M.O.S. FOX 3000 e-nose was used to measure odorlevels before and after treatment. They concluded that the UV treatment did indeedreduce the odor levels, and thus would be useful in summer, or all year around inwarm climates.

17.1.3

Air

Air quality has been the primary target of e-nose research projects in environmentalmonitoring [16, 17]. The e-nose can monitor odorous emissions at their source, suchas paper mills, animal production sites, power-plant stacks, vehicle exhaust pipes,compost facilities, wastewater treatment plants, animal rendering plants, paintshops, printing houses, dry cleaning facilities, and sugar factories. The e-nose alsoholds promise for monitoring emissions from near-source or remote locations in apopulated area. Currently, available sensor arrays have not proven efficient at remo-tely located sites, owing to their lack of adequate sensitivity to many of the offendingVOCs in odorant mixtures. However, e-nose measurements made at the source couldserve as input to mathematical emission dispersion models that can predict VOC con-centrations at remote locations given accurate meteorological data for a specific geo-


graphic location [18]. As sensor-array technology improves, the measuring of odorousVOCs at remote locations will become a significant market for hand-held e-nose de-vices (see Chapter 9).Although in most cases annoying atmospheric emissions do not menace public

health, they do greatly reduce the quality of life [4, 19]. Measuring these odors atthe site of complaints is very difficult due to the transient nature of the odorousevents. The e-nose offers the promise of being able to make accurate and repeatablemeasurements of odor profiles at sites of complaint. These e-nose measurements canbe correlated with those of human panels in order to calibrate the odor quality andperception scales [20] (see Case Study 3 in this chapter).Now we discuss several examples of the application of the e-nose to monitoring air

quality. Odor abatement and control is a major issue facing municipal sewage treat-ment facilities. The odors emitted from these facilities can be monitored by an e-nose.Gostelow et al. [21] reviewed various sensory, analytical, and e-nose methods for mon-itoring sewage facility emissions. Stuetz et al. [22, 23] employed a Neotronics NOSE toinvestigate emissions from ten sewage treatment facilities. Odor levels measured bythe NOSE unit were compared with those of an independent human panel, measuredin odor units per cubic meter. The effect of biofilters was also considered. A linearrelationship was observed between the NOSE measurement and the human panelresults for data at each independent site. At low odor levels, the results were also ex-tended to themultiple site case. Hydrogen sulfide concentrations, although commonlyused as a measure of odor strength, were also compared with the human panel resultsand were found not to be a reliable marker compound for measuring sewage odorconcentrations.The perception of the quality of indoor air by building inhabitants is addressed by

Schreiber and Fitzner [24, 25]. Delpha et al. [26, 27] investigated the use of an e-noseusing metal-oxide TGS sensors for the detection of a leaking refrigerant gas (ForaneR134a) in an air-conditioned atmosphere. First the researchers showed that the timeresponse of the TGS sensors to Forane R134a gas in humidity varying from 0 to 85%could be represented by a double exponential model. The authors then demonstratedthe ability to identify the target gas by discriminant factorial analysis, even for cases inwhich the relative humidity or the gas temperature were outside the range of the train-ing database. In a similar study, Sarry and Lumbreras [28] investigated the detection ofcarbon dioxide, Forane R134a, or their mixtures, without a sensor dedicated to carbondioxide measurement. They used an array of five tin-dioxide sensors. Discriminantfactorial analysis was used for processing the data. The authors report a reliable sys-tem can be designed for this application.Ramalho [29] analyzed the characteristics of indoor paints and their effect on per-

ceived indoor air quality. Ten different indoor paints were presented to an e-nose andto 13 trained panelists. Significant differences among panelists were found, whereasthe sensors displayed little difference. However, some similarities were found betweensome sensors and individuals.Feldhoff et al. [30] compared the ability of an Alpha M.O.S. FOX 4000 and a LDZ

Laboratory Smart Nose GA200 to differentiate between twenty Diesel fuels from threedifferent refineries. The authors reported that both units were able to correctly identify


the production site of the 20 samples. However, the Smart Nose uses a mass spectro-meter and its data were easier to obtain and were more reproducible. In a similarstudy, Lauf and Hoffheins [31] illustrated that a selected array of chemical sensorscan produce unique signatures for many aviation and automotive fuels. Patternsfor aviation fuel are readily identified by visual inspection. The differences amongautomotive fuels with different octane ratings are subtle but perceptible. Gasohol mix-tures have strikingly different signatures from pure gasoline. The results indicate thatan e-nose can distinguish between various classes of petroleum-based fuels.Automotive ventilation may also be monitored and controlled by an e-nose. Menzel

and Goschnick [32] investigatedmethods for improving the time response of an e-noseinstrument intended for on-line discrimination applications. Their method combinedthe classification of the steady-state and transient response via time-series analysis.Rapid signal transients were detected by appropriate digital filters, while steady-statesignals were classified by standard statistical methods. To illustrate the method, theyinvestigated automatic control of the ventilation flap of an automobile. Steams of badair were detected in one to two seconds. The error in the detection of pollutants wasreduced from the original 25% to only 10% for their new method.E-nose systems have also been studied for detection of hazardous materials and

gases. For example, Hopkins and Lewis [33] investigated the use of arrays of car-bon-black/organic-polymer composite chemiresistive vapor detectors for detectingnerve agents. Chapter 23 of this handbook is devoted to the detection of explosives.Odorousemission fromanimalproduction facilitieshasbeenextensively studiedover

the last few years.We present several case studies in this area later in the chapter. Otherresearch groups have also studied this important problem.Hobbs et al. [20] correlated e-nosemeasurementsofpigmanureodors to thoseofahumanpanel.Fourof theprincipleodorous compounds in pig manure were selected for the study. Thirty-one differentmixtures of hydrogen sulfide, 4-methyl phenol, ammonia, and acetic acid were used tosimulate the livestock waste odor. A radial-basis-function neural network was used forsignalprocessing.Predictionsusinga linear regressionmodelwereonaverage20%lessthan observed values. The authors reported that this approach using the four mainodorants is appropriate for determining the odor concentration of pig manure.An e-nose can frequently be employed to identify specific VOCs and mixtures of

VOCs. Hudon et al. [34] compared the effectiveness of three different e-nose instru-ments in measuring the odor intensity of n-butanol, CH3COCH3, and C2H5SH, andbinary mixtures of n-butanol and CH3COCH3. Two commercial e-nose systems, theAromaScan A32S (conducting-polymer sensors) and the Alpha M.O.S. Fox 3000 (me-tal-oxide sensors), and an experimental unit with Taguchi-type tin-oxide sensors wereemployed. The e-nose measurements were processed using linear regression analysisand neural networks. Very strong correlation (q ¼ :99) was obtained between the sen-sory data and the two commercial units when using neural network analysis. In arelated study, Negri and Reich [35] used an e-nose with commercially available tin-oxide sensors to analyze a mixture of gases containing carbon monoxide, ethanol,methane and/or isobutane. They modeled the theoretical response function of thearray and designed a pattern recognition scheme for the simultaneous identificationof a given gas and its concentration in the mixture.


The growth of bacteria and fungi on organic matter generates a broad range of vo-latile organic compounds and fixed gases. Wessen and Schoeps [36] and Sunesson etal. [37] showed that the presence of certain VOCs could be used as an indicator of thepresence and of the identity of microorganisms. Holmberg [38], in a dissertation atLinkoping University in Sweden, used an e-nose with 15 sensors to classify five typesof bacteria (Escherichia coli, Enterococci sp., Proteus mirabilis, Pseudomonas aeruginosa,and Staphylococcus saprophyticus). The 15 sensors included nine metal-oxide semicon-ductor field-effect transistors, four Taguchi-type devices, one carbon dioxide sensor,and one oxygen monitor. The volatile compounds generated by the bacteria weresampled from agar plates. The results suggested that this e-nose could successfullyclassify Escherichia coli and Enterococci sp. but was less successful with the other bac-teria.Gardner et al. [39] used an e-nose that contained six commercial metal-oxide sen-

sors, a temperature sensor, and a humidity sensor to predict the class and growthphase of two types of bacteria, Escherichia coli and Staphylococcus aureus. The six sen-sors were designed to detect hydrocarbons, alcohols, aldehydes/heteroatoms, polarmolecules, and non-polar compounds. The best mathematical model correctly identi-fied 100% of the unknown S. aureus samples and 92% of the unknown E. coli samples.Other studies have also found that bacteria can be discriminated using an e-nose. In

an evaluation of seven bacterial strains, Vernat-Rossi et al. [40] were able to correctlydiscriminate 98% of a training set with a cross-validation estimate (test set) of 86%using six semiconductor gas sensors. Studies at AromaScan PLC [unpublished datafrom Dr. Krishna Persaud] showed that polymer sensors performed well in discrimi-nating multiple samples of five different types of bacteria.Keshri et al. [41] used an e-nose consisting of 14 polymer sensors to classify six

spoilage fungi (four Eurotium sp., a Penicillium sp., and a Wallemia sp.). The head-space was sampled after 24, 48, and 72 hours of growth. The e-nose discriminatedthe fungi at the 24-hour mark (prior to the visible signs) with an accuracy of 93%.The best results occurred at the 72-hour mark.The measurement of air quality by an e-nose requires a hand-held unit. Several

commercial instruments are available as described in Chapters 7 and 9. Nicolas etal. [42] have also developed a portable prototype e-nose based on tin-oxide sensorsfor field applications; with this device they generate a warning signal when the mal-odor level exceeds some given threshold value, identify the source of an odor detectedon site, or identify on-line and monitor levels of an odor in the field.As outlined above, the field of environmental monitoring is very broad. In this chap-

ter, we will focus on case studies in livestock odors and microbial contamination.


17.2

Special Considerations for Environmental Monitoring

17.2.1

Sample Handling Problems

17.2.1.1 Sample Lifetime

If not properly handled (e.g., long exposure to sunlight), some organic samples maydisintegrate or undergo certain chemical reactions. Therefore, considerable effort isrequired in order to maintain samples in their original state prior to their delivery tothe sniffing device.

17.2.1.2 Humidity

As will be discussed later, it is important that the various odor samples have similarhumidity levels. The humidity of the reference sample should also be adjusted to thatof the odor samples. This is to ensure minimal response due to humidity when switch-ing from reference to odor inputs. A closed-loop humidity control system for the re-ference input is offered in some commercial systems for this reason.

17.2.1.3 Extraction of volatiles

In cases in which the number of volatile molecules is low, onemay be required to boostthese numbers via some preconcentration, activation, or agitation method. In order torecord a meaningful sensor response, the concentration of volatiles in the samplemust be above a minimum threshold. Certain agitation methods may be necessaryfor liquid samples in order to increase the concentration of volatiles in the head-space. Conversely, in the case of highly volatile molecules (e.g., alcohols), one mayneed to dilute samples in order to avoid sensor saturation. Chapter 3 covers precon-centration methods.

17.2.1.4 Tubing system

The acquisition system is generally equipped with a tubing system that delivers volatilecompounds from the sample container to the sensor compartment, and then to theexhaust outlet. The material used in the tubing must be inert to the type of odorantsthat the device handles. In other words, the tubing material should not modify oradsorb the odor of the samples. Similar requirements exist for the sensor compart-ment, valves, and so on.

17.2.1.5 Temperature

The temperature of the sample, sensor chamber, and sensors must be kept constant toachieve repeatable performance of the e-nose system. A temperature perturbation cancause shift/deformation in the generated patterns, by virtue of changes in concentra-tion or sensor behavior. A constant temperature is usually maintained using a feed-back control system. Temperature control is important for all types of sensors.

17.2 Special Considerations for Environmental Monitoring 425425

17.2.2

Signal Processing Challenges

In addition to appropriate sample handling, signal-processing algorithms are requiredto compensate for the variability of conditions in the field. By including temperatureand humidity sensors in the e-nose instrument, it may be possible to compensate forthese effects by means of signal processing algorithms. Sensor baseline drift and un-wanted concentration effects may also be handled by means of preprocessing algo-rithms (see Chapter 5).Due to the large number of sensors and features (e.g., dynamic response record-

ings), the e-nose is subject to “the curse of dimensionality.” A large number of dimen-sions can hinder the true (and useful) information, so the use of dimensionality re-duction procedures (e.g. feature selection, principal components) is often required.These signal processing procedures must be carefully chosen to ensure that memoryand CPU requirements do not become prohibitive for an economical (e.g., hand-held)device [43].

17.3

Case Study 1: Livestock Odor Classification [44]

17.3.1

Background

Livestock industries are expanding rapidly throughout the world, and this expansion iscausing environmental concerns. Modernmethods of confining thousands of animalsin a single facility have led to increased production and profits while creating concernsabout odor and water pollution. Odors associated with livestock operations are gen-erated from a mixture of urine, fresh and decomposing feces, and spilled feed. Inswine operations, for example, odors emanate from the ventilation air of confinementbuildings, waste storage, and handling systems including lagoons and field applica-tions of waste. Anaerobic microbial decomposition of livestock waste appears to be thesource of the most objectionable smells. Odorous compounds identified in livestockwastes include sulfides, disulfides, volatile organic acids, alcohols, aldehydes, amines,fixed gases, nitrogen heterocycles, mercaptans, carbonyls, and esters. Reduction ofodors emanating from livestock operations is necessary to improve the relationshipbetween producers and their neighbors.Sensitive measurement techniques are important to characterize and document

swine odors, as well as evaluate the effectiveness of methods for reducing odors.At present, olfactometry using human odor panels is the most precise approachfor quantifying odors, since the human nose can detect compounds at concentrationsthat cannot be detected by any other method. Human evaluations, however, can betime-consuming, unrepeatable, and expensive. In addition, odor samples degrade ra-pidly, and thus human panels must perform evaluations shortly after collection foraccurate assessment. Because swine odor abatement research is being conducted


all around the world on a 24-hour basis, odor testing with human panels is often im-practical. Rapid, accurate, cost-effective evaluation of techniques to reduce odor pro-duction (such as the manipulation of pig diets to reduce excrement odor) is vitallyimportant to the swine industry. For this reason it would be helpful to determineif an e-nose can substitute for human odor panels in evaluating methods for odorreduction.

17.3.2.

Description of the problem

The objective of the following study was the classification of various odorant samplesrelated to a hog farm. The main task was to gauge the accuracy and the precision of ane-nose in identifying the source of unknown odor samples.

17.3.3.

Methods

Odor samples were collected from three locations at a rural hog farm: lagoon, fan, anddownwind ambient air. The samples were presented to an e-nose, and signal-proces-sing algorithms were used to classify the data. A cross-validation method was em-ployed to measure the performance of the system. At each step of this cross-validationmethod, 70% of the data was used to train the system, while the other 30%was used asan unknown sample set. The e-nose used for the experiments of this section was theAromaScan A32S (see Chapter 7). The core of the A32S system is an array of 32 con-ducting-polymer sensors. Depending on the mode of operation, the sensor compart-ment is exposed to one of the odorant sample, the reference gas, or the cleansing gas.The reference gas was generated by filtering, dehydrating, and humidifying steps. Thehumidity of the reference air was set to match that of the odor samples. The cleansinggas (2% n-butanol bubbler) was used to remove (detach) odorants from the sensorsafter each data acquisition cycle.Various air-samples from two lagoons, a confinement building exhaust fan, and a

downwind site at a hog farm in rural North Carolina were collected using 25-L Tedlar�

bags. The downwind-air sample was collected 1,500 feet from the swine operation.These bags were filled using a pump device and sealed barrel under negative pres-sure. The bags were cleaned using a combination of butanol, methanol, nitrogen,and/or dry air, and reused. The most commonly used cleaning technique was flush-ing with nitrogen, then a methanol vapor, followed by clean dry air. A major drawbackof this sampling method is the shipping and handling of the filled bags. Since theodors degrade over time, the samples should be processed the same day during whichthey are collected. Hence, this technique is adequate for sites that are located in closeproximity (within 150 miles) to the testing facilities. We have found that holding thebags overnight for processing the following day significantly reduces the odor inten-sity, and hence the reliability of the sample collection method.

17.3 Case Study 1: Livestock Odor Classification [44] 427427

17.3.4

Signal Processing Algorithms

The datasets obtained from the e-nose were analyzed using a set of algorithms listedbelow. More detailed explanations of the various algorithms can be found in Chap-ters 5 and 6. The main steps of signal processing in this case study are outlined asfollows:

17.3.4.1 Bias Removal

One of the drawbacks of polymer sensors is their inability to return (within a reason-able time frame) to the baseline after washing. The residual signal will result in agradual shift in the successive data acquisition cycles. The first step of preprocessingwas to remove the bias mathematically. In these experiments, the bias was removed bysubtracting the response of each sensor at the first time point from all the other sub-sequent time points in the dynamic response of that sensor.

17.3.4.2 Humidity

Another major weakness of some conducting-polymer sensors is their high sensitivityto water molecules. If not controlled, the common-mode response that is caused byhumidity could completely overshadow the signal of the odorants. Various approacheshave been proposed to counteract humidity and its effects. One is to model the re-sponse of the sensors to humidity, and then to subtract it from the composite re-sponse. However, due to the low repeatability of the patterns, this was not foundto be a suitable approach for the AromaScan A32S polymer sensors. Another ap-proach is to employ the humidity control features of the AromaScan A32S that allowthe operator to adjust the humidity of the reference signal to that of the odor sample.We should point out that researchers in this field are developing new types of con-ducting-polymer sensors that are much less sensitive to changes in sample humidity.

17.3.4.3 Concentration

One obvious challenge in sample preparation is the control of the volatile concentra-tion. Within certain ranges, the effect of concentration has been shown to be linear.When comparing samples of the same kind, one must be able to either normalize theeffect of concentration, or guarantee that samples contain similar concentrations of theodorant of interest. In the experiments of this study, the response of each sensor ateach time point was divided by the average response of all sensors at that time point.When the sensors operate in the linear range, this method has been shown to normal-ize the response of the sensors with respect to the concentration [44].

17.3.4.4 Dimensionality Reduction

In the following experiments, every sample produces 30 � 32 ¼ 960 data points. Sincea single training session may include several dozens of samples, it is evident that thedimensionality could become overwhelming for this problem. Therefore, in lieu of


supplying the time-series data directly into the processing unit, a reduced set of fea-tures was extracted prior to the main analysis.Data reduction was done in two stages. In the first stage, a series of bell-shaped

curves were used to serve as windowing functions. By using windowing functions,the set of 30 time points of the response of each sensor was reduced to four, the num-ber of windowing functions. The next step of data compression was done by Karhu-nen-Loeve (truncated) expansion (KLE), also known as principal components analysis.KLE is known to be the optimal linear method for data compression [45]. Using KLE, aseries of features, i.e., the significant eigenvectors, was extracted from the time-wind-owed traces of each sample. The dimension of the transformed signal was found dy-namically by analyzing the relationship between the eigenvalues of the covariancematrix [44]. The set of features extracted from the KLE compression was then directedinto an multi-layer perceptron neural network for training and testing. The learningrule of the neural network was based on the Levenberg-Marquardt method [46, 47].The back-propagation method [48] (with a momentum term and adaptive learningrates) was also used for comparison purposes. A genetic-algorithm-based supervisorwas designed to tune the number of neurons in the hidden layer and the learningparameters of the neural network. The genetic algorithm (GA) was also responsiblefor choosing all or a subset of the windowed values and/or features.

17.3.5.

Results

The results are depicted in Fig. 17.1. Aside from the difficulties of sample handling,the results appear to be reasonable. The figure shows the histogram of the perfor-mance of 100 cross-validated runs. The y-axis is the number of runs and the x-axisis the correct recognition in percent. Note that 97 of the runs gave a perfect 100%correct recognition, while the remaining three cases were 97% correct. The overallcorrection recognition rate was 99.92%.

17.3.6.

Discussion

Several alternative signal processing methods, e.g., neural networks with back-propa-gation, with and without the GA supervision, were tried prior to applying the above-mentioned methods. These alternative methods were found to achieve lower perfor-mance metrics. The preprocessing steps were found to be necessary for generatingrepeatable histogram patterns. A neural-network-based classifier with the Leven-berg-Marquardt learning rule was found to be appropriate for this particular pat-tern-recognition application. Using GAs as a supervisor provided a systematic, reli-able, and automated method for feature selection and architectural tuning of the neur-al network.The final hybrid GA-neural network system proved to serve as an effective signal-

processing technique for this application. However, regardless of the efficacy of the

17.3 Case Study 1: Livestock Odor Classification [44] 429429

signal-processing method, the quality of the final outcome is a function of the qualityof the input data. In general, due to their limited sensitivity, conducting-polymer sen-sors were found to be more suitable for odor samples containing high concentrationsof highly volatile molecules such as those found in fragrances.

17.4

Case Study 2: Swine Odor Detection Thresholds

17.4.1.

Description of the Problem

The detection threshold for a specific odorant mixture is related in part to the detectionthresholds of its individual components. In this study, we select one of the odorouscomponents of hog slurry – acetic acid – and compare the detection thresholds of ahuman panel and the AromaScan A32S for this compound.

Fig. 17.1 Histogram showing test results of 100 runs of training/

testing of hog-farm samples using the hybrid of neural-network and

genetic-algorithms in conjunction with the AromaScan A32S. The

number of runs is given on the y-axis, and the percent correct recog-

nition is given on the x-axis. On 97 of the runs, there was a perfect

100% correct recognition, while there was 97% correct recognition for

the remaining three cases


17.4.2

Methods

In this experiment, twelve serial dilutions of acetic acid that differed by a factor of threeand ranged from 5% to 0.0000094% v/v were presented to the human panel at theTaste and Smell Laboratory at Duke University Medical Center and the AromaScanA32S for evaluation. Odorless mineral oil was used as the diluent. The e-nose signalswere processed using the same procedure as Case Study 1 above [44, 49]. The tech-niques used consisted of a preprocessing stage and a data-compression stage. Thepreprocessing stage involved shifting each sensor’s curve, so that the initial resistancechange was adjusted to zero. The data-compression stage consisted of two steps: wind-owed time integration and Karhunen-Loeve expansion (KLE). The windowed timeintegration multiplied each sensor curve by four bell-shaped kernels and then com-puted the area beneath the curves. In this way, each odor sample was reduced from32 � 45 (sensors x seconds) to 32 � 4 (sensors x windows) features. Then the KLE wasperformed to extract the principal components in feature space.

17.4.3

Results

The dilution labels ranged from 13 to 1, for the highest and lowest concentrations,respectively. The resultant two-dimensional KLE scatter plot for the acetic acid dilu-tions in mineral oil is presented in Fig. 17.2. Note that a detection threshold betweenlabels 9 and 10 can be visually determined.

17.4.4

Discussion

Our results indicate that the e-nose has a detection threshold at a concentration that is afactor of three above that of the human panel. The detection thresholds for the fourhuman subjects were at dilutions 8 or 9 (two subjects at each dilution), whereas the e-nose was between dilution 9 and 10, as can be seen in the figure. Since dilution 10 hasan odorant concentration that is three times greater than dilution 9, and dilution 9 hasan odorant concentration that is three times greater than dilution 8, on average thehuman panel’s detection level is at a concentration that is three times lower thanthat of the e-nose. A factor of three in odorant concentration therefore gives the hu-man panel an advantage over the e-nose in this application. However, the e-nose can bedeployed on site and can measure emissions over long time periods, characteristics ofa monitoring system that are not practical for human-panel implementation

17.4 Case Study 2: Swine Odor Detection Thresholds 431431

17.5

Case Study 3: Biofilter Evaluation [50]

17.5.1


The objective of this study was two-fold. First, to develop an experimental procedure toevaluate biofilters for odor remediation in the ventilation exhaust fans of hog confine-ment buildings. Second, to determine if the AromaScan A32S could be utilized topredict the human panel olfactory ratings of malodors, before and after bioremedia-tion.

17.5.2

Methods

In order to rapidly screen the performance of various odor remediation materials, abench-top biofilter setup was developed at the NC State University Animal and PoultryWaste Management Center. The biofilter material consisted of earth, wood chips,small twigs, and straw. This material was placed in a one-inch diameter PVC tube,which was cut to a length of 3.9 inches. This length was selected because of the re-quirement to have the air reside within the filter for 15 seconds, which matches thespecifications of field units at this site. The tube was cemented at each end to a PVCfitting which had screw threads and an O-ring to produce an airtight seal with the

Fig. 17.2 Principal component analysis (PCA) of the e-nose data for

dilutions of acetic acid in mineral oil. The two-dimensional scatter plot

shows that a detection threshold occurs between labels 9 and 10


connecting piece. Wire mesh was placed on each end of the cemented tube fitting toprevent the biofilter material from spilling out of the tube.To test this biofilter setup, we conducted an odor remediation experiment with a

synthetic slurry following the concoction of Persaud et al. [51]. Serial dilutions (1/1, 1/3, 1/9, 1/27 and 1/81) of the headspace above the slurry, as well as serial dilutionsof the biofiltered synthetic slurry and biofiltered blank room air (as a control) werepresented to both the Duke human panel and the e-nose. The experimental setupis depicted in Fig. 17.3.To measure the human perception to the different odors and dilutions, the panelists

were asked to generate scores for intensity, irritation, and pleasantness using the 9-point scale shown in Table 17.1. The e-nose signals were preprocessed by computingthe fractional change in resistance of each sensor with respect to its baseline resistancein reference air (steady-state DR/R). The steady-state response of each sensor wasextracted to form a 32-dimensional feature vector.

Fig. 17.3 Experimental setup for malodor biofiltration assessment.

Air from the synthetic hog slurry and the room-air control is filtered and

delivered to the human sensory panel and e-nose (AromaScan A32S)

for analysis

Table 17.1 Hedonic tone odor rating scales

Scale Odor Intensity Irritation Intensity Pleasantness

8 Maximal Maximal Extremely Unpleasant

7 Very Strong Very Strong Very Unpleasant

6 Strong Strong Moderately Unpleasant

5 Moderately Strong Moderately Strong Slightly Unpleasant

4 Moderate Moderate Neutral

3 Moderately Weak Moderately Weak Slightly Pleasant

2 Weak Weak Moderately Pleasant

1 Very Weak Very Weak Very Pleasant

0 None at all None at all Extremely Pleasant

17.5 Case Study 3: Biofilter Evaluation [50] 433433

17.5.3.

Results

The average response of the human panel and the 32 conducing-polymer sensors inthe e-nose for each of the 15 dilutions (five dilutions for each of three odor sources) isshown in Fig. 17.4. Note that for the human panel, biofiltering reduced the intensity,irritation, and unpleasantness of the odor. In addition, the panel’s ratings of the bio-filtered slurry and blank air were quite similar.In order to establish whether the e-nose could be used to replace a human panel in

odor-remediation scenarios, we performed partial-least-squares regression [52] to pre-dict the average response of the human panel from the 32-dimensional average re-sponse of the e-nose. To establish the predictive accuracy of this model, we performedcross-validation in which one of the fifteen dilutions was removed from the trainingdata and predicted only after the partial-least-squares model had been obtained. Fig-ure 17.5 shows the performance of the model on test data for these fifteen leave-one-out validation runs. The correlation coefficient (between predictions and true values)on test data for intensity, irritation, and pleasantness are 0.90, 0.94 and 0.86, respec-tively.Given the notorious cross-sensitivity of conducting polymers to moisture, we

decided to analyze the response of the built-in humidity sensor of the AromaScanA32S to the different odors and dilution ratios. The transient response of odor and

Fig. 17.4 Average human and e-nose response versus dilution

number in the biofiltration experiment. The labels on the abscissa

for the serial dilutions are defined as follows: 5 (1/1 dilution),

4 (1/3 dilution), 3 (1/9 dilution), 2 (1/27 dilution), and 1 (1/81 dilution).

The human response sale is defined in Table 17.1. As expected,

both human and e-nose (AromaScan A32S) responses decrease

with increasing dilution


humidity sensors to the fifteen samples is shown in Fig. 17.6. Two observations can bemade. First, looking at the humidity sensor response to the slurry before and afterbiofiltration, it can be concluded that the biofilter material is increasing the relativehumidity of the samples. Second, as a result of serial dilutions, the humidity of thesamples is significantly reduced.On the basis of these results, it is necessary to determine if humidity is dominating

the e-nose response. A closer look at the data shows one that the response of the sensorarray to the synthetic slurry has a unique dynamic signature that is different from the

Fig. 17.5 True vs. predicted human panel ratings for intensity, irri-

tation, and pleasantness using the odor sensor array based on the

performance of the model on test data for the fifteen leave-one-out

validation runs. q ¼correlation coefficient

Fig. 17.6 Transient response of the gas sensor array and the humidity

sensor to five serial dilutions per odor using the AromaScan A32S. The

waveforms in both the upper and lower portions of the figure show the

time response of the odor and humidity sensors for each dilution

(labeled in the center of the figure). Note that the humidity sensor

response indicates that the biofilter material is increasing the relative

humidity of the samples. Serial dilutions with dry air reduce the hu-

midity of the samples

17.5 Case Study 3: Biofilter Evaluation [50] 435435

exponential decay to the biofiltered samples. This indicates that, in spite of relativehumidity changes, the odor sensors are able to detect the synthetic slurry. In addi-tion, if the odor sensors were responding only to the humidity, the largest responseof the sensor array would then occur with the 1/1 biofiltered blank since this samplehas the highest response on the humidity sensor.To further rule out the possibility that the e-nose is just detecting differences in

moisture, it was attempted to predict the human olfactory ratings from the humiditysensor response alone. The results are summarized in Fig. 17.7. The correlation coef-ficients between these single sensor predictions and true values by the human panelon test data for intensity, irritation, and pleasantness drop down to 0.40, 0.31 and 0.29,respectively. Hence, the conducting-polymer sensor array is giving much better per-formance, proving that the response of the odor sensors contains information relatedto the presence of synthetic slurry.

17.5.4

Discussion

The main findings of this study are that the AromaScan A32S can differentiate be-tween different dilutions of the components of swine odor, and between syntheticslurry and biofiltered slurry/blank samples. The sensor array response can be usedto predict the intensity and pleasantness olfactory ratings from a human panel. Moist-ure is shown to be a major interferent since biofiltration increases the relative humid-ity of the samples. However, the signal processing routines were able to mediate thisinterference. In the future, this interference might be reduced further by performingserial dilutions with a carrier gas having the same relative humidity as the odor sam-ples.

Fig. 17.7 True versus predicted human panel ratings using only the

humidity sensor. The correlation coefficients between the true and

predicted values for intensity, irritation, and pleasantness are reduced

compared with those in Fig. 17.5, thus the conducting-polymer sensor

array gives much better performance than the humidity sensor alone


17.6

Case Study 4: Mold Detection [53]

17.6.1

Background

Microbial contamination of our environment is an area of increasing concern. An e-nose has the potential to identify and classify microorganisms, including bacteria andfungi. When conditions are favorable and a nutrition source is present, microbialorganisms such as fungi and bacteria can grow almost anywhere. Microorganismshave been shown to generate VOCs while metabolizing nutrients, and these VOCshave been used as indicators of microbial growth. Colonies of microorganisms notonly generate airborne contamination in the form of VOCs, but also generate tox-ins, conidia (spores), and bacterial cells.When microoganisms infest buildings, they can produce a potentially hazardous

environment. Individuals exposed to environments that contain high concentrationsof airborne contaminants from microbial organisms report health symptoms includ-ing eye and sinus irritation, headaches, nausea, fatigue, congestion, sore throat, andeven toxic poisoning. Sick-building syndrom, which includes health symptoms arisingfrom poor indoor air quality, has been correlated with the presence of fungi [54]. Astudy of two housholds reporting indoor environmental complaints correlated thepresence of excessive VOCs with the presence of fungal contamination [55]. Typicalsigns of microbial contamination include water damage, high levels of humidity, andphysical presence. However, these signs are not always present, and therefore cannotbe utilized as sole indicators of microbial contamination.Current methods for detecting microbial contamination include air and material

sampling with culture analysis, air sampling coupled with gas chromatography/mass spectrometry, and visual inspection [56, 57]. These methods, however, can beinconclusive as well as time consuming and expensive. Thus, rapid detection ofthe presence of microbial contamination is needed in order to minimize its impact.

17.6.2


In this study, we explored the ability of the NC State E-Nose, a prototype electronicsystem with 15 metal-oxide sensors, to detect fungi at various stages of growth. Fungithat are typically found in indoor air-conditioning systems were chosen for experimen-tation. The purpose of the experiment was to demonstrate that an e-nose system iscapable of diagnosing the presence of these fungal types in commercial buildingsand residential housing units.

17.6.3

The NC State E-Nose

An e-nose instrument was designed and constructed at North Carolina State Univer-sity [44, 49] that uses an array of metal-oxide sensors for measuring odor in air samples

17.6 Case Study 4: Mold Detection [53] 437437

(see Fig. 17.8). The e-nose consists of a sampling unit, a sensor array, and a signalprocessing system. The sampling unit, which consists of a pump and a mass-flowcontroller, directs the air sample containing the odorant under investigation acrossthe sensor array. The current configuration allows for sampling from a set of 12 odor-ants, a reference sample (filtered odorless dry ambient air), and a washing agent (am-bient air bubbled through a 2% n-butanol solution). The tubing and sensor chamberare made of stainless steel. The sensor chamber is designed to minimize dead volume(see Fig. 17.9). The sensor array is composed of 15 different metal-oxide sensors.Twelve of the 15 metal-oxide sensors are manufactured by Capteur (Didcot, UK)and include sensors for isopropyl alcohol, toluene, hydrogen sulfide, nitrogen diox-ide, chlorine, butane, propane, hydrogen, carbon monoxide, heptane, ozone, and gen-eral VOCs. The remaining three metal-oxide sensors are produced by Figaro USA(Glenview, IL) and include methane, a combustible gas, and a general air-contami-nant sensor. All of the sensor response patterns are digitized and recorded using aNational Instruments Data Acquisition Card controlled by LabVIEWJ.The solenoid valves are normally closed. Solenoid valve s1 (exhaust) and an appro-

priate inlet solenoid valve (s2 to s15) are opened at the beginning of each phase andclosed afterwards. The mass flow controller must also be set at the beginning of

Fig. 17.8 System configuration for the NC State E-Nose. The exhaust

pump pulls air samples through the system. The mass flow controller

(MFC) and exhaust pump can be separated from the system by solenoid

valve S1. The system has 14 sample input ports controlled by solenoid

valves S2 to S15. Ports S2 and S3 are assigned the washing (cleaning) and

reference functions, respectively. Ports S4 through S15 are designated as

odor sample handling inputs. The system includes an inline pressure

sensor, a combined temperature/humidity sensor, and 15 metal-oxide

odor sensors


each cycle to the appropriate set point (between 0.0 and 1.0 Lmin�1). The operationcycle for the NC State E-Nose consists of three phases: wash, reference, and sample.Wash phase: solenoid valves s1 and s2 are opened. Room air is passed through a

charcoal filter (to remove residual ambient odors) and a bubbler with 2% dilutedn-butanol in distilled water. The resulting gas is used to flush tubing and sensorsand remove traces of odorants from previous gas samples.Reference phase: solenoid valves s1 and s3 are opened. Room air is passed through a

charcoal filter (to remove residual odors) and a moisture trap. The resulting odor-free

Fig. 17.9 The sensor chamber of the NC State E-nose. (a) airflow pattern;

(b) photograph. Commercially available metal-oxide sensors are mounted in

a stainless steel chamber. The electrical leads of the sensors are soldered to

printed circuit boards with attached ribbon cables that relay the sensor responses

to interfacing electronics. From the top of the chamber, air enters a cylindrical

tube with holes that ‘jet’ the odor samples directly onto each odor sensor. After

passing over the sensors, the air streams merge and exit the chamber

17.6 Case Study 4: Mold Detection [53] 439439

dry air is used as a reference gas to force the sensor resistances back to their baselinevalues.Sample phase: solenoid valve s1 and one other valve (s4 to s15) are opened. The odor-

ous sample is passed through the e-nose. Return to Wash phase.

17.6.4

Methods

Five fungi (Aspergillus flavus, Aspergillus niger, Penicillium chrysogenum, Cladosporiumcladosporioides, and Stachybotrys chartarum) were incubated at 28 8C on 150-mm dia-meter Petri dishes containing potato dextrous agar (PDA), a complex media rich innutrients, and Czapek-Dox agar (CZ), a minimal media. These two types of mediawere used in order to provide two different growth environments and to produce dif-ferent growth rates. Twenty-four Petri dishes of each media were inoculated with0.5 mL of an individual spore suspension containing 10 000 condia mL�1 fromeach fungus, respectively. The suspensions were prepared using a Spencer hemacyt-ometer with improved Neubauer ruling. Using the autosampler functions of the NCState E-Nose, air samples from the headspace of each Petri dish containing one specieson each medium were randomly sampled ten times, each after 24 hours and everyother day thereafter for two weeks. The headspace above each fungus was sampledthrough a small hole in the center of the lid of the Petri dish using a PVC tubeand an inline 2-lm filter for removing conidia (spores).The data were analyzed with MATLABJ using signal-processing algorithms devel-

oped by Kermani [44] and Gutierrez-Osuna [49]. More specifically, the raw data werefirst compressed using windowing functions that produced a set of four features foreach sensor. Linear-discriminant analysis was then applied to the compressed data tomaximize class separability. Sixty percent of the compressed data was randomly se-lected to form a training set for the classification algorithms. K-nearest-neighbors(KNN) and least-squares (LS) techniques were both employed to classify the remain-ing 40% of the compressed data [58]. This process was repeated 100 times, and theaverage score was used as the final classification score.

17.6.5

Results

The data were analyzed using two classification protocols. In the first protocol, the datawere grouped into 12 classes: five fungal species grown on PDA and CZ, respectively,plus two controls (the two media PDA and CZ without fungal growth). The results areshown in Table 17.2. After 24 hours of growth, the percent classification was 90% forKNN, and 76% for LS. Classification for the 12 classes reached a maximum after fivedays of growth, with an accuracy of 96% for KNN and 94% for LS. After day 5, thepercent classification began to decrease slowly. By day 15, the percent classificationwas reduced to 89% for KNN and 69% for LS.


In the second classification protocol, the data were grouped into seven classes: fivefungal species (independent of media used for growth) plus two controls (the twomedia PDA and CZ without fungal growth). In other words, each of the fungi grownin PDA and CZ were combined into a single class. After 24 hours of growth, the per-cent classification was 89% for KNN, and 79% for LS. Classification reached a max-imum after five days of growth, with an accuracy of 94% for KNN and 93% for LS.After day 5, the percent classification oscillated around an average percent classifica-tion of 92% with a standard deviation of 2%. The results are shown in Table 17.3.

17.6.6

Discussion

The experiment with five fungi showed that the NC State E-Nose using metal-oxidesensors can detect and classify microorganisms on the basis of volatile emissions. Theclassification was independent of the media used to grow the fungi. Furthermore,correct classification was achieved early in the experiment at 24 hours of growth.Thus e-nose instruments of this type have the potential to be used for early detectionof microbial contamination in office buildings and manufacturing facilities.

17.7

Future Directions

The success of laboratory instruments in classifying environmental odors has beendemonstrated by many research groups around the world. This success must nowbe leveraged to build new portable instruments for field use. These portable unitsmust operate in real time, recording odor concentration profiles at specific time inter-vals tailored to individual environmental monitoring applications. These devices must

Table 17.2 Percent classification for 12 classes (five fungal species on

two different media and two control media) [53]

Classification Method Day of Growth

1 3 5 7 9 11 13 15

KNN 90% 91% 96% 94% 89% 93% 93% 89%

LS 76% 90% 94% 90% 93% 86% 80% 69%

Table 17.3 Percent classification of seven classes (five fungal species and two control media) [53].

Classification Method Day of Growth

1 3 5 7 9 11 13 15

KNN 89% 90% 94% 93% 89% 94% 94% 92%

LS 79% 88% 93% 91% 95% 90% 92% 86%


be able to detect odors at very low (parts per billion) levels. Hence, more sensitive gassensors and preconcentration units must be included in instruments that will be usedin on-site, real-time environmental measurements. Chapters 7 and 9 have illustratedsome progress by the instrumentmakers towards reaching these goals. Improvementsin signal-processing algorithms can offer some assistance. Low-power, embedded mi-croprocessors are continually being improved by the electronics industry. Incorporat-ing more powerful real-time data-processing algorithms onboard these portable in-struments will differentiate the different commercial models. If the e-nose manufac-turers can ‘break’ into the environmental monitoring market in a significant way, thefuture of this technology will be guaranteed.

Acknowledgements

The authors wish to acknowledge the support of the National Science Foundation, theUS Agricultural Research Service, the National Pork Producers Council, the NC StateUniversity Animal and Poultry Waste Management Center, and the Center for IndoorAir Research for supporting various portions of the work reported herein.

References

1 R. E. Baby, M. Cabezas E. N. W. de Reca.Electronic nose: a useful tool for monitoringenvironmental contamination. SensorActualB-Chem 69 (3): 214–218 OCT 25 2000.

2 T. Dewettinck, K. Van Hege, W. Verstraete.The electronic nose as a rapid sensor forvolatile compounds in treated domesticwastewater. Water Res 35 (10): 2475–2483JUL 2001.

3 K. Van Hege, T. Dewettinck, W. Verstraete.Pre-evaporative Fenton remediation oftreated municipal wastewater for reusepurposes. Environ Technol 22 (5): 523–5322001.

4 F. Di Francesco, B. Lazzerini, F. Marcelloni,G. Pioggia. An electronic nose for odourannoyance assessment.Atmos Environ 35 (7):1225–1234 2001.

5 R. A. Fenner, R. M. Stuetz. The applicationof electronic nose technology to environ-mental monitoring of water and wastewatertreatment activities.Water Environ Res 71 (3):282–289 1999.

6 R. M. Stuetz, R. A. Fenner, G. Engin. As-sessment of odours from sewage treatmentworks by an electronic nose, H2S analysisand olfactometry.Water Res 33 (2): 453–4611999.

7 R. M. Stuetz, S. George, R. A. Fenner,S. J. Hall. Monitoring wastewater BODusing a non-specific sensor array. J ChemTechnol Biot 74 (11): 1069–1074 1999.

8 W. Bourgeois, R. M. Stuetz. Measuringwastewater quality using a sensor array:prospects for real-time monitoring.Water Sci Technol 41 (12): 107–112 2000.

9 R. M. Stuetz, R. A. Fenner, S. J. Hall. et al.Monitoring of wastewater odours usingan electronic nose. Water Sci Technol 41 (6):41–47 2000.

10 W. Bourgeois, J. E. Burgess, R. M. Stuetz.On-line monitoring of wastewater quality:a review. J Chem Technol Biot 76 (4):337–348 2001.

11 R. Stuetz. Using sensor arrays for on-linemonitoring of water and wastewater quality.Am Lab 33 (2): 2001.

12 C. Di Natale, A. Macagnano, F. Davide,A. D’Amico, A. Legin, Y. Vlasov,A. Rudnitskaya, B. Selezenev. Multi-component analysis on polluted watersby means of an electronic tongue. SensorsActuat B-Chem 44 (1–3): 423–428 1997.

13 J. W. Gardner, H. W. Shin, E. L. Hines,C. S. Dow. An electronic nose system formonitoring the quality of potable water.Sensors Actuat B-Chem 69 (3): 336–34 2000.


14 H. W. Shin, E. Llobet, J. W. Gardner,E. L. Hines, C. S. Dow. Classification ofthe strain and growth phase of cyanobacteriain potable water using an electronic nosesystem. IEE P-Sci MeasTech 147 (4):158–164 2000.

15 E. M. Biey, W. Verstraete. The use of aUV lamp for control of odour decompositionof kitchen and vegetable waste. EnvironTechnol 20 (3): 331–335 1999.

16 P. E. Keller, R. T. Kouzes, L. J. Kangas.Three Neural Network Based SensorSystems for Environmental Monitoring.IEEE Electro 94 Conference Proceedings,Boston, MA, 377–382 1994.

17 C. Mouche. Electronic nose sniffs out,classifies contamination. Pollut Eng 31 (2):31 1999.

18 J. Lee, J. Stewart. “Omonos: A ComputerModel for the Dispersion of Odours in Air”in Clean Air and Environmental Protection,Vol 29, No. 5, Published by the NationalSociety for Clean Air, Brighton, England,140–144 1999.

19 S. S. Schiffman, E. A. Satterly-Miller,M. S. Suggs, B. G. Graham. The effect ofenvironmental odors emanating from com-mercial swine operations on the mood ofnearby residents. Brain Res Bull 37: 369–3751995.

20 P. J. Hobbs, T. H. Misselbrook,M. S. Dhanoa, K.C. Persaud. Developmentof a relationship between olfactory responseand major odorants from organic wastes.J Scs Food Agr 81 (2): 188–193 2001.

21 P. Gostelow, S. A. Parsons, R. M. Stuetz.Odour measurements for sewage treatmentworks. Water Res 35 (3): 579–59 2001.

22 R.M. Stuetz, G. Engin, R. A. Fenner. Sewageodour measurements using a sensory paneland an electronic nose. Water Scs Technol 38(3): 331–335 1998.

23 R. M. Stuetz, R. A. Fenner, G. Engin.Characterisation of wastewater using anelectronic nose. Water Res 33 (2): 442–4521999.

24 F. W. Schreiber, K. Fitzner. Electronic Nose:Investigation of the Perceived Air Quality inIndoor Environments, Indoor Air 99, Vol. 2,624–629 Edinburgh, 8–13.08.1999.

25 F. W. Schreiber, K. Fitzner. Investigation ofthe Perceived Air Quality in an OfficeBuilding with an Electronic Nose, HealthyBuildings 2000, Helsinki, 6–10.08.2000.

26 C. Delpha, M. Siadat, M. Lumbreras.Discrimination of a refrigerant gas in ahumidity controlled atmosphere by usingmodelling parameters. Sensors ActuatB-Chem 62 (3): 226–232 2000.

27 C. Delpha, M. Siadat, M. Lumbreras.An electronic nose for the identificationof Forane R134a in an air-conditionedatmosphere. Sensors Actuat B-Chem 69 (3):243–247 2000.

28 F. Sarry, M. Lumbreras. Gas discriminationin an air-conditioned system. IEEE T InstrumMEAS 49 (4): 809–812 2000.

29 O. Ramalho. Correspondences betweenolfactometry, analytical and electronic nosedata for 10 indoor paints.Analysis 28 (3):207–215 2000.

30 R. Feldhoff, C. A. Saby, P. Bernadet. Dis-crimination of diesel fuels with chemicalsensors and mass spectrometry based elec-tronic noses. Analyst 124 (8): 1167–11731999.

31 R. J. Lauf, B. S. Hoffheins. Analysis of liquidfuels using a gas sensor array. Fuel 70 (8):935–940 1991.

32 R. Menzel, J. Goschnick. Gradient gassensor microarrays for on-line processcontrol – a new dynamic classificationmodelfor fast and reliable air quality assessment.Sensors Actuat B-Chem 68 (1–3): 115–1222000.

33 A. R. Hopkins, N. S. Lewis. Detection andclassification characteristics of arrays ofcarbon black/organic polymer compositechemiresistive vapor detectors for the nerveagent simulants dimethylmethylphospho-nate and diisopropylmethylphosponate.Anal Chem 73 (5): 884–892 2001.

34 G. Hudon, C. Guy, J. Hermia. Measurementof odor intensity by an electronic nose.J Air Waste Manage 50 (10): 1750–17582000.

35 R. M. Negri, S. Reich. Identification ofpollutant gases and its concentrations with amultisensor array. Sensors Actuat B-Chem 75(3): 172–178 2001.

36 B. Wessen, K.-O. Schoeps. Microbial volatileorganic compounds – what substances canbe found in sick buildings? Analyst 121:1203–1205 1996.

37 A.-L. Sunesson. et al. Identification ofvolatile metabolites from five fungal speciescultivated on two media. Appl EnvironMicrobiol 61: 2911–2918 1995.


38 M. Holmberg. Data Evaluation for anElectronic Nose. Dissertation, Depart. Phys.Meas. Tech. Linkoping University, Sweden,1997.

39 J. W. Gardner, M. Craven, C. Dow, E. L.Hines. The prediction of bacteria type andculture growth phase by an electronic nosewith a multilayer perceptron network. MeasSci Tech 9: 120–127 1998.

40 V. Vernat-Rossi, C. Garcia, R. Talon, C. Y.DeLayer, J. L. Berdague. Rapid discrimina-tion of meat products and bacterial strainsusing semiconductor gas sensors. SensorsActuat B-Chem 37: 43–48 1996.

41 G. Keshri, N. Mayan, P. Voysey. Use of anelectronic nose for the early detection anddifferentiation of spoilage fungi. Lett ApplMicrobiol 27: 261–264 1998.

42 J. Nicolas, A. C. Romain, V. Wiertz,J. Maternova, P. Andre. Using the classifi-cation model of an electronic nose to assignunknown malodours to environmentalsources and to monitor them continuously.Sensor Actuat B-Chem 69 (3): 366–371 2000.

43 A. Perera, T. Pard, T. Sundic, S. Marco,R. Gutierrez-Osuna. “IpNose: Electronicnose for distributed air quality monitoringsystem,” in Proceedings of the 3rd EuropeanCongress on Odours, Metrology and Elec-tronic Noses, Paris, France, June 19–21,2001.

44 B. G. Kermani. On using artificial neuralnetworks and genetic algorithms to optimizeperformance of an electronic nose. Ph.D.Dissertation, Department of Electrical Eng-ineering, North Carolina State University,Raleigh, NC, 1996.

45 K. Fukunaga. Introduction to statisticalpattern recognition, 2nd Edition, AcademicPress Inc., San Diego, CA, 1992.

46 K. A. Levenberg. A method for the solutionof certain non-linear problems in leastsquares, Quart Appl Math 2: 164–168, 1944.

47 D. Marquardt. An algorithm for least squa-res estimation of non-linear parameters, JSoc Ind Appl Math 11: 431–441, 1963.

48 D. E. Rumelhart, J. L. McClelland. ParallelDistributed Processing, MIT Press, Cam-bridge, MA, p.318, 1986.

49 R. Gutierrez-Osuna. Signal processing andpattern recognition for an electronic nose.Doctoral Dissertation, Department ElectricalComputer Engineering, North CarolinaState University, USA, 1998.

50 R. Gutierrez-Osuna, S. S. Schiffman,H. T. Nagle. “Correlation of SensoryAnalysis with Electronic Nose Data forSwine Odor Remediation Assessment,” inProceedings of the 3rd European Congresson Odours, Metrology and Electronic Noses,Paris, France, June 19–21, 2001.

51 K. C. Persaud, S. M. Khaffaf, O. J. Hobbs,R. W. Sneath. Assessment of conductingpolymer odour sensors for agriculturalmalodour measurements, Chemical Senses21: 495–505 1996.

52 P. Geladi, B. R. Kowalski. Partial least-squares regression: A tutorial. Anal ChimActa 185: 1–17, 1986.

53 S. S. Schiffman, D. W. Wyrick, R. Gutierrez-Osuna, H. T. Nagle. “Effectiveness of anelectronic nose for monitoring bacterial andfungal growth.” in: Gardner JW, PersaudKC. Electronic Noses and Olfaction 2000,Bristol: Institute of Physics Publishing,2000, pp. 173–180.

54 D. G. Ahearn. et al. Fungal colonization offiberglass insulation in the air distribution ofamultistory office building: VOC productionand possible relationship to sick buildingsyndrome. J Indust Microbiol 16: 280–2851996.

55 G. Strom. et al. Health Implications of Fungiin Indoor Environments, Elsevier, Amster-dam, 291–305, 1994.

56 S. S. Schiffman, J. L. Bennett, J. H. Raymer.Quantification of odors and odorants fromswine operations in North Carolina. AgForest Meteor 108: 213–240, 2001.

57 A. L. Pasanen. et al. Occurrence and mois-ture requirements of microbial growth inbuildings. Int Biodeter Biodegrad 30: 273–283 1992.

58 R. O. Duda, P. E. Hart. Pattern classificationand Scene Analysis, Wiley, New York, 1973.


18

Medical Diagnostics and Health Monitoring

Krishna C. Persaud, Anna Maria Pisanelli, Phillip Evans

18.1

Introduction

Many diseases and intoxications are accompanied by characteristic odors, and theirrecognition can provide diagnostic clues, guide the laboratory evaluation, and affectthe choice of immediate therapy [1–4]. Common observations are the change inbreath odor profile in diabetic patients entering a ketotic state, while the profilesof urinary volatiles from patients with phenylketonuria, maple syrup urine disease,isovaleric acidemia, or trimethylaminuria (fish-odor syndrome) are vastly differentfrom the normal urinary volatiles profile [5]. It is also recognized that many bacteriagrowing on specific media produce characteristic odorous metabolites, and that thesecan be used to diagnose which bacteria species are present in a culture [6]. The realiza-tion that electronic nose technologies may be a useful diagnostic aid has spurred ac-tivity in many research laboratories and companies, one of the earliest clinical trials ofthe technology reported being detection of infections in leg ulcers in patients in 1995[7]. This chapter reviewsmajor activity in the field (see Table 18.1), and then focuses onselected investigations in the area of myopathies and in bacterial vaginosis (BV), toprovide perspective on measurement and sampling requirements for applicationsof electronic noses in clinical measurements and diagnosis.Medical and health-monitoring applications are often cited in the electronic nose

literature. However, converting these potential markets to commercial reality hasyet to be achieved. There are numerous reasons for this, not least being the require-ments for robustness when dealing with the health of a patient, mistakes could becostly for all concerned. There is also the dichotomy between the ability to performthe measurement and the need for measurement. An example might be the case ofmaple syrup urine disease where urine takes on the consistency of maple syrup; thisalone is a reasonably good diagnostic marker so the knowledge that the urine has theodor of burnt sugar and fenugreek [8] is probably redundant.Oral malodor has long been cited as a potential application, having the advantage

that an incorrect diagnosis is unlikely to lead to death of the patient. This is not to saythat serious disease is not detectable by oral malodor [9]. Lung cancer, peritonsillar


445445

Tab. 18.1 Potential electronic nose applications in the medical/healthcare field.

Intended use Author(s)/

References

Sensors employed Algorithms used Sample handling Findings

Breath

monitoring

Halimeter Direct sampling

Cell growth [13] Direct headspace

mass spectrometry

Pirouette� v2.7

(PCA)

Cultures in vials,

headspace by

dedicated

autosampler

Growth phases of

E. coli defined by

changes in volatile

composition

Eye infection [14] Polymer/Carbon

black composite

PCA, FCM, SOM,

MLP, and RBF

Cultures in vials,

handheld sampling

Comparison of data

processing algo-

rithms found RBF

and MLP to be

most applicable

General medical/

healthcare

[42] Karlsruhe

microarray

(KAMINA)

Linear discriminant

analysis (LDA)

Direct Sweat sampling may

be useful in diagno-

stic applications

Medical

environmental

monitoring (e.g. sick

building syndrome)

[15] 15 metal oxide

sensors

LDA, least squares

(LS) and nearest

neighbor neural

network (KNN)

Direct sampling

above pure cultures

Discrimination

between the fungi

was achieved along

with discrimination

between levels

of characteristic

volatiles

Respiratory tract

(e.g. tuberculosis)

[43] MOSES

II þ amperometric

sensors

PCA Headspace

sampling

Discrimination

achieved of

M. tuberculosisfrom controls

Diabetes [9] 2 element MOS Non-supervised

fuzzy clustering

Direct sampling

from patients

expired breath

Discrimination of

diabetics from a

normal population

Breath alcohol [44] 10 MOSFET and

one IR CO2 sensor

Partial least

squares (PLS)

and artificial

neural network

(ANN)

Forced exhalation

into bags followed

by sampling

Evaluation of the

requirements of such

a system for forensic

acceptability of breath

alcohol measure-

ments using an

electronic nose-type

setup

Leg ulcers [7] 20 conducting

polymers

PCA Sampling of leg

ulcer dressings

(presence of

b-haemolyticstreptococci)

Demonstrated

feasibility of the

approach

Cultured bacteria [45] 16 Conducting

polymers

(Bloodhound)

ANN, PCA Headspace from

12 bacteria and

1 yeast

Good discrimination

achieved

18 Medical Diagnostics and Health Monitoring446

abscess, and cancer of the larynx may all manifest themselves via oral malodor. How-ever, despite a great deal of funding, a successful breath odor device has yet to reach aclinic. Perhaps the principal reason for this is the suite of volatiles produced by the,typically, anaerobic bacteria causing malodor such as hydrogen sulfide, sulfur dioxideand methyl and dimethyl disulfide. The Halimeter system (Halimeter Interscan Inc.,Chatsworth Ca, USA) does measure low parts per billion levels of hydrogen sulfide butis prone to several interferences such as ethanol, essential oils, perfumes andmouthwashes. Sulfur compounds have incredibly low human olfactory thresholdsmeaning that most people would become aware of the odor far quicker than thebest of the sensing systems available. Coupled with this are the vast array of variablesthat need to be compensated for before an accurate measurement may be made; pre-sence of environmental contaminants, patient to patient variability, perfumes, food-stuff in the oral cavity, hunger, tiredness etc. The oral malodor model features someimportant rules for the investigator intomedical and health monitoring applications ofelectronic nose/sensor systems. A well-defined and controlled symptom is highly de-sirable. Phenomena such as bad-breath have ill-defined sources and as such are diffi-cult to define sufficiently. This is especially significant when a volatile or combinationof volatiles may characterize one or more phenomena.The use of smell in medical diagnostics and the development of systems for evalua-

tion of odor in a medical context have been reviewed by Pavlou and Turner [10]. Thisarticle also provides a description of various odors associated with disease such as astale beer odor on skin associated with tuberculosis and burnt sugar smells in urineassociated with maple syrup urine disease.Hanson and Thaler patented a system based on an AromaScan A32S system for the

monitoring of patients with lung infections such as pneumonia. The patent also dis-cusses the use of the system in evaluation of fluid samples from the sinus or nose forpresence of cerebrospinal fluid [11]. The authors expand the cerebrospinal fluid workfurther suggesting that electronic nose technology may be used to distinguish cere-brospinal fluid from serum, having applications in the diagnosis of otorrhea or rhi-

Tab. 18.1 Continued

Intended use Author(s)/

References

Sensors employed Algorithms used Sample handling Findings

Cultured bacteria [46] ANN, Feature

extraction

Petri dishes of

Escherichia coli,Enterococcus sp.,Proteus mirabilis,Pseudomonasaeruginosa,Staphylococcussaprophytica

76% classification

Cultured bacteria [47] 6 MOS

(Neotronics)

ANN Headspace from

Escherichia coli Sta-phylococcus aureus

Discrimination and

prediction of growth

phase achieved

Estrus in cows [18] Conducting polymer Wavelet analysis Swab in chamber Initial investigation


norrhea, and may have further application in the field of otorhinolaryngology [12].A novel detection means for diabetes detection based upon measurement of breathsamples for acetone using a two detector metal oxide system is suggested by Ping [9].Experiments describing responses before and after eating suggested good correlationbetween acetone concentration and diabetes. Paulsson et al. describe a breath alcoholanalysis using metal oxide field-effect transistor (MOSFET) sensor technology [44]. Aspart of their evaluation they considered the requirements of applying such a system inthe routine use of breath alcohol detection from a forensic standpoint.Changes in the odor of sweat has been proposed as a potential means of disease

diagnosis using the KAMINA system [42]. Mantini et al. also present a study of sweatas a potential means of following the menstrual state of women, although the studywas merely a demonstration of the idea rather than a clinical study. They also brieflydescribe an approach to a skin-sampling methodology and the evaluation of urinesamples containing blood [48].Esteves et al. [13] describe an investigation of the growth characteristics of Escher-

ichia coli using the Agilent headspace sampling system. The authors present datashowing how distinct growth phases may be monitored using principal componentanalysis (PCA) of selected portions of the mass spectrum acquired. It is suggestedthat the lower-molecular-weight fragments are more indicative of the growth phase(from cellular metabolism) whilst higher molecular weight fragments are derivedfrom cellular components especially when higher sampling temperatures wereused. The application of the Cyrano Sciences handheld electronic nose to the detec-tion of bacteria implicated in eye infections was reported by Boilot et al. [14]. Thebacteria investigated were E. coli, Staphylococcus aureus, Haemophilus influenzae, Strep-tococcus pneumoniae, Pseudomonas aeruginosa, and Moraxella catarrhalis. Simple PCAsuggested broad discrimination between the six bacteria grown in culture and pre-sented at various colony counts (discrimination based upon bacterial count wasnot however, reliably achieved). Further off-line analysis was then undertaken usinga number of data-processing strategies; (PCA), fuzzy c-means (FCM), self-organizingmaps (SOM), multi layer perceptron (MLP), radial basis function neural networks(RBF) and the fuzzy ARTMAP (adaptive resonance theory mapping) paradigm. Com-parisons on the usefulness of all of the approaches were made, with MLP and RBFalgorithms being most useful overall. Significant development was cited as being ne-cessary however, before a system could be developed for a truly near-patient system tobe developed. Schiffman et al. described the use of a MOS-based system for the dis-crimination of cultures of common fungi (Aspergillus flavus, A. niger, Penicillium chry-sogenum, Cladosporium cladosporoides and Stachybotrys chartarum) that are implicatedin sick-building syndrome, toxic poisoning, and allergic reactions [15]. Discriminationof cultured fungi was achieved along with discrimination between volatiles known tobe associated with the fungi (ethanol, 3-octanone, 3-octanol, 3-pentanone and 2-methyl-1-propanol).Dodd proposes the use of electronic noses as monitoring tools in conditions such as

schizophrenia [17]. This differs considerably from the detection of a pathogenic con-dition as described previously, with the author suggesting volatiles from autoxidation


of arachidonic acid might present a diagnostic marker monitorable via mass spectro-metry or electronic noses.Health monitoring is not exclusively used for humans – estrus in cows has also been

studied [18] using a modified Osmetech sensor system coupled to a custom built hu-midity compensation system “the olfactory lens” (a device for measuring dynamicchanges in order) using wavelet analysis to process the data.

18.2

Special Considerations in Medical/Healthcare Applications

Medical samples present all of the standard sample presentation problems and more.Chemical and food samples are relatively straight forward to analyze, providing theyare not subject to biological change i.e. that there are no overt degradation processesoccurring from sample to sample, or a characteristic off-odor or contaminant is pre-sent. In a similar field, food spoilage measurements also suffer from many of theeffects discussed below.One of the principal difficulties is the variability of the sample. This is especially true

if the samples are the patients themselves. Patient to patient variability is a huge factorin any sampling procedure. As described above for oral malodor measurements, anynumber of environmental and habitual factors can affect the measurement. Any ef-fective electronic nose application must either select out these unaccountable varia-tions or compensate for them by anticipating them. It is easy to envisage that thelatter approach is fraught with difficulty although it is given that no measurementis truly free from interferences. Hence, the more commonly encountered broad-se-lectivity electronic nose model is not the optimum system design.When developing any system and approach, the final application of the systemmust

be considered from the start. An at-patient system must be capable of being exactlythat, delivering a reliable and reproducible result within a typical consultation timewith the minimum of calibration and user expertise required.The presentation of the sample and its acquisition are critical parts of the process.

Developing a system for the discrimination of bacteria in culture, for example, is not aviable end product since standard culture techniques will take the same length of timeand produce equally valid results for less resource and probably higher reliability.Enhanced identification through the use of selective media might be a considera-tion, but this is equally achievable without resort to electronic nose technology, anti-biotic loaded culture plates for resistance checking are a simple example. Additionally,enhanced techniques such asmatrix assisted laser desorption/ionization time-of-flightmass spectrometry (MALDI-TOF MS) would give superior characterization over a si-milar time span.Consequently, at-patient or direct patient-derived samples with the minimum of

sample preparation are the most desirable approaches since they lend themselvesto rapid turnaround, even if the technique is simply a screen to eliminate negativesprior to further investigation. An example where this is an attractive option is inscreening for urinary tract infection, where, typically 60–80% of the presenting po-

18.2 Special Considerations in Medical/Healthcare Applications 449449

pulation is in fact negative. A technique for pre-screening samples prior to cultureautomatically saves time and frees resource. Further, in typical populations, 50 to80% of the infected population (i.e. between 20–40% of those initially presenting)is infected with E. coli. A system capable of screening out E. coli positives would offerenormous potential benefit in cost and time saving.Once a technique has been identified, a number of other factors arise. The robust-

ness of the technique and its performance at a clinical level must be considered. To beviable, any clinical/healthcare application would have to at least approach the perfor-mance of the current optimum methodology. As an example, a screen for urine in-fection would be no good no matter how quick it was if it was wrong 50% of the time.Other factors such as cross-contamination, sample reproducibility, user and patientsafety must also be considered.

18.3

Monitoring Metabolic Defects in Humans Using a Conducting Polymer Sensor Array toMeasure Odor

18.3.1

Background

The odor of the human body and excreted or secreted products of metabolism is re-lated to many complex factors associated with sex, age, genetics, diet, and metaboliccondition. In many cases, bacterial or viral infection, or metabolic diseases modifythese odors. Typical examples are bromidrosis in patients affected by rheumatismand uremia and diseases of the respiratory and digestive tract [19].Somemyopathies induce alterations in themetabolic pathway that cause anabnormal

secretionofmetabolites inblood as ketones andacids [20]. Thediagnosis of suchgeneticdiseases is based on gene analysis, muscle biopsy and testing muscle performance.Biochemical tests are carried out by enzyme analysis or by determining metabolites byHPLC gas chromatography coupled to mass spectrometry (GC-MS) or immunologicalmethods [21]. The main objectives of this research were: (a) to determine whether it ispossible to use an electronic nose as a diagnostic method for detection andmonitoringmetabolic diseases such as myopathies, (b) to carry out screening of samples frompatients and controls, using GC-MS to identify chemical species that could beused as markers, for which an electronic nose device could be focused. In the courseof this research we have been able to achieve the following breakthroughs in under-standing how to apply methods based on odor recognition to medical diagnostics.We have identified specific volatile chemical markers in the urine of patients with

specific metabolic disorders that are not present in controls, or are present at verydifferent concentrations.We have been able to discriminate populations of diseased persons from controls by

their odor fingerprint measured by an electronic nose, using urine samples. We haveapplied statistical and neural network methods to process data from such systems toenable the future on-line recognition of disease states.


18.3.2

Methodology

One useful set of materials that may be utilized as sensors in an electronic nose is thatof electrically conducting organic polymers based on heterocyclic molecules such aspyrroles, thiophenes and anilines. These display reversible changes in conductivitywhen exposed to polar volatile chemicals. Rapid adsorption and desorption kineticsare observed at ambient temperatures. The materials do not display high specificityto individual gases. However, they can be chemically tailored to enhance differences inresponse to particular classes of polar molecules. For single chemical species, theconcentration-response profiles can be fitted to Langmuir type adsorption models.This is advantageous as simple computational methods may be used for informationprocessing [22–24].Different polymers made from modified monomer units show broad overlapping

response profiles to different volatile compounds. Hence, arrays of these sensorsshould behave very similarly to olfactory sensor arrays in the biological system. Min-iature arrays consisting of up to 48 different conducting polymer materials have nowbeen realized by Osmetech plc (see Fig. 18.1). A microprocessor-driven circuit, mea-suring changes in resistances of individual sensor elements interrogates the sensorarray at user-defined intervals, and data are stored in memory. Each sensor elementchanges in resistance when exposed to a volatile compound. However, the degree ofresponse to a given substance depends on the type of polymer element used, so that apattern of resistance changes can be recorded and processed to produce a set of de-scriptors for that particular substance. The sensor responses are normalized to repre-sent relative changes in resistance and thus approximately concentration-independentpatterns can be produced. Taken over the whole array, there are enough statisticaldifferences for many compounds to be differentiated from each other. Functionalityof the system depends on devising robust computer programs that will allow the sys-tem to operate under adverse conditions whereby background odors may be present,

Fig. 18.1 Osmetech sensor

array and electronics

18.3 MonitoringMetabolic Defects in Humans using a Conducting Polymer Sensor Array to Measure Odor 451451

temperature and humidity may be cycling up and down and sensor-aging effects mayalso be interfering.For these experiments, a sensor array with 32 different conducting polymers (Os-

metech plc) was used for detecting odors from urine. A Hewlett Packard GS-MS(HP5890 GC/HP5971MS) apparatus was used for analyzing the volatiles and Supelcosupplied fibers for solid phase micro-extraction (SPME). Odors were commonly mon-itored by static headspace GC and occasionally by thermal desorption or purge and traptechniques.We opted for use of the SPMEmethod after testing static headspace, purge and trap,

and thermal desorption methods for sampling odors. SPME is a powerful techniquefor introducing analytes into a GC. The technique utilizes a 1 cm length of fused silicacoated with an adsorbent. The coated fused silica (SPME fiber) is immersed directlyinto an aqueous sample or into the headspace above a liquid or solid sample. Organiccompounds in the sample are subsequently adsorbed onto the fiber. Finally, the fiberis inserted into a GC injector where the analytes are thermally desorbed and separatedon the GC column. This technique is rapid and minimizes any sample manipulation.

18.3.3

Results

Replicate urine samples were taken from ten people affected by different musculardiseases and thirteen from healthy subjects over several days and frozen until theywere analyzed. An electronic nose system was used to analyze the headspace fromurine samples.To study the individual urine odor of a particular person, it is important to consider

their temporary differences, caused by different diet, state of health, physiologicalcondition etc. Thus urine samples were collected over a period of several days. Var-iance between urinary headspace of different individuals is significant, whereas for thesame individual the profile over different days remains constant, as measured usingthe electronic nose system. We analyzed urine headspace in a normal population aswell as in patients with myopathies by using the electronic nose and the GC-MS. Thepatterns obtained from the sensor array were recorded on a computer and stored forfurther processing. Urine samples collected from normal and diseased populationsgenerated patterns that slightly differed between each person and showed some var-iation due to the physiological condition and to the diet. In order to process the data weadopted the Sammon map method [25]. The Sammon non-linear mapping algorithmreduces multidimensional pattern space by mapping onto two-dimensional or three-dimensional pattern space based on a distance measure such as the Euclidean distanceand produces axes that are meaningful in terms of distances of one cluster from an-other. By using this method it was possible to differentiate the normal population fromthat with myopathies. Moreover we obtained subclusters within the population due toslight differences between them. This can be due to the different myopathies or degreeof the pathological status. Figure 18.2 shows the population distribution obtained be-tween controls and patients. Each point on themap represents an odor pattern reduced


to two dimensions, and clusters represent how close each odor pattern is to another inthe same area (the further away points are from each other, the greater the differencebetween them).The results obtained from the GS-MS analysis show that the composition of the

urine headspace is markedly different within normal and diseased populations.Key volatile components found in the profiles of normal urine were 2-heptanoneand 4-heptanone. The amount of these volatiles increases in urine samples from peo-ple affected with myopathies. Compounds such as 2(3H)phenanthrene-4-4a-9,10-ter-tahydro-4a-methyl and phenyl-isopropylphenyl ether are present in different quanti-ties only in urine from patients and not in normal controls.We performed the GC-MS analysis of urine in order to validate the results obtained

from the electronic nose. The different patterns obtained from the gas sensor appa-ratus are correlated with the different volatiles detected by the GC-MS, and their quan-tities. Knowing the composition of urine headspace will allow us to build specificsensors for diagnostic purposes.

Fig. 18.2 Analysis of urine headspace: the population distribution obtained

between controls and myopathic patients. The Sammon map represents in two

dimensions the averaged Euclidean distance between urine headspace patterns

for each individual tested, each point representing one individual. It is seen that

the majority of patients group together, and the controls also group together

separately, but there are two patients who group with the controls and one

control individual who groups with the patients

18.3 MonitoringMetabolic Defects in Humans using a Conducting Polymer Sensor Array to Measure Odor 453453

18.4

The Use of an Electronic Nose for the Detection of Bacterial Vaginosis

18.4.1

Background

Bacterial vaginosis (BV) is a particularly ill-defined phenomenon with uncertain symp-toms. Numerous reports [28–30] cite as much as 50% of the affected population beingasymptomatic. The consequence of this is that at time of presentation only 50% of thestory is known. The remaining 50% of the population either go undetected or presentduring routine examination for another associated or uncorrelated problem. Initialinvestigations were performed by Chandiok et al. [26] using a standard AromaScan(now Osmetech) system at Withington Hospital, Manchester, UK.The consequences of BV are wide and varied and are not completely understood.

This is understandable given the difficulties in getting reliable BV data for a popula-tion. The primary challenge facing any prospective diagnostic technique (or aid todiagnosis) is finding a unique indicator against which BV may be detected. Cur-rently, the Amsel test is the benchmark for determining the problem. The criteriafor the test rely on at least three out of four conditions being met [27]. These are:

* pH of vaginal fluid > 4.5;* Presence of a typical thin, homogenous vaginal discharge;* Release of strong fishy smell on addition of alkali (10%KOH) to a sample of vaginal

fluid (whiff test);* Clue cells present on microscopic examination of a wet mount of vaginal fluid.

Individually none of these tests are diagnostic. pH variation of the vaginal fluid isnearly always present in BV positive patients but it is a non-specific test and the varia-tion is equally likely to be caused by another infection or problem. Additionally, con-tamination of the sample by cervical mucus (typical pH 7) can lead to false diagnoses insome cases. pH variation also occurs as part of the natural menstrual cycle. Ethnicbackground is also a factor affecting vaginal pH and this has been used as a reasonfor the relatively higher number of black American women who present with symp-toms of the disease. According to Hay, pH is highly sensitive (97%) but very non-specific giving false positives in 47% of cases [28, 29].Conversely, discharge is very accurately recognized by clinicians giving false positivesat 3% but only has a specificity of 67%. Following this, the ‘whiff’ test also gives lowfalse positives (1%) but is non-specific (43%). Finally clue cells are typically found in81% of positive BV cases whereas 6% of non-BV cases have positive clue cell tests.Other trials report variation on these figures but all concur with the non-specificity andreliability of any one individual test 30].BV is commonly thought to arise as a result of fluctuation of the normal vaginal

flora. In some cases the flora can fluctuate naturally over the menstrual cycle withno adverse effects. It is thought that one of the primary controlling mechanisms con-trolling BV-causative bacteria is the presence of adequate colonies of Lactobacillus sp.


that produce hydrogen peroxide which limits the growth of anaerobes associated withBV. The most common organisms associated with BV are: Gardnerella vaginalis, Bac-teroides (Prevotella) spp., Mobiluncus spp. and Mycoplasma hominis. However, the pre-sence or absence of these flora is not reliably diagnostic.Treatment after diagnosis is usually quite effective and usually comprises oral doses

of metronidazole. Topical treatments withmetronidazole or clindamycin are also com-mon [31].Originally thought to be a benign infection, recent studies have linked the problem

to increased risk of:

* intra-amniotic infection [32]* choroamnionitis [33]* post-caesarean [34] and post-partum endometritis [35]* adverse pregnancy outcome [36]* pre-term labor [36–38]and birth [39]* premature rupture of membranes at term [40]* post-hysterectomy cuff cellulitis [41].

The data presented here are merely an overview and the reader is directed to the lit-erature cited for a more comprehensive discussion of the occurrence, diagnosis, andtreatment of this phenomenon.

Fig. 18.3 Example of BV swab in vial and presentation of

dual concentric needle sampling system

18.4 The Use of an Electronic Nose for the Detection of Bacterial Vaginosis 455455

18.4.2

Methodology

Patient samples were taken as swabs during normal examination. Swabs were weighedbefore and after sampling to determine the amount of sample collected during theexamination so that variation in sample collection may be evaluated. After collectionthe stem of the swab was cut off and the vial sealed with a standard septum and crimptop (see Fig. 18.3).After collection the vials may be stored for later analysis or analyzed immediately.

The sample ismounted on the carousel of the autosampler system and held at constanttemperature until its place in the sequence is reached. The sample is then lowered to apre-heated platen and its temperature stabilized for a predetermined period before thedual concentric needle is lowered into the vial through the septum and the dynamicheadspace extracted using a constant humidified gas flow. The headspace is trans-ferred across the Osmetech sensor array (see Fig. 18.1) where the signal is transducedand recorded for processing.

18.4.3

Results

Results can be produced from the Osmetech Microbial Analyser (OMA) within20 minutes (as can results from the Amsel test). The microbiology (Nugent score)results can take much longer and in some cases it can be five days before the resultsare transmitted back to the Genito-Urinary Medicine (GUM) clinic.The results from Table 18.2, which were derived from Fig. 18.4, give an overall sen-

sitivity of 89% and a specificity of 87% versus Amsel and Nugent scores with a ne-gative predict value of 96.8% and a positive predict value of 65%. Samples projectedon the PCA map labeled as Suspect BV were the result of indeterminate microbiologyand Amsel results (i.e. the two were not in agreement). As a result of the data availableduring the conduct of the clinical evaluation it was not possible to follow the suspectpatients up to confirm any further clinical diagnosis.

Tab. 18.2 Results from a clinical pre-trial carried out at the depart-

ment of Genito-Urinary Medicine, Withington Hospital, Manchester,

UK using the OMA instrument (BV ¼ bacterial vaginosis, STDs ¼ se-

transmitted diseases). The results in this table and Fig. 18.3 are for 89

newly registered non-pregnant females between the ages of 18 and 60.

Total Positives False positives Negatives False negatives

BV 16 15 1

Suspect BV 3 2 1

Yeasts 13 3 10

Negative 48 6 42

STDs 9 9


It is clear from the PCAmap presented that sexually transmitted diseases (STDs) arenot confused with BV status with all STDs projected in the negative BV sector. Aspreviously stated in the earlier discussions in this chapter it is highly desirablethat any clinical electronic nose application should display a high degree of selectivityto the target application. In the case of STDs this is clearly the case, although yeasts didhave a tendency to produce false positives (approximately 23% of yeasts analyzed pro-duced a false positive for BV). However, it is highly likely that the healthcare profes-sional carrying out the test would discriminate between yeast and BV before testing forBV using the OMA system.

18.4.4

Discussion

It is clear from the results presented in Fig. 18.4 and Table 18.1 that BV-positive pa-tients differ from the normal population when the data is processed as a PCAmap. Thedata can be seen to ‘branch’ into two categories away from the defined normal popula-tion. These two branches may be described by the use of two standards that are differ-entiated by means of sensor elements in the Osmetech array responding orthogonallyto the test chemicals used. These chemical standardsmay be used to define a PCAmaponto which the experimental data is processed. The threshold between positive andnegative results is subsequently defined by means of experimentally defined para-meters in the first instance, and then by adjustment of the chemical standards toreflect the threshold giving the clearest distinction between BV positive and negativepatients.

Fig. 18.4 PCA map of patient swabs from a GUM clinic, positives

are based upon combined agreement of Nugent and Amsel scoring

systems. Intermediate BV is assigned to tests where the Nugent

and Amsel scores do not concur


The use of standardized test chemicals is critical to the success of the procedure. Thestandards may be used to produce a master projection against which all subsequenttest chemical data may be projected. Subsequent system calibrations or checks maythen be projected against thismaster such that the current system performancemay becompared against the original blueprint. Significant change to the mapping of thestandards will flag substandard performance. Failure to pass this system check pro-cedure would prevent further use of the system until a suitable remedy has taken place.This imparts an inherent strength to the system allowing greater faith in the accuracyof any prediction made.Hence it can be seen that through the use of surrogate test chemicals a medical

device can be successfully used without resort to complicated drift correction andstandardization algorithms.

18.4.5

Conclusion

The OMA system in this case offers clear potential in the rapid diagnosis of BV. It canclearly compete with the established means of detection and in the case of Nugentscoring is a much more rapid technique. With further development it should provesuperior to the existing Amsel technique offering the advantages of ease of use andremoval of doubt from interpretative testing such as sniffing (the whiff test) and visualinspection (examination for the presence of clue cells).

18.5

Conclusion

It will be apparent from the information presented in this chapter that enormouspotential exists for the application of electronic nose technology in medical applica-tions. However, the field is still in the research and development stage, where clini-cally proven robust applications are still to come. There is now rapid growth in cap-ability of the technology and it is clear that many future diagnostic tools for selectedapplications will be available for physicians to utilize. Indeed, Osmetech plc has sub-mitted an application for approval of its urinary tract infection technology to the FDAafter a series of successful clinical trials.

Acknowledgements

This work was in part supported by Osmetech plc, Crewe, UK. AMPwas funded by theWellcome Trust for work on myopathies. We thank Dr. Ros Quinlivan, of OswestryHospital for co-operation and help with patient samples, Prof. Robert Beynon and Dr.Duncan Robertson for much help with GC-MS analysis.


References

1 G. F. Hayden. Postgraduate Medicine 1980,67(4), 110–5, 118.

2 W. Z. Stitt, A. Goldsmith. A. Archives OfDermatology 1995, 131(9), 997–999.

3 A. Zlatkis, R. S. Brazell, C. F. Poole. ClinicalChemistry 1981, 27(6), 789–797.

4 A. Zlatkis, C. F. Poole, R. Brazell, K. Y. Lee,F. Hsu, S. Singhawangcha. Analyst 1981,106(1260), 352–360.

5 D. G. Burke, B. Halpern, D. Malegan,E. McCairns, D. Danks, P. Schlesinger,B. Wilken. Clinical Chemistry 1983, 29(10),1834–1838.

6 P. Grametbauer, S. Kartusek, O. Hausner.Ceskoslovenska Epidemiologie, Mikrobiologie,Imunologie 1988, 37(4), 216–223.

7 A. D. Parry, P. R. Chadwick, D. Simon,B. Oppenheim, C. N. McCollum. JournalOf Wound Care 1995, 4(9), 404–406.

8 K. Monastiri, K. Limame, N. Kaabachi,H. Kharrat, S. Bousnina, H. Pousse,M. Radhouane, M. N. Gueddiche,N. Snoussi. Journal Of Inherited MetabolicDisease 1997, 20(4), 614–615.

9 W. Ping, T. Yi, H. B. Xie, F. R. Shen. Bio-sensors & Bioelectronics 1997, 12(9–10),1031–1036.

10 A. K. Pavlou, A. P. F. Turner. ClinicalChemistry and Laboratory Medicine 2000,38(2), 99–112.

11 C. W. Hanson, R. Thaler. 09965386 WO,1999.

12 E. R. Thaler, F. C. Bruney, D. W. Kennedy,C. W. Hanson. Archives of OtolaryngologyHead and Neck Surgery 2000, 126(1), 71–74.

13 R. Esteves de Matos, D. J. Mason, C. S. Dow,J. W. Gardner. Investigation of the GrowthCharacteristics of E. coli using HeadspaceAnalysis, in Electronic Nose and Olfaction2000, Gardner, J. W.; Persaud, K. C., editors;IOP Publishing: Bristol, UK, 2000;pp. 181–188.

14 P. Boilot, E. L. Hines, S. John, J. Mitchell,F. Lopez, J. W. Gardner, E. Llobet, M. Hero,C. Fink, M. A. Gonogora. Detection ofBacteria Causing Eye Infections using aNeural Network Based Electronic NoseSystem, in Electronic Nose and Olfaction 2000,Gardner, J. W.; Persaud, K. C., editors;IOP Publishing: Bristol, UK, 2000;pp. 189–196.

15 S. S. Schiffman, D. W. Wyrick, G. A. Payne,G. O’Brian, H. T. Nagle. DetectingMicrobialContamination using an Electronic Nose, inISOEN200 abstracts, Persaud, K. C.; Gard-ner, J. W., editors; ECRO, Indigo Lithoprint:Manchester, UK, 2000.

16 R. T. Marsili. Journal Of Agricultural AndFood Chemistry 1999, 47(2), 648–654.

17 G. H. Dodd. Prostaglandins, LeukotrienesEssential Fatty Acids 1996, 55(1 þ 2), 95–99.

18 T. T. Mottram, R. M. Lark, A. J. P. Lane,D. C. Wathes, K. C. Persaud, M. Swan,J. M. Cooper. Techniques to Allow theDetection of Oestrus in Dairy Cows with anElectronic Nose, in Electronic Nose andOlfaction 2000, Gardner, J. W.; Persaud,K. C., editors; IOP Publishing: Bristol, UK,2000; pp. 201–208.

19 M. Inaba, Y. Inaba. Human Body Odor.Etiology, Treatment and Related Factors.;Springer Verlag: Berlin, 1992.

20 H. Chen, F. Aiello. Amer. J. of Med Genetics1993, 45, 335–339.

21 M. A. Hollinger, B. Sheikholislam. TheJournal of International Medical Research1991, 19, 63–66.

22 P. Pelosi, K. C. Persaud. Gas sensors:Towards an artificial nose. In: Sensors andSensory Systems for Advanced Robots.,in NATO ASI Series F: Computer and SystemsScience, Dario P, editor; Springer-Verlag:Berlin, 1988; pp. 361–382.

23 K. C. Persaud. Analytical Proceedings(London) 1991, 28(10), 339–341.

24 K. C. Persaud. Trends in Analytical Chemistry1992, 11(2), 61–67.

25 J. W. Sammon Jr.. IEEE Transactions onComputers 1969, 5(C-18), 401–409.

26 S. Chandiok, B. A. Crawley, B. A. Oppen-heim, P. R. Chadwick, S. Higgins, K. C.Persaud. Journal Of Clinical Pathology 1997,50(9), 790–791.

27 R. Amsel, P. A. Totten, C. A. Spiegel,K. C. Chen, D. Eschenbach, K. K. Holmes.American Journal Of Medicine 1983, 74(1),14–22.

28 P. E. Hay, D. Taylor-Robinson, R. F. Lamont.British Journal Of Obstetrics And Gynaecology1992, 99(1), 63–66.

29 P. E. Hay. Dermatologic Clinics 1998 16(4),769–773.


30 J. L. Thomason, S. M. Gelbart,R. J. Anderson, A. K. Walt, P. J. Osypowski,F. F. Broekhuizen. American Journal OfObstetrics And Gynecology 162(1), 155–160.

31 P. E. Hay. Journal of Antimicrobial Chemo-therapy 1998, 41(1), 6–9.

32 D. H. Watts, M. A. Krohn, S. L. Hillier,D. A. Eschenbach. Obstetrics And Gynecology1990, 75(1), 52–58.

33 D. H. Watts, D. A. Eschenbach, G. E. Kenny.Obstetrics And Gynecology 1989, 73, 52–60.

34 M. G. Gravett, H. P. Nelson, T. DeRouen,C. Critchlow, D. A. Eschenbach,K. K. Holmes. JAMA 1986, 256(14),1899–1903.

35 S. Faro. Journal Of Reproductive Medicine1989, 34(8 Suppl), 602–604.

36 M. G. Gravett, D. Hummel, D. A. Eschen-bach, K. K. Holmes,. Obstetrics And Gyne-cology 1986, 67(2), 229–237.

37 J. A. McGregor, J. I. French, R. Richter,A. Franco-Buff, A. Johnson, S. Hillier,F. N. Judson, J. K. Todd. American JournalOf Obstetrics And Gynecology 1990,163(5 Pt 1), 1465–1473.

38 J. A. McGregor, J. I. French. Obstetrical AndGynecological Survey 2000, 55(5 Suppl 1),S1–19.

39 D. E. Soper, R. C. Bump, W. G. Hurt.American Journal Of Obstetrics And Gyneco-logy 1990, 163(3), 1016–1021.

40 C. A. Spiegel. Clinical Microbiology Reviews1991, 4(4), 485–502.

41 R. L. Cook, G. Reid, D. G. Pond, C. A.Schmitt, J. D. Sobel. Journal Of InfectiousDiseases 1989, 160(3), 490–496.

42 U. Kruger, R. Korber, J. Ziegler,J. Goschnick. Prospective experimentsto determine sweat odour with a gradientmicroarray, in ISOEN 2000 Abstracts,Persaud, K. C.; Gardner, J. W., editors;ECRO Indigo Lithoprint: Manchester, 2000;pp. 47–48.

43 J. R. Stetter, W. R. Penrose, C. McEntegart,R. Roberts. Prospects for infectious diseasediagnosis with sensor arrays, in ISOEN 2000Abstracts, Persaud, K. C.; Gardner, J. W.,editors; ECRO Indigo Lithoprint: Manche-ster, 2000; pp. 101–104.

44 N. Paulsson, E. Larsson, F. Winquist. SensorsAnd Actuators A-Physical 2000, 84(3),187–197.

45 T. D. Gibson, O. Prosser, J. N. Hulbert,R. W. Marshall, P. Corcoran, P. Lowery,E. A. Ruck-Keene, S. Heron. Sensors AndActuators B-ChemicalK 1997, 44(1–3),413–422.

46 M. Holmberg, F. Gustafsson,E. G. Hornsten, F. Winquist, L. E. Nilsson,L. Ljung, I. Lundstrom. BiotechnologyTechniques 1998, 12(4), 319–324.

47 J. W. Gardner, M. Craven, C. Dow,E. L. Hines. Measurement Science &Technology 1998, 9(1), 120–127.

48 A. Mantini, C. DiNatale, A. Macagnano,R. Paolese, A. Finazzi-Agro, A. D’Amico.Critical Reviews in Biomedical Engineering2000, 28(3–4), 481–485.


19

Recognition of Natural Products

Olivia Deffenderfer, Saskia Feast, Francois-Xavier Garneau

Abstract

The application of sensor-array analysis to natural products is still in its infancy. Thischapter seeks to provide an overview of the work that has been accomplished on nat-ural products, and to discuss various sampling and instrument setup considerationsthat apply in this arena. Two examples of the application of a polymer-composite sen-sor-array-based electronic nose to the identification of natural products are described.In one study, the CyranoseTM 320 accomplishes the sorting of wood species, jack pine,balsam fir, and black spruce, used in the lumber industry. In the second study thevolatile natural compounds from essential oils are used to distinguish closely relatedspecies of plants.

19.1

Introduction

Electronic noses provide a powerful modern analytical technique that addresses manysafety, quality, and process challenges facing manufacturers. Since their introductionin the early 1990s there have been many advances in sensor technology and data pro-cessing procedures used in electronic noses, coupled with a much greater understand-ing of the appropriate applications for this technology. This chapter provides an over-view of this modern analytical tool for applications in natural products. Many of thenatural products we shall discuss are also used in the food industry and are coveredfrom a food quality perspective in Chapter 21.Using an electronic nose in natural product applications can be challenging. Sam-

pling, sensor technology, sensitivity, and the inherent variability of natural productsare some of the concerns.


461461

19.2

Recent Literature Review

Electronic noses have been used for many applications from identifying solvents tonatural products. A summary of the recent publications on the applications of elec-tronic noses to natural products is included in Table 19.1. It is clear that most of thework to date on natural products has focused on those that we consume. The applica-tions include discrimination of spirits, beverage quality, fruit ripeness and quality,grain quality, meat and fish freshness, and oil quality; all types of sensors and a varietyof data processing tools are used. The electronic nose sensor technology used mostoften to sniff natural products, were metal oxide sensors (MOS) and conducting poly-mers (CP), or combinations of different sensing technology.Quartz crystal microbalance (QMB), surface acoustic wave (SAW), and mass spec-trometery (MS)-based electronic noses have also been tested. The most commondata analysis tool used was either principal component analysis (PCA) or cluster ana-lysis to easily visualize the differences between samples. Neural networks (NN) andfactor or discriminant analysis (DA) as well as regression techniques were used to testmodels.One of the main differences between the various studies is the sampling technique.

Though the general process of sampling, such as placing the samples in a sealed con-tainer, allowing headspace to equilibrate, and presenting the sample to the electronicnose was similar for many applications, the method in which this was done variedgreatly.

19.3

Sampling Techniques

One of the major components of successfully using an electronic nose is sample pre-paration. Sample containment, treatment, conditioning, storage, and seasonal varia-tions all impact the results of experiments performed with electronic noses.

19.3.1

Sample Containment

Typically all the samples need to be contained. These containers vary from simple vialsor jars to more sophisticated headspace vials for auto samplers. Electronic noses sam-ple headspace, hence knowledge of headspace generation and consistency is necessaryto develop the methods.

19 Recognition of Natural Products462

Tab. 19.1 Review of recent literature on electronic-nose applications

in natural products

Application Sensor Sampling Data Analysis Findings

Toasting level

of oak wood

barrels [1]

6 MOS Headspace sample taken

from above hot barrel

immediately after toasting.

PCA,

discriminant

function

analysis

(DFA), NN

An electronic nose would be

useful in process

monitoring of the toasting

level of oak wood barrels.

Fermentation-

bioprocess

monitoring [2]

eNOSE 4000

(Neotronics)

12 CP

10 mL samples were

placed in 500 mL glass

sample vessels and

tested at 30 8C.

DA Media spoilage, contamina-

tion, and microbial conta-

mination could be detected

earlier than other conven-

tional methods using an

electronic nose. Sterilization

level and inoculation level

could not be discriminated.

Freshness

of soybean

curd [3]

6MOS 10 s baseline, 50 s

sample, 40 8C s

ample extraction.

PCA Sensitivity decreased with

higher temperatures. The

electronic nose was able to

predict freshness of

soybean curd over time.

Cheese

ripening [4, 5]

eNOSE 5000

(12CP 8

MOS);

6QMB;

10MOS-

FET þ5MOS;

Smart

Nose (MS)

Various electronic noses

were used to test the ripe-

ning of four Swiss Emmen-

tal cheeses over a period of

one year. Static heasdspace

measurements were taken

in first study. SPME was

used for pre-concentration

in second study.

CDA MOS discriminated well but

were ‘poisoned’, CP and MS

poor sensitivity resulting in

poor discrimination, QMB

no discrimination, MOS-

FET alone gave poor discri-

mination but with MOS was

good system. MS with

SPME was best method in

discriminating cheeses be-

cause of repeatability, sim-

plicity, autosampler

capability.

Milk spoilage

(yeast/bacteria)

[6]

14 CP

(Bloodhound)

Samples allowed to

equilibrate for 30 min.

A charcoal filter was used

and the samples were

‘bubbled’.

BP-NN,

DFA, PCA,

canonical

analysis (CA)

Study shows promise in

using an electronic nose for

detecting milk spoilage.

Espresso

(seven blends)

[7]

Pico-1

(five thin-

film MOS)

Coffee was ground and

static headspace was

sampled.

PCA, ANN There was noticeable drift

that needed to be corrected.

95% correct predictions

when two similar classes

were combined as one class.

Espresso

beans/ground/

liquid [8]

Four thin-

film tin

oxide

Espresso beans and

ground beans were placed

in 20-mL vials. Liquid

coffee was extracted at high

pressure then placed in

vials. Samples equilibrated

in vial at 50 8C for 30 min

before sampling.

PCA, MLP

ANN, data is

drift

corrected

Whole beans: 100%

correct classification with

two sensors. Ground

coffee: 87.5% correct clas-

sification.

Liquid coffee: unsuccessful.


Tab. 19.1 Continued


Coffee [9] 12MOS Static headspace sampling

of roughly 30 samples of

three roasted coffees

(data from 1992 article).

Fuzzy

ARTMAP

97% of samples were

accurately classified as a

result of data processing.

Vanillin

fortified grape-

fruit Juice [10]

Ion-trap MS

chemical

sensor

Juice samples were

spiked with 40 to

2000 ppm vanillin.

PCA, DFA Vanillin limit of detection

was 40 ppm with classifica-

tion possible at 100 ppm.

Fruit ripeness

monitoring [11]

Tin oxide Peaches, pears and apples

placed into a plastic box.

150 mL headspace was

pulled out with a gas tight

syringe after 1 hr equilibra-

ting. Sensors were allowed

to stabilize for 10 min. They

were purged with synthetic

dry air.

NN A sealed chamber was used

to increase signal. Peach and

pear ripeness could correctly

be determined more than

92% of the time. Apple ri-

peness could not be deter-

mined well.

Fruit quality

[12]

Thickness

shear mode

quartz

resonators

coated with

pyrrolic

macrocycle

Slices of peaches and

nectarines in sealed glass

bottles and allowed to

equilibrate for

10 min at 30 8C.

PCA and

Learning

Vector

Quantization

neural network

Discrimination evident be-

tween fruits that had been

classified by a sensory panel.

Tomato aroma

[13]

e-NOSE 4000

(12CP)

Ripe tomatoes were stored

at 5, 10, 12.5, and 20 8Cand tested over 12 days. 20 g

of frozen tomato puree was

placed in sealed 113-mL

cups and thawed in 25 8Cwater bath. Then the sample

was placed into the electro-

nic nose sampling glass. The

electronic nose was purged

for 4 min, allowing head-

space to equilibrate.

MVDA

(CDA)

The electronic nose was able

to detect differences be-

tween ripe tomatoes stored

in different conditions. The

results from the electronic

nose corresponded with

sensory panel results.

Soft-rot

detection in

potato tubers

[14]

Two MOS

and three

MOS

(two experi-

ments)

Ambient conditions were

4 8C and 85% RH. 1 Kg te-

sted in Quickfit jar, 25 Kg in

paper sack with diseased

tuber at bottom of sack,

100 Kg tested in storage

crate.

Threshold One tuber with soft rot in a

storage crate of 100 kg good

tubers could be detected. An

inoculated tuber, not sho-

wing signs of soft rot, could

also be detected within 10 kg

of good tubers.


Tab. 19.1 Continued


Oatmeal

oxidation [15]

Fox 3000

(Alpha MOS)

1 g of oatmeal was

placed in 10 mL vial and

incubated at 100 8C for

30 min. Compressed air

was used as carrier gas.

Triplicate/quadruplicate

analyses performed for

each sample. Oatmeal had

been packaged in four

different pouches, some

designed to prevent/delay

rancidity.

PCA, SIMCA Hexanal is main rancidity

marker. Small variations in

volatile profile were seen

among samples analyzed

with an electronic nose.

After six weeks of storage,

differences could be seen

between different pack-

aging. Two to four weeks

was not long enough.

Barley grain

quality [16]

10MOSFET,

six MOS, one

CO2 monitor

10 samples with normal

odor and 30 with off

odor. 3 33 g samples of

each class were heated to

50 8C. Baseline and purge

with zero air.

PCA, PLS,

PLS-DA,

SIMCA

SIMCA used to classify if

samples had off odor. E-

nose: 3/40 misclassified,

GC-MS: 6/40 misclassified.

PLS used to predict ergo-

sterol with high confidence

and CFU level, which could

not be predicted well from

naturally infected grain.

Cereal quality

[17]

BH114,

Blood hound,

14 surface-

responsive

polymer

arrays

Cultures grown for 48,

72, and 96 hours on

wheat meal agar. Single

replicate petri plate cultures

placed in 500-mL sampling

bags filled with 300 mL

sterile air. Samples equili-

brated for 1 hr at 25 8C.Sampled in a 25 8C constant

temperature room.

PCA, DA,

CA

Classification of grain

quality may be a possibility

using electronic-nose tech-

nology. May be a simple and

fast way to detect and diffe-

rentiate between strains and

species of fungus.

Wheat

classification

by grade [18]

16

electro-

chemical

40 g of grain heated to

60 8C in sealed glass

container. 10 L of air

circulated through contai-

ner, ice trap and liquid N2

trap. Volatiles from traps

evaporated into air and

saved into tedlar bags.

Nearest

neighbor

(k-NN), NN

(k-NN) classified 68% cor-

rectly and NN classified

65% correctly. After data

correction for instrument

changes NN improved to

83%. NN outperformed

k-NN.

Wheat quality

[19]

CP array Wheat samples were

made artificially moldy

in the laboratory.

RBF-ANN

(92 samples

in training)

92.3% correct classification

(40 samples) with no bad

samples misclassified as

good.


Tab. 19.1 Continued


Rice quality

[20]

10MOSFET

and 12MOS

Rice varieties from two

crop years were studied.

5 g placed into 20-mL

vials sealed with Teflon

lined septa and caps.

Sample kept at room

temperature prior to

analysis and at 50 8Cduring sampling.

PCA Differences related to the

rice variety and age were

observed, but the varietal

differences were small in

comparison to differences in

age. The electronic nose may

be used to monitor aging or

shelf-life of rice.

Capelin

spoilage for

fishmeal

production

[21]

FreshSense

(Nine electro-

chemical gas

sensors)

Headspace gas above

capelin was sampled at

0 and 5 8C in storage

conditions. 1 kg of capelin

was placed in 5.2 L contai-

ner. Sensors reached

steady state within 10 min.

PLS1,

saturated

generalized

linear model

The total volatile base value

of capelin stored under dif-

ferent conditions could be

predicted with an electronic

nose.

Mahi-mahi

freshness [22]

AromaScan

(32CP)

The fish was stored at

1.7, 7.2, 12.8 8C for 0, 1,

3, 5 days and analyzed

with AromaScan. 10 g of

fish was placed in a bag.

The bag was evacuated

and filled with carbon-filte-

red air and allowed to equi-

librate for 10 min at 35 8C.The baseline was dried with

Silica gel. Carbon-filtered

ambient air was reference

air. Sensors were purge-

dwith headspace from 2%

2-propanol and allowed to

react with reference air for

2.5 min before next sample

MDA using

AromaScan

A32S

Windows

software

v. 1.3

The quality changes in ma-

hi-mahi using the electronic

nose correlated with sensory

panel results and microbio-

logical analysis. The elec-

tronic nose was also able to

predict different grades of

mahi-mahi stored at 7.2 8C.

Chicken

freshness [23]

8MOS The chicken was placed in

glass sampling containers.

NN The electronic nose could

predict freshness within

40 min of actual time using

one sensor and 20 min

using eight sensors

Minced-meat

rancidity [24]

HP 4440 Minced beef was stored

at 4 8C with lighting and

storage equivalent to a

retail store.

PCA The electronic nose was able

to measure the development

of rancidity in minced beef

over 17 days.

Swine products

[25]

FOX 2000

six MOS

Samples of subcutaneous

adipose tissue were min-

ced and frozen. For testing,

0.5 g was placed in a 10-mL

glass vial. Synthetic air in-

jected to remove ambient air

and the sample equilibrated

at 35 8C for 7 min.

LDA,

SIMCA

Swine products could be

classified with an electronic

nose based on what the

swine were fed. (feed,

feed þ acorn, acorns alone).


Tab. 19.1 Continued


Olive oil

quality [26]

Eight CP 2 mL of oil was placed in

10 mL vials for static head-

space sampling. The samp-

les equilibrated at 50 8C for

9 min prior to testing.

PCA Five oil qualities could be

discriminated with 90%

confidence interval. Five

different oils could be dis-

criminated with 90% confi-

dence interval.

Frying fat

quality [27]

Four MOS The fats were aged in

air at 180 8C. Vaporpassed through a GC

separation column, then

the flow of gas was split

between FID and MOS

chamber. Fat was hot

during sampling.

Line plots

were used to

compare

results of

MOS sensors

to reference

food oil

sensor.

Fat is deteriorated if level of

polar compounds exceeds

24–27%. The results show

good correlation with the

Food oil sensor. Water in-

fluence could be removed

and there was no interfe-

rence from different foods

cooked in oil.

Corn oils [28] AromaScan

(32 CP)

Samples were stored in

50 mL beakers at 60 8C in

dark. Testing occurred on

days 0, 4, 8. The total sample

time was 200 s with a 30 s

purge with 2% IPA vapor

followed by a 30 s purge with

water vapor.

PCA The electronic nose was

successful in detecting off-

odors that were produced by

oxidation.

Maize corn oil

rancidity [29]

MOSES II:

eight MOS,

eight QMBs

Sampling parameters not

outlined in article but are the

same as the parameters used

in similar GC/MS headspace

analysis.

PCA The limit of detection was

1 ppm of aldehyde in oil.

Tansy essential

oil [30]

32CP 0.5–1 mL of oil was placed

in 8-mL glass vials with

Teflon septum cap. Samples

were left for 1 hr at room

temperature. Each sample

was sampled five times in

random order.

PCA Good discrimination was

seen between three chemi-

cal varieties of Tansy es-

sential oil using an

electronic nose.

Golden Rod es-

sential oil [31]

32CP Refer to Tansy Oil

sample preparation [31]

PCA In less than 30 s per sample,

essential oils of three Gol-

den Rod species could be

discriminated using an

electronic nose.

Wood chip sor-

ting [32]

32CP Pieces of wood were broken

and placed into 250 mL

sealed glass jars. The

samples were tested at room

temperature. Each sample

was sampled five times in

random order.

PCA An electronic nose was ra-

pidly able to discriminate

and identify black spruce,

balsam fir, and jack pine.


19.3.2

Sample Treatments

Although an electronic nose may be a ‘point and sniff’ device for certain applications,additional sample treatment is often required for natural-product applications. Heat-ing, preconcentration, and grinding are methods used to increase the volatiles in theheadspace. Cooling can be used to prevent or slow spoilage over time. Removing a basecomponent can improve sensitivity to slight differences in samples. In the followingparagraphs, applications using these sample treatments are discussed.

19.3.2.1 Heating

A natural-product application for electronic noses is determining oil quality, which isoften done organoleptically. Cooking oils tend to have little or no odor, are not volatile,and have a low vapor pressure; it is therefore difficult to use electronic noses to detectoil. However, off odors in oil can be volatile. In several studies, electronic noses havebeen used to detect the rancidity of oil. Shen determined that an array of 32CP sensorscould detect odors produced by oxidation of corn oil [28] and Frank determined that anarray of eight MOS and eight QMB could detect as little as 1 ppm of aldehyde in cornoil [29]. In another study, discrimination of flat, musty, rancid, fusty, and muddy oliveoil could be determined with 90% confidence using conducting polymer sensors andPCA [27]. All studies were conducted in sealed containers and the samples were heatedto a minimum of 50 8C.

19.3.2.2 Cooling

Because meat can spoil rapidly it is essential to keep the samples cool. Process-linemonitoring would also require the sensors to perform at cool temperatures. Severalstudies have been done on fish freshness, while keeping the fish in cool conditions. Anelectronic nose using an array of electrochemical gas sensors, FreshSense, has beenspecifically designed to detect the volatiles resulting from the spoilage of fish. Thesestudies were usually done at normal storage conditions, between 0–7.2 8C. In anotherstudy, an electronic nose with CP sensors was used to evaluate the freshness of mahi-mahi fillets [22]. The electronic nose results correlated with sensory panel results aswell as microbiological analysis, and were successfully used to predict different sen-sory grades of mahi-mahi stored at 7.2 8C.

19.3.2.3 Removal of Base Component

Another sampling technique was used to discriminate different brands of beer. Etha-nol is present in beer in high concentrationsmasking slight differences between beers.In this case, the ethanol was ‘pre-separated’ from the beer. The remaining componentswere presented to an electronic nose with eight QMB sensors resulting in good dis-crimination between brands using PCA [33].


19.3.2.4 Preconcentration

Preconcentration is a technique used to concentrate volatiles prior to testing, and ismost commonly used for gas chromatography (GC)-MS headspace analysis. Types ofpreconcentration include solid-phase micro extraction (SPME), direct thermal deso-rption, purge-trap, and cyrotrapping. Preconcentration using SPME was typically per-formed for the electronic nose applications studied. Using this technique, there wasimprovement in the ability to discriminate cheeses. Schaller examined a variety ofelectronic nose technologies to test the ripening of Swiss Emmental cheese overthe period of a year [4]. A MOS sensor array alone and a MOS-field-effect transistor(MOS-FET) plus MOS sensor array resulted in a good assessment of cheese ripeness.However, the MOS sensors were ‘poisoned’ over time by the vapor. No discriminationwas seen using a QMB array or CP array. The MS-based electronic nose was not sen-sitive enough. However, when the cheese vapor was pre-concentrated on an SPMEfiber, good discrimination was seen using the MS electronic nose [5]. Another exam-ple where preconcentration is used is in the discrimination of similar wines. Gooddiscrimination between different types of alcoholic beverages such as beer, wine, spir-it, and samshu could be obtained with a relatively simple sampling method and eightSAW sensors [34]. Predictions of unknown samples using a back-propagated ANNwere also successful. However, discrimination between similar alcohols, such as or-ganoleptically similar wines [35] or beer [34] required sampling technique improve-ment. Wines from the same region with a similar taste were discriminated usingSPME fiber to concentrate the headspace before being presented to an electronicnose with 12 CPs [35].

19.3.2.5 Grinding

Grinding or crushing a solid sample creates more surface area, therefore a greaterconcentration of volatiles can be released into the headspace. This will reduce pro-blems created by headspace depletion and low volatile solids. In one study, differentbrands of espresso were classified by looking at whole beans, ground coffee, andbrewed coffee. Classification by espresso brand was 100% correct for whole beansamples and 87.5% correct for ground coffee using a NN. Classification was not suc-cessful when brewed coffee was sampled [8]. Two other studies showed similar resultswith 95% [7] to 97% [9] prediction accuracy for roasted coffee samples. In this casegrinding the coffee did not enhance prediction ability over using whole beans. How-ever, grinding coffee is a better sample preparation method than brewing coffee. Thisexample illustrates the importance of finding the best sample preparation techniquefor the application.

19.3.3

Instrument and Sample Conditioning

Instrument and sample conditioning are also important parts of the sampling tech-nique when using an electronic nose. This section refers to the pathway between thesample and the sensors. Modification of the baseline, purge technique, and tempera-ture control in the instrument are discussed.


19.3.3.1 Modifying Baseline

Some electronic-nose systems use gas cylinders to supply a constant baseline. How-ever, many electronic noses, including portable ones, draw the baseline from the am-bient air. Many modifications can be made to the baseline measurement includingdrying, humidifying, and filtering. A dry baseline is important when samplingvery dry products, such as dried spices, with a sensor array that responds to ambientmoisture in the baseline. For example, the baseline air was dried with Drierite (calciumcarbonate) when discriminating between two types of whole dried black peppercornsfrom different origins. Using a dry baseline improved the response allowing discri-mination and identification of unknown samples over a period of 13 days [36]. A hu-midified baseline can improve the sensitivity of an electronic nose to similar com-pounds in aqueous solutions such as beverages. Filtering the baseline can be espe-cially important when using a portable electronic nose in the field. A charcoal filtercleans the baseline air that may be contaminated by factory, fuel, or other strong odors,preventing the sensors from responding to the baseline vapor.

19.3.3.2 Purge Technique

Following a sample, the sensors need to be cleaned to return back to baseline prior tothe next sample. This is imperative in order to prevent cross contamination of samplesor carryover. Different methods are used to wash or purge the sensors after sampling.Often ambient or dry air is passed over the sensors for a period of time to clean thesensors of any remaining sample vapor. However, in determining mahi-mahi fresh-ness [22] and corn-oil freshness [28], the sensors were purged with 2% isopropyl al-cohol in water vapor followed by a second purge of only water vapor.

19.3.3.3 Temperature Control

In some electronic noses the sampling pathway before the vapor reaches the sensors isheated. This ensures that the sample temperature is always consistent regardless ofambient temperature.

19.3.4

Sample Storage

A great challenge of working with natural products is that they change over time. Byunderstanding the mechanism of change in natural products, for example spoiling orripening, sample quality can be maintained over time. An electronic nose can be usedto track the quality of natural products, such as grain, over time. Grain quality para-meters including rancidity and the presence of microorganisms have been studiedwith various electronic noses (see Chapter 21, reference 5). In one study an electronicnose trained on wheat made artificially moldy was used to identify commercial wheatsamples (of which 24 where good and 17 bad) with a 92.3% correct prediction rate [19].Importantly, no bad samples were misclassified as good.Another example of the effect of storage on natural products is shown in Maul’s

study of tomato flavor and aroma [13]. Tomatoes stored at lower temperatures had


a less flavorful aroma than tomatoes stored at higher temperatures. The electronicnose used for this study was able to classify ripe tomatoes based on storage condi-tions. Consideration of the variation of the quality of natural products as a resultof storage is therefore necessary in developing methods to use an electronic nosefor natural-product applications.

19.3.5

Seasonal Variations

Electronic noses have been used to study the quality or ripeness of fruits. Brezmesstudied the ripeness of peaches, pears, and apples, using whole fruit, an array of me-tal-oxide sensors and a NN [11]. Over 92% of the time, peach and pear ripeness couldbe determined. Unfortunately, the same results were not seen for apples.DiNatale was able to discriminate the quality of sliced peaches and nectarines based

on sensory markers, such as size and color, and QMB array [12]. Maul used an elec-tronic nose with CPs to detect differences in ripe tomatoes stored in different condi-tions [13]. Though electronic noses potentially can be used to monitor the quality ofsome fruits over one season, seasonal variations need to be addressed before there iswidespread use of the electronic nose in fruit quality monitoring.In another example, the seasonal variations over two crop years of different varieties

of rice were found to be greater than the differences in the rice varieties [20]. It wassuggested that the electronic nose might be more useful for shelf-life studies of ricethan for determining the variety of rice.

19.3.6

Inherent Variability of Natural Products

Natural products vary from season to season, by country of origin, and by species. Eventwo plants growing next to each other are different. Like humans, each plant and an-imal and therefore natural product, is unique although the major characteristics aresimilar. Due to this inherent variability it is critical that a large enough data set be takento capture as much variability as possible resulting in a more robust model.

19.4

Case Study: The Rapid Detection of Natural Products as a Means of Identifying PlantSpecies

Natural products have often been used to characterize and differentiate plants. Oneexample is the sorting of wood of different species of trees in the lumber industry bythe detection of species-specific marker compounds. Volatile natural compounds havealso been used to distinguish closely related species of plants or chemical varieties(chemotypes) of a particular species of plant by GC analysis of their essential oils.We have applied electronic-nose technology in both of these cases.

19.4 Case Study: The Rapid Detection of Natural Products as a Means of Identifying Plant Species 471471

19.4.1

Wood Chip Sorting

The pulp and paper industries in eastern Canada have a need to differentiating blackspruce, balsam fir, and jack pine because their proportions in wood chips affect thequality of the pulp and paper produced. A prerequisite to determining their propor-tions is to be able to rapidly identify the wood of the three conifers. Several attemptshave beenmade and the few that have succeeded were mainly directed to the sorting oflumber. The methods developed involved recognition of the heartwood of the threespecies by spectroscopic and/or visual differentiation [37]. These methods failed todistinguish the sapwood of these conifers which makes up the major proportion ofthe wood chips used by paper mills. Pichette et al. [38] were able to distinguish thethree woods using a combination ofmarker compounds and GC profiles (fingerprints)of the hexane extracts, however the method is too slow to be of any use to paper mills.The rapid sapwood differentiation of these conifers has now been achieved using anelectronic nose based on sensor-array technology. In addition, the heartwood of thethree trees was also differentiated in the same manner.

19.4.2

Experimental Procedure

Pieces of the sapwoodmeasuring 3 � 5 cm from seven jack pine, eight balsam fir, andeight black spruce trees were sampled using the CyranoseTM 320. The wood chips wereplaced in 23 250-mL glass jars, randomly ordered, and kept at room temperature. Thesamples were sealed with a Teflon-lined lid for storage. The lid was removed for testingand replaced with a two-port Teflon covering. One port was fitted to the snout of theCyranoseTM and used for sampling while the other port was open to the atmosphere.The headspace of each jar was sampled five times in succession using the samplingconditions listed in Table 19.2.A total of 115 smell prints were acquired from the 23 logs. The smell prints were

analyzed by PCA and eight smell prints that were identified as outliers with 95%confidence were removed. Canonical analysis was then applied to the data. The cano-nical plot (Figure 19.1) shows separation between the different woods. The sampleswere correctly classified 95% of the time as shown in Table 19.3.

Tab. 19.2 Cyranose 320 Sampling Conditions for wood chips

Baseline Time 15 s

Sample Time 25 s

Purge Time 60 s

Sample Flow Rate 75 mL min�1

Sample Temperature Room temperature

Sensor Temperature 41 8C


19.4.3

SPME-GC Analysis of the Sapwood of the Conifers Used in Pulp and Paper Industries

An electronic nose essentially analyzes the headspace of a sample, and SPME-GCanalysis can indicate whether a difference exists in the headspace between differentmaterials. SPME-GC analyses were carried out on samples of ground sapwood fromindividual trees of balsam fir, jack pine, and black spruce. One gram of wood wasplaced in headspace vials and heated for 3 minutes at 70 8C. Then a polyacylateSPME fiber (85lm) was inserted into the sample vial for 5 minutes at 70 8C. The fiberwas desorbed for 2 minutes at 280 8C in the injection port of the GC. GC analysis wasperformed using a non-polar DB-5 capillary column (25 mm � 0.25 mm � 0.25 lm)using the time settings listed in Table 19.4.

Fig. 19.1 Canonical plot pro-

jections of the 114 smell prints

of wood chips from fir, spruce,

and pine

Tab. 19.3 Number of correct identifications for wood chips sampled

with a CyranoseTM 320. The value in parentheses is the percentage

correct

Identified as Fir Identified as Pine Identified as Spruce

Fir 36 (100) 0 (0) 0 (0)

Pine 0 (0) 30 (88) 4 (12)

Spruce 0 (0) 1 (3) 36 (97)

Tab. 19.4 GC Temperature setting for study of ground sapwood.

Temperature 8C Time

Injection Temperature 280

Detector Temperature 320

Temperature Program

Step 1 60 2 min

Step 2 (ramping) 220 5 8C min�1

Step 3 220 5 min

Step 4 (ramping) 320 10 8C min�1

Step 5 320 40 min

19.4 Case Study: The Rapid Detection of Natural Products as a Means of Identifying Plant Species 473473

Fig. 19.2 GC profiles of the headspace of balsam fir, jack pine,

and black spruce sapwoods obtained by SPME


The GC profiles, also referred to as fingerprints, are shown in Fig. 19.2 and repre-sent the average of the individual tree profiles obtained for each of the three species ofconifers studied. As can be seen, the differences observed in the three GC profilescorrelate to the clusters shown in the PCA plots obtained (Fig. 19.1) using the Cyra-noseTM 320.

19.4.4

Conclusion: Wood Chip Sorting

This procedure, if extended to a chip-by-chip analysis of samples representative of apile of sawmill wood chips, should lead to a means of determining the proportions ofthe three conifers present in the mixture.

19.5

Case Study: Differentiation of Essential Oil-Bearing Plants

19.5.1

Golden Rod Essential Oils

The essential oils of three species of Golden Rod, Solidago canadensis, S. rugosa and S.graminifolia, were analyzed by GC using a non-polar and a polar capillary column andby GC-mass spectrometry. As can be seen from the results shown in Table 19.5, thechemical compositions are quite different. The major constituents of S. canadensis area-pinene (26.9%) and myrcene (28.3%). Sabinene (10.1%), limonene (14.8%) and b-Phellandrene (18.9%) are the major components of S. graminifolia whereas a-pineneis by far the most important constituent of S. rugosa at 49.4%. Other differences arealso noticeable in the percentage composition and the presence or absence of certainminor compounds. Approximately two hours of experimental work were required toperform these analyses.The essential oils of these three species of Golden Rod were also analyzed using the

CyranoseTM 320 unit. A 0.5–1 mL sample of the essential oil from each of the threespecies of Golden Rod was placed in an 8-mL glass bottle fitted with a Teflon-facedrubber-lined cap. A small hole in the cap was covered and the oil was allowed tostand for 1 hr at room temperature. The headspace of each of the oils was thensampled five times in a random order. The sampling conditions are shown in Ta-ble 19.6. A total of 15 smell prints were acquired from the three essential oils.Each print required less than 30 s. The 15 smell prints were analyzed by PCA andthe plot projections (Fig. 19.3) show a clear distinction of the essential oils of the threespecies of this plant.


Tab. 19.5 Percentage composition of essential oils of three species of

Solidago

Compounds R.I. (DB-5)a) S. canadensis S. graminifolia S. rugosa

a-pinene 941 26.9 1.8 49.4

camphene 954 0.8 0.7 0.4

sabinene 977 0.8 10.1 13.5

b-pinene 978 4.2 5.8 5.8

myrcene 993 28.3 4.7 3.5

a-phellandrene 1002 1.3 3.1

limonene 1033 11.1 14.8 3.1

b-phellandrene 1033 1.2 18.9 14.4

(E)-b-ocimene 1058 0.7 3.9

bornyl acetate 1295 3.5 3.5 1.0

b-elemene 1389 1.1

a-gurjunene 1402 4.8

a›-caryophyllene 1459 0.9 0.4

c-gurjunene 1472 1.1

germacrene D 1488 5.5 3.1 3.4

germacrene A 1512 2.8

d-cadinene 1531 1.2

cubenol 1632 1.2

m/e:

105,147,161,148,218

1799 6.4

a) R.I.: retention indices; DB-5: non-polar capillary column.

Fig. 19.3 PCA plot projections

of 15 smell prints of essential

oils of Solidago.

Squares are S. graminifolia;

circles are S. rugosa;

triangles are S. Canadensis


19.5.2

Essential Oils of Tansy

Several different chemical varieties of Tansy (Tanacetum vulgare) have been reported[31]. The chemical compositions of the three varieties observed close to Chicoutimi inthe Saguenay Region of northern Quebec, Canada, are shown in Table 19.7. The threechemotypes are characterized by the predominance of either b-thujone (75.3%) orchrysanthenone (54.8%) or similar amounts of 1,8-cineol (16.9%), camphor(17.5%), and borneol (19.3%).The three varieties of Tansy essential oil were also analyzed by the Cyrano Sciences

electronic nose. The sampling conditions and the procedure used were the same asthose described above for the Golden Rod essential oils. A total of 15 smell prints were

Tab. 19.6 Cyranose 320 sampling conditions for essential oils

Baseline Time 2 s

Sample Time 2 s

Purge Time 20 s

Sample Flow 120 mL min�1

Sample Temperature Room temperature

Sensors Temperature 35 8C

Tab. 19.7 Percentage composition of essential oils of three chemo-

types of Tanacetum vulgare (T.v.)

Compounds R.I. (DB-5)a) T.v. 538b) T.v. 540c) T.v. 541d)

a-pinene 941 0.5 4.8 3.8

camphene 954 0.5 1.1 7.0

sabinene 977 2.5 2.4 5.3

b-pinene 978 0.5 4.4 2.3

myrcene 993 0.3

a-phellandrene 1002 1.3 3.1

para-cymene 1028 0.40 0.9 0.9

1,8-cineol 1034 4.1 6.8 16.9c-terpinene 1068 0.3 0.4 0.6

linalool 1112 0.9 1.1 0.2

a-thujone 1117 0.2 0.8 0.3

b-thujone 1123 75.3 3.1

chrysanthenone 1130 3.3 54.8 5.9

camphor 1146 0.9 1.3 17.5pinocarvone 1163 0.2 1.2 0.9

borneol 1166 2.3 4.3 19.3bornyl acetate 1295 0.7 1.0 7.6

germacrene D 1488 2.3 2.7 3.0

a) R.I.: retention indices; DB-5: non-polar capillary column.b) Tanacetum vulgare, b-thujone chemotype.c) Tanacetum vulgare, chrysanthenone chemotype.d) Tanacetum vulgare, camphor, borneol and cineol chemotype.


acquired from the three essential oils. The 15 smell prints were analyzed by PCA andthe plot projections (Fig. 19.4) show a clear distinction of the essential oils of the threevarieties of this plant.

19.5.3

Conclusion: Essential Oils

These results show promise for the rapid identification of essential oils from differentspecies of plants, and of oils from different chemical varieties of a specific species of aplant. This is particularly important when one considers that the various applicationsof essential oils require consistency in their chemical composition. An extension ofthis method would be a plant-by-plant identification in the field by sampling the head-space volatile compounds using this electronic nose technology.

19.6

Conclusion and Future Outlook

The application of electronic noses to the classification and identification of naturalproducts provides a large potential market. There are opportunities to classify plantspecies by aroma, identify and sort raw materials, check for consistency among nat-ural oils used in perfumes and as flavors. With careful sample preparation and control,electronic noses can be usefully applied to the recognition of natural products. In orderfor the full potential of the electronic nose to be realised in this field, we need to devel-op library-type applications whereby the instrument could be taught the patterns of aspecies and a database developed that spans the seasons. For this to become reality very

Fig. 19.4 PCA plot projections

of 15 smell prints of essential

oils of Tanacetum vulgare.

Squares are camphor chemo-

type; circles are chrysanthenone

chemotype; triangles are

b-thujone chemotype


stable systems, or systems that are readily calibrated, are required. Both of these solu-tions are being developed. Noses using mass spectometry are more stable but ulimt-ately they are too expensive for widespread implementation and not yet versatile forpoint-of-need deployment. In addition, software capable of handling hundreds of re-sponse patterns needs to be provided.

Acknowledgments

The valuable contributions to these studies by Steve Hobbs, Bernard Riedl, AndrePichette and Helene Gagnon are gratefully appreciated. We also thank Guy Collinfor the reproduction of the Tansy essential oil percent composition Table from hispublication [31].

References

1 P. Chatonnet. American Journal of Enologyand Viticulture 1999, 50, 479–494.

2 P. Namdev, Y. Alroy, V. Singh. BiotechnologyProgress 1998, 14, 75–78.

3 E. Y. Park, K. Y. Han, S. H. Ho, S. S. Kim,B. S. Noh. 2000 IFT Annual Meeting:Prediction of Freshness for Soybean Curd bythe Electronic Nose, June 10–14, 2000,Dallas, TX.

4 E. Schaller, J. Bosset, F. Escher. ChemicalSensors, Biosensors, Bioarrays 1999, 53,98–102.

5 E. Schaller, S. Zenhausern, T. Zesiger,J. Bosset, F. Escher. ANALUSIS 2000, 28,743–749.

6 N. Magan, A. Pavlou, I. Chrysanthakis.Sensors and Actuators B 2001, 72, 28–34.

7 G. Sberveglieri. The Knowledge Founda-tion’s Conference on Electronic NoseTechnologies: Advances in Engineering,Integrating and Commercial Novel SensorTechnologies, October 26–27, 2000,San Diego, CA

8 M. Pardo, G. Niederjaufner, G. Benussi,E. Comini, et al. Sensors and Actuators B2000, 69, 397–403.

9 E. Llobet, E. L. Hines, J. W. Gardner,P. N. Bartlett, T. T. Mottram. Sensors andActuators B 1999, 61, 183–190.

10 K. Goodner, R. Rouseff. Journal of Agricul-tural and Food Chemistry 2001, 49, 250–253.

11 J. Brezmes, E. Llobet, X. Vilanova, G. Saiz,X. Correig. Sensors and Actuators B 2000, 69,223–229.

12 C. DiNatale, A. Macagnano, E. Martinelli,E. Proietti, R. Paolesse, L. Castellari,

S. Campini, A. D’Amico. Sensors andActuators B 2001, 77, 561–566.

13 F. Maul, S. Sargent, C. Sims, E. Baldwin,M. Balaban, D. Huber. Journal of FoodScience 2000, 65, 1228–1237.

14 B. Costello, R. Ewen, H. Gunson,N. Ratcliffe, P. Spencer-Phillips. Measure-ment Science and Technology 2000, 11,1685–1691.

15 C. Wessling, T. Nielsen, J. Giacin. Journalof the Science of Food and Agriculture 2001, 81,194–201.

16 B. T. Olsson J, T. Lundstedt, J. Schnurer.International Journal of Food Microbiology2000, 59, 167–178.

17 G. Keshri, N. Magan. Journal of AppliedMicrobiology 2000, 89, 825–833.

18 J. R. Stetter, M. W. Findlay, K. M. Schroeder,C. Yue, W. R. Penrose. Analytica ChimicaActa 1993, 284, 1–11.

19 P. Evans, K. Persaud, A.McNeish, R. Sneath,N. Hobson, N. Magan. Sensors and ActuatorsB 2000, 69, 348–358.

20 A. Kramer, C. Grimm, E. T. Champagne.2000 IFT Annual Meeting.

21 G. Olafsdottir, A. Hognadottir, E. Martins-dottir, H. Jonsdottir. Journal of Agriculturaland Food Chemistry 2000, 48, 2353–2359.

22 W. Du, T. Huang, J. Kim, M. Marshall,C. Wei. Journal of Agricultural and FoodChemistry 2001, 49, 527–534.

23 A. Galdikas, A. Mironas, D. Senuliene,V. Strazdiene, A. Setkus, D. Zelenin. Sensorsand Actuators B 2000, 69, 258–265.

24 V. P. Shiers, A. D. Squibb. LeatherheadTechnical Note 1999, 130, 1–18.

19.6 Conclusion and Future Outlook 479479

25 I. Gonzalez-Martin, J. Perez-Pavon,C. Gonzalez-Perez, J. Hernandez-Mendez,N. Alvarez-Garcia. Analytica Chimica Acta2000, 424, 279–287.

26 A. Guadarrama, M. L. Rodriguez-Mendez,J. A. de Saja, J. L. Rios, J. M. Olias. Sensorsand Actuators B 2000, 69, 276–282.

27 M. Muhl, H. Demisch, F. Becker, C. Kohl.European Journal of Lipid Science and Tech-nology 2000, 102, 581–585.

28 N. Shen, S. Duvick, P. White, L. Pollak.Journal of the American Oil Chemists Society1999, 76, 1425–1429.

29 M. Frank, T. Hermle, H. Ulmer, J. Mitrovics.Sensors and Actuators B 2000, 65, 88–90.

30 G. J. Collin, H. Gagnon. Personal Com-munication.

31 G. J. Collin, H. Deslauriers, N. Pageau,M. Gagnon. Journal of Essential Oil Research1993, 5, 629.

32 S. Hobbs, F. Garneau, B. Riedl. “Distin-guishing Spruce, Fir & Pine woods for thepulp and paper industry”, Cyrano SciencesApplication Note, March 2000.

33 I. Heberle, A. Liebminger, U. Weimar,W. Gopel. Sensors and Actuators B 2000, 68,53–57.

34 Y. Yang, P. Yang, X. Wang. Sensors andActuators B 2000, 66, 167–170.

35 Guadarrama, J. A. Fernandez, M. Iniguez,J. Souto, J. A. de Saja. Sensors and Actuators B2001, 77, 401–408.

36 O. Koper, T. Zhang. ‘Discrimination ofBlack Peppers’, http://cyranosciences.com/applications/F_PepperIdentification_14.pdf,2000.

37 A. H. Lawrence, R. J. Barbour, R. Sutcliffe.Analytical Chemistry 1991, 63, 1217.

38 A. Pichette, F.-X. Garneau, F.-I. Jean, B.Riedl, M. Girard. Journal of Wood ChemistryTechnology 1998, 18(4) 427.


20

Process Monitoring

Thomas Bachinger and John-Erik Haugen

Abstract

Electronic noses have the potential to prepare new ground for non-invasive on-linemonitoring of biological processes. In this article we outline their applicability forprocess monitoring on the basis of selected examples in the areas of food- and bio-technology. Specific case studies on bioprocessmonitoring are presented showing thatan investigation of the odor of cell cultures can provide the bioprocess operator withvaluable information on cell and process state changes. The second application pre-sented outlines the use of electronic noses at-line for monitoring industrial processesin the food and feed industry. For both applications we show that the implementationof electronic noses represents a cost-effective tool for rapid assessment of the chemicaland microbial status of raw materials, process streams and end products. Extensiveand costly rework or disposal of products that do not fulfill their specifications couldbe prevented.

20.1

Introduction

The quality of biological products has today become of increasing concern to society.Based on concerns like the potential threat of BSE in food products or the cross trans-ferability of viruses between vertebrates this is especially true for biopharmaceuticalproducts, which is also expressed in the existing vast amount of public regulations.This draws attention to the importance of the monitoring of batch processes to ensuretheir safe operation and to assure that they produce consistent high-quality products.Most biological processes that can be found in the food and biotechnology industries

are probably suited for the application of electronic noses. This is because they involvehigh concentrations of aromatic compounds or microorganisms producing a widerange of volatiles. However, the demands put on real-time monitoring methods bysuch processes are high regarding information content, system integration and sta-bility. One reason is that traditional chemical and biological plants are complexnon-linear dynamical systems with multiple input and output variables. Often they


481481

are also composed of numerous sub-processes closely integrated with interconnectedmass and energy balances.Figure 20.1 illustrates the simplicity of system integration for electronic noses into

such processes. The emission from the process is sampled on-line and analyzed by thegas sensor array in regular time intervals. Besides the ease of system integration sev-eral other advantages are obvious: eventual process barriers, e.g. the sterile barrier in abioprocess, are not violated and system maintenance as well as operator interferenceare minimized due to a possible high degree of automation.In this review we will focus on two application areas: the on-line monitoring of

bioprocesses and the at-line monitoring of food processes. A short introduction tothese fields will be followed by a review of previous works. On the basis of recentresults we will then outline the capacity of the electronic nose for process monitoring.

20.1.1

On-line Bioprocess Monitoring

In a typical bioprocess cells are grown under sterile conditions in tanks on liquidmedia that provide, for example, essential nutrients, and vitamins. The productsfrom bioprocesses range from enzymes and single cell protein to biopharmaceuti-cals, which naturally all impose high demands on product quality and safety.Today, most bioprocesses still operate at relatively low yields despite the fact that

microbial transformations often reach yields close to the theoretical maximum.One of the reasons is that sensors that acquire real-time information about the

Fig. 20.1 Schematics on the integration of an electronic nose into a

biological process. The reference gas, which can be the same as the

process gas, is humidified before reaching the gas sensors. The

sampling interface protects for liquid entry and compensates for flow

variations. VS, Sample gas valve. VR, Reference gas valve

20 Process Monitoring482

cells’ state and their interaction with the bioreactor environment are rarely available.Consequently, the implementation of sophisticated process control is prevented.Since the experienced operators have long used the odor from bioprocesses for state

identification, it could be expected that relevant information can be extracted from thebioprocess off-gas. The application of non-invasive on-line monitoring methods likeelectronic noses could therefore certainly contribute to improve the quality of biopro-cess products.

20.1.2

At-line Food Process Monitoring

A typical production line in the food industry is characterized by several productionsteps/stages on the way from raw material to final product. In order to keep productquality high throughout the whole production line there may be quality properties thatare not measurable on-line and therefore would require at-line sampling and off-lineanalysis at the production line or in the QC laboratory of the factory. The properties tobe investigated off-/at-line will therefore not necessarily coincide with the on-line re-quirements to quality control analysis. In the case of at-line gas sensor array applica-tions for food process monitoring such properties may be related to the food chemistryof the product and can be measured directly or indirectly by analyzing the vapor phaseof the product at different production stages. They may represent product propertiesrelated to, for instance, odors, flavor, rancidity, and spoilage. Since such properties areof importance for a variety of processed food products, electronic nose technologyshould have a wide application range in the food industries.

20.2

Previous Work

A large number of investigations on biological activity monitoring using electronicnose technology can be found in the literature. Examples range from the classificationof microbial strains [1, 2] and grains [3] to bacterial contamination of meat [4] andmedical applications like the diagnosis of diabetes [5]. However, only a few are directlyrelated to at-line or even on-line process monitoring (see Tab. 20.1).

20.2.1

Quantitative Bioprocess Monitoring

The applicability of electronic noses to bioprocess monitoring has only recently beenpresented [6–8]. The main focus was initially on using multivariate methods to relatethe gas sensor responses to key metabolite concentrations or cell growth. This is be-cause such variables can be expected to be directly associated with the aroma from thecell culture. For instance, the concentration of ethanol and the cell growth could be


Tab. 20.1 Listing of process monitoring applications presented in the literature. BM,

bioprocess monitoring. FM, food process monitoring

Application Sensors Algorithms Comment

BM – estimation

of key metabolites

[11, 21]

MOSFET, MOS, IR Forward selection [28] for

signal parameter selection

applied [11]

ANN models for glucose [11, 21], acetate

[11], ethanol [11, 21], acetaldehyde [11],

glycerol [11] in S. cerevisiae batchprocesses

BM – estimation

of cell growth

[10–12]

MOSFET, MOS, IR Forward selection [11, 12].

Component correction [27]

for drift compensation [12].

ANN models established in E. coli batch

[10]; CHO perfusion [12]; S. cerevisiae

batch [11] processes

BM – estimation

of product

concentration [12]

MOSFET, MOS, IR Forward selection

and component

correction

ANN model for rFVIII estimation in

long-term perfusion CHO process

BM – quality of

bioprocess media

[17, 19]

MOSFET, MOS, IR

[19].

CP [17, 19]

Forward selection

[19], PCA [17, 19],

ANN [19]

Discrimination of casein hydrolysate

for E. coli growth [17]. Prediction of

fermentability of lignocellulose media

for S. cerevisiae [19]

BM – estimation

of preculture quality

[20]

MOSFET, MOS, IR Forward selection, PLS Preculture quality and state estimation

for a rec. E. coli strain

BM – process and

cell state determination

[9, 13, 14]

MOSFET, MOS, IR PCA [9, 13, 14] Process state visualization in rec. E. colifed-batch [13], and S. cerevisiae large-scaleprocesses [9]. Cell transition state visua-

lization in perfusion CHO cell process

[14]

BM – cell physiology

prediction [15]

MOSFET, MOS, IR Forward selection,

PLS

Semi-quantitative estimation of physio-

logical state variables in E. coli and

S. cerevisiae processes

BM – observation

of metabolic burden

[18]

MOSFET, MOS, IR – Visualization of cell stress caused by

strong overexpression of rec. protein in E.coli

BM – detection of

infection [14, 16, 17]

MOSFET, MOS, IR

[14, 16].

CP [17]

– Identification of Micrococcus sp. infectionin 500 L CHO process [14]. Identification

of B. cereus, P. aeruginosa in 2 L CHO

process [16]. Shake flask tests with E. coli[17]

FM – aroma quality

of cured ham [22]

MOS, Electro-chemical ANN Identification of off-flavor in Serrano

type dry cured hams

FM – quality control

of drying process in

ham production [23]

MOS Critical level of accumulated

sensor response

Control of drying process of Iberian

hams in chambers

FM – quality control

of sugar beet [24]

Ion mobility

spectrometry, MOS

Critical level of sensor

response

Identification of spoiled sugar beet

FM - sorting of fresh

fruit juices [25]

MOS PCA Identification of grape juices with

off-flavor

FM – off-flavors in

cow’s milk [26]

MOSFET, MOS, IR DPLSR, ANN Identification of feed off-flavor in

cow’s milk


estimated with an accuracy of about 10% in a 200 m3 Saccharomyces cerevisiae fermen-tation process using artificial neural network (ANN) technology [9]. Improvement ofthe electronic nose system and the sampling method allowed the estimation of cellgrowth (biomass) in a 2 L Escherichia coli batch process to as high as 1.46% accuracy[10]. Such cell growth estimates correlate almost perfectly with the accuracy of stan-dard reference methods (see Section 20.4, study 1 for details).The same improved system was also used to measure cell growth and metabolites

like ethanol, glucose, or acetate in a 2.5 L Saccharomyces cerevisiae batch process [11].Consistently high accuracies between 2.4 and 5% for the process variables were va-lidated by adding a total of 6 batches at extended batch duration of 35 h to the artificialneural network (ANN) training set. In a different study, the viable cell count was es-timated instead of biomass in a five-week production-scale perfusion process. There, itwas shown that the viable cell count of Chinese Hamster Ovary (CHO) cells can beestimated accurately at � 10% despite the typical low cell concentration of such pro-cesses (� 106 cells mL�1) [12].The successful measurement of glucose in the studies described above is of course

not related to a direct measurement of glucose in the process off-gas, since glucose isnon-volatile. However, the presented results suggest that it is possible to predict suchmetabolites because they are correlated with other volatile compounds from the pro-cess via stoichiometric or other complex correlations. In the same context, it was pos-sible to measure the product concentration in the above-described CHO cell process[12]. The therapeutic high molecular weight protein ‘human blood coagulation factorVIII’ could be estimated accurately to about the same value as the viable cell count,despite the fact that it is non-volatile.

20.2.2

Qualitative Bioprocess Monitoring

To improve current control strategies in bioprocesses it is often not necessary to mea-sure all key metabolite concentrations accurately. Instead it can be of great advantage ifthe sensor signal changes in time can reveal simple process state deviations or meta-bolic changes of the cells. With such on-line information available the operator couldreact faster to process faults or unfavorable conditions in the bioreactor.This principle was described when following simple process phases in a small-scale

Escherichia coli fed-batch fermentation producing recombinant human growth hor-mone [13], as well as in a fed-batch bakers yeast production process on 200 m3 scale[9]. Again, with an improved measurement system cell transition states could be vi-sualized more accurately in a 500 L perfusion mammalian cell (CHO) cultivation pro-cess for production of recombinant human blood coagulation factor VIII [14]. It waspossible to follow batch, fed-batch and perfusion stages on-line in the process. Also,states of high and low factor VIII productivity as well as lactate formation (high lactateconcentrations are inhibiting to the metabolism of mammalian cells) could be visua-lized.The above principle was further extended by a quasi-quantification of the different

metabolic states of cells during a bioprocess. In laboratory-scale Escherichia coli fed-


batch and Saccharomyces cerevisiae batch processes semi-quantitative estimation of thephysiological and metabolic states of the cells was realized using simple partial leastsquares (PLS) models [15]. For a detailed description see Section 20.4, study 2.Another important area of application for electronic noses is the detection of con-

tamination in bioprocess. Suitable on-linemethods for identification of foreign growthin bioprocesses rarely exist. Instead, routine checks are made usually once a day bytime-consuming incubations of media samples. In a production-scale CHO cell perfu-sion process it was shown that a bacterial infection with a Micrococcus sp. could beidentified at least 1 day before the in-process analysis [14]. Intentional contaminationsof laboratory-scale CHO cell perfusion processes with Bacillus cereus and Pseudomonasaeruginosa supported the above findings [16]. In shake-flask cultivations of Micromo-nospora carbonacea the pure culture could be discriminated from contaminated culture[17].In order to optimize the productivity in recombinant protein fermentations, max-

imization of the replication and protein expression rate is desired in order tomatch thebiosynthetic capacity of the cell. Such process optimization is much easier to achieve ifa sensor technology is available that can identify metabolic burden on-line in the bio-process. The applicability of an electronic nose to detect metabolic burden was as-sessed in a series of small-scale fed-batch fermentations using Escherichia coli produ-cing human recombinant superoxide dismutase [18].The quality of complex growth media is decisive for high growth rate and product

yield. Successful discrimination of casein hydrolysates with high quality for growth inrecombinant Escherichia coli from lots with low quality has been shown recently [17].Also the quality of lignocellulose hydrolysates for production of ethanol with Sacchar-omyces cerevisiae has been predicted [19]. The outcome of the anaerobic yeast fermenta-tion could be predicted concerning ethanol productivity by analyzing the hydrolysatesbefore fermentation start.An application with great impact on the performance of final-stage production bio-

processes is the determination of preculture quality. A preculture is the precedingfermentation stage of the production scale fermentation and its quality is thereforeof high importance for product quality and yield. The quality and state of inoculumfor a 2.5 L recombinant Escherichia coli fed-batch fermentation was assessed success-fully in a recent study [20].

20.2.3

At-line Food Process Monitoring

Only a few comprehensive studies exist on at-line food process monitoring applica-tions. One promising application is the work done by Abass et al. who applied anelectronic nose for at-line quality monitoring of cured hams [22]. They could demon-strate that the system successfully rejected all the hams that had been assessed as “bad”according to off-flavors by a trained panel. In another at-line food process monitoringapplication an electronic nose was applied for monitoring and controlling the aromaduring the drying process of Iberian hams in chambers [23]. An application that has


been implemented recently in the food industry is the use of an ion-mobility based gassensor system for at-line quality sorting of spoiled sugar beet [24]. Additional at-linefood process monitoring examples are listed in Tab. 20.1.

20.3

Special Considerations

Reproducibility and repeatability is an issue in sensor technology due to sensor drift(see Chapter 12). Most chemical sensors do not remain stable over time due to loss insensitivity and require a frequent recalibration to obtain stable pattern recognition andprediction models. In cases where the sensor drift exceeds the variation in the realmeasurement data a drift algorithm would be required. Different mathematical ap-proaches have been used recently to handle this problem and they are based onthe temporal variation in the sensor signal of repeated identical reference samplesthat are being measured together with the real samples [27–29]. In bioprocesses ithas been shown that the background of the non-inoculated growth media can beused as a stable reference and sensor drift of up to 30% over 1 year could be correctedfor [12]. For the case studies investigated in this paper the drift of the sensors did notrepresent any major problem due to the fact that the real measurements by far ex-ceeded the magnitude of the sensor drift and drift compensation was therefore notemployed.Important considerations for instrument design are to include liquid protection and

foam traps when measuring liquid samples on-line over a long period of time. Alsoheated gas transfer lines should be installed to avoid condensation. To reduce theinfluence of water and to minimize the difference in response intensity between sam-ple and reference, the reference gas should be humidified (see Fig. 20.1).

20.4

Selected Process Monitoring Examples

20.4.1

On-line Monitoring of Bioprocesses

Conventional bioprocess monitoring still suffers from a lack of suitable on-line mon-itoring methods that can reveal process states, identify the concentrations of key me-tabolites or determine cell growth. The complexity of the metabolic network of cellsresults in a large amount of chemical compounds that could be analyzed in a biopro-cess in order to obtain information about the cells’ metabolic or physiological state.However, measurement of such a vast number of analytes requires several differentsensor systems to be connected to the bioprocess, many of which are difficult or im-possible to operate in on-linemode. The lack of such on-line sensors that could capturecomprehensive data about the metabolic state of the cell culture therefore impedesefficient process control.


One of the most important parameters to measure in bioprocesses is certainly thetotal cell mass. However, today’s existing cell mass monitoring methods can onlypartly cope with the requirements from modern bioprocesses, e.g., changing highaeration rates in the bioreactor, changing media composition or low cell mass. Thefirst case study presented will outline the potential of the electronic nose for quanti-tative estimation of biomass in bioprocesses.Refined process control algorithms could be implemented in bioprocesses if it

would be possible to on-line measure biomass, substrate, product and inhibitor con-centrations. By calculating the uptake/production rates (physiological variables) there-of, the physiological state of the cell culture would be revealed and the culture couldtheoretically be controlled towards highest possible yield and product quality. Thesecond case study will show the successful semi-quantitative estimation of the phy-siological state of a cell culture using an electronic nose.

20.4.2

At-line Monitoring of a Feed Raw Material Production Process

The third case study is a feasibility study that focuses on the use of an electronic nosefor monitoring the quality of slaughter waste. Waste from slaughterhouse’s representan important raw material that is being utilized for production of different animalfeeds. Due to the possible link between animal cannibalism and BSE, quality controlof the waste processing is of great importance in order to obtain products based onpure raw material from the same type of animal. At the waste processing plant thequality of the delivered waste will differ due to different extents of bacterial decayof the slaughter waste, and type of material depending on transport time and sea-son. The off-odor perceived at delivery will be a combination of volatile compoundsderived from body effluents (urine and feces), lipid oxidation and bacterial spoilageprocesses. Accordingly, the major components in the headspace will be volatileacids, aldehydes, ketones, sulfides and amines. With increasing onset of spoilage,the volatile secondary metabolic products (sulfides and amines) will be dominatingthe off-odor of the waste.Waste from animal slaughterhouses consists of blood and a slurry of particulate

(matter) slaughter waste with a high water content, which have been separated beforethey enter the plant. The different process steps of slaughter waste processing are: (a)the blood is coagulated by water vapor and the dry matter is separated; (b) the waterphase is recycled and the drymatter is mixed with the slaughter waste after the grinder;(c) the slurry with particulate matter is delivered by truckloads from different slaugh-terhouses to the processing plant and fed via a huge funnel into a grinder (particle sizeof 0.5 cm) representing the first processing step; (d) the waste is dried thereafter, byheating at 100 8C at atmospheric pressure for 20 minutes. The drying process de-creases the water content to 42%; (e) the material then goes into the dry smelter(autoclave) where it is heated to 136 8C with pressure up to 3.2 bar for another20 minutes. The pressure is decreased and the mass is vaporized until the remaininghumidity is 5–8%. The mass from the dry smelter contains about 40% fat and 60%


drymatter, which are sent through a press for separation. The final products are a purelipid phase and bone flour, which are used as raw material for animal feed production.

20.4.3

Monitoring Setup

The gas sensor arrays used in the bioprocess monitoring studies (study 1 and 2) wereequipped with a set of 10 metal-oxide semiconducting field effect transistor sensors(MOSFET), up to 19 metal oxide semiconductor sensors (MOS) and 1 CO2-monitorbased on infrared adsorption. The MOSFET sensors were produced in-house at Lin-koping University (Linkoping, Sweden) with different catalytic metal gates of Pd, Ptand Ir at metal film depths between 70 and 400 A. The MOS sensors were commer-cially available sensors of Taguchi (TGS) or FIS type fabricate. The electronic nose usedin case study 3 was a commercial on-line sensor array system (NST 3210, Nordic Sen-sor Technologies AB, Sweden) consisting of 10MOSFET and 5MOS sensors (Taguchitype).In the presented case studies the electronic noses had a built-in membrane pump

and a mass flow controller to supply the sensor array with a constant flow of gas at alltimes. Repetitive cycles of, alternately, reference gas and sample gas were measured inorder to be able to relate the sensor signal to a stable baseline value, and hence to obtainaccurate and reproducible measurements.In case studies 1 and 2 a compensator vessel formed the interface to the bioreactor

exhaust gas stream in order to compensate for minor variations in flow rate or gasconcentration and to trap condensation (see Fig. 20.1). The reference gas used wasthe same as the process air to the bioreactor (compressed and filtered air). The hu-midity of the reference gas was adjusted to approximately the same value as the bior-eactor exhaust gas by bubbling the reference gas through distilled and sterile water. Incase study 3, dehumidified and active charcoal filtered ambient air was used as refer-ences gas.

20.4.4

Signal Processing

The definitions of the signal parameters that have been extracted from the gas sensorsare illustrated in Fig. 20.2. The frequency of collecting the sensor signals was set to1 Hz in all case studies. The total measurement cycle time was 10 and 15 minutes incase studies 1 and 2, respectively. The interval for measuring the sample from thebioreactor exhaust gas was between 20 and 30 seconds. The mean value of the last20 seconds of the baseline measurement was taken as sensor baseline value foreach cycle. The sensor response, on-derivative and on-integral values were all calcu-lated relative to the baseline. The response is the average over the last 6-second samplemeasurement period. The on-derivative is the value of the fourth measurement pointof the sample measurement, and the on-integral is the average of the first 21 secondsof the sample measurement. The off-derivative and off-integral values were calculated


relative to the response, as the fourth measurement point after the sample measure-ment and the average over the first 21 seconds after the sample measurement, respec-tively.Measurement conditions used in case study 3 were as follows: storage experiment –

100 sec baseline, 10 sec sampling (12 ml sample volume) and 40 min recovery. Inaddition to the three batches of slaughter waste, also the background air was mea-sured repeatedly. Each of the waste batches was measured every 2.8 hour over 5.5days, i.e. 47 measurements of each sample batch; field experiment – 20 sec base-line, 40 sec sampling (50 ml sample volume) and 40 sec recovery. Between measure-ments of every new waste batch, both ambient air and water vapor were alternatelypumped over the sensors in order to flush the sampling tube and inlet system.Time between each new sample (truck delivery) lasted from 5 minutes up to onehour. Accordingly, the minimum recovery time was about 5 minutes. Average sensorresponses (signal height relative to baseline) of the last two replicate measurementswere used for the data analysis.

Fig. 20.2 Signal parameter extraction for gas sensors. The sensors are

continuously exposed to reference gas and interrupted by short

sampling periods (Sample gas/Reference gas). Response and baseline

values are calculated as the mean value over a defined time interval.

On- and Off-integral values represent times over which the signal is

integrated. The derivates on the rise and fall of the signal are the on- and

off-derivative values


20.4.5

Chemometrics

The structure of the ANN used in case study 1 was a one-hidden layer back-propaga-tion network with a sigmoidal activation function and one output node. In the storageexperiment of case study 3 the responses of the five MOS sensors were used as inputsto a back-propagation network. A four-hidden layer network with a sigmoidal transferfunction and three output nodes was used. Minimization of the network output errorwas in all cases performed using the Levenberg-Marquardt algorithm.For efficient sensor variable selection a forward selection algorithmwas used in case

study 2 [30]. The objective of this algorithm is to find a subset of the original sensorsignals that minimizes a selection criterion. The selection criterion is the predictionerror from a multiple linear regression model towards the desired model output (theprocess variable). A forward selection adds one variable at a time to the model until theselection criterion reaches a minimum.The PLS models in case study 2 were built using the NIPALS algorithm. All calcula-

tions in case studies 1 and 2 were performed using MATLAB� (The MathWorks Inc.,MA, USA) and PLS-toolbox for MATLAB� (Eigenvector Technologies, Manson, WA).The PLS calculation in study 3 was performed using The Unscrambler (v7.5, Camo,Trondheim).

20.4.5.1 Study 1: Estimation of Cell growth in Escherichia coli Fermentations

This study was performed using a recombinant Escherichia coli strain producing hu-man carbonic anhydrase. A total of five batch cultivations on a 2 L scale were carriedout with a fermentation time of 22 h. Details on this study can be found in Bachinger etal. [10].Investigation of the raw sensor signals from the gas sensor array reveals several

interesting aspects. In Fig. 20.3a, selected sensor signals, biomass and dissolved oxy-gen level for one of the batch processes are shown. The response pattern frommost ofthe sensors can be directly associated with the three phases of a typical batch process:the lag phase (0–2 h), a phase associated with growth (2–11 h) and the stationaryphase (11–22 h). This characteristic sensor pattern can be related either to the cellmetabolism or the physical parameters in the broth. For example, some sensors mir-rored the dissolved oxygen level in the broth, while the increase in MF8resp after6 hours occurred at the same time as the depletion of the carbon source in the med-ium. Since a nutrient rich 2 � LB-medium was used in this process metabolic activitydid not stop after carbon source depletion, which is reflected in the signal of, for ex-ample, MF6resp. The infrared sensor (IR) followed the cell mass evolution proportion-ally with time during the exponential phase of the fermentation.In order to estimate biomass ANN technology was used to relate the gas sensor

signal pattern to the cell mass in this process. A trial-and-error procedure was per-formed to identify the best set of input variables and the structure for the ANN.The input pattern that resulted in the lowest training error for the biomass estimationwas identified as MF1(resp), MF2(resp), MF3(resp), MF4(resp), MF7(resp), MF8(resp),MF9(resp), MOS3(resp) and IR(resp).


Fig. 20.3 (a) Selected gas sensor signals and the dissolved oxygen

concentration (pO2) from a 22 h Escherichia coli batch fermentation. (b)

ANN validation and off-line values for biomass in a 22 h Escherichia coli

batch fermentation


A 9-8-1 network with the biomass as network output was trained using the abovesensor signal parameters from four Escherichia coli fermentations. The result of themodel validation on the fifth fermentation is shown in Fig. 20.3b. In the figure,the off-line biomass values for the fermentation are compared with the estimated bio-mass values from the ANNmodel. Themean deviation between off-line and estimatedbiomass values was 0.043 gL�1 and the accuracy reached 1.46%.

20.4.5.2 Study 2: Physiologically Motivated Monitoring of Escherichia coli Fermentations

An on-line approach of monitoring the physiological changes of the cells in a biopro-cess is presented in this study. The basic idea was that for the task of a simple phy-siological state (PS) description it should be sufficient to focus on state identificationinstead of quantification. We are therefore not specifically interested in the exact va-lues given in the physiological variables instead we would like to trace fast changes inmetabolic state. A semi-quantitativemethod for PS identification is therefore proposedthat can be performed without the need for sensor calibration. In this method thetrajectory representation of the gas sensors is directly related to the physiological stateof the cell culture. Thereby the precise response height or intensity values of the sen-sors are not critical. Details on this study can be found in Bachinger et al. [15].The principle is explained on the basis of five 35 h fed-batch fermentations with a

recombinant Escherichia coli strain producing b-galactosidase. Figure 20.4a shows se-lected gas sensor signals, dissolved oxygen level and biomass for one fed-batch pro-cess. Similar sensor response characteristics can be observed in this fermentationcompared to study 1. Several of the sensors follow the dissolved oxygen concentrationand the stages of the fed-batch process can be clearly associated with the sensor re-sponse pattern.The strategy developed for PS characterization is relying on PLS methods. For every

physiological variable a specific PLS-model is calculated from selected sensor responsesignals in a standard fermentation. The latent variable with highest correlation towardsthe desired physiological variable is identified as the models output. The resultingPLS-models serve as the base models for respective physiological variable predictionsin subsequent fermentations. More specific, sensor signals from a new fermentationare projected on-line onto a defined PLS-model resulting in new latent variable scoresthat represent the physiological variable of interest.As first example the physiological variable ‘growth rate’ was modeled accurately by

this approach as can be seen in Fig. 20.4b. The sensor signals MF7(onder.), MOS3(off int.),MF7(abs resp.), MOS12(abs resp.), MOS16(abs resp.), MOS5(on int.), IR(resp.), MF5(abs resp.),MOS7(off der.), and MOS19(abs resp.) were selected by a forward selection method forPLS-model building [30]. The PLS-model was built with the data from the first fermen-tation and a latent variable was selected from the model by visual evaluation to repre-sent the growth rate. Both actual growth rate and the LV score from a new fed-batchprocess are shown in the plot. The arrows in the figure indicate the coincidence ofchanges in direction in time in both actual and modeled growth rate.A second model was calculated for the physiological variable ‘glucose uptake rate’.

Figure 4d shows LV scores for modeled physiological variables of growth and glucose


Fig. 20.4 (a) Selected gas sensor signals, dissolved oxygen concen-

tration (pO2) and biomass from a 35 h Escherichia coli fed-batch fer-

mentation. (b) Actual growth rate (GR) and latent variable (LV) re-

presenting growth rate in the E. coli process. (c) Actual growth rate and

glucose update rate values for the E. coli process. (d) Latent variables

representing growth rate and glucose update rate in the E. coli process


uptake rates, plotted next to each other. The selected sensor signals for glucose uptakerate were IR(resp.), MOS18(offder.), MOS12(on int.), MOS5(offder.), MF5(on int.), andMOS2(offder.). It can be seen that the changes in both trajectories occur at the same instances intime as the original physiological variables seen in Fig. 20.4c.

Fig. 20.4c, d


20.4.5.3 Study 3: Quality Control of a Slaughter Waste Process

The traditional quality criteria for animal feed products is based on the content of freefatty acids (FFA) in the lipid product. A high FFA value corresponds to a poor productquality. The objective of the study was to investigate whether the electronic nose tech-nology could be used to determine the quality of the waste rawmaterial before it entersthe processing plant. Mixing of raw material of different spoilage quality would resultin poor quality of the final product. One of the objectives was therefore also to use thetechnique to sort out waste of similar quality in terms of spoilage status and final FFAvalue.Two different experiments have been carried out. One was a small-scale experiment

for simulating the spoilage processes taking place in the waste during storage or trans-port on trucks to the waste processing plant. The second was a field experiment, mea-suring the truckloads of slaughter waste directly at delivery by the processing plant. Forthe storage experiment batches of fresh slaughter waste from pure pork, pure cattleand a mixture of both were investigated. The off-gas production from the bacterialdecay of the waste was monitored continuously. In the second experiment a qualitymonitoring was performed on the waste directly on the truckloads before they were fedinto the processing plant. The waste consisted of pure pork, pure cattle, mixture ofcattle and pork and pure poultry.

Storage experiment

Three batches of 30 L fresh slaughter waste were stored indoors in open tanks of 80 cmin diameter at 8 8C over 5.5 days. Each of the batches was covered by odorless plasticlids that were connected to tubing under the room ceiling, where the off-odors werepassively drained through an outside ventilation system into the open air. This wasdone to prevent contamination of off-odors between the batches. Off-gases were mon-itored continuously in the process with the electronic nose. The sampling tubes to thesensor array were located about 30 cm above the waste at the center of each tank.Typical sensor responses are shown for selected sensors in Fig. 20.5 for the pork

waste. The other sensors of the array showed a similar distribution over time forthe different waste types. There is a period of 62 hours (the bacterial lag phase) beforethe bacteria enter the exponential growth phase, which is reflected in a simultaneousincrease in the gas production. After 5 days the stationary phase was still not reachedwhen the experiment had to be terminated due to a high concentration of sulfides thatcaused poisoning of some of the sensors so they did not recover back to baseline. Aprincipal component analysis (PCA) plot based on the MOS sensor responses(Fig. 20.6) shows that the different categories of slaughter waste at the given condi-tions show different off-gas profiles throughout the whole measurement periodfrom the start and far into the exponential phase. At the end of the experiment thepork and cattle waste seem to become similar indicating a production of similaroff-gases at this stage of the bacterial decay process. In Fig. 20.7 the PCA plot is basedentirely on the MOSFET-sensor responses that show a slightly different distribution.At the start of the experiment the wastes are very similar. After some time, however,the gas sensor profiles become separated and proceed in different directions, which


corresponds to the onset of the exponential bacterial growth phase. The onset of thegrowth phase could also easily be perceived by the change in odor of the samples.During the lag phase the odor had a fresh note and at the onset of the growth phasecharacteristic strong off-odors were perceived related to increased production of sul-fides. Later on, the wastes are clearly separated and only the pork and the mixturewastes are becoming similar at the end of the measurement period, whereas the cattlewaste stays significantly different in the off-gas profile. PLS modeling was used toobtain freshness prediction models based on the sensor responses and storagetime. The results are listed in Table 20.2. The storage time could be predicted withan error of about 5 hours, which represents an error of 3.7–4.2%. A back-propagationneural network model has been applied to the sensor responses in order to obtain aprediction model for classification of the different waste types. The responses of thefive MOS sensors were used as inputs to the network. A four-hidden layer networkwith a sigmoidal transfer function and three output nodes was used. The outputsrepresented the three waste classes (C, P and P þ C). 30% of the measurementdata were used as the training set and the rest was used for validation. The resultsare shown in Table 20.3. A high classification rate (96–98%) was obtained for allthree waste types. For the pork and cattle waste there was only one measurementthat was wrongly classified as belonging to the mixed pork and cattle waste, whereastwo measurements of the mixture (P þ C) was undefined in that they could not befitted to any class.

Field experiment

The measurement device was set up in a room next to the feeding funnel. A 7 mstainless tube (1.5 mm inner diameter) that was connected to the sampling inlet of

Fig. 20.5 Selected sensor responses from pork measurements in the storage experiment


the instrument led outdoors and positioned at about 3 m height above ground, wasused for sampling the off-gases directly on the truckload. The inlet of the samplingtube was positioned inside, below the cover of the truckload into the headspace abovethe slaughter waste where measurement took place. The indoor temperature aroundthe measurement device was consequently higher than outside, where the sampling

Fig. 20.6 PCA plot based on the MOS sensor responses from the

storage experiment (C ¼ cattle, P ¼ pork, P þ C ¼ pork and cattle

mixture)

Fig. 20.7 PCA plot based on the MOSFETsensor responses from the

storage experiment (C ¼ cattle, P ¼ pork, P þ C ¼ pork and cattle

mixture)


took place, hence preventing condensation of gases in the sampling tube. Ambient airwas dehumidified, coal filtered, and used as reference air in order to keep a stablesensor baseline.Three replicate measurements were performed on each sample (on truckload de-

livery). The samples consisted of pure pork, cattle and poultry waste, and mixturesof cattle and pork waste. Due to different transport times the waste was deterioratedto a different extent representing different states of bacterial spoilage. In parallel to theelectronic nose analysis at delivery of the waste rawmaterial, the quality of thematerialwas also assessed by giving it a sensory score according to color, odor quality (bad,good), and intensity. The waste material differed in quality from fresh “pleasant” smel-ling to different extents of spoiled and unpleasant (sulfide, ammonia like) smellingsamples due to bacterial spoilage processes.Figure 20.8 shows the output signals of selected sensors for the different waste types

measured. It is seen that the response is increasing with increasing FFA values of finalproduct. Increasing sensor signals were also in accordance with the sensory assess-ment of the raw material. Fresh material having good odor and low odor intensityshowed low sensor responses in comparison to samples with unpleasant off-odorsand discolor that showed increasing sensor responses. The off-odors and discoloringof the pork and cattle waste was similar to what had previously been observed duringthe storage experiments.The pork samples showed lower responses at low FFA values compared to the cattle

samples. PLS regression between sensor responses and FFA values was used to obtainprediction models for the FFA value for the different types of waste. The results aresummarized in Table 20.4. The results show that the quality of the waste raw materialis correlated with the quality of final product in terms of the FFA values of the lipidproduct.

Tab. 20.2 Results from PLS regression between the MOS sensor

responses and storage time from storage experiment. Number of

samples used for the model (n), correlation coefficient (r), and root

mean square error of prediction (RMSEP) in hours

Waste n r RMSEP

Cattle 47 0.98 5.6

Pork 47 0.99 5.1

Pork/cattle mix 47 0.99 5.0

Tab. 20.3 Results from ANN classification of different types of

slaughter waste. Number of samples is given in brackets

Predicted class Pork þ Cattle Pork Cattle Undefined Total

Pork þ Cattle 96% (46) 0% (0) 0% (0) 4% (2) 48

Pork 2% (1) 98% (47) 0% (0) 0% (0) 48

Cattle 2% (1) 0% (0) 98% (47) 0% (0) 48


20.4.5.4 Discussion

Several conclusions can be drawn from the presented bioprocess monitoring studies.Specific compounds (cell metabolites) that are non-volatile or of a very low concentra-tion below the detection limit of the gas sensor array can be monitored indirectly bymeasuring the vapor phase. The results suggest that it may be possible to predictmetabolites of a biological system because they are correlated with volatile compoundsvia stoichiometric or other complex correlations. In both case studies all the cell tran-sition states could be predicted with a constant and high accuracy including the verysmall biomass values at the beginning of the cultivation in case study 1 Even thoughthe result was obtained at low biomass concentration the derived neural network mod-el gave an accuracy similar to that for conventional wet chemical techniques. Physio-logical state changes could be tracked in case study 2. It was not necessary to achievequantitative resolution of the PS, instead fast cell transition states were monitored in a

Fig. 20.8 Sensor responses of selected sensors for cattle, pork, cattle

and pork mixture and poultry slaughter waste samples with increasing

content of free fatty acids (FFA)

Tab. 20.4 Results from PLS regression between sensor responses and

FFA values from field experiment. Number of samples used for the

model (n), correlation coefficient (r), and RMSEP as a percentage of the

measurement range

Waste n r RMSEP

Cattle 30 0.95 12.5

Pork 20 0.83 12.5

Cattle/pork mix 20 0.92 6.2

Poultry 16 0.93 5.0


semi-quantitative approach. Themethod has the advantage that a representativemodelcan already be built on the basis of a single fermentation. Since quantitative informa-tion is not acquired, drift counteraction and calibration problems are not a major com-plication. The possibilities of on-line and non-invasive operation of the measurementmake it a simple and fast method for the monitoring of industrial bioprocesses.The results from the storage experiment in case study 3 suggest that the electronic

nose has a potential for sorting out slaughter waste based on the extent of spoilage. Thebacterial transient responses expressed in the sensor signals were strongly correlatedto characteristic odor differences varying between fresh and spoiled odor. In addition,the results indicate the possibility to determine the purity of the slaughter waste interms of animal content. Results obtained in the field experiment indicate that elec-tronic nose technology can have a potential in quality control of slaughter waste interms of spoilage status of the waste before it is fed as raw material into the wasteprocessing plant. In addition, it could be demonstrated that the quality in terms ofFFA values of the final product can be predicted by analyzing the odor of the rawmaterial before it enters the process. This demonstrates the importance the qualityof raw material may have for the quality of final product.

20.5

Future Prospects

Even thoughmany attempts aremade to employ electronic noses for quantitativemon-itoring, direction of application focuses mainly onto the more successful qualitativemonitoring approaches. This favors biological process monitoring, since detection ofprocess abnormalities or cell/process states does not rely purely on quantitative infor-mation. Further development and success of the electronic nose technology in processmonitoring applications would profit greatly from sensors with improved stability,selectivity, less signal drift, and faster update speeds. There is a rapidly advancingresearch and development going on both on sensors and instrument hardware andsoftware in order to enhance selectivity, sensitivity and reproducibility of the gas sen-sors. Application-specific sensor selection, improved calibration modeling andadapted pattern recognition analysis will enable us to expand the area of applicabilityeven further.It becomes clear from the presented material that this technology has a potential for

process control by monitoring the volatile compounds produced throughout a processthat will allow fast/rapid detection of process abnormalities/deviations in order toensure the final product quality. However, since this technology does not provide spe-cific chemical information due to the limited selectivity of chemical sensors, it mostlyprovides little insight into the causes when deviations occur. For some applications themonitoring of the vapor phase may therefore not be sufficient to obtain the essentialprocess information and additional sensors would be required. Sensor fusion withother on-line/at-line measured process parameters could, especially in biopro-cesses, lead to a better understanding of the signal responses. A fully automated mul-ti-sensor system methodology consisting of different sensor technologies to monitor


the essential process parameters required for assuring the quality of both rawmaterial,process and final product may therefore be the future solution for some applications[31]. Gas-sensors would make up a vital part of such a multi-sensor system. Thisthought leads to the integration of the electronic nose into knowledge-based systemssupporting process control [32]. Process reproducibility and in turn product qualityand safety could be improved in the first place and the technology could even be usefulin supporting process development. This may be realized in industry in the not sodistant future.

Acknowledgments

Drs Carl-FredrikMandenius, Per Martensson, Tomas Eklov, Helena Liden andMartinHolmberg are acknowledged for their valuable contributions to the development of theelectronic nose technology for bioprocess monitoring. Process engineer Oliver Tomicis acknowledged for his contribution in the slaughter waste field study.

References

1 T. D. Gibson, O. Prosser, J. N.Hulbert, R.W.Marshall, P. Corcoran, P. Lowery, E. A.Ruck-Keene, S. Heron. Sensors Actuators B1997, 44, 413–422

2 M. Holmberg, E. G. Hornsten, F. Winquist,I. Lundstrom, L. E. Nilsson, F. Gustafsson, L.Ljung. Biotechnol. Techn. 1998, 12(4), 319–324

3 T. Borjesson, T. Eklov, A. Jonsson, H.Sundgren, J. Schnurer. Cereal Chem. 1996,73, 457–461

4 F. Winquist, E. G. Hornsten, H. Sundgren,I. Lundstrom. Meas. Sci. Technol. 1993, 4,1493–1500

5 W. Ping, T. Yi, X. Haibao, S. Farong. Biosens.Bioel. 1997, 12, 1031–1036

6 C. F. Mandenius, I. Lundstrom, T. Bachin-ger. 1st Eur. Symp. Biochem. Eng. Sci. 1996,104

7 C. F. Mandenius. Adv. Biochem. Eng. Bio-technol. 1999, 66, 65–83

8 T. Bachinger, C. F. Mandenius. Trends Bio-technol. 2000, 18, 494–500

9 C. F. Mandenius, T. Eklov, I. Lundstrom.Biotechnol. Bioeng. 1997, 55, 427–438

10 T. Bachinger, P. Martensson, C. F. Mande-nius. J. Biotechnol. 1998, 60, 55–66

11 H. Liden, T. Bachinger, L. Gorton, C. F.Mandenius. Analyst 2000, 125, 1123–1128

12 T. Bachinger, U. Riese, R. K. Eriksson, C. F.Mandenius. Bioproc. Eng. 2000, 23 (6), 637–642

13 C. F. Mandenius, A. Hagman, F. Dunas, H.Sundgren, I. Lundstrom.Biosens. Bioel. 1998,13, 193–199

14 T. Bachinger, U. Riese, R. K. Eriksson, C. F.Mandenius. J. Biotechnol. 2000, 76, 61–71

15 T. Bachinger, C. F. Mandenius. Eng. in LifeSciences 2001, 1, 33–42

16 T. Bachinger, U. Riese, R. K. Eriksson, C. F.Mandenius. Biosens. Bioel. 2002, 17, 395–403

17 P. K. Namdev, Y. Alroy, V. Singh. Biotechnol.Prog. 1998, 14, 75–78

18 T. Bachinger, C. F. Mandenius, G. Striedner,F. Clementschitsch, E. Durrschmid, M.Cserjan-Puschmann, O. Doblhoff-Dier, K.Bayer. Chem. Technol. Biotechnol. 2001, 76,885–89

19 C. F. Mandenius, H. Liden, T. Eklov, M.Taherzadeh, G. Liden. Biotechnol. Prog. 1999,15, 617–621

20 C. Cimander, T. Bachinger, C. F. Mande-nius. Biotechnol. Prog. 2002, 18, 380–386

21 T. Bachinger, H. Liden, P. Martensson, C. F.Mandenius. Seminars Food Anal. 1998, 3,85–91

22 A. K. Abass, L. D. Coper et al.. ElectronicNoses & Sensor Array Based Systems, Designand Applications. W. J. Hurst. (Ed.) 1999,Pennsylvania, USA, Technomic PublishingCompany Inc

23 M. C. Horillo, I. Sayago et al.. ISOEN 2000,Brighton, UK


24 A. Kaipanen. Electronic Noses in the FoodIndustry, A state of the art symposium 1998,49–52, Stockholm, Sweden

25 P. Mielle, F. Marquis. Sensors Actuators B2001, 3795, 1–7

26 J. E. Haugen, O. Tomic, F. Lundby, K. Kvaal,E. Strand, L. Svela, K. Jørgensen. In: Elec-tronic Noses and Olfaction 2000, ISBN0750307641, pp. 265–271

27 M. Fryder, M. Holmberg, F. Winquist, I.Lundstrom, Proc. Transducers ’95 and Euro-sensors IX 1995, Stockholm, 683–686

28 J. E. Haugen, O. Tomic, K. Kvaal. Anal.Chim. Acta 2000, 407, 23–39

29 T. Artursson, T. Eklov, I. Lundstrom, P.Martensson, M. Sjostrom, M. Holmberg. J.Chemometrics 2000, 14, 711–723

30 T. Eklov, P. Martensson, I. Lundstrom. Anal.Chim. Acta 1999, 381, 221–232

31 V. Steinmetz, F. Sevila, V. Bellon-Maurel. J.Agric. Engng. Res. 1999, 74, 21–312

32 M. D. Naish, E. A. Croft. Mechatronics 2000,10, 19–51


21

Food and Beverage Quality Assurance

Corrado Di Natale, Roberto Paolesse, Arnaldo D’Amico

Abstract

Among the numerous applications of electronic nose technology, the analysis of food-stuff is one of the most promising, and also the most traveled road towards industrialapplications for this technology. Because human senses are strongly involved in anindividual’s interaction with foods, the analysis of food provides an excellent fieldto compare the performances of natural and artificial olfaction systems. Becausethe electronic nose is non-destructive and directly correlates, in principle, to theway the consumer perceives food products, it is a good candidate for use as an evalua-tion tool for quality assessment.In this chapter, a review of the applications of the electronic nose (and its liquid

counterpart the electronic tongue) to the evaluation of quality in foods and beveragesis given. Also included is an example case study: the measure of the quality of fish. Theexperiment described was performed with an electronic nose developed by theauthors, a description of which is also provided in the text.

21.1

Introduction

Food analysis is a complex discipline involving many different basic sciences. A multi-tude of different principles of instrumental analysis is currently being investigated andused for the analysis of foods and beverages. At the industrial level, the objectives ofthese measurements are directed towards safety (e.g. the search for contaminants),biochemical composition (to identify the basic constituents), and the effects offood treatment and processing. For each of these concerns, a number of techniquesare currently being studied and developed. They span from the classical analyticalchemistry to the more advanced diagnostic imaging techniques such as nuclear mag-netic resonance (NMR) [1]. In order to optimize the evaluation of quality and to en-hance themarketability of the products, there is an increasing interest for non-destruc-tive methods to assist in the complex classification of fresh products.


505505

Besides the classical objectives mentioned above, recently there has been increasedemphasis on the certification of quality. In a time of increasing globalization of man-ufacturing and markets, quality improvement is becoming one of the latest trends infood marketing. For instance, the consuming public wants to recognize those protec-tion classified products that may be identified with certain animal or vegetable speciesor with particular production methods. In this regard, we are witnessing the extensionof the same classification criteria traditionally adopted for wines, to foodstuffs likemeats and fruit.Quality is a global character of a food – it is concerned with all aspects of the inter-

action between food and consumers. Hence, the perfect instruments to determinefood quality are the human senses. Actually, trained panels of tasters are used toset the criteria of quality, to assess the quality of food, and to help in the developmentof new products.Although the science of food assessment by a human panel is well understood and

accepted as the ‘gold’ standard of sensory analysis, the actions of panels are affected byimprecision, are scarcely repeatable, and should not be used for routine operations.One of themajor difficulties with panels is the comparison of analysis done at differenttimes. For instance, the evaluation of wines performed in two different years may giveinconsistent results. Although these limitations are widely known, the importance ofpanels is growing. As an example, the European Union employs human panels toassign market values for olive oils.For these reasons, it is important to replace evaluation by panels with an accurate

instrumental technique that can performmeasurements in real-time and generate thesame information as a panel, but in a reproducible and stable way.An example of the complexity of the measure of food quality is demonstrated by the

case of fruit. Currently, fruit quality is assessed by measuring mechanical properties(texture, firmness, and acoustic properties) [2, 3], external images (visible and infrared(IR)) [4], internal images (NMR) [5], electrical properties (complex impedance) [6],sound-wave propagation [7], reflectance spectroscopy (visible and infrared) [8], and,of course, sensory analysis as the reference method to which the instrumental mea-surements are correlated [9]. Each of these techniques provides partial information,and only through the integration (fusion) of all of them is it possible to achieve quality.Many measurements are made by creating a headspace above a fruit sample. This

headspace is studied with conventional analytical chemistry equipment (such as gaschromatography and mass spectroscopy). Correlation between the quality aspects offoods and beverages and the composition of their headspaces (both in quantitative andqualitative terms) has been found in many different cases. Specific biochemical mod-eling of the production of volatile compounds is also available in many cases (e.g. forfish [10]). Despite these encouraging findings, the measure of the composition ofheadspaces has not resulted in any practical industrial instrumentation to measurefood quality. Currently, the information from the headspace is mostly exploited bythe senses of human panels, who provide their judgments about the quality of pro-ducts.The development of artificial olfaction machines (electronic noses) that are easy to

use, portable, and provide a simplified sampling method, appears extremely appealing

21 Food and Beverage Quality Assurance506

in this field. It could make possible the practical exploitation of a fundamental sourceof information to determine food and beverage quality. Recently entering the scene isthe liquid counterpart of electronic noses, namely arrays of sensors working in solu-tion: the so-called electronic tongue. Such devices are of extreme interest to research-ers who want to characterize beverages and, in some cases, foodstuff.For all these reasons, food and beverage quality is the most practiced application of

the electronic noses. Research results reported in the literature are finding their wayinto industrial practice. In the next section, a survey of food application studies avail-able in literature is given with a discussion of general arguments about using electro-nic noses in this field. Following that, a selected case study will be presented with adescription of an electronic nose developed by the authors.

21.2

Literature Survey

Table 21.1 lists a number of foods and beverages that have been the subject of elec-tronic nose analysis. Many of the papers appearing in this area originated from thedesire of electronic nose researchers to test the recognition capabilities of their sensorarrays, so that many of these papers deal with questions of little interest to food in-dustries. This is the case, for instance, in the classification of wines of different vari-eties. Although researchers have focused on this question, the wine industry wouldrather study differences occurring among wines of the same variety.For each different foodstuff, attention has been devoted to particular aspects. In the

case of meat, the effects of processing and the microbial quality have been investigated[11, 12]. As a specific example, boar taint detection is an important factor in the qualityof pork meat [13]. It is interesting to note that the boar taint is due to the presence ofandrostenone, this is a typical compound present in human male sweat. Electronicnose sensitivity to androstenone helped in the analysis of human skin for medicaldiagnosis purposes [14]. Because the presence of this compound is related to the sex-ual status of the animal, the counteraction of the boar taint is achieved by the castrationof the animal. The relation between castration and meat quality has been found inother animals; for instance, it has been used in evidence by an electronic nose studyof South American camelids meat [15].In the case of fruit and vegetables, attention has mostly been given to measuring the

headspace composition variations due to post-harvest processes [16–20] and theircorrelation with the presence of defects, such as mealiness in apples [21]. From anindustrial point of view there is also a strong requirement for the identificationand selection of cultivars. Recently, the problem of the identification of the optimalharvest time has been addressed in the case of apples, with an electronic nose obtain-ing results comparable with the most widely used destructive methods [22].Olive oil is another special case where electronic noses are requested to be compe-

titive with sensory analysis panels. The European Community requires that humanpanels assign each olive oil to a market-value category [23]. The use of an economicalinstrument able to overcome the scarcity of trained human panels has become an


Tab. 21.1 List of some applications of electronic noses to food and

beverage quality. Some electronic tongue applications are also listed

Food Description Reference

Meat Fermentation of sausages De Meyer et al, 2000 [61]

Eklov et al., 1998 [62]

Processed chicken meat Pfannahuser, 1999 [11]

Packaged beef meat Blixt et al., 1999 [12]

Ground meat Di Natale et al., 1997 [33]

Ground meat Winquist et al., 1193 [34]

Boar taint Annor-Frempong at al., 1998 [13]

Alpaca and llama meat quality Neely et al., 2001 [15].

Fruit and vegetables Aroma of pears Oshita et al., 2000 [63]

Quality of tomatoes Sinesio et al., 2000 [39]

Maul et al., 1998 [16]

Bacteria infection in potatoes De Lacy Costello et al., 2000 [27]

Quality of straweberries Hirschfelder et al., 1998 [64]

Ripeness detection Benady et al, 1995 [17]

Peaches: correlation with sensory analysis Di Natale et al., 2000 [40]

Apple ripeness Hines et al., 1999 [18]

Apple picking time Saeveles et al., 2001 [22]

Quality of apples and citruses Di Natale et al., 2000 [21]

Banana ripeness Llobet et al., 1999 [19]

Blueberries – quality sorting Simon et al., 1996 [20]

Vegetable oils Defects and rancidity of olive oil Aparicio et al., 2000 [24]

Classification of vegetable oils Martin et al., 1999 [65]

Classification of olive oil Stella et al., 2000 [66]

Di Natale et al. 2000 [68]

Cereals Mite infestation Ridgway et al, 1999 [25]

Microbial quality Jonsson et al, 1997 [26]

Odor classification Borjesson et al, 1996 [67]

Wine Toasting of barrels Chatonnet, 1999 [69]

Vintage years Di Natale et al., 1995 [70]

Vineyards of production Di Natale et al., 1996 [71]

Denomination (electronic tongue) Legin et al. 1997 [72]

Correlation with sensory analysis

(electronic tongue)

Legin et al., 1999 [37]

Correlation with chemical analysis

(electronic nose and tongue)

Di Natale et al., 2000 [38]

Vinegar Anklam et al., 1998 [73]

Dairy products Cheese ripening Schaller et al., 1999 [74]

Off flavors in milk Marsili, 1999 [75]

Cheddar cheese aroma Muir et al., 1997 [76]

Aroma of UHT milk Di Natale et al., 1998 [35]

Milk freshness (electronic nose and tongue) Di Natale et al., 2000 [77]

Milk freshness (electronic tongue) Winquist et al., 1999 [36]

Coffee Aroma discrimination Gretsch et al., 1998 [78]

Discrimination of blends Gardner et al., 1992 [79]

Discrimination of blends (electronic tongue) Fukunaga et al., 1996 [80]

Discrimination of blends (electronic tongue) Legin et al. 1997 [81]

Brewery Aroma detection in brewery Tomlinson et al, 1995 [82]

Flavor detection Pearce et al., 1993 [83]


urgent issue. In this direction, the detection of defects and rancidity (the two maindescriptors of human panel scores) by means of an electronic nose represents a po-sitive result [24].Other interesting applications of great social and industrial relevance are those re-

lated to the safety of food. As an example, the infestation of mites in cereals [25], themicrobial quality of grain [26], and potatoes [27]. The detection of spoilage processes infish [28–32], meat [33, 34], andmilk [35, 36] are also of great importance most of all forprocessing industry.Food analysis also offers the possibility to compare the electronic nose evaluation

with those of expert panels, namely with the human senses at their best. This parti-cular field is fully detailed elsewhere this book. Here it is interesting to note that inthose applications where sensory analysis has a long and established tradition, thedescriptor used by panelists are so specialized that poor correlation with electronicnose data is found. Typical examples are found in wine [37, 38]. In other cases,when more simple descriptors are used that are less involved with fine human per-ceptions but rather linked to general quality, the correlation is found to be much better[39, 40]. This suggests that to pursue the utilization of electronic noses a reformulationof sensory profiles is, in some cases, perhaps necessary.From amethodological point of view, all these applications can be classified into two

main categories: static classifications and dynamic classifications. Static classificationis related to those applications where the electronic nose is expected to recognize sam-ples of foods and beverages as belonging to definite classes. Dynamic classificationconsiders the capability of electronic noses to monitor the evolution of foods fromthe fresh product. Often in this case, samples are represented along a ‘freshness lad-der’, going from perfectly fresh up to the state of non-edibility.Many of the applications listed in Table 21.1 belong to two opposite classes. The first

is the class of studies done by electronic noses researchers. In these papers, the choiceof the application and the sample treatment are often nave. Also, in some cases, greatattention is devoted to sensor development and sometimes to data analysis, so theresults are of little interest to food scientists. However, there are studies by food scien-tists using commercially available electronic noses. In these cases, the major focus isdevoted to the sample, with insufficient attention being paid to the sensors and dataanalysis, resulting also in reports with little practical use. In those cases where both theresearchers and industrial scientists co-operate, the most promising results areachieved.

Tab. 21.1 Continued

Food Description Reference

Fishes Trout freshness Schweizer-Berberich et al., 1994

[29]

Freshness of cod Di Natale et al., 2000 [30]

Cod-fillet storage time Di Natale et al., 1996 [31]

Freshness of capelin Olafsdottir et al., 1997 [32]

Spirits Sake (electronic tongue) Arikawa et al., 1996 [84]


21.3

Methodological Issues in Food Measurement with Electronic Nose

From the point of view of measurement methodology, electronic nose measurementshave some peculiar issues to be considered. In particular, it should be clear to theelectronic nose user that the sensor’s signal is a combination of the sensor sensitivityand the concentration of volatile compounds. Preliminary knowledge of these twoquantities is a fundamental pre-requisite to foresee the meaning of the data obtainedfrom the electronic nose. The sensors used should not be very sensitive to volatilescarrying low information about the sample under measurement. A typical exampleof this is found in olive oil, which is characterized by a large difference betweenthe compositions of the oil and its headspace. Dominant compounds in the headspaceare methanol and ethanol, whose presence in liquid is scarce and of no importance todefining the oil characteristics. On the other hand, those substances responsible forthe sensory properties (e.g. hexanal, trans-2-hexanal, and ethylacetate among theothers), and which are abundant in the oil due to their high boiling points, are foundat low concentrations in the vapor phase [41]. In this situation, sensors with highsensitivity to alcohols (e.g. metal-oxide semiconductors and conducting polymers)may give rise to signals that are poorly correlated with the relevant properties ofthe samples. On the other hand, the sensor nose should be sufficiently sensitive tobe able to capture the variations of relevant compounds in the different classes ofthe inspected samples.Environmental parameters, such as temperature, may greatly affect, directly or in-

directly, the sensor responses. We can call direct disturbances those related to thesensitivity of the sensors to the environmental parameters, whereas indirect distur-bances are those concerned with the effects of the environment on the samples undertest. This last aspect is associated with the performances of the samplingmethodology.Generally, attention is paid to insulating the sensors from the actions of the environ-ment, e.g. with proper temperature conditioning, making the direct disturbances al-most negligible. On the other hand, it has to be clear that what the electronic nosereally measures is an image of a solid or liquid foodstuff. The image (i.e. the composi-tion of the headspace) may be, in some cases, very different from the sample itself;furthermore, it is strongly dependent on the environmental parameters.The concentration in the headspace of a compound present, for instance, in a liquid

phase, is related to the vapor pressure and to the liquid phase concentration of thecompound, and is a function of the temperature. This means that more volatile com-pounds tend to be more abundant in the headspace than their relative abundance inthe sample. Furthermore, the headspace changes dynamically with the variation oftemperature. It is well known that for each foodstuff an optimal temperature existsat which the richest expression of the aroma is achieved. A classic example of thisis found in red wine and spirits such as cognac.The framework outlined above holds for ideal solutions, namely those for which the

mixing enthalpy is zero. In this case Raoult’s law applies and changes of temperatureproduce a scaling of all the headspace concentrations [42]. For non-ideal liquids (e.g.water-ethanol mixtures) significant deviations from Raoult’s law can occur, and tem-


perature variations result in a progressive distortion of the headspace composition.Foods are, almost always, complex and non-ideal mixtures. Therefore, samplinghas to be carefully designed, and when sensory analysis is involved as a referencemethod, the use of the same temperature range is a constraint to achieving significantresults.Optimal sampling systems should completely isolate the sample from the outside

environment. In practice this condition is not completely fulfilled, and changes of theenvironmental parameters results in variations of both quantity and quality of theheadspace. These give rise to an additional signal source that can sometimes comple-tely hide the resolution of the electronic nose. The straightforward way to counteractthe problem is to increase the performance of the sampling system, improving thesample temperature conditioning and using synthetic carriers. On the other hand,portability and economic requirements are in contrast with a sampling system thatis too sophisticated. It is worth noting that, except for a few exceptions, previouswork did not pay sufficient attention to the difference between the intrinsic sensordrift and the disturbances induced by the experimental set-up [43].For some food, the interaction with the environment can also induce irreversible

modification of the sample itself. A typical example of this effect is found in winethat is gradually oxidized when exposed to air. As a result of the effect, successivemeasurements of the same sample are not reproducible. A way to avoid this problemin wine consists of introducing two needles in the cork and using nitrogen as a carrierto sample the bottle headspace. The use of nitrogen as the carrier does not modify thechemical state of the wine [37]. With this arrangement, the wine is measured withoutopening the bottle.

21.4

Selected Case

As an example of applications in food quality analysis, the freshness of fish will bedescribed in some detail. This example is concerned with measurements performedwith an electronic nose conceived, designed, and fabricated by the authors at the Uni-versity of Rome ‘Tor Vergata’. These activities, started in 1995, resulted in a full op-erative instrument named LibraNose in 1999. In the following, a detailed descriptionof the instrument is given, followed by the selected case study.

21.4.1

LibraNose

LibraNose is based on an array of thickness shearmode resonators (TSMR) also knownin literature as quartz microbalance (QMB) sensors. The chemical sensitivity is givenby a molecular film of pyrrolic macrocycles (mostly metalloporphyrins and similarcompounds). In the current configuration eight sensors are used [44].


The most well known pyrrolic macrocycles are porphyrins and pthalocyanines. Thesensing properties of phtalocyanines have been studied extensively in the past [45].Porphyrins have only rarely been used, however, and even then only their optical prop-erties were exploited to fabricate sensors for working in solutions. In spite of this,porphyrins are among the most important molecules in nature, their functions ascomplexing ligands or redox catalysts are essential for all organisms [46]. The mimick-ing of their biological functions in an electronic nose sensor array has been so attrac-tive that these molecules have become the fundamental component of sensor system.Figure 21.1 shows the basic porphyrin molecule.A number of features make porphyrins eligible as a ‘sensing material’ able to detect

the volatile organic compounds. Porphyrins are rather stable and their properties canbe finely tuned by simple modifications of their basic molecular structure. The coor-dinated metal, the peripheral substituents, and the structure of the macrocyclic ske-leton influence the coordination and the related sensing properties [47, 48].The adsorption properties of solid-state porphyrins are characterized by large sen-

sitivities and wide selectivities: both of these features are particularly appealing forelectronic nose applications. While the wide selectivity is generally related to weakinteractions (such as Van der Waals force and hydrogen bonding), an additionalterm, due to the coordination of analytes, has be taken into account. Both the inter-actions are expected to co-operate.The double interaction is expected to give rise to a non-linear adsorption isotherm

resulting from specific p-p-interaction between the aromatic systems of porphyrin andan aromatic analyte (such as benzene). This double interaction has been recently in-troduced to model interactions in analog molecules [49]. This interaction takes place atlow concentrations of benzene and is ruled by a Langmuir isotherm. At higher con-centration, after the saturation of the specific sites, only the non-specific adsorptionoccurs and the shape of the isotherm becomes linear (Henry-type behavior).In general, the selectivity frame of metalloporphyrins towards different analytes

depends on several factors, such as peripheral substituents, solid-phase packing, de-position techniques and so on. Among them, a key factor is the metal coordinated to

Fig. 21.1 The basic porphyrin molecule.

The molecule can be functionalized by adding

lateral substituents at the R R 0 positions,

and a metal ion at the core of the ring


the porphyrin ligand; coordination of the analyte to the central metal in this case con-tributes significantly to sensing material-volatile compound interactions. The strengthof these interactions can be broadly predicted by the hard-soft acid-base theory(HSAB): hard acid (metals) prefers to interact with hard base (ligands) and vice ver-sa. In our case, for example, Cr, Mo and V porphyrins (containing hard metals)strongly bind hard ligands, such as alcohols or organic acids, while soft metals (Cuand Ni for example), prefer to bind soft ligands, such as sulfur compounds.In order to be exploited as a sensor, the porphyrins need to be deposited as a solid

film onto a substrate. Different techniques are available for this purpose and, amongthem, the following have been used: solvent casting, Langmuir-Blodgett [50, 51], self-

Fig. 21.2 The figure shows a PCA biplot of scores and loadings of an

experiment aimed at evaluating the volatile organic compounds

(VOCs) discrimination of arrays of porphyrin-based QMB sensors.

Scores are indicated by circles and loadings by crosses. Loadings from

1 to 7 are related to a tetraphenylporphyrin functionalized with different

metals (in order: cobalt, molybdenum, copper, iron, vanadium,

nickel and chromium) while from 8 to 14 the same cobalt-tetraphe-

nylporphrin with different functional groups at lateral positions.

The figure shows that metal ions are responsible of a different behavior

of the sensitivity. It is also worth noting that compounds are separated

in four main groups: amine, aromatic, alcohol and acid, and aldehyde

and alkane. The separation indicates the way VOC interact with

porphyrin film [76]


assembled monolayers [52, 53], and electropolymerization. The adsorption of analytesinto solid-state porphyrin layers induces the variation of physical quantities. Each ofthese quantities can be transformed in an electrical signal matching the porphyrinlayer with a proper transducer. Porphyrin-based sensors have been demonstratedwith different basic transducers such as TSMR [47], surface acoustic wave [54], con-ductivity [48], work function [55], and optical [56].TSMR sensors have been chosen for the implementation of a porphyrin-based elec-

tronic nose. These sensors consist of a thin slab of crystalline quartz, cut along a cer-tain symmetrical axis (usually the crystallographic AT direction is used) to obtain ma-terial able to sustain bulk electroacoustical oscillation at frequencies from 5 to 30 MHz[57]. The quartz property that makes it interesting as a sensor is that the resonancefrequency is, in a limited linear range, inversely proportional to the mass gravitatingonto the surface of the quartz. This behavior is exploited to turn the quartz into achemical sensor when some chemically interactive material, able to capture moleculesfrom the environment, is used as a coating.

Fig. 21.3 The currently availa-

ble versions of LibraNose

Tab. 21.2 List of the main features of the LibraNose instrument

Sensors Eight thickness shear mode resonators, fundamental frequency:

20 MHz

Sensor chamber Stainless steel, volume: 10 cm�3

Internal tubing Stainless steel

Pneumatic components Peristaltic pump, flux:0–0.2 sccm

Three two-ways electrostatic valves

Sample channels 2 input (sample and cleaning) 1 output

Electronics Eight ‘Pierce Oscillators’ at large dynamics

Motherboard: microcontroller (Motorola HC05) and programmable logics

(Xylinx)

Surface-mounted components

RS232 serial connection to an external computer

Software Cþþ builder for MS/Windows


First studies on a porphyrin-coated QMB showed the fundamental properties ofthese materials, namely the role played in defining the performances of the sensorby the metal, both in metalloporphyrin complexes and the lateral substituents [47].Results always confirmed a wide selectivity range that can be adjusted by changingthe metal and the peripheral substituents. This property of these sensors satisfiesthe general requirement of sensors to be employed in electronic noses. Figure 21.2shows, as a PCA score plot, the ability of the LibraNose to distinguish among differentvolatile compounds.Figure 21.3 shows the LibraNose. The instrument is linked to an external computer

that manages measurements, data collection, and analysis. Pneumatic components(pump and valves) are installed on-board to provide the necessary sample deliveryto the sensor chamber. Table 21.2 gives the technical specification of the electronicnose.

21.4.2

Case Study: Fish Quality

For fish it is important to determine the freshness degree, defined as the number ofstorage days at a certain temperature. For this kind of product, issues such as thedistinction between fresh and thawed samples and the maintenance of a constanttemperature during storage are of great importance. Currently, many methods basedon different measurement principles are available to give a measure of fish freshness[58]. The physical properties of the fish such as the rheological characteristics (firm-ness and texture) and the electrical properties (impedance) can sometimes be corre-lated with storage days. For instance, the impedance of fish is, for many species suchas cod and salmon, a good indicator of the time after catch. Nonetheless, this method isnot effective in case of frozen and thawed fish.The composition of the fish headspace is a source of information about the fresh-

ness degree of a sample. Spoilage in fish can be detected through the measure of theamount of amines, such as trimethylamine. Somemethods, based on analytical chem-istry procedures, are currently available to get information about the content of volatiletrimethylamine in the headspace. Nevertheless, the formation of amines due to de-composition starts some days after the catch. Chemical investigations using gas chro-matographic techniques have shown that there are five sources of odors, which whencombined, give rise to the overall odor of fish [59]. Fresh fish odor is a characteristicrelated to the individual species. Long-chain alcohol and carbonyls, bromophenols,and N-cyclic compounds are the basic contributors. Opposite to the fresh fish odoris the microbial spoilage odor – caused by compounds that are microbially formedduring the spoilage processes. These compounds are short-chain alcohol and carbo-nyls, amines, sulfur compounds, aromatics,N-cyclic compounds, and some acids. Theconcentration of these volatiles increases with time as the fish spoils; in fact, some ofthese are often used as indicators of spoilage [59]. Other sources of odors can be en-vironmental (such as petroleum odors), or due to the processing of fish, and fromproducts of lipid oxidation.


Due to the high number of volatile compounds involved in the process, and to thefact that they also dynamically change, themeasure of fish freshness over a long periodof storage can be achieved with amulticomponent approach. This is a typical electronicnose application where a number of non-selective and partially cross-correlated sen-sors are used to get a qualitative analysis of samples.Different electronic noses have been applied in the past to the detection of fish

freshness. Interesting results have been obtained with different sensor technologiessuch as metal-oxide semiconductor gas sensors [28, 29], electrochemical sensors [32]and TSMRs [31]. In the following, the application of the LibraNose to themeasurementof freshness, expressed as storage days, of a number of samples of cod fillets is de-scribed.It is useful to discuss some properties of metalloporphyrin-based sensors with re-

gard to the fish-freshness application. As stated earlier, some of the selectivity proper-ties of metalloporphyrins can be derived from the HSAB principle. In this context theuse of Mn(III) ion, a hard acid, is expected to provide greater sensitivity to oxygen-based ligands, while a metal ion like Co(II) is expected to give higher sensitivity to-wards amines or sulfur-containing metals. This scheme is simplified because it doesnot consider the role of the porphyrin ligand, but experiments have shown that it offera good explanation just for the selectivity towards amines, alcohols, and sulfur [47].Therefore, metalloporphyrins offer a way to design sensors optimized to catch fishodor at earlier and late stages of storage.As a reference method, trimethylamine (TMA) and total volatile bases nitrogen

(TVB-N) have been measured in the same samples. The data discussed here are re-lated to an experiment performed at the Icelandic Fisheries Laboratory in Reykjavikfrom 15–20 Nov 1999. Three batches of Atlantic Cod were collected for the experi-ment. Fish was caught with long line, gutted, and iced immediately after catch andbrought to the Icelandic Fisheries Laboratories the following day. Fish was kept at 0 8Cbefore being analyzed. Samples were filleted and de-skinned prior tomeasuring on thestorage days: 1, 2, 3, 4, 7, 9, 11, 15, and 17. Eight samples per storage day were mea-sured; a total of 72 fish. Themeasurements were performed on fillets. For each fish theright side fillet was measured, and the other side reserved for experiments not de-scribed here. Fillets were prepared about one hour before the analysis and wereheld constantly on an ice-bed until measured.

Fig. 21.4 The fish odor

sampler. The probe has a

diameter of 5 cm. Air refill is

provided by a series of small

holes immediately over the

fish surface, so that the odor

concentration in the supplied

air is very close to the equilibri-

um value. Measurement of a

salmon is shown on the right


Fish odormeasurements were done using a suitably designed fish odor sampler (seeFig. 21.4), which is a metallic capsule with an internal volume of 10 mL, approximatelyequal to the volume of the sensor chamber. The capsule is endowed with a series ofsmall orifices for air refilling. The sampler works in contact with the fish fillet, and astable and reproducible (from the point of view of sensor response) headspace is es-tablished in five minutes.During the experiment, the bone side of the right fillets was measured for each fish,

and each sample measured twice. The variation in resonant-frequency of QMB, con-sidered in steady state, was used as the sensor feature. Filtered ambient air was used asa reference. Ethanol, at its saturated pressure, was measured before and after eachmeasurement session, in order to control the stability of the sensors. The temperatureof the fillet surface, monitored during themeasurement, varied from 7 8C to 10 8C, andno correlation of sensor responses with the fillet temperature was observed. TMA andTVB-N, extracted from fish muscles, were measured using a conventional flow injec-tion analysis-gas diffusion method [60].Electronic nose data were analyzed by partial least square discriminant analysis

(PLS-DA). All calculations were carried out in Matlab 5.0. PLS-DA is a supervisedclassification method where the search for optimal discriminant directions is per-formed using PLS. Class membership is numerically represented with a so-calledone-of-many encoding. Namely, the y-block in PLS contains a number of variablesequal to the number of classes, and the membership of a single data point is expressedby putting the corresponding variable to one and all the others to zero. An unknownsample is then assigned to the class whose output is higher than the others. Thisprocedure is standard when quantitative oriented classifiers are used, such as neuralnetworks.PLS-DA provides both a quantitative estimation of class discrimination, and score

and loading plots for a visual inspection of data separation, and the contribution ofsingle sensors to the array. The meaning of these plots is different from those ob-tained by principal component analysis. In this case, the latent variables are deter-mined in a supervised procedure aimed at fitting the declared class membership,so that, even if the score plot of the first two latent variables may show class overlap-ping, the globality of all the latent variables can achieve a class separation. Nonetheless,these score plots, being linear projection over some basis, are indicative of the distri-bution of data in the sensor space. An evaluation of the classification properties can beobtained through a training and validation procedure using the one-leave-out valida-tion technique.Figure 21.5 shows the LibraNose data plotted on a basis identified by the first two

latent variables. Samples stored up to three days are clearly gathered in close clusters,the fourth day is overlapped with days 11 and 17, while the days 7,9, and 15 are alsooverlapping. This tendency to overlap the last days of storage with the first days,namely the inability to distinguish fish at two very different stages of storage, willbe shown to be consistent in this experiment. Here we have to keep in mind thatthere were three batches of fish and evidently there was a slight variation in the spoi-lage rate of the different batches. There may have been slight variations in handlingduring the first 24 hours after catch resulting in different spoilage rates of the batches.


The result of sensory analysis (data not shown here) confirms this effect and in fact thespoilage rate of the second batch appears to be slower than the first one. To clarify, days1, 2, 3 and 4 are from the first batch, days 7, 9 and 11 are from the second batch andfinally days 15 and 17 are from the oldest batch.Class identification is shown in Table 21.3 as a confusion matrix. The validation has

been performed on the whole data set because the one-leave-out validation techniquehas been used. Almost 90% of the samples were correctly identified. Nevertheless,errors, although numerically few, are qualitatively not negligible. Indeed, some sam-ples belonging to storage days from 7 to 15 are classified as belonging to the first day.An interpretation of the errors can be obtained by considering the values of TMA

and TVB-N. Figure 21.6 shows the measured values of these two important indicators.As reported in the literature, TMA values become considerably different from zeroonly after 9 days of storage, whereas TVB-N shows a non-linear and a non-monotonicbehavior with time. At the beginning of storage, TVB-N increases to reach amaximumafter approximately 4 days, then reaches the same levels as the very fresh fish after 7days, and then increases following the behavior of TMA. Figure 21.7 shows the plot ofTVB-N versus TMA, a log-log scale has been chosen in order to avoid the differentevolution of the two indicators. The plot shows basically the same distribution exhib-

Fig. 21.5 Plot of the first two latent variables of the PLS-DA for

the LibraNose data. Days 1-3 are separated while days 4-11-17

and 7-9-15 form grouped clusters


ited by the electronic nose systems, namely a straight evolution from days 1 to 4 and afolding back from 7 to 11 and then a net separation of the last days. It is worth men-tioning that the similarity of the log-log plot with the electronic nose score plot sug-gests that a logarithmic-like relationship between sensor response and volatile concen-tration should exists for the sensors considered here.

Tab. 21.3 Confusion matrix, estimated versus true, storage days

in fish freshness experiment

1 2 3 4 7 9 11 15 17

1 8

2 7 1

3 8

4 8

7 1 7

9 2 5 1

11 8

15 1 7 1

17 8

Fig. 21.6 Measure of TMA and TVB-N on the samples. TMA becomes

important after day 11, whereas TVB-N shows a non-monotonic

behavior during the first part of the evolution. In both the plots, the

inter-class dispersion grows with the number of storage days


The results of TMA and TVB-N show that the sensors are mostly correlated withthese two parameters, and most of all that the evolution of the chemical composition(qualitative and quantitative) does not provide a straightforward indication of the fresh-ness represented as storage days. This may be explained by slightly different spoilagerates of the three batches used, indicating that days of storage may not give the bestinformation about the freshness status of the fish when different batches of fish areconsidered.

21.5

Conclusions

The quality of foods and beverages is certainly among the most explored area of ap-plications of electronic noses. Nonetheless, the reported studies have been mostlyperformed at academic institutions. In many cases the results are certainly interestingfor the improvement of the field, but only rarely do they constitute a basis for immedi-ate industrial exploitation. The field still requires more basic research. Most of the

Fig. 21.7 The log-log plot of TVB-N versus TMA reveals a class

distribution very similar to that achieved by the electronic noses.

This result confirms that the class overlapping (a sort of folding back

effect) may be considered as intrinsic to the examined samples


research reports have concentrated on the improvement of sensors, while other im-portant areas, like the reliability of the sampling systems, have been neglected.However, a couple of conclusions can be made. The first is that the results achieved

so far are a sound basis for continuing towards reliable and industrially applicablequality measurement systems. To make rapid progress, the co-operation of electronicnose researchers and food scientists is necessary in order to customize a general-pur-pose technology like the electronic nose to the specific requirements of food and bev-erage industries. The second more general conclusion is that the electronic nose is notan analytical instrument, because it does not provide separation of volatile organiccomponents.The future is bright. For the first time, the principles of natural olfaction are being

exploited to obtain a chemical measurement. A cultural revolution is emerging thathas still to permeate the academic and industrial organizations, as well as thementalityof end users.

21.6

Future Outlook

All the participants in the food chain (producers, processors, and consumers) are po-tential users of electronic nose technology. Each step of the food chain has peculiarneeds that an electronic nose approach can satisfy in principle. As an example, atproducer level the increment of quality and yield, at processor level the screeningof quality of incoming products to optimize the processing and to sort processedfood, and finally at consumer level the control of quality and safety both on the marketand at home. All these applications require instruments that work on-site.Food-related sites are usually highly contaminated from the point of view of odor. At

the current state of the art, sensors are not able to distinguish between background andrelevant odor. From this perspective, portable systems without any conditioning of thesamples are of limited use in food analysis. For example, measuring the peculiar odorof a fish in a typical storage room among dozens of stacks of fish crates would bedifficult. However, there are certainly applications, interesting at industrial level,where existing electronic noses can be specialized, in terms of sampling and datapresentations, in order to fulfill user requirement. For this it is necessary to havestrong co-operation between electronic nose producers and end users in order to op-timize practical solutions. At this level it is important to have a correct and carefulanalysis of user needs and expectations, and an educational effort towards the usersin order to disseminate the intrinsic novelty carried by the artificial olfactionmachines.It is also important that developers and users are aware of the intrinsic limit of

information that is carried by the volatile part of a food. For instance, it is importantto consider that sensory analysis is almost never just confined to olfactory perception.Actually, synergetic action among the senses is required to form a full judgment over aparticular food sample. As an example, in fish analysis, a quality index, linearly cor-related with the days in ice, is calculated considering visual, tactile, and olfactory per-ceptions [60]. This suggests that, to fully reproduce the perceptions of humans with


artificial sensors, the electronic nose has to be compared and integrated with instru-ments providing information about visual aspects, texture, and firmness. This opens afurther novel investigation direction involving researchers from different areas, con-firming that the interdisciplinary nature is the most strong added value for food ana-lysis.

References

1 J. R. J. Pare, J. M. R. Belenger (Eds.). ‘In-strumental techniques in food analysis’, Else-vier, Amsterdam (The Netherlands), 1997.

2 F. Paoletti, E. Moneta, F. Sinesio. FoodScience and Technology, 1993, 26, 264–270.

3 N. Galili, I. Shmulevich, N. Benichou.Transactions of ASAE, 1998, 41, 399–407.

4 Y. Tao, C. T.Morrow, P. H.Heinemann, J H.Sommer. Transactions of ASAE 1990 90,3531–3554.

5 S. I. Cho, G. W. Krutz, H. G. Gibson, K.Haghighi. Transactions of ASAE, 1990, 3,1043–1050.

6 C. Curt. Science des aliments, 1997, 17, 435–456.

7 N. Sarkar, R. R. Wolfe. Transactions of ASAE,1983, 26, 624–629.

8 V. Bellon, J. L. Vigneau, M. Leclerq. AppliedSpectroscopy, 1992, 47, 1079–1083.

9 G. Jellinek. Sensory evaluation of food theoryand practice, Ellis Horwood Ltd. Publ. Chi-chester (UK), 1985.

10 T. Kawai. Critical Reviews in Food Science andNutrition, 1996 36, 257–298.

11 W. Pfannhauser. European Food Research andTechnology 1999 208, 336–341.

12 Y. Blixt, E. Borch. International Journal ofFood Microbiology 1999 46, 123–134.

13 I. E. Annor-Frempong, G. R. Nute, J. D.Wood, F. W. Whittington, A. West. MeatScience 1998 50, 139–151.

14 C. Di Natale, A. Macagnano, R. Paolesse, E.Tarizzo, A.Mantini, A. D’Amico. Sensors andActuators B, 2000 65, 216–219.

15 K. Neely, O. Prosser, P. F. Hamlyn. MeatScience, 2000 58, 53–58.

16 F. Maul, S. A. Sargent, D. J. Huber, M. O.Balaban, D. A. Luzuriaga, E. A. Baldwin.Proceedings of the Florida State HorticulturalSociety 1997 110, 188–194.

17 M. Benady, J. E.Simon, D. J. Charles, G. E.Miles. Transactions of the ASAE, 1995 38,251–257.

18 E. Hines, E. Llobet, J. W. Gardner. ElectronicLetters, 1999 35, 821–823.

19 E. Llobet, E. L Hines, J. W Gardner, S.Franco.Measurement Science and Technology,1999 6, 538–548.

20 J. E. Simon, A. Hertzoni, B. Bordelon, G. E.Miles, D. J. Charles. Journal of Food Science,1996 61, 967–969.

21 C. Di Natale, A. Macagnano, E. Martinelli, R.Paolesse, E. Proietti, A. D’Amico. Sensorsand Actuators B, 2001 78, 26–31.

22 S. Saevels, C. Di Natale, B. Nicolai. Procee-dings of the 6th International Symposium onFruit, Nut, and Vegetable Production Engin-eering, Potsdam (Germany) 11–14 Sept.2001.

23 European Communities Regulation 2598/91,Off;. J. Eur. Communities Legis. 248 (1991)1–33.

24 R Aparicio, S. M. Rocha, I. Delgadillo, M. T.Morales. Journal of Agricultural and FoodChemistry 2000 48, 853–860.

25 C. Ridgway, J. Chambers, E. Portero-Larra-gueta, O. Prosser. Journal of the Science ofFood and Agriculture 1999 79, 2067–2074.

26 A. Jonsson, F. Winquist, J. Schnuerer, H.Sundgren, I. Lundstrom. International Journalof Food Microbiology 1997 35, 187–193.

27 B. P. DeLacy Costello, P. J. Ewan, H. EGunsam, W. M. Ratcliffe, P. T. N. SpencerPhilips. Measurement Science Technology2000 11, 1685–1691.

28 R. Olafsson, E. Martinsdottir, G. Olafsdottir,T. I. Sigfusson, J. W. Gardner. in Sensors andsensory systems for an electronic nose, J. W.Gardner and P. Bartlett (eds.). Kluwer Aca-demic Publishers, Dordrecht (Netherlands),1992.

29 M. Schweizer-Berberich, S. Vahinger, W.Gopel. Sensors and Actuators B 1994 18, 282–290.

30 C. Di Natale, G. Olafsdottir, S. Einarsson, E.Martinelli, R. Paolesse, A. D’Amico. Sensorsand Actuators B 2001 77, 572–578.


31 C. Di Natale, J. A. J. Brunink, F. Bungaro, F.Davide, A. D’Amico, R. Paolesse, T. Boschi,M. Faccio, G. Ferri.Measurement Science andTechnology 1996 7, 1103–1114.

32 G. Olafsdottir, E. Martinsdottir, E. H.Jonsson. Journal of Agriculture and FoodChemistry 1997 45, 2654–2659.

33 C. Di Natale, A. Macagnano, F. Davide, A.D’Amico, R. Paolesse, T. Boschi, M. Faccio,G. Ferri. Sensors and Actuators B 1997 44,521–526.

34 F. Winquist, E. G. Hornsten, H. Sundgren,I. Lundstrom. Measurement Science andTechnology 1993 4, 1943–1950.

35 C. Di Natale, A. Macagnano, A. Mantini, E.Tarizzo, A. D’Amico, R. Paolesse, T. Boschi,F. Sinesio, F. M. Bucarelli, E. Moneta, G. B.Quaglia. Sensors and Actuators B 1998 50,246–252.

36 F. Winquist, C. Krantz-Rulcker, P. Wide, I.Lundstrom. Measurement Science and Tech-nology, 1998 9, 1937–1946.

37 A. Legin, A. Rudnitskaya, Yu. Vlasov, C. diNatale, E. Mazzone, A. D’Amico. Sensors andActuators B 2000 65, 232–234.

38 C. Di Natale, R. Paolesse, A. Macagnano, A.Mantini, A. D’Amico, M. Ubigli, A. Legin, L.Lvova, A. Rudnitskaya, Yu. Vlasov. Sensorsand Actuators B 2000 69.

39 F. Sinesio, C. Di Natale, G. Quaglia, F. Bu-carelli, E. Moneta, A. Macagnano, R. Pao-lesse, A. D’Amico. Journal of the Science ofFood and Agriculture, 2000 80, 63–61.

40 C. Di Natale, A. Macagnano, E. Martinelli, E.Proietti, R. Paolesse, L. Castellari, S. Cam-pani, A. D’Amico. Sensors and Actuators B2000 77, 561–566.

41 M. T. Morales, A. J. Berry, P. S. McIntyre, R.Aparicio. Journal of Chromatography A, 1998819, 267–275.

42 R.A.Alberty.PhysicalChemistry (sixth edition),J. Wiley and sons (New York, USA) 1983.

43 P. Mielle, F. Marquis. Sensors and ActuatorsB, 1999 58, 526–535.

44 A. D’Amico, C. Di Natale, A. Macagnano, F.Davide, A. Mantini, E. Tarizzo, R. Paolesse,T. Boschi. Biosensors and bioelectronics 199813, 711–721.

45 C. C. Lezenoff, A. B. P. Lever (eds.). Phta-locyanines: Properties and Applications, VCHPubl. (Weinheim, Germany);, 1989.

46 D. Dolphine (ed.). The Porphyrins, Vol. VIpart A and Vol. VII part B, Academic Press(New York, USA) 1978.

47 J. A. J. Brunink, C. Di Natale, F. Bungaro, F.A. M. Davide, A. D’Amico, R. Paolesse, T.Boschi, M. Faccio, G. Ferri. Analytica Chi-mica Acta 1996 325, 53–60.

48 C. Di Natale, A. Macagnano, G. Repole, G.Saggio, A. D’Amico, R. Paolesse, T. Boschi.Material Science and Engineering C, 1998 5,209–214.

49 K. Bodenhofer, A. Hierleman, M. Juza, V.Schurig, W. Gopel. Analytical Chemistry1997 69, 4017–4031.

50 G. Roberts. Langmui-Blodgett films , PlenumPress (New York, USA) 1990.

51 C. Di Natale, R. Paolesse, A.Macagnano, V.I.Troitsky, T. S. Berzina, A. D’Amico. Analy-tica Chimica Acta, 1999 384, 249–259.

52 C. D Bain, G. M. Whiteside. AngewandteChemistry International Edition, English 1989101, 522–525.

53 C. Di Natale, R. Paolesse, A. Mantini, A.Macagnano, T. Boschi, A. D’Amico. Sensorsand Actuators B 1998 48, 369–373.

54 C. Caliendo, P. Verardi, E. Verona, A. D’A-mico, C. Di Natale, G. Saggio, M. Serafini, R.Paolesse, S. E. Huq. Smart Materials andStructures 1997 6, 689–698.

55 C. Di Natale, D. Salimbeni, R. Paolesse, A.Macagnano, A. D’Amico. Sensors and Ac-tuators B 2000 65, 220–226.

56 D. S. Ballantine (ed.). Acoustic Wave Sensors:Theory, Design, and Physico-Chemical Appli-cation, Academic Press (New York, USA)1996.

57.G. Olafsdottir, E. Martinsdottir, J. Oehlen-schlager, P. Dalgaard, B. Jensen, I. Unde-land, I. Mackie, G. Henehan, J. Nielsen, H.Nilsen. Trends in Food Science Technology1997 8, 258–265.

58 D. B. Josephson, R. C. Lindsay and G.Olafsdottir. in D. F. Kramer, L. Liston. (eds);Seafood quality determination Symposium,Nov 10–14, 1986, Elsevier, Amsterdam,1986.

59 S. Sadok, R. Uglow, S. Haswell. AnalyticaChimica Acta 1996 334, 279–285. 60 J. B.Luten, E. Martinsdottir. in Methods to deter-mine the freshness of fish in research and in-dustries, Institut International du Froide,Paris 1997.

61 D. Demeyer, M. Raemaekers, A. Rizzo,A. Holck, A. Smedt, B. de Brink, B. tenHagen, C. Montel, E. Zanardi, E. Murbrekk,F. Leroy, F. Vandendriessche, K. Lorentsen,K. Venema, L. Sunesen, L. Stahnke, L. Vuyst,


R. de Talon, R. Chizzolini, S. Eerola. FoodResearch International 2000 33, 171–180.

62 T. Eklov, G. Johansson, F. Winquist, I.Lundstrom. Journal of the Science of Food andAgriculture 1998 76, 525–532.

63 S. Oshita, K. Shima, T. Haruta, Y. Seo, Y.Kawagoe, S. Nakayama, H. Takahara. Com-puters and Electronics in Agriculture 2000 26,209–216.

64 M. Hirschfelder, D. Ulrich, E. Hoberg, D.Hanrieder. Gartenbauwissenschaft (in eng-lish) 1998 63, 185–190.

65 Y. G. Martin, J. L. Perez-Pavon, B.M. Cor-dero, C. G. Pinto. Analytica Chimica Acta,1999 384, 83–94.

66 R. Stella, J. Barisci, G. Serra, G. G. Wallace,D. De Rossi. Sensors and Actuators B, 200063, 1–9.

67 T. Borjesson, T. Eklov, A. Jonsson, H.Sundgren, J. Schnurer. Cereal Chemistry1996 73, 457–461.

68 C. Di Natale, A. Macagnano, S. Nardis, R.Paolesse, C. Falconi, E. Proietti, P. Siciliano,R. Rella, A. Taurino, A. D’Amico. Sensorsand Actuators B 2001 78, 303–309.

69 P. Chatonnet, D. Dubordieu. Journal ofAgriculture and Food Chemistry 1999 47,4319–4322.

70 C. Di Natale, F. Davide, A. D’Amico, G.Sberveglieri, P. Nelli. Sensors and Actuators B1995 25, 801–804.

71 C. Di Natale, F. Davide, A. D’Amico, P. Nelli,S. Groppelli, G. Sberveglieri. Sensors andActuators B 1996 33, 83–88.

72 A. Legin, A. Rudnitskaya, Y. Vlasov, C. DiNatale, E. Mazzone, A. D’Amico. Electroa-nalysis 1999 11, 1–7.

73 E. Anklam, M. Lipp, B. Radovic, E. Chiavaro,G. Palla. Food Chemistry 1998 61, 243–248.

74 E. Schaller, J. O. Bosset, F. Escher. Chimia1999 53, 98–102.

75 R. T. Marsili. Journal of Agricultural and FoodChemistry 1999 47, 648–654.

76 D. D. Muir, E. A. Hunter, J. M. Banks.Milchwissenschaft (in English) 1997 52, 85–88.

77 C. Di Natale, R. Paolesse, A. Macagnano, A.Mantini, A. D’Amico, A. Legin, L. Lvova, A.Rudnitskaya, Y. Vlasov. Sensors and ActuatorsB 2000 64, 15–21.

78 C. Gretsch, A. Toury, R. Estebaranz, R.Liardon. Seminars in Food Analysis 1998 3,37–42.

79 J. W. Gardner, H. V. Shurmer, T. T. Tan.Sensors and Actuators B, 1992 6, 71–75.

80 T. Fukunaga, K. Toko, S. Mori, Y. Naka-bayashi, M. Kanda. Sensors and Materials,1996 8, 47–56.

81 A. Legin, A. Rudnitskaya, Y. Vlasov, C. DiNatale, F. Davide, A. D’Amico. Sensors andActuators B 1997 44, 291–296.

82 J. B. Tomlinson. Ferment 1996 9, 85–89.83 T.C. Pearce, J.W. Gardner, S. Friel, P.N.

Bartlett, N. Blair. Analyst, 1993 118, 371–377.

84 Y. Arikawa, K. Toko, H. Ikezaki, Y. Shinha,T. Ito, I. Oguri. Journal of FermentationBioengineering 1996 82, 371–376.


22

Automotive and Aerospace Applications

M. A. Ryan, H. Zhou

22.1

Introduction

The trainability of an electronic nose, along with the ability to select sensors for re-sponse to a suite of compounds, has made this type of device useful in several applica-tions for monitoring air quality in an environment where the possible contaminantsare known. In this chapter we will discuss its application to monitoring the presence ofhazardous compounds for breathing air in an enclosed space. The application of anelectronic nose as an air quality monitor is as an event monitor, where events of lowconcentration that do not present a hazard are not reported, but events of concentra-tions approaching a hazardous level are reported so remedial action can be taken. Theelectronic nose used in these applications is not an analytical device that analyzes theair for all compounds present, but neither is it an alarm that sounds at the presence ofany change in the atmosphere. The device described here was used as an air-qualitymonitor in an experiment aboard NASA’s space shuttle Flight STS-95, and was de-signed to fill the gap between an alarm with no ability to distinguish between com-pounds and an analytical instrument.

22.2

Automotive Applications

Use of an electronic nose in the automotive industry is primarily conceptual today, butthere are several areas in which such a device can be used. These include monitoringthe exhaust for combustion efficiency, monitoring the cabin air for passenger safety,and monitoring the engine compartment for other conditions such as leaking oil orother fluids. Owing to offgassing of fabrics and materials (‘new car smell’), to leaks ofcoolant from the air-conditioning system, and intake of air from the roadway and theengine compartment, the passenger cabin of an automobile can be significantly morehazardous to human health than the outside air [1, 2]. Improvement of the air quality


525525

in an automobile cabin can be accomplished rather simply, but as cabins will remainwell sealed for climate control and energy conservation, a need to monitor the interiorwill remain. As environmental concerns spur development of more efficient combus-tion, it will be useful to monitor the exhaust for combustion products as well. Severalautomobilemanufacturers have discussed the possibility of using an electronic nose ina system in which the exhaust is monitored for the presence of compounds indicativeof incomplete combustion, and feedback to the engine will adjust engine settings toimprove combustion efficiency.

22.3

Aerospace Applications

Electronic noses have been proposed for many applications in aerospace; some ofthose applications are realistic within the limits of today’s technology, and somewill require more development. In the area of space exploration, electronic noseshave been proposed for planetary atmospheric studies on landers. This applicationvaries from addition of an electronic nose to a rover to study the atmosphere asthe rover moves, to stationary devices, which will study the variations in atmosphereover days or seasons. In the search for evidence of life on other planets, electronicnoses have been proposed as desirable sensors because the sensing media in the arraycan be selected to make it possible to distinguish between isomers and enantiomers[3], and because the sensor array can be configured to span a broad range of com-pounds. These applications require development of methods that will allow the elec-tronic nose to deconvolute target vapors from an unknown background; work to devel-op devices with these capabilities is underway at the Jet Propulsion Laboratory (JPL).An immediate, and perhaps the most important, application is monitoring air qual-

ity in human habitats. The ability to monitor the recycled breathing air in a closedchamber is important to NASA for use in enclosed environments such as the crewquarters in the space shuttle and the International Space Station (ISS). Today, airquality in the space shuttle is generally determined anecdotally by crewmembers’ re-ports, and is determined after flight by collecting an end-of-mission sample and ana-lyzing it in an analytical laboratory using gas chromatography-mass spectrometry (GC-MS). The availability of a miniature, low-power instrument capable of identifying con-taminants in the breathing environment at part-per-million (ppm) and sub-ppm levelswould enhance the capability to monitor the quality of recycled air and thus to protectcrew health. Such an instrument is envisioned for use as an incident monitor, to notifythe crew of the presence of potentially dangerous substances from spills and leaks, andto provide early warning of heating in electrical components that could lead to a fire. Inaddition to notification of events, it is necessary to have a reliable method by whichjudgments on the use of breathing apparatus can be made; if the crew has put onbreathing apparatus while repairing a leak or cleaning a spill, it is necessary toknow whether it is safe to remove the apparatus. These needs have led to the devel-opment of an electronic nose at JPL [4–6], with ultimate application to ISS intendedand experiments on the space shuttle in the near-term.

22 Automotive and Aerospace Applications526

The qualities required for an incident monitor to be used in spacecraft are that itshould be capable of identifying and quantifying target compounds at determinedlevels in a fairly wide range (see Table 22.1), that it be a low mass and volume devicewhich uses low power, and that it require little crew time for maintenance, calibration,and air analysis. There are several possible sensing devices that could be used in thespace shuttle or ISS, but all have limitations in terms of their requirements. Thesedevices include GC-MS, volatile organic carbon analyzers, flame ionization detec-tors, and smoke alarms. Of these, only GC-MS discriminates among compounds;it also has the greatest sensitivity. However, it generally requires crew time in samplepreparation, maintenance and calibration. An electronic nose does not, in general,have the sensitivity of GC-MS, however, for most target compounds ppm and sub-ppm sensitivity is required, but not the parts per trillion level found with GC-MS.An electronic nose meets the requirements for an incident monitor. It can identify

and quantify compounds in its target set with a dynamic range of about 0.01 to10 000 ppm, depending on the compound, it lends itself to miniaturization, and be-cause it measures deviation from a background it does not require frequent calibrationand maintenance.The electronic nose developed at JPL was designed to detect a suite of compounds

and is suitable for use in the crew habitat of a spacecraft. The habitat is an enclosedspace where air is recycled and where it is unlikely that unknown and unexpectedvapors will be released into the air. It can be assumed that the air is clean at the begin-ning of a period of enclosure, and it is deviations from that state that the electronicnose will monitor, thus it is not necessary to have detailed knowledge of the consti-tuents of the air initially. In addition, the contaminants which are likely to be present,and for which it is important to monitor, are well known, the number of compounds isnot large (50 or so), and the probability of mixtures of 5 or more such compoundsappearing at one time is small. It is possible, then, to design and train a device tomonitor the air for deviation from a clean baseline and to analyze those deviationsfor the appearance of a set of target compounds.The air quality conditions in the crew quarters of a spacecraft are not radically dif-

ferent from the conditions in an aircraft cabin, or in the passenger cabin of a bus orautomobile. In all those cases, it is reasonable to assume the air is clean at the begin-ning of a monitoring period, and there is a set of contaminants of concern to be mon-itored. With such conditions in mind, the JPL electronic nose was designed for a flightexperiment where the crew habitat in the space shuttle was monitored continuouslyfor six days.The JPL electronic nose is a low power, miniature device which, in its current ex-

perimental design, has the capability to distinguish among, identify and quantify 10common contaminants which may be present as a spill or leak in the recirculatedbreathing air of the space shuttle or space station. It has as its basis an array of con-ductometric chemical sensors made from polymer/carbon composite sensing filmsdeveloped at Caltech [7, 8]. It is an array of 32 sensors, coated with 16 polymers/carboncomposites. The polymers were selected by analyzing polymer responses to the targetcompounds and selecting those that gave the most distinct fingerprints for the targetanalytes. The JPL development model was used in a flight experiment on the space

22.3 Aerospace Applications 527527

shuttle flight STS-95 (October-November 1998) to determine whether it could be usedas a continuous air quality monitor. A block diagram and photo of the JPL electronicnose are shown in Fig. 22.1. The device used in the flight experiment has a volume of2000 mL and a mass of 1.4 kg including the HP200 LX computer used for control anddata acquisition, and uses 1.5 W average power. The mass and volume were deter-

Fig. 22.1 The JPL electronic nose used in the flight experiment

on STS-95 is shown as a block diagram and as a photo. The

developmental device occupies a volume of 2000 mL and has a mass

of 1.4 kg, including the HP 200 LX computer


mined primarily by the spaceflight-qualified container required for the device to beused in an experiment; the volume and mass can be reduced by a factor of 4 withno modifications to the sensor head or the electronics and minor modifications tothe pneumatic system.

22.4

Polymer Composite Films

The polymer/carbon composite films developed at Caltech are the sensing media usedin the JPL electronic nose [7–10]. These films are made from insulating polymersloaded with a conductive medium such as carbon to make resistive films. When apolymer film is exposed to a vapor, some of the vapor partitions into the film andcauses the film to swell; the degree of swelling is proportional to the change in resis-tance in the film because the swelling decreases the number of connected pathways ofthe conducting component of the composite material [7]. The electrical resistance ofeach sensor is thenmeasured and the response of each sensor in the array is expressedas the change in resistance, dR.Using commercially available organic insulating polymers as the basis for conducto-

metric sensing films allows ready incorporation of broad chemical diversity into thesensing array. The sensors respond differently to different vapors, based on the dif-ferences in such properties as polarizability, dipolarity, basicity or acidity, and mole-cular size of the polymer and the vapor.The polymer/carbon composite sensing films are sensitive to temperature and pres-

sure change as well as to changes in the composition of the atmosphere. In a measur-ing mode where the device is sniffing the atmosphere and comparing it to a cleanbackground with measurements of each a few minutes apart, temperature changesare generally not significant.However, in the case of continuous monitoring over several hours or days, both

temperature and pressure changes will influence the location of the baseline, andit is necessary to distinguish among temperature and/or pressure change, slow build-up of compounds, and baseline drift. All of these issues were addressed in the devicedeveloped at JPL. Neither changes in pressure nor humidity which might be found innormal habitat have a significant effect on the differential sensor response, but tem-perature changes greater than 4–8 8C influence the magnitude of response across thesensing array as well as the fingerprint of individual analytes. While it is possible tomeasure temperature, pressure, and humidity and to subtract any effect of changes inthese conditions from the sensor response data, the JPL electronic nose was built withthe capability to control temperature, and pressure and humidity were measured se-parately. Temperature was controlled on the sensor substrates to stay constant at 28,32, or 36 8C, both to eliminate apparent baseline drift (film resistance changes) causedby temperature change, and to aid the sensing process. Temperatures around 30 8Cwill assist the process of desorption of analytes from the films and will prevent hydro-gen bonds from forming between analytes and the polymers.

22.4 Polymer Composite Films 529529

22.5

Electronic Nose Operation in Spacecraft

While it is reasonable to assume clean air at the beginning of an enclosed period in thespace shuttle, there are two scenarios in which a clean air baselinemust be established.In one scenario, the electronic nose might be used to determine whether it is safe toenter a chamber that has been enclosed for some time without crew use, such as amodule in ISS. In the other scenario, a background of clean air must be established todetermine whether there has been a slow buildup of a contaminant. This second sce-nario is among the most likely for contamination of the air. Contaminants may buildup slowly as offgassing, slow leaks in vapor and liquid containers, from inadequate airrevitalization or filter breakthrough, and as human metabolic products such asmethane or carbon dioxide. In both of these scenarios, a system by which a baselineof clean air can be established is necessary.Contamination from offgassing may be considered of minor importance for aircraft

or automobile cabins because the air is exchanged frequently in the course of use andfresh air can be brought inside during use, but in cabins where air is not exchanged forseveral hours, the buildup can be considerable. Often the offgassed molecules aresmall, such as formaldehyde, and are not well scrubbed in the air revitalization sys-tem. In the space shuttle where air might not be exchanged for several days or, moreimportantly in ISS, where the air is not exchanged, offgassing becomes an importantconsideration. Flight qualification includes establishment that the offgassing rate ofcomponents be below a set level, but there are as yet no data for offgassing over periodsof months to years, as will be found on ISS.The JPL electronic nose pneumatic system includes a diaphragm pump, which pulls

atmosphere at 0.25 L/min over the sensors and two filters, an activated charcoal filterand a filter of inert material, before the sample chamber. The atmosphere to be ana-lyzed travels through a filter that is selected by a solenoid valve, which switches be-tween the two. During usual monitoring intervals, the air travels through the ‘dummy’filter made of inert material to provide a pressure drop equivalent to the pressure dropacross the charcoal filter. The charcoal filter cleans air without removing humidity, anda baseline of cleaned air can be constructed and used to determine the degree of base-line drift. The constructed baseline allows the analysis program to distinguish betweendrift and slow change in atmosphere. Figure 22.2 shows how drift and slow buildupcan be distinguished after the charcoal filter is switched off; the sensor films respondby rising rapidly and creating a ‘virtual peak,’ and the sensor responses can then beanalyzed against the cleaned air background. The analysis of the responses of thesensing array can then be used to determine whether the slow change in the atmo-sphere is caused by contamination.For the flight experiment, 6 days of continuous operation, the charcoal filter was

switched on for 20 minutes out of every 210 minutes. This frequency was sufficientto determine the baseline in this application. If an electronic nose is to be used todetermine whether a chamber is safe to enter after a closed period, the cleaned airbaseline must be established for several minutes, and the virtual peak analyzedwhen the charcoal filter is turned off. A schedule for filter changeout must be estab-


lished; for space shuttle air and no unusual events, changing the filter every 2–3months is sufficient. If there has been an incident found by the filter, it should bechanged after the cause of the incident has been fixed.In other applications, where the pressure and temperature are changing rapidly, or

where the composition of the atmosphere changes frequently, the filters can be pro-grammed to switch at different frequencies. In the passenger cabin of an aircraft, forexample, filtering can be frequent during the loading and taxi stages, when the con-centration of combustion products and of fuel can be high, and less frequent duringcruise.The responses of the electronic nose were not influenced significantly by meals or

activities in the crew quarters because the device was placed under the air intake ventfor the entire cabin; odors were significantly diluted when they reached the sensors.This condition was chosen in order to monitor the average concentration in the cabinrather than localized concentrations.

Fig. 22.2 a) A virtual peak is

created at time 21:08 when the

airflow is switched from the

charcoal filter, which determines

the clean air baseline, to the inert

filter that is used during normal

measurements. The baseline

drift can be determined by fitting

the trend of the clean air base-

line; in this case the virtual peak

can be attributed to baseline

drift. b) A virtual peak, which is

not attributable to baseline drift,

can be analyzed for the presence

of hazardous materials


22.5.1

The JPL Enose Flight Experiment

For the application of adverse event monitoring in the space shuttle, the JPL electronicnose was trained to respond to 12 compounds; 10 of these were compounds likely toleak or spill and the other two were humidity change and vapor from amedical swab (2-propanol and water), which was used daily to confirm that the device was operating.The electronic nose was trained to identify and quantify the 10 contaminant com-pounds at the 1-hour spacecraft maximum allowable concentration (SMAC) levelsthat are shown in the upper section of Table 22.1.The 10 contaminants were drawn from a list of compounds of concern and for which

air samples are tested after a shuttle flight. In the second-generation device, now underdevelopment, there will be 10–12 additional compounds. The sensitivity required forthe device was set at the 1 hour SMAC in the flight experiment, and is set at the24 hour SMAC for the second-generation device. The upper section of Table 22.1shows the 24-hour SMAC and the lowest level detected reliably by the first generation

Tab. 22.1 Upper Section: Compounds targeted in the first-generation

electronic nose, with their 1-hour and 24-hour SMACs, and the lower

level detected at JPL with that device. Lower Section: compounds

considered for the second-generation electronic nose, with their

24-hour SMACs

Compound SMAC 1 hr (ppm) [**] SMAC 24 hr (ppm) [**] Detected at JPL (ppm)

Methanol 30 10 5

Ethanol 2000 500 50

2-Propanol 400 100 50

Methane 5300 5300 3000

Ammonia 30 20 20

Benzene 10 3 10

Formaldehyde 0.4 0.1 10

Freon 113 50 50 20

Indole 1 0.3 0.03

Toluene 16 16 15

Acetaldehyde 6

Acetone 270

Acetonitrile 4

2-Butanone 150

Chlorobenzene 10

Dichloromethane 35

Furan 0.1

Hexamethyltricyclosilane 25

Hydrazine 0.3

Methyl hydrazine 0.002

Tetrahydrofuran 40

1,1,1-Trichloroethane 11

o,p-Xylenes 100

* Source [11]


electronic nose at JPL, where lower levels were determined by SMACs and are notnecessarily detection limts. The lower section of Table 22.1 shows a list of compoundsconsidered for the second set and their 24-hour SMACs; sensor response data on thesecompounds are not yet available. As an event monitor, it is not necessary to be sig-nificantly more sensitive than the 24 hour SMAC level; when the concentration of acontaminant approaches � 35% of the SMAC, measures can be taken to remove thecompound from the air and to take action on the source of the contamination. Furthertraining of the software is possible in situ, but for accurate identification and quanti-fication, the training must be done in an environment where it is possible to deliverprecise concentrations of the compound in the range of interest.For all cases except formaldehyde, the electronic nose is able to detect the compound

at or below the 1 hour SMAC. The sensitivity limit for formaldehyde in the flightexperiment device is 10 ppm; by selection of a different polymer set with polymersmore likely to sorb formaldehyde, it should be possible to detect that compound belowthe 24-hour SMAC level. The electronic nose is also able to deconvolute signals toidentify and quantify mixtures of two compounds with moderate success (about60%). It is expected that with further training and a more selective group of poly-mers, it will be possible to detect lower concentrations of compounds and to decon-volute mixtures of three or four compounds.

22.5.2

Data Analysis

The data analysis software development portion of the JPL electronic nose flight ex-periment considered several different approaches. The primary constraint in softwaredevelopment was the requirement that gas events of single or mixed gases from the 10target compounds be identified correctly and quantified accurately. The co-investigatorin the flight experiment, Dr. John James of the Toxicology Branch at NASA-JohnsonSpace Center (JSC), defined accurate quantification as þ/� 50% of the known con-centration measured in the laboratory. This degree of error was defined based on theSMACs; the toxic level of most of the compounds is not known more accurately thanþ/� 50%, so the SMACs have been set at the lower end. For the flight experiment,constraints in telemetry and communication prevented real-time analysis, and so thedevelopment process did not include full capability for immediate resistance vs. timedata analysis.A series of software routines was developed usingMATLAB (fromMathWorks, Inc.)

as a programming tool. MATLAB is a flexible program, and thus appealing for devel-opment of software, though it runs relatively slowly. For future use, where real-time orquasi-real time analysis is called for, the routines can be translated into C and run on adesktop or laptop computer.For sensing media such as the conducting polymer/carbon films used in this pro-

gram, relative response changes (in magnitude) have been found to be more reliablethan the response shapes, especially at the low gas concentration range targeted in thisprogram (1–100 ppm). Hence, the task of identifying and quantifying a gas event isroughly a two-step procedure:


1) Data pre-processing, to extract the response pattern of a gas event from raw time-series resistance data for subsequent analysis.

2) Pattern recognition, to identify and quantify a gas event based on the responsepattern extracted.

22.5.2.1 Data Pre-Processing

When presented with continuous monitoring data, a response pattern must be ex-tracted by use of software. This process of extracting a response pattern from rawresistance data involves four sequential steps: 1) noise removal, 2) baseline drift ac-commodation, 3) gas event occurrence determination, and 4) resistance change cal-culation.

Noise removal

Despite the best effort in choosing sensor films with the consideration of low noiselevel, the noise level can be quite large. Some polymer films were found to be noisierthan others. The reasons one polymer/carbon composite film might be noisier thananother are not well understood; noise may be attributed to high sensitivity of thepolymer film to small changes in pressure caused by air flow, to differences in thecarbon dispersion in the film, or to inhomogeneities in the thickness or even composi-tion of the film itself. In general, the fluctuation in resistance (or noise) is fast com-pared to the response to a gas event. Therefore digital filtering may be used to filter outthis high frequency fluctuation. The length of the filter may be different for differentsensors and can be determined by analysis of the noise in each sensor.

Baseline drift accommodation

Baseline drift is one of the most difficult problems to be solved in extracting electronicnose resistance data from the time data. The causes for baseline drift can be multiple,and include variations in temperature, humidity, pressure, aging of the sensors, andsensor saturation. However, at present there is no clear understanding of the under-lying mechanism of each one of the causes, whichmakes attempts to compensate driftvery difficult. Nevertheless, the baseline drift is generally slowly varying in naturecompared to the response time of a detectable gas event. This difference in time scaleenables us to use a long-length digital filter to determine the approximate baseline driftand then subtract it from the raw data. The result is further adjusted by piecewisefitting using the baseline information from the clean air reference cycles describedabove. Although this approach will not accommodate the drift fully, it will reducethe effect to a manageable degree. Figure 22.3 shows resistance data that has beenprocessed. The dark, smooth trace in the upper plot shows the baseline variation de-termined through the use of low frequency filters. The gray, noisy trace in the lowerplot is the data after baseline variation has been subtracted, and the dark line is theprocessed data, with baseline variation subtracted and after filtering for noise accom-modation.


Gas event occurrence determination

Because data analysis in the flight experiment of the JPL electronic nose was not real-time owing to constraints unrelated to the technology development, it was not neces-sary for the analysis to be automatic, but a preliminary software routine for automateddetermination of whether and when a gas event occurs was developed. It is basedprimarily on threshold calculation, in which the resistance change over a certaintime interval is calculated, and a time-stamp is registered if the change exceeds apre-set threshold. This routine can detect most gas events; however, it was also foundthat it might identify noise, and sometimes baseline drift, as gas events. For the flightexperiment, events identified by the automated routine were confirmed by visual in-spection of the time domain data; future development of the data analysis software willrefine the identification method.

Resistance change calculation

Since the sensors’ relative responsiveness to a vapor determines the fingerprint of thatgas – the response pattern – it is important to preserve this relative responsiveness.

Fig. 22.3 a) Grey, noisy trace: raw resistance as recorded; dark line:

baseline drift determined by low frequency digital filtering. b) Grey trace:

resistance after baseline drift subtracted; dark line: processed data, resistance

after noise accommodation by smoothing and high frequency filtering,

and baseline drift corrected


This means any calculation method of the resistance change should be taken at thesame time-stamp after the initial onset of a gas. Both the relative resistance change,R=R0, and the fractional resistance change, (R� R0Þ=R0 were tested, and the latter wasadopted as it maximizes the difference between the signatures of different gas com-pounds.

22.5.3

Pattern Recognition Method

Although many pattern analysis methods exist in the general field of electronic noseand other array-based sensor data analysis [12; also see Chapter 6 of Part A, and Part C],no single method appears to be readily applicable to the task of identifying and quan-tifying single gases as well as mixtures of up to three of the 12 compounds (10 targetcompounds plus water, humidity change and the propanol wipe) at levels about 1–100 ppm. Most of the widely used methods have demonstrated their effectiveness, butnot to a combination of all three scenarios found here: a large number of target com-pounds, some of which are of very similar chemical structure (e.g., ethanol andmetha-nol), low target concentrations, and both single gases and mixtures.

22.6

Method Development

For reasons stated above, three parallel approaches to electronic nose data analysiswere used during the early stages of software development: discriminant functionanalysis (DFA), neural networks with back propagation (NNBP), and linear algebra(LA). Principal component analysis (PCA) was initially used, but was later replacedby DFA because DFA tends to do better at discriminating similar signatures that con-tain subtle, but possibly crucial, gas-discriminatory information. DFA is also better inclass labeling than PCA.NNBP, or more specifically, multilayer perceptron (MLP), was selected as an ap-

proach because it has good generalization of functions to cases outside the trainingset, is capable of finding a best-fit function (linear or nonlinear; no models needed),and is also more suitable than DFA when the sensor signatures of two gases are notseparable by a hyperplane (e.g., one gas has a signature surrounding the signatures ofanother gas). However, NNBP is inferior to DFA in classifying data sets that mayoverlap.The reason to use LA, which is not as commonly used as other methods, is that

neither DFA nor NNBP were found to be well suited to recognizing the sensor sig-natures from combinations of more than one gas. This method tries to solve the equa-tion x=Ac, where vector x is an observation (a response pattern), vector c is the cause forthe observation (concentrations of a gas or combinations of gases), and matrix Adescribes system characteristics (gas signatures obtained from training data, or sen-sitivity coefficients). For electronic nose data analysis where the response pattern can


be noise corrupted, whichmeans theremay exist no exact solution, least squares fittingis the preferred way to solve the equation [13, 14].The idea of developing three parallel methods is that one can first use the LAmethod

to deconvolute an unknown response pattern as a linear combination of target com-pounds; unknown compounds are expressed as a combination of up to four com-pounds. If a single compound is found, additional verification can then be carriedout by NNBP and DFA methods for increased success rate and accuracy. Howeverwe have found the LA method to perform consistently best among the three methodseven for single gases, while DFA was consistently the worst, which prompted us todiscard the use of the two verification methods of NNBP and DFA in the process.LA is suitable only if the training data are linear, which is not the case for all sensors

at the concentration ranges considered (see Table 22.2). For a nonlinear scenario, it isthen reasonable to use some nonlinear least squares fitting methods such as that ofLevenberg and Marquart (LM-NLS). This is the one of the two new methods that wereinvestigated for nonlinear analysis. The other method, a differential evolution (DE)approach, was also investigated because it promises fast optimization (the LM-NLSmethod can be rather slow). DE represents some recently emerged so-called geneticalgorithms [15]. It is a parallel direct search optimization tool, and begins with aninitial randomly chosen population of parameter vectors, adding random vector dif-ferentials to the best-so-far solution in order to perturb it. A one-way crossover opera-tion then replaces parameters in the targeted population vector with some (or all) of theparameter values from this ‘noisy’ best-so-far vector. In essence it imitates the prin-ciples of genetics and natural evolution by operating on a population of possible solu-tions using so-called genetic operators, recombination, inversion, mutation and selec-tion. Various paths to the optimum solution are checked and information about themcan be exchanged. The concept is simple, the convergence is fast and the requiredhuman interface is minimal: no more than three factors need be selected for a specificapplication. However the last advantage is also its disadvantage: limited control forelectronic nose data analysis. Finally, the LM-NLS method was selected as the besttool for electronic nose data analysis.

22.6.1

Levenberg-Marquart Nonlinear Least Squares Method

For nonlinear models the technique of choice for least-squares fitting is the iterativedamped least-square method LM-NLS. Similar to LA, LM-NLS tries to find the best-fitparameter vector c from an observation vector x, which is related to c through a knownlinear or nonlinear function, x ¼ f ðA; cÞ, where A covers system characteristics (sen-sitivity coefficients) obtained from training data. This method usually begins from agiven starting point of c, and calculates the discrepancy of the fit:

residual ¼ ðcomputed � observedÞ=r;

where r is the standard deviation, and updates with a better-fitted parameter c at eachstep. LM-NLS automatically adjusts the parameter step to assure a reduction in the


residual: increase damping (reduce step) for a highly nonlinear problem; decreasedamping (increase step) for a linear problem. Because of this ability to adjust dam-ping, LM-NLS is adaptive to both linear and nonlinear problems. How this methodadjusts damping is discussed in detail elsewhere [16].In the course of this work, it was found that the response of the films to the target

compounds is linear with concentration only within a limited range. The nonlineari-ties in the training data generated are of low order, but successful identification andquantification of gas eventsmust take the nonlinearities into account. To obtain sensorcharacteristics without further knowledge of sensor nonlinearities, a second-orderpolynomial fit was used to model the nonlinearities. For each sensor response toeach gas, the program finds the best-fit sensitivity coefficients A1 and A2 (in theleast-squares sense) to the following equation:

resistance change ¼ A1c þ A2c2

where c is gas concentration vector. The fit is constrained to pass through the origin.A1 and A2 are 13� 32 matrices characterizing the sensors’ response to ten targetedgases plus water, humidity change, and the propanol wipe.Several modifications were made to the standard LM-NLS method to suit the elec-

tronic nose data analysis problem. First, sets of starting points of vector c were usedinstead of a single set of starting points of vector c. The purpose of doing this is to avoida local residual minimum, which is common in many optimization algorithms, in-cluding the LM-NLS method. These initial sets of vector c can be randomly assignedfrom within each element’s allowed range. The total number of initial sets will bedetermined by the speed desired and the complexity of local minimum problem.In our case, about 200 initial sets were found (� 15 N, where N ¼ 13 is the numberof target compounds) to be a good compromise.Second, instead of always updating c for a smaller residual, we modify the update

strategy to favor a smaller number of gases within certain ambiguity ranges of theresidual. The reason is that signature patterns for a given gas compound generatedby the electronic nose sensors have been observed to have large variations. The simpleupdating strategy tends to minimize the residual with a more than reasonable largenumber combination of gas when the residual is simply the variation in recordedresponse pattern itself and should be ignored. The amount of the final residual isan indicator of how large the fitting error is and the confidence level of the fitting.Finally, the sensors’ response pattern was weighted to maximize the difference be-

tween similar signatures. As seen in Fig. 22.4, which shows representative signaturesof the ten target gas compounds plus the medical wipe at a median concentration level(because of the nonlinearity, there is no single signature for one gas at all concentra-tions), it is clear that ethanol and methanol have very similar signature patterns. Re-gression analysis also pointed out linear dependency to certain degrees. This meansthat the signature pattern of one gas could be expressed as a linear combination of theresponse pattern generated by some other target gases. To reduce this similarity, thesensors’ raw resistance responses must be modified by different weights in the dataanalysis procedure.


22.6.2

Single gases

For lab-controlled gas events, the overall success rate reaches � 85% for targeted sin-gles where success is correct identification and quantification L. Broken down intoindividual singles, the successes are listed below in Table 22.2. The concentrationranges used in the training sets for each single gas are also given.

Fig. 22.4 Representative signatures of ten targeted gas compounds

plus wipe generated by electronic nose sensors. Notice the

similarity between ethanol and methanol, and the significant

difference between benzene and toluene

Tab. 22.2 Identification and quantification success rates for single

gases. The ranges shown here are ranges used in LM-NLS analysis

Compound Concentration Range (ppm) Success Rate (%)

Ammonia 10–50 100

Benzene 20–150 88

Ethanol 10–130 87

Freon 113 50–525 80

Formaldehyde 50–510 100

Indole 0.006–0.06 80

Methane 3000–7000 75

Methanol 10–300 65

Propanol 75–180 80

Toluene 30–60 50

%Relative Humidity 5–65 100

Medical Wipe 500–4000 100


Considering that the raw data are often very noisy at low concentrations, nonlinear athigh concentrations, highly correlated in some cases, and weakly additive in somemixtures, these results demonstrate that the LM-NLS method is an effective techniquefor analysis of an array of sensors. Future work on the electronic nose will attempt toremovemany of the impediments to data analysis, with focus on noise and correlation.Correlation will be addressed in polymer film selection.The ability of the data analysis software to identify and quantify single and multiple

gas events in clean air was tested in the laboratory. The targeted concentration rangefor quantification was 30% to 300% of the one hour SMAC for each compound. Ascan be seen from Table 22.2, in some cases it was possible to identify and quantify

Fig. 22.5 Identification and quantification of four single gases using LM-NLS.

The shaded area is the target þ/� 50% detection range

Fig. 22.6 Identification and quantification of three single gases using LM-NLS


substantially below the 30% SMAC concentration; however, in a few cases quantifica-tion was successful only as low as 100% of the 1-hour SMAC. In one case, formalde-hyde, we were unable to identify and quantify reliably below several times the 1-hourSMAC. Figures 22.5 and 22.6 show some results of single gas identification and quan-tification graphically.

22.6.3

Mixed Gases

Deconvolution for identification and quantification of mixtures relies on the additivityof the sensor responses. Here, additivity means that the strength of the response to amixture of gas 1 at level c1 and gas 2 at level c2 equals the response of the single gas 1 atlevel 2 plus the response of the single gas 2 at level c2.Identification and quantification of mixtures in clean air was moderately successful.

Additive linearity holds for some combinations in concentration ranges near theSMAC level of the lower SMAC compound. The success rate for double gases (about60%) was less than that of single gases, as would be expected. An exhaustive set of gaspairs was not run because of time constraints; only a selected group of mixture pairswere run to test the additivity. For this relatively small pool of data, additivity holds forthe following gas combinations:

methanol þ toluene ammonia þ benzene ethanol þ formaldehydemethanol þ benzene ammonia þ ethanol propanol þ benzene.

Although data obtained on some other combinations of gas compounds, e.g., {ben-zene þ formaldehyde} and {methanol þ propanol}, did not validate their additivity inthese tests, this does not necessary mean the additivity does not hold for those gascombinations. In fact, in many of the gas combination tests, often one of the gaseswas run at a very low concentration and its response was overwhelmed by the othergas’s strong response. In other words, the detectable concentration of a gas might behigher if there exist other highly responsive gases.

22.6.4

STS-95 Flight Data Analysis Results

The resistance vs. time data that were returned from STS-95 showed that there wereseveral gas events in addition to the daily marker. The daily marker, exposure to apropanol and water medical wipe, was added to the experiment so that operationof the device over the entire period could be confirmed. The initial analysis selectedthe daily markers and identified them as 2-propanol plus a humidity change. Theseidentifications were confirmed by comparison of crew log times with the time of theevent in the data. While the hope in an experiment such as this one is that there will beseveral events that test the ability of the device, such events would certainly be anom-alous events in the space shuttle environment. Software analysis identifies all events


that were not propanol wipe events as humidity changes. Most of those changes can bewell correlated with the humidity changes recorded by the independent humidity mea-surements provided to JPL by JSC. The events are not completely correlated in timebecause the humidity sensor was located on the stairway between themid-deck and theflight deck, and the electronic nose was located in the mid-deck locker area near the airrevitalization system intake. Those events identified as humidity changes but not cor-related with cabin humidity change are likely to be caused by local humidity changes;that is, changes in humidity near the electronic nose that were not sufficient to cause ameasurable change in cabin humidity.Figure 22.7 shows the correlation of cabin humidity with electronic nose response in

several cases. There are visible dips in the traces at times 19:00, 20:52, and 0:07 CST,November 2–3, 1998. These dips are the changes in air composition, and thus resis-tance, during the baselining cycle, when air is directed through the charcoal filter.Piecewise baseline fitting is based on the resistance during the baselining cycle.Software analysis of the flight data did not identify any other target compounds as

single gases or as mixtures. The independent analysis of collected air samples, inwhich the samples were analyzed at JSC by GC-MS, confirmed that no target com-pounds were found in the daily air samples in concentrations above the electronicnose detection threshold. It is not surprising that the only changes the electronic

Fig. 22.7 Sample data from STS-95 electronic nose flight experiment.

Circles are the independent humidity measurements in the

stairway from mid-deck to flight deck. Polymer sensor responses:

(A) poly (2,4,6-tribromostyrene), (B) polyamide resin, (C) poly(ethylene

oxide), (D) poly(4-vinylphenol)


nose saw were humidity changes, and it is because events were not expected that theexperiment included the relatively uncontrolled daily marker events. There were nocompounds that the electronic nose would have indicated as unidentified events pre-sent in the air samples.

22.7

Future Directions

22.7.1

Sensors

The number of sensors in the second-generation electronic nose will remain at 32. Thenumber of polymers may be expanded beyond 16 in order to make sub-groups ofpolymers that have been selected for response to particular classes of compoundswithin the set of 32 sensors.To determine the set and sub-groups of polymers for the set of some 20 target com-

pounds, a model of polymer-analyte interaction is under development. This modeltakes account of such parameters of equilibrium constant of solvation of the analytein the film, analyte diffusion in the film, and the effect of the conductive medium. Themodel will be used to select polymer suites with maximum separation in patterns forparticular analyte suites. This type of selection may result in using some subset of the32 sensors for various patterns.It is possible that the use of carbon as the conductive medium is responsible for the

nonlinearity of responses at low concentrations. Studies of the use of metals such asgold or oxides of transition metals as the conductive medium are underway. It hasbeen found that alcohols and ketones desorb from metals more rapidly than theydo from carbon.

22.7.2

Data Acquisition

Current research in data acquisition is investigating the use of frequency dependentmethods for data acquisition. ACmethods are generally more sensitive than DCmeth-ods of measurements; AC methods may allow the use of thinner, higher resistancefilms, thus increasing film sensitivity. Some sensors exhibit high frequency noise,which may be caused by local heating while resistance is measured, by inhomogen-eously distributed carbon, or by variable thickness of the film. Thinner sensors couldeliminate some sources of noise, and AC measurements may filter out some of thenoise.To test whether high frequency noise can be filtered by AC methods, a single sen-

sing film of polyethylene oxide/carbon was exposed to 2500 ppm methanol and theimpedance measured at several frequencies, including DC resistance. As shown inFig. 22.8, there is substantially less baseline drift when sensor response is plotted


as dI=I0 where I is the impedance, than there is in the same sensor measured at DC,but higher frequency noise is not diminished at the frequencies at which impedancewas measured. The decision whether to change over to using AC measurement tech-niques will consider the efficiency of removing baseline drift through digital filteringin the data analysis process vs. the electronic requirements for AC measurements. Itmay be sufficient to measure DC resistance and remove the high frequency noise byincreasing the number of signal averages from 16 to 32 or 64 and remove the lowfrequency noise by digital filtering in data processing, as described above.

22.7.3

Data Analysis

Though the data analysis software developed for this electronic nose program washighly successful for its application, several improvements can be made in the fu-ture. The overall approach to data analysis will not be modified in the second-genera-tion device. The major change will be the addition of real-time or quasi-real-time ana-lysis. For the flight experiment, data were stored and analyzed after the flight. Forground test experiments in which events are manufactured to challenge the electronicnose, the goal is to have data analyzed within minutes of detection. For faster dataanalysis, it will be necessary to implement a reliable automated event identificationroutine and to translate the identification and quantification routines from Matlabinto C.There will also be some adjustments to the identification and quantification rou-

tines. First, the current data analysis software uses all 32 sensors’ responses as in-

Fig. 22.8 Response of a polymer/carbon film of polyethylene

oxide to 2500 ppm of methanol, at three frequencies of impedance

measurement and DC resistance measurement


put. Though each sensor’s response was weighted in the analysis in order to maximizethe differences between similar signature patterns observed for different gas com-pounds, it was not done systematically and therefore was not necessarily optimal.In the second generation, the selection of the to-be-used sensor set and their corre-sponding weights will be optimized by maximizing distances between gas signatures.The distance between the signatures for gas m and gas n, dmn, is defined as

dmn ¼1

N

XN

i

jdRm;i � dRn;ij

where Rm;i is the ith sensor’s normalized (fractional) resistance change for them th gasand the summation is over N numbers of sensors used.Second, the core of our data analysis software is the modified LM-NLS method,

which is heavy with matrix operations and largely determines the entire data analysisspeed. Matrix operation speed is known to be exponentially slower as the matrix sizeincreases. One way to increase speed is to reduce the size of the matrix dynamically inoperation by incorporating sensors’ characteristic response information, such asknown negative or no responses to certain gas compounds.This characteristic response information can also be used for compounds that can-

not be identified by the software; sensors which are known to respond or not to re-spond to particular functional groups can be sampled for a match. Thus, while it maynot be possible to identify unexpected compounds, it will be possible to classify themby functional group.In the first generation electronic nose, data analysis is performed on the steady-state

signal produced by changes in the atmosphere. For air quality monitoring, using thesteady-state signal is, in general, acceptable, as a transient will not remain in the en-vironment long enough to do harm.However, there are toxins that can be hazardous astransients. With automated event determination, analysis can begin as soon as theresistance measurement passes the preset threshold rather than waiting for steady-state to be reached. In addition, if desorption time is a function of the conductivemedium, then it may be possible to use the kinetics of sensor film response for iden-tification and quantification. Several compounds, such as ammonia, can be identifiedby the shape of the response curve upon visual inspection of the curve. Quantificationof the kinetics of response may enable identification of transients.

22.8

Conclusion

The results of the flight experiment were somewhat disappointing to the experimen-ters, while satisfying to the crew. There were no anomalous events, and the electronicnose was not challenged to identify compounds for which it had been trained. Never-theless, the experiment was successful. The electronic nose detected changes in hu-midity and the presence of the daily marker, was able to identify and quantify thechanges, and was able to use the training set made in the laboratory to do the data

22.8 Conclusion 545545

analysis. Further work in development of the JPL electronic nose will involve substan-tial challenge to the device and to the analysis software, with blind testing, mixtures,and unknowns that can be identified by functional group.

Acknowledgements

The research reported in this paper was carried out at the Jet Propulsion Laboratory,California Institute of Technology under a contract with the National Aeronautics andSpace Administration, and was supported by NASA Code UL.

References

1 C. C. Chan, H. Ozkaynak, J. D. Spengler, L.Sheldon. ‘Driver Exposure To Volatile Or-ganic-Compounds, CO, Ozone, and NO2

Under Different Driving Conditions,’ Envi-ron. Sci. Technol., 25, 964 (1991).

2 P. L. Leung, R. M. Harrison. ‘Roadside andIn-vehicle Concentrations of MonoaromaticHydrocarbons,’ Atmospheric Environment,33, 191 (1999).

3 M. A. Ryan, N. S. Lewis. ‘Low Power andLightweight Vapor Sensing Using Arrays ofConducting Polymer Composite Chemical-ly-Sensitive Resistors,’ Enantiomer, 6, 159(2001).

4 M. A. Ryan, M. L. Homer, M. G. Buehler, K.S. Manatt, F. Zee, J. Graf. ‘Monitoring theAir Quality in a Closed Chamber Using anElectronic Nose,’ Proceedings of the 27th In-ternational Conference on Environmental Sy-stems, Society of Automotive Engineers, 97-ES84 (1997).

5 M. A. Ryan, M. L. Homer, M. G. Buehler, K.S. Manatt, B. Lau, D. Karmon, S. Jackson.‘Monitoring space shuttle Air for SelectedContaminants Using an Electronic Nose,’Proceedings of the 28th International Conferenceon Environmental Systems, Society of Auto-motive Engineers, 981564 (1998).

6 M. A. Ryan, M. L. Homer, H. Zhou, K. S.Manatt, V. S. Ryan, S. P. Jackson. ‘Operationof an Electronic Nose Aboard the spaceshuttle and Directions for Research for aSecondGenerationDevice,’ Proceedings of the30th International Conference on Environmen-tal Systems, Society of Automotive Engin-eers, 00ICES-259 (2000).

7 M. S. Freund, N. S. Lewis. ‘A ChemicallyDiverse Conducting Polymer-Based Elec-

tronic Nose’, Proc. National Academy of Sci-ence, 92, 2652, (1995).

8 M. C. Lonergan, E. J. Severin, B. J. Doleman,R. H. Grubbs, N. S. Lewis. ‘Array-BasedSensing Using Chemically Sensitive, Car-bon Black-Polymer Resistors’, Chem. Mate-rials, 8, 2298 (1996).

9 E. J. Severin, B. J. Doleman, N. S. Lewis. ‘AnInvestigation of the Linearity and Responseto Mixtures of Carbon Black-Insulating Or-ganic Polymer Composite Vapor Detectors’,Anal. Chem., 72, 658 (2000).

10 K. J. Albert, N. S. Lewis, C. L. Schauer, G. A.Sotzing, S. E. Stitzel, T. P. Vaid, D. R. Walt.‘Cross-Reactive Chemical Sensor Arrays,’Chem. Rev., 2595 (2000).

11 Spacecraft Maximum Allowable Concentrati-ons for Selected Airborne Contaminants, Vols.1 & 2, National Academy Press,Washington,DC (1994).

12 P. N. Bartlett, J. W. Gardner. Electronic Noses:Principles and Applications, Oxford Univer-sity Press, Oxford (1999).

13 G. Stang. Linear Algebra and its applications,2nd edition, Academic press, New York,1980.

14 C. Lawson, R. Hanson. Solving Least SquaresProblems, S.I.A.M. Press, Philadephia, 1995.

15 R. Storn. ‘On the usage of differential evo-lution for function optimization,’ BiennialConference of the North American Fuzzy In-formation Processing Society, NAFIPS, IEEE,519 (1996).

16 M. Lampton. ‘Damping-Undamping Strate-gies for the Levenberg-Marquart NonlinearLeast-Squares Method,’ Comput. Phys., 11,110 (1997).


23

Detection of Explosives

Vamsee K. Pamula

Abstract

Detection of explosives is one of the problems for which an electronic nose is the mostappropriate technological solution. Currently, landmines are detected by dogs, whichuse their noses to sniff explosive vapors or particles. With the current technology itwould take about a thousand years and hundreds of billions of dollars to clear all themines in the world [1]. An electronic nose used in this context would save human lives,work round the clock without getting tired, and could improve security for all humans.In this chapter, a review of different state-of-the-art technologies developed for sen-

sing explosives for the detection of landmines is presented. Various sensors are com-pared with respect to their detection limits of explosives such as trinitrotoluene anddinitrotoluene, because they are found to be the predominant explosives found inlandmines.The system developed by Nomadics is identified to be the best of the currently avail-

able detection devices. Future success of the electronic nose in this area depends onthe ability of these devices to outperform the dogs. Such systems will emerge withinthe next decade.

23.1

Introduction

There are some horrifying facts about landmines [1].Around the world they claim the life of a victim or maim one victim every 22 min-

utes. There are about 120 million unexploded landmines lurking in 70 countriesaround the world. With the current technology, 4.6 square miles of landmine infestedarea can be cleared per year. For every mine that is cleared, 20 newmines are laid. Thecost of a mine ranges from $3–$5, whereas clearing it costs $1000. On average, forevery 5000 mines removed, one mine-clearer is killed and two others are injured. Itwould cost about $120 billion and take a thousand years to clear all the mines in theworld with the current technology.


547547

The insidious nature of mines has stimulated significant research – spanning overhalf a century – on techniques for mine detection, identification, and remediation. Inthe context of detection, technologies that have been explored include magnetic metaldetectors, ground penetrating radars, optical, infrared, acoustic, X-ray, and thermalneutron analysis. The identification problem is even more daunting, requiring tech-nology and algorithms that can selectively detect landmines among the clutter. Devel-opment of a sensor that is both sensitive and selective for all kinds of landmines underall conditions is almost impossible. Realizing that there is no single sensor that works,a multi-sensor approach needs to be taken for the detection of the mines.Currently, demining is done by humans with simple metal detectors or a human-

dog team. Metal detectors have very high false alarm rates due to the metallic junk in aminefield. Also, they cannot detect plastic mines, which have almost no metallic con-tent. Dogs have proven to be the most effective mine detectors to date, although theyhave limited attention spans measured in tens of minutes. Mine clearing by teamsusing metal detectors proceeds at 200 meters/day, whereas a human-dog team allows2–4 kilometers/day to be cleared [2]. Dogs search by placing their noses close to theground and inhaling vapors as well as solid particles of the material to be detected. It isnot clearly understood whether they detect the pure explosive, some impurities asso-ciated with the explosive, or some signature of the odorant [3]. A sensor that combinesboth the vapor and particle detection will be the closest approximation to a dog’s nose.Such a sensor will work round the clock!One of the most important military and humanitarian applications of the electronic

nose is to sniff out landmines. Most of this chapter will concentrate on various tech-nologies developed to date for sensing explosive vapors in this context.

23.2

Previous Work

Semiconductor vapor sensors have been developed in the past [4, 5]. A complementaryapproach to obtaining increased sensitivity is to detect the particles of explosive resi-dues in addition to the vapors around a landmine. This particle sensor, used in con-junction with the vapor sensor, would approximately mimic a dog’s functionality indetecting the landmines. Most of the commonly used landmines contain 2,4,6-trini-trotoluene (TNT) and/or 1,3,5-trinitro-1,3,5-triazocyclohexane (RDX) as the explosivecharge. It has been observed from experiments that at least a few nanograms of TNTexplosive particles are present in the vicinity of landmines. For a buried landmine,vapors of the explosive charge emanate from the casing of the mine into the soiland further into the air above it. Many explosives have very low vapor pressures, in-cluding TNT and RDX. The equilibrium vapor concentration of TNT is about 70 pi-cograms/mL of air at 298 K [6]. Due to low vapor pressures of the explosives, theconcentration of the vapors above a landmine are very low. Most of the contaminantspresent in TNT have a higher vapor pressure than TNT itself. For a particular mine,2,4-dinitrotoluene (DNT) vapors were found to be 20 �more concentrated than thoseof TNT vapors, even though DNT accounted for less than 1% of the explosive by mass

23 Detection of Explosives548

[7]. The mixture of the compounds escaping the landmine form a ‘chemical signature’indicative of the explosive present in the landmine.Significant success has been reported by using trace explosive particles for sub-

stance identification [8]. Indeed, both RDX and TNT have been detected at higherlevels than expected, when the vapor sampling system was augmented with a traceparticle collector [9]. At Auburn University, the researchers found that dogs thatwere trained to detect TNT learn to use DNT as a detection odor signature. In theirexperiments the dogs were able to sense DNT in fractions of the parts per billion range.While evaluating the nature of olfaction to determine whether the particles or vaporsplay the main role in detection, they found that particles did not reach the olfactoryepithelium of the dogs, which suggests that the particles may not be a likely basis for adetection scheme [10]. Researchers at Penn State University studied the flow patternsof air generated by a dog while sniffing. They observed that the sniffer must approachthe scent source in close proximity to avoid dilution of the scent and disruption bywind. They also point out that particles may play a role in scenting as they observethat the particles on the surface become airborne while the dog is sniffing. Basedon this research, the electronic dog’s nose should be aerodynamically designed tosniff efficiently [11]. A single solution does not exist for the landmine problem, there-fore a variety of sensors would be needed to successfully replace dogs.Sandia National Labs’ studies [12, 13] indicate that the dogs seem to work better in

wet conditions because water competes for soil sorption sites thereby enabling releaseof explosive vapors. Also, their experiments on a buried landmine made of TNT re-vealed that the vapor above the soil is that of DNT, and also that DNT passes throughthe mine casing more easily than TNT, therefore DNT ends up in a higher concen-tration on the surface of the soil.

23.3

State-of-the-art of Various Explosive Vapor Sensors

In this section, we will cover the work performed in developing electronic noses forexplosive detection. In 1997, the Defense Advanced Research Projects Agency (DAR-PA) developed a high-risk technology development program to detect mines throughtheir chemical signature. In view of the arguments presented in the previous section, anumber of researchers concentrated their efforts in developing a sensor whichmimicsa dog’s nose, if not exactly at least functionally, for the detection of explosives. Researchwas performed on mammalian olfaction which stimulated new ideas for chemicalsensing.

MIT

Swager’s group from MIT has developed fluorescent conjugated polymer thin-filmsthat have high affinities for DNT, TNT, and related compounds. The incorporation ofrigid three-dimensional pentiptycene moieties in the conjugated polymer backboneprevent p-stacking or excimer formation, which allows the diffusion of analytes


into the dense polymer films. The fluorescence of the films reduces in a few secondsdue to the vapors of TNT and DNT. The authors believe that the reduction in fluor-escence is due to the exchange of the excited electrons of the polymer film with theelectron-deficient DNT or TNT molecules [14]. In this process, TNT short circuits themigrating electron by allowing it to jump back to the valence band without the emis-sion of light as shown in Fig. 23.1. Since the polymer molecules are wired serially, theTNT short circuit amplifies the reduction in fluorescence.

Duke University

At Duke University, a microelectromechanical systems-based explosive particulatesensor was developed [15]. The purpose of this sensor is to complement the vaporsensors by detecting the explosive particles from the soil to aid more accurate detec-tion. As mentioned earlier, a few nanograms of DNT and TNT are present on thesurface of the soil near a buried landmine.The sensor comprises of a bimetallic gold (0.5-lm-thick)/polysilicon (1.5-lm-thick)

surfacemicromachined cantilever. Due to a large difference in the thermal coefficientsof expansion between gold and polysilicon, the cantilever deflects down upon heating.

Fig. 23.1 Fluorescence quenching mechanism in polymer chemosensor films

Fig. 23.2 Schematic of a cantilever’s response to a deflagrating explosive particle


A few nanograms of pure DNT was placed on the pad of the cantilever as shown inFig. 23.2. When the cantilever is heated without an explosive particle, it deflects down-wards monotonously. But when the cantilever is heated with the explosive particle, thecantilever’s deflection shows an additional dip around the temperature when the ex-plosive particle disappears from the pad. The cantilever is heated at 6 8C/sec. It isassumed that the deflagrating explosive particle is generating this additional heat.The magnitude of the dip in deflection corresponds to the size of the DNT particle[16]. For nanograms of DNT particles, it was always observed that the particles releaseenergy giving rise to the dip around 110–120 8C.

Tufts

Dickinson et al. from Tufts University have developed the first optical artificial nose[17]. As explained in Chapter 8 (Optical electronic noses), thousands of bead sensorsare randomly dispersed across an etched fiber optic tip. Each bead sensor within thearray is a porous silica bead impregnated with the environmentally sensitive dye, NileRed, which is a solvatochromic dye (highly sensitive to the polarity of its local envir-onment) as shown in Fig. 23.3. The sensor array is connected to a charge-coupleddevice (CCD) camera detector which monitors the fluorescence with an imaging sys-tem. On exposure to a particular vapor, the bead sensors undergo characteristic andreproducible fluorescence intensity and wavelength shifts that are used to generatetime-dependent fluorescence response patterns. Each of these sensor beads iscross-reactive (not analyte specific and broadly selective) and produces a unique fluor-escence signature in response to different analytes. These patterns can be used to trainpattern recognition computational networks. On subsequent exposure to the same

Fig. 23.3 Fabrication of a sensor bead array on the tip of a fiber optic cable

[19]. Reproduced with permission from Anal. Chem., (1999), 71, 2192–2198.

Copyright 1999 Am. Chem. Soc


analyte vapor, the system identifies the vapor by the characteristic response pattern ofthe sensors. They have found that the surface chemistry of the sensor favors attractionbetween the electron-accepting nitroaromatic compounds such as DNT and TNT, andthe highly adsorptive surface of the porous silica beads thus maximinzing the analyte-dye interactions [18, 19].It was demonstrated that the detection limit can be enhanced due to the increase in

the signal-to-noise ratio when the signal is collected over a thousand sensors and aver-aged as shown in Fig. 23.4. The sensors were able to respond to vapor concentrationsof DNT and TNT up to tens of ppb (parts per billion).

Draper Laboratories Caltech

Carbon black-insulating organic polymer composite films are employed in an array ofvapor detectors. These vapor detectors are cross-responsive and respond by exhibiting

Fig. 23.4 Comparison of the response due to

250 and 1000 sensors for DNT vapor [18].

Reproduced with permission from Anal. Chem.

(2000) 72, 1947–1955. Copyright 2000 Am.

Chem. Soc

Fig. 23.5 Caltech/Draper sniffer assembly for landmine detection


a change in their resistance on exposure to a particular vapor. Each element of the arraycontains a different organic polymer as the insulating material. The resistance be-tween the electrodes of an element changes due the swelling of the polymer and variesdue to the differing gas-solid partition coefficients for the various polymers of thedetector array. No individual sensor is uniquely responsive to a given analyte, butthe swelling pattern across all the elements of the array is unique for each odor.The response is matched to an existing pattern that aids in the classification andquantification of analytes in the vapor phase. The pattern type of the response allowsidentification of the vapor and the steady-state pattern height allows quantification ofthe analyte.In association with Draper Laboratories, Caltech’s vapor sensors were incorporated

into a sniffer that collects the volume of air above a mine and delivers it to the sensorarrays. The sniffer head has two sensor chips opposite each other through which thesniffed vapor is investigated as shown in Fig. 23.5. They were able to detect DNT in thelow ppb range in less than 5 seconds of exposure to the vapor [20].

Rockwell Science Center

Rockwell Science Center developed a miniaturized mass detection system, which hasan array of polymer-coated thin-film resonators (TFR) operating at 2 GHz as shown inFig. 23.6. An array of eight TFR sensors, which change their resonance frequencies asa function of the mass of the vapor adsorbed in the polymer coatings, has been devel-oped to detect vapors of TNT and its decomposition products for landmine detection.The surface coatings of these sensors preferentially adsorb specific types of chemicalvapors. The TFRs were fabricated using AlN as the piezoelectric film with a thicknessof � 1.5 lm. The polymer coating was sprayed onto the TFR into thin films becausethicker coatings degrade the quality of the acoustic resonance [21]. Out of the eightsensors, three were coated with polymers that have affinity for aromatic nitrates,one with affinity to water, three for varying degrees of adsorption of organic materi-als, and one left uncoated as a general reference [22]. The sensors recognize the targetvapors and quantify their concentration by comparing the pattern of the response,which is based on the magnitude and time-dependence response of all the coateddetectors, with a known pattern for that particular vapor. The system was able to detectDNT at few ppb concentration in air in the absence of large background levels ofinterference.

Fig. 23.6 Cross-section of the thin film

resonator microbalance for vapor detection

adapted from [22]


Naval Research Laboratory

The US Naval Research Laboratory has developed polymer coatings for surface acous-tic wave (SAW) sensors to be used for explosive vapor detection. SAW resonator de-vices (acquired from SAWTEK Inc, Orlando FL) were spray-coated with various poly-mer films to evaluate the most promising polymers for the vapor detection of nitroaro-matic explosive compounds. As the coatings absorb the vapors, the resonance fre-quency of the polymer-coated SAW device decreases due to increased mass load-ing. Several hexafluorisopropanol-functionalized aromatic silicon-based polymershave been prepared and coated on the SAW devices to enhance the detection of ni-troaromatic analytes. The polymers are strongly hydrogen-bond acidic which reversi-bly sorbs nitroaromatics and other hydrogen-bond basic vapors. They estimate that thedetection limit for these sensors will be < 100 ppt (parts per trillion) for DNT [23].

Texas Instruments

Texas Instruments’ Spreeta� sensor, when used in conjunction with a sniffer fromSPEC (Systems & Processes Engineering Corporation), closely mimics the dynamicsof a dog’s nose. The SPEC sniffer has six exhaler orifices from where the particulatesare stirred up and then drawn through the sampling orifice, as shown in Fig. 23.7.These particulates then impinge on a membrane. TNT and DNT from the sampledissolve into the membrane and rapidly diffuse to the liquid side. An automated mi-crofluidics system mixes the sample with antibodies, which can then be delivered tothe Spreeta sensor for analysis using a bioassay.The Spreeta sensor is based on the principle of surface plasmon resonance (SPR).

SPR can be employed to study the kinetics of molecular binding events in real-time.

Fig. 23.7 SPEC’s explosive particulate sampler


On an active gold surface, the refractive index changes can be monitored by SPR. Asshown in Fig. 23.8, the liquid sample containing dissolved TNT and DNT is deliveredto the top of the gold’s surface through a flow cell. An AlGaAs infrared light-emittingdiode with a transverse magnetic polarizer excites the surface plasmons in the goldfilm at the liquid interface. The reflected light is captured on a photodiode, whichyields the refractive index of the liquid on the sensor. It also has a built-in temperaturesensor because refractive indices vary with temperature [24].A glass slide coated with gold is used as the SPR active surface. The slides were

coated with trinitrobenzene (TNB) and bovine serum albumin (BSA), which servesas an immobilized ligand. Since TNT is a small molecule and SPR detects changesin the surface mass concentrations of an analyte, a competition assay is used. A TNTantibody with a large molecular weight is used, the binding of which is competitivelyprevented by free TNT in the solution. When the TNT antibody binds to the TNB-BSAgroups in the gold surface, an increase in refractive index is observed. When TNT ispresent in the solution, however it reduces the rate of antibody adsorption leading to areduced value of the refractive index. The limit of detection of this sensor is 1 ppm(parts per million) of TNT (1 mg of TNT/1 kg of soil) [26].

ETC Laboratories

EIC Laboratories have made a vapor sensor based on surface-enhanced raman spectro-scopy. A laser interrogates an area of a microscopically roughened metal for adsorbedanalytes. The vibrational modes of the analyte adsorbed on the metal are enhancedcompared to their nonresonant Raman intensities. The metal surface can be madeto selectively adsorb compounds of similar chemical structure by choosing a combina-tion of the metal surface, the degree of roughness, the degree of oxidation of the sur-face, and other factors. The Raman spectra are collected using an echelle spectrographcoupled to an air-cooled CCD camera. The raw vapor spectra are presented to a soft-ware algorithm which creates a curve fit and compares it to the anticipated curve forDNT, and therefore ascertains the presence or absence of the analyte. The laser signalis delivered through a fiber-optic probe and the spectrometer was packaged for use inthe minefield [27].

Fig. 23.8 SPR sensor for

detecting DNT and TNT

dissolved in a liquid. Adapted

from [25]


Figure 23.9 shows the Raman spectra of TNT and its primary impurities, DNT andDNB. EIC sensors were able to detect the presence of sub-ppb concentration of DNTover aqueous solutions.

Quantum Magnetics

Quantum Magnetics (QM) has developed a sensor based on quadrupole resonance,which is similar to the magnetic resonance imaging technique used in the medicalindustry. QM in a subsidiary In Vision Technologies, which produces X-ray com-puted-tomography machines for scanning airport cargo and baggage. The QM instru-ment is not specifically an electronc nose in that it does not detect the explosivesthrough vapors or particles, but it is chemically specific enough to detect explo-sives. The device sends short pulses of radio waves at specific frequencies that reso-nate with the atomic nuclei of the explosive molecules. At the end of the pulsing, thenuclei send out a weak radio signal. Out of 10 000 compounds studied, there has notbeen an overlap in the responses. 14N nuclei gives the characteristic signal in the caseof TNT and RDX. There are no false alarms due to other nitrogen-containing com-pounds available in the background because the signal is either not given or is givenat a sufficiently different resonance frequency. This signal depends on the molecularstructure of the atoms, which is analyzed by a computer to identify the material [28].

Fig. 23.9 Raman spectra of (a) TNT, (b) DNT, and (c) DNB. Raman

spectra were normalized. The intensity axis was not plotted for

illustrative purposes [29]. Reproduced with permission from Anal.

Chem. (2000) 72, 5834–5840. Copyright 2000 Am. Chem. Soc


Field tests performed to detect RDX- and TNT-based nonmetallic antitank andantipersonnel mines, yielded 100% probability of detection with very low false alarmrates [29].

Nomadics

Another scheme for detecting explosives is with electrochemical sensors, which yieldqualitative information about the presence of these compounds. In voltammetry, thepotential of a sensor is held constant and the sensor detects the current resulting fromelectrochemical oxidation or reduction. In this case, the signal may be disturbed due tothe presence of other substances, or adsorbates may form on the electrode surfacerendering the sensor less sensitive over time. To overcome these problems, cyclicvoltammetry was employed where a time-varying potential was applied on a gold elec-trode in sulfuric acid and the resulting current recorded as a function of the potential.This sensor is in its early stages and the detection was demonstrated only for TNT inthe gaseous phase [30].

23.4

Case Study

Nomadics Inc., in Stillwater, Oklahoma, is developing a highly sensitive and selectivelandmine detector based on the detection of the trace amounts of TNT vapors emanat-ing from a landmine. Nomadics’ Fido (Fluorescence Impersonating Dog Olfaction)

Fig. 23.10 Nomadics’ Fido

landmine detector [31]

23.4 Case Study 557557

landmine detector has demonstrated the ability to detect landmines under field con-ditions, and is perhaps one of the most promising explosive detection technologies onthe market. It is based on the fluorescent polymer beads developed by Swager’s groupat MIT as detailed in 23.3 above [14].Nomadics’ landmine detector uses the same technology as shown in Fig. 23.0, which

is based on amplifying fluorescent polymers. The fluorescence of many polymers de-crease when a single molecule of the nitroaromatic compound binds to a polymer. Inits handheld configuration, the system consists of a small sensor module, detectorelectronics, operator display/control panel, battery pack, and mounting arm. Asshown in Fig. 23.10, a blue-light fluorescence excitation laser is collimated and filteredto pass a narrow band of light around 405 nm. This beam is normally incident throughtwo borosilicate glass substrates coated on the surfaces with spin-cast thin films of thepentiptycene polymer. The coated substrates are held in a cassette that can be easilyremoved from the device to facilitate the replacement of the polymer films. A small gapis maintained between the two substrates by a thin-U-shaped spacer. The spacer formsa seal along three edges of the polymer-coated inner faces of the substrates. The sub-strates are not sealed along the fourth edge. This opening serves as a sample inlet.Vapor is drawn through the inlet into the sampling volume between the two sub-

strates by a small pump. The pump is connected to an exit port bored through thespacer on the side opposite the inlet. Transmitted incident light, along with theemitted fluorescent light, is passed through a filter which passes only the fluorescencesignal at 460 nm. The intensity of the emission from the films is thenmeasured with aphotomultiplier tube (PMT) [32].Seventy one soil and water samples containing landmine explosives with potential

interferants and blanks were presented to this detector, which has successfully iden-tified each of them without any single error in the laboratory conditions. Blind fieldtesting was performed by DARPA at Fort LeonardWood test-field over real landmines.The probability of detection was 0.89 with a probability of false alarms of 0.27. Noma-dics soon hopes to be in full production of field-deployable Fido landmine detectors.The current sensor prototype can instantly detect in the parts per quadrillion range,

Tab. 23.1 Vapor Detection limits of various systems

Detection Method Limits (/mL)

High Performance Liquid Chromatography Ultraviolet (HPLC-UV) 1 nanogram (ng)

Mass Spectrometer 800 picogram (pg)

High Performance Liquid Chromatography 600 pg

Electrochemical (HPLC-EC)

Thermal Energy Analysis (TEA) 30–50 pg

Mass Spec – Chemical Ionization (MS-CI) 20 pg

Airport Sniffers 20 pg

Electron Capture Detector (ECD) 10 pg

Micro Electron Capture Detector (lECD) 1 pg

Ion Mobility Spectrometer (IMS) 50–100 femtogram (fg)

Nomadics Amplifying Fluorescent Polymer 1 fg


which is better than most of the current explosive detection methods (Table 23.1). Tothe author’s knowledge, this is the first sniffer capable of detecting landmines in thefield with performance comparable to that of dogs.

23.5

Conclusions

In conclusion, we have presented an overview of sensors that detect either the vaporsor particles of the explosives commonly found in landmines. The sensors aremade in avariety of technologies, each having their own advantages and disadvantages for fielddeployment. Most of the sensors presented in this chapter are beyond the proof-of-concept stage and many are driven by the industry for commercialization, thoughthere are no commercial products available in the market yet that can sniff landminesin the field. Among the electronic noses made for explosive detection, currently No-madics’ FIDO landmine detector has shown capabilities that match those of dogs. Thesuccessful detector will have characteristics such as portability, high sensitivity to theexplosive vapors and selectivity to detect only those vapors among clutter, a friendlyinterface for the deminers, very low false alarm rates with lowmaintenance, and will bevery robust.

23.6

Future Directions

Once a commercially viable electronic nose for landmine explosive detection is avail-able, the potential customers include professional deminers, humanitarian demininggroups like the United Nations and the International Committee of the Red Cross,various non governmental organizations, land and economic developers, and govern-ments of countries affected by landmines. Currently, there are about 120 millionmines deployed around the world, which would cost about 120 billion dollars to de-mine. This presents a huge market opportunity for any company that comes up with asuitable solution. Some research groups have already demonstrated that the elctronicnoses developed by them have comparable sensitivity to that of a dog’s nose. With theincreasing awareness of the landmine problem and various companies and university-based research groups around the world working on the problem, it may not be longbefore a commercial electronic nose, which provides a better solution than a dog inmany ways other than just the nose aspect, will successfully emerge.An electronic nose to detect landmines and explosives would be required to operate

in situations that will be dangerous to human life. One that is integrated with tele-operation capability or a robot will be far more attractive in such situations, butsuch an autonomous electronic nose is still years away from becoming a reality.


References

1 http://www. icrc. org/eng/mines2 A. M. Prestrude, J. W. Ternes. Proc. of SPIE,2093, (1994) 633–643.

3 L. J. Myers. Critical Reviews. CR42, (1994),93–103.

4 R. T. Howe, R. S. Muller. IEEE Transactionson Electron Devices, ED-33, 18–19; (1986)499–506.

5 A. Wilson, M. Tamizi, J. D. Wright. Sensorsand Actuators B (Chemical), 18–19, (1994)511–514.

6 J. Yinon, S. Zitrin. Modern Methods andApplications in Analysis of Explosives, JohnWiley and Sons, New York, (1993).

7 V. George, T. F. Jenkins, J. M. Phelan, D. C.Leggett, J. Oxley, S. W. Webb, P. H. Miyares,J. H. Cragin, J. Smith, T. E. Berry. Proc. ofSPIE, 3710, (1999) 258–269.

8 T. G. Sheldon, R. J. Lacey, G. M. Smith, P. J.Moore, L. Head. Proc. of SPIE, 2092, (1994),145–160.

9 W. R. Davidson, W. Scott. Proc. of SPIE,2092, (1994), 108–119.

10 J. M. Johnston, M. Williams, L. P. Waggo-ner, C. C. Edge, R. E. Dugan, S. F. Hallowell.Proc. of SPIE, 3392, (1998), 490–501.

11 G. S. Settles, D. A. Kester. Proc. of SPIE,4394, (2001).

12 S. W. Webb, J. M. Phelan. Proc. of SPIE,4394, (2001), 474–488.

13 V. George, T. F. Jenkins, J. M. Phelan, D. C.Leggett, J. Oxley, S. W. Webb, P. H. Miyares,J. H. Cragin, J. Smith, T. E. Berry. Proc. ofSPIE, 4038, (2000), 590–601.

14 J.-S. Yang, T. M. Swager. J. Am. Chem. Soc,120, (1998), 11864–11873.

15 V. K. Pamula. Ph. D. Thesis. Department ofElectrical and Computer Engineering, DukeUniversity, (2001).

16 V. K. Pamula, R. B. Fair. Proc. of SPIE, 4038,(2000), 547–552

17 J. White, J. S. Krauer, T. A. Dickinson, D. R.Walt. Nature, 382, (1996), 697–700.

18 K. J. Albert, D. R. Walt. Anal. Chem., 72,(2000), 1947–1955.

19 T. A. Dickinson, K. L. Michael, J. S. Krauer,D. R. Walt. Anal. Chem., 71, (1999), 2192–2198.

20 S. M. Briglin, M, C, Burl, M. S. Freund, N. S.Lewis, A. Matzger, D. N. Ortiz, P. Toku-maru. Proc. of SPIE, 4038, (2000), 530–538.

21 P. Kobrin, C. Seabury, C. Linnen, A. Harker,R. Chung, R. A. McGill, P. Matthews. Proc.of SPIE, 3392, (1998), 418–423.

22 C. Linnen, P. Kobrin, C. Seabury, A. B.Harker, R. A. McGill, E. J. Houser, R.Chung, R. Weber, T. Swager. Proc. of SPIE,3710, ( 1999), 328–334.

23 E. J. Houser, R. A. McGill, V. K. Nguyen, R.Chung, D. W. Weir. Proc. of SPIE, 4038,(2000), 504–510.

24 J. Mendelez, R. Carr, D. U. Bartholomew, K.Kukanskis, J. Elkind, S. Yee, C. Furlong, R.Woodbury. Sensors and Actuators B, 35–36,(1996), 212–216.

25 R. G. Woodbury, C. Wendin, J. Clenden-ning, J. Mendelez, J, Elkind, D. U. Bartho-lomew, S. Brown, C. Furlong. Biosensorsand Bioelectronics, 13, (1998), 1117–1126.

26 A. A. Strong, D. I. Stimpson, D. U. Bar-tholomew, T. F. Jenkins, J. Elkind. Proc. ofSPIE, 3710, (1999), 362–372.

27 J. M. Sylvia, J. A. Janni, J. D. Klein, K. M.Spencer. Anal. Chem., 72, (2000), 5834–5840.

28 A. D. Hibbs, G. A. Barrall, P. V. Czipott, A. J.Drew, D. Gregory, D. K. Lathrop, Y. K. Lee,E. E. Magnuson, R. Matthews, D. C. Skvo-retz, S. A. Vierkotter, D. O. Walsh. Proc. ofSPIE, 3710, (1999), 454–463.

29 A. D. Hibbs, G. A. Barrall, S. Beevor, L. J.Burnett, K. Derby, A. J. Drew, D, Gregory, C.S Hawkins, S. Huo, A. Karunaratne, D. K.Lathrop, Y. K. Lee, R. Matthews, S. Milber-ger, B. Oehmen, T. Petrov, D. C. Skvoretz, S.A. Vierkotter, D. O. Walsh, C. Wu. Proc. ofSPIE, 4038, (2000), 564–571.

30 T. Berger, H. Ziegler, M. Krausa. Proc. ofSPIE, 4038, (2000), 452–461.

31 M. la Grone, C. Cumming, M. Fisher, D.Reust, R. Taylor. Proc. of SPIE, 3710, (1999),409–420

32 M. la Grone, C. Cumming, M. Fisher, M.Fox, S. Jacob, D. Reust, M. Rockley, E.Towers. Proc. of SPIE, 4038, (2000), 553–562.


24

Cosmetics and Fragrances

P. A. Rodriguez, T. T. Tan, and H. Gygax

Abstract

The use of electronic noses in the cosmetic and fragrance industry appears limitedwhen compared to other industries and areas of application, such as the food andbeverage industry, or the chemical, polymer, and plastic industries, or in environmen-tal and medical applications. However, the literature and the work we present in thischapter show that, with optimization, many challenging problems in the cosmetic andfragrance industry can be successfully addressed using electronic nose technology.In this chapter we describe key challenges and limitations of analytical instruments

expected when correlating their output with the human response to perfume-relatedsamples. We also include two industrial applications addressed by the use of commer-cially available instruments; one based on a chemical sensor, the other on a massspectrometer. They provide insights into the ability of electronic noses to matchand mimic the perception of odor by humans, as well as their ability to competewith well-established analytical methods. Good sensitivity, selectivity, and reproduci-bility were obtained in the two cases presented here.

24.1

Introduction

Perfumes, derived from plants or flowers, have been used for millennia as a means toenhance the quality of life. Today, perfumes are ubiquitous in society, we encounterthem in cosmetics, in the home environment, and in virtually every cleaning product.As a consequence, perfumery has become a global, multibillion-dollar industry.Although the industry employs modern, sophisticated analytical tools to ensure the

quality of their products, the creation of a winning fragrance is still an art. Skilledperfume designers (also known as perfumers), rely on intuition, market research,and knowledge of raw materials to create perfumes designed to meet the require-ments of a particular product. An important requirement, in addition to meetingcost constraints, is to deliver a fragrance that reinforces the product image. Thus,if you are developing a cleaning product, the perfume is likely to be required to delivera ‘clean’ fragrance.


561561

Translating words like ‘clean’ and ‘fresh’ into chemical formulas useful in a productis part of the art of perfumery. But, unlike music or painting, the art of perfumery hasfewer standardized tools to accomplish the objective of creating a masterpiece. Thedifficulty in developing the necessary tools is a consequence of our inadequacy inusing words to define odors. The closest we come to defining an odor is throughthe description of how it resembles other familiar, well-known odors. Interestingly,although there have been many attempts to describe or classify odors, no schemehas survived the test of time. Nevertheless, a successful classification scheme wouldprovide a useful framework for understanding odors, and would facilitate effortsaimed at identifying elusive, primary odorants.In addition to its academic importance, the need to describe and classify odors has

enormous economic consequences. This is true because description and classificationwould provide a link to understanding preference. Importantly, consumers through-out the world often use odor preference to discriminate between products that other-wise offer similar price/performance attributes. Description, classification, and under-standing preference are areas where ‘electronic noses’ or ‘chemical classification tools’could make unique contributions to perfumery.In this chapter we discuss the requirements, characteristics, and usage of commer-

cial electronic noses in the perfume and cosmetics industry. A comparison with a gaschromatography (GC)-based approach is also presented. The chapter ends with anassessment of the technology for future applications in this market segment.

24.2

The Case for an Electronic Nose in Perfumery

Perfumes are complex mixtures of volatile and semi-volatile organic compounds [1].Today, it is not uncommon for a commercial perfume to be prepared by mixing fifty toone hundred or even more perfume raw materials (PRMs). Furthermore, PRMs arenot pure chemical compounds. Becausemany are obtained by complex processes and/or derived from complex raw materials, PRMs may contain many isomers or evencompounds unrelated to the main odorant in the PRM. For example, the main odor-ant in GalbanumPRM is a pyrazine accounting for less than 0.01% of the total mass inthis PRM. As a consequence, we find it is not uncommon for a finished perfume tocontain hundreds or even thousands of distinct chemical compounds.Unfortunately, perfume complexity quickly adds to the perfume cost. This is true

because in addition to inventory costs, specifications for each PRM must be estab-lished and confirmed by analysis, and safety evaluations must be performed onthe many possible compounds present above a certain percentage.Presently, the chemical complexity of perfumes is mainly a consequence of the de-

sire by perfume designers to deliver perceptual complexity to consumers. It is therichness of the perception that makes perfumes so attractive to humans. Unfortu-nately, the link between chemical complexity and perceptual complexity has notbeen thoroughly examined. As a consequence, the optimum number of compoundsin a perfume has not been established. Recent work suggests it is possible to reduce

24 Cosmetics and Fragrances562

the chemical complexity of existing perfumes without any measurable reduction inperceptual complexity [2]. Reductions in the number and quantity of chemicals canbe achieved through the use of psychophysical principles and the use of humansto establish the relative importance of individual odorous compounds to the overallperfume fragrance.Although reductions of 20–40% in the mass and number of PRMs can be achieved

when examining existing perfumes, the process is not straightforward and requires anumber of iterations. Because the reasons to do the work are so compelling, i.e. costreductions, raw material inventory simplification, and the elimination of tens of thou-sands of metric tons of materials from the environment, the industry and the planetwould certainly benefit from a rapid and simple process to do the work.There are reasons to be optimistic about the use of an electronic nose as a tool to help

simplify existing perfumes and help design new, cost- and material-efficient fra-grances. Although the initial report by Axel and Buck [3] on the identification of hu-man genes that code for olfaction, suggested the existence of perhaps 1000 such genes,or corresponding ‘sensor molecules’, recent work suggests that number is signifi-cantly smaller. If true, the number of required sensors may approach the numberof ‘sensing channels’ measurable with a mass spectrometer as discrete ions, and re-ported as mass/charge (m/z). In addition, our work to understand the relative impor-tance of perfume odorants suggests that only 10 to 15 compounds contribute most ofthe intensity and character to any given perfume. We could call those compounds‘principal odorants’, and although they would be different for each perfume, wefind that most perfumes contain many of the same compounds as principal odor-ants. Thus, if sensors could be developed to be quasi-selective for those com-pounds, we would expect the resulting electronic nose to have near-perfect correlationwith humans judging variations in perfumes.

24.3

Current Challenges and Limitations of Electronic Noses

Humans are highly sensitive and selective sensors of perfume components. For ex-ample, odor detection thresholds (ODTs) are in the low- or sub-part-per-billion (ppb(volume/volume, v/v)) range for many compounds used as principal odorants in cur-rent perfumes. In addition, the human selectivity for certain odorous materials allowsperception of those odorants when in the presence of much higher concentrations ofother compounds. For example, 10 ppb (v/v) of a jasmonate in the headspace of aproduct would deliver a fresh, floral fragrance to consumers. Humans would perceivethe jasmonate fragrance even in the presence of 1000-fold excess (10 ppm (v/v)) oflimonene (or orange terpenes), a widely used PRM.The jasmonate/limonene example is by no means a rare case or exception. Indeed,

key odorous compounds classified as principal odorants of a perfume often account fora small fraction of the product headspace composition. The reality is that humans haveno problem ignoring the bulk of the compounds in the headspace and sensing prin-cipal odorants in product. Perfumers take advantage of human selectivity towards the

24.3 Current Challenges and Limitations of Electronic Noses 563563

principal (and other) odorants to deliver desirable fragrances to products. Indeed, hu-man selectivity is of paramount importance to perfumery.Sensitivity and selectivity are also important to efforts aimed at developing an elec-

tronic nose for perfumery. State-of-the-art electronic noses employing a few sensors orsensing strategies, with selectivities vastly different from those of the human, are likelyto be limited to perfumes where the principal odorants are a major fraction of theheadspace composition.

24.4

Literature Review of Electronic Noses in Perfumery and Cosmetics

The use of electronic noses in cosmetics and perfumery appears limited compared toother areas, e.g. food, beverages, chemicals, polymers, and plastics. The limited use isreflected in the number of publications. We found about twenty publications speci-fically addressing cosmetics and perfumery, while the published reports in other areasreach into the hundreds.Interestingly, references including perfumery applications often include the devel-

opment of new chemical sensors designed to enhance sensitivity and selectivity. Forexample, Kusumoputro and Rivai of Indonesia University [4] used quartz resonatorcrystals with lipid membranes to discriminate fragrance odor. Quartz resonatorsare also known as quartz microbalances (QMB) or quartz crystal microbalances(QCM). Using those sensors and an artificial neural network, they achieved high re-cognition accuracy when determining the correct percentage of aroma from MarthaTilaar cosmetics products and five flavors from Splash Cologne products.Byfield et al. [5] also demonstrated the use of quartz crystal resonators in the fra-

grance and petrochemical industries, and in another case [6] demonstrated chiral dis-crimination with a QMB sensor. This development is especially important to perfum-ery were optical isomers may have clear differences in odor.Chiral recognition was achieved by coating the crystals with compounds such as

heptakis (2,3,6-tri-o-methyl)-beta-cyclodextrin, and octakis (6-o-methyl 2, 3-di-o-pen-tyl) gamma-cyclodextrin dissolved (as 50% and 20% (w/w) solutions) in OV1701,a widely used stationary phase in GC. The sensors showed preferential binding forenantiomers of a- and b-pinene and cis- and trans-pinane. By comparing to elutiontime in gas chromatography, the observed separation factor was seen to be dependentupon the chiral stationary phase concentration. The results suggest that on-line deter-mination of enantiomeric excess and concentration of certain monoterpenes is pos-sible at room temperature using QMB sensors coated with chiral GC stationaryphases.Cao et al. [7] and Yokoyama and Ebisawa [8] have published results related to the

development of QMB sensors for use in the fragrance and perfume industry. Bothgroups concluded that their sensors could correlate with sensory perception and dis-criminate between different fragrances.Alternative approaches to the use of QMB have also been reported. Hyung-Ki-Hong

et al. [9] developed an electronic nose with a micro gas-sensor array. The chemical


sensors were made using thin-film metal oxides. As with the work discussed above,good discrimination between samples was reported for both flavor and fragrances.Two fragrances, a women’s perfume (eau de cologne) and a man’s perfume (eaude toilette) were correctly identified.Penza et al. [10] classified food, beverages, and perfumes using an electronic nose

based on the use of a thin-film sensor array and pattern recognition. Using tungstenoxide (WO) with different catalysts, e.g. Pd, Au, Bi, and Sb, good selectivity and sen-sitivity were obtained to correctly classify the samples in question. The authors con-cluded the arrays show promise for use in a variety of industries and applications.Letant et al. [11] used porous silicon chips in an electronic nose designed to measure

a series of solvent vapors, ethyl esters, and perfumes. The chemical information fromthe porous silicon sensors was obtained by measuring changes in reflectivity andphotoluminescence. Good reversibility and reproducibility were obtained. Theyalso compared results with those obtained using metal-oxide sensors.Recently, a new technique to discriminate Yves Saint Laurent (YSL) perfumes by

means of an electronic nose was described by Carrasco et al. [12]. The authors ad-dressed an off-odor problem reported by an expert panel at Sanofi Beaute. ThreeYSL perfumes, Paris eau de toilette, Paris eau de toilette with an off-odor and Opiumeau de toilette were analyzed. The differences between samples were also apparent intheir GC profiles. However, to meet the needs of a perfume quality control laboratory,where the analysis would need to be faster than possible by GC, GC-mass spectrometry(GC-MS) and/or sensory analysis, an electronic nose was considered.The methodology included the use of Fox4000 Electronic nose (AlphaMOS, France)

and an autosampler. The system was equipped with 18 metal-oxide sensors. The onlysample preparation technique used was to allow alcohol evaporation before analysis,because the sensors are sensitive to alcohol. The procedure allowed 35 ll of eau detoilette samples, deposited onto a 2-cm2 paper strip placed inside a 10 ml headspacevial, to evaporate in air.The authors concluded that the electronic nose could correctly identify 100% of all

the samples in their respective perfume families, within 30 min, and without usingelaborate sample preparation techniques. They also recommended that the electronicnose be considered, along with classical techniques such as GC-MS or infrared spec-troscopy, as another useful tool for studying perfume volatiles.Feldoff et al. [13] studied the use of electronic noses with metal-oxide sensors and

MS-based sensors as tools for the discrimination of diesel fuels. No sample pre-paration other than the use of a static autosampler was necessary for both the chemi-cal and MS-based sensors. Good correlation was found between the samples, whichcorresponded to the origin of the fuel for both types of instrument. In this particularapplication, data obtained with the MS-based sensor was reported to be easier to ob-tain, and more reproducible, compared to data obtained by the use of chemical sen-sors.In summary, the literature review reveals a number of approaches, ranging from the

use of QMB, metal-oxide semiconductors and new sensor types, in conjunction withthe use of a number of pattern recognition and sample preparation methods, havebeen used in the perfume and cosmetic industry. In general, good correlations are

24.4 Literature Review of Electronic Noses in Perfumery and Cosmetics 565565

reported between analytical data, obtained by means of a number of sensor-basedstrategies – including MS, and the human response to the sample odor. The useof autosamplers simplifies the tasks within quality control laboratories, while helpingto achieve good reproducibility.Although some of the applications reported in the literature are based on the use of

experimental sensors or sensing strategies, it is clear that commercially available in-strumentation may offer a viable alternative to those sensors or strategies. In addition,commercial instrumentationmay offer a viable alternative, or be a powerful adjunct, toconventional GC and or GC-MS analysis, as illustrated by the work of Carrasco et al.[12] and Feldoff et al. [13].

24.5

Special Considerations for using Electronic Noses to Classify and Judge Quality ofPerfumes, PRMs, and Products

Today, the human nose is the ultimate judge of the quality of a perfume, PRM orproduct. This is true even after samples are examined by high-resolution, multi-di-mensional chromatographic tools, such as capillary GC/FID/MS (FID – flame ioniza-tion detector) or GC/MS/IRD (IRD-inhared detection). The primary reason for the useof humans as judges is that, as mentioned before, they have exquisite sensitivity andselectivity towards certain odorous compounds. Thus, a peak seemingly insignificantin a chromatogram, e.g. the pyrazine in Galbanum, may be the most important odor-ous compound in a perfume or PRM. As a consequence, it is not uncommon for aperfume or PRM to meet analytical specifications and fail sensory evaluation, or viceversa. Therefore, to successfully address odor issues in the perfume and cosmeticindustry it is essential to combine results from analytical and sensory measurements[14].Unfortunately, human judgment is subjective and somewhat variable. In addition,

for any given odorant, a fraction of the population would have ODTs significantlyhigher/lower (> 10–100 �) than the average population. Thus, to use humans asan analytical tool to judge perfumes one must go through a process designed to:

* select humans for their ability to smell,* teach how to scale intensity and name odorants,* calibrate people over time and correct for ‘drift’.

Such a process is often used to identify and train a number of judges who work in-dividually or as a group, i.e. as in an ‘expert panel’. As a consequence, developing andmaintaining expert judges and expert panels is an expensive, laborious, and time-in-tensive activity. In addition, human fatigue (adaptation) and habituation require spe-cial attention be given to testing protocols. Thus, even under the best of circumstances,it is possible to encounter artifacts that hinder the human capacity to judge odors. Forthose reasons, there is great interest in developing alternatives to the use of expertjudges or expert panels in perfumery.


Although we addressed the selectivity and sensitivity requirements above, we havenot addressed the instrumental analogs of human drift, fatigue, and habituation. First,we must define the terms. Panel drift is a change in panel judgment towards a given,standard stimulus presented over time. Detector drift is a change in output in theabsence of an input. However, in addition to detector drift, there is an instrumental‘classification drift’ similar to that experienced by expert panels. Human fatigue(adaptation) is a decrease in perceived odor intensity as a result of exposure to a con-stant odorant concentration. In addition to a decrease in perceived intensity, fatiguemay also produce changes in the perceived character of an odor when a complex mix-ture of odorants is used. Fatigue in an instrumental detector is a change (typically adecrease) in output when the device is exposed to a constant input. Detector fatigue is aunique function of detector design, sensitivity and selectivity. It may be the primaryfactor responsible for instrumental ‘classification drift’. Habituation in humans, as isalso true for fatigue, is a decrease in perceived intensity as the human brain growsaccustomed to a constant stimulus. Because it is strictly a consequence of how thehuman brain processes stimuli, it has no corresponding instrumental-sensor analog.Thus, for successful use of electronic noses in perfumery, the detectors must have

adequate sensitivity and selectivity, have minimum drift and fatigue, and the signal-processing package must address the problem of ‘classification drift’.To measure how well those requirements are met by available electronic noses,

analysts typically use a training set consisting of samples selected to encompassthe range of odors expected. The number of samples to be used depends on the abilityof the electronic nose to differentiate between extremes, e.g. best-worst, or most si-milar from most different odor. The following two case studies were selected to illus-trate the use of electronic noses and other classification tools to address perfume-re-lated questions.In both cases, special care was given to the sample-introduction phase of the mea-

surement. Autosamplers were used to ensure high reproducibility in generating head-space and introducing the sample into the different detectors. The second case studydescribes the use of the electronic nose within a production environment wherebyresults obtained were compared to the current quality methods being used.

24.6

Case Study 1: Use in Classification of PRMs with Different Odor Character but of SimilarComposition

24.6.1

The Problem

Because of their high selectivity, humans may perceive odorous compounds in thepresence of 103, 106 or even larger excess of other non-odorous compounds in air.In other words, the human response towards odorous compounds may exceed theresponse towards non-odorous compounds by many orders of magnitude. This isin contrast to two common analytical detectors, the FID and the MS in electron ioniza-

24.6 Case Study 1: Use in Classification of PRMs with Different Odor Character but of Similar Composition 567567

tion (EI) mode, which would have roughly the same response factor for odorous andnon-odorous compounds.Thus, we reasoned that electronic noses and other classification tools, utilizing de-

tectors having roughly comparable sensitivity towards organic compounds, wouldhave problems dealing with samples that have very different odors but have similarbulk chemical composition. An experiment to assess this perceived limitation wasdesigned by C. L. Eddy of The Procter and Gamble Co.For the experiment, eight PRM samples with distinct odor characters but similar

bulk composition were selected: bergamot, clementine, grapefruit, lime, lemon, man-darin, orange, and tangerine. Typically, the samples contained>85%D-limonene. Fortwo samples, orange and grapefruit, limonene together with myrcene, and a-, and b-pinene, accounted for 99 þ % and 96 þ % of the mass, respectively. Importantly, therelative abundance of those four compounds is virtually identical in the two samples.Therefore, we would expect that those two samples would be the most difficult todistinguish. Samples were analyzed by means of an HP4440 (Hewlett-Packard) che-mical sensor and by capillary GC-FID. Results obtained with the HP4440 were pro-vided by D. R.White Jr., and GC-FID data analysis was performed by K. D. Juhlin, bothof The Procter and Gamble Co.

24.6.2

Methods

The HP4440 is a device that combines a headspace analyzer and a bench-top MS. Toperform an analysis, the PRMheadspace was injected directly into theMSwhere it wassubjected to EI. A mass range was rapidly scanned, and ion currents at each m/z weresummed over the duration of the run time, e.g. 1 min. Data were analyzed by tools inthe Pirouette suite of chemometricmethods (Infometrix, Inc.). For the GC analysis, wechose to analyze the samples as neat oils, using a conventional HP-GC equipped withan autosampler for liquids. We justified this choice, as opposed to using headspace,because the PRM samples were similar in composition and volatility. Analysis timewas kept at ca. 15 min, although it could certainly be decreased if desired. A 30-m DB-1, 0.5-lm-thick, 0.32 mm id column and FID were used to separate and detect thecompounds.

24.6.3

Results

We compared results obtained by the two approaches. The HP4440 discriminated thePRMs, with the exception of some overlap of orange and grapefruit, as shown in thedendrogram in Fig. 24.1. Repeat analysis on Day 2 showed good reproducibility. BothSIMCA and K-nearest-neighbors (KNN) classificationmodels predicted Day 2 sampleswith 100% accuracy. Mass fragments, (m/z) in decreasing order of discriminationpower (DP, a ratio of between-class to within-class variances), are listed as m/z of


the ion followed by (DP), as follows: 43(4320), 90(2853), 154(2759), 150(2554),69(1727), 68(1696), 70(1677), 67(1629), 41(1067), 89(1003). Those ions are character-istic of terpene-like compounds, the most likely class of compounds responsible forthe odor of the PRMs.As expected, the most prominent ions in the raw data are due to limonene, as this

compound accounts for most of themass in the headspace. Therefore, the general lookof the raw data resembles the limonene mass spectrum. Limonene would have a massspectrum with prominent ions at (listed as m/z followed by relative abundance inparenthesis) 136(25), 121(23), 107(22), 94(27), 93(70), 92(22), 91(18), 79(31),68(100), 67(63). Therefore, those ions would not be expected to be among the listof ions with high discrimination power. Surprisingly, ions at m/z 67 and 68 appearto have high discrimination (DPs 1629 and 1696, respectively) probably because theirrelative abundances are a sensitive function of terpene structure.As expected, the highest discrimination power was exhibited by ions of low abun-

dance or absent from the limonenemass spectrum. Thus, two ions at m/z 150 and 154should not be present in limonene (MW136), while ions at m/z 90 and 43, if present,should be low abundance ions, i.e.� 1%. Consequently, it may be possible to enhancediscrimination between orange and grapefruit PRMs by selecting ions with the highestdiscrimination power for the analysis.GC-FID chromatograms of the orange and grapefruit PRMs are shown in Fig. 24.2.

Peaks labeled ‘A’ are virtually superimposable in the two samples. They correspond tolimonene (the largest peak in the chromatogram), myrcene, and a-, and b-pinene.However, a number of other peaks, labeled ‘B’, represent peaks distinctly differentin the two samples. Because each peak can be viewed as an independent measure-

Fig. 24.1 HCA cluster dendrogram of training set (Day 1) of eight

PRMs. Data autoscaled

24.6 Case Study 1: Use in Classification of PRMs with Different Odor Character but of Similar Composition 569569

ment of a compound in a given PRM, the discrimination of those two samples issimple. The classification results are shown in the dendrogram in Fig. 24.3. Having20–30 peaks, representing independent variables (because the peaks are a measure ofvirtually ‘pure’ compounds), probably over-defines this system. This is in contrast tothe use of the total ion current at a given m/z, which may depend on the presence ofinterfering, structurally related compounds in the sample.

24.6.4

Conclusions for Case Study 1

Two instrumental approaches, GC- and MS-based, were used to successfully classifyand differentiate odorous samples of similar chemical composition but different odorcharacter. Because the samples were chosen to challenge instrumental capabilities tomatch the odor recognition abilities of humans, we conclude that the future is indeedbright for instrumentally based approaches to evaluate and mimic the perception ofodors by humans.

24.7

Case Study 2: Use in Judging the Odor Quality of a Sunscreen Product

24.7.1

Background

Established practice in the industry requires the use of various analytical measure-ments to ensure the quality of every aspect of a perfumed product. On delivery of

Fig. 24.2 GC-FID chromatograms of samples of orange (red trace)

and grapefruit (blue trace) PRMs. Peaks labeled “A” are nearly identical

in the two samples and account for most of the mass under the peaks.

Peaks labeled “B” differ significantly between the two samples


Fig. 24.3 HCA cluster dendrogram based on all 185 time points

in the chromatograms of eight PRMs. A principal components

analysis (PCA) of the 185 points produced 12 factors and

explained 89% of the variance. The first two principal components

separate most of the PRMs, however, to separate orange from

grapefruit we needed to go to principal components

4 through 6


the product, specifications and measurements are attached as a written record. How-ever, asmentioned before, it is not uncommon for a perfume, PRM, or product tomeetanalytical specifications and fail sensory evaluation (or vice versa). Thus, sensory eva-luation is an important additional quality control task, most often performed by anexpert panel. Unfortunately, as discussed previously, development and maintenanceof expert panels are costly and time consuming. Thus, we wanted to establish the useof an electronic nose as a tool to qualify the sensory properties of a product.

24.7.2

The Problem

We wanted to establish if an objective and sensitive electronic nose could free expertpanels from tedious quality control activities, thus freeing them to judgemore difficultsamples. In our example, the best product would be judged to be ‘odorless’. The pro-duct samples would have already passed analytical tests prior to undergoing sensoryevaluation. We used a Fox4000 electronic nose with 18 chemical sensors for correla-tion with sensory evaluations. Expert panel evaluations were made on � 150 samplesjudged to fall in three categories:A: does not meet odor standard quality, but it is sufficiently good to be used as

‘diluent’ when adjusting bulk qualityB: good (BON) odor quality, meets sensory standardM: rejected quality (MAUvais)To demonstrate the ability of the electronic nose Fox4000 to function in both re-

search and development and production environments, two systems were evalu-ated. To function in both environments the electronic nose must be:

* As sensitive as the expert panel* Selective* Reproducible over time (short- and long-term, to allow the generation of databases)* Reproducible following sensor exchange or array replacement (to allow transferabil-

ity of databases)* Robust, and simple to use and maintain

The following experiments were carried out to evaluate the performance of the elec-tronic nose on a compound designed to serve as a sunscreen. The work was carried outover a six-month period in parallel with the standard quality control operating proce-dures at Givaudan Vernier. The initial work was carried out at the research facilitylocated at Givaudan Dubendorf.


24.7.3

Equipment and Methods

24.7.3.1 Equipment

Supplied and manufactured by Alpha MOS

2 Fox4000 electronic nose units (18 sensors), operated with zero-grade air.ACU500 humidifier, operated with HPLC-grade water.Fox4000 software. (Calibration methods)HS100 headspace autosampler.

Fox4000 EN

The system used for this study at Givaudan R&D Dubendorf was a Fox4000 electronicnose with three metal oxide sensor chambers (18 sensors). The equipment is shown inFigure 24.4. All the chambers had their temperature controlled at 55 � 0.1 8C. Thecarrier gas was synthetic air (P ¼ 5 psi) and humidity was controlled by anACU500 (RH ¼ 20%, T ¼ 36 8C) using pure water. The samples were injected tothe Fox by an autosampler from 10 ml sealed vials, the acquisition time and timebetween subsequent analyses were 120 s and 20 min, respectively, and the flowrate was kept at 300 mLmin�1. The second Fox4000 was used in a factory environ-ment at Givaudan Vernier, using the database developed at the R&D facility in Du-bendorf.

Specific parameters for oil injections:* Headspace generation time: 20 min at 100 8C.* Injection volume: 2500 lL.* Volume of sample: 2 mL.

Fig. 24.4 Alpha MOS Fox4000 electronic nose


Specific parameters for injection of standards:

* Headspace generation time: 2 min at 60 8C.* Injection volume: 100 lL.* Volume of sample: 1 mL.

Most of the standards used as calibration products were chosen from selected controlsensory samples that were used as odor standards. Selected samples were also used asreference compounds.

24.7.4

Results

24.7.4.1 Sensory Correlation and Long Term Repeatability

Analytical results are shown in Fig. 24.5. The PCA clustering of good (red) and rejected(blue) samples shows an excellent correlation with the expert panel judgments. Onlythree sensors (out of eighteen) were required to achieve those results demonstratingsufficient sensitivity and selectivity. Importantly, ten weeks later it was necessary toaddress calibration drift to interpret sample quality. This was achieved using a built-incalibration option available in the standard instrument software. This option allows theacquisition of data on new standard samples selected to track the drift and compensatefor it. As a consequence, good results were obtained over a six month period whencomparing electronic nose results with those obtained by standard sensory methods.

24.7.4.2 Database transfer from Dubendorf to Vernier

Database transfer from Dubendorf to Vernier was carried out with help from AlphaMOS Toulouse. At the present stage of development, the successful use of the software

Fig. 24.5 (a) The PCA-clustering of good (red) and rejected (blue)

samples shows an excellent correlation with the assessment of the

sensory expert panel. Only three sensors are necessary to achieve this

discrimination model. (b) The discriminant function model is capable

identifying all unknowns


used to address calibration drift required help from Alpha MOS. While good resultswere obtained, the present methodology is not plug-and-play. New developments, aim-ing at addressing this difficult problem, are in progress.

24.7.5

Conclusions for Case Study 2

The six-month evaluation of the electronic nose in quality control (Vernier facility), inparallel with standard sensory evaluations by an expert panel, demonstrated the abilityof the Fox4000 electronic nose to carry out sensory analyses. Over the study period, thesystem accurately classified ‘good’ and ‘bad’ batches of the tested product. Althoughthis was a remarkable result, further improvements would have to be made to justifyreplacing current practice. Some of the improvements include a reduction of capitalinvestment for the plant and a simpler software calibration option (i.e. a ‘plug-and-play’software) as well as a significant reduction in the required measurement time.Since this work was performed, a number of improvements have been made avail-

able by the manufacturer. The improvements include faster sample throughput(5 min), and a significant reduction in the level of expertise and labor required torun the instruments. Finally, there is a ’plug-and-play’ database transferability be-tween units.

24.8

Conclusions

The exquisite sensitivity and selectivity humans exhibit towards ‘key’ components ofperfumes presents a challenging problem when attempting to predict human percep-tion based on data derived from instrumental measurements. Ideally, to predict thehuman response to perfumes our instruments would need to approach the sensitivityand selectivity exhibited by humans. However, while state-of-the-art electronic nosesmay differ from humans in both selectivity and sensitivity, they can be trained to per-form the function of a highly skilled sensory panel.Furthermore, there may only be a few hundred ’key’ compounds we would need to

measure to obtain near-perfect correlations with the human response to virtually allperfumes. The number would drop to less than fifty within any given perfume family.Those key compounds, and their respective concentration, could bemeasured by high-resolution techniques such as GC-FID. Alternatively, markers of those compoundscould well serve the purpose. This could be done, without prior separation, by mon-itoring key ions with a mass spectrometer or by the use of quasi-selective sensors.General-purpose instruments and sensors would work in cases where the bulk gas-

phase composition is determined by key compounds or marker compounds. This isoften the case in samples expected to have little or no odor, such as bases for cosmeticsand raw materials used in the industry, e.g. plastics.A number of studies, reflecting the status of the field, are listed under references.


24.9

Future Directions

The sensitivity of analytical instruments has increased dramatically over time. Thus,analytical detection limits reported using electronic-based instruments have droppedby roughly 1000-fold per decade starting in the 1970s. In that decade, concentrationunits (or mass) reported in the literature, and instrumental specifications were inparts-per-million (or micrograms). The literature and specifications changed toparts-per-billion (or nanograms) in the 1980s, and parts-per-trillion (or picograms)in the 1990s. The increased sensitivity is, to a large extent, a consequence of develop-ments in the semiconductor and computer industries and their application to analy-tical chemistry. This trend is likely to continue in the foreseeable future.Importantly, those advances in sensitivity often translated into advances in selectiv-

ity. Today, the selectivity of state-of-the-art GC-MS instrumentation equipped withlarge-volume-injection systems can be used to identify and measure hundreds of per-fume compounds present in the gas-phase at, or above 1 ppb (v/v). This capability canbe used to identify ‘key’ compounds in perfumes and should facilitate the developmentof new, highly sensitive quasi-selective sensors.Advances in solid-state chemistry and ionization mechanisms, as well as advances

in microfabrication techniques are likely to produce large detector arrays with en-hanced sensitivity and selectivity. Those advances, coupled with the low power require-ments of small arrays should produce portable electronic noses with capabilities com-parable to those of humans.›

References

1 R. R. Calkin, J. S. Jellinek. ‘Perfumerypractice and principles’, Wiley & Sons, 1994.

2 A. Jinks, D. Laing. Perception 28: 395–4041999.

3 L. Buck, R. Axel. A novel multigene familymay encode odor recognition: a molecularbasis for odor recognition, Cell 65: 175 1991.

4 B. Kusumoputro, M. Rivai. ‘Discriminationof fragrance odor by arrayed quartz resona-tor and a neural network’. Proceedings ofInternational Conference on ComputationalIntelligence and Multimedia Applications(Eds. H. Selvaraj, B. Verma), Gippsland,Victoria, Australia, 1998, pp.264–269.

5 M. P. Byfield, L. Wunsche, C. R. Vuil-leumier. ‘Development and applications ofan electronic nose based on arrays of pie-zoelectric sensors’. Proceedings of the Se-venth Conference on Sensors and theirApplications. (Ed. A.T. Augousti) Institute ofPhysics Publishing, Bristol, UK, 1995,pp.52–57.

6 M. P. Byfield, M. Lindstrom, L. F. Wunsche.Chiral discrimination using a quartz crystalmicrobalance and comparison with gaschromatographic retention data, Chirality1997.

7 Z. Cao, H. G. Lin, B. F. Wang, D. Xu, R. Q.Yu. A perfume odor-sensing system usingan array of piezoelectric crystal sensors withplasticized PVC coatings, Fresenius Journal ofAnalytical Chemistry 355 (2): 194–199 1996.

8 K. Yokoyama, F. Ebisawa. Detection andevaluation of fragrances by human reactionsusing a chemical sensor based adsorbatedetection, Analytical Chemistry 65 (6): 673–677 1993.

9 Hyung-Ki-Hong, Hyun-Woo-Shin, Dong-Hyun-Yun, Seung-Ryeol-Kim, Chul-Han-Kwon, Kyuchung-Lee, T. Moriizumi-T.Electronic nose system with micro gas sen-sor array, Sensors and Actuators B (Chemical)36 (1–3): 338–341 1996.


10 M. Penza, G. Cassano, F. Tortorella, G.Zaccaria. Classification of food, beveragesand perfumes byWO thin-film sensors arrayand pattern recognition techniques, Sensorsand Actuators B (Chemical) 73 (1): 76–872001.

11 S. E. Letant, S. Content, Tze-Tsung-Tan, F.Zenhausern, M. J. Sailor. Integration ofporous silicon chips in an electronic artificialnose, Sensors and Actuators B (Chemical) 69(1–2): 193–198 2000.

12 A. Carrasco, C. Saby, P. Bernadet. Discri-mination of Yves Saint Laurent perfumes byan electronic nose, Flavour and FragranceJournal 13 (5): 335–348 1998.

13 R. Feldhoff, C. A. Saby, P. Bernadet. De-tection of perfumes in diesel fuels with se-miconductor and mass spectrometry-basedelectronic noses, Flavour and FragranceJournal 15 (4): 215–222 2000.

14 (a) N. Neuner-Jehle, F. Etzweiler. in ‘Per-fumes art, science and technology’, (Eds. P.M. Muller, D. Lamparsky), Elsevier, London,New York, 1991, p.153. Updated in: (b) H.Gygax, H. Koch, Chimia 55 (5): 401 2001.


Chapter 27: Automotive and Aerospace Applications

M. A. Ryan and Hanying Zhou Jet Propulsion Laboratory, California Institute of Technology

Pasadena CA 91109

INTRODUCTION

By the nature of their trainability to a broad range of compounds, electronic noses

are a good choice for air quality monitoring in an environment where the possible

contaminants are known. The trainability of an electronic nose, along with the ability to

select sensors for response to a suite of compounds has made this type of device

useful in several applications; in this chapter we will discuss its application to monitoring

the breathing air in an enclosed space for the presence of hazardous compounds. The

application of an electronic nose as an air quality monitor is as an event monitor, where

events of low concentration which do not present a hazard are not reported, but events

of concentration approaching a hazardous level are reported so remedial action can be

taken. The electronic nose used in these applications is not an analytical device which

analyzes the air for all compounds present, but neither is it an alarm which sounds at

the presence of any change in the atmosphere. The device described here was used

as an air quality monitor in an experiment aboard NASA’s Space Shuttle Flight STS-95,

and was designed to fill the gap between an alarm with no ability to distinguish among

compounds and an analytical instrument.

AUTOMOTIVE APPLICATIONS

Use of an electronic nose in the automotive industry is primarily conceptual today,

but there are several areas in which such a device can be used. These include

monitoring the exhaust for combustion efficiency, monitoring the cabin air for passenger

safety, and monitoring the engine compartment for other conditions such as leaking oil

or other fluids. Owing to offgassing of fabrics and materials (“new car smell”), to leaks

of coolant from the air-conditioning system, and intake of air from the roadway and the

engine compartment, the passenger cabin of an automobile can be significantly more

hazardous to human health than the outside air [Chan, Leung]. Improvement of the air

quality in an automobile cabin can be accomplished rather simply, but as cabins will

remain well-sealed for climate control and energy conservation, a need to monitor the

interior will remain. As environmental concerns spur development of more efficient

combustion, it will be useful to monitor the exhaust for combustion products as well.

Several automobile manufacturers have discussed the possibility of using an ENose in

a system in which the exhaust is monitored for the presence of compounds indicative of

incomplete combustion, and feedback to the engine will adjust engine settings to

improve combustion efficiency.

AEROSPACE APPLICATIONS

Electronic Noses have been proposed for many applications in aerospace; some

of those applications are realistic within the limits today‘s technology, and some will

require more development. In the area of space exploration, electronic noses have

been proposed for planetary atmospheric studies on landers. This application varies

from addition of an electronic nose to a rover to study the atmosphere as the rover

moves, to stationary devices which will study the variations in atmosphere over days or

seasons. In the search for evidence of life on other planets, electronic noses have

been proposed as desirable sensors because the sensing media in the array can be

selected to make it possible to distinguish isomers and enantiomers [Ryan and Lewis,

2001], and because the sensor array can be configured to span a broad range of

compounds. These applications require development of methods which will allow the

electronic nose to deconvolute target vapors from an unknown background; work to

develop devices with these capabilities is underway at the Jet Propulsion Laboratory

(JPL).

An immediate, and perhaps the most important, application is monitoring air

quality in human habitats. The ability to monitor the recycled breathing air in a closed

chamber is important to NASA for use in enclosed environments such as the crew

quarters in the Space Shuttle and the International Space Station (ISS). Today, air

quality in the Space Shuttle is generally determined anecdotally by crew members’

reports, and is determined after flight by collecting an end-of-mission sample and

analyzing it in an analytical laboratory using gas chromatography-mass spectrometry

(GC-MS). The availability of a miniature, low power instrument capable of identifying

contaminants in the breathing environment at part-per-million (ppm) and sub-ppm levels

would enhance the capability to monitor the quality of recycled air and thus to protect

crew health. Such an instrument is envisioned for use as an incident monitor, to notify

the crew of the presence of potentially dangerous substances from spills and leaks and

to provide early warning of heating in electrical components which could lead to a fire.

In addition to notification of events, it is necessary to have a reliable method by which

judgements on the use of breathing apparatus can be made; if the crew has put on

breathing apparatus while repairing a leak or cleaning a spill, it is necessary to know

whether it is safe to remove the apparatus. These needs have led to the development

of an electronic nose at JPL [Ryan et al., 1997, 1998, 2000], with ultimate application to

ISS intended and experiments on the Space Shuttle in the near-term.

The qualities required for an incident monitor to be used in spacecraft are that it

be capable of identifying and quantifying target compounds at determined levels in a

fairly wide range (see Table I), that it be a low mass and volume device which uses low

power, and that it require little crew time for maintenance, calibration and air analysis.

There are several possible sensing devices which could be used in the Space Shuttle

or ISS, but all have limitations in terms of the requirements. These devices include GC-

MS, Volatile Organic Carbon Analyzer, Flame Ionization Detectors, and Smoke Alarms.

Of these, only GC-MS discriminates among compounds; it also has the greatest

sensitivity. However, it generally requires crew time in sample preparation,

maintenance and calibration. An electronic nose does not, in general, have the

sensitivity of GC-MS; however, for most target compounds ppm and sub-ppm sensitivity

is required, but not the parts per trillion level found with GC-MS.

An electronic nose meets the requirements for an incident monitor. It can identify

and quantify compounds in its target set with a dynamic range of about 0.01 to 10,000

ppm, depending on the compound, it lends itself to miniaturization, and because it

measures deviation from a background it does not require frequent calibration and

maintenance.

The electronic nose developed at JPL was designed to detect a suite of

compounds and is suitable for use in the crew habitat of a spacecraft. The habitat is an

enclosed space where air is recycled and where it is unlikely that unknown and

unexpected vapors will be released into the air. It can be assumed that the air is clean

at the beginning of a period of enclosure, and it is deviations from that state that the

electronic nose will monitor; thus, it is not necessary to have detailed knowledge of the

constituents of the air to start. In addition, the contaminants which are likely to be

present and for which it is important to monitor are well known, the number of

compounds is not large (50 or so), and the probability of mixtures of 5 or more such

compounds appearing at one time is small. It is possible, then, to design and train a

device to monitor the air for deviation from a clean baseline and to analyze those

deviations for the appearance of a set of target compounds.

The air quality conditions in the crew quarters of a spacecraft are not radically

different from the conditions in an aircraft cabin, or in the passenger cabin of a bus or

automobile. In all those cases, it is reasonable to assume the air is clean at the

beginning of a monitoring period, and there is a set of contaminants of concern for

which to monitor. With such conditions in mind, the JPL electronic nose was designed

for a flight experiment where the crew habitat in the Space Shuttle was monitored

continuously for six days.

The JPL ENose is a low power, miniature device which, in its current

experimental design, has the capability to distinguish among, identify and quantify 10

common contaminants which may be present as a spill or leak in the recirculated

breathing air of the space shuttle or space station. It has as its basis an array of

conductometric chemical sensors made from polymer/carbon composite sensing films

developed at Caltech [Freund; Lonergan]. It is an array of 32 sensors, coated with 16

polymers/carbon composites. The polymers were selected by analyzing polymer

responses to the target compounds and selecting those which gave the most distinct

fingerprints for the target analytes. The JPL development model was used in a flight

experiment on the Space Shuttle flight STS-95 (October-November 1998) to determine

whether it could be used as a continuous air quality monitor. A block diagram and

photo of the JPL ENose are shown in Figure 27.1. The device used in the flight

experiment has a volume of 2000 mL and a mass of 1.4 kg including the HP200 LX

computer used for control and data acquisition, and uses 1.5 W average power. The

mass and volume were determined primarily by the spaceflight-qualified container

required for the device to be used in an experiment; the volume and mass can be

reduced by a factor of 4 with no modifications to the sensor head or the electronics and

minor modifications to the pneumatic system.

POLYMER COMPOSITE FILMS

The polymer-carbon composite films developed at Caltech are the sensing media

used in the JPL ENose [Freund; Lonergan; Severin; Albert]. These films are made from

insulating polymers loaded with a conductive medium such as carbon to make resistive

films. When a polymer film is exposed to a vapor, some of the vapor partitions into the

film and causes the film to swell; the degree of swelling is proportional to the change in

resistance in the film because the swelling decreases the number of connected

pathways of the conducting component of the composite material [Freund]. The

electrical resistance of each sensor is then and the response of each sensor in the

array is expressed as the change in resistance, dR.

Using commercially available organic insulating polymers as the basis for

conductometric sensing films allows ready incorporation of broad chemical diversity into

the sensing array. The sensors respond differently to different vapors, based on the

differences in such properties as polarizability, dipolarity, basicity or acidity, and

molecular size of the polymer and the vapor.

The polymer-carbon composite sensing films are sensitive to temperature and

pressure change as well as to change in the composition of the atmosphere. In a

measuring mode where the device is sniffing the atmosphere and comparing it to a

clean background with measurements of each a few minutes apart, temperature

changes are generally not significant. However, in the case of continuous monitoring

over several hours or days, both temperature and pressure changes will influence the

location of the baseline, and it is necessary to distinguish among temperature and/or

pressure change, slow buildup of compounds, and baseline drift. All of these issues

were addressed in the device developed at JPL. Neither changes in pressure nor

humidity which might be found in normal habitat have a significant effect on the

differential sensor response, but temperature changes greater than 4 – 8 oC influence

the magnitude of response across the sensing array as well as the fingerprint of

individual analytes. While it is possible to measure temperature, pressure and humidity

and to subtract any effect of changes in these conditions from the sensor response

data, the JPL ENose was built with the capability to control temperature, and pressure

and humidity were measured separately. Temperature was controlled on the sensor

substrates to stay constant at 28, 32 or 34 oC, both to eliminate apparent baseline drift

(film resistance changes) caused by temperature change and to aid the sensing

process. Temperatures around 30 oC will assist the process of desorption of analytes

from the films and will prevent hydrogen bonds from forming between analytes and the

polymers.

ENOSE OPERATION IN SPACECRAFT

While it is reasonable to assume clean air at the beginning of an enclosed period

in the Space Shuttle, there are two scenarios in which a clean air baseline must be

established. In one scenario, the ENose might be used to determine whether it is safe

to enter a chamber that has been enclosed for some time without crew use, such as a

module in ISS. In the other scenario, a background of clean air must be established to

determine whether there has been a slow buildup of a contaminant. This second

scenario, slow buildup of a compound, is among the likely scenarios for contamination

of the air. Contaminants may build up slowly as offgassing, slow leaks in vapor and

liquid containers, from inadequate air revitalization or filter breakthrough, and as human

metabolic products such as methane or carbon dioxide. In both of these scenarios, a

system by which a baseline of clean air can be established is necessary.

Contamination from offgassing may be considered of minor importance for

aircraft or automobile cabins because the air is exchanged frequently in the course of

use and fresh air can be brought inside during use, but in cabins where air is not

exchanged for several hours, the buildup can be considerable. Often the offgassed

molecules are small, such as formaldehyde, and are not well scrubbed in the air

revitalization system. In the Space Shuttle where air might not be exchanged for

several days or, more importantly in ISS, where the air is not exchanged, offgassing

becomes an important consideration. Flight qualification includes establishment that

the offgassing rate of components be below a set level, but there are as yet no data for

offgassing over periods of months to years, as will be found on ISS.

The JPL ENose pneumatic system includes a diaphragm pump which pulls

atmosphere over the sensors at 0.25 L/min and two filters, an activated charcoal filter

and a filter of inert material, before the sample chamber. The atmosphere to be

analyzed travels through a filter that is selected by a solenoid valve which switches

between the two. During usual monitoring intervals, the air travels through the “dummy”

filter made of inert material to provide a pressure drop equivalent to the pressure drop

across the charcoal filter. The charcoal filter cleans air without removing humidity, and

a baseline of cleaned air can be constructed and used to determine the degree of

baseline drift. The constructed baseline allows the analysis program to distinguish

between drift and slow change in atmosphere. Figure 27.2 shows how drift and slow

buildup can be distinguished after the charcoal filter is switched off; the sensor films

respond by rising rapidly and creating a “virtual peak,” and the sensor responses can

then be analyzed against the cleaned air background. The analysis of the responses of

the sensing array can then be used to determine whether the slow change in the

atmosphere is caused by contamination.

For the flight experiment, 6 days of continuous operation, the charcoal filter was

switched on for 20 minutes out of every 210 minutes. This frequency was sufficient to

determine the baseline in this application. If an electronic nose is to be used to

determine whether a chamber is safe to enter after a closed period, the cleaned air

baseline must be established for several minutes, and the virtual peak analyzed when

the charcoal filter is turned off. A schedule for filter changeout must be established; for

Space Shuttle air and no events, changing the filter every 20-30 days is sufficient. If

there has been an incident found by the filter, it should be changed after the cause of

the incident has been fixed.

In other applications, where the pressure and temperature are changing rapidly,

or where the composition of the atmosphere changes frequently, the filters can be

programmed to switch at different frequencies. In the passenger cabin of an aircraft, for

example, filtering can be frequent during the loading and taxi stages, when the

concentration of combustion products and of fuel can be high, and less frequent during

cruise.

The responses of the ENose were not influenced significantly by meals or

activities in the crew quarters because the device was placed under the air intake vent

for the entire cabin; odors were significantly diluted when they reached the sensors.

This condition was chosen in order to monitor the average concentration in the cabin

rather than localized concentrations.

THE JPL ENOSE FLIGHT EXPERIMENT

For the application of adverse event monitoring in the Space Shuttle, the JPL

ENose was trained to 12 compounds; 10 of these were compounds likely to leak or spill

and the other two were humidity change and vapor from a medical swab (2-propanol

and water), which was used daily to confirm that the device was operating. The ENose

was trained to identify and quantify the 10 contaminating compounds at the 1 hour

Spacecraft Maximum Allowable Concentration (SMAC) levels which are shown in the

upper section of Table I.

The 10 contaminants were drawn from a list of compounds of concern and for

which air samples are tested after a shuttle flight. In the second generation device now

under development, there will be 10-12 additional compounds. The sensitivity required

for the device was set at the 1 hour SMAC in the Flight Experiment and is set at the 24

hour SMAC for the second generation device. The upper section of Table I shows the

24 hour SMAC and the lowest level detected reliably by the first generation ENose at

JPL. The lower section of Table I shows a list of compounds considered for the second

set and their 24 hour SMACs. As an event monitor, it is not necessary to be

significantly more sensitive than the 24 hour SMAC level; when the concentration of a

contaminant approaches ~35% of the SMAC, measures can be taken to remove the

compound from the air and to take action on the source of the contamination. Further

training of the software is possible in situ, but for accurate identification and

quantification, the training must be done in an environment where it is possible to

deliver precise concentrations of the compound in the range of interest.

Table I Upper Section: Compounds targeted in the first generation ENose, with their 1-hour and 24-hour SMACs, and the lower level detected at JPL with that device.

Lower Section: compounds considered for the second generation ENose, with their 24 hour SMACs.

Compound SMAC 1hr SMAC 24 hr Detected at JPL (ppm) [**] (ppm) [**] (ppm) alcohols methanol 30 10 5 ethanol 2000 500 50 2-propanol 400 100 50 methane 5300 5300 3000 ammonia 30 20 20 benzene 10 3 10 formaldehyde 0.4 0.1 10 Freon 113 50 50 20 indole 1 0.3 0.03 toluene 16 16 15 acetaldehyde 6 acetone 270 acetonitrile 4 2-butanone 150 chlorobenzene 10 dichloromethane 35 furan 0.1 hexamethyltricyclosilane 25 hydrazine 0.3 methyl hydrazine 0.002 tetrahydrofuran 40 1,1,1-trichloroethane 11 o,p-xylenes 100

** source: Spacecraft Maximum Allowable Concentrations for Selected Airborne Contaminants; Space Physiology and Medicine

For all cases except formaldehyde, the ENose is able to detect the compound at

or below the 1 hour SMAC. The sensitivity limit for formaldehyde in the flight

experiment device is 10 ppm; by selection of a different polymer set with polymers more

likely to sorb formaldehyde, it should be possible to detect that compound below the 24-

hour SMAC level. The ENose is also able to deconvolute signals to identify and

quantify mixtures of two compounds with moderate success (about 60%). It is expected

that with further training and a more selective group of polymers, it will be possible to

detect lower concentrations of compounds and to deconvolute mixtures of three or four

compounds.

DATA ANALYSIS

The data analysis software development portion of the JPL ENose flight

experiment considered several different approaches. The primary constraint in software

development was the requirement that gas events of single or mixed gases from the 10

target compounds be identified correctly and quantified accurately. The co-investigator in

the flight experiment, Dr. John James of the Toxicology Branch at NASA-Johnson Space

Center (JSC), defined accurate quantification as +/- 50% of the known concentration

measured in the laboratory. This degree of error was defined based on the SMACs; the

toxic level of most of the compounds is not known more accurately than +/- 50%, so the

SMACs have been set at the lower end. For the flight experiment, constraints in

telemetry and communication prevented real-time analysis, and so the development

process did not include full capability for immediate resistance vs. time data analysis.

A series of software routines was developed using MATLAB (from MathWorks,

Inc.) as a programming tool. MATLAB is a flexible program, and thus appealing for

development of software, though it runs relatively slowly. For future use, where real-

time or quasi-real time analysis is called for, the routines can be translated into C and

run on a desk top or lap-top computer.

For sensing media such as the conducting polymer/carbon films used in this

program, relative response changes (in magnitude) have been found to be more reliable

than the response shapes, especially at the low gas concentration range targeted in this

program (1–100 ppm). Hence, the task of identifying and quantifying a gas event is

roughly a two-step procedure:

1) Data pre-processing, to extract the response pattern of a gas event from raw

time-series resistance data for subsequent analysis, and

2) Pattern recognition, to identify and quantify a gas event based on the response

pattern extracted.

Data Pre-processing

When presented with continuous monitoring data, a response pattern must be

extracted by use of software. This process of extracting a response pattern from raw

resistance data involves four sequential steps: 1) Noise removal, 2) Baseline drift

accommodation, 3) Gas event occurrence determination, and 4) Resistance change

calculation.

Noise removal Despite the best effort in choosing sensor films with the

consideration of low noise level, fluctuation in the sensors’ responses are still seen to

be quite large. Some polymer films were found to be noisier than others. The reasons

one polymer-carbon composite film might be noisier than another are not well

understood; noise may be attributed to high sensitivity of the polymer film to small

changes in pressure caused by air flow, to differences in the carbon dispersion in the

film, or to inhomogeneities in the thickness or even composition of the film itself. In

general, the fluctuation in resistance (or noise) is fast compared to the response to a

gas event. Therefore digital filtering may be used to filter out this high frequency

fluctuation. The length of the filter may be different for different sensors and can be

determined by analysis of the noise in each sensor.

Baseline drift accommodation Baseline drift is one of the most difficult

problems to be solved in extracting ENose resistance data from the time data. The

causes for baseline drift can be multiple, and include variations in temperature,

humidity, pressure, aging of the sensors, and sensor saturation. However, at present

there is no clear understanding of the underlying mechanism of each one of the causes,

which makes attempts to compensate drift very difficult. Nevertheless, the baseline drift

is generally slowly-varying in nature compared to the response time of a detectable gas

event. This difference in time scale enables us to use a long-length digital filter to

determine the approximate baseline drift and then subtract it from the raw data. The

result is further adjusted by piecewise fitting using the baseline information from the

clean air reference cycles described above. Although this approach will not

accommodate the drift fully, it will reduce the effect to a manageable degree. Figure

27.3 shows resistance data which has been processed. The dark, smooth trace in the

upper plot shows the baseline variation determined through the use of low frequecy

filters. The grey, noisy trace in the lower plot is the data after baseline variation has

been subtracted, and the dark line is the processed data, with baseline variation

subtracted and after filtering for noise accomodation.

Gas event occurrence determination Because data analysis in the flight

experiment of the JPL ENose was not real-time because of constraints unrelated to the

technology development, it was not necessary for the analysis to be automatic, but a

preliminary software routine for automated determination of whether and when a gas

event occurs was developed. It is based primarily on threshold calculation, in which the

resistance change over a certain time interval is calculated, and a time-stamp is

registered if the change exceeds a pre-set threshold. This routine can detect most gas

events; however, it was also found that it may identify noise, and sometimes baseline

drift, as gas events. For the flight experiment, events identified by the automated

routine were confirmed by visual inspection of the time domain data; future

development of the data analysis software will refine the identification method.

Resistance change calculation Since the sensors’ relative responsiveness to

a vapor determines the fingerprint of that gas, the response pattern, it is important to

preserve this relative responsiveness. This means any calculation method of the

resistance change should be taken at the same time stamp after the initial onset of a

gas. Both relative resistance change, R/Ro, and fractional resistance change, (R-Ro)/Ro

were tested, and the latter was adopted as it maximizes the difference between the

signatures of different gas compounds.

Pattern Recognition Method

Although many pattern analysis methods exist in the general field of electronic

nose and other array-based sensor data analysis [Bartlett, 1999; also see Chapter 8 of

Part B, and Part C], no single method appears to be readily applicable to the task of

identifying and quantifying single gases as well as mixtures of up to three of the 12

compounds (10 target compound plus water, humidity change and the propanol-wipe)

at levels about 1 –100 ppm. Most of the widely used methods have demonstrated their

effectiveness, but not to a combination of all three scenarios found here: a large

number of target compounds, some of which are of very similar chemical structure

(e.g., ethanol and methanol), low target concentrations, and both single gases and

mixtures.

METHOD(S) DEVELOPMENT

For reasons stated above, three parallel approaches to ENose data analysis

were used during the early stages of software development: Discriminant Function

Analysis (DFA), Neural Networks with Back Propagation (NNBP), and Linear Algebra

(LA). Principal Component Analysis (PCA) was initially used , but was later replaced by

DFA because DFA tends to do better at discriminating similar signatures that contain

subtle, but possibly crucial, gas-discriminatory information. DFA is also better in class

labeling than PCA.

NNBP, or more specifically, multilayer perceptron (MLP), was selected as an

approach because it has good generalization of functions to cases outside the training

set, is capable of finding a best-fit function (linear or nonlinear; no models needed), and

is also more suitable than DFA when the sensor signatures of two gases are not

separable by a hyperplane (e.g. one gas has a signature surrounding the signatures of

another gas). However, NNBP is inferior to DFA in classifying data sets which may

overlap.

The reason to use LA, which is not as commonly used as other methods, is that

neither DFA nor NNBP were found to be well suited to recognizing the sensor

signatures from combinations of more than one gas. This method tries to solve the

equation x=Ac, where vector x is an observation (a response pattern), vector c is the

cause for the observation (concentrations of a gas or combinations of gases), and

matrix A describes system characteristics (gas signatures obtained from training data,

or sensitivity coefficients). For ENose data analysis where the response pattern can be

noise corrupted, which means there may exist no exact solution, least squares fitting is

the preferred way to solve the equation [Stang; Lawson].

The idea of developing three parallel methods is that one can first use the LA

method to deconvolute an unknown response pattern as a linear combination of target

compounds; unknown compounds are expressed as a combination of up to four

compounds. If a single compound is found, additional verification can be then carried

out by NNBP and DFA methods for increased success rate and accuracy. However we

have found the LA method to perform consistently best among the three methods even

for single gases, while DFA was consistently the worst, which prompted us to discard

the use of the two verification methods of NNBP and DFA in the process.

Linear Algebra is suitable only if the training data are linear, which is not the case

for all sensors at the concentration ranges considered (see Table II) . For a nonlinear

scenario, it is then reasonable to use some Nonlinear Least Squares fitting methods

such as that of Levenberg and Marquart (LM-NLS). This is the one of the two new

methods which were investigated for non-linear analysis. The other method, a

Differential Evolution (DE) approach, was also investigated because it promises fast

optimization (the LM-NLS method can be rather slow). DE represents some recently

emerged so-called genetic algorithms [Storn; also see Chapter 15 of Part C]. It is a

parallel direct search optimization tool. It begins with an initial randomly-chosen

population of parameter vectors, adding random vector differentials to the best-so-far

solution in order to perturb it. A one-way crossover operation then replaces parameters

in the targeted population vector with some (or all) of the parameter values from this

“noisy” best-so-far vector. In essence it imitates the principles of genetics and natural

evolution by operating on a population of possible solutions using so-called genetic

operators, recombination, inversion, mutation and selection. Various paths to the

optimum solution are checked and information about them can be exchanged. The

concept is simple, the convergence is fast and the required human interface is minimal:

no more than three factors need be selected for a specific application. However the

last advantage is also its disadvantage: limited control for ENose data analysis. Finally,

the LM-NLS method was selected as the best tool for ENose data analysis.

Levenberg-Marquart Nonlinear Least Squares Method

For nonlinear models the technique of choice for least-squares fitting is the

iterative damped least-square method of Levenberg and Marquart (LM-NLS). Similar to

LA, LM-NLS tries to find the best-fit parameter vector c from an observation vector x,

which is related to c through a known linear or nonlinear function, x=f(A,c), where A is

system characteristics (sensitivity coefficients) obtained from training data. This

method usually begins from a given starting point of c, calculates the discrepancy of the

fit:

residual =(computed-observed)/�,

where � is the standard deviation, and updates with a better-fitted parameter c at each

step. LM-NLS automatically adjusts the parameter step to assure a reduction in the

residual: increase damping (reduce step) for a highly nonlinear problem, decrease

damping (increase step) for a linear problem. Because of this ability to adjust damping,

LM-NLS is adaptive to both linear and nonlinear problems. How this method adjusts

damping is discussed in detail elsewhere [Lampton].

In the course of this work, it was found that the response of the films to the target

compounds is linear with concentration only within a limited range. The nonlinearities in

the training data generated are of low order, but successful identification and

quantification of gas events must take the nonlinearities into account. To obtain sensor

characteristics without further knowledge of sensor nonlinearities, a second order

polynomial fit was used to model the nonlinearities. For each sensor response to each

gas, the program finds the best-fit sensitivity coefficients A1 and A2 (in the least-squares

sense) to the following equation:

resistance change = A1c + A2c2

where c is gas concentration vector. The fit is constrained to pass through the origin.

A1 and A2 are 13x32 matrices characterizing the sensors’ response to ten targeted

gases plus water, humidity change, and the propanol wipe.

Several modifications were made to the standard LM-NLS method to suit the

ENose data analysis problem. First, sets of starting points of vector c were used

instead of a single set of starting points of vector c. The purpose of doing this is to

avoid a local residual minimum, which is common in many optimization algorithms,

including the LM-NLS method. These initial sets of vector c can be randomly assigned

from within each element’s allowed range. The total number of initial sets will be

determined by the speed desired and the complexity of local minimum problem. In our

case, about 200 initial sets were found (~15N, where N=13 is the number of target

compounds) to be a good compromise.

Second, instead of always updating c for a smaller residual, we modify the

update strategy to favor a smaller number of gases within certain ambiguity ranges of

the residual. The reason is that signature patterns for a given gas compound

generated by the ENose sensors have been observed to have large variations. The

simple updating strategy tends to minimize the residual with a more-than-reasonable-

large-number combination of gas when the residual is simply the variation in recorded

response pattern itself and should be ignored. The amount of the final residual is an

indicator of how large the fitting error is and the confidence level of the fitting.

Finally, the sensors’ response pattern was weighted to maximize the difference

between similar signatures. As seen in Figure 27.4, which shows representative

signatures of the ten target gas compounds plus the medical wipe at a median

concentration level (because of the nonlinearity, there is no single signature for one gas

at all concentrations), it is clear that ethanol and methanol have very similar signature

patterns. Regression analysis also pointed out linear dependency to certain degrees.

This means that the signature pattern of one gas could be expressed as a linear

combination of the response pattern generated by some other target gases. To reduce

this similarity, the sensors’ raw resistance responses must be modified by different

weights in the data analysis procedure.

Single gases

For lab-controlled gas events, the overall success rate reaches ~85% for targeted

singles. Broken down into individual singles, the success are listed below in Table II.

The concentration ranges used in the training sets for each single gas are also given.

Table II Identification and quantification success rates for single gases. The ranges shown here are ranges used in LMNLS analysis.

Compound Concentration Success Range (ppm) Rate (%) Ammonia 10 – 50 100 Benzene 20 – 150 88 Ethanol 10 – 130 87 Freon 113 50 – 525 80 Formaldehyde 50 – 510 100 Indole .006 – 0.06 80 Methane 3000 – 7000 75 Methanol 10 – 300 65 Propanol 75 – 180 80 Toluene 30 – 60 50 %Relative Humidity 5 – 65 100 Medical Wipe 500 – 4000 100

Considering that the raw data are often very noisy at low concentrations, nonlinear

at high concentrations, highly correlated in some cases, and weakly additive in some

mixtures, these results demonstrate that the LM-NLS method is an effective technique for

analysis of an array of sensors. Future work on the ENose will attempt to remove many

of the impediments to data analysis, with focus on noise and correlation. Correlation will

be addressed in polymer film selection.

The ability of the data analysis software to identify and quantify single and

multiple gas events in clean air was tested in the laboratory. The targeted

concentrations range for quantification was 30% to 300% of the one hour SMAC for

each compound. As can be seen from Table II, in some cases it was possible to

identify and quantify substantially below the 30% SMAC concentration; however, in a

few cases quantification was successful only as low as 100% of the one-hour SMAC.

In one case, formaldehyde, we were unable to identify and quantify reliably below

several times the one-hour SMAC. Figures 27.5 and 27.6 show some results of single

gas identification and quantification graphically.

Mixed gases

Deconvolution for identification and quantification of mixtures relies on the

additivity of the sensor responses. Here, additivity means that the strength of the

response to a mixture of gas 1 at level c1 and gas 2 at level c2 equals the response of

the single gas 1 at level c1 plus the response of the single gas 2 at level c2.

Identification and quantification of mixtures in clean air was moderately

successful. Additive linearity holds for some combinations in concentration ranges near

the SMAC level of the lower SMAC-compound. The success rate for double gases

(about 60%) was less than that of single gases, as would be expected. An exhaustive

set of gas pairs was not run because of time constraints; only a selected group of

mixture pairs were run to test the additivity. For this relatively small pool of data,

additivity holds for the following gas combinations:

methanol + toluene ammonia + benzene ethanol + formaldehyde

methanol + benzene ammonia + ethanol propanol + benzene.

Although data obtained on some other combinations of gas compounds, e.g.,

{benzene + formaldehyde} and {methanol + propanol}, did not validate their additivity in

these tests, this does not necessary mean the additivity does not hold for those gas

combinations. In fact, in many of those gas combination tests, often one of the gases

was run at a very low concentration and its response was overwhelmed by the other

gas’s strong response. In other words, the detectable concentration of a gas might be

higher if there exist other highly responsive gases.

STS-95 Flight Data Analysis Results

The Resistance vs. Time data that were returned from STS-95 showed that there

were several gas events in addition to the daily marker. The daily marker, exposure to a

propanol and water medical wipe, was added to the experiment so that operation of the

device over the entire period could be confirmed. The initial analysis selected the daily

markers and identified them as 2-propanol plus a humidity change. These identifications

were confirmed by comparison of crew log times with the time of the event in the data.

While the hope in an experiment such as this one is that there will be several events

which test the ability of the device, such events would certainly be anomalous events in

the space shuttle environment. Software analysis identifies all events which were not

propanol wipe events as humidity changes. Most of those changes can be well

correlated with the humidity changes recorded by the independent humidity

measurements provided to JPL by JSC. The events are not completely correlated in

time because the humidity sensor was located on the stairway between the mid-deck

and the flight deck, and the ENose was located in the mid-deck locker area near the air

revitalization system intake. Those events identified as humidity changes but not

correlated with cabin humidity change are likely to be caused by local humidity

changes; that is, changes in humidity near the ENose which were not sufficient to cause

a measurable change in cabin humidity.

Figure 27.7 shows the correlation of cabin humidity with ENose response in

several cases. There are visible dips in the traces at times 19:00, 20:52, and 0:07 CST,

Nov 2 -3, 1998. These dips are the changes in air composition, and thus resistance,

during the baselining cycle, when air is directed through the charcoal filter. Piecewise

baseline fitting is based on the resistance during the baselining cycle.

Software analysis of the flight data did not identify any other target compounds, as

single gases or as mixtures. The independent analysis of collected air samples, in which

the samples were analyzed at JSC by GC-MS, confirmed that no target compounds were

found in the daily It is not surprising that the only changes the ENose saw were humidity

changes, and it is because events were not expected that the experiment included the

relatively uncontrolled daily marker events. air samples in concentrations above the

ENose detection threshold. There were no compounds that the ENose would have

indicated as unidentified events present in the air samples.

FUTURE DIRECTIONS

Sensors

The number of sensors in the Second Generation ENose will remain at 32. The

number of polymers may be expanded beyond 16 in order to make sub-groups of

polymers which have been selected for response to particular classes of compounds

within the set of 32 sensors.

To determine the set and sub-groups of polymers for the set of some 20 target

compounds, a model of polymer-analyte interaction is under development. This model

takes account of such parameters of equilibrium constant of solvation of the analyte in

the film, analyte diffusion in the film, and the effect of the conductive medium. The

model will be used to select polymer suites with maximum separation in patterns for

particular analyte suites. This type of selection may result in using some subset of the

32 sensors for various patterns.

It is possible that the use of carbon as the conductive medium is responsible for

the non-linearity of responses at low concentrations. Studies of the use of metals such

as gold or oxides of transition metals as the conductive medium is underway. It has

been found that alcohols and ketones desorb from metals more rapidly than they do

from carbon.

Data Acquisition

Current research in data acquisition is investigating the use of frequency

dependent methods for data acquisition. AC methods are generally more sensitive

than DC methods of measurements; AC methods may allow the use of thinner, higher

resistance films, thus increasing film sensitivity. Some sensors exhibit high frequency

noise which may be caused by local heating while resistance is measured, by

inhomogeneously distributed carbon, or variable thickness in the film. Thinner sensors

could eliminate some sources of noise, and AC measurements may filter out some of

the noise.

To test whether high frequency noise can be filtered by AC methods, a single

sensing film of polyethylene oxide/carbon was exposed to 2500 ppm methanol and the

impedance measured at several frequencies, including DC resistance. As shown in

Figure 27.8, there is substantially less baseline drift when sensor response is plotted as

dI/I0 where I is the impedance, than there is in the same sensor measured at DC, but

higher frequency noise is not diminished at the frequencies at which impedance was

measured. As would be expected, the magnitude of the response is not substantially

different when measured by AC or DC methods, as the film is equally thick and

probably equally sensitive. The decision whether to change over to using AC

measurement techniques will consider the efficiency of removing baseline drift through

digital filtering in the data analysis process vs. the electronic requirements for AC

measurements. It may be sufficient to measure DC resistance and remove the high

frequency noise by increasing the number of signal averages from 16 to 32 or 64 and

remove the low frequency noise by digital filtering in data processing, as described

above.

Data Analysis

Though the data analysis software developed for this ENose program was highly

successful for its application, several improvements can be made in the future. The

overall approach to data analysis will not be modified in the Second Generation device.

The major change will be the addition of real time or quasi-real time analysis. For the

flight experiment, data were stored and analyzed after the flight. For ground test

experiments in which events are manufactured to challenge the ENose, the goal is to

have data analyzed within minutes of detection. For faster data analysis, it will be

necessary to implement a reliable automated event identification routine and to

translate the identification and quantification routines from Matlab into C.

There will also be some adjustments to the identification and quantification

routines.

First, the current data analysis software uses all 32 sensors’ responses as input.

Though each sensor’s response was weighted in the analysis in order to maximize the

differences between similar signature patterns observed for different gas compounds, it

was not done systematically and therefore was not necessarily optimal. In the second

generation, the selection of the to-be-used sensor set and their corresponding weights

will be optimized by maximizing distances between gas signatures. The distance

between the signatures for gas m and gas n, dmn , is defined as

dN

R Rmn m i n ii

N

� ��1

� �, ,

where �Rm,i is the ith sensor’s normalized (fractional) resistance change for the mth gas

and the summation is over N sensors used.

Second, the core of our data analysis software is the modified LM-NLS method,

which is heavy with matrix operations and largely determines the entire data analysis

speed. Matrix operation speed is known to be exponentially slower as the matrix size

increases. One way to increase speed is to reduce the size of the matrix dynamically in

operation by incorporating sensors’ characteristic response information, such as known

negative or no responses to certain gas compounds.

This characteristic response information can also be used for compounds which

cannot be identified by the software; sensors which are known to respond or not to

respond to particular functional groups can be sampled for a match. Thus, while it may

not be possible to identify unexpected compounds, it will be possible to classify them by

functional group.

In the First Generation ENose, data analysis is performed on the steady state

signal produced by changes in the atmosphere. For air quality monitoring, using the

steady state signal is, in general, acceptable, as a transient will not remain in the

environment long enough to do harm. However, there are toxins which can be

hazardous as transients. With automated event determination, analysis can begin as

soon as the resistance measurement passes the pre-set threshold rather than waiting

for steady state to be reached. In addition, if desorption time is a function of conductive

medium, then it may be possible to use the kinetics of sensor film response for

identification and quantification. Several compounds, such as ammonia, can be

identified by the shape of the response curve upon visual inspection of the curve.

Quantification of the kinetics of response may enable identification of transients.

CONCLUSION

The results of the flight experiment were somewhat disappointing to the

experimenters, while satisfying to the crew. There were no anomalous events, and the

ENose was not challenged to identify compounds to which it had been trained.

Nevertheless, the experiment was successful. The ENose detected changes in

humidity and the presence of the daily marker, was able to identify and quantify the

changes, and was able to use the training set made in the laboratory to do the data

analysis. Further work in development of the JPL ENose will involve substantial

challenge to the device and to the analysis software, with blind testing, mixtures, and

unknowns which can be identified by functional group.

REFERENCES

K. J. Albert, N. S. Lewis, C. L. Schauer, G. A. Sotzing, S. E. Stitzel, T. P. Vaid and D. R.

Walt, “Cross-Reactive Chemical Sensor Arrays,” Chem. Rev., 2595-2626 (2000).

P. N. Bartlett and J.W. Gardner, Electronic Noses : Principles and Applications, Oxford

Univ Press, Oxford (1999).

M. G. Buehler and M.A. Ryan, “Temperature and Humidity Dependence of a Polymer-

Based Gas Sensor,” Proc. SPIE Conf. on Electro-Optical Tech. for Chemical

Detection, (1997).

C. C. Chan, H. Ozkaynak, J. D.Spengler and L. Sheldon, “Driver Exposure To Volatile

Organic-Compounds, CO, Ozone, and NO2 Under Different Driving Conditions,”

Environmental Science & Technology, 25, 964 (1991).

M. S. Freund and N. S. Lewis, “A Chemically Diverse Conducting Polymer-Based

“Electronic Nose”, Proc. National Academy of Science, 92, 2652, (1995).

J. T. James, T.F. Limero, H.J. Leano, et al., “Volatile Organic Contaminants Found in

the Habitable Environment of the Space-Shuttle: STS-26 TO STS-55,” Aviation,

Space Environ. Med., 65, 851 (1994).

M. Lampton, "Damping-Undamping Strategies for the Levenberg-Marquart Nonlinear

Least-Squares Method," Computers in Physics, 11, 110 (1997).

C. Lawson and R. Hanson, Solving Least Squares Problems, S.I.A.M. Press,

Philadephia, 1995.

P .L. Leung and R. M. Harrison, “Roadside and In-vehicle Concentrations of

Monoaromatic Hydrocarbons,” Atmospheric Environment, 33, 191 (1999).

M. C. Lonergan, E. J. Severin, B. J. Doleman, R. H. Grubbs and N. S. Lewis “Array-

Based Sensing Using Chemically Sensitive, Carbon Black-Polymer Resistors”,

Chem. Materials, 8, 2298 (1996).

M. A. Ryan, M.L. Homer, M.G. Buehler, K.S. Manatt, F. Zee, and J. Graf, “Monitoring

the Air Quality in a Closed Chamber Using an Electronic Nose,” Proceedings of the

27th International Conference on Environmental Systems, Society of Automotive

Engineers, 97-ES84 (1997).

M. A. Ryan, M. L. Homer, M. G. Buehler, K. S. Manatt, B. Lau, D. Karmon and S.

Jackson, “Monitoring Space Shuttle Air for Selected Contaminants Using an

Electronic Nose,” Proceedings of the 28th International Conference on Environmental

Systems, Society of Automotive Engineers, 981564 (1998).

M. A. Ryan, M. L. Homer, H. Zhou, K. S. Manatt, V. S. Ryan and S. P. Jackson,

“Operation of an Electronic Nose Aboard the Space Shuttle and Directions for

Research for a Second Generation Device,” Proceedings of the 30th International

Conference on Environmental Systems, Society of Automotive Engineers, 00ICES-

259 (2000).

M. A. Ryan and N. S. Lewis, “Low Power, Lightweight Vapor Sensing Using Arrays of

Conducting Polymer Composite Chemically-Sensitive Resistors,” Enantiomer, in

press

E. J. Severin, B. J. Doleman and N. S. Lewis, “An Investigation of the Linearity and

Response to Mixtures of Carbon Black-Insulating Organic Polymer Composite

Vapor Detectors”, Anal. Chem., 72, 658 (2000).

Spacecraft Maximum Allowable Concentrations for Selected Airborne Contaminants,

Vols. 1 & 2, National Academy Press, Washington, DC (1994).

Space Physiology and Medicine, A.E. Nicagossian, C.L. Hunton & S.L. Pool, eds., Lea

and Febiger, Philadelphia (1994).

G. Stang, Linear Algebra and its applications, 2nd ed, Academic press, New York,

1980.

R. Storn, “On the usage of differential evolution for function optimization,” Biennial

Conference of the North American Fuzzy Information Processing Society, NAFIPS,

IEEE, 519 (1996).

ACKNOWLEDGEMENTS

The research reported in this paper was carried out at the Jet Propulsion

Laboratory, California Institute of Technology under a contract with the National

Aeronautics and Space Administration, and was supported by NASA Code UL.

FIGURE CAPTIONS

Figure 27.1 The JPL Electronic Nose used in the flight experiment on STS-95 is shown

as a block diagram and as a photo. The developmental device occupies a volume

of 2000 mL and has a mass of 1.4 kg, including the HP 200 LX computer.

Figure 27.2: a) A virtual peak is created at time 21:08 when the air flow is switched

from the charcoal filter, which determines the clean air baseline, to the inert filter

which is used during normal measurements. The baseline drift can be determined

by fitting the trend of the clean air baseline; in this case the virtual peak can be

attributed to baseline drift.

b) A virtual peak which is not attributable to baseline drift can be analyzed for the

presence of hazardous materials.

Figure 27.3: a) Grey, noisy trace: raw resistance as recorded; dark line: baseline drift

determined by low frequency digital filtering.

b) Grey trace: resistance after baseline drift subtracted; dark line: Processed data,

resistance after noise accomodation by smoothing and high frequency filtering, and

baseline drift corrected.

Figure 27.4 Representative signatures of ten targeted gas compounds plus wipe

generated by ENose sensors. Notice the similarity between ethanol and methanol,

and the significant difference between benzene and toluene.

Figure 27.5 Identification and quantification of four single gases using LM-NLS. The

shaded area is the target +/- 50% detection range.

Figure 27.6 Identification and quantification of three single gases using LM-NLS

Figure 27.7 Sample data from STS-95 ENose Flight Experiment. Circles are the plot of

independent humidity measurements in the stairway from mid-deck to flight deck.

Polymer sensor responses: (A) Poly (2, 4, 6-tribromostyrene), (B) Polyamide resin (C)

Poly(ethylene oxide), (D) Poly(4-vinylphenol).

Figure 27.8 Response of a polymer/carbon film of polyethylene oxide to 2500 ppm of

methanol, at three frequencies of impedance measurement and DC resistance

measurement.

ENoseChamber

DCPower

switch

AIR IN

28 V in

carbon filterfor baselining

solenoid valve(choose filter)

8-bit Microcontroller

Data Acquisition Subystem

Computer HP200 LX

Pump(250 mL/min)

AIR OUT

teflon filterfor pressure equalization

Figure 27.1

Photo of ENose See jpg file

-0.002

-0.001

0.000

0.001

0.002

20:00 20:30 21:00 21:30 22:00

Time of Day (Nov 2, 1998)

delta

R /

R0

Poly(2, 4, 6-tribromostyrene)Poly(4-vinylphenol)Poly(ethylene oxide)Polyamide resin

Figure 27.2 a, b

-0.005

0.000

0.005

0.010

0.015

0.020

0.025

0.030

15:45 16:00 16:15 16:30 16:45 17:00 17:15 17:30

Time of Day (Feb 25, 1997)

delta

R /

R0

poly(ethylene oxide)poly(styrene -co-allyl alcohol)poly(ethylene-co-vinyl acetate)poly(styrene -co-maleic anhydride)

Figure 27.3

Figure 27.4

Concentration delivered (ppm)

Con

cent

ratio

n de

tect

ed (p

pm)

0

20

40

60

80

100

120

140

160

180

200

0 20 40 60 80 100 120 140 160 180 200

2-propanol methanol ethanol benzene

Figure 27.5

0 20 40 60 80 100Concentration delivered (ppm)

Con

cent

ratio

n de

tect

ed (p

pm)

0

20

40

60

80

100

indole (x 103)NH3

toluene

Figure 27.6

Figure 27.7

time (min)

dR/R

0

-0.03

-0.025

-0.02

-0.015

-0.01

-0.005

0

0.005

0.01

0.015

0 5 10 15 20 25 30 35 40

2 kHz, dt = 5 sec6 kHz, dt = 1 sec

1 kHz, dt = 5 sec

DC, dt = 1 sec

Figure 27.8

Documents

T.C. Pearce, S.S. Schiffman, H.T. Nagle, J.W. Gardner ...read.pudn.com/downloads120/ebook/510257/Handbook of machine … · 8.2.6 Scanning Light-Pulse Technique 191 8.3 The Tufts