Evaluation of analytical methodologies for fluoride

Evaluation of analytical methodologies for fluoride determination and speciation of

fluoro complexes of aluminium

By

JIHYANG NOH

DISSERTATION

submitted in fulfillment

of the requirements for the degree

MAGISTER SCIENTIA

IN

CHEMISTRY

IN THE

FACULTY OF SCIENCE

AT

UNIVERSITY OF JOHANNESBURG

SUPERVISOR: PROF. P.P. COETZEE

MAY 2005

i

Abstract

The regulations for water fluoridation of South African municipal waters up to the

optimum fluoride (F-) concentration of 0.7 mg/L, (Government Gazette, 2001) have been

legislated. Fluoridation processes need accurate analytical methodology for the

determination of F- , because F- has a narrow margin of safety between beneficial and

toxic levels. In this work the analytical chemistry of F- was investigated comprehensively

and guidelines compiled for the accurate determination of low level F- (between 0.05 to 1

mg/L) in aqueous systems.

The first part of this study focused on method validation and the evaluation of the ISE

and IC methods. The analytical methodologies were applied to the analysis of natural

waters such as river water (Vaal and Crocodile Rivers), dam water (Hartbeespoort Dam),

and drinking water (Johannesburg municipal tap water) to evaluate the performance of

the chosen methods in the analysis of real samples and to assess the effect of the sample

matrix on the accuracy of F- determinations. An inter-laboratory study in collaboration

with the South African Bureau of Standard (SABS) was carried out to evaluate the

proficiency of South African analytical laboratories and to check the proficiency of the

procedures developed in this study.

In the second part of this study, the development of an IC-ICP-OES and IC-ICP-MS

method was investigated for the speciation of fluoro-aluminium complexes. This work

was motivated by the fact that the water fluoridation could lead to remobilisation of scale

from municipal pipes. Scale may contain aluminium hydroxide or oxide precipitates that

can dissolve as fluoroaluminates or hydroxofluoro aluminates. The speciation of cationic

fluoro-aluminates, free Al3+, AlF2+ and, AlF2+ together with the neutral AlF3

0, was based

on cation exchange Ion chromatography (IC) coupled with Inductively Coupled Plasma-

Optical Emission Spectroscopy (ICP-OES) and Inductively Coupled Plasma Mass

spectroscopy (ICP-MS).

ii

Acknowledgements

I would like to express my sincere gratitude to:

• My Lord God and My Savior Jesus Christ for His unchanging love, amazing

grace and giving me the opportunity to prepare this Thesis

• My supervisor, Professor Paul Coetzee, for his insight and guidance as well as his

financial support

• Dr. Fischer and Herman for their help when I used the ICP-OES and ICP-MS

instruments

• My beloved parents and brother for all their love and trusting in me

• My family in Jesus Christ, Missionary Andrew, Missionary Mercy, Sarah, John &

Paul, for their love and prayer support

• My husband, Jacob Kim, for his love, prayer and encouraging me to finalize this

Thesis

• My lovely UBF Church members for their love

• Lovely lady, Nishya, for helping me to correct my English

• My lab fellows:

Cynthia, Simon and Mingsong for their encouragement

Raymond, Jeanine, Tarryn, Nick and Gert for making the office worth sharing

To the on l y w i se Go d b e g l o ry fo re v e r

th ro ug h Jesus Chr i s t ! Ame n –Romans 16 :27 -

iii

Contents

Abstract i

Acknowledgements ii

Contents iii

List of abbreviations x

List of figures xii

List of tables xiv

Chapter 1: Introduction 1

1.1 Fluoridation in South Africa 1

1.2 Importance of accurate F- determination 2

1.3 The current status of F- determination In South Africa 2

1.4 Objectives of the study 3

Chapter 2: Literature review of fluoride (F-) determination 5 2.1 Introduction 5

2.2 Analytical methods for F- analysis in water matrices 5

2.2.1 Determination of F- by chromatographic analysis 5

a. Ion chromatography (IC) 5

i) Conductivity detection 6

ii) Spectrophotometric detection 11

iii) ICP-MS detection 12

b. Reversed-Phase High Performance Liquid Chromatography 13

c. Chromatographic F – determination 15

2.2.2 Determination of F- by spectrophotometric analysis 16

2.2.3 Determination of F- by micro fluidic analysis (FIA, SIA) 18

a. Introduction 18

b. Flow Injection Analysis (FIA) 19

ivc. Sequential Injection Analysis (SIA) 23

2.2.4 Determination of F- by potentiometric analysis 24

a. Direct potentiometric method 24

b. Extended potentiometric method 25

2.2.5 Other methods 26

a. Pervaporation 26

b. Extraction 27

2.2.6 Comparative studies 27

Chapter 3: Theory of analytical methods 28

3.1 Ion Selective Electrodes (ISE) 28

3.1.1 Introduction 28

3.1.2 Important concepts of ISE 29

a. Activity 29

b. Nernst equation 30

c. Range of linear response 31

d. Slope 31

e. Liquid junction potential 31

f. Hysteresis (electrode memory) 31

g. Response time 32

3.1.3 Components of ISE 32

a. Reference electrode 32

b. Sensing electrode (Indicator electrode) 34

3.1.4 Fluoride Ion Selective Electrode 34

a. Introduction 34

b. The mechanism of the F-ISE 35

c. Total Ionic Strength Adjustment Buffer (TISAB) for F-ISE 37

3.2 Ion Chromatography (IC) 38

3.2.1. Introduction 38

3.2.2. Instrumentation and process of IC 38

a. Flow schematic 38

b. Sample injection: Rheodyne injection valve 39

vc. Operation and maintenance 39

d. The separation columns 40

e. Ion echange equilibria 40

f. The chemical suppression 42

g. Conductivity detection 44

3.2.3 The important parameters in IC 46

a. Resolution 46

3.3 Inductively Coupled Plasma-Optical Emission Spectroscopy 48


3.3.2 Instrumentation for ICP-OES 48

a. Sample introduction 49

b.Inductively coupled plasma (ICP) torch 49

3.3.3 ICP process 52

3.3.4 Optimization of experimental condition 54

a. Element wavelength selection 54

b. Plasma observation height 55

c. Plasma power 55

d. Plasma gas flow rate 55

e. Auxiliary gas flow rate 55

f. Nebuliser pressure 56

3.4. Inductively Coupled Plasma Mass spectroscopy (ICP-MS) 56


3.4.2 Instrumentation and process of the ICP-MS 56

a. Mass discrimination 57

b. Detection system – counting Ions 59

Chapter 4: Evaluation of instrumental methods

for low-level F- determination in laboratory test samples 60 4.1 Objective 60

4.2 Ion Selective Electrode (ISE) method 61

4.2.1 Instrumentation 61

4.2.2 Standards and reagent solutions 61

via. Stock fluoride solution and F- standards 61

b. Total ionic strength adjustment buffers (TISAB) for F-ISE 61

4.2.3 Result and discussion 62

a. Electrode drift 62

b. Investigation of electrode response 63

c. TISAB and calibration study 69

i) High-level F- calibration with different TISAB 70

ii) Low-level F- calibration and sample measurement: TISAB III 72

iii) Low-level F- calibration and sample measurement:

Low Level TISAB 79

iv) Low level sample measurement (Tap water matrix) 80

d. Interference study 82

i) Single compound ion interference 82

ii) Single compound colloid interference 93

iii) Multi-compound interferences 93

e. Analytical parameters 95

i) Reproducibility and repeatability 95

ii) Control chart 95

iii) Method Detection Limit (MDL) 96

iv) Linear range and working range 97

4.3 Ion Chromatography (IC) method 97


4.3.2 Standards and reagent solutions 98

a. Standards 98

b. Eluent for IC 99

4.3.3 Results and discussion 99

a. Effect of base line drift 99

b. Calibration study 100

c. Interference study 105

d. Analytical parameters 107

i) Repeatability 107

viiii) Method Detection Limit (MDL) 107

iii) Linear range and working range 108

4.4 Comparison of two methods (ISE, IC) 108

4.5 Methodology recommendations 110

4.5.1 ISE 110

4.5.2 IC 110

4.6 Conclusion 111

Chapter 5: Comparison of IC and ISE

for the analysis of natural and drinking water 112

5.1 Introduction 112

5.2 Laboratory Fortified sample Matrix (LFM) 112

5.3 Experimental 113


a. ISE 113

b. IC 113

c. ICP-OES 113

d. ICP-MS 114

5.3.2 Reagents, standard solutions and sample preparation 114

5.3.3 Experimental procedure 115

a. ISE 115

b. IC 116

5.4 Results and discussion 116

5.5 Conclusion 121

Chapter 6: Inter-laboratory study

(SABS water-check proficiency testing programme) 123


6.1.1 Programme design 124

6.2 Major constituents in water: (Group 3) 124

6.2.1 Samples for the Group 3 programme 124

a. Sample preparation 124

viiib. Sample dispatch 126

6.2.2 Statistical evaluation of results 126

a. Introduction 126

b. Z-scores 126

6.3 Results and discussion of F- determination 127

6.3.1 F- data collection 127

6.3.2 Statistical summary 127

6.3.3 Comparison of analytical methods 129

6.3.4 Determination of composition of natural water samples 132

a. Experimental method 132

b. Results 132

6.3.5 Determination of F- in water check samples by ISE and IC 133

a. Results and discussion 133

6.4 Conclusion 136

Chapter 7 Speciation of fluoro-aluminum species

by chromatography coupled ICP-OES and ICP-MS 137


7.2 Literature review 139

7.3 Modelling of the Al-F-H system

and calculation of distribution curves for cationic fluoro-aluminates 142

7.3.1 Theoretical distribution curve 142

7.3.2 Calculation of the solubility of Al(OH)3 145

7.4 Cation chromatography–ICP-OES: experimental 147



7.4.3 Reagents and standard solutions 148


a. Optimisation of ICP-OES conditions 149

b. Optimisation of the chromatographic conditions 149

i) Al concentration and pH optimisation 149

ii) Eluent and column optimisation 150

ixiii) Flow rate optimisation 151

c. Experimental distribution curve 153

7.4.5 Conclusion 157

7.5 Anion chromatography – ICP-OES: experimental 157





a. Theoretical distribution curves 159

b. Optimisation of ICP-OES conditions 159

c. Optimisation of chromatographic condition 159


7.6 Cation chromatography – ICP-MS: experimental 160





a. Optimisation of ICP-MS conditions 162

b. Preparation of the container 162

c. Optimisation of the chromatographic conditions 163

i) Eluent optimization 163

d. Typical chromatogram 164

e. Pitfall of this method 165


Chapter 8 Conclusion 166

References 168

x

List of abbreviations

8 HQS: 8-hydroxyquinoline-5-sulphonate

AAS: Atomic Adsorption Spectrophotometry

AOAC: Association of Analytical Chemists

APHA: American Public Health Association

ASTM: American Society of Testing and Material

AWWA: American Water Works Association

CDTA: (trans-1,2-Cyclohexylendinitrilo)tetra-acetic acid

DHLL: Debye-Hückel Limiting Law

EDTA: Ethylenediamine tetra-acetic acid

HETP: Height Equivalent to a Theoretical Plate

IC: Ion Chromatography

ICP-OES: Inductively Coupled Plasma-Optical Emission Spectroscopy

IRZ: Initial radiation zone

ISE: Ion Selective Electrode

ISO: International Organization for Standardization

LFM: Laboratory Fortified Matrix

LLT: Low Level TISAB

MDL: Method Detection Limit

MNO: Methylnaphthlo Orange

NAZ: Normal analytical zone

NPDES: National Pollution Discharge Elimination System

NRC: United States National Academy of Science’s National Research Council

NV: Naphthol Violet

PHZ: Preheating zone

PS/DVB: Polystyrene/divinylbenzene

R: Resolution

RF: Radio frequency

xi

RP-HPLC: Reverse-Phase High Performance Liquid Chromatography

SABS: South African Bureau of Standards

SBCO: Semi-Bromocresol Orange

SMTB: Semi-Methylthymol Blue

SPANDS: sodium 2-(p-sulfophenylazo)-1,8-dihydroxynaphthalene-3, 6-disulfonate

SRS: Self-Regenerating Suppressor

TAC buffer: Tri-Ammonium citrate buffer

TISAB: Total Ion Strength Adjustment Buffer

UJ: University of Johannesburg

US EPA: United. States Environmental Protection Agency

USGS: United States Geological Survey

xii

List of figures Chapter 2 Figure 2.1 5-Br-PADAP 14

Chapter 3 Figure 3.1 Electrochemical measuring system of ISE 32

Figure 3.2 Construction of F- electrode 34

Figure 3.3 Crystalline membrane constructions 35

Figure 3.4 IC flow schematic 39

Figure 3.5 The structure of the anion exchange resins 41

Figure 3.6 The structure of a cation exchange resin 42

Figure 3.7 ASRS suppression mechanism 43

Figure 3.8 Parameters for assessing the chromatographic separations 46

Figure 3.9 ICP instrumentation 49

Figure 3.10 Inductively coupled plasma 50

Figure 3.11 Torch-excitation stages within the torch 51

Figure 3.12 ICP process 53

Figure 3.13 Main reaction of the processes in plasma 54

Figure 3.14 Schematic of an ICP-MS system 57

Figure 3.15 ICP-MS interface 58

Figure 3.16 Quadrupole mass analyzer 58

Figure 3.17 ICP-MS detector 59

Chapter 4 Figure 4.1 Electrode response: TISAB III at 0.1 mg/L F- 64

Figure 4.2 Electrode response: TISAB III at 0.1 mg/L F- 65

Figure 4.3 Electrode response: TISAB III at 0.02 mg/L and 0.1 mg/L F- 66

Figure 4.4 Electrode response: TISAB III at 0.02 mg/L and 0.1 mg/L F- 67

Figure 4.5 Comparison of calibration line in related to electrode cleaning 68

Figure 4.6 Comparison electrode responses with respect to electrode age 68

Figure 4.7 Potential with different types of TISAB 71

Figure 4.8 Multi-calibration graph using TISAB III 73

xiii

Figure 4.9 Comparison of calibration line in related to electrode cleaning 78

Figure 4.10 Calibration comparison with TISAB after cleaning 79

Figure 4.11 TISAB efficiency for F- determination in presence of Al 87

Figure 4.12 TISAB efficiency for F- determination in presence of Fe 87

Figure 4.13 F- % recovery verse Al concentration (TISAB III) 88

Figure 4.14 Acetate and formate interference 106

Figure 4.15 Correlation plot of ISE and IC data 109

Chapter 5

Figure 5.1 Method comparison of the LFM % Recovery in Crocodile River 117

Figure 5.2 Method comparison of the LFM % Recovery in Vaal River 117

Figure 5.3 Method comparison of the LFM % Recovery in Hartbeespoort Dam 117

Figure 5.4 Method comparison of the LFM % Recovery in Tap water 117

Figure 5.5 Typical anion chromatogram of the Hartbeespoort Dam water 118

Figure 5.6 Typical anion chromatogram of the Crocodile River water 119

Figure 5.7 Typical anion chromatogram of the Vaal River water 119

Figure 5.8 Typical anion chromatogram of Tap water 119

Figure 5.9 Distribution curves for Al-F-OH system 120

Figure 5.10 Distribution curves for Al-F-OH system 121

Chapter 6 Figure 6.1 Average absolute z –score 131

Chapter 7

Figure 7.1 Theoretical distribution curves for the species in the Al-F-H system

at pH 3 and 10 mg/L Al 145

Figure 7.2 The influence of flow rate on the chromatographic factors 152

Figure 7.3 Typical chromatogram of the Al-F speciation 153

Figure 7.4 Experimental distribution curves for Al-H-F system in 10mgL Al 156

Figure 7.5 Theoretical anionic distribution curves of Al(OH)mFn 3-m-n 159

Figure 7.6 The deposition place 163

Figure 7.7 Typical chromatogram of the Al-F speciation using ICP MS detect 165

xiv

List of tables Chapter 2 Table 2.1 Single operator precision and recovery using AS4A column

(US EPA method 300.0, 1993) 6

Table 2.2 Single operator precision and recovery using AS9-HC column

(US EPA method 300.1, 1997) 7

Table 2.3 A summary of analytical parameters for F- determination

using the IC method 9

Table 2.4 Recovery results obtained for F- spiked in environmental water 10

Table 2.5 A summary of performance data using spectrophotometric

and ICP-MS detectors 13

Table 2.6 Determination of F- in tap water (n=5) 14

Table 2.7 A summary of chromatographic condition for F- determination 15

Table 2.8 A summary of comparative study of F- analysis 27

Chapter 3 Table 3.1 Types of ion selective membrane electrode 34

Table 3.2 Limiting Equivalent Conductivities at 25 ° C 45

Chapter 4 Table 4.1 Technical specifications for the F-ISE 61

Table 4.2 Potential Drift (mV) at each F- concentration vs. time 63

Table 4.3 Potential with different TISAB 70

Table 4.4 Analytical parameter (slope and R2) 74

Table 4.5 Influence of calibration parameters 75

Table 4.6 Influence of calibration parameters 76

Table 4.7 F- determination in curvature region (< 0.2 mg/L, second run) 76

Table 4.8 F- determinations in curvature region (< 0.2 mg/L, third run) 77

Table 4.9 Electrode slope with regard to electrode cleaning 77

xv

Table 4.10 Sample measurement after cleaning the electrode 78

Table 4.11 Sample measurement with Low Level TISAB 80

Table 4.12 LFM measurement using two TISABs in tap water matrix 81

Table 4.13 Stock solutions for Interferences study 82

Table 4.14 Effects of interference on the determination of F- no TISAB III 83

Table 4.15 Ionic strength and activity coefficient for Al interference 84

Table 4.16 The log of stability constant for F- ligand 85

Table 4.17 Effects of interference on the determination of F-

with TISAB III (single compound interference) 86

Table 4.18 The log of stability constant for CDTA ligand 88

Table 4.19 F- % recovery in presence of low-level Al (TISAB III) 89

Table 4.20 The efficiency of TAC buffer on the F- determination 91

Table 4.21 The effectiveness of time allowance on F- determination

in presence of 1 mg/L and 10 mg/L Al at 1 mg/L F- 92

Table 4.22 Effects of colloid interference on the determination of F-

with TISAB III 93

Table 4.23 Effects of interference on the determination of F- with TISAB III

(more than 2 compounds interference-Synergy effect) 94

Table 4.24 The repeatability of ISE 95

Table 4.25 Control chart for 0.2 mg/L and 1 mg/L F- solution 96

Table 4.26 Masses of compounds used to prepare 1000 mg/L stock solution 98

Table 4.27 Base line drift in IC 99

Table 4.28 Calibration factor with different set of standards 100

Table 4.29 Optimisation of the injection volume 101

Table 4.30 Column efficiency on F- determination 101

Table 4.31 Influence of decimal factor on accuracy 102

Table 4.32 Evaluation of calibration type on accuracy (1st run) 103

Table 4.33 Evaluation of calibration type on the F- determination (2nd run) 103

Table 4.34 Optimisation of peak threshold 105

Table 4.35 Acetate and formate interference on F- determination 106

Table 4.36 The repeatability of IC 107

xvi

Table 4.37 Determination of MDL using 0.05 mg/L F- 108

Table 4.38 Comparison of F- determination by ISE and IC 109

Chapter 5

Table 5.1 ICP-OES operational conditions of multi-element determination 114

Table 5.2 % Recovery in the analysis of the LFM samples 116

Table 5.3 Summary of the composition in each matrix 118

Chapter 6

Table 6.1 SABS water-check proficiency testing programme design 124

Table 6.2 Sample contents 125

Table 6.3 Concentrated solutions of synthetic samples 125

Table 6.4 Interferences added to the synthetic samples 125

Table 6.5 Z-score criteria 126

Table 6.6 Statistical summary 127

Table 6.7 Composition of the synthetic samples 128

Table 6.8 Analytical methods for F- determination in this study 129

Table 6.9 Average Z-score and analytical method 129

Table 6.10 Average absolute Z-score with Method 131

Table 6.11 Laboratory performance 131

Table 6.12 ICP-OES operational conditions 132

Table 6.13 Composition of the natural samples 133

Table 6.14 Sample background measurement 133

Table 6.15 LFM % recovery 134

Chapter 7

Table 7.1 International water quality standards for Al in drinking water 137

Table 7.2 Equilibria and equilibration constant 143

xvii

Table 7.3 Solubility of Al(OH)3 at different pH value 146

Table 7.4 Solubility of Al(OH)3(s) in a F- solution at specified pH 146

Table 7.5 ICP-OES operating condition 149

Table 7.6 The influence of flow rate on the chromatographic factors 152

Table 7.7 Summary of chromatographic conditions 153

Table 7.8 Summary of the typical results: 10 mg/L Al at pH 3 155

Table 7.9 ICP-MS typical operating condition 162

Table 7.10 Summary of chromatographic conditions 163

Introduction ______________________________________________________________________________

1

Chapter 1

Introduction 1.1 Fluoridation in South Africa The Department of Health in South Africa has legislated regulations in respect of

compulsory fluoridation of potable water supplies in September 2001 (Government

Gazette, 2001). Water fluoridation is a form of mass medication or diet supplementation,

where fluoride (F-) is added to the public water supplies so that all water consumers are

treated, irrespective of individual age, state of health or needs (Pontius, 1991). F- was

declared an essential element in the recommended daily dietary allowance by the United

States National Academy of Science’s National Research Council (NRC) in 1980,

particularly because of its beneficial effects on reducing the incidence of dental caries

(Underwood, 1977; Murray et al., 1991; WHO, 2002). Tooth decay is one of the well-

known health problems in South Africa (SA). Van Wyk (1995) reported that caries

affected 90-93% of the SA population. Regulations for the fluoridation of SA potable

water supplies to an optimum concentration of not more than 0.7 mg F-/L in order to limit

the development of dental caries were published in Government Gazette of 8 September

2000 (Department of Health 2000). The addition of F- to drinking water up to this level is

thought to be beneficial for tooth development in children up to the age of about 8 years.

The possible negative effects of over-exposure to F- are dental and skeletal fluorosis

(WHO, 2002; Liu.1995; Chen et al., 1993,1996; Butler et al., 1985; Richards et. al.,

1967). Fluorosis is caused by high fluoride intake predominantly through drinking water

containing F- concentrations more than 1 mg/L. Increased F- levels lead to increased

levels of dental mottling (Dean, 1934; Pontius, 1993; Reeves, 1994), and changes in bone

structure, namely skeletal fluorosis (Bayley et al., 1990; Riggs 1991; Pak et al., 1994;

Rizzoli et al., 1995; Meunier et al., 1998; Alexandersen et al., 1999).

Introduction ______________________________________________________________________________

2

1.2 Importance of accurate F- determination

Compared with many other chemicals, F- has a relatively narrow range between intake

associated with beneficial effects and exposures causing negative effects (White, 1988).

As levels of fluoride are increased, the risk of dental fluorosis increases more rapidly than

the decrease in dental decay. Because of the small margin of safety between beneficial

and toxic levels of F-, the consequence of accidental overdosing will be very serious. This

makes the need for careful control and choosing of optimum levels of fluoride critical.

Fluoridation processes need accurate analytical methods for the determination of F-. A

systematic study of the determination of fluoride in SA source waters (natural waters and

processed waters) is therefore an urgent requirement before the fluoridation process can

be managed with confidence. A very important aspect is therefore to assess the accuracy

of current analytical procedures.

1.3 The current status of F- determination in South Africa

All work in this project would rely heavily on accurate and reliable analytical procedures.

Analysis of results for the determination of F- by SA laboratories over the last 30 years

gives a rather unconvincing picture of the reliability of F-determination. The following

facts came to light in a statistical analysis of results of F- determinations in the period

1970 to 2002 (Haarhoff, 2003).

• Huge differences were found in the results for F- determinations in laboratory

water and natural water samples. Results for natural water show significantly

lower reproducibility with a strong negative bias

• No systematic difference is evident between the methods used. It can therefore

be concluded that the main cause of discrepancies could be traced to matrix and

interference effects

A very good accuracy for F- measurement is a prerequisite for staying within the narrow

dosing tolerances allowed by the SA Regulations. The SA Fluoridation Manual seems to

Introduction ______________________________________________________________________________

3

recommend the use of the antiquated SPADNS method which is prone to interferences

more so than the modern methods such as Ion Chromatography (IC) and Ion Selective

Electrode (ISE). For work that is more accurate, this manual indeed recommends ISE. It

is, however, known that ISE has several limitations that could lead to erroneous results if

analytical procedures are not followed diligently. IC, which is the method of choice in the

USA fluoridation programme, is more reliable at lower F- levels, but is not mentioned as

an alternative in the Fluoridation Manual.

1.4 Objectives of the study This study was motivated by the necessity for accurate F- determination with regard to

water fluoridation. The objective of the study is to investigate the chemistry and

analytical chemistry of F-. This study focused on the pitfalls in analysing low levels of F-

in natural waters. Low levels of F- between 0.05 to 1 mg/L are important in the

fluoridation context.

The specific objectives of this study are as follows:

• Objective 1: Method validation for low-level F- (0.05-1 mg/L) determinations

(Chapter 4)

To achieve this objective the analytical methodology to ensure that accurate, precise

and reliable F- determinations using ISE and IC are obtained was to be investigated.

The following aspects are important and are discussed in Chapter 4.

1. Effect of electrode drift on accuracy

2. Effect of baseline drift in IC

3. Effect of different TISAB solutions (Total Ion Strength Adjustment Buffer)

4. Evaluation of calibration procedures for low level fluoride

5. Determination of analytical parameters: repeatability, reproducibility, LOD, range,

and method detection limit

6. Study of the effectiveness of interference prevention procedures.

Introduction ______________________________________________________________________________

4

• Objective 2: Evaluation of methods for the determination of F- in natural

water samples with different matrix composition (Chapter 5)

It is necessary to determine F- in a variety of water matrices to establish public

exposure to F-. An important objective is therefore the study of the analytical chemistry

of F- determination in various water matrices. The evaluation of methods for the

determination of F- in various matrices of natural water (tap water, river water and dam

water) using two methods (ISE and IC) are reported in Chapter 5.

• Objective 3: Inter-laboratory study (Chapter 6)

Inter-laboratory comparison of F- determinations from 66 laboratories in South Africa

on specially prepared samples in conjunction with SABS (The South African Bureau of

Standards) is presented in Chapter 6. The statistics of the results obtained in the

analysis of samples which contain major interferences (synthetic samples) and typical

South Africa source water (natural samples), are reported in detail.

• Objective 4: Speciation of fluoro-aluminium complexes (Chapter 7)

The fluoridation of water could lead to remobilisation of scale from municipal pipes.

This scale contains aluminium hydroxide or oxide precipitates that can dissolve as

floroaluminates or hydroxofluoro aluminates. Chapter 7 discusses the development of

an IC-ICP-OES and IC-ICP-MS method for the determination of cation fluoro-

aluminium species.

Literature review of Fluoride (F-) determination ______________________________________________________________________________

5

Chapter 2

Literature review of Fluoride (F-) determination 2.1 Introduction

Many methods have been reported for the determination of F- in water matrices. The aim

of this chapter is to critically review the analytical methodology for the determination of

F- in water matrices, which include drinking water, natural waters and water subjected to

water treatment processes.

2.2 Analytical methods for F- analysis in water matrices

Analytical methods for the F- analysis can be classified into the three categories:

chromatographic analysis, spectrophotometric analysis and electrochemical analysis.

2.2.1 Determination of F- by chromatographic analysis

Chromatography is a separation method that relies on differences in partitioning

behavior between a flowing mobile phase (eluent) and a stationary phase to separate the

components in a mixture. It is a convenient technique and allows for the method to be

hyphenated with various detection methods, such conductivity, spectrophotometry and

ICP-MS detection.

a. Ion Chromatography (IC)

In this technique F- is separated from other components on ion exchange columns. It is a

well-established, mature methodology and applicable to the determination of inorganic

anions, including F-, in environmental waters such as wastewater’ drinking, ground, and

surface waters (US EPA 300.0, 1993; US EPA 300.1, 1997). This technique

conventionally used a combination of analytical column and suppressor system to


6

decrease the conductivity of the eluent to enable highly sensitive conductivity detection

(US EPA 300.0, 1993; US EPA 300.1, 1997; Weiss, 1995; Hirayama et al., 1996;

Moskvin et al., 1998; Jackson et al., 2000; Jackson, 2001; Stefanović, 2001; Application

note 133,135,140 and 154). Other detection methods, such as spectrophotometric and

ICP-MS can also be used (Bayón et al., 1998; Barnett et al., 1993; Jones, 1992;

Oszwaldowski et al. 1998).

i) Conductivity detection

US EPA method 300.0 (1993) specifies the use of an IonPac AS4A anion-exchange

column with a carbonate-hydrogen carbonate eluent and suppressed conductivity

detection for the determination of inorganic anions, including F-, in environmental waters.

It documents that F- concentrations of < 1.5 mg/L are subject to interference from mg/L

levels of small organic acids, such as formate and acetate, when using the AS4A column.

The single operator precision and recovery is given in Table 2.1.

Table 2.1 Single operator precision and recovery using AS4A column

(US EPA method 300.0, 1993) Analytical Parameter

Reagent Water

Ground Water

Surface Water

Drinking Water

Domestic and industrial waste Water

Added F- (mg/L) 2.0 0.4 1.0 1.0 1.0 Recovery (%) 91 95 73 92 92

SD (mg/L) 0.05 0.07 0.05 0.06 0.06 Acceptable recovery range: 80-120 % (US EPA 300.0, 1993)

The performance of methods used for environmental analysis is typically validated

through single operator precision and recovery studies on spiked samples. The data in

Table 2.1 shows acceptable recoveries (i.e., 80-120 %) except for surface water, 73 %.

US EPA 300.1 (1997) was published as an update to Method 300.0 in 1997. This method

specified the use of a high capacity Ion Pac AS9-HC column and suppressed conductivity

detection for the determination of the inorganic anions in drinking water by IC. The

single operator precision and recovery is given in Table 2.2.


7

Table 2.2 Single operator precision and recovery using AS9-HC column (US EPA method 300.1, 1997)

Reagent water Ground water Surface water Drinking water

% Recovery 89.7 84.3 80.4 89.0 %RSD 1.18 0.85 0.56 0.46

* Amount added: 2 mg/L F-, Acceptable recovery range: 75-125 % (US EPA 300.1, 1997)

The data in Table 2.2 shows acceptable recoveries (i.e., 75-125 %) for all samples.

A number of Dionex application notes deal with F- determination in a water matrix

(Application note 133, 140, 135, 154). Application note 133 and 135 utilized both the

IonPac AS4A-SC and AS14 columns with suppressed conductivity, ASRA-ULTRA in

recycle mode. These systems provided suitable performance for the determination of F-

in drinking waters as outlined in US EPA Method 300.0. The AS4A-SC column is

recommended for the rapid analysis of anions in low-ionic strength, well-characterized

samples such as drinking, raw, and surface water. However, this column is not

recommended for the analysis of F- in more complex wastewater samples that may

contain small organic acids, since low levels of F- are subject to interference from mg/L

levels of small organic acids with the AS 4A column (US EPA Method 300.0, 1993).

The AS14 column provides improved F- resolution from the system void peak and

complete resolution of F- from formate and acetate. The improved selectivity, as well as

higher capacity, makes the AS 14 column a better choice for higher ionic strength water

samples and more complex matrices, such as domestic and industrial wastewaters.

The AS9-HC column system is introduced in application note 135. It has significantly

higher capacity than the AS4A-SC and AS14 columns, so total run times are longer and

peak response is somewhat reduced. However, this column is ideal for the determination

of inorganic anions in high ionic strength wastewater samples.

The use of the IonPac AS 14A anion exchange column with a new Atlas Electrolytic

Suppressor (AES) is described in the application note 140, for the routine high

throughput determination of common inorganic anions in drinking water matrices. The

Atlas suppressor lowers baseline noise, which lowers detection limits and therefore


8

improves the detection of trace-level ions. Good long-term performance was realized

using the AS14A with Atlas suppression at 0.8 mL/min.

Traditionally, these columns, AS4A-SC, AS14A and AS9-HC, are designed for use with

carbonate/bicarbonate eluent for determining inorganic anions, including F-, in

environmental samples. A hydroxide-selective column, AS18, which has not been as

widely used for routine analysis of F- due to the lack of appropriate selectivity and

difficulty in preparing contaminant-free hydroxide eluent, has been described for anion

determination. The difficulty in preparing hydroxide eluent is eliminated by using of

automated, electrolytic eluent generation. The AS18 provides improved retention for F-

from the column void volume, overall improved selectivity, and a significantly higher

capacity compared to the AS4A column.

Jackson et al. (2000) tested a modification of Method 300.0 for F- determination. The use

of IC with a hydroxide-selective IonPac AS17 column, automated eluent generation and

potassium hydroxide gradient, was investigated. The AS 17 column provides superior

retention of F- from the column void volume and improved resolution from small organic

acids, such as formate and acetate, compared to the AS 4A column. Jackson (2001)

shows a chromatogram of inorganic anions, of which F- concentration was at 1 mg/L,

obtained using the IonPac AS14A column. No quantification data of F- is provided. The

higher capacity AS14A column provides better overall peak resolution compared to the

AS4A, complete resolution of F- from acetate, and improved resolution of F- from the

void peak.

Stefanović et al. (2001) describes a non-suppressor ion chromatographic method with

conductometric detection for the simultaneous determination of anions, including F-. The

proposed method in this paper shows numerous advantages such as higher selectivity,

shorter analysis time, lower quantitation and detection limits. Moreover, no regeneration

step has to be included and no special equipment is needed. The separation can be

achieved on any liquid chromatograph equipped with a conductometric detector.


9

The performance of methods used for F- analysis is typically validated through single

operator precision and bias studies on spiked samples. The validity of the method is

established by determining the validation parameter, such as range, linearity, method

detection limit and precision, and the quantitative recovery, obtained from

environmental water matrix. A summary of validation parameters, given in these papers

are tabulated in Table 2.3 and Table 2.4.

Table 2.3 A summary of analytical parameters for F- determination using the IC method

System

aColumn

Range (mg/L)

Linearity(r2)

bMDL (µg/L)

cRt Precision (% RSD)

Area Precision (% RSD)

Reference

DX 120 IonPac AS4A-SC 0.1-100 0.9971 5.9 0.48 0.67

DX 500 IonPac AS 14 0.1-100 0.9980 3.5 0.23 0.17

Dionex dAN 133,135

DX 500 IonPac AS9-HC 0.1-100 0.9980 7.4 0.36 0.58 Dionex

AN135 ICS-2000

Reagent-Free IC

IonPac AS 18 Hydroxide-

selective 0.1-100 0.9991 2.3 0.13 0.27 Dionex

AN154

DX 500 IonPac AS17 Hydroxide-

selective 0.1-100 0.9996 2.9 0.21 0.18 Jackson et al.

2000

Metrohm 690 IC

Metrohm Anion Super

Sep 0.5-50 0.9998 4.0 - - Stefanović et al.,

2001 aColumn : conjunction with relevant guard column bMDL : Method detection limit, cRt : Retention time, dAN: Application Note

The Table 2.3 shows the linear concentration ranges investigated, linearity (r2), and

calculated Method Detection Limit (MDL), which was defined to be the concentration

below which the analytical method cannot reliably detect a response (USEPA, 1984).

The retention time and peak area precision (expressed as % RSD) were determined from

seven replicate injections of a quality control standard, 2mg/L F- solution, as described

in US EPA Method 300.0 (1993) and 300.1 (1997). These results demonstrate that the

hydroxide columns, such as IonPac AS17 and AS 18, which were operated with

electrolytically generated hydroxide eluent, improve the analytical parameters,

compared to conventional IC columns such as the IonPac AS4A, AS 14 and AS 9HC

that use carbonate/bicarbonate eluent.


10

Table 2.4 Recovery results obtained for F- spikeda in environmental water

Column DrinkingWater

Raw Water

SurfaceWater

Domestic Wastewater

Industrial Wastewater Reference

AS4A-SC 93.9 96.5 109.0 57.0 88.0 AS 14 91.5 85.1 101.0 90.8 90.1

Dionex Application Note 133,135

AS9-HC 94.8 93.3 108.3 106.8 103.6 Dionex Application Note 135

AS18 115.5 99.2 103.4 114.5 103.1 Dionex Application Note 154

AS 17 115.5 109.5 110.5 103.8 119.0 Jackson et al., 2000 F- spikeda at 1 mg/L F-

The data in Table 2.4 show acceptable recoveries (i.e., 75-125 %) for the majority of

water samples, except the domestic wastewater sample using the AS4A-SC column,

where recovery is 57 %. Because organic acids are likely to be present in domestic

wastewater samples, the lower recovery is presented although no interference was

observed when using the other column system (US EPA 300.0, 1993). That is why

AS4A-SC column is not recommended for the analysis of F- in more complex wastewater

samples that may contain small organic acids (Application note 133). Except for this case,

all the other systems are validated for F- determination in various water matrices.

Weiss et al. (1995) developed a pellicular anion-exchange column, AS12A column, for

the determination of inorganic anions including F-. The separation between F-, chlorite,

bromate, chloride and nitrite on AS12A is superior to that obtained with AS9-SC in

simulated drinking water. Even short-chain fatty acids such as acetic acid, which co-

elutes with F- on an AS9-SC column, can be partly separated from F- on an AS12A

column. Tap water analysis was conducted using this column, by varying the

carbonate/bicarbonate ratio in the eluent, 2.7 mM Na2CO3 and 0.3 mM NaHCO3 eluent

system and 10.5 mM Na2CO3 and 0.5 mM NaHCO3 eluent system. The resulting

chromatogram shows that F- can be separated well and the analysis speed can be

significantly increased with the latter eluent. Comparison of isocratic and gradient

separation of anions on this column using a borate eluent is also provided. The

chromatogram shows that borate gradients on the AS12A column are suitable for a

number of sample types.


11

Moskvin et al. (1998) suggested a procedure for preparing an alkaline eluent with low

concentration of F- impurity for the IC determination of weakly retained anions present in

trace amounts in high-purity water. Because of the weak eluent power of OH- ions, the

use of a hydroxide eluent is advantageous only for determining weakly retained ions.

However, hydroxide eluent is difficult to handle, primarily due to the absorption of

carbon dioxide and other acidic gases from air. An anion-exchange purification of the

eluent was designed for extending the analytical range of F- and considerably decreasing

the background electric conductivity of the eluent. The method showed a method

detection limit of 0.05 µg/L at a sample volume of 10 mL.

ii) Spectrophotometric detection

Spectrophotometric methods for the determination of F- are often based on the bleaching

of the colour of various dye complexes, which results in a decrease in their absorbance,

due to the stronger complexes formed between the metal and F- ions (Jones, 1992; Bayón

et al., 1998; Barnett et al., 1993).

Jones (1992) detailed the development of an IC method for the trace determination of F-

in water samples based on the spectrophotometric detection of the AlF2+ species in the

presence of excess Al, 5mg/L, utilizing the post-column formation of an Al complex with

8-hydroxyquinoline sulphonate (8-HQS). This method showed very high sensitivity and

good selectivity. The possible interferences were Zn2+ and Mg2+, overlapping the F- peak,

if present in relatively large amounts, though modification of the eluent and/or post-

column reaction pH can considerably reduce the interferences. The results of F- analysis

for the three samples by the IC and F-ISE methods are in good agreement, yielding

slightly lower level by IC, and certainly within the tolerance of ± 0.1 mg/L F-.

Barnett et al. (1993) presented some results to illustrate the analytical utility of the

zirconyl xylenol orange complex as a post-column reagent for F- detection. A detection

limit of 0.05 µg/mL was achieved with a linear calibration function up to 10 µg/mL. The


12

observed cationic interferences, such as Mg2+, Ca2+, Al3+ and Fe3+, could be avoided by

the inclusion of a small cation exchange column prior to the analytical column.

Bayón et al. (1999) employed spectrophotometric detection, utilizing a post-column

reaction of the species with 8-hydroxyquinoline-5-sulfonic acid in a micellar medium of

cetyltrimethylammonium bromide. The method for the determination of trace levels of F-

was based on the formation of the AlF2+ complex in excess of Al3+ and separation of the

two species formed (AlF2+ and Al3+) in an ion exchange column. The final determination

was accomplished by post column derivatisation with fluorimetric detection. It showed

good detection limits, but interferences from cations such as Fe3+, Mg2+ and Zn2+ required

the use of the longer CS2 ion exchange column and the addition of EDTA to the sample

solution to eliminate the interferences. However, the chromatogram, obtained for a real

water sample using this condition with photometric detection, yielded 30 minutes Al3+

retention time, while the ICP-MS method, which will be discussed in next paragraph,

showed less than 8 minutes retention time of Al3+. This meant that the photometric

detection was not adequate for F- analysis in the analytical point of view.

iii) ICP-MS detection

ICP techniques have not been used so far for direct F- determination because of the high

excitation and ionization potentials presented by F- which resulted in extremely poor

sensitivity.

Bayón et al. (1998) reported the indirect method for the determination of trace levels of F-

in drinking and seawater samples. It was based on the formation of AlF2+ complex in

excess of Al3+, which allowed complete formation of the AlF2+ complex, and separation

of the two species formed (AlF2+ and Al3+) in an ion exchange column (CG2), after

thermal treatment of the sample for 1h at 50 °C, followed by ICP-MS detection,

monitoring atomic Al at m/z 27. In aqueous acidic solution, Al ions are present as

[Al(H2O)63+], which can react with F- to form the AlF2+ complex. The optimized pH,

where the complex AlF2+ proved to be stable, was 2.6-3. Compared to the


13

spectrofluorimetric detection method, ICP-MS detection proved to be extremely sensitive,

with a detection limit of 0.1 ng/g of F-, and free from interferences form other cations and

anions in natural water samples. The results obtained for the F- analysis in water samples

using ICP-MS detection were in good agreement with the values found by F-ISE and the

recoveries of 100 ± 10% showed the applicability of the proposed methodology to

perform F- determinations at extremely low levels, at least two orders of magnitude more

sensitive than the ISE method.

Performance data, using IC method with Spectrophotometric detection and ICP-MS

detection, are summarized in Table 2.5.

Table 2.5 A summary of performance data using spectrophotometric and ICP-MS detectors

Reference and detection method Jones, 1992 Bayón et al. 1999 Barnett et al. 1993Analytical

Parameter Spectro -Photometry

Spectro -Photometry ICP-MS Spectro

-Photometry Post-column

Reagent 8-hydroxyquinoline –5-sulfonic acid

8-hydroxyquinoline –5-sulfonic acid - Zirconyl xylenol

orange Wave

Length 360 nm

And 512 nm 390 nm

And 500 nm - 550 nm

Detection Limit 0.2 µg/L 0.6 ng/mL 0.1 ng/g 0.05 µg/mL

1.2 % (n=6, 50 µg/L) Precision 8.9 % (n=6, 1 µg/L) 2.3 %

(n=5, 20 ng/mL) 4 % -

Linear Range 1-400 µg/L Up to 2000 ng/mL

(Al at 10 ug/g) Up to 100 ng/g (Al at 500 ng/g) Up to 10 µg/mL

Linearity 0.9994 0.9995 0.9993 0.9999

b. Reversed-Phase High Performance Liquid Chromatography (RP-HPLC)

The chromatographic technique, which utilizes a non-polar adsorbent surface and a polar

eluent, has been named Reversed-Phase HPLC (RP-HPLC). Retention is the result of the

interaction of the non-polar components of the solutes and the non-polar stationary phase.

Typical stationary phases are non-polar hydrocarbons, waxy liquids or bonded

hydrocarbons (such as C18, C8, C4, etc.) and the solvents are polar aqueous-organic

mixtures such as water-methanol or water-acetonitrile.


14

i) RP-HPLC separation with UV-VIS detector

Oszwaldowski et al. (1998) utilized the direct RP-HPLC determination of F-, using a UV-

VIS spectrophotometric detector, based on the ternary M-F- -(5-Br-PADAP) complexes

[M=ZrIV or HfIV and 5-Br-PADAP=2-(5-bromo-2-pyridylazo)-5-diethylaminophenol

(Figure 2.1)]. The aim of this paper was to find a ternary coloured system which could

form the basis for a sensitive, selective and rapid chromatographic method. The selected

metal ions, TiIV, VV, NiII, ZnII, ZrIV, NbV, MoVI, SbIII, HfIV, TaV, WVI and UVI, were

examined using spectrophotometry and RP-HPLC over a wide pH range, 2-8, and ZrIV or

HfIV , at pH 4.0 x± 0.3, were found as good ternary coloured systems since these metals

resulted in strong signals, which meant a good sensitivity.

Figure 2.1 5-Br-PADAP

NBr N

N NCH3

CH3HO

Chromatographic separation was performed with C18 end-capped columns, and the eluate

was detected spectrophotometrically at max 585nmλ = . The detection limit in the ZrIV –F- -

(5-Br-PADAP) and HfIV -F- -(5-Br-PADAP), using 20 µL loop, were 0.7 ng/mL and 0.8

ng/mL respectively. Using the proposed methods, F- was determined in tap water based

on the ternary zirconium system and a 20µL loop. The result is tabulated in Table 2.6.

Table 2.6 Determination of F- in tap water (n=5)

Sample F- (ng/mL) RSD Confidence limit Recovery Volume (mL) Added Found (%) (Probability level 0.95) (%)

1.0 - 14.2 1.8 14.2 ± 0.3 - 1.0 15.0 30.0 2.9 30.0 ± 1.1 105.6 1.0 30.0 44.0 1.2 44.0 ± 0.7 98.6

This result indicates that the determined F- content is far below the recommended level of

0.7-1.2mg/L (WHO 1984). The recoveries are within the reasonable recovery range (75-

125%) (USEPA. 1997) and the reproducibility is good.


15

c. Chromatographic F- determination

Up to so far, the chromatographic method with various detection techniques has been

reviewed. A summary of chromatographic conditions for F- determination is given in

Table 2.7.

Table 2.7 A summary of chromatographic condition for F- determination

Column Eluent Fra Iv

b(µL) Detection Reference

IonPac AG4A +AS4A-SC 1.8 mM Na2CO3+ 1.7mM NaHCO3

2.0 50 Conductivity US EPA 300.0, 1993

IonPac AG17+AS17 1-40 mM KOH gradient 2.0 25 Conductivity Jackson et al., 2000 IonPac

AG14A+AS14A 8.0 mM Na2CO3+ 1.0 mM NaHCO3

0.5 25 Conductivity Jackson, 2001

IonPac AG9HC+ AS9HC 9.0 mM Na2CO3 0.4 10 Conductivity US EPA 300.1, 1997

IonPac AG4A+ AS4A-SC

1.8 mM Na2CO3+ 1.7mM NaHCO3

2.0 50 Conductivity

IonPac AG14A+ AS14A

3.5 mM Na2CO3+ 1.0 mM NaHCO3

1.2 50 Conductivity

Dionex Application

Note 133, 135

IonPac AG9 HC+ AS9 HC 9.0 mM Na2CO3 1.0 50 Conductivity

Dionex Application

Note 135

IonPac AG 18+ AS 18

22-40 mM KOH Gradient 1.0 25 Conductivity

Dionex Application

Note 154 Metrohm Anion

Super Sep 1.5-3.5 mM Phtalic acid 1.0-

2.0 100 Conductivity Stefanović et al., 2001

2.7 mM Na2CO3+ 0.3 mM NaHCO3 10.5mM Na2CO3 +0.5mM NaHCO3

IonPac AG 12A+AS12A

20mM Na2B4O7 +18 mM NaOH

1.5 25 Conductivity Weiss et al., 1995

KRS-8P cation guard +KhISK-1 anion analytical column

1 mM NaOH 3.0 250-1000 Conductivity Moskvin et al., 1998

TSKgel IC-Anion-PW

5.0 mM acetylacetone[2,4-

pentanedione]-sodium hydroxide(pH 9)

1.0 100 Conductivity Hirayama et al., 1996

0.1 M K2SO4 (pH 3) 1.0 Spectro- PhotometryIonPac CG2

0.45 M HNO3 0.5100

ICP-MS Bayón et al., 1999

IonPac AS4 5 mM Na2B4O7 (pH 8.2) 1.0 100 Spectro-

Photometry Barnett et al., 1993

IonPac CG2 0.09 M K2SO4 (pH3) 1.0 100 Spectro- Photometry Jones, 1992

C18 end-capped column Acetonitrile-water (85+15 v/v, pH4) 1.0 20 Spectro-

Photometry Oszwaldowski et al., 1998

Fra: flow rate, mL/min, Iv

b: injection volume


16

2.2.2 Determination of F- by spectrophotometric analysis

Spectrophotometric methods for the determination of F- ion are either based on metal

displacement from a colored complex or the formation of a mixed-ligand complex,

Zr(IV)-F-Alizarin (Standard method, 1955; Crosby et al., 1968), Zr(IV)-F-SPANDS

(Bellack and Schouboe, 1958; Standard method 1975 and 1998; US EPA Method 13 A

and 340.1, 1971; Devine and Partington, 1975; Mac Leod and Crist, 1973; Crosby et al.,

1968), Th(IV)-F-Bromocresol orange (Khalifa and Hafez, 1998), Eriochrome cyanine R

(Crosby et al., 1968), Semi-Methylthymol Blue (SMTB), Semi-Bromocresol Orange

(SBCO), Methylnaphthlo Orange (MNO), Naphthol Violet (NV) (Yuchi et al., 1995).

This method is subject to much interference, most of which are supposed to be eliminated

by the preliminary distillation step (Standard method 1975; Standard method 1998; US

EPA Method 13 A and 340.1, 1971) or micro diffusion (Sara 1975).

Water samples were analyzed for F- by one of the several alizarin-zirconium methods,

either visually or by using a colorimeter (Standard Method, 1955). These methods had a

serious drawback in that the colour development is progressive, requiring accurate timing

to achieve consistent results. The waiting period was also a source of constant annoyance.

The distillation step was necessary to avoid interferences.

Bellack and Schouboe (1958) developed a simple and rapid SPANDS (2-(para-

sulfophenylazo)-1,8-dihydroxynaphthalene-3,6-disulfonate) method, in place of the usual

alizarin-zirconium photometric methods, which yielded an accuracy within 0.02 mg/L in

the F- concentration range of 0.00 to 1.40 mg/L. Compared to alizarin-zirconium methods,

it saved considerable time and increased precision. It was, however, still subject to many

interferences, so that the distillation step had to be applied to eliminate interferences. A

distillation step severely restricts the number of samples that can be analysed in a given

period, so that this distillation step is often omitted when large numbers of samples have

to be analysed (Harwood, 1969). The SPANDS method became the reference

spectrophotometric method for the F- analysis in water. (Bellack and Schouboe, 1958;

Standard method 1975 and 1998; US EPA Method 13 A and 340.1, 1971).


17

Leod and Crist (1973) stated that although the SPANDS method was generally regarded

as accurate and sensitive, it was extremely time consuming requiring a sulfuric acid or

perchloric acid distillation step to remove interferences. This paper describes the results

of a study to assess the equivalence of the ISE and SPANDS methods for the

determination of soluble F-. The data presented in this paper indicate that the

determination of F- by the ISE gives results which are essentially equivalent to the

SPANDS method and require much less analysis time. The conclusion of this paper is

that the ISE method offers a number of significant advantages over the SPANDS method

by virtue of its simplicity, high precision, and speed. Devine and Partington (1975) also

suggests that the F-ISE method following distillation be adopted as the method of choice.

They found that serious errors have been experienced in the analysis of wastewater

samples for F- concentration by the SPANDS spectrophotometric method. The source of

positive interference in this method was due primarily to sulfate ion carry-over during the

preliminary distillation step. The chemical analysis of F- in water is subject to many

errors, due to interference of constituents present in the water. A distillation step is

therefore often included in the procedure to remove these interferences.

Crosby et al. (1968) evaluated four of the most frequently used reagents, such as Alizarin

red S, Eriochrome cyanine R, SPANS and Alizarin complexone, for the F- determination

in water with respect to reproducibility, sensitivity, range, stability of colored products

and of reagents, specificity and effect of temperature. The Alizarin complexone

procedure had many advantages over the spectrophotometric method, and was

particularly suitable for samples containing only very small amounts of F-. The analytical

results of F- determination from the spectrophotometric method and those from the ISE

method are compared and the F-ISE surpasses all the spectrophotometric methods with

regard to speed, accuracy and convenience, and is recommended for the routine

monitoring of fluoridated water supplies.

Complexes of tetravalent metal ions, such as zirconium (IV), thoruim (IV) and hafnium

(IV), with chromogenic chelating reagents having only one methyliminodiacetated group


18

ortho to the OH group were examined for spectrophotometric determination of F- by

Yuchi et al. (1995). Several compounds [Semi-Methylthymol Blue (SMTB), Semi-

Bromocresol Orange (SBCO), Methylnaphthlo Orange (MNO), Naphthol Violet (NV)]

have been examined for the validity of the improvement in sensitivities of

spectrophotometric determination of F-. Zirconium (IV) and hafnium (IV) were superior

to thorium (IV) as a central metal ion. The reaction with F- in this study was described as

follows:

Zr2(OH)4(HL)2+2F- ↔ 2ZrLF +2OH- + 2H2O

Khalifa and Hafez (1998) investigated the ternary purple coloured complex formed

between Th4+, bromocresol orange (BCO) in acidic medium. A spectrophotometric

method, based upon the decrease in colour intensity of the Th-BCO complex on mixing it

with F- ion, is also described. The proposed method was convenient, rapid and sensitive

for F-, and could be used for the determination of F- in the 0.02-3.00 µg/mL ranges with

standard deviation ±0.9 %. 2.2.3 Determination of F- by micro fluidic analysis (FIA, SIA) a. Introduction

Microfluidics for automated sample processing can be based either on continuous flow,

such as Flow Injection Analysis (FIA), or on programmable flow, such as Sequential

Injection Analysis (SIA). FIA uses constant forward motion of the carrier to transport

sample from injector to detector. The reagents are added continuously to the carrier

stream near the injection zone and samples are injected into a carrier/reagent solution,

allowing the sample and reagent to mix, which transports the sample zone into a detector

while desired chemical reactions take place. The resulting reaction product forms a

concentration gradient corresponding to the concentration of the analyte throughout the

entire sample zone length. SIA, on the other hand, uses flow reversals, and flow

acceleration to mix sample with reagents, and stop flow to accommodate reaction time,

and to monitor the reaction rate based response. The advantage of FIA is transparency of


19

operation, while SIA is more versatile and that a reduced sample and reagent volume are

required since sample manipulation is automated.

b. Flow Injection Analysis (FIA)

i) Potentiometric detector (ISE)

The F-ISE has been widely used as a detector in FIA because of its simplicity, selectivity

and relatively low cost. Extensive studies concerning the use of F-ISE in FIA have been

performed by various research groups (Slanina et al., 1980; Trojanowicz and

Matuszewski, 1982; Frenzel and Brätter, 1986; Najib and Othman, 1992; Borzitsky et al.,

1993). The advantages associated with the coupling of the F-ISE as sensor to an FIA

system, enables this combination to be used in a reliable low cost, robust analyzer for the

determination of F- in toothpaste as well as in effluent streams, natural and borehole

water samples. By using FIA systems, the selectivity of the electrode is improved due to

the very short contact time of the interfering ions with the electrode membrane. The

application of the F- analysis in water samples are reported by several authors (Frenzel

and Brätter, 1986; Wang et al., 1995; Trojanowicz et al., 1998; Papaefstathiou et al.,

1995; Lopes da Conceição et al. 2000; Stefan et al. 1999).

The F-ISE has been widely used for monitoring F-, but its application in FIA systems has

been limited because of its memory effect (Slanina et al., 1980; Trojanowicz and

Matuszewski, 1982). Van Oort and Van Eerd (1983) obtained a fast response from the F-

ISE in a FIA system by using a mixture of methanol and total ionic-strength adjusting

buffer containing 10-6M F- in the carrier and by polishing the electrode surface with fine

wet alumina powder. However, no application to real samples was presented.

Frenzel and Brätter (1986) employed the FIA with a combination F-ISE, yielding an

excellent sensitivity down to 1ug/L, to the determination of traces of F- in tap water.

However, for the analysis of samples with high inherent ionic strength and fairly large

amounts of interfering elements, the optimization of the particular TISAB was necessary.


20

Emphasis was laid on the requirement for particular TISAB solutions to obtain accurate

results at low F- levels. The citrate TISAB was found to be effective to overcome the

interferences, but it reduces the sensitivity and causes peak broadening when compared to

the commonly used TISAB III. Recoveries after spiking tap water with F- concentrations

ranging from 0.01 to 1 mg/L were in the range 91-106 %.

Although the ISE detector has merits, such as simplicity, selectivity and low cost, one of

the major inconveniences in the applications, however, is its relatively slow response in

solutions of low F- ion concentration. Wang et al. (1995) aimed to establish a simple and

fast method, overcoming the pitfall, for determining low F- concentration using FIA by

applying an extrapolation procedure to predict the electrode equilibrium potential,

without the necessity of waiting for the establishment of the steady-state.

This work clearly showed that FIA with the extrapolation method was faster compared

with the conventional method, and had higher sensitivity. And also the accuracy of the

method was good, making it particularly suitable for the determination of large numbers

of samples at low F- concentration, such as in environmental analysis.

Stefan et al. (1999) reported on the incorporation of F-ISE in an FIA system for the

determination of F- in effluent streams, natural and borehole water. This system was

suitable for the on-line monitoring of F- at a sampling rate of 60 samples/h in the linear

range 10-4-10-1 (approximately 1.9-1900 mg/L) with an RSD of better than 0.6 %. But the

disadvantage of the proposed method was the relatively high detection limit of 0.34 mg/L.

Lopes da Conceição et al. (2000), proposed the use of the entire time-dependent response,

instead of the final potential reading alone (Trojanowicz et al. (1998), as an improved ion

activity assessment with ISE and a flow through system. They explored the concentration

gradient in a flow injection system for the potentiometric determination of F- in water

samples based on a gradient flow injection titration. Determination of F- is accomplished

by switching the injection valve between a standard and a sample solution. The

methodology developed gave results with a relative standard deviation of about 3 %,


21

detection limit of 0.19 mg/L, and the average recoveries after spiking natural samples

with F- were in the range 100-102%.

ii) Spectrophotometric detector

The FIA method for the F- determination using spectrophotometric detection can be

applied either without pretreatment (Wada et al., 1985; Arancibia et al., 2004) or with

pretreatment such as extraction (Garrido et al., 2002), gas diffusion (Fang et al., 1988;

Cardwell et al., 1994).

• Without pretreatment

Wada et al. (1985) reported the spectrophotometric determination of F- in the range of

0.03-1.2 mg/L, with lanthanum/alizarin complexone (1:1) in 70 % acetone by the FIA

method, in conjunction with a 500 cm reaction coil at 60°C at 24 samples per hour. This

method was applied to tap water samples, using standard addition method.

The presence of high sulphate constitutes a serious interference in the SPANDS method

for the determination of F- in water. Thus, Arancibia et al. (2004) explored the Zr-

SPANDS method incorporating full spectral measurements and a suitable multivariate

calibration methodology. Partial least-squares (PLS) regression appeared to be the

candidate of choice, due to the quality of its predictive models, the availability of

software and the ease of its implementation. They combined this strategy with flow

injection analysis, in order to develop an automated method for routine F- monitoring. By

coupling the FIA system to a diode-array spectrophotometric detector it is possible to

obtain the spectra from the recorded FIA peaks. The comparison between the obtained

results by using the FIA/PLS method and those obtained with an ISE showed a good

agreement.

• With pretreatment

1) Gas diffusion


22

Many reports have appeared on the use of gas diffusion (GD) in flow injection analysis

for different gaseous species but few have given details of this methodology for the

analysis of F-. Fang et al. (1988) gives a detailed description of an automated GD-FIA

method. This procedure involved separation and on-line preconcentration of the F-. The

sample was acidified and heated to 78°C while passing through a coil. HF formed in the

donor stream, diffused through a Teflon membrane in the gas diffusion unit to be

preconcentrated for a set time period in a static alkaline recipient solution enclosed in the

sample loop of an injection valve. At the end of that time, the injector was activated and

the carrier transported the absorbed sample down-stream where it merged with

lanthanum/alizarin complexone solution to be determined spectrophotometrically at 620

nm. F- was determined in wastewater at a sample rate of 40/hr with a detection limit of

0.1 mg/L. Al still interfered but there was considerable improvement in the concentration

of Al ion, which could be tolerated in the system, compared with that by the direct

spectrophotometric procedure without separation by gas diffusion.

Cardwell et al. (1994) shows the application of the GD, as a pre-treatment technique, with

the FIA method prior to spectrophotometric detector. The pre-treatment of F- was

achieved on-line by converting it to trimethylsilane, which then diffuses through a gas

permeable membrane to be absorbed in a stationary sodium hydroxide acceptor stream.

This stream was enclosed in the sample loop of an injection valve and after pre-treatment,

the F- sample was flushed into a flow injection manifold for spectrophotometric analysis

by the zirconium/alizarin procedure at 520 nm. The method was suitable for F- analysis in

the range 0.1-10 mg/L at a sampling rate of 17/hour. The detection limit was calculated to

be 0.055 mg/L and the quantification limit was found to be 0.18 mg/L. The accuracy and

precision of the method for real water samples, collected from boreholes, a drainage farm,

reservoirs and reticulation sites, have proven to be comparable with the F-ISE batch

method for the samples tested. The results obtained by the FIA method correlated well

with those obtained by the F-ISE method, giving a correlation coefficient of 0.9970.

2) Solid phase extraction (SPE)


23

Garrido et al. (2002) suggested a spectrophotometric determination of F- in a flow

assembly integrated on-line to an open/closed FIA system to remove interference by solid

phase extraction. The method is based on the enhanced fluorescence of quercitin-Zr (IV)

complex when F- ion is present in the sample. An open/close FIA manifold with a mini-

column of Dowex 50W X8 resins was used to remove the most important interference of

aluminium. The proposed method was applied for F- determination in several water

samples and the results compared with the SPANDS method. Good agreement between

the two methods was obtained.

c. Sequential Injection Analysis (SIA)

i) Potentiometric detector (ISE)

Applications of SIA to environmental analysis is an attractive field due to the ability to

perform multicomponent determinations using several chemical reactions involving

various reagents in different zones, various types of detectors or even diverse multivariate

techniques. Specifically, SIA can be used to obtain necessary control measurements of

water in a rapid and simple way. Alpízar et al. (1996) illustrated the possibility of SIA

on-line implementation with potentiometric detection for simultaneous chloride and F-

determination in waters, using two ion selective electrodes in two serial flow-through

cells. The necessary ionic strength adjustment was obtained by mixing the TISAB

solution with the sample by diffusion during the propelling process to the detection cells.

The proposed method gave results with a relative standard deviation of 3.7 % at 1µg/mL

F- and the average recoveries after spiking tap water samples with F- were in the range

92-109%. This methodology was applied successfully to drinking tap water samples.

The versatility of the SIA technique is centered on a selection valve where each port of

the valve allows a different operation to be performed. Van Staden et al. (2000)

investigated the peak profiles of four different buffer-sample SIA configurations, e.g.

buffer-sample; sample-buffer; buffer-sample-buffer and sample-buffer-sample, with a F-

ISE and the application to the determination of F- in tap water. The best response


24

characteristics and peak shapes as well as recovery and precision values were obtained

for the buffer-sample configuration.

2.2.4 Determination of F- by potentiometric analysis The potentiometric technique, mainly F-ISE, is considered the simplest and most reliable

for F- determination (Durst, 1969; Fry and Taves, 1970; Hall et al., 1972; Hallsworth et

al., 1976; Retief et at., 1985; Kissa, 1987; Villa, 1988; Vogel et al., 1990). The F-ISE

method for the F- determination can be applied either without pretreatment technique,

namely conventional potentiometric method, (Babcock and Johnson, 1968; Frant and

Ross, 1968; Harwood, 1969) or with pretreatment technique, the extended method, such

as co-precipitation (Okutani et al., 1989) and steam distillation (Analytical working group,

1986).

a. Conventional potentiometric method

Babcock and Johnson (1968) used the electrode without the addition of any ionic buffer

for the determination of F- in municipal water. Their particular application was basically

that of monitoring a municipal water supply for F-. Under such conditions, the

composition of the water, and hence the total ionic strength, would remain fairly constant

since the water originated form the same source. Consequently there was no great

necessity to add electrolytes to ensure constant ionic strength.

Frant and Ross (1968), on the other hand, have used the electrode on samples from

different sources, and so had to make use of an ionic buffer. This method was selected for

routine F- analysis. Later tests indicated that the composition of the ion buffer should be

changed in order to overcome interference from aluminium. Results obtained indicate

that with the new ionic buffer, satisfactory analysis for F- can be obtained even in the

presence of up to 5 mg Al /L.

Crosby et al. (1968) evaluated ISE method and five spectrophotometric procedures for

the determination of F- in potable waters and other aqueous solutions, with respect to


25

reproducibility, sensitivity, range, stability of coloured products and of reagents,

specificity and effect of temperature. F- was determined directly, without the need for any

separation step. The ISE method was shown to be less susceptible than the colorimetric

methods to interference from other ions in solution.

Harwood (1969) evaluated a commercially available F-ISE for use in routine F-

determination on water samples. Tests on samples showed that aluminium interfered

seriously and an ionic buffer containing CDTA was recommended in order to control this

interference, since both citrate and EDTA had only limited applicability. With CDTA

present, reliable F- determination were possible in the presence of up to 5 mg/L

aluminium. The F- analysis results were compared using both SPANDS method and ISE

and the results, obtained with the F-ISE, were more trustworthy because of the inherent

difficulties of the SPANDS method.

b. Extended potentiometric method

i) Co-precipitation

Okutani et al. (1989) performed the F- determination in natural water, such as sea, tap,

well and river waters, using F-ISE after co-precipitation with aluminium phosphate. The

value, obtained by the proposed method, agreed well with those obtained by IC.

ii) Steam distillation

The analytical working group of the comité technique européen du fluor (1986)

determined the F- in rain water and aqueous effluents using both direct potentiometric and

extended potentiometric methods. All the methods are based on final measurement of F-

by means of the F-ISE. Interferences which could not be eliminated by the masking

power of the TISAB buffer solution (1M sodium chloride and 0.1 M citrate buffer at pH

5.5) were avoided by separation of F- by steam distillation. F- was distilled from a

mixture of perchloric and phosphoric acid. Several samples of waste water were


26

determined by using both methods, and the results showed that the extended method

yielded satisfactory results.

2.2.5 Other methods

a. Pervaporation

Pervaporation is an energy efficient combination of membrane permeation and

evaporation and is considered an attractive alternative for many separation

processes. The process, which requires only low temperatures and pressures, has cost

and performance advantages compared with distillation for constant-boiling

mixtures. Pervaporation is particularly suitable for the dehydration of organic solvents

and the removal of organics, such as methanol and acetone, from aqueous streams. In the

pervaporation process the liquid (mixture) is in contact with the membrane. At the

permeate side, a partial pressure is generated by means of a vacuum pump, or by means

of an inert gas flow. The components of the liquid that move through the membrane are

vaporized by the low pressure, removed and condensed.

Papaefstathiou et al. (1995) developed a selective method for the determination of F- in

contaminated samples based on pervaporation of a volatile derivative and potentiometric

monitoring of the anion after collection in a basic solution. F- was converted to volatile

trimethylfluorosilane by reaction with hexamethyldisilazane, and absorbed in dilute

NaOH solution. Both the continuous and stopped-flow modes were used in order to

accomplish a variable efficiency of the separation process thus enlarging the

determination range which was established between 5 and 100 mg/L. The precision was

between 2.60 and 3.58 %. The determination of the analyte in samples from different

sources, such as tap water, well water, fertilizers and ceramic industry wastewater,

testified to its usefulness, giving the recovery range between 87.23 and 106.09% with

RSD values between 0.72 and 4.17 %.

b. Extraction


27

Extraction is primarily used to separate analytes from a matrix thus eliminating or

reducing interferences from other components. Secondly, it is used to concentrate the

analyte up to a detectable concentration level.

Nishimoto et al. (2001) employed spectrophotometry after solvent extraction of F- into

chloroform using [2,23-diethy-8,17-bis(2-ethoxycarbonylethyl)-3,7,12,13,18,22-

hexamethylsapphyrin (H3 sap), sapphyrin] in the presence of 0.1 M sodium nitrate. F- ion

was extracted from chloroform with the positively charged sapphyrin, as ion-pair

complexes of H5sap2+·F-·NO3-, that gave a new absorption maximum wavelength at 448

nm. The results provided the determination of F- ion at concentrations as low as 0.01

mg/L. Metal ions like Al(III) and Fe(III) did not interfere with the determination of F-.

2.2.6 Comparative studies

The references to results of comparative studies are summarized in Table 2.8.

Table 2.8 A summary of comparative study of F- analysis

Comparison method Reference Section Spectrophotometry ISE Crosby et al. (1968) 2.2.2, 2.2.4.a

SPANDS ISE Harwood (1969) 2.2.4.a SPANDS ISE Leod and Crist (1973) 2.2.2

GD-FIA ISE Cardwell et al. (1994) 2.2.3.b.ii 2.2.4.a

IC ISE Van den Hoop et al. (1996) 2.2.4.a

SPANDS Solid Phase Extraction Garrido et al. (2002) 2.2.3.b.ii

FIA/PLS ISE Arancibia et al. (2004) 2.2.3.b.ii 2.2.4.a

Theory of analytical methods ______________________________________________________________________________________

28

Chapter 3 Theory of analytical methods

3.1 Ion Selective Electrodes (ISE)

3.1.1 Introduction

The Ion Selective Electrode (ISE) is a sensor for the potentiometric determination of

ionic species and one of the most frequently used devices during laboratory analysis in

industry, process control, physiological measurements, and environmental monitoring.

The functioning of an ISE is based on the selectivity of passage of charged species from

one phase to another, leading to the creation of a potential difference. The most

commonly used ISE is the pH probe. Other ions that can be measured include anions such

as, F-, Br-, I-, CN-, NO3-, NO2

-, the cations such as, Pb2+, K+, Ca2+, Cu2+ and Na+, and

gasses in solution such as NH3, CO2, N2 and O2.

Some advantages of the ISE method are:

• It is considerably less expensive than other analytical techniques, such as Atomic

Adsorption Spectrophotometry (AAS) or Ion Chromatography (IC).

• It is simple to use and measurement is quick.

• It has a large range of applications and can be used over a wide concentration

range.

• It is robust and durable and ideal for use in either field or laboratory environments.

• Accuracy and precision levels of ± 2% for some ions compare favourably with

analytical techniques which require more complex and expensive instrumentation.

3.1.2 Important concepts of ISE (Skoog et al., 1997)


29

a. Activity

Activity is the effective concentration of a free ion in a solution. It is dependent on the

total ionic strength of a solution. In dilute solutions with low ionic strength, the ions are

relatively far apart and free to move so that the activity and concentration are virtually

identical. In more concentrated solutions, or high ionic strength, however, the ions are

more closely packed and their charges are shielded. Thus activity may differ from

concentration and is usually lower up to ionic strengths of about 0.1 M. In solutions with

concentrations of more than 1 M other interactions such as hydration and inter-ionic

interactions, which can impede movement, become important, and activity can actually

be larger than the concentration. In the case of ISE measurements, the activity of the

measured ion determines the proportion of ions passing through the ion-selective

membrane and thus the magnitude of the voltage developed by the electrode.

The activity of an ion in solution for any known concentration may be calculated from

the formula:

x xa Cγ= × (3.1) Where, xγ is the activity coefficient and C is the Molar concentration

i) Ionic strength (µ)

The ionic strength of a solution is a measure of the total effect of all the ions, both

positive and negative, present in the solution and is given by the equation:

20.5 i ii

C Zµ = ×∑ (3.2)

Where, Ci = the species molar concentration Zi = charges of ions

ii) Activity coefficient

The activity coefficient is dependent on the ionic strength of the solution. It becomes

progressively lower as the ionic strength increases, due to inter-ionic interactions. The

activity coefficient for any ion in solution can be calculated using the Debye-Hückel

equation that permits the calculation of activity coefficients of ions from their charge and

their average size.


30

-Log xγ = µαµ

x

xZ3.31

51.0 2

+ (3.3)

Where: Z x = the ionic charge, µ = the ionic strength of the solution αx = effective diameter of the hydrated ion X in nanometers (10-9m), approximately 0.3nm for most singly charged ions. The constants 0.51 and 3.3 are applicable to aqueous solutions at 25 °C

In solutions of very low ionic strength, when µ is less than 0.01, the denominator of the

equation becomes 1 and the equation can be simplified to:

-Log xγ = 0.51 Zx2 µ (3.4)

b. Nernst equation

This is the fundamental equation, which relates the electrode potential to the activity of

the measured ions in the test solution.

log0592.0n

EE −°= (a) (3.5)

Where, E = the total potential (in mV) developed between the sensing and reference electrodes E0 = a constant which is characteristic of the particular ISE/reference pair n = the charge on the ion (with sign) Log (a) = the logarithm of the activity of the measured ion

Note that the activity is equivalent to the concentration in dilute solutions but becomes

increasingly lower as the ionic strength increases. This is effectively the equation of a

straight line:

y mx c= + (3.6)

Where, y = E = the measured electrode response in mV, x = Log (a), c = E0 = the intercept on the y axis. m = -0.0592/n = the slope of the line (the electrode slope)

An electrode is said to have a Nernstian response over a given concentration range if a

plot of the potential difference versus the logarithm of the ionic activity of a given

species in the test solution, is linear with a slope factor that is given by the Nernst

equation.


31

c. Range of linear response

At high and very low ion activities there are deviations from linearity. Typically, the

electrode calibration curve exhibits linear response range between 10-1M and 10-5M.

d. Slope

The slope of an electrode is the gradient of the line formed by plotting the electrode

response in millivolts against the logarithm of the activity (or concentration) of the

measured ion. The theoretical Nernstian slope at 25°C is 59.16 mV/decade change in

activity for monovalent ions and 29.58 mV/decade for divalent ions. In practice the slope

is generally lower than this theoretical value due to inefficiency of the ion selective

membranes and failure to meet ideal conditions.

Measured slopes generally lie in the range 54±5 and 26±3 respectively.

e. Liquid junction potential

Liquid junction potential is the potential formed at the interface between any two-

electrolyte solutions of different compositions. In ISE measurements, the most important

liquid junction is that between the reference electrode filling solution and the sample

solution. Ideally, this potential should be as low and as constant as possible, despite

variations in the external solution. Reference electrode filling solutions are chosen to

minimize this potential.

f. Hysteresis (electrode memory)

The occurrence of a different value in the potential difference after the concentration of

the test solution has been changed and then restored to its original value. This systematic

error is generally in the direction of the concentration of the previous solution. Thus, if

the electrodes are washed with water between each sample measurement, successive

readings of the same solution can be expected to become progressively lower for cation

and higher for anions.


32

g. Response time

The length of time that is necessary to obtain a stable electrode potential when the

electrode is removed from one solution and placed in another of different concentration.

The electrode type, the magnitude and direction of the concentration change, hysteresis,

temperature, and the presence of interfering ions, affect response time. For ISE

specifications it is defined as the time to complete 90% of the change to the new value

and is generally quoted as less than ten seconds.

3.1.3 Components of ISE

The essential components of an ion selective measuring system include a reference

electrode, a sensing electrode (or indicator electrode) and a potential measuring device

(Figure 3.1).

Figure 3.1 Electrochemical measuring system of ISE

a. Reference electrode

To make a measurement, a second unvarying electrical potential is needed to compare

with the sensing membrane potential. The reference electrode provides this function. A

filling solution within the reference electrode completes the electrical circuit between the

sample and internal cell of the reference electrode. The point of contact between the

sample and filling solution is called the liquid junction. Typical examples of commonly


33

used reference electrodes are the calomel electrodes and the silver/silver chloride

electrodes.

i) Silver/silver chloride electrode

It is the most common and simplest reference system. This generally consists of a

cylindrical glass tube containing a 4 M solution of KCl saturated with AgCl. In

electrochemical terms, the half-cell can be represented by:

Ag | AgCl (Satd), KCl (Satd) || (3.7)

And the electrode reaction is:

AgCl (s) + e- ↔ Ag (s) + Cl- (aq) (3.8)

The electrode potential for this half-cell is + 0.2046 V relative to the Standard Hydrogen

Electrode at 25°C

ii) Calomel electrode

It consists of mercury in contact with a solution that is saturated with mercury (I) chloride

and also contains a known concentration of KCl. In electrochemical terms, the half-cell

can be represented by:

Hg | Hg2Cl2 (Satd), KCl (xM) || (3.9)

Where, x is the molar concentration of KCl

And the electrode reaction is:

Hg2Cl2 (s) + 2e- ↔ 2Hg (l) + 2Cl- (aq) (3.10)

The electrode potential for this half-cell is + 0.2444 V relative to the Standard Hydrogen

Electrode at 25°C.

b. Sensing electrode (indicator electrode)


34

Membrane electrodes are called ion selective electrodes because of their high selectivity.

The membranes must have minimal solubility, electrical conductivity, and selective

reactivity with the analyte of interest. Several types of sensing electrodes are

commercially available (Orion Research). Table 3.1 shows different types of ISE and

some of the analytes they are used for.

Table 3.1 Types of ion selective membrane electrode

Types Example

Single crystal LaF3 for F- Crystalline membrane electrode Polycrystalline or mixed

crystal Ag2S for S2- and Ag+

Glass Silicate glasses for Na+ and H+

Liquid Liquid ion exchangers for Ca2+ and neutral carriers for K+

Noncrystalline membrane electrode

Immobilized liquid in a rigid polymer

Polyvinyl chloride matrix for Ca2+ and NO3

-

Attention will be focused on crystalline membrane electrodes used for the determination

of fluoride (F-), since F- is the target element of this study.

3.1.4 Fluoride ion selective electrode

a. Introduction

The F- Ion-Selective Electrode is designed for the detection of F- in aqueous solutions and

is suitable for use in both field and laboratory applications. The F- electrode consists of a

solid-state inorganic membrane of LaF3 crystal, bounded into an epoxy body (Figure 3.2).

Figure 3.2 Construction of fluoride electrode


35

Within this crystal lattice, F- is a mobile ion whereas the lanthanum ions are in fixed

lattice positions. A lattice vacancy will not accept ions other than F- due to differences in

size, shape or charge. Therefore, the LaF3 membrane is highly selective for conduction of

F- ions. At the two interfaces, ionization creates a charge on the membrane surface as

given by:

LaF3 (solid) ↔ LaF2+ (solid) + F- (solution) (3.11)

When the sensor is in contact with a solution containing F- ion, a potential, dependent

upon the level of F- ion in the solution, develops across the membrane. The construction

of the membrane is shown in Figure 3.3.

Figure 3.3 Crystalline membrane constructions

b. The mechanism of the F-ISE (Skoog et al., 1998)

The magnitude of the charge is dependent upon the F- ion concentration of the solution.

Thus, the side of the membrane encountering the lower F- ion concentration becomes

positive with respect to the other surface. It is this charge difference that provides a

measure of the difference in F- concentration of the two solutions. The membrane

potential (Em) consists of two potentials E1 and E2 on both sides of membrane-inner

solution and membrane-outer solution interfaces. This can be written as,

Em = E2 - E1 (3.12)


36

The F- ion activities at each face are related to the Nernst–like relationships.

E1 = j1-1

1'log0592.0aa

n (3.13)

E2 = j2-2

2 'log0592.0aa

n (3.14)

Where, j1 and j2 are constants and a1 and a2 are activities of F- in the solutions on the

external and internal sides of membrane respectively. The terms a1’ and a2’ are the

activities of F- at the external and internal surfaces of the membrane. Since the two

membrane surfaces are usually identical, so j1 and j2 as well as a1’ and a2’ are also

identical. By substituting two equations and employing the equalities j1 = j2, a1’ = a2’ and

n=1, the membrane potential (Em) can be rearranged like,

Em = E2 - E1 = 0.0592 log 1

2

aa (3.15)

Thus, the membrane potential Em depends only on the F- ion activities of the solutions on

either side of the membrane. For an ISE, the ion activity of the internal solution a2 is a

constant at temperature 25 °C, the equation can be simplified and the potential of a cell

containing a lanthanum F- electrode is given by the equation to

Em = L + 0.0592 logaF- = L - 0.0592 pF- (3.16)

Where, L = 0.0592 log a2

The net potential is measured against a reference electrode.

Ecell = Eise – Eref (3.17)

The results are usually given in milli volts or even directly as concentrations of the

species being measured. The strength of this charge is directly proportional to the

concentration of the selected ion.


37

c. Total Ionic Strength Adjustment Buffer (TISAB) for F-ISE (Front and Ross, 1968;

Durst, 1969)

The problems with Ion Selective Electrode measurements are the effect of interferences

from other ions in solution, the effect of the ionic strength of the solution reducing the

measured activity relative to the true concentration at high concentrations. TISAB (Total

Ionic Strength Adjustment Buffer) is normally added to samples and standards in order to

solve these problems. The function of the TISAB is discussed below.

i) Adjustment of a constant ionic strength for both standards and samples

As discussed in Section 3.1.2.a, the F- electrode actually measures activity. The measured

result is therefore strongly dependent on the ionic strength of the medium. TISAB

ensures a constant ionic strength background, minimizing variations between samples

and standards, which may otherwise influence the electrode potential. A constant ionic

strength also reduces errors due to liquid junction potentials (Butler, 1969).

ii) Optimization of the pH

Of the commonly occurring ionic species (other than F-) the F- electrode responds

directly only to hydroxide ions. This effect causes serious positive errors at high pH

values. On the other hand under the acidic condition, below pH 5, the formation of HF

and HF2- reduces the activity of the F- ions, which lead to a negative error. Use of TISAB

ensures constant pH of around 5.4 is maintained eliminating the effect of pH changes

(Vesely and Stulik, 1974).

iii) De-complexation of the interference ions

Many TISAB solutions also include a masking agent to preferentially complex any

potentially interfering species, e.g. di and trivalent cations such as aluminium, iron,

magnesium and calcium. TISAB contains certain metal-complexing agents (masking

agents) to release free F- ion from certain metal-F- complexes. CDTA has found wide

usage (Nicolson and Duff, 1981).


38

3.2 Ion Chromatography (IC)

3.2.1. Introduction

Small et al. (1975) pioneered the IC method for separation and quantitative determination

of inorganic ions and is the trade name for the Dionex system. It is now a well-

established method for the analysis of ionic species and has been approved for

compliance monitoring of common anions in US drinking water since the mid-1980s, as

described in USEPA method 300.1 (1997). Many regulatory and standard organizations,

such as ASTM, AOAC, ISO and US EPA, have approved methods of analysis based

upon IC, most of which have been published within the last 10 years (Jackson, 2000;

2001). It is a form of liquid chromatography that uses ion-exchange resins to separate

atomic or molecular ions based on an ion exchange process between the mobile phase

(eluent) and the exchange groups (the analyte) covalently bound to the stationary phase

on the column. As the eluent flows through the column, the components of the analyte

will move down the column at different speeds according to the affinity of each analyte

to the stationary phase, and therefore separate from one another. Subsequently, the

detector, which can be conductivity detector, ICP-MS detector and spectrofluorimetric

detector etc., measure the different anions as they emerge from the column. The

conductivity detector that is most commonly used in ion chromatography is introduced in

this section. Compared to non-chromatography techniques, the IC method has the

advantages of separation before detection, increased sensitivity, simultaneous analysis in

a single run, simple sample preparation and faster analysis time, which is less than 15

minutes (Dasgupta, 1992; Romano and Krol, 1992).

3.2.2 Instrumentation and process of IC

a. Flow schematic

The main components of an IC system are a high-pressure pump, a column and an

injector system as well as a detector. The flow path through a single-column Ion

chromatography system is shown in Figure 3.4.


39

Figure 3.4 IC Flow Schematic

The eluent is filtered and pumped through a chromatographic column, the sample is

loaded and injected to the column and the effluent is monitored using a detector and

recorded as peaks.

b. Sample injection: Rheodyne injection valve

The Rheodyne injection valve has two operating positions, which are Load and Inject. In

the Load position, sample is loaded into the sample loop, where it is held until injection.

In the Inject position, the sample is swept to the column for analysis. Eluent flows

through one of two paths, depending on the valve position.

c. Operation and maintenance

i) Degassing and filtering eluents

Degassing all eluents and storing them in reservoirs pressurized with filtered inert gas

helps prevent bubbles (resulting from eluent out gassing) from forming in the pump and

the detector cell. Several degassing procedures can be used, including vacuum degassing,

sparing with helium, or sonication without vacuum. Filtration of the eluents before

operation works to remove small particulates that may contaminate the pump check

valves and cause erratic flow rates or loss of prime.


40

ii) Diluting

The concentrations of ionic species in different samples can vary widely from sample to

sample, no single dilution factor can be recommended for all samples of one type. In

many water samples concentrations are so low that dilution is not necessary. Diluting

with eluent, also using eluent to prepare the calibration blank and standards, minimizes

the effect of the water dip at the beginning of the chromatogram. This is most important

for F- and chloride, which elute near the water dip. To improve the accuracy of early

eluting peak determinations, such as F-, at concentrations below 50 µg/L, dilute standards

in eluent or spike the samples with concentrated eluent to minimize the water dip.

d. The separation columns

Heavy-wall glass, stainless steel and plastic are among materials that can withstand high

pressures and are thus used to construct HPLC columns. They must also be able to resist

the chemical action of the mobile phase. Wall irregularities will cause a well-packed

column to channel near the wall or packing interface thus the tubing must have a smooth,

precision bore internal diameter. Channels would cause peak broadening and a decrease

in efficiency. Column connections are made with low dead-volume fittings, which

prevent stagnant pockets of eluent.

Usually a short guard column, which prevents poisoning of the separator column by

sorbing organic contaminants and removing particulates, is placed in front of the

analytical column. This serves to increase the life of the analytical column by removing

particulate matter and contaminants from the solvents.

e. Ion exchange equilibria In ion chromatography, the support material is a polystyrene/divinylbenzene (PS/DVB)

based resin (Joachim, 1986). The column material (resin) is synthesized to serve as cation

exchange or anion exchange columns. The most common site on anion exchanger is the


41

tertiary amine group –N(CH3)3+OH-, a strong base and the sulphonic acid group –SO3

-H+

for a cation.

i) Anion-exchange resin

The anion-exchange resins used by Dionex are composed of a surface sulphonated

PS/DVB core (10-25 µm) and a totally porous latex particle which is completely

aminated. The latex particles have a considerably smaller diameter of about 0.1 µm and

carry the actual ion exchange function –NR3+. The structure of the anion exchange resins

used by Dionex and developed by Small is illustrated in Figure 3.5.

Figure 3.5 The structure of the anion exchange resins

The exchange process of a solution, which contains an anion, AX- on an anion exchange

column, is described by the following reaction.

xN(CH3)3+OH-(solid)+AX-(solution)↔(N(CH3)3

+)xAX-(solid)+xOH-(solution) (3.18)

The affinity of the resin for the anion relative to OH- ion is SO42->C2O4

2->I->NO3->Br-

>Cl->HCO2->OH->F-. These conditions depend on various factors, such as the type of

resin, size or the hydrated ion, and so on.


42

ii) Cation exchange resin

The stationary phase of a cation exchange column is based on inert, surface sulphonated,

crosslinked polystyrene (Figure 3.6; Weiss, 1986).

Figure 3.6 The structure of a cation exchange resin

The exchange process for a cation, M x+, can be described by the equilibrium

xRSO3-H+ (solid) + Mx+ (solution) ↔(RSO3

- )xMx+ (solid)+xH+ (solution) (3.19)

Since the core of a cation exchange resin is strongly hydrophobic, the diffusion into the

resin of highly dissociated and hydrated species, such as Na+, K+, and Mg2+, can be

neglected. Consequently, the diffusion paths are short and high efficiencies are achieved.

f. The chemical suppression

The conductance of a solution is proportional to the concentration of the ions dissolved in

the solution. Separation methods for ions require eluents containing strong electrolytes,

which presents a problem, because the analyte anions must be detected without the

detector being overwhelmed by the ions in the eluent. They present only a tiny fraction of

the concentration of the eluent, approximately 100 times less than the eluent. The best

solution for removing the eluent anions from the solution is to neutralize the eluent in a

suppressor.

Figure 3.7 illustrates the suppression mechanism in the Dionex Anion Self-Regenerating

Suppressor (ASRS) that was used in this study.


43

Figure 3.7 ASRS suppression mechanism

Analyte anions elute from the column with sodium counter ions. Two electrodes, one

beside each membrane (on the side opposite the eluent) hydrolyzes water to hydrogen

and hydroxide ions. Hydrogen ions diffuse across the membrane next to the anode,

neutralizing the eluent hydroxide to water, while sodium ions from the eluent diffuse

across the other membrane, providing counter ions to the hydroxide being generated at

the cathode.

In effect, sodium hydroxide from the eluent is transferred across the membrane and does

not reach the detector. The resulting eluent background conductivity is near zero,

considerably lower than before suppression. Also, the counter ion to the anion analytes is

now a hydrogen ion with conductivity seven times higher than the original sodium

counter ion. Since both the anion analyte and the cation counter ion produce the detector

response, response is increased. The suppressor lowers the background conductivity (and

thus the baseline noise and drift) and increases the signal. Suppression can also be

accomplished without water electrolysis by pumping a dilute sulfuric acid solution (the

regenerant) through the suppressor on the side of the membranes opposite the eluent.

For ion chromatography of cations, the suppressor membranes are anion exchange

polymers that allow anions to pass freely but exclude cations. The eluent uses dilute acids

such as methanesulfonic acid. In the Dionex Cation Self-Regenerating Suppressor

(CSRS), methanesulfonate counter ions are replaced by hydroxide generated by the


44

electrolysis of water. This neutralizes the acidic eluent and provides the highly

conductive hydroxide counter ion to the analyte cations.

g. Conductivity detection

Conductivity detectors are commonly used in ion chromatography mainly because of

their excellent sensitivity and predictable response to concentration changes. Ions in

solution conduct electrical current when a voltage is applied between electrodes in

contact with the solution. Since the magnitude of this current is proportional to the

concentration of dissolved ions, conductivity detection is useful for quantifying ionic

analytes.

i) Principle

The conductivity of a solution is measured by applying an alternating voltage between

two electrodes in a conductivity cell. At any instant in time, negatively charged anions

migrate toward the positive electrode and positively charged cations migrate toward the

negative electrode. Since the detector applies a known voltage to the cell electrodes and

the current is measured, the solution resistance, R, is calculated from Ohm’s law.

R = iE (3.20)

The inverse of the measured solution resistance is the conductance, G, measured in siemens (S).

G = R1 =

Ei (3.21)

The conductivity cell constant, K, corrects the measured conductance. The conductivity

of the solution is the conductance, which would be measured in a standard cell containing

electrodes of 1 cm2 surface area held 1 cm apart. This quantity is the conductivity, κ. The

units of conductivity are siemens per cm (S/cm).

K Gκ = × (3.22)


45

According to Kohlraush’s law of independent migration, conductivity is directly

proportional to concentration (that is, the conductivity of a dilute solution is the sum of

the individual contributions to conductivity of all the ions in the solution multiplied by

their concentration). Kohlraush’s law further states that each ion carries its portion of the

total conductivity without being affected by any of the other ions in solution. It is stated

as an equation:

К = 1000

∑ °

iiicλ (3.23)

Where: К = measured conductivity(S/cm), ci = concentration of the ions in equivalents/L (Equivalents/L equals moles/L times the charge on the ion)

The ionic limiting equivalent conductivity ( °iλ ) is specific for each ion. The unit for °

iλ is

S.cm2/equivalent. It is the conductivity of the ion divided by the concentration and

extrapolated to infinite dilution. Limiting equivalent conductivities for a number of

organic and inorganic ions are given in Table 3.2.

Table 3.2 Limiting equivalent conductivities at 25 ° C Anions °

iλ Cations °iλ

OH- 198 H+ 350

F- 54 Li+ 39

Cl- 76 Na+ 50

Br- 78 K+ 74

I- 77 NH4+ 73

NO3- 71 Mg2+ 53

HCO3- 45 Ca2+ 60

SO42- 80 Sr2+ 59

Acetate 41 CH3NH3+ 58

Benzoate 32 N(CH3CH2)4+ 33

Values of °iλ from Table 3.2 can be used to calculate conductivities of solutions

containing ions. For example, the limiting equivalent conductivity for NaCl at 25 ºC is

126.5. This is the sum of the ionic limiting equivalent conductivity for Na+, which is 50.1,

plus that of Cl-, which are 76.4. A 0.1 mM solution of NaCl at 25 ºC has a conductivity of

0.1×126.5, or 12.65 µS/cm.


46

ii) Species detected by conductivity

Conductivity detection is best suited to anions and cations of strong acids and bases, such

as chloride, sulfate, trifluoroacetate, sodium, and potassium. Ions of weaker acids are

detected, provided that the eluent pH is chosen to maximize analyte dissociation

Sensitivity is best for anions with pKa values below 6. As analyte ionization

(dissociation) decreases, so does sensitivity. Anions with a pKa above 7 can be detected

under certain conditions, but signal-to-noise ratios are generally poorer. All organic acids

with carboxylate, sulfonate, or phosphonate functional groups have pKa’s below 4.75, so

conductivity is the preferred detection method for these species. Common inorganic

strong acid anions include chloride, nitrate, phosphate, and sulfate.

3.2.3 The important parameters in IC (Skoog et al., 1997 and 1998) a. Resolution

Resolution, R, provides a quantitative measure of the ability of a column to separate two

analytes. It is expressed as the ratio of the distance between the two peak maxima to the

mean value of the peak width at the baseline from the chromatogram. Figure 3.8 shows

those parameters for assessing the chromatographic separations.

Figure 3.8 Parameters for assessing the chromatographic separations


47

The equation of the resolution is given below.

R =

221

12

wwtt msms

+− =

21

2wwt

+∆ (3.24)

Where, tms2 and tms1: Gross retention times for peaks 1 and 2 respectively w1 and w2: Peak widths along the baseline of peak 1 and 2 respectively

The resolution should be at least 0.5 to confirm two peaks as separate peaks and if R is

greater than 1.5 two peaks are considered as completely separate peaks. The resolution

can be improved by lengthening the column but this will also increase the analysis time.

i) Capacity factor (k’)

Capacity factor is a dimensionless measure of retention, which is independent of column

length or eluent flow rate. It is calculated as follows:

k’ = m

ms

ttt − (3.25)

Where: ts = retention time of solute tm = retention time of unretained solute (column void volume)

Small values of k’ imply that the compound elutes near the void volume. Thus, the

separation will be very poor. Large k’ values indicate a good separation but also imply

longer analysis times, an associated peak broadening, and a decrease in sensitivity.

ii) Column efficiency (N) (theoretical plates)

It is a measure of the narrowness of analyte bands as they elute from the column. High

efficiency is desirable because resolution between closely spaced bands improves with

greater efficiency. For a symmetrical (Gaussian) peak, column efficiency can be

determined by the following:

N = 5.54 2

2/1

1⎟⎟⎠

⎞⎜⎜⎝

⎛W

t (3.26)

Where: t1 = the peak retention time, in seconds W1/2 = the peak width at 1/2 height, in seconds


48

Column efficiency is proportional to column length, for a given resin and column

diameter, increasing the column length increases the column efficiency.

3.3 Inductively Coupled Plasma-Optical Emission Spectroscopy (ICP-OES)

3.3.1 Introduction Plasma is a partly ionized gas (often argon), which contains particles of various types

such as electrons, atoms, ions and molecules, maintained by an external field. As a whole,

it is electrically neutral. The most common plasma source is the ICP torch (Lajunen,

1991). The introduction of ICP as a source for analytical atomic emission spectrometry

constituted a revolutionary advance in that field. The performance characteristics of ICP-

OES, specifically its versatility, wide applicability, and ease of use, were almost

unparalleled among methods of elemental analysis.

ICP-OES is a major technique for quantitative multi-element analysis and very powerful

analytical tool. In terms of sensitivity, ICP-OES is generally comparable to flame atomic

absorption, i.e. detection limits are typically at the µg/L level in aqueous solutions

(Varma, 1991).

3.3.2 Instrumentation for ICP-OES (Lajunen , 1991; Varma, 1991; Liberty, 1991) A typical arrangement for ICP-OES is shown in Figure 3.9 and consists of a sample

introduction system, the ICP torch and its associated gas supplies, a radio-frequency

generator, an optical spectrometer, detectors and associated electronics and computerized

instrument control, data collection and analysis.


49

Figure 3.9 ICP instrumentation (http://icp-oes.com/instrument_layout.htm, Moore, 1989)

a. Sample introduction

For most analysis the liquid sample is pumped into a pneumatic nebuliser (Figure3.9),

which aspirates the sample with high velocity argon, forming a fine aerosol. The aerosol

then passes into a spray chamber where larger droplets are removed via a drain (Jarvis et

al., 1992). Typically, only 2% of the original aerosol passes through the spray chamber

(Olesik, 1996). This process is necessary to produce droplets small enough to be

vaporized in the plasma torch. The finer droplets in the vapor are transported to the

plasma.

b. ICP torch It is formed when energy is transferred to a gas, usually argon, by means of an induction

coil. The plasma forms above the inner tube (Plasma tube) and within the outer tube

(coolant tube) and its location are largely determined by the position of the induction coil

(Figure 3.10).


50

Figure 3.10 Inductively coupled plasma (Moore, 1989)

i) The formation mechanism of the ICP

The induction coil, which is part of an oscillator circuit, commonly a radio frequency

generator, induces a strong high-frequency magnetic field. The magnetic field, generated

by the RF-currents, induces an electric field in the coil region. When seed electrons and

ions, formed by the generation of a spark in the argon stream using a Tesla coil, are

introduced, the electrons accelerate and encounter resistance through collisions with

argon atoms. This collision yields an avalanche of charged particles in sufficient numbers

to absorb the energy from the radio frequency field. Once the electrons reach the

ionization potential of the plasma support gas (usually argon), further ionization takes

place, and stable self-sustaining plasma is formed. A vortex flow of argon is used to cool

the inside walls of the torch for protecting the outer tube from the very hot plasma (6000

to 10000 K) and stabilizes and thermally isolates the plasma.

ii) Zones in the ICP

The various processes occurring in the ICP lead to several rather distinct zones, with very


51

different characteristics. The excitation stages within the torch as well as the different

region of ICP are shown in Figure 3.11.

Figure 3.11 Torch-excitation stages within the torch (Color: from yttrium) (Liberty, 1991)

The Pre Heating Zone (PHZ) is the place where desolvation, evaporation and dissociation

take place. The most obvious zone is the intensely luminous region, called the induction

zone, where energy from the induction coil is coupled into the plasma. When solutions of

certain elements are introduced, the emission of visible light allows the various zones in

the axial region of the ICP. The specific atom radiation is emitted in the Initial Radiation

Zone (IRZ), which shows that the salts are dissociated into free atom in that region. Just

above the IRZ, the specific atomic radiation emission is no longer visible because of the

ionization of the atom in this region. The majority of elements emit most intensely in this

region and it is therefore called the Normal Analytical Zone (NAZ). Above the NAZ, the

specific atomic radiation becomes visible again as the gas stream cools. It is called

plasma tail.

iii) Radio-frequency supply

The basic circuit for a RF generator is simple, consisting of a capacitor and inductor in

either a series or parallel configuration. ICP generally require 1-2 kW maximum output at


52

Radio Frequency (RF) to maintain the plasma. The RF generator in the Varian ICP-OES

instrument is based on a 40.68 MHz crystal-controlled oscillator and has automatic

tuning and automatic level control and is rated at 1.7 kW.

iv) Optical spectrometer

A spectrometer consists of three main parts: An emission source, which produces the

spectrum, an optical system, which scatters the spectrum and the device to measure the

emitted lines that processes the result. A common spectrometer optical system consists of

a polychromator or a monochromator, which scatters the spectrum and isolates the

analytical lines of the elements to be analyzed.

v) Detectors and readout systems

In optical emission spectrometry, photo-multipliers are commonly used as detectors.

They are photocell detectors. The incident photons coming from the exit slit liberate

electrons from the photo cathode and the electron flow is then amplified by a set of

dynodes. The final anode current is proportional to the incident photon signal received by

the photo cathode. The measurement dynamic range is very broad, i.e. 1015, and

sensitivity is high, as the dark current is low. These detectors allow the detection of low

intensities emitted by trace elements, as well as strong signals from major elements. They

have very fast response times, typically 1-2 ns for a 10%-90% change in signal. A fatigue

lamp (a small incandescent light source) is often used with photo-multipliers to keep the

temperature of the tube and its associated electronics constant. The fatigue lamp is

switched on when the emission source is off and switched off when the emission source

is on.

3.3.3 ICP process (Liberty p10)

The sample being analysed is introduced into the plasma as a fine droplet aerosol. The

processes, which occur while the aerosol moves up through the plasma, are described in

Figure 3.12.


53

Figure 3.12 ICP process (Moore, 1989)

The hot plasma removes any remaining solvent and causes sample atomization followed

by ionization. In addition to being ionized, sample atoms are excited in the hot plasma, a

phenomenon that is used in ICP emission spectroscopy.

The aerosol droplets introduced to plasma after nebulization is desolvated to solid salt

particles and vaporized to produce gas-phase molecular species. These species

subsequently undergo dissociation to free atoms, namely atomization. With sufficient

energy, these free atoms are excited to higher energy stages and further to higher energy

states of the ions called excitation and ionization. At this condition, most elements emit

light of characteristic wavelengths, which can be measured and used to determine the

concentration. The following reactions are shown the main reaction of the processes

occurred in the plasma (Figure 3.13).


54

Figure 3.13 Main reactions of the processes in plasma (Liberty, 1991)

Where, M is an atom, M+ is an ion and M+* is an excited energy state ion

Light from the different elements is separated into different wavelengths by means of a

grating and is captured by light-sensitive detectors (Figure 3.9). This permits

simultaneous analysis of up to 40 elements and ICP-OES is consequently a multi-element

technique (http://icp-oes.com/sitering.htm).

3.3.4 Optimization of experimental conditions (Monstaser and Golightly, 1992; Liberty, 1992)

An analytical instrument should be operated at the optimum conditions. The optimum

conditions are those that give the best analytical accuracy, the lowest detection limits, the

fastest sample throughput, or the lowest running costs is a matter for the analyst to decide.

From the analytical point of view, accuracy and low detection limits are of paramount

importance although speed and cost certainly cannot be completely ignored. The power o

f the plasma, observation height, plasma gas and nebuliser gas flows are the predominant

parameters in the optimization procedure. Auxiliary gas flow rate plays a secondary role.

a. Element wavelength selection The primary wavelength of an element is typically the first wavelength selected to

provide maximum sensitivity and avoid potential problems such as spectral interferences.

If spectral overlap occurs, an alternative interference-free wavelength can be selected

which have minimal interferences from nearby lines as well as high emission intensity so

that a high sensitivity is achieved. Soft lines are atomic lines of elements, which have a

low to medium ionization potential and give their greatest intensities low in the plasma.


55

Hard lines are atomic or ionic lines, which have greater excitation potentials and their

greatest intensity occurs characteristically at greater observation heights in the plasma.

b. Plasma observation height The intensity of the emission at individual wavelengths varies according to the region in

the plasma that is observed. Observation height is a critical parameter that can

significantly enhance sensitivity. When analyzing high concentration levels it is best to

optimize by intensity, as the background signal will be insignificant when compared to

the large analyte signal. Optimization by signal to background ratio should be used when

analyzing low concentration levels as the background signal becomes significant.

c. Plasma power The plasma power is another critical parameter that can significantly enhance the signal

of selected wavelengths. Higher power settings increase the intensity of hard wavelength.

However, the background is also enhanced at higher power settings. The extent to which

the background is raised depends on the wavelength. Thus it is important to monitor the

effect of increased power settings on background as the increase in gross analyte signal

may be caused by an increase in background levels and not net analyte signal. The

intensity of soft lines on the other hand, generally do not increase significantly with

higher powers because they correspond to low energy transitions and therefore are

already near their maximum sensitivity at lower powers.

d. Plasma gas flow rate

Increasing the plasma gas increases the size of the hot, analytical zone within the plasma.

Thus the intensity of hard wavelengths increases with high plasma gas flow rates as

higher energy transitions can occur due to an increase in available energy.

e. Auxiliary gas flow rate

The main effect of the auxiliary gas is to lift the plasma away from the torch. Higher

auxiliary gas flows can reduce band structures.


56

f. Nebuliser pressure

The intensity of hard wavelengths increased significantly by intensity reducing the

nebuliser pressure. This occurs because the reduced pressure allows an increase in the

residence time of the analyte in the plasma. Thus the analyte is given a longer time to

acquire the energy for high-energy transitions. However, when the nebuliser pressure is

too low instability occurs in the nebuliser operation.

3.4 Inductively Coupled Plasma Mass Spectroscopy (ICP-MS) 3.4.1 Introduction Inductively Coupled Plasma Mass Spectrometry (ICP-MS) is a very powerful tool for

trace elemental analysis. It was developed in the 1980’s and has grown rapidly to be one

of the most important techniques for elemental analysis because of the following benefits:

(Houk, 1986; Monstaser, 1992; Vela et al., 1993)

• A very low detection limit (ng/L-mg/L)

• A very simple and unique mass spectrum of the elements

• The ability to measure elemental isotope ratios

• A good precision and accuracy

• Multielement determination

ICP-MS has been used widely over the years, finding applications in a number of

different fields including drinking water, wastewater, natural water

systems/hydrogeology, geology and soil science, mining/metallurgy, food sciences, and

medicine. 3.4.2 Instrumentation and process of the ICP-MS

In principle, the ICP-MS technology is very similar to the ICP-OES with regard to the

process that samples are decomposed to neutral elements in high temperature argon

plasma. However, unlike ICP-OES, ICP-MS accomplished the elemental determination

by counting the number of ions at a certain mass of the element instead of the light


57

emitted by the element. The interface system is of unique importance in ICP-MS and this

subject is treated extensively in this section.

An ICP-MS can be thought of as four main processes, including sample introduction and

aerosol generation, ionization by an argon plasma source, mass discrimination, and the

detection system. Since the first two processes were already discussed in Section 3.3.2.a

and 3.3.2.b. The mass discrimination and the detection system are considered here. The

schematic below illustrates this sequence of processes.

Figure 3.14 Schematic of an ICP-MS system (Skoog et al. 1998)

a. Mass discrimination

i) ICP-MS interface-sampling ions

The entire mass spectrometer must be kept in a vacuum so that the ions are free to move

without collisions with air molecules. Since atomization/ionization occurs at atmospheric

pressure, a pumping system is needed to continuously pull a vacuum inside the

spectrometer. The Figure 3.15 shows the Mass Spectroscopy interface.


58

Figure 3.15 ICP-MS interface (http://las.perkinelmer.com/content/ApplicationNotes/D-6355A.pdf)

Ions flow through a small orifice, approximately 1 millimeter in diameter, into a pumped

vacuum system. Here a supersonic jet forms and the sample ions are passed into the MS

system at high speeds, expanding in the vacuum system (Jarvis et al., 1992).

ii) The lens system-focusing ions

The ion lens is positioned immediately behind the interface (Figure 3.15) to focus ions

into the quadruple region by keeping the ion beam from diverging. Since the charge on

the lens is the same as the charge on the ions, the ions are repelled back toward each

other to form a focused ion beam.

iii) Mass analyzer-separating ions

The mass spectrometer separates the singly charged ions from each other by mass. The

quadrupole mass analyzers the type most commonly used in analytical instrumentation

(Figure 3.16).

Figure 3.16 Quadrupole mass analyzer (http://www.missouri.edu/~rjse10/icpms.htm , Perkin,)


59

It is made up of four metal rods aligned in a parallel diamond pattern. A combined DC

and AC electrical potential is applied to the rods with opposite rods having a net negative

or positive potential. Ions enter into the path between all of the rods. When the DC and

AC voltages are set to certain values only one particular ion is able to continue on a path

between the rods and the others, which are unstable at this voltage combination, are

forced out of this path. This ion will have a specific mass-to-charge (m/z) ratio. Many

combinations of voltages are chosen which allows an array of different m/z ratio ions to

be detected.

b. Detection system – counting ions

Ions leaving the quadrupole strike the active surface of the detector, known as a dynode,

and generate a measurable electronic signal. The detection system operates on the

principle of electron multiplication. The drawing below is an ICP-MS detector.

Figure 3.17 ICP-MS detector

An ion striking the first dynode releases electrons, which accelerate towards and into the

dynode beneath which in turn releases, more electrons. This cascading of electrons

continues until a measurable pulse is created. By counting the pulses generated by the

detector, the system counts the ions that hit the first dynode.

Evaluation of instrumental methods for low-level F- determination in laboratory test samples ______________________________________________________________________________________

60

Chapter 4

Evaluation of instrumental methods for low-level Fluoride (F-) determination in laboratory test samples

4.1 Objective

Fluoride (F-) in water can be determined in a number of ways, using instrumental

methods such as potentiometry with Ion-Selective Electrodes (ISE) (Richard 1969;

Cardwell et al., 1994; Wang, 1995; McCaffrey, 1994; Lopes da Conceição, 2000; van

Staden, 2000; US EPA method 13B), spectrophotometry (Bellack 1958; Crosby et al.,

1968; Cabello-Tomas and west, 1969; Wada et al. 1985; Yuchi et al., 1995; Khalifa and

Hafez, 1998; Oszwaldowski et al., 1998; Faraj-Zadeh and Kalhor, 2001; Nishimoto et al.,

2001; Garrido et al., 2002; Arancibia et al., 2004; US EPA method 13A), inductively

coupled plasma-mass spectrometry (Bayón et al., 1999; Okamoto, 2001), capillary

conductometry (Buffle et al., 1985), complexometry (Pickering, 1986; Saha, 1993), gas

diffusion (Cardwell et al., 1994), capillary ion analysis (Bondoux and Jones, 1995; Saad

et al., 1998), and ion chromatography (IC) (Jones, 1992; McCarthy, 1994; Hirayama et al.,

1996; Moskvin et al., 1998; Jackson et al., 2000; Jackson, 2001; Application note

133,135,140 and 154).

For the determination of F-, F- Ion Selective Electrode (F-ISE) has been widely used due

to its simplicity and short analysis time. The Ion Chromatography (IC) technique has also

come into prominence in terms of selectivity and sensitivity. Although the SPANDS

method achieved very high sensitivity, this method was displaced by the ISE due to its

main disadvantages; poor selectivity, labour-intensiveness and being time consuming due

to the distillation step (Crosby et al., 1968; Mac Leod and Crist, 1973; Oszwaldowski et

al., 1998). This method was not included in this comparative study. F-ISE and IC

methods have been used for achieving the objectives given in Chapter 1.


61

4.2 Ion Selective Electrode (ISE) method 4.2.1 Instrumentation

The F- concentration was measured using a F-ISE (9404 sc Orion Research Inc., MA,

USA) in combination with an Ag/AgCl single junction Reference electrode (Orion Model

90-01) connected to the read out device, ORION 960 Auto Chemistry System (Orion

Research Inc., MA, USA).

The Technical Specifications for the F-ISE is shown in Table 4.1.

Table 4.1 Technical Specifications for the F-ISE

Parameter Specification Optimum pH Between 5 to 6

Optimum Temperature Optimum temperature: 25°C Time for stable reading after

immersion 2 to 3 minutes

Potential drift (in 1000 ppm) < 3 mV/day (8 hours)

Recommended reference electrode Single Junction AgCl (ELIT 001) or Double Junction lithium acetate (ELIT 003)

Electrode slope at 25°C 54±5 mV/decade

4.2.2 Standards and reagent solutions

a. Stock F- solution and F- standards

A 1000 mg F-/L sodium F- stock solution was prepared by dissolving 0.221 g NaF (pro

analysi, MERCK) in a 100 mL polystyrene volumetric flask with deionised water.

Standards at the required concentration were prepared by appropriate dilution of the stock

solution.

b. Total Ionic Strength Adjustment Buffers (TISAB) for F-ISE

As discussed in Section 3.1.4.c, the TISAB is an important reagent for the determination

of F- using ISE. There are several different recipes for TISAB. The preparation procedure

of each TISAB solution used in this study is described below.


62

i) TISAB III

TISAB III is commercially available from Thermo Orion (Orion Research Inc., Beverly,

MA, USA, Cat. No.940911). It contains CDTA (trans-1,2-cyclohexylendinitrilo) tetra-

acetic acid) and ammonium acetate (CH3COONH4) buffer at pH 5.5. The recommended

volume ratio between TISAB and test solution was 1:10. 20 mL of test solution and 2 mL

of TISAB III mixed in a plastic beaker was used in this study.

ii) TISAB IV

It contains TRIS buffer for high levels of Al interference at pH 8.5. 84 mL concentrated

HCl (36-38%), 242g tris(hydroxymethyl)aminomethane, and 230 g sodium tartrate

(Na2C4H4O6 •2H2O), were added to 500 mL deionised water, allowed to dissolve, and

then made up to 1 L in a volumetric flask with deionised water. A recommended volume

ratio between TISAB and test solution, written in the manual (Orion, 1982) was 1:1. In

this study, 10 mL of test solution and 10 mL of TISAB IV mixed in a plastic beaker.

iii) Low Level TISAB (LLT)

This is used when measuring samples containing less than 0.4 mg/L F- and no F-

complexing agents, such as Fe or Al, are present. 57 mL acetic acid and 58 g sodium

chloride were added in 500 mL distilled water and the pH was adjusted to 5.5 by adding

drops of 5 M NaOH after cooling the solution, then made up to 1 L in a volumetric flask

with deionised water. The recommended volume ratio between TISAB and test solution

was 1:1. 10 mL of test solution and 10 mL of Low Level TISAB mixed in a plastic

beaker was used in this study.

4.2.3 Results and discussion

a. Electrode drift

One of the main problems in ISE measurement is the drift in electrode potential during a

sequence of measurements. 0.02, 0.1, 1, 2 mg/L spiked samples of F- in deionised water

were prepared in 100 mL polystyrene volumetric flasks. 20 mL of sample and 2 mL of

TISAB III were added into a 50 mL polystyrene beaker for measurement. The uncertainty,

due to fluctuating potential values during the first a few seconds after immersion of the


63

electrode, was controlled by the following procedure. The reference electrode was placed

into the beaker first, followed after 3 minutes by the F- electrode. The potential was

recorded from 3 minutes after immersion of the F- electrode for 1 hour and drift values

were calculated by subtraction. These experiments were repeated 3 times for each sample.

The electrodes were rinsed by spraying with a jet of deionised water and gently dabbed

dry with a soft tissue before measurement.

A summary of results, which shows the average drift (mV/h) and standard deviation

obtained in this experiment, is given in Table 4.2.

Table 4.2 Potential Drift (mV) at each F- concentration vs. time

[F] Known 0.02 mg/L 0.1 mg/L 1 mg/L 2 mg/L Time (min) P*1 P2 P3 P1 P2 P3 P1 P2 P3 P1 P2 P3

3 154.6 153.4 159.4 139.5 138.2 138.8 90.4 90.6 89 73.7 73.9 74.25 154.5 153.4 159.4 139.4 139.1 138.7 90.3 90.5 88.9 73.7 73.9 74.17 154.5 153.3 159.4 139.2 139 138.6 90.2 90.4 - 73.7 73.8 74.1

10 154.7 153.3 159.0 139.1 138.9 138.5 90.2 90.4 - 73.7 73.7 74.115 154.8 153.3 158.8 139 138.8 138.4 90.2 90.3 88.7 73.7 73.6 74.030 154.6 153.4 158.3 138.6 138.5 138.2 90.1 90.2 88.4 73.6 73.3 73.860 153.7 152.5 157.5 137.9 138 137.8 89.4 89.6 87.9 73.1 73 73.5

Drift 1.1 0.9 1.9 1.6 1.1 1 1.0 1.0 1.1 0.6 0.9 0.7 Average

Drift (mV/h) 1.30 ± 0.53 1.23± 0.32 1.03± 0.06 0.73± 0.15

P* = Potential (mV)

These results show that the average drift values of the F- electrode are bigger when the

concentrations of F- are lower. This trend is evident in the standard deviations. The

standard deviations for the 0.02 and 0.1 mg/L solutions are larger than those for 1 and 2

mg/L F- solutions. This means that lower concentrations of F- are more severely

influenced by drift and show poor precision. Therefore, steps must be taken to minimize

the drift for the determination of F- at levels less than 1 mg/L. Further investigation was

done in order to determine the best procedure to minimize electrode drift and decrease

response time at the 0.02 to 1 mg/L F- concentration level.

b. Investigation of electrode response

The objective of this study was establishing the most effective way to minimize the

electrode drift. 0.02, 0.1 mg/L spiked samples of F- in deionised water were prepared


64

using 100 mL of polystyrene volumetric flasks. 20 mL test solution and 2 mL TISAB III

(10:1 recommended volume ratio, section 4.2.2.b) and in the case of Low Level TISAB,

10mL test solution and 10mL Low Level TISAB (1:1 recommended volume ratio, see

Section 4.2.2.b), were added into the polystyrene beaker (50mL). The potential was

measured for an hour.

The following solutions were made for these experiments.

• Electrode storage solution

This solution was made up for storing the electrodes while the electrodes were not

used. The recommended concentration of the storage solution), was 0.01M NaF

(100mg/L) (Thermo Orion, Electrode Manual).

• Pre-conditioning solution

This solution, consisting of 0.1 mg/L F-, was made for allowing the electrodes to be

equilibrated in the low-level F- environment before measurement. The procedure was

expected to reduce the electrode memory effect (see Section 3.1.2.f).

Four different procedures were investigated as described below.

i) Procedure 1

Both electrodes, reference electrode and F-ISE, were immersed in the storage solution

(100 mg/L F-) over night in between measurements. The potential of the 0.1 mg/L F- test

solution was measured, without preconditioning, for 1 hour using TISAB III on three

consecutive days. These results are shown graphically in Figure 4.1.

Figure 4.1 Electrode responses: TISAB III at 0.1 mg/L F--

105

110

115

120

125

0 10 20 30 40 50 60

Time (min)

p(m

V)

P[mV] (1st run) P[mV](2nd run) P[mV] (3rd run)


65

The drift in each run was quite different, although the same electrodes and solutions were

used. The first part of the graph, between 0 to 5 minutes shows a similar trend in each run.

In a subsequent experiment the intra-day electrode response was investigated. This is

discussed in the next paragraph.

ii) Procedure 2

Both electrodes (reference electrode and F-ISE) were immersed in the storage solution

(100 mg/L F-) over night. Before the first potential measurement of the test solutions, the

electrodes were put into the pre-conditioning solution (0.1 mg/L) for 20 minutes. The test

solution, 0.1 mg/L, was then measured for 1 hour using TISAB III the potential recorded.

After finishing the first measurement, the electrodes were immersed into the storage

solution again for 50 minutes to allow the electrodes to be equilibrated under similar

conditions as before (Procedure 1). The procedure was then repeated twice on the same

day. The results are summarized in Figure 4.2.

Figure 4.2 Electrode response: TISAB III at 0.1 mg/L F-

100

110

120

130

140

150

160

0 10 20 30 40 50 60

Time(min)

Pote

ntia

l(mV)

P1 P2 P3

This graph shows that the response during the same day is reproducible. Compared to the

Procedure 1 (Figure 4.1), procedure 2 yielded more reproducible results (Figure 4.2). The

response time to obtain a stable reading was, however, not improved as was expected

using a preconditioning solution.


66

iii) Procedure 3

Preconditioning was used between every single measurement in Procedure 2, but this was

not so effective to reduce drift, as was expected. Thus, in Procedure 3, preconditioning

was used on the test solutions (0.02 mg/L and 0.1 mg/L) only for the first measurement in

a series. The results are given in Figure 4.3.

Figure 4.3 Electrode response: TISAB III at 0.02 mg/L and 0.1 mg/L F-

100

110

120

130

140

0 10 20 30 40 50 60Time(min)

Pote

ntia

l(mV)

P1(0.1mg/l) P2(0.1mg/l) P3(0.1mg/l)P1(0.02mg/l) P2(0.02mg/l) P3(0.02mg/l)

Each graph of the first run displays a longer equilibration time and a bigger change of

potential. This means that the electrode takes a long time to adjust after exposure to high

F- concentrations. The potential of the first run varied from low to high value, while the

potential of the second and the third run varied from high to low value. This was because

the electrode was not exposed to high concentration of the storage solution between

measurements. After the first measurement, the electrodes were rinsed by spraying with a

jet of deionised water, which did not contain any F-. The initial high potential values in

the second run were caused by the exposure of the ion selective membrane to the non-F-

environment of deionised water.

iv) Procedure 4

F- electrode become sluggish over time. Brushing the membrane with F- toothpaste can

restore the performance of the electrodes (Thermo Orion, Application note). This

experiment was done after F-ISE had been used for 8 months and 3 months after the

experiment of Procedure 2.


67

• Cleaning process

The cleaning process of F-ISE is described as follows:

A small amount of F- toothpaste (Mentadent P Herbal toothpaste, F- System, Lever

Pond’s (Pty) Ltd., Durban) was squeezed from the tube on to a toothbrush. The F-ISE

membrane was then brushed softly using this toothbrush and then the electrode was

rinsed by spraying with a jet of deionised water until there was no toothpaste on the

membrane and gently dabbed dry with a soft tissue.

After cleaning the F-ISE, it was stored dry in air. The reference electrode was stored in

deionised water. The test solutions (0.02 mg/L and 0.1 mg/L) were measured 3 times

without any preconditioning or exposure to the storage solution using TISAB III. The

results using this method are given in Figure 4.4.

Figure 4.4 Electrode response: TISAB III at 0.02 mg/L and 0.1 mg/L F-

130

140

150

160

170

180

190

0 10 20 30 40 50 60Time(min)

Pote

ntia

l(mV)

P1(0.1mg/l) P2(0.1mg/l) P3(0.1mg/l)

P1(0.02mg/l) P2(0.02mg/l) P3(0.02mg/l)

This graph shows that electrode response was very stable. The potentials measured during

the third run at 0.1 mg/L displayed the most stable potential reading. At lower

concentration of the test solution, 0.02 mg/L, the equilibration time was longer than that

for 0.1 mg/L but shorter than for the other methods. The drift in potential over 1 hour was

small. This indicates that cleaning of F-ISE is a very efficient way for minimizing the

electrode drift and improving equilibration time. Procedure 4 thus enabled a minimization

of drift.

v) Effect of electrode cleaning

This experiment was done to compare with the data obtained before cleaning the


68

electrode with that after cleaning. The cleaning was done for the first time after the

electrode has been in use for 8 months. A series of F- test solutions, 0.1, 0.15, 0.2, 0.3, 1,

3 mg/L, was prepared by appropriate dilution of the stock solution. Potential

measurements were done on two consecutive days using Procedure 4. The results are

given in Figure 4.5.

Figure 4.5 Comparison of electrode response before and after cleaning

The graph, obtained before cleaning electrode, clearly shows that F-ISE can become

sluggish over time. It took a longer time to reach equilibrium in each test solution and in

the case of the test solutions containing 0.1 mg/L and 0.15 mg/L F-, the drift was

continuous and no stable signal was obtained. The results after cleaning the electrode

showed shorter equilibration times and a significant decrease in drift. The overall time for

measurement was about 40 min compared to 1h 25 min before cleaning.

vi) Effect of electrode age

The potential of a 0.1 mg/L test solution was measured over a period of 1 h using

Procedure 3 and Low Level TISAB with two electrodes; an old one that had been used

for 2 years and a new one in use for 3 months. The results are given in Figure 4.6.


69

Figure 4.6 Comparison electrode responses with respect to electrode age

80

90

100

110

120

130

140

150

0 10 20 30 40 50 60 70Time (min)

Pote

ntia

l (m

V)

P 1 (old) P 2 (old) P 3 (old)P 1 (new) P 2 (new) P 3 (new)

The readings obtained with the old electrode were inconsistent and the drift in the 1st and

3rd run was so severe that no stability was reached. The results obtained with new

electrode, however, showed stable readings, after 18 min for first run, 5 min for second

and 10 min. This result emphasizes the importance of electrode condition on low-level F-

determination.

c. TISAB and calibration study

Accurate calibration is an important factor in the determination of analyte concentration

in any analytical technique. In the case of ISE methods calibration is carried out by

immersing the electrodes in a series of solutions of known concentration and by plotting a

graph of the mV reading versus the log of the activity. According to the Nernst equation,

described in Section 3.1.2.b, this should give a straight line over the whole linear

concentration range. The objectives of the investigation in this section were:

• To study calibration techniques and the effect of different TISAB solutions on the

accuracy of F- determinations

• To investigate multi-standard calibration for low-level F- determinations by

comparing the efficiency of the two TISAB solutions, TISAB III and Low Level

TISAB

• To apply the optimized method to the analysis of low-level F- sample analysis

using deionised water and tap water


70

• To investigate the effect of regular electrode cleaning on low-level F-

determination.

i) High-level F- calibration with different TISAB solutions

The calibration graphs, given in Figure 4.7 for the F- concentration range 0.02 mg/L to

1000 mg/L, were obtained using three different TISAB solutions, Low Level TISAB,

TISAB III and TISAB IV. The concentrations in the standard series were: 0.02, 0.05, 0.1,

1, 10, 50, 100 and 1000 mg/L. 20 mL of standard solution without TISAB, 20 mL of

standards and 2 mL of TISAB III, 10mL of standard and 10 mL of TISAB IV and Low

Level TISAB were prepared in 50 mL plastic beakers The potential of standards was

measured from low to high concentration and the meter reading was taken after a

constant value has been attained (drift < 0.1 mV/min). The results obtained from this

experiment are presented in Table 4.3.

Table 4.3 Potential with different TISAB

TISAB type No TISAB TISAB III Low Level TISAB TISAB IV [F] (mg/L) P (mV) P (mV) P (mV) P1 (mV) P2 (mV)

0.02 - 141.3 170.0 60.3 66.0 0.05 - 141.5 168.0 64.3 70.0 0.1 136 135.5 159.2 68.1 71.2 1 81.3 94.0 109.8 64.3 60.3 10 24.3 34.3 50.6 40.0 37.5 50 -16.2 -6.4 9.8 6.3 5.8

100 -33.4 -24.3 -7.5 -11.1 -10.3 1000 -89.5 -82.4 -65.6 -68.6 -67.5

The potential of the 0.02 and 0.05 mg/L F- standard solution, without TISAB, could not

be determined because the potential of these two solutions were consistently decreasing

over a period of an hour. This drift was probably caused by the low ionic strength of 0.02

and 0.05 mg/L F- standards. A stable reading at these levels of F- was obtained after

adding TISAB that increased the ionic strength.

The results are summarized in Table 4.3 and shown in Figure 4.7.


71

Figure 4.7 Potential with different types of TISAB

-100

-60

-20

20

60

100

140

1800.01 0.1 1 10 100 1000

[F] logarism scale (mg/l)

P (m

V)

P (mV): No TISABP (mV): T IIIP (mV): LLTP (mV): T IV(1)P (mV): T IV(2)

TIII: TISAB III, LLT: Low Level TISAB, T IV: TISAB IV

As mentioned in Section 3.1.2.b, an electrode is said to have a Nernstian response, when

the potential difference versus the logarithm of the ionic activity of a given species shows

linear relationship. All graphs, except the graphs obtained using TISAB IV, showed the

linear relationship above 0.1 mg/L F- concentration. The graph obtained without TISAB

showed a straight line above 0.1mg/L F-. Although the electrode response showed the

linear relationship, the ISE measurement without TISAB was not practical since it could

not yield stable potentials at lower than 0.1 mg/L F-.

The standard series with TISAB IV was measured twice because the graph showed

excessive curvature at concentrations below 10 mg/L. The duplicate measurement

produced the same result. This meant that TISAB IV could not be used for low-level F-

determination. Orion recommends the use of TISAB IV for the purpose of de-

complexation of Al, but it has been considered unsuitable for the accurate measurement


72

of low-level F- concentration (Orion, 1982). This experimental result can be proof of this

statement. It is well known that at low F- levels, the ISE response at pH>8 can be in error

because of interference by hydroxide ions (Orion, 1982). The TISAB IV contains TRIS

buffer for high levels of Al interference at pH 8.5 (see Section 4.2.2). Because of this

high pH value of TISAB IV, the erroneous results obtained for the low-level F-

determination.

The graphs for Low Level TISAB and TISAB III showed straight lines above 0.1mg/L F-

concentration. This meant that both Low Level TISAB and TISAB III could be used for

low-level F- determination.

ii) Low-level F- calibration and sample measurement: TISAB III

Since the focus of this work was on low-level F- determination calibration in the

concentration range 0.01 to 3 mg/L was investigated in detail.

• Low-level calibration

A two-point calibration is sufficient when the measurement is done in the linear range of

the calibration curve. Three or more standards are, however, recommended in order to

confirm the linearity and to detect any errors in diluting the standards. If the curve is non-

linear more standards are needed in order to define the curve in the non-linear range. The

slope of the calibration graph is the mV response per decade of concentration change.

This is typically around 54 mV/decade for monovalent ions and 27 for divalent ions and

will have a negative value for negative ions (Nico, 2000). The calibration graph (Figure

4.7) at low-level concentration shows the curved portion where the electrode response

becomes progressively less as the concentration reduces. In this section, multi-point

calibrations were repeated three times using TISAB III. TISAB III, rather than Low

Level TISAB, was chosen initially since TISAB III contains a de-complexation agent and

is more practical in natural samples that might contain interfering elements such as Al

and Fe. Calibration graphs were obtained by measuring the potential of three different

sets of F- standard solutions in order from low to high concentration. The meter reading

was taken after a constant value has been attained (Drift < 0.1 mV/min).


73

a) Multipoint standards 1: 0.05, 0.1, 0.2, 0.3, 0.4, 0.5, 1 and 3mg/L F-

b) Multipoint standards 2: 0.05, 0.06, 0.08, 0.1, 0.2, 0.3, 1 and 3mg/L F-

c) Multipoint standards 3: 0.05, 0.07, 0.1, 0.15, 0.2, 0.3, 1 and 3 mg/L F-

The calibration graph in each run is given in Figure 4.8.

Figure 4.8 Multi-calibration graph using TISAB III

3

1

0.5

0.40.3

0.2

0.1

0.05

0.06

0.080.15

0.07

50

60

70

80

90

100

110

120

130

140

150

1600.01 0.1 1 10


P (m

V)

P1 (mV)

P2 (mV)

P3 (mV)

The graph of the first run, which has many points between 0.1mg/L to 1mg/L F-, clearly

displayed the linear relationship above 0.2mg/L F- and non-linear relationship below 0.2

mg/L. But it was difficult to determine the lowest point where the calibration line is still

linear, since there was no standard point between 0.1 and 0.2 mg/L and between 0.05 and

0.1 mg/L. Therefore, it was necessary to measure more standards in the regions between

0.05 to 0.2mg/L. The results are given in Figure 4.8, for determination of the lowest

linear point.


74

The graph of the second run seems to show linearity above 0.05 mg/L and the third one

above 0.1 mg/L. To see the linearity more clearly, the analytical parameters, slope and R2,

for the second and third run, according to different range of calibration standards points,

is given in Table 4.4.

Table 4.4 Analytical parameter (slope and R2)

Experimental running

Standards Range (mg/L) Slope Y-intercept R2

0.05-3 -48.156 84.995 0.9940 0.06-3 -49.039 84.890 0.9951 0.08-3 -50.014 84.830 0.9959 0.1-3 -51.038 84.850 0.9963

Second run

0.2-3 -53.323 85.022 0.9993 0.05-3 -49.890 88.240 0.9937 0.07-3 -50.670 88.118 0.9934 0.1-3 -52.447 87.997 0.9972 0.15-3 -53.523 88.028 0.9980

Third run

0.2-3 -54.742 88.208 0.9991

The slope, which must ideally be within 54 ± 5 mV/decade is improved, from –48.156 to

–53.323 in second run and from –49.890 to –54.742 in the third run, as the starting

position of the standard series were taken at higher concentration values. It therefore

seemed reasonable to use all the standards in the range except the 0.05-3 in the second

run, where the slope was –48.156 and outside the ideal range.

The value of R2 shows the linearity of each graph. Although the graph of the second run

seems to be straight above 0.05 mg/L, the correlation coefficient for this graph from 0.05

mg/L is only 0.9940, which is not very good. R2 is improved, from 0.9940 to 0.9993 in

the second run and from 0.9937 to 0.9991 in the third run, as the beginning point of

calibration standard is higher. It is necessary to have R2 at least 0.999 for confirmation of

linearity. When 0.15 mg/L standard was included for calibration in the third run, R2 was

less than 0.999, which could not give an assurance of linearity. Including the 0.1 mg/L

standard for the calibration in both runs made correlation worse. The value of R2 > 0.999

only could be obtained for the range from 0.2 to 3 mg/L. This means that the lowest point


75

that can show good linearity is 0.2 mg/L. This fact now presents a problem for

determination of low-level F- samples in the concentration range below 0.2 mg/L. The

influence of the calibration factors, such as slope and R2, were investigated by duplicate

measurement of samples with known concentration.

• Measurement of samples with known concentration in deionised water matrix

Table 4.5 Influence of calibration parameters: linear regression (second run)

Calibration range 0.05-3 mg/L 0.1-3 mg/L 0.2-3 mg/L

Function Y=-48.156x+84.995 Y=-51.038x+84.85 Y=-53.323x+85.022 R2 0.9940 0.9963 0.9993

Known [F] (mg/L)

Calculated[F] (mg/L)

SD (N=2)

% Error


SD (N=2)

% Error

Calculated [F] (mg/L)

SD (N=2)

% Error

0.07 0.062 0.005 -10.9 0.072 0.005 3.6 0.082 0.006 16.70.09 0.076 0.006 -15.2 0.088 0.007 -2.6 0.098 0.007 8.9 0.12 0.093 0.001 -22.2 0.106 0.001 -11.7 0.118 0.001 -2.0 0.15 0.114 0.001 -23.8 0.128 0.001 -14.4 0.141 0.001 -5.8 0.18 0.137 0.000 -23.8 0.152 0.000 -15.3 0.166 0.001 -7.5 0.4 0.343 0.005 -14.4 0.362 0.005 -9.6 0.380 0.005 -4.9 0.5 0.442 0.010 -11.5 0.460 0.010 -7.9 0.479 0.010 -4.1 0.8 0.761 0.010 -4.8 0.768 0.010 -4.0 0.783 0.010 -2.2 1.5 1.567 0.000 4.5 1.518 0.000 1.2 1.502 0.000 0.1 2.0 2.123 0.007 6.2 2.021 0.006 1.1 1.976 0.006 -1.2 2.5 2.775 0.009 11.0 2.602 0.008 4.1 2.517 0.008 0.7

Table 4.5 shows the importance of using the proper calibration function for accurate

determination of sample concentration. The results, obtained by using the function for the

range 0.05-3 where the slope was outside the ideal range and the linearity < 0.999, show

large errors. The results for the calibration line based on the range 0.1-3 mg/L and 0.2-3

mg/L are reasonable, but not very good. An additional problem was that sample

concentrations, less than 0.1 mg/L for the calibration range 0.1-3 mg/L and less than 0.2

mg/L for calibration range from 0.2-3 mg/L, could not be determined accurately since

these were out of calibration range. The results in Table 4.6, obtained from the third run,

shows as similar trend as second run.


76

Table 4.6 Influence of calibration parameters: linear regression (third run)

Calibration range 0.05-3 mg/L 0.1-3 mg/L 0.2-3 mg/L


Known [F] (mg/L)


SD (N=2)

% Error


SD (N=2)

% Error


SD (N=2) % Error

0.06 0.052 0.003 -13.0 0.065 0.003 7.9 0.073 0.003 22.1 0.09 0.073 0.000 -19.4 0.088 0.000 -2.6 0.098 0.000 8.8 0.12 0.095 0.003 -21.2 0.112 0.003 -6.8 0.124 0.004 3.1 0.18 0.148 0.003 -17.9 0.169 0.004 -6.3 0.183 0.004 1.8 0.25 0.211 0.002 -15.5 0.234 0.002 -6.3 0.251 0.002 0.5 0.4 0.349 0.001 -12.8 0.372 0.001 -7.1 0.391 0.001 -2.3 0.5 0.452 0.006 -9.7 0.471 0.004 -5.8 0.490 0.004 -1.9 0.8 0.788 0.016 -1.5 0.785 0.015 -1.8 0.801 0.014 0.1 1.5 1.612 0.000 7.5 1.517 0.000 1.2 1.504 0.000 0.3 2 2.312 0.008 15.6 2.114 0.007 5.7 2.067 0.006 3.3

Therefore a way for the calculation of the sample concentrations below 0.2 mg/L was

investigated, using non-linear regression functions. A summary of the results is given in

Table 4.7 and 4.8.

Table 4.7 F- determination in curvature region (< 0.2 mg/L, second run)

Calibration range Low level between 0.05-0.2 mg/L for measuring < 0.2mg/L sample

Exponential function Polynomial function Linear regression Function

Y=99.279 e-0.2942x Y=704.1x2-331.26x+159.68 Y=-39.047x+94.484 R2 0.9964 0.9999 0.9984

Known [F] (mg/L)


SD (N=2)

% Error


SD (N=2)

% Error


SD (N=2)

% Error

0.07 0.057 0.005 -18.6 0.057 0.006 -18.6 0.057 0.005 -18.30.09 0.073 0.007 -18.9 0.075 0.008 -16.7 0.073 0.008 -18.60.12 0.092 0.001 -23.3 0.095 0.001 -20.8 0.094 0.001 -21.60.15 0.119 0.002 -20.7 0.118 0.001 -21.3 0.121 0.002 -19.60.18 0.150 0.001 -16.7 0.144 0.001 -20.0 0.151 0.001 -16.1


77

Table 4.8 F- determination in curvature region (< 0.2 mg/L, third run)

Calibration range Low level between 0.05-0.2 mg/L for measuring < 0.2mg/L sample

Exponential function Polynomial function Linear regression Function

Y=107.09 e-0.2423x Y=148.05x2-169.36x+153.44 Y=-31.118x+104.41

R2 0.9609 0.9933 0.9982 Known

[F] (mg/L) Calculated [F] (mg/L)

SD (N=2)

% Error


SD(N=2)

%Error


SD (N=2)

% Error

0.06 0.040 0.003 -33.6 0.019 0.006 -69.1 0.033 0.003 -44.30.09 0.062 0.000 -30.9 0.062 0.000 -30.7 0.056 0.000 -38.20.12 0.091 0.004 -24.6 0.101 0.005 -16.2 0.084 0.004 -30.00.18 0.176 0.006 -2.0 0.173 0.004 -3.8 0.168 0.006 -6.9

All the results in both tables are very inaccurate. It meant that other functions, such as

exponential and polynomial, should not be used for determination of F- below 0.2 mg/L

concentration. Since the other functions could not be the solution for the determination of

the F- below 0.2 mg/L, another way, cleaning the electrode, was tried to solve this matter.

• Comparison with regard to electrode cleaning

In this section the effect of the electrode cleaning, referred in Section 4.2.3.b, on

calibration line is discussed. The potential of each standard was measured after cleaning

the electrode and the results together with comparative results before cleaning are shown

in Table 4.9. The calibration standards were chosen as the second run, excluding 0.3

mg/L point.

Table 4.9 Electrode slope with regard to electrode cleaning

Status Before cleaning (second run) After cleaning [F] (mg/L) Log F P (mV) Slope P (mV) Slope

0.05 -1.301 144.8 - 154.0 - 0.06 -1.222 142.4 -30.3 150.5 -44.2 0.08 -1.097 137.8 -36.8 145.0 -44.0 0.1 -1.000 133.5 -44.4 140.3 -48.5 0.2 -0.699 121.6 -39.5 124.7 -51.8 1 0.000 85.9 -51.1 85.9 -55.5 3 0.477 59.0 -56.4 58.6 -57.2

The slope calculation was done by this equation: Slope = ][FLog

Potential∆∆

This table clearly shows that the slope was very much improved after cleaning the

electrode. This proves that electrode cleaning can restore the efficiency of electrode. The


78

calibration graph, before cleaning and after cleaning the electrode, is shown in Figure 4.9.

Figure 4.9 Comparison of calibration line in related to electrode cleaning

50

60

70

80

90

100

110

120

130

140

150

1600.01 0.1 1 10


P (m

V)

Before cleaningAfter cleaning

The graph after electrode cleaning clearly shows a better linearity than that obtained

before cleaning. The calibration function and the calculated F- values are given in Table

4.10.

Table 4.10 Sample measurement after cleaning the electrode

Calibration range 0.05-3 ppm 0.08-3 ppm 0.1-3 ppm


Known [F] (mg/L)


SD (N=2)

% Error


SD (N=2)

% Error


SD (N=2)

% Error

0.07 0.074 0.003 5.7 0.078 0.003 11.10 0.079 0.003 12.9 0.09 0.090 0.004 0.1 0.094 0.004 4.8 0.096 0.004 6.7

0.15 0.153 0.000 1.7 0.158 0.000 5.4 0.161 0.000 7.3

0.5 0.482 0.017 -3.6 0.489 0.017 -2.2 0.493 0.017 -1.4

0.8 0.788 0.005 -1.5 0.791 0.005 -1.1 0.794 0.005 -0.8

1.5 1.523 0.014 1.6 1.510 0.013 0.7 1.509 0.013 0.6

2 2.089 0.006 4.5 2.059 0.006 3.0 2.052 0.006 2.6


79

Seven samples with known F- concentration were measured and their concentrations

calculated based on the different calibration ranges: 0.05-3mg/L, 0.08-3mg/L and 0.1-3

mg/L. Compared to the results in Table 4.6, the linearity was very much improved after

electrode cleaning. The values for R2, were 0.9990, 0.9997 and 0.9998, respectively, in

each calibration range, and were all reasonable good. It means that even the very low

concentration 0.05 mg/L could be included in the calibration function. The slope in each

range, -53.937, -54.989 and –55.411, was and within the ideal range. All calculated F-

values based on these calibration curves were very accurate. These results emphasize the

importance of using a reliable calibration function for the low level F- determination and

the crucial role of electrode conditioning and cleaning in attaining accurate analytical

results.

iii) Low-level F- calibration and sample measurement: Low Level TISAB

• Low-Level calibration

The same calibration experiments were done using Low-Level TISAB instead of TISAB

III and were done after electrode cleaning. The calibration graphs, using TISAB III and

Low Level TISAB, are shown in Figure 4.10 for comparison of TISAB efficiency at low-

level F- concentration.

Figure 4.10 Calibration comparison with TISAB after cleaning

0.2

1

3

0.15

0.070.050.1

3

0.05

0.080.06

0.1

0.2

1

50

60

70

80

90

100

110

120

130

140

150

160

170

1800.01 0.1 1 10

[F] ppm Logarism scale

P(m

V)

Low LevelTISABTISAB III


80

The graph for Low Level TISAB displays curvature below 0.1mg/L. Better linearity is

obtained by using TISAB III. TISAB III is therefore the preferred buffer to be used for

low-level F- determination. The results of F- determination in a set of samples with

known concentration using Low Level TISAB are shown in Table 4.11.

• Known sample measurement (Deionised water matrix) Table 4.11 Sample measurement with Low Level TISAB

Calibration range 0.07-3 mg/L 0.1-3 mg/L Function Y=-51.665x+105.57 Y=-53.12x+105.48

R2 0.9962 0.9987 Known

[F] (mg/L) Calculated[F] (mg/L)

SD (N=2)

% Error


SD (N=2)

% Error

0.06 0.071 0.000 18.3 0.076 0.000 26.7 0.09 0.094 0.004 4.4 0.100 0.004 11.1 0.12 0.122 0.005 1.7 0.129 0.005 7.5 0.18 0.173 0.009 -3.9 0.181 0.009 0.6 0.25 0.240 0.011 -4.0 0.249 0.011 -0.4 0.4 0.370 0.005 -7.5 0.378 0.005 -5.5 0.5 0.472 0.015 -5.6 0.480 0.015 -4.0 0.8 0.776 0.002 -3.0 0.779 0.002 -2.6 1.5 1.573 0.000 4.9 1.548 0.000 3.2 2 2.169 0.014 8.5 2.115 0.013 5.8

The function for the range 0.07-3mg/L was not suitable for calibration because of their

poor linearity. The calibration lines for the range 0.1-3mg/L were linear and used for

calculation of analyte concentrations. The recoveries, obtained using Low Level TISAB,

were reasonable but not very accurate for the low concentrations.

iv) Low level sample measurement (Tap water matrix)

Up to so far, all the results were obtained in deionised water matrix. However, this matrix

was not relevant to environmental samples. Therefore, further investigations for

environmental samples, using tap water matrix, which was the most accessible, were

done. This section describes the TISAB efficiency, using tap water matrix, in terms of the

Laboratory Fortified Matrix (LFM).

According to the results obtained previously, the lowest concentration positioned on the

linear portion of the calibration graph was 0.05 mg/L for TISAB III and 0.1 mg/L for

Low Level TISAB. The calibration standards, 0.1, 1 and 3mg/L, positioned within the

linear part of the calibration line for both TISAB, were chosen. Raw tap water was taken


81

for the measurement of the background and 5 LFM samples 0.1, 0.2, 0.5, 1 and 2 mg/L

were prepared using tap water. Two batches of the standards and samples, one for using

TISAB III, the other for using Low Level TISAB, were prepared (Test solution: TISAB

III = 20 mL: 2mL and test solution: Low Level TISAB=10mL: 10mL). The results are

given in Table 4.12.

Table 4.12 LFM measurement using two TISABs in tap water matrix

TISAB type TISAB III Low Level TISAB Calibration

Range 0.1-3 mg/L 0.1-3 mg/L

Function Y = -54.627x+84.945 Y = -54.872x+107.14 R2 0.9994 0.9993

Matrix Type

*Back Ground[F](C)

Added [F] (s)

FortifiedConc. (Cs)

LFM **R %

*Back Ground [F](C)

Added [F] (s)

Fortified Conc. (Cs)

LFMR %

0.1 0.270 103.0 0.1 0.232 73.0 0.2 0.360 96.5 0.2 0.347 94.0 0.5 0.640 94.6 0.5 0.618 91.8 1 1.178 101.1 1 1.103 94.4

Tap water

0.167

2 2.208 102.1

0.159

2 2.200 102.1*Background [F]: concentration before adding F-, ** % R: % Recovery

The percent recovery(R) was calculated by the expression given below. The recovery

limit should be within the range from 75 to 125% (US EAP method 300.1).

R = 100×−S

CCs (4.2)

After calibration, the background value, which is the F- concentration in raw tap water

before adding F-, was measured and found to be 0.167mg/L for TISAB III and 0.159mg/L

for Low Level TISAB. The value, 0.167 mg/L, obtained from TISAB III, was considered

to be more reliable since more accurate results had been achieved using TISAB III in

previous experiments that had been done with laboratory-deionised water. In addition,

the % recovery, obtained using TISAB III, was better and within the recovery limit. The

results from Low Level TISAB show one outlier at 73 %, for the 0.1 mg/L F- sample.

Compared to the results for TISAB III, the % recovery from Low Level TISAB was

worse than that of TISAB III. This confirms the use of TISAB III instead of Low Level

TISAB for F- determination in real samples.


82

d. Interference study

The cations, Al3+, Fe3+, Ca2+, Mg2+, Cu2+ as well as the anions, Cl-, NO3-, SO4

2-, PO43-

and a colloid (SiO2) were studied to determine their influence on the analysis of low-level

F-. Two approaches were used.

• Sample solutions were spiked with one interfering species at a time and the eff

ect on F- determination assessed

• Sample solutions were spiked with a mixture of interfering species to a

ssess whether synergistic effects are present.

Details of the concentrated stock solutions of possible interfering species used in

preparation of the test samples are given in Table 4.13.

Table 4.13 Stock solutions for Interferences study

Interference Interference added as

Stock solution (1000* mg/L) (Constituted to 100 mL)

Calcium (Ca2+) CaCl3 0.277 g Magnesium (Mg2+) MgSO4 0.495 g

Aluminium (Al3+) AlCl3·6H2O 0.894 g

Cupper (Cu2+) CuCl2·2H2O 0.268 g

Cations

Iron (Fe3+) Fe (NO3) 3·9H2O 0.723 g Chloride (Cl-) KCl 0.210 g

Nitrate (NO3-) NaNO3 0.137 g

Sulfate (SO42-) Na2SO4 0.148 g

Anions

Phosphate (PO43-) NaH2PO4·6H2O 0.164 g

Collide Silica (SiO2) SiO2 powder (300 mg/L) 30g

* The concentration of all stock solutions, except silica (300 mg/L), were 1000 mg/L

A series of standard and synthetic sample mixtures were prepared by the appropriate

dilution of the stock solutions with deionised water. TISAB III, which contained CDTA

as a masking agent, was added in each test solution and the F- concentration was

measured.

i) Single compound ion interference

One chemical at a time was chosen for investigating the interference effect on the F-

determination. The results are summarized in Table 4.14 without TISAB III.


83

Table 4.14 Effects of interference on the determination of F- no TISAB III

Known [F] (mg/L) 0.5 1 2

Interfering Substance

Interference Conc. (mg/L)

Measured [F](mg/L) % R* Measured

[F](mg/L) % R Measured [F](mg/L) % R

0 0.504 100.9 1.003 100.3 2.009 100.51 0.496 99.3 0.999 99.9 2.009 100.5

10 0.518 103.7 1.015 101.5 2.017 100.950 0.531 106.2 1.047 104.7 2.082 104.1

PO4 (As NaH2PO4·6H2O)

100 0.537 107.4 1.060 106.0 2.115 105.80 0.498 99.6 0.982 98.2 1.954 97.71 0.504 100.8 0.994 99.4 1.961 98.1

10 0.527 105.4 1.037 103.7 2.054 102.7SO4

(As Na2SO4) 50 0.552 110.4 1.073 107.3 2.126 106.30 0.493 98.6 1.014 101.4 2.004 100.21 0.495 99.0 1.014 101.4 1.996 99.8

10 0.519 103.9 1.059 105.9 2.085 104.2Cl

(As KCl) 50 0.547 109.4 1.068 106.8 2.187 109.40 0.196 98.0 0.498 99.6 0.982 98.21 0.206 103.0 0.490 98.0 0.998 99.8

10 0.216 108.0 0.529 105.8 1.029 102.9NO3

(As NaNO3) 50 0.225 112.5 0.548 109.6 1.077 107.70 0.486 97.2 1.026 102.6 2.008 100.41 0.006 1.2 0.043 4.3 0.484 24.2

10 0.000 0.0 0.001 0.1 0.001 0.150 0.000 0.0 0.000 0.0 0.000 0.0

Al (As AlCl3·6H2O)

100 0.000 0.0 0.000 0.0 0.000 0.00 0.515 103.1 0.992 99.2 2.005 100.21 0.623 124.7 1.324 132.4 2.637 131.8

10 0.109 21.7 0.350 35.0 0.839 41.9Fe

(As Fe(NO3)3·9H2O) 50 0.008 1.6 0.016 1.6 0.032 1.60 0.502 100.3 1.043 104.3 2.029 101.51 0.526 105.2 1.089 108.9 2.091 104.65 0.552 110.4 1.109 110.9 2.163 108.2

10 - - 1.224 122.4 2.388 119.450 - - 1.184 118.4 2.320 116.0

Mg (As MgSO4)

100 - - 1.145 114.5 2.189 109.40 0.502 100.3 1.043 104.3 2.029 101.51 0.539 107.8 1.092 109.2 2.145 107.35 0.569 113.8 1.125 112.5 2.228 111.4

10 - - 1.209 120.9 2.489 124.550 - - 1.250 125.0 2.521 126.0

Ca (As CaCl3)

100 - - 1.255 125.5 2.459 122.90 0.498 99.6 0.982 98.2 1.954 97.71 0.576 115.2 1.098 109.8 1.961 98.1

10 0.587 117.4 1.172 117.2 2.218 110.9Cu

(As CuCl2·2H2O) 50 0.590 118.0 1.190 119.0 2.349 117.5

R*=Recovery


84

Table 4.14 shows the effect of a single compound interference on the determination of F-

without TISAB III. All the anion interferences show positive errors and the recoveries are

over 100 %. Considering the effect of ionic strength and activity, the % recoveries should

have been less than 100% (see Section 3.1.2.a). When the ionic strength increases an ion

loses some of its effectiveness, and its activity coefficient decreases. This indicates that

there must be another reason for the over-estimation of F- in the presence of anions tested.

The possible reason could be the F- electrode response to other anions when the

concentration of anion interference is high enough.

Among all the cation interferences, Al showed the most serious influence on

determination of F-. The % recoveries of 0.5 mg/L F- in presence of Al interference were

1.2 % for 1 mg/L Al, 0 % for 10, 50 and 100mg/L Al. Table 4.15 shows that ionic

strength effects can only account for a very small part of the observed interference.

Table 4.15 Ionic strength and activity coefficient for Al interference

[F] (mg/L) Added [Al] (mg/L) *Ionic strengthµ

**Activity coefficient γ

Error (-) in Concentration

1 0.00023 0.983 1.74 % 10 0.00211 0.950 4.99 % 50 0.01047 0.898 10.19 % 0.5

100 0.02092 0.865 13.54 % *Ionic strength was calculated by the equation 3.2 in Section 3.1.2.a.i

**Activity coefficient was calculated by the Debye-Hückel equation (see Section 3.1.2.a.ii)

It is well known that Al ions readily form stable complexes with F- ions (Vanlengenhaghe,

1966; Durrant and Durrant, 1970; Srinivasan and Rechnitz, 1988). Several different

forms of Al-F can be found with various F/Al ratios (Wells, 1975). Durrant (1970)

reported that one Al ion in the complex structure is surrounded by a maximum of 6

octahedrally arranged F- ions. Therefore, the significant interference by Al ions is caused

by the strong affinity of F- ions for Al ions and the formation of an octahedral structure of

Al-F with which one Al ion can combine up to a maximum of six F- ions, unless

decomplexing agents such as CDTA, the citrate-nitrate medium, are used (Nakajima,

1997). Fe also seemed to cause a negative error, except at 1 mg/L Fe, which shows a

positive error. This positive error was caused by the NO3- interference, which came from

the chemical (Fe (NO3)3·9H2O) used to prepare the Fe interference stock solution. The


85

extent of the Fe interference is much less than that of Al. The reason is the much smaller

stability constants for the Fe complexes with F- (see Table 4.16).

Table 4.16 The log of stability constant* for F- ligand

Metal Stability constant* Al3+ Fe3+ Fe2+ Mg2+ Ca2+ Cu2+

Log K1 6.10 5.16 0.83 1.32 0.70 0.84

Log K2 4.93 3.91 - - - -

Log K3 6.69 2.93 - - - -

Log K4 2.50 - - - - -

* From Stability Constants of Metal-Ion Complexes Part A INORGANIC LIGANDS (Erik H., 1982)

The stability constants for both Fe2+ and Fe3+ complexes with F- are less than that of Al.

At the1mg/L Fe level, the % recoveries were 124.7%, 132.4% and 131.8 % respectively.

This overestimation shows that the effect of the Fe-F complex, at 1mg/L Fe, was less

than that of anion (NO3-) effect that caused the overestimation. As the Fe concentration

increased, the % recoveries were decreased.

Ca2+, Mg2+ and Cu2+, in Table 4.14, show a positive error. This positive error could not

be interpreted in terms of ionic strength and activity, because the increase of the ionic

strength causes a decrease in the activity, which should lead to a negative error. All the

positive errors for these cation interferences seem to be caused by anion interference

from the counter ion. The logs of stability constants of these cations (see Table 4.16)

were 1.32, 0.70 and 0.84. This meant that Ca2+, Mg2+ and Cu2+ did not form strong

fluoro-complexes like Al3+ or Fe3+. These cations, Ca2+, Mg2+ and Cu2+, therefore did not

present major problems on the F- determination. The results, obtained by using TISAB III,

are summarized in Table 4.17.


86

Table 4.17 Effects of interference on the determination of F- with TISAB III (single compound interference)

Known [F] (mg/L) 0.5 1 2 Interfering

Substance

Concentration Of

Interference (mg/L)



0 0.494 98.9 1.009 100.9 2.015 100.71 0.538 107.6 0.956 95.6 1.946 97.310 0.465 92.9 0.978 97.8 1.963 98.150 0.465 92.9 0.978 97.8 1.929 96.5

PO43-

(as NaH2PO4·6H2O)

100 0.460 92.1 0.996 99.6 1.946 97.30 0.499 99.8 0.980 98.0 2.038 101.91 0.481 96.1 0.971 97.1 2.097 104.910 0.479 95.7 0.988 98.8 2.115 105.7

SO42-

(as Na2SO4)

50 0.481 96.1 0.988 98.8 2.106 105.30 0.528 105.7 1.026 102.6 2.003 100.11 0.511 102.2 1.006 100.6 2.028 101.410 0.507 101.4 1.026 102.6 2.061 103.1

Cl-

(as KCl)

50 0.505 101.0 0.997 99.7 2.028 101.40 0.205 102.5 0.503 100.6 0.998 99.81 0.204 102.0 0.478 95.6 0.990 99.010 0.207 103.5 0.476 95.2 0.994 99.4

NO3-

(as NaNO3)

50 0.199 99.5 0.480 96.0 0.990 99.00 0.505 101 0.994 99.4 2.044 102.21 0.477 95.4 0.876 87.6 1.923 96.210 0.353 70.6 0.677 67.7 1.363 68.250 0.161 32.2 0.295 29.5 0.575 28.8

Al3+

(as AlCl3·6H2O)

100 0.107 21.4 0.168 16.8 0.310 15.50 0.507 101.4 0.981 98.1 2.028 101.41 0.441 88.1 0.900 90.0 1.829 91.410 0.444 88.8 0.911 91.1 1.867 93.4

Fe3+

(as Fe(NO3)3·9H2O) 50 0.461 92.2 0.914 91.4 1.844 92.20 0.495 99.1 0.989 98.9 2.100 105.01 0.503 100.6 1.058 105.8 2.043 102.2

Mg2+

(As MgSO4) 5 0.502 100.3 1.030 103.0 2.035 101.70 0.495 99.1 0.989 98.9 2.100 105.01 0.506 101.2 1.019 101.9 2.083 104.1

Ca3+

(As CaCl3) 5 0.515 102.9 1.039 103.9 2.159 108.00 0.494 98.9 1.009 100.9 2.015 100.71 0.538 107.6 0.956 95.6 1.946 97.310 0.465 92.9 0.978 97.8 1.963 98.1

Cu2+

(as CuCl2·2H2O) 50 0.465 92.9 0.978 97.8 1.929 96.5

*R=Recovery


87

Compared to the previous results in Table 4.14 (without TISAB III), all the results in this

table display better recoveries because of using TISAB III. Over-estimation from the

anion interferences could be corrected by adding TISAB III.

However, the cation interferences, Al and Fe, were problematic even after adding TISAB

III. The following histograms, obtained from the experimental results, (Figure 4.11 and

4.12) show the efficiency of TISAB III for de-complexation of Al and Fe at the level of 1

mg/L Al and 10mg/L Fe.

Figure 4.11 TISAB efficiency for F- determination in presence of Al*

0

20

40

60

80

100

0.5 1 2

[F] (mg/l)

Fluo

ride

Rec

over

y(%

)

NoTISAB With TIII

*Al concentration = 1 mg/L

Figure 4.12 TISAB efficiency for F- determination in presence of Fe*

0

20

40

60

80

100

0.5 1 2

[F] (mg/l)

Fluo

ride

Rec

over

y(%

)

NoTISAB With TIII

* Fe concentration = 10 mg/L


88

These histograms clearly show the efficiency of TISAB III for de-complexation of Al-F

and Fe-F complexes. The stability constant of CDTA is given in Table 4.18.

Table 4.18 The log of stability constant* for CDTA ligand

Metal Stability constant*

Al3+ Fe3+ Fe2+ Mg2+ Ca2+ Cu2+

Log K 17.63 for Al3++L4- ↔ AlL- 29.49 9.30 10.20 13.15 1.92

Log K 7.58 for AlLOH2-+H+↔ AlL- - -

Log K 3.93 for AlL-+H+↔ AlLH - -

Where, CDTA=H4L, *From Stability Constants of Metal-Ion Complexes Part B ORGANIC LIGANDS (Douglas D.P., 1979)

The stability constants of Metal-CDTA complexes are higher than those of Metal-F-

complexes in each case. This meant that CDTA could effectively de-complex metal from

metal-F- complexes, which would release free F- ions. Therefore the % recoveries of each

test solution were very much improved after adding TISAB III. However, the %

recoveries of F-, in presence of 1 mg/L Al were 95.4 %, 87.6% and 96.2 % and, in

presence of 10 mg/L Al were 70.6 %, 67.7 % and 68.2% for 0.5 mg/L 1 mg/L and 2 mg/L

F- respectively. This indicates that TISAB III could not perfectly solve the Al interference.

The effect of increasing Al concentration, for the F- determination, is illustrated in Figure

4.13.

Figure 4.13 F- % recovery verse Al concentration (TISAB III)

0

20

40

60

80

100

120

0 10 20 30 40 50 60 70 80 90 100

[Al] (mg/L)

Fluo

ride

Rec

over

y(%

) 0.5 mg/L F1 mg/L F2 mg/L F

As the Al concentration increases, 1, 10, 50 and 100 mg/L, the % recovery of F- decrease

drastically. A very similar graph has been reported by Duff and Stuart (1975) using a

different type of TISAB, (triethanolamine (TEA) buffer and citrate buffer). Harwood


89

(1968) showed that iron exhibited an effect, but at the concentrations normally found in

natural waters this effect will be insignificant. He also referred that Al interference is a

major obstacle, even at the levels normally found in water. It was therefore necessary to

determine the tolerable level of Al as well as a way to overcome Al interference on the F-

determination. To find the value of Al tolerance, low level Al concentration, below 1

mg/L, were examined with 0.2, 0.5 and 1 mg/L F- solutions. The results, obtained for the

Al tolerance study, are given below (Table 4.19).

Table 4.19 F- % recovery in presence of low-level Al (TISAB III)

Known [F] (mg/L) 0.2 0.5 1 Interfering

Substance




0 0.198 99.0 0.489 97.8 0.995 99.50.1 0.193 96.5 0.470 94.0 0.956 95.60.2 0.188 94.0 0.430 86.0 0.956 95.60.5 0.187 93.5 0.430 86.0 0.884 88.4

Al (as AlCl3·6H2O)

1 0.186 93.0 0.445 89.0 0.855 85.5

From the Table 4.19, the Al tolerance for 95% F- recovery is 0.1 mg/L for 0.2 mg/L and

0.5 mg/L F-, 0.2mg/L for 1 mg/L F-. Since Al interference is a major problem, many

articles, concerning of Al interference on the F- determination, have been reported

(Harwood, 1968; Duff and Stuart, 1975; Nicholson and Duff, 1981; Frenzel and Brätter,

1986; Pickering, 1986; Okutani et al., 1989; Hiroshi et al., 1997; Trojanowicz et al.,

1998). In the determination of F- with an ISE in the presence of Al, release of the F- from

Al-F complex is essential and for this purpose several masking agents have been

examined.

Fritz et al. (1958) used sulphosalycylic acid and 2, 4-pentanedione as agents for masking

Al. Citrate masked iron at pH 7, but Al was incompletely masked. Sulphosalicyclic acid

did not mask iron but masked Al at pH 4.5 while the pentanedione masked Al at pH 6.5-

7.5.

Hardwood (1968) studied the effect of masking agents, such as citrate, EDTA and CDTA,

on overcoming Al interference with 1 mg/L F-, using different Al concentrations. He


90

found citrate and EDTA were not satisfactory, while CDTA released F- at Al levels up to

3 or 5 mg/L, depending on the accuracy required. The result obtained by Crosby et al.

(1968) agreed with the fact that citrate is not completely reliable for masking Al.

However, the buffer recommended by them, contained EDTA.

Yuchi et al. (1986), on the other hand, reported that citrate is more effective than DCTA

or EDTA for masking Al, because it does not form appreciable amounts of mixed ligand

complexes with Al and F- at pH 6.

Nakajima et al. (1997) suggested that when a solution contains more than 0.5 mg/L F- and

up to 20 mg/L Al, the use of TISAB with sodium citrate-potassium nitrate was

recommended. On the other hand, when a solution contains more than 0.1 mg/L F- and

less than 5 mg/L Al, TISAB contained CDTA could be used.

The different views expressed, arise in part from the fact that the degree of release varies

with buffer pH. At low F- levels, the ISE response at pH>8 can be in error because of

interference by hydroxide ions (Orion, 1982), but use of higher pH reduces Al

interference. At pH ~8, recommended masking agents include 0.25 M sodium citrate

(Vickery and Vickery, 1976), 1M sodium citrate (Oliver and Clayton, 1970) and

tris(hydroxymethyl)aminomethane (Kauranen, 1977). At pH 12, formation of the

tetrahydroxo-aluminate ion tends to eliminate Al interference (Oliver and Clayton, 1970).

Okutani et al. (1989) has been illustrated the effect of pH on masking of Al, using citrate

buffer. The masking ability of citrate is markedly enhanced by increase in pH from 6.0 to

8.5, and even at a concentration of 0.015 M (405 mg/L) Al can be masked at pH 8.5 with

citrate in the determination of 0.2-2.0 µg/mL F-. Based on this fact, they developed a

method, which could mask large amounts of Al and determine low levels of F- in natural

waters after co-precipitation with aluminum phosphate.

Nicholson and Duff (1981) also have compared the effectiveness of various masking

agents, which were Citrate, CDTA, Tiron, Mannitol, Salicylate and Tris(hydroxymethyl)

methylamine. Of all the buffer masking agent combinations examined, tri-Ammonium

citrate (TAC) buffer, employing CDTA and tri-ammonium citrate as complexing agents,

was most efficient in terms of masking ability. At 1 mg/L F-, an Al concentration up to


91

100 mg/L was tolerated for 95% recovery. This was greatly superior to TISAB III, which

at the same F- concentration and recovery level tolerated only 2.7 mg/L Al. An important

aspect is the time required for the masking of Al to become complete. In their detailed

study, they noted that 24 hours following buffer addition, would, for each species and

each buffer, at 1 mg/L F- be beneficial in terms of an increased F- recovery. These results

were very remarkable so that the experiments were repeated to confirm. Firstly, TAC

buffer was prepared using TISAB III. The calculated value of TAC was 1M. The results

of F- determination with this buffer as well as the previous work from Table 4.17 are

summarized in Table 4.20.

Table 4.20 The efficiency of TAC buffer on the F- determination

1 mg/L F-

TISAB Type *TISAB III TAC buffer

Interfering Substance

InterferenceConc. (mg/L)

Measured [F](mg/L) % R*

Measured [F](mg/L) % R

0 0.994 99.4 1.024 102.4 1 0.876 87.6 0.962 96.2 5 - - 0.942 94.2

10 0.677 67.7 0.904 90.4 50 0.295 29.5 0.666 66.6

Al (as AlCl3·6H2O)

100 0.168 16.8 0.509 50.9 * TISAB III results was cited from Table 4.17

This result obviously shows that Al concentration up to 5 mg/L was tolerated for 94%

recovery at 1 mg/L F-. Compared to the previous work in Table 4.17, the F- recovery was

very much improved employing TAC buffer. Although this tolerance value was far less

than that of the article, which is 100 mg/L Al, this TISAB can be applied to F-

determinations in the environmental samples that contain Al below 1 mg/L. This

experiment confirmed the fact that TAC buffer is greatly superior to TISAB III in terms

of masking Al interference.

Secondly, the experiment was designed to confirm the effectiveness of time allowance

using TISAB III. A reference solution, 1 mg/L F-(Fr) and a sample solution, mixture of

1mg/L F- and 1 mg/L Al (Fs) were made up. After calibration, using standards 0.1, 0.5

and 3 mg/L F- concentrations of these two solutions were determined at several times


92

from 1 minute to 1 hour (15 points, see Table 4.21). Since this experiment took a long

time, the electrode drift had to be considered. Therefore, the ratio (Fs/Fr), instead of single

measurement (Fs), was calculated to compensate the electrode drift. The results are

shown in Table 4.21.

Table 4.21 The effectiveness of time allowance on F- determination in presence of 1 mg/L and 10 mg/L Al at 1 mg/L F-

Time (h) Fs/Fr (1Al+1F) Fs/Fr (10Al+1F) 0.02 (1minute) 0.85 0.60

0.17 (10minutes) 0.86 0.60 0.33 (20 minutes) 0.86 0.60 0.5 (30minutes) 0.87 0.61

1 0.87 0.61 2 0.88 0.62 3 0.91 0.62 4 0.93 0.62 5 0.94 0.62 6 0.93 0.63 7 0.95 0.63 8 0.95 0.63

9.5 0.95 0.63 10.5 0.95 0.63 24 0.95 0.64

The decomplexing ability of the buffer was followed from the moment the buffer was

added to the F- containing sample, until a stable potential plateau was obtained. This

result show that in TISAB III, 24 hours decomplexing gave improved F- recovery in

presence of 1 mg/L Al, but there was no significant improvement at 10 mg/L Al. It seems

there is sufficient advantage in analyzing for F- in natural samples, which normally

contains Al below 1 mg/L.

Two experiments were done for finding a way to overcome Al interference in F-

determination. The recommendation for F- determination in presence of 1 mg/L level Al

interference is to use TAC buffer and to measure the sample after 24 hours after adding

TISAB buffer.


93

ii) Single compound colloid interference

Colloid particles usually occur in natural samples. A colloid is a system in which finely

divided particles, which are approximately < 0.1 µm in size, are dispersed within a

continuous medium in a manner that prevents them from being filtered easily or settled

rapidly. 60 mg/L colloidal SiO2 was chosen to check the effect of the colloid interference

(Table 4.22).

Table 4.22 Effects of colloid interference on the determination of F- with TISAB III

Known [F] (mg/L) 0.2 0.5 1 Interfering

Substance


Measured [F](mg/L) % R Measured


0 0.196 97.8 0.485 96.9 0.982 98.260 0.201 100.7 0.490 97.9 1.003 100.3SiO2

(As SiO2 powder) 180 0.195 97.5 0.480 96.0 0.973 97.3

This indicates that SiO2 colloidal particle has little effect on the F- result, even at 180

mg/L.

iii) Multi-compound interferences

Up to so far, single compound ion and colloid interference on F- determination was tested.

However, there were numerous elements in the natural sample. So sample solutions were

spiked with more than 2 compounds, with various mixtures of interfering species, to

assess whether synergistic effects are present. The concentrations of interfering

substances were Al3+ 1mg/L, Mg2+ 10mg/L, Ca2+ 10mg/L, Fe2+ 1 mg/L and PO43-

10mg/L at 0.2, 0.5 and 1 mg/L F-.


94

Table 4.23 Effects of interference on the determination of F- with TISAB III (more than 2 compounds interference-synergy effect)

Interfering Compound*

Numbers

Interfering Substances

[F] Known(mg/L)

[F] Measured (mg/L) %R

0 F Only 0.199 99.4 Al + Mg 0.195 97.3 Al + Fe 0.189 94.6 Al + Ca 0.192 96.0 Ca + Mg 0.203 101.4 Ca + Fe 0.192 95.8

2

Mg + Fe 0.200 100.2 0 F only 0.197 98.7

Al + Ca + Mg 0.181 90.4 Al + Ca + Fe 0.175 87.4 Al + Mg + Fe 0.176 88.1 3

Ca + Mg + Fe 0.181 90.4 4 Al + Ca + Mg + Fe 0.185 92.7 5 Al + Ca + Mg + Fe + PO4

0.2

0.177 88.6 0 F only 0.492 98.4

Al + Mg 0.462 92.5 Al + Fe 0.449 89.9 Al + Ca 0.459 91.8 Ca + Mg 0.482 96.4 Ca + Fe 0.468 93.7

2

Mg + Fe 0.482 96.4 0 F only 0.492 98.4

Al + Ca + Mg 0.442 88.5 Al + Ca + Fe 0.446 89.2 Al + Mg+ Fe 0.441 88.3 3

Ca + Mg + Fe 0.470 94.1 4 Al + Ca + Mg + Fe 0.437 87.4 5 Al + Ca + Mg + Fe + PO4

0.5 0.439 87.7

0 F only 1.004 100.4 Al + Mg 0.880 88.0 Al + Fe 0.887 88.7 Al + Ca 0.896 89.6 Ca + Mg 0.975 97.5 Ca + Fe 0.975 97.5

2

Mg + Fe 0.979 97.9 0 F only 1.004 100.4

Al + Ca + Mg 0.853 85.3 Al + Ca + Fe 0.850 85.0 Ca + Mg + Fe 0.953 95.3 3

Al + Mg + Fe 0.860 86.0 4 Al + Ca + Mg + Fe 0.839 83.9 5 Al + Ca + Mg + Fe + PO4

1

0.829 82.9

*Interfering compound: See Table 4.13


95

Compared to the single compound interference test in Table 4.17, a decrease in trend of

F- recovery was observed when the number of interfering elements increased; in other

words, synergistic effects were observed in the presence of many interfering substances.

e. Analytical parameters

i) Reproducibility and repeatability

The term of reproducibility and repeatability are very similar. Reproducibility is the

variation in average measurements obtained when two or more people measure the same

parts or items using the same measuring technique. Repeatability is the variation in

measurements obtained when one person takes multiple measurements using the same

instrument and techniques on the same parts or items. Since only one person was

involved in this experiment, the reproducibility was unavailable. To determine

repeatability, 0.02, 0.05, 0.1, 0.5, 1.0 and 2.0 mg/L F- test solutions were measured 10

times each. The results are given in Table 4.24 and the repeatability was expressed in

terms of % RSD.

Table 4.24 The repeatability of ISE

[F] Conc. (mg/L) Mean potential (mV) SD (N=10) % RSD (repeatability)

0.02 156.0 2.17 1.4

0.05 148.2 0.78 0.5

0.1 138.5 0.60 0.4

0.5 105.5 0.24 0.2

1.0 89.7 0.20 0.2

2.0 73.4 0.18 0.3

ii) Control chart

0.2 mg/L and 1 mg/L F- test solution were made up and each sample measured for a long

term, 9 months, and the result is presented in Table 4.25.


96

Table 4.25 Control chart for 0.2 mg/L and 1 mg/L F- solution

[F] Known 0.2 mg/L 1 mg/L Date

[F] Measured % Recovery [F] Measured % Recovery 11/08/2003 0.202 101.0 0.970 97.0 30/09/2003 0.195 97.5 0.982 98.2 12/03/2004 0.195 97.5 0.982 98.2 11/05/2004 0.200 100.0 0.986 98.6

This table shows that samples containing F- 0.2 and 1 mg F-/L can be stable for at least 9

months.

iii) Method Detection Limit (MDL)

A Method Detection Limit (MDL) is defined to be the concentration below which the

analytical method cannot reliably detect a response (USEPA, 1984). A well-known

method for calculation of the detection limits is to use the standard deviation of several

blank measurements. But it could not be applied because in ISE a blank sample results in

consistent drift potential, which meant no equilibrium potential value could be obtained,

and thus cannot be determined. Therefore, various other methods were used and will be

explained below.

Firstly, a calibration curve was set up from a set of successively diluted standards and the

detection limits were determined experimentally. The qualitative detection limit for F-

was determined as the lowest concentration at which a equilibrium potential could be

obtained. This was determined at 0.02 mg/L. The quantitative detection limit was

determined as the lowest concentration of F- needed to produce a potential that could be

obtained in equilibrium and produce a linear calibration curve. The quantitative detection

limits were determined as 0.05 mg/L.

Secondly, some guidelines suggest that the criteria for the limit of detection should be

that analyte concentration which gives a signal equal to the blank signal (YB) plus three

times the standard deviation of the blank (SB).

3B BY Y S= + (4.3)


97

Miller and Miller (1993) proposed a method for the calculation of the detection limits

from the above criteria. This method assumes that the residual standard deviation about

regression (Sy/x) can replace SB because the regression line assumes that each point on the

plot has a normally distributed variation in the y-direction only. The intercept of the

regression line (a) is used as an estimate of YB itself. Thus Equation 4.1 becomes:

/3 y xY a S= + (4.4)

Where,

2

/

[ ( )]

2

i ii

y x

y a mxS

N

− +=

−

∑

And yi and xi = the individual pairs of data for signal and concentration, a = the intercept of the regression line, m = the slope of the regression line, N-2 = number of degrees of freedom

For using this method, the regression line was obtained by measuring F- standards, 0.02,

0.05, 0.1, 0.5, 1 and 2 mg/L, 10 times each. The average potential of each standard

solution was then used to calculate the statistical parameters for the calibration line

parameters with Statistical Methods in Analytical Chemistry (SMAC) programme (Peter

and Richard, 2000). Using this data, the detection limits for F- was calculated as

0.46mg/L. This method yielded higher detection limit than the value determined

experimentally.

iv) Linear range and working range

The calibration curve for F- (see Figure 4.7) shows linearity over a range of 0.1-1000

mg/L. With a well-conditioned electrode, the lowest calibration point can be extended up

to 0.05 mg/L (see Figure 4.8). Below the lowest point, 0.05 mg/L, curvature of the

calibration curve occurs and this can cause inaccurate results for F- determination (see

Table 4.5, 4.6 and 4.10 in Section 4.2.3.c). A working range of 0.1-1000 mg/L is normally

recommended for F- determination (Orion, 1982), but for low-level measurement, a

working range 0.1-3 mg/L is recommended to get accurate results.

4.3 Ion Chromatography (IC) method 4.3.1 Instrumentation

The anion separation has been done using a Dionex Ion chromatograph, DX 120 (Dionex,

Sunnyvale, CA, USA). The Dionex DX-120 Ion Chromatograph performs isocratic ion


98

analysis applications using conductivity detection and a Dionex anion exchange column

system (AG 14+AS 14A). The IonPac AS14 column provides complete resolution of F-

from formate and acetate in addition to improved resolution of F- from the column void

peak. The improved selectivity and higher capacity AS 14 column also allows improved

resolution of chloride and nitrite, which is important in environmental water analysis

(Application Note 135). Therefore, the AS 14 analytical column, in conjunction with AG

14 guard column, was chosen for this experimental work. The DX-120 is an integrated

system, which includes a pump, detector, and injection valve. The chromatography

components, including the column(s), Self-Regenerating Suppressor (SRS), and

conductivity cell are ordered separately. The flow rate used was 1.2 mL/min.

4.3.2 Standards and reagent solutions

All reagents were of AR grade and all solutions were prepared from high quality

deionised water (18 MΩ). Samples and eluents were filtered through 0.22 µ membrane

filters.

a. Standards

1000 mg/L stock standards were prepared by dissolving the appropriate amounts of target

analytes in 100 mL of deionised water according to details given in Table 4.26.

Table 4.26 Masses of compounds used to prepare 1000 mg/L stock solution

Anion Compound Mass (g) / 100mL Fluoride (F-) NaF 0.221 Acetate (Ac-) CH3COONa 0.139

Formate (HCOO-) HCOONa 0.151 Chloride (Cl-) KCl 0.210 Nitrite (NO2

-) NaNO2 0.150 Bromide (Br-) KBr 0.149 Nitrate (NO3

-) NaNO3 0.137 Phosphate (PO4

3-) NaH2PO4·6H2O 0.164 Sulfate (SO4

2-) K2SO4 0.181

A series of standards was prepared by the appropriate dilution of the stock solutions with

deionised water.


99

b. Eluent for IC

• 0.5 M CO32- stock solution: 26.49 g Na2CO3 (AnalaR, BDH Chemicals Ltd, Poole,

England) was dissolved in 500 mL deionised water

• 0.5 M HCO3- stock solution: 21g NaHCO3 (Saarchem (Pty) Ltd) was dissolved in

500mL deionised water

• Eluent (3mM HCO3-, 9mM CO3

2-): 12mL 0.5 M CO32-, 4 mL 0.5 M HCO3

-

solutions were combined and diluted to 2000mL


a. Effect of base line drift

US EPA method 300.1 specifies that in order to achieve comparable detection limits, an

ion chromatographic system must utilize suppressed conductivity detection, be properly

maintained and must be capable of yielding a baseline with no more than 5 nS noise/drift

per minute of monitored response over the background conductivity. The stabilisation

time needed at least 30 minutes. After stabilisation of the instrument, the conductivity

was recorded for 10 minutes three times and the results are given in Table 4.27.

Table 4.27 Base line drift in IC Time Conductivity 1 Conductivity 2 Conductivity 3 (min) (µS) (µS) (µS)

1 18.258 18.292 18.319 2 18.254 18.293 18.316 3 18.258 18.295 18.316 4 18.260 18.296 18.322 5 18.256 18.290 18.318 6 18.260 18.293 18.319 7 18.259 18.291 18.318 8 18.257 18.293 18.319 9 18.259 18.292 18.320

10 18.257 18.294 18.322

Base line drift (nS/minutes)

0.6 0.6 0.6


100

The base line drift was calculated by the following equation:

Base line drift = 100010

H LC C−× (4.5)

Where, CH = the highest conductivity value, CL = the lowest conductivity value, 1000 is a conversion factor from µS to nS

The calculated value of the base line drift is 0.6 nS, which is acceptable.

b. Calibration study

Correct calibration is an essential factor to achieve accurate measurements of F-

concentration. Calibration lines for different standard series and the calculated linearity in

terms of correlation coefficient are presented in Table 4.28. It was found that the IC

software was set to yield the calibration function, forcing the y-intercept through zero.

Since the calibration line is normally obtained not forcing y-intercept through zero, this

table shows the calibration function and linearity using both calculation methods.

Table 4.28 Calibration factor with different set of standards

Force through zero Ignore zero Standards set Function Linearity Function Linearity

1, 5, 10 Y = 792377x 0.9996 Y = 804977x-99226 0.9999 0.05, 1, 3 Y = 198473x 0.9999 Y = 198475x-4.9656 0.9999 0.1, 0.5, 3 Y = 202925x 0.9998 Y = 204151x-3153.2 0.9999

0.05, 0.5, 3 Y = 201141x 0.9999 Y = 202093x-2481.2 1.0000 0.1, 1, 10, 50 Y = 750234x 0.9994 Y = 745853x+186479 0.9995

0.05, 0.1, 1, 5, 10 Y = 825592x 0.9989 Y = 837317x-91487 0.9994 0.05, 0.2, 0.5, 1, 3 Y = 177497x 0.9997 Y = 177708x-455.95 0.9997

A better linearity was obtained when the y-intercept was not forced through zero and

when three standards were used, rather than 4 or 5 standards. As was expected, increasing

the analytical range to include higher concentrations, such as 10 mg/L or 50 mg/L F-,

reduced the linearity. Since this study focused on low-level F- determinations, the

analytical range 0.05 to 5 mg/L was selected for further study. To evaluate the effect of

injection volume, two sample loops, 25 µL and 50 µL, were tested using a calibration


101

standard series with concentrations 0.05, 0.2, 0.5, 2 and 5 mg/L F- and a set of spiked

samples with concentration 0.1, 0.7, 1 and 4 mg/L. The results are given in Table 4.29.

Table 4.29 Optimisation of the injection volume

25 µL loop 50 µL loop [F] Known (mg/L) [F] Measured

(mg/L) %

Recovery[F] Measured

(mg/L) %

Recovery 0.1 0.102 102.4 0.110 110.4 0.7 0.668 95.4 0.670 95.8 1 0.966 96.6 0.923 92.3 4 3.959 99.0 4.004 100.1

A better recovery was found for the lower concentration levels, 0.1, 0.7 and 1 mg/L, with

the 25 µL loop. Therefore, the 25 µL loop was used for further work.

A brand new analytical column (AS 14) was tested, with and without guard column, to

evaluate its efficiency. Usually a short guard column is placed in front of the analytical

column to increase the life of the analytical column by removing particulate matter and

contaminants from the solvents (see Section 3.2.2.d). 0.2, 0.7, 1 and 2 mg/L F- test

solutions were measured after calibration and the results are given in Table 4.30.

Table 4.30 Column efficiency on F- determination

AS 14 column Only

AG 14 + AS 14 Together [F] Known

(mg/L) [F] Measured (mg/L)

% Error

[F] Measured (mg/L)

% Error

0.2 0.184 -8.2 0.189 -5.5 0.7 0.653 -6.7 0.650 -7.1 1 0.965 -3.5 0.952 -4.8 2 1.974 -1.3 1.950 -2.5

The AS 14 analytical column, used without guard column, gave better accuracy than the

combined column system. However, to use only the analytical column for environmental

samples could reduce column life since they may contain many unknown contaminants.


102

A 3-point calibration curve was prepared using the standards 0.05, 1 and 3 mg/L for the

first run and 0.05, 0.5 and 3 mg/L for the second run. A sample series with F-

concentrations between 0.06 and 2.5 mg/L was then analysed. The results were calculated

using the Peak Net Chromatography software (force through zero option) and by manual

calculation using an Excel spread sheet. The results obtained from the first run are given

in Table 4.31.

Table 4.31 Influence of decimal factor on accuracy

Calculation method By IC software (Peak Net) By manual (Excel Programme) Known Calculated % Error Calculated % Error

[F] (mg/L) [F] (mg/L) Accuracy [F] (mg/L) Accuracy 0.06 0.06 0.0 0.057 -5.1 0.07 0.07 0.0 0.067 -4.2 0.09 0.09 0.0 0.086 -4.9 0.1 0.10 0.0 0.098 -1.8

0.12 0.11 -8.3 0.112 -6.6 0.15 0.15 0.0 0.148 -1.5 0.18 0.18 0.0 0.178 -1.2 0.2 0.20 0.0 0.197 -1.4 0.3 0.28 -6.7 0.278 -7.2 0.4 0.38 -5.0 0.377 -5.6 0.5 0.47 -6.0 0.468 -6.4 0.8 0.76 -5.0 0.761 -4.9 1.5 1.44 -4.0 1.442 -3.9 2 1.96 -2.0 1.956 -2.2

2.5 2.47 -1.2 2.468 -1.3

The table shows the effect of rounding to 2 decimal places by the Peak Chromatography

software. For instance, although the accuracy is calculated to be 0 % at 0.09 mg/L F-, it

does not really mean there is no error. Once the third decimal place is taken into account,

the accuracy is –4.9 % as shown by the manual calculation.

The results of F- determination using the two sets of standards are given in Table 4.32

and 4.33.


103

Table 4.32 Evaluation of calibration type on the F- determination (1st run)

Function type Force through zero Ignore zero Function Y=198473x (R2=0.9999) Y=198475x-4.9656 (R2=0.9999) Known Calculated % Calculated %

[F] (mg/L) [F] (mg/L) Error [F] (mg/L) Error 0.06 0.057 -5.1 0.057 -5.1 0.07 0.067 -4.2 0.067 -4.2 0.09 0.086 -4.9 0.086 -4.8 0.1 0.098 -1.8 0.098 -1.7 0.12 0.112 -6.6 0.112 -6.6 0.15 0.148 -1.5 0.148 -1.5 0.18 0.178 -1.2 0.178 -1.2 0.2 0.197 -1.4 0.197 -1.4 0.3 0.278 -7.2 0.278 -7.2 0.4 0.377 -5.6 0.377 -5.6 0.5 0.468 -6.4 0.468 -6.3 0.8 0.761 -4.9 0.761 -4.8 1.5 1.442 -3.9 1.442 -3.9 2 1.956 -2.2 1.956 -2.2

2.5 2.468 -1.3 2.468 -1.3 Standards: 0.05, 1 and 3 mg/L

Table 4.33 Evaluation of calibration type on the F- determination (2nd run)

Function type Force through zero Ignore zero Function Y=201141x (R2=0.9999) Y=202093x-2481.2 (R2=1) Known Calculated %Error Calculated %Error

[F] (mg/L) [F] (mg/L) Accuracy [F] (mg/L) Accuracy 0.06 0.056 -6.7 0.068 12.6 0.07 0.062 -11.4 0.074 5.1 0.09 0.079 -12.2 0.091 1.0 0.10 0.092 -8.0 0.104 3.7 0.12 0.110 -8.3 0.122 1.3 0.15 0.135 -10.0 0.146 -2.6 0.18 0.168 -6.7 0.179 -0.3 0.2 0.185 -7.5 0.196 -1.9 0.3 0.278 -7.3 0.289 -3.6 0.4 0.365 -8.8 0.376 -6.0 0.8 0.748 -6.5 0.757 -5.4 1.5 1.403 -6.5 1.409 -6.1 2.0 1.903 -4.8 1.906 -4.7

Standards: 0.05, 0.5 and 3 mg/L


104

For the first run, the concentration values were accurate but not for the second run where

concentrations were underestimated when calculated using a “force through zero”

calibration line. The calibration line, which ignored zero, gave acceptable results for the

second run. In the case of first run, the results for both calibration lines were good. In this

case, the y-intercept of the calibration function was only –4.9656 and close to zero.

Therefore, even after forcing the y-intercept through zero, the function was not changed

much. However, for the second run, the y-intercept was –2481.2, which was a quite big

value. Thus, when the y-intercept was forced through zero, the linearity of calibration line

was reduced and low concentrations, especially, show large errors.

These results indicate that establishing a proper calibration function is of utmost

importance, for achieving an accurate F- determination. Therefore, every effort must be

made to establish the best calibration line possible. Experimental aspects to consider

include using very accurate and properly calibrated volumetric equipment such as

pipettes, clean containers, and correctly prepared standards. If the calibration line is well

established like the first run in Table 4.38, very accurate results can be obtained.

On the other hand, when calibration line is not well established like second run, then the

appropriate choice of calibration function, for example ignoring zero on the calibration

line, becomes important. It was found that peak threshold value could affect the F-

determination. The threshold is a value derived from the averaged slopes of data bunches

along a baseline sector. It is applied as a limiting slope to determine peaks. It marks the

potential start of a peak when two consecutive data bunches have a slope higher than the

peak threshold value (Dionex, 1997). Several values of peak threshold were tested to

optimise the best value for accurate F- determination. It is given in Table 4.34.


105

Table 4.34 Optimisation of peak threshold

Threshold 1 0.5 0.12 0.06 [F] (mg/L) [F] (mg/L) [F] (mg/L) [F] (mg/L) [F] (mg/L)

Known Measured %

Error Measured%

Error Measured%

Error Measured %

Error0.3 0.25 -16.7 0.25 -16.7 0.3 0.0 0.32 6.7 0.7 0.65 -7.1 0.65 -7.1 0.68 -2.9 0.7 0.0 2 1.92 -4.0 1.94 -3.0 1.99 -0.5 1.98 -1.0

Threshold 0.03 0.01 0.0025 0.001 [F] (mg/L) [F] (mg/L) [F] (mg/L) [F] (mg/L) [F] (mg/L)

Known Measured %

Error Measured%

Error Measured%

Error Measured %

Error0.3 0.32 6.7 0.30 0.0 0.30 0.0 0.30 0.0 0.7 0.70 -0.4 0.66 -5.7 0.66 -5.7 0.66 -5.7 2 1.98 -0.8 1.94 -3.0 1.94 -3.0 1.94 -3.0

0.3, 0.7 and 2 mg/L F- test solutions were examined with various peak threshold values

and this table shows that the results were varying according to the peak threshold value.

The results obtained using the peak threshold value 1 and 0.5 are very similar and also

between 0.12 and 0.03. The results obtained using the peak threshold value 0.01, 0.0025

and 0.001 are exactly same, but the best value in this table is 0.12. It is reasonable to use

the peak threshold between 0.12 and 0.03 but 0.12 is recommended.

c. Interference study

IonPac AS4A, recommended in US EPA 300.0 (1993), is appropriate for the analysis of

anions in low ionic strength, well-characterized samples, such as drinking and surface

waters. However, this column is not recommended for the analysis of F- in more complex

samples, e.g. containing organic acids, such as acetate and formate. However, the AS 14

column provides improved F- resolution from the system void peak and complete

resolution of F- from formate and/or acetate (Application Note 133; Jackson, 2001). The

improved selectivity, along with higher capacity, makes the AS 14 column a better choice

for higher ionic strength water samples and more complex matrices, such as domestic and

industrial wastewaters.


106

Acetate and formate ions elute close to F- peak, acetate is followed by the F- peak and

formate followed by acetate. The chromatogram of 1 mg/L F- solution, containing

interfering substances, 5 mg/L acetate and 5 mg/L formate, is shown in Figure 4.14.

Figure 4.14 Acetate and formate interference

F- solutions spiked with the interfering substances, acetate and formate, were measured

and the results are presented in Table 4.35.

Table 4.35 Acetate and formate interference on F- determination Interfering [F] Known [F] Measured [F] Average % Substances

acetate+formate (mg/L)

(mg/L) (mg/L) (mg/L) Error

5 2 1.918 5 2 1.874

1.896 -5.2

5 1 0.962 5 1 0.957

0.960 -4.3

1 2 1.937 1 2 1.947

1.942 -2.7

1 1.5 1.450 1 1.5 1.468

1.459 -2.1

The results show negligible interference at the 1 mg/L acetate and formate concentration.

At higher levels of interfering substances, 5 mg/L acetate and 5 mg/L formate, slightly

lower recoveries were measured. This meant that the acetate and formate became

significant interference for the F- determination. Natural samples normally do not contain

the organic acids, formate and acetate, but if present the analyst should be aware of

possible interference effects.


107

d. Analytical parameters

The definition of each analytical parameter, given below, is given in Section 4.2.3.e.

i) Repeatability

To determine repeatability, 0.2, 0.3, 0.7, 1.0 and 2.0 mg/L F test solutions were measured

10 times each. The results are given in Table 4.36 and the repeatability was expressed in

terms of % RSD.

Table 4.36 The repeatability of IC

[F] Conc. Mean intensity % RSD (mg/L) (µS)

SD (N=10) (Repeatability)

0.2 40541.8 2264.9 5.6 0.3 57566.5 2049.4 3.6 0.7 132797.4 1747.2 1.3 1.0 188935.9 2910.4 1.5 2.0 388150.6 2754.2 0.7

As the concentration increases, the repeatability becomes better, as was expected. This

phenomenon was also observed in the ISE measurements. The repeatability in the case of

ISE, given in Table 4.32, was 0.2 % and 0.3 % at 1 mg/L and 2 mg/L, respectively. These

values are significantly better than the values found for the IC method.

ii) Method Detection Limit (MDL)

MDL could not be determined using the procedure based on 3x the standard deviation of

several blank measurements, because in IC a blank sample results in no peak and thus

cannot be integrated. Therefore, alternative procedures for the determination of detection

limit were investigated.

In the first procedure a calibration curve was derived from a set of successively diluted

standards and the detection limits obtained experimentally. The qualitative detection limit

for F- was taken as the lowest concentration at which a peak could be distinguished from

the background. This was 0.05 mg/L. The quantitative detection limit was determined as

the lowest concentration of F- needed to produce a peak that could be integrated and

plotted on a linear calibration curve. The quantitative detection limit was determined as

0.03 mg/L.


108

In the second procedure, the Method Detection Limit (MDL) was calculated according to

the procedure described in US EPA method 300.1 (1997). The MDL is estimated by

injecting seven replicates of reagent water fortified at a concentration of three to five

times the estimated instrument detection limit. The MDL is then calculated as follows:

( ) ( )

Average

t SMDLI×

= (4.6)

Where, t = Student’s t value for a 99% confidence level and a standard deviation estimate with n-1 degrees if freedom [t = 3.14 for seven replicates], S = standard deviation of the replicates analyses, I = Intensity

Table 4.37 Determination of MDL using 0.05 mg/L F-

Injection Intensity (µS) 1 12195 2 11058 3 12275 4 12077 5 12078 6 11978 7 11571

Average Intensity 11890.3 Standard deviation 430.75

MDL 0.11

MDL value according to this procedure is 0.11 mg/L.

iii) Linear range and working range

The calibration line for F- shows linearity from 0.05 mg/L (Table 4.28). A working range

of 0.1-100 mg/L is recommended for F- determination (Application Note 133). However,

when low-level F- is measured, working range of 0.1-5 mg/L is recommended.

4.4 Comparison of two methods (ISE, IC)

The two methods, ISE and IC, applied to F- determination, are compared below.

Calibration curves for both methods were determined using the same set of standards, 0.1,


109

0.5 and 3 mg/L F- and the same samples with known F- concentration, 0.2, 0.7, 1 and 2

mg/L, were analysed. The results are summarised in Table 4.38.

Table 4.38 Comparison of F- determination by ISE and IC Method ISE IC

[F] Known [F] Avg. SD (n=5) % Error [F] Avg. SD (n=5) % Error (mg/L) Conc. Precision Accuracy Conc. Precision Accuracy

0.2 0.199 0.002 -0.5 0.189 0.008 -5.5 0.7 0.684 0.004 -2.3 0.650 0.012 -7.1 1 0.987 0.004 -1.3 0.952 0.015 -4.8 2 2.004 0.005 0.2 1.950 0.008 -2.5

These results seem to suggest that ISE method is superior to IC method for F-

determination in terms of both precision and accuracy. The correlation plot of ISE and IC

for these results is given in Figure 4.15.

Figure 4.15 Correlation plot of ISE and IC data

y = 0.9775x - 0.0115R2 = 0.9999

0.0

0.5

1.0

1.5

2.0

2.5

0 0.5 1 1.5 2 2.5

ISE data (mg/L)

IC d

ata

(mg/

L)

Figure 4.15 shows that the results obtained in this laboratory by the ISE method correlate

well with those obtained by the IC method. The IC method gives a slightly lower

concentration than that obtained by the ISE method (slope of regression line=0.9775)

with a correlation coefficient 0.9999.


110

4.5 Methodology recommendations

4.5.1 ISE ISE is a simple and accurate method for low level F- determination. Electrode drift is one

of the main problems, especially at low concentrations. To minimise the electrode drift

and improve the equilibrium time, F- ion selective electrodes are recommended to be

stored dry in air and the reference electrode in deionised water. Cleaning the electrode

regularly with F- toothpaste is a very effective way to minimize the electrode drift as well

as improve the linearity of electrode response. It is recommended to use low level

standards, such as 0.1, 1 and 3 mg/L, for low level determination. TISAB is normally

added to samples and standards in ISE measurement in order to maintain a constant ionic

strength for both standards and samples, optimise the pH, and de-complex interference

ions. Low Level TISAB and TISAB III are available for low-level measurement, but

TISAB III is recommended in natural samples, which contain many unknown

interferences. Since Low Level TISAB does not contain de-complexation agent, such as

CDTA, it can not be used in natural samples. Most ionic interferences, except the Al ion,

which readily form stable complexes with F- ions, can be removed by using TISAB III.

The Al tolerance for 95 % F- recovery is 0.2 mg/L and 0.1 mg/L for 1 mg/L F- and 0.2

mg/L F- , respectively. Tri-ammonium citrate (TAC) buffer, can improve the Al tolerance

up to 5 mg/L for 94% F- recovery, and is recommended in presence of high

concentrations of Al. However, TISAB III can be applied for most of natural samples

since the Al level is mostly below 0.2 mg/L. Synergistic effects in ISE have been

demonstrated when many interfering substances are present.

4.5.2 IC It is necessary to allow stabilisation time on the IC instrument of at least 30 minutes. IC

base line drift, determined experimentally, was 0.6 nS/minute, which was acceptable. A

3-point calibration at low concentrations, such as 0.1, 1 and 3 mg/L, is recommended. A

25 µL injection volume yielded better results than a 50 µL sample loop. The optimised

peak threshold value was 0.12. IC software calculates the calibration function by forcing


111

the y-intercept through zero. Because of this fact, it is essential to have proper calibration

line where the y-intercept is close to zero. If not, manual calibration line calculation must

be performed. 5 minutes run time is enough to analyse the samples for F- or even acetate

and formate, since these ions are also eluted within 5 minutes. Possible interference on IC

is co-elution of organic acid anions, such as acetate and formate. The interference of 1

mg/L acetate together with 1 mg/L formate was not significant. The IC method is

appropriate for the analysis of anions in low ionic strength, well-characterized samples,

such as drinking and surface waters, since natural water samples normally do not contain

formate and acetate.

4.6 Conclusion

Method validation and the development of standard ISE and IC procedures for the

accurate determination of low-level F-, was conducted. Although the results, obtained by

the ISE and IC, produced a good correlation, ISE gave more accurate and precise results.

As mentioned in Chapter 1, the accuracy of the F- determination is a very important

aspect in terms of fluoridation. Therefore, ISE is recommended for low-level F-

determination.

Comparison of IC and ISE for the analysis of natural and drinking water ______________________________________________________________________________________

112

Chapter 5

Comparison of IC and ISE for the analysis of natural and drinking water

5.1 Introduction

Chapter 4 focused on the evaluation of the ISE and IC methods for the determination of

F-. A detailed study of calibration and measurement procedures and ways of dealing with

matrix interferences have been carried out to ensure that the best possible analytical

methodologies are followed throughout this study. In Chapter 5 the methodologies

developed in Chapter 4 are applied to the analysis of natural waters such as river water

(Vaal and Crocodile rivers), dam water (Hartbeespoort Dam), and drinking water

(Johannesburg municipal tap water). The main objective was to evaluate the performance

of the chosen methods in the analysis of real samples and to assess the effect of the

sample matrix on the accuracy of F- determinations.

5.2 Laboratory Fortified sample Matrix (LFM) A useful procedure to assess whether a method is prone to matrix effects, is the

laboratory fortified sample matrix method, which is a quality control procedure used in

the EPA approved method for F- determination (EPA Method 300.1, Daniel P.H., 1997).

The Laboratory Fortified sample Matrix (LFM) is an aliquot of an environmental sample

to which known quantities of the method analytes are added in the laboratory. The LFM

is analyzed exactly like a sample, and its purpose is to determine whether the sample

matrix contributes bias to the analytical results. The background concentrations of the

analytes in the sample matrix must be determined in a separate aliquot and the measured

values in the LFM corrected for background concentrations. The LFM should be

prepared at concentrations no greater than five times the highest concentration observed

in any field sample.


113

Percent Recovery is calculated using the following equation:

R = 100×−S

CCs (5.1)

Where, R = % recovery, Cs = fortified sample concentration, C = sample background concentration, S = concentration equivalent of analyte added to sample

According to the U.S. EAP method 300.1, the recovery limit ranges are from 75 to 125%.

If the recovery of any analyte falls outside the designated LFM recovery range, the

recovery problem encountered with the LFM is judged to be matrix induced and the

results for that sample and the LFM are reported with a “matrix induced bias” qualifier.

5.3 Experimental 5.3.1 Instrumentation a. ISE

An ORION 940 auto chemistry system, equipped with a 9404 sc ORION F- electrode

connected to an Ag/AgCl Model 90-01 single junction reference electrode, was used for

the measurement of F- in each sample. TISAB III (Orion) was used as a Total Ionic

Strength Adjustment Buffer.

b. IC

A Dionex DX-120 Ion Chromatographic system, equipped with a Dionex anion exchange

column system (AG 14A+AS14A), was used for the determination of F- in each sample.

The Standard HCO3-/CO3

2- (1.0 mM HCO3- + 3.5mM CO3

2-) eluent, recommended for

anion determinations with this column, was used in IC measurement. All the samples and

eluent were filtered through 0.22 µm membrane filters, prior to analysis.

c. ICP-OES

ICP-OES was used to determine the cation elemental composition in tap water. Multi-

element calibration was used for the determination of the elements: Li, B, Na, Mg, Al, Ca,

Cr, Mn, Fe, Co, Ni, Cu, Zn, Sr, Ba, Cd, Pb and Bi using a Varian Liberty 110 ICP


114

Emission spectrometer. A 1000 mg/L ICP multi-element standard solution IV (Merck)

was used to prepare calibration standards by appropriate dilution with 2% HNO3. The

experimental conditions used in ICP-OES measurements are summarized in Table 5.1.

Table 5.1 ICP-OES operational conditions of multi-element determination

Parameter Setting

Viewing Height 6 mm

Search window 0.04 nm

Generator power 1.20 kW

Plasma gas flow rate 15.0 L/min

Auxiliary gas flow rate 1.50 L/min

Pump speed 25.0 rpm

d. ICP-MS

Inductively coupled plasma mass spectrometry (ICP-MS) was used to determine the

cation elemental composition in the Crocodile and Vaal rivers and Hartbeesport Dam

water matrices. Multi-element calibration was also used for the determination of the

elements: Na, Mg, Al, Si, K, Ti, V, Cr, Mn, Fe, Ni, Zn, Cu, Ga, As, Se, Sr, Rh, Cd, Sb

and Pb using a X Series ICP-MS spectrometer (Thermo Elemental). A 1000 mg/L ICP

multi-element standard solution IV (Merck) was used to prepare calibration standards by

appropriate dilution with 2% HNO3.

5.3.2 Reagents, standard solutions and sample preparation

All reagents were of AR grade and high quality deionised water (18 MΩ) was used for

preparing all solutions. Samples and eluents were filtered through 0.22 µm membrane

filters.

The following solutions were prepared:

• 1000 mg/L F- Stock solution: 0.221 g NaF was dissolved in deionised water and

diluted to 100 ml in a polystyrene volumetric flask.

• TISAB lll (Thermo Orion, Cat. No. 940911)

• Single Junction Reference Electrode Filling solution (contents 60 ml, 900001)


115

• 0.5 M CO32- stock solution: 26.49 g Na2CO3 (AnalaR, BDH Chemicals Ltd, Poole,

England) was dissolved in 500 ml deionised water.

• 0.5 M HCO3- stock solution: 21.00 g NaHCO3 (Saarchem(Pty)Ltd) was dissolved

in 500 ml deionised water.

• Eluent (3.5 mM CO32-, 1.0 mM HCO3

-): 12 ml 0.5 M CO32- and 4 ml 0.5 M HCO3

-

solutions were combined and diluted to 2000 ml.

• 1000 mg/L ICP Multi-element standard solution IV (Merck)

• LFM samples: The filtered natural water samples were spiked with appropriate

amounts of NaF to obtain a series of LFM samples for each of the sample types,

containing 0.1, 0.2, 0.5 and 1 mg/L F-.

5.3.3 Experimental procedure

The methods, ISE and IC, were used for the determination of F- concentrations in the

fortified samples. The background F- concentration was determined in the unspiked

sample. Calibration Check Standards (CCS) were measured as recommended in the U.S.

EPA method 300.1. The 3 different kinds of CCS were:

• Initial calibration check standard: It verified the previously established calibration

curve.

• Continuing calibration check standard: It confirmed accurate analyte quantitation

for the previous samples analyzed. It should be measured in every tenth sample

measurement.

• End calibration check standard: It verified the previously established calibration

curve and confirmed accurate analyte quantitation for all the samples since the last

one.

a. ISE

2 ml of TISAB III was added to the 20 ml of test and standard solutions, as recommended

in the manufacturer’s procedure (TISAB III : sample volume = 1:10 for the determination

of F-).


116

b. IC

The standard HCO3-/CO3

2- (1.0 mM HCO3- + 3.5mM CO3

2-) eluent, recommended for

anion determination with AG 14A+AS14A column, was used in every determination.

5.4 Results and discussion

The % recovery obtained in the analysis of the LFM samples is summarized in Table 5.2.

Table 5.2 % Recovery in the analysis of the LFM samples

LFM % recovery Method ISE IC Matrix Back Added Fortified R* Back Added Fortified R* Type Ground [F] Conc. % Ground [F] Conc. %

[F](C) (s) (Cs) [F](C) (s) (Cs) 0.1 0.353 96.3 0.1 0.336 80.50.2 0.452 97.4 0.2 0.424 84.50.5 0.739 96.5 0.5 0.692 87.3

Crocodile River Water

0.257

1.0 1.25 99.3

0.255

1.0 1.196 94.10.1 0.349 93.9 0.1 0.339 92.00.2 0.45 97.5 0.2 0.432 92.30.5 0.748 98.7 0.5 0.712 93.0

Vaal River Water

0.255

1.0 1.235 98.0

0.247

1.0 1.19 94.30.1 0.356 90.2 0.1 0.339 85.20.2 0.454 93.9 0.2 0.43 88.00.5 0.764 99.6 0.5 0.758 100.7

Hart- beesport Dam water

0.266

1.0 1.265 99.9

0.254

1.0 1.258 100.40.1 0.27 102.7 0.1 0.245 82.80.2 0.36 96.4 0.2 0.349 93.30.5 0.64 94.5 0.5 0.659 99.3

Tap water 0.167

1.0 1.178 101.1

0.162

1.0 1.155 99.3* R is calculated by the equation 5.1 in section 5.2.

The % recoveries in Table 5.2 ranged between 75 to 120 %, which is the acceptable

criterion described in the U.S. EPA method 300.1. This meant that there were no “matrix

induced bias” qualifiers. The results obtained by the ISE method showed better accuracy

than those of IC. In general, as the added F- concentration decreased, the % recoveries

also decreased. This fact indicated that careful attention was needed when low

concentration of F- was determined by these methods. The % recoveries of each matrix

graphically are shown in Figure 5.1 to Figure 5.4 for comparison of the two methods.


117

Figure 5.1 Method comparison of the LFM % Recovery in Crocodile River

0

20

40

60

80

100

120

0.1 0.2 0.5 1Added [F] (mg/l)

LFM

% R

ecov

ery

ISE IC

Figure 5.2 Method comparison of the LFM % Recovery in Vaal River

020406080

100120

0.1 0.2 0.5 1

Added [F] (mg/l)

LFM

% R

ecov

ery

ISE IC

Figure 5.3 Method comparison of the LFM % Recovery in Hartbeespoort Dam

020406080

100120

0.1 0.2 0.5 1Added [F] (mg/l)

LFM

% R

ecov

ery

ISE IC

Figure 5.4 Method comparison of the LFM % Recovery in Tap water

0

2 0

4 0

6 0

8 0

1 0 0

1 2 0

0 .1 0 .2 0 .5 1A d d e d [ F ] (m g /l )

LFM

% R

ecov

ery

IS E IC


118

All histograms, except 0.5 mg/L fortified sample from Tap water, clearly show that the %

recovery obtained by the ISE method was closer to 100% recovery than that by IC

method. This indicated that F- determination in natural samples using ISE yielded more

accurate result.

The composition of each matrix is summarized in Table 5.3 and the typical

chromatograms of IC for each matrix are given in Figure 5.5-5.8.

Table 5.3 Summary of the composition in each matrix

Matrix type Types Elements Crocodile

River Hartbeespoort

Dam Vaal River

Tap water

*Cations Na+ 24.6 25.78 33.38 12.2 (mg/L) Mg2+ 14.42 14.84 21.62 5.86

Al3+ 0.13 0.09 0.1 <DL*** Ca2+ 30.99 29.62 63.23 13.63 Fe2+ <DL 0.053 0.25 0.12

**Anions Cl- 47.2 46.5 50.5 10.67 (mg/L) NO2

- <DL <DL <DL 0.10 Br- 0.41 0.40 0.47 0.04 NO3

- 1.81 0.08 10.23 1.64 PO4

3- 0.21 <DL 0.72 <DL SO4

2- 35.3 35.2 124.6 11.91 *Cations were determined by ICP-OES for tap water and ICP-MS for Crocodile and Vaal Rivers and Hartbeesport Dam. **IC (DIONEX) determined anions. ***<DL: below detection limit

Figure 5.5 Typical anion chromatogram of the Hartbeespoort Dam water

10 times dilution was done since the concentration of Cl- and SO4

2- were too high, which were out of

the calibration range (section 3.2.2.c.ii)


119

Figure 5.6 Typical anion chromatogram of the Crocodile River water

Figure 5.7 Typical anion chromatogram of the Vaal River water

Figure 5.8 Typical anion chromatogram of Tap water

The Table 5.3 shows that there were various kinds of cations, Na+, Mg2+, Al3+, Ca2+ and

Fe2+, and anions, Cl-, NO2-, Br-, NO3

-, PO43- and SO4

2-, in each matrix. Among them,

there were Al3+ and Fe2+, which were found as influential interferences for the

determination of F- using ISE method in Section 4.2.3.d. However, the Al concentrations,

0.13, 0.09, 0.10 and below detection limit, for Crocodile river, Hartbeesport Dam, Vaal

river and Tap water, respectively, were not high enough to cause significant interference


120

in the measurement of F- using ISE. Similarly, the Fe concentrations, below detection

limit, 0.053, 0.25 and 0.12, for Crocodile River, Hartbeesport Dam, Vaal River and Tap

water, respectively, were also not expected to cause significant interference. Therefore

the % recoveries, obtained by ISE method, were very good.

Compared to the results obtained by ISE, IC results showed an underestimation of F-

concentration. As described in Section 4.3.3.c, the underestimated results from IC

measurement could be explained in terms of interfering anions, such as the formate and

acetate anions, eluting close to the F- peak. However, there was no significant

interference from formate and acetate in each matrix in Table 5.3 and Figure 5.5-5.8.

Another possible reason could be Al interference. According to the description in Table

4.19, a 100% F- recovery could not be obtained in the presence of Al, if the concentration

was more than 0.2 mg/L, even using TISAB III that contained a decomplexing agent such

as CDTA. Since there was no decomplexing agent in the IC eluent, underestimated

results could be expected. However, it was found that no Al-F complex forms at pH 9 (IC

conditions) with the less than 19 mg/L F- and 1 mg/L or 10 mg/L Al according to the

theoretical distribution curve for the species in the Al-F-OH system calculated using the

computer programme Species-Calc (Prof Hans Rohwer, Mandela Metropolitan

University of Port Elizabeth) and shown in Figure 5.9 and Figure 5.10.

Figure 5.9 Distribution curves for Al-F-OH system

Where, the values of x-axis, 1=190000mg/L, 0=19000 mg/L, -2=190mg/L, -4=1.9 mg/L F-

[Al] = 10 mg/L at pH 9


121

Figure 5.10 Distribution curves for Al-F-OH system

Where, [Al] = 1 mg/L at pH 9

These diagrams show a very similar shape at 1 mg/L and 10 mg/L Al and were assumed

to be also valid at 0.2 mg/L Al which was the approximate background concentration of

each matrix, 0.257, 0.255, 0.266 and 0.17 in Crocodile River, Hartbeesport Dam, Vaal

River and Tap water, respectively. This meant that Al could not be the reason for the

underestimated value of F- determination using IC method.

The comparison study of two methods ISE and IC, on laboratory deionised water matrix

in Section 4.4, showed that the accuracy of IC method for F- determination at low level,

was poor. The values of accuracy at 0.2 mg/L and 0.7 mg/L were –5.5 % and –7.1 %,

respectively. When the IC method was applied to the natural sample matrix, the accuracy

at low level, such as 0.3 mg/L and 0.4 mg/L after fortification, became worse (see Table

5.2). This meant that the IC method for F- determination seemed to be more prone to the

influence of the matrix than the ISE method at low concentration levels.

5.5 Conclusion

The evaluation of the ISE and IC methods for the determination of F- in a deionised water

matrix was investigated in Chapter 4. In Chapter 5, the methodologies developed in

Chapter 4 were applied to the analysis of natural waters such as river water (Vaal and


122

Crocodile Rivers), Dam water (Hartbeespoort Dam), and drinking water (Johannesburg

municipal tap water). The performance of two methods, ISE and IC, in the analysis of

real samples, was investigated by analysing LFM samples. The % recoveries of all LFM

samples were within the acceptable range, which was between 75 to 120 % given in the

U.S. EPA method 300.1. All the values, except the 0.5 mg/L fortified sample from Tap

water, in Table 5.2, clearly show that the % recovery obtained by the ISE method was

closer to 100% recovery than that by the IC, which meant ISE method in natural sample

yielded a more accurate result. The IC method seemed to be more affected by the

influence of the matrix. ISE methodology was found to be preferable for the accurate

determination of F- in natural water sample matrices as well as a deionised water matrix

as was concluded in Chapter 4.

Inter-laboratory study (SABS water-check proficiency testing programme) ______________________________________________________________________________

123

Chapter 6

Inter-laboratory study (SABS water-check proficiency testing programme)

6.1 Introduction

The South African Bureau of Standards (SABS), Test House runs an ongoing Water

Check Programme. Water-Check is a high frequency inter-laboratory proficiency testing

programme with the objective of providing a rapid report-back service to participants for

self evaluation. Participation in the scheme was an important part of this study because it

provided the opportunity to obtain information about the current status of fluoride (F-)

determination in South African laboratories. The SABS was also requested to include

additional F- containing solutions, specially designed to test possible matrix effects

during F- determination. This provided invaluable data and proved to be very useful in

formulating the final recommendations and conclusions. Participation in this scheme was

therefore done for the following reasons:

i) To have the opportunity for the inclusion of special samples in the Water Check

programme to test the response of different analytical methods for

F- determination to matrices containing known interfering elements such as Al

ii) To evaluate the proficiency of South African analytical laboratories for

F- determination

iii) To explain trends in the results in terms of analytical pitfalls in F- determination

and to make recommendations for accurate F- determinations

iv) To check proficiency of the procedures developed in this study

The determination of F- was included in this study, because F- was the target element in

the public water fluoridation programme envisaged for South Africa. It was important to

investigate the chemistry and analytical chemistry of F- before implementation of the

fluoridation.


124

6.1.1 Programme design The programme is divided into three categories, each category being scheduled on a

quarterly basis. Real, spiked and /or synthetic samples are prepared for each study.

Table 6.1 SABS water-check proficiency testing programme design Group number Determinants Scheduled

1

Heavy metals in water: aluminium, barium, beryllium, boron, cadmium, chromium, cobalt, copper, iron, lead, manganese, molybdenum, nickel, silicon, strontium, vanadium, zinc, mercury, arsenic and selenium.

January April July

October

2

Nutrients and oxygen demand: nitrogen, nitrate, ammonia, total phosphate, orthophosphate, oxygen absorbed , chemical oxygen demand: dissolved organic carbon and total organic carbon

February May

August November

3

Major constituents in water: pH, conductivity, dissolved solids, calcium, magnesium, sodium, potassium, chloride, fluoride, sulphate, alkalinity, turbidity and nitrate

March June

SeptemberDecember

The programme structure therefore allows a laboratory to choose participation in one, two

or all three groups. All participants are coded for confidentiality purposes.

6.2 Major constituents in water: (Group 3) Participation in Group 3 of the March 2004 Water Check programme was scheduled for

the purposes of this MSc study. The Group 3 programme consists of the analysis of river,

borehole water and synthetic water samples. This group covers the analysis of pH,

conductivity, dissolved solids, calcium, magnesium, sodium, potassium, chloride, sulfate,

alkalinity, fluoride, nitrate and turbidity.

6.2.1 Samples for the Group 3 programme

a. Sample preparation

Two natural river and borehole water samples and 6 synthetic samples listed in Table 6.2

were included in the sample batch. The synthetic samples, listed in Table 6.3, were

supplied in concentrated form. The interferences added to the samples intended for


125

F- analysis are given in Table 6.4. The addition of interferences was requested by

University of Johannesburg (UJ) for the purposes of the current research.

Table 6.2 Sample contents Sample name Contents

2004/03/1 an unpreserved river water 2004/03/2 an unpreserved borehole water

2004/03/3/4/5/6/7and 8 six synthetic concentrated solutions

Table 6.3 Concentrated solutions of synthetic samples Concentrate solutions (constituted to 10 L)

Determinant Chemicals

dried at 105 ºC

2004/

03/3

2004/

03/4

2004/

03/5

2004/

03/6

2004/

03/7

2004/

03/8

Calcium CaCO3 13.6552 g - - - - -

Magnesium MgSO4 13.8805 g - - - - -

Sodium NaCl 10.4875 g - - - - -

Potassium KCl 2.6322 g - - - - -

Chloride NaCl - - - - - 10.1439 g

Chloride AlCl3 - 4.9418 g - - - -

Sulfate Na2SO4 - 5.0851 g 7.4100 g 10.1413 g 11.8018 g 15.0867 g

Fluoride NaF - 0.1387 g 0.0565 g 0.2798 g 0.8284 g 0.4981 g

Table 6.4 Interferences added to the synthetic samples Concentrate solutions(Constituted to 10 L)

Interference Chemicals dried at 105 ºC

2004/ 03/3

2004/ 03/4

2004/ 03/5

2004/ 03/6

2004/ 03/7

2004/ 03/8

Aluminium AlCl3 Al = 1000 mg/L - 250 ml - - - -

Nitrate

Phosphate

KNO3 + KH2PO4 (Chemicals) NO3+PO4

10000 mg/L

- - 100 ml 100 ml 100 ml -

Formate NaHCO2

HCO2 = 10000 mg/L

- - - 100 ml

100 ml -

Calcium Magnesium

CaCl2 + MgCl2 (Titrisol) Ca+Mg

= 2000 mg/L

100 ml 100 ml -

Sodium Chloride NaCl - - 9.9013 g 8.0182 g 12.238 g -

Samples for analyses were prepared by the participating laboratories by pipetting 20 mL

each of concentrated solution into separate 500ml volumetric flasks and dilution to


126

volume with distilled/ deionised water. All volumetric glassware used in the preparation

of stock standards and samples are Grade A and have been verified as complying with

their relevant standard specification. All chemicals used are AR grade chemicals supplied

by Merck, Riedel-de Haën, Fluka and BDH. Prepared stock standard solution, where used,

are supplied by Merck (Titrosol), Riedel-de-Haën, (Fixanal) or Fluka.

b. Sample dispatch

Samples were dispatched to the participation laboratories on 1 March 2004. The return

date for results was set for 31 March 2004.

6.2.2 Statistical evaluation of results

a. Introduction

The use of Robust Statistics was evaluated in the June 1997 study and found to be

effective in down weighting outlying data without excluding such data from the

evaluation. The Water-Check program now makes use of such statistics for data

evaluation. For this study, all the original values, outliers included, were recaptured

electronically in the reporting process adopted by the SABS.

b. Z-scores

The report on each study consists of a robust statistical analysis of results, with the

inclusion of outliers, against the median or true value. The Z-score for each result is

tabulated and then summarized for each laboratory. The Z-scores are calculated for all

data based on distribution of the results around the Robust Mean (Median), (x- x )/s, or

around the assigned or true value, (x-t)/s. The assigned or true value will only be used

when a synthetic sample is analyzed. The mean value will be used for all other samples,

including spiked samples.

Table 6.5 Z-score criteria Z-score Status

Z<2 Satisfactory 2<Z<3 Questionable Z> 3 Unsatisfactory


127

6.3 Results and discussion of F- determination

6.3.1 F- data collection

For the purposes of the current project, UJ requested the SABS to include a method

information sheet with the samples sent to the participating laboratories. On these

information sheets the laboratories were asked to provide information on the procedures

and methods used for F- determination. All the results and the method information sheets

were collected from SABS. The F- data consisted of 446 F- values obtained from the 66

participating laboratories and the method information sheets were returned from 44

laboratories.

6.3.2 Statistical summary

The statistical summary, which focused on F- data, is presented in Table 6.6. True values

were not available in the natural samples, which were river water (sample 1) and borehole

water (sample 2). Sample 3 that did not contain F-, as shown in Table 6.3, was omitted in

this study.

Table 6.6 Statistical summary

Determinant Sample Number

Truemg/L

Median % Error Robust SD

n*

1 - 0.16 - 0.07 62 2 - 0.20 - 0.06 63 4 0.25 0.22 -12.0 0.12 61 5 0.10 0.13 30.0 0.10 63 6 0.54 0.50 -7.4 0.10 66 7 1.50 1.42 -5.3 0.19 66

Fluoride

as F- in mg/L

8 0.90 0.91 1.1 0.10 65

In natural samples there are many more unknowns that could affect the F- measurements.

Although the Robust SDs are expected to be higher than for synthetic samples, the results

show that Robust SDs are lower in natural samples except for sample 4 and 5 which

contained Al and phosphate interferences, respectively. This information, however, is too


128

limited to confirm that the natural samples are more narrowly distributed, since there are

only 2 natural samples. In the synthetic samples, F- was under-estimated in sample 4, 6

and 7 by –12.0 %, -7.4 % and –5.3 % respectively and over-estimated by 30.0 % in

sample 5. The error in the results for sample 8 was small at only 1.1 %.

To find the reason for the inaccuracies in the determination of F-, the composition of the

synthetic sample matrices were looked at carefully. The concentration of the constituents

of in each synthetic sample, given in Table 6.7, was calculated from the data in Table 6.3

and 6.4

Table 6.7 Composition of the synthetic samples in mg/L Sample number Types Elements

#4 #5 #6 #7 #8 Ca2+ - 0.8 0.8 0.8 - Mg2+ - 0.8 0.8 0.8 -

Cations concentration*

(mg/L) Al3+ 1 - - - - F-(true) 0.25 0.10 0.54 1.50 0.90

F-(median) 0.22 0.13 0.50 1.42 0.91 Formate - - 4 4 -

Cl- 18.77 27.23 22.66 32.9 24.61 NO3

- - 4 4 4 - PO4

3- - 4 4 4 -

Anions

Concentration(mg/L)

SO42- 13.77 20.09 24.47 32.00 40.86

*Concentration values are obtained after 25 times dilution as per instruction

Sample 4 is the only one to contain Al. The under-estimated value in sample 4 was

caused by Al interference (see Section 4.2.3.d). The matrix composition of sample 5, 6

and 7 is similar and contained Ca2+, Mg2+, Cl-, NO3-, PO4

3- and SO42-. Formate was found

in samples 6 and 7 but not in sample 5. F- concentrations were 0.10, 0.54 and 1.50,

respectively. The under-estimated value in samples 6 and 7 seemed to be caused by

formate interference, when the F- is determined using IC as mentioned in section 4.3.3.c.

Better accuracy is shown at higher F- concentrations where a 1.5 mg/L F- concentration in

sample 7 resulted in a small error of 5.3 %. The best accuracy, 1.1 % is shown in sample

8, having a very simple composition with no noticeable interferences. A huge over-

estimated value of 30% was found for sample 5. This sample had a low F- concentration

of 0.1 mg/L in a similar matrix than sample 6 and 7, but no formate. It is significant to


129

notice that the F- concentration in sample 5 was the lowest value of all the synthetic

samples. Therefore a seemingly small change in the second decimal digit can cause a

significant relative error of 30% at this level. There are no obvious interferences that can

cause over-estimated F- value in sample 5, therefore the overestimated results seem to be

caused by measuring error at low-level determination of F-.

6.3.3 Comparison of analytical methods

Data from the method information sheets, returned from 44 out of 66 laboratories, were

analyzed to compare the correlation between accuracy and analytical method. The

methods used by the different laboratories are listed in Table 6.8.

Table 6.8 Analytical methods for F- determination in this study Analytical methods Number of laboratories %

ISE 18 40 IC 16 36

SPAND 8 18 Others 3 7

Total numbers* 45 100 *Lab X: Double-counted since it used both ISE and IC method, therefore total number became 45

Of all the results returned to the SABS, 40 % was obtained by the ISE method, 36 % by

the IC, 8 % by the SPANDS method, and the remaining 7% by other methods. Absolute

Z-scores and analytical method used by each laboratory are summarized in Table 6.9.

Table 6.9 Average Z-score and analytical method Order Lab code Average Z-score (Absolute value) Analytical method

1 X(ISE) 0.18 ISE 2 B104 0.22 ISE 3 B1 0.23 Undisclosed 4 B49 0.24 ISE 5 X(IC) 0.27 IC 6 B108 0.27 ISE 7 B10 0.28 ISE 8 B6 0.30 ISE 9 B22 0.31 ISE

10 B86 0.36 IC 11 B65 0.38 Undisclosed 12 B25 0.40 Undisclosed 13 B63 0.43 IC 14 B23 0.44 SPANDS


130

15 B26 0.44 Undisclosed 16 B3 0.47 ISE 17 B5 0.48 ISE 18 B91 0.49 IC 19 B109 0.49 Undisclosed 20 B11 0.51 ISE 21 B28 0.55 SPANDS 22 B8 0.56 Undisclosed 23 B59 0.58 SPANDS 24 B68 0.59 ISE 25 B130 0.59 Undisclosed 26 B14 0.62 IC 27 B33 0.70 ISE 28 B70 0.72 Undisclosed 29 B100 0.72 Undisclosed 30 B80 0.76 Undisclosed 31 B47 0.77 DR Lange 32 B9 0.79 ISE 33 B34 0.79 IC 34 B39 0.86 Undisclosed 35 B102 0.87 Undisclosed 36 B103 0.93 Undisclosed 37 B19 0.95 Undisclosed 38 B41 0.96 SPANDS 39 B21 1.07 IC 40 B125 1.11 Undisclosed 41 B44 1.13 Probe 42 B99 1.18 SPANDS 43 B112 1.19 SPANDS 44 B101 1.20 SPANDS 45 B123 1.20 Undisclosed 46 B105 1.21 Undisclosed 47 B76 1.28 Undisclosed 48 B117 1.32 IC 49 B45 1.55 Undisclosed 50 B73 1.58 CIA 51 B42 1.62 IC 52 B94 1.65 IC 53 B92 1.70 ISE 54 B132 1.89 IC 55 B16 1.95 IC 56 B74 1.96 ISE 57 B31 2.13 IC 58 B122 2.14 ISE 59 B36 2.23 IC 60 B131 2.37 Undisclosed 61 B115 2.83 ISE 62 B71 3.15 IC 63 B29 3.48 ISE 64 B27 3.88 SPANDS 65 B78 6.66 Undisclosed 66 B116 7.72 Undisclosed 67 B119 13.08 IC


131

The Z-score can be a useful indication for the accuracy of measurement. A smaller Z-

score indicates a better accuracy. The results using the ISE method show good accuracy

and low Z-scores. Among the top 10 lowest Z-scores, 7 laboratories used ISE and 2 used

IC method. The overall average absolute Z-score according to the analytical method is

given in Table 6.10.

Table 6.10 Average absolute Z-score with method Method Average z-score

ISE 0.98 IC 1.33

SPANDS 1.25

The IC Z-score value, 1.33, was obtained after excluding the value of lab B119, being too

high, at 13.08. The average Z-score of SPANDS is slightly lower than that of IC.

The performance of the analytical laboratories in South Africa, participating in this

proficiency testing exercise for the measurement of F-, is summarized in Table 6.11. 84%

of laboratories in this study reported satisfactory results. A graphical presentation of Z-

scores obtained by the 66 laboratories is given in Figure 6.1.

Table 6.11 Laboratory performance Z-score Status Numbers of laboratories %

Z<2 Satisfactory 56 84 2<Z<3 Questionable 5 7 Z> 3 Unsatisfactory 6 9

Total 67 100 Figure 6.1 Average absolute Z–score

Average absolute Z-score for all participants

0

1

2

3

X(ISE)B10

4 B1B49

X(IC)B10

8B10 B6

B22 B86 B65 B25 B63 B23 B26 B3 B5B91

B109B11 B28 B8

B59 B68B13

0B14 B33 B70

B100B80 B47 B9

B34 B39B10

2B10

3B19 B41 B21

B125B44 B99

B112B10

1B12

3B10

5B76

B117B45 B73 B42 B94 B92

B132B16 B74 B31

B122B36

B131B11

5B71 B29 B27 B78

B116B11

9

Lab code

Aver

age

abso

lute

Z-s

core


132

Lab X, the first ranked, was the UJ lab and both results using ISE and IC methods from

this lab are excellent. This showed the proficiency of the procedures developed in this

study.

6.3.4 Determination of composition of natural water samples

a. Experimental method

The water-check samples, received from SABS, were analyzed further to determine

whether elements present in the samples could cause interferences. ICP-OES was used to

determine the cation elemental composition in natural samples 1 and 2. Multi-element

calibration using a Varian Liberty 100 ICP-OES spectrometer was used for the

determination of the elements: Li, B, Na, Mg, Al, Ca, Cr, Mn, Fe, Co, Ni, Cu, Zn, Sr, Ba,

Cd, Pb and Bi. A 1000 mg/L Multi IV (Merck SA) multi-element ICP-OES standard was

used to prepare calibration standards by appropriate dilution with 2% HNO3. The

experimental conditions used in ICP-OES measurements are summarized in Table 6.12.

Table 6.12 ICP-OES operational conditions Parameter Setting

Viewing height 6 mm Search window 0.04 nm

Generator power 1.20 kW Plasma gas flow rate 15.0 L/min

Auxiliary gas flow rate 1.50 L/min Pump speed 25.0 rpm

A Dionex DX-120 Ion Chromatographic system, equipped with a Dionex anion exchange

column system (AG 14A+AS14A), was used for the determination of the anions:: F-,

Formate, Cl-, Br-, NO3-, NO2

-, PO43-, SO4

2- in the natural samples. The standard HCO3

-

/CO32- (1.0 mM HCO3

- + 3.5mM CO32-) eluent recommended for anion determinations

with this column was used in all determinations.

b. Results

The results are summarized in Table 6.13 for those species in sample 1 and 2 where the

concentrations were more than the Method Detection Limit (MDL).


133

Table 6.13 Composition of the natural samples Composition of the natural samples

Types Elements Detection limit #1 #2 Ca2+ 0.0003 36.87 70.83 Mg2+ 0.0010 20.07 45.03 Sr2+ 0.0002 0.038 0.13 Na+ 0.0100 8.52 11.15 Cu2+ 0.0200 <DL 0.020

Cations (mg/L)

Al3+ 0.0150 0.059 0.15 Cl- - 6.27 20.77 Anions

(mg/L) NO3- - 3.39 26.41

SO42- - 12.43 29.11

** <DL: below detection limit

6.3.5 Determination of F- in Water Check samples by ISE and IC (UJ analytical chemistry lab)

The experimental procedure of LFM, used in this analysis, is described in Section 5.2.

a. Results and discussion

The experimental results in this Section were obtained by the author in Lab X, the UJ

Analytical Chemistry Laboratory using the procedures developed and optimized in this

study and are summarized in Table 6.14.

Table 6.14 Sample background measurement Method ISE IC

Samples [F] True

[F] Measured SD %

Recovery[F]

Measured SD % Recovery

#1 0.16 0.141 0.003 88.1 0.152 0.015 95.0 #2 0.20 0.189 0.000 94.5 0.169 0.011 84.5 #4 0.25 0.232 0.004 92.7 0.246 0.012 98.2 #5 0.10 0.103 0.001 103.3 0.109 0.011 108.7 #6 0.54 0.487 0.009 90.1 0.474 0.022 87.8 #7 1.50 1.450 0.012 96.6 1.391 0.015 92.7 #8 0.90 0.879 0.002 97.6 0.849 0.030 94.3

* SD: n=3, F concentration unit: mg/L

True values for sample 1(river water) and 2(bore hole water) could not be determined,

since these samples were natural samples. Therefore, the recoveries for sample 1 and 2

were calculated using the median values shown in Table 6.6. A better % recovery was


134

obtained using ISE, except for sample 1 (natural sample) and 4, which contained Al. The

results show that the ISE method achieved a better accuracy and precision.

To check the possible effect of the matrix on the determination of F- in the Water Check

samples, the samples were spiked with known amounts of F, to prepare a set of laboratory

fortified matrix (LFM) check samples. The recoveries obtained in the analysis of these

samples are given in Table 6.15.

Table 6.15 LFM % recovery Analytical method

Ion selective electrode Ion chromatography Samples Fortified bB Added cR Samples Fortified B Added R Number aConc. (Cs) [F] (C) [F](s) % Number Conc. (Cs) [F](C) [F](s) %

Unfiltered #1 0.607 0.141 0.3 155.0 - - - - -

Unfiltered after 24h

#1 1.033 0.138 0.3 298.3 - - - - -

Filtered #1 0.412 0.140 0.3 90.7 #1 0.445 0.152 0.3 97.7

Filtered UV #1 0.415 0.141 0.3 91.3 - - - - -

#2 0.566 0.189 0.4 94.2 #2 0.557 0.169 0.4 97.0#4 0.576 0.232 0.4 86.2 #4 0.656 0.246 0.4 102.5#5 0.288 0.103 0.2 92.4 #5 0.292 0.109 0.2 91.6#6 1.472 0.487 1 98.5 #6 1.437 0.474 1 96.3#7 4.476 1.450 3 100.9 #7 4.422 1.391 3 101.0#8 2.879 0.879 2 100.0 #8 2.874 0.849 2 101.3

aConc.: F- concentration unit: mg/L, bB: Backgraound, cR: Recovery

The results, except sample 1 using ISE, displayed acceptable recoveries (75-125%, US

EPA method 300.1) for the F- in all matrices. This indicates that the analytical methods,

ISE and IC, are both acceptable methods to measure F- in water samples.

The initial % recovery of sample 1 (river water) was 155.0 %, which was out of the

reasonable recovery range (75-125%). Therefore sample 1 was further analyzed using

three different methods with ISE. In the first, the sample was measured 24 hours after

fortification, in the second, the sample was filtered first and then measured after

fortification, while in the third the sample was measured after fortification, filtration and

UV treatment. The filtration was chosen with the expectation that removal of any

unknown factors that could affect the F- determination in the natural samples, would

improve the accuracy. UV lamp exposure was tested with the expectation of breaking up


135

the organic material in natural samples. The recoveries improved to 90.7% and 91.3%,

respectively, after filtration and filtration with UV lamp exposure. However, the result 24

hours after fortification was the least impressive, 298 %, and was in fact worse than

before. This indicated that filtration was necessary in natural river water samples for

accurate F- measurement, although not the case in the synthetic samples with ISE. Sample

2 (borehole water), another natural sample, showed a reasonable LFM recovery value,

even without filtration. It is therefore clear that the matrix could affect the analytical

results for F- determination, even though the analytical method was accurate enough.

When the ion composition of these samples was compared (Table 6.13), sample 2

contained a larger concentration of each element. Even though a lower concentration of

ions occurred in the river water sample, the LFM recovery value was out of the

acceptable recovery range. This means that there were other factors that caused the

erroneous LFM value in the river water sample. This problem could be solved by

filtration.

In the case of synthetic samples, the recovery for sample 4, which contained Al

interference, was lower 86.2%, for ISE determination, than for the other samples. This

occurred because of the formation of Al-F complexes and the less than 100% efficiency

of the decomplexing agent to break up the complex (see Section 4.2.3.d). In the case of

IC the recovery of sample 4 was 102.5 %. This is because the Al can complex with F-

strongly in the acidic medium used in ISE applied to, but no Al-F complex forms at pH 9

(IC conditions) with the less than 19 mg/L F- and 1 mg/L or 10 mg/L Al as shown in

Figure 5.9 and Figure 5.10. This means that Al is not problematic in the determination of

F-, using IC in terms of interference, but the precipitation of Al hydroxide in basic

medium should be considered, since precipitation could damage the column.

The recoveries for sample 5 were lower for both methods 92.4 % and 91.6 %, because of

a very low-level of F-, 0.10 mg/L.


136

The other synthetic samples, 6, 7 and 8, yielded good results, close to 100 % recovery.

These samples did not display any significant interference that could result in serious

error.

6.4 Conclusion The Inter-laboratory study undertaken in collaboration with SABS (Chapter 6) was

carried out to evaluate the proficiency of South African analytical laboratories and to

check the proficiency of the procedures developed in this study as well as to establish

how different analytical methods for F- determination cope with interferences such as Al.

SABS water check F- samples data, from 66 participant laboratories, were analyzed

statistically. The most serious errors (+30%) were obtained in sample 5 containing 0.1

mg/L F-, which was very low-level and in sample 4 which contained 2.5 mg/L F- together

with 1mg/L Al interference. The F- results were obtained using various methods: 40%

from ISE, 35% from IC and 7% from SPADNS method. The ISE method is mostly used

in South African laboratories for determination of F-. The 84% of participants gave a

satisfactory result where the Z-score, the parameter for the accuracy, was less than 2. The

best average absolute Z-score was achieved when the ISE method was applied to

determine the F- in the water samples. This showed a good agreement with the conclusion

from Chapter 4 and Chapter 5, which indicated that ISE was the recommended

methodology for the accurate F- determination at low level. Lab X (UJ Analytical

chemistry lab) which applied the procedure developed in Chapter 4, showed the best

accuracy using ISE. This proved the proficiency of the procedures developed in this study.

It should be kept in mind that it is to be expected that it would be the “best” laboratories

or accredited laboratories that would be participating in proficiency testing schemes. The

relatively good results obtained in this study for the proficiency of laboratories in South

Africa to determine low level F-, could therefore reflect an over optimistic picture.

Speciation of fluoro-aluminium species by IC coupled ICP-OES and ICP-MS ______________________________________________________________________________

137

Chapter 7

Speciation of fluoro-aluminium species by Chromatography coupled ICP-OES and ICP-MS

7.1 Introduction The presence of Aluminium (Al) species in drinking water constitutes a serious health

risk because the possibility of an association between Al and neurophathological diseases,

including presenile dementia and Alzheimer’s disease, is frequently hypothesised

(Schecher and Driscoll, 1988; Crapper and Boni, 1980; Davidson et al., 1982; Martyn et

al., 1989; Gardner and Gunn, 1991; Jekel, 1991). This is a well-known problem and most

countries have water regulations that permit only very low levels of Al to be present in

drinking water. The international quality standards for Al in drinking water are shown in

Table 7.1. The South African Drinking Water Standard sets an Al maximum level of 0.5

mg/L.

Table 7.1 International water quality standards for Al in drinking water

Organisation/ Government

Guideline levels

(mg/L)

Maximum acceptable Concentration

(mg/L) Reference

South Africa 0.3 0.5 Rand water

WHO 0.2 ------ Letterman and Driscoll (1994)

EEC 0.05 0.2 Letterman and Driscoll (1994)

Belgium 0.05 0.1 Nilsson (1990) FRG 0.05 0.2 Nilsson (1990)

Sweden ------ 0.1 Nilsson (1990) Switzerland 0.05 0.5 Nilsson (1990)

USEPA 0.05 0.2 Federal Register (1991)AWWA Recommended Operating

Level ------ 0.2 Letterman and Driscoll (1994)

1986 Proposed Illinois Regulation 0.1 ------ Driscoll and Letterman (1988)

New York State guidelines on Al in filtered water for pilot-plant studies.

Minimum percent of recorded values

95 75

<0.15 <0.09

------ ------

Letterman and Driscoll (1994)


138

50 <0.05 ------ Finland ------ 0.2 Nilsson (1990)

Denmark 0.05 0.2 Nilsson (1990) Austria ------ 0.2 Nilsson (1990)

California Code of Regulations (maximum contaminant level) ------ 1.0 Letterman and Driscoll

(1994) This table was cited from Srinivasan et al. (1999)

The fluoridation of water could lead to a remobilisation of scale from municipal pipes.

This scale contains Al hydroxide or oxide precipitates that can dissolve as fluoro-

aluminium (AlFn(3-n)+) or hydroxo-fluoro aluminium species (Al(OH)mFn

3-m-n), if F- is

present. This process could increase the levels of dissolved Al in the water through the

formation of fluoro-aluminium complexes. Baker and Schofield (1982) indicated that the

OH- and F- complexes of Al are highly labile and may be more bioavailable and harmful

than organic or particulate forms of Al. Al migration in soils can be enhanced by the

presence of these fluoro-aluminium species increasing the overall risk. Therefore it was

necessary to design the modelling of the Al-F distribution system.

Alkaline conditions will favour the formation of mixed fluoro-hydroxo complexes (Al

(OH)mFn 3-m-n) while acid conditions will favour fluoro-aluminates (AlFn

(3-n)+). Accurate

formation constants for these mixed complexes are required in order to model the system.

It would also be beneficial to develop speciation methods to determine the individual

species. Therefore, this chapter focuses on the development of an IC-ICP-OES and IC-

ICP-MS method for the determination of fluoro-aluminium species. The specific

objectives were to:

• Develop a chromatographic method based on cation exchange for the

determination of cationic fluoro-aluminates, free Al 3+, AlF2+ and possibly, AlF2+

together with the neutral AlF3 using ICP-OES and ICP-MS.

• Develop an anion exchange chromatographic method for the determination of the

mixed complexes, Al(OH)2F2- and AlF4

- using ICP-OES.


139

7.2 Literature review

Dissolved Al species are complex, and can include complexes with natural organic matter,

F-, PO43-, SO4

2-, and OH-. Many authors recognized that the presence of F- could

markedly enhance Al transport in natural surface waters and in crustal fluids (Mitrovic

and Milacic, 2000; Moine et al., 1998; Zaraisky, 1994; Tennakone et al., 1988). The

presence of F- in aqueous solutions results in an increase in dissolution rates of

aluminium oxides and oxyhydroxides (Kraener et al., 1998; Nordin et al., 1998).

Roberson and Hem (1969) reported that for acidic pH (pH 5.8) and high F- concentration

(F ≥ 0.23 mg/L), complexation reactions between Al and F- are very efficient. Al-F

complexes are soluble and could potentially increase residual Al concentrations. Due to

the toxicity of Al and the existence of various fluoro-aluminium complexes, it is

generally agreed that knowledge of the form of Al species in the water system is of vital

importance.

Fluoride Ion Selective Electrode (F-ISE) has been applied to the estimation of fluoro-

aluminium complexes (Driscoll, 1984; Lazerte, 1984; Hodges, 1987; Ritchie et al., 1988).

Moore and Ritchie (1988) estimated fluoro-aluminium complexes by measuring F- with a

F-ISE before and after adding TISAB. Different methods for determining monomeric

fluoro-aluminium species have been tried, but they were indirect and usually required

speciation calculations (Hodges, 1987).

One way of achieving direct separation is to use ion chromatography (IC).The main

concern using this approach is the kinetic stability of the various fluoro-aluminium

species during the chromatography. The advantage of using IC is the direct determination

of the complexes in stead of relying solely on thermodynamic calculations.

A number of papers have been published concerning IC separation and detection by

UV/VIS spectrophotometry of fluoro-aluminium species using a variety of post-column

reactions to generate chromophores (Bertsch and Anderson, 1989; Willett, 1989; Jones,

1991; Motellier and Pitsch, 1994; Borrmann and Seubert, 1996). A post-column reaction


140

technique for Al involves the in-line addition of an indicator reagent, such as Tiron (4,5-

dihydroxy-m-benzene disulfonic acid) and Pyrocatechol Violet, to the separator column

effluent (Kawase et al., 1981; Application Note 42, 1986; Bertsch and Anderson, 1988;

Dean, 1989; Tapparo and Bombi, 1990). Both of these reagents are very selective for Al,

and they are useful for minimizing matrix interferences and determining Al in the

presence of other metals.

The chromatographic separation of fluoro-aluminium species was first proposed by

Bertsch and Anderson (1989), who determined the stability constants of several possible

AlFx species. They evaluated an IC method for fluoro-, oxalato-, and citrato- complexes

of Al. This method in which Al was quantitatively determined via post-column

derivatization with Tiron (4,5-dihydrozy-m-benzenedisulfonic acid) was evaluated for its

utility as a method for speciating Al in aqueous solution. However, adjustments to bring

sample ionic strength and pH to that of the eluent were required to prevent redistribution

of the species during determination. This separation was achieved by using a Dionex CS-

3 (4.6×200mm) separator column with 0.7 M NH4Cl eluent, adjusted to pH 2.0, 3.2, or

4.2 with HCl, at a flow rate of 1.0 mL/min. A CG 3 (4.6×50 mm) was also utilized with

weaker eluents (0.4 M NH4Cl). Improved chromatographic resolution of the AlF2+ and

AlF2+ species was achieved when a lower ionic strength (µ) eluent (0.4 M) was used,

though the use of the lower µ eluent tended to broaden the Al3+ peak. The comparison

with thermodynamic calculations gave results consistent with the formation of AlF2+ and

AlF2+ species.

In a similar way Willet (1989) also employed an IC method, using CG-2 column based

on a sulfate eluent, followed by the post-column reaction with pyrocatechol violet, but

without the adjustment of sample ionic strength and pH to match the eluent. The

chromatograms showed the effects of increasing F- concentrations on 40 µM Al, and the

appearance of peaks for fluoro-aluminium species at the expense of Al3+. However, the

resolution of first two peaks, AlF2+ and AlF2+ species, was not good enough for accurate

integration.


141

Jones (1991) investigated a highly sensitive and selective short-column ion

chromatography method for the direct determination of Al species in water samples. The

eluent used was 0.09 M K2SO4, adjusted to pH 3.0 with dilute nitric acid. The pH was

adjusted to ensure that all Al ions were present as the simple Al 3+ hydrated ion that can

yield only a single peak on the chromatogram of Al standards in absence of F-. It was

concluded that if hydroxy species such as Al(OH)2+ and Al(OH)2+ were present, they

were rapidly converted to the simple hydrated Al3+ ion in the pH 3 eluent stream, thus

giving only one peak. Otherwise, fluoro-aluminium speciation can be very complicated

due to the multi peaks from Al(OH)2+ and Al(OH)2+ species. The chromatogram in this

paper revealed that the first two peaks, AlF2+ and AlF2+ species, were not quite resolved

so the author recommended that for more precise determinations of the fluoro-aluminium

species, longer columns may be required to improve resolution.

A speciation method described by Motellier and Pitsch (1994) was performed with a CS-

2 cation-exchange column protected by a CG-2 guard column. The mobile phase

consisted of 0.5 M NH4Cl - 0.01 M HCl at a flow rate 1.1 mL/min. The post-column

reagent, 3× 10-4 M Tiron in 3 M ammonium acetate, was introduced into the eluent

stream via a T-piece located at the outlet of the column, at a rate of 0.75 mL/min. This

method requires adjustment of the sample pH and ionic strength to match the mobile

phase. The acidic media was chosen to prevent aluminium-hydroxide precipitation for

cation speciation. Total Al could be determined by the sum of the areas of the three peaks

representing the Al species because the Al species are conserved whatever the amount of

F- in the sample. This suggests that sufficiently rapid kinetics of AlFn3-n dissociation and

Al-Tiron complex formation occur as these species successively react with Tiron before

reaching the detector cell. If the sample pH is low (<4.5), ionic strength and total F-

concentration have no significant influence on the accuracy of the experimental results.

With initial sample pH values higher than 5, acidification of the sample before injection

favours rapid dehydroxylation followed by fluorination of Al3+. Although the total Al

concentration is correctly deduced, the method was not suitable for speciation since the

first two peaks, AlF2+ and AlF2+, were not fully resolved.


142

Isocratic elution, which used a single eluent, has been employed in the papers mentioned

above. Since the first two peaks have not been separated completely by an isocratic

elution system, some authors tried to carry out the speciation using gradient elution

system. This system uses different eluents in a stepwise elution to improve peak shape

and the separation of peaks.

Borrmann and Seubert (1996) applied IC with a gradient elution system using the eluent,

perchloric acid and ethlenediamine perchlorate, for speciating Al and its fluoro, oxalate

and citrate complexes. The detection method was a post-column reaction with Tiron

followed by UV detector. The chromatogram shows a good separation of three species.

Another application of gradient elution is found in the paper written by Hara et al. (2001).

The gradient elution procedure, using the eluent HNO3 and Ca(NO3)2, produces three

distinctive peaks for an aluminium F- solution. Equilibrium distribution of fluoro-

aluminium species with variation in the concentration ratio of F- to Al is conducted

experimentally and confirmed to agree with the theoretical calculation of species

distribution.

7.3 Modelling of the Al-F-H system and calculation of

distribution curves for cationic fluoro-aluminates

7.3.1 Theoretical distribution curve

The theoretical distribution curve for the species in the Al-F-H system was calculated

using the computer programme Species-Calc (Prof Hans Rohwer, Mandela Metropolitan

University of Port Elizabeth). The species that were taken into account for this

calculation in acidic media were Al3+; AlF2+; AlF2+; AlF3; AlF4

-. The equilibria and

equilibrium constants expressions of the fluoro-aluminium species used in this

calculation are given in Table 7.2.


143

Table 7.2 Equilibria and equilibration constant

Equilibria Equilibration Constant

Equation Number

−+ +⇔ OHAlsOHAl 3)()( 33

3433 103]][[ −−+ ×== OHAlKsp Eq. 7.1

+−+ ⇔+ 23 AlFFAl ]][[

][3

2

1 −+

+

=FAl

AlFK )10( 02.71 =β Eq. 7.2

+−+ ⇔+ 22 AlFFAlF

]][[][

22

2 −+

+

=FAlF

AlFK )10( 76.122 =β Eq. 7.3

)(32 aqAlFFAlF ⇔+ −+ ]][[

][

2

33 −+=

FAlFAlFK )10( 03.17

3 =β Eq. 7.4

−− ⇔+ 43 AlFFAlF ]][[

][

3

44 −

−

=FAlF

AlFK )10( 73.194 =β Eq. 7.5

−+ +⇔ FHHF 4102.7][

]][[ −−+

×==HF

FHKa Eq. 7.6

−+ +⇔ OHHOH2 1410]][[ −−+ == OHHKw Eq. 7.7

The alpha values, which are used in the distribution curves of Al-F-H system, are the

relative concentrations of the species i.e. if CAl is the sum of the analytical concentrations

of Al3+, AlF2+, AlF2+, AlF3 and AlF4

-,

CAl = [Al3+] + [AlF2+] + [AlF2+] + [AlF3] + [AlF4

-]) Eq. 7.8

Then, definitions of alpha values are:

AlC

Al ][ 3

0

+

=α Eq. 7.9 AlC

AlF ][ 2

1

+

=α Eq. 7.10

AlC

AlF ][ 22

+

=α Eq. 7.11 AlC

AlF ][ 33 =α Eq. 7.12

AlC

AlF ][ 44

−

=α Eq. 7.13

Alpha values are unitless ratios whose sum must equal unity i.e. 14310 =+++ αααα .

The alpha values are derived as follows:

Equation 7.2 is rearranged to:

]][[][ 31

2 −++ = FAlAlF β Eq. 7.14


]][[][ 222

−++ = FAlFKAlF Eq. 7.15

Substitution of Equation 7.8 into 7.9 gives:


144

232

322 ]][[]][][[][

1

−+−−++ == FAlFFAlKKAlF β Eq 7.16

Similarly,

3333 ]][[][ −+= FAlAlF β Eq. 7.17 43

44 ]][[][ −+− = FAlAlF β Eq. 7.18

4344 ]][[][ −+− = FAlAlF β Eq. 7.19

But,

CAl = [Al3+] + [AlF2+] + [AlF2+] + [AlF3] + [AlF4

-] Eq. 7.19

Therefore,

)][][][][1]([ 44

33

221

3 −−−−+ ++++= FFFFAlCAl ββββ Eq. 7.20

The alpha values can be expressed in terms of [F-] and β by substitution of Equation

7.20 into Equation from 7.9 to 7.13. Thus:

DFFFF1

][][][][11

44

33

221

0 =++++

= −−−− ββββα Eq. 7.21

(If, D= 4

43

32

21 ][][][][1 −−−− ++++ FFFF ββββ )

DF

DAlFAl ][

][]][[ 1

3

31

1

−

+

−+

==ββα Eq. 7.22

DAlFAl

][]][[

3

232

2 +

−+

=βα =

DF 2

2 ][ −β Eq. 7.23

DAlFAl

][]][[

3

333

3 +

−+

=βα =

DF 3

3 ][ −β Eq. 7.24

DAlFAl

][]][[

3

434

4 +

−+

=βα =

DF 4

4 ][ −β Eq. 7.25

Other equations used in this calculation are charge balance equation and mass balance

equations.

Charge balance equation:

][][][][][][][2][3 4223 −−−+++++ ++=++++ AlFOHFNaHAlFAlFAl Eq. 7.26

Mass balance equation:

][4][3][2][][][ 4322 −++− +++++= AlFAlFAlFAlFHFFCF Eq. 7.27


145

Substitution of Equation 7.6, 7.14, 7.16, 7.17, 7.18 and 7.19 into Equation 7.27 gives:

⎟⎟⎠

⎞⎜⎜⎝

⎛+++++= −−−+

+− )][4][3][2]([][1][ 3

42

3213 FFFAl

KHFC

aF ββββ Eq. 7.28

The theoretical distribution curve for the fluoro-complexes of Al at pH 3 is given in

Figure 7.1.

Figure 7.1 Theoretical distribution curves for the species in the Al-F-H system at pH 3 and 10 mg/L Al

Where, the values of x-axis, 0=19000 mg/L, -2=190mg/L, -4=1.9 mg/L F-

7.3.2 Calculation of the solubility of Al(OH)3

The solubility of Al(OH)3 in H2O at different pH values is derived as follows:

For the hydrolysis reaction: −+ +⇔ OHAlOHAl 3)( 33

The solubility product constants expression is:

* 343 103]][[ −−+ ×== OHAlKsp Eq. 7.30

* Ksp value is from Skoog et al. 1997


[Al 3+] = 3][ −OHKsp Eq. 7.31

Substitution of Equation 7.7 into 7.31 gives:

[Al 3+] = 33

3

3 ].[][

][+

+

− == HconstKHK

OHK

w

spsp Eq. 7.32


146

Therefore, the solubility calculation at specific pH is given in Table 7.3.

Table 7.3 Solubility of Al(OH)3 at different pH value

pH [Al 3+]/M [Al 3+]/mg/L 7 3.0×10-13 8.1×10-9 6 3.0×10-10 8.1×10-6 5 3.0×10-7 8.1×10-3 4 3.0×10-4 8.1

3.4 0.019 510 3 0.3 8100

The F- concentration was not take into account in this calculation. Since the concern of

this study was remobilisation of Al in presence of F-, recalculation of the solubility of

Al(OH)3(s) in presence of F- at specified pH, was performed.

The total Al concentration, mentioned in Equation 7.20, can be expressed in terms of Ksp

and [OH-]. Thus:

)][][][][1]([ 44

33

221

3 −−−−+ ++++= FFFFAlCAl ββββ

= )][][][][1(][

44

33

2213

−−−−− ++++ FFFF

OHKsp ββββ Eq. 7.33

The solubility of Al(OH)3(s) in a F- solution at specified pH, calculated using the method

of successive approximations, is given in Table 7.4.

Table 7.4 Solubility of Al(OH)3(s) in a F- solution at specified pH [F] /mg/L pH [OH]/M [F]/M Al total/M Al total/ mg/L

0 7 10-7 0 3.00 ×10-13 8.10 ×10-9 0 6 10-8 0 3.00 ×10-10 8.10 ×10-6 0 5 10-9 0 3.00 ×10-7 0.01 0 4 10-10 0 3.00 ×10-4 8.10 1 7 10-7 5.26 ×10-5 9.74 ×10-9 2.63 ×10-4 1 6 10-8 4.04 ×10-5 5.10 ×10-6 0.14 1 5 10-9 3.26 ×10-6 2.95 ×10-5 0.80 1 4 10-10 1.72 ×10-8 3.52 ×10-4 9.51 5 7 10-7 2.61 ×10-4 7.65 ×10-7 0.02 5 6 10-8 1.05 ×10-4 5.85 ×10-5 1.58 5 5 10-9 7.44 ×10-6 1.31 ×10-4 3.55 5 4 10-10 8.03 ×10-8 5.52 ×10-4 14.91


147

10 7 10-7 5.08 ×10-4 5.73 ×10-6 0.15 10 6 10-8 1.43 ×10-4 1.36 ×10-4 3.68 10 5 10-9 1.04 ×10-5 2.55 ×10-4 6.87 10 4 10-10 1.50 ×10-7 7.89 ×10-4 21.30 100 7 10-7 2.30 ×10-3 8.51 ×10-4 22.98 100 6 10-8 3.38 ×10-4 1.65 ×10-3 44.54 100 5 10-9 2.86 ×10-5 2.26 ×10-3 61.05 100 4 10-10 8.69 ×10-7 4.23 ×10-3 114.26

The last column in Table 7.4, Al total /mg/L, means the maximum Al concentration level

that can prevent Al hydroxide precipitation. The Al solubility in F- solutions of, 0, 1, 5,

10 and 100 mg/L, at pH 4 are 8.10, 9.51, 14.91, 21.30 and 114.26 mg/L, respectively.

This indicates that Al solubility increases as F- concentration increases in the solution.

The table shows that the Al concentrations are above permissible levels for F-

concentrations equal to and above 1 mg/L for pH values less than 6.

7.4 Cation chromatography–ICP-OES: experimental

7.4.1 Introduction

It is well known that Al ions readily form a series of complexes with F- of the general

type AlFn(3-n)+ (Vanlengenhaghe, 1966; Durrant and Durrant, 1970; Srinivasan and

Rechnitz, 1988). As mentioned above, Al solubility increases as F- concentration

increases in solution. This means that when F- concentration is increased in the municipal

water, through the process of fluoridation, the levels of dissolved Al, originating from a

remobilisation of scale from municipal pipes, could also increase through the formation

of F-Al complexes. The scale contains Al hydroxide or oxide precipitates that can

dissolve as fluoro-aluminium (AlFn(3-n)+) or hydroxo-fluoro aluminium species

(Al(OH)mFn 3-m-n). The modelling of the Al-F-H system and calculation of theoretical

distribution curves for cationic fluoro-aluminates has been introduced in Section 7.3. It

was necessary to perform the modelling of the distribution curve experimentally to

confirm the possibility of practical application of the Al-F-H theoretical distribution

system. This is going to present in this Section.


148

Since Bersch and Anderson (1989) proposed the separation of fluoro-aluminium species,

many authors reported this speciation mainly using the IC method (Willet, 1989; Jones,

1991; Motelier and Pitsch, 1994). However, their methods were not suitable for

speciation since the first two peaks, AlF2+ and AlF2+, were not completely resolved. The

purpose of the current work was to develop a method for the simultaneous determination

of fluoro-aluminium species, free Al 3+, AlF2+ and possibly, AlF2+ together with the

neutral AlF3. The separation was carried out on a cation exchange column and the fluoro-

aluminium species were detected by inductively coupled plasma-optical emission

spectrometry (IC-ICP-OES). The distribution curves for the fluoro-aluminium complexes,

determined experimentally by the IC-ICP-OES method, were compared with theoretical

curves based on the calculations with formation constants.

7.4.2 Instrumentation

A Dionex Gradient Pump equipped with a Rheodyne Model 9126 Injection Valve (100

µ L sample loop) was coupled to a Dionex Ion Pac CG5A and CG 5 (Serial number

5914) cation exchange column. The chromatographic columns were connected to a

Varian Liberty 110 ICP Emission Spectrometer. The flow rate was 2 mL/min. Data

capturing and peak analyses were performed with the use of Star Chromatography

Software (1995, Varian Associates Inc.).

7.4.3 Reagents and standard solutions


preparing all solutions. Samples and eluents were filtered through 0.22 µm membrane

filters.


• 1000 mg/L Al stock solution: 0.894 g AlCl3·6H2O was dissolved in deionised

water and diluted to 100 mL in a volumetric flask


149


diluted to 100 mL in a polystyrene volumetric flask.

• Eluent (0.5 M NH4Cl, 1mM HCl): 26.745 g NH4Cl was dissolved in deionised

water and diluted to 1L with a 1 mM HCl matrix (100µ L of 32% HCl (10 M)) in

a volumetric flask.

A series of synthetic sample mixtures, 10 mg/L Al and various concentrations of F-,

were prepared by the appropriate dilution of the stock solutions with 1 mM HCl.


a. Optimisation of ICP-OES conditions

A summary of the optimised parameters for ICP-OES is given in Table 7.5.

Table 7.5 ICP-OES operating condition Parameter Setting

Al Emission Wavelength 396.152 nm Plasma power 1.00 kW

Search window 0.040 nm Viewing height 6 mm

Plasma gas flow rate 15.0 L/min Nebuliser pressure 150 kPa

Pump speed 250 rpm Auxiliary gas flow rate 1.50 L/min

b. Optimisation of the chromatographic conditions

i) Al concentration and pH optimisation

A 1 mg/L and 10 mg/L Al solution in 0.01 M HCl was tested with the eluent 0.5M NH4Cl

and 0.01 M HCl (Motellier and Pitsch 1994) using two Dionex CS5 (Serial number 5754,

6223) analytical columns. No peak was observed at the 1 mg/L Al concentration level

with CS5 (S/N6223) and only a very small peak with CS5 (S/N 5754). This meant that

the IC coupled ICP-OES method was not sensitive enough to analyse at 1 mg/L Al level.

One big Al3+ peak, however, was obtained with the 10mg/L Al solution. In further work

and the experimental verification of the distribution diagram given in Figure 7.1, 10 mg/L

Al solution was used.


150

As briefly summarized in the literature review, Jones (1991) has demonstrated the

importance of pH on the fluoro-aluminium speciation (see Section 7.2). Another aspect

concerning pH, which the current work has pointed out, is the Al hydroxide precipitation

at higher pH. This precipitation can damage the column. The solubility of Al(OH)3 in

H2O and in F- solution at different pH values is described in Section 7.3.2. The last

column in Table 7.4, Al total /mg/L, presents the solubility of Al(OH)3 in various

concentrations of F- at specific pH values. The purpose of this experiment was to measure

the distribution curve over the whole range of F- at a fixed concentration of Al, 10mg/L.

It was necessary to choose a suitable pH value, which would prevent Al(OH)3

precipitation. The solubility of Al(OH)3, given in Table 7.4, in presence of 0 mg/L and 1

mg/L F- at pH 4 was 8.10 mg/L and 9.51,mg/L respectively. These values were less than

the chosen Al concentration, 10 mg/L, in previous paragraph. Therefore, the fluoro-

aluminium speciation at pH 4 or higher than pH 4 could not be performed owing to the

Al(OH)3 solubility, which was lower than the chosen Al concentration, 10 mg/L.

According to the solubility data in Table 7.3, the solubility of Al(OH)3 in 0 mg/L F- at pH

3 was 8100 mg/L. This value was high enough to prevent precipitation of Al(OH)3.

Therefore, pH 3 was chosen for further work. Wang et al. (1996) has demonstrated that in

highly acidic medium, F- reacts with H+ to form HF, leading to a decrease in the rate of

complexation of Al3+ with F-. At pH>3.0, the hydrolysis reaction of Al3+ can take place

with the formation of Al(OH)i(3-i)+, which reduces free Al and therefore the concentration

of complex.

ii) Eluent and column optimisation

Dionex CG12A and CG5A guard columns, which are shorter in length than the full

analytical column and would allow the analysis time to remain short resulting in a quick

and cost effective analysis, were chosen for fluoro-aluminium speciation studies of the

cationic species.

As previously mentioned in Section 7.2, the published methods were not suitable for

speciation since the first two peaks, AlF2+ and AlF2+, were not fully resolved to allow

accurate integration (Willet, 1989; Jones, 1991; Motellier and Pitsch, 1994). Therefore, it


151

was important to separate the first two peaks, i.e. for AlF2+ and AlF2+ species. The

theoretical distribution curve in Figure 7.1 shows the intersection point of these species,

AlF2+ and AlF2+, at about 10 mg/L Al and 10 mg/L F-. The intersection was a suitable

point to optimize eluent and column, since two peaks with similar sensitivity were

expected. The peak of Al3+ was not expected to show on the chromatogram, since the

alpha value of Al3+ at this point was very small, about 0.02. Therefore, the concentrations

of this point, 10 mg/L Al and 10 mg/L F-, were chosen for optimization of eluent and

column.

0.5 M NH4Cl and 0.001 M HCl, at pH 3, and a flow rate 1.2 mL/min were chosen for

speciation of Al-F species. No peak was found with the CG 12A column and overlapping

peaks with poor resolution were displayed with the CG 5A column. Therefore, these

guard columns were not suitable for separation of Al-F species. Thus, the CG 5A guard

column, which showed overlapping peaks, was used in conjunction with the CS5

analytical column. Various CS5 analytical columns, (serial number 6223, 5914 and 5754

from Dionex), were evaluated to determine the column, that provided the best

chromatographic separation, for further work. The two peaks, AlF2+ and AlF2+, were

separated with good resolution with a CG5A guard column in conjunction with a CS5

analytical column. The CS5 with serial number 5914 in conjunction with a CG5A was

found to give the best chromatographic separation.

To confirm the efficiency of these column systems in the F- concentration range, that

could display three peaks on the chromatogram, a test solution containing 10mg/L Al and

6mg/L F- was analyzed as well. The three peaks, AlF2+, AlF2+ and Al3+, were separated

with good resolution using this column system. Therefore, CG5A guard column in

conjunction with CS 5 (S/N 5914) analytical column was chosen for further work.

iii) Flow rate optimisation

A sample containing 10 mg/L Al and 10 mg/L F-, for which 2 similar size

chromatographic peaks were displayed, was used for optimisation of flow rate. Flow rates


152

were varied between 1.2 and 2.0 mL/min. Peak widths and resolution were calculated for

each flow rate. The results are summarised in Table 7.6 and Figure 7.2.

Table 7.6 The influence of flow rate on the chromatographic factors

Flow rate

Retention time (min) Peak Area 1/2 Peak width (sec)

mL/min AlF2+ AlF 2+ AlF2

+ AlF 2+ AlF2+ AlF 2+

*Resolution

1.2 1.551 2.311 4174 6473 8.7 8.4 2.7 1.5 1.321 1.967 4489 6000 6.5 8.1 2.7 1.7 1.151 1.709 4265 4054 5.2 7.2 2.7 1.8 0.993 1.507 3902 5559 3.8 5.7 3.2 1.9 0.972 1.432 5202 7675 3.5 5.4 3.1 2.0 0.949 1.388 4648 7413 6.9 6.0 2.0

*Resolution - see Section 3.2.3.a

Figure 7.2 The influence of flow rate on the chromatographic factors

0

0.5

1

1.5

2

2.5

1 1.2 1.4 1.6 1.8 2flow rate(ml/min)

rete

ntio

n tim

e(m

in)

1.5

2.0

2.5

3.0

3.5

Res

olut

ion

1 2 Resolution

Where, 1= AlF2

+, 2= AlF 2+

As mentioned in Section 3.2.3.a quantitative measure of the ability of a column to

separate two peaks can be determined by resolution. The resolution should be greater

than 1.5 to confirm two peaks as completely separate peaks. All the resolutions in Table

7.6 are greater than 1.5, which means that all the flow rates could be used for the

speciation of Al-F species. To achieve the shortest analysis time, a flow rate of 2.0

mL/min was chosen for further work since the sensitivity and the half-peak width were

still acceptable at this flow rate.

Chromatographic conditions were optimised so far. A summary of all the optimised

chromatographic conditions is presented in Table 7.7.


153

Table 7.7 Summary of chromatographic conditions

Experimental parameter Optimum condition Al concentration 10 mg/L

pH 3 Eluent 0.5 M NH4Cl + 1mM HCl

Column CG5A-CS5 (S/N 5914) Flow rate 2 mL/min

c. Experimental distribution curve

Samples, mixtures of 10 mg/L Al and various concentrations of F-, were made up and

analysed by IC-ICP-OES, using the optimised conditions in Table 7.7. The appearance

and number of peaks for the Al-F species varied, depending on F- concentration. The

typical chromatograms are shown from Figure 7.3.

Figure 7.3 Typical chromatogram of the Al-F speciation

10mg/L Al + 0.05mg/L F- (*RT: 8.2 minutes)

10mg/L Al + 1mg/L F- (RT: 9.5 minutes)

10mg/L Al + 6mg/L F (RT: 9.1 minutes)

10mg/L Al + 10mg/L F- (RT: 9.1 minutes)

Speciation of fluoro-aluminium species by IC coupled ICP-OES and ICP-MS ______________________________________________________________________


154

10mg/L Al + 80mg/L F- (RT: 3.0 minutes) * RT: Run Time

These chromatograms clearly show that the effect of increasing F- concentration on the

appearance of peaks for Al-F species at 10mg/L Al. When F- concentration was 0.05

mg/L, only 1 peak, Al 3+, was found. However, as the F- concentration was increased to 1

mg/L, a new peak was detected, namely AlF2+, as well as Al3+ species. 6 mg/L F- resulted

in another peak at earlier retention time so that 3 peaks were displayed, which were AlF2+,

AlF2+ and Al3+. When the F- concentration was even more increased to 10 mg/L, Al3+

species was disappeared while AlF2+ was increased. For 80 mg/L F- only 1 peak, single

charged species, AlF2+ was found.

As mentioned in Section 7.3.1, the alpha values, relative concentrations of the species,

were calculated using the Equation 7.9, 7.10 and 7.11.

AlCAl ][ 3

0

+

=α Eq. 7.9 AlC

AlF ][ 2

1

+

=α Eq. 7.10 AlC

AlF ][ 22

+

=α Eq. 7.11

CAl, the sum of the analytical concentrations of all species, was obtained by adding the

peak area of all peaks on the chromatogram. For instance, the alpha value of Al3+ species

was calculated as follows:

*)( 3

3PAAlPA

Al ∑=

+

+α Eq. 7.34

Where, PA* is peak area


155

In a similar way, the alpha values of AlF2+, AlF2+ also could be expressed as:

PAAlFPA

AlF ∑=

+

+

)( 2

2α Eq. 7.35 PAAlFPA

AlF ∑=

+

+

)( 22

α Eq. 7.36

The results, peak area, retention time as well as alpha (α ) values with various F-

concentration, are summarised in Table 7.8.

Table 7.8 Summary of the typical results: 10 mg/L Al at pH 3

AlF30 +AlF2

+ AlF2+ Al3+ [F]Total (mg/L)

[F]Total (M)

Log [F] (M) *PA **RT PA RT PA RT

α1 AlF2

+ α2

AlF2+ α3

Al3+ 0.05 2.6 ×10-06 -5.58 - - - - 10971 6.30 0.00 0.00 1.00 0.1 5.3×10-06 -5.28 - - 119 1.50 11005 6.53 0.00 0.01 0.99 0.2 1.1×10-05 -4.98 - - 396 1.53 11949 7.24 0.00 0.03 0.97 1 5.3×10-05 -4.28 - - 2377 1.57 13570 7.51 0.00 0.15 0.85 2 1.1×10-04 -3.98 332 1.06 5253 1.57 14118 7.57 0.02 0.27 0.72 4 2.1×10-04 -3.68 519 1.08 8287 1.59 8332 7.70 0.03 0.48 0.49 6 3.2×10-04 -3.50 1303 0.97 10005 1.48 3378 7.67 0.09 0.68 0.23 8 4.2×10-04 -3.38 2614 1.04 9347 1.54 1241 7.83 0.20 0.71 0.09 10 5.3×10-04 -3.28 4632 1.05 8095 1.55 - - 0.36 0.64 0.00 12 6.3×10-04 -3.20 7331 1.06 5551 1.58 - - 0.57 0.43 0.00 14 7.4×10-04 -3.13 8468 1.03 3437 1.54 - - 0.71 0.29 0.00 18 9.5×10-04 -3.02 13692 1.11 1610 1.61 - - 0.89 0.11 0.00 30 1.6×10-03 -2.80 10856 1.02 - - - - 1.00 0.00 0.00 50 2.6×10-03 -2.58 10416 0.96 - - - - 1.00 0.00 0.00 80 4.2×10-03 -2.38 14201 0.89 - - - - 1.00 0.00 0.00

120 6.3×10-03 -2.20 20228 0.88 - - - - 1.00 0.00 0.00 180 9.5×10-03 -2.02 18554 0.86 - - - - 1.00 0.00 0.00 250 1.3×10-02 -1.88 1492 0.83 - - - - 1.00 0.00 0.00 500 2.6×10-02 -1.58 122 0.82 - - - - 1.00 0.00 0.00 2000 1.1×10-01 -0.98 482 0.84 - - - - 1.00 0.00 0.00 5000 2.6×10-01 -0.58 228 0.85 - - - - 1.00 0.00 0.00

* PA=Peak Area ** RT=Retention time

Initially, only a single peak, Al3+ species, was found at very low-level F-, 0.05 mg/L.

However, as the F- concentration increased, two peaks, AlF2+ and Al3+, were displayed

between 0.1 mg/L and 1 mg/L F-and three peaks, AlF2+, AlF2+ and Al3+ species between

2 mg/L and 8 mg/L F-. This suggested that Al could not be free triple charge ion, Al3+, in

the presence of F-, because of the strong fluoro-aluminium complexes. When F-

concentration was even higher than 8 mg/L, the free Al ion, Al3+, disappeared and only

two peaks, AlF2+ and AlF2+species were found between 10 mg/L and 20 mg/L F-

concentration. Finally, only one peak, AlF2+, was shown on the chromatogram at

concentration more than 30 mg/L F-.


156

All these results are illustrated graphically in Figure 7.4.

Figure 7.4 Experimental distribution curves for Al-H-F system in 10mgL Al

0.00.10.20.30.40.50.60.70.80.91.0

-6.0 -5.0 -4.0 -3.0 -2.0

Log[F] total (M)

α (A

l fra

ctio

n)

Al 3+

AlF2+

AlF2++AlF3

0

A very good agreement was found between theoretical distribution curves (Figure 7.1)

and the experimental one (Figure 7.4) with regard to AlF2+ and Al 3+ species. The AlF2+,

AlF30, and AlF4

- species, however, were different from the theoretical distribution curves

in Figure 7.1. The AlF4- species, of course could not be uniquely identified in the cation

exchange method because it was a negatively charged species, which could not be

retained on the cation exchange resins. The first peak was initially thought to be AlF2+

species alone, however, this peak seemed to be AlF2+ peak together with the neutral

species, AlF30. The reason for this assumption could be demonstrated in terms of peak

area and retention time. According to the theoretical distribution curves in Figure 7.1,

AlF2+ species peak was not supposed to be detected above the point of x-axis -2. If the

peak was only the AlF2+ species, the alpha value must be 0 because of no peak integration

above that point. However, the peak area on the point of x-axis at -2.02 was still very

high and the peak was still detected even up to the point of x-axis at -0.58 (see Table 7.8).

In addition to this explanation, another clue, which could support this assumption, was

the decreasing tendency of the retention time in Table 7.8. As described in Figure 3.6 (see

Section 3.2.2.e), the surface of cationic exchange resin is composed of negatively charged

SO3-, groups. Therefore, the affinity for the resin will be higher the larger the positive

charge on a species. That was why the retention time for Al3+ species was the longest.

The singly charged species AlF2+ could be retained on the cationic resin longer than the


157

neutral species, AlF30. In other words, the neutral species, AlF3

0, seemed to be eluted a

little bit earlier than the singly charged species, AlF2+. A good correspondence between

this explanation and the data of retention time for the first peak was found. Therefore, the

first peak was not only AlF2+ species but also AlF2

+ together with AlF30 species.

Concerning this first peak, Hara et al. (2001) also demonstrated that the first peak

represents species with a negative, neutral or 1+ charge. According to the current

experiment, this column system could not separate the neutral and singly charge species

perfectly. Therefore, the distribution curve shown in Figure 7.4, which was drawn for

high F- concentration, depicts AlF2+ together with AlF3

0 species.

7.4.5 Conclusion

Speciation of fluoro-aluminium complexes has been demonstrated using a cation

chromatographic column system, CG5A in conjunction with CS5, coupled to an ICP-

OES as the detector. The species, Al 3+, AlF2+ and possibly, AlF2+ together with the

neutral AlF3,0

were well resolved with this column system and the experimental

distribution curves correlate well with theoretical curves, obtained by calculation from

the complexation constants.

7.5 Anion chromatography – ICP-OES: experimental 7.5.1 Introduction

The purpose of the current work was to develop a speciation method of the mixed

complexes, Al(OH)2F2- and AlF4

-. This was to be carried out using a hyphenated

technique between anion chromatography and inductively coupled plasma-optical

emission spectrometry (IC-ICP-OES). No literature was found on this matter. The

development of a method using an anion exchange chromatographic method was to be

considered. Because of the fact that it would be performed under basic conditions, meant

that it had the risk of blocking the column due to aluminium hydroxide precipitation.


158



µL sample loop) was coupled to a Dionex Ion Pac AG 5 and AG 14 anion exchange

column. The chromatographic columns were connected to a Varian Liberty 110 ICP

Emission Spectrometer. The flow rate was 1.5 mL/min. Data capturing and peak analyses

were performed with the use of Star Chromatography Software (1995, Varian Associates

Inc.)



preparing all solutions. Samples and eluent were filtered through 0.22 µm membrane

filters.


• 1000 mg/L Al stock solution: 0.894 g AlCl3·6H2O was dissolved in deionised water

and diluted to 100 mL in a volumetric flask.


diluted to 100 mL in a polystyrene volumetric flask.

• Eluent (1mM HCO3-, 3.5mM CO3

2-): 4 mL 0.5 M HCO3-, 14 mL 0.5 M CO3

2-

solutions were combined and diluted to 2000mL.

- 0.5 M CO32- stock solution: 26.49 g Na2CO3 (AnalaR, BDH Chemicals Ltd,

Poole, England) was dissolved in 500 mL deionised water.

- 0.5 M HCO3- stock solution: 21g NaHCO3 (Saarchem (Pty) Ltd) was dissolved

in 500mL deionised water.

• A series of synthetic sample mixtures, 10 mg/L Al and various concentration of F-,

were prepared by the appropriate dilution of the stock solutions adjusted the pH at 9

using NaOH.


159


a. Theoretical distribution curves

The theoretical distribution curve for the species in basic media was calculated using the

computer programme Species-Calc and are displayed in Figure 7.5.

Figure 7.5 Theoretical anionic distribution curves of Al(OH)mFn 3-m-n

Where, the values of x-axis, 0=19000 mg/L, -2=190mg/L, -4=1.9 mg/L F-, Al:10mg/L at pH 9

The fluoro-aluminium species in basic media are Al(OH)2F2- and AlF4

-. In the current

work it was tried to separate these two species for F- concentrations between 10-2 and 10-1

M.

b. Optimisation of ICP-OES conditions

Similar conditions to those used in cation chromatography coupled ICP-OES method

were found and given in Table 7.5.

c. Optimisation of chromatographic condition

A Dionex AG 14 guard column was selected for the separation of the two species,

Al(OH)2F2- and AlF4

-.


160

10-1M NaOH solution was initially selected to elute the species at the flow rate 2.0

mL/min. However, no peak was detected with this eluent, so it was thought that this

eluent was too weak to elute the species. Thus, sodium bicarbonate/carbonate solution,

which is a stronger eluent, and is the typical eluent for anion chromatographic separations,

was selected as the eluent A single peak was detected at a F- concentration 250 mg/L, at

which two peaks were expected.

To find a column, that was appropriate for separating the two peaks, the AG 5 guard

column was evaluated. Only one peak was obtained as before. This was probably because

the two species had the same single-charge as well as about the same size and thus the

affinity of these two species for the stationary phase seemed to be very similar. This

made the anion separation impossible with the columns tried in this investigation. Owing

to this fact as well as the risk of blocking the columns by aluminium-hydroxide

precipitation, this separation work was discontinued.

7.5.5 Conclusion

An anionic chromatographic procedure with ICP-OES detection was investigated for the

simultaneous separation and determination of Al(OH)2F2- and AlF4

-. However, it was

discontinued because the two species had the same charge and the similar size, which

made separation impossible with the column available in this study.

7.6 Cation chromatography – ICP-MS: experimental 7.6.1 Introduction The cationic fluoro-aluminium speciation, Al 3+, AlF2+ and possibly, AlF2

+ together with

the neutral AlF3, has been demonstrated at 10 mg/L Al concentration, using ICP-OES

detector in Section 7.4. Using this method, all the species, were well resolved and the

comparison with theoretical curves gave results consistent with the experimental

distribution curves. However, the concentration used in ICP-OES detection, 10 mg/L Al,


161

was quite high compared to the environmental Al concentration level. Thus, the Al-F

speciation using ICP-MS detection, which provides very low detection limit, as described

in Section 3.4.1, was investigated at low level of Al and especially tried to find the 3

peaks configuration.



µL sample loop) was coupled to a Dionex Ion Pac CG5A and CG 5 (Serial number 5914)

cation exchange column. Al-specific ICP-MS detection was carried out using a Thermo

Elemental X 7 (Thermo electron corporation, USA) fitted with a concentric nebuliser

(conical) and a Peltier cooled (2°C) spray chamber (Impact bead type). All

chromatograms were obtained monitoring Al at m/z 27 using time resolved analysis. The

flow rate was 2 mL/min. Thermo electron Plasma Lab software was used for system

control and data collection.


All reagents were of AR grade, ultra-pure HNO3 (Merck), which contain no Al impurity,

and deionised water (18 MΩ cm), obtained from a combination MilliQ Elix and Milli Q-

element water purification system, was used for preparing all solutions. Samples and

eluent were filtered through 0.22 µm membrane filters.


• 1000 mg/L Al stock solution: Dissolve 0.894 g AlCl3·6H2O in deionised water

and dilute to 100 mL in a volumetric flask.

• 1000 mg/L F- Stock solution: Dissolve 0.221 g NaF in deionised water and dilute

to 100 mL in a polystyrene volumetric flask.

• Eluent (0.5 M NH4NO3, 1mM HNO3): Dissolve 40.02 g NH4NO3 in deionised

water and dilute to 1L with a 1 mM HNO3 matrix (10mL of 0.1M HNO3) in a

volumetric flask.


162

• 0.1 M HNO3: Add 0.357 mL of 65 % (14M) ultra pure HNO3 into the 50 mL

volumetric flask.

• A series of synthetic sample mixtures, various concentrations of Al and F-, were

prepared by the appropriate dilution of the stock solutions with a 1mM HNO3

matrix.


a. Optimisation of ICP-MS conditions

ICP-MS is a very sensitive instrument and needs to be optimised the every day before

using it. The ICP-MS operating conditions are summarised in Table 7.9.

Table 7.9 ICP-MS typical operating condition

Parameter Setting Rf power 1093 W

Al m/z ratio 27 Sampling depth 201 mm

Pump speed 45 rpm Nebuliser gas flow rates 0.81 L/min

Uptake time 30 seconds Cool gas flow rates 14.9 L/min

Auxiliary gas flow rates 0.93 L/min Extraction -619 V

Deflection ion -18.5 V Pole Bias 0.5 V

Hexapole Bias -1.0 V Lens 1 2.9 V Lens 2 -20.7 V Lens 3 -168.3 V

Focus lens 18.7 V Oxide level (CeO+/Ce+) < 0.02 %

Double charged level (Ba2+/Ba+) < 0.04 %

b. Preparation of the container

As described in Section 3.4.1, ICP-MS is a very powerful tool for trace elemental

analysis, since it shows a very low detection limit (ng/L-mg/L). Therefore, the reagents as

well as containers should not contain Al or F- impurities. All containers, such as plastic

volumetric flasks, pipette tips, polytops etc., were washed with 2% HNO3 and rinse with

the Mili Q deionised water several times to remove possible contamination.


163

c. Optimisation of the chromatographic conditions

As mentioned in Section 3.4.2, the principle of the ICP-MS technology is very similar to

the ICP-OES with regard to the process that samples are decomposed to neutral elements

in high temperature argon plasma. However, unlike the ICP-OES technique, ICP-MS

accomplishes the elemental determination by counting the number of ions at a certain

mass of the element instead of the light emitted by the element. Due to the similarity, the

same separation conditions, which were optimized in ICP-OES detection, were selected.

A summary of chromatographic conditions from ICP-OES method is given in Table 7.10

(see Table 7.7).

Table 7.10 Summary of chromatographic conditions

Experimental parameter Optimum condition pH 3

Eluent 0.5 M NH4Cl + 1mM HCl Column CG5A-CS5 (S/N 5914)

Flow rate 2 mL/min

i) Eluent optimization

• 0.5 M NH4Cl + 1mM HCl eluent system

Firstly, a 0.5 M NH4Cl mixed with 1mM HCl solution, used in ICP-OES method, was

selected to perform the separation of the fluoro-aluminium complexes. The speciation,

however, was discontinued because of a white deposition on the central channel of the

torch (see Figure 7.6).

Figure 7.6 The deposition place


164

Figure 7.6 clearly shows that the deposition positioned at the place where desolvation

traditionally occurs. This meant that the component of deposition came from the eluent.

Instead of using 0.5M ammonium chloride, lower concentrations of ammonium chloride,

such as 0.2M, 0.3M and 0.4M, were tested, however, these also yielded the deposition.

Thus, ammonium chloride eluent could not be used in further work.

• 0.45M HNO3 eluent system

Bayón et al. (1999) stated that saline solutions, such as K2SO4, are not recommended in

ICP-MS, because of the possibility of clogging the central channel of the torch and

deposits on the cones. 0.45 M nitric acid, using CG 2 guard column, was selected to

separate the two species, AlF2+ and Al3+. Although only two species were shown in this

paper and the pH was lower than 3, which was the optimized value in the current work,

this eluent was tested to separate the 3 peaks in the current experiment. However, only 2

peaks were found on the chromatogram using nitric acid eluent. This meant that H+ alone,

involved in the process of exchange on the resin, was not strong enough to elute 3 fluoro-

aluminium species, Al3+, AlF2+ and possibly, AlF2+ together with the neutral AlF3.

• 0.5 M NH4NO3 + 1mM HNO3 eluent system

An appropriate eluent, which was stronger than HNO3 and could solve the deposition

problem, was necessary. Ammonium nitrate eluent was tested because all components of

the eluent would be converted into the gaseous state in the plasma and so deposition

should be minimal. This eluent system worked well with the Al-F speciation; however, it

also yielded the deposition problem after instrument being used for a long time. A

possible way to solve this problem was to use the 2% HNO3 solution between sample

measurements, which could prevent the high concentration eluent from flowing through

the entire ICP-MS system. This eluent system was better than the ammonium chloride

eluent in terms of preventing blocking the central channel of torch; therefore this eluent

was introduced for further Al-F speciation work.

d. Typical chromatogram

Samples, mixtures of various concentrations of Al and various concentrations of F-, were

made up and analysed by Ion chromatography coupled ICP-MS, using the optimised ICP-


165

MS conditions in Table 7.9, IC conditions in Table 7.7 with ammonium nitrate eluent

system. This method worked well for Al-F speciation and the typical chromatogram,

having 3 peaks, is shown in Figure 7.7.

Figure 7.7 Typical chromatogram of the Al-F speciation

using ICP MS detector

e. Pitfall of this method

Although the Al-F speciation at low levels could be performed successfully with this

method, deposit formation in the torch after extended use could occur. This required

regular cleaning of the torch.

7.6.5 Conclusion

The speciation of Al-F was done by using a cation chromatographic column system,

CG5A in conjunction with CS5, coupled to an ICP-MS or ICP-OES as the detector. The

species, Al 3+, AlF2+ and possibly, AlF2+ together with the neutral AlF3,

0 were well

resolved with this column system and the ammonium nitrate eluent system.

Conclusion _____________________________________________________________________________

166

Chapter 8 Conclusion

The investigation of the chemistry and analytical chemistry of low-level (0.05 – 1

mg/L) F- determination was intensively discussed in this thesis. Method validation and

the development of standard ISE and IC procedures for the determination of low-level

F- in deionised water brought new insights in the requirements for accurate analytical

procedures for F- (Chapter 4). It was found that ISE gave more accurate and precise

results. Therefore, ISE is recommended for low level F- determination. In Chapter 5, the

procedures for using ISE and IC determinations, developed in Chapter 4, were evaluated

for the analysis of natural waters. The % recoveries of laboratory fortified samples

(LFM) samples were within the acceptable criterion (75-120 %) as described in the U.S.

EPA method 300.1. The % recovery obtained by the ISE method showed better

accuracy than that by the IC method and the IC method was more susceptible by the

influence of the matrix. Therefore, overall the ISE methodology was found to be

preferable for the accurate determination of F- in both deionised water and natural water

sample matrices.

An Inter-laboratory study in conjunction with SABS (Chapter 6) showed that the ISE

method is the most commonly used method in South African laboratories for

determination of F-. The best accuracy, calculated from average absolute Z-scores, was

achieved by using ISE methodology. This result supported the conclusion from Chapter

4 and Chapter 5. The proficiency of the procedures developed in this study was proven

through the best results obtained by Lab X (UJ Analytical chemistry lab), which

obtained the most accurate results among 66 participants in South Africa laboratories.

The speciation of cationic fluoro-aluminium complexes, Al 3+, AlF2+ and, AlF2+

together with the neutral AlF30, has been successfully accomplished using a cation

Conclusion _____________________________________________________________________________

167

chromatographic column system coupled to an ICP-OES or ICP-MS. The experimental

distribution curves at 10mg/L Al concentration level, achieved by the developed method,

IC-ICP-OES with ammonium chloride eluent system, correlated well with theoretical

curves, obtained by calculation from the complexation constants. However, the

separation and determination of Al(OH)2F2- and AlF4

- , an anionic chromatographic

procedure with ICP-OES detection, was found impossible because the two species had

the same charge (-1) and the similar size.

References ______________________________________________________________________________

168

References

1. ALEXANDERSEN P, RIIS BJ and CHRISTIANSEN C (1999)

Monofluorophosphate combined with hormone replacement therapy induce a

synergistic effect on bone mass by dissociating bone formation and resorption in

postmenopausal women: a randomized study. J. Clin. Endocrinol Metab. 84 2013-

3020

2. ALPÍZAR J, CRESPÍ A, CLADERA A, FORTEZA R and CERDÁ V (1996)

Simultaneous determination of chloride and fluoride ions in waters by sequential

injection analysis electroanalysis 8 1051-1054

3. ANALYTICAL WORKING GROUP OF THE COMITÉ TECHNIQUE EUROPÉEN

DU FLUOR (1986) The determination of fluoride in environmentally relevant

matrices Analytica Chimica Acta 182 1-16

4. Annual Book of ASTM Standards Part 31, “Water”, Standard D 1179-72, Method A,

(1976) Am. Soc. Test Master 310

5. BABCOCK RH and JOHNSON KA (1968) Selective ion electrode system for

fluoride analyses. J. Am. Water Works Assoc. 60 953-961

6. BAKER JP and SCHOFIELD CL (1982) Aluminum toxicity to fish in acidic waters.

Water, Air and Soil Pollut. 18 289-309

7. BARNETT NW, JONES P and HANDLEY HW (1993) The use of zirconyl xylenol

orange for the post-column spectroscopic detection of fluoride in ion chromatography

Analytical letters 26 2525-2529

8. BAYLEY TA, HARRISON JE, MURRAY TM, JOSSE RG, STURTRIDGE W,

PRITZKER KP, STRAUSS A, VIETH R and GOODWIN S (1990) Fluoride-induced

fractures: relation to osteogenic effect. J. Bone Miner. Res. 5 (suppl 1): S217-S222

9. BAYÓN MM, GARCIA AR, ALONSO JIG and SANZ-MEDEL A (1999) Indirect

determination of trace amounts of fluoride in natural waters by ion chromatography: a

comparison of on-line post-column fluorimetry and ICP-MS detectors Analyst 124

27-31

10. BELLACK E (1958) Simplified Fluoride Distillation Method J. Am. Water Works

Assoc. 50

References ______________________________________________________________________________

169

11. BELLACK E and SCHOUBOE PJ (1958) Rapid photometric Determination of

Fluoride in Water Anal. Chem. 30 2032

12. BERSCH PM and ANDERSON MA (1989) Speciation of aluminum in aqueous

solutions using ion chromatography Anal. Chem. 61 535-539

13. BONDOUX G and JONES T (1995) Analysis of Anions and cations in deionised

water at parts-per-billion levels LC-GC 13 144-148

14. BORRMANN G and SEUBERT A (1996) Aluminum speciation by liquid

chromatography concerning hydro- and geo-chemical aspects Analytica Chimica

Acta 332 233-239

15. BORZITSKY JA, DVININ A, PETRUKHIN OM and UROSOV YI (1993) Flow

cell with double slope factor for potentiometric determination of fluoride at low

concentrations Analyst 118 859-861

16. BUTLER JN (1969) Ion Selective Electrodes (DURST RA ed.) Chapter 5 National

bureau of standards special publication 314 Washington D.C.

17. BUTLER WJ, SEGRETO V and COLLINS E (1985) Prevalence of dental mottling

in school-aged lifetime residents of 16 Texas communities Am. J. Public health 75

1408-1412

18. CABELLO-TOMAS ML and WEST TS (1969) Kinetochromic spectrophotometry-

I Determination of fluoride by catalysis of the zirconium-xylenol orange reaction

Talanta 16 781-788

19. CARDWELL TJ, CATTRALL RW, HAUSEN PC and HAMILTON IC (1987)

Buffer systems for use with the fluoride electrode in flow injection analysis Anal.

Chem. 59 206-208

20. CHEN XY, LIN MQ, HE ZL, CHEN C, MIN D, LIU YQ and YU MH (1996)

Relationship between total fluoride intake and dental fluorosis in areas polluted by

airborne fluoride. Fluoride 29 7-12

21. CHEN Y, LIN M, HE Z, XIE X, LIU Y, XIAO Y, ZHOU J, FAN U, XIAO Z and

XU F (1993) Air pollution-type fluorosis in the region of Pingxiang, Jiangxi,

People’s Republic of China. Arch. Environ. Health 48 246-249

22. CRAPPER MDR and BONI D (1980) Aluminum in human brains disease-an

overview. Neurotoxicol 1 3-16

References ______________________________________________________________________________

170

23. CROSBY NT, DENNIS AL and STEVENS JG (1968) An evaluation of some

methods for the determination of fluoride in potable waters and other aqueous

solutions Analyst 93 643-652

24. DAVIDSON AM, WALKER GS and LEWINS AM (1982) Water supply

aluminum concentrations, dialysis dementia, and effects of reverse osmosis water

treatment. Lancet. 2 785-787

25. DEAN HT (1934) Classification of mottled enamel diagnosis J. Am. Dent. Assoc.

21 1421-1426

26. DEPARTMENT OF HEALTH (2000) Regulation No. 873 Regulations under the

Health Act, 1977 (Act No. 63 of 1977) Regulations on Fluoridating Water Supplies.

Government Gazette, 8 September 2000, 423 No 21533, Pretoria.

27. DEVINE RF and PARTINGTON GL (1975) Interference of sulfate ion on

SPANDS colorimetric determination of fluoride in wastewaters Environ. Sci.

Technol. 9 678-679

28. DIONEX (1997) PeakNet Software User’s Guide Document No. 034914

29. DIONEX APPLICATION NOTE 42 (1986) Determination of aluminum by ion

chromatography Dionex Corporation Sunnyvale, CA

30. DIONEX APPLICATION NOTE 133 Determination of inorganic anions in

drinking water by ion chromatography Dionex

31. DIONEX APPLICATION NOTE 135 Determination of Inorganic Anions in

Wastewater by Ion Chromatography Dionex

32. DIONEX APPLICATION NOTE 140 Fast Analysis of anions in drinking water by

ion chromatography Dionex

33. DIONEX APPLICATION NOTE 154 Determination of inorganic anions in

environmental waters using a hydroxide-selective column Dionex

34. DOUGLAS DP (1979) Stability Constants of Metal-Ion Complexes Part B:

ORGANIC LIGANDS” IUPAC Chemical Data Series, No. 22. PERGAMON

PRESS, OXFORD, NEW YORK, TORONTO, SYDNEY, PARIS, FRANDFURT

35. DRISCOLL CT and LETTERMAN RD (1988) Chemistry and fate of Al III in

treated drinking water. Int. J. Environ. Eng. Div. 114(1) 21-37

References ______________________________________________________________________________

171

36. DUFF EJ and STUART JL (1975) The successive determination of chloride,

fluoride and sodium in single samples of orthophosphate minerals by means of ion-

selective electrodes Talanta 22 823-826

37. DURRANT PJ and DURRANT B (1970) Sub-group III N boron, aluminum,

indium and thallium In introduction to advanced Inorganic Chemistry, 2nd edn. New

York John Wiley 522-600

38. DURST RA (1969) Determination of fluoride by analyte addition potentiometry.

Microchim. Acta 31 611-614

39. DURST RA (1969) Ion Selective Electrodes National bureau of standards special

publication 314 Washington D.C.

40. ERIK H (1982) Stability Constants of Metal-Ion Complexes Part A: INORGANIC

LIGANDS” IUPAC Chemical Data Series, No 21. PERGAMON PRESS,

OXFORD, NEW YORK, TORONTO, SYDNEY, PARIS, FRANDFURT

41. FANG Z, ZHU Z, ZHANG S, XU S, GUO L and SUN L (1988) On-line separation

and preconcentration in flow injection analysis Analytica Chimica Acta 214 41-55

42. FARAJ-ZADEH MA and KALHOR EG (2001) Extraction-spectrophotometric

method for determination of fluoride in the range of microgram per liter in natural

waters Mikrochim. Acta 137 169-171

43. FEDRAL REGISTER,1991 Jan 30, 56:20:3573

44. FETIEF DH, SUMMERLIN DJ, HARRIS BE, BRADLEY EL (1985) An

evaluation of three procedures for fluoride analysis. Caries Res. 19 248-254

45. FRANT MS and ROSS JW (1968) Use of a total ionic strength adjustment buffer

for electrode determination of fluoride in water supplies. Anal. Chem. 40 1169-1171

46. FRANT MS and ROSS UW (1966) Electrode for sensing fluoride ion activity in

solution. Science 154 1553-1554

47. FRENZEL W and BRÄTTER (1986) Fluoride ion-selective electrode in flow

injection analysis Analytical Chimica Acta 187 1-16

48. FRENZEL W and BRÄTTER P (1986) The fluoride ion-selective electrode in flow

injection analysis: part 3. Application Analytica Chimica Acta 188 151-164

49. FRITZ JS, RICHARD MJ and KARRAKER SK (1958) Potentiometric titrations

with EDTA. Use of masking agents to improve selectivity Anal. Chem. 30 1347-

1350

References ______________________________________________________________________________

172

50. FRY BW and TAVES DR (1970) Serum fluoride analysis with fluoride electrode. J.

Lab. Clin. Med. 756 1020-1025

51. GARDNER MJ and GUNN AM (1991) Bioavailability of Al from food and

drinking water. Proc. Roual Soc. Med. Round Table Series: Alzheimer’s Diseases

and the Environment, London, UK

52. GARRIDO M, LISTA AG, PALOMEQUE M and FERNÁNDEZ BAND BS

(2002) Fluorimetric determination of fluoride in a flow assembly integrated on-line

to an open/closed FIA system to remove interference by solid phase extraction

Talanta 58 849-853

53. HAARHOFF J (2003) The accuracy of fluoride measurement in water and its

implications for water fluoridation (Rapid communication) Water S.A. 29 219-224

54. HALL LL, SMITH FA, DE LOPEZ OH, GARDNER KE (1972) Direct

potentiometric determination of total ionic fluoride in biological fluids. Clin. Chem.

18 1455-1458

55. HALLSWORTH AS, WEATHERELL JA, DAN DEUTSCH D (1976)

Determination of subnanogram amounts of fluoride with the fluoride electrode.

Anal. Chem. 48 1660-1664

56. HARA H and HUANG CC (1997) Buffer composition suitable for determining

very low fluoride concentrations using a fluoride ion-selective electrode and its

application to the continuous analysis of rain water Analytica. Chimica Acta 338

141-147

57. HARWOOD JE (1969) The use of an ion-selective electrode for routine fluoride

analyses on water samples Water Research 3 273-280

58. JACKSON PE (2000) in: MEYERS RA (Editor) Encyclopedia of Analytical

Chemistry John Wiley and Sons Ltd., Chichester 3 2779

59. JACKSON PE (2001) Determination of inorganic ions in drinking water by ion

chromatography trends in analytical chemistry 20 320-329

60. JACKSON PE, WEIGERT C, POHL CA and SAINI C (2000) Determination of

inorganic anions in environmental waters with a hydroxide-selective column

Journal of Chromatography A 884 175-784

61. JAVIS KE, GRAY AL and HOUK RS (1992) Handbook of Inductively Coupled

plasma mass spectrometry Chapman and Hall New York

References ______________________________________________________________________________

173

62. JEKEL MR (1991) Aluminum in water: How it can be removed? Use of aluminum

salts in treatment Proc. of the Int. Water Supply Assoc. Copenhagen, Demark, May

25-31

63. JOACHIM W (1986) Handbook of Ion Chromatography Dionex Corporation

publishers Sunnyvale California

64. JONES P (1991) The investigation of aluminum speciation in natural and potable

water using short-column ion chromatography Int. J. Environ. Anal. Chem. 44 1-10

65. JONES P (1992) Development of a high-sensitivity ion chromatography method for

the determination of trace fluoride in water utilizing the formation of the AlF2+

species Analytica Chimica Acta 258 123-127

66. KATHRYN EML and HOWARD LC (1973) Comparison of the SPANDS-

Zirconium Lake and Specific Ion Electrode Methods of Fluoride Determination in

Stack Emission Samples Anal. Chem. 45, 1272

67. KAURANEN P (1977) The use of buffers in the determination of fluoride by an Ion

Selective Electrode at low concentration s and in the presence of aluminum

Analytical Letters 10 451-465

68. KHALIFA ME and HAFEZ MAH (1998) Spectrophtometric and complexometric

methods for the determination of thorium and fluoride using bromocresol orange

reagent Talanta 47 547-559

69. KISSA E (1987) Determination of inorganic fluoride in blood with a fluoride ion-

selective electrode. Clin.Chem. 332 253-255

70. LAJUNEN LHJ (1991) Spectrochemical Analysis by Atomic Absorption and

Emission, University of Oulu, Finland

71. LETTERMAN RD and DRISCOLL CT (1994) Control of residual aluminum in

filtered water. AWWA RF Report, AWWA, Denver, CO 80235. 1-93

72. LIBERTY (1991) Analytical methods for the Liberty spectrometer system, Varian

Australia Pty.Ltd., publication N0. 85 100938 00

73. LIU Y ed. (1995) Human exposure assessment of fluoride. An international study

within the WHO/UNEP Human Exposure Assessment Location (HEAL)

Programme. Beijing, Chinese Academy of Preventive Medicine, Institute of

Environmental Health Monitoring, Technical Cooperation Centre of

Fluoride/HEAL Programme, 64

References ______________________________________________________________________________

174

74. LOPES DA CONCEIÇÃO AC, CORREIA DOS SANTOS MM, SIMÕES

GONÇALVES and SANTOS FJV (2000) Gradient flow titration for the

determination of fluoride ion in natural waters Talanta 50 1245-1252

75. MAC LEOD KE and CRIST HL (1973) Comparison of the SPANDS-Zirconium

lake and specific ion electrode methods of fluoride determination in stack emission

samples Anal. Chem. 45 1272-1273

76. MARTYN CN, OSMAND C, EDWARDSON JA, BARKER KJP, HARRIS EC

and LACEY RFC (1989) Geotraphical relatioin between Alzheimer’s disease and

aluminum in drinking water. The Lancet. I (8629) 59-62

77. MCCAFFREY LP (1994) Fluoride electrode vs ion chromatographic analysis of

groundwater. Abstract Volume Analytika 94 87

78. MEUNIER PH, SEVERT JL, REGINSTER JY, BRIANCON D, APELBOOM T,

NETTER P, LOEB G, ROUILLON A, BARRY S, EVREUS JC, AVOUAC B and

MARCHANDISE X (1998) Fluoride salts are no better at preventing new vertebral

fractures than calcium-vitamin D in postmenopausal osteoporosis: the FAVO Study.

Osteoporosis Int, 8 4-12

79. MILLER JC and MILLER JN (1993) Statistics for Analytical Chemistry 3 rd Ed.,

Prentice Hall

80. MONSTER A and GOLIGHTLY DW (1992) Inductively Coupled Plasmas in

Analytical Atomic spectrometry 2nd Ed. VCH Publishers Cambridge

81. MOORE GL (1989) Introduction to Inductively coupled plasma atomic emission

spectrometry Elsevier Amsterdam

82. MOSKVIN LN, KATRUZOV AN and NIKITINA TG (1998) Ion-chromatographic

determination of fluoride and chloride ions in high-purity water Journal of Anal.

Chem. 53 173-177

83. MOTELLIER S and PITSCH H (1994) Determination of aluminum and its fluoro

complexes in natural waters by ion chromatography Journal of Chromatography A

660 211-217

84. MURRY JJ, RUGG0GUNN AJ and JENKINS GN (1991) Fluorides in Caries

Prevention (3rd edn.). Butterworth-Heineman Ltd, Oxford

85. NAJIB FM and OTHMAN S (1992) Computer-aided flow-analysis for laboratory

use and process analysis. Anal. Instrum. 20 79-100

References ______________________________________________________________________________

175

86. NAKAJIMA H, KOMATSU H and OKABE T (1997) Aluminum ions in analysis

of released fluoride from glass ionomers. Journal of Dentistry 25 137-144

87. NICHOLSON K and DUFF EJ (1981) Fluoride determination in water: An

optimum buffer system for use with the fluoride-selective electrode Analytical

letters 14(A7) 493-517

88. NICO (2000) Guide to ISE measurements, www.nico2000.net (accessed May/2004)

89. NILSON R (1990) Residual aluminum concentration in the drinking water after

treatment with aluminum or iron salts. In: HH Hahn and R Klute(eds.) Chemical

Water and Wastewater Treatment, Springer-Verlag, Berlin, Germany. 329-410

90. NISHIMOTO J, YAMADA T and TABATA M (2001) Solvent extraction and

fluorometric determination of fluoride ion at ppb level in the presence of large

excess of aluminum (III) and iron (III) by using an expanded porphyrin, sapphyrin

Analytica Chimica Acta 428 201-208

91. OKAMOTO Y (2001) Determination of fluorine in aqueous samples by

electrothermal vaporization inductively coupled plasma mass spectrometry (ETV-

ICP-MS). J. Anal. At. Spectrom. 16 539-541

92. OKUTANI T, TANAKA C and YAMAGUCHI Y (1989) Determination of fluoride

in natural waters by ion-selective electrode potentiometry after co-precipitation with

aluminum phosphate Talanta 36 973-976

93. OLESIK JW (1996) Fundamental Research in ICP-OES and ICP-MS Anal. Chem.

68 1996

94. OLIVE RT and CLAYTON AG (1970) Direct determination of fluoride in

miscellaneous fluoride materials with the Orion fluoride electrode Analytica

Chimica Acta 51 409-415

95. ORION (1982) Instruction manual, fluoride electrodes, model 94-09 Orion

Research, Cambridge, MA

96. OSZWALDOWSKI S, LIPKA R, MAJEWSKI T and JAROSZ M (1998) Sensitive

reversed-phase liquid chromatographic determination of fluoride based on its

ternary systems with zirconium (IV) or hafnium (IV) and 2-(5-bromo-2-

pyridylazo)-5-diethylaminophenol Analyst 123 1529-1533

97. PAK CYC, SAKHAEE K, PIZIAK V, PETERSON RD, BRESLAU NA, BOYD P,

POINDEXTER JR, HERZOG J, HEARD-SAKHAEE A, HAYNES S, ADAMS-

References ______________________________________________________________________________

176

HUET B and REISCH JS (1994) Slow-release sodium fluoride in the management

of postmenopausal osteoporosis. A randomised controlled trial. Ann. Intern. Med.

120 625-632

98. PAPAEFATATHIOU J, TENA MT and LUQUE DE CASTRO MD (1995) On-line

pervaporation separation process for the potentiometric determination of fluoride in

dirty samples Analytica Chimica Acta 308 246-252

99. PERKIN ELMER SCIEX instruments. Technical note, The 30-Minute Guide to

ICP-MS ( http://las.perkinelmer.com/content/ApplicationNotes/D-6355A.pdf )

100. PICKERING WF (1986) The effect of hydrolyzed aluminum species in fluoride ion

determinations. Talanta 33 661-664

101. PONTIUS FW (1991) Fluoride regulation and water fluoridation J. AWWA 83

20,22,96

102. PONTIUS FW (1993) NRC determines current fluoride MCL appropriate. Opflow

19 3-4

103. REEVES TG (1994) Water Fluorididation: A Manual For Water Plant Operators.

U. S. Department of Health and Human Services, Centers for Disease Control and

Prevention, Atlanta, Georgia.

104. RICHARD AD (1969) Ion selective electrode Department of commerce National

Bureau of Standards Special Publication 314, Washington, D.C.

105. RICHARD FD and GERALD LP (1975) Interference of Sulfate Ion on SPANDS

Colorimetric Determination of Fluoride in Wastewaters. 9 678

106. RICHARDS LF, WESTMORELAND WW, TASHIRO M, MCKAY CM and

MORRISON TJ (1967) Determination of optimum fluoride levels for community

water supplies in relation to temperature. J. Am. Dent. Assoc. 74 389-397

107. RIGGIS BL (1991) Treatment of osteoporosis with sodium fluoride or parathyroid

hormone. Am. J. Med. 91 37S-41S

108. RIZZOLI R, CHEVALLEY T, SLOSMAN DO and BONJOUR JP (1995) Fluoride

intake from foods, beverages and dentifrice by young children in communities with

negligibly and optimally fluoridated water: a pilot study. Community Dent. Oral

Epidemiol 27 288-297

References ______________________________________________________________________________

177

109. ROBERSON CE and HEM JD (1969) Solubility of aluminum in the presence of

hydroxide, fluoride and sulfate. US Geological Survey Water Supply, Paper 1827-C,

US Government Printing Office, Washington, DC.

110. SAAD B, POK RW, SUJARI ANA and SALEH MI (1998) Analysis of anions and

cations in drinking water samples by capillary ion analysis Food chemistry 61 249-

254

111. SAHA S (1993) Treatment of aqueous effluent for fluoride removal. Water Res. 27

1347-1350

112. SARA R and WÄNNINEN E (1975) Separation and determination of fluoride by

diffusion with hexamethyuldisiloxame and use of a fluoride-sensitive electrode

Talanta 22 1033

113. SCHECHER WD and DRISCOLLCT (1988) An evaluation of equilibrium

calculations within acidification models: The effect of uncertainty in measured

chemical components. Water Resour. Res. 24 533-542

114. SKOOG DA, NIEMAN TA and HOLLER FJ (1998) Principles of Instrumental

Analysis, 5th Ed., Harcourt Brace College Publishers, USA

115. SKOOG DA, WEST DM and HOLLER FJ (1997) Fundamentals of Analytical

Chemistry, 7th Ed., Saunders college Publishers, USA

116. SLANINA J, LINERAK WA and BAKKER F (1980) The use of ion-selective

electrodes in manual and computer-controlled flow injection systems Analytica


117. SRINIVASAN K and RECHNITZ GA (1988) Reaction rate measurements with

fluoride ion-selective membrane electrode. Formation kinetics of ferrous fluoride

and aluminum fluoride complexes Anal. Chem. 40 1818-1825

118. SRINIVASAN PT, VIRARAGHAVAN T and SUBRAMANIAN KS (1999)

Aluminum in drinking water: An overview. Water S.A. 25(1) 47-55

119. STANDARD METHODS FOR THE EXAMINATION OF WATER AND

WASTEWATER METHOD NO. 414A, Preliminary Distillation Step 389-390 and

Method 414C SPANDS 14th Edition (1975) American Public Health Association,

Washington 393-394

References ______________________________________________________________________________

178

120. STANDARD METHODS FOR THE EXAMINATION OF WATER AND

WASTEWATER METHOD 4500 FD 20th ed., (1998) Am. Pub. Health Assoc.

Washington 4-62

121. STANDARD METHODS FOR THE EXAMINATION OF WATER, SEWAGE,

AND INDUSTRIAL WASTE METHOD (1955) 10th ed. Am. Pub. Health Assoc.,

Am. Water Works Assoc., Am. Federation of Sewage and Industrial Wastes New

York 98-107

122. STEFAN RI, VAN STADEN JF and ABOUL-ENEIN HY (1999) Determination of

fluoride in toothpaste, effluent streams and natural and borehole water using a flow

injection system with a fluoride-selective membrane electrode Pharmaceutica Acta

Helvetiae 73 307-310

123. STEFANOVIĆ ŠC, BOLANĆA T and ĆURKOVIĆ L (2001) Simultaneous

determination of six inorganic anions in drinking water by non-suppressed ion

chromatography Journal of chromatography A 918 325-334

124. TROJANOWICZ M and MATUSZEWSKI W (1982) Limitation of linear response

in flow-injection systems with ion-selective electrodes Analytica Chimica Acta 138

71-79

125. TROJANOWICZ M, ALEXANDER PW and HIBBERT DB (1998) Flow-Injection

analysis wit potentiometric detection for the speciation of fluoride and calcium

Analytica Chimica Acta 366 23-33

126. UNDERWOOD EJ (1977) Trace Elements In Human And Animal Nutrition (4th

edn.) Academic Press, New York, San Francisco, London.

127. US EPA 300.0 (1993) The determination of inorganic anions in water by Ion

Chromatography US Environmental Protection Agency, Cincinnati, OH 45268

128. US EPA 300.1 (1997) Determination of inorganic anions in water by Ion

Chromatography Revision 1.0 US Environmental Protection Agency, Cincinnati,

OH 45268

129. US EPA 340.1 Colorimetric, SPANDS with Bellack Distillation” STORET NO.

Total 00951 Dissolved 00950, 1971-Retrieved May/2004 from

http://www.epa.gov/cgi-bin/climage.sh/ima-link/ORD/images/epa-cin8

References ______________________________________________________________________________

179

130. US EPA METHOD 13A Determination of total fluoride emissions from stationary

sources-SPANDS ZIRCONIUM LAKE METHOD-Retrieved May/2004 from

www.epa.gov/ttn/emc/promgate/m-13b.pdf

131. US EPA METHOD 13B Determination of total fluoride emissions from stationary

sources(specific ion electrode method)-Retrieved May/2004 from

www.epa.gov/ttn/emc/promgate/m-13b.pdf

132. US EPA (1984) Appendix B to Part 136- Definition and Procedure for the

Determination of the Method Detection Limit-Revision 1.11, Federal Register

49(209), 43430, October 26

133. VAN OORT WJ and VAN EERD EJJM (1983) Effect of polishing the fluoride-

selective electrode on the response time and sensitivity in flow systems Analytica


134. VAN STADEN JF, STEFAN F and VIRGHILA S (2000) Evaluation of different

SIA sample-buffer configurations using a fluoride-selective membrane electrode as

detector Talanta 52 3-11

135. VAN WYK PJ (1995) Personal communication. Department of Health, Pretoria

136. VANLENGENHAGHE C, VALENSI G and POURBAIX M (1966) Fluorine. In:

Pourbaix m, ed. Atlas of Electrochemical Equilibria in Aqueous Solutions Oxford

Pergamon Press 579-589

137. VARMA A (1991) Handbook of Inductively Coupled plasma atomic emission

Spectroscopy, CRC Press , Florida

138. VICKERY B and VICKERY ML (1976) Suppression of interfering ions in the

analysis of plants to determine fluoride using the fluoride ion selective electrode

Analyst 101 445-454

139. VILLA AE (1988) Rapid method for determination very low fluoride concentration

using an ion-selective electrode. Analyst 113 1299-1303

140. VOGEL GL, CAREY CM, CHOW LC, EKSTRAND J (1990) Fluoride analysis in

nanoliter and microliter-size fluid samples. J. Dent. Res. 69 (special issue) 522-528

141. WADA H, MORI H and NAKAGAWA G (1985) Spectrophotometric

determination of fluoride with lanthanum/alizarin complexone by flow injection

analysis Analytica Chimica Acta 172 297-302

References ______________________________________________________________________________

180

142. WANG H, ZHANG Z, SUN A, LIU D and LIU R (1996) Stopped-flow kinetic

determination of aluminum in Chinese tea leaves by fluoride ion-selective electrode

potentiometry Talanta 43 2067-2072

143. WANG XD, SHEN W, CATTRALL RW, NYBERG GL and LIESEGANG J

(1995) An extrapolation method for the fast evaluation of the fluoride Ion Selective

Electrode equilibrium potential in continuous flow analysis Electroanalysis 7 221-

224

144. WEISS J (1986) Handbook of Ion Chromatography Dionex Corporation California

145. WEISS J, REINHARD S, POHL C, SAINI C and NARAYARAN L (1995)

Stationary phase for the determination of fluoride and other inorganic anions

Journal of chromatography A 706 81-92

146. WELLS AF (1975) Complex, oxy- and hydroxyl-halides. In: Structural Inorganic

Chemistry. 4th edn. London: Oxford University Press 377-413

147. WHO ENVIRONMENTAL HEALTH CRITERIA (1984) Fluorine and fluorides

148. WHO ENVIRONMENTAL HEALTH CRITERIA 227 (2002) Fluorides Geneva

149. WILLET IR (1989) Direct Determination of Aluminum and its Cationic Fluoro-

Complexes by Ion Chromatography Soil Sci. Soc. Am. J. 53 1385-1391

150. WOLFGANG F and PETER B (1986) Fluoride ion-selective electrode in flow

injection analysis Analytical Chimica Acta 187 1-16

151. YUCHI A, MURASE H and WADA H (1995) Structural Features of Organic

Reagents Suitable for Spectrophotometric or Fluorometric Determination of

Fluoride Based on Mixed-Ligand Complex Formation Analytical Sciences 11 221-

226

Documents

Evaluation of analytical methodologies for fluoride