The Multivariate Optical Element Platformcirtemo.com/pdf/CIRTEMO_Technology_Overview.pdf · 2013. 10. 24. · Multivariate Optical Elements can increase the sensitivity and specificity

Technology Overview

The Multivariate Optical Element Platform

What Does CIRTEMO Do?

CIRTEMO designs and manufactures patented optical filters, called Multivariate

Optical Elements (MOE), which are encoded to detect/measure very complex chemical signatures/attributes.

© CIRTEMO, LLC 2013. All Rights Reserved. CIRTEMO Products and Services are protected by U.S. and International issued patents and pending patent applications. 2

33 Issued Patents and 9 Pending Patents

CIRTEMO™ Corporate Overview • CIRTEMO was founded in December 2012 and is

headquartered in Columbia, SC

• CIRTEMO designs and manufactures patented optical filters: – Called Multivariate Optical Elements (MOE)

– Encoded to detect and measure complex chemical signatures or attributes.

• MOEs enable optical systems to – Detect and measure specific chemicals or attributes that

cannot be achieved with traditional optical filters

– Achieve better performances from optical components and systems

• CIRTEMO has 40+ patents around MOE technology platform

• CIRTEMO is partnering with – Optical Filter Manufacturers (OFMs)

– Optical System Manufacturers (OSMs)

3

Patented optics platform called Multivariate Optical Computing (MOC) licensed from University of South Carolina

Successfully commercialized MOC in many markets-Chemicals, Pharma, Food, Mining, Oil and Gas

CIRTEMO founded to commercialize MOC to all industries and applications outside of oil and gas

Commercializing new markets/applications with partners /customers (eg. life science, and medical devices,)

.

OMETRIC Founded OMETRIC sold to Halliburton in 2011 for

$XXM

New business model established to license

technology to partners/customers

Establishing new Intellectual Property

OMETRIC ™

CIRTEMO™

© CIRTEMO, LLC 2013. All Rights Reserved. CIRTEMO Products and Services are protected by U.S. and International issued patents and pending patent applications.

700

nm

Each spectrum is a vector in space

An optimal regression is observed

400 500 600 700 800

Mixture Spectra

Wavelength, nm

Inte

nsity

, a.u

. Optical Spectroscopy + Multivariate Calibration


• Optical spectroscopy is the study of the interaction between light and matter where each wavelength (or color) may provide insight into an unknown material’s composition

• Multivariate calibration is the utilization of many variables in order to predict a chemical/physical property of interest (i.e. analyte concentration)

• In complex chemical systems, a number of wavelengths at least equal to the number of independent chemical species is required for a calibration

• A special direction (or spectroscopic pattern) exists inside the data set that is related to the chemical measurements of interest but insensitive to spectroscopic interferences.

400 500 600 700 800

Pure Component Spectra

Wavelength, nm

Inte

nsity

, a.u

.

species 1 species 2 interferent

Find the pattern

Finding the Optimal Spectral Pattern…

5

Recognizing spectral patterns allow us to develop weighted regression vectors thus converting optical spectra into chemical/physical properties of interest


Multivariate Optical Computing

• Chemometrics – is a method for modeling multivariate data (eg. optical spectra) – Model parameters can be applied to data from a spectrometer (or series

of bandpass measurements) to estimate the composition of unknowns

6 © CIRTEMO, LLC 2013. All Rights Reserved. CIRTEMO Products and Services are protected by U.S. and International issued patents and pending patent applications.

Wavelength (nm)

Inte

nsi

ty

Sample #

Pre

dic

tion

Wavelength (nm)

Reg

ress

ion

• Multivariate Optical Computing (MOC) – is an alternative method for modeling multivariate optical spectra – Is the optical equivalent of a dot product in which simple optical systems

may achieve the sensitivity/specificity of a laboratory grade spectrometer.

– is mostly achieved by refinement of optical interference filter structures that we call Multivariate Optical Elements (MOEs)

– MOEs can be installed in a photometer to estimate or predict the composition of unknowns.

+

-

x =

The Multivariate Optical Element (MOE) Platform

• Multivariate Optical Computing is the optical equivalent of a dot product

– ŷ - estimated analytical property (eg. concentration)

– t - scaled regression vector – λ - analytical spectroscopic response

(eg. SWIR spectrum)

7

M.P. Nelson, J.F. Aust, J.A. Dobrowolski, P.G. Verly and M.L. Myrick "Multivariate Optical Computation for Predictive Spectroscopy " Anal. Chem. 70, 73-82 (1998).

ŷ = 𝒕𝒕 • 𝜆𝜆 = �𝑡𝑡𝒊𝒊 • 𝜆𝜆𝒊𝒊

𝑵𝑵

𝒊𝒊

• Multivariate Optical Elements (MOEs)

– are patented, wide-band, optical interference filters encoded with an application-specific regression (or pattern) to detect/measure complex chemical signatures.

– realize the measurement advantages of Multivariate Optical Computing (MOC)

– enable a filter based instrument to achieve the sensitivity/specificity of a laboratory spectrometer as well as convert a focal plane array into a real-time hyperspectral imager.

Multiplication Addition

λ1

Optical Filter (t1 = 0.9; t2 = 0.5)

∝ 0.9λ1

Detector

λ2

∝ (0.9λ1+ 0.5λ2)

∝ 0.5λ2

Wavelength (nm)

% T

ran

smis

sion

MOE


The Multivariate Optical Element (MOE) Platform

MOEs can be incorporated into optical systems in a variety of ways


MOE DetectorT

DetectorR

50 100 150 2000

20

40

60

80

100

Interested Wavelength

%R

50 100 150 2000

20

40

60

80

100


%T

Beamsplitter Configuration

MOE1 Detector

50 100 150 2000

20

40

60

80

100


%R

50 100 150 2000

20

40

60

80

100


%T

Filter Photometer Configuration

MOE2

ND

Snapshot Array Configuration

I1,1 I1,2 H1,1 H1,

2 G1,1

G1,2

E1,1 E1,2 F1,1 F1,2 D1,1 D1,2

C1,1 C1,2 B1,1 B1,2 A1,1 A1,2

I2,1 I2,2 H2,

1

H2,2

G2,1

G2,2

E2,1 E2,

2 F2,1

F2,2

D2,1 D2,

2

C2,1 C2,

2 B2,1

B2,2

A2,1 A2,

2

Example Spectral Regression Encoding with an MOE


Transmission (T)

Reflection (R)

Transmission -Reflection

50 100 150 200 0

20

40

60

80

100


%T

50 100 150 200 0

20

40

60

80

100


%R

50 100 150 200 -100

-50

0

50

100


T-R

A multivariate spectral regression may be

constructed by utilizing the transmission & reflection

profiles of the MOE

Multivariate Optical Elements vs. Bandpass Filters

• Multivariate Optical Elements (MOE) are not bandpass (BP) filters – MOEs possess a higher overall throughput than individual BP filters yielding a higher analyte

sensitivity based on superior SNR – MOEs sample more spectral wavelengths than discrete BP filters yielding a higher analyte specificity – MOEs are physically less complex than BP filters

• MOEs tend to exhibit fewer layers and overall filter thicknesses less than traditional band pass filters.

– Unlike well defined quarter wave optical thickness (QWOT) deposition recipes used for BP filter fabrication, there are multiple MOE solutions possible for any application

– Optimal MOE designs are selected based on a set of performance criteria inclusive of overall physical thickness and number of layers

• MOEs are fabricated via the same methods as traditional BP filters

10

BP1 BP2 BP3 BP4 BP5 BP6

Wavelength (nm)

% T

ran

smis

sion

Bandpass Filters

Wavelength (nm)

% T

ran

smis

sion

MOE

Multivariate Optical Element


Multivariate Optical Element Features & Benefits


Feature Benefit(s)

• Higher sensitivity than traditional bandpass filters

• Pure optical amplification of analyte signal permits lower detection limits

• Higher specificity than traditional bandpass filters

• Reduced crosstalk • Multiplexing opportunities (more analytes

can be detected simultaneously) in complex mixtures

• Higher signal-to-noise ratio measurement than traditional narrow bandpass filters

• Less sample material (smaller volume) can be used

• Less expensive/powerful subcomponents may be used

• Measurement flexibility • Environmental interference compensation may be rolled up into the MOE design

Multivariate Optical Elements can increase the sensitivity and specificity of analyte detection compared to bandpass filters.

Designing a Multivariate Optical Element (MOE)

• Traditional chemometric modeling identifies and exploits the variance within spectral (and reference) data to correlate with a feature/analyte of interest

• A definitive model is achieved most often by deconvolving the spectroscopic data into a projection in N-dimensional space (i.e. score)

• An MOE is designed through an iterative, non-linear optimization routine. – A local minimum response is

achieved based upon a random starting point

– a Newton-Raphson nonlinear optimization method is typically employed


Step 1:

Technical Feasibility

Step 2:

MOE Design

Step 3:

MOE Fabrication

Step 4:

System Integration

Working with CIRTEMO

• CIRTEMO can determine via modeling whether or not an MOE can provide value before fabricating actual filters

• Although we have had commercial success in a range of industries, each application is unique


• Step 1: Technical Feasibility – Collection of spectroscopic

calibration data – Convolution of radiometric data

• Step 2: MOE Design – Determination of spectral shapes – Optimize optical filter recipe

• Step 3: MOE Fabrication – Traditional hard coating deposition

(eg. RMS, IBD, etc.)

Visit us at www.cirtemo.com Drop us a line at [email protected]

For more information:

Documents

The Multivariate Optical Element Platformcirtemo.com/pdf/CIRTEMO_Technology_Overview.pdf · 2013. 10. 24. · Multivariate Optical Elements can increase the sensitivity and specificity