Epa Epd06084 Final Report

“sf” RTA P2005#84 EPA SBIR Phase II CN: EP-D-06-084 Phase II Final Report May 2010

1

Report Type: Phase II EPA SBIR Final Report Date of the report: May 2010 EPA Contract Number: EP-D-06-084 Title: Multiplexed Chemical Sensor for Water Security PI: Dr. Stuart Farquharson Business name: Real-Time Analyzers, Inc. EPA contact: James Gentry Project period: May 1, 2006 to May 31, 2010 Research category: SBIR, Topic I1. Drinking Water and Wastewater Security (2004) Purpose of the research: The overall goal of this proposed program (through Phase III) is to provide the EPA with a chemical sensor that can be multiplexed into water distribution systems to provide early warning of poisoned water supplies. Brief description of the research carried out: The overall goal of the Phase II program was to fully develop the proposed analyzer and improve sensitivity to detect poisons at 10 µg/L (10 parts-per-billion, ppb) in 10 minutes. This was accomplished by optimizing the surface-enhanced Raman active sol-gel chemical selectivity, ruggedizing the capillaries, developing a universal sampling system with a stream-to-capillary interface and a capillary-to-fiber optic probe interface, and developing a comprehensive analysis that included a searchable spectral library of 96 poison related chemicals capable of rapidly identifying these chemicals. Research findings: Twenty target chemicals, consisting of chemical agents, their hydrolysis products, simulants, pesticides and toxic industrial chemicals, were measured reproducibly at 10 µg/L (10 ppb) in 10 minutes, with a statistical confidence of 95% or greater! A universal and automated sampling system that controlled flow, pressure and delivery of water samples to the surface-enhanced Raman active capillaries, was successfully designed and used to measure samples from New York City’s Kensico Reservoir. This included measurement of methyl phosphonic acid (75 ppb), thiodiglycol (100 ppb), and cyanide (100 ppb), the primary hydrolysis products of the nerve agents, mustard gas, and cyanide salts, close to or exceeding the required detection limits (10, 100, and 6000 ppb, respectively). Potential applications: In addition to the proposed application of monitoring water supplies to ensure safe drinking water, the proposed analyzer could be used by first responders to assess safety of any water supply. It could also be used to evaluate groundwater contaminated by cyanide (leaching operations) or chromium (plating operations).

This entire document is considered Proprietary Information, except for this Cover Page and the Executive Summary.

Providing Chemical Information When & Where You Need It

362 Industrial Park Road Middletown, CT 06457


2

“Multiplexed Chemical Sensor for Water Security”

EXECUTIVE SUMMARY The overall goal of this proposed program (through Phase III) is to provide the EPA with a chemical sensor that can be multiplexed into water distribution systems to provide early warning of poisoned water supplies. This will be accomplished by developing surface-enhanced Raman (SER) sensors that can be integrated into water supply systems and coupled to a central Raman analyzer via fiber optics. The overall goal of the Phase II program was to fully develop the proposed analyzer and improve sensitivity to detect poisons at 10 µg/L (10 parts-per-billion, ppb) in 10 min (~the 5 day/5L values in Table E.1). This included optimizing the SER-active sol-gel chemical selectivity, ruggedizing the capillaries, developing a universal sampling system with a stream-to-capillary interface and a capillary-to-fiber optic probe interface, and developing a comprehensive analysis that includes rapid chemical identification. These goals were largely met as summarized by the following accomplishments: 1) The sol-gel chemistry was successfully optimized to achieve the required sensitivity of at least 10 µg /L (ppb) for all of the 20 target chemicals within 10 minutes using a solid phase extraction cartridge that was included in the sampling system. The Lowest Measured Concentrations (LMC) for these chemicals are summarized in Table E.1. SER spectra are shown in Figure E.1 for methyl phosphonic acid, thiodiglycol, and cyanide, the primary hydrolysis products of the nerve agents, mustard gas, and cyanide salts. Table E.1. Lowest Measured Concentrations (LMC) for the 20 primary chemicals studied compared to the Required Detection Limit (RDL, Military Drinking Water Guidelines, Short Term, 19991), along with the parent chemical agents, stimulants, pesticides, chlorinated by-products and their selected hydrolysis products.

Chemical Agents & Simulants (Abbreviation)

Hydrolysis Products

RDL (5-day/5L (µg/L (ppb)

LMC (µg/L (ppb))

Sarin (GB) isopropyl methylphosphonic acid (IMPA) 28 10 Soman (GD) pinacolyl methylphosphonic acid (PMPA) 12 10 Tabun (GA) ethyldimethyl-phosporamidate (EDMPA) 140 Cyclohexyl Sarin (GF) cyclo methylphosphonic acid (CMPA) 28 10 VX ethyl methylphosphonic acid (EMPA),

EA2192, methylphosphonic acid (MPA), di-isopropylamino ethanethiol (DIASH)

15 10 1 10

EA2192 same as for VX 30 Mustard (HD) thiodiglycol (TDG), 1,4-dithiane 100 10 2-chloroethyl ethylsulfide (CEES, half mustard)

2-hydroxyethyl ethyl sulfide (HEES) 10 10

2-chloroethyl methyl sulfide (CEMS, HD simulant)

same as for CEES 10

cysteamine S-phosphate sodium salt (CSPS, VX simulant)

10

Hydrogen Cyanide (HCN) cyanide (CN) 6000 0.1 Pesticides

chlorpyrifos (CP) trichloropyridinol (TCP) 40 10 fonofos (FON) O,O-dimethyl hydrogen thiophosphate,

potassium salt (DMHTP) 30 10

10 methyl parathion (MP) dimethylthiophosphoric acid DMTPA 300 10 disulfoton (DS) disulfoton sulfoxide (DS-SO) 140 10, 10

Chlorination By-products 3,5-dichlorobenzoic acid (DCBA) 1500 10 4,4-dichlorobiphenyl (DCBP) 1400 1

2) The two most active sol-gel chemistries were successfully developed to withstand flow rates of 5 mL/min and pressures of 30 psi. The sample system was successfully designed to reduce flow and pressure to at least these values.


3

3) Receiver operator characteristic (ROC) curves were used to demonstrate that the required sensitivity could be reproducibly achieved 95% of the time with a 3 minute spectral acquisition for methyl phosphonic acid, thiodyglycol, cyanide, fonofos, dichlorobenzoic acid, and sunset yellow (a food dye selected for field studies). However, this required scanning the length of the capillary (rastering, Figure E.2), which was not implemented in the final sampling system. 4) Software was written that successfully identified any of 96 chemicals within a spectral library data base, consisting of chemical agents, pesticides, toxic industrial chemicals, and hydrolysis products. The spectral search software ranks all of the chemicals based on the closest match to the unknown. The analysis is virtually instantaneous (<< 1 sec, Figure E.3)

5) A computer controlled sample system was designed and successfully built that was capable of being connected to virtually any water supply. It controls the water flow rate and pressure, directs the sample first through a solid phase extraction cartridge (for 5 min) into a waste container, then switches flow to pass methanol through the cartridge and transfer the concentrated sample to the SERS-active capillary (for 5 min). The flow is reset to introduce the next sample, while the SER spectrum is acquired using a fiber optic coupled Raman spectrometer (Figure E.3). Analysis is updated every 10 minutes.

0 20 40 60 80 100 Probability of False Positive, %

P

roba

bilit

y of

Det

ectio

n, %

100

80

60

40

20

0

Figure E.1. Surface-enhanced Raman spectra of 10 µg/L (ppb) methyl phosphonic acid (MPA), thiodyglycol (TDG), and sodium cyanide (CN).

Figure E.2. Receiver Operator Characteristic curves for 5 µg/L (ppb) methyl phosphonic acid using the raster program (green) and discrete points (red). The black line is the probability of a random guess (50/50).

Figure E.3. Spectral search software showing identification of unknown sample (10 ppb MPA in water) as MPA using the 96 component library. Conditions: 80 mW, 785 nm, 1-min. Note the following ranking of phosphate containing chemicals (a score greater than 0.4 represents a mismatch): Hit Quality Name 0 0.174 MPA 1 0.477 DEHDTP 2 0.498 EMPA 3 0.512 CMPA 4 0.522 DMHP 5 0.524 VX

95% Confidence Line

Methyl Phosphonic Acid

Spectral Match (MPA) Measured Unknown

Cyanide

Thiodiglycol


4

Figure E.4. A) Photograph of prototype Raman analyzer with fiber optic probe connected to B) the sampling system. The yellow line shows the sample flow through the SPE during the concentration step. The red line shows the flow of methanol through the SPE to the SERS Capillary during the extraction and delivery step. C) User Interface software used to control the sample and solvent flow. D) User interface of software used during operation (“More” shows spectral match as shown in Figure E.3). 6) The automated sample system in conjunction with a Raman analyzer was successfully used to detect 75 µg/L (ppb) methyl phosphonic acid artificially added to water samples obtained from the Kensico Water Reservoir, which supplies New York City its drinking water (Figure E.5). The raster method was NOT used, which improved sensitivity by more than a factor of 10 (and would therefore achieve the required sensitivity).

Figure E.5. Surface-enhanced Raman spectrum of 75 ppb MPA extracted from a Kensico Reservoir water sample using the Figure E.4B sample system (SPE and SERS Capillary). The total analysis time is approximately 10 minutes.

Sample In

Bypass

MeOH Waste

SPE

SERS Capillary

Waste

A B

D C


5

Limitations and Suggestions. Although 17 of 20 milestones were met, 3 were not. First, the proposed fiber-optic to SERS-capillary interface, which would use a Parker-Hannifin “Intraflow” machined block, was not pursued. This was largely due to the fact that Parker-Hannifin delayed delivery by more than 1 year of the sample system Intraflow components that they presumably already manufactured. Nevertheless, we successfully built a suitable (non- integrated) probe to perform the measurements. This probe could be readily modified to be a permanent component of the sample system. Second, the proposed measurements of actual nerve agents at the US Army’s Edgewood Chemical and Biological Research Center were never performed. Although the US Army provided a letter indicating that they would perform such measurements, and we mailed SERS-active capillaries to them for these measurements, they were not able to fit these measurements into their schedule. We understand their priorities have changed to detecting biological warfare agents. Third, the at-site measurements at Kensico Reservoir were never performed. This was due to the fact that the prototype system requires a number of modifications to perform these tests correctly. These modifications include: 1) developing a motorized fiber optic probe system to “scan” the SERS capillary to achieve the necessary sensitivity and/or 2) incorporating additional solid phase extraction material into the SERS-active sol-gel to overcome sample dilution due to the sample system channels to achieve the necessary sensitivity, 3) mounting the probe to the sample system, 4) completing the top level software user interface so that it incorporates a) the flow control software, b) the Raman analyzer control software, c) the chemical identification software, d) the ROC curve concentration software with alarms. It should be noted that the personnel at Kensico Reservoir were willing to perform the proposed measurements using the food dye sunset yellow, which we measured at 1 µg/mL (ppb). Fourth, although the proposed prototype was built using matching funds and tested using the Commercialization Option funds, due to the limitations cited above (primarily sensitivity), the proposed additional field tests were not performed. Finally, for these same reasons, the Verification Option was not exercised. Finally, it is worth noting that we (Real-Time Analyzers) have continued talks with Hach and GE Power & Water (March, 2010, April 2010, respectively), and will pursue Phase III commercialization with these companies as partners.


6

TABLE OF CONTENTS

Cover Page ................................................................................................................................................... 1 Executive Summary ..................................................................................................................................... 2 Table of Contents ......................................................................................................................................... 4 Program Description .................................................................................................................................. 4

1. Identification and Significance of the Problem or Opportunity ................................................................... 6 2. The Phase II Program Schedule (verbatim from the Phase II proposal) ...................................................... 8 3. Detailed Summary of Phase II Work ........................................................................................................... 8

Task 1. Refine sol-gel selectivity and sensitivity ..................................................................................... 9 Task 2. Develop SER-active capillary durability .................................................................................. 20 Task 3. Develop Spectral Library.......................................................................................................... 30 Task 4. Test Sol-Gel Capillaries with Real Water Samples ................................................................... 59 Task 5. Establish Performance (ROC Curves) ...................................................................................... 63 Task 6. Design and Test Sample System .............................................................................................. 75 Task 7. Field Tests ................................................................................................................................ 87

4. Conclusions ............................................................................................................................................... 89 5. Publications ............................................................................................................................................... 89 6. Commercialization .................................................................................................................................... 90

References ................................................................................................................................................. 90 Appendix (Publications) ............................................................................................................................ 92

PROGRAM DESCRIPTION

1. Identification and Significance of the Problem or Opportunity The nation’s drinking water supply is a potential target for terrorists. Chemical sensors or sensor systems are needed to provide an early warning of poisoned water supplies to protect US citizens. This program will develop such a sensor to meet this need. The remainder of this section is verbatim from the Phase I Proposal. The Phase II summary begins on page 8. 1.a. The Problem or Opportunity - Countering terrorist attacks requires recognizing likely deployment scenarios and having the required technology to rapidly detect the deployment event. In addition to the expected use of chemical agents released into the air, terrorists may also poison water supplies. The National Strategy for Homeland Security designates the Environmental Protection Agency with the task of securing the nations drinking water.2 In response the EPA has defined four broad categories of both public and private labs based on analysis type: environmental, radiochemical, biotoxins, and chemical weapons.3 In the case of attack (real or suspected) the EPA has developed protocols for collecting samples, sending them to EPA designated labs, and performing analyses. In the case of chemical warfare agents (CWAs) there are only two labs with the surety to perform these analyses. Furthermore, the analytical methods employed (e.g. gas chromatography coupled with mass spectrometry) require sample extraction and calibration, and are time consuming. The positive determination that a water supply is contaminated is likely to take as much as a day. This is entirely inadequate for the prevention of widespread illness, death, and terror. To overcome this limitation, the EPA has identified a number of field test kits, but unfortunately, they lack chemical selectivity and yield positive response to other chemicals. Clearly, a system of integrated sensors with chemical specificity is needed to monitor the safety of drinking water in real-time. 1.b. The Innovation - We at Real-Time Analyzers (RTA) believe that a series of chemical sensors, based on surface-enhanced Raman (SER) spectroscopy, can be multiplexed into water distribution systems to provide early warning of poisoned water supplies. The proposed analyzer would employ sensors composed of 1 mm diameter windows coated with chemically selective sol-gels that will extract and concentrate target chemical agents, pesticides or harmful industrial chemicals from flowing streams (Figure 1.1). The sol-gel sensor will also contain metal particles to generate SER spectra of these analytes allowing detection of 10 microg/L in 10 minutes. The coated sensor windows will be hermetically mounted to a stainless steel flange, such that it can handle high pressures (100 psi) and variable temperatures (5-40 C). Fiber optics will allow placement of sensors at critical nodes throughout the distribution system at distances up to 100 m from a central Raman analyzer. A 10-position fiber optic coupling wheel interfaced to the Raman analyzer will allow sequential analysis of the nodes. Multiplexing software and supervisory control and data acquisition (SCADA) system software will provide the appropriate

Confidential Proprietary Information


7

Water Pipe SER-Active

Window

Fiber OpticCoupled Probe

Fibers to RamanAnalyzer

warnings to EPA and/or water authority personnel. The proposed SERS-active, chemically-selective sensors are based on several important advances developed at RTA. First, we have developed and patented a sol-gel process that incorporates silver and/or gold nanoparticles into a stable porous silica matrix.4,5 Second, no special reagents, conditions or sample treatments are required, and aqueous solutions ranging in pH from 2 to 11 can be used. Third, the sol-gel process is highly reproducible, and we guarantee SER-activity at 20% RSD for our Simple SERS Sample Vials, now sold commercially for 2 years (Figure 1.2). Fourth, we have used these metal-doped sol-gels to measure SER spectra of several hundred chemicals,6-11 with typical detection limits of 10 mg/L using 100 mW of 785 nm and 3-min acquisition time. Fifth, we have measured cyanide (CN), mustard (HD), VX, as well as several CWA hydrolysis products, (e.g. pinacolyl methyl phosphonic acid), and numerous pesticides at 10 mg/L in 1 minute, with estimated limits of detection of 1 mg/L (Figure 1.3). Sixth, the choice of metal and alkoxide can be used to develop chemically selective sensors. We have successfully developed sol-gels that select for polar-positive, polar-negative, weakly polar-positive, and weakly polar-negative chemical species.12 Seventh, we can coat a variety of surfaces to produce a wide range of sensor designs, including the glass discs proposed here. We have filled capillaries and micro-channels to detect flowing samples, as well as to perform chemical separations.13 Finally, it is worth noting that RTA has developed an extremely rugged, compact Raman instrument that employs interferometry for absolute wavelength accuracy and an avalanche Si detector that improves sensitivity by ~100 times.14 And we are currently developing a portable, battery powered version of the system for the Navy, which will weigh 21.5 pounds, occupy 0.5 cubic foot, require 23.5 W, and will be capable of wireless communication.15

Figure 1.1. Idealized concept of a SERS-based chemical agent detection node integrated into a water distribution system.

0 30 60 90 120 150 180 210 24

0 270 300 33

0

6

9

12

15

051015202530354045 40.0-45.0

35.0-40.030.0-35.025.0-30.020.0-25.015.0-20.010.0-15.05.0-10.00.0-5.0

Height along vial (mm)

1008

cm

-1 b

and

inte

nsity

for B

A

15o increments around vial Figure 1.2. Reproducible SER-intensity response for benzoic acid over entire surface of a Simple SERS Sample Vial. Average = 29.1± 4.26 (14.6%) for 240 points (10 sec per point).

C

A

B

Figure 1.3. SER spectra of A) cyanide, B) pinacolyl methyl phosphonic acid and C) fonofos, all at 10g/L in water. Conditions: 100 mW of 785 nm, 1-min.


8

2. The Phase II Program Schedule (verbatim from the Phase II proposal) Performance Schedule (24 Months) Quarters 0 1 2 3 4 5 6 7 8 Task 1 - Refine sol-gel chemistry (6 mo) 1 2 3 . Task 2 - Develop capillary durability (4 mo) 4 5 . Task 3 - Develop spectral library (2 mo) 6 7 . Task 4 - Identify real sample interferents (3mo) 8 9 10 . Task 5 - Define performance (ROC) (4 mo) 11 12 . Task 6 - Design and test sampling system (3 mo) 13-16. Task 7 - Perform field tests (2 mo) 17 18 19 20.

Milestones 1. Prepare various Chem. 2 sol-gel capillaries 2. Test extraction and SER-activity for 20 analytes

11. Measure 5 conc, 5 samples 12. Plot ROC curves, calc. K

3. Determine baseline LODs 4. Develop capillary coating process

13. Finalize sample system design 14. Build SERS-capillary interface

5. Test flow and pressure integrity 6. Measure chemicals, calc LODs 7. Test library search algorithms 8. Measure artificially prepared interferents 9. Measure real sample interferents 10. Design solution (filter) as needed

15. Build FO interface 16. Test sample system 17. Obtain Kensico samples 18. Perform Edgewood measurements 19. Perform Kensico measurements 20. Summarize Data in a Final Report.

All data will be summarized in terms of existing and estimated (Phase III product) capabilities in the Final Report.

3. Detailed Summary of Phase II Work The overall goal of Phase II will be to fully develop the proposed analyzer and improve sensitivity to allow detection at 10 microg/L in 10 min (~the 5 day/5L values in Table 2). This will include ruggidizing the SER-active sol-gel capillaries, developing a universal sampling system with a stream-to-capillary interface and a capillary-to-fiber optic probe interface, and developing comprehensive analysis that includes rapid chemical identification. Capabilities will be developed using real-world samples and performing two field tests. The American Water Works will supply numerous samples from the NJ drinking water distribution system. One field test will involve the analysis of a food dye artificially added to a NYC water supply, while the other field test, performed at the US Army’s Edgewood Chem Bio Research Center, will involve the analysis of HD and VX added to a close-loop water test system.


9

Task 1. Refine sol-gel selectivity and sensitivity. The overall objective of this task is to finalize the sol-gel chemistry so that the target chemicals can be detected at the required concentrations (e.g.10 microg/L (10 ppb)). This will be accomplished by improving chemical extraction and taking advantage of improved SER-active sol-gels to detect 20 targeted chemicals.

Twenty primary target chemicals listed in Table 1.1 (Bold) were tested in this Task (as well as in Task 3) with a refined subset of the SERS-active sol-gel chemical libraries summarized in Table 1.2. The SER activities and lowest measured concentrations for the 20 chemicals and the 6 sol-gel libraries are summarized in Table 1.3. Table 1.1. Chemical agents, stimulants, pesticides, chlorinated by-products and selected hydrolysis products with the Military Drinking Water Guidelines (Short Term, 1999).16 Bold indicates the 20 primary chemicals to be included in this study.

Chemical Agents & Simulants

5-day/5L (mg/L)

5-day/15L (mg/L)

Hydrolysis Half-Life*

Hydrolysis Products

Sarin (GB) 0.028 0.0093 21.3 hours IMPA Soman (GD) 0.012 0.004 2.3 hours PMPA Tabun (GA) 0.14 0.046 4.1 hours CN/EDMPA Cyclohexyl Sarin (GF) 0.028 0.0093 21.3 hours CMPA VXL 0.015 0.005 82.1 hours EMPA, EA2192, MPA, DIASH EA2192 0.03 0.01 9 years same as for VX Mustard (HD)L 0.1 0.05 2-30 hours TDG, 1,4-dithianeL CEES (1/2mustard) 2-4hrs HEES CEMS (HD simulant) 1-2 hrs same as for CEES CSPS (VX simulant) HCN 6 2 (0.2)** Stable CN

Pesticides CP 0.04 0.014 (0.09)** 35-78 days TCPL FON 0.03 0.0009 2 days DMHTP (EEPA) MP 0.3 0.1 (0.002)** 11.2 days DMTPAL DS 0.014 0.005 (0.01)** 5-12 hrs DS-SO Chlorination By-products

DCBA 1.5 0.5 (0.06)** DCBP 1.4 0.5

* pH 7 to 7.5 and 20 to 25 oC, ** EPA MCL, L = to be measured as part of the spectral library. CN = cyanide, IMPA = isopropyl methylphosphonic acid, PMPA = pinacolyl methylphosphonic acid, EDMPA = ethyldimethyl-phosporamidate, EMPA = ethyl methylphosphonic acid, CMPA = cyclo methylphosphonic acid, MPA = methylphosphonic acid, DIASH = di-isopropylamino ethanethiol, TDG = thiodiglycol, CEES = 2-chloroethyl ethylsulfide, HEES = 2-hydroxyethyl ethyl sulfide, CEMS = 2-chloroethyl methyl sulfide, CSPS = cysteamine S-phosphate sodium salt, TCP = trichloropyridinol, EEPA = O-ethyl ethylphosphonothioic acid, DMHTP = O,O-dimethyl hydrogen thiophosphate, potassium salt, DMTPA = dimethylthiophosphoric acid, DS = disulfoton, DS-SO = disulfoton sulfoxide, DCBA = 3,5-dichlorobenzoic acid, DCBP = 4,4-dichlorobiphenyl, CP = chlorpyrifos, FON = fonofos, and MP = methyl parathion. Italics mean not in SERS library; of these GA, GB, GD, EEPA and DMTPA have not yet been measured. We are in process of measuring G-agents at Aberdeen. Chemicals in parentheses mean could not be purchased. Table 1.2. Updated summary of chemically-selective, SERS-active, sol-gel libraries (capillaries) used in Task 1.

Chemistry Selectivity/Analyte Metal Solution A/ Metal precursor A (µL) B (µL) Solution B/ Sol-gel precursor Type M Component ratio volume volume Si-Alkoxide ratio + additional

components standard P-C Ag 1N AgNO3/28%NH3OH/CH3OH/H2O L1 polar - negative silver 5:1:10:0 100 120 TMOS/MTMS (1:6) L2 non-polar - negative silver 5:5:10:0 100 100 MTMS L3 L4 L6

non-polar – negative P-C polar - positive OTC polar-negative

silver Au gold Ag silver

5:5:10:0 0.25N HAuCl4 /70% HNO3

4:1 1N AgNO3 /0.026M NaBH4

3:1

100 100 120

175 100 120

MTMS/ODS/TMOS (5:1:1) TMOS APTES (95/5 in EtOH)

modified L1_PEG L2_PEG L3_PDMS

P-C (polymer) polar - negative non-polar – negative non-polar – negative

Ag silver silver silver

1N AgNO3/28%NH3OH/CH3OH/H2O 5:1:10:0 5:5:10:0 5:5:10:0

100 100 100

120 120 100

TMOS/MTMS (1:6) + 10 PDMS TMOS/MTMS (1:6) + 10 PEG MTMS/ODS/TMOS (5:1:1) + 10 PDMS

Note: Initial metal precursors/reagents prepared in water unless noted. Sol-gel chemistries in Italics are new chemistries developed during two NASA PII SBIRs. P-C designates filled packed-capillaries and OTC designates open tubular capillaries. Si-alkoxide precursors: TMOS (tetramethyl-orthosiloxane), MTMS (methyltrimethoxy-siloxane), ODS (octadecyl-trimethoxysilane) and APTES (3-aminopropyltrimethoxysilane). Polymer additives: PDMS (polydimethyl-siloxane) and PEG (polyethyleneglycol).


10

Table 1.3. SERS-activity (+/-) and LMC (ppb) summary of the 20 target chemicals measured (on L1-L6). CN CSP

S MPA DIASH EMPA IMPA CMPA PMPA CEES CEMS TDG HEES

L1 + + + - + + + + + +

peg + + + + + + + + + + + -

L2 + + + + + + +

peg + + + + - + - + + + -

L3 + + + + + + + + + + + +

pdms + + + +

2/1 +

L4 + - - + - - - - + + - -

L6 + + + + + + + + + + + +

Static L2 1k L6 1k

L1 1k L6 10

L3 1k L6 10

L3100 L6 10

L3 1k L6 100k

L3 10k L61000k

L3100k L61000k

L3 10k L6 100k L3pdms10k

L3100k L6 1k L4 1k

L6 1k L4 1k

L1 10k -1kL3

Flow L2 1 L32/1

500

L2 1k L3 1k L2 100

-10 L3

L3 1k -500L1peg

-1kL2

-1k L2 -1k L3pdms

L3 1k L2 100 L4 50

L3 500wk L11k L2 1k

SPE AIEX

L2 0.1

L6 10 L3 10 L6 10

L3 1k L2 10 L6 1

L3 10

L3 10

L3 10

L2 10

L2 10 L2 10

L1 10 L4 10 L3 10

L2 10 L1peg 10

L1peg10

L6 10

CP MP FON DS DSSO DMHTP DCBA 44DCBP

L1 + + + + +

peg + + + +

L2 + + + + +

peg + +

L3 + + + + + + +

pdms -

2/1 +

L4 - - wk - - - - +

L6 + + + + + + + +

Static L3 1k L6 100

L3 1k L6 50 L32/1500

L2 1k L6 10

L6 10

L3 1k -1k L3 L4 10

Flow L3 1k -1k L2 L1peg 1k

L2 1k

L2 100 L32/150

L3 1k -1k L2 -1kL1peg

-1k L2 -1k L1peg

-10 L3 -1k L2

L2 1k L4 0.1

SPE AIEX

L3 10 L3 10

L3 10 L1peg10

L3 10 L3 10 L6 10

L6 10 L2 10 L3 10

L2 10 L2 10

Red indicates successful low ppb measurements. SERS measurements were performed in glass capillaries (1.1 mm outer diameter, 0.8 mm inner diameter) filled with metal-doped sol-gels. The basic design and use of the SER-active capillaries is shown in Figure 1.1, and are prepared as follows. The alkoxide and amine precursors are prepared according to Table 1.2, mixed, and then drawn into the capillary by syringe. Typically 0.1 mL of solution coats a 4 cm length of capillary. The sol solution gels in 5 minutes, and a more rigid structure is obtained after 24 hours. A solution of 0.1g/100mL NaBH4 is drawn through the capillary to reduce the metal. This is followed by a 0.035% HNO3 acid wash, and then the capillary is ready to be used.


11

Fig.1.1. Photograph of silver-doped sol-gel coated melting point capillaries attached to syringes, before (top) and after (bottom) reduction with sodium borohydride. Note: capillaries filled with gold-doped sol-gels are similarly prepared, but reduced twice. Initial screening of the sol-gel SER-activity employed 1 mg samples in 1 mL HPLC grade methanol or water. For “static” measurements, 50 µL of the samples were drawn into the capillaries, which were mounted on an XY sample stage above a fiber optic probe coupled to RTA’s Industrial Raman Analyzer. Spectra were obtained using 80 to 100 mW of 785 nm excitation at the sample and 1 minute acquisition time unless noted otherwise. Once the initial screening was performed, the stock samples were serially diluted to determine sensitivity. Also, normal Raman (NR) spectra of the analytes were acquired as pure solids, neat liquids or solutions (in appropriate solvents) in glass capillaries (or vials) using 300 mW at 785 nm for 5 minutes. Table 1.3 above shows the activity table for the 20 primary chemicals screened on the different chemistries, as well as the lowest measured concentrations (LMC) using the static, flow and solid phase extraction methods. Figures 1.2 to 1.21show a stack plot of the NR (solid), NR (solution) and SERS respectively for the 20 target chemicals. Most of the SERS are for 100 or 1000 ppm (0.1 or 1 mg/mL) and are intended to clearly show the Raman peaks (high signal-to-noise ratios) and are not intended to show the lowest possible concentrations (which come later).

Fig.1.2. A) NR (solid), B) NR solution (20 mg/mL in HPLC water) and C) SERS of 100 ppm sodium cyanide in water on chem. L2_PEG. Conditions: A) and B) 200mw, 785 nm, 5-min, and C) 80mW, 785 nm, 1-min.

A B C

Fig.1.3. A) NR (solid), B) NR solution (30 mg/mL in HPLC water) and C) SERS of 1000 ppm cysteamine S-phosphate sodium salt in water on chem. L2. Conditions: as in Fig.1.2.

A B C

Fig.1.4. A) NR (solid), B) NR solution (400 mg/mL in HPLC water) and C) SERS of 1 ppm methylphosphonic acid in water on chem. L6. Conditions: as in Fig.1.2.

A B C

Fig.1.5. A) NR (solid), B) NR solution (1000 mg/mL in HPLC water) and C) SERS of 1000 ppm di-isopropylamino ethanethiol in methanol on chem. L2_PEG. Conditions: as in Fig.1.2.

A B C


12

Fig.1.6. A) NR (solid), B) NR solution (200 mg/mL in 3:1 HPLC water/MeOH) and C) SERS of 1000 ppm ethyl methylphosphonic acid in methanol on chem. L3. Conditions: as in Fig.1.2.

A B C

Fig.1.7. A) NR (solid), B) NR solution (250 mg/mL in HPLC water) and C) SERS of 1000 ppm isopropyl methylphosphonic acid in methanol on chem. L2. Conditions: as in Fig.1.2.

A B C

Fig.1.8. A) NR (solid), B) NR solution (saturated in 1N KOH) and C) SERS of 1000 ppm cyclo methylphosphonic acid in methanol on chem. L2. Conditions: as in Fig.1.2.

A B C

Fig.1.9. A) NR (solid), B) NR solution (200 mg/mL in 3:1 HPLC water/MeOH) and C) SERS of 1000 ppm pinacolyl methylphosphonic acid in methanol on chem. L2. Conditions: as in Fig.1.2.

Fig.1.10. A) NR (neat liquid) and B) SERS of 1000 ppm 2-chloroethyl ethylsulfide in methanol on chem. L3. Conditions: as in Fig.1.2.

Fig.1.11. A) NR (neat liquid), and B) SERS of 1000 ppm 2-chloroethyl methyl sulfide in methanol on chem. L3_PDMS. Conditions: as in Fig.1.2.

A B C

A B

A B


13

Fig.1.12. A) NR (neat liquid), and B) SERS of 1000 ppm thiodiglycol in methanol on chem. L3. Conditions: as in Fig.1.2.

A B Fig.1.13. A) NR (neat liquid), and B) SERS of 1000 ppm

2-hydroxyethyl ethyl sulfide in water on chem. L3. Conditions: as in Fig.1.2.

Fig.1.14. A) NR (solid), B) NR solution (100 mg/mL in methanol) and C) SERS of 1000 ppm chlorpyrifos in methanol on chem. L2. Conditions: as in Fig.1.2.

A B

A B C

Fig.1.15. A) NR (solid), B) NR solution (neat liquid) and C) SERS of 500 ppm methyl parathion in methanol on chem. L2_PEG. Conditions: as in Fig.1.2.

A B C

Fig.1.16. A) NR (neat liquid) and B) SERS of 1000 ppm fonofos in methanol on chem. L3. Conditions: as in Fig.1.2.

A B

Fig.1.17. A) NR (neat liquid) and B) SERS of 1000 ppm disulfoton in methanol on chem. L1. Conditions: as in Fig.1.2.

A B


14

From the above it was found that chem. L2, L3 and L6 are universal chemistries for routine screening of the 20 target chemicals at nominal concentrations. Once the screening was performed, the chemicals were serially diluted to 10 ppb (desired sensitivity). Initially, two methods were experimented with to achieve the desired sensitivity. The first method was the static method, which simply involved loading a small sample plug over the sol-gel and making SERS measurement. The second method was the flow method, which involved flowing a fixed volume of the sample through the sol-gel capillaries. The static method, in general, allowed measuring a few chemicals in the range of 100 ppm to 10 ppb (for e.g. MPA, DS and DSSO at 10 ppb on chem. L6, CN at 100 ppb on chem. L2 and CEES at 1 ppm on chem. L4) as can be seen from Figures 1.22 (A) to 1.41 (A). Unfortunately, the static method could not achieve the desired sensitivity of 10 ppb for all the 20 target chemicals. Thus flow experiments were performed to test the ability of the sol-gels to extract the target analytes by measuring the SERS signal as the sample flowed through a capillary as a function of time. For the flow experiments, a syringe pump (Sage model 341B, Thermo Electron, Waltham, MA) was used to flow the sample (10-50 mL) through the capillary at a rate of 1 mL/min until a signal was observed. SER spectral collection was initiated when the sample solution entered the capillary and spectra were collected continuously (20 sec/spectrum) for ~5-30 minutes. This was followed by stop-flow measurements (i.e. measure 9 points along the capillary after flowing the fixed volume of sample). The flow method, in general, allowed measurements in the range of 1ppm to 10 ppb for few of the target chemicals, (for e.g. CN on chem. L2, DIASH on chem.L3, and DCBP on chem. L4 at 10 ppb, CEES on chem. L4 at 50 ppb and EMPA on chem. L3 at 1 ppm). See Figures 1.22 (B) to 1.41 (B). Unfortunately, the flow method also did not achieve the desired sensitivity of 10 ppb for all the 20 target analytes.

Fig.1.18. A) NR (neat liquid) and B) SERS of 1000 ppm disulfoton sulfoxide in methanol on chem. L6. Conditions: as in Fig.1.2.

A B

Fig.1.19. SERS of 1000 ppm dimethyl hydrogen thiophosphate in methanol on chem. L3. Conditions: as in Fig.1.2. Note; no pure sample was available to generate NR.

Fig.1.20. A) NR (solid) and B) SERS of 1000 ppm 3,5-dichlorobenzoic acid in methanol on chem. L3. Conditions: as in Fig.1.2.

Fig.1.21. A) NR (solid) and B) SERS of 1000 ppm 4,4-dichlorobiphenyl in methanol on chem. L3. Conditions: as in Fig.1.2.

A B

A B


15

In an effort to improve sensitivity several variations to the 3 chemistries (L3, L6 and L4) were performed by changing the ratios of the metal to the alkoxide and by the addition of polymers like polydimethyl siloxane (PDMS) and polyethylene glycol (PEG). Table 1.4 shows the initial modifications made on chem. L3. NOTE: not all permutations were tried. Table 1.4. Initial modifications for making chem.L3 more polar.

Chem Selectivity/Analyte Metal Solution A/ Metal precursor A(µL) B (µL) Solution B/ Sol-gel precursor Type M Component ratio volume volume Si-Alkoxide ratio P-C Ag 1N AgNO3/28%NH3OH/MeOH MTMS/ODS/TMOS L3

non-polar - negative

Silver

5:5:10 A = 25 + 25 + 50

100

175

5:1:1 B = 125 + 25 + 25 5:1:1

L31

less non-polar – neg. more polar – neg.

Silver

A = 25 + 25 + 50

100 100 100 100 100

175 175 175 175 175

B = 125 + 20 + 30 B = 125 + 15 + 35 B = 125 + 10 + 40 B = 125 + 5 + 45 B = 125 + 0 + 50 5:0:2

L32

less non-polar – neg. more polar – neg.

Silver

A = 25 + 25 + 50

100 100 100 100 100

170 165 160 155 150

B = 125 + 20 + 25 B = 125 + 15 + 25 B = 125 + 10 + 25 B = 125 + 5 + 25 B = 125 + 0 + 25 5:0:1

Two alkyl phosphonic acids (APA’s), PMPA and EMPA were chosen for preliminary screening with the modified chem.L3. The preliminary results obtained for PMPA indicate that it is enhanced on the standard non-polar chem.L3 as opposed to the more polar L.3 chemistries (L.3_PDMS>L3>L32&L31). In contrast, the SERS-response of EMPA appeared to be improved with these new modified chem. L3 polar subsets (L32>L31>L3). Note: L3_PDMS = inclusion of 10 microL of PDMS within chemistry L3. (chem.L3_PDMS) provided a static detection of PMPA in water at 10 ppm, as opposed to 100 ppm for chem.L3. Similarly, 3 modifications were performed on chem.L6. In the first modification, 0.5% APTMS solution was used in place of a 1% APTMS. This produced a better signal at 1 ppm, but the thinner coat resulted in less reproducibility. In the second modification, a 10/1 ratio of 1% APTMS and 1% MTMS was used to make a more hydrophobic coating. In this case also MPA was only detected at 1 ppm. In the third modification, 1% amino propyl 3-ethoxysilane (APTES) was used instead of APTMS. This resulted in a more uniform coat and better reproducibility and MPA was reproducibly detected at 1 ppm to 125 ppb, and sporadically at 100 ppb.

Finally, modifications to chem.L4 were made. This included making the chemistry more non-polar with the addition of MTMS (6/1 TMOS/MTMS, chem3b), 1/1 TMOS/MTMS (chem3c) and PTMS (1/1 TMOS/PTMS chem4b_PTMS). CEES, CEMS and 44DCBP were measured on the modified chemistries. The SERS-response is much weaker on chem3c and chem4b_PTMS with respect to chemL4. Static chemL4 LMC = 10 ppb for 44DCBP while flowing 10 mL aqueous sample improved the LMC to 100 ppt. These modifications also however did not allow detection at the required sensitivity for all the 20 target chemicals using the flow method. However, it is worth pointing out that the standard and modified L1-L6 libraries were used to further test and extensively screen over 71 additional chemicals in Task 3, spanning a wide range of different classes and subclasses of relevance. As an alternative method to flow we investigated the use of solid phase extraction (SPE) and ion-exchange (IEX) techniques as a way to pre-concentrate the sample to achieve the desired sensitivity. SPE is a form of chromatography designed to extract, partition, and/or adsorb one or more components (sample) from a liquid phase (sample matrix) onto stationary phase (sorbent). The steps involve 1) conditioning the SPE sorbent with methanol and water, 2) flowing the sample through the sorbent and 3) eluting the extracted sample with a solvent. As part of the method development, we experimented with various SPE sorbents (C2, C8 and C18), IEX sorbents (anion exchange and cation exchange), different flow rates, and eluting solvents (methanol, acetonitrile, dichloromethane and hexane). These experimental conditions allowed us to develop SPE/IEX methods for all the 20 target analytes. Table 1.5 lists the different SPE/IEX conditions for all the target chemicals, while Figures 1.22 (C) to 1.42 (C) shows the SERS of the 20 chemicals at10 ppb obtained using the SPE/IEX pre-concentration method.


16

Table 1.5. SPE/IEX sample pre-concentration conditions for the 20 target chemicals. Chemical Sorbent Elution Solvent Procedure CN AEX 0.01M HNO3 Load 50 mL sample at 0.5mL/min, wash sorbent with water, elute with 0.4 mL of 0.01 M

HNO3

CSPS ENVI-CARB

MeOH Load 50 mL sample at 1mL/min, wash sorbent with water, elute with 0.2 mL of MeOH

MPA mixed mode (C8+AEX)

0.01M HCl in MeOH Load 50 mL sample at 0.5mL/min, wash sorbent with water, elute with 0.4 mL of 0.01M HCl in MeOH

DIASH C8 MeOH Load 50 mL sample at 1mL/min, wash sorbent with water, elute with 0.2 mL of MeOH EMPA AEX 0.01 M NaCl Load 50 mL sample at 0.5mL/min, wash sorbent with water, elute with 0.3 mL of 0.01 M NaCl IMPA AEX 0.01 M NaCl Load 50 mL sample at 0.5mL/min, wash sorbent with water, elute with 0.3 mL of 0.01 M NaCl CMPA AEX 0.01 M NaCl Load 50 mL sample at 0.5mL/min, wash sorbent with water, elute with 0.3 mL of 0.01 M NaCl PMPA AEX 0.01 M NaCl Load 50 mL sample at 0.5mL/min, wash sorbent with water, elute with 0.3 mL of 0.01 M NaCl CEES C8 MeOH Load 50 mL sample at 1mL/min, wash sorbent with water, elute with 0.2 mL of MeOH CEMS C8 MeOH Load 50 mL sample at 1mL/min, wash sorbent with water, elute with 0.2 mL of MeOH TDG DPA-6S MeOH Load 50 mL sample at 1mL/min, wash sorbent with water, elute with 0.2 mL of MeOH HEES C8 MeOH Load 50 mL sample at 1mL/min, wash sorbent with water, elute with 0.2 mL of MeOH CP C18 DCM/MeOH (4:1v/v) Load 50 mL sample at 1mL/min, wash sorbent with water, elute with 0.2 mL of

DCM/MeOH (4:1v/v) MP C18 DCM/MeOH (4:1v/v) Load 50 mL sample at 1mL/min, wash sorbent with water, elute with 0.2 mL of

DCM/MeOH (4:1v/v) FON C18 DCM/MeOH (4:1v/v) Load 50 mL sample at 1mL/min, wash sorbent with water, elute with 0.2 mL of

DCM/MeOH (4:1v/v) DS C8 MeOH Load 50 mL sample at 1mL/min, wash sorbent with water, elute with 0.2 mL of MeOH DSSO C8 MeOH Load 50 mL sample at 1mL/min, wash sorbent with water, elute with 0.2 mL of MeOH DCBA C8 MeOH Load 50 mL sample at 1mL/min, wash sorbent with water, elute with 0.2 mL of MeOH 44DCBP C18 MeOH Load 50 mL sample at 1mL/min, wash sorbent with water, elute with 0.2 mL of MeOH AEX=anion-exchange, MeOH=methanol, C8=silica based reversed-phase packing with monomerically bonded octyl (9%) carbon load, DPA-6S=polyamide resin with reverse phase retention mechanism, C18= silica based reversed-phase packing with monomerically bonded octadecyl (18%) carbon load, DCM=dichloromethane. Figures 1.22 to 1.41 shows the stack plot of the SER spectra of the LMC obtained using the static, flow and the SPE/IEX methods, respectively.

Fig.1.22. LMC for CN obtained by A) static (100 ppb, chem.L2), B) flow (10 ppb, chem.L2), and C) SPE (10 ppb, chem.L2). Conditions: as in Fig.1.2.

A B C

Fig.1.23. LMC for CSPS obtained by A) static (10 ppb, chem.L6), B) flow (1000 ppb, chem.L2), and C) SPE (10 ppb, chem.L3). Conditions: as in Fig.1.2.

A B C


17

A B C

Fig.1.24. LMC for MPA obtained by A) static (10 ppb, chem.L6), B) flow (100 ppb, chem.L3), and C) SPE (1 ppb, chem.L6). Conditions: as in Fig.1.2.

A B C

Fig.1.25. LMC for DIASH obtained by A) static (1000 ppb, chem.L3), B) flow (10 ppb, chem.L3), and C) SPE (10 ppb, chem.L3). Conditions: as in Fig.1.2.

Fig.1.26. LMC for EMPA obtained by A) static (1000 ppb, chem.L3), B) flow (1000 ppb, chem.L3), and C) SPE (10 ppb, chem.L3). Conditions: as in Fig.1.2.

A B C

Fig.1.27. LMC for IMPA obtained by A) static (100 ppm, chem.L3), and B) SPE (10 ppb, chem.L3). Conditions: as in Fig.1.2. NOTE: No SERS from flowing 1 ppm.

A B

Fig.1.28. LMC for CMPA obtained by A) static (100 ppm, chem.L3), and B) SPE (10 ppb, chem.L2). Conditions: as in Fig.1.2. NOTE: No SERS from flowing 1 ppm.

Fig.1.29. LMC for PMPA obtained by A) static (10 ppm, chem.L3_PDMS), and B) SPE (10 ppb, chem.L2). Conditions: as in Fig.1.2. NOTE: No SERS from flowing 1 ppm.

A B

A B


18

A B C

Fig.1.30. LMC for CEES obtained by A) static (1000 ppb, chem.L4), B) flow (50 ppb, chem.L4), and C) SPE (10 ppb, chem.L1). Conditions: as in Fig.1.2.

Fig.1.31. LMC for CEMS obtained by A) static (1000 ppb, chem.L4), B) flow (500 ppb, chem.L3), and C) SPE (10 ppb, chem.L2). Conditions: as in Fig.1.2.

A B C

A B C

Fig.1.32. LMC for TDG obtained by A) static (10 ppm, chem.L1), B) flow (1 ppm, chem.L1), and C) SPE (10 ppb, chem.L1_PEG). Conditions: as in Fig.1.2.

Fig.1.33. LMC for HEES obtained by A) static (100 ppm, chem.L3), B) flow (1 ppm, chem.L2), and C) SPE (10 ppb, chem.L6). Conditions: as in Fig.1.2.

A B C

A B C

Fig.1.34. LMC for CP obtained by A) static (10 ppm, chem.L3), B) flow (1 ppm, chem.L3), and C) SPE (10 ppb, chem.L3). Conditions: as in Fig.1.2.

A B C

Fig.1.35. LMC for MP obtained by A) static (1 ppm, chem.L6), B) flow (1 ppm, chem.L2), and C) SPE (10 ppb, chem.L3). Conditions: as in Fig.1.2.


19

Task 1 Summary: Chem. L2, L3 and L6 are the universal chemistries for screening of the 20 target chemicals at nominal concentrations. Polymer modified chemistries provided some selectivity towards a few classes of chemicals (e.g. L1_PEG was better for pesticides and L2_PEG was better for Blister and Blood agent simulants like CEES, CEMS, TDG and CN). Although a few chemicals could be detected at 10 ppb using the static or the flow methods, none of the modifications provided this detection limit for all the 20 target chemicals. To achieve the desired sensitivity of 10 ppb, SPE or AIEX methods were necessary. SPE sorbents like C8 and C18 provided a universal pre-concentration method for the purpose of detecting the various chemical, toxic industrial and pesticide agents of concern, while the AIEX sorbents provide for the pre-concentration of nerve agent hydrolysis products. In conclusion, all of the 20 target chemicals could be detected at 10 ppb using SPE/AIEX methods with sol-gel chemistries L2 and L6.

Fig.1.36. LMC for FON obtained by A) static (1 ppm, chem.L2), B) flow (500 ppb, chem.L3), and C) SPE (10 ppb, chem.L3). Conditions: as in Fig.1.2.

A B C

Fig.1.37. LMC for DS obtained by A) static (100 ppm, chem.L6), B) flow (1 ppm, chem.L3), and C) SPE (10 ppb, chem.L2). Conditions: as in Fig.1.2.

A B C

A B C

Fig.1.38. LMC for DS-SO obtained by A) static (10 ppb, chem.L6), B) flow (1 ppm, chem.L1_PEG), and C) SPE (10 ppb, chem.L6). Conditions: as in Fig.1.2.

Fig.1.39. LMC for DMHTP obtained by A) static (1 ppm, chem.L3), and B) SPE (10 ppb, chem.L3). Conditions: as in Fig.1.2.

Fig.1.40. LMC for 3,5-DCBA obtained by A) static (100 ppm, chem.L6), B) flow (1 ppm, v), and C) SPE (10 ppb, chem.L2). Conditions: as in Fig.1.2.

A B C

A B

A B C

Fig.1.41. LMC for 4,4-DCBP obtained by A) static (10 ppb, chem.L4), B) flow (1 ppb, 4), and C) SPE (10 ppb, chem.L4). Conditions: as in Fig.1.2.


20

Task 2. Develop SER-active capillary durability. The overall objective of this task is to develop the SER-active capillary fabrication procedure so that modest flow rates, pressure, and temperature can be used for extended periods of time. This will be accomplished by investigated cure and coating procedures, and performing flow, pressure and temperature tests. As part of this task, we performed studies to evaluate the stability and shelf-life of our SER active sol-gels. The primary issues that must be addressed include determining the optimal conditions (e.g. temperature, time) for curing specific sol-gel chemistries, under what conditions can the sol-gels (e.g. in capillaries) be reduced and stored with minimal loss of performance (e.g. 3 months) and their resistance to degradation during sample flow. Preliminary results have indicated that a constant cure temperature (e.g. cure at 20-25°C for up to 24 hrs) with the sol-gel sensors properly sealed (to minimize solvent loss and subsequent drying out) is critical. We performed extensive durability studies on chem.L2, L3, and L6, the three universal chemistries determined during Task 1. A. chem.L6 Durability Studies: As described before chem.L6 is an open-tubular chemistry (OTC), where silver nanoparticles are embedded into a thin Si-alkoxide film functionalized on the internal glass capillary surface. Since the coating is thin, it was important to know if this open tubular format is durable during continuous water flow and how frequently the substrates need to be replaced. Hence, to determine the optimum fabrication conditions for manufacturing durable chem.L6 OTCs we performed the following experiments. 1. The durability of chem.L6 was demonstrated by flowing pure water for one hour (at 1 mL/min) using a

peristaltic pump through such a coated capillary. This was followed by measuring a 0.1 mg/mL sample of MPA, which exhibited no apparent decrease in the SERS signal as compared to similar pre-flow control measurements (see Figure 2.1). A similar set of experiments were performed in which HPLC water was flowed through the standard chem.L6 coated capillaries for 1 and 2 hours prior to the introduction of analyte samples (at RT, 10 mL/min). After 1 hour of flowing pure water, a 1 ppm sample of MPA in HPLC water was drawn into this capillary and measured. No MPA signal was detected (see Figure 2.2). To verify if SER-activity was completely extinguished, a 1000 ppm sample was measured on this same capillary. Again no signal was observed. Flowing water through capillaries coated with chem.L6 at 10 mL/min eliminates the SERS response.

2. We examined the shelf-life of chem.L6. Two batches of chem.L6 OTC capillaries (40 each) were made initially (1 week apart) using the standard procedure. All measurements (here and in Task 5) were performed on multiple capillaries (and 9 spots for each capillary). This allowed evaluating variations within capillaries, between capillaries, and batch-to-batch capillaries. For the shelf-life test, a 30-min reduction was applied to the sol-gel, water removed, and sealed with parafilm. Measurements were made on the first day, and then on the following 4th and 7th days. In each case, MPA at 1 ppm was used to test the response over time (see Figures 2.3-2.4). The results indicate that 1 ppm can still be detected at the 9 equidistant points along the sol-gel coated capillary after 7 days. However, the signal appears to start dropping off substantially after 2 days.

Fig.2.1. SERS of MPA at 0.1 mg/mL in HPLC water; A) sample loaded and measured (static) after flowing 50 mL of pure HPLC water at 1 mL/min, and B) sample loaded and measured (static) on a different capillary with no prior flowing of water. Conditions: OTC chem.L6 (APTMS based), 90 mW, 785 nm, 1-min.

A B

A B

Fig.2.2. SERS of MPA after flowing HPLC water at 10 mL/min for 1 hr on chem.L6A) 1 ppm and B) 1000 ppm. Conditions: 80 mW, 785 nm, 1-min.


21

3. We examined the sol-gel curing process as a function of both temperature and time in our continuing effort to

further improve the overall performance (sensitivity, reproducibility and durability) of chem.L6 capillaries. Initial test were carried out on capillaries prepared as above with the exception that after curing for 24-hrs at RT, 4 of the chem.L6 capillaries were placed in a preheated oven set at 35 °C. A single capillary was removed after heating for a period of 10, 30, 60 or 180-min, respectively, which was allowed to cool to room temperature for 45-min, and then reduced (30-min) by the standard method. In each case, MPA at 1 ppm was used to test the response. The 9-point averaged SER spectra obtained for each capillary heated for the specified time at 35 °C are presented in Figure 2.5. This experiment was repeated on a different set of 4 capillaries at 50 °C (see Figure 2.6), and again at 65 °C (see Figure 2.7). These results are summarized in Figure 2.8 where the intensity of the 759 cm-1 peak (9 points averaged) is plotted for each temperature as a function of time. Two additional experiments were carried out at 35 °C and 50 °C. In these 2 cases, the capillaries were reduced first then heated. These results are also shown in Figures 2.9-2.10. Capillaries coated with chem.L6 1) heated for 10-min at 35 °C afforded an improvement in sensitivity by a factor of ~2 times that of the standard RT base-line response of 1 ppm MPA (this was confirmed in repeat measurements), and 2) extended heating over time (greater than 10-min) or at elevated temperatures (e.g. 50 °C) diminished the SERS response. It is important to point out that degradation generated artifacts in the spectra at 65 °C, which greatly enhanced the MPA signal over time. Although improvements were observed for some higher cure temperatures and periods, the improvements were not consistent, nor were they substantially better than RT cure. Since the RT conditions gave consistent results, they were used.

Fig.2.3. SERS of MPA at 1 ppm on chem.L6, 9-points, which was parafilmed and capped for 7 days. Conditions: in water, 785 nm, 80 mw, 1-min (batch 1).

Fig.2.4. SERS of MPA at 1 ppm on chem.L6, for the averaged 9-point spectra: aged A) 1 (red), B) 4 (black), and C) 7 days (blue); 785 nm, 80 mw, 1-min (batch 1).

A B C

Fig.2.5. SERS of MPA at 1 ppm on chem.L6 (9 point average), cured at 35 oC for A) 10, B) 30, C) 60, and D) 180-min. Conditions: in HPLC water, 80 mW, 785 nm, 1-min.

A B C D


22

0

0.02

0.04

0.06

0.08

0.1

0.12

0 50 100 150 200

Peak

Hei

ght

Curing time (min)

50 C

35 C

65 C

35 C Red

50 C Red

4. In order to determine the operational range of the chem.L6 capillaries, a temperature bath was used to set the

temperature of aqueous samples of MPA, drawn through a capillary and measured. The analyte samples were measured at 20 and 40 °C (Figure 2.11). The response did not appear affected at 40 °C.



Fig.2.8. Intensity of the peak at 759 cm-1 (baseline at 720 cm-1 subtracted) of MPA at 1 ppm as a function of curing time on chem.L6 (9 point average), at different curing temperatures 35 oC, 50 oC, and 65 oC. Conditions: 80 mW, 785 nm, 1-min.

Fig.2.9. SERS of MPA at 1 ppm on chem.L6 (9 point average), reduced then cured at 35o C for A) 10 and B) 30-min. Conditions: 80 mW, 785 nm, 1-min.

Fig.2.10. SERS of MPA at 1 ppm on chem.L6 (9 point average), reduced then cured at 50o C for A) 10, B) 30, C) 60, and D) 180-min. Conditions: 80 mW, 785 nm, 1-min.

A B C D

A B C D

A B

A B C D


23

5. We continued to evaluate the shelf-life of our standard chem.L6 capillaries by investigating the optimum storage conditions. Three methods for storing these capillaries were investigated. A) Initially, a 30-min addition of Ag-colloids was applied to the sol-gel, water removed, and the capillary ends sealed with parafilm. The results indicate that 1 ppm MPA can still be detected at 9 equidistant points along the sol-gel coated capillary after 7 days (Figure 2.12), but exhibited no activity after 14 days. B) A similar experiment was also performed, but in this case water was added to the capillary prior to sealing. The activity is significantly diminished on the 4th day, and completely gone on the 7th day (Figure 2.13). C) A 3rd experiment was carried out where the APTES coating solution was removed after 24-hr, and then the capillaries sealed. After 7 days, the standard method for adding the Ag-colloids to the sol-gel coat was followed. SER-activity for MPA at 1 ppm was still observed (Figure 2.14). It is worth pointing out that the initial chem.L6 vials, which can be prepared and ready for measurement within 24-hrs, is still capable of detecting 250 ppb MPA even after 4 days of storage (see Figure 2.15).

A B

Fig.2.11. SERS of MPA (static)1 ppm in HPLC water, with sample at A) 20 °C and B) 40 °C. Conditions: 80 mW, 785 nm, 1-min.

Fig.2.12. SERS of MPA on chem.L6 at 1 ppm after A) 1, B) 4 and C) 7 days, sealed with no water following Ag-colloid addition. Conditions: HPLC water, 80 mW, 785 nm, 1-min.

Fig.2.13. SERS of MPA on chem.L6 at 1 ppm after A) 4, and B) 7 days, sealed with water following Ag-colloid addition. Conditions: HPLC water, 80 mW, 785 nm, 1-min.

Fig.2.14. SERS of MPA on chem.L6 at 1 ppm after coating step, sealed empty for 7 days, then Ag-colloids added, and tested at 9 spots. Conditions: HPLC water, 80 mW, 785 nm, 1-min.

Fig.2.15. SERS of MPA on chem.L6 coated vials at 250 ppb after 4 days, A) single spot and B) 9 points averaged. Coated with 1% APTES for 24-hr, Ag-colloids added for 1-min, sealed empty, tested 4 days later. Conditions: HPLC water, 80 mW, 785 nm, 1-min.

A B

A B

A B C


24

6. Finally, we evaluated the shelf-life of our new chem.L6 vials (Figures 2.16-2.19), which were sent to the ECBC facilities at Aberdeen for preliminary testing as part of Task 7. Our preliminary results show that if Ag-colloid solution is added immediately following the APTES coating procedure, then sealed and stored empty, MPA can be detected at 25 ppb if measured on the first day (Figure 2.17A), at 250 ppb after being stored for 1-2 days (Figure 2.17B), and at 1 ppm following 7 days of storage (see Figure 2.18). It is worth pointing out that a very weak signal was observed sporadically for MPA at 1 ppm after 10 days of storage. However, as shown by the SERS of MPA at 250 ppb (Figure 2.19B) obtained with coated vials stored empty for 7 days, adding the Ag-colloids prior to making a measurement is the best method for maintaining sensitivity.

Summary for the durability studies of chem.L6: 1. The shelf-life of chem.L6 standard capillaries was found to be 1 week, though the signal started to drop off

substantially after 2 days (Figures 2.1-2.3). 2. Capillaries coated with chem.L6 1) heated for 10-min at 35 °C afforded an improvement in sensitivity by a

factor of ~2 times that of the standard RT base-line response, and 2) extended heating over time (greater than 10-min) or at elevated temperatures (e.g. 50 °C) diminished the SERS response. Signal improvements were observed at further elevated cure temperatures (e.g. 65°C) and extended heating (greater than 60-min) but it also produced degradation (of sol-gel) generated artifacts in the spectra (Figures 2.5-2.8). Although improvements were observed for some higher cure temperatures and periods, the improvements were not consistent, nor were they substantially better than RT cure. Since the RT conditions gave consistent results, they were used.

3. Flowing water through capillaries coated with chem.L6 at 1 mL/min did not affect the SERS response, whereas at 10 mL/min the response was eliminated (Figures 2.1 and 2.2). However none of these sol-gels became

A B

Fig.2.16. Images of A) standard chem.L1 coated vials (left: Ag-TMOS sol-gel unreduced, right: after reduction) as compared to B) chem.L6 coated vials (left: after APTES coating, right: after silver colloids added).

Fig.2.19. SERS of MPA on chem.L6 coated vials at A) 1 ppm and B) 250 ppb; after 1-min spin coating step (1400 rpm), sealed empty for 7 days, then Ag-colloids added for 1-min (spin at 1400 rpm). Conditions: as in Fig.2.17. Note 889 and 1418 cm-1 peaks indicate degradation of the APTES/colloid.

Fig.2.18. SERS of MPA on chem.L6 coated vials, 1 ppm (on 2 spots); after spin coated with 1% APTES for 1-min (1400 rpm), then Ag-colloids added for 1-min (spin at 1400 rpm), sealed empty for 7 days, and then measured. Conditions: as in Fig.2.17.

Fig.2.17. SERS of MPA on chem.L6 coated vials at A) 25 ppb 1st day and B) 250 ppb 2nd day; after spin coated with 1% APTES for 1-min (1400 rpm), then Ag-colloids added for 1-min (spinning at 1400 rpm), sealed empty. Conditions: HPLC water, 80 mW, 785 nm, 1-min.

A B

A B


25

detached. 4. Storing the chem.L6 capillaries without any solvent and sealed with parafilm gave better shelf life 7 days

compared to those stored in water (signal seen after 4 days no signal seen on the 7th day- Figures 2.12-2.13). 5. Initial results suggest that chem.L6 vials, which can be prepared and ready for measurement within 24-hrs, is

capable of detecting 250 ppb MPA even after 4 days of storage (see Figure 2.15). 6. For chem.L6 vials, our preliminary results show that if a Ag-colloid solution is added immediately following

the APTES coating procedure, then sealed and stored empty, MPA can be detected at 25 ppb if measured on the first day (Figure 2.17A), at 250 ppb after being stored for 1-2 days (Figure 2.17B), and at 1 ppm following 7 days of storage (see Figure 2.18). It is worth pointing out that a very weak signal was observed sporadically for MPA at 1 ppm after 10 days of storage. However, as shown by the SERS of MPA at 250 ppb (Figures 2.14 and 2.19B) obtained with chem.L6 coated capillaries and vials stored empty, the best way to prolong the life of chem.L6 substrates is to store the APTEOS functionalized substrates (capillaries and vials) sealed and add silver colloid prior to making a measurement.

B. chem.L3 Durability Studies: 1. We examined the sol-gel curing process as a function of both temperature and time in our continuing efforts to

improve the overall performance (sensitivity, reproducibility and durability) of sol-gel plugs in “packed” capillaries filled with chem.L3 and its variations of (e.g. chem.L32). We prepared chem.L3 and chem.L32 filled capillaries, allowed them to gel over night (24-hrs at RT), and then placed them into a pre-heated oven set initially at 40 °C for 10-min. Following this, the capillaries were taken out of the oven, allowed to equilibrate to RT for ~30-min, then reduced by the standard method, and tested for SER-activity (static) with FON at 1 ppm in HPLC water. In both instances, no SER-activity was observed (see Figure 2.20).

2. In a second set of experiments, we prepared chem.L3 and chem.L32 filled capillaries, and immediately placed part of them into a pre-heated oven set initially at 26 °C and the rest in a refrigerator, for overnight curing. Following this, the two different sets of capillaries were taken out, allowed to equilibrate to RT for ~30-min, then reduced by the standard method, and tested for SER-activity (static) with FON at 1 ppm in HPLC water. For each case, SER-activity was observed (see Figure 2.21). An important observation was that all of the sol-gels cured at 26 °C formed “half-filled” capillaries similar to that of chem.2c. Such a capillary format is generally more reproducible than the standard filled format, but unfortunately is not as amenable to analyte pre-concentration by the continuous flow method.

3. To investigate the durability of the capillaries, experiments were also performed in which HPLC water (at RT)

was flowed continuously at a rate of 5 mL/min using a peristaltic pump through a standard series of reduced chem2d capillaries for 1, 2 and 5 hours prior to the introduction of an analyte sample (see Figure 2.22). The 3 point averaged “static” spectra of FON (10 ppm in HPLC water) following flow of water for 2-hr (Fig.2.22A) and 5-hr (Fig.2.22B) are nearly identical.

4. A similar experiment was then carried out on a second set of chem.L3 capillaries, where ordinary tap water was continuously flowed through the reduced sol-gels (at RT, and 5 mL/min) for 24-hrs. None of these sol-gels became detached or appeared bleached out. The subsequent addition of a 10 ppm FON sample revealed that SER-activity was retained after 24-hrs of flowing just water (Fig.2.22C), but the signal was relatively weak.

A B

Fig.2.20. SERS of FON, on A) chem.L3, and B) chem.L32 after curing 24-hrs at RT, heated at 40 °C for 10-min, equilibrated, reduced, and tested. Conditions: 1 ppm in HPLC water, 80 mW, 785 nm, 1-min.

Fig.2.21. SERS of FON, on chem.L3 initially cured overnight in A) oven at 26 °C, and B) refrigerator at 4 °C. Conditions: 1 ppm in HPLC water, 80 mW, 785 nm, 1-min (see Fig.2.20).

A B


26

However, we believe an IR resin (as described in Task 4) can be used to remove ions in tap water that over time may have accumulated on the silver surface, and diminished the SERS response.

Summary for the durability studies of chem.L3: 1. For chem.L3 high temperature (40 °C) curing eliminated the SER-activity. The RT curing gave better

performance than curing at 4 °C. (Figures 2.20 and 2.21). 2. Flowing HPLC water did not diminish the SER activity. The SERS responses of 10 ppm FON samples

following flow of HPLC water for 2-hr (Fig.2.22A) and 5-hr (Fig.2.22B) are nearly identical. 3. Flowing tap water diminished the SER activity after prolonged flow. However none of these sol-gels became

detached or appeared bleached out. The subsequent addition of a 10 ppm FON sample revealed that SER-activity was retained after 24-hrs of flowing tap water at 5 mL/min (Fig.2.22C), but the signal was relatively weak. However, we believe an IR resin (as described in Task 4) can be used to remove ions in tap water that over time may have accumulated on the silver surface, and diminished the SERS response.

C. chem.L2 Durability Studies: 1. Flow and durability experiments were also performed on chem.L2. In these experiments ordinary tap water was

flowed through the SER-active capillaries for 24 hours at a rate of 5 mL/min at room temperature using a peristaltic pump. After this point 50 mL of 10-100 microg/L (ppb) aqueous samples were flowed through the capillaries at a rate of 1mL/min and the SER spectra recorded using a program (Raster) that can continuously collects spectra while moving along a 1cm length of the capillary as opposite averaging 9 discrete points (Figures 2.23-2.27). None of these sol-gels became detached or appeared bleached out. CEES, MPA, FON, and Sunset Yellow could be successfully detected at 100 microg/L (100 ppb), whereas HEES was detected at 1000 microg/L (1000 ppb). However, no SER-activity was observed for TDG, CEMS, CMPA, IMPA, PMPA and EMPA at 1000 microg/L.

Fig.2.22. SERS of FON (static) measured after flowing pure HPLC water at 5 mL/min on chem.L3 for A) 2 and B) 5-hr (3 points averaged); C) after flowing tap water at 5 mL/min for 24-hr. Conditions: 10 ppm in HPLC water, 80 mW, 785 nm, 1-min.

A B C

A B

A B

Fig.2.23. SERS of CEES, A) 100 ppm (as reference) and B) 100 ppb after flowing tap water at 5 mL/min for 24-hr. Conditions: 80 mW of 785 nm, 1-min. chem.L2

Fig.2.24. SERS of MPA, A) 100 ppm (as reference) and B) 100 ppb after flowing tap water at 5 mL/min for 24-hr. Conditions: as in Fig.2.23.except 4-min Raster .


27

2. Flow experiments were continued with ordinary tap water as a part of the proposed durability studies. An IR

resin (immobilized in a glass capillary using blank MTMS sol-gel plugs as frits and placed in front of the SER-active capillaries) was incorporated to eliminate the small inorganic ions present in tap water possibly responsible for weak SERS response and the rising background.

3. After this, ordinary tap water flowed through IR resin and into the SER-active capillaries then 50 mL of 10-100 microg/L (ppb) aqueous samples were flowed through the capillaries at a rate of 1mL/min and the SER spectra recorded (Figures 2.28-2.32). None of these sol-gels became detached or appeared bleached out. CP, MP, CSPS, DS, and DS-SO could be successfully detected at 1000 microg/L (1 ppm). However, no SER-activity was observed for DMHTP and DIASH at 1000 microg/L (lowest detection limits to be determined). This completes the flow experiments for the 20 primary chemicals in this program.

4. Shelf life: Since our commercially available vials (Simple SERS Sample Vials) have a long shelf life (4-6 weeks) which is attributed to a good seal, we investigated various methods for standardizing procedures and automating the fabrication of our capillary sol-gel sensors. Particularly, the use of the butane torch (Idaho Products) to melt and completely seal the ends of the sol-gel coated or filled glass capillaries was investigated. For this, the capillaries were prepared, cured for 24 hrs, reduced with 1mg/mL sodium borohydride and loaded with water. Then they were sealed with a butane torch and stored at RT till further use. Figure 2.33 show the SERS response of 50 ppm Bacillus cereus spores on chem.L2 in acetic acid at week 1 and at week 5. The data clearly indicate that the shelf life of sol-gel capillaries can be significantly improved by melting and sealing the ends of the sol-gel glass capillaries versus capillaries stored with rubber tips.

A B

A B

Fig.2.25. SERS of FON, A) 100 ppm (as reference) and B) 100 ppb after flowing tap water at 10 mL/min for 24-hr. Conditions: as in Fig.2.23.

Fig.2.26. SERS of Sunset Yellow, A) 1ppm (as reference) and B) 100 ppb after flowing tap water at 5 mL/min for 24-hr. Conditions: as in Fig.2.23.

A B

Fig.2.27. SERS of HEES, A) 100 ppm and B) 1000 ppb after flowing tap water at 5 mL/min for 24-hr. Conditions: as in Fig.2.23.


28

Summary for the durability studies of chem.L2: 1. Flowing tap water through the SER-active capillaries for 24 hours at a rate of 5 mL/min at room temperature

did not affect the sol-gel stability, though the activity was slightly diminished after tap water flowing. None of these sol-gels became detached or appeared bleached out. CEES, MPA, FON, and Sunset Yellow could be successfully detected at 100 microg/L (100 ppb), whereas HEES was detected at 1000 microg/L (1000 ppb). However, no SER-activity was observed for TDG, CEMS, CMPA, IMPA, PMPA and EMPA at 1000 microg/L. (Figures 2.23-2.27).

2. The rising background and weak SER response from the ions in the tap water were successfully overcome by flowing the tap water through a IR resin prior to loading aqueous samples in the SER-active capillaries. CP,

A B

A B

Fig.2.28. SERS of CP, A) 1000 ppm (as reference) and B) 1 ppm after flowing IR treated tap water at 5 mL/min for 24-hr. Conditions: 80 mW of 785 nm, 1-min. chem.L2

Fig.2.29. SERS of MP, A) 1000 ppm (as reference) and B) 1 ppm after flowing IR treated tap water at 5 mL/min for 24-hr. Conditions: as in Figure 2.28.

A B

A B

Fig.2.30. SERS of CSPS, A) 100 ppm (as reference) and B) 1 ppm after flowing IR treated tap water at 10 mL/min for 24-hr. Conditions: as in Figure 2.28.

Fig.2.31. SERS of DS, A) 1000ppm (as reference) and B) 1 ppm after flowing IR treated tap water at 5 mL/min for 24-hr. Conditions: as in Figure 2.28.

Fig.2.32. SERS of DS-SO, A) 1000 ppm (as reference) and B) 1 ppm after flowing IR treated tap water at 5 mL/min for 24-hr. Conditions: as in Figure 2.28.

A B

Fig.2.33. SERS of 50 ppm Bacillus cereus spores, A) at week 1 and B) at week 5 on chem.L2, sealed with a butane torch. Conditions: as in Figure 2.28, same scale, but offset.

A B


29

MP, CSPS, DS, and DS-SO could be successfully detected at 1000 microg/L (1 ppm). However, no SER-activity was observed for DMHTP and DIASH at 1000 microg/L (lowest detection limits to be determined). (Figures 2.28-2.32).

3. The shelf-lives of sol-gel capillaries can be significantly improved (more than a month) by melting and sealing the ends of the sol-gel glass capillaries versus capillaries stored with rubber tips.

E. Temperature stability study for chem.L4: 1. Initial temperature study carried out on chem4b cured for 24-hrs (sealed) at 22°C, and then heated at 37°C for

2-hrs prior to being reduced (Figure 2.34). CEMS was used to test these heat-treated gold capillaries. It does not appear that the SERS-response is affected at this temperature and time.

F. Flow studies to re-evaluate preconcentration: 1. As part of this task, a larger sample volume was used to further test the continuous flow preconcentration

method. 50 mL was used as opposed to 10 mL in an effort to improve the sensitivity of this pre-concentration method. Unfortunately, the majority of the target analytes were not amenable to pre-concentration and detection at the required ppb levels with flowing. For example, 50 mL of CEMS and HEES in water at 500 ppb were flowed on chem.L3 (HCl washed) at 1 mL/min, but again as with previous experiments using 10 mL samples, the analytes were not detected. Flowing 50 mL of 10 ppb samples of DIASH and DMHTP on chem.L3 and chem.L2 also produced negative results (data not shown). As we have shown in Task 1, the alternative pre-concentration methods using IEX and SPE resins are clearly superior to the continuous flow method, in that all 20 primary targets proposed can be extracted from water and detected at 10 ppb.

Overall summary for the durability of sol-gel capillaries: 1. The best flow rate was determined to be 1 mL/min that can be universally applied for an extended period for all

the above three chemistries (chem.L2, chem.L3, and chem.L6) without any sol-gel degradation and reduction of the SERS activity.

2. Pre-treating tap water with ion retardation resin eliminated interfering ions. This improved the SERS response and retention of SERS activity of the sol-gel capillaries even after water flow for an extended period.

3. RT curing gave the most consistent and reproducible response, however the sol-gel coating was found to be SER-active (e.g. chem.L6) and stable even at elevated temperatures (none of these sol-gels became detached or appeared bleached out).

4. Good sealing (melting and sealing the ends of the glass capillaries) was identified as a critical manufacturing step to prolong the shelf-life.

Figure 2.34. SERS of CEMS on chem.L4 A) heated in oven for 2-hrs at 37°C and B) control at RT, following standard 24-hr cure at 22°C. Conditions: 10,000 ppm in MeOH, 80 mW, 785 nm, 20-sec.

A B


30

Task 3. Develop Spectral Library. The overall objective of this task is to build a spectral database such that chemical agents, their hydrolysis products, pesticides, and the highest priority toxic industrial chemicals can be rapidly identified by spectral matching or functional group analysis. Although this program will measure several chemical agents and a number of agent hydrolysis products, there exists numerous industrial toxic chemicals, and a practical or near term approach to spectral matching will be based on a limited library of those of greatest concern. The spectral library is comprised of a total of 96 chemicals from three major groups, 1) the proposed Primary Chemicals (20), Secondary Chemicals (30), and Additional Chemicals (46). All chemicals were obtained from Sigma Aldrich or Cerilliant. The latter chemicals were added to expand the capabilities of the chemical detection and identification software (spectral search and match) to include additional potential poisons and their degradation products that may be used in a terrorist attack. The larger library also served to better test and optimize the spectral search and match algorithms. Each group is described below. Primary Chemicals (20): CN (cyanide, the hydrolysis product of KCN or NaCN), IMPA (isopropyl methylphosphonic acid, the hydrolysis product of GB), PMPA (pinacolyl methylphosphonic acid, the hydrolysis product of GD), EMPA (ethyl methylphosphonic acid, the hydrolysis product of VX), CMPA (cyclo methylphosphonic acid, the hydrolysis product of GF), MPA (methylphosphonic acid, final nerve agent hydrolysis product), DIASH (di-isopropyl aminothiol, a hydrolysis product of VX), TDG (thiodiglycol, the hydrolysis product of HD), CEES (2-chloroethyl ethylsulfide, simulant of HD), CEMS (2-chloroethyl methylsulfide, simulant of HD), CSPS (cysteamine S-phosphate as sodium salt, simulant of VX), HEES (2-hydroxyethyl ethylsulfide, the hydrolysis product of CEES), CP (chlorpyrifos, organophosphate (OP) pesticide), FON (fonofos, OP pesticide), MP (methylparathion, OP pesticide), DS (disulfoton, OP pesticide), DS-SO (disulfoton sulfoxide, the metabolite product of DS), DMHTP (potassium salt of O,O-dimethyl hydrogen thiophosphate, an OP oxon model), 35DCBA (3,5-dichlorobenzoic acid, a chlorinated toxic industrial chemical (TIC)), and 44DCBP (polychlorobiphenyl 4,4-dichloro biphenyl, a chlorinated TIC). Secondary Chemicals (30): HD (bis(2-chloroethyl)sulfide, a blister agent), VX (O-ethyl S-[2-(diisopropylamino)ethyl] methylphosphonothioate, a nerve agent), 14DT (1,4-dithiane, a degradation product of HD), EtSH (ethane thiol, a degradation product of CEES), TCP (trichloropyridinol, a hydrolysis product of CP), DMHP (dimethyl hydrogen phosphate, the final OP degradation product), DIAZ (diazinon, an OP pesticide), AZ (atrazine, a pesticide), ALD (aldicarb, a pesticide), CAR (carbaryl, a pesticide), CD (chlordane, a pesticide), DAL (dalapon, a pesticide), DC (dicamba, a pesticide), DQT (diquat, a pesticide), ENDT (endothall, a pesticide), GP (glyphosate, a pesticide), OX (oxamyl, a pesticide), SIM (simazine, a pesticide), TBC (thiobencarb, a pesticide), TOX (toxaphene, a pesticide), 24DNT (2,4 dinitrotoluene, a herbicide simulant of DCNT class, also an explosive precursor), BCAA (bromochloroacetic acid, a chlorinated TIC), CB (chlorobenzene, a chlorinated TIC), CPIC (chloropicrin, a chlorinated TIC), 2CT (2-chlorotoluene, a chlorinated TIC), 14DCB (1,4 dichlorobenzene, a chlorinated TIC), 135TCB (1,3,5 trichlorobenzene, a chlorinated TIC), PIC (picloram, a chlorinated TIC), DCAA (dichloroacetic acid, a chlorinated TIC), and DCAN (dichloroacetonitrile, a chlorinated TIC). Additional Chemicals (46): EA2192 (diisopropylaminoethyl methylphosphonothioic acid, a lethal hydrolysis product of VX ), EDMAPA (ethyldimethyl amidophosphoric acid, hydrolysis product of GA), IBMPA (isobutyl methylphosphonic acid, hydrolysis product of Russian-VX), DMMP (dimethyl methylphosphonate, simple nerve agent simulant), H3PO4 (phosphoric acid, simple phosphate model), amifostin (EA2192 simulant), cysteamine (degradation product of CSPS), DEICNMP (Diethylisocyanomethylphosphonate, a GA simulant), HOEtSH (2-hydroxyethane thiol, a vesicant degradation product), EES (ethylethyl sulfide) and MMS (methylmethyl sulfide) as simple aliphatic sulfide models, 3ClPrSH (3-chloropropane thiol, relevant to the important but unobtainable 2-chloroethane thiol degradation product of HD), 2CEPhS (2-chloroethyl phenyl sulfide, relevant to HD, CEES and CEMS vesicant series), ESSE (ethylethyl disulfide, simple aliphatic disulfide model), TDG-SO (oxidation product of TDG); we also measured simple models of pesticides and degradation products such as DEHDTP (sodium salt of diethyl hydrogen dithiophosphate), DEHTP (potassium salt of diethyl hydrogen thiophosphate), DEHP (diethyl hydrogen phosphate, a final OP degradation product), DMHDTP (O,O-dimethyl hydrogen dithiophosphate, note methyl analogs of DEHTP and DEHP (i.e. DMHTP and DMHP) are a primary and secondary target, respectively), EHEP (ethyl hydrogen ethylphosphonate, final oxidative and hydrolysis product common to FON type OP pesticides), actual OP degradation products including PhSH (thiophenol, degradation product of FON and 2CEPhS), 4NOPh (4-nitrophenol, degradation product of MP), CPO (chlorpyrifos oxon), MPO (methylparathion oxon), DEM-S (demeton-S, oxon of DS) and EPO (ethylparathion oxon, ethyl analog of MPO), plus other additional OP


31

pesticides related to CP type CPM (chlorpyrifosmethyl, methyl analog of CP), EP (ethylparathion, ethyl analog of MP), PIRI (pirimifos methyl) and DEM-O (demeton-O), methyl analogs of the DS type PMT (phosmet), MAL (malathion) and DIM (dimethoate), other additional non-OP pesticides such as TBZ (thiobendazole) and those chlorinated PERM (permithrin), DDT (dichlorodiphenyltrichloroethane), ALA (alachlor), PQT (paraquat), ENDO (endosulfan), ENDO-S (endosulfan sulfate, metabolite of ENDO), and finally additional simple TIC models AA (acetic acid) and TFA (trifluoroacetic acid), and MELA (melamine, toxic contaminant). 2CAcPh-on and 2CPhMeS-on (2-chloroacetophenone of the incapacitating chemical agent class and 2-chlorophenylmethyl sulfone relevant as an oxidation product). DPA (dipicolinic acid, released from anthrax spores when in water). Tryp and Phe (tryptophan and phenylalanine, potential biological interferents). Although not included in spectral library, we also measured a series of dyes to determine the best for potential field tests during Task 7. R6G (rhodamine6G), CV (crystal violet), sunset yellow, allura red, tartrazine, patent blue, and erythrosineB. In addition, the standard test chemicals used in Phase I to evaluate sol-gel libraries are included in this report for completeness; AN (aniline), BA (benzoic acid), PABA (p-aminobenzoic acid), PA (phenylacetylene), PYR (pyridine) and KSCN (potassium thiocyanate). All of the proposed and additional chemicals measured in this program are listed in Table 3.1. Table 3.1. All chemicals measured (see above text for abbreviations).

Agents & Simulants

Degradation products Secondary and Additional chemicals Other Chemical Classes

CWAs Tabun (GA) EDMAPA DEICNMP’ Incapacitating agents’ Sarin (GB) IMPA DMMP 2CAcPh-on’, 2CPhMeS-on’ Soman (GD) PMPA Cyclohexyl Sarin (GF) CMPA H3PO4 Biologicals RVX IBMPA DPA Trp Phe VXL EA2192 DIASH Amifostin, EMPA/ MPA Mustard (HD)L TDG/1,4-DTL TDG-SO 2CEPhS CEES (1/2mustard) HEES / EtSHL HOEtSH 3ClPrSH CEMS (HD simulant) EES MMS’ ESSE CSPS (VX simulant) cysteamine CN as CN- from NaCN HCN KSCN Dyes Pesticides sunset yellow chlorpyrifos (CP) TCPL CPO, DEHP DIAZL GPL24DNTL CARBL AZL SIM L

DIQL OXL DALL TOX L DICL ALDIL

CLORL ENDTL TBCL TBZ PQT PERM DDT ALA ENDO ENDO-S

allura red tartrazine, patent blue, erythrosinB R6G CV

fonofos (FON) EEPA EHEPA, HSPh MAL DIM PMT CPM PIRI DEM-O’ methylparathion (MP) DMTPA DMHPL

4NOPh, MPO, EPO DMHTP EP

Std test chemicals AN BA PABA PA PYR

disulfoton (DS) DS-SO DEM-S DEHDTP DEHTP DMHDTP TICS 35DCBA 44DCBP

BCAAL DCAA L CBL DCBL TCBL CTL CPICL PICL DCAN L AA TFA MELA

Italics indicate chemicals not in SERS search library; of these GA, GB, GD, GF, HCN, EEPA and DMTPA not yet measured. Note, bold = 20 primary and L= 30 secondary chemicals proposed, all others 41 + 5’ = 46 additional chemicals measured, to give 96 component SERS library. Also note that 2 additional classes were measured, explosives and polyaromatic hydrocarbons, but since less relevant, spectra are not shown. . The surface enhanced Raman spectra for all of the chemicals are arranged according to their chemical class with their degradation products (e.g. hydrolysis). The degradation pathway (if appropriate) is given for each class showing the associated products, followed by the spectra. In most cases the Raman spectra are also included. However, many of the more poisonous chemicals were only available as forensic samples (1 mg/mL), which is too low a concentration to produce a normal Raman spectrum. Finally, the normal and SER spectra of similar chemicals, such as the hydrolysis products of nerve agents, are grouped together in the figures to highlight similarities and differences, which may affect the spectral search and match algorithms.


32

The hydrolysis pathways, NR and SERS of the 3 broad classes of CWAs are shown in Figures 3.1 -3.23. The hydrolysis of hydrogen cyanide, the simplest hydrolysis pathway is shown in Figure 3.1, followed by the normal and SER spectra of this blood agent in Figure 3.2. The hydrolysis pathways for VX and the G-series of nerve agents are presented in Figure 3.3.

Class 1. Blood Agent, CN & Hydrolysis Product

HCN

H2OH3O+ + CN-

Fig.3.2. NaCN A) NR, and SERS B) gold capillary chem.4b (TMOS); C) silver capillary chem.L3 (ODS/MTMS based), Conditions: SERS; 1mg/mL in water, 100 mW of 785 nm, 1-min; NR; solid, 300 mW of 785 nm, 5-min. NOTE: all spectra background subtracted, and solvent contributions removed.

A B C

Class 2. CWA - Nerve Agents & Hydrolysis Products

OHP

O

OH

SP

O

OH

N

SHN

OHP

O

O

SP

O

OH

OHN

SP

O

O

SP

O

O

NVX

ethanolH2O

MPA

+H2O DIASH

H2O EtOH MPA+

EA2192

+DIASHH2O

EMPA70%

20%

+

ethanolH2O

MPT

+H2O DIAOH EMPT10%

+

PO

O N

PO

O N

PO

OH N

C N

GA

OH

HCN +H2O H2Oethanol +

OHEDMPAA

DMPAA

PO

O

PO

O

PO

O

PO

O

PO

O

PO

O

F

F

F

GB

GD

GF

OH

OH

HF

HF

HF

2-propanol

2-pinacolyl

cyclohexanol

MPA

MPA

MPA

IMPA

PMPA

CMPA

H2O

H2O

H2O

+

+

+

+

+

+

OH

H2O

H2O

H2O

A. CWNA V-Series

B. CWNA G-Series

Fig.3.3. Hydrolysis degradation pathways for nerve agents A) V-series and B) G-series.

Fig.3.1. Hydrolysis degradation pathways for blood agents.


33

Figure 3.4 shows the NR and SERS of VX and one of its hydrolysis products EA2192. These are shown in a separate figure, as the spectra are quite different than the phosphonic acid hydrolysis products (shown below). CSPS, a VX simulant is shown in Figure 3.5, along with it’s potential degradation product cysteamine, included for comparison. Figures 3.6 and 3.7 show the NR and corresponding SERS of the primary nerve agent hydrolysis products MPA, IMPA, PMPA, CMPA, EMPA and DIASH. The ability to distinguish these chemicals (excluding

DIASH) is illustrated in Figure 3.8. The figure includes the sodium salt of ethyl hydrogen dimethylamidophosphate (EDMAP), the primary hydrolysis product of the nerve agent Tabun (GA), and isobutyl methylphosphonic acid (IBMPA), which is the primary hydrolysis product of the Russian variant of VX (designated RVX or VR). It is worth pointing out that the final hydrolysis product of GA is dimethylamidophosphonic acid (DMAPA), which is uniquely different than methylphosphonic acid (MPA), the final hydrolysis product of the other nerve agents listed above. Unfortunately, DMAPA is currently not available. Note that both IBMPA and EDMAPA were obtained as forensic samples (at 1mg/mL from Cerilliant), and the NR was not measured. It is clear (Figure 3.8) that the hydrolysis products can easily be identified as a class by peaks in the 700 cm-1 region, but quantifying each in a water mixture that also contains the parent nerve agent and MPA may require chemometrics.

Fig.3.7. SERS of A) MPA, B) IMPA, C) PMPA, D) CMPA, E) EMPA and F) DIASH. Conditions: 1mg/mL in MeOH on chem.L3 (ODS/MTMS based), 100 mW at 785 nm, 1-min

Fig.3.6. NR of A) MPA, B) IMPA, C) PMPA, D) CMPA, E) EMPA and F) DIASH. Conditions: all neat, except MPA and DIASH in saturated water solution, CMPA as solid, 300 mW at 785 nm, 5-min.

A B C D E F

A B C D E F

A B C D

Fig.3.4. A) NR, B) SERS of EA2192, C) NR, D) SERS of VX, Conditions: 1mg/mL in water on silver TMOS vials, 100 mW at 785 nm, 1-min. Note: obtained with RTA’s portable 785 nm analyzer, and 30 ft fiber optic cables, at ECBC (2003).

Fig.3.5. CSPS A) NR solid, B) NR sat water solution and C) SERS; cysteamine D) NR and E) SERS; Conditions: chem.L1_PEG, 1mg/mL in MeOH , 30 mW, 785 nm, 1-min; NR 300 mW at 785 nm, 5-min.

A B C D E


34

Fig.3.8. Surface-enhanced Raman spectra of the primary hydrolysis products of nerve agents with molecular structures. A) methylphosphonic acid (MPA), B) isopropyl methylphosphonic acid (IMPA), C) pinacolyl methylphosphonic acid (PMPA), D) cyclohexyl methylphosphonic acid (CMPA), E) ethyldimethyl amidophosphoric acid (EDMAPA), F) ethyl methylphosphonic acid (EMPA), and G) isobutyl methylphosphonic acid (IBMPA). Conditions: all spectra collected at 100 ppm or lower, using silver-doped sol-gel filled 1 mm glass capillaries, 75 mW of 785 nm laser excitation, 1 minute (or less, one position) acquisition. Grey line, included for comparison, is at 750 cm-1. As shown below in Figures 3.9-3.11, we have expanded our spectral library to include 4 additional secondary chemicals; amifostine (a simulant of EA2192), phosphoric acid and dimethylmethylphosphonate (as simple phosphonate models), and diethylisocyanomethylphosphonate (a simulant of GA). Although the former two chemicals produce intense peaks at ~700 cm-1, they are sufficiently removed from the phosphonic acid peaks at ~750 cm-1, and should easily be differentiated by the search and match algorithms.

A) MPA B) IMPA C) PMPA D) CMPA E) EDMAPA F) EMPA G) IBMPA

Ram

an In

tens

ity (R

elat

ive)

600 800 1000 1200 1400 Wavenumber (cm-1)


35

The hydrolysis pathway for HD, the mustard gas chemical warfare agent is shown in Figure 3.12. Figure 3.13 presents the NR and SERS of HD. Figure 3.14 and Figure 3.15 compare the NR and SERS, respectively, of HD Figure 3.12. The hydrolysis pathways for HD, A) the primary hydrolysis products (and oxidation), B) the corresponding hydrolysis mechanism, and C) secondary hydrolysis products: 1) sulfonium salts produced in reactions I and II with TDG, and 2) other possible byproducts. Also note that hydrolysis of CH (2-chlorohydroxy ethylethyl sulfide) to TDG is faster than HD to CH, and that thiolates (e.g. formed from C-S cleavage of mono-sulfides on metal surface via photolysis) can also oxidize to di-sulfides.

Class 3. CWA - Blister Agent, HD & Hydrolysis Products

SCl Cl

SOH OH

SOH OH

HD TDG TDG-SOO

H2O[Ox]

2 HCl +

SOH OH

O

O

S

S

S

O

SOH

SOH

(HOEtS)2 TDG-SO2

1,4-dithiane 1,4-thioxane

SCl S+

OH OH

SOH S+

OH OH

SS+ S+

OH OH

OH

OH

H-TDG CH-TDG

H-2TDG

SCl OH

SOH

SOH OH

SCl Cl

SCl

SCl OH

H2O HClH2O TDGCl -

+

HDH2O H2O

Cl -+

+

+

+

+

HCl CH

I

II

A B C

(1)

(2)

Fig.3.9. Amifostine A) NR powder, 188 mW, 785 nm, 5-min; and B) SERS 1000 ppm in water on chem.L3, at 80 mW, 785 nm, 1-min.

Fig.3.10. H3PO4 A) NR powder, 188 mW, 785 nm, 5-min and B) SERS in water; Dimethyl methylphosphonate (DMMP) C) NR neat, 300 mW, 785 nm, 25-min and D) SERS in MeOH. Conditions: 1000 ppm, chem.L3, 80 mW, 785 nm, 1-min.

A B

A B C D

SP

O

OHOH

N NH2

OP

O

O

OHP

O

OHOH

Fig.3.11. Diethylisocyanomethylphosphonate A) NR neat, 200 mW, 785 nm, 5-min; SERS B) 1 mg/mL in MeOH, chem.L2, 80 mW, 785 nm, 1-min.

A B


36

and the related series of aliphatic monosulfides CEES, TDG, HEES and EES. The NR and SERS of relevant thiols (as potential degradation products of this monosulfide series) is shown in Figure 3.16 and Figure 3.17. Other chemicals relevant to the HD vesicant agent class are provided in Figures 3.18 – 3.23. This includes two other chlorinated vesicants, CEMS (Figure 3.18) and 2CEPhS (Figure 3.19), PhSH as a potential thiol degradation product of 2CEPhS (also shown in Figure 3.19), TDG-SO as an oxidative byproduct of TDG (Figure 3.20), and 14DT as a possible secondary hydrolysis product of HD (Figure 3.21). Figure 3.22 and 3.23 show MMS, the simplest aliphatic monosulfide, and ESSE, a simple disulfide, respectively, that can result from oxidation of the parent sulfide.

A B C D E

Fig.3.14. NR of the primary aliphatic mono-sulfide series; A) HD, B) CEES, C) TDG, D) HEES, and E) EES. Conditions: spectra measured as pure neat liquids in glass capillaries, at 8 cm-1 resolution, with 300 mW of 785 nm, and acquisition for 5-min (150 scans). HD at 0.4 cm-1 resolution, 100 mW of 785 nm, 1 min.

Fig.3.16. NR of relevant thiols; A) 3ClPrSH, B) HOEtSH and C) EtSH. Conditions: neat liquids in glass capillaries, at 8 cm-1 resolution, 300 mW of 785 nm, 5-min (150 scans).

A B C

Fig.3.15. SERS of sulfides (see Fig.3.14 for corresponding NR); A) HD, B) CEES, C) TDG, D) HEES and E) EES. Conditions: 1% v/v in MeOH with chem.L3 (ODS/MTMS based capillary), 8 cm-1 resolution, 100 mW of 785 nm, 1-min (30 scans). HD measured with std silver-doped TMOS vials, 1 mg/mL in MeOH (0.1% v/v), at 2 cm-1 resolution, 100 mW of 785 nm, 1 min.

A B C D E

A B C

Fig.3.17. SERS of thiols (see Fig.3.16 for corresponding NR); A) 3ClPrSH, B) HOEtSH and C) EtSH. Conditions: 1% v/v in MeOH with chem.L3 (ODS/MTMS based capillary), at 8 cm-1 resolution, 100 mW of 785 nm, 1-min (30 scans).

Fig.3.13. HD A) NR, neat liquid, 100 mW of 785 nm, 1-min; and SERS B) gold-doped and C) silver-doped TMOS vials, in isopropanol/water 1mg/mL, 100mW of 785nm, 16 sec. Note: collected at ECBC (2003).

A B C


37

Fig.3.22. SERS of MMS at A) 100 mW and B) 50 mW, 1% v/v in MeOH, chem.L3 (ODS/MTMS based capillary), 785 nm, 1-min (224 scans). C) NR of MMS, neat liquid in glass vial, 260 mW of 785 nm, 5-min (1124 scans).

A B C

Fig.3.23. ESSE A) NR neat liquid in glass vial, 300 mW of 785 nm, 5-min (150 scans) and B) SERS 1% v/v in MeOH on chem.L1_PEG, 80 mW, 785 nm, 1-min (30 scans).

A B

A B C D E

Fig.3.19. NR A) and SERS B)-C) of 2CEPhS. Conditions: A) neat liquid in glass capillary, 300 mW, 785 nm, 5-min; 1% v/v in MeOH on B) gold-doped TMOS capillary, 25 mW, and C) chem.L3, 100 mW, both at 785 nm, 1-min. Also shown, SERS D) and NR E) of PhSH. Conditions: D) 1% v/v in MeOH with TMOS (95-PEG) Ag vial, 100 mW of 785 nm, 1-min; E) neat liquid in glass vial, 300 mW of 785 nm, 5-min.

Fig.3.18. CEMS, SERS A) and NR B); and for comparison C) NR of CEEO, an oxygen analog of CEES (but SERS-inactive). Conditions: 1% v/v in MeOH at 100 mW 1-minute 785 nm, with chem.L3; NR of neat liquids at 300 mW, 785 nm, 5-min. Note: other ethers, e.g. diethylether (EEO, oxygen analog of EES) are typically not SERS-active in contrast to their sulfur analogs.

A B C

Fig.3.20. SERS of TDG-SO. Conditions: 0.1% v/v in MeOH, chem.L3 (ODS/MTMS based capillary), 100 mW of 785 nm, 1-min (30 scan).

Fig.3.21. 1,4-dithiane A) NR solid, 300 mW, 785 nm, 5-min; and B) SERS 1mg/mL in MeOH, chem.L1_PEG, 80 mW, 785 nm, 1-min.

A B


38

The oxidative and hydrolysis degradation pathways for organophosphate pesticides of the form (RO)2P=S(SR’) are shown in Figure 3.24. Here, the main products formed are shown specifically for the diethyl-ester pesticides (i.e. R = Et). The significance for the inclusion of pesticide oxons is reflected by the fact that during water treatment, oxidative degradation of parent OP pesticides{P=S} to their more toxic oxons {P=O} is prevalent in the disinfection process (accelerated in presence of HOCl/free chlorine below neutral pH).1 Subsequent hydrolysis of the oxons occurs readily during water softening (under alkaline conditions). Therefore, an apparent absence of OP pesticides as indicated by screening tests following water treatment, may in fact lead to a false-negative response if formation of extremely toxic oxons (and hydrolysis products) are not included in the analysis. The sodium salt of diethyl hydrogen dithiophosphate (DEHDTP, as a free acid), provides a simple structural model of these pesticides. The SER spectrum of a forensic sample of this analyte is presented in Figure 3.25A. As illustrated in Figure 3.24, these pesticides can hydrolyze to form a common degradation product O,O-diethyl phosphorothioic acid (DEPTA), which cannot be purchased. Alternatively, the parent pesticide can be oxidized to form an oxon, modeled here by the potassium salt of diethyl hydrogen thiophosphate (DEHTP, Figure 3.25B). These oxons may subsequently be hydrolyzed to form a common product diethyl hydrogen phosphate (DEHP), which could also be produced from oxidation of the primary parent hydrolysis product DEPTA. The SERS of DEHP is presented in Figure 3.25C for comparison with DEHTP (in Figure 3.25B) and DEHDTP (in Figure 3.25A). This general degradation scheme (see Figure 3.24) applies to the dimethyl-ester analogs as well (i.e. R = Me). The SERS of the simple representative model O,O-dimethyl hydrogen dithiophosphate (DMHDTP) is shown in Figure 3.26A. Unfortunately, the common hydrolysis product O,O-dimethyl phosphorothioic acid (DMPTA) could not be purchased for measurement as proposed. Instead, the potassium salt of O,O-dimethyl hydrogen thiophosphate (DMHTP) was acquired, and represents an oxon model for these pesticides (see Figure 3.26B). The oxon hydrolysis product O,O-dimethyl hydrogen phosphate (DMHP) is shown in Figure 3.26C. Note, with the exception of DMHTP, each of these structural models are additional secondary chemicals used to expand the SERS library. This also includes other oxons and degradation intermediates presented below.

Class 4A. Organophosphate Pesticides & Hydrolysis Products: (RO)2P=S(SR’)

OP

SO

OH OP

OO

OH

OP

OOSR'

OP

SOSR'

[Ox]-R'SH

-R'SH

+H2O

+H2O

1)

DEPTA DEHP

[Ox]

(RO)2P=S(SR')

Fig.3.24. Oxidative and Hydrolysis degradation pathways for OP pesticides of type (RO)2P=S(SR’). Note: R = Et or Me, and R’ is unique functional group coordinated via the thiol sulfur donor atom of the hydrolysable P-S bond.


39

The NR and SER spectra for specific examples of pesticides with diethyl ester groups, represented by disulfoton (DS) and its metabolite disulfoton sulfoxide (DS-SO) are shown in Figure 3.27. Figure 3.28 presents the NR and SERS of demeton-S (DEM-S), which is the corresponding oxon of DS.

We have included three additional secondary representative pesticides with dimethyl ester groups, dimethoate, phosmet and malathion, in our spectral library (see Figures 3.29-3.31).

O SP

SOS

O SP

SOSO

O SP

OOS

OP

SOS

OP

OO

OH

OP

OOS

O SP

SO

O

NH O S

PSO

N

O

O

Figure 3.30. Phosmet A) NR (solid), 295 mW at 785 nm, 5-min, and SERS B) chem.L3 HCl washed, 1mg/mL in MeOH, and C) chem.L2 1 ppm; 80 mW, 785 nm, 1-min. 512 cm-1 pk consistent with model (Fig.3.26); C) possible oxon contribution.

Fig.3.25. SERS of A) O,O-diethyl hydrogen dithiophosphate potassium salt, B) O,O-diethyl hydrogen thiophosphate potassium salt, and C) diethyl hydrogen phosphate. Conditions: chem.L3, 1mg/mL in MeOH (forensic), 80 mW, 785 nm, 1-min.

Figure 3.27. disulfoton A) NR and B) SERS; and disulfoton sulfoxide C) NR and D) SERS. Conditions: NR (neat) 295 mW at 785 nm, 5-min, B) chem.L1 with HCl wash, D) chem.L1_PEG; 1mg/mL in MeOH, 80 mW at 785 nm, 1-min.

Figure 3.28. demeton-S A) NR (neat), 280 mW, 785 nm, 5-min., SERS 1 mg/mL in MeOH, chem.L1_PEG at B) 80 mW, C) 140 mW, and after prolonged exposure D) 150 mW; 785 nm, 1-min. Note: possibility of C-S (or P-S) scission requires further study.

A B C

A B C D

A B C D

A B C

A B C

Fig.3.26. SERS of A) O,O-dimethyl hydrogen dithiophosphate potassium salt, B) O,O-dimethyl hydrogen thiophosphate potassium salt, and C) dimethyl hydrogen phosphate. Conditions: same as in Fig.3.25, but C) on chem.L1 (similar spectra on chem.L3).

A B C

OP

OO

OH

OP

OOS

OP

SOS

Figure 3.29. Dimethoate NR A) (powder), 300 mW at 785 nm, 5-min, and B) saturated in MeOH, 375 mW, 785 nm, 8-min, and SERS C) 1 mg/mL chem.L2. Conditions: 1mg/mL in MeOH, 100 mW at 785 nm, 1-min.


40

Figure 3.32 depicts a similar degradation pathway for pesticides represented by chlorpyriphos (and methyl parathion). The NR and SERS of CP, its oxon (CPO) and hydrolysis product (TCP) are compared in Figures 3.33 and 3.34. Figures 3.35 and 3.36 present the NR and SERS of MP, its oxon MPO, and hydrolysis product (4-NOPh). Note that both TCP and 4-NOPh are secondary chemicals prposed. Fig shows Diazinon (a proposed secondary OP pesticide), while Figures 3.37-3.42 present the NR and SERS of additional secondary chemicals used to further expand the library; and includes Demeton-O (DEM-O) , parathion (EP) and paraoxon (EPO), which are the ethyl analogs of MP and MPO, Pirimiphos-methyl (PIRI) and Chlorpyriphos-methyl (CPM).

Class 4B. Organophosphate Pesticides & Hydrolysis Products: (RO)2P=S(OR’)

O N

Cl

Cl

Cl

OP

SO

OR'OP

SO

O

OOP

OR'

OP

SO

OH OP

OO

OH

-R'OH

-R'OH

+H2O

2)

[Ox]

(RO)2P=S(OR')

+H2O [Ox]

DEHPDEPTA

Fig.3.32. Hydrolysis and Oxidative degradation pathways for OP pesticides of type (RO)2P=S(OR’). Note: R = Et or Me, and R’ is unique functional group coordinated to the oxygen donor atom of the hydrolysable P-O bond.

NCl

Cl

OH

Cl

O N

Cl

Cl

Cl

OP

OO

Fig.3.33. NR of A) chlorpyriphos (powder) 280 mW, B) chlorpyriphos oxon (solid) 180 mW and C) trichloropyridinol (powder) 295 mW; at 785 nm, 5-min.

A B C

A B C

Fig.3.34. SERS of A) chlorpyriphos on chem.L1_PEG, B) chlorpyriphos oxon on chem.L3, and C) trichloropyridinol on chem.L3; 1mg/mL in MeOH, 80 mW at 785 nm, 1-min.

Fig.3.31. Malathion NR A) (neat), 300 mW at 785 nm, 5-min, B) saturated in MeOH, 375 mW, 785 nm, 8-min, and SERS C) 0.1 mg/mL chem.L2, D) and E) on chem.L1_PEG. Conditions: 1mg/mL in MeOH, 80 mW at 785 nm, 1-min. Note: E) 140 mW, and additional bands in C) may be due to degradation.

A B C D E

O SP

SO

O O

O O


41

O OP

SONO2

O OP

OONO2

Fig.3.36.SERS of A) methyl parathion on chem2a_PEG, B) methyl parathion on chem.L3, C) methyl parathion oxon on chem2d, and D) 4-nitrophenol on chem.L3; 80 mW at 785 nm, 1-min. Note the degradation of MP to 4-NOPh in B).

Fig.3.35. NR of A) methyl parathion (powder), B) methyl parathion (neat, tech.), C) methyl parathion oxon (neat), and D) 4-nitrophenol (solid), 180 mW, at 785 nm, 5-min.

Fig.3.39. NR of A) ethyl parathion (neat), and B) ethyl parathion oxon (neat), 180 mW, at 785 nm, 5-min.

O OP

OONO2

O OP

SONO2

OH

NO2

A B C D

A B C D

A B

A B C D

Fig.3.40. SERS of ethyl parathion A)-B), and ethyl parathion oxon C)-D); chem.L3, 80 mW, 785 nm, 1-min. Note: EP degradation not clear if occurring here.

A B

Fig.3.38. Demeton-O NR A) (neat), 300 mW at 785 nm, 5-min, and SERS on B) chem.L2, C) chem.L1, D)-E) chem.L3. Conditions: 1mg/mL in MeOH, 100 mW at 785 nm, 1-min, and E) 20-sec.

A B C D E

O OP

SOS

Fig.3.37. Diazinon A) NR (neat), 200 mW at 785 nm, 4-min, and SERS on B) chem.L3 with HCl wash. Conditions: 1 mg/mL in MeOH, 80 mW at 785 nm, 20-sec.

O

N N

OP

SO


42

For comparative purposes, the oxidative and hydrolysis pathways for fonofos is provided in Figure 3.43. The SER spectra of relevant analytes that could be obtained are shown in Figure 3.44. Note that the oxon fonofoxon or primary hydrolysis product O-ethyl ethylphosphonothioic acid (EEPA) of fonofos could not be purchased. EEPA was a primary target for Task 1, and has been replaced DMHTP. The final common product, ethyl hydrogen ethylphosphonate (EHEP) was also obtained (forensic sample) and measured.

Class 4C. Organophosphate Pesticides & Hydrolysis Products: (R”)(RO)P=S(SR’)

SH

SP

SO

OP

O

OH

OP

S

SPh OP

O

SPh

OP

S

OH

3)

-PhSH

+H2O[Ox]

-PhSH

+H2O

EEPA EHEP

[Ox]

(R")(RO)P=S(SR')

OP

O

OH

Fig.3.44. SERS of A) fonofos, B) thiophenol, C)-D) ethyl hydrogen ethylphosphonate (2 different spots), on chem.L2, 1mg/mL in MeOH, 80 mW at 785 nm, 1-min.

Fig.3.43. Oxidative and Hydrolysis degradation pathways for OP pesticides of type (R”)(RO)P=S(SR’); EHEP similar to the nerve agent hydrolysis products (e.g. EMPA).

A B C D

A B C D E

N

O OP

SON

NA B C

Fig.3.41. Pirimiphos-methyl NR A) (powder), 300 mW at 785 nm, 5-min, and SERS B) chem.L1_PEG, and C) chem.L2_PEG. Conditions: 1mg/mL in MeOH, 80 mW, 785 nm, 1-min.

O N

Cl

Cl

Cl

OP

SO

Fig.3.42. Chlorpyriphos-methyl NR A) (powder), 300 mW at 785 nm, 5-min; and B) saturated in methanol, 375 mW, 785 nm, 10-min, and SERS on C) chem.L3_PDMS, D) and E) on chem.L1_PEG, Conditions: 1mg/mL in MeOH, 80 mW at 785 nm, 1-min.


43

Class 5. Non-Organophosphate Pesticides

Raman and SER spectra are given for 20 additional pesticides in Figures 3.45 – 3.64. In some cases SER spectra are shown for more than 1 sol-gel chemistry library. The hydrolysis products were not studied.

Fig.3.45. Glyphosate A) NR (powder), 280 mW at 785 nm, 5-min, and SERS on B) chem.L2 with HCl wash. Conditions: 1 mg/mL in MeOH, 80 mW at 785 nm, 20-sec.

A B

Fig.3.46. Carbaryl NR A) (powder), 300 mW at 785 nm, 5-min, and SERS on B) chem.L4, C) chem.L2, D) chem.L3 with HCl wash, E) chem.L3, and F) chem.L1_PEG. Conditions: 1mg/mL in MeOH, 100 mW at 785 nm, 1-min, with E)-F) at 80 mW.

A B C D E F

Fig.3.47. Oxamyl A) NR (powder), 280 mW at 785 nm, 5-min, and SERS on B) chem.L3. Conditions: 1 mg/mL in MeOH, 80 mW at 785 nm, 1-min.

A B

Fig.3.48. Aldicarb NR A) (powder), 300 mW at 785 nm, 5-min, and SERS on B) chem.L4, and C) chem.L1. Conditions: 1mg/mL in MeOH, 80 mW and 150 mW at 785 nm, 1-min and 20-sec, respectively.

A B C

Fig.3.49. Chlordane NR A) (neat), 300 mW at 785 nm, 10-min, B) NR spectrum corrected for fluorescence, and SERS on C) chem.L3_PDMS. Conditions: 1mg/mL in MeOH, 80 mW at 785 nm, 1-min. Note: fluorescence, which is not apparent in SERS, can be minimized in NR using 1064 nm excitation (see ENDO).

A B C

A B

Fig.3.50. Dalapon A) NR (neat), 280 mW at 785 nm, 5-min, and SERS on B) chem.L3. Conditions: 1 mg/mL in MeOH, 80 mW at 785 nm, 1-min.

O

ONH2+ P

OO

O

O

NH

O

NO

O NS

NO

NO

O N S

Cl

Cl

O

OHCl

ClCl

Cl Cl

Cl

ClCl


44

Fig.3.51. Atrazine A) NR (powder), 200 mW at 785 nm, 4-min, and SERS on B) chem.L3 with HCl wash. Conditions: 1 mg/mL in MeOH, 80 mW at 785 nm, 20-sec.

A B

A B C

Fig.3.52. Dicamba NR A) (powder), 280 mW at 785 nm, 5-min, and SERS on B) chem.L3, and C) chem.L3. Conditions: 1mg/mL in MeOH, 80 mW at 785 nm, 1-min. Note: possible photo-degradation may be occurring.

Fig.3.55. Endothall NR A) (powder), 280 mW at 785 nm, 5-min, and SERS on B) chem.L1, and C) chem.L3. Conditions: 1mg/mL in MeOH, 80 mW at 785 nm, 1-min.

A B C

Fig.3.56. Diquat NR A) (powder), 280 mW at 785 nm, 5-min, and SERS on B) chem.L1, and C) chem.L3. Conditions: 1mg/mL in MeOH, 80 mW at 785 nm, 1-min.

A B C

N

N

N

Cl

N N

Cl

OHO

O

Cl

COOH

COOHO

N+

N+

Br

Br

Fig.3.53. Toxaphene A) NR (solid), 295 mW at 785 nm, 5-min, and SERS on B) chem.L1_PEG. Conditions: 1 mg/mL in MeOH, 80 mW at 785 nm, 1-min.

A B

Cl8

Fig.3.54. Thiobencarb NR A) (neat), 280 mW at 785 nm, 5-min, and SERS on B) chem.L1 with HCl wash, and C) chem.L3 with HCl wash. Conditions: 1mg/mL in MeOH, 80 mW at 785 nm, 1-min.

A B C Cl S

ON


45

Following are 7 additional secondary pesticides (not proposed) we have included in the library; paraquat, thiobendazole, DDT, alachlor, permithrin, endosulfan, and endosulfan-sulfate (not reported).

Fig.3.59. Paraquat NR A) (neat), 280 mW at 785 nm, 5-min, and SERS on B) chem.L1_PEG with HCl wash, 80 mW, C) chem.L1_PEG with HCl wash, 10 mW, D) chem.L1_PEG, and E) chem.L3_PEG. Conditions: 1mg/mL in MeOH, 80 mW at 785 nm, 1-min (unless noted).

A B C D E

Fig.3.61. Endosulfan NR A) (powder), 400 mW at 1064 nm, 10-min, and B) (powder), 300 mW at 785 nm, 5-min; SERS on C) chem..L3, D) chem.L3_PDMS, and E) chem.L1_PEG. Conditions: 1mg/mL in MeOH, 80 mW at 785 nm, 1-min, E) 20-sec. Note: Fluorescence in NR at 785 nm, but minimized in NR using 1064 nm; possible degradation in C).

A B C D E

Fig.3.62. Endosulfan sulfate NR A) (powder), 280 mW at 785 nm, 5-min, and SERS B) chem.L4, and C) chem.L1_PEG. Conditions: 1mg/mL in MeOH, 80 mW at 785 nm, 1-min. Note: hydrolysis product of endosulfan (Fig. 3.61).

A B C

N+

N+ CH3CH3

ClCl

SO

O

O

ClCl

ClCl

ClCl

SO

O

ClCl

ClCl

ClClOO

Fig.3.60. Thiobendazole NR A) (powder), 280 mW at 785 nm, 5-min, and SERS on B) chem.L1_PEG with HCl wash, 1-min and C) chem.L1_PEG with HCl wash, 20-sec, D) chem.L1_PEG, and E) chem.L3 with HCl wash. Conditions: 1mg/mL in MeOH, 80 mW at 785 nm, 1-min (unless noted).

A B C D E

N

S

N

NH

A B

A B C Fig.3.57. Simazine. A) NR (powder) 180 mW at 785 nm, 5-

min; B) SERS (1 mg/mL) on chem.L3 HCl washed, 80 mW at 785 nm, 1-min.

Fig.3.58. 2,4-dinitrotoluene. A) NR (powder), B) NR 0.026 g in 0.4 mL MeOH; 295 mW at 785 nm, 5-min; C) SERS (1 mg/mL) on chem.L3 HCl washed, 80 mW at 785 nm, 1-min.


46

Class 6. Toxic Industrial Chemicals Raman and SER spectra are given for 11 chlorinated TICs, and for AA and TFA, in Figures 3.65 – 3.76. The spectra for Melamine are shown in Figure 3.77. In some cases SER spectra are shown for more than 1 sol-gel chemistry library. The hydrolysis products were not studied.

Fig.3.67. Chloropicrin. A) NR (neat) 295 mW at 785 nm, 5-min; B) SERS (1 mg/mL) on chem.L1, 90 mW at 785 nm, 1-min.

A B

A B

Fig.3.65. 3,5-dichlorobenzoic acid. A) NR (powder) 180 mW at 785 nm, 5-min; B) SERS (1 mg/mL) on chem.L3, 80 mW at 785 nm, 1-min.

A B

Fig.3.68. 2-chlorotoluene. A) NR (neat) 295 mW at 785 nm, 5-min; B) SERS (1 mg/mL) on chem.L1_PEG, 80 mW at 785 nm, 1-min.

Fig.3.66. 4,4-dichloro biphenyl. A) NR (powder) 180 mW at 785 nm, 5-min; SERS (1 mg/mL in DCM) on B) chem.L3 HCl washed, and C) chem.L4, 80 mW at 785 nm, 1-min.

A B C

OO

O

Cl

Cl

ClCl

ClCl

Cl

N OCl

O

Fig.3.64. DDT NR A) (powder), 300 mW at 785 nm, 5-min, and SERS B)-C) chem.L4 and D) chem..L3. Conditions: 1mg/mL in MeOH, 100 mW at 785 nm, 60-sec. Note: common peaks in C and D suggests previous artifact observed in some instances on silver may also occur in gold. Although differences in B) suggest this may be DDT, further studies are required.

Fig.3.63. Permethrin A) NR and B) SERS; Alachlor C) NR and D) SERS; Conditions: NR (powders), 300 mW, 785 nm, 5-min; SERS on chem.L4, 1mg/mL in MeOH, 80 mW, 785 nm, 1-min. Note neither analyte appears active on silver, and it is not certain if the activity exhibited on gold here is real or due to an artifact or degradation product; more rigorous testing of gold with various analytes is in progress.

A B C D

A B C D


47

Fig.3.69. 1,3,5-trichlorobenzene. A) NR (powder) 295 mW at 785 nm, 5-min; B) SERS (1 mg/mL) on chem.L1, 90 mW at 785 nm, 1-min.

Fig.3.70. Bromochloroacetic acid. A) NR (neat) 300 mW at 1064 nm, 10-min; B) SERS (1 mg/mL) on chem.L1, 90 mW at 785 nm, 1-min.

A B

A B

A B

A B

A B

A B C

Fig.3.74. Dichloroacetonitrile. A) NR (neat) 295 mW at 785 nm, 5-min; B) SERS (1 mg/mL) on chem.L1, 80 mW at 785 nm, 1-min.

Fig.3.71. Chlorobenzene. A) NR (neat) 295 mW at 785 nm, 5-min; B) SERS (1 mg/mL) on chem.L1_PEG, 80 mW at 785 nm, 1-min.

Fig.3.72. 1,4-dichlorobenzene. A) NR (powder) 295 mW at 785 nm, 5-min; SERS (1 mg/mL) on B) chem.L1, and C) chem.L4, 80 mW at 785 nm, 1-min.

Fig.3.73. Picloram. A) NR (powder) 180 mW at 785 nm, 5-min; B) SERS (1 mg/mL) on chem.L2 HCl washed, 80 mW at 785 nm, 1-min.


48

Figure 3.76 shows acetic acid and trifluoroacetic acid, relevant to BCAA and DCAA (Figures 3.70 and 3.75, respectively). Melamine (Figure 3.77), a fertilizer, was added to this list of toxic industrial chemicals, as it has been used in China to artificially increase the apparent amount of protein in foods (pet food, baby formula), but has also caused deaths of pets in the USA.

Class 7. Incapacitating Agents We have measured 2-chloroacetophenone (an incapacitating agent employed in tear gas) and 2-chlorophenylmethyl sulfone (see Figures 3.78 and 3.79).

A B

Fig.3.75. Dichloroacetic acid. A) NR (neat) 300 mW at 1064 nm, 10-min; B) SERS (1 mg/mL) on chem.L1, 80 mW at 785 nm, 1-min.

N

NN

NH2

NH2NH2

A B C

Fig.3.77. Melamine A) NR powder, in cap 188 mW, 785 nm, 5-min; SERS B) chem.L6.3 (APTEOS) and C) chem.L6.4 (1:1 APTEOS/MTMS). Conditions: ~1mg/mL in HPLC water, 80 mW, 785 nm, 1-min.

Fig.3.78. 2-chloroacetophenone. A) NR (powder) 295 mW at 785 nm, 5-min; B) NR at 500 mW, 1064 nm, 5-min; C) SERS (0.1 mg/mL in MeOH) on chem.L2, 80 mW at 785 nm, 1-min.

Fig.3.79. 2-chlorophenylmethyl sulfone. A) NR (powder) 295 mW at 785 nm, 5-min; B) SERS (1 mg/mL in MeOH) on chem.L2, 80 mW at 785 nm, 1-min.

Fig.3.76. Acetic acid A) NR (neat) at 295 mW, 785 nm, 5-min, and B) SERS on chem.L1, 80 mW, 785 nm, 1-min; Trifluoroacetic acid C) NR (neat) at 295 mW, 785 nm, 5-min, and D) SERS on chem.L1, 80 mW, 785 nm, 1-min.

A B C D

A B C

A B


49

Class 8. Biologicals We also included three biological chemicals as potential interferents, and included them in the spectral library. Dipicolinic acid is a biomarker for anthrax, as it is released when the spores germinate (which can be initiated by water). Tryptophan and phenylalanine were also included, since they are two highly SERS-active amino acids that may occur in water from biomass degradation. The normal Raman and SER spectra for these biochemicals are shown in Figures 3.80 and 3.81, respectively.

Class 9. Dyes We also investigated seven dyes to use as a safe chemical for testing water at Kensico reservoir (Figures 3.82-3.93). As shown, a number of these gave good sensitivity, especially sunset yellow and allura red at 1 part-per-billion. But sunset yellow, used in Gatorade, was chosen as the safest. Note that R6G and CV are carcinogens, and not appropriate for field tests. In all cases, the Raman spectra are included with the SER spectra for peak comparison. In addition, SER spectra are shown for various sol-gel chemistries (libraries) and at various concentrations. These chemicals were NOT included in the search library (although it wouldn’t produce false positives or negatives).

A B C D

Fig.3.82. Sunset yellow (FD&C Yellow No.6), sodium salt A) NRS, 100 mg/mL, 180 mW, 785 nm, 5-min; SERS static B) 1 mg/mL on chem.L3; C) 1 mg/mL, D) 1 ppm on chem.L3. Conditions: HPLC H2O, 80 mW, 785 nm, 1-min.

Fig.3.83. SERS of Sunset yellow at A) 1 ppm, B) 10 ppb and C) 1 ppb on chem.L6 (vials). Conditions: as in Fig.3. Note: signal very weak at 100 parts-per-trillion.

A B C

Fig.3.80. NR of A) dipicolinic acid (DPA), B) tryptophan (Trp) and C) phenylalanine (Phe). Conditions: powders, 300 mW, 785 nm, 5-min.

Fig.3.81. SERS of A) dipicolinic acid, B) tryptophan and C) phenylalanine. Conditions: 1 mg/mL, chem.L2, 80 mW, 785 nm, 1-min.

A B C

A B C


50

Fig.3.89. CV A) NRS, 1 mg/mL, 180 mW, 785 nm, 10-min; SERS static B) 1 ppm chem.L6 (vial); C) 100 ppb chem.L3 and D) 1 ppm chem.L4. Conditions: HPLC water, 80 mW, 785 nm, 1-min. Note active on chem2a.

Fig.3.87. SERS of patent blue V on chem.L6 (vials) at A) 10 ppm, B) 100 ppb, and C) 10 ppb. Conditions: as in Fig.14. Note: very weak signal was observed at 1 ppb on chem.L6; negative at 1 ppm on chem.L2 and chem.L3.

Fig.3.86. Patent blue V (sodium salt) A) NRS, 1 mg/mL, 180 mW, 785 nm, 5-min; SERS static B) 1 mg/mL on chem.L2, and C) 1 ppm on chem.L4. Conditions: in HPLC water, 80 mW, 785 nm, 1-min.

A B C

A B C

Fig.3.88. Tartrazine A) NRS, 1 mg/mL, 180 mW, 785 nm, 5-min; SERS on chem.L6 (vials) B) 1 ppm, and C) 100 ppb. Conditions: as in Fig.14. Note negative on chem.L4 at 1 ppm, on chem2d and chem2a at 1000 ppm.

A B C

A B C D

Fig.3.84. Allura red AC (FD&C Red No.40) sodium salt A) NRS, 100 mg/mL, 180 mW, 785 nm, 5-min; SERS static B) 1 ppm and C) 10 ppb on chem.L6 (vials). Conditions: in HPLC water, 80 mW, 785 nm, 1-min. Note: no activity observed on chem.L3.

A B C

A B C

Fig.3.85. SERS of Allura red AC at A) 1 ppb, B) 1 ppb (with artifact subtracted) and C) 10 ppt on chem.L6 (vials spin-coated with colloids only). Conditions: as in Fig.5 Note: artifact pks at 885/1406 cm-1 in A) are due to reduced Ag and not the APTES sol-gel (see RPT#18).


51

Fig.3.91. SERS of R6G on chem.L2 (HCl washed), static A) 1 ppm, B) 100 ppb, C) 10 ppb and D) 1 ppb. Conditions: as in Fig.17.

A B C

Fig.3.90. R6G A) NRS, 0.7 mg/mL, 180 mW, 785 nm, 5-min; SERS static B) 1 ppm, chem.L6 (vial) and C) 1 ppm on chem.L4. Conditions: as in Fig.17.

A B C D

Fig.3.92. Erythrosin B (FD&C Red No.3), sodium salt A) NRS, 50 mg/mL, 180 mW, 785 nm, 10-min; SERS static B) 1 mg/mL and C) 1 ppm on chem.L2. Conditions: in HPLC MeOH, 80 mW, 785 nm, 1-min.

A B C

A B

Fig.3.93. SERS of Erythrosin B at A) 1 ppm, and B) 500 ppb on chem.L6 (vials). Conditions: as in Fig.1. Note: signal very weak at 100 parts-per-billion.


52

Class 10. Standard Test Chemicals Figures 3.94-3.96 show the standard test chemicals used as SERS calibrants and a SERS internal reference from Phase I. These chemicals were NOT included in the search library (although it wouldn’t produce false positives or negatives).

Spectral Library Search and Match Algorithms As part of this program, we evaluated the ability of four spectral library search routines to identify the SER spectra of the twenty proposed primary targets representing various chemical warfare agents (CWAs), organophosphate (OP) pesticides, and toxic industrial chemicals (TICs). The Euclidean Distance Algorithm (EDA), the Absolute Value Algorithm (AVA), the Least Squares Algorithm (LSA), and the Correlation Algorithm (CA) have been incorporated into a LabVIEW program at RTA and successfully applied to both Raman and SER spectra. These four different search algorithms are summarized in Figure 3.97.

Fig.3.94. NR of A) aniline (neat), benzoic acid (powder), C) p-aminobenzoic acid (powder), D) phenylacetylene (neat) and E) pyridine (neat). Conditions: 300 mW, 785 nm, 5-min.

Fig.3.95. SERS of A) aniline , benzoic acid, C) p-aminobenzoic acid, D) phenylacetylene and E) pyridine. Conditions: 1 mg/mL in MeOH, chem.L3, 80 mW, 785 nm, 1-min.

A B C D E

A B C D E

Fig.3.96. KSCN. A) NR (solid) 180 mW at 785 nm, 5-min; B) SERS (1 mg/mL) on chem.L1, 80 mW at 785 nm, 1-min.

A B


53

Euclidean Distance Algorithm (EDA) Absolute Value Algorithm (AVA) Least Squares Algorithm (LSA)

HQI 2 1Lib Unkn⋅

Lib Lib⋅ Unkn Unkn⋅⋅−⋅

Correlation Algorithm (CA)

HQI 1Libm Unknm⋅( )2

Libm Libm⋅( ) Unknm Unknm⋅( )⋅−

Where: In all cases, Lib is the library entry being searched and Unkn is the unknown sample spectrum. Each method scores the spectral match in terms of a hit quality index (HQI), where the best score or match is 0 (Lib = Unkn) and the worst score or match approaches 2½ (1 for CA). Initially, we tested these algorithms against a SERS library composed of the 20 primary targets and 28 of the 30 proposed secondary chemicals. The library spectra were collected using 80 mW of 785 nm, 1-min, and 1 mg/mL concentrations. In this preliminary evaluation, spectra different from that used in the library were employed for these tests (the HQI scores are presented in Table 3.2). Generally, we have found that the best results are obtained by subtracting the glass background produced by the capillary from both the sample and the spectra in the library. Smoothing the data improved scores, but not significantly. In all cases, as summarized in Table 3.2, the CA method resulted in a positive match for each of the 20 targets. More importantly, the CA method gave the best overall results for correctly identifying each chemical (reasonably low HQI score) as well as discriminating against structurally similar analytes (relatively higher HQI scores). Table 3.2. Summary of Hit Quality Index Scores for Spectral Match Search Algorithms on proposed 48 component library (20 primary reference subLib + 28 secondary reference subLib (missing HD and VX)) using non-reference test spectra of each primary target. Note, + implies a spectral match, - implies no match. Ssub = background subtracted spectra, dSsub= background subtracted and first-derivative spectra.

Analyte

Euclidean Distance Absolute Value Least Squares Correlation

(Ssub) (dSsub) (Ssub) (dSsub) (Ssub) (dSsub) (Ssub) (dSsub) IMPA - + 0.519 - - - + 0.001 + 0.286 + 0.768 PMPA - + 0.439 - + 0.017 - + 0.001 + 0.191 + 0.685 CMPA - + 0.489 - + 0.017 - + 0.001 + 0.183 + 0.739 EMPA + 0.025 - - - - - + 0.209 - MPA - + 0.364 + 0.062 + 0.011 + 0.006 + 0.000 + 0.232 + 0.595 DIASH - + 0.651 - + 0.016 - + 0.001 + 0.243 + 0.878 CSPS + 0.036 + 0.126 - + 0.007 + 0.010 + 0.000 + 0.066 + 0.236 CEES + 0.007 + 0.484 - + 0.014 - + 0.000 + 0.114 + 0.733 CEMS + 0.007 - + 0.040 - + 0.003 - + 0.316 - TDG + 0.021 + 0.245 + 0.064 + 0.009 + 0.006 + 0.000 + 0.098 + 0.430 HEES + 0.001 - + 0.025 - + 0.001 - + 0.021 - CN + 0.063 - - - + 0.008 - + 0.209 - DS + 0.049 + 0.449 - - - + 0.000 + 0.300 + 0.696 DS-SO - + 0.690 - - - - + 0.125 + 0.904 CP + 0.001 + 0.213 + 0.018 + 0.005 + 0.000 + 0.000 + 0.037 + 0.381 MP - + 0.365 - + 0.009 - + 0.001 + 0.380 + 0.597 FON + 0.134 + 0.062 - + 0.010 - + 0.000 + 0.064 + 0.121 EHEPA + 0.008 + 0.438 + 0.045 + 0.014 + 0.004 + 0.001 + 0.144 + 0.685 35DCBA + 0.004 + 0.034 + 0.020 + 0.005 + 0.001 + 0.000 + 0.038 + 0.066 44DCBP + 0.053 + 0.033 + 0.051 + 0.003 + 0.004 + 0.000 + 0.391 + 0.066

Note, we later substituted EHEPA with a new primary target, DMHTP.

HQI1

n

i

Libi Unkni−∑=

nHQI

1

n

i

Libi Unkni−( )2∑=

n

Libm Libn

n

i

Libi∑=

n− Unknm Unkn

n

n

i

Unkni∑=

n−

Figure 3.97. Equations for the spectral library search algorithms.


54

As a specific example, the results obtained for identifying disulfoton sulfoxide (DS-SO), and the ability to discriminate between this metabolite and the parent OP pesticide disulfoton (DS) are presented in Figures 3.98 and 3.99, respectively. For the Correlation Algorithm, the HQI score is 0.125, while the next lowest score is 0.270, which is the parent pesticide. The third best score is for another sulfur containing chemical, CEES, but it has a significantly higher score of 0.471.

Fig.3.98. Display of Spectral Library Search and Match software. The program ranks the best matches (lowest scores) and overlays the unknown and matched spectra for visual comparison (SERS reference library = 48 chemicals). Both the sample and library spectra were 80 mW of 785 nm, 1-min, and 1 mg/ml concentration.

Fig.3.99. SERS of A) Disulfoton sulfoxide on chem.L6 as the unknown sample, and B) Disulfoton on chem.L1 (see Fig.3.98). Conditions: 1 mg/mL in MeOH, 80 mW, 785 nm, 1-min.

A B

DS-SO DS CEES


55

The ability of CA to discriminate against chemicals NOT included in the SERS reference library, we also measured the spectrum of benzoic acid (on L3). As shown in Figure 3.100, the best matches were dichlorobenzoic acid (DCBA) for CA and chlorpicrin for the first derivative CA. As can be seen, the HQI values are 0.853 and 0.780, respectively, and clearly indicate that a match has not been found. In general it was found that the HQI scores above 0.4 represent no match (Table 3.2). Also, our preliminary results indicate that the CA search routine is superior to the other algorithms in providing such discrimination. After we were satisfied that the Correlation Algorithm correctly identified the 20 target chemicals, the library was expanded to 96 chemicals and the algorithm was re-tested. As shown in Figure 3.101, 1 mg/ml VX could be identified from this 96 component library.

Fig.3.100. Display of Spectral Library Search and Match software using 1 mg/ml BA as unknown chemical not in 48 component reference library. The results are shown for the Correlation method with non-treated (left) and first derivative (right) corrected spectra.

Fig.3.101. Results using 96 component library for previous SERS of VX obtained in silver-doped TMOS vials at Aberdeen. Conditions: 1mg/mL in HPLC water, 100 mW at 785 nm, 20-sec. VX EA2192 EMPA

DCBA

CPIC


56

The ability of this program to perform a functional group analysis is demonstrated in Figure 3.102, where EMPA is identified and ranked accordingly relative to other alkyl phosphonic acids (APAs) and parent CWAs. In addition to these measurements, the search routines were challenged by using SERS of analytes at different pH and concentration, on different SERS-active sol-gels, and with the static and continuous flow and AIEX/SPE preconcentration sampling methods. In general, the alkoxide chemistry of the sol-gels produces insignificant changes in the SER spectra. Whereas an acid wash (pH adjustment) or choice of metal precursor (i.e. silver vs gold) can result in significant spectral differences for many, but not all chemicals. Nevertheless, for these chemicals that had such spectral differences, proper identification can be achieved if both spectral versions were in the library. As an example, the ability to identify MPA extracted from water at 10 ppb (pH > 7) on the open-tubular chem.L6, as opposed to the static reference spectrum measured on PC chem.L3 (1mg/mL, pH = 2.24) is shown in Figure 3.103. Figure 3.104 shows the results for TDG (10 ppm) static.

Fig.3.103. Results for MPA obtained on chem.L6 at 10 ppb in water (static) using the 96 component library. Conditions: 80 mW, 785 nm, 1-min. Note the following ranking of phosphate containing chemicals: Hit Quality Name 0 0.174 MPA 1 0.477 DEHDTP 2 0.498 EMPA 3 0.512 CMPA 4 0.522 DMHP 5 0.524 VX

Fig.3.102. Results for EMPA obtained on chem.L3 with 96 component library. Conditions: 1mg/mL, 80mW, 785 nm, 1-min. Note the following ranking: Hit Quality Name 0 0.209 EMPA (VX hydrolysis product) 1 0.376 IBMPA (RVX hydrolysis product) 2 0.384 PMPA (GD hydrolysis product) 3 0.431 IMPA (GB hydrolysis product) 4 0.435 EA2192 (VX hydrolysis product) 5 0.450 CMPA (GF hydrolysis product) 6 0.455 MPA (VX final hydrolysis product) 7 0.471 VX (parent of EMPA )


57

The next set of Figures 3.105 and 3.106 show the ability to correctly identify samples flowed at lower concentration with this 96 component reference library.

Fig.3.105. Ability to identify and discriminate A) parent pesticide DS and B) metabolite DS-SO, both flowed at 1 ppm (50 mL) on chem.L2.

A B

DS DS-SO

DS-SO DS

Fig.3.104. Results for TDG obtained on chem.L1 at 10 ppm in water (static) using the 96 component library. Conditions: 80 mW, 785 nm, 1-min. Note the following ranking of these sulfur containing chemicals: Hit Quality Name 0 0.258 TDG 1 0.366 HD 2 0.392 EtSH 3 0.409 CEES 4 0.440 CEMS 5 0.558 HEES 6 0.570 DS 7 0.600 14DT 8 0.607 DS-SO 9 0.609 DEM-S

FON PhSH 2CEPhS

Fig.3.106. Ability to detect FON flowed at 100 ppb (50 mL) on chem.L2.


58

The last test demonstrates the ability to identify targets at 10 ppb pre-concentrated by AIEX, SPE and IR methods with this 96 component library (Figures 3.107-3.110).

Fig.3.107. NaCN at 10 ppb using IR method on chem.L2. Fig.3.108. 35DCBA 10 ppb by AIEX method on chem.L2.

.

Fig.3.110. Results for 10 ppb CMPA using AIEX method on chem.L2; using the 96 component library. Conditions: 80 mW, 785 nm, 1-min. Note the following ranking of phosphonates: Hit Quality Name 0 0.056 CMPA 1 0.350 IBMPA 2 0.356 EMPA 3 0.386 PMPA 4 0.448 IMPA 5 0.464 EA2192 6 0.464 MPA 7 0.468 DS 8 0.472 DEM-S 9 0.473 VX

35DCBA

NaCN

Fig.3.109. Results for 10 ppb CEMS using SPE method on chem.L2; using the 96 component library. Conditions: 80 mW, 785 nm, 1-min. Note the following ranking of sulfur containing chemicals: Hit Quality Name 0 0.091 CEMS 1 0.400 CEES 2 0.436 14DT 3 0.546 HD 4 0.611 TDG 5 0.624 EtSH


59

Task 4. Test Sol-Gel Capillaries with Real Water Samples. The overall objective of this task is to build a sample system that will ensure that the SERS-active capillaries can detect poisons in water. This will be accomplished by measuring the SERS of some 20 water samples supplied by American Water Works obtained from various locations of their drinking water distribution systems (see letter on page 28). These measurements will identify potential interferents (chemical and physical) and determine if and how they will be removed. Water samples were prepared to quantify the effect of two major interferents (minerals and chlorinated species) on the performance of the SERS substrates. In the case of minerals, calcium carbonate (CaCO3) was used at a high concentration to produce a hard water sample. Specifically for this experiment, 50 ppm of CaCO3 was dissolved in a 1 ppm aqueous solution of MPA. Although CaCO3 is insoluble in water, the presence of an acid (MPA) provided a mechanism for its dissolution. The SERS response of this methyl phosphonic acid sample was then measured. As seen from Figure 4.1A, the hard water sample reduces the SERS-response of MPA. To improve the SERS-response, we used ion-retardation resins (Bio-Rad) to remove the ions. These resins contain paired anion and cation exchange sites, which are extremely efficient in extracting small inorganic ions (e.g. alkaline salts, Cl- and OCl- etc) from both aqueous as well as organic sample matrices. The ion-retardation (IR) resin was prepared as a slurry in HPLC grade water and loaded into a capillary. Sol-gel frits were then formed on each end of the capillary to serve as physical supports for the resin. After this the hard water sample doped with MPA (0.5 mL) was passed through an IR resin at a flow rate of 0.5mL/min, which extracted the Ca2+ and the CO3

2- ions allowing a

more intense SERS-response for MPA (Figure 4.1B). A similar experiment was performed with another target chemical, dichlorobenzoic acid. Briefly, 100 ppm of CaCO3 was dissolved in a 100 ppm aqueous solution of DCBA. The SERS-responses were measured before and after passing through an IR resin. As can be seen from Figure 4.2 (A), 100 ppm of CaCO3 obscures the SERS of DCBA. After passing this sample solution through the IR resin removes the interfering ions, allowing a normal spectrum to be observed (Figure 4.2B). Clearly, the IR resin improves the SERS-response of chemicals in hard water.

Next several experiments were performed for determining the effects of chlorinated water on the SERS-response. First, we measured several chemicals in ordinary tap water containing low levels of chlorine. In general, it was observed that low chlorine concentrations do not diminish the SERS-response (at least for static measurements). As an example, the SERS of CEES and CN in ordinary tap water and HPLC water are shown in Figures 4.3 and 4.4A. However, MPA at 100 ppm on chem.L6 exhibited some fluorescence in the SER spectrum, as seen from the rising background from Figure 4.4 (B). The SERS-response could be improved by passing the sample through an IR resin as seen from Figure 4.4 (C).

Fig.4.1. SERS of 1 ppm MPA in aqueous solution containing 50 ppm CaCO3 A) before and B) after passing through an ion-retardation resin. Conditions: 80 mW, 785 nm, 1-min, chem.L6.

A B

Fig.4.2. SERS of 100 ppm DCBA in aqueous solution containing 100 ppm CaCO3 A) before and B) after passing through an ion-retardation resin. Conditions: 80 mW, 785 nm, 1-min, chem.L2.

A B


60

Second, water samples containing hypochlorite ion (ClO-) as NaOCl were prepared to mimic the presence of the ClO- ions, which is present during the water disinfection process through the use of chlorine. For this experiment, a 1 ppm MPA aqueous sample was prepared containing 50 ppm ClO- and the SERS was measured. The spectrum is dominated by the ClO- peak at 700 cm-1 (see Raman spectrum in Figure 4.5A), and the MPA peak at 750 cm-1 is only observed if a spectrum of ClO- without MPA is subtracted (Figure 4.5B). MPA is still barely discernable after the use of an IR resin, but clearly visible after spectral subtraction (~ 4x as intense, Figure 4.4C). It is important to note that the ClO- ions at this concentration level do not inactivate the sol-gel SERS-activity, at least for chem.L6.

• • •

Dr. Farquharson visited Yves Mikol (Director of Early Water Surveillance at the New York City Department of Environmental Protection, NYCDEP; 465 Columbus Ave, Valhalla, NY 10595) with John Canning at the Kensico Reservoir water supply in New York. The purpose of the visit was to collect various samples, untreated water from the Kensico Reservoir Monitoring Station and treated (chlorinated) water from a downstream location, for the initial tests to be performed at RTA. Four water samples were provided to Dr. Farquharson labeled EWS -2008-002, EWS -2008-03, EWS -2008-04 and EWS -2008-005. Water samples EWS -2008-03 and EWS -2008-04 had pH (7.07 and 7.81), turbidity (1.47 NTU and 1.35 NTU) and specific conductivity (58 μS/cm and 392 μS/cm), respectively.

A B

Fig.4.3. SERS of 100 ppm CEES in A) ordinary tap water from RTA, and B) in pure HPLC water. Conditions: chem.L1_PEG, 80 mW, 785 nm, 1-min.

Fig.4.5. A) NR of stock 10% NaOCl solution in a blank capillary (188 mW, 785 nm, 5-min); and SERS of 1 ppm MPA prepared in an aqueous solution containing 50 ppm ClO- B) before and C) after passing through the ion-retardation resin. Conditions: chem.L6, 80 mW, 785 nm, 1-min.

A B C

Fig.4.4. SERS of A) CN on chem.L2, B) MPA on chem.L6 at 100 ppm in ordinary tap water, and C) MPA after passing through an IR resin. Conditions: 80 mW, 785 nm, 1-min.

A B C


61

All four water samples were tested on chemistries L6 and L3 to obtain a reference SERS response and to determine the presence of any interference (Figures 4.6 and 4.7). Figure 4.6A shows intense bands at 552, 906 and 1030 cm-1

for water sample EWS-2008-002, (measured on chem.L6). These peaks are attributed to the phosphate anion based on our previous measurements of phosphoric acid and cyclophosphamide (Figure 4.8). This water sample was then passed through an IR resin to remove the phosphate anion and tested again on chem.L6 and L. As can be seen from Figures 4.9-11, the IR resin successfully removed the phosphate ions from all 4 Kensico water samples.

RL

A B C D

A B C D

Fig.4.6. SERS of water samples from Kensico water reservoir. A) EWS-2008-002, B) EWS-2008-03, C) EWS-2008-04 and D) EWS-2008-005 on chem.L6; RL = Room Lights. Conditions: as in Fig.4.1.

Fig.4.7. SERS of water samples from Kensico water reservoir. A) EWS-2008-002, B) EWS-2008-03, C) EWS-2008-04 and D) EWS-2008-005 on chem.L3; Conditions: as in Fig.4.1.

A B C

A B

Fig.4.8. SERS A) EWS-2008-002 on chem 6c, B) phosphoric acid, and C) cyclophosphamide on chem.L3; Conditions: as in Fig.4.1.

Fig.4.9. SERS of EWS-2008-002 , A) before passing through IR and B) after passing through IR on chem.L6. Conditions: as in Fig.4.1. A

B C Fig.4.10. SERS A) EWS-2008-003 after IR, B) EWS-2008-04,

after IR and C) EWS-2008-05 after IR on chem.L6; Conditions: as in Fig.1. Note Y-axis, these spectra are noise.

RL

Fig.4.11. SERS A) EWS-2008-002 after IR, B) EWS-2008-03, after IR and C) EWS-2008-04 after IR and D) EWS-2008-005 after IR on chem.L3; RL = Room Lights. Conditions: as in Fig.1.

A B C D


62

The next set of experiments involved doping EWS-2008-002 with 10 ppb of sunset yellow dye. For this experiment a 50 mL sample solution of this dye was sequentially passed through IR resin cartridge and then through an isopropyl alcohol (IPA) pre-conditioned C-18 SPE cartridge at a flow rate of 1 mL/min. This served as a matrix “cleanup” (IR resin) as well as a pre-concentration step (SPE). Since C-18 is a non-polar phase, the organic dye is retained on the cartridge (hydrophobic interactions between the C-H bonds of sunset yellow and the C18 of the sorbent). After this the cartridge was washed with 3 mL of HPLC water to remove the sample matrix and then the extracted dye was eluted off the cartridge with 1 mL of 13% isopropyl alcohol (IPA). Figure 4.12A shows the SERS of 10 ppb sunset yellow extracted from water sample EWS-2008-002, while Figure 4.12B shows the SERS of a 10 ppb sunset yellow pre-concentrated from HPLC grade water and Figure 4.12C shows the static SERS of 1ppm sunset yellow for comparison. All measurements used chem.L6.

For the next set of experiments, the remaining Kensico water samples, EWS-2008-03, EWS-2008-04 and EWS-2008-005, were doped with 10 ppb sunset yellow dye, SPE pre-concentrated, and measured using chem.L6 and L3 (Figures 4.13 and 4.14). The above set of experiments demonstrated our ability to measure chemicals in real water samples using a combination of IR and SPE sample clean-up methods. SERS of other target chemicals in Kensico water samples are presented in Task 6 as part of the Parker sampling System.

The results of this task clearly indicate that the presence of inorganic ions (CaCO3, ClO- and phosphate anion) as interferents in water can reduce the SERS-response of the chemicals. Fortunately, passing water samples through ion retardation resins can substantially improve the SERS-response. It should be noted that the use of IR resin will also extract and retain CN and to some extent MPA from water samples. Yet, ion retardation resins are not necessary for chem.L3, since this chemistry exhibits very low SERS response towards such inorganic ions, especially phosphate (Figure 4.7). In instances where chem.L3 cannot be used (e.g. detection of dyes, or MPA) the IR resin must also be tested to ensure that CN and MPA are not extracted. This can be done by sequentially washing the IR resin bed with 0.01M HNO3 (for eluting off any CN) and with 0.01M NaCl (for eluting off MPA). To avoid any complications of specific chemicals being retained on the IR resin, the sampling system in Task 6 was designed so as to minimize the use on the IR resin.

A B C

Fig.4.12. SERS of Sunset Yellow in A) 10 ppb in EWS- 2008-002 water sample (extracted by C-18 pre-concentration), B) 10 ppb in HPLC water (extracted by C-18 pre-concentration), and C) 1 ppm static in HPLC water. Conditions: as in Fig.1. chem.L6. Note: The peak at 810 cm-1 in pre-concentrated EWS 2008 002 sample may be a contribution from IPA which was used as a solvent to elute the dye off from the C18 column.

A B C

A B C

Fig.4.13. SERS of 10 ppb Sunset Yellow in A) EWS-2008-03, B) EWS-2008-04 and C) EWS-2008-005. Conditions: 80 mW of 785 nm, 1-min, chem.L6

Fig.4.14. SERS of 10 ppb Sunset Yellow in A) EWS-2008-03, B) EWS-2008-04 and C) EWS-2008-005. Conditions: as in Fig.4.12, chem.L3


63

Task 5. Establish Performance (ROC Curves). The overall objective of this task is to characterize the ability of the sol-gel capillaries (and instrument) to successfully detect poisons with virtually no false alarms (false positives). This will be accomplished by defining performance in terms of receiver operator characteristics (ROC) curves. Limit of detection and reproducibility (capillary position) Initially we examined the limit of detection and reproducibility (false responses) for static measurements using MPA on the new OTC chem.L6 capillaries. A series of capillaries were prepared and MPA was measured at 1000, 500, 250, 125, 100, and 10 ppb. In each case 9 points spaced 0.375 mm apart along the length of the coated capillary were measured. All 9 spectra were averaged with no attempt to identify and discard outliers (Figure 5.1). As can be seen, the P-C stretching mode at 760 cm-1 is barely discernable at 125 ppb. Note also that the SERS relationship between the concentration and signal is not linear. This is expected since the signal levels off as the coverage on the available surface of the metal approaches a monolayer.

The performance of the SERS-active capillaries compared to the SERS-active vials was also evaluated. Previously we measured MPA at 10 ppm using the SERS-active vials with a predicted detection limit of 100 ppb based on the signal-to-noise ratio. Although this measurement was repeatable, detection was not obtained on every vial. Here, the situation is similar for sub ppm concentrations. The measurements for 50, 25, and 10 ppb gave sporadic results. The standard deviation for the 250 ppb 9-point values was ±0.036 or 12%, and as shown in Figure 5.2A, this was largely due to one point, which could be considered an outlier. Furthermore, at the lowest concentration several of the positions along the SERS-capillary produced MPA spectra at 10 ppb (Figure 5.2 B).

Fig.5.2. A) 250 ppb MPA measured in SERS-active capillary at 9 positions. Note 1 position produced a signal 50% greater than other positions. Inset: Magnified view of 760 cm-1 PC stretching mode for clarity. B) 10 ppb MPA measured at one position of a capillary. Conditions: 75 mW of 785 nm, 1 min each position, chemL6.

SERS MPA

ppb

1000

500

250

125

0

0.025

0.05

0.075

0 100 200 300 400 500 600 700 800 900 1000Concentration (ppb)

Peak

hei

ght

A B

Fig.5.1. MPA at 1000, 500, 250, and 125 ppb (μg/L) measured in SERS-active capillaries. Conditions: 75 mW of 785 nm at the sample, 9 averaged positions, 1 min acquisition each. Inset: plot of 760 cm-1 peak height vs. concentration, chemL6.


64

Limit of detection and reproducibility (batch-to-batch) As a further analysis of detection limits and reproducibility, we constructed concentration curves for MPA on chem. L6 using two different batches of capillaries. Serial dilution of the stock aqueous MPA solution (1 mg/mL) was used to prepare the following concentration standards in HPLC water (with measured pH):1000 (2.24), 100 (3.08), 50 (3.44), 10 (4.23), 5 (4.54), 1 (4.7), 0.5 (5.43), 0.25 (5.69), 0.125 (6.54), and 0.100 ppm. As shown in Figures 5.3 and 5.4, both batches yielded similar detection.

Fig.5.3. SERS and calibration plot of MPA (Batch 1): 50, 10, 5, 1, 0.5, and 0.25 ppm in water. Conditions: chem.L6, 80 mW, 785 nm, 1-min.

Fig. 5.4. SERS and calibration plot of MPA (Batch 2): 50, 10, 5, 1, 0.5, and 0.25 ppm in water. Conditions: chem.L6, 80 mW, 785 nm, 1-min. Limit of detection and reproducibility (ROC Curves) The military developed a method to quantify the likelyhood that a radio signal sent by an operator would be received. A receiver operator characteristic (ROC) curve is a graphical plot of the sensitivity vs. (1 – specificity) for a binary system used to establish the threshold that a signal can be confidently detected above background noise (for an overview see http://en.wikipedia.org/wiki/Receiver_operating_characteristic). In general, the goal is a 95% confidence level (here, the chemical concentration that can be detected 95% of the time). The confidence level is defined in terms of a separation value, K (which indicates statistically significant separation between a true and false response and is also a function of the ROC curve): K = [(M+S)-M]/σ = S/σ where M = mean response of the blank, S = the mean response of the sample, and σ = standard deviation of the signal. In signal detection theory, the responses to a blank are assumed to have a normal distribution with mean M and standard deviation σ (the noise distribution). When there is a detectable signal, the responses are assumed to be equal to the blank responses plus a constant S. Thus, the distribution of responses when the signal is present has a normal distribution with mean, M+S, and standard deviation, σ. A K value of 3.29 is equivalent to a 95% confidence level (95% signal detection with 5% false positives).

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0 10 20 30 40 50

Peak

Hei

ght

Concentration (ppm)

0

0.05

0.1

0.15

0.2

0.25

0.3

0 10 20 30 40 50

Peak

Hei

ght

Concentration (ppm)


65

Six chemicals were chosen for ROC curve analysis based on importance and chemical class representation. They were sodium cyanide, thiodiglycol, methyl phosphonic acid, fonofos, dichlorobenzoic acid, and sunset yellow. The latter chemical was included as the likely test chemical for Task 7. Each of these chemicals can be detected at 10 ppb with either the IEX or SPE pre-concentration method (Figures 5.5, sunset yellow not shown).

Each of the six chosen chemicals was measured at five concentrations (solid phase extracted), centered around the required detection limits (Table 5.1). SERS measurements were obtained at 9 equidistant points along the length of the sol-gel plugs using an automated positioning device, which minimized any user induced bias, to determine the concentration that can be detected with 95% confidence. For each analyte, the SPE pre-concentration and SER measurements were repeated at least 3 times to generate ~ 27 measurements per concentration for the ROC analysis. Table 5.1. Target Analytes, the concentration sample sets, and sol-gel chemistries employed for ROC analysis.

Analyte Required Concentration (ppb)

Concentration Samples Measured (ppb)

Sol-gel Chemistry

SPE Material

Cyanide 2000 100, 1000, 6000, 10000, and 50000

chem.L2 AIEX

Methylphosphonic Acid

5 0.5, 1, 5, 10, and 50 chem.L6 C8+AIEX

Thiodiglycol 100 10, 50, 100, 1000, and 10000 chem.L1_PEG DPA-6S Fonofos 10 1, 5, 10, 50, and 100 chem.L2 C18

Dichlorobenzoic Acid

10 1, 5, 10, 50, and 100 chem.L2 C18

Sunset Yellow Dye

10 1, 5, 10, 50, and 100 chem.L6 C18

ROC Analysis for Cyanide The CN stretch mode of cyanide at 2137 cm-1 was used to calculate relative signal response (Figure 5.6). The peak height for each response was calculated by subtracting the peak intensity at 2137 cm-1 from the baseline response at 2500 cm-1. The mean peak height of each substrate was calculated. Next the standard deviation of the mean response from all the substrates at a particular concentration was determined. These steps were followed for all other concentration and blank samples. The data and statistics for cyanide on chem.L2 at 100, 1000, 6000, 10000, and 50000 ppb (all samples pre-concentrated with SPE) are given in Table 5.2, along with the K values. Note: Signal detection theory and common statistical methods such as analysis of variance, require that the data have the same variance at different levels of response. Unfortunately the variances usually vary at different target concentrations. To normalize the different degree of variance between different concentration levels, a mean standard deviation value was used for the analysis.

A B C D E

Fig.5.5. SERS of A) NaCN, B) TDG, C) MPA, D) FON and E) DCBA, pre-concentrated at 10 ppb from HPLC water with IEX method A) and C), and SPE method B), D) and E). Conditions: 80 mW, 785 nm, 1-min.


66

Table 5.2. Measurement and ROC statistics for Cyanide using chem.L2 Capillaries. Concentration Number of

Capillaries Mean Peak

Height Standard Deviation

Mean Standard Deviation (σ)

K Value

Blank 4 0.006 0.0025 0.022 100 ppb 4 0.124 0.011 5.481

1000 ppb 4 0.302 0.058 13.503 6000 ppb 3 0.376 0.015 16.840 10000 ppb 3 0.919 0.025 41.287 50000 ppb 3 0.583 0.023 26.164

As mentioned before the ROC curve is a graphical plot of the sensitivity (y, probability of signal detection) vs. 1 – specificity (x, probability of false positives), and can be generated according to the following equations: y = Φ((M+S-t)/σ)), where Φ is the normal cumulative distribution function applied to the background, M, plus the signal, S, minus the threshold t, all divided by the standard deviation, σ. The threshold is defined as: t = M-σz, where z = Φ-1(x), and x = Φ((M-t)/σ), and was calculated from the background, signal, and standard deviation values given in Table 5.2 above, then the distribution functions were applied (x goes from 0 to 1, i.e. normalized). This is best done in a spreadsheet. The ROC curves for 100, 1000, 6000, 10000, and 50000 ppb Cyanide samples measured on chem.L2 substrates are shown in Figure 5.7. The orange line is 50000 ppb, blue is 10000 ppb, green line is 6000 ppb, yellow is 1000 ppb, and red is 100 ppb, while the black line is the probability of a random guess (50/50 chance).

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

1.1

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Prob

abili

ty o

f Det

ectio

n

Probability of False Positive

ROC Curve

All concentration samples passed the ROC test and cyanide was easily detected with a 95% confidence level at the required detection limit of 2000 ppb. The mean signal and the K-value at 50000 ppb are lower than that at 10000 ppb. This may be because the signal saturation and monolayer coverage are achieved at 10000 ppb, and a higher cyanide concentration may cause possible dissolution of the metal surface and subsequent substrate degradation, and hence diminishing the SER response.17

A B C D E F

Fig.5.6. Mean static SERS Response on chem.L2 for A) water blank, B) 100, C) 1000, D) 6000, E) 10000, and F) 50000 ppb Cyanide solutions, Conditions: 785 nm, 80 mW, 1-min. Note: Only raw data are shown above, which were used to construct ROC curve. No background subtractions were performed since the process can introduce additional variabilities. The major vibration mode of Cyanide SERS is at 2137 cm-1, so only the selected region 1500-2500 cm-1 is displayed above.

Fig.5.7. ROC curves (all overlapping) for 100 ppb (red), 1000 ppb (yellow), 6000 ppb (green), 10000 ppb (blue), and 50000 ppb (orange) Cyanide solutions on chem.L2. Black line is the probability of a random guess (50/50 chance). All concentration samples pass the 95% confidence ROC test.


67

ROC Analysis for Methyl Phosphonic Acid The P-C symmetric stretching mode of methyl phosphonic acid at 756 cm-1 was used to calculate relative signal response (Figure 5.8). The peak height for each response was calculated by subtracting the peak intensity at 756 cm-1 from the baseline response at 721 cm-1. The mean peak height of each substrate was calculated. Next the standard deviation of the mean response from all the substrates at a particular concentration was determined. These steps were followed for all other concentration and blank samples. The data and statistics for methyl phosphonic acid on chem.L6 at 0.5, 1, 5, 10, and 50 ppb (all samples pre-concentrated with SPE) are given in Table 5.3, along with the K values, and plotted as ROC curves (Figure 5.9).

Table 5.3. Measurement and ROC statistics for Methyl Phosphonic Acid using chem.L6 Capillaries. Concentration Number of




K Value

Blank 4 -0.001 0.0007 0.025 0.5 ppb 4 0.004 0.008 0.197 1 ppb 3 0.004 0.008 0.210 5 ppb 3 0.009 0.008 0.411

10 ppb 3 0.241 0.054 9.643 50 ppb 3 0.444 0.072 17.750

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

1.1

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Prob

abili

ty o

f Det

ectio

n


ROC Curve

Only the K values for 10 and 50 ppb samples are greater than 3.29. This means that the statistically significant detection limit at 95% confidence level is somewhere in between 5 and 10 ppb. The actual detection limit at 95 % confidence can be obtained by linear interpolation in K and log C as shown in Figure 5.10 (assumes a linear response). This yields a value of 6 ppb, very close to the required detection limit of 5 ppb.

A B C D E F

Fig.5.8. Mean static SERS Response on chem.L6 for A) water blank, B) 0.5, C) 1, D) 5, E) 10, and F) 50 ppb Methyl Phosphonic Acid solutions, Conditions: 785 nm, 80 mW, 1-min. Note: Only raw data are shown above, which were used to construct ROC curve. No background subtractions were performed since the process can introduce additional variabilities. The major vibration mode of MPA SERS is at 756 cm-1, so only the selected region 500-1000 cm-1 is displayed above.

Fig.5.9. ROC curves for 0.5 ppb (red), 1 ppb (yellow), 5 ppb (green), 10 ppb (blue), and 50 ppb (orange) Methyl Phosphonic Acid solutions on chem.L6. Black line is the probability of a random guess (50/50 chance). Note: 10 ppb blue line overlaps with 50 ppb orange line and hence is not visible.


68

y = 30.666x + 254.97

0

1

2

3

4

5

6

7

8

9

10

-8.4 -8.3 -8.2 -8.1 -8 -7.9 -7.8

ROC Analysis for Thiodiglycol (TDG) The C-S mode of thiodiglycol at 635 cm-1 was used to calculate relative signal response (Figure 5.11). The peak height for each response was calculated by subtracting the peak intensity at 635 cm-1 from the baseline response at 560 cm-1. The mean peak height of each substrate was calculated. Next the standard deviation of the mean response from all the substrates at a particular concentration was determined. These steps were followed for all other concentration and blank samples. The data and statistics for thiodiglycol on chem.L1_PEG at 10, 50, 100, 1000, and 10000 ppb (all samples pre-concentrated with SPE) are given in Table 5.4, along with the K values, and plotted as ROC curves (Figure 5.12). Note: TDG showed peak shifts at different concentration levels, where the observation window for the main vibration mode was anywhere in the range of 628-637 cm-1.

Table 5.4. Measurement and ROC statistics for Thiodiglycol using chem.L1_PEG Capillaries. Concentration Number of




K Value

Blank 4 0.004 0.0008 0.003 10 ppb 3 0.0048 0.002 0.161 50 ppb 6 0.0049 0.003 0.191

100 ppb 4 0.012 0.007 2.247 1000 ppb 3 0.051 0.001 13.528 10000 ppb 4 0.062 0.006 16.862

K=3.29

Log C

K

Fig.5.10. Plot of K versus Log Concentration (at K=3.29 (95% confidence), log C = -8.20713, and C = 6 ppb).

A B C D E F

Fig.5.11. Mean static SERS Response on chem.L1_PEG for A) water blank, B) 10, C) 50, D) 100, E) 1000, and F) 10000 ppb Thiodiglycol solutions, Conditions: 785 nm, 80 mW, 1-min. Note: Only raw data are shown above, which were used to construct ROC curve. No background subtractions were performed since the process can introduce additional variabilities.


69

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

1.1

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Prob

abili

ty o

f Det

ectio

n


ROC Curve

Only the K values for 1000 and 10000 ppb samples are greater than 3.29. This means that the statistically significant detection limit at 95% confidence level is somewhere in between 1000 and 100 ppb. The actual detection limit at 95 % confidence is 124 ppb, based on linear interpolation (Figure 5.13), again very close to the required detection limit of 100 ppb. ROC Analysis for Fonofos The aromatic ring breathing mode of fonofos at 991 cm-1 was used to calculate relative signal response (Figure 5.14). The peak height for each response was calculated by subtracting the peak intensity at 991 cm-1 from the baseline response at 968 cm-1. The mean peak height of each substrate was calculated. Next the standard deviation of the mean response from all the substrates at a particular concentration was determined. These steps were followed for all other concentration and blank samples. The data and statistics for fonofos on chem.L2 at 1, 5, 10, 50, and 100 ppb (all samples pre-concentrated with SPE) are given in Table 5.5, along with the K values, and plotted as ROC curves (Figure 5.15).

y = 11.281x + 81.216

2

3

4

5

6

7

8

9

10

11

12

13

14

-7.2 -7 -6.8 -6.6 -6.4 -6.2 -6 -5.8

Fig. 5.14. Mean static SERS Response on chem.L2 for A) water blank, B) 1, C) 5, D) 10, E) 50, and F) 100 ppb Fonofos solutions, Conditions: 785 nm, 80 mW, 1-min. Note: Only raw data are shown above, which were used to construct ROC curve. No background subtractions were performed since the process can introduce additional variabilities.

A B C D E F

K=3.29

K

Log C


Fig.5.12. ROC curves for 10 ppb (red), 50 ppb (yellow), 100 ppb (green), 1000 ppb (blue), and 10000 ppb (orange) Thiodiglycol solutions on chem.L1_PEG. Black line is the probability of a random guess (50/50 chance). Note: 1000 ppb blue line overlaps with 10000 ppb orange line and hence is not visible.


70

Table 5.5. Measurement and ROC statistics for Fonofos using chem.L2 Capillaries. Concentration Number of




K Value

Blank 4 0.0004 0.0007 0.017 1 ppb 6 0.005 0.003 0.287 5 ppb 3 0.019 0.008 1.120

10 ppb 5 0.046 0.027 2.776 50 ppb 3 0.121 0.011 7.258

100 ppb 3 0.149 0.050 8.959 The ROC curves for 1, 5, 10, 50, and 100 ppb Fonofos samples measured on chem.L2 substrates are shown in Figure 5.15. The orange line is 100 ppb, blue is 50 ppb, green line is 10 ppb, yellow is 5 ppb, and red is 1 ppb, while the black line is the probability of a random guess (50/50 chance).

Only the K values for 50 and 100 ppb samples are greater than 3.29. This means that the statistically significant detection limit at 95% confidence level is somewhere in between 50 and 10 ppb. The actual detection limit at 95 % confidence is 12 ppb, based on linear interpolation (Figure 5.16), which is again very close to the required detection limit of 10 ppb.

y = 6.4122x + 54.074

2

2.5

3

3.5

4

4.5

5

5.5

6

6.5

7

7.5

8

-8.1 -8 -7.9 -7.8 -7.7 -7.6 -7.5 -7.4 -7.3 -7.2

ROC Analysis for DCBA The ring breathing mode of DCBA at 991.57 cm-1 was used to calculate relative signal response (Figure 5.17). The peak height for each response was calculated by subtracting the peak intensity at 991.57 cm-1 from the baseline response at 960 cm-1. The mean peak height of each substrate was calculated. Next the standard deviation of the

Log C

K

K=3.29


Fig.5.15. ROC curves for 1 ppb (red), 5 ppb (yellow), 10 ppb (green), 50 ppb (blue), and 100 ppb (orange) Fonofos solutions on chem.L2. Black line is the probability of a random guess (50/50 chance). Note: 50 ppb blue line overlaps with 100 ppb orange line and hence is not visible.


71

mean response from all the substrates at a particular concentration was determined. These steps were followed for all other concentration and blank samples. The data and statistics for DCBA on chem.L2 at 10, 50 and 100 ppb (all samples pre-concentrated with SPE) are given in Table 5.6, along with the K values, and plotted as ROC curves (Figure 5.18).

Table 5.6. Measurement and ROC statistics for DCBA using chem.L2 Capillaries. Concentration Number of




K Value

Blank 3 -0.0046 0.0006 0.023 1 ppb 4 0.032 0.023 1.416 5 ppb 4 0.103 0.019 4.497

10 ppb 3 0.113 0.024 4.906 50 ppb 3 0.129 0.024 5.589

100 ppb 3 0.147 0.024 6.394

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

1.1

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Prob

abili

ty o

f Det

ectio

n


ROC Curve

All concentration samples except 1 ppb passed the ROC test. The actual detection limit at 95 % confidence is 2.7 ppb, based on linear interpolation (Figure 5.19), which is better than the required detection limit of 10 ppb.

A B C

D E

F

Fig.5.17. Mean static SERS Response on chem.L2 for A) water blank, B) 1, C) 5, D) 10, E) 50, and F) 100 ppb DCBA solutions, Conditions: 785 nm, 80 mW, 1-min. Note: Only raw data are shown above, which were used to construct ROC curve. No background subtractions were performed since the process can introduce additional variabilities.

Fig.5.18. ROC curves for 1 ppb (red), 5 ppb (yellow), 10 ppb (green), 50 ppb (blue), and 100 ppb (orange) DCBA solutions on chem.L2. Black line is the probability of a random guess (50/50 chance). All concentration samples except 1 ppb pass the ROC test. Hence 5 ppb (yellow), 10 ppb (green), 50 ppb (blue), and 100 ppb (orange) are overlapping.


72

y = 4.4075x + 41.084

1

1.5

2

2.5

3

3.5

4

4.5

5

-9.1 -9 -8.9 -8.8 -8.7 -8.6 -8.5 -8.4 -8.3 -8.2 -8.1

ROC Analysis for Sunset Yellow The aromatic ring vibration mode of sunset yellow at 1589 cm-1 was used to calculate relative signal response (Figure 5.20). The peak height for each response was calculated by subtracting the peak intensity at 1589 cm-1 from the baseline response at 1558 cm-1. The mean peak height of each substrate was calculated. Next the standard deviation of the mean response from all the substrates at a particular concentration was determined. These steps were followed for all other concentration and blank samples. The data and statistics for sunset yellow on chem.L6 at 1, 5, 10, 50, and 100 ppb (all samples pre-concentrated with SPE) are given in Table 5.7, along with the K values, and plotted as ROC curves (Figure 5.21).

Table 5.7. Measurement and ROC statistics for Sunset Yellow using chem.L6 Capillaries. Concentration Number of




K Value

Blank 4 -0.002 0.0011 0.021 1 ppb 3 0.126 0.015 6.058 5 ppb 3 0.129 0.017 6.178

10 ppb 3 0.171 0.013 8.209 50 ppb 3 0.225 0.057 10.729

100 ppb 4 0.234 0.021 11.174

A B C D E F

Fig.5.20. Mean static SERS Response on chem.L6 for A) water blank, B) 1, C) 5, D) 10, E) 50, and F) 100 ppb Sunset Yellow solutions, Conditions: 785 nm, 80 mW, 1-min. Note: Only raw data are shown above, which were used to construct ROC curve. No background subtractions were performed since the process can introduce additional variabilities.

Fig.5.19. Plot of K versus Log Concentration (at K=3.29 (95% confidence), log C = -8.574929098, and C = 2.7 ppb).

K=3.29

Log C

K


73

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

1.1

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Prob

abili

ty o

f Det

ectio

n


ROC Curve

All concentration samples, including 1 ppb, passed the ROC test and sunset yellow was easily detected with a 95% confidence level at the required detection limit of 10 ppb. Limit of detection and reproducibility (ROC curves with raster spectral acquisition) As shown above, some, but not all analytes were measured at the required detection limits 95% of the time. In an effort to improve sensitivity, we examined how the capillary could be better analyzed. This led to the averaging of 9 discrete points along the capillary, which helped and reduced the number of false positives. Nevertheless, as shown in Figure 5.2, some points produce more intense signals than other. These are attributed to “hot spots” , metal structures that produce extraordinary enhancements. However, these hot spots could be missed using the current 9-point analysis. Consequently, a method was developed to scan back and forth (raster) along a predefined length of capillary. A software program was written to move the capillary above a fixed laser spot using a Conix plate reader. The raster approach is demonstrated for 500 ppb CN flowing through a chem.L3 capillary sol-gel plug (Figure 5.22). This represents a concentration that could NOT be detected without the use of SPE. In fact, measuring and averaging the signal for 9 discrete points on the same capillary before and after performing the raster measurement does NOT produce a detectable CN peak.

In the following experiment we investigated the performance of Raster program acquired data for ROC analysis. Discrete 9-point measurements have shown that the ROC 95% detection limit for MPA is 6 ppb, close to the required detection threshold of 5 ppb. Hence, an experiment was designed to evaluate if we can improve the ROC detection limit beyond the required concentration. A 50 mL sample solution of MPA at 5 ppb was pre-concentrated using the SPE procedure (see Table 1.5, Task 1). The pre-concentrated MPA solution was then loaded onto our chem.L6 capillaries, and the capillary was measured at 9 discrete points and scanned (raster) along the same 1 cm length. The 9 1-min points were averaged to produce a spectrum for analysis, whereas, the raster collected spectrum was used. However, it should be noted that the raster was only a 3 minute collection. The spectra were analyzed as before to generate the ROC curves (721 cm-1 baseline intensity subtracted from the 756 cm-1 peak

Fig.5.21. ROC curves (all overlapping) for 1 ppb (red), 5 ppb (yellow), 10 ppb (green), 50 ppb (blue), and 100 ppb (orange) Sunset Yellow solutions on chem.L6. Black line is the probability of a random guess (50/50 chance). All concentrations pass the ROC test.

Fig.5.22. SERS of CN at 500 ppb on chem.L3, 5-min scan-sweep spectrum from the Raster program measured over a 3-mm region, after flowing sample for 5-min (at 5 mL/min). Conditions: in HPLC water, 80 mW, 785 nm. Note: previous measurements (on 9 equidistant points) in the same 3-mm region after flowing revealed no signal.


74

intensity to determine the peak height). It was found that the peak intensity obtained for the rastor spectrum was significantly higher than the averaged 9-points (0.082 vs 0.009), indicating that hot-spots were missed for the latter approach. Furthermore, the capillary reproducibility (standard deviation) was found to be little better (0.018 vs 0.025). As a result, the K value significantly improved by more than a factor of 10 (4.55 vs 0.411 (discrete point measurements)), and easily passed the 95:5 critical K value 3.29. The following Figure 5.30 shows a comparison of the 5 ppb ROC curves obtained using A) Raster scan program (green) and B) the 9 discrete point measurements (red). It should be cautioned, that these same improvements may not occur in every case, especially if the discrete measurements include the same hot-spots. Finally, it should be noted that a program could be written to seek hot-spots and measure at the “hottest” location. This could reduce analysis time and minimize false-negatives.

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

1.1

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Prob

abili

ty o

f Det

ectio

n


ROC Curve

Overall Summary: 1) SPE pre-concentration coupled with SERS sensing helped to achieve detection of four target analytes at concentration levels below the desired values with 95% confidence. For MPA, the 95:5 detection limit was 6 ppb, just 1 ppb above the desired detection limit (5 ppb), and for TDG the detection limit was 124 ppb, 24 ppb above its desired detection limit (100 ppb). 2) Preliminary studies show that the existing sensitivity and reproducibility of the SER-active capillaries can be further improved by using a Raster scan program. This provided better ROC performance and the required concentration level of MPA (5 ppb) was detected at 95% confidence.

Fig.5.23. ROC curves for 5 ppb (red) obtained from discrete point measurements, and 5 ppb (green) obtained using Raster scan program, Methyl phosphonic acid solutions on chem.L6. Black line is the probability of a random guess (50/50 chance).


75

Task 6. Design and Test Sample System. This overall objective of this task is to design and build a fully functional system with a universal interface. This will be accomplished by selecting sampling interface components, designing and building the capillary interface, and integrating the sample system to the Raman analyzer. We began to design the sampling system in two stages. In the first stage, we selected the major components that would be required based upon the results of Tasks 1, 2 and 4. This included an SPE/AIEX sorbent to pre-concentrate the chemicals to achieve detection at 10 ppb, pressure regulating valves so that the sol-gel sensor would not be subjected to extreme flow conditions, and physical filters to remove any debris in the water samples. In the second stage we began to search for a platform that would enable us to integrate all of the above components with our SER sensor and Raman analyzer. The Parker IntraFlow Modular System based on the New Sampling/Sensor Initiative (NeSSI) concept appeared to be an ideal option. During several telephonic and face to face meetings (at trade shows etc.) with the Parker sales and design engineers, it was assured to us that Parker would indeed be able to design and manufacture such a system in a timely manner. (This turned out not to be the case, and delivery of required components were delayed by more than 1 year!) The Parker IntraFlow System employs 1.5x1.5x0.25” SS building blocks with standardized hole and thread placement (ANSI/ISA 76.00.03 specifications). This allows mounting valves, pressure regulators and gauges, etc. The system consists of three layers, a substrate to attach the building blocks, the building blocks that contain various flow pattern options, and the control/measuring devices. The building blocks allow directing flow to and from the devices. Figure 6.1 shows a simplified block diagram of the system. Each block represents a NeSSI substrate (with the 2/3 sized blocks being end-pieces). The sampling system consists of two parts, a stream conditioning line and a stream/reagent switching line. The conditioning line is substrates 1-8, and is designed to control the flow of a water stream (drinking water quality) that enters at 5-10 gal/min and 20-30 psi and regulate it to 5 ml/min at 20 psi. The stream/reagent line is substrates 9-18. The switching is performed by three 3-way valves (9, 10 and 13). The threaded plugs that fit IF-4G5-END-SS will be machined, so we can connect the Extractor (SPE) and SERS Sensors.

Operational sequence of the system is as follows: 1) the water sample is flow and pressure regulated (1 through 8) prior to passing through the Extractor (sample collection) for 5 minutes into a waste bottle (9-14), 2) Valves (9, 10, 13) switch the water stream to flow (15) into a sample bottle, and allow elution solvent (e.g. methanol) to flow (16) through the Extractor into SERS Sensor and into waste for 10 seconds (16, 10-13, 17, 18), and 3). The valves switch back to their original position loading the next sample on the Extractor, while SERS Sensor is read.

Fig.6.1. Illustration of The Parker-Hannifin building blocks showing the flow pattern of the proposed sampling system. This Figure is Proprietary. A list of mounts and attachments to control the flow are as follows: Mounts: 1, 11, 12, 14, 15, 16, 17 and 18 are IF-4G5-END-SS, 2,3,4,7 and 8 are IF-A1B3, 5 and 6 are IF-E-A1B4 or IF-EA-A1B4, 9,10 are IF-TR-A1B23 and 13 is IF-TL-A1B34. Attachments: 2: IF-B2LJ2-V-SS (ball Valve), 3 and 8: IF-FT4-1-SS (In-Line Filter), 4: IF-CO4-1/3-V-SS (Check Valve), 5: SMSQMICRO3010RING (pressure reducing regulator), 6: GAUGE-9118128 and 7: IF-HOL-V-SS-K (metering valve).

SPE


76

The above flow sequence was carefully examined and designed to eliminate the need for an ion-retardation (IR) resin in the sampling system. As seen from the Task 4 results, the use on an IR resin may extract CN, and the nerve agent hydrolysis products to some extent. In such a scenario, the IR resin would have to be tested to ensure the presence or absence of the chemicals of interest. This would have involved extensive method development to elute chemicals from the IR resin, which is very difficult as the IR resin is not as efficient as a regular AIEX resin. Thus selectively eluting CN or MPA without co-eluting interfering ions is very complex without the use of precise elution solvent dispensers, automated pH and conductivity sensors. This in turn would have caused difficulties with the entire design of final sampling system, with considerable cost escalation. Hence we decided on the above operational flow sequence. For this sequence, the sample passes through the Extractor (SPE) cartridge and subsequently flows into a waste bottle. If C8 or C18 is used as the SPE material, it will only retain organic non-polar chemicals from the water (e.g. pesticides, TICS etc) due to the hydrophobic interactions, while allowing the small inorganic ions (interferents) to pass through unhindered into the waste bottle, thus never contacting the sol-gel SER sensor. If an anion exchange resin (AIEX) is used (for CN and nerve agent hydrolysis products), it will retain inorganic ions like CO3

-2 , phosphates, Cl- as well. However a proper choice of the eluting solvent and type of the AIEX resin helped us to selectively elute CN and nerve agent hydrolysis products over these commonly found interferents in water. The order of selectivity for the AIEX resin we used is: HPO4

- <CO3 - < Cl- < ClO-1 < CN < PO3

-.18 and 19 Thus this resin has the highest selectivity for the nerve agent hydrolysis products. This is due to the interaction between the PO3

- group of the nerve agent and the functional group of the resin -(CH3)N+. If the water sample contains all of the above inorganic chemicals, PO3

- and CN will be selectively retained over the others. The un-retained chemicals in this case also will flow into the waste bottle thus never contacting the sol-gel SER sensor. The retained chemicals can then be eluted off the AIEX as shown later in this Task. The above process thus allowed us to utilize the flow system without using an IR resin. Once the system was conceptualized, we had a telephone conference with Parker, after which Parker prepared a drawing according to the design and a formal quote of $4000 for the system. Figure 6.2 shows the drawing of the Parker sampling system.

Fig.6.2. Drawing of the Parker Sampling System. This Figure is Proprietary. The field connector end access are labeled as 1,10,12,15 and 17, the pneumatic diaphragm valve is labeled as 2, the filters are labeled as 3 and 8, the check valve is labeled as4, the pressure reducing regulator is labeled as 5, the pressure indicator meter is labeled as 6, the metering valve is 7, the 3-way ball valves are labeled as 9, 11 and 14, the SPE cartridge is labeled as 13 and the SERS Sensor is labeled as 16.

Fig.6.3. Drawing of the modified Parker Sampling System with the 2-way solenoid valves. This Figure is Proprietary.


77

After the drawing was approved by RTA, Parker began work on manufacturing the components of the sampling system. After a few weeks, RTA was informed of Parker’s inability to supply the 3-way ball valves in a timely manner (apparently this was the first electrically controlled 3-way NeSSI valve that they produced). In an effort to speed up the process, Parker offered to supply 2-solenoid valves at no extra cost. This forced RTA to modify the design of the sampling system to incorporate the solenoid valves. Figure 6.3 above shows the new design of the sampling system. A few weeks after the modified design was approved by RTA, we began to receive a few components of the system. Since RTA had already paid for the manual 3-way ball valves, Parker shipped them to us with the entire system, while we were waiting for the solenoid valves, so that we could start testing the system. Figure 6.4 shows a picture of the flow system with all components. The components are numbered and described as follows:

To test the critical components of the sample flow system (#2 through #9) we attached it to our laboratory faucet through a field connector (#1) and flowed water through it (see Figure 6.5). The water inlet flow entering the system was about 142 mL/sec at a pressure in excess of 30 psi. The water pressure and flow rate were regulated to 30 psi and 5 mL/min, respectively, by the system components as it came out of a field connector (#10). No leaks were detected, and the flow regulator held constant pressure throughout the experiment which demonstrated that the components functioned correctly. A SERS-active capillary (chem.L2) was attached to the field connector (#10) and water was flowed through the capillary at 5 mL/min for 30 minutes. Under these conditions the sol-gel plug did not detach and the silver particles were retained in the sol-gel. This capillary was then tested for SERS activity by introducing 100 ppm fonofos (see Figure 6.6).

5

4

3

11

2

10

20

18

9

14

1

12 13

15

17

8

6

7

16

19

Fig.6.4. Photograph of the Parker Sampling System with manual valves for initial testing. This Figure is Proprietary. 1, 10, 14, 18 and 20 are field connector end access with 1/4 inch compression fittings. 2, 9, 11, 12, 13, 15 and 17 are manual 2-way ball valves. 3 and 8 are 0.5 micron filters. 4 is a check valve. 5 is a pressure reducing regulator. 6 is a pressure indicator meter (0-100 psi). 7 is a metering valve and 16 and 19 are flow substrates.


78

This above test demonstrated that the sampling system was functioning as planned and more importantly, chem.L2 could withstand tap water flow for 30 minutes at 5mL/min, without any effect on the SER activity. After this test, we began experiments to incorporate the SPE sorbent into the IntraFlow sampling system. For this the sorbent (C18 or anion exchange resin) was placed inside a 3/8 inch plastic tube supported by frits at both ends. The plastic tube was connected above the usual flow plain of the substrates, so that replacing the sorbent in the tubing after each experiment is easily done. 1/8” tubing was attached to the field connector end access to introduce the sample into the flow system. Figure 6.7 shows the SPE incorporated into the Parker sampling system.

Fig.6.5. Experimental setup of the Parker IntraFlow sample flow system attached to a faucet for flowing water.

Water inlet from the lab faucet.

Pressure indicator showing water pressure from the faucet entering the sample flow system.

SERS capillary attached to the sample flow system .

Fig.6.6. SERS of 100 ppm fonofos obtained on SERS sol-gel capillary (chem.L2) after flowing tap water for 30 minutes at 5mL/min. Conditions: 80 mW, 785 nm, 1-min.


79

After this an experiment was performed to test the ability of the incorporated SPE cartridge to extract chemicals from a flowing water sample. For this a 500 mL sample solution of MPA was prepared at 10 ppb in (ion-retardation conditioned) water sample EWS-2008-002 obtained from Kensico water reservoir. Using a syringe pump the sample was introduced into the flow system at a flow rate of 9 mL/min through the field connector and flowed through the substrates containing the conditioning line (micron filter, check valve, pressure reducing regulator, pressure indicator meter and a metering valve) and through the SPE sorbent (anion exchange) and into the waste bottle. After this the sorbent was washed with 10 mL of HPLC water to remove the sample matrix and then the extracted MPA was eluted off the sorbent with 5 mL of methanol containing 0.1 M HCl (at a flow rate of 1mL/min) and collected in a sample bottle. A syringe was used to load a SERS-active capillary, and then measure the spectrum. A similar procedure was followed for extracting 10 ppb sunset yellow from the water sample, except that C18 was used as a SPE sorbent and 13% isopropyl alcohol (IPA) was used as the eluting solvent. Figure 6.8 shows the SERS of MPA and sunset yellow. These experiments showed that the SPE sorbent was successfully incorporated into the Parker IntraFlow system and MPA and sunset yellow at 10 ppb could be extracted and detected in Kensico water sample.

SPE Sorbent

MeOH for Elution

Sample

Eluted Analyte Collection Bottle

Fig.6.7. SPE sorbent incorporated into the Parker IntraFlow sampling system.

Fig.6.8.SERS of A) 10 ppb MPA and B) 10 ppb sunset yellow extracted from water sample EWS-2008-002 from Kensico water reservoir and detected using the SPE sorbent incorporated into the Parker IntraFlow sampling system. Conditions: 80 mW at 785 nm, 1-min, chem.L6.

A B

Elution solvent MeOH

MPA 756 cm-1


80

In the following weeks, we received the solenoid valves from Parker for the IntraFlow sample system and began working towards incorporating them in the system. These valves would allow building an automated system controlled by software for a field unit, as opposed to the manual valves initially used. Figure 6.9 shows a picture of the 24 VDC 10 Watt solenoid valves.

In this period we also investigated National Instruments (NI) instrumentation flow control systems for a single interface to configure solenoid valves to control the pneumatics, and a logic controller for the entire flow system. For the working of the valves, they were wired to the NI USB-6525; both powered by a Triad 24 VDC /150 Watt power source. The NI 6525 contains solid state relay switches which will trigger each solenoid valve to open or close and thus direct the flow. This lets the system be completely controlled and run by software (described below) allowing automatic flow switching at pre-determined set points or as necessary. Figure 6.10 shows the NI 6525 USB 8 Channel I/O with built in solid state relays (left) powered with a 24V/150W power source (right), while Figure 6.11 shows the solenoid valves now incorporated into the Parker IntraFlow sampling system.

To control the solenoid valves, a software routine was designed to drive the NI 6525 device that controls the valves. The software driver for the NI 6525 device operates as follows: On start-up, ALL valves are switched OFF (Figure 6.12, Closed). This prevents the system from starting up in an unwanted state. Once the software starts, the user can use the switch to change the flow. There are two flow paths in the operational sequence as shown in Figures 6.13 and 6.14. In the first flow path, (Figure 6.13) the water sample passes through the Extractor (sample collection) for 5 minutes and then into a waste bottle. In the second flow path, (Figure 6.14) valves switch the water stream to flow into a sample bottle, and allow methanol to flow through the Extractor into SERS Sensor and into waste bottle.

Fig.6.9. Picture of the 24 VDC/10 Watt solenoid valves supplied by Parker.

Fig.6.10. NI 6525 USB 8 Channel I/O w/ built in solid state relays (left) powered with a Triad 24V/150W power source (right)

Fig.6.11. Parker IntraFlow System with electronic solenoid valves


81

After this we performed a preliminary test of the entire system (hardware and software) to ensure that it was performing as designed. For this the sampling system was again connected to a high-pressure water faucet in our laboratory which allowed tap water to be flowed continuously into the system at a rate of 142 mL/sec with a pressure in excess of 30 psi. The water pressure and flow rate were regulated successfully by the components to 30 psi and 5 mL/min, respectively. During the testing, no leaks were observed in this system. The entire process was completely controlled and run by software as described above. Figure 6.15 shows the setup of the Parker IntraFlow System with the computer interface.

Fig.6.12. Image of the Process Control software at system startup.

Fig.6.13. Process Control software showing the Flow from the Sample Inlet through the Extractor into waste bottle.

Fig.6.14. Process Control software showing the Flow from the methanol bottle through the Extractor into SERS Sensor and into waste bottle


82

After this we began actual measurements of the 5 target chemicals using the Parker IntraFlow system. In the first experiment, we extracted thiodiglycol (TDG) at 1000 and 100 ppb from a water sample obtained from the Kensico water reservoir using the Solid Phase Extraction (SPE) sorbent integrated into the automated Parker IntraFlow sampling system. The entire process was completely controlled and run by software as described above. The experimental procedure was as follows: A normal phase SPE sorbent (DPA-6S) was packed inside a 3/8 inch plastic tube connected to the IntraFlow system. A 600 mL sample solution of TDG was prepared at 1000 ppb in water sample EWS-2008-002 obtained from Kensico water reservoir. Using a syringe pump the sample was introduced into the flow system at a flow rate of 9 mL/min and flowed through the substrates containing the conditioning line (micron filter, check valve, pressure reducing regulator, pressure indicator meter and a metering valve) and through the SPE sorbent and into the waste bottle. After this the sorbent was washed with 10 mL of HPLC water to remove the sample matrix and then the extracted TDG was eluted off the sorbent with 4 mL of methanol at a flow rate of 1mL/min and collected in a sample bottle. A syringe was used to load a SERS-active capillary, and then measure the spectrum. A similar experiment was performed with a 100 ppb TDG sample as well. Figure 6.16 shows the SERS of TDG at 1000 and 100 ppb from water sample obtained from Kensico reservoir.

Fig.6.15. Picture of the Parker IntraFlow System with electronic solenoid valves with the computer interface.


83

Using a similar experimental set-up as described above, we tested the 3 remaining target chemicals using the flow system. Figure 6.17 shows a stack plot of 3,5-dichlorobenzoic acid (DCBA), fonofos (FON) and cyanide (CN) extracted from water sample EWS-2008-002 from Kensico water reservoir and detected using the SPE sorbent incorporated into the Parker IntraFlow sampling system (see figure caption for specific SPE conditions).

NOTE: for the above experiments, the SER capillary was not incorporated into the system. The chemicals were eluted in a sample bottle and a syringe was used to draw the sample into a SERS-active capillary, which was then placed on the Raman analyzer plate reader and then the spectrum was measured. After this, we began working towards incorporating the SER capillary into the Parker IntraFlow sampling system (Figure 6.18). For this 1/8” Tygon tubing sleeves were coated with a fast setting epoxy and then inserted into the 3/8” diameter tubes used to connect the Parker flow plates. The connector tubes have embedded o-rings which fit snugly with the Tygon tubing after the epoxy has set in. After this the SERS sol-gel capillaries were attached to the Tygon tubing. Once the capillaries were interfaced, we performed a preliminary test of the entire system to ensure that it functioned as designed. For this the system was attached to the outdoor water line at RTA. This line has a regulated pressure of 60 psi and a flow rate of 2 gallons/min. The water pressure and flow rate were regulated successfully by the components to 30 psi and 5 mL/min, respectively. During the testing, no leaks were observed in this system. The entire process was completely controlled and run by software as described previously. The ability of the sample system to slow flow and reduce pressure and maintain constant flow through a SERS capillary was thus demonstrated.

Fig.6.16. SERS of TDG at A) 1000 ppb and B) 100 ppb extracted from water sample EWS-2008-002 from Kensico water reservoir detected using the SPE sorbent incorporated into the Parker IntraFlow sampling system. Conditions: 80 mW at 785 nm, 1-min.

A B

Fig.6.17. SERS of A) DCBA, B) FON, and C) CN at 100 ppb on chem.L2, extracted from water sample EWS-2008-002 from Kensico water reservoir detected using the SPE sorbent incorporated into the Parker IntraFlow sampling system. Conditions: 80 mW at 785 nm, 1-min. SPE conditions: A) C8, MeOH elution, B) C18, MeOH elution and C) anion-exchange, 0.01 M HNO3elution.

A B C


84

After this we also integrated the Raman analyzer with the IntraFlow sampling system using a preliminary fiber optic probe interface. For this the probe is set up on top of the SERS sol-gel capillaries so that the capillaries can be easily removed after each measurement without having to disturb the probe alignment. Figures 6.19 and 6.20 show the sampling system with the probe attached to the Raman System.

SERS capillary

SPE Cartridge

Fig.6.18. SER capillary and Solid Phase Extraction (SPE) sorbent incorporated into the Parker IntraFlow sampling system.

Fig.6.19. Parker IntraFlow sampling system incorporated with the Raman System attached to a fiber optic probe.

Fiber optic probe attached above the SERS capillary.

Prototype Raman Analyzer.


85

After this we tested the Parker IntraFlow sampling system with the incorporated SER capillary. The experimental procedure was similar to that described previously. Briefly, 750 mL sample solution of MPA was prepared at 75 ppb water sample EWS-2008-002 obtained from Kensico water reservoir. Using a syringe pump the sample was introduced into the flow system at a flow rate of 5 mL/min through the field connector and flowed through the substrates containing the conditioning line (micron filter, check valve, pressure reducing regulator, pressure indicator meter and a metering valve) and through the SPE sorbent and into the waste bottle. Then the sorbent was washed with 10 mL of HPLC water to remove the sample matrix and then the extracted MPA was eluted off the sorbent into the SER active sol-gel capillary with 8 mL of methanol containing 0.1 M HCl (at a flow rate of 0.5mL/min). Figure 6.21 shows the SER spectrum of the MPA eluted on the SER substrate.

During the next few weeks we continued to test the system with the target chemicals. It was noticed that the primary analytes (sp. TDG, DCBA, MPA, FON and CN) could only be detected at 100 ppb, and not the required 10 ppb

Fig.6.20. Close up shot of the probe over the SER substrate.

Laser Focus Point

Fig.6.21. SERS of 75 ppb MPA extracted from water sample EWS-2008-002 from Kensico water reservoir detected using the SPE sorbent and SER capillary incorporated into the Parker IntraFlow sampling system. Conditions: 80 mW at 785 nm, 1-min, chem.L6.


86

with the SER capillary incorporated into the Parker IntraFlow sampling system. It was suspected that this was due to the extended flow path from the solvent elution bottle to the SPE tube to the SERS substrate. Initially the extracted sample is eluted off the SPE cartridge as a thin plug (band). But as it flows through the NeSSI channels the sample broadens and becomes diluted. It was found that it takes about 8-10 mL volume of the elution solvent from the entry point of the SPE cartridge to the SER substrate. As a reminder, all of the samples could be detected when the SERS capillary was not integrated into the system, substantiating our suspicions. In an effort to address the sample dilution problem, we evaluated the flow rate of the sample and the eluent through the SPE cartridge. Slower flow rates, 0.5 and 0.25 mL/min, were examined assuming that it would increase the retention of the analytes on the SPE, and increase sample recovery during elution. It was found that using a sample flow rate at 0.5 mL/min and by using an elution flow rate of 0.25 mL/min, SERS was obtained at 50 ppb for MPA and sunset yellow (Figure 6.22), and at 75 ppb for DCBA and fonofos (Figure 6.23). However, the required lower concentrations of 10 ppb remained undetectable.

SUMMARY: A functional Parker IntraFlow sampling system was designed, built and successfully tested. The Parker IntraFlow system appears to be a good platform for integrating the SER sensors for use in harsh field conditions. Some limitations were encountered during the experiments which are listed below. Limitation 1: The extended flow path from the solvent elution bottle to the SPE tube to the SERS substrate results in the extracted sample band getting diluted by the elution solvent. Thus using the Parker sampling system in the fully automated mode, we could not detect the selected chemicals (CN, MPA, TDG, FON and DCBA) at 10 ppb. Limitation 2: Results from the previous tasks dictate the use of two different types of SPE sorbents. SPE sorbents like C8 and C18 for the various non-polar chemical, toxic industrial and pesticide agents of concern while the AIEX sorbents for the pre-concentration of more polar CN and nerve agent hydrolysis products. Thus both the SPE sorbents must be used in parallel or simultaneously when testing for poisons in water. Both limitations can be addressed by incorporating some SPE material in the SERS-active sol-gel. This would allow using both SPE materials, as well as allow re-concentrating the diluted chemical. As an alternative to this chemical solution, an engineering solution is also possible. Again, both limitations can be addressed by employing more traditional flow injection analysis designs as opposed to the NeSSI platform. Such FIA designs use HPLC pumps and 1 mm diameter tubing that minimize tubing volumes and allows for switching multiple streams, i.e. between the two SPEs. Analysis would be performed in sequence. Pursuit of either approach is beyond the current funding.

Fig.6.22. SERS of 50 ppb A) MPA and B) Sunset yellow extracted from water and detected using the SPE sorbent, SER capillary and Raman analyzer incorporated into the Parker IntraFlow sampling system at a sample flow rate of 0.5 mL/min and an elution flow rate of 0.25 mL/min. Conditions: L6 SER substrate, 80 mW at 785 nm, 1-min.

A B

Fig.6.23. SERS of 75 ppb A) DCBA and B) FON extracted from water and detected using the SPE sorbent, SER capillary and Raman analyzer incorporated into the Parker IntraFlow sampling system at a sample flow rate of 0.5 mL/min and an elution flow rate of 0.25 mL/min. Conditions: as in figure 6.25, except L2 SER substrate.

A B


87

Task 7. Field Tests. The overall objective of this task is to demonstrate real-world capabilities. This will be accomplished by performing two field tests, measurements of HD and VX in flowing water samples at the U.S. Army’s Edgewood Chemical Biological Center and measurements of a test dye added to New York’s Kensico Reservoir water supply. At the end of the first year of this Phase II project we began discussions with U.S. Army’s Edgewood Chemical Biological Center (ECBC) to perform measurements of HD and VX at their facilities. Preliminary field tests at the ECBC facilities in Aberdeen, MD (under the direction of Dr. Steve Christesen and Dr. Augustus W. Fountain) were tentatively scheduled for May 07. The purpose was to demonstrate improvements made in the capillary sol-gel chemistry (sensitivity) and coating procedures (reproducibility). We had also planned to measure the SERS of the G-series of nerve agents at this time for the spectral library in Task 3. Due to the busy work schedule at the ECBC facilities, we were instead asked to just ship our SERS substrates as opposed to us personally going to ECBC to perform the measurements. On 10/16/07, we provided the ECBC (Dr. Steve Christesen and Dr. Jason A. Guicheteau) with 40 vials coated with chem.L6 for preliminary evaluation. Also 15 of our improved chem.L1 standard vials (HCl washed) were sent to obtain SERS of the G-series of nerve agents for the spectral library. These vials are similar to those used previously to measure VX, EA2192 and HD. VX and HD were to be measured on capillaries at a later date. Unfortunately, the planned measurements were not performed. Instead the vials were used to measure methyl phosphonic acid and phenylalanine. Once we were aware of their decision, we used vials prepared at the same time to measure MPA at 1000 and 250 µg/L (ppb, Figure 7.1), so that we could compare results (they used a different Raman analyzer). We requested these measurement results (numerous times, even as recent as May 2010), but they were never supplied. Nevertheless, we continued to improve the capabilities of our SERS capillaries for a future ECBC visit. Again, ECBC personnel agreed to a tentative date (04/01/2008) to measure VX and HD on our capillaries using the IEX and SPE methods. Once again, the trip was postponed. In follow-up discussions with ECBC personnel, we were informed that the focus was on bioagents, and chemical agent work was put on hold. This situation has not changed as of May 2010.

Field measurements at New York City’s Kensico Reservoir were originally to be conducted towards the end of the Phase II program, when all of the other tasks were completed. Nearly all of the 1.2 billion gallons of water supplied daily to New York City and its neighboring counties come from the Catskill, Delaware, and Croton Watersheds. And virtually all of this water passes through the Kensico Reservoir via aqueducts. Due to the delay of the sample system, these measurements were postponed by over a year. Nevertheless, communication with Dr. Yves Mikol, Director of the Early Warning Surveillance Program of the City of New York Department of Environmental Protection (CNYDEP), continued, and Dr. Farquharson visited Dr. Mikol and John Canning at the Kensico Reservoir February 4, 2008). During the visit flow charts showing the water system and the analyzer house were discussed. The following was learned. A sample loop is in place located two floors above the dam in a small house (a new one was being built at the time). A 1 horsepower pump drives the water up to the sample loop through a 1” pipe. The water flow and pressure are modest at 20-30 psi and 5-10 gal/min. The sample loop has several in-line and at-line monitors, the latter through T

Figure 7.1. SERS of A) 1000 and B) 250 µg/L (ppb) MPA measured using SERS vials produced the same time as the vials that were sent to the ECBC (same batch). Conditions: 80 mW of 785 nm, 1-min. Measurements were 10 days after production. A

B


88

connectors. The analyses include conductivity, pH, temperature, and turbidity. RTA could tie into one of these sample lines. As described in Task 5, the pressure is okay, and flow will be reduced to 5 mL/min using the sample system. Since the water will be slightly contaminated by the 0.1 ml methanol used per analysis (~ 1 mL/hour) , it was decided to collect the waste in a 5 gallon container. It was determined that at 5 mL/min, it would take 67 hours to fill the container (5 gallons = 18,950 mL), which was acceptable. An additional concern was the 1 week frequency that the SERS capillary (and SPE cartridge) would have to be replaced. This replacement period could be extended, if the SERS system only samples when other analyzers (e.g. pH, conductivity) produced a cautionary alarm (yellow), instead of monitoring the water continuously. This would be considered when implementation took place. Other concerns were raised in regard to fouling of the SPE and SERS capillary due to particulate matter, bacteria, and fuels from highway spills (albeit the latter is considered rare). The sample system has an in-line particulate filter, but fouling by bacteria is unknown, and would have to be studied. During this meeting, the use of a dye for testing the system was also discussed. In the past, CNYDEP used Gatorade or Sprite. They do not add the dye directly to the reservoir, but into the sample loop that enters the analyzer house using a “mixing box”. Upon returning to RTA we learned which food dyes were used in these consumer drinks. Primarily, Gatorade employs Allura Red, Tartrazine, and Sunset Yellow in their most common drinks. Consequently, these dyes, as well as rhodamine6G, crystal violet, patent blue, and erythrosine B were studied (see task 3). Ultimately we selected Sunset Yellow for potential field tests. An additional purpose of the visit was to collect various samples, untreated water from the Kensico Reservoir Monitoring Station and treated (chlorinated) water from a downstream location, for the initial tests to be performed at RTA. Four water samples were provided labeled EWS -2008-002, EWS -2008-03, EWS -2008-04 and EWS -2008-005. These samples were used for Tasks 4-6, as described earlier. Finally, during this program, the software user interface was continually updated. The software consists of several components: the software that operates the sample system (Figure 6.12), the software that scans the capillary (raster), the software that operates the Raman analyzer (not shown), the software that performs the unknown identification (Figure 3.98), and the software that determines the concentration (Figure 5.7). A user interface was designed that controls the analyzer and performs the unknown identification was developed for our commercial RamanID product during this program (Figure 7.2A). A user interface software program (Raman Water Security) that encompasses all of the required programs is shown in Figure 7.2B. It employs the color codes used for the Homeland Security Advisory System. It also includes most of the same information provided by the RamanID software, including the National Fire Protection Association diamond commonly used by emergency responders. The Set-Up button links to the sample system software, while the More button links to the unknown identification software. The raster and concentration software has not been incorporated at this time, but will become part of the Set-Up software. Suggested alarm concentrations are given in Table 7.1.

Figure 7.2. Screen shots of the user interface software for A) our RamanID product and B) the future Raman Water Security product developed during this program.

A B


89

4. Conclusions The overall goal of Phase II was to fully develop the proposed analyzer and improve sensitivity to detect poisons at 10 µg/L (10 parts-per-billion, ppb) in 10 min (~the 5 day/5L). These goals were largely met as summarized by the following accomplishments: 1) The sol-gel chemistry was successfully optimized to achieve the required sensitivity of at least 10 µg /L (ppb) for all of the 20 target chemicals within 10 minutes using a solid phase extraction cartridge that was included in the sampling system. 2) The two most active sol-gel chemistries, were successfully developed to withstand flow rates of 5 mL/min and pressures of 30 psi. The sample system was successfully designed to reduce flow and pressure to at least these values. 3) Receiver operator characteristic (ROC) curves demonstrated that the required sensitivity could be reproducibly achieved 95% of the time with a 3 minute spectral acquisition, so long as the length of the capillary was scanned. 4) Software was written that successfully identified any of 96 chemicals within a spectral library data base, consisting of chemical agents, pesticides, toxic industrial chemicals, and hydrolysis products. 5) A computer controlled sample system was designed and successfully built that was capable of being connected to virtually any water supply. It controls the water flow rate and pressure, and allows updated analysis every 10 minutes. 6) The automated sample system in conjunction with a Raman analyzer was successfully used to detect 75 µg/L (ppb) methyl phosphonic acid artificially added to water samples obtained from the Kensico Water Reservoir, which supplies New York City its drinking water. The raster method was NOT used, which improved sensitivity by more than a factor of 10 (and would therefore achieve the required sensitivity). Future work. In order to complete the proposed prototype, the following must be addressed: 1) A fiber optic probe with rastering ability must be coupled to the sample system. 2) The sample system must be modified to a) minimize the dilution effects caused by the channels, and b) incorporate both types of SPEs. 3) All of the software routines must be incorporated into one program easily used by water security personnel.

5. Publications During the course of this program, 5 papers were published. They are listed below and attached as appendices at the end of this document. 1. Inscore, F, S Farquharson, “Detecting hydrolysis products of blister agents in water by surface-enhanced Raman

spectroscopy”, SPIE, 5993, 19-23 (2005). 2. Farquharson, S, F Inscore, S Christesen “Detecting chemical agents and their hydrolysis products in water”, in

Surface-Enhanced Raman Scattering – Physics and Applications 103, Eds. K Kneipp, M Moskovitz, and H Kneipp, Springer, Berlin/Heidelberg, 447-460 (2006)

3. Inscore, F, S Farquharson, “Surface-enhanced Raman spectral analysis of blister agents and their hydrolysis products” SPIE, 6378, 63780X (2006).

4. Farquharson, S, F Inscore, “A SERS-based analyzer for point and continuous water monitoring of chemical agents and their hydrolysis products”, IJHSES, 20, 719-728 (2007).

5. Inscore, F, C Shende, A Sengupta, S Farquharson, “Water security: continuous monitoring of water distribution systems for chemical agents by SERS”, SPIE, 6540, (2007).


90

6. Commercialization During this program, Dr. Farquharson worked with Foresight Technologies to evaluate commercialization of the proposed technology. A focus was placed on identifying end-users who would be willing to evaluate the technology, and potential partners that would bring the technology to market. A 40 page report was delivered by James Davis. In this report, the greatest concerns raised by end-users regarding the proposed application were 1) sensitivity, 2) specificity (minimum false positives), and 3) reliability (mean-time-to-maintenance/failure). We believe that this Phase II program has demonstrated that the approach can provide the necessary sensitivity (10 µg/L (ppb) in 10 minutes) and selectivity (unknown identification software). However, mean-time-to-failure is still unknown. Jim Davis identified several potential end-users that would be willing to test the proposed analyzer. Three very high profile operations were selected, and in each case initial discussion with positive response has taken place. These are the City of New York Department of Environmental Protection (CNYDEP, Valhalla, NY, Yves Mikol, Director, Pathogen Field Operations), American Water Works (Voorhees, NJ, Ricardo DeLeon, Director, Innovation and Environmental Excellence), and JEA (Jacksonville, FL, Mark Hollifield, Manager of Water Operations and Maintenance). New York City has been in the spotlight as far as Homeland Security measures are concerned. As a priority area for Federal Aid, it is believe that they will be on the list of five cities funded under the President’s Water Sentinel Initiative. American Water Works operates a total of 450 water systems in the US and Canada, while JEA is the 8th largest water utility in the US. JEA also serves the Jacksonville Naval Air Station, which may be a source of additional funding. Yves Mikol is a member of the National Drinking Water Advisory Council Water Security Working Group (WSWG), a working group tasked with identifying ways to improve the security of the nation’s water supply. The other contacts are also believed to be members of this working group. It is clear that successful demonstrations at any of these sites would provide rapid access to the market. Several potential partners were identified, including (in order of most desirable) Hach (Loveland, CO), who is a leading supplier of water instrumentation, YSI (Yellow Springs, OH), who is a manufacturer of monitoring systems for water testing, and Source Sentinel (E. Syracuse, NY), who is a developer of a water monitoring system using various sensor inputs and processing to detect alarm conditions. Terrance Madden of Source Sentinel expressed interest in teaming possibilities. During the past year we were contacted by Richard Godec, Market Development Manager of GE Power & Water. He has expressed interest in the program and a potential to continue this product development. Discussions are continuing. Potential partnering will be pursued with all four companies.

References 1 Hauschild, V. et al. “Short-Term Exposure Guidelines for Deployed Military Personnel”, USACPPM TG 230A

(May, 1999) available at http://chppm-www.apgea.army.mil/imo/ddb/dmd/DMD/TG/TECHGUID/Tg230.pdf 2 Whitman, CT, “EPA’s Strategic Plan for Homeland Security”, 2002, available at

http://www.epa.gov/epahome/downloads/epa_homeland_security_strategic_plan.pdf 3 EPA, “Response Protocol Toolbox: Planning for and Responding to Drinking Water Contamination Threats and

Incidents. Module 3: Site Characterization and Sampling Guide, Module 4: Analytical Guide” December 12, 2003, available at http://www.epa.gov/safewater/security/ertools.html#compendium

4 Lee, Y.-H., Farquharson, S., Kwon, H., Shahriari, M., Rainey, P., "Sol-Gel Chemical Sensors for Surface-Enhanced Raman Spectroscopy", SPIE, 3537, 252-260 (1998)

5 Farquharson, S.; Lee, Y. H., and C. Nelson "Material for SERS and SERS sensors and method for preparing the same", U.S. Provisional Patent Number AFU-17-PROV (1999)

6 Farquharson, S., and W. W. Smith, W. H. Nelson and J. F. Sperry, ”Biological agent identification by nucleic acid base-pair analysis using surface-enhanced Raman spectroscopy,” SPIE, 3533, 207-214 (1998)

7 Farquharson, S., and W. W. Smith, S. Elliott and J. F. Sperry, "Rapid biological agent identification by surface-enhanced Raman spectroscopy", SPIE, 3855, 110-116 (1999).


91

8 Lee, Y.H. and S. Farquharson, "Rapid chemical agent identification by surface-enhanced Raman spectroscopy",

SPIE, 4378, 21-26 (2001) 9 Lee, Y.H., and S. Farquharson, "SERS sample vials based on sol-gel process for trace pesticide analysis", SPIE,

4206, 140-146 (2001) 10 Farquharson, S., W.W. Smith, Y.H. Lee, S. Elliott and J. F. Sperry, "Detection of biological signatures: A

comparison of electrolytic and metal-doped sol-gel surface-enhanced Raman media", SPIE, 4575, 62-72 (2002) 11 Farquharson, S. and Lee, Y. H., "High Throughput Identification: Drug Analysis by Surface-Enhanced Raman

Spectroscopy", D&MD Focus Reports, 4-7 (March 2001) 12 Farquharson, S., P. Maksymiuk, K. Ong and S.D. Christesen, "Chemical agent identification by surface-

enhanced Raman spectroscopy", SPIE, 4577, 166-173 (2001). 13 Farquharson, S. and P. Maksymiuk, “Simultaneous chemical separation and surface-enhancement Raman

spectral detection using silver-doped sol-gels”, Applied Spectroscopy, 57, 479-482 (2003). 14 Farquharson, S. PI, "High Sensitivity Raman Spectrometer" NSF CN DMI-0296116. 15 Farquharson, F. PI, "A portable Raman instrument for fuel characterization", Navy, CN M6785-03-M-5043. 16 Hauschild, V. et al. “Short-Term Exposure Guidelines for Deployed Military Personnel”, USACPPM TG 230A

(May, 1999) available at http://chppm-www.apgea.army.mil/imo/ddb/dmd/DMD/TG/TECHGUID/Tg230.pdf 17 P.M. Tesler, S.D. Christesen, K.K. Ong, E.M. Clemente, A.M Lenhoff, E.W. Kaler, and O.D. Velev, “On-line

Spectroscopic Characterization of Sodium Cyanide with Nanostructured Gold Surface-Enhanced Raman Spectroscopy Substrates” Applied Spectroscopy, 56, 1524-1530 (2002).

18 Bio-Rad, Life Science Research Products Catalogue, 2004/05, Page 61. 19 DOWEX™ Ion Exchange Resins, Tech Sheet, Form No. 177-01755-0207, The Dow Chemical Company.

SPIE-2005-5993 19

Detecting hydrolysis products of blister agents in water by surface-enhanced Raman spectroscopy

Frank Inscore and Stuart Farquharson

Real-Time Analyzers, Middletown, CT, 06457

ABSTRACT Protecting the nation’s drinking water from terrorism, requires microg/L detection of chemical agents and their hydrolysis products in less than 10 minutes. In an effort to aid military personnel and the public at large, we have been investigating the ability of surface-enhanced Raman spectroscopy (SERS) to detect microgram per liter (part-per-billion) concentrations of chemical agents in water. It is equally important to detect and distinguish the hydrolysis products of these agents to eliminate false-positive responses and evaluate the extent of an attack. Previously, we reported the SER spectra of GA, GB, VX and most of their hydrolysis products. Here we extend these studies to include the chemical agent sulfur-mustard, also known as HD, and its principle hydrolysis product thiodiglycol. We also report initial continuous measurements of thiodiglycol flowing through a SERS-active capillary. Keywords: chemical warfare agent detection, CWA, hydrolysis, SERS, Raman spectroscopy

1. INTRODUCTION The July 2005 terrorist bombings of the London transit system are a stark reminder that such attacks on the United Kingdom and the United States will continue. Countering such attacks requires recognizing likely deployment scenarios and having the required technology to rapidly detect the deployment event. In addition to the expected use of chemical agents released into the air, terrorists may also poison water supplies with chemical warfare agents (CWAs). The National Strategy for Homeland Security designates the Environmental Protection Agency with the task of securing the nations drinking water.1 Presently, the EPA employs several field test kits to monitor drinking water supplies, and gas chromatography coupled with mass spectrometry in supporting laboratories to confirm positive responses.2 Unfortunately, these test kits are prone to false-positive responses, and follow-up analysis typically takes a day. This is entirely inadequate for the prevention of widespread illness and potential fatalities. In the past several years we have been investigating the use of surface-enhanced Raman spectroscopy (SERS) to be used as a field-usable analyzer that can detect chemical agents in water at the required microg/L sensitivity and 10 minute timeframe.3,4,5,6,7 The expected success of SERS is based on the million-fold or more Raman signal increase obtained when a molecule interacts with surface plasmon modes of metal nanoparticles.8 In the case of cyanide, an industrial-based CWA and methyl phosphonic acid, the final hydrolysis product for the nerve agents, we have measured at or below 10 microg/L in one minute.9 The expected success of SERS is also based on the unique set of Raman spectral peaks associated with the molecular vibrational modes of each molecule. The unique SER spectra should not only reduce false-positive responses, but also allow discriminating hydrolysis products of CWAs. This is important, since CWAs can hydrolyze rapidly in the presence of water,10 and detection of the hydrolysis products could allow determining 1) the state of an attack (ratio of CWA to hydrolysis product(s)), 2) the point of attack initiation, and 3) the continued extent and severity of the CWA attack throughout a water distribution system. Previously, we used SERS to measure sarin, tabun, VX, and EA2192, and their respective hydrolysis products.3,4,6,7 Here we extend these studies to include the chemical warfare agent sulfur-mustard, designated HD, and its primary hydrolysis product thiodiglycol (TDG, Figure 1). The physical and chemical properties of this blister agent are well known. It’s solubility in water is 0.92 g/L with a hydrolysis half-life of 8.5 min (both at 25 C).10 HD has an oral LD50 of 0.7 mg/kg in humans,11 and the military drinking water guideline places the 5-day 5L limit at 100 microg/L.12 TDG is relatively non-toxic, very water soluble at 690 g/L, and stable in water with a hydrolysis half-life of approximately 6 days. Accordingly, a reasonable sensitivity goal to ensure safe water is placed at 10 microg/L for HD and an equivalent goal to map HD usage is placed at 10 microg/L for TDG.13

SPIE-2005-5993 20

2. EXPERIMENTAL Highly distilled sulfur mustard, designated HD (bis(2-chloroethyl)sulfide), was measured at the U.S. Army’s Edgewood Chemical Biological Center (Aberdeen, MD). Thiodiglycol, designated TDG here (bis(2-hydroxyethyl)sulfide), was purchased as an analytical reference material from Cerilliant (Round Rock, TX). TDG was measured at Real-Time Analyzers, Inc. (RTA, Middletown, CT). All solvents, including methanol, ethanol, and HPLC water, as well as all sol-gel precursor chemicals including AgNO3, tetramethyl orthosilicate, methyltrimethoxysilane, HNO3 and NaBH4, were purchased from Sigma-Aldrich (St. Louis, MO). HD samples prepared for SERS analysis consisted of 0.1% v/v HD in methanol. The methanol was used to minimize hydrolysis. The final concentration is 1000 parts-per-million (ppm, EPA definition). TDG samples were prepared for SERS analysis using methanol for static measurements and HPLC grade water for flow measurements. All HD measurements were performed in SERS-active vials (Simple SERS Sample Vials, RTA),14 while all TDG measurements were performed in SERS-active capillaries (1-mm diameter glass capillaries filled with silver-doped sol-gels).15,16 In the case of flow measurements, a peristaltic pump (variable flow mini-pump, Control Co., Friendswood, TX) was used to flow the 1 and 10 ppm TDG samples through a SERS-active capillary at 1 mL per min. The vials or capillaries were mounted horizontally on an XY positioning stage (Conix Research, Springfield, OR), such that the focal point of an f/0.7 aspheric lens was positioned just inside the glass wall. The probe optics and fiber optic interface have been described previously.16 In all cases a 785 nm diode laser was used to deliver ~100 mW of power to the SERS samples and 300 mW to the Raman samples. A Fourier transform Raman spectrometer equipped with a silicon photo-avalanche detector (RTA, model IRA-785), was used to collect both the RS and SERS at 8 cm-1 resolution. 3. RESULTS AND DISCUSSION The surface-enhanced and normal Raman spectra of HD have been measured and are shown in Figure 2. The SER spectrum is dominated by a peak at 630 cm-1 with an extended high frequency shoulder composed of two or more peaks (695, 830 cm-1), as well as a moderately intense peak at 1045 cm-1. It is possible to assign these peaks based on the normal Raman spectrum of HD, and previous assignments.17 Theoretical studies assigned the 640, 655, 700 cm-1 peaks to C-Cl stretching modes and the 740, and 760 cm-1 peaks to C-S stretching modes. Additional peaks are observed at 1040, 1190, 1270, 1295, 1410, 1425, and 1440 cm-1. The first peak is assigned to a C-C stretch, while the remaining peaks are all CH2 deformation modes (scissors, twists, and wags). Based on these assignments, then only the C-Cl peak maintains significant intensity in the SER spectrum occurring at 630 cm-1. If the C-Cl assignments are correct, then the SER spectra suggest that the molecule to metal interaction is strongest through the chlorine end groups. Alternatively, the electron lone pairs of the tetrahedrally coordinated sulfur of HD could interact with the silver surface. Consequently, the 630 cm-1 SERS peak could be assigned to CS or CSC stretching modes (see below).18 The surface-enhanced and normal Raman spectra of TDG have been measured and are shown in Figure 3. The SER spectrum is dominated by three peaks at 630, 715, and 1010 cm-1 with minor peaks at 400, 820, 930, 1210, 1275, 1410, and 1460 cm-1. Similarly, the Raman spectrum contains two intense peaks at 660 and 1010 cm-1, while moderately intense peaks occur at 400, 680 (shoulder), 735, 770, 830, 950, 1040, 1230, 1290, 1420, and 1465 cm-1. In both spectra, the assignment of the peaks near 1000 cm-1 can be confidently assigned to C-C stretching modes, while the peaks from 1200 to 1465 cm-1 can be confidently assigned to various CH2 deformation modes. Here, however, it is difficult to assign the 630 cm-1 SERS peak to a C-Cl mode, since the chlorines have been replaced by hydroxyl groups.

Figure 1. Hydrolysis of bis(2-chloroethyl)sulfide (HD) to bis(2-hydroxyethyl)sulfide (TDG).

H2O+ 2HCl

SPIE-2005-5993 21

Consequently, in the case of HD and TDG, assigning the 630 cm-1 peak to a CS or a CSC stretch, is favored. Although the 620-680 cm-1 peaks are normally assigned to C-Cl modes, and the 700-750 cm-1 peaks to CS modes, most authors concede that the reverse assignments are possible.

Of possibly greater importance, is that the TDG SER spectrum is of high quality, with three distinct peaks. With the goal of detecting this hydrolysis product of HD in water, a number of samples of decreasing concentration were prepared and measured. As Figure 4 shows, these peaks are evident even at 10 ppm (0.001% v/v in methanol). However, repeated measurements of 1 ppm did not yield any discernable peaks (lowest trace in Figure 4). Notwithstanding, measurements were also performed in a flowing stream. Initial measurements of a 10 ppm sample yielded quality spectra and prompted measurements of a 1 ppm sample. As Figure 5 shows, reasonable spectra are obtained, even at 1 minute resolution. It is worth stating that the 630 cm-1 peak was evident in all spectra collected over a 12 minute period. There is an important difference between the TDG spectra recorded for static and flowing samples, namely that the 715 cm-1 peak is noticeably more intense in the static sample. This suggests that it may represent a photo-degradation product. Further studies are required to clarify this point.

Figure 2. A) SERS and B) RS of HD. A) 0.1% v/v (1000 ppm) in MeOH in a SERS-active vial, 100 mW of 785 nm, 1-min, B) neat sol. in glass container, 300 mW of 785 nm, 5-min.

Figure 3. A) SERS and B) RS of TDG. A) 0.1% v/v in MeOH in SERS-active capillary, 100 mW of 785 nm, 1-min, B) neat sol. in glass capillary, 300 mW of 785 nm, 5-min.

Figure 4. SERS of 1000, 100, 10 and 1 ppm TDG in water (top to bottom). All in SERS-active capillaries, 100 mW of 785 nm, 1-min.

Figure 5. SERS of 1 ppm TDG in water flowing through a SERS-active capillary at 1, 2, 3, 4, and 5 min. (top to bottom), 100 mW of 785 nm, 1-min each.

A B

A B

SPIE-2005-5993 22

4. CONCLUSIONS The ability to measure and distinguish HD and TDG using SERS-active capillaries has been demonstrated. Specifically, the peak at 715 cm-1 is unique to TDG, as both chemicals produce an intense SERS peak at 630 cm-1. The latter peak is likely due to CS or CSC stretching modes favorably enhanced by the interaction of the sulfur lone electron pairs to silver surface. Measurements of similar chemicals, such as diethylsulfide, are ongoing to clarify this assignment. Detection of TDG at 1 mg/L in 1 minute in a flowing system suggests that the goal of 10 microg/L in 10 minutes is possible. Improvements in the enhancement achieved by the SERS-active capillaries, as well as their durability, are the focus of current research and product development.

5. ACKNOWLEDGMENTS The authors are grateful for the support of the U.S. Army (DAAD13-02-C-0015, Joint Service Agent Water Monitor program) and the Environmental Protection Agency (EP-D-05-034). The authors would also like to thank Dr. Steve Christesen for helpful discussions, and Mr. Chetan Shende for sol-gel chemistry development.

6. REFERENCES 1 Whitman, CT, “EPA’s Strategic Plan for Homeland Security”, 2002, available at

http://www.epa.gov/epahome/downloads/epa_homeland_security_strategic_plan.pdf 2 Creasy, W., Brickhouse, M., Morrissey, K., Stuff, J., Cheicante, R., Ruth, J., Mays, J., Williams, B., O’Connor, R.,

Durst, H., “Analysis of chemical weapons decontamination waste from old ton containers from Johnston atoll using multiple analytical methods”, Environ. Sci. Technol., 33, 2157-2162, (1999).

3 Farquharson, S., Maksymiuk, P., Ong, K., Christesen, S., “Chemical agent identification by surface-enhanced Raman spectroscopy”, SPIE, 4577, 166-173 (2001).

4 Farquharson, S., Gift, A., Maksymiuk, P., Inscore, F., Smith, W., Morrisey, K., Christesen, S., “Chemical agent detection by surface-enhanced Raman spectroscopy”, SPIE, 5269, 16-22 (2004).

5 Farquharson, S., Gift, A., Maksymiuk, P., Inscore, F., Smith, W., “pH dependence of methyl phosphonic acid, dipicolinic acid, and cyanide by surface-enhanced Raman spectroscopy”, SPIE, 5269, 117-125 (2004).

6 Inscore, F., A. Gift, P. Maksymiuk, S. Farquharson, “Characterization of chemical warfare G-agent hydrolysis products by surface-enhanced Raman spectroscopy” SPIE, 5585, 46-52 (2004).

7 Farquharson, S., A. Gift, P. Maksymiuk, F. Inscore “Surface-enhanced Raman spectra of VX and its hydrolysis products”, Appl. Spectrosc., 59, 654-660 (2005).

8 Jeanmaire, D.L., and R.P. Van Duyne, J. Electroanal. Chem., 84, 1-20 (1977). 9 Inscore, F., P. Maksymiuk, and S. Farquharson, “SERS detection of chemical agents in flowing streams”, in

preparation. 10 Munro, N.B., Talmage, S.S., Griffin, G.D., Waters, L.C., Watson, A.P., King, J.F., and Hauschild V. “The Sources,

Fate, and Toxicity of Chemical Warfare Agent Degradation Products”, Environ. Health Perspect. 107, 933-974 (1999).

11 Committee on Toxicology, Review of Acute Human-Toxicity Estimates for Selected Chemical-Warfare Agents, Nat. Acad. Press (Washington, D.C.) 1997

12 Hauschild, V. et al. “Short-Term Exposure Guidelines for Deployed Military Personnel”, USACPPM TG 230A (May, 1999) available at http://chppm-www.apgea.army.mil/imo/ddb/dmd/DMD/TG/TECHGUID/Tg230.pdf

13 These values are only estimates made by the authors. 14 Farquharson, S., Y.H. Lee, and C. Nelson, “Material for surface-enhanced Raman spectroscopy and SER sensors,

and method for preparing same", U.S. Patent Number 6,623,977 (2003) 15 Farquharson, S. and P. Maksymiuk “Simultaneous chemical separation and surface-enhanced Raman spectral

detection using metal-doped sol-gels” and “Separation and Plural-point surface-enhanced Raman spectral detection using metal-doped sol-gels”, U.S. Patent Numbers 6,943,031 and 6,943,032 (2005)

16 Farquharson, S., Gift, A., Maksymiuk, P., Inscore, F., “Rapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopy”, Appl. Spectrosc. 58, 351-354 (2004).

17 Sosa, C., R.J. Bartlett, K. KuBulat, and W.B. Person, “A theoretical study of harmonic vibrational frequencies and infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H, Cl)”, J. Phys. Chem., 93, 577-588 (1993).

18 Joo, T., K. Kim, M. Kim “Surface-enhanced Raman study of organic sulfides adsorbed on silver”, J. Molec. Struct.,16, 191-200 (1987).

Springer Book Kneipp Editor Draft 1

25. Detecting chemical agents and their hydrolysis products in water

Stuart Farquharson, Frank E. Inscore and Steve Christesen Real-Time Analyzers, Middletown, CT, 06457

25.1 INTRODUCTION The use of chemicals as weapons was introduced during World War I. It is estimated that chlorine, phosgene and sulfur-mustard (HD) resulted in an estimated death of 100,000 soldiers and 1 million injuries [1]. Over the next 20 years, chemicals designed specifically for warfare were developed; this included the substantially more toxic nerve agents, tabun, sarin, and soman (GA, GB, and GD, respectively). Fortunately, these abhorrent chemicals were not used in WWII, as world leaders feared reprisal attacks on their cities. During the Iran-Iraq war in the 1980s, the Iraqis used HD, GA, GB, and GF (cyclo-sarin), and in 1988, Saddam Hussein used mustard and possibly nerve agents in killing several thousand Kurds [1]. In more recent years, chemical agents have been used by terrorists. In Japan, the Aum Shinrikyo religious cult released GB within the Tokyo subway system in 1995 [2]. The release of GB in this confined space had devastating effects resulting in 12 fatalities and hospitalization of thousands. This event and the mailing of anthrax causing spores through the US Postal System in 2001 demonstrated that deployment of chemical and biological agents do not require sophisticated delivery systems, and a wide range of attack scenarios must be considered. Among these scenarios is the deliberate poisoning of drinking water. This includes water supplies used in military operations and water delivered to major cities from reservoirs and through distribution systems. Countering such an attack requires detecting poisons in water rapidly, and at very low concentrations. The required detection sensitivity for each agent depends on several factors, such as toxicity and hydrolysis (Table 25.1). In the case of cyanide (AC) it known that 4 milligrams per liter of water produces detectable changes in human blood chemistry and 8 mg L-1 causes severe, but reversible symptoms [3]. The military has used this and other toxilogical data to set a field drinking water standard (FDWS) for cyanide at 2 mg L-1 [4]. The FDWS represents the maximum allowable concentration that is assumed safe when 15 L of water per day is consumed over 5 days (expected soldier intake in arid climates). Human toxicity data for the other chemical warfare agents in water have, in general, not been determined. The normal route of exposure for chemical warfare agents is inhalation, and most of the toxicity data is given as the LCt50s [1], the concentration that is lethal to 50% of an exposed population as a function of exposure time. In the case of mustard, animal studies along with the inhalation LCt50, the oral lethal dosage of 0.7 mg per kg of body mass (LD50), and modeling studies [5], have been used to set the FDWS at 0.047 mg L-1. Similar analyses of LCt50s and LD50s for GB and VX have been used to set their FDWS at 0.0046 and 0.0025 mg L-1, respectively. The FDWS concentrations have also been used by the military to set the minimum detection requirement for poisons in water. Table 25.1. Military field drinking water standard [4], lethal exposures and dosages [1,3,5], and water properties for selected chemical warfare agents.

Chemical FDWS 5-day/15L (mg L-1)

LCt50 inhalation

(mg-min m-3)

LD50 oral

(mg kg-1)

Water Solubility at 25°C


Hydrolysis Product

HCN (AC) NaCN

2 2000 -

480 g L-1

- CN

Mustard (HD) 0.047 900 0.7 0.92 g L-1 2-30 hours TDG Sarin (GB) 0.0046 70 2 completely

miscible 20-40 hours IMPA, MPA

VX 0.0025 35 0.07 150 g L-1 82 hours DIASH, EMPA, EA2192, MPA

In order to detect these poisons in water, their properties in water must also be considered, i.e. the solubility, rate of hydrolysis, and hydrolysis products formed. In the case of cyanide, as HCN, KCN, or NaCN, all of these chemicals are extremely soluble in water (completely miscible, 716, and 480 g L-1, respectively) [6]. In solution the cyanide


ion is formed in equilibrium with the conjugate acid, HCN (Figure 25.1A), according to the Ka of 6.15x10-10 [7 ]. In the case of cyanide then it is important to know the pH, if one form of the chemical is to be detected versus the other. For example, if 2 mg L-1 of NaCN is added to water (the FDWS), then 1.25 mg L-1 of CN- and 0.75 mg L-1 of HCN will be present.

Figure 25.1. Hydrolysis reaction pathways for A) CN, B) HD, C) GB, and D) VX. In the case of sulfur-mustard, the situation is somewhat more complex. It is marginally soluble in water tending to form droplets, and hydrolysis occurs at the droplet surface. This property has made measuring the hydrolysis rate constant difficult, and half-lives anywhere from 2 to 30 hours are reported [8]. Chemically, the hydrolysis of HD involves the sequential replacement of the chlorine atoms by hydroxyl groups through cyclic sulfonium ion intermediates to form thiodiglycol (TDG, Figure 25.1B) [9]. If a median hydrolysis rate is assumed, then early detection of poisoned water will require measuring HD, while post-attack or downstream monitoring will require measuring TDG. For sarin, the analysis is more straightforward, since it dissolves readily into water and it is stable for a day or more. In this case, detecting poisoned water will largely require measuring sarin, while monitoring the attack will require detecting its sequential hydrolysis products, isopropyl methylphosphonic acid (IMPA) and methyl phosphonic acid (IMPA, MPA, respectively, Figure 25.1C) [8,10,11]. The other hydrolysis products, hydrofluoric acid and 2-propanol, are too common to provide definitive evidence of water poisoning and their measurement would be of limited value. VX is reasonably soluble, and like sarin, is fairly persistent with a hydrolysis half-life greater than 3 days [12]. Unfortunately, one of its hydrolysis products, known as EA2192, is considered just as toxic as VX, more soluble and more persistent in water [13]. Consequently, detecting the early stages of poisoning water should focus on measuring VX, while longer term monitoring should focus on EA2192. The earliest technologies developed for CWA detection were based on electrochemical, ionization, or colorimetric analysis. Examples of the latter include phosgene, M8 and M9 tape, which change color when in contact with a sample like pH paper. Although these tapes are easy to use, they are not generally agent specific and suffer from a high percentage of false-positives [14]. For example, M8 changes color when in contact with common solvents such as acetonitrile, ethanol, methanol, or common petroleum products such as brake fluid, lighter fluid, or WD-40 [15]. More rigorous laboratory methods have been successfully developed to detect chemical agents with minimum false-positive responses. More than a decade ago, Black et al. demonstrated the ability of combining gas chromatography

A B C D


with mass spectrometry detection (GC/MS) to measure sarin and mustard [16]. Sega at al. used GC with a phosphorous-selective flame ionization detector to analyze nerve agent hydrolysis products in groundwater [17], while several researchers used capillary electrophoresis (CE) to measure chemical warfare agents and their hydrolysis products [18,19,20]. The sensitivity of these techniques has improved by two orders of magnitude from 1 mg L-1 to 0.01 mg L-1 in 10 years. A comprehensive development of these techniques was undertaken by Creasy et al. in analyzing chemical weapon decontamination waste from the Johnston Atoll [11,21]. These researchers used GC/MS for nerve agents, GC coupled atomic emission detection for arsenic compounds, LC/MS for mustard compounds, and CE with ultraviolet absorption detection for alkyl phosphonic acids. Detection limits of 0.02 and 0.140 mg L-1 were reported for nerve agents and mustard, respectively. Detection of the alkyl phosphonic acids have proven more difficult, and Liu, Hu and Xie recently used GC/MS to detect mg L-1 concentrations of these degradation products [22]. However, they concede that all of these separation methods require extraction, derivatization, and repeated column calibration, making them labor intensive, time consuming (typically 30 to 60 minutes), and less than desirable for field use. Another variant of these separation/mass detection technologies is ion mobility spectrometry (IMS) [23]. This technology has been successfully developed to measure explosives in air samples, and commercial products can be found at most airports [24]. Eiceman et al. have investigated the ability of IMS to measure organophosphorous compounds in air [25], while Steiner et al. have investigated IMS to measure chemical agent simulants in water [26]. In the latter case, electrospray ionization was coupled to the sample entry point of an IMS, and a time-of-flight MS was added as an orthogonal detector. Water samples spiked with 10 mg L-1 diisopropylmethylphosponate and thiodiglycol could be measured in 1-min, once sample pretreatment was accomplished. It is worth noting that with proper care these MS-based technologies are likely to detect chemical agents with virtually no false-positives, but detection limits are still insufficient by 1 to 2 orders of magnitude in the case of nerve agents and their hydrolysis products. More rapid analysis of agents in the solid, liquid and gas phase has been demonstrated by vibrational spectroscopy [27-31]. Hoffland et al. reported infrared absorbance spectra and absolute Raman cross sections for several chemical agents [27], while Christesen measured Raman cross sections for sarin, tabun, mustard gas, and VX [28]. Again, however these technologies also have limitations. Raman spectroscopy is simply not a very sensitive technique, and detection limits are typically 0.1% (1000 ppm). And infrared spectroscopy would have limited value in analyzing poisoned water, since the very strong infrared absorption of water would obscure most other chemicals present. Nevertheless, efforts to overcome these limitations have been demonstrated. Braue and Pannella quantified the G-series nerve agents (tabun, sarin, and soman) in terms of infrared attenuated total reflectance using a circle-cell [29]. Enormous improvements in sensitivity for Raman spectroscopy can be achieved through surface-enhancement [32]. The interaction of surface plasmon modes of metal particles with target analytes can increase scattering efficiency by as much 14-orders of magnitude, although 6-orders of magnitude are more common. The details of surface-enhanced Raman spectroscopy (SERS) can be found in the beginning of this book. The utility of SERS to measure chemical agents was first demonstrated by Alak and Vo-Dinh by measuring several organophosphonates as simulants of nerve agents on a silver-coated microsphere substrate [33]. Spencer, et al. used SERS to measure cyanide, MPA, HD and EA2192 on electrochemically roughed gold or silver foils [34,35,36]. However, in all of these measurements, the sample needed to be dried on the substrates to obtain the best sensitivity (e.g. 0.05 mg L-1 for MPA). More recently, Tessier et al. obtained SERS of 0.04 mg L-1 cyanide in a stream flowing over a substrate formed by a templated self-assembly of gold nanoparticles [37]. However, optimum sensitivity required introduction of an acid wash and the measurements were irreversible. In the past few years, we have also been investigating the ability of SERS to measure chemical agents at 0.001 mg L-1 in water and with sufficient spectral uniqueness to distinguish the agent and its hydrolysis products [38-43]. In our work, we have developed silver-doped sol-gels as the SERS-active medium. These sol-gels can be coated on the inside walls of glass vials, such that water samples can be added to perform point-analysis, or they can be incorporated into glass capillaries, such that flowing measurements can be performed [44]. Here, both sampling devices were used to measure and compare SER spectra of AC, HD, VX and several of their hydrolysis products, TDG, EA2192, EMPA, and MPA. In addition, a field-usable Raman analyzer was used to measure 0.01 mg L-1 cyanide flowing in water with a detection time of less than 1-min.


25.2 EXPERIMENTAL Sodium cyanide, 2-hydroxyethylethyl sulfide (HEES), 2-chloroethylethyl sulfide (CEES) and methylphosphonic acid (MPA) were purchased from Sigma-Aldrich (St. Louis, MO) and used as received. Ethyl methylphosphonic acid (EMPA), isopropyl methylphosphonic acid (IMPA), 2-(diisopropylamino) ethanethiol (DIASH), and thiodiglycol (TDG, bis(2-hydroxyethyl)sulfide) were purchased from Cerilliant (Round Rock, TX). Highly distilled sulfur mustard (HD, bis(2-chloroethyl)sulfide), isopropyl methylphosphonofluoridate (GB), ethyl S-2-diisopropylamino ethyl methylphosphonothioate (VX), and ethyl S-2-diisopropylamino methylphosphonothioate (EA2192) were obtained at the U.S. Army’s Edgewood Chemical Biological Center (Aberdeen, MD) and measured on-site. All samples were initially prepared in a chemical hood as 1000 parts-per-million (1 g L-1 or 0.1% by volume, Environmental Protection Agency definition) in HPLC grade water (Fischer Scientific, Fair Lawn, NJ) or in some cases methanol or ethanol (Sigma-Aldrich) to minimize hydrolysis. Once prepared, the samples were transferred into 2-ml glass vials internally coated with a silver-doped sol-gel (Simple SERS Sample Vials, Real-Time Analyzers, Middletown, CT) or drawn by syringe or pump into 1-mm diameter glass capillaries filled with the same SERS-active material [45,46,47]. In the case of flow measurements, a peristaltic pump (variable flow mini-pump, Control Co., Friendswood, TX) was used to flow the various cyanide solutions through a SERS-active capillary at 1 mL min-1. The vials or capillaries were placed on aluminum plates machined to hold the vials or capillaries on a standard XY positioning stage (Conix Research, Springfield, OR), such that the focal point of an f/0.7 aspheric lens was positioned just inside the glass wall. The probe optics and fiber optic interface have been described previously [40]. SER spectra were collected using a Fourier transform Raman spectrometer equipped with a 785 nm diode laser and a silicon photo-avalanche detector (IRA-785, Real-Time Analyzers). All spectra were nominally collected using 100 mW, 8 cm-1 resolution, and 1-min acquisition time, unless otherwise noted. Complete experimental details can be found in Reference 48. For added safety, all samples were measured in a chemical hood. In the case of actual agents measured at Edgewood, the FT-Raman instrument was placed outside the laboratory and 30 foot fiber optic and electrical cables were used to allow remote SERS measurements and plate manipulation. 25.3 RESULTS AND DISCUSSION 25.3.1 Cyanide. Sodium cyanide completely dissolves in water forming the ions in equilibrium with the conjugate acid, HCN as described above. Concentrations of 1.0, 0.1, and 0.01 mg L-1 result in CN- concentrations of 0.52, 0.016, and 0.00021 mg L-1 as the corresponding pH decreases from just above the pKa of 9.21 at 9.24 to 8.48 and 7.54. This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum, and no spectral signal is observed below pH 7 (except on electrodes at specific potential conditions [49]). The SER spectra of cyanide are dominated by an intense, broad peak at 2100 cm-1 attributed to the C≡N stretch (Figure 25.2). This mode occurs at 2080 cm-1 in Raman spectra of solutions, and the frequency shift in SER spectra is attributed to a strong surface interaction, which is supported by the appearance of a low frequency peak at 135 cm-1 due to a Ag-CN stretch (not shown). It is also observed that as the concentration decreases, the CN peak shifts to 2140 cm-1. This shift has been attribute to the formation of a tetrahedral Ag(CN)3

2- surface structure [50], as well as to CN adsorbed to two different surface sites [51]. Alternatively, it has also been suggested that at concentrations near and above monolayer coverage, the CN- species is forced to adsorb end-on due to crowding, and at lower concentrations the molecule can reorient to lie flat. This suggests that the 2100 and 2140 cm-1 peaks correspond to the end-on and flat orientations, respectively. However, a previous concentration study of cyanide on a silver electrode observed the reverse trend, i.e. greater intensity was observed for the 2100 cm-1 peak at low concentration [49]. Repeated measurements of cyanide in the SERS-active vials consistently allowed measuring 1 mg L-1 (1 ppm), but rarely below this concentration (Figure 25.2A). Nevertheless, this sensitivity is in general sufficient for point sampling of water supplies. In the case of continuous monitoring of water, the capillaries are a more appropriate sampling format, and they also allowed routine measurements at 0.01 mg L-1 and repeatable measurements at 0.001 mg L-1 (1 ppb, Figure 25.2B). Employing this format, a 50 mL volume of 0.01 mg L-1 cyanide solution was flowed at 2.5 mL min-1 through a SERS-active capillary, and spectra were recorded every 20 seconds. As Figure 25.3 shows, the cyanide peak was easily discerned as soon as the solution entered the capillary and remained relatively stable over the course of the experiment. It is worth noting, as indicated above, that the SERS peak in Figure 25.3 is in fact due to 210 ng L-1!


Figure 25.2. Surface enhanced Raman spectra of CN in water in silver-doped sol-gel A) coated glass vials and B) filled glass capillaries. All spectra were recorded using 100 mW of 785 nm in 1-min and at a resolution of 8 cm-1.

Figure 25.3. 2100 cm-1 peak height measured during continuous flow of a 0.01 mg L-1 (10 ppb) cyanide in water. Surface-enhanced Raman spectra are shown for 1 and 6 min after sample introduction. A 2.5 mL min-1 flow rate was used and spectra were recorded every 20 sec using 100 mW of 785 nm.

A B


25.3.2 HD and CEES. The surface-enhanced Raman spectrum of HD is dominated by a peak at 630 cm-1 with an extended high frequency shoulder composed of at least two peaks evident at 695 and 830 cm-1, as well as a moderately intense peak at 1045 cm-1 (Figure 25.4A). The latter peak is assigned to a CC stretching mode, based on the assignment for a peak at 1040 cm-1 in the Raman spectrum of HD [52]. The assignment of the 630 cm-1 peak is less straightforward, since the Raman spectrum of HD contains five peaks in this region at 640, 655, 700, 740, and 760 cm-1 [40,52]. Theoretical calculations for the Raman spectrum of HD indicate that the first three peaks are due to CCl stretching modes, and the latter two peaks to CS stretching modes [53]. Based on these calculations, and the expected interaction between the chlorine atoms and the silver surface, it is reasonable to assign the 630 cm-1 SERS peak to a CCl mode [40]. However, recent SERS measurements of diethyl sulfide produced a very simple spectrum with an intense peak at 630 cm-1 [54,55], strongly suggesting CS or CSC stretching modes as the appropriate assignment for this peak [56]. The authors of the theoretical treatment concede that the CCl and CS assignments could be reversed [53]. The CS assignment also indicates that HD interacts with the silver surface through the sulfur electron lone pairs. But, interaction between chlorine and silver is still possible and may be responsible for the 695 cm-1 peak. The 830 cm-1 peak is left unassigned.

Figure 25.4. Surface-enhanced Raman spectra of A) HD in methanol and B) TDG in water. Spectral conditions as in Fig. 25.2, samples were 1 g L-1. The surface-enhanced Raman spectrum of TDG is also dominated by a peak at 630 cm-1 with minor peaks at 820, 930, 1210, and 1275 cm-1 (Figure 25.4B). Again, the 630 cm-1 peak is preferably assigned to a CSC stretching mode versus a CCl mode, especially since the chlorines have been replaced by hydroxyl groups. Furthermore, the lack of a 695 cm-1 peak in the TDG spectrum supports the assignment of this peak in the HD spectrum to a CCl mode. The 930, 1210 and 1275 cm-1 SERS peaks are assigned to a CC stretch with CO contribution, and two CH2 deformation modes (twist, scissors, or wag) based on the assignments for the corresponding peaks at 940, 1230 and 1290 cm-1 in the Raman spectrum of TDG [52,54 ]. It is worth noting that irradiation at high laser powers or for extended periods produces peaks at 715 and 1010 cm-1, which are attributed to a degradation product, such as 2-hydroxy ethanethiol [54]. The SERS of CEES is very similar to HD, dominated by a peak at 630 cm-1 that is accordingly assigned to a CS or CSC stretching mode (Figure 25.5A). This peak also has a high frequency shoulder centered at 690 cm-1, and a third peak appears at 720 cm-1 in this region. Again, these can be assigned to CCl or CS modes. The quality of this spectrum also reveals weak peaks at 1035, 1285, 1410, and 1445 cm-1. Peaks at 1035, 1285, 1425, and 1440 cm-1

A B


appear in the Raman spectrum of CEES, and the previous peak assignments are used here [52], i.e. the first peak is assigned to a CC stretch, while the remaining peaks are assigned to various CH2 deformation modes.

Figure 25.5. Surface-enhanced Raman spectra of A) CEES and B) HEES. Spectral conditions as in Fig. 25.2, samples were 1 g L-1 in methanol. Replacing the chlorine atom of CEES by a hydroxyl group in forming HEES produces SER spectral changes analogous to those cited above for HD and TDG. Again, the SER spectrum is dominated by an intense peak at 630 cm-1 attributed to a CS or CSC stretching mode, and the other CEES peaks in this region, specifically the 720 cm-1 peak, disappear (Figure 25.5B). Peaks with modest intensity at 1050 and 1145 cm-1 are assigned to a CC stretching mode and CH2 deformation, respectively. A new peak at 550 cm-1 is likely due to a skeletal bending mode, such as CSC, SCC, or CCO. Finally, it is worth stating that HD, TDG, CEES, and HEES all produce moderately intense peaks at 2865 and 2925 cm-1 (not shown), that can be assigned to symmetric and asymmetric CH2 stretching modes. Only a limited number of measurements of HD were performed to evaluate sensitivity, due to the safety requirements. HD was repeatedly observed at 1 g L-1 and usually observed at 0.1 g L-1 (100 ppm) in the SERS-active vials [40] But even at the latter concentration, substantial improvements in sensitivity are required to approach the required 0.05 mg L-1 (50 ppb) sensitivity. More extensive experiments were performed on HD’s hydrolysis product, TDG since this chemical is safely handled in a regular chemical lab. Flowing TDG through SERS-active capillaries allowed repeatable measurements at 10 mg L-1, and routine measurements at 1 mg L-1 (1 ppm) [55]. These SERS measurements of TDG suggest that the required HD sensitivity may be achievable using this technique. Similar flowing measurements in capillaries for HD, CEES, and HEES have not been performed. 25.3.3 Sarin. SERS measurements of GB have not been made, but its primary hydrolysis products, IMPA and MPA, have been measured using the SERS-active vials. The SERS of IMPA is very similar to its Raman spectrum [42], which in turn is very similar to the Raman spectrum of sarin [28]. The SER spectrum is dominated by a peak at 715 cm-1 (Figure 25.6A), which is assigned to a PC or PO plus skeletal stretching mode, as is a weak peak at 770 cm-1. These assignments are also consistent with a theoretical treatment of the Raman spectrum for sarin [57]. Similarly, a modest peak at 510 cm-1 can be assigned to a PC or PO plus skeletal bending mode. Other SERS peaks of modest intensity occur at 875, 1055, 1415, and 1450 cm-1, and based on the spectral analysis of sarin and the Raman spectrum of IMPA with peaks at 880, 1420, and 1455 cm-1, are assigned to a CCC bend, a PO3 stretch, a CH3 bend, and a CH2 rock, respectively.

A B


Figure 25.6. Surface-enhanced Raman spectra of A) IMPA, B) MPA, and C) EMPA. Spectral conditions as in Fig. 25.2, samples were 1 g L-1 in water. MPA has been well characterized by infrared and Raman spectroscopy [58,59], as well as normal coordinate analysis [60], and the literature assignments are used here for the SERS of MPA. The SER spectrum is dominated by a peak at 755 cm-1, which is assigned to the PC symmetric stretch (Figure 25.6B). In comparison to IMPA, it is clear that removing the isopropyl group shifts this frequency substantially (40 cm-1), as the mode becomes a purer PC stretch. Additional peaks with comparatively little intensity occur at 470, 520, 960, 1040, 1300, and 1420 cm-1, and are assigned to a PO3 bending mode, a C-PO3 bending mode, a PO3 stretching mode, another PO3 bending mode, and two CH3 deformation modes (twisting and rocking). SERS-active vials allowed repeatable measurements of MPA at 10 mg L-1 and routine measurements at 1 mg L-1, and repeatable measurements of IMPA at 100 mg L-1 and routine measurements at 10 mg L-1. Again, however, substantial improvements in sensitivity are required to achieve the minimum requirement of 0.004 mg L-1. 25.3.4 VX. The hydrolysis of VX can occur along two pathways (Figure 25.1D) [11,22], either being converted to DIASH and EMPA or EA2192 and ethanol with the former pathway favored four to one. These products also hydrolyze, and EMPA forms MPA and ethanol, while EA2192 forms DIASH and MPA. Here the SER spectra of VX, EA2192 and DIASH are compared, while EMPA is compared to IMPA and MPA. The SER spectrum of VX is similar to its Raman spectrum with corresponding peaks at 375, 460, 540, 730, 1095, 1300, 1440, and 1460 cm-1 (Figure 25.7A). Since a computer predicted Raman spectrum contains most of the measured Raman spectral peaks [43,61], it is used to assign the above SERS peaks respectively to an SPO bend, a CH3-P=O bend, a PO2CS wag, an OPC stretch, a CC stretch, and three CHn bends. As previously described for CEES and HD, the 730 cm-1 peak could alternatively be assigned to a CS stretch, but the SER spectra of these chemicals suggest otherwise. The SER spectrum of EA2192 is somewhat different than VX with the PO modes having limited intensity and the NC3 modes having significant intensity (Figure 25.7B). Specifically, the EA2192 spectrum has moderately intense peaks at 480, 585, 940, and 1125 cm-1 that can be assigned to an NC3 breathing mode, an NCC bending mode, another NC3 stretching mode, and a NCC stretching mode. Two additional peaks with significant intensity at 695

A B C


and 735 cm-1 are assigned to a CS stretching mode and an OPC stretching mode, respectively. Two peaks of modest intensity at 525 and 970 cm-1 are attributed to a PO2S bending mode and a PO2 stretching mode.

Figure 25.7. Surface-enhanced Raman spectra of A) VX, B) EA2192, and C) DIASH. Spectral conditions as in Fig. 25.2, samples were 1 g L-1 in water. The SER spectrum of DIASH contains most of the NC3 modes cited previously for EA2192 (Figure 25.7C), specifically peaks appear at 480, 585, 940, and 1120 cm-1, and can be assigned as above. Additional peaks at 740, 810, and 1030 cm-1, are assigned to CH bending, a combination of SC stretching and NC3 bending, and SCCN bending modes, based on the Raman spectrum of DIASH [43]. A broad peak centered at 695 cm-1 also occurs that has previously been assigned to an SC stretch, but the frequency and intensity of this mode in the HD and CEES spectra above, makes this assignment less certain. It is worth noting the similarity between the EA2192 and DIASH SER spectra, the principle difference being the addition of the SCCN bending mode at 1030 cm-1 for the latter. This may simply be due to the fact that both molecules interact through the sulfur with the metal surface to similar extents resulting in similar spectra. However, it is also possible that the EA2192 spectrum is of DIASH formed either by hydrolysis or photo-degradation. Since the sample was measured within one hour of preparation, and the hydrolysis half-life is on the order of weeks [12], the former explanation seems unlikely. Since the peak intensities did not change during these measurements, photo-degradation catalyzed by silver also seems unlikely. Further experiments are required to clarify this point. The SER spectrum of the other hydrolysis product formed from VX, EMPA, is shown in Figure 25.6. It is included with MPA and IMPA, the hydrolysis products of GB, for convenient spectral comparison of these structurally similar chemicals. The spectrum is dominated by a peak at 745 cm-1 with a substantial low frequency shoulder at 725 cm-1. Both are assigned, similarly to IMPA, to PC or PO plus skeletal stretching modes. In fact, virtually all of the peaks in the SER spectrum correspond to peaks of similar frequency in the SER spectrum of IMPA, and are assigned as follows: the peaks at 480 and 500 cm-1 to PC or PO plus skeletal bends; 890, 1415, and 1440 cm-1 to CHn deformations; 945 and 1060 cm-1 to POn stretches; and 1095 to a CO or CC stretch. A peak at 1285 cm-1 is assigned to a CHn deformation based on the MPA spectral assignment for a peak at 1300 cm-1. In this series of chemicals VX and EA2192 were routinely measured at 100 mg L-1, and on occasion at 10 mg L-1 using the SERS-active vials. Again, however, only a limited number of measurements were attempted. More

A B C


extensive measurements of EMPA using the SERS-active capillaries allowed repeatable measurements of 10 mg L-1 and routine measurements of 1 mg L-1. No concentration studies of DIASH were undertaken. 25.4 Conclusions The ability to obtain surface-enhanced Raman spectra of several chemical agents and their hydrolysis products has been demonstrated using silver-doped sol-gels. Two sampling devices, SERS-active vials and capillaries, provided a simple means to measure water samples containing chemical agents. No sample pretreatment was required and all spectra were obtained in 1 minute. It was found that the SER spectra can be used to identify chemical agents by class. Specifically, cyanide contains a unique peak at 2100 cm-1, HD and CEES both have a unique peak at 630 cm-1, while VX has a unique peak at 540 cm-1. In the case of HD and CEES, their hydrolysis products produce very similar spectra, and it may be difficult to determine relative concentrations in an aqueous solution. In the case of the VX hydrolysis products, EA2192 and DIASH were spectrally similar, as was IMPA and MPA. However, there appears to be sufficient differences when comparing entire spectra, such that chemometric approaches might allow successful compositional analysis of aqueous solutions. The SERS-active vials and capillaries provided sufficient sensitivity to measure cyanide below the required 2 mg L-1 sensitivity either as a point measurement or as a continuous flowing stream measurement. Measurements of TDG suggest that the sensitivity requirements for it and HD may be attainable with modest improvements. In contrast, the vials and capillaries did not provide sensitivity sufficient to meet the requirements of VX. In this case substantial improvements in sensitivity are required and are being pursued. 25.5 Acknowledgements The authors are grateful for the support of the U.S. Army (DAAD13-02-C-0015, Joint Service Agent Water Monitor program) and the Environmental Protection Agency (EP-D-05-034). The authors would also like to thank Mr. Chetan Shende for sol-gel chemistry development. 25.6 References 1 S.L. Hoenig: Handbook of Chemical Warfare and Terrorism. (Greenwood Press, 2002) p. 8, 19, 34-63 2 H. Nozaki, N. Aikawa: Sarin poisoning in Tokyo subway. Lancet 345, 1446 (1995) 3 Committee on Toxicology: Review of Acute Human-Toxicity Estimates for Selected Chemical-Warfare Agents.

(Nat Acad Press, 1997) 4 Committee on Toxicology: Guidelines for Chemical Warfare Agents in Military Field Drinking Water. (Nat

Acad Press, 1995) 5 T.C. Marrs, R.L. Maynard, F.R. Sidell: Chemical Warfare Agents: Toxicology and Treatment. (John Wiley and

Sons, 1996) 6 Material Safety Data Sheets, available at www.msds.com 7 D.R. Lide, Ed: Handbook of Chemistry and Physics: (CRC Press, 1997) p. 8-43 8 N.B. Munro, S.S. Talmage, G.D.Griffin, L.C. Waters, A.P. Watson, J.F. King, V. Hauschild: Environ. Health

Perspect. 107, 933 (1999) 9 A.G. Ogsten, E.R. Holiday, J.St.L.Philpot, L.A. Stocken: Trans. Faraday Soc. 44,45 (1948) 10 G. Wagner, Y. Yang: Ind. Eng. Chem. Res. 41, 1925 (2002) 11 W. Creasy, M. Brickhouse, K. Morrissey, J. Stuff, R. Cheicante, J. Ruth, J. Mays, B. Williams, R. O’Connor,

H. Durst: Environ. Sci. Technol. 33, 2157 (1999) 12 Y. Yang: Acc. Chem. Res. 32, 109 (1999) 13 Y. Yang, J. Baker, J. Ward: Chem. Rev. 92, 1729 (1992) 14 B. Erickson: Anal. Chem. News & Features, 397A (1998) 15 Product literature at http://www.wmdetect.com/Library/M8/M8%20Paper.htm 16 R.M. Black, R.J. Clarke, R.W. Read, M.T. Reid: J. Chromat. 662, 301 (1994) 17 G.A. Sega, B.A. Tomkins, W.H. Griest: J Chromat. A 790, 143 (1997) 18 S.A. Oehrle, P.C. Bossle: J. Chromat. A 692, 247 (1995) 19 J.E. Melanson, B.L-Y. Wong, C.A. Boulet, C.A. Lucy: J. Chromat. A 920, 359 (2001) 20 J. Wang, M. Pumera, G.E. Collins, A. Mulchandani: Anal. Chem. 74, 6121 (2002)


21 W.R. Creasy: Am. Soc. Mass. Spectrom. 10, 440 (1999) 22 Q. Liu, X. Hu, J. Xie: Anal. Chim. Acta 512, 93 (2004) 23 G.A. Eiceman, Z. Caras: Ion Mobility Spectrometry. (CRC Press, 1994) 24 See products from Smiths Detection, Bruker Daltronics, etc. 25 N. Krylova, E. Krylov, G.A. Eiceman: J. Phys. Chem. 107, 3648 (2003) 26 W.E. Steiner, B.H. Clowers, L.M. Matz,W.F. Siems, H.H. Hill Jr.: Anal. Chem. 74, 4343 (2002) 27 L.D. Hoffland, R.J. Piffath, J.B. Bouck: Opt. Eng. 24, 982 (1985) 28 S.D. Christesen: Appl. Spectrosc. 42, 318 (1988) 29 E.H.J. Braue, M.G. Pannella: Appl. Spectrosc. 44, 1513 (1990) 30 C-H. Tseng, C.K. Mann, T.J. Vickers: Appl. Spectrosc. 47, 1767 (1993) 31 S. Kanan, C. Tripp: Langmuir 17: 2213 (2001) 32 D.L. Jeanmaire, R.P. Van Duyne: J. Electroanal. Chem. 84, 1 (1977) 33 A.M. Alak, T. Vo-Dinh: Anal. Chem. 59, 2149 (1987) 34 K.M. Spencer, J. Sylvia, S. Clauson, J. Janni: Proc. SPIE 4577,158 (2001) 35 S.D. Christesen, M.J. Lochner, M. Ellzy, K.M. Spencer, J. Sylvia, S. Clauson: 23rd Army Sci. Conf. (2002) 36 S.D. Christesen, K.M. Spencer, S. Farquharson, F.E. Inscore, K. Gosner, J. Guicheteau: In: S. Farquharson, Ed.

Applications of Surface-Enhanced Raman Spectroscopy. (CRC Press, in preparation) 37 P. Tessier, S. Christesen, K. Ong, E. Clemente, A. Lenhoff, E. Kaler, O. Velev: Appl. Spectrosc. 56, 1524

(2002) 38 Y. Lee, S. Farquharson: Proc. SPIE 4378, 21 (2001) 39 S. Farquharson, P. Maksymiuk, K. Ong, S. Christesen: Proc. SPIE 4577, 166 (2001) 40 S. Farquharson, A. Gift, P. Maksymiuk, F. Inscore, W. Smith, K. Morrisey, S. Christesen: Proc. SPIE 5269, 16

(2004) 41 S. Farquharson, A. Gift, P. Maksymiuk, F. Inscore, W. Smith: Proc. SPIE 5269, 117 (2004) 42 F. Inscore, A. Gift, P. Maksymiuk, S. Farquharson: Proc. SPIE 5585, 46 (2004) 43 S. Farquharson, A. Gift, P. Maksymiuk, F. Inscore: Appl. Spectrosc. 59, 654 (2005) 44 S. Farquharson, P. Maksymiuk: Appl. Spectrosc. 57, 479 (2003) 45 S. Farquharson, Y.H. Lee, C. Nelson: U.S. Patent Number 6,623,977 (2003) 46 S. Farquharson, P. Maksymiuk: U.S. Patent Numbers 6,943,031 and 6,943,032 (2005) 47 S. Farquharson, A. Gift, P. Maksymiuk, F. Inscore: Appl. Spectrosc. 58, 351 (2004) 48 F. Inscore, A. Gift, P. Maksymiuk, J. Sperry, S. Farquharson: In: S. Farquharson, Ed. Applications of Surface-

Enhanced Raman Spectroscopy. (CRC press, in preparation) 49 D. Kellogg, J. Pemberton: J. Phys. Chem. 91, 1120 (1987) 50 J. Billmann, G. Kovacs, A. Otto: Surf. Sci. 92,153 (1980) 51 C.A. Murray, S. Bodoff: Phys. Rev. B 32,671 (1985) 52 S.D. Christesen: J. Raman Spectrosc. 22, 459 (1991) 53 C. Sosa, R.J. Bartlett, K. KuBulat, W.B. Person: J. Phys. Chem. 93, 577 (1993) 54 F. Inscore, S. Farquharson: J. Raman Spectrosc. (submitted) 55 F. Inscore, S. Farquharson: Proc. SPIE 5993, accepted (2005) 56 T. Joo, K. Kim, M. Kim: J. Molec. Struct. 16, 191 (1987) 57 H. Hameka, J. Jensen: CRDEC-TR-326 (1992) 58 R. Nyquist: J. Molec. Struct. 2:123 (1968) 59 B.J. Van der Veken, M.A. Herman: J. Molec. Struct. 15, 225 (1973) 60 B.J. Van der Veken, M.A. Herman: J. Molec. Struct. 15, 237 (1973) 61 H. Hameka, J. Jensen: ERDEC-TR-065 (1993)

SPIE 6378-27 2006 1

Surface-enhanced Raman spectral analysis of blister agents and their hydrolysis products

Frank Inscore and Stuart Farquharson

Real-Time Analyzers, Middletown, CT, 06457

ABSTRACT Protection of military personnel and civilians from water supplies poisoned by chemical warfare agents requires an analyzer that has sufficient sensitivity (µg/L), selectivity (differentiate the warfare agents from its hydrolysis products), and speed (less than 10 minutes) to be of value. In an effort to meet these requirements, we have been investigating the ability of surface-enhanced Raman spectroscopy (SERS) to detect these chemicals in water. The expected success of SERS is based on reported detection of single molecules, the one-to-one relationship between a chemical and its Raman spectrum, and the minimal sample preparation requirements. It is equally important to detect and distinguish the hydrolysis products of these agents to eliminate false-positive responses and evaluate the extent of an attack. Previously, we reported the SER spectra of GA, GB, VX and most of their hydrolysis products, as well as a preliminary study of HD, and its principle hydrolysis product thiodiglycol. Here we expand this study to include half-mustard, its hydrolysis product, 2-hydroxyethyl ethylsulfide, and ethyl ethylsulfide to better characterize the observed SER spectra. We also report the measurement of 10 µg/L of thiodiglycol as we continue to improve sensitivity. Keywords: chemical warfare agent detection, CWA, hydrolysis, SERS, Raman spectroscopy

1. INTRODUCTION Since the terrorist attacks of September 11, 2001, the threat of further attacks of the United States is ever present. The purported use of liquid explosives to destroy US bound airliners in August 2006 is a reminder that many attack scenarios must be considered. One such scenario is the poisoning of water supplies with chemical warfare agents (CWAs). The National Strategy for Homeland Security designates the Environmental Protection Agency with the task of securing the nations drinking water.1 Presently, the EPA employs several field test kits to monitor drinking water supplies, and gas chromatography coupled with mass spectrometry in supporting laboratories to confirm positive responses.2 Unfortunately, these test kits are prone to false-positive responses, and follow-up analysis typically takes a day. This is entirely inadequate for the prevention of widespread illness and potential fatalities. In the past several years we have been developing surface-enhanced Raman spectroscopy (SERS) based sample systems coupled with a field-usable Raman analyzer to detect chemical agents in water at the required μg/L sensitivity and 10 minute timeframe.3-9 The expected success of SERS is based on the million-fold or more increase in Raman signal obtained when a molecule interacts with surface plasmon modes of metal nanoparticles.10 In the case of cyanide, an industrial-based CWA, we have measured below 10 μg /L in one minute.11 The expected success of SERS is also based on the unique set of Raman spectral peaks associated with the molecular vibrational modes of each molecule. The unique SER spectra should not only reduce false-positive responses, but also allow discriminating the CWA hydrolysis products. This is important, since CWAs can hydrolyze rapidly in the presence of water,12 and detection of the hydrolysis products could allow determining 1) the state of an attack (ratio of CWA to hydrolysis product(s)), 2) the point of attack initiation, and 3) the continued extent and severity of the attack throughout a water distribution system. Previously, we used SERS to measure the chemical warfare agent sulfur-mustard (HD) and its primary hydrolysis product thiodiglycol (TDG, Figure 1).9 Here we expand our study to include half-mustard, 2-chloroethyl ethylsulfide (CEES), its hydrolysis product, 2-hydroxyethyl ethylsulfide (HEES), and ethyl ethylsulfide (EES) to better characterize the observed SER spectra. The toxicities of HD and CEES have been studied in mice (LD50s of 19.3 and 566 mg/kg,13 respectively) and are estimated to be very similar for humans (20 mg/kg for HD).14 The military provides a 5-day 5-liter drinking water limit for HD of 140 µg/L,15 but none for CEES. Both HD and CEES are slightly soluble in water at 0.92 and 1.39 g/L, respectively, with hydrolysis half-lives of 8.5 and ~1 min (both at 25 C).12,16 In comparison, TDG is

SPIE 6378-27 2006 2

relatively non-toxic, very water soluble at 690 g/L, and stable in water with a hydrolysis half-life of approximately 6 days (similar properties are assumed for HEES). Accordingly, detection goals to ensure safe water are placed at 10 and 100 µg/L for HD and CEES, respectively, as well as their hydrolysis products.17

HD TDG

CEES HEES Figure 1. Hydrolysis of bis(2-chloroethyl)sulfide (HD) to bis(2-hydroxyethyl)sulfide (TDG); 2-chloroethyl ethylsulfide EES (CEES), to 2-hydroxyethyl ethylsulfide (HEES), and the structure of ethyl ethylsulfide (EES).

2. EXPERIMENTAL

Highly distilled sulfur mustard (bis(2-chloroethyl)sulfide), was measured at the U.S. Army’s Edgewood Chemical Biological Center (Aberdeen, MD). Thiodiglycol (bis(2-hydroxyethyl)sulfide) was purchased as an analytical reference material from Cerilliant (Round Rock, TX). TDG was measured at Real-Time Analyzers, Inc. (RTA, Middletown, CT). All solvents, other analytes, including 2-chloroethyl ethylsulfide (CEES), 2-hydroxyethyl ethylsulfide (HEES), ethyl ethylsulfide (EES), and the chemicals used to prepare the silver-doped sol-gels were obtained from Sigma-Aldrich (St. Louise, MO) and also used as received. The glass vials and capillaries were prepared according to US Patents 6,623, 977, 6,943,031 and 6,943,032.18,19,20 Simply stated, a silver amine precursor and an alkoxide precursor were prepared and mixed. Then 100 μL was placed in a vial and spun or drawn into a 1-mm diameter glass capillary and allowed to gel. After sol-gel formation, the incorporated silver ions were reduced with dilute sodium borohydride. For the purpose of safety, samples were prepared in a chemical hood, transferred to a sampling device and sealed prior to being measured. All samples were measured initially by Raman in their pure state at room temperature as neat liquids. All samples prepared for SERS analysis consisted of 0.1% v/v (1000 ppm) in methanol. The methanol was used to minimize hydrolysis. TDG samples were also prepared at much lower concentrations down to 10 parts-per-billion in HPLC water. HD was measured in SERS-active vials (Simple SERS Sample Vials, RTA), while all other analytes were measured in SERS-active capillaries. Both sampling configurations (vials and capillaries) were mounted horizontally on an XY positioning stage (Conix Research, Springfield, OR), such that the focal point of an f/0.7 aspheric lens was positioned just inside the glass wall. The lens focused the excitation beam into the sample and collected the scattered radiation back along the same axis. A Fourier transform Raman spectrometer (Real-Time Analyzers, model IRA-785, Middletown, CT) was used to acquire the Raman and SER spectra at 8 cm-1 resolution, with 5-min and 1-min acquisition times, and 300 and 100 mW of laser power at the sample respectively, unless otherwise indicated in the figure captions.

H2O + 2HCl

H2O + HCl

SPIE 6378-27 2006 3

3. RESULTS AND DISCUSSION The assignment of the surface-enhanced Raman spectral peaks for HD is not straightforward in that the spectrum is significantly different than the Raman spectrum.21 These differences are attributed to the metal-to-molecule surface interactions, which can shift and enhance various vibrational modes to different extents. The Raman spectrum of HD is dominated by five peaks at 631, 649, 699, 734, and 757 cm-1, in which theoretical calculations assign the first three peaks to CCl stretching modes, and the latter two peaks to CS stretching modes. However, previous assignments of these modes has been questioned,22 and the SER spectrum only contains two distinct peaks in this region, an intense peak at 630 cm-1 and a moderately intense peak at 695 cm-1.23 Consequently, it is difficult to assign the SER peaks. For this reason, the Raman and SER spectra of chemicals similar and simpler in structure have been measured. The Raman and SER spectra of HD, CEES, TDG, HEES and EES are shown in Figures 2 and 3. The Raman spectra for all of these compounds contain peaks in five distinct regions; 300-400 cm-1, 600-800 cm-1, 950-1050 cm-1, 1400-1500 cm-1, and 2800-3100 cm-1. The vibrational mode assignments for each region can in general be assigned to 1) CHn deformation and CSC bending modes, 2) CS, CCl, and CSC stretching modes, 3) CC stretching modes, 4) CHn rocking, twisting, wagging, and scissor modes, and 5) CHn stretching modes, respectively (Table 1). Here the focus is to make vibrational mode assignments to the peaks in the 600-800 cm-1 region. The spectral analysis begins with EES, the simplest chemical with the most confident assignments. The Raman spectrum of EES is dominated by three peaks at 640, 656, and 693 cm-1, which have been assigned to CSC stretching modes. 21,23,24,25, Theoretical calculations performed for the all-trans (TT) C2v structure of EES predict only two peaks at ~655 and 640 cm-1, due to the in- and out-of-phase CSC stretching modes respectively.23 While other studies suggest that the 693 cm-1 peak is due to a TT isomer,25 density function calculations suggest an alternate assignment to CSC stretching that includes a CH3 rocking mode contribution.26 The replacement of a terminal methyl H of EES by an OH group to form HEES reduces the molecular symmetry (C2v to CS) and has the effect of coalescing the three peaks to ~640 (shoulder), 656 and 688 cm-1. Density function calculations suggest that the lower two peaks are CSC stretching modes with CCH3 and COH contributions (i.e. backbone modes), respectively, while the higher frequency peak is again due to CSC stretching with a CH3 rocking contribution. Two peaks at 758 and 776 cm-1 with weak to moderate intensity have been assigned to the asymmetric counterparts of the 640 and 656 cm-1 peaks, but assignment to CH2 wags is also possible. These modes were very weak or absent in the EES spectrum. The replacement of the other terminal methyl H by a second OH group to form TDG regains the C2v molecular symmetry, but additional contribution to the primary CSC mode by the CCOH groups further coalesces the three peaks into a primary peak at 658 cm-1 with high and low frequency shoulders at ~640 and 685 cm-1. The asymmetric modes again gain intensity and now appear at 734 and 770 cm-1. Replacement of the terminal methyl H of EES by a Cl group to form CEES again reduces the molecular symmetry. The Cl strongly influences the CSC backbone modes at 653 cm-1 with 635 and 670 cm-1 shoulders, and produces intense peaks at 699 and 753 cm-1. These peaks can be assigned to CCl stretching modes (symmetric and asymmetric) with significant CSC backbone contribution, but density function calculation suggest that one of these modes could again be due to CSC with a CH2 wag contribution. Finally, adding Cl to both ends of EES to produce HD yields a symmetric molecule with the five peaks previously indicated. However, based on the spectral analysis of the above molecules, the five peaks are now assigned as follows. The 631 and 649 cm-1 peaks are assigned to CSC backbone stretching modes, 699 cm-1 to a backbone mode with CH2 contribution, and 734, and 757 cm-1 to backbone modes with CCl stretching contributions. In the past 5 years, the surface-enhanced Raman spectra have been measured for HD and CEES,3,5,9,11,27 while in the past 20 years, SER spectra of several simple thiols have been measured, and in some cases vibrational modes assigned. 28-34 These measurements and assignments are used to analyze the SER spectra of EES, HEES, TDG, CEES, and HD. The SERS spectra of all of these chemicals, similar to their Raman counterparts, contain peaks in the same five regions described above, and assignments can be made accordingly. However, except for the 600-800 cm-1 and the 950-1050 cm-1 regions, the peaks are usually quite weak in intensity or completely absent. In the specific case of EES, three peaks have reasonable intensity at 629, 967, and 1046 cm-1. The latter two peaks are assigned to CC stretching modes as were the 975 and 1047 cm-1 peaks in the Raman spectrum. The intense 629 cm-1 peak is assigned to the CSC stretching modes (in- and out-of-phase), in which the interaction of the molecule with the silver surface makes these

SPIE 6378-27 2006 4

Figure 2. RS of A) HD, B) CEES, C) TDG, D) HEES and E) EES. Conditions: neat liquids.

Figure 3. SERS of A) HD, B) CEES, C) TDG, D) HEES and E) EES. Conditions: 0.1% v/v in methanol.

A B C D E

A B C D E

SPIE 6378-27 2006 5

modes indistinguishable and shifts their frequency significantly. A similar 26 cm-1 shift has also been observed between the Raman and SERS spectra for ethanethiol (CH3CH2SH) and in other simple alkanethiols.35 This shift has been attributed to the electron donor properties of sulfur and subsequent redistribution of electron density in the CSC bond upon adsorption of the molecule via sulfur to the electropositive silver surface.28,36,37,38 The SER spectrum of HEES is nearly identical to EES, except that the peak intensities for the same concentration are significantly reduced, so much so that only the CSC stretching mode at 629 cm-1 is easily observed. This lack in intensity suggests that the OH group may influence the molecular orientation to the surface. The SER spectrum of TDG is also very similar to the previous two molecules dominated by the CSC stretching mode at 629 cm-1 and minor peaks at 931, 1209 and 1274 cm-1. The relative intensities of these peaks are also reduced possibly due to weakened sulfur-to-silver interaction. It is also worth noting that irradiation of TDG with laser powers at 100 mW and above or for long periods cause photo-degradation resulting in peaks at 715 and 1008 cm-1, possibly due to 2-hydroxy ethanethiol.39 The SER spectrum of CEES also contains an intense peak at 629 cm-1, but also a shoulder at 668 cm-1 and a second moderately intense peak at 722 cm-1. The former peak is assigned as above, while the 722 cm-1 peak is assigned to a CCl stretching mode. This is consistent with the Raman peak assignments above that place the CCl mode at higher frequency than the CSC mode. This mode is also significantly shifted by interaction with silver at 31 cm-1. The 668 cm-1 is tentatively assigned to the CSC stretch with a CH2 contribution. The SER spectrum of HD is also dominated by a peak at 629 cm-1. But the spectrum also contains an extended high frequency shoulder composed of at least two peaks evident at 695 and 830 cm-1, as well as a moderately intense peak at 1045 cm-1. The latter peak is assigned to a CC stretching mode, based on the assignment for a peak at 1040 cm-1 in the Raman spectrum of HD, and the 1046 cm-1 peak in the SER spectrum of EES. The 830 cm-1 peak is left unassigned, while the 695 cm-1 peak is assigned to a CCl stretching mode based on the SER spectrum of CEES, the Raman spectrum of HD, and the fact that this peak does not occur in the TDG, HEES, or EES SER spectra. Table 1. Tentative vibrational mode assignments for Raman and SER spectral peaks for HD, CEES, TDG, HEES, and EES.

EES HEES TDG CEES HD Tentative Assignments Raman SERS Raman SERS Raman SERS Raman SERS Raman SERS

334 340w 348 360w 343m C(CH) deformation 383 380w 403w 402m 379w CSC + CCCl bend

478w 479w 420w 415w 640s 640sh 640sh 635sh 631s CSC stretch 656s 629s 656s 629s 658s 629 s 653s 629s 649s 629 s SC + CCl stretch

670m 693s 688m 685s 685m 699s 668m 699s CH3 rock + backbone

734s 695 m 762w 758w 715w 734m 753s 722s 757s CCl + backbone 779w 776w 770m

807w 822w 826br 819w 853w 830w 947w 947m 931w 930w

975m 967m 975w 977w CC stretch 1018sh 1010m 1012s

1047m

1046m

1047m

1020m 1050m

1043m

1038m 1054m

1019w1054w

1039m 1047m CC stretch

1074m 1064sh 1062m 1171w 1143w 1171w 1197w 1226w 1231m 1209w 1216w CH2 twist/ CH2 wag

1250w 1247w CH2 wag 1273w 1287w 1266w 1289s 1274w 1270w

1289m 1285w 1272m

1293s

1380w 1377w 1382m 1378s 1382w 1408w

1408w

CH3 def

1428m 1413w 1426m CHn def/ CH2 scissor 1430m 1428w 1428m 1424s 1409w 1443m 1442m 1452s 1448w 1453m 1448w 1467s 1463w 1454m 1447w

SPIE 6378-27 2006 6

Due to the hazards associated with HD and CEES, only a limited number of experiments have been performed to evaluate sensitivity. HD was repeatedly observed at 1 g/L and usually observed at 0.1 g/L (100 ppm) in the SERS-active vials,5 while CEES has been measured at similar concentrations. More extensive experiments were performed on TDG since this chemical is safely handled in a regular chemical lab. Previously, TDG was measured in a SERS-active vial at 1 mg/L (Figure 4A). However, the goal for HD is 10 μg/L. To this end a number of functionalized sol-gels were examined to chemically extract TDG from water and concentrate it at the silver surface to improve sensitivity. Figure 4B shows the SER spectrum of a 10 μg/L TDG sample in HPLC grade water extracted and measured using such a sol-gel. This measurement could easily be repeated. Note that 80 mW of laser power causes some photodegradation of the sample, as is evident by the peaks at 715 and 1008 cm-1. These SERS measurements of TDG suggest that the required HD sensitivity may be achievable using this technique. Similar flowing measurements in capillaries for HD, CEES, and HEES are underway.

Figure 4. SER spectra of TDG at A) 1 mg/L in a SER-active vial, and B) at 10 μg/L in a functionalized SER-active capillary. Both spectra obtained using 80 mW of 785 nm laser excitation and 1-min accumulation time.

4. CONCLUSIONS Raman and surface-enhanced Raman spectra of the chemical agents, sulfur-mustard and half-mustard, and their hydrolysis products have been obtained. Comparison of the Raman spectra of sulfur-mustard, half-mustard, thiodiglycol, 2-hydroxyethyl ethylsulfide, and ethyl ethylsulfide provide strong evidence for the following HD assignments; the 631 and 649 cm-1 peaks to CSC backbone stretching modes, the 734, and 757 cm-1 peaks to backbone modes with CCl contributions, and the 699 cm-1 peak to a backbone mode with CH2 contributions. While this region of the Raman spectra is sufficiently unique to identify and differentiate this series of chemicals, the SER spectra are less unique. The SER spectra for all five chemicals are dominated by a 629 cm-1 peak assigned to the CSC stretching mode. This mode appears favorably enhanced by the interaction of the sulfur lone electron pairs to the silver surface. Nevertheless, there are additional peaks in this region that would allow distinguishing HD from CEES and their hydrolysis products, but it may be difficult to differentiate thiodiglycol from 2-hydroxyethyl ethylsulfide based on simple spectral analysis. Further SER measurements and the use of chemometrics are underway. Towards the goal of developing a simple, field-usable method to measure chemical agents in water, we obtained the

A B

SPIE 6378-27 2006 7

SER spectrum of 10 μg/L of thiodiglycol in water in less than 5-minutes using a SER-active capillary. It is likely that similar concentrations can be obtained for HD and CEES, such that water supplies can be monitored for security.

5. ACKNOWLEDGMENTS The authors are grateful for the support of the Environmental Protection Agency (EP-D-05-034). The authors would also like to thank Dr. Steve Christesen for helpful discussions, and Mr. Chetan Shende for sol-gel chemistry development.

6. REFERENCES 1 Whitman, CT, “EPA’s Strategic Plan for Homeland Security”, 2002, available at

http://www.epa.gov/epahome/downloads/epa_homeland_security_strategic_plan.pdf 2 Creasy, W., Brickhouse, M., Morrissey, K., Stuff, J., Cheicante, R., Ruth, J., Mays, J., Williams, B., O’Connor, R., Durst, H.,

“Analysis of chemical weapons decontamination waste from old ton containers from Johnston atoll using multiple analytical methods”, Environ. Sci. Technol., 33, 2157-2162, (1999).

3 Lee, Y.H. and S. Farquharson, "Rapid chemical agent identification by surface-enhanced Raman spectroscopy ", SPIE, 4378, 21-26 (2001).

4 Farquharson, S., Maksymiuk, P., Ong, K., Christesen, S., “Chemical agent identification by surface-enhanced Raman spectroscopy”, SPIE, 4577, 166-173 (2002).

5 Farquharson, S., Gift, A., Maksymiuk, P., Inscore, F., Smith, W., Morrisey, K., Christesen, S., “Chemical agent detection by surface-enhanced Raman spectroscopy”, SPIE, 5269, 16-22 (2004).

6 Farquharson, S., Gift, A., Maksymiuk, P., Inscore, F., Smith, W., “pH dependence of methyl phosphonic acid, dipicolinic acid, and cyanide by surface-enhanced Raman spectroscopy”, SPIE, 5269, 117-125 (2004).

7 Inscore, F., A. Gift, P. Maksymiuk, S. Farquharson, “Characterization of chemical warfare G-agent hydrolysis products by surface-enhanced Raman spectroscopy” SPIE, 5585, 46-52 (2004).

8 Farquharson, S., A. Gift, P. Maksymiuk, F. Inscore “Surface-enhanced Raman spectra of VX and its hydrolysis products”, Appl. Spectrosc., 59, 654-660 (2005).

9 Inscore, F, S Farquharson, “Detecting hydrolysis products of blister agents in water by surface-enhanced Raman spectroscopy”, SPIE, 5993, 19-23 (2005).

10 Jeanmaire, D.L., and R.P. Van Duyne, J. Electroanal. Chem., 84, 1-20 (1977). 11 Christesen, S, K Spencer, S Farquharson, F Inscore, K Gonser, J Guicheteau “Surface-enhanced Raman detection of chemical

agents in water”, in Applications of Surface-Enhanced Raman Spectroscopy, Ed. S Farquharson, CRC Press, Boca Raton, FL, in press.

12 Munro, N.B., Talmage, S.S., Griffin, G.D., Waters, L.C., Watson, A.P., King, J.F., and Hauschild V. “The Sources, Fate, and Toxicity of Chemical Warfare Agent Degradation Products”, Environ. Health Perspect. 107, 933-974 (1999).

13 Gautam, A., Vijayaraghavan, R., Sharma, M., Ganesan, K. “Comparative toxicity studies of sulfur mustard (2,2’-dichloro diethyl sulfide) and monofunctional sulfur mustard (2-chloroethyl ethyl sulfide), administered through various routes in mice”, J. Med. CBR Def., 4 (2006). 14 Committee on Toxicology, Review of Acute Human-Toxicity Estimates for Selected Chemical-Warfare Agents, Nat. Acad.

Press (Washington, D.C.) 1997 15 Hauschild, V. et al. “Short-Term Exposure Guidelines for Deployed Military Personnel”, USACPPM TG 230A (May, 1999)

available at http://chppm-www.apgea.army.mil/imo/ddb/dmd/DMD/TG/TECHGUID/Tg230.pdf 16 Wagner, G., Koper, O., Lucas, E., Decker, S., Klabunde, K. “Reactions of VX, GD, and HD with nanosize CaO: autocatalytic dehydrohalogenation of HD”, J. Phys. Chem.B., 104, 5118-5123 (2000). 17 These values are only estimates made by the authors. 18 Farquharson, S., Lee, Y. H., and C. Nelson "Material for SERS and SERS sensors and method for preparing the same", U.S.

Patent Number 6,623,977 (2003). 19 Farquharson, S. and P. Maksymiuk, “Simultaneous chemical separation and surface-enhanced Raman spectral detection using

metal-doped sol-gels”, U.S. Patent Number 6,943,031 (2005). 20 Farquharson, S. and P. Maksymiuk, “Chemical separation and plural point, surface-enhanced Raman spectral detection using

metal-doped sol-gels”, U.S. Patent Number 6,943,032 (2005). 21 Christesen, S. “Vibrational Spectra and Assignments of Diethyl Sulfide, 2-Chlorodiethyl Sulfide and 2,2’-Dichlorodiethyl

Sulfide”, J. Ram. Spec., 22, 459-465 (1991). 22 Donovan, W., Famini, G., “ Conformational analysis of sulfur mustard from molecular mechanics, semiempirical, and ab-initio

methods”, J. Phys. Chem., 98, 3669-3674 (1994). 23 Sosa, C., R.J. Bartlett, K. KuBulat, and W.B. Person, “A theoretical study of harmonic vibrational frequencies and infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H, Cl)”, J. Phys. Chem., 93, 577-588 (1993). 24 M. Ohsaku, H. Murata, Y. Shiro, “Part XV: C-S stretching vibrations of aliphatic sulfides”, J. Molec. Struct.,42,

SPIE 6378-27 2006 8

31-36 (1977). 25 Ohta, M., Ogawa, Y., Matsuura, H., Harada, I., Shimanouchi, T., “Vibration spectra and rotational isomerism of chain molecules

IV: diethyl sulfide, ethyl propyl sulfide, and butyl methyl sulfide”, Bull. Chem. Soc. Jpn., 50, 380 (1977). 26 Inscore FE, and S Farquharson, in preparation. 27 Spencer, K., Sylvia, J., Clauson, S. Janni, J., Surface-enhanced Raman as a water monitor for warfare agents, in Proc. SPIE Vol.

4577, 158-165 (2002). 28 T. Joo, K. Kim, M. Kim, “Surface-enhanced Raman scattering (SERS) of 1-propanethiol in silver sol”, J. Phys. Chem. 90, 5816-

5819 (1986). 29 C. Kwon, M. Kim, K. Kim, “Raman spectroscopic study of 2-methyl-1-propanethiol in silver sol”, J. Molec. Struct.,16, 201-210

(1987). 30 S. Lee, K. Kim, M. Kim, W. Oh, Y. Lee, “Structure and vibrational properties of methanethiolate adsorbed on silver”, J. Molec.

Struct., 296, 5-13 (1993). 31 M. Schoenfisch, J. Pemberton “Effects of electrolyte and potential on in situ structure of alkanethiol self-assembled monolayers on silver”, Langmuir, 15, 509-517 (1999). 32 S. Cho, E. Park, K. Kim, M. Kim, “Spectral correlation in the adsorption of aliphatic mercaptans on silver and gold surfaces: Raman spectroscopic and computational study” J. Molec. Struct.,479, 83-92 (1999). 33 Kudelski, A., Hill, W., “Raman study on the structure of cysteamine monolayers on silver”, Langmuir, 15, 3162- 3168 (1999). 34 Michota, A., Kudelski, A., Bukowska, J., “ Chemisorption of cysteamine on silver studied by surface-enhanced Raman scattering”, Langmuir, 16, 10236-10242 (2000). 35 C. Kwon, D. Boo, H. Hwang, M. Kim, “Temperature dependence and annealing effects in surface-enhanced Raman scattering on chemically prepared silver island films”, J. Phys. Chem. B., 103, 9610-9615 (1999). 36 Tarabara, V., Nabiev, I., Feofanov, A., “Surface-Enhanced Raman Scattering (SERS) Study of Mercaptoethanol Monolayer Assemblies on Silver Citrate Hydrosol. Preparation and Characterization of Modified Hydrosol as a SERS- Active Substrate”Langmuir, 14, 1092-1098, (1998). 37 Wehling, B., Hill, W., Klockow, D., “Crosslinking of organic acid and isocyanate to silver SERS substrates by mercaptoethanol”, Chem. Phys. Lett., 225, 67-71 (1994). 38 A. Kudelski “Chemisorption of 2-mercaptoethanol on silver, copper, and gold: direct Raman evidence of acid-induced changes in adsorption/desorption equilibria”, Langmuir, 19, 3805-3813 (2003). 39 Inscore FE, and S Farquharson, “Surface-enhanced Raman spectroscopic characterization of the chemical warfare agent HD and

related mono-sulfides” JRS, in preparation.

A SERS-BASED ANALYZER FOR POINT AND CONTINUOUS WATER MONITORING OF CHEMICAL AGENTS AND THEIR HYDROLYSIS

PRODUCTS

STUART FARQUHARSON

FRANK E. INSCORE

Real-Time Analyzers,362 Industrial Park Road (#8), Middletown, Connecticut 06457, United States of America1

stu@rta_biz

Accepted (Day Month Year)

Protection of military personnel and civilians from water supplies poisoned by chemical warfare agents (CWAs) requires an analyzer that has sufficient sensitivity (µg/L, ppb), specificity (differentiate the CWA from its hydrolysis products), and speed (less than 10 minutes) to be of value. In an effort to meet these requirements, we have been investigating the ability of surface-enhanced Raman spectroscopy (SERS) to detect cyanide and sulfur mustard in water. In our work, we have developed a novel SERS-active material that consists of a porous glass with trapped metal particles. Previously, we coated the inside walls of glass vials and measured cyanide at 1 mg/L in water in as little as 1 minute. However, measurements of sulfur mustard have only been measured to 10 mg/L. Recently, we filled glass capillaries with the SERS-active material. Here we describe measurements of cyanide, sulfur mustard, and it’s hydrolysis product, thiodiglycol, using these capillaries and a portable Raman analyzer suitable for point and continuous water monitoring.

1. Introduction The use of chemicals as weapons was introduced during World War I. It is estimated that chlorine, phosgene and sulfur-mustard (HD) resulted in an estimated death of 100,000 soldiers and 1 million injuries.1 In more recent years, chemical agents have been used by terrorists. In Japan, the Aum Shinrikyo religious cult released Sarin (GB) within the Tokyo subway system in 1995.2 The release of GB in this confined space had devastating effects resulting in 12 fatalities and hospitalization of thousands. Countering such attacks requires considering the potential deployment scenarios and having methods and/or devices to detect the chemical warfare agents rapidly, and at very low concentrations. Here we consider the deliberate poisoning of drinking water. This includes water supplies used in military operations and water delivered to major cities from reservoirs and through distribution systems. The required detection sensitivity for each agent depends on several factors, such as toxicity and hydrolysis properties (Table 1). In the case of cyanide, it is known that 4

1 Real-Time Analyzers, 362 Industrial Park Road (#8), Middletown, Connecticut, 06457, United States of America.

102

milligrams per liter of water produces detectable changes in human blood chemistry and 8 mg/L causes severe, but reversible symptoms.3 The military has used this and other toxilogical data to set a field drinking water standard (FDWS) for cyanide at 2 mg/L.4 The FDWS represents the maximum allowable concentration that is assumed safe when 15 L of water per day is consumed over 5 days (expected soldier intake in arid climates). Human toxicity data for the other chemical warfare agents in water have, in general, not been determined. The normal route of exposure for chemical warfare agents is inhalation, and most of the toxicity data is given as the LCt50s,1 the concentration that is lethal to 50% of an exposed population as a function of exposure time. In the case of mustard, animal studies along with the inhalation LCt50, the oral lethal dosage of 0.7 mg per kg of body mass (LD50), and modeling studies,5 have been used to set the FDWS at 0.047 mg/L. The FDWS concentrations have also been used by the military to set the minimum detection requirement for poisons in water.

Table 1. Military field drinking water standard,4 lethal exposures and dosages, 1, 3, 5 and water properties for selected chemical warfare agents.

Chemical FDWS 5-day/15L (mg/L)

LCt50

inhalation (mg-min m-3)

LD50

oral (mg kg-1)

Water Solubility at 25°C


Hydrolysis Product

NaCN (CN) 2 2000 - 480 g/L - CN-

Mustard (HD) 0.047 900 0.7 0.92 g/L 2-30 hours TDG In order to detect these poisons in water, their properties in water must also be considered, i.e. the solubility, rate of hydrolysis, and hydrolysis products formed. In the case of cyanide, as HCN, KCN, or NaCN, all of these chemicals are extremely soluble in water (completely miscible, 716, and 480 g/L, respectively).6 In solution the cyanide ion is formed in equilibrium with the conjugate acid, HCN Reaction (1), according to the Ka of 6.15x10-10. 7 In the case of cyanide then it is important to know the pH, if one form of the chemical is to be detected versus the other. For example, if 2 mg/L of NaCN is added to water (the FDWS), then 1.25 mg/L of CN- and 0.75 mg/L of HCN will be present.

HCN + H2O ↔ CN- + H3O- . (1)

Cl-C2H4-S-C2H4-Cl + 2H2O → HO-C2H4-S-C2H4-OH + 2HCl . (2)

(HD) (TDG) In the case of sulfur-mustard, the situation is somewhat more complex. It is marginally soluble in water tending to form droplets, and hydrolysis occurs at the droplet surface. This property has made measuring the hydrolysis rate constant difficult, and half-lives anywhere from 2 to 30 hours are reported.8 Chemically, the hydrolysis of HD involves the sequential replacement of the chlorine atoms by hydroxyl groups through cyclic sulfonium ion intermediates to form thiodiglycol (TDG), Reaction (2).9 If a median hydrolysis rate is assumed, then early detection of poisoned water will require measuring HD, while post-attack or downstream monitoring will require measuring TDG.

103

The earliest technologies developed for CWA detection were based on electrochemical, ionization, or colorimetric analysis. Examples of the latter include phosgene, M8 and M9 tape, which change color from contact with a sample just like pH paper does. Although these tapes are easy to use, they are not generally agent specific and suffer from a high percentage of false-positives.10 For example, M8 changes color when in contact with common solvents such as acetonitrile, ethanol, methanol, or common petroleum products such as brake fluid, lighter fluid, or WD-40.11 More rigorous laboratory methods have been successfully developed to detect chemical agents with minimum false-positive responses. More than a decade ago, Black et al. demonstrated the ability of combining gas chromatography with mass spectrometry detection (GC/MS) to measure mustard.12 A comprehensive study was performed by Creasy et al. in analyzing chemical weapon decontamination waste from the Johnston Atoll.13, 14 These researchers used liquid chromatography with MS detection and achieved a detection limit of 0.140 mg/L for mustard. Although close to the FDWS, these researchers concede that chromatography coupled MS methods require extraction, derivatization, and repeated column calibration, making them labor intensive, time consuming (typically 30 to 60 minutes), and less than desirable for field use. Another variant of these separation/mass detection technologies is ion mobility spectrometry (IMS).15 Steiner et al. coupled electrospray ionization to the sample entry point of an IMS and a time-of-flight MS was added as an orthogonal detector.16 Water samples spiked with 10 mg/L (10 ppm) thiodiglycol could be measured in 1-min, once sample pretreatment was accomplished. It is worth noting that with proper care these MS-based technologies are likely to detect chemical agents with virtually no false-positives, but detection limits are still insufficient by 1 to 2 orders of magnitude. More rapid analysis of agents in the solid, liquid and gas phase has been demonstrated by vibrational spectroscopy.17, 20, 21 Hoffland et al. reported infrared absorbance spectra and absolute Raman cross sections for several chemical agents,17 while Christesen measured Raman cross sections for mustard gas.18 Again, however these technologies also have limitations. Raman spectroscopy is simply not a very sensitive technique, and detection limits are typically 0.1% (1000 ppm). And infrared spectroscopy would have limited value in analyzing poisoned water, since the very strong infrared absorption of water would obscure most other chemicals present. Nevertheless, efforts to overcome these limitations have been demonstrated by using attenuated total reflectance in the sample cell.19

Enormous improvements in sensitivity for Raman spectroscopy can be achieved through surface-enhancement.22 The interaction of surface plasmon modes of metal particles with target analytes can increase scattering efficiency by as much 14-orders of magnitude, although 6-orders of magnitude are more common. Spencer, et al. used SERS to measure cyanide and mustard on electrochemically roughed gold or silver foils.23, 24, 25 However, the sample needed to be dried on the substrates to obtain the best sensitivity. More recently, Tessier et al. obtained SERS of 0.04 mg/L cyanide in a stream flowing over a

104

substrate formed by a templated self-assembly of gold nanoparticles.26 In this case, optimum sensitivity required introduction of an acid wash and the measurements were irreversible. In the past few years, we have also been investigating the ability of SERS to measure chemical agents at 0.001 mg/L in water and with sufficient spectral uniqueness to distinguish the agent and its hydrolysis products.27, 28, 29, 30, 31, 32 In our work, we have developed silver-doped sol-gels as the SERS-active medium. These sol-gels can be coated on the inside walls of glass vials, such that water samples can be added to perform point-analysis, or they can be incorporated into glass capillaries, such that flowing measurements can be performed.33 Here, both sampling devices were used to measure and compare SER spectra of cyanide, sulfur mustard, and thiodiglycol. In addition, a field-usable Raman analyzer was used to measure 0.001 mg/L cyanide flowing in water with a detection time of less than 1-min. 2. Experimental Sodium cyanide was purchased from Sigma-Aldrich (St. Louis, MO) and thiodiglycol (TDG, bis(2-hydroxyethyl)sulfide) was purchased from Cerilliant (Round Rock, TX). Both chemicals were used as received. Highly distilled sulfur mustard (HD, bis(2-chloroethyl)sulfide) was obtained at the U.S. Army’s Edgewood Chemical Biological Center (Aberdeen, MD) and measured on-site. All samples were initially prepared in a chemical hood as 1000 parts-per-million (1 g/L or 0.1% by volume, Environmental Protection Agency definition) in HPLC grade water (Fischer Scientific, Fair Lawn, NJ) or in some cases methanol or ethanol (Sigma-Aldrich) to minimize hydrolysis. Once prepared, the samples were transferred into 2-ml glass vials internally coated with a silver-doped sol-gel (Simple SERS Sample Vials, Real-Time Analyzers, Middletown, CT) or drawn by syringe or pump into 1-mm diameter glass capillaries filled with the same SERS-active material.34, 35, 36 In the case of flow measurements, a peristaltic pump (variable flow mini-pump, Control Co., Friendswood, TX) was used to flow the various cyanide solutions through a SERS-active capillary at 1 mL min-1. The vials or capillaries were placed on aluminum plates machined to hold the vials or capillaries on a standard XY positioning stage (Conix Research, Springfield, OR), such that the focal point of an f/0.7 aspheric lens was positioned just inside the glass wall. The probe optics and fiber optic interface have been described previously.29 SER spectra were collected using a Fourier transform Raman spectrometer equipped with a 785 nm diode laser and a silicon photo-avalanche detector (IRA-785, Real-Time Analyzers). All spectra were nominally collected using 100 mW, 8 cm-1 resolution, and 1-min acquisition time, unless otherwise noted. Complete experimental details can be found in Reference. 37.37 For added safety, all samples were measured in a chemical hood. In the case of actual agents measured at Edgewood, the FT-Raman instrument was placed outside the laboratory and 30 foot fiber

105

optic and electrical cables were used to allow remote SERS measurements and plate manipulation. 3. Results and Discussion 3.1. CyanideThe surface-enhanced and normal Raman spectra of sodium cyanide are shown in Figure 1A. Sodium cyanide completely dissolves in water forming the ions in equilibrium with the conjugate acid, HCN as described above. Concentrations of 1.0, 0.1, and 0.01 mg/L result in CN- concentrations of 0.52, 0.016, and 0.00021 mg/L as the corresponding pH decreases from just above the pKa of 9.21 at 9.24 to 8.48 and 7.54. This is significant in that only CN- appears to interact sufficiently with silver to produce a SER spectrum, and no spectral signal is observed below pH 7 (except on electrodes at specific potential conditions).38 The SER spectra of cyanide are dominated by an intense, broad peak at 2100 cm-1 attributed to the C≡N stretch (Figure 1). This mode occurs at 2080 cm-1 in Raman spectra of solids, and the frequency shift in SER spectra is attributed to a strong surface interaction, which is supported by the appearance of a low frequency peak at 135 cm-1 due to a Ag-CN stretch (not shown). It is also observed that as the concentration decreases, the CN peak shifts to 2140 cm-1. This shift has been attribute to the formation of a tetrahedral Ag(CN)3

2- surface structure,39 as well as to CN adsorbed to two different surface sites.40 Alternatively, it has also been suggested that at concentrations near and above monolayer coverage, the CN- species is forced to adsorb end-on due to crowding, and at lower concentrations the molecule can reorient to lie flat. This suggests that the 2100 and 2140 cm-1 peaks correspond to the end-on and flat orientations, respectively. However, a previous concentration study of cyanide on a silver electrode observed the reverse trend, i.e. greater intensity was observed for the 2100 cm-1 peak at low concentration.39

Repeated measurements of cyanide in the SERS-active vials consistently allowed measuring 1 mg/L (1 ppm), but rarely below this concentration (Figure 1B). Nevertheless, this sensitivity is in general sufficient for point sampling of water supplies. In the case of continuous monitoring of water, the capillaries are a more appropriate sampling format, and they also allowed routine measurements at 0.01 mg/L and repeatable measurements at 0.001 mg/L (1 ppb, Figure 1C). Employing this format, a 50 mL volume of 0.01 mg/L cyanide solution was flowed at 2.5 mL min-1 through a SERS-active capillary, and spectra were recorded every 20 seconds. As Figure 2 shows, the cyanide peak was easily discerned as soon as the solution entered the capillary and remained relatively stable over the course of the experiment. It is worth noting, as indicated above, that the SERS peak in Figure 2 is in fact due to 210 ng/L!

106

Fig. 1. Surface enhanced Raman spectra of CN in water in silver-doped sol-gel A) coated glass vials and B) filled glass capillaries. All spectra were recorded using 100 mW of 785 nm in 1-min and at a resolution of 8 cm-1.

Fig. 2. 2100 cm-1 peak height measured during continuous flow of a 0.01 mg/L (10 ppb) cyanide in water. Surface-enhanced Raman spectra are shown for 1 and 6 min after sample introduction. A 2.5 mL min-1 flow rate was used and spectra were recorded every 20 sec using 100 mW of 785 nm.

3.2. HD and TDGThe surface-enhanced and normal Raman spectra of HD are shown in Figure 3. The SER spectrum of HD is dominated by a peak at 630 cm-1 with an extended high frequency shoulder composed of at least two peaks evident at 695 and 830 cm-1, as well as a moderately intense peak at 1045 cm-1 (Figure 3A). The latter peak is assigned to a CC stretching mode, based on the assignment for a peak at 1040 cm-1 in the Raman spectrum of HD.41 The assignment of the 630 cm-1 peak is less straightforward, since the Raman spectrum of HD contains five peaks in this region at 640, 655, 700, 740, and 760 cm-1. 29,

41 Theoretical calculations for the Raman spectrum of HD indicate that the first three peaks are due to CCl stretching modes, and the latter two peaks to CS stretching modes.42

Based on these calculations, and the expected interaction between the chlorine atoms and the silver surface, it is reasonable to assign the 630 cm-1 SERS peak to a CCl mode.29 However, recent SERS measurements of diethyl sulfide produced a very simple spectrum with an intense peak at 630 cm-1, 43, 44 strongly suggesting CS or CSC stretching modes as the appropriate assignment for this peak.45 The authors of the theoretical treatment concede that the CCl and CS assignments could be reversed.42 The CS assignment also indicates that HD interacts with the silver surface through the sulfur electron lone pairs. But, interaction between chlorine and silver is still possible and may be responsible for the 695 cm-1 peak. The 830 cm-1 peak may be due to a deformation mode (e.g. CCH). The surface-enhanced and normal Raman spectra of TDG are shown in Figure 4. The SER spectrum of TDG is also dominated by a peak at 630 cm-1 with minor peaks at 820, 930, 1210, and 1275 cm-1 (Figure 4A). Again, the 630 cm-1 peak is preferably assigned to a CSC stretching mode versus a CCl mode, especially since the chlorines have been replaced by hydroxyl groups. Furthermore, the lack of a 695 cm-1 peak in the TDG

107

spectrum supports the assignment of this peak in the HD spectrum to a CCl mode. The 930, 1210 and 1275 cm-1 SERS peaks are assigned to a CC stretch with CO contribution, and two CH2 deformation modes (twist, scissors, or wag) based on the assignments for the corresponding peaks at 940, 1230 and 1290 cm-1 in the Raman spectrum of TDG.41, 43 It is worth noting that irradiation at high laser powers or for extended periods produces peaks at 715 and 1010 cm-1, which are attributed to a degradation product, such as 2-hydroxy ethanethiol.43

Fig. 3. A) SERS and B) RS of HD. A) 0.1% v/v (1000 ppm) in MeOH in a SERS-active vial, 100 mW of 785 nm, 1-min, B) neat sol. in glass container, 300 mW of 785 nm, 5-min.

Fig. 4. A) SERS and B) RS of TDG. A) 0.1% v/v in MeOH in SERS-active capillary, 100 mW of 785 nm, 1-min, B) neat sol. in glass capillary, 300 mW of 785 nm, 5-min.

Only a limited number of measurements of HD were performed to evaluate sensitivity, due to the safety requirements. HD was repeatedly observed at 1000 ppm and usually observed at 100 ppm in the SERS-active vials.29 But even at the latter concentration, substantial improvements in sensitivity are required to approach the required 0.05 mg/L (50 ppb) sensitivity. More extensive experiments were performed on HD’s hydrolysis product, TDG since this chemical is safely handled in a regular chemical lab. Repeated measurements of TDG in the SERS-active vials consistently allowed measuring 10 ppm, but repeated measurements of 1 ppm did not yield any discernable peaks (lowest trace in Figure 5). Flowing TDG through SERS-active capillaries allowed repeatable measurements at 10 ppm, and routine measurements at 1 ppm as shown in Figure 6.44 There is an important difference between the TDG spectra recorded for static and flowing samples, namely that the 715 cm-1 peak is noticeably more intense in the static sample. This suggests that it may represent a photo-degradation product, which was recently verified.46

108

Fig. 5. SERS of 1000, 100, 10 and 1 ppm TDG in water (top to bottom). All in SERS-active capillaries, 100 mW of 785 nm, 1-min.

Fig. 6. SERS of 1 ppm TDG in water flowing through a SERS-active capillary at 1, 2, 3, 4, and 5 min. (top to bottom), 100 mW of 785 nm, 1-min each.

4. Conclusions The ability to obtain surface-enhanced Raman spectra of two chemical agents, cyanide and mustard, and their hydrolysis products has been demonstrated using silver-doped sol-gels. Two sampling devices, SERS-active vials and capillaries, provided a simple means to measure water samples containing chemical agents. No sample pretreatment was required and all spectra were obtained in 1 minute. The SERS-active vials and capillaries provided sufficient sensitivity to measure cyanide below the required 2 mg/L sensitivity either as a point measurement or as a continuous flowing stream measurement. Measurements of TDG suggest that the sensitivity requirements for it and HD may be attainable with modest improvements. 5. Acknowledgements The authors are grateful for the support of the U.S. Army (DAAD13-02-C-0015, Joint Service Agent Water Monitor program) and the Environmental Protection Agency (EP-D-05-034). The authors would also like to thank Mr. Chetan Shende for sol-gel chemistry development.

References 1. S. L. Hoenig, Handbook of Chemical Warfare and Terrorism (Greenwood Press, London,

2002) 2. H. Nozaki and N. Aikawa, Sarin poisoning in Tokyo subway, Lancet 345, 1446 (1995). 3. Committee on Toxicology, Review of Acute Human-Toxicity Estimates for Selected Chemical-

Warfare Agents (Nat Acad Press, Washington, DC, 1997). 4. Committee on Toxicology, Guidelines for Chemical Warfare Agents in Military Field

Drinking Water (Nat Acad Press, Washington, DC, 1995). 5. T. C. Marrs, R. L. Maynard and F. R. Sidell, Chemical Warfare Agents: Toxicology and

Treatment (John Wiley and Sons, 1996).

109

6. Material Safety Data Sheets, available at www.msds.com. 7. D. R. Lide, Ed, Handbook of Chemistry and Physics (CRC Press, Boca Raton, 1997). 8. N. B. Munro, S. S. Talmage, G. D. Griffin, L. C. Waters, A. P. Watson, J. F. King and V.

Hauschild, The Sources, Fate, and Toxicity of Chemical Warfare Agent Degradation Products, Environ. Health Perspect. 107, 933 (1999).

9. A. G. Ogsten, E. R. Holiday, J. St. L. Philpot and L. A. Stocken, The replacement reactions of b,b'-dichlorodiethyl sulphide and of some analogues in aqueous solution: the isolation of b-chloro-b'-hydroxydiethyl disulphide, Trans. Faraday Soc. 44, 45 (1948).

10. B. Erickson, The Chemical Weapons Convention Redefines Analytical Challenge, Anal. Chem. News & Features June 1: 397A (1998).

11. Product literature at http://www.wmdetect.com/Library/M8/M8%20Paper.htm. 12. R. M. Black, R. J. Clarke, R. W. Read and M. T. Reid, Application of gas chromatography-

mass spectrometry and gas chromatography-tandem mass spectrometry to the analysis of chemical warfare samples, found to contain residues of the nerve agent sarin, sulphur mustard and their degradation products, J. Chromat. 662, 301 (1994).

13. W. Creasy, M. Brickhouse, K. Morrissey, J. Stuff, R. Cheicante, J. Ruth, J. Mays, B. Williams, R. O’Connor and H. Durst, Analysis of chemical weapons decontamination waste from old ton containers from Johnson atoll using multiple analytical methods, Environ. Sci. Technol. 33, 2157 (1999).

14. W. R. Creasy, Postcolumn Derivatization Liquid Chromatography/Mass Spectrometry for Detection of Chemical-Weapons-Related Compounds, Am. Soc. Mass. Spectrom. 10, 440 (1999).

15. G. A. Eiceman and Z. Caras, Ion Mobility Spectrometry (CRC Press, Boca Raton, 1994). 16. W. E. Steiner, B. H. Clowers, L. M. Matz, W. F. Siems and H. H. Hill Jr., Rapid screening of

aqueous chemical warfare agent degradation products: ambient pressure ion mobility mass spectrometry, Anal. Chem. 74, 4343 (2002).

17. L. D. Hoffland, R. J. Piffath and J. B. Bouck, Spectral signatures of chemical agents and simulants, Opt. Eng. 24, 982 (1985).

18. S. D. Christesen, Raman cross sections of chemical agents and simulants, Appl. Spectrosc. 42, 318 (1988).

19. E. H. J. Braue and M. G. Pannella, Circle Cell FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutions, Appl. Spectrosc. 44, 1513 (1990).

20. C-H. Tseng, C. K. Mann and T. J. Vickers, Determination of Organics on Metal Surfaces by Raman Spectroscopy, Appl. Spectrosc. 47, 1767 (1993).

21. S. Kanan and C. Tripp, An infrared study of adsorbed organophosphonates on silica: a prefiltering strategy for the detection of nerve agents on metal oxide sensors, Langmuir 17, 2213 (2001).

22. D. L. Jeanmaire and R. P. Van Duyne, Surface Raman Spectroelectrochemistry. J. Electroanalytical Chem. 84, 1 (1977).

23. K. M. Spencer, J. Sylvia, S. Clauson and J. Janni, Surface Enhanced Raman as a Water Monitor for Warfare Agents in Water, Proc. SPIE 4577, 158 (2001).

24. S. D. Christesen, M. J. Lochner, M. Ellzy, K. M. Spencer, J. Sylvia and S. Clauson, Surface Enhanced Raman Detection and Identification of Chemical Agents in Water, 23rd Army Science Conference (2002).

25. S. D. Christesen, K. M. Spencer, S. Farquharson, F. E. Inscore, K. Gosner and J. Guicheteau, Surface Enhanced Raman Detection of Chemical Agents in Water, In: S. Farquharson, Ed. Applications of Surface-Enhanced Raman Spectroscopy (CRC Press, Boca Raton, in preparation).

26. P. Tessier, S. Christesen, K. Ong, E. Clemente, A. Lenhoff, E. Kaler and O. Velev, On-line spectroscopic characterization of sodium cyanide with nanostructured gold surface-enhanced Raman spectroscopy substrates, Appl. Spectrosc. 56, 1524 (2002).

110

27. Y. Lee and S. Farquharson, Rapid chemical agent identification by SERS. Proc. SPIE 4378,

21 (2001). 28. S. Farquharson, P. Maksymiuk, K. Ong and S. Christesen, Chemical agent identification by

surface-enhanced Raman spectroscopy, Proc. SPIE 4577, 166 (2001). 29. S. Farquharson, A. Gift, P. Maksymiuk, F. Inscore, W. Smith, K. Morrisey and S. Christesen,

Chemical agent detection by surface-enhanced Raman spectroscopy, Proc. SPIE 5269, 16 (2004).

30. S. Farquharson, A. Gift, P. Maksymiuk, F. Inscore and W. Smith, pH dependence of methyl phosphonic acid, dipicolinic acid, and cyanide by surface-enhanced Raman spectroscopy, Proc. SPIE 5269, 117 (2004).

31. F. Inscore, A. Gift, P. Maksymiuk and S. Farquharson, Characterization of chemical warfare G-agent hydrolysis products by surface-enhanced Raman spectroscopy, Proc. SPIE 5585, 46 (2004).

32. S. Farquharson, A. Gift, P. Maksymiuk and F. Inscore, Surface-enhanced Raman spectra of VX and its hydrolysis products, Appl. Spectrosc. 59, 654 (2005).

33. S. Farquharson and P. Maksymiuk, Simultaneous chemical separation and surface-enhancement Raman spectral detection using silver-doped sol-gels, Appl. Spectrosc. 57, 479 (2003).

34. S. Farquharson, Y. H. Lee and C. Nelson, Material for surface-enhanced Raman spectroscopy and SER sensors, and method for preparing same, U.S. Patent Number 6,623,977 (2003).

35. S. Farquharson and P. Maksymiuk, Simultaneous chemical separation and surface-enhanced Raman spectral detection using metal-doped sol-gels, U.S. Patent Number 6,943,031. Separation and Plural-point surface-enhanced Raman spectral detection using metal-doped sol-gels, U.S. Patent Number 6,943,032 (2005).

36. S. Farquharson, A. Gift, P. Maksymiuk and F. Inscore, Rapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopy, Appl. Spectrosc. 58, 351 (2004).

37. F. Inscore, A. Gift, P. Maksymiuk, J. Sperry and S. Farquharson, In: S. Farquharson, Ed. Applications of Surface-Enhanced Raman Spectroscopy (CRC press, Boca Raton, in preparation).

38. D. Kellogg and J. Pemberton, Effects of solution conditions on the surface-enhanced Raman scattering of cyanide species at silver electrodes, J. Phys. Chem. 91, 1120 (1987).

39. J. Billmann, G. Kovacs and A. Otto, Enhanced Raman effect from cyanide adsorbed on a silver electrode. Surf. Sci. 92, 153 (1980).

40. C. A. Murray and S. Bodoff, Depolarization effects in Raman scattering from cyanide on silver island films, Phys. Rev. B 32, 671 (1985).

41. S. D. Christesen, Vibrational spectra and assignments of diethyl sulfide, 2-chlorodiethyl sulfide and 2,2’-dichlorodiethyl sulfide, J. Raman Spectrosc. 22, 459 (1991).

42. C. Sosa, R. J. Bartlett, K. KuBulat and W. B. Person, A theoretical study of harmonic vibrational frequencies and infrared intensities of XCH2CH2SCH2CH2X and XCH2CH2SH (= H, Cl), J. Phys. Chem. 93, 577 (1993).

43. F. Inscore and S. Farquharson, Surface-enhanced Raman Spectroscopic characterization of sulfur mustard, half-mustard, their hydrolysis products and related mono-sulfides, J. Raman Spectros. (in preparation).

44. F. Inscore and S. Farquharson, Detecting hydrolysis products of blister agents in water by surface-enhanced Raman spectroscopy, Proc. SPIE, 5993, 19-23 (2005).

45. T. Joo, K. Kim and M. Kim, Surface-enhanced Raman study of organic sulfides adsorbed on silver, J. Molec. Struct. 16, 191 (1987)

46. S. Farquharson, F. E. Inscore and S. Christesen, In K. Kneipp, M. Moskovits and H. Kneipp, Eds., Surface-Enhanced Raman Scattering – Physics and Applications (Springer-Verlag, Berlin, in press, 2006).

111

SPIE 6540-06 2007 1

Water Security: Continuous Monitoring of Water Distribution Systems for Chemical Agents by SERS

Frank Inscore, Chetan Shende, Atanu Sengupta, and Stuart Farquharson

Real-Time Analyzers, Inc. 362 Industrial Park Rd (#8), Middletown, CT 06457

ABSTRACT Ensuring safe water supplies requires continuous monitoring for potential poisons and portable analyzers to map distribution in the event of an attack. In the case of chemical warfare agents (CWAs) analyzers are needed that have sufficient sensitivity (part-per-billion), selectivity (differentiate the CWA from its hydrolysis products), and speed (less than 10 minutes) to be of value. We have been investigating the ability of surface-enhanced Raman spectroscopy (SERS) to meet these requirements by detecting CWAs and their hydrolysis products in water. The expected success of SERS is based on reported detection of single molecules, the one-to-one relationship between a chemical and its Raman spectrum, and the minimal sample preparation requirements. Recently, we have developed a simple sampling device designed to optimize the interaction of the target molecules with the SERS-active material with the goal of increasing sensitivity and decreasing sampling times. This sampling device employs a syringe to draw the water sample containing the analyte into a capillary filled with the SERS-active material. Recently we used such SERS-active capillaries to measure 1 ppb cyanide in water. Here we extend these measurements to nerve agent hydrolysis products using a portable Raman analyzer. Keywords: water security, chemical agents, CWA, hydrolysis, SERS, Raman spectroscopy

1. INTRODUCTION The terrorist attacks of 2001 were a stark reminder that such attacks can take many forms. Consequently, likely deployment scenarios must be considered, followed by the development of technology to detect such events and treat victims. One such scenario is the poisoning of water supplies by toxic chemicals, such as chemical warfare agents (CWAs). The Environmental Protection Agency, responsible for securing the nations drinking water, has developed protocols for collecting samples, sending them to designated labs, and performing analyses in the case of an attack (real or suspected). 1,2 However, it is highly desirable to have portable monitors to perform analysis at the point of a suspected attack, as well as permanent monitors installed at major drinking water sources and distribution systems to ensure public safety on a continuous basis. There are three major classes of chemical warfare agents, blood agents, such as cyanide, blister agents, such as mustard gas, and nerve agents, such as sarin. The nerve agents are a particular concern due to their previous use in Iraq and Japan,3,4 their extreme toxicity (LD50 man for GB = 25 mg/kg),5 and their persistence (hydrolysis half-life of several days)6 and relatively high solubility in water (5-25 g/L, see Table 1). Consequently, any technology designed to detect nerve agents in water must also be able to detect their hydrolysis products to accurately characterize the extent of an attack.7,8 The nerve agents, GB, GD, and GF are very similar in structure, each an alkyl substituted oxy methyl-phosphonofluoridate, where the alkyl groups, represented by R, are isopropyl, pinocolyl, and cyclohexyl groups, respectively (Figure 1A). Hydrolysis occurs in two steps, first hydrofluoric acid and the corresponding alkoxy methylphosphonic acid are formed (isopropyl methylphosphonic acid (IpMPA), pinacolyl methylphosphonic acid (PMPA), and cyclohexyl methylphosphonic acid (CMPA), respectively), and second the alcohols and a common final product, methylphosphonic acid (MPA), are formed at a much slower rate.9,10 GA hydrolyzes in a similar fashion (Figure 1B, first hydrogen cyanide and ethyldimethyl amidophosphoric acid (EDMAPA) are formed followed by ethanol and dimethylamido phosphoric acid (DMAPA). In contrast, VX can hydrolyze according to two different pathways (Figure 1C).10,11 For the preferential pathway, 80% of VX is hydrolyzed to 2-(diisopropylamino) ethanethiol (DIASH), which is stable in water and ethyl methylphosphonic acid (EMPA), which further hydrolyzes to form MPA and ethanol. For the less favorable pathway, 20% of VX is hydrolyzed to EA2192 (common name) and ethanol, and at a much slower rate, EA2192 hydrolyzes to DIASH and MPA. In this study we have also measured one of the primary hydrolysis products of RVX (Russian VX). For RVX, the ethoxy group of VX has been replaced by an isobutoxy group. Here the corresponding hydrolysis product, isobutyl methylphosphonic acid (IbMPA), was measured. Finally, the required

SPIE 6540-06 2007 2

sensitivity to perform adequate measurements to characterize an attack is on the order of 10 part-per-billion (EPA definition for µg/L). This is based on the short term (< 7 days) US military maximum exposure guidelines for drinking water (Table 1).12

Table 1. Properties of chemical agents and their primary hydrolysis products investigated in the present study.2 Chemical Agent Lethal Dose (LD50) Hydrolysis ½ life Water Solubility (25°C) Sensitivity Sarin (GB) 25 mg/kg 39 hr (pH 7) completely miscible 12 ppb IMPA stable 4.8 g/L MPA very stable >1000 g/L Soman (GD) 5 mg/kg 45 hr (pH 6.6) 21 g/L (@20°C) 12 ppb PMPA stable no data Cyclosarin (GF) 5 mg/kg slower than GB 3.7 g/L 12 ppb CMPA no data no data Tabun (GA) 4.1 hours 12 ppb EDMAPA VX 0.012 mg/kg > 3 days 150 g/L 20 ppb EMPA > 8 days 180 g/L RVX 0.011 mg/kg IbMPA

Figure 1. Molecular structure of the nerve agents and their hydrolysis products. A) GB, GD, and GF have near identical structures except for the change in the R group as shown, and all produce MPA, B) GA is somewhat different and does not produce MPA, while C) VX hydrolyzes via two different pathways, and both produce MPA. Several technologies have recently been investigated as potential at-site analyzers for chemical warfare agents, as well as their hydrolysis products.10,13 This includes liquid chromatography combined with mass spectrometry (LC/MS),14-17 infrared spectroscopy18,19,20 and Raman spectroscopy.21 However, LC/MS remains a labor intensive technique, infrared

A B C

SPIE 6540-06 2007 3

is limited by the strong absorption of water which obscures much of the spectrum, while Raman spectroscopy does not have sufficient sensitivity.21 In the past few years, we have explored the potential of surface-enhanced Raman spectroscopy (SERS) to detect CWAs,22 and their degradation products.23 The utility of SERS is based upon the extreme sensitivity of this technique and the ability to identify molecular structure through the abundant vibrational information provided by Raman spectroscopy. SERS employs the interaction of surface plasmon modes of metal particles with the target analytes to increase scattering efficiency by as much as 1 million times.24 In our studies metal-doped sol-gels were employed to promote the SERS effect. The porous silica network of the sol-gel matrix offers a unique environment for immobilizing and stabilizing SERS-active metal particles of both silver and gold.25 Previously, we used this SERS-active sol-gel to internally coat glass vials and measured MPA, IpMPA, PMPA, CMPA,22,23,26 as well as cyanide,22,23 HD,27 and VX.28,29 However, except for cyanide, the necessary sensitivity for CWA hydrolysis products was not achieved. To improve sensitivity we have been investigating the use of different alkoxides used to synthesize the sol-gel in terms of their ability to chemically select the target analytes. Furthermore, we have incorporated the SERS-active sol-gel into glass capillaries for point, as well as flow measurements. In the former case a syringe is used to rapidly draw a sample into the capillary, which is then placed on a portable Raman spectrometer for analysis. Recently, we used these capillaries to measure thiodiglycol, the primary hydrolysis product of HD, at 1 ppm, close to the requirement of 0.1 ppm.30 Here we provide analysis of the nerve agent hydrolysis products using these SERS-active capillaries with a focused study on the detection limits for methylphosphonic acid.

2. EXPERIMENTAL Except for methylphosphonic acid (MPA), the hydrolysis degradation chemicals measured in this study, isopropyl methylphosphonic acid (IpMPA), pinacolyl methylphosphonic acid (PMPA), cyclohexyl methylphosphonic acid (CMPA), ethyldimethyl amidophosphoric acid, EDMPA, ethyl methylphosphonic acid (EMPA), and isobutyl methyl-phosphonic acid (IbMPA), were obtained as analytical reference materials from Cerilliant (Round Rock, TX) and used without further purification. MPA, methanol and ethanol, and all chemicals used to prepare the silver-doped sol-gel coated capillaries were acquired from Sigma-Aldrich (St. Louis, MO) and used as received. HPLC grade water was obtained from Fischer Scientific (Fair Lawn, NJ). All samples were either obtained at forensic concentrations at 1 mg/mL in methanol (e.g. IbMPA) and diluted by a factor of 10, or prepared in a chemical hood at 0.1 mg/mL or 0.01% by volume (100 parts-per-million, ppm) in HPLC grade water, or in some cases methanol to minimize hydrolysis. For MPA lower concentrations were prepared by serial dilution, and all solutions were stored at 10°C until needed. SERS-active capillaries were prepared using the following general procedure. A silver-doped sol-gel solution was prepared from a mixture of two precursors, a silver amine precursor prepared from ammonium hydroxide and AgNO3, and an alkoxide precursor prepared from tetramethyl orthosilicate, octadecyltrimethoxysilane, and methyltrimethoxy-silane. Details are provided in US Patents 6,943,031 and 6,943,032.31,32 Then 100 μL of the mixture was drawn into a 1-mm diameter glass capillary and allowed to gel. After sol-gel formation, the incorporated silver ions were reduced with dilute sodium borohydride. The capillaries were then ready for performing SERS. For the purpose of safety, samples were prepared in a chemical hood, transferred to the SERS-active capillaries, which were sealed prior to being measured. In each case a 50 µL sample of analyte was drawn into a SERS-active capillary mounted horizontally on an XY positioning stage (Conix Research, Springfield, OR), and measured (Figure 2).33 A Fourier transform Raman spectrometer (Real-Time Analyzers, model IRA-785, Middletown, CT) equipped with a 785 nm diode laser (Process Instruments Inc. model 785-600, Salt Lake City, UT) and a silicon photo-avalanche detector (Perkin Elmer model C30902S, Stamford, CT) was used to deliver 75 or 300 mW of power to the samples generating SERS or Raman spectra with 8 cm-1 resolution, respectively. A software program was used to automatically collect spectra from either single or 9 consecutive points spaced 1 mm apart (Figure 2A).

SPIE 6540-06 2007 4

3. RESULTS AND DISCUSSION

The SER spectra of the seven nerve agent hydrolysis products measured are shown in Figure 3. The hydrolysis product common to almost all of the nerve agents is methylphosphonic acid. MPA also has the simplest molecular structure, which has been well characterized by infrared, Raman, and recently surface-enhanced Raman spectroscopy.34,35,26 Just as the Raman spectrum, the SERS spectrum of MPA is dominated by the symmetric PC stretch, which produces a peak at 756 cm-1 (Figure 3A). Other Raman modes are enhanced to a much lesser extent, and only the CH3 twist is evident at 1300 cm-1. Like MPA, the SERS of isopropyl methylphosphonic acid, is dominated by a peak at 716 cm-1 (Figure 3B). However, this peak is not simply due to the PC stretch, but includes a considerable amount of the backbone CPOCC mode created by the addition of the isopropyl group. The spectrum also contains a moderate peaks at 772 cm-1 that may also be a PC containing backbone mode, as has been suggested by a theoretical treatment of the vibrational modes for sarin.36 It is also worth noting that the Raman spectrum of IpMPA is very similar to that of a published spectrum of sarin.21 Two peaks of modest intensity at 508 and 1055 cm-1 are assigned to PO3 modes, based on assignments to peaks observed for the Raman spectrum of MPA. Similarly, the moderately intense peaks at 874 and 1416 cm-1 are assigned to methyl rocking and bending modes based on MPA. Not surprisingly, the isopropyl group not only increased the intensity of these bands, but also gives rise to additional CH3 and CH2 wagging modes at 1388 cm-1 and 1451 cm-1. The SER spectrum of pinacolyl methylphosphonic acid is also dominated by a peak at 750 cm-1 with a shoulder at 729 cm-1 (Figure 3C). These peaks are both likely PC stretching modes with different backbone contributions. Only three other peaks have any intensity, the CH3 wag at 543 cm-1, the PO3 stretch at 1037 cm-1, and the CH2 bend at 1444 cm-1. The SERS of cyclohexyl methylphosphonic acid is in many ways like IpMPA with the addition of cyclohexane modes (Figure 3D). This includes peaks at 622, 1023, and 1262 cm-1, that are attributed to ring CC stretching modes, and peaks at 811 and 1443 cm-1 that are assigned to ring CH2 modes. The most intense peak observed at 749 cm-1 is again assigned to a PC stretch with backbone contribution. The structure of ethyldimethyl amidophosphoric acid has the most significant changes compared to the other hydrolysis products, and even though the PC mode is replaced by a PN mode, the spectrum is still dominated by a peak at 782 cm-1

with a shoulder at 805 cm-1 (Figure 3E). This suggests that the backbone contribution to the peak is significant. Nevertheless four peaks appear at 566, 1022, 1098, and 1399 cm-1 that are attributed to NC2 stretch, CNC bend, NC stretch, and NC2 bending modes, respectively. The peaks at 1441 cm-1 and its shoulders are assigned to CH2 and CH3 bending modes as before.

Figure 2. A) Screen capture of capillary positioning and spectral acquisition software, B) photograph of portable Raman spectrometer and C) XY plate reader, and D) 1-mm SERS-active capillary.

A B

D C

SPIE 6540-06 2007 5

Figure 3. Surface-enhanced Raman spectra of the primary hydrolysis products of nerve agents with molecular structures. A) methylphosphonic acid (MPA), B) isopropyl methylphosphonic acid (IpMPA), C) pinacolyl methylphosphonic acid (PMPA), D) cyclohexyl methylphosphonic acid (CMPA), E) ethyldimethyl amidophosphoric acid, EDMAPA, F) ethyl methylphosphonic acid (EMPA), and G) isobutyl methylphosphonic acid (IbMPA). Conditions: all spectra collected at 100 ppm or lower, using silver-doped sol-gel filled 1 mm glass capillaries, 75 mW of 785 nm laser excitation, 1 minute (or less, one position) acquisition. Grey line, included for comparison, is at 750 cm-1.

A) MPA B) IpMPA C) PMPA D) CMPA E) EDMAPA F) EMPA G) IbMPA

Ram

an In

tens

ity (R

elat

ive)

600 800 1000 1200 1400 Wavenumber (cm-1)

SPIE 6540-06 2007 6

The structure of ethyl methylphosphonic acid is very similar to isopropyl methylphosphonic acid, but the SER spectrum is somewhat different. The dominant PC stretching mode is split into two peaks at 727 and 746 cm-1, presumably because of the backbone contributions (Figure 3F). The peaks at 1059, 1416 and 1441 cm-1 are assigned to a PO3 stretch, and CH3 and CH2 bending modes as before. Again, although the structure of isobutyl methylphosphonic acid is quite similar to both ethyl and isopropyl methylphosphonic acid, it produces a unique SERS spectrum. In particular, a relatively intense peak appears at 821 cm-1, which is assigned to a CCC2 bend or stretching mode of the isobutyl group (Figure 3G). Similarly, a peak at 485 cm-1 may also be assigned to this functionality. The other peaks at 743, 1063, and 1457 cm-1 are assigned to the PC plus backbone mode, a PO3 stretch, and CH3 and/or CH2 bending modes as before. The importance of the SER spectra for these seven nerve agent hydrolysis products is not the vibrational mode assignments, but the fact that these molecules are SERS-active (not all chemicals are), and the spectra are sufficiently different for identification. Although it must be stated, that it would be difficult to determine the concentration of MPA relative to PMPA, CMPA, EMPA or IbMPA in a mixture. For these hydrolysis reactions it would be more appropriate to compare the SER spectra of the primary hydrolysis products to the original nerve agent spectra. At this point, only VX has been measured using the silver-doped sol-gels, and fortunately, the spectrum is quite different than EMPA.28 Next, the limit of detection for these SERS-active capillaries was examined. MPA was used as a representative molecule. A series of capillaries were prepared and MPA was measured at 1000, 500, 250, 125, 50, 25, and 10 ppb. In each case 9 points spaced 1 mm apart along the length of the capillary were measured. All 9 spectra were averaged in each case with no attempt to identify and discard outliers. The averaged spectra for 1000, 500, 250, and 125 ppb are shown in Figure 4. As can be seen, the PC stretching mode at 760 cm-1 is barely discernable at 125 ppb. The measurements for 50, 25, and 10 ppb gave sporadic results. The uniformity of the capillaries is reflected in the standard deviation of the 9 values for each concentration, which are included in a plot of concentration versus 760 cm-1 peak height (Figure 4 inset). The standard deviation for the 250 ppb values was ±0.036 or 12%, and as shown in Figure 5, this was largely due to one point, which could be considered an outlier. Note also that the SERS relationship between the concentration and signal is not linear. This is expected since the signal levels off as the coverage on the available surface of the metal approaches a monolayer.

Figure 4. MPA at 1000, 500, 250, and 125 ppb (μg/L) measured in SERS-active capillaries. Conditions: 75 mW of 785 nm at the sample, 9 averaged positions, 1 min acquisition each. Inset: plot of 760 cm-1 peak height vs. concentration.

SERS MPA

ppb

1000

500

250

125

0

0.025

0.05

0.075

0 100 200 300 400 500 600 700 800 900 1000Concentration (ppb)

Peak

hei

ght

SPIE 6540-06 2007 7

The performance of the SERS-active capillaries compared to the SERS-active vials was also evaluated. Previously we measured MPA at 10 ppm using the SERS-active vials with a predicted detection limit of 100 ppb based on the signal-to-noise ratio.23 Although this measurement was repeatable, detection was not obtained on every vial. Here, the situation is similar in that some positions on the capillaries did produce spectra, but as mentioned, the signals were not consistent. For example, several of the positions along the SERS-capillary produced MPA spectra at 10 ppb (Figure 5B).

Figure 5. A) 250 ppb MPA measured in SERS-active capillary at 9 positions. Note 1 position produced a signal 50% greater than other positions. Inset: Magnified view of 760 cm-1 PC stretching mode for clarity. B) 10 ppb MPA measured at one position of a capillary. Conditions: 75 mW of 785 nm, 1 min each position.

4. CONCLUSION The ability to obtain surface-enhanced Raman spectra of most of the primary hydrolysis products of nerve agents were obtained using silver-doped sol-gel coated capillaries at 100 ppm or lower in one minute. In general, the SER spectra for the alkyl methylphosphonic acid hydrolysis products were dominated by one or two peaks between 715 and 765 cm-

1, which have been assigned to PC stretching modes with varying amounts of backbone mode contributions. It is clear from the present study that the hydrolysis products can easily be identified as a class by these 700 cm-1 peaks, but quantifying each in a mixture that also contains MPA and the parent nerve agent may require chemometric approaches. In the case of methylphosphonic acid a detailed concentration study demonstrated reproducible measurements to 125 ppb, close to the required detection requirement of 12 ppb. Although SER spectra of 10 ppb MPA were obtained, the results were sporadic. Efforts to improve consistency and reliability are ongoing. It is worth noting that these measurements were all performed in water without sample pretreatment, and therefore the capillaries could be used for point measurements used to establish the extent and distribution of an agent during an attack,7 or they could be incorporated into a sampling system for continuous analysis. It is also worth noting that an analyzer capable of measuring these hydrolysis products at such low concentrations would also be valuable in establishing prior presence of nerve agents through soil and groundwater analysis,15,37 or in verifying successful destruction during decommissioning operations,9,38,39 and establishing extent of exposure during an attack.

5. ACKNOWLEDGMENTS The authors are grateful for the support from the Environmental Protection Agency (EP-D05-034 and EP-D-06-084) and the National Science Foundation (DMI-0215819).

A B

SPIE 6540-06 2007 8

6. REFERENCES 1. Whitman, CT, “EPA’s Strategic Plan for Homeland Security”, 2002, available at

http://www.epa.gov/epahome/downloads/epa_homeland_security_strategic_plan.pdf 2. EPA, “Response Protocol Toolbox: Planning for and Responding to Drinking Water Contamination Threats and

Incidents. Module 3: Site Characterization and Sampling Guide, Module 4: Analytical Guide” December 12, 2003, available at http://www.epa.gov/safewater/security/ertools.html#compendium

3. Hoenig, S.L. Handbook of Chemical Warfare and Terrorism, Greenwood Press (Westport, CT) 2002. 4. Nozaki, H. and Aikawa, N. “Sarin poisoning in Tokyo subway”, Lancet, 345 1446-1447 (1995). 5. Committee on Toxicology. Review of Acute Human-Toxicity Estimates for Selected Chemical-Warfare Agents,

Nat. Acad. Press (Washington, D.C.) 1997. 6. Munro, N.B., Talmage, S.S., Griffin, G.D., Waters, L.C., Watson, A.P., King, J.F., and Hauschild V. “The

Sources, Fate, and Toxicity of Chemical Warfare Agent Degradation Products”, Environ. Health Perspect., 107, 933-974 (1999).

7. Hui, D.-M. and Minami, M. “Monitoring of fluorine in urine samples of patients involved in the Tokyo sarin disaster, in connection with the detection of other decomposition products of sarin and the by-products generated during sarin synthesis”, Clin. Chim. Acta, 302, 171-188 (2000).

8. McKone, T.E., Huey, B.M., Downing, E., and Duffy, L.M., Editors. Strategies to Protect the Health of Deployed U.S. Forces: Detecting, Characterizing, and Documenting Exposures, Nat. Acad. Press (Washington, D.C.) p.207, 1999.

9. Wagner, G. and Yang, Y. “Rapid nucleophilic/oxidative decontamination of chemical warfare agents”, Ind. Eng. Chem. Res., 41, 1925-1928, (2002).

10. Creasy, W., Brickhouse, M., Morrissey, K., Stuff, J., Cheicante, R., Ruth, J., Mays, J., Williams, B., O’Connor, R., and Durst, H. “Analysis of chemical weapons decontamination waste from old ton containers from Johnston atoll using multiple analytical methods”, Environ. Sci. Technol., 33, 2157-2162, (1999).

11. Qin Liu, Xuying Hu and Jianwei Xie “Determination of nerve agent degradation products in environmental samples by liquid chromatography–time-of-flight mass spectrometry with electrospray ionization” Analytica Chimica Acta, Volume 512, Issue 1, 4 June 2004, Pages 93-101

12. Hauschild, V. et al. “Short-Term Exposure Guidelines for Deployed Military Personnel”, USACPPM TG 230A (May, 1999), but see 2003 values at http://chppm-www.apgea.army.mil/documents/TG/TECHGUID/TG230RD.pdf

13. “The Chemical Weapons Convention Redefines Analytical Challenge”, Analytical Chemistry News & Features, June 1, 397A (1998).

14. Sega, G.A., Tomkins, B.A., and Griest, W.H. “Analysis of methylphosphonic acid, ethyl methylphosphonic acid and isopropyl methylphosphonic acid at low microgram per liter levels in groundwater” J. Chromatography A, 790, 143-152 (1997).

15. D’Agustino, P.A, Hancock, J.R., and Provost, L.R. “Determination of sarin, soman and their hydrolysis products in soil by packed capillary liquid chromatography-electrospray mass spectrometry”, J. Chromatography A, 912, 291-299 (2001).

16. Creasy, W.R. “Postcolumn Derivatization Liquid Chromatography/Mass Spectrometry for Detection of Chemical-Weapons-Related Compounds” Am. Soc. Mass Spectrom., 10, 440-447 (1999).

17. Liu, Q., Hu, X., and Xie, J. “Determination of nerve agent degradation products in environmental samples by liquid chromatography–time-of-flight mass spectrometry with electrospray ionization”, Analytica Chimica Acta, 512, 93-101 (2004).

18. Hoffland, L.D., Piffath, R.J., and Bouck, J.B. “Spectral signatures of chemical agents and simulants”, Optical Engineering, 24, 982-984, (1985).

19. Braue, E.H.J., and Pannella, M.G. “CIRCLE CELL FT-IR Analysis of Chemical Warfare Agents in Aqueous Solutions”, Applied Spectroscopy, 44, 1513-1520, (1990).

20. Kanan, S. and Tripp, C. “An infrared study of adsorbed organophosphonates on silica: a prefiltering strategy for the detection of nerve agents on metal oxide sensors”, Langmuir, 17, 2213-2218, (2001).

21. Christesen, S.D. “Raman cross sections of chemical agents and simulants”, Appl. Spec., 42, 318-321 (1988). 22. Farquharson, S., Gift, A., Maksymiuk, P., Inscore, F., Smith, W., Morrisey, K., and Christesen, S. “Chemical agent

detection by surface-enhanced Raman spectroscopy”, SPIE, 5269, 16-22 (2004). 23. Farquharson, S., Gift, A., Maksymiuk, P., Inscore, F., and Smith, W. “pH dependence of methyl phosphonic acid,

dipicolinic acid, and cyanide by surface-enhanced Raman spectroscopy”, SPIE, 5269, 117-125 (2004).

SPIE 6540-06 2007 9

24. Weaver, M.J., Farquharson, S., and Tadayyoni, M.A. “Surface-enhancement factors for Raman scattering at silver

electrodes”, J. Chem. Phys., 82, 4867-4874 (1985). 25. Farquharson, S.; Lee, Y. H., and C. Nelson “Material for SERS and SERS sensors and method for preparing the

same”, U.S. U.S. Patent Number 6,623,977 (2003). 26. Inscore, F, A Gift, P Maksymiuk, and S Farquharson, “Characterization of chemical warfare G-agent hydrolysis

products by surface-enhanced Raman spectroscopy”, SPIE, 5585, 46-52 (2005). 27. Inscore, F, S Farquharson, “Surface-enhanced Raman spectral analysis of blister agents and their hydrolysis

products” SPIE, 6328, (2006). 28. Farquharson, S, A Gift, P Maksymiuk, and F Inscore, “Surface-enhanced Raman spectra of VX and its hydrolysis

products”, Applied Spectroscopy, 59, 654-660 (2005). 29. Christesen, S, K Spencer, S Farquharson, F Inscore, K Gonser, J Guicheteau “Surface-enhanced Raman detection

of chemical agents in water”, in Applications of Surface-Enhanced Raman Spectroscopy, Ed. S Farquharson, CRC Press, Boca Raton, FL, accepted

30. Farquharson, S, F Inscore, “A SERS-based analyzer for point and continuous water monitoring of chemical agents and their hydrolysis products”, IJHSES, 20, 102-111 (2007).

31. Farquharson, S. and P. Maksymiuk, “Simultaneous chemical separation and surface-enhanced Raman spectral detection using metal-doped sol-gels”, U.S. Patent Number 6,943,031 (2005).

32. Farquharson, S. and P. Maksymiuk, “Chemical separation and plural point, surface-enhanced Raman spectral detection using metal-doped sol-gels”, U.S. Patent Number 6,943,032 (2005).

33. Farquharson, S., Gift, A., Maksymiuk, P., and Inscore, F. “Rapid dipicolinic acid extraction from Bacillus spores detected by surface-enhanced Raman spectroscopy”, Appl. Spec. 58, 351-354 (2004).

34. Nyquist, R. “Vibrational spectroscopic study of (R-PO3)2ˉ”, J. Mol. Struct., 2, 123-135, (1968). 35. Van der Veken, B.J. and Herman, M.A. “Vibrational analysis of methylphosphonic acid and its anions: I.

Vibrational spectra”, J. Molec. Struct., 15, 225-236 (1973). 36. Hameka, H. and Jensen, J. “Theoretical prediction of the infrared spectra of nerve agents”, CRDEC-TR-326, 1992. 37. Johnston, R.L., Hoefler, C.M., Fargo, J.C., and Moberley, B. “The Defense Nuclear Agency’s Chemical/

Biochemical Weapons Agreements Verification Technology Research, Development, Test and Evaluation Program and its Requirements for On-Site Analysis”, AT-ONSITE, 5-8 (1994).

38. Yang, Y., Baker, J., and Ward, J. “Decontamination of chemical warfare agents”, Chem. Rev., 92, 1729-1743 (1992).

39. Christesen, S., MacIver, B., Procell, L., Sorrick, D., Carrabba, M., and Bello, J. “Nonintrusive analysis of chemical agent identification sets using a portable fiber-optic Raman spectrometer”, Appl. Spec., 53, 850-855 (1999).

Epa Epd06084 Final Report

Documents

Transcript of Epa Epd06084 Final Report