Surface detection of chemical warfare agent simulants and degradation products

Analytica Chimica Acta 553 (2005) 148–159

Surface detection of chemical warfare agent simulants and degradation products Abu B. Kanu a , Paul E. Haigh b , Herbert H. Hill a,∗ a

Department of Chemistry, Washington State University, Pullman, WA 99164-4630, USA b GE Ion Track, 205 Lowell Street, Wilmington, MA 01887, USA

Received 17 February 2005; received in revised form 15 July 2005; accepted 3 August 2005 Available online 5 October 2005

Abstract Chemical warfare agent (CWA) simulants as well as their degradation and hydrolysis products were detected from surfaces using thermal desorption ion mobility spectrometry (TD-IMS). CWA simulant materials that closely mimic the chemical structures of real CWA G/V-type nerve and S-type vesicant simulants were used in this study. Reduced mobility constants (K0 ) in air were reported for 20 compounds studied. Spectra for sample materials as low as 1 ng deposited on a paper filter were produced for most of the compounds. Detection limits as low as 15 pg of sample material with a sensitivity of 3.2 × 102 ampere per gram (A g−1 ) were reported. TD-IMS, which is normally used for the detection of explosives and drugs of abuse, demonstrated the capability of separating and detecting mixtures of CWA simulants, degradation and hydrolysis products from surface samples. TD-IMS demonstrated clear advantages of speed, high throughput and versatility over chromatographic methods of analysis for detecting CWA simulants, degradation and hydrolysis products. Successful development of the technique may lead to a practical and simple sensor for CWA and related compounds that could be installed and used at sensitive locations around the USA and throughout the world. © 2005 Elsevier B.V. All rights reserved. Keywords: Chemical warfare agent simulants; Degradation products; Thermal desorption ion mobility spectrometry; Surface detection; Sensors

1. Introduction Throughout history, the periodic use of chemical weapons for war and terror has created a need for the development of rapid detection and sensitive analytical methods and instrumentation. The use of nerve agents in 1988 that killed thousands in Kurdish villages and the 1991 gulf war further emphasized the threat of chemical warfare. The events in the United States in 2001, of course, again focused the world’s attention on the specter of terrorist attack. The chemical weapons convention, [1] enforced in April 1997, bans the development, production, stockpiling and use of chemical weapons by member nations and produced a requirement for rapid detection schemes for nerve agents and their degradation and hydrolysis products. Onsite verification procedures to monitor suspected production and storage facilities require methods for detecting and identifying chemical weapons, their degradation and hydrolysis products

∗

Corresponding author. Tel.: +1 509 335 5684; fax: +1 509 335 8867. E-mail address: [email protected] (H.H. Hill).

0003-2670/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.aca.2005.08.012

which are rapid, portable, selective and sensitive. Monitoring surfaces for trace quantities of deposited agents and/or their degradation products would provide a relative simple method for the determination of the presence or past presence of chemical warfare agents. CW agents are commonly classified as nerve, vesicant or blood born agents. Nerve agents – Sarin (GB), Soman (GD), Tabun (GA), and VX – disrupt neurological regulation within biological systems through the inhibition of organophosphorous cholinesterase [2,3]. Vesicant agents – sulfur mustard gas (H, HD, HT), lewisite (L), nitrogen mustard gas (HN-1, HN-2, HN-3), and phosgene-oxime (CX) – are typically responsible for blistering action, damaging eyes, mucus membranes and the respiratory tract [4]. Blood born agents – hydrogen cyanide (AC), or cyanogen chloride (CK) – prevent tissue utilization of oxygen by inhibition of cytochrome oxidase [5,6]. All of these compounds hydrolyze to form stable degradation products [7–9]. The degradation pathway for G-type nerve agents is a quick hydrolysis to form alkyl phosphonic acid and esters [10]. V/VX type nerve agents degrade to form alkyl phosphonothioic acids, and various alkyl amino ethanol compounds

A.B. Kanu et al. / Analytica Chimica Acta 553 (2005) 148–159

[11]. The degradation products of sulfur mustard are thiodiglycol, thiodiglycol sulfone, thiodiglycol sulfoxide, thiodiglycol disulfide, 1,4-dithiane and 1,4-thioxane [12]. Finally, blood born agents such as AC initially hydrolyze to formamide, and subsequently to ammonium formate; while CK readily hydrolyzes to hydrogen chloride and unstable cyanic acid. The cyanic acid further decomposes to carbon dioxide and ammonia [13]. A widely established method for detecting these compounds has been derivatization followed by chromatography. Derivatization has been used to form fluorescing chromophores [14] for detection after liquid chromatographic separation or derivatization to make these compounds sufficiently volatile for analysis by gas chromatography–mass spectrometry (GCMS) [15]. Both methods require extensive sample preparation and are generally expensive and time consuming. In recent years a number of rapid separation and sensitive detection systems have become available. Capillary zone electrophoresis (CZE) combined with indirect UV detection at 210 nm has been demonstrated to be another means to separate and detect hydrolysis products of nerve agents as well as hydrolytic or oxidative and degradation products of HD. Because of its relative ease of analysis its utility in the field and on-site analysis is feasible [16]. Liquid chromatography–electrospray ionisation–MS (LC-ESI-MS) was utilized as a rapid screening method for the hydrolysis products of nerve agents in aqueous samples and extracts [17]. In this same report the application of LC-ESI/APCI-MS and LC-ESI/APCI-MS-MS to the identification of dialkyl esters of phosphonic acids was described. High-resolution atmospheric pressure ESI-IMS-MS has also been used for the analysis of chemical warfare degradation products in which quantitative studies were reported [18]. Ion mobility spectrometry (IMS) is a rapidly advancing technique that has wide applications for explosives, narcotics and CWAs. While attempts to place some analytical methods into field venues have been frustrated by weight-power requirements, IMS instruments have been developed which are portable and hand-held analyzers [19,20]. Changes in the gas composition in the IMS can be used to enhance sensitivity and selectivity in the IMS response. For example, gas phase chloride ions from a suitable dopant increase sensitivity of IMS towards vapors of explosives. Doping the ion source of the IMS with acetone improves its selectivity towards organo-phosphorous compounds, including nerve agents as used in chemical agent monitors [21–23]. Detection of CWA with IMS has, for the most part, been limited to gas phase samples using a 63 Ni or corona ionization source. For verification of the past presence of CWA it is necessary to detect these compounds or their degradation products on the surfaces on which they have settled or condensed. ESI has been used for the detection of CWAs in water, but the ability of IMS to separate and detect CWAs and their degradation products deposited on surfaces has not been investigated. In this paper we described a novel use of IMS for the detection of chemical warfare agent simulants and degradation products from surfaces.

149

2. Experimental 2.1. Materials and reagents Because of the high toxicity of CWAs, less toxic structural analogs that directly mimic or imitate the actual CWAs were utilized to evaluate instrumental response. A total of 20 compounds comprising of five CWA simulants, two CWA degradation products and 13 CWA hydrolysis products were studied. Five of the 20 compounds studied (dimethyl methyl phosphonate, DMMP; pinacolyl methylphosphonate, PMP; diethyl phosphoramidate, DEPA; 2-chloroethyl ethyl sulfide, 2-CEES; and 2-(butylamino) ethanethiol, 2-BAET) were chosen to mimic or imitate the actual CWAs. DMMP was chosen as a simulant for the presence of a P O, P-CH3 and P-OCH3 functional groups that are found in both GB and GD. PMP was chosen as a simulant because its structure are identical to that of GD, except that the fluorine atom in GD is substituted by an OH group. DEPA was chosen as a simulant because it contains a P O, P-NH2 and P-OCH2 CH3 functional groups which is found in GA. Similarly, 2-BAET was selected as a simulant because it possesses both C S and C N bonds found in the functional groups of VX which are attached to the phosphorus atom. One of these simulants, 2-CEES, was used as a structurally identical simulant for HD with the exception of a missing chloride atom. Thus, the CWA simulants used in this study were the following: 97% DMMP, 97% PMP, 98% DEPA, 97% 2BAET, and 98% 2-CEES were obtained from Sigma Aldrich Chemical Company (St. Louis, MO). CWA hydrolysis and degradation products – 1,4-dithiane (1,4-DT), 1,4-thioxane (1,4TO), thiodiglycol sulfoxide (TDS), ethyl methyl phosphonic acid (EMPA), cyclohexyl methyl phosphonic acid (CHMPA), and diisopropyl methyl phosphonate (DIMP) were obtained as 1 mg ml−1 certified reference materials (CRM) from Cerilliant (Auston, TX). Alkyl amines – cyclopentylamine (CPA), cyclohexylamine (CHA), cycloheptylamine (CHPA), cyclooctylamine (COA), dipropylamine (DPA), tripropylamine (TPA), nbutylamine (NBA) dihexylamine (DHA) and decylamine (DA) were obtained from Sigma Aldrich Chemical Company (St. Louis, MO), purity ≥ 99%. Standards were prepared using a micropipette (Brinkmann, Westbury, NY). A known amount of the analyte was pipetted into 30 ml polypropylene nalgene vials (Fisher Scientific, Tustin, CA). The resulting transfer was diluted with methanol (J.T. Baker, Phillipsburg, NJ) to 10 ml total volume. This was followed by serial dilutions where necessary. The stock solutions were stored in the refrigerator at 4 ◦ C when not in use. In Table 1, the structures and CAS numbers for all compounds are shown. Sample introduction into the instrument was achieved by pipetting a 1 ␮l aliquot of sample onto a paper disk, 50 ␮m thick × 90 mm wide (GE Ion Track, Wilmington, MA) or a gold seal® micro glass slide (VWR, Brisbane, CA). Concentrations of the liquid test standards were the following: DMMP (0.12 mg cm−3 ), PMP (0.10 mg cm−3 ), DEPA (0.09 mg cm−3 ), 2-CEES (0.11 mg cm−3 ), 2-BAET (0.09 mg cm−3 ), 1,4-DT (1 mg cm−3 ), TDS (1 mg cm−3 ), EMPA (1 mg cm−3 ), CHMPA

150

A.B. Kanu et al. / Analytica Chimica Acta 553 (2005) 148–159

Table 1 Chemical structures and CAS numbers of compounds studied Name

DMMP (dimethyl methyl phosphonate)

PMP (pinacolyl methyl phosphonate)

DEPA (diethyl phosphoramidate)

2-CEES (2-chloroethyl ethylsulfide)

2-BAET (2-(butylamino) ethanethiol)

1,4-DT (1,4-dithiane)

1,4-TO (1,4-thioxane)

TDS (thiodiglycol sulfoxide)

EMPA (ethyl methyl phosphonic acid)

CHMPA (cyclohexyl methyl phosphonic acid)

DIMP (diisopropyl methyl phosphonate)

Chemical structure

A.B. Kanu et al. / Analytica Chimica Acta 553 (2005) 148–159

151

Table 1 (Continued ) Name

Chemical structure

2,2 -TDE (2,2 -thiodiethanol)

CPA (cyclopentyl amine)

CHA (cyclohexyl amine)

CHPA (cycloheptyl amine)

COA (cyclooctyl amine)

DPA (dipropyl amine)

TPA (tripropyl amine)

NBA (n-butyl amine) DHA (dihexyl amine) DA (decyl amine) Me: methyl; Et: ethyl.

(1 mg cm−3 ), EMPA (1 mg cm−3 ), DIMP (1 mg cm−3 ), CPA (0.09 mg cm−3 ), CHA (0.08 mg cm−3 ), CHPA (0.09 mg cm−3 ), COA (0.09 mg cm−3 ), DPA (0.07 mg cm−3 ), TPA (0.08 mg cm−3 ), NBA (0.08 mg cm−3 ), DHA (0.08 mg cm−3 ) and DA (0.08 mg cm−3 ). All injections on filters or glass slides were allowed to dry before being introduced to the instrument. The solvent methanol, evaporated rapidly at room temperature leaving the analyte of interest on the surface of the filters or glass slides. The filters or slides were then placed in the detection slot of the vapor desorption unit. Filters were preconditioned in the desorber until clean. They were allowed to cool before they were used for sample introduction. The gold seal® micro glass slides were baked in an oven overnight at 200 ◦ C before use. 2.2. Instrumentation The IMS used in this study was a thermal desorption ion mobility spectrometer (TD-IMS) (Itemiser, GE Ion Track, Wilmington, MA) [24–26]. The TD-IMS instrument had a drift length of 3.9 cm, a drift voltage of 980 V, and a drift tube temperature of 205 ± 5 ◦ C. The drift voltage was also applied to the

field free region. The pulse electric field established across the drift tube was 251 V cm−1 and flow rate into the TD-IMS was 1000 ml min−1 . Pressure in the drift tube during mobility measurements was 703–708 torr. Fig. 1 is a schematic cross-section of the TD-IMS detector. A thermal desorption chamber attached to the front end of the instrument allowed samples to be volatized into the TD-IMS. The desorber was operated at 185 ◦ C when the instrument was in the explosive detection mode and at 220 ◦ C when it was in the drug detection mode. For this investigation the drugs mode desorber was used, and temperature of the desorber was held constant at 220 ◦ C. Desorbed vapors were drawn into the TD-IMS through a semi-permeable elastomeric membrane. The membrane, excludes dust and dirt from the system, protecting the detector from contamination. The air-flow to the TD-IMS drift tube was scrubbed air containing a low concentration of ammonia or methylene dichloride as a dopant. The dopant was necessary because interfering organic vapors can be easily transmitted through the membrane into the sample air stream and would produce responses in the detector. These responses are eliminated by the addition of a trace of dopant vapor in the gas stream entering the detector. The dopant was

152

A.B. Kanu et al. / Analytica Chimica Acta 553 (2005) 148–159

Fig. 1. Schematic cross-section of the thermal desorption ion mobility spectrometry detector.

chosen to extract charge from background ions of lesser proton affinity or electronegativity so that CWA simulants, degradation and hydrolysis products produced a single-response peak in the spectrum. Sample molecules passing through the membrane were carried into the detector by a carrier gas. The gas flowed through an ionization chamber, 6-mm long, whose walls were lined with a 10 mCi 63 Ni radioactive foil, emitting low energy beta particles. As the carrier gas flowed through the chamber, both positive and negative reactant ions were formed. These reactant ions then interacted with neutral sample vapors to produce product ions. The electric field in the detector’s reaction chamber was normally zero, but at 20 ms intervals, short potential pulses were applied across the chamber. A kick-out pulse of 0.2 ms created an electric field that forced the ions through an open grid electrode “E1” (Fig. 1) and into the ion drift region where they were propelled towards the collector electrode by a constant and continuous electric field. The kick-out pulse was square in shape with amplitude of 1600 V. The open grid electrode defines the field free region for ionization and prevents the field in the drift region from getting into the ionization region, thus it maintained the integrity of the field free region. The velocities of the ions are related to their size to charge ratio; thus, a measurement of its velocity makes substance identification possible. The collector and related electronics passed an analogue signal to the system’s computer for digital conversion, analysis, and identification. The TD-IMS pseudo real-time software provided procedures to change various operational parameters and select display properties. Pseudo in this case means the instrument took an average of four scans, i.e. each plasmagram displayed was an average of four scans. Real time would mean that every scan could be observed and that the response would be displayed as it was detected. The report used all averaged ion mobility spectra. Four view modes were available (select scan, plasmagram, intensity map, and processed 3D). The plasmagram view mode was used to collect all data in this study which displayed an

average of 70 IMS spectra on the screen. The data shown in this study, however, were of a single spectral scan, i.e. the average of four scans. Raw spectral data were converted into text files and copied onto a floppy disk drive connected to the back of the instrument. Resulting files were then processed using Microsoft Excel 2000. 2.3. Experiments conducted In this study, a sampling time of 5 s, and an analysis time of 2 s were used; thus, the total analysis time for each sample was 7 s. The following experiments were conducted to determine ion mobilities, reproducibility, resolution, response and surface effects on CWA simulants and their degradation products. 2.4. Mobility measurements Each compound listed above was introduced into the TDIMS to determine their reduced mobility value. These values were determined for compounds desorbed from both filters and glass slides. K0 values were calculated based on a standard (see Eq. (1)). Cocaine was selected as the standard and instrument calibration was conducted every 24 h. 2.4.1. Repeatability measurements Triplicate mobility measurements for each compound were obtained using both filters and glass slides, taken over a 3-day period. Reproducibility of K0 values were obtained in both the positive ion mode using NH3 as the dopant gas and the negative ion mode with CH2 Cl2 as the dopant gas. 2.4.2. Calibration and sensitivity studies Calibration studies for two CWA simulants detected on both filters and gold seal® micro glass slides were conducted. From the calibration graphs sensitivity of TD-IMS was estimated.

A.B. Kanu et al. / Analytica Chimica Acta 553 (2005) 148–159

2.4.3. IMS resolving power Data obtained from experiments 1 and 2 were used to calculate resolving power and the number of theoretical plates for each compound. 2.4.4. IMS resolution Experiments were conducted to investigate how well TDIMS can separate mixtures. 2.4.5. Comparison of surfaces The final studies were conducted by applying a known amount of DMMP (117 ng), EMPA (50 ng), COA (92 ng) and DHA (80 ng) in methanol individually as well as mixtures on four surfaces (briefcase, stainless steel, wood and 1/2 holland pavers concrete). Application solutions were allowed to dry on the surface before filters were used to wipe the surface and desorbed in the instrument. 2.5. Calculations All reduced mobility constants (K0 ) were determined by comparison to known mobilities and from experimentally determined drift times (td ). The reduced mobility constants were calculated with reference to cocaine using Eq. (1) to correct for variations in temperature and pressure: K0(unknown) =

drift timecocaine K0(cocaine) drift timeunknown

(1)

where 1.16 ± 0.02 cm2 V−1 s−1 is the K0 value for cocaine [27]. In IMS, separation efficiencies are typically reported as resolving power and are calculated based on the following

153

equation: Rp =

td wh

(2)

where Rp is the IMS resolving power, td is the drift time, and wh is the ion pulse duration measured at half the maximal intensity [28,29]. Eq. (3) shows the relationship between Rp and the number of theoretical plates. N = 5.55(Rp )2

(3)

where N is the number of theoretical plates and Rp is the resolving power. In IMS systems, resolution (a measure of the ability to resolve two compounds) is different from Rp (a measure of the sharpness of the ion pulse), and can be determined from: √ Rp α − 1 N α−1 = (4) R= 1.70 α 4 α where R is resolution, N is number of theoretical plates, α is the selectivity factor defined as K01 /K02 . 3. Results and discussion 3.1. Mobility measurements Tables 2 and 3 list K0 values obtained using both glass slides and filters for sample introduction to the TD-IMS for all compounds whose responses were discernable. The K0 values reported in Tables 2 and 3 were calculated from drift times obtained when CWA simulants or degradation products were introduced to the instrument. For many of these compounds

Table 2 Reduced mobilities (K0 ), resolving power (Rp ) and theoretical plates (N) obtained for ions from studies using glass slides Compound

Day 1

Day 2

Day 3

K0 (lit K0 )

Rp

N

K0 (lit K0 )

Rp

N

K0 (lit K0 )

Rp

N

DMMP DEPA

1.74 1.61

16 16

1421 1421

1.74 1.61

14 18

1088 1798

1.74 1.61

15 16

1249 1421

2-BAET

1.67 1.20

21 13

2448 938

1.67 1.20

18 18

1798 1798

1.67 1.21

19 18

2004 1798

TDS EMPA

1.73 1.61

14 19

1088 2004

1.73 1.61

13 19

938 2004

1.74 1.61

14 19

1088 2004

DIMP

1.51 1.03a

12 4

799 89

1.52 1.03

11 4

672 89

1.52 1.03

12 4

799 89

2,2 -TDE CPA CHA CHPA COA DPA TPA NBA DHA DA

1.80 1.90 (1.94) 1.81 (1.84) 1.71 (1.72) 1.62 (1.64) 1.84 (1.87) 1.61 (1.66) 1.97 (1.97) 1.31 (1.34) 1.35 (1.35)

26 16 16 18 15 20 19 19 12 18

3752 1421 1421 1798 1249 2220 2004 2004 799 1798

1.79 1.91 (1.94) 1.81 (1.84) 1.71 (1.72) 1.63 (1.64) 1.85 (1.87) 1.61 (1.66) 1.97 (1.97) 1.31 (1.34) 1.35 (1.35)

20 15 17 14 17 17 15 17 13 16

2220 1249 1604 1088 1604 1604 1249 1604 938 1421

1.80 1.91 (1.94) 1.82 (1.84) 1.72 (1.72) 1.63 (1.64) 1.85 (1.87) 1.61 (1.66) 1.97 (1.97) 1.31 (1.34) 1.35 (1.35)

21 15 17 16 17 14 17 17 15 15

2448 1249 1604 1421 1604 1088 1604 1604 1249 1249

Note: lit K0 are reduced mobilities from literature. a Dimer ions.

A.B. Kanu et al. / Analytica Chimica Acta 553 (2005) 148–159

154

Table 3 Reduced mobilities (K0 ), resolving power (Rp ) and theoretical plates (N) obtained for ions from studies using filters Compound

Day 1

Day 2

Day 3

K0 (lit K0 )

Rp

N

K0 (lit K0 )

Rp

N

K0 (lit K0 )

Rp

N

DMMP PMP DEPA 2-CEES

1.74 1.84 1.61 1.80

15 15 15 7

1249 1249 1249 272

1.74 1.85 1.61 1.80

14 18 17 7

1088 1798 1604 272

1.74 1.84 1.61 1.80

14 17 17 7

1088 1604 1604 272

2-BAET

1.66 1.20

11 11

672 672

1.67 1.20

12 11

799 672

1.67 1.20

14 13

1088 938

1,4-DT TDS EMPA CHMPA

1.09 1.73 1.60 0.95

15 16 11 12

1249 1421 672 799

1.73 1.61

14 12

1088 799

1.73 1.61

14 10

1088 555

DIMP

1.51 1.03a

11 5

672 139

1.51 1.03

11 6

672 200

1.52 1.03

12 5

799 139

2,2 -TDE CPA CHA CHPA COA DPA TPA NBA DHA DA

1.80 1.91 (1.94) 1.81 (1.84) 1.71 (1.72) 1.62 (1.64) 1.85 (1.87) 1.61 (1.66) 1.97 (1.97) 1.31 (1.34) 1.35 (1.35)

18 12 12 13 12 12 13 13 11 13

1798 799 799 938 799 799 938 938 671 938

1.79 1.91 (1.94) 1.81 (1.84) 1.72 (1.72) 1.63 (1.64) 1.85 (1.87) 1.61 (1.66) 1.97 (1.97) 1.31 (1.34) 1.35 (1.35)

16 13 12 13 14 12 14 13 12 12

1421 938 799 938 1088 799 1088 938 799 799

1.79 1.91 (1.94) 1.82 (1.84) 1.72 (1.72) 1.62 (1.64) 1.85 (1.87) 1.61 (1.66) 1.97 (1.97) 1.31 (1.34) 1.35 (1.35)

14 13 14 13 15 12 13 14 12 13

1088 938 1088 938 1249 799 938 1088 799 938

Note: lit K0 are reduced mobilities from literature. a Dimer ions.

this was the first time their mobilities had been measured and reported. All five simulants studied were detected by TD-IMS. 2-BAET showed a monomer and an unknown peak at K0 = 1.67 cm2 V−1 s−1 . Responses for PMP and 2-CEES on filters were much lower than the other compounds and responses for PMP and 2-CEES from glass slides were not observed. The lack of response for PMP and 2-CEES on the glass slides may have been due to their high vapor pressure. They may have evaporated from the glass surface with the solvent. Out of 13 hydrolysis products studied (TDS, EMPA, DIMP, 2,2 -TDE, CPA, CHA, CHPA, COA, DPA, TPA, NBA, DHA and DA), TDS, EMPA, DIMP, 2,2 -TDE and the nine amines were detected by the TD-IMS with DIMP showing both a monomer and a dimer. Both compounds classified as degradation products, 1,4-DT and CHMPA studied were also detected by TD-IMS. Fig. 2 shows example spectra produced for the CWA simulants, degradation and hydrolysis products on glass slides. Responses were obtained in the positive ion mode with ammonium as the dopant gas. For those compounds whose mobility values had been reported previously, comparison to literature data was made. Some of these K0 values compared well with the literature values [18,27,30–34] and some did not. Amines gave values closer to literature values but DMMP, DEPA, 2-BAET, EMPA, TDS, 2,2 -TDE and DIMP gave values that differ from literature values by 6–8%. Unfortunately, references which contained K0 values that did not match the ones observed in this experiment did not provide sufficient information (drift length, temperature, volt-

age, pressure, etc.) to verify the calculations. Thus, it was not possible to determine source of the discrepancy from this source. However, in this study, the effect of water content in the scrubbed air and humidity changes in the atmosphere may have been the cause of the differences in some of the K0 values. When comparing K0 values reported for the same compounds in air, differences of K0 units of 2% is considered an acceptable variation [35]. The amines gave a K0 difference between those measured in this study and those reported in the literature of between 1 and 3%. A monomer and dimer is reported for DIMP (K0 = 1.45 ± 0.005 and 0.98 ± 0.03 cm2 V−1 s−1 ). Previous studies conducted in our laboratory [30], however, reported K0 values (0.96 cm2 V−1 s−1 ) consistent with the dimer of DIMP. 3.2. Repeatability measurements Repeatability studies were conducted by introducing the samples from both glass slides and filters. Studies on glass slides were used to represent non-adsorptive surfaces while the filters represented surfaces with high adsorptivity. Table 4 is a summary of mean drift times together with the coefficient of variation (CV) for three replicate measurements. The CV which is the same as the relative standard deviation is the relative error estimate divided by the estimate of the absolute value (drift times) of the measured quantity. If we consider the DMMP measurement on day 1 in Table 4, the average drift time reported is 5.37 ms ± 0.04%. Repeatability limits (RL) were estimated using all nine measurements obtained over the 3-day period

A.B. Kanu et al. / Analytica Chimica Acta 553 (2005) 148–159

155

results demonstrated that TD-IMS responses to CWA simulants, degradation and hydrolysis products were highly reproducible for both filters and slides. 3.3. Calibration and sensitivity studies Calibration studies were conducted for DMMP and 2-BAET using glass slides. Linear regression on the response of DMMP on the glass slides gave the following equation: Peak area (103 au) = 2.46 (103 au ng−1 ) [DMMP] (ng) − 2.93 (103 au) R2 = 0.9922 where au are arbitrary units. The limit of detection (LOD), based on three times the noise level, for this study was estimated to be 30 pg of sample material deposited on glass slides with a sensitivity of 3.1 × 102 A g−1 . The negative intercept indicated some sample loss in the transfer process from the thermal desorber to the IMS. In the case of 2-BAET linear regression on response for the monomer response gave: Peak area (103 au) = 2.50 (103 au ng−1 ) × [2 − BAET monomer] (ng) Fig. 2. Example ion mobility spectra for: (a) CWA simulants, hydrolysis and degradation products, and (b) CWA hydrolysis products detected on glass slides. The dotted line for DIMP was the response for the dimer ion whereas for 2-BAET this was an unknown ion that develops at higher concentrations of 2-BAET.

measurements were taken. The RL value was determined for us to verify at which point there is a 95% probability that two single test results are significantly different. It was determined from t × (2)0.5 × sr , where t is the student two tailed value at n-1 degrees of freedom and sr is the repeatability standard deviation. From Table 4, RL of 0.03 ms was obtained for DMMP. This value could be interpreted to mean at the 95% confidence level, any two drift times of DMMP that differ by not more than 0.03 ms is a good result. A similar conclusion can be drawn from all compounds listed in Table 4. Spectra obtained for the samples desorbed from the filter paper were similar to those obtained with glass slides. Due to a higher surface area and a more polar surface, greater adsorption of analytes on surfaces of filters was expected relative to that on the glass slides. Results demonstrating the effect of the amount of analyte introduced on a surface will be discussed under resolving power measurements. However, the filters were more suitable for studies with faster diffusing analytes and analytes like PMP and 2-CEES that were not observed on glass slides were seen on filters. Results obtained for this study are displayed in Table 5. From Table 5, DEPA drift time was at 5.79 ms ± 0.13% on day 3. The RL at the 95% confidence level demonstrated that any two DEPA results differing by not more than 0.05 ms was a good result. Results obtained for both glass slides and filters were identical with respect to drift times, %CV and RL. These

− 8.53 (103 au) R2 = 0.9951 The linear regression on the unknown ion response gave the following equation: Peak area (103 au) = 1.45 (103 au ng−1 ) × [2 − BAET unknown] (ng) − 2.38 (103 au) R2 = 0.9999 LOD was estimated at 15 and 35 pg of sample material deposited on glass slides, with sensitivity of 3.2 × 102 and 1.8 × 102 A g−1 for the monomer and the unknown ion responses, respectively. As before, the negative intercepts indicate that the transfer between the thermal desorber and the IMS was not 100%. These results demonstrated the possibility of quantitative studies on the TD-IMS system. On the basis of comparison of sensitivities for both filters and slides, the two studies showed comparable sensitivities with higher boiling point compounds for both glass slides and filters. However, compounds with lower boiling point and hence high volatility showed increased sensitivity on filters compared to glass slides. Studies with filters were also associated with band broadening phenomenon and higher background noise. As these filters age their background noise increased, presumably due to breakdown of the fibers and adsorption of background compounds.

A.B. Kanu et al. / Analytica Chimica Acta 553 (2005) 148–159

156

Table 4 Average experimental drift times obtained for ions from three replicate measurements for studies using glass slides Compound

Day 1

Day 2

Day 3

Mean td (ms)

%R.S.D.

Mean td (ms)

%R.S.D.

Mean td (ms)

%R.S.D.

RL

DMMP PMP DEPA 2-CEES

5.37 nd 5.79 nd

0.04

5.35

0.01

5.35

0.12

0.03

0.02

5.80

0.03

5.81

0.02

0.02

2-BAET

5.60 7.76

0.12 0.01

5.60 7.74

0.12 0.01

5.58 7.73

0.12 0.01

0.03 0.04

1,4-DT TDS EMPA CHMPA

nd 5.38 5.80 nd

0.07 0.09

5.38 5.79

0.23 0.15

5.37 5.78

0.31 0.26

0.04 0.04

DIMP

6.16 9.06a

0.09 0.28

6.14 908

0.20 0.18

6.14 9.10

0.43 0.13

0.06 0.07

2,2 -TDE CPA CHA CHPA COA DPA TPA NBA DHA DA

5.19 4.90 5.15 5.45 5.74 5.07 5.80 4.74 7.14 6.91

0.13 0.10 0.11 0.10 0.20 0.14 0.18 0.14 0.03 1.73 × 10−6

5.20 4.89 5.15 5.45 5.74 5.05 5.79 4.73 7.12 6.89

0.11 0.01 0.01 0.11 0.10 0.13 0.02 0.13 0.10 0.01

5.19 4.88 5.14 5.44 5.73 5.04 5.78 4.72 7.10 6.88

0.15 0.12 0.01 0.20 0.01 0.11 0.12 0.14 0.10 0.01

0.03 0.03 0.02 0.03 0.05 0.03 0.06 0.03 0.03 0.03

Note: td is drift time (ms), R.S.D. is relative standard deviation (%) also known as coefficient of variation (CV), RL is repeatability limit for nine measurements over 3 days and nd is not discernable. a Dimer ions. Table 5 Average experimental drift times obtained for ions from three replicate measurements for studies using filters Compound

Day 1

Day 2

Day 3

Mean td (ms)

%R.S.D.

Mean td (ms)

%R.S.D.

Mean td (ms)

%R.S.D.

RL

DMMP PMP DEPA 2-CEES

5.36 5.06 5.79 5.18

0.09 0.14 0.09 0.04

5.37 5.05 5.79 5.18

0.02 0.41 0.46 0.06

5.35 5.06 5.79 5.18

0.09 0.30 0.13 0.10

0.03 0.05 0.05 0.01

2-BAET

5.60 7.77

0.06 0.06

5.60 7.77

0.01 0.07

5.59 7.74

0.02 0.02

0.02 0.04

1,4-DT TDS EMPA CHMPA

8.55 5.39 5.83 9.77

0.05 0.05 0.11 0.05

5.38 5.80

0.17 0.11

5.39 5.78

0.03 0.12

0.10 0.07

DIMP

6.16 9.08a

0.11 0.20

6.16 9.08

0.16 0.16

6.13 9.07

0.20 0.18

0.06 0.05

2,2 -TDE CPA CHA CHPA COA DPA TPA NBA DHA DA

5.19 4.89 5.15 5.45 5.74 5.05 5.80 4.73 7.14 6.92

0.01 0.21 0.04 0.13 0.01 0.12 0.09 0.01 0.01 0.01

5.22 4.89 5.14 5.44 5.73 5.05 5.80 4.73 7.12 6.90

0.09 0.02 0.13 1.6 × 10−6 0.22 0.01 0.01 0.10 0.02 0.09

5.21 4.87 5.14 5.42 5.72 5.04 5.78 4.72 7.11 6.89

0.27 0.24 1.6 × 10−6 0.12 0.25 0.23 0.12 0.14 0.09 0.07

0.05 0.03 0.02 0.05 0.04 0.03 0.03 0.02 0.04 0.04

Note: td is drift time (ms), R.S.D. is relative standard deviation (%) also known as coefficient of variation (CV), RL is repeatability limit for nine measurements over 3 days and nd is not discernable. a Dimer ions.

A.B. Kanu et al. / Analytica Chimica Acta 553 (2005) 148–159

157

3.4. IMS resolving power Theoretical resolving power for the TD-IMS was estimated to be 44 ± 0.05, using operating parameters from the TD-IMS. Eqs. (3) and (4) describe the separation efficiency in terms of theoretical plates and resolving power similar to equations predominantly used in chromatography [36]. Rp and N values for the compounds detected are shown in Tables 2 and 3 where it can be seen that Rp and hence N values for studies with glass slides were higher than those from filters. Studies with glass slides shown in Table 2, gave Rp values between 11 and 26, the maximum obtained for 2,2 -TDE on day 1 giving N of 3752. In the case of filters, Table 3, Rp values between 7 and 18 were obtained. The minimum of 7 occurred with 2-CEES with N of 272 and the maximum of 18 occurred with PMP and 2,2 -TDE giving N of 1798. These Rp range demonstrated band broadened peaks for filters. The cause of this band broadening may have been due to introduction of water into the drift region of the spectrometer. Filters, with a higher surface area are capable of sequestering more water than glass slides. Because the TD-IMS instrument used in these studies did not have a gas flowing counter to the ion flow, neutral water molecules could be passed directly into the drift region of the spectrometer. Ion molecule reactions with water in the drift region of the spectrometer could have resulted in broader ion peaks and thus a reduced resolving power. However, an increase in ion–molecule reactions between the analyte ion and water in the drift region of the spectrometer would have also been expected to cause a decrease in the ion’s mobility, which was not observed. Thus, the cause of this band broadening remains unclear. On the whole, the experimental resolving power of the instrument differed from the theoretical value by about 41%. 3.5. IMS resolution In high-resolution IMS systems, resolving power of 172 and 150 have been reported [37,38]. For such systems td of 0.03 ms is enough to obtain resolution of two compounds. In the case of conventional IMS systems with resolving powers of approximately 30, resolution of two peaks needs much higher td values. For the TD-IMS system at least 0.6 ms was needed to obtain baseline resolution of two peaks. To demonstrate the capability of the TD-IMS to separate CWA simulants three CWA simulants and four of the CWA hydrolysis products were introduced as mixtures. Fig. 3a shows a TD-IMS separation of DMMP, DEPA and 2-BAET desorbed from filters. Drift times of 5.39 ± 0.01, 5.85 ± 0.01, and 7.77 ± 0.02 ms (K0 = 1.67 ± 0.005, 1.54 ± 0.007, and 1.16 ± 0.01 m2 V−1 s−1 ) corresponds to the mobilities of DMMP, DEPA and the monomer of 2-BAET measured individually, and presented in Fig. 2. Fig. 3b shows the separation of DMMP and 2-BAET desorbed from glass slides. Drift times of 5.38 ± 0.01, and 7.78 ± 0.02 ms (K0 = 1.66 ± 0.006, and 1.15 ± 0.01 cm2 V−1 s−1 ) corresponds to the mobilities of DMMP, and the monomer of 2-BAET measured individually, and presented in Fig. 2. Fig. 3c shows the separation of DEPA and 2-BAET desorbed from

Fig. 3. Ion mobility spectra showing separation of CWA simulants mixtures for: (a) DMMP, DEPA and 2-BAET on traps; (b) DMMP and 2-BAET on slides; (c) DEPA and 2-BAET on slides. These results demonstrate that TD-IMS technology can be used for separating CWA simulants in real world situations. Three plasmagrams were displayed in each spectrum and a plasmagram is the average of four scans.

glass slides. Drift times of 5.77 ± 0.01, and 7.83 ± 0.01 ms (K0 = 1.55 ± 0.01, and 1.15 ± 0.01 cm2 V−1 s−1 ) corresponds to the mobilities of DEPA, and the monomer of 2-BAET measured individually, and presented in Fig. 2. In all cases the response for the unknown ion of 2-BAET was masked. A possible reason could be the abstraction of a proton from this ion converting it to a neutral and thus undetectable molecule.

158

A.B. Kanu et al. / Analytica Chimica Acta 553 (2005) 148–159

Fig. 4. Ion mobility spectra of CWA hydrolysis products mixtures. Note that CPA and CHA could not be resolved from each other. This result demonstrates that TD-IMS technology can be used for separating CWA hydrolysis products in real world situations. Each plasmagram displayed is the average of four scans.

Fig. 4 shows a TD-IMS response of CPA, CHA, COA and DHA desorbed from filters. Drift times of 4.91 ± 0.01, 5.10 ± 0.01, 5.72 ± 0.02, and 7.14 ± 0.01 ms (K0 = 1.83 ± 0.004, 1.76 ± 0.006, 1.57 ± 0.01, and 1.40 ± 0.01 cm2 V−1 s−1 ) corresponds to the mobilities of CPA, CHA, COA and DHA measured individually, and presented in Fig. 2. However, baseline resolution was not achieved between CPA and CHA with td value of 0.19 ms. Figs. 3 and 4 demonstrate the separation capability of TD-IMS enabling CWA to be separated. The advantages of developing a method for detecting CWA simulants, degradation and hydrolysis products with the TD-IMS technique is that it will provide a convenient means of detecting as well as separating these compounds with much more sensitivity and selectivity than other conventional methods in real world applications. 3.6. Detecting and separating CWA deposits on surfaces by TD-IMS Responses for the test compounds sampled from the surfaces of a briefcase, wood and stainless steel by wiping with the sampling filters are shown in Fig. 5a–c. Drift time of DMMP (5.38 ± 0.01), EMPA (5.75 ± 0.03), COA (5.79 ± 0.01), and DHA (7.13 ± 0.02) corresponds to the mobilities of the individual compounds desorbed from filters and glass slides and presented in Tables 4 and 5. Deposits of CWA wiped off from concrete produced no discernible responses, presumably because the CWA material irreversibly absorbed into concrete. Because of the porous nature of concrete detectable quantities of the material may not have remained on the surface of concrete when the deposits were allowed to dry. Wood which is also porous showed a similar effect although not with all compounds. This explains why a reduced response was obtained with wood, Fig. 5b. A more sensitive response was obtained from stainless steel, due to the fact that its surface is not rough and the filters were able to more efficiently wipe up the sample from the surface. The results demonstrate the technique is capable of picking individual components of CWA deposits from surfaces by TD-IMS. Of more

Fig. 5. Ion mobility spectra of DMMP (117 ng) and COA (92 ng) deposits wiped off using filters from (a) briefcase wipes; (b) wood wipes; and (c) stainless steel metal wipes. Similar spectrums were obtained for individual components of DMMP, EMPA, COA and DHA.

importance mixtures of CWA can be wiped from surfaces desorbed, detected, analyzed and separated. 4. Conclusion A TD-IMS instrument, which is commonly used for the detection of explosives and drugs, has the capability to rapidly detect and separate CWAs and/or their degradation products from environmental surfaces with good sensitivity and selectivity. All 20 compounds studied responded at the 1 ng level or below in the positive ion mode. Drift times and hence reduced mobility values were reproducible within 1–3%. Many of these mobility values are reported for the first time. In summary, this

A.B. Kanu et al. / Analytica Chimica Acta 553 (2005) 148–159

paper demonstrated that with little or no modification to an existing commercial instrument, TD-IMS can be used for the detection of CWAs and their degradation products on a variety of surfaces. Acknowledgements The authors would like to acknowledge GE Ion Track for partial support of this project. We also thank Dr. Wes E. Steiner, Dr. Ching Wu and Mr. Robert Baker for their support in this study. References [1] Convention on the Prohibition of the Development, Stockpiling and use of Chemical Weapons and on their Destruction. United States Control and Disarmament Agency, Washington, DC, 1993. [2] A.S.V. Burgen, S. Hobbiger, Br. J. Pharmacol. Chemother. 6 (1951) 593. [3] J.A.F. Compton, Military Chemical and Biological Agents: Chemical and Toxicological Properties, Telford press, Caldwell, NJ, 1987. [4] M. Fox, D. Scott, Mutat. Res. 75 (1980) 131. [5] R.E. Gosselin, R.P. Smith, H.C. Hodge, Clinical Toxicology of Commercial Products, fifth ed., Williams & Wilkins, Baltimore, MD, 1984. [6] G.J. Hathaway, N.H. Proctor, J.P. Hughes, M.L. Fischman, Proctor and Hughes Chemical Hazards of the Workplace, third ed., Van Nostrand Reinhold, New York, 1991. [7] R.L. Cheicante, J.R. Stuff, H.D. Durst, J. Cap. Elec. 4 (1995) 157. [8] A.F. Kingery, H.E. Allen, Toxicol. Environ. Chem. 47 (1995) 155. [9] G.W. Wagner, Y.C. Yang, Ind. Eng. Chem. Res. 41 (8) (2002) 1925. [10] R.I. Ellin, W.A. Groff, A.J. Kaminskis, Environ. Sci. Health, Part B B16 (6) (1981) 713. [11] Y.C. Yang, L.L. Szafraniec, W.T. Beaudry, C.A. Banton, J. Org. Chem. 58 (1993) 6964. [12] W.M. Meylan, P.H. Howard, J. Pharm. Sci. 84 (1) (1995) 83. [13] S. Franke, Manual of Military Chemistry, vol. 1, Chemistry of Chemical warfare Agents, Deutscher Militˆırverlag: Berlin (East), 1967. Translated from German by US Deptartment of Commerce, National Bureau of Standards, Institute for Applied Technology, NIST no. AD-849866.

159

[14] M.C. Roach, L.W. Unger, R.N. Zare, L.M. Reimer, D.L. Pumpliano, J.W. Frost, Anal. Chem. 59 (1987) 1056. [15] J.A. Tornes, B.A. Johnson, J. Chromatogr. 467 (1989) 129. [16] A.F. Nassar, S.V. Lucas, Anal. Chem. 71 (1999) 1285. [17] R.A. Black, R.W. Read, J. Chromatogr. A 794 (1998) 233. [18] G.R. Asbury, C. Wu, W.F. Siems, H.H. Hill Jr., Anal. Chim. Acta 404 (2000) 273. [19] G.A. Eiceman, Guest editorial report, Field Anal. Chem. Technol. 4 (5) (2000) 217. [20] M. Utrianen, E. K¨arp¨anoja, H. Paakkanen, Sensor Actuat. B 93 (2003) 17. [21] C.J. Proctor, J.F.J. Todd, Anal. Chem. 56 (1984) 1794. [22] A.H. Lawrence, P. Neudorfl, Anal. Chem. 60 (1988) 104. [23] G.E. Spangler, D.N. Campbell, J.P. Carrico, Acetone reactant ions for IMS, in: Proceedings of the PittCon Conference on Analytical Chemistry and Applied Spectroscopy, Atlantic City, 1991, pp. 13–15. [24] E. Geraghty, C. Wu, W. McGann, Int. J. IMS 5 (3) (2002) 41. [25] W. McGann, Proceedings of SPIE-The Int. Soc. for Optical Eng., 2937 (1997) 78, Chem. and Bio. Based Tech. for contraband detection. [26] TD-IMS contraband detection and identification system user’s manual, GE Ion Track Instruments, Document no. MA001043, Wilmington, MA, USA, 1991. [27] C. Shumate, R.H. St. Louis, H.H. Hill Jr., J. Chromatogr. A 373 (1986) 141. [28] W.F. Siems, C. Wu, E.E. Tarver, H.H. Hill Jr., P.R. Larsen, D.G. McMinn, Anal. Chem. 66 (1994) 4195. [29] L.M. Matz, P.S. Tornatore, H.H. Hill Jr., Talanta 54 (2001) 171. [30] W.E. Steiner, B.H. Clowers, P.E. Haigh, H.H. Hill Jr., Anal. Chem. 75 (22) (2003) 6068. [31] Z. Karpas, Anal. Chem. 61 (7) (1989) 684. [32] F.W. Karasek, H.H. Hill, S.H. Kim, J. Chromatogr., Sci. 117 (1976) 327. [33] C.S. Hoaglund, S.J. Valentine, D.E. Clemmer, Anal. Chem. 69 (1997) 4156. [34] R.W. Purves, R. Guevremont, S. Day, C.W. Pipich, M.S. Matyjaszcayk, Rev. Sci. Inst. 69 (1998) 4094. [35] L.M. Matz, H.H. Hill Jr., Anal. Chem. 73 (8) (2001) 1664. [36] G.R. Asbury, H.H. Hill Jr., J. Microcolumn Sep. 12 (2000) 172. [37] P. Dugourd, R.R. Hugdins, D.E. Clemmer, M.F. Jarrold, Rev. Sci. Instrum. 68 (2) (1997) 1122. [38] C. Wu, W.F. Siems, J. Klasmeier, H.H. Hill, Anal. Chem. 72 (2000) 391.