Article pubs.acs.org/ac
Sensitive Monitoring of Volatile Chemical Warfare Agents in Air by Atmospheric Pressure Chemical Ionization Mass Spectrometry with Counter-Flow Introduction Yasuo Seto,*,† Mieko Kanamori-Kataoka,† Koichiro Tsuge,† Isaac Ohsawa,† Kazumitsu Iura,†,⊥ Teruo Itoi,†,⊥ Hiroyuki Sekiguchi,†,⊥ Koji Matsushita,†,⊥ Shigeharu Yamashiro,†,⊥ Yasuhiro Sano,†,⊥ Hiroshi Sekiguchi,†,⊥ Hisashi Maruko,†,⊥ Yasuo Takayama,†,⊥ Ryoji Sekioka,†,⊥ Akihiko Okumura,*,‡ Yasuaki Takada,‡ Hisashi Nagano,‡ Izumi Waki,‡ Naoya Ezawa,§ Hiroyuki Tanimoto,§ Shigeru Honjo,§ Masumi Fukano,§ and Hidehiro Okada§ †
National Research Institute of Police Science, Kashiwa, Chiba 277-0882, Japan Central Research Laboratory, Hitachi, Ltd., Kokubunji, Tokyo 185-8601, Japan § Hitachi, Ltd., Defense Systems Company, Chiyoda, Tokyo 101-8608, Japan ‡
ABSTRACT: A new method for sensitively and selectively detecting chemical warfare agents (CWAs) in air was developed using counter-flow introduction atmospheric pressure chemical ionization mass spectrometry (MS). Four volatile and highly toxic CWAs were examined, including the nerve gases sarin and tabun, and the blister agents mustard gas (HD) and Lewisite 1 (L1). Soft ionization was performed using corona discharge to form reactant ions, and the ions were sent in the direction opposite to the airflow by an electric field to eliminate the interfering neutral molecules such as ozone and nitrogen oxide. This resulted in efficient ionization of the target CWAs, especially in the negative ionization mode. Quadrupole MS (QMS) and ion trap tandem MS (ITMS) instruments were developed and investigated, which were movable on the building floor. For sarin, tabun, and HD, the protonated molecular ions and their fragment ions were observed in the positive ion mode. For L1, the chloride adduct ions of L1 hydrolysis products were observed in negative ion mode. The limit of detection (LOD) values in real-time or for a 1 s measurement monitoring the characteristic ions were between 1 and 8 μg/m3 in QMS instrument. Collision-induced fragmentation patterns for the CWAs were observed in an ITMS instrument, and optimized combinations of the parent and daughter ion pairs were selected to achieve real-time detection with LOD values of around 1 μg/m3. This is a first demonstration of sensitive and specific real-time detection of both positively and negatively ionizable CWAs by MS instruments used for field monitoring.
C
hemical warfare agents (CWAs)1 were used in World War I, World War II, and the Cold War, and production and stockpiling continued after these wars finished.2 In the 1980s, Iraq used sarin (GB) and mustard gas (HD) in the Iran−Iraq War.3 In 1992, a treaty prohibiting the development, production, stockpiling, and use of chemical weapons and mandating their destruction was ratified, and in 1997, it came into force.4 The Japanese cult Aum Shinrikyo used GB in Matsumoto in 1994 and in the Tokyo subway in 1995, and many defenseless people were poisoned or killed,5 which brought the threat of chemical terrorism to public attention worldwide. For public safety and security, stronger national crisis management systems are required for civil defense.6 In responding to chemical warfare terrorism and in disposal of chemical weapons, CWAs need to be detected in a variety of areas, including in public places and at territorial borders, airports, event venues, executive facilities, and demilitarization facilities.7,8 Monitoring is performed for protection against © 2013 American Chemical Society
terrorism and to ensure the health and safety of workers involved in weapons disposal. To manage the consequences of CWA use in terrorism, on-site detection of CWAs is performed by first-responders to protect exposed individuals. On-site samples are transported to laboratories for analysis for criminal investigation and to provide information for treatment of exposed individuals. Laboratory analysis is also performed to provide evidence for court to prevent future crimes. Current laboratory analysis for identifying toxic substances from on-site and exposure victim specimens9,10 consists of a preliminary rapid screening test, sample pretreatment, and instrumental analysis by hyphenated mass spectrometry (MS). Rapid on-site detection is important to reduce the impact of any CWA Received: October 11, 2012 Accepted: January 23, 2013 Published: January 23, 2013 2659
dx.doi.org/10.1021/ac303373u | Anal. Chem. 2013, 85, 2659−2666
Analytical Chemistry
Article
situation, because transfer of samples to a laboratory for analysis takes too long. We previously evaluated the performance of commercially available portable air-monitoring CWA detectors,11−14 including Draeger safety gas detection tubes,15 ion mobility spectrometry (IMS) instruments,16−19 and an arrayed surface acoustic wave detection instrument (ChemSentry).20 However, none of these detection apparatuses met all the requirements of field utilization by first responders.14 Low sensitivity, low specificity (frequent false alarms), and long response and recovery times because of strong adsorption of CWAs on the devices were particularly problematic. Commercially available portable equipment based on gas chromatography (GC)21 and GC/mass spectrometry (GC/MS)22 can provide highly sensitive and specific detection of CWAs in air. However, real-time and rapid detection cannot be achieved by GC-based technologies. Miniature MS instruments have been developed, one of which enables continuous detection of CWA simulants in air by glow discharge electron ionization cylindrical ion trap MS with a dual-tube thermal desorption system.23 For sensitive and specific real-time detection, various technologies based on direct air-sampling MS have been developed and used for the detection of CWAs. These technologies include membraneinlet electron-ionization MS,24 atmospheric pressure chemical ionization MS (APCI-MS),25−27 proton transfer reaction MS (PTR-MS),28,29 selected ion flow tube MS (SIFT-MS),30,31 and APCI-ion mobility spectrometry (IMS)-MS.32 Some of these technologies provided real-time and sensitive detection. However, the range of compounds that can be detected is limited, and the instruments are very large and/or complicated for on-site use. Recently, a research group at Hitachi Ltd. developed novel on-site detection technology using counterflow introduction APCI-MS (CFI-APCI-MS).33 In the CFI configuration, sample air is introduced into the corona discharge region, and either of the positive or negative reactant ions generated are extracted out of the region against the incoming airflow by the strong electric field. Interfering neutral byproducts generated in corona discharge, such as nitrogen oxides and ozone, are effectively removed out of the system by the airflow, and target analytes in the air are efficiently ionized. The instruments used are suitable for field usage because of their compactness and sturdiness. This technology has been applied for the detection of dioxin precursors in flue gas,34,35 polychlorinated biphenyls in air,36 explosives in air,33,37−39 and illegal drugs in tablets.40 In this paper, we evaluated the detection ability of CFIAPCI-MS for representative CWAs, including the nerve gases GB and tabun (GA) and the blister agents mustard gas (HD) and Lewisite (L1). Two instruments based on quadrupole mass spectrometry (QMS) and ion trap tandem mass spectrometry (ITMS), both of which were movable on the building floor, were examined. Ultrasensitive and specific real-time detection of not only positively ionizable GB, GA, and HD but also negatively ionizable L1 in air has been achieved manually by switching the polarity of applied voltages according to the target.
Figure 1. Atmospheric pressure chemical ionization source with counter-flow introduction (a) and experimental setup for sample vapor generation and introduction into mass spectrometry instrument (b).
metal sewing needle (ø 0.84 mm, Clover Mfg. Co. Ltd., Osaka, Japan) and a stainless steel disk electrode (aperture ø 3 mm) separated by 4 mm from each other. The disk electrode was placed to face down on the inlet electrode in the vacuum system for mass analysis. Sample air (1.2 L/min) was drawn into the discharge region through the aperture and then pumped out from the root of the needle (0.7 L/min). For positive ion analysis, the voltages on the disk and the inlet electrode were fixed at 1 kV and 30−60 V, respectively. The applied voltage (2−5 kV) on the needle was automatically controlled to yield a certain discharge current (1−20 μA). The positive reactant ions were generated through the aperture, forced toward the inlet electrode against the counter air flow by the strong electric field, and allowed to react with the analyte molecules. Then, the positive ions generated from the analytes were drawn into the vacuum system along with the sample air because of the pressure difference. Negative ion analysis was performed by reversing the polarity of the voltages used for positive ion analysis. In positive ionization, analyte molecules that have higher proton affinities than the hydronium ion or its water cluster are efficiently protonated to form MH+, where M denotes an analyte molecule, by the proton transfer reaction .41 M + H3O+ → MH+ + H 2O
In negative ionization, analyte molecules that have higher electron affinities than the oxygen anion or its water cluster are possibly charged to form M− by the electron transfer reaction. M + O2− → M− + O2
The deprotonated molecule, [M − H]−, which is possibly generated through proton abstraction by the oxygen anion,42 is also observed.34
■
EQUIPMENT AND EXPERIMENTAL CONDITIONS CFI-APCI. The CFI-APCI source is shown in Figure 1. Corona discharge was used to produce reactant ions, hydronium ion (H3O+) and oxygen anion (O2−) for positive and negative chemical ionization of the analyte molecules, respectively. The corona discharge was generated between a
M + O2− → [M − H]− + HO2
The term CFI arises from the fact that the sample airflow in the discharge region is counter to the movement of ions toward the MS instrument.33 In negative ionization, the CFI 2660
dx.doi.org/10.1021/ac303373u | Anal. Chem. 2013, 85, 2659−2666
Analytical Chemistry
Article
configuration provides much higher O2− yield than in a conventional configuration where the airflow is reversed. This is apparent when comparing the mass spectra obtained for background air with CFI and conventional configurations (Figure 2). In the mass spectrum with the conventional
field gradient was formed between the aperture plate and the skimmer electrode to transmit the ions efficiently into the second chamber and to remove water molecules from water− ion clusters by accelerating the ions to collide with residual neutral molecules.33 The bare ions exiting from the skimmer were focused through a static lens system onto the entrance of a conventional quadrupole mass filter. Mass analysis was conducted by scanning the m/z value of the ion to be transmitted onto the ion detector. The CFI-APCI-ITMS instrument (DS-1000, Hitachi) was 270 kg in weight and 2 kW in power consumption. It could also be easily moved on casters. It used a stainless steel capillary (I.D. 0.3 mm, length 21 mm) as the inlet electrode. The vacuum system was separated into three chambers by two skimmer electrodes and evacuated by differential pumping. The first and the second skimmers had aperture diameters of 0.9 and 0.6 mm, respectively. The first chamber was evacuated at 342 L/min by a rotary pump (E2M18, Edwards, Crawley, United Kingdom). The second and the third chambers were evacuated at 205 and 280 L/s, respectively, by a split-flow turbo molecular pump (TMH261-250-101, Pfeiffer Vacuum GmbH, Asslar, Germany). The temperatures of the ion source, the capillary, and the second skimmer were controlled at 120 °C. Voltages of 35, 30, and 5 V were applied on the capillary, the first, and the second skimmers, respectively, to form a field gradient to efficiently transmit the ions through the apertures and to promote collisional declustering of the analyte−water cluster ions. These voltages were for positive ion analysis, and the polarity was reversed for negative ion analysis. The instrument used a conventional three-dimensional quadrupole ion trap. Ions exiting the second skimmer were transferred into the ion trap through a static lens focusing system. Helium gas (>99.99% purity, Tomoe Shokai Co. Ltd., Tokyo, Japan) was continuously introduced into the trap to decrease the kinetic energy of incoming ions through collisions and improve the ion trapping efficiency. Ions were loaded into the trap for a certain period by switching the voltage on the gate electrode placed between the focusing system and the trap. After stopping ion loading, mass analysis of the trapped ions was carried out by ejecting and detecting the ions in order of increasing m/z. This was realized by increasing the amplitude of the trapping radio frequency (RF) voltage corresponding to the cutoff m/z or the minimum m/z of the trapped ions. Tandem mass spectrometry was carried out by isolating the ions with m/z of interest and then fragmenting the isolated ions before mass analysis. The ions were isolated by increasing the cutoff m/z almost to the m/z of interest and applying the supplemental RF voltage to mass-selectively energize and eject the ions with m/z values higher than those of interest. Fragmentation of the isolated ions by collision-induced dissociation was achieved by energizing the ions with the supplemental RF voltage to collide with the helium gas. Chemicals. GB (o-isopropyl methylphosphonofluoridate), GA (o-ethyl N,N-dimethylphosphonocyanidate, tabun), HD (bis(2-chloroethyl) sulfide, mustard gas), and L1 (2-chlorovinyldichloroarsine, Lewisite 1) were obtained from the TNO Prins Maurits Laboratory (Rijswijk, Netherlands) or synthesized in the laboratory of National Research Institute of Police Science (Kashiwa, Japan). All of these compounds were >98% pure. The other reagents were of analytical grade. Sample Vapor Generation. Standard solutions of the CWAs were prepared in n-hexane (Wako Pure Chemical Industries, Osaka, Japan) at desired concentrations. Figure 1b
Figure 2. Negative atmospheric pressure chemical ionization mass spectra of laboratory air obtained with (a) counter-flow introduction and (b) reversed air-flow.
configuration (Figure 2b), the base peak at m/z 62 was dominant. This peak could be assigned to NO3− by tandem mass spectrometry (MS/MS), which showed the m/z 62→ 46 transition for loss of O. Because of the higher electron affinity of NO3 (3.9 eV43) compared with O2 (0.5 eV43) and its lower proton affinity (13.8 eV43) compared with O2− (15.0 eV43), NO3− is stable and chemical ionization of the analytes is suppressed. Formation of NO3− and the decrease in O2− can be explained by the reaction of O2− with nitrogen monoxide (NO) generated in the corona discharge.33 Another process for formation of NO3− involving O2−, ozone, and NO2 generated in the corona discharge has been reported.44 In any case, in the CFI configuration, neutrals generated in the discharge region are effectively removed from the system. This reduces formation of NO3− and increases the yield of O2− as shown in Figure 2a, which leads to high ionization efficiency for the analyte molecules. A large sensitivity enhancement with the CFI configuration compared with the conventional configuration was demonstrated for the proton abstraction reaction for a test compound.37 In contrast, in positive ionization, there was no obvious difference in the mass spectra, where H+(H2O)2 was dominant, for the background air obtained with CFI and conventional configurations. For the test compounds, the intensity of the protonated molecule was similar in each of the configurations. The ion source was heated to 100−200 °C to suppress adsorption of molecules in the sample air onto the instrument surfaces. Heating also removed water molecules from the water−ion clusters, which simplified the mass spectra. MS Instrument. A QMS instrument and an ITMS instrument were compared in the experiments. The CFIAPCI-QMS instrument (DS-100, Hitachi, Tokyo, Japan) was 140 kg in weight and 1.5 kW in power consumption. It could be easily moved on casters by one person and used for monitoring CWAs in a field monitoring post or on the building floor. The inlet electrode was a stainless steel plate (0.3 mm thick) with an aperture diameter of 0.08 mm. The vacuum system was separated into two chambers by a grounded skimmer electrode (aperture ø 0.3 mm) and evacuated by differential pumping. A 2661
dx.doi.org/10.1021/ac303373u | Anal. Chem. 2013, 85, 2659−2666
Analytical Chemistry
Article
Table 1. Detection of Nerve and Blister Chemical Warfare Agents in Air by Counter-Flow Introduction Atmospheric Pressure Chemical Ionization Quadruple-Mass Spectrometry Instrument in Real-Time or a 1 s Measurement Cycle ion monitored agenta GB GA HD L1
ion polarity + + + −
assignment
limit of detectionb (μg/m3)
MH MH+ (M − Cl)+ [(M − 2Cl + 2OH + Cl)]−
1.1 2.3 1.9 7.7
m/z
ions observed [m/z] (relative abundance) 99 (100), 141 (5) 135 (70), 163 (100) 123 (100), 159 (2), 175 (2) 187 (50), 205 (100), 223 (30)
141 163 123 205
+
a GB = sarin; GA = tabun; HD = mustard gas; L1 = Lewisite 1. bDefined as the concentration giving the ion signal of 3 × σBG, where σBG is the standard deviation of the background signal measured for the laboratory air.
vapors prepared in a 500 mL container have been successfully used to evaluate other detectors.12,20−26 However, in the present study, the observed signal intensity showed poor reproducibility at the low concentrations that were used for the evaluation. This occurred because of the limited volume of the container, which limited the amount of agent injected for analysis. Therefore, a 10 L glass container was tried with the PFA connecting tube. Unfortunately, the agents strongly adsorbed on the surfaces of the container and tubing, and no CWA specific signals were observed at subppm levels. Replacing the PFA tube with a heated stainless steel tube was not effective. However, the method was successful when the 10 L glass container was replaced with a 10 L stainless steel container coated with PTFE. For calibration, the air flow through the discharge region was set to 0.5 L/min and the signal intensity was measured at 1 s intervals while introducing the vapor. The maximum signal intensity was plotted against the vapor concentration to form a calibration plot. The vapor concentration was determined by dividing the amount of agent injected by the container volume. For the detection of GB, the ion of m/z 141 was monitored because the fragment ion of m/z 99 could be obtained from the other fluorophosphonic nerve gases such as soman. The calibration plot for m/z 141 showed good linearity. The limit of detection (LOD) is defined as the GB concentration giving the ion (m/z 141) signal of 3 × σBG, where σBG is the standard deviation of the background signal measured for the laboratory air. This value, which was calculated using the slope of the calibration curve and σBG, was 1.1 μg/m3. Vapors made from GA, HD, and L1 were examined using the above method. Table 1 lists the mass spectral patterns and the LODs for the CWAs tested. For GA, in addition to the fragment ion at m/z 135, the protonated molecule of agent was observed at m/z 163 as the base peak, and this was used for the detection of GA. For HD, the base ion peak at m/z 123 ([M − Cl]+) and weak peaks at m/z 159 (MH+) and 175 ([M + O + H]+) were observed, and the base peak was used for the detection of HD. The mass spectral pattern was different from that presented by the PTR-time-of-flight (TOF)-MS using the radioactive source ionization for producing the reactant H3O+ where not only MH+ but also [M − Cl + 2H]+ at m/z 125 were observed.46 In the negative ion mass spectra, GB, GA, and HD provided no particular ions, although APCI-IMS-MS using radioactive ionization observed O2− adduct ion of HD.47 For L1, peaks at m/z 187, 205, and 223 were observed in negative ion mode, and the peaks at m/z 205 and 223 could be assigned to the chloride adducts of the L1 hydrolysis products, as was observed in APCI-IMS-MS using radioactive ionization.47 Because L1 is known to easily degrade, even in ambient air, hydrolyses likely occurred in the instrument. The base peak at m/z 205 was used
shows the experimental setup. Each CWA was vaporized by injecting a drop of its standard solution into one of three different heated containers, each with a sealed syringe injection port and valves for the inlet and outlet gases. Laboratory air (23 °C, relative humidity 30−54%) was sucked into the container through the inlet. The first and second heated containers were a 500 mL glass gas-sampling cylinder (GL Sciences, Tokyo, Japan) and custom-made 10 L glass cylinder (Hitachi Kyowa Engineering Co., Hitachi-naka, Japan), that were heated with a dryer for 1 min (temperature not measured). The third heated container was a custom-made stainless steel 10 L cylinder (Hitachi Kyowa Engineering Co.) with its inner surface coated with PTFE (plytetrafluoroethylene). This container was heated to be 120 °C with an electric heating mantle. The CWA vapor was drawn into the MS instrument through a heated PFA (tetrafluoroethylene-perfluoroalkylvinylether copolymer) or stainless steel connecting tube (I.D. 0.64 cm, length 50 cm). CWA vapor with the designated moisture level was generated by the following procedure and introduced into the MS instrument at the flow rate of 1.2 L/min. The humidified air stream was generated by passing the pressurized clean and dried air (Tomoe Shokai Co. Ltd.) through a water bath incubated at 50 °C and mixed with an additional clean and dry air stream to make the 1.2 L/min stream with the designated moisture level using mass flow controllers (SFC280E, Hitachi Metals Ltd., Tokyo, Japan; KOFLOC 3660, Kojima Instruments Inc., Kyoto, Japan). The combined stream was finally mixed with the third dry air stream at the flow rate of 1 mL/ min using a syringe pump (MSPE-1, AS ONE Corporation, Osaka, Japan,) where a 2 μL drop of CWA solution was injected and vaporized. Safety Considerations. CWAs are highly toxic and were carefully handled by specially trained personnel. All experiments for CWAs were performed in a specific facility. Sample vapor was prepared in a fume hood with alkaline scrubber system to prevent personnel from being exposed to vapors and CWAs from being released outside the facility. The use and synthesis of CWAs for this research were approved by the Minister of Economy, Trade and Industry of Japan (http:// www.meti.go.jp/polycy/chemical_management/cwc/ 200kokunai/202horitu_gaiyo.htm).
■
RESULTS AND DISCUSSION CFI-APCI-QMS. GB vapor (0.1 mg/m3) was prepared in a heated 500 mL glass container and introduced into the instrument via PFA tubing. A mass spectrum for GB was measured by QMS. The base peak at m/z 99 was attributed to a fragment ion ([M − C3H5]+) and a weak peak (5% of the intensity of the base peak) at m/z 141 was attributed to MH+, as was observed in conventional flow APCI-quadrupole MS/ MS using corona discharge ionization.45 In earlier studies, 2662
dx.doi.org/10.1021/ac303373u | Anal. Chem. 2013, 85, 2659−2666
Analytical Chemistry
Article
Figure 3. Detection of chemical warfare agents by counter-flow introduction atmospheric pressure chemical ionization ion trap mass spectrometry instrument for (a) sarin, (b) tabun, and (c) mustard gas in positive ion mode and (d) Lewisite 1 in negative ion mode. The mass spectra and the corresponding product ion mass spectra are shown in the upper and lower figures. The mass spectrum for GB was measured in low mass (m/z 100) regions with ion loading times of 3 and 30 ms, respectively. Possible assignment and fragmentation pathways for the ions are shown in the inset.
GA in the air with humidity of 0.01−50 g/m3 were observed to be more than 50% compared to those in the room air. The mass spectrum of HD (Figure 3c) showed a base peak at m/z 159 (MH+) and peaks at m/z 123 ([M − Cl]+) and 175. The m/z 175 could be assigned to the protonated molecule of oxidized HD, probably mustard sulfoxide ((ClC2H4)2SO). A high concentration of strongly oxidative ozone can be generated by corona discharge in air,37 and oxidation may occur to some extent by reaction with the residual ozone, although the generated ozone is effectively removed from the system by the CFI configuration. The product ion mass spectrum for the MH+ ion (data not shown) showed a product ion peak at m/z 123 ([M − Cl]+) because of loss of HCl. The product ion mass spectrum for the [MO]H+ ion showed a base peak at m/z 77 possibly because of loss of C2H4Cl2 as illustrated in the figure. Other fragment ion peaks characteristic of HD could be attributed to [MO]H+ by elemental composition considerations. The signal intensities of the [MO]H+ ion in the air with humidity of 0.01−50 g/m3 were observed more than 50% compared to those in the room air, in contrast to that of the MH+ ion which intensity fell significantly above the humidity of 10 g/m3. Accordingly, the intensity of the MS/MS transition m/z 175→77 was monitored for HD. L1 (Figure 3d) formed many characteristic ion peaks in the negative ion mass spectrum. Among these, the ions at m/z 205 and 223 could be assigned to the chloride adduct ions of L1 hydrolysis product [(M − 2Cl + 2OH) + Cl]− and [(M − Cl + OH) + Cl]−, respectively, with the aid of MS/MS transitions. The ions at m/z 241 are probably chloride adduct ions of L1, as was also observed in literature,47 although there were no observable collision-induced product ions other than Cl−. The
for the detection of L1. The calibration plots for these CWAs showed good linearity (data not shown). The LODs for the CWAs were between 1 and 8 μg/m3. In positive ion mode, L1 provided no particular ions. CFI-APCI-ITMS. In QMS analysis, high detection sensitivity was obtained for the CWAs by monitoring their characteristic ions (Table 1). However, in on-site practical use, interference from other compounds should be considered because they may give false positives or increase the noise. In the presence of contaminants such as gasoline vapor, we found detection of the CWAs deteriorated (data not shown). Consequently, we evaluated MS/MS using an ITMS instrument. The mass spectra and product ion mass spectra for GB, GA, HD, and L1 are shown in Figure 3. For GB (Figure 3a), the mass spectrum was similar to that from QMS, with the base peak observed at m/z 99. However, compared with QMS, an additional weak peak was observed by ITMS at around m/z 140. This peak could be caused by a mass-shift of the MH+ ion from m/z 141 which was observed in QMS, as shifts toward lower m/z values are known to occur in ITMS with fragile ions.48 The MH+ ion of GB was much lower in intensity than the m/z 99 fragment ion probably because of fragmentation of the fragile MH+, as was observed by QMS. The product ion mass spectrum for MH+ gave a single product ion peak at m/z 99, and this transition was monitored for GB. For GA (Figure 3b), the mass spectrum gave a strong peak at m/z 163, which was assigned to the MH+ ion of GA, and another strong peak at m/z 135 that was attributed to a fragment ion. The MS/MS for the MH+ ion provided one product ion peak at m/z 135 possibly because of loss of CO, C2H4, or CH2N, and this transition was monitored for GA. The signal intensities of the MH+ ions from GB and 2663
dx.doi.org/10.1021/ac303373u | Anal. Chem. 2013, 85, 2659−2666
Analytical Chemistry
Article
time for GA (m/z 163→135) as its vapor was introduced and the concentration increased (Figure 4a). The corresponding calibration plot is shown in Figure 4b. For each concentration, the signal intensities were averaged over 2 min and plotted against the concentration. The background signal of laboratory air (humidity 10 g/m3) was repeatedly measured 250 times. Table 2 lists the LODs obtained by MS/MS for the CWAs examined. The LODs obtained by ITMS were much lower than those obtained by QMS for GA, HD, and L1 (see Tables 1 and 2) and about two times higher than that obtained by QMS for GB. Although the LOD for GB increased slightly in ITMS compared with QMS, detection with ITMS was more specific than with QMS because of suppression of interferences from the matrix. Because CWAs are acutely toxic and easily vaporize, positive detection of a CWA vapor needs to occur at a vapor level one hundredth of the median lethal dosage (human, LCt50, mg/ min/m3),1 and this needs to be performed rapidly (i.e., within 1 min of sample air intake), in consideration of the on-site situation where casualties are exposed to a high level of hazardous vapor in the “counterterrorism” aspect.14 These concentrations (referred to as 1/100 median lethal concentration (LC50) value) for GB, GA, HD, and L1 can be 1.5, 3, 15, and 15 mg/m3, respectively, which are calculated from the LCt50 values.1 In the “health and safety” aspect, the time weighted averages need to be considered for people who are exposed to a low level of hazardous vapor in a CWA dispersal area for long time, and these values are approximately 1/1 000 000 of the LC50.49 In this case, positive detection is allowed to occur within several minutes of sample air intake. The LODs obtained in real-time or a 1 s monitoring cycle by CFI-APCIITMS (Table 2) were much lower than the 1/100 LC50 values for all the agents examined. The LODs were similar to the 1/1 000 000 LC50 values for all agents except GB. These results indicate that the established detection system could fulfill requirements for CWA detection (1/100 LC50, 1 min) in counterterrorism and for health and safety (1/1 000 000 LC50, several minutes) such as chemical weapons demilitarization facilities. In these facilities, a longer detection time is acceptable. Averaging over a longer period will give lower LODs than for the 1 s detection. Assuming there is randomness in the signal, the LOD will be inversely proportional to the square root of the number of times of averaging. For example, by averaging one hundred times or about 2 min, the LOD will decrease to one tenth that at 1 s. However, the linearity of the calibration curve must be confirmed at concentrations lower than those tested here. Comparison of the Detection Performance with the Other Technologies. Among the existing field-deployable instruments for detecting CWAs in air,14 IMS equipment50,51
source of adducting chloride ion may be hydrogen chloride generated through the hydrolysis of L1. The ion at m/z 187 was also observed, which could be assigned to [(M − Cl + O]−. This ion is possibly a fragment of the ion at m/z 205 and 223. The ion at m/z 185 could be assigned to [M − 2Cl + 3O + H]−. This ion was monitored for the detection of L1 as it provided the lowest LOD by MS/MS, and this signal intensity in the air with the humidity of 0.01−30 g/m3 was observed more than 25% compared to those in the room air. The MS/ MS of the ion showed two intense product ions at m/z 123 and 149 because of loss of C2H3Cl and HCl, respectively. The transition to the former product ion was monitored for L1. In positive ion mode, L1 formed an ion peak that could be attributed to protonated molecule of L1 hydrolysis products. However, its intensity was much weaker than the intensities of the negative-ion peaks. For calibration, the airflow through the discharge region was set to 0.7 L/min. The ion loading time into the ion trap was set to 300 ms, and three successive measurements were averaged. These conditions resulted in a measurement cycle of 1.2 s. Figure 4 shows changes in the recorded signal intensity with
Figure 4. Monitoring of tabun (GA) by counter-flow introduction atmospheric pressure chemical ionization ion trap-mass spectrometry instrument. (a) Changes in the signal intensity for GA (m/z 163→ 135) while introducing GA vapor. (b) Calibration plot.
Table 2. Detection of Nerve and Blister Chemical Warfare Agents in Air by Counter-Flow Introduction Atmospheric Pressure Chemical Ionization Ion Trap-Tandem Mass Spectrometry Instrument in Real-Time or a 1 s Measurement Cycle transition agent GB GA HD L1
a
ion polarity + + + −
assignment
daughter ion (m/z)
limit of detectionb (μg/m3)
MH MH+ [M + O + H]+ [M − 2Cl + 3O + H]−
99 135 77 123
1.8 0.14 0.63 0.66
parent ion (m/z) 141 163 175 185
+
GB = sarin; GA = tabun; HD = mustard gas; L1 = Lewisite 1. bDefined as the concentration giving the ion signal of 3 × σBG, where σBG is the standard deviation of the background signal measured for the laboratory air. a
2664
dx.doi.org/10.1021/ac303373u | Anal. Chem. 2013, 85, 2659−2666
Analytical Chemistry
Article
provides real-time and sensitive detection at around sub mg/m3 levels for nerve gases and several mg/m3 levels for blister agents.16−19 Although IMS equipment fulfills the counterterrorism requirement for detection sensitivity (1/100 LC50) and provides real-time measurement, frequent false positives occur because of poor peak resolution. A GC/QMS instrument combined with microtrap concentration device (Hapsite) provides high sensitivity with LODs of 0.1−2.4 μg/m3 12 and excellent specificity. This is because the process involves GC separation followed by mass spectrometric discrimination. However, real-time monitoring or rapid detection (
