Article pubs.acs.org/ac
Direct Real-Time Detection of Vapors from Explosive Compounds Robert G. Ewing,* Brian H. Clowers, and David A. Atkinson Pacific Northwest National Laboratory, Richland, Washington 99352, United States ABSTRACT: The real-time detection of vapors from low volatility explosives including PETN, tetryl, RDX, and nitroglycerine along with various compositions containing these substances was demonstrated. This was accomplished with an atmospheric flow tube (AFT) using a nonradioactive ionization source coupled to a mass spectrometer. Direct vapor detection was accomplished in less than 5 s at ambient temperature without sample preconcentration. The several seconds of residence time of analytes in the AFT provided a significant opportunity for reactant ions to interact with analyte vapors to achieve ionization. This extended reaction time, combined with the selective ionization using the nitrate reactant ions (NO3− and NO3−·HNO3), enabled highly sensitive explosives detection from explosive vapors present in ambient laboratory air. Observed signals from diluted explosive vapors indicated detection limits below 10 ppqv using selected ion monitoring (SIM) of the explosive-nitrate adduct at m/z 349, 378, 284, and 289 for tetryl, PETN, RDX, and NG, respectively. Also provided is a demonstration of the vapor detection from 10 different energetic formulations sampled in ambient laboratory air, including double base propellants, plastic explosives, and commercial blasting explosives using SIM for the NG, PETN, and RDX product ions.
T
when using a nonvolatile nitrate dopant over a volatile chloride dopant.10 They reported detection limits of 5.3 μg L−1 of RDX solution added at a rate of 10 μL min−1 into a nitrogen gas stream of 573 mL min−1 and further mixed with nitrogen gas through the drift region at a rate of 800 mL min−1. This correlates to a detection limit of about 4 pptv, slightly below the 5 pptv equilibrium vapor pressure of RDX at 25 °C. In 2009, Martinez-Lozano et al. used SESI with mass spectrometry to demonstrate detection limits of 0.2 ppt v for PETN, approximately 50 times lower than the 11 pptv equilibrium vapor pressure of PETN at 25 °C.9 The authors noted that they observed the detection of vapors from residue deposited in a glass flask, but due to issues with particles being entrained in the gas stream, this method of vapor generation was abandoned. The primary method they employed to generate explosive vapors involved introducing a flow of an explosive solution into a second electrospray ionization source with a supporting gas and introducing these vapors into the SESI region. Additional improvements to the SESI method reported by Vidal-de-Miguel et al. used a lower sample flow and a differential mobility analyzer prior to detection with mass spectrometry to achieve 20 fg detection of TNT.11 They reported an improvement of about a factor of 70 over that of Martinez-Lozano et al. for the detection of TNT. In spite of achieving remarkable detection limits at levels below the equilibrium vapor pressures of many explosives at 25 °C, the perception remains that real-time vapor detection of
he continued development of trace explosives detection technologies for security and forensics remains a high priority. On the basis of world events, there is a growing need to deploy detection solutions outside of controlled environments such as airport checkpoints and into less structured environments such as mass transit stations, military checkpoints, and large public events. The nature of these environments necessitates effective noncontact sampling and detection methods, as typical contact based surface sampling techniques are invasive and time-consuming. Sampling for particles rather than vapor has been the explosives detection paradigm for decades,1−4 due to the exceedingly low vapor pressures of many explosive compounds.5 The equilibrium vapor pressures for explosives such as RDX and PETN are in the single digit pptv range, and actual concentrations in the air from an uncontained explosive mass are likely 3 orders of magnitude or more lower due to dilution and losses to surfaces in the environment.6,7 However, from an operational standpoint, the ability to detect vapors directly rather than collect particles is highly attractive. Most methods developed to sample and detect vapors from explosives with vapor pressures below that of TNT (e.g., RDX, PETN) depend upon preconcentration and subsequent thermal desorption,8 which is often too time-consuming to be effective in a highthroughput environment. The development of secondary electrospray ionization (SESI) has demonstrated improved ionization efficiency with the formation of specific adducts, suggesting the potential for vapor detection of low-volatility explosives.9−11 In 2004, Tam and Hill demonstrated the use of SESI with ion mobility spectrometry for explosives detection noting a 20-fold improvement in the detection limits for RDX © XXXX American Chemical Society
Received: August 8, 2013 Accepted: October 3, 2013
A
dx.doi.org/10.1021/ac402513r | Anal. Chem. XXXX, XXX, XXX−XXX
Analytical Chemistry
Article
(5 mm in diameter and 1 cm in height, Eiko A3C, Shawnee, Kansas) with one of the inner leads connected to one end of a rf power supply which produced a 42 kHz waveform with a 2.5 kV p-p voltage using approximately 60 mA of current. The other lead from the power supply was connected to a stainless steel wire mesh wrapped around the outside of the bulb. The ionization source was operated in air, producing nitrate ions (NO3−, m/z 62), the hydrated nitrate ion (H2O·NO3−, m/z 80), and the nitric acid adduct of the nitrate ion (HNO3·NO3−, m/z 125) as the predominant background species.13,14 The exit of the copper tube was inserted into a cylindrical aluminum housing (approximately 15-cm inner diameter) that was fastened onto the front end of the mass spectrometer. The copper tube was inserted so that the end of the tube was approximately 0.5 cm away from the mass spectrometer orifice. The aluminum housing contained a 1/8-in. port to which the house vacuum was connected. As a result of the applied suction at the exit end of the AFT, the ions were conveyed from the ionization source to the inlet of the mass spectrometer by the bulk flow of air at a flow rate of approximately 3.5 L/min. At this flow rate, ion residence times within the reaction region were approximately 5.5 s. No heat was applied to either the gas or the AFT device and both were held at ambient temperature, approximately 22 °C. Further, with this flow rate and the entrance to the 1 in. o.d. tube open to the laboratory, the reaction chamber was close to ambient pressure of ∼750 Torr. Mass Spectrometer. An API-5000 (AB-Sciex, Framingham, Massachusetts) triple quadrupole mass spectrometer was interfaced to the AFT. Minimal modifications to the instrument included removing both the commercially supplied ionization source and the interface plate providing the plenum gas. All spectra acquired under this effort were of negative ions with interface voltages minimized to reduce dissociation that may occur at the aperture. Using the vendor-defined nomenclature the following potentials were used: DP at −5 V, EP at −2 V, interface heater off, and the AFT was held at ground potential. For experiments using selected ion monitoring (SIM), a dwell time of either 100 or 200 ms was used and noted in the text. Chemicals and Materials. Stock solutions of 0.1 mg/mL PETN in methanol, 1.0 mg/mL tetryl in methanol/acetonitrile (1:1), 1.0 mg/mL RDX in methanol/acetonitrile (1:1), and 0.1 mg/mL NG in ethanol were obtained from AccuStandard (New Haven, Connecticut). Purified air, used for sample vapor generation, was provided by compressed lab air and purified by a Parker Domnick Hunter pure air generator (model CO2RP140, Gateshead, United Kingdom). Moisture content of the purified air was approximately 40 ppm as measured by a GE Panametrics Moisture Image series I hygrometer (Billerica, Massachusetts). The flow of gas used for sample introduction was controlled using a Cole-Parmer mass-flow controller (model 32907-69, Vernon Hills, Illinois). A total of 10 other commercial explosive and propellant formulations were used in solid form. These samples included PE4, a plastic explosive similar to C-4 containing RDX; Primasheet 1000 and Semtex 1A, both plastic explosives containing PETN; Semtex-H, a plastic explosive containing a mixture of RDX and PETN; Red Dot and Blue Dot, both smokeless powders containing 4−40% NG; Geldyne and Powerfrac, both gelatin dynamites containing 1−20% NG; Dynomax Pro and Unimax, both gelatin dynamites contain 3−30% NG. Sample Introduction and Vapor Generation. For the tetryl, RDX, PETN, and NG pure vapors, 1−2 μg (from stock solutions) were deposited on a few milligrams of quartz wool
explosive compounds such as RDX and PETN is not possible. To address this misconception, we have recently demonstrated the direct detection of RDX vapor, in real time, at parts-perquadrillion (ppqv) levels at ambient temperatures.12 Those results, combined with the experimental evidence presented within this manuscript, convincingly demonstrate direct realtime vapor detection of low volatility explosives well below saturated vapor levels. More specifically, the current experiments were designed to sample ambient laboratory air and maintain all regions of the instrumentation at ambient temperature (20−25 °C) to minimize the potential impact of particles, all while achieving ppqv levels of detection. The work presented in this manuscript extends the earlier publication12 to include additional analytes and demonstrates detection of these in common formulations. This work also explores the operational parameters for this new detection system. The key result from the previous work was the ppqvlevel sensitivity for RDX resulting from atmospheric pressure ionization reaction times of several seconds, combined with selective ionization via a nitrate reactant ion. RDX sensitivities of approximately 25 ppqv were demonstrated using single ion monitoring of the stable RDX adduct with NO3−. In this work, the same nitrate based ion chemistry was used in the atmospheric flow tube mass spectrometer (AFT-MS) described previously12 to achieve direct, real time detection of the explosives compounds RDX, PETN, tetryl, nitroglycerin. Also, since explosives are rarely encountered outside the laboratory in pure form, a number of different explosives formulations were evaluated for vapor based detection of the main explosive component directly from mixtures. These included military and commercial explosives formulations as well as smokeless powders. This approach not only allowed for confirmation of detection capabilities of the explosive constituent directly from the formulation but also demonstrated no observable effects from either the background laboratory air or the inert components in the formulations.
■
EXPERIMENTAL SECTION Atmospheric Flow Tube (AFT). Shown in Figure 1 is a schematic of the AFT-MS. The design of this apparatus has
Figure 1. Schematic of the atmospheric flow tube (AFT) mass spectrometer. Room air was drawn down the tube at a rate of about 3.5 L/min.
been described previously.12 The reaction region, consisting of a 1-in. outer diameter copper tube (2.36 cm inner diameter and 71 cm long), was connected to a 1-in. tee and maintained at or near ambient temperature and pressures. The ionization source was positioned in the center of the tee providing an overall reaction region of approximately 74 cm long with a volume of about 324 mL. The ionization source used was a dielectric barrier discharge and consisted of a miniature neon light bulb B
dx.doi.org/10.1021/ac402513r | Anal. Chem. XXXX, XXX, XXX−XXX
Analytical Chemistry
Article
inside a 1/4-in. stainless steel tube and allowed to dry. Purified air was supplied to the inlet of the 1/4-in. tube from a mass flow controller with the tube outlet placed just inside the entrance of the AFT. A pump on the downstream end of the AFT pulled the combination of both laboratory air and explosive vapor at a rate of 3.5 L/min. Vapors were generated at ambient temperature and pressure (normally 22 °C in the laboratory and about 750 mmHg for Richland, Washington). For the 10 explosive formulations, a few milligrams (less than 20 mg) of each was placed in the bottom of separate 2 mL glass vials with screw top caps. To sample the vapor, the cap was removed and the vial was placed on a stand with the top just inside a 1 in. elbow connection on the inlet to the AFT-MS to draw headspace vapors from the vial. The AFT-MS was sampling ambient laboratory air at a rate of approximately 3.5 L/min. Safety Considerations. This manuscript describes experimentation with explosive materials. Care should be exercised when handling any explosive compounds by eliminating all potential sources of initiation. Further, the risks can be mitigated by minimizing the amount of material used and eliminating sources of external energy such as friction, impact, or static discharge. Researchers should seek out accepted best practices for guidance in handling explosive materials.
■
RESULTS AND DISCUSSION In contrast to most mass spectrometry-based systems, the ion source used here is not directly adjacent to the atmospheric pressure interface. Rather, the ion source was distantly oriented (74 cm) from the MS inlet. This configuration increases the probability of reactant ion interactions with the analyte molecules, enabling charge transfer or adduct formation to occur. While the absolute signal intensity is diminished by distantly locating the ion source from the MS inlet, the large number of collisions, combined with selective ionization, enhances the signal-to-noise ratio. Examination of the negative ions produced in unfiltered laboratory air illustrates three predominant peaks at m/z 62, 80, and 125, which correspond to NO3−, H2O·NO3−, and HNO3·NO3−, respectively. Because of the high electron affinity of the nitrate (∼3.9 eV),15,16 these ion species represent the terminal products of ionized air. The nitrate ions have not been observed to give up their charge to other species and appear to form adducts with only few analytes. Thus, even with the extended reaction times studied here, the background spectra of ambient air are relatively simple and yield only a few minor peaks in addition to the three nitrate based ions present. Nitro-organic explosives tend to be highly polar and commonly form stable adducts with a variety of anions at atmospheric pressure including chloride, nitrite, and nitrate ions.17 Here the nitrate based ions serve as the reactant ion species for adduct formation with explosive analytes. A representative spectrum for PETN vapor and the corresponding background (inverted spectrum) are shown in Figure 2A. Each of these spectra is the result of 1 min signal averaging. The peaks directly attributed to the PETN ion clusters are labeled in Figure 2A along with their calculated signal-to-noise ratios (SNR). For the raw spectrum, the SNR for the PETN·NO3− and the PETN·NO3−·HNO3 were 25.8 and 12.6, respectively. Signal to noise ratios for both the raw and backgroundsubtracted spectra were calculated by dividing the analyte peak intensity by 3 times the standard deviation of the ion signal originating between 450 and 500 m/z. The background spectrum of ambient laboratory air was collected first, followed by the spectrum of PETN vapor at a flow of 500 mL/min
Figure 2. Spectra of PETN vapor at room temperature diluted with additional air flow by a factor of 1:7: (A) 1-min average for both background room air (inverted) and PETN vapor and (B) background subtracted spectra of the two showing an increase in the signal/noise ratio.
supplied to the inlet of the AFT. The PETN vapor (presumed near the saturated level at 25 °C = 11 pptv)5 was thus diluted by a factor of 1:7 providing a concentration of 1.6 pptv. Background Subtraction. While analyte adducts may be observed in the raw spectra, these signals are quite small when compared to the reactant ion peaks but may be improved by means of background subtraction. As shown in Figure 2B, the background subtracted signal is improved as a result of reducing peaks unrelated to the analyte. First, the background subtraction reduces the magnitude of the reactant ions which dominate the spectra, allowing the product ions to be more prominent on the same vertical scale. This provides for rapid visual discovery of the appearance of analyte ions. Since low levels of chemical noise occur at nearly every mass, background subtraction reduces this type of noise providing a higher signal/ noise ratio, thus improving detection limits. By using background subtraction, the SNR was increased by a factor of ∼1.5 over the nonsubtracted spectra for both PETN product ions. Random variations in signal intensity prohibit background subtraction from being 100% effective, resulting in some negative values. The most notable are some negative signals occurring for the reactant ion species at m/z 62 and 125. This could be a result of overall ion signal reduction during sample introduction or it could result from a direct reduction of the reactant ions through the conversion to the explosive adduct. However, in either case, peaks resulting from low intensity signals do not display these artifacts and the resulting spectra provide clearly identifiable product ions for PETN. Vapor Detection. Additional spectra demonstrating the vapor detection of tetryl, NG, and RDX are provided in Figure 3 at flow rates of 1000, 500, and 1000 mL/min, respectively, C
dx.doi.org/10.1021/ac402513r | Anal. Chem. XXXX, XXX, XXX−XXX
Analytical Chemistry
Article
response to tetryl vapor at different flow rates (20−800 mL/ min as indicated) for the SIM m/z 349 is presented in Figure 4.
Figure 4. Selected ion monitoring (SIM) with a 200 ms dwell time of the tetryl-nitrate adduct ion of m/z 349 at various sample flow rates (mL/min as indicated on the graph above each response) and diluted into a sampling flow of 3500 mL/min. The sample flow was shut off between each change to the sample flow rate.
The displayed SIM spectra were smoothed using the Loess method with a factor of 0.5 and a second order polynomial. The combination of the generated vapor and laboratory air (total flow of 3500 mL/min) along with the saturated vapor pressure for tetryl of 7.4 pptv5 can be used to estimate the vapor concentration within the AFT. This value represents a maximum possible concentration as it excludes analyte losses to the surfaces of the AFT and assumes initial saturated vapor from the source. Using these reasonable estimates, the observed ion signals correspond to tetryl concentrations ranging from 0.04 to 1.69 ppt. A full accounting of these values is highlighted in Table 1 along with the similar data collected for PETN, RDX, and NG. Along with vapor concentrations determined from vapor pressures and dilution flow ratios, estimates of vapor concentration based upon signal intensities and reaction time can be determined from eq 1 and provide an alternative assessment of detection limits.18
Figure 3. Spectra of tetryl, NG, and RDX vapor at room temperature diluted by a factor of 2:7, 1:7, and 2:7, respectively. These spectra are displayed as background subtracted spectra similar to that of Figure 2.
diluted into an overall sampling flow rate of 3500 mL/min. These dilutions, using estimated saturated levels at 25 °C,5 correspond to analyte concentrations of 2.1 pptv, 92 ppbv, and 1.4 pptv for tetryl, NG, and RDX, respectively. The spectra here result from the background subtraction; the background spectra used for subtraction were obtained by summing the spectra for 30 s prior to sample introduction. It should be noted that an equal number of background and analyte-containing spectra were used to produce the background subtracted spectra. For the SNR reported, the maximum peak heights were compared to 3 times the standard deviation of the signal from 450 to 500 m/z. In addition to the analyte peaks indicated, the nitrate based reactant ions appear on the left of each spectrum at m/z 62 and 125. The product ions for tetryl are at m/z 349 and 412, for NG at m/z 289 and 352, and for RDX at m/z 284 and 347, which are adducts of each explosive with two reactant ions. The presence of two peaks for each explosive provides additional confirmation over a single peak for the identification of these explosive materials. Further, these product ion peaks provide the basis for selected ion monitoring (SIM) to enhance sensitivity. Detection Limits. An attempt to estimate the detection limits of several explosives was based upon the assumption that saturated levels of explosive vapors were generated from the 1 /4-in.-outer diameter sample tube containing explosive residue on quartz wool. The saturated concentration levels used for calculations were from values reported in the literature.5 A controlled flow rate of air was passed through the sample tube and added at the front of the AFT. The total flow entering the reaction region was kept close to a constant level of 3.5 L/min from suction (a combination of generated vapor and laboratory air). Data were collected using selected ion monitoring (SIM) of the reactant ion at m/z 62 and the corresponding product ions of the analytes with a 200 ms dwell time each. The
[A−] = [R−]0 [A]kt
(1)
In eq 1, k is the reaction rate constant, t is the reaction time of 5.5 s, [A] is the concentration of the analyte, [A−] is the concentration of the analyte ions (measured signal), and [R−]0 is the initial reactant ion concentration (measured signal). Since ionization is occurring at atmospheric pressure, the high rate of collisions allows the approximation of the third-order clustering rate constant as an effective second-order rate constant, assumed here to be the collision rate constant, kc. An estimated value for k of 2 × 10−9 cm3 molecule−1 s−1 was used. Generating and measuring representative vapor concentrations at ppqv levels is challenging, and thus careful validation of detection limits at these levels needs to occur. Therefore, the two independent calculation methods described above (flow based dilution and ion intensity with eq 1) were used to quantify the detection capabilities of the AFT mass spectrometer measurement approach. The results of both of these calculations are provided in Table 1 in units of ppqv. Vapor concentrations based upon dilution ratios used saturated equilibrium vapor pressures at 25 °C of 11 pptv for PETN, 7.4 pptv for tetryl, 4.9 pptv for RDX, and 650 ppbv for NG.5 Vapor concentrations based upon eq 1 used the average analyte ion intensity and the average initial reactant ion intensity along with D
dx.doi.org/10.1021/ac402513r | Anal. Chem. XXXX, XXX, XXX−XXX
Analytical Chemistry
Article
Table 1. Calculated Concentrations of PETN, Tetryl, RDX, and NG with a Constant Total Flow (Both Sample and Laboratory Air) of 3500 mL/mina [PETN] ppqv
[Tetryl] ppqv
[RDX] ppqv
[NG] ppqv
sample flow mL/min
A
B
A
B
A
B
A
B
20 50 100 200 300 500 800 100 50
63 157 314 629 943 1570 2510 314 157
3.6 7.2 15.6 31.5 51.3 95.7 233 18.3 7.9
42 106 211 423 634 1060 1690 211 106
2.9 6.6 10.6 16.9 27.0 40.9 74.2 11.0 7.3
28 70 140 280 420 700 1120 140 70
7.9 14.6 24.7 50.9 70.6 134 192 24.1 13.4
3.69 × 106 9.21 × 106 18.4 × 106 36.9 × 106 55.3 × 106 92.1 × 106 147 × 106 18.4 × 106 9.21 × 106
8.3 29.0 81.4 199.9 305.8 525.9 1136.6 158.5 94.1
Two methods are used to calculate vapor concentration at each flow: A, the left column, is based upon the dilution flow ratio and corresponding vapor pressure values, and B, the right column, is based upon the ion intensities using eq 1. a
an ion residence time t = 5.5 s and an estimated k = 2 × 10−9 cm3 molecule−1 s−1. With the exception of nitroglycerine (which will be discussed later), the concentration values calculated from the ion signal are approximately an order of magnitude lower than the concentrations predicted by dilution of the saturated vapor. The discrepancy between these values is reasonable, since the calculated value based upon the ion signal is lower than the value predicted using the vapor pressure, which represents the theoretical maximum level. Discrepancies in the values between the two methods are likely a result of losses of the vapor to the walls or the vapor source not being at equilibrium. Smaller errors could also result from the estimation of the collision rate constant, kc. To determine if losses of analyte to the walls were significant, the upstream distance between the sample introduction location and the ionization source (left of the ionization source in Figure 1) was varied while monitoring the signal intensities of m/z 62 and 284. A glass microscope slide with a few micrograms of RDX residue was placed just inside the front opening of a 1 in. o.d. sample introduction tube, which was replaced with similar tubing of different length to provide a varying distance of analyte travel before the source. Using the AFT with a 74 cm reaction chamber (the distance between the ionization source and the mass spectrometer), a constant air flow of 5 L/min of purified air was supplied to the AFT, streaming across the glass slide and toward the mass spectrometer. The reaction distance, ionization source conditions, explosive vapor source term, and flow rate were fixed. The varying tube length value (upstream of the ion source) along with the resulting ion intensities are listed in Table 2. As shown in the table, there are measurable losses of RDX vapor to the walls of the system prior to the sample reaching the ionization source. This is evident since the reactant ion
species intensity remains unchanged while the RDX signal drops, indicating that the RDX concentration at the ionization source decreases with the increases in length of the inlet tube. When plotted, these losses follow an exponential decay. This demonstrates that even if saturated vapor is presented to the AFT, some losses do occur, as the sample vapor traverses the inlet tube, ion source, and reaction region. Sample losses to the walls could possibly be overcome by heating the inlet tube and the reaction regions in a future design. Part of the goal in this work was to maintain all surfaces at ambient temperature to avoid volatilizing any particles and artificially enhancing the direct vapor concentration measurement. Vapor concentration calculations between the two methods (dilution flow ratio labeled A in Table 1 and ion intensity method labeled B) vary by a factor of about 20 for PETN and tetryl and only about a factor of 6 for RDX. If a loss to the walls was the only contributing factor, the expected differences should be nearly the same for all three analytes. Additional factors impacting these results could be related to differences in the reaction rate constant or different binding energies (or affinities) between the nitrate ion and the various analytes. Alternately, the effect could also be related to challenges in vapor generation (e.g., vapor not at saturated levels or surface effects related to analyte release from substrate). These factors will require further investigation. As shown in Table 1, the calculated concentrations for NG are substantially different from each other. The vapor pressure of NG is approximately 5 orders of magnitude higher than the other three explosives studied, yet the response at low flow rates produces similar ion intensities. The significantly higher vapor pressure of NG makes the generation of vapors by the same method difficult. Depositing a few micrograms onto quartz wool was chosen for the low volatility species to provide a large surface area, with the hopes of maintaining saturated vapor over a variety of flow rates. The losses of NG from the source term appear to be significant. This was tested by placing a glass slide in the inlet of the AFT. About 8 μL of the 0.1 mg/ mL NG solution was placed on the slide and the NG response (SIM 289) was monitored over time. The signal showed an initial spike during deposition, followed by a very low signal that corresponded to the time period during which the solvent was evaporating. Once the solvent evaporated, a significant NG response was observed that lasted for approximately 1 h after which the signal decayed down to near zero levels within about 90 min. This indicates that the NG evaporates relatively quickly, and it is surprising that any vapor remained on the
Table 2. Effects of Sample Distance Prior to the Ionization Source within a 1 in. o.d. Metal Tube on Nitrate and RDX Ion Signal Intensities inlet tube length (cm)
NO3− intensity (cps)
RDX·NO3− intensity (cps)
7.62 30.5 91.4 185
× × × ×
4.7 × 103 3.5 × 103 1.9 × 103 0.28 × 103
4.4 4.5 4.5 4.5
5
10 105 105 105
E
dx.doi.org/10.1021/ac402513r | Anal. Chem. XXXX, XXX, XXX−XXX
Analytical Chemistry
Article
in signal with a corresponding increase in sample flow (and thus increasing concentration). The detectable levels shown here and those published earlier12 indicate real-time detection levels below ten parts per quadrillion. These concentrations were based upon calculations derived from signal intensity, reaction time, and an estimated reaction rate constant. These calculated concentrations were about an order of magnitude lower than the values estimated from vapor pressure and dilution flows. With the evidence that sample losses occur to the walls, the ion signal calculated values should always be lower than those determined by dilution ratios. To validate concentration levels by yet another method, desorption of a known mass was investigated. In this experiment, a small coil of Nichrome wire was placed just inside the inlet of the AFT. A suction flow of ∼6 L/min was applied to the AFT. A 1 μL drop of a solution containing 0.2 pg/μL RDX in methanol was deposited on the wire and allowed to evaporate for several minutes, leaving 200 fg of RDX residue on the wire. Upon desorption of the wire, a sharp peak with a width at baseline of about 6.5 s was observed using SIM of the RDX ion at m/z 284. A rough concentration of 32 ppqv was estimated from the 200 fg of RDX present in the 650 mL of air passing through the system in the 6.5 s desorption. The peak maximum was 70 cps with an average noise prior to desorption of 6 ± 5 cps. The value of 32 ppqv is similar to the value of 28 ppqv for RDX vapor determined by dilution flow ratios. This pulsed method of sample introduction also does not account for vapor losses to the walls. Thus, actual detection limits of the system will be lower than these values and likely approaching those calculated (
