ARTICLE pubs.acs.org/ac
Modular Ion Mobility Spectrometer for Explosives Detection Using Corona Ionization Kristyn M. Roscioli,† Eric Davis,† William F. Siems,† Adrian Mariano,‡ Wansheng Su,‡ Samar K. Guharay,*,‡ and Herbert H. Hill, Jr.*,† † ‡
Department of Chemistry, Washington State University, Pullman, Washington 99164, United States The MITRE Corporation, 7515 Colshire Drive, McLean, Virginia 22102, United States ABSTRACT: Ion mobility spectrometry (IMS) has become the most widely used technology for trace explosives detection. A key task in designing IMS systems is to balance the explosives detection performance with size, weight, cost, and safety of the instrument. Commercial instruments are, by and large, equipped with radioactive 63Ni ionization sources which pose inherent problems for transportation, safety, and waste disposal regulation. An alternative to a radioactive source is a corona discharge ionization source, which offers the benefits of simplicity, stability, and sensitivity without the regulatory problems. An IMS system was designed and built based on modeling and simulation with the goal to achieve a lightweight modular design that offered high performance for the detection of trace explosives using a corona ionization source. Modeling and simulations were used to investigate design alternatives and optimize parameters. Simulated spectra were obtained for 2,4,6trinitrotoluene (TNT) and cyclo-1,3,5-trimethylene-2,4,6-trinitramine (RDX) and showed good agreement with experimentally measured spectra using a corona ionization source. The reduced mobilities for TNT and RDX obtained with corona ionization were 1.53 and 1.46 cm2/(V s), respectively, and this agreed well with literature values.
T
he detection of explosives is critical for security applications. Ion mobility spectrometry (IMS) has proven to be one of the effective analytical methods for detection of explosives due to the low detection limits, fast response, simplicity, and portability.1 A variety of ionization sources can be used in IMS.2 Several research groups38 have been actively engaged in activities on ionization sources, and many important results have come out of these activities. Radioactive materials, such as 63Ni, are commonly used as the ionization source in commercial ion mobility spectrometers. 63Ni has a half-life of 100.1 years, and this yields continuous and steady streams of energetic β particles over a long time and sustains ionization. While these sources are stable and do not require any electrical power, the use of radioactive materials requires stringent regulatory compliance in terms of scheduled monitoring of radiation leakage and disposal. Corona ionization sources have the advantage of being relatively inexpensive. In addition, they are simple to construct, operate continuously while consuming little power, and generate a total ion current an order of magnitude greater than that produced by 63Ni.6 A potential disadvantage to using corona ionization is that the continuous operation of the source can lead to the production of ozone and neutral oxides of nitrogen which generate CO3, O3, and NO3 reactant ions. It is believed that the RDX 3 CO3 adduct has too short a lifetime to survive the typical transit time of several milliseconds through the drift tube, and when CO3 is the predominant reactant ion, the detection of RDX is hindered.3 A pulsed source ionization method was developed so that the buildup of these ions does not reach a sufficient concentration to r 2011 American Chemical Society
diminish responses of target explosives.9 Ross addressed the problem by conducting the corona ionization in the presence of a counter buffer gas similar to that developed in our laboratory to sweep the neutral ozone and nitrogen oxides away from the reaction region before the formation of the aforementioned short lifetime ions occurs.10 There are questions on degradation in corona source performance and eventual failure due to erosion at the needle tip.4,11 A significant effect of this change in surface character as a function of time can be seen in the continuous operation of a dispersed plasma ionization source; after 33 days of continuous operation, corrosion of the entire electrode surface was observed.4 Besides the elimination of the 63Ni ionization source, another criterion for the construction of a desirable IMS instrument for explosive detection is that the instrument be light and durable. In traditional stacked ring designs of ion mobility instruments, each electrode must be electrically insulated from the adjacent electrode and voltages on these IMS electrodes must be insulated from the cover of the instrument. While attempts have been made to use various types of polymers and air gap insulators, the best approach has been the use of ceramic insulators. However, ceramic insulators can add a significant fraction of weight to the instrument and make the overall unit bulky. A third criterion is that the IMS have as high a resolution as possible in order to reduce false positive responses. The ion separation Received: April 12, 2011 Accepted: June 17, 2011 Published: June 17, 2011 5965
dx.doi.org/10.1021/ac200945k | Anal. Chem. 2011, 83, 5965–5971
Analytical Chemistry
ARTICLE
Table 1. Simulation Conditions Used for Five Different Cases Examining the Effect of Electrode Number and Dimensions design label
A
B
C
D
E
electrode thickness (mm)
4.3
2.7
6
10
15
insulator thickness (mm)
0.9
2.4
2
2
2
number of electrodes
23
23
15
10
7
drift region total length (cm)
12.1
12.26
12.5
12.7
12.85
potential across drift region (V) peak width for beam with 1 cm
3002 328
3000 312
2995 336
2996 362
3004 364
308
300
324
362
362
316
318
332
398
460
diameter (μs) peak width for beam with 2 cm diameter (μs) peak width for beam with 4 cm diameter (μs)
capability of an ion mobility instrument is expressed quantitatively as resolving power. This value can be theoretically calculated and is often used to show how operating conditions and instrument dimensions can affect the peak separation in an ion mobility experiment.12 More specifically, a “conditional” resolving power can be used to quickly approximate realistic resolving powers for specific instrumental operating parameters and compounds. The efficiency of an ion mobility spectrometer (expressed as a percent) is used to describe the relationship of the theoretical and experimentally measured resolving powers.13 It is desirable to achieve the highest efficiency possible when building an IMS, and this value is affected by factors such as voltage, drift tube length, gate pulse width, temperature, and pressure. The final criterion for the construction of an IMS is that the instrument be modular in design. A majority of commercial IMS instruments developed for explosive detection are sealed systems due to the 63Ni ionization source.14 If the 63Ni source can be eliminated, then a modular IMS can be constructed with interchangeable parts. Modular IMS instruments offer the flexibility to adapt a variety of ionization sources, such as corona ionization source and electrospray ionization source, as well as different detection methods, such as Faraday plate, mass spectrometry, and surface enhanced Raman spectroscopy. The work presented in this paper uses a stable corona ionization source with a modular IMS design. Systematic modeling and simulation studies were undertaken to determine the sensitivity to geometrical parameters in predicting the IMS resolving power. One of the key objectives of the modeling and simulations was to determine the maximum occupancy of the drift tube cross-section for achieving the highest throughput of ions without any degradation of the resolving power. The results of the modeling and simulation studies were used to build a lightweight IMS system. The performance of a corona ionization needle was tested over time and used to detect IMS standards in the positive mode and explosives in the negative mode. The simulated and experimental results for TNT and RDX were compared, and mobility values were calculated for comparison with literature values.
’ MODELING AND SIMULATION With the use of SIMION as previously described,15 simulations were conducted to aid in the design of the IMS. Simulations included a systematic study of the dependence of resolving power
Figure 1. Simulations: (a) design C with the equipotential lines showing the extent of field nonuniformity; (b) design C showing the ion trajectories; (c) design E showing equipotential lines; (d) trajectories for design E. Note that the ions at the edges travel along a significantly longer path, broadening the IMS peak. The characteristic parameters for different designs are given in Table 1.
on electrode diameter and input beam size. Other system parameters were held constant, namely, the gate pulse width of 200 μs, the drift tube temperature of 500 K, the gas pressure of 760 Torr, and the potential difference across the drift region of about 3 kV. The effect of changing the number and dimensions of the electrodes on resolving power was calculated for a drift region about 12 cm long and 5 cm in diameter. The input beam diameter was varied, and its effect on the analyte peak width was noted. For a fixed length of the drift cell, five different electrode configurations were considered with different electrode sizes and correspondingly different numbers of electrodes. Table 1 shows several modeling and simulation case studies examining the effect of electrode number and dimensions. The simulations show similar peak widths of 328 and 336 μs for the cases with 23 and 15 electrodes, respectively. Instrument performance in these cases does not deteriorate even when 80% of the drift tube diameter is filled by the beam. In the case of 10 and 7 electrodes, overall performance deteriorated and a significant 5966
dx.doi.org/10.1021/ac200945k |Anal. Chem. 2011, 83, 5965–5971
Analytical Chemistry
ARTICLE
Figure 2. SolidWorks IMS drawings showing (a) outer casing, (b) electrode rings, (c) insulator rings, (d) sample inlet ring, (e) gas manifold, and (f) Faraday plate detector.
broadening of the IMS peak was observed. Some representative illustrations showing these effects of electrode geometry are given in Figure 1. A comparison of design C and design E is made to illustrate how the ion trajectories are affected by the number and size of electrodes; the characteristic parameters for these designs are described in Table 1. The ions in design E travel a longer path along the edge because of electric field nonuniformity resulting in a broadening of the IMS peak. Fabrication is simplified by using the smallest number of elements possible, and design C was selected because it has the fewest electrodes and yielded reasonably good ion trajectories.
’ EXPERIMENTAL SECTION IMS Construction. All components of the IMS were modeled in SolidWorks 2004 CAD (Figure 2). This modular IMS system consisted of a stacked alumina insulator (Alsint 997; 99.7% alumina; outer diameter (o.d.) of 6.1 cm; inner diameter (i.d.) of 4.7 cm) and stainless steel rings (o.d., 5.1 cm; i.d., 4.3 cm). The insulator and stainless steel rings formed an interlocking tube design that was encased in an anodized aluminum outer casing with a total length of 24.6 cm. Coating the aluminum outer casing of the IMS with a layer of aluminum oxide film eliminates the need for an extra ceramic layer between the electrodes and outer casing. Hard anodization is the process of producing a hard aluminum oxide (Al2O3) film on aluminum by electrolytic oxidation. The hard anodized aluminum process was originally developed in the early 1960s and has applications in numerous commercial and high-tech industries, including military and defense, aerospace, construction, electronics, and home cookware.16 The anodization proved to be an insulator in itself that can handle voltages of (5000 V. To protect the gate wires from arcing to the outer casing with this new design, a 127 μm thin layer of kapton foil was used. A kapton sheet heater was used in place of heating rods, eliminating the need for a thicker outer casing. This design allows for a lightweight instrument which is desirable for handheld systems and field use. The system was divided into two sections, a desolvation region and drift region, separated by a BradburyNeilsen type ion gate. The stainless steel rings were connected in series with 0.5 MΩ (desolvation region) and 1.0 MΩ (drift region) high temperature resistors ((1%). When a voltage of 5000 V was applied to the first ring in the series, an electric field of approximately 254 V/cm was created throughout the drift region of the IMS tube. A preheated drift gas was introduced at the Faraday plate
detector end of the drift region at a flow rate of 1 L/min. The drift gas used was air which created a counter flow of heated gas that exited the IMS tube at the front of the desolvation region. The gas line was heated over a range of approximately 100150 °C before entering the drift region. The tube itself was heated over a range of approximately 90150 °C using a kapton sheet heater that was sealed to the outside of the anodized outer casing. The IMS was operated at ambient pressure. The ion gate was made with parallel 75 μm alloy 46-stainless steel wires. The gate was open when all the wires have the same voltage as if it was in series with the electric field in the drift tube. The gate was closed when a voltage of (18.5 V was applied between each set of wires. This creates an electric field different from that of the drift tube, and it stops positive and negative ions from entering the drift region. A gate pulse width of 0.2 ms and repetition rate of approximately 20 Hz was used to allow pulses of ions to enter the drift region. The corona ionization source was constructed using a conical corona needle, with a tip radius of curvature of ∼25 μm. A 1000 MΩ resistor was connected to the end of the needle to ensure a stable ionization process. This allowed the needle to be positioned closely to the front target screen without arcing. The front of the IMS was a closed plate design with four 1.14 mm i.d. holes drilled in a Macor center piece to allow the drift gas to exit the drift tube. A small hole (3.18 mm i.d.) was drilled directly in the center of the plate to hold the corona needle. This mechanical arrangement ensures proper placement of the corona needle and optimization of the source performance. The front plate was held in place by threading it into the outer casing, allowing the ionization source to be easily changed. Sealing the ionization region of the drift tube and having a counter flow of drift gas allows continuous operation of the corona source without the buildup of neutral ozone and nitrogen oxides which hinder the detection of target analytes.10 The gas manifold (Figure 3) was held in place by four set screws and served as the closing end of the drift tube. The gas manifold held the Faraday plate in proper position and also allowed removal of the plate providing an easy way to interface the instrument to another detector if desired. Analyte samples were introduced in the desolvation region of the IMS through a sample inlet ring (Figure 3). This ring had the same dimensions as an electrode ring and was kept in series with the alternating electrode/insulator pattern. The ring was hollow with a small piece of tubing exiting the IMS. The sample was introduced through the tubing and escaped into the desolvation region through a series of small holes placed along the edges of the electrode. 5967
dx.doi.org/10.1021/ac200945k |Anal. Chem. 2011, 83, 5965–5971
Analytical Chemistry
ARTICLE
Table 2. Ion Mobility Spectrometer Operating Conditions Used during Experimental Data Collection parameter
setting
desolvation region length
8.90 cm
drift region length corona needle voltage
15.74 cm (8500 V
voltage at first ring
(5000 V
voltage at gate
(4089 V
gate closure potential
(18.5 V
gate pulse width
0.2 ms
scan time
47 ms
drift gas
air
drift gas temperature drift tube temperature
100 °C for + mode and 150 °C for mode 90 °C for + mode and 150 °C for mode
drift gas flow
1 L/min
’ RESULTS AND ANALYSIS
Experimental Arrangement for Detection of Explosives Using Corona Ionization. The operating conditions for this in-
Explosive Detection. The measured IMS spectra are shown in Figure 4. Each spectrum shows a labeled set of reactant ion peaks (RIP) and the target analyte ion peak. The reactant ion peaks in the negative mode spectra are a large peak at a drift time of 14.95 ms (K0 = 2.42 cm2/(V s)) and a relatively weaker peak at a drift time of 17.48 ms (K0 = 2.07 cm2/(V s)). The reduced mobility (K0) of each of the product ions was calculated using eq 1 and was reported in units of centimeters squared/(volt second): P 273 ð1Þ K0 ¼ K 760 T
strument for the positive mode were the following: corona needle voltage, +8500 V; voltage at gate, +4089 V; gate pulse width, 0.2 ms; scan time, 47 ms; ambient pressure; drift gas, air; drift gas temperature, 100 °C; drift tube temperature, 90 °C; drift gas flow, ∼1 L/min. The operating conditions in negative mode were the same as listed above except for the following: corona needle voltage, 8500 V; voltage at gate, 4089 V; drift gas temperature, 150 °C; drift tube temperature, 150 °C (Table 2). Chemicals standards included 2,4-lutidine and ditertbutylpyridine (dTBP) (ACS reagent grade, g 98% purity), 2,4,6-trinitrotoluene (TNT), and cyclo-1,3,5-trimethylene-2,4,6-trinitramine (RDX). A vapor generator was used to introduce 2,4-lutidine and dTBP. A capillary from the vapor generator was placed into the sample entry ring of the IMS. A 1.0 μL aliquot of each standard was placed inside a small piece of glass tubing filled with glass wool. Heated air gas flowed over the sample tube at a rate of 500 mL/min, and the sample was swept into the desolvation region of the IMS. The IMS drift tube temperature for 2,4 lutidine and dTBP was kept at 90 °C, and the drift gas temperature was 100 °C. TNT and RDX (1 mg/mL in acetonitrile solvent) were introduced into the IMS desolvation region as liquid through a syringe and capillary setup at a flow rate of 3 μL/min. The capillary was placed in the sample entry ring of the IMS. The IMS drift tube was held at a temperature of 150 °C, which allowed evaporation of the sample from the tip of the capillary. The drift gas at 150 °C then swept the samples toward the corona ionization source where the samples were ionized. The ions were gated into the drift region at 0.2 ms pulse width and collected with a Faraday plate at the opposite end of the drift tube.
Mass identification of the ions was not possible with the standalone system, but it has been shown that in the negative mode and with a high level of discharge gas, the NO3 species is the only ion present and appears as two peaks in the IMS spectrum, NO3 and the NO3 3 HNO3 adduct, with separate mobilities.3 The K0 values for the two reactant ion peaks (RIP) in the negative mode are estimated to be 2.43 cm2/(V s) for the first peak and 2.07 cm2/(V s) for the second peak. Previous work reported RIPs in the negative mode with K0 values of 2.52 8 and 2.56 cm2/(V s) 3 for the first peak and 2.15 cm2/(V s) 3 for the second peak. On the basis of the reduced mobility values and corresponding mass identification in previous work,3 the K0 values for RIPs in the current work (smaller by about 5% with respect to the literature values) can be attributed to water clustering with the RIPs in the present work; however, mass analysis along with IMS will be required to ascertain this point. In the positive mode, the reactant ion peaks were two peaks with a drift time of 18.00 ms (K0 = 2.01 cm2/(V s)) and 19.22 ms (K0 = 1.88 cm2/(V s)). In a previous study, it was shown that the reactant ions generated using corona ionization in air in the positive mode were water clusters of NH4+, NO+, and H3O+.8 The ions present in this system were most likely two of these three ions. The addition of ammonia vapor to this system could identify which peak correlates to the ammonium ion. The above K0 values for the RIPs agree reasonably well with the literature K0 values of 2.07 and 1.89 cm2/(V s).17 The reduced mobility for the target analyte ions were 1.44, 1.93, 1.53, and 1.46 cm2/(V s) for the ions [dTBP + H]+, [lutidine + H]+, [TNT H], and [RDX 3 NO3], respectively. These K0 values agreed with literature values within (0.02 cm2/(V s).1,3,17
Figure 3. (A) Faraday plate (dark gray) with gas manifold (light gray) setup and (B) sample inlet ring. The post leading to both parts is hollow making it possible to bring in the drift gas and sample vapors without interruption of the homogeneous electric field and allowing a modular design without bulky components for gas introduction.
5968
dx.doi.org/10.1021/ac200945k |Anal. Chem. 2011, 83, 5965–5971
Analytical Chemistry
ARTICLE
Figure 4. (1) Positive mode spectrum of ditertbutylpyridine, (2) positive mode spectrum of 2,4-lutidine, (3) negative mode spectrum of TNT, and (4) negative mode spectrum of RDX.
Table 3. Summary of Experimental Results and Comparison to Literature Values1,3,14 calculated analyte K0(cm2/
literature K0 experimental (cm2/(V s))
resolving
(V s))
theoretical
efficiency
resolving power
ratio
power
(%)
dTBP lutidine
1.44 1.93
1.43 1.95
48 47
70 66
68 70
TNT
1.53
1.54
49
67
73
RDX
1.46
1.46
44
72
61
The efficiency of the IMS instrument for detecting TNT and RDX was calculated, and these results are summarized in Table 3. Theoretical resolving powers (Rp) were calculated specifically for TNT and RDX using eq 2. The use of the theoretical resolving power equation has been studied and described elsewhere.13 L2 P 273 VK0 760 T ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi R p ¼ v0 u !2 1 u 16kT ln 2 L2 P 273 A u@ 2 t tg + Vez VK0 760 T
ð2Þ
Figure 5. Corona needle performance showing RIP intensity over 100 h of operation over several months; RIP for a new needle (red trace); RIP after 100 h of operation over several months (blue trace). No significant change of RIP intensity is noted indicating stable detection sensitivity. Additionally, no significant erosion of the needle is noted after several months of use.
The resolving power efficiency of an ion mobility spectrometer was determined by expressing the experimental resolving power as a percentage of the theoretical resolving power as seen in eq 4.
IMS measured resolving power (Rm), defined as the drift time (td) of the ion divided by the peak width at half-height (w1/2) of the ion mobility peak, was calculated for TNT and RDX. This relationship is given by eq 3. Rm ¼
td w1=2
ð3Þ
efficiency ¼
experimental Rm 100% theoretical Rp
ð4Þ
While it is desirable to obtain a resolving power that matches closely with the theoretical resolving power, this modular IMS yields an average efficiency of 68%. 5969
dx.doi.org/10.1021/ac200945k |Anal. Chem. 2011, 83, 5965–5971
Analytical Chemistry
ARTICLE
Figure 6. Experimental and calculated IMS spectra for (a) TNT, (b) lutidine, (c) dTBP, and (d) dinitro-orthocresol.
Table 4. Comparison of Measured and Calculated Peak Position and Resolving Power for the Cases Shown in Figure 6 peak location td(ms) analyte
resolving power R
measured
calculated
ratio
measured
calculated
ratio
TNT
23.5
24.1
0.975
49
67
0.73
lutidine
21.75
22.17
0.981
47
65
0.72
dTBP
29.05
29.68
0.979
48
71
0.68
DNOC
25.1
25.54
0.982
44
68
0.65
For stable IMS operation using corona ionization sources and achieving reliable performance characteristics, the characteristics of the corona needle, namely, tip geometry and surface smoothness, need to be preserved. In view of this, the corona needle was optically examined by a microscope before using it and again after 6 months of its use; additionally, the intensity of RIP was compared (Figure 5). No significant degradation of the corona needle was noted after about 100 hours of experimental use; most of the experimental runs pertained to IMS operation in the negative mode. There were no noticeable changes in the background current and sensitivity over time. Comparison of Measured and Simulated Spectra. The IMS system parameters and respective operating conditions were used in SIMION to calculate and predict expected spectra for the analyte ions introduced experimentally. The calculated and measured spectra are shown in Figure 6 for four compounds including TNT, 2,4-lutidine, dTBP, and dinitro-orthocresol (DNOC). The results are summarized in Table 4. The calculated peak position and shape reasonably agree with the respective measured values. The measured resolving power also matched predictions with a ratio of 0.73 for TNT, 0.72 for lutidine, 0.68 for dTBP, and 0.65 for DNOC. Further studies of the dependency of the resolving power on beam parameters, namely, beam distribution at the input, loss of particles, and space charge effects are warranted. As explained in a previous article,15 space-charge effects can play an important role here.
’ CONCLUSIONS Development of an alternative source in the detection of trace explosives by IMS is desirable due to safety and regulatory concerns associated with the use of radioactive ionization sources. The research presented here demonstrates the use of corona discharge ionization as a safer and less costly alternative with the added advantage of ease of operation. The common problem of neutral ozone and nitrogen oxide buildup was eliminated with a counter flow of drift gas and an enclosed IMS design. The neutrals were swept out of the IMS and, as a result, were prevented from interfering with the detection of TNT and RDX. The modular IMS system presented here has several advantageous characteristics. A threaded corona ionization needle assembly enabled interchange of ionization sources and optimization and reproducible positioning of the corona needle. A novel gas manifold ring was used to introduce the drift gas into the IMS and also hold the Faraday plate in proper placement. The Faraday plate, held in place by set screws, can be easily removed to allow an interface to a different detector. A sample inlet ring was used to introduce sample vapors without interruption of the homogeneous electric field. The modeling and simulation studies provide the design of an ion mobility spectrometer with optimal geometrical and operational characteristics for achieving a high resolving power. The instrument performance was tested with two IMS standards, namely, 2,4-lutidine and dTBP, in the positive mode and two explosives, TNT and RDX, in the negative mode. Reduced mobilities were calculated from measured spectra and agreed with literature values within error. The resolving power of the instrument was determined from the measured spectra, and it was estimated to be in the range of 44 to 49. The average efficiency of the instrument was determined to be about 68%. The predicted spectra for four compounds (TNT, 2,4-lutidine, dTBP, and DNOC) reasonably matched with the experimentally measured spectra. ’ AUTHOR INFORMATION Corresponding Author
*Samar K. Guharay: e-mail,
[email protected]; phone, 703983-1787. Herbert H. Hill, Jr.: e-mail,
[email protected]; phone, 509-595-1492. 5970
dx.doi.org/10.1021/ac200945k |Anal. Chem. 2011, 83, 5965–5971
Analytical Chemistry
ARTICLE
’ ACKNOWLEDGMENT This work is supported by the MITRE Innovation Program. The article is approved for Public Release: 11-1231. ’ REFERENCES (1) Khayamian, T; Tabrizchi, M; Jafari, M. T. Talanta 2003, 59, 327–333. (2) Guharay S. K., Dwivedi P, Hill H. H. (2008) Ion mobility spectrometry: ion source development and applications in physical and biological sciences. IEEE Trans. Plasma Sci. 36. (3) Ewing, R. G.; Waltman, M Int. J. Ion Mobility Spectrom. 2009, 12, 65–72. (4) Waltman, M. J.; Dwivedi, P; Hill, H. H.; Blanchard, W. C.; Ewing, R. G. Talanta 2008, 77, 249–255. (5) Ewing, R. G.; Waltman, M. J. Int. J. Mass Spectrom. 2010, 296, 53–58. (6) Tabrizchi, M.; Khayamian, T.; Taj, N. Rev. Sci. Instrum. 2000, 71, 2321–2328. (7) Tabrizchi, M. Anal. Chem. 2003, 75, 3101–3106. (8) Tabrizchi, M.; Ilbeigi, V. J. Hazard. Mater. 2010, 176, 692–696. (9) Hill, C. A.; Thomas, C. L. P. Analyst 2003, 128, 55–60. (10) Ross, S. K.; Bell, A. J. Int. J. Mass Spectrom. 2002, 218, L1–L6. (11) Dindosova, D.; Skalny, J. D. Acta Phys. Univ. Comenianae 1992, 33, 77. (12) Siems, W. F.; Wu, C.; Tarver, E. E.; Hill, H. H. Anal. Chem. 1994, 66, 4195–4201. (13) Kanu, A. B.; Gribb, M. M.; Hill, H. H. Anal. Chem. 2008, 80, 6610–6619. (14) Ewing, R. G.; Atkinson, D. A.; Eiceman, G. A.; Ewing, G. J. Talanta 2001, 54, 515–529. (15) Mariano, A.; Su, W.; Guharay, S. K. Anal. Chem. 2009, 81, 3385–3391. (16) Lee, W. JOM 2010, 62, 57–63. (17) Eiceman, G. A.; Karpas, Z. Ion Mobility Spectrometry; CRC Press: Boca Raton, FL, 1994.
5971
dx.doi.org/10.1021/ac200945k |Anal. Chem. 2011, 83, 5965–5971
