Fast Orthogonal Separation by Superposition of Time of Flight and

Dec 22, 2017 - This novel field asymmetric time of flight ion mobility spectrometer (FAT-IMS) allows high repetition rates and reaches limits of detec...
0 downloads 0 Views 1MB Size
Article Cite This: Anal. Chem. XXXX, XXX, XXX−XXX

pubs.acs.org/ac

Fast Orthogonal Separation by Superposition of Time of Flight and Field Asymmetric Ion Mobility Spectrometry Alexander Bohnhorst,* Ansgar T. Kirk, Marc Berger, and Stefan Zimmermann Institute of Electrical Engineering and Measurement Technology, Department of Sensors and Measurement Technology, Leibniz Universität Hannover, Appelstrasse 9A, 30167 Hannover, Germany ABSTRACT: Ion mobility spectrometry is a powerful and lowcost technique for the identification of chemical warfare agents, toxic chemicals, or explosives in air. Drift tube ion mobility spectrometers (DT-IMS) separate ions by the absolute value of their low field ion mobility, while field asymmetric ion mobility spectrometers (FAIMS) separate them by the change of their ion mobility at high fields. However, using one of these devices alone, some common and harmless substances show the same response as the hazardous target substances. In order to increase the selectivity, orthogonal data are required. Thus, in this work, we present for the first time an ambient pressure ion mobility spectrometer which is able to separate ions both by their differential and low field mobility, providing additional information for selectivity enhancement. This novel field asymmetric time of flight ion mobility spectrometer (FAT-IMS) allows high repetition rates and reaches limits of detection in the low ppb range common for DT-IMS. The device consists of a compact 44 mm drift tube with a tritium ionization source and a resolving power of 70. An increased separation of four substances with similar low field ion mobility is shown: phosgene (K0 = 2.33 cm2/ (V s)), 1,1,2-trichlorethane (K0 = 2.31 cm2/(V s)), chlorine (K0 = 2.24 cm2/(V s)), and nitrogen dioxide (K0 = 2.25 cm2/(V s)). Furthermore, the behavior and limits of detection for acetonitrile, dimethyl methylphosphonate, diisopropyl methyl phosphonate in positive polarity and carbon dioxide, sulfur dioxide, hydrochloric acid, cyanogen chloride, and hydrogen cyanide in negative polarity are investigated.

I

vd = K(E)·Ed

on mobility spectrometry (IMS) is a powerful tool for fast detection of trace gases in air. Some of the outstanding features provided by the IMS are the relatively low complexity, short response times, and low limits of detection.1 Therefore, it is well-suited for applications in which size and weight of the analytical system, and a fast detection of low concentrations is a crucial factor, e.g. the on-site detection of chemical warfare agents,2 toxic industrial chemicals,3 explosives,4,5 or substance abuse.6 Nevertheless, most commercially available systems still suffer problems from false positives due to a lack of analytical power.3 Therefore, we propose a method based on orthogonal separation to improve the analytical performance of IMS especially for compact and fast systems. Basically, ion mobility spectrometers can be divided by their principle of ion separation. Drift tube ion mobility spectrometers (DT-IMS) separate ions by the absolute value of their ion mobility as a result of an ion motion in a static electrical field along the axis of a drift tube.1 Field asymmetric ion mobility spectrometers (FAIMS) act as ion filters, separating ions by their differential mobility by employing an alternating, dynamic electrical field.7 A simplified schema of a DT-IMS is illustrated in Figure 1. A discrete portion of ions, located in the ion source, is injected into the drift cell and accelerated by the present drift field Ed. The drift velocity vd is given by eq 1 with the ion mobility K in the bath gas as a function of E. © XXXX American Chemical Society

(1)

Due to the low electrical fields used in common DT-IMS, the field dependent term of the ion mobility is negligible. Thus, the mobility can be expressed by the time td the ions need to pass the drift cell, the drift voltage Ud, and the drift length Ld. Normalized to standard conditions (T0 = 298.15 K, p0 = 101.3 Pa) with the working temperature T and the working pressure p yields the reduced mobility K0 (eq 2) under low field conditions. K0 =

Ld2 T0 p · · td·Ud T p0

(2)

By resolving the time dependent ion current, a mobility spectrum can be obtained in which the peak maxima mark the characteristic mobilities of the corresponding ion species. Since DT-IMS can gather an entire spectrum within a few milliseconds, the response time is extremely short, while the limits of detection are in the lower ppbv to pptv range.8 Nevertheless, a major drawback of the most commercial available IMS devices is the low resolving power. The resolving Received: August 9, 2017 Accepted: December 12, 2017

A

DOI: 10.1021/acs.analchem.7b03200 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Figure 1. Schema of a DT-IMS (left) and a FAIMS (right).

power Rp is a measure for the sharpness of the generated peaks and is given by Rp =

td w0.5

zero yields a shift to one of the closely spaced electrodes where the corresponding ions recombine (circles and rectangles). By superimposing the dispersion field with a weak compensation field, this motion caused by the differential mobility can be compensated. Thus, by sweeping the compensation field, an ionogram, describing the alpha function of the ions in the present dispersion field, can be captured. A separation of overlapping substances with closely spaced low field or differential mobilities and even of isotopologues11 in a DT-IMS or FAIMS is possible by increasing the resolving power.12,13 However, a high resolving power requires, in the case of a DT-IMS, long drift cells, high electrical field, and, in the case of the FAIMS, a special bath gas, which is not feasible for applications in which the portability of the analytical system is a crucial factor. Furthermore, a separation of substances possessing nearly the same low field or differential mobility remains impossible, even in IMS or FAIMS with outstanding resolving power. In order to overcome this problem, a well-known solution is the coupling of two orthogonal separation techniques, e.g. gas chromatography (GC) with DT-IMS.14 In GC-IMS, the IMS is used as a detector for a GC, which preseparates substances by their retention time. However, this significantly increases the measuring time and system complexity. Another possibility is the coupling of FAIMS with DT-IMS. Here, the FAIMS is used as ionization source and ion filter. However, common FAIMS IMS systems15,16 only analyze the ions passed through the FAIMS at a given time. To record the total IMS spectrum, the compensation voltage of the FAIMS has to be swept through a certain range leading to low repetition rates in nontargeted analysis. Hence, alternative separation techniques are needed. In the literature, the high kinetic energy ion mobility spectrometer (HiKE-IMS) has been presented as a possible solution.17,18 The HiKE-IMS consists of a DT-IMS, which operates at variable reduced electric field strength up to 120 Td. Like a common DT-IMS, the HiKE-IMS generates a spectrum of all present ions within a few milliseconds. Thus, a drift field sweep reveals the alpha function of the entire spectrum. Nevertheless, the complexity of a HiKE-IMS is high and currently not suitable for hand-held devices. Since there is no other known concept matching all needed requirements, an innovative approach is of high interest. Thus, we introduce the field asymmetric time-of-flight ion mobility spectrometer (FAT-IMS) first described by Allers et al. and Münchmeyer et al.19,20 combining the benefits of FAIMS and

(3)

with the drift time td and the peak full width half-maximum w0.5. In linear systems, the resolution between two peaks is proportional to the resolving power Rp. Thus, two ion species with nearly the same mobility might entail overlapping peaks in an IMS with low resolving power. Eq 2 predicts a strict linear relationship of drift velocity vd and field Ed. However, when high ratios of electrical field E to neutral density N, where E/N is given in Townsend (Td, 10−21 Vm2), are applied, the mobility tends to change.9 High field effect can be expected in the range of 10−120 Td. Thus, the low field mobility K0 needs to be expended by a field dependent term. The field dependent mobility K0(E) can by described by eq 4 with the alpha function α(E). Thus, ions with an alpha function of zero will adhere to eq 2 for all electrical fields. However, when compared to the low field mobility K0, a positive alpha function will induce a higher mobility when high electrical fields are applied. Consequently, a negative alpha function causes a lower mobility in high electrical fields. The mobility difference occurring between low and high field conditions is known as the differential mobility. K 0(E) = K 0·(1 + α(E))

(4)

The drift field utilized in a common DT-IMS is in the range of a few Townsend − the low field regime. Since the alpha function α(E) does not correlate with K0,7 the differential mobility can be used to gather additional mobility data. A device utilizing this technique is the field asymmetric ion mobility spectrometer (FAIMS). There are several types of FAIMS mentioned in the literature.10 A schema of a planar FAIMS is illustrated in Figure 1 (right). This device uses a continuous flow of ions carried by a gas flow passing two closely spaced electrodes.4 Since ions of both polarities can pass the device, two biased detectors, one for each ion polarity, are used. While the drift field utilized in a DT-IMS is a constant electrical field, the ions in a FAIMS are exposed alternately to a low and a high electrical field perpendicular to the gas flow − the dispersion field. The dispersion field applied in a FAIMS can be described by an asymmetric function. As the function has no DC component, there should be no net ion movement (Figure 1, diamonds). However, an alpha function not equal to B

DOI: 10.1021/acs.analchem.7b03200 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Figure 2. Schema of a FAT-IMS. Three different ion species with identical K0 but positive (circles), negative (rectangular), and zero alpha coefficients (diamond) are shown. The radial motion is for illustrative reasons only.

IMS to separate ions by their low field and differential mobility without reducing the repetition rate.

turning off Udv the device acts as a normal DT-IMS, hereinafter referred to as the DT mode. Similar to a common DT-IMS, the ions in a FAT-IMS are injected into the drift cell and start to move due to the drift voltage Ud (Figure 2, dashed ion trajectories) creating the drift field. When the ions reach a certain point in the FAIMS region, the alternating dispersion voltage Udv is switched on, forcing the ions into a vibrating motion (Figure 2, solid ion trajectories). In the case of positively charged ions, with the conditions shown in Figure 2, the ions will be attracted to the dispersion electrode in the high field regime and repelled in the low field regime. Thus, the axial drift motion is overlapped by an oscillation entailing a small spatial shift. When the ions reach the dispersion electrode, Udv is switched off, canceling the oscillations so the ions continue to drift toward the detector. This is hereinafter referred to as the FAT mode. Since the waveform of the dispersion field has no DC component, ions with an alpha function equal to zero (Figure 2, diamonds) will not execute a net shift toward the dispersion or one of the guard electrodes. Therefore, these ions will reach the detector as predicted by eq 1, only separated by their low field mobility K0. However, ions with a positive alpha function (Figure 2, circles) will travel further in the high field period than in the low field period − they will execute a net shift toward the dispersion electrode. This will affect the obtained spectrum, since a net shift to the dispersion electrode will decrease the effective drift length and therefore shifting the peak of the corresponding ion to lower drift times. The same argumentation is valid for ions with a negative alpha function (Figure 2, circles), but their corresponding peaks will shift to higher drift times. This way, previously overlapping peaks in the drift time spectrum can be separated. When investigating the mobility spectrum, substances with a positive alpha function will shift to higher mobilities, and substances with a negative alpha function will shift to lower mobilities. Utilizing the FAT mode as described so far to separate two overlapping peaks might lead to ambiguous results when the peaks are located next to each other but do not completely overlap, as illustrated in Figure 3. The spectrum captured in the DT mode consists of two undistinguishable peaks (gray and red) and the sum signal (black line). The first peak (gray line) has an alpha function of zero, while the second peak (red line) has a positive alpha function. Thus, using the FAT mode shifts



FIELD ASYMMETRIC TIME OF FLIGHT IMS (FAT-IMS) In FAIMS, a gas flow transports a continuous ion current among two electrodes, while the dispersion voltage, applied to the electrodes, deflects the ions depending on their alpha function. If the gas flow is stopped, there is no axial ion motion. Thus, the dispersion voltage will yield a shift to one of the electrodes, depending on the magnitude and sign of their alpha function. Deactivating the dispersion voltage immediately before one of the ion species reaching the electrodes could thus result in a spectrum of spatial spaced ions separated by their differential mobility. However, the analytical power of a device based on this principle is limited by two parameters. First, a reliable separation requires a large gap width leading to unwieldy high dispersion voltages. Second, a sectored detector with a high spatial resolution which is able to resolve the closely spaced ion species is currently not feasible as described by Shvartsburg.7 Nevertheless, there is a well-known approach for converting closely spatial separated ion portions into a measurable signal. By applying an electrical field and capturing the ions with a Faraday detector, a time-resolved ion current can be measured. Thus, a small spatial shift is converted to a shift Δtd in the time domain with electrical field strength and ion mobility as conversion factors. A device already utilizing this type of measurement is the DT-IMS. Thus, by using a DT-IMS and inserting a FAIMS region which shifts the ions in axial direction, we would obtain a device that is able to separate ions both by their low field mobility in the drift region and by their differential mobility in the FAIMS region. A possible integration of a FAIMS region into a DT-IMS is illustrated in Figure 2. To insert the dispersion electrode of a planar FAIMS into the drift cell, it is rotated by 90° and replaced by a metal grid. Two additional grids located next to the dispersion electrode act as guard electrodes ensuring a homogeneous dispersion field. Furthermore, they avoid interferences between the alternating field and the constant drift field outside the FAIMS region. Guard electrodes and power supply of the dispersion voltage are referenced to the potential of drift electrodes relating to their positions. Thus, by C

DOI: 10.1021/acs.analchem.7b03200 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

choosing an initial ion position close to one of the electrodes. As a measure for the peak shift gained in the FAT mode we use the resolution RDT‑FAT between the peaks obtained in the DT mode and the shifted peaks in the FAT mode with the peak position in DT mode td,DT and in the FAT mode td,FAT and the width at half-maximum of the shifted peak w0.5,FAT. Since the peak width in IMS is not measured at the baseline, the conversion factor 1.18 is necessary to retain the known definition of the resolution. RDT ‐ FAT = 1.18·

td,DT − td,FAT 2·w0.5,FAT

(5)

A resolution of 1 is necessary for a baseline separation, but with a resolution of 0.5 two peaks with equal amplitudes are already clearly distinguishable. The sign of RDT‑FAT is in this case a measure for the sign of the alpha function. A positive resolution yields an alpha function greater and a negative resolution yields an alpha function smaller than the alpha function of the fixed ion species. Unless otherwise stated, all measurements are obtained with a fixed reactant ion peak. The relationship between the resolution R of two peaks (between DT and FAT modes) and the resolving power Rp is described by eq 6. Thus, the resolving power Rp,min necessary for a baseline separation of two peaks with given peak positions can be calculated (eq 6) by inserting a resolution R of 1.

Figure 3. Schematic example spectrum of a substance with an alpha function equal to zero (dotted, gray) and a negative alpha function (dashed, red) and the sum of both signals (solid, black) in the DT mode (top) and the FAT mode in the first (middle) and second FAIMS region (bottom).

R p,min = R ·

td,DT + td,FAT |td,DT − td,FAT|

(6)



the second peak to lower drift times (Figure 3, middle), and the sum signal still looks like a single peak shifted slightly to the left. Hence, a method for inverting the direction of the shift is needed, amplifying the offset and accomplishing an even greater separation. Inverting the shift could be done by a simple inversion of the voltages used to generate the dispersion field. However, this would also entail a greater instrumental effort and a reduction of the signal-to-noise ratio, since no spectra can be captured during the polarity switch. Therefore, a second FAIMS region, as illustrated in Figure 2, between the dispersion and the second guard electrode is introduced. The field gradient of the dispersion field will point into the other direction, while the strength is the same. Thus, the magnitude of the shift is equal in both FAIMS regions, but the direction is inverted. Since the initial position of the ions in the first or second FAIMS region is only determined by the delay between the injection and the leading edge of the dispersion voltage, a measurement of both modes, FAT and inverted FAT mode, can be done within two spectra in a row. Using this feature, the second peak shown in Figure 3 can be shifted to higher drift times, resulting in a significantly better separation (Figure 3, bottom). To carry out a measurement with the FAT-IMS, the DC component of the dispersion waveform is adjusted to compensate the motion of one peak in the spectrum, thus, acting as the compensation voltage known from common FAIMS devices. However, in contrast to a common FAIMS, the compensation voltage of the FAT-IMS is constant during a measurement and depends only on the fixed ion species. As the shift of the other ions species is now relative to the fixed species, fixing the species with the highest or lowest alpha function is practical. Doing so will lead to a predicable shift direction enabling an optimization of the available space by

EXPERIMENTAL SECTION To implement the setup described in the theoretical section, a drift tube previously presented by Kirk et al.12 was modified and used. The experimental parameters are summarized in Table 1. Ions are generated by a radioactive tritium ion source and injected by a field switching shutter.21 The drift tube is operated under ambient pressure and has a total length of 44 mm including the FAIMS regions. The resolving power of the Table 1. Operational Parameters of the FAT-IMS parameter drift length source diameter source activity drift field injection field resolving power averaging time dispersion field (high/low) FAIMS repetition rate AC duty cycle IMS repetition rate drift gas flow sample gas flow dew point drift/sample drift/sample humidity operating pressure operating temperature membrane material membrane thickness membrane temperature D

value 44 10 300 2.55 10 70 1 70−100/2.50 10 1−5 22 216 66 −62.5 5 1018 21 PDMS 10 60

mm mm MBq Td Td s Td kHz % Hz mL·min−1 mL·min−1 °C ppmv mbar °C μm °C

DOI: 10.1021/acs.analchem.7b03200 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

FAIMS cycles nFAIMS, it is possible to increase the shift to a maximum of K0,FAT = 1.4 cm2/(V s) (dotted line) which equals a resolution of −0.61. The increasing amplitude of the dimer is a result of the additional reaction time in the FAIMS regions. It should be noted that the negative sign of RDT‑FAT is a consequence of the negative alpha function while the magnitude is a measure for the magnitude of the alpha function related to the fixed ion species. The impact of the FAIMS, thus the resolution, depends primarily on the magnitude of the electrical field and the number of FAIMS cycles nFAIMS. Increasing nFAIMS will increase the resolution in a linear fashion (Figure 5). However, the highest achievable shift

IMS is 70 with a drift voltage of 2800 V. The ion current is amplified by a custom transimpedance amplifier with a gain of 5.3 GOhm and a bandwidth of 30 kHz.22 If not stated otherwise, all shown spectra are captured with an averaging time of 30 s and interpolated by a factor of 10 using a sincinterpolation without smoothing. The FAIMS region implemented into the IMS consists of a stack of three stainless steel grids and two 2 mm PTFE spacers. The dispersion field is generated by applying a voltage of up to 5 kV to the middle grid entailing a dispersion field of up to 100 Td under normal conditions. The dispersion field in the low field period is 2.5 Td for all presented measurements. Dispersion and drift voltages are supplied by custom power supplies. The rectangular FAIMS waveform is generated by a solid-state switch, switching the dispersion voltage with a frequency of 10 kHz. The duty cycle is in the range of 1% to 5%, depending on the FAT mode and magnitude of the dispersion voltage. As drift and sample gas, purified air with a humidity of 5 ppmv is used. The sample injection is accomplished by utilizing a membrane inlet attached to the IMS. The used membrane is a heated 10 μm polydimethylsiloxane (PDMS) foil with a working temperature of 60 °C. All chemicals were purchased from Sigma-Aldrich Germany with a purity of >99%.



RESULTS AND DISCUSSION To carry out a measurement with the FAT-IMS, the parameters are adjusted to ensure that no ions enter or leave the FAIMS regions or hit the grids while the dispersion fields are applied. Thus, all ions are exposed to the same number of FAIMS cycles. Furthermore, the DC component of the dispersion field is regulated to maintain the position of the positive reactant ion peak in both spectra (DT and FAT modes). Figure 4 shows a mobility spectrum in the DT mode with the dimer of dimethyl methylphosphonate (DMMP, solid line) at a reduced mobility of 1.37 cm2/(V s). Utilizing the FAT mode, the dimer peak shifts to a higher reduced mobility (dashed line) − the modified mobility K0,FAT. By raising the number of

Figure 5. Resolution RToF‑FAT of the DMMP monomer (circles), dimer (triangle), and a linear fit (dashed line) assigned to nFAIMS with EFAIMS,h = 95 Td, EFAIMS,l = 2.5 Td, f FAIMS = 10 kHz, and a duty cycle of 1.85%.

is, in the first order, limited by the gap width, since the ions might annihilate at the electrodes when the spatial shift is too high. The highest possible resolution RDT‑FAT, for ions with K0 = 2.3 cm2/(V s) is RDT‑FAT = 1.5 in the presented setup. To prove the presence of the alpha behavior, a spectrum of the positive RIP+ and the acetone monomer in the DT mode and in the normal and inverted FAT modes is shown in Figure 6. By switching between the normal and the inverted FAT modes, we are able to shift the monomer in different directions, while the RIP+ is almost stable. Such a behavior is only possible if the ions are not only separated by their constant mobility in static electrical fields. As a real-world application of the FAT-IMS, the masking usually occurring when analytes with similar mobilities are measured is investigated. In Figure 7, a spectrum of the DT mode (dashed line) and the FAT mode (solid line) of 0.5 ppmv phosgene with a reduced mobility of 2.33 cm2/(V s) and 0.5 ppmv 1,1,2-trichlorethane with a reduced mobility of 2.31 cm2/ (V s) is shown. The DT mode reveals only two peaks: The negative RIP− with a reduced mobility of 2.16 cm2/(V s) and an analyte peak with a maximum at 2.31 cm2/(V s). By taking a closer look at the analyte peak, a small shoulder, which gives a hint at the presence of a second substance, is noticeable, but neither the position nor the peak amplitude of the second substance can be estimated. Furthermore, such a shoulder might also result from a small contamination or ion

Figure 4. Reduced mobility of the DMMP dimer in the DT mode (solid line) and the modified mobility in the inverted FAT mode for nFAIMS = 8 (dashed line), nFAIMS = 23 (dotted line), EFAIMS,h = 95 Td, EFAIMS,l = 2.5 Td, f FAIMS = 10 kHz, and a duty cycle of 1.85%. E

DOI: 10.1021/acs.analchem.7b03200 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

highest ever reported resolving power of 250 for a DT-IMS.8 However, using the FAT mode phosgene is shifted to higher drift times, respectively a lower modified mobility K0,FAT (K0,FAT = 2.29 cm2/(V s)), and 1,1,1-trichlorethane to lower drift times, respectively a higher modified mobility K0,FAT (K0, FAT = 2.40 cm2/(V s)), while the RIP− is stable due to the compensation. Thus, the resolving power Rp,min necessary to distinguish phosgene and 1,1,1-trichlorethane drops from 302 to 44 with the additional resolution RDT‑FAT accomplished by utilizing the FAT mode. The integral of the RIP− decreases from −11.8 fC for the DT mode to −6 fC for the FAT mode, which is a reduction of nearly 50%. In contrast, the integral of the analyte peaks decreases from −10.3 fC to −9 fC, which is a reduction of only 12.8%. The significant reduction of the RIP− is a result of the parameters optimized to preserve the analyte ions and additional reactions occurring in the FAIMS region. However, analyte peaks are only reduced by 12.8%, thus, the limits of detection in the FAT mode are comparable to the value obtained in the DT mode. Resolution and Limits of Detection. To investigate the opportunity of distinguishing substances by their resolution RDT‑FAT between the DT and FAT modes, a scope of different substances needs to be investigated. In Table 2, the resolution RDT‑FAT of acetonitrile, DMMP, 2-butoxyethanol (EGMBE), and diisopropyl methyl phosphonate (DIMP) is shown. All data are obtained within the first FAIMS region for nFAIMS = 25, EFAIMS,h = 95 Td, EFAIMS,l = 2.5 Td, f FAIMS = 10 kHz, and positive IMS polarity. The limits of detection (LOD) are measured in the DT mode with an averaging time of 1 s. Since the impact of the FAT mode strongly depends on the alpha function, magnitude and sign of RDT‑FAT are a direct measure for magnitude and sign of the alpha function at the given electrical field. The acetonitrile dimer entails the lowest measured resolution of RDT‑FAT = −0.10, while the highest resolution RDT‑FAT = −0.71 was observed at the DMMP dimer. Thus, especially for the dimers a baseline separation might be possible by utilizing the FAT mode. Furthermore, the phosphonates (DMMP and DIMP) reveal a similar behavior for both the mono- and dimer. Since DMMP is a precursor for the chemical warfare agent sarin and sarin is a phosphonate, too, it might be possible to predict the behavior of sarin from the obtained data. The resolution RDT‑FAT for EGMBE on the other side - a potential interference for sarin and common substance - differs, especially in the case of the monomer, from DMMP and DIMP. Thus, distinguishing sarin and EGMBE by their resolution RDT‑FAT might be a possible application for the FAT-IMS to reduce the rate of false positives when detecting chemical warfare agents. To evaluate the negative IMS mode, carbon dioxide (CO2), nitrogen dioxide (NO2), sulfur dioxide (SO2), chlorine (Cl), hydrochloric acid (HCl), cyanogen chloride (NCl), hydrogen cyanide (HCN), 1,1,1-trichlorethane (1,1,1-TCA), 1,1,2-

Figure 6. Acetone monomer in the normal and inverted FAT modes with nFAIMS = 15, EFAIMS,h = 90 Td, EFAIMS,l = 2.6 Td, f FAIMS = 10 kHz, and a duty cycle of 2.66%.

Figure 7. 0.5 ppmv phosgene and 0.5 ppmv 1,1,2-trichlorethane in the DT mode (solid line) and in the FAT mode (dashed line) with nFAIMS = 25, EFAIMS,h = 100 Td, EFAIMS,l = 2.54 Td, f FAIMS = 10 kHz, and a duty cycle of 2.26%.

decomposition in the presence of a single ion species. Relating to eq 5, a resolving power Rp,min of at least 302 is necessary to separate phosgene and 1,1,2-trichlorethane, which is above the

Table 2. Resolution RDT‑FAT and Limits of Detection of Some Positive Ion Species monomer 2

RIP+ acetonitrile DMMP EGMBE DIMP

K0 in cm /(V s)

RDT‑FAT

1.98 1.88 1.74 1.58 1.47

−0.10 −0.49 −0.30 −0.53

dimer LOD 0.19 0.17 1.70 4.80

ppbv ppbv ppmv ppbv

F

2

K0 in cm /(V s)

RDT‑FAT

1.79 1.37 1.25 1.07

−0.13 −0.71 −0.62 −0.62

LOD 1.10 ppbv 0.42 ppbv 3.70 ppmv 14 ppbv

DOI: 10.1021/acs.analchem.7b03200 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

a higher gap width might be able to separate all substances investigated in this work. Furthermore, by utilizing the data listed in Table 3, it is possible to distinguish substances by RDT‑FAT. An example of a potential issue for a conventional DT-IMS is the spectrum shown in Figure 9. Again, only two separated peaks are revealed

trichlorethane (1,1,2-TCA), and phosgene (CG) were measured. The reduced mobility and resolution RDT‑FAT are summarized in Table 3. All data were captured with nFAIMS = 25, EFAIMS,h = 95 Td, EFAIMS,l = 2.5 Td, and f FAIMS = 10 kHz in the second FAIMS region and an averaging time of 1 s. Table 3. Resolution RDT‑FAT and Limits of Detection of Some Negative Ion Species monomer 2

HCN phosgene 1,1,2-TCA NCl 1,1,1-TCA NO2 Cl HCl SO2 RIP− CO2

K0 in cm /(V s)

RDT‑FAT

2.38 2.36 2.31 2.31 2.31 2.25 2.24 2.15 2.15 2.13 2.15

0.25 0.76 −0.38 0.78 0.26 0.50 −0.45 0.83 1.19 1.08

LOD 0.16 ppbv 0.27 ppbv 0.22 ppbv 37 ppmv 0.41 ppmv 36 ppmv 0.55 ppbv 5.60 ppmv 0.11 ppbv −

To separate all substances listed in Table 3, an extremely high resolving power is needed. Especially in the case of HCN, CG, 1,1,2-TCA, NCl, 1,1,1-TCA, and HCl, a resolving power Rp,min of at least 420 would be necessary. However, by utilizing the FAT mode the shift can be used for a better separation. Thus, the necessary resolving power Rp,min can be minimized, and even NCl and 1,1,1-TCA can be separated to a limited extent, despite possessing virtually identical mobilities. Figure 8

Figure 9. 10 ppmv nitrogen dioxide and 3 ppmv chlorine in the DT mode (solid line) and in the FAT mode (dashed line) with nFAIMS = 25, EFAIMS,h = 100 Td, EFAIMS,l = 2.54 Td, f FAIMS = 10 kHz, and a duty cycle of 2.31%.

by utilizing the DT mode (solid line). The reactant ion peak RIP− with a reduced mobility K0 of 2.14 cm2/(V s) and a second peak at K0 = 2.24 cm2/(V s). Both nitrogen dioxide and chlorine could match to this mobility. Since Rp,min = 588 is needed to resolve chlorine and nitrogen dioxide, even the DTIMS with the highest resolving power reported so far would be insufficient for this application.12 Utilizing the FAT mode (dashed line) in the first FAIMS region with nFAIMS = 25, EFAIMS,h = 100 Td, EFAIMS,l = 2.54 Td, f FAIMS = 10 kHz, and adjusting the duty cycle to 2.31% - to fix the analyte peak - the RIP− is shifted to a higher modified mobility (K0,FAT = 2.17 cm2/(V s)). Furthermore, a second peak appears on the right site of the analyte peak, shifting to a higher modified mobility, too. Since the nitrogen dioxide (RDT‑FAT = 0.50, s. Table 3) has a positive alpha function and chlorine (RDT‑FAT = −0.45) has a negative alpha function, fixing the nitrogen dioxide would have shifted the chlorine to lower mobilities. Therefore, the shifted peak (K0,FAT = 2.31 cm2/(V s)) must be the nitrogen dioxide, and the fixed peak (K0,FAT = 2.24 cm2/(V s)) is the chlorine. Thus, a separation and identification of previously completely overlapping peaks is possible by using the FAT-IMS.

Figure 8. Comparison of the investigated substances in negative IMS polarity. The light gray areas indicate substances separable utilizing a DT-IMS with a resolving power of 70. The green areas indicate substances which are not separable in a common DT-IMS but in a FAT-IMS. The red hatched areas indicate substances not even separable by utilizing the FAT-IMS.



shows a summary of the substances measured in this work. The light gray areas are substances separable employing only a common DT-IMS with a resolving power of 70. However, the 13 colored (green and red) substances are not separable by the DT mode; but with the added resolution gained by the FAT mode, it is possible to achieve a baseline separation of the substance highlighted by the green areas. Thus, the FAT-IMS is able to separate 9 of the 13 substances investigated in this work. Since, the extra resolution generated by the FAT mode is, in the first order, only limited by the gap width, a new device with

CONCLUSION In this work, a functional FAT-IMS utilizing the separation techniques of both a DT-IMS and a FAIMS was presented for the first time. Employing the FAT-IMS, substances usually not separable in a conventional IMS can be distinguished with a low additional instrumental effort and without significantly increasing the measurement time. Thus, the presented technique is an ideal expansion for hand-held systems with G

DOI: 10.1021/acs.analchem.7b03200 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

(15) Anderson, A. G.; Markoski, K. A.; Shi, Q.; Coy, S. L.; Krylov, E. V.; Nazarov, E. G. DMS-IMS2, GC-DMS, DMS-MS: DMS Hybrid Devices Combining Orthogonal Principles of Separation for Challenging Applications. In SPIE Defense and Security Symposium; Fountain, A. W., III, Gardner, P. J., Eds.; SPIE: 2008; DOI: 10.1117/ 12.782429. (16) Nazarov, E. G.; Anderson, A. G.; Krylov, E. V.; Coy, S. L.; Miller, R. A.; Burchfield, D.; Eiceman, G. A. Miniature DMS-IMS detector for enhanced revolving power. In Proceedings of 16th International Conference on Ion Mobility Spectrometry: Mikkeli, Finland, 22−25 July 2007. (17) Langejuergen, J.; Allers, M.; Oermann, J.; Kirk, A. T.; Zimmermann, S. Anal. Chem. 2014, 86, 7023−7032. (18) Langejuergen, J.; Allers, M.; Oermann, J.; Kirk, A. T.; Zimmermann, S. Anal. Chem. 2014, 86, 11841−11846. (19) Allers, M.; Bohnhorst, A.; Kirk, A. T.; Ungethüm, B.; Walte, A.; Zimmermann, S. Int. J. Ion Mobility Spectrom. 2015, 18, 107−115. (20) Münchmeyer, W.; Ungethüm, B.; Walte, A. Method and device for detection and identification of gases. (21) Kirk, A. T.; Zimmermann, S. Int. J. Ion Mobility Spectrom. 2014, 17, 131−137. (22) Cochems, P.; Kirk, A. T.; Zimmermann, S. Rev. Sci. Instrum. 2014, 85, 124703.

small IMS. Furthermore, even for IMS with an extremely high resolving power, the orthogonal data provided by the FAT mode can be used to reduce the rate of false positives while still providing a low response time, which is, especially for systems used in medical or security applications, crucial. In addition to this, even a measurement of the alpha coefficients might be possible. Besides the improved analytical power, an investigation of the fragmentation in high fields might be possible by using the dispersion field to heat the ions, accomplishing fragmentation and then utilizing an IMS with high resolving power to separate and investigate the resulting fragments.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Alexander Bohnhorst: 0000-0002-9710-3254 Ansgar T. Kirk: 0000-0001-7152-3077 Author Contributions

A.B. designed the theoretical model and measurement setup. A.B. and M.B. performed most of the experiments and the data analysis. S.Z. and A.T.K. gave scientific and conceptual advice. S.Z. supervised the research project. All authors contributed to discussions and the manuscript. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS We would like to thank Airsense Analytics for providing the laboratory for the measurements. This work has been supported by the German Federal Ministry of Economics and Technology (BMWi), under the Grant KF3238301NT3 upon decision of the German Bundestag.



REFERENCES

(1) Eiceman, G. A.; Karpas, Z. Ion mobility spectrometry; CRC Press: Boca Raton, FL, 1994. (2) Puton, J.; Namieśnik, J. TrAC, Trends Anal. Chem. 2016, 85, 10− 20. (3) Ungethüm, B.; Walte, A.; Münchmeyer, W.; Matz, G. Int. J. Ion Mobility Spectrom. 2009, 12, 131−137. (4) Buryakov, I. A.; Krylov, E. V.; Nazarov, E. G.; Rasulev, U. K. Int. J. Mass Spectrom. Ion Processes 1993, 128, 143−148. (5) Mäkinen, M.; Nousiainen, M.; Sillanpäa,̈ M. Mass Spectrom. Rev. 2011, 30, 940−973. (6) Fink, T.; Baumbach, J. I.; Kreuer, S. J. Breath Res. 2014, 8, 027104. (7) Shvartsburg, A. A. Differential mobility spectrometry; CRC; Taylor & Francis [distributor]: Boca Raton, FL, London, 2008. (8) Kirk, A. T.; Allers, M.; Cochems, P.; Langejuergen, J.; Zimmermann, S. Analyst 2013, 138, 5200−5207. (9) Krylov, E. V.; Nazarov, E. G. Int. J. Mass Spectrom. 2009, 285, 149−156. (10) Guevremont, R. J. Chrom. A 2004, 1058, 3−19. (11) Kirk, A. T.; Raddatz, C.-R.; Zimmermann, S. Anal. Chem. 2017, 89, 1509. (12) Kirk, A. T.; Zimmermann, S. Int. J. Ion Mobility Spectrom. 2015, 18, 17−22. (13) Shvartsburg, A. A.; Seim, T. A.; Danielson, W. F.; Norheim, R.; Moore, R. J.; Anderson, G. A.; Smith, R. D. J. Am. Soc. Mass Spectrom. 2013, 24, 109−114. (14) Cohen, M. J.; Karasek, F. W. J. Chromatogr. Sci. 1970, 8, 330− 337. H

DOI: 10.1021/acs.analchem.7b03200 Anal. Chem. XXXX, XXX, XXX−XXX