Article pubs.acs.org/ac
High Kinetic Energy Ion Mobility Spectrometer: Quantitative Analysis of Gas Mixtures with Ion Mobility Spectrometry Jens Langejuergen,* Maria Allers, Jens Oermann, Ansgar Kirk, and Stefan Zimmermann Leibniz University Hannover, Institute of Electrical Engineering and Measurement Technology, Appelstrasse 9a, 30167 Hannover, Germany ABSTRACT: We present a high kinetic energy ion mobility spectrometer (HiKE-IMS) for quantitative gas analysis. Drift tube and reaction tube can be operated at reduced fields up to 110 Td. At such conditions the distribution of reactant ion water clusters is shifted toward smaller clusters. Due to the resulting presence of bare reactant ions (e.g., H3O+) and the kinetic control of the ionization process with decreasing reaction time, unlike conventional IMS, a quantitative detection with ppbv detection limits of low proton affine analytes even in humid gas mixtures containing high proton affine compounds is possible using a direct sample gas inlet. A significantly improved dynamic range compared to conventional IMS is achieved. An incremental change in reduced fields enables the observation of parameters like field dependent ion mobilites or analyte fragmentation. Also, the characteristic of the analyte signal with respect to the reduced reaction field gives insight into the ionization process of the analyte. Thus, HiKE-IMS enables substance identification by ion mobility and additional analytical information that are not observed with conventional IMS. The instrumental effort is similar to conventional desktop IMS with overall dimensions of the drift and reaction tube of 4 cm × 4 cm × 28.5 cm. However, the mobility resolution is limited and between 30 and 40. Because of the moisture independent ionization and the decrease in competing ion−molecule reactions, no preseparation or membrane inlet is necessary when the compounds of interest are distinguishable either by a significant difference in ion mobility or the additional analytical information.
I
time-consuming gas chromatographic or other preseparation technique. This is mainly caused by two things: First, the competing ion−molecule reaction sequences within the ionization region are leading to a discrimination of compounds in the presence of other more reactive compounds (e.g., with higher proton affinity). The same applies to mixtures with very different analyte concentrations. Second, for many compounds the sensitivity of an IMS is vastly decreasing with increasing water content of the introduced sample gas. While the moisture dependence of IMS signals can be partly circumvented using heated water impermeable membrane inlets for sample introduction13 the issues related to compound discrimination when analyzing gas mixtures are still not solved. In this work, we present a method to decrease both, the moisture dependence and the extent of competing ion− molecule reactions, featuring fast and quantitative detection of compounds even in gas mixtures with ppbv-detection limits independent of the sample gas humidity. This method does not require any kind of enrichment, preseparation or membrane inlet. A detection even of low proton affine compounds is possible by increasing the kinetic ion energy, which leads to a
on mobility spectrometry (IMS) is a well-known technology based on the ionization and separation of analytes mostly at atmospheric pressure. Various applications are possible, such as IMS-based preseparation in mass spectrometry with the ability to separate isomeric compounds,1 IMS detectors in gas chromatography, for example, for breath analysis,2 and handheld IMS for fast detection of toxic industrial compounds, chemical warfare agents,3 and explosives.4 Multiple IMS designs and ionization methods are used depending on the specific application. Examples are high-resolution drift tube IMS,5 field asymmetric IMS (FAIMS),6 and aspirator-type IMS (A-IMS).7 Commonly used ionization methods are atmospheric pressure chemical ionization by weak radioactive 63Ni, 3H, and 241Am8 or nonradioactive corona sources,9 electrospray ionization (ESI),10 and direct photoionization of the analytes.12,11 When using IMS with atmospheric pressure chemical ionization (APCI) for the detection of volatile organic compounds (VOCs) extremely low detection limits in the range of pptv to ppbv (parts per trillion to parts per billion) can be achieved. This is due to the chemical gas phase reactions in which analyte ions are efficiently generated by collisions of neutral analytes with reactant ions. Even though this renders APCI-IMS a very sensitive, versatile, and compact method, for most compounds, quantitative analysis of gas mixtures is not possible without a © XXXX American Chemical Society
Received: March 31, 2014 Accepted: June 17, 2014
A
dx.doi.org/10.1021/ac5011662 | Anal. Chem. XXXX, XXX, XXX−XXX
Analytical Chemistry
Article
decrease of the average size of water clusters.14 Furthermore, the decrease in operating pressure in combination with reaction times in the order of 100 μs to 1 ms are causing a decrease in cross sensitivity because the ions are injected into the drift tube before the thermal equilibrium establishes. The kinetic control of the ionization process and the dissociation of larger water clusters by high electric fields are known from SIFT (selected ion flow tube) and PTR (proton transfer reaction) mass spectrometry and are well described by Spanel 15 and Lindinger.16 The reaction rate constants which describe the formation of water clusters and the ionization process by different types of reactant ions in such reaction tubes have been published among others by Good,17 Kebarle,18 and Zhao.19 However, high kinetic energy chemical gas phase ionization with the presented benefits coupled to a stand-alone ion mobility spectrometer has not been described before. Furthermore, the arising effects related to a gradual increase in kinetic ion energy in both the reaction tube and the drift tube can be used to gather additional information about the different components of the analyzed gas mixture besides the ion mobility which improves compound identification.
time dependent analyte ion concentration [A+] can be calculated by the reaction rate constant, the initial reactant ion concentration [R+0 ] and the analyte concentration: [A+] = [R +0 ](1 − e−k[A]t )
The analyte concentration is assumed to be constant since it is significantly higher than the initial reactant ion concentration. The relative analyte ion abundance [A+]/[R+0 ] is dependent on the reaction time, the reaction rate and [A], which again is pressure dependent. At atmospheric pressure and reaction times in the order of several 10 ms the relative ion abundances of different analytes present in ppbv concentrations are in thermal equilibrium. Thus, when multiple analytes are present due to competing ion−molecule reactions the relative analyte ion abundances do not reflect the analyte molecule concentrations and a simultaneous quantitative measurement is not possible. In HiKE-IMS the pressure is decreased to 20 mbar and the reaction times are in the order of several 100 μs depending on the reaction field. Thus, a kinetic control of the competing ion−molecule reactions is possible and the cross sensitivity is significantly reduced. Even though this results in a decrease in relative analyte ion abundance compared to atmospheric pressure ionization detection limits in the low ppbv range are still possible because of an increased signal-tonoise ratio per time. However, the amount of water present in the sample gas is often significantly higher than the amount of analyte and a number of water molecules can be attached to the reactant ions before a collision with an analyte molecule occurs. The corresponding equilibrium establishes within microseconds.26 The resulting thermal water cluster distribution is a function of the water concentration27 and the kinetic ion energy.28 The sensitivity of a conventional IMS for most VOCs vastly decreases with an increasing amount of water introduced with the sample gas. For an ionization via H+(H2O)n, many compounds cannot be ionized when the proton bound water clusters exceed a certain size. This maximum number is increasing with increasing proton affinity as well as with increasing polarity of the specific compound to be ionized.29 For a rather nonpolar compound with comparably low proton affinity like toluene only the proton transfer with H3O+ and H+(H2O)2 is observed.30 Furthermore, NO+ which is a second possible reactant ion especially for alkanes and is present in higher quantities in particular when corona discharge ionization is used, reacts to HNO2 and H+(H2O)n−1 with increasing humidity.31 However, with increasing kinetic energy the reaction rate for water cluster formation decreases and the effect of collision induced separation of water clusters increases. Thus, even under moist conditions the cluster distribution is shifted toward smaller clusters and a large number of, for example, H3O+ or NO+ can be present when the kinetic energy of the reactant ions is sufficiently high. In this work a constant and relatively high amount of water is added into the reaction tube. Thus, the water concentration within this volume is assumed to be nearly constant and is not significantly increased or decreased by the water content of the introduced sample gas. This implies that at constant temperature the size distribution of water clusters is primarily defined by the kinetic energy gained from the electric field within the reaction tube. Therefore, the number of water molecules attached to the reactant ions and thus the ability to ionize the analyte of interest can be tuned with the applied
■
THEORETICAL BASIS Ion Mobility. In a drift tube IMS, the ions of different analytes are separated according to the analyte-specific drift time tD the ions need to traverse a drift tube in which they collide with neutral molecules of the drift gas. The ion drift velocity is proportional to the electric field by an analytespecific ion mobility K vD = K ·E
(1)
Often the reduced ion mobility K0 is given in an attempt to correct the mobility to standard temperature T0 = 273 K and standard pressure p0 = 1013 mbar. However, since ion temperature can differ from the drift tube temperature and both the ion temperature and the water content of the drift gas affect cluster formation reduced mobilities need to be compared with care. K0 = K
T0p p0 T
(3)
(2)
A fundamental theory of ion mobility on a molecular scale has been introduced by Revercomb and Mason.20 It is noteworthy that the ion mobility of a specific analyte is constant at low field conditions (E < 50 V/mm at atmospheric pressure) and constant ambient conditions but field dependent ion mobilities can be observed and used to separate analytes at high-field conditions.21 Ionization. The most common ionization method used in IMS is atmospheric pressure chemical ionization (APCI). For APCI a primary ionization, for example, corona discharge or beta-radiation from 63Ni or 3H is used to generate reactant ions (e.g., H+(H2O)n) from the main components of air.22 By collision with analyte molecules the reactant ions can form analyte ions either by proton transfer, charge transfer, ligand switching or association reactions16.23 For example a proton transfer from H3O+ to an analyte molecule is possible when the proton affinity of the analyte exceeds the proton affinity of water which is 691 kJ/mol.24 This is true for most volatile organic compounds. The reaction rate constant k for such a reaction is on the order of magnitude of the collision rate constant kc25 leading to extremely low detection limits. The B
dx.doi.org/10.1021/ac5011662 | Anal. Chem. XXXX, XXX, XXX−XXX
Analytical Chemistry
Article
Figure 1. Description of gas flows and main components of the high kinetic energy IMS.
Figure 2. Chart of voltages applied to the high kinetic energy IMS.
electric field within the reaction tube and is independent of the water content of the sample gas. Kinetic Energy. The kinetic energy EKin of an ion with the charge q within an electric field E can be approximately written as E Kin = 3/2·kB·T + q·E ·d
where N is the gas number density (particles/cm3). The unit of the reduced field is Td (Townsend, 1 Td = 10−17 V cm2).
■
EXPERIMENTAL METHODS The experimental setup is shown in Figure 1 and consists of a reaction tube in which the primary and secondary ionization processes occur coupled to a drift tube in which the ion species are separated because of their different ion mobilities. If not otherwise specified, the setup is operated at room temperature (295 K). All applied voltages can be easily inverted to detect negative ions. All necessary dimensions and a definition of the applied voltages are given in Figure 2. Default values are given in Table 1. Reaction tube and drift tube are both operated below atmospheric pressure and are evacuated via a membrane pump (MVP 40, Pfeiffer Vacuum). A gas dosing valve (EVN 116, Pfeiffer Vacuum) is used to adjust the suction rate. The pressure within the reaction tube and drift tube is monitored via a capacitive pressure gauge (Pfeiffer Vacuum, CMR 362). Considering the overall size of the HiKE-IMS smaller pumps, such as Pfeiffer MVP 006, can be used thus the complete setup fits into a 19 in. housing. A corona discharge needle (corona
(4)
where T is the temperature in Kelvin and kB is Boltzmann’s constant. The distance d describes the interval in space between two intermolecular collisions. This assumes the ion to be an ideal gas molecule and the kinetic energy gained due to the acceleration within the electric field to be fully dissipated with every impact. At atmospheric pressure the mean free path between two collisions is in the order of 70 nm.32 When thermal effects are neglected the kinetic energy of an ion can be increased when either the electric field or the distance d is increased. Assuming an ideal gas the mean free path and thus the average kinetic energy gained between two collisions in a constant electric field is inversely proportional to the pressure. Thus, a suitable parameter to describe the kinetic energy according to the applied electric field is the reduced field E/N C
dx.doi.org/10.1021/ac5011662 | Anal. Chem. XXXX, XXX, XXX−XXX
Analytical Chemistry
Article
species have a different dependence of their corresponding ion mobilities on the reduced drift field as known from FAIMS. The voltage between detector and aperture grid is 20 V. A constant drift gas flow of 5 sccm dry clean air is introduced at the end of the drift tube. The experimental setup for sample gas preparation is as follows. Two permeation ovens (Model 150, VICI) are used to provide known and constant concentrations of the chemical compounds to be analyzed. The gas flows from the permeation ovens are further diluted with dry clean air provided by a zero gas generator (ZA350S, JAG). The humidity of the sample gas can be increased either by adding a defined amount of gas through a gas washing bottle filled with water or by using the permeation oven loaded with a water containing tube. The moisture content of the analyzed gas is measured by a dew point meter (LPDT, Xentaur). A list of used chemical compounds is given in Table 2.
Table 1. Default Parameters for Gas Flows, Applied Voltages, and Ion Gating drift gas flow sample gas flow reaction chamber additive water pressure corona voltage (UCorona) reaction tube voltage (URT) gate transmission voltage (UGate) gate closing voltage (UClose) gate opening time drift tube voltage (UDT) aperture voltage (UAperture) repetition rate
5 sccm dry clean air ( 99,9 (Sigma-Aldrich) analytical standard (Fluka) analytical standard (Fluka) analytical standard (Fluka)
References are given when they are different from the references given in the column heading. D
dx.doi.org/10.1021/ac5011662 | Anal. Chem. XXXX, XXX, XXX−XXX
Analytical Chemistry
Article
current increases to over 10 nA, while the noise level remains constant. This is because the peak becomes narrower with shorter drift times while the number of ions within one injected ion plug remains approximately constant. However, the mobility of the reactant ion peak increases significantly by over 10% with increasing drift voltage. This is because the average number of water molecules attached to the reactant ions within the drift tube decreases with an increasing reduced drift field resulting in higher ion mobilities for smaller reactant ion water clusters.27,26 It is noteworthy that the reaction tube is operated at 100 Td leading to a shift in average cluster size in the reaction tube but not in the drift tube when operated at lower reduced fields. As will be shown later, the effect of field dependent ion mobility as known from FAIMS or DMS21 can be used to separate and identify peaks that have similar ion mobilities under low field conditions. In Figure 5, the positive reactant ion peak at different reduced reaction fields is shown. An increase in reduced
Figure 3. Mobility spectrum of the positive reactant ion peak. Dry (5400 ppmv water) and dry air is used as sample gas.
reaction field is leading to a significant increase in ion current at the detector. This is caused by coulomb repulsion and ion discharge at the metallic guard rings, which is, because of the high space charge density, the limiting factor for an increase in ion current within the reaction tube. Ions generated close to the axis of the tube migrate to the walls within a certain time which is dependent on the radial charge distribution. Since this effect dominates the ion loss in the reaction tube further increasing the corona current does not lead to an increased ion current at the detector. However, an increase in reaction voltage lowers the drift time necessary for ions to traverse the reaction tube and is leading to an increased ion current passing the open ion gate and entering the drift tube. Cyclooctane. In Figure 6 the mobility spectra of cyclooctane at different concentrations and 100 Td within the reaction tube are shown. The water concentration of the sample gas is >2700 ppmv, which corresponds to 10% relative humidity at room temperature. The limit of detection (LOD) can be calculated by linearly fitting the maximum of the cyclooctane peaks versus the corresponding concentrations and dividing three times the standard deviation of noise (3σ) by the resulting slope m as in eq 5.
Figure 4. Ion mobility spectra showing the positive reactant ion peak changing its position and height at different reduced drift fields. The change in reduced mobility with increasing reduced drift field is displayed in the inset. The sample gas is dry (5400 ppmv water) clean air as sample gas at different reduced drift fields from 15 to 84 Td. The reduced reaction field is 100 Td. The change in reduced mobility with increasing reduced drift field is displayed in the inset.
toluene (4 ppmv) at 5400 ppmv sample gas moisture at different reduced drift fields are shown. When higher drift voltages are applied the mobility of the RIP increases as noted above while the mobility of toluene slightly decreases. Thus, separation and quantification of toluene at reduced drift fields above 50 Td is favored. The limits of detection at these parameters for toluene in dry (5400 ppmv water) sample gas only differ slightly (moist 7.0 ppbv and dry 5.5 ppbv) hence a reduction of suitable reactant ions by water cluster formation within the reaction tube is sufficiently suppressed even at elevated analyte sample gas humidity. However, when decreasing the gate opening time to 6 μs at 77 Td reduced reaction field and 51 Td reduced drift field, three peaks T1 (2.10 cm2/(V s)), T2 (2.07 cm2/(V s)), and T3 (1.89 cm2/(V s)) become visible that correlate with the toluene concentration. In Figure 9, the corresponding spectra at reduced reaction fields of 77 Td, 82 and 87 Td are shown. We expect these peaks to be caused by the different toluene ionization pathways. We suspect T1 to be caused by the charge transfer with NO+, T2 by a proton transfer with H3O+ and H+(H2O)2 and T3 by NO+ association. The relative signal intensities at different reduced reaction fields of these three peaks which are displayed in the inset of Figure 9 are leading to this conclusion. At reduced reaction fields below 60, Td no significant toluene signal is observed. With increasing kinetic energy the abundance of NO+ and H+(H2O)2 increases and an onset in all toluene signals is observed. Above 80 Td, the T3 signal is decreasing because of decomposition of the associated ion. Above 100 Td the NO+ signal and the T1 signal are almost saturated while T2 is significantly increasing due to a shift in proton bound water cluster distribution toward smaller clusters and an increase in H3O+ and H+(H2O)2 abundance. However,
Figure 7. Relative signal of cyclooctane and NO+ at different reduced reaction fields.
maximum) of 2.4 ppmv cyclooctane and NO+ at different reduced reaction fields is shown. The reduced drift voltage is 50 Td which enables a separation of the NO+ peak from the RIP. Below 60 Td reduced reaction field no significant cyclooctane is measured because of a reduction in NO+ concentration by building larger water clusters as described by Kebarle.31 Above 60 Td an onset in cyclooctane signal and correlating increase in NO+ are visible. Toluene. Like cyclooctane toluene can be ionized via charge transfer with NO+. However, ionization via H3O+ and H+(H2O)2 is also possible and the resulting protonated toluene F
dx.doi.org/10.1021/ac5011662 | Anal. Chem. XXXX, XXX, XXX−XXX
Analytical Chemistry
Article
Figure 9. Spectra of toluene (4 ppmv) in dry (2700 ppmv water) clean air with 350 ppbv acetone is used as sample gas. The reduced drift field is 20 Td. The reduced mobility of the acetone monomer is highly dependent on the reduced drift tube field and is at the given conditions is K0 = 1.91 cm2/(V s). The reduced mobility of the dimer is K0 = 1.79 cm2/(V s). The sensitivity plot for the acetone monomer is displayed in the inset.
especially at higher reduced reaction fields T1 and T2 cannot adequately be separated and an evaluation of the signal must be performed by curve fitting. Thus, the resulting absolute values must be compared with care. Acetone. In contrast to largely nonpolar hydrocarbons, such as toluene or cyclooctane, polar analytes like acetone, ethanol, 1-heptanol, or n-butylamine are ionized even by larger proton bound water clusters. Thus, it is not necessary to generate large amounts of H3O+ to ionize these analytes. Nevertheless, it is known that there is a certain limit to the size of proton bound water clusters by which a certain compound can be ionized; for example, for acetone, this is H+(H20)4.29 However, the average size of water clusters at 10% relative humidity at room temperature is significantly larger.23 This is why the sensitivity for such compounds using conventional IMS with atmospheric pressure chemical ionization and direct inlets significantly differs with the moisture content in the ionization region yet not as severely as observed with nonpolar compounds having low proton affinities. However, with increasing kinetic energy of the reactant ions within the reaction tube the average cluster size can be decreased to such an extent that mostly all present water clusters can potentially ionize acetone. Furthermore, the necessary reduced reaction field is nearly independent of the sample gas moisture since the amount of water vapor that is intentionally introduced into the reaction tube at the corona discharge source is kept constant and dominates the absolute moisture content within the reaction tube. Thus, the introduced HiKE-IMS allows a quantitative measurement of acetone independent off the moisture content of the sample gas even though a direct inlet is used. In Figure 10, the spectrum of a sample gas containing 350 ppbv acetone and 2700 ppmv water is shown. The scale of the ion current is logarithmic to visualize the acetone dimer peak which at these conditions is less pronounced compared to conventional IMS. The mobility of the monomer peak does significantly change with the reduced drift field. Above 30 Td, the monomer peak and the RIP merge and cannot be separated. The dimer peak however is not influenced by the reduced drift field and remains almost constant (K0 = 1.79 cm2/ (V s)). It is known that the ion mobility of the acetone
monomer in contrast to the acetone dimer is strongly dependent on the effective ion temperature as well as on the water concentration within the drift tube due to a change in the number of water molecules attached to the monomer.36 The sensitivity is nearly independent of the water content of the sample gas. The acetone monomer detection limits are calculated to 970 pptv in humid (>2700 ppmv water) and 920 pptv in dry (2700 ppmv water) clean air with ∼350 ppbv acetone is used as sample gas. The reduced drift field is 12 Td. The logarithmic relative acetone monomer signal is shown versus the reaction time, which is calculated by the acetone ion mobility, the reaction field and the length of the reaction tube. The pseudo-first-order reaction rate constant v is determined by linear interpolation. The resulting reaction rate constant k = v/[acetone] is 4.0 × 10−9 cm3/s. G
dx.doi.org/10.1021/ac5011662 | Anal. Chem. XXXX, XXX, XXX−XXX
Analytical Chemistry
Article
considerably higher than that of acetone. Even relatively high concentrations of n-butylamine have no significant impact on the acetone peak or the RIP. Surprisingly, the amine signal is lower than the acetone signal even at higher concentrations. A similar effect is visible when ethanolamine is used instead of nbutylamine (not shown here). However, analyzing the same sample gas mixture with a conventional IMS using atmospheric pressure chemical ionization no reactant ion peak is present in the spectrum and the acetone peak is suppressed significantly.5 Fragmentation. With increasing reduced fields some compounds start fragmentation.37 As known from mass spectrometry, these fragments are specific for a certain compound and can be used for substance identification. Furthermore, the onset of a certain fragmentation reaction can be used to estimate the binding energy of a specific complex. In Figure 13 zoomed spectra of 200 ppbv 1-heptanol with an increasing reduced reaction field are shown. As expected, the
increase. This is because acetone can be ionized by the majority of water clusters present at reduced reaction fields, even at 20 Td. However, when the reaction voltage is increased, the time a specific reactant ion needs to traverse the reaction tube decreases as noted above. The linear correlation as given in Figure 11 is clear evidence that the ion distribution within the reaction tube is not in thermal equilibrium and the measured ion abundances are controlled by the reaction kinetic. The average reaction rate constant for an ionization of acetone with the present reactant ions can be calculated by the pseudo first order rate constant as given in Figure 11 and the acetone concentration. Considering a possible 15% deviation in analyte concentration and fitted slope the calculated reaction rate k = 4.0 × 10−9 cm3/s is in good agreement with values given in the literature.29 Compound Mixtures. Another major problem using atmospheric pressure chemical ionization is not only the formation of large water clusters inhibiting the ionization of specific analytes but also rivaling ionization processes of compounds with highly different concentrations or highly different proton affinities. The impact of both issues and thus the cross sensitivity is tremendously decreased using the presented HiKE-IMS. As shown above, because of a relatively short reaction time, for example, of less than 200 μs at 100 Td the ratio of measured analyte ion currents does not reflect the thermodynamic equilibrium of analyte concentrations at which the reaction products would be shifted toward ion species with higher proton affinities. The impact of analyte concentrations below 1 ppmv on the reactant ion peak intensity can be neglected. Only a minor saturation of the analyte signal is visible below 1 ppmv for all analyzed compounds leading to a much wider dynamic range compared to conventional atmospheric pressure IMS. In Figure 12, the spectra of sample gas mixtures containing 150 pbbv acetone, 2700 ppmv water, and different concentrations of n-butylamine from 30 ppbv to 250 ppbv are shown. As listed in Table 2 the proton affinity of n-butylamine is
Figure 13. Spectra of 200 ppbv 1-heptanol in dry (
