Anal. Chem. 1996, 68, 2097-2101
Analysis of Polar Organic Compounds Using Charge Exchange Ionization and Membrane Introduction Mass Spectrometry M. E. Cisper,† A. W. Garrett,† D. Cameron,‡ and P. H. Hemberger*,†
Chemical Sciences and Technology Division, MS J565, Los Alamos National Laboratory, Los Alamos, New Mexico 87545 and Department of Chemistry and Geochemistry, Montana Tech of the University of Montana, Butte, Montana 59701
Charge exchange ionization in conjunction with membrane introduction mass spectrometry provides a sensitive method for the detection of polar volatile organic compounds and semivolatile compounds in air. Sample introduction into an ion trap mass spectrometer was accomplished with a hollow fiber silicone membrane assembly. Atmospheric oxygen, which diffuses through the membrane, was used as the charge exchange reagent. Chemical ionization parameters were optimized using methyl ethyl ketone (2-butanone) standards in air. Several other oxygen-containing compounds, including acetone (2-propanone), methyl isobutyl ketone (4-methyl2-pentanone), propanal, isopropyl alcohol (2-propanol), cyclohexanol, dimethyl sulfoxide (sulfinylbismethane), 2-(diethylamino)ethanol, and dimethyl methylphosphonate were analyzed with this technique. This method was used to obtain mass spectra for a variety of classes of compounds and produced a 4-20-fold improvement in response for all of the polar compounds we examined when compared to signal obtained from electron ionization. The simplicity, sensitivity, low cost, and near real-time detection offered by membrane introduction mass spectrometry (MIMS)1 make it an especially attractive method for the analysis of volatile organic compounds (VOCs) in air samples. Current work indicates that MIMS may be very suitable for such vapor monitoring applications as stack and process monitoring and soil gas screening. Conventional analysis of air samples by EPA method TO-142 requires time-consuming collection and preconcentration steps. MIMS, which does not require any sample preparation steps, utilizes flow injection procedures for sample analysis and can be implemented for continuous VOC characterization in air. Air monitoring by MIMS has been accomplished using flat sheet membranes3,4 and hollow fiber membranes.5 A two-stage hollow fiber interface developed by Cooks’ group,6,7 which features †
Los Alamos National Laboratory. Montana Tech of the University of Montana. (1) Kotiaho, T.; Lauritsen, F. R.; Choudhury, T. K.; Cooks, R. G.; Tsao, G. T. Anal. Chem. 1991, 63, 875A. (2) Compendium Method TO-14. The Determination of Volatile Organics in Ambient Air using SUMMA Passivated Canister Sampling and Gas Chromatography Analysis; U.S. Environmental Protection Agency: Research Triangle Park, NC, 1988. (3) Matz, G.; Trinks, H. Int. J. Mass Spectrom. Ion Phys. 1982, 43, 79. (4) Cameron, D.; Hemberger, P. H.; Alarid, J. E.; Leibman, C. P.; Williams, J. D. J. Am. Soc. Mass Spectrom. 1993, 4, 774. (5) LaPack, M. A.; Tou, J. C.; Enke, C. G. Anal. Chem. 1990, 62, 1265. ‡
S0003-2700(96)00039-X CCC: $12.00
© 1996 American Chemical Society
pneumatically assisted transport of membrane permeate as demonstrated by Slivon et al.,8 provides additional sample enrichment by both analyte transport and the use of a jet separator to remove low molecular weight gases. We reported recently on the first use of this two-stage membrane-jet separator for the detection of VOCs in air at parts-per-trillion levels by ion trap mass spectrometry.9 Interest in improving detection limits for airborne polar VOCs and semivolatiles using silicone membranes led to the work discussed in this paper. Although there may be several reasons for the reduced sensitivity of the MIMS experiment toward polar compounds when silicone membranes are used, it is generally observed that polar compounds are not detected with the same efficiency as nonpolar compounds.10 Researchers have explored different techniques for improving the detection of polar or oxygen-containing VOCs in solution by MIMS. Xu et al. used a chemically modified membrane that selectively binds and releases analytes containing a specific functional group as a means to detect aldehydes in aqueous solutions.11 Sensitivity can also be enhanced through the use of ionization techniques other than electron ionization. Two of the important advantages of chemical ionization are that molecular ion abundance and control over ion fragmentation may be tailored through the choice of an appropriate reagent gas.12 Brodbelt13 demonstrated that proton transfer CI in the ion trap provides up to 1 order of magnitude improvement in sensitivity over electron impact ionization (EI) for compounds introduced via gas chromatography. Bier and Cooks14 demonstrated a CI method in combination with the MIMS technique; using isobutane as the reagent gas for proton transfer CI, they analyzed aqueous mixtures of alcohols and ketones introduced via a direct insertion membrane probe. For the analysis of aqueous solutions, Lauritsen and co-workers15 took advantage of the passage of water through a microporous membrane in a direct insertion probe; using water vapor as the (6) Dejarme, L. E.; Bauer, S. J.; Cooks, R. G.; Lauritsen, F. R.; Kotiaho, T.; Graf, T. Rapid Commun. Mass Spectrom. 1993, 7, 935. (7) Cooks, R. G.; Bauer, S. J. U.S. Patent 5,448,062, September 5, 1995. (8) Slivon, L. E.; Bauer, M. R.; Ho, J. S.; Budde, W. L. Anal. Chem. 1991, 63, 1335. (9) Cisper, M. E.; Gill, C. G.; Townsend, L. E.; Hemberger, P. H. Anal. Chem. 1995, 67, 1413. (10) Wong, P. S. H.; Cooks, R. G.; Cisper, M. E.; Hemberger, P. H. Environ. Sci. Technol. 1995, 29, 215A. (11) Xu, C.; Patrick, J. S.; Cooks, R. G. Anal. Chem. 1995, 67, 724. (12) Harrison, A. G. Chemical Ionization Mass Spectrometry; CRC Press: Boca Raton, FL, 1992. (13) Brodbelt, J. S.; Louris, J. N.; Cooks, R. G. Anal. Chem. 1987, 59, 1278. (14) Bier, M. E.; Cooks, R. G. Anal. Chem. 1987, 59, 597. (15) Lauritsen, F. R.; Choudhury, T. K.; Dejarme, L. E.; Cooks, R. G. Anal. Chim. Acta 1992, 266, 1.
Analytical Chemistry, Vol. 68, No. 13, July 1, 1996 2097
CI reagent gas served to improve detection limits for polar compounds in aqueous solution by at least 1 order of magnitude over a silicone membrane CI system. In analogous work with a porous polypropylene membrane, Lauritsen16 analyzed VOCs dissolved in organic solvents; again, the matrix containing the analytes of interest was used to advantage as the CI reagent. A corresponding technique has been used for the analysis of polar compounds in air. Gordon and Miller17 used a constituent of airswater vaporsas the CI reagent for determining polar VOCs in air; separation and analysis were accomplished with a gas chromatograph/ion trap detector. In this paper, we discuss an approach that parallels some of these features: the method uses membrane-transported oxygen as the charge exchange (CE) reagent gas for analysis of polar compounds in air. In earlier MIMS work, we reported the observation of a charge exchange process occurring as air pressure in the ion trap increased.4 However, the mechanism of this process was unknown. This paper clarifies the reaction mechanism leading to oxygen charge exchange and demonstrates the analytical utility of this process. In contrast to the sheet membrane interface used in our earlier work, the present study employed the simple two-stage hollow fiber membrane interface.6,9 Oxygen diffused through the silicone membrane in adequate amount to serve as the charge exchange reagent in the ion trap mass spectrometer. This technique improved instrument response for all of the oxygen-containing compounds we tested; in addition, our studies demonstrated its suitability for analysis of airborne hydrophobic VOCs such as benzene, carbon tetrachloride, chlorobenzene, and toluene. While the detection of polar compounds could have been accomplished by proton transfer CI with a reagent gas such as CH4, the utilization of membrane-transported O2, a matrix constituent, required no additional gas cylinders and was simple to implement. EXPERIMENTAL SECTION Apparatus and Instrumentation. The membrane/jet separator interface used in these experiments has been described in detail previously.6 The hollow fiber, polymeric silicone membrane (0.020 in. i.d. × 0.037 in. o.d., Dow Corning, No. 508-002) was attached at both ends to stainless steel hypodermic tubing after being briefly cleaned in hexane to remove residual organic contaminants18 and allowed to air-dry. Hexane was briefly observed in the mass spectrum after connection to the ion trap but rapidly disappeared. For these experiments, the fiber membrane was enclosed in 1/4 in. o.d. Teflon (DuPont) or stainless steel tubing. The membrane interface was heated to approximately 40 °C by wrapping heating tape around the exterior of the tubing enclosing the membrane. This feature was used during the analysis of polar and semivolatile compounds. Ultrapure helium flow through the membrane interior was used to transport VOCs to the ion trap manifold and served as buffer gas. Helium flow through the membrane was regulated at approximately 45 mL/min by a manual flow controller (Veriflo, SC440FC, Richmond, CA). (16) Lauritsen, F. R.; Kotiaho, T.; Choudhury, T. K.; Cooks, R. G. Anal. Chem. 1992, 64, 1205. (17) Gordon, S. M.; Miller, M. Report, EPA/600/3-89/070. NTIS Order No. PB90106451, 1989. (18) Cooks, R. G. Personal communication, Jan 1994.
2098 Analytical Chemistry, Vol. 68, No. 13, July 1, 1996
Table 1. Chemical Ionization Parameters Used for Charge Exchange MIMS Experimentsa stage
parameter
available settings
A
ionization period (µs) ionization rf level (Da) reagent reaction time (µs) analyte reaction time reaction rf level reagent ion ejection level (Da) mass scan (Da)
10-2500 0-50 400-25000 1-100 0-50 10-650 10-650
A′ B C D
experimental values 1500 17.5 25000 100 18 43 see text
a Stages A-D correspond to those in Figure 1 and are described in the Results and Discussion section.
Additional sample enrichment was provided by a heated metal jet separator (P/N 113617, SGE, Inc., Austin, TX) which linked the membrane assembly to a custom manifold interface on a Finnigan Magnum ion trap spectrometer (Finnigan MAT, San Jose, CA). The ion trap manifold was held at 40 °C for the work described here. Magnum software (ver. 2.4) operating in chemical ionization mode with axial modulation was used to obtain data. The parameters used in the chemical ionization scan function are listed in Table 1. A more complete discussion of these conditions is found in the next section. The scan range for all compounds was broad enough to include low-mass fragment ions and the molecular ion. Filament emission current was set at 10 µA, the recommended setting for CI experiments. Manifold pressure with helium flow on was measured at 2.4 × 10-5 Torr (uncorrected). Experiments to confirm the N2•+/O2•+ charge exchange reaction were done on a Finnigan MAT ion trap mass spectrometer (ITMS); a Teledyne HST 1000 broadband waveform synthesizer (Teledyne-Hitachi, Mountain View, CA) was used to selectively eject ions by resonant excitation to confirm the reaction mechanism. This instrumentation has been described previously.19 Gases were introduced to this instrument via a batch inlet system using Granville-Phillips variable leak valves (Model 203, Boulder, CO). Sample Preparation. We prepared all air samples described in this paper using a VICI Metronics Dynacalibrator (Model 34024-Y, Santa Clara, CA) and diffusion vials (sizes B and C) or permeation tubes. Chemicals analyzed from diffusion vials were chlorobenzene, acetone, methyl isobutyl ketone, isopropyl alcohol, cyclohexanol, propanal, dimethyl sulfoxide, diethylaminoethanol, and dimethyl methylphosphonate. Permeation tubes as supplied by VICI Metronics were used for methyl ethyl ketone, certified at 59.1 ( 3 ng/min; benzene, certified at 21.9 ( 2 ng/min; carbon tetrachloride, certified at 135 ( 5 ng/min; and toluene, estimated at 0.8 ng/min. Concentrations in air were calculated based on chamber dilution flow and temperature, and diffusion vial capillary length and opening, if used. Zero-grade compressed air (Matheson, Secausus, NJ) was used for CI parameter optimization experiments; otherwise, room air supplied the Dyncalibrator permeation chamber after passing through an internal charcoal filter. A small diaphragm pump at the sample outlet regulated sample air flow at 0.2 L/min over the membrane. Excess sample flow (greater than 0.2 L/min from the Dynacalibrator) and sample effluent were vented through charcoal filters. (19) Hemberger, P. H.; Nogar, N. S.; Williams, J. D.; Cooks, R. G.;. Syka, J. E. P. Chem. Phys. Lett. 1992, 191, 405.
Figure 1. Operation of the ion trap for charge exchange ionization. Stages A-D are described in the text. Reagent ion eject level was lowered to 40 Da when necessary to store lower mass fragment ions.
RESULTS AND DISCUSSION Chemical Ionization Parameters. The scan function in an ITMS is used to create, store, and detect ions through a programmed sequence of events. The duration and amplitude of the fundamental radio frequency (rf) voltage during each of these steps are adjusted to effect a specific process on the ions of interest. Much of this is handled automatically by the Magnum ion trap but chemical ionization scan functions require user tuning of these individual steps via the software for optimal analyte detection. Because reagent ion formation, reaction with sample, and analyte ion storage take place within the same trapping volume, the rf voltage level is adjusted to minimize space-charge effects and unwanted ion/molecule reactions. Figure 1 shows the changes in the amplitude of the fundamental rf voltage during mass spectral analysis. The rf voltage level is set during step A for the creation of the O2•+ charge exchange reagent ions. A large portion of the O2•+ ions is not formed directly by electron ionization but through the charge exchange reaction of O2 with N2•+ as described below. The charge exchange reaction of O2•+ with the analyte occurs in step B. After this reaction, residual O2•+ is ejected from the trap in step C by adjusting the rf voltage so that O2•+ is no longer stable within the ion trap and is ejected. The rf voltage is then ramped (step D) to obtain the mass spectrum. The presence of residual water in the ion trap lead to the formation of H3O+ during the reagent gas ionization and reaction periods. Reagent reaction time (step A′) was found to be related to the magnitude of the H3O+ signalslonger delay period led to disappearance of hydronium ion completely. If the rf voltage levels and times were not carefully adjusted in steps A and B, the M + H+ ion arising from the protonation of the analyte by H3O+ was observed for three analytes: acetone, methyl isobutyl ketone, and propanal. At higher analyte concentrations, self-protonation could be occurring, but at the low analyte concentrations in these experiments, water CI is more likely. Even though protonation was observed, it should be noted that [O2•+] was several orders of magnitude greater than [H3O+]. The values in the last column of Table 1 represent the parameters used to optimize the formation of the molecular ion of methyl ethyl ketone in air (180 ppb by volume) and were changed slightly for acetone, methyl isobutyl ketone, and propanal to minimize the protonation reaction. Transport of Air through Silicone Membranes. When the membrane inlet (with helium flow on) was attached and EI was employed, the observed ratio of oxygen to nitrogen was approximately 2:1. However, when the ion trap manifold was isolated, i.e., the membrane assembly was not attached, a conventional mass spectrum of residual air was observed with
Figure 2. O2•+/N2•+ as a function of time at constant partial pressure of air in the ion trap. Selective ejection of N2•+ indicates that it is the reactant ion in the formation of O2•+ by charge exchange.
oxygen and nitrogen in the expected ratio (nominal 1:3.7). There are likely to be two factors leading to the enhanced oxygen-tonitrogen ratio. First, oxygen permeation through silicone rubber is favored over nitrogen transport by a ratio of 2.2.20 Second, O2•+ concentration is enhanced by the charge exchange reaction of oxygen with N2•+. During our studies of charge exchange parameters, nitrogen was not observed at any significant level resulting, perhaps, from a combination of these conditions. We have shown previously that the oxygen-to-nitrogen ratio increases as the pressure of air in the trap increases.21 However, because these earlier studies were done with a Finnigan ITD 800, which did not allow selective ion storage or reaction period control, we were not able to confirm then that the increase in O2•+ was due to CE with N2•+. With another ion trap mass spectrometer, the Finnigan ITMS, we found that extending the reaction period, at a constant partial pressure of air, following EI also leads to an increasing ratio of O2•+/N2•+. Data from an experiment in which reaction times were varied while 1.0 × 10-5 Torr air was present in the ion trap (introduced through a variable leak valve) are shown in Figure 2. In this experiment, O2•+ was ejected by resonant ejection after the ionization period. During the subsequent reaction period, O2•+ was replenished by the ion/molecule reaction between N2•+ and neutral O2. However, if N2•+ was ejected following ionization, the concentration of O2•+ remained fairly constant during the reaction period. The ITMS experiments explain the sequence of events that occurs in the MIMS experiment. In the first step, both O2 and N2, which are resident in the ion trap at a membrane-mediated ratio, are ionized by electron impact. Following ionization, a charge exchange reaction between N2•+ and O2 leads to the depletion of N2•+ and the formation of additional O2•+, the production of which is favored by the exothermicity of the reaction between the reagent N2•+ and neutral oxygen (∆H ) -3.51 eV). Analytical Results. As a measure of the reproducibility of the CE-MIMS technique, three calibration plots for methyl ethyl (20) Stern, S. A. J. Membr. Sci. 1994, 94, 1. (21) Cameron, D.; Hemberger, P. H. Proceedings, 38th ASMS Conference on Mass Spectrometry and Allied Topics, 1990; p 61.
Analytical Chemistry, Vol. 68, No. 13, July 1, 1996
2099
Figure 3. Three calibration plots for methyl ethyl ketone obtained over a 5-day period with the (95% confidence limits for the average of the three plots. The response from m/z 43 + 57 + 72 is plotted against the concentration of MEK in air in ppbv. Table 2. Results of the Comparative Study of CE and EI Responses for Selected Polar and Volatile Organic Compoundsa compound
concn (ppb)
IP (eV)
CE/EIa
acetone methyl ethyl ketone methyl isobutyl ketone isopropyl alcohol cyclohexanol propanal
Hydrophilic 900 30 23 132 19 1800
9.69 9.5 9.3 10.15 9.8 9.98
4 5 5 15 22 6
benzene carbon tetrachloride chlorobenzene toluene
Hydrophobic 16 10 40 0.070
9.3 11.47 9.07 8.82
2 5 3 4
a The CE/EI ratio represents the ratio of response heights (total analyte ion current) for each ionization mode after background subtraction.
ketone were obtained over a 5-day period. Figure 3 shows the (95% confidence limits for the average of the three plots. The calibration points were taken at one temperature setting on the Dynacalibrator, which limits the concentration range available. Each point on the independent calibration curves represents the average of three samples of methyl ethyl ketone in air. Percent relative standard deviations for the independent points range from 1.2 to 11.7. Table 2 lists all compounds systematically analyzed by membrane introduction for this study. The time required to reach steady-state permeation through the membrane varied between 15 s for propanal and 7.7 m for cyclohexanol; for most of the volatile polar compounds, the rise time (10%-90%) was about 1 m, as observed previously.9 The results of the comparative study of CE and EI responses are given in the fourth column of the table, which presents the ratio of total analyte ion current obtained in each ionization mode. For each compound, we attempted to maximize response in each ionization mode by tuning scan function settings. Generally, the automatic gain control option (AGC or MCI) provided by the instrument software produced a higher response over a fixed ionization time. Charge exchange produced a greater signal over EI for all compounds as indicated by the response ratios, which ranged from 2 to 22. We observed 2100 Analytical Chemistry, Vol. 68, No. 13, July 1, 1996
Figure 4. Mass spectra of methyl isobutyl ketone from direct air sampling by MIMS. The upper spectrum was obtained using conventional EI (2 µs ionization time); the lower spectrum was obtained by CE ionization (70 µs ionization time; O2•+ reagent ion).
over a 1 order of magnitude improvement in response for the two alcohols we tested. Wong and Cooks22 used water CI with a microporous membrane and reported 4-75-fold detection limit improvements for polar compounds in aqueous solutions over results obtained using EI and a semipermeable silicone membrane. As expected, comparison of mass spectra obtained by the two ionization methods often revealed significant differences. The experimental EI mass spectra generally agreed with library spectra; however, experimental CE spectra showed qualitative and quantitative differences in most cases. Figures 4 and 5 contrast the experimental spectra of methyl isobutyl ketone and chlorobenzene, respectively, obtained in EI and CE modes. The MIBK spectra have different base peaks (m/z 43 for ionization by EI vs m/z 58 by CE), the lower energy CE process favoring the formation of the rearrangment ion at m/z 58. The relative abundance of the molecular ion at m/z 100 is 400% greater in the CE spectrum than in the EI spectrum. The CE chlorobenzene spectrum exhibits only molecular ions in contrast with the EI spectrum. These spectral variances are explained by the energy deposition processes in the two ionization modes. Both processes lead to the formation of odd-electron molecular ions, but the internal energy deposited is process-dependent. More energy is transferred to the analytes examined here in the EI process than by charge exchange with O2•+. Ionization by EI imparts a distribution of internal energies while the degree of fragmentation in the charge exchange process depends solely on the difference between the ionization energy and recombination energy of the (22) Wong, P. S. H.; Cooks, R. G. Anal. Chim. Acta 1995, 310, 387.
Figure 5. Mass spectra of chlorobenzene from direct air sampling by MIMS. The upper spectrum was obtained using conventional EI (500 µs ionization time); the lower spectrum was obtained by CE ionization (300 µs ionization time; O2+• reagent ion).
reactant and sample species.12 Energy differences in CE with oxygen span 0.59-3.24 eV for the compounds listed in Table 2. A discrete, lower level of energy goes into fragmenting the analyte species in the CE process, generally resulting in a different distribution of ion fragments. For four of the compounds (toluene, carbon tetrachloride, chlorobenzene, benzene), significant fragmentation was not observed in the CE mass spectra. A higher mass base peak was observed for the ketones and toluene; in addition, more prominent higher mass ions for the ketones, cyclohexanol, and propanal were detected in the CE mass spectra than in the EI spectra. Three semivolatile compounds ranging in boiling point from 161 to 189 °C were also analyzed to demonstrate the versatility of both the membrane introduction system and the application of CE to different compound classes. preliminary observations for dimethyl sulfoxide, diethyl aminoethanol, and dimethyl methylphosphonate indicate results similar to those for volatile polar compoundssmore intense molecular ions and higher response with charge exchange ionization. Mass spectra of dimethyl sulfoxide obtained under EI and CE conditions are shown in Figure 6. The dimethyl sulfoxide concentration in air was calculated to be 140 ppbv, based on its vapor pressure at 70 °C (9.4 Torr),23 which was the chamber temperature of the Dynacalibrator in this experiment. Despite the longer ionization time in the EI experiment, charge exchange resulted in a higher signal. Here again, we see the expected trends of lower fragmentation and increased sensitivity with CE ionization. (23) Jose, J.; Philippe, R.; Clechet, P. Bull. Soc. Chim. Fr. 1971, 8, 2860.
Figure 6. Mass spectra of dimethyl sulfoxide (bp 189 °C) from direct air sampling by MIMS. The upper spectrum was obtained using conventional EI (25 ms ionization time); the full-scale intensity is 2300 counts. The lower spectrum was obtained by CE ionization (1.5 ms ionization time; O2+• reagent ion); the full-scale intensity is 6200 counts. Despite an ionization time differential favoring EI by a factor of 17, charge exchange leads to greater sample response.
CONCLUSIONS Conventional semipermeable silicone membranes are suited for the introduction of both polar volatile organic compounds and certain semivolatile organics to an ion trap mass spectrometer. We have shown that membrane-transported oxygen can be used to enhance the detection of polar, hydrophilic compounds in air through charge exchange ionization. Silicone, a hydrophobic material, has been conventionally thought to be less suitable as a membrane material for the analysis of hydrophilic species. Our study here may indicate that difficulties surrounding the analysis of polar compounds in water by silicone membrane introduction may be related more to interactions of the matrix and analyte than to interactions of the membrane and analyte. “Fingerprint” information that is available in conventional EI mass spectra may be diminished when analysis by charge exchange ionization is selected. However, when mixtures of compounds are analyzed where there is no chromatographic separation as in the MIMS technique, such fingerprint information is of lesser value. More intense molecular ions were observed in most cases when charge exchange ionization was implemented, making the method more suitable for MS/MS speciation of mixtures. Quantitation may be more easily accomplished when fewer ions are observed or when only the molecular ion is detected. Received for review January 16, 1996. Accepted April 12, 1996.X AC960039F X
Abstract published in Advance ACS Abstracts, May 15, 1996.
Analytical Chemistry, Vol. 68, No. 13, July 1, 1996
2101
