Direct Analysis of Semivolatile Organic Compounds in Air by

scribed and detection of semivolatile compounds is shown. The analytical performance of the technique is established with methyl salicylate, including...
0 downloads 0 Views 64KB Size
Anal. Chem. 2001, 73, 5061-5065

Direct Analysis of Semivolatile Organic Compounds in Air by Atmospheric Pressure Chemical Ionization Mass Spectrometry Laurence Charles, Leah S. Riter, and R. Graham Cooks*

Department of Chemistry, Purdue University, West Lafayette, Indiana 47907

Atmospheric pressure chemical ionization is employed for direct air analysis, without ion source modification, by using the sheath gas as the sample transport agent. A simple modification of the sheath gas inlet line allows introduction of gaseous samples into a commercial atmospheric pressure chemical ionization source. Optimization and testing of this novel air sampling method are described and detection of semivolatile compounds is shown. The analytical performance of the technique is established with methyl salicylate, including a limit of quantification of 100 pptr, a limit of detection of 50 pptr, a linear response from 100 pptr to 20 ppb, and rise and fall times of 12 and 20 s, respectively. Using reagent ion monitoring, it is shown that the protonated methanol dimer is the principal CI reagent ion leading to protonated dimethyl methylphosphonate, while the monomer is mainly responsible for protonating methyl salicylate. Since the formation of the CI reagent (methanol clusters) can be controlled by simple variation of experimental parameters, the selectivity of the method can be easily adjusted to suit the targeted analyte. Performance is found to be independent of the choice of air or nitrogen as the sheath gas (and thus as the sample matrix) and this, together with the sensitivity and speed of the technique, make it promising for field studies. Analysis of trace organics in air, including compounds of environmental significance, is becoming increasingly important. Gas chromatography (GC) is the most commonly used analytical tool to solve this problem.1 However, this technique normally requires sample preconcentration via solid-phase extraction (SPE),2 solid-phase microextraction (SPME),3 or sorbent trapping.4 Such experiments are usually performed off-line and are time-consuming, a limitation for field analysis. To improve performance in air analysis, new methods using atmospheric pressure chemical ionization mass spectrometry (APCI-MS) are of interest. * Corresponding author: (e-mail) [email protected]; (fax) 765-494-9421. (1) Helmig, D. J. Chromatogr., A 1999, 843, 129-146. (2) Stuff, J. R.; Cheicante, R. L.; Durst, H. D.; Ruth, J. L. J. Chromatogr., A 1999, 849, 529-540. (3) Koziel, J.; Jia, M. Y.; Khaled, A.; Noah, J.; Pawliszyn, J. Anal. Chim. Acta 1999, 400, 153-162. (4) Wang, J. L.; Chen, W. L.; Lin, Y. H.; Tsai, C. H. J. Chromatogr., A 2000, 896, 31-39. 10.1021/ac010606l CCC: $20.00 Published on Web 09/26/2001

© 2001 American Chemical Society

APCI-MS is an analytical technique that is characterized by its simplicity, speed, sensitivity, and wide range of applicability.5,6 First reported by Horning et al.,7 it was developed by Henion and co-workers8 and by others9-11 as an interface for LC-MS. The APCI process involves the formation of reagent ions from the reagent molecules (usually solvent molecules from the LC mobile phase) that are ionized by a corona or similar discharge.12 Vaporization is performed by nebulizing the LC effluent in a heated chamber with the assistance of a sheath gas (usually nitrogen). In the positive ion mode, protonation is the usual ion/molecule reaction performed, the reagent ion transferring a proton to those analyte molecules present in the sample vapor with a higher proton affinity. Most compounds produce protonated molecules, [M + H]+, although clustering with water is also commonly observed in APCI. A variant on APCI involving proton-transfer reactions (PTR) has also been developed.13 In this experiment, typically H3O+ ions are produced in a high-pressure electron impact ion source containing water vapor, extracted from the source, preselected by a quadrupole analyzer, and injected into the drift region of a selected ion flow drift tube (SIFDT). The gas sample to be analyzed enters the flow-drift section through a reactant gas inlet, and proton transfer from H3O+ occurs in the course of collisions with trace constituents of the sample. Monitoring of the product ion signals is achieved by mass analysis using a quadrupole mass spectrometer, and this allows one to calculate the concentrations of the neutral trace compounds in the gas sample. This technique has been very successful for in situ on-line monitoring of trace (pptr level) volatile organic compounds in air.14 Very low level analysis of xenobiotic compounds in air is an (5) Niessen, W. M. A. J. Chromatogr., A 1998, 794, 407-435. (6) Abian, J. J. Mass Spectrom. 1999, 34, 157-168. (7) Horning, E. C.; Horning, M. G.; Carroll, D. I.; Dzidic, I.; Stillwell, R. N. Anal. Chem. 1973, 45, 936-943. (8) Covey, T. R.; Lee, E. D.; Bruins, A. P.; Henion, J. D. Anal. Chem. 1986, 58, 1451A-1461A. (9) Dzidic, I.; Carroll, D. I.; Stillwell, R. N.; Horning, E. C. Anal. Chem. 1976, 48, 1763-1768. (10) Kambara, H. Anal. Chem. 1982, 54, 143-146. (11) Garcia, D. M.; Huang, S. K.; Stansbury, W. F. J. Am. Soc. Mass Spectrom. 1996, 7, 59-65. (12) Rafaelli, A. In Selected topics and mass spectrometry in the biomolecular sciences; Caprioli, R. M., Ed.; Kluwer Academic Publishers: Amsterdam, 1997; pp 17-31. (13) Hansel, A.; Jordan, A.; Holzinger, R.; Prazeller, P.; Vogel, W.; Lindinger, W. Int. J. Mass Spectrom. Ion Processes 1995, 150, 609-619. (14) Lindinger, W.; Hansel, A.; Jordan, A. Chem. Soc. Rev. 1998, 27, 347-354.

Analytical Chemistry, Vol. 73, No. 21, November 1, 2001 5061

increasingly important application of PTR.15 The technique is characterized by good sensitivity and precise ionization control; however, the instrumentation is not simple and thus is not widely available. An alternative approach to the analysis of volatile compounds has been reported using an APCI source, modified to allow introduction of a gaseous sample.16,17 Flow of sample in the ionization source was achieved through a venturi made by sheathing the sample capillary in another tube carrying a flow of nitrogen. No liquid phase was employed, and water from the moisture in air was used as the CI reagent. To control the concentration of reactant ions in the source, and so achieve quantitative results, additional water was introduced in the form of humidified sheath gas.18 A practical way to perform gas analysis in APCI-MS without any source modification or sheath gas humidification is to retain the liquid mobile phase as a constant source of reactant ions and to utilize the sheath gas as the sample carrier. This idea is explored in this paper, and this method of sample introduction into the APCI source is evaluated for mass spectral analysis of semivolatile organic compounds in air. The common analytical criteria for judging performance of a trace analysis technique, (1) the rise and fall times (10-90%) of the analyte signals, (2) the limit of detection (LOD), (3) the limit of quantification (LOQ), and (4) the linear dynamic range, are utilized for characterization of the APCI air analysis method. Optimization of the experimental parameters, including flow rates of liquid and gas, sample matrix, and collision-induced dissociation (CID) conditions, was performed to achieve the maximum analytical performance. EXPERIMENTAL SECTION Sample Preparation. Methyl salicylate and dimethyl methylphosphonate were obtained from Sigma-Aldrich Corp. (Milwaukee, WI) and were used without purification. A 1-mL sample of the analyte of interest was placed into a sealed container containing ambient air and maintained at 25 °C for 1 h to allow liquidvapor equilibrium to be reached. The analyte concentration in the gaseous headspace solution was calculated based on the known vapor pressure of the analyte.19,20 A 3-mL aliquot of this gas mixture was sampled with a gastight syringe fitted with a shutoff valve (Hamilton, Reno, NV). The sample was injected at a controlled flow rate (0.5-2.0 mL/min) using a syringe pump (model 22, Harvard Apparatus, Holliston, MA) into the flowing stream of sheath gas (2.0-3.3 L/min.) supplied to the APCI source. APCI-MS Operating Conditions. All data were recorded using a Finnigan TSQ 700 triple-quadrupole mass spectrometer (Thermo Finnigan, San Jose, CA). The instrument was equipped with an atmospheric pressure ion source, which was operated with excess reagent gas, i.e., in the chemical ionization (APCI) mode. A Gilson HPLC pump (model 305) equipped with a manometric (15) Lindinger, W.; Hansel, A.; Jordan, A. Int. J. Mass Spectrom. 1998, 173, 191241. (16) Linforth, R. S. T.; Taylor, A. J. European Patent EP0819 937 A2, 1998. (17) Linforth, R. S. T.; Taylor, A. J. U.S. Patent 5,869,344, 1999. (18) Zehentbauer, G.; Krick, T.; Reineccius, G. A. J. Agric. Food Chem. 2000, 48, 5389-5395. (19) Tevault, D. E.; Keller, J.; Parsons, J., Proceedings of the ERDEC Scientific Conference on Chemical and Biological Defense Research, Aberdeen Proving Ground, MD 1998; 815-822. (20) Weast, R. C. In CRC Handbook of Chemistry and Physics; Chemical Rubber Co.: Cleveland, OH, 1971.

5062 Analytical Chemistry, Vol. 73, No. 21, November 1, 2001

Figure 1. Single ion monitoring traces of (a) 20 ppb methyl salicylate (m/z 153) and (b) 90 ppb dimethyl methylphosphonate (m/z 125) in air. Insets: mass spectra extracted from the total ion current trace. [M + H]+ ions of each analyte (m/z 153 methyl salicylate and m/z 125 dimethyl methylphosphonate) are observed in the corresponding mass spectra, as are protonated clusters of methanol [CH3OH]nH+, n ) 1-3, formed in the APCI source at m/z 33 (n ) 1), 65 (n ) 2), and 97 (n ) 3).

module (model 805) (Gilson Inc., Middleton, WI) was used to deliver the water/methanol (50:50, v/v) liquid phase into the APCI source (0.05-1.0 mL/min). The vaporizer temperature was 400 °C, and the transfer capillary was set at 200 °C. Either grade D OSHA breathable air or nitrogen (BOC gases, Murray Hill, NJ) was used as the sheath gas. No auxiliary gas was added. Mass spectra were recorded in the positive ion mode using the following conditions: corona discharge needle voltage 5 kV, plasma current 5 µA, and capillary inlet 15 V. Tandem mass spectrometry (MS/MS) experiments were based on collisioninduced dissociation using a collision energy of 15 eV. Argon (BOC gases, Murray Hill, NJ) was used as the target gas at a collision gas pressure of 2.3 mTorr, corresponding to multiple collision conditions. Analyte response was measured as the average height of the plateau from the baseline. RESULTS AND DISCUSSION Figure 1 shows single ion monitoring (SIM) traces for air samples injected into the APCI source through the nitrogen sheath gas. Protonated methyl salicylate (Figure 1a) and dimethyl methylphosphonate (Figure 1b) were monitored. As seen in the mass spectra (insets in Figure 1), protonated clusters of methanol

Figure 2. Reactant monitoring profiles for protonated methanol (m/z 33) and the [M + H]+ ion of methyl salicylate (m/z 153). As methyl salicylate is injected, the abundance of m/z 153 increases, while m/z 33 decreases in a statistically significant fashion (more than 2σ from the mean). Higher clusters of protonated methanol m/z 65 (n ) 2), and 97 (n ) 3) were also monitored, but were not plotted since the abundances did not change in the presence of the methyl salicylate.

[CH3OH]nH+ (n ) 1-3) are formed in the APCI source, and these are therefore candidates to be the reagent ions responsible for analyte protonation. The relative abundances of these cluster ions are highly dependent on source conditions. It was determined experimentally that the methyl salicylate signal was inversely proportional to the liquid flow rate and directly proportional to the sheath gas flow rate. The optimal conditions were found be 0.05 mL/min for the methanol/water solution and 3.3 L/min for the sheath gas. These conditions were also found to be those that favor the formation of protonated methanol over its dimer and trimer (less methanol is introduced into the APCI source at lower liquid flow rates, while a high sheath gas flow rate might cause more efficient collisional dissociation of large clusters). The inverse correlation between protonated methanol monomer and the methyl salicylate signal was confirmed by monitoring the intensity of the methanol clusters and the analyte during sample injection using reagent ion monitoring.21 As shown in Figure 2, as the methyl salicylate analyte is introduced into the APCI source, [M + H]+ ion forms readily, while the protonated methanol monomer intensity decreases. The intensities of the higher clusters of protonated methanol at m/z 65 (n ) 2), and 97 (n ) 3) were also monitored but are not shown since they did not change significantly (more than 2σ from the mean). It should be noted that the magnitude of the depletion in the [CH3OH]H+ signal in Figure 2 is much smaller than the increase of [M + H]+, suggesting that proton transfer from the protonated methanol monomer is not the only process involved in methyl salicylate ionization. However, no alternative proton sources could be identified and we note that the experiment in Figure 2 was performed at high analyte concentration to allow the observation of reagent ion depletion, and this might induce self-ionization of the analyte. In contrast to methyl salicylate, the dimethyl methylphosphonate signal was directly proportional to the liquid flow rate and indirectly proportional to the sheath gas flow rate, which favors (21) Hatch, F.; Munson, B. Anal. Chem. 1977, 49, 731-733.

Figure 3. Effect of liquid flow rate on the dimethyl methylphosphonate and the methanol cluster ion intensities (the latter plotted as the value in the blank signal minus the value during sample injection). The changes in intensity of protonated methanol dimer (m/z 65) follow that of the [M + H]+ ion of DMMP (m/z 125) while those of the other clusters of protonated methanol m/z 33 (n ) 1) and 97 (n ) 3) do not. This suggests that the protonated methanol dimer is the primary CI reagent ion yielding protonated dimethyl methylphosphonate.

the formation of methanol cluster ions (optimal conditions at 1.0 mL/min for liquid and 2.0 L/min for the sheath gas). These ions must therefore protonate dimethyl methylphosphonate; in particular, the correlation between the dimethyl methylphosphonate signal and the depletion of the methanol dimer as a function of liquid flow rate (Figure 3) suggests that the protonated methanol dimer is the main CI reagent for dimethyl methylphosphonate. This interpretation is further supported by the fact that increasing depletion of the protonated methanol trimer with increasing liquid flow rate during dimethyl methylphosphonate sample injection is not accompanied by an observable increase of any other peak in the mass spectrum. These conclusions were tested by considering the reaction thermochemistry of the protonation reactions. Since the PA values of methyl salicylate and dimethyl methylphosphonate are not available in the literature, they were estimated using data for similar compounds.22 Methyl salicylate proton affinity was estimated as 855 ( 5 kJ/mol by consideration of the values for methyl benzoate (PA ) 850.5 kJ/mol), methyl 3-hydroxybenzoate (PA ) 850.0 kJ/mol), methyl 4-hydroxybenzoate (PA ) 863.4 kJ/ mol), methyl 2-methylbenzoate (PA ) 858.3 kJ/mol), methyl 3-methylbenzoate (PA ) 857.7 kJ/mol), methyl 4-methylbenzoate (PA ) 861.5 kJ/mol), methyl 3-methoxybenzoate (PA ) 856.7 kJ/ mol), and methyl 4-methoxybenzoate (PA ) 870.6 kJ/mol). An estimate of 925 ( 10 kJ/mol for the PA of dimethyl methylphosphonate was made by considering the data for the following compounds: acetone (PA ) 812.0 kJ/mol), methyl acetate (PA ) 821.6 kJ/mol), dimethyl carbonate (PA ) 830.2 kJ/mol), and trimethylphosphine oxide (PA ) 909.7 kJ/mol). The binding energy of CH3OH2+ and CH3OH is reported to be 135 ( 5 kJ/ mol.23 Methanol proton affinity is reported as 755 kJ/mol.22 With these values, the reaction enthalpies of the proton transfer to the (22) Hunter, E. P.; Lias, S. G. J. Phys. Chem. Ref. Data 1998, 27, 413. (23) Keesee, R. G.; Castleman, A. W. J. Phys. Chem. Ref. Data 1986, 15, 10111071.

Analytical Chemistry, Vol. 73, No. 21, November 1, 2001

5063

Figure 4. Sensitivity of MS vs MS/MS experiments for methyl salicylate. The slope of the calibration curves was measured as 1 × 107 counts/ppb for SIM (m/z 153) in single-stage mass spectrometry and 4 × 106 counts/ppb for SRM (m/z 153 f 121) in tandem mass spectrometry. The limit of quantification was lower for MS/MS (100 pptr) than for MS (200 pptr).

methyl salicylate from the methanol monomer is calculated to be exothermic (755 - 855 ) -100 kJ/mol), while that from the dimer is endothermic (755 -855 + 135 ) +35 kJ/mol) and that from the trimer is expected to be even more endothermic. In contrast, the reaction enthalpy of the proton transfer to the dimethyl methyl phosphonate is estimated to be exothermic for both the methanol monomer (755 - 925 ) -170 kJ/mol) and dimer (755 - 925 + 135 ) -35 kJ/mol). The value for the trimer reaction is not known but should be slightly exothermic. Experimental data are consistent with the calculated reaction enthalpies. It was found that the methyl salicylate is only protonated by the protonated methanol monomer, and not by the dimer or trimer, which is consistent with the calculation that the proton-transfer reaction from the protonated methanol monomer is exothermic, while those from the dimer and trimer are endothermic. Likewise, it was experimentally determined that dimethyl methylphosphonate can be protonated by any methanol clusters (to different extents, depending on the reaction conditions; see Figure 3), which is consistent with proton-transfer reactions each of the protonated methanol clusters being exothermic. Since the extent of formation of the methanol clusters (the CI reagents) can be controlled by simple variation of relative liquid and sheath gas flow rates, the selectivity of the method can be adjusted to suit the targeted analyte. The cluster ion/analyte ion/molecule reaction yielded protonated methyl salicylate, [M + H]+, at m/z 153, which could be isolated and examined by MS/MS. The product ion spectrum of m/z 153, recorded under multiple collision conditions, showed a single major fragment ion at m/z 121, which is ascribed to the loss of methanol from the protonated ester to give the substituted benzoyl ion. The sensitivity (response per unit change in concentration) of single-stage MS is higher than that of MS/MS, as shown in Figure 4. However, the added specificity of MS/MS experiments results in lower limits of quantitation (100 pptr, vs 200 pptr for singlestage MS, calculated from the calibration curve as the concentration corresponding to a signal intensity 10σ of the blank). This situation is well known24 but has seldom been as explicitly 5064 Analytical Chemistry, Vol. 73, No. 21, November 1, 2001

Figure 5. Flow injection analysis of methyl salicylate in the single reaction monitoring mode (transition m/z 153 f 121). The average rise and fall times were 12 and 20 s, respectively. The limit of quantification was shown to be 100 pptr.

demonstrated. Due to this added specificity and lower limit of quantitation, MS/MS single reaction monitoring (SRM) was used to acquire all of the following data. The analytical performance of the method was evaluated with methyl salicylate by recording the criteria of rise and fall times, linear dynamic range, limit of detection, and limit of quantification. As established by flow injection analysis (FIA) experiments (Figure 5), the average 10-90% rise and fall times were 12 and 20 s, respectively. The potential for interferences in SRM experiments, compared to that of full-scan tandem mass spectrometry, is recognized. These results suggest the potential utility of the method for real-time analysis of semivolatile organic compounds in gaseous samples. Since air is the targeted matrix, the linear dynamic range of the method for methyl salicylate was measured for both nitrogen and air as the sheath gas (and hence also as the sample matrix). The method was found to be linear from 100 pptr to 20 ppb with a slope of 8 × 106 (R2 ) 0.9982) for nitrogen and 8 × 106 (R2 ) 0.995) for dry air. The performance of the technique is independent of the sample matrix to the extent tested, an advantage since air is the desired sheath gas for in-field studies. Although dry air was utilized for these studies because of availability, it is anticipated that the moisture content and variation of ambient air will not be of concern due to the copious amounts of water added to the air in the ion source of this experiment. This has been shown in previous APCI work where ambient air was utilized and then humidified to circumvent problems associated with variation of water content in air samples.25 The upper value of the linear dynamic range was limited by the technique used to prepare the sample and not by the method itself. The limit of quantification, 100 pptr, was evaluated as that concentration which yields a signal equivalent to 10 times the standard deviation (σ) of the blank signal (Figure 5). The limit of detection, 50 pptr, was calculated from the calibration curve as the concentration corresponding to a signal as 3σ the blank. (24) Busch, K. L.; Glish, G.; McLuckey, S. Mass Spectrometry-Mass Spectrometry: Techniques and Applications of Tandem Mass Spectrometry; VCH Publishers: Weinheim, Germany, 1989. (25) Taylor, A. J.; Linforth, R. S. T.; Harvey, B. A.; Blake, A. Food Chem. 2000, 71, 327-338.

CONCLUSIONS A new method for introduction of air samples containing traces of semivolatile organic compounds into a mass spectrometer has been described and tested. The method allows gaseous samples to be introduced directly into an APCI source, via the sheath gas stream. No other modifications are made to the mass spectrometer. The sensitivity and speed of the technique may help expand the applicability of APCI to on-line detection of semivolatile organic compounds in a gaseous matrix. Since the formation of methanol clusters, shown to be the CI reagents involved, can be controlled by simple variation in the relative liquid and sheath gas flow rates, the selectivity of the method can be easily adjusted to suit the targeted analyte. High performance was achieved with air as the gas matrix, demonstrating that the method has promise as an in situ technique. In the future, development of additional chemical

selectivity will be investigated by (1) changing the CI reagent by making modifications in the composition of the liquid phase and (2) utilizing the negative ion mode to extend the scope of the method to explosives surrogates, such as nitroaromatic compounds. In addition, field studies with the APCI method will be attempted. ACKNOWLEDGMENT This work was supported by NAVSEA/NSWC Crane N0016400-C-0047. We thank a reviewer for helpful comments on the thermochemistry. Received for review May 30, 2001. Accepted August 23, 2001. AC010606L

Analytical Chemistry, Vol. 73, No. 21, November 1, 2001

5065