Online Volatile Organic Compound Measurements Using a Newly

Jun 17, 2005 - in detail by Lovejoy and Wilson (10), and recent modifications are discussed by Curtius et al. (11). With a system of four lenses the i...
0 downloads 0 Views 647KB Size
Download PDF
Environ. Sci. Technol. 2005, 39, 5390-5397

Online Volatile Organic Compound Measurements Using a Newly Developed Proton-Transfer Ion-Trap Mass Spectrometry Instrument during New England Air Quality StudysIntercontinental Transport and Chemical Transformation 2004: Performance, Intercomparison, and Compound Identification C A R S T E N W A R N E K E , * ,†,‡ S H U J I K A T O , § J O O S T A . D E G O U W , †,‡ PAUL D. GOLDAN,† WILLIAM C. KUSTER,† MIN SHAO,| E D W A R D R . L O V E J O Y , † R A Y F A L L , ‡,§ A N D F R E D C . F E H S E N F E L D †,‡ National Oceanic and Atmospheric Administration, Aeronomy Laboratory, 325 Broadway, Boulder, Colorado 80305, Cooperative Institute for Research in Environmental Sciences, University of Colorado, Boulder, Colorado 80309, Department of Chemistry and Biochemistry, University of Colorado, Boulder, Colorado 80309, and Department of Environmental Sciences, Peking University, Beijing, China

We have used a newly developed proton-transfer ion-trap mass spectrometry (PIT-MS) instrument for online trace gas analysis of volatile organic compounds (VOCs) during the 2004 New England Air Quality StudysIntercontinental Transport and Chemical Transformation study. The PIT-MS instrument uses proton-transfer reactions with H3O+ ions to ionize VOCs, similar to a PTR-MS (proton-transfer reaction mass spectrometry) instrument but uses an ion trap mass spectrometer to analyze the product ions. The advantages of an ion trap are the improved identification of VOCs and a near 100% duty cycle. During the experiment, the PIT-MS instrument had a detection limit between 0.05 and 0.3 pbbv (S/N ) 3 (signal-to-noise ratio)) for 2-min integration time for most tested VOCs. PIT-MS was used for ambient air measurements onboard a research ship and agreed well with a gas chromatography mass spectrometer). The comparison included oxygenated VOCs, aromatic compounds, and others such as isoprene, monoterpenes, acetonitrile, and dimethyl sulfide. Automated collisioninduced dissociation measurements were used to determine the contributions of acetone and propanal to the measured signal at 59 amu; both species are detected at this * Corresponding author phone: 001-303-497-3601; e-mail: [email protected]. † National Oceanic and Atmospheric Administration. ‡ Cooperative Institute for Research in Environmental Studies, University of Colorado. § Department of Chemistry and Biochemistry, University of Colorado. | Peking University. 5390

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 14, 2005

mass and are thus indistinguishable in conventional PTRMS.

1. Introduction Volatile organic compounds (VOCs) are emitted into the atmosphere in large quantities from a variety of different natural and anthropogenic sources (1). VOCs are key ingredients in the formation of ozone and aerosols in polluted air, and play a significant role in determining regional air quality, in the chemistry of the global troposphere, and possibly in the global carbon cycle. There is continuing interest in the development of new methods for analysis of complex mixtures of VOCs in ambient air. Proton-transfer reaction mass spectrometry (PTR-MS) is such a method and was developed by Lindinger and co-workers at the University of Innsbruck (2). This method allows fast online measurements of some important VOCs in air, including many oxygenated VOCs. PTR-MS utilizes proton-transfer reactions of hydronium ions, H3O+, to ionize VOCs, in combination with mass spectrometric detection of the product ions. PTRMS is well suited for studying the atmospheric chemistry of organic compounds, because it allows many important VOCs from natural and man-made origin to be measured along with their atmospheric oxidation products. In a PTR-MS measurement, only the mass of the product ions is determined, which is a valuable but certainly not a unique indicator of the identity of trace gases. It is clear, for example, that different isomers cannot be resolved in this manner. Recently a method was developed that separates the contributions from different VOCs to a single mass signal by coupling a gas chromatographic (GC) column to a PTRMS instrument (3). It was shown that at many mass signals of interest only one VOC contributed to the signal. The disadvantage of this approach is that the acquisition of a sufficiently large air sample and the subsequent GC separation of the compounds may take as long as 30 min, and the ability of performing fast online measurements is lost. PTRMS was further validated in an intercomparison with a GCMS instrument, and it was demonstrated that PTR-MS can measure many VOCs and several isomeric classes of VOCs without significant interferences (4). This intercomparison was part of the New England Air Quality Study 2002 (NEAQS2002), where the NOAA ship Ronald H. Brown was deployed in the New England area. Here we present results obtained with a PIT-MS (protontransfer ion-trap mass spectrometer) instrument that is similar to a PTR-MS but uses an ion-trap mass spectrometer instead of a quadrupole mass filter for the detection of ions. This technique was first introduced by Prazeller et al. (5) under the name PTR-ITMS (proton-transfer reaction ion-trap mass spectrometer). PIT-MS and its potential advantages over PTR-MS have recently been discussed by Prazeller et al. (5) and Warneke et al. (6). These are (1) the analytical capabilities of an ion trap: collision-induced dissociation (CID) and ion-molecule reactions can be performed in the ion trap to separate some ions that are indistinguishable in PTR-MS, and (2) an ion trap has a near 100% duty cycle for a scan over a wide mass range which eliminates the need for selecting the specific mass signals to be monitored prior to an experiment. During July and August of 2004, the New England Air Quality Study was repeated as part of the NEAQS-ITCT 2004 (ITCT, Intercontinental Transport and Chemical Transformation) campaign. VOCs were measured by several techniques onboard the NOAA research ship Ronald H. Brown. Here we compare mixing ratios of several hydrocarbons and some 10.1021/es050602o CCC: $30.25

 2005 American Chemical Society Published on Web 06/17/2005

The ship carried an extensive set of instruments to characterize the gas-phase and aerosol properties of the atmosphere. Research goals included a detailed characterization of (1) the primary emissions of gas-phase and aerosol species on the North American continent, (2) the chemical transformation leading to the formation of secondary pollutants (ozone and aerosol), and (3) the transport processes involved. The gray scale of the cruise track in Figure 1 is the toluene/benzene ratio, which gives an indication of the photochemical age of the sampled air. Benzene and toluene are emitted with a typical ratio from mostly traffic but removed from the atmosphere in reactions with OH at different rates so that the ratio gives an indication of the photochemical age. Estimated photochemical ages ranged from a very fresh plume close to Boston, MA, to about a few days close to Chebogue Point in Nova Scotia, where one of the NEAQS-ITCT 2004 ground sites was located. During the cruise, the ship was carefully maneuvered to always face the wind and not measure the ship exhaust. The few occasions, when ship exhaust was measured, were identified using NO measurements and removed from the data set.

FIGURE 1. Cruise track of the NOAA Ronald H. Brown during the NEAQS-ITCT study in 2004. The gray scale indicates the benzene/ toluene ratio, which is an indicator of photochemical age of the air mass. oxygenated VOCs, measured simultaneously by PIT-MS and GC-MS to validate the newly developed PIT-MS technique. In addition, the improved analytical capabilities of PIT-MS were tested by performing CID measurements of ions at 59 amu. Acetone and propanal are both detected at this mass and are therefore indistinguishable by standard PTR-MS.

2. Field Measurements 2.1. Cruise Track. The cruise track of the NOAA ship Ronald H. Brown is shown in Figure 1.

2.2. Instrument Setup. The PIT-MS instrument used in this work was deployed in a field experiment for the first time during this cruise. This home-built instrument was described in detail by Warneke et al. (6), and only a brief description is given here. The PIT-MS instrument, shown in Figure 2, consists of four parts: (1) an all-Teflon gas inlet and handling system, (2) an ion source for the production of H3O+ primary ions, (3) a drift-tube reaction chamber, and (4) the detection system with the ion-trap mass spectrometer and a secondary electron multiplier (SEM). The first three parts, similar to PTR-MS, were described in detail previously (7-9). H3O+ ions, produced in the ion source, undergo proton-transfer reactions with VOCs in a drift-tube reactor, and the product and primary ions are measured with an ion trap. H3O+ only reacts with compounds that have a higher proton affinity than water,

FIGURE 2. Schematic drawing of the PIT-MS instrument including the gas inlet system. VOL. 39, NO. 14, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

5391

which include most of the VOCs and oxygenated VOCs. H3O+ does not react with the main constituents of air (nitrogen, oxygen, and rare gases) and also not with most alkanes, ethylene, and acetylene. The sample air (20 STP mL min-1) (STP ) standard temperature of 273.15 K and pressure of 1.013 bar) is pumped through the drift-tube reactor in which it has a residence time of about 0.2 s. The residence time is the most limiting factor for the response time of the instrument with memory effects being another one. Typical values in the drift tube used here were 600 V electric field and 2.4 mbar pressure. This results in an E/N value (electric field over gas number density) of about 108 Td (1 Td ) 10-17 V cm2). This setting results in the production of [H3O‚H2O]+ ions in the drift tube, and as these ions are also reactive toward VOCs, their presence has to be accounted for as described by de Gouw et al. (4) and Warneke et al. (9). The ion trap used in the PIT-MS instrument was described in detail by Lovejoy and Wilson (10), and recent modifications are discussed by Curtius et al. (11). With a system of four lenses the ions are focused into the ion trap. The bias of the last lens can be controlled to either focus ions into the trap or block them when analyzing the trapped ions. The ion trap has an internal radius of 1 cm and stretched end cap geometry. Pressure in the ion trap was kept at 1 × 10-3 mbar during all measurements by adding ultrapure He to the iontrap chamber. About 5% of the gas in the ion trap chamber is sample air from the drift-tube reactor. Each ion trap end cap has a 1 mm diameter centered aperture. Ions ejected through the exit end cap are accelerated onto a discrete dynode SEM, which is located in a differentially pumped chamber at a pressure of 1 × 10-5 mbar. The ion trap electronics consists of National Instruments PCI boards and a “LabView” (National Instruments) program that controls the amplitude of the ring electrode RF voltage, the lens gate bias, and end-cap waveforms and also acquires data during the mass scan. Two different waveforms are applied to the ion trap end caps: (1) filtered noise fields (FNF) (12) used for mass filtering during the ion-trapping period and also exciting trapped ions for fragmentation using CID (also called MS-MS mode), and (2) sine waves used for axial modulation during the mass scan. The data acquisition electronics consists of a discrete dynode electron multiplier, a preamplifier, and an analogue to digital converter. The software allows the user to adjust the timing and the amplitude of the RF voltage, the end-cap waveforms, and the voltage that gates the ions into the ion trap. This flexibility allows the user to: choose the trapping time, perform fast mass scans of all trapped ions, selectively trap specific ions, do CID on the selected ions, and carry out ion-molecule reactions in the ion trap by adding reactant gases to the He buffer in the trap chamber. The PIT-MS instrument is built into a double-width 19in. rack with a height of 24 in. It weighs about 130 kg and has a total power consumption of about 1.2 kW. The PIT-MS and GC-MS were mounted inside a sea container on the front deck of the Ronald H. Brown. Air was sampled through a 7.5-m long double-walled perfluoroalkoxy (PFA) tube that was attached to a beam pointing forward from the sea container. Air was pumped through the inner 1/ in. outside diamater (o.d.) tube to the GC-MS instrument 4 by a 7 STP L min-1 diaphragm pump (STP ) standard temperature of 273 K and pressure of 1 atm). The excess flow from the GC-MS (about 6 STP L min-1) was exhausted through the outer 1/2 in. o.d. tube. By use of this arrangement, concentration gradients across the PFA material of the inlet line were avoided and thus the permeation of impurities through the PFA was minimized. A flow of 0.4 STP L min-1 branched off the main inlet toward the PIT-MS. This flow first passed through a Teflon needle valve, and after this point the pressure in the line was maintained to a constant 5392

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 14, 2005

350 mbar by a pressure controller that regulated the PIT-MS inlet flow. A small portion of this inlet flow (20 STP mL min-1) branched off to the drift-tube reactor through a 1/16 in. o.d. Silcosteel (Restek Bellefonte, PA 16823) line. In this setup, the pressure in the drift-tube reactor was kept at a constant value of 2.4 mbar, and the sample air is only exposed to Teflon surfaces before reaching the drift tube. About 2 × 104-1 × 105 ions can be stored in the ion trap before space charge effects distort the peak shapes. To estimate the number of ions that are overfilling the ion trap, the measured gain of the SEM was used together with an estimated efficiency of the extraction of the ions from the ion trap. The ion trap is filled up with primary ions in about 20 ms, which is too short to detect small mixing ratios of product ions. Therefore primary and product ions were measured separately (6). The primary ions (H3O+, [H3O‚H2O]+, and impurities O2+ and NO+) were measured with a trapping time of 20 ms every 73 min. During the cruise, the ion trap was set up to measure mass scans of the product ions with a trapping time of 5 s. The trapping voltage was chosen to trap all masses from 25-160 amu so that H3O+ is not trapped, and all other primary ions ([H3O‚H2O]+, O2+, and NO+) were made unstable in the ion trap using FNF during the trapping time. Twenty mass scans were averaged to reduce the noise on the signal, resulting in a 110-s measurement frequency. The actual trapping time for each measurement point is 100 s, and the rest is used for acquiring the mass scan, emptying the ion trap, and integrating the mass peaks (6). For most species there is a nonzero offset in the signal due to impurities in the system. These offsets are determined by passing the ambient air through a catalyst, which consists of a stainless steel tube filled with Pt-coated quartz wool (Shimadzu) heated to 350 °C. The instrument backgrounds were determined every 73 min, immediately following the primary ion signal measurements. CID measurements were performed every 73 min on mass 59 during most of the cruise. A CID measurement is done by isolating mass 59 ions in the ion trap. This is achieved by applying a FNF to the end caps so that all other ions are unstable and therefore not trapped. After the trapping, a different FNF is applied to the end caps (width about 1 amu), which adds kinetic energy to ions of mass 59 so that fragmentation occurs. The amplitude of the FNF controls the degree of fragmentation that occurs. Since there is no energy added to ions other than mass 59, the fragment ions (masses 31 and 41, as will be shown later) remain trapped. In the CID measurements reported here. the FNF amplitude is increased stepwise until all primary mass 59 ions are fragmented. 2.3. Calibrations. The response of the instrument to 10 different VOCs was calibrated by standard addition of a calibration gas to the inlet every 8 h. Calibrations at several different mixing ratios were performed every 2 days during the entire cruise. Calibration procedures were similar to the ones described by Warneke et al. (3) and de Gouw et al. (4). The response of the PIT-MS instrument was linear in the calibrated range of 5-70 ppbv. Results from all calibration measurements are shown in Figure 3, and an average calibration factor is given in Table 1. The calibration factors for some compounds (methanol, isoprene, and the monoterpenes) are clearly lower than the average. In the case of isoprene and the monoterpenes this is caused by fragmentation, because only about 50% of the signal is detected at the parent mass for these compounds. The detection efficiency of the ion trap is smaller at the lower end of the mass range and therefore methanol at mass 33 is detected less efficient. The calibration factors for all compounds in the standard gas varied slightly during the cruise and these changes were taken into account for the calculation of the mixing ratios. After the cruise it was

FIGURE 3. Results from all PIT-MS calibration measurements during NEAQS-ITCT 2004. The dashed lines give the calibration factors that were used to determine the mixing ratios.

TABLE 1. Parameters Describing the Intercomparison between the PIT-MS and GC-MS Measurements during the NEAQS-ITCT Study in 2004 (Detection Limits for a 2-min PIT-MS Measurement Compared with the Detection Limits for PTR-MS Measurements during NEAQS in 2002) intercomparison

detection limits (pptv)

compound

m/z (amu)

slope

offset (pptv)

r2

methanol acetonitrile acetaldehyde MTBE acetone + propanal DMS isoprene MVK + MACR MEK benzene toluene C8 aromatics monoterpenes

33 42 45 57 59 63 69 71 73 79 93 107 137

0.77 ( 0.11 0.97 ( 0.14 0.92 ( 0.13 1.22 ( 0.06 0.97 ( 0.015 1.11 ( 0.04 1.28 ( 0.04 1.02 ( 0.003 1.30 ( 0.07 2.35 ( 0.4 1.40 ( 0.005 1.54 ( 0.06 2.16 ( 0.05

-44 ( 21 -12 ( 13 160 ( 64 23 ( 3 7(8 5(3 30 ( 7 12 ( 3 -4 ( 5 5 ( 17 18 ( 6 7(5 19 ( 15

0.96 0.80 0.82 0.89 0.97 0.97 0.87 0.96 0.92 0.76 0.94 0.90 0.75

c

PIT-MS 2004

PTR-MS 2002 (4)

sensitivity (mV/ppbv)

550 73 350 93 200 75 210 140 95 110 75 105 205

170 24 150 n.a. 59 n.a. 30 24 290 32 23 76 28

1.03 ( 0.1a 5.37b 4.14 ( 0.23a 4.04b 5.37 ( 0.53a,c 3.58b 1.60 ( 0.13a 2.43 ( 0.28a,c 4.40b 3.43 ( 0.52a 4.14 ( 0.50a 4.08 ( 0.47a,c 1.56 ( 0.17a,c

a Measured during NEAQS-ITCT2004. b Measured before and after NEAQS-ITCT2004 and scaled to the calibrated response of the instrument. The weighted average calibration factor, as described by de Gouw et al. (4) is given.

discovered that laminar flow conditions existed in parts of the gas inlet system, which caused incomplete mixing. This introduced extra uncertainties in the calculated mixing ratios. Scaling factors for the incomplete mixing were thus determined in laboratory tests after the cruise with a different gas inlet setup. The measured scaling factors are mass dependent and varied from 0% for methanol on mass 33 and a maximum of 20% for the monoterpenes on mass 137. For VOCs not contained in the standard mixture, we used calibration factors determined from other mixtures in laboratory measurements before and after the campaign. The response of the instrument changed during the cruise, as was shown in Figure 3, and the same time variation was assumed for the calibration factors obtained in the laboratory measurements. At some masses, the PIT-MS measures more than one VOC. In these cases, we have used a weighted average of the calibration factors for the different VOCs (acetone + propanal, mass 59; methyl vinyl ketone (MVK) + methacrolein (MACR), mass 71; aromatics, mass 107 and mass 121; monoterpenes, mass 137). This procedure is described in detail by de Gouw et al. (4). The uncertainty in the calibration is about 25%, which is mainly determined by the uncertainty of the calibration standard. In addition to direct calibration, the sensitivity of PIT-MS can be estimated using the reaction time and the proton-transfer reaction rate coefficients. Most of the reaction rate coefficients are only known to within ∼40%. The calculated and measured calibration factors agreed within

the stated uncertainties. Here the measured calibration factors are used, because they are more accurate and also might better account for some possible inlet effects. 2.4. GC-MS Instrument. Nonmethane hydrocarbons and oxygenated VOCs were measured by online GC-MS using an automated instrument. A detailed description of this instrument and its analysis procedure is given elsewhere (4, 13). The GC-MS instrument analyzed 350-mL air samples with a 5-min acquisition time every 30 min. More than 100 VOCs including many oxygenated compounds, hydrocarbons, halocarbons, and alkyl nitrates can be identified and quantified with this instrument. The detection limit of the GC-MS instrument is 0.75. At low mixing ratios, however, the intercomparisons for acetaldehyde, MTBE, DMS, isoprene, and the monoterpenes diverge from the 1:1 relation. This is likely caused by other ions contributing to the PIT-MS mass signals used to determine the mixing ratios (3). MTBE (mass 57) can have contributions from butenes or butanol, the monoterpenes (mass 137) from unknown compounds but likely other biogenic VOCs, and isoprene (mass 69) from 2-methyl-3-buten-2-ol, pentanal, and pentenol. The measurement of DMS at mass 63 is complicated by the presence of H+‚CH3CHO‚H2O cluster ions at that same mass. The contribution to the mass 63 signal from the acetaldehyde cluster was estimated from laboratory measurements and subtracted from the DMS signal and it is these corrected data that are intercompared with the GC-MS results

in Figure 5 and Table 1. This correction is small and changed only the offset in the intercomparison by about 15 pptv. There is no known interference for mass 45 (acetaldehyde) (3, 4). The GC-MS acetaldehyde measurements have been corrected for a known ozone artifact produced by ambient levels of ozone (13) and possibly overcorrected in the measurements presented here. With the exception of acetaldehyde, all the intercepts reported here are similar in magnitude to those found for the NEAQS2002 intercomparison of PTR-MS and GC-MS. The larger offset for acetaldehyde, seen in the present data set, may have resulted from a change in the ozone induced artifact in the GC-MS measurements. 3.2. PIT-MS Detection Limit. The detection limits of the PIT-MS instrument are determined by the background signals and the calibration factors. In Table 1 detection limits are given for the investigated compounds and compared with the detection limits for PTR-MS as determined during NEAQS in 2002 (4). The detection limits are defined as three times the standard deviation in the background measurements (S/N ) 3) and given in Table 1 for a 2-min measurement together with the respective sensitivity (4, 9). In Figure 5 also data below the detection limit for 2-min measurements are shown. Most of these data still agree well with the GC-MS. This is because (1) using a S/N ) 3 as a detection limit is rather conservative, (2) the variation in the background is also influenced by trends in the background measurement, which increases the detection limit, even though short-term measurements might have lower detection limits, and (3) in Figure 5 an average over the GC-MS acquisition time is shown. Data below the detection limit for a 2-min measurement could be averaged longer to achieve smaller detection limits and therefore also data below the 2-min detection limit are useful and shown in Figure 5. At this point, the PTR-MS has better detection limits for all investigated compounds except MEK. Since, the offsets for all compounds are clearly lower than the detection limits, and therefore the stated detection limits seem to be justified for the ambient air measurements presented here. 3.3. What Fraction of the Total Signal Has Been Identified? During NEAQS-ITCT 2004 the PIT-MS acquired full mass spectra in the range from 25-160 amu. All masses that can be attributed to specific compounds or classes of compounds are summed up in Figure 6b, which shows a time series for the second half of the study period. Also shown in Figure 6b is the sum of all ion signals that cannot be attributed unambiguously to specific compounds. The total detected VOC signal in Figure 6b reaches a maximum of about 32 ppbv, and only ∼5 ppbv of this maximum VOL. 39, NO. 14, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

5395

FIGURE 7. (a) CID measurement of pure acetone. (b) CID measurement of pure propanal. (c) One CID measurement of mass 59 of ambient air taken during the cruise. The measurement time is on the top axis and the FNF amplitude on the bottom axis.

FIGURE 8. (a) The time series of the mixing ratio of the sum of acetone and propanal measured during the first leg of the NEAQS-ITCT cruise with PIT-MS. (b) Ratio of mass 31/mass 41 (the product ions of mass 59 fragmentation) from the CID measurements vs time. The solid lines show the same ratios that were determined for pure acetone and propanal. (c) Ratio acetone/(acetone + propanal) determined from the CID method and from the GC-MS. is unidentified. These high mixing ratios were found in the Boston harbor. An average of all calibration factors were used for the quantification of the unidentified signals. From the totals shown in Figure 6b the identified fraction of the signal can be calculated, and the result is shown in Figure 6a. The identified fraction is around 90% of the total measured VOC signal for most of the time (average ) 86 ( 10%). The unidentified fraction can be as large as 30%. This occurred mainly in clean conditions when low mixing ratios were measured. During such times, signals are summed up that are around the detection limits resulting in large uncertainties. Such measurements took place many hours downwind of the emission sources (anthropogenic emissions from Boston and New York, or biogenic emissions from the forested areas in Maine and New Hampshire), and the highly reactive VOCs were already removed from the air mass before they reached the ship. Very close to strong VOC sources, closer than during this experiment, such as biomass burning, forests, and cities, more reactive compounds can be expected that could contribute to a larger unidentified fraction. Close to such 5396

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 14, 2005

emission sources, the higher time resolution and the improved selectivity of PIT-MS compared to PTR-MS will be of even greater importance. The lower sensitivity of PIT-MS will be less critical than in the measurements presented here. It should be mentioned again that most VOCs are detectable with PIT-MS except formaldehyde, alkanes, ethylene and acetylene, and most chlorofluorocarbons. We are able to identify the majority of the detectable VOCs, but still a significant part remains unidentified. Important atmospheric oxidation products of the identified VOCs that are expected downwind of cities and forested areas have not been identified and are likely to contribute to the unidentified fraction of the signal. The analytical capabilities of PIT-MS should help to identify some of these oxidation products and therefore extend our understanding of atmospheric processes. In the following section, the method for VOC identification with PIT-MS is described. 3.4. VOC Identification with CID. As described earlier, the ion trap has extended analytical capabilities compared to the quadrupole mass spectrometer used in conventional

PTR-MS (5, 16). By use of PIT-MS, CID measurements were performed every 73 min on mass 59 during most of the cruise. Both acetone and propanal are detected at this mass and are thus indistinguishable in PTR-MS. Acetone typically is much higher than propanal (17), and thus this mass is often associated with acetone only. Ions with mass 59 were isolated in the ion trap and subsequently excited and fragmented using a FNF at increasing amplitudes. In parts a and b of Figure 7, the CID fragmentation patterns of pure acetone and propanal are shown, which were measured before the field campaign. The relative abundances of the precursor ion, mass 59, and the product ions, mass 31 and mass 41, are plotted. The ratio of mass 31/mass 41 at higher FNF amplitudes is substantially different for the two compounds (0.6 for propanal and 2.3 for acetone), and this difference can be used to estimate the contribution from the two compounds to the total signal (5, 16). Figure 7c shows one typical CID measurement of mass 59 of air measured during NEAQSITCT2004, which was sampled during an episode with high acetone mixing ratios. The top axis is the time of the measurement and the bottom axis shows the respective amplitude of the CID FNF field. Each CID measurement in the mode used here takes about 3 min. To reduce the time for CID, fewer or only one sufficiently high amplitude could be chosen to determine the fragment ions. The fragmentation pattern of mass 59 in most ambient air measurements (Figure 7c) looks similar to that from acetone (Figure 7a). Figure 8a shows part of the time series of the sum of acetone and propanal measured with PIT-MS. The ratio of mass 31/mass 41, averaged over the FNF amplitudes 0.12, 0.14, and 0.16 V, is plotted vs time in Figure 8b. Each point in Figure 8b represents one CID measurement as presented in Figure 7c. For almost the entire measurement period the observed mass 31/mass 41 ratios are very close to that of acetone. This shows that the signal measured on mass 59 can be attributed mainly to acetone with a small contribution of propanal. This is expected, because of the longer atmospheric lifetime and higher emission of acetone compared to propanal (17). The error bars in Figure 8b are determined by the measured atmospheric mixing ratios and are larger, when clean air masses were sampled. In Figure 8c the ratio acetone/(acetone + propanal) calculated using the mass 31/mass 41 ratios from the CID measurements is compared with acetone/(acetone + propanal) determined separately from the GC-MS measurements. Again it can be seen that most of the mass 59 signal can be attributed to acetone except in the few periods when the contribution of propanal was enhanced to about 20%. The CID patterns of pure acetone and propanal were determined in laboratory experiments before the field campaign. These were assumed to be constant throughout the field campaign. Such CID patterns may change under the variable conditions of field experiments and should be measured during future field campaigns to improve the accuracy of the CID method. Even though the CID results have some uncertainties at this point, these results demonstrate the potential value of the CID method for the identification of VOCs or for the separation of the contribution of two compounds to a single mass signal. PIT-MS may prove especially useful in the analysis of highly polluted air masses, in which case the identification by PTR-MS becomes problematic, but the higher detection limit of PIT-MS is not an issue.

Acknowledgments This work was funded by the National Science Foundation (Project ATM-0207587). We acknowledge useful discussions with Peter Prazeller and Michael Alexander of the Pacific Northwest National Laboratory. We also acknowledge Paul

Murphy from the NOAA Aeronomy Laboratory for writing the LabView code that made automated measurements possible.

Literature Cited (1) )Hewitt, C. N. Reactive Hydrocarbons in the Atmosphere; Academic Press: San Diego, 1999. (2) Lindinger, W.; Fall, R.; Karl, T. Environmental, food and medical applications of proton-transfer-reaction mass spectrometry (PTR-MS). Adv. Gas-Phase Ion Chem. 2001, 4, 1-48. (3) Warneke, C.; De Gouw, J. A.; Kuster, W. C.; Goldan, P. D.; Fall, R. Validation of atmospheric VOC measurements by protontransfer-reaction mass spectrometry using a gas-chromatographic preseparation method. Environ. Sci. Technol. 2003, 37, 2494-2501. (4) de Gouw, J. A.; Goldan, P. D.; Warneke, C.; Kuster, W. C.; Roberts, J. M.; Marchewka, M.; Bertman, S. B.; Pszenny, A. A. P.; Keene, W. C. Validation of proton-transfer reaction-mass spectrometry (PTR-MS) measurements of gas-phase organic compounds in the atmosphere during the New England Air Quality Study (NEAQS) in 2002. J. Geophys. Res. 2003, 108. (5) Prazeller, P.; Palmer, P. T.; Boscaini, E.; Jobson, T.; Alexander, M. Proton-transfer reaction ion trap mass spectrometry. Rapid Commun. Mass Spectrom. 2003, 17, 1593-1599. (6) Warneke, C.; de Gouw, J. A.; Lovejoy, E. R.; Murphy, P.; Kuster, W. C.; Fall, R. Development of Proton-Transfer Ion Trap-Mass Spectrometry (PIT-MS): On-line Detection and Identification of Volatile Organic Compounds in Air. J. Am. Soc. Mass Spectrom. 2005, in press. (7) Lindinger, W.; Hansel, A.; Jordan, A. On-line monitoring of volatile organic compounds at pptv levels by means of ProtonTransfer-Reaction Mass Spectrometry (PTR-MS): Medical Applications, food control and environmental research. Int. J. Mass Spectrom. Ion Proc. 1998, 173, 191-241. (8) de Gouw, J. A.; Warneke, C.; Karl, T.; Eerdekens, G.; van der Veen, C.; Fall, R. Sensitivity and Specificity of Atmospheric Trace Gas Detection by Proton-Transfer-Reaction Mass Spectrometry. Int. J. Mass Spectrom. 2003, 223-224, 365-382. (9) Warneke, C.; van der Veen, C.; Luxembourg, S.; de Gouw, J. A.; Kok, A. Measurements of benzene and toluene in ambient air using proton-transfer-reaction mass spectrometry: calibration, humidity dependence, and field intercomparison. Int. J. Mass Spectrom. 2001, 207, 167-182. (10) Lovejoy, E. R.; Wilson, R. R. Kinetic studies of negative ion reactions in a quadrupole ion trap: absolute rate coefficients and ion energies. J. Phys. Chem. A 1998, 102, 2309-2315. (11) Curtius, J.; Froyd, K. D.; Lovejoy, E. R. Cluster ion thermal decomposition (I): Experimental Kinetics Study and ab initio calculations for HSO4-(H2SO4)x(HNO3)y. J. Phys. Chem. A 2001, 105, 10867-10873. (12) Cheng, L.; Wang, T. C. L.; Ricca, T. C.; Marshall, A. G. Phasemodulated stored waveform inverse Fourier transform exitation for trapped ion mass spectrometry. Anal. Chem. 1987, 59, 9-454. (13) Goldan, P. D.; Kuster, W. C.; Williams, E. J.; Fehsenfeld, F. C. Nonmethane hydrocarbon measurements during the 2002 New England Air Quality study. J. Geophys. Res. 2004, 109, doi:10.1029/ 2003JD004455. (14) Bakwin, P. S.; Hurst, D. F.; Tans, P. P.; Elkins, J. W. Anthropogenic sources of halocarbons, sulfur hexafloride, carbon monoxide, and methane in the southeastern United States. J. Geophys. Res. 1997, 102, 15915. (15) Holzinger, R.; Lee, A.; Paw U, K. T.; Goldstein, A. H. Observations of oxidation products above a forest imply biogenic emissions of very reactive compounds. Atmos. Chem. Phys. Discuss. 2004, 4, 5345-5365. (16) Warneke, C.; Rosen, S.; Lovejoy, E. R.; de Gouw, J. A.; Fall, R. Two additional advantages of proton-transfer ion trap mass spectrometry. Rapid Commun. Mass Spectrom. 2004, 18, 133-134. (17) Williams, J.; Po¨schl, U.; Crutzen, P. J.; Hansel, A.; Holzinger, R.; Warneke, C.; Lindinger, W. An Atmospheric Chemistry Interpretation of Mass Scans obtained from a Proton-Transfer Mass Spectrometer flown over the Tropical Rainforest of Surinam. J. Atmos. Chem. 2001, 38, 133-166.

Received for review March 28, 2005. Revised manuscript received May 17, 2005. Accepted May 18, 2005. ES050602O VOL. 39, NO. 14, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

5397