Environ. Sci. Technol. 2003, 37, 4664-4671
Hydroxyaldehyde Products from Hydroxyl Radical Reactions of Z-3-Hexen-1-ol and 2-Methyl-3-buten-2-ol Quantified by SPME and API-MS FABIENNE REISEN,† SARA M. ASCHMANN, R O G E R A T K I N S O N , * ,†,‡,§ A N D J A N E T A R E Y * ,† Air Pollution Research Center, University of California, Riverside, California 92521
Hydroxyaldehyde products of the OH radical-initiated reactions (in the presence of NO) of two volatile vegetative emissions, Z-3-hexen-1-ol and 2-methyl-3-buten-2-ol, were examined to assess the qualitative and quantitative potential of two analysis techniques (1) sampling by SolidPhase MicroExtraction (SPME) with on-fiber derivatization followed by gas chromatographic analyses and (2) in situ analysis by negative ion mode atmospheric pressure ionization mass spectrometry (API-MS). The compounds were chosen because reaction mechanisms predict hydroxyaldehyde products, and reliable coproduct yield data are available. The API-MS analyses showed product ion peaks attributed to the NO2- adducts of 3-hydroxypropanal and dihydroxynitrates from Z-3-hexen-1-ol, and a formation yield of 3-hydroxypropanal of 44% was derived. Product ion peaks attributed to NO2- adducts of glycolaldehyde [HOCH2CHO], 2-hydroxy-2-methylpropanal [(CH3)2C(OH)CHO], and dihydroxynitrates were observed by API-MS from 2-methyl3-buten-2-ol, and a formation yield of 2-hydroxy-2methylpropanal of 16% was obtained. In experiments with SPME sampling, the formation yields of hydroxycarbonyls measured as their oxime derivatives were as follows: from Z-3-hexen-1-ol, propanal, 56 ( 8%; 3-hydroxypropanal, 101 ( 24%; and from 2-methyl-3-buten-2-ol, 2-hydroxy-2methylpropanal, 31 ( 4%. Both the API-MS and SPME analyses provided product information, and hydroxycarbonyl yields from the SPME data are in reasonable agreement with previously measured formation yields of coproducts.
Introduction Gas-phase reactions of volatile organic compounds (VOCs) present in the troposphere, whether with OH radicals, NO3 radicals, or O3, can produce products more oxidized than the original VOC (1). Studies carried out over the past few years show that the atmospheric reactions of VOCs can lead to the formation of 1,4-hydroxycarbonyls (2-5) and hy* Corresponding author phone: (909)787-3502; e-mail: janet.arey@ ucr.edu (J.A.) and phone: (909)787-4191; e-mail:
[email protected] (R.A.). † Also Interdepartmental Graduate Program in Environmental Toxicology. ‡ Also Department of Environmental Sciences. § Also Department of Chemistry. 4664
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 20, 2003
droxyaldehydes (6-10) in significant yield. Although combined gas chromatography-mass spectrometry (GC-MS) and gas chromatography with flame ionization detection (GCFID) are generally convenient means of identifying and quantifying VOC reaction products, these particular classes of carbonyl-containing compounds do not typically elute from GC columns without prior derivatization (2, 7, 8, 10, 11). Recently the use of Solid-Phase MicroExtraction (SPME) fibers (12) coated with O-(2,3,4,5,6-pentafluorobenzyl)hydroxylamine (PFBHA) hydrochloride for the analysis of formaldehyde has been reported (13-15). In this work, SPME fibers coated with PFBHA have been used to allow on-fiber derivatization of hydroxycarbonyls (and other carbonylcontaining compounds) for analysis as their oxime derivatives by GC-FID and GC-MS. The hydroxycarbonyl products formed from the gas-phase reactions of OH radicals with two VOCs emitted from vegetation (7, 8, 16), Z-3-hexen-1-ol [CH3CH2CHdCHCH2CH2OH] and 2-methyl-3-buten-2-ol [(CH3)2C(OH)CHdCH2 ], were identified by GC-MS. Relative response factors for the on-fiber derivatization and GC-FID analysis of the oxime derivatives of a series of carbonylcontaining compounds were measured to allow quantitative analysis of hydroxycarbonyl products for which standards are not available. In addition, the Z-3-hexen-1-ol and 2-methyl-3-buten-2-ol reactions were analyzed by in situ direct air sampling atmospheric pressure ionization mass spectrometry (API-MS) using their NO2- adducts for quantification (4, 5). Z-3-Hexen-1-ol and 2-methyl-3-buten-2-ol were chosen for study because reaction mechanisms predict hydroxyaldehyde product formation (7, 8, 10, 17), and reliable product yields are available for more readily quantified coproducts expected to be formed in equal yield to the hydroxyaldehydes (7, 8, 17).
Experimental Section Experiments were carried out in ∼7000 L Teflon chambers at 296 ( 2 K and 740 Torr total pressure of purified air at ∼5% relative humidity. Each chamber is equipped with two parallel banks of blacklamps for irradiation and with a Teflon-coated fan to ensure rapid mixing of the reactants during their introduction into the chamber. SPME Sampling and Response Factors. Carbonyl products were identified by using on-fiber derivatization with SPME as described by Pawliszyn and co-workers (12-15). A 65 µm poly(dimethylsiloxane)/divinylbenzene (PDMS/DVB) fiber was coated with PFBHA. This involved maximally coating the fiber by a 60 min headspace extraction from 4 mL of an aqueous solution (∼40 mg of PFBHA per 100 mL of water) placed in a 20 mL vial and agitated with a magnetic stirrer. The PFBHA coating of the fiber was carried out under a nitrogen atmosphere to minimize contamination from, for example, acetone in the laboratory air. The coated fiber was then exposed to the reactants/standards in the chamber for 5 min (with the mixing fan on) to form a PFBHA-carbonyl oxime. Note that for asymmetic carbonyls, Z- and E-oximes may be produced. To develop response factors relative to 3-pentanone, chosen to be the internal standard for the chamber reactions, analyses of the carbonyl-containing compounds (aldehydes, ketones, hydroxyketones, and glycolaldehdye) listed in Table 1 were carried out using the on-fiber derivatization technique. For each analysis, two or three carbonyl compounds (one of 10.1021/es034142f CCC: $25.00
2003 American Chemical Society Published on Web 09/10/2003
TABLE 1. GC-FID Responses of Oxime Derivatives Relative to Oxime of 3-Pentanone
compound
SPME/GC-FID response factor relative to that of the oxime of 3-pentanone
3-pentanone, CH3CH2C(O)CH2CH3 3-hexanone, CH3CH2C(O)CH2CH2CH3 3-heptanone, CH3CH2C(O)CH2CH2CH2CH3 3-octanone, CH3CH2C(O)CH2CH2CH2CH2CH3 2-butanone, CH3C(O)CH2CH3 2-pentanone, CH3C(O)CH2CH2CH3 2-hexanone, CH3C(O)CH2CH2CH2CH3 2-heptanone, CH3C(O)CH2CH2CH2CH2CH3 2-octanone, CH3C(O)CH2CH2CH2CH2CH2CH3 2-decanone, CH3C(O)CH2CH2CH2CH2CH2CH2CH2CH3 4-octanone, CH3CH2CH2C(O)CH2CH2CH2CH3 3-methyl-2-pentanone, CH3C(O)CH(CH3)CH2CH3 4-methyl-2-pentanone, CH3C(O)CH2CH(CH3)2 5-methyl-2-hexanone, CH3C(O)CH2CH2CH(CH3)2 2,6-dimethyl-4-heptanone, (CH3)2CHCH2C(O)CH2CH(CH3)2 propanal, CH3CH2CHO butanal, CH3CH2CH2CHOa pentanal, CH3CH2CH2CH2CHOb hexanal, CH3CH2CH2CH2CH2CHOb 2-methylpropanal, (CH3)2CHCHO 2,2-dimethylpropanal, (CH3)3CCHO 2-methylbutanal, CH3CH2CH(CH3)CHO 3-methylbutanal, (CH3)2CHCH2CHO glycolaldehyde, HOCH2CHOc 3-hydroxy-2-butanone, CH3C(O)CH(OH)CH3 3-hydroxy-3-methyl-2-butanone, (CH3)2C(OH)C(O)CH3 1-hydroxy-2-butanone, CH3CH2C(O)CH2OH 4-hydroxy-2-butanone, CH3C(O)CH2CH2OHa 4-hydroxy-4-methyl-2-pentanone, (CH3)2C(OH)CH2C(O)CH3 4-hydroxy-3-hexanone, CH3CH2C(O)CH(OH)CH2CH3 4-hydroxy-3-methyl-2-butanone, CH3C(O)CH(CH3)CH2OHa,d 5-hydroxy-2-pentanone, CH3C(O)CH2CH2CH2OH hydroxyacetone, CH3C(O)CH2OH 1,3-dihydroxy-2-propanone, HOCH2C(O)CH2OH
1.0 1.7 3.9 7.3 1.25 2.4 5.5 9.6 14.2 29.1 3.3 0.74 1.1 8.3 0.16 4.4 7.6; 6.7 16.0 22.3 5.3 2.0 9.5 11.8 18.8 7.3 1.5 5.6 12.2; 12.8 3.1 2.9 10.7; 10.3 15.0 6.4 12.3
a Two independent measurements. b Relative to the average butanal response factor of 7.2. c Obtained from coated SPME/GC-FID analysis of 3 irradiated CH3ONO-NO-2-methyl-3-buten-2-ol-air mixtures, with 4-hydroxy-3-hexanone and (in one experiment) 1-hydroxy-2-butanone added after the irradiation as an internal standard(s) and using our previously measured glycolaldehyde formation yield of 58 ( 4% [a weighted average of the measured formation yields of glycolaldehyde and its coproduct acetone (8)] and taking into account the small loss of glycolaldehyde (300 nm (18), and NO was added to the reactant mixtures to suppress the formation of O3 and NO3 radicals (18). Each experiment consisted of a single irradiation resulting in CH3C(O)R > CH3CH2C(O)R for a given carbon number are consistent with the higher reactivity of aldehydes and the expected steric effect (from replacing the aldehydic H with larger alkyl groups) on the transition state leading to the formation of the tetrahedral carbinolamine intermediate (see Figure 2). The nearly one-to-one ratio of the Z- and E-isomers (these ratios varied from 0.7 to 1.2 for straight chain aldehydes and ketones) suggests that the dehydration step is less influenced by the size of R or R′. While the FID response generally increases proportionately with carbon number (21), this effect is obviously too small to explain the data shown in Figure 1. The observed increased response with carbon number in a given series is difficult to understand if, as has been postulated for formaldehyde sampling, all sorption sites of the fiber are occupied by the PFTBA and the gaseous carbonyl is bound to the fiber by forming the oxime (13, 14). Although the mixed coating PDMS-DVB fiber is expected to extract mainly via adsorption (15, 22), because it has PDMS supporting the DVB on the fused-silica rod of the fiber it is possible that some absorption also occurs. Our data can be rationalized by assuming that in addition to the reaction of gaseous carbonyl with the PFBHA on the surface, some carbonyl sorbs and then reacts with the derivatization agent immobilized on the surface. The increasing sorption equilibrium constant with an increasing carbon number would then explain the enhanced response with the carbon number in a given series. It should be noted that small amounts of underivatized carbonyls were also observed in some instances. Inspection of the data shown in Table 1 indicates that response decreases with branching [for example, from butanal (7.2) to 2-methylpropanal (5.3) and from pentanal (16.0) to 3-methylbutanal (11.8), 2-methylbutanal (9.5), and 2,2-dimethylpropanal (2.0)]. The decrease in response with
branching can again be explained by steric effects on the rate of formation of the carbinolamine intermediate. For example, the C5-aldehyde trimethylacetaldehyde has a response only twice that of 3-pentanone, while pentanal has a relative response factor of 16 (see Table 1). Additionally, steric effects were observed on the ratio of the oxime isomers from trimethylacetaldehyde, with one isomer being favored by a factor of >5. Also evident from inspection of the data shown in Table 1, the response of the carbonyl increases significantly with hydroxyl substitution. Hydroxyl substitution may increase the rate of nucleophilic attack on the carbonyl by the lonepair electrons of the amine (19c). As shown in Table 2, the SPME/GC-FID response factors for hydroxycarbonyls are consistently higher than those for the corresponding carbonyl with the OH group replaced by a methyl group, by a factor of 5.1 (with an uncertainty of a factor of ∼2). The enhanced response due to substitution of a methyl with an alcohol group may also be seen from the solid symbols in Figure 1. Note that a similar approximately 5-fold enhancement occurs for hydroxyacetone [CH3C(O)CH2OH] relative to 2-butanone (C4- solid and open diamonds, respectively), 4-hydroxy-2-butanone relative to 2-pentanone (C5- solid and open diamonds, respectively), and 1-hydroxy2-butanone relative to 3-pentanone (C5- solid and open circles, respectively). In contrast, the enhancement in the response factors of glycolaldehyde [HOCH2CHO] relative to propanal (C3- solid and open squares, respectively) is 4.3 and that for 5-hydroxy-2-pentanone relative to 2-hexanone (C6- solid and open diamonds, respectively) is only 2.7, suggesting there is an upper limit to the response factors under our sampling conditions. The response for 2-decanone (see Table 1; not shown in Figure 1) falls well below the line for the CH3C(O)R compounds, again indicating that a maximum relative response value is reached. It has been assumed that formation of the carbinolamine is much slower than diffusion of the compounds to the fiber (14). However, VOL. 37, NO. 20, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4667
TABLE 3. Product Formation Yields for the OH Radical-Initiated Reaction of Z-3-Hexen-1-ol in the Presence of NO product
from ref 7
SPME/GC-MS
SPME/GC-FIDa
Tenax/GC-FIDa
propanal 3-hydroxypropanal CH3CH2CH(OH)CH(OH)CH2CHO dihydroxynitrate(s) (MW 179)
74.6 ( 6.7b 48+48-24e,f observedf observedf
observedc observedc trace observedc
56 ( 8d 101 ( 24g
72 ( 18
API-MSa 44h,i observedi observedi
a Product yields corrected for secondary reactions with OH radicals using rate constants (in units of 10-11 cm3 molecule -1 s-1) of Z-3-hexen-1-ol, 10.8 (23); propanal, 2.0 (24); and 3-hydroxypropanal, 3.0 [estimated based on data in ref 25]. Corrections were
