Article pubs.acs.org/est
Formation Yields of C8 1,4-Hydroxycarbonyls from OH + n‑Octane in the Presence of NO Sara M. Aschmann,† Janet Arey,*,†,‡ and Roger Atkinson*,† †
Air Pollution Research Center and ‡Department of Environmental Sciences, University of California, Riverside, California 92521, United States ABSTRACT: 1,4-Hydroxycarbonyls are major products of the OH radicalinitiated reactions of ≥C5 n-alkanes in the presence of NO. However, because of a lack of commercially available standards of 1,4-hydroxycarbonyls and difficulties in using gas chromatography for their analysis without prior derivatization, quantification of 1,4-hydroxycarbonyls in OH + alkane reactions has proven difficult. We have used an annular denuder coated with XAD resin and further coated with O-(2,3,4,5,6-pentafluorobenzyl)hydroxylamine for in situ derivatization of the 1,4-hydroxycarbonyls formed from the OH + n-octane reaction in the presence of NO. Quantification was achieved by using 2,5-hexanedione as an internal standard. Formation yields for (7-hydroxy-4-octanone + 6-hydroxy-3octanone + 5-hydroxy-2-octanone) and of 4-hydroxyoctanal of 61 ± 11% and 2.1 ± 0.5%, respectively, were obtained. When combined with previously measured or estimated formation yields for octyl nitrates and hydroxyoctyl nitrates, 93 ± 15% of the overall reaction products are accounted for, indicating that no additional reaction pathways remain to be identified.
■
INTRODUCTION Alkanes comprise ∼60% of reformulated gasolines in California and ∼40% of nonmethane volatile organic compounds emitted in vehicle exhaust.1,2 In the atmosphere, alkanes react dominantly with hydroxyl (OH) radicals, by H-atom abstraction from the various C−H bonds to form alkyl radicals (R•), which rapidly react with O2 to form alkyl peroxy (RO2•) radicals.3,4 OH + RH → H 2O + R•
sequent to alkoxy radical isomerizations (Scheme 1), with 1,4hydroxycarbonyl formation accounting for ≥50% of the total first-generation products for the C5−C8 n-alkanes.6−8 However, quantification of 1,4-hydroxycarbonyls has proven difficult because only 5-hydroxy-2-pentanone (formed from OH + npentane6−8) is commercially available and 1,4-hydroxycarbonyls cannot be readily analyzed by gas chromatography unless they have been derivatized.7,8 We previously used solid-phase microextraction (SPME) fibers precoated with O-(2,3,4,5,6-pentafluorobenzyl)hydroxylamine (PFBHA) for on-fiber derivation of carbonyl groups, followed by combined gas chromatography−mass spectrometry (GC-MS) to identify 1,4-hydroxycarbonyls from n-pentane, n-hexane, n-heptane, and n-octane.8 Quantification by gas chromatography with flame ionization detection (GCFID) necessitated the use of SPME/GC-FID response factors involving the collection and derivatization efficiency on the PFBHA-coated SPME fiber together with the GC-FID response factor for the oximes of the 1,4-hydroxycarbonyls,8 and since only 5-hydroxy-2-pentanone was available as a standard the derived formation yields from the n-hexane, nheptane and n-octane reactions had significant associated uncertainties.8 In this work, we have extended our previous study8 by using a denuder coated with XAD resin and further coated with PFBHA for in situ derivatization of the 1,4hydroxycarbonyls from the OH + n-octane reaction. Quantification was achieved by using 2,5-hexanedione as an
(1)
R• + O2 + M → RO2• + M
(2)
•
In the presence of NO, RO2 radicals react with NO to form alkyl nitrates (RONO2) [reaction 3a] or an alkoxy radical (RO•) plus NO2 (reaction 3b).3,4 RO2• + NO + M → RONO2 + M
(3a)
RO2• + NO → RO• + NO2
(3b)
Under atmospheric conditions, alkoxy radicals react by unimolecular decomposition, unimolecular isomerization (generally through a 6-member transition state), and by reaction with O2,3−5 noting that not all of these potential reactions may be feasible for a given alkoxy radical. At room temperature, except for the 3-pentoxy radical, isomerization is the dominant reaction for the alkoxy radicals formed from ≥C5 n-alkanes, leading ultimately to formation of 1,4-hydroxyalkyl nitrates and 1,4-hydroxycarbonyls, as shown in Scheme 1 for the reactions subsequent to formation of the 2-octyl peroxy radical (firstgeneration products are shown in boxes).3−8 Indeed, the only products observed from ≥C7 n-alkanes at room temperature are the alkyl nitrates formed from reaction 3a and the 1,4hydroxyalkyl nitrates and 1,4-hydroxycarbonyls formed sub© 2012 American Chemical Society
Received: Revised: Accepted: Published: 13278
October 9, 2012 November 27, 2012 December 3, 2012 December 7, 2012 dx.doi.org/10.1021/es3041175 | Environ. Sci. Technol. 2012, 46, 13278−13283
Environmental Science & Technology
Article
Scheme 1a
a
Observed products are shown in boxes.
onto a 30 m DB-1701 megabore column, initially held at −40 °C and then temperature programmed to 250 °C at 8 °C min−1. Samples were also collected after the reaction for 30 min at 15.9 L min−1 using an XAD-coated denuder, further coated with PFBHA prior to sampling to derivatize carbonyls to their oximes, and extracted as described previously.11 The sampling entrance of the denuder extended into the chamber, thereby eliminating any sampling line upstream of the denuder. The extracts were analyzed by combined gas chromatography−mass spectrometry in positive chemical ionization (PCI GC-MS) mode, and by GC-FID, with both analyses using DB-5 columns (60 m for the GC-MS analyses and 30 m for the GC-FID analyses). The GC-MS analyses used an Agilent 5973 Mass Selective Detector operated in the scanning mode with methane as the reagent gas. Each carbonyl group derivatized to an oxime adds 195 mass units to the compound’s molecular weight, and methane-PCI gives characteristic protonated molecules ([M + H]+) and smaller adduct ions at [M + 29]+ and [M + 41]+.8,11 Methane-PCI of oximes also gives [M + H − 198]+ fragment ions,11 and derivatized hydroxycarbonyls exhibit strong [M + H − H2O]+ fragment ions.8 The presence of weak fragment ions attributed to minor charge-transfer ionization,8 at m/z 253, 267, and 281, has been suggested to be diagnostic of 2-, 3- and 4-ketones, respectively, and the presence of a m/z 239 fragment ion to be diagnostic of aldehydes.8 In 4 of the 5 experiments, 2,5-hexanedione was added prior to reaction. In these 4 experiments, replicate GC-FID analyses
internal standard, with the concentration of 2,5-hexanedione being measured by GC-FID.
■
EXPERIMENTAL METHODS Experiments were carried out at 298 ± 2 K in the presence of 735 Torr of purified air containing 3.3 × 1017 molecules cm−3 of H2O (∼50% relative humidity) in a ∼7000 L Teflon chamber equipped with two parallel banks of black lamps for irradiation. OH radicals were generated from the photolysis of CH3ONO at wavelengths >300 nm, and NO was included in the reactant mixtures to suppress formation of O3 and hence of NO3 radicals. In all cases the light intensity corresponded to an NO2 photolysis rate of 0.14 min−1. The initial concentrations (molecules cm−3) of CH3ONO, NO and n-octane were ∼4.8 × 1013, ∼4.8 × 1013, and (2.32− 2.44) × 1013, respectively. 2,5-Hexanedione (∼2.4 × 1012 molecules cm−3) was added to the chamber prior to reaction or (in one experiment) after reaction to serve as an internal standard. Irradiations (with a single irradiation period per experiment) were carried out for 4−7 min, resulting in 14.0− 20.3% consumption of the initially present n-octane. Water vapor was present to minimize cyclization and dehydration of the 1,4-hydroxycarbonyls.8−10 The concentrations of n-octane and 2,5-hexanedione were measured during the experiments by gas chromatography with flame ionization detection (GC-FID). Gas samples of 100 cm3 volume were collected from the chamber onto Tenax-TA adsorbent, with subsequent thermal desorption at ∼205 °C 13279
dx.doi.org/10.1021/es3041175 | Environ. Sci. Technol. 2012, 46, 13278−13283
Environmental Science & Technology
Article
of Tenax samples were conducted before and after the single irradiation period to measure the concentrations of n-octane and 2,5-hexanedione in the chamber, and sampling onto the PFBHA-coated denuder started immediately after the lights were turned off. In the other experiment, 2,5-hexanedione was added to the chamber after the single irradiation period, and in this experiment replicate GC-FID analyses of Tenax samples were conducted before and after the single irradiation period (to measure the concentration of n-octane pre- and postreaction) and again after addition of 2,5-hexanedione to the chamber (to measure the concentration of 2,5-hexanedione in the chamber). Sampling onto the PFBHA-coated denuder commenced after addition of the 2,5-hexanedione, this being 80 min after the end of the reaction. The C8 1,4-hydroxycarbonyls and 2,5-hexanedione were quantified as their mono- and dioximes, respectively, from replicate GC-FID analyses of the extracts from the PFBHAcoated denuder samples. 2,5-Hexanedione served as an internal standard, and corrections for the differing FID responses of the oximes of the hydroxycarbonyls and the dioximes of 2,5hexanedione were made using the effective carbon numbers (ECNs) of Scanlon and Willis12 and Nishino et al.13 We used 2,5-hexanedione because (a) it is fairly unreactive toward OH radicals (see below)14 and hence could be included in reactant mixtures prereaction, thereby allowing denuder sampling to commence immediately after an irradiation period, (b) with the procedure employed here, we observe efficient derivatization resulting in only diderivatives of dicarbonyls, and (c) 2,5hexanedione can be readily quantified by GC-FID analyses of gas samples collected onto Tenax solid adsorbent (without prior derivatization), thus allowing the concentration of 2,5hexanedione to be accurately measured postreaction. We have previously shown that when using PFBHA-coated denuders and our analysis procedure to measure 5-methyl-2-hexanone, 2octanone, and 2-decanone as their oximes and 2,5-hexanedione as its dioximes, we get, on a relative basis after correcting for differing ECNs, equal molar responses to within 10%.15 The chemicals used and their stated purity levels were: noctane (99+%) and 2,5-hexanedione (98+%), Aldrich; O(2,3,4,5,6-pentafluorobenzyl)hydroxylamine hydrochloride (99+%), Alfa Aesar; and NO (≥99.0%), Matheson Gas Products. Methyl nitrite was prepared as described by Taylor et al.16 and stored at 77 K under vacuum.
Figure 1. Top: Methane-PCI GC-MS total ion chromatogram (TIC) of an extract of a PFBHA-coated denuder sample from an irradiated CH3ONO−NO−n-octane−2,5-hexanedione−air mixture. The peak assignments are as follows:8 peaks 1 and 2, oximes of 7-hydroxy-4octanone; peaks 3 and 4, oximes of 6-hydroxy-3-octanone; peaks 5 and 6, oximes of 5-hydroxy-2-octanone; peak 7, oxime of unidentified C8hydroxyketone; peak 8, oxime of 4-hydroxyoctanal; peak 9, dioxime of methylglyoxal; peaks 10 and 11, dioximes of CH3C(O)CH2CHO formed from OH + 2,5-hexanedione; and peaks 12, dioximes of C8dicarbonyls formed from OH + C8-hydroxycarbonyls. Bottom, PCI mass spectrum of peak 6 (5-hydroxy-2-octanone), exhibiting [M + H]+, smaller adduct ions at [M + 29]+ and [M + 41]+, and [M + H − H2O]+ and [M + H − 198]+ fragment ions. The m/z 181 ion is attributed to the pentafluorotropylium ion [C7F5H2]+, generally the base peak in the electron impact mass spectra of PFBHA oximes, but absent in their isobutane-PCI mass spectra.8 The presence of the minor fragment ion at 253 u is indicative of a 2-ketone.8.
■
small fragment ion at m/z = 253 identifies the spectrum in Figure 1 as that of a 2-octanone.8 Also labeled on the chromatogram in Figure 1 (top) are the dioximes of 2,5-hexanedione (no mono-oximes were present) and the dioximes of dicarbonyls of molecular weight 72, 142 and, when 2,5-hexanedione was initially present in the reactant mixtures, 86. On the basis of previous studies in this laboratory, the molecular weight 72 and 86 products were methylglyoxal (presumably a second-generation product) and 3-oxo-butanal [CH3C(O)CH2CHO], respectively, and the molecular weight 142 dicarbonyls are attributed to C8-dicarbonyls formed as second-generation products from OH + C8-hydroxycarbonyls, after H-atom abstraction from −CH(OH)− groups. The C8 1,4-hydroxycarbonyls were quantified as their monooximes from GC-FID analyses of extracts of PFBHA-coated denuder samples. Because the GC peaks of the hydroxyoctanones (peaks 1−6 in Figure 1) were not baseline resolved, they were integrated as the sum of the oximes of 7-hydroxy-4octanone + 6-hydroxy-3-octanone + 5-hydroxy-2-octanone. The much smaller peak identified as the mono-oxime of 4-
RESULTS An example of a PCI GC-MS analysis of an extract of a PFBHA-coated denuder sample collected from the chamber after reaction is shown in Figure 1. Note that in general there are 2 mono-oximes formed from each hydroxycarbonyl, the E and Z isomers, although one may dominate as is the case for 4hydroxyoctanal. Peaks 1−8 are mono-oximes of products of molecular weight 144. These are attributed to 5-hydroxy-2octanone (peaks 5 and 6), 6-hydroxy-3-octanone (peaks 3 and 4), 7-hydroxy-4-octanone (peaks 1 and 2), and 4-hydroxyoctanal (peak 8) based on the GC retention time pattern and the mass spectra reported by Reisen et al.8 The mono-oxime of 4-hydroxyoctanal was, as expected, formed in much lower yield than the hydroxyoctanones.8 The lower panel of Figure 1 shows the methane-PCI mass spectrum of peak 6, identified as 5hydroxy-2-octanone. Ion peaks common to all the C8hydroxycarbonyl oximes include m/z = 340, 368, 380, and 322 corresponding respectively to [M + H]+, [M + 29]+, [M + 41]+, and [M + H − H2O]+ for M = 339 = 144 + 195. The 13280
dx.doi.org/10.1021/es3041175 | Environ. Sci. Technol. 2012, 46, 13278−13283
Environmental Science & Technology
Article
Table 1. Products of the OH Radical-Initiated Reaction of nOctane in the Presence of NO and Their Formation Yields
hydroxyoctanal was quantified separately. As noted, most experiments were carried out with 2,5-hexanedione present as an internal standard in the initial reactant mixtures (i.e., before reaction). The rate constant for the reaction of OH radicals with 2,5-hexanedione has been measured to be (7.13 ± 0.34) × 10−12 cm3 molecule−1 s−1 at 298 K,14 and hence 2,5hexanedione is of comparable reactivity as n-octane toward OH radicals (k(OH + n-octane) = 8.11 × 10−12 cm3 molecule−1 s−1 at 298 K3). On the basis of GC-FID analyses of Tenax samples collected before and after reaction, the percentages of 2,5-hexanedione and n-octane which were consumed by reaction were indeed similar in each of the four experiments in which 2,5-hexanedione was added initially. However, because the added 2,5-hexanedione was only ∼10% of the n-octane concentration, products formed from 2,5-hexanedione were minor. The major expected product of OH + 2,5-hexanedione is CH3C(O)CH2CHO, formed after H-atom abstraction from the CH2 groups, and dioximes of this mw 86 dicarbonyl were observed in all of the PCI GC-MS analyses when 2,5hexanedione was present (peaks 10 and 11 in Figure 1), but not in the experiment in which 2,5-hexanedione was added after the reaction. The concentrations of the C8-hydroxycarbonyls in the chamber after reaction were obtained by ratioing the GC-FID peaks areas of the mono-oximes of the C8-hydroxycarbonyls to those of the dioximes of 2,5-hexanedione, taking into account the differing FID responses of the mono-oximes of the C8hydroxycarbonyls (ECN = 15.4212,13) and of the dioximes of 2,5-hexanedione (ECN = 22.3412,13). Two to five replicate postreaction GC-FID analyses of the extract of each PFBHAcoated denuder sample were conducted, with good reproducibility (to within 9%) relative to the internal standard. The concentrations of the C8-hydroxycarbonyls measured in the chamber after the reaction were corrected for reaction with OH radicals, using rate constant ratios of k(OH + hydroxyoctanones)/k(OH + n-octane) = 2.63 and k(OH + 4hydroxyoctanal)/k(OH + n-octane) = 5.12, based on the rate constants estimated by Reisen et al.8 for OH + 5-hydroxy-2octanone, 6-hydroxy-3-octanone, 7-hydroxy-4-octanone and 4hydroxyoctanal of 2.34 × 10−11, 2.28 × 10−11, 2.24 × 10−11, and 4.46 × 10−11 cm3 molecule−1 s−1, respectively, combined with that for n-octane of 8.71 × 10−12 cm3 molecule−1 s−1 (from Atkinson,17 to be consistent with the estimation method used by Reisen et al.8). The estimated rate constants for 5-hydroxy2-octanone, 6-hydroxy-3-octanone and 7-hydroxy-4-octanone are consistent, within the large experimental uncertainties, with the values measured by Baker et al.9 The multiplicative correction factors, F, increase with the rate constant ratio k(OH + hydroxycarbonyl)/k(OH + n-octane) and with the extent of reaction. For the small extents of reaction used here (≤20.3%), the values of F for (7-hydroxy-4-octanone + 6hydroxy-3-octanone + 5-hydroxy-2-octanone) were in the range 1.22−1.34, and those for 4-hydroxyoctanal were in the range 1.45−1.73 (see footnotes c and d of Table 1). Plots of the amounts of C8 1,4-hydroxycarbonyls formed, corrected for reaction with OH radicals, against the amounts of n-octane reacted are shown in Figure 2. The hydroxyoctanone formation yield in the experiment in which 2,5-hexanedione was added after the reaction (MTC-2556) and sampling commenced 80 min after the lights were turned off is lower by a factor of 1.08 ± 0.08 than the hydroxyoctanone yields in the experiments in which 2,5-hexanedione was present initially and sampling started immediately after the lights were turned off. A
molar yield (%)
product 2- + 3- + 4-octyl nitrate hydroxyoctyl nitrates 7-hydroxy-4-octanone + 6-hydroxy-3-octanone + 5-hydroxy-2octanone 4-hydroxyoctanal a
22.6 ± 3.2a 5.4,a 9b 61 ± 11c,d 2.1 ± 0.5c,e
7
Measured yields from Arey et al. The uncertainty in the hydroxyoctyl nitrate yield was estimated to be a factor of ∼2.7 b Estimated (see text). cThis work. Obtained from least-squares analyses of the data from the experiments in which 2,5-hexanedione was initially present (data points denoted by ○ in Figure 2), with formation yields for (7-hydroxy-4-octanone + 6-hydroxy-3-octanone + 5-hydroxy-2-octanone) and 4-hydroxyoctanal of 60.7 ± 4.4% and 2.14 ± 0.30%, respectively, where the indicated error are two least-squares standard deviations. The indicated errors in the table are two leastsquares standard deviations combined with estimated uncertainties in the GC-FID response factors for n-octane and 2,5-hexanedione of ±5% each and an estimated uncertainty in the GC-FID responses for the mono-oximes of the C8-hydroxyoctanones relative to those for the dioximes of 2,5-hexanedione of ±15%. Unweighted averages of the individual yields from the same 4 experiments leads to formation yields for (7-hydroxy-4-octanone + 6-hydroxy-3-octanone + 5-hydroxy-2octanone) and 4-hydroxyoctanal of 60.9 ± 3.8% and 2.03 ± 0.31%, respectively, where the errors are two standard deviations. dFor the small extents of reaction used here (≤20.3%), the values of F for 7hydroxy-4-octanone + 6-hydroxy-3-octanone + 5-hydroxy-2-octanone were in the range 1.22−1.34., and a ±30% change in the rate constant ratio k(OH + hydroxyoctanones)/k(OH + n-octane) would change the multiplicative factor F by
