Environ. Sci. Technol. 1092, 26, 1165-1173
Atmospheric Photochemical Oxidation of l-Octene: OH, 03,and O(3P) Reactions Suzanne E. Paulsontss and John H. Seinfeld*p§
Environmental Engineering Science and Chemical Engineering, California Institute of Technology Pasadena, California 9 1 125 c _ _
8 The photooxidation chemistry of l-octene is examined
in detail. Formation of OH from the 0, reaction was examined with the use of methylcyclohexane as an OH radical tracer. The 0,-l-octene reaction is found to produce, apparently directly, significant quantities of OH, 0.45 & 0.2 on a per molecule of reacted l-octene basis. Almost 10090 of the reacted l-octene could be accounted for as 80 f 10% heptanal, 10 f 6% thermally stabilized Criegee biradical, and 1% hexane. The OH-l-octene reaction was found to produce only 15 f 5% heptanal. The remainder is assumed to result in the formation of alkyl nitrates, and possibly multisubstituted products. A separate experiment examining the 0(3P)-l-octene reaction showed that the sum of hexyloxirane and octanal accounted for -75% of the reacted l-octene. A photochemical model was developed for l-octene oxidation and compared favorably with smog chamber results from NONO,-l-octene experiments, predicting ozone maxima within *25%.
-
Introduction 1-Octeneis an anthropologically derived contributor to the formation of photochemical smog. l-Octene itself has been observed at ambient concentrations ranging from 40 ppt to 48 ppb (1).Further, l-octene is expected to undergo photooxidation in a manner which is similar to that of the larger l-alkenes, including, for example, l-hexene, 1nonene, and 2-ethyl-l-hexene, which have also been observed in the atmosphere (I, 2). Important questions arise with regard to the photochemical oxidation of the long-chain alkenes. Compounds with at least four- or five-carbon backbones may have additional reaction pathways available to them beyond those available to the smaller hydrocarbons. For example, it has been proposed that an alkoxy radical (formed from OH addition to the double bond, addition of 02, and subsequent reaction with NO) with a four-carbon or larger backbone and an available hydrogen atom may undergo isomerization via a six-membered ring (3-6). The ozone reaction is at least as uncertain as the OH reaction (e.g., ref 7). Of particular interest is the formation of OH radical from the O3 reaction with l-octene, as this reaction is currently assumed to be a nearly complete radical sink. b e n t studies of the 0, reaction with tetramethylethylene and isoprene have revealed -70% OH yields from these reactions (8, 9). In addition, an earlier study of propene mdicated an unspecified radical with a yield of 30%; this adid may have been OH (10). Several other studies have the authors to propose OH formation from 0, reactions ?th isobutene, propene, and several terpenes (11,12). It '8,OfParticularand timely interest to determine the OH h l d s from additional alkenes. OH production impacts both the effect of alkenes on urban ozone formation and
''Present Environmental Engineering Science. address: ASP Fellow, Box 3000, National Center for
\
Research, Boulder, CO 80307-3000. 'Chemical Engineering. 0013.~36X/92/0926-1 165$03.00/0
experimental results that have assumed little or no OH formation from the Os-alkene reaction. No studies have been carried out to date examining the atmospheric chemistry of l-octene. In an earlier study in this laboratory, aerosol dynamics and aerosol chemical composition from l-octene were examined (13,14). Aerosol formation and oxidation chemistry of l-heptene by a mixture of OH, O,, and O(,P) have been examined by Grosjean (15), who identified several aldehyde products (dominated by formaldehyde and hexanal), in addition to small quantities of hexanoic acid. We have performed a series of smog chamber experiments examining the photooxidation of l-octene using techniques specific for studying OH, O,, and O(,P) reactions, focusing on identification and quantification of oxygenated products using GC-FID and GC/MS. We have also used l-octene-NO, photooxidation experiments as a tool to refine the overall oxidation mechanism for l-octene. These results are presented in the order O(,P), OH, 03, and the mixed oxidant experiments, so that each section depends only on those that precede it. Experimental System The experiments were performed in a flexible outdoor Teflon smog chamber, with protocols that have been described previously (9,16-18). The initial conditions of the experiments are detailed in Table I. The fully inflated chamber volume was approximately 7 m3 for all experiments except the INOX series, which were carried out in a 60-m3chamber in conjunction with aerosol measurements that have been reported elsewhere (13). The gas-phase species measured on-line include NO, NO, (chemiluminescence), 0, (UV absorption), and 1octene and its major products (GC-FID quantification and GC/MS identification). Chamber temperature, UV radiation, and total solar radiation were measured on-line as well. The GC was calibrated for each experiment using certified cylinders containing mixtures of ultrapure air and l-octene (30.1 ppm) and methylcyclohexane (31.3 ppm). Relative response factors for organic products were formulated by vaporizing small quantities of the liquid compounds into an 8-L Teflon bag with l-octene as an internal standard (17). l-Octene used in these experiments was certified 98% pure (Aldrich) and was used without further purification. We were able to detect a slight impurity, amounting to -1% of the l-octene signal. GC/MS analysis revealed that this impurity was n-octane. n-Octane is relatively unreactive when compared to l-octene (with an OH rate constant that is slower by a factor of 5; see Table 11);by the end of a typical experiment, during which 90% or more of the l-octene had reacted, the n-octane remained at 88% of its original signal. While separation of l-octene and n-octane was not sufficient in some analyses to reliably monitor n-octane, for a few experiments, n-octane provides a rough additional check on the OH formed in the system. GC/MS identification of heptanoic acid, 2-ethylhexanal, and 3-octanone were made by comparison with the National Institute of Standards and Technology library of
0 1992 American Chemical Society
Environ. Sci. Technol., Vol. 26, No. 6, 1992
1165
Table I. Summary of Initial Conditions
a
expt
primary reaction
HC
103P1 lOHl 10H2 lNOXl lNOX2 1NOX3 1NOX4 1NOX5 1NOX6 1NOX7 1031 1032 M1031 MI031 + methylcyclohexane
I-octene + O(3P) I-octene + OH 1-octene + OH 1-octene + NO, I-octene + NO, 1-octene + NO, 1-octene + NO, 1-octene + NO, 1-octene + NO, 1-octene + NO, I-octene + O3 1-octene + O3 1-octene + O3
8.25 1.50 1.81 0.782 1.59 0.77 0.88 0.56 1.46 20.0 14.36 14.4 10.94 60.9
ppm C/ppm NO,.
initial concn, ppm NO NOz CH30N0 0.062 3.49 4.42 0.343 0.166 0.075 0.097 0.134 0.174 1.69
5.5 0.51 1.1 0.180 0.075 0.342 0.049 0.021 0.074 1.86
HC/NOXa
0.120 0.014 0.026 0.005 0.35 0.024 0.28 0.185* 0.292b 0.738
12.0 52.8 14.8 48.2 26.3 47.1 45.1
0.6 1.4
Experiment was terminated before ozone had peaked. 44
I
W
0.0
max concn 03,ppm
0.5
Time (hrs) Flgure 1. NO, NO2, 03, and 1-octene data and simulations for experiment 103P1. Note that O3is plotted with respect to the right-hand axis.
mass spectral data. Identification of hexyloxirane, octanal, 2-octanone, and heptanal were made by comparison to standards obtained from Aldrich.
0(3P)-I-OcteneReaction The reaction between O(3P)and 1-octene, while likely to be of minor importance in the atmosphere, is of significant importance in both the mixed oxidant (1NOXn) and the methyl nitrite (10Hn) experiments (discussed below). Earlier studies of the 0(3P)-alkene reaction have shown that this reaction leads to the formation of oxiranes, resulting from direct addition of the O(3P) atom to the double bond, and aldehydes and/or ketones resulting from a 1,2 hydrogen shift (19). A fraction of many 0(3P)-alkene adducts has also been shown to decompose, forming a variety of organic radicals (19,20). The organic radicals can lead to the formation of OH both in the reaction vessel and in the atmosphere. We carried out one experiment to examine the reaction between O(3P)and 1-octene. The chamber was filled with ultrahigh purity N2, 1-octene and NO2 were added and allowed to mix, and the chamber was exposed to sunlight. The chamber contained approximately 1% residual O2 (17). The 1-octene,NO2,NO, and O3data appear in Figure 1,and the organic products are shown in Figure 2. The summed products, which include octanal, hexyloxirane, 2-octanone, and heptanal, as well as trace quantities of hexane, 2-ethylhexanal, and 3-heptanone, account for 73% of the reacted 1-octene. Octanal and hexyloxirane,
-
1166 Envlron. Scl. Technol., Vol. 26, No. 6, 1992
I
I
I
I
3
4
5
6
e 0.8
2-Ethyl Hexanal
1
0
1.0
I
2
I-Octene Reacted (ppm) Flgure 2. Product concentrations and simulations from experiment 103P1. Note that the combined octanal and hexyloxirane is plotted with respect to the left-hand axis, and the other products with respect to the right-hand axis.
closely related isomers, were not completely separated in any of our analyses and hence are reported as a sum. As expected, octanal and hexyloxirane are the dominant products, arising from the following reactions: P-CHz
CH3(CH215/
CH.dCH&
HC-
+
CH,
(1)
CH3(CHz15'
Addition of O(3P)to the terminal carbon is both less sterically hindered and produces the more stable radical, hence this adduct should be dominant. This adduct can relax or isomerize with a 1,2 hydrogen shift, forming hexyl oxirane and octanal, respectively:
(Hexyl oxirane)
(Octanal)
2-0ctanone, a much less abundant product, is expected to arise from the minor isomer:
9 HCCH3(CHzl/
CH,
-
/"\
HC-CH,
CH3(CHz15' (Hexyl oxiranel
0 +
0
,C-CHI CH3(CHz)5
(3)
(2-Octanone)
There are a number of possible decomposition pathways for the 0(3P)-l-octene adduct, many of which can form heptanal along with other products (23). Heptanal is also formed from the OH and 0,reactions. It should be noted
Gble 11. 1-Octene Oxidation Mechanism Lumped Species lumped alkylperoxy radical, mainly secondary and assumed to react as such lumped alkyl radical resulting from OH attack on the alkyl backbone of heptanoic acid alkyl radical intermediate resulting from reaction of HACRl with NO terminal alkyl radical; six or seven carbons lumped nitroalkyl radical radical products of the 0(3P)-l-octene reaction
MCHR HACRl HACR2 6R1 NCRl OR k(298 K),cm3 molecule-' s-l
EJR, K
rep
(1)4.0 X lo-'' (2) 1.7 x 10-17
-467 1713
e,f g,f
(3) 1.1 x lo-" (4) 7.6 X lo-'' (5) 7.6 X lo-''
-180 -180
h, f 6, f 6, f
(6) 7.6 x
-180
6, f
lo-''
-180
6, f
(8) 7.6 X lo-''
-180
6, f
(7) 7.6
X
(9) 3.25 X lo-'' (10) 7.6 X (11)8.0 X lo-'' (12) 3.04 X lo-"
-180
i, f
6, k
i, f 1, 1
m,m
26.7 -180 -180 -180 -180 -180 -180 -180 -180 -180
m, m 6, m 6, f 6, f 6, f 6, f m, m m,m 6, f 15, 24 6, f 6, f 6, f 31, 31 31, 31 31, 31 31, 31 31, 31 31, 31 n, f
reactionCsd
-- -
1-octene + OH 0.7CH3(CH2)5CHOO'CH20H+ 0.3CH3(CH2)5CH(OH)CH200* 1-octene + O3 0.7HCHO + 0.8heptanal + O.llHzC'OO' + 0.11CH3(CHz)5C'H00' + 0.Olhexane + 0.450H + 0.20(3P) 1-octene + O(3P) 0.75octanal + 0.170R + 0.052-octanone 0.05heptanal OR + NO 0.lorganic nitrate + 0.9HOZ+ 0.9N02 + 0.9hydroxy carbonyl cpd CH3(CH2),CHOO'CH20H + NO 0.3organic nitrate + 0.7NO2 + 0.15HO2 + 0.15heptanal+ 0.15HCHO + 0.55CH3(CHz)&HOaCHZOH CH3(CH2)5CHO'CH20H+ NO 0.3organic nitrate + 0.7N02 + 0.7H02
+
+ + + -
+
0.7CH3(CH2)2CH(OH)(CH2)2C(O)CHOH
CH3(CHz)5CH(OH)CH200' NO 0.15heptanal + 0.55CH3(CH2),CH(OH)CHzO' + 0.3organic nitrate + 0.7N02 + O.15HCHO 0.15HOz CH3(CH2)5CH(OH)CH200' NO 0.3organic nitrate + 0.7NOz + 0.7HO2 +
- --
0.7CH3(CH.J3CH(O)(CH2)CH(OH)CHO
organic nitrate + OH 0.3hydroxy carbonyl cpd + 0.3NO2+ 0.7NCR1 NCRl + NO O.3organic nitrate + 0.7NO2 + 0.7hydroxy carbonyl cpd 2-octanone + OH MCHR heptanal + OH 0.68CH3(CHz)5C(=0)00' + 0.027CH3(CH2)4CH'CH0+ 0.155CH3(CHz)3CH'CH&HO + 0.045CH3(CH2)ZCHo(CH2)2CHO+ 0.046CH3(CH2)CH'(CHz)3CH0 + 0.O35CH3CH'(CH2),CHO(the last two are lumped with CH3(CH2)3CH'CH&HO) CH3(CHZ)5C(=O)OOo + NO2 CH3(CH2),C(=O)OONOz CH3(CH2)5C(=O)OON02 CH3(CHZ)&(=O)OO' + NO2 CH3(CHZ)$(=O)OO' + NO CH3(CHZ)dCH' + NO2 + COZ CH3(CH2)4CH'CH0+ NO hexanal + HOz + NOz + CO CH3(CH2)3CH'CH2CH0+ NO 0.3organic nitrate + 0.7N02 + 0.7CH3(CH2)3CHO'CHzCH0 CH3(CHz)3CHO'CHzCH0+ NO hydroxy carbonyl cpd + HOz + NOz CH3(CHz)2CH'(CHz)zCH0+ NO NOz + CH3(CH2),C(=O)OO' (lumped) heptanal + hu 0.92 6R1 + 0.92Ho2 + CO + 0.08hexane octanal + hv and 0.92 6R1 + 0.92Ho2 + CO + 0.08hexane (lumped) 6R1 + NO 0.79hydroxy carbonyl cpd + 0.79NO2+ 0.79HO2+ 0.2lorganic nitrate heptanoic acid + OH 0.12CH3(CH2),CH' + 0.12CO + 0.88HACR1 HACRl NO .--* NO2 + HACR2 HACR2 + NO substituted heptanoic acid + NOz + HOz CH3(CH2)4CH' + NO NO2 + HOz 2CH3(CH2)5CHOO'CH20H same products as 27b CH3(CH2)5CHOO'CHz0H+ CH3(CHz)5CH(OH)CHz00' 0.2heptanal+ 0.2H02 + 0.2HCHO + 0.4dihydroxy cpd + 0.4hydroxy carbonyl cpd + CH3(CH2)5CHOO'CH20H 2CH3(CHz)5CH(OH)CH200' 0.2heptanal + 0.2H02 + 0.2HCHO + 0.4dihydroxy cpd + 0.4hydroxy carbonyl cpd + CH3(CHz)5CH(OH)CHz00' CH3(CH2)5CH(OH)CH200'+ MCHR 0.12heptanal + 0.2H02 + O.GHCHO+ 0.7dihydroxy cpd + 0.7hydroxy carbonyl cpd + 0.5CH3(CHz)&H(OH)CH200' CH3(CHz)5CHOO'CHz0H+ MCHR 0.12heptanal + 0.2HOZ+ O.6HCHO + 0.7dihydroxy cpd + 0.7hydroxy carbonyl cpd + 0.5CH3(CHz)5CHOO'CH20H CH3(CHz)5CHOO'CH20H+ HOz and CH3(CHZ),CH(OH)CH200'+ HOz peroxide peroxide + OH 0.28CH3(CHz)5CHOO'CH20H+ 0.12CH3(CH2)5CH(OH)CH200' + O.6MCHR + 0.10H CH3(CHz)5C'HOO'+ NO heptanal + NOz CH3(CHz)4C'HOO'+ NOz heptanal + NO3 CH3(CHz)5C'HOO'+ HzO heptanoic acid Criegee biradicals + HCHO or heptanal --c ozonide n-octane + OH MCHR methylcyclohexane + OH MCHR MCHR + NO 0.84dihydroxy cpd + 0.16organic nitrate + 0.78HO2+ 0.84NOz hexane + OH MCHR MCHR + MCHR HOz MCHR + HOz peroxide2 peroxide2 + OH 0.10H + MCHR octanal + OH 0.65CH3(CHz)&(=O)00. (lumped) + 0.35MCHR
-
--
+
-
+
-
-
--
+
-
-
--
-
-
-
23, 23 23, 23 23, 23 23, 23 6, f 6, f -1 .80 6, o 6, f 31, f 31, f h, f 1, 1 others are indicated with names or structures. bThe first reference is for the rate constant, and the second is for the product bMbution. Addition of molecular oxygen to alkyl and acyl radicals is implicit in many reactions. Octanal is used to represent the sum '*al and hexyloxirane. eEstimated from trend of C2-C7 1-alkene-OH rate constants, which approach 4.0 X lo-" cm3 molecule-' S-', ~thm asymptotic trend. The activation energy was assumed to be the same as that for the 1-butene-OH reaction. 'This work. #The rate at 298 K for 1-heptene and the activation energy for 1-butene were used. hEstimated from correlations between the rate Constants of OH md developed by Atkinson (21). [Estimated from structure-activity relationships (28), based on the major hydroxy nitrate isomer. ' k d on the assumption that if OH attacks either the a- or &carbons, NOz is lost. Same assumption as G);additionally assuming that &Oxy radical isomerizes. 'Rate constant and branching ratios estimated from structure-activity relationships (28). "Analogous to cm3 molecule-' s-l [based on the wudehyde. We have assumed that the rate of OH attack on the hydroperoxide H atom is 3.7 X late a d branching ratio of methyl hydroperoxide (35)] and have added the hexane-OH rate (7) to account for OH attack on the carbon backbone. Based on extensive estimates of the branching ratios, products, and rate constants from structure-activity relationships de*by Atkinson (28).
- ---
Environ. Sci. Technol., Vol. 26, No. 6, 1992
1167
Table 111. Summary of Product Yields reaction OH
+ 1-octene
product
yield"
heptanal organonitrates*
15f8 30 f 8 45
octanal + hexyloxirane 2-octanone organic radicals
75 f 20 5f3 8 f 3c 88
heptanal thermally stabilized C7 Criegee biradical' hydroxyl radical unknown hexane
80 f 20 11 f 5
total O(3P) + 1-octene total
O3+ 1-octene
total
45 f 15 6 f 4d 0.8 f 0.4 92
a Percent of 1-octene reacted leading to product. bAssumed to equal n-octane yield. e Inferred from the observed heptanoic acid concentration. "This is not included in the total as its origin is unknown.
that as the decomposing compound (reactant in reaction 3) is a biradical, two radical species (or another biradical) will result from each decomposition. We have assumed that these products react with NO to produce 0.9NOZ(with arbitrary 10% organic nitrate formation), plus heptanal and a lumped oxygenated product (Table 11). The source of the minor (1% yield or less) products 2-ethylhexanal and 3-heptanone remains very unclear. It is possible that they are the products of oxidation reactions of impurities in the 1-octene used in these experiments, although pure liquid samples of 1-octene did not reveal any contaminant peaks other than n-octane. Simulations of experiment 103P1 were carried out using the mechanism in Table I1 and are included in Figures 1 and 2. The 0(3P)-l-octene rate constant was estimated from the O(3P)-OH rate constant correlations developed by Atkinson (21). We have examined the reaction between NOpand 1-octene in a separate experiment and found the cm3molecule-' s-l, too slow rate to be less than 1.2 X to be of importance even in this high-N02 system (18). While the NO and NOz simulations of this low-oxygen system are in good agreement with the data, the O3 data are difficult to simulate. The observed O3 concentration increases throughout the run (although it does not exceed 3% of the initial NO2 concentration), even after the NOz concentration has dropped and the NO concentration is several ppm; hence the explanation may lie in a product compound that absorbs at the same wavelength as the ozone analyzer (254 nm). Many oxygenated and nitrogenated compounds absorb to some degree in this region (22). Regardless, the majority of the 1-octene has reacted before the O3 has reached 25 ppb. Before we can derive the yields of the observed products of the O(3P)reaction, we must approximately derive the amount of OH forming in the system. OH can form from the ROz-HOz-NO, reactions resulting from the radical decomposition products. Calculations show that OH from the 0, photolysis pathway is minimal, even when the O3 concentration is artificially increased to the observed concentration. The NO and NOz concentrations are very sensitive to the amount of decomposition-initiated ROZ-HOz-NO, chemistry that occurs, as this is the dominant pathway for oxidation of NO to NOz. A simulation assuming zero decomposition predicts a maximum NO concentration of 5.45 ppm (the observed maximum was 4.47 ppm), and the predicted NOz concentration decreases to 0.3 ppm after 11min (the observed minimum was 0.87 ppm), indicating 1168 Envlron. Scl. Technol., Vol. 26, No. 6, 1992
a process converting NO to NOz is taking place. The simulations shown in Figures 1 and 2 were generated by assuming an organic radical yield of 0.16, translating into an 0.08 decomposition yield (since the decomposing corn. pound is a biradical; e.g., reactant in reaction 3). The combined octanal + hexyloxirane yield was 0.75, acd 2octanone, 0.05. The heptanal yield from the OH react ions is quite low, 15% (see below); hence it was necessary to assume some heptanal formation from the O(3P)reaction, via decomposition. The heptanal yield that provided the simulations in Figures 1 and 2 was 0.05. The yields, dong with the estimated uncertainties, are summarized in 'iable 111. N
Hydroxyl Radical Reaction If the OH reaction with 1-octene proceeds in the wme manner as with its smaller homologues, the reactm is expected to proceed as follows,beginning with OH addition across the double bond (e.g., ref 7):
This alkyl radical quickly adds 02,e.g.: H, /C-C%
OH
/
+
0,
CH&CH,),
-
,OH
''0,
/c\-
CHJCH,),
(5)
CH, H
In polluted atmospheres, this alkylperoxy radical reacts with NO, resulting in partial formation of an alkyl nitrate and an alkoxy radical, and in partial oxidation of NO to NOz: ''0,
CHJCH,),/
,OH C,-CH, H
+ NO
-
9 (1-XI
OPO,
,OH
/C\-CH,
CHJCH,),
(6) +
(1-X)
NO,
/W
C\-CH' CH3(CH2j5/ H
H
+ (X)
The value of x (the hydroxy nitrate formation yirid) has not been measured €or 1-octene or any other alkene except isoprene (23) and propene (24). Tuazon and Atkinson (23) estimated the value for isoprene (a C5 dialkene) at 0.12 0.05; however, this did not account for interference in the methyl nitrite experiments from O(3P) reactions. Taking this interference into account, the value becomes 0.14 f 0.05 (17), in agreement with the same size alkane, n-pentane, for which an alkyl nitrate yield of 0.13 f 0.004 Was measured [2-alkylperoxyradicals (25,26)],indicatiilg that the alkenes may have yields similar to the same size alkanes. The nitrate yields for propene were 0.017 r 0.008 and 0.015 f 0.007 for secondary and primary alkyiperoxY radicals, respectively (24), compared to 0.042 f 0.003 and 0.02 f 0.009 for propane secondary and primary radicals (%), respectively. The propene values may also have been underestimated somewhat, as they were also measured methyl nitrite experiments. Even so, the secondary propene value may in fact be smaller than its propane counterpart, while the primary value in is reasonable agreement. The alkyl nitrate yield for n-wtane is 0.33 (25); we have used 0.3 for 1-octene in our simulations. The alkoxy radical may decompose, e.g.:
9C,-CHz P H CH,CH,),/
H
-.--
CH3(CH,j5-C. (Heptanal)
2
yy-:
+ H,C
HCH
Ho2
(7)
(Formaldr' ( d e )
Decomposition is the dominant pathway for a variety Of smaller (up to C,) hydroxy alkoxy radicals (27).
6 b l e IV. Average Product Yields (ppb) for Low-Concentration Experiments total reacted 1-octene
expt
hexyloxirane + octanal
heptanal
iOHl 850 85 f 40 93 f 40 43 f 20 10H2 1330 100 f 50 70 f 0.30 INOX1 300 220 f 80 1031" 1390 350 f 150 11 i 5 olncluded here because data were collected only at the end of
2-octanone
hexane
15 f 10 11 f 5 16 i 8
10 f 5 9f5
n-octane reacted
3-heptanone 2-ethylhexanal 9f4
12 f 6
7f2
7f3 5f3 1 f 0.5 27 f 15 the experiment. *Heptanoic acid. 'Unknown.
Alternatively, this alkoxy radical may isomerize in the same manner as has been proposed for alkanes (3-6):
20 f 5
Octanal
and Hexyl oxirane
120-
The resulting alkylperoxy radical, in the presence of NO, ie expected to form the alkoxy radical with concomitant NOz production and partial alkyl nitrate formation. The alkoxy radical is then expected to isomerize again, abstracting the labile 2-carbon hydrogen and forming the following trisubstituted product along with H02:
-
OH '\CH
CH,CH,'
'CH,
/CH\, C,$ HC, OH
(9)
--02
OH
HQ,
CH,CH,/
CH /Cyc,CH2 'CH,
+HOP
0
A third alternative reaction pathway is possible for the 1scteneOH reaction, involving hydrogen abstraction from the alkyl backbone, e.g.: (10) CH,CH,
0.05
0.10
0.15
0.20
0.25
0.30
1-Octene Reacted (ppm)
OH
1 ,
0.00
;cy
/cy2
cH,
p
HC
2
+ H,O + other isomers
The alkoxy radical that results from this compound may then isomerize, as in reaction 9, react with 02,or decomPose. The structure-activity relationships developed by Atkinson (28) predict that in the neighborhood of 25% of the initial OH reaction involves H abstraction from the alkyl backbone. The same assumptions predict that H abstraction accounts for -5% of the 1-butene-OH reaction,consistent with the experimental observation that this Pathway is less than 5% (29). I t should be noted that, While the allylic C-H bond is -13 kcal/mol weaker than alkyl (secondary) C-H bond (30),the allylic H does not to have enhanced reactivity with respect to OH, b h g the possibility of OH attack on the alkyl backbone uncertain. ~._. A fourth alternative is reaction of the hydroxy alkoxy with 02:
Nowever,this reaction is expected to account for less than of the decomposition pathway (reaction 7) (27). The degree to which reaction 7 vs the alternate pathways 8, 10,
Figure 3. Heptanal and combined octanal and hexyloxlrane vs 1octene reacted from experiment 10H1, 10H2, and 1NOX1. Data are shown as symbols and simulations as lines.
and 11 occurs is explored below. OH was the dominant oxidant in both the methyl nitrite experiments (10H1 and -2), and 1NOX1. This latter experiment was designed with a low hydrocarbon/NO, ratio so that O3formation would be minimal (Table I). Major product data from samples concentrated on Tenax and run on a GC/FID for experiments l O H l and -2 and lNOXl are shown with simulated results in Figure 3. The final yields of all products are summarized in Table IV. The products include heptanal, hexyloxirane and octanal, and trace quantities of 2-octanone, as well as hexane, 3-heptanone, and 2-ethylhexanal in some experiments. Several other small peaks were observed, although none was larger than the heptanal peak, and we were unable to identify their mass spectra. The total yields of identified products were very low; about 22,16, and 21% for 10H1,10H2, and lNOX1, respectively. The dominant product is heptanal, an expected product of reaction 7. The formation of hexyloxirane, octanal, and 2-octanone is probably due to O(3P) formation in the chamber, (see reactions 2 and 3). Small quantities of hexane are expected from the photolysis of heptanal. Simulations of experiments l O H l and -2 and lNOXl were carried out using the mechanism listed in Table 11. Simulations for 1NOX3, an experiment similar to 1NOX1, are also shown in Figure 6. We found that the best agreement with all three experiments was obtained by assuming that the heptanal yield (via reaction 7) was 0.15. The mechanism was also successful in predicting the hexyloxirane + octanal concentrations from the O(3P) reaction; with the low hydrocarbon/NO, ratios used in 1NOX1, the OH concentration remains fairly low; hence the O(3P) reaction is significant. The final yields from the OH reaction with 1-octene are 0.15 f 0.07 heptanal, which, combined with an assumed alkyl nitrate yield of 0.3, totals 0.45. The remaining products of the 1-octene0H reaction probably isomerized Environ. Sci. Technol., Vol. 26, No. 6, 1992
1189
to form multisubstituted products via reactions 8 and 10, and the similar reactions for the l-alkoxy radical, and a small fraction may have reacted via reaction 11. Some of these products may condense to form aerosol. In an earlier study, we observed that -3% of the reacted l-octene was converted to aerosol (13),although those experiments did not make clear whether OH or O3reactions were responsible for aerosol formation. The low heptanal yield indicates that the l-octene-OH reaction may proceed via the following process. About 80% of the OH adds across the double bond (reaction 5), and 20% abstracts an H atom from the backbone (reaction 11). If -30% of each results in formation of organic nitrates, the 70% alkoxy and hydroxy alkoxy radicals, react -21% via channel 7 to produce heptanal, and the remainder react via (9) and (10). The total yields may then be 30% organic nitrate, 15-20% products from reaction 10, 15% heptanal from reaction 7, 35% from reactions 8 and 9, and possibly 1 or 2% via reaction 11.
Ozone Reaction Ozone experiments were carried out in the dark by adding externally generated O3 in O2to the chamber after the l-octene concentration had stabilized. On the basis of results for l-butene and other alkenes, O3is expected to add across the double bond (ref 7 , and references therein): H.
f
0/90
1*
This energetically excited adduct is expected to decompose. Cleavage through the C-C bond and one of the 0-0bonds yields two pairs of products: heptanal and a C1 Criegee biradical, and formaldehyde and a C7 Criegee biradical:
A 0
cyclohexane Reacted MlO3lUnknown .... M I 0 3 1 Heptanoic Acid
____________._ .......o'......*A. . 2
l-Octene Reacted (ppm)
column in broad peaks and hence do not develop into integratable peaks until they have reached subs:antiaI concentrations. The total product yields, on a per molecule l-octene reacted basis, were 43 (1031), 53 (1032), and 75% (M1031). Some of the variation in the total yields results from variation in the detection limits in each of the ex. periments. That all of the products were not observed in each of the experiments indicates that they did not exceed the detection limits of that experiment; they may still have formed. Simulations were carried out with the mechanism in Table I1 to determine the OH, carbonyl, and hexane yields needed to explain the data. The mechanism includes chemistry for the alkylperoxy radicals that form from OH addition-abstraction reactions with l-octene; these reac. tions produce minimal OH and heptanal (e.g., reaction 5 above; see also Table I1 and refs 9 and 18); the dominant fate of alkyl peroxy radicals as well as HOz is the following reaction: /OH
+ Hoz CH,(CHz15/C;-CH2 H
The fate of the C, Criegee biradical and the ratio of the two pairs of products are areas of significant uncertainty, particularly for an alkene as large as l-octene. From results for l-butene we expect that the C7 Criegee biradical decomposes to form hexane and COP,or becomes thermally stabilized, and may isomerize to form heptanoic acid (24). From the results for isoprene (9),we can expect that the C7Criegee will result in the formation of OH, more than 50% heptanal, and other species. Three experiments were carried out to examine the O3 reaction with l-octene: 1031,1032, and M1031. The first two experiments involved mixing externally generated O3 with l-octene in the dark, and the third also included methylcyclohexane (MCH), added to scavenge and track OH formation from the 03-l-octene reaction. The product concentration data from 1032 and M1031, as well as the MCH reacted, are shown as a function of l-octene reacted in Figure 4. The yield data from the end of experiment 1031 are shown in Table IV. The reaction of n-octane was also observed in experiment 1031 (Table IV).As expected from recent studies, significant OH formation is evident. The dominant product species is heptanal, but quantities of heptanoic acid, an unidentified peak, and trace quantities of hexane and 3-heptanone were also observed. The unknown and heptanoic acid eluted from the analytic 1170 Envlron. Scl. Technol., Vol. 26, No. 6, 1992
*I
Figure 4. Product concentration and methylcyclohexanereai:ted vs l-octene reacted for experiments 1032 and M1031. Data ;are indicated with symbols and simulations with lines.
'-0 \
(Formaldehyde))
f .
4
-
-'H
0,
/C,-CHz CHZ(CH215 H
,OH r
02
(14)
To find the OH yield necessary to account for the methylcyclohexane loss, we ran simulations of experiment M1031, adjusting the OH yield until it agreed Rith the data. An OH yield of 0.55 was used to generate the curve shown in Figure 4. However, the methylcyclohexane data do not extrapolate to zero (the maximum change in the methylcyclohexane signal was 5%, because the methylcyclohexane-OH rate constant is much slower than 1. octene), and if we assume that the OH is derived from the slope of the methylcyclohexane reacted vs l-octene we obtain an OH yield of -0.4. As the best fit of the mixed oxidant experiments (below) is obtained with an OH yield of 0.45, we recommend an OH yield of 0.45 0.2. an independent check on the OH yield, we simulated the impurity n-octane in experiment 1031, which began at ppm and ended at 0.15 f 0.005 ppm. Simulation-. of this reaction showed a final n-octane concentration of 0.1651 in good agreement with the data. The heptanal yield necessary to account for the concentrations observed (Figure 4) is 0.8 per molecule of1octene reacted. The simulation results for both 1032ad M1031 are included in Figure 4. The heptanal Yield'' higher than the observed yield because it accounts for the l-octene that reacted with OH rather than 03,and the reaction of heptanal with OH. About 2% of the heptad that formed came from the alkylperoxy radical reactions' S The hexane yield from the O,-I-octene reaction W ~ rathe'
*
os17
low; calculations showed that a yield of 0.012 was sufficient to account for the 1031 data. Heptanoic acid is expected to form from isomerization of the thermally stabilized C7 Criegee biradical, thought to be facilitated by water (8):
The heptanoic acid concentration was successfully simulated with a thermally stabilized C7Criegee biradical yield ofO.l. The estimated uncertainty for this yield is f0.06, resulting largely from the fact that the GC calibration factor for heptanoic acid was estimated. In summary, we recommend the following yields from the 0,-l-octene reaction: 0.80 f 0.15 heptanal, 0.012 f 0.004 hexane, 0.45 f 0.2 OH, and 0.1 f 0.06 thermally stabilized C7 Criegee biradical. We observed large quantities of O(3P)formation (a yield of 0.45) from the isoprene03 reaction (9),and there is evidence that this pathway is by no means restricted to isoprene; oxirane formation has been observed from the reaction of O3 with ethene, trans-2-butene, and tetramethylethylene (31). We did not observe evidence for the formation of O(3P) from the Opl-octene reaction, in the form of the known products of the 0(3P)-l-octene reaction, shown (above) to be octanal, hexyloxirane, and 2-octanone. However, the rate constant for the O(3P)reaction with l-octene is slower than isoprene by a factor of 5; hence l-octene is expected to be much less successful in competing for O(3P) with 02. Simulations showed that an O(3P) yield of 0.2 would produce less than 0.04 ppm octanal and hexyloxirane combined; this is below the detection limit of -0.06 ppm for these experiments. We propose the following mechanism for the 03-l-0ctene reaction. After the initial addition of O3to the double bond in l-octene, reaction 11, the ozonide decomposes aspmetically. r
H
(Formaldehyde))
The ratio is arbitrary, but since the C7 Criegee biradical is more stable than the C1 Criegee biradical, the formddehyde/C, Criegee pair is assumed to dominate. As it appears that the C1 Criegee does not result in OH formation (ref 7, and references therein), assuming the 0.3/0.7 hnching ratio allows room to explain OH formation. The c, Criegee biradical then breaks down in the following manner:
-
0-0 CH3(CH2)5--C(
0-0
0 1 5 CH,(CH,),-C,’
(thermally s t a b ~ l ~ r e d i
(Heptanal)
+ 002 (Hexane)
(HeptanalJ
n e first two products are derived directly from our obemations of heptanoic acid and hexane formation. The mechanism of OH formation is not well understood (7,9); we are assuming a pathway that produces one OH and one hePtanal, necessary to explain the large heptanal yield. Consistent with O(3P)formation in other systems (9),we b e assumed that the remaining C7 Criegee biradical
r~ 1.4 16[
y I
0
1
. I
I
, I
2
3
4
Time (hrs) Flgure 5. NO, 03, and l-octene data (symbols and solid lines) and simulations (broken lines) from experiment 1NOX2. The conditions for the simulations are indicated in the text. Some of the simulations for NO have been omitted for clarity.
decomposes to give heptanal and O(3P). This mechanism produces 0.105 thermally stabilized C7 Criegee biradical, 0.014 hexane, 0.42 OH, 0.16 O(3P),and 0.88 heptanal. Combined with the additional stabilized biradical from the C, biradical fragment [37% (31,the total yield of stabilized biradical is 0.22. This value is consistent with indirect measurements of stabilized biradical yields in other systems. Hatakeyama et al. (32) estimated ethene, propene, and trans-2-butene yielded 39, 25.4, and 18.5% stabilized biradical, respectively, and Niki et al. (8)found 25-30% for 2,3-dimethyl-2-butene. The most similar alkene of these is propene, which, like 1octene, has a monosubstituted double bond, and which compares favorably with our stabilized biradical yields. Mixed Oxidant Experiments In l-octene-NO, experiments OH, Os,and O(3P) are present, allowing us to evaluate the mechanism for the full photooxidation system. The NO, experiments can be divided into three sets. lNOXl was discussed in the OH section above. 1NOX2-6 are smog simulation runs similar to runs that have been made with many hydrocarbons for mechanism verification purposes. These runs were carried out (except for 1NOX3) as part of a set of experiments designed to study aerosol formation; this aspect is discussed elsewhere (13,14). In the present study, we have run simulations of each of those experiments that involved only l-octene, NO,, and in some cases small quantities of ammonium sulfate (seed) aerosol. Experiment 1NOX7 was run at a particularly high concentration and includes product data. The l-octene, NO, and O3data for the chamber experiments (1NOX2-5) with simulation results are shown in Figures 5 and 6. 1NOX6 is similar to 1NOX2 and hence has been omitted. The (NO,-NO) data were not simulated as the instrument response for the organic nitrates was not known. Simulations were carried out using the mechanism in Table 11, and the computational chemical kinetics scheme and inorganic and formaldehyde mechanisms of Carter and Atkinson (33). The inorganic portion of the mechanism includes the updates of Atkinson et al. (34). Wall effects were also included and are discussed in Paulson (18). Simulations using a series of assumptions about the mechanism are compared to the data from 1NOX2 in Figure 5. The base case (A), detailed in Table 11, makes the following assumptions: from the OH reaction, 15% decomposition with heptanal formation via reaction 7, 30% organic nitrates, and 55% isomerization and/or H atom abstraction via reactions 8-10. The O3 Environ. Sci. Technol., Vol. 26, No. 6, 1992
1171
2.0
I
I
I
I
Time (hrs)
0.2
Flgure 8. NO and O3data (solid and broken lines) and sirrrulations (dotted and dashed lines) from experiment 1NOX7.
0.0
0
2
1
3
Time (hrs)
Flgure 6. NO, O,, and 1-octene data (symbols and solid and dashed lines) and simulations (dotted lines) from experiments 1NOX3, 1NOX4, and 1NOX5. Note that O3is plotted with respect to the right-hand axis for 1NOX3.
I
-I
E % E.
+
1-
Combined Octanal and Hexyl oxirane
Time (hrs) Figure 7. l-Octene, heptanal, and combined hexyloxirane and octanal data (symbols) and simulations (lines) from 1NOX7.
reaction is assumed to produce 80% heptanal, 45% OH, 20% O(3P),and 70% formaldehyde. From the O(3P)reaction we assumed a 75% yield for the sum of octanal and hexyloxirane, 5 % 2-octanone, and 8% decomposition (leading to 16% organic radicals). Figure 5 shows the sensitivity of the predictions to assumptions about the amount of organic nitrate formed (B), the effect of assuming dominance of reaction 7 vs reactions 8-11 (C), and the importance of the amounts of OH (D)and O(3P)(E) from the O3reaction. The base case, A, is somewhat fast relative to 1NOX2 but provides good agreement with the 0,maximum (within 8%);the timing is better simulated for some other experiments (Figures 6-8). A reduction in organic nitrate formation from 30 to 20% (B) leads to an overestimation of the O3 peak, and a reaction system that is much too fast. Curve C shows the effect of assuming the OH reaction channels 65% through reaction 7 (forming heptanal) and 2% through reactions 8-10, with 30% nitrate formation. The result is similar to curve B; too much 03,too fast. This case is also shown for 1NOX7 in Figures 7 and 8 (discussed below). Assuming an OH yield from the O3 reaction of 55% rather than 45% produces curve D, also somewhat fast. Assuming no O(3P) from the O3reaction results (E) in an accurate prediction of the timing, but with some underestimation of the O3 peak. This case is also shown for experiment 1NOX5 (Figure 6), where it dramatically underpredicts both O3 and timing. 1172 Environ. Scl. Technol., Voi. 26,
No. 6, 1992
The mechanism in Table I1 performs well against the chamber data for runs 1NOX2-5 (Figures 5 and 6) and 1NOX6 (not shown). The predicted O3 levels are within 15% of the observed levels for all experiments, and the reaction timing is generally within -15% of what is observed. Additionally, the NO, 1-octene, and O3 curves agree with the predictions equally well; these variables do not behave independently. Simulations of the experiments with high hydrocarbon/NO, ratios (1NOX2, -4, and -6) me consistently too fast. This portion of these experiments is dominated by OH chemistry, indicating that the assumptions in the model for this reaction are reasonable, but may not be completely accurate. Consistent disagreement appears late in most of the runs, after the O3 has peaked in a regime that can be characterized by lowNO, chemistry. In this portion of the experiments, the predicted O3decays faster than the observed O3 This type of behavior has also been observed for isoprene (18). The low-NO, chemistry is very sensitive to the available NO,, which is largely controlled by the thermal release of NO2 from PAN-type compounds; in this system, peroxyheptanoic nitric anhydride (from heptanal). The rate for this process has not been measured (we used the decomposition rate for PAN); this may be the source of :ome of the discrepancy after the O3has peaked, although other parts of the low-NO, chemistry are uncertain as well. Experiment 1NOX7 is shown in Figures 7 and 8 and allows testing of the ability of our mechanism to predict the product data. The mechanism in Table I1 provides reasonable agreement with the ozone curve (Figure 8) in terms of timing and magnitude, but the O3concentration quickly drops below the observed value as the a\ ailable NO, is diminished. The 1-octene, heptanal, and NO are all in very good agreement. The combined hexyloxirane and octanal signal is underpredicted by -5070, possiblY indicating that the 0(3P)-l-octene rate constant is somewhat faster than estimated. If we instead assume case c, where the OH-1-octene adduct is assumed to decompose (reaction 7), we find that the model performance is much worse; heptanal is overpredicted by 35%, the timing is fast by -25%, and the combined hexyloxirane and &mal signal is underpredicted by -60%. These results COP roborate the OH experiments, indicating that the 0H-loctene reaction is not dominated by the subsequent de. composition reaction (reaction 7).
Summary The reactions of I-octene are in some ways very idar to those of smaller alkenes, and in others dramatically different. The O(3P) reaction is similar to thnr of its smaller homologues; this reaction is dominated by oxirme,
@mal, and 2-octanone formation. The decomposition pathway, -8%, is less important than it is for 1-butene, where it accounts for -20% (24). The majority of the OH-initiated alkoxy radical reaction appears to proceed via a different pathway than the , d e r hydrocarbons; the decomposition pathway results in only a 15% heptanal yield. The alkyl nitrate yield from thi(~reaction may account for 30% of the products, leaving 55% unaccounted for by known pathways. Likely reaction for the remaining alkoxy radicals is an internal isomeriation via a six-membered ring. Some fraction of the initial I.dne-0H reaction may also involve H atom abstraction from the carbon backbone. Some of these products may enter the aerosol phase, which was shown to account for H3% of the reacted 1-octene in a previous study (13). The 1-octene reaction with Os produces significant quantities of OH, -0.45 per molecule of 1-octene reacted, 88 well as about 80% heptanal, 10% thermally stabilized C, biradical, and 1%hexane, accounting for nearly 100% of the reacted 1-octene. Some O(3P)may also form, possibly 20%, although we have no direct evidence for this. The OH yield is not at all consistent with most Os-alkene studies, but as discussed in detail in Paulson et al. (9),OH formation in Os reaction systems has often been overlooked. The OH yield is consistent with those studies that have inferred or measured OH formation (8-10) and begins to fill out the picture of the quantities of OH that can be expected to form from the alkene family. The heptanal yield is much more than the 50% carbonyl yield expected from the standard 0,-alkene reaction mechanism and exceeds the rough estimate of hexanal yield from the 1heptene-Os reaction reported by Grosjean (15). The standard mechanism and Grosjean's study, however, did not account for OH formation. The heptanal yield is consistent with findings for isoprene (9) where carbonyl yields totaled 93%. The decomposition pathway that leads to hexane formation appears to be very minor, at -1% compared to 6% for 1-butene (21). The complete mechanism, comprised of the constituent OH, O(,P), and Os reactions compares favorably to smog chamber data, and supports conclusions from the experiments that isolate individual reactions. In particular, assuming a different amount of decomposition from the OH reaction predicts too much ozone, too fast, and overpredicts the heptanal yield. However, agreement is not perfect, particularly under low-NO, conditions. Registry No. 0,17778-80-2; OH, 3352-57-6; NO,, 11104-93-1; 08,10028-15-6; 1-octene, 111-66-0; heptanal, 111-71-7; octanal, 124-13-0;hexyloxirane, 2984-50-1; 2-octanone, 111-13-7; hexane, 110-54-3;2-ethylhexanal, 123-05-7; 3-heptanone, 106-35-4; n-och e , 111-65-9; nitric acid, 7697-37-2.
(19) (20) (21) (22) (23) (24) (25) (26) (27) (28) (29) (30) (31) (32) (33)
(34) (35)
Isidorov, V. A. Organic Chemistry of the Earth's Atmosphere; Springer-Verlag: Berlin, 1990. Grosjean, D.; Fung, K. J.-Air Pollut. Control Assoc. 1984, 34, 527. Carter, W. P. L.; Darnall, K. R.; Lloyd, A. C.; Winer, A. M.; Pitts, J. N. Chem. Phys. Lett. 1976, 42, 22. Baldwin, A. C.; Barker, J. R.; Golden, D. M.; Hendry, D. G.J. Phys. Chem. 1977,81, 2483.
Carter, W. P. L.; Atkinson, R. J. Atmos. Chem. 1985,3,377. Dobe, S.; Berces, T.; Marta, F. Znt. J. Chem. Kinet. 1986, 18, 329. Atkinson, R. Atmos. Environ. 1990, 24A, 1. Niki, H.; Maker, P. D.; Savage, C. M.; Breitenbach, L. P.; Hurley, M. D. J . Am. Chem. SOC.1987, 91,941. Paulson, S. E.; Flagan, R. C.; Seinfeld, J. H. Znt. J. Chem. Kinet. 1992,24, 103. Japar, S. M.; Wu, C. H.; Niki, H. J . Phys. Chem. 1976,80, 2057. Herron, J. T.; Huie, R. E. Znt. J. Chem. Kinet. 1978, 10, 1019. Atkinson, R.; Hasegawa, D.; Aschmann, S. M. Znt. J. Chem. Kinet. 1990,22, 871. Wang, S. C.; Paulson, S. E.; Grosjean, D.; Flagan, R. C.; Seinfeld, J. H. Atmos. Emiron. 1992, 26A, 403. Wang, S. C.; Flagan, R. C.; Seinfeld, J. H. Atmos. Environ. 1992,26A, 421. Grosjean, D. Sci. Total Environ. 1984, 37, 195. Pandis, S. N.; Paulson, S. E.; Flagan, R. C.; Seinfeld, J. H. Atmos. Environ. 1991, 25A, 997. Paulson, S. E.; Flagan, R. C.; Seinfeld, J. H. Znt. J . Chem. Kinet. 1992,24, 79. Paulson, S. E. Ph.D. Thesis, California Institute of Technology, Pasadena, CA, 1991. Cvetanovic. R. J. Adv. Photochem. 1963. 1. 115. Atkinson, R.; Lloyd, A. C. J . Phys. Chem.'Ref. Data 1984, 13, 315. Atkinson, R. Chem. Rev. 1985,85, 799. Hirayama, K. Handbook of Ultraviolet and Visible Absorption Spectra of Organic Compounds; Plenum Press: New York, 1967. Tuazon, E. C.; Atkinson, R. Int. J. Chem. Kinet. 1990,22, 1221. Shepson, P. B.; Edney, E. 0.;Kleindienst, T. E.; Pittman, J. H.; Namie, G. R. Environ. Sci. Technol. 1985, 19, 849. Atkinson, R.; Aschmann, S. M.; Carter, W. P. L.; Winer, A. M.; Pitts, J. N. J . Phys. Chem. 1982, 86, 4563. Atkinson, R.; Cater, W. P. L.; Winer, A. M. J. Phys. Chem. 1983,87, 2012. Atkinson, R.; Carter, W. P. L. J. Atmos. Chem. 1991,13, 195. Atkinson, R. Int. J . Chem. Kinet. 1987, 19, 799. Atkinson, R.; Tuazon, E.; Carter, W. P. L. Int. J . Chem. Kinet. 1985, 17, 725. McMillen, D. F.; Golden, D. M. Annu. Rev. Phys. Chem. 1982, 33, 493. Martinez, R. I.; Herron, J. T.; Huie, R. E. J . Am. Chem. SOC.1981, 92, 4644. Hatakeyama, S.;Kobayashi, H.; Akimoto, H. J.Phys. Chem. 1984, 88, 4736. Carter, W.; Atkinson, R. Development and implementation of an up-to-date photochemical mechanism for use in airshed modeling. Summary final report, California Air Resources Board, 1988. Atkinson, R.; Baulch, D. L.; Cox, R. A.; Hampson, R. F., Jr.; Kerr, J. A,; Troe, J. Znt. J. Chem. Kinet. 1989,21,115. Vaghjiani, G. L.; Ravishankara, A. R. J. Phys. Chem. 1989, 93, 1948.
Received for review September 9, 1992. Revised manuscript received January 21, 1992. Accepted February 24,1992. This work was supported by National Science Foundation Grant ATM-9003186 and a Dissertation Fellowship from the American Association of University Women.
Environ. Sci. Technol., Vol. 26,
No. 6, 1992 1173
