Hydroxyl radical production from the gas-phase reactions of ozone

Suzanne E. Paulson, Jill D. Fenske, Atish D. Sen, and Tyrone W. Callahan ..... Marc in het Panhuis , Adam J. Trevitt , Todd W. Mitchell , Stephen J. B...
0 downloads 0 Views 829KB Size
Envlron. Sci. Technot. 1993, 27, 1357-1363

OH Radical Production from the Gas-Phase Reactions of Alkenes under Atmospheric Conditions

0 8

with a Series of

Roger Atklnson'pt and Sara M. Aschmann Statewide Air Pollution Research Center, University of Californla, Riverside, Californla 92521

The gas-phase reactions of 0 3 with a series of alkenes have been investigated at 296 f 2 K and atmospheric Pressure of air in the presence of cyclohexane at concentrations sufficient to scavenge any OH radicals formed. The expected products of the OH radical-initiated reaction of cyclohexane, cyclohexanone and cyclohexanol, were observed in all cases. From a knowledge of the chemistry of cyclohexane in these reaction systems and the cyclohexanone and cyclohexanol formation yields obtained, the following formation yields of OH radicals in these 0 3 alkene reactions were derived propene, 0.33; l-butene, 0.41; 2-methylpropene, 0.84; cis-2-butene, 0.41; trans-2butene, 0.64; 2-methyl-l-butene, 0.83; 2-methyl-2-butene, 0.89; 2,3-dimethyl-2-butene, 1.00; cyclohexene, 0.68; 1,3-butadiene,0.08; all with estimated overall uncertainties of a factor of -1.5. Introduction The kinetics, mechanisms,and reaction products of the gas-phase reactions of 0 3 with the 5C7 alkenes have been extensively studied (1-32), in part because of their importance in the formation of photochemical air pollution (33,34). While the rate constants for these reactions are now reliably known (1,28,31,32),there are still significant uncertainties concerning the detailed reaction mechanisms and the products formed (18, 25, 32). The gas-phase reaction of 0 3 with an alkene proceeds by initial addition to the >C=C< bond to form an energy-rich ozonide [ 3 *, which rapidly dissociates to carbonyl(s) and initially energy-rich Criegee biradical(s). The energy-rich birad-

Experimental Section

J icals can decompose or be collisionally stabilized R,R.&6 [RyR&6]+ decomposition produds

with the stabilizationldecomposition ratio depending on the total pressure (4, 7). At present, one of the major uncertainties concerns the reactions under atmospheric conditions of the biradicals and the extent and nature of the decomposition products (18,25,32). For several years, there has been evidence from both product and kinetic studies that reactive radical species ~

* Author to whom correspondence should be addressed. + Also at the Department of Soil and Environmental Sciences, University of California, Riverside. 0013-936X/93/0927-1357$04.00/0

e

are formed during these reactions (9, IO, 15,35-39), and OH radicalshave been implicated playinga major role in the chemistrs of ozone-alkene reactions (g,10,15,3639). It should be noted that in the low-pressure studies of Herron and Huie (36, 37), Martinez et al. (381, and Martinez and Herron ( 10,151,any H atoms produced would have been rapidly and efficiently converted to OH radicals by reaction with 03,and hence these studies (10,15,3638) could not differentiate between initial formation of H atoms or OH radicals. Recently, however, Paulson et al. (26)and Paulson and Seinfeld (30) have monitored the decay of methylcyclohexane (which reacts only with the OH radical) in reacting 03-alkene-air mixtures and concluded from computer modeling that OH radicals are produced directly in the reactions of 0 3 with isoprene [2-methyl-l,3-butadienel (26)and l-octene (32)with yields of 0.68 f 0.15 and 0.45 f 0.2, respectively. Using a somewhat similar approach, Atkinson et al. (29) used excess concentrations of cyclohexane to scavenge any OH radicals formed in the reactions of OSwith ethene, isoprene, and a series of monoterpenes and derived the OH radical formation yields from these reactions from the cyclohexanone and cyclohexanol products formed from the OH radical reaction with cyclohexane. The OH radical formation yields derived varied from 0.12 for ethene to -1.0 for several of the monoterpenes containing internal >C=C< bonds, and that for isoprene was 0.27 (allwith estimated overall uncertainties of a factor of -1.5) (29). In this work, we have extended our previous study (29) to investigate the formation of OH radicals from the gasphase reactions of 0 3 with propene, l-butene, 2-methylpropene, cis- and trans-2-butene, 2-methyl-l-butene, e-methyl-%-butene,2,3-dimethyl-2-butene, cyclohexene, and 1,3-butadiene at room temperature and 1 atm total pressure of air.

0 1993 Amerlcan Chemical Society

The experimental methods employed were similar to those used and described by Atkinson et al. (29). Two seta of experiments were carried out. In the first set of experiments, the products arising from the OH radicalinitiated reactions of cyclohexane were measured together with the amounts of the alkenes reacted in the presence of sufficient cyclohexane that any OH radicals formed from the 03-alkene reactions were scavenged by reaction with cyclohexane. Cyclohexane does not react with Os, and in these systems the only loss process of cyclohexane was by reaction with the OH radical (18). Under these conditions of excess cyclohexane, the amount of cyclohexane reacted could not be quantified. In the second set of experiments, much lower cyclohexane concentrations were used, and the initial cyclohexane concentrations were similar to those of the alkenes. While these experimentswere not suitable for determining the OH radical formation yields from the Os-alkene reactions since the majority of the OH radicals formed Envlron. Scl. Technol., Vol. 27, No. 7, 1993 1357

reacted with the alkene, the amounts of cyclohexane reacted could be measured together with the amounts of cyclohexanone and cyclohexanol formed from the OH radical-initiated reaction of cyclohexane. Thus, the ratio [(cyclohexanone + cyclohexanol formed)/cyclohexane reacted] could be experimentally determined. All experiments were carried out at 296 f 2 K and 740 Torr total pressure in a -6900-L all-Teflon chamber equipped with a mixing fan rated at 300 L s-l. Pure air was used as the diluent gas, with a water vapor concentration of 2.7 X 10l6molecule ~ m(-4 - ~% relative humidity) being employed for the majority of the experiments. A water vapor concentration a factor of 9 higher (2.4 X lo1' molecule ~ m - was ~ ) used for an experiment with propene. Reactions with Excess Concentrations of Cyclohexane. The concentrations of the reactant alkene and the cyclohexanone and cyclohexanol products were measured by gas chromatography with flame ionization detection (GC-FID) in 03-alkene-cyclohexane-air mixtures. The cyclohexane concentrations were sufficiently high so that >95% of any OH radicals formed from the 03-alkene reactions were scavenged by reaction with cyclohexane, as calculated from the OH radical reaction rate constants (40) and the alkene and cyclohexane concentrations. The initial reactant concentrations (in molecule ~ munits) - ~ were as follows: propene, (2.29-5.51) X 1,3-butadiene, (4.76-4.80) X 1013; other alkenes, (1.95-2.59) X 1013;cyclohexane, -9 X 10l6;and 4 or (for the higher concentration propene experiments) 5 additions of O3 in 02 diluent (corresponding to initial O3 concentrations in the chamber of -5 X molecule ~ m per - ~ addition) were made to the chamber during each experiment. Rapid mixing of the 0 3 / 0 2 additions was achieved by use of the chamber mixing fan (31). For the analysis of the alkenes, 100 cm3 gas samples were collected from the chamber into gas-tight, all-glass syringes for injection into a gas-sampling loop connected to a 30-m megabore DB-5 fused silica column held at -25 OC and then temperature programmed at 8 "C min-l (a 15-m megabore DB-5 column was used with the same temperature program for the first propene experiment carried out, with an initial propene concentration of 2.29 x 1013molecule ~ m - ~For ) . the analysis of cyclohexanone and cyclohexanol, 100cm3gas samples were collected from the chamber onto Tenax-GC or -TA solid adsorbent. These Tenax samples were then thermally desorbed at -225 OC onto the head of a 15- or 30-m megabore DB-5.625 fused silica column held at 0 OC and then temperature programmed at 8 "C min-l. With the 15-m column, cyclohexanone and cyclohexanol were only marginally resolved under these GC conditions, but their GC-FID response factors were experimentally determined to be identical to within 5 % ,and reliable quantifications of (cyclohexanone + cyclohexanol) were obtained. The 30-m column was used for all but the initial three experiments with propene, and cyclohexanoneand cyclohexanolwere more completely resolved with this column, enabling the concentrations of each of these products to be determined. Thermal desorption analyses of Tenax adsorbent samples were also carried out for selected experiments by combined gas chromatography-mass spectrometry (GCMS), using a Hewlett-Packard (HP) 5890 GC interfaced to a HP 5790 mass selective detector (MSD) with a 60-m N

1358 Envlron. Scl. Technol., Vol. 27, No. 7, 1993

DB-5 capillary column, which resolved cyclohexanoneand cyclohexanol. Measurement of Cyclohexanone and Cyclohexanol Formation Yields in Reacting Alkene-Cyclohexane 03-Air Mixtures. The concentrations of cyclohexane, cyclohexanone, and cyclohexanol were measured by GCFID in 03-alkene-cyclohexane-air mixtures. The initial reactant concentrations were (in molecule ~ m units) - ~ as follows: cyclohexane, (2.20-2.51) X 1013;propene, (4.91other alkenes, (2.06-2.55) X and 4 or (for 4.94) X propene) 5 additions of O3 in 02 diluent, each equivalent to an 03 concentration in the chamber of -5 X 10l2 molecule ~ m - were ~ , made to the chamber during each experiment. Cyclohexane and the alkenes were analyzed using the 30-m DB-5 megabore column and loop system described above. The 15-m DB-5.625 column/thermal desorption system was used for the analyses of cyclohexanone and cyclohexanol in the first propene experiment of this series, with the 30-m DB-5.625 column being used in all subsequent experiments. Chemicals. The sources of the chemicals used, and their stated purities, were as follows: cyclohexane (highpurity solvent grade), American Burdick and Jackson; 1,3butadiene (299.0%),1-butene(199.0%),2-methylpropene (199.0%), and propene (199.0%), Matheson Gas Products; cis-2-butene and trans-2-butene (both CP grade), Linde; 2,3-dimethyl-2-butene (99+ % ) and 2-methyl-2butene (99+%), Aldrich Chemical Co.; and cyclohexene (99%) and 2-methyl-1-butene (99.9%), Chem Samples. O3in 02 diluent was prepared as needed using a Welsbach T-408 ozone generator. Results

Reactions with Excess Concentrations of Cyclohexane. A series of reactions of 0 3 with propene, 1-butene, 2-methylpropene, cis-2-butene, trans-2-butene, 2-methyl1-butene, 2-methyl-2-butene, 2,3-dimethyl-2-butene, cyclohexene, and 1,3-butadiene were carried out in the presence of 1 atm of air and with sufficient added cyclohexane to scavenge >95 % of any OH radicals formed from the 0 3 reactions. In these experiments, the alkene, cyclohexanone, and cyclohexanol concentrations were measured by GC-FID during the reactions. Plots of the amounts of (cyclohexanone+ cyclohexanol)formed against the amounts of alkene reacted are shown in Figures 1-3. Generally good straight line plots are observed, and the ratios [(cyclohexanone + cyclohexanol formed)/ (alkene reacted)] obtained from least-squares analyses of these data are given in Table I. Because of the very low conversions of cyclohexane W0.3 % 1, any corrections to the measured cyclohexanone and cyclohexanol concentrations to take into account secondary reactions of these species were totally negligible. Three independent sets of experiments were conducted for propene over a 1-year interval with differing GC columns and calibration factors (see above), with agreement between these sets of experiments within the overall experimental uncertainties (Figure 1). Although a constant yield of [(cyclohexanone + cyclohexanol formed)/ (alkene reacted)] is assumed here, the data from the propene reactions plotted in Figure 1 provide some evidence for a slight decrease in this yield with increasing extent of reaction. Experiments were also carried out for propene with the water vapor concentration being varied

12 r

m

r

-A [ALKENE] molecule cm3

+

Flgurr 1. Plots of the amounts of (cyclohexanone cyclohexanol) formed against the amounts of propene and 1,3-butadlene reacted in O3-alkene-cyclohexane (In excess)-air mixtures. For the propene units) reactions, the initial propene concentrations (in molecule andthe GC columns usedin the analyses for propeneand cyclohexanol and cyclohexanone were as follows: ( 0 )2.29 X loi3, 15-m DE51 loop, 15-mD&5.625/thermaidesorption;(0)(4.92-5.02) X 30-m D&S/loop, 15mDB5.625/thermaidesorptkn;(A)(4.79-4.88) X lo? 30-m D&5/loop, 30-m D&5.625/thermal desorption.

0.5

1.o 1.5 -A [ALKENE] mdecule cm4

2.0

2.5~10'~

+

Flgurr 3. Plots of the amounts of (cyclohexanone cycbhexanol) formed against the amounts of 1-butene, 2-methyl-2-butene, 2,s dlmethyL2-butene, and cyckhexene reacted InOAkene-cyckhexane (In excess)-air mixtures. The data for 2,3dlmethyl-2-butene have been displaced vertically by 1.0 X loi2 molecule cm" units for ciarity.

Table 1. Measured Formation Yields of (Cyclohexanone Cyclohexanol)/(Alkene Reacted) and Calculated Formation Yields of OH Radicals in Reacting 08-Alkene-CyclohexaneAir Mixtures at 296 f 2 K and 740 Torr Total Pressure'

+

2-METHYL-l-BUTENE

alkene

cyclohexanone + cyclohexanolb alkene reacted

OH radical formation yieldc

propene

0.167 t 0.022 0.33 0.140 t 0.013d 0.28d 1-butene 0.206 0.018 0.41 0.84 2-methylpropene 0.419 0.032 cis-2-butene 0.204 f 0.026 0.41 trans-2-butene 0.318 f 0.029 0.64 2-methyl-1-butene 0.413 t 0.034 0.83 2-methyl-2-butene 0.445 t 0.040 0.89 2,3-dimethyl-2-butene 0.501 t 0.044 1.00 cyclohexene 0.338 t 0.032 0.68 1,3-butadiene 0.042 t 0.004 0.08 a Water vapor concentration 2.7 X 10lsmolecule cm4 unless noted otherwise. * Indicated errors are two least-squareastandard deviations combined with estimated overall uncertainties in the GC-FID calibration factors for the alkenes, cyclohexanoneand cyclohexanol of *5% each. Overall uncertainty is estimated to be a factor of -1.5. At a water vapor concentration of 2.4 X lo1' molecule cm4.

* *

I

I

I

I

0.5

1.o

1.5

2.0

_I

2.5~10'~

-A [ALKENE] molecule m3

+

Figure 2. Plots of the amounts of (cyclohexanone cyclohexanol) formed against the amounts of 2-methylpropene, cls-2-butene. trans2-butene, and 2-methyl-1-butene reacted in 03-alkene-cyclohexane (in excess)-air mixtures. The data for 2-methylpropgne and 2-methyl1-butene have been displaced vertically by 1.0 X loi2 and 2.0 X loi2 molecule ~ m units, - ~ respectively, for ciarity.

over the range (2.7-24) X 1OI6 molecule ~ m - and ~ , it can be seen from Table I that the (cyclohexanone + cyclohexanol)/(propene reacted) yield was invariant over this range of water vapor concentrations. The cyclohexanone/cyclohexanolconcentration ratios were also determined and varied from 0.72 (cyclohexene) to 1.4(propene), with the remainder of the alkene reactions leading to concentration ratios of 0.8-1.2. These cyclo-

hexanone/cyclohexanol concentration ratios can be compared with ratios ranging from 1.0 to 1.9 for the reactions in which the concentrations of cyclohexane were a factor of -10oO lower and in which the cyclohexane reacting was measured (see below). GC-MS measurements of the cyclohexanone/cyclohexanol concentration ratios were in agreement with the GC-FID data. Measurement of Cyclohexanoneand Cyclohexanol FormationYields in Reacting AlkendyclohexaneOa-Air Mixtures. Reactions of O3with all of the alkenes, apart from 1,3-butadiene, in air in the presence of cyclohexane were carried out in which the disappearances of Envron. Sci. Technol., Vol. 27, No. 7, 1993 1959

Discussion

+

G i -

B w

0.4 -

r

9

D

Y

Y

0

Y

x

$

o

0

,

I

1

,

I

I

I

2

b

J

3x101*

-A [CYCLOHEXANE]molecule ~ r n - ~ Figure 4. Plot of the amounts of (cyclohexanone cyclohexanol) formed, correctedfor reaction with the OH radical, against the amount of cyclohexane reacted in 03-alkene-cyclohexane-air mixtures.

+

cyclohexane and the alkene and the formation of cyclohexanone and cyclohexanol were measured by GC-FID. In each reaction, four or five additions of 08 in 02 diluent were added to the chamber. Since only small amounts of cyclohexane reacted under these conditions (110%), dilution due to the 03/02 additions (0.07 % per addition) was taken into account in the data analyses. No reaction was carried out for 1,3-butadiene because the low OH radical formation yield (Table I) would have led to only small amounts of cyclohexane reacted. No useful data were obtained from the cyclohexene reaction because of GC interferences with the analyses of the cyclohexanone and cyclohexanol. In these experiments, the ratio of the cyclohexanone/ cyclohexanol formed was in the range 1.0-1.9. As discussed previously (29,41),it was necessary to take into account the reactions of the cyclohexanone and cyclohexanol products with the OH radical during these experiments, and this was carried out as described by Atkinson et al. (41), using rate constants (in units of cm3 molecule-1 s-l) of cyclohexane, 7.49 (40);cyclohexanone, 6.39 (42);and cyclohexanol, 17 (43). A plot of the amounts of (cyclohexanone + cyclohexanol) formed, corrected for reaction with the OH radical, against the amount of cyclohexane reacted is shown in Figure 4. While a significant amount of scatter is evident in the data shown in this plot, this is largely due to the small fraction of cyclohexane reacted (110%). A least-squares analysis of these data yields a formation yield of (cyclohexanone + cyclohexanol) of cyclohexanone + cyclohexanol formed/cyclohexane reacted = 0.50 f 0.07 where the indicated error is two least-squares standard deviations of the slope of the plot in Figure 4 combined with the estimated overall uncertainties in the GC-FID calibration factors for cyclohexane, cyclohexanone, and cyclohexanol of f 5 % each. 1360 Environ. Sci. Technoi., Vol. 27, No. 7. I993

The experimental data presented above show that OH radicals are formed from the gas-phase reactions of Os with the alkenes studied under the experimental conditions utilized in this study. Furthermore, the formation yields of cyclohexanone and cyclohexanol from cyclohexane are independent of the alkene concentration during the experiments, as evidenced by the generally good straight line plots in Figures 1-3, and our data for propene suggest that these formation yields are also independent of the water vapor concentration. However, the formation yield of cyclohexanone plus cyclohexanol from cyclohexane is less than unity in reacting 03-alkene-cyclohexane-air mixtures (0.50 f 0.07 and 0.55 f 0.09 from our present and previous (29) studies, respectively). While expected based on the chemistry of cyclohexane in these reacting Os-alkene systems (refs 29 and 44; see also below), this variable and less than unit yield leads to uncertainties in deriving the yield of OH radicals from the measured (cyclohexanone + cyclohexanol)/(alkene reacted) yields in the 03-alkene reactions with added excess concentrations of cyclohexane. Furthermore, the possibility of generation of OHradicals from HOz radicals, formed either directly in the Os-alkene reactions or from the OH radical reaction with cyclohexane, needs to be assessed. Based on the current knowledge of the OH radical reaction with cyclohexane in the absence of NO, (18,32, 44,451, the following reaction sequence is anticipated if OH radicals are produced in 03-alkene reactions OH + C6H1, -.+ HZO + C6Hll*

+ - + + + -

(1)

M

C6H11'

+ 0,

C6H110C)

C6H11OC) ROC) C,HllC) + RC) + 0,

(2)

(3)

C6H110C)+ ROC) a cyclohexanone + (1 - a)cyclohexanol products of ROC)+ 0,(4) C6Hl100H + 0,

(5)

cyclohexanone + HO,

(6a)

C6H1,0C) HO,

-

C6H11C) 0,

C6H11C)

decomposition and/or isomerization products (6b) where ROC) is organic peroxy radicals including the cyclohexyl peroxy radical. These reactions lead to the formation of cyclohexanone and cyclohexanol from the reacted cyclohexane, with a (cyclohexanone + cyclohexanol)/(cyclohexane reacted) yield of less than unity (29, 44).

The parameters of interest are then /3 [the fractional yield of cyclohexanone from the CsH11C) alkoxy radical; /3 = ksa[Ozl/(k~e[Ozl f k6b)l, the value of CY in reaction 4, and theratios y = kdk4and 6 =kslHOzl/((ks f kd[ROol}. We have previously determined that 0 = 0.42 f 0.05 from measurements of the cyclohexanone yield from the OH radical-initiated reaction of cyclohexane in the presence of NO, (29). Additionally, Rowley et al. (44) have determined a (cyclohexanone + cyclohexanol) formation yield of 0.67 f 0.05 and a cyclohexanone/cyclohexanol concentration ratio of 1.23 f 0.14 from the C1 atominitiated reaction of cyclohexane in 700Torr total pressure of air at room temperature using in situ Fourier transform infrared absorption spectroscopy. These data of Rowley et al. (44) are, however, not expected to be quantitatively

applicable to the present situation because of the presence of additional ROz' and/or HO2 radicals in 03-alkene reaction systems compared to the C1 atom-initiated reaction of cyclohexane alone. Based on reactions 1-6, then in reacting 03-alkenecyclohexane-air mixtures ([cyclohexanone1 + [cyclohexanoll )/ [cyclohexane reacted1 = (1 + By)/[(l + y ) ( l + 611 (1)

At the present time, the precise source(s) of these OH radicals in the reactions of O3 with the alkenes studied here is not known, although it is expected (see,for example, refs 9 and 10) that the energy-rich biradicals undergo decomposition by three pathways:

-

[R1CH26(Rz)06]~

[cyclohexanone]/[cyclohexanol] = (by + a ) / ( l- a ) (11) where y = k3/k4 and 6 = ks[HOzl/{(ka + kd[ROOIj. Asuming a value of a rrr 0.5 (32, 45), the data obtained from the alkene reactions with O3 in which the amounts of cyclohexanone and cyclohexanol formed and the amount of cyclohexane reacted were measured lead to values of y (=k3/k4) N 0.5 f 0.5 and 6 [= k6[HOz]/((ka + kd[R063) N 0.7 f 0.3. The uncertainties in y and 6 allow for the range of values of ([cyclohexanonel/ [cyclohexanoll) observed. In view of the possible occurrence of HO2 radical reactions with the CeH11OC) peroxy radicals (reaction 5) and the uncertainties in the value of a in reaction 4 and in the relative importance of reactions 3-5, the (cyclohexanone + cyclohexanol) formation yield from cyclohexane in reacting 03-alkene-cyclohexane-air mixtures measured in this work of 0.50 has been used to calculate the OH radical formation yield from the measured (cyclohexanone + cyclohexanol)/(alkene reacted) data. These calculated OH radical formationyields are also given in Table I and have estimated uncertainties of a factor of -1.6 associated with them. As noted above, there is the possibility that HO2 radicals, generated either directly from the 03-alkene reaction or from the OH radical reaction with cyclohexane (through reaction 6a), can be converted to OH radicals by the reaction: HO,

+ 0,--, OH + 20,

(7) This reaction is in competition with reaction 5 and with the self-reaction of HO2 radicals:

HO, + HO,

-

H202+ 0, (8) Using the rate constanta for the reactions of 0 3 with the alkenes studied (28, 31, 32) and the rate constants for reactions 7 and 8 under atmospheric conditions of 2 X 10-l6cm3 molecule-' s-l and 3 X 10-l2cm3 molecule-l s-l, respectively (45), then the importance of reactions 7 and 8 as loss processes for HO2 radicals (and of reaction 7 for the regeneration of OH radicals) can be assessed. For the 0 3 concentrations employed, and using H02 radical formation yields identical to the OH radical formation yields given in Table I, the dominant HO2 radical loss process was then calculated to be by the self-reaction 8. Hence for these reaction systems, the formation of OH radicals from HO2 radicals was not important, and the data given in Table I are for the direct formation of OH radicals without the intermediary of HOz radicals. The calculations indicate that an OH radical formation yield of 10.05 was due to OH radical generation from HO2 radicals by reaction 7 for these alkenes. This conclusion is consistent with the fact that OH radical formation yields as low as 0.08 have been measured (for l,&butadiene).

-

-

and

[R,CH2C(0)O&]*

(ester channel)

[R~CH$(RZ)O~]~

-

R1CH2C(O)R2 + O(3P) [R1CH26(R2)O$

(O-atom elimination channel)

[RICH =C( R2)OOH$

(hydroperoxide channel)

with the ester (for R2 = H) and hydroperoxide channels leading to OH radical formation. Thus, Niki et al. (9)and Martinez and Herron (10)have proposed that the reaction of O3 with 2,3-dimethyl-2-butene proceeds by

0,+ (CH3),C=C(CH3),

-

-

[ozonide]* CH3C(0)CH3+ [(CH3),c0O1*

followed by stabilization of the [(CH3)&001 * biradical or its reaction through the hydroperoxide channel to produce OH radicals: