Theory-Based Mechanism Development - CiteSeerX

The OH-Initiated Atmospheric Oxidation of α-Pinene: Theory-Based Mechanism Development Contribution to subproject CMD-GPP

Jozef Peeters, Luc Vereecken, and Gaia Fantechi Department of Chemistry, University of Leuven, Celestijnenlaan 200F, B -3001 Leuven, Belgium e-mail: [email protected]

Summary A detailed mechanism was developed for the OH-initiated atmospheric oxidation of α-pinene in the presence of NOx, based solely on quantitative structure-activity relationships (SARs) and on theoretical quantum chemistry methods. On objective theoretical grounds, the fate of some 40 organic (oxy) radical key intermediates was predicted. Addition of OH onto αpinene's double bond accounts for ~90 % of the reaction and leads to chemically activated P1OH (~44%) and P2OH (~44%) radicals. The subsequent chemistry of these radicals is described, and the quantitative importance of the pathways leading to first-generation products such as pinonaldehyde, acetone, formaldehyde, formic acid, and nitrates is discussed. H-atom abstraction by OH from α-pinene is a minor route (~12%) but contributes importantly to the overall yield of formaldehyde. Total product yields were obtained by propagating the product fractions of each step in the mechanism. Introduction Several laboratory studies on the oxidation of α-pinene initiated by the OH radical are reported in the literature (Arey et al., 1990; Hatakeyama et al., 1991; Hakola et al., 1994; Aschmann et al., 1998; Noziè re et al., 1999; Fantechi, 1999; Orlando et al., 2000; Larsen et al., 2001; Wisthaler et al., 2001). While the reaction rate constant of the oxidation process is reasonably well known, reaction products and their yields, and especially reaction mechanisms, are still insufficiently known or understood. Moreover, there is still a lack of theoretical information to support the validity of the several mechanisms proposed. Recently, we performed a theoretical analysis of the OH-initiated oxidation of α-pinene to elucidate the mechanism of acetone formation (Vereecken and Peeters, 2000). Dibble (2001) very recently reported a theoretical study of the β C-C bond scission reactions of the alkoxy radicals resulting from the α-pinene + OH reaction. However, several sub-mechanisms in the OHinitiated oxidation of α-pinene still need to be clarified, among others the nature and quantitative importance of the pathways leading to products such as acetone, formaldehyde, and formic acid. Activities We developed a detailed and quantitative chemical mechanism for first-generation product formation from α-pinene, based solely on either objective chemical-kinetics knowledge, quantitative structure-activity relationships (SARs), or on first principles, i.e. ab initio / DFT (density functional theory)-calculated barrier heights, in combination with proven statistical rate theories (transition state theory and RRKM theory, for some cases in conjunction with master equation analysis (ME)). We made extensive use of SARs, including the previously developed site-specific SAR for addition of OH to (poly)alkenes (Peeters et al., 1994 and 1996), and the recently proposed mini SAR for H-abstraction from (poly)alkenes (Vereecken and Peeters, 2001). New SARs for barrier heights to dissociation, and for isomerization by 1,5-H shifts, of (OH and/or oxo-substituted) alkoxy radicals were developed and validated

(Peeters and Vereecken, 2002). These SARs predict barrier heights with an accuracy of 0.5-1 kcal/mol. However, for critical cases, barrier heights were computed by ab-initio / DFT. In this way, the competing isomerization and/or dissociation reactions of some 40 alkoxy radicals were predicted. Organic nitrates were estimated on the basis of a recently revised SAR, and for organic α- and β-hydroxynitrates, the predicted yields were taken as ~25% of the corresponding alkyl nitrate amount (Arey et al., 2001). Total yields of products were finally derived, both for normal laboratory conditions and for real atmospheric conditions, by propagating the relevant product fractions of each step in the mechanism (Peeters et al., 2001). Results and Conclusions Figure 1 summarizes the critical steps of the proposed α-pinene + OH reaction mechanism, including the major pathways involved in the formation of the relevant oxidation products and the yields thereof in laboratory conditions, where high NO concentrations are usually added to the system.

0.7% Other carbonyls 0.3 %Nitrates ~1% H-abstrac

O

1% Nitrates 11 % 89 % + O2 + NO - NO2

~9 % H-abstrac

RO1 60%

OH-addition

2% CH2O 0.5% CO2 0.8 % Nitrates 1.5% Other carbonyls

~2%

+ OH

H-abstrac

α -pinene ~44%

~44 %

OH-addition

40% 1,5-H OH

O

H OH

OH

P2OH + O2 + NO + O2 + NO 72 % - NO2

RO2

28 %

- NO2

0.5% CH2O 0.11%Nitrates 2.2% Other carbonyls

RO3 50% 2.1% Acetone 1.6% CH2O 1.6% CO2 0.5% Nitrates

50%

3.2% CH2O 3.5% CO2 0.8% Nitrates

3% total other carbonyls

3.1 % 7 % Nitrates - NO2 93 %

0.7 % Nitrates

RO8 + O2 + NO 0.5 % 11 % Nitrates 89 %

P1OH

+ O2 + NO

- NO2

50 %

+ O2 + NO

RO28

O O

O2 60%

8% CH3COOH 6% Acetone 6.2% CH2O 14% CO2 2% Nitrates 6 % Other carbonyls

11 %

50 % 0.7 % H 3% Nitrates OH + O2 + NO 97 % - NO2 Thermal P1OH

2.4 % Nitrates

- NO2 89%

40%

+ O2 Pinonaldehyde HO2 + NO 24.5 % 0.5 % 3 % Nitrates - NO2 97 % 8% CO2 1,7-H 2.2% HCOOH RO29 50% 2.3% Nitrates 3.5% Other carbonyls 50%

60 % anti 40 % syn

RO23

12.5% 87.5% 1,5-H 0.4 % Nitrates 2.3% CH2O

RO20 anti (60 %)

syn (40 %)

O2

+ Acetone 7.8 %

O

22 %

O

60 %

40% Pinonaldehyde + O2 11.2 % + NO 0.5 % 7% Nitrates 93 % - NO2

RO24 H

+ O2 + NO

78 % - NO2 2.9% CH2O 3% CO2 0.2% Nitrates 9 % Other carbonyls

HO2

H OH

O

O O2 HO2

OH 2% Nitrates

O

8-hydroxymenthen-6-one HCA168

11.3 %

FCH2

+

HCOOH 7%

RO21

Figure 1: General outline of the α-pinene + OH reaction mechanism (NOx) in laboratory conditions, showing the relevant steps, first-generation products and their molar yields.

Overall predicted α-pinene + OH product yields for usual conditions of laboratory experiments are shown in Table 1. It must be stressed again that these yields are based solely on validated quantitative SARs or on first-principle theoretical predictions, with the single exception of HC(O)OH, of which the yield in laboratory conditions was adopted.

Product Pinonaldehyde Acetone CH2O Organic nitrates HCA168 CO2 HC(O)OH CH3C(O)OH other carbonylsa

Theoretically Predicted Yields (%) Laboratory Atmosphere 35.7 59.5 17.9 11.9 18.8 12.6 19 13.1 11.3 11.3 30.7 8.7 9.2 0 8 0 25.9 16.4

Table 1: Molar product yields in the reaction between α-pinene and OH radicals predicted in this study, both for laboratory and real atmospheric conditions. a unsaturated dicarbonyls, unsaturated (poly)hydroxycarbonyls (cyclic and non-cyclic), methyl vinyl ketone, methacrolein, glycolaldehyde, malonaldehyde

Our SAR- and theory-based yields are generally in good agreement with literature data (Table 2), including those concerning formaldehyde and organic nitrates. Our theoretical CH2O yield of ~19 % is in agreement with the experimental yields reported by Noziè re et al. (1999) and by Orlando et al. (2000). The lower formaldehyde yield reported by Larsen et al. (2001) is probably due to CH2O photolysis by the intense UV light in their experimental set-up, and therefore cannot be used for comparison.

Reference Arey et al., 1990 Hatakeyama et al., 1991 Hakola et al., 1994 Aschmann et al., 1998 Fantechi, 1999 b Noziè re et al., 1999 Orlando et al., 2000 Larsen et al., 2001 Wisthaler et al., 2001 Peeters et al., 2001 (theoretical)

Product Yield (%) Pinonaldehyde Acetone CH2O 29 ± 5 a 78.5 28 ± 5 11 ± 2.7 5 87 ± 20 9±6 23 ± 9 5±2 19 ± 5 c 6±2 11 ± 3 8±1 34 ± 9 11 ± 2 35.7 17.9 18.8

HCOOH

6 7±2 28 ± 3c 9.2 (adjusted)

Table 2: Summary of the available literature regarding molar product yields in the reaction between α-pinene and OH radicals in the presence of NO x. a see also Noziè re et al.; b Unpublished results; c due to strong oxidative conditions.

Our predicted yield of 19 % total nitrates is in excellent accord with the total nitrates (18 ± 9%) quantified by Noziè re et al. (1999). The acetone yield derived in this study, however, is high, about double the average of the reported yields. As shown in Table 2, one of the major divergences regarding the product yield measurements of the α-pinene + OH reaction concerns the reported data for pinonaldehyde. In general, most of the laboratory data on the pinonaldehyde yield are significantly lower than that reported recently by Noziè re et al. (1999) and earlier by Hatakeyama et al. (1991). Whilst our "laboratory" pinonaldehyde yield of 35.7 % is in excellent agreement with the majority of the values listed in Table 2 and in good agreement with their average (42%), it differs markedly

both from the very high yield of 87 ± 20 % (Noziè re et al., 1999) and from the very low yield of 6 ± 2 % (Larsen et al., 2001). Acknowledgements This work was carried out in part in the frame of the ongoing Belgian research program on Global Change and Sustainable Development, funded via the Federal Office for Scientific, Technical and Cultural affairs. The authors are also indebted to the Fund for Scientific Research (FWO-Vlaanderen) and to the KULeuven Research Council (BOF) for continuing support.

References Arey J., R. Atkinson and S.M. Aschmann, J. Geophys. Res., 1990, 95, 18539. Arey J., S.M. Aschmann, E.S.C. Kwok and R. Atkinson, J. Phys. Chem. A, 2001, 105, 1020. Aschmann S. M., A. Reissel, R. Atkinson and J. Arey, J. Geophys. Res., 1998, 103, 25553. Dibble T. S., J. Am. Chem. Soc., 2001, 123, 4228. Fantechi G., Ph. D. Thesis, KULeuven, 1999. Hakola H., J. Arey, S.M. Aschmann and R. Atkinson, J. Atmos. Chem., 1994, 18, 75. Hatakeyama S., K. Izumi, T. Fukuyama, H. Akimoto and N. Washida, J. Geophys. Res., 1991, 96, 947. Larsen B.R., D. Di Bella, M. Glausius, R. Winterhalter, N.R. Jensen and J. Hjorth, J. Atmos. Chem., 2001, 38, 231. Noziè re B., I. Barnes and K.H. Becker, J. Geophys. Res., 1999, 104, 23645. Orlando J.J., B. Noziè re, G. Tyndall, G.E. Orzechowska, S.E. Paulson and Y. Rudich, J. Geophys. Res., 2000, 105, 11561. Peeters J., W. Boullart and J. Van Hoeymissen, 1994, Proceedings of EUROTRAC Symposium '94, p. 110. Peeters J., W. Boullart, V. Pultau and S. Vandenberk, 1996, Proceedings of EUROTRAC Symposium '96, , p. 471. Peeters J, L. Vereecken and G. Fantechi, Phys. Chem. Chem. Phys., 2001, 3, 5489. Peeters J. and L. Vereecken, 2002, manuscript in preparation. Vereecken L. and J. Peeters, J. Phys. Chem. A, 2000, 104, 11140. Vereecken L. and J. Peeters, Chem. Phys. Lett., 2001, 333, 162. Wisthaler A., N.R. Jensen, R. Winterhalter, W. Lindinger and J. Hjorth, Atmos. Environ., 2001, 35, 6181.

Theory-Based Mechanism Development of the OH-Initiated Atmospheric Oxidation of Pinonaldehyde Contribution to subproject CMD-GPP

Gaia Fantechi, Luc Vereecken and Jozef Peeters Department of Chemistry, University of Leuven, Celestijnenlaan 200F, B -3001 Leuven, Belgium e-mail: [email protected]

Summary A detailed mechanism was developed for the OH-initiated atmospheric oxidation of pinonaldehyde in the presence of NOx, based solely on quantitative structure-activity relationships (SARs) and on theoretical quantum chemistry methods. On objective theoretical grounds, the fate of some 30 organic (oxy) radical key intermediates was predicted. The pinonaldehyde + OH reaction proceeds via H-atom abstraction by OH, which can occur on six different carbon atoms. The subsequent chemistry of these radicals is described, and the quantitative importance of the pathways leading to first-generation products is discussed. Total product yields were obtained by propagating the product fractions of each step in the mechanism. Introduction Pinonaldehyde (3-acetyl-2,2-dimethyl-cyclobutyl-ethanal) is the major primary carbonyl product in the atmospheric oxidation of the ubiquitous monoterpene α-pinene, and has been reported in laboratory yields ranging from 6% to 87% (Arey et al., 1990; Hatakeyama et al., 1991; Hakola et al., 1994; Aschmann et al., 1998; Noziè re et al., 1999a; Orlando et al., 2000; Larsen et al., 2001; Wisthaler et al., 2001). Recently, we have reported a theoretically predicted yield of 60%, applicable to realistic atmospheric conditions (Peeters et al., 2001). The atmospheric fate of pinonaldehyde is still fairly uncertain, but its reaction with the OH radical should be the only loss process of importance besides photo-oxidation. There have been only a few reported kinetic studies on the pinonaldehyde + OH reaction (Glasius et al., 1997; Hallquist et al., 1997; Alvarado et al., 1998; Noziè re et al., 1999b), and measured rate values are in the range (4-9) × 10-11 cm3 molecule-1 s-1. Our reported theoretical rate for this reaction, in fair agreement with the most recent experimental values, is 3.5 × 10-11 cm3 molecule-1 s-1 (Vereecken and Peeters, 2001). To our knowledge, the only product study of the pinonaldehyde + OH oxidation reaction in the presence of NO is the one reported by Noziè re and co-workers (1999a), and formaldehyde and acetone were found to be major products, formed in important yields together with high yields of unidentified carbonyl compounds. A mechanism was also proposed, to account for the above-mentioned products. There is, however, a lack of theoretical information to support its validity. We therefore developed a (detailed) quantitative chemical mechanism for first generation product formation from pinonaldehyde. Total yields of products such as formaldehyde, acetone, norpinonaldehyde, CO2, nitrates, and other carbonyl compounds were also finally derived, both for normal laboratory conditions and for real atmospheric conditions, by propagating the relevant product fractions of each step in the mechanism (Fantechi et al., 2002). Activities We developed a detailed and quantitative chemical mechanism for first-generation product formation from pinonaldehyde, based solely on either objective chemical-kinetics knowledge, quantitative structure-activity relationships (SARs) (Peeters et al., 2002), or on first principles, i.e. ab initio / DFT (density functional theory)-calculated barrier heights, in

combination with proven statistical rate theories (transition state theory and RRKM theory, for some cases in conjunction with master equation analysis (ME)). The extensively used SARs include the recently proposed mini SAR for H-abstraction from (poly)alkenes (Vereecken and Peeters, 2001). New SARs for barrier heights to dissociation, and for isomerization by 1,5-H shifts, of (OH and/or oxo-substituted) alkoxy radicals were also developed and validated (Peeters and Vereecken, 2002). These SARs predict barrier heights with an accuracy of 0.5-1 kcal/mol. However, for critical cases, barrier heights were computed by ab-initio / DFT. In this way, the competing isomerization and/or dissociation reactions of some 30 alkoxy radicals were predicted. Organic nitrates were estimated on the basis of a recently revised SAR, and for organic α- and β-hydroxynitrates, the predicted yields were taken as ~25% of the corresponding alkyl nitrate amount (Arey et al., 2001). Results and Conclusions The oxidation of pinonaldehyde by OH is initiated by H-atom abstraction at different C-sites. The partial rate coefficients depend on the individual C-H bond strengths and on the type of resonance or (de)stabilization effect of the resulting radicals (Vereecken and Peeters, 2001). The main site of H-abstraction in pinonaldehyde is the aldehyde carbon atom Cc (~59%) (Figures 1a and 1b), followed by Cd (~23%) (Figure 2), Ce (~8.5%), Cg (~5.7%), Ci/j (~2.3%) and Cf (~1%). Pinonaldehyde O f

H2O

O i

O OONO

+ O2 + NO

c

e

g

O

O

OH

d

O

O

Nitrates 14 %

O 1,5-H-shift OH

- NO2

O

>90%

86 %

O

14 %

O2

HO2

O ~90%

+ O2 + NO

+ CO2

R'O4 (FCH2O) + O2 + NO

+ NO2

O R'O1

j

O

Nitrates

O

Thermal FCH2

FCH2

1,7-H-shift

O O

86 % - NO2

~10%

O

O

O

+ CH2O

OH

Figure 1b

OH

O R'O5 1,5-H-shift O OH OH 60% O2

40% + O2 + NO

O

O

73% - NO2

27% Nitrates

O OH

OH

+

HCOOH

OH

R'O6

HO2

O O OH

Figure 1a: The pinonaldehyde + OH reaction mechanism (NOx) in laboratory conditions: H-atom abstraction from the Cc atom and fate of the thermally stabilized FCH2 radical.

The key intermediate in the Hc-abstraction channel is the thermal FCH2 radical, that gives rise to the FCH2O oxy radical (Figure 1a). For this oxy radical, four reaction pathways were considered: H-abstraction by O2 (kO2~5×104 s-1), CH2O elimination (DFT barrier of 13.5 kcal/mol; kdiss≤5×103 s-1), 1,7-H shift (DFT barrier of 11 kcal/mol; kshift~3.5×103 s-1) and 1,5-

H shift (DFT barrier of 8.5 kcal/mol; kshift~5.5×106 s-1).This latter channels far outruns all others. For the "d" site, the C-H bond strength was found to be much lower than expected for a secondary abstraction site featuring a β-carbonyl function. This lower bond strength can be ascribed to a combination of a vinoxy-type resonance and a strain-enhanced hyperconjugation, and B3LYP-DFT calculations have clearly shown the influence of both effects and indicate that the H-abstraction from this site rivals that from the aldehyde site (Vereecken and Peeters, 2001). The complete and detailed mechanism can be found in Fantechi et al., 2002. Nitrates O FCH2

10%

10%

O

O

+ O2 - NO2 + NO 90 %

O

+

R'O2 O

+ O2 + NO

16% Nitrates 84% - NO2

CH2O

CO2 + CH3 - NO2 + O2 + NO CH3CO +

O O O R'O3

Figure 1b: H-atom abstraction from the Cc atom: prompt dissociation (~10%) of the FCH 2‡ radical. O

O OH O

H2O O pinonaldehyde-d-yl

+ O2 + NO

Nitrates 28% 72% - NO2

O

O

O O + CHO O

R'O7

Norpinonaldehyde

Figure 2: The pinonaldehyde + OH reaction mechanism (NO x) in laboratory conditions: H-atom abstraction from the Cd atom.

Table 1 shows total molar product yields in comparison with experimental yields available in the literature (Noziè re et al., 1999a); our theoretical yields, based solely on validated SARs or on first-principle theoretical predictions, were derived by propagating the product fractions of each step in the mechanism. The yields listed in Table 1 are for both "laboratory" and "real atmospheric" conditions. Product Acetone CH2O Organic nitrates CO2 HC(O)OH Norpinonaldehyde other carbonylsa

Theoretically Predicted Yields (%) Laboratory Atmosphere 17.6 17.6 19.8 19.8 32 27.8 83.3 83.3 11.4 0 16.6 16.6 56.8 84.6

Experimental (%) 15 ± 7 152 ±56 107

Table 1: Molar product yields in the reaction between pinonaldehyde and OH radicals predicted in this study, both for laboratory and real atmospheric conditions. The results are compared with the only experimental quantitative study available to date. a (di)(tri)(hydroxy)(di)(tri)carbonyls, unsaturated carbonyls, tricarbonyls

Our theoretically predicted primary acetone yield is in good agreement with the experimental yield reported by Noziè re et al. (1999a). However, we predict a much lower yield of CH2O. The higher yield reported by Noziè re et al. (1999a) might be due to further photolysis of carbonyl compounds by the 254 nm lamps used in the reaction system to photolyse H2O2 to OH. Acknowledgements This work was carried out in part in the frame of the ongoing Belgian research program on Global Change and Sustainable Development, funded via the Federal Office for Scientific, Technical and Cultural affairs. The authors are also indebted to the Fund for Scientific Research (FWO-Vlaanderen) and to the KULeuven Research Council (BOF) for continuing support. References Alvarado A., J. Arey, R. Atkinson, J. Atmos. Chem., 1998, 31, 281. Arey J., R. Atkinson and S.M. Aschmann, J. Geophys. Res., 1990, 95, 18539. Arey J., S.M. Aschmann, E.S.C. Kwok and R. Atkinson, J. Phys. Chem. A, 2001, 105, 1020. Aschmann S. M., A. Reissel, R. Atkinson and J. Arey, J. Geophys. Res., 1998, 103, 25553. Fantechi G., L. Vereecken and J. Peeters, "The OH-initiated atmospheric oxidation of pinonaldehyde: detailed theoretical study and mechanism construction", 2002, to be submitted. Glasius M., A. Calogirou, N.R. Jensen, J. Hjorth, C.J. Nielsen, Int. J. Chem. Kinet., 1997, 29, 527. Hakola H., J. Arey, S.M. Aschmann and R. Atkinson, J. Atmos. Chem., 1994, 18, 75. Hallquist M., I. Wanberg, E. Ljungstrom, Env. Sci. Technol., 1997, 31, 3166. Hatakeyama S., K. Izumi, T. Fukuyama, H. Akimoto and N. Washida, J. Geophys. Res., 1991, 96, 947. Larsen B.R., D. Di Bella, M. Glausius, R. Winterhalter, N.R. Jensen and J. Hjorth, J. Atmos. Chem., 2001, 38, 231. Noziè re B., I. Barnes and K.H. Becker, J. Geophys. Res., 1999a, 104, 23645. Noziè re B., M. Splitter, L. Ruppert, I. Barnes, K.H. Becker, M. Pons, K. Wirtz, Int. J. Chem. Kinet., 1999b, 31, 291. Orlando J.J., B. Noziè re, G. Tyndall, G.E. Orzechowska, S.E. Paulson and Y. Rudich, J. Geophys. Res., 2000, 105, 11561. Peeters J, L. Vereecken and G. Fantechi, Phys. Chem. Chem. Phys., 2001, 3, 5489. Peeters J. and L. Vereecken., 2002, manuscript in preparation. Vereecken L. and J. Peeters, "Enhanced H-atom abstraction from pinonaldehyde, pinonic acid, pinic acid, and related compounds: theoretical study of C-H bond strengths", 2001, accepted for publication in Phys. Chem. Chem. Phys. Wisthaler A., N.R. Jensen, R. Winterhalter, W. Lindinger and J. Hjorth, Atmos. Environ., 2001, 35, 6181.

The Acetic-Acid forming Channel in the Acetone + OH Reaction Contribution to subproject CMD-GPP

Sabine Vandenberk, Luc Vereecken, Jozef Peeters University of Leuven, Leuven, Belgium e-mail : [email protected]

Summary Recent atmospheric chemistry studies have highlighted the importance of acetone in the chemistry of the upper troposphere and the lower stratosphere (UTLS). An important loss process is the reaction with hydroxyl radicals. The rate constant of the reaction with OH has been measured recently down to ≈ 200 K (Wollenhaupt et al., 2000) and it was found that the temperature dependence is not described by a simple Arrhenius expression, suggesting that simple H-atom abstraction may not be the only reaction pathway, and that a second channel occurs, involving addition of OH on the carbonyl double bond, leading to the formation of acetic acid. In this work, an experimental and theoretical study of the acetone + OH reaction is presented, in an attempt to elucidate the product distribution. In a flow reactor – molecular beam sampling mass spectrometry investigation of the elementary reaction of acetone with OH at 290 K, no significant production of acetic acid could be measured; absolute calibrations result in a branching fraction of the OH-addition/CH3-elimination channel of at most ≈5%. In a theoretical study of the acetone + OH reaction, the potential energy profiles of the OHaddition/CH3-elimination channel, the direct H-abstraction channel, and the indirect Habstraction path via an hydrogen-bonded OH-acetone complex, were characterized. The barrier for OH-addition is found to be at least 2.5 kcal/mol higher than that for the Habstraction channels. Transition State Theory and RRKM - Master Equation calculations indicate that the OH-addition channel is negligible at all relevant atmospheric temperatures. Introduction Acetone is believed to be a major source of HOx radicals and an important precursor of peroxyacetylnitrate in the UTLS (e.g. Wennberg et al., 1998). Atmospheric removal of acetone occurs via photolysis, reaction with OH and deposition. In the UTLS, the major sink of acetone is photolysis, although recent measurements of the rate coefficient of the reaction of acetone + OH at low temperatures by Wollenhaupt et al. (2000) show that the title reaction can still account for about 30 % of the total loss in the upper troposphere. These researchers observed strong non-Arrhenius behavior for the overall reaction of OH with acetone; the negative temperature dependence below 240K was explained by the existence of two parallel reaction routes. According to Wollenhaupt et al. (2000) the reaction proceeds mainly via Hatom abstraction at higher temperatures (1a), while below room temperature OH-addition followed by methyl elimination dominates (1b). OH + CH3COCH3

→ →

CH3COCH2 + H2O (CH3)2C(O )OH → CH3 + CH3COOH

(1a) (1b)

In a subsequent publication, Wollenhaupt and Crowley (2000) presented indirect experimental evidence for the formation of CH3 in the reaction of OH with acetone. They derived a branching fraction k1b/k1 of 0.5 ± 0.15 at room-temperature, and of 0.3 ± 0.1 at 233K.

Vasvári et al.(2001) determined the branching ratio for the H-abstraction reaction channel to be 0.50 ± 0.04 at 298 K, by quantifying the acetonyl radicals (CH3COCH2 ) formed in the Habstraction channel. Vasvári et al. (2001) also characterized the two different reaction pathways (1a and 1b) by ab initio calculations. Both the OH-addition to the C-atom of the >C=O carbonyl group and the H-abstraction reaction were found to occur through loosely bound complexes preceding the transition states. Objectives It was aimed to elucidate the product distribution of the acetone + OH reaction by direct measurement of the CH3COOH that would be formed in the OH-addition/CH3-elimination channel and by characterization of the potential energy profiles for the different possible reaction channels. Experimental study The experimental set-up consists of a multi-stage fast-flow reactor and a Molecular Beam sampling Mass Spectrometry (MBMS) apparatus. Hydroxyl radicals were generated by reacting H atoms with NO2. Acetone was added to the OH-flow via a central injector. All experiments were carried out at 290 K and a total reactor pressure of 2 Torr using He as carrier gas. The contribution of reaction 1b was determined on an absolute basis by direct MBMSmeasurement of the acetic acid formed in the reaction of acetone with OH. The principle of this method is to react a known amount of OH with a very large excess of acetone to ensure fast and complete conversion of OH into primary products, and to measure the amount of CH3COOH formed. To convert the mass spectrometric CH3COOH+ signals, measured at 70 eV, into absolute concentrations, an instrumental calibration for acetic acid was carried out. For a given temperature, total pressure and average molecular mass of the gas in the reactor, the mass spectrometric output signal iX is directly proportional to the absolute concentration [X]s of the given species at the sampling point : iX = SX [X]s. The sensitivity SX for CH3COOH was determined by bubbling He through pure liquid acetic acid in a glass container, thus carrying the vapor into the flow reactor. The amount of acetic acid carried into the reactor per unit time and hence the concentration of gas phase acetic acid in the reactor was determined from the weight loss of the liquid acetic acid over a period of several hours. The weight loss measurement was only started when the adsorption/desorption equilibrium of acetic acid on the tubing and reactor walls was established. The acetic acid parent ion signal at m/z = 60 was corrected for the contribution of isotopes of acetone, which is added in a high concentration to the reactor. In almost all experiments the initial concentrations were [OH]i ≈ 1 × 1013 molecule cm-3 and [CH3COCH3] = 8.3 × 1015 molecule cm-3. Given the rather low rate constant of the acetone + OH reaction, k = 1.7 × 10-13 cm3 molecule-1 s-1 (Atkinson et al., 1999), versus the much higher expected rate constant for the secondary reaction CH3COCH2 + OH of ≈ 1 × 10-10 cm3 molecule-1 s-1, it can be calculated that about 10% of the OH will disappear by secondary reactions (taking into account also the loss of CH3COCH2 by its reaction with the remaining H), which has to be taken into account in evaluating the acetic acid yield. To reduce the OHloss to 2% would require an initial acetone concentration as high as ≈ 4 × 1016 molecules cm-3 (≈ 1.2 Torr), which is unrealistic in our conditions. The secondary reactions of acetic acid with OH and other radicals are all sufficiently slow to assume that no loss occurs. The measured instrumental sensitivity for acetic acid of 1.5 × 10-11 µV cm3 molec-1, together with the noise on the m/z = 60 acetone isotopes signal, implies that in the given experimental

conditions, with [OH]i = 1 × 1013 molecule cm-3 and [CH3COCH3] = 8.3 × 1015 molecule cm-3, acetic acid yields as low as 2-3% can be measured. Five separate determinations of the acetic acid yield of the acetone + OH reaction were carried out. In most of these experiments, the CH3COOH signal from the reaction of acetone + OH was below the detection limit, implying a branching fraction for the additionelimination path ≤ 3%. Only in one experiment, with a higher initial OH concentration, a statistically significant signal could be measured. For an initial OH concentration of 1.5 × 1013 cm-3, a CH3COOH+ signal of 5.3 ± 9.2 µV was measured. At the stated sensitivity for CH3COOH of 1.5 × 10-11 µV cm3 molec-1 and at the given [OH]i this gives a result for the contribution of the addition–elimination channel of 2.5 ± 4 %, including a correction for OHloss by secondary reactions. Theoretical study The potential energy surface of the acetone + OH reaction was characterized first at the B3LYP-DFT/6-31G(d,p) level of theory, followed by calculations on the B3LYP-DFT/6311++G(d,p) level of theory, and single point CCSD(T)/6-311++G(2d,2p) calculations on the B3LYP/6-311++G(d,p) geometries. All calculations were performed using the Gaussian quantum chemical program (Frisch et al., 1998). The B3LYP-DFT relative energies were corrected for basis set superposition errors using the counterpoise method by Bernardi et al. (Van Duijneveldt et al., 1994; Boys and Bernardi, 1970). Three pathways were characterized : addition of OH on the carbonyl-C, direct abstraction of a hydrogen, and formation of an initial hydrogen-bonded complex followed by H-abstraction. At all levels of theory used, it was found that addition of OH to the carbonyl-C of acetone faces a barrier of 6 ± 0.5 kcal/mol. The (CH3)2C(O )OH intermediate formed is calculated at our best level of theory to lie 13.8 kcal/mol below the reactants CH3COCH3+OH. This intermediate can eliminate a CH3 radical over a barrier of only 8.5 kcal/mol, forming acetic acid as coproduct. Transition State Theory (TST) calculations on the rate of OH addition, using our B3LYP/6-311++G(d,p) harmonic vibrational data and the single point CCSD(T) energy barrier, lead to a rate coefficient at room temperature of k(298 K)=1 × 10-18 cm3 molecule-1 s-1. Compared to the total rate coefficient for the reaction of acetone + OH, this k(298K) value implies a contribution of the addition channel of less than 6 × 10-4 % at room temperature; because the barrier for OH-addition is substantially higher than that for H-abstraction, the contribution of the OH-addition will decrease even further at lower temperatures. The fate of the chemically activated (CH3)2C(O )OH intermediate formed in the addition was examined in an RRKM - Master Equation analysis. It was found that the rate of dissociation of the chemically activated (CH3)2C(O )OH intermediate to acetic acid + CH3 is much faster (kdiss ≈ 1 × 1011 s-1) than collisional energy loss at 1 atmosphere or lower, such that only a negligible fraction of the (CH3)2C(O )OH intermediate will be stabilized, only to undergo thermal decomposition to the same products on a (sub-)µs time-scale. Thus, the fast and quantitative dissociation of (CH3)2C(O )OH to acetic acid in all relevant conditions (i) implies that the acetic acid forming channel should be independent of pressure, and (ii) makes the CH3COOH production a rigorous measure of the addition fraction in the reaction of acetone with OH radicals. Two pathways for H-abstraction from acetone by OH radicals were found, differing in whether or not the OH radical first forms a hydrogen bonded complex with acetone. The Hbonded OH-acetone complex is made by a barrierless association; the complex-stabilization energy of about 5 kcal/mol at our two highest levels of theory agrees with the acetone+OH

result of 6.3 kcal/mol by Vasvári et al. (2001). After hydrogen-abstraction in this complex, an H-bonded complex (-26 kcal/mol) between the acetonyl+H2O products is formed, but the energy released in the H-transfer should readily break this H-bond of about 5 kcal/mol, leaving the separated acetonyl and H2O fragments as end products. For direct H-abstraction by OH from acetone, i.e. without prior formation of an H-bonded acetone-OH complex, direct formation of the separated H2O + acetonyl fragments was found. At all levels of theory the Habstraction transition states are calculated to lie at least 2.5 kcal/mol below the OH-addition transition state. Since, moreover, the TST frequency factor for direct H-abstraction is some three orders of magnitude higher than that for OH-addition, H-abstraction should be much faster than OH-addition, even at elevated temperatures. Conclusions The quantum chemical data, combined with Transition State Theory calculations, indicate that the OH-addition channel in the acetone+OH reaction only contributes for a very minor fraction at room temperature and below. This conclusion is substantiated in the experiments, in which no significant signal could be detected for acetic acid as a product of the acetone + OH reaction. CH3COOH should be a very good indicator for the addition fraction, given the complete conversion of the (CH3)2C(O•)OH adducts to acetic acid as shown in the RRKMME calculations, the high sensitivity for detecting acetic acid in our experimental setup, and the absence of reactions removing acetic acid at an appreciable rate. Hence, it is our conclusion that the reaction of acetone + OH proceeds for (almost) 100% by H-abstraction, at all tropospheric and stratospheric temperatures. Acknowledgments The authors thank the Belgian federal office for Science, Technology and Culture for support of this research in the frame of the Belgian programme on Global Change and Sustainable Development. SV is indebted to the Flemish Institute for Science and Technology (IWT) for granting her a doctoral fellowship. LV and JP thank the Fund for Scientific Research Flanders (FWO-Vlaanderen) and the KULeuven Research Council (PDM and BOF Fund) for continuing support. References Atkinson, R., D. L. Baulch, R. A. Cox, R. F. Hampson, J. A. Kerr, M. J. Rossi, J. Troe, 1999, Evaluated kinetic and photochemical data for atmospheric chemistry, organic species : supplement VIII. J. Phys. Chem. Ref. Data 28, 191-393. Boys, S. F. and F. Bernardi, 1970, The calculation of small molecular interactions by the differences of separate total energies. Some procedures with reduced errors. Mol. Phys. 19, 553-566. Frisch, M. J. et al., 1998, Gaussian 98, Revision A.9, Gaussian Inc.: Pittsburgh PA. Van Duijneveldt, F. B., J. G. C. M. Van Duijneveldt-van de Rijdt and J. H. Van Lenthe, 1994, State of the art in Counterpoise theory. Chem. Rev. 94, 1873-1885. Vasvári, G., I. Szilágyi, A. Bencsura , S. Dó bé , T. Bérces, E. Henon, S. Canneaux , F. Bohr , 2001, Reaction and complex formation between OH radical and acetone. Phys. Chem. Chem. Phys. 3, 551-555. Wennberg, P. O. et al., 1998, Hydrogen radicals, nitrogen radicals, and the production of O3 in the upper troposphere. Science 279, 4-53. Wollenhaupt, M., J. N. Crowley, 2000, Kinetic studies of the reactions CH 3 + NO2 → products, CH3O + NO2 → products, and OH + CH3C(O)CH3 → CH3C(O)OH+CH3, over a range of temperature and pressure. J. Phys. Chem. A 104, 6429-6438. Wollenhaupt, M., S. A. Carl, A. Horowitz, J. N. Crowley, 2000, Rate coefficients for reaction of OH with acetone between 202 and 395 K. J. Phys. Chem. A 104, 2695-2705.