Sulfur dioxide oxidation induced mechanistic branching and particle formation during the ozonolysis of b-pinene and 2-butene w

Recent studies have suggested that the reaction of stabilised Criegee Intermediates (CIs) with sulfur dioxide (SO2), leading to the formation of a carbonyl compound and sulfur trioxide, is a relevant atmospheric source of sulfuric acid. Here, the significance of this pathway has been examined by studying the formation of gas phase products and aerosol during the ozonolysis of b-pinene and 2-butene in the presence of SO2 in the pressure range of 10 to 1000 mbar. For b-pinene at atmospheric pressure, the addition of SO2 suppresses the formation of the secondary ozonide and leads to highly increased nopinone yields. A complete consumption of SO2 is observed at initial SO2 concentrations below the yield of stabilised CIs. In experiments using 2-butene a significant consumption of SO2 and additional formation of acetaldehyde are observed at 1 bar. A consistent kinetic simulation of the experimental findings is possible when a fast CI + SO2 reaction rate in the range of recent direct measurements [Welz et al., Science, 2012, 335, 204] is used. For 2-butene the addition of SO2 drastically increases the observed aerosol yields at higher pressures. Below 60 mbar the SO2 oxidation induced particle formation becomes inefficient pointing to the critical role of collisional stabilisation for sulfuric acid controlled nucleation at low pressures.

The atmospheric abundance of sulfuric acid (H 2 SO 4 ) is closely related to observations of new particle formation. 1The dominant H 2 SO 4 forming pathway of atmospheric relevance is assumed to start with the oxidation of sulfur dioxide (SO 2 ) to sulfur trioxide (SO 3 ) by the OH radical. 2 Low-volatility organic compounds are believed to contribute to the subsequent growth of the clusters to nanoparticles. 3 Such highly oxidized species are formed in the atmospheric degradation of many hydrocarbons from biogenic and anthropogenic sources by OH, NO 3 and ozone.
5][6][7][8] Hatakeyama et al. analysed the particle formation during the ozonolysis of 2-butene and other alkenes in the presence of SO 2 . 6An intriguing result of this early work was the reduction of sulfuric acid yields in aerosol samples from trans-2-butene with decreasing pressure, reaching zero at the lowest pressures.The authors concluded that SO 2 is oxidized by stabilised Criegee Intermediates (CIs).20 years later Berndt et al. concluded for the ozonolysis of 2-butene (as well as a-pinene and tetramethylethylene) in the presence of SO 2 that the oxidation of SO 2 by OH is the dominant pathway for H 2 SO 4 formation but at the same time reported significant contributions from CI + SO 2 .We note that Berndt et al. used much lower reactant concentrations being close to atmospheric conditions.For cyclohexene + O 3 we found in previous work a relation of generated particle number densities and SO 2 concentrations, which were at low H 2 SO 4 /SO 2 concentrations consistent with SO 2 + OH being the only source of sulfuric acid. 9n a recent paper the first direct kinetic measurements of the reaction of the C1-CI (CH 2 OO) with SO 2 were reported employing single photon, near threshold ionization for specific CI detection. 10The measured rate coefficient was significantly faster than previously believed (3.9 Â 10 À11 cm 3 s À1 compared to 0.04 to 8 Â 10 À13 cm 3 s À1 , see discussion in ref. 10), providing experimental evidence for theoretical work of Kurte´n et al. on the oxidation of SO 2 to SO 3 by gas-phase organic oxidants. 11These results suggested that CI + SO 2 may contribute significantly to atmospheric H 2 SO 4 production.This interpretation is bolstered by a current combined field and laboratory study. 12owever, due to the experimental approach the direct kinetic measurements were performed at 4 Torr and have not been extended to substituted CIs at atmospheric pressure. 13The aim of the present study is to bridge this gap by studying the CI + SO 2 reaction in the context of pressure dependent mechanistic branching and aerosol formation during alkene ozonolysis.The alkenes under study are b-pinene and 2-butene (mixture of cis and trans), which produce predominantly 14 or solely substituted CIs in the reactions with ozone.
Direct kinetic measurements are highly desirable but unfortunately limited by the availability of suitable precursors for the photodissociation induced specific preparation of larger CIs.5][16] In the case of b-pinene, a recent extensive theoretical study provided crucial kinetic information which is needed for robust modelling of the first reaction steps. 14For the ozonolysis of 2-butene the results of absolute rate coefficient determinations of reactions involving the C2-CI (CH 3 CHOO) are available. 17his journal is c the Owner Societies 2012 All ozonolysis experiments were performed in our static variable pressure reaction chamber with a volume of 64 L and two premixing chambers of 40 L volume.One of them is used as an ozone reservoir and is equipped with UV optics for a continuous measurement of O 3 concentrations.Concentrations of reactants and the product yields were determined using infra-red spectroscopy (Bruker IFS 66, 12 m path length in the cell).SO 2 concentrations were determined by means of IR spectroscopy and volumetry as described in detail in a recent paper. 9A particle classifier (TSI SMPS 3936 with NDMA (3085) and LDMA (3081) as well as a 3022 CPC (condensation particle counter)) was used for particle measurements (operation parameters: 2 litres per minute (lpm) sheath flow and 0.2 lpm sample flow).The pressure variation was achieved by expanding the reactants from the premixing chambers into the evacuated reaction cell within 3 s.For starting pressures between 25 and 1000 mbar this procedure results in final pressures between 10 and 450 mbar.For experiments at higher pressures additional synthetic air is filled in the reaction cell directly after the expansion (approx.15 s filling time).The fast reactant mixing allows the use of relatively high initial concentrations which effectively suppress the heterogeneous loss of reactants and products to the wall.Further details on the apparatus and the applied procedures are given in recent papers. 9,16All chemicals used were of commercial grade (see ESIw).
First, we discuss the results for the ozonolysis of b-pinene.This terpene is of atmospheric relevance and many experimental and theoretical studies have been published on this reaction (ref.14, 16, 18-22 and references therein) with the focus on both stable gas phase product formation and the production of SOA.The extensive theoretical study of Nguyen et al. revealed that two distinct, non-interconvertible conformers of C9-CI (-COO function either points to or away from internal ring, see ref. 14)  are formed.Only 5% CH 2 OO yield was predicted.These properties of b-pinene ozonolysis indicate that this is a suitable system for studying mechanistic features related to the reactions of larger substituted, biogenic CIs with SO 2 .The key observations for such experiments are illustrated in the upper panels of Fig. 1.Addition of 3 ppm SO 2 to an initial mixture of 9 ppm O 3 and 30 ppm b-pinene has two significant effects on the IR final product spectrum: the additional formation of a carbonyl compound and the reduced formation of a product with a characteristic double peak near 1100 cm À1 (features A and B in the IR difference spectrum, upper panel of Fig. 1).Feature B was previously assigned to a secondary ozonide (SOZ). 16,19The carbonyl compound is most likely nopinone, which is deduced from the fit of the CQO stretch band and other characteristic nopinone features when compared to a pure product spectrum (see Fig. S1 in the ESIw).In addition, we observe a complete consumption of SO 2 at 1 bar (see the lower trace in the left middle panel of Fig. 1).At 10 ppm initial concentration unreacted SO 2 is observed in the end product spectrum.At this initial SO 2 concentration the characteristic SOZ features are completely suppressed (see the right middle panel of Fig. 1).All these findings can be explained by the direct oxidation of SO 2 to SO 3 via the fast reaction with a stabilised CI. 7,10,11 If SO 2 is predominantly consumed via CI + SO 2 the picture will change at low pressure where a much lower fraction of CIs is stabilised enough to take part in bimolecular reactions. 14The upper trace in the left middle panel of Fig. 1 belongs to an experiment conducted at 50 mbar with an initial SO 2 concentration of 3 ppm.The comparison with the pure substance spectrum of the same amount of SO 2 indicates that only a minor fraction is consumed.This finding strongly supports the interpretation that CI + SO 2 is the dominant pathway for SO 2 consumption at 1 bar.Similar results were obtained for the 2-butene + O 3 reaction.In the lower panels of Fig. 1 the key results of these experiments are illustrated.At an initial concentration of 3 AE 0.2 ppm SO 2 a consumption of 1.3 AE 0.2 ppm SO 2 (lower left panel of Fig. 1) and additional formation of acetaldehyde are observed (lower right panel of Fig. 1).The less efficient consumption of SO 2 is consistent with a lower fraction of stabilised CIs for 2-butene ozonolysis (see e.g.ref. 6    24 ).An in-depth analysis of this interesting effect should include the isomer distinction of 2-butene, a rigorous calibration of the additional carbonyl compound formation as a function of pressure and -where possible -relative to adduct formation. 7This is beyond the scope of this communication and will be addressed in a subsequent full length paper.Several test experiments were performed for all alkenes using cyclohexane as an OH scavenger.No measurable effect on the SO 2 consumption was observed.
In the next section the experimental results are compared to predicted product yields derived by kinetic modelling of the ozonolysis reaction.To this end the mechanism applied in the preceding work 9 has been adopted to the alkenes under study using kinetic data provided by Nguyen et al. 14 for b-pinene and Fenske et al. for 2-butene. 17We have now included the CI + SO 2 reaction which was not considered for cyclohexene + O 3 due to the absence of stabilised CIs. 15 Further details on the kinetic modeling are given in the ESI.w We simulated the ozonolysis of b-pinene for the conditions given in the caption of Fig. 1 with an initial amount of 3 ppm SO 2 using different rate coefficients for the CI + SO 2 reaction.A complete consumption of SO 2 as observed in the experiment is only predicted when a value larger than 1 Â 10 À11 cm 3 s À1 is applied (see Fig. 2).Use of a rate coefficient of 1 Â 10 À11 cm 3 s À1 would according to the simulation result in a final SO 2 concentration of 0.3 ppm, which is not observed in the experiment (see the left middle panel of Fig. 1).Similar results were obtained for 2-butene + O 3 and the partial consumption of SO 2 (see Fig. S3 in the ESIw).Also for this alkene the best agreement between experimental findings and model predictions is found when a high value for k(CI + SO 2 ), larger than 1 Â 10 À11 cm 3 s À1 , is used.These findings suggest that the direct measurements of Welz et al. for k(CI + SO 2 ) at 4 Torr give a suitable estimate for larger CIs at higher pressure.However, we note here that the modelling results depend to some degree on the yields of stabilised CIs.If the added amount of SO 2 exceeds the applied fraction of stabilised CIs the simulation predicts that SO 2 is not completely consumed for simple stoichiometric reasons.This implies an intrinsic uncertainty of model predictions regarding the final SO 2 concentrations.We further note that the conditions of our experiment (relatively high initial reactant concentrations) favor bimolecular chemistry over the consumption of CIs by unimolecular reactions.Assuming unimolecular rates in the range of 1 to 100 s À1 would only slightly change the predicted consumption of SO 2 in case of b-pinene, for 2-butene the influence would be larger.However, Nguyen et al. 14 found a consistency of their predicted and experimentally determined 17 isomerisation rates of thermalised CIs for the two alkenes under study.In the light of these considerations we hesitate to use a simple best fit of k(CI + SO 2 ) to match the point where the final SO 2 concentration is below the detection limit to pinpoint the rate coefficient.Nonetheless, the combined results of experiments and kinetic modelling suggest that k(CI + SO 2 ) is likely above 1 Â 10 À11 cm 3 s À1 and clearly above 3 Â 10 À12 cm 3 s À1 for both alkenes under study.Independent evidence for this lower limit in the case of b-pinene comes from the observed suppression of SOZ formation (see Fig. 1 and Fig. S4 in the ESIw) where the relative rate of CI + carbonyl and CI + SO 2 is probed.The effective suppression shows that the latter reaction is much faster.However, we have to state that the evidence from kinetic modelling is limited in case fast, yet unexplored SO 2 consuming reactions exist.Furthermore, alkene specific rates have to be considered (see ref. 12 for results on a-pinene and limonene).A critical point for both kinetic modelling and experimental studies is the unimolecular rate coefficients of isomerisation reactions of stabilised CIs.Assuming in the kinetic simulation significantly lower unimolecular reaction rates than reported by Fenske et al. 17 would in case of 2-butene allow a consistency with a lower range of values for k(CI+SO 2 ).We note that measurements at low initial reactant concentrations, which employ the detection of the product H 2 SO 4 , are more closely linked to unimolecular CI chemistry.Here, the rate coefficient k(CI + SO 2 ) is measured relative to the CI loss rate.Another important point is that measuring H 2 SO 4 formation instead of CI or SO 2 consumption would lead to lower k(CI + SO 2 ) values when CI-SO 2 adducts are stabilised to a significant degree (see also discussions in ref. 12, 14 and 26).
In the following the key results on aerosol yields as a function of SO 2 addition will be discussed.All experiments were performed at 450 mbar to achieve a high reproducibility by fast reactant mixing. 9For 2-butene the observed SO 2 consumption and acetaldehyde formation suggest an effective production of H 2 SO 4 .The kinetic model predicts H 2 SO 4 formation via CI + SO 2 to be at least two orders of magnitude faster than via OH + SO 2 .This should manifest itself in a large influence of SO 2 addition on the observed aerosol formation.In the left panel of Fig. 3  This journal is c the Owner Societies 2012 2-butene ozonolysis (1 ppm ozone, 10 ppm 2-butene) as a function of the added amount of SO 2 are shown.In the absence of SO 2 no particle formation is observed, above 0.0004 Pa (4 ppb at 1 bar) SO 2 aerosol formation starts and is highly intensified when the fraction of SO 2 in the mixture is further increased.This finding suggests that SO 2 oxidation by CIs provides an efficient and kinetically controllable gas phase source of sulfuric acid, which can be used to study H 2 SO 4induced nucleation dynamics in laboratory experiments.When reducing the pressure a slight decrease in particle numbers between 450 and 60 mbar is found.This effect can be related to the pressure dependence of stabilised CI yields (see results for the related alkenes, ethylene and tetramethylethylene 15,25 ).Below 60 mbar, however, the particle formation is significantly reduced which is a similar finding to our previous work on SO 2 oxidation induced aerosol formation during cyclohexene ozonolysis, pointing to the critical role of collisional stabilisation for H 2 SO 4 induced particle formation. 9The effect on observed particle numbers upon adding SO 2 during b-pinene ozonolysis is much weaker because particles are effectively formed in the absence of SO 2 (see Fig. S5 in the ESIw). 16ifferent from a-pinene ozonolysis 9 we find additional aerosol mass build up below 100 mbar for b-pinene which again can be explained by the much higher abundance of stabilised CIs in the case of b-pinene at low pressure. 14,15n summary, we have characterized several effects of SO 2 addition on final gas phase product and aerosol formation during the ozonolysis of b-pinene and 2-butene.The experimental findings, namely the efficient SO 2 consumption, additional carbonyl and suppressed SOZ formation in combination with the kinetic modelling result in a consistent picture when a fast reaction of stabilised CIs with SO 2 at ambient pressure is assumed.Thus the direct kinetic results of Welz et al. on the CH 2 OO + SO 2 reaction at low pressure 10 seemingly provide a reliable estimate for larger, substituted CIs at tropospheric pressure.At the same time we have to be aware that uncertainties concerning the unimolecular reaction rates of stabilised CIs and an incomplete understanding of H 2 SO 4 formation pathways may imply systematic errors which cause discrepancies between different experimental approaches. 10,12,14,17,26Therefore additional direct measurements of CI + SO 2 reactions rates are desirable.The observed effects on particle formation under SO 2 addition give additional experimental evidence for both the important role of the CI + SO 2 reaction in atmospheric aerosol formation 12 and the inefficiency of sulfuric acid based nucleation at low pressures. 9

Fig. 1
Fig. 1 IR spectra of final products from b-pinene (upper and middle panels) and 2-butene (lower panels) ozonolysis at initial concentrations of 30 AE 2 ppm alkene and 9 AE 0.5 ppm ozone at 295 AE 0.5 K. Upper panel: difference product spectrum presence-absence of 3 ppm SO 2 .Middle panels: (left) complete (lower trace, 1 bar) and minor (upper trace, 50 mbar) consumption of SO 2 at 3 ppm initial concentration; middle traces show spectra of 0.3 and 3 ppm SO 2 for comparison.(right) Suppression of the characteristic band of the secondary ozonide upon SO 2 addition.Lower panels: (left) partial consumption of SO 2 at 3 AE 0.2 ppm initial concentration.(right) Additional acetaldehyde formation upon SO 2 addition in the difference IR product spectrum (presence-absence of 3 AE 0.2 ppm SO 2 ).Published on 23 October 2012.Downloaded by University of Goettingen on 12/06/2014 15:22:37.

Fig. 2
Fig. 2 Simulated mole fraction profiles of SO 2 for an initial mixture of 30 ppm b-pinene and 9 ppm ozone at 298 K and 1 bar using different rate coefficients for CI + SO 2 .
and 17 and discussion therein).This finding suggests that the fast CI + SO 2 reaction can give direct access to stabilised CI yields on the basis of a titration with SO 2 .To bolster this conclusion we performed similar experiments using ethylene at 1 bar and initial concentrations of 100 AE 5 ppm C 2 H 4 , 8.5 AE 0.5 ppm O 3 and 6 AE 0.3 ppm SO 2 (see Fig. S2 in ESIw).The observed consumption of 4.2 AE 0.3 ppm SO 2 indicates an yield of 50 AE 15% of stabilised CIs in agreement with the extensive study of Horie and Moortgat 23 but larger than other