Far~d~y 1995,100 39-54 D~SCUSS. Role of Adducts in the Atmospheric Oxidation of Dimethyl Sulfide Stephen B. Barone Andrew A. Turnipseed and A. R. RavishankaraT National Oceanic and Atmospheric Administration Aeronomy Laboratory 325 Broadway Boulder CO 80303 USA and Department of Chemistry and Biochemistry and the Cooperative Institute for Environmental Sciences University of Colorado Boulder CO 80309 USA Based on the results from recent laboratory studies and field observations we suggest that adduct formation may play a key role in the atmospheric oxidation of dimethyl sulfide (DMS). Experimental evidence for the forma- tion of the DMS * OH and CH,SO * 0 adducts are described and used in a simple ‘box model’ to calculate the ratio of [MSA]/[S0,2-] observed in the atmosphere.These calculated ratios are compared with those inferred from field measurements. Based on these comparisons we suggest that the changes with temperature in the mechanism of the OH + DMS reaction one of the primary initiation steps in atmospheric DMS oxidation are unlikely to be the source of the variation in the [MSA]/[S0,2-] ratio with temperature. We suggest that the competition between the thermal decom- position of adducts such as CH,SO and their bimolecular reactions with atmospheric species is the likely cause of the variation. Finally a few observ- ations needed to better our understanding of the DMS oxidation are identi- fied. Dimethyl sulfide (DMS CH,SCH,) emitted by oceanic phytoplankton is the major natural source of sulfur in the troposhere.It is a steady source unlike the sporadic volcanic emissions. Much effort has been focussed during the past two decades on understanding the atmospheric oxidation (see for example ref. 1) of DMS via atmo-spheric field measurements modelling studies and laboratory (direct and chamber) investigations of the kinetics and product identities. However many uncertainties still remain in quantifying the eventual fate of DMS in the atmosphere. This uncertainty is primarily due to a lack of understanding of the oxidation mechanism. The extent of branching to stable end-products and the variation of this branching with temperature and composition are unclear. Such information is essential for evaluating the role of DMS in the atmosphere and assessing its impact on Earth’s climate.The likely major end-products of DMS oxidation are species such as sulfur dioxide sulfur trioxide and methanesulfonic acid (CH,SO,H MSA). SO and MSA have been measured in chamber studies and in the The effect of DMS on the climate is critically dependent on the production of gas-phase sulfuric acid and new particles. If SO and/or SO3 is produced new particles can be produced because both will lead to gas-phase H,SO, at least part of the time. In contrast the formation of MSA is not expected to lead to new particles. There are subtle but important differ- ences even between SO2 and SO production. If SO2 is produced in DMS oxidation it t Address correspondence to this author at NOAA/ERL R/E/AL2 325 Broadway Boulder CO 80303 USA.39 Atmospheric Oxidation of Dimethyl Sulfide is expected to lead to a more diffuse distribution of H2S04 while SO3 generation should lead to immediate conversion into H2S04. In the latter case the emitted DMS will likely influence only the marine boundary layer. This expectation is based on the lifetime of SO3 being very short owing to reaction with water and that of SO being a few days. In addition if the DMS oxidation mechanism is well characterized and the variation of the yields of end-products (and intermediates) with temperature and atmospheric composition are quantified atmospheric abundances of products will provide insight into the concentrations of oxidants and temperature. This would be valuable for predict- ing the future as well as for understanding the pre-industrial atmosphere.In the latter case DMS must have been the predominant source of sulfur. Interpretation of ice-core data would also be greatly facilitated by such an understanding of the DMS oxidation mechanism. In this paper we argue that sulfur-containing molecules and free radicals are prone to forming adducts with OH and 02,and the thermal stability relative to their bimolecular reactions is responsible for the changes in the branching between MSA and sulfate in the atmosphere. This thesis rests on the simple fact that unimolecular decom- position reactions important in the atmosphere have very large activation energies and hence the decomposition rate can change by orders of magnitude for small changes in temperature.In contrast most atmospheric bimolecular reactions are not very sensitive to temperature. When the product branching depends on the competition between unimolecular reactions and bimolecular reactions it can change rapidly with tem-perature. To show the importance of these adducts we have carried out a simple modelling analysis of the DMS atmospheric oxidation based on kinetic and product studies. Field measurement data have been used to define the range of possibilities for the parameters. Thus to some extent our approach links observations from the atmosphere and ice- cores with the results of the numerous laboratory studies via a proposed DMS oxidation mechanism. Surprisingly we can draw quite a few conclusions about the DMS oxida- tion mechanism and identify future studies that will be most productive.Laboratory Studies of Adduct Formation in DMS Oxidation Laboratory studies of the reactions involved in the atmospheric oxidation have been carried out by many The rate coefficients for the initiation reactions needed to quantify the atmospheric lifetime of DMS are now reasonably well defined However the subsequent reactions which influence the end-product yields are not well understood. Here we briefly discuss the approach we have taken in our laboratory to understand the mechanism of the OH-DMS reaction and the reactions of the CH3S radical. These studies are of particular significance since both involve formation of adducts. Based on these studies we speculate on the bond strengths of other possible adducts in DMS oxidation.Experimental All measurements discussed here were carried out using the technique of pulsed laser photolysis-pulsed laser-induced fluorescence (PLP-PLIF). The PLP-PLIF apparatus and the experimental procedures used to measure rate coeficients have been described in detail else~here.'~.'' The OH and CH,S radicals were detected by LIF in an excess of other reagents. OH was monitored by exciting the (A 2C+ u' = 1) +(X 211,u" = 0) band at ca. 282.1 nm using a pulsed XeCl excimer laser pumped dye laser (0.10-1.2 mJ pulse-' cmP2 pulse width CCI. 8 ns 10 Hz). CH,S radicals were detected by exciting the (A 2A, v; = 1) + (X 2E ulj = 0) transition at 371.4 nm. OH radicals were typically generated cia the photolysis of H202 at 248 nm while CH,S radicals (when S.B. Barone A. A. Turnipseed and A. R. Ravishankara they were reactants) were generated by the 248 nm photolysis of DMS or DMDS (CH,SSCH, dimethyl disulfide). The OH (or CH,S) temporal profile was obtained by measuring the LIF signal at various delay times between the photolysis and probe lasers (5 ps-50 ms). The detection limit defined as SIN = 1 where S is the time-zero signal and N is equal to twice the standard deviation in the mean value of the measured back- ground was typically ca. 4 x lo8 molecules cm- for OH and 4 x lo9 molecule cm-3 for CH,S in 100 Torr He upon integration of 100 laser shots. All experiments were carried out by slowly flowing the gas mixture through the reactor.The concentrations of the individual stable gases were determined from pressure and flow-rate measurements or via UV absorption measurements. Status of Results in the OH + DMS Oxidation The reaction of OH with DMS is a key step for the initiation of DMS oxidation. This is also the first place in the mechanism where branching to yield different products can take place.g-' ' Recently we have focussed on the rate coeficients and products of the 9 OH reaction with DMS. The following is a short synopsis of experiments which have been recently completed in our laboratory. A complete description of the experiments and the obtained results will be published in the near future along with detailed dis- cussion of the chemical simulations involved in obtaining most of the values reported here.Along with some recent results from other laboratories these studies lay the frame- work for understanding the branching in the initiation step in the oxidation of DMS in the atmosphere. OH + DMS C2H,]DMS (k, k2) The OH + DMS reaction can proceed via several possible channels OH + DMS( +M) @ CH,S(OH)CH,( + M); AHo = -11.0 kcal mo1-' (If lr) OH + DMS + CH,SCH2 + H,O; AHo = -25.5 kcal mol-' (la) OH + DMS -+ CH + CH,SOH AHo = 0 3 kcal mol-' (1b) OH + DMS + CH,S + CH30H; AHo = -18.8 kcal mol-' (W Initially the overall rate coefficients for reaction (1) and the reaction of fully deuteriated DMS OH + [2H,]DMS -+ products (2) were measured under pseudo-first-order conditions ([DMS] 2 100[OH]) at 298 K in the absence of 0,.[Note that reaction (2) has the same possible product channels as reaction (l).] The values k1(298 K) = (4.95 & 0.35) x lopt2 and k,(298 K) = (1.75 & 0.25) x are in good agreement with other recent measure-ments.127 16.17 At low temperatures (T < 240 K) [OH] temporal profiles were characterized by an initial rapid exponential decay followed by a slower exponential decay. The slower decay is due to channels la-lc accounted for the rate coefficient kbi and is slower for C2H,]DMS than for DMS. The faster decay is due to the rapid approach to equilibrium oia the addition pathway and is controlled by the rate coefficients for addition of OH to DMS (k,) and the thermal decomposition of the adduct back to reactants (kf). By fitting the temporal profiles with a non-linear least-squares fitting routine FACSIMILE,' the parameters k, k and kbi were determined and the equilibrium constant K, 42 Atmospheric Oxidation of Dimethyl Sulfide measured from the ratio of k and k,.In the above equation AN = -1. The equilibrium constants were determined over the temperature range 217-245 K. A second-law analysis of the K us. 1/T data shown in Fig. 1 yields the following thermochemical values at 229.5 K A,, H" = -10.2 f2.0 kcal mol-'; A,, So = -28.4 f.6.4 cal mol-' K-l. Using AS" ( = -31.1 at 229.5 K) and AC; calculated from statistical mechanics a third-law calculation yields A,, H"(298 K) = -10.7 & 2.5 kcal mol- '. This leads to A Ho[2H6]DMS * OH) of -10.3 f2.5 kcal mol-' at 298 K. Equilibration of OH with C2H6]DMS has also been recently observed by Wine and co-w~rkers'~ who report A,, H" = -14.2 2.6 kcal mol-I and A,, So = -42.2 10.8 cal mol -' K -' obtained from a second-law analysis.However their measured equi- librium constants are not very different from those of the present study. There are also two contradictory ab initio studies of reaction (If). Turecek2' predicts that there is no bound DMSeOH species; however McKee2' predicts a DMSsOH bond strength of ca. 6 kcal mol-' and states that this may be underestimated by ca. 4 kcal mol-'. Our observation of the equilibration clearly supports the existence of the DMS-OH adduct and supports the calculation of McKee. The DMS-OH adduct is a bound species with a bond dissociation energy of ca.10-14 kcal mol-'. It is conceivable that the adduct could fall apart to give species other than DMS and OH. For example it may be possible to form CH,SOH + CH or CH,OH + CH,S. As shown later the possibility of the formation of CH,OH + CH,S can be neglected. Such reactions of the adduct would manifest as bimolecular reactions in our system. Products from Reaction (1). When OH was generated at 298 K in the presence of DMS production of CH,S radicals was not observed. Based on the relative detection sensitivities of our apparatus for OH and CH,S an upper limit for the branching ratio for kl was determined to be QlC = k,,/k < 0.04. Neither CH,SCH or CH were detected directly by this study but they were converted into CH,S and OH respectively by the addition of 0 and NO.CH,S and OH were detected by LIF. The results of Fig. 1 K us. 1000/T (van't Hoff plot) for reaction (1). K was measured over a range of pressures (30-100 Torr) and bath gases (He N and SF,). The error bars shown are the 95% confidence limits in K and are obtained from the data analysis. The solid line drawn is the weighted least- squares fit to the data (the second-law analysis) and the dashed line represents the fit determined from a third-law analysis (see text for details). S. B. Barone A. A. Turnipseed and A. R. Ravishankara these studies in 20 Torr of 0 at 298 K showed that k,,/k = Q1 = 0.86 f0.26 and k:' = k, + k, = (5.4_+ 1.1)x lo-' cm3 molecule-' s-'. The obtained value of mla is in excellent agreement with that measured recently by Stickel et al.,, who determined Old = 0.84 f0.15by measuring the HOD appearance from the reaction of OD radicals with DMS.Channel (lb),which appears to be small cannot be ruled out from the current experiments because OH regeneration was observed at long reaction times. However the OH regeneration in this system could be due to the reaction CH,SCH,O + 02 +CH3SCHO + HOz (3) followed by the conversion of HO to OH via reaction with NO. Reaction (3) was suggested by Butkovskaya and LeBras2 to explain their observed products in the NO, + DMS reaction in 0,. Our study indicates that under the experimental conditions used here (20 Torr 0, 298 K) channel (la)(the H atom abstraction) is the dominant channel and primarily leads to CH,S formation.0 + DMS * OH + Products (k4) Previous kinetic studies have shown that k and k increase with [02]16*24 due to the reaction of DMS -OH with 0, DMS * OH + 0 +products (4) The rate coefficients for this reaction were determined by monitoring the OH temporal profiles in the presence of DMS (or C2H6]DMS) and 0 and modelling the obtained profiles using the FACSIMILE program. A value of k = (9.7 f3.3) x lo-' cm3 molecule-' s-l was measured between 225 and 234 K and at 30 and 100 Torr of N,. Thus in the atmosphere the only two important loss processes for the adduct are reaction with 0 or thermal decomposition back to reactants. No other reactions of the adduct need to be considered because of the overwhelming abundance of 0,in the atmosphere. Our value of k does not agree with the earlier reported value of Hynes et a1.,16 who report a value of k4 = 4.2 x lo-' cm3 molecule-' s-l; however more recent work'' in their laboratory gives a value of k4 w 8 x 10-' cm3 molecule- 's-which is in good agreement with our measured value.One of the major products of this reaction was shown to be HO . At both 234 and 258 K a branching ratio for the.formation of H0 in the DMS OH + 0,reaction WHO,) = 0.50 f0.15was determined independent of temperature. This is in excellent agreement with recent values from Hynes et al. who also report a branching ratio for channel (4a) of ca. 50%.'' It is assumed that the co-product of HO is DMSO. In the atmosphere it is very likely that DMSO is either deposited in the ocean and aerosols or reacts with OH.The fate of DMSO in the atmo- sphere is not well understood. Reactions of the CH,S Radical We have shown above that in the atmosphere CH,S is a major product of the abstraction part of the OH + DMS reaction. It also has to be the dominant product in the reaction of NO with DMS in the atmosphere because it is known that the net result of this reaction is formation of CH,SCH,., The CH3S radical may also be produced by the reactions of the DMSaOH adduct. The reactions of the CH3S radical with various atmospheric reactants such as 0,and NO have been studied in various labor- at~ries.~ Recently we discovered that CH,S adds to O2 to form the CH,SOO radical.25 When CH,S was produced via DMDS or DMS photolysis in the presence of 0 at low temperatures the CH3S profiles were non-exponential and its concentration reached a 44 Atmospheric Oxidation of Dimethyl Sulfide constant value which decreases as [O,] is increased.The rate of approach to this con- stant value also increases with [O,]. These observations clearly show the occurrence of the equilibrium kt CH3S + 02+ M CH3SOO + M (3) kr CH3SOO + M ____* CH3S + 02+ M (3-1 The temporal profiles of CH3S measured at various concentrations of 0 were analysed to obtain the equilibrium constant at each temperature. From this data the following thermodynamic quantities about CH,SOO were calculated A,," H"(298 K) = -11.7 &-0.9 kcal mol- ' and A' Ho(CH3SO0,298 K) = 18.1 & 1.0 kcal mo1- ' using a calculated A,, S"(237 K) = -36.8 2.6 cal mol- ' K-'and the calculated values of the heat capacities.The adduct formed by this process must be CH3S-00 and not CH3S02 (where both oxygen atoms are bound to the sulfur) since it rapidly decomposes back to CH3S and 0 and does not appear to add another 0 molecule. Formation of the CH3S.00 adduct suggests that other weakly bound adducts such as CH,SO(OO) CH3S02(00) and HS(O0) could exist. Barnes et and Jensen et ul.,,' have observed CH3S(0)OON02which could be formed from addition of NO to CH3SO(OO). Some recent theoretical calculations also suggest that 0 could add to HS.28 There are also some indirect experimental observations that are consistent with HS(O0) formation.25 Thus it appears very likely that other S-containing free radicals could also add 0 and that the bond strengths of these adducts would be ca.10-15 kcal mol-'. In particular we assume that formation of CH,S(O)O and CH,S(O,)O are likely. Field Evidence for Changes in End-products with Temperature Field measurements of DMS and its possible oxidation products in the atmosphere have cast considerable light on key aspects of the DMS oxidation mechanism. A diurnal variation in the DMS concentration in the atmosphere has been reported in several studies all of which report a night-time maxim~m.~.~*~~ This observation suggests that the rate of daytime loss processes exceeds those at night if the strength of DMS emission is independent of time of day. Another observation with significant implications for the DMS oxidation mechanism lies in the [MSA]/[S042-] ratio measured in the atmo- sphere and in ice-cores.is assumed to be formed with unit yield in the atmo- sphere from SO or SO3). Atmospheric measurements of the MSA/S042- ratio have been made by numerous investigators. The general finding is that this ratio changes with latitude and temperature. For example Staubes and Georgii4 report the [MSA]/[S042-] ratio to be ca. 0.01 between 50"N and 30"N but increasing to ca. 0.25 between 53"s and 62"sand 0.44 at 70"s.This variation in the latitude must be accompa- nied by a temperature change in the vicinity of 300 to ca. 250 K. A more direct impres- sion of the variation in the [MSA]/[S042-] ratio with temperature can be gained from the ice-core data. For example Legrand et observed a change in the [MSA]/[S0,2-] ratio from ca.0.03 to ca. 0.18 when the surface temperature must have changed from ca. 210 to 200 K. Even though there has not been a very coherent mea- surement of the variation in the MSA/S042- ratio as a function of temperature it is clear that this ratio changes rapidly with temperature. The magnitude of this change appears to be at least a factor of four or greater in going from 300 to ca. 250 K and another factor of ca. four in going from ca. 210 to 200 K. If both MSA and non-sea salt (nss) SO,' -are primarily produced from the oxidation of DMS the oxidation mecha- nism of DMS must involve steps which reflect this temperature sensitivity. S. B. Barone A. A. Turnipseed and A. R. Ravishankara The Role of Adducts in the DMS Oxidation Mechanism A simplified DMS oxidation mechanism is shown in Fig.2. A more detailed mechanism is listed in Table 1. Fig. 2 highlights various branching points in the mechanism where weakly bound adducts play a role. We will discuss how these reactions can alter the [MSA]/[S0,2-] ratio and justify the assumptions in the mechanism. In this section CH 3s(02)00N02 Fig. 2 Schematic representation of the atmospheric oxidation mechanism of DMS. The numbers on the left indicate branching points in the proposed mechanism where weakly bound adducts may play an important role. The letters A and B indicate potential unimolecular decomposition reactions important in the oxidation mechansim. Table 1The reactions and their rate coefficients used in the box model calculations to predict the variation of the MSA/nss sulfate ratios as a function of temperature reaction rate constant/cm3 molecule-' s-' ref.OH + DMS +(abstraction) 9.6 x lO-"exp(-234/T) 16 OH + DMS +(addition in 760 air) a 16 NO + DMS +CH,SCH + HNO 1.9 x lo-' exp(-500/T) 12 CH,S + 0,+CH,SO + 0 1.98 x lo-'' exp(290/T) 15 CH,S + NO +CH,SO + NO 2.06 x lo-" exp(320/T) 15 CH,SO + 0,+CH,SO + 0 6 x lo-'' 39 CH,SO + NO +CH,SO + NO 8 x 40 CH,SO2 ACH + SO 5 x lo" expC(17.2 kcal mol-' + RT)/RT] see text CH,SO + 0,+CH,SO + 0 3 x 10-13 estimated CH,SO + NO +CH,SO + NO 4 x 10-12 estimated CH,SO 5CH + SO 5 x lOI3 exp[-(22 kcal mol-' + RT)/RT] see text CH,SO + CH,O +CH,SO,H + CHO 1.6 x 10-15 11 CH,SO + HO +CH,SO,H + 0 5 x lo-" 11 CH,S(O),OO + NO +CH,S(O),OO + NO 2.4 x lo-" estimated [T exp(-234/T) + 8.46 x lo-'' exp(7230/T) + 2.68 x lo-'' exp(7810/T)]/[1.04 x 10"T + 88.1 exp(7460/T)J.Atmospheric Oxidation of Dimethyl Suljide CH,S CH,SO CH,SO and CH,S03 are called CH,SO, while the peroxy species CH,S(O),OO refers to CH,SOO CH,S(O)OO and CH,S(O),OO. Reactions of DMS with OH and NO The results of previous studiesg*’ suggest that two reactions NO + DMS +products (6) OH + DMS +products (1) will dominate the initiation because of the large rate coeficents for these reactions and the abundances of OH and NO in the troposphere. Other initiation reactions are unlikely to be important.’ Reaction (6) has been shown to proceed through adduct formation followed by HNO elimination i.e.equivalent to an H atom abstra~tion.~~~~’ In contrast the results presented in this paper combined with those of previous investi- gations show that reaction (I) proceeds through a complex mechanism involving H abstraction and electrophilic addition.16 Work presented in the present study has estab- lished that H-atom abstraction primarily generates the CH,S radical. The reaction product of the OH addition channel (DMS-OH) has been shown to react rapidly with O2 but all the products in this reaction and their subsequent fate in the atmosphere have not fully been elu~idated.’~*’’ The effective overall branching between the addition and abstraction channels of reaction (1) has been characterized and is dependent on tem- perature.Frevious investigations3’ have suggested that NO + DMS will be of little signifi- cance in the oxidation of DMS in the remote marine boundary layer due to the pre- dicted low ambient NO concentrations in this region. However if the concentrations of NO exceed ca. six times that of OH the rate of reaction (6) will equal that of reaction (1) at 298 K. The rate coefficients for both OH and NO reactions increase with decreas- ing temperature. However the rate coefficient for the OH reaction increases slightly faster than that for NO, such that at 250 K for example the NO concentration has to be ca. 10 times that of OH for the two pathways to be equal (assuming the same number of hours of exposure to both reactants). Field measurements and modelling studies suggest that NO concentrations can easily be 10 times the diurnally averaged OH value in the marine boundary As one approaches high latitudes and winter time the [N03]nigh,-timc/[OH]daytime will only get larger.Therefore a complete quantification of the initiation step in the atmospheric oxidation of DMS requires a better knowledge of ambient NO3 concentrations in the oceanic marine boundary layer. As noted earlier the diurnal cycling of DMS concentrations has been observed in the remote marine boundary layer. If this is a common phenomenon the NO concentra-tions in the remote boundary layer are less than the field measurements indicate or more likely DMS concentrations are enhanced at night. The latter possibility could be due to meteorological differences between day and night and the height of the marine boundary layer and may not require an increase in flux of DMS.This is clearly an important issue that needs resolution. The competition between OH addition and hydrogen abstraction in reaction (1) has been suggested in several studies to be responsible for the observed temperature depen- dence of the [MSA]/[SO,’-] ratio in the atmo~phere.~*~~.~~ For the [MSA]/[SO,’-] ratio to increase as temperature decreases MSA must be a product of the addition channel and SO be a product of the H-atom abstraction. We use the rate coefficients for the addition us. abstraction given by Wine and co-workers16 and assume that the addition channel leads only to MSA and the subsequent steps leading to MSA from the DMS .OH adduct are insensitive to temperature.The temperature dependence of the OH + DMS reaction rate coefficient is such that the [MSA]/[SO,’-] ratio will not S. B. Barone A. A. Turnipseed and A. R. Ravishankara change enough with temperature (see Fig. 3). For example from 250 to 270 K this ratio increases by at most 50% while field data suggest a factor of two or more. In addition to the above arguments two points are worth noting (1) The fractional change in the [MSA]/[S0,2-] ratio for a given change in T AT will depend on T and it will become quite insensitive to AT at low T if it depended only on the changes in the branching ratios of reaction (1). Fig. 3 shows that the temperature dependence of the [MSA]/[S042-] ratio tends to decrease at lower temperatures (i.e.the slope decreases at lower temperatures). (2) It is known from chamber studies that CH,S does lead to MSA.8.26-27 Including reaction (6) decreases the temperature dependence of the [MSA]/[S042 -1 ratio even further. Fig. 3 demonstrates that the effect of NO + DMS is only to lessen the importance of the branching between OH addition cf. abstraction. Based on these arguments we propose that the cause of this temperature-dependent branching is unlikely to be due to changes in the OH reaction mechanism with temperature and has an entirely different origin. The lifetime of DMS due to reactions with OH and NO is of the order of a couple of days. OH exists during daytime while NO peaks at night. At night the concentra- tions of species such as HO and NO will be essentially zero.Therefore DMS oxidation initiated at night will probably lead to species such as CH,SCH202N02 and CH,S(O),OONO (see below). Unless these species are removed by deposition (wet or dry) they should be sources of CH,S (or CH,SO,) radicals after sunrise and undergo further reactions. Therefore it appears that even though NO initiates DMS oxidation at night and OH during daytime the oxidation to gas-phase MSA and sulfate takes place during daytime only. Product-yield Branching following the Initiation Reactions We propose that the branching in DMS oxidation arises from the oxidation of CH,S radicals involving temperature-sensitive processes such as unimolecular decomposition and reversible adduct formation with 02.To test this proposal we developed a simple oxidation mechanism of DMS based largely on available laboratory studies of CH3S 220 240 260 280 300 TiK Fig.3 The branching between MSA and SO,’-in 760 Torr air as a function of temperature for a variety of NO conditions for the simple model where only the mechanism for the OH + DMS reaction changes with T. (a)-(d) represent ambient NO concentrations of 0 3 6 and 9 x lo6 molecules cm-, respectively. The diurnally averaged OH concentration is assumed to be 1 x 1O6 molecules cm -’. Atmospheric Oxidation of Dimethyl SulJide and its higher oxides. Where kinetic data is unavailable we have estimated rate coeffi- cients by analogy with other known reactions. The reactions involved in this simplified mechanism are listed in Table 1.The ambient concentrations used in our model are listed in Table 2. We have treated the initiation chemistry in the most simple fashion possible. H-atom abstraction from OH and NO3 reactions is presumed to yield entirely CH,S whereas the addition of OH followed by the O2 reaction forms a reaction product different from SOz and MSA (probably DMSO DMSO,) OH + DMS + +products other than SO and MSA (addition) (14 OH + DMS -+ +CH,S (abstraction) (lb) We realize that it is very likely that the atmospheric reactions of the DMS -OH adduct could lead to CH,SO, MSA or SO2. This occurrence will not greatly affect our argu- ments. When these reactions are better studied one can easily incorporate them into our mechanism without major qualitative changes in the outcome.Role of Higher Oxides Our mechanism involves CH,S and of higher oxides CH,SO, reacting with O3 and NO2 in competition with their unimolecular decomposition or addition of 02.We investigate the role of higher oxides via first modelling the thermal decomposition of CH,SO and then the addition of 0 to these species. The chemistry of each CH,SO is treated analogously to that of CH,S. To calculate potential unimolecular decomposition rates of the higher oxides we have assumed an A factor of 5 x 1013 s-’ and taken the endothermicity for the decomposition as the activation energy. The simple mechanism given in Table 1 leads to a temperature sensitivity in the [MSA]/[S0,2 -J significantly steeper than that observed in the OH initiation reaction.Fig. 4 shows this temperature dependence for the conditions outlined in Tables 1 and 2. The steep temperature-dependent branching between MSA and SO2 is a result of the competition between the unimolecular decomposition of the CH,S02 radical and the reaction of CH3S02with 0 to oxidize sulfur further without breaking the C-S bond. Table 2 Assumed abundance of various species in the clean marine environment daytime species mixing ratio ref. 4 (-8) 41 3 (-11) 41 1 (-10) 42 2 (-1 1) 38 3 (-13) calculated 1 (-11) calculated 2 (-12) calculated 4 (-14) calculated The sources of the data are given in the references shown. These are ‘typical’ mixing ratios during daytime and some of them have been rounded off to one significant figure.The ratios of HO to OH and NO to NO are kept consis- tent with the above values of 0 and the known J values for NO,. S. B. Barone A. A. Turnipseed and A. R. Ravishankara t 0.1 + I I I I I 270 275 280 285 290 295 300 TIK Fig. 4 Calculated [MSA]/[S0,2-J ratios from DMS oxidation as a function of temperature. (a) All branching due to changes in initiation reaction mechanism; (b)model conditions outlined in Tables 1 and 2. The temperature dependence of the branching reflects the activation energy for the decomposition of CH3S02. We estimate the decomposition of CH3S03 to be more endothermic than that of CH3S02. This is because CH3S0 must be oxidized to CH3S03 by a species such as 03,NO or OJNO combination and these reactions have to be at least thermo-neutral to be important in the atmosphere.Other investigators have also estimated the C-S bond strength in CH3S02 to be less than in CH3S03." Because of the larger endo- thermicity one could expect it to contribute more to the temperature dependence of the branching between MSA and However we do not expect the unimolecular decomposition of CH3S03 to be a very important process in the oxidation for two reasons (1) A large amount of the CH3S02 decomposes to SO2 before it can form CH3S03 and (2) most of the CH,S03 reacts with H02 and CH20. Thus if CH3S03 is formed in the atmosphere it will very likely end up as MSA. Therefore if our assumed rate coefficients are correct it appears that the direct formation of SO3 in DMS oxida-tion as suggested by Bandy et al.is likely to be minor.35 Note,'however that the thermochemistry of CH3S0 is very poorly defined. If the C-S bond strength in CH3S03is much smaller say <20 kcal mo1-' this species could become a significant branching point in DMS oxidation. If the CH,SO radical is stable it could react with NO to form CH3S03N02; this species may be hydrolysed in the particles to give MSA in the liquid phase (see below). Also the rate coefficient for the CH3S03+ HOz reac-tion needs to be determined to estimate this branching. A systematic variation of the ambient NO concentration in our model showed a very small effect on the oxidation products of CH3S in the atmosphere. Although the rate constants for reactions of CH3S CH,SO and CH3S02 with NO are large the low NO levels in the remote troposphere prohibit these reactions from competing with 0 reactions.An order of magnitude increase in the NO concentration used in our model results in a less than 8% change in the [MSA]/[S042-] ratio at 271 K. Therefore we conclude that changes in the amount of NO in the boundary layer has the greatest effect on the initiation chemistry of DMS oxidation and little effect on the subsequent reactions for this proposed scheme of DMS oxidation until the NO level reaches > ca. 300 pptv. Molecular oxygen has been observed to add reversibly to CH3S radicals under atmospheric condition^.^^ [Reactions (5f 5r).] The rate constants for the reactions of the Atmospheric Oxidation of Dimethyl Suljide CH3SO0 radical with NO NO and 0 have been measured." In the remote atmo- sphere the CH3S + O3 reaction should dominate over all other loss processes of the CH,SOO adduct and CH,S.To predict the effect of addition of O2 to CH,S(O), we added the following reactions to the model; kr CH,S(O) + 0 + M -CH,S(O),OO + M (7f) kr CH3S(0),00 + M -CH,S(O) + 02 + M (7r) CH3S(0),00 + NO -CH,SO + NO (8) Note that CH,SO would not be expected to add 0 since the sulfur in this species is fully oxidized. Reactions (7) and (8) comprise a potentially efficient pathway for convert- ing CH3S02 to CH,S03. The equilibrium constant for the reversible addition of 0 to CH3S0 would be very sensitive to the temperature and hence this process could be another candidate in explaining the [MSA]/[SO,' -1 ratio observations in the field.Owing to the abundance of 0 in the atmosphere it is likely that the equilibrium in reaction (7) will be established rapidly and reaction (8) would be the rate-limiting step in determining the overall efficiency of this mechanism. The reaction mechanism presented up to this point is somewhat analogous to that given originally by Yin et d." and modified by Bandy et 121.~' However there are some major differences. First we can show that NO and not HO reaction followed by pho- tolysis of the peroxide is responsible for converting CH,S(O),O to CH,S(O),O. Based on simple steady-state calculations and measured abundances of NO ,36-38 the abun- dance of NO is shown to be approximately an order of magnitude greater than that of H02 in the marine boundary layer.The rate coefficients for the reaction of the larger peroxy radicals with NO is greater than those with H0 and CH302,'2*'3 and hence the rate coefficient for the reaction of CH,S(O),OO with NO must be larger than that with HO,. Therefore the majority of the CH,S(O),OO must react with NO. Even if hydroperoxides are formed since they do not usually absorb very strongly in the near- UV they are likely to be washed out of the atmosphere. We have estimated the rate constant for reaction (8) to be 2.4 x lo-'' cm3 molecule-' s-' and calculated yields for a variety of different CH,SO,-00 bond strengths. Bond strengths < 10 kcal mol-' have little effect on the outcome.However as the bond strength is increased from 10.5 kcal mol- ' the addition reactions of CH,S(O) radicals have a pronounced effect on the absolute values of the branching to MSA and SO2 and the temperature dependence of their ratio. Fig. 5 shows the effect of adduct bond strengths of 13 and 14 kcal mol-' on the temperature dependence of the [MSA)/[SO,'-] ratio. The increase in the [MSA]/[S0,2-] ratio with temperature is a result of a shift toward larger equilibrium concentrations of the CH,S0200 at low temperatures. This in turn increases the rate of reaction (8) relative to the unimolecular decomposition of CH3S02 and hence adds to the overall temperature sensitivity of the [MSA)/[S0,2 -1 ratio. Note that the incorporation of this process in the model adds to the temperature dependence in the [MSA]/[SO,'-] ratio determined by the unimolecular decomposition of the CH,S02 radical alone.Aside from influencing the temperature dependence of the branching in DMS oxida-tion the inclusion of the 0 adduct formation reaction has several other consequences. First because the NO + CH3S(0,)O0 reaction becomes the rate limiting step in the cycling of CH3S02 to CH,SO, NO will play a larger role in determining the [MSA]/[S0,2-] ratio. The branching to stable end-products in DMS oxidation will be strongly NO dependent. Secondly the increased efficiency in the conversion of CH,S02 to CH,SO makes the unimolecular decomposition of CH,SO more impor- tant. Thus branching to form SO can become more likely. However SO formation is still likely to be minor because the rate of CH,SO decomposition is likely to be slow S.B. Barone A. A. Turnipseed and A. R. Ravishankara 100.0 270 275 280 285 290 295 300 TI Fig. 5 Calculated [MSA]/[SO,’-] ratios as a function of temperature from a DMS oxidation mechanism which includes reactions (7) and (8). (a),(b)and (c) are calculated from CH,S(O,). 00 adduct bond strengths of 0 13 and 14 kcal mol-’ respectively. The other rate coefficients and ambient concentrations are listed in Tables 1 and 2. with respect to its reactions with H02 and CH,O. Lastly formation of an 0 adduct with CH3S0 is also a potentially important pathway to form CH,SO,. In our simple model all CH3S0 is converted to CH,SO,. However isomerization or other removal processes of this radical could play key roles in the oxidation of DMS.The formation of nitrate from NO addition to the peroxy-adducts; NO + CH,S(O),OO + CH,S(O),OONO (9) could be an important process in the atmosphere. Nitrates such as these have pre- viously been observed in chamber studies of DMS ~xidation.~~.’~ This observation is an indication of the likelihood of 0,-adduct reactions in the atmospheric oxidation of DMS. The formation of these peroxynitrate radicals could be especially important if they are easily hydrolysed in cloud drops. The hydrolysis of these nitrates could prove to be an important source of MSA in the atmosphere. Also if the nitrates are long-lived they could play a role in the transport of sulfur to the free troposphere.It is possible that reactions of CH,S and CH,SO with 0 and of CH,SO with NO could result directly in the cleavage of the C-S bond. These channels are exothermic and would have a significant effect on the results of the mechanism developed thus far. To evaluate this possibility we have included the following reaction in our model; CH,SO + 0 + +SO (10) Note that the above reaction merely represents the cleavage of the C-S bond to lead to SO2formation and is not meant to be an elementary reaction. Inclusion of this chem- istry significantly effects the calculated [MSA]/[S0,2 -1 ratio. Aside from increasing the absolute magnitude of SO formation including reaction (10) decreases the temperature dependence of the SO yield. Fig. 6 demonstrates the effect on the [MSA]/[S0,2-] ratio of 25% 50% and 75% channels in the reaction CH3S0 + O3 to cleave the C-S bond.Reaction (10) channels significant amounts of the sulfur to end-products by bypassing the CH,SO species and hence avoids the temperature-dependent mecha- nism for SO production via unimolecular decomposition. The effect of CH3S + O3-+ -,SO2 (1 1) Atmospheric Oxidation of Dimethyl SulJide 0.01 5 270 275 280 285 290 295 300 TIK Fig. 6 [MSA]/[S0,2-] as a function of temperature from DMS oxidation calculated with the inclusion of reaction (10). (a)-(d) reflect branching ratios of the reaction CH,SO + O3 to yield SO of 0,0.25,0.50 and 0.75 respectively.All other relevant conditions are in Tables 1 and 2. would be identical to reaction (10).However the effect of; CH3S0 + 0 + CH + SO3 + 0 (12) could be less than reactions (10) and (ll) depending on the rate coefficient for the unimolecular decomposition of CH,SO .Note that the addition of O2 to CH,SO fol- lowed by the reactions of CH,S(O)OO could mediate the effects of the C-S bond cleavage reactions involving 03. We have focussed on the potential branching reactions of the CH3S02 and CH3S03 radicals however it is important to note that the chemistry of the preceding CH,SO radical could also be a source of the branching between MSA and SO,,-in DMS oxidation. Branching at CH3S0 would most likely occur through a competition between O2 addition followed by reaction with NO and reaction (10) resulting in C-S bond cleavage.C-S bond cleavage would eventually lead to SO formation whereas 0 addition could enhance the oxidation state of the sulfur without breaking the C-S bond; CH3S0 + 0 -+CH3S(0)O0 (13) CH,S(O)OO + NO + CH3SOZ + NO2 (14) CH,SO + O3-+ SO + CH + 02. (15) Role of Heterogeneous Reactions There is a possibility that MSA is formed only via heterogeneous reactions of species such as CH3S(0),00N02. The MSA measured in the atmosphere has always been extracted from the aerosol phase. If this is indeed the case the [MSA]/[S0,2-J ratio may be related to the heterogeneous reaction probability i.e. availability of aerosols fog clouds etc. and the stability of the gas-phase species which are taken up by the con- densed medium. Both these factors are enhanced at lower temperature and hence are consistent with the observed increases in the [MSA]/[S0,2-] ratio with decreases in S.B. Barone A. A. Turnipseed and A. R. Ravishankara 53 temperature. Measurement of gas-phase MSA abundances is essential for shedding light on this problem. Summary of Modelling Studies Our simple modelling analysis of DMS atmospheric oxidation has led to several basic conclusions (1) The temperature dependence of OH addition us. H-atom abstraction in the initiation chemistry of DMS oxidation is not strong enough to rationalize completely the variations in the MSA/SO,’- ratios observed in the ice-cores and with latitude. The inclusion of NO chemistry greatly reduces the overall temperature dependence of the OH addition us.H-atom abstraction. An accurate knowledge of the ambient NO concentrations in the marine boundary layer is the key to better understanding of the initiation chemistry. (2) It appears that the abstraction pathway in the OH/NO + DMS reaction can also lead to the formation of MSA contrary to what has been assumed in some past studies but in accordance with many chamber studies. (3) It appears that the major role of NO in DMS oxidation is limited to the initiation step unless CH,SO -0 adducts are important. (4) CH,S oxidation via the formation of higher oxides that can unimolecularly decom- pose (cleave the C-S bond) or raise the oxidation state of sulfur without breaking the C-S bond can lead to a large temperature dependence of the branching between MSA and SO,.This temperature dependence will be heightened if 0 adds reversibly to some of the CH,SO species. If this were to happen NO could play a role in determining the branching between different endproducts. (5) The potential for C-S bond cleavage via other bimolecular reactions of CH,S or CH,SO lessens the calculated temperature dependence in the [MSA]/[S0,2 -3 ratio. This is especially the case if it occurs before the formation of CH,SO in the oxidation mechanism. (6) Even in the clean marine air it is unlikely that species such as CH,SO,H [CH,S(O)-OH or CH,SOOH] are formed via the involvement of HO radicals. (7) Heterogeneous reactions of species such as CH,S(O),OONO could be key steps in MSA formation.The simple modelling exercise highlights areas which need further effort (1) Measurement of the ambient levels of NO and NO in the marine boundary layer and free troposphere is essential. The observed diurnal variation in DMS may be contradictory to the measured levels of NO in the marine troposphere. (2) Determinations of the unimolecular decomposition of CH,SO radicals via experi- mental or computational investigations are needed. (3) The detection of peroxy and peroxy nitrate intermediates in the atmosphere and measurements of MSA as a functions of NO and temperature could provide valu- able insight into the key aspects of the oxidation mechanism. (4) Measurement of gas-phase MSA is needed to evaluate the role of heterogeneous processes.(5) Development of detection techniques such as photoionization mass spectrometry and chemical ionization mass spectrometry for various intermediates and end- products are needed to decipher the DMS oxidation mechanism. This work was carried out as a part of NOAA’s Climate and Global Change Pro- gramme. We thank Professor Michael McKee for assistance in calculating rotational and vibrational constants for CH,S(OH)CH . Atmospheric Oxidation of Dimethyl SulJide References 1 Dimethylsulphide Oceans Atmosphere and Climate ed. G. Restelli and G. Angeletti Kluwer Dordrecht 1992. 2 H. Berresheim J. Geophys. Res. 1987,92 13. 3 H. Berresheim M. 0. Andreae G. P. Ayers R. W. Gillett J. T. Merrill V. J. Davis and W. L. Cha- meides J.Atmos. Chem. 1990 10 341. 4 R. Staubes and H-W. Georgii Measurements of Atmospheric and Seawater DMS Concentrations in the Atlantic the Arctic and the Antarctic region Dimethylsulphide Oceans atmosphere and climate Belgi-rate Italy 1992 p.95. 5 R. W. Gillet G. P. Ayers J. P. Ivey and J. L. Gras Measurement of Dimethyl Suljide Sulfur Dioxide. Methane Sulfonic acid and Non-sea Salt Sulfate at the Cape Grim Baseline Station Dimethylsulphide Oceans Atmosphere and Climate Belgirate Italy 1992. 6 M. 0. Andreae R. J. Ferek K. Bermond K. P. Byrd R. T. Engstrom S. Hardin P. D. Houmere F. LeMarrec H. Raemdonck and R. B. Chatfield J. Geophys. Res. 1985,90 12. 7 I. Barnes K. H. Becker and N. Mihalopoulos FTlR Products Study of the Photolysis of CH,SSCH, Reactions of the CH,S Radical Dirnethylsulphide Oceans Atmosphere and Climite Belgirate Italy 1992.8 S. Hatakeyama and €1. Akimoto J. Phys. Chem. 1983,87,2387. 9 A. A. Turnipseed and A. R. Ravishankara The Atmospheric Oxidation of Dimethyl Suljide Elementary Steps in a Complex Mechanism Dimethylsulphide Oceans Atmosphere and Climate Belgirate Italy 1992 p. 185. 10 G. S. Tyndall and A. R. Ravishankara Int. J. Chem. Kinet. 1991,23,483. 11 F. Yin D. Grosjean and J. H. Seinfeld J. Atmos. Chem. 1990,11,309. 12 W. B. DeMore S. P. Sander D. M. Golden R. F. Hampson M. J. Kurylo C. J. Howard A. R. Ravishankara C. E. Kolb and M. J. Molina Chemical Kinetics and Photochemical Data for use in Stratospheric Modeling Jet Propulsion Laboratory JPL Pub. No. 92-20 1992.13 R. Atkinson D. L. Baulch R. A. Cox R. F. Hampson J. A. Kerr and J. Troe J. Phys. Chem. ReJ Data. 1992 21 1125. 14 G. L. Vaghjiani and A. R. Ravishankara J. Phys. Chem. 1989,93 1948. 15 A. A. Turnipseed S. B. Barone and A. R. Ravishankara J. Phys. Chem. 1993,97,5926. 16 A. J. Hynes P. H. Wine and D. H. Semmes J. Phys. Chem. 1986,90,4148. 17 J. P. D. Abbatt F. F. Fenter and J. G. Anderson J. Phys. Chem. 1992 % 1780. 18 A. M. Malleson H. M. Kellet R. G. Myhill and W. P. Sweetenham A.E.R.E. Harwell Publication R 13729 A.E.R.E. Publications Oxford 1990. 19 A. J. Hynes T. Pounds T. McKay J. D. Bradshaw and P. H. Wine 12th International Symposium on Gas Kinetics Reading 1992. 20 F. Turecek J. Phys. Chem. 1994,98,3701. 21 M. L. McKee J. Phys. Chem. 1993,97 10971.22 R. E. Stickel Z. Zhao and P. H. Wine Chem. Phys. Lett. 1993,212,312. 23 N. I. Butkovskaya and G. LeBras J. Phys. Chem. 1994,98,2582. 24 I. Barnes V. Bastian and K. H. Becker Int. J. Chem. Kinet. 1988,20,415. 25 A. A. Turnipseed S. B. Barone and A. R. Ravishankara J. Phys. Chem. 1992,% 7502. 26 I. Barnes V. Bastian K. H. Becker and H. Niki Chem. Phys. Lett. 1987 140,451. 27 N. R. Jensen J. Hjorth C. Lohse H. Skov and G. Restelli J. Amos. Chem. 1992 14,95. 28 D. Laakso C. E. Smith A. Goumri J-D. Rocha and P. Marshall Chem. Phys. Lett. 1994,227 377. 29 E. S. Saltzman and D. J. Cooper J. Atmos. Chem. 1988,7 191. 30 M. Legrand C. Feniet-Saigne E. S. Saltzman C. Germain N. I. Barkov and V. N. Petrov Nature (London) 1991,350 144. 31 R. Atkinson J.N. Pitts and S. M. Aschmann J. Phys. Chem. 1984,88 1584. 32 J. F. Noxon J. Geophys. Res. 1983,88 1 1017. 33 J. A. Logan J. Geophys. Res. 1983,88 10785. 34 A. J. Hynes and P. H. Wine in Biogenic Sulfur in the Environment ed. E. S. Saltzmann and W. J. Cooper American Chemical Society Washington DC 1989. 35 A. R. Bandy S. B. W. Blomquist S. M. Chen and D. C. Thornton Geophys. Res. Lett. 1992 11 1125. 36 B. A. Ridley M. A. Carroll D. D. Dunlap M. Trainer G. W. Sachse G. L. Gregory and E. P. Condon J. Geophys. Rex 1989 94 5043. 37 B. A. Ridley S. Madronich R. B. Chatfield J. G. Walega R. E. Shetter M. A. Carroll and D. D. Montzka J. Geophys. Res. 1992 97 10. 38 M. A. Carroll B. A. Ridley D. L). Montzka G. Hubler J. G. Walega R. B. Norton B. J. Huebert and F.E. Grahek J. Geophys. Res. 1992 97 10. 39 F. Domine A. R. Ravishankara and C. J. Howard J. Phys. Chem. 1992,% 2171. 40 G. S. Tyndall and A. R. Ravishankara in Biogenic. Sulfur in the Environment ed. E. S. Saltzman and W. J. Cooper American Chemical Society Washington DC 1989 p. 450. 41 B. A. Ridley and E. Robinson J. Geophys. Rex. 1992,97 10. 42 B. Heikes J. Geophys. Res. 1992 97 18. Paper 5:00197H Rcceived 9th January 1995