Rate coefficients for the room temperature reaction of C1 atoms with dimethyl sulfide and related alkyl sulfides David J. Kinnison, Wolfgang Mengon and J. Alistair Kerr* EA WAG, Swiss Federal Institute for Environmental Science and Technology, E TH Zurich, CH-8600 Dubendorf, Switzerland Rate coefficients for the reaction of C1 atoms with a selection of alkyl sulfides have been determined using the relative rate technique at 298 & 3 K. Based on an absolute value of 1.94 x 10-'' cm3 molecule-' sK1 for the reaction of Cl with n-butane, the following values for the reaction-rate coefficients were determined in the presence of nitrogen (in units of lo-'' cm3 molecule-' s-') to be: dimethyl sulfide, 3.61 & 0.21; ethylmethyl sulfide, 4.89 & 0.29; diethyl sulfide, 5.98 & 0.23; and dipropyl sulfide, 7.02 0.81.The relative-rate experiments were conducted in an atmosphere of either nitrogen or synthetic air to investigate the effect of the presence of oxygen on the rate coefficient; only the rate coefficient for the reaction of C1 with dimethyl sulfide showed an appreciable difference, yielding a value of (4.19 0.16) x fO-" cm3 molecule-' s-'. Further experiments were undertaken to measure the relative-rate coefficients of the series of alkyl sulfides using cyclohexane as the reference compound. In most cases, the value of ksulfide/k,,-butane divided by the value of ksulfide/kcyclohexane produced a value of kcycIohexane/kn-butane consistent with our experimentally determined value of kcyclohexane/k,,-butane= 1.69 0.04.The results are discussed with respect to the literature data, structure activity relationships, chemical mechanisms and the chemistry of the marine atmosphere. Dimethyl sulfide, DMS, is an important atmospheric sulfur Reactions of the OH radical with organic compounds have species; it is the main natural source of sulfur in the atmo- been studied extensively and structure-activity methods are sphere.' DMS is produced in the oceans and is volatised to available for estimating OH rate coefficients with some accu- the atmosphere., The majority of kinetic and mechanistic racy, based on a well defined database for OH rate coeffi- studies have focused on the reaction of DMS with the main cients. Aschmann and Atkinson' have applied similar daytime oxidant, the hydroxy radical,, and the night-time methods to the reactions of C1 with a selection of alkanes and species, the nitrate radical.,^^ Stickel et aL5 have indicated that have produced a structure-activity relationship (SAR).Data the removal rate of DMS from the atmosphere is too rapid to for the reactions of C1 with alkyl sulfides are limited and to be accounted for entirely by OH and NO,. As an explanation develop a meaningful structure-activity relationship requires a of the short lifetime of DMS, I0 radicals were proposed as a well defined dataset, and hence, an additional aim of this work possible sink for DMS.6 It has been shown, however, that the is to supplement the database for the reactions of C1 with removal of DMS by I0 radicals is negligible.'.' Recently, it alkyl sulfides.has been suggested8-*' that the atmospheric concentration of C1 atoms in the marine atmosphere could be significant com- Experimental pared with the atmospheric concentration of OH. Generally, Kinetic experiments were carried out in a previously described the reaction-rate coefficients of C1 with organics are greater ~ystem'~which was modified for the purposes of this work by than the reaction-rate coefficients of OH with organics by at replacing the calibrated volume, glass manifold, pressureleast an order of magnitude, owing to the higher reactivity of transducers and photolysis lamps. Chlorine atoms were pro- C1. If the atmospheric concentration of C1 reaches lo4 mol-duced from the photolysis of carbonyl dichloride (phosgene): ecule ern-,, two orders of magnitude lower than the atmo- spheric concentration of OH, then the rates of removal of COC1, + hv -+ CO + 2C1 (2) DMS by C1 and OH will become competitive.Earlier work by Molecular chlorine is unsuitable as a source of C1 atoms, Nielsen et a!." and Stickel et aL5 has established that the owing to a fast dark reaction with alkyl sulfides.'' The possi- reaction of C1 with DMS is very fast, approaching the gas- bility of signficant amounts of C1, being formed oia the reac- kinetic limit, and that two reaction channels exist. tion C1 + C1 + M -+ C1, + M from the photolysis of COCl, C1+ CH,SCH, --* CH3SCH, + HCl (la) in our system, can be discounted on the basis of the known rate coefficient for the C1 recombination reaction, and the + CH,SClCH, rapid rate of photolysis of C1, under our experimental condi- It has been shown that reaction (1) is pressure dependent, tions.The kinetic experiments were conducted in the absence and that channel (la) dominates (lb) at low pressure (1 Torr) of NO,, because it has been shown that, in the presence of but with increasing pressure, channel (lb) becomes significant NO,, unsatisfactory kinetic data are obtained for the reaction owing to the stabilisation of the add~ct.~ The atmospheric fate of OH with alkyl sulfides.' The relative-rate measurements of CH,SClCH, is unknown, although possible reaction path- were undertaken at room temperature (298 & 3 K) and at ways based on thermodynamical arguments have been sug- atmospheric pressure (760 & 10 Torr) in a Teflon bag with a ge~ted.~?'~Part of the motivation behind this work is to volume of ca.200 dm3, surrounded by 16 Germicidal lamps investigate the reactions and mechanisms of C1 with alkyl sul- (Osram HNS 15 W OFR, A,,,,, = 254 nm). The bag was fides using end-product analysis, especially for DMS with covered with a black cloth to prevent pre-photolysis of the respect to atmospheric chemistry. Knowledge of the reaction reactants, and two electrical fans circulated air around the bag rates of C1 with symmetric and asymmetric dialkyl sulfides in to sustain a reasonably uniform reaction temperature atmospheres of N, and synthetic air will aid future studies of throughout the irradiation of the reactants.The reaction mix- these species. tures were prepared by sweeping known amounts of the J. Chem. SOC.,Faraday Trans., 1996,92(3), 369-372 369 sulfide, reference compound and phosgene from a calibrated volume into the Teflon bag with a flow of either N, or syn- thetic air passed through an air purifier (Quantitech, QD100). Pressures of each reactant were measured using a capacitance manometer (MKS Baratron 220, 100 Torr range). In general, the Teflon bag was filled with 200 dm3 of either N, or purified synthetic air. The Teflon bag was subsequently shaken and allowed to stand for ca. 1 h in order to mix thoroughly the reactants. The mixture was then irradiated for ca. 1 h and the decay of the sulfide and the reference compound was moni- tored by gas chromatography (Carlo Erba HRGC 4160) with flame-ionisation detection (FID).Gas samples were injected into the gas chromatograph, GC, via a 3 cm3 stainless-steel gas sampling valve and the GC was equipped with a 20 m glass-capillary column with 60% OV-1701 as the stationary phase and operated at an oven temperature of 363 K. Nitrogen was supplied by Carba Gas with a stated purity of 99.999%.The synthetic air (Pangas), a mixture of 20% 0, and 80% N,, was passed through an air purifier (Quantitech, QD100) to remove methane. Dimethyl sulfide, DMS, and diethyl sulfide, DES, were supplied by Merck with stated purities of >99% and >98%, respectively. The following compounds were supplied by Fluka with purities as stated by the manufacturer: ethylmethyl sulfide (EMS) (> 97%); dip-ropy1 sulfide (DPS) (~90%);n-butane (>99%); cyclohexane (99.7%);and phosgene (> 99Y0).All reactants were degassed by repeated freeze-pump-thaw cycles.Results In general, the relative-rate measurements for the reaction of C1 with the alkyl sulfides (RSR') were carried out at least three times against the reference hydrocarbon RH (n-butane in the first instance): C1 + RSR' -,products (3) C1 + n-butane +products (4) Further experiments were conducted with cyclohexane as the reference hydrocarbon to test the internal consistency of the system and to confirm any 0, effect on k, . C1 + cyclohexane -+ products (5) Provided the alkyl sulfide and the reference hydrocarbon are consumed solely by C1 atoms, the ratio k,/k, can be deter- mined from the equation: where the subscripts 0 and t refer to the concentration at the start of the experiment and at a time t, respectively.A plot of ln([RSR'],/[RSR'],) us. ln([RH],/[RH],) therefore yields a graph with a slope of k,/k,, and an intercept through the origin. The mixing ratios of the reactants were all in the range 1-4 parts per million (ppm) (1 ppm = 2.46 x 1013 molecules from photolysis was insignificant ( -=z2%) compared with the loss due to the reaction of the substrate with C1 on the same timescale. The relative-rate ratios were determined using eqn. (I) by linear-regression fitting of the data. Example plots are shown in Fig.1 for the sulfide-butane measurements, and the resulting rate-coefficient ratios determined in this work are shown in Table 1, where the errors quoted are for 95% con-fidence limits. To test for internal consistency of the system, for each of the sulfides the ratio k,/k, (using cyclohexane as the reference hydrocarbon) was divided by the value of k,/k, (using n-butane as a reference) to yield calculated values of the relative rate coefficient ratio kcy,ohexane/kn-butane, k,/k, , as shown in Table 1. These calculated values of ratio k,/k, compare well with our measured ratio, k,/k, = 1.69 & 0.04. This value is in good agreement with previous studies of Atkinson and Aschmann16 (1.58 k0.07), Wallington et al.17 (1.60 0.07) and Nielsen et al." (1.66 f0.04), and adds further confidence to the experimental technique employed in this study.The rate-coefficient ratios were calculated on an absolute basis by taking a recently evaluated value13 for the rate coefficient for the reaction of C1 with n-butane of (1.94 x lo-'' cm3 molecule-' s-l) and the results are shown in Table 2. A second set of absolute rate coefficients in N, was calculated from our measured ratios of k,/k, and k,/k,, together with the evaluated rate coefficient k,. As seen from Table 2, these results are consistent with our first set of absol- ute rate coefficients, but the first set are recommended since they are calculated in a more direct manner. The errors I 0.8 0.4 E cn c 'a0 IJDES 0 0.1 0.2 0.3 0.4 0.5 In ([n-butane]d[n- butane],) cmP3 at 295 K and 740 Torr).The reactant mixtures were Fig. 1 Sample plots of ln([RSR'],/[RSR'],) us. ln([n-butane],/stable in the Teflon bag in the dark for several hours. It was [n-butane],) for the reaction of C1 with the following alkyl sulfides noted that in the absence of the C1 atom precursor, phosgene, (RSR'): (0)dimethyl sulfide, (m) ethylmethyl sulfide, (+) diethyl the alkyl sulfides were photolysed, but the loss of the substrate sulfide and (A)dipropyl sulfide. Table 1 Rate coefficient ratios measured in this work DMS CH,SCH, 1.86 f0.11 2.16 f0.08 1.23 f0.08 1.48 f0.12 1.51 f0.14 EMS C2H,SCH3 2.52 f0.15 2.37 f0.24 1.40 f0.08 1.55 f0.08 1.80 f0.17 DES C2H 5SC2H 5 3.08 f0.12 3.44 f0.16 2.02 f0.08 1.91 0.16 1.52 f0.14 DPS C,H,SC,H, 3.62 f0.42 3.83 f0.25 1.98 f0.19 2.18 0.09 1.83 f0.46 370 J.Chem. Soc., Faraday Trans., 1996, Vol.92 Table 2 Absolute rate coefficients for the reaction of C1 with alkyl sulfides at room temperature sulfide formula k" ref. methodb DMS CH3SCH3 3.61 f0.21' this work RR 4.03 f0.17' this work RR 3.3 k0.5' 5 FP-RF 3.22 fO.3OJ 11 RR EMS C,H3SCH3 4.89 f0.29' this work RR 4.60 f0.17" this work RR DES C,H,SC,H, 5.98 f0.23' this work RR 6.62 0.17d this work RR 3.52 k0.28g 5 FP-RF 4.41 f0.401 11 RR DPS C3H,SC3H, 7.02 _+ 0.81' this work RR 6.50 -t 0.41' this work RR 5.18 f0.401 11 RR " In lo-'' cm3 molecule-' s-'. RR = relative rate and FP-RF = flash photolysis-resonance fluorescence.' 760 Torr N, , calcu-lated from k3/k, taking kn-butane+C,= 1.94 x lo-'' cm3 molecule-' s-'.'760 Torr N,, calculated from k3/k, x k,/k,, taking /cn& e+C, = 1.94 x lo-'' cm3 molecule-' s-'. '700 Torr N,. 760 Torr N,, based on kcyclohexane+C, 3.11 x lo-'' cm3 molecule-' s-l. = 500 Torr N, . quoted for the absolute rate coefficients are for 95% con-fidence limits and reflect only the precision of the data. Inclu- sion of the errors from the evaluated rate coefficient for the reaction of C1 with n-butane adds at least a further 25% uncertainty to the absolute rate coefficients determined in this work. We consider the effect of secondary chemistry on the relative-rate coefficients to be negligible because of the inter- nal consistency shown by the application of a second reference hydrocarbon, cyclohexane.Furthermore, the linear In-ln plots all have zero intercepts, which lends confidence to the general correctness of the method and, therefore, suggests that the reactants were solely consumed by C1. Discussion Comparison of the absolute rate coefficients derived in this work with the literature data in Table 2 shows that only the rate coefficient for the reaction of DMS with C1, in the pres- ence of N,, is in agreement. The discrepancies in the other sulfide data may be due, in part, to interferences from second- ary chemistry. It is difficult to make comparisons when only two other studies have been reported for the series of alkyl sulfides studied here.Nielsen et al.' ' measured k, employing cyclohexane as the reference compound and their rate coeffi- cients are ca. 20-40% lower than the values reported here, the reason for which we have no explanation. However, Nielsen et a/.' ' did not report their measured relative-rate-constant ratios and, consequently, a direct comparison with our values of k,/k, is not possible. To the best of our knowledge, there are no values available in the literature for the rate coefficient for the reaction of C1 with EMS and for the reaction of DMS with C1 in the pres- ence of 0,. The rate coefficient for the reaction of EMS with C1 is consistent with the data presented here, lying between the value of kDES+C,and k,. The presence of 0, has an effect on the value of k, which was observed in two separate sets of experiments with n-butane and cyclohexane as the reference hydrocarbons.In each case, the value of k, measured in the presence of air is higher than when k, is measured solely in the presence of N, ,and furthermore, there is no overlap of the relative-rate-constant ratios when the error limits are taken into account. The most likely explanation of this observation is that the adduct, CH,SClCH, ,reacts with O,, thus remov- ing the adduct, and hence, the rate of the back reaction of the adduct to form reactants is reduced, resulting in a net increase in the relative-rate ratio. No 0, effect was observed or confirmed for the other sulfides, for which essentially the same values of the relative-rate ratios were obtained for the experiments in the presence of either N, or air.In the case of DES and DPS, the value of k,/k, is higher in the presence of air than in the presence of N, ,but the value of k,/k, is essen- tially the same in the presence of either N, or air. Fig. 2 shows a plot of k,/k, us. the number of carbon atoms in the alkyl sulfide. The correlation of the ratio k,/k, to the number of carbon atoms in the alkyl sulfide is clearly non- linear; with increasing number of carbon atoms, the ratio k,/k, becomes essentially constant, presumably because the rate coefficient for the reaction of C1 with the higher alkyl sulfides is approaching the gas-kinetic limit. Taking the slope of the linear portion of the curve (GC,) shown in Fig.2, we are able to estimate the rate-coefficient ratio, k,/k,, and hence the rate coefficient k, by multiplying the slope by k,. In this way, k,, per C atom equals 1.51 x lo-'' cm3 molecule-' s-'. Furthermore, adopting a similar approach to that of the SAR for the reactions of C1 with hydrocarbons,13 we can estimate the substituent factor [F(-S-)] for the S atom by taking values for the substituent factors [F(-CH,) and F(-CH,-)I and the rate coefficients for reaction of C1 with primary, kprim, and secondary, kSec,hydrogen atoms as presented by Asch- mann and Atkinson.I3 For example, in the case of EMS; where k,,, is the estimated rate coefficient for the reaction of C1 with EMS. We find that F(-S-) is approximately equal to 4 and the rate coefficients kDMS+Cl,kEM,+cl and kDES+C1are estimated from the SAR method to be 2.7, 4.9 and 7.1, respec- tively, (in units of 10-lo cm3 molecule- ' s-').The SAR esti- mation predicts k, for DMS, EMS, and DES to within *35% of the measured rate coefficients reported here, which is con- sistent with previous SAR estimations of C1 rate coefficients.'3 As noted by Nielsen et al.," the sulfur atom appears to acti- vate only C-H bonds of the C atoms directly bonded to the S atom, unlike the OH case where the C-H bonds are acti- vated over three to four carbon atoms. Because the rate coefi- cients for the reactions of C1 with larger alkyl sulfides approach the gas-kinetic limit, any increase in the reactivity is likely to be small and difficult to observe.The atmospheric lifetimes of the alkyl sulfides studied in this work with respect to C1 are estimated to be 77, 57,47 and 40 h for DMS, EMS, DES, and DPS, respectively, assuming an atmospheric concentration for C1 of lo4 molecule ern-,. We consider these lifetimes to be lower limits because we have assumed a high atmospheric concentration of C1. If the atmo- spheric concentration of C1 in the marine boundary layer is as high as Pszenny et aL8 suggest (i.e. 104-105 molecule cm3) 4 3 +9 I 01-0 4 6 8 10 number of C atoms Fig. 2 k,/k, at room temperature us. the number of carbon atoms in the alkyl sulfide. Values of k,/k, for H,S, CH,SH and (C4H9)*S were calculated using the rate coefficients for the reaction of C1 with H,S,'* CH,SH,19 and (C4H9),Sl1 at 298 K, respectively.The value of the slope of the linear portion of the curve d C, is equal to 0.78. J. Chem. SOC.,Faraday Trans., 1996, Vol.92 371 then such a short lifetime of DMS is to be expected. Only DMS, of the alkyl sufides studied in this work, is of any atmo- spheric importance. The observed effect of 0, on the values of k, implies that there is a reaction between the adduct, CH,SCICH, and 0,. Stickel et aL5 have suggested reaction (6). CH,SClCH, + 0, +(CH,),SO + C10 Thus, reaction (6) could be an additional source of (CH,),SO (DMSO) in the atmosphere, although there are no reported mechanistic studies of the Cl-DMS system employing end- product analysis to support this claim.It has been reported, however, that the analogous reaction of the bromine analogue of the adduct (replacing C1 with Br) is slow5 which might imply that reaction (6) is slow. Obviously, further work is required for the determination of the atmospheric fate of CH,SClCH, and the branching ratio for reaction (1) under atmospheric conditions. References 1 R. J. Charlson, J. E. Lovelock, M. 0.Andrea and S. G. Warren, Nature (London), 1987, 326, 655. 2 T. S. Bates, J. D. Cline, R. H. Gammon and S. R. Kelly-Hansen, J. Geophys. Res., 1987, 92, 2930. 3 R. Atkinson, J. Phys. Chem. Ref: Data, 1994, monograph no. 2 and references therein. 4 N. I. Butkovskaya and G. Le Bras, J. Phys. Chem., 1994,98,2582. 5 R.E. Stickel, J. M. Nicovich, S. Wang, Z. Zhao and P. H. Wine, J. Phys. Chem., 1992,96,9875. 6 P. Carlier, in Atmospheric Ozone, ed. C. S. Zerefos and A. Ghazi, 1985, D. Reidel, Hingham, p. 815. 7 I. Barnes, in Dimethylsulphide: Oceans, Atmosphere, and Climate, ed. G. Restelli and G. Angeletti, ECSC, EEC, EAEC, Brussels and Luxembourg, 1993, p. 223. 8 A. A. P. Pszenny, W. C. Keene, D. J. Jacob, S. Fan, J. R. Maben, M. P. Zetwo, M. Springer-Young and J. N. Galloway, Geophys. Res. Lett., 1993,20, 699. 9 J. A. Ganske, H. N. Berko and B. J. Finlayson-Pitts, J. Geophys. Res., 1992, 97, 7651. 10 B. J. Finlayson-Pitts, Res. Chem. Intermed., 1993, 19,235. 11 0. J. Nielsen, H. W. Sidebottom. L. Nelson, 0. Rattigan, J. J. Treacy and D. J. O'Farrell, Znt. J. Chem. Kinet., 1990,22, 603. 12 N. I. Butkovskaya, G. Poulet and G. Le Bras, J. Phys. Chem., 1995,99,4536. 13 S. M. Aschmann and R. Atkinson, Int. J. Chem. Kinet., 1995, 27, 613. 14 J. A. Kerr and D. W. Stocker, J. Atmos. Chem., 1986,4263. 15 0.J. Nielsen, H. W. Sidebottom, L. Nelson, J. J. Treacy and D. J. OFarrell, Znt. J. Chem. Kinet., 1989,21, 1101. 16 R. Atkinson and S. Aschmann, Int. J. Chem. Kinet., 1985,17, 33. 17 T. J. Wallington, L. M. Skewes, W. 0. Siegl, C-H. Wu and S. M. Japar, Int. J. Chem. Kinet., 1988,20, 867. 18 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 Sratospheric Modeling, 1994, evaluation number 11, JPL pub- lication 94-26. 19 J. M. Nicovich, S. Wang and P. H. Wine, Int. J. Chem. Kinet., 1995,27, 359. Paper 5/061531; Received 18th September, 1995 372 J. Chem. SOC.,Faraday Trans., 1996, Vol. 92