2570
J. Phys. Chem. 1981, 85, 2570-2575
Kinetics of the Gas-Phase Reaction between Hydroxyl and Carbonyl Sulfide over the Temperature Range 300-517 K Ming-Taun Leu" and Roland H. Smith+ Jet Propulslon Laboratoty, California Institute of Technology, Pasadena, California G I 103 (Recelved: January 14, 198 1; In Flnal Form: Aprll29, 198 1)
By use of a discharge-flow resonance-fluorescence method the rate constant for the title reaction has been measured at five temperatures in the range 300-520 K. The Arrhenius expression is ~ O H + O C S= (1.3 f 0.3) X exp(-(2300 f l O O ) / T ) cm3s-l. Mass spectrometry has been used to detect the product HS and to collect some information about its reactivity. This study has carefully avoided the pitfalls associated with possible photolysis of reactants and complications due to H2S impurity in carbonyl sulfide that may have marred previous studies. This study has confiied that the rate constant for this reaction is so much lower than the value originally used in computer modeling of the upper and lower atmosphere that conclusions about the relative importance of photolysis of OCS and of the reaction OH + OCS in the stratosphere must now be reassessed. The reaction OH + OCS has little significance for atmospheric chemistry.
Introduction Since the detection of carbonyl sulfide, OCS, at significant concentrations in the earth's troposhere and stratosphere,l+ there has been considerable interest in determining both its source and its fate and in calculating to what extents oxidation and photolysis of OCS contribute to the production of sulfur dioxide and sulfate aerosol in the s t r a t ~ s p h e r e . Based ~ ~ upon the value obtained by KurylolO for the rate constant for the reaction OH + OCS products (1) (with the assumption that the ultimate product is SOz), these computations have shown that this reaction is a major removal process for OCS and that it would contribute significantlyto the production of atmospheric SOz. Kurylo's room temperature value for this rate constant, (5.7 f 1.2) X cm3s-', was a factor of eight higher than the upper limit previously reported by Atkinson, Perry, and Pitts.'l Both were flash-photolysis-resonancefluorescence studies with the difference in rate constant arising primarily from a difference in interpretation of fairly similar experimental results. Cox and Sheppardl2 in a steady-state photolysis, relative rate study obtained an upper limit of 4 X cm3 s-l though suggested that cm3 s-l. In a reinthe value could be as low as 8 X vestigation of the kinetics of the reaction (again by the flash-photolysis-resonance-fluorescence technique) Ravishankara et al.13 attempted to eliminate complications due to chemiluminescence from OCS and to secondary reactions of photofragments from OCS by using laser photolysis of nitric acid at 193 nm as the source of hydroxyl and obtained upper limits for the rate constants of 8.8 X 10-15, 1.89 X 10-14,and 3.27 X cm3s-l at 298,343, and 369 K, respectively. If the rate constant is as low as this, then reaction 1 is of only minor significance in atmospheric chemistry.14 In order to try to shed further light upon what is the true value for this rate constant and so help decide upon the significance of reaction 1 in atmospheric chemistry, we have investigated the kinetics and mechanism of the reaction by discharge-flow-resonance-fluorescenceand mass spectrometric methods.
-
+Onstudy leave from Macquarie University, North Ryde, N.S.W.,
2113,Australia. 0022-3654/81/2085-2570$01.25/0
Experimental Section The 2.5-cm id., 120 and 160 cm long Pyrex flow tubes had fixed hydroxyl sources (excess NO2added to H atoms produced by a microwave discharge through dilute mixtures of H2 in helium) and a fixed detector system comprising a resonance-fluoresence cell of conventional design just upstream of a pinhole leak into a two-stage molecular beam-sampling inlet to an Extranuclear EMBA I1 mass spectrometer. Carbonyl sulfide was admitted through a movable Pyrex injector (6 mm 0.d.) with the reaction distance variable from 5 or 12 cm to about 60 cm. One flow tube was fitted with a jacket for circulating a thermostatting liquid for use between 300 and 420 K while the other had a snugly fitting electrically heated aluminum tube furnace for measurements at 517 K. By measuring the temperature profile along the axis of the flow tube with carrier gas flowing by using a fine thermocouple fitted with a radiation shield, it was established that over the reaction zone used for kinetics the temperature was constant to *2 K in the liquid thermostatted assembly and to f 5 K in the electrically heated one. Flow tube and injedor surfaces were coated with phosphoric acid and conditioned by prolonged heating and evacuation followed by exposure to quite high OH concentrations for several hours before use for kinetic measurements. (1)P. L.Hanst, L. L. Spiller, D. M. Watts, J. W. Spence, and M. F. Miller, J. Air Pollut. Control Assn., 25,1220 (1975). (2)F. J. Sandalls and S. A. Penkett, Atmos. Enuiron., 11, 197 (1977). (3)P.J. Maroulis, A. L. Torres, and A. R. Bandy, Geophys. Res. Lett., 4, 510 (1977). (4)W. G. Mankin. M. T. Coffev. and D. W. T. Griffith.. GeoDhvs. - - Res. Lett.; 6 , 853 (1979).' (5)N.D.Sze and M. K. W. KO,Nature (London),278,231 (1979);280, 308 (1979). -.-, (6)J. A. Logan, M. B. McElroy, S. C. Wofsy, and M. J. Prather, Nature (London),281,185 (1979). (7)M. B. McElroy, S. C. Wofsy, and N. D. Sze, Atmos. Enuiron., 14, 159 (1980). (8) P. J. Crutzen, Geophys. Res. Lett., 3, 73 (1976). (9)R. P: Turco, R. C. Whitten, 0. B. Toon, J. B. Pollack, and P. Hamill, Nature (London),283,283 (1980). --, (10) M. J. Kurylo, Chem. Phys. Lett., 58, 238 (1:178). (11)R.Atkinson, R. A. Perry, and J. N. Pitts, Ch?m.Phys. Lett., 54, 14 (1978). (12)R.A. Cox and D. Sheppard, Nature (London),284,330 (1980). (13)A. R.Ravishankara, N. M. Kreutter, R. C. Shah, and P. H. Wine, Geophys. Res. Lett., 7 , 861 (1980). (14)N.D.Sze, private communication. " I
.--
0 1981 American Chemical Society
Kinetics of Reaction between Hydroxyl and Carbonyl Sulfide
An interference filter (Arn= = 309 nm) and photoncounting equipment were used to measure fluorescence. Typically in the absence of OCS, the ratio of fluorescence to scattered light was 4 to 5 for 10” hydroxyl radicals per cm3. On the mass spectrometer ion currents were measured by pulse counting the output of an electron multiplier. Where direct calibration was possible sensitivities were typically in the range 10-50 cps (counts per second) per 10’l molecule cm4. Other details of the apparatus were published previously.l6J6 Gas flow rates were generally measured with Teledyne-Hastings-Raydist mass flowmeters which were regularly calibrated either with a soap film-type flowmeter or by timing the pressure drop in a known volume. At 517 K a capillary flowmeter was used for OCS. Pressures were measured with MKS Baratron capacitance manometers. Since hydrogen sulfide is three or four orders of magnitude more reactive than OCS toward hydro~yl,l’-~ it was necessary to minimize that amount of HzS present in the OCS and to measure it quantitatively. Because of the natural abundance of 34Sin OCS, it was not possible to use mass spectrometry for sensitive analysis of this HzS. Instead, HzS was measured by gas chromatography on Porapak Q in a 2 m X 6 mm 0.d. stainless steel column using 2-cm3samples of OCS at atmospheric pressure. The HzS peak was identified and calibrated by subsequently injecting similar samples of OCS to which known small concentrations of HzS had been added. One cylinder of OCS contained (0.01 f 0.003)% HzS and, as a separate chromatographic analysis (on silica gel at room temperature), had shown only 40 ppm of Nz, 50 ppm of CO, and 0.5% CO,; it was thus used without purification. This cylinder was used for all measurements from 300 to 421 K. However, a second cylinder contained 0.5% HzS. Neither of the suppliers, Matheson or Linde, would provide OCS with a specification more precise than “total impurities 10.5%”. Over 90% of this HzS impurity was removed as follows: a quantity of OCS was distilled from the supply cylinder at room temperature into a 500-cm3stainless steel cylinder at 195 K which was then maintained at room temperature while 80% of its contents was allowed to slowly evaporated (by venting the cylinder to air). Finally, the sample was liquefied, outgassed, and distilled from 195 to 77 K with rejection of the last 40% of the distillate, then transferred back to the (thoroughly heated and evacuated) stainless steel cylinder for storage. After this procedure, the H2S concentration was (0.04f 0.01)% and CSz was still not detectable in the gas phase (> [OH],. Preliminary measurements at room temperature showed that the reaction was comparatively slow and that the OH decay deviated from a first-order rate law in the direction indicative of interfering secondary reactions which consumed OH. This combination of slow primary reaction, fast secondary reactions, and strong fluorescence quenching by OCS placed severe constraints upon the precision with which the rate could be measured. To minimize these problems we made the most extensive sets of measurements at 517 K (the highest temperature accessible with our equipment) and with the lowest possible initial concentrations of hyroxyl, typically 2 X 1Oll-4 X loll ~ m - The ~ . initial hydroxyl concentration, [OH],, is defined as the concentration of OH at the furthest point upstream of the fluorescence cell to which the OCS injector was withdrawn during an experiment; this was typically 60 cm. [OH], was calculated from [OH]OCeu,the value at the fluorescence cell with zero OCS present, by the equation [OH], = [OHIOeu exp(k&/u) where k, is the first-order rate constant for wall loss of OH, x the reaction distance, and u the linear flow velocity. (21)K.R. German, J. Chem. Phys., 62, 2584 (1975). (22)P. M. Selzer and C. C. Wanp, -. J. Chem. Phvs.. 71,3786 (1979).and references contained therein.
Leu and Smith
The Journal of Physical Chemlstty, Vol. 85, No. 17, 1981
2572 80
I
I
sumption of OH was calculated by a least-squares fit to the computed In [OH] vs. time data. Use of the experimental range of k2[OCS] values with k3 = 2 X lo-" cm3 s-l and [OH], = 4 X 10" cmJ causes kdcd to exceed k2[OCS] by 10-20%. With [OH], = 2 X 10" cm-3 k exceeds k2[OCS] by 5 1 1 % . Because 2 X lo-" cm3 s*$ probably high for k3 (if X is SH as deduced below), it is reasonably safe to conclude that the slope of curve c in Figure 1 approximates fairly closely to the rate constant cm3 s-l with a standard k,'. This slope is 1.67 X deviation of 3%. Use of a fixed radical source, fixed detector, and moveable injector for the other reactant under pseudofirst-order conditions should cancel out the contribution from wall reactionsB so that plots of kl vs. [OCS] should pass through the origin. Positive intercepts were obtained in Figure 1. Similar positive intercepts have occasionally been observed for other reactions by previous workers using similar techniques. Interfering secondary reactions could cause this effect by increasing the observed kl at low [OCS] but this possibility was eliminated by the computer modeling described above; for reasonable values of k3 calculated kl values were plotted against [OCS] with the resulting intercepts being too small to be the full explanation. Two recent attribute such intercepts to a change in the first-order surface removal rate constant for OH due to the presence of the added reactant. This could be the explanation in our study also, though no positive supporting evidence could be obtained. At 517 K if k,' is calculated from kl/[OCS], then from the 10 experiments of curve a in Figure 1, k,' is 1.94 X lO-" cm3 s-l compared with 1.67 X cm3 s-l from the least-squares slope. The standard deviations are 0.10 X and 0.04 X 10-14,respectively, so that the difference is only just significant. Since we were unable to establish with certainty the cause of the positive intercepts in Figure 1, we took as our best estimate of kd the average of the slope and the mean kl/[OCS] value and assigned to it a statistical error of two standard deviations in the slope: cm3 s-l at 517 K. that is, k,' = (1.8 f 0.1) X Since for these experiments the OCS contained (0.04 f 0.01)% H2S, a correction for this was applied. At this temperature we have obtainedm a value of 5.5 X cm3 s-l for kOH+Hfi in agreement with Perry et al.l9 (and in disagreement with Westenberg and deHaasl') and thus the cms s-l to give k2 value of k,' was reduced by 0.22 X cm3 s-l = 1.6 X cm3 s-l at 517 K. That 0.22 X and not twice this value was the proper correction to apply was confirmed by numerical integration by adding OH + H2S X to the scheme above (where X is SH). To estimate the systematic error in this value, we write the equations
':40y//PRESSURE (torr)
0 5.9
30 20
0 1.3
0 2.0 A 3.8
A 1.65
(c) 1.8
3.6 0 6.1 "
0
1 .o
2.0
3.0
1 4.0
10-l~ [OCS] (cm-3)
Figure 1. Dependence of the apparent flrst-order rate constant k , upon [OCS] at various values of [OH], and total pressures at 517 K. For (a) and (b) the origin has been displaced veiticaliy by 10 and 5 units, respectively.
and [OCS] 1 6 X At 517 K with [OH], I 2 X 10" 1014~ m hydroxyl - ~ decay was approximately first order for 2 to 3 half-lives. At slightly higher [OH], (3 X 1Oll-7 X loll) curvature was detectable in the firsborder decay plots (in the direction of increasing first-order rate constant with increasing reaction time) and in these cases an average slope was calculated over about the first half-life of reaction. To determine the order of the reaction in [OCS] several sets of runs were performed. In each set constant flowrate, pressure, and [OH], were used while the first-order decay constant, kl, was measured for each of several different OCS concentration and kl plotted against [OCS]. The linearity of such plots, shown in Figure 1, established that the reaction was first order in [OCS]. To determine whether or not the apparent second-order rate constant ki was dependent upon total pressure, we performed such sets of experiments at different total pressures but with constant [OH],. When [OH], = 3.0 X 10" ~ m -sets ~ , were performed at totalpressures of 1.3,2.0, and 3.8 torr and, as curve b in Figure 1 shows, there is no apparent pressure dependence. Similarly with [OH], = 1.8 X 10'l cm-3 sets were performed at 1.65, 3.6, and 6.1 torr and again (curve c) there is no evidence for pressure dependence. Finally, to check whether secondary reactions were contributing to the rate of removal of OH, sets were performed at different [OH],. Comparison of the slopes of curves a, b, and c in Figure 1 shows that decreasing [OH], does decrease the apparent second-order rate constant k,' (from 2.40 X to 2.02 X to 1.67 X cm3 9-l). Since we had to work with rather low values of k2[OCS] (typically 15-40 s-l), partly because k i was comparatively small and partly because only modest values of [OCS] could be used in order to avoid excessive quenching of the OH fluorescence, it was necessary to check that secondary reactions were not interfering. To assess the possible contribution from such reactions the scheme OH + OCS 4 X + Y (k2) (6) (k3) (7) OH + X products was integrated numerically (fourth-order Runge-Kutta method with stepsize adjusted by algorithm to meet a specified precision of 0.1%) and kdd, an average firstorder rate constant (= -d In [OH]/dt) over 75% con-
-
-
V =
[OCS] =
(760 torr)FbMT ?rr2(293K ) P
FoCsP(9.662X 10l8 cm-3 K-l) FtotalT
k i = -(u/[OCS]) d In [OH] dx
(8)
(9) (10)
where u is the linear flow velocity, Fbd and FOCSare the total flow rate and the flow rate of OCS (expressed in volume of gas at 760 torr and 293 K per unit time), r is (23)A. A. Westenberg and N. deHaas, J. Chem. Phys., 46,490 (1967). (24) U. C. Sridharan, B. Reimann, and F. Kaufman, J. Chem. Phys., 73, 1286 (1980). (25)L.F. Keyser, J.Phys. Chem., 84,1659 (1980).
The Journal of Physlcal Chemlstty, Vol. 85, No. 17, 1981 2573
Kinetics of Reaction between Hydroxyl and Carbonyl Sulfide
TABLE I: Rate Constants for OH + OCS no. temp, of 10-15[OCS], K runs cm3 k,’a
300 364 389 421 517
16 7 6 15 10
4.3-18.2 1.0 5.5-19.8 2.5 3.8-13.1 4.2 3.2-16.2 5.4 0.6-2.3 18.0
6
1015k,a
0.6 f 2.1 f 3.8 f 5.0 r 16.0 f
0.4 0.9 2.0 2.0 4.0
weighting
1 1 1 2 5
5
k , is the corrected rate constant; that is the average measured rate constant h , ’ less the contribution from the known concentration of H,S in the OCS. h o ~ + H , oS; 5.9 x lo-’, exp(- 89/T)was used for this correction. Units for both k , and k,’ are cm3 s-l.
-
4
N
Y
‘0 0 =.
-C
the radius of the flow tube, and P and Tare the pressure and temperature of the experiment. Hence
k; =
aFbM2T2 d In [OH]/dx r2P2Focs
3
(11)
where a incorporates all the numerical constants. Errors of 0.5% in r, 0.5% in P , 1% in T, and 1% in each Fbd and Foes lead to an overall systematic error of 7% in ki. Combination of this plus two standard deviations for statistical scatter plus 3% for uncertainty in the H2S correction plus a possible 8% contribution from secondary reactions at this low [OH], leads to our final estimate: cm3 s-l at 517 K k2 = (1.6 f 0.4) X with k2 independent of pressure in the range 1-6 torr. In order to obtain reasonable precision with the discharge-flow technique, we found that the pseudo-first-order rate constant for the reaction being studied should be significantly greater than the rate constant for wall loss of OH which in this case was about 10-20 s-l. At lower temperatures the slow rate of the OH + OCS reaction required comparatively high values of [OCS] to achieve this (see Table I) which in turn means that the OH fluorescence signal is so strongly quenched that reasonable precision is virtually unattainable with [OH], I5 X 1011 ~m-~. At lower temperatures then, an alternative procedure for minimizing the effect of secondary reactions was adopted. It was found that relatively high concentrations of NO, appeared to remove the interference. Normally OH was produced by adding [NO,], = 2 X 10l2to 3 X lo1, cm4 to the stream of hydrogen atoms in helium. At 300 and 421 K under these conditions, the first-order decay plots were markedly c w e s in the direction of increasing kl with increasing reaction time. However, as [NO,], was increased the curves became to about 2 X 1013 or 3 X 1013 approximately linear with the overall slopes similar to the initial slopes in the low [NO,], experiments and with the overall slopes independent of [NO2lo. This implies that in addition to reactions 2 and 3 there is a reaction X+NO,-R+S (12) which at high [NO,], can compete so effectively with (3) that (3) contributes only negligibly to the overall decay of [OH]. These observations are consistent with the mass spectrometric measurements described below. Hence over the temperature range 300-421 K high [NO,], was used and the rate constant calculated from the average slope. Values of [OH], used were generally in the range 3 X loll to 8 X loll ~ m - At ~ . 517 K, where we were able to use lower [OH],, the measured pseudo-first-order rate contant was independent of [NO,], in the range 2 X 10” to 6 X lo1, ~m-~.
2
1
0.0015
0.0020
0.0025
0.0030
0.0035
1/T (K-’)
Figure 2. Dependence of the second-order rate constant k, upon temperature.
At 300 and 421 K when kl was plotted against [OCS], least-squares fits to the results again produced positive intercepts, but, because of the greater scatter of the results at these temperatures, the significance of these intercepts was even more dubious than at 517 K. However, as at 517 K the final value of k; was taken as the average of the slope and the mean kl/[OCS] value. At 389 and 364 K, where fewer experiments were performed, k i was calculated only from kl/[OCS] in order not to give excessive weight to experiments at the ends of the [OCS] ranges used. These k i values were corrected for the contribution of the 0.01% H2S present in the OCS cylinder used for these experiments. Results are summarized in Table I. At 421 K totalpressure ranged from 1.6 to 10.2 torr with four experiments below 3 torr and six above 7 torr; there was no evidence of any pressure dependence for ka. Similarly at 389, 364, and 300 K the pressure ranges were 3.8-9.2, 2.2-10.5, and 3.6-7.5 torr, respectively, and again no pressure dependence was detected. Although the accuracy at these temperatures is insufficient to establish beyond doubt the absence of some pressure dependence, the results are consistent with the more accurate 517 K results which tended to establish that the reaction was in a true second-order regime. The results are plotted in Figure 2 and were fitted to an Arrhenius equation by the least-squares method using the weightings given in Table I. These weightings were chosen in an arbitrary way to reflect both the precision and the number of experiments at each temperature. Provided the 517 K rate constant received a weighting about equal to all the others combined (because measurements were much more precise there), the slope was not particularly sensitive to the weightings used. Over the temperature range 300-517 K our results can be summarized by the equation
k2 = (1.3 f 0.3) X lo-’, exp(-(2300 f 100)/T} cm3 s-l (13)
2574
The Journal of Physical Chem;stry, Vol. 85,No. 17, 1981
where the errors are 2 standard deviations of the leastsquares fit. Product Analysis. Product peaks were most clearly detected with the mass spectrometer by injecting OCS well upstream of the mass spectrometer leak (to allow long reaction times for OH + OCS) and by performing discharge on (i.e., OH present), discharge off (zero OH) experiments while monitoring suspected product peaks one a t a time. Such experiments were performed only at 517 K. In this way products were seen at m l e values of 33 (at an electron impact voltage of 15 V to eliminate the 33 peak from OCS), 49, and 64 (both at 30 V). These are presumably HS, HSO, and SOz. With [OH], N 7 X 10l1 12m-~,[OCS] N 8 X 1014~ m - ~ , ~ ) was detected and relatively low [NO,], (2 X lo1, ~ m - HS a t about 30-40 cps which, by means of an indirect calibration using H2S + H as a source of approximately calculable HS concentrations, corresponded to about l X loll cm-3 (probably &50%). HSO was detected at about 4 cps though this was barely distinguishable about the noise and SO,could not be seen. When [NO,] was increased to about 3.4 X lo1, cm4 the HS signal dropped to about 12 cps while HSO became clearly visible at 20 cps. At a still higher [NO,] (6 X lo1, to 7 X lo’, ~ m - HS ~ ) could not be seen, HSO fell to a barely detectable level, but SO, could be clearly detected at about 40 cps (-1 X loll ~ m - ~ ) . At the low [NO,] the amount of NO, consumed (as measured by discharge on/off) when OCS was present was virtually identical with that consumed without OCS present (this was the amount of NO, which reacted with H to produce OH). However, at higher [NO,] significantly increased amounts of NO, were consumed when OCS was present which lent support to the argument that NOz was oxidizing one or more products of the primary reaction. Under some conditions (e.g., [OH], = 7 X lo1’ ~ m - ~ , [OCS] = 2 X 1015 ~ m - [NO,] ~ , = 3.5 X 10l2~ m - it~ was ) possible to measure both HS and HSO as a function of reaction time (injector distance from ms leak). Such experiments showed a steady buildup of HSO but a rapid production of a maximum in the concentration of HS followed by a distinct fall. Attempts to see an adduct, OC(S)OH, at m l e 77 were unsuccessful. The two most likely sets of products for the OH + OCS reaction are CO, + HS and CO HSO. Because of high mass spectrometer backgrounds at m l e 44 and 28 it was not possible to detect C02or CO so conclusions about the path of the reaction had to be based upon the HS and HSO observations. Our experiments indicated fairly conclusively that HS was the primary reaction product: HO + OCS HS + CO, (14) and that it was rapidly oxidized by NO2 to HSO and ultimately to SO,: HS + NO2 HSO + NO (15) (16) HSO- -SO, Interference by secondary reactions in the kinetic measurements indicated that OH reacted with HS also, presumably by HS + OH H20 + S (17) Competition between (15) and (17) is consistent with our earlier observation at the lower temperatures that excess NO, could minimize the kinetic interference by secondary reactions and result in near linear first-order OH decay curves. These conclusions about the mechanism of the OH + OCS reaction are consistent with the conclusions drawn
+
+
4
-+
Leu and Smith
+
about the reactivity of HS in our study of OH HzS.,O Our rate constants at 300 and 364 K are less than one tenth of the values reported by Ravishankara et al.13 Hence we investigated the possibility that in our system a chain reaction had been initiated which regenerated OH so that the net rate of OH removal was much less than the rate of reaction 14. One example of such a chain would be reactions 14,15, HSO H + SO, and H + NO2 OH + NO. Regardless of the details, it is hard to imagine any source for OH other than NO,, since apart from OCS itself there is no other source of elemental oxygen. Hence as described above A[N02] was measured (discharge on/off) both with and without OCS present under a variety of conditions. While the extra NO2consumed when OCS was present varied from zero (at low [NO,],) to 50-100% of the amount consumed with zero OCS (at higher [NO2]), this observation is entirely consistent with the proposed reaction scheme and fairly convincingly eliminates the possibility that a chain reaction has been inadvertently initiated.
-
-
Discussion The discrepancy between our rate constant and that obtained by Ravishankara et aI.,l3 a factor of more than 10, is of considerable concern. Those workers showed that the measurements previous to theirs were erroneous because of chemiluminescence from OCS and interference from fragments from the photolysis of OCS. They therefore produced OH by the laser photolysis of nitric acid at 193 nm where HN03 had an absorption cross section about 1000 times that of OCS. They used ratios of [OCS] to [HN03]ranging from 30 to 200 which meant that OCS absorbed 3-20% as many photons as “OB. Since the primary product of such OCS photolysis is S, which at this wavelength may be excited as well,mthere is the possibility of interference by reactions of S leading to increased rate of removal of OH. However, this could hardly explain a factor of 10. On the other hand, H2Shas an absorption cross section so that, if at 193 nm not much less than that of 0.1% H2S were present in their OCS, its concentration ~, would typically have been 1 X 1012-7 X 10l2~ m - while ~ ; could have absorbed about [HN03]was 3 X 1013~ m -H2S 10% as many photons as HN03 which would have meant [HS] = [HI N O.l[OHjo. Our experience here and in ref 20 is that HS reacts rapidly with OH and in addition the reaction of H with H,S to produce more HS would not be insignificant compared with the rate of OH + OCS (based on our rate constant). Hydrogen sulfide at a level of 0.1% in their OCS could conceivably explain the discrepancy. The purification procedure used by Ravishankara et al. would not have removed H,S if it had been present, and, as mentioned above, cylinders of OCS can contain up to 0.5% H2S. Following Kurylo’s suggestionloof adduct formation, we can interpret our results in terms of the scheme OH + OCS F? OC(S)OH* (ka,12.J OC(S)OH* C02 + HS (kb) OC(S)OH* + M OC(S)OH + M (12,)
--
where OC(S)OH* is an excited adduct which can (a) fall apart to reactants, (b) fall apart to products, or (c) become collisionally stabilized. Application of the steady-state hypothesis to OC(S)OH* leads to (26) H.Okabe, “Photochemistry of Small Molecules”, Wiley, New York, 1978, p 215. (27) Reference 26, p 204.
J. Phys. Chem. 1981, 85, 2575-2582
This scheme is consistent with the observed second-order rate law and with the absence of any pressure dependence for k,. If k-, >> kb, then k2 = k , k b / k a with the experimental activation energy given by Ez = E, - E-, + E b As explained (p 111 (case (a)(i)) in the analysis by Eb - E-, may be positive or negative so that this scheme is consistent with a positive E2. Alternatively, if k-
