Kinetics and mechanism of the reaction of oxygen ... - ACS Publications


Kinetics and mechanism of the reaction of oxygen...

0 downloads 141 Views 804KB Size

The Journal of Physical Chemistry, Vol. 83, No. 17, 7979

Reaction of Oxygen Atoms with Hydrogen Sulfide (25) (a) E. C. Avery, J. R. Remko, and B. Smaller, J. Chem. Phys., 49, 951 (1968); (b) F. P. Sargent and E. M. Gardy, Chem. Phys. Lett., 39, 188 (1976). (26) R. W. Fessenden and R. H. Schuler, J. Chem. Phys., 39,2147 (1963). (27) J. E. Bennett, B. Mile, and B. Ward, Chem. Commun., 13 (1969). (28) R. Livingston and H. Zeldes, J. Chem. Phys., 44, 245 (1966). (29) R. Livingston and H. Zeldes, J. Chem. Phys., 45, 1946 (1966). (30) “High Resolution NMR Spectra Catalog”, Varian Associates, Palo Alto, Calif., 1962. (31) “The Aldrich Library of NMR Spectra”, C. J. Pouchert and J. R. Campbell, Ed., Aldrich Chemical Co., Milwaukee, Wisc., 1974. (32) I n view of the low ratio of acetyl radical which is expected to be

2195

present, this intermediate seems to play an inordinately important role in product formation and polarization. Correspondingly, the contributions of the hydrate radical to these processes are less than would be expected. These observations have led us to conskler the possibility that the hydrate radical dehydrates to the acetyl radical: CH,C(OD),

* CH,CO

-F D20

While this process woukl contribute to the acetyl radical concentration at the expense of that of the hydrate radical and explain the above observations, it must be noted that, for it to be significant, a rate constant greater than 10’ is required.

Kinetics and Mechanism of the Reaction of Oxygen Atoms with Hydrogen Sulfide’ Donald L. Singleton,* Robert S. Irwin, Wing S,, Nip,t and R. J. Cvetanovl6 Division of Chemistry, National Research Council of Canada, Ottawa, Ontario, Canada K I A OR9 (Received April 3, 1979) Publication costs assisted by the National Research Council of Canada

Rate constants were determined by a phase shift technique for the reaction of oxygen atoms with hydrogen sulfide. Over the temperature interval 297-502 K, the results are consistent with the Arrhenius equation k l = (1.56 f 0.83) X 1O’O exp[(-2171 f 202)/T] L mol-’ s-l, where the uncertainties are 95% confidence limits for three degrees of freedom. Potential sources of systematic errors are discussed and their magnitudes estimated. In separate experiments, oxygen atoms were generated by mercury photosensitized decomposition of NzO at 700 torr, and the products of their reaction with HzS were analyzed by gas chromatography. Besides N2, the only detected products were H20,H2, and a very small amount of 02.The results suggest that about 52% of the reaction proceeds by way of hydrogen abstraction from H2S,and less than 11%by way of addition of oxygen atoms to H2S followed by decomposition of the adduct to form H and HSO.

Introduction The kinetics and mechanisms of the reactions of ground state oxygen atoms with hydrogen sulfide and organic sulfides have been the subject of several recent studies which produced somewhat surprising results. Mass spectrometric detection of the products resulting from crossed beams of oxygen atoms and organic sulfur compounds suggested an addition mechanism’ in which the initial adduct fragmented or was stabilized, analogous to the well-known addition of O(3P) to olefim2 The very large rate constants and the negative Arrhenius activation energies for the organic sulfides could be accomodated by this rne~hanism.~ It has been suggested that the reaction with hydrogen sulfide also could proceed by an addition mechanism (Ib) 0 + H2S OH + SH (la) 0 + H2S [H2SO]* HSO + H (1b) instead of entirely by direct abstraction (la), although it was emphasized that not enough data existed to reach definite concl~sions.~ Evidence that an addition mechanism could occur, at least under very special conditions, came from a matrix isolation study in which the species HSOH was identified as a product of the photolysis of mixtures of O3 and H2S in an argon matrix? Presumably the direct abstraction route, reaction la, has too high an activation energy to occur at the temperature of the matrix, 8 K, and it was suggested that HSOH was formed by decomposition and secondary reactions of an HzSO adduct in a matrix cage, Le., 0 + H 2 0 [H2SO] HSO + H HSOH. Subsequent studies of chemiluminescence from 4

+

-+

-

-

-

NRCC No. 17548. NRCC Research Associate. 0022-3654/79/2083-2195$0 1.OO/O .. -

this system were consistent with this m e c h a n i ~ m . ~ ! ~ The reported room temperature rate con~tants~*~-’l for reaction 1vary by more than a factor of 4, although the two most recent ~ a l u e s are ~ 3 ~within 20%. However, the same two studies yielded quite different temperature dependences, such that the reported rate constants at 500 K differed by loo%, and both results differed significantly from the only other reported Arrhenius parametersS8 The present work was done to provide additional information, using quite different experimental methods, which could help resolve some of the questions concerning the kinetics and mechanism of reaction 1. Rate constanints were determined with a phase shift technique, and several separate experiments were carried out to determine the products.

Experimental Section Rate Constant Determinations. Oxygen atoms were generated by modulated mercury photosensitized decomposition of nitrous oxide.l2-I4 The relevant reactions are

+ + + + + + - +

Hg6(’So) + hv(254 nm) Hg6(3Pl) N20

Hg6(lSo)

O(3P) H2S o ( 3 ~ ) NO

Hg6(3P1)

M

N2

products

NO^

M

O(3P) (1) 2)

The phase shift between the modulated 254-nm light incident on the reaction cell and the chemiluminsecence resulting from reaction 2 was measured with photomultipliers and a lock-in amplifier. Pressures and flow rates of the gases were measured as before14 with the exception of HzS, for which the pressure drop across the capillary

0 1979 American

Chemical Society

2196

The Journal of Physical Chemistry, Vol. 83, No. 17, 1979

used to determine its flow rate was measured with a Validyne transducer. The rate of absorption of 254-nm light in the reaction cell was about 2 X einstein s-l, as determined by reaction of O(3P) with l-butene.' For the longest residence time of 1 s, this led to a 0.02% consumption of H2S. The hydrogen sulfide, stated to be 99.5% pure, was used directly from the cylinder and was passed through a silica gel trap a t 273 K. Analysis of the H2S by gas chromatography on a Porapak QS column (50-80 mesh, 30 cm, 343 K) with a thermal conductivity detector indicated 0.01% COz and a trace of water and air. Carbon disulfide and methanethiol were not detected, but upper limits of 0.05% were established. Nitrous oxide and nitric oxide had stated purities of 99.0%. The latter was passed through Linde 13X molecular sieve to remove NO2. Product Analysis. Oxygen atoms were produced by mercury photosensitized decomposition of nitrous oxidea2 Mixtures of H2S and N20,each degassed and distilled from trap to trap, were exposed to 254-nm light in a 250-cm3 reaction cell containing a drop of mercury and fitted with a circulating pump to ensure constant mercury concentrations. A Corning 9-54 filter blocked 184.9-nm light emitted by the low pressure mercury lamp. Products were identified and quantitatively determined by gas chromatography. Noncondensible (at 77 K) products were collected with a Toepler pump and analyzed. A trap containing the condensible products and reactants was slowly warmed from 77 K, and the volatile constituents (HzO,H2S, and NzO) were qualitatively analyzed directly, or were passed through two traps at 195 K to collect H20. Quantitative analysis of HzO was otherwise impossible with large amounts of H2S and NzO present. Propane was then added as an internal standard, and the water was determined on a Porapak QS column.

Results Rate Constants. Because the rate constant for reaction 1is rather small, high concentrations of H2Swere necessary in order to obtain accurately measurable phase shifts. This meant that, in some experiments, H2S contributed significantly as a third body in reaction 2, and the formerly used treatment of the data had to be modified. The phase shift, 4, in the present work is related to the various rates of oxygen atom reactions by the equation 2TV -= kl[HzS] + k2[NO]{[NzO] + a[HZS] + P[NO]) tan 4 (1) The third body efficiency, p, of NO relative to NzO for reaction 2 can be obtained from the combined results of ref 15 and 16, but there is no reported value for a, the third body efficiency for M = H2S in reaction 2. By rearranging eq I, we obtain

(11)

The ratio [N20]/[H2S]was held constant in a series of measurements in which [NO] was varied. Thus a value of k l was obtained from the intercept of a plot of the left-hand side of eq I1 vs. [NO]. Examples of plots of eq I1 at 297 K and 30 torr are given in Figure 1 for three different ratios of [N20]/[H2S]. Good linearity was consistently observed. Values of a could in principle be determined from the slopes of plots of eq I1 by plotting the slope against

Singleton et ai.

0

I

2

3

4

5

[ N O ] ( I O - ~ ~ Oi') I Figure 1. Plots of eq I1 used to determine values of k, at 297 K and 30 torr. For the filled symbols, the intensity of the 253.7-nm light was 10% of its usual value: (0) [N20]I[H2§]= 10.48; (0) [N20]/[H2S] = 5.06; (A)[N20]/[H2S]= 1.94.

[N,O]/[H,S]. The slopes of these new plots give kz,and the intercepts, ahz. The values of k2 obtained a t each pressure at 297 K were consistent with the previously determined va1ue,14but the values of a varied from 1.5 to 0.6, depending on the pressure. The correction term, Pk2[N0]2/[H2S],which accounts for M = NO in reaction 2 had a maximum value of 25% of 2 ~ v / t a n4[HzS], but was usually much less. The value of 0 used was 0.82, but the intercepts, kl, were not very sensitive to the value of 0.A change of 20% in P produced typically a 2% change in kl. The results are given in Table I. At room temperature, the average values of kl did not vary significantly for total cell pressures of 30, 70, and 180 torr. Also, within the experimental uncertainty, there was no significant effect upon increasing the residence time in the reaction cell a t 30 torr to its value at 180 torr. Reduction of the intensity of the 254-nm light incident on the reaction cell to 10% of its usual value did not affect the results, as shown in Figure 1. The possibility of an extraneous emission affecting the phase shift measurements was examined by using different sharp cutoff filters between the reaction cell and the photomultiplier viewing the chemiluminescence. The filters (and the wavelengths at which the transmission was 40%) were Pyrex (-300 nm), Corning 3-70 (500 nm), and Corning 2-61 (620 nm). At 30 torr and 297 K, the variation in the value of the left-hand side of eq I1 was less than 10% among the three filters, both for large and for small concentrations of [NO], as shown in Table 11. Also, the relative intensities of the chemiluminescence determined with the three filters were similar to those in the absence of H2S (Le., as a result of reaction 2 alone), except at the higher concentration of NO with [N20]/[H2Sl = 0.71, where there appears to be slightly more intensity with the 3-70 and 2-61 filters. For values of [N20]/[HzS] < 1,the results were not reliable because of the low signal-to-noise ratio (due to quenching of Hg6(3Pl) by H2S) and lengthy extrapolation to [NO] = 0. In one set of experiments an ascarite and molecular sieve trap was inserted in the N 2 0 line to remove any NO2 impurity, which could cause the measured intercept to be

The Journal of Physical Chemistry, Vol. 83, No. 17, 1979 2197

Reaction of Oxygen Atoms with Hydrogen Sulfide

40

TABLE I : Intercepts, k , , and Slopes, k,([N,O]/[H,Slt a),of Plots of Eq

press., [N,Ol/ torr [H,S]

Nb

30

10.63‘ 10.48 5.06 1.94

15 36 19 22

3Od

7.72 7.64 7.42 6.09 5.68 5.28 3.98 3.70 2.72

8 5 7 6 8 6 6 5 6

70

10.28 5.08 1.93

14 17 21

2-3 2-6 2-6

1.11i 1.22 2 1.14 * 1.09 i

0.06 0.21 0.06 0.16

5.73 f 0.44 3.22 i 0.12 1.59 i 0.10

180

8.75 7.30 6.20 4.93

17 11 10 10

6 6 6 6

1.14 * 0.93 f 1.02 f 1.20 i 1.15 i

0.11 0.12 0.08 0.44 0.28

4.52 i 3.81 i 3.03 f 2.69 f

10.61 5.69 1.99

13 5 10

331 K 0.5-2 2 . 5 6 i 0.13 2 2.25 i 0.31 2 1.99 f 0.13

10 5 5 5

373 0.5-2 3 3 3

I

I

I

I

I

I

I

I

I

-1

slopeg

kHz interceptf 297 K 0.5-2 1.17 i. 0.06 0.5-2 1.32 f 0.08 2 1.112 0.07 2 1.01 i 0.08 1.15 i: 0.18 2 0 . 8 6 i 1.50 2 1.09 2 0.32 2 0.98 2 0.23 2 1.27 c 0.28 1.09 i 0.08 2 2 1.15 f 0.20 2 1.27 f 0.16 1.11f 0.44 2 2 1.07 c 0.08

v,

I

6.57 c 6.28 i 3.59 f 1.96 f

0.11 0.10 0.08 0.06

5.42 i 4.64 i 4.76 f 3.94 c 3.77 i 3.69 i 2.69 2 2.51 c 2.31 i

1.49 0.39 0.23 0.35 0.07 0.24 0.19 0.44 0.08

0.5 2

+

I

0.6

+_

5

4

Flgure 2. Arrhenius plot of rate constants for reaction 1, 0 HpS: ( 0 )this work; (0)ref 7; (0)ref 8; ( 0 ) ref 9; (A) ref 10; (V)ref 11; (+) ref 3.

0.15 0.14 0.43 0.25

5.45 f 0.17 3.19 0.30 1.69 c 0.07

3

IOOO/T (K-’)

Oe5

1.01 f 0.11 1.10 i O.Oge 30

I

X

IIU

I

I

AI

1

t

I

I

A

A

K 0.2

2.25 c 0.80 30

16.52 8.47 2.48 1.38

K 4.40 4.39 2 4.23 t 4.16t

6.91 c 0.17 3.80 2 0.36 1 - 5 2 ? 0.40 1.07 f 0.16

0.20 0.41 0.47 0.35

4.31 f 0.20 30

18.89 5.56 2.89 2.01 1.66

7 8 6 5 5

426 0.5-2 2-3 3 3 3

K 9.25 -t 8.62 f 8.66f 8.15 i 7.75 f

0.23 0.36 0.53 0.79 0.53

6.50 f 2.40 c 1.24 c 1.22i 1.22 i

0.24 0.33 0.44 0.74 0.25

7.90 i 3.84 f 2.01 f 0.57 i

0.59 0.78 2.25 2.78

8.78 c 0.75 30

30.8 13.8 6.81 3.16

5 5 5 5

2 2 2 3

502 K 22.02 21.2f 20.3 c 21.6 f

0.5 0.7 2.1 2.6

21.6 2 0.8 The indicated uncertainties are the 95% confidence limits calculated from Student’s t distribution for N - 2 degrees of freedom. Number of points. An ascarite trap was inserted in the N,O line for these experiments. The total flow rates were 0.1-0.2 times the flow rates in the other sets of experiments at 30 torr. e Weighted average of the four mean values at 297 K. X 10” L mol-’ s - l . g x IO-” Lz mol-, s - l . a

*



greater than hl, particularly for the larger ratios of [N20]/[HzS]. As seen in Table I, the intercept obtained with the trap in place a t room temperature (30 torr, [NzO]/[H2S] = 10.6) was about 13% lower than that obtained without the trap. An NOz impurity of 26 ppm in the NzO would account for the difference, assuming negligible loss of NOz by reaction with mercury in the gas

0

I

I 2

I 4

I 6

1 8

I IO

I

I 12

14

YIELD OF Ne (pmoles)

+

Figure 3. Relative rates of formation of products in the 0 HS , reaction = 10 torr; PNIO= 700 system as a function of the yield of N2. PHzS torr: (A) H20; (0)HP; (0) 02.The values for R,,/RN, have been multiplied by 10.

handling system. The concentration of NOz in several cylinders of N20 was found to range from 2.8 to 0.01 ppm, and the larger value was reduced to less than 0.03 ppm by inserting the ascarite and molecular sieve trap in the sampling line. It could be possible that a cylinder of N20 could contain as much as 26 ppm of NOz. If a correction is made for this amount of NOz, the average of the four values of hl in Table I a t 297 K and 30 torr (with the shorter residence time) would be lowered by 4.5%. At higher temperatures, the average value would be lowered even less, 1.4% at 500 K. At other pressures, where the residence time is increased, there is increasing opportunity for NOz to react with mercury, and any correction would be smaller. The Arrhenius equation for kl, calculated from the average values for the five temperatures in Table I, weighted by l / a 2 , is

hi = (1.56 f 0.83) X 1O1O exp[(-2171 f 202)/T] L mol-’ s-l

where the indicated uncertainties are 95 % confidence limits for three degrees of freedom, using a Student’s t value of 3.182. The equation is plotted in Figure 2. Potential systematic errors are estimated in the Discussion section.

2198

The Journal of Physical Chemistry, Vol. 83,

No. 17,

1979

Singleton et al.

TABLE 11: Effect of Different Glass Filters, Used to Isolate the Chemiluminescence, on Values of Pnvltan @[H,S]- D ~ , [ N O ] ~ / [ H , S ] ~ [N,01/ [H,S]

[NOIC 0.99

filter Pyrex 3-70 2-61

0.1

0

--OM

c

--------..........................

.................................. ........................... O

O

0.01

0.02

$1

0.03

1h 2 d

+

(.e*)

= 0.

sumption is made that reaction 1 occurs entirely by way of channel l a , that OH and H formed in reactions 1and 4 immediately react with HzS to form SH, and that the only secondary reaction of importance involving oxygen atoms is 0 + SH .-+ SO f H (8) for which the rate constant is reported to be 9.6 X 10loL mol-1 s-1,9bthen, for steady-state conditions a t room temperature, the ratio of the rates of reactions 8 and 1, rg/rl,is less than 0.033 for [N20]/[HzS]= 10.6, and less = 1.94. This treatment igthan 0.0084 for [N20]/[H2S] nores any effect that the NO present in the kinetic experiments might have by scavenging the reactive radicals to form nonreactive products. Reactive impurities in the H2S,if present, could affect the measured values of kl. The reactions of oxygen atoms with CH3SH3J7and with CS218are 100 and 200 times faster, respectively, than reaction 1. Although the manufacturer of the H2Sused in the present work lists CH3SH and CS2 as possible impurities, neither was detected by gas chromatography, but upper limits of 0.05% were established. They could therefore make, as an upper limit, a 15% contribution at room temperature to the measured rate constant. Also, as discussed in the previous section, the average room temperature value of kl could be 4.5% too large because of a possible impurity of NOz in the N20. The systematic errors considered above would tend to make the measured value of kl larger than the true value. As an upper limit, these factors could have introduced a maximum combined potential systematic error of +23 % at 297 K. Possible systematic errors that could operate kqOoa

Hg6('S0) t hv(253.7 nm) -+ Hg6(3Pl) Hg6(3P,) t N,O -+ HgG('S,) + N , t 0 Hg6(3Pl)+ H,S -+ HgG('S,) t H t SH

ref

1.43 X 10" 2.9 X 10''

26

1.10x 107

this work

3.1 x 109 4.3 x l o 8

28 18

lo9

18

26

OH t SH 0 t H,S%

(51 (6)

[H,SO]* OH t H,S -+ H,O + SH H t H,S -+ H, + SH

(7)

SH

+ SH

[H,SO]*

Units are L mol-' s-'.

H

-+

?

+ HSO

-rHZS+3

7.2 X

+ H , t S,

a

+

0.04

Figure 4. Relative rates of formation of products in the 0 H2Sreaction system as a function of the ratio of H,S and N20 concentrations: (A) H20; (0) H,. The lines are calculated values of RH,IR,, as described in the text. klblkl = 0,k7,1k7 = 0.13;(---) kl,lkl = 0.11, k7Jk7

TABLE 111: Model Reaction Mechanism reaction

(1)

I

LL CH

Discussion Estimation of Systematic Errors in kl. The reaction mechanism in Table 111, which will be discussed in a later section, can be used to estimate the extent of reaction of oxygen atoms with products of the primary reaction, reaction 1, in the rate constant determinations. If the as-

(4)

-_---

O . 1

Product Analysis. The relative rates of formation of products are given in Figures 3 and 4. The only products detected were N2,H20,Hz,and a small amount of OD The rates of product formation, relative to that of N2, are essentially independent of the extent of reaction (i.e., the yield of N2, which is a direct measure of the number of oxygen atoms formed by mercury photosensitized decomposition of N20). Also, the relative rates of H 2 0 and O2 formation are independent of the pressure of H2S,but the relative rate of H2 increases with increasing concentration of H,S. The major detected product, and the only one containing oxygen, is H 2 0 ,but it constitutes only about 50% of the reacted atomic oxygen. No sulfur containing products were detected, except for a black deposit on the window of the reacAn cell which was probably a mercury-sulfur compound.

(3)

A

(2nvltan @ Dh,[N012)/ intb [HZSld

100 1.67 66 1.52 29 1.55 10.5 10.8 Pyrex 100 8.45 3-70 70 8.27 8.06 2-61 23 0.72 9.0 Pyrex 100 1.86 3-70 68 1.99 2-61 28 2.02 0.71 34 Pyrex 100 4.73 3-70 75 5.09 2-61 33 4.64 to 15 Pyrex 100 3-70 66 2-61 23 Conditions: total pressure, 30 torr and temperature, 296 K. Intensity of the chemiluminescence ( i q O 2 t i o 2 ) ' " ,normalized to the results for the Pyrex filter, where i,, and i o are the signals 90" out-of-phase and inphase, respectively, with the modulated 253.7-nm light. Units of mol L-' x 105. d Units of L mol-' s-l x io-'. 10.7

1

I

A

0.5

-

6

The Journal of Physical Chemistry, Vol. 83, No. 17, 1979 2199

Reaction of Oxygen Atoms with Hydrogen Sulfide

in the opposite direction (i.e,, make the observed value smaller than the true value of kl) might include a large amount of nonreactive impurity in the H2S or an extraneous emission in the chemiluminescence. However, as described in the Experimental Section, the former effect was insignificant, and the latter could be as much as 10%. Thus the average of the four mean values of kl at 297 K in Table I and its maximum uncertainty due to potential systematic errors considered above are (l.lO?$) X lo7 L mol-l s-l. Comparison with Literature Values of kl. The value of kl at 297 K determined in the present work is in close agreement with the value determined by Cupitt and Glassg in a computer simulation of reaction profiles of 0, H, and SO in discharge flow-ESR studies of reaction 1. Their work may have been affected by translationally hot hydrogen atoms formed by the reaction 0 + OH 02 + H and possibly by reaction 8.19 Because hot H atoms are inefficiently quenched by He (the carrier gas in their experiments), this could account for the high value obtained for k6 in their simulations, but should not have introduced an error into kl,since kl and k6 were independently optimized.lg All other values a t room temperature which have been reported previously are larger, as shown in Figure 2. With the exception of the values at 263 and 298 K, the results of Whytock et al.7are in good agreement with the present work. The measured values of kl in their flash photolysis-resonance fluorescence experiments were found to vary with the energy of the flash (ie., the initial concentration of oxygen atoms). Over a range of low flash energies, the observed value of kl appeared to remain constant, and these values were averaged at each temperature. This “plateau region” extended to higher energies of the flash a t higher temperatures, so that the results a t lower temperatures were obtained from a smaller range of flash energies. Slagle et aL3used a discharge flow technique with mass spectrometric detection to monitor H2S in an excess of oxygen atoms. Their results are -90% larger than the present results a t room temperature, and -55% larger near 500 K. Both Whytock et al.7 and Slagle et al.3 were aware of the discrepancy between their results at higher temperatures during the course of their work, but could not find any systematic errors. It is interesting to note that the results of Slagle et al. could be brought into reasonable agreement (within -20%) with the present results if a constant stoichiometric factor of 2 were used at all temperatures (although the results of Whytock et al. indicate that a constant stoichiometric factor may not be valid), The OH formed by reaction 1 in their experiments will react rapidly with oxygen atoms to form 0, and H, with the H possibly reacting on the time scale of their experiments with H2S to increase the observed rate of decay of H2S. However, if the differential equations for these reactions along with reactions 5 , 7, and 8 (and our value for kl)are integrated, using the initial concentrations of reactants in their experiments, it appears that secondary reactions would cause the observed value of kl to be only 15-70% too large in one experiment, and 10-22% in the other at 301 K. The uncertainty in the percentages is due to the slight curvature in the computed plots of In [H,S] against time for the initial 50 ms (approximately two half-lives). Also, the appropriate value of k6 in the simulations is uncertain because of the possible influence of hot H atoms. The results at 300 K of Hollinden et al.8are higher than the present value at 297 K, and their results at lower

-

temperatures are above the extrapolated Arrhenius line determined in this work. They used a discharge flow technique with ESR detection of oxygen atoms in excess H2S, and corrected the observed rate constants with an assumed stoichiometric factor of 3.5. Whytock et, aL7have pointed out that the stoichiometry is likely to vary with temperature, however. Mechanism of Reaction 1. The results of the experiments in which the products of reaction 1were analyzed can be rationalized by the mechanism in Table 111. Reactions l b and ICare indicated as proceeding by an addition mechanism, as proposed by Gutman and co-workers for organic sulfides and possibly H2S.lr3 The exothermicity of reaction l b has been discussed by several authors recently. Schurath, Weber, and BeckerZ0reported an upper limit for AHf(HSO)of 14.9 kcal rnol-l, and a “suggestive” lower limit of 13 kcal mol-l from spectroscopic data. The latter value would make reaction l b endothermic by 10 kcal mol-l. However, Slagle, Baiocchi, and Gutman3 have suggested that AHf(HSO) is -3 kcal mol-l, and RensonZ1 has suggested -5 kcal moll1, both of which would make reaction l b exothermic. White and Gardiner22recently calculated thermochemical properties of HSO, based on the data for AHf(HSO) reported by Schurath et al. They pointed out, however, that the species of composition HSO may be different in the experiments of Schurath (0, + SH HSO 0,) from that postulatedl~~ in the oxygen atom reaction (0+ H2S H + HSO), with the latter possibly having the theoretically calculatedz3more stable structure, SOH. The present work cannot supply information on the energetics of reaction lb, or on the structure of HSO, but an upper limit for klb/kl, based on the yield of H,, will be suggested later in the discussion. Reaction l a represents direct hydrogen abstraction, although the products in principle could result by fragmentation of the adduct postulated in reactions l b and IC. Direct abstraction is consistent with the fairly high activation energy for reaction 1. Reaction IC simply represents any unspecified reaction channels, and could include stabilization and isomerization of the postulated adduct. HartecklO has shown in studies with mixtures of H2 -+ H2S and D2Sthat the reaction channel 0 H,S SO is not significant, although an upper limit was not reported. In the steady state, the rate of formation of water relative to N2is RH20/RN2= kla/kl. The results in Figures 3 and 4 give an average value kla/kl = 0.52 f 0.05, where the uncertainty is the 95% confidence limits for four degrees of freedom with a Student’s t value of 2.8. The potential systematic error could be much larger. The background water determined in nonirradiated reaction mixtures amounted to about 30% of the total water detected in the irradiated mixtures. (In the experiment in Figure 3 with the lowest yield of N2,the background was 70% of the total, and was not used in determining the average value of kla/kl.) Also, water could be lost on the walls of the glass vacuum system, or by association with nonrecovered reaction products. The increase in RH2/RN2as the relative Concentration of H2Sincreases, shown in Figure 4, is apparently due to competition between N20 and H2S for excited mercury atoms, reactions 3 and 4. The steady state expression for RH21RN2is

-

+

-+

+

-

2288

The Journal of Physical Chemistry, Vol. 83, No. 17, 1979

where k7 = k7, + k7b. A value of 0.13 has been reportedz4 for k7b/k7,although this has been considered too large.25 The ratio k3/k4 was taken as 0.49,26and kl,/kl, as 0.52. The dotted line in Figure 4 was calculated with klb/kl = 0 and k7b/k7 = 0.13. The calculated line lies below the experimental points, and therefore if kYb/k7I 0.13, reaction 7b alone cannot account for the observed rate of hydrogen formation. With klb/kl = 0.11 and kTb/k7= 0, the dashed line in Figure 4 is obtained, which fits the experimental results reasonably wen. Since an extreme value of kTb/k7was used, the value used for klb/kl would be an upper limit. Inclusion of any HzS2formationz7in reaction 7 would not alter this conclusion. Thus the present work would suggest that klb/k? 5 0.11, depending on the actual value of k7b/k7, but reaction l b is clearly much less important than reaction la. However, the present experiments were done only at 700 torr. It would be interesting to determine if klb/kl varies with pressure. The present results cannot discriminate between Hz that could be formed by reactions l b and 6, and by elimination of molecular hydrogen, 0 + H2S H2 + SO. As mentioned above, the work of HartecklO indicated that the latter route was not significant, but the possibility that it might contribute to some extent to the rate of Hzformation observed in the present work cannot be excluded. Such a small contribution may not have been detectable in Harteck’s experiments with H2S and DzS. The small amount of Oz detected was not an impurity in the reactants, nor did it arise from a leak in the gas handling system, since neither N2 nor Oz was detected by gas chromatography in an experiment in which the usual procedures were followed, except that the reactants were not irradiated. The steady-state concentrations of 0 and OH were too low for the reactions 0 + OH 02 + H and 0+0 M Oz M to account for the rate of O2 formation. With the assumption of a 5-cm3 absorption volume (corresponding to a path length of only 0.25 cm for the 253.7-nm radiation) and a steady state, Ro,/RN,= 5 X lo4 for each of the reactions. Also, direct photolysis of N20 to yield O(lD) which could generate Oz by reaction with N 2 0 is unlikely, since the 9-54 filter blocks the 184.9-nm line, and the absorption coefficient of Nz0 at 253.7 nm is too small. A potential source of the O2 is the secondary reaction 0 + HSO O2 + SH, which is about 20 kcal mol-l exothermic (using Benson’s valuez1 of aH,(HSO)), However, the other reaction channels yielding SQ OH and SOz H are 44 and 73 kcal mol-’ exothermic, respectively, and may dominate. This is a purely speculative source of Oz, of course, since there is uncertainty regarding the formation of HSO in reaction 1as well as uncertainty in its chemistry and kinetics. The Oz is only a very small fraction of the total recovered products, and does not affect the conclusions regarding hl,/kl and klb/k1.

-

+

-

-

+

-

+

+

Singleton et al.

The absence of any detectable products containing sulfur, other than the material deposited on the window, is not inconsistent with the mechanism in Table 111. The SH formed in reaction l a is converted to HzS, S, and possibly some Hzand S2 in reaction 7. The consumption of H2S was not determined in the experiments, and elemental sulfur would not be recovered by the techniques used in this work. The uncertainty in the nature of the sulfur containing products in reactions 1b and ICprecludes speculation as to the possibility of their recovery. Acknowledgment. The authors are grateful to M. E. Bednas for the analysis of impurities in H2S, and to R. Ironside for the analysis of NO2in NzO. The authors also thank Dr. G. P. Glass for providing data prior to publication, and Dr. D. Gutman for supplying experimental data for the numerical integrations. References and Notes I. R. Slagle, R. E. Graham, and D. Gutman, Int. J . Chem. Kinet., 8, 451 (1976). R. J. Cvetanovie, Adv. Photochem., 1, 115 (1963). I. R. Slagle, F. Baiocchi, and D. Gutman, J. Phys. Chem., 82, 1333 (1978). R. R. Smardzewski and M. C. Lin, J . Chem. Phys., 66, 3197 (1977). R. R. Smardzewski, J . Chem. Phys., 68, 2878 (1978). D. E. Tevault and R. R. Smardzewski, J , Chem. Phys., 69, 3182 (1978). D. A. Whytock, R. B. Timmons, J. H. Lee, J. V. Michael, W. A. Payne, and L. J. Stlef, J . Chem. Phys., 65, 2052 (1976). G. A. Hollinden, M. J. Kurylo, and R. 8. Timmons, J . Phys. Chem., 74, 988 (1970). (a) L. T. Cupitt and G. P. Glass, Trans. Faraday Soc., 66, 3007 (1970); (b) Int. J. Chem. Kinet., S1, 39 (1975). G. Liuti, S. Dondes, and P. Harteck, J. Am. Chem. Soc., 88, 3212 (1966). S. Takahashi, Mem. Def. Acad., Math., Phys., Chem. Eng., Yokosuka, Jpn., 10, 369 (1970). R. Atkinson and R. J. CvetanoviE, J. Chem. Phys., 55, 659 (1971). S. Furuyama, R. Atkinson, A. J. Colussi, and R. J. CvetanoviE, Int. J . Chem. Kinet., 6, 741 (1974). D. L. Singleton, S. Furuyama, R. J. Cvetanovi6, and R. S. Irwin, J . Chem. Phys., 63, 1003 (1975). F. Stuhl and H. Niki, J . Chem. Phys., 55, 3943 (1971). F. Kaufman, Annu. Rev. Phys. Chem., 20, 45 (1969). W. S. Nip, R. J. CvetanoviE, R. S. Irwin, and D. L. Singleton, unpublished data. R. F. Hampson, Jr., and D. Garvin, Ed., Natl. Bur. Stand., Spec. Pub., No. 513 (1977). G. P. Glass, personal communication. U. Schurath, M. Weber, and K. H. Becker, J. Chem. Phys., 67, 110 (1977). S. W. Benson, Chem. Rev., 78, 23 (1978). J. N. White and W. C. Gardiner, Jr., Chem. Phys. Lett., 58, 470 (1978). A. B. Sannigrahi, S. D. Peyerimhoff, and R. J. Buenker, Chem. Phys., 20, 381 (1977). B. de B. Darwent and R. Roberts, Proc. R . SOC.London, Ser. A , 216, 344 (1953). K. Schofield, J. Phys. Chem. Ref. Data, 2, 25 (1973). R. J. CvetanoviC, Prog. React. Kinet., 2, 39 (1964). D. L. Baulch, D. D. Drysdale, J. Duxbury, and S. Grant, “Evaluated Kinetic Data for High Temperature Reactions”, Vol. 3, Butterworths, London, 1976, pp 485-495. R. A. Perry, R. Atkinson, and J. N. Pitts, Jr., J . Chem. Phys., 64, 3237 (1976).