Reaction of OH with Dimethyl Sulfide (DMS). 1. Equilibrium Constant for OH + DMS Reaction and the Kinetics of the OH*DMS + O2 Reaction

Article abstract of DOI:10.1021/jp960866k

The formation of a weakly bound adduct in the reaction of OH with DMS-d6 was observed between 217 and 240 K using the technique of pulsed laser photolysis/pulsed laser-induced fluorescence.The equilibrium constant for this process, OH + DMS-d6 <--> OH*DMS-d6, was measured as a function of temperature.The bond strength of this adduct was determined to be 10.7 +/- 2.5 kcal mol^-1.The weakly bound adduct was observed to react rapidly with O2.The rate constant for the reaction OH*DMS-d6 + O2 -> products was determined to be (1.00 +/- 0.33)E-12 cm³ molecules^-1 s^-1, independent of pressure and temperature.The atmospheric implications of the formation of this adduct and its reaction with O2 to the mechanism of DMS oxidation in the atmosphere are discussed.

Full text of DOI:10.1021/jp960866k

1
4694
J. Phys. Chem. 1996, 100, 14694-14702
Reaction of OH with Dimethyl Sulfide (DMS). 1. Equilibrium Constant for OH + DMS
Reaction and the Kinetics of the OH‚DMS + O2 Reaction
Stephen B. Barone, Andrew A. Turnipseed, and A. R. Ravishankara*
National Oceanic and Atmospheric Administration, EnVironmental Research Laboratories, Aeronomy
Laboratory, Boulder, Colorado 80303, and Department of Chemistry and Biochemistry and The CooperatiVe
Institute for EnVironmental Sciences, UniVersity of Colorado, Boulder, Colorado 80309
X
ReceiVed: March 22, 1996; In Final Form: June 18, 1996
The formation of a weakly bound adduct in the reaction of OH with DMS-d
40 K using the technique of pulsed laser photolysis/pulsed laser-induced fluorescence. The equilibrium
constant for this process, OH + DMS-d T OH‚DMS-d , was measured as a function of temperature. The
bond strength of this adduct was determined to be 10.7 ( 2.5 kcal mol . The weakly bound adduct was
observed to react rapidly with O . The rate constant for the reaction OH‚DMS-d + O f products was
6
was observed between 217 and
2
6
6
-1
2
6
2
-12
3
-1 -1
determined to be (1.00 ( 0.33) × 10 cm molecules s , independent of pressure and temperature. The
atmospheric implications of the formation of this adduct and its reaction with O
oxidation in the atmosphere are discussed.
2
to the mechanism of DMS
Introduction
An electronic ab-initio structure calculation of McKee9
yielded a bound geometry for the OH‚DMS adduct with a bond
Measurements in the atmosphere suggest that natural sources
of organosulfur compounds comprise ∼60% of the total sulfur
flux into the marine boundary layer in the southern hemisphere
-1
strength of 6.0 kcal mol . Contrary to these results, theoretical
1
0,11
calculations and experiments by Gu and Turecek
suggested
that OH‚DMS was not a stable intermediate but a transition
state in reaction 1.
1
and ∼15% in the northern hemisphere. The primary organo-
sulfur compound emitted to the atmosphere is dimethyl sulfide
CH3SCH3, DMS).1 The mechanism of the atmospheric oxida-
The present work was initiated to investigate the existence
of the OH‚DMS species and quantify the rate coefficients
involved in its formation, decomposition, and reaction with O2.
(
tion of DMS is very complex and the branching to final stable
end products will depend on the temperature, pressure, and
composition of the atmosphere. Quantification of the yields of
stable end products is vital in evaluating the role of DMS in
climate regulation and in understanding the biogeochemical
cycling of sulfur.
1
2
The accompanying paper describes our attempts to identify
the products and measure their branching ratios in the OH
reaction with DMS. Temperature and O2 concentrations were
varied to obtain data relevant to modeling DMS oxidation in
the atmosphere. In combination, the two papers present a
description of the first several steps in the OH-initiated oxidation
of DMS. In this paper we compare our results on the OH +
DMS reaction kinetics with those from previous studies and
briefly discuss their importance in the atmospheric oxidation
of DMS. A more detailed discussion of the atmospheric
consequences of these findings will be presented in the
companion paper.
2
The atmospheric oxidation of DMS has been the subject of
many investigations in the past few decades and has been
reviewed recently.3
-5
The reaction of OH with DMS is
currently believed to be the most important process that initiates
DMS oxidation in the daytime. Hynes et al. observed that the
loss rate of OH in the presence of DMS increased significantly
upon the addition of O2. They proposed the following mech-
anism for reaction 1 to account for their observations:
6
OH + DMS T OH‚DMS
(1f, 1r)
(1a)
Experimental Section
OH + DMS f H O + CH SCH
2
All experiments utilized pulsed laser photolysis for generating
OH and pulsed laser-induced fluorescence (LIF) for detecting
it. The apparatus used is identical to that previously employed
2
3
The symbol “T” indicates the combination of forward and
reverse reactions. The enhancement in the OH loss rate in the
presence of O2 was explained by the occurrence of a reaction
between the OH‚DMS adduct and O2:
13
14
to detect CH3S and CF3O and was modified for the detection
of OH. The detection of OH via LIF is routine in our laboratory.
Therefore, we relate here only the aspects essential to the
understanding of the present study.
OH‚DMS + O f products
(2)
3
2
The reaction vessel was a ∼200-cm jacketed Pyrex cell with
six ports: four were 5-cm long with a 1-in. diameter for
transmitting of the probe and photolysis laser beams, one was
an inlet for inserting a movable thermocouple, and one was for
collecting the laser-induced fluorescence. A series of light
baffles were used to minimize scattered probe laser radiation
in the cell. The reactor temperature was regulated over the range
In the absence of O2, the adduct would decompose rapidly to
re-form OH and, hence, channels 1f and 1r would not be
observable at long reaction times (i.e., after equilibration).
7
8
Chamber studies, by Barnes et al. and Wallington et al. , have
confirmed the pronounced dependence of the rate coefficient
for removal of DMS on the partial pressure of O2.
2
17-359 K by flowing either cooled methanol or heated
ethylene glycol from a temperature-regulated bath through the
jacket of the cell. The temperature of the gases within the
reaction volume (defined by the intersection of the photolysis
*
Address correspondence to this author at NOAA/ERL, R/E/AL2, 325
Broadway, Boulder, CO 80303.
X
Abstract published in AdVance ACS Abstracts, August 1, 1996.
S0022-3654(96)00866-0 CCC: $12.00 © 1996 American Chemical Society

Reaction of OH with DMS. 1
J. Phys. Chem., Vol. 100, No. 35, 1996 14695
and probe beams) was measured by inserting a calibrated
chromel-alumel thermocouple through a movable injector under
gas flow conditions identical to those during the experiment.
For experiments using He as the bath gas, we conservatively
estimate the uncertainty in the reactor temperature to be (0.5
K because there was no measurable temperature gradient within
the reaction volume. However, a measurable temperature
gradient was observed when SF6 and N2 were used as bath gases.
We conservatively estimate the reaction temperatures to be
accurate to ∼(2 K under these conditions.
absorption at 213.9 nm (zinc line) in a 100-cm long absorption
-18
2
-1
cell [σ213.9(DMS-d6) ) 1.23 × 10 cm molecule , measured
-
18
2
-1
in this study, and σ213.9(DMS) ) 1.70 × 10
cm molecule
1
7
from Hearn et al. ]. Although not essential for the kinetic
measurements, knowledge of the photolyte concentration is
useful in evaluating the potential role of secondary reactions.
H2O2 concentration was monitored on line by absorption at
213.9 nm prior to entering the cell [σ213.9nm(H2O2) ) 3.16 ×
-19
2
-1 18
10
cm molecule ].
HONO concentrations were not
monitored on line directly but were inferred from the measured
rate coefficient for OH radical loss due to the OH + HONO
OH radicals were generated by pulsed excimer laser pho-
tolysis of either H2O2 at 248 nm (KrF, 5-30 mJ pulse- cm ,
1
-2
reaction, in the absence of DMS-d or DMS.
6
Because it enables us to easily separate equilibration of OH
with DMS from the irreversible loss of OH, we used DMS-d6,
pulse width ∼20 ns, 10 Hz) or HONO at 351 nm (XeF, 10-30
mJ pulse- cm , pulse width ∼20 ns, 10 Hz) on an axis
perpendicular to the direction of the gas flow. The photolysis
beam was passed through a 5-cm side arm and reduced to a
diameter of 0.80 cm by an aperture before traversing through
the cell and exiting via another 5-cm side arm. The fluence of
the photolysis beam was measured by a disk calorimeter after
the beam exited the cell. H2O2 was introduced into the cell by
passing a small flow of helium through a glass bubbler
containing H2O2. A 90% H2O2 solution was purified further
by bubbling He through it for several days prior to use. HONO
was prepared, on-line, in a flask by adding 0.1 M NaNO2
solution dropwise into a solution of 10% H2SO4 and swept out
of the flask directly into the reactor by a small flow of helium.
1
-2
rather than DMS, as the reactant in most of our experiments:
OH + DMS-d f products
(3)
6
In the present experiments, OH loss is governed by both its
reversible addition to DMS and its irreversible reaction with
DMS. If the time scales for these two processes are different,
their rate coefficients can be unambiguously extracted from the
measured OH temporal profiles. The rate coefficient for the
irreversible reaction of OH with DMS is approximately a factor
6
of 3 faster than the that of analogous reaction with DMS-d6.
Therefore, we used DMS-d6 in place of DMS. Use of DMS-d6
has an added advantage in that it eliminates any possible
production of OH radicals from secondary reactions involving
the methyl groups in DMS.
OH free radical concentrations were monitored by exciting
2
+
2
the (A Σ , V′ ) 1) r (X Π, V′′ ) 0) electronic transition at
∼
-
1
-2
281.9 nm (0.10-1.2 mJ pulse cm , pulse width ∼8 ns, 10
Hz) from a dye laser pumped by a pulsed XeCl excimer laser.
The excitation beam was orthogonal to the photolysis beam.
The fluence of this pulse was measured after the beam traversed
through the reactor using a disc calorimeter to monitor its
Results
OH + DMS T OH‚DMS (k1f,r). The reaction of OH with
DMS can proceed via several channels:
2
+
stability. The red-shifted fluorescence, resulting from the (A Σ ,
OH + DMS (+ M) T
2
2
+
2
V′ ) 1) f (X Π, V′′ ) 1) and (A Σ , V′ ) 0) f (X Π, V′′ ) 0)
transitions, was collected on an axis orthogonal to both the
photolysis and probe laser beams by a 5-cm focal length UV
quartz lens. The fluorescence was passed through an aperture
and band-pass filter (λ ) 308.7 nm, with a FWHM of 5.9 nm)
before being imaged onto a slit in front of the photomultiplier
tube (PMT). The output pulse from the PMT was fed into a
gated charge integrator and then to a microcomputer, via an
A/D converter, for data analysis.
CH S(OH)CH (+ M) ∆°H ) (determined here) (1f, 1r)
3
3
OH + DMS f
-
1
CH SCH + H O ∆°H ) -25.5 kcal mol (1a)
3
2
2
-
1
OH + DMS f CH + CH SOH ∆°H ) 0 ( 3 kcal mol
3
3
(1b)
The temporal profiles of the OH radical concentrations were
acquired by measuring the LIF signal at various delay times
between the photolysis and probe lasers (5 µs-50 ms). The
nascent rotational and vibrational energy distributions in the OH
radical produced by photolysis of H2O2 and HONO are
OH + DMS f
CH S + CH OH ∆°H ) -18.8 kcal mol (1c)
-1
3
3
The heats of formation of the reactants and products of reactions
1
9,18
1
a-1c are taken from the literature.
Reaction 3 (the
1
5,16
nonthermal.
The use of an excess of buffer gas rapidly
reaction of OH with the deuterated analog of DMS) has the
same possible product channels as reaction 1 with approximately
the same reaction enthalpies.
thermalized these rotational and vibrational energy distributions
to the ambient temperature (before our earliest observation times,
t < 5 µs). The detection limit for OH, defined as S/N ) 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-
Initially, the overall rate coefficients for reaction 1 and its
deuterated analog (reaction 3) were measured at 298 K in the
absence of O2 under pseudo-first-order conditions in [OH] (i.e.,
8
-3
ground, was 4 × 10 molecules cm in 100 Torr of He upon
[DMS] g 100[OH]). The postphotolysis temporal profiles of
integration of 100 laser shots.
OH at 298 K strictly obeyed the equation
The photolyte, bath gas, and reactants were slowly flowed
-
1
(
linear flow velocities of 2-10 cm s ) through the reactor
perpendicular to the direction of the photolysis beam. He
Bureau of Mines, g99.9997%), N2 (Scott Gases, g99.9995%),
[OH] ) [OH] exp(-k′ t)
(4)
t
0
i
(
In this expression [OH]t and [OH]0 are, respectively, the
concentration of OH radicals at times t and 0 (i.e., after it is
generated but before any reactive loss occurred). k′i represents
the pseudo-first-order decay rate constant for OH loss and is
related to the second-order rate coefficient (ki) by the expression
O2 (Scott Gases, g99.99%), and SF6 (Scott Gases, g99.9%)
were added to the reactor through calibrated electronic mass
flow meters. O2 concentrations were determined from the
measured mass flow rates and the cell pressure, measured by a
capacitance manometer. Concentrations of DMS-d6 (CD3SCD3,
k′ ) k [DMS] + k
d
(5)
>
99.9%) and DMS (99.5%) were monitored on line by UV
i
i

1
4696 J. Phys. Chem., Vol. 100, No. 35, 1996
Barone et al.
profiles in the presence of DMS-d6 were not strictly single
exponential; they were characterized by an initial rapid decay
followed by a slower exponential decay. The rate of the initial
decay and the point of inflection were observed to vary with
the concentration of DMS-d6. This behavior is indicative of
the occurrence of an equilibrium between OH and DMS-d6 in
combination with a slower first-order loss process for OH. The
observed temporal profiles are consistent with the following
series of reactions:
mechanism 1
OH + DMS-d T OH‚DMS-d6
(3f, 3r)
6
OH + DMS-d f products other than OH‚DMS-d6 (3bi)
6
OH f loss
(6)
(7)
OH‚DMS-d f loss
6
3
bi ) 3a + 3b + 3c. We fit the obtained OH temporal profiles
to mechanism 1 using a nonlinear least-squares fitting routine,
FACSIMILE. The fits were constrained by using the mea-
sured first-order rate coefficient for the loss of OH in the absence
of DMS, k6, and assuming that k7 was <100 s . Parameters
k3f, k3r, and k3bi were variables in the fit. The equilibrium
constant, Kc, was determined from the ratio of the calculated
values of k3f and k3r. Table 1 summarizes the values of Kc
obtained in this study.
2
0
Figure 1. Examples of temporal profiles of [OH] measured at 230 K
in 100 Torr (25 Torr of N and 75 Torr of He): (A) [DMS-d ] ) 1.70
-1
2
6
1
5
-3
15
-3
×
10 molecule cm ; (B) [DMS-d
The initial concentrations were the same for both profiles. The lines
are the nonlinear least-squares fits of the data to mechanism 1.
6
] ) 4.38 × 10 molecule cm
.
We measured equilibrium constants only between 217 and
2
45 K. At temperatures >245 K, equilibration was completed
before 5 µs. We did not utilize reaction times <5 µs because
thermalization of photolytically produced OH cannot be assured.
At temperatures <217 K the concentrations of photolyte required
to generate a measurable amount of OH exceeded their vapor
pressure.
The measured values of Kc were independent of the photolysis
-
2
-1
and probe laser fluences (10-30 mJ cm , 0.1-1 mJ pulse ,
respectively) and identity (N2, SF6, and He) of bath gas and its
pressure (30-100 Torr). Kc was also independent of the identity
1
3
and concentration of the photolyte ([H2O2] ) (1-40) × 10
-3
13
-3
molecules cm ], [HONO] ) (2-20) × 10 molecules cm ).
The obtained equilibrium constants are displayed as a plot of
∆n
Kp, on a log scale, versus 1/T in Figure 2 [Kp ) Kc(RT) , where
∆
n ) -1]. The slope and intercept of this plot (second-law
analysis) correspond to the ∆°H/R and ∆°S/R for reaction 3f.
Figure 2. Plot of K
reaction 3f,r in different bath gases [He (O), SF
photolysis at 248 nm was used in most cases; experiments with
HONO photolysis are indicated by ]. The error bars are the 2σ
precision obtained from the nonlinear least-squares fits (see text) of
the data. The solid line is the weighted linear least squares fit to the
p
(on a log scale) vs 1000/T (van’t Hoff plot) for
Fitting the entire data set with a linear least-squares fitting
6
(4), and N (0)].
2
-1
H
2
O
2
program yields ∆°H3f ) -10.2 ( 2.0 kcal mol and ∆°S3f )
-28.4 ( 6.4 cal mol-
1
K
-1
at 229.5 K, the mean value of the
temperatures that were used.
An alternative means of obtaining ∆°H3f is the third-law
method in which the equilibrium constant at a particular
temperature is used in conjunction with the calculated entropy
change of the reaction at that temperature to obtain ∆°Hrxn.
equation ln K
p
) -∆°H
r
r
/RT + ∆°S /R (second-law analysis). The
dashed line is the weighted least-squares fit of the data using ∆°S
calculated from statistical mechanics (third-law analysis).
r
Using ∆°S3f () -31.1 cal mol-
1
K
-1
at 229.5 K) and ∆Cp
where kd is the first-order OH loss rate coefficient in the absence
of DMS (or DMS-d6). By measuring k′i at various DMS (or
DMS-d6) concentrations, the rate coefficients for reactions 1
and 3 at 298 K were measured to be k1 ) (4.95 ( 0.35) ×
calculated from statistical mechanics yields a ∆°H3f(298 K) )
-1
-10.7 ( 2.5 kcal mol . The vibrational frequencies and
rotational constants used in calculating ∆°S3f and ∆Cp are given
in Table 2. The agreement between the calculated value of
∆°H3f and that obtained from the second-law analysis indicates
a lack of any major systematic errors. The error quoted is twice
the standard deviation of the mean with estimates of systematic
uncertainty in the [DMS-d6]. Due to the limited temperature
range of our Kc measurements, which makes the second-law
analysis (taking the slope) less reliable, the third-law analysis
of our data is preferred. Our analysis leads to a ∆°Hf(DMS-
d6‚OH) of -10.3 ( 2.5 kcal mol- at 298 K.
-
12
3
-1 -1
-12
3
1
0
cm molecule
s
and k3 ) (1.75 ( 0.25) × 10 cm
-
1
-1
molecule
s
at 298 K. The reported uncertainties are the
2σ errors in fitting the measured k′i vs [DMS] data to eq 5 using
a weighted linear least-squares routine. The weighting was
according to the uncertainties in the measured values of k′i. The
obtained values are in good agreement with other recent
3
,6
measurements.
Temporal profiles of [OH] were measured at various tem-
peratures. As shown in Figure 1, at T e 240 K, OH temporal
1

Reaction of OH with DMS. 1
J. Phys. Chem., Vol. 100, No. 35, 1996 14697
TABLE 1: Summary of Equilibrium Constants Measured in the Present Study along with Experimental Conditions Used To
Acquire the Data
[
DMS-d
6
],
k
3bi
,
k
3f
,
k
3r
,
c
K ,
T, K buffer gas P, Torr 1015 molecules cm
-3
10-12 cm molecule s-1 10-12 cm molecule s-1
3
-1
3
-1
10 s
4
-1
10-17 cm molecule
3
-1
217
222
224
225
225
227
228
228
229
229
229
229
230
230
233
234
235
236
240
241
He
He
He
100
100
55
20/80
30/70
55
100
30
100
100
100
100
50
25/75
50
100
30/70
55
100
50
1.3-8.2
1.3-5.8
2.4 ( 0.8
1.6 ( 0.4
1.3 ( 0.1
1.7 ( 0.1
1.6 ( 0.1
1.4 ( 0.1
1.8 ( 0.3
1.6 ( 0.2
1.4 ( 0.1
1.8 ( 0.5
1.8 ( 0.3
1.5 ( 0.1
1.4 ( 0.1
1.5 ( 0.2
1.4 ( 0.1
1.4 ( 0.3
1.3 ( 0.2
1.4 ( 0.1
1.4 ( 0.1
1.4 ( 0.1
5.49 ( 2.46
4.96 ( 3.32
2.56 ( 0.35
5.04 ( 2.00
6.10 ( 2.20
2.60 ( 0.27
10.6 ( 4.6
3.70 ( 0.84
3.57 ( 1.40
4.32 ( 2.24
4.81 ( 1.02
3.79 ( 0.47
2.34 ( 0.28
7.10 ( 3.30
2.68 ( 0.14
9.70 ( 4.64
4.65 ( 3.90
2.46 ( 0.22
2.26 ( 0.70
1.83 ( 0.23
2.04 ( 0.46
2.71 ( 0.94
1.68 ( 0.21
3.62 ( 0.95
1.57 ( 0.13
2.16 ( 0.17
6. 9 ( 1.3
2.29 ( 0.50
4.02 ( 0.80
4.5 ( 1.8
2.04 ( 0.46
4.00 ( 0.41
2.48 ( 0.34
6.3 ( 1.3
3.44 ( 0.23
13.0 ( 3.1
9.0 ( 3.4
4.51 ( 0.33
6.0 ( 1.1
5.21 ( 0.58
26.7 ( 3.5
19.2 ( 3.7
15.3 ( 1.4
13.9 ( 3.2
14.1 ( 2.6
12.6 ( 2.1
15.2 ( 3.7
16.2 ( 1.1
8.8 ( 1.8
9.9 ( 3.9
10.1 ( 1.6
9.2 ( 0.8
9.4 ( 0.4
11.1 ( 3.1
7.8 ( 0.3
7.0 ( 3.4
5.0 ( 3.0
5.5 ( 0.2
3.8 ( 1.1
3.5 ( 0.2
1.2-8.7
SF
/He
He
6
/He
1.2-12.2
1.2-11.8
3.6-11.2
3.4-10.0
1.4-8.4
N
2
N
N
2
2
He
He
He
He
He
2.6-11.3
1.4-6.4
1.9-6.4
2.4-8.9
3.2-14.0
12.0-9.3
3.9-18.3
4.1-10.40
5.3-24.4
6.4-18.5
6.4-18.9
6.7-20.9
2
N /He
He
N
N
2
2
/He
He
He
He
a
2
2
30
34
N
N
2
2
100
100
2.3-5.7
4.4 ( 0.6
4.6 ( 0.6
9.1 ( 7.7
7.5 ( 5.6
6.0 ( 3.2
6.8 ( 1.7
14.0 ( 6.0
9.5 ( 2.3
a
1.3-6.8
a
DMS], 1015 molecules cm-3
[
.
TABLE 2: Vibrational Frequencies (cm-1) and Rotational
Constants (GHz) of DMS-d
6
6
and OH‚DMS-d (from McKee,
Personal Communication) Used To Calculate the Entropy
and Heat Capacity Changes in Reaction 3
OH‚DMS-d
6
DMS-d
6
OH‚DMS-d
6
DMS-d
6
vibr freq
vibr freq
15
16
17
18
19
20
21
22
23
24
25
26
1
2
3
4
5
6
7
8
9
3705
2353
2352
2340
2339
2194
2193
1086
1079
1077
1070
1060
1042
853
2330
2330
2319
2314
2184
2182
1093
1089
1082
1075
1068
1049
850
763
730
712
696
655
636
254
235
228
166
142
128
88
726
701
696
634
227
133
126
1
1
1
1
1
0
1
2
3
4
27
767
rotn const
A
B
C
5.64
4.76
3.43
17.56
7.39
5.57
Figure 3. Plot of k3bi (k3bi ≡ k3a + k3b + k3c) (on a log scale) vs 1/T.
OH was generated by 248 nm photolysis of H . The line is the linear
2 2
O
least-squares fit of all the data for T > 250. The obtained values of A
The long time (t > 50 µs) loss of OH in the presence of an
excess of DMS-d6 yields information on the sum of the rate
coefficients for reactions 3a, 3b, and 3c (the bimolecular channel,
k3bi). Values of k3bi as a function of temperature were obtained
by fitting the acquired OH temporal profiles to mechanism 1.
Figure 3 shows the values of k3bi obtained in experiments when
H2O2 was used as the OH precursor as a function of temperature.
Also shown in the figure are the values of kbi at higher
temperatures (T > 260 K), where the bimolecular rate coefficient
was directly measured (i.e., under conditions where the adduct
formation was negligible). As shown in the figure, a slight
and E /R are given in the text. k values for T < 250 K deviate from
a
3bi
the above fit by ∼20%. Note the highly expanded scale for the rate
coefficient in this figure.
reactants or photolytes. Possible explanations of these observa-
tions are given under Discussion. We prefer to quote the values
obtained at T > 260 K.
The 248-nm photolysis of HNO3 is another commonly used
OH radical source. By photolyzing a mixture of HNO3 [(1-
1
4
-3
15
12) × 10 molecules cm ], DMS-d6 [(1-9) × 10 molecules
-3
cm ], and He (100 Torr) at 298 K, we determined the rate
-12
3
coefficient for reaction 3bi to be (1.72 ( 0.6) × 10
cm
1 -1
(
∼20%) deviation from the Arrhenius behavior was observed
molecule- s . This value is in excellent agreement with our
measurements of k3 using H2O2 and HONO at 298 K. Attempts
were made to observe the addition of OH to DMS-d6 using
HNO3 photolysis to generate OH radicals. However, the
photolysis of a mixture of HNO3 and DMS-d6 at 229 K resulted
in very rapid single-exponential decay profiles of OH with time
for temperatures much lower than 260 K. When these points
-12
are not included, our data yield k3bi ) (6.5 ( 1.1) × 10 exp-
-(380 ( 60)/T] cm molecule s . The error bars reflect
σ of the standard deviation of the fit to the Arrhenius
3
-1 -1
[
2
expression. The values of k3bi measured using HONO as the
OH precursor were slightly higher (∼30%). The deviations in
the values of k3bi obtained at T < 260 K are likely due to
reactions involving the photolytes and/or impurities in the
4
-1
constants of (1.5-7.0) × 10 s . In addition, the OH loss rate
constant was observed to increase with increasing HNO3
concentrations. In the absence of DMS-d6, the OH temporal

1
4698 J. Phys. Chem., Vol. 100, No. 35, 1996
Barone et al.
to secondary reactions involving the OH + DMS reaction
products) was unrelated to the biexponential portion of the OH
temporal profiles, we subtracted the small long time regeneration
and fit the residual profiles to mechanism 1 to determine values
of k1f, k1r and k1bi. The obtained equilibrium constants are
summarized in Table 1 and are, within the combined uncertain-
ties of our measurements, the same as those for the DMS-d6
measurements.
O2 + OH‚DMS-d6 f Products (k8). Reaction 8 was initially
studied at T < 235 K, at which the equilibration between OH
and DMS-d6 was observable and a large fraction of the initial
OH could be tied up as OH‚DMS-d6. Under these conditions,
the long time (t > 50 µs) loss rate coefficients of OH in the
presence of an excess of DMS-d6 increased upon the addition
of O2. The OH temporal profiles showed an initial rapid decay
due to equilibration followed by a second decay, the rate of
which increased with increasing concentrations of O2. This
behavior is indicative of
a
reaction between the
OH‚DMS-d6 adduct and O2, which leads to an irreversible loss
of the adduct and, hence, OH:
OH‚DMS-d + O f products
(8)
6
2
For the [OH] temporal profile to be sensitive to the occurrence
of reaction 8, a significant fraction of OH must be tied up in
the OH‚DMS-d6 adduct. This requirement limited measure-
ments of meaningful values of k8 to 220-235 K.
Figure 4. Plot of k′
K and [DMS-d
] ) 5.4 × 10 molecules cm in approximately 30
Torr of N . The line is the weighted linear least-squares fit. The error
bars are the 95% confidence limits derived from the FACSIMILE
simulations (see text).
9
vs [O
2
] for the O
2
6
+ OH‚DMS-d reaction at 228
1
5
-3
6
2
Temporal profiles of [OH], in an excess of DMS-d6, were
monitored while the total O2 pressure was varied [(0.2-2.0) ×
TABLE 3: Rate Coefficients for Reactions 8 and 2 Obtained
in the Present Study along with the Pertinent Experimental
Conditions
1
7
-3
1
0
molecules cm ]. The observed [OH] temporal profiles
could be accounted for by the following series of reactions:
temp, K
bath gas pressure, Torr k, 10- cm molecule s-1
13
3
-1
mechanism 2
DMS-d
6
222
228
228
229
230
230
233
233
234
234
247
258
258
He
100
100
100
100
30
100
75
8.9 ( 1.4
8.5 ( 2.7
9.1 ( 0.6
7.8 ( 3.6
8.6 ( 1.6
8.8 ( 0.9
11.8 ( 3.0
11.5 ( 2.6
10.5 ( 2.2
8.4 (2.4
OH + DMS-d T OH‚DMS-d6
(3f, 3r)
6
N
N
2
2
OH + DMS-d f products other than OH‚DMS-d6 (3bi)
6
He
N
2
OH f loss
OH‚DMS-d f loss
(6)
(9)
He
N
N
2
2
80
6
He
He
100
100
100
100
200
Mechanism 2 is nearly identical to mechanism 1, and here
reaction 9 includes contributions from reactions of OH‚DMS-
d6 with O2. Reaction 9 represents the loss of OH‚DMS-d6
expressed as a first-order process. The first-order rate coefficient
for this reaction in an excess of O2 (over OH‚DMS-d6) can be
expressed as k′9 ) k8[O2] + k9dif, where k9dif is the first-order
rate constant for loss of OH‚DMS-d6 in the absence of O2.
Reaction 6 represents the OH loss in the absence of DMS-d6
and O2 and is mostly due to reaction with the photolyte and, to
a lesser degree, diffusion out of the reaction zone.
N
2
N
2
N
2
10.9 ( 0.7
10.1 ( 0.5
9.9 ( 0.3
DMS
2
2
47
58
N
N
2
100
100
10.9 ( 0.5
13.7 ( 2.7
2
profiles were exponential and the OH loss rate constant was
consistent with the known bimolecular rate coefficient for the
reaction of OH with HNO3 at 229 K. Our observations indicate
that either HNO3 itself or an impurity (such as NO2) in it reacts
efficiently with the OH‚DMS-d6 adduct. Due to the existence
of this interfering reaction(s), we were unable to determine the
equilibrium constant using HNO3 as the OH source.
We also attempted to observe equilibration between OH and
DMS in a manner similar to that for OH with DMS-d6. The
approach to equilibrium was more difficult to separate from
the longer time loss, which was controlled by the bimolecular
component of reaction 1. However, at high pressures (P g 100
Torr) and low temperatures (T e 230), the equilibration between
OH and DMS (nondeuterated) was indeed observed. Under
these conditions, the OH temporal profiles were characterized
by biexponential decays superimposed upon a small regeneration
of OH at long reaction times. We were unable to eliminate the
slight OH regeneration in the OH + DMS system. Assuming
that the regeneration of OH at long times (which is likely due
First, [OH] temporal profiles were measured in an excess of
DMS-d6 in the absence of O2. Values for k3f, k3r, and k3bi were
obtained by fitting these profiles to mechanism 1 using a
2
0
nonlinear least-squares routine (FACSIMILE). Next, [OH]
temporal profiles were monitored in the presence of an excess
of DMS-d6 and O2 while the total pressure was kept the same.
By using prescribed values of k3bi and k6, the OH temporal
profiles were fit to mechanism 2 (using FACSIMILE) to yield
values of k3f, k3r, and k′9. The rate coefficient of reaction 6
could be measured directly by monitoring the OH loss rate
constant subsequent to its generation via photolysis of H2O2 in
the bath gas alone. k8 was calculated from the slope of the
plot of k′9 vs [O2]. Figure 4 shows a typical plot of k′9 vs [O2]
obtained at 230 K in 100 Torr of He. The obtained values of
k3f, k3r, and Kc were in excellent agreement ((5%) with those
measured in the absence of O2. Conversely, fixing k3f and k3r

Reaction of OH with DMS. 1
J. Phys. Chem., Vol. 100, No. 35, 1996 14699
to the values obtained in the absence of O2 resulted in a <5%
change in the calculated values of k′9. k8 was independent of
-2
the photolysis and probe laser fluences (10-30 mJ cm and
-1
0
.1-1.0 mJ pulse ) and bath gas identity (SF6, N2, and He) or
pressure (30 -100 Torr). Table 3 summarizes the results of
the measured k8 values.
To extend the limited temperature range over which k8 was
measured, a second approach was taken. At higher reactor
temperatures and in the presence of a large concentration of
O2, the OH‚DMS-d6 adduct can be assumed to be in steady
state. The following mechanism (which is identical to mech-
anism 1 with the exception that reaction 9 is represented by
two separate loss processes, reactions 8 and 10) adequately
describes the temporal profiles of OH:
mechanism 3
OH + DMS-d T OH‚DMS-d
(3f, 3r)
6
6
OH + DMS-d f products other than OH‚DMS-d6 (3bi)
6
OH‚DMS-d + O f products
(8)
(6)
6
2
OH f loss
'
Figure 5. Plots of kobs [kobs ≡ (kobs - k
6 6 2
)/[DMS-d or DMS]] vs [O ]
OH‚DMS-d f
for experiments at 247 K and 100 Torr (N + O ) for DMS and DMS-
d . The error bars shown are the 95% confidence limits for each kobs.
6
6
2
2
loss other than to produce OH or via reaction 8 (10)
A significant difference in the intercept and asymptote (A * B) is
observed for two cases and interpreted to be the kinetic isotope effect
in the “bimolecular” channel.
Assuming steady state in [OH‚DMS-d6] in the above mecha-
nism, the following rate expression can be derived:
k [O ] + k
10
kobs (kobs ≡ (k′obs - k6)/[DMS-d6]), as a function of [O2] at 247
K and 100 Torr of (N2 + O2). The kobs vs [O2] data were fit to
eq 13 by varying the magnitude of k8, k3f, and k3r while keeping
k3f/k3r ) Kc. Kc values at these temperatures were obtained via
extrapolation of the low-temperature equilibrium constant data.
A trial and error procedure was used in determining the best
values of k8, k3f, and k3r by minimizing the standard deviation
from a nonlinear least-squares fit of the data. Table 3
summarizes the obtained values of k8. The error bars are the
95% confidence limits from the fits to eq 13. Experiments at
higher temperatures were not possible because the concentrations
of O2 required to adequately characterize the enhancement of
kobs due to reaction 8 severely degraded the OH detection
sensitivity. Our results indicate that over the range 220-258
K, reaction 8 is essentially temperature independent.
d[OH]
dt
8
2
-
) k3 + k3f
[OH] ×
bi
[
(
)
]
k + k [O ] + k
3
f
8
2
10
[DMS-d ] - k [OH] (11)
6
6
The above expression can be integrated to yield
[
OH] ) [OH] exp - k
+
3bi
t
0
(
{
[
k [O ] + k
8
2
10
k
[DMS-d ] - k6
t
3
f
6
(
)
]
})
k + k [O ] + k
3
f
8
2
10
)
[OH] exp(-k′obst)
(12)
0
Analogous experiments were performed at 247 and 258 K
using DMS in place of DMS-d6, which yielded similar plots of
kobs vs [O2]. Figure 5 also shows one such plot at 247 K for
DMS. Assuming that extrapolations of our equilibrium constant
data for reaction 3f,r are valid for the OH + DMS system, the
data were fit to eq 13, where k3f, k3r, k3bi, and k8 are replaced
by k1f, k1r, k1bi, and k2, respectively. The obtained values of k2
are summarized in Table 3. Similar to the case of the DMS-
d6, experiments at higher temperatures were impossible because
of the degradation of the OH signal at large O2 concentrations.
The results show that the rate coefficient for O2 + OH‚DMS is
independent of the isotopic composition of DMS.
Equation 12 connects the measured second-order rate coefficient
(
kobs ≡ (k′obs - k6)/[DMS-d6]) to the elementary reaction rate
coefficients k3bi, k3f, k3r, k10, and k8. Here kobs is the observed
second-order rate coefficient given by the relation
k [O ] + k
10
8
2
kobs ) k + k3f
(13)
3bi
[
(
)
]
k + k [O ] + k
10
3
f
8
2
Assuming a steady state in OH‚DMS-d6 is valid because the
experimental conditions were maintained such that k3r g k3f-
[
DMS-d6] × (100) and the time required to reach steady state
was short compared to the time scale in which OH was lost.
Temporal profiles of [OH] were acquired in an excess of DMS-
d6 and O2 at 247 and 258 K. At these temperatures, the
equilibration between OH and OH‚DMS-d6 was fast and not
observable at times >5 µs. While a constant total pressure (N2
Reactions with very slight temperature dependencies, such
as reactions 2 and 8, require determinations over a large
temperature range to adequately define their activation energies.
Due to the limited temperature range over which reactions 2
and 8 could be studied, we prefer to average all measurements
+
O2) was maintained, [O2] was varied over the range 1-100
-
13
3
-1
Torr. All [OH] temporal profiles were exponential for at least
three lifetimes, and the first-order rate constant (k′obs) was
observed to increase systematically with the partial pressure of
O2. Figure 5 displays the measured second-order rate constant,
and report k8 ) k2 ) (10.0 ( 3.3) × 10
cm molecule
-1
s , independent of temperature. The error bars reflect 2σ of
the standard deviation of the mean and include estimates of
systematic uncertainty in the [O2], [DMS-d6], and [DMS].

1
4700 J. Phys. Chem., Vol. 100, No. 35, 1996
Barone et al.
Discussion
appears that the presence of ∼0.1% DMDS (possibly deuterated)
in DMS-d6 is a likely explanation for the deviation from
Arrhenius behavior in k3bi. Note that as the temperature
increases, with the accompanying increase in kbi, the contribution
of the impurity reaction decreases significantly. This is because
the impurity reaction has to be very fast and, hence, temperature
independent to make a significant contribution.
OH + DMS-d6 T OH‚DMS-d6. We have directly observed
OH temporal profiles at T e 240 K which show that OH radicals
add to DMS-d6 to form a stable, thermalized adduct, OH‚DMS-
d6. There are several experimental observations which suggest
that systematic errors did not influence our measurements of
Kc. Kc values were the same with two OH sources (H2O2 and
HONO) and did not depend on photolysis fluence or [DMS-
d6]. Although we did not systematically measure k3f and k3r as
functions of temperature and pressure, we did measure them at
different temperatures and pressures. It is clear that k3r increases
with bath gas pressure and temperature while k3f increases with
pressures and decreases only slightly with temperature. These
behaviors are consistent with the reaction of OH with DMS
being an addition process and the decomposition being hindered
by a thermal barrier. Even though k3f and k3r changed with
pressure and bath gas identity, their ratio, Kc, did not change.
This invariance of Kc suggests thermalization of the adduct
before it underwent unimolecular decomposition back to
reactants.
6
-12
Hynes et al. report k3bi(T) ) (10.3 ( 1.7) × 10
exp[-
3
-1 -1
(
4
498 ( 51)/T] cm molecule s , measured between 253 and
13 K. Although their values of A and Ea are slightly different
from ours, all of their measured rate constants agree with ours,
within the combined error bars, in the overlapping temperature
range.
Considering the indirect nature of the present determinations
(i.e., OH‚DMS-d6 was not directly monitored), a variation in
k3bi by 20% is not very large. Constraining the value of k3bi at
T < 250 K to that obtained by extrapolating the higher
temperature data increased the calculated equilibrium constants
by <20%, but the OH temporal profiles did not fit mechanism
1, as indicated by the residual sum of the squares. Incorporating
an extra loss process for OH‚DMS-d6 in our model and
constraining the value of k3bi to that determined from T > 250
K data resulted in k3f and k3r values that were <5% different
than those obtained from fits to mechanism 1. On the basis of
the insignificant sensitivity to such side reactions, we believe
that our equilibrium constants are good to ∼25%. Experiments
in which both OH and OH‚DMS-d6 are monitored and purer
samples of DMS-d6 are used would aid in identifying the cause
of the observed slight enhancement of k3bi at lower temperatures.
Our low-temperature values of k3bi, which were measured by
producing OH from 248-nm photolysis of H2O2, do not
extrapolate to the values obtained at T > 258 K, at which adduct
formation is minimal. The temperature dependence of reaction
3
bi was calculated using data at T g 258 K. k3bi values
measured at T < 250 K are approximately 20% larger than the
values obtained by extrapolating the higher temperature data
(see Figure 3). The values of k3bi determined in experiments
in which HONO was photolyzed to generate OH were slightly
larger than those measured using H2O2. There are several
possible explanations for the observed departure in kbi from
Arrhenius behavior at low T.
As mentioned earlier, Kc for DMS (nondeuterated) was not
extensively measured. The rough value of the measured Kc was
essentially the same as those for DMS-d6. This agreement is
expected because the methyl groups in DMS are unlikely to be
directly involved in the addition process. By looking at the
addition of OD to DMS, the regeneration of OH, if it is related
to the methyl groups of DMS, may be avoided (OD and OH
can be easily monitored independently). Such experiments are
planned for future studies in our laboratory.
It is possible that we have not accounted for one or more
loss pathways for the OH‚DMS-d6 adduct in our model. The
exponential nature of our postequilibrium decays in the OH
signal suggests that the loss process, if occurring, is first-order
with respect to the OH‚DMS-d6 adduct. (However, within the
experimental uncertainty of our OH temporal profiles, a small
second-order process cannot be ruled out.) One possible loss
process for OH‚DMS-d6 adduct is its reaction with the photolytic
precursor; however, the value of k3bi did not change significantly
6
Hynes et al. have previously suggested that there is no
significant isotope effect in the addition of OH to DMS because
the enhancement in their OH loss rate coefficients upon addition
of O2 to their reactor was the same in DMS and DMS-d6. Our
observation that Kc for the OH + DMS reaction is the same as
for OH + DMS-d6 reaction supports the proposal of Hynes et
al. Further, our results for DMS-d6 should be applicable to
nondeuterated DMS in the atmosphere.
1
3
with [H2O2] or [HONO], which were varied from 1 × 10 to
1
3
-3
2
0 × 10 molecules cm . The presence of an NO2 impurity
in our HONO source may, at least partly, account for the higher
values of k3bi measured using this OH source, if NO2 reacts
very rapidly with the OH‚DMS-d6 adduct. Another possible
loss process for the OH‚DMS-d6 is its unimolecular decomposi-
tion to products other than OH and DMS-d6. In the next section,
based on the observed kinetic isotope effect in reaction 1 in
large concentrations of O2, we argue against this possibility.
A more likely explanation for this discrepancy is the presence
of trace amounts of impurities in our DMS-d6 sample. In past
investigations of OH + DMS reaction, the presence of reactive
impurities in DMS samples has been a significant source of
6
Hynes et al. measured the loss rate of OH as a function of
DMS-d concentration at 261 K in N by generating OH via
6
2
the pulsed laser photolysis of H O or broad band flash
2
2
photolysis of H O. At low DMS-d concentrations ([DMS-d ]
2
6
6
1
6
-3
< 5 × 10 molecules cm ), their observed OH loss rate
coefficients were linearly proportional to the concentration of
DMS-d6 (i.e., obeyed pseudo-first-order kinetics; k′3bi )
k3bi[DMS-d6]). However, as the concentration of DMS-d6 was
raised above ∼5 × 10 molecules cm , their observed OH
loss rate coefficients were significantly lower than that expected
from the lower [DMS-d6] data. They attributed the deviation
to the formation of the OH‚DMS-d6 adduct and determined k3f
2
1,22
16
-3
systematic error.
Likely impurities include CH3SH and
dimethyl disulfide (DMDS), which react very rapidly with OH
-
11
-10
3
-1 -1
(3 × 10
and 2 × 10
cm molecule
s
at 298 K,
respectively). We used high-purity samples of DMS-d6 (>99.9%)
and DMS (99.5%). Even if all of the impurities in our samples
were DMDS, its reaction with OH would increase the measured
rate coefficients for the approach to equilibrium by <5%.
Although such impurities would not significantly affect the
measurement of k3f and k3r, it would have a pronounced effect
on k3bi. A 20% error in the value of kbi is possible if DMDS
-11
3
-1 -1
and k3r to be (1.15 ( 0.20) × 10
cm molecule
s
and
6
-1
(3.5 ( 2.0) × 10 s , respectively, at 261 K by fitting the
observed variation of the first-order rate coefficient with [DMS-
d6] to mechanism 1. These values of k3f and k3r yield a Kc value
-18
3
-1
of (3.3 ( 2.0) × 10
cm molecule , which is a factor of 2
smaller than that obtained by extrapolating our data to 261 K,
-
18
3
-1
(or deuterated DMDS) is present at a 0.1% and 0.4% level,
(6.1 ( 2.3) × 10 cm molecule . However, the two values
respectively, for the case of DMS-d6 and DMS. Therefore, it
overlap given the large error bars.

Reaction of OH with DMS. 1
J. Phys. Chem., Vol. 100, No. 35, 1996 14701
The small discrepancy in Kc between the two studies could
be due to the indirect nature of Hynes et al.’s measurements or
uncertainties in extrapolating our data. Unlike the present study,
Hynes et al. required the value of k3bi to obtain values of k3f
the calculation of the bond strength for the OH‚CS2 adduct, it
is reasonable that the OH‚DMS adduct’s binding energy was
underestimated by approximately 4 kcal mol . If this were
true, the result of McKee’s study would be in excellent
agreement with our experimental determination. In any case,
the results of McKee, Hynes et al., and the present study agree
-1
-12
and k3r from their data. Hynes et al. used k3bi ) 1.53 × 10
3
-1 -1
cm molecule s , which they determined in a separate series
of flash photolysis experiments in which OH was detected by
that OH forms a stable adduct with DMS. In contrast, two
6
10,24
resonance fluorescence. As mentioned by Hynes et al., and
computational studies by Gu and Turecek
suggest that the
noted during data analysis, the value determined for k3r is very
sensitive to the value chosen for k3bi. If the value of k3bi is
OH‚DMS adduct is unstable. The origin of the disagreement
between the ab initio studies of McKee and of Gu and Turecek
is not clear. Gu and Turecek also saw no evidence for the
formation of OH‚DMS adduct when protonated dimethyl
sulfoxide ion was neutralized and argued that OH‚DMS is not
-
12
3
-1 -1
taken to be 1.72 × 10
cm molecule s , the value at 261
6
K measured by Hynes et al. using laser-induced fluorescence
-18
3
detection of OH, their data would yield Kc ) 4.6 × 10
cm
-1
10
molecule . This value agrees better with our data and clearly
shows the difficulty in deducing the value of Kc without
observing the approach to equilibrium.
a bound species. The enthalpy change for the neutralization
reaction that Gu and Turecek used in an attempt to form the
OH‚DMS adduct
7
Barnes and co-workers have also measured the rate coef-
+
-
ficients for the formation and decomposition of the OH‚DMS
(CH ) SOH + e f (CH )SOH
(15)
3
2
3
(not OH‚DMS-d6), which, as discussed above, should be directly
-1
comparable to our values of Kc obtained in the DMS-d6 + OH
system. Barnes et al. measured the enhancement of k1bi as a
function of O2 partial pressure in their chamber. Assuming a
steady state in the OH‚DMS adduct, they derive an analytical
expression for the enhancement in the DMS loss rate in terms
is approximately -128 kcal mol . Given the large exother-
micity of reaction 15 and the weakly bound nature of the
OH‚DMS adduct, it is very likely that any OH‚DMS formed
would be highly excited and decompose very rapidly. Further,
the structure of protonated DMSO is unclear and its neutraliza-
tion may not lead to the OH‚DMS adduct formed in OH addition
to DMS.
-
12
3
of k1f, k1r, and k2. They report k1f ) (9.0 ( 0.5) × 10
cm
molecule-
1
s
-1
and estimate k1r ) 1.5 × 10 s at 298 K,
-19 3 -1
7
-1
yielding Kc ∼ 6 × 10 cm molecule , which is in agreement
OH‚DMS-d6/OH‚DMS + O2 f Products. We measured
the rate coefficient for reaction 8 to be k8 ) (1.00 ( 0.33) ×
with an extrapolation of our data for DMS-d6 to 298 K, Kc(298
-
19
3
-1
-12
3
-1 -1
K) ) (4.8 ( 2.2) × 10
cm molecule . However, this
agreement is likely to be fortuitous because the value of k2 (2
10
cm molecule s , independent of temperature (222-
258 K) and pressure (30-200 Torr of N2). One advantage of
our experimental approach is that k8 and k3f/k3r are determined
together for each measurement. The obtained values of k3f and
k3r agree well ((5%) with those measured in the absence of O2
and imply that mechanism 2 accurately describes the OH
temporal profile in the presence of DMS-d6 and O2. The rate
coefficient for reaction 8 (and reaction 2) was independent of
the concentrations of DMS-d6 (or DMS) and O2. The invariance
of the reaction rate coefficient with the isotopic composition of
DMS (DMS-d6 or DMS) suggests that the methyl groups in
the OH‚DMS adduct are not directly involved in reaction 2.
There are several possible OH-producing reactions that could
interfere with our measurements of k8 and k2. The direct
production of OH via the reaction of DMS‚OH with O2 is shown
in the accompanying paper to be a minor channel, if it occurs
at all. OH regeneration via the reaction of HO2
-12
3
-1 -1
×
10
cm molecule s ) assumed by Barnes et al. is a
factor of 2 larger than our directly determined value. Using
-
12
3
-1 -1
our value of k2 ) (1.00 ( 0.33) × 10
cm molecules s ,
we have fit the O2 enhancement data of Barnes et al. to the
expression
k [O ]
2
2
kobs ) k + k1f
(14)
1bi
[
(
)
]
k + k [O ]
1
r
2
2
which is the steady state solution for the effective bimolecular
loss rate coefficient (kobs) for DMS loss measured in Barnes et
al.’s chamber. Our reanalysis of Barnes et al.’s data yields k1f
-
11
3
-1 -1
7
-1
)
1.05 × 10
Kc ) 9.5 × 10
98 K determined from our study [k1bi ) (4.95 ( 0.35) × 10
cm molecule s ] and fit the O2 enhancement data of Barnes
cm molecule s , k1r ) 1.1 × 10 s , and
-19
3
-1
cm molecule . Further, if we use k1bi at
-12
2
-
3
-1 -1
HO + DMS-d f CD S(O)CD + OH
(16)
2
6
3
3
-11
3
et al. to eq 14, we obtain values of k1f ) 1.09 × 10
cm
-1
-1
7
-1
-19
molecule s , k1r )1.46 × 10 s , and Kc ) 7.5 × 10
is also unlikely because HO2 does not react with DMS.25 It is
possible that HO2 reacts with the OH‚DMS-d6 adduct to
regenerate OH. However, because very low concentrations of
3
-1
cm molecule , which are, though larger, still in fair agreement
with an extrapolation of our results.
Recently, Hynes et al.23 reported their observation of OH
addition to DMS-d6 using the PLP-PLIF technique and
OH were initially produced, the concentrations of HO and the
2
adduct were too small for such a radical-radical reaction to
-1
preliminary values of ∆°H3f ) -14.2 ( 2.6 kcal mol and
influence the OH profiles.
∆
°S3f ) -42.2 ( 10.8 cal mol- K , which are in fair
1
-1
Originally, Hynes et al. reported k ) (4.2 ( 2.2) × 10
6
-12
8
3
-1 -1
agreement with our results. However, because their work is
only a preliminary report, a detailed comparison of the individual
equilibrium constants is not possible.
cm molecule
s
at 261 K in 700 Torr of O , which is
2
significantly larger than our pressure-independent value Sub-
.
2
3
sequently, Hynes et al. reported a preliminary value of k )
8
-13
It is important to note that the discrepancies in the measure-
ments of Kc between the present study and those of previous
measurements are not very large. In all cases these differences
do not exceed a factor of 2. For reaction 3, a factor of 2 change
in Kc corresponds to an uncertainty in ∆°H3f of ∼0.35 kcal
8 × 10
2 2
at 258 K and 110 Torr of N and O , which is in
excellent agreement with the present results. The apparent
pressure dependence of k of Hynes et al. disagrees with our
8
measurements. As Hynes et al. pointed out, their measured k8
at 700 Torr is indirect and relies on separate and indirect
-
1
mol .
measurements of k , k , and k .
3
f
3r
3bi
9
A recent computational study by McKee yielded a binding
The results of the kinetics of the OH‚DMS + O2 reaction
also provide insight into the mechanism of the OH + DMS
reaction. Past studies of the mechanism of the OH + DMS
-
1
energy in the OH‚DMS adduct of 6.0 kcal mol relative to
OH and DMS. The author pointed out that, by analogy with

1
4702 J. Phys. Chem., Vol. 100, No. 35, 1996
Barone et al.
reaction have assumed that the reaction proceeds via two
separate reaction channels (addition and H atom abstraction).
This assumption excludes a possible mechanism consisting of
an initial addition of OH to DMS followed by rearrangement
to form the “bimolecular” products, H2O + CH3SCH2:
channel) in the atmosphere will be dependent on what fraction
of the oxidation proceeds via reaction 1 as opposed to reaction
18. As the contribution of reaction 18 increases, the fraction
of DMS that is lost to form DMS‚OH decreases. Therefore, to
calculate the extent of OH addition to DMS, we need the
ambient nighttime NO3 concentrations, which, at present, are
poorly defined in the marine boundary layer.
OH + DMS (+ M) T OH‚DMS (+ M)
(1f)
The atmospheric fate of the adduct will be its reaction with
O2 or thermal decomposition back to reactants. No other
reactions of the adduct need be considered due to the over-
whelming abundance of O2 in the atmosphere. The products
of the OH‚DMS + O2 reaction are described in the accompany-
ing paper. The implication of these kinetic studies to the
chemistry of DMS in the atmosphere is also described in the
accompanying paper.
OH‚DMS (+ M) f H O + CH SCH (+ M) (17)
2
3
2
Our measurements of the observed bimolecular rate constants
for OH + DMS and OH + DMS-d6 reaction at high O2
concentrations provide the first qualitative evidence that indeed
two uncoupled reaction channels exists. The reaction of OH
with DMS, in the absence of O2, exhibits a significant kinetic
isotope effect, which indicates a direct reaction involving H atom
abstraction at least part of the time. As the concentration of
O2 increases, more of the OH‚DMS adduct will react with O2
at the expense of reaction 17. Therefore, if reaction 17 is the
primary path for the “bimolecular” reaction, the observed OH
loss rate constant in an excess of O2 would be independent of
the isotopic nature of DMS. This behavior arises because the
adducts, once formed, are scavenged faster then they can
decompose to reactants. (The above interpretation relies on the
assumption that the O2 + OH‚DMS rate constant exhibits no
isotope effect, a contention which is supported by our measured
values of k1 and k8 being equal at 247 and 258 K.) Figure 5
clearly shows that as the O2 concentration is raised, kobs increases
and levels off at two different asymptotic values for DMS and
DMS-d6. The difference between the asymptotes, A and B in
Acknowledgment. We thank Prof. M. McKee for providing
the vibrational frequencies and rotational constants used in our
calculations and Prof. P. H. Wine for sharing data before
publication. This work was funded in part by NOAA’s Climate
and Global Change Research Program. S.B.B. held an NSF
Atmospheric Traineeship during part of this work.
References and Notes
(
1) Bates, T. S.; Lamb, B. K.; Guenther, A.; Dignon, J.; Stoiber, R. E.
J. Atmos. Chem. 1992, 14, 315.
2) Charlson, R. J.; Lovelock, J. E.; Andreae, M. O.; Warren, S. G.
(
Nature 1987, 326, 655.
(3) Turnipseed, A. A.; Ravishankara, A. R. The atmospheric oxidation
of dimethyl sulfide: elementary steps in
a complex mechanism.
Dimethylsulphide: oceans, atmosphere and climate. Proceedings of the
International Symposium, Belgirate, Italy; Restelli, G., Angeletti, G., Eds.;
Kluwer Academic Publishing: Dordrecht, The Netherlands, 1992; p 185.
(4) Tyndall, G. S.; Ravishankara, A. R. Int. J. Chem. Kinet. 1991, 23,
483.
-
12
3
-1 -1
Figure 5, is approximately 3 × 10 cm molecule s . This
value is also the difference between the measured values of k1
-
12
3
-1 -1
and k3 [k1 - k3 ) (3.2 ( 0.5) × 10
cm molecule s ]
(
09.
5) Yin, F.; Grosjean, D.; Seinfeld, J. H. J. Atmos. Chem. 1990, 11,
measured at 298 K in the absence of O2. The experiments
carried out at 258 K yield similar results. Therefore, we suggest
that the OH + DMS/DMS-d6 reaction indeed proceeds through
two independent reaction pathways, addition and H atom
abstraction.
3
(6) Hynes, A. J.; Wine, P. H.; Semmes, D. H. J. Phys. Chem. 1986,
90, 4148.
(
7) Barnes, I.; Bastian, V.; Becker, K. H. Int. J. Chem. Kinet. 1988,
2
0, 415.
(8) Wallington, T. J.; Atkinson, R.; Tuazon, E. C.; Aschmann, S. M.
This observation has important consequences with respect
to the effective branching between OH addition and abstraction
in the atmosphere. Because the addition and abstraction
channels are uncoupled, the effective branching can be calcu-
lated simply by subtracting the bimolecular rate coefficient for
reaction 1 from the observed bimolecular rate coefficient in an
atmosphere of air:
Int. J. Chem. Kinet. 1986, 18, 837.
(9) McKee, M. L. J. Phys. Chem. 1993, 97, 10971.
10) Gu, M.; Turecek, F. J. Am. Chem. Soc. 1992, 114, 7146.
11) Turecek, F. J. Phys. Chem 1994, 98, 3701.
12) Turnipseed, A. A.; Barone, S. B.; Ravishankara, A. R. J. Phys.
(
(
(
Chem. 1996, 100, 14703.
(13) Turnipseed, A. A.; Barone, S. B.; Ravishankara, A. R. J. Phys.
Chem. 1993, 97, 5926.
(14) Barone, S. B.; Turnipseed, A. A.; Ravishankara, A. R. J. Phys.
Chem. 1994, 98, 4602.
Φaddition ) (kobs - k )/k
1bi
obs
(15) Ondrey, G.; van Veen, N.; Bersohn, R. J. Chem. Phys. 1983, 78,
3
732.
(16) Vasudev, R.; Zare, R. N.; Dixon, R. N. J. Chem. Phys. 1984, 80,
4863.
Such bimolecular reaction rate coefficients in air have been
measured as a function of temperature by Hynes et al.
6
(
17) Hearn, C. H.; Turcu, E.; Joens, J. A. Atmos. EnViron. 1990, 24A,
1
939.
Atmospheric Implications
(18) DeMore, W. B.; Sander, S. P.; Golden, D. M.; Hampson, R. F.;
The results of previous studies have suggested that reaction
in combination with its reaction with NO3
Kurylo, M. J.; Howard, C. J.; Ravishankara, A. R.; Kolb, C. E.; Molina,
M. J. Chemical Kinetics and Photochemical Data for Use in Stratospheric
Modeling; JPL Publication 94-26; Jet Propulsion Laboratory: Pasadena,
CA, 1994.
1
NO + DMS f products
(18)
3
(
19) Benson, S. W. Chem. ReV. 1978, 78, 23.
(20) Malleson, A. M.; Kellet, H. M.; Myhill, R. G.; Sweetenham, W.
will dominate the initiation of DMS oxidation in the troposphere.
This is due to the relatively large abundances of NO3 and OH
in the troposphere in combination with the large rate coefficients
for these reactions. In this study we have shown that reaction
P. AERE Harwell Publication R 13729; AERE Publications Office:
Oxfordshire, U.K., 1990.
(
21) Atkinson, R.; Perry, R. A.; Pitts, J. N., Jr. Chem. Phys. Lett. 1978,
5
4, 14.
(22) Kurylo, M. J. Chem. Phys. Lett. 1978, 58, 238.
1
proceeds through a complex mechanism involving at least
(23) Hynes, A. J.; Pounds, T.; McKay, T.; Bradshaw, J. D.; Wine, P.
H. Detailed mechanistic studies of the OH-initiated oxidation of biogenic
sulfur compounds under atmospheric conditions. Presented at the 12th
International Symposium on Gas Kinetics, Reading, U.K., 1992.
two active channels. Therefore, the initiation reaction involving
OH reactions with DMS is the first possible branching point,
which can lead to different products, in its atmospheric
oxidation. This is in contrast to initiation via reaction with NO3,
which has been shown to react exclusively to form CH3SCH2
(
24) Turecek, F. J. Phys. Chem. 1994, 98, 3701.
(25) Mellouki, A.; Ravishankara, A. R. Int. J. Chem. Kinet. 1994, 26,
55.
(26) Butkovskaya, N. I.; LeBras, G. J. Phys. Chem. 1994, 98, 2582.
3
2
6
and H2O. The extent of branching between OH addition and
H atom abstraction” (the most likely active bimolecular
“
JP960866K