FT-IR kinetic and product study of the bmnitiated oxidation of dimethyl sulfide

Article abstract of DOI:10.1002/(SICI)1097-4601(1999)31:12<883::AID-KIN7>3.0.CO;2-V

An FT-IR kinetic and product study of the Br-atom-initiated oxidation of dimethyl sulh'de (DMS) has been performed in a large-volume reaction chamber at 298 K and 1000mbar total pressure as a function of the bath gas composition (N2 + O2). In the kinetic investigations using the relative kinetic method, considerable scatter was observed between individual determinations of the rate coefficient, suggesting the possibility of interference from secondary chemistry in the reaction system involving dimethyl sulfoxide (DMSO) formation. Despite the experimental difficulties, an overall bimolecular rate coefficient for the reaction of Br atoms with DMS under atmospheric conditions at 298 K of 1 × 10^-13 cm³ molecule-' s^-1 can be deduced. The major sulfur products observed included SO₂, CH3SBr, and DMSO. The kinetic observations in combination with the product studies under the conditions employed are consistent with rapid addition of Br atoms to DMS forming an adduct that mainly re-forms reactants but can also decompose unimolecularly to form CH₃SBr and CH₃ radicals. The observed formation of DMSO is attributed to reactions of BrO radicals with DMS rather than reaction of the Br-DMS adduct with O₂ as has been previously speculated and is thought to be responsible for the variability of the measured rate coefficient. The reaction CH₃O₂ + Br → BrO + CH₂O is postulated as the source of BrO radicals.

Full text of DOI:10.1002/(SICI)1097-4601(1999)31:12<883::AID-KIN7>3.0.CO;2-V

FT-IR Kinetic and Product
Study of the Br-Initiated
Oxidation of Dimethyl
Sulfide
T. MAURER, I. BARNES, K. H. BECKER
Bergische Universit a¨ t-GH-Wuppertal, Physikalische Chemie/FB 9, 42097 Wuppertal, Germany
Received 1 March 1999; accepted 30 August 1999
ABSTRACT: An FT-IR kinetic and product study of the Br-atom-initiated oxidation of dimethyl
sulfide (DMS) has been performed in a large-volume reaction chamber at 298 K and 1000-
mbar total pressure as a function of the bath gas composition (N ϩ O ). In the kinetic inves-
2
2
tigations using the relative kinetic method, considerable scatter was observed between indi-
vidual determinations of the rate coefficient, suggesting the possibility of interference from
secondary chemistry in the reaction system involving dimethyl sulfoxide (DMSO) formation.
Despite the experimental difficulties, an overall bimolecular rate coefficient for the reaction
Ϫ
13
3
Ϫ1
of Br atoms with DMS under atmospheric conditions at 298 K of Յ1 ϫ 10 cm molecule
s
Ϫ
1
can be deduced. The major sulfur products observed included SO , CH SBr, and DMSO.
2 3
The kinetic observations in combination with the product studies under the conditions em-
ployed are consistent with rapid addition of Br atoms to DMS forming an adduct that mainly
re-forms reactants but can also decompose unimolecularly to form CH SBr and CH radicals.
3
3
The observed formation of DMSO is attributed to reactions of BrO radicals with DMS rather
than reaction of the Br–DMS adduct with O as has been previously speculated and is thought
2
to be responsible for the variability of the measured rate coefficient. The reaction CH O ϩ Br
3
2
:
BrO ϩ CH O is postulated as the source of BrO radicals. ᭧ 1999 John Wiley & Sons, Inc. Int J
3
Chem Kinet 31: 883–893, 1999
INTRODUCTION
and BrO radicals, which has been observed at concen-
trations up to ϳ17 pptv [2–4], (ii) the measurement
of high levels of filterable bromine [5,6], and (iii) the
change in the pattern of non-methane hydrocarbons
measured in air plumes, which is more in keeping with
Cl-atom-initiated or in some cases Br-atom-initiated
rather than OH-radical-initiated oxidation [7,8]. In-
creased attention has been given in recent years to the
possibility that reactions occurring on sea-salt surfaces
can lead to production of halogen atoms in the marine
boundary layer [3,9,10].
Over the past few years episodicdestrut ci on of
ground-level ozone has been observed in the spring-
time Arctic boundary layer [1]. Evidence is now ac-
cumulating that catalytic cycles involving reactive hal-
ogens, in particular, BrO chemistry, are responsible
for the sudden ozone decay. The evidence for the in-
x
volvement of BrO during the episodes is based on (i)
x
the anticorrelation observed between the ozone decay
Dimethyl sulfide (CH SCH ; DMS) is one of the
3
3
major natural organicsulfur co mpounds emitted to the
atmosphere, and its possible involvement in the reg-
Correspondence to: I. Barnes
1999 John Wiley & Sons, Inc.
᭧
CCC 0538-8066/99/120883-11

8
84
MAURER ET AL.
ulation of the Earth’s radiation budget and its oxida-
tion mechanisms have been the subject of considerable
research [11–14]. Recent field measurements have
shown that the Arctic Ocean is a potentially substantial
source of DMS, which is likely to become important
as sea ice melts in early summer [15]. It is already
reasonably well established that the reaction of BrO
with DMS is relatively fast, results quantitatively in
the formation of dimethyl sulfoxide (CH SOCH ;
room temperature is reported where adduct formation
dominates [20,21]. The aim of the study was to inves-
tigate the reaction of Br atoms with DMS as a function
of the O partial pressure in order to test for possible
2
kinetic effects and also to establish whether the reac-
tion of the (CH ) SBr adduct with O leads to the for-
3
2
2
mation of DMSO or if other reaction channels are op-
erative.
3
3
DMSO) [16–18], and may be of significance for the
oxidation of DMS [19].
EXPERIMENTAL
The reaction of Br atoms with DMS has not been
extensively investigated. The kinetics of the reaction
of Br with DMS has been investigated by Wine and
co-workers [20,21] using laser flash photolysis–res-
onance fluorescence as a function of temperature and
pressure. The studies show that between 260 and 310
K the main reaction pathway is reversible adduct for-
mation and at temperatures above 375 K the
The kineticinvestigations on the rea ct ion of Br atoms
with DMS were performed in a 1080-L quartz glass
reaction chamber equipped with 32 super actinic flu-
orescent lamps (Philips TL05/40 W: 320 Ͻ ␭ Ͻ 480
nm, ␭max ϭ 360 nm) and 32 low-pressure mercury
lamps (Philips TUV 40 W: ␭max ϭ 254 nm). Only the
fluorescent lamps were used in the present experi-
ments. The experimental setup has been described in
detail elsewhere [23]. The relative kineticte ch nique
(
CH ) SBr decomposition is so rapid that addition
3 2
does not effectively occur.
[24] was applied to determine the rate coefficient for
Br ϩ (CH ) S ϩ M !: (CH ) SBr ϩ M
the reaction of Br atoms with DMS at 298 ϩ 2 K and
a total pressure of 1000 mbar (O ϩ N ). The O partial
pressure was varied between 50 and 400 mbar. The
visible photolysis (hereafter referred to as VIS pho-
3
2
3 2
2
2
2
(dominating pathway below 310 K)
(
CH ) SBr ϩ M !: Br ϩ (CH ) S ϩ M
3 2 3 2
tolysis) of Br vapor (approximately 50 ppmV: 1 ppm
2
ϭ 2.46 ϫ 1013 molecule cm at 298 K),
Ϫ3
The results suggest that at high temperatures a hydro-
gen abstraction pathway dominates.
Br ϩ h␯ !: 2Br
2
Br ϩ CH SCH !: CH SCH ϩ HBr
3
3
3
2
was used to generate Br radicals. The Br radical con-
centration was typically 10 radicals cm . The initial
(dominating pathway above 375 K)
9
Ϫ3
concentration of molecular Br was typically 20 ppmV
2
From an investigation of the fate of the (CH ) SBr
3
2
and those of both reactants ϳ4 ppmV. The concentra-
tion–time behaviors of the organic co mpounds were
monitored in situ using long-path FT-IR spectroscopy
(Bruker IFS-88, 492-m path length, HgCdTe detector).
adduct in the presence of O , Wine and co-workers
2
Ϫ16
3
Ϫ
1
Ϫ1
placed a limit of 3 ϫ 10 cm molecule
s
on the
rate of reaction of the adduct with O and its possible
2
reaction to form dimethyl sulfoxide and BrO radicals:
Ϫ1
Generally, infrared spectra were recorded at 1-cm
resolution and 10–15 spectra were collected over ir-
radiation periods of 20–30 min.
(
CH ) S9Br ϩ O !: (CH ) SO ϩ BrO
3 2 2 3 2
The bromine atoms generated in the photolysis sys-
tem will react with DMS and the reference hydro-
carbon:
Barnes et al. [22] have reported preliminary results
from a kinetic and product study on the reaction of Br
with DMS at 298 K and atmosphericpressure. A rate
Ϫ
14
3
Ϫ1 Ϫ1
coefficient of (9.6 Ϯ 1.0) ϫ 10 cm molecule
s
Br ϩ DMS !: products, k1
was reported in syntheticair and the results suggested
a possible dependence upon the O partial pressure.
Br ϩ reference !: products, k2
2
The products observed included HBr, DMSO,
DMSO , SO , and what was tentatively identified as
Control experiments for all of the compounds inves-
tigated showed that other removal processes such as
loss to the wall and/or photolysis were negligible. The
control experiments consisted of observing the con-
centration–time behavior of DMS and the reference
2
2
CH SCH Br. The product yields and the mechanism
3
2
of their formation were not reported.
In the present work a detailed product and kinetic
investigation of the reaction of Br atoms with DMS at

Br-INITIATED OXIDATION OF DIMETHYL SULFIDE
885
hydrocarbon without the Br radical source over a pe-
riod of 1 h either in the dark or with the photolysis
lamps switched on. Since the reaction of Br with DMS
is known to proceed through adduct formation, pho-
tolysis of DMS–Br might also pose a complication.
However, the UV spectrum for DMS does not extend
below 280 nm. It is very unlikely that the addition of
an auxochrome such as Br would result in a large
enough shift of the absoprtion to longer wavelengths
to cause a significant overlap with the spectral region
of the lamps (320 Ͻ ␭ Ͻ 480 nm, ␭max ϭ 360 nm). In
addition, photolysis has not been observed for DMS
and compounds such as S Br and CH SO Br. It is
therefore extremely unlikely that photolysis of the
DMS–Br will be a loss pathway for the adduct and
consequently this pathway has not been taken into ac-
count in the analysis of the data.
Since all other processes for the loss of reactants,
apart from reaction with Br atoms, appear to be neg-
ligible, the following rate laws will govern the decay
of DMS and the reference:
used was 50 mbar; below this pressure, the rate coef-
ficient for the reaction of Br with acetylene rapidly
declines and the change in the measured acetylene
concentration becomes too small for accurate analysis
with FT-IR.
The product studies were performed on DMS/Br2/
N /O mixtures at a total pressure of 1000 mbar using
2
2
approximately 4 ppm DMS and 10 ppm Br . The con-
2
centration–time behavior of reactants and products
was again monitored using in situ long-path FT-IR
spectroscopy. Typically, infrared spectra were re-
corded over a 2-min period by coadding 64 interfer-
ograms and 10 such spectra were collected over 20
min.
Molecular bromine (grade “pure”) and dimethyl
sulfoxide (Ͼ99.5% (GC)) were purchased from Fluka
Chemie, and dimethyl sulfide (Ͼ99.5% (GC)) was
purchased from Sigma-Aldrich Chemie.
2
2
3
2
RESULTS AND DISCUSSION
d[DMS]
Ϫ
ϭ k [Br][DMS]
(I)
Kinetic Investigations
1
dt
The kineticdata obtained from the VIS photolysis of
d[reference]
DMS/acetylene/Br mixtures gave reasonably linear
2
Ϫ
ϭ k [Br][reference]
(II)
2
plots when analyzed according to eq. (III). An exam-
ple for a measurement made in syntheticair is shown
in Figure 1. The rate coefficients extracted from the
dt
Equations (I) and (II) can be rearranged to give eq.
III):
plots for the various O partial pressures are listed in
2
(
Table I. The error limits are 1␴ and represent the ex-
perimental precision from a minimum of five experi-
ments. In the data set the rate coefficients appear to go
[
DMS]t0
[reference]t0
[reference]t
k1
ln
ϭ
ln
(III)
[
DMS]t
k2
If the mechanism is valid, then plots of ln{[DMS] /
t
0
[
DMS] } vs. ln{[reference] /[reference] } should give
t
t
t
0
straight lines with zero intercepts. The measurements
were performed relative to the reaction of Br atoms
with acetylene. At conditions of constant pressure and
temperature, the rate coefficients for the reaction of Br
atoms with unsaturated hydrocarbons depend on the
partial pressure of O . The rate coefficient for the re-
2
action of Br with acetylene at the respective O partial
2
pressures have been calculated using the expression
for the temperature and O dependence of the rate co-
2
efficient given in the study of Barnes et al. [25]:
1
.15 ϫ 10 34e1888/RT [O2 ]
Ϫ
kBrϩacetylene
ϭ
Ϫ3932/RT
5
8.3e
ϩ [O ]/[M]
2
Figure 1 Example of the experimental data plotted ac-
cording to eq. (III) for experiments performed on a DMS/
where R is the gas constant (cal K 1 mol ), T is the
Ϫ
Ϫ1
temperature (K), and [O ] and [M] are concentrations
acetylene/Br mixture in 1000 mbar of syntheticair, i.e., an
O partial pressure of 200 mbar.
2
2
2
Ϫ3
(molecules cm ). The lowest partial pressure of O
2

8
86
MAURER ET AL.
Table I Examples of the Kinetic Data Obtained for the Reaction of Br Atoms with DMS as a Function of the O2
Partial Pressure at Room Temperature and 1000-mbar Total Pressure (N ϩ O ) Using the Competitive Kinetic
2
2
Technique and Acetylene as the Reference Hydrocarbon
[
O2]
K
kBrϩDMS/
kBrϩacetylene
k
Brϩ
Br
3
ϩ
acetylene
DMS
Ϫ
14
Ϫ
1
Ϫ
1
Ϫ
13
3
Ϫ
1
Ϫ1
T(K)
(mbar)
(10 cm molecule s )
(10 cm molecule s )
300
299
295
298
299
50
100
200
300
400
2.60 Ϯ 0.41
3.80 Ϯ 0.60
4.97 Ϯ 0.75
5.45 Ϯ 0.83
5.42 Ϯ 0.79
4.25 Ϯ 0.16
3.24 Ϯ 0.07
2.98 Ϯ 0.07
2.63 Ϯ 0.05
2.35 Ϯ 0.03
1.09 Ϯ 0.18
1.23 Ϯ 0.18
1.50 Ϯ 0.22
1.43 Ϯ 0.22
1.27 Ϯ 0.19
through a maximum and decrease again at higher O2
partial pressures; however, the variation lies within the
combined experimental error limits of the individual
rates.
which has been used to describe the O dependence of
2
the reactions of OH radicals with DMS [26] and CS2
[27], is apparently not valid in this case, or at least
under the present experimental conditions.
Ϫ
13
3
The value of k ϭ (1.48 Ϯ 0.22) ϫ 10 cm mol-
The results from the kineticmeasurements suggest
that the relative competitive technique under the
present experimental conditions is not providing reli-
able kinetic data. As will be discussed later, reaction
of BrO radicals with DMS forming DMSO is thought
to be responsible for the variation in the measured rate
coefficients. Ideally, experiments in pure nitrogen
should eliminate any interference from BrO chemistry.
Unfortunately, as previously indicated, the value of the
rate coefficient for the reference reaction of Br with
Ϫ1
Ϫ1
ecule
s
obtained at 200 mbar O partial pressure
2
is approximately 50% higher than the value of k ϭ
Ϫ
14
3
Ϫ
1
Ϫ1
(
9.6 Ϯ 1.0) ϫ 10 cm molecule
s
obtained in a
previous study in this laboratory [22], where a similar
behavior of the rate coefficients with variation in the
O partial pressure was noted. However, as in the
2
present study, the variation remained within the com-
bined experimental error limits.
In the previous and present investigations from this
laboratory, the molecular bromine and acetylene con-
centrations employed were approximately the same for
both studies; however, the initial level of DMS was
approximately a factor 5–8 higher than in the present
study. To test for potential artifacts in the kinetic mea-
surements, the initial concentration of DMS was var-
ied over the range 3–25 ppm while keeping the ref-
erence hydrocarbon concentration constant and the O2
partial pressure at 200 mbar. The measured rate co-
acetylene depends on the O partial pressure and be-
2
comes too slow at pressures below about 50 mbar.
Interestingly, of the measured values, the value ob-
Ϫ
13
3
Ϫ1
tained with 50 mbar (1.09 ϫ 10
cm molecule
Ϫ
1
s ) is the lowest, possibly reflecting reduced interfer-
ence from BrO chemistry. Despite the experimental
difficulties, an upper limit can be calculated from the
results for the rate coefficient at atmospheric pressure.
With the yield of DMSO as an approximate measure
of the involvement of BrO, it has been estimated that
the overall rate coefficient for the reaction of Br with
Ϫ
13
efficient was found to vary from 0.95 ϫ 10 to 1.85
Ϫ
13
3
Ϫ
1
Ϫ1
ϫ 10 cm molecule s , which is well outside the
Ϯ15% error limits of the individual experiments; how-
ever, no systematic dependence on the DMS concen-
tration was evident.
Ϫ
13
DMS in air at 298 K will be k
molecule s .
Յ 1 ϫ 10 cm3
BrϩDMS
Ϫ
1
Ϫ1
The results from the previous study combined with
the present observations show that the DMS/alkene/
Br /O /N system, although obeying pseudo-first-
order kinetics reasonably well, does not provide re-
producible results. The lack of any pronounced
Product Studies
2
2
2
Figure 2, trace a, shows an infrared product spectrum
Ϫ
1
in the range 3600–700 cm obtained from the Br-
atom-initiated oxidation of DMS in 1000 mbar of syn-
theticair after subtra ct ion of absorptions due to DMS.
In this spectrum, absorptions due to CO, HCHO, HBr,
dependence on the O partial pressure also indicates
2
that the simple kineticmodel
SO , and DMSO are readily visible. Other products
2
Br ϩ (CH ) S ϩ M ;: (CH ) SBr ϩ M
3
2
3 2
also detected in the spectrum include CH OOH,
3
CH OH, DMSO , HCOOH, and OCS. Figure 2, trace
3
2
(
CH ) SBr ϩ O !: (CH ) SO ϩ BrO
b, is a reference spectrum of methanesulfenyl bromide
3
2
2
3 2

Br-INITIATED OXIDATION OF DIMETHYL SULFIDE
887
[29]. As can be seen, the absorptions are very similar
in both cases, giving confidence that the observed
product is indeed CH SBr. For the purposes of cali-
3
bration, it has been assumed that one molecule of
DMDS results in the formation of two molecules of
CH SBr. Since the spectrum appears to be devoid of
3
other products, this is probably a reasonable assump-
tion. The error in the concentration has been estimated
to be approximately 20%. The following reactions are
thought to lead to the formation of CH SBr in close
3
to unit yield:
CH SSCH ϩ h␯ !: CH S ϩ CH S
3
3
3
3
CH S ϩ Br !: CH SBr ϩ Br
3
2
3
CH S ϩ Br ϩ M !: CH SBr ϩ M
3
3
A comparison between Figure 2, panels a and b, shows
quite clearly that CH SBr is also a product of the Br-
3
atom-initiated oxidation of DMS.
Figure 2, panel c, shows the residual spectrum ob-
tained after subtraction of all the identified products
listed previously. The spectrum shows two strong ab-
Ϫ1
sorption peaks centered around 1394 and 1182 cm
and a number of moderately strong peaks around
Ϫ
1
3
000, 1057, 962, and 750 cm . Table III, part a, com-
Ϫ
1
pares the SO bond stretching frequencies (in cm ) for
DMSO and other halogenated sulfoxides for both the
gas and liquid phases and part b the gas-phase asym-
metricand symmetricSO stretching frequencies for
2
DMSO and other halogenated sulfones. From the re-
2
ported infrared spectrum for methanesulfonyl bromide
(
CH SO Br), it is clear that the absorptions at 1394,
3 2
Table II Molecular Frequencies (cm 1) of
Ϫ
Methanesulfenyl Bromide (CH SBr) Generated by the
3
Figure 2 Infrared absorption spectra in the range 3600–
a
Photolysis of DMDS and Br in 1000 mbar of N2
2
Ϫ
1
7
00 cm . Trace a shows an infrared product spectrum ob-
CH SBr
(cm )
CH SCl
3
(cm )
tained after 10-min irradiation of a DMS/Br mixture in 1000
3
2
Ϫ
1
Ϫ1
Assignment
mbar of syntheticair and subtra ct ion of absorptions due to
DMS. Trace b is a reference spectrum of methanesulfenyl
␯6
␯5
␯11
␯4
␯10
␯3
␯3 (␯s,CH3)
(␯ )
(rCH3)
(rCH3)
(␦s,CH3)
(␦a,CH3)
(␦a,CH3)
713
709.7
958.2
965.4
1318.6
143.1
1444.9
CS
bromide (CH SBr). Trace c is the residual spectrum obtained
3
954.8
963.8
1318
1415.5
1445.2
2873.5
2927
from trace a after subtraction of all the identified products.
(
CH SBr). The infrared spectrum of CH SBr was ob-
3
3
tained by the 254-nm photolysis of dimethyl disulfide
2
with molecular bromine in 1000 mbar of N . A similar
2
␯2 (␯a,CH3)
␯9 (␯a,CH3)
2929.1
3004.6
process using the photolysis of DMDS in the presence
of Cl has recently been used to produce CH SCl in
3005.2
2
3
a
The assignments and approximate descriptions of the modes
are based on those for the gas-phase spectrum of methanesulfenyl
close to unit yield [28]. The infrared absorption bands
obtained for CH SBr are listed in Table II, where they
3
chloride (CH SCl) reported in ref. [29]. The molecular frequencies
3
are compared with the literature values for CH SCl
for CH SCl as measured in this laboratory are given for comparison.
3
3

8
88
MAURER ET AL.
Table III S"O Bond Stretching Frequencies of Some Sulfoxide/Sulfone Molecules
a) Sulfoxides
␯(Gas Phase)
(
Compound
␯(Liquid)
Dimethyl sulfoxide (CH SOCH )
Methanesulfinyl
1103
1166 (est.)
1050
—
3
3
bromide (CH SOBr)
3
Methanesulfinyl
1180.7
1150
chloride (CH SOCl)
3
Thionyl bromide (SOBr2)
Thionyl chloride (SOCl2)
1129 (est.)
1255.5
1202.7
1229.2
(
b) Sulfones
Compound
␯ (SO str)
␯ (SO str)
as
2
s
2
Dimethyl sulfone (CH SO CH )
Methanesulfonyl
1355
1391a
1162
1178a
3
2
3
bromide (CH SO Br)
3
2
Methanesulfonyl
1404
1192
chloride (CH SO Cl)
3
2
Sulfuryl bromide (SO Br )
1426 (est.)
1436.2
1190 (est.)
1206.7(?)
2
2
Sulfuryl chloride (SO Cl )
2
2
a
Geiseler, G.; Nagel, B. J Mol Struct 1973, 16, 79.
182, and 962 cm 1 (Figure 2c) must be principally
Ϫ
1
due to this compound. However, from the concentra-
Ϫ1
tion behavior of the band centered around 1182 cm ,
it is also evident that other compounds must also con-
tribute to this absorption. From the absorptions listed
in Table III, contributions from methanesulfinyl bro-
mide (CH SOBr) and sulfuryl bromide (SOBr ) are
3
2
possibilities. The further oxidation of CH SCl is
3
known to lead to the chlorinated analogs [28].
Figure 3 shows an example of the concentration–
time profiles for DMS and the identified products dur-
ing the photolysis of molecular bromine and DMS in
1
000 mbar of syntheticair (i.e., 200 mbar O ). Similar
2
profiles were obtained for all of the other O partial
2
pressures investigated. Even for the supposedly “pure”
N case (contains Ͼ100 ppm O due to leaks), for-
2
2
mation of DMSO and SO was observed, albeit at
2
much lower yields than in the systems containing con-
trolled partial pressures of O . The concentration–
2
time profiles for the measurement in N are shown in
2
Figure 4.
The upward curvature of the concentration–time
profiles of CH SBr, DMSO, CO, HCHO, and HBr in
3
the initial phase of the reaction is typical for primary
products. The profiles for CH SBr, DMSO, and
3
HCHO go through a maximum, indicating further ox-
idation in the reaction system. The oxidation of HCHO
is particularly efficient; however, this is not surprising
considering the fast reaction with Br atoms for this
Figure 3 Concentration–time profiles for the products ob-
served in the reaction of Br atoms with DMS in 1000 mbar
of syntheticair at 298 K. The lines are just best empiri ca l
polynomial fits to the experimental points and do not rep-
resent model simulations.
Ϫ
12
3
Ϫ1
compound (kBr
HCHO ϭ 1.0 ϫ 10
cm molecule
ϩ
Ϫ
1
s
at 298 K [30]. The profiles for SO , CH OOH,
2
3

Br-INITIATED OXIDATION OF DIMETHYL SULFIDE
889
the yield of CH SBr converged to a constant value for
3
all the O partial pressures.
2
In contrast, the yield of DMSO shows a pronounced
dependence on the O partial pressure. The lowest ini-
2
tial yield of ϳ28% was obtained for the experiments
carried out in “pure” N . With increasing O partial
2
2
pressure, the initial DMSO yield gradually increased
to values of around 50–55%. As with the yield of
CH SBr, the yield of DMSO converged to a constant
3
value at longer reaction times for all the O partial
2
pressures. The yield of the other important product,
CO, was independent of reaction time and the O par-
2
tial pressure, with an average yield of approximately
4
0% C for all the conditions. The yield of HBr was
variable, with values between 0.8 and 1.2 molecules
per molecule of DMS consumed.
Mechanistic Considerations
As discussed in the introduction, the reaction can pro-
ceed either via a direct H-atom abstraction to produce
CH SCH radicals and HBr or addition to form a Br–
3
2
Figure 4 Concentration–time profiles for the products ob-
served in the reaction of Br atoms with DMS in 1000 mbar
of N at 298 K. The lines are just best empirical polynomial
2
fits to the experimental points and do not represent model
simulations.
CH OH, DMSO , HCOOH, and OCS all show de-
3
2
layed formation, indicating that they are secondary
products.
The yields of SO and CH OOH only gain in im-
2
3
portance toward the end of the irradiation period and
that of HCHO remains low over the whole irradiation
period. The yields of CH OH, HCOOH, and OCS
3
were always low. The yield of OCS was typically
0
.5%, which is similar to the yields recently reported
for the OH-initiated oxidation of DMS.
Initially, CH SBr and DMSO have approximately
3
the same rates of formation; however, as CH SBr con-
3
tinues to increase, the concentration of DMSO levels
out to an approximately constant value. After reaching
a maximum value, CH SBr begins to decrease again.
3
The yields of CH SBr and DMSO are plotted in Figure
3
5
, panels a and b, respectively, as a function of reaction
time and O partial pressure. The yields of CH SBr as
2
3
a function time are very similar for all of the O partial
2
pressures investigated, even in the “pure” N (Ͼ100
2
ppmV O ) experiments. The scatter in the values in
2
the initial stages of the reaction is caused by difficulty
in accurately determining the CH SBr concentration
3
Figure 5 Formation yields of CH SBr and DMSO as a
3
early in the reaction due to the relatively small change
in the DMS concentration. At longer reaction times,
function of reaction time and O partial pressure, panels a
and b, respectively.
2

8
90
MAURER ET AL.
DMS adduct, which can either (i) decompose back to
reactants, (ii) decompose to new products, or (iii) react
CH SCH radicals with O are responsible for the for-
3 2 2
mation of CH SBr since its yield would be expected
3
with molecules such as O to form products.
to dramatically decrease with increasing O partial
2
2
Extrapolation of the high-temperature (386–604
pressure and vary with the initial concentration of mo-
lecular bromine in the system. Neither of these effects
was observed. The concentration–time profile of
K) Arrhenius expression of Jefferson et al. (k ϭ (9.0
Ϫ
11
Ϯ 2.9) ϫ 10 exp((Ϫ2386 Ϯ 151)/T), [20]) for the
Ϫ
14
cm3
abstraction reaction gives a value of 3 ϫ 10
CH SBr is also more consistent with its production via
3
Ϫ
1
Ϫ1
molecule
s
for this channel at 298 K assuming
a primary reaction channel.
linearity at lower temperatures. Based on the kinetic
observations of the present work, this would imply a
possible contribution of ϳ30% from this direct ab-
straction reaction. Since the mechanism is in the cross-
over region between abstraction and addition at room
temperature and linearity of the Arrhenius expression
has only limited validity in this region, the estimation
for the contribution of the abstraction is an upper limit.
There are three possible decomposition channels
for the Br–DMS adduct:
The observation of the formation of DMSO would
be consistent with a reaction of O with a Br–DMS
2
adduct as has been proposed previously.
CH S(Br)CH ϩ O !: CH SOCH ϩ BrO
3
3
2
3
3
Although the product studies show a dependence
of the DMSO formation yield on the O partial pres-
2
sure, the kineticexperiments do not support any sys-
tematicdependen ce on the O
partial pressure. Fur-
2
thermore, in a study of the reaction of Br atoms with
DMS using time-resolved long-path absorption detec-
tion in the near-ultraviolet coupled with 248-nm laser
CH S(Br)CH !: CH SCH ϩ HBr
3
3
3
2
flash photolysis of CF Br/DMS/H /N /O of CF Br /
CH S(Br)CH !: CH Br ϩ CH S
3
2
2
2
2
2
3
3
3
3
DMS mixtures, Wine et al. [21] failed to detect BrO
Ϫ
16
radicals and from the results put a limit of Ͼ3 ϫ 10
CH S(Br)CH !: CH SBr ϩ CH
3
3
3
3
3
Ϫ
1
Ϫ1
cm molecule
s
on the reaction. Wine et al. also
estimated a lifetime of the (CH ) SBr adduct toward
3
2
The first channel is equivalent to a direct abstraction.
Although HBr is observed as a product, there are sev-
eral other channels for its formation, which are more
consistent with its measured yield. The further reac-
tions of CH SCH produced in the reaction are known
unimolecular decomposition of between 0.01 and
.001 s. Based on these kineticdata, in the present
0
investigations, a pronounced dependence should have
been observable on the overall rate constant for the
3
2
reaction of Br atoms with DMS on varying the O from
2
to lead to formation of SO , HCHO, and MSA. How-
2
16
18
Ϫ3
Ͼ1.5 ϫ 10 to 8.8 ϫ 10 molecules cm . The ab-
ever, this channel is considered negligible for reasons
that will be given later.
sence of any “O effect” suggests that the reaction of
2
the (CH ) SBr adduct with O is probably very much
3
2
2
In the second possible reaction channel of the Br–
Ϫ
16
lower than the limit of k Ͼ1 ϫ 10 given by Wine
et al. [21].
DMS adduct, CH Br is a product. Methyl bromide has
3
been searched for among the products but no evidence
could be found for its formation. Therefore, this chan-
nel is also considered to be negligible.
There is an alternative explanation for the forma-
tion of DMSO that is more consistent with both the
kinetic and product observations. The reactions of Br
atoms with peroxy radicals are reasonably fast. For
The observation of high yields of CH SBr supports
3
that the third possible pathway for the Br–DMS ad-
example, the reaction of Br atoms with HO has a rate
2
duct is probably the major fate of the adduct. CH SBr
3
Ϫ
12
3
Ϫ
1
Ϫ1
coefficient of 2.0 ϫ 10 cm molecule
s
at 298
could also be formed in further reactions of CH S rad-
3
K and is thought to result exclusively in the formation
icals, produced in the oxidation of CH SCH radicals,
3
2
of HBr and O [31]. A rate coefficient of k Ͻ 8.3 ϫ
2
for example:
Ϫ10
3
Ϫ
1
Ϫ1
1
0
cm molecule
s
[32] has been estimated for
the reaction of Br atoms with CH O . However, a
3
2
CH SCH ϩ O !: 2CH SCH O
Ϫ
13
cm3
3
2
2
3
2
much lower value of k ϭ (4.4 Ϯ 0.4) ϫ 10
!
: CH S ϩ HCHO
Ϫ
1
Ϫ1
3
molecule
s
has been measured for the reaction in
a recent determination using the discharge flow
method with detection by mass spectroscopy [33]. No
products were determined in the study. Cl atoms are
known to react with CH OO and C H OO radicals via
CH S ϩ Br !: CH SBr ϩ Br
3
2
3
CH S ϩ Br ϩ M !: CH SBr ϩ M
3
3
3
2
5
two exothermic ch annels [34–37]; for example, for
the methylperoxy radical
However, it is highly improbable that reactions of

Br-INITIATED OXIDATION OF DIMETHYL SULFIDE
891
CH OO ϩ Cl !: CH O ϩ ClO
tude or less. Therefore, it is to be expected that reaction
of Br with the various peroxy radicals can compete
effectively with the self- and cross-reactions of the
peroxy radicals. This would lead to production of BrO
radicals and HBr in the system.
3
3
CH OO ϩ Cl !: CH OO ϩ HCl
3
3
In the case of Br, on grounds of the thermochemistry
and experimental observations, it is currently thought
The 254-nm photolysis of the alkyl iodides methyl
that the channel producing CH O and BrO dominates
iodide and 2-iodobutane in the presence of O
used to produce CH and CH CH(O )CH CH
3
2
can be
3
[33,38].
3
O
2
3
2
2
rad-
If in the present Br /DMS/air reaction system it is
icals, respectively. Tests in this laboratory have shown
quite conclusively that, when irradiations are per-
2
assumed that the major product producing channel of
the initially formed Br–DMS is decomposition to
form CH SBr and CH ,
formed in the presence of Br
that photolyze bromine, the expected yields of CH
and CH CH(OH)CH CH from the peroxy radical
with additional lamps
2
OH
3
3
3
3
2
3
self-reactions are drastically reduced. Low concentra-
CH S(Br)CH !: CH SBr ϩ CH
3
3
3
3
tions of Br are sufficient to bring about the change.
2
Furthermore, in experiments with ppm levels of NO
then under the present experimental conditions, the
CH radicals would add O to form methylperoxy rad-
icals,
in the DMS/Br /air reaction mixture, the formation of
2
3
2
DMSO was completely suppressed. This can be un-
derstood again in terms of the very effective compe-
tition between NO and Br for the CH O radicals.
CH ϩ O !: CH OO
3
2
3
2
3
In the concentration–time profiles of the products
from the Br /DMS system shown in Figure 3, the con-
2
On the basis of the preceding discussion, both HO2
centration of CH OH remains very low throughout the
3
and CH OO radicals can, apart from the conventional
3
course of the whole reaction and that of CH OOH
3
self-reaction or reaction with HO radicals, also un-
2
shows delayed production, only becoming important
dergo reaction with Br atoms:
late in the oxidation process when Br is low. These
2
are indications that the expected reactions of CH O2
3
HO ϩ Br !: HBr ϩ O2
2
and HO are being suppressed.
2
Ϫ12
3
Ϫ
1
1
1
Ϫ
1
1
1
(
k ϭ 2.0 ϫ 10 cm molecule
s
[31])
[30])
[33])
[30])
The rate coefficient of the reaction of BrO radicals
with DMS has been measured several times (k ϭ 2.6
HO ϩ HO ϩ M !: H O ϩ M
Ϫ
13
2
2
2
2
ϫ 10 at 298 K) and is known to result in the for-
mation of DMSO in unit yield ([16] and references
therein). The reactions of BrO radicals with DMS,
formed in the Br ϩ CH O reaction, could then ac-
Ϫ12
3
Ϫ
Ϫ
(
k ϭ 1.6 ϫ 10 cm molecule
s
CH OO ϩ Br !: CH O ϩ BrO
3
3
3
2
Ϫ13
3
Ϫ
Ϫ
count for the formation of DMSO.
(
k ϭ 4.4 ϫ 10 cm molecule
CH OO ϩ CH OO !: CH O ϩ CH O ϩ O
2
s
CH SCH ϩ BrO !: CH SOCH ϩ Br
3
3
3
3
3
3
3
3
Ϫ13
3
Ϫ1
Ϫ1
(
k ϭ 3.7 ϫ 10 cm molecule
s
If we assume that decomposition of Br9CH SCH to
3
3
form CH SBr and CH is the major product producing
3
3
!
: CH OH ϩ HCHO ϩ O2
3
pathway for the adduct and that under O -rich condi-
2
tions CH is converted quantitatively to CH O and
3
3
CH OO ϩ HO !: CH OOH ϩ O
2
3
2
3
BrO, then the initial yields of CH SBr and DMSO
3
Ϫ12
3
Ϫ1
Ϫ1
(
k ϭ 5.2 ϫ 10 cm molecule
s
[30])
should be equal, at least in the early stages of the re-
action. Although the yields of DMSO and, in partic-
The relative importance of the individual reactions
will depend on not only the rate coefficients but also
the concentration of the Br and the peroxy radicals.
The steady-state concentration of Br atoms in such a
system, as monitored by the decay of an alkene with
known rate of reaction with Br atoms, is generally of
ular, CH SBr are subject to a large degree of uncer-
3
tainty, the initial yields are approximately 1:0.8
(CH SBr:DMSO) at high O partial pressures. The
3
2
yield of DMSO falls off with increasing reaction time.
This would be expected with the suggested mechanism
due to a decrease in production associated with the
decreasing Br atom steady-state concentration and
consequently BrO radical concentration. The yield of
9
Ϫ3
the order of ca. 10 Br atoms cm . Model simulations
of such systems show that the concentrations of both
HO and CH O will be of the same order of magni-
CH SBr also decreases, however; in this case it is due
2
3
2
3

8
92
MAURER ET AL.
to further oxidation. Tests have shown that DMSO is
further oxidized only very slowly in the system.
The yields of DMSO are only much lower at very
varying degrees, which also contribute to the decay of
DMS. The product data support that the reaction
mainly proceeds over the formation of an Br–DMS
adduct which, apart from re-forming the reactants, can
also decompose to form CH SBr and CH radicals.
low O (Fig. 5), that is, under conditions where loss
2
paths for CH radicals other than reaction with O be-
3
2
3
3
come important. Note that the yield of CH SBr, al-
Addition of O to CH produces CH O radicals that
3
2 3 3 2
though subject to some scatter, does not show any
in further reaction with Br atoms produce BrO radi-
cals. An analysis of the kinetic and product data gives
a lifetime of the Br–DMS adduct toward thermal de-
composition at 298 K of approximately 0.05 s.
variation with the O partial pressure, which is also in
2
line with the proposed mechanism.
The further reaction of CH O with O will yield
3
2
HCHO and HO , both of which can react with Br at-
oms to yield HBr. The reaction of HCHO will also
yield CO.
At 298 K the tropospheric Br atom concentration
2
7
Ϫ3
will be too low (ϳ10 cm , [2]) for the reaction of
Br with DMS to compete against the dominant day-
time sink, reaction with OH radicals.
CH O ϩ O !: HCHO ϩ HO
2
3
2
HO ϩ Br !: HBr ϩ O2
2
Financial support of this work by the EC (European Com-
mission) is gratefully acknowledged.
HCHO ϩ Br !: HBr ϩ HCO
HCO ϩ O !: CO ϩ HO2
2
Bibliography
In the early stages of the reaction the sum of ⌬CO and
⌬
HCHO are approximately equal to ⌬DMS/2. The
1
. Barrie, L. A.; Botenheim, J. W.; Schnell, R. C.; Crutzen,
P. J.; Rasmussen, R. A. Nature 1988, 334, 138.
yield of CO should be equal to or larger than that of
HBr, which is the case.
The mechanism can account for the observed var-
iations in the measured rate for Br ϩ DMS.
Assuming the simple mechanism
2
3
. Platt, U.; Janssen, C. Faraday Discuss 1995, 100, 175.
. Hausmann, M.; Platt, U. J Geophys Res 1994, 99,
2
5399.
. Platt, U.; Hausmann, M. Res Chem Intermed 1994, 20,
57.
4
5
CH SCH ϩ Br ϩ M !: CH S(Br)CH ϩ M, k
3
3
3
3
1
5. Barrie, L. A.; Li, S.-M.; Toom, D. L.; Landsberger, S.;
Sturges, W. J Geophys Res 1994, 99, 25453.
CH S(Br)CH ϩ M !: CH SCH ϩ Br ϩ M, k
Ϫ1
6. Li, S.-M.; Yokouchi, Y.; Barrie, L. A.; Muthuramu, K.;
Shepson, P. B.; Bottenheim, J. W.; Sturges, W. T.;
Landsberger, S. J Geophys Res 1994, 99, 25415.
3
3
3
3
CH S(Br)CH !: CH SBr ϩ CH , k
2
3
3
3
3
7
. Jobson, B. T.; Niki, H.; Yokouchi, Y.; Bottenheim, J.;
Hopper, F.; Leaitch, R. J Geophys Res 1994, 99, 25355.
. Yokouchi, Y.; Akimoto, H.; Barrie, L. A.; Bottenheim,
J. W.; Anlauf, K.; Jobson, B. T. J Geophys Res 1994,
the overall rate coefficient (koverall) for loss of DMS is
given by k k /(k ϩ k ). Since k ϽϽ k , the overall
8
1
2
Ϫ1
2
2
Ϫ1
rate coefficient measured by the relative method,
99, 25379.
neglecting possible secondary chemistry, will be k k .
eq
2
9. Mozurkewich, M. J Geophys Res 1995, 100, 14199.
Ϫ
15
3
Using a value of k ϭ 5.0 ϫ 10 cm from Wine et
eq
10. Vogt, R.; Crutzen, P. J.; Sander, R. Nature 1996, 383,
327.
11. Barone, S. B.; Andrew, A. A.; Ravishankara, A. R. Far-
aday Discuss 1995, 100, 39.
12. Turnipseed, A. A.; Ravishankara, A. R. The Atmo-
sphericOxidation of Dimethyl Sulfide: Elementary
Steos in a Complex Mechanism. Dimethyl Sulfide:
Oceans, Atmosphere and Climate, Belgirate, Italy,
Ϫ
13
3
Ϫ
1
Ϫ1
al. [21] and koverall Ϸ 1 ϫ 10
cm molecule
s
results in a value for the decomposition channel of the
adduct of k Ϸ 20 s , corresponding to a lifetime
toward unimolecular decomposition of approximately
Ϫ1
2
0
.05 s.
1
992, p. 185.
3. Tyndall, G. S.; Ravishankara, A. R. Int J Chem Kinet
991, 23, 483.
4. Yin, F.; Grosjean, D.; Seinfeld, J. H. J Atmos Chem
990, 11, 309.
CONCLUSIONS
1
1
1
1
The relative kineticte ch nique when applied for the
study of the reaction of Br atoms with DMS in a large-
volume reaction chamber does not result in a constant
value for the rate. The variability in the rate coefficient
has been attributed to the formation of BrO radicals in
1
5. Ferek, R. J.; Hobbs, P. V.; Radke, F.; Herring, J. A.;
Sturges, W. T.; Cota, G. F. J Geophys Res 1995, 100,
26093.

Br-INITIATED OXIDATION OF DIMETHYL SULFIDE
893
1
1
1
6. Bedjanian, Y.; Poulet, G.; Le Bras, G. Int J Chem Kinet
996, 28, 383.
7. Barnes, I.; Bastian, V.; Becker, K. H.; Overath, R. D.
Int J Chem Kinet 1991, 23, 579.
8. Barnes, I.; Becker, K. H.; Martin, D.; Carlier, P.; Mou-
vier, G.; Jourdain, J. L.; Laverdet, G.; Le Bras, G. In
ACS Symposium Series No. 393: BiogenicSulfur in the
Environment; Saltzman, E. S., Cooper, W. J., Eds.;
American Chemical Society, Washington, DC, 1989; pp
26. Hynes, A. J.; Stoker, R. B.; Pounds, A. J.; McKay, T.;
Bradshaw, J. D.; Nicovich, J. M.; Wine, P. H. J Phys
Chem 1995, 99, 16967.
27. Becker, K. H.; Nelsen, W.; Su, Y.; Wirtz, K. Chem Phys
Lett 1990, 168, 559.
1
2
2
3
8. Patroescu, I. V.; Barnes, I.; Becker, K. H. J Phys Chem
996, 100, 17207.
9. Winther, F.; Guarnieri, A.; Nielsen, O. F. Spectrochim
Acta 1975, 31A, 689.
0. Atkinson, R.; Baulch, D. L.; Cox, R. A.; Hampson, R.
F., Jr.; Kerr, J. A.; Troe, J. J Phys Chem Ref Data 1992,
1
4
64–476.
1
2
9. Toumi, R. Geophys Res Lett 1994, 21, 117.
0. Jefferson, A.; Nicovich, J. M.; Wine, P. H. J Phys Chem
21, 1125.
31. Laverdet, G.; Le Bras, G.; Mellouki, A.; Poulet, G.
1
994, 98, 7128.
Chem Phys Lett 1990, 172, 430.
2
1. Wine, P. H.; Nicovich, J. M.; Stickel, R. E.; Zhao, Z.;
Shackelford, C. J.; Kreutter, K. D.; Daykin, E. P.; Wang,
S. In The TroposphericChemistry of Ozone in the Polar
Regions: NATO ASI Series, Vol. 17; Niki, H., Becker,
K. H., Eds.; Springer-Verlag, Berlin, 1993; pp 385–395.
2. Barnes, I.; Becker, K. H.; Overath, R. D. In The Tro-
posphericChemistry of Ozone in the Polar Regions:
NATO ASI Series, Vol. 17; Niki, H., Becker, K. H.,
Eds.; Springer-Verlag, Berlin, 1993; pp 371–383.
3. Barnes, I.; Becker, K. H.; Mihalopoulos, N. J Atmos
Chem 1994, 18, 267.
3
3
2. Kondo, O.; Bensen, S. W. J Phys Chem 1984, 88, 6675.
3. Aranda, A.; Laverdet, G.; Le Bras, G.; Poulet, G. J Chim
Phys, 1998, 95, 963.
3
3
4. Da e¨ le, V.; Poulet, G. J Chim Phys 1996, 93, 1081.
5. DeMore, W. B.; Sander, S. P.; Golden, D. M.; Hamp-
son, R. F.; Kurylo, M. J.; Howard, C. J.; Ravishankara,
A. R.; Kolb, C. E.; Molina, M. J. Chemical Kinetics and
Photochemical Data for Use in Stratipheric Modeling,
Evaluation Number 11, JPL Publication 94-26, 1994.
6. Maricq, M. M.; Szente, J. J.; Kaiser, E. W.; Shi, J.
J Phys Chem 1994, 98, 2083.
2
2
2
2
3
4. Barnes, I.; Bastian, V.; Becker, K. H.; Overath, R.; Zhu,
T. Int J Chem Kinet 1989, 21, 1069.
5. Barnes, I.; Bastian, V.; Becker, K. H.; Overath, R.;
Tong, Z. Int J Chem Kinet 1989, 21, 499.
37. Jungkamp, T. P. W.; Kukui, A.; Schindler, R. N. Ber
Bunsenges Phys Chem 1995, 99, 1057.
38. Aranda, A.; Da e¨ le, V.; Le Bras, G.; Poulet, G. Int J
Chem Kinet 1998, 30, 249.