Atmospheric chemistry of 1,4-dioxane. Laboratory studies

Article abstract of DOI:10.1039/a700598i

A pulse radiolysis technique was used to measure the UV absorption spectra of c-C₄H₇O₂ and (c-C₄H₇O₂)O₂ radicals over the range 220-320 nm, σ(c-C₄H₇O₂)_{250 nm} = (5.9 ± 0.6) × 10^-18 and σ[(c-C₄H₇O₂)O₂]_{240 nm} = (4.8 × 0.8) × 10^-18 cm² molecule^-1. The observed self-reaction rate constants for the c-C₄H₇O₂ and (c-C₄H₇O₂)O₂ radicals, defined as -d[c-C₄H₇O₂]/dt = 2k₄[c-C₄H₇O₂]² and -d[(c-C₄H₇O₂)O₂]/dt = 2k_{5 obs[}(c-C₄H₇O₂)O₂]² were k₄ = (3.3 ± 0.4) × 10^-11 and k_{5 obs} = (7.3 ± 1.2) × 10^-12 cm³ molecule^-1 s^-1. The rate constants for reactions of (c-C₄H₇O₂)O₂ radicals with NO and NO₂ were k₆ (1.2 ± 0.3) × 10^-11 and k₇ = (1.3 ± 0.3) × 10^-11 cm³ molecule^-1 s^-1, respectively. The rate constants for the reaction of F atoms with 1,4-dioxane and the reaction of c-C₄H₇O₂ radicals with O₂, were k₃ = (2.4 ± 0.7) × 10^-10 and k₂ = (8.8 ± 0.9) × 10^-12 cm³ molecule^-1 s^-1, respectively. A relative rate technique was used to measure the rate constant for the reaction of Cl atoms with 1,4-dioxane, k₁₇ = (2.0 ± 0.3) × 10^-10 cm³ molecule^-1 s^-1. A long-pathlength FTIR spectrometer coupled to a smog chamber system was used to show that the sole atmospheric fate of the alkoxy radical (c-C₄H₇O₂)O is decomposition via C-C bond scission leading to the formation of H(O)COCH₂CH₂OC(O)H (ethylene glycol diformate).

Full text of DOI:10.1039/a700598i

View Article Online / Journal Homepage / Table of Contents for this issue
Atmospheric chemistry of 1,4-dioxane
Laboratory studies
J. Platz,a J. Sehested,a T. MÊgelberg,a O. J. Nielsena* and T. J. Wallingtonb*
a Section for Chemical Reactivity, Environmental Science and T echnology Department,
RisÏ National L aboratory, DK-4000, Roskilde, Denmark
b Ford Motor Company, 20000 Rotunda Drive, Mail Drop SRL -3083, Dearborn,
Michigan 48121-2053, USA
A pulse radiolysis technique was used to measure the UV absorption spectra of c-C H O and (c-C H O )O radicals over the
4
7
2
4
7
2
2
range 220È320 nm, p(c-C H O )
2 250 nm
The observed self-reaction rate constants for the c-C H O and (c-C H O )O radicals, deÐned as -d[c-C H O ]/dt \ 2k [c-
\ (5.9 ^ 0.6) ] 10~18 and p[(c-C H O )O ]
\ (4.8 ] 0.8) ] 10~18 cm2 molecule~1.
4
7
4 7 2
2 240 nm
4
7
2
4
7
2
2
4
7
2
4
C H O ]2 and -d[(c-C H O )O ]/dt \ 2k
[(c-C H O )O ]2 were k \ (3.3 ^ 0.4) ] 10~11 and k
\ (7.3 ^ 1.2) ] 10~12
4
7
2
4
7
2
2
5 obs
4
7
2
2
4
4
5 obs
cm3 molecule~1 s~1. The rate constants for reactions of (c-C H O )O radicals with NO and NO were k (1.2 ^ 0.3) ] 10~11
7
2
2
2
6
and k \ (1.3 ^ 0.3) ] 10~11 cm3 molecule~1 s~1, respectively. The rate constants for the reaction of F atoms with 1,4-dioxane
7
and the reaction of c-C H O radicals with O , were k \ (2.4 ^ 0.7) ] 10~10 and k \ (8.8 ^ 0.9) ] 10~12 cm3 molecule~1 s~1,
4
7
2
2
3
2
respectively. A relative rate technique was used to measure the rate constant for the reaction of Cl atoms with 1,4-dioxane,
k
\ (2.0 ^ 0.3) ] 10~10 cm3 molecule~1 s~1. A long-pathlength FTIR spectrometer coupled to a smog chamber system was
17
used to show that the sole atmospheric fate of the alkoxy radical (c-C H O )O is decomposition via CwC bond scission leading
to the formation of H(O)COCH CH OC(O)H (ethylene glycol diformate).
4
7
2
2
2
1
Introduction
Ethers are an important class of automotive fuel additives.
For example, methyl tert-butyl ether (MTBE), ethyl tert-butyl
ether (ETBE), tert-amyl methyl ether (TAME), and diiso-
propyl ether (DIPE) are widely used octane-enhancing addi-
tives in gasoline. In addition, ethers such as dimethyl ether
and dimethoxy methane are under consideration as alterna-
tive diesel fuels.1 To predict the environmental impact of ether
fuels and fuel additives it is desirable to understand how the
structure of such molecules impacts their atmospheric reacti-
vity. As part of a collaborative study of the atmospheric chem-
istry of automotive fuel additives we have undertaken an
investigation of 1,4-dioxane. This compound is a cyclic ether
FTIR studies were performed to determine the fate of the
alkoxy radical formed in reaction (6a).
2
Experimental
with four equivalent CH groups a to the oxygen ether
2
linkage. A study of the chemistry of 1,4-dioxane provides
Two di†erent experimental systems were used, both have been
described previously2,3 and are only discussed brieÑy here.
insight into the chemical behaviour of peroxy and alkoxy rad-
icals formed a to the oxygen ether linkage.
The atmospheric degradation of 1,4-dioxane is initiated by
reaction of OH radicals to produce alkyl radicals which react
2.1 Pulse radiolysis systems
with O to give alkylperoxy radicals:
c-C H O radicals were generated from 1,4-dioxane by the
2
4
7 2
pulsed radiolysis of SF È1,4-dioxane gas mixtures in a 1 l
6
stainless-steel reaction cell with a 30 ns pulse of 2 MeV elec-
trons from a Febetron 705 Ðeld emission accelerator. SF was
6
always in great excess and was used to generate Ñuorine
atoms:
We have used pulse radiolysis coupled with time-resolved
UVÈVIS absorption spectroscopy, to determine the UV
absorption spectra of the alkyl and alkyl peroxy radicals
derived from 1,4-dioxane, and the kinetics of reactions (2)È(7).
The radiolysis dose was varied by insertion of stainless-steel
attenuators between the accelerator and the reaction cell. In
this work we refer to the radiolysis dose as a fraction of the
maximum dose that is achievable. The Ñuorine atom yield was
calibrated by monitoring the transient absorption at 260 nm
by CH O radicals produced by pulse radiolysis of
3
2
J. Chem. Soc., Faraday T rans., 1997, 93(16), 2855È2863
2855

View Article Online
SF ÈCH ÈO mixtures, as described previously.4 Using
mixtures of 8È12 mTorr of 1,4-dioxane, 100È200 mTorr of
Cl , 5.2È700 Torr of O and 0È16 mTorr of NO in 700 Torr
6
4
2
p(CH O ) \ 3.18 ] 10~18 cm2 molecule~1 5 the F atom
3
2
2
2
yield was determined to be (3.16 ^ 0.35) ] 1015 molecules
total pressure with N diluent at 296 K (760 Torr \ 1013
2
cm~3 at full radiolysis dose and 1000 mbar of SF . The
mbar \ 101 kPa). Loss of reactants and the formation of pro-
6
quoted error includes 10% uncertainty in p(CH O ) and two
ducts were monitored by FTIR spectroscopy, using an analys-
ing pathlength of 27 m and a resolution of 0.25 cm~1. IR
spectra were derived from 32 co-added spectra. Reactants
were monitored using their characteristic features over the
wavenumber range 800È2000 cm~1.
3
2
standard deviations from the experimental absorption mea-
surements; unless otherwise stated, all uncertainties reported
in this paper are two standard deviations. Standard error pro-
pagation methods were used where appropriate.
The analysis light was provided by a 150 W pulsed Xenon
arc lamp and was multi passed through the reaction cell using
internal White cell optics to give total optical pathlengths of
80 or 120 cm. After leaving the cell the light was guided
through a monochromator and detected with either a photo-
multiplier (to record absorption transients) or a diode array
(to record absorption spectra). All absorption transients were
derived from single-pulse experiments. The spectral resolution
was 0.8 nm with the photomultiplier detector and 3.2 nm with
the diode array.
3
Results and Discussion
3.1 Reaction of F atoms with 1,4-dioxane
To measure k , the maximum transient absorbance at 260 nm
3
was measured following the radiolysis of SF È1,4-
6
dioxaneÈCH mixtures. The radiolysis dose (52.7% of full
4
dose) and the total concentration (1000 mbar) were Ðxed. The
concentrations of 1,4-dioxane and CH were varied over the
ranges 0È4.6 mbar and 0È40.0 mbar, respectively. Fig. 1 shows
the observed maximum absorbance as a function of the con-
centration ratio [CH ]/[1,4-dioxane]. In the reaction system
4
Reagent concentrations used were: SF , 976È997 mbar;
6
1,4-dioxane, 1È5 mbar; O , 1.7È20 mbar; NO , 0.31È1.29
2
2
mbar; and NO, 0.25È1.21 mbar. All experiments were per-
4
reactions (3) and (9) compete for the available F atoms:
formed at 296 K and 1000 mbar total pressure. Chemicals
were supplied by: SF (99.9%) Gerling and Holz; 1,4-dioxane
6
([99%) E. Merck; O , (ultrahigh purity) LÏAir Liquide; NO
2
2
([98%) Linde Technische Gase; and NO (99.8%) Messer
Griesheim. 1,4-Dioxane was distilled over CaH before use.
All other chemicals were used as received.
2
Six sets of experiments were performed using the pulse-
radiolysis system. First, the rate constant for reaction of F
atoms with 1,4-dioxane was determined using the pulse
radiolysis of SF È1,4-dioxaneÈCH mixtures. Second, the UV
CH radicals do not absorb signiÐcantly at 260 nm. In con-
trast the alkyl radicals derived from 1,4-dioxane absorb appre-
ciably at 260 nm (see Section 3.3). As shown in Fig. 1, as the
3
6
4
absorption spectra of the alkyl and alkyl peroxy radicals
derived from 1,4-dioxane were obtained by using the diode
array camera to capture the UV absorption following
radiolysis of SF È1,4-dioxane and SF È1,4-dioxaneÈO mix-
concentration ratio [CH ]/[1,4-dioxane] increases the
4
maximum absorbance decreases because a greater fraction of
the F atoms react with CH to give CH radicals. The solid
4
3
line in Fig. 1 is a three-parameter Ðt of the following expres-
sion to the experimental data:
6
6
2
tures. Third, the rate constants for reactions (4) and (5) were
determined from the rate of decay of the absorption at 250
and 240 nm, following the radiolysis of SF È1,4-dioxane and
A ] ACH3(k /k )[CH ]/[1,4-dioxane]
6
R
9
3
4
A
\
(I)
SF È1,4-dioxaneÈO mixtures, respectively:
max
1 ] (k /k )[CH ]/[1,4-dioxane]
6
2
9
3
4
where A
max
is the observed maximum absorbance, A is the
R
maximum absorbance of alkyl radicals expected if only alkyl
radicals from 1,4-dioxane were produced and A
is the
CH3
maximum absorbance expected if only CH radicals were pro-
3
duced. We performed model calculations with and without
radicalÈradical reactions that showed an e†ect on A
of the
max
Fourth, the rate constant for the association reaction between
and the alkyl radical derived from 1,4-dioxane was mea-
O
2
sured by monitoring the loss of alkyl radicals in the presence
of O . Fifth, the rate of NO formation following radiolysis of
2
2
SF È1,4-dioxaneÈO ÈNO mixtures was used to measure k .
6
2
6
Sixth, k was determined by monitoring the rate of NO loss
following radiolysis of SF È1,4-dioxaneÈO ÈNO mixtures:
7
2
6
2
2
2.2 FTIR-smog chamber system
Fig. 1 Plot of the maximum absorbance observed at 260 nm follow-
ing the radiolysis of 1,4-dioxaneÈCH ÈSF mixtures vs. the concentra-
The FTIR system was interfaced to a 140 l Pyrex reactor.
Radicals were generated by UV irradiation (22 blacklamps) of
4
6
tion ratio [CH ]/[1,4-dioxane]
4
2856
J. Chem. Soc., Faraday T rans., 1997, V ol. 93

View Article Online
radicalÈradical reactions of \2.3%. The 2.3% is well within
under conditions where 100% of the F atoms are converted
into alkyl radicals, it is necessary to consider potential inter-
fering radicalÈradical reactions such as (4) and (10):
the reported uncertainty. The parameters A , A
and k /k
R
CH3
9
3
were varied simultaneously, the best Ðt was obtained with
k /k \ 0.28 ^ 0.06, A \ 0.27 ^ 0.01 and
A
CH3 \ 0.009
9
3
R
^ 0.009. Using k \ (6.8 ^ 1.4) ] 10~11 cm3 molecule~1 s~1
9
(ref. 6) gives k \ (2.43 ^ 0.71) ] 10~10 cm3 molecule~1 s~1.
3
The A
value is not statistically di†erent from zero. It
should be noted that the CH radical does not absorb at 260
nm. However, the 0.02% O in SF could give rise to a small
CH3
3
2
6
absorption from CH O at 260 nm.
3
2
To check for such complications, experiments were performed
3.2 Spectrum of alkyl radicals derived from 1,4-dioxane
using mixtures of 997 mbar of SF and 3 mbar of 1,4-dioxane,
6
with the maximum transient absorption at 250 nm measured
Following the pulse radiolysis of SF È1,4-dioxane mixtures, a
6
as a function of radiolysis dose. Fig. 3(A) shows the observed
maximum absorbance from experimental transients ascribed
to alkyl radicals as a function of the dose. As seen from Fig.
3(A), the absorption observed in experiments using maximum
dose was signiÐcantly less than that expected based upon a
linear extrapolation of the low-dose data. This observation
suggests that unwanted radicalÈradical reactions such as reac-
tions (4) and (10) become important at high radical concentra-
tions.
rapid (complete within 0.5È1.5 ls) increase in absorption was
observed at 250 nm, followed by a slower decay. Fig. 2(A)
shows the transient absorption at 250 nm following the
radiolysis of a mixture of 997 mbar SF and 3 mbar 1,4-
6
dioxane. No absorption was observed when either 997 mbar
SF or 3 mbar 1,4-dioxane was radiolysed separately. We
6
ascribe the absorption shown in Fig. 2(A) to the formation of
alkyl radicals and their subsequent loss by self-reaction. In
this work we assume that F atoms react exclusively with 1,4-
dioxane by H atom abstraction to give alkyl radicals. To work
The line through the experimental data in Fig. 3(A) is a
linear least-squares Ðt to the low-dose data and has a slope of
0.964 ^ 0.035. Combining this slope with the F atom yield of
(3.16 ^ 0.35) ] 1015 molecules cm~3 (full dose and [SF ] \
6
1000 mbar), we obtain p
(alkyl) \ (5.85 ^ 0.58) ] 10~18
250 nm
cm2 molecule~1. The quoted uncertainty includes both sta-
tistical and potential systematic uncertainties and so reÑects
the accuracy of the measurement.
The spectrum of the alkyl radical was measured by record-
ing the initial absorbance following the pulsed radiolysis of
SF È1,4-dioxane mixtures using a diode array. The delay was
6
1 ls, the integration time was 2 ls, and the spectral resolution
was 3.2 nm. The initial absorbances were scaled to that at 250
nm and placed upon an absolute basis using p
\ 5.85
250 nm
] 10~18 cm2 molecule~1. The results are given in Table 1 and
Fig. 4(A). Nese et al.7 have studied the UV spectrum of the
c-C H O radical in aqueous solutions. The gas-phase spec-
4
7 2
trum measured in the present work appears to be indistin-
guishable from that reported by Nese et al. which was
obtained by pulse radiolysis of N O-saturated aqueous solu-
2
tions of 1,4-dioxane.7 For comparison the UV absorption
Fig. 2 (A) Transient absorbance at 250 nm following the pulsed
Fig. 3 (A) Maximum transient absorbance at 250 nm following the
radiolysis (maximum dose) of a mixture of 3 mbar 1,4-dioxane and
pulsed radiolysis of mixtures of 3 mbar 1,4-dioxane and 997 mbar SF
6
997 mbar SF . The UV pathlength was 120 cm. (B) Transient absorb-
vs. the radiolysis dose. The UV pathlength was 120 cm. (B) Maximum
6
ance at 240 nm following the pulsed radiolysis (maximum dose) of a
transient absorbance at 240 nm following the pulsed radiolysis of mix-
mixture of 3 mbar 1,4-dioxane, 20 mbar O and 977 mbar SF . The
UV pathlength was 120 cm. The smooth lines are second-order decay
Ðts to the experimental data.
tures of 3 mbar 1,4-dioxane, 20 mbar O , and 997 mbar SF vs. the
radiolysis dose. The UV pathlength was 120 cm. The solid lines in A
and B are linear regressions of the low-dose data (Ðlled circles).
2
6
2
6
J. Chem. Soc., Faraday T rans., 1997, V ol. 93
2857

View Article Online
Table 1 Absorption cross-sections measured in this work
wavelength
/nm
p(c-C H O )
p[(c-C H O )O ]
4
7
2
4
7
2
2
/10~20 cm2 molecule~1
/10~20 cm2 molecule~1
220
230
240
250
260
270
280
290
300
310
320
295
415
541
585
538
415
272
163
101
63
236
422
480
411
324
245
194
158
124
96
44
65
spectrum of the cyclohexyl radical is also given in Fig. 4(A).8
As expected for such structurally similar species, the absorp-
tion spectra of the alkyl radicals derived from 1,4-dioxane and
cyclohexane are similar in both intensity and shape.
3.3 Kinetic study of the self-reaction of c-C H O radicals
4
7 2
Fig. 2(A) shows a typical absorption transient obtained by
pulse radiolysis of a mixture of 997 mbar SF and 3 mbar 1,4-
6
dioxane. The trace shows a rapid increase in absorption at 250
nm followed by a slower decay. We ascribe the decay to the
self-reaction of the alkyl radical, via reaction (4):
Fig. 5 (A) Plot of the reciprocal half-life for self-reaction of
The decays were observed to follow second-order kinetics and
c-C H O radicals as a function of the maximum absorbance at
were Ðtted with the expression A(t) \ (A [ A )/[1 ] 2k(A
4 7 2
250 nm. (B) Plot of the reciprocal half-life for the self-reaction of
(c-C H O )O radicals as a function of maximum absorbance at
0
inf
0
[ A )t] ] A , where A(t) is the measured absorbance at
inf
inf
4
7
2
2
time t, A and A are the absorbance extrapolated to time
240 nm.
0
inf
zero and inÐnite time, respectively. k \ (k ] ln 10)/
[p(alkyl)L ] where L is the optical pathlength. A Ðt using this
4
second-order decay expression is shown in Fig. 2(A). No
residual absorption was observed, indicating the absence of
any long-lived products which absorb signiÐcantly at 250 nm.
where p(alkyl) is the absorption cross-section of the c-C H O
4
7 2
radical, and L is 120 cm. The intercept of the linear regression
of the data is (0.99 ^ 1.30) ] 104 s~1 and is not signiÐcantly
di†erent from zero. The slope is (2.18 ^ 0.39) ] 105 s~1 and
can be combined with p(alkyl) \ (5.82 ^ 0.58) ] 10~18 cm2
The reciprocal half-lives, 1/t
(derived from the Ðts), are
1@2
plotted as a function of A in Fig. 5(A). The alkyl radical con-
0
centration was varied by varying the initial F atom concentra-
molecule~1 to give k \ (3.32 ^ 0.39) ] 10~11 cm3
tion. A linear least-squares Ðt of the data in Fig. 5(A) gives a
4
molecule~1 s~1. The quoted error includes statistical uncer-
slope of (2.18 ] 0.39) ] 105 s~1 \ (k ] 2 ln 10)/[p(alkyl)L ],
4
tainties in the slope in Fig. 5(A) and the absorption cross-
section of the alkyl radical. The value of k is similar to the
4
rate constants reported for the self-reaction of alkyl radicals
derived from MTBE, (2.5 ^ 1.1) ] 10~11,9 and di-tert-butyl
ether, (2.9 ^ 0.2) ] 10~11 cm3 molecule~1 s~1.10
3.4 Association reaction between O and the alkyl radical
2
c-C H O
4
7 2
The kinetics of reaction (2) were studied by monitoring the
absorbance at 260 nm following pulse radiolysis of mixtures of
500È995 mbar of SF , 3.0 mbar of 1,4-dioxane and 1.7È10.1
6
mbar of O . Using k \ 2.4 ] 10~10 cm3 molecule~1 s~1 (see
2
3
Section 3.1) it follows that in the presence of 3 mbar of 1,4-
dioxane the lifetime of F atoms with respect to conversion
into c-C H O radicals is 0.06 ls. Assuming that c-C H O
4
7
2
4 7 2
radicals behave like other alkyl radicals, and add O , with a
2
rate constant of the order 10~12È10~11 cm3 molecule~1 s~1,
the lifetime of c-C H O radicals with respect to conversion
4
2
7 2
into (c-C H O )O radicals will be of the order of 1 ls.
4
7
2
Consistent with expectations, the pulsed radiolysis of SF È1,4-
6
dioxaneÈO mixtures produced absorption transients which
2
consisted of an initial rapid increase in absorption (complete
Fig. 4 (A) UV absorption spectra of the alkyl radicals derived from
1,4-dioxane and cyclohexane.8 (B) UV absorption spectra of the
peroxy radicals derived from 1,4-dioxane and cyclohexane.14
within 0.2 ls) followed by a rapid decay which occurred on a
timescale of 1È2 ls and followed Ðrst-order kinetics. The insert
2858
J. Chem. Soc., Faraday T rans., 1997, V ol. 93

View Article Online
in Fig. 6 shows a typical absorption transient. A Ðrst-order
decay expression Ðt to the transients at 0.2È5.0 ls to give
pseudo-Ðrst-order decay rates. As seen from Fig. 6, the
In Section 3.1 we reported k \ (2.43 ^ 0.71) ] 10~10 cm3
3
molecule~1 s~1. Using
k
\ (1.9 ^ 0.3) ] 10~13 cm3
13
molecule~1 s~1,12 we calculate that 0.5% of the F atoms are
pseudo-Ðrst-order rate, k1st, increased linearly with the O
converted into FO radicals and 99.5% into alkyl peroxy rad-
2
2
concentration. It seems reasonable to ascribe the decay of
icals. Corrections were computed for the presence of FO rad-
2
icals using literature data for the UV spectrum of FO
2
absorption to loss of c-C H O radicals via reaction with O .
4
7
2
2
Experiments were performed in the presence of either 500 or
radicals.13 The corrections were small (\2.5%).
1000 mbar of SF diluent. As seen from Fig. 6, there was no
There are no literature data concerning the kinetics of reac-
tions (10), (11) and (12) and we cannot calculate their impor-
tance. To check for these unwanted radicalÈradical reactions,
the transient absorbance at 240 nm was measured using mix-
tures of 977 mbar SF , 3 mbar 1,4-dioxane, and 20 mbar O
6
discernible e†ect of total pressure suggesting that reaction (2)
is at, or near, the high-pressure limit in 500È1000 mbar of
SF . This behaviour is consistent with the available database
6
for the addition of O to large ([C ) alkyl radicals.11 Linear
2
3
6
2
least-squares analysis of the data in Fig. 6 gives k \ (8.8
with the radiolysis dose varied over an order of magnitude.
The UV pathlength was 120 cm. Fig. 3(B) shows the observed
maximum absorbance of the experimental transients at 240
nm as a function of dose. As seen from Fig. 3(B), in experi-
ments employing doses of 0.5 or greater the transient absorb-
ance falls below that expected from a linear extrapolation of
the low-dose data. We ascribe this to incomplete conversion
of F atoms into alkyl peroxy radicals caused by secondary
radicalÈradical reactions (4), (10), (11) and (12) at high radical
concentrations. The solid line drawn through the data in Fig.
3(B) is a linear least-squares Ðt of the low-dose data. The slope
is (0.79 ^ 0.03). From this value and the yield of F atoms of
2
^ 0.9) ] 10~12 cm3 molecule~1 s~1. The y-axis intercept is
(1.5 ^ 1.5) ] 105 s~1 and is not signiÐcant.
3.5 Spectrum of alkyl peroxy radicals derived from
1,4-dioxane
Following the pulse radiolysis of mixtures of 977 mbar SF , 3
6
mbar 1,4-dioxane and 20 mbar O , a rapid increase (complete
2
with 0.5È2.0 ls) in the UV absorbance was observed, followed
by a slower decay. An example is shown in Fig. 2(B). We
ascribe this UV absorbance to the formation of the peroxy
radical via reaction (2). To work under conditions where a
known fraction of the F atoms are converted into alkyl peroxy
radicals, it is necessary to consider potentially interferent sec-
ondary chemistry. Potential complications include unwanted
radicalÈradical reactions, such as reactions (4), (10), (11) and
(12):
(3.16 ^ 0.35) ] 1015 molecule cm~3 (full dose and [SF ] \
6
1000 mbar) we derive p(240 nm) \ (4.80 ^ 0.72) ] 10~18 cm2
molecule~1. The quoted uncertainty includes both statistical
and potential systematic uncertainties and so reÑects the accu-
racy of the measurement.
To map out the spectrum of the alkyl peroxy radical, the
maximum transient absorbance between 220 and 320 nm, fol-
lowing the pulsed radiolysis of SF È1,4-dioxaneÈO mixtures,
6
2
was recorded using a diode array. The delay was 1 ls, the
integration time was 2 ls, and the spectral resolution was 3.2
nm. The initial absorbances were scaled to that at 240 nm and
placed on an absolute basis using p(240 nm) \ 4.80 ] 10~18
cm2 molecule~1. Results are given in Table 1 and Fig. 4(B).
For comparison the UV spectrum of cyclohexyl peroxy rad-
icals, reported by Rowley et al.,14 is also given in Fig. 4(B).
The UV absorption spectra of alkyl peroxy radicals typically
consist of a broad featureless band in the region 200È320 nm
with a maximum absorption cross-section at ca. 240È250 nm
of 4È5 ] 10~18 cm2 molecule~1.14 Inspection of Fig. 4(B)
shows that the UV spectra of the peroxy radicals derived from
1,4-dioxane and cyclohexane are typical for this class of
radical species.
and loss of F atoms by reaction with molecular oxygen, reac-
tion (13):
3.6 Kinetic study of the self-reaction of (c-C H O )O
2
4
7
2
radicals
Following the pulse radiolysis of SF È1,4-dioxaneÈO mix-
6
2
tures a fast increase in the absorbance at 250 nm was observed
followed by a slower decay [see Fig. 2(B)]. We ascribe the
decay of the absorption to loss of alkyl peroxy radicals via
reaction (5):
The decay followed second order decay kinetics and was Ðtted
with the expression A(t) \ (A [ A )/[1 ] 2k
(A [ A )t]
0
inf
5 obs 0 in
] A ; where A(t) is the measured absorbance at time t, A is
inf
0
the absorbance at time zero, A is the absorbance at inÐnite
inf
time, and k
is the observed second-order rate constant for
5 obs
reaction (5). From the Ðt, the half-life (t ) for the self-reaction
1@2
Fig. 6 Pseudo-Ðrst-order rate constants for the formation of
was calculated. Experiments were performed using mixtures of
(c-C H O )O radicals as a function of the O concentration in 500
977 mbar SF , 3 mbar 1,4-dioxane, and 20 mbar O . The
4
7
2
2
2
(L) and 1000 (…) mbar of SF diluent. The insert shows an experi-
6
2
6
initial alkyl peroxy radical concentration was varied by
varying the radiolysis dose. Fig. 5(B) shows the reciprocal
half-lives (1/t ) plotted as a function of A , where A is the
mental transient obtained using a mixture of 3.4 mbar 1,4-dioxane,
7.7 mbar of O , 989 mbar of SF . The UV pathlength was 80 cm
2
6
and the radiolysis dose was 42% of maximum.
1@2
0
0
J. Chem. Soc., Faraday T rans., 1997, V ol. 93
2859

View Article Online
extrapolated absorbance of the alkyl peroxy radical at t \ 0.
A linear least-squares Ðt gives a slope of (5.95 ^ 0.42) ] 104
s~1 \ (k
5 obs
absorption cross-section of the (c-C H O )O radical, and
] 2 ln 10)/(p
L ), where p
is the
alkyl peroxy
L \ 120 cm. Hence, we derive k
alkyl peroxy
4
7
2
2
\ (7.3 ^ 1.2) ] 10~12 cm3
5 obs
molecule~1 s~1. The quoted error includes statistical uncer-
tainties from analysis of the data in Fig. 5(B) and the absorp-
tion cross-section of the (c-C H O )O radical. The intercept
4
7
2
2
of the linear regression of the data is (1.3 ^ 4.2) ] 103 s~1 and
is not signiÐcant. As shown in Section 3.10, reaction (5) pro-
duces the alkoxy radical which undergoes ring opening
leading to the formation of the peroxy radical
H(O)COCH CH OCH O . In the absence of information
2
2
2 2
concerning the UV spectrum and reactivity of
H(O)COCH CH OCH O radicals we are unable to correct
2
2
2
2
our measured value of k
for complications caused by for-
5 obs
mation and reactions involving H(O)COCH CH OCH O
2 2
2 2
radicals. Our reported value of k
stant which describes the rate of decay of absorption at 250
nm in the present chemical system.
is the observed rate con-
5 obs
3.7 Kinetics of the reaction between (c-C H O )O radicals
4
7
2
2
and NO
The kinetics of reaction (6) were studied by monitoring the
increase in absorbance at 400 nm following the radiolysis of
mixtures of 3 mbar 1,4-dioxane, 20 mbar O , 0.25È1.21 mbar
2
NO, in 1000 mbar of SF diluent. Fig. 7(A) shows the absorb-
6
ance as a function of time after the pulsed radiolysis of a
mixture containing 0.25 mbar of NO. NO absorbs strongly
2
at 400 nm and we ascribe the increased absorbance to NO
formation via reaction (6a).
2
Fig. 7 (A) Transient absorbance at 400 nm observed following
pulsed radiolysis (dose \ 42% of maximum) of a mixture of 0.60 mbar
NO, 3 mbar 1,4-dioxane, 20 mbar O and 977 mbar SF . The three
2
6
Ðts are from three simulations using three di†erent values for k .
This method of measuring the kinetics of the reactions of alkyl
peroxy radicals with NO has been used previously in our
laboratory and is discussed in detail elsewhere.16
6
Short dash, k \ 1.5 ] 10~11; full line, k \ 1.2 ] 10~11; long dash,
6
6
k \ 0.9 ] 10~11 (see text for details). (B) Transient absorbance at 400
6
nm observed following pulsed radiolysis (dose \ 42% of maximum) of
The maximum absorbances due to NO , measured from
a mixture of 0.314 mbar NO , 3 mbar 1,4-dioxane, 20 mbar O and
2
2
2
the experimental transients, were 90È130% of those expected
977 mbar SF .
6
if all the alkyl peroxy radicals reacted via reaction (6a). The
NO yields displayed no systematic trend when compared to
2
the NO concentration. The observation of NO yields can be
2
explained by the rapid decomposition of the alkoxy radical
shows the results of simulations using the chemical mecha-
formed in reaction (6a) by CwC bond scission, as discussed in
Section 3.10. Decomposition of the alkoxy radical via CwC
bond scission gives H(O)COCH CH OCH radicals and by
nism with k \ 0.9, 1.2, 1.5 ] 10~11 cm3 molecule~1 s~1. All
6
experimental absorption transients were well Ðtted using k \
6
1.2 ] 10~11 cm3 molecule~1 s~1. From the sensitivity of the
Ðt of the simulated to the experimental absorption transient
2
2
2
reaction with O , H(O)COCH CH OCH O radicals, which
2
2
2
2 2
react with NO to give more NO .
[as shown in Fig. 7(A)] we estimate the uncertainty on k to
2
To determine the rate constant for reaction (6) the experi-
mental transients were modelled using the CHEMSIMUL
6
be ^0.3 ] 10~11 cm3 molecule~1 s~1 and choose to report
k \ (1.2 ^ 0.3) ] 10~11 cm3 molecule~1 s~1.
computer modelling program17 with
consisting of reactions (2), (3), (4), (5), (6), (7), (20),
(22), (23), HC(O)OCH CH OCH O ] NO and
a
mechanism
6
As a check of the kinetic model used above we varied
k[HC(O)OCH CH OCH O ] NO] in the range (2È
2
2
2 2
12) ] 10~12 cm3 molecule~1 s~1. Satisfactoring Ðts were
only obtained using k[HC(O)OCH CH OCH O ] NO] \
2
2
2
2
2
HC(O)OCH CH OCH O ] NO; k , k , k , k and k were
2
2
2
2
3
4
5
7
determined in this work. The rate constants for the other reac-
2
2
2 2
(4È9) ] 10~12
cm3
molecule~1
s~1,
hence
tions were chosen as follows: k \ O, k \ 8.8 ] 10~12,
k[HC(O)OCH CH OCH O ] NO] \ (7 ^ 3) ] 10~12 cm3
20
22
k
\ 9.0 ] 10~12 [assumed equal to k(CH OCH O
2
2
2 2
molecule~1 s~1.
23
3
2 2
] NO)18],
k[HC(O)OCH CH OCH O ] NO ] \ 7.9
2
2
2
2
2
] 10~12 [assumed equal to k(CH OCH O ] NO )18] and
3
2
2
2
k[HC(O)OCH CH OCH O ] NO] \ 3.6 ] 10~11 [assumed
3.8 Kinetics of the reaction between (c-C H O )O radicals
2
3
2
2
4
7
2
2
equal to k(CH O ] NO)].19 In addition, we assumed that the
and NO
2
rate constants for the reactions of the alkyl radical from 1,4-
dioxane and the H(O)COCH CH OCH radical with NO
The kinetics of reaction (7) were studied by monitoring the
decrease in absorbance at 400 nm following the pulsed
2
2
2
and NO are the same as the rate constants for the reactions
2
of the acetonyl radicals with NO and NO 20 of 2.6 ] 10~11
radiolysis of mixtures of 3 mbar 1,4-dioxane, 20 mbar O ,
2
2
and 1.6 ] 10~11 cm3 molecule~1 s~1, respectively. Fig. 7(A)
0.314È1.29 mbar NO , and SF added to 1000 mbar total
2
6
2860
J. Chem. Soc., Faraday T rans., 1997, V ol. 93

View Article Online
pressure. Fig. 7(B) shows the absorbance as a function of time
after the pulsed radiolysis of a mixture containing 0.314 mbar
The observed loss of 1,4-dioxane vs. those of the reference
compounds following the UV irradiation of 1,4-
of NO . The rate of decay of the absorbance at 400 nm
dioxaneÈreferenceÈCl mixtures, in 700 Torr total pressure of
2
2
increased linearly with the NO concentration. It seems rea-
N
or air diluent are shown in Fig. 9. There was no discern-
2
2
sonable to ascribe the loss of absorbance to the loss of NO in
ible di†erence between data obtained in N or air diluent.
2
2
Linear least-squares analysis gives k /k \ 2.07 ^ 0.10 and
17 18
the system. Three reactions could be responsible for the NO
2
loss:
k
k
/k \ 1.04 ^ 0.09. Using k \ 9.4 ] 10~11 (ref. 21) and
17 19
18
\ 1.9 ] 10~10 (ref. 22) gives k \ (1.9 ^ 0.1) ] 10~10
19
17
and k \ (2.0 ^ 0.2) ] 10~11 cm3 molecule~1 s~1, respec-
17
tively. Consistent results were obtained using the di†erent ref-
erence compounds. We choose to quote a value for k which
17
is the average of the results above the error limits which
encompass the extremes of the ranges, hence k \ (2.0 ^ 0.2)
17
] 10~10 cm3 molecule~1 s~1. We estimate that potential sys-
tematic errors associated with uncertainties in the reference
rate constants could add an additional 10% to the uncertainty
range. Propagating this additional 10% uncertainty gives
However, the timescale of the decay shown in Fig. 7(B) is only
k
\ (2.0 ^ 0.3) ] 10~10 cm3 molecule~1 s~1.
17
consistent with reaction (7) causing the loss of NO (the life-
2
times for both F atoms and c-C H O radicals are \0.3 ls,
4
7 2
3.10 FTIR study of the atmospheric fate of the alkoxy radical
see Sections 3.1 and 3.3).
(c-C H O )O~
The smooth curve in Fig. 7(B) is the pseudo-Ðrst-order Ðt
which gives pseudo-Ðrst-order rate constant, k1st,
4
7 2
a
\1.26 ] 105 s~1. To allow sufficient time for conversion of F
atoms into the peroxy radicals the analysis of the decay traces
was performed starting 2 ls after the radiolysis pulse. The k1st
values for Ðve experiments with di†erent concentrations of
To determine the atmospheric fate of the alkoxy radical
(c-C H O )O~ formed from reaction (6a), the FTIR-smog
4
7 2
chamber system was used to study the products resulting from
the Cl atom initiated oxidation of 1,4-dioxane with and
without NO present. All experiments were performed at 700
Torr total pressure at 295 K. In the absence of NO the
(c-C H O )O~ alkoxy radical is formed via channel (5a).
NO are shown in Fig. 8. Linear least-squares analysis of the
2
data in Fig.
8
gives k \ (1.31 ^ 0.25) ] 10~11 cm3
7
molecule~1 s~1. The y-axis intercept in Fig.
(1.5 ^ 4.0) ] 104 s~1 and is not statistically signiÐcant.
8
is
4
7 2
With NO present it is formed via reaction (6a).
3.9 Kinetics of the reaction of Cl atoms with 1,4-dioxane
Prior to investigating the atmospheric fate of (c-C H O )O~
4
7 2
radicals, relative rate experiments were performed using the
FTIR system to investigate the kinetics of the reactions of Cl
atoms with 1,4-dioxane. Photolysis of molecular chlorine was
used as the source of Cl atoms.
The kinetics of reaction (17) were measured relative to reac-
tions (18) and (19).
Fig. 9 Decay of 1,4-dioxane vs. C H and CH OCH when mixtures
2
4
3
3
Fig. 8 Pseudo-Ðrst-order rate constants for loss of NO following
of these compounds were exposed to Cl atoms in 700 Torr of N
(open symbols) or air (Ðlled symbols) diluent
2
2
pulsed radiolysis of SF È1,4-dioxaneÈO ÈNO mixtures vs. [NO ]
6
2
2
2
J. Chem. Soc., Faraday T rans., 1997, V ol. 93
2861

View Article Online
The aim of these experiments was to determine the relative
importance of reactions (20) and (21) in the atmospheric
chemistry of (c-C H O )O~ radicals.
carbon-containing product was identiÐed; ethylene glycol
diformate. The yield of ethylene glycol diformate in experi-
ments employing NO was slightly (ca. 10%) larger than in
those experiments where NO was absent. This observation is
not unreasonable for two reasons. First, some of the peroxy
radicals are expected to react via channel (5b), which does not
4
7 2
lead to ethylene glycol diformate. Second, HO radials are
2
generated by reaction (24) and can react with peroxy radicals
to give hydroperoxides thereby reducing the ethylene glycol
diformate yield.
As shown in Fig. 11, variation of the O partial pressure by
2
a factor of 114 had no discernible e†ect on the yield of ethyl-
ene glycol diformate. We conclude that decomposition via
reaction (20) is the sole fate of (c-C H O )O~ radicals under
4
7 2
The loss of 1,4-dioxane and the formation of products were
monitored using FTIR spectroscopy. Fig. 10 shows FTIR
spectra acquired before (A) and after (B) a 40 s irradiation of a
atmospheric conditions. Product features at 1640È1680 cm~1
characteristic of alkyl nitrates were searched for but were not
found suggesting that reaction (6) proceeds predominantly, if
not exclusively, via channel (6a). Unfortunately, we do not
have a sample of ethylene glycol diformate and so we cannot
calibrate its IR features. In the light of the absence of any
other IR product features and in the absence of any e†ect of
mixture of 11.8 mTorr of 1,4-dioxane and 135 mTorr of Cl in
2
700 Torr of O diluent. Subtraction of the IR features
2
attributable to 1,4-dioxane from panel (B) gives the infrared
product features shown in panel (C). By comparison with the
available literature spectrum23 we assign the features at 850,
1025, 1092, 1163, 1279, 1460 and 1753 cm~1 in panel (C) to
ethylene glycol diformate, H(O)COCH CH OC(O)H, formed
O
partial pressure it seems reasonable to assume that reac-
2
tion (6) proceeds essentially 100% via channel (6a) and that
the subsequent conversion of the alkoxy radical (c-C H O )O~
2
2
4
7 2
by the following reactions:
into ethylene glycol diformate is 100% e†ective. These
assumptions were used to derive the absolute y-axis scale in
Fig. 11.
Nese et al.7 have studied the behaviour of the alkoxy
radical (c-C H O )O~ in oxygenated aqueous solutions and
4
7 2
report that decomposition via reaction (20) is the dominant
fate of (c-C H O )O~ radicals in such systems. Our Ðndings in
4
7 2
the gas phase are entirely consistent with those reported by
Nese et al.7 in the liquid phase.
To investigate the dependence of the ethylene glycol difor-
mate yield on the O partial pressure, experiments were per-
formed using mixtures of 12 mTorr 1,4-dioxane, 200 mTorr
4
Implications for atmospheric chemistry
2
Cl , 16 mTorr NO and 5.2È595 Torr O in 700 Torr total
A substantial body of kinetic and mechanistic data pertaining
to the atmospheric chemistry of 1,4-dioxane is presented here.
The atmospheric lifetime of 1,4-dioxane is determined by reac-
tion with OH radicals. While the OH radical concentration in
the atmosphere varies with location, time of day, season, and
meteorological conditions, a reasonable 24 h global average is
(0.5È1.0) ] 106 cm~3.24h26 At 296 K, the rate constant for
reaction of OH radicals with dimethoxymethane is
1.09 ] 10~11 cm3 molecule~1 s~1,27 hence, the atmospheric
2
2
pressure with N diluent. In these experiments only one
2
Fig. 10 IR spectra acquired before (A) and after (B) a 40 s irradiation
of a mixture of 11.8 mTorr of 1,4-dioxane and 135 mTorr of Cl in
2
700 Torr of O diluent. During the irradiation 69% of the 1,4-dioxane
2
Fig. 11 Formation of ethylene glycol diformate vs. loss of 1,4-
dioxane following the UV irradiation of mixtures of 12 mTorr of 1,4-
was consumed. Subtraction of the IR features attributable to 1,4-
dioxane from panel B gives the product spectrum shown in panel C
which is assigned to ethylene glycol diformate.
dioxane, 16 mTorr of NO, 200 mTorr of Cl , 5.2 (L), 147 (|) or 595
2
(…) Torr of O , in 700 Torr total pressure of N diluent at 296 K
2
2
2862
J. Chem. Soc., Faraday T rans., 1997, V ol. 93

View Article Online
9
S. Langer, E. Ljungstrom, T. Ellermann, O. J. Nielsen and J.
Sehested, Chem. Phys. L ett., 1995, 240, 499.
lifetime of 1,4-dioxane is 1È2 days. Reaction with OH gives
the alkyl radical c-C H O which in 1 atm of air has a lifetime
4
7 2
10 O. J. Nielsen, J. Sehested, S. Langer, E. Ljungstrom and I. Vang-
berg, Chem. Phys. L ett., 1995, 238, 359.
11 W. G. Mallard, F. Westley, J. T. Herron and R. F. Hampson,
NIST Chemical Kinetics Database, V er. 6.0, NIST Standard Ref-
erence Data, Gaithersburg, MD, 1994.
12 T. Ellermann, J. Sehested, O. J. Nielsen, P. Pagsberg and T. J.
Wallington, Chem. Phys. L ett., 1994, 218, 287.
13 M. M. Maricq and J. J. Szente, J. Phys. Chem., 1992, 96, 4925.
14 D. M. Rowley, P. D. Lightfoot, R. Lesclaux and T. J. Wallington,
J. Chem. Soc., Faraday T rans., 1991, 87, 3221.
15 P. D. Lightfoot, R. A. Cox, J. N. Crowley, M. Destriau, G. D.
Hayman, M. E. Jenkin, G. K. Moortgat and F. Zabel, Atmos.
Environ., 1992, 26, 1805.
16 J. Sehested, O. J. Nielsen and T. J. Wallington, Chem. Phys. L ett.,
1993, 213, 257.
17 O. L. Rasmussen and E. B. Bjergbakke, RisÏ-Report-395, 1984.
18 S. Langer, E. Ljungstrom, T. Ellermann, O. J. Nielsen and J.
Sehested, Chem. Phys. L ett., 1995, 240, 53.
19 R. Atkinson, D. L. Baulch, R. A. Cox, R. F. Hampson, J. A. Kerr
and J. Troe, J. Phys. Chem. Ref. Data, 1992, 21, 1125.
20 J. Sehested, L. K. Christensen, M. Bilde, O. J. Nielsen and T. J.
Wallington, J. Phys. Chem., 1997, submitted.
of 0.02 ls with respect to addition of O to give the corre-
2
sponding peroxy radical (c-C H O )O . We have shown here
that (c-C H O )O radicals react rapidly with NO to produce
4
7
2
2
4
7
2
2
NO and, by inference, (c-C H O )O radicals. Using k \ 1.2
2
4
7
2
6
] 10~11 cm3 molecule~1 s~1 together with an estimated
background tropospheric NO concentration of 2.5 ] 108 mol-
ecule cm~3, the lifetime of (c-C H O )O radicals with respect
4
7
2
2
to reaction with NO is calculated to be 6 min. Reaction (6) is
likely to be an important atmospheric loss of (c-C H O )O
4
7
2
2
radicals. The sole fate of the alkoxy radical derived from 1,4-
dioxane is decomposition via CwC bond scission leading to
the formation of H(O)COCH CH OC(O)H (ethylene glycol
diformate). From the available database, it appears that where
possible decomposition via CwC bond scission is the domi-
nant atmospheric fate of alkoxy radicals a to the oxygen atom
in the ether linkages (e.g. in diethyl ether,28 2-ethoxy ethanol29
and 2-butoxy ethanol30).
2
2
We thank Roscoe Carter (Ford Motor Company) for helpful
discussions regarding the IR spectrum of ethylene glycol difor-
mate and the Commission of the European Communities for
Ðnancial support.
21 W. B. DeMore, S. P. Sander, D. M. Golden, R. F. Hampson,
M. J. Kurylo, C. J. Howard, A. R. Ravishankara, C. E. Kolb and
M. J. Molina, JPL Publ. 94-26, Jet Propulsion Laboratory, Pasa-
dena, CA, 1992.
22 S. Langer, E. Ljungstrom, I. Vangberg, T. J. Wallington, M. D.
Hurley and O. J. Nielsen, Int. J. Chem. Kinet., 1996, 28, 299.
23 Sadtler Research Laboratories Standard Spectra Collection,
1980, ISBN 0-8456-0045-1.
References
24 A. R. Ravishankara and E. R. Lovejoy, J. Chem. Soc., Faraday
1
2
M. A. Roubi, Chem. Eng., 1995, 44, 37.
T rans., 1994, 90, 2159.
K. B. Hansen, R. Wilbrandt and P. Pagsberg, Rev. Sci. Instrum.,
1979, 50, 1532.
25 H. P. Dorn, U. Brandenburger, T. Brauers and D. H. Ehhalt,
Geophys. Res. L ett., 1996, 23, 2537.
3
4
T. J. Wallington and S. M. Japar, J. Atmos. Chem., 1989, 9, 399.
M. Bilde, T. E. MÔgelberg, J. Sehested, O. J. Nielsen, T. J. Wall-
ington, M. D. Hurley, S. M. Japar, M. Dill, V. L. Orkin, T. J.
Buckley, R. E. Huil and M. J. Kurylo, J. Phys. Chem., 1997, 101,
3514.
T. J. Wallington, P. Dagaut and M. J. Kurylo, J. Chem. Rev.,
1992, 92, 667.
T. J. Wallington, M. D. Hurley, J. Shi, M. M. Maricq, J. Sehested,
O. J. Nielsen and T. Ellermann, Int. J. Chem. Kinet., 1993, 25,
651.
26 A. Hofzumahaus, U. Aschmutat, M. Hessling, F. Holland and
D. H. Ehhalt, Geophys. Res. L ett., 1996, 23, 2541.
27 P. Dagaut, R. Liu, T. J. Wallington and M. J. Kurylo, J. Phys.
Chem., 1990, 94, 1881.
28 T. J. Wallington and S. M. Japar, Environ. Sci. T echnol., 1991, 25,
410.
5
6
29 K. Stemmler, W. Mengon and J. A. Kerr, Environ. Sci. T echnol.,
1996, 30, 3385.
30 K. Stemmler, W. Mengon, D. J. Kinnison and J. A. Kerr, Environ.
Sci. T echnol., 1997, 31, 1496.
7
8
C. Nese, M. N. Schuchmann, S. Steenken and C. von Sonntag, J.
Chem. Soc., Perkin T rans 2, 1995, 1037.
J. Platz, J. Sehested and T. J. Wallington, unpublished work.
Paper 7/00598I; Received 27th January, 1997
J. Chem. Soc., Faraday T rans., 1997, V ol. 93
2863