Product distributions from the OH radical-induced oxidation of but-1-ene, methyl-substituted but-1-enes and isoprene in NO(x)-free air

Article abstract of DOI:10.1039/b002053m

Product distributions resulting from the OH-induced oxidation of but-1-ene, 2-methylbut-1-ene, 3-methylbut-1-ene and isoprene in air were measured in the absence of nitrogen oxides and compared with predictions based on currently accepted oxidation mechanisms. In the case of butenes, the observed distributions of carbonyl compounds, hydroxyketones, hydroxyalkanals and diols were evaluated to obtain probabilities for the initial attack of OH radical on the outer position of the double bond (y = 0.90 ± 0.03 for 2-Me-but-1-ene and y = 0.76 ± 0.05 for both but-1-ene and 3-Me-but-1-ene), for the probability of formation of stable products in the self-reaction of secondary β-hydroxyperoxyl radicals (k(ssb)/k(ss) = 0.29 ± 0.07 for but-1-ene and k(ssb)/k(ss) = 0.19 ± 0.06 for 3-Me-but-1-ene), and for the ratio of the reaction with oxygen vs. decomposition of β-hydroxyalkoxyl radicals, k₃[O₂]/(k₄ + k₃[O₂]) = 0.25 ± 0.04 for but-1-ene and = 0.38 ± 0.04 for 3-Me-but-1-ene. The last two values disagree with other published data, which suggest a smaller effect of oxygen. The oxidation of isoprene produced methacrolein and methyl vinyl ketone with a ratio 0.93 ± 0.10, the ratio of methyl vinyl ketone and 3-methylfuran was 7.3 ± 1.0. Other products were 1-hydroxy-3-methylbut-3-en-2-one (identified by mass spectrometry) and 3-methyl-3-oxo-butane (tentatively identified). The overall product distribution was complex and could not be fully elucidated. Computer simulations based on several mechanisms applied the relative probabilities for OH addition found for the but-1-enes. Comparison with the experimental data suggests probabilities for OH addition to the methylated double bond of 0.504 ± 0.027 (outer position) and 0.056 ± 0.003 (inner position), and to the non-methylated double bond of 0.335 ± 0.023 (outer position) and 0.105 ± 0.008 (inner position).

Full text of DOI:10.1039/b002053m

View Article Online / Journal Homepage / Table of Contents for this issue
Product distributions from the OH radical-induced oxidation of
but-1-ene, methyl-substituted but-1-enes and isoprene in NO -free
x
air¤
Heinz-Jurgen Benkelberg,a Olaf Boge,b Ralph Seuwena and Peter Warneck*a
a Max-Planck-Institut fur Chemie, 55020 Mainz, Germany. E-mail: biogeo=mpch-mainz.mpg.de
b Institut fur T ropospharenforschung, 04303 L eipzig, Germany
Received 14th March 2000, Accepted 20th July 2000
Published on the Web 31st August 2000
Product distributions resulting from the OH-induced oxidation of but-1-ene, 2-methylbut-1-ene, 3-methylbut-
1-ene and isoprene in air were measured in the absence of nitrogen oxides and compared with predictions
based on currently accepted oxidation mechanisms. In the case of butenes, the observed distributions of
carbonyl compounds, hydroxyketones, hydroxyalkanals and diols were evaluated to obtain probabilities for
the initial attack of OH radical on the outer position of the double bond (c \ 0.90 ^ 0.03 for 2-Me-but-1-ene
and c \ 0.76 ^ 0.05 for both but-1-ene and 3-Me-but-1-ene), for the probability of formation of stable
products in the self-reaction of secondary b-hydroxyperoxyl radicals (k /k \ 0.29 ^ 0.07 for but-1-ene and
ssb ss
k
/k \ 0.19 ^ 0.06 for 3-Me-but-1-ene), and for the ratio of the reaction with oxygen vs. decomposition of
ssb ss
b-hydroxyalkoxyl radicals, k [O ]/(k ] k [O ]) \ 0.25 ^ 0.04 for but-1-ene and \0.38 ^ 0.04 for 3-Me-but-
3
2
4
3
2
1-ene. The last two values disagree with other published data, which suggest a smaller e†ect of oxygen. The
oxidation of isoprene produced methacrolein and methyl vinyl ketone with a ratio 0.93 ^ 0.10, the ratio of
methyl vinyl ketone and 3-methylfuran was 7.3 ^ 1.0. Other products were 1-hydroxy-3-methylbut-3-en-2-one
(identiÐed by mass spectrometry) and 3-methyl-3-oxo-butane (tentatively identiÐed). The overall product
distribution was complex and could not be fully elucidated. Computer simulations based on several
mechanisms applied the relative probabilities for OH addition found for the but-1-enes. Comparison with the
experimental data suggests probabilities for OH addition to the methylated double bond of 0.504 ^ 0.027
(outer position) and 0.056 ^ 0.003 (inner position), and to the non-methylated double bond of 0.335 ^ 0.023
(outer position) and 0.105 ^ 0.008 (inner position).
RO ] R@ ] aldehyde
(IV)
Introduction
where R@ is a carbon-centered radical of lower carbon number
than R and the aldehyde formed in reaction (IV) contains one
less carbon than that formed in reaction (III). Hydroxyalkoxyl
radicals undergo reactions similar to those of alkoxyl radicals,
but the fragmentation pathway (IV) has been found to domi-
nate in most cases studied.5 Only for the HOCH CH O
The degradation of hydrocarbons in the atmosphere is largely
initiated by reactions with the hydroxyl radical, OH, which
has multiple sources such as ultraviolet photolysis of ozone in
the presence of water vapor, the reaction of hydroperoxyl rad-
icals with NO, and photolysis of nitrous acid.1,2 Whereas OH
reacts with alkanes by hydrogen abstraction, the reactions
with alkenes and alkadienes proceed primarily by the addition
of OH to a double bond. In both cases, the reactions are fol-
lowed by further addition of molecular oxygen to the newly
created reactive site. This leads to the formation of alkyl-
peroxyl radicals in the Ðrst case, and hydroxy-alkylperoxyl
radicals in the second case. It is generally assumed that in the
2
2
radical derived from ethene has hydrogen abstraction by
oxygen been found to be important.6,7 Larger hydroxyalkoxyl
radicals may also isomerise via internal hydrogen
abstraction,5,8 and this reaction path can lead to additional
complications. In atmospheric regions where the abundance
of NO is low or in laboratory studies of the oxidation of
hydrocarbons in the absence of NO (\NO ] NO ), peroxyl
atmosphere both types of alkylperoxyl radicals, RO , react
x
2
2
radicals react with themselves or with other peroxyl radicals
with NO
present in the reaction system,9 for example
RO ] NO ] RO ] NO (major)
(I)
2
2
R O ] R O ] alcohol ] aldehyde/ketone ] O
(V)
(VI)
1
2
2
2
2
] RONO (minor)
(II)
2
] R O ] R O ] O
1
2
2
because such reactions have been found to be rapid for a
R O /R O ] HO ] R OOH/R OOH ] O
(VII)
number of alkylperoxyl radicals,3h5 and NO is present in day-
1
2
2
2
2
1
2
2
light due to the photodissociation of NO in addition to
where, in reaction (VI), RO radicals appear again as products.
The partitioning between reactions (III) and (IV) and reac-
tions (V) and (VI) can be determined, in principle, from the
distribution of alcohols, aldehydes and ketones occurring as
end products in appropriate laboratory studies.
2
direct emissions. The alkoxyl radicals, RO, derived from
alkanes undergo further reactions such as
RO ] O ] HO ] aldehyde/ketone
(III)
2
2
The present paper deals with the oxidation of methyl-
substituted but-1-enes induced by OH radicals, which we have
studied in the absence of nitrogen oxides in order to determine
¤ Electronic Supplementary Information available. See http://
www.rsc.org/suppdata/cp/b0/b002053m/
DOI: 10.1039/b002053m
Phys. Chem. Chem. Phys., 2000, 2, 4029È4039
This journal is ( The Owner Societies 2000
4029

View Article Online
the relative signiÐcance of pathways (III) and (IV) on the one
hand, and (V) and (VI) on the other, in the hope that the
information thus obtained might improve our understanding
of the oxidation mechanism of important alkadienes such as
isoprene. Fig. 1 shows the main reaction scheme for the oxida-
tion of but-1-ene. Similar schemes apply to the methyl-
substituted but-1-enes. The OH-induced oxidation of but-1-
bon (mole fraction) in synthetic air. The pressure was slightly
above atmospheric. The reaction was started by placing the lit
mercury lamp in the quartz Ðnger to generate OH radicals by
photolysis of H O . Under these conditions, taking into
2
2
account the known rate coefficients for reactions of OH rad-
icals with H O , the but-1-enes, and isoprene,3,5 more than
2
2
95% of the OH radicals were calculated to have reacted with
the hydrocarbon. Following irradiation, the bulb was con-
nected to the sample loop of a gas chromatography via a thin
TeÑon tube pushed through a hole pierced in the septum.
Samples were transferred by means of a suction pump.
ene in the presence of NO has previously been studied by
x
Atkinson et al.10 and Aschmann et al.,11 that of 3-methylbut-
1-ene was studied by Atkinson et al.12 In the present work we
attempted an unambiguous analytical characterization of the
diols and mixed hydroxyaldehydes or hydroxyketones
expected to arise from reaction (III)È(V). Model calculations
were performed to assist in the data evaluation and to
compare calculated with observed product distributions.
The oxidation of isoprene induced by OH radicals has been
The sample loop of the gas chromatography was heated to
160 ¡C. A Ñame ionization detector was used in conjunction
with a 50 m long CP-WAX 52-CB capillary column, 0.32 mm
id, 1.2 lm Ðlm thickness. The nitrogen carrier gas Ñow was 6.9
cm3 min~1. The following column temperature program was
mostly used: 40 ¡C constant for 2 min, followed by a rise of
20 ¡C min~1 up to 160 ¡C, which was held for 10 min, fol-
lowed by a further rise of 30 ¡C min~1 up to 210 ¡C.
The alkenes but-1-ene, 2-methylbut-1-ene and 3-methylbut-
1-ene, isoprene and the oxidized compounds propanal, 2-
methylpropanal, butanone, 1-buten-3-one (methyl vinyl
ketone), 2-methylpropenal (methacrolein), 3-methylfuran, 1,2-
dihydroxybutane and 1-hydroxybutan-2-one were available
commercially. The diols 1,2-dihydroxy-2-methylbutane and
1,2-dihydroxy-3-methylbutane were prepared by hydroxyl-
ation of the corresponding alkenes with performic acid as
described by Criegee et al.;21 1-hydroxy-3-methylbutan-2-one
was prepared by reacting glycolnitrile with isopropyl-
magnesium bromide as described by Pfeil and Barth.22
Product identiÐcation was made by comparison of peak
retention times with those of authentic samples, as far as pos-
sible. In a number of cases a combination of gas chromatog-
raphy and mass spectrometry (GC-MS) with conventional
electron impact ionization was used for peak identiÐcation. In
this case, the sample was injected by the freeze trapÈÑash
heating procedure. Although the instrument was Ðtted with a
similar separating column as that used for quantitative
analysis as described above, a di†erent carrier gas Ñow was
required, which gave rise to changes in retention times and
resulted in some uncertainty. Peak calibration was performed
with gaseous samples as far as possible. Samples were pre-
pared in the reaction bulb by successive dilution of a mixture
of the pure substance and air to cover the concentration range
encountered. Mixing ratios were determined by means of pres-
sure measurements using capacitance manometers and apply-
ing the ideal gas law. Diols and the mixed hydroxycarbonyl
compounds, which were not sufficiently volatile for the deter-
repeatedly studied in the presence of NO .13h16 Although the
x
yields of the products methacrolein, methyl vinyl ketone, 3-
methylfuran and formaldehyde are well characterized, most
studies showed a deÐcit in the product balance indicating the
occurrence of additional products. Mechanistic considerations
suggest these products to be mainly hydroxycarbonyl com-
pounds. Yu et al.17 and Kwok et al.18 have identiÐed several
of these compounds by mass spectrometry employing soft ion-
ization techniques. We have oxidized isoprene under NO -free
x
conditions similar to those in the oxidation of the but-1-enes
in an attempt to detect some of these products and to deter-
mine the relative yields of the three main products under con-
ditions di†erent from those used previously. Here again,
model calculations were performed to compare observed with
expected product distributions. Ruppert and Becker19 have
also studied the oxidation of isoprene in the absence of NO
and found unsaturated diols as evidence for the occurrence of
x
the interaction of peroxyl radicals in this system.
Experimental
Apparatus and experimental procedures were similar to those
described previously.20 Reactions were carried out in glass
bulbs of 2 L capacity. The interior walls were treated with
dimethyl-dichlorosilane to minimize losses of hydroxy-
carbonyl compounds and diols. Each bulb was provided with
TeÑon-stoppered shut-o† valves, a silicone rubber septum,
and a quartz Ðnger reaching into the center of the bulb, into
which a penray mercury lamp was inserted. Specialty quartz
was used to block ozone-forming radiation (j \ 235 nm
wavelength). The bulbs were Ðlled with a gas mixture contain-
ing approximately 150 ppm H O and 0.01È0.1% hydrocar-
2
2
Fig. 1 Oxidation mechanism for but-1-ene.
Phys. Chem. Chem. Phys., 2000, 2, 4029È4039
4030

View Article Online
mination of vapor pressures, were injected directly into the
reaction bulb in known amounts by means of a calibrated
syringe, followed by the admixture of air and further dilution.
In cases where authentic samples were unavailable, the cali-
bration factors were estimated by comparison with the signal
response for similar compounds. The detector sensitivity
varied insigniÐcantly during the time of the experiments so
that consistency checks were required only occasionally.
suggests that primary and secondary species do not survive
passage through the GC column at elevated temperatures.
Tertiary hydroperoxides are more stable. A tertiary hydroxy-
hydroperoxide is expected to be formed in the oxidation of
2-methylbut-1-ene. It is possible in this case that the peak
identiÐed as the diol, which is considerably larger than that of
the corresponding b-hydroxyalkanal, contains a contribution
from 1-hydroxy-2-hydroperoxy-2-methylbutane. Finally, one
must consider the possibility of products formed by hydrogen
abstraction from the side chain of the but-1-enes. The most
facile abstraction process involves the tertiary hydrogen atom
in 3-methylbut-1-ene. The tertiary peroxyl radical thus formed
is expected to produce methyl vinyl ketone. Acetone and a
vinyl radical are conceivable alternative products. Methyl
vinyl ketone was not detected, and acetone, which on our
column was masked by 2-methylpropanal, may arise by a dif-
ferent route (see below). Atkinson et al.,12 who looked for pro-
ducts resulting from H-abstraction from 3-methylbut-1-ene,
did not observe methyl vinyl ketone, but found small amounts
of methacrolein, which they interpreted to arise from H-
abstraction following a complex isomerization pathway. They
concluded that H-abstraction occurs to about 5% of the
overall reaction. We did not detect signiÐcant amounts of
methacrolein and have ignored the abstraction pathway.
Atkinson et al.12 have also found that the 1-hydroxy-3-
methyl-2-butoxyl radical resulting from 3-methylbut-1-ene can
decompose in two ways, forming either 2-methylpropanal and
HCHO or glycolaldehyde and acetone.
Results
Oxidation of but-1-enes
Product distribution. Fig. 2 shows a set of chromatograms
that are typical for the product mixtures arising from the oxi-
dation of but-1-ene and the two methyl-substituted but-1-enes.
The patterns are similar. We have concentrated on the four
major peaks and ignored other peaks that were marginal. Car-
bonyl compounds elute after about 3 min, diols after about 12
min. With but-1-ene and 3-methylbut-1-ene one Ðnds
hydroxyketones eluting after about 8 min, in the case of 2-
methylbut-1-ene the hydroxyketone is missing. This is
expected from the tertiary hydroxyperoxyl radical involved.
Whereas these products were unambiguously identiÐed, an
additional peak with lesser intensity occurred in each chro-
matogram slightly prior to the hydroxyketones. We have
assigned these peaks to b-hydroxyalkanals for the following
reasons: these compounds are among the expected products
in each case (see Fig. 1); they occur with abundances compa-
rable to the products already identiÐed; and the retention
times are close to those of the structurally similar hydroxy-
ketones. The b-hydroxyalkanals could not be synthesized,
however, so that positive conÐrmation by means of authentic
samples was not possible. Breakdown patterns of the mass
spectra supported our assignment, but the parent peaks were
missing so that some ambiguity remained. In addition to the
products mentioned the interaction between hydroxy-
alkylperoxyl and hydroperoxyl radicals lead to the formation
of hydroxyhydroperoxides. Little is known about the stability
of these compounds. Our experience with alkylhydroperoxides
HOCH CH(O~)CH(CH ) ] HOCH ] (CH ) CHCHO
2
3 2
2
3 2
] HOCH CHO ] É CH(CH )
3 2
2
Their data suggest that 20È25% of the decomposition follows
the second pathway. On the column used in this study acetone
and 2-methylpropanal eluted with almost identical retention
times so that they could not be separated. Glycolaldehyde had
a retention time of 8.0 min, and should have appeared shortly
before 2-hydroxy-3-methylbutanal. Glycolaldehyde may have
been present as a minor product, but we estimate that its yield
after 30 min irradiation time was \0.2 ppm.
Table 1 summarizes product distributions derived from
quantitative analysis of the chromatograms. The response
factor for b-hydroxyalkanals was estimated from that of the
1,2-diols. The response of the hydroxyketones was about 12%
lower. Table 1 shows product concentrations obtained 15 and
30 min after starting the reaction. The evolution of products
was linear with time for about 20 min. The yield after 30 min
was slightly lower, but the relative distribution did not change
and an average was taken. The product distributions were
used to determine the branching ratios of reactions appearing
in Fig. 1.
Model calculations. Fig. 1 shows that OH radicals may
attach to the but-1-ene double bond either at the outer or at
the inner position. The subsequent addition of oxygen leads to
the formation of primary and secondary b-hydroxyperoxyl
radicals in the cases of but-1-ene and 3-methylbut-1-ene,
whereas the oxidation of 2-methylbut-1-ene involves primary
and tertiary b-hydroxyperoxyl radicals. Products evolve from
the mutual interactions of these radicals. Table 2 lists reac-
tions that need to be considered. Model calculations were
based on the FACSIMILE computer code,23 and involved an
iteration scheme to derive consistent values for the probability
c of OH attack at the outer position of the double bond, and
the branching ratios k /k , k [O ]/(k [O ] ] k ) and the
ssb ss
overall product distribution.
3
2
3 2 4
A
major problem in this
approach was a proper choice of rate coefficients for reactions
of hydroxyperoxyl radicals. It has been shown by numerous
studies summarized in recent reviews9,24 that the rate coeffi-
cients for self-reactions of primary, secondary and tertiary
alkylperoxyl radicals are of the order of k B 10~13, k B
Fig. 2 Chromatograms of gas mixtures resulting from the OH
radical-induced oxidation of but-1-ene, 2-methylbut-1-ene and 3-
methylbut-1-ene in air.
pp
ss
Phys. Chem. Chem. Phys., 2000, 2, 4029È4039
4031

View Article Online
Table 1 Relative product distributions from the oxidation of methyl-substituted but-1-enes
Retention time
Exp.
Yield (ppm)
15 min
Relative yield (%)
Average
Product
Auth.
30 min
But-1-ene:
Propanal
2.44
8.04
8.59
2.44
È
8.56
12.49
5.94 ^ 0.14
0.83 ^ 0.06
2.85 ^ 0.15
2.04 ^ 0.07
11.66 ^ 0.42
10.25 ^ 0.25
1.52 ^ 0.05
4.92 ^ 0.15
3.87 ^ 0.08
20.49 ^ 0.53
50.6
7.0
24.2
18.2
2-Hydroxybutanal
1-Hydroxybutan-2-one
1,2-Dihydroxybutane
Total
2-Me-but-1-ene:
Butanone
2-Hydroxy-2-Me-butanal
1,2-Dihydroxy-2-Me-butane
Total
12.49
3.64
7.47
11.74
3.64
È
11.76
11.09 ^ 0.14
0.65 ^ 0.03
1.93 ^ 0.03
13.67 ^ 0.2
18.31 ^ 0.33
1.09 ^ 0.02
3.31 ^ 0.06
22.71 ^ 0.41
80.8
4.6
14.3
3-Me-1-butene:
2-Me-propanala
2-Hydroxy-3-Me-butanal
1-Hydroxy-3-Me-butan-2-one
1,2-Dihydroxy-3-Me-butane
Total
2.65
8.19
8.40
2.65
È
8.40
13.66
6.13 ^ 0.30
0.85 ^ 0.04
3.68 ^ 0.07
1.80 ^ 0.08
12.46 ^ 0.22
10.8 ^ 0.25
1.61 ^ 0.08
6.76 ^ 0.14
3.82 ^ 0.07
22.99 ^ 0.54
48.1
6.9
29.5
15.5
13.69
a The signal may include a contribution of acetone.
10~15 and k B 2.5 ] 10~17 (unit: cm3 molecule~1 s~1). The
leading either to hydroxyalkoxyl radicals or to diols and
mixed hydroxy-carbonyl compounds. The Ðrst channel is
designated a, the other(s) b or b and c, respectively. As the
observed product distribution with four identiÐed products
allows only three independent parameters to be determined,
further assumptions are necessary for the branching ratios of
the self- and cross-reactions of primary peroxyl radicals. The
branching ratio for the self-reaction of hydroxyethylperoxyl
tt
rate coefficients for a cross-combination reaction may be esti-
mated by taking the root over the product of the rate coeffi-
cients for the two self-reactions.25 Experimental data for the
rate coefficients of b-hydroxyperoxyl radicals have also been
obtained: rate coefficients for the self-reactions of hydroxy-
ethylperoxyl, 2-hydroxy-3-butylperoxyl and 2-hydroxy-2,3-
dimethyl-3-butylperoxyl were 2.1 ] 10~12, 6.9 ] 10~13 and
4.9 ] 10~15 cm3 molecule~1 s~1, respectively.26,27 The b-
hydroxy functionality evidently accelerates the rates of the
self-reactions compared to the simple alkylperoxyl radicals.
On the basis of these data, we have adopted for the self-
reactions of the primary, secondary and tertiary hydroxy-
radicals according to Barnes et al.6 is k /k \ 0.50 ^ 0.06 at
ppb pp
298 K. Thus, we have assumed standard conditions
k
/k \ 0.5, k /k \ 0.5, k \ k as shown in Table 2,
ppa pp
psa ps
psb
psc
but in order to determine their inÑuence on the calculations
we have also varied k /k and k /k within the range 0.3È
ppa pp
psa ps
peroxyl
radicals
involved
in
the
oxidation
of
0.7. For the interaction of primary with tertiary peroxyl rad-
methyl-substituted but-1-enes the rate constant values listed in
Table 2. Values for the rate constants of cross-combination
reactions were estimated by the method outlined above for
alkylperoxyl radicals.25 For comparison, we also performed
calculations using rate coefficients applicable to simple alkyl-
peroxyl radicals. The di†erences in the results were minor.
icals, we have assumed k /k \ 0.67, k /k \ 0.33. In this
pta pt
ptc pt
case
k /k \ 0, because the reaction site lacks an
ptb pt
abstractable hydrogen atom. Table 3 shows values for c and
two other branching ratios calculated from the measured
product distributions. The range attached to each individual
result is that derived by varying the branching ratios of reac-
tions involving primary peroxy radicals within the indicated
range. Experimental uncertainties are not included here. Table
3 includes the calculated fractional yields of hydroperoxides.
These products contribute appreciably to the total product
yield. The e†ect is greatest in the oxidation of 2-methylbut-1-
ene because of the comparatively low rate of the self-reaction
Reactions with HO radicals, which are formed in the reaction
2
sequences following formation of b-hydroxyalkoxyl radicals,
generally are rapid. We have used rate coefficients similar to
those reported by Jenkin and Hayman.26
All the reactions involving two organic peroxyl radicals
admit two (in the case of cross-reactions three) channels
Table 2 Rate coefficients used for reactions of hydroxyperoxyl radicalsa
OH ] Bu ] O ] R O
1-c
2
p
2
2
OH ] Bu ] O ] R O
c
c
a
b
a
b
R
B 9 ] 1011b
2
s
OH
OH ] Bu ] O ] R O
2
t
2
R O ] R O ] 2 R O ] O
k
\ 2.0 ] 10~12
\ 4.0 ] 10~13
p
2
p
2
p
2
pp
R O ] R O ] R OH ] RCHO ] O
p
2
p
2
p
2
R O ] R O ] 2 R O ] O
k
s
2
s
2
s
2
ss
R O ] R O ] R OH ] R 2CO ] O
s
2
s
2
s
s
2
R O ] R O ] R O ] R O ] O
k
k
\ 4.5 ] 10~13
\ 2.25 ] 10~13
\ 2.25 ] 10~13
p
2
s
2
p
s
2
psa
R O ] R O ] R OH ] R 2CO ] O
p
2
s
2
p
s
2
2
psb
R O ] R O ] RCHO ] R OH ] O
k
p
2
s
2
s
psc
R O ] R O ] 2 R O ] O
k \ 5.0 ] 10~15
t
2
t
2
t
2
tt
R O ] R O ] R O ] R O ] O
k
k
\ 6.7 ] 10~14
t
2
p
2
t
p
2
pta
R O ] R O ] R OH ] RCHO ] O
\ 3.3 ] 10~14
t
2
p
2
t
2
ptc
R O ] HO ] R OOH ] O
k
pHO2 \ 1.5 ] 10~11
p
2
2
p
2
R O ] HO ] R OOH ] O
k
sHO2 \ 1.5 ] 10~11
s
2
2
s
2
R O ] HO ] R OOH ] O
k
tHO2 \ 1.5 ] 10~11
t
2
2
t
2
HO ] HO ] H O ] O
k
HO2 \ 2.5 ] 10~12
2
2
2
2
2
a Rate coefficients in cm3 molecule~1 s~1. b Rate of radical formation in molecule cm~3 s~1.
4032
Phys. Chem. Chem. Phys., 2000, 2, 4029È4039

View Article Online
Table 3 Oxidation of but-1-enes induced by reaction with OH radicals: probability c for OH attack at the outer position of the double bond;
probability for the formation of stable products in the self-reactions of secondary hydoxyperoxyl radicals; ratio of reaction pathways of secondary
hydroxyalkoxyl radicals, and fraction of products, f (ROOH), appearing as hydroperoxidesa,b
Alkene
c
k
/k
k [O ]/(k ] k [O ])
k /s~1c
f (ROOH)
ssb ss
3
2
4
3
2
4
But-1-ene
3-Me-but-1-ened
0.76 ^ 0.05
0.76 ^ 0.04
0.78 ^ 0.05
0.90 ^ 0.03
0.29 ^ 0.07
0.19 ^ 0.06
0.16 ^ 0.05
È
0.25 ^ 0.04
0.38 ^ 0.04
0.32 ^ 0.04
È
1.3 ] 105
6.9 ] 104
9.0 ] 104
È
0.36
0.38
0.38
0.45
2-Me-but-1-ene
a For deÐnition of rate coefficients see Fig. 1. b Assumed range for k /k \ 0.5 ^ 0.2, k /k \ 0.5 ^ 0.2, k \ k . c Rate constant, if k \ 8
ppa pp
psa ps
psb
psc
3
] 10~15 cm3 molecule~1 s~1. d First line: according to the mechanism of Fig. 1; second line: assuming two pathways for the decomposition of
the 1-hydroxy-3-methylbutan-2-oxy radical with yields of 78% 2-methylpropane and 22% glycolaldehyde.
of tertiary hydroxy-2-methyl-2-butylperoxyl radicals. In view
of the Ðnding of Atkinson et al.12 that the
HOCH CH(O~)CH(CH ) radical appearing as intermediate
the highest mass number that is clearly above the background
appears at 86 u. If this were the parent peak, the most reason-
able assignment of the mass spectrum would be to 2-
methylbutan-3-one. This ketone, however, should elute at a
much earlier time, that is, with a retention time closer to that
of methyl vinyl ketone, whereas the observed retention time is
more characteristic of that of a difunctional compound. We
have considered butan-2,3-dione, but found that the spectral
intensity distribution is not fully consistent with the published
mass spectrum.27 In addition, it is difficult to identify a route
to its formation. Yu et al.17 have shown that 1-hydroxy-3-
buten-2-one (hydroxymethyl vinyl ketone) occurs as a product
2
3 2
in the oxidation of 3-methylbut-1-ene may enter into two
decomposition pathways, additional calculations were per-
formed assuming a 22 : 78 partitioning between the products
glycolaldehyde and 2-methylpropanal. The rate constant and
the branching ratio for the self-reaction of the 2-propylperoxyl
radicals formed concurrently with glycolaldehyde was taken
from Heimann and Warneck,20 the corresponding values for
the cross-reactions with other peroxyl radicals were estimated
as described above. The results are included in Table 3. The
calculations were mainly done to obtain yields of products
predicted to occur in addition to those detected and listed in
Table 1, namely, acetone, propan-2-ol and glycolaldehyde.
Their calculated yields after 30 min of irradiation are (in ppm)
0.086 ^ 0.02, 0.028 ^ 0.006 and 2.3 ^ 0.4, respectively. The
yields of acetone and propan-2-ol are low, because most of the
in the oxidation of isoprene in the presence of NO . This
x
compound also features a parent peak at 86 u. The fragmenta-
tion pattern is expected to include a peak at 55 u from the
elimination of CH OH. Although this is observed, the com-
2
panion peak expected to occur at 31 u is weak. In contrast,
the mass spectrum of 1-hydroxybutan-2-one shows strong
signals at 31 and 29 u in addition to the peak at 57 u resulting
2-propylperoxyl radicals react with HO to form 2-propyl
2
hydroperoxide. The yield of glycolaldehyde is substantial, so
from the loss of CH OH. Also inconsistent with 1-
2
that this substance should have been observed as a signal in
our chromatograms.
hydroxybut-3-en-2-one are the peak at 71 u, which signals the
loss of methyl, and the strong peak at 43 u. We have looked
for alternatives and note that 3-oxo-butanal would feature a
strong mass 43 fragment peak and a parent peak at 86 u.
However, elimination of the CHO group would require a frag-
ment peak at 57 u, which is missing from the mass spectrum.
We conclude that an assignment based on a parent peak at 86
u is difficult to justify on the basis of the observations. The
two compounds identiÐed in the oxidation of 3-methylbut-1-
ene: 1-hydroxy-3-methylbutan-2-one (shown in Fig. 4) and 2-
hydroxy-3-methylbutanal, feature mass spectra in which the
parent peaks at 102 u are practically absent and either the
Oxidation of isoprene
Product distribution. Isoprene was reacted under experimen-
tal conditions similar to those described above for the but-1-
enes. Relatively high mole fractions of isoprene, namely 100È
1000 ppm, were used in order to suppress losses of the major
products methacrolein and methyl vinyl ketone by secondary
reactions with OH radicals. The use of much smaller concen-
trations in the range 5È50 ppm produced an undesirably large
scatter of the results, which we were unable to remove. Fig. 3
shows a chromatogram of the product series. The temperature
programme was slightly modiÐed compared to that described
earlier, resulting in a slower elution of products at later times.
The leading peaks are due to methacrolein and methyl vinyl
ketone with retention times 5.4 and 6.7 min, respectively; 3-
methylfuran eluted after 5.7 min. Most of the peaks appearing
later are small and difficult to identify. In contrast to the study
of the but-1-enes we had no authentic samples for hydroxy-
carbonyl compounds, so that GC-MS was the only method
available for product identiÐcation. Mass spectra obtained for
the largest peaks, numbered 6, 10, 14, are shown in Fig. 4.
Included for comparison are mass spectra for methacrolein,
1-hydroxybutan-2-one and 1-hydroxy-3-methylbutan-2-one.
Peak d4 was identiÐed as 3-methylbut-3-enal. As the inten-
sity of the product was low, it was included only in the total
count but not in the mechanism to be discussed later. Peak
d6 can be assigned to 3-methylbut-3-en-2-one.28 This
product may arise from a rearrangement of the radical
resulting from the addition of OH to the third carbon atom of
isoprene as shown in Fig. 5. The assignment of the next peak
is more problematic. The mass spectrum for peak d10 lacks
the fragmentation pattern in the range 37È42 u that is charac-
teristic of the CH C2CH group (compare with methacrolein),
CH OH or the CHO group is split o† to produce peaks at 71
2
and 73 u, respectively. For comparison and future reference,
we present here the spectral intensity distributions (mass 102
Fig. 3 Chromatogram of products resulting from the oxidation of
isoprene in NO -free air: Peak identiÐcation (1) methacrolein, (2) 3-
x
methylfuran, (3) methyl vinyl ketone, (6) 2-methylbut-1-en-3-one; (10)
3
2
and it features a strong signal at 43 u, which indicates either
2-methyl-3-oxo-butanal (tentative, see text), (14) 1-hydroxy-3-
methylbut-3-en-2-one.
CH CO` or C H ` as fragments, or both. The signal with
3
3 7
Phys. Chem. Chem. Phys., 2000, 2, 4029È4039
4033

View Article Online
There is no elimination of CH OH, because the correspond-
2
ing signals at 69 and 31 u are absent. Accordingly, the mass
spectrum suggests an assignment to either one of the two
compounds 2-hydroxy-3-methylbut-3-enal or 2-hydroxy-2-
methylbut-3-enal. The Ðrst would be inconsistent with the
high intensity of the signal at 43 u and the absence of the
CH C2CH group fragmentation pattern. 2-hydroxy-2-
3
2
methylbut-3-enal would be more consistent with these fea-
tures, but it is not expected to be formed in high yield, because
addition of OH to the inner carbon atom of the methylated
double bond of isoprene has a low probability. We have,
therefore, considered yet another possible candidate com-
pound, namely 2-methyl-3-oxo-butanal, which also features a
parent peak at 100 u. The mass spectrum of this compound is
expected to contain fragment peaks at 85, 71 and 43 u
resulting from elimination of CH , CHO and CH CO. It
3
3
appears that this compound is most consistent with the
observed mass spectrum. We conclude that peak d10 may be
assigned to 2-hydroxy-2-methylbut-3-enal as well as to 2-
methyl-3-oxo-butanal. Whereas the former is an expected
product, the latter is unexpected. Fig. 5 shows a possible route
to its formation.
Peak d14, as the gas chromatogram in Fig. 3 shows, con-
sists of a not fully resolved doublet (14 and 14a). The
resolution achieved by GC-MS was similar to that evident in
Fig. 3, so that the mass spectrum of peak d14 may contain
contributions from two compounds. The mass of peak d14a
was complex and could not be evaluated. In contrast to peak
d10, the spectrum associated with peak d14 features a dis-
tinct parent peak at 100 u, and the fragment peak at 69 u
suggests elimination of CH OH. In this case the elimination
2
of CH OH is conÐrmed by the appearance of a signal at 31 u.
2
The mass spectrum also shows the characteristic fragmenta-
tion pattern of the CH C2CH group in the range 37 to 42 u,
3
2
with signals at 39 and 41 u being dominant. In many respects
the spectrum is similar to that of 1-hydroxy-3-methylbutan-2-
one, so that it can be conÐdently assigned to 1-hydroxy-3-
methylbut-3-en-2-one. The presence of a peak at 71 u suggests
loss of CHO. Although this may be a fragment from the iden-
tiÐed compound, we cannot preclude that it is due to contami-
nation.
Fig. 4 Mass spectra of peaks number 6, 10, 14 and methacrolein
observed as products of the oxidation of isoprene, and of 1-hydroxy-
butan-2-one and 1-hydroxy-3-methylbutan-2-one observed in the oxi-
dation of but-1-ene and 3-methylbut-1-ene, respectively.
Table 4 summarizes product yields observed in the oxida-
tion of isoprene. Because of the deliberately chosen high con-
centrations of isoprene it was not possible to derive absolute
product yields based on the consumption of isoprene. The
yields for methyl vinyl ketone (MVK), methacrolein (MAC)
and 3-methylfuran were calculated from experimental cali-
bration factors. For the products that were characterized only
by mass spectrometry, the calibration factors were estimated
by means of compounds with similar structures. The relative
yields thus obtained were found to vary little between each
run. Averaged values are shown in the last column of Table 4.
The molar ratio of methacrolein to methyl vinyl ketone is
0.93 ^ 0.10, markedly larger than that found in the presence
plus the six most prominent peaks); 1-hydroxy-3-
methylbutan-2-one: 1, 22, 43, 17, 4, 9, 4% for mass numbers
102, 71, 43, 41, 39, 31, 29, respectively; 2-hydroxy-3-methyl-
butanal: nil, 21, 15, 21, 16, 12, 15% for mass numbers 102, 73,
60, 55, 43, 41, 29, respectively. The mass spectra of the corre-
sponding hydroxy-carbonyl compounds formed in the oxida-
tion of isoprene would have parent peaks at 100 u, but
according to our experience with products from 3-methylbut-
1-ene the parent peaks are expected to be extremely weak. We
now consider this possibility. The fragment peak at 71 u may
then be assigned to the elimination of CHO and the peaks at
86 and 85 u to the elimination of CH and CH , respectively.
of NO , which is about 0.71, but it decreased slightly with
2
3
x
Fig. 5 Suggested reaction schemes for the formation of the products 3-methylbut-3-en-2-one and 2-methyl-3-oxo-butanal.
Phys. Chem. Chem. Phys., 2000, 2, 4029È4039
4034

View Article Online
Table 4 Relative product distributions obtained from the oxidation of isoprene
Yield/lmol mol~1
15 mina
Relative yield
Average
Product
Retention time/min
30 mina
Methacrolein
3-Methylfuran
Methyl vinyl ketone
Peak d4b
Peak d6c
Peak d10
5.43
5.74
6.67
7.33
8.63
14.24
17.40
3.96 ^ 0.11
0.52 ^ 0.08
4.05 ^ 0.10
7.28 ^ 0.15
1.21 ^ 0.12
8.21 ^ 0.20
0.14 ^ 0.03
0.45 ^ 0.02
2.70 ^ 0.14
1.30 ^ 0.20
21.31 ^ 0.85
0.350 ^ 0.010
0.052 ^ 0.005
0.376 ^ 0.015
È
0.22 ^ 0.02
1.50 ^ 0.08
0.80 ^ 0.12
11.05 ^ 0.51
0.021 ^ 0.001
0.131 ^ 0.007
0.067 ^ 0.010
Peak d14d
Total
a Irradiation time. b 3-Methylbut-3-enal. c 3-Methylbut-3-en-2-one. d 1-Hydroxy-3-methylbut-3-en-2-one.
time when the reaction bulb was left standing after irradiation
and before analysis. 1-hydroxy-2-methylbut-3-en-2-hydro-
was assumed to form 20% of 1,2-dihydroxy-2-methylbut-3-ene
in cross-reactions with other peroxyl radicals. Alkoxyl radicals
resulting from species C and F were assumed to undergo frag-
mentation leading to methyl vinyl ketone and methacrolein,
respectively. The alkoxyl radicals derived from A, B, E and D
may follow alternative routes of decomposition and reaction
with oxygen. The associated product distributions were
described by the parameters a, i and j, which are deÐned as
follows:
peroxide is expected to be a major product under NO -free
x
conditions (see calculations). This compound is expected to
decompose by forming methyl vinyl ketone, which would
explain the decrease in the MAC/MVK ratio.
Calculated product distribution. The FACSIMILE computer
program23 was used to calculate product distributions. The
overall reaction mechanism was similar to the scheme
described by Jenkin et al.29 with certain modiÐcations out-
lined below. A list of reactions is given in the Electronic Sup-
plementary Information.¤ The oxidation mechanism for
isoprene involves six isomeric radicals resulting from the addi-
tion of OH as shown in Fig. 6. Further addition of oxygen
leads to the corresponding peroxyl radicals. These include the
tertiary species A, a secondary radical D and the primary rad-
icals B, C, E and F. In order to account for the results of our
product studies we have tentatively added another peroxyl
radical G originating from the rearrangement of the precursor
radical of F (see Fig. 5). Self- and cross-reactions of the
primary species B and E resulting from a shift of the double
bond are expected to yield 4-hydroxymethylbutenaldehydes
and 1,4-diols in addition to 3-methylfuran. Gu et al.13 and
Paulson and Seinfeld30 assigned 3-methylfuran to arise from
isomerization of the hydroxyalkoxyl radicals derived from
species B and E, and we followed their suggestion. Jenkin et
al.29 adopted an alternative route Ðrst suggested by Atkinson
et al.,31 assuming 3-methylfuran to be formed from the
hydroxyalkoxyl radicals derived from species C and F. It is
currently not possible to decide which one of the two path-
ways is more realistic. The secondary species D is expected to
produce 1,2-hydroxyketone and 1,2-diol in addition to meth-
acrolein. Primary peroxyl radicals were assumed to yield 50%
aldehydes and alcohols and 50% alkoxyl radicals in most
cases. For the secondary species D we applied the partitioning
between ketone/alcohol and alkoxyl radical that was derived
for 3-methylbut-1-ene: 0.175 : 0.825. The tertiary radical A
A: 1-hydroxy-2-methylbut-3-en-2-oxy
a
ÈÈ methyl vinyl ketone
1~a
ÈÈ 1-hydroxybut-3-en-2-one
B: 1-hydroxy-2-methylbut-2-en-4-oxy
i
ÈÈ 3-methylfuran
1~i
ÈÈ 4-hydroxy-3-methylbut-2-enal
E: 1-hydroxy-3-methylbut-2-en-4-oxy
i
ÈÈ 3-methylfuran
1~i
ÈÈ 4-hydroxy-2-methylbut-2-enal
D: 1-hydroxy-3-methylbut-3-en-2-oxy
j
ÈÈ methacrolein
1~j
ÈÈ 1-hydroxy-3-methylbut-3-en-2-one
Calculations were carried out for two initial conditions: one
case assumed the presence of NO under conditions similar to
x
those reported by Tuazon and Atkinson,14 who used ethyl
nitrite as an indirect source of OH; in the second case NO -
x
Fig. 6 Isomeric peroxyl radicals produced by the addition of the OH radical to isoprene: AÈF are commonly assumed, G is tentative.
Phys. Chem. Chem. Phys., 2000, 2, 4029È4039
4035

View Article Online
Table 5 Calculated values for y, the site of OH addition to isoprene,
and x \ z, the degree of 1,4-isomeric shift to form radicals B and E,
for several parametersa
free conditions were assumed similar to those used for the
but-1-enes. The Electronic Supplementary Information¤ lists
the participating reactions in a simple code based on the
assignment of peroxyl radicals in Fig. 6. The associated rate
coefficients were initially assumed to be the same as in Table
2, except for the self-reactions of the radicals B and E, for
which a value of 6.8 ] 10~13 was used, similar to that given
by Jenkin et al.32 for the self-reaction of the allyl-peroxyl
radical. Subsequently, the rate coefficients reported more
recently by Jenkin et al.29 were used. The results di†ered little,
however.
The relative probability of OH addition at the inner and
outer position of the isoprene double bonds was assumed to
be the same as for the but-1-enes: 90 : 10 for the methylated
double bond and 76 : 24 for the non-methylated one. In order
to estimate the probability for the addition of OH radicals to
either of the two double bonds, we made use of the experi-
mentally observed ratios of methyl vinyl ketone to meth-
k
j
a
p
y
x \ z
0.5
0.5
0.5
0.5
0.5
0.5
0.62
1.0
0.62
1.0
1.0
1.0
1.0
1.0
0.75
0.75
1.0
1.0
1.0
1.0
1.0
0.2
0.0
0.508
0.590
0.569
0.648
0.529
0.511
0.858
0.840
0.875
0.862
0.853
0.857
1.0
a See text for deÐnitions of parameters.
methyl vinyl ketone and 3-methylfuran are: 6.44 (Tuazon and
Atkinson14), 7.19 (Gu et al.13) and 8.88 (Paulson et al.15). The
average is about 7.5. These data were then used in com-
bination with eqn. (1) and (2) to evaluate y and x \ z for
several values of the parameters a, j and p. Table 5, which
summarizes the results, shows that little variation occurs for
x \ z B 0.85, whereas y covers a wider range of values
between 0.51 and 0.65.
acrolein and to 3-methylfuran in the presence of NO . Under
x
these conditions one expects the seven peroxyl radicals shown
in Fig. 6 to be largely converted to the corresponding alkoxyl
radicals. The formation of nitrates in the reactions of peroxyl
radicals with NO, for which a uniform yield of 12% was
assumed, would not greatly alter the peroxyl radical distribu-
tion. If, as shown in Fig. 6, the probability for OH addition to
the left-hand side is designated y, the probability for 1,4-isom-
eric shift of the double bond is designated z and x, respec-
tively, and the fraction of peroxyl radical F derived from its
precursor is p, the ratio of the yield of the major products is
Table 6 summarizes the calculated product distributions.
One important result is that, in the absence of NO , methyl
x
vinyl ketone and methacrolein are predicted to occur in nearly
equal yields, which agrees with the experimental observations.
In fact, the calculated yields of methyl vinyl ketone, meth-
acrolein and 3-methylfuran di†er little, regardless of assump-
tions made in the individual runs, so that it is not possible to
distinguish between the di†erent cases that were considered by
comparing the results with experimental data.
Hydroperoxides are predicted to be formed in high yields,
in agreement with the results of Jenkin et al.29 Most of the
many other products are formed in small yields making it dif-
Ðcult to quantify them under any experimental conditions. 1-
Hydroxybut-3-en-2-one and 1-hydroxy-3-methylbut-3-en-2-
one are the only compounds that would occur in fairly high
yield under favorable conditions. The former compound was
not observed, although we should have been able to detect it
experimentally, indicating that the parameter a must be close
[MVK]
y(0.9za ] 0.1)
[MAC] (1 [ y)(0.76jx ] 0.24p)
[MVK] (y/i)(0.9za ] 0.1)
[MFU] 0.9y(1 [ z) ] 0.76(1 [ y)(1 [ x)
\
(1)
(2)
\
We have provisionally set i \ 0.5 and z \ x. The ratios of
yields [MVK]/[MAC] reported in the literature as 1.35 (Gu
et al.13), 1.38 (Tuazon and Atkinson14), 1.44 (Paulson et al.15),
1.4 (Grosjean et al.33) and 1.45 (Miyoshi et al.16). The average
value is 1.4. The ratios of yields that have been reported for
Table 6 Calculated percentage product distributions resulting from the OH-induced oxidation of isoprene in the presence and absence of NO
(source of OH is either ethyl nitritea or hydrogen peroxideb), depending on the branching factory y
x
C
H
ONO a
H O b
2
5
2
2
y
0.508 0.590 0.569 0.648 0.529 exp.c
0.508 0.590 0.569 0.648 0.529 exp.d
Methacrolein
Methyl vinyl ketone
3-Methylfuran
Formaldehyde
MAC
MVK
MFU
23.88 27.21 21.09 23.61 24.70 22 ^ 5
33.69 38.37 29.75 33.40 34.87 32 ^ 7
18.62 21.89 16.74 18.69 19.72 19.1 ^ 1.0
18.14 21.87 18.90 22.22 18.60 20.5 ^ 1.2
4.24
4.85
3.75
4.20
4.42 4.8 ^ 0.6
2.54
3.04
2.38
2.77
2.66
2.8 ^ 0.3
HCHO
58.19 66.17 59.93 67.19 60.18 63 ^ 10 36.81 43.02 38.20 44.15 38.37
1,2 Dihydroxy-2-methylbut-3-en
1,2 Dihydroxy-3-methylbut-3-en
1,4-Dihydroxy-2-methylbut-2-en
2-Hydroxy-2-methylbut-3-enal
2-Hydroxy-3-methylbut-3-enal
4-Hydroxy-2-methylbut-2-enal
4-Hydroxy-3-methylbut-2-enal
1-Hydroxy-3-methylbut-3-en-2-one
1-Hydroxybut-3-en-2-one
Methanol
AOH ] COH
DOH ] FOH
BOH ] EOH
CCHO
FCHO
ECHO
BCHO
DCO
HMVK
È
È
È
È
È
2.05
2.50
9.40
È
È
È
È
È
1.92
3.28
È
È
È
È
È
È
È
È
È
È
È
È
È
È
È
È
1.41
3.08
È
9.63
È
È
È
È
È
È
È
È
È
2.02
2.69
È
È
È
0.79
6.26
2.29
5.37
1.91
1.05
2.47
2.14
2.63
10.76
È
È
È
È
È
2.67
3.44
2.16
1.23
2.08
2.09
3.58
2.72
4.83
1.69
1.20
2.19
1.77
2.78
3.10
3.90
1.86
1.38
1.80
1.60
3.51
3.18
3.99
0.80
2.78
3.49
1.98
1.10
0.46
2.14
2.06
3.74
È
È
0.90
6.40
1.72
1.57
2.46
8.36
8.60
3.37 10.00
È
3.7 ^ 0.6
È
3.18
0.66
È
È
È
È
CH OH
È
È
È
È
È
È
È
È
È
È
3
3-Methylbut-3-en-2-one
2-Methyl-3-oxo-butanale
1-Hydroxy-2-methylbutan-3-one
Organic nitrates
È
È
È
È
1.1 ^ 0.1
7.1 ^ 0.4
GCHO
GOH
È
È
È
È
10.51 10.49 10.47 10.45 10.40 D11
È
È
OH ] MAC secondary products
OH ] MVK secondary products
1-Hydroxy-2-hydroperoxy-2-methylbut-3-ene AOOH
1-Hydroxy-2-hydroperoxy-3-methylbut-3-ene DOOH
Hydroxyhydroperoxides total
0.55
0.46
È
È
È
0.61
0.51
È
È
È
0.48
0.41
È
È
È
0.65
0.55
È
È
È
0.53
0.45
È
È
È
0.06
0.03
0.07
0.04
0.05
0.03
0.06
0.04
0.06
0.03
22.09 23.83 23.64 25.14 22.59
5.33 4.09 4.20 3.13 5.02
31.72 31.87 31.40 31.41 31.00
a Initial concentrations (unit: molecule cm~3): [C H ONO] \ 2.4 ] 1014, [NO] \ 1.9 ] 1014, [NO ] \ 2.3 ] 1013, [C H ] \ 2.4 ] 1014, after
2
5
2
5 8
4 min reaction time. b Initial concentrations (unit: molecule cm~3): [H O ] \ 3 ] 1014, [C H ] \ 2.5 ] 1016, after 15 min reaction time. c From
2
2
5 8
Tuazon and Atkinson14 as corrected by Atkinson.5 d Prorated observed product distribution. e Tentative assignment.
4036
Phys. Chem. Chem. Phys., 2000, 2, 4029È4039

View Article Online
to unity. The yield of 1-hydroxy-3-methylbut-3-en-2-one was
less than that predicted by setting the parameter j \ 0.62, so
that competition between reaction with oxygen and decompo-
sition of the alkoxyl precursor radical is less favorable than
wefound for the oxidation of 3-methylbutene. Evidently,
decomposition toward methacrolein is preferred, and 1-
hydroxy-3-methylbut-3-en-2-one is formed primarily by cross-
combination reactions of peroxyl radicals involving species D
as one of the reactants. Our assignment of peak d10 as 2-
methyl-3-oxo-butanal must be considered tentative. Neverthe-
less, when allowance is made for the formation of this product
along the suggested pathway, it is possible to bring calculated
and experimentally estimated yields of the two products
3-methylbut-3-en-2-one and 2-methyl-3-oxo-butanal into
approximate agreement.
known for alkylperoxyl radicals,9 but the radical-terminating
reaction pathway is still important. Similarly, k /(k
4
4
] k [O ]) is 0.75 ^ 0.04 and 0.62 ^ 0.04, respectively, indi-
3
2
cating that the reaction with oxygen cannot be entirely
neglected. Atkinson et al.,10 who studied the oxidation of
but-1-ene in the presence of NO , found a propanal yield of
x
0.94 ^ 0.12 relative to hydrocarbon consumption. From the
present data one calculates a propanal yield of 0.81 ^ 0.04
under similar conditions, where the range reÑects mainly the
estimated analytical uncertainties. Both values are consistent
within the margin of error. Aschmann et al.11 recently report-
ed 1-hydroxybutan-2-one to occur as a product from the OH-
induced oxidation of but-1-ene in the presence of NO . Gas
x
chromatographic analysis indicated a molar yield of 0.005 in
air at atmospheric pressure, and of 0.024 in a gas mixture con-
taining about 90% of oxygen, whereas our results would
suggest a much higher yield of about 0.19. The di†erence is
difficult to explain. In our system, the fraction of 1-hydroxy-2-
butoxyl radicals reacting with oxygen is assessed mainly from
the di†erence between the yields of hydroxyketones and diols
relative to that of the alkanal decomposition product. It does
not depend much on the probability of OH addition to either
side of the double bond. Varying the parameters in the calcu-
lations within a fairly wide range did not greatly change the
results. In order to reduce the ratio k [O ]/(k ] k [O ])
Discussion
Methyl-substituted but-1-enes
It has been possible to observe and quantify the mixed
hydroxy-carbonyl compounds and diols expected as products
from the oxidation of but-1-enes. The observed fraction of 2-
hydroxyalkanals provided an indication for the relative prob-
ability of OH addition at the two sites of the double bond
(Table 3). The behavior of but-1-ene and 3-methylbut-1-ene is
similar: OH attack occurs to 76 ^ 5% at the outer position,
to 24 ^ 5% at the inner position. With 2-methylbut-1-ene the
probability in favor of the terminal position is higher, about
90%. The probability in the case of the Ðrst two but-1-enes is
similar to that found by Tuazon and Atkinson34 for the reac-
tion of OH with methyl vinyl ketone, where OH addition to
the terminal carbon atom occurs with 72 ^ 21% probability.
The result that addition at the terminal position atom is
favored is in agreement with previous estimates based on
structural considerations.3 It is also known that the rate coef-
Ðcient for OH addition to methyl-substituted double bonds is
about twice that for unsubstituted double bonds.35 For
3
2
4
3
2
derived for but-1-ene from 0.25 to 0.1, the yield of 1-
hydroxybutan-2-one would have to be about halved or that of
butane-1,2-diol raised by about 50%. We consider this
unlikely in view of the 5È7% precision observed in calibrating
the signals. The calibration curves for 1-hydroxyketones and
diols were linear and met the zero point with little deviation
indicating that losses on the walls of the reactor were not
overly signiÐcant. Aschmann et al.11 worked with a 7900 L
TeÑon chamber, which also should have minimized wall
losses, but the gas mixture was stirred by fan so that the pro-
ducts experienced wall contact more frequently.
For the oxidation of 3-methylbut-1-ene the results of Atkin-
son et al.12 indicate that the yield of 2-methylpropanal and
glycolaldehyde combined is 0.85 ^ 0.12 relative to hydrocar-
bon consumption. From the present data we calculate a yield
of 0.75 ^ 0.05 under similar conditions assuming that the
glycolaldehyde/2-methylpropane ratio is similar. We did not
detect glycolaldehyde, although the calculations showed that
it should have been observable. If the yield of glycolaldehyde
were signiÐcantly lower than that reported by Atkinson et
example, for but-1-ene k \ 3.1 ] 10~11,for 3-methylbut-1-
OH
ene k \ 3.2 ] 10~11, whereas for 2-methylbut-1-ene k
\
OH
OH
6.5 ] 10~11 cm3 molecule~1 s~1 (ref. 3, 34). Peeters et al.,36
following earlier work of Atkinson37 aimed at establishing
group additivity rules, have argued that the di†erences in the
rate coefficients should depend mainly on the type of radical
generated, i.e. whether it is a primary, secondary or tertiary
radical, and they assigned site-speciÐc rate coefficients of
0.45 ] 10~11, 3.0 ] 10~11 and 5.5 ] 10~11 cm3 molecule~1
s~1, respectively, which they derived from the known rate
coefficients for the reactions with ethene, but-2-ene, and 2,3-
dimethylbut-2-ene.3,35 The total rate coefficient for each com-
pound is then obtained from the sum of the site-speciÐc rate
coefficients. Accepting the scheme of Peeters et al.36 leads to
the following predicted percentages for OH addition at the
outer and inner positions of the double bonds: 87 : 13 for
but-1-ene and 3-methylbut-1-ene, and 92 : 8 for 2-methylbut-
1-ene. The observed values 76 : 24 and 90 : 10 agree approx-
imately with prediction, the former less well than the latter.
Our values for c depend critically on the observed relative
yields for the 2-hydroxyalkanals. If we had overestimated their
yields from but-1-ene and 3-methylbut-1-ene by a factor of
two, the values for OH addition to the outer and inner posi-
tions of the double bond would be 0.87 : 0.13, in much better
agreement with prediction.
al.,12 the yield of 2-methylpropanal in the presence of NO
x
would be 0.71. While both results lie within common error
limits, our ratio k /(k ] k [O ]) is lower than that derived
4
4
3
2
from the data of Atkinson et al.12 by a margin similar to that
observed for the oxidation of but-1-ene.
The assumption, based on data available for several alkoxyl
radicals,3,5 that the rate coefficient for the reaction with
oxygen is approximately 8 ] 10~15 cm3 molecule~1 s~1 sug-
gests values for k of 1.3 ] 105 s~1 and 0.7 ] 105 s~1 for the
4
decomposition of the 1-hydroxybutan-2-oxy radicals resulting
from but-1-ene and 3-methylbut-1-ene, respectively. If glycol-
aldehyde were formed to the extent observed by Atkinson et
al.,12 the latter rate coefficient would be D1.0 ] 105 s~1. The
rate constant for decomposition of the hydroxyethoxyl radical
is similar, namely 1.5 ] 105 s~1 (ref. 3, 6, 7). Orlando et al.7
have shown that in this case about 25% of the HOCH CH O
2
2
radicals decompose promptly while the others are thermalized
and undergo competition between decomposition and reac-
tion with oxygen. For 1-hydroxypropyl-2-oxy radicals derived
from propene Vereecken et al.38 calculated that 80% decom-
pose promptly, while the others are thermalized but also
decompose at a high rate, 2.4 ] 107 s~1 at 298 K and atmo-
The product distributions observed for the oxidation of
but-1-ene and 3-methylbut-1-ene provide branching ratios for
the self-reaction of secondary hydroxy-alkylperoxyl radicals
and relative probabilities k /(k ] k [O ]) for the break-up of
4
4
3
2
hydroxyalkoxyl radicals vs. their reaction with oxygen. The
spheric pressure, whereas Atkinson39 calculated a total
data in Table 3 suggest values for k /k of 0.29 ^ 0.07 and
decomposition rate of D1 ] 105 s~1, more in line with the
ssb ss
0.19 ^ 0.06, respectively. These values are lower than those
above data.
Phys. Chem. Chem. Phys., 2000, 2, 4029È4039
4037

View Article Online
Isoprene
that of methyl vinyl ketone is much smaller initially and rises
with time until it approaches that of methacrolein. Accord-
ingly, the ratio of both yields changes with time. The experi-
mental data conÐrmed this behavior by a slight decrease of
the MAC/MVK ratio when the irradiation time was increased
from 15 to 30 min. The reason is that tertiary peroxyl radicals
Our search for products that are suggested to arise from the
oxidation of isoprene in addition to methyl vinyl ketone,
methacrolein and 3-methylfuran has been only partly suc-
cessful, but we were able to identify the 1,2-hydroxy-carbonyl
compound that according to our mechanism is expected to
arise from species D (see Fig. 6). Comparison with the results
of calculations indicates that decomposition of the 1-hydroxy-
3-methylbut-3-en-2-oxy radical to form methacrolein is pre-
ferred over its reaction with oxygen. Ruppert and Becker19
reached the same conclusion because they found that varying
the concentration of oxygen has little inÑuence on the yield of
methacrolein. Another identiÐed product is 2-methylbut-1-en-
3-one. The mass spectrum of peak d10indicates 2-methyl-3-
oxo-butanal to be a major product. We reiterate that identiÐ-
cation of the latter product is tentative. Paulson and
Seinfeld30 have postulated a similar product, namely 1-
hydroxybut-1-en-3-one, to be formed from species E. This
product would undergo enolisation and relax to 3-oxo-
butanal. The mass spectrum observed for peak d10 was not
consistent with 3-oxo-butanal, however. Provided our inter-
pretation of the mass spectrum of peak d10 is correct, the
products assigned to peaks numbered 6 and 10 suggest a
rearrangement of the radical preceding the formation of
species F (see Fig. 5), so that another species G would be
formed. The calculations show that by including this route
into the mechanism it is possible to simulate the experimen-
tally observed yields of peaks numbered 6 and 10 in Fig. 3, if a
partitioning between F and G is chosen that favors G.
There was no evidence for the formation of 1-hydroxybut-3-
en-2-one, so that we cannot support the suggestion of Jenkin
et al.29 for a loss of methyl from the 1-hydroxy-2-methylbut-3-
en-2-oxyl radical at high yield. The generation of methyl rad-
icals in the system is considered necessary for the formation of
methanol, which Ruppert and Becker19 found to occur with a
yield of 1.7 ^ 0.3% in the absence of NO . Neither the calcu-
lations of Jenkin et al.29 nor our own calculations, which are
based on similar assumptions, led to methanol yields of the
required magnitude. An alternative route to the formation of
methyl radicals may be a decomposition of the 2-hydroxy-2-
methylbut-3-enyl radical resulting from OH addition to iso-
prene at the 2-position (the precursor to species C in Fig. 6).
of type A initially prefer to react with HO to form hydro-
2
peroxide before their concentration builds up sufficiently to
support the self-reaction and cross-reactions with other
peroxyl radicals. The yield of tertiary hydroperoxide, in turn,
is calculated to be high initially, before it decreases as methyl
vinyl ketone builds up. Calculated relative yields of meth-
acrolein, methyl vinyl ketone and 3-methylfuran in the
absence of NO depend somewhat on the parameter y, which
x
was derived from the observed relative yields in the presence
of NO . Although the calculated relative yields agree well
x
with the experimental data regardless of the assumptions
made, the lowest MAC/MVK ratio, about 0.86, was found for
a \ 0.75. The alternative route to the formation of methyl
radicals discussed above would not a†ect the MAC/MVK
ratio.
Peeters et al.27 predicted for the addition of OH to isoprene
a probability y \ 0.65 at the methylated double bond and
1 [ y \ 0.35 at the non-methylated one. They assumed that
the addition occurred mainly at the outer ends of the double
bond and only to 5% each at the inner position. Jenkin et
al.,29 who adopted these factors in their calculations, found it
necessary to assume a \ 0.75 and a correspondingly high yield
of 1-hydroxybut-3-en-2-one in order to account for [MVK]/
[MAC] B 1 in the absence of NO . We have applied our
x
results obtained for the but-1-enes, that is, we have assumed a
ratio of 90 : 10 for the addition of OH to the methylated
double bond (outer and inner position) and a ratio of 76 : 24
to the non-methylated bond, taking y to be a parameter deter-
mined from the [MVK]/[MAC] ratio in the presence of NO .
x
The highest value, y \ 0.65, was obtained with a \ 0.75.
However, the observed yield of 1-hydroxybut-3-en-2-one was
too low as that the high value y \ 0.65 can be supported. The
lowest value, y \ 0.51, can also be disregarded in view of the
fairly low yield of 1-hydroxy-3-methylbut-3-en-2-one observed
compared to that predicted by the calculations based on
j \ 0.62. It appears that 0.53 \ y \ 0.59, the precise value
depending on the choice of mechanism. These results suggest
the following probabilities for OH addition to the four
unsaturated carbon atoms in isoprene (see Fig. 6):
0.504 ^ 0.027 (Ðrst), 0.056 ^ 0.003 (second), 0.105 ^ 0.008
(third), 0.335 ^ 0.023 (fourth). The value for y clearly is some-
what smaller than that predicted by Peeters et al.36 but prob-
ably still within the common error margins. It remains to be
veriÐed whether conjugated alkadienes can be assumed to
behave in a similar way as alkenes, so that the relative prob-
abilities of OH addition to the individual carbon atoms in
isoprene suggested here require further conÐrmation.
x
~CH3
~CH C(OH)(CH )CH2CH ÈÈÈ CH 2C(OH)CH2CH
2
3
2
2
2
CH 2C(OH)CH2CH ] CH COCH2CH
2
2
3
2
where 2-hydroxybuta-1,3-diene formed as the second product
would undergo enolisation to methyl vinyl ketone.
Ruppert and Becker19 also observed 1,2-hydroxy-2-
methylbut-3-ene and 1,2-hydroxy-3-methylbut-3-ene as pro-
ducts with yields of 4.7 ^ 1.4% and 2.4 ^ 0.9%, respectively.
Our calculations indicated yields of 2.3È3.1% for the former
compound and of 3.4È5.4% for the latter. The Ðrst is smaller
and the second greater than the observed values. Jenkin et
al.29 found yields of 3.9% and 1.7%, which agree better with
the observations. The formation of 1,2-hydroxy-2-methylbut-
3-ene occurs predominantly in cross-reactions of the tertiary
peroxyl radical A, that of 1,2-hydroxy-3-methylbut-3-ene
arises mainly from the self- and cross-reactions of the second-
ary peroxyl radical D, and the yields of both diols depend
greatly on the chosen branching ratios.
The following % yields of methacrolein and methyl vinyl
ketone were previously reported: Miyoshi,16 22 and 17,
Ruppert and Becker,19 17.8 ^ 1.4 and 15.3 ^ 1.2. The corre-
sponding ratios are 1.29 and 1.16 ^ 0.18, respectively. The
ratio observed here is 0.93 ^ 0.10, which is smaller but still in
agreement with the previous data within common error
margins. The calculations show that while the yield of meth-
Acknowledgements
Financial support of this work by the German Federal Minis-
try for Research and Technology is gratefully acknowledged.
References
1
P. Warneck, Chemistry of the Natural Atmosphere, Academic
Press, San Diego, 2nd edn., 1999.
2
B. J. Finlayson-Pitts and J. N. Pitts, Jr.., Chemistry of the Upper
and L ower Atmosphere: T heory, Experiments and Applications,
Academic Press, San Diego, 1999.
3
4
5
6
R. Atkinson, J. Phys. Chem. Ref. Data, Monograph 2, 1994.
J. Eberhard and C. J. Howard, J. Phys. Chem., 1997, 101, 3360.
R. Atkinson, J. Phys. Chem. Ref. Data, 1997, 26, 215.
I. Barnes, K. H. Becker and L. Ruppert, Chem. Phys. L ett., 1993,
203, 295.
acrolein (in the absence of NO ) is fairly constant with time,
x
4038
Phys. Chem. Chem. Phys., 2000, 2, 4029È4039

View Article Online
7
J. J. Orlando, G. S. Tyndall, M. Bilde, C. Ferronato, T. J. Wall-
ington, L. Vereecken and J. Peeters, J. Phys. Chem. A, 1998, 102,
8116.
E. S. C. Kwok, R. Atkinson and J. Arey, Environ. Sci. T echnol.,
1996, 30, 1048.
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. A, 1992, 26, 1805.
24 T. J. Wallington, P. Dagaut and M. J. Kurylo, Chem. Rev., 1992,
92, 667.
25 S. Madronich and J. G. Calvert, J. Geophys. Res., 1990, 95, 5697.
21.
8
9
26 M. E. Jenkin and G. D. Hayman, J. Chem. Soc., Faraday T rans.,
1995, 91, 1911.
27 A. A. Boyd and R. Lesclaux, Int. J. Chem. Kinet., 1997, 29, 323.
28 Eight Peak Index of Mass Spectra, The Royal Society of Chem-
istry, The Mass Spectrometric Data Centre, Nothingham, Unwin
Brothers Ltd, Old Woking, Surrey, 3rd edn., 1983, vol. 1, p. 22.
29 M. E. Jenkin, A. A. Boyd and R. Lesclaux, J. Atmos. Chem., 1998,
29, 267.
10 R. Atkinson, E. C. Tuazon and W. P. L. Carter, Int. J. Chem.
Kinet., 1985, 17, 725.
11 S. M. Aschmann, J. Arey and R. Atkinson, Environ. Sci. T echnol.,
2000, 34, 1702.
12 R. Atkinson, E. C. Tuazon and S. M. Aschmann, Int. J. Chem.
Kinet., 1998, 30, 577.
30 S. E. Paulson and J. H. Seinfeld, J. Geophys. Res., 1992, 97,
20 703.
13 C. L. Gu, C. M. Rynard, D. G. Hendry and T. Mill, Environ. Sci.
T echnol., 1985, 19, 151.
31 R. Atkinson, S. M. Aschmann, E. C. Tuazon, J. Arey and B. Ziel-
inska, Int. J. Chem. Kinet., 1989, 21, 593.
14 E. C. Tuazon and R. Atkinson, Int. J. Chem. Kinet., 1990, 22,
32 M. E. Jenkin, T. P. Murrells, S. J. Shalliker and G. D. Hayman,
J. Chem. Soc., Faraday T rans., 1993, 89, 433.
1221.
15 S. E. Paulson, R. C. Flagan and J. H. Seinfeld, Int. J. Chem.
Kinet., 1992, 24, 79.
33 D. Grosjean, E. L. Williams II and E. Gorjean, Environ. Sci.
T echnol., 1993, 27, 830.
16 A. Miyoshi, S. Hatakeyama and N. Washida, J. Geophys. Res.,
1994, 99, 18 779.
34 E. C. Tuazon and R. Atkinson, Int. J. Chem. Kinet., 1989, 21,
1141.
17 H. Yu, H. E. Je†ries and R. M. Le Lacheur, Environ. Sci.
T echnol., 1995, 29, 1923.
35 T. Ohta, Int. J. Chem. Kinet., 1984, 16, 879.
36 J. Peeters, W. Boullart and J. Van Hoeymissen, in T ransport and
T ransformation of Pollutants in the T roposphere, EUROT RAC
Symp. Ï94, ed. P. M. Borrell, P. Borrell, T. Cvitas and W. Seiler,
SPB Academic Publishing, Den Haag, The Netherlands, 1994, p.
110.
18 E. S. C. Kwok, R. Atkinson and J. Arey, Environ. Sci. T echnol.,
1995, 29, 2467.
19 L. Ruppert and K. H. Becker, Atmos. Environ., 2000, 34, 1529.
20 G. Heimann and P. Warneck, J. Phys. Chem., 1992, 96, 8403.
21 R. Criegee, E. Hoger, G. Huber, P. Kruck, F. Marktsche†el and
H. Schellenberger, L iebigs Ann. Chem., 1956, 599, 81.
22 E. Pfeil and H. Barth, L iebigs Ann. Chem., 1955, 593, 81.
23 A. R. Curtis and W. P. Sweetenham, FACSIMIL E/
CHECKMAT Users Manual, Aere-R-12805, Her MajestyÏs Sta-
tionary Office, London, 1988.
37 R. Atkinson, Int. J. Chem. Kinet., 1987, 19, 799.
38 L. Vereecken, J. Peeters, J. J. Orlando, G. S. Tyndall and C. Fer-
ronato, J. Phys. Chem., A, 1999, 103, 4693.
39 R. Atkinson, Int. J. Chem. Kinet., 1997, 29, 99.
Phys. Chem. Chem. Phys., 2000, 2, 4029È4039
4039