OH-initiated oxidation of benzene part II. Influence of elevated NOx concentrations

Article abstract of DOI:10.1039/b204398j

The present work represents a continuation of part I of this series of papers, in which we investigated the phenol yields in the OH-initiated oxidation of benzene under conditions of low to moderate concentrations of NO_x, to elevated NO_x levels. The products of the OH-initiated oxidation of benzene in 700-760 Torr of N₂/O₂ diluent at 297 ± 4 K were investigated in 3 different photochemical reaction chambers. In situ spectroscopic techniques were employed for the detection of products, and the initial concentrations of benzene, NO_x, and O₂ were widely varied (by factors of 6300, 1500, and 13, respectively). In contrast to results from previous studies, a pronounced dependence of the product distribution on the NO_x concentration was observed. The phenol yield decreases from approximately 50-60% in the presence of low concentrations (<5 ppb) of NO_x to values below 5% in the presence of extremely high (>10 000 ppb) NO_x concentrations. In the presence of high concentrations of NO_x, the phenol yield increases with increasing O₂ partial pressure. The rate constant of the reaction of hydroxycyclohexadienyl peroxyl radicals with NO was determined to be (1.7 ± 0.6) x 10^-11 cm³ molecule^-1 s^-1. This reaction leads to the formation of E,E-2,4-hexadienedial as the main identifiable product (29 ± 16%). The reaction of the hydroxycyclohexadienyl radical with NO₂ gave phenol (5.9 ± 3.4%) and E,E-2,4-hexadienedial (3.4 ± 1.9%), no other products could be identified. The residual FTIR product spectra indicate the formation of unknown nitrates or other nitrogen-containing species in high yield. The results from the present work also show that experimental studies aimed at establishing/verifying chemical mechanisms for aromatic hydrocarbons must be performed using NO_x levels which are representative of those found in the atmosphere.

Full text of DOI:10.1039/b204398j

View Article Online / Journal Homepage / Table of Contents for this issue
OH-initiated oxidation of benzene
Part II.y Influence of elevated NOx concentrations
a
b
c
d
¨
Bjorn Klotz,*z Rainer Volkamer, Michael D. Hurley, Mads P. Sulbaek Andersen,
Ole John Nielsen,e Ian Barnes,f Takashi Imamura,a Klaus Wirtz,g Karl-Heinz Becker,f
Ulrich Platt,b Timothy J. Wallingtonc and Nobuaki Washidah
a
National Institute for Environmental Studies, Atmospheric Environment Division, 16-2 Onogawa,
Tsukuba-shi, Ibaraki-ken 305-8506, Japan. E-mail: bjoern.klotz@nies.go.jp
Universita¨t Heidelberg, Institut fu¨r Umweltphysik, Im Neuenheimer Feld 229, D-69120,
Heidelberg, Germany
Ford Motor Co., 20000 Rotunda Drive, Mail Drop SRL-3083, Dearborn, Michigan 48121-2053,
USA
University of Southern Denmark, Department of Chemistry, Campusvej 55, DK-5230,
Odense M, Denmark
Department of Chemistry, University of Copenhagen, Universitetsparken 5, DK-2100,
Copenhagen, Denmark
Bergische Universita¨t Wuppertal, Physikalische Chemie - FB 9, Gaußstraße 20, D-42097,
Wuppertal, Germany
b
c
d
e
f
g
´
´
Fundacion Centro de Estudios Ambientales del Mediterraneo (CEAM), Parque Tecnologico,
Calle Charles R. Darwin 14, E-46980, Paterna (Valencia), Spain
Kyoto University, Department of Chemistry, Graduate School of Science, Sakyo-ku,
Kyoto 606-8502, Japan
h
Received 7th May 2002, Accepted 10th July 2002
First published as an Advance Article on the web 13th August 2002
The present work represents a continuation of part I of this series of papers, in which we investigated the phenol
yields in the OH-initiated oxidation of benzene under conditions of low to moderate concentrations of NOx , to
elevated NOx levels. The products of the OH-initiated oxidation of benzene in 700–760 Torr of N2/O2 diluent at
297 ꢀ 4 K were investigated in 3 different photochemical reaction chambers. In situ spectroscopic techniques
were employed for the detection of products, and the initial concentrations of benzene, NOx , and O2 were
widely varied (by factors of 6300, 1500, and 13, respectively). In contrast to results from previous studies,
a pronounced dependence of the product distribution on the NOx concentration was observed. The phenol yield
decreases from approximately 50–60% in the presence of low concentrations (<5 ppb) of NOx to values below
5% in the presence of extremely high (>10 000 ppb) NOx concentrations. In the presence of high concentrations
of NOx , the phenol yield increases with increasing O2 partial pressure. The rate constant of the reaction of
hydroxycyclohexadienyl peroxyl radicals with NO was determined to be (1.7 ꢀ 0.6) ꢁ 10ꢂ11 cm3 moleculeꢂ1 sꢂ1
.
This reaction leads to the formation of E,E-2,4-hexadienedial as the main identifiable product (29 ꢀ 16%). The
reaction of the hydroxycyclohexadienyl radical with NO2 gave phenol (5.9 ꢀ 3.4%) and E,E-2,4-hexadienedial
(3.4 ꢀ 1.9%), no other products could be identified. The residual FTIR product spectra indicate the formation
of unknown nitrates or other nitrogen-containing species in high yield. The results from the present work also
show that experimental studies aimed at establishing/verifying chemical mechanisms for aromatic
hydrocarbons must be performed using NOx levels which are representative of those found in the atmosphere.
pollution. Unfortunately, despite numerous studies, substan-
1. Introduction
tial uncertainties persist in our understanding of the atmo-
spheric chemistry of aromatic compounds, making them a
priority field for mechanistic studies.3,4 The simplest aromatic
hydrocarbon, benzene, reacts exclusively with OH radicals
under tropospheric conditions. This reaction is rather slow
when compared to those of alkylbenzenes, the rate constant
kOH+benzene ¼ 1.2 ꢁ 10ꢂ12 cm3 moleculeꢂ1 sꢂ1 is a factor of
five to fifty lower than kOH of toluene and trimethylbenzenes,
respectively.5 The relatively low reactivity of benzene gives
The importance of aromatic hydrocarbons in local and regio-
nal air pollution is well established.1 Aromatic hydrocarbons
contribute significantly to the formation of ozone and second-
ary organic aerosols in urban air.1,2 An understanding of the
atmospheric chemistry of aromatic compounds is therefore
crucial for our understanding of the chemistry governing air
y For Part I see ref. 7.
rise to
a significant residence time in the atmosphere
z Present address: Bergische Universita¨t Wuppertal, Physikalische
Chemie - FB9, Gaußstraße 20, D-42097, Wuppertal, Germany. E-mail:
E-mail: bjoern.klotz@uni-wuppertal.de
(several weeks) and consequently a rather low photochemical
activity. Benzene oxidation is therefore not a major source of
DOI: 10.1039/b204398j
Phys. Chem. Chem. Phys., 2002, 4, 4399–4411
This journal is # The Owner Societies 2002
4399

View Article Online
Fig. 1 Proposed pathways for the initial steps in the OH-radical initiated degradation of benzene in the gas-phase. Thick lines indicate the path-
ways studied in this work, thin lines those studied by Volkamer et al.7 and dashed lines those probably not operative.7 ‘‘Other products’’ denotes
products other than phenol.
photooxidants in urban air. However, there is another incen-
tive to study the benzene + OH reaction. At ambient tempera-
tures this reaction proceeds almost exclusively via addition of
OH to the aromatic ring, see reaction (1) of Fig. 1. The general
mechanistic conclusions from benzene studies serve as a model
for the OH addition pathway for the alkylated monocyclic aro-
matics, which are considerably more reactive.6
degradation channels of 2/3. The formation of phenol follow-
ing OH initiated oxidation of benzene in the presence of sev-
eral ppm NOx was investigated by Atkinson et al.,16 but no
significant dependence of the phenol yield on NOx concentra-
tion was found and a phenol yield of (23.6 ꢀ 4.4)% was
reported. In recent studies by Bjergbakke et al.10 and Berndt
et al.,13 phenol yields of (25 ꢀ 5)% and (23 ꢀ 7)%, respectively,
were reported. These later studies were conducted under NOx
free conditions, apparently confirming the absence of a NOx
dependence of the phenol yield. Similar results were obtained
by Berndt and Bo¨ge,17 who reported results of an investigation
of the OH initiated degradation of benzene as a function of
NOx concentration, pressure and temperature.
In part I of this series, Volkamer et al.7 extensively reviewed
the initial steps of benzene oxidation and investigated the for-
mation of phenol under atmospheric conditions (i.e. low to
moderate NOx levels). In Fig. 1 these initial steps are re-drawn
to summarize our previous conclusions and illustrate the con-
nection of the present work to part I. Briefly, addition of OH
results in the formation of a hydroxycyclohexadienyl radical 2,
which under atmospheric conditions will reversibly add O2 to
give a hydroxycyclohexadienyl peroxyl radical 3. This equili-
brium adjusts rapidly8 and a multitude of subsequent reaction
channels have been proposed, see reactions (3)–(12) in Fig. 1.
Under most atmospheric conditions species 2 and 3 are lost
predominately via reactions involving molecular oxygen
through channels (3), (4), (6), (8) and (9).8,9 Of these, channels
(3) and (4) lead to the formation of phenol. Volkamer et al.7
presented a method to distinguish both pathways on the basis
of temperature dependent considerations and concluded that
the majority of phenol is probably formed via channel (3).
The experimental evidence10 for channel (5) has been dis-
puted8,11,12 and it is unlikely that this pathway is actually
operative. Channel (6) is now believed to be of minor, or neg-
ligible, importance.7,13,14
However, our re-calibration7 of the studies of Bjergbakke
et al.10 and Berndt et al.13 showed that considerable scatter
exists among the apparently conclusive values if the data are
placed on a common basis of calibration.18 Further, we deter-
mined the phenol yield from the reaction of benzene with OH
radicals under realistic atmospheric conditions to be
(53.1 ꢀ 6.6)%.7 Part of that difference of more than a factor
of two was attributed to the influence of channels (10) and
(12) which were expected to decrease the phenol yield under
conditions of elevated NOx . However, for high NOx condi-
tions (several ppm) the predicted dependence of the phenol
yield on the NOx concentration was found to contrast with
the rather constant yield found by Atkinson et al.16
Here we investigate the formation of phenol and hexadiene-
dials in the OH-initiated oxidation of benzene under condi-
tions where the loss processes of 2/3 are dominated by
reactions involving NOx rather than oxygen. Even though such
conditions are unlikely to be observed in the atmosphere, they
have been widely applied in smog chamber studies in the past.
The objective of this study is to elucidate whether the reaction
steps drawn in Fig. 1 allow a general description of the initial
steps of benzene oxidation in the presence of NOx . The NO,
NO2 and oxygen concentrations were varied systematically
to differentiate between these reaction channels and to obtain
insight into their products.
In the presence of high concentrations of NOx channels (10),
(11) and (12) become significant loss processes for species 2/3.
Channel (11) is negligible under all but the most extreme
laboratory conditions.9,15 Channels (10) and (12) are impor-
tant loss processes for species 2/3 at NOx concentrations
above 0.1 ppm8,9 (parts per million, 1 ppm ¼ 2.46 ꢁ 1013
cmꢂ3 at standard temperature and pressure). The dependence
of the phenol formation yield on the NOx concentration has
the potential to offer valuable insights into the various further
4400
Phys. Chem. Chem. Phys., 2002, 4, 4399–4411

View Article Online
NIES have been discussed previously7 and are re-stated in
Table 1 for ease of reference.
2. Experimental
The experimental data reported herein were obtained using
three different experimental facilities; indoor photoreactors at
NIES in Tsukuba/Japan and at FORD in Dearborn/USA,
and an outdoor chamber (EUPHORE) at CEAM in Valen-
cia/Spain. Detailed descriptions of the EUPHORE,7,19
NIES7,20 and FORD21 facilities are available in the literature.
Under the high NOx concentrations employed in many of
the NIES experiments, the reaction of OH radicals with NO2
acted as an efficient scavenger of OH, which resulted in very
low benzene degradation rates. To alleviate this, most of the
high-NOx experiments were conducted with added CH3ONO
(1–2 ppm) (experiment type (6)). In the presence of ultraviolet
light, NO and molecular oxygen, methyl nitrite forms OH radi-
cals through the following sequence of reactions:
2.1 NIES reaction chamber
The NIES photochemical reaction chamber is a 6.065 m3
Teflon-coated steel vessel with a surface/volume (S/V) ratio
of 3.7 mꢂ1. It is equipped with a high-performance pump sys-
tem for efficient cleaning and can be temperature controlled
over the range 15–90 ꢃC. The light source is a solar simulator
that consists of nineteen 1 kW high pressure Xe arc lamps
which were operated at 80% of their maximum power. The
lamps are fitted with 2 mm Pyrex glass filters (cutoff wave-
length 290 nm) to simulate the sunlight reaching the lower
troposphere. The photolysis frequency of NO2 in this system
CH3ONO þ hn ! CH3OꢄþNOꢄ
CH3OꢄþO2 ! CH2O þ HO2ꢄ
HO2ꢄþNOꢄ! OHꢄþNO2ꢄ
The NIES benzene/CH3ONO/NOx experiments can be
divided into 2 subgroups. In the first group the NOx was lar-
gely present as NO initially, and NO to NO2 conversion
occurred during the irradiation. Initial NO concentrations of
up to 2.25 ꢁ 1014 molecule cmꢂ3 were employed. These experi-
ments were designed to study the impact of elevated concentra-
tions of NO on the OH-initiated oxidation of benzene. Under
these experimental conditions, reaction of the hydroxycyclo-
hexadienyl peroxyl radicals 3 with NO was the major loss pro-
cess for 2/3. In the second group of experiments a large excess
of NO2 was added to the initial mixture and excess NO2 was
maintained throughout the run. These conditions were
designed to investigate the impact of high NO2 concentrations
on the OH-initiated oxidation of benzene, reaction of the
hydroxycyclohexadienyl radical 2 with NO2 would be the
dominating loss process of 2/3 here. To scavenge O3 and
NO3 radicals, NO (ca. 2.5 ꢁ 1013 molecule cmꢂ3) was added
to the high-NO2 experiments.
In this type of experiment, the photolysis of NO2 can lead to
the formation of significant amounts of ozone, which in turn
can react with the excess NO2 to form NO3 radicals. Phenol
reacts rapidly with NO3 radicals22 so the experimental condi-
tions had to be adjusted to minimize the importance of this
reaction. Using a simple steady state analysis19 and the rele-
vant reaction rate coefficients of Atkinson5 it can be calculated
that for experiment NB40, which had the highest average
ozone and consequently the highest expected NO3 concentra-
tion, about 3 ꢁ 106 molecule cmꢂ3 of NO3 are formed. At this
concentration the loss of phenol through NO3 reaction is <5%
of that through OH reaction, as calculated from the average
OH concentration and the rate constant for the reaction of
phenol with OH.5 Reaction of phenol with NO3 radicals is
therefore concluded to be negligible under the present experi-
mental conditions. Other reactions that might influence the
results are those of intermediates with ozone or HO2/RO2
radicals. However, the high NO concentrations employed here
effectively suppress the formation of these species in the experi-
ments. In the high-NO2 experiments, the maximum ozone con-
centration is around 1012 molecule cmꢂ3, in the high-NO
experiments ozone is not detectable (<1011 molecule cmꢂ3).
In the low-NOx experiments already presented in part I,7 max-
imum ozone concentrations reached (5–10) ꢁ 1012 molecule
cmꢂ3. Even these comparably high concentrations did not
influence the measured phenol yields. Due to the slow build-
up and the widely varying maximum concentrations of ozone,
product yields would otherwise have been expected to vary
considerably, not only between experiments but also within
individual runs.
is J(NO2) ¼ 4.25 ꢁ 10ꢂ3
s
ꢂ1, while that of methyl nitrite is
J(CH3ONO) ¼ 8.7 ꢁ 10ꢂ4 sꢂ1
.
Concentrations of reactants and products were determined
using an FTIR spectrometer Nicolet Nexus 670, which was
operated at a spectral resolution of 1 cmꢂ1 and coupled to a
White-type mirror system allowing an absorption path of
221.5 m. The spectrometer was equipped with a liquid nitrogen
cooled MCT (mercury cadmium tellurium) detector.
2.2 EUPHORE reaction chamber
The outdoor European Photo-Reactor (EUPHORE) is part of
the Centro de Estudios Ambientales del Mediterraneo
(CEAM), located near Valencia/Spain. The hemispherical
FEP chamber used in the present study had a volume of
ca. 187 m3 and an S/V ratio of 1 mꢂ1. FTIR spectrometry
and differential optical absorption spectroscopy (DOAS) were
the primary analytical techniques employed. The irradiation
source for the experiments is natural sunlight.
2.3 FORD reaction chamber
The reaction chamber at FORD Motor Co. is an evacuable
140 L Pyrex glass vessel with a surface to volume ratio of
14.4 mꢂ1. 22 Fluorescent blacklamps (GE F15T8-BL) provide
the light source, the irradiation intensity corresponds to an
NO2 photolysis frequency of J(NO2) ¼ 1.2 ꢁ 10ꢂ3
s
ꢂ1. Pro-
ducts and reference compounds were monitored in situ by
FTIR spectroscopy using a spectral resolution of 0.25 cmꢂ1
and an analysing path-length of 27 m. The experiments were
conducted using an O2 partial pressure of 50–650 Torr in
700 Torr total pressure of N2 diluent. Experiments were per-
formed at a temperature of (296 ꢀ 2) K.
The FTIR and DOAS absorption cross sections used in the
evaluations of benzene and phenol in all three laboratories
were taken from Etzkorn et al.18 NO and NO2 were deter-
mined by chemiluminescence in the EUPHORE experiments
and by FTIR spectrometry at NIES and FORD. Integrated
FTIR absorption cross sections of NO and NO2 were deter-
mined independently in both laboratories and were in agree-
ment within 4%.
2.4 Types of experiments performed
The NIES experiments were performed in (760 ꢀ 20) Torr of
purified air at (298 ꢀ 3) K, irradiation times were 2–4 h for the
benzene/NOx experiments and 45–60 min for the benzene/
CH3ONO/NOx experiments. Spectra were recorded by co-
adding 100 scans in the benzene/CH3ONO/NOx experiments
and 270 scans in the benzene/NOx experiments. All but 2 of
The experimental type refers to the OH-radical source applied
in the different experiments, the nomenclature used is consis-
tent with that used in part I of this series.7 Initial conditions
of the complete data set are listed in Table 1. The experiments
performed under low-NOx conditions at EUPHORE and
Phys. Chem. Chem. Phys., 2002, 4, 4399–4411
4401

View Article Online
Table 1 Summary of the conditions and results of the experiments conducted
Experiment Experiment [benzene]initial
1013 cmꢂ3
/
Tracer
compound
[CH3ONO]initial
1013 cmꢂ3
/
code
type
[NO]avg./ppb [NO2]avg./ppb Yphenol/% Yphenolcorr/%
NIES
NBE1
NBE2
NBE3
NBE4
NB17
(1)
(1a)
(6)
(6)
(6)
(6)
(1)
(1)
(1)
(6)
(6)
(6)
(6)
(6)
(6)
(6)
(6)
(6)
(6)
(6)
(6)
(6)
(1)
12.2
12.3
5.0
Di-n-butyl ether
—
858
638
170
350
33.5
34.6
25.0
25.8
18.1
16.7
42.7
44.8
43.1
14.6
11.7
5.1
33.0 ꢀ 5.4
33.7 ꢀ 2.9
23.8 ꢀ 4.9
24.4 ꢀ 2.1
15.9 ꢀ 1.4
15.0 ꢀ 1.4
42.4 ꢀ 4.3
44.6 ꢀ 5.1
42.8 ꢀ 6.4
12.8 ꢀ 1.3
10.0 ꢀ 1.3
3.8 ꢀ 1.8
Di-n-butyl ether 0.10
2.5
—
1288
1217
1606
3686
97
627
835
12.3
17.8
18.4
18.6
18.3
18.6
18.5
18.4
18.8
18.2
18.1
24.6
48.2
48.6
48.9
48.8
48.6
42.1
48.2
24.8
Di-n-butyl ether 2.3
Di-n-butyl ether 5.3
Di-n-butyl ether 5.2
Di-n-butyl ether 0.02
Di-n-butyl ether 0.02
Di-n-butyl ether 0.02
Di-n-butyl ether 5.2
Di-n-butyl ether 5.2
Di-n-butyl ether 5.5
Di-n-butyl ether 5.3
Di-n-butyl ether 2.7
Di-n-butyl ether 7.2
Di-n-butyl ether 7.5
Di-n-butyl ether 7.5
Di-n-butyl ether 7.7
Di-n-butyl ether 7.6
Di-n-butyl ether 7.6
Di-n-butyl ether 7.8
Di-n-butyl ether 7.7
1634
1687
95
NB18
NB19
NB20
NB21
35
170
57
123
NB22
NB23
2734
4294
8426
7681
3678
1320
243
1509
1789
2032
1844
6172
3660
2279
3503
1855
2451
3351
4202
4805
41
NB24
NB25
8.2
8.1
6.9 ꢀ 1.8
NB29
NB30
4.7 ꢀ 3.7
17.7
20.5
21.5
20.6
21.5
18.6
16.6
14.9
42.1
14.4 ꢀ 1.7
17.4 ꢀ 1.7
18.3 ꢀ 1.7
17.8 ꢀ 1.7
18.5 ꢀ 1.5
15.2 ꢀ 1.5
13.2 ꢀ 1.6
11.5 ꢀ 1.5
42.0 ꢀ 6.9
NB31
NB32
1215
382
NB33
NB34
446
889
NB40
NB41
1534
2249
20
NB42
NB43
—
0.02
EUPHORE
BEN1
BEN2
BEN3
BEN4
BEN5
BEN6
BEN7
BEN8
BEN9
BE10
(1)
(1)
(1)
(1)
(1)
(1)
(1)
(1)
(1)
(1)
(1)
(2)
(2)
(4)
(5)
(1)
(1)
(5)
(3)
(3)
(3)
(3)
(3)
(3)
(3)
(3)
5.8
2.9
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
18
54
33
32
47
33
78
75
55
21
30
16
19
2
48.0
49.9
44.4
48.8
47.7
52.1
47.4
49.6
55.3
55.7
56.1
55.8
58.4
66.6
55.0
48.3
43.9
56.7
54.5
50.5
57.8
50.4
50.3
42.1
26.6
53.5
47.9 ꢀ 6.9
49.8 ꢀ 10.1
44.3 ꢀ 11.5
48.7 ꢀ 7.4
47.4 ꢀ 5.2
51.9 ꢀ 4.4
47.3 ꢀ 6.0
49.5 ꢀ 4.1
55.2 ꢀ 5.7
55.6 ꢀ 7.9
56.0 ꢀ 10.6
55.8 ꢀ 7.5
58.4 ꢀ 8.5
66.6 ꢀ 8.1
55.0 ꢀ 7.2
48.1 ꢀ 4.8
43.6 ꢀ 4.4
56.6 ꢀ 9.2
54.4 ꢀ 4.5
50.3 ꢀ 9.0
57.7 ꢀ 8.5
50.1 ꢀ 7.9
50.1 ꢀ 5.8
41.6 ꢀ 5.2
25.7 ꢀ 4.4
53.1 ꢀ 7.0
2.9
2.9
—
—
84
60
5.7
2.8
Di-n-butyl ether
Di-n-butyl ether
100
135
124
25
2.8
2.9
—
Di-n-butyl ether
1.2
2.9
Di-n-butyl ether
—
65
15
BE11
BE12
0.16
1.0
—
—
22
2
BE13
BE14
0.16
0.50
2.4
—
—
2
2
2
2
BE15
BE16
BE17
—
1
5
8.4
Di-n-butyl ether
Di-n-butyl ether
—
108
145
1
66
86
4
11.3
2.5
BE16-II
BE17-II
BE18
1.4
2.7
—
—
64
75
34
67
30
96
69
191
445
110
BE19
BE24
0.11
10.6
4.9
—
55
95
Mesitylene
Mesitylene
p-Cresol
p-Cresol
p-Cresol
BE26
BE28
62
44.2
41.2
37.8
600
1360
115
BE28-II
BE31a
FORD
05–14–01b
05–24–27b
04–30–01
02–05–01
04–25–14
05–01–01
05–11–01c
01–30–24c
(6)
(6)
(6)
(6)
(6)
(6)
(6)
(6)
961
965
Propene
Propene
Propene
Propene
Propene
Propene
Propene
Propene
327
327
330
327
301
327
314
350
671
1340
592
796
1533
791
20.5
14.1
31.0
22.8
21.5
20.5
35.0
35.8
17.5 ꢀ 4.4
10.5 ꢀ 1.7
29.3 ꢀ 0.2
20.6 ꢀ 2.2
19.1 ꢀ 1.4
18.6 ꢀ 1.8
34.6 ꢀ 3.3
35.2 ꢀ 3.3
1000
291
1313
1055
1528
737
1421
1548
1116
663
285
1003
971
288
1196
1327
Experiment types: (1) benzene/NOx , (1a) benzene/NOx with small amount of added CH3ONO, (2) ozone-photolysis, (3) HONO photolysis, (4)
H2O2 photolysis, (5) HCHO/NOx system, (6) benzene/CH3ONO/NOx .a 300 Torr O2 in 760 Torr,b 50 Torr O2 in 700 Torr,c 650 Torr O2 in 700
Torr, all other experiments are conducted with the atmospheric O2 mixing ratio of 20% in 700 Torr (FORD) or 760 Torr (NIES, EUPHORE).
4402
Phys. Chem. Chem. Phys., 2002, 4, 4399–4411

View Article Online
the NIES experiments were conducted with an added OH tra-
cer substance (di-n-butyl ether) to determine the concentra-
tion–time profile of OH radicals.
determine the rate coefficient for the reaction of the hydroxy-
cyclohexadienyl peroxyl radical 3 with NO (reaction (12) in
Fig. 1). Under these conditions, reaction (12) will be the domi-
nant loss channel of 2 and 3. If phenol formation from reaction
(10) is negligible, then
All the experiments conducted at FORD were of the ben-
zene/CH3ONO/NOx type. The OH-radical tracer compound
used was propene. Consumption of propene was measured
and used to calculate the consumption of benzene. In all
experiments benzene consumption was limited to <5%. Mix-
tures of benzene, propene, CH3ONO and NO were subjected
to consecutive 1–10 s irradiations. After each irradiation the
concentrations of propene, phenol, NO and NO2 were deter-
mined using FTIR spectroscopy. The experimental conditions
were chosen so that reactions of propene with O3 and phenol
with NO3 did not influence the results. The initial concentra-
tions were; benzene, (3–15) ꢁ 1015; propene, 1.9 ꢁ 1014;
CH3ONO, (1–5) ꢁ 1014, and NO, either 3.2 ꢁ 1013 or
6.5 ꢁ 1013 molecule cmꢂ3. Experiments were conducted at
700 Torr total pressure and 3 oxygen concentrations; 50, 140
and 650 Torr (7, 20 and 93%).
ktotal ¼ kphenol=Yphenol
ð4Þ
The assumption that phenol formation from the reaction of
2 with NO2 (reaction (10)) does not contribute significantly to
the overall phenol formation in these experiments is justified
based on (a) the low phenol formation yield in this reaction
determined in the present study and (b) the use of only those
experiments where NO was added in excess. Reaction (10) will,
however, be a non-negligible loss process of 2/3, and ktotal can
be corrected according to:
k10 ½NO2ꢅ kphenol k10 ½NO2ꢅ
NO2-corrected
k
¼ ktotal
ꢂ
¼
ꢂ
ð5Þ
total
Keq ½O2ꢅ
Yphenol
Keq ½O2ꢅ
Since, considering the definitions of ktotal and kclean
,
NO2-corrected
The experimental data were evaluated by the methods
described previously,7 the new NIES and the FORD data were
evaluated using method (b2) therein.
ktotal
¼ kclean + k12[NO] it follows:
kphenol k10 ½NO2ꢅ
ꢂ
¼ kclean ꢂ k12 ½NOꢅ
ð6Þ
Yphenol
Keq ½O2ꢅ
2.5 NOx dependence of the phenol yield
Thus, a plot of kphenol/Yphenol ꢂ k10[NO2]/(Keq[O2]) as a
function of the NO concentration should yield a straight line
with a slope of k12 and an intercept of kclean . The parameters
On the basis of the mechanistic scheme shown in Fig. 1 Volk-
amer et al.7 derived the following formula to describe the
dependence of the phenol yield on the concentrations of
NOx . The simplifications made in this approach7 also apply
under the high NOx conditions discussed in this work and
make formula (1) a valid starting point for further discussion:
k
10 , Keq and kclean are available in the literature,8,9 kphenol has
been determined in part I of this series7 and the other para-
meters are measured directly.
2.7 Determination of the phenol yield from reaction (10)
k3
k10 ½NO2ꢅ
Keq ½O2ꢅ
þ k4 þ Yð10Þ
phenol
The phenol yield from the reaction of the hydroxycyclohexa-
dienyl radical 2 with NO2 (Yðp1h0eÞnol) can be determined by re-
arranging eqn. (2):
Keq
Yphenol
¼
ð1Þ
k3
k9
k10 ½NO2ꢅ
þ
þ k4 þ k8 þ
þ k12 ½NOꢅ
Keq Keq
Keq ½O2ꢅ
ꢀ
ꢁ
Keq ½O2ꢅ
ð10Þ
Y
¼ Yphenolktotal ꢂ kphenol
ð7Þ
In this equation the numerator represents the rate of forma-
tion of phenol from reactions (3) and (4) involving oxygen
(henceforth kphenol ¼ k3/Keq + k4) and reaction (10) involving
NO2 , Yðp1h0eÞnol denotes the phenol yield from reaction (10), Keq
phenol
k10 ½NO2ꢅ
Only those experiments conducted in the presence of a large
ð10Þ
excess of NO2 were used in the determination of Y
. In
phenol
addition, due to the decay of NO2 during the experiments, only
phenol formation yields measured during the initial 15–30 min
of the experiments were used. The measured phenol yield
between any 2 data points is given by:
is the equilibrium constant Keq ¼ k2/k of the equilibrium
ꢂ2
reactions (2) and (ꢂ2). The denominator represents the total
loss rate for the intermediates 2/3 and will be called ktotal for
simplification (ktotal is equal to the lifetime of 2/3):
ꢂ1
corrected
D½phenolꢅtðnÞꢂtðnꢂ1Þ
k10 ½NO2ꢅ
kphenol þ Yð10Þ
phenol
Yphenol
¼
ð8Þ
Keq ½O2ꢅ
D½benzeneꢅtðnÞꢂtðnꢂ1Þ
Yphenol
¼
ð2Þ
ð3Þ
ktotal
corrected
Here, D[phenol]
is the amount of phenol formed
tðnÞꢂtðnꢂ1Þ
Rearrangement gives:
between times t(n ꢂ 1) and t(n), corrected for losses
through OH-radical reaction and wall deposition, and
D[benzene]t(n)ꢂt(nꢂ1) is the benzene reacted in the same time
interval. The correction of the phenol concentrations was per-
formed as described earlier, see Volkamer et al.7 Combining
eqn. (7) and (8) with the definition of ktotal gives the following
formula for the phenol yield from reaction (10):
k10 ½NO2ꢅ
total
kphenol
Yphenol ꢂ Yð10Þ
¼
phenol
k
Keq ½O2ꢅ
ktotal
Under conditions of low concentrations of NOx , as preva-
lent in the atmosphere, ktotal is equal to kclean ¼ (k3 + k9)/
K
eq + k4 + k8 ¼ (760 ꢀ 80) sꢂ1 8
.
In evaluations of the experiments, the NOx concentrations
ð10Þ
Y
ðtðnÞ ꢂ tðn ꢂ 1ÞÞ
were taken as the averages over the measurement time.
According to eqn. (3) a plot of the phenol yield, corrected
for phenol formed in reaction (10), as a function of ktotal
should yield a straight line through the origin, with a slope
equal to the formation rate of phenol from 2/3, kphenol . The
ratio of kphenol to the total loss rate of 2 and 3 under atmo-
spheric conditions, kclean , gives the phenol yield under atmo-
spheric conditions.
phenol
(
corrected
D½phenolꢅtðnÞꢂtðnꢂ1Þ
ꢂ1
¼
D½benzeneꢅtðnÞꢂtðnꢂ1Þ
!
)
k10 ½NO2ꢅ
Keq ½O2ꢅ
Keq ½O2ꢅ
ꢁ
kclean þ k12 ½NOꢅ þ
ꢂ kphenol
k10 ½NO2ꢅ
ð9Þ
The time intervals t(nꢂ1) to t(n) correspond to the spectra
2.6 Determination of k12
recorded during the experiments. Phenol yields from reaction
(10) are calculated for the respective intervals of an experiment
The experiments conducted under excess NO concentrations
allow the measured phenol formation yields to be used to
ð10Þ
and then averaged to calculate Y
for that experiment. The
phenol
Phys. Chem. Chem. Phys., 2002, 4, 4399–4411
4403

View Article Online
average of the yields of the individual ðe1x0Þperiments is then cal-
culated to give the overall value for Yphenol
determined values. The chemical scheme was that shown in
Fig. 1 and 2, the kinetic parameters for the reactions in Fig. 1
were those given above, the kinetic parameters for the reac-
tions shown in Fig. 2 were taken from Klotz et al.24 Reactions
(5), (6), and (11) in Fig. 1 were ignored, see above and Volka-
mer et al.7
Since reactions (10) and (12) can both lead to the formation
of E,E-2,4-hexadienedial, an iterative procedure had to be
employed. First approximations for the yield of E,E-2,4-hexa-
dienedial 9 from reaction (12) were obtained using the high-
NO experiments (NB17, NB18, NB22–25) and assuming the
E,E-2,4-hexadienedial yield from reaction (10) to be negligible.
The results for the individual experiments were averaged, and
that average was used in simulation runs of 3 high-NO2 experi-
ments (NB40–42) in which the yield of E,E-2,4-hexadienedial
from reaction (10) was fitted to its observed concentration–
time profile. Those yields were then averaged and used to
determine a second approximation to the yield of 9 from reac-
tion (12). The iterations were repeated until self-consistency
was achieved.
.
For 298 K and tropospheric concentrations of O2 , the fol-
lowing kinetic parameters have been reported in the
literature:8,9 Keq ¼ (2.7 ꢀ 0.4) ꢁ 10ꢂ19 cm3 moleculeꢂ1
,
kclean ¼ (k3 + k9)/Keq + k4 + k8 ¼ (760 ꢀ 80)sꢂ1, k10 ¼ (2.75 ꢀ
0.2) ꢁ 10ꢂ11 cm3 moleculeꢂ1 sꢂ1
,
and k12 ¼ (1.1 ꢀ 0.4) ꢁ
10ꢂ11 cm3 moleculeꢂ1 ꢂ1. For k12 , the value determined in
s
the present study has been employed instead of the literature
value, see below.
2.8 The yields of E,E-2,4-hexadienedial from reactions (10)
and (12)
E,E-2,4-hexadienedial (species 9 in Fig. 2) is a highly reactive
unsaturated dicarbonyl, and its steady state concentrations
are consequently very low. It reacts rapidly with OH radicals
and photolyses under irradiation with ambient sunlight.23,24
In the OH initiated oxidation of benzene, E,E-2,4-hexadiene-
dial can be formed via the formation pathways proposed by
Zellner et al.25 and Hoshino et al.26 The degradation scheme
of E,E-2,4-hexadienedial is given in Klotz et al.24 Fig. 2 gives
an overview of the postulated formation pathways of 2,4-hexa-
dienedial 9 in reactions (10) and (12), as well as the subsequent
degradation mechanism. The degradation scheme involves a
reversible photoisomerisation, and photolysis and OH radical
reaction of both E,E-2,4-hexadienedial and its photoisomerisa-
tion product.
Due to the complex behaviour of E,E-2,4-hexadienedial 9 it
is not possible to derive a simple kinetic equation to calculate
its formation yield in the benzene photooxidation system.
Therefore the formation yield of E,E-2,4-hexadienedial has
been fitted to its observed concentration–time profiles using
the chemical simulation software FACSIMILE. In the simula-
tion runs, the OH radical concentration–time profiles were
fitted to the decay of the added OH tracer substance, concen-
trations of NO and NO2 were fitted to the experimentally
3. Results and discussion
An overview of the experiments conducted, their conditions
and results is given in Table 1. The experiment type nomencla-
ture is the same as that used in part I.7 Fig. 3 shows typical
FTIR product spectra of the 3 types of experiments performed
at NIES, and selected reference spectra. The product spectra
are normalised by the benzene consumption, which leads to
different levels of spectral noise. Notably a low benzene con-
sumption in (b) causes a relatively high noise level in the nor-
malised spectrum.
In the low-NOx experiment (a) the product spectrum is
dominated by the absorptions of phenol, a result that is in line
with the experiments shown in part I of this series, see
Volkamer et al.7 In contrast, as seen in panel (b), very little
Fig. 2 Proposed mechanisms for the reaction of the hydroxycyclohexadienyl radical 2 with NO2 and the hydroxycyclohexadienyl peroxyl radical
3 with NO. The formation and removal pathways of E,E-2,4-hexadienedial 9 are also shown.
4404
Phys. Chem. Chem. Phys., 2002, 4, 4399–4411

View Article Online
Fig. 4 Plots of the phenol formed as a function of benzene reacted
for experiments with different average NO concentrations (solid circles:
NIES, open circles: FORD, open diamond: EUPHORE). Among
these experiments NB20 has the lowest average NO concentration fol-
lowed by NBE2, NBE4, NB18 and NB24, see Table 1.
3.1 Determination of k12
As outlined in section 2.6, the phenol yields measured in the
high-NO experiments can be used to measure the rate coeffi-
cient of the reaction of 3 with NO (k12). Fig. 5 shows a plot
according to eqn. (6), the slope of which gives k12 . Only the
experiments with average NO concentrations above 1013 mole-
cule cmꢂ3 have been considered. Experiments in which either
the correction for the NO2 reaction of 2 or the uncertainty
in the phenol yield exceed 35% are not plotted. The solid line
in Fig. 5 represents a non-weighted linear least squares fit
through the data plotted, with the intercept fixed at
kclean ¼ 760 sꢂ1. The slope gives:
Fig. 3 Product spectra recorded during the photolysis of a benzene/
NOx mixture in air (NIES). Product spectra of (a) a low-NOx
([NOx] ¼ 60 ppb), (b) a high-NO ([NO] ¼ 10 000 ppb) and (c) a
high-NO2 ([NO2] ¼ 7500 ppb) experiment, after 75 min irradiation
and subtraction of the absorptions of the reactants and well charac-
terised products; reference spectra of (d) nitrobenzene, (e) E,E-2,4-
hexadienedial and (f) phenol. (g)–(i): Spectra (a)–(c) after subtraction
of all identified absorptions. Note that for easier comparison, the spec-
tra are normalised by the benzene consumption.
k12 ¼ ð1:7 ꢀ 0:6Þ ꢁ 10ꢂ11 cm3 moleculeꢂ1 sꢂ1
:
phenol is formed in the presence of high concentrations of NO,
which is in line with the expectation from eqn. (1). Instead, a
new product dominates the product spectrum; E,E-2,4-hexa-
dienedial 9, compare panel (e). The high-NO2 experiment
shown in panel (c) shows significant absorptions of both phe-
nol and E,E-2,4-hexadienedial, qualitatively confirming the
indirect results of Atkinson et al.16 that reaction (10) does con-
tribute to phenol formation. Panels (g)–(i) show spectra (a)–(c)
after subtraction of the absorption of all identified products.
From the spectra shown in Fig. 3 it is evident that under the
different experimental conditions employed different products
are formed. The dependence of the phenol formation yield
on the NO concentration is illustrated in Fig. 4, which shows
plots of the phenol formed as a function of the benzene
reacted. All phenol concentrations were corrected for losses
through OH radical reaction and wall deposition, the amount
of benzene reacted has been calculated from the degradation of
di-n-butyl ether, which served as the OH tracer substance. Of
the experiments plotted in Fig. 4, NB24 has the highest average
NO concentration. The average NO concentration in the
experiments decreased in the order
cNO(NB24) > cNO-
Fig. 5 Plot according to eqn. (6). The slope gives the rate coefficient
of the reaction of the hydroxycyclohexadienyl peroxyl radical 3 with
NO, see text. The dotted line shows previous results of Bohn and
Zetzsch.8
(NB18) > cNO(NBE4) > cNO(NBE2) > cNO(NB20), see Table 1.
From Fig. 4 it is evident that as the NO concentration
increases, the phenol yield decreases.
Phys. Chem. Chem. Phys., 2002, 4, 4399–4411
4405

View Article Online
Including the low-NOx experiments would have the same
effect as fixing the intercept at kclean , since the value for
kphenol ¼ (404 ꢀ 50) sꢂ1 employed is itself based on
kclean ¼ 760 sꢂ1. The dotted line represents the previous deter-
mination of k12 ¼ (1.1 ꢀ 0.4) ꢁ 10ꢂ11 cm3 moleculeꢂ1 sꢂ1 by
Bohn and Zetzsch.8 The value of k12 determined herein is,
within the admittedly large uncertainties, indistinguishable
from that reported by Bohn and Zetzsch.8 In Fig. 5 the error
limit of k12 reported by Bohn and Zetzsch is shown for one
data point. The uncertainty in the present study is the result of
reported by Berndt et al.13 were re-evaluated to match the cali-
bration of this work, see Volkamer et al.7 The NOx concentra-
tions for Atkinson et al.16 were calculated as the averages of
the data points reported for each experiment, the yields were
calculated by dividing the reported amount of phenol formed,
corrected for losses through OH radical reaction, by the
reported amount of benzene reacted. The phenol yields of
Atkinson et al.16 are also corrected for phenol formation
through reaction (10), using the phenol yield from this reaction
determined in the present study. The data of Berndt and
Bo¨ge17 were also corrected to match the IR absorption cross
sections18 used in the present work.
the combined uncertainties of the parameters kphenol , Yphenol
,
k
k
10 , [NO2], Keq and kclean , which all affect the determination of
12 , see eqn. (6). None of these individual sources of uncer-
For the data of the present study as well as those of Atkinson
et al.16 the error limits for ktotalꢂ1 are calculated using the stan-
dard deviation of the measured NOx concentrations as the
only source of uncertainties. In addition to these non-systema-
tic errors the inaccuracies of the parameters k10 , k12 , kclean and
tainty dominate the final uncertainty associated with k12
,
and consequently a more precise determination of k12 by the
method used here will require more precise measurements of
all or most of the above parameters. In the discussions below,
we choose to adopt the value of k12 measured herein as the
value is more compatible with our experimental results
(see section 3.2). Finally, it is worth noting that k12 is of a
similar magnitude to that of the reaction of other peroxyl
radicals which typically lie in the range (1–2) ꢁ 10ꢂ11 cm3
K
eq add to the overall errors. The errors of the phenol yields of
Atkinson et al.16 were assumed to be the same as those
reported for the average phenol yield, i.e. 4.4%. The errors
of the phenol yields of the present study were calculated as
described earlier.7
moleculeꢂ1 sꢂ1 27
.
The observed strong dependence of the phenol yield on the
concentration of NOx , while supporting the work of Knispel
et al.9 and Bohn and Zetzsch,8 is in contradiction to the find-
ings of Atkinson et al.,16 the only previous study that
employed high concentrations of NOx under comparable
experimental conditions. As evident from Fig. 6, the phenol
yields reported by Atkinson et al. do not show a dependence
on the concentration of NOx . This absence of a NOx depen-
dence was supported by two recent studies13,10 under NOx-free
conditions, in which measured phenol yields were in apparent
agreement with those reported by Atkinson et al.16 However,
as demonstrated in Table 3 of Volkamer et al.7 considerable
scatter becomes apparent among these values if the latter
two studies are placed on the same basis of calibration.
The absence of NOx dependence of the phenol yield in the
data of Atkinson was explained by speculating that a signifi-
cant yield of phenol in the reaction of the hydroxycyclohexa-
dienyl radical 2 with NO2 (reaction (10)) can offset the effect
of NOx reactions to lower the phenol yield. A possible
mechanism for this is shown in Fig. 2. There is no reasonable
mechanism to explain phenol formation from reaction (12),
hydroxycyclohexadienal peroxyl + NO, see Fig. 2. Though
the results of Atkinson et al.16 also qualitatively confirm eqn.
(1) under conditions of high NOx considerable differences are
found with respect to the phenol yield from channel (10)
(see ection 3.4).
3.2 Phenol yield, NOx dependence
As outlined in section 2.5, the phenol yield in the OH initiated
oxidation of benzene should be a linear function of the inverse
total degradation rate of the hydroxycyclohexadienyl and
hydroxycyclohexadienyl peroxyl radicals 2 and 3, ktotal , see
eqn. (3). Fig. 6 shows a plot according to eqn. (3), with the
phenol yields being corrected for phenol formed through chaꢂn1-
nel (10) in Fig. 1, as determined below. Low values for ktotal
represent highꢂ1concentrations of NOx , while the maximum
value of ktotal ¼ 1.32 ꢁ 10ꢂ3 s represents NOx-free condi-
tions. The phenol yield of Bjergbakke et al.10 and the yield
It may be speculated that these different results are caused
by the fact that Atkinson et al.16 used an off-line analytical
technique. Atkinson et al.16 used Tenax tubes to adsorb phenol
and other compounds, followed by thermal desorption into a
GC/FID instrument. The thermal desorption process may
have facilitated chemical reactions on the Tenax. For example,
intermediates like the nitro hydroxycyclohexadiene 7 or the
hydroxycyclohexadiene nitrate 12 (Fig. 2) are probably ther-
mally unstable and may readily eliminate nitrous and nitric
acid, respectively, to form phenol. The nitrate 12 may be
formed by reaction of the oxyl radical 7 with NO2 in both
the high-NO and the high-NO2 systems, while 7 can be formed
by addition of NO2 to the hydroxycyclohexadienyl radical 2,
see Fig. 2. These compounds may also photolyse or decompose
thermally in the gas-phase.
The hypothesis of thermal/photolytic decomposition of the
intermediate 7 may also explain the fact that Atkinson
observed the formation of nitrobenzene, which was not
observed in the present study, even though it is known to be
relatively unreactive.5 Nitrobenzene can be formed by elimina-
tion of H2O from 7. Atkinson et al.16 reported yields of nitro-
benzene of up to 10% under conditions of very high NOx
Fig. 6 Plot of the phenol yield, corrected for phenol formed through
reaction (10) (see Fig. 1), as a function of the inverse total degradation
rate of 2/3, see eqn. (3). Data were taken from: NIES experiments,
solid circles; EUPHORE experiments, solid diamonds; FORD experi-
ments, hollow circles; Atkinson et al.,16 hollow diamonds; re-evaluated
value7 of Berndt et al.,13 hollow square; re-evaluated value7 of Bjerg-
bakke et al.,10 crossed square; Berndt and Bo¨ge17 (100 mbar, 295 K),
hollow triangle with tip facing down; Berndt and Bo¨ge17 (500 mbar,
293 K), hollow triangle with tip facing up. The line represents a non-
weighted linear least squares fit to the data obtained in the present
study (NIES, EUPHORE, FORD).
4406
Phys. Chem. Chem. Phys., 2002, 4, 4399–4411

View Article Online
concentrations (10 ppm NO2), and extrapolated a nitroben-
zene yield of (3.4 ꢀ 0.8)% for NOx-free conditions. Atkinson
et al. noted that this result was not compatible with accepted
mechanisms for benzene photooxidation and suggested several
alternative schemes, one of them being further reaction of a
precursor species. If this precursor species reacts predomi-
nately on the sampling material, this hypothesis can explain
why the present study, which employs in situ analytical techni-
ques, did not observe nitrobenzene. However, it cannot explain
why Berndt and Bo¨ge17 also observed formation of nitro-
benzene, even though they used in situ FTIR detection, and
NOx/O2 ratios that were in the same range as in the present
study. The fact that at the spectral resolution of 8 cmꢂ1
employed in Berndt and Bo¨ge17 many of the nitrate absorp-
tions observed in the present study may become indistinguish-
able from nitrobenzene absorptions may play a role here. In
contrast to the result of Berndt et al.13 the phenol yields of
Berndt and Bo¨ge17 are in broad agreement with the present
study. This is somewhat surprising because they were mea-
sured with the same experimental set-up. However, it should
be noted that at high NOx/O2 ratios the 100 mbar data show
phenol yields up to 2 times higher than observed in the present
study, while the phenol yield at the lowest NOx/O2 ratio is
below our values. This becomes clearer in Fig. 4 of Berndt
and Bo¨ge,17 where a phenol yield of around 20% is extra-
polated for low concentrations of NOx . This is consistent with
the previous NOx-free study from the same laboratory,13 but
inconsistent with our results. However, the better agreement
observed among their experiments conducted in the presence
of NOx and this work is in line with speculations by Volkamer
et al.7 about high radical concentrations to explain remaining
differences in the absence of NOx and 100 mbar total pressure.
In the presence of NOx the concentration of peroxyl radicals
would be expected to be considerably lower. It is also note-
worthy that the 500 mbar results of Berndt and Bo¨ge, which
were measured at lower NOx/O2 ratios, show a significantly
higher phenol yield than the NOx-free data13 and the 100 mbar
data17 from the same laboratory. This may be an indication of
pressure dependences of reactions in the system which deserve
future study.
a higher O2 concentration is expected to result in a higher yield
of phenol, even when the NOx concentration remains constant.
To test this prediction of eqn. (1) a series of experiments was
conducted using the smaller chamber at FORD, where the
O2 concentration could be easily varied. In the EUPHORE
BE31 experiment the oxygen concentration was raised to
40%, twice the value in normal air. At FORD, oxygen concen-
trations of 7%, 20% and 93% were employed at a total pressure
of 700 Torr, see Experimental section.
As evident from the data given in Table 1, changes in the
oxygen concentration were found to have a strong effect on
the phenol yield in the high-NOx experiments conducted at
FORD, but not in the low-NOx EUPHORE experiment
BE31. In the presence of high concentrations of NOx ,
decreased oxygen concentration leads to decreased phenol
yield while increased oxygen concentration leads to increased
phenol yield, in line with the prediction from eqn. (1). Also,
the effect is observed to be stronger at higher concentrations
of NOx . In Fig. 6 the low and high O2 FORD-data are plotted
together with the other data obtained in this study and no
deviation from the expected quantitative behaviour is
observed. The results of these experiments serve as an addi-
tional quantitative confirmation of the validity of eqn. (1).
The O2 dependence of the phenol yield at high concentra-
tions of NOx indicates that pathway (5) in Fig. 1, formation
of phenol by elimination of an H-atom from the hydroxycyclo-
hexadienyl radical 2, does not play a significant role in the OH
initiated oxidation of benzene. This finding is in line with the
results of Bohn and Zetzsch,8 but contradicts the report by
Bjergbakke et al.10
3.4 Phenol yield from reaction (10)
When comparing the product spectra of the high-NO and
high-NO2 experiments (Panels b and c in Fig. 3) it becomes
clear that in the high-NO2 experiments substantial amounts
of phenol are formed even though channels (3)–(6) in Fig. 1
are only of secondary importance, see Table 2. This shows that
phenol is a product of the reaction of the hydroxycyclohexa-
dienyl radical 2 with NO2 . A quantitative calculation using
the methodology described in section 2.7 gives an average phe-
ð10Þ
The solid line in Fig. 6 represents the results of a linear non-
weighted least squares fit through the data of this study (NIES,
EUPHORE, FORD), giving:
nol yield for this reaction of Y
phenol
nol yields calculated for each individual experiment are given
in Table 2.
¼ (5.9 ꢀ 3.4)%. The phe-
corr
ꢂ1
Yphenol
¼ ð421 ꢀ 24Þ sꢂ1 ktotal ꢂ ð0:010 ꢀ 0:023Þ
The reaction of the hydroxycyclohexadienyl radical 2 with
NO2 contributes to phenol formation, but not as much as
required for the high-NOx data of Atkinson et al.16 to be com-
patible with our results. Using the data of Atkinson et al.16 the
The errors given above are 2 standard deviations of the
regression analysis and represent statistical precision only.
Including the statistical error above and the average uncertain-
ties in the phenol yields determined in this study, we arrive at
kphenol ¼ (421 ꢀ 39) sꢂ1. The intercept is not statistically signif-
icant, in line with the expectation from eqn. (3). The phenol
phenol yield from channel (10) is calculated to be
Y
ð10Þ
¼ (18 ꢀ 6)%,28 this is not compatible with the results
phenol
from this work. As discussed in section 3.2, we believe that
there were confounding sources of phenol in the experiments
of Atkinson et al.16 Our hypothesis about thermal decomposi-
tion of nitrohydroxycyclohexadiene 7 and/or hydroxycyclo-
hexadiene nitrate 12 contributing to the phenol formation in
their work is supported by an experiment conducted under
conditions of extremely high concentrations of NO2 and NO
(25 ppm each), performed at FORD. Under these conditions
the phenol yield became significantly higher at longer reaction
times (higher benzene decompositions) rising from 4.3% initi-
ally to about 20%. This value is essentially the value28 esti-
mated from the data of Atkinson et al.16 but cannot be
explained by the chemical mechanisms shown in Fig. 1 and 2.
The NO2 concentration fell somewhat during the experiment,
while the concentration of NOx rose due to the supply of
NO by photolysis of CH3ONO. Consequently phenol yields
would be expected to become lower with the falling concentra-
tion of NO2 and rising concentration of NO, unless an
additional mechanism for phenol formation is operative under
these extreme conditions.
yield under clean (NOx-free) conditions is given by kphenol
k
/
clean . Combining the value of kphenol ¼ (421 ꢀ 39) sꢂ1 derived
above with kclean ¼ (760 ꢀ 80) sꢂ18 gives Yphenol(NOx-
free) ¼ (55.4 ꢀ 7.8)%. This value is consistent with the phenol
yield of (53.1 ꢀ 6.6)% calculated from the low-NOx experi-
ments reported in part I of this series.7 Possible explanations
for why the NOx-free data of Bjergbakke et al.10 and Berndt
et al.13 deviate significantly from our results have been given
by Volkamer et al.,7 and are not repeated here.
3.3 Phenol yield, O2 dependence
The yield of phenol in the OH initiated oxidation of benzene
was found to be independent of the concentration of O2 in a
single experiment (BE31) conducted at EUPHORE at low con-
centrations of NOx .7 This behaviour is in agreement with eqn.
(1) under conditions where reactions (10) and (12) are essen-
tially unimportant. However, at high concentrations of NO2
Phys. Chem. Chem. Phys., 2002, 4, 4399–4411
4407

View Article Online
Table 2 Summary of the results of the experiments conducted with large excesses of either NO or NO2 . Columns 3 and 4, respectively, give the
phenol and E,E-2,4-hexadienedial yields from reaction (10) in Fig. 1. Column 5 gives the yield of E,E-2,4-hexadienedial from reaction (12)
Branching ratio (%) for reaction of 2/3 with
Experiment
code
Experiment
type
ð10Þ
ð10Þ
ð12Þ
Y
(%)
Y
(%)
Y
(%)
O2
NO
NO2
phenol
E;E-2;4-hexadienedial
E;E-2;4-hexadienedial
NB17
NB18
NB22
NB23
NB24
NB25
NB30
NB31
NB32
NB33
NB34
NB40
NB41
NB42
Averages
High-NO
High-NO
High-NO
High-NO
High-NO
High-NO
High-NO2
High-NO2
High-NO2
High-NO2
High-NO2
High-NO2
High-NO2
High-NO2
—
—
—
—
—
—
—
—
27.3 ꢀ 14.0
34
24
29
22
14
16
24
38
25
41
35
27
22
18
—
30
49
43
52
67
66
18
5
37
27
29
26
19
19
58
57
58
51
56
60
60
59
—
31.2 ꢀ 15.9
—
—
29.3 ꢀ 15.4
30.5 ꢀ 15.6
—
—
26.4 ꢀ 13.2
31.3 ꢀ 15.7
4.2 ꢀ 2.7
4.1 ꢀ 2.5
6.5 ꢀ 3.5
6.2 ꢀ 3.8
6.6 ꢀ 3.8
6.1 ꢀ 3.3
6.8 ꢀ 3.7
6.5 ꢀ 3.5
5.9 ꢀ 3.4
—
—
—
—
—
—
—
—
17
9
—
—
—
9
2.9 ꢀ 1.7
4.0 ꢀ 2.3
3.3 ꢀ 1.8
3.4 ꢀ 1.9
13
19
23
—
—
—
29 ꢀ 16
3.5 Yields of E,E-2,4-hexadienedial from reactions (10)
and (12)
high-NO series. Experiment NB17 is the one with the lowest
initial NO concentration in that series, and consequently the
reaction of NO with 3 only accounts for ca. 30% of the decay
of 2/3. This represents the experiment with the lowest NO con-
centration that could be reliably evaluated for E,E-2,4-hexa-
dienedial. Experiment NB25, together with NB24, represents
the highest NO concentrations employed in this study, about
66% of 2/3 is lost through reaction of 3 with NO. Fig. 7 shows
that the chemical mechanism shown in Fig. 2 gives reasonably
good fits to the experimental data. The observed deviations,
notably the faster rise of the E,E-2,4-hexadienedial concentra-
tion in the fits, are probably due to the fact that the kinetic
parameters employed were obtained under different light condi-
tions that were scaled to the J(NO2) value in the NIES reaction
chamber. These uncertainties are estimated at 50% and domi-
nate the errors reported in Table 2.
The formation of E,E-2,4-hexadienedial in the high-NO2
experiments was larger than expected from the reaction of 3
with NO alone, indicating that the reaction of 2 with NO2 also
gives E,E-2,4-hexadienedial. A proposed pathway to E,E-2,4-
hexadienedial formation from this reaction is shown in Fig. 2,
Table 2 includes calculated yields for E,E-2,4-hexadienedial
from the reaction of the hydroxycyclohexadienyl radical 2
As evident from Fig. 3(b), the main product of the OH
initiated oxidation of benzene identified under conditions of
extremely high NO concentrations is E,E-2,4-hexadienedial 9,
which is presumably formed through reaction of the hydroxy-
cyclohexadienyl peroxyl radical 3 with NO (reaction (12)), see
Fig. 2. The yield of E,E-2,4-hexadienedial following this reac-
tion was calculated using the methodology described in section
2.8, individual results are shown in Table 2, together with the
branching ratios between the loss processes for 2/3 with O2 ,
NO and NO2 . An average value of YðE1;2EÞ-2;4-hexadienedial
(29 ꢀ 16)% was determined for the yield of E,E-2,4-hexadiene-
dial following reaction (12). Two examples of experimental and
fitted concentration–time profiles of E,E-2,4-hexadienedial are
given in Fig. 7, showing the first and last experiments of the
¼
with NO2 , see experimental section. An average value of
Y
ð10Þ
¼ (3.4 ꢀ 1.9)% is calculated. E,E-2,4-hexa-
E;E-2;4-hexadienedial
dienedial is therefore a minor product formed from the reac-
tion of 2 with NO2 (reaction (10) in Fig. 1).
A simple mechanism for the formation of E,E-2,4-hexadie-
nedial 9 in reaction (10) is shown in the left column of Fig. 2.
The reaction of 2 with NO2 can proceed via oxygen transfer to
the carbon ring, giving a hydroxycyclohexadienyl oxyl radical
8 and NO, as proposed by Zellner et al.25 The hydroxy-
cyclohexadienyl oxyl radical 8 can then give Z,Z-2,4-hexadie-
nedial 9a by ring cleavage and abstraction of an
H-atom by molecular oxygen. Z,Z-2,4-hexadienedial 9a will
rapidly isomerise into its more stable E,E-form 9.23,24 Alterna-
tively, the delocalised pentadienyl-type radical formed after
ring-opening of 8 can isomerise before reaction with O2 ,
resulting in the direct formation of E,E-2,4-hexadienedial 9.
The most likely formation mechanism for E,E-2,4-hexadiene-
dial from reaction (12), based on the work of Hoshino
et al.,26 is given in the right column of Fig. 2. Reaction of
NO with 3 gives a hydroxycyclohexadienyl oxyl radical 8,
which can then give E,E-2,4-hexadienedial as described above.
The yields determined for the individual experiments shown
in Table 2 do not show any systematic variation with the
Fig. 7 Experimental (symbols) and fitted (lines) concentration–time
profiles for E,E-2,4-hexadienedial in benzene/CH3ONO/NOx experi-
ments in the presence of high NO concentrations (NIES). The first
(NB17, lowest NO) and last (NB25, highest NO) of the series of
high-NO experiments are shown.
4408
Phys. Chem. Chem. Phys., 2002, 4, 4399–4411

View Article Online
concentration of NOx , even though in the high-NO experi-
ments the relative importance of the reaction of 2/3 with O2
and NO changes by about a factor of 2. This indicates that
within the experimental errors E,E-2,4-hexadienedial is not
formed in the absence of NOx . While this is expected from
the reaction mechanism shown in Fig. 1, it appears to be in
contradiction to the results of Berndt et al.13 and Berndt and
Bo¨ge.17 The FTIR spectra shown by Berndt et al.,13 measured
under NOx-free conditions, seem to show significant absorp-
tions of E,E-2,4-hexadienedial. In part I of this series7 it was
speculated that the very high radical concentrations in that sys-
tem may have led to a different chemistry of the intermediates
2/3, leading to the formation of E,E-2,4-hexadienedial
through pathways not occurring under atmospheric condi-
tions. In the more recent NOx dependent study of Berndt
and Bo¨ge,17 a maximum E,E-2,4-hexadienedial yield of about
36% of the benzene reacted was found at very high concentra-
tions of NOx , using an estimated absorption cross section.
When this yield is re-calibrated using the FTIR absorption
cross section of Etzkorn et al.18 a value of 17% is obtained.
When it is further considered that a maximum of about 70%
of 2/3 reacts with NO under the experimental conditions of
Berndt and Bo¨ge,17 a yield of 24% is obtained for the forma-
tion of E,E-2,4-hexadienedial from reaction (12). This com-
pares well with the value of (29 ꢀ 16)% found here. The E,E-
2,4-hexadienedial yields of Berndt and Bo¨ge drop significantly
at low NOx concentrations (see their Fig. 4), though there
appears to be a statistically significant E,E-2,4-hexadienedial
formation of about 10% (re-calibrated 7%) under NOx-free
conditions. This is consistent with Berndt et al.,13 but not with
the results of the present study.
The error limits of the E,E-2,4-hexadienedial yields reported
in Table 2 are a combination of an estimated uncertainty of
50% resulting from the degradation kinetics of this compound,
and the 2-sigma errors from the fitting procedure described in
section 2.8. The atmospheric chemistry of E,E-2,4-hexadiene-
dial was not determined in our laboratory but was adapted
from the literature,23,24 the error is therefore largely systematic
in nature. However, the good agreement with the value deter-
mined by Berndt and Bo¨ge,17 whose experimental set-up is far
less susceptible to secondary chemistry than that reported
herein, indicates that our estimated uncertainty of 50% is very
conservative. The statement on the absence of E,E-2,4-hexa-
dienedial formation under NOx-free conditions is more pre-
cise. Our data regarding the absence of hexadienedials at low
concentrations of NOx is in agreement with two recent stu-
dies7,14 which observed no indication for the formation of
hexadienedials in the presence of low concentrations of NOx
(several tens ppb). Under these conditions an upper limit for
the yield of hexadienedials from benzene of 8% was derived.14
spectral noise. It is noteworthy that the residual spectra in the
two types of high-NOx experiments are similar, both in terms
of the absorptions and their relative intensities, while the resi-
dual spectrum of the low-NOx experiment in panel (g) is signi-
ficantly less intense and shows different absorptions. This
reflects the fact that 97% of the loss of 2 and 3 in this experi-
ment is through reaction with O2 , and only 3% through reac-
tions with NO and NO2 . The similarity between the residual
spectra of the 2 types of high-NOx experiments shows that per-
oxynitrates, which cannot be formed at high concentrations of
NO, are not formed in significant yields. Some of the other
bands observed in panels (h) and (i) of Fig. 3 are indicative
of nitrates, which typically exhibit absorptions in the regions
around 1650 cmꢂ1, 1300 cmꢂ1, 850 cmꢂ1 and 1000 cmꢂ1 (some
only). The residual spectra of the high-NOx experiments show
bands around those regions, the nitrate band expected around
1650 cmꢂ1 may be obscured by saturated absorptions of NO2 .
The formation of nitrates in the reaction of peroxyl radicals
with NO has been known for some time.29 It is also established
that the nitrate yield increases with size of the peroxyl radical.
Nitrate yields of approximately 20% are observed in the reac-
tion of hexylperoxyl radicals with NO.27,30 It is reasonable to
expect a substantial nitrate yield in reaction (12).
Other absorptions are visible in the carbonyl region. The
band visible at 1830 cmꢂ1 is typical of electron rich carbonyl
groups, e.g. in R–CO–O–R. Additional strong bands at
1680, 1707 and 1750 cmꢂ1 indicate the presence of further car-
bonyls and unsaturated carbonyls. In contrast to the region
below 1600 cmꢂ1 the carbonyl regions exhibit similar absorp-
tions in the 3 systems studied. There are a few similarities
between the residual product spectra recorded here under
high-NOx conditions and the residual product spectra of the
OH initiated oxidation of hexadienedial isomers reported in
the literature.23 It seems likely that some of the residual
absorptions shown in panels (h) and (i) of Fig. 3 are attribut-
able to products of the further reaction of E,E-2,4-hexadiene-
dial.
One product that has been reported previously as arising
from the OH initiated oxidation of benzene in the presence
of NOx is nitrobenzene 13, which can be formed as shown in
Fig. 2. By comparison of panel (d) with panels (a)–(c) in
Fig. 3 it is evident that nitrobenzene, which is essentially
unreactive, is not formed above its detection limit. It is con-
cluded that, even at extremely high NOx concentrations, nitro-
benzene is formed in a yield below the value of (3.4 ꢀ 0.8)%
extrapolated for NOx-free conditions by Atkinson et al.,16
who determined yields of up to 10% under conditions of high
concentrations of NOx . Possible reasons for this and other dis-
crepancies are discussed in section 3.2. The absence of nitro-
benzene in the present study is in contradiction to the recent
results of Berndt and Bo¨ge,17 who found a yield of 11% at high
concentrations of NOx . This may again be an indication that
additional mechanisms are operative under the conditions of
that study. The results of Berndt and Bo¨ge are generally com-
patible with those of Atkinson, with the exception of nitroben-
zene formation at low concentrations of NOx , which Berndt
and Bo¨ge do not observe.
In summary, less than one third of the products of the reac-
tion of the hydroxycyclohexadienyl peroxyl radical 3 with NO
and less than 10% of the products of the reaction of the hydro-
xycyclohexadienyl radical 2 with NO2 could be identified in the
present study. Fig. 2 contains other possible pathways for
these reactions, however, phenol and E,E-2,4-hexadienedial
were the only products that could be identified. 1,2-Dihydroxy-
benzene 11, which may result from the reaction of 3 with NO,
is highly reactive31 and may be formed in considerable yield
without being detectable under the present experimental condi-
tions. On the basis of the residual product spectra shown in
Fig. 3, we speculate that nitrates and possibly other nitrogen
containing compounds are formed in high yields.
3.6 Other products of reactions (10) and (12)
The only product identified in the reaction of the hydroxy-
cyclohexadienyl peroxyl radical 3 with NO (reaction (12))
was E,E-2,4-hexadienedial 9 with a yield of (29 ꢀ 16)%, no
other products were observed which could be unequivocally
attributed to this reaction. In the case of the reaction of the
hydroxycyclohexadienyl radical 2 with NO2 (reaction (10)),
only two products were observed; phenol 4 and E,E-2,4-hexa-
dienedial 9 in yields of a few percent each. Thus the major
fraction of the products of both reactions could not be identi-
fied, and only minor products of reaction (10) were found.
Panels (g), (h), and (i) of Fig. 3 show residual product spec-
tra obtained in the 3 types of experiments performed at NIES.
These residual spectra were obtained by subtracting the
absorptions of all reactants and identified products from the
product spectra shown in panels (a), (b) and (c). The product
spectra were normalised by the benzene consumption to allow
an easier comparison, which again results in different levels of
Phys. Chem. Chem. Phys., 2002, 4, 4399–4411
4409

View Article Online
which is in contradiction to previous studies. The simple for-
mula (1) introduced in part I of this series7 is confirmed under
conditions of high concentrations of NOx (up to >10 ppm).
A study of the dependence of the phenol yield on the concen-
tration of O2 further confirmed the results and supports previous
conclusions8 that channel (5) in Fig. 1 is not a significant loss
process for the hydroxycyclohexadienyl radical, 2. With this
work a complete and consistent set of rate constant and yield
data is available to calculate the phenol yield in the OH radical
initiated oxidation of benzene in the presence of NOx .
The changes in the chemical mechanism of the OH initiated
oxidation of benzene at high concentrations of NOx were con-
firmed by the products observed under these conditions, which
are distinctly different from those observed at low NOx con-
centrations. Increased importance of the reactions of the
hydroxycyclohexadienyl radical 2 with NO2 and the hydroxy-
cyclohexadienyl peroxyl radical 3 with NO are concluded to be
responsible for these differences and dedicated product studies
of these reactions were carried out for the first time. E,E-2,4-
Hexadienedial was observed as a significant product of the lat-
ter reaction, small amounts of phenol and E,E-2,4-hexadiene-
dial are formed in the former. The kinetic parameters for the
initial steps of the OH initiated oxidation of benzene (see
Fig. 1) determined by Knispel et al.9 and Bohn and Zetzsch8
are consistent with the product yields determined in this study.
The rate constant for reaction of the hydroxycyclohexadienyl
peroxyl radical 3 with NO measured herein is higher than
the value by Bohn and Zetzsch,8 their value is at the lower limit
of our error margins. The uncertainty in the rate-constant of
this reaction limits the conclusions that can be drawn from
experiments conducted at high concentrations of NOx .
Fig. 8 Plot according to eqn. (3) showing linear fits through the
experimental data when the upper and lower limits of k10 , k12 , Keq
and kclean are used instead of the recommended values. The inset shows
an expanded view of the high-NOx region (low values of ktotalꢂ1). The
calculations of ktotalꢂ1 were performed with: upper and lower limits of
k12 , dash-dot-dash lines; upper and lower limits of k10 , short dashed
lines; upper and lower limits of kclean , long dashed lines; upper and
lower limits of Keq , dotted lines; maximum (lower limits of k12 , k10
and kclean and upper limit of Keq) and minimum (upper limits of k12
,
k10 and kclean and lower limit of Keq) values of ktotal , solid lines; recom-
mended values for k12 , k10 , kclean and Keq , thick line (inset only).
The observed strong NOx-dependence of the degradation
chemistry is in contrast to the chemistry of alkanes and most
alkenes, where the formation of peroxyl radicals is essentially
irreversible. This highlights the special role aromatic hydro-
carbon mechanisms hold in atmospheric chemistry.
3.7 Error considerations
The influence of the uncertainties in the kinetic parameters k10 ,
k12 , kclean and Keq on the data evaluation has been examined.
The results also have important consequences for atmo-
spheric modelling studies. Most chemical models used for the
prediction of tropospheric photooxidant formation are verified
by comparing their predictions to experimental data from
smog-chamber studies conducted at high levels of NOx . Cur-
rently, levels of NOx in such studies are typically in the range
100–1000 ppb.32–34 The results of the present study show that
benzene exhibits a distinctly different degradation chemistry in
the presence of 1–10 ppm NOx than in the presence of more
atmospherically relevant concentrations of 1–100 ppb NOx .
This finding is qualitatively transferable to other aromatic
hydrocarbons.28,35 The effect of high concentrations of NOx
needs to be taken into account in the verification of chemical
models for tropospheric chemistry. Thus, the high-NOx chem-
istry needs to be included in models that are to be validated
against existing smog-chamber studies. The new data on pro-
duct yields and kinetics presented here should allow some pro-
gress in this field. Also, future experimental studies aimed at
establishing/verifying chemical mechanisms for aromatic
hydrocarbons should be performed using NOx levels which
are representative of those found in the atmosphere.
Fig. 8 shows plots of linear fits through the experimental data
according to eqn. (3) when the upper and lower limits,
respectively, of k10 ¼ (2.75 ꢀ 0.2) ꢁ 10ꢂ11 cm3 moleculeꢂ1 sꢂ1
,
¼
k12 ¼ (1.7 ꢀ 0.6) ꢁ 10ꢂ11 cm3 moleculeꢂ1 sꢂ1
,
Keq
(2.7 ꢀ 0.4) ꢁ 10ꢂ19 cm3 moleculeꢂ1 and kclean ¼ (760 ꢀ 80) sꢂ1
are used instead of the recommended values. The values and
their errors are taken from Knispel et al.,9 Bohn and Zetzsch8
and this work. The inset shows an expanded view of the
high-NOx region at low values of ktotalꢂ1. The individual mea-
sured data points are not shown. As expected, kclean is the
dominant source of error in the low-NOx region, while the
errors associated with k10 , k12 and Keq are more important
in the high-NOx region.
At high-NOx the uncertainties in Keq and especially k12
widely determine the overall uncertainty. Despite the fact that
the value of k10 is about twice as high as the value of k12 the
uncertainty in k10 only adds a minor error. This is mainly
due to the about five times lower relative uncertainty in k10
.
For Keq , its larger uncertainty increases its significance over
that of k10 . The black line in Fig. 8 represents upper and lower
limits for ktotalꢂ1, using either the lower limits of k10 , k12
kclean and the upper limit of Keq or the upper limits of k10
,
,
k
12 , kclean and the lower limit of Keq , respectively. The thick
black line shows a linear fit to the experimental data using
the above values for these constants, to avoid cluttering the
graph it is shown in the inset only. It is evident from Fig. 8 that
the major contribution to the overall error under conditions of
Acknowledgements
The authors are grateful for financial support of this work by
the ‘‘Bundesministerium fur Bildung, Wissenschaft, For-
¨
schung und Technologie’’ (BMBF, grant 07TFS30) and the
European Commission (project EXACT, EVK2-1999-00332).
B. K. gratefully acknowledges successive fellowships of the
Japan Science and Technology Agency (STA)/Alexander von
Humboldt Foundation and the Environment Ministry of
Japan. R. V. acknowledges a Marie Curie fellowship from
high-NOx is due to uncertainty in k12
.
4. Conclusions and atmospheric implications
A strong dependence of the phenol yield on the concentrations
of NOx was observed in the OH initiated oxidation of benzene,
4410
Phys. Chem. Chem. Phys., 2002, 4, 4399–4411

View Article Online
16 R. Atkinson, S. M. Aschmann, J. Arey and W. P. L. Carter, Int. J.
Chem. Kinet., 1989, 21, 801.
17 T. Berndt and O. Bo¨ge, Phys. Chem. Chem. Phys., 2001, 3,
4946.
18 T. Etzkorn, B. Klotz, S. Sørensen, I. V. Patroescu, I. Barnes,
K. H. Becker and U. Platt, Atmos. Environ., 1999, 33, 525.
19 B. Klotz, S. Sørensen, I. Barnes, K. H. Becker, T. Etzkorn,
the European Commission, DG-XII. O. J. N. acknowledges
financial support from the Danish Natural Science Research
¨
Council. Further, the support of Jens Ucker and Milagros
´
Rodenas in the EUPHORE experiments is acknowledged.
The CEAM Foundation is supported by the Generalitat
´
Valenciana and the Fundacio BANCAIXA.
´
R. Volkamer, U. Platt, K. Wirtz and M. Martın-Reviejo, J. Phys.
Chem. A, 1998, 102, 10 289.
20 H. Akimoto, M. Hoshino, G. Inoue, F. Sakamaki, N. Washida
and M. Okuda, Environ. Sci. Technol., 1979, 13, 471.
21 T. J. Wallington and S. M. Japar, J. Atmos. Chem., 1989, 9, 399.
22 W. P. L. Carter, A. M. Winer and J. N. Pitts, Jr., Environ. Sci.
Technol., 1981, 15, 829.
References
1
J. G. Calvert, R. Atkinson, K. H. Becker, R. M. Kamens, J. H.
Seinfeld, T. J. Wallington and G. Yarwood, Mechanisms of Atmo-
spheric Oxidation of Aromatic Hydrocarbons, Oxford University
Press, Oxford, 2002.
23 B. Klotz, A. Bierbach, I. Barnes and K. H. Becker, Environ. Sci.
Technol., 1995, 29, 2322.
2
3
4
J. R. Odum, T. P. W. Jungkamp, R. J. Griffin, R. C. Flagan and
J. H. Seinfeld, Science, 1997, 276, 96.
L. Wang, J. B. Milford and W. P. L. Carter, Atmos. Environ.,
2000, 34, 4337.
L. Wang, J. B. Milford and W. P. L. Carter, Atmos. Environ.,
2000, 34, 4349.
24 B. Klotz, I. Barnes and K. H. Becker, Int. J. Chem. Kinet., 1999,
31, 689.
25 R. Zellner, B. Fritz and M. Preidel, Chem. Phys. Lett., 1985, 121,
412.
26 M. Hoshino, H. Akimoto and M. Okuda, Bull. Chem. Soc. Jpn.,
1978, 51, 718.
5
6
R. Atkinson, J. Phys. Chem. Ref. Data, 1994, Monograph No. 2, 1.
S. D. Piccot, J. J. Watson and J. W. Jones, J. Geophys. Res., 1992,
97, 9897.
27 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.
7
R. Volkamer, B. Klotz, I. Barnes, T. Imamura, K. Wirtz,
N. Washida, K. H. Becker and U. Platt, Phys. Chem. Chem.
Phys., 2002, 4, 1598.
B. Bohn and C. Zetzsch, Phys. Chem. Chem. Phys., 1999, 1, 5097.
R. Knispel, R. Koch, M. Siese and C. Zetzsch, Ber. Bunsen-Ges.
Phys. Chem., 1990, 94, 1375.
28 R. Volkamer, PhD thesis, University of Heidelberg, http://
www.ub.uni-heidelberg.de/archiv/1939, 2001.
29 R. Atkinson, J. Phys. Chem. Ref. Data, 1997, 26, 215.
30 J. Arey, S. M. Aschmann, E. S. C. Kwok and R. Atkinson,
J. Phys. Chem. A, 2001, 105, 1020.
31 R. Olariu, I. Barnes, K. H. Becker and B. Klotz, Int. J. Chem.
Kinet., 2000, 32, 696.
32 W. R. Stockwell, F. Kirchner, M. Kuhn and S. Seefeld, J. Geo-
phys. Res., 1997, 102, 25 847.
8
9
10 E. Bjergbakke, A. Sillesen and P. Pagsberg, J. Phys. Chem., 1996,
100, 5729.
11 R. Koch, J. Phys. Chem. B, 1997, 101, 293.
12 P. Pagsberg, J. Phys. Chem. B, 1997, 101, 294.
13 T. Berndt, O. Bo¨ge and H. Herrmann, Chem. Phys. Lett., 1999,
314, 435.
33 M. C. Dodge, Atmos. Environ., 2000, 34, 2103.
34 M. Jang and R. M. Kamens, Environ. Sci. Technol., 2001, 35,
3626.
14 R. Volkamer, U. Platt and K. Wirtz, J. Phys. Chem. A, 2001, 105,
7865.
15 F. Berho and R. Lesclaux, Phys. Chem. Chem. Phys., 2001, 3, 970.
35 R. Volkamer, J. Uecker, K. Wirtz and U. Platt, Proc. EURO-
TRAC II Symposium, 11th–15th March 2002, Garmisch-Parten-
kirchen, Germany.
Phys. Chem. Chem. Phys., 2002, 4, 4399–4411
4411