Kinetics and branching ratio studies of the reaction of C2H 5O2 + HO2 using chemical ionisation mass spectrometry

Article abstract of DOI:10.1039/b703038j

The overall rate coefficient for the reaction of C₂H ₅O₂ with HO₂ was determined using a turbulent flow chemical ionization mass spectrometer (TF-CIMS) system over the pressure range of 75 to 200 Torr and temperatures between 195 and 298 K. The temperature dependence of the overall rate coefficient for the reaction between C ₂H₅O₂ and HO₂ was fitted using the following Arrhenius expression: k(T) = (2.08_-0.62^+0.87) × 10^-13 exp [(864 ± 79)/T] cm^-3 molecule ^-1 s^-1. The upper limits for the branching ratios for reactive channels leading to O₃ and OH production were quantified for the first time. A tropospheric model has been used to assess the impact of the experimental error of the rate coefficients determined in this study on predicted concentrations of a number of key species, including O₃, OH, HO₂, NO and NO₂. In all cases it is found that the propagated error is very small and will not in itself be a major cause of uncertainty in modelled concentrations. However, at low temperatures, where there is a wide discrepancy between existing kinetic studies, modelling using the range of kinetic data in the literature shows a small but significant variation for [C₂H₅O₂], [C₂H ₅OOH], [NO_x] and the HO₂:OH ratio. Furthermore, a structure-activity relationship (SAR) was developed to rationalise the reactivity of the reaction between RO₂ and HO₂. the Owner Societies.

Full text of DOI:10.1039/b703038j

View Article Online / Journal Homepage / Table of Contents for this issue
This paper is published as part of a PCCP Themed Issue on:
Synergies between Experimental and Theoretical
Studies of Gas Phase Reactions
Guest Edited by: Paul Seakins (Leeds) and Struan Robertson (Accelrys,
Cambridge and Leeds)
Published in issue 31, 2007 of PCCP
Image reproduced by permission of Dr Taatjes and Linda A.Houston for Sandia National
Laboratories from Phys.Chem.Chem.Phys., 2007, 9, 4315-4331
Other papers published in this issue include:
Direct detection of polyynes formation from the reaction of ethynyl radical (C2H) with propyne (CH3–
C
CH) and allene (CH2 C CH2)
Fabien Goulay, David L. Osborn, Craig A. Taatjes, Peng Zou, Giovanni Meloni and
Stephen R. Leone, Phys. Chem. Chem. Phys., 2007
DOI: 10.1039/b614502g
Master equation methods for multiple well systems: application to the 1-,2-pentyl system
Struan H. Robertson, Michael J. Pilling, Luminita C. Jitariu and Ian H. Hillier, Phys. Chem. Chem. Phys.,
2007
DOI: 10.1039/b704736c
Ab initio methods for reactive potential surfaces
Lawrence B. Harding, Stephen J. Klippenstein and Ahren W. Jasper, Phys. Chem. Chem. Phys., 2007
DOI: 10.1039/b705390h
Visit the website for both cutting edge research papers and authoritative
review articles by leaders in a range of fields
www.rsc.org/pccp

PAPER
www.rsc.org/pccp | Physical Chemistry Chemical Physics
Kinetics and branching ratio studies of the reaction of C2H5O2 + HO2
using chemical ionisation mass spectrometry
a
a
a
´
M. Teresa Raventos-Duran, Carl J. Percival,* Max R. McGillen,
Paul D. Hamerb and Dudley E. Shallcrossb
Received 28th February 2007, Accepted 4th May 2007
First published as an Advance Article on the web 29th May 2007
DOI: 10.1039/b703038j
The overall rate coefficient for the reaction of C2H5O2 with HO2 was determined using a
turbulent flow chemical ionization mass spectrometer (TF-CIMS) system over the pressure range
of 75 to 200 Torr and temperatures between 195 and 298 K. The temperature dependence of the
overall rate coefficient for the reaction between C2H5O2 and HO2 was fitted using the following
Arrhenius expression: k(T) = (2.08+ꢀ00..6827) ꢁ 10ꢀ13 exp [(864 ꢂ 79)/T] cmꢀ3 moleculeꢀ1
s
ꢀ1. The
upper limits for the branching ratios for reactive channels leading to O3 and OH production were
quantified for the first time. A tropospheric model has been used to assess the impact of the
experimental error of the rate coefficients determined in this study on predicted concentrations of
a number of key species, including O3, OH, HO2, NO and NO2. In all cases it is found that the
propagated error is very small and will not in itself be a major cause of uncertainty in modelled
concentrations. However, at low temperatures, where there is a wide discrepancy between existing
kinetic studies, modelling using the range of kinetic data in the literature shows a small but
significant variation for [C2H5O2], [C2H5OOH], [NOx] and the HO2 : OH ratio. Furthermore, a
structure–activity relationship (SAR) was developed to rationalise the reactivity of the reaction
between RO2 and HO2.
At high NO concentrations, the propagating reaction of
Introduction
C2H5O2 with NO will become important in the ozone forma-
tion cycle.
Ethane is an important hydrocarbon in the Earth’s lower
atmosphere, with a lifetime of up to a few months and source
strength of around 15 Tg yearꢀ1. The main anthropogenic
source is the petroleum industry having an emission flux of
B6 Tg yearꢀ1, but there are also natural emissions, arising
from forest fires of around 7 Tg yearꢀ1, oceans, terrestrial
biogenic sources such as soils, wetlands and vegetation etc. of
C2H5O2 þ NO ! C2H5O þ NO2
NO2 þ hvðl o 410 nmÞ ! NO þ O
O þ O2 þ M ! O3 þ M
ð3Þ
ð4Þ
ð5Þ
1 Tg yearꢀ1 1
. In the lower stratosphere, ethane plays an
In remote areas, where NOx concentrations are low, the
reaction of C2H5O2 with HO2 is assumed to yield C2H5OOH
exclusively, which is water soluble, acting as a net sink of
peroxy radicals in the atmosphere and therefore reducing the
oxidative capacity of the troposphere. However, previous
studies have suggested that the reaction of C2H5O2 with
HO2 can proceed via the four possible reaction channels,
namely
important role in determining the budgets of HOx and NOx
2 as well as being a precursor to peroxy acetyl nitrate (PAN) in
the troposphere. Once emitted into the atmosphere, the OH
radical oxidizes ethane to form the peroxy radical, C2H5O2, as
shown in reactions (1) and (2).
C2H6 þ OH ! C2H5 þ H2O
ð1Þ
C2H5O2 þ HO2 ! C2H5OOH þ O2
! C2H5OH þ O3
ð6aÞ
ð6bÞ
ð6cÞ
C2H5 þ O2 þ M ! C2H5O2 þ M
ð2Þ
C2H5O2 is known to have a significant impact on tropospheric
ozone concentrations.3–7 The fate of C2H5O2 is determined by
its reaction with NOx, HO2 or other RO2 radicals. Thus, the
relative importance of those reactions will be determined by
the concentrations of NOx, HOx and RO2 as well as their rate
coefficients.
! C2H5O þ OH þ O2
! CH3CHO þ H2O þ O2 ð6dÞ
The determination of both kinetics and mechanism of reaction
(6) has been carried out by a large number of studies employ-
ing UV absorption spectroscopy,8 which have significantly
enhanced our understanding of ethyl peroxy radical budgets
in the remote atmosphere. The study of the reactions of peroxy
radicals with HO2 are extremely difficult and rely on the
a The School of Earth, Atmospheric and Environmental Science, The
University of Manchester, Sackville Street, Manchester, UK M60
1QD. E-mail: C.Percival@manchester.ac.uk
b Biogeochemistry Research Centre, School of Chemistry, University
of Bristol, Cantock’s Close, Bristol, UK BS8 1TS
ꢃc
This journal is the Owner Societies 2007
4338 | Phys. Chem. Chem. Phys., 2007, 9, 4338–4348

selective detection of radicals. However it is very difficult to
resolve the RO2 and HO2 concentrations using the UV
absorption technique as a result of broad band spectral over-
lap, which adds significant errors to the retrieved rate coeffi-
cients. Furthermore, the calculated rate coefficients are
dependent on the absolute UV absorption cross sections of
both C2H5O2 and HO2 and previous reviews8 have reported
that the absorption cross section of C2H5O2 is not well
defined. Despite these difficulties, experimental studies are in
agreement with recent theoretical studies, indicating that
C2H5OOH is the principal product for the reaction of
C2H5O2 with HO2. However, the UV absorption method has
been unable to determine the branching ratios for the minor
channels (6b), (6c) and (6d).9 CIMS has proved to be a
powerful tool for the study of peroxy radical kinetics, facil-
itating the selective quantification of peroxy radicals and the
unambiguous quantification of product channels, thus avoid-
ing the problems associated with the UV absorption techni-
que.9,10
In this paper, the overall rate coefficient for reaction (6) was
investigated over an extended temperature and pressure range,
195–298 K and 75–200 Torr, also channels (6b) and (6c) were
monitored directly and their branching ratios quantified for
the first time. In addition, from the theoretical study presented
in this work, a correlation between stabilization energies and
the logarithm of experimental rate coefficients for the reactions
of alkyl peroxy radicals with hydroperoxy radicals was estab-
lished and a comparison of our results with other SARs shows
the reliability of our method.
Experimental
Fig. 1 A schematic diagram of the turbulent flow tube. C2H5O2 was
A schematic diagram of the experimental apparatus is shown
in Fig. 1. The flow tube was constructed from 23 mm id Pyrex
tubing, the walls of which were coated with Halocarbon wax
(Halocarbon Products Inc.). A large flow of nitrogen (ranging
from 50 to 130 STP l minꢀ1) was injected upstream of the flow
tube. The flow tube was pumped by a rotary pump (Varian
DS1602). HO2 radicals were injected through a moveable
double injector constructed from 9 mm id Pyrex. A propel-
ler-shaped Teflon piece (a ‘‘turbulizer’’)—designed to enhance
turbulent mixing—was fixed to the end of the moveable
injector. All the experiments were performed under turbulent
flow conditions.
generated at the side arm.
same temperature by passing it through a copper coil im-
mersed in liquid nitrogen. The carrier gas temperature was
maintained with heating tape (Omega, Heavy duty) regulated
by an electronic controller (Carel Universal Infrared control
type W) in conjunction with a thermocouple. The temperature
profile along the flow tube was checked by placing a thermo-
couple at the end of the moveable injector, and moving it
along the length of the flow tube.
HO2 was produced in the double nested sliding injector via
reaction (7)
The ion-molecule region was constructed from 23 mm id
Pyrex tubing and a quadrupole mass spectrometer (ABB
Extrel Merlin) was located at the end of the flow tube. All
gas flows were monitored with calibrated mass flow meters
(MKS, 1179). The pressures in the flow tube were monitored
using a 0–1000 Torr capacitance manometer (MKS Baratron).
The temperature within the flow tube was maintained within
2 K by placing the flow tube into an insulated chamber that
was filled with dry ice. All temperatures were monitored by
type K thermocouples. The flow tube temperature was main-
tained with heating tape (Omega, Heavy duty) regulated by an
electronic controller (Carel Universal Infrared control type W)
in conjunction with a thermocouple. The flow tube had 5
thermocouples along its length to monitor the temperature of
the system. The nitrogen carrier gas was pre-cooled to the
H þ O2 þ M ! HO2 þ M
ð7Þ
Hydrogen atoms were produced by combining a 2.0 STP l
minꢀ1 flow of He with a 0.1 to 3 STP cm3 minꢀ1 flow of 1% H2
in helium, which was then passed through a microwave
discharge produced by a Beenaker (Ophtos Instruments,
Inc.) cavity operating at 50 W. To produce HO2 radicals,
the H atoms were injected into the flow tube via a moveable
4 mm id inner quartz tubing injector and mixed with a 1.0 STP
l minꢀ1 flow of O2 that was added through the moveable 7 mm
id Pyrex outer injector. The double nested injector enables the
contact time for reaction (7) to be varied to ensure that all H
atoms have been titrated whilst minimising the loss of HO2 to
ꢃc
This journal is the Owner Societies 2007
Phys. Chem. Chem. Phys., 2007, 9, 4338–4348 | 4339

reaction (8).
molecules. The front vacuum chamber was pumped by a
mechanical pump (Varian DS402) and held at approximately
2 Torr. The ions were further focused by a 3 cm od and 0.2 mm
id stainless steel plate held at ꢀ15 V and passed into a second
chamber containing the quadrupole mass filter (ABB Extrel,
Merlin). This second chamber was pumped by a turbomole-
cular pump (Varian V250) backed by a rotary pump (Edwards
E2M8). The rear chamber which held the multiplier assembly
was pumped by a further turbomolecular pump (Varian V250)
backed by a rotary pump (Edwards E2M2). Under typical
operating conditions the rear chamber was at a pressure of
approximately 9 ꢁ 10ꢀ6 Torr. Ions were detected using a
channeltron (Dtech 402A-H) via negative ion counting.
HO2 þ HO2 þ M ! H2O2 þ O2 þ M
ð8Þ
At the pressures and flow conditions used in this study, it is
calculated that the H atoms have been completely titrated
before entering the flow tube. H atoms were periodically
titrated with NO2 to determine the concentration of H atoms
produced. In the absence of oxygen, known amounts of NO2
were added. The mass spectrometer signal (at m/z 46) was
monitored with the microwave turned off and then turned on.
The decrease in the NO2 signal observed when the microwave
was turned on was attributed to the reaction (9)
H þ NO2 ! NO þ OH
ð9Þ
Using this method it was found that on average, 20% of H2
was dissociated into H atoms. Under normal operating con-
ditions the initial concentration of hydrogen atoms in the flow
Materials
Helium (BOC, CP Grade) was passed through a gas clean
oxygen filter (Chrompak) cartridge to remove traces of oxy-
gen, a gas clean moisture filter cartridge (Chrompak) to
remove H2O, a GateKeeper Inert gas purifier (Aeronex, model
CE500KIAR) to remove H2 and a trap maintained at 77 K
containing a molecular sieve (Alfa Aesar, 4A) to remove
water. NO2 (Sigma Aldrich, 99.54%) and SF6 (BOC, electro-
nic grade) were purified by repeated freeze–pump–thaw cycles.
NO (BOC, Technical grade) was purified by freeze-pump-thaw
cycles, and selective freezing of NO2 impurities. Fluorine (5%
F2/He, BOC), N2, O2 (BOC, 99.6%), H2 (BOC, 99.999%) and
C2H6 (BOC, 99.995%) were used as supplied.
tube ranged from (1–10) ꢁ 1012 molecule cmꢀ3
.
C2H5O2 radicals were produced upstream of the flow tube
via reactions (10) and (11).
C2H6 þ F ! C2H5 þ HF
ð10Þ
ð11Þ
C2H5 þ O2 þ M ! C2H5O2 þ M
Fluorine atoms were produced by combining a 1.0 STP
l minꢀ1 flow of He with a 0.1 to 0.5 STP cm3 minꢀ1 flow of
5% F2 in helium, which was then passed through a microwave
discharge produced by a Surfatron (Sairem) cavity operating
at 60 W. To produce C2H5 radicals, the F atoms were injected
into the flow tube via a side arm inlet located at the rear of the
flow tube and mixed with an excess of C2H6, to ensure that all
F atoms were titrated via reaction (10). C2H5O2 was then
produced by the addition of a 1.0 STP l minꢀ1 flow of O2 just
downstream from the production of ethyl radicals.
Results and discussion
Assessment of detector sensitivity
Dilute mixtures of NO2 were injected via the moveable injector
ꢀ
All experiments were performed using an excess of HO2,
[HO2] 4 [C2H5O2]. Blank runs (in the absence of HO2) were
carried out to ensure that the C2H5O2 signal (measured at m/z
61, i.e. C2H5O2ꢀ) was not affected by movement of the
injector.
into the flow tube with no other gases present and the NO2ꢀ
signal was monitored. From a linear plot of [NO2] vs. NO2
signal it is estimated that the sensitivity for NO2 was 3.5 ꢁ
107 molecule cmꢀ3 (counts per second, cps)ꢀ1 for a signal to
noise ratio of one, a time constant of 1 s and an initial SF6ꢀ of
1 ꢁ 106 (cps)ꢀ1. The NO2 concentrations were corrected to
take into account equilibrium concentrations of N2O4 in the
gas mixtures used and were typically of the order of 10%.
Under the experimental conditions the lifetime of N2O4
formed by the equilibrium (12)
Trace constituents within the flow tube were chemically
ꢀ
ꢀ
ionized using SF6 as the reagent ion. SF6 was generated
by combining a 10 STP l minꢀ1 flow of N2 with a 2.5 STP cm3
minꢀ1 flow of SF6 and passing it through a 210Po Nuclecel
ionizer (NRD Inc.). The generated reagent ion was then
carried into the ion-molecule region through an injector con-
structed from 6 mm od stainless steel tubing. A fan shaped
turbulizer was attached to the end of the inlet to enhance
mixing of the reagent ion with the sampled flow from theꢀflow
tube. O3, OH, C2H5O2 and NO2 were ionized by SF6 via
electron transfer enabling detection by the parent ion. HO2
NO2 þ NO2 þ M Ð N2O4 þ M
ð12Þ
is comparable with the time of mixing.12
Calibration of the RO2 (where R = H or C2H5) signal was
achieved by adding a large excess of NO (using a Z 50% NO/
He mixture) via the moveable injector at a constant contact
time and by monitoring the resultant NO2 formed by reaction
with RO2. Sufficient NO was added to ensure completꢀe
removal of RO2 that was confirmed by a constant NO2
signal with increasing [NO]. It is assumed that [NO2]observed
= [RO2]. This procedure was repeated for several different
ꢀ
was detected as SF4O2 presumably through a multi step
pathway as reported by previous studies.11 All experiments
ꢀ
were carried out under conditions such that the initial SF6
signal was not depleted by more than 10%.
Ions were detected with a quadrupole mass spectrometer in
a three-stage differentially pumped vacuum chamber. A sam-
ple of the bulk gas flow containing reactant ions is drawn into
the front chamber through a 0.6 mm id aperture which was
held at a potential of ꢀ70 V to further focus charged reactant
ꢀ
fluorine atom concentrations and yielded a linear plot of RO2
signal vs. [NO2]observed. It is estimated that the sensitivity for
HO2 and C2H5O2 was 1.50 ꢁ 107 and 2.8 ꢁ 107 molecule cmꢀ3
ꢃc
This journal is the Owner Societies 2007
4340 | Phys. Chem. Chem. Phys., 2007, 9, 4338–4348

(cps)ꢀ1 for a signaꢀl to noise ratio of one, a time constant of 1 s
and an initial SF6 of 1 ꢁ 106 (cps)ꢀ1, respectively.
Dilute mixtures of O3 were injected via the moveable
injector into the flow tube with no other gases present and
the O3ꢀ signal was monitored. O3 was generated by passing a
1 STP L minꢀ1 O2 flow through a UVP ozone generator. The
output from the O3 UVP generator was calibrated using a
Perkin Elmer (Lamꢀda 40) UV-Vis spectrometer. From a linear
plot of [O3] vs. O3 signal it is estimated that the sensitivity
for O3 was 3.8 ꢁ 107 molecule cmꢀ3 (cps)ꢀ1 for a signal to
noise ratio of one, a time constant of 1 s and an initial SF6ꢀ of
1 ꢁ 106 (cps)ꢀ1
.
Calibration of the OH sensitivity was performed in a
manner described by Canosa-Mas and Wayne.13 OH radicals
were produced via reaction (9). H atoms were injected into the
flow tube via a side arm inlet located at the rear of the flow
tube and in the absence of oxygen, known amounts of NO2
were added via the moveable injector. The mass spectrometer
OHꢀ signal (at m/z 17) was monitored as a function of the
NO2 concentration added at various contact times. It is
estimated that the sensitivity for OH was 3.2 ꢁ 107 molecule
cmꢀ3 (cps)ꢀ1 for a signalꢀto noise ratio of one, a time constant
Fig. 2 Kinetic plot of [C2H5O2] versus time for the C2H5O2 + HO2 at
298 K, 75 Torr, [C2H5O2]0 = 1.25 ꢁ 1012 molecule cmꢀ3 and [HO2]0 =
2.61 ꢁ 1013 molecule cmꢀ3. (E) experimental values, (—) predicted
values from the model and (— —) upper and lower values of the
decay.
of 1 s and an initial SF6 of 1 ꢁ 106 (cps)ꢀ1
.
represent the error in all of the experimentally determined rate
coefficients adequately and thus the maximum error has been
quoted throughout the work presented in this paper.
Rate coefficient determination
The rate coefficients in the model can be varied by a factor
of ten without affecting the final value of k6 with the notable
exception of the HO2 self-reaction. However, within the
experimental uncertainty reported by Sander et al.,8 no effect
on the retrieved rate coefficient for reaction (6) is observed.
The bimolecular rate coefficients used in this study are listed in
Table 1. This approach for the determination of the effective
bimolecular rate coefficient assumes that deviations from the
plug flow approximation are negligible. Under the experimen-
tal conditions used, Seeley et al.,18 estimated that deviations
from the plug flow approximation result in apparent rate
coefficients that are at most 8% below the actual values.
Hence, flow corrections were neglected, as they are smaller
than the sum of other systematic experimental errors.
The overall rate coefficient was evaluated by simultaneously
monitoring the C2H5O2 and HO2 concentration time profiles.
All experiments were performed under an excess of [HO2],
with [C2H5O2] = 1–5 ꢁ 1012 molecule cmꢀ3 and [HO2] =
1–10 ꢁ 1012 molecule cmꢀ3. HO2 was used in excess as the
product of the self-reaction is a molecular product that will not
result in any complicated secondary chemistry.3–9 Whilst, the
C2H5O2 self-reaction is slower, the major product is C2H5O
which could potentially lead to unwanted secondary chemis-
try. The HO2 decay (monitored at m/z = 140) was dominated
by loss via the HO2 self-reaction as a consequence of the [HO2]
being in excess. Thus the rate coefficient for reaction (6) was
determined by monitoring C2H5O2 (at m/z = 61). A typical
kinetic plot of [C2H5O2] versus time is shown in Fig. 2. An
extensive reaction mechanism was used to analyse the con-
centration-time profiles of C2H5O2 (ACUCHEM, 1995). The
relevant chemical reactions and rate parameters used in the
mechanism are given in Table 1. The model was used to
generate predicted curves for the loss of C2H5O2 and HO2
based on experimental conditions and precursor concentra-
tions, and then fitted to the experimental data. Fig. 2 shows a
typical fit of the modelled [C2H5O2] concentration–time pro-
file, showing excellent agreement between the experimental
and modeled data. The dashed lines represent changes in the
modelled [C2H5O2] concentration time profile assuming an
experimental error of 0.6 ꢁ 10ꢀ12 cm3 moleculeꢀ1 sꢀ1 in the
rate coefficient for reaction (6). The experimental error reflects
the range of rate coefficients for reaction (6) that could be used
to model these data satisfactorily. Experimental error in the
rate coefficient was determined by altering the rate coefficient
for reaction (6) in the kinetic model until modelled fit exceeded
the standard deviation of the C2H5O2 signal. A maximum
uncertainty of 0.6 ꢁ 10ꢀ12 cm3 moleculeꢀ1 sꢀ1 was found to
Table 2 shows the overall rate coefficient for experiments
carried out at several pressures and 298 K. Within error, each
data set produces the same effective bimolecular rate coeffi-
cient indicating that the rate coefficient is invariant with
pressure. From all model runs, a rate coefficient at 298 K of
(4.0 ꢂ 0.6) ꢁ 10ꢀ12 cm3 moleculeꢀ1 sꢀ1 is determined, where
the quoted error reflects the range of rate constants that could
be used to model these data satisfactorily. There is a consider-
able range in the previously reported experimental rate coeffi-
cients ranging from (5.3 ꢂ 1.0) to (11.0 ꢂ 2.1) ꢁ 10ꢀ12 cm3
moleculeꢀ1 ꢀ1. Our room temperature rate coefficient is in
s
good agreement with the lower values of the previous studies.6
All previous experimental studies3,4,6,7 have measured the rate
coefficients using the UV-absorption technique and thus de-
convolution procedures were necessary in order to obtain
radical concentrations. As noted by Sander et al.,8 the value
of the absorption cross section utilized for C2H5O2 is the most
likely explanation of the inconsistency between the previous
determinations. Interestingly, Cattell et al.,7 reported two rate
coefficients, a value of (6.3 ꢂ 0.9) ꢁ 10ꢀ12 cm3 moleculeꢀ1 sꢀ1
ꢃc
This journal is the Owner Societies 2007
Phys. Chem. Chem. Phys., 2007, 9, 4338–4348 | 4341

Table 1 A summary of the chemical reactions and their rate parameters of the modelled mechanism to predict C2H5O2 curves
Reaction
k/cm3 moleculeꢀ1 sꢀ1
4.0 ꢁ 10ꢀ12
Ref.
a
C2H5O2 + HO2 - C2H5OOH + O2
- C2H5OH + O3
- C2H5O + OH + O2
14
15
8
- CH3CHO + H2O + O2
C2H5O2 + C2H5O2 - C2H5O + C2H5O + O2
C2H5O2 + C2H5O2 - C2H5OH + CH3CHO
C2H5O + O2 - HO2 + CH3CHO
HO2 + HO - Products
C2H5O + C2H5O - Products
C2H5O + C2H5O2 - CH3CHO + C2H5OOH
C2H5O - CH3 + CH2O
C2H5O - CH3CHO + H
H + O2 - HO2
CH3 + O2 - CH3O2
CH3O2 + HO2 - Products
4.18 ꢁ 10ꢀ14
2.45 ꢁ 10ꢀ14
1.01 ꢁ 10ꢀ14
3.02 ꢁ 10ꢀ12
7.00 ꢁ 10ꢀ12
1.36 ꢁ 10ꢀ12
1.86 ꢁ 10ꢀ02
1.78 ꢁ 10ꢀ03
1.74 ꢁ 10ꢀ13
4.93 ꢁ 10ꢀ13
3.80 ꢁ 10ꢀ13
1.00 ꢁ 10ꢀ10
2.48 ꢁ 10ꢀ13
1.22 ꢁ 10ꢀ13
1.50 ꢁ 10ꢀ15
2.60 ꢁ 10ꢀ12
1.30 ꢁ 10ꢀ11
6.70 ꢁ 10ꢀ15
9.28 ꢁ 10ꢀ13
9.54 ꢁ 10ꢀ12
2.00 ꢁ 10ꢀ13
8
14
15
8
8
b
16
17
8
8
15
8
CH3O2 + OH - Products
CH3O2 + CH3O2 - CH3O + CH3O + O2
CH3O2 + CH3O2 - CH2O + CH3OH + O2
CH3O + O2 - CH2O + HO2
CH3O + CH3O2 - CH2O + CH3OOH
CH3O + CH3O - CH2O + CH3OH
HO2 + CH2O - Adduct
CH3O + C2H5O2 - Products
CH3O + C2H5O - Products
CH3O2 + C2H5O2 - CH3O + RO + O2
8
8
b
b
21
b
b
21
a
b
Refer to text. Calculated using geometric mean rule.
utilising infrared diode-laser spectrometric detection of HO2
and ultraviolet absorptions detection of C2H5O2 at low pres-
sures (2 Torr) where spectral deconvolution is not an issue and
a value of 9.3 ꢁ 10ꢀ12 cm3 moleculeꢀ1 sꢀ1 employing ultra-
violet-absorption detection of radicals for experiments at
ambient pressures. This had previously been attributed to a
pressure dependence by Fenter et al.4 However, this work and
recent theoretical work (see later) indicate that the overall rate
coefficient would not display a pressure dependence. In con-
trast with the previous studies, the detection system utilized in
this work enables the unambiguous quantification of both
reactants (C2H5O2 and HO2) and thus reduces the errors in
determining the experimental rate coefficient significantly and
would further support the hypothesis that the errors asso-
ciated with UV absorption cross sections and subsequent
spectral deconvolution can be significant. Furthermore, the
sensitivity of the CIMS detection enables much lower radical
concentration to be used (e.g. the maximum [C2H5O2]0 was
5 ꢁ 1012 molecule cmꢀ3 in comparison to 20 ꢁ 1013 molecule
cmꢀ3 utilised by Fenter et al.,4) which significantly reduces the
secondary chemistry. In this work we only had to consider the
effect of the HO2 self-reaction. However, previous studies also
need to include the C2H5O2 radical self-reactions kinetics
which can add further uncertainties in the experimentally
retrieved rate coefficients.
The rate coefficient for reaction (6) was also studied over the
temperature ranges 195–298 K. Table 3 summarizes the rate
coefficients obtained in this study. The rate coefficients in-
creased by approximately a factor of 4 as the temperature was
lowered from 298 to 195 K. Over the temperature and pressure
range studied, no effect of pressure on the measured effective
bimolecular rate coefficient was observed. Using the data in
Table 3, it is possible to carry out an Arrhenius type analysis
of the temperature dependence yielding an ‘‘Arrhenius’’ ex-
pression of k(T) = (2.08+0.87
ꢀ0.62 ) ꢁ 10ꢀ13 exp [(864 ꢂ 79)/T]
cmꢀ3 moleculeꢀ1 ꢀ1, i.e. an apparent negative activation
s
energy is observed. The uncertainty associated with the rate
coefficient is given at the one standard deviation level.
Three other experimental studies have been carried out to
investigate the temperature dependence of reaction (6) and all
three used the flash photolysis-UV absorption technique; at
higher pressures, (760 Torr), Fenter et al.,4 monitored HO2
and C2H5O2 decays at 220 nm and 260 nm and analysed the
experiments by fitting both decays simultaneously. Dagaut
et al.,6 monitored the decay curve at 250 nm and at 100 Torr
and Maricq and Szente3 covered the widest range of tempera-
tures (220–363 K) monitoring the absorbance over the range
200–250 nm and at different pressures (176–698 Torr). Fig. 3
shows a comparison of the temperature dependence data
reported here and from previous studies. The negative activa-
tion energy obtained in this work is in fair agreement with the
Ea/R values of (ꢀ650 ꢂ 125) K determined by Dagaut et al.,6
and (ꢀ700 ꢂ 200) K by Maricq and Szente,3 whereas the value
of (ꢀ1260 ꢂ 130) K reported by Fenter et al.,4 is well above
the previous determinations and that of this work. In a recent
Table 2 Experimentally determined rate coefficients for the reaction
of C2H5O2 + HO2 at 298 K and pressures ranging from 75 to 200 Torr
T/K
P/Torr
k6/10ꢀ12 cm3 moleculeꢀ1 sꢀ1
298
298
298
298
298
298
298
75
4.50 ꢂ 0.5
2.00 ꢂ 0.5
4.50 ꢂ 0.5
3.10 ꢂ 0.5
4.50 ꢂ 0.5
4.70 ꢂ 0.5
4.50 ꢂ 0.5
100
100
100
125
150
200
ꢃc
This journal is the Owner Societies 2007
4342 | Phys. Chem. Chem. Phys., 2007, 9, 4338–4348

Table 3 Experimentally determined bimolecular rate coefficients and upper limits of branching ratios for C2H5O2 + HO2 reaction at T r 298.
The quoted rate coefficients are in units of 10ꢀ12 cm3 moleculeꢀ1 sꢀ1
Branching ratio upper limits
Branching ratioa upper limits
T/K
P/Torr
k6
6b
6c
6b
6c
298
243
233
223
205
195
a
75–200
75–150
75–150
75–150
75–150
75–150
3.97 ꢂ 0.5
0.014
0.010
0.007
0.007
0.006
0.005
0.052
0.026
0.019
0.013
0.013
0.011
0.020
0.018
0.016
0.015
0.015
0.014
0.103
0.057
0.043
0.025
0.024
0.019
6.19 ꢂ 0.5
8.88 ꢂ 0.5
11.0 ꢂ 0.5
13.0 ꢂ 0.5
18.4 ꢂ 0.5
Revised branching ratios if secondary ion molecule chemistry is included (refer to text).
review Tyndall et al.,19 suggest that given the difficulty asso-
ciated with measuring cross-reaction kinetics that the agree-
ment is fair and that the spread in the reported values of the
Arrhenius parameters is once again as a result of errors in the
UV absorption cross sections. The ionization schemes utilized
in this study enable the selective quantification of all peroxy
radical concentrations without interference and thus reduce
the errors associated with the measured rate parameters.
short-lived to be affected by collisions with the bulk gas (N2)
even at the extended temperatures and pressures of this study.
Our results would indicate that the rate determining step for
reaction (6) is actually reaction (13), i.e. the rate of formation
of the C2H5O4H intermediate. Since the re-dissociation reac-
tion is negligible at all pressures over the temperature range
studied, the rate coefficient for the overall reaction (6) cannot
be pressure dependent and in addition the rate coefficient for
reaction (6) might be expected to decrease with increasing
temperature, because the re-dissociation of the C2H5O4H
could begin to compete with channel (14).
Mechanism
The negative activation energy of the rate coefficient for
reaction (6) suggests that it proceeds by a tetraoxide inter-
mediate [C2H5O4H]*. The fate of the intermediate will then
determine the product distribution. The elements of the me-
chanism may be summarised as:
All previous experimental studies6,7,20–23 based on the end
product analysis for the reaction (6), have mainly focused on
C2H5OOH formation and suggest that within error the reac-
tion proceeds solely via channel (6a). Such experiments do
observe small concentrations of CH3CHO, however the
authors can easily account for this as a product from second-
ary reactions and discredited its formation by channel (6d).
Analogous to the reaction of CH3O2 + HO2 it is expected that
channel a, i.e. ROOH, would dominate on thermodynamic
grounds (as it is the most exothermic path) and channels b–d
should be negligible. Indeed, recent theoretical and experi-
mental studies by Hasson et al.,21 (see later) have verified this.
However, it was not possible to detect any of the products
associated with channel 6d with the current experimental setup
and additional studies of the direct quantification would be
useful. Similar to our previous work,24 we present the first
direct study to quantify the branching ratio to channels (6b)
and (6c). The reaction products O3 (m/z = 48) and OH (m/z =
17) from reactions (6b) and (6c) were simultaneously mon-
itored with C2H5O2 and HO2 by CIMS. Over the range of
pressures and temperatures used in this study, the signals
corresponding to O3 and OH were not observed, even at the
longest contact times. Therefore, it is only possible to quantify
an upper limit for channel (6b) and (6c). The branching ratios
were evaluated by using the reaction scheme given in Table 1
and varying the branching ratio of reaction (6) until the
modelled O3 or OH was above the limit of detection. For this
work, the limit of detection was assumed to be the concentra-
tion that would produce signal greater than three times the
standard deviation of the background signal at the corre-
sponding mass to charge ratio. The limit of detection for
O3 and OH were 5.45 ꢁ 109 molecule cmꢀ3 and 2.98 ꢁ
109 molecule cmꢀ3, respectively. Table 3 summarizes the upper
limits of the branching ratios for both channels quantified in
this study.
C2H5O2 þ HO2 ! ½C2H5O4Hꢄꢅ
ð13Þ
ðꢀ13Þ
ð14Þ
½C H5O4Hꢄꢅ ! C2H5O2 þ HO2
2
½C H5O4Hꢄꢅ ! prods
2
The absence of any experimentally observed pressure depen-
dence, and the excellent agreement with the low pressure rate
coefficients19 suggests that the [C2H5O4H]* intermediate is too
Fig. 3 Comparison of Arrhenius plots of the rate coefficient for the
reaction C2H5O2 + HO2 from (E) Fenter et al.;4 (n) Maricq and
Szente;3 (m) Dagaut et al.,6 and (’) this work. The solid lines
represent an Arrhenius fit to each experimental study and the dashed
line is the temperature dependence as suggested by Sander et al.8
ꢃc
This journal is the Owner Societies 2007
Phys. Chem. Chem. Phys., 2007, 9, 4338–4348 | 4343

The branching ratios reported in Table 3 assumes an
identical limit of detection for O3 and OH during a kinetics
experiment as evaluated during a calibration experiment. The
C2H5O2 is produced via reactions (10) and (11) leading to the
formation of HF as one of the co-products of reaction (10).
Experimental data25 indicates that the following competing
ion molecule reaction can occur at the gas kinetic limit
chanism of RO2 + HO2 reactions. Hasson et al.,26 suggest
that the reactions proceed via hydrogen-bonded complexes
and that once formed, the [C2H5O4H]* intermediate is too
short-lived to be affected by collision with the N2 bath gas and
that re-dissociation back to reactants is insignificant, thus
formation of the [C2H5O4H]* is the rate limiting step and no
pressure dependence of the overall reaction is expected which
is in good agreement with this study. Hasson et al.,26 suggest
that channel (6a) should dominate as [C2H5O4H]* would not
dissociate to form products associated with channels (6b–6d)
as the exit barriers are too large, which is in agreement with
this study.
HOꢀ þ HF ! products
ð15Þ
If one assumes that both OHꢀ and O3ꢀ, react with HF at the
gas kinetic limit; the limit of detection of the CIMS instrument
towards OH and O3 will be reduced. The maximum [C2H5O2]0
used in this study was 5 ꢁ 1012 molecule cmꢀ3, which equates
to the maximum HF that can be produced in the flow tube. If
one assumes that there is a pseudo-first order loss, as HF is in
In analogy to our previous work which demonstrated that
the stabilization energy will control the rate of reaction for
RO2 with RO2 and RO2 with CH3O2, it is also likely that the
stabilization energy for the cross-reaction RO2 with HO2 will
determine reactivity as the rate of formation of [RO4H]* is the
rate limiting step. Experimental and recent theoretical stu-
dies26,27 have shown the formation of the hydrotetraoxide
intermediate is the rate limiting step. Thus, it is conceivable
that, the stabilization energy (DHstab.RO +HO ) will govern the
ꢀ
excess, of product ions, i.e. O3 and OHꢀ, our experimental
limit of detection for O3 and OH is reduced to 1.30 ꢁ
1010 molecule cmꢀ3 and 5.1 ꢁ 109 molecule cmꢀ3, respectively.
Table three also summarizes the upper limits of the branching
ratios for both channels, assuming this reduction in sensitivity
as a result of the possible reaction of product ions.
2
2
reactivity. Table 4 shows the peroxy radicals studied in this
work, the rate coefficients of the reaction of RO2 with HO2
and the heats of formation of the peroxy radical and the
tetraoxide intermediate, which were calculated via simple PM3
semi-empirical methods.24 Fig. 4 shows a plot of the logarithm
of the room temperature rate constants for the RO2 with HO2
reactions as a function of the stabilisation energy {DHf(RO4H)
ꢀ (DHf(RO2) + DHf(HO2))}. The graph displays a correlation
implying that the larger the stabilisation energy, the higher the
rate of reaction. The data were fitted using a linear least-
squares routine and can be represented by eqn (I).
The recent study by Hasson et al.,21 determined the product
yields of reaction (6) via end product analysis using Fourier
transform infrared spectroscopy and high-performance liquid
chromatography with fluorescence detection. The experimen-
tal analyses of Hasson et al.,21 were carried out by using large
initial concentrations of peroxy radicals (E1015 molecules
cmꢀ3) which could lead to an increase of secondary reactions.
By using TF-CIMS, our experimental studies were carried out
with [RO2] three orders of magnitude lower than those used by
Hasson et al.21 However, as shown in Table 3, the maximum
values of branching ratios for OH-product channel and
O3-product channel were 0.05 and 0.01, respectively. The
insignificant amount of OH and O3 is in agreement with the
dominant ROOH production of 93% determined by Hasson
et al.21
fDHf ðRO4HÞ ꢀ ðDHf ðRO2Þ þ DHf ðHO2ÞÞg
¼ ꢀ69:66 log10 k ꢀ 808:95
ðIÞ
Despite the relatively limited range of the kinetic data, the
correlation is quite striking, with a correlation coefficient, r, of
0.87 for peroxy radicals derived from simple hydrocarbons.
However, if the approach is extended to halogenated peroxy
Theoretical studies
In a recent theoretical study, Hasson et al.,26 used B3LYP/
6-311G(2d,d,p) and CB2-QB3 scheme to rationalize the me-
Table 4 Summary of the peroxy radicals studied in this work, the rate coefficients of the reaction of RO2 with HO2 and the heat of formation of
the peroxy radical and the tetraoxide intermediate
k ꢁ10ꢀ12
(cross reaction)/
A = DHf (RO4H)/ B = DHf (RO2)/ C = DHf (HO2)/ D = B + C/ AꢀD/
kJ molꢀ1 cm3 moleculeꢀ1 sꢀ1 Ref.
R
kJ molꢀ1
kJ molꢀ1
kJ molꢀ1
kJ molꢀ1
CH3
C(CH3)3CH2
ꢀ24.79
12.50
ꢀ73.90
ꢀ259.20
ꢀ39.76
ꢀ222.87
ꢀ218.13
ꢀ176.80
ꢀ171.62
ꢀ67.52
ꢀ149.9
128.96
ꢀ11.20
ꢀ11.20
ꢀ11.20
ꢀ11.20
ꢀ11.20
ꢀ11.20
ꢀ11.20
ꢀ11.20
ꢀ11.20
ꢀ11.20
ꢀ11.20
ꢀ11.20
ꢀ11.20
ꢀ11.20
ꢀ11.20
1.30
ꢀ85.10
ꢀ270.40
ꢀ50.96
ꢀ234.07
ꢀ229.33
ꢀ188.00
ꢀ182.82
ꢀ78.72
ꢀ161.12
117.76
ꢀ26.09
ꢀ55.91
ꢀ61.24
ꢀ61.09
ꢀ52.40
ꢀ50.87
ꢀ60.00
ꢀ54.72
ꢀ67.28
ꢀ37.61
ꢀ40.93
ꢀ28.68
ꢀ17.51
ꢀ58.98
ꢀ59.00
5.80
14.70
20.10
18.00
14.00
15.00
12.00
15.00
17.00
14.00
10.10
9.00
29
30
31
29
32
31
29
31
29
29
33
34
ꢀ141.01
(CH3)2C(OH)C(CH3)2 ꢀ331.64
c-C5H9
C(OH)(CH3)2CH2
ꢀ112.05
ꢀ286.47
CH3CH(OH)CH(CH3) ꢀ280.20
HOCH2
ꢀ248.00
ꢀ237.54
ꢀ146.00
ꢀ198.73
76.83
ꢀ170.86
ꢀ38.31
ꢀ190.82
29.58
HOCH2CH2
c-C6H11
CH3C(O)
C6H5CH2
CH3C(O)CH2
C2H5
ꢀ130.98
ꢀ142.18
ꢀ20.80
ꢀ131.84
88.58
ꢀ9.60
4.50
20.00
11.00
This work
35
36
(CH3)3C(CH3)2CH2
CH2CHCH2
ꢀ120.64
99.78
ꢃc
This journal is the Owner Societies 2007
4344 | Phys. Chem. Chem. Phys., 2007, 9, 4338–4348

radicals, no correlation is observed, which could possibly
indicate a change in reaction mechanism.
In a recent theoretical study, Hou and Wang27 used
CCSD(T)/cc-pvdz//B3lYP/6-311G(d,p) to rationalize the me-
chanism of RO2 with HO2 reactions (where R = CH3 and
CH2F). Their work suggests that the RO2 and HO2 cross-
reactions can occur either on a singlet or a triplet surface, as
shown in Fig. 5. For peroxy radicals derived from simple
hydrocarbons, reaction (6) occurs primarily on the triplet
surface via formation of an energised complex (a seven-mem-
bered-ring adduct). The work would indicate that reaction via
the singlet surface, i.e. the [RO4H]* considered in this study,
will be insignificant as a result of large exit barriers to product
formation. However, as Hou and Wang27 pointed out, the
intermediate wells for both surfaces are similar in depth and
therefore, the collision complex can sample both wells. Thus
the stabilisation energy calculated in this work will represent
the changes in reactivity for peroxy radicals derived from
simple hydrocarbons as, to a rough approximation, the
Fig. 4 A plot of the stabilisation energies versus the natural logarithm
of rate coefficients for RO2 + HO2 reactions of halogenated peroxy
radicals (n) and hydrocarbon peroxy radicals (&).
Fig. 5 Schematic diagram for the potential energy surface of the CH2FO2 + HO2 and CH3O2 + HO2 reactions. Adapted from Hou and
Wang.23
ꢃc
This journal is the Owner Societies 2007
Phys. Chem. Chem. Phys., 2007, 9, 4338–4348 | 4345

Table 5 A summary of various SAR approaches and the corresponding r and r2 calculated with the correlations of molecular properties in
function of RO2 + HO2 reactivity
Regression coefficient (r) of peroxy radicals correlation
Regression coefficient (r2) of peroxy radicals correlation
Ref.
Hydrocarbons only
Reported for over all study
Hydrocarbons only
Reported for over all study
This work
37
38
0.87
0.77
0.30
0.87
0.97
0.95
0.75
0.59
0.09
0.75
0.94
0.90
stabilization energy corresponds to the energy difference on
the triplet surface.
approaches assume that all RO2 and HO2 reactions proceed
via the same mechanism, which is in contradiction to the
findings of more recent theoretical studies of Hou and Wang27
and Hasson et al.,26 so it is unclear why a correlation would
exist for the range of peroxy radicals studied. Nevertheless,
applying these methods exclusively to peroxy radicals derived
from simple hydrocarbons, a direct comparison of the perfor-
mance of the techniques can be made. Table 5 shows for each
different SAR approach the corresponding r and r2 associated
with the correlation. For the small range of peroxy radicals
considered, the approach developed in this work returns a
value of r2 = 0.75, whereas the frontier orbital approach of
King et al.,37 has an r2 = 0.59 and the Johnson’s approach has
r2 = 0.09. Our approach was found to correlate as well, if not
better than previously reported SAR approaches. Whilst our
approach is limited to peroxy radicals derived from simple
alkanes they are, by far, the most dominant peroxy radicals in
the troposphere.39
However, for the fluorinated peroxy radical, CH2FO2, two
pathways for the peroxy radical cross-reaction were described,
as shown in Fig. 5. The triplet pathway proceeds via the
formation of a seven-membered-ring complex with two hydro-
gen bonds and subsequent hydrogen abstraction leading to the
formation of ROOH, i.e. channel (6a). However, as a result of
the reduction of barrier heights the singlet pathway can also
operate. The singlet surface is more complicated, there is a
direct hydrogen abstraction pathway leading to the formation
of ROOH that is kinetically accessible, i.e. channel (6a). Or
reactive channels that proceed via the formation of the tetra-
oxide intermediate [RO4H]* that decomposes via addition/
elimination to form RCHO, ROH or RO, i.e. channels
(6b–6d). The energies for the singlet and triplet tetraoxide
intermediate [RO4H]* are no longer similar, so for the reaction
of CH2FO2 with HO2, the calculated stabilisation energy given
in Table 4, would not accurately represent the reaction surface.
Furthermore, as other reaction pathways can now operate the
formation of the [RO4H]* would not be rate limiting and thus
the stabilization energy would not be expected to govern
reactivity. In Fig. 4 it can be seen that our calculations suggest
that the stabilisation energy does not correlate for all haloge-
nated peroxy radicals, which is consistent with the findings of
Hou and Wang27 that for halogenated peroxy radicals, i.e. the
reaction surface cannot be assumed to proceed via the forma-
tion of a single energised adduct. Whilst it has been shown that
the use of TF-CIMS can be utilised to quantify branching
ratios experimentally9,28 either using positive ionisation
schemes to quantify channel (6a) and (6d) or negative ionisa-
tions schemes to quantify channels (6b) and (6c), it is not
possible to differentiate between the triplet or singlet reaction
surface. In particular, it is important to identify the electronic
state of O2. As we have generalised the study of CH2FO2 from
Hou and Wang27 to explain a possible behaviour for all the
halogenated peroxy radicals, further theoretical work is
needed on more halogenated, especially the chlorinated and
brominated peroxy radicals.
Atmospheric implication
In general, the rate coefficient for reaction (6) determined in
this study (e.g. k298 K = 4.0 ꢁ 10ꢀ12 cm3 moleculeꢀ1
s
ꢀ1) is a
lot smaller than that recommended by the most recent JPL
evaluation (k298 K = 8.0 ꢁ 10ꢀ12 cm3 moleculeꢀ1
s
ꢀ1).8 The
overall impact of this difference in rate coefficient on key
atmospheric species is small but worthy of note. A number
of simulations have been performed using a 2D-box model
described by Fish et al.,40 simulations were carried out at the
temperatures, 205, 213, 233, 243 and 298 K, with NOx levels
ranging from 30 to 1200 ppt, at pressures appropriate to the
mid troposphere (2–8 km) in summer at a latitude of 40 N.
Comparing integrations using the NASA evaluation and the
smaller rate coefficient (as measured in this study) shows,
unsurprisingly, that a reduction in k6 leads to elevated
C2H5O2 and HO2 radical concentrations and decreased levels
of C2H5OOH. The impact is most pronounced at the lowest
temperatures where C2H5O2 rises by as much as 80% (0.5 ppt
absolute) and HO2 rises by between 5–20% (around 1 ppt
absolute). At the lowest temperatures there is also a reduction
in NOx of around 15% (5–20 ppt) and of C2H5OOH by as
much as 50% (down by between 25 and 60 ppt). At low
temperatures, an increase in HO2 and C2H5O2 will reduce NO
and NO2 levels via the reactions
Other recent studies have developed SARs for RO2 and
HO2 reactions. Most notably, King et al.,37 have reported
correlations for the reaction of a range of peroxy radicals with
HO2, based on a frontier orbital approach. They report
correlation coefficients similar to that obtained in this study,
with a correlation coefficient, r, of 0.97. More recently John-
son et al.,38 have reported a structure–activity type relation-
ship for the reactions of self and cross reactions for a series of
peroxy radicals. Their analysis shows a correlation (r = 0.95)
when log k is plotted vs. calculated electrostatic potential of a
closed shell molecular analogue of the peroxy radical. Both
HO2 þ NO ! OH þ NO2
C2H5O2 þ NO ! C2H5O þ NO2
HO2 þ NO2 þ M ! HO2NO2 þ M
ð16Þ
ð17Þ
ð18Þ
ꢃc
This journal is the Owner Societies 2007
4346 | Phys. Chem. Chem. Phys., 2007, 9, 4338–4348

where the stability of HO2NO2 at low temperature is a key
factor. In addition the HO2 : OH ratio will also increase at low
temperature (i.e. OH will decrease slightly). As temperature
increases (at 298 K for example) the importance of reaction
(16) is diminished as the decomposition of HO2NO2 is rapid.
Therefore reactions (14) and (15) operate and rapidly recycle
NOx through photolysis of NO2.
References
1 D. Jaffe, T. C. D. Anderson, B. Trost, J. S. W. Danielson, D. H. J.
Blake and D. Streets, J. Geophys. Res., 2001, 106, 7449.
2 D. J. Lary and D. E. Shallcross, J. Geophys. Res., 2000, 105, 19771.
3 M. M. Maricq and J. J. Szente, J. Phys. Chem., 1994, 98, 2078.
4 F. F. Fenter, V. Catoire, R. Lesclaux and P. D. Lightfoot, J. Phys.
Chem., 1993, 97, 3530.
5 A. A. Boyd, P. M. Flaud, N. Daugey and R. Lesclaux, J. Phys.
Chem. A, 2003, 107, 818.
NO2 þ hnðl ꢆ 410 nmÞ ! NO þ O
ð4Þ
6 P. Dagaut, T. J. Wallington and M. J. Kurylo, J. Phys. Chem.,
1988, 92, 3836.
7 F. C. Cattell, J. Cavanagh, R. A. Cox and M. E. Jenkin, J. Chem.
Soc., Faraday Trans. 2, 1986, 82, 1999.
leading to ozone formation (nearly a ppb in these simulations
at 298 K) via reaction (5).
8 S. P. Sander, R. R. Friedl, A. R. Ravishankara, D. M. Golden, C.
Kolb, R. E. Huie, L. T. Molina, G. K. Moortgat and B. J.
Finlayson-Pitts, Chemical Kinetics and Photochemical Data for
use in Stratospheric Modeling: Evaluation 14, The Jet Propulsion
Laboratory, Pasadena, CA, 2003.
9 M. J. Elrod, D. L. Ranschaert and N. J. Schneider, Int. J. Chem.
Kinet., 2001, 33, 363.
10 A. Bacak, M. W. Bardwell, M. T. Raventos, C. J. Percival, P. D.
Hammer and D. E. Shallcross, Chem. Phys. Lett., 2006, 419,
125.
3
Oð PÞ þ O2 þ M ! O3 þ M
ð5Þ
Therefore, changes to the kinetic parameters of reaction (6)
will have a small but significant impact on species at low NOx
levels in the colder regions of the troposphere.
Conclusions
11 M. W. Bardwell, A. Bacak, M. T. Raventos, C. J. Percival, G.
Sanchez-Reyna and D. E. Shallcross, Phys. Chem. Chem. Phys.,
2003, 5, 2381.
The rate coefficient and the branching ratios of the reaction of
C2H5O2 with HO2 have been determined using TF-CIMS. The
experimental results represent an extension in the range of
temperatures over which reaction (6) has been studied over the
temperatures range 195 and 298 K and at pressures ranging
from 75 to 200 Torr. In agreement with previous studies (e.g.
Maricq and Szente3), the results show that, within error, the
rate coefficient is independent of the pressure. The TF-CIMS is
capable of detecting peroxy radicals selectively for the first
time experimentally thus avoiding the problem of overlapping
spectra and reducing the experimental errors associated with
rate coefficient determination. Moreover, in comparison with
previous experimental studies, the initial peroxy radicals con-
centrations used in this work were at least one order of
magnitude lower reducing the amount of secondary reactions
and reducing further the associated experimental errors. In
general, the rate coefficients determined in this study are lower
than those recommended by the most recent JPL evaluation.8
However, the negative activation energy is in good agreement
with the Ea/R values reported by Dagaut et al.,6 and Maricq
and Szente.3 This study represents the first real time quantifi-
cation of the branching ratios for reactive channels (6b) and
(6c) and the experimental data reveal that channel (6a) is the
dominant one, which is in agreement with the 93% yield
reported by recent product studies.21
12 P. Borrell, C. J. Cobos and K. Luther, J. Phys. Chem., 1988, 92,
4377.
13 C. E. Canosa-Mas, M. D. King, R. Lopez, C. J. Percival, R. P.
Wayne, D. E. Shallcross, J. A. Pyle and V. Daele, J. Chem. Soc.,
Faraday Trans., 1996, 92, 2211.
14 R. Atkinson, D. L. Baulch, R. A. Cox, J. N. Crowley, R. F.
Hampson, R. G. Hynes, M. E. Jenkin, M. J. Rossi and J. Troe,
Atmos. Chem. Phys. D, 2005, 5, 6295.
15 D. Hartmann, J. Karthauser, J. P. Sawerysyn and R. Zellner, Ber.
Buns.-Ges. Phys. Chem., 1990, 94, 639.
16 D. L. Baulch, C. J. Cobos, R. A. Cox, C. Esser, P. Frank, T. Just,
J. A. Kerr, M. J. Pilling, J. Troe, R. W. Walker and J. Warnatz, J.
Phys. Chem. Ref. Data, 1992, 21, 411.
17 J. Heicklen, Adv. Photochem., 1988, 14, 177.
18 J. V. Seeley, J. T. Jayne and M. J. Molina, J. Phys. Chem., 1996,
100, 4019.
19 G. S. Tyndall, R. A. Cox, C. Granier, R. Lesclaux, G. K.
Moortgat, M. J. Pilling, A. R. Ravishankara and T. J. Wallington,
J. Geophys. Res.: Atmos., 2001, 106, 12157.
20 T. J. Wallington and S. M. Japar, Chem. Phys. Lett., 1990, 166,
495.
21 A. S. Hasson, G. S. Tyndall and J. J. Orlando, J. Phys. Chem. A,
2004, 108, 5979.
22 M. Spittler, I. Barnes, K. H. Becker and T. J. Wallington, Chem.
Phys. Lett., 2000, 321, 57.
23 T. J. Wallington, C. A. Gierczak, J. C. Ball and S. M. Japar, Int. J.
Chem. Kinet., 1989, 21, 1077.
24 D. E. Shallcross, M. T. Raventos-Duran, M. W. Bardwell, A.
Bacak, Z. Sloman and C. J. Percival, Atmos. Environ., 2005, 39,
763.
25 Y. Ikezoe, S. Matsuoka, T. Takebe and A. Viggiano, Gas Phase
Ion-Molecule Reaction Rate Constants Through 1986, Ion Reaction
Research Group of The Mass Spectroscopy Society of Japan,
Tokyo, Japan, 1987.
26 A. S. Hasson, K. T. Kuwata, M. C. Arroyo and E. B. Petersen, J.
Photochem. Photobiol., A, 2005, 176, 218.
27 H. Hou and B. S. Wang, J. Phys. Chem. A, 2005, 109, 451.
28 M. W. Bardwell, A. Bacak, M. T. Raventos-Duran, C. J. Percival,
G. Sanchez-Reyna and D. E. Shallcross, Int. J. Chem. Kinet., 2005,
37, 253.
29 P. D. Lightfoot, R. A. Cox, J. N. Crowley, M. Destriau, G. D.
Hayman, M. E. Jenkin, G. K. Moortgat and F. Zabel, Atmos.
Environ., 1992, 26A, 1805.
Sensitivity studies, using a tropospheric trajectory model,
were performed to assess the impact of the experimental errors
associated with k6 from this study on tropospheric O3 produc-
tion. It was found that at low NOx concentrations and at
205 K, a lower k6 increases the amount of C2H5O2 and HO2
(up to a ppt) and lowers the concentration of C2H5OOH
(down by 25 ppt). Meanwhile, at 298 K and low NOx levels
the recommended rate coefficient presented in this work would
lead to an increase in the O3 concentration of up to a ppb in
comparison with the value recommended by Sander et al.,8
This represents the impact of just one reaction between a
peroxy radical and HO2, further studies are warranted at low
temperature to characterise this interaction for a range of RO2
species.
30 D. M. Rowley, R. Lesclaux, P. D. Lightfoot, K. Hugues, M. D.
Hurley, S. Rudy and T. J. Wallington, J. Phys. Chem., 1992, 96,
7043.
31 M. E. Jenkin and G. D. Hayman, J. Chem. Soc., Faraday Trans.,
1995, 91, 1911.
ꢃc
This journal is the Owner Societies 2007
Phys. Chem. Chem. Phys., 2007, 9, 4338–4348 | 4347

32 A. A. Boyd, R. Lesclaux, M. E. Jenkin and T. J. Wallington, J.
Phys. Chem., 1996, 100, 6594.
36 A. A. Boyd, B. Noziere and R. Lesclaux, J. Chem. Soc., Faraday
Trans., 1996, 92, 201.
33 B. Noziere, R. Lesclaux, M. D. Hurley, M. A. Dearth and T. J.
Wallington, J. Phys. Chem., 1994, 98, 2864.
37 M. D. King, C. E. Canosa-Mas and R. P. Wayne, Atmos. Environ.,
2001, 35, 2081.
34 T. J. Wallington, O. J. Nielsen and J. Sehested, Reaction
of organic peroxy radicals in the gas phase, Wiley, Chischester,
1997.
35 G. Le Brass, Chemical Process in Atmospheric Oxidation, Springer,
Heidelberg, 1997.
38 D. Johnson, D. W. Price and G. Marston, Atmos. Environ., 2004,
38, 1447.
39 N. M. Donahue and R. G. Prinn, J. Geophys. Res., 1990, 95, 18387.
40 D. J. Fish, D. E. Shallcross and R. L. Jones, Atmos. Environ., 1999,
33, 687.
ꢃc
This journal is the Owner Societies 2007
4348 | Phys. Chem. Chem. Phys., 2007, 9, 4338–4348