Mechanisms for the formation of secondary organic aerosol components from the gas-phase ozonolysis of α-pinene

Article abstract of DOI:10.1039/b803283a

Gas-phase ozonolysis of α-pinene was studied in static chamber experiments under 'OH-free' conditions. A range of multifunctional products-in particular low-volatility carboxylic acids-were identified in the condensed phase using gas chromatography coupled to mass spectrometry after derivatisation. The dependence of product yields on reaction conditions (humidity, choice of OH radical scavengers, added Criegee intermediate scavengers, NO₂etc.) was investigated to probe the mechanisms of formation of these products; additional information was obtained by studying the ozonolysis of an enal and an enone derived from α-pinene. On the basis of experimental findings, previously suggested mechanisms were evaluated and detailed gas-phase mechanisms were developed to explain the observed product formation. Atmospheric implications of this work are discussed. the Owner Societies.

Full text of DOI:10.1039/b803283a

View Article Online / Journal Homepage / Table of Contents for this issue
PAPER
www.rsc.org/pccp | Physical Chemistry Chemical Physics
Mechanisms for the formation of secondary organic aerosol components
from the gas-phase ozonolysis of a-pinenew
Yan Ma, Andrew T. Russell and George Marston*
Received 26th February 2008, Accepted 24th April 2008
First published as an Advance Article on the web 6th June 2008
DOI: 10.1039/b803283a
Gas-phase ozonolysis of a-pinene was studied in static chamber experiments under ‘OH-free’
conditions. A range of multifunctional products—in particular low-volatility carboxylic
acids—were identified in the condensed phase using gas chromatography coupled to mass
spectrometry after derivatisation. The dependence of product yields on reaction conditions
(humidity, choice of OH radical scavengers, added Criegee intermediate scavengers, NO2 etc.) was
investigated to probe the mechanisms of formation of these products; additional information was
obtained by studying the ozonolysis of an enal and an enone derived from a-pinene. On the basis
of experimental findings, previously suggested mechanisms were evaluated and detailed gas-phase
mechanisms were developed to explain the observed product formation. Atmospheric implications
of this work are discussed.
spheric aerosol samples, supporting the role of multifunctional
acids in ambient aerosol formation.20–24
1. Introduction
Monoterpenes are an important class of biogenic volatile
ozonolysis follows the ‘Criegee mechanism’.25 The cycloaddi-
organic compounds (VOCs) emitted into the troposphere,
It is generally accepted that the initial step of a-pinene
with an estimated emission rate comparable to that of the
total anthropogenic VOCs.1 The atmospheric oxidation of
tion of ozone to the a-pinene double bond leads initially to an
energy-rich primary ozonide (POZ), which promptly decom-
these compounds is known to be a significant source of
poses to two excited carbonyl substituted Criegee intermedi-
secondary organic aerosol (SOA), which is currently of intense
ates (CIs), CI1 and CI2, as shown in Scheme 1.
research interest due to its potential implications for human
Possible fates of CIs have been the subject of numerous
health and climate change.1–6 While the oxidation processes
studies, mainly on simple alkenes.26,27 The most important
can be initiated by OH and NO3 radicals and O3, experimental
pathways are suggested to be stabilisation by collision and
evidence indicates that for a number of monoterpenes the
decomposition via the hydroperoxide channel and/or ester
reaction with O3 is the most efficient in generating low-
channel, as illustrated in Scheme 2.
volatility products that can be incorporated into SOA.7,8 In
The hydroperoxide channel is believed to be the dominant
order to better understand and predict atmospheric SOA
path for syn- and di-substituted CIs (with allylic hydrogens syn
formation, the detailed mechanisms of monoterpene ozonoly-
to the terminal oxygen); this process gives an OH radical and a
co-radical which can further react and decompose.28–30 Recent
sis, especially those leading to the condensable products, need
to be understood.
time-resolved experiments suggest that this channel can occur
The most naturally abundant monoterpene in the tropo-
with both excited and stabilised CIs, albeit on different time-
sphere is a-pinene;9 hence its reaction with ozone has received
much attention in recent years. Studies on the composition of
SOA from the reaction have revealed the formation of a
variety of multifunctional products, among which, compounds
with acid functionalities are observed in large measure.10–18
Due to their low volatility, the higher molecular weight
carboxylic acids—dicarboxylic acids, in particular—have been
suggested as key species in gas-to-particle conversion pro-
cesses.10,12,19 Furthermore, field studies have shown that sev-
eral organic acids observed from a-pinene ozonolysis,
including pinic acid, pinonic acid, norpinic acid, norpinonic
acid and its isomers, are important components of atmo-
scales.30,31 The stabilised CIs (SCIs) are also thought to react
with other species, often referred to as SCI scavengers, includ-
ing water, alcohols, acids, carbonyls, etc.32–35 In the light of
these studies and to avoid confusion, in this article SCI refers
to the Criegee intermediate which is stabilised and capable of
undergoing bimolecular reactions with other molecules.
Department of Chemistry, Reading University, Whiteknights, PO Box
224, Reading, UK RG6 6AD. E-mail: g.marston@rdg.ac.uk;
Fax: +44 118 3788703; Tel: +44 118 3786343
w Electronic supplementary information (ESI) available: Mass
spectral data of the products identified from the gas-phase ozonolysis
of a-pinene. See DOI: 10.1039/b803283a
Scheme 1
ꢀc
This journal is the Owner Societies 2008
4294 | Phys. Chem. Chem. Phys., 2008, 10, 4294–4312

View Article Online
acid.36,37 Here, detailed information on the full range of acidic
products observed in the ozonolyses, combined with results of
other experiments, is reported.
2. Experimental
2.1 Chemicals
Dry synthetic air (comprising nitrogen and oxygen, certified
99.995% pure) was supplied by BOC, and nitrogen dioxide
(purity 4 99.9%) was supplied by Sigma-Aldrich. All other
reagents employed were of analytical grade and were pur-
chased from recognised suppliers except pinonaldehyde, the
enal and the enone, which were synthesised from the ozono-
lysis of a-pinene in the condensed phase. Gaseous reagents
were used without further processing; all liquid reagents
employed in ozonolysis reactions were degassed by repeated
freeze–pump–thaw cycles prior to use.
Scheme 2 (a) Stabilisation by collision, (b) hydroperoxide channel,
(c) ester channel.
The detailed gas-phase processes that follow the initial
ozone attack and lead to the formation of SOA products,
such as organic acids, are, however, much less well under-
stood. While there have been several suggested mechanisms
for the formation of such products in a-pinene ozonolysis,
little experimental evidence has been provided to validate these
mechanisms. The aim of the present study was therefore to use
appropriate experimental design to test and evaluate the
previously postulated mechanisms and develop probable for-
mation pathways to the observed acid products on the basis of
experimental findings. A series of controlled laboratory ex-
periments were performed to examine the dependence of
product yields on various reaction conditions which in turn
gave indications of the mechanism of formation of individual
products. Such studies also allowed a better understanding of
what factors mainly control the SOA product yields. In
addition, ozonolysis experiments were carried out using an
enal (A) and an enone (B) which are ‘cleaner’ sources of either
one or the other of the two CIs formed in a-pinene ozonolysis.
As indicated in Scheme 3, the enal only gives rise to the
disubstituted CI2 when ozonised, while the enone only gives
the monosubstituted CI1. An investigation of the product
formation from the ozonolysis of these two compounds com-
pared with that from a-pinene ozonolysis then allowed us to
identify the early mechanistic steps in the formation of the
observed products. We have previously reported some results
obtained from the enal and the enone experiments with focus
on the formation of pinic acid, pinonic acid and norpinonic
2.2 Experimental set-up and procedure
The experimental apparatus employed for this study comprises
an 80 L collapsible Teflon chamber with attached gas chro-
matography (GC) coupled to mass spectrometry (MS) and
flame ionisation detector (FID). Ozonolysis experiments were
carried out at atmospheric pressure (760 Æ 10 Torr, synthetic
air) at a temperature of 295 Æ 4 K in the Teflon chamber.
Concentrations of the unsaturated compounds were on the
order of 15–20 ppmv, with ozone concentrations being typi-
cally 5 ppmv lower in order to minimise secondary ozone
reactions. Experiments were carried out in the presence of
sufficient concentrations of OH scavengers (cyclohexane and/
or methanol) to scavenge 495% of any OH radicals formed in
the reactions. Typical concentrations of cyclohexane and
methanol used were 2250–3000 ppmv and 16 500–22 000
ppmv, respectively.
Prior to each experiment, the Teflon chamber was cleaned
and flushed with plenty of dry cylinder air (o0.01% RH), and
the remaining hydrocarbon level was checked by drawing air
samples from the chamber for analysis with GC/FID until the
background alkene concentration was indiscernible. A mixture
of the alkene under study and the relevant OH scavenger and
SCI scavenger, when used, was prepared in the Teflon cham-
ber using synthetic air as the diluent gas. The air used could
either be admitted dry, or wet by passing through a series of
bubblers containing deionised water; by varying the fraction
of dry to wet air it was possible to control the relative humidity
of the reaction mixture. It should be noted that water vapour
could permeate though Teflon, thus a low concentration of
water may be present in experiments reported as having been
performed at 0% RH; however, as we cleaned the chamber
with large quantities of dry synthetic air and made the reactant
mixture shortly before the reaction, the water content in the
chamber is expected to be insignificant. A known concentra-
tion (ca. 20 ppmv) of methane—which is relatively inert to
reactions within the ozonolysis system and also was not
formed in any significant yields (below detection limit of B1
ppmv)—was used as an internal standard to measure the
chamber volume. Hydrocarbon concentrations were deter-
mined using online GC/FID before and after each reaction.
Scheme 3
ꢀc
This journal is the Owner Societies 2008
Phys. Chem. Chem. Phys., 2008, 10, 4294–4312 | 4295

View Article Online
Ozone was generated using a Fischer (Model 502) ozoniser,
and was trapped onto silica gel at 225 K, before being desorbed
into a Pyrex bulb. Its purity was determined by measuring its
ultraviolet absorption at 254 nm. A sample of ozone at the
the collected samples were analysed with the same procedure as
for products. The wall-loss rate thus determined for pinonalde-
hyde was B25–35% hÀ1. It was assumed that this wall-loss rate
applied to all low-volatility products. However, on the timescale
that the experiments were performed (typical reaction and
sampling time being several minutes), the wall losses were
estimated to be insignificant compared to statistical measure-
ment uncertainties (estimated to be B10–20% on the basis of
replicate experiments), and thus the calculated product yields
were not corrected for wall losses.
desired pressure was prepared in
a
0.5
L
boro-
silicate bulb and added to the chamber last by flowing synthetic
air from a vacuum line through the bulb. The reaction was
allowed to occur long enough that it was simulated to be more
than 99% complete, but not so long that subsequent slower
heterogeneous processes could significantly affect the results.
Under the experimental conditions employed, it is assumed that
the condensable products we observed were formed as a result
of gas-phase reactions. To test this assumption, a series of
diagnostic experiments on the ozone/cyclohexene and ozone/
a-pinene systems have been carried out before the main study
and will be described in section 3.2. A known volume of
reactant/product mixtures was then withdrawn from the cham-
ber through adsorption onto a PTFE membrane filter (Schlei-
cher and Schuell TE 36, 0.45 mm pore size) by the vacuum. To
minimise the sample-to-sample bias, in all experiments carried
out with similar reagent concentrations, filter samples were
collected at a constant reaction time. The filter sample was
immediately treated with 14% BF3/methanol at 60 1C for 30
min to generate the methyl esters of the acids, which were
extracted with hexane and then separated using GC (Thermo-
Finnigan, Trace GC) and detected using quadrupole MS
(ThermoFinnigan, Trace MS) in EI (70 eV) or CI(À) mode
(methane). Note that the studies of Hoffmann and co-
workers15,38 suggested that acid dimers may be formed in the
ozonolysis of a-pinene; nevertheless, since these compounds are
only held together by hydrogen bonds, they are expected to
have dissociated during the workup procedure and are there-
fore detected as derivatives of the corresponding monomers.
To investigate the potential interferences and contributions
from background, a blank filter sample was obtained before
each reaction. In these experiments, a sample of the reactant
mixture (without the addition of ozone) was collected onto a
PTFE filter and this filter sample was then treated and
analysed with the same procedure as for reaction products.
Generally very low carryover (below detection limit S/N = 3)
was found from previous experiments.
Secondly, while this study was aimed at obtaining the total
acid yields including the particulate and gaseous phase via
condensation/adsorption, negative sampling artefacts may oc-
cur due to evaporation or desorptive loss of products. For the
higher molecular weight organic acids investigated, the sam-
pling efficiency was estimated as 85–98% under typical experi-
mental conditions by collecting the reaction mixture with two
filters in line and calculating the ratio of front to back-up filter
concentrations. Since the scatter in product measurements was
of the same order as the potential sampling errors, measured
yield for these relatively low-volatility acid products could be
taken as an acceptable approximation of the total yield.
For the acidic products studied, measured yield values from
five replicate experiments showed general agreement within
Æ10–15% of the average value (see Table 2). This value may
be taken as an estimate of statistical uncertainties associated
with experimental acid yield measurements. Measurements for
carbonyl products showed more scatter (one standard devia-
tion B15–20%, see Table 2), and thus the relative uncertainty
in quantification of these compounds can be assumed to be
higher than for acids.
Based on the above measurements, the absolute uncertainties
derived from wall losses and sample collection were estimated to
be r10% and r15%, respectively, under typical experimental
conditions. Other sources that contribute to the uncertainty in
measured acid yields include mainly uncertainties for quantifica-
tion of ozone reacted (BÆ5%), sample volume determination
(BÆ5%), sample derivatisation and extraction (BÆ15%), GC/
MS analysis and product peak quantification (BÆ5%), and
calibration uncertainties (BÆ10% for compounds with stan-
dards available and BÆ20–40% for compounds quantified
using surrogate calibration). After considering all these contri-
buting factors, the absolute uncertainty for experimentally
measured acid yields was calculated as the square root of the
sum of squares for individual uncertainties, which is approxi-
mated at BÆ27% for compounds quantified with standards
and BÆ32–47% for compounds that do not have authentic
standards. Nevertheless, since the strategy of this study was to
determine how the product yields vary with reaction conditions,
the absolute errors associated with individual product measure-
ments should not significantly affect the results discussed below.
In order to confirm that the observed methyl esters were
generated in the derivatisation procedure from the corres-
ponding acids but not directly in the ozonolysis reaction,
direct injection of the sampled products was performed for
comparison. The procedure was as follows: the contents of the
filter were extracted by sonication with dichloromethane after
collection, and the solution was then concentrated to 1 mL
under a dry gentle stream of nitrogen and 1 mL of this was
injected directly into the GC/MS. The instrument method used
was the same as for the derivatised samples.
There are a number of potential sources of error associated
with product yield measurements. First, the wall losses of semi-
and low-volatility products cannot be neglected in our reaction
chamber which has a relatively high surface-to-volume ratio.
For an estimation of the influence of wall losses, a synthesised
pinonaldehyde sample was evaporated into the chamber and the
bag was then made up to full capacity. The chamber mixture
was then sampled for three times with an interval of 30 min and
3. Results and discussion
3.1 Product identification and quantification
Eight organic acids and two dicarbonyl compounds were identi-
fied in the collected a-pinene ozonolysis products as their
corresponding methyl ester and/or acetal derivatives. These
ꢀc
This journal is the Owner Societies 2008
4296 | Phys. Chem. Chem. Phys., 2008, 10, 4294–4312

View Article Online
compounds were cis-pinic acid, cis-pinonic acid, norpinic acid,
norpinonic acid, pinalic-3-acid, norpinalic acid, 10-OH-pinonic
acid, 4-OH-pinalic-3-acid, pinonaldehyde and norpinonalde-
hyde. Among them, cis-pinic acid, cis-pinonic acid and pinonal-
dehyde were positively identified by comparison of GC retention
times and mass spectra with those of the derivatised authentic
standards (purchased or synthesised); other products were iden-
tified using their respective GC/EI and CI(–) mass spectra
combined with retention time information. Names (based on
the nomenclature for terpene oxidation products by Larsen
et al.39), molecular structures, and the method of identification
of these compounds are indicated in Table 1. Mass spectral data
of the identified products are given in the ESIw together with a
detailed discussion of structure elucidation. By comparison with
the blank filter results, these compounds could be attributed to
products derived from ozonolysis reactions. None of the ob-
served methyl esters or acetals was found in the directly injected
sample, and thus it can be confirmed that they were not formed
directly in the ozonolysis system but during the derivatisation
procedure from their corresponding acids or carbonyls. Further-
more, in order to examine if the observed product formation
originated from reactions of unscavenged OH radicals, some
experiments were performed with elevated cyclohexane concen-
tration to scavenge Z99% of OH formed. The composition of
the collected products showed no marked difference, indicating
that the products identified were formed from the a-pinene/O3
reaction and not from secondary OH reactions.
Table 1 Products identified from the gas-phase ozonolysis of a-
pinene
Identification
Name
Structure
Positive identification by
comparison with authentic
standards
Pinonic acid
Pinic acid
Pinonaldehyde
Identification by analogy
with homologous reference
compounds and by matching
the spectra reported in
literature
Norpinonic acid
Norpinic acid
Pinonic acid, pinic acid and pinonaldehyde have been
previously identified as major aerosol components during a-
pinene ozonolysis (e.g. ref. 10–15). Norpinonic acid, norpinic
acid, norpinonaldehyde, pinalic-3-acid, norpinalic acid and
10-OH-pinonic acid have also been tentatively identified in a
number of product studies using various analytical techni-
ques.11–18 Tentative identification of 4-OH-pinalic-3-acid was
reported by Jaoui and Kamens40 in one study of b-pinene
ozonolysis, while its presence is proposed here for the first time
in a-pinene ozonolysis.
Quantification of pinic acid, pinonic acid and pinonaldehyde
was performed using the calibration of relevant standards by
GC/EI-MS. For other products that have no standards avail-
able, a semi-quantification method was used: pinic acid was used
as a surrogate for norpinic acid and pinonaldehyde was used for
norpinonaldehyde; for all other acids identified, the calibration
of pinonic acid was used. The determined or estimated molar
yields of the above-identified products are presented in Table 2,
along with the results from several previous studies.
Identification by analogy
with homologous compounds
Norpinonaldehyde
Norpinalic acid
Pinalic-3-acid
Identification without
reference data
10-OH-Pinonic
acid
The reported product yields in the literature showed a fair
degree of scatter, which might be a result of the different
reaction conditions such as reagent concentrations, reaction
times, and the sampling and analytical methods. At the
relatively high concentration levels used in our study, the
majority of the acid products investigated—which have rela-
tively low volatility—can be expected to have transferred to
the adsorbent filter by condensation and/or adsorption. As
discussed earlier, although some desorptive loss may occur
during sampling leading to less than complete recovery,
measurements have indicated that this is not very significant
compared to experimental uncertainties for the determined
acids. It can be seen in the table that our measured yields of
4-OH-Pinalic-3-
acid
pinic acid and pinonic acid are well within the range of
previous results. Also the estimated yields for norpinic acid,
norpinonic acid and OH-pinonic acid are in reasonable agree-
ment with other studies. Our yield of pinalic-3-acid appears
ꢀc
This journal is the Owner Societies 2008
Phys. Chem. Chem. Phys., 2008, 10, 4294–4312 | 4297

View Article Online
Table 2 Comparison of molar yields (in %) of reaction products from the gas-phase ozonolysis of a-pinene
Yu et al.41
Christoffersen et al.10
Glasius et al.14
Winterhalter et al.18
Koch et al.12
This workg
OH scavenger
Seed aerosol
[a-Pinene]0/ppmv
[O3]0/ppmv
Relative humidity
2-Butanol
(NH4)2SO4
0.06–0.1
0.074–0.269
NM
Cyclohexane
—
0.1(L),1(H)
0.08(L),0.8(H)
Dry or 13%
Cyclohexane
—
1.9
0.82
40–55%
—
—
Cyclohexane
—
4.2
1.7
NM
Cyclohexane
—
15–20
10–15
Dry
0.1
0.13–0.14
Dry or 45%
Pinic acid
3.0–6.0a
0.3–0.5 (L)
1–3 (H)b
0.06 (L)
1.4b
1.55 Æ 0.28b
0.71 Æ 0.24b
0.06 Æ 0.02b
0.37 Æ 0.13bc
0.37 Æ 0.13bc
3.2b
2.85 Æ 0.31
1.86 Æ 0.31
0.14 Æ 0.04
0.09 Æ 0.03
2.33 Æ 0.33
1.8–3.9b
Pinonic acid
Norpinic acid
Norpinonic acid
Pinalic-3-acid
2.2–7.9a
1.5b
1.2b
1.3–1.7b
0.2–0.4 (H)b
0.09–0.1a
0.05–0.09b
4.3–12.6ac
2.1–4.8bc
4.3–12.6ac
0.04b
0.19bc
0.19bc
bf
bf
bf
2.1–4.8bc
e
Norpinalic acid
OH-Pinonic acid
0.18 Æ 0.04
0.27 Æ 0.07
1.5–3.7ad
1.3–2.1b
0.86bd
1.9b
0.51 Æ 0.18b
0.29 Æ 0.09b
OH-Pinalic acid
Pinonaldehyde
0.10 Æ 0.03
1.00 Æ 0.30
6–19a
11–26a
0.3–0.9b
1.2–2.6a
0.1–0.2b
Norpinonaldehyde
0.10 Æ 0.04
a
b
Total yield (gas + aerosol). In aerosol samples. Sum of norpinonic acid, pinalic-3-acid isomers. Sum of possible isomers. Tentatively
c
d
e
f
g
identified but yields not reported. Tentatively identified, sum of yields o2%. Uncertainties listed are 2 standard deviations of the calculated
yields for replicate experiments. NM: not mentioned. L: low concentration experiments; H: high concentration experiments.
much higher than that reported by Glasius et al.14 and
Winterhalter et al.18 as the sum of norpinonic acid isomers,
but is of the same order as the result of Yu et al.41
potential interferences from secondary heterogeneous pro-
cesses. A series of diagnostic experiments on adipic acid
(HO(O)C(CH2)4C(O)OH) and glutaric acid (HO(O)C(CH2)3-
C(O)OH) formation in the ozone-cyclohexene reaction were
carried out before the main study; in these preliminary studies,
we investigated the time dependence of product yields, mea-
sured products on the walls of glass vessels and varied the
surface-to-volume ratio of the reactor (22.4–60.9 mÀ1). The
measured yields for both acids were found to be dependent on
reaction time, indicating that in addition to a ‘prompt’ for-
mation pathway, some slower secondary processes also exist
that can contribute to acid production. It was also noted that
both wall losses and the total acid yields (filter products +
bulb wall products) increased with increasing surface-to-vo-
lume ratio of the glass bulb; the latter would imply that the
reactions leading to these acids took place, at least partially, at
the bulb wall as a result of heterogeneous chemistry beyond
the gas-phase ozonolysis. Nevertheless, experiments carried
out in the 80 L Teflon chamber—which has a relatively small
surface-to-volume ratio (B8 mÀ1) and an unreactive Teflon
surface—showed that the measured acid yields in such a
system by filter collection are close to the extrapolation of
bulb experiment yield plots (as sum of filter and bulb wall
products) to zero surface-to-volume ratio. This would indicate
that the interferences from wall losses and heterogeneous wall
reactions were not significant in the chamber system under the
experimental conditions employed.
Quantitative investigation of carbonyl products was not the
main focus of this study; nevertheless, for the purpose of
comparing with acid products and providing additional me-
chanistic information, the collected pinonaldehyde and norpi-
nonaldehyde in filter samples were also determined. These
compounds have previously been identified as both gas- and
aerosol-phase products, but predominantly existing in the gas
phase as expected from their relatively high vapour pressures.
As a result it is likely that only a small portion of these
carbonyl compounds have been collected by filters. Indeed,
our measured yields of pinonaldehyde and norpinonaldehyde
were found to be in reasonable agreement with the reported
aerosol-phase yields of Yu et al.,41 indicating that the present
experimental setup was mainly recovering the particle-bound
fraction of these carbonyl products. Nevertheless, the general
data trends obtained from average filter measurements can be
used to provide mechanistic information regarding the reac-
tion routes to these products.
For clarity, in the following, results are first described for
the investigation of the formation mechanisms of the four
major condensed products, pinic acid, pinalic-3-acid, pinonic
acid and pinonaldehyde. The experimental observations and
possible formation mechanisms for other identified products
are discussed in the section after.
Experiments were then carried out in the chamber system
for a-pinene ozonolysis to further examine if the observed
products were formed as a result of gas-phase chemistry, or if
they were involved in secondary formation processes. In this
respect, it is interesting to note that the previous study by
3.2 Effect of secondary heterogeneous chemistry
This study focused on gas-phase chemistry, and thus relatively
short reaction and sampling times were used to limit the
ꢀc
This journal is the Owner Societies 2008
4298 | Phys. Chem. Chem. Phys., 2008, 10, 4294–4312

View Article Online
Christoffersen et al.10 indicated that pinic and pinonic acid
formation is highly dependent on concentrations of reactants,
with measured yields reduced by ca. a factor of 5 in experi-
ments where initial reagent concentrations were lowered by an
order of magnitude. The authors suggested that the possible
explanation for this observation could be that the sampling
efficiency is lower for smaller aerosol particles generated at
lower concentration levels, or that the chemistry at elevated
concentrations differs somewhat from that in low concentra-
tion systems; e.g., if these compounds were also formed in
secondary heterogeneous reactions, their yields can be ex-
pected to increase in higher concentration reactions with
higher aerosol surface areas available. Indeed, several recent
studies have pointed out that secondary reactions could occur
in the condensed phase following the gas-phase ozonolysis
reactions; e.g., the hydroperoxides generated from SCI reac-
tions or RO2–HO2 reactions may subsequently react with
aldehydes leading to peroxyhemiacetals.34,35,41 Such products
are thermally labile and may convert back to the parent
compounds or potentially decompose to alcohol and acid
compounds during the analysis processes.
pinonaldehyde plot showed that the yield decreased slightly
with decreasing initial ozone concentration, which could sug-
gest that some secondary chemistry might contribute to pino-
naldehyde formation. However, considering its much higher
volatility compared to acidic products, an alternative explana-
tion is that the sampling recovery for pinonaldehyde dropped
at lower concentrations due to the smaller amount of
partitioning into SOA mass.
To further address the implications of heterogeneous chem-
istry in the reaction system, product time evolution following
the ozonolysis was studied. After the introduction of a-pinene
(B20 ppmv) and ozone (B15 ppmv), the reaction mixture was
allowed to age in the chamber and three samples were taken
after different periods of time, i.e. 6, 30 and 60 min. As some
products would have been lost to the chamber walls, a loss rate
of 35% hÀ1 was taken into account when calculating product
yields here based on the measured wall losses of the synthe-
sised pinonaldehyde sample (see Experimental). For compar-
ison, another filter sample was collected immediately after the
first sample (at 10 min) and was sealed in a glass container and
left to age for 1 h before treatment and analysis. Illustrated in
Fig. 2 are the calculated yields for pinic acid, pinonic acid,
pinalic-3-acid and pinonaldehyde in these four samples.
Although errors associated with these experiments are ex-
pected to be large due to the uncertainty in determination of
wall losses, a trend was observed that during the ageing process
the yields of pinonic acid and pinic acid somewhat increased
whereas those of pinonaldehyde and pinalic-3-acid decreased.
Since the major gas-phase reaction should have been essentially
completed (499% of the ozone initially present has been
consumed) within 6 min under the experimental conditions,
the observed variation in product yields over longer times
suggests that slower secondary oxidation reactions may take
place in the chamber beyond the gas-phase ozonolysis. Pre-
sumably, the oxidation of pinalic-3-acid could result in pinic
acid, and pinonic acid can be generated from the oxidation of
pinonaldehyde. While secondary oxidation of first-generation
products via reactions with O3 or OH radicals is possible, in the
presence of excess a-pinene and cyclohexane scavenger, such
In order to minimise the influence of physical factors such as
sampling efficiency and wall losses but focus on the potential
role of heterogeneous reactions in the formation of individual
products, experiments were carried out with the initial a-
pinene concentration kept constant at ca. 25 ppmv and the
ozone concentration varied from ca. 3.5–20 ppmv (the reac-
tion and sampling times were maintained constant), and a
stack of two filters was used for sampling. Excess cyclohexane
was added as OH scavenger. Measured yields for four main
products as the sum of both filter samples in these experiments
are presented in Fig. 1.
Pinic acid, pinonic acid and pinalic-3-acid plots showed only
minor fluctuations which lie within the experimental errors;
i.e., the formation of these products was not significantly
affected by the initial concentration of ozone under the
experimental conditions. This would suggest that with the
current experimental setup, these acids were formed as direct
products from the gas-phase ozonolysis reaction and also were
not undergoing further reactions to any significant extent. The
Fig. 2 Time evolution of pinic acid, pinonic acid, pinonaldehyde and
pinalic-3-acid from a-pinene ozonolysis in the reaction chamber or on
adsorbent filter.
Fig. 1 Molar yields of pinic acid, pinonic acid, pinonaldehyde and
pinalic-3-acid from a-pinene ozonolysis as a function of initial ozone
concentration.
ꢀc
This journal is the Owner Societies 2008
Phys. Chem. Chem. Phys., 2008, 10, 4294–4312 | 4299

View Article Online
Scheme 4 Suggested mechanisms by Jenkin et al.19 and Koch et al.12 for pinic acid formation from a-pinene ozonolysis.
reactions should be minimal. Another possible mechanism is
via the particle-phase association reactions, such as reactions of
hydroperoxide species with aldehydic products as mentioned
above, which may transform aldehyde functionalities to acids
via decomposition of the resultant peroxyhemiacetals.35 In-
deed, in comparison with the chamber experiments, a more
pronounced change in product yields was observed in the
ageing experiment on the adsorbent filter, implying that the
potential secondary oxidation reactions are likely the most
significant in the condensed phase. Since the reaction and
sampling times in most chamber studies are on the order of a
few hours or longer, the effect of particle phase reactions in
these systems can be expected to be large, as reported by, for
example, Docherty et al.42 and Gao et al.;43 and this may also
explain some of the discrepancies in the measured product
yields among different studies as a result of the time-dependent
SOA evolution. Nevertheless, judged from the product forma-
tion behaviour presented above, the interferences from second-
ary heterogeneous chemistry should not be significant (within
statistical measurement uncertainties) in the short timescale
experiments (6 min) with our experimental setup and thus it
can be concluded that the observed formation of these pro-
ducts resulted predominantly from direct gas-phase reactions.
radical, a subsequent reaction of which with either RO2 or
HO2 radicals would then give the diacid.
The relevance and relative importance of these potential
reaction channels have yet to be clarified, but can be expected
to be sensitive to the ratio of [RO2] to [HO2] in the system, and
this ratio can be modified by the choice of OH scavenger used.
As illustrated in Scheme 5, the use of methanol scavenger gives
exclusively HO2 radicals. In contrast, when cyclohexane is
used as the OH scavenger, the major radical product is
cyclohexylperoxy radicals which will contribute to RO2 radical
population in the system. Depending on the specific reaction
conditions, some HO2 production may also occur from the
reaction of O2 with cyclohexyloxy radicals; however, this
amount is expected to be small.44,45
These additional radicals generated in the OH–scavenger
reactions can thus alter the peroxy radical population in the
ozonolysis system, not simply by adding to the radicals
produced in the ozonolysis reaction (e.g. RO2 radicals formed
from the reaction of O2 with the alkyl radicals co-generated
with OH), but as a result of the fast RO2–HO2 reactions,
which lead to shorter lifetimes of RO2 radicals and thus lower
[RO2] when [HO2] is elevated, and vice versa.45,46 Hence the
[RO2]/[HO2] ratios in the reaction system can be expected to
increase with increasing concentration ratios of cyclohexane to
methanol. It is noted that methanol may form a complex with
the HO2 radical and accelerate the HO2 self reaction;47,48
however, we do not expect this effect to alter the general
trends in the [RO2]/[HO2] ratio when the [cyclohexane]/
[methanol] ratio changes.
3.3 Effect of OH scavengers
Pinic acid has been the subject of a number of studies due to its
very low volatility and hence potential importance in SOA
formation. Two slightly different mechanisms have been pro-
posed by Jenkin et al.19 and Koch et al.12 to explain the
observed gas-phase formation of pinic acid. As illustrated in
Scheme 4, both authors suggested that the diacid is produced
via the unimolecular decomposition of the di-substituted CI2
through one of the possible hydroperoxide channels; the
resulting a-carbonyl alkyl radical adds O2 and leads, via a
series of peroxy radical reactions, to the formation of a C9
acylperoxy radical. This radical may react with RO2 or HO2
radicals under NOx-free conditions. Koch proposed that the
oxo peracid formed in the HO2 channel undergoes an intra-
molecular isomerisation to generate pinic acid; while Jenkin
suggested that, in addition to decomposition via loss of CO2,
the C9 acyl-oxy radical formed in the RO2 channel may
isomerise by a 1,7 H-shift leading to another acylperoxy
The effect of varying OH scavengers on pinic acid and other
products was examined in a series of experiments with similar
Scheme 5
ꢀc
This journal is the Owner Societies 2008
4300 | Phys. Chem. Chem. Phys., 2008, 10, 4294–4312

View Article Online
with the RO2–HO2 reactions, and thus would promote the
production of pinic acid. (Interesting, very recent results49
indicate that the yield of RO from the reaction of RO2 with
HO2 is non-zero for acyl peroxy radicals, although the yield is
lower than for RO2 + RO2 reaction). Qualitatively, the turn-
ing point of the yield plot may then correspond to the situation
under which the permutation reactions of the key intermediate
peroxy species start to compete favourably with their reactions
with HO2 radicals. Note that the electronic structure calcula-
tions by Jenkin et al.19 have indicated that the abstraction of
aldehydic hydrogen is fast enough to compete with the decom-
position of the above-mentioned C9 acyl-oxy radical which
provided support for their proposed mechanism. However, the
exact mechanism is still elusive as the experimental detection
of the key intermediate species, such as Criegee intermediates
and acylperoxy radical intermediates, is extremely difficult;
nevertheless, product studies with added NO2 and the enal and
enone experiments could provide a partial test of the proposed
mechanisms, as we shall discuss in later sections.
Fig. 3 Molar yields of pinic acid and pinalic-3-acid from a-pinene
ozonolysis as a function of [cyclohexane] : [methanol].
initial reagent concentrations (ca. 20 ppmv a-pinene and ca. 15
ppmv O3) but varying ratios of [cyclohexane]/[methanol]. The
sum of the concentration of OH scavengers was maintained
such that 495% of OH was scavenged in all experiments.
Both pinic acid and pinalic-3-acid were found to be sensitive
to the relative amounts of cyclohexane to methanol, as shown
in Fig. 3. For pinic acid, the measured yield was lowest in the
presence of pure methanol scavenger; as the concentration of
cyclohexane was increased relative to methanol, the yield value
rose rapidly at the beginning and then appeared to level off
somewhat. While one possible explanation for this effect
involves reactions of stabilised CIs as methanol may also act
as an SCI scavenger, our experiments (described in section 3.4)
have indicated that the pinic acid yield is not obviously
affected by SCI reactions. Therefore, the most likely explana-
tion lies in the variation in [RO2]/[HO2] ratios caused by the
different amounts of cyclohexane to methanol used. This
indicates that radical pathways play a role in the formation
of pinic acid, for example, via the mechanisms suggested by
Koch et al.12 or Jenkin et al.19 The observed increase in pinic
acid yield at elevated [RO2]/[HO2] further indicates that
increased concentrations of RO2 and/or decreased concentra-
tions of HO2 can promote its formation. Since the reaction
path proposed by Koch et al. necessitates one RO2–RO2
reaction and one RO2–HO2 reaction, the production of diacid
via this channel can be expected to be effectively inhibited at
high relative concentrations of cyclohexane where the domi-
nant fate for the majority of RO2 radicals (with the exception
of tertiary RO2 radicals which have particularly slow self-
reactions and reactions with other RO2 radicals) formed in the
system would be reactions with RO2 radicals.19 This however
disagrees with the experimentally observed pinic acid forma-
tion behaviour and would appear to suggest that the Koch
et al. mechanism makes at most a minor contribution to pinic
acid formation under our experimental conditions. A mechan-
ism which is more consistent with the above finding is that
proposed by Jenkin et al. Because the formation of diacid by
this route involves two permutation reactions of RO2 inter-
mediates with available RO2 radicals, the increase in [RO2]/
[HO2] would allow such reactions to compete more efficiently
Formation of pinalic-3-acid from a-pinene ozonolysis fol-
lows a generally similar trend to pinic acid. This behaviour is
qualitatively consistent with a previously suggested mechan-
ism via the RO2/HO2 reaction of the C9 acylperoxy radical
that also potentially leads to pinic acid.12,13,18,19 This pathway
to pinalic-3-acid involves only one RO2-dependent step, and
thus one would expect that pinalic-3-acid is less influenced by
OH scavengers compared to pinic acid, assuming the above
pinic acid formation mechanism suggested by Jenkin et al.
This indeed agrees with the experimental observation here.
Again, further tests are needed to evaluate such a mechanism
and will be addressed later.
In contrast, pinonic acid and pinonaldehyde showed no
clear dependence on the identity of the OH radical scavenger,
as illustrated in Fig. 4. A similar conclusion has been made by
Baker et al.,50 who showed that the measured pinonaldehyde
yield was insensitive to the OH scavenger type (cyclohexane or
2-butanol). Thus the mechanisms to form pinonic acid and
pinonaldehyde would not appear to involve the RO2 depen-
dent chemistry to any significant extent.
Fig. 4 Molar yields of pinonic acid and pinonaldehyde from a-pinene
ozonolysis as a function of [cyclohexane] : [methanol].
ꢀc
This journal is the Owner Societies 2008
Phys. Chem. Chem. Phys., 2008, 10, 4294–4312 | 4301

View Article Online
arrangement of CI1 via the ester channel, leading initially to a
vibrationally excited acid, a part of which may be deactivated
to the ground-state pinonic acid.53
To evaluate the suggested mechanisms and, in particular,
the role played by SCI in the a-pinene/O3 system, a series of
experiments were carried out at three different relative humid-
ity levels (dry conditions, 30% RH and 80% RH) and in the
presence or absence of acetic acid, a well-known SCI scaven-
ger. Either cyclohexane or methanol was used to scavenge
495% of the OH radicals formed. The results obtained for
pinonic acid, pinonaldehyde, pinic acid and pinalic-3-acid as a
function of relative humidity are presented in Fig. 5.
Scheme 6 Suggested mechanism for pinonic acid and pinonaldehyde
formation from a-pinene ozonolysis.
Different formation behaviour was observed for these pro-
ducts. In the absence of acetic acid, pinonic acid and pino-
naldehyde sampled showed a fairly strong dependence on RH,
while pinic acid and pinalic-3-acid showed a much weaker
dependence. With the use of cyclohexane scavenger, the yield
of pinonic acid was 1.86 Æ 0.31% under dry conditions and
3.35 Æ 0.40% at 80% RH, and the pinonaldehyde yield was
1.00 Æ 0.30% for dry conditions and 2.16 Æ 0.33% at 80%
RH. These observations appear to support the proposed
mechanism where the stabilised CI interacts with water vapour
to yield acids and carbonyls, probably via the decomposition
of a hydroxyalkylhydroperoxide intermediate. In line with our
finding, Warscheid and Hoffmann54 and Fick et al.55 also
observed a positive correlation between pinonaldehyde for-
mation and RH; however, Berndt et al.56 reported a lower
yield of gaseous pinonaldehyde at higher humidity, and Baker
et al.50 reported a pinonaldehyde yield independent of RH in
the range of B5–50%. The reasons for these discrepancies are
not clear, but could depend on the differences in experimental
setup and reaction conditions, such as initial concentrations
and reaction time. The relatively high reactant concentrations
used in our experiments may have caused more competitive
Several recent experimental studies have indicated that the
use of different OH scavengers can largely affect the SOA yield
from the ozonolysis reactions, and this observation has been
suggested to arise from the differing radical products formed
in the OH–scavenger reactions.42,45,46 Furthermore, Jenkin51
has recently reported a modelling simulation of the SOA
formation and composition during a- and b-pinene ozonolysis
with cyclohexane or 2-butanol as OH scavenger. For a-pinene,
the use of 2-butanol scavenger was simulated to have a slight
lowering effect on yields of SOA and some products including
pinic acid and pinalic-3-acid as a result of the elevated [HO2]/
[RO2]. Our experimental results are qualitatively consistent
with the modelling results of Jenkin and support the impact of
the OH scavenger on certain condensable products.
3.4 SCI scavenger experiments
Pinonic acid and pinonaldehyde have usually been assumed to
arise from the mono-substituted CI1 via its reaction with
water, as indicated in Scheme 6.18,19,52 An additional pathway
proposed for pinonic acid formation involves the direct re-
Fig. 5 Molar yields of (a) pinonic acid, (b) pinonaldehyde, (c) pinic acid and (d) pinalic-3-acid from a-pinene ozonolysis under various conditions.
ꢀc
This journal is the Owner Societies 2008
4302 | Phys. Chem. Chem. Phys., 2008, 10, 4294–4312

View Article Online
SCI scavenging in other processes, e.g. reactions with OH–
scavenger reaction products or certain ozonolysis products, or
possibly SCI self reactions. Few literature studies have focused
on the impact of water on pinonic acid formation. In contrast
to the observation here, Fick et al.55 reported a dramatic
reduction in pinonic acid yield at increased RH in the absence
of OH scavengers and the disappearance of pinonic acid when
an OH scavenger (TMB) was added.
a negative water dependence of the gas-phase HO2 concentra-
tion, whereas the interaction between RO2 radicals and water
has not been reported.61 As a consequence, the addition of
water vapour can be expected to reduce HO2 radical concen-
trations and accordingly increase the relative concentrations of
RO2 versus HO2, thereby leading to the enhanced formation of
pinic and pinalic acids according to their RO2 mediated
mechanisms. The more pronounced effect of RH in the
methanol scavenger system is indeed consistent with the find-
ing (see Fig. 3) that the increase in [RO2]/[HO2] has a more
significant effect on acid production at the lower end of
[cyclohexane]/[methanol].
Also noted is that the measured pinonic acid and pinonal-
dehyde yields under humid conditions were in all cases lower
when methanol scavenger was used compared with cyclohex-
ane. This may be accounted for by noting that methanol could
also act as an SCI scavenger; in this case, methanol would
compete with water for the SCI, thereby reducing the water-
enhanced formation of pinonic acid and pinonaldehyde ac-
cording to the above SCI–water reaction mechanism. In the
presence of acetic acid, the water enhanced formation of both
pinonic acid and pinonaldehyde was removed, which is con-
sistent with acetic acid competing more effectively for SCI
than water or methanol.57 These findings provide more evi-
dence for the water–SCI reaction as the source of the RH-
dependent formation of pinonic acid and pinonaldehyde.
However, it is also evident that a significant fraction of
pinonic acid and pinonaldehyde formed irrespective of the
presence or absence of water vapour or acetic acid, showing
that there are additional routes to these products that are not
dependent on SCI chemistry. Furthermore, the OH scavenger
experiments have indicated that these channels do not signifi-
cantly rely on the RO2 radical chemistry. For pinonic acid, one
possibility is the suggested ester channel of CI1. The formation
of pinonaldehyde under conditions of OH scavenging and in
the absence of water vapour has been observed in a number of
studies,10–14,50,54–56,58,59 while the exact mechanism still re-
mains unclear. The O-atom elimination channel has been
reported to be of minor importance for the ozonolysis of a-
pinene.60 Other mechanisms such as the CI self reaction or the
reaction of CI with peroxy radicals as postulated by Berndt
et al.56 might be possible, but no experimental evidence has
been provided for these paths. In the absence of convincing
evidence, no further speculation will be made here on the
formation mechanism of pinonaldehyde other than the reac-
tion of SCI with water.
3.5 Impact of NO2 addition
As indicated above, the key species in the suggested formation
mechanisms for pinic acid and pinalic-3-acid are the acylper-
oxy radical intermediates. The experimental detection of such
intermediates is extremely difficult due to their short lifetimes
and highly reactive nature; nevertheless, the assumed mechan-
isms indicate that these acids would be largely suppressed in
the presence of NO2 because acylperoxy radicals will react
with NO2 forming relatively stable peroxyacyl nitrates (the
alkyl peroxynitrates formed in reactions of alkyl peroxy
radicals with NO2 are known to rapidly decompose back to
reactants at room temperature).26 Experiments were thus
carried out by examining product formation with added
NO2 in order to probe the importance of acylperoxy inter-
mediates in the formation of these products. Such a study of
the NO2 dependence of ozonolysis products is also interesting
due to the lack of experimental data and the potential atmo-
spheric relevance.
As O3 can react with NO2 to form NO3 radicals which
would react readily with a-pinene and lead to additional
products,26 relatively high concentrations of a-pinene (B25
ppmv) and low concentrations of NO2 (up to 0.6 ppmv) were
used in these experiments. Based on the concentrations em-
ployed and the rate constant ratio k(O3 + a-pinene)/k(O3 +
NO2) B 3 at 298 K and atmospheric pressure,26 499% of O3
should have reacted with a-pinene in these reactions and so the
competing reactions of O3 with NO2 against a-pinene can be
deemed to be minimal. It was also noted that NO2 may react
with a-pinene at a slow but not negligible rate;26 hence NO2
was not premixed with the a-pinene mixture but added to the
reaction chamber immediately prior to O3. All reactions were
carried out under dry conditions and in the presence of excess
cyclohexane to scavenge 499% of OH radicals to eliminate
any complication from the OH–NO2 reaction system.
As shown in Fig. 5c and d, yields of pinic acid and pinalic-3-
acid also appeared to be somewhat dependent on water
vapour concentration, in particular in the presence of metha-
nol scavenger; this is unexpected given the assumption that
they are formed from the unimolecular decomposition of CI2
via reactions of radical intermediates. Nevertheless, experi-
ments carried out in the presence and absence of acetic acid as
SCI scavenger gave very similar results, implying that SCI
reactions are not involved in the pathway to these acids and so
the observed RH dependence did not result from the SCI
reaction. Considering the sensitivity of pinic acid to radical
chemistry, a possible explanation is that water vapour may
alter pinic and pinalic-3-acid formation by affecting their
radical-mediated mechanism, e.g. by changing the peroxy
radical populations in the reaction system. It is known that
HO2 radical and water can form an adduct in the gas phase,
which accelerates the HO2 self reactions and so would produce
Results from these experiments are illustrated in Fig. 6.
Addition of 0.02 ppmv of NO2 appeared to have no significant
effect on the formation of the above products. However, the
increase of NO2 mixing ratio to 0.2 ppmv and further to 0.6
ppmv led to a marked reduction of pinic acid and a modest
decrease of pinalic-3-acid and pinonic acid. The formation of
pinonaldehyde, in contrast, was unchanged within experimen-
tal uncertainties. While the reaction of NO2 with SCI could
potentially occur and may be implicated in the observed yield
variation, our experiments with added SCI scavenger have
indicated that SCI reactions are not important in the
ꢀc
This journal is the Owner Societies 2008
Phys. Chem. Chem. Phys., 2008, 10, 4294–4312 | 4303

View Article Online
sively from the decomposition of the more substituted
CI2.11–13,18,19,53 In order to test these assumptions, we have
synthesised an enal (A) and an enone (B) derived from a-
pinene that can only give rise to CI2 and CI1, respectively,
when ozonised. Some key findings concerning pinic, pinonic
and norpinonic acid formation have been reported pre-
viously36,37 and will be mentioned briefly for completeness.
As expected, pinonaldehyde was identified from both the
enal and the enone ozonolysis; its main source in these
reactions is presumably via the decomposition of primary
ozonides; although other paths may be possible. Pinic acid
and pinalic-3-acid were identified from the ozonolysis of the
enal but not from the enone, thus confirming the suggestion
that these acids are generated exclusively from CI2 during a-
pinene ozonolysis. In contrast to the general expectation,
pinonic acid was observed from both the enal and the enone
and hence can be formed via both CIs.
Fig. 6 Molar yields of pinonic acid, pinic acid, pinonaldehyde and
pinalic-3-acid from a-pinene ozonolysis at different NO2 mixing ratios;
note that for clarity, the pinonic acid yield plot has been displaced
vertically by À0.1%.
Fig. 7 illustrates the yields of pinic acid, pinalic-3-acid, pinonic
acid and pinonaldehyde from the enal ozonolysis under various
reaction conditions (cyclohexane or methanol as OH scavenger,
different RH, presence or absence of acetic acid as SCI scaven-
ger), which can be compared with the results obtained in similar
experiments for a-pinene ozonolysis (Fig. 5).
formation of the above products under dry conditions. There-
fore, a more likely interpretation for the observed NO2 effect
lies in its reaction with acylperoxy radical intermediates; and
the diverse role played by NO2 may be taken as an indication
that the acylperoxy radical chemistry is involved to different
degrees in the formation of various products. For pinalic-3-
acid, the likely formation mechanism is via the reaction of the
corresponding C9 acylperoxy radical with either RO2 or HO2
radicals, which occurs in direct competition with the removal
by NO2; thus its yield can be expected to decrease with
increasing [NO2], in accord with the experimental observation.
Comparatively the much more significant change in pinic acid
yield implies that its production is more strongly dependent on
reactions of acylperoxy radical intermediates. This is qualita-
tively consistent with the mechanism proposed by Jenkin
et al.19 where the pathway leading to pinic acid relies on the
participation of two acylperoxy intermediates, each of which
can be suppressed by reaction with NO2. However, the
observed decrease in pinonic acid yield is unexpected based
on the usually assumed formation mechanism via the direct
rearrangement of CI1 and would suggest that at least part of
the route to pinonic acid also involves acylperoxy chemistry; a
possible mechanism is discussed in the following section. The
observation that pinonaldehyde yield was insensitive to NO2
implies that its formation route under the experimental con-
ditions does not depend upon acylperoxy radical reactions,
although the exact mechanism is not explicit.
The pinic acid plot shows clearly that its formation yield
from the enal ozonolysis was primarily dependent on the
identity of the OH scavenger with some dependence on the
presence of added water vapour, similar to that observed in
the a-pinene ozonolysis experiments. In the presence of cyclo-
hexane scavenger, yields of pinic acid from the enal ozonolysis
seem to agree well with those for a-pinene ozonolysis; how-
ever, the measured yields from the enal with methanol sca-
venger were much lower than those from a-pinene in similar
experiments. A possible interpretation may be that in the a-
pinene/O3 system, the interaction of the other CI (CI1) with
methanol can make additional contributions to pinic acid, yet
to envisage such a mechanism is difficult. Given the sensitivity
of pinic acid to radical chemistry, however, an alternative
explanation lies in the differing radical concentrations in the
two systems. For example, the decomposition of a-pinene-CIs
gives rise to OH and alkyl radicals and the latter will add O2
forming RO2 radicals, while the decomposition of the C1–CI
formed in the enal ozonolysis is known to lead partially to OH
and HCO (and possibly H) radicals,27 and further reactions of
HCO and H with O2 would give HO2 radicals instead. As a
consequence the [RO2]/[HO2] ratio can be expected to be much
lower in the enal reactions; such difference may be unimpor-
tant in the cyclohexane system but could have a marked
inhibiting effect in the methanol system where the diacid
formation is more sensitive to the change in peroxy radical
populations. Formation of pinic acid from the enal ozonolysis
also presented a dependence on RH as observed in the a-
pinene ozonolysis experiments, albeit the magnitude of this
effect is again greater in the enal system, particularly in the
presence of methanol scavenger. Since this water dependence
appeared undisturbed by the addition of acetic acid, the route
to pinic acid likely does not rely on SCI reactions, a finding
consistent with that obtained for a-pinene. Hence, the same
reasoning as discussed above for a-pinene experiments could
explain the RH enhanced pinic acid formation observed for
3.6 Enal and enone experiments
As we have previously described,36,37 one particular difficulty
in interpreting the ozonolysis mechanism is to determine
which CI generates which products. For the ozonolysis of a-
pinene, two CIs are generated and both of them can go on to
decompose and/or react, producing a large array of products.
It has usually been assumed that pinonic acid, norpinonic acid,
pinonaldehyde and norpinonaldehyde are generated exclu-
sively via CI1, while most of the other products including
pinic acid, pinalic-3-acid and norpinic acid are formed exclu-
ꢀc
This journal is the Owner Societies 2008
4304 | Phys. Chem. Chem. Phys., 2008, 10, 4294–4312

View Article Online
Fig. 7 Molar yields of (a) pinic acid, (b) pinalic-3-acid, (c) pinonic acid and (d) pinonaldehyde from the enal ozonolysis under various conditions.
the enal, while the apparent greater change in the methanol
system compared to a-pinene results may also be rationalised
by assuming a lowered [RO2]/[HO2] ratio in the enal ozono-
lysis reaction. From these findings it can be concluded that
pinic acid is generated via RO2 radical mediated chemistry
following the unimolecular decomposition of CI2 during a-
pinene ozonolysis; this is entirely consistent with the postula-
tion made by Jenkin et al.19
The mechanisms for pinonic acid formation have been
discussed in our previous papers.36,37 From the results of the
enal and the enone experiments, it can be concluded that the
water-dependent formation of pinonic acid in the a-pinene
ozonolysis is due to CI1, while the water-independent forma-
tion is mainly due to CI2, with a very small fraction to CI1.
Thus the suggested ester channel of CI1 is at most a minor
source of the acid. The formation of pinonic acid via CI2 is
unexpected and may be explained by our proposed mechanism
in ref. 36. This pathway to pinonic acid does not rely very
significantly on the relative amounts of RO2 to HO2 radicals in
the system, so it is consistent with the finding in Fig. 7c that
pinonic acid formation is not markedly affected by the use of
different OH scavengers. Furthermore, since acylperoxy radi-
cals can be sequestered by reaction with NO2, the proposed
mechanism which involves an acylperoxy radical reaction with
other peroxy radicals can also explain the observed reduction
in pinonic acid yield upon the NO2 addition as presented in
Fig. 6. Thus this mechanism to form pinonic acid is qualita-
tively consistent with the experimental observations.
Formation of pinalic-3-acid from the enal ozonolysis also
showed a dependence on OH scavengers similar to that
observed for a-pinene ozonolysis. Like pinic acid, the pina-
lic-3-acid yields from the enal in the presence of cyclohexane
scavenger were similar to those from a-pinene, whereas the
yields measured for methanol scavenger appeared much smal-
ler. This again may be explained by the above hypothesis
concerning the differing peroxy radical chemistry in the enal
and a-pinene systems. Also, the general trends of pinalic-3-
acid formation with added water vapour and acetic acid were
similar to those observed for a-pinene in similar experiments.
Thus the source of pinalic-3-acid can also be attributed to the
RO2 mediated unimolecular decomposition of CI2. The fact
that pinic acid exhibited a greater sensitivity to radical chem-
istry than pinalic-3-acid is consistent with, and supportive of,
the above discussed mechanisms where the production of
pinalic-3-acid is only partially influenced by RO2–RO2 radical
reactions, while the pinic acid formation involves more RO2
permutation reactions.
The pinonaldehyde results are also interesting. Since it
forms as a primary carbonyl product from the ozonolysis of
the enal and the enone, it would not be possible to determine
whether its measured yields under dry conditions were exclu-
sively from the decomposition of POZ or whether other routes
also exist. Fig. 8 illustrates that the pinonaldehyde yield from
the enone ozonolysis increased with increasing RH, which
ꢀc
This journal is the Owner Societies 2008
Phys. Chem. Chem. Phys., 2008, 10, 4294–4312 | 4305

View Article Online
Fig. 8 Molar yields of pinonaldehyde from the enone ozonolysis at
different relative humidities in the presence of cyclohexane as OH
scavenger.
Fig. 9 Time evolution of a-pinene ozonolysis products in the reaction
chamber or on adsorbent filter; (1) norpinonic acid, (2) norpinic acid,
(3) norpinonaldehyde, (4) norpinalic acid, (5) 10-OH-pinonic acid, (6)
OH-pinalic acid.
provides further evidence for the suggested mechanism that
water reacts with SCI1 and contributes to pinonaldehyde
formation. However, it is unexpected that the formation of
pinonaldehyde from the enal ozonolysis also appeared to be
water dependent, as illustrated in Fig. 7d. Such an increase
(B2.40% at dry conditions and B2.95% at 80% RH in the
presence of cyclohexane scavenger; note that the data with
methanol scavenger may be complicated with the SCI–metha-
nol reaction) may not be very significant considering the
statistical uncertainty in pinonaldehyde yield measurements;
however, the addition of acetic acid apparently suppressed this
water enhanced fraction, implying that the additional pino-
naldehyde formed under humid conditions resulted from the
interaction of SCI with water. This would suggest that there
also exists stabilisation and bimolecular reactions of CI2. This
finding contradicts the general assumption that the di-substi-
tuted CI2 decomposes exclusively to generate OH, but is
consistent with the conclusion drawn by Warscheid and
Hoffmann54 in a study of pinonaldehyde formation from a-
pinene ozonolysis using isotopically labelled water (H218O).
products from the gas-phase ozonolysis of a-pinene under
the experimental conditions applied.
The time dependences of these products are depicted in
Fig. 9 (from the same experiments as for Fig. 2). Again, a
general trend is observed that during the ageing experiment the
compounds containing aldehyde groups (norpinonaldehyde,
norpinalic acid and OH-pinalic acid) decreased while the more
highly oxidised acids (norpinonic acid, norpinic acid and OH-
pinonic acid) increased, and such yield variance appeared
more significant on the adsorbent filter compared with the
evolution in the chamber. These observations further support
the notion that secondary condensed phase reactions could
occur following the gas-phase ozonolysis and change the
formation and composition of SOA products. However, as
explained for the other products, these processes are not
believed to be important on the timescale of our experiments.
Illustrated in Fig. 10 are formation yields of these products
from the ozonolysis of a-pinene, the enal precursor and the enone
precursor under various reaction conditions. Note that the data
from a-pinene and the enal ozonolysis are the average of those
obtained in the presence and absence of acetic acid as these values
were found to be insensitive to the addition of this compound.
For the ozonolysis of the enone, only one set of experiments with
cyclohexane scavenger at different RH was performed.
3.7 Formation mechanisms for other identified products
In this section, the formation behaviour of other identified
products with minor yields in the condensed phase, including
norpinic acid, norpinonic acid, norpinalic acid, OH-pinonic
acid, OH-pinalic acid and norpinonaldehyde, is examined so
as to elucidate the probable pathways leading to their produc-
tion during a-pinene ozonolysis. Literature studies have usual-
ly assumed that the decarboxylated acids such as norpinic acid
and norpinonic acid derive from the further oxidation of first-
generation aldehydic products by secondary OH radicals or
O3.11,13,18 However, under conditions where OH and O3 have
been essentially reacted, the exact mechanism is not clear.
While the particle-phase secondary reactions may contribute
to their formation (see discussion in section 3.2), decreasing
the initial ozone concentration by a factor of ca. 6 (B25 ppmv
a-pinene, B20–3.5 ppmv O3) resulted in no statistically very
significant change in the formation of the above products,
suggesting that these products are likely formed as direct
3.7.1 Other products generated via CI2. As shown in
Fig. 10, norpinic, norpinalic, OH-pinalic and OH-pinonic
acids were observed from the ozonolysis of a-pinene and the
enal, but none of them were identified for the enone; this gives
conclusive evidence that all these acids are generated exclu-
sively via CI2 during a-pinene ozonolysis. Norpinonaldehyde
has usually been assumed to arise from CI1. Here unexpect-
edly, it was also observed from the enal ozonolysis but with a
very low yield compared to that from a-pinene ozonolysis; this
indicates that norpinonaldehyde can also be formed from CI2
but this channel is likely to be minor.
Among these products, norpinic acid, norpinalic acid and
OH-pinalic acid appeared to behave similar to pinic acid in
that their formation from a-pinene ozonolysis is strongly
ꢀc
This journal is the Owner Societies 2008
4306 | Phys. Chem. Chem. Phys., 2008, 10, 4294–4312

View Article Online
Fig. 10 Molar yields of (a) norpinic acid, (b) norpinalic acid, (c) OH-pinalic acid, (d) OH-pinonic acid, (e) norpinonic acid and (f)
norpinonaldehyde from the ozonolysis of a-pinene, the enal precursor and the enone precursor under various conditions. ’, from a-pinene
(cyclohexane scavenger); &, from a-pinene (methanol scavenger); m, from the enal precursor (cyclohexane scavenger); n, from the enal precursor
(methanol scavenger); ~, from the enone precursor (cyclohexane scavenger). Absence of a data point indicates that the yield is zero.
dependent on the identity of OH scavenger used, with an
average decrease of 30–40% in yield when methanol was used
compared with cyclohexane; they appeared weakly dependent
on water vapour concentration (especially in the presence of
methanol scavenger) but were unaffected by the presence of
acetic acid as SCI scavenger. Their formation from the enal
ozonolysis showed similar dependences on OH scavengers and
RH as observed in a-pinene experiments, although the mag-
nitude of these effects appeared even greater in the enal system;
also, formation yields of these products from the enal reac-
tions were insensitive to the presence of acetic acid and were
similar to those from a-pinene reactions in the presence of
cyclohexane scavenger, while in the presence of methanol
scavenger, relatively lower yields were measured in the enal
system.
via a number of permutation reactions of RO2 intermediates. To
further elucidate the probable mechanisms, formation behaviour
of these acids from a-pinene ozonolysis in the presence of NO2
was examined (from the same experiments as described in
section 3.5); results obtained under three NO2 levels (0.02 ppmv,
0.2 ppmv and 0.6 ppmv) are illustrated in Fig. 11.
Yields of norpinic acid, norpinalic acid and OH-pinalic acid
all decreased with increasing NO2 concentration while the
extent of the change differs between compounds, which implies
that the formation of all these acids is influenced by reaction of
acylperoxy radicals but the importance of such reactions is
likely different among their routes of formation. A quantita-
tive inspection of the yield data shows that norpinic acid
behaves like pinic acid; these two are the most sensitive to
NO2, with 450% reduction in yield when NO2 concentration
was raised from 0.02 to 0.6 ppmv, whereas OH-pinalic acid
and norpinalic acid were decreased by B32% and B38%
respectively, similar to that for pinalic-3-acid (B30%
These observations indicate that the routes to these three acids
might follow a RO2 radical mediated chemistry similar to that
for pinic acid, i.e., from the unimolecular decomposition of CI2
ꢀc
This journal is the Owner Societies 2008
Phys. Chem. Chem. Phys., 2008, 10, 4294–4312 | 4307

View Article Online
from the carbon adjacent to the carbonyl functionality. This
possibility is supported by a recent theoretical study indicating
that the C–H bond strength at such an exocyclic CH2 group is
much lower than expected for a secondary abstraction site
with a neighbouring carbonyl group.62 The authors of this
paper suggested that this is ascribed to a combination of ring-
strain-enhanced hyperconjugation and the vinoxy-type reso-
nance stabilisation in the product radical. Here, the resulting
vinoxy radical (3) may exist in resonance structures (3a), (3b)
and (3c), which can then add O2 forming a peroxy radical (4)
and its reaction with RO2 radicals would generate 4-OH-
pinalic-3-acid as well as the corresponding oxy radical (5).
Further decomposition of this oxy radical (5) would directly
produce norpinalic acid by loss of HCO or it might eliminate
HCHO leading to a C8 acylperoxy radical (6), a subsequent
reaction of which with either RO2 or HO2 radicals would then
yield norpinic acid. Alternatively, norpinalic acid may be
generated from a permutation reaction of the acylperoxy
precursor (7) to pinic acid, followed by decomposition and
peroxy radical reactions of the resulting acyl-oxy radical (8).
These mechanisms can then explain the direct formation of
norpinalic, norpinic and OH-pinalic acids from the gas-phase
ozonolysis of both a-pinene and the enal. Since these reaction
paths all involve a series of RO2 permutation reactions, the
acid production can be expected to be strongly affected by the
use of different OH scavengers, in accord with the experimen-
tal observations. These are also consistent with the NO2
experimental results: the formation of OH-pinalic acid in-
volves the participation of only one acylperoxy radical inter-
mediate, the same as pinalic-3-acid, and the magnitude of
depression of these two products upon NO2 addition was in
Fig. 11 Molar yields of a-pinene ozonolysis products at different
NO2 levels.
decrease). While these measurements have experimental un-
certainties associated with them, the relative data trends
observed can provide useful information in terms of the extent
of involvement of acylperoxy radical reactions in the genera-
tion of individual products. Combining all the above experi-
mental findings, possible mechanisms describing the formation
of these acids are proposed in Scheme 7, starting from the
same C9 acylperoxy radical (1) that also potentially leads to
pinic acid and pinalic-3-acid (see Scheme 4).
In the proposed mechanism, it is assumed that in addition to
thermal decomposition and the transfer of the aldehydic
hydrogen that leads subsequently to pinic acid, the C9 acyl-
oxy radical (2) may also isomerise by abstraction of a H atom
Scheme 7 Proposed formation mechanisms for norpinic acid, norpinalic acid and 4-OH-pinalic-3-acid from CI2 (postulated reaction steps are
shown in dashed arrows), along with the suggested mechanism by Jenkin et al.19 for pinic acid and pinalic-3-acid.
ꢀc
This journal is the Owner Societies 2008
4308 | Phys. Chem. Chem. Phys., 2008, 10, 4294–4312

View Article Online
fact observed to be similar. Formation of both norpinic acid
and pinic acid requires reactions of two acylperoxy radicals,
which agrees with the observed more significant effect of NO2
on these compounds compared with pinalic-3-acid and OH-
pinalic acid. While norpinalic acid could be generated in two
paths involving, respectively, one or two acylperoxy radicals
and this may explain a NO2 effect between the above two
categories of products.
seems to be a trend for its formation from the enal ozonolysis
to increase when OH scavenger was changed from methanol to
cyclohexane. If these differences are real, the formation of
norpinonaldehyde from CI2 might be explained by the follow-
ing reaction scheme, via the further reaction and decomposi-
tion of the acylperoxy radical (13) which is proposed to give
rise to pinonic acid.
3.7.2 Other products generated via CI1. In addition to
pinonic acid and pinonaldehyde, norpinonic acid and norpi-
nonaldehyde were also detected from the enone ozonolysis,
indicating that these products can be formed via CI1. No
norpinonic acid was observed from the enal ozonolysis; thus
its only source in a-pinene ozonolysis is from CI1. The
norpinonic acid yield presented a weak dependence on the
OH scavenger type and NO2 concentration but appeared
undisturbed by the addition of acetic acid (see Fig. 10e and
11), which may be rationalised by a radical chain mechanism
following the unimolecular decomposition of CI1. A possible
reaction route consistent with the above experimental findings
is shown in Scheme 10: application of the hydroperoxide
channel to CI1 results in an a-carbonyl peroxy radical (18),
which may undergo a permutation reaction to give an oxy
radical (19). Decomposition of (19) by elimination of HCO
would generate norpinonaldehyde, as has been suggested in
several previous studies.11,13,18 In addition, it is proposed
(represented with a dashed arrow) that the oxy radical (19)
may also lose HCHO and react with O2 leading to an
acylperoxy radical (20), whose reaction with either RO2 or
HO2 radicals would thus account for the direct formation of
norpinonic acid from CI1. Furthermore, as we have previously
discussed,37 the observation that norpinonic acid yields from
the ozonolysis of a-pinene and the enone (in the presence of
cyclohexane as OH scavenger) are almost indistinguishable
indicates that the branching ratio for CI1 in both systems is
close to 0.5. Norpinonaldehyde was observed from the enone
ozonolysis with similar yields to those from a-pinene in the
presence of cyclohexane scavenger (see Fig. 10f), and thus it
can be concluded that the formation of norpinonaldehyde in
a-pinene ozonolysis is predominantly due to CI1 (likely via the
mechanism shown in Scheme 10), with a small fraction to CI2
(likely via the mechanism shown in Scheme 9).
In comparison, OH-pinonic acid appeared to behave more
like pinalic-3-acid in that both of them can be affected by OH
scavengers, RH and NO2 but the extent of these effects was
smaller compared with the above mentioned pinic acid type
products. The only direct gas-phase mechanism proposed for
10-OH pinonic acid is via an isomerisation reaction (1,8-H
shift) of a C10 a-carbonyl oxy radical (10) formed in the
hydroperoxide channel of the decomposition of CI2,19 as
illustrated in Scheme 8. This mechanism is consistent with
the present experimental findings that OH-pinonic acid was
formed in both a-pinene and the enal ozonolysis under ‘OH-
free’ conditions and irrespective of the presence of SCI sca-
vengers. Similar to pinalic-3-acid, the suggested pathway to
OH-pinonic acid involves one permutation reaction and one
reaction of acyl RO2 intermediate with RO2 or HO2 radicals,
which could explain the observed similar degree of lowering
effect of methanol and NO2 on the yields of these compounds.
The experimental results obtained are thus in general agree-
ment with this postulated reaction path. Note that electronic
structure calculations by Jenkin et al.19 indicate that the
isomerisation of (10) to (11) is thermodynamically feasible
but is likely to be a minor channel compared with the decom-
position of (10) to HCHO and the acylperoxy radical (1), the
latter being the potential precursor to pinic and pinalic-3-acid
(and possibly to norpinalic acid, norpinic acid and OH-pinalic
acid as discussed above). This accords with the lower forma-
tion yield of OH-pinonic acid compared with pinic and
pinalic-3-acids.
Measurements of norpinonaldehyde showed relatively large
uncertainties compared to acid products, but it is still possible
to glean some general data trends from the average filter
measurements. The yield of norpinonaldehyde from a-pinene
ozonolysis appeared to be weakly dependent on OH scaven-
gers but not very significantly affected by SCI scavengers,
implying that its formation pathway may involve RO2 radical
chemistry but not likely rely on SCI reactions. Also there
4. Conclusions
A detailed study has been carried out to investigate the
formation of aerosol-phase products in the gas-phase ozono-
lysis of a-pinene under ‘OH-free’ conditions. In the filter-
collected reaction products a number of C8–C10 multifunc-
tional carboxylic acids and carbonyl compounds were identi-
fied using GC/MS analysis after derivatisation. The
mechanisms of formation of these products have been inves-
tigated under a variety of reaction conditions and using
compounds that only generate one or other of the two Criegee
intermediates. Results of these experiments allowed previously
postulated mechanisms to be evaluated and, based on experi-
mental observations and mechanistic considerations, a num-
ber of new mechanisms have been suggested to explain the
observed product formation.
Scheme 8 Proposed formation mechanism for 10-OH-pinonic acid
by Jenkin et al.19
ꢀc
This journal is the Owner Societies 2008
Phys. Chem. Chem. Phys., 2008, 10, 4294–4312 | 4309

View Article Online
Scheme 10 Proposed mechanisms for norpinonic acid and norpino-
naldehyde formation from CI1 (postulated reaction steps are shown in
dashed arrows).
Scheme 9 Proposed mechanism for norpinonaldehyde formation
from CI2 (postulated reaction steps are shown in dashed arrows).
The results described above can be summarised in reaction
Scheme 11, which simply shows the early mechanistic steps in
the formation of the investigated products. Briefly, the gas-
phase ozonolysis of a-pinene leads to two CIs, CI1 and CI2,
with virtually identical yields. A portion of the mono-substi-
tuted CI1 is stabilised and forms pinonaldehyde and pinonic
acid when reacting with water; the remaining undergoes un-
imolecular decomposition/isomerisation to give OH radicals
and other products including norpinonic acid and norpino-
naldehyde. The di-substituted CI2 presumably decomposes via
the hydroperoxide channel forming OH radicals and radical
co-products and the latter undergoes further radical chain
reactions leading to a variety of products including pinic acid,
pinalic-3-acid, pinonic acid, norpinic acid, norpinalic acid,
OH-pinalic acid, OH-pinonic acid and norpinonaldehyde. In
addition, a fraction of CI2 is also stabilised and can react with
water to generate pinonaldehyde.
The mechanisms that we have proposed are consistent with
a wide range of experimental observations, although a number
of the proposed steps are difficult to validate. Further experi-
mental and theoretical studies are required to more firmly
establish the proposed reaction pathways.
5. Atmospheric implications
This study confirmed the previous observation that a series of
low-volatility multifunctional acids are formed during the gas-
phase reaction of ozone with a-pinene. The experimental
results described improved our understanding of the reaction
mechanisms that lead to these important SOA components
Scheme 11
ꢀc
This journal is the Owner Societies 2008
4310 | Phys. Chem. Chem. Phys., 2008, 10, 4294–4312

View Article Online
and can be used in detailed atmospheric chemistry models
such as the Master Chemical Mechanism.63
References
1. J. H. Seinfeld and S. N. Pandis, Atmospheric Chemistry and
Physics: From Air Pollution to Climate Change, Wiley-Inter-
science, New York, 1998.
It is worth noting that the concentrations of reactants used
in this study are very much greater than those observed in the
atmosphere—as is the case in the majority of laboratory
studies. Nevertheless, the results are applicable under condi-
tions where the fate of peroxy radicals is reaction with RO2 or
HO2 rather than NOx (NO + NO2); i.e. the results apply to
the chemistry of the rural, unpolluted atmosphere. More
specifically, since the acids investigated in this work showed
different dependences on reaction parameters, they are likely
to be differently affected by ambient conditions. Pinonic acid
presented a relatively strong water dependence but minor
dependence on NO2 and RO2, and hence its formation can
be expected to readily occur in both polluted and clean
environments. In contrast, the other acids investigated—in
particular the diacids—were significantly affected by [RO2]/
[HO2] ratios and NO2 concentration, indicating a formation
mechanism in which RO2 radicals, including acylperoxy radi-
cals, play a key intermediate role. The yields of these com-
pounds would thus depend strongly on reactions of these
radical intermediates. In the atmosphere, a series of photo-
chemical reactions allow NO and NO2 to be constantly inter-
changed and coexist with O3,64 and so important reactions of
the RO2 radicals produced following ozonolysis are with NOx,
HO2 radicals and other RO2 radicals, and the competition
between these reactions would thus determine the nature and
yields of the reaction products. While the experimental in-
vestigation of the impact of NO is difficult due to the rapid
reaction of O3 with NO to form NO2, it can be expected that
under normal atmospheric conditions NO will react with
alkyl- and acyl-peroxy radicals forming mainly NO2 and
corresponding alkoxy radicals, and NO2 will react with acyl-
peroxy radicals to form peroxyacyl nitrates.64 Conversion of
alkylperoxy radicals to alkoxy radicals is a key process for the
formation of many of the acids, but conversion of acylperoxy
radicals to PAN type species reduces their formation. Thus,
the exact impact of polluted conditions on acid formation may
depend on the [NO]/[NO2] ratio, as well as the NOx concen-
tration. Under high [RO2] and low [NOx] conditions, such as
might be expected in forested areas with relatively large
terpene emissions, formation of acids should be favoured. In
the polluted environment, the gas-to-particle conversion may
also occur to some extent via the phase partitioning of higher-
volatility compounds into preexisting aerosol phase. It can
thus be concluded that SOA yields and composition are very
variable in the real atmosphere, depending in a complex way
on ambient conditions such as the abundances of parent
hydrocarbon, ozone, nitrogen oxides, HO2 radicals and RO2
radicals.
2. I. G. Kavouras, N. Mihalopoulos and E. G. Stefanou, Nature,
1998, 395, 683.
3. R. W. Atkinson, H. R. Anderson, D. P. Strachan, J. M. Bland and
S. A. Bremner, Eur. Respir. J., 1999, 13, 257.
4. C. Pilinis, S. N. Pandis and J. H. Seinfeld, J. Geophys. Res.,
[Atmos.], 1995, 100, 18739.
5. M. O. Andreae and P. J. Crutzen, Science, 1997, 276, 1052.
6. IPCC Climate Changes. The Science of Climate Changes, ed.
L. G. Houghton,
F Meira, B. A. Callendar, N Harris,
A Katenberg and K Maskel, Cambridge University Press, Cam-
bridge, 1995.
7. T. Hoffmann, J. R. Odum, F. Bowman, D. Collins, D. Klockow,
R. C. Flagan and J. H. Seinfeld, Atmos. Chem., 1997, 26, 189.
8. R. J. Griffin, D. R. Cocker, R. C. Flagan and J. H. Seinfeld,
J. Geophys. Res., [Atmos.], 1999, 104, 3555.
9. A. Guenther, C. N. Hewitt, D. Erickson, R. Fall, C. Geron,
T. Graedel, P. Harley, L. Klinger, M. Lerdau, W. A. Mckay,
T. Pierce, B. Scholes, R. Steinbrecher, R. Tallamraju, J. Taylor
and P. Zimmerman, J. Geophys. Res., [Atmos.], 1995, 100, 8873.
10. T. S. Christoffersen, J. Hjorth, O. Horie, N. R. Jensen, D. Kotzias,
L. L. Molander, P. Neeb, L. Ruppert, R. Winterhalter, A. Virkkula,
K. Wirtz and B. R. Larsen, Atmos. Environ., 1998, 32, 1657.
11. J. Yu, R. C. Flagan and J. H. Seinfeld, Environ. Sci. Technol.,
1998, 32, 2357.
12. S. Koch, R. Winterhalter, E. Uherek, A. Kolloff, P. Neeb and
G. K. Moortgat, Atmos. Environ., 2000, 34, 4031.
13. M. Jang and R. M. Kamens, Atmos. Environ., 1999, 33, 459.
14. M. Glasius, M. Lahaniati, A. Calogirou, D. Di Bella,
N. R. Jensen, J. Hjorth, D. Kotzias and B. R. Larsen, Environ.
Sci. Technol., 2000, 34, 1001.
15. U. Kuckelmann, B. Warscheid and T. Hoffmann, Anal. Chem.,
¨
2000, 72, 1905.
16. B. Warscheid and T. Hoffmann, Rapid Commun. Mass Spectrom.,
2001, 15, 2259.
17. B. Warscheid and T. Hoffmann, Rapid Commun. Mass Spectrom.,
2002, 16, 496.
18. R. Winterhalter, R. Van Dingenen, B. R. Larsen, N. R. Jensen
and J. Hjorth, Atmos. Chem. Phys. Discuss., 2003, 3, 1.
19. M. E. Jenkin, D. E. Shallcross and J. N. Harvey, Atmos. Environ.,
2000, 34, 2837.
20. I. G. Kavouras, N. Mihalopoulos and E. G. Stephanou, Nature,
1998, 395, 683.
21. I. G. Kavouras, N. Mihalopoulos and E. G. Stephanou, Geophys.
Res. Lett., 1999, 26, 55.
22. I. G. Kavouras, N. Mihalopoulos and E. G. Stephanou, Environ.
Sci. Technol., 1999, 33, 1028.
23. J. Yu, R. J. Griffin, D. R. Cocker, R. C. Flagan and J. H. Seinfeld,
Geophys. Res. Lett., 1999, 26, 1145.
24. C. O’Dowd, P. Aalto, K. Hameri, M. Kulmala and T. Hoffmann,
¨
Nature, 2002, 416, 497.
25. R. Criegee, Angew. Chem., Int. Ed. Engl., 1975, 14, 745.
26. R. Atkinson, J. Phys. Chem. Ref. Data, 1997, 26, 215.
27. J. G. Calvert, R. Atkinson, J. A. Kerr, S. Madronich,
G. K. Moortgat, T. J. Wallington and G. Yarwood, The mechan-
isms of atmospheric oxidation of the alkenes, Oxford University
Press, Oxford, 2000.
28. R. Gutbrod, E. Kraka, S. N. Schindler and D. Cremer, J. Am.
Chem. Soc., 1997, 119, 7330.
29. J. D. Fenske, K. T. Kuwata, K. N. Houk and S. E. Paulson,
J. Phys. Chem. A, 2000, 104, 7246.
30. J. H. Kroll, S. R. Sahay, J. G. Anderson, K. L. Demerjian and
N. M. Donahue, J. Phys. Chem. A, 2001, 105, 4446.
31. J. H. Kroll, J. S. Clarke, N. M. Donahue, J. G. Anderson and
K. L. Demerjian, J. Phys. Chem. A, 2001, 105, 1554.
32. P. Neeb, O. Horie and G. K. Moortgat, Tetrahedron Lett., 1996,
37, 9297.
Acknowledgements
The authors wish to thank NERC for support through the
COSMAS Programme (NER/T/S/2000/01083). Y.M. wishes
to thank the University of Reading for an Overseas Post-
graduate Research Studentship.
33. P. Neeb, F. Sauer, O. Horie and G. K. Moortgat, Atmos. Environ.,
1997, 31, 1417.
34. H. J. Tobias and P. J. Ziemann, Environ. Sci. Technol., 2000, 34,
2105.
ꢀc
This journal is the Owner Societies 2008
Phys. Chem. Chem. Phys., 2008, 10, 4294–4312 | 4311

View Article Online
35. H. J. Tobias, K. S. Docherty, D. E. Beving and P. J. Ziemann,
Environ. Sci. Technol., 2000, 34, 2116.
49. M. E. Jenkin, M. D. Hurley and T. J. Wallington, Phys. Chem.
Chem. Phys., 2007, 9, 3149.
36. Y. Ma, T. J. Wilcox, A. T. Russell and G. Marston, Chem.
Commun., 2007, 13, 1328.
50. J. Baker, S. M. Aschmann, J. Arey and R. Atkinson, Int. J. Chem.
Kinet., 2002, 34, 73.
37. Y. Ma, T. Luciani, R. A. Porter, A. T. Russell, D. Johnson and
G. Marston, Phys. Chem. Chem. Phys., 2007, 9, 5084.
38. T. Hoffmann, R. Bandur, U. Marggraf and M. Linscheid,
J. Geophys. Res., [Atmos.], 1998, 103, 25569.
39. B. R. Larsen, M. Lahaniati, A. Calogirou and D. Kotzias, Chemo-
sphere, 1998, 37, 1203.
40. M. Jaoui and R. M. Kamens, J. Atmos. Chem., 2003, 43, 101.
41. J. Yu, D. R. Cocker, R. J. Griffin, R. C. Flagan and J. H. Seinfeld,
J. Atmos. Chem., 1999, 34, 207.
42. K. S. Docherty, W. Wu, Y. B. Lim and P. J. Ziemann, Environ.
Sci. Technol., 2005, 39, 4049.
51. M. E. Jenkin, Atmos. Chem. Phys., 2004, 4, 1741.
52. R. Kamens, M. Jang, C. J. Chien and K. Leach, Environ. Sci.
Technol., 1999, 33, 1430.
53. S. Hatakeyama, K. Izumi, T. Fukuyama and H. Akimoto,
J. Geophys. Res., [Atmos.], 1989, 94, 13013.
54. B. Warscheid and T. Hoffmann, Atmos. Environ., 2001, 35, 877.
55. J. Fick, L. Pommer, C. Nilsson and B. Andersson, Atmos.
Environ., 2003, 37, 4087.
56. T. Berndt, O. Boge and F. Stratmann, Atmos. Environ., 2003, 37,
3933.
57. P. Neeb, O. Horie and G. K. Moortgat, Int. J. Chem. Kinet., 1996,
28, 721.
43. S. Gao, M. Keywood, N. L. Ng, J. Surratt, V. Varutbangkul,
R. Bahreini, R. C. Flagan and J. H. Seinfeld, J. Phys. Chem. A,
2004, 108, 10147.
58. J. D. Fenske, A. S. Hasson, A. W. Ho and S. E. Paulson, J. Phys.
Chem. A, 2000, 104, 9921.
44. R. Atkinson, S. M. Aschmann, J. Arey and B. Shorees,
J. Geophys. Res., [Atmos.], 1992, 97, 6065.
59. H. Hakola, J. Arey, S. M. Aschmann and R. Atkinson, J. Atmos.
Chem., 1994, 18, 75.
45. M. D. Keywood, J. H. Kroll, V. Varutbangkul, R. Bahreini,
R. C. Flagan and J. H. Seinfeld, Environ. Sci. Technol., 2004, 38, 3343.
46. K. S. Docherty and P. J. Ziemann, Atmos. Sci. Technol., 2003, 37,
877.
47. L. E. Christensen, M. Okumura, S. P. Sander, R. J. Salawitch,
G. C. Toon, B. Sen, J.-F. Blavier and K. W. Jucks, Geophys. Res.
Lett., 2002, 29, 1299.
60. A. Alvarado, E. C. Tuazon, S. M. Aschmann, R. Atkinson and
J. Arey, J. Geophys. Res., [Atmos.], 1998, 103, 25541.
61. C. C. Kircher and S. P. Sander, J. Phys. Chem., 1984, 88, 2082.
62. L. Vereecken and J. Peeters, Phys. Chem. Chem. Phys., 2002, 4,
467.
63. Master Chemical Mechanism, version 3 (MCM v3), http://mcm.
leeds.ac.uk/MCM/.
48. L. E. Christensen, M. Okumura, J. C. Hansen, S. P. Sander and
J. S. Francisco, J. Phys. Chem. A, 2006, 110, 6948.
64. R. P. Wayne, Chemistry of Atmospheres, Oxford University Press,
3rd edn, Oxford, 2000.
ꢀc
This journal is the Owner Societies 2008
4312 | Phys. Chem. Chem. Phys., 2008, 10, 4294–4312