Carboxylic acids in secondary aerosols from oxidation of cyclic monoterpenes by ozone

Article abstract of DOI:10.1021/es990445r

A series of smog chamber experiments have been conducted in which five cyclic monoterpenes were oxidized by ozone. The evolved secondary aerosol was analyzed by GC-MS and HPLC-MS for nonvolatile polar oxidation products with emphasis on the identification of carboxylic acids. Three classes of compounds were determined at concentration levels corresponding to low percentage molar yields: i.e. dicarboxylic acids, oxocarboxylic acids, and hydroxyketocarboxylic acids. Carboxylic acids are highly polar and have lower vapor pressures than their corresponding aldehydes and may thus play an important role in secondary organic aerosol formation processes. The most abundant carboxylic acids were the following: cis-pinic acid AB1 (cis-3- carboxy-2,2-dimethylcyclobutylethanoic acid) from α-and β-pinene; cis- pinonic acid A3 (cis-3-acetyl-2,2-dimethylcyclobutylethanoic acid) and cis- 10-hydroxypinonic acid AB6 (cis-2,2-dimethyl-3- hydroxyacetylcyclobutylethanoic acid) from α-pinene and β-pinene; cis-3- caric acid C1 (cis-2,2-dimethyl-1,3-cyclopropyldiethanoic acid), cis-3- caronic acid C3 (2,2-dimethyl-3-(2-oxopropyl)cyclopropanylethanoic acid), and cis-10-hydroxy-3-caronic acid C6 (cis-2,2-dimethyl-3-(hydroxy-2- oxopropyl)cyclopropanylethanoic acid) from 3-carene; cis-sabinic acid S1 (cis-2-carboxy-1-isopropylcyclopropylethanoic acid) from sabinene; limonic acid L1 (3-isopropenylhexanedioic acid), limononic acid L3 (3-isopropenyl-6- oxo-heptanoic acid), 7-hydroxylimononic acid L6 (3-isopropenyl-7-hydroxy-6- oxoheptanoic acid), and 7-hydroxylimononic acid L6' (7-hydroxy-3-isopropenyl- 6-oxoheptanoic acid) from limonene. A series of smog chamber experiments have been conducted in which five cyclic monoterpenes were oxidized by ozone. The evolved secondary aerosol was analyzed by GC-MS and HPLC-MS for nonvolatile polar oxidation products with emphasis on the identification of carboxylic acids. Three classes of compounds were determined at concentration levels corresponding to low percentage molar yields: i.e. dicarboxylic acids, oxocarboxylic acids, and hydroxyketocarboxylic acids. Carboxylic acids are highly polar and have lower vapor pressures than their corresponding aldehydes and may thus play an important role in secondary organic aerosol formation processes. The most abundant carboxylic acids were the following: cis-pinic acid AB1 (cis-3-carboxy-2,2-dimethylcyclobutylethanoic acid) from α- and β-pinene; cis-pinonic acid A3 (cis-3-acetyl-2,2-dimethylcyclobutylethanoic acid) and cis-10-hydroxypinonic acid AB6 (cis-2,2-dimethyl-3-hydroxyacetylcyclobutyl-ethanoic acid) from α-pinene and β-pinene; cis-3-caric acid C1 (cis-2,2-dimethyl-1,3-cyclopropyldiethanoic acid), cis-3-caronic acid C3 (2,2-dimethyl-3-(2-oxopropyl)cyclopropanylethanoic acid), and cis-10-hydroxy-3-caronic acid C6 (cis-2,2-dimethyl-3-(hydroxy-2-oxopropyl)cyclopropanyl-ethanoic acid) from 3-carene; cis-sabinic acid S1 (cis-2-carboxy-1-isopropylcyclopropylethanoic acid) from sabinene; limonic acid L1 (3-isopropenylhexanedioic acid), limononic acid L3 (3-isopropenyl-6-oxo-heptanoic acid), 7-hydroxy-limononic acid L6 (3-isopropenyl-7-hydroxy-6-oxoheptanoic acid), and 7-hydroxylimononic acid L6′ (7-hydroxy-3-isopropenyl-6-oxoheptanoic acid) from limonene.

Full text of DOI:10.1021/es990445r

Environ. Sci. Technol. 2000, 34, 1001-1010
compounds such as terpenes, estimated on a global scale to
Carboxylic Acids in Secondary
Aerosols from Oxidation of Cyclic
Monoterpenes by Ozone
M A R I A N N E G L A S I U S , M A R I A L A H A N I A T I , †
A G G E L O S C A L O G I R O U , D A R I O D I B E L L A ,
N I E L S R . J E N S E N , J E N S H J O R T H ,
D I M I T R I O S K O T Z I A S , A N D
amount to 120-480 Tg y-1 (13, 14), has triggered investiga-
tions of the SOA forming potential of these compounds. The
first results indicate that biogeochemical sources of SOA may
play an important role in atmospheric chemistry (15-17).
Although very little is known about the mechanism for
the photochemical gas-to-particle conversion of terpenes it
is conceivable that the formation of oxidation products with
considerably lower vapor pressures than the parent com-
pounds plays an important role. The main products of gas-
phase reactions of terpenes with O3, OH, or NO3 are carbonyl
compounds, which are produced in molar yields up to 60%
but only found in low concentrations in SOA (1, 5, 7-9, 18-
20). Carbonyls generally possess relatively high vapor pres-
sures and cannot reach supersaturation to nucleate or
condense under the experimental conditions used. This
applies even for high molecular weight (C10) ketoaldehydes,
such as e.g. the main oxidation product of R-pinene and of
3-carene (21). Therefore, to explain the particle burst in most
SOA experiments attention has been turned to multifunc-
tional secondary products such as carboxylic acids, oxocar-
boxylic acids, hydroxyketones, hydroxycarboxylic acids, diols,
and dicarboxylic acids with stronger intermolecular forces
and much lower vapor pressures.
B O R . L A R S E N *
European Commission, Joint Research Centre, Ispra,
Environment Institute, TP 290 I-21020, Ispra (VA), Italy
A series of smog chamber experiments have been
conducted in which five cyclic monoterpenes were oxidized
by ozone. The evolved secondary aerosol was analyzed
by GC-MS and HPLC-MS for nonvolatile polar oxidation
products with emphasis on the identification of carboxylic
acids. Three classes of compounds were determined at
concentration levels corresponding to low percentage molar
yields: i.e. dicarboxylic acids, oxocarboxylic acids, and
hydroxyketocarboxylic acids. Carboxylic acids are highly
polar and have lower vapor pressures than their corresponding
aldehydes and may thus play an important role in secondary
organic aerosol formation processes. The most abundant
carboxylic acids were the following: cis-pinic acid AB1 (cis-
3-carboxy-2,2-dimethylcyclobutylethanoic acid) from R-
and â-pinene; cis-pinonic acid A3 (cis-3-acetyl-2,2-
dimethylcyclobutylethanoic acid) and cis-10-hydroxypinonic
acid AB6 (cis-2,2-dimethyl-3-hydroxyacetylcyclobutyl-
ethanoic acid) from R-pinene and â-pinene; cis-3-caric
acid C1 (cis-2,2-dimethyl-1,3-cyclopropyldiethanoic acid), cis-
3-caronic acid C3 (2,2-dimethyl-3-(2-oxopropyl)cyclopro-
panylethanoic acid), and cis-10-hydroxy-3-caronic acid C6
(cis-2,2-dimethyl-3-(hydroxy-2-oxopropyl)cyclopropanyl-
ethanoic acid) from 3-carene; cis-sabinic acid S1 (cis-2-
carboxy-1-isopropylcyclopropylethanoic acid) from sabinene;
limonic acid L1 (3-isopropenylhexanedioic acid), limononic
acid L3 (3-isopropenyl-6-oxo-heptanoic acid), 7-hydroxy-
limononic acid L6 (3-isopropenyl-7-hydroxy-6-oxoheptanoic
acid), and 7-hydroxylimononic acid L6′ (7-hydroxy-3-
isopropenyl-6-oxoheptanoic acid) from limonene.
Pinonic acid A3, a ketocarboxylic acid (3-acetyl-2,2-
dimethylcyclobutylethanoic acid) is a well-known minor
oxidation product of R-pinene, found both in the gas-phase
and the particle phase (1, 7, 10, 11, 18-21). Pinic acid AB1,
a dicarboxylic acid (3-carboxy-2,2-dimethylcyclobutyletha-
noic acid), was identified for the first time during smog
chamber experiments in 1997 in SOA from R-pinene/ O3 (22,
23) and â-pinene/ O3 (22) together with a decarboxylated
analogue, norpinic acid AB2 (3-carboxy-2,2-dimethylcy-
clobutylmethanoic acid). The reported yields of these
compounds (0.1-3%) can only account for a small proportion
of the evolved aerosol mass. By the use of various chemical
derivatization approaches combined with gas chromatog-
raphy-mass spectrometry (24-26) and gas chromatography-
Fourier transform infrared spectroscopy (25) to obtain
information on molecular mass and functional groups, recent
investigations have steadily increased the number of identi-
fied nonvolatile terpene oxidation products in SOA from
R-pinene (24-26), â-pinene (26), 3-carene (24, 26), sabinene
(26), and limonene (26). Nevertheless, a large proportion of
the aerosol mass still cannot be accounted for.
In this study characterization of the particle phase reaction
products from the reaction of the cyclic monoterpenes
R-pinene, â-pinene, 3-carene, sabinene, and limonene with
ozone was undertaken. A newly developed analytical method
was employed based on liquid chromatography interfaced
with tandem mass spectrometry (LC-MSn) using electrospray
ionization and atmospheric pressure chemical ionization (27).
The aim was primarily to identify new products in the particle
phase and to confirm previous tentatively identified products,
but attempts were also made to obtain quantitative infor-
mation on the particle composition. The choice of terpenes
was based upon their high emission rates from vegetation
and their proven SOA forming potential (11, 17).
Introduction
Smog chamber experiments carried out over the past 2
decades have identified secondary organic aerosol (SOA) from
gas-phase oxidation of volatile organic compounds (VOC) as
a potentially important contributor to the atmospheric
burden of particulate matter (1-11). Although observations
of blue haze in the atmosphere deriving from compounds
emitted by trees date back to 1960 (12), organic aerosol
research has mainly been focused on anthropogenic VOC
and their SOA forming potential with relevance to urban air
quality. The growing interest about biogenically emitted
Experimental Section
Sm og Cham ber Experim ents. The experiments were carried
out in 600-1200 L Teflon (PTFE) reaction chambers. Indi-
vidual terpenes were oxidized by ozone and the evolved SOA
was sampled and analyzed. The investigated compounds
were the cyclic monoterpenes (+)-R-pinene, (+)-â-pinene,
(+)-3-carene, (-)-limonene, and (+)-sabinene (Figure 1) used
* Corresponding author phone: +39-0332-789647; fax: +39-0332-
785704; e-mail: bo.larsen@jrc.it.
† Present address: National Centre for Scientific Research,
Demokritos, 1531-AG, Paraskevi Attikis-Greece POB 60228.
9
10.1021/es990445r CCC: $19.00
Published on Web 02/04/2000
2000 Am erican Chem ical Society
VOL. 34, NO. 6, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 1 0 0 1

FIGURE 1. Terpenes used in the present investigation. In the text
the identified ozonolysis products from each terpene are nominated
with a letter (A, B, C, S, L, for r-pinene, â-pinene, 3-carene,
sabinene, and limonene, respectively) and a number in bold.
FIGURE 2. Typical evolution of a smog chamber experiment.
in initial mixing ratios from 80 ppbV to 13 ppmV. All terpenes
were obtained from Fluka (Milan, Italy). Stereochemical
characterization of ozonolysis products of terpenes has
recently been published by Griesbaum et al. (28). In the
present study pure stereoisomers terpenes were used merely
due to the fact that these could be obtained in a higher
chemical purity than racemic terpenes. Pinic acid AB1 and
pinonic acid A3 were obtained from Aldrich (Milan, Italy);
pinonaldehyde A7 and limononaldehyde L7 were synthesized
in the laboratory (29). All compounds including terpenes,
solvents, and reagents were used in 99% purity. It is important
to note that, although the initial terpene concentrations are
orders of magnitudes above ambient levels over natural
vegetation (13, 14, 28), the ozone and hydroxyl radical
chemistry occurring in the chambers is largely independent
of concentration (30). Hence, the products formed and the
mechanism of formation are expected to be valid for ambient
conditions. However, since the reactant concentrations will
influence the relative importance of competing reactions
(e.g. reactions of organic species with O2 and H2O compared
to reactions with other organic species), the reaction yields
determined in this investigation do not necessarily give
accurate estimates of ambient yields.
The experiments were carried out in purified air at 740
( 5 Torr, 298 ( 5 K, and at medium humidity (40-55% RH).
The desired terpene mixing ratios were obtained by direct
injection of a few microliters of the liquid terpene into the
reaction chamber (high levels) or into a dilution cell (low
levels).
Ozone was prepared by silent discharge and afterward
introduced into the reaction chamber. To address the
importance of secondary OH evolved during ozonolysis of
terpenes the experiments were carried out with and without
addition of cyclohexane as OH scavenger (31). SOA was
monitored by a condensation nucleus counter. Experimental
details are summarized in Table 1.
Sam ple Collection. SOA was collected on PTFE filters
(0.5 µm Fluoropore membranes from Millipore, Rome, Italy).
Terpenes are known to react rapidly with ozone not only in
the gas-phase but also in the adsorbed state during sampling
(32) thereby forming significant yields of artifact reaction
products on the adsorbent (33, 34). Ozone scrubbers have
been developed for sampling of reactive products in the gas-
phase. However, these tend to discriminate toward large
diameter particles and are therefore not ideal for SOA
sampling (34). To minimize artifacts from ozone during
sampling, a simple approach has been used to draw samples
at the end of each experiment when only low concentrations
of residual ozone was present, typically after 6 h. Possible
artifacts from residual ozone at these concentrations were
addressed in separate experiments by purging PTFE fiber
filters impregnated with terpene with air containing ozone.
No evidence of carboxylic acid formation was observed.
Terpenes are known to undergo rearrangement reactions
on acidic surfaces (35). Thus, the use of glass or quartz fiber
material was avoided. Air, with its content of SOA, was drawn
from the reaction chambers at a rate of 1-2.5 L min-1 through
Limonene (90 ppbv) + O (122 ppbv) + methylcyclohexane (1.5 ppmv)
3
as OH scavenger. Black bullets ) limonene mixing ratio; dotted
line ) particle number concentration; solid line ) ozone mixing
ratio.
PTFE filters (2.5 cm in diameter with a 0.5 µm pore size) for
the duration of 30-200 min. The sampling volume cor-
responded to 40-500 NL. A minimum of two samples was
collected for each experiment. One filter was extracted and
analyzed immediately by gas chromatography-MS (GC-MS).
The other filters were stored at -20 °C in 10 mL sealed glass
vials for later analysis by HPLC-MSn. The filter samplers were
cleaned by sonication in distilled water followed by sonication
in methanol between experiments. Tests with the PTFE filters
proved the absence of particle breakthrough. However, it
cannot be excluded that artifacts in the particle sampling
may have occurred due to size dependent charge distribution
or sedimentation after 6 h in the smog chambers.
Extraction and Analysis. A sample preparation protocol
directed toward GC-MS eletron impact (EI) and chemical
ionization (CI) analysis of carboxylic acids was adapted from
Christoffersen et al. (22). The PTFE filters were sonicated
with 7 mL of dichloromethane at room temperature, the
filters were removed, the solvent was evaporated under a
gentle stream of nitrogen, and the residues were treated with
300 µL 10% BF3-methanol at 50 °C for 30 min in order to
esterify the organic acids. The solution was cooled, 300 µL
distilled water was added, and the esters were extracted with
2 mL of pentane containing 2 ng µL-1 bornyl acetate as
internal reference compound. A second extraction step with
methanol-water (1:1) was applied occasionally to look for
highly polar compounds such as e.g. tricarboxylic acids. No
such compounds were found. Stored filters were analyzed
with a newly developed HPLC-MSn method using atmo-
spheric pressure chemical ionization (APCI) and pneumati-
cally assisted electrospray chemical ionization (ESI), which
have proved efficient for polar terpene oxidation products
(27). Detailed descriptions of the mass spectrometric methods
are given in the Supporting Information.
Monitoring of the terpene concentrations was done by
drawing air samples at the beginning, the end, and oc-
casionally during the experiments onto Tenax tubes through
ozone scrubbers (33, 34). The Tenax tubes were later analyzed
by thermal desorption GC-MS according to Calogirou et al.
(33).
Results and Discussion
A typical evolution of a smog chamber experiment is shown
in Figure 2. Shortly after contact of ozone with terpene a
particle burst of SOA occurs, which reaches a maximum in
number concentration after a few minutes. In the ensuing
period ozone and terpene are consumed, but coagulation of
already existing particles proceeds more rapidly than the
formation of new particles, and as a result, decay curves are
obtained. If wall losses are minor, the mass concentration
of SOA will increase during the decay of the reactants. To
9
1 0 0 2 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 34, NO. 6, 2000

TABLE 1. Experimental Conditions for Smog Chamber Experiments and Resulting Yields of Reaction Products in Organic Aerosol
[HC]o -
[HC]end
ppmV
[O ] -
3
particles
max.,
3 o
,
[O ]end
,
[MCH],
ppm
terpene
ppmV
cm-3
molar yielda of reaction products detected in organic aerosol, %
pinic
norpinic
pinonic
norpinonicd
OH-pinonicd
pinonaldehyde/OH-
pinonaldehyde
2.1/0.51
sum
R-pinene
1.8-0.29
1.9-0.28
0.91-0.005
0.82-0.005
0
16
18300
17200
1.9
0.09
2.1
0.32
1.7
8.7
1.4
3-caric
0.04
nor-3-caric
1.5
3-caronic
0.19
nor-3-caronic
0.86
1.9/0.32
6.2
sum
OH-3-caricd
3-caronaldehyde/OH-
3-caronaldehyde
nab
3.7/0.64
1.4/0.17
nopinone/OH-
pina ketoned
1.1/nab
0.19/0.34
0.02/0.05
sabina ketone/OH-
sabina ketoned
1.3/nab
0.80/0.31
0.08/0.03
3-carene
â-pinene
sabinene
6.5-6.0
0.103-0.029
0.105-0.027
0.64-0.005
0.24-0.036
0.13-0.064
0
0
1.3
nab
41000
40500
1.6
2.9
0.67
pinic
0.13
0.09
trace
norpinic
0.91
4.8
0.21
pinonic
0.01
0.03
nab
5.1
2.6e
17
2.6
trace
0.13
norpinonicd
OH-pinonicd
sum
1.9-1.7
0.046-0.014
0.042-0.016
0.89-0.020
0.11-0.044
0.10-0.059
0
0
1.6
23700
13000
2000
3.9
1.2
0.46
sabinic
0.18
0.11
0.01
norsabinic
0
0
0
trace
0
0
nab
0.14
0.05
5.2e
2.0
0.6
pinic
sum
2.0-1.0
0.75-0.005
0.49-0.009
0.5c- 0.009
0.5c-0.084
20
0
0
nab
38300
13500
8700
2.7
1.1
0.26
0.20
lim onic/
keto-
lim onic
0.06/0.01
0.36/0.04
0.91/0.04
0.30/0.02
0.08
0.03
trace
trace
0.09
1.4
0.64
0.22
lim ononic/
keto-
lim ononic
0.22/trace
0.64/0.01
1.3/trace
0.23/trace
4.2
3.7
1.0
0.5
sum
1.1-0.26
0.40-0.10
0.51-0.08
1.6
0.05/0.03
norlim onic
nor-
OH-lim ononicd/
OH-keto-
lim ononicd
0.36/0.03
0.90/0.08
nab
lim ononaldehyde/
OH-lim onon-
aldehyde
lim ononicd
lim onene
13.3-9.5
3.6-1.9
0.090-0.033
0.080-0.002
3.8-0.130
1.8-0.154
0.12-0.064
0.14-0.094
0
0
0
6.5
nab
140000
40000
18000
0.003
0.01
0.004
trace
0.01
0.09
0.19
0.07
0.60/0.09
1.4
0.91/0.08
3.1
nab
2.3e
0.6e
nab
nab
a
b
c
d
The propagated estim ated error on yields am ounts to less than a factor 2. na ) not analyzed. Malfunctioning ozone analyzer, the value is estim ated based on elapsed tim e with UV light on. Sum of all isom ers.
Not all com pound classes were analyzed. MCH ) m ethyl-cyclohexane, used as OH radical scavenger. OH as prefix denom inates a hydroxy derivate.
e

FIGURE 3. Dicarboxylic acids and oxocarboxylic acids in secondary organic aerosol formed by gas-phase reactions between terpene
and ozone.
sample as much particle mass as possible SOA was collected
at the end each experiment.
cis-pinic acid AB1 (cis-3-carboxy-2,2-dimethylcyclobutyl-
ethanoic acid) from R- and â-pinene; cis-3-caric acid C1 (cis-
2,2-dimethyl-1,3-cyclopropyldiethanoic acid) from 3-carene;
cis-sabinic acid S1 (2-carboxy-1-isopropylcyclopropyletha-
noic acid) from sabinene; and limonic acid L1 (3-isopro-
penylhexanedioic acid) and ketolimonic acid L1′ (3-acetyl-
hexanedioic acid) from limonene.
In the following, data are first presented to support
tentative identifications of nonvolatile polar oxidation prod-
uct in the SOA extracts. Second, semiquantitative results of
the analyses are discussed. Mass spectra of a large number
of oxidation products in SOA have been recorded during this
study. Examples of representative mass spectra are shown
in the Supporting Information together with a detailed
discussion of their interpretation. Three classes of carboxylic
acids were determined (dicarboxylic acids, oxocarboxylic
acids, and hydroxyketocarboxylic acids) together with the
well-known classes of ketones and ketoaldehydes. The
proposed structures of individual compounds are shown in
Figures 3 and 4 together with shorthand names according
to Larsen et al. (37). For clarity, the corresponding IUPAC
names and a letter and a number in bold (Figure 1) are given
in the text first time a new reaction product is mentioned.
Dicarboxylic Acids. For all terpenes the HPLC-MSn
analysis of the SOA extract gave chromatograms with a few
major peaks and a number of minor peaks (Figure 5 and 6).
For the major peaks it was possible to obtain high quality
GC-EI-MS and GC-CI-MS spectra, which supported tentative
structure assignments. The most abundant class of com-
pounds was dicarboxylic acids. The proposed structures are
By comparing mass spectral data and retention times with
those of an authentic standard, cis-pinic acid AB1 was
unambiguously identified as a main oxidation product in
SOA not only from R-pinene and â-pinene but surprisingly
also from sabinene. Even in SOA from 3-carene minor yields
of cis-pinic acid AB1 were found. A number of control
experiments including blank runs with newly prepared PTFE
reaction chambers and new batches of terpene standards
excluded that cross-contamination could be the reason for
these findings. For more than a century, terpenes have been
known to undergo (Wagner-Meerwein) rearrangements of
their carbon skeleton either in solution or when adsorbed
onto active surfaces (35, 38-39). The more strain there is in
the terpene structure the stronger the tendency is to rearrange
(39-41). A recent example of this has been given by Coeur
et al. who observed a significant transformation (72%) of
sabinene into other terpenes after adsorption to steel and
Tenax surfaces (42). Likewise it has been demonstrated that
9
1 0 0 4 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 34, NO. 6, 2000

FIGURE 4. Hydroxycarboxylic acids, ketoaldehydes, hydroxyketoaldehydes, ketones, and hydroxyketones in secondary organic aerosol
formed by gas-phase reactions between terpene and ozone.
intermediate (radical) reaction products of terpenes after an
OH attack can undergo skeleton rearrangements (20, 43, 44),
and it is not unlikely that also the exited Criegee biradical
intermediate products of terpenes after an O3 attack may
undergo rearrangements. Hence, the finding of dicarboxylic
acids in SOA with carbon skeletons that have lost most of
their resemblance with the precursor terpenessurprising as
it may besis perfectly explainable. The four-member ring in
the structure of pinic acid AB1 is considerably less strained
than the three-member rings in the structure of sabinic acid
S1 and 3-caric acid C1 and thus, during rearrangement
reactions, pinic acid AB1 is energetically favored over the
latter two dicarboxylic acids. The formation of pinic acid
AB1 in SOA from sabinene and 3-carene has also been
observed by others (45).
Based on GC-EI and GC-CI mass spectral data 3-caric
acid C1, sabinic acid S1, and limonic acid L1 have previously
been tentatively identified in SOA from 3-carene, sabinene,
and limonene by Glasius et al. (26) using esterification with
BF3/ MeOH. These dicarboxylic acids have also been reported
by Yu et al. (24, 45) using derivatization with O-(2,3,4,5,6-
pentafluorobenzyl)hydroxyamine and N,O-bis(trimethylsi-
lyl)trifluoroactamide. Both methods have their advantages.
A high degree of fragmentation with detailed structural
information is obtained by EI analysis of methyl ester
derivatives. However, the introduction of methyl groups into
the analytes may disconcert their identification by diminish-
ing spectral differences of methylated SOA components and
nonmethylated components containing an extra -CH2-
group in its molecular structure. By introducing the larger
groups O-(2,3,4,5,6-pentafluorobenzyl)oxime and trimeth-
ylsilyl into the analytes such ambiguity is avoided. Yet, the
large groups adsorb most of the energy during the electron
impact ionization, which results in considerably less frag-
mentation of the carbon skeleton. The use of both GC-MS
after methylation of the SOA extracts and direct HPLC-MSn
analysis of the extracts, as in the present investigation, gave
complementary information on the analytes and confirmed
the previous tentative identification of 3-caric acid C1, sabinic
acid S1, and limonic acid L1.
In addition to the C9-dicarboxylic acid less abundant peaks
were present in the chromatograms with mass spectral data
also pointing to C8-dicarboxylic acids. The proposed struc-
tures for these compounds (Figure 3) are cis-norpinic acid
AB2 (cis-3-carboxy-2,2-dimethylcyclobutylmethanoic acid)
from R- and â-pinene, cis-nor-3-caric acid C2 (3-carboxy-
2,2-dimethylcyclopropanylethanoic acid) from 3-carene, cis-
norsabinic acid S2 (2-carboxy-1-isopropylcyclopropylmeth-
anoic acid) from sabinene, and norlimonic acid L2 (2-
isopropenylpentanedioic acid) from limonene. A decar-
boxylated form of pinic acid AB1, namely norpinic acid AB2,
has been tentatively identified in SOA from R-pinene/ O3 in
9
VOL. 34, NO. 6, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 1 0 0 5

In SOA from R-pinene/ O3 small amounts of an isomeric
form of norpinonic acid A4 has previously been detected
(24, 27). In Figure 5 the ESI(-)MS extracted ion chromatogram
at m/ z 169 shows traces of an isomeric form of norpinonic
acid A4. A strong signal from neutral loss of H2O by ESI(-)-
MS2 (and GC-CI) points to an aldo-acid rather than a keto-
acid. Two possible structures fit the mass spectrometric data,
namely pinalic 4-acid (2,2-dimethyl-3-formylcyclobutyl-
ethanoic acid) and pinalic 3-acid A5 (2,2-dimethyl-3-formyl-
methylcyclobutylmethanoic acid). It is difficult to distinguish
between these two isomers on the basis of mass spectral
data alone. In Figure 3 we have shown the structure of pinalic
3-acid A5 as representative of these two isomers. In SOA
from 3-carene/ O3 and limonene/ O3 traces of similar al-
docarboxylic acids were present. They are tentatively identi-
fied as 3-caralic acid C5 (2,2-dimethyl-3-formylmethylcy-
clopropylethanoic acid), limonalic acid L5 (3-formylmethyl-
2-methylenylhexanoic acid), and ketolimonalic acid L5′ (3-
formylmethyl-2-oxohexanoic acid).
Hydroxyketocarboxylic Acids. C10-hydroxyketocarboxylic
acids were determined in SOA from the terpenes with an
endocyclic CdC double bond. They eluted as major peaks in
the HPLC chromatograms 5-7 min before C10-keto-car-
boxylic acid (Figures 4 and 5). In SOA from R-pinene and in
a smaller amount from â-pinene cis-10-hydroxypinonic acid
AB6 (cis-2,2-dimethyl-3-hydroxyacetylcyclobutylethanoic acid)
was identified by comparison of retention times and mass
spectral data with those of a standard (27) produced by
ozonolysis of myrtenol (10-hydroxy-R-pinene). The other
representatives of this class of nonvolatile terpene oxidation
products are tentatively identified as cis-10-hydroxy-3-
caronic acid C6 (cis-2,2-dimethyl-3-(hydroxy-2-oxopropyl)-
cyclopropanylethanoic acid) from 3-carene and 7-hydroxy-
limononic acid L6 (7-hydroxy-3-isopropenyl-6-oxoheptanoic
acid) and 7-hydroxyketolimononic acid L6′ (3-acetyl-7-
hydroxy-6-oxoheptanoic acid) from limonene (Figure 4). The
position of the hydroxy-group in the proposed structures for
the hydroxyketocarboxylic acids from 3-carene and limonene
is based upon similarities with 10-hydroxypinonic acid AB6
in retention times and MS data. Other isomers may also be
possible.
A mechanism for the R-pinene/ O3 reaction has been
proposed by Yu et al. (24) to explain the formation of
hydroxyketoaldehydes such as 10-hydroxypinonaldehyde A8
and 2-hydroxypinonaldehyde in SOA from R-pinene/ O3 (24).
According to this mechanism such compounds may arise by
isomerization of Criegee radicals to form enehydroperoxides
through intramolecular hydrogen migration followed by
subsequent isomerization to hydroxyketoaldehydes. A further
oxidation of these compounds would explain the formation
of hydroxyketocarboxylic acids observed in the present study.
Interestingly, also a number of minor peaks are present in
the HPLC-MS chromatograms with mass spectral data
corresponding to hydroxyketocarboxylic acids (Figures 4 and
5). They may correspond to 2-hydroxy-3-caronic acid (2,2-
dimethyl-3-(2-oxopropyl)cyclopropanylhydroxyethanoic acid)
and 2-hydroxylimononic acid (2-hydroxy-3-isopropenyl-6-
oxoheptanoic acid) deriving from the O3-mechanism sug-
gested above or they may be other isomers deriving from
mechanisms involving secondary OH radical. Traces of a
compound assigned the structure of 10-hydroxypinonic acid
AB6 has also been reported in SOA from R-pinene/ O3 by
Jang and Kamens (25) and Yu et al. (45).
FIGURE 5. HPLC-ESI-MS chromatograms (TIC and extracted ions)
of carboxylic acids in SOA from gas-phase oxidation of r-pinene
with ozone. Black peaks at 30 s derive from postcolumn flow
injection.
other studies based upon its online-APCI-MSn spectral data,
(23), HPLC-MSn data (27), GC-MS (EI and CI) data (25, 26),
and GC-FTIR data (25). Nor-3-caric acid C2 and norsabinic
acid S2 are reported for the first time in the present study
and by Yu et al. (45). Norlimonic acid L2 is reported here for
the first time.
Oxocarboxylic Acids. The second class of abundant
compounds identified in SOA is the oxocarboxylic acids. The
proposed structures of the tentatively identified compounds
are shown in Figure 3 together with their shorthand names.
The most abundant compounds are the C10-ketocarboxylic
acids. They are cis-pinonic acid A3 (cis-3-acetyl-2,2-di-
methylcyclobutylethanoic acid) from R-pinene; cis-3-caronic
acid C3 (2,2-dimethyl-3-(2-oxopropyl)cyclopropanylethanoic
acid) from 3-carene; and limononic acid L3 (3-isopropenyl-
6-oxoheptanoic acid) and ketolimononic acid L3′ (3-acetyl-
6-oxoheptanoic acid) from limonene. The analogue C9-
ketocarboxylic acid were also present in the chromatograms
(Figure 2). The intensities of the peaks were lower than those
of the C8-dicarboxylic acids. The compounds are tentatively
identified as cis-norpinonic A4 acid (cis-3-acetyl-2,2-di-
methylcyclobutylmethanoic acid) from R-pinene, cis-nor-
3-caronic acid C4 (2,2-dimethyl-3-(2-oxopropyl)cyclopro-
panylmethanoic acid) from 3-carene, and norlimononic L4
(2-isopropenyl-5-oxo-hexanoic acid) from limonene. C9-keto-
carboxylic acids are not very important contributors to the
total aerosol load but are mentioned here for completeness.
Pinonic acid A3 has been reported in SOA from smog
chamber experiments by a number of investigators and most
recently also in ambient aerosol over a forested area (17).
Limononic acid L3 and 3-caronic acid C3 have been
tentatively identified in SOA from limonene/ O3 (26) and
3-carene/ O3 (24). Norpinonic acid A4 has been tentatively
identified in SOA from R-pinene/ O3 by Hatakeyama et al.
(18), Yu et al. (24, 45), and Jang and Kamens (25). Nor-3-
caronic acid C4 was tentatively identified in SOA from
3-carene/ O3 by Yu et al. (24, 45), whereas norlimononic acid
L4 is proposed here for the first time.
Nonacidic Com ponents of SOA. Ketones and ketoalde-
hydes are major gaseous products from the reaction of
monoterpenes and ozone (20). Although the analytical
procedures were not optimized for such compounds, HPLC-
APCI(+)MS chromatograms and GC-MS chromatograms
showed the presence of a number of compounds with mass
spectral data corresponding to well-known main oxidation
9
1 0 0 6 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 34, NO. 6, 2000

products of terpenes such as cis-pinonaldehyde A7 (cis-3-
acetyl-2,2-dimethylcyclobutylethanal) from R-pinene, cis-
3-caronaldehyde C7 (cis-2,2-dimethyl-3-(2-oxopropyl)cy-
clopropanylethanal) from 3-carene, pina ketone B9 (6,6-
dimethylbicyclo[3.1.1]heptan-2-one), also known as nopinone,
from â-pinene, sabina ketone S9 (5-isopropylbicyclo[3.1.0]-
hexan-2-one) from sabinene, and limona ketone L9 (4-acetyl-
1-methylcyclohexene), limononaldehyde L7 (3-isopropenyl-
6-oxo-heptanal), and ketolimononaldehyde L7′ (3-acetyl-6-
oxoheptanal) from limonene.
Hydroxyketoaldehydes have previously been tentatively
identified as products from R-pinene/ O3 (24, 45, 47) and
3-carene/ O3 (24, 45). We have recently (27) confirmed the
identification of 10-hydroxy-pinonaldehyde A8 (cis-2,2-
dimethyl-3-hydroxyacetylcyclobutylethanal) from the exactly
matching retention times and mass spectrometric data with
those of the major oxidation product from ozonolysis of
myrtenol (10-hydroxy-R-pinene). Due to the extra hydroxy
group in the molecular structure their equilibrium vapor
pressures are considerably lower than those of the corre-
sponding ketoaldehydes. As such, hydroxyketoaldehydes may
accumulate in the particle phase. The tentatively identified
hydroxyketoaldehydes were cis-10-hydroxypinonaldehyde A8
(cis-2,2-dimethyl-3-hydroxyacetylcyclobutylethanal) from
R-pinene, cis-10-hydroxy-3-caronaldehyde C8 (cis-2,2-di-
methyl-3-(hydroxy-2-oxopropyl)cyclopropanylethanal) from
3-carene, and 7-hydroxylimononaldehyde L8 (7-hydroxy-
3-isopropenyl-6-oxoheptanal) and 7-hydroxyketolimonon-
aldehyde L8′ (3-acetyl-7-hydroxy-6-oxoheptanal) from li-
monene (Figure 4). The position of the hydroxy-group in the
proposed structures for the hydroxyketoaldehydes from
3-carene and limonene is based upon similarities with 10-
hydroxypinonaldehyde A8 in retention times and MS data.
Other isomers may also be possible.
FIGURE 6. HPLC-ESI-MS chromatograms (TIC) of carboxylic acids
in aerosol from ozonolysis of 3-carene (top), limonene (second),
sabinene (third), and â-pinene (bottom).
In the gas-phase low amounts of hydroxy-ketones were
also detected such as 3-hydroxy-pina ketone B10 (3-hydroxy-
6,6-dimethylbicyclo[3.1.1]heptan-2-one), and 3-oxo-pina
ketone (6,6-dimethylbicyclo[3.1.1]heptan-2,3-dione) from
â-pinene and tentatively identified 3-hydroxy-sabina ketone
S10 (3-hydroxy-5-isopropylbicyclo[3.1.0]hexan-2-one) and
3-oxo-sabina ketone (5-isopropylbicyclo[3.1.0]hexan-2,3-
dione). These oxidized forms of pina ketone and sabina
ketone have previously been reported in the gas-phase (45-
47). Only hydroxy-ketones (Figure 4) were found in measur-
able concentrations in SOA.
Binary Diacid Clusters. Hoffmann et al. (23) have recently
hypothesized on formation of binary, tertiary, and higher
clusters of nucleating multifunctional carboxylic acids in the
reactions of terpenes with ozone. Experimental indication
for binary clusters derived from online APCI-MS analysis of
SOA from R-pinene/ O3 and was based upon a signal at m/ z
357 corresponding to the pseudomolecular cluster ion [M1
+ M2 - H]- of pinic acid AB1 and norpinic acid AB2 (Figure
4). Furthermore, offline HPLC-APCI-MS analysis of SOA
sampled on filters gave a peak in the ion-chromatogram at
m/ z 357. MS2 analysis of the ion m/ z 357 revealed fragment
ions corresponding to pinate (m/ z 185) and norpinate (m/ z
171).
In contrast to Hoffman et al. no evidence of such clusters
has been found, neither in the present nor in other studies
of SOA from terpene/ O3 reactions (24-27, 45). A number of
unanswered questions in the study of Hoffmann et al. point
to measurement artifacts as an alternative explanation of
their observations. It is unclear why clusters of a minor
oxidation product (norpinic acid AB2) are observed and not
clusters between the main carboxylic acids (pinonic A3, pinic
AB1, and hydroxypinonic acid AB6). Furthermore a binary
diacid cluster only held together by hydrogen bonds, if
existing, would dissociate in the offline analysis during 2-15
h adsorptive sampling on Tenax, extraction with methanol,
and elution through a reversed phase LC column with a 30
mM ammonium acetate buffered solution. It is well-known
that atmospheric pressure ionization techniques suffer from
inherent artifact formation of adducts and clusters. The
critical step in online APCI analysis of particles is the
evaporation/ desolvation of the analyte to generate significant
gaseous concentrations available for ionization in the corona
discharge region. The APCI is originally developed as an
interface of liquid chromatography with mass spectrometry,
and the evaporation/ desolvation processes have been op-
timized for trace concentrations of analytes in volatile solvents
such as methanol, water, etc. To reduce the artifact formation
of clusters and adducts the chemist typically operate the
HPLC-APCI-MS at low concentration levels and at optimized
source conditions (temperature, sheath gas flow, choice of
solvent). When using the APCI source for online analysis of
particles consisting of bulk concentrations of nonvolatile
analytes evaporation/ desolvation processes are by no means
facilitated and problems with adducts and cluster are
augmented. A demonstration of the strong formation of
clusters of multifunctional carboxylic acids in the APCI source
is shown in Figure 7. The data were obtained by measuring
the ion intensity over a source temperature range of 500-
850 K, after direct injection of dilute methanolic solutions of
mixtures of pinic acid AB1 and pinonic acid A3. The heat of
formation calculated from the Arrhenius plots points to
intermediate strength hydrogen bonds as the governing forces
of these binary clusters A clear demonstration of such an
artifact is shown in Figure 5. Pinic acid AB1 (m/ z 185) and
norpinic acid AB2 (m/ z 169) are well separated, and no signal
for their binary cluster (m/ z 357) in the chromatogram can
be seen. However in the peak obtained by postcolumn flow
injection of the SOA extract (no chromatographic separation)
a clear artifact signal is formed.
Unidentified Com pounds. In the chromatograms a
number of minor peaks were present, to which no chemical
9
VOL. 34, NO. 6, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 1 0 0 7

is already saturated with reaction products. Finally, organic
vapors may condense/ adsorb onto the filter surface contain-
ing the already collected particles. This may change the
composition of SOA and lead to overestimations of yields.
No data are available in the literature, which allows an
estimation of the magnitude of this problem. To overcome
this, denuders may be used to strip the gas phase out of the
sampling flow (17, 23, 25). Possible drawbacks of this solution
is an elevated evaporative loss of semivolatile oxidation
products in SOA and artifact formations of terpene oxidation
products by reaction of residual ozone with gas-phase
compounds collected on the denuder walls similar to the
artifacts demonstrated for adsorptive sampling of terpenes
without ozone scrubbers (32-34). The relatively high pro-
portion of pinic acid AB1 reported in the gas-phase of recent
smog chamber experiments by Jang and Kamens (25) may
have been caused by denuder artifacts.
FIGURE 7. Arrhenius plots for the ion formation by APCI(-). Gray
bullets: pinate [M - H]-; black bullets: pinate-pinic acid cluster
[2M - H]-; gray squares: pinonate [M - H]-; black bullets:
pinonate-pinonic acid cluster [2M - H]-; white triangles: pinate-
pinonic acid cluster [M1 + M2 - H]-.
Quantitative determination of carboxylic acids in SOA is
possible for compounds that have neat standards available.
A comparison of the quantitative results by the LC-MS
method and the GC-MS method for pinonic A3 and pinic
acid AB1 (R-pinene and â-pinene experiments) showed
agreements within 20-35%. In the absence of pure standards
for other terpene oxidation products than pinic acid AB1,
pinonic acid A3, and pinonaldehyde A7, a semiquantitative
approach was taken. Dicarboxylic acids (C9 and C8) and
ketocarboxylic acids (C10 and C9) were quantified as their
methyl-ester derivatives by GC-EI-MS (total ion current, TIC)
using respectively cis-pinic acid AB1 and cis-pinonic acid A3
as surrogate standards. Hydroxyketocarboxylic acids show
low recoveries in the methylation procedure due to losses
during cleanup of the derivatized extracts and could therefore
not be quantified by GC-MS. Instead HPLC-ESI-MS (TIC)
was used with cis-pinonic acid A3 as surrogate. Ketones,
ketoaldehydes, and hydroxy-ketones were quantified by
HPLC-APCI-MS (TIC). Pinonaldehyde A7 and 2-hydroxy-3-
pinanone were used as surrogate standard for ketoaldehydes
and hydroxy-ketones, respectively. Pina ketone was quanti-
fied with authentic standards. The response for a chemical
compound by atmospheric pressure ionization MS is more
dependent on its chemical structure than by GC-EI-MS, thus
the uncertainty in quantification of hydroxy compounds can
be assumed to be higher than the uncertainty for the non-
hydroxylated analogues. Table 1 presents the molar yields of
the quantifiable oxidation products in SOA from the smog
chamber experiments. The uncertainty associated with the
estimated yields comprises uncertainties for a number of
steps in the entire analytical procedure. The most significant
error sources are estimated to be quantification of the
monoterpene starting concentration ((10%), sample volume
determination (( 5%), interference of ozone during sampling
(( 10-20%), particle collection and wall loss (<40%),
extraction loss (<5-10%), derivatization loss (<10%), and
surrogate calibration uncertainty ((20-50%).
FIGURE 8. Desorptive loss of cis-pinic acid (squares), trans-pinic
acid (triangles), and cis-pinonic acid (bullets) from spiked filters
after purging with clean air.
structure could be assigned. They point to carboxylic acids
with a molecular mass of m/ z 172 (not C8-dicarboxylic acids),
m/ z 184 (not corresponding to any identified ketocarboxylic
acid), m/ z 186 (not corresponding to any identified C9-
dicarboxylic acid), m/ z 188, m/ z 202, and m/ z 216. It is not
possible to quantify unknown compounds. Based on the
intensities in the chromatograms the sum of these unidenti-
fied carboxylic acids may represent up to 20-30% of the
mass of polar reaction products in SOA.
Product Yields. Sampling of aerosol is associated with a
number of inherent errors. The relatively high surface-to-
volume ratio of the PTFE bags used as reaction chambers
may lead to loss of particles by coagulation and subsequent
settling on the reactor walls. An upper limit of this loss has
been estimated to be in the order of 30-40% (39), when
sampling after 6 h. Also, organic vapors may be lost to the
reactor walls. Only scarce data is available in the literature
on this topic. Based on measurements of loss of pinonal-
dehyde A7 to the walls of a glass reactor (21) we estimate that
between 10 and 30% of the nonvolatile polar reaction
products may be lost to the walls in the present study. Organic
compounds may also evaporate (desorb) from already
sampled particles. To get a worst case estimate of desorptive
loss during sampling, PTFE filters were impregnated with 10
µg of pinic acid AB1 and pinonic acid A3 and purged at 298
K with large volumes of clean air. As seen in Figure 8, the
dicarboxylic acid remained unaffected on the filters, whereas
the ketocarboxylic acid was gradually lost. After 180 L purging
approximately 60% of pinonic acid A3 was lost. During
sampling of SOA in the present smog chamber experiments
desorptive loss is much lower, since the air passing the filters
It appears in Table 1, that the load of organic compounds
to SOA formed by ozonolysis of terpenes amount to a few
percent of the reacted mass. For each terpene there is a high
variability in the total yield of reaction products, which cannot
be accounted for by measurement error alone, but may be
explained by different reaction conditions such as initial
concentrations of terpene, ozone, and OH scavenger. There
seems to be a trend in the data of the absolute yields to
decrease with scavenging of OH radicals. This is consistent
with current theories for terpene oxidation (20) and secondary
OH formation, which may be responsible for a part of the
observed products in particular the decarboxylated species
such as C8-dicarboxylic acids and C9-ketocarboxylic acids
and C9-hydroxyketocarboxylic acids. There is also a tendency
for the absolute yields to decrease with a decreasing initial
9
1 0 0 8 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 34, NO. 6, 2000

terpene concentration, which is consistent with recent
observations on SOA formation governed by adsorptive
partition processes (4, 5, 9).
TABLE 2. Incorporationa of 18O into Major Oxidation Products
in Aerosol from r-Pinene/O /H 18O
3
2
A remarkable product analogy was observed in SOA from
the five investigated terpenes. For the terpenes with an
endocyclic CdC double bond (R-pinene and 3-carene) the
molar fraction distribution was 0.2-0.4 of C9-dicarboxylic
acids, 0.3-0.5 of C10-ketoaldehydes, 0.2-0.3 of C10-hydroxy-
ketocarboxylic acids, 0.1-0.2 of C10-ketocarboxylic acids, and
0.04-0.06 of C10-hydroxyketoaldehydes. For the terpenes with
an exocyclic CdC double bond (â-pinene, and sabinene) the
molar fraction distribution was 0.5-0.7 of C9-dicarboxylic
acids, 0.05-0.2 of C9-ketones, 0.05-0.15 of C9-hydroxy-
ketones, 0.05-0.1 of C10-hydroxyketocarboxylic acids, and
0.02-0.06 of C8-dicarboxylic acids. Limonene, containing
both an endocyclic and an exocyclic CdC double bond,
showed a product distribution between the two groups, with
an additional group of compounds (molar fraction from 0.01
to 0.02) deriving from oxidation of both double bonds.
To date, only a few studies have been published on the
composition of SOA after terpene/ ozone reactions, which
includes detailed quantitative data on carboxylic acids (25,
45). The experimental conditions used in the investigations
are too different to allow for a meaningful comparison of
results. However, they all point to the conclusion, that a
number of multifunctional carboxylic acids are accumulating
in SOA at low percentage yields (1-5%), which together with
other polar reaction products such as multifunctional car-
bonyls may account for a substantial fraction of the SOA
mass. More quantitative data is needed to investigate the
role of these compounds in aerosol formation processes.
Mechanistic Considerations. Based on the detailed data
on main reaction products and key intermediates, which
has become available over the latest few years (11, 23-27,
45), it has become possible to suggest chemical reaction
mechanistic pathways for aerosol formation. Generally
approved gas-phase mechanisms over Criegee biradical
intermediates lead directly to C10-ketoaldehydes and C10-
ketocarboxylic acids such as e.g. pinonaldehyde A7 and
pinonic acid A3 from R-pinene. Pathways leading to C9-
dicarboxylic acids such as pinic acid AB1 are less evident
and necessitate a number of rearrangement reactions of the
Criegee biradicals and subsequent oxidation either in the
gas phase (24, 25, 45) or the aerosol phase (25). The
hydroperoxide channel of the excited C9-Criegee intermediate
as proposed by Winterhalter et al. (48) is the only pure gas-
phase pathway proposed to data that can lead to formation
of C9-dicarboxylic acids. However the experimental observa-
tion of key intermediates is still lacking. The present
observations of significant yields of C10-hydroxyketoalde-
hydes and C10-hydroxyketocarboxylic acids may be taken as
an evidence of Criegee biradical rearrangements. Although
it is out of the scope of the present investigation to discuss
formation mechanisms, an interesting observation of mecha-
nistic relevance was done during some of the experiments.
When H218O was added to the smog chambers at a 5% relative
humidity after drying out all nonlabeled water (H216O)
precisely one 18O atom was incorporated into the molecular
structure of dicarboxylic acids, oxocarboxylic acids, and
hydroxyketocarboxylic acids at fractions between 10 and 20%,
but not in significant amounts (2-3%), into the molecular
structure of ketoaldehydes or hydroxyketoaldehydes (Table
2). When secondary OH radicals were not scavenged the
fraction of 18O-analogues increased, possibly due to a OH-
18OH shift reaction (H218O + OH T H2O + 18OH). The
incorporation of 18O was observed in the MS2 daughter scans
of the quasi-molecular ion by fragments from neutral loss
of H2O, H218O, CO2, and OdCd18O. An example is given in
Figure 9 for [M - H]- ) 187 of 18O-pinic acid from the
R-pinene/ O3/ H218O reaction. Further experiments with H218O
secondary OH radicals
scavenged with
methyl-c-hexane
no OH
scavenger
pinic acid
pinonic acid
10-hydroxypinonic acid
pinonaldehyde
21%
9%
19%
2%
34%
18%
28%
3%
a
Expressed as the fraction of the oxidation product containing one
18O m olecule relative to its total yield, YM+2/(YM+YM+2).
FIGURE 9. MS2 daughter scan of the quasi-molecular ion [M - H]-
) 187 of 18O-pinic acid from the r-pinene/O /H 18O reaction. Note
3
2
the fragments from neutral loss of H O, H 18O, CO , and OdCd18O.
2
2
2
are in progress in our laboratory. These preliminary results
indicate that mechanistic pathways for carboxylic acid
formation through ozone/ terpene reactions may include a
reaction step with H2O addition and that water vapor may
play an important role in SOA formation, not only physically
but also chemically. This factor has been neglected in the
experimental setup of SOA experiments performed till date
and should be investigated further.
Acknowledgments
The financial support of the European Commission during
the Ph. D. projects of M.G. and A.C. and during the Post Doc.
work of M.L. is gratefully acknowledged. The authors are
indebted to B. Nicollin and C. Ottobrini for their skilful
technical assistance in preparing and executing the smog
chamber experiment and to M. Duane for the mass spec-
trometric analysis. The present work has been carried out as
part of the EU funded NUCVOC project (Nucleation Processes
from Oxidation of Biogenic Volatile Organic Compounds)
under contract no. ENV4-CT97-0391.
Supporting Information Available
Mass spectral data for tentative identification of polar
oxidation products in secondary aerosol from the gas-phase
reaction of terpene and ozone. This material is available free
of charge via the Internet at http:/ / pubs.acs.org.
Literature Cited
(1) Atmos Biog. Hydrocarbons; Bufalini, J. J., Arnts, R. R., Eds.; Ann
Arbor Science: Ann Arbor, MI, 1981; Vol. 2.
(2) Yokouchi, Y.; Ambe Y. Atmos. Environ. 1985, 19, 1271-1276.
9
VOL. 34, NO. 6, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 1 0 0 9

(3) Paulson, S. E.; Pandis, S. N.; Baltensperger, U.; Seinfeld, J. H.;
Flagan, R.; Palen, E. J.; Allen, D. T.; Giger, W.; Portmann, A. J.
Aerosol. Sci. 1990, 21, 245-248.
EUROTRAC Symposium 98′. Garmisch-Partenkirchen 23-27
March 1998; Borrel, P. M., Borrel, B., Eds.; Witpress: Southamp-
ton, 1999; pp 4361-4364.
(4) Pankow, J. F. Atmos. Environ. 1994, 28, 189-193.
(5) Pandis, S. N.; Harley, R. A.; Cass, G. R.; Seinfeld, J. H. Atmos.
Environ. 1992, 26A, 2269-2282.
(6) Grosjean, D. Atmos. Environ. 1992, 26A, 953-965.
(7) Palen, E. J.; Allen, D. T.; Pandis, S. N.; Paulson, S. E.; Seinfeld,
J. H. Atmos. Environ. 1992, 26A, 1239-1251.
(27) Glasius, M.; Duane, M.; Larsen, B. R. J. Chromatogr. 1998, 833,
121-135.
(28) Griesbaum, K., Miclaus, V.; Jung, I., C. Environ. Sci. Technol.
1998, 32, 647-649.
(29) Glasius, M.; Calogirou, A.; Jensen, N. R.; Hjorth, J.; Nielsen, C.
J. Int. J. Chem. Kinet. 1997, 29, 527-536.
(30) Atkinson, R. J. Phys. Chem. Ref. Data 1997, 26, 215-290.
(31) Chew, A. A.; Atkinson, R. J. Geophys. Res. 1996, 101, 28649-
28653.
(32) Helmig, D. Atmos. Environ. 1997, 31, 3635-3651.
(33) Calogirou, A.; Larsen, B. R.; Brussol, C.; Duane, M.; Kotzias, D.
Anal. Chem. 1996, 68, 1499-1506.
(34) Calogirou, A.; Duane, M.; Kotzias, D.; Lahaniati, M.; Larsen, B.
R. Atmos. Environ. 1997, 31, 2741-2751.
(8) Zhang, S.-H.; Shaw, M.; Seinfeld, J. H.; Flagan, R. J. Geophys.
Res. 1992, 97, 20717-20729.
(9) Odum, J. R.; Hoffmann, T.; Bowman, F.; Collins, D.; Flagan, R.
C.; Seinfeld, J. H. Environ. Sci. Technol. 1996, 30, 2580-2585.
(10) Lahaniati, M.; Nanos, C.; Konidari, C.; Nicollin, B.; Kotzias, D.
In Physico-Chemical Behaviour of Atmospheric Pollution. Pro-
ceedings of the 7th European Symposium, Venice, Italy, 2-4
October; Larsen, B. R., Versino, B., Angeletti G., Eds.; European
Commission: Luxembourg, EUR 17482 EN. 1997; pp 542-546.
(11) Hoffmann, T.; Odum, J. R.; Bowman, F.; Collins, D.; Klockow,
D.; Flagan, R. C.; Seinfeld, J. H. J. Atmos. Chem. 1997, 26, 189-
222.
(35) Hogeveen, A.; von Kruchten, V. Top. Curr. Chem. 1979, 80, 89-
94.
(36) Van Dingenen, R.; Jensen, N- R.; Hjorth, J.; Raes, F. J. Atmos.
Chem. 1994, 18, 211.
(12) Went, F. W. Nature 1960, 187, 641-643.
(13) Fehsenfeld, F.; Calvert, J.; Fall, R.; Goldan, P.; Guenther, A.;
Hewitt, N. C.; Lamb, B.; Liu, S.; Trainer, M.; Westberg, H.;
Zimmerman, P. Global Biogeochem. Cycles 1992, 6, 389-430.
(14) Guenther, A.; Hewitt, N. C.; Erickson, D.; Fall, R.; Geron, C.;
Graedel, T.; Harley, P.; Klinger, L.; Lerdau, M.; Mckay, W. A.;
Pierce, T.; Scholes, B.; Steinbrecher, R.; Tallamraju, R.; Taylor,
J.; Zimmerman, P. J. Geophys. Res. 1995, 100, D5, 8873-8892.
(15) Novakov, T.; Penner, J. E. Nature 1993, 365, 823-826.
(16) Andreae, M., O.; Crutzen P. J. Science 1997, 276, 1052-1058.
(17) Kavouras, I. G.; Mihalopoulos, N.; Stephanou, E. G. Nature 1998,
395, 683-686.
(37) Larsen, B. R.; Lahaniati, M.; Calogirou, A.; Kotzias, D. Chemo-
sphere 1998, 37, 1207-1220.
(38) Enklaar, M. C. J. Recl. Trav. Chim. Pays-Bas 1908, 27, 423-434.
(39) Chemistry of Terpenes and Terpenoids; Newman, A. A. E.;
Academic Press: London, 1972.
(40) Rothweiler, H.; Wager, P. A.; Schlatter, C. Atmos. Environ. 1991,
25, 231-239.
(41) Cao, X. L.; Hewitt, C. N. Chemosphere 1993, 27, 695-703.
(42) Coeur, C.; Jacob, V.; Denis, I.; Foster, P. J. Chromatogr. A 1997,
786, 185-187.
(43) Eberhard, J.; Mu¨ lle, C. L.; Stocker, D. W.; Kerr, J. A. Environ. Sci.
Technol. 1995, 29, 232-241.
(18) Hatakeyama, S.; Izumi, K.; Fukuyama, T.; Akimoto, H. J. Geophys.
Res. 1989, 94, 13013-13024.
(19) Hakola, H.; Arey, J.; Aschmann, S.; Atkinson, R. J. Atmos. Chem.
1994, 18, 75-102.
(20) Calogirou, A.; Larsen, B. R.; Kotzias, D. Atmos. Environ. 1999,
33, 1352-2310.
(21) Hallquist, M.; Wa¨ngberg, I.; Ljungstro¨m, E. Environ. Sci. Technol.
1997, 31, 3166-3172.
(22) Christoffersen, T. S.; Hjorth, J.; Horie, O.; Jensen, N. R.; Kotzias,
D.; Molander, L. L.; Neeb, P.; Ruppert, L.; Winterhalter, R.;
Virkkula, A.; Wirtz, K.; Larsen, B. R. Atmos. Environ. 1998, 32,
1657-1661.
(44) Kwok, E. S. C.; Atkinson, R.; Arey, J. Environ. Sci. Technol. 1996,
30, 1048-1052.
(45) Yu, J.; Cocker, D. R., III; Griffin, R. J.; Flagan, R. C.; Seinfeld, J.
H. J. Atmos. Chem. 1999, 34, 207-258.
(46) Jay, K.; Stieglitz, L. In Proceedings of air pollution and ecosystems;
Grenoble, France, 18-22 May; 1987; pp 91-95.
(47) Hull, L. A. In Atmos Biog. Hydrocarbons; Bufalini, J. J., Arnts, R.
R., Eds.; Ann Arbor Science: Ann Arbor, MI, 1981; Vol 2, pp
161-186.
(48) Winterhalter, R.; Neeb, P.; Grossmann, Kollof, A.; Horie, O.;
Moortgat, G. J. Atmos. Chem. 2000, 35, 165-197.
(23) Hoffmann, T.; Bandur, R.; Marggraf, U.; Linscheid, M. J. Geophys.
Res. 1998, 103, 25569-25578.
(24) Yu, J.; Flagan, R. C.; Seinfeld, J. H. Environ. Sci. Technol. 1998,
32, 2357-2370.
(25) Jang, M.; Kamens, R. M. Atmos. Environ. 1999, 33, 459-474.
(26) Glasius, M.; Lahaniati, M.; Calogirou, A.; Di Bella, D.; Jensen,
N. R.; Hjorth, J.; Kotzias D.; Larsen, B. R. In Proceedings of
Received for review April 21, 1999. Revised manuscript re-
ceived November 1, 1999. Accepted December 6, 1999.
ES990445R
9
1 0 1 0 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 34, NO. 6, 2000