Aerosol formation in the cyclohexene-ozone system

Article abstract of DOI:10.1021/es001180f

Atmospheric oxidation of certain VOCs can yield products with low vapor pressure resulting in the partitioning of the products into the aerosol phase, producing secondary organic aerosol. As cyclohexene is known to produce aerosol upon photooxidation, it was used as a model system for the molecular identification of gas- and aerosol-phase products via derivatization and MS detection. Adipic acid, hydroxyglutaric acid, and hydroxyadipic acid were among the compounds identified in the aerosol phase. Pentanal was the predominant product in the gas phase at a molar yield of 17%. Possible explanations for the presence of relatively high vapor pressure compounds in the aerosol phase are presented.

Full text of DOI:10.1021/es001180f

Environ. Sci. Technol. 2000, 34, 4894-4901
Aerosol Formation in the
TABLE 1. Experimental Conditions
cyclohexene
O (initial) (initial) (final)a hexene
Cyclohexene-Ozone System
∆cyclo-
[ppb] [µg/m3] [%]
Mob
yieldc
3
experiment
[ppb]
[ppb]
[ppb]
M . K A L B E R E R , J . YU , D . R . C O C K E R ,
R . C . F L A G A N , A N D J . H . S E I N F E L D *
5/13/99a
5/13/99b
5/17/99a
5/17/99b
5/19/99a
5/19/99b
539
388
634
415
203
131
87
186
120
48
1.3
2.5
2.8
3.2
4.2
0d
130
85
183
117
44
43
36
126
43
9.6
12.3
20.0
10.7
9.3
Department of Chemical Engineering and Environmental
Engineering Science, California Institute of Technology,
Pasadena, California 91125
14
1011
137
137
36
7.6
a
b
Concentration when denuder/filter sam pling began. ∆Mo ) based
on SEMS m easurem ents, assum ed F ) 1.4 g/cm 3, see text. Yield )
∆Mo/∆cyclohexene. Below detection lim it.
c
Atmospheric oxidation of certain volatile organic compounds
can lead to products of sufficiently low vapor pressure
that the products partition into the aerosol phase, producing
what is termed secondary organic aerosol. Cyclohexene
has long been known to produce aerosol upon photooxidation.
In this study, we obtain molecular product information
especially for those species that partition between the gas
and particle phases. Compounds identified in the aerosol
phase include hydroxyglutaric acid, adipic acid, and
hydroxyadipic acid. In accord with previous studies,
pentanal is found to be the predominant identified gas-
phase products, at a molar yield of 17%. Possible explana-
tions for the presence of relatively high vapor pressure
compounds in the aerosol phase are advanced.
d
The goal of the present paper is to revisit the cyclohexene-
ozone system. Cyclohexene-ozone has been selected as a
model system because it has been intensively studied, gas-
phase products are well-characterized, and a reasonable
amount is known about reaction mechanisms in the system.
Although cyclohexene is not likely to be an important
precursor for secondary organic aerosol in the ambient
atmosphere, the cyclohexene-ozone system is highly relevant
for its use as a model system that is relatively well understood.
We seek molecular identification of gas- and aerosol-phase
products, using recently developed methods of derivatization
and mass spectrometric detection. Such data will, in the
future, serve as a basis for evaluating the ability to predict
SOA formation from first principles; that is, if all the
semivolatile oxidation products of a parent hydrocarbon were
known, together with their physical properties, such as their
vapor pressures, then the yield of SOA from that parent
molecule could be predicted for any given set of conditions.
Introduction
Volatile organic compounds, when photooxidized in the
atmosphere, can lead to products that have sufficiently low
vapor pressure to partition between the gas and aerosol
phases. The fraction of organic particulate matter arising
from this path is referred to as secondary organic aerosol
(SOA). Parent organic compounds that, upon oxidation, lead
to SOA generally contain six or more carbon atoms. While
overall yields of SOA have been measured in the laboratory,
investigation of aerosol composition at the molecular level
have only recently begun to resolve an appreciable fraction
of SOA. Molecular identification of SOA is difficult because
atmospheric oxidation pathways of g C6 compounds lead to
a relatively large number of products that are generally highly
polar, complicating their analysis. Cyclohexene, a cyclic
alkene that is known to produce aerosol upon oxidation,
was, in fact, one of the first organic compounds investigated
for its aerosol-forming potential. Hatakeyama et al. (1)
identified pentanedial, 5-oxopentanoic acid, glutaric acid,
hexanedial, 6-oxohexanoic acid, and adipic acid as SOA
products of cyclohexene oxidation. Gaseous compounds
measured included formic acid, pentanal, pentanedial, and
hexanedial. Since no actual aerosol measurements were
performed in that study, there is no information as to what
fraction of the total particle mass was identified. In a later
study, Izumi et al. (2) measured the aerosol yield of
cyclohexene in the presence of ozone and OH radicals in the
cyclohexene concentration range between 0.5 and 5 ppm
but determined only the total carbon and nitrogen content
of the aerosol particles. Grosjean et al. (3) measured gas-
phase products of the cyclohexene-ozone reaction, in the
presence of an OH scavenger (cyclohexane), and found
pentanal as a principal oxidation product.
Experimental Section
Sm og Cham ber Experim ents. Most of the experimental
procedures associated with the Caltech outdoor smog
chamber have been described previously (4-6), and only a
brief summary is given here. Reactant concentrations and
general experimental conditions for the cyclohexene-O3
system are given in Table 1. Experiments were performed in
the dark in two outdoor Teflon chambers of about 22 m3
volume each. The temperature for all experiments was
maintained at 298 K ((2 K). Before the reactants were
introduced into the chambers, (NH4)2SO4 seed aerosol, of
mean diameter of approximately 100 nm, was injected, at a
number concentration of about 104 cm-3. Particle number
and size measurements were performed with a differential
mobility analyzer (Model 3071, TSI, St. Paul, MN) and a
condensation nucleus counter (Model 3760, TSI, St. Paul,
MN). To prevent OH oxidation by OH generated in alkene-
O3 reactions, CO was added as an OH scavenger (see for
example ref 7), at levels (268 to 1040 ppm) sufficient to ensure
that >95% of any OH generated was scavenged. (The current
experiments were not designed to measure OH yield from
the cyclohexene-O3 reaction.) Samples for gas- and particle-
phase analysis were taken after the hydrocarbon was es-
sentially consumed. The mixing ratio of cyclohexene was
measured with a Hewlett-Packard 5890 GC/ FID. All experi-
ments were carried out under dry conditions (RH < 5%), so
the dependence of product levels on aerosol water was not
addressed.
Gas and Particle Sam pling. Since many reaction products
are present in both gas and particle phases, the sampling
* Corresponding author phone: (626)395-4635; fax: (626)796-2591;
e-mail: seinfeld@its.caltech.edu.
9
4 8 9 4 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 34, NO. 23, 2000
10.1021/es001180f CCC: $19.00
2000 Am erican Chem ical Society
Published on Web 10/11/2000

and chemical ionization mode. The gas used in the CI mode
was methane. Samples were separated in a RTX-5MS column
(Restek, Bellefonte, PA).
TABLE 2. Denuder Collection Efficiencies for Individual
Cyclohexene/Ozone Reaction Products
reaction product
denuder collection efficiencya
Most compounds identified were quantified, either with
an authentic standard or with surrogate compounds of similar
structure. Calibration mixtures that included all standards
were produced at five concentrations in the range of 10-50
ng µL-1. Compound specific ions or, in the case of surrogates,
mass fragments typical for the functional group were used
for quantification of the different compounds. Surrogates
were chosen to have the same type and number of functional
groups and the same carbon number, if possible. Since the
compounds with surrogates were quantified with mass
fragments typical for the functional groups, the number of
functional groups was important for selection of a surrogate
compound.
Recovery experiments to account for losses during
sampling were performed by spiking standard mixtures with
known amounts of all standard compounds on the denuders
and filters. The denuder collection efficiency, defined by the
amount of a given compound collected by the two denuders
preceding the filter, is shown for each compound in the
cyclohexene-O3 system in Table 2. The collection efficiency
was > 91% for all compounds, except for pentanal. Because
it is expected that pentanal should exist exclusively in the
gas-phase, the amount of pentanal found on the filter and
on the denuder behind the filter was considered to be part
of the gas-phase fraction. Because analytes can be lost to
extraction vessels and by evaporation during rotary evapora-
tion, “recovery” refers to the fraction of a compound that is
recovered at the end of the treatment process.
oxalic acid
m alonic acid
succinic acid
adipic acid
0.95 ( 0.02
0.98 ( 0.01
0.96 ( 0.02
0.98 ( 0.01
0.99 ( 0.01
0.91 ( 0.08b
0.99 ( 0.01
0.95 ( 0.02
0.93 ( 0.02
0.98 ( 0.03
0.99 ( 0.01
0.98 ( 0.01
0.99 ( 0.01
0.98 ( 0.02
0.92 ( 0.06
0.77 ( 0.18
glutaric acid
4-hydroxy-1-butanal
hydroxypentanoic acid
hydroxyglutaric acid
hydroxyadipic acid
6-oxohexanoic acid
5-oxopentanoic acid
4-oxobutanoic acid
1,6-hexanedial
1,5-peritanedial
1,4-butanedial
pentanal
a
Denuder collection efficiencies for individual products are calculated
b
using six sets of sam ples. Only five data points.
system consisted of a series of two annular denuders
(University Research Glassware, Chapel Hill, NC) to remove
the gaseous reaction products, followed by a Teflon-coated
quartz fiber filter (Pallflex, Putnam, CT), which collected all
particles. The first denuder is 40 cm long, and the second is
20 cm long. Both denuders consist of 5 annular channels
with 2 mm spacing between channels. An additional denuder
was placed behind the filter to trap compounds that might
volatilize from the filter during sampling. The gaseous
concentration of a compound was calculated from the sum
of the amounts found in the first two denuders, and the
particle phase concentration was determined from the
amounts found on the filter and the third denuder. Samples
were taken at a flow rate of 25 L min-1 for 1 h resulting in
a total sampling volume of 1.5 m3. We evaluated particle loss
to the denuders and found it to be less than 5% by mass.
Regarding artifacts from blowing off a compound from a
denuder to the gas phase, our experiments showed that two
denuders retain more than 99.4% of all compounds with the
exception of pentanedial, for which the retention was 95%.
The composition of secondary aerosols may include artifacts
from filter absorption of gas-phase species that evade denuder
collection. The contribution from this effect can be estimated
for 1,4-butanedial, the most volatile product detected in the
aerosol phase. The first denuder has a collection efficiency
for 1,4-butanedial of 92% (Table 2); the two denuders then
have a combined collection efficiency of 99.4% for this
compound. If the 0.6% of gaseous 1,4-butanedial that escaped
collection on the denuders was retained on the filter, this
amounted to 0.6% × 0.45% ) 0.0027% of the 1,4-butanedial
gas-phase yield. The measured aerosol-phase yield of 1,4-
butanedial was 0.05% (to be presented subsequently). Thus,
only 5% (0.0027/ 0.05) of the 1,4-butanedial collected on the
filter can be attributed to artifact. Further details about
denuder coating and preparation are given by Yu et al. (4).
Results and Discussion
Aerosol Yield. Figure 1 shows cyclohexene and total particle
mass concentrations for experiment 5/ 13/ 99a; this behavior
is typical of all experiments. Cyclohexene was almost
completely consumed within the first 2 h, and the organic
particle mass increased correspondingly. Typical aerosol size
distributions before the start of the experiment and after
particle growth ceased are shown in the insert in Figure 1.
The volume of organic aerosol material produced, ∆Mo, was
determined from the two size distributions, after accounting
for wall losses of particles.
For a wide variety of parent VOCs, the fractional aerosol
yield by mass that is Y as a function of ∆Mo, can be fit by
assuming two empirical oxidation products (5, 8). Fractional
aerosol yields for the six cyclohexene-O3 experiments are
given in Table 1. Parameters (5, 8) of the two-component
best fit correlation to these data are R1 ) 0.36, R2 ) 0.32, Kom,1
) 0.009 m3 µg-1, and Kom,2 ) 0.001 m3 µg-1
.
Identification of Products. Identified products in the
cyclohexene-ozone system are given in Table 3. In the mass
spectrometer, derivatized compounds fragment in a char-
acteristic way, as described by Yu et al. (6) and LeLacheur
et al. (9). Compounds with a carbonyl functional group are
identified by the mass m/ z 181, and those with a carboxylic
or hydroxy group show a typical mass fragment at m/ z 73
and m/ z 75. In addition, the fragments at m/ z M - 15, M -
73, M - 89, and M - 117 (M - 117 is a characteristic ion only
for carboxyl groups, not for hydroxyl groups), which denote
a loss of [CH3], [Si(CH3)3], [OSi(CH3)3], and [C(O)OSi(CH3)3]
from the derivatized molecule, respectively, were used in
identification. In the CI mode the fragment M - 89 was often
stronger compared to EI mode spectra, where M - 117 was
often prominent. Also the molecular ion and the adduct ion,
M + 1 and M + 73, were often detected in the CI mode
spectra. The fragment m/ z 147 was always associated with
compounds with two carboxyl or hydroxyl groups, and m/ z
117 (C(O)OSi(CH)3) was prominent in spectra of compounds
with only one carboxylic group. For compounds with carbonyl
Sam ple Treatm ent and Analysis. Following elution of
the denuders with a dichloromethane/ acetonitrile/ hexane
mixture and extraction of the filter in a Soxhlet apparatus
(dichloromethane/ acetonitrile mixture), the samples were
derivatized with PFBHA (O-(2,3,4,5,6-pentafluorobenzyl)-
hydroxyamine) and BSTFA (N,O-bis(trimethylsilyl)trifluo-
roacetamide). The derivatization reactions convert polar into
less polar compounds, i.e., carbonyl groups react with PFBHA
to form oxime derivatives; carboxyl and hydroxyl groups react
with BSTFA to form trimethylsilyl derivatives. Sample analysis
was performed with a Varian Star 3400/ Saturn 2000 gas
chromatograph/ mass spectrometer in electron ionization
9
VOL. 34, NO. 23, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 4 8 9 5

FIGURE 1. Cyclohexene and total aerosol mass concentration ∆Mo as a function of time. As the cyclohexene concentration decreases
with time, the organic aerosol mass increases correspondingly. The organic aerosol mass, ∆Mo, is calculated from the difference of the
aerosol size distributions at the beginning and end of the reaction, as shown in the insert.
FIGURE 2. Chromatogram of a filter sample: (a) m/z 181 chromatogram showing all carbonyl compounds identified and (b) m/z 73+75
chromatogram showing the compounds with a carboxylic and/or hydroxy functional group.
groups, the mass fragment M - 197, which results from a
loss of [OCH2C6F5] from the derivatized molecule, was
generally used for identification in the EI mode mass
spectrum along with the molecular ion in the CI mode
spectrum.
Figure 2a shows the m/ z 181 ion chromatogram of a filter
sample indicating all nine compounds identified in the
cyclohexene-O3 system with a carbonyl group, and Figure
2b shows the m/ z 73+75 ion chromatogram of the same
sample with all 12 compounds identified containing a
9
4 8 9 6 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 34, NO. 23, 2000

TABLE 3. Chemical Structures of Products of the Cyclohexene-Ozone Reaction
a
Molecular weight for derivative is for full derivatization of all the possible functional groups in the m olecule.
carboxylic and/ or hydroxyl group. In total, 17 compounds
were identified, for eight of which an authentic standard was
available, the others identified tentatively by their mass
spectra (see Table 4).
(M + 1) of the two compounds (i.e., m/ z 477 for the 1,4-
butanedial derivative and m/ z 505 for the 1,6-hexanedial
derivative). The retention time supports the identification,
i.e., 0.98 min earlier than that of 1,5-pentanedial for 1,4-
butanedial and 0.9 min later for 1,6-hexanedial. 1,5-Pen-
tanedial was quantified with compound- specific ions,
whereas 1,4-butanedial and 1,6-hexanedial were quantified
using the carbonyl specific m/ z 181 (see Table 4).
Three dialdehydes, 1,4-butanedial, 1,5-pentanedial, and
1,6-hexanedial, were found in both the particle and gas
phases. Of these, an authentic standard was available only
for 1,5-pentanedial; the two others were tentatively identified
by comparing their mass spectra with that of 1,5-pentanedial.
The EI spectra give compound specific information, i.e., for
carbonyl compounds a strong peak at m/ z M - 197. The
spectra of the compounds identified as 1,4-butanedial and
1,6-hexanedial showed prominent peaks at m/ z 279 and 307,
respectively, which are -14 and +14 mass units (-CH2-
group) compared to the M - 197 peak of 1,5-pentanedial.
In addition, the CI spectra showed the molecular ion peaks
Three oxoalkanoic acids, 4-oxobutanoic acid, 5-oxopen-
tanoic acid, and 6-oxo hexanoic acid, were identified in both
the gas and particle phases. An authentic standard was
available only for 4-oxobutanoic acid; however, the two other
compounds could be identified, similar to the dialdehydes,
by comparing their mass spectra with that of 4-oxobutanoic
acid. Molecular peaks as well as characteristic fragments were
found. Again the M - 197 fragments typical for carbonyls
9
VOL. 34, NO. 23, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 4 8 9 7

TABLE 4. Chemical Compounds Used as Standards or Surrogates for Quantification
compds found in samples
standard or surrogates
quantification ions retention time [min] density [g/cm3] surrogate for density
oxalic acid
oxalic acid
219
233
247
261
275
73,75,181
73,75
73,75
73,75
354,188,172
73+75+181
73+75+181
181
309,293
181
8.533
9.931
1.9
m alonic acid
succinic acid
glutaric acid
adipic acid
4-hydroxy-1-butanal
hydroxypentanoic acid 3-OH-3-m e-glutaric acid
hydroxyglutaric acid
hydroxyadipic acid
4-oxobutanoic acid
5-oxopentanoic acid
6-oxohexanoic acid
1,4-butanedial
1,5-pentanedial
1,6-hexanedial
pentanal
m alonic acid
succinic acid
glutaric acid
adipic acid
1.619
1.572
1.424
1.36
0.817
0.9391
1.424
1.36
11.082
12.348
13.748
14.983
11.390
14.716
15.817
16.014
17.049
18.149
20.284
21.166
21.149
11.698
4-oxobutanic acid
butanal
pentanoic acid
glutaric acid
adipic acid
3-OH-3-m e-glutaric acid
3-OH-3-m e-glutaric acid
4-oxobutanoic acid
4-oxopentanoic acid
5-oxohexanoic acid
1,5-pentanedial
1,5-pentanedial
1,5-pentanedial
pentanal
0.9391
0.93
0.817
0.8095
0.8095
0.8095
pentanoic acid
butanal
pentanal
hexanal
181
were prominent for 5-oxopentanoic acid and 6-oxohexanoic
acid at m/ z + 14 and m/ z +28, compared to the spectrum
of 4-oxobutanoic acid. Fragments also characteristic for the
carboxyl group (i.e., M - 15 and M - 89) were found in both
compounds. Again, the retention time supports the iden-
tification as 5-oxopentanoic acid and 6-oxohexanoic acid
(see Table 4). For quantification, m/ z 73+75+181 using the
surrogates 4-oxopentanoic acid and 5-oxohexanoic acid,
respectively, were used for 5-oxopentanoic acid and 6-oxo-
hexanoic acid. 4-Oxobutanoic acid was quantified with the
compound-specific ions.
The peak at 15.05 min was characterized by the presence
of m/ z 181 and m/ z 73+75, indicating that the compound
has a carbonyl as well as a carboxyl or hydroxyl group. The
CI spectrum shows weak peaks at m/ z 356 and m/ z 340 and
the base peak at m/ z 266, which were assigned to M + 1, M
- 15, and M - 89, respectively, of 4-hydroxy-1-butanal. In
the EI spectrum the fragment m/ z 158 (M -197) strengthened
the identification of this peak as 4-hydroxy-1-butanal. For
quantification, the surrogate 4-oxobutanoic acid was used.
Pentanal, found only in the gas phase, was identified and
quantified with an authentic standard.
All five dicarboxylic acids identified, oxalic acid, malonic
acid, succinic acid, glutaric acid, and adipic acid, were verified
with authentic standards and were quantified with com-
pound-specific ions (Table 4).
At a retention time of 15.96 min, 5-oxopentyl formate was
tentatively identified. In the CI spectrum the molecule ion
(m/ z 325) was prominent with 16% relative intensity. The
main peak in the CI spectrum was m/ z 280 (M - 45),
corresponding to the loss of formate from the molecule. In
addition, the fragment M - 197, prominent in the CI
spectrum, and m/ z 181, especially strong in the EI spectrum,
both typical for carbonyls, supported the identification.
5-Oxopentyl formate was the only compound identified that
was not quantified, since no appropriate standard was
available.
Four different compounds with a hydroxyl functional
group were tentatively identified: hydroxypentanoic acid,
hydroxyglutartic acid, hydroxyadipic acid, and 4-hydroxy-
1-butanal. The compound occurring in the chromatogram
at 11.04 min has been tentatively identified as hydroxypen-
tanoic acid. In the CI spectrum the molecular ion, m/ z 262,
was visible as well as the adduct M + 73, which is typical for
compounds with a derivatized carboxyl or hydroxyl group.
In the EI mode the spectrum showed a strong peak at mass
m/ z 247, which is the typical M - 15 fragment observed for
all carboxyl and hydroxyl compounds, assuming m/ z 262 is
the molecular ion as found under CI conditions. This
molecular mass can be attributed theoretically to three
different compounds in this context, i.e., succinic acid,
hexanediol, and hydroxypentanoic acid. Standards were
available for the first two compounds; however, their
retention times did not match with the peak at 11.04 min.
Glutaric acid was used as a surrogate to quantify the amounts
of hydroxypentanoic acid using the ions m/ z 73 and 75.
Hydroxyglutaric acid was similarly identified by rational-
izing the mass spectrum occurring at 14.7 min in the
chromatogram. The CI spectrum showed again a prominent
peak at m/ z M + 73 and at m/ z M + l, although the latter
was rather weak. In the El mode spectrum the corresponding
fragments, m/ z 349 (M - 15), m/ z 247 (M - 117), and m/ z
147, were found. 3-Hydroxy-3-methylglutaric acid was used
as a surrogate to quantify the amounts of hydroxyglutaric
acid using the ions m/ z 73+75.
Quantification of Reaction Products. The molecular yield
of cyclohexene-O3 reaction products in the gas and particulate
phases and the total yield are given in Table 5. The largest
single yield was measured for pentanal at 17%. This agrees
with that reported by Grosjean et al. (3) of 16% and by
Hatakeyama et al. (1) of 17%. The two other groups of
compounds with high molar yields are dicarboxylic acids
and oxocarboxylic acids, at about 22% and 16%, respectively.
Among the dicarboxylic acids, oxalic acid, malonic acid, and
glutaric acid are the most abundant, whereas succinic acid
and adipic acid were found in smaller quantities. All three
oxocarboxylic acids (4-oxobutanoic acid, 5-oxopentanoic
acid, 6-oxohexanoic acid) were formed in about equal
amounts (4-7%). Dialdehydes and hydroxylated compounds
were produced in only minor amounts. The overall average
carbon yield, both gas and particle phase, of the experiments
was about 47% (see Table 5), that is, only about one-half of
the carbon atoms originally present in the cyclohexene were
recovered in the 16 compounds quantified in this study. Much
of the missing carbon can likely be attributed to volatile
products that could not be identified with the technique
used in this study. For example, Hatakeyama et al. (1) found
significant amounts of CO, CO2, and HCOOH. If one assumes
that for all compounds with fewer than six carbon atoms
corresponding compounds with one or two carbon atoms
were produced, the carbon balance increases to about 62%.
The average distribution of aerosol composition is shown
in Figure 3. Total particle mass was calculated from the
The peak at 15.90 min retention time showed a similar
spectrum to that for hydroxyglutaric acid, except that all
relevant mass fragments were shifted by 14 mass units toward
higher masses; mass unit 14 is typically assigned to a -CH2-
group, and thus this compound was tentatively identified as
hydroxyadipic acid and quantified with 3-hydroxy-3-meth-
ylglutaric acid as surrogate.
9
4 8 9 8 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 34, NO. 23, 2000

TABLE 5. Average Molar Yield (%) of All Quantified Compounds
total molar yield
gas-phase molar yield
aerosol molar yield
compd name
oxalic acid
m alonic acid
succinic acid
glutaric acid
adipic acid
4-hydroxybutanal
hydroxypentanoic acid
hydroxyglutaric acid
hydroxyadipic acid
4-oxobutanoic acid
5-oxopentanoic acid
6-oxohexanoic acid
1,4-butandial
av
std dev
av
std dev
av
std dev
carbon yield av
6.16
6.88
0.63
5.89
2.20
2.60
1.02
2.33
1.19
6.90
4.52
4.16
0.53
0.44
1.64
17.05
64.14
3.15
3.68
0.34
3.37
1.10
2.54
0.62
1.51
0.26
4.62
3.09
3.30
0.41
0.20
1.15
9.39
4.83
6.25
0.49
5.20
1.46
1.96
0.84
0.34
0.23
6.48
4.13
3.77
0.45
0.40
1.62
17.05
55.48
2.69
3.46
0.31
3.36
0.82
2.08
0.56
0.24
0.09
4.59
3.07
3.34
0.39
0.18
1.15
9.39
1.33
0.63
0.14
0.69
0.74
0.64
0.18
1.99
0.97
0.42
0.39
0.39
0.08
0.04
0.03
0.35
0.23
0.04
0.92
1.08
0.31
0.20
2.56
0.48
0.25
0.23
0.22
0.06
0.03
0.02
2.04
3.41
0.41
4.87
2.18
1.58
0.73
1.46
1.27
4.56
3.73
4.13
0.30
0.36
1.72
14.1
46.86
61.9a
1,5-pentanedial
1,6-hexanedial
pentanal
sum
8.66
a
Percent recovery if sm all C1, C2 com pounds are included (see text).
By summing all the individual compounds, the average
total aerosol mass was 104% of the volume calculated from
the DMA size measurements based on an estimated aerosol
density of 1.4 g cm-3. This agreement between the two
independent methods indicates that the entire aerosol mass
has been accounted for on a molecular level, subject to
uncertainties inherent in the chemical analysis and the
assumed particle density.
Cyclohexene-Ozone Reaction Mechanism . The initial
reaction between cyclohexene and ozone is the well-
established addition of ozone to the double bond leading to
the energy-rich Criegee biradical (Figure 4a) (10). Loss of
CO2 from the Criegee biradical leads to the formation of
pentanal (b), the most abundant identified reaction product
(Figure 4b). The Criegee biradical can also react with water
and by subsequent loss of H2O2 and H2O lead to 1,6-
hexanedial (c) and 6-oxohexanoic acid (d), respectively.
Adipic acid (e) might be formed by oxidation of 6-oxohexanoic
acid, although an explicit mechanism for this reaction is not
obvious (Figure 4c). Similarly, we speculate that other acid
products, h, i, n, o, p, q, are formed from their corresponding
aldehyde precursors.
Intramolecular isomerization of the Criegee biradical leads
to an excited intermediate (f) and by subsequent loss of CH2O
can lead to 1,5-pentanedial (g), which is oxidized to 4-oxo-
pentanoic acid (h) and further to glutaric acid (i) (Figure 4d).
The intermediate (f), however, can also lead to the C4
compounds identified in this study as outlined in Figure 4e.
This reaction chain involves the cleavage of a C-C bond of
the intermediate alkoxy radical (j), which leads either to a C4
alkyl radical (k) or the C5 compound 1,5-pentanedial (g).
After reaction with O2 and H2O the alkyl radical (k) is
converted to a second alkoxy radical (l), which can react to
1,4-butanedial (m) and by further oxidation to 4-oxobutanoic
acid (n) and succinic acid (o). Another pathway the inter-
mediate alkoxy radical (1) can undergo is a cleavage and
subsequent reaction to malonic (p) and oxalic acid (q) (see
Figure 4f).
A possible reaction mechanism for the formation of
hydroxyadipic acid is shown in Figure 4g. After an initial H
abstraction from adipic acid, the reactions with O2 and H2O
lead to the formation of hydroxyadipic acid. The path of
formation of hydroxyglutaric acid is thought to be equivalent
to that leading to hydroxyadipic acid. CO was present in the
reaction chamber, which should have scavenged about 95%
of the OH radicals generated in the cyclohexene-O3 reaction.
FIGURE 3. Molecular distribution of the particle composition. 104%
of the particle mass could be resolved on a molecular level assuming
a weighted particle density of 1.4 g cm-3.
measured increase of the particle diameter using a density
of the organic mass of F ) 1.4 g cm-3, which is the weighted
average of the densities of the individual compounds listed
in Table 5. Dicarboxylic acids contribute about 41% of the
aerosol mass. The two hydroxylated dicarboxylic acids,
hydroxyglutaric acid and hydroxyadipic acid, contribute
about one-third to the aerosol mass (38%). Although their
overall molar reaction yields are rather small (3.5%), they
contribute significantly to the particle mass because of their
low vapor pressures. About 17% of the particle mass consists
of oxocarboxylic acids; dialdehydes account for only a small
amount, about 1.1%.
9
VOL. 34, NO. 23, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 4 8 9 9

FIGURE 4. (a)-(g) Proposed reaction mechanism of the cyclohexene-ozone reaction leading to the products observed in this study. For
details see text. The combined mass of all individual compounds is 104% of the mass determined from DMA measurement, based on an
assumed aerosol density of 1.4 g cm-3.
9
4 9 0 0 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 34, NO. 23, 2000

position reactions are probably too slow to contribute to a
significant loss for the ozonides (14). We do not exclude the
possibility that products of relatively high volatility, such as
hexanedial and pentanedial, were present in the aerosol
phase. Even if the presence of these relatively high vapor
pressure compounds in the aerosol phase resulted from
experimental artifacts, they represent only about 4% of the
aerosol mass and therefore do not compromise the general
results of this study.
Acknowledgments
This work was supported by the U.S. Environmental Protec-
tion Agency Center on Airborne Organics and the Swiss
National Science Foundation.
Literature Cited
(1) Hatakeyama, S.; Tanonaka, T.; Weng, J.; Bandow, H.; Takagi,
H.; Akimoto, H. Environ. Sci. Technol. 1985, 19, 935.
(2) Izumi, K.; Murano, K.; Mizuochi M.; Fukuyama, T. Environ. Sci.
Technol. 1988, 22, 1207.
(3) Grosjean, E.; Grosjean, D.; Seinfeld, J. H. Environ. Sci. Technol.
1996, 30, 1038.
(4) Yu, J.; Cocker, D. R.; Griffin, R. J.; Flagan, R. C.; Seinfeld, J. H.
J. Atmos. Chem. 1999, 34, 207.
(5) Griffin, R. J.; Cocker, D. R.; Flagan, R. C.; Seinfeld, J. H. J. Geophys.
Res. 1999, 104, 3555.
FIGURE 5. (a) The Criegee biradical can react with aldehydes to
a second ozonide. (b) Ozonides, like the reaction product of (a),
might react with the derivatization agent PFBHA, a primary amine,
to form an aldehyde and an acid.
(6) Yu, J.; Flagan, R. C.; Seinfeld, J. H. Environ. Sci. Technol. 1998,
32, 2357.
(7) Gutbrod, R.; Meyer, S.; Rahman, M. M.; Schindler, R. N. Int J.
Chem. Kinet. 1997, 29, 717.
(8) Odum, J. R.; Hoffmann, T.; Bowman, F.; Collins, D.; Flagan, R.
C.; Seinfeld, J. H. Environ. Sci. Technol. 1996, 30, 2580.
(9) LeLacheur, R. M.; Sonnenberg, L. B.; Singer, P. C.; Christman
R. F.; Charles, M. J. Environ. Sci. Technol. 1993, 27, 2745.
(10) Bailey, P. S. Ozonation in organic chemistry; Academic Press:
New York, 1978.
(11) Loan, L. D.; Murray, R. W.; Story, P. R. J. Am. Chem. Soc. 1965,
87, 737.
(12) Bauld, N. L.; Thompson, J. A.; Hudson, C. E.; Bailey, P. S. J. Am.
Chem. Soc. 1968, 90, 1822.
(13) Ellam, R. M.; Padbury, J. M. Chem. Commun. 1971, 1094.
(14) Hull, L. A.; Hisatsune, I. C.; Heicklen, J. J. Phys. Chem. 1972, 76,
2659.
It is not clear whether the formation of hydroxyglutaric acid
and hydroxyadipic acid (molar yields: 1.6 and 0.9%, respec-
tively) was a result of unscavenged OH radicals or whether
another reaction mechanism is responsible for the formation
of these two compounds.
The Criegee biradical can undergo reaction with aldehydes
to produce ozonides (11, 12) (Figure 5a). Since different
aldehydes are present as reaction products, formation of
ozonides is likely. No ozonides were detected; however, a
possible reason might be their decomposition during sample
treatment. PFBHA is a primary amine that can react with
ozonides leading to an aldehyde and an acid (13) (Figure
5b). The occurrence of the relatively high vapor pressure
particle-phase aldehydes and acids might actually be a result
of ozonide decomposition products. While another possibility
could be thermal decomposition of ozonides during the
Soxhlet extraction of the filter samples, which was performed
at about 70 °C, at temperatures below 100 °C the decom-
Received for review April 14, 2000. Revised manuscript re-
ceived August 29, 2000. Accepted August 29, 2000.
ES001180F
9
VOL. 34, NO. 23, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 4 9 0 1