Products and mechanisms of ozone reactions with oleic acid for aerosol particles having core-shell morphologies

Article abstract of DOI:10.1021/jp049759d

Heterogeneous reactions of oleic acid aerosol particles with ozone are studied below 1% relative humidity. The particles have inert polystyrene latex cores (101-nm diameter) coated by oleic acid layers of 2 to 30 nm. The chemical content of the organic layer is monitored with increasing ozone exposure by using an aerosol mass spectrometer. The carbon-normalized percent yields of particle-phase reaction products are 20-35% 9-oxononanoic acid, 1-3% azelaic acid, 1-3% nonanoic acid, and 35-50% other organic molecules (designated as CHO_T). There is approximately 25% evaporation, presumably as 1-nonanal. To explain the formation of CHO_T molecules and the low yields of azelaic and nonanoic acids, we suggest a chemical mechanism in which the Criegee biradical precursors to azelaic acid and nonanoic acid are scavenged by oleic acid to form CHO_T molecules. These chemical reactions increase the carbon-normalized oxygen content (z/x) of the C_xH _yO_z layer from 0.1 for unreacted oleic acid to 0.25 after high ozone exposure. Under the assumption that oxygen content is a predictor of hygroscopicity, our results suggest an increased cloud condensation nuclei activity of atmospherically aged organic particles that initially have alkene functionalities.

Full text of DOI:10.1021/jp049759d

6686
J. Phys. Chem. A 2004, 108, 6686-6695
Products and Mechanisms of Ozone Reactions with Oleic Acid for Aerosol Particles Having
Core-Shell Morphologies
Yasmine Katrib,† Scot T. Martin,*,† Hui-Ming Hung,† Yinon Rudich,‡ Haizheng Zhang,§
Jay G. Slowik,§ Paul Davidovits,§ John T. Jayne, and Douglas R. Worsnop
DiVision of Engineering and Applied Sciences, 29 Oxford St., Pierce Hall, Room 122, HarVard UniVersity,
Cambridge, Massachusetts 02138, Department of EnVironmental Sciences, Weizmann Institute,
RehoVot 76100, Israel, Chemistry Department, Boston College, Chestnut Hill, Massachusetts 02467, and
Aerodyne Research, Inc., 45 Manning Road, Billerica, Massachusetts 08121-3976
ReceiVed: January 16, 2004; In Final Form: April 28, 2004
Heterogeneous reactions of oleic acid aerosol particles with ozone are studied below 1% relative humidity.
The particles have inert polystyrene latex cores (101-nm diameter) coated by oleic acid layers of 2 to 30 nm.
The chemical content of the organic layer is monitored with increasing ozone exposure by using an aerosol
mass spectrometer. The carbon-normalized percent yields of particle-phase reaction products are 20-35%
9-oxononanoic acid, 1-3% azelaic acid, 1-3% nonanoic acid, and 35-50% other organic molecules
(designated as CHOT). There is approximately 25% evaporation, presumably as 1-nonanal. To explain the
formation of CHOT molecules and the low yields of azelaic and nonanoic acids, we suggest a chemical
mechanism in which the Criegee biradical precursors to azelaic acid and nonanoic acid are scavenged by
oleic acid to form CHOT molecules. These chemical reactions increase the carbon-normalized oxygen content
(z/x) of the CxHyOz layer from 0.1 for unreacted oleic acid to 0.25 after high ozone exposure. Under the
assumption that oxygen content is a predictor of hygroscopicity, our results suggest an increased cloud
condensation nuclei activity of atmospherically aged organic particles that initially have alkene functionalities.
1. Introduction
the products formed are uncertain, although they generally
contain additional oxygen atoms.
Atmospheric particles affect the environment in many ways.
On the global scale, they directly and indirectly affect radiative
forcing and the hydrological cycle.1-5 On the regional scale,
they have a primary role in air pollution, visibility, and human
health.6 Quantifying these roles is challenging, in part because
atmospheric particles have a complex structure and chemical
composition. For example, organic coatings can form on
inorganic particles such as sea salt and dust.7-12 In some cases,
these organic coatings control particle hygroscopicity and
heterogeneous chemical reactivity.13,14 Widespread occurrence
of organic and inorganic species, which are also often mutually
and homogeneously mixed within the same particle,15-17 implies
a broad impact for the organic chemical fraction of atmospheric
particles.18-26
The formation of oxygenated products during atmospheric
residence leads to a general increase in the carbon-normalized
oxygen content (z/x) of the average chemical composition
CxHyOz of the particle. An associated increase in particle
hygroscopicity and cloud condensation nuclei (CCN) effective-
ness is expected.13,37 With the current state of knowledge,
however, many assumptions are required for the formulation
of kinetic models and of reaction mechanisms relevant to the
aging of particles and the conversion from hydrophobic to
hydrophilic properties. A need exists to understand and quantify
the chemical aging processes of organic particles.
As a model system for complex multicomponent atmospheric
particles, in this paper we examine the ozone reaction with
particles having a surface layer of varying thickness of oleic
acid (C18H34O2). Particle oleic acid concentrations are ap-
proximately 1 ng m-3 in urban aerosol.38 The double bond of
oleic acid is susceptible to attack by ozone, which is believed
to lead to the formation of a primary ozonide (molozonide)
(Figure 1).
The molozonide decomposes when a pair of O-O and C-C
bonds break, which allows for two alternative reaction pathways.
Azelaic acid (AA) and 1-nonanal (NN) are believed to form in
pathway 1, while pathway 2 is expected to yield 9-oxononanoic
acid (OA) and nonanoic acid (NA). NN has a higher vapor
pressure and is expected to volatilize. Although AA, OA, and
NA have low vapor pressures and are surmised to remain in
the particle, a summary of previous work (Table 1) shows that
a thorough analysis of the possible particle-phase products has
not been undertaken (i.e., the numerous “n/a” designations).
Furthermore, there is an absence of quantitative particle-phase
Heterogeneous oxidation reactions are an important process
in the aging of the organic fraction in atmospheric particles,
specifically in affecting their chemical composition and hygro-
scopic properties. A key unknown is an understanding of the
chemical reactions between the organic particulate matter and
atmospheric oxidants such as O3, NO3, and OH.27-34 Condensed-
phase organic molecules typically react much faster than their
gas-phase counterparts: 1 in 103 collisions between O3 mol-
ecules and liquid oleic acid molecules leads to chemical reaction,
which compares to 1 in 106 in the gas phase.30,35,36 However,
* Author to whom correspondence should be addressed. E-mail:
scot_martin@harvard.edu. Web: http://www.deas.harvard.edu/∼smartin.
† Harvard University.
‡ Weizmann Institute.
§ Boston College.
Aerodyne Research, Inc.
10.1021/jp049759d CCC: $27.50 © 2004 American Chemical Society
Published on Web 07/16/2004

Ozone Reactions with Oleic Acid
J. Phys. Chem. A, Vol. 108, No. 32, 2004 6687
TABLE 1: Comparison of Techniques, Oleic Acid Layer Thickness, Ozone Exposure, Water Partial Pressure, Phase Analyzed
(i.e., particle versus gas), and Products Identified for Oleic Acid Reactions with Ozonea
oleic acid
thickness
ozone exposure
PH O
phase
studied
technique
AMS
(atm‚s)
1 × 10-5
(To2rr)
identified products
OA (major)
AA (minor)
NA (minor)
NN (not present)
other CxHyOz products (major)
OA (n/a)
AA (n/a)
NA (n/a)
NN (n/a)
OA (identified)
AA (n/a)
this work
aerosol study
2-30 nm
0.2
condensed
Morris et al.29
aerosol study
AMS
200-600 nm
>1 µm
1 × 10-5
8 × 10-4
10-7
-
-
-
-
condensed
condensed
Smith et al.33
aerosol study
SMS
NA (n/a)
NN (n/a)
Moise and Rudich39
coated wall flow tube study
GMS, IC
millimeters
condensed and gas OA (n/a)
AA (identified, off-line)
NA (n/a)
NN (gas, 25% yield)
NN (gas, 25% yield)
Thornberry and Abbatt30
coated wall flow tube study
GMS
millimeters
10-7
gas
a The designation “n/a” indicates that the analysis method is not performed for these chemical species. Key: AMS, aerosol mass spectrometer;
SMS, single-particle mass spectrometer; GMS, gas-phase mass spectrometry; and IC, ion chromatography of condensed-phase products. (1 atm of
O3 is 2.46 × 1019 molecules cm-3.)
of product yield on oleic acid layer thickness can identify the
reaction pathways that occur mostly in the surface region.
2. Experimental Section
An overview of the experimental setup and protocol is as
follows. Nonvolatile core particles of polystyrene latex (PSL)
are generated by using an atomizer. Following a drying stage
that decreases relative humidity below 1%, the particles are
coated with oleic acid by passing through a heated tube having
a linear hot-to-cool temperature gradient along its length. The
layer thickness of oleic acid is variable and is adjusted by
controlling the concentration of the oleic acid at the entry of
the heated tube. The coated particles are then exposed to variable
ozone concentrations. The loss of parent oleic acid and the
appearance of chemical products in the organic layer are
Figure 1. Reaction pathways of oleic acid with ozone. Shown are
quantified with an Aerodyne aerosol mass spectrometer (AMS).
pathways to form AA and NA by carboxylate formation on the right
2.1. Generation of Coated Particles. Oleic acid particles,
side of the double bond and to form OA and NA by carboxylate
formed by vapor condensation over a heated oleic acid liquid
formation on the left side of the double bond. Also shown is a third
reservoir (50 to 70 °C), are transported at 0.1 Lpm and externally
mixed downstream with nonvolatile core particles (Figure 2).
These core particles are 101 nm polystyrene latex (PSL) (Duke
Scientific), which are atomized into an aerosol flow (1 Lpm).
The combined externally mixed aerosol passes through a tube
furnace having a linear hot-to-cool temperature gradient (78 to
25 °C): the oleic acid particles vaporize in the hot region, and
the vapor subsequently condenses in the cool region onto the
surfaces of the PSL particles.40 Internally mixed coated particles
result. The temperature of the reservoir, the gradient of the oven,
and the adjustable flow are the levers controlling the layer
thickness of oleic acid. Using this apparatus, 101 nm PSL
particles coated with oleic acid layers varying from 2 to 30 nm
are generated in a reproducible and controlled manner in number
concentrations of 103 particles cm-3. (In the experiments for
thicker layer thickness, a second mode near 60-nm diameter is
also observed, presumably arising from homogeneous nucleation
of oleic acid. The relative contribution to total aerosol oleic
acid mass, however, is approximately 5%. This homogeneous
mode is omitted in our data analysis.)
pathway hypothesized by the results of the current work, which involves
the reaction of ozone, oleic acid, and a C9 product to form CxHyOz
products (see text). Key: AA, azelaic acid; NN, 1-nonanal; OA,
9-oxononanoic acid; NA, nonanoic acid.
product yields in all previous studies. An important aspect of
our current work is the capability to directly identify and
quantify the condensed-phase products of the in situ aerosol.
We determine the chemical composition of the particles and
the product yields after ozone exposure by using an Aerodyne
aerosol mass spectrometer (AMS). In addition to pathways 1
and 2 (Figure 1), we find that other organic molecules form,
indicating complex reaction pathways possibly mediated by
Criegee biradicals.29,33,39 Another important aspect of the current
study is that we probe the formation of products in the particles
as a function of the layer thickness from 2 to 30 nm. To carry
out this work, we develop a vaporization-condensation method
for generating an inert polystyrene latex particle core (101-nm
diameter) coated with oleic acid layers of variable thickness (2
to 30 nm). An important conclusion from previous studies is
that the diffuso-reactive length of O3 in oleic acid is ap-
proximately 10 nm.29,33,39 Our hypothesis is that a dependence
A scanning mobility particle sizer (SMPS) with isokinetic
sampling is employed to measure the number size distribution

6688 J. Phys. Chem. A, Vol. 108, No. 32, 2004
Katrib et al.
of PSL particles prior to atomization (i.e., in the aqueous PSL
suspension in the reservoir) and dried nonvolatile inorganic
impurities initially present in the water. An example of
coagulation is apparent in the second mode centered at 143 nm
for the PSL particles (Figure 3).
Complementary to the mobility diameter measured by the
SMPS, the AMS measures the particle aerodynamic diameter.
Whereas the mobility diameter decreases (Figure 3), the
aerodynamic diameter increases (not shown) after reaction with
ozone. Morris et al.29 and Smith et al.33 report that aerodynamic
diameter increases upon ozone exposure, whereas Broekhuizen
et al.41 report a decrease in mobility diameter similar to our
observations. For spherical particles, a decrease in mobility
diameter coupled to an increase in aerodynamic diameter implies
that particle mass density increases. Oxygenated products of
oleic acid are expected to increase density. A fuller report on
this topic is forthcoming from us.
2.2. Chemical Composition of the Organic Layer. The mass
loading of particle-phase oleic acid and its oxidation products
is obtained with use of an aerosol mass spectrometer (AMS).42
The AMS is composed of three main sections: (1) particle
sampling into high vacuum (10-5 Torr) through a critical orifice,
(2) aerodynamic sizing via particle time-of-flight (TOF) mea-
surement, and (3) particle analysis with mass spectrometry (MS).
Figure 2. Top: Experimental apparatus for particle generation, coating,
processing, and analysis. Bottom: Coating process inside the tube
furnace followed by reaction with ozone. Key: AMS, aerosol mass
spectrometer; SMPS, scanning mobility particle sizer.
For the MS analysis, the particle beam is directed onto a
resistively heated hotplate (ca. 350 °C), where the volatile and
semivolatile constituents of the particle flash vaporize upon
impact. The volatilized molecules are ionized by electron impact
at 70 eV, and the positive ions of the fragments pass through a
quadrupole mass filter and are detected as a current at the
electron multiplier. The detection process has sufficient mass
sensitivity and time response to detect single particle vaporiza-
tion events for a fixed m/z value. The AMS, therefore, provides
primary measurements of particle number size distribution
(number of particles cm-3‚air/d log Dp) and mass loading at a
specific m/z value (µg m-3/d log Dp) for any particle yielding
a positive ion fragment at the tuned m/z value. (The connection
between the electron multiplier current and the mass loading is
discussed in section 2.4.) Given an aerosol input of constant
composition, as is true for our laboratory studies, the m/z value
can be scanned to provide size-resolved mass spectra of the
particle ensemble. For our experimental conditions (103 particle
cm-3, 0.1 Lpm sampling, 75% transmission efficiency, and 300
ms m/z dwell time), each mass loading at specific m/z value is
an average of approximately 300 individual particles.
In the current paper, MS results are integrated across the
aerosol size distribution, and the size-resolved capability is not
fully exploited. (The measurement of the particle aerodynamic
diameter is employed to determine that the density of a reacted
particle increases.) The analytical capability of the AMS is
employed to measure the loss of oleic acid and the appearance
of oxidation products occurring with ozone exposure. Although
the SMPS measurements show 2-, 4-, 6-, and 8-nm coatings,
only the 8 nm and thicker coatings have sufficient mass for
detection by the AMS when operated in its size-resolving mode.
For thinner coatings, we leave open the AMS chopper for 100%
duty cycle. With this approach, we obtain the mass spectrum
without a particle sizing capability.
Figure 3. SMPS number size distribution measurements of uncoated
polystyrene (PSL) particles, particles coated by oleic acid (20 nm),
and particles coated by oleic acid and products after 1.1 normalized
ozone exposure. The original layer thickness of 20 nm decreases to 15
nm following ozone exposure.
of coated PSL particles. At low PSL concentrations, the
distribution is symmetric and is limited by the DMA transfer
function.35 Examples of uncoated PSL particles and PSL
particles having a 20 nm oleic acid coating are shown in Figure
3. The increase of the particle mobility diameter beyond that
of the PSL core (101 nm) is twice the oleic acid layer thickness,
assuming a uniform coating on spherical particles. The original
layer thickness of 20 nm decreases to 15 nm following ozone
exposure (Figure 3).
At the high PSL concentrations necessary for sufficient mass
loading in the AMS analysis, PSL coagulation occurs in some
of our experiments. Larger particles apparent in these distribu-
tions arise from a combination of factors, including coagulation
2.3. Ozone Flow Tubes Studies. Ozone reaction with oleic
acid is studied by coupling an aerosol flow reactor (1/2 in. i.d.,
85 cm length) to the SMPS/AMS system. At the exit of the
tube furnace (Figure 2), the coated particles are exposed to
ozone, which is produced by a 254-nm UV source and oxygen
using a Jelight Model 600 ozone generator. The ozone exposure

Ozone Reactions with Oleic Acid
J. Phys. Chem. A, Vol. 108, No. 32, 2004 6689
TABLE 2: Selected Values of the Normalized
and depends on the amount of the organic molecule present.
We obtain the R j values by a linear combination analysis of
the recorded spectrum, using the fragmentation patterns as the
basis set. The governing i equations requiring simultaneous
solution are as follows:
j
Fragmentation Pattern (fi ) Measured with the AMS,
Where i is the m/z Value and j is a Specific Chemical Species
(e.g., oleic acid)a
m/z
OL
OA
AA
NA
NNb
CHOT
44
0.006
0.035
0.008
0.002
0.072
0.001
0.02
0.0008
0.002
R jfi
j
142
144
152
155
158
221
265
282
0.0002
0.0002
0.002
0.0009
0.0001
0.001
0.0003
0.002
m )
(1)
∑
i
0.025
RIE j
j
0.03
0.001
0.0004
0.002
0.001
In matrix form, [m] ) [f][â], where [m] is a column matrix of
i rows, [f] is the fragmentation pattern matrix of i rows and j
columns, and [â] is the column matrix of unknown chemical
mass loadings having j rows (i.e., j products). Elements of [â]
are â j ) R j/RIE j. Once the system of equations is solved for
the R j values, the total mass loading of each product (C j, µg
0.001
0.0004
RIE
MW
n
1.4
282
18
1.4
172
9
1.4
188
9
1.4
158
9
1.4
142
9
1.4
see text
see text
m-3) is given by C j ) ∑imi ) ∑iR jfi .
j
j
a See further discussion in text. Also given are relative ionization
efficiency (RIE) compared to nitrate, molecular weight (MW), and
number of carbon atoms (n). Key: OL, oleic acid; OA, 9-oxononanoic
acid; AA, azelaic acid; NA, nonanoic acid; NN, 1-nonanal; and the
products CHOT. b NIST gas-phase fragmentation pattern given for NN
(see text).
The percent yield (Y j) of each condensed-phase product
relative to the initial oleic acid concentration is determined on
a carbon number basis, as follows:
n j/MW
j
C j
Y j ) 100
(2)
(atm.sec) is adjusted by varying the ozone concentration from
1 to 30 ppmV while the interaction time is constant at 3 s.
Reaction conditions are 298 K, 1 atm of N2, and under 1%
relative humidity. Ozone concentration is measured by light
extinction at 254 nm, using a Hg line lamp and a photodiode
detector along a 1-cm path. An absorption cross-section of
1.15 × 10-17 cm2 molecule-1 is employed. Ozone is injected
at a variable concentration, and the changes in particle reaction
products are determined by AMS analysis.
nOL/MOWL CO0 L
where n j is the number of carbon atoms in the molecule j, MW
j
OL
is the molecular weight of molecule j (see Table 2), and C0
is the oleic acid mass loading at zero ozone exposure. The
percent evaporation (YE) of the particle is given as YE ) 100 -
∑jY j.
3. Results and Discussion
2.4. Analysis Methods for Quantification of the Products.
Ozone reactions modify the particle-phase chemical composition
and hence the aerosol mass spectrum. To identify and quantify
the molecules present, we deconvolute the mass spectrum using
the principles of linear analysis, as explained in detail below.
The electron multiplier current of the AMS at each m/z value
is a direct measure of the number of ions. The two key operative
physical principles for mass determination are (1) that the
number of ions generated by electron impact through a thin path
is proportional to the number of electrons in a bolus of a material
(i.e., a particle) and (2) that the number of electrons is
proportional to the mass of the bolus. The first proportionality
constant is the ionization cross section, while the second is
determined by the approximate 1 electron to 1 proton to 1
neutron relationship among light elements. On the basis of these
principles, the AMS software employs the ionization efficiency
(IE) of nitrate, which is obtained in a daily calibration procedure
with a test ammonium nitrate aerosol having a measured mass
of nitrate, to convert the current at the electron multiplier into
the nitrate equivalent mass (mi, µg‚NO3 m-3) at each m/z value
(i). (For this approach to be accurate, the m/z values of all ions
generated from the bolus of material must be scanned.) The mi
values are shown as the y-axes of Figures 4, 5, and 7. They
must be corrected by the ionization efficiency of each organic
molecule relative to nitrate (RIEj, no units) (Table 2) to obtain
3.1. Control Experiments. The following control experi-
ments set the framework for the interpretation of our data. No
organic material is detected by the AMS for PSL particles alone;
PSL particles (C8H8)n do not vaporize until an AMS heater
temperature of 850 °C, which is 500 °C above our normal
operating conditions. Similarly, no organic material is detected
when PSL particles alone are exposed to ozone. When oleic
acid is included in the experiments, only its mass spectrum is
observed until ozone is added. The oleic acid fragmentation
patterns obtained for pure OL particles versus OL-coated PSL
particles are the same, from which we conclude that the
morphology of the studied particles does not affect the mass
spectra. For particles coated by oleic acid and reacted with
ozone, the total organic mass detected is independent of the
heater temperature for 150 to 600 °C, which indicates that we
quantitatively vaporize all semivolatile organics in the reacted
outer layer of the particles. Another possible complication ruled
out is that reacted organic material could over time condense
on the inner walls of the tube furnace when cooled. When next
heated, these products could volatilize and complicate our
product analysis. However, heating the tube furnace after several
weeks of experiments shows neither organic material in the
AMS nor particles in the SMPS, both in the presence and
absence of ozone.
the organic molecule mass loadings (mi , µg‚organic m-3),
3.2. Aerosol Mass Spectrometry of Authentic Standards.
The sources of our authentic standards are as follows: OL (CAS
112-80-1, NIST MS No. 228066, above 99% purity, Aldrich),
AA (CAS 123-99-9, NIST MS No. 113078, 99% purity,
Aldrich), OA (NIST MS No. 144254, 98%, Larodan), NA (CAS
112-05-0, NIST MS No. 227949, 98% purity, Aldrich), and NN
(CAS 124-19-6, NIST MS No. 232162, 98% purity, Aldrich).
The particle-phase MS fragmentation patterns (Figure 4) of the
authentic standards of OL, AA, OA, and NA compare well with
j
where j is a specific chemical species (viz. j ∈ OL, OA, AA,
NA, NN, and CHOT). The RIE j values OL, OA, AA, and NA
are determined from pure compounds.42 (Products CHOT are
discussed in detail later.) The recorded mass at each m/z value
j
is given by mi ) ∑j(mi /RIE j).
Each organic molecule has a specific fragmentation pattern
fi j (µg‚organic m-3), normalized such that 1 ) ∑i fi . We have
j
mi ) Rjfi , where R j is a must-be-determined scaling factor
j
j

6690 J. Phys. Chem. A, Vol. 108, No. 32, 2004
Katrib et al.
Figure 5. (a) Mass spectrum of pure oleic acid particles. (b) Mass
spectrum after ozone exposure of 1.98 × 10-5 atm‚s. (c) Mass spectrum
after ozone exposure of 6.6 × 10-5 atm‚s. Identified peaks are 152
amu for AA, 144 amu for OA, 155 amu for CHOT, and 282 amu for
OL. The numbers next to each label are the y-axis values at the specific
m/z value. Conditions: 11 nm coating on PSL core.
Figure 4. Reference mass spectra for OL, AA, OA, and NA. The
principal peaks for the chemical species are 282 amu for OL, 152 amu
for AA, 144 amu for OA, and 158 amu for NA. See also Table 2.
the gas-phase MS patterns of the national institute and technol-
ogy issues (NIST) library (not shown), although there is some
variation in the relative intensities of the peaks, which could
be explained by differing ionization settings and mass filters
(i.e., quadrupole versus magnetic sector) of the AMS and the
NIST procedures.43 Table 2 summarizes the observed fragmen-
tation patterns at important m/z values. The relative intensities
of fragmentation patterns in the AMS analysis are reproducible
within 5%.
Aerosol is obtained for AMS analysis by atomizing the
authentic standards individually in methanol; whether atomized
or obtained by vapor condensation, the OL mass spectrum is
similar. NN also prepared in methanol does not yield an aerosol
mass spectrum, presumably due to its high volatility. The NIST
database provides the gas-phase NN fragmentation pattern. On
the basis of this fragmentation pattern, NN is never detected
by us when oleic acid is exposed to ozone, viz. RNN ) 0.
3.3. Reaction Pathways upon Ozone Exposure. In the
following sections, we discuss the formation of OA, AA, NA,
and CHOT products (section 3.3.1 and section 3.3.2), the effects
of layer thickness on those products (section 3.3.3), and the
increase in chemical oxygen content with increasing ozone
exposure (section 3.3.4).
3.3.1. OA, AA, and NA Products. An example of raw data
demonstrating the disappearance of parent OL and the appear-
ance of OA, AA, and NA products upon ozone exposure is
shown in Figure 5. The oleic acid peak (282 amu) decreases
steadily from panel a to panel c in Figure 5 as ozone exposure
increases from 0 to 6.6 × 10-5 atm‚s. Product peaks for OA
(144 amu), AA (152 amu), and NA (158 amu) concomitantly
increase, although to observe these results the raw data in Figure
5 must be corrected for remaining oleic acid as by eq 1. NN
(142 amu) is not detected, possibly due to its volatility. Moise
and Rudich39 and Thornberry and Abbatt30 identify NN as a
major volatile reaction product.
The loss of parent compound and the appearance of products
with increasing ozone exposure are shown in Figure 6. The
percent of oleic acid reacted increases steadily with increasing
ozone exposure (Figure 6a). We define 95% loss as 1.0
normalized exposure. The normalization is useful for comparison
of results among different layer thickness (e.g., Figure 9).
Identification and quantification of the condensed-phase
products of the in situ aerosol shows that as OL decreases with
increasing ozone exposure, OA, AA, and NA form (Figure 6b,c).
The rapid initial formation of the products OA, AA, and NA
parallels the loss of the OL parent compound (Figure 6a). The
product yields shown in Figure 6c are 35% for OA, 3% for
AA, and 1% for NA after 1.0 normalized ozone exposure.
Although the run-to-run variability of an individual m/z value
is 5%, we can nevertheless reproducibly observe AA and NA
in yields of 1% to 3% by using the information content of the
entire mass spectrum in conjunction with the pattern analysis
methods implied in eq 1 (i.e., pseudoinverse matrix to obtain
the matrix of chemical mass loadings â).
During ozone exposure, the particles evaporate by up to 30%
by carbon number as determined by our analysis of the mass
spectra. This result is consistent with the decrease in measured
particle mobility diameter (Figure 3). If NN is assumed to be
the volatile product, then evaporation of 25% by carbon number
is expected based upon NN gas-phase product yields reported
in the literature.30,39
Our results show that AA and NA are minor products. The
OA:AA and OA:NA relative yields at 1.0 normalized ozone
exposure range from 9 to 18 for the 2 to 30 nm coating thickness
studied. This result can be compared to the 1:1 product ratios
expected for the gas-phase reactions of ozone with oleic acid.44
Our results for the oxidation reaction of condensed-phase oleic
acid by ozone can be compared to those reported in the literature
(Table 1). Morris et al.29 do not clearly identify products. Moise
and Rudich39 in an ex situ chromatographic analysis identify
AA as a condensed-phase reaction product and NN as a volatile
reaction product. Moise and Rudich39 do not analyze for OA.
Smith et al.33 report a major product, which is tentatively
assigned as OA, as well as a small amount of NN. They employ
particles several times larger than ours, which may explain why
NN does not fully volatilize. Thornberry and Abbatt30 identify
NN as a volatile reaction product. The results of Morris et al.,29
Moise and Rudich,39 Smith et al.,33 and Thornberry and Abbatt30
are consistent with our observations.
3.3.2. Other CxHyOz Products. After accounting for OL, OA,
AA, NN, and NA fragments, a residual mass spectrum remains.
Important marker peaks occur at 155, 221, and 265 amu (Figure
7). The intensities of these peaks increase linearly with ozone
exposure.

Ozone Reactions with Oleic Acid
J. Phys. Chem. A, Vol. 108, No. 32, 2004 6691
Figure 7. Example of the mass spectrum of the combined products
(CHOT) shown for 30 nm coating and 10-5 atm‚s ozone exposure. Note
that the y-axis is log to show the full range of peak intensities.
To establish the minimum number of major components
necessary to explain our data in the reacted particles, we select
16 mass spectra for products having normalized ozone exposure
greater than 1.0 and 16 mass spectra for unreacted oleic acid.
We obtain three singular values above the noise threshold, which
we believe correspond to OL, OA, and CHOT. In addition to
these major products, the singular value analysis does not rule
out additional minor products below the noise threshold. As
discussed below, we believe AA and NA occur as minor
products.
To explore the physical basis of the singular value compo-
nents, we first try to obtain [â] by using [â] ) [f]-1[m] for [f]
constituted by the authentic standards OL, OA, AA, and NA.
The results show significant contributions by OL and OA;
however, the residuals obtained are not satisfactory and there
are negative concentrations for AA. We conclude that there must
be one or more unidentified chemical species present. When
taken together, these species constitute the CHOT group of
molecules.
To isolate the mass spectra of CHOT, we select m/z values
unique to the known species and assume that the CHOT products
do not contribute significantly at these selected m/z values.
Specifically, we subtract the pure OL standard from all the
recorded mass spectra using the 282 peak as the proportionality
factor. From these residuals, we then subtract OA using the 144
peak, followed by AA using the 152 peak, followed by NA
using the 158 peak. The resulting set of residuals is shown by
singular value decomposition to have three singular values above
the noise threshold of 0.04. We conclude CHOT is composed
of at least three molecules CHOa, CHOb, and CHOc. Our efforts
to deconvolute the CHOT spectrum (Figure 7) into CHOa, CHOb,
and CHOc component spectra yield nonunique results. We
believe additional experiments are necessary to better identify
CHOa, CHOb, and CHOc.
Figure 6. (a) Percent reacted oleic acid (100 - YOL; eq 2) with
increasing ozone exposure. We define 95% loss as 1.0 normalized
exposure in parts b and c. The normalization is useful for comparison
of results among different layer thickness (e.g., Figure 9). (b) Percent
yield (Y j of eq 2) of condensed-phase products (AA, OA, and CHOT;
left-hand axis) for particles initially having 8 nm OL coatings with
increasing ozone exposure. Also shown is the percent change in particle
mass (0) (YE, right-hand axis), which arises from the evaporation of
volatile products. (c) Same as part b except for 30 nm coating thickness.
Key: (9) AA, (b) OA, (1) NA, (2) CHOT. Lines are drawn in the
figures to aid the eye and do not represent a model fit. (The AA and
NA data points overlap and are nearly indistinguishable.)
3.3.2.1 MultiVariate Analysis. We employ singular value
decomposition to explore the variability in the set of mass
spectral data from 0 to 2 normalized ozone exposure for 8- to
30-nm coatings. Using 50 mass spectra for pure oleic acid, we
establish a variance threshold of 0.89 (i.e., the first singular value
divided by the sum of all singular values). The remaining 0.11
variance diagnoses the reproducibility of our experiment. The
noise threshold is 0.04, which is the second singular value for
pure oleic acid. This result is consistent with our direct
observations that the relative intensities making up the frag-
mentation pattern of oleic acid are reproducible within 5%.
3.3.2.2. Percent Yield. The uncertainties in the chemical
structures of the CHOT components, especially their oxidation
states, provide a challenge for estimating their percent yields.
To approach this problem, we assume an OL molecule combines
with one of the C9 radical intermediates. With this assumption,
we obtain nCHO ) 27 and 472 < MCWHO < 518. We can then
T
T
use eq 2 to calculate percent yield. As shown in eq 2, the ratio
CHOT
nCHO /MW
is the critical value, rather than our absolute
T

6692 J. Phys. Chem. A, Vol. 108, No. 32, 2004
Katrib et al.
assumptions of nCHO and MWCHO
. We have 0.62 <
T
T
Although we do not rule out the formation of secondary
ozonides by the reaction of the aldehyde products with Criegee
biradicals, we do not need to invoke this explanation for our
data set. Secondary ozonide formation by the reaction of the
Criegee biradical with the CdO groups of NN and OA would
be inconsistent with the reported 25% yield of 1-nonanal.30 This
reasoning suggests that Criegee biradicals preferentially react
with the double bond of OL despite the initial proximity of the
CdO groups of NN and OA in a cage prior to diffusion away
from the Criegee biradicals.
CHOT
12nCHO /MW < 0.68, which reflects the uncertainty in the
T
oxidation state of the residual organic component. We also
assume RIECHO ) 1.4. The uncertainty bars in Figures 6b,c
T
and 9 reflect this analysis. Panels b and c in Figure 6 show that
as OL decreases with increasing ozone exposure, CHOT forms
as a major product comparable with OA. CHOT has a higher
percent yield for thinner rather than thicker coatings.
3.3.2.3. Proposed Mechanism. Any proposed mechanism is
constrained by the following observations:
1. Work by Moise and Rudich39 and Thornberry and Abbatt30
shows that the only volatile product formed is NN. Its yield is
25%. Pointedly, CO2(g) is not observed.
2. We find OA is the major identified condensed-phase
product. Its yield is above 25%. The NN yield accounts for
25% yield of OL; the additional OA yield requires that OA be
formed by additional pathways.
The observations also allow several other gas-phase pathways
to be ruled out in the condensed phase. A unimolecular
decomposition pathway of the Criegee biradical in the gas phase
is COO to OCO isomerization followed by release of CO2,
which in the case of OL ozonolysis would lead to the formation
of octane instead of NA and octanoic acid instead of AA.47 In
contrast, observations of the gas-phase products show neither
octane nor CO2.30,39 The absence of this pathway further
supports our hypothesized mechanism that the Criegee biradicals
react sufficiently rapidly with OL such that the gas-phase
reaction pathways are at most minor pathways in the condensed
phase.
3. AA and NA form as minor products below 3% yield. The
similarities among the chemical structures of OA, AA, and NA
indicate similar chemical reactivities in regard to possible further
reactions with ozone or Criegee biradicals. To reconcile this
similar reactivity with the observations of high OA yield and
low AA and NA yields, we infer that AA and NA must never
form in great yield. Their precursor species must be scavenged.
4. Additional CHOT products form at yields above 25%. The
mass spectrum of the CHOT component (Figure 7) shows that
these molecules have fragments at high amu values (greater than
200 amu). They therefore contain carbon chains longer than C9
(i.e., the products of pathways 1 and 2 shown in Figure 1).
The mechanism we propose below is consistent with these
four observations.
The formation of CHOT molecules is hypothesized by us to
involve reactions of the Criegee biradical in the condensed
phase. The initial products of ozone attack on OL are a Criegee
biradical and an aldehyde.45 In the gas phase, the biradical
decomposes into a suite of chemical species, such as RCO,
RCOOH, CO2, CO, H2O, and OH.45,46 In solvents, however,
the highly energetic Criegee biradical may become stabilized,
in which case the decomposition pathways analogous to gas-
phase reactions are less important.
For the reaction mechanism in the condensed phase, we
propose that ozone adds across the double bond of oleic acid.
The molozonide then decomposes by 50:50 channels to form
the radical species shown in Figure 8. The key point in our
proposed mechanism is that the resulting Criegee biradicals add
to the oleic acid double bond as a major pathway to form C27
molecules. Rearrangements to AA and NA are minor path-
ways: this condensed-phase behavior contrasts to gas-phase
reactions because the high density of double bonds in condensed-
phase oleic acid intercepts the Criegee biradicals before
significant AA and NA can form. Cleavage of the five-
membered-ring peroxide in the C27 molecule can lead to many
compounds, including ketones, alkenes, and carboxylic acids
(Figure 8). The symmetry of OL leads to the result that the
Criegee biradical precursor to AA can instead add to OL, form
a peroxide, cleave, and yield OA. As such, the OA yield can
exceed 25%, which could explain the OA yields observed in
Figure 6. Similarly, the precursor to NA can form C27 products
(Figure 8), which may be stable or decompose to many possible
products, including NN. This NN-formation pathway, however,
may not be favorable because there is some evidence against
it: gas-phase studies show only a 25% NN yield, which can be
fully accounted for by direct decomposition of the primary
ozonide (Figure 8).
After 1.0 normalized ozone exposure, the mass spectrum of
CHOT continues to change (not shown). We probe the change
at 155 amu for normalized ozone exposure up to 3.0 and observe
a continuous increase of the signal. This observation suggests
that new reaction pathways open when the supply of OL double
bonds is depleted. Possible reactions at this point could include
secondary ozonides, although any OA decrease is not detectable
within our uncertainty limits (Figure 6). Alternatively, O3 could
react directly with the carboxylic acid groups.48 Continued direct
reactions of CHOT products with O3 are thus hypothesized by
us as sufficient to continue to process the particles and to alter
the mass spectrum of CHOT. The report by Ziemann49 of
numerous oxygenated organic molecular products including
peroxides for the reaction of OL with O3 is consistent with our
observations.
3.3.3. Effect of Layer Thickness. The reacto-diffusive length
of ozone in oleic acid is reported as 5,29 6,39 and 20 nm.33 Hence,
varying the coating thickness of oleic acid layers from 2 to 30
nm allows us to probe the transition between reactions that occur
mainly in the surface region versus those that occur predomi-
nantly in the bulk volume. The overall uptake coefficient γ can
be partitioned into contributions from uptake due to reaction at
the surface (Γsurf) versus in the bulk (Γbulk).33 A future report
from us will provide kinetic results (e.g., Figure 6a) for the
dependence of γ on layer thickness and hence the partitioning
into Γsurf and Γbulk
.
In addition to the different possible rates for the surface versus
bulk reaction channels, it is further possible that the relative
yield of products is influenced by surface versus bulk chemical
environments. As a limiting example to make this point, consider
that product A forms only in the bulk chemical environment
whereas product B forms only in a surface region of fixed depth.
In the case of a 2 nm layer thickness, a 5 nm assumed surface
region, and a 10 nm assumed reacto-diffusive length, the yields
of product A and B would be 0% and 100%, respectively.
Increasing the layer thickness to 10 nm would increase the yield
of A and decrease that of B. Increasing the layer thickness
beyond 10 nm would continue to produce a mix of A and B,
approaching some limiting ratio as determined by the diffusion
of oleic acid from the deep interior of particle into the reaction
zone (of depth equal to the reacto-diffusive length), which is
subdivided into surface and bulk chemical environments. This

Ozone Reactions with Oleic Acid
J. Phys. Chem. A, Vol. 108, No. 32, 2004 6693
Figure 8. Proposed condensed-phase reaction pathways. The Criegee biradical reacts with oleic acid.
where n jO is the number of oxygen atoms in the molecule j.
example assumes that the reacto-diffusive length is invariant
x
with regard to the extent of reaction and that the reacto-diffusive
length is greater than the depth of the surface region.
The effect of layer thickness on percent yield of condensed-
phase products is shown in Figure 9 for 1.0 normalized ozone
exposure. CHOT and OA percent yields appear to decrease and
increase respectively with greater layer thickness, suggesting
that a surface reaction for CHOT formation occurs at greater
yield than its bulk counterpart and vice versa for OA. Within
the uncertainty bars, no trend can be discerned for the AA and
NA percent yields for layer thickness varying from 2 to 30 nm.
3.3.4. Particle Oxygen Content. When ozone reacts with oleic
acid, there is an increase in the carbon-normalized oxygen
content (z/x) in the average chemical composition CxHyOz of
the layer. We use the definition
The numerator of eq 3 is the total moles of oxygen. The
denominator of eq 3 is the total moles of carbon. The ratio z/x
is anticipated to correlate positively with the hygroscopicity and
the density of the overall particle chemistry.
The carbon-normalized oxygen content (z/x) in the average
chemical composition CxHyOz of the particle increases with
higher ozone exposure (Figure 10). CH4 has z/x ) 0.0 while
CO2 has z/x ) 2. To prepare Figure 10, we employ eq 3 for
which limiting assumptions on the molecular structures of CHOT
are necessary. As discussed above for Figure 6b,c, we have
CHOT
0.62 < 12nCHO /MW
< 0.68. We further employ 0.11 <
T
T
CHOT
16nCOHO /MW < 0.20 for 3.5 < nCOHO < 6.5, which is based
T
x
x
on our description that one O3 molecule adds to oleic acid to
yield C27 structures via the mechanisms depicted in Figure 8.
Namely, these structures have either 4 (NA precursor radical
pathway) or 6 (AA precursor radical pathway) oxygen atoms
for 27 carbon atoms. An assumption of these structures restricts
(n jO /MWj)C j
∑
j
x
(z/x) )
(3)
(n j/MWj)C j
∑
j

6694 J. Phys. Chem. A, Vol. 108, No. 32, 2004
Katrib et al.
4. Conclusions and Atmospheric Implications
In this paper the reaction products and mechanisms of ozone
with multicomponent aerosol particles are studied with oleic
acid coated particles as a model system. The results of our study
have several atmospheric implications:
1. In our proposed mechanism (Figure 8), condensed-phase
ozonolysis leads to radical coupling reactions. Specifically, these
products are proposed to form by the reactions of Criegee
intermediates with alkenes present in the particles. We speculate
that long-chain multifunctional organic molecules result and that
these high molecular weight products have low vapor pressures
and therefore remain in the condensed phase. Although the
formation of high molecular weight compounds such as humic
acids52 has been suggested for reactions of organic molecules
in acidic medium,53 our suggestion of these molecules in the
ozonolysis reactions of organic aerosol particles is new.
2. The organic molecules formed by the reactions of oleic
acid and ozone are more oxygenated than the parent molecule
(i.e., (z/x) increases). The oxygen-enriched organic molecules
are expected to have small contact angles with water.27,34,54 As
a result, the reacted particles may act as improved CCN
compared to the unreacted hydrophobic particles.55 The forma-
tion of water-soluble molecules such as carboxylic acids may
further enhance the hygroscopicity and hence the CCN activity
of the reacted particles. The high molecular weight molecules
may also be surface active and thus may lower the surface
tension of partially activated droplets, thus possibly altering the
Ko¨hler curve growth properties of the reacted particles.14
3. This study demonstrates that relative product yields depend
on whether the reactions occur in the surface or bulk regions
of the particle. By extension, we suggest that the reaction
products of thin coatings, such as those encapsulating dust
particles, may differ from the products formed inside particles
having a high fraction of organic matter.
Figure 9. Effect of layer thickness on percent yield (Y j of eq 2) of
condensed-phase products (AA, OA, NA, and CHOT; left-hand axis)
at 1.0 normalized ozone exposure. Also shown is the percent change
in particle mass (0) (YE, right-hand axis), which arises from the
evaporation of volatile products. Key: (9) AA, (b) OA, (1) NA, (2)
CHOT. Lines are drawn in the figures to aid the eye and do not represent
a model fit. (The AA and NA data points overlap and are nearly
indistinguishable.)
4. This study finds that ozonolysis reactions lead to the
formation of volatile products, which are readily released to
the gas phase. The ozonolysis reactions studied here lead to
25% mass loss. The particle mass loss is accompanied by
unequal changes in the mobility and aerodynamic diameters of
the particle, which implies that the density of the reacted
particles also changes, apparently due to the addition of oxygen.
The heterogeneous chemistry of organic molecules in atmo-
spheric aerosol particles is complex. Condensed- and gas-phase
products and mechanisms differ. Inside the condensed phase,
surface and bulk regions have differing chemical reactivities.
Our results thus show that the reaction products and mechanisms
of organic molecules with ozone may depend not only on the
chemical composition of the molecules but also on their physical
state and geometrical configuration. Further studies with mul-
ticomponent aerosol particles with varying relativity humidity
are warranted to further elucidate this complex problem.
Figure 10. Effect of normalized ozone exposure on carbon-normalized
oxygen content (z/x) in the average CxHyOz chemical composition of
the layer for 8 (9) and 30 nm (2) coatings. Shown are the central
values with uncertainty estimates showing minimum and maximum
values assuming specific limiting molecular structures discussed in the
text. Inset: (z/x) versus 100m44/∑jCj for 30 nm coating (see text).
the x-axis range of Figure 10 to an upper limit of approximately
1.0 normalized ozone exposure because the calculation assumes
only one net O3 molecule present in the proposed C27 structures.
Other structures having additional net O3 content are suggested
by the continuous development of the CHOT mass spectrum
for high ozone exposures (section 3.3.2.3).
Interpretations of AMS measurements of atmospheric par-
ticles employ the intensity of the 44-amu peak as an indicator
of oxidized organic compounds.50,51 The intensity is small when
sampling primary particles believed to be composed of soot and
other materials having small z/x, whereas the intensity increases
for aged particles, including secondary organic aerosol. These
aged particles are more hygroscopic and have larger z/x. The
inset in Figure 10 shows the mass percent of the 44-amu peak
in our analyses (given by 100m44/ ∑jC j) directly correlates with
z/x.
Acknowledgment. We are grateful for support received from
the National Science Foundation Atmospheric Chemistry Pro-
gram and the Dreyfus Foundation Postdoctoral Program in
Environmental Chemistry.
References and Notes
(1) Seinfeld, J. National Research Council, Panel on Aerosol RadiatiVe
Forcing and Climate Change. A Plan for a Research Program on Aerosol
RadiatiVe Forcing and Climate Change; National Academy Press: Wash-
ington, DC, 1996.
(2) Heintzenberg, J. Chem. Unserer Zeit 1999, 33, 158.
(3) Haywood, J.; Boucher, O. ReV. Geophys. 2000, 38, 513.

Ozone Reactions with Oleic Acid
J. Phys. Chem. A, Vol. 108, No. 32, 2004 6695
(30) Thornberry, T.; Abbatt, J. P. D. Phys. Chem. Chem. Phys. 2004, 6,
(31) Tobias, H. J.; Ziemann, P. J. Anal. Chem. 1999, 71, 3428.
(32) Grassian, V. H. Int. ReV. Phys. Chem. 2001, 20, 467.
(33) Smith, G. D.; Woods, I. E.; DeForest, C. L.; Baer, T.; Miller, R.
E. J. Phys. Chem. A 2003, 106, 8085.
(34) Bertram, A. K.; Ivanov, A. V.; Hunter, M.; Molina, L. T.; Molina,
M. J. J. Phys. Chem. A 2001, 105, 9415.
(4) Ramaswamy, V.; Boucher, O.; Haigh, J.; Hauglustaine, D.; Hay-
wood, J. M.; Myhre, G.; Nakajima, T.; Shi, G. Y.; Solomon, S. Radiative
Forcing of Climate Change. In Climate Change 2001: The Scientific Basis.
Contribution of Working Group I to the Third Assessment Report of the
IntergoVernmental Panel on Climate Change; Houghton, J. T., Ding, Y.,
Griggs, D. J., Noguer, M., Linden, P. v. d., Dai, X., Maskell, K., Eds.;
Cambridge University Press: Cambridge, UK, 2001; pp 349ff.
(5) Ramanathan, V.; Crutzen, P. J.; Kiehl, J. T.; Rosenfeld, D. Science
2001, 294, 2119.
84.
(6) Samet, J. Research Priorities for Airborne Particular Matter. 1.
Immediate Priorities and a Long-Range Research Porfolio; National
Academy Press: Washington, DC, 1998.
(7) Cruz, C. N.; Pandis, S. N. J. Geophys. Res. 1998, 103, 13111.
(8) Russell, L. M.; Maria, S. F.; Myneni, S. C. B. Geophys. Res. Lett.
2002, 29, 1029.
(9) Posfai, M.; Xu, H.; Anderson, J. R.; Buseck, P. R. Geophys. Res.
Lett. 1998, 25, 1907.
(10) Falkovich, A. H.; Schkolnik, G.; Ganor, E.; Rudich, Y. J. Geophys.
Res. 2004, 109, D022008.
(35) Mulholland, G. W.; Bryner, N. P.; Croarkin, C. Aerosol Sci.
Technol. 1999, 55, 31.
(36) Wadia, Y.; Tobias, D. J.; Stafford, R.; Finlayson-Pitts, B. J.
Langmuir 2000, 16, 9321.
(37) Weingartner, E.; Burtscher, H.; Baltensperger, U. Atmos. EnViron.
1997, 31, 2311.
(38) Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.;
Simonelt, B. R. T. EnViron. Sci. Technol. 1991, 25, 1112.
(39) Moise, T.; Rudich, Y. J. Phys. Chem. A 2002, 106, 6469.
(11) Tervahattu, H.; Juhanoja, J.; Kupiainen, K. J. Geophys. Res. 2002,
107, 4319.
(12) Ellison, G. B.; Tuck, A. F.; Vaida, V. J. Geophys. Res. 1999, 104,
11633.
(13) Saxena, P.; Hildemann, L. M.; McMurry, P. H.; Seinfeld, J. H. J.
Geophys. Res. 1995, 100, 18755.
(14) Facchini, M. C.; Mircea, M.; Fuzzi, S.; Charlson, R. J. Nature 1999,
401, 257.
(40) Han, J. H.; Martin, S. T. Aerosol Sci. Technol. 2001, 34, 363.
(41) Broekhuizen, K. E.; Kumar, P. P.; Abbatt, J. P. D. American
Geophysical Union Fall Meeting, San Francisco, 2003.
(42) Jayne, J. T.; Leard, D. C.; Zhang, X.; Davidovits, P.; Smith, K.
A.; Kolb, C. E.; Worsnop, D. R. Aerosol. Sci. Technol. 2000, 33, 49.
(43) NIST Chemistry Webbook; http://Webbook.Nist.Gov/Chemistry/,
2003.
(44) Atkinson, R.; Carter, W. P. L. Chem. ReV. 1984, 84, 437.
(45) Criegee, R. Angew. Chem., Int. Ed. Engl. 1975, 14, 745.
(15) Middlebrook, A. M.; Murphy, D. M.; Thomson, D. S. J. Geophys.
Res. 1998, 103, 16475.
(16) Noble, C. A.; Prather, K. A. Mass Spectrom. ReV. 2000, 19, 248.
(17) Allan, J. D.; Coe, H.; Bower, K. N.; Williams, P. I.; Galanter, M.
W.; Alfarra, M. R.; Jimenez, J. L.; Nernitz, E.; McDonald, A. G.;
Canagaratna, M. R.; Jayne, J. T.; Worsnop, D. R. J. Geophys. Res. 2003,
108, 4284.
(18) Novakov, T.; Penner, J. E. Nature 1993, 365, 823.
(19) Novakov, T.; Corrigan, C. E. Geophys. Res. Lett. 1996, 23, 2141.
(20) Yu, S. C. Atmos. Res. 2000, 53, 185.
(21) Jacobson, M. C.; Hansson, H. C.; Noone, K. J.; Charlson, R. J.
ReV. Geophys. 2000, 38, 267.
(22) Charlson, R. J.; Seinfeld, J. H.; Nenes, A.; Kulmala, M.; Laaksonen,
A.; Facchini, M. C. Science 2001, 292, 2025.
(46) Finlayson-Pitts, B. J.; Hemminger, J. C. J. Phys. Chem. A 2000,
104, 11463.
(47) Eliason, T. L.; Aloisio, S.; Donaldson, D. J.; Cziczo, D. J.; Vaida,
V. Atmos. EnViron. 2003, 37, 2207.
(48) Nepotchatykh, O. V.; Ariya, P. A. EnViron. Sci. Technol. 2002,
36, 3265.
(49) Ziemann, P. J. American Association for Aerosol Research Annual
Meeting, Anaheim, 2003.
(50) Allan, J. D.; Jimenez, J. L.; Williams, P. I.; Alfarra, M. R.; Bower,
K. N.; Jayne, J. T.; Coe, H.; Worsnop, D. R. J. Geophys. Res. 2003, 108,
4283.
(23) Seinfeld, J. H.; Pandis, S. N. Atmospheric Chemistry & Physics:
From Air Pollution to Climate Change; Wiley: New York, 1998.
(24) Andreae, M. O. Nature 2001, 409, 671.
(25) Andreae, M. O.; Crutzen, P. J. Science 1997, 276, 1052.
(26) Rudich, Y. Chem. ReV. 2003, 103, 5097.
(27) Moise, T.; Rudich, Y. J. Geophys. Res. 2000, 105, 14667.
(28) Moise, T.; Talukdar, R. K.; Frost, G. J.; Fox, R. W.; Rudich, Y. J.
Geophys. Res. 2002, 107, 1029.
(51) Allan, J. D.; Alfarra, M. R.; Bower, K. N.; Williams, P. I.;
Gallagher, M. W.; Jimenez, J. L.; McDonald, A. G.; Nemitz, E.; Canaga-
ratna, M. R.; Jayne, J. T.; Coe, H.; Worsnop, D. R. J. Geophys. Res. 2003,
108, 4284.
(52) Gelencser, A.; Hoffer, A.; Kiss, G.; Tombacz, E.; Kurdi, R.; Bencze,
L. J. Atmos. Chem. 2003, 45, 25.
(53) Duncan, J. L.; Schindler, L. R.; Roberts, J. T. J. Phys. Chem. B
1999, 103, 7247.
(54) Moise, T.; Rudich, Y. Geophys. Res. Lett. 2001, 28, 4083.
(29) Morris, J. W.; Davidovits, P.; Jayne, J. T.; Jime´nez, J. L.; Shi, Q.;
Kolb, C. E.; Worsnop, D. R.; Barney, W. S.; Cass, G. Geophys. Res. Lett.
2002, 29, 9.
(55) Raymond, T. M.; Pandis, S. N. J. Geophys. Res. 2002, 107, 4787.