Isomeric product detection in the heterogeneous reaction of hydroxyl radicals with aerosol composed of branched and linear unsaturated organic molecules

Article abstract of DOI:10.1021/jp508378z

(Figure Presented) The influence of molecular structure (branched vs linear) on product formation in the heterogeneous oxidation of unsaturated organic aerosol is investigated. Particle phase product isomers formed from the reaction of squalene (C₃₀H₅₀, a branched alkene with six C=C double bonds) and linolenic acid (C₁₈H₃₀O₂, a linear carboxylic acid with three C=C double bonds) with OH radicals are identified and quantified using two-dimensional gas chromatography-mass spectrometry. The reactions are measured at low and high [O₂] (～1% vs 10% [O₂]) to understand the roles of hydroxyalkyl and hydroxyperoxy radical intermediates in product formation. A key reaction step is OH addition to a C=C double bond to form a hydroxyalkyl radical. In addition, allylic alkyl radicals, formed from H atom abstraction reactions by hydroxyalkyl or OH radicals play important roles in the chemistry of product formation. Functionalization products dominate the squalene reaction at ～1% [O₂], with the total abundance of observed functionalization products being approximately equal to the fragmentation products at 10% [O₂]. The large abundance of squalene fragmentation products at 10% [O₂] is attributed to the formation and dissociation of tertiary hydroxyalkoxy radical intermediates. For linolenic acid aerosol, the formation of functionalization products dominates the reaction at both ～1% and 10% [O₂], suggesting that the formation and dissociation of secondary hydroxyalkoxy radicals are minor reaction channels for linear molecules. The distribution of linolenic acid functionalization products depends upon [O₂], indicating that O₂ controls the reaction pathways of the secondary hydroxyalkyl radical. For both reactions, alcohols are formed in favor of carbonyl functional groups, suggesting that there are some key differences between heterogeneous reactions involving allylic radical intermediates and those reactions of OH radicals with simple saturated hydrocarbons.

Full text of DOI:10.1021/jp508378z

Article
pubs.acs.org/JPCA
Isomeric Product Detection in the Heterogeneous Reaction of
Hydroxyl Radicals with Aerosol Composed of Branched and Linear
Unsaturated Organic Molecules
Theodora Nah,†,‡ Haofei Zhang,‡,§ David R. Worton,§,∥ Christopher R. Ruehl,‡ Benjamin B. Kirk,‡
Allen H. Goldstein,§,⊥,# Stephen R. Leone,†,‡,○ and Kevin R. Wilson*,‡
§
#
†Department of Chemistry, Department of Environmental Science, Policy, and Management, Department of Civil and
Environmental Engineering, and ○Department of Physics, University of California, Berkeley, California 94720, United States
‡Chemical Sciences Division and Environmental and Energy Technologies Division, Lawrence Berkeley National Laboratory,
⊥
Berkeley, California 94720, United States
∥Aerosol Dynamics Inc., Berkeley, California 94710, United States
S
* Supporting Information
ABSTRACT: The influence of molecular structure (branched vs linear) on product formation in the heterogeneous oxidation of
unsaturated organic aerosol is investigated. Particle phase product isomers formed from the reaction of squalene (C30H50, a
branched alkene with six CC double bonds) and linolenic acid (C18H30O2, a linear carboxylic acid with three CC double
bonds) with OH radicals are identified and quantified using two-dimensional gas chromatography−mass spectrometry. The
reactions are measured at low and high [O2] (∼1% vs 10% [O2]) to understand the roles of hydroxyalkyl and hydroxyperoxy
radical intermediates in product formation. A key reaction step is OH addition to a CC double bond to form a hydroxyalkyl
radical. In addition, allylic alkyl radicals, formed from H atom abstraction reactions by hydroxyalkyl or OH radicals play
important roles in the chemistry of product formation. Functionalization products dominate the squalene reaction at ∼1% [O2],
with the total abundance of observed functionalization products being approximately equal to the fragmentation products at 10%
[O2]. The large abundance of squalene fragmentation products at 10% [O2] is attributed to the formation and dissociation of
tertiary hydroxyalkoxy radical intermediates. For linolenic acid aerosol, the formation of functionalization products dominates the
reaction at both ∼1% and 10% [O2], suggesting that the formation and dissociation of secondary hydroxyalkoxy radicals are
minor reaction channels for linear molecules. The distribution of linolenic acid functionalization products depends upon [O2],
indicating that O2 controls the reaction pathways of the secondary hydroxyalkyl radical. For both reactions, alcohols are formed
in favor of carbonyl functional groups, suggesting that there are some key differences between heterogeneous reactions involving
allylic radical intermediates and those reactions of OH radicals with simple saturated hydrocarbons.
reaction mechanisms explain heterogeneous aerosol oxidation,
which is controlled to a large extent by particle phase properties
such as radical chain propagation reactions, thermodynamic
phase, and interfacial molecular orientation.3−6
1. INTRODUCTION
The heterogeneous oxidation of organic aerosol in the
atmosphere is complex due in large part to both the rich
array of organic molecules present in ambient aerosol and the
numerous free radical transformation pathways initiated by gas
phase oxidants (e.g., hydroxyl (OH), nitrate (NO3), and
chlorine (Cl) radicals).1,2 Gas and condensed phase reaction
mechanisms are typically used to interpret the formation of
particle phase products, yet it remains unclear how well these
Received: August 19, 2014
Revised: November 11, 2014
Published: November 11, 2014
© 2014 American Chemical Society
11555
dx.doi.org/10.1021/jp508378z | J. Phys. Chem. A 2014, 118, 11555−11571

The Journal of Physical Chemistry A
Article
To better understand fundamental OH-particle reactions,
several studies have investigated the products formed in the
OH reaction with saturated organic aerosol.4,5,7−15 These
studies found that oxidation proceeds via the abstraction of a
hydrogen (H) atom by an OH radical to form an alkyl radical.
In the presence of O2, the alkyl radical rapidly forms a peroxy
radical, which reacts through a series of functionalization and
fragmentation pathways to form higher and lower molecular
weight oxygenated products. Functionalization reactions
produce less volatile higher molecular weight products via the
addition of oxygenated functional groups (e.g., carbonyl and
hydroxyl groups) to the carbon backbone. Conversely,
fragmentation reactions produce lower molecular weight
oxygenated products via C−C bond scission along the carbon
backbone.
This work aims to provide new molecular and mechanistic
insights into the reaction pathways in the heterogeneous OH-
initiated oxidation of unsaturated organic aerosol through
detailed product analysis. Aerosols of squalene (C30H50, a liquid
branched alkene with six CC double bonds) and linolenic
acid (C18H30O2, a liquid linear carboxylic acid with three CC
double bonds) are photo-oxidized in a flow reactor and
analyzed using two-dimensional gas chromatography coupled
to electron impact ionization and vacuum ultraviolet photo-
ionization mass spectrometry. This technique separates and
quantifies the products by their volatility, polarity and mass
spectra. Figure 1 shows the molecular structures of squalene
The distribution of products formed from heterogeneous
oxidation depends strongly on molecular structure. Ruehl et al.4
analyzed the products formed by the reaction of OH with
squalane (a liquid branched C30 alkane) and octacosane (a solid
normal C28 alkane) aerosols and observed that the first
generation squalane and octacosane functionalization products
consist primarily of carbonyls and alcohols. Additionally, they
observed that the squalane reaction yielded significantly more
fragmentation products than octacosane. The carbon number
distribution of the squalane fragmentation products is
consistent with C−C bond scission reactions involving tertiary
carbons, so the smaller amount of octacosane fragmentation
products is attributed to the lack of tertiary carbons in
octacosane. Zhang et al.13 showed that the heterogeneous
reaction of OH with cholestane (a C27H48 solid cyclic alkane
with a branched aliphatic side chain) yielded few ring-opening
oxidation products. This is in contrast to gas phase reaction
mechanisms,16−21 which predict that ring-opening products
would be formed in large abundance due to the presence of
highly reactive tertiary and secondary carbons in the aliphatic
ring. These studies clearly indicate that the influence of
molecular structure on heterogeneous reaction mechanisms is
still not well understood and should be an area of focus for
future studies.
Unlike H atom abstraction by OH of aerosol composed of
saturated organic molecules, the heterogeneous OH-initiated
oxidation of unsaturated organic species mainly proceeds
primarily via OH addition to a CC double bond to form a
hydroxyalkyl radical.22−25 Previous kinetic studies of the
oxidation of unsaturated C18 fatty acids (oleic acid, linoleic
acid and linolenic acid)24 and squalene25 aerosols reported that
the hydroxyalkyl radical reacts though a sequence of
intermediates to form functionalization and fragmentation
products, a result broadly consistent with previous gas16,17,21
and monolayer22,23 studies. In addition, the effective uptake
coefficients, defined as the number of particle phase molecules
reacted per OH-particle collision, are found to be larger than
one for these reactions, providing clear evidence for particle-
phase secondary chain chemistry. These secondary chain
reactions may in turn alter the distribution of functionalization
and fragmentation particle phase products. Furthermore, the
formation of hydroxyperoxy radicals (from the reaction of
hydroxyalkyl radicals with O2), and their subsequent reaction
pathways to form stable products, provides a means of
understanding how reaction mechanisms might be altered by
the presence of adjacent oxygenated functional groups, which is
a common molecular feature in highly oxygenated organic
aerosol.10,26
Figure 1. Molecular structures of squalene and linolenic acid. The
carbon atoms in the carbon backbones of both molecules are labeled
and will be referred to accordingly in this paper.
and linolenic acid. The carbon atoms in the carbon backbones
of both molecules are labeled and referred to accordingly
throughout this paper. By measuring both reactions at high and
low [O2] (∼1% vs 10% [O2]), information can be obtained
about the roles of hydroxyalkyl and hydroxyperoxy radical
intermediates in product formation. This work builds on our
previous investigations, which focused broadly on the
heterogeneous chemical kinetics of the linolenic acid24 and
squalene25 reactions. Here we focus on the molecular and
positional isomeric distributions of the particle phase
functionalization and fragmentation products and how their
relative populations depend upon [O2]. As such the identities
and relative abundances of squalene and linolenic acid
oxidation products provide a more comprehensive under-
standing of how molecular structure (branched vs linear)
influences product distributions in the heterogeneous oxidation
of unsaturated organic aerosol. Even though gas phase products
are not directly measured in this study, the importance of
fragmentation reactions can still be inferred through detailed
analysis of the particle phase fragmentation products.
2. EXPERIMENT
2.1. Flow Tube Reactor Experiments. The experimental
setup used in this study has been described in detail
previously24,25 and will be reviewed briefly here. An
atmospheric pressure flow reactor (1.4 m long, 5.8 cm i.d.,
quartz tube) is used to investigate the heterogeneous oxidation
of squalene and linolenic acid aerosol. OH radicals are
generated in the flow reactor via the photolysis of H2O2 gas
using four continuous output 130 cm long Hg lamps (λ = 254
nm) that are placed outside and along the length of the flow
reactor. Aerosol particles are generated by homogeneous
nucleation in a N2 stream flowing through a pyrex tube
(heated in a furnace at 135 and 112 °C for squalene and
linolenic acid, respectively) containing either liquid squalene or
linolenic acid. The aerosol stream is passed through an annular
activated charcoal denuder to remove any residual gas phase
organics that may have been formed in the furnace. A N2/H2O2
gas mixture is introduced into the reactor by passing 0.15 L/
min of N2 through a heated bubbler (50 °C) that is packed with
11556
dx.doi.org/10.1021/jp508378z | J. Phys. Chem. A 2014, 118, 11555−11571

The Journal of Physical Chemistry A
Article
a 50:50 mixture of sand and urea-hydrogen peroxide.
Additional flows of humidified N2, O2 and tracer gas hexane
(∼150 ppb hexane in the flow reactor) are added to give a total
flow rate of 1 L/min entering the flow reactor. The relative
humidity in the flow reactor is fixed at 30%. The [H2O2] in the
flow reactor prior to reaction is ∼1 ppm. The [O2] in the flow
reactor are reported here as flow ratios (e.g., 10% [O2] is
obtained when 0.1 L/min of the total 1 L/min flow is O2).
The OH exposure in the flow reactor is changed by adjusting
the photon flux of the Hg lamps while keeping the total flow
rate constant. The OH exposure in the flow reactor is
determined from the decay of the tracer gas hexane, which is
measured using a gas chromatograph equipped with a flame
ionization detector (SRI Instruments), as described previously
by Smith et al.12 On the basis of the illuminated volume of the
reactor and the total flow rate going into the reactor (1 L/min),
the reaction time is ∼220 s. OH concentrations in the flow
reactor range from 7 × 107 to 9 × 108 mol cm−3. Particle size
distributions exiting the flow reactor are measured by a
scanning mobility particle sizer (SMPS, TSI model 3772). The
squalene and linolenic acid particle size distributions are log-
normal, with total mass concentrations of ∼8000 μg/m3, and
mean surface weighted diameters of ∼170 and 190 nm,
respectively. The mass concentrations of unreacted squalene
and linolenic acid (∼8000 μg/m3) are obtained from the SMPS
data using the densities of squalene (0.858 g/cm3) and linolenic
acid (0.914 g/cm3), respectively. At each OH exposure, aerosol
samples are collected for ∼30 min at 0.6 L/min onto quartz
filters (4.7 cm diameter Tissuquartz, Pall Life Science, prebaked
at 600 °C). The aerosol stream is passed through a charcoal
denuder (8 in. 480-channel MAST Carbon) before the filters to
prevent gas phase compounds from adsorbing onto the filters.
Particle phase organic compounds with less than ∼10 carbon
atoms are expected to be removed from the particle via
evaporation by the charcoal denuder.
(ToFWerk) using either EI ionization (70 eV) or VUV
photoionization (10.5 eV). The VUV radiation is produced by
the Chemical Dynamics Beamline at the Advanced Light
Source.
EI-MS and external standards are used with MSTFA
derivatization to calculate the response factors for GC × GC
chromatographic analysis and quantify oxidation products.
Since analytical standards are not commercially available for the
quantification of squalene and linolenic acid oxidation products,
external standards with retention times similar to these
products are used instead. This adds to the uncertainty of the
product quantification since the squalene and linolenic acid
oxidation products may have different response factors from the
external standards. This is especially true for linolenic acid
oxidation products, which have three or more oxygenated
functional groups, since the external standards used in this
study have only one or two oxygenated functional groups and
may not be suitable for accurate quantification of highly
oxygenated compounds due to differing response factors. In
addition, the response factors tend to decrease with decreasing
volatility and increasing polarity for the GC system used in this
study.4 Highly oxygenated compounds (with three or more
oxygenated functional groups) may also have lower GC column
transfer efficiencies.13 In this study, external standards of 12-
hydroxyoctadecanoic acid, palmitic acid, margaric acid, stearic
acid and behenic acid are used to correct for the volatility
response of linolenic acid products with alcohol functional
groups. Standards of triacontane, 1-hexadecanol, 1-octadecanol
and 1-eicosanol are used to correct for the volatility response of
squalene oxidation products with alcohol functional groups.
These standards are chosen since they have similar polarities as
these oxidation products (with alcohol functional groups) when
MSTFA is used as the derivatization reagent. 2-Pentadecanone
and 2-octadecanone standards are used correct for the polarity
response of squalene and linolenic acid oxidation products with
carbonyl functional groups.
2.2. Two-Dimensional Gas Chromatography with
High-Resolution Time-of-Flight Mass Spectrometry.
The gas chromatograph used to analyze the filters has been
described in detail previously4,13,27 and will be discussed briefly
here. Prior to analysis the filters are stored at −20 °C. The
filters are analyzed using two-dimensional gas chromatography
and detected by high-resolution time-of-flight mass spectrom-
etry utilizing either electron impact (EI) ionization or vacuum
ultraviolet (VUV) photoionization (GC × GC-HRTOF-EI/
VUVMS). The filters (sampled area = 0.41 cm2) are thermally
desorbed at 320 °C in helium using a thermal desorption
system and autosampler (TDS3 and TDSA2, Gerstel) and
introduced into a two-dimensional gas chromatograph (GC ×
GC, Agilent 7890). The helium stream is saturated with the N-
methyl-N-trimethylsilyl trifluoroacetamide (MSTFA) derivati-
zation reagent to convert hydroxyl groups into trimethylsilyl
ethers that are amenable to GC analysis.
Analytes are first separated by volatility using a 60 m × 0.25
mm × 0.25 μm nonpolar capillary column (Rxi-5Sil MS,
Restek, 85 min retention time), then by polarity using a 1 m ×
0.25 mm × 0.25 μm medium polarity column (Rtx-200MS,
Restek, 2.3 s retention time). The first column is connected to
the second column using a dual loop modulator (1.5 m × 0.25
mm Rxi guard column, Zoex Corp, 2.4 s modulation time).
Internal standards of deuterated n-alkanes are used to correct
for GC column transfer efficiency and chromatogram
reproducibility. Mass spectra are obtained using a high-
resolution (Δm/m ∼ 4000) time-of-flight mass spectrometer
VUV-MS is used to identify specific compounds and
structural isomers. The VUV mass spectrum generally features
a more prominent molecular ion peak and significantly fewer
fragment ions compared to EI. Although the molecular
formulas of the compounds are determined from high-
resolution VUV mass spectra, for simplicity, only integer m/z
values are presented in this paper. Shown as an example in
Figure S1 (Supporting Information) are VUV mass spectra
obtained for derivatized linolenic acid (Figure S1a) and its diol
products (Figure S1b−d). Using MSTFA as the derivatization
reagent results in the acidic H atoms in alcohol and carboxylic
acid functional groups to be substituted by a trimethylsilyl
(Si(CH3)3, labeled as “TMS”) group. For simplicity, the TMS-
derivatized hydroxyl (OH) groups in alcohol and carboxylic
acid functional groups are labeled as “OTMS” in this paper. In
Figure S1 (Supporting Information), the molecular ions of
linolenic acid (C17H29COOH, m/z 278) and its diol product
isomers (C17H29CO(OH)3, m/z 312) appear as C17H29CO-
(OTMS) (m/z 350) and C17H29CO(OTMS)3 (m/z 528),
respectively.
3. RESULTS AND DISCUSSION
The product analysis of the squalene and linolenic acid
reactions are discussed in sections 3.1 and 3.2, respectively.
Section 3.1.1 provides an overview of the squalene two-
dimensional GC (GC × GC) chromatogram. In this section the
oxidation products are presented in broad groupings by
11557
dx.doi.org/10.1021/jp508378z | J. Phys. Chem. A 2014, 118, 11555−11571

The Journal of Physical Chemistry A
Article
Figure 2. Relative abundance of squalene and its oxidation products shown in GC × GC space at OH exposure =1.62 × 1011 molecules cm−3 s at (a)
1
2
∼ 1% and (b) 10% [O2]. The x-axis is the first dimension (tR ) and the y-axis is the second dimension (tR ) retention time. In panels a and b, each
circle represents a single compound and its relative abundance is represented by the size of the circle. Squalene C30H50 is the orange circle located at
tR1 ∼ 67.5 min and tR2 ∼ 0.7 s. The oxidation products are separated into four groups by their chemical formulas and delineated with different colors:
all fragmentation products (black); derivatized C30H49(OTMS) alcohols (green); derivatized C30H51(OTMS) alcohols (blue); derivatized
C30H50(OTMS)2 diols (red).
Figure 3. Total mass concentrations of squalene, oxidation products, functionalization products and fragmentation products as a function of OH
exposure at (a) ∼ 1% [O2], and (b) 10% [O2]. Error bars for the squalene mass concentrations originate from the filter sampling process. The
squalene decays in parts a and b are fit to an exponential function. Total mass concentrations of C30H49OH alcohol, C30H51OH alcohol and
C30H50(OH)2 diol products as a function of OH exposure at (c) ∼ 1% [O2], and (d) 10% [O2].
molecular formula (i.e., functionalization and fragmentation) to
show in a general way how the aerosol composition evolves as a
function of OH exposure. This overview is followed by detailed
isomeric analysis of individual products: C30H51OH and
C30H49OH alcohols (3.1.2) and C30H50(OH)2 diols (3.1.3).
In section 3.1.4, the carbon number distribution of the squalene
11558
dx.doi.org/10.1021/jp508378z | J. Phys. Chem. A 2014, 118, 11555−11571

The Journal of Physical Chemistry A
Article
fragmentation products is presented. Section 3.2 on the
linolenic acid reaction is similarly subdivided into an overview
of the linolenic acid GCXGC chromatogram (3.2.1), isomeric
product analysis of C17H30CO(OH)2 and C17H28CO(OH)2
alcohols (3.2.2), C17H28CO2(OH)2 hydroxycarbonyls and
C17H29CO(OH)3 diols (3.2.3), and fragmentation products
(3.2.4).
products are functionalization products, and approximately 11%
are fragmentation products. This indicates that the addition of
oxygenated functional groups to the carbon backbone (i.e.,
functionalization) dominates the heterogeneous reaction at
∼1% [O2].
Figure 3b shows that at 10% [O2] the total mass
concentration of detected oxidation products reaches a
maximum value of ∼4900 μg/m3 at an OH exposure of ∼9.6
× 1010 molecules cm−3 s, before decreasing upon further
oxidation. Under these conditions, the total abundance of
observed functionalization products is approximately equal to
the fragmentation products, indicating that C−C bond cleavage
(i.e., fragmentation) is important in the presence of 10% [O2].
The total mass concentration of detected oxidation products
is compared to the reactive loss of squalene. At ∼ 1% [O2]
(Figure 3a), ∼ 90% of the original squalene mass concentration
can be accounted for (by particle phase products and unreacted
squalene) over the course of the reaction. The remaining mass
can be attributed to gas phase fragmentation products (which
are not collected on filters). At 10% [O2] (Figure 3b), only
∼80% of the original squalene mass concentration can be
accounted for during oxidation. This is likely due to particle
volatilization being an important reaction channel at high [O2],
as demonstrated by the detection of higher mass concentrations
of fragmentation products at 10% [O2]. This is also consistent
with aerosol mass measurements presented previously by Nah
et al.,25 which showed that significant particle volatilization
occurs at high [O2] in the squalene reaction (∼5% particle mass
loss at ∼0% [O2] vs ∼20% particle mass loss at 10% [O2]).
The squalene mass concentration vs OH exposure (Figure 3,
parts a and b) can be fit by an exponential function. Similar to
the previous kinetics study of the squalene reaction,25 these
exponential decays indicate that the decay rate is directly
proportional to the concentration of particle phase squalene
and the reaction is not limited by the diffusion of squalene
molecules to the particle surface. As described previously by
Nah et al.,25 the squalene decay rate constants can be used to
3.1. Squalene Oxidation Product Analysis. 3.1.1. Squa-
lene GC × GC Chromatogram. Single ion chromatograms
(extracted from the GC × GC run performed using VUV-MS)
are used to identify product isomers via their molecular ions
and resolve coeluting species. These product isomers are then
quantified in EI-MS using their total GC volumes. Figure 2
shows the distribution and relative abundance of particle phase
squalene oxidation products in GC × GC space (volatility tR1 vs
polarity tR ) at an OH exposure of 1.6 × 1011 mol cm−3 s for
2
∼1% and 10% [O2] (Figure 2, parts a and b, respectively). Each
circle signifies a single compound while its relative abundance is
represented by the size of the circle.
Squalene C30H50 (m/z 410, shown in orange in Figure 2)
1
elutes at first-dimensional retention time (tR ) ∼ 67.6 min and
2
second-dimensional retention time (tR ) ∼ 0.7 s. Fragmentation
products (shown in black in Figure 2) have carbon numbers
1
less than 30 and elute at shorter tR relative to squalene,
indicating that they are more volatile than squalene. The most
prominent fragmentation products have carbon numbers 17,
22, and 27 (discussed in detail in section 3.1.4). Functionaliza-
tion products with carbon number 30 and new oxygenated
1
functional groups (formed in the reaction) elute at longer tR
relative to squalene.
The functionalization products are grouped based on their
molecular formulas. At both ∼1% and 10% [O2], two distinct
groups of TMS-derivatized alcohol products are detected:
C30H49(OTMS) (m/z 498, shown in green in Figure 2) and
C30H51(OTMS) (m/z 500, shown in blue in Figure 2). Twelve
C30H49(OTMS) alcohol isomers are resolved at tR2 ∼ 0.7 s with
tR1 from ∼68 to 73 min. Three C30H51(OTMS) alcohol isomers
are detected at tR2 ∼ 0.7 s with tR1 from ∼70 to 72 min. Three
TMS-derivatized C30H50(OTMS)2 diol isomers (m/z 588,
calculate an effective uptake coefficient (γSOqHe ). Using the
Sqe
OH
methodology detailed in Nah et al., γ measured at ∼1% and
2
shown in red in Figure 2) are also detected at tR ∼ 0.7 s with
10% [O2] are 2.92 0.18 and 3.65 0.22, respectively. This
indicates that more than one particle phase squalene molecule
is lost for every OH-particle collision and is clear evidence for
1
tR from ∼73 to 75 min. MSTFA derivatization decreases the
polarity of the alcohol and diol products, causing them to have
2
approximately the same tR as squalene.
secondary chain chemistry. Because of unknown amounts of
Sqe
OH
The relative abundance of particle phase reaction products at
∼1% and 10% [O2] suggests that [O2] controls the formation
pathways of functionalization and fragmentation products. To
obtain a more comprehensive understanding of how [O2]
influences the competition between functionalization and
fragmentation pathways, the mass concentrations of detected
squalene particle phase oxidation products are quantified using
external standards. Parts a and b of Figure 3 show the mass
concentrations of squalene and its detected oxidation products,
which are subdivided into total functionalization and
fragmentation products, as a function of OH exposure at
∼1% and 10% [O2], respectively. Error bars for the oxidation
product mass concentrations are not shown since there is some
uncertainty on how the detection efficiencies of the external
standards differ from that of the products.
secondary chemistry, the γ values are not corrected for gas
phase diffusion.
Sqe
OH
The measured γ values are larger than those previously
reported by Nah et al.25 (1.55 0.05 and 2.59 0.12 for 1.4
ppm [H2O2] at ∼0% and 10% [O2], respectively). These
differences most likely arise from differences in the radical chain
propagation length, whose magnitude is sensitive to reaction
time and oxidant concentration. For example, previously it has
Sqe
OH
been shown that γ is highly dependent on O2, H2O2 and OH
concentrations in the reactor.25 In this study, the reaction time
is longer and the average [OH] in the flow reactor is smaller
than those used in Nah et al.25 Che et al.14 observed larger
effective uptake coefficients in the OH + squalane reaction
when longer reaction times and lower average [OH] are used,
and they attributed this to secondary reactions. Here it is
possible that the differences in the experimental conditions led
Figure 3a shows that at ∼1% [O2] the total mass
concentration of detected oxidation products increases and
reaches a maximum value of ∼5370 μg/m3 at an OH exposure
of ∼9.6 × 1010 molecules cm−3 s, before decreasing upon
further oxidation. Approximately 89% of the observed oxidation
to differences in the radical chain propagation length, and
Sqe
consequently differences in the γ values. Nonetheless the
OH
Sqe
OH
increase in γ when [O2] increases from ∼1% to 10% in this
study is broadly consistent with the previous study of the
11559
dx.doi.org/10.1021/jp508378z | J. Phys. Chem. A 2014, 118, 11555−11571

The Journal of Physical Chemistry A
Article
squalene reaction25 and indicates that O2 promotes chain
propagation.
Although the hydroxyalkyl/alkyl radical + hydroxyperoxy/
peroxy radical (R + RO2), hydroxyalkyl/alkyl radical +
hydroperoxyl radical (R + HO2) and hydroxyperoxy/peroxy
radical + hydroperoxyl radical (RO2 + HO2) reactions (which
form organic peroxides ROOR and ROOH) have been
observed in gas phase reactions,16,17,21 we see no evidence of
ROOR and ROOH products in this study despite the estimated
average [HO2] being approximately equal to the average [OH]
in the flow reactor (estimated using a kinetic model). ROOR
and ROOH products are also not detected in our previous
study of the squalene reaction.25 Therefore, these chain
termination reactions appear to be minor channels, within
our ability to detect these products, in the squalene reaction,
and are not included in the reaction mechanism presented in
Figure 4. It is possible that the measurement techniques used in
this study and our previous study of the squalene reaction25
may not be sensitive to ROOR and ROOH products, however
similar observations that the R + RO2, R + HO2, and RO2 +
HO2 reactions are potentially minor channels in heterogeneous
reactions have been made in previous heterogeneous aerosol
oxidation studies by Smith et al.,12 McNeill et al.,11 and Ruehl
et al.4 The reactions of hydroxyalkoxy and alkoxy radicals with
HO2 radicals to form hydroxycarbonyls and carbonyls,
respectively, are also important pathways in gas phase
reactions.16,17,21 However, carbonyls and hydroxycarbonyls
are not detected in the squalene GC × GC chromatograms.
This suggests that they are likely minor channels in the
squalene reaction, and are therefore not included in the
reaction mechanism shown in Figure 4.
It is important to note that high [OH] and short reaction
times (with respect to ambient conditions) are used in this
product formation study. These experimental conditions may
cause the measured branching ratios among the various
reaction pathways to differ from that observed in the
atmosphere. For example, the importance of hydroxyperoxy
radical self- and cross-reactions may be larger in this study than
in ambient conditions due to the higher radical concentrations
used here. Nonetheless, since the experimental conditions
employed in this study for squalene and linolenic acid oxidation
are similar, we can directly compare the identities and relative
abundances of squalene and linolenic acid oxidation products to
gain an understanding of how molecular structure (branched vs
linear) influences product distributions in the heterogeneous
oxidation of unsaturated organic aerosol.
The observed functionalization products are composed of
TMS-derivatized C30H49(OTMS), C30H51(OTMS) and
C30H50(OTMS)2 product isomers. The underivatized molecular
formulas of these products (obtained by substituting the
“TMS” group with a H atom) are C30H49OH, C30H51OH and
C30H50(OH)2, respectively. Previous kinetic studies25 indicate
that these are first generation functionalization products. Parts c
and d of Figure 3 show the evolution of C30H49OH, C30H51OH,
and C30H50(OH)2 as a function of OH exposure at ∼1% and
10% [O2], respectively. At ∼ 1% [O2], C30H49OH, C30H51OH
and C30H50(OH)2 make up approximately 46%, 22% and 32%
of the total mass concentration of detected functionalization
products, respectively (Figure 3c). At 10% [O2], C30H49OH,
C30H51OH and C30H50(OH)2 make up approximately 47%,
11%, and 42% of the total mass concentration of detected
functionalization products, respectively (Figure 3d). Carbonyls
and TMS-derivatized hydroxycarbonyls are not detected at
either ∼1% or 10% [O2], indicating that they are minor
reaction products.
The formation of first generation C30H49OH, C30H51OH,
and C30H50(OH)2 functionalization products, together with the
absence of hydroxycarbonyl products in the GC × GC
chromatogram and the increased abundance of fragmentation
products at 10% [O2] (relative to at ∼1% [O2]) oxidation can
be explained by the reaction mechanism proposed previously
by Nah et al.25 In that work, it is shown that the OH + squalene
reaction is initiated predominantly by OH addition to the less
substituted carbon of the CC double bond to form tertiary
hydroxyalkyl radicals. Figure 4 shows the reaction mechanism
3.1.2. C30H51OH and C30H49OH Alcohol Product Isomers.
Three TMS-derivatized C30H51(OTMS) alcohol isomers are
detected at ∼1% and 10% [O2]. The underivatized molecular
formula of these alcohol isomers C30H51OH indicates that one
O atom and two H atoms are added to the squalene molecule.
On the basis of the reaction mechanism shown in Figure 4,
these C30H51OH alcohol products are formed via R2 where a H
atom is abstracted from a neighboring squalene molecule by the
hydroxyalkyl radical.
At ∼ 1% [O2], the total mass concentration of C30H51OH
alcohol isomers increases to a maximum of ∼1150 μg/m3 (OH
exposure ∼9.6 × 1010 molecules cm−3 s, Figure 3c). At 10%
[O2], the total mass concentration of C30H51OH alcohol
isomers reaches a maximum of ∼260 μg/m3 (OH exposure
∼7.3 × 1010 molecules cm−3 s, Figure 3d). Thus, the total
abundance of these alcohol isomers is approximately four times
larger at ∼1% [O2] compared to 10% [O2], indicating that
increasing O2 suppresses their formation pathways. This is
consistent with the reaction mechanism (Figure 4), where the
Figure 4. Reaction mechanism for the OH-initiated oxidation of
squalene (Sqe) aerosol. R• represents the generic alkyl radical formed
by the abstraction of a H atom from a squalene molecule. The
underivatized (bold text) and derivatized (regular text) molecular
formulas of the detected first generation functionalization products are
also shown.
for the OH-initiated oxidation of squalene aerosol. For clarity,
the derivatized and underivatized molecular formulas of the
products are also shown. This reaction scheme is used to
understand the distribution of isomeric products described in
sections 3.1.2−3.1.4.
11560
dx.doi.org/10.1021/jp508378z | J. Phys. Chem. A 2014, 118, 11555−11571

The Journal of Physical Chemistry A
Article
formation of hydroxyperoxy radicals (R5) becomes the
dominant reaction pathway for hydroxyalkyl radicals at high
[O2], effectively slowing down R2, the chain propagation
reaction that produces C30H51OH alcohol products.
Parts a−c of Figure 5 show the VUV mass spectra of the
three detected TMS-derivatized C30H51(OTMS) (m/z 500)
the hydroxyl group) along the squalene carbon backbone can
be ascertained. For example, Figure 5a shows the VUV mass
spectrum of a TMS-derivatized C30H51(OTMS) alcohol isomer
with an OTMS group on C3 of the squalene carbon backbone.
The α-cleavage process is the dominant ion fragmentation
pathway, producing m/z 145 (C4H8(OTMS)+) and 457
(C27H44(OTMS)+) fragment ions. Using this approach, the
VUV mass spectra shown in Figure 5, parts b and c, are
determined to be C30H51(OTMS) alcohol isomers with an
OTMS group on C7 and C11 of the squalene carbon backbone,
respectively. Together these mass spectra indicate that these
three alcohol isomers all originate from OH addition to the less
substituted carbon of the CC double bond.
C30H51(OTMS) alcohol isomers with the OTMS group on
C11 of the squalene carbon backbone make up the majority
(∼50%) of the structural isomers. This suggests that C30H51OH
alcohol isomers with the hydroxyl group on C11 of the
squalene carbon backbone are formed preferentially (compared
to the other two isomers) during the reaction. It is currently
unclear why this is the case and should be an area of focus for
further investigation.
C30H51(OTMS) alcohol isomers with OTMS groups located
on C2, C6 and C10 of the squalene carbon backbone are not
detected, indicating that C30H51OH isomers formed via OH
addition to the more substituted carbon of the CC double
bond are minor reaction products. These results suggest more
generally that OH addition occurs at the less substituted carbon
of the CC double bond (to form tertiary hydroxyalkyl
radicals), while OH addition to the more substituted carbon of
the CC double bond (to form secondary hydroxyalkyl
radicals) is a minor channel. OH addition to the less substituted
carbon of the CC double bond is also the dominant channel
in the oxidation of gas phase unsaturated organics by OH
radicals.19
The higher propensity for OH addition to the less
substituted carbon of the CC double bond is also consistent
with the absence of TMS-derivatized C30H49O(OTMS)
hydroxycarbonyl products from the GC × GC chromatogram
and the large abundance of fragmentation products at 10%
[O2]. On the basis of the reaction mechanism shown in Figure
4, the preference for OH addition to the less substituted carbon
of the CC double bond leads to high yields of tertiary
hydroxyperoxy radicals (relative to secondary hydroxyperoxy
radicals) at high [O2] (R5). The formation of hydroxycarbonyl
products (via the Russell and Bennett−Summers mecha-
nisms28−32) is suppressed in the tertiary hydroxyperoxy radical
self-reaction due to the lack of H atoms adjacent to the peroxy
radical site. Instead tertiary hydroxyalkoxy radicals are likely the
dominant products from tertiary hydroxyperoxy radical self-
reaction (R7). These tertiary hydroxyalkoxy radicals can
dissociate to form fragmentation products (R10), which may
explain the high yields of fragmentation products at 10% [O2]
(relative to ∼1% [O2]).
In addition to self-reaction, tertiary hydroxperoxy radicals can
also react with secondary hydroxyperoxy radicals and peroxy
radicals (formed from the reaction of O2 with alkyl radicals that
are formed in reactions R2 and R8) to form diols (R6) and
tertiary hydroxyalkoxy radicals (R7). As discussed above, our
results suggest that secondary hydroxyalkyl and hence
secondary hydroxyperoxy radicals are formed in low abundance
relative to tertiary hydroxyalkyl and tertiary hydroxyperoxy
radicals. Therefore, the reaction between tertiary and secondary
hydroxyperoxy radicals is expected to be minor channel, while
Figure 5. (a−c) VUV mass spectra of the TMS-derivatized
C30H51(OTMS) alcohol product isomers (m/z 500). Figure insets in
panels a−c show the ion fragmentation patterns that explain the
characteristic ion peaks observed in the mass spectra. The McLafferty
rearrangement and H rearrangement are labeled as “ML” and “H
rearrang.” respectively. In panels a−c, the m/z 410 fragment ion is
formed from the loss of HOTMS. The ion peak m/z 131 in panel c is a
C3H6OTMS fragment. Also shown are the labeled carbon atoms in the
squalene carbon backbone.
alcohol isomers. Also shown are the ion fragmentation patterns
that explain the characteristic ion peaks observed in the VUV
mass spectra. The observed fragment ions in the mass spectra
are produced by α-cleavage, β-cleavage, McLafferty rearrange-
ment, and H rearrangement. The m/z 410 ion is a prominent
fragment ion in the three VUV mass spectra, and it corresponds
to the loss of (CH3)3SiOH (labeled “HOTMS”). The loss of
HOTMS is similar to the dehydration of alcohols. Similar
fragment ions in the VUV mass spectra of TMS-derivatized
alcohols have also been observed by Zhang et al.13 in their
study of cholestane aerosol oxidation.
By identifying distinct fragment ions in the VUV mass
spectra of these three TMS-derivatized C30H51(OTMS) alcohol
isomers, the location of the OTMS group (and consequently
11561
dx.doi.org/10.1021/jp508378z | J. Phys. Chem. A 2014, 118, 11555−11571

The Journal of Physical Chemistry A
Article
reactions between tertiary hydroxyperoxy radicals and peroxy
radicals are expected to play more important roles in product
formation. This is another reason for the absence of TMS-
derivatized C30H49O(OTMS) hydroxycarbonyl products from
the GC × GC chromatogram.
The role that tertiary hydroxyalkoxy radical intermediates
plays in radical chain chemistry at high [O2] is demonstrated by
the measured γSOqHe , which increases from 2.92 0.18 at ∼1%
[O2] to 3.65 0.22 at 10% [O2]. On the basis of the reaction
mechanism shown in Figure 4, pathways R2 and R8 are chain
propagation reactions that accelerate the reactive depletion of
squalene molecules at high [O2]. At low [O2], pathway R8 is
suppressed since hydroxyperoxy radicals (and consequently
hydroxyalkoxy radicals) are formed in low abundance. There-
maximum of ∼2010 μg/m3 and ∼1130 μg/m3 at an OH
exposure of ∼9.6 × 1010 molecules cm−3 s at ∼1% and 10%
[O2], respectively. On the basis of the reaction mechanism
shown in Figure 4, these C30H49OH alcohol product isomers
are formed by the reaction of alkyl (R) radicals (formed by H
atom abstraction reactions R2 and R8) with O2 and OH
radicals (R3 and R9). The R radicals react with O2 to form
peroxy radicals, which undergo self- and cross-reactions to form
C30H49OH alcohol products.28−32 It is possible that the R
radicals are also formed by H atom abstraction reactions by OH
radicals (for brevity this pathway is not explicitly shown in
Figure 4), even though studies by Atkinson et al.33,34 suggest
that this is a minor reaction pathway. It remains unclear how
well these studies on small gas phase alkenes (≤C5) extrapolate
to the heterogeneous oxidation of large alkenes (i.e., squalene).
Squalene has 44 H atoms in C(sp3)−H bonds available for
abstraction. Abstraction of H atoms attached to the CC
double bonds are expected to be energetically unfavorable.35,36
The 44 H atoms from the C(sp3)−H bonds in squalene are
allylic as they are attached to C atoms adjacent to CC double
bonds. Equation 1 illustrates resonance stabilization of allylic R
radicals formed from H atom abstraction.
fore, the radical chain chemistry in the particle is slowed with
Sqe
OH
decreasing [O2], causing γ
to decrease when [O2] is
decreased.
Twelve TMS-derivatized C30H49(OTMS) alcohol isomers
are detected at ∼1% and 10% [O2]. The underivatized
molecular formula of these alcohol isomers C30H49OH
corresponds to the addition of one O atom to the squalene
molecule. Figure 3, parts c and d show that the total mass
concentration of C30H49OH alcohol isomers reaches a
As a result of resonance stabilization, 16 allylic R radicals
could be formed from H atom abstraction reactions (shown in
Figure S2 in the Supporting Information). To qualitatively
characterize the probability at which each resonance structure
forms, quantum chemical calculations are performed using the
Gaussian 0937 program (shown and described in Figure S3 in
the Supporting Information). A smaller C7H14 squalene
analogue with three allylic positions is used for these
calculations, as high-level calculations of the larger molecule
are unfeasible. The quantum chemical calculations show that
the unpaired electron resulting from H atom abstraction is
delocalized across the adjacent double bond (i.e., the electron
density is evenly distributed) resulting in two regions of
significant electron density as expected of an allylic R radical.
This suggests that the resonance-stabilized allylic R radicals
formed from intermolecular H atom abstraction reactions
during squalene oxidation are delocalized. As a result one might
expect to observe 16 C30H49OH alcohol isomers formed from
the reaction of allylic R radicals with O2 and OH radicals.
However, only 12 TMS-derivatized C30H49(OTMS) alcohol
isomers are detected at ∼1% and 10% [O2].
resulting in only 12 C30H49OH alcohols being formed. It is
important to note that without the resonance stabilization of
allylic R radicals, only a maximum of eight different TMS-
derivatized C30H49(OTMS) alcohol isomers would be
observed. Therefore, the detection of 12 TMS-derivatized
C30H49(OTMS) alcohol isomers demonstrates that these allylic
R radicals undergo resonance stabilization. Since the mass
spectra of these 12 isomers contain similar fragment ions, we
are not able to assign structural isomers.
Previous studies28−32,38 have shown that R radicals react with
O2 to form peroxy radicals, which self- and cross-react to form
alcohol (C30H49OH) or carbonyl (C30H48O) products.
C30H48O carbonyls are not detected in the GC × GC
chromatograms, indicating that they are minor reaction
products. Previous condensed phase studies32,38−40 have
similarly observed the formation of excess alcohol products
(relative to carbonyl products) from peroxy radical self-
reaction. These studies reported that a labile tetroxide
intermediate is formed by peroxy radical self-reaction. The
subsequent homolytic cleavage of this tetroxide intermediate
into carbonyls, alcohols and alkoxy radicals (i.e., Russell and
Bennett−Summers mechanisms28−32,38−40) depends on the
structure of the initial peroxy radical, the temperature of the
reaction and the viscosity of the solution. Thus, the
alcohol:carbonyl product ratios are also observed to be
influenced by the temperature of the reaction and the viscosity
of the solution. Although previous condensed phase stud-
One possible explanation for the detection of only 12
(instead of 16) TMS-derivatized C30H49(OTMS) alcohol
isomers is that some of these alcohol isomers may have similar
volatilities, causing them to coelute. It is also possible that some
of the allylic R radicals (i.e., tertiary vs secondary vs primary)
are more reactive toward O2 and OH radicals than others,
11562
dx.doi.org/10.1021/jp508378z | J. Phys. Chem. A 2014, 118, 11555−11571

The Journal of Physical Chemistry A
Article
ies32,38,39 have reported alcohol:carbonyl ratios of up to 2, our
study suggest that this ratio is significantly higher for the
squalene reaction. Since the formation of alcohols via peroxy
radical chemistry requires the abstraction of H atoms by peroxy
radicals, the high alcohol:carbonyl ratio in this study suggests
that the peroxy radicals may be abstracting H atoms from other
species other than peroxy radicals. While it is possible that,
analogous to condensed phase studies, molecular structure,
temperature of the reaction and the viscosity of the particle may
be responsible for the very low yields of C30H48O carbonyl
products, further work is clearly needed to better understand
peroxy radical chemistry in systems that form allylic alkyl
radicals.
lower molecular weight oxygenated products may evaporate
from the particle into the gas phase, leading to a detectable loss
in aerosol mass.10,12,15
To identify which C−C bonds in squalene are most
susceptible to bond cleavage during oxidation, the mass
concentrations of the particle phase fragmentation products
are plotted as a function of carbon number in Figure 6a. These
3.1.3. C30H50(OH)2 Diol Product Isomers. Three TMS-
derivatized C30H50(OTMS)2 (m/z 588) diol isomers are
detected at ∼1% and 10% [O2]. The underivatized molecular
formula of these diol isomers C30H50(OH)2 corresponds to the
addition of two O atoms and two H atoms to the squalene
molecule. The formation of three isomers is consistent with the
number of unique CC double bonds in squalene. At ∼ 1%
[O2], the total mass concentration of C30H50(OH)2 diol
isomers reaches a maximum of ∼1810 μg/m3 (OH exposure
∼1.5 × 1011 molecules cm−3 s, Figure 3c). At 10% [O2], the
total mass concentration of C30H50(OH)2 diol isomers reaches
a maximum of ∼1140 μg/m3 (OH exposure ∼9.6 × 1010
molecules cm−3 s, Figure 3d). On the basis of the reaction
mechanism shown in Figure 4, these C30H50(OH)2 diol
product isomers are formed by chain termination reactions
between OH radicals and hydroxyalkyl radicals (R4) at ∼1%
and 10% [O2]. At high [O2], C30H50(OH)2 diols can also be
formed by self- and cross-reactions of hydroxyperoxy radicals
(R6) and H atom abstraction reactions by hydroxyalkoxy
radicals (R8).
Figure S4a−c (Supporting Information) shows the VUV
mass spectra of the three TMS-derivatized C30H50(OTMS)2
diol isomers, and Figure S4d (Supporting Information) shows
the relative abundance of the diol isomers. C30H50(OTMS)2
diol isomers with OTMS groups on C2 and C3 of the squalene
carbon backbone make up the majority (∼50%) of the
structural isomers, suggesting that these diol isomers with
hydroxyl groups located at the ends of the molecule are formed
preferentially during the reaction. Ruehl et al.4 also observed an
overall enhancement in hydroxyl groups formed toward the
ends of the molecule in the heterogeneous reaction of OH
radicals with squalane aerosol. They attributed this to alkoxy
radical sites located toward the ends of the molecule being
more exposed to other molecules and are thus more likely to
undergo intermolecular H atom abstraction. Analogous to the
squalane reaction, it is possible that intermolecular H atom
abstraction reactions by hydroxyalkoxy radicals in the squalene
reaction (R8 in Figure 4) contribute to the observed
distribution of diol product isomers.
3.1.4. Squalene Fragmentation Products. Figure 3 shows
that there are larger quantities of particle phase fragmentation
products formed during oxidation at 10% [O2] relative to ∼1%
[O2]. The higher abundance of particle phase fragmentation
products at 10% [O2] is consistent with the reaction
mechanism shown in Figure 4 and the previous study of the
OH + squalene reaction,25 both of which show that O2
promotes fragmentation reactions and aerosol mass loss.
These fragmentation reactions produce more volatile lower
molecular weight oxygenated products via C−C bond scission
along the carbon backbone. In some cases, the more volatile
Figure 6. (a) Squalene fragmentation products as a function of carbon
number at 10% [O2] and an OH exposure of 7.32 × 1010 molecules
cm−3 s (oxidation conditions at which maximum fragmentation
product concentration is obtained). (b) Schematic of the C−C bonds
in hydroxyalkoxy radicals that are most likely to undergo scission to
form fragmentation products with carbon numbers (I) 27, (II) 22 and
24, and (III) 17 and 19. Also shown in part b are the labeled carbon
atoms in the squalene carbon backbone.
classifications are made based on their retention times and
parent molar masses. This analysis does not include
fragmentation products with less than 10 carbon atoms since
they are too volatile to be collected on filters. The analysis also
does not include gas phase products (which are not measured)
that may be formed during fragmentation reactions. However,
previous studies of the squalene system25 show that substantial
aerosol mass loss occurs even at low OH exposures, indicating
that the formation and subsequent loss of volatile lower
molecular weight oxygenated products to the gas phase is an
important reaction channel.
Figure 6a shows that the particle phase fragmentation
products are made up of alcohols and hydroxycarbonyls. The
majority of fragmentation products contain multiple hydroxyl
(OH) and carbonyl (CO) functional groups. The observa-
tion that most of the fragmentation products are multifunc-
tional is expected, since these fragmentation products still have
11563
dx.doi.org/10.1021/jp508378z | J. Phys. Chem. A 2014, 118, 11555−11571

The Journal of Physical Chemistry A
Article
Figure 7. Relative abundance of linolenic acid and its oxidation products shown in GC × GC space at OH exposure =1.62 × 1011 molecules cm−3
s
1
2
at (a) ∼ 1% and (b) 10% [O2]. The x-axis is the first dimension (tR ) and the y-axis is the second dimension (tR ) retention time. In panels a and b,
each circle represents a single compound and its relative abundance is represented by the size of the circle. Derivatized linolenic acid
1
2
C17H29CO(OTMS) is the orange circle located at tR ∼ 54.5 min and tR ∼ 0.7 s. The oxidation products are separated into five groups by their
chemical formulas and delineated with different colors: all fragmentation products (black); derivatized C17H28CO(OTMS)2 alcohols (green);
derivatized C17H30CO(OTMS)2 alcohols (blue); derivatized C17H29CO(OTMS)3 diols (red); derivatized C17H28CO2(OTMS)2 hydroxycarbonyls
(purple).
CC double bonds that are susceptible to further reaction
with OH radicals and O2. Majority of the fragmentation
products have carbon number 20 and above. The measured
abundance of products with carbon numbers less than 20 may
be lower than their actual abundance (formed during the
reaction) since some of these products may have evaporated
from the particle prior to filter collection and analysis.
Approximately 78% of the total fragmentation product mass
concentration consists of products with carbon numbers 17, 22
and 27, indicating that the C10−C11, C6−C7, and C2−C3
bonds in the squalene carbon backbone are most susceptible to
cleavage. This suggests that hydroxyalkoxy radical dissociation
occurs primarily at C−C bonds that are adjacent to both the
alkoxy radical site and hydroxyl group. The close proximity of
the hydroxyl group to the alkoxy radical site likely increases the
rates of C2−C3, C6−C7, and C10−C11 bond cleavage.
Quantum chemical calculations37 on barrier heights and
enthalpies for tertiary hydroxyalkoxy radical dissociation
(shown and described in Figure S5 in the Supporting
Information) support this hypothesis. While the reaction
energy is largely consistent between the three fragmentation
pathways, cleavage of the C−C bond adjacent to both the
alkoxy radical site and hydroxyl group has the lowest barrier
height. This observation is also consistent with previous aerosol
oxidation studies, which showed that the presence of
oxygenated functional groups (e.g., alcohols and carbonyls)
may increase the rates of alkoxy dissociation reactions.10,12,15
3.2. Linolenic Acid Oxidation Product Analysis.
3.2.1. Linolenic Acid GC × GC Chromatogram. Figure 7
shows the distribution and relative abundance of particle phase
linolenic acid oxidation products in GC × GC space at an OH
exposure of 1.6 × 1011 molecules cm−3 s for ∼1% and 10% [O2]
(Figure 7, part a and b, respectively). Each circle signifies a
single compound and its relative abundance is represented by
the size of the circle. TMS-derivatized linolenic acid C17H29CO-
Figure 6a shows that the five most prominent carbon
numbers of the fragmentation products, in order of decreasing
abundance, are 22, 27, 17, 19, and 24. The large abundance of
these fragmentation products can be explained by the
dissociation of tertiary hydroxyalkoxy radicals (R10 in Figure
4). OH adds preferentially to the less substituted carbon of the
CC double bond (i.e., OH addition to C3, C7 and C11 of
the squalene carbon backbone) during heterogeneous
oxidation, leading to high yields of tertiary hydroxyalkoxy
radicals at 10% [O2]. In these tertiary hydroxyalkoxy radicals,
the alkoxy radical sites are located at C2, C6 and C10 of the
squalene carbon backbone (shown in Figure 6b). Previous
studies4,7−10,12,13,30,31 have shown that bond cleavage occurs at
C−C bonds adjacent to the alkoxy radical site. Here cleavage of
the C2−C3 bond in the squalene carbon backbone (which is
adjacent to the alkoxy radical at C2 of the squalene carbon
backbone) leads to the formation of fragmentation products
with carbon number 27 (I in Figure 6b). Similarly, cleavage of
the C5−C6 and C6−C7 bonds in the squalene carbon
backbone produce fragmentation products with carbon
numbers 24 and 22, respectively (II in Figure 6b). Cleavage
of the C9−C10 and C10−C11 bonds in the squalene carbon
backbone produce fragmentation products with carbon
numbers 19 and 17, respectively (III in Figure 6b).
1
(OTMS) (m/z 350, shown in orange in Figure 7) elutes at tR
2
∼ 54.5 min and tR ∼ 0.7 s. The fragmentation products
(shown in black in Figure 7) have carbon numbers less than 18
11564
dx.doi.org/10.1021/jp508378z | J. Phys. Chem. A 2014, 118, 11555−11571

The Journal of Physical Chemistry A
Article
Figure 8. Total mass concentrations of linolenic acid oxidation products, functionalization products and fragmentation products as a function of OH
exposure at (a) ∼ 1% [O2], and (b) 10% [O2]. Total mass concentrations of C17H28CO(OH)2 alcohol, C17H30CO(OH)2 alcohol, C17H29CO(OH)3
diol and C17H28CO2(OH)2 hydroxycarbonyl products as a function of OH exposure at (c) ∼ 1% [O2], and (d) 10% [O2].
and elute at earlier tR1 than linolenic acid. The functionalization
products (which elute at later tR than linolenic acid) are
grouped based on their molecular formulas.
8, respectively) suggests that the total abundance of linolenic
acid products that are detected is significantly smaller than the
squalene products. However, this may not be the case since the
external standards used in this study have only one or two
oxygenated functional groups, which may not be suitable for
accurate quantification of highly oxygenated linolenic acid
products due to differing response factors. The majority of
linolenic acid products have three or more oxygenated
functional groups, while most of the squalene products only
have one or two oxygenated functional groups. Highly
oxygenated compounds (with three or more oxygenated
functional groups) may also have lower GC column transfer
efficiencies.13 While these limitations prevent the accurate
quantification of linolenic acid oxidation products, it is still
useful to quantify and compare the linolenic acid products at
∼1% and 10% [O2] to gain insights into the role of [O2] on the
product formation chemistry. Because of the presence of a
contaminant (linoleic acid, which is present as an impurity in
the linolenic acid solution used to generate aerosol) that
coelutes with the TMS-derivatized linolenic acid, the effective
uptake coefficient for linolenic acid (γLONHA) could not be
determined from the EI-MS measurements. However, previous
1
Two groups of TMS-derivatized alcohol products are
detected as functionalization products: C17H28CO(OTMS)2
(m/z 438, shown in green in Figure 7) and C17H30CO-
(OTMS)2 (m/z 440, shown in blue in Figure 7). Ten
C17H28CO(OTMS)2 alcohol isomers are detected at ∼1%
[O2] (Figure 7a), while only five C17H28CO(OTMS)2 isomers
are detected 10% [O2] (Figure 7b), indicating that [O2] has a
considerable influence on the formation of these alcohol
isomers. These C17H28CO(OTMS)2 alcohol isomers elute at
tR2 ∼ 0.7 s with tR1 from ∼58 to 61 min at both ∼1% and 10%
[O2]. At both ∼1% and 10% [O2], six C17H30CO(OTMS)2
isomers are detected at tR2 ∼ 0.7 s with tR1 from ∼58 to 61 min.
Three TMS-derivatized diol isomers C17H29CO(OTMS)3 (m/z
528, shown in red in Figure 7) are detected at tR2 ∼ 0.7 s with
1
tR from ∼61 to 64 min, while six TMS-derivatized
hydroxycarbonyl isomers C17H28CO2(OTMS)2 (m/z 454,
1
shown in purple in Figure 7) are detected at tR from ∼61 to
2
64 min and tR from ∼0.8 to 0.9 s. MSTFA derivatization
decreases the polarities of the TMS-derivatized linolenic acid
and its alcohol and diol functionalization products, causing
kinetic studies of the linolenic acid reaction24 reported that γ
LNA
OH
2
them to have shorter tR than the TMS-derivatized hydrox-
decreases when [O2] increases, implying that O2, unlike for the
ycarbonyl functionalization products.
squalene reaction, promotes chain termination.
The mass concentrations of particle phase linolenic acid
products are quantified using external standards. Parts a and b
of Figure 8 show the total mass concentrations of detected
oxidation products, functionalization products and fragmenta-
tion products as a function of OH exposure at ∼1% and 10%
[O2], respectively. Error bars for the oxidation product mass
concentrations are not shown since there is some uncertainty
on how the detection efficiencies of the external standards differ
from that of the products.
Figure 8a shows that at ∼1% [O2] the total mass
concentration of detected oxidation products increases with
OH exposure. The total mass concentration of detected
oxidation products reaches a maximum value of ∼760 μg/m3
at an OH exposure of ∼1.8 × 1011 molecules cm−3 s.
Approximately 95% of the observed products are functionaliza-
tion products, and approximately 5% are fragmentation
products.
Figure 8b shows that at 10% [O2] the total mass
concentration of detected oxidation products increases and
reaches a maximum value of ∼280 μg/m3 at an OH exposure of
A comparison of the mass concentrations of detected
squalene and linolenic acid oxidation products (Figures 3 and
11565
dx.doi.org/10.1021/jp508378z | J. Phys. Chem. A 2014, 118, 11555−11571

The Journal of Physical Chemistry A
Article
∼1.1 × 1011 molecules cm−3 s. Approximately 88% of the
observed oxidation products are functionalization products, and
approximately 12% are fragmentation products. The large
abundance of functionalization products accompanied by very
low yields of fragmentation products at ∼1% and 10% [O2]
indicate that functionalization pathways play more prominent
roles in linolenic acid oxidation irrespective of [O2] (relative to
squalene oxidation).
The total mass concentration of oxidation products is larger
at ∼1% [O2] than at 10% [O2]. The differences in the total
mass concentration of oxidation products can be attributed to
differences in the functionalization product distribution at ∼1%
and 10% [O2]. For example, 10 TMS-derivatized C18H28O-
(OTMS)2 alcohol isomers are detected at ∼1% [O2], while
only five such alcohol isomers are detected 10% [O2]. This
suggests that [O2] controls the formation of these alcohol
products and consequently, the contribution of these alcohol
products to the total mass concentration of oxidation products.
The role that O2 plays in controlling the distribution of
functionalization products will be discussed further in sections
3.2.4 and 3.2.3.
Figure 9. Reaction mechanism for the OH-initiated oxidation of
linolenic acid (LNA) aerosol. R• represents the generic alkyl radical
formed by the abstraction of a H atom from a linolenic acid molecule.
The underivatized (bold text) and derivatized (regular text) molecular
formulas of the detected first generation functionalization products are
also shown.
The functionalization products are composed of TMS-
derivatized C17H28CO(OTMS)2, C17H30CO(OTMS)2,
C17H29CO(OTMS)3, and C17H28CO2(OTMS)2 product iso-
mers. The underivatized molecular formulas of these products
(obtained by substituting the “TMS” group with a H atom) are
C17H28CO(OH)2, C17H30CO(OH)2, C17H29CO(OH)3, and
C17H28CO2(OH)2. Previous kinetic studies24 indicate that
these are all first generation functionalization products. Figures
8c and 8d show the evolution of C17H28CO(OH)2, C17H30CO-
(OH)2, C17H29CO(OH)3, and C17H28CO2(OH)2 mass con-
centrations as a function of OH exposure at ∼1% and 10%
[O2], respectively. At ∼ 1% [O2], C17H28CO(OH)2,
C17H30CO(OH)2, C17H29CO(OH)3, and C17H28CO2(OH)2
make up approximately 29%, 41%, 9% and 21% of the total
mass concentration of observed functionalization products,
respectively (Figure 8c). At 10% [O2], C17H28CO(OH)2,
C17H30CO(OH)2, C17H29CO(OH)3, and C17H28CO2(OH)2
make up approximately 12%, 13%, 13%, and 62% of the total
mass concentration of observed functionalization products,
respectively (Figure 8d).
The formation of first generation C17H28CO(OH)2,
C17H30CO(OH)2, C17H29CO(OH)3, and C17H28CO2(OH)2
functionalization products can be explained by the reaction
mechanism proposed previously by Nah et al.24 Figure 9 shows
this reaction mechanism for the OH-initiated oxidation of
linolenic acid aerosol. Similar to the OH + squalene reaction,25
the OH + linolenic acid reaction is initiated by OH addition to
the CC double bond. For clarity, the derivatized and
underivatized molecular formulas of the products are also
shown. Product isomers are analyzed in sections 3.2.4 − 3.2.4
within the context of the reaction mechanism.
This suggests that they are likely minor channels in the
linolenic acid reaction, and are therefore not included in the
reaction mechanism shown in Figure 9.
3.2.2. C17H30CO(OTMS)2 and C17H28CO(OH)2 Alcohol
Product Isomers. Six TMS-derivatized C17H30CO(OTMS)2
alcohol isomers are detected at both ∼1% and 10% [O2].
The underivatized molecular formula of these alcohol isomers
C17H30CO(OH)2 indicates that one O atom and two H atoms
are added to the linolenic acid molecule. These C17H30CO-
(OH)2 alcohol product isomers are formed by the chain
reaction where a H atom is abstracted from a neighboring
linolenic acid molecule by the hydroxyalkyl radical (R2 in
Figure 9). The detection of six TMS-derivatized C17H30CO-
(OTMS)2 alcohol isomers can be explained by the molecular
structure of linolenic acid. Linolenic acid is a linear molecule
with six carbon sites that OH can add to form secondary
hydroxyalkyl radicals. Since OH addition is predicted to be
approximately equally rapid at either of the carbons in the C
C double bonds of linolenic acid,19 six different secondary
hydroxyalkyl radicals are expected to be formed, resulting in the
formation of six C17H30CO(OH)2 alcohol product isomers.
Since the mass spectra of these six isomers contain similar
fragment ions, we are not able to assign structural isomers.
At ∼ 1% [O2], the total mass concentration of C17H30CO-
(OH)2 alcohol isomers reaches a maximum of ∼400 μg/m3
(OH exposure ∼1.8 × 1011 molecules cm−3 s, Figure 8c). At
10% [O2], the total mass concentration of C17H30CO(OH)2
alcohol isomers reaches a maximum of ∼30 μg/m3 (OH
exposure ∼1.6 × 1011 molecules cm−3 s, Figure 8d). Thus, the
total abundance of these alcohol isomer products is
approximately 13 times larger at ∼1% [O2] compared to 10%
[O2], indicating that O2 suppresses the formation of
C17H30CO(OH)2 products. This suggests that the formation
of secondary hydroxyperoxy radicals (R5 in Figure 9) becomes
the dominant reaction pathway for hydroxyalkyl radicals at high
[O2], effectively slowing down the chain propagation reaction
that produces the C17H30CO(OTMS)2 alcohol products (R2 in
Figure 9).
Similar to the squalene reaction, we see no evidence of
organic peroxide products formed from hydroxyalkyl/alkyl
radical + hydroxyperoxy/peroxy radical (R + RO2), hydrox-
yalkyl/alkyl radical + hydroperoxyl radical (R + HO2) and
hydroxyperoxy/peroxy radical + hydroperoxyl radical (RO2 +
HO2) reactions. Therefore, these chain termination reactions
appear to be minor channels, within out ability to detect these
products, in the linolenic acid reaction, and are not included in
the reaction mechanism shown in Figure 9. Carbonyls, which
can be formed from the reaction of alkoxy radicals with HO2
radicals, are also not detected in the GC × GC chromatograms.
Ten TMS-derivatized C17H28CO(OTMS)2 alcohol isomers
are detected at ∼1% [O2], while only five such alcohol isomers
11566
dx.doi.org/10.1021/jp508378z | J. Phys. Chem. A 2014, 118, 11555−11571

The Journal of Physical Chemistry A
Article
are detected at 10% [O2]. The underivatized molecular formula
of these alcohol isomers C17H28CO(OH)2 corresponds to the
addition of one O atom to the linolenic acid molecule. At ∼ 1%
[O2] the total mass concentration of C17H28CO(OH)2 alcohol
isomers reaches a maximum of ∼280 μg/m3 (OH exposure
∼1.8 × 1011 molecules cm−3 s, Figure 8c). At 10% [O2] the
total mass concentration of C17H28CO(OH)2 alcohol isomers
reaches a maximum of ∼30 μg/m3 (OH exposure ∼1.6 × 1011
molecules cm−3 s, Figure 8d). These C17H28CO(OH)2 alcohol
isomers can be formed by the reaction of alkyl (R) radicals
(formed by H atom abstraction by OH, hydroxyalkyl and
hydroxyalkoxy radicals) with O2 and OH radicals (R3 and R9 in
Figure 9). Since the mass spectra of these C17H28CO(OTMS)2
alcohol isomers contain similar fragment ions, we are not able
to assign structural isomers.
The decrease in the number and total mass concentration of
these alcohol product isomers with increasing [O2] can be
explained by the reaction mechanism shown in Figure 9. At low
[O2], secondary hydroxyalkyl radicals can propagate the
particle-phase secondary chain chemistry by abstracting a H
atom from a neighboring linolenic acid molecule to form R
radicals (R2), which can react with O2 and OH radicals to form
C17H28CO(OH)2 alcohol products (R3). When increasing
amounts of O2 are added to the reaction, the secondary
hydroxyalkyl radicals can react instead with O2 to form
secondary hydroxyperoxy radicals (R5), effectively slowing
down the formation of R radicals via R2 and suppressing the
formation of these C17H28CO(OH)2 alcohol products from R3.
The formation of R radicals from H atom abstraction reactions
by hydroxyalkoxy radicals (R8) is a minor channel at high [O2]
since low yields of hydroxyalkoxy radicals are formed from self-
and cross-reactions of hydroxyperoxy radicals (R7).
C17H28CO2(OTMS)2 hydroxycarbonyl product isomers are
detected at both ∼1% and 10% [O2]. The underivatized
molecular formula of these hydroxycarbonyl isomers is
C17H28CO2(OH)2. Figure S7a−f (Supporting Information)
shows the VUV mass spectra of the six TMS-derivatized
C17H28CO2(OTMS)2 hydroxycarbonyl product isomers, while
Figure S7g (Supporting Information) shows the relative
abundance of the six hydroxycarbonyl isomers.
C17H28CO2(OTMS)2 hydroxycarbonyl isomers with the
OTMS group positioned at C12 and carbonyl group positioned
at C13 make up the majority (∼30%) of the structural isomers.
This suggests that C17H28CO2(OH)2 hydroxycarbonyl isomers
with the hydroxyl and carbonyl groups at C12 and C13,
respectively, are formed preferentially (compared to the other
five isomers) during the reaction.
To determine more generally which CC double bonds in
linolenic acid (i.e., C15C16 vs C12C13 vs C9C10) are
most susceptible to the addition of hydroxyl and carbonyl
groups, the relative abundances of C17H28CO2(OH)2 hydrox-
ycarbonyl isomers are plotted as a function of carbon atom
position along the linolenic acid backbone in Figure 10. There
The detection of ten TMS-derivatized C17H28CO(OTMS)2
alcohol isomers at ∼1% [O2] is consistent with the number of
allylic H atoms in the linolenic acid molecule. While linolenic
acid has 23 H atoms in C(sp3)−H bonds (along the carbon
backbone) available for H abstraction, there are only 4 allylic
C(sp3)-H sites. Allylic C(sp3)−H bonds have significantly
lower bond energies than nonallylic C(sp3)−H bonds.35,36
Therefore, the rate constants for H atom abstraction reactions
may be higher at these 4 sites. The allylic H atoms of linolenic
acid are located at C8, C11, C14 and C17 of the linolenic acid
carbon backbone (shown in Figure S6a in the Supporting
Information). The allylic R radicals formed from H atom
abstraction reactions are resonance stabilized. A total of 10
resonance-stabilized allylic R radicals can be formed (shown in
Figure S6b in the Supporting Information), which results in the
formation of 10 C17H28CO(OH)2 alcohol products at ∼1%
[O2].
Figure 10. Distribution of C17H28CO2(OH)2 hydroxycarbonyl
product isomers as a function of carbon atom position along the
linolenic acid backbone. Error bars represent the different fractions of
positional isomers under different OH exposure conditions. Also
shown are the labeled carbon atoms in the linolenic acid carbon
backbone.
is an enhancement in the addition of hydroxyl and carbonyl
groups to CC double bonds located at the terminus of the
linolenic acid molecule (i.e., C15C16 and C12C13 double
bonds). Isomers formed from the addition of hydroxyl and
carbonyl groups to the C15C16 and C12C13 double
bonds make up ∼36% and 45% of the total hydroxycarbonyl
mass concentration, respectively. One possible explanation for
the observed product distribution is that particle phase linolenic
acid molecules are oriented perpendicular to the particle
surface, with the C15C16 and C12C13 double bonds
preferentially exposed to (and consequently susceptible to
reaction with) gas phase OH radicals and O2. This proposed
molecular orientation of linolenic acid molecules at the particle
surface is similar to that proposed previously by Hearn et al.41
of oleic acid (linear C18 carboxylic acid with one CC double
bond) molecules at the particle surface.
Previous studies28−32 have shown that R radicals can react
with O2 to form peroxy radicals, which self- and cross-react with
other peroxy and hydroxyperoxy radicals to form particle phase
products with one hydroxyl group (C17H28CO(OH)2) or one
carbonyl group (C18H28O3). Similar to the squalene reaction,
C18H28O3 carbonyls are not detected, indicating that they are
minor reaction products. Analogous to the squalene reaction,
while it is possible that the temperature of the reaction and the
viscosity of the linolenic acid particle are responsible for the
very low yields of C18H28O3 carbonyl products, better
understanding of the oxidation chemistry at allylic R radical
sites is clearly needed.
At ∼ 1% [O2], the total mass concentration of
C17H28CO2(OH)2 hydroxycarbonyl isomers reaches a max-
imum of ∼60 μg/m3 (OH exposure ∼1.7 × 1011 molecules
cm−3 s, Figure 8c). At 10% [O2], the total mass concentration
3.2.3. C17H28CO2(OH)2 Hydroxycarbonyl and C17H29CO-
(OH)3 Diol Product Isomers. Six TMS-derivatized
11567
dx.doi.org/10.1021/jp508378z | J. Phys. Chem. A 2014, 118, 11555−11571

The Journal of Physical Chemistry A
Article
of C17H28CO2(OH)2 hydroxycarbonyl isomers reaches a
maximum of ∼170 μg/m3 (OH exposure ∼1.1 × 1011
molecules cm−3 s, Figure 8d). Since C17H28CO2(OH)2
hydroxycarbonyls make up the majority (∼62%) of the
functionalization products at 10% [O2] but are minor
(∼21%) functionalization products at ∼1% [O2], this indicates
that O2 promotes the formation of these hydroxycarbonyl
products. On the basis of the reaction mechanism shown in
Figure 9, C17H28CO2(OH)2 hydroxycarbonyl products are
formed from self- and cross-reactions of secondary hydrox-
yperoxy radicals (R6) and the reaction of O2 with
hydroxyalkoxy radicals (R11). At low [O2], the formation
rates of secondary hydroxyperoxy (R5) and hydroxyalkoxy
radicals (R7) are slowed, leading to low yields of hydrox-
ycarbonyl products. Conversely secondary hydroxyperoxy and
hydroxyalkoxy radicals are formed in high yields at high [O2].
The secondary hydroxyperoxy radicals can then self- and cross-
react via the Russell and Bennett−Summers mechanisms25−29
(R6) and the secondary hydroxyalkoxy radicals can react with
O2 (R11) to form C17H28CO2(OH)2 hydroxycarbonyl
products.
hydroxycarbonyl isomers, which show that hydroxyl and
carbonyl groups are added preferentially to CC double
bonds located at the terminus of the linolenic acid molecule
(Figure 10). It is currently unclear why this is the case and
should be an area of focus for further investigation.
At ∼ 1% [O2] the total mass concentration of C17H29CO-
(OH)3 diol isomers increases and reaches a maximum of ∼30
μg/m3 (OH exposure ∼1.7 × 1011 molecules cm−3 s, Figure
8c), and is ∼9% of the total mass concentration of
functionalization products. At 10% [O2] the total mass
concentration of C17H29CO(OH)3 diol isomers increases and
reaches a maximum of ∼30 μg/m3 (OH exposure ∼1.1 × 1011
molecules cm−3 s, Figure 8d), and is ∼13% of the total mass
concentration of functionalization products. This indicates that,
unlike the hydroxycarbonyl products, the total mass concen-
tration of diol products appears to be independent of [O2] in
the reactor.
One possible explanation for the relatively low diol product
mass concentrations at 10% [O2] is that hydroxycarbonyls are
formed in excess of diol products in the self- and cross-
reactions of secondary hydroxyperoxy radicals (R6 in Figure 9),
consistent with Hearn et al.28 who also observed excess
carbonyl formation (relative to alcohols) in the Cl oxidation of
dioctyl sebacate aerosol. They attributed this to the importance
of the Bennett−Summers mechanism at high [O2]. Unlike the
Russell mechanism26−28 that produces equal quantities of
alcohols and carbonyls in peroxy radical self- and cross-
reactions, only carbonyls are produced in the Bennett−
Summers mechanism.28,30−32
The excess formation of hydroxycarbonyl products (relative
to diol products) at high [O2] can also be attributed to the fact
that the H atom abstraction reaction by secondary hydrox-
yalkoxy radicals (R8 in Figure 9) is a minor reaction channel at
high [O2]. As discussed previously by Nah et al.,25 low yields of
hydroxyalkoxy radicals are formed from secondary hydroxyper-
oxy radical self- and cross-reactions (R7 in Figure 9) at high
[O2]. This is consistent with the results of Docherty et al.,42
which showed that alkoxy radical formation is a minor reaction
pathway in the NO3 oxidation of oleic acid aerosol.
3.2.4. Linolenic Acid Fragmentation Products. At ∼ 1%
[O2] the total mass concentration of detected fragmentation
products reaches a maximum of ∼10 μg/m3 (OH exposure
∼9.8 × 1010 molecules cm−3 s, Figure 8a). At 10% [O2] the
total detected fragmentation product mass concentration
reaches a maximum of ∼30 μg/m3 (OH exposure ∼1.6 ×
1011 molecules cm−3 s, Figure 8b). Fragmentation products
makes up approximately 5 and 12% of the total mass
concentration of oxidation products at ∼1% and 10% [O2],
respectively. The formation of gas phase fragmentation
products is expected to be a minor channel as evident by the
small detectible loss in aerosol mass measured previously by
Nah et al.24 Therefore, the very low yields of fragmentation
products at ∼1% and 10% [O2] indicate that, unlike squalene,
fragmentation is only a minor channel in linolenic acid
oxidation.
The minor role that fragmentation plays in linolenic acid
oxidation can be explained by the reaction mechanism shown in
Figure 9. The only route for fragmentation product formation is
via hydroxyalkoxy radical dissociation (R10). At ∼ 1% [O2],
pathway R10 is effectively suppressed since hydroxyperoxy
radicals (and consequently hydroxyalkoxy radicals) are formed
in low abundance. At 10% [O2], secondary hydroxyperoxy
radicals are formed in high yields. These secondary
The importance of hydroxycarbonyl product formation at
LNA
high [O2] is consistent with previous measurements of γ
.
OH
Nah et al.24 showed that γ
decreases when [O2] increases
LNA
OH
(6.21
0.17 at ∼0% [O2] vs 5.73
0.14 at 10% [O2]),
implying that chain termination reactions become more
prominent at higher [O2]. This is consistent with the reaction
mechanism shown in Figure 9, which shows that secondary
hydroxyalkyl radicals propagate the particle-phase secondary
chain chemistry at low [O2] by abstracting a H atom from the
neighboring linolenic acid molecules (R2). When increasing
amounts of O2 are added to the reaction, the secondary
hydroxyalkyl radicals can react with O2 to form secondary
hydroxyperoxy radicals (R5) instead. These secondary
hydroxyperoxy radicals can then self- and cross-react via the
Russell and Bennett−Summers mechanisms28−32 (R6) to form
stable oxygenated products (via chain termination). This slows
the radical chain chemistry in the particle, causing a decrease in
LNA
γ
with increasing [O2]. It should be pointed out that the
vs [O2] trend is strongly influenced by the ratio of the
OH
LNA
OH
γ
hydroxyalkyl radical + linolenic acid (R2) reaction rate to the
hydroxyalkyl radical + O2 (R5) reaction rate. Preliminary
results suggest that the ratio has to be ≥10 to obtain the
LNA
OH
observed γ
vs [O2] trend, and it can change based on the
reaction rates of the other reaction pathways. Detailed
modeling of the branching ratios of the different pathways
responsible for the measured effective uptake coefficients and
product distributions will be presented in an upcoming
publication.
In addition to the C17H28CO2(OH)2 hydroxycarbonyl
products, self- and cross-reactions of secondary hydroxyperoxy
radicals (R6 in Figure 9) also produce C17H29CO(OH)3 diols.
C17H29CO(OH)3 diols (detected as C17H29CO(OTMS)3) can
also be formed by the reaction of hydroxyalkyl radicals with
OH (R4 in Figure 9), and H atom abstraction reaction by
secondary hydroxyalkoxy radicals (R8 in Figure 9). Figure
S1b−d (Supporting Information) shows the VUV mass spectra
of the three TMS-derivatized C17H29CO(OTMS)3 diol product
isomers, while Figure S1e (Supporting Information) shows the
relative abundance of the three diol isomers. The relative
abundances of the three diol isomers are approximately the
same (∼33% each), suggesting that these diol isomers are
formed at equal rates. This result differs from that of the
11568
dx.doi.org/10.1021/jp508378z | J. Phys. Chem. A 2014, 118, 11555−11571

The Journal of Physical Chemistry A
Article
hydroxyperoxy radicals can preferentially self- and cross-react
via the Russell and Bennett−Summers mechanisms28−32 (R6)
to form terminal stable oxygenated products (i.e., chain
termination). Even though secondary hydroxyalkoxy radicals
can be formed from self-and cross-reactions of secondary
hydroxyperoxy radicals (R7), the low abundance of particle
phase fragmentation products at 10% [O2] indicates that
pathway R7 is a minor reaction channel. This is consistent with
the previous study of the OH + linoleic acid reaction,24 which
showed that the average number of C atoms in the linoleic acid
molecule stays roughly constant at 18 during oxidation at 10%
[O2], indicating little C−C bond cleavage. These measure-
ments are also consistent with that by Docherty et al.42 in the
NO3 oxidation of oleic acid aerosol, which showed that the
formation of oxygenated products is a major reaction pathway
while the formation of alkoxy radicals is of minimal importance.
bonds that are adjacent to both the alkoxy radical site and
hydroxyl group.
In contrast to squalene, functionalization products are the
dominant products in linolenic acid aerosol oxidation at both
∼1% and 10% [O2]. At ∼ 1% [O2], approximately 95% and 5%
of the total observed oxidation product mass concentration are
functionalization and fragmentation products, respectively. At
10% [O2], approximately 88% and 12% of the total observed
oxidation product mass concentration are functionalization and
fragmentation products, respectively. At ∼ 1% [O2],
C17H28CO(OH)2 alcohols, C17H30CO(OH)2 alcohols,
C17H29CO(OH)3 diols and C17H28CO2(OH)2 hydroxycarbon-
yls make up 29%, 41%, 9% and 21% of the total observed
functionalization products, respectively. At 10% [O2],
C17H28CO(OH)2 alcohols, C17H30CO(OH)2 alcohols,
C17H29CO(OH)3 diols, and C17H28CO2(OH)2 hydroxycarbon-
yls make up 12%, 13%, 13%, and 62% of the total observed
functionalization products, respectively. The significantly lower
yields of linolenic acid fragmentation products indicate that
fragmentation is a minor reaction channel irrespective of [O2].
Analysis of linolenic acid functionalization products shows that
oxidation is initiated by OH addition to either of the carbons in
the CC double bonds, leading to the formation of secondary
hydroxyalkyl radicals. The distribution of linolenic acid
functionalization products is governed by [O2], suggesting
that O2 controls the functionalization pathways taken by the
secondary hydroxyalkyl radical intermediate.
In summary, the differences in the squalene and linolenic
acid product distributions can be understood by comparing the
roles that the hydroxyalkyl, hydroxyperoxy and hydroxyalkoxy
radical intermediates play in the two reaction systems. At ∼ 1%
[O2] (i.e., low [O2]), hydroxyalkyl radicals (formed by OH
addition to the CC double bonds) react mainly with OH
radicals to form diols, and with neighboring squalene and
linolenic acid molecules to form alcohols. In both reaction
systems, the formation of hydroxyperoxy radicals is suppressed
at ∼1% [O2] due to the low concentration of O2. At 10% [O2]
(i.e., high [O2]), hydroxyperoxy radicals are formed via the
reaction of O2 with hydroxyalkyl radicals.
In the squalene reaction, tertiary hydroxyperoxy radicals are
the dominant products in the O2 + hydroxyalkyl radical
reaction at 10% [O2]. Because of the lack of H atoms adjacent
to the peroxy radical site in the tertiary hydroxyperoxy radicals,
diols and tertiary hydroxyalkoxy radicals are likely the dominant
products from the self- and cross-reactions of tertiary
hydroxyperoxy radicals. The lack of H atoms adjacent to the
peroxy radical site also explains the absence of hydroxycarbonyl
products in the squalene GC × GC chromatograms. The
tertiary hydroxyalkoxy radicals can subsequently dissociate to
form lower molecular weight products. Hydroxyalkoxy radical
dissociation is an important channel in the OH + squalene
reaction at 10% [O2], as evident by the large abundance of
fragmentation products.
In the linolenic acid reaction, secondary hydroxyperoxy
radicals are formed from the O2 + hydroxyalkyl radical reaction
at 10% [O2]. Hydroxycarbonyl and diol products are the
dominant products from the self- and cross-reactions of
secondary hydroxyperoxy radicals. Unlike the squalene reaction,
the low abundance of fragmentation products indicates that
hydroxyalkoxy radicals are likely minor products from the self-
and cross-reactions of secondary hydroxyperoxy radicals in the
linolenic acid reaction.
4. CONCLUSIONS
The oxidation products from two model reaction systems, OH
+ squalene (C30H50, a branched alkene with six CC double
bonds) and OH + linolenic acid (C18H30O2, a linear carboxylic
acid with three CC double bonds), are analyzed to
investigate the effect of molecular structure (branched vs
linear) on product formation in the heterogeneous OH-
initiated oxidation of unsaturated organic aerosol. The
oxidation products of squalene and linolenic acid are identified
and quantified using two-dimensional gas chromatography−
mass spectrometry. In both squalene and linolenic acid
oxidation, the reaction is initiated by OH addition to the
CC double bonds to form hydroxyalkyl radicals. Allylic alkyl
radicals, formed from H abstraction reactions by hydroxyalkyl
radicals, play important roles in particle-phase secondary chain
chemistry and product distributions. Oxidation products arising
from secondary chain reactions are detected in large abundance
in both reaction systems, indicating that secondary chain
chemistry plays a major role in the heterogeneous oxidation of
unsaturated organic aerosol.
Functionalization products are the dominant products in
squalene oxidation at ∼1% [O2], with 89% and 11% of the total
observed oxidation product mass concentration being function-
alization and fragmentation products, respectively. C30H49OH
alcohols, C30H51OH alcohols and C30H50(OH)2 diols make up
46%, 22% and 32% of the observed functionalization products,
respectively. At 10% [O2], the total mass concentration of
functionalization products is approximately equal to the
fragmentation products. C30H49OH alcohols, C30H51OH
alcohols and C30H50(OH)2 diols make up 47%, 11% and 42%
of the observed functionalization products, respectively.
Approximately 90% and 80% of the original squalene mass
concentration can be accounted for (by particle phase products
and unreacted squalene) over the course of the reaction at ∼1%
and 10% [O2], respectively. Analysis of squalene functionaliza-
tion products reveals that oxidation is initiated predominantly
by OH addition to the less substituted carbon of the CC
double bond. Functionalization products originating from OH
addition to the more substituted carbon of the CC double
bond are not observed, indicating that it is a minor reaction
channel. The large abundance of squalene fragmentation
products at 10% [O2] is attributed to the formation and
dissociation of tertiary hydroxyalkoxy radicals in the presence of
O2. Analysis of squalene fragmentation products indicates that
hydroxyalkoxy radical dissociation occurs primarily at C−C
11569
dx.doi.org/10.1021/jp508378z | J. Phys. Chem. A 2014, 118, 11555−11571

The Journal of Physical Chemistry A
Article
(10) Kroll, J. H.; Smith, J. D.; Che, D. L.; Kessler, S. H.; Worsnop, D.
R.; Wilson, K. R. Measurement of Fragmentation and Functionaliza-
tion Pathways in the Heterogeneous Oxidation of Oxidized Organic
Aerosol. Phys. Chem. Chem. Phys. 2009, 11, 8005−8014.
(11) McNeill, V. F.; Yatavelli, R. L. N.; Thornton, J. A.; Stipe, C. B.;
Landgrebe, O. Heterogeneous Oh Oxidation of Palmitic Acid in Single
Component and Internally Mixed Aerosol Particles: Vaporization and
the Role of Particle Phase. Atmos. Chem. Phys. 2008, 8, 5465−5476.
(12) Smith, J. D.; Kroll, J. H.; Cappa, C. D.; Che, D. L.; Liu, C. L.;
Ahmed, M.; Leone, S. R.; Worsnop, D. R.; Wilson, K. R. The
Heterogeneous Reaction of Hydroxyl Radicals with Sub-Micron
Squalane Particles: A Model System for Understanding the Oxidative
Aging of Ambient Aerosols. Atmos. Chem. Phys. 2009, 9, 3209−3222.
(13) Zhang, H. F.; Ruehl, C. R.; Chan, A. W. H.; Nah, T.; Worton, D.
R.; Isaacman, G.; Goldstein, A. H.; Wilson, K. R. Oh-Initiated
Heterogeneous Oxidation of Cholestane: A Model System for
Understanding the Photochemical Aging of Cyclic Alkane Aerosols.
J. Phys. Chem. A 2013, 117, 12449−12458.
(14) Che, D. L.; Smith, J. D.; Leone, S. R.; Ahmed, M.; Wilson, K. R.
Quantifying the Reactive Uptake of Oh by Organic Aerosols in a
Continuous Flow Stirred Tank Reactor. Phys. Chem. Chem. Phys. 2009,
11, 7885−7895.
(15) Wilson, K. R.; Smith, J. D.; Kessler, S. H.; Kroll, J. H. The
Statistical Evolution of Multiple Generations of Oxidation Products in
the Photochemical Aging of Chemically Reduced Organic Aerosol.
Phys. Chem. Chem. Phys. 2012, 14, 1468−1479.
(16) Atkinson, R. Gas-Phase Tropospheric Chemistry of Organic-
Compounds. J. Phys. Chem. Ref. Data 1994, R1−R217.
ASSOCIATED CONTENT
* Supporting Information
Vacuum ultraviolet mass spectra used for the product
identification, quantum chemical calculations of alkoxy radical
decomposition, and resonance structures of allylic free radicals.
This material is available free of charge via the Internet at
http://pubs.acs.org.
■
S
AUTHOR INFORMATION
Corresponding Author
*(K.R.W.) E-mail: krwilson@lbl.gov.
Notes
■
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
T.N., S.R.L., and K.R.W. are supported by the Director, Office
of Energy Research, Office of Basic Energy Sciences, Chemical
Sciences Division, of the U.S. Department of Energy under
Contract No. DE-AC02-05CH11231. K.R.W. and C.R.R. are
additionally supported by the Department of Energy, Office of
Science Early Career Research Program. H.Z. is supported by
the Camille and Henry Dreyfus foundation postdoctoral
program in environmental chemistry. This research used
resources of the National Energy Research Scientific Comput-
ing Center, which is also supported by the Office of Science of
the U.S. Department of Energy under Contract No. DE-AC02-
05CH11231.
(17) Atkinson, R. Gas-Phase Tropospheric Chemistry of Volatile
Organic Compounds 0.1. Alkanes and Alkenes. J. Phys. Chem. Ref. Data
1997, 26, 215−290.
(18) Kwok, E. S. C.; Atkinson, R. Estimation of Hydroxyl Radical
Reaction-Rate Constants for Gas-Phase Organic-Compounds Using a
Structure-Reactivity Relationship - an Update. Atmos. Environ. 1995,
29, 1685−1695.
(19) Peeters, J.; Boullart, W.; Pultau, V.; Vandenberk, S.; Vereecken,
L. Structure-Activity Relationship for the Addition of Oh to
(Poly)Alkenes: Site-Specific and Total Rate Constants. J. Phys.
Chem. A 2007, 111, 1618−1631.
(20) Ziemann, P. J.; Atkinson, R. Kinetics, Products, and Mechanisms
of Secondary Organic Aerosol Formation. Chem. Soc. Rev. 2012, 41,
6582−6605.
(21) Atkinson, R.; Arey, J. Atmospheric Degradation of Volatile
Organic Compounds. Chem. Rev. 2003, 103, 4605−4638.
(22) D’andrea, T. M.; Zhang, X.; Jochnowitz, E. B.; Lindeman, T. G.;
Simpson, C. J. S. M.; David, D. E.; Curtiss, T. J.; Morris, J. R.; Ellison,
G. B. Oxidation of Organic Films by Beams of Hydroxyl Radicals. J.
Phys. Chem. B 2008, 112, 535−544.
(23) Moussa, S. G.; Finlayson-Pitts, B. J. Reaction of Gas Phase Oh
with Unsaturated Self-Assembled Monolayers and Relevance to
Atmospheric Organic Oxidations. Phys. Chem. Chem. Phys. 2010, 12,
9419−9428.
(24) Nah, T.; Kessler, S. H.; Daumit, K. E.; Kroll, J. H.; Leone, S. R.;
Wilson, K. R. Oh-Initiated Oxidation of Sub-Micron Unsaturated Fatty
Acid Particles. Phys. Chem. Chem. Phys. 2013, 15, 18649−18663.
(25) Nah, T.; Kessler, S. H.; Daumit, K. E.; Kroll, J. H.; Leone, S. R.;
Wilson, K. R. Influence of Molecular Structure and Chemical
Functionality on the Heterogeneous Oh-Initiated Oxidation of
Unsaturated Organic Particles. J. Phys. Chem. A 2014, 118, 4106−
4119.
(26) Kessler, S. H.; Nah, T.; Daumit, K. E.; Smith, J. D.; Leone, S. R.;
Kolb, C. E.; Worsnop, D. R.; Wilson, K. R.; Kroll, J. H. Oh-Initiated
Heterogeneous Aging of Highly Oxidized Organic Aerosol. J. Phys.
Chem. A 2012, 116, 6358−6365.
REFERENCES
■
(1) Rudich, Y. Laboratory Perspectives on the Chemical Trans-
formations of Organic Matter in Atmospheric Particles. Chem. Rev.
2003, 103, 5097−5124.
(2) Hallquist, M.; Wenger, J. C.; Baltensperger, U.; Rudich, Y.;
Simpson, D.; Claeys, M.; Dommen, J.; Donahue, N. M.; George, C.;
Goldstein, A. H.; et al. The Formation, Properties and Impact of
Secondary Organic Aerosol: Current and Emerging Issues. Atmos.
Chem. Phys. 2009, 9, 5155−5236.
(3) Liu, C. L.; Smith, J. D.; Che, D. L.; Ahmed, M.; Leone, S. R.;
Wilson, K. R. The Direct Observation of Secondary Radical Chain
Chemistry in the Heterogeneous Reaction of Chlorine Atoms with
Submicron Squalane Droplets. Phys. Chem. Chem. Phys. 2011, 13,
8993−9007.
(4) Ruehl, C. R.; Nah, T.; Isaacman, G.; Worton, D. R.; Chan, A. W.
H.; Kolesar, K. R.; Cappa, C. D.; Goldstein, A. H.; Wilson, K. R. The
Influence of Molecular Structure and Aerosol Phase on the
Heterogeneous Oxidation of Normal and Branched Alkanes by Oh.
J. Phys. Chem. A 2013, 117, 3990−4000.
(5) Renbaum, L. H.; Smith, G. D. The Importance of Phase in the
Radical-Initiated Oxidation of Model Organic Aerosols: Reactions of
Solid and Liquid Brassidic Acid Particles. Phys. Chem. Chem. Phys.
2009, 11, 2441−2451.
(6) Renbaum, L. H.; Smith, G. D. Artifacts in Measuring Aerosol
Uptake Kinetics: The Roles of Time, Concentration and Adsorption.
Atmos. Chem. Phys. 2011, 11, 6881−6893.
(7) George, I. J.; Abbatt, J. P. D. Heterogeneous Oxidation of
Atmospheric Aerosol Particles by Gas-Phase Radicals. Nat. Chem.
2010, 2, 713−722.
(8) George, I. J.; Vlasenko, A.; Slowik, J. G.; Broekhuizen, K.; Abbatt,
J. P. D. Heterogeneous Oxidation of Saturated Organic Aerosols by
Hydroxyl Radicals: Uptake Kinetics, Condensed-Phase Products, and
Particle Size Change. Atmos. Chem. Phys. 2007, 7, 4187−4201.
(9) Kessler, S. H.; Smith, J. D.; Che, D. L.; Worsnop, D. R.; Wilson,
K. R.; Kroll, J. H. Chemical Sinks of Organic Aerosol: Kinetics and
Products of the Heterogeneous Oxidation of Erythritol and
Levoglucosan. Environ. Sci. Technol. 2010, 44, 7005−7010.
(27) Isaacman, G.; Chan, A. W. H.; Nah, T.; Worton, D. R.; Ruehl,
C. R.; Wilson, K. R.; Goldstein, A. H. Heterogeneous Oh Oxidation of
Motor Oil Particles Causes Selective Depletion of Branched and Less
Cyclic Hydrocarbons. Environ. Sci. Technol. 2012, 46, 10632−10640.
11570
dx.doi.org/10.1021/jp508378z | J. Phys. Chem. A 2014, 118, 11555−11571

The Journal of Physical Chemistry A
Article
(28) Hearn, J. D.; Renbaum, L. H.; Wang, X.; Smith, G. D. Kinetics
and Products from Reaction of Cl Radicals with Dioctyl Sebacate
(Dos) Particles in O2: A Model for Radical-Initiated Oxidation of
Organic Aerosols. Phys. Chem. Chem. Phys. 2007, 9, 4803−4813.
(29) Ingold, K. U. Peroxy Radicals. Acc. Chem. Res. 1969, 2, 1−2.
(30) von Sonntag, C.; Schuchmann, H. P. The Elucidation of Peroxyl
Radical Reactions in Aqueous Solution with the Help of Radiation-
Chemical Methods. Angew. Chem., Int. Ed. 1991, 30, 1229−1253.
(31) von Sonntag, C.; Dowideit, P.; Fang, X. W.; Mertens, R.; Pan, X.
M.; Schuchmann, M. N.; Schuchmann, H. P. The Fate of Peroxyl
Radicals in Aqueous Solution. Water Sci. Technol. 1997, 35, 9−15.
(32) Bennett, J. E.; Summers, R. Product Studies of Mutual
Termination Reactions of Sec-Alkylperoxy Radicals - Evidence for
Non-Cyclic Termination. Can. J. Chem.Rev. Can. Chim. 1974, 52,
1377−1379.
(33) Atkinson, R.; Tuazon, E. C.; Aschmann, S. M. Products of the
Gas-Phase Reaction of the OH Radical with 3-Methyl-1-Butene in the
Presence of NO. Int. J. Chem. Kinet. 1998, 30, 577−587.
(34) Atkinson, R.; Tuazon, E. C.; Carter, W. P. L. Extent of H-Atom
Abstraction from the Reaction of the Oh Radical with 1-Butene under
Atmospheric Conditions. Int. J. Chem. Kinet. 1985, 17, 725−734.
(35) Lide, D. R. CRC Handbook of Chemistry and Physics, 85th ed.;
CRC Pr I Llc: Boca Raton, FL, 2004.
(36) Tesa-Serrate, M. A.; King, K. L.; Paterson, G.; Costen, M. L.;
McKendrick, K. G. Site and Bond-Specific Dynamics of Reactions at
the Gas-Liquid Interface. Phys. Chem. Chem. Phys. 2014, 16, 173−183.
(37) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.;
Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci,
B.; Petersson, G. A., et al., Gaussian 09, Revision D.01;, Gaussian, Inc.:
Wallingford CT, 2009.
(38) Khursan, S. L.; Martem’yanov, V. S.; Denisov, E. T. Mechanism
of the Recombination of Peroxyl Radicals. Kinet. Catal. 1989, 31,
1031−1040.
(39) Bennett, J. E.; Brunton, G.; Smith, J. R. L.; Salmon, T. M. F.;
Waddington, D. J. Reactions of Alkylperoxyl Radicals in Solution 0.1.
A Kinetic-Study of Self-Reactions of 2-Propylperoxyl Radicals between
135-K and 300-K. J. Chem. Soc. Faraday Trans. 1987, 83, 2421−2432.
(40) Denisov, E. T.; Afanas’ev, I. B. Oxidation and Antioxidants in
Organic Chemistry and Biology. CRC Press: Boca Raton, FL, 2005.
(41) Hearn, J. D.; Lovett, A. J.; Smith, G. D. Ozonolysis of Oleic Acid
Particles: Evidence for a Surface Reaction and Secondary Reactions
Involving Criegee Intermediates. Phys. Chem. Chem. Phys. 2005, 7,
501−511.
(42) Docherty, K. S.; Ziemann, P. J. Reaction of Oleic Acid Particles
with No3 Radicals: Products, Mechanism, and Implications for
Radical-Initiated Organic Aerosol Oxidation. J. Phys. Chem. A 2006,
110, 3567−3577.
11571
dx.doi.org/10.1021/jp508378z | J. Phys. Chem. A 2014, 118, 11555−11571