Interaction of gas-phase ozone at 296 K with unsaturated self-assembled monolayers: A new look at an old system

Article abstract of DOI:10.1021/jp046604x

The oxidation of organics adsorbed on surfaces by ozone is of fundamental chemical interest and potentially important in the lower atmosphere. Studies of the oxidation of the three-carbon and eight-carbon vinylterminated self-assembled monolayers (SAMs, C3= and C8=) on a silicon ATR (attenuated total reflectance) crystal by gas-phase O₃ at 296 K are reported. Oxidation of the SAMs was followed in real time by ATRFTIR using ozone concentrations that spanned 5 orders of magnitude, from ～10¹¹ to 10¹⁶ molecules cm^-3. For comparison, some studies of the saturated C8 SAM were also carried out. The films were also characterized by atomic force microscopy and water contact angle measurements. The loss of C=C and the formation of C=O were measured in real time and shown to be consistent with a Langmuir-Hinshelwood mechanism in which O₃ is rapidly adsorbed on the surface and then reacts more slowly with the alkene moiety. This is supported by molecular dynamics (MD) calculations which show that O₃ does not simply undergo elastic collisions but has a significant residence time on the surface. However, the kinetics measurements indicate a much longer residence time than the MD calculations, suggesting a chemisorption of O ₃. Formaldehyde was observed as a gas-phase product by infrared cavity ring down spectroscopy. Possible mechanisms of the ozonolysis and its atmospheric implications are discussed.

Full text of DOI:10.1021/jp046604x

J. Phys. Chem. A 2004, 108, 10473-10485
10473
Interaction of Gas-Phase Ozone at 296 K with Unsaturated Self-Assembled Monolayers:
A New Look at an Old System
Yael Dubowski, John Vieceli, Douglas J. Tobias,* Anthony Gomez, Ao Lin,
Sergey A. Nizkorodov, Theresa M. McIntire, and Barbara J. Finlayson-Pitts*
Department of Chemistry, UniVersity of California, IrVine, IrVine, California 92697-2025
ReceiVed: July 29, 2004; In Final Form: September 2, 2004
The oxidation of organics adsorbed on surfaces by ozone is of fundamental chemical interest and potentially
important in the lower atmosphere. Studies of the oxidation of the three-carbon and eight-carbon vinyl-
terminated self-assembled monolayers (SAMs, C3d and C8d) on a silicon ATR (attenuated total reflectance)
3
crystal by gas-phase O at 296 K are reported. Oxidation of the SAMs was followed in real time by ATR-
1
1
16
-3
FTIR using ozone concentrations that spanned 5 orders of magnitude, from ∼10 to 10 molecules cm .
For comparison, some studies of the saturated C8 SAM were also carried out. The films were also characterized
by atomic force microscopy and water contact angle measurements. The loss of CdC and the formation of
CdO were measured in real time and shown to be consistent with a Langmuir-Hinshelwood mechanism in
which O
supported by molecular dynamics (MD) calculations which show that O
collisions but has a significant residence time on the surface. However, the kinetics measurements indicate
a much longer residence time than the MD calculations, suggesting a chemisorption of O . Formaldehyde
3
is rapidly adsorbed on the surface and then reacts more slowly with the alkene moiety. This is
3
does not simply undergo elastic
3
was observed as a gas-phase product by infrared cavity ring down spectroscopy. Possible mechanisms of the
ozonolysis and its atmospheric implications are discussed.
Introduction
gets trapped inside the porous network of fluid organic chains
13,15
on the surface,
whereas for surfaces of bulk liquids, trapping
Aerosols play an important role in the chemistry and radiation
balance of the atmosphere. They provide surfaces for hetero-
geneous reactions and affect global climate both directly, by
absorbing and scattering solar and terrestrial radiation, and
indirectly, by acting as cloud condensation nuclei (CCN).
Organic matter accounts for a significant portion of the total
mass of atmospheric aerosols, in both continental and marine
environments. These organics are commonly found in the form
of mixed aerosols together with inorganic compounds. The
organic fraction in mixed aerosols often forms an outer coating
over an inorganic core that can be either solid (e.g., mineral
of ozone on the surface or lowering of the activation energy
20,23,24
for the reaction has been suggested.
For alkenes adsorbed
on solid substrates, the loss of ozone to the surface has been
followed and shown to be faster than expected based on
1
-3
1
6,20,25
analogous gas-phase O3-alkene reactions,
and gas-phase
1
7
products have been identified. However, in situ real-time
monitoring of alkenes adsorbed on solid substrates has only
recently been reported. This is important, because ozone has
been shown to physisorb on some organics as the initial step in
4,5
2
6
2
2,27
a multistep mechanism.
In such a case, loss of gas-phase
ozone may not occur at the same rate as formation of products
on the surface and representation of reactions by a single value
for the reaction probability under all conditions may not be
appropriate.
6
-8
dust) or liquid (e.g., aqueous nitrate and sulfate mixtures).
Reactive uptake of atmospheric trace gases onto such particles
may alter the chemical and physical properties of the aerosols.
For example, oxidation of an organic aerosol may affect its
reactivity toward other trace gases and its hygroscopic proper-
ties, in turn affecting the aerosol particles’ optical properties
and their CCN activity.
Unsaturated hydrocarbons are widespread in the troposphere.3
The reaction between unsaturated hydrocarbons and ozone has
In the present study, alkylsilane self-assembled monolayers
(SAMs) were used as proxies for organics adsorbed on solid
substrates such as airborne dust particles or urban surfaces (e.g.,
2,9,10
28-30
buildings).
The SAMs were attached directly to the surface
of an attenuated total reflection (ATR) crystal, and their
1
1
16
-3
oxidation by gaseous ozone (10 -10 molecules cm ) was
monitored in real time using FTIR at 1 atm pressure and room
temperature (296 ( 2 K). Although the reactions of ozone with
a variety of alkene SAMs on silica surfaces have been previously
been studied extensively in the gas phase and in bulk liq-
uids.3
,11,12
More recently, the kinetics and mechanisms of the
ozone surface oxidation of condensed alkenes, either as pure
liquids (bulk, aerosol particles, and thin films) or adsorbed on
1
6,17,20
studied,
kinetics measurements of the condensed phase
solid or liquid substrates have received attention.9,13-24,67
A
oxidation in real time have not been reported previously. In
addition to the real-time FTIR monitoring of the surface, gas-
phase products of the ozonolysis of SAMs were probed by
infrared cavity ring-down spectroscopy (IR-CRDS). The surfaces
were characterized additionally by atomic force microscopy
common conclusion of these studies is that surface alkenes react
with ozone faster than expected based on the analogous gas-
phase reactions. In the case of Langmuir films on water, ozone
*
To whom correspondence should be addressed. B. J. Finlayson-Pitts
949) 824-7670, e-mail bjfinlay@uci.edu; D. J. Tobias (949) 824-4295,
e-mail dtobias@uci.edu.
(AFM), and changes in their hygroscopicity upon oxidation were
(
probed using contact angle measurements. Finally, molecular
1
0.1021/jp046604x CCC: $27.50 © 2004 American Chemical Society
Published on Web 11/02/2004

10474 J. Phys. Chem. A, Vol. 108, No. 47, 2004
Dubowski et al.
mercury lamp (Jelight company Inc., Double Bore 78-2046);
3
the 185 nm line causes dissociation of O2 to generate O( P)
that adds to O2, generating O3. The O3 concentration in the
mixture was controlled by varying the exposure of oxygen gas
to the UV light using an adjustable opaque sleeve surrounding
the UV lamp. The O3/O2 mixture was introduced into a 5-cm
cell with quartz windows, and the O3 concentration was
determined using a UV-vis spectrometer (Hewlett-Packard
model 8452A) based on its absorption at 254 nm (σ ) 1.15 ×
-
17
2
-1
32
1
0
cm molecule , base e). Before each experiment, the
O3/O2 mixture was flowed through the lamp compartment and
the UV-vis spectrometer for over an hour to ensure that the
ozone concentration was stable. The gas flow was then diluted
with N2 and introduced into the ATR cell whose headspace is
Figure 1. Schematic diagram of the flow system in the oxidation
experiments.
∼
500 µL. The stainless steel parts of the ATR cell were coated
with halocarbon wax, and the lines were Teflon, materials for
which there is very little loss of O3 and for which precondition-
ing is therefore not required. In the experiments with [O3] <
dynamics simulations were used to investigate the nature of these
SAMs and their physical interaction with gaseous ozone. The
combination of all of these approaches yields new molecular
level insights into the interaction of O3 with unsaturated SAMs.
Specifically, we show that the mechanism involves the initial
adsorption of O3 on the SAM followed by a slower reaction
with the double bond. The atmospheric implications are
discussed.
1
4
-3
1
0 molecules cm , a mixture of O2 in He (Oxygen Services
Co, UHP, >99.995%) was irradiated and then further diluted
with He downstream. The O3 concentration in the cell was
calculated from that in the O3/O2 mixture and the factor by
which it was diluted with N2 or He. In the experiments with
11
-3
[O3] ≈ 10 molecules cm , the ozone mixture was introduced
3 -1
into the ATR system by withdrawing it (at ∼30 cm min )
from a premixed collapsible Teflon chamber. As discussed
below, the effective reaction probability for O3 with the SAMs,
although larger than that for analogous gas phase reactions, is,
on an absolute scale, relatively small. Under our conditions,
the loss of O3 in the ATR cell is calculated to be less than 5%
of its initial concentration so that there should not be significant
gradients of ozone along the crystal. In any case, the ATR
measurements average over the entire length of the crystal so
that the reported reaction probabilities reflect that from an
average concentration of O3 in the cell.
Experimental Section
1
. Deposition of SAMs. Three different alkylsilanes were
used to form SAMs on silica substrates: (1) methyl-terminated
n-octyltrichlorosilane, C8 (Gelest, 95%); (2) vinyl-terminated
allyltrichlorosilane, C3d (Aldrich, 95%); (3) vinyl-terminated
7
9
-octenyltrichlorosilane, C8d (United Chemicals Technologies,
5%). The SAMs were deposited directly on a 45° Si ATR
crystal (8 cm × 1 cm × 0.4 cm, 10 reflections) according to a
well-established technique.31 Briefly, the crystal was cleaned
with boiling ethanol and then with boiling chloroform. The dry
crystal was further cleaned for ca. 30 min with an argon (Oxygen
Services Co., ultrahigh purity (UHP), >99.999%) plasma
discharge (Harrick Scientific Plasma Cleaner/Sterilizer PDC-
A background spectrum of the surface was taken before
introducing the O3/O2 mixture into the ATR cell. Sample spectra
of the surface were recorded approximately every 40 s during
exposure to ozone, averaging 64 scans with a resolution of 4
3
2G, medium power). Upon removal from the plasma cleaner,
the substrate was rinsed with Nanopure water (18 MΩ cm) to
hydrate the surface and then dried with a flow of N2 (Oxygen
Services Co., UHP, 99.999%). The Si crystal is known to have
a layer of silica on it under these conditions. To coat only the
crystal area that is exposed to ozone during the experiments,
the deposition of the SAM was carried out with the dried crystal
placed in a fabricated Teflon holder. The latter was designed
-1
cm using an FTIR spectrometer (Cygnus 100, Mattson). Over
-1
the spectral range of interest, 1500-4000 cm , the penetration
depth (dp) of the evanescent wave into the ATR surrounding in
the present experimental system was calculated (eq I) to be 0.5-
3
3
0
.2 µm, respectively.
λ
to have an open slot on its top side with identical geometry
dp )
(I)
2
2
2 1/2
(
area of 3.6 cm ) to the one in the horizontal ATR holder used
n 2π(2 sin θ - n
)
21
1
during the experiments. About 2 mL of a millimolar solution
of alkylsilane in hexadecane (Aldrich, 99%) was introduced onto
the slot for ca. 10 min, during which time the SAM deposited
on the crystal. The SAM-coated crystal was then placed in
boiling chloroform to remove any unreacted physisorbed
alkylsilane. The coating and chloroform extraction steps were
repeated two additional times to ensure a smooth, closely packed
coating, which was characterized as described below. Samples
prepared for AFM imaging and cavity ring-down spectroscopy
In eq I, λ is the wavelength of the light, n1 is the refractive
index of the denser medium (3.4 for the Si ATR crystal), θ is
the incident angle, and n21 is the ratio between the refractive
index of the rarer medium (n2) and the crystal (n1). In the case
of a thin film, the refractive indices of the ATR crystal and of
the media above the film control the electric field more than
the refractive index of the film itself; therefore n(air) was taken
3
3
as n2. This depth is much larger than the estimated height of
the studied SAMs (∼13 and 7 Å for C8d and C3d, respec-
tively), and hence the ATR-IR spectra of these SAMs should
(
see below) were deposited on Si wafers following the same
procedure and were stored under dry N2 until analysis.
. Reaction with O3. After placement of the Si crystal in a
2
3
3
resemble their transmission spectra.
flow-through horizontal ATR holder (Pike Technologies), the
system was thoroughly purged with dry nitrogen overnight. A
schematic diagram of the experimental flow system is shown
in Figure 1. Ozone was generated by irradiating a dry flow of
O2 (Oxygen Services Co., UHP, >99.993%) with a low-pressure
As discussed below, exposure of the reacted SAMs to gaseous
ammonia was used to probe for the generation of carboxylic
acids in the reaction. Ammonia vapor was collected over an
NH4OH solution (29.5%, Fisher) after two cycles of freeze-

Interaction of Ozone with Unsaturated SAMs
J. Phys. Chem. A, Vol. 108, No. 47, 2004 10475
pump-thaw. The glass cell containing 401 Torr of ammonia
vapor (at 23 °C, including 18 Torr of water vapor) was then
flushed into the ATR system with a flow of nitrogen at 1 atm.
SAMs on a silica surface differ only slightly from SAMs of
37,38
similar carbon length on a gold surface.
On a silica surface,
the thickness and tilt angle of a SAM composed from octade-
cyltrichlorosilane (18 carbons long) is 25.2 Å and 10°,
3
. Atomic Force Microscopy (AFM) Imaging. Specimens
3
7
respectively. On a gold surface, the thickness of a SAM
composed from hydrocarbons that are 18 alkane carbons long
is 29 Å and the tilt angle of chains with 16-22 alkane carbons
were imaged in air at ambient pressure and humidity by tapping
mode using a Park Scientific AutoProbe CP scanning probe
microscope equipped with a piezoelectric scanner with a range
up to 10 µm. The scanner was calibrated in the xy directions
using a 1.0 µm grating and in the z direction using several
conventional height standards such as highly oriented pyrolytic
graphite. The tips employed were V-shaped silicon 2 µm
cantilevers (Ultralevers, model no. ULNC-AUNM, Thermo-
Microscopes) or ultrasharp V-shaped noncontact silicon (Ul-
trasharp cantilevers, model no. NSC11, MikroMasch). Topo-
graphs were obtained as 256 × 256 pixels and were flattened
line by line and analyzed using AutoProbe image processing
software supplied by the manufacturer of the AFM. The root-
mean-square (rms) surface roughness was calculated from Rrms
3
8
is 20-30°.
a. The Neat Systems. Molecular dynamics simulations were
performed for two different SAMs in order to characterize their
structures. Each SAM consisted of either 64 1-propenethiolate
(CH2dCHCH2S) molecules or 64 1-octenethiolate (CH2dCH-
(CH2)6S) molecules, where the sulfur atom was chemisorbed
to a gold (111) surface. The simulation box dimensions were x
40.08 Å, y ) 34.72 Å, and z ) 56.00 Å for the 1-propenethi-
)
olate SAM and x ) 40.08 Å, y ) 34.72 Å, and z ) 62.00 Å for
the 1-octenethiolate SAM. The z dimension was perpendicular
to the SAM/vapor interface, and its magnitude was reduced with
a shorter hydrocarbon chain to make the gas-phase volume the
same in both systems. The gold surface was located at z ) 0 Å
in both systems. The simulation box geometry resulted in a
N
2
1/2
)
[∑ (zN - jz ) /(N - 1)] where jz is the average z height and
N is the number of points sampled.
. IR-CRDS Spectroscopy. The gas-phase products of the
4
2
-1
surface area of 22 Å molecule . The force field for the
hydrocarbons was taken from previously published work.39 An
adsorption potential and surface corrugation potential were
included in the simulations, following the work of Mar and
ozonolysis of SAMs were probed in a separate experiment using
IR-CRDS. Commercial Si(111) wafers optically polished on one
side were used as substrates for SAMs. The SAM-coated wafers
were placed inside a vacuum chamber just below the axis of
the CRDS cavity, where they were exposed to ozone at a
concentration of ∼10 molecules cm . The CRDS cavity was
equipped with 99.98% reflective mirrors optimized for the 3.3
µm range (Los Gatos Research). The cavity mirrors were spaced
by about 60 cm and protected by a constant purging flow of
dry helium, resulting in empty cavity ring-down times of the
order of 5-10 µs. A commercial pulsed optical parametric
3
5
Klein.
14
-3
b. Ozone. Molecular dynamics simulations of the SAMs
described above in the presence of ozone were performed in
order to characterize the adsorption of ozone to the SAM
surfaces and the collision rate between ozone molecules and
double bonds. The oxygen atoms of ozone were described using
4
0
the CHARMM22 force field, where each oxygen atom was
considered to be a carbonyl oxygen atom with no atomic charges
-
1
oscillator laser (0.1 cm spectral resolution) was used to
optically pump the cavity. The region between 2910 and 2930
(
the hybridization in ozone is more similar to that of a carbonyl
2
-
1
oxygen atom, sp , than to that of an oxygen in an alcohol or
water, sp ). The electrostatic interactions were ignored in these
cm , which contains easily identifiable lines of formaldehyde
and formic acid, was used to detect these two products. All
experiments were carried out at room temperature (295 K) under
slow flow conditions in an atmosphere of mostly helium at 50-
3
systems because the hydrocarbon molecules that compose the
SAMs were nonpolar and the molecular dipole moment of ozone
4
1,42
is known from experiment to be small (0.53 D).
The gas-
1
30 Torr. An optoacoustic reference spectrum of HCOOH was
recorded in parallel with CRDS spectra for wavelength calibra-
tion purposes.
phase equilibrium geometry was used for the ozone molecules,
where the two bond lengths were 1.278 Å and the bond angle
43
was 117°. A reflecting wall was included to prevent the escape
of gas-phase ozone molecules. It was placed in the xy plane at
z ) 56 Å in the 1-propenethiolate SAM and at z ) 62 Å in the
5
. Contact Angle Measurements. The surface wettability
of the SAM-coated crystal was probed before and after exposure
to O3 via contact angle measurements using water droplets.
Quasi-equilibrium contact angles of 1 µL Nanopure water sessile
droplets were measured under ambient conditions with a Kodak
DCS 315 camera equipped with a long-range microscope
1
-octenethiolate SAM.
c. Simulation Details. After 400 ps equilibrations, 800 ps
simulations of the neat systems were performed. Then, four
ozone molecules were added to the final configuration from
the simulation of each neat system. The ozone molecules were
placed approximately 20 Å above the SAM surfaces, and 300
ps equilibrations were run. After equilibration, the SAMs with
ozone were simulated for 1600 ps. All of the simulations were
performed with a constant number of molecules, temperature,
(
Infinity Optics). The shape of the droplet depends on its
34
interaction with the surface. The line tangent to the curve of
the droplet at the point where it intersects the solid surface forms
the contact angle. A water droplet resting on a hydrophobic
surface would form a spherical droplet having a high contact
angle, but it would have a much smaller contact angle when
placed on a more hydrophilic surface.
44
and volume using CHARMM. The system temperature of 300
K was controlled using Nose-Hoover chain thermostats.45 The
van der Waals interactions were truncated at 10 Å. A 1.0 fs
integration time step was used, and all bonds lengths involving
hydrogen atoms were constrained using the SHAKE/RATTLE
6
. Molecular Dynamics Simulations. Molecular dynamics
simulations were carried out to provide insight into the
experimental results. The SAMs used in the simulations were
alkylthiolates because the surface potentials for these are well
developed. This is justified because experimental data from
the literature suggest that the structural differences between
SAMs composed from long alkane chains on silica and on gold
are minor. For example, the spacing between Si atoms on a
silica surface is 4.97 Å whereas the S atoms on a gold surface
46,47
algorithm.
3
5
Results and Discussion
Deposition of SAMs directly on an ATR crystal enables
detection of various features of even very short monolayers (e.g.,
C3d) and permits real-time monitoring of these features during
3
6
are 4.4 Å apart. Also, the thickness and tilt angle of alkane

10476 J. Phys. Chem. A, Vol. 108, No. 47, 2004
Dubowski et al.
Figure 3. Snapshots from the molecular dynamics simulations of (a)
1
-propenethiolate and (b) 1-octenethiolate SAMs. Sulfur atoms are
shown in black, alkane carbons in red, and alkene carbons in green.
Figure 2. ATR-FTIR spectra of the three SAMs before exposure to
ozone. The spectra shown were obtained by coating the ATR crystal
with the SAM on all sides.
experiments. Typical ATR-FTIR spectra of the three unreacted
SAMs used in the present study are shown in Figure 2. The
-
1
absorption bands at ∼2856 and ∼2926 cm are due to the
-
CH2-symmetric (νs) and asymmetric (νas) stretches, respec-
4
8
-1
tively. The additional band at ∼2962 cm in the spectrum
of the saturated C8 SAM is assigned to the asymmetric stretch
of the methyl groups (νas -CH3).48 These peaks are in excellent
16,17
agreement with previous studies on similar monolayers
and
are indicative that the chains are not as highly ordered as longer
chain alkane SAMs.4 The terminal vinyl groups in C3d and
C8d give rise to two additional bands: (1) a dCH2 asymmetric
9,50
-
1
-1
stretch at ∼3081 cm for C3d and at ∼3077 cm for C8d,
-
1
and (2) a CdC stretch at ∼1635 cm (C3d) and at ∼1642
-
1
cm (C8d). The CdC stretch has a much lower intensity in
the spectrum of C8d than that in the C3d spectrum. The lower
relative intensity of the CdC band in the C8d SAM may be
indicative of a more highly ordered film formed with longer
alkyl chains (where van der Waals forces assist in forming
highly ordered films). The CdC stretch vibrations in well-
oriented SAMs are mainly perpendicular to the surface, an
orientation to which ATR is not sensitive.33,51,52 Thus, a more
highly ordered SAM for C8d may lead to a reduced intensity
of the CdC band.
Figure 4. Probability density distributions from the molecular dynamics
simulations of the angle, θ, between the surface normal and the (a)
S-C1 bond and the (b) CdC bond, in both C8d (dashed line) and
C3d (solid line).
To probe this, molecular dynamics (MD) simulations were
carried out to examine the orientation of the chains and of the
terminal vinyl groups relative to the surface normal. Snapshots
of the neat C3d and C8d SAM simulations are shown in Figure
70°). This broader distribution of tilt angles for the C3d SAM
relative to the C8d SAM demonstrates that there is a greater
degree of disorder in the C3d SAM. The observation of
increasing disorder with decreasing chain length has been
38
3. The top views are looking in the z dimension from the airside
observed previously with both alkylthiol SAMs on gold and
5
2
of the interface. The thickness of the simulated SAMs is
approximately 7 Å for the C3d SAM and 13 Å for the C8d
SAM, which is in excellent agreement with experimental
alkylsilane SAMs. In Figure 4b, the double bond orientation
distributions, defined as the distribution of angles between the
surface normal and the terminal CdC, are shown. Although
the overall distributions are quite similar, the C8d SAM has
somewhat more double bonds oriented perpendicular to the
interface. This suggests that better ordering of the C8d SAM
makes a contribution to the different relative intensities observed
for the CdC and -CH2- bands in the two SAMs.
38
measurements of alkylthiol SAM thickness using ellipsometry.
The probability distributions of tilt angles and double bond
orientations from the simulations are shown in Figure 4. In
Figure 4a, the tilt angle distribution, based on the angles between
the surface normal and the S-C1 bond, spans the range from
0
° to approximately 30° for the C8d system. However, in the
Further evidence supporting the increased order for the C8d
SAM involves measurements of the contact angles for water
C3d system, the tilt angles can be much larger (up to about

Interaction of Ozone with Unsaturated SAMs
J. Phys. Chem. A, Vol. 108, No. 47, 2004 10477
Figure 5. AFM images of (a) C8, (b) C3d, (c) C8d SAMs deposited on Si wafers. Cross sectional profiles measured along lines shown in the
figures are presented on the right.
and AFM imaging of these SAMs. For the unreacted C8d SAM,
the contact angles were in the range of 89° to 96° compared to
opening the sample compartment led to increased water vapor
in the light path.
7
7° to 84° for the C3d SAM, and around 100° for the saturated
In the oxidation experiments, however, the noise was
-4
C8 SAM (1σ for these measurements is 2°). These values are
lower than those reported for saturated long-chain SAMs that
are highly ordered (∼110°), indicating that the unsaturated
SAMs form a more disordered layer, as might be expected. The
contact angles reported in the present study are in good
agreement with previous measurements on similar SAMs
and with the trend of the tilt angle distributions from the MD
simulations.
AFM images of the various SAMs, deposited on Si wafers,
are shown in Figure 5. The averaged rms roughness values
measured were 6.5, 3.1, and 4.2 Å for C3d, C8d, and C8,
respectively. These low roughness values suggest that all three
of the SAMs used here make relatively uniform films. In the
case of C3d, the AFM image shows the presence of small
aggregates, which are most likely due to polymerization of the
C3d chains. This is in agreement with the unrealistically high
surface density obtained for these SAMs based on their IR
absorption (see below), as the ATR measures these polymerized
chains as well.
The FTIR spectra of the SAMs shown in Figure 2 were
obtained by ratioing the spectra of the SAMs to that of the clean
crystal; water vapor contributions due to changing purge in the
sample compartment of the spectrometer were also subtracted.
The noise in those spectra is on the order of 7 × 10 (3σ) and
arises primarily from small changes in the positioning of the
ATR crystal between the background spectrum of the clean
crystal and that of the reinserted, coated crystal. In addition,
significantly improved (3σ ∼1 × 10 ), which allowed changes
even in the weak CdC band in C8d to be measured during the
oxidation. For example, as seen in Figure 6, a clear decrease in
the absorption bands associated with the terminal vinyl groups
is observed for both the C3d and the C8d along with a
53
1
7,54
-1
simultaneous increase in a broad peak centered at ∼1710 cm
-
1
(for C3d) and at 1728 cm (for C8d) that is indicative of the
formation of a carbonyl group. The improved signal-to-noise
ratio was achievable by purging the system overnight with the
coated ATR mounted in the sample compartment to minimize
changes in water vapor and by using the spectrum of the SAM
before exposure to O3 as the reference, rather than the spectrum
of the clean crystal. The improved signal-to-noise ratio obtained
in this manner thus allows the oxidation of the SAMs to be
followed in real time with high sensitivity.
The width of the carbonyl peak for the oxidized C3d SAM
is about twice that for the oxidized C8d SAM. This may be
because of two factors. First, as discussed above, there is
evidence that the unreacted C3d SAM is initially less ordered
than the C8d SAM, and hence its oxidized form would be
expected to be less ordered as well. Both the positions and the
bandwidths are known to shift as a SAM becomes more
disordered; for example, the bandwidth for the νas(CH2) band
-
4
-1
of a tightly packed octadecyltrichlorosilane SAM is 16 cm
-
1
50
but ∼25 cm for less highly ordered films. Second, it may
be indicative of the presence of comparable amounts of both
aldehyde and carboxylic acid products on the surface. Thus,

10478 J. Phys. Chem. A, Vol. 108, No. 47, 2004
Dubowski et al.
Figure 6. Typical changes in the absorption spectra of (a) C3d and
1
4
-3
(
b) C8d SAMs upon exposure to ozone (8 × 10 molecules cm ).
Spectra were recorded at progressive ozone exposure times (light gray
to black, respectively), for C3d 0, 0.5, 1, and 5 min; for C8d 0, 1,
and 6 min.
Figure 7. Changes in absorbance at the (a) CdO and (b) CdC bands
1
4
-3
as a function of exposure time to O
3
(8 × 10 molecules cm ), for
C3d (open circles), C8d (solid triangles), and C8 (open squares).
-
1
the carbonyl peak for aldehydes is blue-shifted by about 20 cm
-
1
compared to that for acids (∼1730 versus 1710 cm ). This
possibility is discussed in more detail below in conjunction with
the IR-CRDS results on the gas-phase products.
by Maoz and Sagiv51 for longer SAMs on silicon (absorbance
)
0.00021 per CH2 group per ATR reflection) and the surface
1
4
density for such well-packed SAMs (4.2 × 10 molecules
Kinetics. As seen in Figure 7, the CdO peak increases
-
2 55,56
cm ).
3.1 ( 0.5) ×10 (1σ) molecules cm for both the C8d and
C8 SAMs. Given a typical surface density of free OH groups
The average surface density was calculated to be
(Figure 7a) and the CdC band decreases (Figure 7b) in the first
1
4
-2
(
few minutes upon exposure to ozone. No such pattern was
observed for the saturated C8 SAM, as expected based on the
very low reactivity of alkanes toward ozone.3,11,12 Because the
product CdO absorbance is linearly correlated to its surface
coverage, the following should hold
1
4
-2 57,58
on the silica surface to be ∼4.5 × 10 molecules cm ,
this suggests occupation of about 70% of the binding sites. The
values obtained for surface densities of C3d were higher with
1
4
-2
a larger uncertainty, (8 ( 4) ×10 molecules cm ; this may
be due to the higher disorder of the C3d SAM, as well as the
presence of some polymerized aggregates as indicated by the
AFM image (Figure 5). Therefore, we chose to use for C3d
the same surface density that was calculated for C8 and C8d.
The rate of change of the CdO absorbance, d[abs(CdO)]/dt,
was calculated using the derivative of the best fit to the
experimental absorbance data (Figure 7a). Because active
surface sites are consumed during the reaction, only the initial
rates were considered.
abs(CdO)t
[CdO]t
)
(II)
abs(CdO)∞
[CdO]∞
where abs(CdO)t and [CdO]t represent the CdO absorbance
and the total surface concentration of CdO groups at time t,
respectively, and t ) ∞ represents the time at which the SAM is
fully oxidized. Because in a completely oxidized SAM [CdO]∞
is expected to be equal to the surface density of the original
monolayer, [SAM]0, this correlation can be expressed as
Gas-surface reactions are normally treated in terms of a
reaction probability (γ), defined as the fraction of collisions of
gas with the surface that lead to reaction. In this framework,
abs(CdO) × [SAM]
t
0
[
CdO] )
(III)
(IV)
t
2
abs(CdO)∞
the rate of formation of CdO per cm is given by eq V
d[CdO]
dt
d[abs(CdO)]
[SAM]
0
d[CdO]
RT
2πM
)
×
) γ × [O3]g
(V)
(
)
(
)
(
)
t
dt
t
abs(CdO)∞
dt
t
x
The initial surface densities of the SAMs were estimated
where [O3]g is the gas-phase ozone concentration adjacent to
the surface, R is the gas constant, T is the temperature, and M
is the molecular weight of ozone. This assumes that diffusion
of O3 to the surface is not a limiting factor; given the fact that
based on the absorbance of the -CH2- asymmetric stretch at
-
1
∼
2926 cm (after deconvoluting the spectra and normalizing
the absorption to the number of reflections along the ATR
crystal), using the nonpolarized ATR measurements reported
-
5
the measured reaction probabilities are small, e10 , and that

Interaction of Ozone with Unsaturated SAMs
J. Phys. Chem. A, Vol. 108, No. 47, 2004 10479
Figure 8. Calculated initial reaction probabilities, γ(initial), for the oxidation of C3d (open circles) and C8d (solid squares) as a function of the
gas-phase ozone concentration. The data points were calculated using eq VI based on measured values of the carbonyl absorbances and initial O
and SAM concentrations.
3
the cell geometry is expected to favor turbulent flow, this
assumption is reasonable. Combining eqs IV and V gives an
expression for γ in terms of the measured experimental
parameters, eq VI
the surface had been rapidly oxidized before the first FTIR scans
were completed and the measurements of reaction probabilities
were made during further oxidation of the initial products.
Clearly, treating the reaction of gas-phase O3 with the
unsaturated SAMs as a simple collision with the surface that
has a fixed probability of reaction does not adequately describe
the experimental data. An alternative is a multistep process in
which O3 is first physisorbed onto the surface and hence has a
finite residence time on the SAM before reacting. Reaction of
O3 with the SAM then occurs in competition with its desorption
from the surface. This is the well-known Langmuir-Hinshel-
d[abs(CdO)]
dt
[
SAM]0
γ )
×
(VI)
(
)
t
abs(CdO) [O ]
x
RT/2πM
∞
3 g
Because the reaction rate (i.e., molecules cm- s-1 of product
formed) changes linearly with the gas-phase ozone concentra-
tion, the reaction probability is expected to be independent of
2
59,60
wood type of mechanism.
[O3]g.
Assuming that ozone cannot be adsorbed onto a previously
adsorbed ozone, then the rate of adsorption (Ra) and rate of
desorption (Rd) of ozone can be expressed in the form of eqs
VII and VIII
Figure 8 shows the values of the initial reaction probabilities,
γinit, calculated from eq VI as a function of the gas-phase ozone
concentration (note the log-log scale). The error bars, typically
of the order of 20%, are the statistical errors only (1 σ) and do
not include potential systematic errors. The greatest uncertainty
is in the initial surface concentration of the SAM, which can
vary from one coating to another; in addition, the value for the
SAM surface concentration is based on the absorption coefficient
for -CH2- groups reported by Maoz and Sagiv51 for a C18
SAM, which was assumed to apply to our C3 and C8 SAMs.
We estimate that inclusion of such potential systematic errors
could lead to an overall error for an individual experiment of
the order of 50%. However, it is clear that γinit is inversely and
nonlinearly dependent on [O3]g and levels off as the ozone
R ) k [O ] ([S] - [O ] )
(VII)
a
a
3 g
3 s
R ) k [O ]
3 s
(VIII)
d
d
3
-1 -1
where ka is the adsorption rate constant (cm molecules s ),
[S] is the total surface density of O3 adsorption sites (taken as
1
4
-2
5 × 10 molecules cm for ozone), [O3]s is the surface density
-
2
of adsorbed ozone (molecules cm ), and kd is the desorption
rate constant (s ). At equilibrium, the two rates are equal, and
-
1
the steady-state ozone coverage on the surface can be written
as
11
-3
concentration approaches 10 cm . The overall error associated
with the point at the smallest ozone concentration may be larger
than those at higher concentrations due to the longer duration
of the exposure, ∼17 h, during which a greater variation in the
experimental conditions (e.g., O3 concentration) would be
expected. Thus, when these factors are taken into account, the
data point at the lowest O3 concentration is consistent with a
leveling off of the measured reaction probability.
k [O ] [S]
[O ] [S]
a
3 g
3 g
[
O ] )
)
(IX)
3
s
k [O ] + k
[O ] + B
a
3 g
d
3 g
where B ) kd/ka. In the present system, the surface reaction
-2 -1
rate (molecules cm s ) is given by eq X
Because these reaction probabilities are calculated for t f 0,
passivation of the surface by reaction of a significant fraction
of the alkene groups cannot be responsible for this dependence
of the reaction probability on ozone. Consistent with this, the
initial product formation was accompanied by corresponding
decreases in the CdC peaks, which would not be the case if
d[CdO]
)
)
t
k [O ] [SAM] (1 - f )
(X)
(
s
3 s
0
t
dt
2
-1
where ks is the second-order reaction rate constant (cm s
molecules ) and f t is the fraction of the SAM that has reacted
with ozone at time t (which can be estimated based on the ratio
-
1

10480 J. Phys. Chem. A, Vol. 108, No. 47, 2004
Dubowski et al.
on silica particles coated with C8d. An initial uptake
2
5
-
5
coefficient for O3 of (7 ( 2) × 10 was reported. However,
2
5
the data in Figure 5 of the Usher et al. paper show a small,
but continuing, uptake of O3 with production of gas-phase
products as indicated by a peak at m/e ) 29. We estimate from
these data that the uptake coefficient during this time is ∼7 ×
-
6
1
0 . The gas-phase ozone concentration in those experiments
1
1
-3
was 1.9 × 10 molecules cm . Applying eq XII with our
14
values of B and ks, a SAM concentration of 3 × 10 molecules
cm , and assuming f ) 0.1, an uptake coefficient of 7 × 10
is calculated using the Langmuir-Hinshelwood mechanism for
-
2
-6
2
5
the experimental conditions of Usher et al. Given the approx-
imations made, this is in very good agreement with the slower
uptake of ozone at longer reaction times that they observed.
The values of ks reported for the reaction of surface-adsorbed
ozone with BaP-coated soot, anthracene on aqueous solutions,
and SAMs on a solid substrate are surprisingly similar,
suggesting that there is a common, rate-determing step. All three
organics have electron-rich π systems, and the formation of
weak π complexes between ozone and alkenes has been reported
Figure 9. Calculated reaction probabilities for the oxidation of C3d
at the time when half of the SAM has been oxidized (f ) 0.5) as a
function of the gas-phase ozone concentration. The solid line is the
least-squares fit of the data to eq XII.
of abs(CdC)t to the final absorption abs(CdC)∞). Combining
eqs IX and X gives
d[CdO]
dt
[O
3
]
g
[S]
)
k [SAM] (1 - f )
(XI)
(
)
s
0
t
t
[O ] + B
62-64
3
g
at low temperatures,
suggesting that perhaps partial electron
transfer is a key step in the oxidation.
Substituting eq V in eq XI and rearranging the latter provides
a new expression for the reaction probability (equation XII),
which contains two unknowns, B and ks
-
1
The residence time of ozone on the surface is equal to kd
-
1
(
)[B×ka] ). The adsorption rate constant ka is given by eq
XIII
k [SAM] (1 - f )
[S]
s
0
t
γ )
× [
(XII)
RT
2πM
[S]
O ] + B
S0
x
RT/2πM
3 g
x
3
-1 -1
k (cm molecules s ) )
(XIII)
a
For consistency over the large range of O3 concentrations
used in these studies (which covers 5 orders of magnitude), the
experimental data were analyzed by calculating values of γ at
the point where half of the SAM had reacted (i.e., ft ) 0.5)
using eq VI. The values obtained for γ(f)0.5) are plotted as a
function of [O3]g in Figure 9. The best fit value for B is (4 (
where S0 is the average uptake coefficient. Ammann et al.60
have demonstrated that as long as the surface coverage of the
adsorbed gas species (O3 in our case) is well below saturation,
its uptake coefficient equals S0 and is independent of its gas-
1
3
-3
16
3
1
) × 10 molecules cm , and the best fit value for ks is (2 (
phase concentration. Moise and Rudich report uptake coef-
-
17
2
-1 -1
-4
-4
) × 10 cm molecules
s
(l σ). A value of ft of 0 or 0.25
ficients for O3 on C3d and C8d of 2.7 × 10 and 1.7 × 10 ,
instead of 0.5 does not alter significantly the values obtained
for B or ks.
respectively, which were independent of [O3]g over the range
9
11
-3
of 10 to 10 molecules cm . Thus, it seems that these
measured uptake coefficients at very low ozone concentrations
likely represent the surface accommodation of these SAMs with
respect to O3. This conclusion is further supported by the B
The calculated B for the present system is about an order of
magnitude larger than the value obtained for ozone on benzo-
1
2
-3
[
a]pyrene (BaP)-coated soot (3.6 × 10 molecules cm ; the
-1 27
13
K value reported in that study is equal to B ). Because B is
the ratio of the desorption rate constant to that for adsorption,
its magnitude is related to the strength of binding of O3 to the
substrate. Soot is known to have free radical sites61 with which
ozone is expected to strongly interact, leading to smaller rates
of desorption relative to adsorption and hence smaller values
of B as is the case for soot compared to the SAMs. However,
the B value from these studies is smaller than that for anthracene
value obtained from our experimental data (∼4 × 10
-
3
-1
molecules cm s ), which indicates a fractional surface
coverage by ozone (i.e., [O3]s/[S]; eq IX) on the order of only
11
-3
0.3% at [O3]g ) 10 molecules cm . Use of an average uptake
-
4
coefficient of 2 × 10 and surface density of adsorption sites
1
4
-15
3
-1
of ∼5 × 10 gives a value for ka of ∼4 ×10
cm s
-
1
molecule for O3 at 298 K (eq XIII). On the basis of these
-
1
values of B and ka, a value of 0.15 s is calculated for kd,
indicating a residence time of ca. 7 s. This relatively long
residence time is of the same order of magnitude as the residence
times reported by P o¨ schl et al. for O3 on BaP-coated soot.
In the gas phase, a typical reaction probability (γg) for ozone
15
3
-1
adsorbed on either water (2.1 × 10 cm molecule ) or an
1
4
3
-1 22
aqueous octanol solution (5.1 × 10 cm molecule ),
implying that ozone is more strongly adsorbed on the SAMs in
the present study than on the polycyclic aromatic hydrocarbon
2
7
-
8
(
PAH) coating on aqueous solutions. In the latter case,
with terminal alkenes (e.g., 1-hexene) is about 3 × 10 , based
-
17
3
-1 -1
significant amounts of water vapor are present, and this may
alter the surface and hence the strength of binding of O3 to it.
Indeed, P o¨ schl et al.27 observed water competed with O3 for
surface sites on the BaP-coated soot.
The value obtained for ks is very similar to the rate constant
reported previously for the surface reaction of ozone with benzo-
on rate constant of 1 × 10
cm molecules
s
and a
-10
3
-1
diffusion-controlled rate constant of ∼3 × 10 cm molecule
-1 3
s . Substitution of this reaction probability in eq V suggests
1
5
-3
that at [O3]g ∼ 10 molecules cm oxidation of a SAM with
1
4
-2
a surface density of 3.8 × 10 molecules cm should be
complete after ∼30 min, about 10 times slower than the time
observed experimentally (Figure 7). This discrepancy increases
with decreasing ozone concentrations; other studies on the
heterogeneous oxidation of unsaturated alkenes by ozone have
-17
2
-1
[
a]pyrene on soot aerosols, (2.6 ( 0.8) × 10 cm molecule
-1 27
s , and with anthracene adsorbed at the air-water interface,
-
17
2
-1 -1 22
2
.6 × 10
cm molecule s . The value for ks reported
1
3,15,16,22,25,27
here is consistent with Knudsen cell data on the uptake of ozone
indicated similar trends.
Unlike the case of very

Interaction of Ozone with Unsaturated SAMs
J. Phys. Chem. A, Vol. 108, No. 47, 2004 10481
TABLE 1: Collision Rates(s-1 × ppm ×cm-2) of Ozone with the C3d and C8d SAMs
-1
theoretical
simulation
residence
time (ps)
simulation
collision
lifetime (ps)
a
ozone-surface b
ozone-alkenec
system
ozone-surface
1
7
17
17
C3d
C8d
2.2 × 10
2.1 × 10
14
17
16 × 10
0.19
0.23
1
7
17
17
2.2 × 10
1.7 × 10
20 × 10
a
b
Rates calculated based on gas-phase collision theory. Surface collision defined as ozone molecule entering and exiting the vicinity of the
SAM (see text). Alkene collision defined as distance of approach e4 Å between center of masses of ozone and double bond.
c
SCHEME 1: Schematic Diagram of Heterogeneous
a
Ozonolysis of Alkenes
a
The labels (s) and (g) represent surface and gas-phase species,
respectively.
calculations suggesting that a stable complex is formed on the
64
reaction pathway at a separation of 3.291 Å. The rates obtained
for O3 collision with the surface are slower than its collision
rates with the terminal alkene by factors of 8 and 12 for C3d
and C8d, respectively (Table 1). This apparent difference
between collision rates of O3 with the surface and with alkene
groups is due to inelastic collisions of O3 with the surface,
yielding multiple collisions with the terminal alkenes in every
collision with the SAM surface. The MD simulations indicate
a residence time of O3 near the C3d and C8d interface of about
Figure 10. Simulated density profiles of sulfur (thin solid line), alkane
carbon (dashed line), alkene carbon (solid line), and ozone (dotted line)
in (a) 1-propenethiolate and (b) 1-octenethiolate SAMs-ozone systems.
porous organic surfaces,15 the present SAMs form a relatively
packed and rigid surface layer (Figures 3 and 5), and therefore
an enhanced reaction probability due to significant ozone
trapping between the SAMs chains is not likely. This assumption
is supported by depth profiles (Figure 10) from the MD
simulations for such SAMs, showing a very limited penetration
of the ozone molecules into the SAM.
1
4 and 17 ps, respectively. This residence time is due to van
der Waals interaction between the SAM and the O3 molecules.
Thus, both the experimental data and the MD simulations
support a Langmuir-Hinshelwood type mechanism for the
interaction of gaseous ozone and the unsaturated SAMs.
However, the multiple encounters of an ozone molecule with
the terminal alkenes upon each collision with the surface, as
suggested by the MD simulations (about a factor of 10), can
only account for part of the observed enhancement in oxidation
rate. The residence time of ozone on the surface as calculated
from the experimental data, ∼7 s, is much longer than the time
obtained from the MD simulations (∼17 ps). This discrepancy
suggests that the adsorption of ozone on the surface involves
not just van der Waals attraction but also chemisorption.6
To check whether the enhancement in the oxidation rate of
the SAMs by ozone observed in the present study is due to an
enhancement in reaction probability or in residence time on the
terminal alkenes, molecular dynamic simulations of the interac-
tion of ozone with similar SAMs were conducted. Table 1
summarizes the collision rates of ozone with the surface as a
whole and with the terminal alkene in particular as calculated
either from gas-phase collision theory or from the MD simula-
tions. In these simulations, collision between an O3 molecule
and the surface was defined as an O3 molecule that goes below
a threshold value in the z dimension and then returns above
that threshold value. The z threshold selections of 15 Å for the
C3d SAM system and 20 Å for the C8d SAM system were
made to be greater than the maximum z coordinate of the ozone
center of mass observed while an ozone molecule was adsorbed
to the SAM surface. A collision of O3 with an alkene group
was defined as a distance of 4 Å or less between the ozone
center of mass and the center of the double bond. Four
angstroms was selected as the cutoff distance based on ab initio
2-64
Products and Mechanism. The products of the oxidation
of alkenes by ozone are expected to be aldehydes and the
3
,11,12
Criegee intermediates (Scheme 1).
In the gas phase, the
Criegee intermediate is known to decompose to form stable
products such as CO2 and free radicals such as OH or to
rearrange to a carboxylic acid. The increase in the carbonyl band
-
1
in the 1700 cm region during the reaction was accompanied
by a small simultaneous increase in the absorption around 3400
-
1
cm , suggesting the formation of a carboxylic acid on the
surface. To probe this further, the oxidized SAMs were exposed

10482 J. Phys. Chem. A, Vol. 108, No. 47, 2004
Dubowski et al.
Figure 11. Changes in the absorption spectra of oxidized C8d (solid
line) and pentanoic acid (dotted line) upon exposure to NH vapors,
showing the conversion of the carboxylic acid (∼1710 cm ) to
3
-
1
-
1
carboxylate (1555 cm ). Spectra were obtained by ratioing the single
beam spectra after exposure to the ones before exposure to NH
3
.
to gaseous NH3, which reacts quickly with acids to form an
ammonium salt; in separate experiments we showed that the
reaction with aldehydes was much slower under these condi-
tions. As seen in Figure 11 when the SAMs were exposed to
NH3, the CdO peak decreases sharply and a new absorption
-
1
band appears at 1555 cm , indicating the formation of
carboxylate ions. Similar results were observed upon exposure
of an authentic sample of pentanoic acid to gaseous NH3 (Figure
1
1). Thus, carboxylic acids are formed in the ozone reaction
with vinyl-terminated SAMs.
As discussed earlier, the larger width of the carbonyl peak
for the reacted C3d may be in part due to the presence of a
significant amount of aldehyde in addition to the carboxylic
acid.
However, the presence of aldehydes cannot be definitively
established based on the ATR spectra of the reacted films.
Further insight into the mechanism can be obtained from
studies of the gas phase products. Cavity ring-down spectroscopy
14
Figure 12. CRDS spectra obtained during the reaction of O
3
(∼10
-
3
molecules cm ) of (a) C8, (b) C8d, and (c) C3d (y axis is absorption
-
1
coefficient in cm ). Also shown are (d) the reference optoacoustic
spectrum of HC(O)OH vapor and (e) the reference spectrum of HCHO
downloaded from the National Solar Observatory database convoluted
to 0.1 cm resolution. (Note: stronger lines in this spectrum are
saturated.)
-
1
(
CRDS) was used for real-time detection of gas phase products
released during the ozonolysis of C3d, C8d, and C8 SAMs
Figure 12). In the control experiments on the ozonolysis of
(
C8 SAMs, no HCHO or HCOOH gas-phase products were
observed in CRDS spectra except for a small background from
outgassing of the CRDS chamber walls (Figure 12a). On the
contrary, ozonolysis of alkene-terminated SAMs under identical
conditions reproducibly resulted in considerably larger signals
from HCHO in agreement with the expected mechanism of the
terminal CdC bond cleavage by O3 (Scheme 1, path b). For
example, characteristic HCHO bands are clearly visible in both
C8d and C3d spectra. The amount of HCHO produced in
ozonolysis of C3d and C8d SAMs is comparable, suggesting
the primary ozonide splitting pattern is not too different in these
two films. No significant signal from HCOOH was observed
in either case; an upper limit for the formation of HCOOH is
about 10% of that of HCHO. The small yield observed for the
formation of HCOOH is consistent with two possibilities. First,
the decomposition of the primary ozonide mainly follows
pathway b in Scheme 1, releasing HCHO to the gas phase and
leaving the Criegee intermediate on the surface. This rearranges
and stabilizes to form the carboxylic acid, which was confirmed
to be present on the surface by reaction with NH3. A second
possibility is that the Criegee intermediate released to the gas
phase in pathway a of Scheme 1 rapidly decomposes to form
CO2, CO, H2, and H2O rather than isomerizing to HCOOH.
The latter occurs, for example, in the gas-phase ozone-propene
reaction where yields of (CO + CO2) were measured to be 52%
compared to a yield of 11% for formic acid.65 The wall loss for
HCOOH is expected to be larger than that for HCHO, which
may also contribute to the lower apparent HCOOH yield.
The products observed here can be compared to those reported
1
7
25
by Rudich and co-workers and Grassian and co-workers. In
the former case, HCHO was observed as the major gas-phase
product of both the C3d and C8d oxidations, with a yield of
about 50% of the ozone reacted. Smaller amounts of CO2, and
CO in the case of C3d were also measured. Formic acid was
only observed in the C3d oxidation and only at total pressures
below 80 Torr; even then, the yield was less than 5%. If small
amounts of formic acid were released to the gas phase, the for-
mation of an aldehyde group on the surface would be expected
(Scheme 1, path a). The broader carbonyl peak we observed in
the ATR spectra of oxidized C3d compared to oxidized C8d
suggests that there could be a higher HCOOH yield for C3d
than that for C8d as reported by Rudich and co-workers,17
although it was below the IR-CRDS detection limit.
25
13
Usher et al. used C solid-state NMR to search for products
on the surface. Although signals due to the alkene carbons
decreased upon reaction, no signals due to aldehyde or car-
boxylic acid groups were observed. They concluded that oxygen-
containing products did not remain on the surface and proposed
that the reaction proceeded via formation of gas-phase HCHO
and a Criegee biradical on the surface; the latter was hypoth-
esized to have decomposed to generate gas-phase CO2 and a
surface alkane. Such decompositions may be more important

Interaction of Ozone with Unsaturated SAMs
J. Phys. Chem. A, Vol. 108, No. 47, 2004 10483
Figure 13. A 1 µL water droplet resting on a C8d monolayer (a)
14
-1
3
before and (b) after exposure to O
angles are marked with black lines.
3
(8 × 10 molecules cm ). Contact
under the low-pressure conditions of their Knudsen cell than at
atmospheric presssure under which the present experiments were
carried out.
Conversion of the alkene to carboxylic acids and aldehydes
upon reaction with ozone is expected to make the surface more
hydrophilic. The contact angles of water droplets on the SAM-
coated crystal were measured before and after exposure to ozone
Figure 14. Change in absorbance at the CdO band (solid squares)
and CdC band (open squares) of C3d SAM as a function of exposure
time to 6 × 10 molecules cm of ozone.
(
Figure 13). After oxidation at higher concentrations of O3
15
-3
(∼10 molecules cm ), the contact angle of both the C3d
16
-3
and C8d (Figure 13) SAMs decreased by about 20°, whereas
relatively little change was observed for the saturated C8 SAM.
Both the values of the initial contact angles and the amplitude
of their reduction upon oxidation are in good agreement with
previous studies on similar SAMs.1
significantly higher yield than that for HCOOH. This can be a
result of preferable formation of formaldehyde during the
decomposition of the primary ozonide or because of rapid
decomposition of the Criegee intermediate rather than stabiliza-
tion to form HCOOH.
The formation of more polar species on the surface contributes
to an increase in surface hydrophilicity, as indicated by the
observed decrease in the contact angle of the water droplet on
the unsaturated SAM after oxidation (by ∼20°). Similar
oxidation processes are expected to occur on organic aerosols
during their residence in the atmosphere. Thus, the optical
properties of some organic aerosols and their potential to become
active CCN under atmospheric conditions may also depend on
their residence time in the atmosphere, the concentration of
atmospheric oxidants, and the rate of such heterogeneous
oxidation reactions.
6,17
At the CH-stretching frequency range of the ATR-FTIR
spectra (data not shown), a small decrease in the CH2 bands of
the unsaturated SAMs is observed upon ozone oxidation (no
such changes are observed when saturated C8 is used). Because
ozone has very low reactivity toward saturated alkanes, these
results suggest the formation of OH during the ozonolysis.
However, the same trends were observed when conducting the
experiments in the presence of excess ethane, indicating that
increasing disorder of the unsaturated monolayers during their
oxidation, rather than secondary OH oxidation, is the dominant
factor affecting the absorption bands in this region. This is in
agreement with previous studies on heterogeneous oxidation of
1
5,17,20,21
alkenes by ozone
and suggests that the excess energy
An interesting picture of the oxidative processing kinetics of
organic aerosols is beginning to emerge as experiments have
of the intermediate left attached to the surface is quickly damped
before it decomposes.
been combined with molecular dynamics computer simula-
Interestingly, during experiments with O3 concentrations
15,24
tions.
In the case of SAMs, which are model systems for
1
6
-3
above 5 × 10 molecules cm , the absorption of the carbonyl
band continues to rise slowly even after most of the CdC bonds
appear to have been consumed (Figure 14). This may be due to
oxidation of alkene bonds in the interior of what appears to be
polymer aggregates seen on the surface by AFM (Figure 5)
and/or secondary oxidation.
organic molecules adsorbed on urban surfaces and solid particles
25
(
e.g., mineral dust), the oxidation by ozone occurs via a
Langmuir-Hinshelwood type mechanism involving a rapid
equilibration between gaseous and adsorbed ozone and a slower
surface reaction between ozone and the terminal alkene group
of the SAMs. The best fit to the experimental data yields a
1
3
-3
Langmuir constant (B) of (4 ( 3) × 10 molecules cm and
Conclusions and Atmospheric Implications
a second-order rate constant for the surface oxidation reaction
-
17
2
-1 -1
The present work demonstrates the utility of ATR-FTIR to
monitor heterogeneous oxidation of organic surfaces (SAMs in
this case) by ozone. One of the advantages of using this
unintrusive spectroscopic technique is that it permits real-time
monitoring of both reactants and products on the surface under
ambient temperature and pressure. These spectra show that upon
exposure to ozone the terminal alkene groups of these SAMs
are quickly oxidized to form carbonyls. Some of these carbonyl
groups are present as carboxylic acid as indicated by the
formation of ammonium salt upon exposure of the oxidized
sample to NH3 vapor. The relatively broad carbonyl peak
observed in the ATR-FTIR spectra of oxidized C3d relative to
oxidized C8d suggests the presence of a significant amount of
aldehyde in addition to the carboxylic acid. Cavity ring down
spectroscopy indicates the formation of gas-phase HCHO in
(ks) of (2 ( 1) × 10
cm molecule s . These values are
similar to those reported previously for the heterogeneous
27
reactions of ozone with benzo[a]pyrene on soot aerosols and
anthracene on water or aqueous octanol solutions.22 One
important implication of the fact that the heterogeneous ozo-
nolysis of surface alkene (present work) and aromatic com-
22,27
pounds
follows a Langmuir-Hinshelwood type mechanism
is that the use of a simple reaction probability independent of
the ozone concentration to calculate the loss of the organic or
uptake of ozone is not appropriate. A more complex approach
that represents the molecular level mechanism must be used
when incorporating laboratory data for such heterogeneous
reactions into atmospheric models.
Although the MD simulations predict a residence time of ∼17
ps for ozone on the surface, the experimental data suggest a

10484 J. Phys. Chem. A, Vol. 108, No. 47, 2004
Dubowski et al.
much longer residence time, of the order of several seconds.
This discrepancy between simulations and experiments, and the
long residence time suggested by the latter, suggests that
chemisorption (e.g., through the formation of a π complex) plays
a significant role in these heterogeneous processes. Both the
van der Waals attraction, which leads to multiple “encounters”
of ozone with the surface during a collision, and the chemi-
sorption likely contribute to the rapid oxidation of the SAM
relative to what is expected based on analogous gas-phase
reactions. For example, in urban atmospheres, where terminal
alkenes account for a significant fraction of the unsaturated
compounds, ozone levels are on the order of 80 ppb. On the
basis of the values of B and ks obtained in the present study, an
initial oxidation rate for such thin alkene films is calculated
(13) Lai, C. C.; Yang, S. H.; Finlayson-Pitts, B. J. Langmuir 1994, 10,
4
637.
14) de Gouw, J. A.; Lovejoy, E. R. Geophys. Res. Lett. 1998, 25, 931.
15) Wadia, Y.; Tobias, D. J.; Stafford, R.; Finlayson-Pitts, B. J.
Langmuir 2000, 16, 9321.
(
(
(
16) Moise, T.; Rudich, Y. J. Geophys. Res. 2000, 105, 14667.
(17) Thomas, E. R.; Frost, G. J.; Rudich, Y. J. Geophys. Res. 2001,
1
06, 3045.
18) Morris, J. W.; Davidovits, P.; Jayne, J. T.; Jimenez, J. L.; Shi, Q.;
Kolb, C. E.; Worsnop, D. R.; Barney, W. S.; Cass, G. Geophys. Res. Lett.
002, 29.
19) Smith, G. D.; Woods, E.; DeForest, C. L.; Baer, T.; Miller, R. E.
(
2
(
J. Phys. Chem. A 2002, 106, 8085.
(20) Moise, T.; Rudich, Y. J. Phys. Chem. A 2002, 106, 6469.
(
21) Eliason, T. L.; Aloisio, S.; Donaldson, D. J.; Cziczo, D. J.; Vaida,
V. Atmos. EnViron. 2003, 37, 2207.
22) Mmereki, B. T.; Donaldson, D. J. J. Phys. Chem. A 2003, 107,
1038.
(23) Thornberry, T.; Abbatt, J. P. D. Phys. Chem. Chem. Phys. 2004, 6,
4.
(
1
8
5
-
4
-1
(
using eq XI) to be on the order of 5 × 10 molecules s per
surface site, indicating a lifetime for the SAM of about 0.6 h.
However, considering a typical rate constant for the reaction
(
24) Vieceli, J.; Ma, O. L.; Tobias, D. J. J. Chem. Phys. A 2004, 108,
806.
(25) Usher, C. R.; Michel, A. E.; Grassian, V. H. Chem. ReV. 2003,
103, 4883.
3
-
17
3
of ozone with gaseous aliphatic alkene of about 1 × 10 cm
-
1 -1
molecule
s
(e.g., 1-butene), a lifetime for gas-phase terminal
(26) Uosaki, K.; Quayum, M. E.; Nihonyanagi, S.; Kondo, T. Langmuir
alkenes of about 15 h is obtained. It should be noted that this
rapid oxidation of unsaturated SAMs on surfaces could also
lead to artifacts in sampling such organics on particles under
2
004, 20, 1207.
(27) P o¨ schl, U.; Letzel, T.; Schauer, C.; Niessner, R. J. Phys. Chem. A
2001, 105, 4029.
66
(28) Diamond, M. L.; Gingrich, S. E.; Fertuck, K.; McCarry, B. E.; Stern,
G. A.; Billeck, B.; Grift, B.; Brooker, D.; Yager, T. D. EnViron. Sci. Technol.
000, 34, 2900.
some conditions, as has been observed for PAH, for example.
Despite their importance in the atmosphere, heterogeneous
reactions on organic aerosols are not yet considered in most
atmospheric models. This is mainly due to the large diversity
of organic species present in atmospheric aerosols as well as a
significant lack of information regarding the kinetics and
mechanisms of such reactions. This study demonstrates the
importance of real-time monitoring of reactions at surfaces
involved in such heterogeneous reactions and the potential of
using both ATR-FTIR and MD simulations to enhance our
understanding of these important surface processes.
2
(29) Gingrich, S. E.; Diamond, M. L.; Stern, G. A.; McCarry, B. E.
EnViron. Sci. Technol. 2001, 35, 4031.
30) Finlayson-Pitts, B. J.; Wingen, L. M.; Sumner, A. L.; Syomin, D.;
Ramazan, K. A. Phys. Chem. Chem. Phys. 2003, 5, 223.
31) Sagiv, J. J. Am. Chem. Soc. 1980, 102, 92.
(
(
(32) DeMore, W. B.; Sander, S. P.; Golden., D. M.; Hampson, R. F.;
Kurylo, M. J.; Howard, C. J.; Ravishankara, A. R.; Kolb, C. E.; Molina,
M. J. Chemical Kinetics and Photochemical Data for Use in Stratospheric
Modeling, EValuation No. 12; Jet Propulsion Laboratory: Pasadena, CA,
1997; Vol. JPL Publ. No. 97-4.
(33) Harrick, N. J. Internal Reflection Spectroscopy; Interscience
Publishers: New York, 1967.
(
(
(
34) Ulman, A. ACS Symp. Ser. 1991, No. 447, 144.
35) Mar, W.; Klein, M. L. Langmuir 1994, 10, 188.
36) Ulman, A. Chem. ReV. 1996, 96, 1533.
Acknowledgment. We are grateful to the National Science
Foundation for support of a Collaborative Research in Chemistry
award (CHE-0209719) and an Environmental Molecular Sci-
ences Institute (CHE-0431512) under which this research was
carried out, Professor Jim Rutledge for assistance with the
contact angle measurements, and Lee Moritz and Jorg Meyer
for technical assistance. S.A.N. thanks the Research Corporation
for a Research Innovation Award.
(37) Allara, D. L.; Atul N. P.; F., R. Langmuir 1995, 11, 2357.
(38) Porter, M. D.; Bright, T. B.; Allara, D. L.; Chidseyi, C. E. D. J.
Am. Chem. Soc. 1987, 109, 3559.
(
39) Tobias, D. J.; Tu, K.; Klein, M. L. J. Chem. Phys. 1997, 94,
482.
(40) MacKerell, A. D.; Bashford, D.; Bellott, M.; Dunbrack, R. L.;
1
Evanseck, J. D.; Field, M. J.; Fischer, S.; Gao, J.; Guo, H.; Ha, S.; Joseph-
McCarthy, D. J.; Kuchnir, L.; Kuczera, K.; Lau, F. T. K.; Mattos, C.;
Michnick, S.; Ngo, T.; Nguyen, D. T.; Prodhom, B.; Reiher, W. E.; Roux,
B.; Schlenkrich, M.; Smith, J. C.; Stote, R.; Straub, J.; Watanabe, M.;
Wiorkiewicz-Kuczera, J.; Yin, D.; Karplus, M. J. Phys. Chem B 1998, 102,
3586.
References and Notes
(41) Meerts, W. L.; Stolte, S.; Dymanus, A. Chem. Phys. 1977, 19, 467.
(42) Mack, K. M.; Muenter, J. S. J. Chem. Phys. 1977, 66, 5278.
(43) Trambarulo, R.; Gosh, S. N.; Burrus, C. A.; Gordy, W. J. Chem.
(1) Change Climate 2001: The Scientific Basis; Houghton, J. T., Ding,
Y., Griggs, D. J., Noguer, M., van der Linden, P. J., Dai, X., Maskell, K.,
Johnson, C. A., Eds.; Intergovernmental Panel on Climate Change,
Cambridge University Press: Cambridge, 2001.
Phys. 1953, 21, 851.
44) Brooks, B. R.; Bruccoleri, R. E.; Olafson, B. D.; States, D. J.;
Swaminathan, S.; Karplus, M. J. Comput. Chem. 1983, 4, 187.
45) Martyna, G. J.; Tuckerman, M. E.; Tobias, D. J.; Klein, M. L. Mol.
(
(
2) Ramanathan, V.; Crutzen, P. J.; Kiehl, J. T.; Rosenfeld, D. Science
001, 294, 2119.
3) Finlayson-Pitts, B. J.; Pitts, J. N., Jr. Chemistry of the Upper and
2
(
(
Phys. 1996, 87, 1117.
(46) Ryckaert, J. P.; Ciccotti, G.; Berendsen, H. C. J. Comput. Chem.
1977, 23, 327.
Lower Atmosphere: Theory, Experiments and Applications; Academic
Press: San Diego, 2000.
(
4) Jacobson, M. C.; Hansson, H. C.; Noone, K. J.; Charlson, R. J.
ReV. Geophys. 2000, 38, 267.
5) Murphy, D. M.; Thomson, D. S.; Mahoney, T. M. J. Science 1998,
82, 1664.
(47) Andersen, H. C. J. Comput. Chem. 1983, 52, 24.
(48) Roeges, N. P. G. A Guide to the Complete Interpretation of Infrared
Spectra of Organic Structures; John Wiley & Sons: New York, 1994.
(49) Flinn, D. H.; Guzonas, D. A.; Yoon, R. H. Colloids Surf., A 1994,
87, 163.
(50) Angst, D. L.; Simmons, G. W. Langmuir 1991, 7, 2236.
(51) Maoz, R.; Sagiv, J. J. Colloid Interface Sci. 1984, 100, 465.
(52) Netzer, L.; Iscovici, R.; Sagiv, J. Thin Solid Films 1983, 100, 67.
(53) Wasserman, S. R.; Tao, Y. T.; Whitesides, G. M. Langmuir 1989,
5, 1074.
(54) Dordi, B.; Schonherr, H.; Vancso, G. J. Langmuir 2003, 19, 5780.
(55) Banga, R.; Yarwood, J.; Morgan, A. M.; Evans, B.; Kells, J.
Langmuir 1995, 11, 4393.
(56) Banga, R.; Yarwood, J.; Morgan, A. M.; Evans, B.; Kells, J. Thin
Solid Films 1996, 285, 261.
(
2
(
(
6) Husar, R. B.; Shu, W. R. J. Appl. Meteorol. 1975, 14, 1558.
7) Tervahattu, H.; Hartonen, K.; Kerminen, V.-M.; Kupiainen, K.;
Aarnio, P.; Koskentalo, T.; Tuck, A. F.; Vaida, V. J. Geophys. Res. 2002,
1
07, AAAc1.
(8) Tervahattu, H.; Juhanoja, J.; Kupiainen, K.J. Geophys. Res. 2002,
1
07, Art. No. 4319.
(
9) Rudich, Y. Chem. ReV. 2003, 103, 5097.
10) Weingartner, E.; Burtscher, H.; Baltensperger, U. Atmos. EnViron.
997, 31, 2311.
(
1
(
11) Atkinson, R.; Arey, J. Atmos. EnViron. 2003, 37, S197.
(12) Atkinson, R.; Arey, J. Chem. ReV. 2003, 103, 4605.

Interaction of Ozone with Unsaturated SAMs
J. Phys. Chem. A, Vol. 108, No. 47, 2004 10485
(
57) Iler, R. K. The Chemistry of Silica; Wiley: New York, 1979;
(63) Hull, L. A.; Hitsatsune, I. C.; Heicklen, J. J. Am. Chem. Soc. 1972,
Chapter 6.
94, 4856.
64) Gillies, C. W.; Gillies, J. Z.; Suenram, R. D.; Lovas, F. J.; Kraka,
E.; Cremer, D. J. Am. Chem. Soc. 1991, 113, 2412.
(
(
58) Zhuravlev, L. T. Langmuir 1987, 3, 316.
59) Adamson, A. W.; Gast, A. P. Physical Chemistry of Surfaces, 6th
(
ed.; Wiley-Interscience: New York, 1997.
60) Ammann, M.; P o¨ schl, U.; Rudich, Y. Phys. Chem. Chem. Phys.
003, 5, 351.
61) Chughtai, A. R.; Atteya, M. M. O.; Kim, J.; Konowalchuck, B.
K.; Smith, D. M. Carbon 1998, 36, 1573.
62) Bailey, P. S.; Ward, J. W.; Hornish, R. E. J. Am. Chem. Soc. 1971,
3, 3552.
(
(
65) Disselkamp, R. S.; Dupuis, M. J. Atmos. Chem. 2001, 40, 231.
66) Schauer, C.; Niessner, R.; P o¨ schl, U. EnViron. Sci. Technol. 2003,
(
2
3
7, 2861.
(
(67) Katrib, Y.; Martin, S. T.; Hung, H. M.; Rudich, Y.; Zhang, H. Z.;
Slowik, J. G.; Davidovits, P.; Jayne, J. T.; Worsnop, D. R. J. Phys. Chem.
A 2004, 108, 6686.
(
9