Heterogeneous reaction of NO2 on hexane soot: A Knudsen cell and FT-IR study

Article abstract of DOI:10.1021/jp002918i

NO₂ can react with soot particles to produce HONO. The heterogeneous reaction of NO₂ on freshly prepared hexane soot was investigated using a Knudsen cell reactor and FTIR spectroscopy. Initial uptake coefficients were determined using gas-diffusion models that take into consideration the surface area of the top layer of soot particles as well as the accessible underlying layers of soot particles. Under dry conditions, the initial uptake coefficient was near 5 x 10^-5 at a gas concentration of 2.5 x 10¹¹ molecules/cc and 295 K. The adsorbed products remained on the surface in agreement with previous results. HONO was a gas-phase reaction product, accounting for 36% of the NO₂ reacted. Knudsen cell data were in better agreement with the values of the uptake coefficient determined from other experimental methods when the BET areas of the soot samples were taken into account and average values of the uptake coefficient were compared.

Full text of DOI:10.1021/jp002918i

11926
J. Phys. Chem. A 2000, 104, 11926-11933
Heterogeneous Reaction of NO2 on Hexane Soot: A Knudsen Cell and FT-IR Study
Hind A. Al-Abadleh and V. H. Grassian*
Departments of Chemistry and Chemical and Biochemical Engineering, The UniVersity of Iowa,
Iowa City, Iowa 52242
ReceiVed: August 10, 2000; In Final Form: September 21, 2000
2
In this study, the reaction of NO on freshly prepared hexane soot has been studied using a Knudsen cell
reactor and FT-IR spectroscopy. Initial uptake coefficients were determined using gas-diffusion models that
account for the surface area of the top layer of soot particles as well as the accessible underlying layers of
-
5
soot particles. Under dry conditions, the initial uptake coefficient was found to be near 5 × 10 at a gas
11
-3
concentration of 2.5 × 10 molecules cm and a temperature of 295 K. In addition to reporting an initial
uptake coefficient, uptake coefficients averaged over time and/or coverage are reported so as to compare to
other values in the literature. Accounting for the increased surface area due to the underlying layers of soot
particles, there is better agreement between the value of the uptake coefficient reported here using a Knudsen
cell reactor and values determined from other experimental methods. The total amount of NO
2
reacted per
unit surface area of soot and the conversion of NO to HONO have been quantified. From these data, the
2
2
atmospheric implication of the heterogeneous reaction of NO on soot particles to form HONO is discussed.
4
Introduction
fellow et al. reported that the uptake coefficient at 262 K, 27%
11
relative humidity, and NO2 gas concentration of 2 × 10
Heterogeneous reactions have been reported to be of some
importance in the troposphere. In particular, carbonaceous
-3
-4
molecules cm to be in the range of (2-4) × 10 for propane
1
-
5
-5
and methane soot, 5 × 10 for kerosene soot, and 1 × 10
aerosol may provide a reactive surface for trace atmospheric
for hexane soot using the geometric surface area of the soot
surface. In another study using a Knudsen cell reactor to measure
the heterogeneous kinetics, the initial uptake coefficient for the
gases including O3,2 SO2,3 and nitrogen oxides.
4-15
The
heterogeneous reaction of ozone with soot aerosol has been
investigated by Kamm et al. They report that the kinetics of
2
reaction of NO2 on ethylene soot was reported to be 0.095 (
.007 by Gerecke et al. at 296 K using a NO2 gas concentration
ozone decomposition on soot is complex and that ozone loss
on soot particles may be nonnegligible in the polluted boundary
layer, where the mass concentration of soot particles is extremely
high. However, under most conditions in the troposphere and
lower stratosphere, the reaction is of negligible importance.
5
0
1
1
-3
of 7 × 10 molecules cm . The geometric area of the sample
holder was used to determine the initial uptake coefficient in
that study. The heterogeneous reaction of NO2 on commercial
and freshly prepared soot generated from the combustion of
toluene, diesel, and kerosene was also investigated using a static
3
Recently, Koehler et al. studied the heterogeneous uptake of
SO2 on n-hexane soot over the temperature range from -130
to -40 °C using FT-IR spectrometry. It was shown that there
is rapid uptake of SO2 on soot under dry conditions and low
temperatures (-70 °C), similar to the conditions of the upper
troposphere and lower stratosphere.
6
reactor. Mean uptake coefficients that accounted for the BET
surface area of the commercial soot samples were determined
-
6
13
to be ∼10 at 296 K for the consumption of 10 NO2
-
2
molecules cm at a gas concentration of 25 ppm.
From the discussion above, it is clear that there is a range of
values, from 0.095 to 10 , reported for the heterogeneous
uptake coefficient of NO on soot, a range that spans nearly 5
orders of magnitude. The discrepancy in the values of the uptake
coefficient arises in part because the soot samples investigated
were from different sources. Commercial soot differs from
freshly prepared soot, and even the later can be prepared in
different ways using different hydrocarbons. This means that
from study to study soot particles can have different sizes,
different BET surface areas, different chemical composition,
and, therefore, different reactivities. Experimental conditions
such as temperature, relative humidity, and pressure of NO2
were also varied from one study to study.
The heterogeneous reaction of nitrogen oxides, in particular
NO2, on carbonaceous aerosol has been studied by several
groups using a variety of experimental techniques.4-14 It has
been shown that NO2 can react with soot particles to produce
-
6
2
4
-8
HONO. It has been proposed that this reaction may account
for the high concentration of HONO observed during night time,
16
when the homogeneous reaction, shown below, is insignificant:
NO(g) + OH(g) f HONO(g)
To assess the atmospheric importance of the heterogeneous
reaction of NO2 and soot, the kinetics of the reaction must be
known. Most often a heterogeneous uptake coefficient, γ, is
reported for surface reactions. The heterogeneous uptake coef-
ficient, also referred to as a sticking coefficient, is defined as
the ratio of the rate of molecules lost from the gas phase to the
total gas-surface collision rate. Using a flow reactor, Long-
Another source of discrepancy is that the accessible surface
area for reaction must be accurately accounted for when
calculating uptake coefficients. NO2 may diffuse through the
interparticle void areas and access underlying layers of par-
6,17
4,5,11
ticles. However, in some of the above studies,
only uptake
*
To whom correspondence should be addressed.
on the first layer of soot was considered, and thus, the exposed
1
0.1021/jp002918i CCC: $19.00 © 2000 American Chemical Society
Published on Web 12/01/2000

Reaction of NO2 on Hexane Soot
J. Phys. Chem. A, Vol. 104, No. 51, 2000 11927
TABLE 1: Knudsen Cell Reactor Parameters
is O-ring sealed by a blank flange. Table 1 lists the Knudsen
cell parameters used in this study. The most important param-
eters for the Knudsen cell are the escape constant, kesc, and the
residence time of the molecules inside the reactor, τ, where τ
) 1/kesc.
Reactor parameter
Value
Volume, V
Total calculated surface area
964 cm3
1002 cm
2
2
Geometric area of the sample holder, A
Effective area of the escape aperture, Ah,eff
s
11.95 cm
a
2
0.0677 cm
When the sample compartment is opened to a steady-state
flow of NO2, loss of NO2 to soot surface competes with escape
through the exit aperture, causing a decrease in the mass
spectrometer signal. Thus, the geometric uptake coefficient, γg,
also referred to as the observed uptake coefficient, γobs, is
0.651 s (τ ) 1.5)c
-
1 b
k
esc for NO
2
.370 s (τ ) 2.7)c
-1 d
0
a
b
Effective area after accounting for the Claussing factor.
k
esc
)
1/2
c
d
cj Ah,eff/4V; cj )(8RT/πM) . τ ) 1/kesc
. kesc experimentally determined.
2
0,21
determined from eq 1
geometric area of soot was used in calculating γ. This may
explain the very high γ values reported in some studies
compared to others. In addition, in all of the studies discussed
above, surface areas were not measured for the soot samples
under investigation. This is usually done using the BET method
Ah,eff Io - I
γg )
) γobs
(1)
(
)
As
I
1
8
where Ah,eff and As are listed in Table 1, Io is the mass
spectrometer signal for NO2 prior to opening the sample
compartment, and I is the mass spectrometer signal when the
soot is exposed.
and N2. Instead, values were taken from the literature, which
may or may not accurately represent the true values because of
differences in sample preparation. The fractal nature of soot
also makes the characterization of the available surface area
4
Freshly prepared samples of n-hexane or toluene soot were
deposited directly on the sample holder by burning the vapor
of about 30 mL of n-hexane (EM Science, purity 98.5%) or
toluene (EM Science, purity 99.5%) in a small beaker. The
sample was placed at a height of ∼5 cm from the top of the
beaker so that soot is collected from the top part of the flame.
Particle characterization studies of hexane and toluene soot were
performed in order to have some insight about the nature of
these particles. These characterization studies include particle
size distribution analysis using transmission electron microscope
images, specific BET surface areas, SBET, and bulk density, Fb,
measurements. Figure 1 shows representative TEM images of
n-hexane soot and toluene soot. Table 2 lists the physical
parameters of hexane and toluene soot. It should be noted that
the size of the particles used here is close to the 30-50 nm
diameter carbon particles found in aircraft exhaust.22
difficult. As noted by Longfellow et al., in order to apply
laboratory measurements of the heterogeneous uptake measure-
ments to the atmosphere, the surface area of the soot must be
considered.
Another difference in the literature values is that the initial
uptake coefficient is reported in some cases, whereas a steady
state or an average value is reported in others. This means that
the reactivity of a surface covered with adsorbed molecules is
compared to the reactivity of the unreacted surface. Because of
site-blocking, adsorbate-adsorbate interactions, and electronic
effects, uptake coefficients are typically coverage dependent and
usually decrease as a function of coverage.1
8,19
In this study, the reaction of NO2 with freshly prepared hexane
soot has been studied using a Knudsen cell reactor and FT-IR
spectroscopy. A limited number of experiments were done on
toluene soot as well. The uptake coefficient for the NO2‚soot
reaction was measured using a Knudsen cell reactor. Using
models that take gas diffusion into account, an uptake coefficient
that takes into consideration the accessible surface area and the
increased number of gas surface collisions within the underlying
layers is determined. We report here initial and average values
of the uptake coefficient in order to compare to literature values.
In addition, we have quantified gas-phase reactants and products
in order to determine the total amount of NO2 reacted with the
soot surface as well as the branching of NO2 uptake on the
hexane soot surface to the production of HONO. FT-IR
experiments were done in order to gain some additional insight
into the heterogeneous reaction of NO2 on soot and the possible
For the Knudsen cell experiments, soot samples were evacu-
-
8
ated overnight to reach a pressure of ∼5 × 10 Torr inside
the reactor. The sample holder was then sealed and NO2
(
Matheson, 99.95% purity, used as received) was introduced to
the reaction chamber through a leak valve to passivate the walls
of the reactor. Experiments were run at NO2 pressures near 8
µTorr (equivalent to 11 ppb or 2.5 × 10 molecules cm ).
The gas pressure inside the reaction chamber was monitored
by an absolute pressure transducer (MKS 690 A.1TRC, range
.1-10 Torr). All experiments were done at 295 K.
Prior to each experiment, calibration of pressure vs mass
spectral intensity data for pure gaseous NO2 (m/e ) 46) was
made. By using pure NO2 for this calibration, the contribution
1
1
-3
-
6
0
4,6
role of adsorbed water that has been reported for this reaction.
15
of NO2 (m/e ) 47) to the m/e ) 47 ion signal monitored during
the course of the experiment can be calculated. The calibration
of pure NO2 also provides information about the fragmentation
of NO2 to NO (m/e ) 30). Calibration of HONO is more
difficult, since there is no readily available source of pure
gaseous HONO. The mass spectrum of HONO shows a parent
ion peak at m/e ) 47 and two major fragments at m/e ) 17
Experimental Section
A Knudsen cell reactor coupled to a quadruple mass
spectrometer was used to study the kinetics of the uptake of
NO2 on hexane and toluene soot and to quantify gas-phase
reactants and products. The reactor consists of a stainless steel
cross that serves as the reaction chamber. The region between
the reaction chamber and the mass spectrometer is separated
by a gate valve and the escape aperture or escape hole. The
size of the aperture can be changed in a vacuum using a linear-
rotary motion feedthrough. The quadruple mass spectrometer
+
+
(OH ) and m/e ) 30 (NO ); there is no apparent fragment at
+
23
m/e ) 46 (NO2 ). However, the parent ion peak (m/e ) 47)
15
has two contributions, the mass spectral intensity of NO2 and
1
5
HONO. Since the contribution of NO2 to the m/e ) 47 signal
1
4
is known from the NO2-signal, the mass spectral intensity of
HONO can be determined. In addition, calibration of pure
gaseous NO (Matheson, 99%) pressure versus mass spectral data
of m/e ) 30 ion signal was also done. The production of NO
can be calculated by correcting for contributions to the ion
(
UTI, 100C) is pumped by a 150 L/s ion pump, and the region
between the reactor and the mass spectrometer is pumped by a
0 L/s turbo pump (both from Varian). The stainless steel sample
holder sits on top of a tee support in the reaction chamber and
7

11928 J. Phys. Chem. A, Vol. 104, No. 51, 2000
Al-Abadleh and Grassian
Figure 2. (a) Knudsen cell data for the heterogeneous uptake of NO
on soot (15.1 mg sample). The soot particles are exposed to NO when
2
2
the sample lid is open. The QMS intensity for m/e ) 46 is shown. (b)
Observed uptake coefficient calculated via eq 1 from the data shown
in part a.
molecules (NO and HONO). Data were acquired every 0.5 s
for each ion monitored during the course of the experiment.
The IR cell used in this study has been modified from that
2
4
previously described. The new infrared cell now consists of a
Teflon-coated stainless steel cube which is placed on a linear
translator inside the sample compartment of the FT-IR spec-
trometer (Mattson RS-1, MCT narrow band detector). Freshly
prepared hexane or toluene soot was deposited directly on half
of a tungsten grid (3 cm × 2 cm) (Accumet Materials Co.).
The grid is then mounted inside the IR cell and evacuated
overnight. The linear translator allows each half of the sample
grid to be scanned by the IR beam in order to obtain spectra of
gas-phase and adsorbed reactants and products. Each spectrum
was recorded by averaging 250 scans at an instrument resolution
-
1
of 4 cm . NO2 pressures ranging from 5 to 100 mTorr
1
5
(
equivalent to 7-132 ppm or (1.6-32.0) × 10 molecules/
Figure 1. (a) TEM image of n-hexane soot taken at 100 000×
magnification. (b) TEM image of toluene soot taken at 100 000×
magnification. The bar in the left corner of each micrograph is
equivalent to 100 nm.
3
cm ) were used in the FT-IR experiments. Distilled H2O (Milli-
Q) was subjected to several cycles of freeze-pump-thaw before
use.
TABLE 2: Physical Parameters of n-Hexane and Toluene
Soot
Results
Parameter
n-Hexane soot
Toluene soot
A. Knudsen Cell Measurements for the Heterogeneous
39 ( 13a
Uptake of NO2 on Soot. Soot was directly deposited onto the
Knudsen cell sample holder, which is 11.95 cm in area. The
Particle diameter, nm
58 ( 17
2
-
3
-2
-2
Bulk density, g cm
Specific BET area, m g
(13 ( 6) × 10
(8 ( 2) × 10
2
-1
76 ( 3
2.0
55 ( 5
2.0
initial uptake coefficient for the reaction of NO2 with hexane
soot was measured as a function of sample mass (and thus
sample thickness). Representative Knudsen cell data are shown
in Figure 2. The QMS intensity of NO2 (m/e ) 46) was
monitored during the experiment (Figure 2a). When the sample
lid is open and the soot particles are exposed to the reactive
gas, there is a decrease in the QMS intensity. The QMS intensity
of NO2 is then converted to the observed uptake coefficient via
eq 1, as shown in Figure 2b. The initial observed uptake
coefficient, γo,obs, is taken as the maximum value for γ and found
b
-3
True density, g cm
a
The errors are 1 - σ statistical error. b Reference 27.
channel (m/e ) 30) from fragmentation of NO2 to NO. The
fragmentation of HONO to NO is taken from the literature.
By using gas kinetic theory in conjunction with the experimen-
tally determined cell parameters and calibration data, mass
spectral intensities of ions can then be converted to flow
2
3
-
1
(
molecule s ) for the reactant molecule (NO2) and product

Reaction of NO2 on Hexane Soot
J. Phys. Chem. A, Vol. 104, No. 51, 2000 11929
observed initial uptake coefficient is independent of the sample
mass. The region above 8 mg shows that the probe depth of
the gas-phase molecules measured during this initial time is
constant and equivalent to the probe depth reached for a sample
mass of 8 mg. In other words, on the time scale of the
measurement of the initial uptake coefficient, NO2 molecules
are diffusing into only a portion of the soot sample estimated
to be on the order of 1000 layers. For thin samples, the time
scale for the observed uptake coefficient to reach its maximum
value is on the order of the residence time (approximately 2 s).
For a sample mass of 7.7 mg, this time scale is slightly longer,
on the order of 3.5 s for the molecules to diffuse through all of
the underlying layers. The diffusion constant can then be
calculated from these data, the measured soot density, and eq
Figure 3. Mass dependence (thickness dependence) of the initial uptake
coefficient for the reaction of NO with hexane soot. Filled circles (b)
2
represent experimental data, and the solid line (s) is a fit to the data
2
9
2
that corrects for gas diffusion into the soot powder, giving a true initial
-
5
uptake cofficient of 7 × 10 (see text for further details).
1
/2
l ) (2Dt)
(2)
TABLE 3: Observed, BET, and Average Uptake Coefficient
2
Values of NO on n-Hexane Soot Using Different
where l is the root-mean-square of the distance traveled, D is
the diffusion constant, and t is the time. The parameter l can be
determined from the sample mass where the plateau region
begins (taken as 7.7 mg) using the geometric area of the sample
holder and the experimentally determined density of the soot
material (given in Table 2). The parameter t is taken as the
experimentally determined time for γ to reach its maxiumum
value (i.e. t ) 3.5 s) at the plateau region. Using a value of l )
Experimental Conditions
-
o,BET (×10-5)b
Sample mass (mg)
γ
o,obs (×10 3)a
γ
0
0
2
7
.6
.8
.0
.7
1.80
1.50
7.22
15.9
17.9
18.4
4.72
2.95
5.68
3.25
1
5.1
9.9
1.86c
1.45c
1
-
3
5
.0 × 10 cm and a time of 3.5 s, the diffusion constant for
a
Calculated via eq 1. b Calculated via eq 3. c For these two samples,
-6 2
NO2 in hexane soot is calculated to be 3.5 × 10 cm /s.
Table 3 shows the initial observed uptake coefficient for each
of the masses shown in Figure 3 along with an initial uptake
coefficient that has been recalculated using the entire BET area
the masses were in the plateau region, so using the total BET area of
these samples will result in a smaller uptake coefficient, because the
entire BET area was not accessed by the gas-phase molecules.
2
8
to be 0.018 from the data shown in Figure 2. It can be seen
from the data plotted in Figure 2 that over time the value of the
uptake coefficient decreases as the reaction proceeds. This
decrease occurs because the surface becomes less reactive with
continued exposure of NO2 and is interpreted as a coverage
dependent uptake coefficient.
of the soot samples according to eq 3:
A
S
γo,BET ) γo,obs × A
(3)
BET
Using the entire BET area is an overcorrection for samples that
are above ∼8 mg as not all of the surface area is being accessed
by the NO2 molecules during the time scale of the measurement
of the initial uptake coefficient. Indeed after correcting for the
entire BET surface area, the two lowest values obtained are for
the two samples whose masses are above 8 mg. The average of
The value of the observed initial uptake coefficient was found
to depend on the number of layers of soot present. The observed
initial uptake coefficient measured as a function of the sample
mass is plotted in Figure 3. Since the sample holder is of a
2
fixed area, 11.95 cm , increasing the sample mass increases the
number of layers of soot. The line through the data is determined
from the gas diffusion model that Keyser et al. adapted from
-5
the first four values of γo,BET is 4.2 × 10 .
For the same experimental data shown in Figure 3, average
observed uptake coefficients were calculated over a 140 s time
period. The data plotted in Figure 4 show γave,obs as a function
of mass of n-hexane soot. It is seen that γave,obs is a linear
function of sample mass over the entire range. The average
uptake coefficient can be corrected for the accessible BET area
using a linear least-squares fit of the line through the data points.
This correction is then obtained from eq 4:
25
the catalysis area to atmospherically relevant heterogeneous
26
reactions. This is a semiempirical model that accounts for gas
diffusion into porous samples and corrects the observed uptake
coefficient for contribution from underlying layers of particles.
The parameters listed in Table 2 were used as input data in the
diffusion model for the reaction of NO2 with hexane soot. There
are two parameters that are adjusted in order to obtain a good
fit: the true initial uptake coefficient, γo,t, which takes into
account the accessible surface area of the underlying layers,
and the tortuosity, τ, which is related to the diffusion of the
molecules through the sample. A value of 1 for the tortuosity
2
A (cm )
-
1
S
γave,BET ) slope (mg ) ×
(4)
2
-1
SBET (cm mg )
-
5
and a value of 7 × 10 for the true initial uptake gave the best
-4
fit to the data.
The slope of the line in Figure 4 is 7.7 × 10 mg-1 and the
-
5
A limited number of experiments was done on toluene soot.
Initial uptake coefficient values of NO2 on toluene soot were
lower than those measured for hexane soot by a factor of 4.
As discussed in Underwood et al.,28 there are two distinct
regions in the plot shown in Figure 3. The region that extends
from 0 to approximately 8 mg shows that the observed initial
uptake coefficient has a nearly linear dependence on sample
mass. The portion of the plot above 8 mg shows that the
value of γt calculated using eq 3 is 1.2 × 10 . This is a factor
of 6 times lower than the true initial value of the uptake
coefficient. Although the application of the diffusion model in
Figure 3 predicts a nearly linear mass regime to be in the range
from 0 to 8 mg for the initial uptake coefficient, the dependence
of the average uptake coefficient on the mass is still linear up
to 19 mg, as shown in Figure 4. The linear region for any time-
averaged uptake coefficient can be predicted from the diffusion

11930 J. Phys. Chem. A, Vol. 104, No. 51, 2000
Al-Abadleh and Grassian
TABLE 4: Values of the Uptake Coefficient
a) Initial and at Specific Surface Coverages
(
Mass of BET surface
γ
o,BET
γ
0.1,BET
γ
0.25,BET
γ
0.50,BET
2
(×10-5) (×10-5) (×10-5) (×10-5)
soot (mg)
area (cm )
0
0
2
7
5.1
9.9
.6
.8
.0
.7
456
608
1520
5852
11476
15124
4.72
2.95
5.68
3.25
1.86
1.66
1.00
1.26
1.01
0.37
1.12
0.82
0.31
0.73
0.44
0.22
0.76
0.48
a
0.42
a
a
0.32
0.33
1
1
(b) Averaged over Time And Surface Coverages
Figure 4. Linear mass dependence of the average uptake coefficient,
averaged over 140 seconds. Filled circles (b) represent experimental
data, and the solid line (-) represents a linear least-squares fit to the
γ
ave,BET (×10-5)
Mass of
soot (mg)
0.1 coverageb
0.25b
0.50b
at 140 s
at 300 s
-
4
-1
0.6
0.83
0.84
1.86
0.93
1.37
1.15
0.74
0.51
1.03
0.58
1.00
.99
1.41
1.77
1.72
0.80
1.37
1.03
0.69
1.17
0.76
0.48
0.99
0.76
a
0.69
a
a
0.71
0.60
data of the form y ) mx. The slope of the fit is 7.7 × 10 mg , with
an R value of 0.986.
0
2
7
5.1
9.9
.8
.0
.7
1
1
a
Some experiments were ended before a coverage of 0.50 was
reached. b Surface coverage values of 0.1, 0.25, 0.5 were calculated
using the maxim coverage of NO on soot, which under the conditions
of this study (pressure of 8 µTorr) is taken as 1.4 × 10 molecules
2
1
3
-
2
cm
.
to be 36 ( 5%. The remaining NO2 taken to the surface may
correspond to the formation of surface-bound products (vide
infra) and other gas products such as NO. The percentage of
HONO produced in this study is lower than that reported in
5
another Knudsen cell study. This may be a result of using a
different hydrocarbon as a soot generator and collecting soot at
different heights from the flame base both of which has been
shown to effect HONO production.
2
Figure 5. Calibrated flow of NO reacted and HONO produced (7.7
mg sample).
4
,5
-
6
2
constant determined above, i.e., l ) [2 × (3.5 × 10 cm /s)
From the calibrated data it is also possible to determine the
absolute number of NO2 molecules reacted per unit surface area
(or unit mass of soot). For small masses of soot (<1 mg),
complete saturation of the surface can be obtained by continuing
the reaction until no further uptake of NO2 is observed. This
means that over time the uptake coefficient goes to zero. The
data in Figure 5 can be used to determine the total amount of
NO2 reacted per unit surface area. For thin or low mass samples,
complete saturation occurs over the time period of the experi-
ment, whereas for thicker samples with larger mass, the data
are fit to a double exponential form, and the fit is then
extrapolated to the limit of no further uptake of NO2 corre-
sponding to a saturated surface. From all of the data, it is
determined that the total amount of NO2 that can react at a
1
/2
-2
×
140] ) 3.2 × 10 cm. It is predicted that, in a 140 s period,
the uptake coefficient would be linear up to a diffusion length
-
2
3
of 3.1 × 10 cm. From the soot density of 0.13 g/cm and
sample geometric area this diffusion length is equivalent to a
mass of 48 mg. These data show that NO2 molecules can access
additional underlying layers of the soot sample over time and
that the probe depth of NO2 through the powdered sample
increases with time.
In some experiments, the parent peak of HNO3 (m/e ) 63)
was monitored. No change in the mass spectral intensity of
HNO3 was observed, suggesting that HNO3 is not formed as a
gaseous reaction product. This result agrees with the study of
7
Kalberer et al., in which HNO3 was not detected either in the
1
3
gas phase or adsorbed on the aerosol particles or the walls of
the reactor. The HONO mass channel (m/e ) 47) was also
monitored. There was an increase in the m/e ) 47 signal as
NO2 reacted with the surface. Quantification of the HONO
production is described below.
B. Calibrated Flow Experiments Used To Determine the
Branching of NO2 to HONO and the Total Uptake. The QMS
intensity for the parent ions of NO2 and HONO, m/e ) 46 and
pressure near 8 µTorr is determined to be (1.4 ( 0.5) × 10
-2
molecules cm . This value is used as a saturation coverage
for NO2 on soot at 295 K and a pressure of 8 µTorr.
1
3
-2
Using a saturation coverage of 1.4 × 10 molecules cm ,
the uptake coefficient can be calculated at specific coverages.
These data are presented in Table 4, which lists initial uptake
coefficients and those determined for NO2 coverages of one-
tenth, one-quarter, and one-half (0.1, 0.25 and 0.5) of the
1
2
4
7, respectively, were calibrated and converted to molecular
flow through the Knudsen cell. These data are shown in Figure
. The flow of NO2 has been offset and inverted to show the
saturation coverage, which corresponds to 1.4 × 10 , 3.5 ×
12 12 -2
10 , and 7 × 10 molecules cm , respectively. All of the
numbers reported in Table 4 have been corrected for the entire
BET area of the soot sample. It is seen from the data for a
specified mass of soot that γ decreases as the coverage of NO2
increase. The reaction of NO2 with soot is not catalytic under
the conditions used in the study, and therefore active sites are
either blocked by reaction products from irreversible NO2
adsorption or from chemical deactivation.
5
amount of NO2 that reacts per second. Since the m/e ) 47 signal
from the mass spectrometer contains two contributions: the
parent ions of HONO and the natural abundance 15N-labeled
NO2. Calibration of the m/e ) 47 signal for HONO production
was done according to the procedure outlined in the Experi-
mental Section. From the calibrated data presented in Figure 5,
the amount of HONO produced per NO2 reacted is determined
For one set of experimental data, the variation of γBET with

Reaction of NO2 on Hexane Soot
J. Phys. Chem. A, Vol. 104, No. 51, 2000 11931
TABLE 5: Assignment of Vibrational Bands of Surface
Bound Species from the Adsorption of NO
2
on Hexane Soot
Surface species
Wavenumber (cm-1)
a
b
R-ONO
R-NO
1281, 1280
a
b,c
2
1323, 1340
2-
1413a
CO
3
(ad)
a
b,c
R-NO
2
1531, 1540
a,b
R-N-NO
R-O-NO
2
1565 a
c
1653, 1660
a
b
lactone and alkyl carbonyl groups
1779, 1775
a
This work. b Reference 9. c Reference 13.
of the gas phase, the infrared spectrum of the surface remains
nearly identical, with only a small decrease in intensity of the
-
1
Figure 6. Log of γBET versus NO
the data fits the linear equation y ) -4.18 - (1.46 × 10 )x. At higher
coverages of NO
, data fits the linear equation y ) -4.52 - (2.96 ×
2
coverage. At low coverage of NO
2
,
bands between 1400 and 1675 cm . On the basis of literature
assignments, the absorptions bands of the surface-bound species
can be assigned to different functional groups present on the
soot surface after reaction with NO2 (see Table 5).
-12
2
-
13
1
0
)x.
Several additional pressures of NO2 were introduced into the
infrared cell containing the soot sample until there was no longer
any more changes in the surface spectrum. To try and reactivate
the soot sample, the infrared cell was evacuated overnight and
then 5.0 Torr of distilled H2O was added to the infrared cell
for 20 min, as water has been shown to reactivate the soot
4,6
surface. A spectrum of the surfacesafter evacuation of waters
was compared to that before the addition of water to see if water
caused a change to the surface or not. The bending mode of
-
1
adsorbed H2O at 1625 cm was seen in the spectrum. After
that, successive pressures of NO2 were added to the surface in
the same manner. Spectra of the surface were referenced to that
after evacuation of H2O. These spectra show an obvious
decrease in the intensity of the 1625 and 1779 cm- bands due
to the loss of H2O and CdO groups from the surface after the
addition of 52, 75, and 95 mTorr of NO2. Addition of pressures
of NO2 less than 52 mTorr did not show bands of adsorbed
1
Figure 7. FT-IR spectra of hexane soot surface recorded as a function
of NO pressure (5, 11, 23, 46, 63, and 91 mTorr). Spectra were
2
2
recorded in the presence of gas-phase NO (gas-phase absorptions were
subtracted out).
30
NO2 coverage was determined in order to further understand
2
NO on the surface. HONO(g) bands at 1264, 1703, and 3591
-1
the coverage dependence of γ. Figure 6 shows a plot the log of
cm were not observed in any of the gas-phase spectra recorded
-
2
γBET as a function of NO2 coverage in molecules cm . The
data can be fit to two straight lines corresponding to a double
exponential fit. At low coverages of NO2, the linear least-squares
in the presence of NO . This may be a result of using a short
2
IR path length, since relatively small amounts of HONO are
produced from the reaction of NO with hexane soot. However,
2
-
12
fit of the data is y ) -4.18 - (1.46 × 10 )x. However, at
the loss of water from the surface and the growth of the
-
1
higher coverages of NO2, data fits the linear equation y ) -4.52
carbonate band at 1413 cm suggests further oxidation of the
surface.
-
13
-
(2.96 × 10 )x. It is not clear what the exact fundamental
nature of the coverage dependence is due to. This type of
coverage dependence may be explained by several reasons,
1
8,19
Discussion
including surface site heterogeneity and repulsive adsorbate-
adsorbate interactions. As will be discussed in the next section,
adsorbed NO2 does remain on the surface of the soot particles.
C. FT-IR Study of the Heterogeneous Reaction of NO2
on Soot. Figure 7 shows the FT-IR spectra of hexane soot (1
mg) as a function of NO2 pressure. FT-IR experiments were
done at higher pressures (ppm) than the Knudsen cell experi-
ments, because of the lower sensitivity of the technique. The
infrared spectra recorded as a function of increasing pressure
of NO2 from 5 to 91 mTorr are shown in Figure 7. These spectra
were recorded in the presence of gas-phase NO2; the unreacted
soot surface was taken as the reference background spectrum.
The spectra showing only surface-bound products are obtained
by spectral subtraction of gas-phase absorptions. The intensity
In two recent studies of the NO2 reaction on soot aerosol,
the analysis of the data obtained for the same reaction using a
Knudsen cell reactor has been called into question.6 This is
because in Knudsen cell studiessoften referred to as bulk
studiessuptake coefficients are usually calculated using only
,17
5
,11
the geometrical surface area of the sample holder. Using the
geometric area as the effective surface area for heterogeneous
uptake may introduce a major under estimation of the available
surface area for reaction. This will especially be true for small
particles with high surface areas, such as soot. Soot is a porous
material, and with its fractal geometry, a relatively large amount
of surface area may be accessible for NO2 to adsorb and react.
In this study, initial uptake coefficient values calculated using
the geometric area were corrected to take into account the
diffusion of NO2 into the interparticle pores of the soot sample
and its ability to access underlying soot layers. By doing so the
-1
of the bands at 1281, 1323, 1531, 1565, and 1653 cm increase
as the coverage of NO2 is increased. There is a decrease in the
-
1
intensity of the band at 1779 cm , suggesting the loss of the
CdO group from the carbonyl surface, presumably due to the
formation of gas-phase CO2 or CO, as has been observed in
other studies done at high pressures of NO2.1 Upon evacuation
-
5
true value of the initial uptake is determined to be 7 × 10 ,
significantly lower than the value obtained using the geometric
area.
4

1
1932 J. Phys. Chem. A, Vol. 104, No. 51, 2000
Al-Abadleh and Grassian
TABLE 7: Uptake Coefficient Values Reported in the
TABLE 6: Comparison of Initial, Specific Coverage, and
Average Uptake Coefficients
Literature Corrected for BET Surface Area
Geometric
uptake
coefficient, γ
a. Longfellow et al.a
Uptake
BET uptake
coefficient,
coefficient
Average values
Range of values
BET surface
2
-1
Soot
Mass (mg) area (m g
)
g
γ
BET
Initial Uptake Coefficients
-
3
γ
o,obs
(1.8-18) × 10
-
5
-5
-4
-5
-6
-7
-7
γ
o,BET
(3.4 ( 1.6) × 10
(1.5-5.7) × 10
Methane
Propane
Hexane
2.6
4.0
1.0
25
130a
89
76
na
3.0 × 10
2.5 × 10
3.2 × 10
6.2 × 10
7.2 × 10
-
4
3.0 × 10
1.0 × 10
Uptake Coefficients at Different Coverages
-5
-
-
-
6
6
6
-6
γ
γ
γ
0.1,BET
0.25,BET
0.5,BET
(9.3 ( 0.3) × 10
(4.9 ( 0.2) × 10
(3.6 ( 0.6) × 10
(3.7-12.6) × 10
b
-6
-6
(2.2-7.6) × 10
-
-
5
2
c
Kerosene
2.1
5.0 × 10
na
(3.2-4.6) × 10
b. Gerecke et al.d
76
a
Time Averaged Uptake Coefficients
-4
-
-
5
6
-5
Ethylene
8.0
9.5 × 10
3.0 × 10
γ
γ
ave,BET (140 s)
(1.2 ( 0.4) × 10
(0.8-1.9) × 10
-6
(8.2 ( 2.2) × 10
(5.1-10.3) × 10
a
ave,BET (300 s)
For Longfellow et al. study (ref 4), γBET was calculated using eq
2
b
3 with the geometric area of the flow reactor equal to 55 cm . The
Coverage Averaged Uptake Coefficients
2
b
-
5
-5
given BET surface area in Longfellow et al. was 89 m /g. BET area
determined in this study. na: not available. For Gerecke et al. study
γ
γ
γ
ave,BET (0.1)
ave,BET (0.25)
ave,BET (0.5)
(1.4 ( 0.4) × 10
(0.8-1.8) × 10
-
6
-6
c
d
(8 ( 2) × 10
(5-11) × 10
-
6
-6
(6.7 ( 0.6) × 10
(6.0-7.1) × 10
(ref 5), γBET was calculated using eq 3 with the geometric area of the
2
sample holder equal to 19.6 cm .
a
Uptake coefficients averaged from time t ) 0 to the indicated time.
b
Uptake coefficients averaged from a coverage of zero to the indicated
coverage.
The data averaged over time and coverage can be compared
to some of the measurements made using different techniques.
4
Longfellow et al. studied the reaction of NO2 on methane,
It is also typical to report initial uptake coefficients determined
propane, hexane, and kerosene soot using a flow reactor. Uptake
coefficient values reported in their study are corrected for BET
surface area in Table 7. It can be seen that the steady state γBET
for the hexane soot is a factor of 10-20 times lower than the
average (coverage and time) reported in Tables 4 and 6. This
may be due to a temperature effect, since their experiments were
done at 262 K.
for very short reaction times in Knudsen cell experiments. In
other measurements using aerosol chambers, flow reactors, and
other static reactors, average or steady-state values are often
reported. When γave is reported in the literature, it corresponds
to the reaction occurring over longer time scales. From study
to study, uptake coefficients are averaged over different time
scales and, therefore, over different surface coverages. Here we
report γBET averaged over coverage. This is a more fundamental
quantity for surface reactions than time since different soot
samples of varying mass can reach the same coverage at
different times. The major assumption made in this analysis is
that the underlying layers and the top layers of the soot sample
are reaching the same coverage at the same time, when in fact
there may be a coverage gradient with the top layers being at
higher coverages compared to the bottom most layers. Further
analysis of the data would require detailed knowledge of
saturation effects of NO2 adsorbed on soot samples. An analysis
of this type is currently underway.
Kleffmann et al.6 studied the consumption of NO2 on
commercial soot samples with a well-characterized BET surface
area. Mean uptake coefficient values were derived from the
slopes of the first-order decay constants during the first 5 min
of the reaction as a function of the BET surface area of the
-
6
soot samples. Those γ values were determined to be ∼10 for
1
3
-2
the consumption of ∼10 NO2 cm , close to the values that
we obtain for hexane soot. In their study, uptake coefficient
values were not determined for freshly prepared soot samples,
because of the unavailability of the BET surface area values
for these samples.
To determine the atmospheric significance of the reaction of
NO2 to yield HONO, the reactive uptake coefficient must be
known. Under the dry conditions reported here, the initial
reactive uptake coefficient would be the fraction of NO2
converted to HONO times the measured uptake coefficient, i.e.,
Table 6 summarizes the uptake coefficients calculated in
different ways (initial uptake coefficients determined using
geometric and BET areas, uptake coefficients at specified
coverages, and uptake coefficients averaged over coverage and
time). Initial uptake coefficients corrected for the BET surface
area of soot samples have a narrower range than the observed
uptake coefficient using the sample holder geometric area. The
coverage and time-average determined uptake coefficients are
lower than the initial ones. The FT-IR data show that the surface
becomes covered with adsorbed molecules as the reaction
proceeds. Once the surface was covered, it was not possible to
recover the reactivity of the soot particles.
-
5
-5
0
.36 × (7 × 10 ) ) 2.5 × 10 . Atmospheric models to
determine the importance of this reaction under atmospheric
conditions would need to take into account the coverage
dependence of the uptake coefficient and the fact that the surface
becomes deactivated over time. Further analysis of this reaction
would require additional experimentation under more atmo-
spherically relevant conditions that span the temperature and
relative humidity range of the troposphere.
The data generated in this study can be used to reexamine
measurements made in other laboratories. The geometric uptake
Conclusions
5
coefficient reported by Gerecke et al. using a Knudsen cell
reactor can be corrected for diffusion of NO2 (see Table 7).
They reported an initial uptake coefficient for NO2 on ethlylene
soot to be 0.095 ( 0.007. An 8.0 mg sample of soot reported
In this study, the heterogeneous reaction of NO with freshly
prepared hexane soot has been studied using a Knudsen cell
2
reactor and FT-IR spectroscopy at 295 K. The FT-IR data show
that adsorbed products remain on the surface in agreement with
another very recent FT-IR investigation of the NO2 soot
2
in their study has a BET surface area of 6080 cm (assuming
2
-1
that the BET surface area of ethylene soot equals 76 m g ).
Using eq 3 with the geometric area of the sample holder equal
31
reaction. HONO was detected as a gas-phase reaction product,
accounting for 36% of the NO2 reacted. Initial uptake coef-
ficients were determined by accounting for the accessible surface
area of the soot samples on the time scale of the measurement.
2
-4
to 19.6 cm , γo,BET can be calculated to be about 3 × 10 ,
which is approximately a factor of 4 times greater than the initial
uptake determined here of 7 × 10 using a gas-diffusion model.
-5

Reaction of NO2 on Hexane Soot
J. Phys. Chem. A, Vol. 104, No. 51, 2000 11933
The time scale is on the order of the residence time of the gas
molecules inside the Knudsen cell reactor. On this time scale,
diffusion into the bulk powder can occur. Diffusion of gases
into the bulk of the material may be important in other systems
involving solid particle surfaces as well, thus making the
geometric area inappropriate for use in determining the uptake
coefficient. Uptake coefficients averaged over longer times and
higher coverages were nearly an order of magnitude lower than
the initial uptake coefficient. Comparison to values reported in
the literature using other techniques are now in better agreement
with the Knudsen cell data when the BET areas of the soot
samples are considered and average values of the uptake
coefficient are compared. This discrepancy was previously 5
orders of magnitude, making it difficult to fully assess the
importance of the heterogeneous reaction of NO2 on soot to
form HONO in the atmosphere.
(10) Kalberer, M.; Ammann, M.; G a¨ ggeler, H. M.; Baltensperger, U.
Atmos. EnViron. 1999, 33, 2815.
(
11) Rogaski, C. A.; Golden, D. M.; Williams, L. R. Geo. Phys. Res.
Lett. 1997, 24, 381.
12) Kalberer, M.; Tabor, K.; Ammann, M.; Parrat, Y.; Weingarther,
(
E.; Pighet, D.; R o¨ ssler, E.; Jost, D. T.; T u¨ rler, A.; G a¨ ggeler, H. W.;
Baltensperger, U. J. Phys. Chem. 1996, 100, 15487.
(
13) Smith, D. M.; Welch, W. F.; Graham, S. M.; Chughtai, A. R.;
Wicke, B. G.; Grady, K. A. Appl. Spectrosc. 1988, 42, 674.
14) Chughtai, A. R.; Welch, W. F.; JR.; Akhter, M. S.; Smith, D. M.
(
Appl. Spectrosc. 1990, 44, 294.
(15) Choi, W.; Leu, M.-T. J. Phys. Chem. A. 1998, 102, 7618.
(
16) Harrison, R. M.; Peak, J. D.; Collins, G. M. J. Geophys. Res. 1996,
01-D9, 14429.
17) Aumont, B.; Madronich, S.; Ammann, M.; Kalberer, M.; Bal-
1
(
tensperger, U.; Hauglustaine, D.; Brocheton, F. J. Geophys. Res. 1999, 104-
D1, 1729.
(18) Adamson, A. W. Physical Chemistry of Surfaces, 5th ed.; John
Wiley & Sons: New York, 1990.
(19) Masel, R. Principles of Adsorption and Reaction on Solid surfaces;
John Wiley & Sons: New York, 1996.
Acknowledgment. The authors gratefully acknowledge the
Department of Energy-Atmospheric Chemistry Program for
support of this work (DE-FG01-98ER62580).
(
(
20) Beichert, P.; Finlayson-Pitts, B. J. J. Phys. Chem. 1996, 100, 15218.
21) Golden, D. M.; Spokes, G. M.; Benson, S. W. Angew Chem., Ent.
Ed. Engl. 1973, 12, 534.
22) Hagen, D. E.; Trueblood, M. B.; Whitefield, P. D. Particulate Sci.
Technol. 1992, 10, 53.
23) Fenter, F. F.; Caloz, F.; Rossi, M. J. J. Phys. Chem. 1996, 100,
1008.
(24) Miller, T. M.; Grassian, V. H. J. Am. Chem. Soc. 1995, 117, 10969.
Appl. Spec. 1988, 42, 674.
25) Aris, R. The Mathematical Theory of Diffusion and Reaction in
Permeable Catalysts; Clarendon Press: Oxford, 1975.
26) Keyser, L. F.; Moore, S. B.; Leu, M.-T. J. Phys. Chem. 1991, 95,
496.
27) Lide, D. R. CRC Handbook of Chemistry and Physics, 72nd ed.;
CRC Press: Boca Raton, FL, 1991; pp 4-50.
28) Underwood, G. M.; Li, P.; Usher, C. R.; Grassian, V. H. J. Phys.
Chem. A. 2000, 104, 819.
29) Atkins, P. A. Physical Chemistry, 6th ed.; W. H. Freeman and
(
References and Notes
(
(
1) Noble, C. A.; Prather, K. A. Phys. World 1998, 11, 39.
(2) Kamm, S.; M o¨ hler, O.; Naumann, K.-H.; Saathoff, H.; Schurath,
U. Atmos. EnViron. 1999, 33, 4651.
3) Koehler, B. G.; Nicholson, V. T.; Roe, H. G.; Whitney, E. S. J.
Geophys. Res. 1999, 104-D5, 5507.
4) Longfellow, C. A.; Ravishankara, A. R.; Hanson, O. R. J. Geophys.
Res. 1999, 104-D11, 13833.
5) Gerecke, A.; Thielmann, A.; Gutzwiller, L.; Rossi, M. Geophys.
Res. Lett. 1998, 25-13, 2453.
6) Kleffmann, J.; Becker, K.; Lackhoff, M.; Wiesen, P. Phys. Chem.
Chem. Phys. 1999, 1, 5443.
7) Kalberer, M.; Ammann, M.; Arens, F.; G a¨ ggeler, H. M.; Bal-
tensperger, U. J. Geophys. Res. 1999, 104-D11, 13825.
8) Ammann, M.; Kalberer, M.; Jost, D. T., Tolber, L.; R o¨ ssler, E.;
Piguet, H. W.; G a¨ ggeler, H. W.; Baltensperger, U. Nature 1998, 395, 157.
9) Akhter, M. S.; Chughtai, A. R.; Smith, D. M. J. Phys. Chem. 1984,
8, 5334.
(
(
(
(
5
(
(
(
(
(
(
Co.: New York, 1998; p.753.
(
(30) Jacox, M. E. J. Phys. Chem. Ref. Data. 1990, 19, 1444.
(31) Kirchner, V.; Scheer, V.; Vogt, R. J. Phys. Chem. A 2000, 104,
8908.
(
8