Atmospheric heterogeneous reactions of benzo(a)pyrene

Article abstract of DOI:10.1524/zpch.2010.6145

This experimental study deals with heterogeneous reactions of benzo(a)pyrene (BaP) with ozone, nitrogen dioxide and hydroxyl radicals. BaP was adsorbed on silica particles chosen here as a model of mineral atmospheric particles. Compound extractions were assisted by focused microwave and analyses were performed by gas chromatography coupled with mass spectroscopy in single ion monitoring mode. Pseudo-first order rate constants were obtained from the fit of experimental decays of particulate-BaP concentration versus reaction time. Second order rate constants were determined considering the different oxidant gaseous concentrations except for the case of hydroxyl radicals where only a pseudo-first order rate constant was proposed. Values obtained at room temperature are (2.1±0.5)×10^-15 cm³ molecule^-1 s^-1 for (BaP + ozone), (5.8±1.4) ×10^-16 cm³ molecule^-1 s^-1 for (BaP + nitrogen dioxide) and (3.4±0.8)×10^-2 s ^-1 for (BaP + OH) reactions. Products have only been investigated for the NO₂ and the OH (in the presence of NO_x) reactions. 1-, 3- and 6-nitrobenzo(a)pyrenes were detected as degradation products and quantified. Reaction rate constants for product formation are (3.7±0.9)×10^-16 cm³ molecule^-1 s^-1 for 6-NBaP, (2.2±0.6)×10^-17 cm ³ molecule^-1 s^-1 for 1-NBaP and (5.3±1.3)×10^-17 cm³ molecule^-1 s^-1 for 3-NBaP. 1-, 3- and 6-nitroBaP account respectively for approximately 5%, 12% and 83% of total nitrated species. If in thepresence of only nitrogen dioxide, BaP was totally degraded within few minutes, only 20 to 25 % of the initial BaP led to nitrated compounds when reacting with OH (in the presence of NO_x). by Oldenbourg Wissenschaftsverlag.

Full text of DOI:10.1524/zpch.2010.6145

Z. Phys. Chem. 224 (2010) 1151–1170 . DOI 10.1524.zpch.2010.6145
by Oldenbourg Wissenschaftsverlag, München
©
Atmospheric Heterogeneous Reactions of
Benzo(a)pyrene
By M. Cazaunau , K. Le Ménach , H. Budzinski , and E. Villenave1,2,*
1
,2
1,2
1,2
1
Université de Bordeaux, Institut des Sciences Moléculaires, Laboratoire de Physico-Toxico
Chimie de l’Environnement, 33405 Talence cedex, France
2
CNRS, UMR 5255, 33405 Talence cedex, France
th
Dedicated to Prof. Dr. Reinhard Zellner on the occasion of his 65 birthday
(
Received January 16, 2010; accepted March 29, 2010)
PAH . Reactivity . Benzo(a)pyrene . Nitrobenzo(a)pyrene . Ozone .
Nitrogen Dioxide . Hydroxyl Radical . Atmospheric Particle . Silica . GC-MS
This experimental study deals with heterogeneous reactions of benzo(a)pyrene (BaP) with
ozone, nitrogen dioxide and hydroxyl radicals. BaP was adsorbed on silica particles chosen here
as a model of mineral atmospheric particles. Compound extractions were assisted by focused
microwave and analyses were performed by gas chromatography coupled with mass spectro-
scopy in single ion monitoring mode. Pseudo-first order rate constants were obtained from the
fit of experimental decays of particulate-BaP concentration versus reaction time. Second order
rate constants were determined considering the different oxidant gaseous concentrations except
for the case of hydroxyl radicals where only a pseudo-first order rate constant was proposed.
K15
3
K1 K1
Values obtained at room temperature are (2.1±0.5)!10
cm molecule
for (BaP nitrogen dioxide) and
for (BaP + OH) reactions. Products have only been investigated for the
and the OH (in the presence of NO ) reactions. 1-, 3- and 6-nitrobenzo(a)pyrenes were
s
for (BaP +
K16
3
K1
K1
ozone), (5.8±1.4)!10
cm molecule
s
+
K2 K1
(
3.4±0.8)!10
2
s
NO
x
detected as degradation products and quantified. Reaction rate constants for product formation
K16
3
K1 K1
K17
3
K1 K1
are (3.7±0.9)!10
cm molecule
s
for 6-NBaP, (2.2±0.6)!10
cm molecule
s
K17
3
K1 K1
for 1-NBaP and (5.3±1.3)!10
cm molecule
s
for 3-NBaP. 1-, 3- and 6-nitroBaP
account respectively for approximately 5%, 12% and 83% of total nitrated species. If in the
presence of only nitrogen dioxide, BaP was totally degraded within few minutes, only 20 to
2
5 % of the initial BaP led to nitrated compounds when reacting with OH (in the presence of
x
NO ).
1. Introduction
Polycyclic Aromatic Hydrocarbons (PAHs) are ubiquitous contaminants in the
environment. Such compounds are mainly formed during natural and anthropo-
*
Corresponding author. E-mail: e.villenave@ism.u-bordeaux1.fr

1
152
M. Cazaunau et al.
genic incomplete combustion processes [26]. In the atmosphere, almost 90% of
PAH sources are anthropogenic (Finlayson-Pitts et al., 1986). They are well-
known to have mutagenic and.or carcinogenic activities for humans through
metabolic activation pathways [12]. In 1979, US-EPA defined 16 PAHs as prior-
ity pollutants [33] with a particular focus on benzo(a)pyrene (BaP) which is one
of the most mutagenic and carcinogenic PAH [13, 28]. BaP is nowadays used
by regulators as the basis for quantitative risk estimation of PAH-containing
combustion by-products [62].
In the atmosphere, due to its low vapour pressure and its high molecular
weight, BaP is mainly present adsorbed onto particulate matter [19] and may
be adsorbed either on carbonaceous particles, directly emitted from combustion
processes [1, 44, 59], or on mineral particles arising from soil erosion [54].
After formation and emission, particulate-BaP may be subject to various
degradation processes such as photodegradation and chemical oxidation with O3,
NO or OH. These reactions are important to consider as they can lead to the
x
formation of particularly toxic products like oxygenated and.or nitrated com-
pounds. It has already been demonstrated that such species are generally much
more toxic than the BaP itself [30].
Although considerable research has already been performed on the degrada-
tion of BaP under simulated atmospheric conditions [1, 5, 6, 16, 18, 44, 56, 57,
59], there is still a lack of knowledge about the kinetics and fate of PAHs ad-
sorbed on particles. Several studies have been performed on different types of
particulate matters such as natural carbonaceous particles [18, 36, 49], artificial
carbonaceous particles [17, 42, 44, 49, 59, 60], model or natural mineral particles
[5, 6, 25, 50, 52, 56, 57, 66] and salt aerosol [41]. These studies exposed coated-
particles to different gaseous species using various experimental set-ups (reaction
chamber, fast flow and quasi-static reactors…) in order to measure the kinetics
of oxidation reactions of adsorbed compounds and identify or predict their degra-
dation products. For the reaction of BaP with ozone, oxygenated compounds,
such as ketones, quinones, lactones and carboxylic acids, predominate as degra-
dation products [30, 65]. When looking at potential oxygenated particulate pro-
ducts in the atmosphere, it may be noted that benzo(a)pyrene dione has for
instance been detected in natural particulate samples in different countries in the
world [7, 36]. In laboratory studies, experiments have demonstrated the forma-
tion of BaP-6, 12-dione, BaP-1, 6-dione and BaP-3, 6-dione when artificial soot
was exposed to ozone [37]. Lactones can be formed both by photoreaction and.
or ozonolysis and contribute to a high mutagenicity of particulate matter [30].
Reactions with hydroxyl radicals in the presence of nitrogen dioxide can also
lead to the formation of oxygenated compounds such as hydroxylated BaP,
formed by the loss of one molecule of nitrous acid [16, 61]. Nitrated compounds
have also been found on ambient particles [2, 3, 4]. The origin of the formation
of nitro-PAHs may vary: they may be formed directly during combustion proc-
esses or in the gas phase during the atmospheric transport or by heterogeneous
reactions in the atmosphere at the gas.particle interface. During such heterogene-

Atmospheric Heterogeneous Reactions
1153
ous processes, BaP may react with the nitrate radical or nitrogen dioxide by
electrophilic substitution with elimination of nitrous acid [7, 21, 34, 35]. The
formation of nitrated species may also be initiated by the hydroxyl radical, fol-
lowed by reaction with nitrogen dioxide and the loss of one molecule of water
[
9, 15, 20, 61]. Nitro-BaP and dinitro-BaP are only occasionally selected for
chemical analysis despite the fact they are strong direct-acting mutagens [27, 30,
6, 47]. For instance, 6-nitro-BaP has already been detected on atmospheric
particles [2, 3, 4]. In laboratory studies, 1- and 3-nitro-BaP [38] and dinitroBaP
4
[
30] have been detected to be generated by similar processes.
In this work, heterogeneous reactions of O , NO and OH (in the presence
3
2
of NO ) with BaP adsorbed on silica particles have been investigated. Silica was
x
chosen as a model of atmospheric particles: first it belongs to the major elements
as oxides in the continental crust [63] and it is also one of the most used inor-
ganic proxy in heterogeneous reaction studies due to its well-known physico-
chemical properties [5, 6, 10, 50, 56, 57, 66]. Room temperature second order
reaction rate constants have been determined for the reactions with O and
3
NOand compared to previous measurements. Finally, identification and quantifi-
2
cation of some resulting degradation products were performed.
2. Experimental
2.1 Chemicals
The chemicals used in this work were as follows: Benzo(a)pyrene (≥ 97%, Al-
drich), 6-nitro-benzo(a)pyrene (99.8 %, Promochem), benzo(a)pyrene d12 (98 %,
Cambridge Isotope Laboratories), benzo(b)fluoranthene d12 (98 %, Cambridge
Isotope Laboratories). Analytical grade solvents were dichloromethane (Acros
Organics, for residue and pesticides analysis) and isooctane (Scharlau, HPLC
Grade). Standard solutions were prepared by dissolving standard crystals in iso-
octane. Standard and solvent quantities were measured by gravimetry. These
solutions were kept at -20°C and in complete darkness.
2.2 Silica particles
Silica particles IT70–5 were supplied by Interchim. According to their certificate
of analysis, the silica particles have an average particle diameter of 5 μm and an
average pore diameter of 70 Å. The diameter of these particles is situated in the
coarse particle mode. These particles have well-defined properties already de-
2
K1
tailed in a previous work [56]. Their specific surface area is about 500 m g
.
If this value is higher than those reported for natural mineral particles (Saharan
2
K1
sand and Gobi dust BET surface areas are typically around 5 and 10 m g
respectively [64]), it allows us to compare this work with previous reported
data, particularly by our group, on heterogeneous reactions of PAHs adsorbed
on different types of particles [17, 18, 57].

1
154
M. Cazaunau et al.
2.3 Particle preparation and storage
The particles were first cleaned by ultrasonication (3 times) in dichloromethane
(Acros Organics, for residue and pesticides analysis) and allowed to dry at room
temperature. Benzo(a)pyrene was individually coated on silica particles using
liquid–solid adsorption. The coating procedure has already been described in a
previous work [51, 57]. Particulate BaP concentrations were always about
K1
5
00 μg g . The surface coverage was calculated, to be lower than 1 % of a
monomolecular layer. Note that this does not exclude a partial formation of
multilayers of PAHs on the surface.
In order to prevent any photodegradation processes, already observed in the
case of silica particles [10], all the prepared particles were stored at room temper-
ature, in the dark and in amber glass flasks.
2.4 Benzo(a)pyrene
2.4.1. Quantification
Particulate BaP concentration measurements were performed using internal
standard quantification and furthermore, internal standard was quantified by a
syringe standard. Such double quantification allowed the calculation of internal
standard recovery yields and hence to check that internal standard, and therefore
native BaP, was not lost during the analytical procedure. Perdeuterated
benzo(a)pyrene and benzo(b)fluoranthene were respectively used as internal and
syringe standards and were supplied by Cambridge Isotope Laboratories. Quanti-
fications were performed using a calibration solution, prepared with native BaP
and perdeuterated standards (BaP d12 and BbF d12). This solution, injected
before, between and after each analysis, allowed us to calculate the BaP and the
standard response factors and therefore to check and monitor the conditions of
the chromatographic and detection systems.
2.4.2 Analytical procedure
After the reaction (and before the extraction), the samples and the internal stand-
ard were introduced in a “matras” (a pyrex test tube by Prolabo), the internal
standard quantity being controlled by gravimetry. BaP was extracted using fo-
cused microwave assisted extraction (Microdigest 301, Prolabo) in 45 mL of
dichloromethane at 30W during 15 min. The complete extraction procedure was
validated and detailed previously [43]. The extracts were filtered on glasscoton
(VWR) previously cleaned by three successive ultrasonications in dichlorometh-
ane. Dichloromethane was evaporated and change in iso-octane (2,2,4-trimethyl
pentane, HPLC grade, Scharlau) using a vacuum evaporation system (Rapidvap,
Labconco), which uses vortex motion, vacuum and heat (70%, 900 mbars and
50°C during 30 min). Finally, syringe standard was added to the samples just
before the GC-MS analysis.

Atmospheric Heterogeneous Reactions
1155
2.4.3 GC-MS analysis
Sample analyses were performed by gas chromatography coupled to mass spec-
trometry (GC-MS), using a system coupling an HP model series 6890 Gas Chro-
matograph with an HP model 5973 selective detector (Agilent Technologies).
The column was a 30m!0.25mm i.d.!0.25 μm film thickness (HP 5MS, Hew-
lett Packard). 1μL of each sample was introduced via splitless mode injection.
The PTV injector temperature was 270°C, and the oven temperature was held at
K1
7
0 °C for 2 min, then programmed from 70 to 300°C at a rate of 10°C min
and held at 300°C for 10 min. Helium (He 99.9999+% purity, Linde Gas) was
K1
used as carrier gas at a constant flow of 1 mL min . The interface temperature
was kept at 290°C for analysis. Ionisation was achieved by electron impact
(70 eV) and mass detection was performed in selected ion monitoring mode. The
ion molecular masses were chosen to detect BaP and perdeuterated standards.
2.5 Reaction systems
2.5.1 Description of experimental set-ups
All experimental set ups used in this work have already been described in details
elsewhere [15, 17, 18, 45, 48, 51, 56, 57]. Only the main features are described
below for each oxidant:
Ozone
The ozonolysis set-up consists of a quasi-static reaction cell where the particu-
late-BaP, previously deposited on a quartz fiber filter, is placed during different
ozone exposure time. Ozone is generated using a low pressure mercury lamp,
photolysing an Oxygen (99.999+% purity, Linde Gas).Nitrogen (99.999+% pu-
rity, Linde Gas) mixture. The ozone concentration is finally measured using a
home-made UV spectrophotometer. With this set-up, ozone concentration may
typically vary from 7!1012 to 2.4!10 molecule cm . Note that in order to
avoid any interference due to the ambient light, ozone reaction has been investi-
gated in a total darkness.
13
K3
Nitrogen Dioxide
The BaP-coated silica particles are placed using a filter in a fast flow reactor
(
(
2.4 cm inside diameter.48 cm long Pyrex tube). A mixture of nitrogen dioxide
5% NO in Helium 99.995% purity, Alphagaz) diluted in Helium (99.9999 %
2
purity, Alphagaz) is introduced in the reactor via a movable injector (1 cm inside
diameter, 60 cm long quartz tube) sliding into the reactor. Total pressure was
K1
held to (1.8±0.1) Torr with a total volumetric flow rate of 1 L min . NO2
concentration was controlled by Tylan FC2900V mass flow controller and ranged
from 2.0!1013 to 5!10 molecule cm
13
K3
.

1
156
M. Cazaunau et al.
Hydroxyl radical
Experiments were performed in the fast flow reactor quoted above, by monitor-
ing the OH concentration using laser-induced fluorescence (LIF). OH radicals
were generated by the fast (H + NO / OH + NO) reaction (Sander et al.,
2
2
1
2
003). H atoms were produced by hydrogen impurities from Helium (Alphagaz
, Air liquide) passing through a microwave discharge (Raitek LRE 300,
450 MHz). OH radicals were directly introduced into the main reactor via the
movable injector coated with halocarbon wax (Halocarbon Product) in order to
prevent most of the loss of OH radicals on the walls before arriving in the
reactor. NO was in a large excess in order to ensure a complete stoichiometric
2
conversion of H atoms. Total pressure was held to (1.5±0.1) Torr with a total
K1
volumetric flow rate of 0.775 L min
.
OH initial radical concentration was calibrated using two different methods.
The first method was consisting in slowly decreasing the NO concentration until
2
the extinction of the LIF signal, allowing to extrapolate a minimal concentration
12
K3
of NO to generate OH radicals of 1.4!10 molecule cm . The second method
2
was consisting in adding well-known concentrations of H (5% H in Helium
2
2
9
9.995% purity, Air Liquide) and waiting for the LIF signal to be multiplied by
a factor of 2, allowing to extrapolate a value for initial OH concentration of
.5!1012 molecule cm . Finally, the loss of OH radicals on the filter + silica
K3
1
particles system alone was evaluated from the LIF signal to be around 60%,
allowing to conclude to the value of 5.8!1011 molecule cm for the initial OH
radical concentration in our experiments.
K3
Considering all the uncertainties on the calibration method detailed above, it
was decided that only qualitative information will be proposed for the hydroxyl
radical concentration in this work.
2.5.2 Kinetic conditions
For all oxidants, a complete degradation of particulate-bound BaP was observed
after a certain exposure time. For each experiment, the kinetic information was
obtained by monitoring the remaining concentration of benzo(a)pyrene versus
exposure time to O , OH and.or NO . Specific details are provided below.
3
2
Ozone
Ozone concentration was measured all along the reaction time and no variation
was observed under our experimental conditions. Gaseous O residence time in
3
the reactor (about 1 s) was negligible compared to the BaP lifetime.
Nitrogen dioxide
Assuming that all losses on the injector and reactor walls were negligible, NO2
partial pressure before reaching the sample was equal to the value set by the
mass flow controller (taking into account the dilution in Helium flow). In the

Atmospheric Heterogeneous Reactions
1157
presence of water adsorbed on silica particles, NO may be converted to HONO
2
and HNO [22]. However, under our experimental conditions (dried silica parti-
3
cles and gaseous environment), NO loss on surfaces was negligible compared
2
to the amount of NO available for the reactions [50, 56].
2
Hydroxyl radical
OH radicals were produced upstream from the reactor. Loss on injector walls
was negligible [11] but loss on filter and silica (with no BaP) was measured to
be approximately equal to 60 % of the total initial OH radical produced by the
microwave discharge, as already detailed above. Online OH measurement during
the reaction time did not show any fluorescence variation on the detector keeping
the same injector position, meaning that the OH concentration was kept station-
ary all along the reaction.
For each oxidant, secondary reactions implying oxidation products were first
considered as negligible and finally verified a posteriori. NO radicals formed
simultaneously with OH do not react with PAHs [17] and do not further interfere
in the mechanism as their concentration (equal to that of OH) is always too
small compared to that of NO . Experiments were performed for three different
2
concentration of oxidant (except for the hydroxyl radical) in order to verify the
elementary nature of the reaction. Pseudo-first order rate constants derived from
the three experimental data sets were proportional to the oxidant concentration
demonstrating that the reaction of BaP with NO or with O was the predominant
2
3
reaction and the limiting step (see part 3).
2.6 Data analysis
2.6.1 Pseudo-first order rate constant measurement
Concentrations vs. time were normalised i.e. the BaP concentration measured in
K1
μg g of particles after an exposure time t was divided by the corresponding
initial particulate-BaP concentration (measured at t = 0), leading to a normalised
particulate concentration of compound in %. Experimental data points were fitted
by a simple mono-exponential function allowing to determine pseudo-first-order
I’
rate constants (k ) for each oxidant concentration value, using non linear least
square fitting. All experiments were repeated at least three times and for three
different oxidant concentration values (in the case of O and NO only), and
3
2
were very reproducible, resulting in data analyses with a high consistency. Note
that all experimental data obtained (excepted in the case of OH) were finally
presented on a single graph by recalculating the reaction times: concentration of
oxidants (included in the pseudo-first order rate constant values) and reaction
times are interchangeable in first order kinetic laws, allowing to have a maximal
number of data points available to fit the results.
Similarly, experimental data points relative to the NO oxidation product
2
formation were fitted using first-order exponential growths, demonstrating that

1
158
M. Cazaunau et al.
decays corresponding to BaP reaction with NO and growths describing the for-
2
mation of oxidation products present the same global reaction rate (within uncer-
tainties).
2.6.2 Second order rate constants
For gas-phase reactions, second order rate constants are often determined by
dividing pseudo-first order rate constants by the reactant concentration. In most
heterogeneous reactivity studies [23, 24], uptake coefficients are usually ex-
tracted from kinetic measurements, considering the fate of the oxidant vs. reac-
tion time. In some cases, authors considered that the reactions occurred between
two adsorbed species, according to a Langmuir–Hinshelwood mechanism [59].
Second-order surface reaction rate constants are therefore calculated in square
centimeters per second by dividing the pseudo-first order rate constants by the
concentration of ozone adsorbed on free active site on the surface. The active
site surface concentrations for the silica particles used in this study being not
known; such rate constants were not calculated. In other studies, [5, 32, 66],
second order rate constants were calculated similarly to gas-phase reactions con-
stants, suggesting that reactions occurred between a gaseous oxidant and a sur-
I’
face-bounded PAH molecule. Since values of k were demonstrated to be pro-
portional to gaseous oxidant concentration under our experimental conditions
(
and in these concentration ranges), second order rate constants were calculated
considering such gaseous oxidant concentrations [17, 18, 51, 55, 56].
2.6.3 Uncertainties determination
Errors on internal and syringe standards weighting, on silica particles weighting,
on chromatographic peak integration and on GC-MS measurements were consid-
ered as random errors. Total random error was estimated by the standard devia-
tion on the exponential coefficients obtained from the nonlinear least square fit
of the experimental points, and was varying between 5% and 10%. Systematic
errors (oxidant concentration, pressure, temperature, and calibration of mass flow
controllers) were also taken into account. Finally, global uncertainties (random
+
systematic) on the second order rate constant values were evaluated to be about
25%.
3. Results and discussion
3.1 Analytical procedure validation
Analytical procedure validation was already reported by our group for different
types of particles (graphite, silica, diesel exhaust and urban dust…) [43, 57]. In
the present work, silica particles were doped with a solution of well-known
concentration of BaP. Benzo(a)pyrene was then extracted from coated particles
and analysed. Resulting recovery yields for BaP and perdeuterated BaP were

Atmospheric Heterogeneous Reactions
1159
Fig. 1. Normalised concentration of BaP adsorbed on silica particles as a function of ozone
1
3
K3
exposure time, represented for [O
3
] = 1.4!10 molecule cm
.
respectively of (96±2) % (N = 3) and (88±2) % (N = 3) allowing to validate the
analytical protocol.
3.2 Desorption
Prior to the study of BaP oxidation reactions, it was first necessary to verify that
no potential desorption of BaP from the silica particles was occurring during the
reaction time. BaP-coated-silica was exposed under the same conditions as those
used for studying oxidation reactions (total flow, complete darkness) but without
any oxidant. As no loss of BaP was observed under these experimental condi-
tions, desorption of BaP was considered as negligible and degradation of BaP
was fully attributed to the only reactions with the different oxidants.
3.3 Reactions of BaP with ozone
All experimental data for this reaction were plotted on a same single graph,
presented in Figure 1, for a reference ozone concentration of [O ]
=
3
ref
.4!1013 molecule cm . Initial BaP concentration was (353±65) μg g (n =
K3
K1
1
2
4).
Reaction of ozone with BaP adsorbed on silica particles was almost complete
after 3 min under our experimental conditions. 12 replicates for each reaction
I’
time were available for the fit to derive the pseudo-first order rate constant (k ).
The kinetics of BaP degradation was also investigated as a function of O con-
3
centration. In Figure 2 are presented pseudo-first order rate constant values as a
function of O concentration fitted by a straight line. This Figure allows to verify
3
that the reactivity of BaP was proportional to the oxidant concentration under

1
160
M. Cazaunau et al.
Fig. 2. Pseudo-first order rate constants for the reaction of silica-bound-BaP with ozone as a
function of ozone concentration and corresponding linear regressions. Error bars represent 2
standard deviations on the exponential coefficient obtained from the non-linear least square
fitting of experimental points.
our experimental conditions, demonstrating the elementary nature of the reaction.
According to line slopes, second order rate constants were determined for the
K15
3
K1 K1
reactions of O to be (2.1±0.5)!10
cm molecule
s
, at room tempera-
3
ture.
The first order dependence of the BaP decay observed in this study was
already reported for other PAH or for different particulate support [5, 51, 57,
66]. On the contrary, Pöschl et al. [59] and Kwamena et al. [39, 40] observed a
Langmuir-Hinshelwood behaviour and considered a reaction mechanism where
ozone is first adsorbed to the surface prior to reacting with BaP. They specified
that there is a maximum for pseudo-first order constant values when all active
sites of the support are saturated by oxidant molecules. Observation of this mech-
anism was made for laboratory-soot particles [59] and for azelaic acid aerosols
[39, 40]. Up to now, no Langmuir-Hinshelwood behaviour was observed for
silica gel [5] and silica plates [66], experiments being all performed in the same
range of ozone concentration (2.5!1012 to 8!10 molecule cm ). This ques-
14
K3
tion is therefore still under consideration.
Another parameter largely influencing the heterogeneous reactivity of BaP is
the particle loading. A previous study by our group [57] of the reactivity of BaP
adsorbed on silica but in a mixture with 12 other PAH-congeners (with similar
particulate concentrations) and exposed to ozone led to a second order rate con-
stant of (1.4±0.3)!10K16 cm molecule
3
K1 K1
s
at room temperature. Experi-
ments were performed using the same silica particles, the same particulate-BaP
loading and the same experimental set-up. Even accounting for the presence of
other PAHs on the silica to react with ozone (their reactivity being measured

Atmospheric Heterogeneous Reactions
1161
Fig. 3. Normalised concentration of BaP adsorbed on silica particles as a function of NO
2
1
3
K3
exposure time, represented for [NO
2
] = 3.5!10 molecule cm
.
simultaneously) and therefore decreasing the amount of ozone available to only
react with BaP, this does not allow to explain the factor of 15 existing between
the rate constant value previously reported by Perraudin et al. [57] and that
measured in this work. It is clear that the nature (specific surface area, composi-
tion,…) of the coating particle strongly influences heterogeneous reaction mecha-
nisms and therefore the rate of BaP degradation. The influence of the close
environment of the PAH on the particle was already reported and discussed [5,
10, 17, 18] but is still difficult to interpret as the total amount of PAH in almost
all reported works (including this one) corresponds to a total surface coverage
less than 10% of a monomolecular layer. The influence of the BaP particle
loading will be carefully carried out in a future study, focusing particularly on
the possible presence of PAH aggregates and on the different adsorption sites to
be considered on the silica surface.
3.4 Reactions of BaP with NO2
A similar procedure was applied for the reaction of nitrogen dioxide with BaP
adsorbed on silica particles. BaP concentration decay when reacting with NO is
2
presented in Figure 3, for a reference nitrogen dioxide concentration of
.5!1013 molecule cm . Initial BaP concentration was (479±31) μg g (n =
K3
K1
3
2
1). Reaction of nitrogen dioxide with BaP adsorbed on silica particles was
observed to be complete after 3–4 min. Influence of nitrogen dioxide concentra-
tion was also investigated using three different concentrations and the elementary
nature of the reaction was again demonstrated. Results are presented in Figure
4
. According to line slopes, second order rate constant was determined for the
K16
3
K1 K1
reaction to be (5.8±1.4)!10
cm molecule
s
, at room temperature.

1
162
M. Cazaunau et al.
Fig. 4. Pseudo-first order rate constants for the reaction of silica-bound-BaP with nitrogen
dioxide as a function of NO concentration and corresponding linear regression. Error bars
2
represent 2 standard deviations on the exponential coefficient obtained from the non-linear least
square fit of experimental points.
As observed above for ozone, our group also previously reported a different
(
1.6 times higher) rate constant for the NO reaction [56]. Again, BaP was ad-
2
sorbed in that work on particles of different types among a mixture of twelve
other PAHs. Such a difference is still not very clear or easy to explain, despite
the very large number of experiments in both works and the very good reproduci-
bility and repeatability of the results. Note that Miet et al. [51] also recently
showed that the reactivity of pyrene (a 4-rings PAH) is 4 times larger when
adsorbed alone on particles (as a model) than when being present in a complex
mixture.
Different uptake studies previously made on thin films of PAHs did not show
any reaction when exposed to NO but the same reaction appears efficient with
2
PAHs adsorbed on n-hexane soot [24, 39]. The amount of PAHs adsorbed on
partially or completely covered particle surface, strongly influences the reaction
rates. As explained above in the ozone reaction section, considering a high PAH
concentration and according to the specific surface area of the particle, PAH
may form aggregates and become less accessible to the oxidants. On the contrary,
PAHs with low coating concentration are mono-dispersed on the surface and will
more accessible to the oxidant, as previously suggested by Kamens et al. [31]
3.5 Reactions of BaP with OH
A decay of particle-bound BaP concentration versus hydroxyl radical (in the
presence of NO ) exposure time is presented in Figure 5. Initial BaP concentra-
2
K1
tion was (476±62) μg g (n = 15). The reaction was complete after 1.5 min.
Since it was not possible to accurately calibrate OH concentration in this work

Atmospheric Heterogeneous Reactions
1163
Fig. 5. Decay of normalised concentration of silica-bound BaP vs. exposure time to hydroxyl
radicals in the presence of nitrogen dioxide.
and therefore propose a second order rate constant for this reaction, only
I
a pseudo-first order rate constant was derived from the fit, leading k(
=
BaP + OH)
K2 K1
(3.4±0.8)!10
s
. Such a value was obtained after the subtraction of the NO2
contribution to the total BaP decay.
Note that the only estimation of the OH initial concentration proposed in this
work allows us to conclude that the heterogeneous reaction of OH with BaP
II
(
K14
3
K1 K1
(k
≈ 5.9!10
cm molecule
s
) is much faster than those of
BaP + OH)
ozone and NO but slower than the usual corresponding reaction of OH with
2
(lighter) PAHs occurring in the gas phase, meaning that heterogeneous processes
of BaP with OH are negligible under atmospheric conditions (considering an
6
K3
averaged atmospheric OH concentration of ≈ 1!10 molecule cm ).
.6 Oxidation products
3
In this work, degradation products were investigated by GC-MS only for the
reactions of silica-bound BaP with NO and OH. Reaction products arising from
2
ozone initiated oxidation of BaP will be presented in another article.
3.6.1 Reaction with nitrogen dioxide
In order to identify the largest number of products, mass detection was first
performed in the scan mode (mass.charge ratio ranging from 50 to 550). Pulsed
3
K1
splitless mode injection (purge flow = 1 cm s , purge time = 60 s, 25 psi pulse
during 60 s) was added to the previous chromatographic parameters in order to
enhance the analyte transfer in the head of the analytical capillary column.
The chromatograms presented only three news peaks after the reaction of
BaP with NO (Figure 6). A comparison of the corresponding mass spectra with
2

1
164
M. Cazaunau et al.
Fig. 6. Chromatogram of products from the reaction of silica-bound BaP with nitrogen dioxide.
those reported in the mass spectra library (NIST 98, HP Mass Spectral Library,
1998) allowed the identification of these compounds as three nitrobenzo(a)py-
renes (NBaP). 6-nitrobenzo(a)pyrene was identified by its retention time inject-
ing a standard solution. 1- and 3-nitrobenzo(a)pyrenes were only identified by
comparison with chromatograms already reported in the literature [30, 38]. Note
that at longer exposure times than those allowing the total degradation of BaP,
it was observed the further formation of dinitrobenzo(a)pyrenes (diNBaP), paral-
lely to the decrease of the concentrations of 1-, 3- and 6-nitrobenzo(a)pyrenes.
At least two compounds are expected to be formed by this way: 1,6- and 3,6-
dinitrobenzo(a)pyrenes. Unfortunately, their identification was not made possible
in this work.
Ishii et al. [30] have previously shown the formation of all these compounds
studying the reaction of BaP-coated glass fiber filter exposed to 10 ppm of nitro-
gen dioxide. They estimated that 70 % of total initial BaP concentration were
converted in mononitroBaP. It was also shown in that work that amounts of
dinitroBaP increased gradually and then decreased slowly.
In the present study, GC-MS single ion monitoring mode was applied for
the product quantification. NBaP were quantified by external calibration: three
solutions of 6-nitrobenzo(a)pyrene were prepared with quantities ranging from
0
.74 to 9.19 ng injected on the column. These solutions were injected before,
between and after each analysis. The major product for the NO reaction was
2
the 6-NBaP isomer with smaller yields of the 1- and 3- isomers. Experimental
data shows that 6-NBaP accounts for (83±6) % of all nitrated BaP formed, 1-
NBaP for (5±2) % and 3-NBaP for (12±5) % (n = 25). The distribution of
isomeric mononitro-BaP derivatives generally followed the electron densities cal-
culated for various positions in the molecule. Carbon balance for investigated
reaction was therefore complete within uncertainties, confirming that the oxida-
tion products identified in this study are the only products formed in the system.
Note that these results are in very good agreement with those reported previously
by Ishii et al. [30].

Atmospheric Heterogeneous Reactions
1165
Fig. 7. Normalised concentrations of 6-(square), 1-(circle) and 3-(triangle)NBaP for different
1
3
K3
2
exposure time to nitrogen dioxide (with [NO ] = 3.5!10 molecule cm ). Error bars repre-
sent 2 standard deviations corresponding to the statistical uncertainty evaluated from 5 experi-
mental replicates.
The product quantification also allowed us to determine the product forma-
tion rates. Indeed, normalised product concentrations within NO exposure time
2
were fitted using first order exponential functions (Figure 7). Note that it was
logically assumed that nitro- and dinitroBaP were not desorbing along the reac-
tion time, as they present lower vapor pressures than BaP. Second order rate
constant values for product formations were determined to be
K16
3
K1 K1
K17
3
K1 K1
3
.7!10
cm molecule
s
for 6-NBaP, 2.2!10
cm molecule
for 3-NBaP, leading to a global
for all mononitrated BaP formation.
s
K17
3
K1 K1
for 1-NBaP and 5.3!10
value of 4.45!10
cm molecule
s
K16
3
K1 K1
cm molecule
s
This value is still reasonable considering all the uncertainties on the four different
compounds to be considered. After 3–4 min of reaction, NBaP concentrations
start to decrease and a small amount of diNBaP was detected. Second order
rate constants derived for the NO degradation of NBaP were estimated to be
2
K17
3
K1 K1
K17
3
K1 K1
1
.8!10
for 1-NBaP and 1.9!10
being considered to be similar within uncertainties.
cm molecule
s
for 6-NBaP, 2.2!10
cm molecule
s
K17
3
K1 K1
cm molecule
s
for 3-NBaP, all these values
A possible mechanism explaining the formation of the three NBaP could be
the first addition of NO to the most electrophilic carbon atoms (positions 6, 3
2
and 1 for benzo(a)pyrene [58]), followed by a second addition of NO in the
2
alpha position to finally lead to a elimination of HONO occurring by the abstrac-
tion of a hydrogen atom from the first carbon (Figure 8). Such a mechanism has

1
166
M. Cazaunau et al.
Fig. 8. Mechanism proposed for the heterogeneous nitration of benzo(a)pyrene.
already been recently proposed by our group in the similar case of pyrene reac-
tion [51].
3.6.2 Reaction with hydroxyl radical
Product identification was also performed by GC-MS in the scan mode using the
same chromatographic parameters as described above.
Due to the method of OH production in the reactor (a large excess of NO2
reacting with H atoms), nitrogen dioxide was the major reactive species in the
reaction system. Therefore, all the three nitrated-BaP were again identified and
quantified. But this time, the total yield of degraded BaP was estimated to be

Atmospheric Heterogeneous Reactions
1167
not more than 20–25 %. As in the case of the reaction with nitrogen dioxide
only, 6-NBaP relatively accounted for approximately 80 % of total nitrated spe-
cies formed, 1-NBaP for 5% and 3-NBaP for 15 %. Consequently, it is likely
that the nitro-BaP formation was following the same mechanism under our ex-
perimental condition. Another mechanism for the formation of nitrated com-
pounds has been reported for the PAH in the gas phase in the presence of nitro-
gen dioxide [9, 61]. It first involves the addition of OH, then NO is adding in
2
the alpha position to finally lead to a departure of H O occurring by the abstrac-
2
tion of a hydrogen atom from position 1.
In this work, derivation of samples with BSTFA (bis(trimethylsilyl)-trifluoro-
acetamide) (25μl, 30 min at 65°C) allowed to also detect hydroxylated com-
pounds. Samples were injected before and after derivation. Monohydroxylated
(m.z = 268) and a small amount of three different dihydroxylated (m.z = 284)
compounds were detected. Unfortunately, comparison of the mass spectra ob-
tained with those reported from the NIST mass spectra library did not clearly
allow us to better identify these compounds.
4. Conclusions
In this work, rate constants were measured for the reactions of O , NO and OH
3
2
with benzo(a)pyrene adsorbed on silica particles, in the darkness and at room
temperature. Second order rate constants were measured to be
K15
K16
3
K1
K1
(
(
2.1±0.5)!10
5.8±1.4)!10
cm molecule
s
for the reaction with O3 and
cm molecule sK1 for the reaction with NO . Only pseudo-
3
K1
2
I
first order rate constant could be obtained for the reaction of BaP with OH: k =
K2 K1
(
3.4±0.8)!10
s
. Degradation products were investigated in the case of the
reactions with NO and OH. 1-, 3- and 6-nitrobenzo(a)pyrenes were found to be
2
the major oxidation products from the (BaP + NO ) reaction. The global kinetics
2
of product formation has been demonstrated to be similar to that measured for
the BaP degradation, within uncertainties. Dinitrobenzo(a)pyrenes were also
identified when no more BaP was detected but it was not possible for them to
be quantified. Results obtained for nitro-BaP are in very good agreement with
those previously obtained by Ishii et al. [30]. Product identification was also
performed for the reaction of BaP with OH in the presence of a large excess of
NO . 1-, 3-, and 6-nitroBaP were again detected and it appears that their forma-
2
tion was following the same mechanism for both NO and OH (in the presence
2
of NO ) reactions under our experimental conditions, but with a lower efficiency.
2
Mono- and di-hydroxylated compounds were also detected but not identified in
this work.
Further work will be necessary to better calibrate the initial OH concentration
in our experiments, to allow to propose second order rate constants useful to
extrapolate atmospheric lifetimes but also to give rise to evaluate the formation
of HONO and.or H O.
2

1
168
M. Cazaunau et al.
Heterogeneous mechanisms involved for all reactions were not clearly under-
stood in this work but again Langmuir-Hinshelwood mechanism could not be
clearly retained from the observations described in this study. Such mechanisms
are often debated and it is clear that the particulate substrate is one of the most
important parameter. Particle loading and adsorption state of adsorbed-PAHs
(mono-dispersed, aggregates, amorphous or crystalline) are essential parameters
to focus on in the next future to better understand heterogeneous reactivity.
Finally, the nitrated products identified in this work are very important, as
being recognized to be very toxic and particularly mutagenic [30, 46, 47]. Identi-
fication of all other BaP oxidation products will be carried on performing new
analytical developments as it is of major importance to improve the evaluation
of their impact on human health.
Acknowledgement
The authors wish to thank the French National Research Agency and the Aqui-
taine Region for financial support.
References
1
2
3
4
. C. Adelhelm, R. Niessner, U. Pöschl, T. Letzel, Analytical and Bioanalytical Chemis-
try 391 (2008) 2599.
. Albinet, E. Leoz-Garziandia, H. Budzinski, E. Villenave, Sci. Total Environ 384
(2007) 280.
. Albinet, E. Leoz-Garziandia, H. Budzinski, E. Villenave, J. L. Jaffrezo, Atmos.
Environ 42 (2008a) 55.
. Albinet, E. Leoz-Garziandia, H. Budzinski, E. Villenave, J. L. Jaffrezo, Atmos.
Environ 42 (2008b) 43.
5
6
7
. Alebic-Juretic, T. Cvitas, L. Klasinc, Environ Sci. Technol. 24 (1990) 62.
. Alebic-Juretic, T. Cvitas, L. Klasinc, Chemosphere 41 (2000) 667.
. J. O. Allen, N. M. Dookeran, K. Taghizadeh, A. L. Lafleur, K. A. Smith, A. F.
Sarofim, Environ. Sci. Technol. 31 (1997) 2064.
8
. M. Ammann, M. Kalberer, D. T. Jost, L. Tobler, E. Rössler, D. Piguet, H. W.
Gäggeler, U. Baltensperger, Nature 395 (1998) 157.
9
. R. Atkinson, J. Arey, B. Zielinska, S. M. Aschmann, Environ. Sci. Technol. 21
(1987) 1014.
1
0. T. D. Behymer, R. A. Hites, Environ. Sci. Technol. 19 (1985) 1004.
1. A. K. Bertram, A. V. Ivanov, M. Hunter, L. T. Molina, M. J. Molina, J. Phys. Chem.
A. 105 (2001) 9415.
1
1
2. G. L. Borosky, K. K. Laali, Organic and Biomolecular Chemistry 5 (2007) 2234.
3. W. F. Busby Jr, H. Smith, C. L. Crespi, B. W. Penman, Mutation Research - Genetic
Toxicology 342 (1995) 9.
1
1
4. M. A. A. Clyne, P. B. Monkhouse, Journal of the Chemical Society, Faraday Trans-
actions 2: Molecular and Chemical Physics 73 (1977) 298.
15. W. Estève, Thèse de Doctorat de l'Université de Bordeaux 1 (2002).
6. W. Estève, H. Budzinski, E. Villenave, Polycyclic Aromatic Compounds 23 (2003)
1
4
41.
1
7. W. Estève, H. Budzinski, E. Villenave, Atmos. Environ 38 (2004) 6063.
8. W. Estève, H. Budzinski, E. Villenave, Atmos. Environ 40 (2006) 201.
1

Atmospheric Heterogeneous Reactions
1169
1
9. B. J. Finlayson-Pitts, P. J. N. Jr., Atmospheric chemistry, Fundamentals and experi-
mental techniques. Wiley Interscience Editions, New York (1986).
0. B. J. Finlayson-Pitts, J. N. Pitts Jr., Science 276 (1997) 1045.
2
2
1. B. J. Finlayson-Pitts, L. M. Wingen, A. L. Sumner, D. Syomin, K. A. Ramazan,
Physical Chemistry Chemical Physics 5 (2003) 223.
2
2. A. L. Goodman, G. M. Underwood, V. H. Grassian, J. Phys. Chem. A. 103 (1999)
7
217.
23. V. H. Grassian, J. Phys. Chem. A. 106 (2002) 860.
24. S. Gross, A. K. Bertram, J. Phys. Chem. A. 112 (2008) 3104.
25. A. l. Guilloteau, M. L. Nguyen, Y. Bedjanian, G. Le Bras, J. Phys. Chem. A. 112
(2008) 10552.
2
6. M. K. D. Hailwood, E. Leotz-Garzianda, R. Maynard, E. Menichi, S. Moorcroft, J.
Pacyna, P. Perez Ballesta, J. Schneider, R. Westerholm, M. Wichmann-Fiebig, M.
Woodfield, L. Van Bree, C. Conolly, Office for Official Publications of the European
Communities (2001).
2
2
2
3
7. K. Horikawa, N. Sera, K. Murakami, N. Sano, K. Izumi, H. Tokiwa, Toxicology
Letters 98 (1998) 51.
8. IARC. (2008). "Complete List of Agents evaluated and their classification: Agents
rewiewed by the IARC monographs."
9. Retrieved http:..monographs.iarc.fr.ENG.Classification.Listagentsalphorder.pdf
consulted in February 2009.
0. S. Ishii, Y. Hisamatsu, K. Inazu, T. Kobayashi, K. I. Aika, Chemosphere 41 (2000)
1
809.
3
3
3
3
3
3
3
3
3
4
1. R. M. Kamens, Environ. Sci. Technol. 22 (1988) 103.
2. R. M. Kamens, J. M. Perry, D. A. Saucy, Env. Int. 11 (1985) 131.
3. L. H. Keith, W. A. Telliard, Environ. Sci. Technol. 13 (1979) 416.
4. U. Kirchner, V. Scheer, R. Vogt, J. Phys. Chem. A. 104 (2000) 8908.
5. J. Kleffmann, P. Wiesen, Atm. Chem. Phys. 5 (2005) 77.
6. R. Koeber, J. M. Bayona, R. Niessner, Environ. Sci. Technol. 33 (1999) 1552.
7. R. Koeber, R. Niessner, J. Aerosol Sci. 28 (1997).
8. R. L. Kristovich, P. K. Dutta, Environ. Sci. Technol. 39 (2005) 6971.
9. N. O. A. Kwamena, J. P. D. Abbatt, Atmos. Environ. 42 (2008) 8309.
0. N. O. A. Kwamena, M. G. Staikova, D. J. Donaldson, I. J. George, J. P. D. Abbatt,
J. Phys. Chem. A. 111 (2007) 11050.
4
4
1. N. O. A. Kwamena, J. A. Thornton, J. P. D. Abbatt, J. Phys. Chem. A. 108 (2004)
1
1626.
2. S. Lelievre, Y. Bedjanian, N. Pouvesle, J. L. Delfau, C. Vovelle, G. Le Bras, Physical
Chemistry Chemical Physics 6 (2004) 1181.
4
4
4
4
4
4
4
5
5
3. M. Letellier. Thèse de Doctorat de l'Université de Bordeaux 1 (1998).
4. T. Letzel, E. Rosenberg, U. Pöschl, R. Niessner, J. Aerosol Sci. 30 (1999) S605.
5. L. Ley. Thèse de Doctorat de l'Université de Bordeaux 1 (1995).
6. V. Librando, A. Alparone. J. Hazard, Mater. 161 (2009) 1338.
7. V. Librando, A. Alparone, G. Tomaselli, J. Mol. Mod. 14 (2008) 489.
8. J. C. Loison, L. Ley, R. Lesclaux, Chem. Phys. Lett. 296 (1998) 350.
9. Messerer, R. Niessner, U. Pöschl, Carbon 44 (2006) 307.
0. K. Miet. Thèse de Doctorat de l'Université de Bordeaux 1 (2008).
1. K. Miet, K. Le Menach, P. M. Flaud, H. Budzinski, E. Villenave, Atmos. Environ.
4
3 (2009) 837.
5
2. Naspinski, R. Lingenfelter, L. Cizmas, Z. Naufal, L. Y. He, A. Islamzadeh, Z. Li,
Z. Li, T. McDonald, K. C. Donnelly, Env. Int. 34 (2008) 988.
5
5
5
3. M. L. Nguyen. Thèse de Doctorat de l'Université d'Orléans (2008).
4. Papastefanou, Appl. Radiat. Isot. 64 (2006) 93.
5. Perraudin, H. Budzinski, E. Villenave, Analytical and Bioanalytical Chemistry 383
(2005a) 122.

1
170
M. Cazaunau et al.
5
5
5
5
6
6
6. Perraudin, H. Budzinski, E. Villenave, Atmos. Environ. 39 (2005b) 6557.
7. Perraudin, H. Budzinski, E. Villenave, J. Atm. Chem. 56 (2007) 57.
8. J. N. Pitts Jr., Environmental Health Perspectives 47 (1983) 115.
9. U. Pöschl, T. Letzel, C. Schauer, R. Niessner, J. Phys. Chem. A. 105 (2001) 4029.
0. M. S. Salgado, M. J. Rossi, Int. J. Chem. Kinet. 34 (2002) 620.
1. J. Sasaki, S. M. Aschmann, E. S. C. Kwok, R. Atkinson, J. Arey, Environ. Sci.
Technol. 31 (1997) 3173.
6
2. US-EPA. EPA 540.1–86-013 Cincinnati OH:Environmental Criteria and Assessment
Office (1984).
63. C. R. Usher, A. E. Michel, V. H. Grassian, Chemical Reviews. 103 (2003) 4883.
4. G. M. Underwood, P. Li, H. Al-Abadleh, V. H. Grassian, J. Phys. Chem. A. 105
6
(2001) 6609.
6
5. K. A. Van Cauwenberghe, L. Van Vaeck, J. N. J. Pitts, Forensic Environmental
Applications. (1984) 1499.
6
6. C. H. Wu, I. Salmeen, H. Niki. Environ, Sci. Technol. 18 (1984) 603.