UV absorption spectrum of the ClO dimer (Cl2O2) between 200 and 420 nm

Article abstract of DOI:10.1021/jp9065345

The UV photolysis of Cl₂O₂ (dichlorine peroxide) is a key step in the catalytic destruction of polar stratospheric ozone. In this study, the gas-phase UV absorption spectrum of Cl₂O₂ was measured using diode arr

Full text of DOI:10.1021/jp9065345

J. Phys. Chem. A 2009, 113, 13711–13726
13711
UV Absorption Spectrum of the ClO Dimer (Cl O ) between 200 and 420 nm
2
2
Dimitrios K. Papanastasiou, Vassileios C. Papadimitriou,†,‡,§ David W. Fahey,†,‡ and
†
,‡
,
†
James B. Burkholder*
Earth System Research Laboratory, Chemical Sciences DiVision, National Oceanic and Atmospheric
Administration, 325 Broadway, Boulder, Colorado 80305-3328, and CooperatiVe Institute for Research in
EnVironmental Sciences, UniVersity of Colorado, Boulder, Colorado 80309
ReceiVed: July 10, 2009; ReVised Manuscript ReceiVed: September 28, 2009
The UV photolysis of Cl
ozone. In this study, the gas-phase UV absorption spectrum of Cl
spectroscopy and absolute cross sections, σ, are reported for the wavelength range 200-420 nm. Pulsed laser
photolysis of Cl O at 248 nm or Cl /Cl O mixtures at 351 nm at low temperature (200-228 K) and high
pressure (∼700 Torr, He) was used to produce ClO radicals and subsequently Cl via the termolecular ClO
self-reaction. The Cl spectrum was obtained from spectra recorded following the completion of the gas-
phase ClO radical chemistry. The spectral analysis used observed isosbestic points at 271, 312.9, and 408.5
nm combined with reaction stoichiometry and chlorine mass balance to determine the Cl spectrum. The
O
2 2
(dichlorine peroxide) is a key step in the catalytic destruction of polar stratospheric
2 2
O
was measured using diode array
2
2
2
2 2
O
2 2
O
2
O
2
+
0.8
-18
2
-1
Cl
where the quoted error limits are 2σ and include estimated systematic errors. The Cl
sections obtained for wavelengths in the range 300-420 nm are in good agreement with the Cl
O
2 2
UV absorption spectrum peaks at 244.5 nm with a cross section of 7.6-0.5 × 10
cm molecule
absorption cross
spectrum
2
O
2
2 2
O
reported previously by Burkholder et al. (J. Phys. Chem. A 1990, 94, 687) and significantly higher than the
values reported by Pope et al. (J. Phys. Chem. A 2007, 111, 4322). A possible explanation for the discrepancy
2 2
in the Cl O cross section values with the Pope et al. study is discussed. Representative, atmospheric photolysis
rate coefficients are calculated and a range of uncertainty estimated based on the determination of σCl O2(λ)
2
2 2
in this work. Although improvements in our fundamental understanding of the photochemistry of Cl O are
still desired, this work indicates that major revisions in current atmospheric chemical mechanisms are not
required to simulate observed polar ozone depletion.
1
. Introduction
Ozone depletion in the Arctic and Antarctic stratosphere
understanding of the relevant gas-phase and heterogeneous
halogen chemistry under the conditions found in the polar
regions.
during winter and spring is well-established from a variety of
satellite, airborne, and ground-based observations. In the
The majority of polar stratospheric ozone loss in current
1
atmospheric model calculations4 is from the “ClO dimer”
Antarctic, ozone depletion is extensive and leads to what is
commonly called an “ozone hole”. It is recognized that halogen
chemistry involving Cl- and Br-containing species is the
principal cause of ozone depletion observed in the polar regions.2
The link between ozone depletion and emission of man-made
halogen compounds, e.g., chlorofluorocarbons (CFCs) and
bromofluorocarbons (halons), led to the Montreal Protocol and
its subsequent amendments and adjustments which regulate
global production and consumption of these ozone depleting
compounds. Polar ozone depletion alters the composition of
Earth’s atmosphere and has long-term implications through its
impact on climate. Atmospheric models constrained by observa-
tions of ozone-depleting compounds predict that the polar and
global ozone abundance levels will not return to pre-1980s
values (i.e., before the Antarctic stratospheric ozone hole) until
later than 2050.3 Accurate models of current and future
atmospheric ozone depletion and recovery require a thorough
catalytic ozone destruction cycle that was first proposed by
Molina and Molina5
ClO + ClO + M f Cl O + M
(1)
(2)
(3)
(4)
2
2
Cl O + hυ f Cl + ClOO
2
2
ClOO + M f Cl + O + M
2
2
× (Cl + O f ClO + O )
3 2
NET 2O + hυ f 3O
3
2
2 2
where Cl O (dichlorine peroxide, ClOOCl) is referred to as
the ClO dimer. A discussion of other catalytic ozone destruction
cycles is given elsewhere. Each of the elementary reactions in
*
Corresponding author, James.B.Burkholder@noaa.gov.
Earth System Research Laboratory, Chemical Sciences Division,
2
†
the ClO dimer cycle has been studied extensively over the last
several decades using a variety of laboratory techniques and
National Oceanic and Atmospheric Administration.
‡
Cooperative Institute for Research in Environmental Sciences, Uni-
versity of Colorado at Boulder.
6
are considered reasonably well established. Atmospheric model
§
Current address: Laboratory of Photochemistry and Chemical Kinetics,
calculations show that in the polar stratosphere the rate of Cl
2 2
O
Department of Chemistry, University of Crete, Vassilika Vouton, 71003,
Heraklion, Crete, Greece.
UV photolysis, reaction 2, is critical in determining the rate
1
0.1021/jp9065345 CCC: $40.75  2009 American Chemical Society
Published on Web 11/06/2009

1
3712 J. Phys. Chem. A, Vol. 113, No. 49, 2009
Papanastasiou et al.
Pope et al.10 study reported Cl
O
absorption cross sections at
2
2
λ >300 nm that are significantly lower than those reported in
all previous studies. With Cl photolysis rates derived from
2
the Pope et al. spectrum, models produce values of ClO, Cl O ,
2
O
2
10
2
and ozone and its loss rate that are in poor agreement with in
situ and remote-sensing observations.11 The Pope et al. study,
which if correct, would, therefore, have significant implications
for the current understanding of halogen chemistry in the polar
stratosphere. Furthermore, if Pope et al.10 were correct, it would
10
9
mean that the earlier Burkholder et al. study conducted at
NOAA was in error outside the stated uncertainty limits.
Therefore, an objective of this study was to remeasure the Cl O
2 2
spectrum with a technique that was independent of that used
by Burkholder et al.9
In this paper, we present a detailed study of the Cl
absorption spectrum and report absorption cross sections over
the wavelength range 200-420 nm. The Cl UV absorption
spectrum was measured using complementary experimental
techniques and several sources of Cl in order to minimize
possible systematic errors. The primary experimental method
used was pulsed laser photolysis of Cl O or Cl O/Cl mixtures,
at low temperature and high pressure, to produce ClO radicals
and subsequently Cl via reaction 1, combined with diode
array spectroscopy. The Cl cross sections obtained in this
2 2
O UV
2
O
2
2
O
2
2
2
2
2
O
2
2 2
O
Figure 1. Comparison of Cl
section values currently available in the literature. The figure illustrates
the large scatter in Cl cross section values, particularly in the actinic
region. The data for gas-phase Cl spectra recorded at temperatures
in the range 195-296 K were taken from Burkholder et al. (blue line),
2
2
O UV absorption spectra and cross
work at λ >300 nm are in good agreement with the values
9
O
2 2
reported previously by Burkholder et al., substantially higher
2
O
2
than the values reported by Pope et al., and greater than the
10
9
values currently recommended by NASA/JPL6 for use in
1
3
16
DeMore and Tschuikow-Roux (brown line), Huder and DeMore
light blue line, σCl O (λ >310 nm) based on extrapolation), Cox and
2 2
atmospheric models. This work therefore yields larger Cl O
(
2
2
1
5
10
atmospheric photolysis rates and more efficient catalytic ozone
destruction in the polar stratosphere than those obtained by using
the current recommendation. Representative polar stratospheric
photolysis rate coefficients, J, are calculated and the uncertainties
Hayman (dotted line), and Pope et al. (green line). Cross section
values at discrete wavelengths are from Bloss et al. (triangle), Molina
et al. (diamond), and Chen et al. (squares). The data from Molina et
al. and Chen et al. are plotted assuming a Cl photolysis quantum
yield of unity. The Cl spectrum in a Ne matrix at 10 K from von
1
7
7
19
2
O
2
2
O
2
imposed by the estimated Cl
discussed.
2 2
O cross section uncertainty are
1
4
6
Hobe et al. (long dashed line) and the NASA/JPL recommendation
red line) for atmospheric modeling are included for comparison
(
2
4
purposes. The earlier studies by Basco and Hunt and Molina and
Molina have since been shown to have incorrectly identified the UV
spectrum of Cl and are therefore not included.
5
9
2
. Experimental Details
The UV absorption spectrum of Cl
2
O
2
2
2
O was determined from
data obtained using the apparatus shown in Figure 2. The
primary experimental method used pulsed UV laser photolysis
(PLP) of gas-phase Cl O or Cl O/Cl mixtures to produce ClO
2 2
and extent of ozone depletion. The Cl O photolysis rate is
determined from the wavelength, λ, dependence of the overlap
of its absorption spectrum with the solar flux, which is a function
2
2
2
2 2
radicals and subsequently Cl O via the termolecular ClO self-
of solar zenith angle, SZA. For photolysis of Cl
important wavelength region is >300 nm, the actinic region.
The photolysis quantum yield for Cl in this wavelength
region is thought to be unity although only limited studies are
2
O
2
the most
reaction, reaction 1, at low temperatures (200-228 K) and high
pressure (∼700 Torr). Diode array (DA) spectroscopy was used
to measure UV absorption spectra of the reaction mixture that
contained Cl O , Cl O, and Cl after completion of the ClO
2
O
2
2
2
2
2
currently available.7 Although UV photolysis of Cl
,8
O
has been
2
2
2 2
radical gas-phase chemistry and the formation of Cl O .
clearly shown to be critically important in the ClO dimer
catalytic ozone destruction cycle, there exist considerable
discrepancies in its UV absorption spectrum, particularly in the
most critically important wavelength region. An objective of
Sequential pulsed photolysis of the sample and the recorded
UV absorption spectra were used to deduce the Cl O UV
2
2
absorption spectrum and its absolute cross section values. In
the sequence of photolysis steps the partitioning between Cl O ,
2
2
the present study was to resolve discrepancies in the Cl
2
O
2
UV
2 2
Cl O, and Cl changed while the total chlorine content remained
absorption spectrum using laboratory measurements capable of
nearly constant (mass balance). Absolute absorption cross
sections for Cl O were determined at and near the peak of its
2 2
determining the Cl O spectrum in this critically important
2
2
wavelength region.
UV spectrum, 244.5 nm, and other wavelengths in the range
200-420 nm on the basis of three observed isosbestic points,
reaction stoichiometry, and chlorine mass balance. (An isosbestic
point is defined as the waVelength where the total absorbance
does not change during a chemical reaction.) A second
experimental method employed in this study utilized a cold
There are a number of experimental and theoretical studies
for the UV absorption spectrum of Cl currently available in
the literature including a previous study from our laboratory.
A summary of the reported Cl spectra is shown in Figure 1.
It is well established that the Cl UV absorption spectrum is
2
O
2
9
2
O
2
2
O
2
1
0
continuous with a maximum near 245 nm. However, in the
wavelength region λ >300 nm the reported spectra are in poor
agreement, which leads to significant levels of uncertainty in
trapping (CT) technique recently developed by Pope et al. in
which bulk Cl samples are collected in the solid phase.
Thermal desorption (TD) of the samples containing Cl
collected using different trapping temperatures enabled a
2
O
2
2
O
2
2 2
the calculated Cl O atmospheric photolysis rate. The recent

UV Absorption Spectrum of the ClO Dimer
J. Phys. Chem. A, Vol. 113, No. 49, 2009 13713
Figure 2. Schematic of experimental apparatus used to study the UV Cl
2
O
2
absorption spectrum: (a) pulsed laser photolysis diode array (PLP-DA)
setup where M is mirror, RM is a positive position movable dielectric mirror, F is an optical filter, ND is a neutral density filter, BS is a broad band
beam splitter, and BB is beam block; (b) cold trap thermal desorption (CT-TD) setup where TC is thermocouple.
-
4
qualitative measurement of the Cl
wavelength dependence and an evaluation of the influence of
the trapped Cl impurity on the measured spectra. A description
of the experimental setups and procedures used in this study is
given below.
2
O
2
absorption spectrum
an absorption spectrum was typically ∼5 × 10 absorbance
units. Over the duration of an experiment that typically lasted
15 min or less, the drift of the D lamp intensity was <0.001
2
2
absorbance units at all wavelengths.
Absorption spectra were measured in pairs using two separate
lamps as shown in Figure 2 to help minimize the influence
of second order and scattered light effects. One D lamp was
used to measure absorption spectra over the entire wavelength
range of the diode array spectrometer, 190-530 nm. For the
2
other D lamp, a long-pass filter was used in the beam path to
remove short wavelength radiation and minimize second-order
and scattered light effects in the spectrometer. Different
combinations of band-pass and long-pass filters were used over
the course of our study to confirm the accuracy of the absorption
spectra. The majority of the experiments were performed using
a WG280 (λ >280 nm) long-pass filter. To improve optical
system stability, ease of use, and speed with which spectra were
2
.1. Diode Array Spectrometer. UV absorption spectra of
gas-phase species were measured by passing a collimated beam
from a 30 W D lamp through a temperature-regulated reactor/
absorption cell. Two different absorption cells were used as
beam
D
2
2
2
shown in Figure 2 with pathlengths of ∼90 cm. The D
2
exiting the reactor was focused onto the entrance slit of a 0.5 m
spectrometer equipped with a 1024 pixel diode array detector.
The spectrometer used an entrance slit width of 150 µm and a
1
1
50 grooves mm- holographic grating. Spectra were measured
over the wavelength range 190-530 nm at a resolution of ∼1
nm. Emission lines from an Hg pen-ray lamp were used to
calibrate the wavelength scale. Typical detector exposure times
were 0.3 s, and spectra were obtained from the average of 10
detector readings with a total acquisition time of ∼10 s.
Absorption, A, spectra were calculated using Beer-Lambert’s
law
2
recorded, two separate fixed D lamp setups were used rather
than changing optical elements in the beam path during an
experiment. The signal intensities from the two lamps were
balanced by inserting a neutral density filter in the beam path
of the unfiltered lamp. Rapid switching between lamps, a delay
of ∼1 s, was accomplished using automated beam blocks.
Absorption spectra were obtained by combining the two spectra
by averaging the signals in the overlapping wavelength region.
The agreement in the overlapping wavelength region was always
better than 2%.
2.2. Photolysis Sources. An excimer laser operating at 248
(KrF) or 351 (XeF) nm was used as the photolysis source in
the pulsed laser photolysis diode array, PLP-DA, experiments,
method 1. Laser fluences in the reaction cell were in the range
I(λ)
I0(λ)
A(λ) ) -ln
)
(σ (λ)L[x] )
(I)
∑i
i
i
(
)
where I(λ) and I
with and without absorbing species present in the absorption
cell, respectively, σ is the absorption cross section of species i
at wavelength λ, L is the absorption cell path length, and [x] is
the concentration of species i. The peak-to-peak noise level of
0
(λ) are the measured signals at wavelength λ
i
i

1
3714 J. Phys. Chem. A, Vol. 113, No. 49, 2009
Papanastasiou et al.
2
4
-7 mJ cm- pulse-1 for 248 nm and 2-4 mJ cm-2 pulse-1
for the formation of Cl atoms has been reported to be >0.9 at
for 351 nm as measured with a power meter at the exit of the
reactor. The photolysis beam made a single pass through the
reactor and nearly filled the reactor volume. For the cold trap
thermal desorption, CT-TD, experiments, method 2, the pho-
tolysis source was an excimer laser operating at 351 nm, with
two passes through the reactor, or a continuous wave 150 W
Xe arc lamp. The beam from the Xe arc lamp reflected off a
UV cold mirror and passed through a UG5 band-pass filter (220
nm < λ < 430 nm) and a WG320 long-pass filter (λ >320 nm)
before entering the photolysis cell. The effective wavelength
range used for photolysis was 320-430 nm, and the photolysis
beam made a single pass through the reactor. The Xe arc lamp
light source was used in the majority of the CT-DT experiments.
2
248 nm. The initial Cl O concentration was varied over the range
1
5
-3
(4.7-7.7) × 10 molecules cm and the calculated initial ClO
1
3
radical concentrations were in the range (5.6-10.8) × 10
-
3
molecules cm .
2
Cl atoms formed in the photolysis of Cl O react primarily
with Cl
form Cl
2
O following the initial laser pulses of an experiment to
and a ClO radical
2
Cl + Cl O f Cl + ClO
(7)
2
2
where k
7
(T) ) 6.2 × 10 exp(130/T) cm molecule s-1 and
-
11
3
-1
-
10
3
-1 -1
k
7
(200 K) ) 1.2 × 10 cm molecule s . The Cl
2 2
O source
2
.3. Production of Cl
used over the course of the study to produce gas-phase Cl
2 2
O . Several chemical schemes were
chemistry is such that for every two Cl
is one Cl and one Cl molecule formed. This stoichiometry
leads to the following relationships for the changes in Cl O,
Cl , and Cl concentrations
2
O molecules lost there
2
2
O .
2 2
O
2
The chemical schemes used different photolytic sources to
produce the ClO radical in the gas phase and were used to
evaluate the influence of secondary chemistry and spectrally
interfering species. All experiments were performed at low
temperature (200-228 K) and high pressure (∼700 Torr, He)
2
2
2 2
O
∆
[Cl O] ) ∆[Cl ] + ∆[Cl O ]
(II)
2
2
2
2
2 2
to optimize the production of Cl O via reaction 1 and minimize
ClO loss via its bimolecular self-reaction
where ∆[Cl
2
O] is defined as [Cl
2
2
O]initial - [Cl O]final and
ClO + ClO f Cl + O
(5a)
2
2
∆
[Cl ] ) ∆[Cl O ]
(III)
2
2
2
f ClOO + Cl
f OClO + Cl
(5b)
(5c)
These stoichiometric relationships are used in the interpretation
of the measured UV absorption spectra and determination of
-
12
3
-1 -1
2 2
Cl O absorption cross section values described later. The further
photolysis of the reaction mixture leads to a gradual but
predictable breakdown of the reaction stoichiometry due to the
where k5a(T) ) 1.0 × 10 exp(-1590/T) cm molecule s ,
-
11
3
-1 -1
k
k
5b(T) ) 3.0 × 10
5c(T) ) 3.5 × 10
exp(-2450/T) cm molecule s , and
13 3
-
-1
-1
exp(-1370/T) cm molecule s .
secondary loss of Cl
2 2
O via photolysis
(
Thermochemical parameters, rate coefficients, quantum yields,
and absorption cross section data quoted in this paper are taken
from Sander et al.6 unless noted otherwise.) At the low-
temperature and high-pressure conditions used in our experi-
ments reaction 5 was a minor loss for ClO radicals compared
Cl O + hν(248 nm) f Cl + ClOO
(8)
2
2
where the quantum yield for photolysis is unity and by its
reaction with Cl atoms
to reaction 1 where k /k ) 0.0017 at 200 K and 0.007 at 228
5 1
-
13
K for 700 Torr (He), k
1
(200 K, 700 Torr (He)) ) 5 × 10
3
-1
-1
cm molecule
s . At low temperature, the ClO/Cl O
2 2
equilibrium
Cl + Cl O f Cl + ClOO
(9)
2
2
2
ClO + ClO + M T Cl O + M
(1-1)
(228
2
2
(T) ) 1 × 10 cm molecule s-1. The ClOO radical
-10
3
-1
where k
9
formed in reactions 8 and 9 rapidly thermally decomposes
-
28
3
-1
where K
K) ) 6.3 × 10 cm molecule ) highly favors the formation
of Cl such that the ClO concentration at equilibrium is small
c
(T) ) 9.3 × 10 exp(8835/T) cm molecule (K
c
-11
3
-1
ClOO f Cl + O2
(10)
2
O
2
12
-3
(
<4 × 10 molecules cm ) and below our detection limit in
experiments performed at temperatures <210 K.
where k10(200 K) ) 1 × 106 s-1 leading to the nearly
The ClO radical precursor used in the majority of our 248
instantaneous formation of a Cl atom for our experimental
2
nm pulsed photolysis experiments was Cl O photolysis
conditions. The significance of Cl
as the Cl O concentration decreases over the course of an
experiment. As a result, the Cl concentration is expected to
increase initially and go through a maximum. The Cl
photolysis experiments were used to determine the wavelength
2 2
O secondary loss increases
2
Cl O + hν(248 nm) f Cl + ClO
λT < 843 nm (6a)
2 2
O
2
2
O
f O + Cl2
λT < 713 nm (6b)
dependence of the Cl
cross sections.
2 2
O spectrum and its absolute absorption
f O + Cl + Cl
λT < 291 nm
(6c)
A second source that was extensively used was the 351 nm
pulsed laser photolysis of mixtures of Cl /Cl O. In this method,
ClO radical formation was initiated by the photolysis of Cl2
where the photolysis thresholds, λ
standard heats of formation. Cl
T
, were calculated using
2
2
2
O absorbs strongly at 248 nm
-18
2
-1
with a cross section of 1.7 × 10
cm molecule at 298 K
and has a photolysis quantum yield of unity. The formation of
Cl + hν(351 nm) f Cl + Cl (11)
2
Cl + ClO is the major product channel and the quantum yield

UV Absorption Spectrum of the ClO Dimer
J. Phys. Chem. A, Vol. 113, No. 49, 2009 13715
where the quantum yield for Cl atom production is 2. The initial
photolysis light source. The number of laser pulses used varied
depending on the type and conditions of the experiment being
performed. The mirror (RM) was then removed, and spectra
were recorded for the change in composition of the reactor. This
sequence of photolysis and spectrum measurements was then
repeated. At the completion of the photolysis experiment,
typically after 10-30 photolysis steps, the reactor was flushed
1
5
Cl
2
concentration was in the range (3-8.3) × 10 molecules
-
3
1
2
cm and the initial Cl atom concentrations were (6-17) × 10
-3
atom cm for the range of photolysis laser fluences used. The
1
5
-3
Cl
to minimize its initial absorption between 300 and 370 nm. Less
than 5% of the initial Cl atom concentration was from Cl
photolysis at 351 nm. Cl atoms react with Cl O to produce an
equimolar amount of ClO. An advantage of this ClO source
was that Cl photolysis was significantly reduced due to its
2
concentration was kept <9 × 10 molecules cm in order
2
O
2
out and reference spectra, I , recorded again. The total duration
0
of a typical experiment was ∼15 min, although experiments of
2
O
2
shorter and longer durations were also performed. The lamp
smaller absorption cross section at 351 nm compared to 248
nm, the difference in cross section being about a factor of 100.
spectra, I , measured before and after the sequence of photolysis
steps were compared and typically agreed to better than 0.001
0
Therefore, higher Cl
nm photolysis of Cl /Cl
This source was used to determine the wavelength dependence
of the Cl spectrum and absorption cross section values.
Another source used to produce ClO radicals was the reaction
O
2 2
concentrations were obtained from 351
absorbance units at all wavelengths. Reference spectra of Cl2
and OClO were measured immediately following a photolysis
experiment at the same temperature and pressure.
2.5. Cold Trap Thermal Desorption (CT-TD), Method 2.
The cold trap thermal desorption (CT-TD) apparatus is shown
in Figure 2 and was constructed entirely of quartz. The general
2
2
O than from 248 nm photolysis of Cl O.
2
2
O
2
of Cl atoms with O
3
approach used to study the UV absorption spectrum of Cl
with this apparatus follows the methodology developed by Pope
et al.10 to collect bulk samples of Cl
and measure its gas-
2 2
O
Cl + O f ClO + O
(4)
3
2
2
O
2
phase absorption spectra during thermal desorption. The three
main regions of the apparatus consist of a gas-phase photolysis
reactor, a cold trap, and an absorption cell.
-
11
exp(-200/T) cm molecule s-1
3
-1
where k
and the Cl atoms were produced in the 351 nm photolysis of
Cl . This was the only source used in the CT-TD experiments.
In the PLP-DA experiments, the use of high O concentrations
interfered with the UV absorption measurements while using
low initial O concentrations yielded low Cl concentrations.
Therefore, results obtained using this source were primarily
limited to determining the wavelength dependence of the Cl
absorption spectrum in the wavelength range between 200 and
80 nm.
.4. Pulsed Laser Photolysis Diode Array Absorption
PLP-DA), Method 1. In this method, the ClO sources
described earlier were used to produce Cl in the reactor/
4
(T) ) 2.3 × 10
2
The three regions of the apparatus were temperature regulated
independently, and the continuity of the temperature-regulated
regions helped minimize Cl O thermal decomposition during
2 2
3
3
2 2
O
flow through the apparatus. The photolysis reactor was a
jacketed 25 cm long cylinder (3 cm i.d.) that was temperature
controlled by circulating methanol from a temperature-regulated
reservoir through its outer jacket. The temperature of the gas
within the reactor was typically 208 K as monitored with a
thermocouple inserted in the gas flow (withdrawn during sample
preparation). The ends of the reactor were sealed with O-ring
joints with quartz windows as described earlier. The cold trap
region was a 15 cm long jacketed tube (1.5 cm i.d.) that was
2 2
O
2
2
(
2
O
2
absorption cell under static conditions at temperatures in the
range 200-228 K. Experiments were primarily performed at
high pressure, ∼700 Torr (He), although a few test measure-
ments were performed at lower pressures. The experimental
apparatus shown in Figure 2 consists of a temperature-regulated
reactor/absorption cell that is optically coupled to the pulsed
laser photolysis sources and diode array spectrometer. The
photolysis reactor/absorption cell was a jacketed Pyrex tube 105
cm long with an internal diameter of 2.5 cm. The ends of the
reactor were sealed with O-ring joint connectors that held two
quartz windows separated by an evacuated region, which
enabled the entire absorption path to be contained within the
temperature-regulated region and prevented condensation on the
outside windows. The temperature of the reactor was regulated
by circulating methanol from a cooled reservoir through its
jacket. The temperature of the reactor was measured at the fluid
inlet and outlets and was stable to 0.5 K with a gradient of ∼2
K for a cell temperature of 200 K.
temperature-controlled by flowing precooled N
outer jacket. The N gas was cooled by passing it through a
copper coil submersed in a liquid N bath. The temperature of
the cold trap was regulated between 120 and 155 K by varying
the total N flow rate. The temperature of the sample gas flow
2
gas through its
2
2
2
through the trap region was monitored with a thermocouple
located near the center of the trap. Although the temperature of
the trap was stable to within 1 K during an experiment, there
were significant gradients along its length as measured with a
thermocouple inserted along the axis of the gas flow. At 120
K, the difference between the sample gas temperature at the
entrance and exit of the trap region was ∼20 K. From here on
we will refer to the lowest temperature as the effective
temperature of the trap. For sample collection the trap temper-
ature was held constant and independent experiments were
performed at various temperatures between 138 and 155 K
during our study. The absorption cell had an optical path length
of 92 cm, and its temperature was either 200 or 218 K.
PLP-DA experiments were performed using the following
procedure. First, a lamp reference spectrum, I
0
, was recorded
for both D lamps while flowing He bath gas through the reactor.
2
The CT-TD experiments were performed by first recording
3
-1
The reactor was evacuated and filled with reactants while
continuously monitoring the contents of the cell by UV
absorption. The sample was diluted with He bath gas to a total
pressure of >700 Torr (ambient pressure ) 623 Torr). The
reactant concentrations in the cell were monitored by UV
absorption to ensure a well-mixed and homogeneous sample
prior to starting photolysis. A spectrum of the sample, I, was
recorded for each lamp, and then the photolysis beam steering
mirror (RM) was inserted. The sample was then exposed to the
I
0
with only the bath gas flowing (∼15 STP cm s , ∼650 Torr)
and all regions of the apparatus at ∼200 K. O (0.55-0.65 Torr)
and Cl
(∼0.22 Torr) were then added to the flow and the sample
exposed to the photolysis source while monitoring the formation
of Cl and loss of O and Cl in the UV absorption cell. The
3
2
2
O
2
3
2
residence time of the gases in the reactor was ∼6 s. When the
absorption signal stabilized, the trap temperature was reduced
and a sample collected for ∼90 min. The photolysis source and
reactant gas flows were then stopped, the trap temperature was

1
3716 J. Phys. Chem. A, Vol. 113, No. 49, 2009
Papanastasiou et al.
reduced to ∼120 K, and the apparatus was flushed with bath
gas. In several experiments the sample was held at 120 K while
flushing for ∼30 min although this did not significantly affect
the absorption results. With the trap at 120 K, there was no
0
detectable absorption in the absorption cell and I was recorded.
-
1
The trap was then warmed at a rate of ∼0.2 K s while
recording UV absorption spectra of the desorbed sample as it
flowed slowly through the absorption cell. During warming, the
3
-1
total bath gas flow was reduced to ∼12 STP cm s which
resulted in a ∼50 s residence time for the sample in the
2
absorption cell. In these experiments the D lamp intensity was
attenuated to minimize sample photolysis during its residence
in the absorption cell.
2
.6. Materials. He (UHP, 99.999%), N
UHP, 99.99%), and air (Zero grade) were used as supplied.
Cl (UHP grade) mixtures in He or N bath gases were used
from commercial sources, 0.2% Cl /He (UHP) and 5% Cl /N
UHP), or prepared manometrically in a 12 L Pyrex bulb from
Cl (UHP, 99.97%), 5.1% Cl /He. Ozone was generated by
flowing O through an ozonizer and trapped on silica gel at 195
2 2
(UHP, 99.99%), O
(
2
2
2
2
2
(
2
2
2
K. Ozone was added to the gas flow by passing a small flow of
bath gas through the trap.
Dichlorine oxide (Cl
HgO using two methods. Briefly, Cl
in a Pyrex reactor containing dried HgO (heated overnight at
00 K) in a Cl
:HgO mole ratio of ∼1:2. The mixture was
warmed to 195 K and allowed to stand for 24 h. In the second
method, a flow of Cl gas passed through a reactor containing
glass beads coated with dry HgO. The gas stream exiting the
reactor contained Cl O and Cl and was condensed at 195 K.
The Cl O samples were purified by trap-to-trap distillation to
remove the Cl impurity. The Cl O purity, >99%, was measured
by UV absorption. Cl O was added to the apparatus by passing
a small flow of bath gas over the liquid sample held in a Pyrex
reservoir at 195 K. Reference UV absorption spectra for Cl
and Cl were recorded using the diode array spectrometer. The
measured Cl spectrum was in excellent agreement with
currently recommended values at all temperatures used in this
study. For Cl O, the temperature dependence of the UV/vis
2
O) was prepared by oxidizing Cl
2
with
2
was condensed at 77 K
4
2
2
Figure 3. Reference Cl
2
O and Cl
2
spectra (frame a) and absorption
spectra measured following the pulsed laser photolysis of an initially
pure Cl O sample at 248 nm (frame b) (see text for details) at 201 K
and 725 Torr (He). In frame b the spectrum of the initial Cl O sample
2
2
2
2
2
is shown in black. The additional spectra shown in the figure were
obtained following the sequential photolysis of the sample (a static
experiment) and show the evolution of the combined absorbance due
2
2
6
2
2 2 2 2
to Cl O, Cl , and Cl O . Spectra were recorded approximately 10 s
2
O
after each photolysis step (see text for details). The sequence of
absorption spectra from the beginning to the end of the experiment is
the same as the order of absorbance at 350 nm from low to high values.
2
2
1
5
-3
The initial Cl
2
O concentration was 4.7 × 10 molecules cm and the
-2 -1
photolysis laser fluence was ∼7 mJ cm pulse . The inset shows an
expanded view of the isosbestic point observed near 313 nm. An
isosbestic point is also observed near 409 nm (see text for details).
2
absorption spectrum was not available and was measured in a
separate set of experiments. The details of these measurements
are given elsewhere.1
2
Chlorine dioxide (OClO) was prepared online by slowly
2 2
under static conditions at reduced temperature. The two Cl O
sources used in the majority of the experiments utilized the same
gas-phase chemistry but were initiated differently. Source 1 used
flowing a 5% Cl
containing NaClO
reactor was >60 s, and conversion of Cl
as determined by UV absorption. NaClO
2 2
/N
mixture through a 25 cm long Pyrex tube
crystals. The residence time of Cl in the
to OClO was >95%
crystals were changed
2
2
the 248 nm photolysis of Cl
Cl O mixture to produce ClO radicals. In
the third photochemical source, ClO radicals were produced in
reaction following 351 nm photolysis of Cl in a
mixture. Although experimental conditions were usually
optimized for maximum production of Cl , measurements
2
O and source 2 used the 351 nm
2
photolysis of a Cl2/
2
2
frequently to maintain a high OClO production efficiency. Gas
flow rates were measured using calibrated electronic mass flow
meters. Pressures were measured using a 1000 Torr capacitance
manometer.
the Cl + O
Cl /O
3
2
2
3
2
O
2
were performed using a range of experimental conditions
including variations in temperature, total pressure, photolysis
laser fluence, and initial reactant concentrations.
3
. Results and Discussion
Results from the pulsed laser photolysis diode array absorp-
Figure 3 shows an example of the UV absorption spectra
tion (PLP-DA) and cold trap thermal desorption (CT-TD)
experiments are presented separately below. Analysis of the UV
absorption spectra and determination of the wavelength depen-
measured in a 248 nm Cl
performed at 201 K and 725 Torr (He). Prior to photolysis the
spectrum is due to Cl O absorption. The sequence of spectra
2
O photolysis, source 1, experiment
2
dence of the Cl
absorption cross sections follow, and an evaluation of potential
errors in the Cl spectrum determination is given. Finally, a
brief comparison of the present results with previous studies of
the Cl UV absorption spectrum is presented.
.1. PLP-DA Results. Three photochemical sources were
used to produce gas-phase Cl in the reactor/absorption cell
2
O
2
UV absorption spectrum and its absolute
shown in Figure 3 was recorded following stepwise photolysis
of the reactor/absorption cell contents where each spectrum was
recorded ∼10 s after photolysis. In the initial photolysis steps
2
O
2
2
the gas-phase photochemistry leads to the loss of Cl O and
2
O
2
production of Cl
the sequence of spectra by (1) an increase in absorbance in the
330-380 nm region due to the formation of Cl and Cl , (2)
2 2 2
and Cl O . This is qualitatively observed in
3
2
O
2
2
2 2
O

UV Absorption Spectrum of the ClO Dimer
J. Phys. Chem. A, Vol. 113, No. 49, 2009 13717
K and low pressure, ∼10 Torr (He). At low pressure the
bimolecular ClO + ClO reaction increases relative to the
termolecular reaction and leads to the increased formation of
OClO via reaction 5c and correspondingly to lower Cl
production. The formation of OClO and Cl complicates the
determination of Cl absolute absorption cross sections due
2 2
O
2
O
3
2
O
2
to uncertainties in the secondary chemistry and spectral analysis.
Therefore, only experiments performed at temperatures <210
K and high pressure were included in the final spectral analysis
to determine the Cl O UV absorption spectrum.
2 2
PLP-DA experiments were usually performed with the
shortest delay possible between photolysis steps to minimize
2 2
possible losses of Cl O . Experiments performed using longer
delays, up to 60 s, between photolysis steps, however, yielded
very similar results, which indicate that no significant gas-phase
chemistry occurred during the time required to measure the UV
absorption spectra. Experiments were also performed in which
the extent of photolysis per step was altered by varying either
the number of photolysis laser pulses or the photolysis laser
fluence. The absorption spectra measured using sources 1-3
yielded complementary results and provide the basis for the
2 2
determination of the wavelength dependence of the Cl O UV
absorption spectrum and its absolute absorption cross sections
described below.
The sequence of UV absorption spectra given in Figures 3
and 4 clearly shows the presence of isosbestic wavelengths near
Figure 4. Absorption spectra measured following the pulsed laser
photolysis of a Cl O/Cl mixture at 351 nm (see text for details) at 201
2 2
K and 766 Torr (He). The spectrum of the initial sample is shown in
black. The additional spectra shown in the figure were obtained
following the sequential photolysis of the sample (a static experiment)
3
13 and 409 nm. At these wavelengths, the absorbance remains
nearly constant although the concentrations of Cl , Cl O, and
Cl change significantly between photolysis steps. The reaction
stoichiometry given in eqs II and III defines a relationship
O
2 2
2
and show the evolution of the combined absorbance due to Cl
and Cl . Spectra were recorded approximately 10 s after each
2
2
O, Cl ,
2
O
2
2
photolysis step (see text for details). The sequence of absorption spectra
from the beginning to the end of the experiment is the same as the
order of absorbance at 350 nm from low to high values. The initial
between the absorption cross sections for Cl
at the isosbestic wavelength
2 2 2 2
O , Cl O, and Cl
1
5
-3
concentrations were [Cl
2
O] ) 6.8 × 10 molecules cm and [Cl
2
] )
1
5
-3
4
.5 × 10 molecules cm , and the photolysis laser fluence was ∼3.5
-2 -1
mJ cm pulse . The inset shows an expanded view of the isosbestic
point observed near 313 nm. An isosbestic point is also observed near
σCl2O2(λ , T) ) 2σ (λ , T) - σ (λ , T) (IV)
iso
Cl2O iso
Cl2 iso
4
09 nm (see text for details). The observed isosbestic points are
consistent with the results obtained in the 248 nm pulsed photolysis of
Cl O shown in Figure 3.
2
where λiso is the isosbestic wavelength and σCl O(λiso,T) and
2
σ
Cl2(λiso,T) are the absorption cross sections for Cl
respectively, at λiso and temperature T. This relationship provides
a method to directly determine the Cl absorption cross section
at the isosbestic wavelengths from the Cl and Cl O absorption
cross sections at the same wavelength. Although sources 1 and
2 initiate the Cl formation chemistry differently, the reaction
2 2
O and Cl ,
a decrease in absorbance at ∼280 nm, primarily due to the loss
of Cl O, and (3) an increase in absorbance near 250 nm and a
shift in the absorbance maximum toward shorter wavelengths
due to the formation of Cl and loss of Cl O. An analysis of
, Cl O, and Cl
2 2
O
2
2
2
2
O
2
2
the spectra yields the temporal profiles of Cl
that are presented later.
2
O
2
2
2
2 2
O
stoichiometry for these two sources is the same. Therefore, the
isosbestic points are expected to be the same for both sources
as was observed. Note that the observation of an isosbestic point
Figure 4 shows an example of the UV absorption spectra
obtained using source 2, 351 nm pulsed laser photolysis of a
initial Cl
2
O/Cl
2
mixture, at 201 K and a total pressure of 766
at 408.5 nm provides direct evidence that Cl
wavelength region.
2 2
O absorbs in this
Torr (He). The spectra shown are representative of those
recorded over the range of experimental conditions used in this
study. As observed with source 1, the sequence of UV spectra
For perspective, we note that isosbestic points would have
been predicted, for the Cl /Cl O/Cl reaction system even if
Cl did not absorb between 200 and 450 nm. However, in
2
O
2
2
2
shows a loss of Cl O combined with formation of Cl . The Cl O
2 2 2 2
2 2
O
concentration goes through a maximum, although higher Cl
concentrations were achieved using this source due to the much
2
O
2
that case the isosbestic points would have fallen at ∼322 and
∼400 nm (for experiments performed at 200 K). The isosbestic
wavelengths observed here are significantly shifted from these
lower extent of Cl
nm.
2 2
O photolysis at 351 nm compared to 248
2 2
values indicating that Cl O absorbs at these wavelengths.
PLP-DA experiments performed at temperatures >210 K
Upon close examination, the isosbestic point observed near
313 nm is not strictly constant but shifts gradually toward longer
wavelengths during the course of an experiment as shown in
the insets in Figures 3 and 4. The shift toward longer wavelength
is consistent with the breakdown of the reaction stoichiometry
showed OClO formation that increased with increasing tem-
perature. OClO leads to the formation of Cl
2 3
O via
ClO + OClO + M T Cl O + M
(12)
2
3
2 2
due to the loss of Cl O via secondary chemistry, photolysis,
and reaction with Cl atoms. The breakdown in stoichiometry is
systematic and predictable with the shift toward longer wave-
2 3
and evidence for the formation of Cl O was also observed under
some conditions. Test experiments were also performed at 200
2 2
length resulting from the fact that the Cl O absorption cross

1
3718 J. Phys. Chem. A, Vol. 113, No. 49, 2009
Papanastasiou et al.
section at this wavelength is greater than that of Cl
detailed analysis of the systematic dependence of the isosbestic
point shift and the dependence on the Cl absorption cross
2
. A more
2
O
2
section is given in the Supporting Information. A systematic
shift in the 408.5 nm isosbestic point was not observed within
the measurement uncertainty because the Cl
section is similar in magnitude to those of Cl
wavelength.
2
O
2
2
absorption cross
O and Cl at this
2
The observed isosbestic wavelengths were found to be
independent of temperature (200-228 K), the initial concentra-
2 2 2
tions of Cl and Cl O, and addition of O (10 Torr) to the
reaction mixture. Experiments performed using band-pass (309
nm, fwhm ) 10 nm) and cutoff optical filters (WG-320 (λ >320
nm) and WG-375 (λ >375 nm) to isolate the isosbestic
wavelength regions yielded identical results. The 408.5 nm
isosbestic point was also examined in an experiment with a
higher Cl
Figures 3 and 4. In this measurement absorption at wavelengths
300 nm was optically thick. However, the signal in the long
2
O
2
concentration (a factor of ∼4) than that shown in
<
wavelength region was greater and thereby enabled a more
accurate absorption measurement. The isosbestic wavelength
was identical to that observed in the lower concentration
experiments, and no significant shift in the isosbestic wavelength
with extent of reaction was observed which is consistent with
a Cl
and Cl
2
O
2
cross section that is similar in magnitude to that of Cl
O at 408.5 nm.
2
2
The isosbestic wavelength in the limit of zero concentration
change was determined in the source 1 experiments to be 312.9
(
0.16 where the quoted uncertainty is the standard deviation
of seven measurements. The isosbestic wavelength determined
using source 2 was consistent with the source 1 results although
the precision in nine measurements was lower, (0.69 nm. In
addition to the isosbestic points observed at 312.9 and 408.5
nm, a pseudoisosbestic point was observed near 271 nm at low
Cl
Roux reported an isosbestic point at 271 nm in their Cl
photolysis experiments. In our experiments, the behavior of the
absorbance near 271 nm was very sensitive to Cl loss and
the breakdown of the reaction stoichiometry. This sensitivity
made the isosbestic wavelength determination less accurate, and
therefore it was not utilized in our spectral analysis. However,
2
O conversion. In an earlier study, DeMore and Tschuikow-
Figure 5. Results obtained in the cold trap thermal desorption (CT-
TD) experiments. Frame a: Sequence of absorption spectra measured
during thermal desorption of a sample collected at a trap temperature,
13
2
2
/Cl O
O
2
T
C
, of 148 K. The spectra color code is given in frame c. Frame b: The
UV absorption spectra shown in frame a normalized at the peak of the
Cl spectrum, 244.5 nm, showing the change in the Cl /Cl
fractionation of the sample as the trap temperature increased. The heavy
black curve is the Cl spectrum obtained in this work. Frame c: The
2
2
O
2
2
O
2
2
2
O
2
2 2
as shown later, the Cl O cross section determined in this work
ratio, A(330 nm)/A(244.5 nm), as a function of the desorption
is consistent with an isosbestic point near 271 nm.
temperature, T , for samples collected at 138 K (3), 146 K (4), 148 K
D
(
b and solid line) color coded to spectra shown in frames a and b, 151
2 2
On the basis of the observed isosbestic behavior the Cl O
K (0), 153 K (]), and 155 K(O). The error limits shown are 2σ (95%
confidence level) values from the absorption measurements. The larger
size symbols indicate the desorption temperature where the maximum
absorption cross sections at 312.9 and 408.5 nm were determined
-19
2
-1
using eq IV to be (3.69 ( 0.24) × 10 cm molecule and
-
21
2
-1
(
9.84 ( 1.85) × 10 cm molecule , respectively. The Cl
2
O
2 2
Cl O absorption signal was observed. The dashed line and shaded 2σ
absorption cross sections at 200 K used in the isosbestic point
uncertainty range is the ratio from the Cl
PLP-DA experiments in this work.
2 2
O spectrum obtained in the
-
18
2
-1
analysis were σCl O(312.9 nm) ) 2.88 × 10 cm molecule
2
-
20
2
-1 12
and σCl O(408.5 nm) ) 1.11 × 10
cm molecule . The
2
quoted uncertainty is at the 2σ (95% confidence) level and
concentrations used, Cl condensation was shown to be near
2
includes estimated systematic uncertainties in the Cl
cross sections at these wavelengths and temperatures. Combining
the Cl absorption cross section determined at the isosbestic
points with the wavelength dependence of the Cl spectrum
2
O and Cl
2
our detection limit, for T >148 K. The absorption spectra and
C
composition of the gas-phase sample as a function of T and
C
2
O
2
desorption temperature, T , are summarized in Figure 5. Similar
D
2
O
2
results were obtained using the 351 nm excimer laser as the
enables the determination of the peak absorption cross section
as well as the absorption cross sections at other wavelengths
photolysis light source although the production of Cl O was
2
2
significantly less. We have used the ratio of absorbances at 330
and 244.5 nm, A(330 nm)/A(244.5 nm), to evaluate the
fractionation of the sample as a function of collection and
desorption temperature. Smaller values of the ratio correspond
(
see section 3.3).
.2. CT-TD Results. CT-TD experiments were performed
with collection trap temperatures, T , of 138, 146, 148, 151,
53, and 155 K. The experimental conditions including [Cl
], photolysis photon flux, total flow rate, flow
3
C
1
2
]
to higher fractionation and purer gas-phase samples of Cl
In the limit of a pure Cl sample, the ratio corresponds to
the ratio of Cl absorption cross sections at these wavelengths.
As shown in Figure 5 for T ) 148 K, the ratio is nearly constant
2 2
O .
(
∼0.22 Torr), [O
3
2 2
O
velocity, and collection time were similar in each experiment
and only the collection trap temperature was changed. For the
2 2
O
C

UV Absorption Spectrum of the ClO Dimer
J. Phys. Chem. A, Vol. 113, No. 49, 2009 13719
for T
an asymptotic value of 0.037. The ratio also shows a systematic
decrease with increasing T . Measurements made with T >151
K yielded a ratio of 0.027 for all values of T ; that is, the sample
composition did not change with increasing T . The absolute
Cl absorption signal increased with increasing T and reached
maximum values for desorption temperatures in the range
48-155 K, depending slightly on T . For experiments with
<145 K the maximum Cl absorbance was the highest with
values of ∼1.4 at the peak (∼2 × 10 molecules cm ). At T
183 K Cl and Cl were completely removed from the trap
D
D
<140 K but decreases rapidly at higher T and approaches
C
C
D
D
O
2 2
D
1
C
T
C
2 2
O
1
5
-3
D
≈
2
O
2
2
and the measured spectrum returned to the initial background
value ((0.001 absorbance unit). The approach to an asymptotic
ratio with increasing T
distillation of the sample that reduces the amount of Cl
trapped sample at the higher T
D
and T
C
is consistent with a fractional
in the
values. The absorption spectra
2
C
measured throughout the desorption process were analyzed
within the precision of the measurement to be due to contribu-
tions from only two absorbing species, Cl and Cl O , for all
2 2 2
conditions and temperatures shown in Figure 5. The lowest
values of the ratio obtained using the CT-TD method, 0.027,
correspond to an absorption spectrum that is in excellent
2 2
agreement with the Cl O spectrum obtained in our PLP-DA
experiments, as shown in Figure 5.
Figure 6. Data used in the determination of the Cl
section at 244.5 nm from the pulsed laser photolysis diode array
absorption (PLP-DA) experiments using the 351 nm photolysis of Cl O/
Cl mixtures (open symbols) and the 248 nm photolysis of Cl O (solid
symbols). The symbols represent individual experiments that were
performed over a range of experimental conditions. The lines were
obtained from a global least-squares fit to all the data. The initial slope
2 2
O absorption cross
The temperature dependence of the Cl
2
O
2
spectrum was
samples
) 151 K. A high
trap collection temperature was used to minimize the Cl
contribution to the absorption spectrum. The Cl spectrum
2
investigated over the range 201-218 K using Cl
prepared using the CT-TD method with T
2 2
O
2
2
C
2
2
O
2
showed only small, <2%, changes over this very narrow
yields a Cl
44.5 nm of (7.58 ( 0.3) × 10 cm molecule where the quoted
uncertainty is the 2σ precision of the fit.
2
O
2
2
absorption cross section, σCl O ) ∆A/(0.5∆[Cl O]L), at
2
2
-
18
2
-1
2
temperature range. The weak temperature dependence of the
Cl
2
O
2
spectrum, particularly between 200 and 300 nm, implies
spectra measured in this
that direct comparisons of the Cl
2
O
2
work with those reported in other studies but at slightly different
temperatures can be made quantitatively.
experiments and sources 1-3 and the CT-TD experiments using
source 3. The Cl spectrum in this wavelength region was
also determined via the PLP-DA method using the previously
reported location of the Cl absorption peak at 244.5 nm.
The pulsed laser photolysis of Cl /O mixtures at 351 nm, source
3, yielded low Cl concentrations, but reliable measurements
of the Cl spectrum in the 200-280 nm range were obtained.
Both experimental methods and the three Cl sources used
yielded Cl spectra that were in excellent agreement, within
2
O
2
Finally, a weak Cl
was observed when T
than Cl . In addition, near T
volatility condensate desorbed from the trap and condensed in
the absorption cell. Absorption spectra recorded while using
higher absorption cell temperatures eliminated the condensation,
but we were unable to characterize the gas-phase absorption
spectrum of the compound due to the very weak absorption
2
O
D
3
absorption signal (A < 0.01 at 265 nm)
approached 200 K; Cl is less volatile
≈ 215 K an unidentified low
O
2 3
2 2
O
2
O
2
D
2
3
2 2
O
2
O
2
2
O
2
2
O
2
2%, in this wavelength region.
The spectral subtraction of Cl
signal. The formation of Cl
Cl when using the CT-TD method was not reported in
previous studies using this method.
performed using broad band photolysis, 320-800 nm, of the
Cl /O mixture led to significant production of OClO and Cl
The amount of OClO and Cl produced was sensitive to the
initial O concentration in the reactor. It is most likely that
photolysis of O in the Chappuis band (400-700 nm) leads to
the formation of O atoms and secondary chemistry leading
to OClO and Cl , although, additional studies are needed to
2
O
3
or species other than Cl
2
and
2
O in each photolysis step of
O concentra-
O is related to the formation of Cl and
via the reaction stoichiometry (eqs II and III) described
earlier
O
2
the PLP-DA experiments yields the change in Cl
tion. The loss of Cl
Cl
2
2
1
0,14
Test measurements
2
2 2
O
2
2
3
2 3
O .
2
O
3
3
0
.5∆[Cl O] ) ∆[Cl O ]
(V)
2
2
2
3
2
O
3
from which the cross sections for Cl O in the wavelength range
2
2
identify the reaction mechanism. Optically filtering the pho-
tolysis wavelengths to 320-430 nm and minimizing the initial
3
O concentration reduced the production of OClO below our
200-280 nm were determined. Figure 6 shows a summary of
the results obtained in the PLP-DA experiments for both sources
1 and 2. The experimental data follow the linear relationship
detection limit in most experiments.
.3. Cl UV Absorption Spectrum Analysis. The Cl
of eq V at low Cl O conversion and negative curvature at higher
2
3
2
O
2
2
O
2
Cl O conversion due to the loss of Cl O (lower residual
2
2
2
spectrum was determined using knowledge of the gas-phase
chemistry and reaction stoichiometry combined with the isos-
bestic wavelengths observed in the PLP-DA experiments. The
absorption), which is greater in the 248 nm photolysis experi-
ments. Several experiments shown in Figure 6 were performed
using multiple laser pulses per step and are indistinguishable
from the results obtained using less photolysis per step. The
wavelength dependence of the Cl
nm region, where the contribution from Cl
negligible, was obtained using results from the PLP-DA
2
O
2
spectrum in the 200-280
2
absorption is
initial slope, i.e., at low Cl
2
O conversion, obtained from a least-
absorption
squares polynomial fit to the data yielded a Cl
2 2
O

1
3720 J. Phys. Chem. A, Vol. 113, No. 49, 2009
Papanastasiou et al.
Figure 7. Cl
UV absorption spectra measured following the 248 nm pulsed laser
photolysis of Cl O at 201 K and 725 Torr (He) (see text for details).
The data points at 271, 312.9, and 408.5 nm (b) are the Cl
2
O
2
UV absorption spectra obtained from the analysis of
2 2 2 2
Figure 8. Temporal profiles of Cl O (b), Cl O (2), Cl (9), and the
sum ([) obtained in a pulsed laser photolysis diode array absorption
experiment at 201 K and 725 Torr (He). The lines shown were obtained
from a simulation of the reaction system (see text for details).
2
2
O
2
absorption cross section values determined at the observed isosbestic
points (see text). The lower frame shows the residuals of the spectra
from the average. The residuals show deviations that are less than 2%
between 200 and 300 nm, 5% between 300 and 350 nm, and increase
toward longer wavelengths.
increases at longer wavelengths. The consistency of the absorp-
tion spectra for each photolysis step shows that the spectra
measured in each photolysis step can be interpreted, within the
precision of the measurements, by absorption from Cl
Cl and another species that we assign to Cl . The residual
spectra in Figure 7 show a precision that is typical for the various
Cl sources and experimental conditions used in this study.
2
O and
2
2 2
O
-
18
cm2
cross section at 244.5 nm of (7.58 ( 0.3) × 10
molecule- where the quoted error is the 2σ uncertainty of the
1
2 2
O
fit. The absorption cross sections at other wavelengths between
The residual spectra obtained in the PLP-DA experiments and
the spectrum obtained from method 2, for spectra with the ratio
A(330 nm)/A(244.5 nm) ) 0.027, are also in excellent agreement
as shown in Figure 5.
2
00 and 280 nm were obtained relative to the peak value using
the wavelength dependence of the Cl spectrum obtained in
the PLP-DA and CT-TD experiments. The temporal profiles
for Cl , Cl O, and Cl and the observed reaction mass balance
are described later.
2
O
2
O
2 2
2
2
The formation and stability of absorbing species other than
Cl
Cl
2
O
O
2
, such as isomers of Cl
, were evaluated by allowing a Cl
2
O
2
(chloryl chloride, ClClOO) or
/Cl O/Cl mixture
A similar absolute cross section determination method to that
shown in Figure 6, eq V, was used to obtain a peak cross section
2
3
2
O
2
2
2
to stand unperturbed in the absorption cell at 200 K for up to
20 min. The Cl concentration showed a first-order loss with
for Cl
2
, which is well-known. The peak Cl
2
absorption cross
2 2
O
-
4
-1
section obtained using PLP-DA data, source 1, agrees with the
a rate coefficient of ∼5 × 10 s , but the wavelength
dependence of the residual absorption spectrum remained
constant. We conclude from these measurements that other
chlorine species did not make a significant contribution to the
measured absorption spectrum in this study.
-19
2
-1 6
currently recommended value of 2.68 × 10 cm molecule
at 200 K with a standard deviation in seven determinations of
%. The good agreement achieved for the peak cross section
of Cl confirms the self-consistency of the reaction stoichiometry
used in the spectral analysis and that the Cl absorption in our
measured absorption spectra was accounted for accurately.
The determination of the Cl absorption spectrum at
wavelengths >280 nm required the quantitative subtraction of
absorption due to Cl . The Cl absorption spectrum at
7
2
2
3.3.1. Mass Balance. The partitioning and total chlorine mass
balance (total ) Cl
2
O
2
+ Cl
2
2
O + Cl ) for the PLP-DA spectra
2
O
2
shown in Figure 3 is given in Figure 8. This data set is
representative of the profiles obtained using sources 1 and 2. A
simulation of the reaction system that included reactions 1–3,
5-11, and pseudo-first-order rate coefficients for the loss of
2
2 2
O
wavelengths >280 nm was determined using the cross sections
at the isosbestic wavelengths combined with the cross section
values obtained in the 200-280 nm region. Residual spectra
-
4
-1
-5 -1
Cl
2
O
2
(5 × 10 s ) and Cl
2
O (7 × 10 s ), as measured in
this work, is included in Figure 8. The experimentally deter-
were obtained for each photolysis step by subtracting Cl
Cl from the measured absorption spectrum. A typical set of
residual spectra, normalized at 244.5 nm, the peak of the Cl
spectrum, is shown in Figure 7. The agreement for the
2
wavelength dependence of the residual absorption spectra, Cl O ,
2
O and
2 2 2 2
mined Cl O , Cl O, and Cl temporal profiles are consistent with
2
the behavior predicted by the reaction mechanism. Good
agreement between the experimental data and simulated tem-
poral profiles was obtained by allowing the photolysis laser
fluence to vary (5% from the measured value. It is worth noting
2
O
2
2
at each photolysis step is excellent. The percent deviation of
the residual spectra from the average, Figure 7, is less than 2%
between 200 and 300 nm, 5% between 300 and 350 nm, and
2 3
that the concentrations of OClO and Cl O calculated in the
1
2
-3
simulation are <1 × 10 molecules cm and therefore make
negligible contributions to the measured UV absorption spectra.

UV Absorption Spectrum of the ClO Dimer
J. Phys. Chem. A, Vol. 113, No. 49, 2009 13721
Supporting Information. The various experimental methods and
chemical sources used in this study yielded Cl
spectra that were in very good agreement. The Cl
2
O
2
absorption
2
O absorption
2
spectrum reported here was obtained from measurements near
00 K. The temperature dependence of the Cl absorption
spectrum over the range 200-218 K was found to be small.
The Cl UV absorption spectrum in the wavelength range
00-420 nm consists of several overlapping electronic bands.
The spectrum peaks at 244.5 nm with a cross section value of
2
2 2
O
2
O
2
2
-
18
2
-1
7
.6 × 10 cm molecule . Toward shorter wavelengths the
spectrum shows a minimum at ∼218 nm and the onset of
another strong absorption feature whose maximum is outside
the wavelength range of our measurements. The absorption
features at longer wavelength appear as shoulders on the main
absorption band at 270 and 340 nm. The diffuse absorption band
near 340 nm correlates with the Cl
Therefore, the accuracy of the Cl
this wavelength region is dependent on accurately accounting
for absorption by Cl . Although the uncertainty in our measure-
2
absorption spectrum.
2
2
O absorption spectrum in
2
ments at wavelengths longer than 380 nm is large, >50%, the
observation of an isosbestic point at 408.5 nm in our experiments
indicates that Cl
.4. Error Analysis. Systematic and absolute uncertainties
in our measurements affect both the wavelength dependence
of the Cl absorption spectrum and its absolute absorption
cross sections. The high precision of our absorption measure-
ments was the result of the stability of the D lamp intensity
and optical setup. In the wavelength range 200-280 nm, the
uncertainty in the Cl absorbance due to the measurement
precision was small, <1%. At longer wavelengths where the
absorption cross sections for Cl are smaller, the measurement
precision makes a larger contribution to the overall uncertainty:
2% at 300 nm, ∼5% at 350 nm, and ∼50% for λ >400 nm.
Figure 9 shows the Cl absorption spectrum obtained in this
work along with the estimated uncertainty as a function of
wavelength. The estimated uncertainty in the Cl absolute
cross sections is 5-10% (2σ confidence level) at and around
the peak of the Cl spectrum, near 245 nm. A parametrization
2 2
O absorbs in this wavelength region.
3
2 2
Figure 9. The Cl O absorption spectrum obtained in this work (heavy
black line) including the estimated error limits as a function of
wavelength given by the shaded region (see text of discussion of error
analysis). The absorption cross sections at the isosbestic points (271,
2 2
O
2
3
12.9, and 408.5 nm) from this work are also shown (red circles). The
heavy dashed line gives an exponential extrapolation of our Cl cross
2 2
O
O
2
section data beyond the range of our measurements. The light gray
shaded region shows the extrapolation of the estimated cross section
uncertainty. See the Supporting Information for the cross section data
and extrapolation parameters. The absorption cross section data from
2
2 2
O
9
17
10
Burkholder et al. (blue line), Bloss et al. (triangle), and Pope et al.
∼
6
(
green line) as well as the NASA/JPL recommendation (red line) are
included for comparison. The data from Chen et al. (squares) at 308
1
9
2 2
O
and 351 nm are given assuming a Cl
unity at these wavelengths.
2 2
O photolysis quantum yield of
2
O
2
2 2
O
The chlorine mass balance is very good, constant to within 1%,
in the early stages of the experiment. However, following the
maximum in Cl O concentration a small systematic loss of total
2 2
chlorine was observed in nearly all experiments with the loss
at the end of the experiment being ∼8%. The loss of total
chlorine is reproduced reasonably well in the simulation when
for the estimated uncertainties in the actinic region is given in
the Supporting Information. The estimated uncertainty includes
the uncertainty associated with the spectral subtraction of Cl
and Cl , which does not contribute significantly to the overall
uncertainty, <1%. The uncertainty in the isosbestic wavelength
at 312.9 nm is estimated to be +0.3/-0.5 nm. The Cl
spectrum obtained using the lower limit of the isosbestic
wavelength in the spectral analysis would increase the Cl
cross section in the wavelength region 330-350 nm as a result
of smaller Cl subtraction. The increase would be 16% at 350
nm for an isosbestic wavelength of 312.4 nm and a decrease of
2
O
2
2
O
2
Cl
2
O
2
and/or Cl
2
O losses, either homogeneous or heterogeneous,
are included.
that do not lead to the formation of gas-phase Cl
2
2
O
2
Several experiments were performed to examine the chlorine
gas-phase loss in more detail. At the completion of a PLP-DA
experiment the reaction cell was pumped out, sealed off, and
2
warmed to room temperature. Cl
were observed to gradually build up as the cell warmed. The
total amount of Cl accounted for the majority of the observed
2
and a small amount of OClO
1
0% for an isosbestic wavelength of 313.2 nm.
The presence of unaccounted for impurities was examined
2
experimentally and determined to be negligible. The most likely
impurities are OClO and Cl while ClClOO and other chlorine
decrease in chlorine mass balance. We conclude that thermally
unstable nonvolatile chlorine containing species were produced,
either homogeneously or heterogeneously, in the later stages
of the PLP-DA experiments. However, the UV absorption
spectra measured in the later stages of an experiment yielded
Cl O absorption spectra that were identical to spectra recorded
2 2
in the early stages. Therefore, the formation of unidentified
compounds did not significantly influence the measured UV
2 3
O
oxides were also considered. OClO, which has a strong and
readily detectable structured UV/vis absorption band, did not
make a significant contribution to our measured spectra. We
estimate that Cl
nm, to our measured absorption spectra. ClClOO, an isomer of
Cl , absorbs strongly in the UV. On the basis of our detection
2 3
O made a negligible contribution, <1% at 265
2
O
2
11
-3
limit for ClClOO, ∼7 × 10 molecules cm , we estimate that
its contribution to the measured absorption spectra was <1% at
all wavelengths.
absorption spectra or the determination of the Cl
absorption cross sections in our experiments.
2 2
O absolute
3
2 2 2 2
.3.2. Cl O Spectrum. The Cl O absorption spectrum
obtained in this study over the wavelength range 200-420 nm
is shown in Figure 9 and listed in 0.5 nm intervals in the
The influence of secondary chemistry is difficult to quanti-
tatively evaluate. The high degree of consistency for the residual

1
3722 J. Phys. Chem. A, Vol. 113, No. 49, 2009
Papanastasiou et al.
2 2
TABLE 1: Summary of Cl O UV Absorption Spectrum Studies
2
0
2
reference
year
temperature (K)
technique
BB-DA
DF-DA/FTIR
BB-DA
BB-DA
PLP-TA
CT-TD
wavelength (λ) (nm)
10 σ(λ) (cm )
1
5
a
Cox and Hayman
1988
1990
1990
1995
2001
2007
2009
2009
200-300
205-250
206
195
183-245
193
220-360
212-410
190-400
200-310
210
226-355
220-400
308
640 ( 60
650
9
+80a
Burkholder et al.
DeMore and Tschuikow-Roux
-50
1
3
a
680 ( 80
1
6
b, d
Huder and DeMore
-
1
7
Bloss et al.
294 ( 86
1
0
b
Pope et al.
von Hobe et al.
-
1
4
b, c
10
200
MI
PLP-MS
-
1
9
e
Chen et al.
49
2
2
2
50
00
50
308
351
351
200-450
50.9
11.2
12.6
+
80a
this work
2009
200-228
PLP-DA
CT-TD
760
-
50
a
Absorption cross section value at the peak of the spectrum, λ ) 245 nm. Cross section data reported over the range of wavelengths given.
b
c
Reported Cl
a Ne matrix. Cl
measured at shorter wavelengths. Cl
2
d
O
2
absorption spectrum normalized to the recommended peak absorption cross section. Cl
2 2
O absorption spectrum measured in
2 2
O
absorption spectrum reported at λ >310 nm is reported in Huder and DeMore using a log-linear extrapolation of values
e
2
2 2 2
O cross section calculated here assuming a unit Cl O photolysis quantum yield at these wavelengths.
BB, broad band photolysis; DA, diode array spectroscopy; DF, discharge flow; FTIR, Fourier transform infrared spectroscopy; PLP, pulsed
laser photolysis; TA, transient absorption; MS, mass spectrometry, CT-TD, cold trap thermal desorption; MI, matrix isolation.
absorption spectra obtained in the PLP-DA experiments suggests
however that it is small. The photolysis of Cl O at 248 nm is
spheric modeling studies is included in Figures 1 and 9 for
comparison purposes. The IUPAC20 evaluation panel currently
recommends the Cl O spectrum and extrapolation for the long
wavelength region reported in Huder and DeMore. The
discrepancies among the previous studies and comparison with
the results from the present study are discussed here.
2
reported to yield >90% Cl atoms. Therefore, a small yield of O
2
2
1
6
atoms is possible. O atoms would react with the Cl
present in large excess initially
2
O that is
O + Cl O f ClO + ClO
(13)
2 2
Although there have been numerous studies of the Cl O UV
2
absorption spectrum reported over the past 30 years, the overall
level of agreement among the various data sets is rather poor.
The scatter in the data is primarily associated with difficulties
in preparing, purifying, and handling gas-phase samples of the
transient Cl O molecule and measuring its weak absorption
2 2
cross sections in the actinic region. The scatter in the laboratory
data translates directly into uncertainties in the calculated
-11
3
-1 -1
where k13(T) ) 2.7 × 10 exp(-530/T) cm molecule s ,
-
12
3
-1 -1
k
13(200 K) ) 1.6 × 10 cm molecule s . The formation
of two ClO radicals and loss of Cl O lead to the same overall
2
reaction stoichiometry as obtained if no O atoms were formed.
In subsequent photolysis steps O atoms could possibly react
2 2
with Cl O
atmospheric photolysis rate for Cl
agreement among these studies is in and around the peak of
the Cl spectrum near 245 nm, with differences of <10% in
2 2
O . The highest level of
O + Cl O f products
(14)
2
2
2 2
O
most cases. Note that several of the studies shown in Figure 1
do not include absolute cross section determinations and have
been normalized to the peak cross section recommended by
NASA/JPL. As shown in Figure 1, the largest discrepancies
are in the long wavelength region, λ >300 nm, which is the
region most important for calculating atmospheric photolysis
rates.
where the rate coefficient for this reaction is currently unknown.
The agreement between our results obtained using Cl O (248
nm) and Cl /Cl O (351 nm) photolysis sources indicates the
possible small yield of O atoms in reaction 6 did not influence
the determination of the Cl spectrum significantly.
In addition to the uncertainties associated with the absorption
measurements and spectral analysis, the breakdown of the
2 2
reaction stoichiometry due to loss of ClO or Cl O influences
2
2
2
2
O
2
The studies of Cox and Hayman,15 Burkholder et al., and
9
DeMore and Tschuikow-Roux13 report high Cl
O
cross section
values in the 250-300 nm range and may have been influenced
by spectral impurities, the primary impurity being Cl O, which
was used as a radical precursor, while the Cox and Hayman
2
2
our results. It is difficult to estimate this uncertainty, but it is
probably <10%. The stoichiometry was shown to be valid for
2
Cl where we obtained good agreement for its peak cross section
2
1
5
9
and Burkholder et al. studies also show evidence for the
with the recommended value. Unaccounted for loss of Cl
or ClO would lead to an overestimation of the Cl concentra-
tion and consequently an underestimation of the peak Cl
absorption cross section. Finally, the Cl absorption cross
2 2
O
1
0
possible interference from Cl
2 3
O . For λ >300 nm, Pope et al.
2 2
O
2
O
2
reported the lowest Cl cross section values, while Burkholder
2 2
O
9
2
O
2
et al. has reported the highest values. The differences are large,
nearly a factor of 10 at 350 nm, which results in substantial
and significant differences in the calculated atmospheric pho-
tolysis rates. It is worth noting that the studies prior to Pope et
+0.8
-18
2
section at the peak of the spectrum is 7.6-0.5 × 10
cm
molecule- were the quoted uncertainties are at the 2σ level
1
and include estimated systematic errors.
10
3
.5. Comparison with Previous Studies. There are a number
al. also show discrepancies in this wavelength region, Figure
2 2
1, although the differences are much smaller. The Cl O
9,10,13-16
of published studies of the Cl
2
O
2
UV absorption spectrum,
9
,13,15,17
its absolute cross section values,
and photolysis products
that we can compare with the present
work. A summary of previously reported Cl UV absorption
spectrum obtained in the present work is in excellent agreement
and quantum yields7
,8,18,19
19
with results from Burkholder et al. for λ >300 nm. Chen et al.
2
O
2
used a newly developed experimental technique that combined
pulsed laser photolysis in a molecular beam with mass spec-
spectrum data is given in Figure 1 and summarized in Table 1.
6
The spectrum recommended by NASA/JPL for use in atmo-
2 2
trometric detection of Cl O . Their experiments measured the

UV Absorption Spectrum of the ClO Dimer
J. Phys. Chem. A, Vol. 113, No. 49, 2009 13723
able agreement with the Cl spectrum reported by Pope et
al., although the residual spectrum contains slightly negative
values at the longer wavelengths. The difference in the amount
2
O
2
10
2
of Cl subtraction needed to obtain the residuals that agree with
9
10
the Burkholder et al. and Pope et al. spectra was only 4.5%.
This exercise highlights the need for accurate methods to account
2 2 2
for Cl in the determination of the Cl O spectrum.
10
9
Prior to the Pope et al. study, only Burkholder et al. and
DeMore and Tschiukow-Roux13 had reported absorption cross
section data for wavelengths >320 nm. As shown in Figure 10,
the uncertainty in the magnitude of the Cl
2
corrections can
account for the majority of the discrepancies in the reported
9
Cl
titration method to determine the Cl
measured absorption spectra. They estimated the uncertainty in
their Cl absorption cross section at 330 nm to be in the range
0-15%. DeMore and Tschiukow-Roux corrected their
measured spectra for Cl absorption using a subjective iterative
spectral subtraction approach. They report the uncertainty in
O
2 2
spectra. Burkholder et al. used a discharge flow chemical
2
contribution to their
2 2
O
1
3
1
2
1
0
their Cl
fit their measured absorption spectra, which contained both
Cl and Cl , using two Gaussian profiles and a reference Cl
2
2
O
2
cross section to be ∼30% at ∼350 nm. Pope et al.
2
O
2
2
absorption spectrum. The fitting procedure optimized the
parameters of the Gaussian functions (peak location and width,
Figure 10. Illustration of the sensitivity of the Cl
to the Cl spectral subtraction and the level of agreement that can be
obtained with representative Cl spectra reported in the literature.
The experimental spectrum was recorded using the CT-TD method and
contains a large contribution from Cl as seen by the peak in the
spectrum at 330 nm. An optimized agreement of residual spectra (lines)
obtained by subtracting Cl absorption from the experimental spectrum
with the Cl spectra reported by Burkholder et al. (O), Huder and
2 2
O spectral analysis
2
in wavelength space) and the abundance of Cl
their measurement precision, they estimated the uncertainty in
the Cl
cross section at ∼350 nm to be ∼25%. The absolute
uncertainties in their Cl absorption cross sections in the long
2
. On the basis of
2
O
2
2
2 2
O
2 2
O
2
wavelength region are most likely larger and dependent on the
validity of the assumptions applied in the spectral fitting, which
are difficult to quantify. In such a spectral analysis approach, a
9
2
O
2
1
6
10
DeMore (4), and Pope et al. (0) are shown.
correlation between the UV absorption spectra of Cl
would lead to systematic errors in the Cl spectrum obtained.
In our work, we obtained UV absorption spectra using the CT-
2 2 2
and Cl O
product of the Cl
quantum yield, σCl O2(λ)Φ(λ). The experimental method is
2 2
O absorption cross section and its photolysis
2 2
O
2
independent of Cl
associated with possible Cl
Values of σCl O2(λ)Φ(λ) reported for wavelengths of 308 and
2
and therefore not susceptible to uncertainties
TD method that contained Cl
those reported by Pope et al., who used the same experimental
method. Therefore, the source of the discrepancies in the
2
1
and Cl
2
O
2
that are consistent with
2
impurities as discussed below.
0
2
351 nm are in excellent agreement with the present Cl
2 2
O cross
2 2
reported Cl O spectra between the two studies is not the result
of the spectral measurements themselves but lies in the
interpretation of the measured spectra, as illustrated in Figure
section values, within 7%, when Φ(λ) is assumed to be unity.
The good agreement between the Chen et al. results and those
obtained here provide additional evidence that the methods used
19
1
0. In our work, we have also shown that absorption spectra
in the present work to account for Cl
.5.1. Cl Spectral Interference. A source of uncertainty in
the reported Cl spectra is associated with the uncertainties
in corrections for Cl absorption. Cl is present as an impurity
in all gas-phase Cl spectrum studies reported to date, and
2
are valid.
obtained using the CT-TD method at trapping temperatures
greater than 151 K are consistent with our PLP-DA experiments,
Figure 5. In conclusion, the major source of the discrepancy
between our work and that of Pope et al.10 is attributed to the
3
2
2
O
2
2
2
2
O
2
uncertainty in the Cl
2
subtraction. In the present study our
measured spectra must account for its contribution to obtain
accurate cross section values at λ >300 nm. The point that
2
differences in Cl subtraction alone can account for the majority
of the discrepancies in the Cl O cross section at λ >300 nm is
2 2
illustrated in Figure 10 using experimental data obtained in this
study. In this example, an absorption spectrum containing
methods were designed to quantitatively account for Cl
2
interference by using the observed isosbestic wavelengths,
reaction stoichiometry, and chlorine mass balance.
von Hobe et al.14 reported a Cl
O
absorption spectrum
2
2
measured in a 10 K Ne matrix, shown in Figure 1. Cl samples
2 2
O
were prepared using the CT method, and samples purified by
vacuum distillation were shown to contain negligible amounts
of Cl . The wavelength dependence of the spectrum at λ <300
2
nm is in good agreement with that from this work although the
width of the absorption band is narrower due to the much lower
temperature of the matrix measurement. In the long wavelength
2
significant Cl absorption was taken from our CT-TD experi-
ments. We compared the residual normalized spectra obtained
9
following Cl
intermediate (Huder and DeMore ), and lowest (Pope et al. )
Cl spectra reported in the wavelength range >300 nm.
Subtracting various amounts of Cl on a subjective basis, using
a Cl reference spectrum recorded under identical conditions,
we obtain the Cl spectrum obtained in this work and good
agreement with the Cl spectrum reported by Burkholder et
al. Increasing the amount of Cl subtraction yields reasonable
2
subtraction with the highest (Burkholder et al. ),
1
6
10
2
O
2
2
region, von Hobe et al. report Cl
out to 400 nm, which is consistent with our gas-phase observa-
tions that Cl absorbs in this wavelength region. Their Cl
2 2
O absorption at wavelengths
2
2
O
2
O
2
2 2
O
2
O
2
2
9
-18
spectrum (normalized to a peak cross section of 6.5 × 10
2
2 -1
agreement with the exponential extrapolation given by Huder
cm molecule ) is a factor of 2 lower than the value reported
in this work and a factor of 7 greater than the value reported
by Pope et al.10 at 350 nm.
and DeMore16 although systematic deviations are observed.
2
Increasing the amount of Cl subtraction further yields reason-

1
3724 J. Phys. Chem. A, Vol. 113, No. 49, 2009
Absolute cross section values for Cl have been reported
Papanastasiou et al.
2
O
2
1
5
9
by Cox and Hayman, Burkholder et al., DeMore and
13
17
Tschiukow-Roux, and Bloss et al. The peak absorption cross
section values in these studies are systematically lower than
the value obtained in this work, 7.6 +- 0.5 × 10 cm molecule ,
0.8
-18
2
-1
although all the reported values fall within the combined 2σ
17
error limits. The cross sections reported by Bloss et al. and
Chen et al.19 (assuming a unity quantum yield) are in excellent
agreement, within 7%, with the values obtained in this work.
2 2
For the actinic region, the Cl O absorption spectrum reported
in this work is in good agreement with spectra reported by
9
19
Burkholder et al. and the values reported by Chen et al. for
08 and 351 nm. These studies used different experimental
techniques and obtained consistent Cl cross section data. Our
spectrum is systematically larger than that reported by DeMore
3
2 2
O
13
and Tschiukow-Roux although the two data sets overlap within
the combined uncertainties. This work and the studies of
9
19
Burkholder et al., Chen et al., and DeMore and Tschiukow-
Roux13 report Cl
absorption cross section values obtained
using different experimental techniques that are significantly
2
O
2
1
0
greater than those reported in the Pope et al. study.
4
. Atmospheric Implications
Figure 11. Comparison of wavelength-dependent Cl
photolysis rate coefficients, J(λ), calculated for a solar zenith angle
UV absorption
cross obtained in this work (heavy black) where the upper and lower
limits of the shaded region were calculated using the estimated
2 2
O atmospheric
In the winter/spring Antarctic polar vortex, the ClO dimer
catalytic ozone destruction cycle, reactions 1-4, accounts for
(
2 2
SZA) of 86° at an altitude of 20 km using the Cl O
4
the majority of calculated ozone depletion. In the Arctic polar
2 2
stratosphere, the photochemistry of Cl O is also important
although the extent of ozone depletion is less due to the
meteorological variability in the winter vortex. A critical
parameter in determining the efficiency of the ClO dimer cycle
uncertainty limits for the Cl
and shown in Figure 9 (see text for details of the error analysis). J(λ)
values calculated using previously reported Cl cross section data
2 2
O cross section data reported in this study
2
O
2
9
6
are included for comparison: Burkholder et al. (blue), NASA/JPL
is the atmospheric photolysis rate coefficient, J, of Cl
2 2
O which
1
0
(
red), and Pope et al. (green). Solar fluxes were calculated using the
is calculated from the product of the Cl absorption cross
2
O
2
21
online NCAR TUV calculator.
section, its photolysis quantum yield, and the solar flux
integrated over all wavelengths
region longer than the experimentally available values an
exponential extrapolation was used as shown in Figure 9. The
wavelength-dependent solar fluxes were obtained using the
NCAR online TUV calculator.21 J(λ) has a maximum between
J )
∫
J(λ) dλ )
∫
σ(λ)Φ(λ)Ψ(λ,Z,ꢀ) dλ
(VI)
3
30 and 340 nm and falls to nearly zero at 300 nm. The absolute
magnitude of J(λ) and the decrease toward longer wavelength
differ for the different Cl cross section data sets as expected.
where σ(λ) is the absorption cross section of Cl
λ, Φ(λ) is the photolysis quantum yield of Cl
O
2 2
at wavelength
2
2
O , and Ψ(λ,Z,ꢀ)
2 2
O
is the solar flux which is a function of wavelength, altitude (Z),
and solar zenith angle (SZA, ꢀ). It is worth noting that the
efficiency of the ClO dimer cycle also influences the efficiency
of the ClO + BrO catalytic ozone destruction cycle
10
The Pope et al. cross section data set results in the lowest
maximum value of J(λ). The data sets from this work and
9
Burkholder et al. yield similar magnitude J(λ) values while
6
the NASA/JPL values have a similar wavelength dependence
but are systematically lower.
J(λ >350 nm) is significant for the Cl O cross section data
2 2
from this work, Burkholder et al., and NASA/JPL. J(λ >420
nm), i.e., beyond the limit of our experimental data, makes a
non-negligible, ∼10-30%, contribution to the total photolysis
rate coefficient. The long wavelength contribution depends on
the SZA and will also depend on the subjective method used to
Cl + O f ClO + O
(4)
(15)
(16)
3
2
9
6
Br + O f BrO + O
3
2
ClO + BrO f Cl + Br + O2
NET 2O f 3O
10
2 2
extrapolate the cross section data. Pope et al. report Cl O cross
section data that decreases significantly at wavelengths greater
than 300 nm with reported values extending out to 360 nm.
Including an extrapolation of their cross section data to longer
wavelengths increases the calculated total photolysis rate
coefficient only slightly. The differences in the photolysis rate
coefficients calculated for the cross section data from Pope et
3
2
by altering the atmospheric abundance of ClO.
J(λ) values calculated for a SZA of 86° and altitude of 20
km, representative conditions for stratospheric measurements
10
of ClO and Cl
2
O
2
in the Arctic, with the Cl
2
O
2
cross section
al. and this work have significant implications for atmospheric
model calculated ClO abundance and ozone loss rates.
The uncertainties in the Cl absorption cross sections from
data from this work and from several previous studies (Burkhold-
x
9
6
10
er et al., NASA/JPL, and Pope et al. ) is given in Figure 11.
These studies were chosen because they represent the highest,
2 2
O
this work, as a function of wavelength, are shown in Figure 9.
We extended the estimated uncertainty limits beyond 420 nm
using an exponential extrapolation, shown in Figure 9, in order
currently recommended, and lowest Cl
2 2
O cross section data
available for the long wavelength region. For the wavelength

UV Absorption Spectrum of the ClO Dimer
J. Phys. Chem. A, Vol. 113, No. 49, 2009 13725
using the present Cl
spectrum and that of Burkholder et al.9
2
O
2
agree very well, within ∼10%, which is within the combined
uncertainties of the Cl cross section values for the two
2
O
2
studies. The J values from the present work are ∼30% higher
than NASA/JPL and a factor of 9 greater than those obtained
using the Pope et al.10 data for SZA < 80°. At larger SZAs the
6
10
differences between this work and NASA/JPL and Pope et al.
are greater. These differences have a significant impact on
atmospheric model calculated ClO abundance and ozone loss
rates as shown in previous atmospheric model studies of polar
x
stratospheric halogen chemistry.4
J(SZA) values, shown in Figure 12, increases with SZA from
50%/-20% at 86° to a maximum at ∼92°.
The present Cl absorption cross section data are in close
,22,23
The uncertainty in the
+
2
O
2
agreement with the data previously reported by Burkholder et
9
al. for λ >300 nm. We point out the similarity with the previous
9
Burkholder et al. study primarily because previous modeling
studies of polar stratospheric in situ and remote sensing
9
measurements have frequently included the Burkholder et al.
2 2
Cl O cross section data in their analysis. Therefore, we can
draw meaningful conclusions here from the previous model
9
studies by a direct comparison with the Burkholder et al. cross
section data. von Hobe et al.23 recently provided a summary of
ClO
analysis using different photochemical input parameters includ-
ing the Cl absorption cross section data from Burkholder et
al. The available ClO field measurement data sets, unfortu-
nately, do not provide a consistent enough picture of ClO in
the Arctic to make an evaluation of Cl cross section data
from field measurements alone. However, von Hobe et al.
concluded that the “best” agreement between model and
observations was obtained for Cl photolysis rates falling
x
Arctic field measurements and an atmospheric model
Figure 12. The upper frame shows the integrated atmospheric
photolysis rate coefficients, J, calculated for Cl O as a function of
2 2
solar zenith angle (SZA) using the UV absorption cross section data
from this work (black). The upper and lower limits of the shaded region
2
O
2
9
x
were calculated using the estimated uncertainty limits for the Cl
cross section data reported in this study and shown in Figure 9 (see
text for details of the error analysis). J values calculated using the Cl
2
O
2
x
2
O
2
2
O
2
23
9
6
cross section data from Burkholder et al. (blue), NASA/JPL (red),
and Pope et al. (green) are included for comparison. Solar fluxes were
calculated using the online NCAR TUV calculator. The lower frame
shows the same data relative to the values obtained using the NASA/
10
2
1
2 2
O
9
6
between the Burkholder et al. and the NASA/JPL recom-
mendations, while using the Pope et al.10 Cl
data resulted in very poor agreement. If the Pope et al. cross
section data are correct, it would therefore require including
unknown chemical processes to explain the discrepancies
cross section
JPL recommended Cl
2
O
2
cross section data.
2 2
O
1
0
2 2
to quantify the impact on the overall Cl O atmospheric
photolysis rate coefficient. The J(λ) values obtained using the
upper and lower limits of the cross section data in eq VI are
shown in Figure 11. As expected, the uncertainty in J(λ >420
between models and observations. In another study, Frieler et
4
al.
calculated ozone loss in the Arctic and Antarctic vortices
nm) is large due to the large uncertainty in the Cl
cross section at these wavelengths. It is also worth noting that
including the extrapolated Cl cross section data as shown
O
2 2
absorption
using various photochemical input parameters. They obtained
good quantitative agreement with field observations when using
9
2
O
2
Cl
2
2
O photolysis rates calculated using the Burkholder et al.
in Figure 9 leads to asymmetric uncertainty limits in J(λ). For
a SZA of 86°, shown in Figure 11, J(λ >420 nm) accounts for
data combined with increased levels of Br
x
loading, ∼6 ppt.
∼
23% of the total photolysis rate coefficient. The range of
5
. Concluding Remarks
2 2
uncertainty in J(λ) using the Cl O cross section data reported
here is approximately +50%/-20% at the 2σ confidence level.
The estimated uncertainty in J(λ <420 nm) is (18% at the 2σ
confidence level. Thus, our estimated lower limit for the
photolysis rate coefficient is only slightly dependent on the
2 2
The UV absorption spectrum of Cl O was measured in the
present study over the wavelength range 200-420 nm at
200-228 K. The present study was designed to quantitatively
account for Cl
isosbestic wavelengths, reaction stoichiometry, and chlorine
mass balance. The Cl absorption cross section at the peak
of the spectrum, 244.5 nm, was determined to be 7.6-0.5 × 10
2
spectral interference by using the observed
extrapolation of the Cl
2
O
2
cross section data to wavelengths
>
420 nm. The estimated upper limit for J is, however, dependent
2 2
O
+0.8
-18
on the extrapolated Cl
2
O
2
cross section data. The extrapolation
2
-1
method used here most likely yields a maximum to the upper
limit of the Cl cross section values for wavelengths >420
nm. Including the upper limit extrapolated Cl cross section
data increases the overall photolysis rate by ∼50%. Reduction
of the uncertainty in the Cl atmospheric photolysis rate
coefficient below the 50% level requires improved knowledge
of the Cl absorption cross sections and photolysis quantum
cm molecule , which is ∼17% larger than that currently
6 20
2
O
2
recommended by NASA/JPL and IUPAC. The cross sections
in the long wavelength region of the spectrum agree reasonably
well with the values reported by Burkholder et al. but are in
2
O
2
9
2
O
2
poor agreement with those reported in the recent study by Pope
et al.10 The Cl
2 2
O absorption spectrum and cross section values
2
O
2
obtained in this study yield atmospheric photolysis rate coef-
ficients that are of similar magnitude to those calculated using
yields at wavelengths >420 nm.
Figure 12 shows the photolysis rate coefficients, J, integrated
over the wavelength range 290-725 nm as a function of SZA
9
the data from Burkholder et al. The stratospheric photolysis
rate coefficients obtained from this work are ∼30% greater than
6
for Cl
2
9
O
2
cross section data from the present work, Burkholder
those obtained using the NASA/JPL recommended Cl
O
2 2
cross
6
10
et al., NASA/JPL, and Pope et al. The J values obtained
section values. The present results provide strong evidence that

1
3726 J. Phys. Chem. A, Vol. 113, No. 49, 2009
Papanastasiou et al.
the recent study by Pope et al.10 underestimated the Cl
O
2
(6) Sander, S. P.; Friedl, R. R.; Golden, D. M.; Kurylo, M. J.; Moortgat,
G. K.; Keller-Rudek, H.; Wine, P. H.; Ravishankara, A. R.; Kolb, C. E.;
Molina, M. J.; Finlayson-Pitts, B. J.; Huie, R. E.; Orkin, V. L. Chemical
kinetics and photochemical data for use in atmospheric studies; JPL Pub.
06-2, Jet Propulsion Laboratory: Pasadena, CA, 2006; Vol. Evaluation
Number 15.
2
absorption in the wavelength region 300-400 nm, the region
most important for atmospheric photolysis rate calculations. It
1
0
has been suggested that if the Pope et al. results were correct,
then significant gaps in our current understanding of polar
stratospheric chemistry and ozone depletion mechanisms exist.
Although improvements in our fundamental understanding of
(7) Molina, M. J.; Colussi, A. J.; Molina, L. T.; Schlindler, R. N.; Tso,
T. L. Chem. Phys. Lett. 1990, 173, 310.
(8) Plenge, J.; Flesch, R.; K u¨ hl, S.; Vogel, B.; M u¨ ller, R.; Stroh, F.;
R u¨ hl, E. J. Phys. Chem. A 2004, 108, 4859.
2 2
the photochemistry of Cl O are still desired, our work indicates
(
9) Burkholder, J. B.; Orlando, J. J.; Howard, C. J. J. Phys. Chem.
990, 94, 687.
10) Pope, F. D.; Hansen, J. C.; Bayes, K. D.; Friedl, R. R.; Sander,
S. P. J. Phys. Chem. A 2007, 111, 4322.
11) Stratospheric Processes and Their Role in Climate (SPARC), The
that major revisions in current atmospheric chemical mecha-
nisms are not required to simulate observed polar ozone
depletion.
1
(
(
Acknowledgment. We thank SPARC for organizing the “The
Role of Halogen Chemistry in Polar Stratospheric Ozone
Depletion” workshop and supporting discussion on this topic.
This work was supported in part by NOAA’s Climate Goal and
in part by NASA’s Atmospheric Composition, Upper Atmo-
spheric Research Program.
12) Papanastasiou, D. K.; Burkholder, J. B. In preparation.
Chem. Chem. Phys. 2009, 11, 1571.
2 2
Supporting Information Available: Tabulated Cl O ab-
sorption cross section data obtained in this work, parameteriza-
tion of the wavelength-dependent estimated uncertainties in the
2 2
Cl O absorption cross section values, and simulations of
isosbestic point behavior obtained using a chemical reaction
mechanism. This material is available free of charge via the
Internet at http://pubs.acs.org.
(
(
15) Cox, R. A.; Hayman, G. D. Nature 1988, 332, 796.
16) Huder, K. J.; DeMore, W. B. J. Phys. Chem. 1995, 99, 3905.
(17) Bloss, W. J.; Nickolaisen, S. L.; Salawitch, R. J.; Friedl, R. R.;
Sander, S. P. J. Phys. Chem. A 2001, 105, 11226.
(
18) Moore, T. A.; Okumura, M.; Seale, J. W.; Minton, T. K. J. Phys.
Chem. A 1999, 103, 1691.
(19) Chen, H.-Y.; Lien, C.-Y.; Lin, W.-Y.; Lee, Y. T.; Lin, J. J. Science
2
009, 324, 781.
20) Ammann, M.; Atkinson, R.; Cox, R. A.; Crowley, J.; Hynes, R.;
(
Jenkin, M. E.; Mellouki, W.; Rossi, M. J.; Troe, J.; Wallington, T.
International Union of Pure and Applied Chemistry http://www.iupac-
kinetic.ch.cam.ac.uk/index.html, 2008.
References and Notes
(
1) Scientific Assessment of Ozone Depletion: 2006, Global Ozone
(
21) Madronich, S. Tropospheric Ultraviolet and Visible (TUV) Radia-
tion Model; http://cprm.acd.ucar.edu/Models/TUV/, 2009.
22) Kawa, S. R.; Stolarski, R. S.; Newman, P. A.; Douglass, A. R.;
Research and Monitoring Project; World Meteorological Organization:
Geneva, URL http://www.wmo.int/pages/prog/arep/gaw/ozone_2006/ozone_
asst_report.html.
(
Rex, M.; Hofmann, D. J.; Santee, M. L.; Frieler, K. Atmos. Chem. Phys.
Discuss. 2009, 9, 13327.
(
(
2) Solomon, S. ReV. Geophys. 1999, 37, 275.
3) Newman, P. A.; Nash, E. R.; Kawa, S. R.; Montzka, S. A.;
(
23) von Hobe, M.; Salawitch, R. J.; Canty, T.; Keller-Rudek, H.;
Moortgat, G. K.; Grooss, J.-U.; M u¨ ller, R.; Stroh, F. Atmos. Chem. Phys.
007, 7, 3055.
24) Basco, N.; Hunt, J. E. Int. J. Chem. Kinet. 1979, 11, 649.
Schauffler, S. M. Geophys. Res. Lett. 2006, 33, L12814.
4) Frieler, K.; Rex, M.; Salawitch, R. J.; Canty, T.; Streibel, M.;
(
2
Stimpfle, R. M.; Pfeilsticker, K.; Dorf, M.; Weisenstein, D. K.; Godin-
(
Beekmann, S. Geophys. Res. Lett. 2006, 33, L10812.
(
5) Molina, L. T.; Molina, M. J. J. Phys. Chem. 1987, 91, 433.
JP9065345