Chlorine-catalyzed ozone destruction: Cl atom production from ClOOCl photolysis

Article abstract of DOI:10.1021/jp9053204

Recent laboratory measurements of the absorption cross sections of the ClO dimer, ClOOCl, have called into question the validity of the mechanism that describes the catalytic removal of ozone by chlorine. Here we describe direct measurements of the rate-d

Full text of DOI:10.1021/jp9053204

J. Phys. Chem. A 2009, 113, 14099–14108
14099
Chlorine-Catalyzed Ozone Destruction: Cl Atom Production from ClOOCl Photolysis
,†
David M. Wilmouth,* Thomas F. Hanisco,* Richard M. Stimpfle, and James G. Anderson
Department of Chemistry and Chemical Biology, HarVard UniVersity, Cambridge, Massachusetts, 02138
ReceiVed: June 5, 2009; ReVised Manuscript ReceiVed: September 10, 2009
Recent laboratory measurements of the absorption cross sections of the ClO dimer, ClOOCl, have called into
question the validity of the mechanism that describes the catalytic removal of ozone by chlorine. Here we
describe direct measurements of the rate-determining step of that mechanism, the production of Cl atoms
from the photolysis of ClOOCl, under laboratory conditions similar to those in the stratosphere. ClOOCl is
formed in a cold-temperature flowing system, with production initiated by a microwave discharge of Cl
photolysis of CF Cl . Excimer lasers operating at 248, 308, and 352 nm photodissociate ClOOCl, and the Cl
atoms produced are detected with time-resolved atomic resonance fluorescence. Cl , the primary contaminant,
2
or
2
2
2
is measured directly for the first time in a ClOOCl cross section experiment. We find the product of the
quantum yield of the Cl atom production channel of ClOOCl photolysis and the ClOOCl absorption cross
-
20
section, (φσ)ClOOCl ) 660 ( 100 at 248 nm, 39.3 ( 4.9 at 308 nm, and 8.6 ( 1.2 at 352 nm (units of 10
2
-1
cm molecule ). The data set includes 468 total cross section measurements over a wide range of experimental
conditions, significantly reducing the possibility of a systematic error impacting the results. These new
measurements demonstrate that long-wavelength photons (λ ) 352 nm) are absorbed by ClOOCl directly,
producing Cl atoms with a probability commensurate with the observed rate of ozone destruction in the
atmosphere.
Introduction
2ClO + M a ClOOCl + M
ClOOCl + hV f 2ClO
Net: null
(1a)
(2b)
On the global scale, with little doubt, the most successful
intersection of fundamental science and public policy is the case
linking ground-level chlorofluorocarbon (CFC) release with the
chlorine-catalyzed loss of ozone in the Earth’s stratosphere. This
scientific case led to the Montreal Protocol, with its subsequent
London and Copenhagen amendments, which controls the global
release of CFCs. The scientific case was built on what was
thought to be an unequivocal combination of laboratory experi-
ments and in situ and remote observations of the stratosphere.1
ClOOCl is the centerpiece of the catalytic cycle (Figure 1a)
that accounts for more than 50% of the chlorine-catalyzed ozone
loss in the Arctic and Antarctic stratosphere every spring. In
this mechanism, the photolysis of ClOOCl to produce Cl atoms
is the rate-determining step for the removal of O
ClO + M a ClOOCl + M
The photolysis rates of ClOOCl in the atmosphere are directly
related to the absorption cross sections. Recently, Pope et al.
4
reported dramatically smaller absorption cross sections for
ClOOCl in the critical wavelength region between 300 and 400
nm. Taken alone, this result would indicate an unknown
mechanism must be responsible for the dramatic ozone loss
observed in the Arctic and Antarctic every spring. However,
when considered along with the body of laboratory work both
prior and since its publication, this study has instead served to
2
3
3
by chlorine:
2
(1a)
ClOOCl + hV f Cl + ClOO
(2a)
ClOO + M f Cl + O + M
(3)
2
2
(C1 + O f ClO + O )
(4)
(5)
3
2
Net: 2O f 3O
3
2
Critical in this cycle is the production of two Cl atoms from
the photolysis of ClOOCl. If the photolysis reaction produced
ClO, a null cycle would result, and O would not be destroyed:
3
Figure 1. Catalytic removal of ozone by chlorine. (a) Chlorine released
from CFCs removes ozone as it forms ClO. The subsequent formation
of ClOOCl is the critical point in the cycle: thermal decomposition
back to ClO short circuits the cycle, resulting in no net ozone loss,
while photolysis with relatively abundant long-wavelength photons (λ
*
To whom correspondence should be addressed. D.M.W.: e-mail,
wilmouth@huarp.harvard.edu; phone, 617-495-5922; fax, 617-495-4902.
T.F.H.: e-mail, thomas.hanisco@nasa.gov; phone, 301-614-6598.
>
320 nm) completes this cycle every few minutes. (b) The laboratory
†
Currently at NASA Goddard Space Flight Center, Greenbelt, MD
equivalent uses multiple excimer lasers to photolyze a continuous flow
of ClOOCl at stratospheric conditions.
2
0771.
1
0.1021/jp9053204  2009 American Chemical Society
Published on Web 10/23/2009

1
4100 J. Phys. Chem. A, Vol. 113, No. 51, 2009
Wilmouth et al.
Figure 2. Schematic of the experimental apparatus. We photodissociate ClOOCl in a fast-flow system with excimer lasers operating at 248, 308,
and 352 nm, detecting Cl atoms with atomic resonance fluorescence and Cl with nonresonant molecular fluorescence.
2
highlight the inconsistency in measurements of the absorption
cross section using traditional absorption spectroscopy tech-
niques.5
is approximately 60% higher at 250 K than the current JPL
7
panel recommendation (hereafter JPL-06) and a factor of 18
4
higher than that of Pope et al. Clearly, resolving the discrepancy
Virtually all past laboratory measurements of the ultraviolet
cross sections of ClOOCl utilized absorption spectroscopy. The
primary challenge with this technique is that a number of other
absorbing species, either added or produced when chemically
forming ClOOCl, may be present in the absorption cell. In
particular, the uncertainty in correcting ClOOCl spectra for the
in reported values at this atmospherically relevant wavelength
is of substantial importance in quantifying the role of ClOOCl
in stratospheric ozone loss.
Experimental Section
A diagram of the laboratory apparatus is shown in Figure 2.
The main flow tube consists of a 7.6 cm × 7.6 cm square duct,
approximately 2 m in length. Source gases and carrier gases
are added to the right of the flow system as shown, with the
direction of flow from right to left. The entire flow system
resides inside a cooled, insulated box to aid in achieving low-
temperature operation. Cooling originates in three locations: (1)
2
presence of Cl , which has an unstructured absorption signal
that peaks at approximately 330 nm, has led to reported ClOOCl
cross sections that differ by more than an order of magnitude
in some of the most atmospherically relevant wavelength
5
regions.
Clearly, much of the confidence in the Molina mechanism3
is based on the historically good agreement between model
calculations and atmospheric observations of ozone loss, not
on the solidity of the laboratory data set. The fundamental
question remains: What is the photolytic efficiency to produce
chlorine atoms from ClOOCl directly at atmospherically relevant
conditions?
2
the main N carrier gas is cooled before entering the flow system
by passing through a copper coil submerged in a dry ice/ethanol
bath, (2) a Neslab chiller pumps cold ethanol through copper
tubing attached to the metal flow tube walls, and (3) the air
inside the insulated box is cooled with an attached liquid
nitrogen tank. Pressure transducers are located at the center of
the flow tube, and temperature readings are taken inside the
flow system at the front and rear, along with a number of other
thermistors deployed throughout the insulated box. The observed
temperature gradient along the flow system is typically less than
To address this question, we designed an experiment to couple
the photolysis of ClOOCl at three excimer laser wavelengths
(
248, 308, and 352 nm) with direct, time-resolved detection of
chlorine atoms. We measure the photodissociation cross section
of ClOOCl for the photolysis pathway that produces Cl atoms,
i.e., the product of the ClOOCl absorption cross section and
the quantum yield of reaction 2a. This measured quantity is
henceforth referred to simply as the photodissociation cross
section, with the understanding that it is only a measure of
photolysis channel 2a and does not include 2b. As defined and
measured here, the photodissociation cross section is the quantity
of greatest importance for the rate-limiting step of the ClOOCl
catalytic cycle. Our experiments are conducted at conditions
relevant to the atmosphere in a temperature-controlled fast-flow
system that is designed to mimic the essential elements of the
Molina mechanism (Figure 1b).
4
K.
ClOOCl Source Chemistry. We employ two methods of
source chemistry to generate ClOOCl. Our primary method is
passing Cl /He through a microwave discharge to form Cl atoms
eq 6), followed by reaction with O to form ClO (eq 4), which
subsequently self-reacts at low temperatures to form ClOOCl
eq 1a).
2
(
3
(
Cl f 2Cl
(6)
2
This approach has the scientific advantage that the production
of Cl from ClOOCl photolysis is measured directly, thus
providing an unambiguous laboratory measurement of the rate-
limiting step in the ClOOCl mechanism. Moreover, this ap-
proach has the technical advantage of having detection sensi-
tivity to Cl atoms in the parts per trillion range. Importantly,
2 2
Our second method of ClOOCl formation uses CF Cl
photolysis as the source of Cl atoms to react with O :
3
8
CF Cl + hV
f CF Cl + Cl
(7)
2
2
172nm
2
2
Cl is measured for the first time in a study of the cross section
of ClOOCl, here with a nonresonant molecular fluorescence
technique developed for this experiment.
2 2
Photolysis of CF Cl is accomplished using a high-intensity 172-
nm xenon excimer lamp. There is good overlap between the
emission of the Xe lamp and the ultraviolet cross sections of
CF Cl in this wavelength region. Neither CF
fragment and its reaction products (e.g., CF ClO
with O ) affect the measurements, as Cl atom production was
not observed from these molecules at our laser wavelengths.
Since the completion of our laboratory work, measurements
of the ClOOCl cross section at 308 and 351 nm have been
2
9
2
2
Cl
2 2
2
nor the CF Cl
6
reported by Chen et al. using molecular beams with mass
2
2
from reaction
detection rather than absorption spectroscopy. This study reports
the highest 351-nm ClOOCl cross section to date, a value that
2

Cl Atom Production from ClOOCl Photolysis
J. Phys. Chem. A, Vol. 113, No. 51, 2009 14101
In addition to the primary channel for the ClO self-reaction
that forms ClOOCl (eq 1a), several competing bimolecular
reactions can also occur:
2
ClO f Cl + O
(1b)
2
2
2
2
ClO f ClOO + Cl
ClO f OClO + Cl
(1c)
(1d)
The formation of ClOOCl is substantially faster than the
bimolecular reactions at our typical cold-temperature conditions,
e.g., approximately 60 times faster at 240 K. Moreover, of the
bimolecular channels, only reaction 1d has the potential to affect
the experiment. Reaction 1b forms Cl , which we measure, and
2
reaction 1c is effectively a null reaction, as ClO is re-formed in
the source tube via reactions 3 and 4.
Figure 3. Absorption cross sections of ClOOCl from JPL-06 and Pope
4
et al., along with JPL-06 cross sections of ClO and Cl
2
shown for
reference. The excimer laser wavelengths are shown as vertical gray
lines.
We achieve a typical ClO
microwave discharge source and less than 1% using the
photolysis lamp source. Photolyzing CF Cl results in signifi-
cantly smaller Cl concentrations relative to ClOOCl than does
passing Cl through the microwave discharge.
x
yield of 40-50% with the
wavelengths indicated. The JPL-06 cross sections for ClO and
Cl are also shown for reference. As is evident in Figure 3,
2
2
2
2
establishing the ClOOCl cross section at several select wave-
lengths is sufficient to resolve the large uncertainty surrounding
the true value, especially at the atmospherically relevant longer
wavelengths. Also evident in Figure 3, however, is that we do
not photodissociate only ClOOCl at these wavelengths. Our
2
The reactions of the photolysis source take place entirely
inside a 75-cm long, 2.2-cm diameter UV-grade fused silica
(
Suprasil) tube embedded in the center of the main flow duct,
as shown in Figure 2. The chemistry of the microwave discharge
source also primarily occurs inside the Suprasil tube after being
initiated in a fused silica side arm. Using a smaller diameter
source region relative to the main duct enables higher ClO
concentrations to be realized, which is essential for significant
ClOOCl formation.1
measurement of Cl
Here we simply note that the cross section of Cl
2
and correction for ClO are addressed below.
is insignificant
2
at 248 nm, and the cross section of ClO is insignificant at 352
nm, simplifying any correction to the observed Cl signal.
The laser light passes through the duct via the diagonal ports
at the front and back sides of the first detection axis, as shown
in Figure 2. Multiple apertures and knife-edged baffles are used
to reduce scatter from the lasers. The optical setup ensures that
the laser beams are well aligned and underfill the detection
volume. Laser power is measured at the exit port of the flow
duct. The lasers are typically operated at 75 Hz and 7.5 mJ
0,11
At the end of the Suprasil tube, the source gases mix with a
-
1
3
0 standard-L min
duct, diluting all mixing ratios from the source by a factor of
40 and essentially preventing any further source chemistry
2
N carrier gas in the larger 7.6-cm square
∼
from taking place. Diluting the source flow in the main duct
provides for low Cl concentrations in the detector region, which
is necessary for accurate resonance fluorescence detection. Our
-
1
pulse .
There are two detection axes on the flow system, the first for
measuring Cl atoms and the second for measuring Cl , as shown
“
6
standard” experimental conditions include a duct pressure of
5 mbar, flow velocity in the duct of 110 cm s , and
-
1
2
temperature of 240 K. All experimental parameters were
adjusted over a wide range of values during the course of the
experiments.
in Figure 2 and, in detail, in Figure 4. Cl detection at the first
axis is accomplished using the well-established atomic resonance
fluorescence technique.1 The detection axis consists of a radio
frequency-powered chlorine resonance lamp light source with
a photomultiplier tube (KBr photocathode) detector at a right
2,13
Cl
added from a 5% CF
from a 10% O /Ar cylinder through an electric discharge to
form approximately 5% O /O . Some experiments were per-
formed in the absence of O by trapping O on silica gel.
2
is added from a 0.25% Cl
2
/He cylinder, and CF
2 2
Cl is
2
2
Cl /He cylinder. O
3
is formed by passing
O
2
2
2
angle. Gas filter cells fitted with MgF windows and pressurized
with air are attached on both the lamp and PMT to isolate the
118.9-nm Cl lines. The photolysis lasers intersect the volume
illuminated by the lamp and imaged by the PMT to maximize
Cl fluorescence signal, as Cl is removed rapidly by the Cl +
3
2
2
3
Photolysis and Detection. Photolysis of ClOOCl in the flow
system is accomplished using two copropogating excimer lasers
operating with KrF (EX5, GAM Laser) and either XeCl or XeF
fill gases (EX50, GAM Laser). The spectral output of the lasers
was measured with a diode array spectrometer. The KrF laser
output is centered at 248.43 nm and is approximated by a
Gaussian function that is 0.5 nm wide (fwhm). The XeCl laser
output is centered at 308.35 nm and is 0.9 nm fwhm. The XeF
laser has two lines at 351 and 353 nm that are both 0.5 nm
fwhm with an intensity ratio of 3:2. The average intensity-
weighted wavelength of the XeF laser is ∼351.8 nm. The laser
wavelengths are rounded and stated as 248, 308, and 352 nm
herein.
O
3
reaction.14
The Cl signal from the PMT is monitored with a multiscalar
analyzer (Stanford Research Systems SR430). The signals are
averaged into 50-µs bins and averaged for 3000 laser shots.
The ability to resolve the fluorescence signal temporally is
essential to isolating the prompt Cl atom signal from the
photolysis of ClOOCl from any slower secondary processes
affecting the Cl signal after photolysis.
The second Cl lamp axis is for Cl
molecular fluorescence.15 The measurement of Cl
for correcting its contribution to the Cl atom signal at 308 and
352 nm, where the Cl cross section is significant. The axis is
operated with a MgF window on the Cl lamp side, a sapphire
2
detection via nonresonant
2
is required
Figure 3 shows the absorption cross sections of ClOOCl from
2
4
JPL-06 and Pope et al. with the relevant excimer laser
2

1
4102 J. Phys. Chem. A, Vol. 113, No. 51, 2009
Wilmouth et al.
Figure 5. Signal from Cl
2
detected with Cl atom fluorescence after
photolysis at 352 nm and with molecular fluorescence. The concentra-
1
2
-3
tion of Cl
2
is varied between 0 and 5.9 × 10 molecules cm
.
Conditions are P ) 65 mbar and T ) 235 K. The laser power is 8 mJ
-
1
pulse at 75 Hz.
1
2601/2 ) 34 in 1-s integration, and S/N ) 230 in the 45-s
integration time used for these experiments. The detection limit
9
-3
for S/N ) 1 in 45 s is [Cl
2
] ) 5.8 × 10 molecules cm . The
Cl atomic detection is more sensitive, with S/N ) 560 in 45-s
integration; however, we use time-resolved detection with this
method. For each 50-µs bin, the S/N ) 23 for 45-s integration
Figure 4. Detection geometry and signals. (a) Configuration of the
excimer laser beams at the upstream Cl detection axis. The laser beams
and Cl atom detection cross in the center of the flow tube. Cl
measured downstream via molecular fluorescence using a similar
detection axis. (b) Signal from the Cl detector with the microwave
discharge source on and off. The difference in Cl signal is proportional
to the amount of Cl converted to ClOOCl. (c) Time-resolved Cl atom
2
is
2
at a laser repetition of 75 Hz. Thus, for both Cl and Cl , the
2
experiment yields excellent signal-to-noise ratios, which is
important for identifying potential systematic errors.
2
2
Just downstream of the Cl
of the flow in the duct is continuously extracted (Figure 2) and
passed through a multipass absorption cell to measure the O
3
2
detection axis, a small fraction
fluorescence resulting from photolysis of ClOOCl at 248, 308, and 352
nm. Signals are reported in 50-µs time bins and are normalized to the
-
1
number of photons for 10 mJ pulse at 352 nm. The signal due to Cl
2
photolysis has been removed using the signal from panel b. The
intercepts at time ) 0 (open circles) are determined from polynomial
fits to the data.
concentration. An absorption path length of 600 cm is obtained
with 20 passes of a 30-cm White cell. A mercury lamp at 254
nm is used as the light source, and a UV vacuum diode serves
as the detector.
window on the PMT side, and N
2
in the filter cells. Multiple Cl
2 3
Concentrations of ClO, ClOOCl, OClO, Cl , and O are
emission lines between 130 and 140 nm are used to excite the
measured periodically using UV absorption spectroscopy along
the length of the source tube between 200 and 400 nm. The
spectrometer is an Ocean Optics USB4000 with 0.7-nm
bandwidth, fiber-coupled to the front end of the flow system,
as shown in Figure 2. The light source is a Hamamatsu L6302
deuterium lamp located at the end of the duct.
Calculating ClOOCl Cross Sections. A key strategy in this
study is that photodissociation cross sections of ClOOCl are
determined by two independent methods. In method 1, the
observed signal from ClOOCl at 308 or 352 nm is referenced
to the observed signal and known cross section of ClOOCl at
1
+
1
+
16
Cl
2
1 Σ
u
- X Σ
g
Rydberg series near 135 nm. The red-
17
shifted fluorescence of Cl
2
between 150 and 180 nm is detected
with near zero background by using the PMT filtered by a
sapphire window that only transmits >150 nm. The sapphire
window completely blocks all atomic Cl, O, and H fluorescence
signals, allowing for a selective and unambiguous measurement
of Cl
2
.
We verified experimentally that the Cl
2
detector signal is
unaffected by any other chemical species either added to the
flow system or produced when the source is on. Moreover,
because only very small fractions of ClOOCl and Cl
photolyzed by the laser at 308 or 352 nm, the Cl molecular
fluorescence measurement is unaffected when the laser is on at
these wavelengths. Both the Cl and Cl detectors were calibrated
2
are
2
48 nm:
2
λ
S
S
2
λ
λref
ClOOCl
ClOOCl
(φσ)
) (φσ)
(8)
ClOOCl
with absorption at 118.9 nm across the duct and at 330 nm along
the length of the source, respectively. While not used in the
determination of the ClOOCl cross sections, these calibration
runs are useful for scaling observed signals to concentrations,
which can be compared with model results.
λref
ClOOCl
where λ is 308 or 352 nm and λref is 248 nm; (φσ)ClOOCl is the
ClOOCl photodissociation cross section, i.e., the product of
the quantum yield of the Cl production pathway (eq 2a) and
the absorption cross section of ClOOCl at the indicated
wavelength; and SClOOCl is the observed Cl atom signal from
ClOOCl photolysis, normalized for the number of laser photons
at that wavelength.
The ability to simultaneously detect Cl and Cl
sensitivity is an essential aspect of the experiment. An example
of the relative sensitivity to Cl of the molecular fluorescence
axis and the Cl atomic fluorescence axis after photolysis at 352
2
with high
2
1
2
nm is shown in Figure 5. For [Cl
2
] ) 5.9 × 10 molecules
-3
-1
cm , the molecular fluorescence signal is 1200 counts s , with
a background signal of 30 counts s . This gives a S/N ) 1200/
Method 2 is only valid for the microwave discharge source.
Here the photodissociation cross section is determined by
-
1

Cl Atom Production from ClOOCl Photolysis
J. Phys. Chem. A, Vol. 113, No. 51, 2009 14103
referencing the observed signal from ClOOCl to the observed
loss of Cl
cross section of Cl
the conservation of chlorine between Cl
concentration of ClOOCl is equal to that of Cl
2
when the discharge is turned on and to the known
at 308 or 352 nm. This method relies on
and ClOOCl; i.e., the
that is removed.
2
2
2
Model results, absorption spectroscopy, and laboratory diag-
nostic tests confirm that indeed nearly all (typically >95%) of
2
the total Cl produced from Cl is found in ClOOCl, as discussed
below. The signal from the photolysis of ClOOCl is referenced
to the Cl
2
cross section and the signal from the amount of Cl
2
that is removed:
λ
S
λref ClOOCl
λ
(
φσ)
) (φσ)Cl2
(9)
ClOOCl
λref
∆
S
Cl
2
Figure 6. Photochemical model output for P ) 65 mbar and T ) 240
1
4
-3
K with initial Cl
1
2
) 1.8 × 10 molecules cm and initial O ) 6.5 ×
3
14
-3
0 molecules cm . The vertical gray lines mark (1) the dilution from
where λ ) λref when λ is 308 or 352 nm, and λref is 308 or 352
λref
the Suprasil source tube into the main flow duct and (2) the location
of the laser photolysis axis.
nm when λ is 248 nm; (φσ)Cl is the photodissociation cross
2
λref
2
section of Cl
2
at the reference wavelength; ∆SCl is the decrease
in Cl signal (observed at the Cl detection axis and normalized
2
2
The value of interest in each trace in Figure 4c is the
y-intercept, i.e., the prompt Cl atom signal from ClOOCl
photolysis that is unaffected by secondary chemistry. The
y-intercept is determined for each trace from a polynomial fit
to the Cl detection axis using a scaling factor appropriate for
the reference wavelength) when the microwave discharge is
turned on; and other variables are as defined for eq 8.
Because both methods of determining the photodissociation
cross section simply involve a ratio of two signals multiplied
by a reference cross section, there is no need to know absolute
concentrations of the molecules in the flow tube. Using the
to the data. The polynomial fits have a precision of (10-20%.
λref
With these y-intercept values representing SClOOCl, and ∆SCl2
2 2
easily determined from the Cl detector data and the Cl scaling
factor run (not shown), the ClOOCl photodissociation cross
sections in eqs 8 and 9 are calculated.
2
reference cross sections of ClOOCl at 248 nm and Cl at 308
and 352 nm, the cross section of ClOOCl can be determined a
variety of ways for our three laser wavelengths using both
calculation methods and the two source chemistry techniques.
Experimental Procedure. Each ClOOCl cross section mea-
surement is determined from a set of four consecutive laboratory
runs: (1) 248-nm laser photolysis with the ClOOCl source on;
Our Cl detection is only sensitive to ground-state Cl atoms
(2
P3/2). During the course of these experiments, we determined
that approximately 10% of the Cl atoms produced from the
photolysis of ClOOCl at 248 nm were in the excited state
2
(
P1/2). The time required for quenching the excited state with
N
2
in these experiments was ∼500 µs, which is easily detected
(
2) 308- or 352-nm laser photolysis with the ClOOCl source
on; (3) 308- or 352-nm laser photolysis of Cl only with the
ClOOCl source off; and (4) background measurement with
the source, lasers, and Cl off. Runs 1-3 each consist of an
and could lead to an undercounting of the Cl signal from the
2
2
48-nm laser photolysis. Collisions with CF
2
Cl
2
rapidly deac-
2
18
tivate Cl ( P1/2). We found that adding a small amount of
CF
below our sampling time resolution.
2
14
-3
2
Cl
2
(2 × 10 molecules cm ) reduced the quenching time
average of 3000 laser shots. Ozone is maintained at a constant
concentration for each set of runs.
The ratio of signals at the two detectors with only Cl
to the flow in run 3 is used to determine the scaling factor that
relates the relative sensitivity of the detection axes to Cl . The
measured Cl signal at the molecular fluorescence detector in
run 2 is multiplied by this scaling factor to determine the amount
of signal due to Cl photolysis at the Cl atomic fluorescence
detector. This Cl contribution is subtracted from the total Cl
2
added
Results and Discussion
2
Kinetic Model and Spectroscopy of the Source. Because
the reactions initialized by the microwave discharge or Xe2
photolysis lamp set off a complex photochemical mechanism,
a kinetic model of the flow tube is used to aid in understanding
and optimizing the flow chemistry. A reaction set including all
known relevant photochemical reactions is used in the model.
The kinetic rate constants and equilibrium constants are taken
from JPL-06 with the exception of the ClOOCl equilibrium
constant, which is from Plenge et al.19 The Plenge et al. value
is smaller than that of JPL-06 by a factor of approximately 2.6
2
2
2
signal in run 2 to yield the Cl signal from ClOOCl alone at 308
or 352 nm.
Sample signals observed from the Cl
detector using the three photolysis wavelengths are shown in
Figure 4b,c. The signal from the Cl detector, depicted in Figure
b, is used to correct the signal at the Cl detector for Cl that
2
detector and the Cl
2
4
2
at 240 K and is in better agreement with high-altitude aircraft
measurements.2
0-22
The model is run for a broad range of
is photolyzed at 308 or 352 nm. For method 2, the difference
in signals with the microwave discharge off and on is used to
calculate the amount of Cl converted to ClOOCl in the source.
2
Figure 4c shows time-resolved Cl atom fluorescence signals
from the photolysis of ClOOCl. The signal decay as a function
possible temperatures, pressures, and concentrations. Typical
model output for a microwave discharge run at 240 K with the
248-nm laser is presented in Figure 6, where the evolution of
the chlorine species is shown as a function of time.
of time for each trace is due to the reaction of Cl with O
well as Cl moving out of the detection volume. In this example,
the signal resulting from the photolysis of Cl has been removed.
This Cl signal is approximately twice that of ClOOCl at 352
nm and half that of ClOOCl at 308 nm.
3
, as
The modeled concentrations of Cl
shown) agree well with those that we observe in the laboratory
in the source and detector regions. Cl and ClOOCl are roughly
equal for the discharge source. Model runs and laboratory
experiments both show that the CF Cl photolysis source has
2 3
, ClOOCl, and O (not
2
2
2
2
2

1
4104 J. Phys. Chem. A, Vol. 113, No. 51, 2009
Wilmouth et al.
TABLE 1: Concentrations of Molecules at the Laser
Photolysis Axis from the Photochemical Model Run in
Figure 6
a
fClO
x
molecule concentration (molecules cm-3) Plenge et al. JPL-06
1
1
1
1
7
2
2
1
0
Cl
ClOOCl
ClO
2
2.39 × 10
2.00 × 10
1.70 × 10
3.40 × 10
8.03 × 10
95.1
4.04
0.81
96.6
2.65
0.79
OClO
Cl
2
O
3
0.0038
0.0025
a
The fraction (in percent) of Cl released from Cl
x
species for model runs using the Plenge et al. and JPL-06
2
found in each
1
9
ClO
equilibrium constants.
less Cl
2
present, but we find that it is not possible to be free of
7
Figure 7. Absorption spectrum of the ClOOCl source. The spectrum
is acquired along the 110-cm length of the source tube at P ) 65 mbar
2
Cl due to the very fast reaction of Cl atoms with ClOOCl,
2
and T ) 245 K with the Cl /He microwave discharge source. A least-
squares fit to the spectrum with the contributing ClOOCl, ClO, and
Cl + ClOOCl f Cl + ClOO
(10)
2
14
OClO components are shown. The average concentrations (×10
-
3
molecules cm ) obtained from the least-squares fit to the spectrum
are [ClOOCl] ) 1.66, [ClO] ) 0.53, and [OClO] ) 0.01.
For both sources, the concentrations of all other chlorine
oxides are substantially less than that of ClOOCl. Table 1 lists
the concentrations of the chlorine species at the laser photolysis
TABLE 2: Absorption Cross Sections of Molecules in the
Flow System at the Excimer Laser Wavelengths
a
axis for the model run shown in Figure 6. The fraction of ClO
) 2ClOOCl + ClO + OClO + 2Cl ) represented by each
molecule is also shown.
The model indicates that 95-97% of the Cl atoms produced
from Cl are found in ClOOCl at the laser photolysis axis for
these typical experimental conditions. The remainder is found
in ClO (2-4%) and OClO (<1%). The concentration of Cl
x
molecule
σ248nm
σ308nm
σ352nm
(
2 3
O
ClOOCl
632
<0.1
315
62.4
1080
48.6
17.3
47.7
1.8
7.9
Cl
2
17.8
<1
0.22
0.03
ClO
2
ClONO
2
O
3
12.2
2
O
3
a
-20
cm2
Cross sections are from JPL-06 and have units of 10
is negligible. A series of model runs testing all temperatures,
pressures, velocities, and concentrations used in this experiment,
along with varying the ClOOCl equilibrium constant, indicate
-
1
molecule . ClONO
NO is added.
2
is only relevant for the experiments where
2
that ClOOCl typically comprises at least 95% of ClO
x
and
always comprises at least 90% of ClO
temperature is less than 250 K.
x
, as long as the
Consideration of Potential Interferences. Chemical. The
kinetic model and laboratory absorption spectra both show that
Absorption spectra of the source species were obtained in
the laboratory using the 110-cm combined length of the Suprasil
tube and fused silica side arm as the absorption path. The diluted
flow in the main duct (Figure 2) does not significantly affect
the absorption. A typical spectrum is shown in Figure 7 along
with least-squares fits using JPL-06 cross sections. Background
and sample spectra were obtained with the microwave discharge
source off and on, respectively. Ozone concentrations were held
nearly constant; a small ozone residual was removed with the
fitting and is not shown. The concentrations determined from
the least-squares fits represent the mean values within the source
tube, i.e., the first 2.3 s of Figure 6. The concentration of
ClOOCl at the end of the source region is higher than the mean
value, and that of ClO is lower, as ClO forms the dimer as it
progresses down the tube. Spectra obtained with different
velocities and temperatures in the source show different relative
abundances of OClO and ClO, as expected.
2
ClOOCl and Cl are the dominant chlorine-containing molecules
in the flow system, with ClO and OClO present in minor
concentrations. There is no evidence for a significant presence
x
of any other ClO species. These results, along with other
exploratory laboratory diagnostic tests over a wide range of
experimental conditions, give confidence that all significant
chlorine-containing molecules in the flow system have been
identified. This is of critical importance, as the generation of
chlorine atoms from laser photolysis of a molecule that is not
accounted for will bias the chlorine atom yield measurements.
Absorption cross sections of several important molecules in
the flow system are shown in Table 2 at the relevant excimer
laser wavelengths. OClO is not shown in Table 2, as tests at
room temperature, where the concentration is much higher, show
no measurable Cl atom production when OClO is photolyzed.
2
Because Cl is measured, the only significant chlorine species
in the flow system not explicitly accounted for is ClO.
The significance of ClO laser photolysis at 248 and 308 nm
is calculated using results from the kinetic model. The fractional
contribution of Cl atoms from photolysis of ClO relative to that
from ClOOCl is the ratio of the concentrations and cross
sections, assuming unity quantum yield, with a factor of 2
accounting for the two Cl atoms per ClOOCl molecule:
Overall, the spectroscopic results are consistent with the
kinetic model and confirm that the chemistry in the flow system
is understood. The only ClO species observed in the spectra
x
are ClOOCl, ClO, and OClO. The concentrations determined
from the least-squares fits in Figure 7 correspond to a percent
contribution to ClO
x
of 86.0 for ClOOCl, 13.7 for ClO, and
0
.26 for OClO. These values are in excellent qualitative
agreement with the model results in Table 1. The quantitative
differences in ClO and ClOOCl are expected, as the start of
source region averaged into the spectra has high ClO and low
ClOOCl concentrations.
ClClO/ClClOOCl ) 1/2[ClO]/[ClOOCl] × σClO/σClOOCl
(11)

Cl Atom Production from ClOOCl Photolysis
J. Phys. Chem. A, Vol. 113, No. 51, 2009 14105
at 248 nm, due to the compensating 4% Cl signal from ClO
photolysis at 308 nm and the 2% Cl signal from ClO photolysis
at 248 nm.
It is important to emphasize that no large correction to our
measured ClOOCl photodissociation cross sections is necessary
due to unmeasured interfering species. The aforementioned
minor corrections could in fact be ignored without significantly
changing our cross section results. No value is adjusted by more
than 5% from what was measured, and the corrections applied
to most reported values are less than 2%.
One final consideration with regards to chemical interferences
is the potential for secondary chemistry reactions to affect the
observed Cl signal. Two important reactions initiated after laser
photolysis are Cl + O
3
that removes Cl atoms (eq 4) and
O + ClO f Cl + O2
(13)
Figure 8. Photodissociation cross sections of ClOOCl at 352 nm (units
of 10 cm molecule ) with and without NO added to the flow versus
2
-20
2
-1
temperature. The cross sections are determined from the ratio of signals
obtained at 352 and 248 nm (method 1). The difference in the measured
values with and without NO results from the presence of ClO. This
2
that produces Cl atoms. The latter reaction is only important at
48 nm, where the O cross section is substantially greater than
at 308 or 352 nm (Table 2). The photolysis of O with the 248-
nm laser produces O atoms, which react rapidly with ClO
produced from the Cl + O reaction.
2
3
3
difference only becomes significant at T > 250 K.
3
The model indicates that [ClO]/[ClOOCl] ) 0.08 ( 0.04 over
the range of our typical experimental conditions, depending on
which ClOOCl equilibrium constant is used. At 248 nm, we
expect the contribution of ClO photolysis to be 0.5 × 0.08((0.04)
The impact of the O + ClO reaction producing Cl atom signal
after ClOOCl photolysis can be substantial. Chlorine atoms that
are regenerated via cycling through eqs 4 and 13 can contribute
to the fluorescence signal multiple times before passing through
the detection volume. This is the primary reason why having
time-resolved fluorescence detection for this experiment is
essential. Our laboratory results and kinetic model both show
clearly that the use of the prompt Cl signal (y-intercept) in our
time-resolved fluorescence detection eliminates any influence
from secondary chemistry reactions on our results. In general,
no secondary chemistry reactions are fast enough to affect the
prompt Cl signal observed from ClOOCl photolysis. We also
note that the ClOO fragment from ClOOCl photolysis dissoci-
ates so rapidly that we observe both Cl atoms from ClOOCl
promptly, even with time-resolved detection.
×
315/632 ) 0.02 ( 0.01 relative to that of ClOOCl photolysis.
At 308 nm, we expect a larger contribution of 0.04 ( 0.02.
Accordingly, we include a 2% correction for ClO photolysis at
2
48 nm and a 4% correction for ClO photolysis at 308 nm in
our reported ClOOCl cross sections. The inclusion of these
corrections is also factored into the reported uncertainties. We
verified that ClO does not produce measurable Cl atoms at 352
nm in experiments performed at room temperature.
The small impact of ClO photolysis predicted by the model
was confirmed in the laboratory by adding NO
flow to remove ClO via
2
to the source
Wall Loss. If present, loss of ClOOCl on surfaces of the flow
system could significantly impact the measurements. Specifi-
cally, if ClOOCl were lost on the walls, the photodissociation
cross sections determined using method 2 may be biased low,
ClO + NO + M f ClONO + M
(12)
2
2
Since the cross section of ClONO
2
is approximately a factor of
2
as the observed loss of Cl would overestimate the amount of
1
0 smaller (Table 2) than that of ClOOCl, and the Cl photolysis
ClOOCl actually present (eq 9). In general, the method 1 results
should be unaffected by wall loss.
Previous laboratory studies note the potential for ClOOCl to
be lost on surfaces and employ the use of clean quartz tubing
7
yield is less than unity, NO
the system by converting it to ClONO
that the cross section measured at 352 nm using 248 nm as a
reference should be 2% higher and that the cross section at 308
nm should be 2% lower with NO
2
effectively removes ClO from
. We therefore expect
2
or a halocarbon wax coating on the walls to minimize its
present (eq 8). Measurements
destruction.4
,23-25
Similarly, our source tube, where ClOOCl
2
with and without NO are nearly identical for most of the
2
typically resides for 75% of its time in the flow system, is clean
Suprasil. The main duct is electropolished stainless steel with
an amorphous silicon coating from Restek.
conditions of the experiments, indicating that ClO concentrations
in the flow tube are indeed small. The exception is for T > 250
K when the ClO/ClOOCl ratio is larger. Figure 8 shows
photodissociation cross sections determined at 352 nm to
illustrate this result.
To test for the possibility of surface loss affecting the
measurements, we varied velocity in both the main flow duct
and Suprasil source tube by factors of approximately 2 and 4,
respectively. Panels a and b in Figure 9 show the photodisso-
ciation cross sections of ClOOCl at 352 nm for these velocity
runs. If wall loss were significant, the measured photodissocia-
tion cross sections for method 2 would decrease at the lower
velocities and increase at the higher velocities. The absence of
any trend in the method 2 values indicates that the observed
photodissociation cross sections are independent of exposure
time to the walls and, therefore, are very unlikely to be biased
by wall loss of ClOOCl (or any other chlorine species formed
when the microwave discharge is turned on).
In addition to the ClO photolysis correction, a second minor
correction to the method 2 results is necessary to account for
the fact that Cl
approximately 95% (Table 1) of the ClOOCl formed instead of
00%. Accordingly, our reported 352-nm ClOOCl cross section
results with method 2 are increased by 5% from those measured
to account for the 5 ( 2% of ClO in ClO or OClO at the typical
experimental conditions where the measurements were made
eq 9). The correction to the measured values is only 1% for
the ClOOCl cross section at 308 nm referenced to Cl and 3%
2
lost in the microwave discharge equals
1
x
(
2

1
4106 J. Phys. Chem. A, Vol. 113, No. 51, 2009
Wilmouth et al.
measurement is made near the end of the flow system,
interchange between ClOOCl and Cl upstream of the Cl atomic
2
fluorescence axis has no impact on the method 2 cross sections.
Photodissociation Cross Sections. Figure 10 shows our
ClOOCl photodissociation cross section results at 308 nm (upper
grouping of points in each panel) and 352 nm (lower points).
We performed 240 experiments over a wide range of conditions
in search of any parameter that would influence our measured
values. Figure 10 shows (φσ)ClOOCl determined from eqs 8 and
9
for every one of the 240 individual measurement runs acquired
over several months. It is immediately clear from panel a that
the data are tightly grouped with no apparent differences whether
the cross section is measured with the microwave discharge
source or Xe photolysis lamp source or whether the calculation
2
is made using method 1 or method 2. There is also no obvious
temporal trend.
Panels b-e show the photodissociation cross sections plotted
versus a number of variables that could potentially influence
the results. The absence of any trend in these regressions
confirms that a number of potential issues have been properly
addressed. The absence of a slope in panel b indicates that Cl
2
is being measured and accounted for correctly. Panel c shows
that there are no abnormalities in our detection of ClOOCl over
a wide range of concentrations, panel d indicates that the
measured signal is not influenced by any reaction that involves
3
O , and panel e shows no cross section dependence on the
pressure of the flow system. The trend with temperature at 352
Figure 9. Photodissociation cross sections of ClOOCl at 352 nm (units
nm in panel f is significant. When ClO is not fully converted to
-
20
2
-1
of 10
velocity in the Suprasil source tube, and (c) conversion efficiency of
the microwave discharge, as determined from the Cl molecular
cm molecule ) versus (a) velocity in the main duct, (b)
λref
2
ClOOCl at T > 250 K, method 2 is not accurate, i.e., the ∆SCl
term in eq 9 is erroneously large. The scatter in Figure 10 results
from statistical uncertainty of each measurement ((20-30%).
In addition to the regressions shown in Figures 8-10, the
2
fluorescence detector. Data obtained with the photolysis source (+)
are in green, and data obtained with the microwave discharge source
(
O) using method 1 are in blue and using method 2 are in red.
2
data set includes variations in NO concentration, laser power,
laser beam size and position, experimental order of data
acquisition, and chlorine lamp output. We are unable to identify
any experimental parameter at T < 250 K that can significantly
change our results, giving confidence that the chemical flow
system is well understood and that systematic errors have been
identified and properly addressed.
Further evidence that wall loss is not significant is shown in
Figure 9c. The measured photodissociation cross sections of
ClOOCl at 352 nm are independent of the conversion efficiency
of the microwave discharge source, as measured at the Cl
2
detection axis. The absence of a systematic trend demonstrates
that the photodissociation cross sections are not affected by the
relative concentrations of chlorine species in the flow system
and that wall loss is not biasing the observed conversion
efficiency numbers.
A summary of all the cold-temperature ClOOCl cross section
data is shown in Table 3. Because each of the 240 experimental
runs is typically used to determine the ClOOCl cross section in
more than one way, there are 468 total photodissociation cross
section measurements represented here. The results are itemized
on the basis of source chemistry, calculation method, and
wavelength. The method 1 results are those referenced to
ClOOCl at 248 nm, and the method 2 results are those
The experimental evidence demonstrating that surface loss
is not significant is consistent with results from past studies.
Burkholder et al.26 concluded that there was no wall loss of
ClOOCl in their experiment after changing the residence time
in their absorption cell by a factor of 3 and observing no
systematic decrease in the ClOOCl absorption cross sections.
Two studies using static cells quantified the decomposition of
2
referenced to Cl at 308 and 352 nm. The columns listing the
ratios of the photodissociation cross sections of the reference
molecules to those of ClOOCl represent our experimentally
determined values, i.e., the ratios of measured signals from eqs
8 and 9. Uncertainties in the ratios include the standard deviation
of the measurements and an estimate of systematic error. The
corrections that account for the presence of ClO and OClO are
included in the systematic error. We report two values at 248
nm, three at 308 nm, and three at 352 nm.
2
4
ClOOCl: Demore and Tschuikow-Roux found a rate constant
-
4
-1
23
of approximately 1 × 10
s
at 195 K, and Cox and Hayman
reported a ClOOCl half-life of 1000 s at 233 K. These numbers
imply a slow ClOOCl decay on surfaces with far less than 1%
of ClOOCl lost in 3 s, the approximate residence time in our
flowing system. We note that the temperatures used in our
experiment are much too warm to condense ClOOCl on the
Using the listed JPL-06 values for the reference cross sections
and assuming a quantum yield of 1, the ClOOCl photodisso-
ciation cross sections for each of the eight ratio measurements
are determined. The photodissociation cross section results agree
remarkably well for each wavelength, independent of the source
chemistry or calculation method. We note that our photodisso-
ciation cross section values should be revised in the future if
the reference cross sections are changed.
4
walls, as in the Pope et al. study.
In summary, if wall loss of ClOOCl is biasing our measure-
ments, we are unable to identify or quantify it, so we do not
employ any correction for wall effects to the data. We also note
that even if ClOOCl were lost on the walls of our flow system,
our measurements would essentially be unaffected if the loss
2 2
mechanism produced Cl as the product. Because our Cl

Cl Atom Production from ClOOCl Photolysis
J. Phys. Chem. A, Vol. 113, No. 51, 2009 14107
Figure 10. Photodissociation cross sections of ClOOCl (units of 10- cm molecule-1) for the photolysis pathway that produces Cl atoms. Data
obtained with the photolysis source at 308 nm (×) and 352 nm (+) are in green. Data obtained with the microwave discharge source at 308 nm (∆)
and 352 nm (O) using method 1 are in blue and using method 2 are in red. The measurements are shown on a log scale plotted against (a) run
20
2
number, (b) Cl
T < 250 K. In panel f, NO
2
concentration, (c) ClOOCl concentration, (d) O
3
concentration, (e) pressure, and (f) temperature. Data in panels a-e are limited to
2
was added to remove excess ClO for T > 250 K.
TABLE 3: Summary of Measured ClOOCl Photodissociation Cross Sections at T ) 240 ( 10 Ka
48 nm 308 nm
2
352 nm
source
reference
σ
ref
(φσ)ref/(φσ)ClOOCl
(φσ)ClOOCl
(φσ)ref/(φσ)ClOOCl
(φσ)ClOOCl
(φσ)ref/(φσ)ClOOCl
(φσ)ClOOCl
2
48nm
microwave
photolysis
microwave
microwave
ClOOCl
ClOOCl
631.5
631.5
17.32
17.78
s
s
s
s
675
644
16.36 ( 1.82
17.59 ( 2.06
0.431 ( 0.034
s
38.6
35.9
40.2
s
72.67 ( 6.94
71.76 ( 9.54
s
8.69
8.80
s
2
48nm
3
3
08nm
52nm
Cl
Cl
2
0.0257 ( 0.0029
0.0276 ( 0.0029
2
2.08 ( 0.23
8.54
mean value
660 ( 100
39.3 ( 4.9
8.6 ( 1.2
a
Photodissociation cross sections of ClOOCl are for the photolysis channel that produces Cl atoms (eq 2a). Cross sections are in units of
20
2
-1
1
0- cm molecule . Reference cross sections are from JPL-06. Values for (φσ)ClOOCl are calculated assuming φref ) 1. The mean
photodissociation cross section for each wavelength, shown in the lowermost row, is the mean of all the individual measurements: 140
measurements at 248 nm, 92 at 308 nm, and 236 at 352 nm. Uncertainties in the mean values include the measurement and statistical
uncertainties but not uncertainties from the reference cross sections.
The average photodissociation cross section for each wave-
length is shown in the lowermost row of Table 3. The
uncertainties listed are only the measurement and statistical
uncertainties (∼12-15%) and do not include uncertainties for
the reference cross sections from JPL-06.
Perspective on Cross Section Results. The value recom-
mended by JPL-06 for the ClOOCl cross section at each laser
wavelength is shown in Table 2. In comparing these numbers
with those in Table 3 from this work, it is important to note
that the JPL-06 values are absorption cross sections, while our
values are the products of quantum yields and absorption cross
sections. The quantum yield for the Cl atom production channel
of ClOOCl photolysis (eq 2a) recommended by JPL-06 is
absorption cross section in previous measurements. We em-
phasize that the true value of the quantum yield at these
wavelengths has no direct effect on our results; it only affects
our ability to compare our results with JPL-06 and other previous
studies that used absorption spectroscopy. The value we
measure, the product of the quantum yield of the Cl production
pathway and the ClOOCl absorption cross section, is the quantity
of greatest atmospheric importance.
As is evident from Tables 2 and 3, the measurements of the
photodissociation cross section in this study are similar to the
values of the absorption cross section recommended by the JPL-
06 panel at 248 and 352 nm but are slightly lower at 308 nm.
That we agree with JPL-06 at 248 nm and are lower at 308 nm
is not surprising, given the measurement database at those
wavelengths.5
1
.0((0.1) below 300 nm and 0.9((0.1) above 300 nm. If the
quantum yield above 300 nm is indeed 0.9, our average 308-
and 352-nm absorption cross sections are 43.7 and 9.6,
The cross section measurement at 352 nm is in many ways
the most important result. The 352-nm wavelength is the most
atmospherically relevant of those measured, and it is by far the
most difficult to measure accurately with absorption spectros-
respectively (units of 10- cm molecule ).
20
2
-1
To our knowledge, the quantum yield of reaction 2a has never
actually been measured experimentally at 352 nm, nor at any
wavelength above 308 nm. Moreover, there is not a clear
consensus in the literature regarding the true value of the
2
copy, as the cross section of ClOOCl is small and that of Cl is
relatively large. Figure 11 shows the cross section of ClOOCl
at 352 nm determined here and in many of the prior studies.
The most striking feature of this figure is the wide range of
values for the cross section and the corresponding large
uncertainty estimate (factor of 3) in the JPL-06 recommendation.
Of the five previous individual cross section studies at 352
nm shown in Figure 11, Burkholder et al.26 is the only study
with which we have overlapping uncertainties. Our value is a
14
quantum yield at 308 nm, and it certainly may be unity. The
2
7
IUPAC subcommittee recommendation is, in fact, for a
quantum yield of 1.0 throughout the 200-398 nm range. Given
the uncertainty, we have decided to proceed here with the
simplest assumption that the quantum yield is 1.0 at all
wavelengths, i.e., our comparisons below assume that (φσ)ClOOCl
in this study represents the same quantity as the ClOOCl

1
4108 J. Phys. Chem. A, Vol. 113, No. 51, 2009
Wilmouth et al.
conditions. The photolysis rates determined with these new data
will not be substantially different from those determined with
the values of JPL-06 or Burkholder et al.,26 but the directness
of the technique and completeness of this study help solidify
the laboratory foundation supporting our understanding of
chlorine-catalyzed ozone loss.
Acknowledgment. We appreciate the role of the SPARC
initiative and workshop (The Role of Halogen Chemistry in
Polar Stratospheric Ozone Depletion) in drawing focus to this
scientific issue and promoting discussion. In particular, we thank
Stan Sander, Jim Burkholder, and Kyle Bayes for helpful
discussions on this work. This research is supported by NASA
grant NNX09AE29G.
Figure 11. Absorption cross sections of ClOOCl at 352 nm (units of
-
20
2
-1
1
0
cm molecule ) from this work compared with selected
4
-7,25,26
previously published values.
The mean value for each study is
References and Notes
shown in blue with stated error limits in red. The mean cross section
from this work is shown assuming the quantum yield is 1.0 (see the
(1) World Meteorological Organization (WMO), Scientific Assessment
of Ozone Depletion: 2006, WMO Global Ozone Research and Monitoring
Project, Report 50, WMO: Geneva, Switzerland, 2007.
4
text). The Pope et al. uncertainty, which is obscured on the scale of
2
5
this figure, is +20/-10%. Huder and DeMore did not report
(
2) Frieler, K.; Rex, M.; Salawitch, R. J.; Canty, T.; Streibel, M.;
uncertainties.
Stimpfle, R. M.; Pfeilsticker, K.; Dorf, M.; Weisenstein, D. K.; Godin-
Beekmann, S. Geophys. Res. Lett. 2006, 33, doi: 10.1029/2005GL025466.
(3) Molina, L. T.; Molina, M. J. J. Phys. Chem. 1987, 91, 433.
(4) Pope, F. D.; Hansen, J. C.; Bayes, K. D.; Friedl, R. R.; Sander,
S. P. J. Phys. Chem. A 2007, 111, 4322.
2
5
factor of 2.3 higher than that of Huder and DeMore, which is
2
7
also the IUPAC recommendation, and a factor of 12 higher
4
6
than that of Pope et al. The recent Chen et al. value at 250 K
and 351 nm is 46% higher than our result. The source of the
discrepancy is not clear, as their mass detection is not suscepti-
ble to the same impurity issues as the absorption studies, but
we do not observe cross sections as high as those found in the
(
5) von Hobe, M.; Stroh, F.; Beckers, H.; Benter, T.; Willner, H. Phys.
Chem. Chem. Phys. 2009, 11, 1571.
6) Chen, H.; Lien, C.; Lin, W.; Lee, Y. T.; Lin, J. J. Science 2009,
24, 781.
7) Sander, S. P.; Friedl, R. R.; Ravishankara, A. R.; Golden, D. M.;
(
3
(
Kolb, C. E.; Kurylo, M. J.; Molina, M. J.; Moortgat, G. K.; Keller-Rudek,
H.; Finlayson-Pitts, B. J.; Wine, P. H.; Huie, R. E.; Orkin, V. L. Chemical
Kinetics and Photochemical Data for use in Atmospheric Studies EValuation
Number 15, JPL Publication 06-02, Jet Propulsion Laboratory: Pasadena,
CA, 2006.
(8) Burkholder, J. B.; Mauldin, R. L.; Yokelson, R. J.; Solomon, S.;
Ravishankara, A. R. J. Phys. Chem. 1993, 97, 7597.
6
five experimental runs of Chen et al. at these conditions.
Ultimately, we find our best cross section agreement at 352 nm
with the JPL-06 recommended value. Our reported value is
slightly higher than JPL-06, but our uncertainty range encom-
passes the recommendation.
(
9) Ibuki, T.; Hiraya, A.; Shobatake, K. J. Chem. Phys. 1989, 90, 6290.
It is worth considering the impact on these comparisons if
the quantum yield of reaction 2a at 352 nm is actually 0.9 instead
of 1.0. Overall, our cross section difference with most of the
other studies is so substantial that any uncertainty regarding
the quantum yield affecting the comparisons is inconsequential.
One change of note is that our absorption cross section would
(
10) Basco, N.; Hunt, J. E. Int. J. Chem. Kinet. 1979, 11, 649.
(11) Trolier, M.; Mauldin, R. L.; Ravishankara, A. R. J. Phys. Chem.
990, 94, 4896.
1
(
12) Anderson, J. G.; Margitan, J. J.; Stedman, D. H. Science 1977,
1
98, 501.
(13) Brune, W. H.; Anderson, J. G.; Chan, K. R. J. Geophys. Res. 1989,
94, 16649.
2
6
(14) Molina, M. J.; Colussi, A. J.; Molina, L. T.; Schindler, R. N.; Tso,
T.-L. Chem. Phys. Lett. 1990, 173, 310.
15) Rao, Y. V.; Venkateswarlu, P. J. Mol. Spectrosc. 1962, 9, 173.
(16) Lee, L. C.; Suto, M.; Tang, K. Y. J. Chem. Phys. 1986, 84, 5277.
be much closer to that of Burkholder et al. than that of JPL-
6, but our value would still lie between the two. While higher,
0
(
the upper bound of our cross section uncertainty would remain
significantly below the lower bound of the uncertainty from the
(
(
(
17) Yu, Y. C.; Setser, D. W. J. Photochem. Photobiol. 1988, 42, 27.
18) Chichinin, A. I. J. Chem. Phys. 2000, 112, 3772.
6
Chen et al. study.
19) Plenge, J.; K u¨ hl, S.; Vogel, B.; M u¨ ller, R.; Stroh, F.; von Hobe,
Our relative agreement at 352 nm with the JPL-06 and
M.; Flesch, R.; R u¨ hl, E. J. Phys. Chem. A 2005, 109, 6730.
20) Stimpfle, R. M.; Wilmouth, D. M.; Salawitch, R. J.; Anderson, J. G.;
J. Geophys. Res. 2004, 109, doi:10.1029/2003JD003811.
21) von Hobe, M.; Grooss, J.-U.; M u¨ ller, R.; Hrechanyy, S.; Winkler,
Burkholder et al.2 ClOOCl cross sections is consistent with
6
(
1
,20,28
ClO/ClOOCl field observations
and with the calculated
(
ClOOCl photolysis rate that is needed to explain observed O
3
U.; Stroh, F. Atmos. Chem. Phys. 2005, 5, 693.
(22) Avallone, L. M.; Toohey, D. W. J. Geophys. Res. 2001, 106, 10411.
29
loss. We find no need for new photolytic or reactive pathways
to bring agreement between models and measurements of
chemical ozone depletion.
(
(
23) Cox, R. A.; Hayman, G. D. Nature 1988, 332, 796.
24) DeMore, W. B.; Tschuikow-Roux, E. J. Phys. Chem. 1990, 94,
5
856.
(
25) Huder, K. J.; DeMore, W. B. J. Phys. Chem. 1995, 99, 3905.
Conclusions
(26) Burkholder, J. B.; Orlando, J. J.; Howard, C. J. J. Phys. Chem.
990, 94, 687.
1
The photodissociation cross sections measured here are
consistent with the widely accepted mechanism for chlorine
removal by the ClOOCl catalytic cycle. The measurements show
that the photolysis rate of ClOOCl is at least as fast as previously
thought. Significantly, the likelihood of a large systematic error
is greatly reduced by the measurement of the primary contami-
(27) Atkinson, R.; Baulch, D. L.; Cox, R. A.; Crowley, J. N.; Hampson,
R. F.; Hynes, R. G.; Jenkin, M. E.; Rossi, M. J.; Troe, J. Atmos. Chem.
Phys. 2007, 7, 981.
(
28) Wilmouth, D. M.; Stimpfle, R. M.; Anderson, J. G.; Elkins, J. W.;
Hurst, D. F.; Salawitch, R. J.; Lait, L. R. J. Geophys. Res. 2006, 111, doi:
10.1029/2005JD006951.
(
29) von Hobe, M. Science 2007, 318, 1878.
2
nant, Cl , and by a methodical variation in experimental
JP9053204