A Novel Flash Photolysis/UV Absorption System Employing Charge-Coupled Device (CCD) Detection: A Study of the BrO + BrO Reaction at 298 K

Article abstract of DOI:10.1021/jp951825b

A novel flah photolysis/kinetic absorption spectroscopy system has been constructed for the study of gas phase reactions.The experiment incorporates a charge-coupled device (CCD) detector and represents the first application of such devices to the study of gas phase kinetics.The CCD enables the recording of rapid sequential time-resolved UV/visible absorption spectra before, during, and after photolysis of a gas mixture.The unequivocal identification and monitoring of several absorbing components in the reacting mixture is therefore possible, therby maximizing the amount of information gathered from a single flash photolysis experiment.The experimental system is described in full here.Results from a preliminary kinetic study of the BrO self-reaction at 298 K are also described: BrO + BrO -> 2Br + O2 (1a); BrO + BrO -> Br2 + O2 (1b).Experiments were performed to independently determine the rates of the overall reaction, k₁, defined by (-d/dt = 2k₁²) and both individual reaction channels (1a) and (1b), giving k₁ = (2.98 +/- 0.42)E-12 cm³ molecule s^-1; k₁a = (2.49 +/- 0.42)E-12 cm³ molecule^-1 s^-1; and k₁b = (4.69 +/- 0.68)E-13 cm³ molecule^-1 s^-1 at 298 K and 760 Torr total pressure in oxygen bath gas.Errors are 2?.Deviations from the expected second-order kinetic scheme were observed in the experimental system wherein bromine was photolyzed in the presence of excess ozone as a source of the BrO radicals.At low ozone concentrations, the formation of Br2O is proposed, and at very high ozone concentrations the formation of symmetric bromine dioxide, OBrO, is observed.Kinetic schemes for these side reactions have been deduced, and the mechanistic implications of these results are discussed.

Full text of DOI:10.1021/jp951825b

3
020
J. Phys. Chem. 1996, 100, 3020-3029
A Novel Flash Photolysis/UV Absorption System Employing Charge-Coupled Device (CCD)
Detection: A Study of the BrO + BrO Reaction at 298 K
David M. Rowley,* Matthew H. Harwood, Raymond A. Freshwater, and Roderic L. Jones
Centre for Atmospheric Science, UniVersity Chemical Laboratory, Lensfield Road, Cambridge CB2 1EW, U.K.
X
ReceiVed: July 6, 1995; In Final Form: September 25, 1995
A novel flash photolysis/kinetic absorption spectroscopy system has been constructed for the study of gas
phase reactions. The experiment incorporates a charge-coupled device (CCD) detector and represents the
first application of such devices to the study of gas phase kinetics. The CCD enables the recording of rapid
sequential time-resolved UV/visible absorption spectra before, during, and after photolysis of a gas mixture.
The unequivocal identification and monitoring of several absorbing components in the reacting mixture is
therefore possible, thereby maximizing the amount of information gathered from a single flash photolysis
experiment. The experimental system is described in full here. Results from a preliminary kinetic study of
the BrO self-reaction at 298 K are also described: BrO + BrO f 2Br + O
2
(1a); BrO + BrO f Br
1b). Experiments were performed to independently determine the rates of the overall reaction, k , defined
) (2.98 ( 0.42)
2
+ O
2
(
1
2
1 1
by (-d[BrO]/dt ) 2k [BrO] ) and both individual reaction channels (1a) and (1b), giving k
-
12
3
-1 -1
-12
3
-1 -1
×
10 cm molecule s ; k1a ) (2.49 ( 0.42) × 10 cm molecule s ; and k1b ) (4.69 ( 0.68) ×
-
13 3 -1
cm molecule s-1 at 298 K and 760 Torr total pressure in oxygen bath gas. Errors are 2σ. Deviations
1
0
from the expected second-order kinetic scheme were observed in the experimental system wherein bromine
was photolyzed in the presence of excess ozone as a source of the BrO radicals. At low ozone concentrations,
the formation of Br O is proposed, and at very high ozone concentrations the formation of symmetric bromine
2
dioxide, OBrO, is observed. Kinetic schemes for these side reactions have been deduced, and the mechanistic
implications of these results are discussed.
Introduction
parts of the detector window, simultaneous wavelength-resolved
measurements were not possible. Thus, the PMT allowed the
Of the available laboratory techniques for the study of gas
phase reactions, flash photolysis coupled with kinetic absorption
spectroscopy has proved to be one of the most powerful. The
monitoring of a single wavelength of analysis light which was
selected using a monochromator. Use of a single wavelength
limited the ability to identify species, however, and no dis-
crimination between other absorbers was possible. Despite this,
the electronic recording of intensity and the use of more stable
continuous analysis lamps have greatly improved the reliability
of detection over that obtained from the use of photographic
plates, and photomultiplier tubes are still widely used in flash
photolysis systems today (e.g. refs 4, 5).
1,2
technique, originally developed in this laboratory in the 1950s,
involves the rapid generation of reactive species, atoms and free
radicals, which are then monitored, by virtue of their charac-
teristic UV/visible absorptions, as a function of time. The nature
and rates of reactions involved in their removal are then inferred
from the decay trace obtained. Since its conception, refinements
to the flash photolysis technique have come about as a result
of technical improvements, particularly in the monitoring of
transient absorptions.
In the original flash photolysis experiments, photographic
plates were used to measure absorptions.1 The plate was
situated in the dispersive plane of a spectrograph, and the light
intensity from a pulsed analysis lamp was transmitted through
the reaction cell at a given delay time after the photolysis flash
and recorded. The wavelength-resolved signals were converted
manually into absorptions by measuring the degree of darkening
across the photographic plates. Repeated experiments using
different delay times between the photolysis and the analysis
flash were then carried out to build up the temporal profile of
the absorption.
Recently, the availability of photodiode arrays has enabled
the monitoring of analysis light intensity electronically and
simultaneously as a function of both wavelength and time. The
array consists of a row of small, separate light sensitive elements
which can be aligned along the dispersive axis of a spectrograph.
The time resolution of the recording is limited by the readout
time of the device. Since readout takes place sequentially for
each element of the array, full readout takes typically mil-
liseconds, and diode arrays are therefore unsuitable for the study
of fast kinetic processes. Nevertheless, photodiode arrays have
,2
6
proved useful in the study of UV/visible absorption spectra,
7
in studies of slower kinetics, and for product analyses. In
addition, fast electronic time gating of photodiode arrays has
enabled the recording of short exposure spectra at specific times
in a flash photolysis experiment, and, analogously to the
photographic plates method, repeated experiments can be
With the advent of electronic photodetectors, photomultiplier
tubes (PMTs), the absorption signal from a flash photolysis
experiment could be monitored continuously as a function of
5
performed to build up the full temporal variation of absorption.
3
time. This enabled the recording of a decay trace in a single
In this work a two-dimensional detector array has been
incorporated into a flash photolysis system. These arrays,
known as charge-coupled devices (CCDs), convert incident light
into photocharge and also have the facility to rapidly and
efficiently transfer this signal into a storage region on the device.
This enables the recording of many sequential spectra at rapid
experiment. However, because of the size of the PMT and the
inability of the detector to discriminate light falling on different
*
Author to whom correspondence should be addressed. E-mail:
rowley@atm.ch.cam.ac.uk.
X
Abstract published in AdVance ACS Abstracts, January 15, 1996.
0022-3654/96/20100-3020$12.00/0 © 1996 American Chemical Society

BrO + BrO Reaction at 298 K
J. Phys. Chem., Vol. 100, No. 8, 1996 3021
enclosing it in a vacuum for thermal insulation. This dark signal
is recorded for each pixel prior to every readout of the device
and subtracted from the measured intensity. Additionally,
although a number of potential sources of noise exist in CCD
operation, the main source of noise is photonic noise, inherent
to the light incident on the device. For a typical full pixel
6
capacity of 2 × 10 photoelectrons, this corresponds to a noise
level of (0.07%.
Use of a CCD in Flash Photolysis. The charge-coupled
device is used in the flash photolysis system to record rapid
sequential spectra of the analysis light transmitted through the
reaction mixture before, during, and after photolysis.
A
schematic of the arrangement is shown in Figure 2. The analysis
light collimated through the reaction vessel is imaged across
rows at the top of the charge-coupled device, and charge transfer
processes are used to record spectra as a function of time. The
flash photolysis system is discussed in detail below.
Experimental Section
Gases are manipulated using a standard Pyrex line equipped
with greaseless Teflon taps (Young & Co.). The line consists
of two parts: a vacuum line and a gas-mixing line. In the
vacuum line, pressures down to 10- mbar are readily attainable
using a rotary/diffusion pump combination and measured using
a Pirani/cold cathode detector combination (MKS Instruments
Ltd. 953). The gas-mixing line consists of a carrier gas flow
tube and five gas injectors aligned against the carrier flow for
efficient gas mixing. All gas flows into the mixing line are
controlled by mass flow controllers (MKS 1159B).
Figure 1. Schematic diagram of the principles of CCD operation.
Incident light is converted into photocharge (A) and stored in a poten-
tial well on the silicon. Three sets of parallel transfer electrodes (1, 2,
and 3) are then charged cyclically to maintain a potential gradient
across the device and push the photocharge from pixel to pixel (B).
Finally, when photocharge from the first pixel has traversed the whole
device, the charge is read out from the readout register to the output
amplifier (C).
5
rates. Thus, the advantage of recording a range of wavelengths
of absorption as with a diode array is preserved, without
compromising the ability to monitor rapidly and continuously
as with a PMT.
The reaction cell into which gases are passed is made in
Spectrosil quartz tube and is 1 m in length. The cell is double
jacketed: the inner jacket containing flowing liquid (Galden
HT90) from a recirculating thermostating unit (Huber HS 80).
The outer jacket is held at low pressure to prevent condensation
on the cell walls when operating at low temperatures. Similarly,
double-windowed, evacuated end pieces are used to prevent
condensation on optical surfaces. The gas temperature in the
cell is monitored using a ceramic enclosed, fast reacting Pt-
100 resistance thermometer (Thermospeed PQ 93870), which
resides in the gas flow. The gas temperature in the cell is thus
controllable from 193 to 373 K within an accuracy of (0.5 K.
Radicals are generated in the cell using flashlamp photolysis.
The lamp used is a 1 m arc length xenon lamp, filled to 50
Torr (Qarc Ltd. QDX 66). The lamp is powered from a 25
kV, 2.6 µF rapid discharge capacitor (NWL Ltd.) typically
charged to 20 kV using a high-voltage power supply (Glassman
High Voltage Ltd. EH100). Since the high voltage used for
the lamp would cause self-breakdown and uncontrolled lamp
firing if applied directly, a spark gap (EG&G GP12B) is used
to hold off the high voltage until lamp triggering. The spark
gap is ionized using a trigger unit (EG&G TM11A), itself
controlled using an optically isolated TTL (+5 V) pulse from
the control computer. The flashlamp circuit is designed to be
critically damped to give the maximum efficiency, short pulse
duration, and absence of current reversal (“ringing”). The
Principles of CCD Operation. A full account of the theory
8
and operation of charge-coupled devices is outside the scope
of this work. A brief summary of the main features of the CCD
is given here.
The charge-coupled device consists of a grid of linearly
coupled metal oxide semiconductor (MOS) elements (pixels)
embedded in a silicon substrate. These elements convert
incident light into photocharge, which is stored in a potential
well within the pixel. Transfer of the photocharge from one
row of pixels to the adjacent row can then be effected, along
one of the axes of the array, by applying suitably phased voltages
to parallel electrodes running across each row of pixels. When
the charge from the first row has traversed the entire device,
readout of each row of pixels can take place on a slower time
scale. A schematic diagram of a charge-coupled device is
shown in Figure 1, along with the principle of operation of
charge transfer.
The quantum efficiency (QE) of the CCD light-to-charge
conversion peaks at 700 nm, where QE ≈ 60%. At lower
wavelengths, this efficiency decreases, and for detection at
wavelengths below 400 nm, the device is thinly coated with
phosphor, which fluoresces at visible wavelengths. Using this
coating, the effective QE is ca. 20% from 180 to 400 nm. The
phosphor fluorescence lifetime is on the order of nanoseconds,
and the phosphor coating is extremely thin (0.3 µm) compared
to the pixel size (22.5 µm square). Thus, cross talk between
pixels on the device is negligible. The charge transfer efficiency
9
method of Markiewich and Emmet was used to determine the
optimum R, L, and C parameters. The flashlamp pulse energy,
approximately 500 J, is completely discharged in less than 20
µs. The optical pulse duration from the flashlamp was
measured using a rapid response photodiode and storage
oscilloscope and found to be ca. 10 µs fwhm. The spectral
output of the xenon flashlamp was not measured but, in
common with similar lamps operated under analogous condi-
tions, is expected to cover the UV/visible range from 180 to
-6
of the CCD is very high, typically >(1-10 )/transfer, and rates
of charge transfer up to 1 MHz per row of pixels (a 1 µs row-
shifting time) are readily attainable. Alternatively, the charge
transfer time can be made arbitrarily long.
As with diode arrays, CCDs produce a dark current, which
is minimized by Peltier cooling of the device to 200 K and
1
0
700 nm.

3
022 J. Phys. Chem., Vol. 100, No. 8, 1996
Rowley et al.
Figure 2. Schematic diagram of the flash photolysis system. Light from the source lamp is collimated through the reaction cell and transferred
to the spectrograph. Wavelength dispersed light is imaged across the top of the CCD perpendicular to the axis of fast charge transfer. Sequential
spectra of the cell contents are recorded by charge transfer on the CCD.
The flashlamp is situated adjacent and parallel to the reaction
cell and can be moved toward and away from the reaction vessel
transfer. The latter is minimized by the use of a small source
(the fiber-optic cable) and an astigmatic, imaging spectrograph.
This preserves the vertical integrity of the source to a much
greater extent than normal Czerny-Turner designs, by virtue of
toroidal collimating and focusing mirrors and, with the small
image size, thereby maximizes temporal resolution.
(3-10 cm separation) to vary the effective photolysis intensity.
When very short (<280 nm) wavelength photolysis is not
required, the flashlamp is enclosed in a Pyrex envelope. A
slow flow of air through this envelope cools the lamp and
prevents ozone buildup from atmospheric photolysis around the
lamp.
The optical fiber diameter used is 550 µm, and after
magnification by the spectrograph the height of the image
corresponds to 35 pixels (pixels are 22.5 µm square) at the
detector. Thus, using the detector’s minimum shift time of 9
µs/row, a single row of spectral data takes a minimum of 315
µs to acquire. This spectral acquisition time can be further
reduced, at the expense of light throughput, by the use of a
smaller diameter (100 µm) optical fiber.
The reaction cell and flashlamp are enclosed in a light-tight
metal box. The box is baffled to minimize the escape of flash-
lamp light along the optical axis and earthed to minimize
radiofrequency interference from the flashlamp. Similarly, the
flashlamp circuit is enclosed in a Faraday cage, earthed to the
same point. All electrical leads into the flashlamp circuit are
screened, and main power is supplied from a filtered unit,
incorporating an earth choke.
The CCD used is a UV coated 298 × 1152 pixel sensor (EEV
CCD05-10-0-202), mounted in a Peltier cooled camera head
Radicals and molecules in the reaction cell are identified and
monitored by virtue of their UV/visible absorption. The source
is a 30 W high-brightness deuterium lamp (Hamamatsu L5499)
for use in the UV or a 30 W continuous xenon arc lamp (Applied
Photophysics Ltd.) for visible output. The lamp output is
collimated through the cell and focused, using a pair of 100
mm focal length Spectrosil quartz lenses (Comar Ltd.), onto a
(
Wright Instruments Ltd.). The device is oriented with the 298
pixel axis in the dispersive plane of the spectrograph. The
camera is coupled to the spectrograph with a flange which
allows linear motion in two axes for focusing and vertical
translation of the CCD with respect to the spectrograph output.
A baffle is situated on the window of the detector to minimize
the levels of scattered light falling onto the storage region of
the CCD. A shutter in the camera head is also used, to prevent
buildup of photocharge on the device between experiments.
5
50 µm diameter fused silica optical fiber (Ensign-Bickford).
The fibre is f-matched and coupled to a 250 mm focal length
astigmatic Czerny-Turner spectrograph (Chromex 250IS) which
images the dispersed light onto the CCD.
The spectrograph, camera, and triggering are all controlled
from a personal computer. Data files of the transmitted intensity
as a function of wavelength and time are recorded on this
computer, archived onto data storage tape, and transferred to a
separate personal computer for analysis. Two types of spectral
analysis are performed: spectra of gas mixtures in the cell
without photolysis are recorded by ratioing equivalent ac-
cumulated columns of pixels from successive CCD exposures
with and without the gas mixture, and applying Beer’s law:
The spectrograph is equipped with three interchangeable
diffraction gratings, ruled at 150, 300, and 600 grooves/mm,
giving spectral coverages of ca. 120, 60, and 30 nm, respec-
tively. The corresponding limiting spectra resolutions of each
grating were ca. 0.45, 0.22, and 0.11 nm fwhm, respectively.
The instrument function of the spectrograph was recorded for
each grating at a range of slit widths from 10 to 1000 µm. Each
grating was periodically wavelength and dispersion calibrated
using lines emitted from a low-pressure mercury lamp.
The maximum temporal resolution between successive spectra
recorded with the CCD is determined by the minimum charge
shift time of the device and the size of the image of the analysis
UV/visible source at the detector in the axis of fast charge
A ) ln(I /I )
(i)
λ
o,λ g,λ
wherein Aλ is the absorbance of the gas mixture at a given
wavelength λ, Io,λ is the total intensity recorded by a column of

BrO + BrO Reaction at 298 K
J. Phys. Chem., Vol. 100, No. 8, 1996 3023
pixels in the absence of the gas mixture, and Ig,λ is the intensity
recorded with the gas mixture in the cell.
For kinetic (flashed) experiments, Beer’s law is applied
similarly, but with 〈Io,λ〉 as an average intensity recorded in the
preflash period and Ig,λ,t that recorded for each pixel recorded
at a time t after the photolysis flash.
Uncorrected, this leads to small deviations in values returned,
compared to those obtained from the “pure” kinetic analysis.
Corrections to the analyses were therefore developed and have
been applied in both cases. These corrections involve the time
averaging of a simulated decay trace, in an identical way to
that which occurs optically at the CCD, followed by the fitting
of the simulated and averaged decay trace to the observed decay.
An initial kinetic study, carried out to characterize the new
system, is described below.
Aλ,t ) ln(〈I 〉/Ig,λ,t)
(ii)
o,λ
wherein Aλ,t is the absorption relative to the preflash intensity
at a wavelength λ and time t after the photolysis flash. Spectra
therefore represent changes in the absorptions caused by the
photolysis flash as a result of reactant consumption and radical
formation. Sequential spectra show the temporal evolution of
these changes.
Analysis of the measured absorption spectra to determine the
concentrations of the expected absorbers was carried out by
numerically fitting reference cross sections of these species to
the experimental spectra. In this process, the Beer-Lambert
law is used to construct a simulated spectrum from a linear
combination of the expected absorbances:
The BrO Self-Reaction
The self-reactions of halogen monoxide (XO) radicals have
been studied by many groups using a variety of techniques.
13,14
The reactions have a number of product channels, and the
resulting complications in the kinetic analysis of such systems
have led to a wide range of different results. The widespread
use of UV absorption spectroscopy to monitor the radicals, Via
the characteristic structured (A r X) absorption, means that
some of the discrepancy beween results can be attributed to
the cross sections used, which are a function of the instrumental
resolution. For experiments using PMT detection, single-
wavelength monitoring is also susceptible to interferences by
underlying absorptions generated in the reaction. As shown
Bλ ) RσA,λ + âσB,λ + ...
(iii)
5
by Mauldin et al., the use of diode array spectroscopy enables
wherein R ) l[A]; â ) l[B]; Bλ is the simulated absorption at
wavelength λ; σA,λ, σB,λ, ... are the absorption cross sections
for species A, B, ..., respectively; and l is the optical path length.
This spectrum is then fitted to the experimental spectrum by
minimizing the sum of squares of the difference between the
simulated and experimental spectra with respect to the coef-
ficients R, â, ..., allowing the determination of concentrations
the specific identification and quantification of halogen mon-
oxide radicals by using the whole structured UV spectrum. In
this experiment, sequential absorption spectra are recorded
during an experiment, thereby precluding the need to conduct
separate experiments for each time-resolved data point.
BrO radicals play important roles in the atmosphere. Cata-
lytic cycles involving BrO are implicated in the observed
depletions of stratospheric ozone in polar regions and over
midlatitudes. Tropospheric ozone is also affected by bromine
oxide chemistry: for example a near-total loss of surface ozone
in marine polar regions has been observed, which has been
[A], [B], .... Often, spectral regions can be chosen such that
individual absorptions show little overlap and simple scaling
of reference spectra in spectral regions exclusive to that absorber
can be used to determine its concentration.
If a spectral contribution is structured, a differential technique
can be used to determine the concentration of the structured
absorber in the presence of other smooth absorptions. Briefly,
this involves fitting a function to the smooth shape of the
composite absorption and subtracting this expression to leave
the spectral structure alone: the differential spectrum. The
reference spectrum is then treated similarly, and the reference
differential is fitted to the observed differential using linear least-
squares techniques as described above. The differential spec-
trum may also be shifted by small increments in the wavelength
axis, to optimize the fit of the reference spectrum and account
for any small deviations in the wavelength calibration.
The reference spectra for absorbing species used in the
spectral stripping processes are either recorded or taken from
the literature. In the latter case, where required, special care is
taken to smooth the literature spectrum to the same wavelength
resolution as used in this study. This is particularly important
for the structured spectra, where the values of σλ are a strong
function of the spectrometer resolution.
16
correlated with bromine emissions from phytoplankton. Here,
the BrO self-reaction has been directly implicated in the ozone
loss mechanism:
BrO + BrO f 2 Br + O2
(Br + O f BrO + O )
(1a)
(2)
2
3
2
net:
2O f 3O2
3
Modeling these atmospheric processes therefore requires a
knowledge of rate constants for all the channels involved.
Furthermore, in the kinetic studies of the halogen oxide cross
reactions (e.g. BrO + ClO) it is first necessary to fully
characterize the self-reactions, as they will also contribute to
the observed radical loss in the kinetic experiment.
Previous studies of the BrO self-reaction have shown that
two channels exist: one producing atomic bromine (reaction
1
a) and the other producing molecular bromine:
Having converted the measured absorptions into species
concentration Vs time data for the species of interest, kinetic
analysis is undertaken. In cases where the differential equations
governing the temporal variation of measured species are
soluble, a classical analysis can be performed. With more
complex reaction systems, the commercial software package
FACSIMILE is used to numerically integrate the differential
equations and fit the observed data to a simulated scheme,
allowing the optimization of rate constants. A minor complica-
tion to both types of analysis is encountered as a result of the
inherent time averaging of data from the CCD, due to the finite
vertical extent of the image of the analysis source on the device.
BrO + BrO f Br + O2
(1b)
2
Reported values for k1 ) k1a + k1b are in the range (0.66-5.2)
-12
3
-1 -1 5,17-25
× 10
cm molecule s .
Channel 1a is considered
to be the dominant pathway for the self-reaction of BrO radicals,
1
2
17-25
with both direct and indirect studies
giving an average R
5
) {k1a/(k1a + k1b)} ) 0.85 ( 0.03. Thus, although the
partitioning of the two channels is well-established, the overall
rate of the self-reaction is less so.
In this study, the self-reaction was studied using two separate
sources of BrO radicals. In the first, the short wavelength (λ

3
024 J. Phys. Chem., Vol. 100, No. 8, 1996
Rowley et al.
Figure 3. Differential fit of a single postflash absorption spectrum
thick line) to the BrO reference spectrum (fine line). The residuals
(
have been offset by -0.03 absorbance units, for clarity. Instrumental
resolution ) 1.7 nm fwhm.
1
9
<
200 nm) photolysis of molecular oxygen ([O2] ) 2.4 × 10
cm ) in the presence of molecular bromine ([Br2] > 9 × 10
cm ) was used,
-
3
16
-3
O + hν f 2O
(8)
(3)
2
O + Br f BrO + Br
2
whereas in the second system, the longer wavelength (λ ≈ 400
Figure 4. (a, top) Typical experimental buildup and decay trace for
16
-3
nm) photolysis of bromine ([Br2] ) (1-5) × 10 cm ) in the
[
BrO] (data points) with best-fit simulated trace from FACSIMILE
15
17
-3
13
presence of ozone ([O3] ) 5 × 10 to 7 × 10 cm ) was
(solid line). The residuals have been offset by -5.0 × 10 molecules
-3
used,
cm , for clarity. (b, bottom) Second-order plot of the postflash data
from (a).
Br + hν f 2Br
(9)
(2)
2
spectral structure with sufficiently large differential cross
sections in this wavelength region, the retrieval used the tail of
the Hartley band to determine the changes in ozone concentra-
tion. After adding in the contribution of ozone to the spectra
from the bromine/ozone system, the residual spectrum showed
no other significant absorbance at any point in the decay trace:
the absorption cross sections of bromine are too small over the
wavelength range used to show Br2 consumption. Similarly,
in the oxygen photolysis system, no absorbance other than that
due to BrO was observed.
Br + O f BrO + O
3
2
In both cases, the molecular species was maintained in excess
over the photolytically generated atoms, and BrO radicals were
formed on a much shorter time scale than their subsequent
1
3
removal. Initial BrO concentrations varied from ca. 5 × 10
-3
14
-3
cm in the oxygen photolysis system to ca. 4 × 10 cm in
the bromine photolysis system, with photolysis flash energies
ranging from 470 to 690 J. As discussed below, the use of
separate sources for the radicals enabled the direct characteriza-
tion of both reaction channels.
Results/Discussion
The BrO radicals were monitored as a function of time using
the CCD as described above. Post flash absorbance Vs
wavelength Vs time traces were recorded over the wavelength
range 234-367 nm, at a spectral resolution of 1.7 nm fwhm
and for total periods ranging from 60 to 400 ms. Where
necessary, the signal-to-noise ratio of the traces was improved
by coaddition of up to 15 experiments.
Decay traces were recorded under a range of conditions at
98 K and were generally well-described by a second-order
kinetic scheme. Thus,
2
-
d[BrO]
dt
2
) 2kobs[BrO]
(iv)
Traces were analyzed for BrO using the absorption cross
section values of Wahner et al.26 The spectrum was convolved
onto our instrument function using a sliding Gaussian average.
This function was chosen, as it gave the best reproduction of
our instrument function. A differential spectrum was produced
from the convolved BrO spectrum and fitted and stripped from
the experimental spectrum in each temporal recording from the
CCD. Wavelength shift routines were used to finely match the
wavelength calibration of the reference and experimental spectra.
A typical fit to the BrO differential spectrum is shown in Figure
and, integrating,
1
1
)
+ 2kobst
(v)
[
BrO]t
[BrO]0
A typical decay trace and plot of 1/[BrO]t Vs t are shown in
Figure 4. This type of classical kinetic analysis and numerical
integration and fitting using FACSIMILE were used to return
values of kobs. The basic reaction scheme used for the numerical
integration in FACSIMILE is shown in Table 1.
3. After removal of the contribution of BrO to the experimental
spectra in the bromine/ozone photolysis system, the residual
absorbance spectra showed negative absorptions attributed to
the consumption of ozone during the experiment. A reference
ozone spectrum was therefore used, again convolved to our
instrument function, to retrieve the change in ozone concentra-
tion as a function of time. Since the ozone spectrum shows no
Values of kobs were obtained from both chemical systems.
In the case of the oxygen photolysis experiments, this value
represented the BrO self-reaction through both channels: kobs
) k1a + k1b. In the bromine photolysis system, however, the
measured rate constant resulted from channel 1b only, as BrO
radicals were rapidly regenerated, Via reaction 2, from bromine
2
7

BrO + BrO Reaction at 298 K
J. Phys. Chem., Vol. 100, No. 8, 1996 3025
TABLE 1: Reaction Scheme Used for Numerical Integration
and include a 10% uncertainty in the BrO cross sections, as
reported by Wahner et al.
of the BrO + BrO Reaction
27
3
-1
reaction
k
(298 K)/cm molecule s-1 reference
At the limits of ozone concentration used in the Br2/O3
photolysis system, some deviations from the simple kinetic
a
b
-11
O + Br
Br + O
2
f BrO + Br
1.4 × 10
1.2 × 10
14
13
this work
this work
17
1
6
-12
scheme shown in Table 1 were noted. At very low (<10 /
3
f BrO + O
2
3
BrO + BrO f 2Br + O
2
optimized
optimized
7.6 × 10
cm ) ozone concentrations the phenomenological rate constant
BrO + BrO f Br
Br + Br + M f Br
2
+ O
+ M
2
kobs increased, although the BrO decay did not deviate from
-
14c
2
second-order. By analogy with chlorine, and as originally
a
In oxygen photolysis system only. b In bromine photolysis system
25
suggested by Bridier et al., this is attributed to the Br + BrO
only. c At 760 Torr in nitrogen.
reaction, followed by the reaction of Br2O and Br:
Br + BrO + M f Br O + M
(10)
(11)
2
Br + Br O f Br + BrO
2
2
net:
Br + Br + M f Br + M
2
These reactions, along with the Br + Br + M termolecular
combination, effectively reduce the number of Br atoms
available for reconversion back into BrO Via reaction 2. The
observed BrO loss rate is therefore enhanced due to a contribu-
tion from channel 1a. Given this proposed mechanism, any
significant buildup of Br2O would cause a deviation from
second-order kinetics, but this was not observed. Br2O has a
strong absorption spectrum with two peaks around the wave-
2
8,29
length region used in this work:
section of 2 × 10
one at 200 nm with a cross
-17
2
-1
cm molecule and one at 314 nm with
-18
2
-1
a peak cross section of 2.3 × 10
cm molecule . No
evidence for Br2O was observed in the residual spectra after
removal of the BrO and O3 contributions, allowing an upper
1
3
-3
limit of 1 × 10 molecules cm to be placed upon its
concentration. This result implies that the Br + Br2O reaction
is fast: no previous measurements of this reaction rate exist,
but the chlorine analogue reaction Cl + Cl2O is extremely rapid,
-11
3
-1 -1 30
with k298 ) 9.9 × 10
cm molecule s . These findings
were confirmed by adding reactions 10 and 11 to the reaction
scheme for numerical integration and fitting using FACSIMILE.
It was found that, provided that k11 was greater than k10 by a
factor of 20, the simulated Br2O concentration was always low
<10 molecules cm ) and the decay traces were well-
described by the simulation. The returned values of k10 lay in
Figure 5. (a, top) Typical experimental trace of the change in ozone
concentration following the photolysis flash (data points), with best-
fit simulated trace from FACSIMILE (solid line). The residuals have
1
4
-3
been offset by +5 × 10 molecules cm , for clarity. (b, bottom)
1
3
-3
(
Arithmetic analysis of the ozone concentration changes, after Sander
1
7
and Watson.
-12
3
-1 -1
the range (2-4) × 10
cm molecule s , not inconsistent
3
1
-12
3
with the value reported by Hayman et al. of 1.9 × 10 cm
atoms formed in channel 1a. Thus, by coupling these results,
the branching ratio of the BrO self-reaction was computed.
In addition, for experiments in ozone, the consumption of
ozone by bromine atoms generated in channel 1a was also
related to the branching ratio, providing a second check on this
quantity. Again, both numerical integration and an analytical
solution could be used to relate R to the temporally varying
ozone concentration. The latter analysis, after Sander and
molecule- s . Furthermore, the simulations were found to
1
-1
be very insensitive to k11, above a lower limit of k11 of 4 ×
-11
3
-1 -1
1
0
cm molecule s .
1
7
3
At high (>10 /cm ) ozone concentrations, some deviations
from the second-order kinetic scheme were noted, which can
be seen clearly in the second-order plot in Figure 6. This
deviation is indicative of another reaction process for BrO, in
addition to the second-order removal. The extent of the
deviation from second-order was also found to increase with
increasing ozone concentration. This behavior appears consis-
tent with a contribution to the decay of BrO from the slow BrO
+ O3 reaction, giving products other than bromine atoms:
1
8
Watson, gives
k1b
1
1
1
t
)
+
(vi)
2
( )
∆
[O3]t k1a[BrO]0
2k [BrO]
0
1
a
where ∆[O3]t is the change in ozone concentration at time t
from that after the initial production of BrO (t ) 0). Thus, a
plot of 1/∆[O3]t Vs 1/t gives k1a and the branching ratio. Both
analyses are shown in Figure 5.
BrO + O f products
(12)
3
Reaction 12 was therefore incorporated into the reaction scheme
for FACSIMILE to test this hypothesis. As can be seen from
Figure 6, the agreement of the experimental and best-fit
simulated BrO decays was somewhat improved when including
BrO + O3. Furthermore, the value of k12 returned from the
simulation was independent of the ozone concentration used,
supporting the hypothesis that the deviation from second-order
is due to this channel. The average value of k12 obtained was
Results for the BrO self-reaction are summarized in Table 2
along with recent values taken from the literature. As can be
seen, the values obtained in this study are in excellent agreement
5
with those obtained from the recent study of Mauldin et al.,
who used a combined PMT/diode array system to measure k1b
and R. Errors quoted refer to statistical scatter at the 2σ level

3
026 J. Phys. Chem., Vol. 100, No. 8, 1996
TABLE 2: Measured Rate Constants and Comparison with Recent Previous Studies
number of
Rowley et al.
/10-
12a
k
1a/10-12a
k
1b/10-13a
R
determinations
source/reference
k
1
2
.98 ( 0.42b
20
10
oxygen photolysis system
bromine photolysis
system:[BrO] monitoring
bromine photolysis
4
.68 ( 0.68b
2
.49 ( 0.42c
0.85 ( 0.02c
0.84 ( 0.01
0.84 ( 0.03
10
system:[O
Mauldin et al.
Bridier et al.
DeMore et al. (review)
3
] monitoring
5
2
3
2
.78 ( 0.24
4.45 ( 0.82
4.4 ( 1.1
38
9
2
5
.1 ( 0.4
.7 ( 0.7
1
3
a
Units of cm molecule-1 s-1. b From analysis using FACSIMILE. c From graphical analysis.
3
Figure 6. (a, top) Experimental BrO decay at high [O
with best fit FACSIMILE simulations using (i) pure second-order
reaction scheme and (ii) reaction scheme including BrO + O f prod-
ucts. The residuals from both fits are shown, expanded ×4 and offset
3
] (data points),
Figure 7. (a, top) Differential fit of a single postflash absorption
spectrum (thick line) to the OBrO reference spectrum (fine line). The
residuals have been offset by -0.01 absorbance units, for clarity.
Instrumental resolution ) 1.7 nm fwhm. (b, bottom) Experimental
trace of the buildup of OBrO.
3
14
-3
by +3.0 × 10 molecules cm , for clarity. Residual from simulation
i) ) open symbols; residual from simulation (ii) ) filled symbols. (b,
bottom) Second-order plots of data and simulations from (a).
(
the absolute absorption cross sections of OBrO have not been
measured, the amount of OBrO evolved could not be deter-
mined. However, the differential fitting of a scaled relative
spectrum to successive postflash spectra enabled the determi-
nation of the temporal variation of the relative OBrO amounts.
Figure 7 shows a typical differential fit to the OBrO absorption
and the temporal variation of the relative concentration.
The relative amount of OBrO evolved was found to increase
with increasing ozone concentrations over the range (1-7) ×
-
17
cm molecule s-1, with k1b ) (4.57 (
3
-1
(
0
2.1 ( 0.7) × 10
-13
3
-1 -1
.80) × 10
cm molecule
s
(errors are 2σ), which is in
good agreement with the value obtained at lower ozone
concentrations, wherein no deviation from second-order was
observed. It is concluded that the reaction of BrO with ozone
is playing a role in this reaction system at high (>10 molecules
cm ) ozone concentrations.
The molecular products of reactions 1a and 1b were not
quantified by UV spectroscopy, and no other products were
observed in the wavelength range 234-367 nm in either the
O2 or the Br2 photolysis system. However, recent work by
Rattigan et al. on the bromine photosensitized decomposition
of ozone has identified the product OBrO and assigned its gas
phase visible absorption spectrum for the first time. This
spectrum extends from 400 to 600 nm and is highly structured,
with pairs of absorption bands. A search for this absorber was
therefore conducted in this work, as a potential reaction product.
The wavelength region 425-558 nm was selected, again using
the 150 grooves/mm grating. OBrO was not observed in the
oxygen photolysis system, but experiments using bromine
photolysis in ozone did show OBrO in the postflash spectra
17
-3
1
7
-3
1
0
cm . This strongly suggests that the OBrO could be a
33
product of reaction 12. However, as Butkovskaya et al. report,
the reaction of bromine atoms with OBrO is fast (k13(298 K) ) 5
-
11
3
-1 -1
×
10
cm molecule s ), and this effect could simply be
3
2
a reflection of the lower steady state bromine atom concentration
at the higher ozone concentrations:
Br + OBrO f 2BrO
(13)
To determine the source of OBrO, experiments to monitor the
formation of OBrO were carried out over ozone concentrations
1
7
-3
in the range (1-7) × 10 cm . At each ozone concentration,
experiments were also carried out to monitor BrO, and
FACSIMILE was used to model the results to attempt to
establish which reaction was the source of OBrO. Since the
17
-3
when ozone concentrations were high (>1 × 10 cm ). Since

BrO + BrO Reaction at 298 K
J. Phys. Chem., Vol. 100, No. 8, 1996 3027
OBrO cross sections are unknown, the cross section values were
assumed for the modeling on the basis of literature OClO cross
sections and the instrumental resolution used for this study. The
assumed values related to a differential cross section of 1 ×
-17
2
-1
1
0
cm molecule between the peak at 495.5 nm (corre-
3
2
sponding to a (Vn,0,0) r (0,0,0) symmetric stretch transition )
and the trough to short wavelengths at 491.1 nm, at the
experimental instrumental resolution of 1.7 nm fwhm.
The FACSIMILE simulation was set up with OBrO either
as a product of the BrO self-reaction (reaction 1c) or as a product
of the BrO + O3 reaction (reaction 12). However, it was found
that neither system could adequately describe the evolution
kinetics, because the OBrO formation occurred on a longer time
scale than the BrO decay.
Figure 8. Final [OBrO] variation with ozone concentration (filled
squares) and behavior predicted from simulations including OBrO
formed in the BrO self-reaction (open circles) and OBrO formed in
BrO + BrO f OBrO + Br
BrO + O f OBrO + O
(1c)
(12)
3
the BrO + O reaction (open triangles).
of OBrO. As can be seen, the variation of the maximum relative
OBrO] generated is well-described by the mechanism using
3
2
[
We therefore postulate here that an additional reaction of OBrO
is taking place: the reversible combination with BrO to form
Br2O3.
the BrO + O3 source and not matched if OBrO is produced in
the self-reaction, providing further confirmation of this finding.
Given the proposed secondary chemistry of OBrO, particu-
larly the formation of Br2O3, the analysis of the BrO decay traces
recorded at high [O3] deserves reconsideration. Should the BrO
BrO + OBrO h Br O
(14, -14)
2
3
+
O3 reaction lead exclusively to OBrO, which then combines
which is analogous to the ClO + OClO reactions.34 Although
no absorption attributable to Br2O3 was observed, the inclusion
of reaction 14, -14 in the chemical scheme for integration did
allow the observed behavior to be well-described by the
simulation using either reaction 1c or 12 as the source of OBrO.
However, in fixing the BrO self-reaction as the source of OBrO
the value of k1c returned was found to change markedly with
ozone concentration. Conversely, when setting reaction 12 as
the source of OBrO, the value obtained for the rate coefficient
k12 was independent of the ozone concentration. This strongly
suggests that the source of OBrO is reaction 12.
relatively rapidly with further BrO to produce Br2O3, the actual
rate for the BrO + O3 reaction will be 50% of that returned
from the BrO decay kinetics. Furthermore, this rate is also based
on the assumption that the deviation from second-order BrO
decay kinetics is exclusively due to the BrO + ozone reaction
producing non-bromine atom products: other first-order BrO
loss processes could also contribute to this value. Given these
-17
3
-1 -1
assumptions, the value of 2.1 × 10
cm molecule
s
for
the BrO + O3 reaction rate should be taken as an upper limit.
Results for the analysis of the secondary chemistry at both
low and high ozone concentrations are summarized in Table 3.
The results obtained for the individual reaction channels of
the BrO self-reaction in this study show excellent consistency,
regardless of the measurement used to derive values, and agree
The averaged values of the rate coefficients k12, k14 and the
equilibrium constant K14 obtained from FACSIMILE fitting for
each simulation were
with the most recent determinations of the kinetics of this
k ) (1.66 ( 0.11) × 10 cm molecule s-1
-
18
3
-1
5,23-24
12
system.
The reaction products can be explained by the
1
mechanism originally proposed by Porter, involving the
metastable BrOOBr intermediate. This species then forms
unstable BrOO, which subsequently decomposes to form the
atomic products or rearranges to form the molecular products.
k ) (6.3 ( 6.0) × 10 cm molecule s-1
-11
3
-1
14
-
14
3
-1
K ) (4.2 ( 4.0) × 10 cm molecule
14
BrO + BrO T BrOOBr*
BrOOBr* f BrOO + Br
Errors are 2σ and take no account of the systematic errors
involved in the assumption of OBrO cross sections, which affect
the returned value of k12 directly and inversely. Notwithstanding
the large uncertainty in the value of K14, a preliminary
thermodynamic analysis was carried out to assess the feasibility
2
f Br + O2
3
5
0
of this equilibrium. The values of S for OBrO and Br2O3
were estimated from those of OClO and Cl2O3, with adjustments
made due to the difference in the translational partition
BrOO f Br + O2
The absence of OBrO as a product of the self-reaction suggests
that the metastable BrOBrO* intermediate does not form or
rearranges to BrOOBr.
3
5,36
-1
function.
The resulting ∆H° obtained was -65.1 kJ mol ,
considerably more exothermic than that for the chlorine analogue
compounds, for which ∆H° ) -47.6 kJ mol .
-
1 34
Thus,
In reaction systems where insufficient ozone was present to
scavenge bromine atoms, the termolecular Br + BrO + M
addition reaction forms Br2O. It is followed in this system by
the fast reaction of Br with Br2O to regenerate Br2, serving to
convert bromine atoms into Br2 and prevent buildup of Br2O,
as observed. These reactions are analogous to those of the
although Br2O3 is somewhat more stable than Cl2O3, the
difference in enthalpies of formation is not unreasonable.
The values of the rate coefficients obtained from the analysis
of the OBrO evolution kinetics were also used to simulate the
variation with ozone concentration of the maximum OBrO
concentration generated in each experiment. The results are
shown in Figure 8, along with the observed variation (assuming
the OBrO cross sections as discussed above) and that which is
predicted from simulations in which reaction 1c was the source
1
3,37
chlorine counterparts.
Some experimental evidence for the
Br + Br2O reaction has since been observed by Orlando and
Burkholder, who observed BrO in the photolysis of Br2 in
the presence of Br2O.
2
8

3
028 J. Phys. Chem., Vol. 100, No. 8, 1996
Rowley et al.
TABLE 3: Measured Rate Coefficients from Numerical
deviations from the expected kinetic scheme, and the monitoring
of a range of wavelengths allows the ready identification of
additional reaction products.
The novel flash photolysis system described in this work
should prove to be a useful tool for the study of other gas phase
reactions of UV/visible absorbing species. This study of the
BrO self-reaction demonstrates the viability of the technique
and reveals hitherto undiscovered secondary chemistry of these
important reactive species.
Integration of Secondary Chemistry in the Br
System
2
Photolysis
reaction
Br + BrO f Br
Br + Br O f BrO + Br
BrO + O f products
BrO + BrO f Br + O
k
(298 K)/cm molecule s-1
3
-1
note
-
12
2
O
(2-4) × 10
a
a
b
b
c
c
c
-
11
2
2
>4 × 10
-
17
3
(2.1 ( 0.7) × 10
-
-
13
18
2
2
(4.57 ( 0.80) × 10
(1.66 ( 0.11) × 10
BrO + O
3
f OBrO + O
2
-
11
BrO + OBrO T Br
O
2 3
(6.3 ( 6.0) × 10
-
14d
K ) (4.2 ( 4.0) × 10
Acknowledgment. This work was supported by the NERC
Atmospheric Chemistry Special Topic (Contract GST/02/877
A) and the EC Environment programme (Contract EV5V-CT93-
338). We thank all of the technical staff at Cambridge
University Chemistry Department who have participated in the
apparatus development. We thank Oliver Rattigan and Tony
Cox for the communication of results prior to publication and
for several useful discussions. We thank Deb Fish for assistance
with the analysis software development.
a
] (<1016 cm-3) experiments only. In high [O
b
17
In low [O
3
3
] (>10
cm- ) experiments only, monitoring BrO. In high [O
3
c
] (>10 cm
17
-3
)
3
experiments only, monitoring OBrO and using assumed cross sections
for OBrO (see text for details). Units of cm molecule . All errors
d
3
-1
quoted are 2σ and represent statistical scatter only.
In previous works, the products of the reaction of BrO with
ozone have been assumed to be exclusively bromine atoms and
5,17
molecular oxygen.
This results from the lack of any observed
deviation from second-order kinetics in BrO decays recorded
under conditions of excess ozone. In this work, such deviations
have been observed, however, as a result of the increased signal-
to-noise ratio afforded by the multiple wavelength monitoring.
Furthermore, the identification of OBrO in experiments by
References and Notes
(1) Porter, G. Proc. R. Soc. London Ser. A 1950, 200, 284.
(2) Norrish, R. W. G.; Porter, G.; Thrush, B. A. Proc. R. Soc. London
Ser. A 1953, 216, 165.
(
(
3) Porter, G.; West, P. Proc. R. Soc. Ser. A 1964, 279, 302.
4) Sander, S. P.; Friedl, R. R. J. Phys. Chem. 1989, 93, 4764.
3
2
Rattigan and Cox provides a suitable candidate for other
products of this reaction. Observation and modeling of the
OBrO buildup have confirmed this, provided the OBrO under-
goes reversible association with BrO to form Br2O3. Again,
(5) Mauldin, R. L., III; Wahner, A.; Ravishankara, A. R. J. Phys. Chem.
1993, 97, 7585.
6) Harwood, M. H.; Jones, R. L. J. Geophys. Res. 1994, 99, 22955.
7) Rattigan, O. V.; Rowley, D. M.; Wild, O.; Jones, R. L.; Cox, R.
(
(
3
4
the analogous chlorine system is a precedent for this. The
absence of any absorption attributable to Br2O3 could indicate
that this product does not absorb in the wavelength regions used
for these experiments, i.e. 234-367 and 425-558 nm, or that
only a small quantity is formed that is below our detection limits.
A. J. Chem. Soc., Faraday Trans. 1994, 90, 1819.
(8) Dereniak, E. L.; Crowe, D. G. Optical Radiation Detectors; John
Wiley & Sons: New York, 1984; Chapter 9.
(
9) Markiewicz, J. P.; Emmet, J. L. J. Quantum Electron. 1966, QE-2
(
11).
(10) Phillips, R. Sources and Applications of UltraViolet Radiation;
Academic Press Inc.: London, 1983.
(11) Harwood, M. H.; Jones, R. L.; Cox, R. A.; Lutman, E.; Rattigan,
O. V. J. Photochem. Photobiol. A: Chem. 1993, 73, 167.
The mechanism of OBrO formation in the BrO + O3 reaction
remains unclear. Turnipseed et al.23 speculate that the formation
of a weak adduct of BrO and O3 at low temperatures explains
the discrepancy between their results obtained for the BrO self-
reaction using different radical sources. Further experiments
as a function of pressure and temperature are needed to resolve
this.
(
12) Curtis, A. R.; Sweetenham, W. P. FACSIMILE, AERE Harwell
publication R 12805; Computer Science and Systems Division, Harwell
Laboratory: Oxfordshire, U.K., 1987.
(13) DeMore, W. B.; Sander, S. P.; Golden, D. M.; Hampson, R. F.;
Kurylo, M. J.; Howard, C. J.; Ravishankara, A. R.; Kolb, C. E.; Molina,
M. J. Chemical Kinetics and Photochemical Data for Use in Stratospheric
Modeling; JPL publication 92-20; California Institute of Technology:
Pasadena, 1992.
Conclusions
(14) Atkinson, R.; Baulch, D. L.; Cox, R. A.; Hampson, R. F.; Kerr, J.
A.; Troe, J. Evaluated Kinetic and photochemical Data for Atmospheric
Chemistry-Supplement IV, IUPAC Subcommittee on Gas Kinetic Data
Evaluation for Atmospheric Chemistry. J. Phys. Chem. Ref. Data 1992,
21, 1125.
A novel flash photolysis system has been developed using a
charge-coupled device (CCD) detector. This represents the first
such use of these devices in studies of gas phase kinetics and
enables sequential absorption spectra to be recorded from the
reacting mixture on microsecond time scales.
(15) Scientific Assessment of Ozone Depletion, 1994. WMO Global
Ozone Research and Monitoring Project, Report No. 37; World Meteoro-
logical Organisation: Geneva, 1994.
Preliminary studies of the BrO self-reaction have been carried
out to demonstrate the capability of the system in kinetics
studies. The recording of multiple wavelengths enables the
simultaneous monitoring of several UV/visible absorbing com-
ponents in the reaction mixture. In addition, the unequivocal
and accurate quantification of species gives a high signal-to-
noise ratio without the need for excessive experiments.
This study of the BrO self-reaction at 298 K is an indepen-
dent, direct determination of both reaction channels. Results
are in excellent agreement with previously published work.
Some complications to the kinetics have been noted at extreme
values of experimental ozone concentration used, including the
observation and monitoring of the recently identified OBrO
molecule. Our results indicate that this product is formed in
the BrO + O3 reaction and not the BrO self-reaction. This
characterization of the secondary chemistry highlights the
advantages of the CCD multiple wavelength monitoring: the
data is of sufficient quality to enable the analysis of very small
(16) Barrie, L. A.; Bottenheim, J. W.; Schnell, R. C.; Crutzen, P.;
Rasmussen, R. A. Nature 1988, 334, 138.
(
(
17) Sander, S. P.; Watson, R. T. J. Phys. Chem. 1981, 85, 4000.
18) Clyne, M. A. A.; Cruse, H. W. Trans. Faraday Soc. 1970, 66,
2214.
(19) Basco, N.; Dogra, S. K. Proc. R. Soc. London Ser. A 1971, 323,
4
1
975, 71, 336.
(21) Jaffe, S.; Mainquist, W. K. J. Phys. Chem. 1980, 84, 3277.
(
22) Cox, R. A.; Sheppard, D. W.; Stevens, M. P. J. Photochem. 1982,
9, 189.
23) Turnipseed, A. A.; Birks, J. W.; Calvert, J. W. J. Phys. Chem. 1990,
94, 7477.
1
(
(
24) Lan c¸ ar, I. T.; Laverdet, G.; LeBras, G.; Poulet, G. Int. J. Chem.
Kinet. 1991, 23, 37.
(
25) Bridier, I.; Veyret, B.; Lesclaux, R. Chem. Phys. Lett. 1993, 201,
563.
(26) Wahner, A.; Ravishankara, A. R.; Sander, S. P.; Friedl, R. R. Chem.
Phys. Lett. 1988, 152, 507.
27) Atmospheric Ozone 1985. WMO Global Ozone Research and
(
Monitoring Project, Report No. 16; World Meteorological Organisation:
Geneva, 1985.

BrO + BrO Reaction at 298 K
J. Phys. Chem., Vol. 100, No. 8, 1996 3029
(
(
(
28) Orlando, J. J.; Burkholder, J. B. J. Phys. Chem. 1995, 99, 1143.
29) Rattigan, O. V. Private communication, 1995.
30) Burrows, J. P.; Cox, R. A. J. Chem. Soc., Faraday Trans. 1 1981,
(33) Butkovskaya, N. I.; Morozov, I. I.; Tal’Rose, V. L.; Vasiliev, E.
S. Chem. Phys. 1983, 79, 21.
(34) Burkholder, J. B.; Mauldin, R. L., III; Yokelson, R. J.; Solomon,
S.; Ravishankara, A. R. J. Phys. Chem. 1993, 97, 7597.
7
7, 2465.
(
35) Harwood, M. H. Ph.D. Thesis, University of Cambridge, 1995.
(31) Hayman, G. D.; Danis, F.; Thomas, D. A. In Proceedings of Air
(36) Benson, S. W. Thermochemical Kinetics, 2nd ed.; John Wiley and
Pollution Research Report 45: Chemical Mechanisms Describing Tropo-
spheric Processes; Peeters, J., Ed.; CEC Publications: Leuven, Belgium,
992.
Sons: London, 1976.
37) Nickolaisen, S. L.; Friedl, R. R.; Sander, S. P. J. Phys. Chem. 1994,
8, 155.
(
1
9
(32) Rattigan, O. V.; Jones, R. L.; Cox, R. A. Chem. Phys. Lett. 1994,
230, 121.
JP951825B