JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 101, NO. D17, PAGES 23,021-23,033, OCTOBER 20, 1996
UV-visible absorption cross sections of gaseous Br:O
and HOBr
O.V. Rattigan,  D.J. Law, R.L. Jones, and R.A. Cox
Centre for Atmospheric Science, Department of Chemistry, University of Cambridge,
Cambridge, England
Abstract. The bsorption cross-section of gseous HOBr ws determined over the
wvelength rnge 235 to 430 nm with  spectrM resolution of 0.6 nm full width t
hMf mximum (FWHM) using  diode rry spectrometer. The spectrum of HOBr
shows two mMn bsorption bnds with mxim near 282 nm (cr=(3.1:1:0.4)x10 -9
cm 2 molecule - nd 350 nm (cr=12.5 + 1.6)x10 -2ø cm 2 molecule -) extending
out to 430 nm. The bsorption cross-sections in the first bsorption bnd re in
good agreement with  recent determination; the cross-sections in the second band
however, re pproximtely  fctor of 2.5 lrger thn previously determined. In
ddition we provide evidence in support of  wek bnd in HOBr round 440 nm
(cr7.Sx10 -2 cm 2 molecule -) s observed by Barnes et al. [1996]. The bsorption
cross-section of Br20, which ws used to prepare HOBr, ws determined over the
wvelength rnge 230 to 750 nm. The spectrum shows four absorption bnds with
mxim t 314 nm (or=(2.1 4- 0 3)x10 -15 cm  molecule-l), 350 nm (cr=(li 4- 0.2) 10 -$ cm  molecule-), 52 nm (cr--(4.4 4-0.5)x10 -0 m  molecule- and
665 nm (cr-(6.2 + 0.9)x10 -ø cm  molecule-). The visible bands t 520 nm
nd 660 nm hve not been observed previously. The equilibrium constant, for he
reaction BrO + HO  2HOBr ws determined to be 0.037 4- 0.004 a 298
K. Measurement of the equilibrium constant s  function of emperture enabled
values for AH298 K -- (13.0 + 0.5) kJ mol - and AS298 K -- (16 + 2) J mol - K -
to be determined. The absorption cross-section data for HOBr have been used in
a photochemical box model to investigate the significance of these results in the
lower stratosphere. The model results are compared with observations during a
recent Stratospheric Photochemistry, Aerosols and Dynamics Expedition (SPADE)
and show that the revised HOBr cross-section, coupled to the rapid heterogeneous
conversion of BrONO2 to HOBr, can account quantitatively for the abrupt morning
rise in HO.
Introduction (R2) H20  BrONO2 --+ HOBr + HNO3
Interest in the spectroscopy of atmospheric bromine
species has been stimulated by their role as catalysts in
stratospheric ozone depletion [Yun# et al., 1980]. HOBr
is thought to be a major bromine reservoir produced by
the gas phase reaction of BrO with HO2 (Poulet ½t al.
[10921, Bridier et al. [1003]):
(R1) HO2 +BrO  HOBr+ O2
and by the hydrolysis of BrONO2 which occurs hetero-
geneously on atmospheric aerosol particles (Hanson and
Ravishankara [1995], Lary et al. [1996]):
1Now at Department of Chemistry, Boston College, Chestnut
Hill, Boston, Massachusetts.
Copyright 1996 by the American Geophysical Union.
Paper number 96JD02017.
0148-0227/96/96 JD-02017509.00
Reaction (R2) followed by the photolysis of HOBr
has recently been suggested as a possible source of OH
radicals in the lower stratosphere (Hanson and Ravis-
hankara [1995]).
The heterogeneous reaction between HOBr and HC1
which links the C1 and Br cycles is also thought to be
important particularly in polar regions (Abbatt [1994]):
(R3) HOBr + HC1 --+ BrC1 + H20
At midlatitudes however, the following O3 destruc-
tion catalytic cycle can occur (Yung et al. [1980], Gar-
cia and Solomon [1994]), starting with (R1):
(R1) HO2 +BrO  HOBr+02
(R4) HOBr+hv  OH+Br
(R5) OH+O3 ) HO2 +O2
(R6) Br + O3  BrO + O2
Net: 203  302
23,021

23,022 RATTIGAN ET AL.:UV-VISIBLE ABSORPTION CROSS SECTIONS OF BR20 AND HOBR
At altitudes >30 km HOBr can also be destroyed via
the reaction with O(3P) (Nesbitt et al. [1995]).
(R7) HOBr + O(aP) > BrO + OH
However, despite the importance of HOBr in the at-
mosphere, information on the UV visible absorption
spectrum is rather limited. The first determination
of the UV-visible absorption cross-section of gaseous
HOBr has been made only recently by Orlando and
Burkholder [1995]. As part of a study of bromine ox-
ides we have redetermined the UV visible absorption
cross-sections for gaseous HOBr and Br20. The results
are compared with the recent measurements of Orlando
and Burkholder [1995].
The measured cross-,q.ctions for HOBr have been
used to calculate jaoBr in a photochemical box model
which includes heterogeneous bromine chemistry to as-
sess its potential for OH production in the lower strato-
sphere as suggested byHanson and Ravishankara [1995].
The model results are compared to recent observations
of $alawitch et al. [1994].
Experimental Method
Apparatus
The experimental system consisted of a double-jacketed
quartz cell, 100 cm long x 2.0 cm diameter, which was
coupled to a dual-beam diode array spectrometer (Rat-
tigan et al. [1993]). Two different gratings were used
for the spectral measurements. A 150 grooves per mm
grating with a spectral range of  305 nm dispersed
over a 512 element array was used for spectral measure-
ments of Br20 in the wavelength region 460 to 750 nm
and for HOBr in the wavelength range 263 to 569 nm.
In all other cases a 600 grooves per mm grating with a
spectral range of  75 nm was used. An entrance slit
width of 100 um was used in both cases providing res-
olutions of m 2.5 nm (FWHM) and 0.6 nm full width
half maximum (FWHM) respectively. The higher reso-
lution of the 600 groove per mm grating greatly aided
in the spectral subtraction of Br2 and Br20. With this
grating measurements were made over several different
spectral regions from 235 to 590 nm, ensuring a 10 to
15 nm overlap between adjacent segments. For mea-
surements at wavelengths >300 nm a Pyrex filter was
mounted in the monitoring beam to avoid higher or-
der radiation from reaching the detector. For the Br20
spectrum a filter with a cutoff at 420 nm was used in
the spectral measurements of the band from 550 to 750
nm. Wavelength calibrations were made using emission
lines of Hg, Zn and Cd from a Philips 93145 spectral
lamp and an entrance slit width of 10 um. The accu-
racy of the wavelength calibrations are 0.6 nm for the
150 groove per mm grating and 0.15 nm for the 600
groove per mm grating. Gas pressures were measured
on a calibrated 100 torr Baratron capacitance manome-
ter (MKS Instruments).
Sample Preparation
HOBr was prepared by the addition of H.O vapor to
a sample of Br.O in the absorption cell and allowing
the establishment of the following equilibrium
(RS) Br.O + H.O  2HOBr
Thus from the change in Br20 upon addition of ex-
cess H.O it was possible to determine the HOBr concen-
tration. Since this method relies on an accurate knowl-
edge of the Br.O concentration, it was therefore neces-
sary to first prepare Br.O and measure its absorption
cross section in the absence of HO vapor.
Br.O was prepared by the reaction of gaseous Br.
(Aristar grade 99.95%. BDH Ltd.) with dry yellow
HgO powder (99%, Aldrich) as described by Zintl and
Rieniicker [1930]'
(R9) 2Br2 + HgO > Br20 + HgBr2
Approximately 60 torr of Br2 was frozen onto 3 to 4
g of HgO in a trap at 77 K. The trap was then warmed
to 263 K over a period of 30 to 60 mins after which
the contents were frozen into a second trap at 77 K lo-
cated close to the absorption cell. Since reaction (R9)
was found to yield at most 1-2% Br20 it was neces-
sary to concentrate the sample by repeating the above
procedure four to five times to ensure sufficient Br20
for spectral measurements. Careful distillation of the
second trap contents over a period of several hours at
213 K resulted in a small amount of red/brown material
which was typically 70-80% Br20, the remainder being
Br2. The sample of Br20 obtained from this method
was sufficient for three to four experiments. For spec-
tral measurements the trap at 213 K was allowed to
warm up to room temperature and a sample of Br20
was measured into the absorption cell. In between mea-
surements the sample of Br20 was kept in a darkened
trap at 213 K.
Results
Absorption Spectrum of Br20
In all cases samples of Br20 contained considerable
amounts of Br2. Furthermore, samples of Br20 in the
absorption cell were found to undergo thermal decom-
position into molecular Br2. In order to determine the
spectrum of Br20. it was necessary to quantitatively
subtract the absorption due to Br2 using a reference
spectrum of Br2 recorded with the same spectral reso-
lution. Br2 shows vibrational structure in its absorption
spectrum in the region from 515 to 565 nm due to the
electronic transition B3II(Ou +) < X Eg +. Quantita-
tive subtraction of the Br2 absorption present in each
sample was carried out using a least squares fit to the
differential spectra of the sample containing Br20 and
Br2 and the Br2 reference over this structured region.

RATTIGAN ET AL.:UV-VISIBLE ABSORPTION CROSS SECTIONS OF BR_O AND HOBR 23,023
A typical fit is shown in figure 1 at 298 K. Figure 2
shows a sample Br20/Br2 spectrum (line a), estimated
Br2 (0.02 torr) in the sample using the fitting routine
(line b) and the residual spectrum (Br20 - 0.07 torr)
(line c) after spectral subtraction of (line b) from (line
a) in the region 490 to 565 nm.
The amount of Br20 present in each sample was es-
timated by converting the Br20 into Br2 by photolysis.
Absorption cross sections for Br2 of $eery and Brit-
ton [1964] were used to determine the amount of Br2
present in the samples before and after photolysis and
the Br20 concentration was determined by difference.
Small amounts of BrO m 1-2 x 10 TM molecule cm -a were
detected during the illumination period consistent with
conversion by the following reaction mechanism as pro-
posed by Orlando and Burkholder [1995]'
(R10) Br20 + h  Br + BrO
(Rll) Br + Br20 . > Br2 + BrO
(R12) BrO +BrO  2Br+ 02
(R13) BrO + BrO  Br2 + 02
The above mechanism results in the following overall
stoichiomctric reaction:
2Br20 .  2Br2 4- O2
From a knowledge of the Br20 concentration and the
absorption spectrum, the absorption cross-sections for
Br20 could then be determined using the Beer Lam-
bert Law. A complete spectrum for Br20 was sub-
sequently constructed from 240 to 750 nm by record-
ing spectra in several regions working towards shorter
wavelengths and ensuring an overlap of at least 15 nm
between adjacent segments. Correction for the contri-
bution due to Br2 at shorter wavelengths was carried
out from a knowledge of the spectral shape and the ab-
sorption cross-sections of Br20 and Br2 in the overlap
region.
Figure 3 shows the cross-sections of Br20 determined
by the above described method, compared to the recent
0.015
0.010
0.005
0.00
'  I ' ' ' I ' ' ' I ' ' ' I ' ' '
, , , I  , , I , , , I , , , I , , ,
80 500 520 540 560
Wavelength/nm
580
Figure 2. Sample spectrum containing approximately
0.1 torr of Br20 + Br2 (line a), Br2 reference spectrum
(line b) and the Br20 spectrum (line c) after subtrac-
tion of (line b) from (line a). Instrument resolution is
0.6 nm (FWHM).
values reported by Orlando and Burkholder [1995]. Tab-
ulated values are given in Table 1. There is very good
agreement (to within m 10%) between the two studies
at wavelengths horter than 400 nm. The absorption
cross section at the maximum near 314 nm= (2.1 4-
0.3)x10 -s cm 2 molecule - is in good agreement with
the earlier value of (2.3 4- 0.3)x10 -is cm 2 molecule -.
However, at wavelengths greater than 400 nm the val-
ues fi'om this study are significantly higher than those
of Orlando and Burkholder which show a cutoff at m
440 nm. The difference between the two measurements
is approximately an order of magnitude at 430 nm. Or-
lando and Burkholder used a shorter path length cell
of 20 cm and their observed residuals after subtraction
of Br2 from their spectra were m 5x10 -4 (at the de-
tection limit) at wavelengths greater than 440 nm. In
the subtraction of Br2 from their samples of Br20 they
assumed that the absorbance due to Br20 was zero at
440 nm and beyond. In the present experiments (path
length  100 cm) the procedure used to correct for Br2
' 0.002
\" I .... \"'\"\" .... ' .... ' ....
'-' OOOl
 ' 1 o '8 _
o ooo
 -0 001  10 '9 ß -
 -0.002 
-o 520 535 550 565 u 0.20 / o I , , , , I , , , , I .... I , , , , I , , , ,
wavelength/rim c 250 350 450 550 650 750
wavelength/nm Figure 1. Differential fit of a sample spectrum con-
taining Br20 + Br2 (dashed line) to the Br2 reference
spectrum (solid line) and the residuals after subtraction
(thick line). Instrument resolution is 0.6 nm (FWHM).
Figure 3. Absorption cross-sections of Br20 at 298 K
determined in this work (solid line) and selected ata
of Orlando and Burkholder [1995] (squares).

23,024 RATTIGAN ET AL.:UV-VISIBLE ABSORPTION CROSS SECTIONS OF BR20 AND HOBR
Table 1. Absorption cross-sections of Br.O and HOBr at 298 K
Wavelength (nm) Br20 (10 -xs cm 2 molecule -x)
Orlando  Burkholder [1995] This Work
HOBr (10 -20 cm 2 molecule -x)
Orlando  Burkholder [1995] This Work
240 0.869 0.805
245 0.630 0.601
250 0.443 0.429
255 0.322 0.303
260 0.249 0.218
265 0.211 0.169
270 0.206 0.152
275 0.252 0.182
280 0.361 0.279
285 0.571 0.470
290 0.883 0.770
295 1.30 1.16
300 1.72 1.55
305 2.07 1.86
310 2.26 2.04
315 2.29 2.08
320 2.20 1.95
325 2.08 1.87
330 1.99 1.78
335 1.94 1.74
340 1.94 1.74
345 1.94 1.74
350 1.91 1.71
355 1.82 1.64
360 1.64 1.53
365 1.51 1.38
370 1.36 1.23
375 1.15 1.09
380 0.988 0.971
385 0.835 0.868
390 0.704 0.776
395 0.600 0.689
400 0.487 0.603
405 0.382 0.519
410 0.283 0.439
415 0.201 0.365
420 0.126 0.298
425 0.070 0.241
430 0.023 0.194
435 0.155
440 0.123
445 0.099
450 0.079
455 0.063
460 0.051
465 0.044
470 0.038
475 0.035
480 0.035
485 0.035
490 0.036
495 0.038
500 0.040
505 0.042
510 0.043
515 0.045
0.43
1.88
4.53
8.24
12.9
18.5
24.0
28.5
30.8
30.4
27.3
22.3
16.4
10.5
7.02
4.78
3.95
4.19
4.89
5.34
5.69
5.98
5.98
5.89
5.70
5.22
4.53
3.81
3.02
2.25
1.67
1.08
0.18
0.29
0.0
6.68
5.22
6.74
9.93
14.1
18.9
23.9
28.0
30.4
30.8
28.7
25.2
20.9
16.8
13.8
11.8
10.8
10.6
11.0
11.5
12.0
12.3
12.5
12.2
11.6
10.7
9.59
8.37
7.40
6.22
5.08
4.13
3.27
2.56
2.04
1.59
1.28
1.06
0.92
0.84
0.74
0.71
0.67
0.65
0.61
0.53
0.49
0.40
0.34
0.28
0.21
0.14
0.09
0.05
0.0
was by fitting to the vibrational structure in the Br.
spectrum referred to above and appreciable absorption
which was not due to Br. was observed at wavelengths
440 nm as shown in figure 2. The residuals obtained
after spectral subtraction using the differential method
were typically ,, :t: lx10 -4, showing no systematic vari-
ation over an approximate order of magnitude range in
Br2 concentration. Two firther bands were observed in
the visible region of the spectrum; the first one peaking
around 520 nm with a cross-section of ,, 4 x 10 -'ø cm '

RATTIGAN ET AL.:UV-VISIBLE ABSORPTION CROSS SECTIONS OF BR.O AND HOBR 23,025
molecule-  and a second broader band from 580 nm ex-
tending beyond 750 nm with a peak cross-section of 6.2
x 10 -20 cm 2 molecule - at 665 nm. These bands can
with reasonable confidence be assigned to Br20 since
similar bands have also been observed in the chlorine
analogue, C120, at 420 and 540 nm, albeit with cross-
sections somewhat lower. The spectrum of Br20 re-
ported here is in good agreement with an earlier re-
ported spectrum obtained during the decomposition of
OBrO (Rattigan et al. [1994]). The Br20 spectrum, un-
like that of C120 exhibits a vibrational band progression
in the wavelength region 345 to 390 nm; the vibrational
spacing being  300 cm -.
Absorption Spectrum of HOBr
HOBr was prepared by the addition of excess H20 va-
por to samples of Br20 (prepared by the above method)
and allowing equilibrium (RS) to be established. A sam-
ple of Br20 (0.20 torr which contained approximately
30% Br2) was first measured into the absorption cell and
a spectrum was recorded over the wavelength range 355
to 430 nm. Upon addition of water vapor ( 9 torr) a
dramatic change to the spectrum of the mixture was ob-
served as seen in figure 4. The amount of Br20 present
in the sample was quantified by fitting to its vibrational
structure in the differential spectrum as shown in figure
5. The remaining spectrum after subtraction of Br20
was smooth and showed a large contribution due to Br2
as well as HOBr (see figure 6). A scaled subtraction of
the Br2 was carried out assuming that HOBr does not
contribute significantly at wavelengths greater than 430
nm, using a reference spectrum for Br2 recorded under
the same experimental conditions. This method had to
be used because the spectral range did not cover regions
where structured absorption occurs both for Br20 (A <
390 nm) and Br2 (A > 515 nm). Spectra were recorded
over several minutes following the addition of water to
the cell. In all cases the residual spectra assigned to
HOBr in the wavelength range 355-430 nm were found
0.25
'- 0.20
 015
= 0.10
o
 0.05
0.00
350
(a)
(b)
I i i i I i , i I i i i
370 390 410 430
Wavelength/rim
Figure 4. Typical absorption spectrum of a 0.2 torr
sample of Br20 containing 30% Br2 at 298 K (line a)
and the absorption spectrum after the addition of 9 torr
of H20 vapor (line b) to the sample in line a.
 0.004
o.0o2
,- 0.000
o
_ -O.OO2
._
 -0.004
*- 355 360 365 370 375 380 385
.,
-o wavelength/nm
Figure 5. Differential fitting of a sample spectrum
containing 0.2 torr Br20 + Br2 with 9 torr of H20
vapor added (solid line) to the Br20 reference spectrum (dashed line) and the i'esiduals after subtraction (thick
dotted line) at 298 K.
to have similar spectral shapes. Furthermore, experi-
ments with a factor of 6 variation in H20 vapor concen-
tration were carried out in the same wavelength region,
and in all cases there was no systematic change to the
shape of the HOBr absorption.
In order to provide complete coverage over the wave-
length range 235 to 430 nm experiments were carried
out at various other spectral regions, working toward
shorter wavelengths and ensuring an overlap between
adjacent segments of at least 15 nm. Wherever possible,
spectral stripping of the Br20 was carried out using its
vibrational structure as discussed above, except in the
region 235 to 310 nm where a scaled subtraction was
employed. Quantitative subtraction of Br2 at shorter
wavelengths was carried out from a knowledge of the
spectral shape of HOBr in the overlap region. In the
0.03 ' ' '  ' ' ' i .... :
0.02
0.01
0.00
350 370 390 410
Wavelength/rim
, , [ I  ,  i , , ,
430
Figure 6. Typical absorption spectrum after the differ-
ential subtraction of the contribution due to Br20 from
the Br20/Br2/H20 spectrum in Figure 4, (line a), a
scaled Br2 reference spectrum (line b) and the residual
spectrum assigned to HOBr (line c) after subtraction of
line b from line a. See text for details.

23,026 RATTIGAN ET AL.:UV-VISIBLE ABSORPTION CROSS SECTIONS OF BR20 AND HOBR
wavelength region 235 to 310 nm, however, the correc-
tion for Br2 was negligible and the subtraction of Br20
was facilitated from a knowledge of the spectral shape
of the HOBr at longer wavelengths.
After submission of this work for publication we learned
of a new study which employed laser photofragment
spectroscopy to determine the relative yields of hy-
droxyl radicals produced by laser photolysis of mixtures
containing Br2, Br20, H20 and HOBr in the region
440 to 650 nm (Barnes et al. [1996]). This work indi-
cates the presence of a weak absorption band for HOBr
centered at  440 nm. It is suggested that the band
arises from excitation to a triplet state of HOBr. This
result implies that the above assumption of zero con-
tribution of HOBr  h .... ; spectrum at 430
nm is incorrect. In order to investigate the presence of
this weak band, we have analyzed spectra of the Br20-
H20-HOBr mixtures taken in our study with the low
resolution grating (2.5 nm FWHM) covering the wave-
length region 263 to 569 nm. Twenty wideband spectra
were analyzed from two mixtures containing initially
4.6x10 s and 1.8x10 s molecule cm -3 Br20 with 2.88
torr and 2.45 torr H20 respectively. After subtraction
of the Br20 absorption using differential fitting to the
vibrational structure from 350 to 380 mn as described
previously, the Br2 was subtracted off the remaining
spectrum by scaling it to the average absorbance at 435
q- 5 nm and at 505 q- 5 nm. The averaged spectra
obtained in this way are plotted in figure 10. When
zero HOBr contribution at 435 nm was assumed, the
spectrum was identical to that obtained in the narrower
wavelength region and higher resolution grating (0.6 nm
FWHM) and showed a smooth cutoff in absorption in
the tail region. However, if zero HOBr absorbance was
assumed at 505 nm, a shoulder appeared in the tail,
which would be consistent with another weak band cen-
tered around 440 nm. Despite having used a Pyrex fil-
ter data obtained at wavelengths > 510 nm could not
be used because of complications due to second order
effects. Time-resolved experiments showed that the ab-
sorbance in the shoulder region (430 to 500 nm) varied
linearly with that in the band centered around 350 nm,
consistent with it being HOBr. The value of the cross-
section at 350 nm (cr=ll.5x10 -2ø cm 2 molecule -) and
the equilibrium constant Ks (.=0.036) obtained from
these spectra were close to the average values from all
experiments (see below). The absorption at  < 400 nm
was not significantly changed from the previous high-
resolution data.
Absorption Cross section for HOBr
In order to estimate the absorption cross-section for
HOBr it is necessary to accurately determine its con-
centration. This was obtained fi'om estimation of the
Cllange m [Br20] upon the addition of excess H20 va-
por, which results in the equilibrium (R8) being estab-
lished (see below):
(RS) Br20 + H20  2HOBr
4.0
E 30
'
- 20
'
1.0
0, 0
(a)
[] Lights on  [][] []
o%
o
oo o
H20 addition
* O
O
1000 2000 3000 4000
time/sec
8.0 ,
6.0
4.0
2.0
0.0
0
(b)
..... , ..... ,, ,, addition Lights on
ß [] [] ß ß [] ß [] [] []
[]
ø    ; o o
e o
300 600 900 1200 1500
time/sec
Figure 7. Concentration (10 s molecule cm -3 ) time
profile of Br20 (solid circles), Br2 (diamonds), HOBr
(open circles) and total Br2 (squares) at 298 K with the
addition of (a) 4.3 torr of H20 and (b) 9.0 torr H20 to
the Br20/Br2 sample. The [HOBr] was estimated from
the A[Br20] upon addition of the H20 vapor to the
Br20/Br2 sample. See text for details.
The amount of HOBr in the equilibrium nfixture was
assumed to be given by the stoichiometry:
A[HOBr]- 2A[Br20] (1)
Figure 7(a) shows the concentration-time profile of
the various bromine species in the cell when 4.3 torr of
H20 was added to a 0.12 torr sample of Br20 contain-
ing initially  10% Br2. Prior to the addition of water,
Br20 decayed, with production of Br2 with a first or-
der decay constant of (6.6 4- 1.1)x10 -4 s -. This is
believed to be due to a heterogeneous reaction since
the rate was independent of total pressure and tended
to decrease in successive xperiments as the surface be-
canhe conditioned. The concentration of Br20 at the
point of H20 addition was estimated by extrapolation
of this decay.
Upon addition of H20 there was an abrupt drop
in the [Br20] which then underwent a slow decompo-
sition with concurrent production of Br2. The slow

RATTIGAN ET AL.:UV-VISIBLE ABSORPTION CROSS SECTIONS OF BR20 AND HOBR 23,027
loss was first order in [Br.O] as shown by the semi-
log plot in figure 8(a), with a decay rate of (4.91 q-
0.01)x10 -4 s -. The [Br20] immediately after H20 ad-
dition was obtained by a short back extrapolation, and
hence A[Br.O] could be determined.
After the addition of HO the absorption due to
HOBr appeared rapidly and then showed a slow first
order decay. The slope of the semilog plot (figure 8(b))
was (2.2 q- 0.2)x10 -4 s -, i.e. a factor of 2 less than
the slope for Br.O decay, within the experimental er-
ror. This feature was observed in all experiments and
provides a strong indication that the equilibrium (R8)
was maintained throughout the decay. Back extrapola-
tion of the HOBr absorption to the point of H.O ad-
dition gave the absorption corresponding to [HOBr] -
2A[Br20] (see (1)). This was used to calculate O'HOBr ,
which was then used to compute the values of [HOBr]
and total [Br2] plotted for the remainder of the experi-
ment.
Prior to H20 addition, [Br20] declined while Br2 in-
creased. However the Br2 increase was less than the
Br20 loss, as seen from the total Br2 curve in figure
0.6
0.3
00'
'
-0.3
-0.6
-0.9
-1.2
0
' ' ' I ' ' ' I ' ' ' I ' ' ' I ' ' '
(a) Br 20
, , , I , , , I , , , I , , , I
700 1400 2100 2800
time/sec
i
3500
-4.2
43
44
-4.5
-46 ß
-4 7 ß
-4.8
-4.9
-5.0
0
, , I , , , I , , , I , , , I ,  ,
700 1400 2100 2800 3500
time/sec
Figure S. Plots of (a) log [Br20] versus time and (b)
log HOBr absorbance versus time following the addi-
tion of 4.3 torr of H20 vapor to the Br20/Br2 sample
at 298 K. The lines show the least squares fit to the
data.
7(a). Upon addition of H20 the Br2 balance was re-
stored and a distinct rapid rise in [Br2] is evident be-
fore the increase due to the slow decomposition of Br20
and/or HOBr. This behavior was much less apparent
at 308 K and is believed to result from the physical
adsorption of Br2 at the vessel walls, with displace-
ment of the adsorbed Br2 by more strongly absorbed
water molecules. In the subsequent period when Br2 in-
creased, there was a slow loss of total Br2, which could
result from additional surface loss of Br2.
Important differences were observed in experiments
with higher added [H20] (P > 8 torr), an example of
which is shown in figure 7(b). After the addition of
H20 there was a distinct delay period before the [HOBr]
maximized, while the changes in Br2 and Br20 were
much less affected. The time period increased with the
amount of added H20. At first this was thought to be a
mixing effect following the addition of a relatively large
partial pressure of H20 to small amounts of Br20 in
the cell. However, when a similar pressure of N2 was
added to the Br20/Br2 mixtures, small perturbations
due to mixing could be seen, but they restored on a
shorter timescale (<100 s), more in line with diffusive
mixing times for these conditions. The slow growth of
HOBr is indicative of a surface controlled kinetic effect.
probably associated with multilayer adsorption of wa-
ter, which is likely to occur on a glass surface at relative
humidities greater than  40%. During this period the
equilibrium between the gaseous Br20, H20 and HOBr
was not established and the data could not be used to
determine  or Ks. Beyond this period the HOBr max-
imized and then decayed with a first order rate of (6.95
q- 0.35)x10 -4 s -1 i.e. a factor of two lower than that
of Br20, (1.29 q- 0.16)x10 -a s -) as also observed at
lower [H20] in figure 7(a). The 'true' initial HOBr ab-
sorption was again determined by back extrapolation
to the point of H20 addition and rrHOBr determined ac-
cording to (1). There was no systematic variation in
HOBr determined in this way over a range of [H20] as
is seen from the data smmnarized in Table 2.
The second effect of increased [H20] was to maintain
a better mass balance in total Br2 during the decay of
Br20 and HOBr. For example, in figure 7(b) follow-
ing the mixing period the total bromine is reasonably
constant as a function of time. Furthermore, the to-
tal amount of Br2 recovered at the end after complete
conversion of the HOBr and Br20 into Br2 via pho-
tolysis, agrees with the initial total concentrations of
Br20 and Br2, indicating that the estimation of both
Br20 and HOBr are consistent. The better mass bal-
ance can be rationalised in terms of the influence of the
more strongly adsorbing water molecules, lowering the
extent of surface adsorption of Br2.
Once the HOBr concentration is known, the absorp-
tion cross-section can be calculated from the absorbance
measurements using the Beer Lambert law. Table 2
shows the values of HOm. at 360 nm for a variety of
initial conditions at different temperatures in the range

23,028 RATTIGAN ET AL.:UV-VISIBLE ABSORPTION CROSS SECTIONS OF BR20 AND HOBR
Table 2. Absorption cross-section for HOBr and equilibrium constant Ks at various temperatures and [H20]
Temp/K [HOI a**o , Ks 
10 -7 molecule cm - 10 2ø crn 2 molecule -
308 3.76 11.0 4- 1.3 0.047 4- 0.003
2.95 9.56 4- 1.2 0.046 4- 0.002
1.97 12.3 4- 0.6 0.046 4- 0.002
1.70 12.8 4- 0.5 0.0414- 0.002
1.45 12.5 4- 0.4 0.043 4- 0.004
average 11.6 4- 1.2 0.045 4- 0.002
298 4.21 9.84 4- 0.7 0.038 4- 0.003
3.97 10.9 4- 0.4 0.038 4- 0.004
3.30 9.26 4- 1.0 0.038 4- 0.003
3.27 10.9 4- 3.2 0.036 4- 0.004
2.90 11.2 4- 0.6 0.037 4- 0.003
2.82 10.4 4- 2.3 0.039 4- 0.003
1.67 11.7 4- 0.7 0.037 4- 0.002
1.58 11.7 4- 0.3 0.038 4- 0.002
1.38 11.4 4- 0.6 0.044 4- 0.006
0.82 11.2 4- 1.2 0.026 4- 0.003
0.66 13.6 4- 1.8 0.039 4- 0.007
0.55 12.2 4- 1.4 0.017 4- 0.003 b
average 11.2 4- 1.1 0.037 4- 0.004
288 3.28 11.6 4- 1.1 0.033 4- 0.005
2.98 12.3 4- 4.5 0.029 4- 0.005
2.87 13.9 4- 1.6 0.031 4- 0.002
average 12.6 4- 1.0 0.031 4- 0.002
278 1.736 12.8 4- 1.8 0.028 4- 0.003
1.371 12.9 4- 0.7 0.028 4- 0.003
1.041 10.5 4- 1.9 0.023 4- 0.003
average 12.1 4- 1.1 0.026 4- 0.002
aKs =[HOBr] /[Br O][H O].
bnot; included in the average of Ks
278 to 308 K. There was no significant temperature de-
pendence in the cross-section and the mean value of
era00nm was (11.6-t- 1.3)x10 -'ø cm ' molecule - (error
represents tatistical scatter at the la level). As a check
on the values of the cross-section determined on the ba-
sis of A[Br.O], the amount of HOBr present immedi-
ately after the addition of H. O was calculated assuming
a mass balance of total bromine, i.e.
[HOBrlt - 2x([Br20]i - [Br20]t) - ([Br2]t - [Br2]i)
(2)
where i refers to the initial concentrations and t refers
to the concentrations at some time after H20 addition.
Because of the delay in the rise in the HOBr absorption
(at higher H20) and the loss of total bromine at longer
reaction times, it was necessary to extrapolate the val-
ues of rr calculated from [HOBr]t to the point of water
addition. The mean value of rr300nrn calculated in this
way was (9.90 -t- 1.62)x10 -2ø cm 2 molecule - at 298 K,
i.e. 15% lower than the mean value of cra00nm based on
A[Br20], but within the error limits. The lower value
probably arises from an under-estimation of [Br2]t due
to adsorption of Br. at the walls, which will result in
an overestimation of [HOBr] calculated by equation (2)
and hence lower values for aHOBr. The higher value
given above is therefore preferred.
The absorption cross-sections for HOBr calculated in
this study are compared with the previous values of
Orlando and Burkholder [1995] in figure 9. Tabulated
values averaged over 5 nm intervals are shown in Ta-
ble 1. The uncertainty in the estimation of er based on
the spread of the values is -t- 13% at wavelengths <400
nm, increasing rapidly at longer wavelengths due to the
assumption made regarding the point where HOBr ab-
sorbance becomes negligible compared to the Br. im-
purity. At wavelengths < 300 nm the cross-sections
determined here are in good agreement with the previ-
ously determined values. The absorption cross-section
at the maximum near 282 nm a=(3.1 -t- 0.4) x I0 -ø
cm 2 molecule - is in excellent agreement with that ob-
tained by Orlando and Burkholder. However, at longer
wavelengths which is the most important region for pho-

RATTIGAN ET AL.:UV-VISIBLE ABSORPTION CROSS SECTIONS OF BR20 AND HOBR 23,029
270 310 350 390 430
wavelength/nm
Figure 9. Absorption cross-sections of HOBr at 298
K with a wavelength cutoff at 430 nm determined in
this work (solid line) and selected data from Orlando
and Burkholder [1995] (filled triangles). The difference
between this work and that of Orlando and Burkholder
[1995] is shown by the dashed line. The open triangles
show the scaled Br.O cross-sections determined in this
work.
tolysis in the atmosphere, our values are up to a factor
of  2.5 higher and significant absorption due to HOBr
is measured out to 430 nm. The shape of the HOBr
spectrum is very similar to that of the chlorine analogue
HOC1, (Burkholder [1993]), HOBr being red-shifted by
 45 nm. The intensity of the HOBr absorption is how-
ever higher than that of HOC1 in both bands.
The reason for the discrepancy between our spectrum
and that of Orlando and Burkholder [1995] at longer
wavelengths is immediately apparent when the differ-
ences in the cross-section as a function of wavelength
are examined, see figure 9. The difference matches well
the spectrum of Br.O, indicating that the discrepancy
arises from systematic errors in subtraction of Br.O in
one of the two studies. As seen from figure 5, there is no
structure evident in the residual from the least squares
fitting routine: a measure of the accuracy in the fitted
Br20 can be determined from the uncertainty in the
gradient of a plot of the fitted and the scaled reference
spectrum. This was found to be typically better than q-
2% (2rr) in the present study. However, because the ab-
sorption due to Br2 O was an order of magnitude greater
than the HOBr absorption at 360 nm, the HOBr cross-
sections are quite sensitive to small errors in the amount
of BrO subtracted. The q- 2% uncertainty in the BrO
subtraction translates to , q- 10% error in the HOBr
cross-section in the present experiments. Thus the dis-
crepancy in the cross-sections is well outside the error
in the present study.
Orlando and Burkholder [1995] also applied differen-
tial fitting routines to subtract BrO absorption from
HOBr, and subtraction of the Br was carried out by
a similar method to that used here, i.e. by assuming
that only Br2 absorbed at wavelengths  440 nm. In
the experiments of Orlando and Burkholder however the
amount of Br2 was typically  10 x6 molecule cm -3, i.e.
approximately an order of magnitude higher than in the
present work and the absorption due to Br2 dominated
that of HOBr at wavelengths  350 nm. This could
have lead to uncertainties in the spectral subtraction
of Br2 O at longer wavelengths where its structured ab-
sorption lies. Thus small errors in the amount of Br20
would have influenced the derived spectrum and hence
the HOBr cross-sections.
During the course of this work we have learned of
two further investigations of the absorption spectrum
of HOBr. Benter et al. [1995] reported cross-section
measurements in the region 234- 400 nm for HOBr,
produced by bubbling helium through freshly prepared
aqueous 0.2 M HOBr solutions at 293 K. Much lower
[Br.O]/[HOBr] ratios ( 0.02 at 293 K) were achieved
by this method. HOBr concentrations were determined
either by iodometric titration of trapped samples or by
gas phase titration of HOBr with C1 atoms in a fast
flow system. Spectra were recorded with a photodi-
ode array. The cross-sections reported at the two band
maxima were (2.4 q- 0.1) x 10 -9 cm 2 molecule - at
284 mn and 7.7x10 -2ø cm 2 molecule - at 350 nm. Fig-
ure 10 shows a comparison of this and the other HOBr
spectra normalized to a value of (3.1 q- 0.4) x l0 -9
cm 2 molecule - at 282 nm. The shape of the Benter
et al. [1995] spectrum lies closer to the present work
than to the Orlando and Burkholder [1995] spectrum,
but nevertheless the cross-sections are lower in the re-
gion of strong Br20 absorption. Unfortunately, Benter
et al. [1995] did not extend their measurements beyond
400 nm and they did not discuss their methodology of
correction for Br20 absorption.
In a second study Deters et al. [1996] have presented
results from similar experiments to our own, utilising
2 1 0 '9
o
E
E 0.20 --*-- Or,. & Buk. I o 1
c - E - Benter et al.
'' --'-- Deters et al. I ,  '\"\"
ß O' 21 This work(a) \"I This work(b) I
 I
 .
O .... I .... I,,, , I .... I ....
 250 300 350 400 450 500
wavelength/nm
Yi[ure Z0, Absopo oss-seos of HOBr
lille), Orlando and Burkholder (solid squares and dot-
ted line), Benter et al. [1995 (open squares and dashed
line) and Deters et al. [1995 (open triangles and dash-
dotted line). Data of Benter et al. and Deters et al.
[1996] are normalized to a value of 3.1 x 10 -xo cm 2
molecule-  at 282 nm.

23,030 RATTIGAN ET AL.:UV-VISIBLE ABSORPTION CROSS SECTIONS OF BR20 AND HOBR
the reaction of Br20 with H20 to make HOBr. They
report cross-sections at the two band maxima of (2.7
4. 0.4)x10 -ø cm 2 molecule - at 280 nm and (7.0 4.
1.1)x10 -2ø cm 2 molecule - at 355 nm. The spectrum
normalized at the maximum near 282 nm is also shown
in figure 10. It is closer in shape to the Orlando and
Burkholder [1995] spectrum. Clearly, the differences lie
in the subtraction of Br20 absorption, but details of
the methodology for this were not given.
Neither of these studies provides information on the
weak absorption band for HOBr centered about 440 nm,
detected using photofragment spectroscopy by Barnes
et al. [1996]. The cross-section at the maximum was es-
timated by comparing the relative yield of OH, assum-
ing mit quantum yield for dissociation via the process
in (R4)'
(R4) HOBr+hv ) OH+Br
and utilising the value of a(355 nm) from Orlando
and Burkholder [1995]. The value obtained was 8.8x10 -2
cm 2 molecule-  at 440 nm. If scaled to the present value
of 12.5x10 -2ø cm 2 molecule - at 350 nm the value is
18.0x10 -2 cm 2 molecule - at 440 nm which is sub-
stintlilly higher than the cross section obtained from
analysis of our wideband spectra, i.e.  7.5x10 -2 cm 2
molecule -. Clearly there remains considerable uncer-
tainty about this long wavelength absorption band of
HOBr.
Equilibrium Constant Ks--[HOBr] 2/[Br20] [H20]
The equilibrimn constant for (R8) was determined
from the [Br20] and [HOBr] present following the ad-
dition of a known excess [H20]. Values for the equilib-
rium constant Ks at 298 K as a function of time with
an initial I-t20 vapor concentration of2.9x10 7 molecule
cm -3 are shown in Table 3. Once mixing is achieved,
the values of Ks are reasonably constant and yield an
average value of 0.037 4- 0.003. This value is a factor
of  2 larger than previously determined by Orlando
and Burkholder [1995]. The value for Ks remained es-
sentially constant over more than a factor of 6 change
in the initial [H20] added as shown in Table 2. Fur-
thermore, experiments in different spectral regions did
not show any systematic variation in Ks. The average
of all the data at 298 K yield a value for Ks - 0.037
4- 0.004, which is not too dissimilar from the value of
(0.1 4- 0.01) for the corresponding equilibrium involving
C120 and H20 [Burkholder, 1993].
Experiments have also been conducted at tempera-
tures in the range 278 to 308 K in order to determine
the temperature dependence of the equilibrium constant
and hence to obtain thermodynamic parameters (AH
and AS) for (RS). For these experiments the spectral
region  290 nm was selected so that accurate subtrac-
tion of Br20 could be carried out using differential fit-
ting to its vibrational fine structure as discussed above.
Any temperature dependence in the absorption cross-
section of HOBr was assumed to be indistinguishable
from the error uncertainties in a of 4- 13% in the lim-
ited temperature range used, and was not considered
in these calculations. The values obtained for Ks are
shown in Table 2, indicating an approximate factor of
2 change over the experimental temperature range 278
to 308 K. A Van't Hoff plot of In Ks versus 1000/T
as shown in figure 11 yields a value for AHs - (13.0
4- 0.5) kJ mol-  and ASs of about (16 4- 2) J mol- 
K-. This low value for AS is expected for the reaction
between the triatomic species Br20 and H20 forming
two triatomic product molecules. The value agrees with
the calculated value of AS ø - 18 J tool - K - obtained
using Bensoh's [1976] bond additivity method to esti-
mate Sø(Br20)29s K= 289 J mol - K - and a value for
Sø(HOBr)29s g -- 248 J mol - K - from McGrath and
Rowland [1994]. Taking an average of the reported val-
Table 3. Equilibrium constant Ks versus time
Time/sec Ks\" Total [Br2]
x 10  molecule cm-
30 0 6.254
110 0.018 b 6.084
164 0.027 b 5.913
224 0.039 6.280
314 0.036 6.243
434 0.035 6.164
554 0.041 6.234
743 0.032 6.019
914 0.033 5.998
1094 0.035 5.997
1274 0.035 5.970
1394 0.043 6.186
average 0.037 4- 0.003
Temperature is298 K; [H20] = 2.9x10 ? molecule cm -s.
'K,--[HOBr] /[BrO][H O1.
bnot included in the average of Ks, see text for details

RATTIGAN ET AL.:UV-VISIBLE ABSORPTION CROSS SECTIONS OF BR20 AND HOBR 23,031
0.10
0.01
3.2
I I I  I ,   , I   , ] I    [
3.3 3.4 3.5 3.6 3.7
1000/T(K)
Figure 11. A Van't Hoff plot of 1Oge Ks versus 1000/T
(K) over the temperature range 273 to 308 K. The line
shows the least squares fit to the data. Error bars rep-
resent statistical scatter to the data and are expressed
as  lcr.
ues of AHø(HOBr)29s K= -58.5 kJ tool - [McGrath and
Rowland, 1994; Ruscic and Berkowitz, 1994; Gl'ukhovt-
sev et al., 1996; Lock et al., 1996; Thorn et al. 1996] and
using the above value for AHs, a value for AHø(Br20)
of 112 kJ mol - is obtained compared to the recently
reported value of 107 kJ mol - by Thorn et al. [1996].
Atmospheric Implications
The absorption cross-sections for HOBr determined
in this study have been used in a photochemical box
model to calculate the photolysis rate at 70 mbar as a
function of solar zenith angle (SZA). For this calcula-
tion a midlatitude ozone and temperature profile were
used and a photodissociation quantum yield of unity for
HOBr was assumed. The calculated photolysis rates
(jHonr) versus SZA are shown in figure 12. As can
be seen, the calculated photolysis rate for HOBr is in-
creased by a factor of  2.3 or 2.8, compared with the
data of Orlando and Burkholder [1995] depending on
whether the long wavelength band is included. In the
sunlit atmosphere a value of  2.5-3x10 -a s - is ob-
tained, which corresponds to a lifetime for HOBr of 
6 minutes indicating that its photodissociation lifetime
is comparable to that of BrONO. [Lary et al., 1996].
The model used accounts for spherical geometry of the
atmospheric and all orders of multiple scattering as de-
scribed by Lary and Pyle [1991a,b]. The effect of scat-
tering from clouds however is not considered in these
calculations. The model is based on work of Meier et
al. [1982], Nicolet et al. [1982], and Anderson [1983].
Recent measurements during the Stratospheric Pho-
tochemistry, Aerosols and Dynamics Expedition (SPADE)
campaign [$alawitch et al., 1994] indicated a pulse of
HO and HO2 after sunrise at 37.4N, consistent with
the photolysis of a nighttime reservoir of HOx. An ar-
gument was proposed by $alawitch et al. [1994], which
0
10\" -
10-3--_
_
_
'7 -
03 -
ß -- 0-4
0 -
 -
_
..1
10--_
_
_
_
_
10\"
o
10 20 30 40 50 60 70 80 90
Orlando & Burkholder
............. Rattioan et al. (])
Rattioan et al. (2)
,
,'o ;o ;o ,o ;o
Solar Zenith Angle (degrees)
lOO
lOO
10 '2
10 '3
10-4
10 '5
10 '6
igUre 12. The calculated HOBrphotolysis coefficient -) at 70 mbar as function of sola zenith angl (de-
grees) using the cross-section data of this study, with a
wavelength cut offat 430 nm (dashed line), and a wave-
length cutoff at 510 nm (dash-dotted line) and those
reported by Orlando and Burkholder [19951 (solid line).
A midlatitude ozone and temperature profile was used,
and a photodecomposition quantum efficiency of unity
was assumed.
involved the heterogeneous conversion of HO=NO= into
HONO on aerosol particles overnight. The abrupt in-
crease in HOx at dawn was ascribed to the subsequent
photolysis of HONO into OH and NO. However recent
measurements by Zhang et al. (Heterogeneous chem-
istry of HO=NO2 on liquid sulfuric acid, submitted to
J. Phys. Chem., 1995) indicate that the heterogeneous
conversion of HO=NO= into HONO is too slow to ac-
count for the observed production in HO and HO=.
The effect of the revised HOBr cross-sections cou-
pled with heterogeneous bromine chemistry on the cal-
culated levels of HO has also been investigated. Dur-
ing the night the heterogeneous hydrolysis of BrONO2
on sulphate aerosols produces HOBr [Hanson and Rav-
ishankara, 1995] and photolysis of the nighttine accu-
mulated HOBr at sunrise could lead to a rapid rise in
OH. As a result little BrONO= remains if even moder-
ate aerosol loadings are present. Lary et al. [1996] have
indeed shown that this nighttime production of HOBr
leads to a sudden increase in the OH and HO= con-
centrations at dawn, which is similar to those recently
reported by $alawitch et al. [1994], although the mag-
nitude of this increase was somewhat underestimated
in their work. If the HOBr cross-sections reported in
this study, however, are used instead of those of Or-
lando and Burkholder [1995], the rate of HOBr photol-
ysis is increased by a factor of  2.5. Figures 13(a) and
(b) show that this does not influence greatly the day-
time HO radical concentrations; it does give a stronger
point of inflection in OH and HO= at sunrise, which is
in better agreement with the SPADE observations than
the simulations presented by Lary et al. [1996]. In these
calculations HOBr cross sections with a cutoff at 430 nm

23,032 RATTIGAN ET AL.:UV-VISIBLE ABSORPTION CROSS SECTIONS OF BR_O AND HOBR
(b)
Data - No Het. Br ..... i-let Br (dO & JB HOBr)
37.9 ø N, 214.5 K, 66.9 rnb, 5 May, Aerosol area - 6/,m  cm 4.
..... Her. Br (OR HOBr)
-
:0.75
-0.5
-
_
_
.
-O.25
_
_
0.0
Solar Zenith Angle (degrees)
Data No Het. Br ....... Her Br (JO & JB HOBr) ..... Her, Br (OR HOBr)
37.9 ø N, 214.5 K, 68.9 rnb, 5 May, Aerosol area = 6/rn' crn 4.
6
Solar Zenith Angle (degrees)
Figure 13. The effect of HOBr photolysis and hetero-
geneous bromine reactions on sulphate aerosols on the
shape of (a) OH and (b) HO2 diurnal cycles. The initial
conditions were taken from Table 1 of Salawitch et al.
[1994] and an assumed total BrOy of 11.5 pptv at 66.9
mbar. The solid lines uses the cross-sections of Orlando
and Burkholder [1995] and does not include heteroge-
neous bromine chemistry. The dot-dashed line uses the
HOBr cross-sections of Orlando and Burkholder and in-
cludes the heterogeneous bromine reactions as described
by Lary et al. [1996]. The dashed line uses the HOBr
cross-sections determined in this work (wavelength cut-
off at 430 nm) and includes heterogeneous bromine re-
actions as described by Lary et al. [1996].
were used (see figure 10). Allowing for the weak band
around 440 nm from this work gives a 10-15% earlier
risetime in HO and HO2; the agreement with the 'pulse'
in the SPADE data is not greatly affected. If heteroge-
neous bromine chenistry is not included in the model
runs, the model underpredicts the observed HOz and
there is no evidence for an early morning pulse in either
OH or HO2. The simulations hown in figures 13(a) and
(b) used the initial conditions given by Salawitch et al.
[1994, Table 1] and assumed a total BrOy of 11.5 parts
per trillion by volume (pptv) at 66.9 mbar. The model
used was AUTOCHEM, described in detail by Lary et al.
[1996].
The increased photolysis rate of HOBr coupled with
the hydrolysis of BrONO2 on the surface of sulphate
aerosols causes a 10-20% enhancement in the daytime
BrO concentrations calculated by the model. In ad-
dition, the enhanced OH concentration causes a slight
decrease in the HC1 lifetime and hence the HC1/C1Oy
ratio in the lower stratosphere. This, in turn, causes
additional C1Oz and BrOz activation, which enhances
the effectiveness ofthe gas phase C10/BrO cycles. The
increase in OH and HO2 also enhances the HO2/C10
and HO2/BrO catalytic cycles. The catalytic hydroly-
sis of BrONO leads to a direct conversion of NOz (NO
+ NO2) to HNO3. For sunlit conditions at midlatitudes
this direct conversion to HNOs is of comparable mag-
nitude to that caused by the hydrolysis of N2Os. How-
ever, there is also an indirect enhancement in HNO3
that occurs owing to the increase in the rate of reaction
of OH with NO2. These mechanisms, all of which act
together, lead to enhanced loss of 03 at all latitudes in
the lower stratosphere.
This is of interest since a recent World Meteorologi-
cal Organization (WMO) assessment [WMO, 1992] re-
ported that for the first time there were statistically
significant decreases in ozone in all seasons in both the
northern and southern Hemispheres at midlatitudes and
high latitudes during the 1980s, and that most of this
decrease is occurring in the lower stratosphere. rends
derived from ozonesondes by Logan, [1994] support of
these findings. Solomon et al. [1996] showed that
when a two-dimensional model is constrained with time
varying aerosol observations, the shape of the observed
trends in ozone are reproduced but their magnitude is
about 50% larger than that which is observed. This pa-
per shows that at least part of this ozone loss is likely
to be due to insitu heterogeneous bromine reactions.
As the hydrolysis of BrONO2 is not very temperature
dependent, it can occur at all latitudes. For a more
detailed analysis, see Lary et al. [1996].
Conclusions
The absorption cross-sections for HOBr appear to be
larger at wavelengths 310 to 430 nm than those pre-
viously determined by Orlando and Burkholder [1995].
The calculated atmospheric photolysis rate is, corre-
spondingly, a factor of 2.5 faster. Inclusion of het-
erogeneous hydrolysis of BrONO2 into HOBr on sul-
phate aerosols together with the faster photolysis rate
of HOBr brings the model simulations of the details in
the diurnal variation of HO and HO2 into good agree-
ment with he SPADE observations. In addition, the
catalytic hydrolysis of BrONO2 leads to enhanced ozone
loss at all latitudes in the lower stratosphere. This loss
is mainly due to the elevated levels of HO and BrO
and the reduction in NOz which enhances the C10 con-

RATTIGAN ET AL.:UV-VISIBLE ABSORPTION CROSS SECTIONS OF BR20 AND HOBR 23,033
centration, thereby enhancing Oa loss by the catalytic
cycles C10/BrO, HO./C10, and HO./BrO in the lower
stratosphere.
Acknowledgments. We thank D.J. Fish for providing
the software for the spectral fitting routines and R.J. Salaw-
itch for providing a copy of the SPADE data. Thanks to
Amitabha Sinha and Bernd Deters for providing copies of
their manuscripts prior to publication. O.V.R. wishes to
thank the United Kingdom Research Council (NERC) for
financial support.
References
Abbatt, J.P.D., Heterogeneous reaction of HOBr with HBr
and HC1 on ice surfaces at 228 K, Geophys. Res. Lett., 21
(R8), 665-668, 1994.
Anderson, D.E., The troposphere to stratosphere radiation
filed at twilight: A spherical model, Planet. Space Sci., 31
(12), 1517-1523, 1983.
Barnes, R.J., M., Lock, J., Coleman, and A., Sinha, Obser-
vation of a New Absorption Band of HOBr and its Atmo-
spheric Implications, J. Phys. Chem., 100 (2), 453-457,
1996.
Benson, S.W., Thermochemical kinetics, 2nd ed., Wiley.
New York, 1976.
Benter, Th., C., Feldmann, U., Kirchner, M., Schmidt,
Schmidt, and R.N., Schindler, UV/VIS-absorption spec-
tra of HOBr and CH3OBr; Br(2P3/2) atom yields in the
photolysis of HOBr, Bet. Bunsenges. Phys. ahem., 99,
1,144-1,147. 1995.
Bridier, I., Veyret, B., and R., Lesclaux, Flash photolysis
kinetic study of reactions of the BrO radical with BrO
and HO2. Chem. Phys. Lett., 201, 563-568, 1993.
Burkholder, J.B., Ultraviolet absorption spectrum of HOC1,
J. Geophys. Res., 98, 2,693-2,974, 1993.
Deters, B., S., Himmelmann, C. Blindauer, and J.P., Bur-
rows, Gas phase spectra of HOBr and Br20 and their
atmospheric significance, Ann. Geophys., 1 (4), 468-475,
1996.
Garcia, R., and S., Solomon, A new numerical-model of the
middle atmosphere: 2-Ozone and related species, J. Geo-
phys. Res., 99 (D6), 12,937-12,951, 1994.
Glukhovtsev, M.N., A., Pross, and L., Radom, Acidities,
proton affinities, and other thermochemical properties
of hypohalous acids HOX (X = F-I): High-Level Com-
putional Study, J. Phys. Chem., 100 (9), 3,498-3,503,
1996.
Hanson, D.R., and A.R., Ravishankara, Heterogeneous
chemistry of bromine species in sulfuric acid under strato-
spheric conditions, Geophys. Res. Lett., 22, 385-388, 1995.
Lary, D.J., and J.A., Pyle, Diffuse radiation, twilight and
photochemistry- I, J. Atmos. Chem., 13, 373-392, 1991a,
13, 393-406, 1991b.
Lary, D.J., M.P., Chipperfield, and R., Toumi, The potential
impact of the reaction OH + C10 > HC1 + 02 on polar
ozone photochemistry, J. Atmos. Chem.. 21, 61-79, 1995.
Lary, D.J., M.P., Chipperfield, R.. Toumi, and T.M.,
Lenton, Atmospheric heterogeneous bromine chemistry,
J. Geophys. Res., 101 (D1), 1,489-1,504, 1996.
Logan, J.A., Trends in the vertical distribution of ozone: An
analysis of ozonesonde data, J. Geophys. Res., 99 (D12),
22,553-25,585, 1994.
Lock, M., R.J., Barnes, and A., Sinha. Near-threshold pho-
todissociation dynamics of HOBr: Determination of prod-
uct state distribution, vector correlation, and heat of for-
marion, J. Phys. Chem., 100, 7,972-7,980, 1996.
McGrath, M.P., and F.S., Rowland, Ideal Gas thermody-
namic properties of HOBr, J. Phys. Chem., 98, 4,773-
4,775, 1994.
Meier, R.R., D.E., Anderson, and M., Nicolet, The radiation
field in the troposphere and stratosphere from 240 nm to
1000 nm: General Analysis, Planet. Space Sci., 30 (9),
923-933, 1982.
Nesbitt, F.L., P.S., Monks, W.A., Payne, L.J., Stirf, and
R., Toumi, The reaction of O(aP) + HOBr: Temperature
dependence of the rate constant and importance of the
reaction as an HOBr loss process, Geophys. Res. Lett.,
œœ, 827-830, 1995.
Nicolet, M., R.R., Meier, and D.E., Anderson, The radiation
field in the troposphere and stratosphere from 240 nm to
1000 nm: Numerical Analysis, Planet. Space Sci., 30 (9),
935-983, 1982.
Orlando, J.J., and J.B., Burkholder, Gas Phase UV/visible
absorption spectra of HOBr and Br20, J. Phys. Chem.,
99, 1,143-1,150, 1995.
Poulet, G., M., Pirre, F., Maguin, R., Ramaroson, and G.,
Le Bras, Role of the BrO + HO2 reaction in the strato-
spheric chemistry of bromine, Geophys. Res. Lett., 19,
2,305-2,308, 1992.
Rattigan, O.V., O., Wild, R.L., Jones, and R.A.,
Cox, Temperature dependent absorption cross-sections
of CFaCOC1, CF3COF, CHCOF, CCICHO and
CFaCOOH, J. Photochem. Photobiol. A:Chem, 73, 1-9,
1993.
Rattigan, O.V., R.L., Jones, and R.A., Cox, The visible
spectrum of gaseous OBrO, Chem. Phys. Lett., 230, 121-
126, 1994.
Ruscic, B., and J., Berkowitz, Experimental determination
of AH'(HOBr) and ionization potentials (HOBr): Im-
plications for corresponding properties of HOI, J. Chem.
Phys., 101 (9), 7795-7803, 1994.
Salawitch, R.J., et al., The diurnal-variation of hydrogen,
nitrogen and chlorine radicals - Implications for the het-
erogeneous production of HNO2, Geophys. Res. Left., 21
(23), 2,551-2,554, 1994.
Seery, D.J., and D., Britton, The continuous absorption
spectra of chlorine, bromine, bromine chloride, iodine
chloride and iodine bromide, J. Phys. Chem., 68, 2,263-
2,266, 1964.
Solomon, S., R. W. Portmann, R. R. Garcia, L. W. Thoma-
son, L. R. Poole and M.P. McCormick, The role of aerosol
variations in anthropogenic ozone depletion at northern
midlatitudes, J. Geophys. Res., 101 (D3), 6,713-6,727,
1996.
Thorn, R.P., Jr., P.S., Monks, L.J., Steif, S.-C., Kuo,
Z., Zhang, and R.B., Kleinre, Photoionization Efficiency
Spectrum, Ionization Energy, and Heat of Formation of
Br20, J. Phys. Chem., 100, 12,199-12,203, 1996.
World Meteorological Organisation, Rep. 25, Scientific as-
sessment of stratospheric ozone: 1991, WMO Global
Ozone Research and Monitoring Project, Geneva, 1992.
Yung, Y.L., J.P., Pinto, R.T., Watson, and S.P., Sander, At-
mospheric bromine and ozone perturbations in the lower
stratosphere, J. Atmos. Sci., 37, 339-353, 1980.
Zintl, E., and G. Rienicker, 0ber die Existenz eines
fifichtigen Bromoxyds, Bet., 63, 1,098-1,104, 1930.
R.A. Cox, R.L. Jones, and D.J. Lary, Centre For At-
mospheric Science, Department of Chemistry, University of
Cambridge, Lensfield Road, Cambridge, CB2 1EW, Eng-
land. (eraall: rac26@cam.ac.uk, rlj1001@cus.cam.ac.uk,
david @ at m. ch. cam. ac.uk)
O.V. Rattigan, Department of Chemistry, Boston Col-
lege, Chestnut Hill, Boston, MA 02167, USA. (email: ratti-
gao@bc.edu)
(Received August 8, 1995; revised May 25, 1996;
accepted June 19, 1996.)