Near IR photolysis of HO
2
NO
2
: Implications for HO
x
Ross J. Salawitch,
1
Paul O. Wennberg,
2
Geoffrey C. Toon,
1
Bhaswar Sen,
1
and Jean-Francois Blavier
1
Received
27
February
2002;
revised
16
May
2002;
accepted
21
May
2002;
publish
e
d
20
August
2002.
[
1
] We report observations and calculations of peroxynitric
acid, HO
2
NO
2
, in the stratosphere and upper troposphere.
The simulations show that photolysis of HO
2
NO
2
via
excitation of purely vibrational modes at wavelengths
longward of 760 nm (the near IR) can dominate loss of
this species. Consideration of this photolytic pathway
reduces calculated HO
2
NO
2
, resolving a large discrepancy
between standard model calculations and observations of
HO
2
NO
2
at high-latitude spring. The lower calculated
abundance of HO
2
NO
2
reduces the efficiency of the OH +
HO
2
NO
2
sink of HO
x
. Consideration of this process leads to
large increases in calculated HO
x
(20 to 60%) for high-
latitude spring and better agreement with observed
stratospheric abundances of HO
x
. Near IR photolysis of
HO
2
NO
2
alters the coupling between NO
x
and HO
x
in
stratospheric and upper tropospheric photochemical
models.
I
NDEX
T
ERM
:
0340 Atmospheric Composition
and Structure: Middle atmosphere—composition and chemistry
1. Introduction
[
2
] Peroxynitric acid, HO
2
NO
2
, is important to the photo-
chemistry of the upper troposphere and stratosphere. It is
formed by:
HO
2
þ
NO
2
þ
M
!
HO
2
NO
2
þ
M
ð
1
Þ
and removed by photolysis as well as reaction with OH:
OH
þ
HO
2
NO
2
!
Products
:
ð
2
Þ
Assuming reaction (2) produces H
2
O and O
2
, the sequence
(1) + (2) catalyzes the loss of HO
x
.
[
3
] The first measurements of HO
2
NO
2
were based on
spectra obtained by the Atmospheric Trace Molecule Spec-
troscopy experiment [
Rinsland et al.
, 1986]. Subsequent
laboratory measurements of HO
2
NO
2
line parameters [
May
and Friedl
, 1993] improved the precision of these and other
remote observations of HO
2
NO
2
[e.g.,
Sen et al.
, 1998].
Previous studies focused on profiles obtained at mid-lat-
itudes, where the discrepancy between modeled and meas-
ured HO
2
NO
2
is small [
Sen et al.
, 1998], well within the
range of model uncertainty.
[
4
] Recently, new measurements of photodissociation
cross sections of HO
2
NO
2
have been reported for both the
UV [
Knight et al.
, 2002] and the near IR [
Roehl et al.
,
2002]. The near IR study was motivated by the suggestion
that photodissociation by excitation of vibrational overtones
would significantly increase the overall photolysis rate of
HO
2
NO
2
[
Donaldson et al.
, 1997]. In light of these new
laboratory data, we examine here remote observations of
HO
2
NO
2
obtained at mid- and high-latitudes and in situ
measurements of HO
2
from high-latitude spring.
2. Observations and Model Description
[
5
] Measurements of HO
2
NO
2
were obtained by the JPL
MkIV balloon-borne Fourier Transform Infrared spectrom-
eter using solar occultation at 35
N on Sept. 25, 1993
during sunset and at high-latitude (65 to 70
N) on May 8,
1997 during sunrise. Details of these flights are given by
Sen et al.
[1998] and
Osterman et al.
[1999].
[
6
] Absorption of sunlight by the unresolved Q branch of
HO
2
NO
2
is apparent in measured spectra near 802.7 cm
1
(Figure 1). The high spectral resolution of MkIV allows this
broad absorption feature to be separated from the stronger,
narrower absorption lines of O
3
,H
2
O, and CO
2
. Profiles of
the volume mixing ratio (vmr) of HO
2
NO
2
(Figures 2 and 3)
are based on band intensities reported by
May and Friedl
[1993], which have an estimated accuracy of 20%. Meas-
ured profiles of HO
2
NO
2
are provided in tabular form in the
auxiliary material.
1
We also examine in situ measurements
of HO
2
obtained aboard the ER-2 aircraft near 64
N during
spring of 1997 [
Wennberg et al.
, 1999].
[
7
] The photochemical model has been used in many
previous studies [e.g.,
Sen et al.
, 1998;
Osterman et al.
,
1999]. The abundance of radicals (e.g., NO
2
and HO
2
) and
reservoir compounds (e.g., HO
2
NO
2
) are calculated by
balancing the production and loss of each species integrated
over 24 hours. For analyses of MkIV data, abundances of
long-lived precursors such as O
3
,H
2
O, NO
y
, and Cl
y
are
specified from MkIVobservations. Profiles of Br
y
are based
on correlations with N
2
O; aerosol surface area is based on
zonal, monthly mean profiles from the Stratospheric Aero-
sol and Gas Experiment II. Constraints for the MkIV
simulations are provided in tabular form in the auxiliary
material.
1
The model is similarly constrained by observa-
tions of long-lived precursors, aerosol surface area, etc. for
analyses of the ER-2 data. These calculations were carried
out as described by
Wennberg et al.
[1999], using the
constraints given in Table 1 of that paper.
[
8
] The simulations shown below for both the MkIV and
the ER-2 data use three sets of kinetic parameters. The first,
based on the most recent evaluation [
Sander et al.
, 2000], is
referred to as the
JPL00
model. The second set, denoted
Model B
, uses several modifications of potential importance
for HO
2
NO
2
:
a
) the rates for OH + O
3
and HO
2
+O
3
from
1
Auxiliary material is available via Web browser or anonymous FTP
from ftp://kosmos.agu.org, directory ‘‘apend’’; subdirectories in the ftp site
are arranged by paper number. Information on electronic supplements is at
http://www.agu.org/pubs/esupp_about.html.
GEOPHYSICAL RESEARCH LETTERS, VOL. 29, NO. 16, 10.1029/2002GL015006, 2002
1
Jet Propulsion Laboratory, California Institute of Technology,
Pasadena, CA, USA.
2
California Institute of Technology, Pasadena, CA, USA.
C
o
p
y
r
i
g
h
t
2
0
0
2
b
y
t
h
e
A
m
e
r
i
c
a
n
G
e
o
p
h
y
s
i
c
a
l
U
n
i
o
n
.
0
0
9
4
-
8
2
7
6
/
0
2
/
2
0
0
2
G
L
0
1
5
0
0
6
9
-
1
JPL97
[
DeMore et al.
, 1997] are used because they better
describe the ratio of HO
2
to OH observed in the lower
stratosphere and because they are in better agreement with
laboratory studies published subsequent to
JPL00
[
Lanzen-
dorf et al.
, 2001];
b
) the reaction Cl + HNO
4
is included at a
rate of 10
13
cm
3
sec
1
[
Simonaitis and Leu
, 1985];
c
)UV
absorption cross sections for HO
2
NO
2
recently measured by
Knight et al.
[2002] are adopted. A third set of calculations
uses the same set of kinetic parameters as
Model B
, except
photodissociation of HO
2
NO
2
in the near IR is added. This
calculation, referred to as
Near IR
, uses values for cross
sections, quantum yields, and solar flux given in Table 3 of
Roehl et al.
[2002].
3. HO
2
NO
2
Comparisons
[
9
] Figures 2 and 3 show comparisons of calculated and
observed HO
2
NO
2
. Calculated profiles found using
JPL00
kinetics exceed observed levels of HO
2
NO
2
at mid-latitudes
for altitudes between
20 to 30 km. However, this discrep-
ancy is well within the calculation uncertainty, based on
considerations such as a factor of 2 uncertainty in the
photolysis rate of HO
2
NO
2
[
DeMore et al.
, 1997]. For
high-latitude springtime,
JPL00
kinetics over estimates
observed HO
2
NO
2
by as much as a factor of 4. The
discrepancy at high-latitude is significant considering uncer-
tainties in both measured and modeled HO
2
NO
2
.
[
10
] Uncertainties in NO
x
chemistry can not explain the
discrepancy between measured HO
2
NO
2
at high latitude
spring and the
JPL00
simulation. The model simulates
observed profiles of NO
2
and NO quite well (differences
typically less than 10% for all three sets of kinetic param-
eters) for both mid-latitudes and high-latitudes, as shown in
the supporting material.
1
Revisions to the rates of OH +
NO
2
+ M and OH + HNO
3
in
JPL00
account for earlier
discrepancies between modeled and measured NO
2
at both
mid- and high-latitudes [e.g.,
Osterman et al.
, 1999].
[
11
] The kinetic parameters adopted for the
Model B
simulation result in lower values of calculated HO
2
NO
2
and better agreement with observation. The difference in
calculated HO
2
NO
2
compared to the
JPL00
simulation is
almost entirely due to the change in the rate constants of OH
+O
3
and HO
2
+O
3
. This change lowers modeled HO
2
and
increases OH; both of these effects lead to reductions in
calculated HO
2
NO
2
. The rate of the Cl + HO
2
NO
2
reaction,
included here for completeness, is about a factor of 100 too
slow to affect either HO
2
NO
2
or HCl. The
Knight et al.
[2002] UV cross sections result in nearly identical photol-
ysis rates of HO
2
NO
2
as
JPL00
kinetics for both mid- and
high-latitudes: weaker absorption between
280 to 325 nm
using the
Knight et al.
data is compensated by contributions
to photolysis longward of 325 nm. The
Knight et al.
measurements do, however, increase our confidence that
Figure 1.
Bottom Panel
: Measured transmittance (dia-
monds connected by black lines) for a MkIV spectrum
obtained on May 8, 1997 with a tangent altitude of 10.4 km.
The estimated contribution of the unresolved Q branch of
HO
2
NO
2
to the transmittance is shown (red line). Contribu-
tions from other species are also shown, as indicated.
Top
Panel
: Calculated residuals to the measured transmittance
found by the retrieval algorithm, allowing for absorption by
HO
2
NO
2
(black line) and neglecting absorption by HO
2
NO
2
(red line).
Figure 2.
Profile of HO
2
NO
2
measured by MkIV on Sept.
25, 1993 at 35
N (points w/ error bars) compared to model
simulations for three sets of kinetic parameters: 1)
JPL00
;2)
Model B
(see text); 3)
Near IR
photolysis of HO
2
NO
2
plus
Model B. Error bars for observed HO
2
NO
2
represent 1
s
precision. The 1
s
uncertainty in calculated HO
2
NO
2
for the
Near IR
model, based on uncertainties in rate constants of
its formation and loss processes, is shown by the error bars
bounded by vertical line segments.
Figure 3.
Same as Figure 2, for May 8, 1997 at 65 to
70
N.
9
-
2
SALAWITCH ET AL.: NEAR IR PHOTOLYSIS OF HO
2
NO
2
uncertainties in UV photolysis of HO
2
NO
2
can not account
for measured HO
2
NO
2
at high latitudes.
[
12
] Inclusion of near IR photolysis of HO
2
NO
2
leads to
good agreement between measured and modeled HO
2
NO
2
at both mid- and high-latitudes. Contributions from the near
IR dominate photolytic loss of HO
2
NO
2
below
24 km at
mid-latitudes and below about
28 km at high-latitudes.
The calculated photolysis rate due to excitation of purely
vibrational modes is nearly an order of magnitude larger
than estimated by
Wennberg et al.
[1999] because, as
discussed by
Roehl et al.
[2002], excitation of the first
overtone of OH stretching frequency (near 1.4
m
m) was
found to have a significant quantum yield for dissociation.
[
13
] Near IR photolysis has a larger effect at high-latitude
spring because of:
a
) strong attenuation of UV light by the
high ozone slant column (this lowers the UV contribution to
HO
2
NO
2
photolysis compared to mid-latitudes);
b
) long
days (the IR contribution to HO
2
NO
2
photolysis depends
essentially on length of sunlight). Calculated contributions
to loss of HO
2
NO
2
by UV photolysis, near IR photoloysis,
and reaction with OH are shown in the supplemental
material.
1
The fact that near IR photolysis makes the largest
contribution to calculated HO
2
NO
2
for the region of the
atmosphere where standard models (e.g.,
JPL00
or
Model
B
) most strongly overestimate measured HO
2
NO
2
provides
compelling evidence for the importance of this mechanism.
4. Implications for HO
x
[
14
] Figure 4 illustrates the effect of near IR photolysis of
HO
2
NO
2
on HO
x
. The increase in calculated HO
x
is modest
(2 to 6%) below
28 km at mid-latitudes. The increase is
dramatic (20 to 60%), however, for high-latitude spring at
altitudes below
28 km. The perturbation extends into the
upper troposphere (UT) for both regions and is large (50 to
60%) in the UT at high-latitudes.
[
15
] Changes in calculated HO
x
are largest for high
latitude spring below
28 km for reasons described above
(weaker UVand longer days result in a stronger perturbation
to HO
2
NO
2
) plus the presence of characteristically high
levels of NO
x
. The loss of HO
x
by (1)–(2) is catalyzed by
NO
x
; for the
JPL00
simulation, this cycle plays a stronger
role in HO
x
photochemistry at high-latitude spring com-
pared to mid-latitudes. The effect of near IR HO
2
NO
2
photolysis on calculated HO
x
depends, therefore, on ambi-
ent NO
x
. This dependence is important for assessing the
tropospheric implications of this process.
[
16
] The influence of near IR photolysis of HO
2
NO
2
on
HO
x
is further illustrated in Figure 5, which compares
modeled and measured HO
2
in the lower stratosphere near
64
N during spring. The
JPL00
simulation significantly
underestimates measured HO
2
throughout the day. An
important change in the
JPL00
kinetic parameters, relative
to
JPL97
, is a reduction in the reaction probability (
g
)for
BrONO
2
hydrolysis from 0.8 to
0.2 for the temperature
and humidity of these observations. This change reduces the
morning rise of HO
2
due to photolysis of HOBr apparent in
the
JPL97
simulation of
Wennberg et al.
[1999].
Model B
kinetic parameters result in a slight increase in calculated
HO
2
. Allowing for near IR photolysis leads to a
20%
increase in 24 hr average HO
2
and better agreement with the
observed rise of HO
2
in early morning due to rapid
photolysis of HO
2
NO
2
at high solar zenith angles. None-
theless, measured HO
2
and OH (figure shown in supple-
mental material
1
) are still underestimated by this model.
[
17
] The discrepancy between measured HO
2
and OH
and the
Near IR
model calculation is not significant given
the estimated possible 30% systematic error in the HO
x
measurements [
Wennberg et al.
, 1999] (see auxiliary mate-
rial
1
). However, the precision of the HO
x
data is much
better; e.g., any systematic error will not vary with solar
zenith angle (SZA). The shapes of HO
2
vs. SZA and OH vs.
SZA from the
Near IR
simulation agree well with observed
SZA dependencies. The discrepancy between measured and
modeled HO
x
for the
JPL00
and
Model B
simulations is
significant because the measured shape of HO
x
vs. SZA
differs considerably from the model calculations.
[
18
] Nonetheless, it is interesting to speculate on possible
reasons for the shortfall between calculated HO
x
from the
Figure 4.
Calculated change in 24 hour average HO
x
(OH
+HO
2
) when near IR photolysis is added to a simulation
based on
Model B
kinetics (solid line) and to a simulation
based on
JPL00
kinetics (dashed line).
Figure 5.
Observations of HO
2
obtained on the morning of
April 30, 1997 and the afternoon of May 9, 1997 from the
ER-2 aircraft near 64
N in the lower stratosphere compared
to model simulations using the three sets of kinetic
parameters described in Figure 2. A fourth simulation,
allowing for a reaction probability of 0.8 for BrONO
2
hydrolysis within the
Near IR
model, is also shown.
SALAWITCH ET AL.: NEAR IR PHOTOLYSIS OF HO
2
NO
2
9
-
3
Near IR
simulation and the HO
x
observations (assuming the
HO
x
calibration is correct). It is unlikely that this shortfall can
be entirely due to errors in the photolysis rate of HO
2
NO
2
.
Sensitivity studies indicate the integrated morning burst of
HO
x
supplied by photolysis of HO
2
NO
2
is limited once
photolysis becomes the dominant sink of HO
2
NO
2
(e.g.,
the concentration of HO
2
NO
2
becomes inversely propor-
tional to its photolysis rate). If the rate of OH + HO
2
NO
2
is
reduced to its lower limit, then calculated HO
x
within the
Near IR
simulation lies close to the observations. Similarly, if
this reaction does not yield H
2
O with 100% efficiency, model
and measured HO
x
are in better agreement.
[
19
] Another possible explanation of the measured HO
x
is
that
g
of BrONO
2
hydrolysis, for conditions of these obser-
vations, is considerably faster than 0.2. The
JPL00
recom-
mendation for BrONO
2
hydrolysis is based on laboratory
results extrapolated to water activity levels of the lower
stratosphere. The model simulation labeled ‘‘
g
= 0.8’’ in
Figure 5 shows that the observed morning burst of HO
2
might
be supplied by photolysis of both HO
2
NO
2
and HOBr. The
slight timing difference between the measured morning rise
of HO
2
and the
g
= 0.8 simulation could be indicative of
errors in the calculated actinic flux, errors in the photolysis
rate of HOBr and/or HO
2
NO
2
, or supply of HO
x
from some
other precursor. The good agreement between the modeled
and measured early morning rise of NO suggests the actinic
flux calculation is carried out correctly [
Gao et al.
, 2001].
Calculations shown in the supplemental material
1
indicate
that BrONO
2
hydrolysis, in the absence of near IR photolysis
of HO
2
NO
2
, is unable to account for the observed fall off of
HO
2
with increasing SZA in the evening regardless of
assumptions regarding
g
or Br
y
.
5. Concluding Remarks
[
20
] Near IR photolysis of HO
2
NO
2
, first suggested by
Donaldson et al.
[1997], alters the coupling between NO
x
and HO
x
. In the lower stratosphere (LS), HO
x
radicals are
the dominant sink for photochemical loss of O
3
.The
increased levels of HO
x
suggested by this analysis will
likely result in greater sensitivity of calculated O
3
to
perturbations such as increases in stratospheric H
2
O. In
the upper troposphere (UT), production of ozone via oxi-
dation of carbon monoxide and other hydrocarbons pro-
ceeds via coupled HO
x
:NO
x
photochemistry. The efficiency
of this chemistry is thought to become limited at moderate
concentrations of NO
x
due to loss of HO
x
by reactions (1)
and (2) [e.g.,
Jaegle
́etal.
, 2001]. The reduced efficiency of
this sink implies that production of UT ozone will be
positively correlated with the abundance of NO to higher
levels of NO
x
than is found in many models.
[
21
] We conclude by noting that space-borne observa-
tions of HO
2
NO
2
can provide significant advances in
constraints on the photochemistry of the UT and LS.
Measurements of HO
2
NO
2
co-located with observations
of NO
2
provide a means to determine concentrations of
HO
2
. Near IR photolysis reduces the lifetime of HO
2
NO
2
,
simplifying the interpretation of observations. There is hope
that observations of HO
2
NO
2
will become available in the
near future, since it is measured by the Michelson Interfer-
ometer for Passive Atmospheric Sounding instrument
aboard the ESA ENVISAT spacecraft and is a ‘‘special
product’’ of the Tropospheric Emission Spectrometer to be
launched on the NASA Aura spacecraft.
[
22
]
Acknowledgments.
We thank Gary Knight and Coleen Roehl for
making available measurements of HO
2
NO
2
cross sections prior to pub-
lication and the anonymous reviewers for helpful comments. This work was
funded by the NASA Upper Atmosphere Research, Atmospheric Chemistry
Modeling and Analysis, and Atmospheric Effects of Aviation Programs.
Research at the Jet Propulsion Laboratory, California Institute of Technol-
ogy, is performed under contract with the National Aeronautics and Space
Administration.
References
DeMore, W. B., et al., Chemical kinetics and photochemical data for use in
stratospheric modeling, Evaluation No. 12,
JPL Publication 97-4
, Jet
Propulsion Laboratory, 1997.
Donaldson, D. J., et al., Atmospheric radical production by excitation of
vibrational overtones via absorption of visible light,
Geophys. Res. Lett.
,
24
, 2651–2654, 1997.
Gao, R. S., et al., J
NO2
at high solar zenith angles in the lower stratosphere,
Geophys. Res. Lett.
,
28
, 2405–2408, 2001.
Jaegle
́, L., D. J. Jacob, W. H. Brune, and P. O. Wennberg, Chemistry of
HO
x
radicals in the upper troposphere,
Atmos. Envir.
,
35
, 469–489, 2001.
Knight, G., A. R. Ravishankara, and J. B. Burkholder, UV absorption cross
sections of HO
2
NO
2
between 343 and 273 K,
J. Phys. Chem. A.
, sub-
mitted, 2002.
Lanzendorf, E. J., et al., Comparing atmospheric [HO
2
]/[OH] to modeled
[HO
2
]/[OH]: Identifying discrepancies with reaction rates,
Geophys. Res.
Lett.
,
28
, 967–970, 2001.
May, R. D., and R. R. Friedl, Integrated band intensities of HO
2
NO
2
at
220 K,
J. Quant. Spect. Rad. Trans.
,
50
, 257–266, 1993.
Osterman, G. B., et al., Partitioning of NO
y
species in the summer Arctic
stratosphere,
Geophys. Res. Lett.
,
26
, 1157–1160, 1999.
Rinsland, C. P., et al., Evidence for the presence of the 802.7 cm
1
band Q
branch of HO
2
NO
2
in high resolution solar absorption spectra of the
stratosphere,
Geophys. Res. Lett.
,
13
, 761–764, 1986.
Roehl, C. A., et al., Photodissociation of HNO
4
in the near IR,
J. Phys.
Chem. A.
,
106
, 3766–3772, 2002.
Sander, S. P., et al., Chemical kinetics and photochemical data for use in
stratospheric modeling, Evaluation No. 13,
JPL Publication 00-3
, Jet
Propulsion Laboratory, 2000.
Sen, B., et al., Measurements of reactive nitrogen in the stratosphere,
J. Geophys. Res.
,
103
, 3571–3585, 1998.
Simonaitis, R., and M. T. Leu, Rate constant for the reaction Cl + HO
2
N-
O
2
!
Products,
Int. J. Chem. Kin.
,
17
, 293–301, 1985.
Wennberg, P. O., et al., Twilight observations suggest unknown sources of
HO
x
,
Geophys. Res. Lett.
,
26
, 1373–1376, 1999.
R. J. Salawitch, G. C. Toon, B. Sen, and J.-F. Blavier, JPL, Mail Stop
183-601, 4800 Oak Grove Drive, Pasadena, CA 91109, USA.
P. O. Wennberg, MS 150-21, Caltech, Pasadena, CA 91125, USA.
9
-
4
SALAWITCH ET AL.: NEAR IR PHOTOLYSIS OF HO
2
NO
2