Large and unexpected enrichment in stratospheric
16
O
13
C
18
O and its meridional variation
Laurence Y. Yeung
a,1,2
, Hagit P. Affek
b,1
, Katherine J. Hoag
c,3
, Weifu Guo
e,4
, Aaron A. Wiegel
d
, Elliot L. Atlas
f
,
Sue M. Schauffler
g
, Mitchio Okumura
a
, Kristie A. Boering
c,d
, and John M. Eiler
e
Divisions of
a
Chemistry and Chemical Engineering and
e
Geological and Planetary Sciences, California Institute of Technology, Pasadena, CA 91125;
b
Department of Geology and Geophysics, Yale University, New Haven, CT 06511; Departments of
c
Chemistry and
d
Earth and Planetary Science, University of
California, Berkeley, CA 94720;
f
Division of Marine and Atmospheric Chemistry, University of Miami, Miami, FL 33149; and
g
National Center for Atmospheric
Research, Boulder, CO 88307
Edited by Mark H. Thiemens, University of California at San Diego, La Jolla, CA, and approved May 26, 2009 (received for review March 16, 2009)
The stratospheric CO
2
oxygen isotope budget is thought to be
governed primarily by the O(
1
D)
CO
2
isotope exchange reaction.
However, there is increasing evidence that other important phys-
ical processes may be occurring that standard isotopic tools have
been unable to identify. Measuring the distribution of the exceed-
ingly rare CO
2
isotopologue
16
O
13
C
18
O, in concert with
18
O and
17
O
abundances, provides sensitivities to these additional processes
and, thus, is a valuable test of current models. We identify a large
and unexpected meridional variation in stratospheric
16
O
13
C
18
O,
observed as proportions in the polar vortex that are higher than in
any naturally derived CO
2
sample to date. We show, through
photochemical experiments, that lower
16
O
13
C
18
O proportions
observed in the midlatitudes are determined primarily by the
O(
1
D)
CO
2
isotope exchange reaction, which promotes a stochas-
tic isotopologue distribution. In contrast, higher
16
O
13
C
18
O propor-
tions in the polar vortex show correlations with long-lived strato-
spheric tracer and bulk isotope abundances opposite to those
observed at midlatitudes and, thus, opposite to those easily ex-
plained by O(
1
D)
CO
2
. We believe the most plausible explanation
for this meridional variation is either an unrecognized isotopic
fractionation associated with the mesospheric photochemistry of
CO
2
or temperature-dependent isotopic exchange on polar strato-
spheric clouds. Unraveling the ultimate source of stratospheric
16
O
13
C
18
O enrichments may impose additional isotopic constraints
on biosphere–atmosphere carbon exchange, biosphere productiv-
ity, and their respective responses to climate change.
clumped isotopes
CO
2
mesosphere
polar vortex
stratosphere
P
redicting future CO
2
concentrations and carbon cycle–
climate feedbacks depends on accurately quantifying the
contributions from the sources and sinks governing the global
carbon budget and how they may change over time. The bulk
stable isotope composition of CO
2
(i.e., its
13
C/
12
C,
18
O/
16
O and
17
O/
16
O ratios) plays an important role in constraining this
budget (ref. 1 and references therein). In the stratosphere, the
oxygen isotopic composition of CO
2
is thought to be modified by
oxygen isotope exchange reactions with O(
1
D) generated by
ozone photolysis (2), whereas in the troposphere it is controlled
by isotope exchange reactions with liquid water in the oceans,
soils, and plant leaves (3). The interplay between stratospheric
and tropospheric isotope exchange reactions, in principle, could
allow the relative abundances of
12
C
16
O
2
,
16
O
12
C
18
O, and
16
O
12
C
17
O isotopologues to be used as tracers for gross bio-
sphere productivity (4, 5), but the stratospheric CO
2
photochem-
ical system is still underconstrained, and our understanding of it
is incomplete.
Discrepancies between laboratory and stratospheric measure-
ments (6–8) have prompted questions about whether the
O(
1
D)
CO
2
isotope exchange reaction acts alone on strato-
spheric CO
2
or whether other photochemical (9) or dynamical
(10) processes significantly affect the stable isotopologue distri-
bution in CO
2
. Stratospheric oxygen isotope covariations in CO
2
(8, 11, 12) consistently differ from those found in laboratory
experiments simulating stratospheric photochemistry (6, 13–16),
and the origin of this disagreement is still uncertain because bulk
stable isotope measurements (i.e., of
13
C,
17
O, and
18
O
values) alone cannot differentiate extrinsic effects [e.g., O(
1
D)
isotope composition] from intrinsic (i.e., photolytic or kinetic
isotope) effects on the isotope composition of stratospheric CO
2
.
Additional constraints arising from the analysis of multiply-
substituted isotopologues of CO
2
can provide sensitivities to
these processes (17–20). To investigate stratospheric CO
2
chem-
istry, we examine here the proportions of the
16
O
13
C
18
O (re-
ported as a
47
value; see
Materials and Methods
) in stratospheric
CO
2
and in laboratory kinetics experiments.
Field and Laboratory Results
Six samples of stratospheric CO
2
collected from the NASA ER-2
aircraft during the 1999/2000 Arctic winter Stratospheric Aero-
sol and Gas Experiment III (SAGE III) Ozone Loss and
Validation Experiment (SOLVE) campaign (21) and 6 addi-
tional CO
2
samples collected from a balloon flight from Fort
Sumner, NM, on September 29, 2004 were analyzed for bulk
stable isotope compositions and
47
values (see
Materials and
Methods
and Table 1). The stratospheric samples display
47
values both higher and more variable than those exhibited by
tropospheric air at the surface, which has an average
47
value
of 0.92
0.01‰ in remote regions (Cape Grim, Tasmania and
Barrow, Alaska) (18). The high-latitude (
57°N) stratospheric
samples, in particular, are significantly more enriched in
47
than
any material analyzed before in nature (see Fig. 1 and refs. 18
and 20). At high latitudes,
47
varies strongly and increases
monotonically with decreasing mixing ratios of long-lived trace
gases that have tropospheric sources and stratospheric sinks,
such as N
2
O and CH
4
. At midlatitudes,
47
varies relatively little
and shows correlations with trace-gas mixing ratios having the
opposite sign of those observed at high latitudes [e.g., Fig. 2
A
;
see
supporting information (SI)
]. Using the correlation between
simultaneously measured N
2
O mixing ratios and potential tem-
Author contributions: L.Y.Y., H.P.A., M.O., and J.M.E. designed research; L.Y.Y., H.P.A.,
K.J.H., W.G., E.L.A., S.M.S., and K.A.B. performed research; L.Y.Y., H.P.A., W.G., and A.A.W.
performed modeling; L.Y.Y., H.P.A., K.J.H., A.A.W., K.A.B., and J.M.E. contributed new
reagents/analytic tools; L.Y.Y., H.P.A., W.G., A.A.W., M.O., K.A.B., and J.M.E. analyzed data;
and L.Y.Y., H.P.A., W.G., A.A.W., K.A.B., and J.M.E. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
1
L.Y.Y. and H.P.A. contributed equally to this work.
2
To whom correspondence should be addressed. E-mail: lyeung@caltech.edu.
3
Presentaddress:UnitedStatesEnvironmentalProtectionAgency(Region9),SanFrancisco,
CA 94105.
4
Present address: Geophysical Laboratory, Carnegie Institution of Washington, Washing-
ton, DC 20015.
This article contains supporting information online at
www.pnas.org/cgi/content/full/
0902930106/DCSupplemental
.
11496–11501
PNAS
July 14, 2009
vol. 106
no. 28
www.pnas.org
cgi
doi
10.1073
pnas.902930106
perature (22), we found that the high-latitude samples were
collected in either inner polar vortex air (3 samples) or vortex
edge air (1 sample). Thus, differences in the sign and magnitude
of
47
variations with N
2
O mixing ratio in the 2 subsets of
samples are presumably due to differences between vortex and
nonvortex processes. In contrast to these
47
trends, covariations
of
17
O with
18
O and mixing ratios of long-lived trace gases for
our midlatitude samples resemble those of our polar vortex
samples (or show the expected differences for slightly shorter-
lived tracers, e.g., CFC-11) and also those reported in earlier
studies of high-latitude (10, 11) and polar vortex air (23) (see
SI
).
This suggests that
47
records a process to which
17
Oisnot
sensitive.
To gain further insight into the correlations we observed
between
47
,
17
O, and
18
O, we conducted several laboratory
photochemistry experiments. Enrichments of
17
O and
18
Oin
stratospheric CO
2
are generally understood to arise from isotope
exchange with O(
1
D), which is produced by photolysis of
17
O-
and
18
O-enriched stratospheric ozone (2, 24), but the kinetics of
the isotope exchange reactions for multiply-substituted isotopo-
logues of CO
2
, and their effects on
47
, have not yet been
explored. Changes in
47
caused by these reactions should
principally reflect relative changes in
16
O
13
C
18
O and
16
O
12
C
18
O
abundances because the
13
C composition is not affected by
isotope exchange with O(
1
D); variations in
17
O
12
C
18
O and
13
C
17
O
2
abundances will change
47
more subtly (see
SI
). Thus,
reactions 1 and 2 are most relevant.
18
O
1
D
12
C
16
O
2
3
18
O
12
C
16
O
*
2
3
16
O
16
O
12
C
18
O
[1]
18
O
1
D
13
C
16
O
2
3
18
O
13
C
16
O
*
2
3
16
O
16
O
13
C
18
O
[2]
Statistical partitioning of isotope exchange products, through a
process governed by random chance such that
18
O(
1
D) has a 2/3
probability of being incorporated into the product CO
2
, would
drive the CO
2
isotopologues toward a stochastic distribution (i.e.,
47
0). If any one (or more) of the isotope exchange reactions
does not partition products statistically, however,
47
can in-
crease (e.g., if
18
O(
1
D) exchanges isotopes with
13
C
16
O
2
more
readily than with
12
C
16
O
2
) or decrease accordingly.
Two sets of laboratory experiments were conducted. First, the
relative rates of reactions 1 and 2 were probed directly to obtain
the
13
C/
12
C kinetic isotope effect (KIE) for the
18
O(
1
D)
CO
2
isotope exchange reaction at 300 K and 229 K. These experi-
ments used pulsed photolysis of 97% N
2
18
O at 193 nm as a source
of
18
O(
1
D), and the extent of isotope exchange was minimized.
Table 1. Stratospheric air sample data
Air type*
Sample name
Altitude, km Latitude, °N N
2
O, ppbv CH
4
, ppbv
,K
13
C, ‰
18
O, ‰
17
O, ‰
47
,‰
ER-2 samples
V
20000312(25)1120
19.33
79.2
111.5
948 443.61
8.11
44.55
5.76
1.614
V
20000131(30)1001
19.85
73.5
142.1
1,066 448.08
8.08
43.62
4.38
1.555
V
20000203(10)1173
17.63
70.4
192.8
1,250 409.37
8.06
43.06
3.69
1.436
VE
20000127(5)1060
19.45
57.7
227.6
1,385 445.10
8.07
42.79
2.45
1.233
M
20000106(30)1169
19.40
29.3
256.4
1,495 467.43
8.08
42.50
2.13
1.071
M
20000111(25)2021
11.40
43.5
307.9
1,726 358.26
8.06
40.98
0.18
1.075
Balloon samples
M
1-A010-R (2035)
33.34
34.5
54.0
835 931.14
8.07
45.14
6.46
0.976
M
3-A01-R (1141)
32.22
34.6
56.7
843 896.15
7.99
45.44
6.29
0.923
M
7-A026-R (1057)
29.26
34.6
92.1
957 778.08
8.07
44.98
6.12
0.912
M
5-A013-R (2079)
30.78
34.6
77.8
913 861.84
8.03
45.14
5.91
0.913
M
8-A017-E (1113)
28.73
34.6
107.9
1,014 757.89
8.04
44.79
5.00
1.004
M
10-A022-E (1186)
27.27
34.6
155.9
1,176 709.67
8.03
44.09
4.26
0.916
*Air type is abbreviated as V, vortex; VE, vortex edge; and M, midlatitude, based on N
2
O/
correlations, where
is potential temperature in Kelvin.
Fig. 1.
Meridional variation of
47
measured in stratospheric samples.
Typical values in the troposphere are
47
0.9‰. Error bars show 2
standard
errors.
Fig. 2.
Correlations between
47
and stratospheric tracers. (
A
), best-fit lines
are shown for midlatitude (solid line) and high-latitude (dashed line)
47
vs.
N
2
O mixing ratio. Correlations between
47
and other tracers with tropo-
spheric sources and stratospheric sinks are similar (see
SI
). (
B
and
C
) Best-fit
lines of
47
vs.
18
O(
B
) and
47
vs.
17
O(
C
) in midlatitude air, which are used
to estimate the integrated effective isotopic composition of stratospheric
O(
1
D) (see
O(
1
D)
CO
2
Explains Midlatitude but Not Polar Vortex
16
O
13
C
18
O
Variations
and
Materials and Methods
). Error bars show 2
standard errors.
Yeung et al.
PNAS
July 14, 2009
vol. 106
no. 28
11497
GEOPHYSICS
They were performed in an excess of helium to probe a strato-
spherically relevant range of collision energies (see
SI
) because
the reaction dynamics are sensitive to reactant collision energies
(25). Second, we performed continuous irradiation experiments,
using a mercury lamp, on mixtures of isotopically unlabeled O
2
,
O
3
, and CO
2
as a function of different irradiation times to
determine the cumulative effects of the O
2
/O
3
/CO
2
photochem-
ical system on
47
.
In both sets of experiments,
47
of CO
2
decreases with
increasing extent of photochemical isotope exchange, as mea-
sured by the change in
18
O(
18
O; see Fig. 3) or
17
OofCO
2
(see
SI
), indicating the O(
1
D)
CO
2
reaction does not selectively
enrich
16
O
13
C
18
O relative to
16
O
12
C
18
O. Furthermore, these
results are consistent with the statistical partitioning of isotope
exchange products; using the results from the
18
O-labeled ex-
periments at 300 K and 229 K and a comprehensive kinetics
model of the pulsed photolysis experiment, we calculate a
13
C/
12
C KIE of 0.999
0.001 (2
est.), which is not significantly
different from the KIE of 1.000 expected for statistical isotope
exchange branching fractions and is quantitatively consistent
with the results from the continuous irradiation experiments.
Hence, our results are consistent with previous studies asserting
that the O(
1
D)
CO
2
isotope exchange products are partitioned
statistically (14, 16, 25). In contrast, a
13
C/
12
C KIE of
1.01
would be required to explain the high-latitude isotopic correla-
tions shown in Fig. 2
B
and
C
(see
SI
).
It is possible that the measured KIEs depend on reactant
collision energies, which can have a nonthermal distribution in
the stratosphere (26). Previous work has shown that the
O(
1
D)
CO
2
isotope exchange reaction occurs through 2 reaction
channels whose branching ratio varies with collision energy (25):
O
1
D
CO
2
3
O
3
P
CO
2
[3a]
3
O
1
D
CO
2
[3b]
Under typical stratospheric conditions, the channel shown as
Reaction
3a
dominates. However, the contribution from the
nonquenching isotope exchange channel, shown as Reaction
3b
,
increases significantly with reactant collision energy (25). Be-
cause the KIEs for the channels shown as Reactions
3a
and
3b
may differ, both between the 2 channels and as a function of
collision energy (27), the overall KIE in the lab or atmosphere
may also depend on these parameters. High-collision-energy (no
buffer gas) experiments, which tip the 3a/3b branching ratio
toward the reaction shown as Reaction
3b
, were performed to
test this hypothesis. Similar to the low-collision-energy experi-
ments described above, these experiments also showed deple-
tions in
47
as the extent of photochemical isotope exchange
increased, with no significant change in the
47
vs.
18
O trend
(see
SI
). This indicates that the KIE we report is relatively
insensitive to the 3a/3b branching ratio. Still, more experimental
studies will be required to quantify these channel-specific KIEs.
O(
1
D)
CO
2
Explains Midlatitude but Not Polar Vortex
16
O
13
C
18
O Variations
We conclude that intrinsic isotope effects in O(
1
D)
CO
2
iso-
tope exchange cannot produce the elevated values of
47
ob-
served in stratospheric polar vortex air. Instead, the statistical
partitioning of isotope exchange products in the O(
1
D)
CO
2
reaction drives the isotopic composition of CO
2
toward a sto-
chastic distribution, thus decreasing
47
. Indeed, the midlatitude
observations show evidence for decreasing
47
with increasing
extent of photochemical isotope exchange (see Fig. 2
B
and
C
).
As photochemical isotope exchange approaches completion (the
unlikely, but instructive case in which all of the oxygen atoms in
CO
2
have been exchanged), the
17
O and
18
OofCO
2
will
approach the
17
O and
18
OofO(
1
D) because the size of the
oxygen reservoir is
500 times that of CO
2
. Concurrently, the
distribution of stable isotopes will become increasingly random,
so the
47
value of CO
2
will approach zero. The slope of this
approach, then, reflects the integrated effective isotope com-
position of O(
1
D) with which the CO
2
molecules in the air
samples have exchanged since entering the stratosphere; it is
possible, however, that other terms in the stratospheric CO
2
budget (e.g., CO oxidation) may, more subtly, change the
isotopic composition of CO
2
and thus the isotopic composition
of O(
1
D) one would infer from such a trend. Using a box model
and the midlatitude observations, we estimate
17
O
80.6
24.1
59.7
‰ and
18
O
98.0
17.1
43.7
‰(2
est.) as values for the
integrated effective isotopic composition of O(
1
D), which is
consistent with ozone photolysis being the primary O(
1
D) source
in the stratosphere. Although the
18
OofO(
1
D) we calculate is
similar to that found in low- and midlatitude stratospheric ozone,
17
O is larger, on average, by 35‰ (28). This is expected
because
17
O enrichments in ozone are concentrated at terminal
atom positions (29); when ozone is photolyzed to O(
1
D) and O
2
,
one of the terminal atoms is ejected (30), elevating the propor-
tion of heavy oxygen isotopes in O(
1
D). The uncertainty bounds
reported here reflect spatial and temporal variations in O(
1
D)
isotope composition expected to be caused by variations in ozone
concentrations, temperature, pressure, actinic flux, and various
mass-dependent kinetic and photolytic isotope effects that de-
pend on these variables.
Having experimentally excluded intrinsic isotope effects in the
O(
1
D)
CO
2
isotope exchange reaction as the source of high
stratospheric
47
in polar vortex air, we consider the possible
effects of other gas-phase stratospheric processes on the CO
2
isotopologue budget. CO is produced in the stratosphere by CH
4
oxidation and destroyed with an e-folding time of
2–3 months
by reaction with OH radicals to form CO
2
. We estimate that up
to 0.9 ppmv CO
2
could be derived from CH
4
oxidation in our
samples based on the difference between the observed CH
4
concentrations (Table 1) and the average CH
4
concentration in
air entering the stratosphere from the troposphere (1.7 ppmv).
12
C/
13
C KIEs for CH
4
OH, CH
4
O(
1
D), and CH
4
Cl of 1.004,
1.01, and 1.07, respectively (31), should yield
13
C-depleted CH
3
radicals at all latitudes, particularly in the polar vortex, where Cl
levels are elevated. CH
3
radicals then undergo several rapid
oxidation steps to form formaldehyde (CH
2
O). Isotope effects in
CH
2
O photolysis to form CO also favor the light isotopologues
(32). These isotope effects, combined with expected
13
C com-
Fig. 3.
Results of laboratory photochemical experiments. Changes in
47
vs.
18
O after pulsed UV photolysis at 300 K (circles) and 229 K (triangles) are
shown.
18
O
18
O
final
18
O
initial
. Also shown is the modeled
47
vs.
18
O
dependence for
13
C/
12
C KIE
13
k
/
12
k
0.999 and 1.01 (offset by
47
0.13‰
for clarity; the significance of this and of the slopes is discussed below and in
SI
). Error bars show 2
standard errors.
11498
www.pnas.org
cgi
doi
10.1073
pnas.902930106
Yeung et al.
positions for stratospheric CH
4
in these samples (
13
C
47 to
35‰) should produce CO that is depleted in
13
C relative to
background stratospheric CO
2
. Small inverse
12
C/
13
C,
16
O/
18
O,
and
16
C/
17
O KIEs (i.e.,
1) in the reaction CO
OH to form CO
2
at low pressures (33) imply that, in principle, CO
OH reactions
could increase
47
values in the product CO
2
, whereas at higher
pressures, the CO
OH reaction yields
13
C-,
18
O-, and
17
O-
depleted CO
2
(34). Ultimately, however, the size of
47
values in
the up to 0.9 ppmv CO
2
derived from CH
4
oxidation would likely
be orders of magnitude too small to increase
47
values signif-
icantly in the other
365 ppmv of background stratospheric CO
2
.
A possibility remains that unusual timing in the competition
between oxidation and photolysis of species in the polar vortex,
combined with unexpectedly large carbon and oxygen KIEs in
some Cl and Br reactions (e.g., Br
CH
2
O; see ref. 35) or other
gas-phase processes might be responsible for the midlatitude–
polar vortex differences in
47
given the large differences in Cl
and Br concentrations in polar vortex vs. midlatitude air, but this
seems unlikely in light of the known chemistry and KIEs.
Stratospheric mixing of 2 or more reservoirs of isotopically
distinct CO
2
could also produce higher
47
values in the resulting
mixtures because mixing 2 isotopically distinct CO
2
reservoirs
can produce nonstochastic abundances of multiply-substituted
isotopologues. This nonlinear dependence of
47
on mixing
arises because the reference against which the measured
47
CO
2
/
44
CO
2
ratio is compared (i.e., the mixture’s stochastic distribu-
tion) is not constant upon mixing; this reference is defined by,
and therefore varies with, the bulk stable isotope ratios of that
mixture. In contrast, the references used in bulk stable isotope
measurements are external standards whose values are constant
(e.g., VSMOW), so the bulk isotope ratios vary linearly with
mixing (see
SI
). The degree of nonlinearity in
47
is a function
of the differences in bulk isotopic composition between the
mixing end-member reservoirs and their mixing fractions (17,
19). We calculate that bulk isotope compositions of the 2 CO
2
reservoirs must differ in
18
O and/or
13
C by orders of magnitude
not previously observed in the stratosphere to generate
47
enrichments of
0.7‰ observed in polar vortex samples.
Effects of Mesospheric and Heterogeneous Chemistry
Physical or chemical processes in the mesosphere may yield
extreme isotopic enrichments in CO
2
or CO such that subsidence
of mesospheric air into the 1999/2000 stratospheric polar vortex
(36) could explain the observed meridional variation in
47
.
Here, we consider 3 mesospheric processes that have been
previously studied in other contexts: Gravitational separation of
upper atmospheric air, UV photolysis of O
2
, and UV photolysis
of CO
2
. Gravitational separation can concentrate heavy isoto-
pologues of CO
2
into the lower mesosphere (37), but the isotopic
enrichments we calculate, based on the expression reported by
Craig et al. (38), are insufficient to generate significant
47
changes (see
SI
). Alternatively, isotope effects in the photolysis
of O
2
by short-wavelength UV radiation in the upper meso-
sphere might lead to unusually enriched CO
2
. Liang et al. (24)
predicted recently that differences in photolysis cross-sections
between light and heavy isotopologues of O
2
in the narrow solar
Lyman-
region (121.6 nm) may result in a population of
mesospheric O(
1
D) enormously enriched in
17
O and
18
O. This
extreme enrichment in O(
1
D) could then be transferred to
mesospheric CO
2
through the O(
1
D)
CO
2
isotope exchange
reaction. CO
2
photolysis in the mesosphere to form CO, followed
by oxidation of that CO by OH in the polar vortex, might also
significantly affect the
47
values of CO
2
in the polar vortex.
To evaluate the effects of
17
O- and
18
O-enriched mesospheric
CO
2
mixing into the stratospheric polar vortex, we constructed
a 3-component mixing model that included air mass contribu-
tions from the troposphere, stratosphere, and mesosphere. Each
polar vortex datum (i.e., with unique
17
O,
18
O, and
47
values)
was fit individually because of the presence of mesospheric
filaments in the polar vortex (36) and consequent heterogeneity
in the mixing fraction and end-member isotopic composition
between samples. Using the mesospheric
18
O and
17
O enrich-
ments in CO
2
mixing end-member suggested by Liang et al. (i.e.,
18
O
10,603‰,
17
O
3,149‰, and
47
0; see
Materials
and Methods
), we were unable to reproduce the polar vortex data
in both bulk isotope compositions and
47
simultaneously. Only
after
18
O and
17
O of the mesospheric CO
2
end-member were
increased an additional 10-fold (i.e.,
18
O
10
5
‰ and
17
O
10
4
‰) were the
47
-
18
O-
17
O systematics of the polar vortex
samples reproduced. No laboratory measurements of the isoto-
pic fractionations in O
2
due to Lyman-
photolysis are available
to compare with this prediction, however. Furthermore, we note
that the O(
1
D)
CO
2
reaction may not necessarily partition
products statistically (i.e., drive
47
3
0) in the mesosphere, as
we assumed here, because the average collision energy there is
higher than in the stratosphere.
In principle, measurements of the isotopic composition of
mesospheric CO should constrain the isotopic composition of
mesospheric CO
2
downwelling into the polar vortex. Large
elevations in CO above background stratospheric levels, ob-
served at high altitudes in the polar vortex (36), are due to CO
2
photolysis in the mesosphere and subsequent transport to the
stratosphere, so the isotopic composition of CO in these cases
should reflect that of the mesospheric CO
2
population and any
fractionations arising from CO
2
photolysis there. The bulk
isotopic composition in mesospheric CO
2
predicted by our
3-component mixing model (e.g., tens of percentage in
18
O-atom
abundance, or 100 times natural abundance) would result in
isotopic enrichments in mesospheric CO that are detectable with
remote-sensing spectrometers because CO
2
photolysis isotope
effects are expected to be much smaller in magnitude. For
example, Bhattacharya et al. (9) measured fractionations of
100‰ in
17
O and
18
O in their UV-photolysis experiments;
CO
2
-photolysis fractionation in the mesosphere could be larger,
but the wavelength dependence of these across the actinic
spectrum is difficult to estimate because the physical origin of the
laboratory fractionations is not understood (see
SI
). Remote-
sensing measurements by the Jet Propulsion Laboratory MkIV
Fourier-transform spectrometer, however, does not show hun-
dredfold enrichments in
18
O of mesospheric CO observed down-
welling to 30-km altitude in the polar vortex (G. C. Toon,
personal communication). On the basis of this observation,
mesospheric CO
2
does not appear to possess oxygen isotopic
enrichments of sufficient size to produce high
47
values in the
polar vortex upon mixing with stratospheric air masses, although
the long path length and consequent averaging over mesospheric
filaments and background stratospheric air of the MkIV instru-
ment will dilute any mesospheric signal. Additional remote
sensing of CO and/or CO
2
isotopologue abundances in the
mesosphere is thus needed to constrain this hypothesis.
Once in the stratosphere, mesospheric CO will be oxidized by
OH to produce CO
2
in the polar vortex. Oxidation of this CO
could, in principle, contribute to the large polar vortex values of
47
in CO
2
observed because: (
i
) CO mixing ratios as large as 10
ppmv have been observed (36) in mesospheric filaments in the
stratosphere (a thousandfold higher than background strato-
spheric CO abundances in the polar vortex), (
ii
) the known KIEs
for the CO
OH reaction (33, 34) are expected to favor forma-
tion of
13
C
18
O-containing CO
2
molecules in the stratosphere,
and (
iii
) the lifetime of CO with respect to oxidation is several
months and is therefore not immediately quantitative. Measure-
ments of the
13
C
18
O
OH vs.
12
C
18
O
OH KIE by Feilberg et al.
(34) at 298 K and 1 atm suggest that the oxidation of mesospheric
CO alone could account for elevated
47
stratospheric polar
vortex. Measurements of all of the relevant KIEs of CO
OH
reaction at stratospheric temperatures and pressures and a
Yeung et al.
PNAS
July 14, 2009
vol. 106
no. 28
11499
GEOPHYSICS
model of the oxidation of mesospheric CO in the stratospheric
polar vortex will be required to evaluate the impact of this
mechanism quantitatively.
Finally, we consider the potential role stratospheric particles
could play in the stratospheric CO
2
isotopic budget. Oxygen
isotope exchange on particle surfaces could drive the population
of CO
2
toward isotopic equilibrium by catalyzing CO
2
–CO
2
isotope exchange reactions (e.g.,
16
O
12
C
18
O
13
C
16
O
2
º
12
C
16
O
2
16
O
13
C
18
O) that are too slow to occur in the gas phase under
stratospheric conditions. Zero-point energy isotope effects dom-
inate in these reactions at equilibrium, driving the
47
value
toward its equilibrium value at a given temperature. We calculate
that
47
adopts a value of 1.6‰ if 70% of the CO
2
molecules
achieve isotopic equilibrium at temperatures coincident with
PSC formation,
190 K (39), which occurs primarily in the polar
vortex. Thus, in principle, particle-catalyzed equilibration of
CO
2
, perhaps via CO
2
hydration reactions in quasiliquid films at
the surface of ice particles (40), could produce the observed
polar vortex
47
values. Bulk isotopic compositions would be
little affected by these isotope exchange reactions, consistent
with the data presented in Table 1 and Fig. 2, if the catalyst
reservoir (e.g., liquid water layers on a surface) is much smaller
than the CO
2
reservoir; CO
2
would impart its bulk isotopic
composition on the catalyst, whereas the multiply-substituted
isotopologue distribution in CO
2
, which is insensitive to the
isotopic composition of the catalyst, approaches that at isotopic
equilibrium. However, CO
2
–liquid water isotope exchange may
not occur quickly enough at low temperatures and low pH to be
the relevant process. Experiments measuring the rate of CO
2
isotope exchange reactions at laboratory PSC– and sulfuric
acid–air interfaces should determine the plausibility of this
mechanism. Last, we note that, in order for this PSC-catalyzed
isotope-exchange mechanism to explain the polar vortex
47
measurements, the mechanism must result in a strong anticor-
relation with N
2
O mixing ratios (Fig. 2
A
); this would imply that
transport is fast compared with isotope exchange on the PSCs,
whose distributions are highly variable in space and time in the
polar vortex. Future 2D modeling efforts will examine whether
such a mechanism remains consistent with the observations or
whether transport and mixing of a mesospheric isotope signal
into the polar vortex better explains the observed anticorrelation
of
47
with N
2
O.
Conclusions
The signature of a new process preserved in stratospheric
16
O
13
C
18
O proportions reveals that a second mechanism, in
addition to the O(
1
D)
CO
2
isotope exchange reaction, alters the
isotopic composition of stratospheric CO
2
and thus the inter-
pretation of its chemical and transport ‘‘history.’’ This second
mechanism, which elevates
47
values in the polar vortex, is likely
of either mesospheric photochemical or heterogeneous origin.
To constrain further the role of each of these mechanisms,
isotopic fractionations due to broadband UV photolysis of O
2
and CO
2
, the KIEs of the CO
OH reaction under stratospheric
conditions, and the kinetics of CO
2
isotope equilibration on the
surfaces of PSCs and other stratospheric aerosols need to be
studied experimentally. Additionally, kinetic and photolysis-
induced isotope effects that may affect CO and CO
2
should be
incorporated into atmospheric models to quantify their contri-
butions to both the bulk and multiply-substituted isotopologue
budgets of stratospheric CO
2
, whose influence on the isotopo-
logue budgets of tropospheric CO
2
may be significant.
Materials and Methods
Air Sampling.
SOLVE mission samples (January–March 2000, 29–79°N, 11–20
km) were collected as whole air samples (WAS) (41). CO
2
was then isolated
from the
5-L STP of air by a combination of liquid N
2
and ethanol–dry ice
traps, then separated into aliquots of 12–18
mol of CO
2
each and sealed into
glassampoules.Balloonsampleswerecollectedbyusingacryogenicwhole-air
sampler (CWAS) (September 29, 2004; 34.5°N, 103.6°W, 27–33 km) (42, 43) and
purified and stored as above.
Isotopic Notation.
18
O
(R
18
sample
/R
18
reference
1)
1,000, where R
n
is the
abundance ratio of a rare isotope or isotopologue of mass
n
to its most
abundant analogue, e.g., R
18
[
18
O]/[
16
O].
17
O
17
O
0.516
18
O and
13
C and
18
O are reported relative to VPDB and VSMOW, respectively.
47
is
defined as the difference in ‰ between the measured R
47
of the sample
(principally
16
O
13
C
18
O/
12
C
16
O
2
but also, to a lesser extent,
17
O
12
C
18
O/
12
C
16
O
2
and
13
C
17
O
2
/
12
C
16
O
2
) and the R
47
expected for that sample if its stable carbon
and oxygen isotopes were randomly distributed among all CO
2
isotopologues
(20).
Isotopic Analysis.
13
C,
18
O, and
17
OofCO
2
were measured on a Finnigan
MAT 252 isotope ratio mass spectrometer (IRMS) at the University of Califor-
nia, Berkeley.
17
O was measured by using the CeO
2
technique with a preci-
sion of
0.5‰ (44). Aliquots of the same CO
2
samples were analyzed for
47
by using a Finnigan MAT 253 IRMS at the California Institute of Technology
configured to collect masses 44–49, inclusive, and standardized by compari-
son with CO
2
gases of known bulk isotopic composition that had been heated
for2hat
1,000 °C to achieve a stochastic isotopic distribution (17). Masses 48
and 49 were used to detect residual hydrocarbon contamination. For
47
analysis, 12- to 18-
mol aliquots of SOLVE mission and Balloon CO
2
samples
were purified of potential contaminants, such as hydrocarbons, by a pentane–
liquid N
2
slush (
120 °C) as well as by passing them through a gas chromato-
graphic (GC) column (Supleco Qplot, 530-
m i.d., 30-m length) at
20 °C, with
column baking at 150 °C between samples (19, 45). All data were corrected for
the presence of N
2
O by using the method described previously (19). Each
measurement consisted of 5–9 acquisitions (of 10 measurement cycles each),
with typical standard deviations (acquisition-to-acquisition) of 0.06‰ in
47
.
No evidence for organic contaminants was found when 1 representative CO
2
sample [20000131(30)1001] was tested (see
SI
).
Pulsed Photolysis Experiments.
18
O-labeled experiments were initiated with
pulsed excimer laser photolysis (193 nm, 50- to 100-mJ pulse
1
, 1-Hz repetition
rate,
200 pulses) of
1:100:3,000 static mixtures (100-Torr total pressure) of
97% N
2
18
O/CO
2
/He. For the 300 K experiments, a 15-cm-long stainless steel
conflat chamber with quartz windows was used as a reaction chamber.
Low-temperature experiments were performed in a 25-cm-long quartz cham-
ber. O(
1
D) was produced by excimer laser photolysis of N
2
18
O at 193 nm (50-
to 100-mJ pulse
1
). By using the absorption cross-section for N
2
O at 193 nm
(9
10
20
cm
2
) (46) and a O(
1
D) quantum yield of 1, the total fraction of the
CO
2
reservoir undergoing reaction was calculated to be
0.05%; we expect
that reaction cycling was negligible. N
2
18
O was synthesized from the acid-
catalyzed reduction of
18
O-labeled aqueous NaNO
2
(47), and its purity was
assessed by using mass spectrometry and Fourier-transform infrared spec-
trometry.
Starting samples of CO
2
contained a stochastic distribution of isotopo-
logues,generatedinthemannerdescribedintheisotopicanalysissection.The
CO
2
isotopologue distribution was measured, and CO
2
was recollected cryo-
genically for use in the reaction chamber. The reaction chambers were baked
overnight under vacuum before each experiment. Residence time in the
reaction chamber had a negligible effect on
47
in the 300 K experiments,
whereas the chamber cooling process was observed to enrich the starting
material in
18
O and
47
, by 0.5‰ and 0.13‰, respectively. These offset values
were subtracted from final
18
O and
47
values in 229 K experiments.
The reaction products were recollected cryogenically and purified in 2 GC
steps. Separation of residual N
2
18
O and reacted CO
2
was achieved through a
packed-column GC separation at 25 °C (PoraPak Q, 0.25 in o.d., 15-ft length,
30-mL min
1
He flow), after which the samples were purified with a capillary
GC as described above. Because of the presence of N
2
18
O isobars with CO
2
at
masses 46–49, mass 14 was also monitored to measure the extent of N
2
18
O
contamination. Differences in mass 14 between the starting and product
material of
5 mV (a variation typical of ‘‘clean’’ laboratory standards) were
deemed contaminated and the data points rejected.
The difference between the initial and final
18
O(
18
O) and
47
compo-
sitions was calculated assuming that
13
C was unchanged, because there was
no external carbon reservoir in the reaction. GC separation of N
2
18
O and CO
2
yielded a small, yet reproducible,
47
change of
0.13‰, which was present
both when gases of stochastic isotope composition were analyzed and when
aliquots of cylinder CO
2
working standard (
47
0.86‰) were analyzed. The
sourceofthisoffsetislikelyasmallamountoffractionationontheGCcolumn,
because it was relatively constant between experiments. Nevertheless, be-
cause of the apparent insensitivity of the
47
change to initial isotopic com-
11500
www.pnas.org
cgi
doi
10.1073
pnas.902930106
Yeung et al.
position, the sample purification step was treated as a small additive effect on
the measured
47
, with no effect on the overall kinetics measured.
A comprehensive, isotopologue-specific kinetics model of the O(
1
D)-CO
2
photochemical experiment (144 isotope exchange reactions) was constructed
in FACSIMILE. Statistical partitioning of isotope exchange products was used,
as suggested by Yung (2) were used: Reactant O(
1
D) atoms had a 2/3 proba-
bility of isotope exchange, regardless of the isotope exchange reaction. The
isotopic composition of the initial O(
1
D) was treated as a constant; for these
18
O-labeled experiments,
17
O and
16
O abundances were assumed to be neg-
ligible. Our determination of the
13
C/
12
C KIE from this model also included a
correction for the contribution of
17
O
12
C
18
Otothe
47
signal.
Box Model for Estimating the Integrated Effective Midlatitude O(
1
D) Isotope
Composition.
The midlatitude O(
1
D) isotope composition was calculated by
fitting the results of our laboratory photochemistry model to the slope of
linear regressions (weighted by standard deviations) of the midlatitude
47
data. A quenching (Reaction
3a
) and nonquenching (Reaction
3b
) branching
ratio of 9:1 in the isotope exchange reaction was used. Atmospheric
47
enrichments (e.g.,
47
1.1‰) were treated as
16
O
13
C
18
O enrichments exclu-
sively, because contributions to
47
from
17
O
12
C
18
O and
13
C
17
O
2
are
3%.
Uncertainties were estimated by varying
17
O and
18
O to match the 2
uncertainty in the slope of the weighted linear regressions. The upper and
lower uncertainty limits of the reported
17
O and
18
O values are correlated;
they correspond to fits that simultaneously match the upper and lower
uncertainty limits of the midlatitude
47
vs.
17
O and
47
vs.
18
O weighted
linear regressions.
Troposphere–Stratosphere–Mesosphere Mixing Model.
The model contained 4
adjustable parameters to fit each high-latitude datum individually: mixing
fractions and
18
O values of stratospheric air and mesospheric air. Tropo-
spheric isotopic compositions were fixed at typical clean surface troposphere
values (
13
C
trop
8‰,
18
O
trop
41‰,
17
O
trop
21‰, and
47trop
0.92‰).
We assumed oxygen isotope exchange was the dominant nonvortex process
affecting CO
2
isotopologue distributions in the stratosphere, so the initial
stratospheric mixing end-member composition was
13
C
strat
8‰,
17
O
strat
80.6‰,
18
O
strat
98.0‰, and
47strat
0. Because of the uncertainty in our
calculated
17
O
strat
and
18
O
strat
values, however,
18
O
strat
was allowed to vary
freely about our estimate above, and the modeled
18
O
strat
values generally
fellwithinthatrange(see
SI
).ThemesosphericCO
2
mixingendmemberhadan
isotopic composition constrained by the calculations of Liang et al., namely
17
O
meso
0.3
18
O
meso
. The multiply-substituted isotope distribution was
fixed at
47meso
0 for this simple mixing-only scenario because our kinetics
experiments and the midlatitude stratosphere
47
values indicate that isotope
exchange drives the isotopologue distribution toward a stochastic one. No
measurements of mesospheric
13
C exist, so
13
C
meso
8‰ was used; how-
ever, the calculated enrichment in mesospheric
18
O was insensitive to the
choice of
13
C.
ACKNOWLEDGMENTS.
We thank R. Lueb for WAS/CWAS field support and
Y. L. Yung, G. A. Blake, and P. O. Wennberg for manuscript comments. This
workwassupportedbytheDavidowFund(CaliforniaInstituteofTechnology),
the National Science Foundation, the National Aeronautics and Space Admin-
istration Upper Atmosphere Research Program, and the Camille Dreyfus
Teacher–Scholar Award (to K.A.B.).
1. Yakir D (2003) The stable isotopic composition of atmospheric CO
2
.
Treatise on
Geochemistry
, eds Turekian KK, Holland HD (Elsevier, Amsterdam), Vol 4, pp 175–212.
2. Yung YL, Lee AYT, Irion FW, DeMore WB, Wen J (1997) Carbon dioxide in the
atmosphere: Isotopic exchange with ozone and its use as a tracer in the middle
atmosphere.
J Geophys Res
102:10857–10866.
3. Ciais P, Meijer HAJ (1998) The
18
O/
16
O isotope ratio of atmospheric CO
2
and its role in
global carbon cycle research.
Stable Isotopes: Integration of Biological, Ecological and
Geochemical Processes
, ed Griffiths H (BIOS Scientific, Oxford), pp 409–431.
4. Hoag KJ, Still CJ, Fung IY, Boering KA (2005) Triple oxygen isotope composition of
tropospheric carbon dioxide as a tracer of terrestrial gross carbon fluxes.
Geophys Res
Lett
32:L02802.
5. Luz B, Barkan E, Bender ML, Thiemens MH, Boering KA (1999) Triple-isotope compo-
sition of atmospheric oxygen as a tracer of biosphere productivity.
Nature
400:540–
550.
6. Shaheen R, Janssen C, Ro
̈ckmann T (2007) Investigations of the photochemical isotope
equilibrium between O
2
,CO
2
, and O
3
.
Atmos Chem Phys
7:495–509.
7. Thiemens MH, Jackson T, Zipf EC, Erdman PW, van Egmond C (1995) Carbon dioxide
and oxygen isotope anomalies in the mesosphere and stratosphere.
Science
270:969–
972.
8. La
̈mmerzahl P, Ro
̈ckmann T, Brenninkmeijer CAM, Krankowsky D, Mauersberger K
(2002) Oxygen isotope composition of stratospheric carbon dioxide.
Geophys Res Lett
29:1582.
9. Bhattacharya SK, Savarino J, Thiemens MH (2000) A new class of oxygen isotopic
fractionation in photodissociation of carbon dioxide: Potential implications for atmo-
spheres of Mars and Earth.
Geophys Res Lett
27:1459–1462.
10. Boering KA, et al. (2004) Observations of the anomalous oxygen isotopic composition
of carbon dioxide in the lower stratosphere and the flux of the anomaly to the
troposphere.
Geophys Res Lett
31:L03109.
11. Kawagucci S, et al. (2008) Long-term observation of mass-independent oxygen isotope
anomaly in stratospheric CO
2
.
Atmos Chem Phys
8:6189–6197.
12. Liang M-C, Blake GA, Yung YL (2008) Seasonal cycle of C
16
O
16
O, C
16
O
17
O, and C
16
O
18
O
in the middle atmosphere: Implications for mesospheric dynamics and biogeochemical
sources and sinks of CO
2
.
J Geophys Res Atmos
113:D12305.
13. Wen J, Thiemens MH (1993) Multi-isotope study of the O(
1
D)
CO
2
exchange and
stratospheric consequences.
J Geophys Res
98:12801–12808.
14. Johnston JC, Ro
̈ckmann T, Brenninkmeijer CAM (2000) CO
2
O(
1
D) isotopic exchange:
Laboratory and modeling studies.
J Geophys Res
105:15213–15229.
15. Chakraborty S, Bhattacharya SK (2003) Experimental investigation of oxygen isotope
exchange between CO
2
and O(
1
D) and its relevance to the stratosphere.
J Geophys Res
108:4724–4738.
16. Baulch DL, Breckenridge WH (1966) Isotopic exchange of O(
1
D) with carbon dioxide.
Trans Faraday Soc
62:2768–2773.
17. Eiler JM, Schauble E (2004)
18
O
13
C
16
O in the Earth’s atmosphere.
Geochim Cosmochim
Acta
68:4767–4777.
18. Affek HP, Xu X, Eiler JM (2007) Seasonal and diurnal variations of
18
O
13
C
16
O in air:
Initial observations from Pasadena, CA.
Geochim Cosmochim Acta
71:5033–5043.
19. Affek HP, Eiler JM (2006) Abundance of mass-47 CO
2
in urban air, car exhaust and
human breath.
Geochim Cosmochim Acta
70:1–12.
20. Eiler JM (2007) ‘‘Clumped-isotope’’ geochemistry—The study of naturally-occurring,
multiply-substituted isotopologues.
Earth Planet Sci Lett
262:309–327.
21. NewmanPA,etal.(2002)AnoverviewoftheSOLVE/THESEO2000campaign.
JGeophys
Res Atmos
107:8259.
22. Greenblatt JB, et al. (2002) Defining the polar vortex edge from an N
2
O:potential
temperature correlation.
J Geophys Res Atmos
107:8268.
23. Alexander B, Vollmer MK, Jackson T, Weiss RF, Thiemens MH (2001) Stratospheric CO
2
isotopic anomalies and SF
6
and CFC tracer concentrations in the Arctic polar vortex.
Geophys Res Lett
28:4103–4106.
24. Liang M-C, Blake GA, Lewis BR, Yung YL (2007) Oxygen isotopic composition of carbon
dioxide in the middle atmosphere.
Proc Natl Acad Sci USA
104:21–25.
25. Perri MJ, Wyngarden ALV, Lin JJ, Lee YT, Boering KA (2004) Energy dependence of
oxygen isotope exchange and quenching in the O(
1
D)
CO
2
reaction: A crossed
molecular beam study.
J Phys Chem A
108:7995–8001.
26. Takahashi K, Taniguchi N, Sato Y, Matsumi Y (2002) Nonthermal steady state transla-
tional energy distributions of O(
1
D) in the stratosphere.
J Geophys Res
107:4290–4296.
27. Mebel AM, Hayashi M, Kislov VV, Lin SH (2004) Theoretical study of oxygen isotope
exchange and quenching in the O(
1
D)
CO
2
reaction.
J Phys Chem A
108:7983–7994.
28. Krankowsky D, et al. (2007) Stratospheric ozone isotope fractionations derived from
collected samples.
J Geophys Res Atmos
112:D08301.
29. Michalski G, Bhattacharya SK (2009) The role of symmetry in the mass independent
isotope effect in ozone.
Proc Natl Acad Sci USA
106:5493–5496.
30. Dylewski SM, Geiser JD, Houston PL (2001) The energy distribution, angular distribu-
tion, and alignment of the O(
1
D
2
) fragment from the photodissociation of ozone
between 235 and 305 nm.
J Chem Phys
115:7460–7473.
31. Brenninkmeijer CAM, et al. (2003) Isotope effects in the chemistry of atmospheric trace
compounds.
Chem Rev
103:5125–5161.
32. Feilberg KL, D’Anna B, Johnson MS, Nielsen CJ (2005) Relative tropospheric photolysis
rates of HCHO, H
13
CHO, HCH
18
O, and DCDO measured at the European photoreactor
facility.
J Phys Chem A
109:8314–8319.
33. Ro
̈ckmann T, et al. (1998) Mass-independent oxygen isotope fractionation in atmo-
spheric CO as a result of the reaction CO
OH.
Science
281:544–546.
34. Feilberg KL, Johnson MS, Nielsen CJ (2005) Relative rates of reaction of
13
C
16
O,
12
C
18
O,
12
C
17
O and
13
C
18
O with OH and OD radicals.
Phys Chem Chem Phys
7:2318–2323.
35. Feilberg KL, Johnson MS, Nielsen CJ (2004) Relative reaction rates of HCHO, HCDO,
DCDO, H
13
CHO, and HCH
18
O with OH, Cl, Br, and NO
3
radicals.
J Phys Chem A
108:7393–7398.
36. Plumb RA, et al. (2003) Global tracer modeling during SOLVE: High-latitude descent
and mixing.
J Geophys Res Atmos
108:8309.
37. Ishidoya S, Sugawara S, Morimoto S, Aoki S, Nakazawa T (2008) Gravitational separa-
tion of major atmospheric components of nitrogen and oxygen in the stratosphere.
Geophys Res Lett
35:L03811.
38. Craig H, Horibe Y, Sowers T (1988) Gravitational separation of gases and isotopes in
polar ice caps.
Science
242:1675–1678.
39. Wang Z, Schauble EA, Eiler JM (2004) Equilibrium thermodynamics of multiply substi-
tuted isotopologues of molecular gases.
Geochim Cosmochim Acta
68:4779–4797.
40. Dash JG, Rempel AW, Wettlaufer JS (2006) The physics of premelted ice and its
geophysical consequences.
Rev Mod Phys
78:695–741.
41. Flocke F, et al. (1999) An examination of chemistry and transport processes in the
tropical lower stratosphere using observations of long-lived and short-lived com-
pounds obtained during STRAT and POLARIS.
J Geophys Res Atmos
104:26625–26642.
42. Froidevaux L, et al. (2006) Early validation analyses of atmospheric profiles from EOS
MLS on the Aura satellite.
IEEE Trans Geosci Remote Sens
44:1106–1121.
43. Lueb RA, Ehhalt DH, Heidt LE (1975) Balloon-borne low temperature air sampler.
Rev
Sci Instrum
46:702–705.
44. Assonov SS, Brenninkmeijer CAM (2001) A new method to determine the
17
O isotopic
abundance in CO
2
using oxygen isotope exchange with a solid oxide.
Rapid Commun
Mass Spectrom
15:2426–2437.
45. Ghosh P, et al. (2006)
13
C–
18
O bonds in carbonate materials: A new kind of paleother-
mometer.
Geochim Cosmochim Acta
70(6):1439–1456.
46. Sander SP, et al. (2006)
Chemical Kinetics and Photochemical Data for Use in Atmo-
spheric Studies: Evaluation Number 15, JPL Publication 06
-
2
(Jet Propulsion Labora-
tory, Pasadena, CA).
47. McIlvin MR, Altabet MA (2005) Chemical conversion of nitrate and nitrite to nitrous
oxidefornitrogenandoxygenisotopicanalysisinfreshwaterandseawater.
AnalChem
77:5589–5595.
Yeung et al.
PNAS
July 14, 2009
vol. 106
no. 28
11501
GEOPHYSICS