JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 106 NO.

D2, PAGES 1733-1739, JANUARY 27, 2001

A study of the reaction between NaHCO3 and H:

Apparent closureon the chemistry of rnesosphericNa
Rachel M. Cox, Daniel E. Self, and JohnM. C. Plane
Schoolof EnvironmentalSciences,Universityof EastAnglia,Norwich,UnitedKingdom

Abstract. The reactionbetweensodiumbicarbonate(NaHCO3)andatomicH is the only

likely route(apartpossiblyfromdaytimephotolysis)for recyclingthismajorsodium
reservoirbackto atomicNa in theuppermesosphere.
The ratecoefficientwasdeterminedfor
the firsttime by reactingNaHCO3 in a knownexcessof H in a fastflow tube,yieldingk(227

K) = (1.7+ 0.6)x 10-3andk(307K) = (5.2+ 2.5)x 10-3cm3molecule

-s-.These
measurements
arevery well describedby transitionstatetheory,usingab initio quantum
calculations
of the saddlepointson thereactionpotentialenergysurface.The resulting

temverature
dependence
oftheratecoefficient
isthengiven
byk(100-400
K) = 1.84x 10-3
7'78
exp-(1014+13_96
/ T)cm3molecule
' s-.Whenk(T)isextrapolated
to 180-220
K, thereis
excellentagreementwith thevaluepredictedfrom atmospheric
modelsof thenighttimeNa
layer.Hencethismeasurement
has,to a first order,demonstrated
closureof the neutral
chemistrygovemingthe undersideof theNa layerabove80 km.
Ager and Howard [1987] were not able to observethis

1. Introduction

reactionin the laboratory;from their paperone can estimate

A longstanding
challengefor explainingthe appearance
of
the atomic Na layer in the upper mcsospherchas been to
identify the molecular reservoirsand sinks for the metal
below 90 km [e.g., Plane, 1991]. Since only Na atomsand

anupper
limittotheratecoefficient
ofk4(300
K)< 2 x 10-2

spcctromctry,respectively),the mcsosphcricchemistryof
sodiumhas had to bc unraveledthrough laboratorykinetic
studies of elementary reactions, coupled with modeling.
Significantprogresshasbeenmadeoverthe pastdecade,with
models that arc now able to reproducethe characteristic
featuresof the Na layer remarkablywell, over a range of
latitudesand seasons
[Helmerand Plane, 1993;McNeil et al.,
1995;Plane et al., 1998, 1999].
Thesemodelspredictthatbelow the peak of the Na layer at
about 90 km, the major sodium reservoir is sodium
bicarbonate (NaHCO3), formed through the following
sequence
of reactions:

above 80 km.

cm3molecule
- s-. Indeed,sodium
models[e.g.,Planeet al.,
1999] requirereaction(4) to be slow at mesospheric
temperatures,
but withinlimits,so thatNaHCO3providesa
Na+ ionscanbc observed
(by lidarandrocket-borne
mass temporaryreservoirratherthan a permanentsink for Na

Na + 03 -- NaO + 02

(1)

NaO + H20 (or H2) --> NaOH + OH (or H)

(2)

NaOH + CO2+ (M) --> NaHCO3.

(3)

Althoughother reservoirspeciessuch as NaO2, NaO3, and

NaCO3 also form, they are radical specieswhich are broken
down quite rapidly by the large concentrations
of atomic O
above 80 km [Ager and Howard, 1986; Helmer and Plane,
1993]. In contrast,NaHCO3 is a very stable closed-shell
molecule:quantumcalculationsindicatethat the only route
backto Na (apartfrom daytimephotolysis)is by reactionwith
atomicH [Rajasekharand Plane, 1993]:
NaHCO3 + H --> Na + H2CO3 (or H20 + CO2).

(4)

Otherwise, there would be negligible

concentrationsof atomic Na below 90 km, in conflict with

observation;furthermore,the steepundersideof the Na layer

indicatesthat a fairly rapidrecyclingof Na mustoccur[Plane
et al., 1998].
Hence a measurementof k4providesa rather stringenttest
of the current understanding of mesospheric sodium
chemistry.In this paperwe reportthe first suchmeasurement.
It is worth pointingout that this reactionis difficult to study,
for severalreasons.First,thereis at presentno directmethod
for monitoringNaHCO3 in the gas phase,particularlyat the
low concentrations
that must be employedto avoid forming
NaHCO3 polymers. Second,there is no direct sourceof
NaHCO3 in the gasphase. The solid dissociates
on heating,
and sothe NaHCO3 vapormustbe producedfrom a sequence
of gasphase
reactions,ratherasin the mesosphere.Third, the
reaction is relatively slow, and atomic H is difficult to
producequantitativelyat very high concentrations.

2. Experimental technique
Reaction (4) was studied by reacting NaHCO3 under
pseudo-first-order
conditionsin a known excessof atomicH,
using the stainlesssteel flow tube illustrated in Figure 1.
NaHCO3 was producedin situ via a sequenceof reactions.
Atomic Na vaporwas first entrainedin a flow of N2 from the
heatpipe oven and enteredthe flow tube (internaldiameterof
36.9 mm) at a point 600 mm upstreamof the spectroscopic

Copyright2001by theAmericanGeophysical
Union.

detection cell. The Na flow mixed with the main N2 carder

Papernumber2000JD900579.

flow and was then convertedto NaHCO3 by the sequential

additionof N20 (analogousto O3 in reaction(1)), H2, and
CO2. Atomic H, producedby the microwavedischargeof H2,

0148-0227/01/2000JD900579509.00
1733

1734

COX ET AL.: NaHCO3 + H REACTION, CLOSURE ON MESOSPHERIC Na

I Sodium
I

Lamp
] Pump

Beam
Chopper
41 __l

N2 flow

Sodium
Oven II

Oat

oppr

Valve !iii)._ii'+
ll: insulation

fi_.

li Movable
Injector

i ]

N Bath NO H
Gas inlet

CO, CO.
NO

Temp

Pressure,
NO

QuadrupoleMass

Spectrometer

Figure 1. Schematicdrawing(approximatelyto scale)of the fast flow tube employedto studythe reaction
betweenNaHCO3 andH (PMT, photomultipliertube).

coolingthe flow tubewalls aroundthe reactionzonewith dry

ice. The temperatureinsidethe Na ovenwas monitoredby a
permanentlyinserted thermocoupleand kept constantto
within +1 K over the course of an experiment by a
temperaturecontroller. Spheresof sodiummetalwere placed
insidea spiralof stainlesssteelmesh(gauge150) in the oven.
As the Na melted and wettedthe mesh,this greatlyincreased
The downstream end of the tube was connected to a
the available surface area of liquid Na and thereby the
commercialstainlesssteel six-way cross,which formed the efficiencyof the oven as a Na vapor source.The inlet from
spectroscopicdetection point.
The relative Na atom the oven into the flow tube was kept approximately40 K
concentrationwas monitored by phase sensitivedetection hotter than the oven to prevent blocking caused by
with a lock-in amplifier of the modulated resonance condensing
Na vapor.
fluorescencesignalproducedby choppingthe outputfrom a
N2 carder gas (99.9992% pure) was purified by passing
sodiumdischargelamp (Figure 1). Fluorescencewas almost througha trap containingmolecularsieveat 77 K. NO (99%)
exclusively
measured
fromtheD lineat589.6nm[Na(32p1/2was purified by passingthroughmolecularsieveat 153 K to
-- 3281/2
)], afterpassing
through
a narrowband
interferenceremove any traces of NO2. NO2 (99.5%) was purified by
filter (full width at half maximumof 2.0 /!t). This repeated distillation from 196 to 77 K before use. N20
arrangement
provided
a detection
limitforNa of 5 x 106 (99.99%), H2 (99.9999%), CO2 (99.995%), and CO (99.9%)
cm-3,estimated
froma separate
experiment
where
theabsolutewere usedwithoutfurtherpurification. Sodiummetalspheres
atomic Na concentration was measured by resonance were cleaned with solvent before insertion into the oven.
absorptionof the radiationfrom a Na hollow cathodelamp
directedtransversely
to the flow alongthe opticalaxis at the 3. Results
detectionpoint. When studyingreaction(4), measurements
Becausereaction (4) was expectedto be slow, it was
wereperformed
witha totalNaconcentration
of (0.5-7)x 108
moleculescm-3in the lowerhalf of the flow tube.
important to prevent any Na produced by the reaction
In order to measure[H] by titration with NO2, a second reformingNaHCO3. This was achievedby adding a large
photomultipliertube was also coupledto the six-way cross; excessof CO just above the H atom injection point. CO
chemiluminescence from HNO was recorded by a reactsvery rapidly with NaO [Kolb etal., 1992]:
picoammeterafter passingthrougha cut-onfilter at . > 645
NaO + CO - Na + CO2,
(5)
nm. Downstreamof the crosspiece was a quadrupolemass
spectrometer
which sampledalongthe cylindricalaxis of the
therebypreventingformationof NaOH by reaction(2).
tube througha 100-gm orifice in the center of a skimmer
The determinationof k4requiredan absolutemeasurement
cone. The cardergasflow thenexitedthrougha throttlevalve
of [HI. This was achievedby addinga knownexcessof NO2
to a booster/rotarypump combination,providinga volume
(by a factorof 5-10) throughthe inlet indicatedin Figure 1.
displacement
rateof 110L s-1.
The fractionof NO2 removedby stoichiometricreactionwith
The main carder gas and reagent gas flows were set by
H asit exitedthe injector,
calibratedelectronicmassflow controllers. The temperature
NO2 + H --> NO + HO,
(6)
and total pressurein the reaction zone upstreamof the
spectroscopiccell were measuredwith a chromel-alumel
thermocoupleand a capacitancemanometer,at the points was determinedby monitoringthe NO2 signalat mass46 in
shown in Figure 1. Reaction (4) was studiednear room the massspectrometer.
Althoughcompleteremovalof H was
temperature(307 K) and at a lower temperature(227 K) by achieved within the first few centimeters downstream of the
was then introducedfurther downstreamthrough a sliding
Pyrexinjector.Note thatthis injectorwaspositionedalongthe
wall of the tube, rather than along the cylindricalaxis, in
order to minimize the loss of sodiumspeciesby deposition.
The openingat the tip of the injectorwas designedto direct
the atomic H to the center of the flow tube, ensuringmore
rapidmixing of the reactants.

COX ET AL.: NaHCO3 + H REACTION, CLOSURE ON MESOSPHERIC Na

N20, I-12
injected

+0 , .............

reactiontime, a plot of In (1 - [Na]./[Na]0) versus[H] should

be linear, with a slopeof-k4 trcac,
where treatis the reaction
time calculatedfor a paraboliclaminar flow profile [Huggins
and Cahn, 1967]. Figure3 showsfour examplesof suchplots
under quite differentreactiontimes and rates of wall loss of
H. Althoughtheseplotsare reasonablylinear,in practice,k4
was determinedby using a numericalmodel which included

CO2injoeted
H (+CO)injoeted

[ ...... I..................

lo+9

1735

Na

the reaction rate coefficients

and diffusion

coefficients

listed

in Table 1; k4 was varied to minimize the goodness-of-fit

parameter
;2,thesumofthesquared
differences
between
the

10+$

0.000

model and the experimentaldata. Examplesof suchfits are

shownin Figure 3. Note that the model fit is very good in
Figure 3a, whereasthe data sets in Figures 3b, 3c and 3d
exhibit small systematicdeviationsfrom their respectivebest
0.005

0.010

0.015

Reaction time

0.020

/ s

model fits. However, these variations were not consistent

between data sets. Out of 26 experiments, 10 varied as in

Figures3b and 3c, 9 varied in the oppositesenseas in Figure

Figure2. Modelsimulation
of the variationof sodium 3d, and8 showedno systematicdeviationasin Figure3a.

specieswith time (or distance)alongthe fast flow tube used

to measurek4(NaHCO3+ H). Theprofileof atomicNa before
reagents are added is shown as Nao, which decays by
diffusion to the walls. NaHCO3 is then producedby the
sequentialaddition of N20, H2, and CO2 to the Na flow.
Finally, atomicH is injected,producingatomicNa (shownas
Nail) by reaction(4). CO is also injectedhere to prevent
NaO reactingwith H2 and CO2to form NaHCO3.

The final values of k4 at each temperaturewere then

obtained from the weighted mean of the model fits to 25
individual datasets. The values were weighted by the
propagatederror arisingfrom the variablewall lossof atomic

injector,a kineticmodelwas alsoemployedto make minor

correctionsfor the subsequentsecondaryreactionswhich
recycleH:

k4(307
K) = (5.2+ 2.5)x 10-3cm3molecule
-s- andk4(227
K) = (1.7+ 0.6)x10'13cm3 molecule
- s-,wherethequoted

H, the calibration of the absolute H atom concentration,

uncertainty
in thereaction
timetrcac,
andthe;2valuesfrom
the model fit. The uncertaintyin the H atom concentration
was a significant componentof the propagatederror. On
average,this was 23%, althoughit ranged from 7 to 58%,
dependingon experimentalconditions.The final result is

errors are single standard deviations about the weighted

means.

OH + OH --> H20 + O
OH + O

--> 02

+ H

OH + H2 --> H20 + H.

(7)
(8)

4. Theoretical

(9)

In orderto understandthe mechanismof reaction(4), and

to providea frameworkfor extrapolatingk4overthe full range
of uppermcsosphcrictemperatures,wc now presenta set of
ab initio quantumcalculationscombinedwith transitionstate
theory (TST). First, the very accurateCompleteBasis Set
(CBS-Q) methodwas employedfrom within the Gaussian98
suiteof programs[Frischet al., 1998] to estimatethe reaction
enthalpiesfor the followingthreereactionpathways:

It was then necessaryto determine the loss of H from

recombination

on the walls of the flow tube This was found

to vary from day to day, presumablybecause of the

accumulation of metallic debris on the walls, and so was

determinedindividually for each experimentalrun. The

relative [H] at the spectroscopiccell was determinedby
addinga smallflow of NO (seeFigure1) andmonitoringthe
resultingchemiluminescence
from the reaction

NO + H --> HNO* --> HNO + hv

(10)

asthe injectordistancewas varied.

Reaction (4) was then studied by varying both the
concentration
of H producedby the microwavedischargeand
the injection distancedown the tube. At each injection
distancethe total sodium concentration,[Na]0, was first
obtainedwith no reactantspresent. The reagentsN20, H2,
andCO2were then addedto makeNaHCO3. Figure2 shows
a computersimulationof the sodiumchemistrythat occurred

Calculations

NaHCO3
+ H -->Na+ H20+ CO2 AH0
= -27.9kJmol' (4')
-->NaCO3
+ H2
/kH0
= +3.8kJmol- (4")

--> Na+ H2CO3

AH0
=+4.1kJmol'. (4"')

The geometryof the transitionstate for each pathway was

thendetermined
by usinghybriddensityfunctional/HartreeFock B3LYP theorywith the high-level6-31l+G(2d,p) basis
set (which has both polarizationand diffusefunctionsadded
to the atoms). Figure4 illustratesthesegeometriesand their

respective
energybarriers(includingzero-pointenergies
).
This showsthat the pathwayto produceNa + H2CO3,
downthelengthof thetube. Whena flow of atomicH (and althoughessentiallythermoneutral
within the expected
CO) was switchedon, a small sodiumsignal ([Na].) was uncertainty
of +8 kJ mol' at the CBS-Qlevelof theory

recorded. The ratio [Na]./[Na]0 was then measuredas a [Foresman and Frisch, 1996], has the lowest barrier.
functionof [H] to produceonedataset;thewholeprocesswas Furthermore,the expectederrorusingB3LYP/6-31l+G(2d,p)
then repeatedat a different injector distance(or reaction theory
is +13kJmol- [Foresman
andFrisch,1996],sothat
time).
the energybarriersof channelsin Figures4a and4b aremuch
An approximateanalytical solutionof the kinetics due to too high within this uncertaintyfor these channelsto be
gasphasereactionsand wall lossesshows that for a fixed significantlyopen at the low temperatures
consideredhere

1736

COX ET AL.. NaHCO3 + H REACTION, CLOSURE ON MESOSPHERIC Na

0.00

0.00

Model

Experiment
'

-0.01

-0.01

-0.02

'

-0.03

,-

-0.02
!

-0.03

b,

a,
-0.04

-0.04

le+13

2e+13

3e+13

le+13

4e+13

3e+13

[H] atomcm

[H] atomcm

0.00

0.00

2e+13

-3

-3

-0.01

--

-0.02

-0.03

.. -0.01
.Z, -0.02

-0.04

'

-0.03 d,

-0.05

le+13

2e+13

3e+13

[H]atom
cm
-3

5e+12

le+13

[H]atom
cm
-3

Figure3. Plotsof modelfitsto foursetsof experimental

datawithdifferingconditions
of atomicH wall loss

(kHloss)
andreaction
time(t)' (a)kmoss
= 598s-1,t = 0.018s; (b)kmos,
= 497s-,t = 0.011s; (c) kmo,,
= 391
-1

s , t = 0.014s; (d)kmo,s
= 598s-1,t = 0.013s. Errorbarsareshown
onlyin Figure3aforthesakeof clarity.
(notethatif the Figure4b channelwere open,thenthiswould
not be observedin our experimentunlessNaCO3reactedwith
H to regenerateNa, rather than recombiningto yield
NaHCO3). TST was then appliedto the Figure4c channelby
using the molecularparameterslisted in Table 2, together
with a correctionfor tunneling of H throughthe reaction
barher,assumed
to be a symmetricalEckartbarher[Steinfeld
et al., 1989].

5. Discussion

Figure5 is an Arrheniusplot for reaction(4), illustrating

the measurementsof k4 at 227 and 307 K from the present

study. They are comparedwith estimatesof the rate

coefficientderived from fitting a one-dimensional
sodium
model to lidar observationsof the nighttime Na layer at
midlatitudesover an annualcycle [Plane et al., 1999] and at

Table 1. Rate CoefficientsUsedto Model the Flow Tube Chemistry

Reaction

RateCoefficienta

Source

Na + N20 --> NaO + N2

2.8x 1040exp
(-1590/T)

NaO

2.0 x 10-l

NaO + H2 --> Na + H20

1.1x 10-9exp(-1100/T)
1.1x 10-9exp(-1400/T)

1
1

NaO + CO --> Na + CO2

2.0 x 1040

NaO + CO2 + M --> NaCO3 + M

1.3x 10-27
(200/r)
1.9x 10-28
(200/r)
1.9x 10-3exp(-840/T)
1.26x 10-exp(-7600/T)
152x (300/T)
4'8s
141x (300/T)
4.8s
134x (300/T)4.8s
120x (300/T)
4'8s

1
1
3
4
5
5
6
7

+ H --> Na+OH

NaO + H2 --> NaOH + H

NaOH + CO2 + M--> NaHCO3 + M

H+CO+M

--> HCO+M

H + N20 --> OH + N2
Diffusion

of Na

Diffusion

of NaO

Diffusion

of NaOH

Diffusion of NaHCO3

aUnits:bimolecularreactions,cm3 molecule
' s'l; termolecular
reactions,cm

molecule
-2s4;diffusion
coefficients,
torrcm2s4.

bSources:
1, PlaneandHelrner[1994]; 2, Kolbet al. [1992]; 3, Warnatz
[1984]; 4, Hanson and Salirnian [1984]; 5, Measuredusing the techniqueof
Helmer and Plane [1993]; 6, measuredby using the photolysisof NaOH as a
spectroscopic
marker; 7, calculatedfrom gas kinetictheory [Helrnerand Plane,
1993] by usinga theoreticaldipolemomentfor NaHCO3 of 6.9 debye[Rajasekhar
and Plane, 1993].

COX ET AL.. NaHCO3 + H REACTION, CLOSURE ON MESOSPHERIC Na

(a)

1737

Temperature/ K

^E#o- 84.1ld mo1-1

400
le-12

30O

200

.'

150

Tranlition
State
Theor

' \" M..odel

[Plane
etal.,
1998]'

le-13

(b)

Experiment

le-14

-_...... /

. Mod.
el.[7la.
neeta,.,
19,99],...".'"",%.
0.003

AE#0-30.6kJmo1-1
(c)

0.004

0.005

0.006

0.007

Temperature-1
/ K-1
Figure 5. Arrhenius plot for the reaction NaHCO3 + H,
illustratingthe experimentalpoints from the presentstudyat
227 and 307 K, transition state theory fitted to the
experimentalpoints (solid line) with upper and lower limits
(dashedlines), and model predictionsof the rate coefficient

fromthe seasonal
behaviorof thesodiumlayerat 40N[Plane
et al., 1999] and during Januaryat the SouthPole [Plane et
al., 1998].

non-Arrheniuscurvatureat the lowesttemperatures,

mostlyas
Figure 4. Transitionstatesfor the reactionbetweenNaHCO3 a result of tunneling through the barrier. A simple 3and H to yield (a) Na + H20 + CO2, (b) NaCO3 + H2, and (c) parameterfit to the TST resultis givenby
Na + H2CO3.
The arrow indicates the reactant H;
K)= 1.84x 10-11r '78
exp-(1014+3_96/T
)
forming/breakingbondsare shownas solid lines. The saddle k4(100-400
point energies(including zero-point energy) with respectto
cm3molecule
-s-l, (11)
the reactantsare calculated at the B3LYP/6-311+G(2d,p)
level of theory.
where the uncertaintiesin the activation energy yield the
upper and lower boundsto TST that are consistentwith the
experimental uncertainties. These bounds are shown as
the SouthPole duringmidsummer[Plane et al., 1998]. The dashedlines in Figure 5.
modelpredictionof the undersideof the Na layer around85
Equation(11) can now be usedto extrapolatek4 from the
km is particularly sensitive to k4 and its temperature experimentaltemperaturerangeto the lower temperatures
of
dependence,and so thesemodelingstudiesprovide estimates the upper mesosphere. As is shown in Figure 5, there is
of k4 at temperaturesof 180-220 K and about 140 K, excellentagreementwith the atmospheric
modelpredictionof
respectively.
k4around200 K (Figure 5). By contrast,the modelprediction
Also shownin Figure 5 is the very satisfactoryfit of TST of k4 at 140 K is higher than the bounds of the TST
through the experimentaldatapoints,which was achieved extrapolation,but this can now be explainedfollowing some
simply by reducingthe theoreticalbarrier height of 14.3 kJ recentmeasurements
in our laboratoryon the photochemistry
mol- to 10.1kJmol- wellwithintheexpected
errorat this of NaHCO3. Although the mesosphereover the South Pole
level of theory (see above).Note that TST predictsmodest during January is continuouslysunlit, the photolysisof

Table 2. Calculated
MolecularParameters
at theB3LYP/6-31l+G(2d,p)Levelof Theory
Molecule

Rotational
Constants,Vibrational
Frequencies,
cm4
GHz

NaHCO3

12.29, 3.34, 2.63

H2CO3

12.00, 11.27, 5.81

NaHCO3 + H -->

11.18, 2.84, 2.27

Na + H2CO3
transition state

112,248,333,570,581,700,833,1008,1237,
1389,1659,3802
522,548,595,602,800,975,1156,1298,
1817,3800,3802
387i, 85,106,249,329,557,593,679,822,
1018,1232,1383,1638,3789

1459,
910,

1738

COX ET AL.' NaHCO3 + H REACTION, CLOSURE ON MESOSPHERIC Na

i

110
I

Na layerat 40N (annualaverage)

105

Lidar measurement

100
,.
..

<

Model (bestfit)

95

Model
(lower
limit
to
k4)

90
85

Model(upperlimittok4)
8O

1000

2000

3000

4000

5000

Na concentration/cm
'3
Figure6. Verticalprofilesof thenighttime
annualaverage
Na layerat 40N: lidarmeasurements
at UrbanaChampaign(Illinois)andFort Collins(Colorado)[Planeet al., 1999](thicksolidline) compared
with model
predictionsusingthe temperature-dependent
expression
for k4(thin solidline) and upper/lowerlimits to k4
(dashedlines)from equation(11).

NaHCO3 was not consideredin the original modelingstudy References

[Plane et al., 1998] becauseof the apparentstabilityof the
molecule.

Now,

however, we

have used a

laser

photofragmentation
technique[Rajasekharet al., 1989] to
measurethe NaHCO3 photolysiscrosssectionto yield Na.
The onsetof photolysisoccursat wavelengthsbelow 240 nm,
providinga significantalternativepathwayto reaction(4) for
converting NaHCO3 to Na during daylight.
The

photodissociation
coefficient
(JNaCO3
1 x 104S'z) is large

Ager, J. W., andC. J. Howard,The kineticsof NaO + O2 andNaO +

CO2 + M and their role in atmosphericsodium chemistry,
Geophys.Res.Lett., 13, 1395-1398, 1986.
Ager, J. W., and C. J. Howard, Rate coefficientfor the gas phase
reactionof NaOH with CO2, J. Geophys.Res., 92, 6675-6678,
1987.

Foresman,J. B., and A. Frisch,ExploringChemistrywith Electronic

StructureMethods,Gaussian,PittsburghPa., 1996.
Frisch,M. J., et al., Gaussian98, revisionA.7, Gaussian,Pittsburgh

Pa., 1998.
enoughto reducethe value of k4 at 140 K requiredin the
Hanson, R. K., and S. Salimian, Survey of rate constantsin the
model,thusremovingthe discrepancy
in Figure5.
N/H/O system, in CombustionChemistryedited by W. C.
Figure6 illustratesthe sensitivityof a sodiummodel[Plane
GardinerJr., Springer-Verlag,
New York, 1984.
et al., 1999] to the uncertaintyin k4. The modelwas usedto Helmer,M., andJ. M. C. Plane,A studyof the reactionNaO2+ O -

calculate
themonthlynighttimeNa layerat 40N,fromwhich

NaO + 02: Implicationsfor the chemistryof sodiumin the upper

atmosphere,
J. Geophys.Res.,98, 23,207-23,222, 1993.
Huggins, R. W., and J. H. Kahn, Metastablemeasurements
in
flowingheliumafterglow,J. Appl.Phys.,38, 180-188,1967.
C. E., D. R. Worsnop,M. S. Zahniser,G. N. Robinson,X. Shi,
10'14cm3molecule
- s-l),aswellastherecommended
lower Kolb,
and D. R. Herschbach,Chemicalkineticsand dynamicsof the
andupperlimitsto k4(meanvalues
of 2.7 x 10-24and8.8x
mesospheric
sodiumnightglow,in Gas-PhaseMetal Reactions,
10'14 cm3 molecule
- s'z, respectively).Thesemodel editedby A. Fontijn,ElsevierSci.,New York, 1992.
model for
predictionsare comparedin Figure6 with lidar measurements McNeil, W. J., E. Murad, and S. T. Lai, Comprehensive
the atmosphericsodium layer, J. Geophys.Res, 100, 16,847made at Urbana-Champaign(Illinois) and Fort Collins

the annual mean profile was derived. Mean profiles were

computedby usingthe temperature-dependent
expressionfor
k4 from equation(11) (annualmean value at 85 km of 4.6 x

16,855, 1995

(Colorado),bothat closeto 40N[Planeet al., 1999]. This Plane, J. M. C., The chemistryof meteoriticmetalsin the upper

shows that although the peak height and undersidescale

height of the layer are quite sensitiveto k4,within the limits
assignedto k4in equation(11) thereis satisfactory
agreement
betweenthe modelpredictionsandthe observations.
In conclusion,the presentstudydemonstrates
that reaction
(4) will indeedrecycleNa from the reservoirspeciesNaHCO3
at a rate consistentwith the requirementsof atmospheric
models. Hence this measurement has, to a first order,
demonstrated
closureon the neutralchemistrygoverningthe
undersideof the Na layer.

atmosphere,
Int. Rev.Phys.Chern.,1O,55-106, 1991.
Plane,J. M. C., and M. Helmer, Laboratorystudiesof the chemistry
of meteoricmetals, Researchin ChemicalKinetics, vol. 2, edited
by R. G. Comptonand G. Hancock,ElsevierSci., New York,
1994.

Plane,J. M. C., R. M. Cox, J. Qian, M. Pfenninger,G. C. Papen,C.

S. Gardner,and P. Espy, The mesospheric
Na layer at extreme
highlatitudesin summer,J. Geophys.Res.,103, 6381-6390, 1998.
Plane, J. M. C., C. S. Gardner,I. Yu, C. Y. She, R. R. Garcia, and H.
C. Pumphrey,The mesospheric
Na layer at 40N:Modelingand
observations,
J. Geophys.Res.,104, 3773-3788, 1999.
Rajasekhar,
B., andJ. M. C. Plane,An ab initiostudyof dissociative
electronattachmentto NaHCO3 and NaCO3, and the role of these

Acknowledgments.This work, includinga researchstudentship

to D.E.S., was supportedby grants GST/02/1242 and GR3/11753
from the Natural Environment Research Council.

reactionsin formationof suddensodiumlayers, Geophys.Res.

Lett., 20, 21-24, 1993.

Rajasekhar,B., J. M. C. Plane,and L. Bartolotti. Determinationof

COX ET AL.: NaHCO3 + H REACTION, CLOSURE ON MESOSPHERIC Na

theabsolute
photolysis
cross-section
of sodiumsuperoxide
at 230
K: Evidencefor the formationof sodiumtetroxidein the gasphase,or.Phys.Chem., 93, 7399-7404, 1989.
Steinfeld,J. I., I. S. Francisco,and W. L. Hase, ChemicalKinetics

R. M.

1739

Cox, J. M. C. Plane, and D. E. Self, School of

EnvironmentalSciences,University of East Anglia, Norwich NR4

7TJ, United Kingdom. (r.m.coxuea.ac.uk;j.planeuea.ac.uk;
d.e.selfuea.ac.uk)

andDynamics,Prentice-Hall,EnglewoodClifs, N.J., 1989.

Warnatz,I., Rate coefficientsin the C/H/O system,in Combustion
Chemistryeditedby W. C. GardinerJr., Springer-Verlag,
New (ReceivedJune29, 2000; revisedAugust24, 2000;
York, 1984.
acceptedSeptember1, 2000.)