Second Harmonic Generation Studies of Ozone Depletion Reactions on Ice Surfaces under Stratospheric Conditions

Article abstract of DOI:10.1021/jp991559s

Hypochlorous acid, HOCl, an important species in the proposed heterogeneous mechanism for stratospheric ozone depletion, has been observed directly at submonolayer amounts on a single crystalline basal ice surface at 155-188 K, using the nonlinear optical method second harmonic generation. The ice is held in equilibrium with its vapor pressure. Second harmonic generation signals from 290 to 310 nm spectroscopically characterize the species and enable us to follow isothermal desorption kinetics in situ. HOCl desorbs as a single species with a ??G*_des = 48 ?± 4 kJ/mol, close to the cohesive energy of ice itself. The lifetime of HOCl on the clean ice surface at 185 K is estimated to be 4 s and the equilibrium surface coverage at 10^-11 Torr HOCl to be around 4 ?? 10¹¹ molecules/cm², corresponding to about 0.1% of a monolayer. However, these same measurements performed on ice predosed with varying amounts of HNO₃ show that the HOCl lifetime is lengthened by coadsorbed HNO₃, depending on the HNO₃ surface density.

Full text of DOI:10.1021/jp991559s

J. Phys. Chem. B 1999, 103, 8205-8215
8205
FEATURE ARTICLE
Second Harmonic Generation Studies of Ozone Depletion Reactions on Ice Surfaces under
Stratospheric Conditions
Franz M. Geiger, Anthony C. Tridico, and Janice M. Hicks*
Department of Chemistry, Georgetown UniVersity, Washington, DC 20057
ReceiVed: May 11, 1999; In Final Form: July 26, 1999
Hypochlorous acid, HOCl, an important species in the proposed heterogeneous mechanism for stratospheric
ozone depletion, has been observed directly at submonolayer amounts on a single crystalline basal ice surface
at 155-188 K, using the nonlinear optical method second harmonic generation. The ice is held in equilibrium
with its vapor pressure. Second harmonic generation signals from 290 to 310 nm spectroscopically characterize
the species and enable us to follow isothermal desorption kinetics in situ. HOCl desorbs as a single species
with a ∆G/des ) 48 ( 4 kJ/mol, close to the cohesive energy of ice itself. The lifetime of HOCl on the clean
ice surface at 185 K is estimated to be 4 s and the equilibrium surface coverage at 10-11 Torr HOCl to be
around 4 × 1011 molecules/cm2, corresponding to about 0.1% of a monolayer. However, these same
measurements performed on ice predosed with varying amounts of HNO3 show that the HOCl lifetime is
lengthened by coadsorbed HNO3, depending on the HNO3 surface density.
1. Introduction
(ClONO2) and HCl, can be “activated” by ice surfaces,
producing hypochlorous acid (HOCl) or Cl2.7,8,24
The depletion of stratospheric ozone over the Antarctic
continent has been the subject of intense field and laboratory
work in the past decade.1-6 There is agreement that the
phenomenon is due to chlorine reactions that are somehow
catalyzed by the solid particles in the polar stratospheric clouds
(PSCs).3,7,8 While an exact compositional and structural analysis
of the particles is not yet possible, field and laboratory work
suggests that the clouds are composed, at least in part, of ∼10
µm crystals of water ice (Type II PSCs) that form at temper-
atures below 188 K.9-14 Another category includes nitric acid
hydrate (NAH) particles (Type I PSCs) that form at ∼195 K.
As a result of their small size and because they grow slowly
and under near equilibrium conditions, PSC particles are thought
to be single crystals, with low index faces exposed.10,12 PSC
particle lifetimes are on the order of minutes15 to hours.11,16
While the particles are not in complete equilibrium with the
gas phase,17 probably local equilibrium is nearly achieved most
of the time. From the vapor pressure of ice at 185 K (10-4
Torr),18 it can be estimated that the equilibrium evaporation and
condensation rates of clean ice are approximately 100 mono-
layers of water per second.19 However, the presence of foreign
molecules may alter these rates.15,20
ClONO2 (g) + H2O (s) f HOCl (ad) + HNO3 (ad) (1)
HOCl (g) + HCl (ad) f Cl2 (g) + H2O (s)
(2)
ClONO2 (g) + HCl (ad) f Cl2 (g) + HNO3 (ad) (3)
Both HOCl and Cl2 can be photolyzed with the advent of the
polar spring to form Cl radicals which catalytically destroy
ozone by well-known gas-phase reactions.1 Reactions 1-3 play
a key role in the processes that ultimately lead to the annual
formation of the ozone hole, but not much is known about the
reaction mechanisms, their time scales, and the unique role of
ice catalyzing the reactions. In this article, we focus on the
behavior of HOCl adsorbed on the ice surface in an effort to
begin to elucidate the mechanisms and unravel the kinetics of
reactions 1-3.
Reactions 1-3 occur at 25 km altitude where, at the end of
the Antarctic winter, stratospheric temperatures typically fall
to 185-200 K, with ambient pressures around 40 Torr. The
chemical composition of the polar stratosphere determined
during spaceborne and airborne missions is reported in terms
of volume mixing ratios (VMRs) which correspond to partial
molar volumes. Measured H2O mixing ratios are typically in
the parts per million by volume (ppmv) range. HNO3 mixing
ratios are found to be around 10 ppbv,25,26 and ClONO2 and
HCl mixing ratios are around 1 ppbv.25,27,28 The partial pressures
of the trace gases ClONO2, HCl, and HNO3 are thus between
10-8 to 10-7 Torr. HOCl has not yet been measured directly in
the polar stratosphere during a PSC event, and its concentration
has to be inferred from model calculations that are based on
It has been known since the mid 1980s that the reaction rate
constants for the homogeneous gas-phase reactions involving
ClONO2, HCl, and H2O are exceedingly slow21,22 such that they
could be neglected in the computer modeling codes predicting
rates of stratospheric ozone depletion.23 However, it is now
known that “safe” forms of chlorine, namely chlorine nitrate
* To whom correspondence should be addressed. E-mail: hicksj@
gusun.georgetown.edu. Fax: 202-687-6209.
10.1021/jp991559s CCC: $18.00 © 1999 American Chemical Society
Published on Web 09/15/1999

8206 J. Phys. Chem. B, Vol. 103, No. 39, 1999
Geiger et al.
peratures of at least 90 K. From the simulation, the authors
conclude that the ice surface has a full bilayer termination.
Enhanced vibrational amplitudes at the surface of ice were
confirmed experimentally by He scattering experiments.35
Buch, Devlin, and Rowland applied transmission infrared
spectroscopy to study thin films of amorphous D2O ice and
nanometer sized clusters of crystalline D2O ice at 10-120 K
in high vacuum and combined the method with Monte Carlo
computer calculations.36 Surface molecules that were 2-fold and
3-fold coordinated (i.e., dangling OH(D) groups) were identified
by two narrow IR bands assigned to nonhydrogen bonded OH-
(D) groups in microporous amorphous ice. Only one band at
2725 cm-1 was found in the case of crystalline ice, in agreement
with a relatively ordered surface structure. This one band was
also observed by Schaff and Roberts, who used single-reflection
FTIR spectroscopy applied to thin D2O ice films.32 Devlin et
al. and Schaff and Roberts thus determined that crystalline
samples have a low density of free or “dangling” OH groups,
corresponding to a mostly nonpolar surface, whereas amorphous
samples were found to have a polar surface resulting from highly
abundant dangling OH groups.
Related to surface order is the phenomenon of surface melting
or the existence of a quasi-liquid layer, which has become a
controversial issue in explaining the physical and chemical
processes that are involved in the heterogeneous reactions
leading to ozone depletion. A solid-liquid, liquid-gas system
of interfaces is expected to behave differently with adsorbing
species than a solid-gas interface. Fletcher suggested that the
existence of a quasi-liquid layer on ice surfaces, first proposed
by Faraday,37 could reduce the free energy of ice by allowing
for disorder at the surface, with gradually increasing order within
the surface layer.38 Important initial experiments by Fletcher
and Adamson suggested the existence of a quasi-liquid layer
on ice down to temperatures of about 240 K,39 well above
stratospherically relevant temperatures of ∼190 K. Several other
experimental40 and theoretical41-45 approaches suggest that a
quasi-liquid layer does not exist at stratospherically relevant
temperatures. Using glancing angle X-ray scattering techniques,
Lied et al.46 showed that a quasi-liquid layer is nonexistent in
both the basal and the prism faces of ice Ih at 190 K. The surface
melting of ice has been shown to be extremely sensitive to the
presence of salt impurities.47
Figure 1. (a) Crystal structure of ice Ih (adapted from ref 38). Oxygen
atoms are shown, hydrogen atoms are omitted for clarity. (b) Top view
of the basal plane of ice Ih displaying the 3m symmetry of the surface.
Open and closed circles represent oxygen atoms in the upper and lower
bilayer, respectively. Hydrogen atoms are omitted for clarity.
measured chlorine and nitrogen concentrations.29 Typical partial
pressures for HOCl determined in this way are around 10-11 to
10-10 Torr.29
We study the heterogeneous reactions on ice from a surface
science point of view. Under stratospheric conditions, ice is
hexagonal Ih and is uniaxial (see Figure 1a). Perpendicular to
the c-axis lays the very flat basal face or (0001) surface of ice
Ih. This surface has 3m symmetry (Figure 1b), and consists of
a bilayer with water molecules spaced by 0.95 Å in the c-axis
direction. Parallel to the c-axis lays the prism face or (1100)
surface of ice Ih, a rougher surface. The perfectly terminated
ice surface structures serve as models of ice surfaces; under
various temperature and pressure conditions, the actual surface
structures of ice, including possible reconstructions, are not
known.
Ice surfaces are difficult to study due to their high vapor
pressures. Conventional ultrahigh vacuum (UHV) surface sci-
ence methods have been utilized to study deposited water
films19,30,31 and adsorption of foreign molecules on them.32-34
These studies are generally limited to total pressures and
temperatures lower than those in the stratosphere, however, they
have proven to be useful in elucidating structures and binding
motifs on amorphous and polycrystalline films. For instance,
Materer et al. measured six-fold LEED patterns from a thin ice
film crystallized on a clean Pt (111) surface at 90 K,30 assigned
to a (0001) surface. Supporting molecular dynamics simulations
showed that large vibrational amplitudes exist down to tem-
Reactions 1-3 on vapor-deposited ice have been studied
using two related techniques. In flow reactor measurements,
reactants are delivered through a tube of ice by a carrier gas
(the total pressure including the carrier gas is usually a few Torr)
and reactions on ice are inferred by studying the effluent
gases.7,48,49 These studies have the advantage of being able to
screen many reactions and measure their efficiencies. Efficien-
cies are usually expressed in terms of a “reactive uptake
coefficient”, or γ, which is the fraction of reactant molecules
impinging on the surface that results in product molecules. The
kinetic analysis can be complicated, however,50 and the ice is
difficult to characterize.6,48,51,52 In Knudsen cell experiments, a
chamber containing ice is exposed to a known amount of gas,
then pumped out and analyzed for products.8,53 This method is
also useful in screening for reactions and measuring reactive
uptake coefficients. The pressures and temperatures can be
stratospheric, and kinetic measurements can be enhanced by
using a pulsed valve to deliver the reactants.53 As in the flow
tube experiments, the ice is difficult to characterize.
The interaction of HOCl with ice has been studied using both
flow reactor and Knudsen cell techniques. Hanson and Ravis-
hankara observed HOCl adsorption on ice at 191K and, using

Feature Article
J. Phys. Chem. B, Vol. 103, No. 39, 1999 8207
a partial HOCl pressure of around 10-7 Torr, found a fractional
surface coverage of 0.01 or less.48 They measured an adsorption
enthalpy ∆Hads for HOCl on ice held at 191 K of 58 ( 8 kJ/
mol, but it is important to note that the HOCl had been formed
via the ClONO2 hydrolysis on vapor deposited ice. Hence, the
adsorption energy includes the thermodynamics and kinetics of
the ClONO2 adsorption and hydrolysis process, which are not
known, and the other product of the reaction, HNO3, is also
present on the ice. Pulsed valve experiments by Oppliger et al.
found as well that HOCl interacts only weakly with ice at T >
173 K and pressures of 10-6 Torr; only up to 5% of a monolayer
harmonic generation (SHG).79-81 In SHG, light at a fundamental
laser frequency, ω, is frequency doubled to produce light at
2ω. Due to symmetry constraints, SHG is forbidden in cen-
trosymmetric media within the electric dipole approximation.
At the surface, however, symmetry is broken and the second-
order process is allowed,82 thus making the method surface
selective without contribution from the bulk phases. Surface
SHG has been used to study a variety of surfaces, for example,
metals, semiconductors, liquids, and glasses.83-86
In the present application, a single crystal of ice mimicks
PSC particles and allows us to bypass problems associated with
partitioning of adsorbates to grain boundaries, which can affect
the observed kinetics and reactivity. Another feature of the work
is that the ice surface is maintained in equilibrium with water
vapor. The laser method also allows us to probe the structure
of the clean ice surface suggesting that it is indeed crystalline.
SHG allows the spectroscopic identification of adsorbed species
in situ, a key point in this study. Previous work focused on the
adsorption of a nonreactive polar probe molecule, p-tolualdehyde
(TA), on the ice surface.81 At low coverages of approximately
one-tenth of a monolayer, TA was shown to adsorb as a tightly
bound species, probably acting as a hydrogen bond acceptor of
dangling OH groups on the ice surface. SHG from the adsorbed
TA/ice surface is consistent with a 3m symmetry, epitaxial with
the underlying crystal lattice. This species begins to desorb at
185 K and higher. A second species appears to be physisorbed
isotropically and begins to desorb at 170 K and higher.
The present study is intended to address some of the basic
molecular level questions regarding the adsorption of small
molecules on ice surfaces. In particular, the stratospherically
relevant molecule HOCl is probed. Is the adsorbed species
molecular? Are there one or more types of adsorption sites?
What is the lifetime of HOCl at the surface? How does HOCl
desorb? What is the effect of coadsorbed HNO3? We will also
address the reaction of HOCl with HCl on ice surfaces. Our
study of reaction 1 using SHG is reported elsewhere.87
is found to form.53 Using HOCl partial pressures in the 10-5
-
10-7 Torr range, Abbatt and Molina showed that at 200 K, HOCl
adsorption on ice and desorption from ice are reversible.54 Heats
of adsorption ∆Hads were reported to be around 44 ( 8 kJ/mol.
Banham et al. also studied the interaction of HOCl with ice.34
IR spectral features associated with HOCl/H2O mixtures at 140
K decrease with the addition of HCl (presumably by reaction
2) and disappear with heating to 165 K. Popov et al. performed
a thermodynamic analysis of flow tube data by Abbatt et al.55
and Hanson and Ravishankara.48 Using an arbitrarily chosen
desorption preexponential factor of υo ) 1013 s-1, they obtained
a desorption barrier E/des of 50 kJ/mol.57
Theoretical calculations have been performed on HOCl
(density functional theory)58,59 and on HOCl/water complexes
(ab initio theory).60,61 The chlorine is thought to have a very
slight positive (+0.13) partial charge,59 whereas the hydrogen’s
partial charge is +0.32. As a weak acid in aqueous solution
(pKa ) 7.54), HOCl is expected to interact as a hydrogen donor
with the oxygen of water; this was observed in ab initio
calculations of the HOCl‚H2O complex60 and HOCl interacting
with four water molecules mimicking an adsorption site on ice.61
This latter work yielded a bonding energy of 37 kJ/mol.
Molecular dynamics calculations performed on HOCl interacting
with the basal surface of ice yielded a bonding energy of -60
kJ/mol; however, the optimized structure yielded HOCl with
its hydrogen oriented away from the surface.43
Reactions 1-3 on a variety of surfaces (ice,7,8,53,54,56,62,63 nitric
acid/water mixtures,48,54,64-69 and sulfuric acid/water
mixtures65,66,70-73) have been proven to occur using both flow
reactor7,48,54,56,63,65-68,70-72,74,75 and Knudsen cell methods.8,53,62,69
Reactions (1-3) occur readily on water ice surfaces, while on
solid nitric acid/water mixtures, their efficiencies depend on the
humidity.64,76
Some authors have suggested that reaction 2, HOCl with HCl
yielding Cl2 and H2O at the ice surface, occurs through an ionic
mechanism.5,49,77 HOCl may dissociatively adsorb to H+ and
OCl- or HO- and Cl+ at the surface. There is no direct
experimental evidence to support this hypothesis. Chu et al.
examined the reaction of HOCl with HCl on water vapor
deposited ice films held at 188 K.49 HCl and HOCl partial
pressures of 6 × 10-8 to 2 × 10-6 Torr were used, and a
“reactive uptake coefficient” of HOCl on an HCl covered ice
film was determined to be 0.13 ( 0.08.
Some authors have suggested that reaction 3 actually occurs
in two steps (reactions 1 and 2 sequentially), a mechanism in
which HOCl is retained on the ice surface under stratospheric
conditions.7,24,48 Other groups have observed that reaction 1
occurs with a noticeable delay,48,53,78whereas reaction 3 occurs
rapidly.53 This implies that reaction 3 can proceed without HOCl
as an intermediate.
2. The SHG Approach
There are several distinct advantages to employing second-
order nonlinear optical techniques in the study of ice surfaces.
First, the methods derive their surface sensitivities from sym-
metry and do not require UHV, allowing one to study the
dynamic ice/water vapor interface. The methods can be used
to follow processes ranging from ultrahigh vacuum to even
higher than atmospheric pressure. Indeed, we have observed
SHG from the ice/liquid water interface.88 Second, the methods
can be used to probe the surface in a direct manner in real time
and in situ. Third, the methods are nondestructive.
By using various polarizations of the incident light and
measuring specific polarizations of the SHG signal, it is possible
to gain information regarding the symmetry of the surface.84 In
general, the intensity of the reflected SHG is
32π3ω2
2
2
I2ω
)
sec2 θ|bꢀ 2ω‚6ø(2):bꢀωbꢀω| Iω
(4)
c3
where θ is the angle of incidence, bꢀω and bꢀ2ω are the Fresnel-
corrected polarization vectors for the fundamental and second
harmonic light fields at the surface, respectively, 6ø(2) is the
macroscopic second-order nonlinear susceptibility of the me-
dium, and Iω is the intensity of the incoming light.84,89 Surface-
generated SHG signals are typically weak (10 to 100 photons
per second in this study) but are collected above a zero
background.
We have developed a method in which a single crystal of
clean ice held in equilibrium with water vapor is interrogated
directly using the nonlinear optical laser method surface second

8208 J. Phys. Chem. B, Vol. 103, No. 39, 1999
Geiger et al.
SHG originates from the second-order nonlinear susceptibility
of the surface 6ø(2), which is a third rank tensor. Different tensor
elements of 6ø(2) contribute to the polarization-analyzed SHG
output for an anisotropic surface with 3m symmetry compared
to that for an achiral isotropic surface.84,91 For the 3m surface,
ø4 ) øxxx ) -øxyy is nonzero and independent, while it is zero
for an isotropic surface. (We follow the convention that the
surface is the xy plane, z is the perpendicular axis, and the
∑
second-order polarization Pi(2)
)
øijkEjEk). For example, for
jk
input light polarized in the surface plane (s), the reflected
s-polarized SHG generated from the 3m surface is given by
I2sω ) A|sin φ[ø4 sin2 φ - 3ø4 cos2 φ]|
(5)
2
where φ is the rotation angle about z. Rotational anisotropies
in SHG predicted by eq 5 have been observed, for example,
for Si(111).90 For the isotropic surface
I2sω ) 0
(6)
Additionally, for input light polarized perpendicular to the
surface (p), s-polarized SHG signals are not allowed for isotropic
achiral surfaces, yet can result from the component øxyy
characteristic of 3m surfaces. These distinctions are important
in work on the basal ice surface, which is 3m symmetric.30,35
The macroscopic surface response 6ø(2) can be related to the
molecular level by the model 6ø(2) ) Ns R6(2) , where the
microscopic second-order nonlinear polarizability R6(2) is aver-
aged over all orientations of the molecules. Ns is the number
density of molecules at the surface. The square root of the SHG
signal is thus proportional to Ns if the orientation of the
molecules is constant with surface coverage.
Through the use of tunable lasers, second-order nonlinear
optical methods can be utilized as a surface electronic spec-
troscopy.92 The perturbation expansion for the second-order
nonlinear polarizability R6(2) for SHG contains terms related to
electronic transitions in the molecules and is usually described
by electric dipole transition terms such as82
Figure 2. Arrangement for probing the basal ice surface using SHG:
LN2 ) liquid nitrogen Dewar; H ) heater; I ) ice; RGA ) residual
gas analyzer.
filters, and its polarization is analyzed using a quartz Rochon
polarizer. The SHG is further separated from the fundamental
frequency by a monochromator and detected by single photon
counting. The signal is normalized to the laser intensity by a
reference SHG signal generated in transmission through potas-
sium dihydrogen phosphate (KDP) powder in an index matched
fluid. An identical detection system is used to collect the
reference signal, correcting for the wavelength response of the
detection system.
All experiments are carried out in a cylindrical Pyrex vacuum
chamber with a base pressure of 1.0 × 10-8 Torr (Figure 2).
The chamber is pumped by a 170 L/s turbo pump (Balzers TPU
170). The partial pressures of gases present during an experiment
are measured with a residual gas analyzer (RGA, AMETEK,
Dycor MA200M). Ice samples are prepared by growth from
the melt using a modification of the Bridgeman apparatus.79,80
At 273 K, the basal face grows faster than the prism face.38
This anisotropy in the growth rate can be reversed by applying
a temperature gradient during ice growth since the thermal
conductivity of ice is about 5% greater along the c axis than
along the a axis.93 Thereby, it is possible to grow ice single
crystals along the c axis94 with the basal (0001) plane exposed.
After cleaving the ice using a razor blade, the ice sample (1 cm
diameter, 3 mm thick) is adhered to a cold polished aluminum
support. At present, it is positioned at an arbitrary rotation angle
φ with respect to the three mirror planes. The support is in
thermal contact with a liquid nitrogen Dewar and a ceramic
heater (Spectra-Mat E-292) maintaining the sample at temper-
atures ranging from 110 to 270 K. A chromel-alumel thermo-
couple is embedded in the ice for accurate temperature
measurements. The sample is annealed for 30 min at 190 K.
Using a variable control leak valve vacuum sealed to a water
reservoir (Fisher, HPLC grade), the water partial pressure in
the chamber is adjusted such that the ice sample is kept in
equilibrium (i.e., the evaporation rate equals the condensation
rate; the water vapor pressure values were obtained from Marti
and Mauersberger18 and Bryson et al.)95 HOCl could not be
leaked into the vacuum chamber using a stainless steel leak valve
because it reacts with the exposed metal surfaces. Instead, a
Pyrex capillary leak valve consisting of a 10 cm capillary with
4π2e3
a|bµi|b ‚ b|bµj|c ‚ c|bµk|a
(2)
ijk
R6 ) -
+
∑
h2
(ω - ωba + iΓba)(2ω - ωca + iΓca)
b,c
. . . (7)
where Γ represents damping coefficients, bµi is the electric dipole
transition moment, a, b, and c represent the ground, intermediate,
and final state respectively, and the summation is over all excited
states. As ω and/or 2ω, the frequency of the input light or the
second harmonic frequency, approach a natural resonance
frequency of the molecule, R6(2) and thus the SHG efficiency
increases. Surface SHG electronic spectra of molecular mono-
layers as well as surface states in metals not found in the bulk
have been reported.81,84,91
3. Experimental Section
The laser system consists of a picosecond Nd3+-YAG laser
(Spectra Physics 3800S) and a 4 MHz cavity-dumped tunable
dye laser operating with Rhodamine-6G (tunable from 580 to
620 nm) at 100 mW average power. Dye laser pulses are 900
fs fwhm and have ∼20 nJ energy. The beam is focused on the
ice sample to approximately 30 µm. SHG is collected in a
reflection geometry (see Figure 2); the quartz window passing
the SHG is not shown in this diagram. The SHG is then
separated from the fundamental frequency by two Corning 7-54

Feature Article
J. Phys. Chem. B, Vol. 103, No. 39, 1999 8209
TABLE 1: Polarization-Analyzed SHG Signals from Clean
Ice and from HNO3 and HOCl on Iced
a 50 µm orifice, a Teflon stopcock, and a 10 mL Pyrex sample
bulb was used.
s-polarized signal
With a 30 µm focused laser spot, one can find optically flat
areas of the crystal (Rq < λ/10), yielding a reflection that is
round and symmetric with no speckle visible to the eye in a
darkened room. The optical flatness and also the fact that the
SHG signals are well polarized (typical extinction ratio of 9:1)
suggests that the surface is not significantly rough. Further, the
observed SHG intensities for 45 input polarization, s output
polarization are the same as that observed from the liquid water
surface. This suggests but does not confirm that surface
roughness does not contribute to the SHG signal. Willen and
Dash observed unusually large planar facets of ice with the basal
face exposed, and used a molecular level argument that, because
it is difficult to nucleate in 2D on a very flat surface, the basal
ice surface grows preferentially and efficiently. If it were rough
with steps and defects, addition of water molecules on the basal
plane would enlarge the prism plane (and this is not observed.)45
sample
input polarization
(photons/s)
background (dark cts)
clean ice 298 nma
0.2 ( 0.2
2.8 ( 0.2
2.8 ( 0.2
5.4 ( 0.3
2.9 ( 0.8
13.4 ( 0.4
4.9 ( 0.2
14.9 ( 0.4
29 ( 1
s
p
45
s
s
p
45
s
p
45
HNO3 on ice 298 nmb
HOCl on ice 298 nmc
HOCl on ice 289 nm
46 ( 1
85 ( 1
a The SHG response from ice is constant with frequency in the range
290 to 310 nm.80 b The ice was exposed to 1 L HNO3. c HOCl was
prepared in situ by exposure of ice to 1 L of ClONO2. d Wavelength
of the SHG is indicated. Temperature is 158 K for clean ice and HOCl
on ice and 163 K for HNO3 on ice.
Note that when a nanosecond pulsed laser was utilized in
the experiment instead of the picosecond system, we observed
SHG signals only at the onset of surface damage, as evidenced
by an increase in water partial pressure in the chamber with
laser irradiation. The nanosecond damage threshold occurred
at fundamental input wavelength 1064 nm at about 100 mJ/
cm2. This damage probably resulted from heating due to the
very small absorption coefficient of water due to an overtone
near 600 nm.93 With the picosecond laser, the energy density
is <10 mJ/cm2, and no surface damage is observed as evidenced
by mass spectrometry.
chamber (on the order of 1 × 10-8 Torr) is determined by
monitoring mass 63.
Exposure of a surface to a gas is conveniently measured in
langmuir (L), where one L corresponds to exposure of the
surface to 10-6 Torr of gas for 1 s, i.e., 1 L ) 10-6 Torr s.103
Exposures employed in this study are usually below 1 L and as
high as 2.4 L. If all molecules that hit the surface at 1 L exposure
stick, the surface will have approximately one monolayer
coverage. Because the sticking probabilities are often not known,
the data presented here are given as exposure (L) rather than
surface coverages.
HOCl is difficult to prepare without contaminants. To attempt
to rule out perturbations by contaminants in the gas phase,
several methods of preparing HOCl were used. In preparation
1, HOCl is prepared as an aqueous solution by mixing NaOCl
(Aldrich Chemical Co.) with MgSO4‚7H2O (Aldrich Chemical
Co.).54,96 The published syntheses list Cl2, HCl, and Cl2O as
possible impurities.54,96 Gaseous HOCl from the vapor over the
solution is admitted into the chamber through the capillary leak
valve from the HOCl solution reservoir. At a solution temper-
ature of 0 °C, the HOCl partial pressure was 6 × 10-9 Torr.
4. Results and Discussion
4.1. Polarization-Resolved SHG From Clean Ice and
HNO3 Dosed Ice. The structure of the clean basal ice surface
at 158 K was interrogated using s-in, s-out polarization analyzed
SHG signals. As shown in eqs 5-6, this type of signal is not
allowed for an isotropic surface; however, it is permitted for
surfaces of 3m symmetry, such as the bilayer-terminated basal
ice surface.30,35 (Other surface symmetries (1, m, 3) allow this
polarization combination,84 but are unlikely for this system).
Table 1 shows the results obtained for the ice samples used in
this study. It is clear that s-in, s-out SHG signals are significant,
thus the surface is anisotropic. P-in, s-out SHG signals are also
significant. Control surfaces that are achiral and isotropic, such
as fused quartz and liquids, never show the s-in s-out or p-in
s-out polarized SHG signals. The in-plane anisotropy signal
consistent with 3m symmetry on the clean basal ice surface at
163 K does not disappear when the ice surface is exposed to 1
L HNO3. The results in Table 1 suggest that the basal ice surface
is 3m symmetric and crystalline and that the symmetry of the
ice surface structure is retained upon adsorption of HNO3.
When HNO3 is dosed directly onto ice, no change in the SHG
intensity is observed within our signal-to-noise levels. This is
consistent with calculations of predicted SHG signals from eq
4, including resonance (HNO3 has no resonance near ω or 2ω
in this experiment) and Fresnel factors.84 These show that, if
anything, the SHG signal from nitric acid coated ice should
decrease with respect to the signal from clean ice because of
the higher refractive indices of nitric acid hydrates104 with
respect to ice.
In preparation 2, chlorine gas is bubbled through an aqueous
solution of silver nitrate, forming ClONO2 which reacts with
water to form HOCl and HNO3. HOCl yields with this method
are 10 times higher than with the first procedure but still contain
HCl contamination.
Third, HOCl is prepared by hydrolyzing ClONO2 vapor in a
40% sulfuric acid solution.97 With this preparation, HCl, HNO3,
Cl2, and H2SO4 impurities are present as well. ClONO2 can also
be hydrolyzed in situ by admitting ClONO2 to the chamber with
an ice sample, and allowing reaction 1 to proceed.78 ClONO2
is prepared by reacting Cl2O and N2O5 (CAUTION: Many of
these compounds are explosive and should be handled with great
care).98-100 N2O5 is prepared through the reaction of fuming
nitric acid, phosphorus pentoxide, and ozone. Cl2O is prepared
by mixing O2 and Cl2 over HgO. The gas-phase infrared
spectrum of ClONO2 at 298 K was obtained using a Nicolet
7000 series FTIR, and agreed with literature data showing HCl
and HNO3 impurities.101,102 When ClONO2 is exposed to ice
(∼2.4 L exposure), HOCl is formed78,87 and thermal desorption
data indicate that only HOCl, HNO3, and H2O desorb from this
ice sample.
Contributions to the s-in s-out SHG signal from the bulk of
the ice crystal are ruled out since the phases of the signal
sampling the in-plane symmetry of each bilayer are opposite in
sign (the layers making up the basal planes within bulk ice Ih
For experiments in which HNO3 is dosed onto the surface,
HNO3 is taken from the vapor phase over a 69% aqueous HNO3
solution (Baker Analyzed). The HNO3 partial pressure in the

8210 J. Phys. Chem. B, Vol. 103, No. 39, 1999
Geiger et al.
preparation 2 and cooling quickly to 155 K and (2) exposing
an ice sample at 155 K to 2.4 L ClONO2, heating to 180 K,
and then cooling quickly to 155 K. It can be seen that the SHG
spectra resulting from these procedures are identical, indicating
that in both experiments the SHG signal originated from HOCl.
The same spectrum is obtained when the ice sample is exposed
to 0.5 L HOCl from preparation 1. At 155 K, the SHG signal
from the adsorbed HOCl is stable for several hours with the
ice sample held in equilibrium with water vapor.
It is interesting that HOCl can be stabilized on the ice in the
presence of the HCl impurities in preparations 1 and 2. Because
of the complications introduced by these possible reactants for
HOCl, we were unable to deduce an accurate coverage of HOCl
on the ice surface; from the exposure of 0.5 L, it is at most 0.5
monolayer, but it is probably less. Note that by adding HCl to
the chamber after HOCl is formed on the surface by in situ
hydrolysis of ClONO2 we observe a decay of the SHG signal
associated with HOCl until the ice background signal is obtained
(presumably by reaction 2; see section 4.7). We do not observe
Cl2, an impurity in the HOCl preparations and also a product
of reaction 2, on the ice surface. This is not surprising as it was
shown previously that Cl2 has a negligible lifetime on ice at
these temperatures; its desorption temperature is 110 K.34
Comparison of the SHG surface spectrum of HOCl with
known gas-phase spectra suggests the predominance of molec-
ular HOCl at the ice/water vapor interface. The shoulder rising
toward the blue starting at 295 nm is similar to the gas phase
and aqueous spectra of HOCl. The position of the higher energy
peak could not be determined with the laser available.
4.3. Polarization-Resolved SHG from HOCl Dosed Ice.
When interrogating the structure of the HOCl adlayer on the
ice surface using s-in, s-out polarization analyzed SHG signals,
we find significantly increased signals over that from the bare
ice (Table 1). This emphasizes that the origin of the s-in s-out
signal is indeed from the surface (<1 ML HOCl adsorbed) and
also that HOCl is binding to the ice surface in registry with the
underlying ice lattice. If HOCl forms a hydrogen bond by
donating its hydrogen to the oxygen of surface waters arranged
in a bilayer (as indicated by ab initio calculations),60,61 the
resulting pattern would be consistent with the observed anisot-
ropy. These observations argue against the suggestion that the
ice surface at 185 K is so diffuse that it is amorphous111 or
pseudoliquid-like.112 It appears that the adsorption of HOCl or
HNO3 in monolayer or submonolayer amounts does not cause
local surface melting, at least not to the extent that the symmetry
of the surface is altered, as has been suggested by others for
Figure 3. (a) Linear absorption spectrum of gas-phase HOCl (s),105,108
of an aqueous solution of HOCl (- - -), and of an aqueous solution of
ClO-(‚‚‚).108 (b) SHG spectra of HOCl adsorbed on ice at 155 K:
polarization ) 45-in, s-out; filled circles ) 0.5 L exposure of HOCl
from preparation 2, open circles ) in situ preparation of HOCl from
2.4 L ClONO2.
follow an ABAB... stacking pattern35 with each AB bilayer
rotated by 180° with respect to the adjacent AB bilayer);38 the
s-in, s-out bulk signal must therefore cancel.
4.2. Surface Spectroscopy of HOCl at the Ice/Water Vapor
Interface. HOCl is resonant but has a small absorption cross
section at 300 nm (7 × 10-20 cm2);105 thus, the resonant
enhancement of the SHG signal at 300 nm (eq 7) is expected
to be modest. Table 1 shows it is possible to detect HOCl using
SHG and to study its interaction with the ice surface.
The identification of HOCl at the ice surface was addressed
using SHG as an electronic spectroscopy. There have been many
experimental105-108 and theoretical109,110 studies on the electronic
structure of HOCl in the 200-400 nm region. Although
disagreement exists among experimentally measured absorption
cross sections of HOCl, it is generally accepted that the shape
of the HOCl vapor spectrum is centered about two peaks at
252 and 304 nm, respectively (Figure 3a).
113
HNO3 or HCl75 adsorption on ice. Rather, the symmetry of
the surface structure is retained when HNO3 or HOCl adsorb
to the basal ice surface. Adsorption-induced changes in the ice
surface have been observed for acetylene and H2S on ice
nanocrystals;114 however, the surface of nanocrystals may not
be comparable to the basal ice surface here.
4.4. Kinetics of HOCl Desorption from Clean Ice Surfaces.
Unlike the case of TA where two differently bound species were
found on the ice surface desorbing at different temperatures,81
HOCl desorbs as a single species. Figure 4 is an example of
the isothermal desorption kinetics of HOCl at the equilibrium
ice surface. 0.1 L of HOCl (preparation 1) is exposed to the ice
surface at 162.5 K; after dosing is complete, the ice surface
temperature is held constant and the SHG signal is monitored
as a function of time. The SHG signal decays exponentially
and returns to the ice background signal. The ice sample was
then flash heated to confirm that the HOCl had desorbed into
the gas phase and was not incorporated into the bulk.
Surface SHG spectra (Figure 3b) were obtained by (1)
exposing an ice sample at 170 K to 0.5 L of HOCl from

Feature Article
J. Phys. Chem. B, Vol. 103, No. 39, 1999 8211
∆H/des of 31 ( 4 kJ/mol are obtained. The negative entropy of
desorption could indicate that adsorbed HOCl must diffuse or
reorient to find the one degree of freedom that allows it to
desorb. This reduction in the degrees of freedom agrees with a
decrease in entropy.
Using the relation Ed/es ) ∆Hd/es + RT, the activation energy
of desorption E/des is calculated to be 32 ( 4 kJ/mol. The
thermodynamic parameters result in a free activation energy of
desorption ∆G/des of 48 ( 8 kJ/mol for HOCl desorption from
ice at 185 K. The cohesive energy of ice itself is 47 kJ/mol.94
The data are consistent with a hydrogen-bonding interaction
between HOCl and ice.
Thermal desorption spectra (TDS) of TA and HOCl were
recorded as well. In the TDS experiments, the clean ice sample
held at a constant temperature is exposed to a known exposure
of sample gas. The temperature of the ice sample is then raised
linearly with time, while the partial pressure of gas desorbing
from the ice surface is monitored by the mass spectrometer.
The molecules desorbing from other surfaces being heated, such
as the sample mount and heater, are distinguished from those
desorbing from the ice surface by comparison to thermal
desorption measurements from the clean sample mount (no ice
sample present). TDS show rates of desorption; at the maximum
rate, the following relation holds:
Figure 4.
I
versus time from HOCl on ice at an ice surface
x
temperature ofSH1G62.5 K (p (HOCl) ) 1.5 × 10-9 Torr; polarization
was 45-in s-out; λSHG ) 300 nm). The dashed vertical lines indicate
the start and cessation of dosing. (
surface).
I
x
≈ 0.8 for a clean ice
SHG
E/des
-E/des
dt
) υ
exp
(8)
( )
(
)
2
dT
RTmax
RTmax
where dT/dt is the heating rate.118 TA is found to have a Tmax
of 189 K, and HOCl has an average Tmax of 187 K. For the
observed Tmax value for HOCl and a heating rate of 0.2 deg/s,
E/des varies from 38 to 41 kJ/mol for υ ) 109 to 1010 s-1, as
estimated from the Arrhenius treatment. In the case of HOCl,
we varied the heating rate from 0.06 to 0.2 deg/s and found
that Tmax does not change within our experimental error. This
situation is usually interpreted to mean that the desorption is
first order119 (meaning that the desorption is from intact HOCl
molecules and does not involve the association of H+ and OCl-).
One note of caution with TDS measurements is that the surface
structure may change during the heating process. The flux of
water molecules at the equilibrium ice surface changes greatly
near 180-190 K. For this reason, we feel that the isothermal
SHG desorption measurements are more reliable for determining
E/des
.
4.5. Kinetics of HOCl Desorption from HNO3 Dosed Ice
Surfaces. In the preceding section, we showed that isothermal
desorption measurements can be performed using surface SHG,
and that the approach yields valuable kinetic and thermodynamic
information on how HOCl interacts with clean ice. More realistic
stratospheric particles may contain coadsorbed species such as
HNO3. In this section, we study the interaction of HOCl with
the basal ice surface predosed with various amounts of HNO3.
The ice surface at 188 K was exposed to 1 × 10-8 Torr HNO3
for up to 100 s such that exposures between 0.1 and 1.0 L were
obtained. Then, the surface was exposed to 0.1 L HOCl
(preparation 3), and the isothermal desorption SHG traces were
recorded as a function of time using the 45-in, s-out polarization
combination.
Figure 5. Arrhenius plot of isothermal desorption rate constants for
HOCl. The solid line is the weighted fit to an Arrhenius expression.
The exponential decay of the SHG signal corresponding to
desorption of HOCl from the ice surface can be used to
determine a first-order desorption rate constant kdes. To deter-
mine the change in kdes as a function of temperature, isothermal
desorption experiments were repeated at various ice temperatures
between 160 and 175 K.
An Arrhenius analysis of the desorption rate constants is
presented in Figure 5. From the slope (-E/des/R), an activation
energy of desorption, E/des, is calculated to be 36 ( 2 kJ/mol.
From the intercept, an Arrhenius prefactor υ of 3 × 109 s-1 is
obtained (range 1 × 109 to 1 × 1010). Additional thermodynamic
parameters for the desorption of HOCl are obtained from the
desorption rate constants using transition state theory (TST).115-117
From the weighted TST treatment, an activation entropy ∆Sd/es
of -90 ( 20 J/(mol K) and an activation enthalpy of desorption
As in the case presented for HOCl desorption from the clean
basal ice surface, the SHG signals decay exponentially in time
and return to the background signal but now depend on the
amount of predosed HNO3. Isothermal HOCl desorption rate
constants obtained from the data are plotted in Figure 6 as a

8212 J. Phys. Chem. B, Vol. 103, No. 39, 1999
Geiger et al.
Figure 6. Rate constants for desorption of HOCl from the ice basal
plane held at 188 K and predosed with different amounts of HNO3.
The line is the best fit to a single exponential. The experiments were
performed using the 45-in, s-out polarization combination; λSHG ) 300
nm.
Figure 7. Normalized and background subtracted SHG signal recorded
vs time after exposing the ice crystal held at 188 K to 0.3 L ClONO2.
The partial pressure of ClONO2 was 5 × 10-9 Torr. Exposure starts at
0 s and lasts for 60 s and is indicated by the two vertical bars. The
solid line is a guide to the eye. The polarization combination was 45-
in, s-out; λSHG ) 300 nm.
function of HNO3 preexposure. It can be seen that the HOCl
desorption rate constants decrease with increasing amount of
predosed HNO3.
is observed in our experiments. The model is somewhat
plausible but of course neglects the effects of ice dipoles, which
probably could play a role (but possibly a smaller role), and
other effects such as ionization and solvation. In such a complex
system, other possibilities abound. Perhaps the HOCl and HNO3
do not interact directly; for example, HOCl may need to diffuse
to sites at which it can achieve a conformation that allows
desorption, and HNO3 could block these sites. Later, we present
data more consistent with a local bonding picture.87 More
detailed calculations are in progress in our laboratory.
Data in Figure 6 were fit to a single exponential of the form
/
kdes ) (A′ e-(E /RT)) e-(Bx/RT)
(9)
des
where x is the amount of preexposed HNO3 in units of langmuir.
The exponential relationship of kdes with HNO3 exposure from
0.1 to 1.0 L can be justified using a model in which the barrier
to desorption of HOCl from ice increases linearly with HNO3
exposure: E′/ ) Ed/es + Bx, where E′/ is the barrier to
4.6. Formation of HOCl by in situ Hydrolysis of ClONO2
on Clean Ice. To verify our understanding of HOCl interactions
with ice and with HNO3 dosed ice, we produced HOCl in situ
on the clean ice surface with the simultaneous production of
HNO3 via the hydrolysis of ClONO2 (reaction 1). Figure 7
shows the SHG signal vs time trace, with the background
subtracted, as the clean basal ice surface at 188 K is exposed
to ClONO2. At 0 s, the ice is exposed to ClONO2 (partial
pressure of 5 × 10-9 Torr) for 60 s. It can be seen that, during
exposure, the SHG signal intensity does not change. After an
induction time of 100 s, the signal increases sigmoidally and
then remains stable over time. The increased SHG signal is
associated with HOCl formed at the surface, as argued in section
4.2. The induction time is not an artifact of the chamber, as
shown by numerous control experiments and given the very
high pumping speed of the chamber vacuum system. A more
detailed analysis of this reaction is given in another paper.87
From Figure 7, it is clear that once HOCl is formed after
200 s it remains on the ice surface. Recall from Figure 4 that
HOCl desorption from clean ice held at 162.5 K is fast; in
contrast, data in Figure 7 are at 188 K and show that HOCl
desorption from the surface is inhibited. This is due to the
simultaneous production of HNO3 by the reaction and is
consistent with the result that the energy barrier for HOCl
desorption is raised in the presence of HNO3 (previous section).
The association of HOCl with HNO3 is consistent with a 10-
member cyclic cluster proposed as a mechanism for the
hydrolysis of chlorine nitrate on ice by Bianco and Hynes.123
des
des
desorption dependent on HNO3 exposure and B is an interaction
factor. Using E/des ) 36 ( 2 kJ/mol, the fit parameters were A′
) 2 × 108 s-1 (range 1 × 108 to 5 × 108) and B ) 4 ( 1 kJ
mol-1 L-HN1 O
.
3
Why does HNO3 raise the energy barrier for HOCl desorption
from ice? One very simple argument based entirely on local
effects is that electrostatic dipole-dipole interactions between
HNO3 (dipole moment 2.17 D)120 and HOCl (1.30 D)121 raise
the barrier. Previously, we examined the interaction of HOCl
with a cluster of four water molecules mimicking a plausible
and abundant adsorption site on ice, and we found that
electrostatic dipole-dipole interactions are mostly responsible
for binding in the HOCl‚(H2O)4 cluster.61 The interaction energy
E
1/2 between two polar molecules with dipole moments µ1 and
µ2 that are separated by a distance R can be estimated by the
use of the following expression:122
E1/2 ) -(µ1µ2/R3)C
(10)
where C is a factor that is dependent on the angle between the
two dipoles. Here, we compare the complexes HOCl-H2O and
HOCl-HNO3 and assume that the dipoles are in a head to tail
arrangement and R and the angle-dependent factor are the same
in each case. Using the H-bond distance we obtained for HOCl
upon interaction with four water molecules (1.78 Å),61 we
estimate that the HOCl-“surface” interaction energy is roughly
20% larger for HNO3 than for water. This is approximately what

Feature Article
J. Phys. Chem. B, Vol. 103, No. 39, 1999 8213
TABLE 2: Rate Constants kdes for HOCl Desorption from
the Basal Ice Surface Held at 188 K and Predosed with
Various Amounts of HNO3 (also Plotted in Figure 6) and
Corresponding HOCl Lifetimes τ on the HNO3-Coated Ice
Surface from Eq 13
HNO3 pre-exposure [L]
kdes × 103 [s-1
]
τ [s]
0
3a
0.1
0.21
0.5
0.75
1.0
11 ( 3
90 ( 20
112 ( 10
244 ( 30
880 ( 40
670 ( 40
8.9 ( 0.8
4.1 ( 0.5
1.14 ( 0.05
1.5 ( 0.1
a Estimated HOCl lifetime on clean ice at 188 K calculated from eq
11. Note that, for temperatures below 165.8 K, lifetimes calculated
using eq 13 are only in qualitative agreement with the lifetimes using
eq 11. This is due to the errors associated with the determined Arrhenius
parameters.
estimate the adsorption rate by
NA[HOCl]0
/
ads/RT)
ads. rate )
e(-E
(12)
x
2πMRT
Figure 8. Measured rates of the HOCl decay during the reaction HOCl
+ HCl f Cl2 + H2O as a function of the initial ClONO2 exposure
used to make the HOCl. The ice was held at 168 K.
where E/ads is the barrier to adsorption (usually zero for
nondissociative adsorption processes).117 We apply stratospheric
temperatures of 185 K and an HOCl volume mixing ratio for
the stratosphere in the range of 1-10 pptv.29,124-126 At 40 Torr
total pressure in the lower polar stratosphere, the HOCl partial
pressure [HOCl]0 thus ranges from 4 × 10-11 to 40 × 10-11
Torr. Assuming no barrier to adsorption, the adsorption rate is
approximately (0.1-1) × 1011 molecules cm-2 s-1 for the range
1-10 pptv HOCl. The coverage is then 0.4 to 4 × 1011
molecules cm-2; assuming a monolayer on ice would consist
of 6 × 1014 molecules/cm2 (estimated from a model of the
bilayer terminated basal ice surface), the coverage would then
be 0.01% to 0.1% of a monolayer. This is in qualitative
agreement with the results of Oppliger et al.,53 Hanson et al.,48
and Popov et al.57
4.7. Surface Reaction of HOCl with HCl. Beginning with
HOCl adsorbed on ice at 168 K (prepared as in the previous
section), subsequent exposure of 1 L HCl caused an exponential
decrease of the SHG signal. This decay is consistent with the
reaction of HOCl with HCl, resulting in the formation of Cl2
and H2O (reaction 2).
We also measured the dependence of the decay rate rd of the
SHG signal on the initial ClONO2 exposure that was used to
make the HOCl; the experiment was repeated for various
ClONO2 exposures between 0.3 and 2.0 L, all followed by one
L HCl exposure. All SHG traces were fit to a single-exponential
function of the form exp(-rdt). The measured rates are plotted
against ClONO2 exposure in Figure 8. The error bars for the
ClONO2 exposure are 10%. It can be seen that the decay rates
increase with increasing amount of ClONO2 that is hydrolyzed
to HOCl on the ice surface. Kinetic analysis of the data is
complicated, since it involves (1) the adsorption of ClONO2
(which we measure in another indirect way in another contribu-
tion)87 and (2) the subsequent bimolecular reaction of the product
HOCl with HCl. Assuming this latter reaction is a straightfor-
ward bimolecular reaction, the curve in Figure 8 is suggestive
that the initial adsorption of ClONO2 is Langmuirian, and that
1 L ClONO2 exposure is near monolayer coverage (implying
an initial sticking coefficient near unity for ClONO2 on ice at
168 K).
A significant new finding of this work is that HOCl lifetimes
on ice surfaces are highly dependent on the coadsorbed HNO3
coverage. From the experimentally determined rate constants
kdes for the first-order HOCl desorption process from HNO3
dosed ice held at 188 K, we can calculate HOCl lifetimes on
HNO3 dosed ice directly:127
1
kdes
τ )
(13)
These lifetimes are at least 1 order of magnitude longer than
HOCl lifetimes on the clean ice surface (see Table 2). Multiply-
ing the HOCl adsorption rate with the HOCl lifetimes deter-
mined for ice surfaces dosed with 0.1-1.0 L HNO3 (lifetimes
of 90 to 670 s, respectively) and assuming that a monolayer on
ice consists of 6 × 1014 molecules/cm2 as before, the HOCl
coverage would then be 0.2% of a monolayer on HNO3 poor
ice surfaces (low HNO3 exposure around 0.1 L) and 20% of a
monolayer for the case of HNO3 rich ice surfaces (high HNO3
exposure around 1 L).
5. Calculated Lifetimes and Coverages
The lifetime of HOCl on the clean ice surface can be
estimated as follows:
/
des/RT)
τ ) τv e(E
(11)
where τv is the lifetime of a vibration.31 Assuming τv ) 3 ×
10-10 s (inverse of the Arrhenius pre-exponential we determined)
and E/des ) 36 kJ/mol (determined by TST treatment of our
isothermal desorption data), one obtains a lifetime for HOCl
on ice at 185 K of 4 s.
6. Conclusions
Second harmonic generation, and a related method, sum
frequency generation (SFG), in which surface vibrational
information can be obtained,82-86 are suitable for the study of
a multitude of heterogeneous processes that are important in
the chemistry and physics of the troposphere and stratosphere.
Direct information about thermodynamics and surface structures
To determine the HOCl surface coverage on clean ice, the
lifetime of HOCl is multiplied by the adsorption rate.31 We can

8214 J. Phys. Chem. B, Vol. 103, No. 39, 1999
Geiger et al.
can be obtained. Real time measurements yield reaction times
and lifetimes. Fundamental results such as these can improve
our understanding of some of the complex phenomena involved
in global change processes such as stratospheric ozone losses,
greenhouse gas build up, or tropospheric pollutant concentra-
tions.
support. F.M.G. wishes to thank NASA for a Global Change
Graduate Student Fellowship in Earth System Sciences. A.C.T.
acknowledges support from the ARCS Foundation. We thank
Y. Ron Shen, Mario Molina, and Roberto Bianco for fruitful
discussions.
In this study, we observed polarized SHG signals that are
consistent with an ordered ice surface at 158 K: the 3m
symmetric basal face of ice. The observation of a structured
and ordered surface is in contrast to suggestions by others that
ice surfaces are disordered.128 We find that the symmetry of
the structure of the basal surface is retained upon adsorption of
a monolayer or less of HOCl and HNO3. Experimental results
by others suggest surface melting upon adsorption of HNO3 on
ice at 200-223 K.113
References and Notes
(1) Molina, M. J.; Rowland, F. S. Nature 1974, 249, 810.
(2) Farman, J. C.; Gardiner, B. G.; Shanklin, J. D. Nature 1985, 315,
207.
(3) Solomon, S.; Garcia, R. R.; Rowland, F. S.; Wuebbles, D. J. Nature
1986, 321, 755.
(4) McCoustra, M. R. S.; Horn, A. B. Chem. Soc. ReV. 1994, 279,
195.
(5) Molina, M. J. CHEMRAWN VII: Chemistry of the Atmosphere:
The Impact of Global Change; Blackwell Scientific Publisher: New York,
1994.
Exposure of the equilibrium basal ice surface to HOCl,
ClONO2, and HNO3 was carried out using stratospherically
relevant partial pressures and exposure times on the order of
minutes. These exposure times are representative of stratospheric
exposures under some circumstances. Recent field and labora-
tory studies and model calculations show that PSC lifetimes
are on the order of minutes15 to hours;11,16 the lifetime of a stable
PSC particle surface, however, could be much shorter given
the fluctuations in local relative humidity in the PSC’s.129
Despite the fact that the ice surface is dynamic at 185 K,
with 100 monolayers of water simultaneously condensing and
evaporating, the stable SHG signals from monolayer or sub-
monolayer amounts of adsorbates (for hours) show that the
adsorbates are not buried by condensing water molecules to the
extent that they enter the bulk ice. Experiments using an
overpressure of water vapor in the chamber do result in the
burying of an adsorbate such as p-tolualdehyde, and the SHG
signal then disappears. With subsequent heating as the added
water layers are removed, SHG signals from the adsorbate are
regained. In the equilibrium situation, it is not surprising that
an adsorbate, whose presence at the surface lowers the overall
surface tension, remains at the surface despite a large water flux.
Consider a liquid water surface, whose flux is even higher; on
a very short time scale, adsorbates dynamically readjust to new
water orientations in order to lower the overall surface free
energy. Thus, for molecules insoluble in ice (constituting the
majority of molecules), stable surface populations are expected.
HOCl was observed directly on the surface using SHG.
Identification of HOCl was addressed by surface electronic
spectroscopy and by its chemical reactivity with HCl. HOCl
desorbs as a single species with a ∆G/des ) 48 ( 4 kJ/mol,
close to the cohesive energy of ice itself. The combination of
an Arrhenius analysis of isothermal HOCl desorption rate
constants measured on the clean basal ice surface with HOCl
lifetime measurements on HNO3 dosed ice surfaces shows that,
at 188 K, the HOCl lifetime on the basal ice surface is
lengthened by at least one order of magnitude in the presence
of coadsorbed HNO3.
(6) Kolb, C. et al. In Progress and Problems in Atmospheric Chemistry;
Barker, J. R., Ed.; World Scientific: Singapore, 1995; Vol. 3.
(7) Molina, M. J.; Tso, T.-L.; Molina, L. T.; Wang, F. C.-Y. Science
1987, 238, 1253.
(8) Tolbert, M. A.; Rossi, M. J.; Malhotra, R.; Golden, D. M. Science
1987, 238, 1258.
(9) Crutzen, P. J.; Arnold, F. Nature 1986, 324, 651.
(10) Toon, O. B.; Hamill, P.; Turco, R. P.; Pinto, J. Geophys. Res. Lett.
1986, 13, 1284.
(11) Turco, R. P.; Toon, O. B.; Hamill, P. J. Geophys. Res. 1989, 94,
16493.
(12) Pueschel, R. F.; Snetzinger, K. G.; Goodman, J. K.; Toon, O. B.;
Ferry, G. V.; Oberbeck, V. R.; Livingston, J. M.; Varma, S.; Fong, W.;
Starr, W. L.; Chan, K. R. J. Geophys. Res. 1989, 94, 16449.
(13) Deshler, T.; Peter, T.; Mueller, R.; Crutzen, P. J. Geophys. Res.
Lett. 1994, 21, 1327.
(14) Goodman, J.; Verma, S.; Pueschel, R. F.; Hamill, P.; Ferry, G. V.;
Webster, D. Geophys. Res. Lett. 1997, 24, 615.
(15) Biermann, U. M.; Crowley, J. N.; Huthwelker, T.; Moortgat, G.
K.; Crutzen, P. J.; Peter, T. Geophys. Res. Lett. 1998, 25, 3939.
(16) Peter, T.; Mueller, R.; Crutzen, P. J.; Deshler, T. Geophys. Res.
Lett. 1994, 21, 1331.
(17) Chen, J.; Crutzen, P. J. J. Geophys. Res. 1996, 101, 23035.
(18) Marti, J.; Mauersberger, K. Geophys. Res. Lett. 1993, 20, 363.
(19) Haynes, D. R.; Tro, N. J.; George, S. M. J. Phys. Chem. 1992, 96,
8502.
(20) Livingston, F. E.; George, S. M. J. Phys. Chem. 1998, 102, 10280.
Warshawsky, M. S.; Zondlo, M. A.; Tolbert, M. A. Geophys. Res. Lett.
1999, 26, 823.
(21) Atkinson, R.; Tuazon, R. C.; Mac Leod, H.; Aschmann, S. M.;
Winer, A. M. J. Geophys. Res. 1986, 13, 117.
(22) Molina, L. T.; Molina, M. J.; Stachnik, R. A.; Tom, R. D. J. Phys.
Chem. 1985, 89, 3779.
(23) Solomon, S. ReV. Geophys. 1988, 26, 131.
(24) Prather, M. J. Nature 1992, 355, 534.
(25) Schoeberl, M. R.; Douglass, A. R.; Kawa, S. R.; Dessler, A. E.;
Newman, P. A.; Stolarski, R. S.; Roche, A. E.; Waters, J. W.; Russell, J.
M., III J. Geophys. Res. 1996, 101, 20909.
(26) Santee, M. L.; Tabazadeh, A.; Manney, G. L.; Salawitch, R. J.;
Froidevaux, L.; Read, W. G.; Waters, J. W. J. Geophys. Res. 1998, 103,
13285.
(27) Santee, M.; Froidevaux, L.; Manney, G. L.; Read, W. G.; Waters,
J. W.; Chipperfield, M. P.; Roche, A. E.; Kumer, J. B.; Mergenthaler, J.
L.; Russell, J. M., III J. Geophys. Res. 1996, 101, 18835.
(28) Kawa, S. R.; Newman, P. A.; Lait, L. R.; Schoeberl, M. R.;
Stimpfle, R. M.; Kohn, D. W.; Webster, C. R.; May, R. D.; Baumgardner,
D.; Dye, J. E.; Wilson, J. C.; Chan, K. R.; Loewenstein, M. J. Geophys.
Res. 1997, 102, 3921.
(29) Jaegle´, L. W. C. R.; May, R. D.; Scott, D. C.; Stimpfle, R. M.;
Kohn, D. W.; Wennberg, P. O.; Hanisco, T. F.; Cohen, R. C.; Proffitt, M.
H.; Kelly, K. K.; Elkins, J.; Baumgardner, D.; Dye, J. E.; Wilson, J. C.;
Pueschel, R. F.; Chan, K. R.; Salawitch, D. J.; Tuck, A. F. J. Geophys.
Res. 1997, 102, 13235.
Direct surface information obtained using SHG sheds new
light on previous results employing flow reactors, in which
reactive uptake coefficients, γ, are determined. For example,
for the reaction HOCl + HCl f H2O + Cl2, γ was reported to
be smaller for nitric acid coated ices than for initially clean
ices.8,48,64,65,67,68 The smaller γ could be due to retardation of
the reaction rate by the sequestering of HOCl by HNO3 as
suggested by our work.
(30) Materer, N.; Starke, U.; Barbieri, A.; Van Hove, M. A.; Somorjai,
G. A.; Kroes, G.-J.; Minot, C. Surf. Sci. 1997, 381, 190. Materer, N.; Starke,
U.; Barbierii, A.; Van Hove, M. A.; Somorjai, G. A.; Kroes, H.-J.; Minot,
C. J. Phys. Chem. 1995, 99, 6267.
(31) Brown, D. E.; George, S. M. J. Phys. Chem. 1996, 100, 15460.
(32) Schaff, J. E.; Roberts, J. T. J. Phys. Chem. 1996, 100, 14151.
(33) Foster, K. L.; Tolbert, M. A.; George, S. M. J. Phys. Chem. A
1997, 101, 4979.
(34) Banham, S. F.; Horn, A. B.; Koch, T. G.; Sodeau, J. R. Faraday
Discuss. 1995, 100, 321.
Acknowledgment. The authors acknowledge the support of
NSF Grants ATM 9022613 and the Presidential Young Inves-
tigator Award ATM-9158080. J.M.H. wishes to thank the Henry
B. Luce Foundation and the Alfred P. Sloan Foundation for

Feature Article
J. Phys. Chem. B, Vol. 103, No. 39, 1999 8215
(35) Braun, J., Glebov, A., Graham, A. H., Menzel, A.; Toennies, J. P.
Phys. ReV. Lett. 1998, 80, 2638.
(85) Eisenthal, K. B. In Annual ReView of Physical Chemistry; Strauss,
H. L., Babcock, G. T., Leone, S. R., Eds.; Annual Reviews Inc: Palo Alto,
CA, 1992; Vol. 43, p 627.
(86) Corn, R. M.; Higgins, D. A. Chem. ReV. 1994, 94, 107.
(87) Geiger, F. M.; Pibel, C. D.; Hicks, J. M. In preparation.
(88) Bouchez, C.; Hicks, J. M. Laser Techniques in Surface Science II;
SPIE: Bellingham, WA, 1995; Vol. 2547, p 152.
(89) Mizrahi, V.; Sipe, J. E. J. Opt. Soc. Am. B 1988, 5, 660.
(90) Tom, H. W. K., Heinz, T. F.; Shen, Y. R. Phys. ReV. 1983, 51B,
1983.
(36) Devlin, J. P.; Buch, V. J. Phys. Chem. 1995, 99, 16534.
(37) Faraday, M. Philos. Mag. 1859, 17, 162.
(38) Fletcher, N. H. The Chemical Physics of Ice; Cambridge University
Press: London, 1970.
(39) Nason, D.; Fletcher, N. H. J. Chem. Phys. 1975, 62, 4444.
(40) Golecki, I.; Jaccard, C. J. Phys. C 1978, 11, 4229. Mizuno, Y.;
Hanafusa, N. J. Phys. C 1987, 3, 48.
(41) Dzyaloshinskii, I. E.; Lifshitz, E. M.; Pitaevskii, L. P. AdV. Phys.
1961, 10, 165.
(91) Shen, Y. R. Nature 1989, 337, 519.
(42) Elbaum, M.; Schick, M. Phys. ReV. Lett. 1991, 66, 1713.
(43) Kroes, G.-J.; Clary, D. C. J. Phys. Chem. 1992, 96, 7079.
(44) Furukawa, Y.; Nada, H. J. Cryst. Growth 1997, 169, 587.
(45) Baker, M.; Dash, J. G. J. Geophys. Res. 1996, 101, 12929. Wilen,
L.; Dash, J. G. Science 1995, 270, 1184.
(46) Lied, A.; Dosch, H.; Bilgram, J. H. Phys. ReV. Lett. 1994, 72, 3554.
(47) Wettlaufer, J. S. Phys. ReV. Lett. 1999, 82, 2516
(48) Hanson, D. R.; Ravishankara, A. R. J. Phys. Chem. 1992, 96, 2682.
(49) Chu, L. T.; Leu, M.-T.; Keyser, L. F. J. Phys. Chem. 1993, 97,
12798.
(92) Heinz, T. F.; Chen, C. K.; Ricard, D.; Shen, Y. R. Phys. ReV. Lett.
1982, 48, 478.
(93) Hobbs, P. V. Ice Physics; Clarendon Press: Oxford, 1974.
(94) Whitworth, R. W.; Ahmad, S.; Ohtomo, M. J. Phys. 1987, 48, 595.
(95) Bryson, C. E.; Cazcarra, V.; Levenson, L. L. J. Chem. Eng. Data
1974, 19, 107.
(96) D’Ans, J.; Freund, H. E. Z. Elekrtochem. 1957, 61, 10.
(97) Zhang, R. Personal communication.
(98) Molina, L. T.; Spencer, J. E.; Molina, M. J. Chem. Phys. Lett. 1977,
45, 158.
(50) Hanson, D. R. J. Phys. Chem. B 1997, 101, 4998.
(51) Hanson, D. R.; Ravishankara, A. R. J. Phys. Chem. 1993, 97, 2802.
(52) Keyser, L. F.; Leu, M.-T.; Moore, S. B. J. Phys. Chem. 1993, 97,
2800.
(99) Gruenhut, N. S.; Goldfrank, M.; Cushing, M. L.; Caesar, G. V.
Angew. Chem. 1950, 3, 78.
(100) Schmeisser, M.; Braendle, K. Angew. Chem. 1961, 173, 388.
(101) Miller, R. H.; Bernitt, D. L.; Hisatsune, I. C. Spectrochim. Acta
1967, 23A, 223.
(53) Oppliger, R.; Allanic, A.; Rossi, M. J. J. Phys. Chem. A 1997, 101,
1903.
(102) Harwood, M. H.; Jones, R. L.; Cox, R. A.; Lutman, E.; Rattigan,
O. V. J. Photochem. Photobiol. A: Chem. 1993, 73, 167.
(103) Masel, R. I. Principles of adsorption and reaction on solid
surfaces; John Wiley & Sons: New York, 1996.
(54) Abbatt, J. P. D.; Molina, M. J. Geophys. Res. Lett. 1992, 19, 461.
(55) Abbatt, J. P. D.; Beyer, K. D.; Fucaloro, A. F.; McMahon, J. R.;
Wooldridge, P. J.; Zhang, R.; Molina, M. J. J. Geophys. Res. 1992, 97,
15819.
(104) Middlebrook, M.; Berland, B. S.; George, S. M.; Tolbert, M. A.
J. Geophys. Res. 1994, 99, 25655.
(56) Leu, M. T. Geophys. Res. Lett. 1988, 15, 17.
(57) Popov, A. M.; Popovicheva, O. B.; Rakhimova, T. V.; Feoktistov,
V. A. Khim. Fiz. 1995, 14, 137.
(58) Robinson-Brown, A.; Doren, D. J. J. Phys. Chem. B 1997, 101,
6308.
(59) Ghanty, T. K.; Ghosh, S. K. J. Phys. Chem. A 1997, 101, 5022.
(60) Dibble, T. S.; Francisco, J. S. J. Phys. Chem. 1995, 99, 1919.
(61) Geiger, F. M.; de Dios, A. C.; Hicks, J. M. J. Phys. Chem. 1998,
102, 1514.
(105) Burkholder, J. B. J. Geophys. Res. 1993, 98, 2693.
(106) Permien, T.; Vogt, R.; Schindler, R. N. In Mechanisms of Gas
Phase and Liquid-Phase Chemical Transformations; Cox, R. A., Ed.;
Environ. Res. Prog. of the CEC: Brussels, Belgium, 1988.
(107) Mishalanie, E. A.; Rutkowski, C. J.; Hutte, R. S.; Birks, J. W. J.
Phys. Chem. 1986, 90, 5578.
(108) Molina, L. T.; Molina, M. J. J. Phys. Chem. 1978, 82, 2410.
(109) Guo, H. J. Phys. Chem. 1993, 97, 2602.
(62) Tolbert, M. A.; Rossi, M. J.; Golden, D. M. Science 1988, 240,
1018.
(63) Leu, M. T. Geophys. Res. Lett. 1988, 15, 851.
(64) Abbatt, J. P. D.; Molina, M. J. J. Phys. Chem. 1992, 96, 7674.
(65) Hanson, D. R.; Ravishankara, A. R. J. Geophys. Res. 1991, 96,
5081.
(110) Nanbu, S.; Iwata, S. J. Phys. Chem. 1992, 96, 2103.
(111) Graham, J. D.; Roberts, J. T. Geophys. Res. Lett. 1995, 22, 251.
(112) Davidovits, P. Faraday Discuss. 1995, 100, 338.
(113) Zondlo, M. A.; Barone, S. B.; Tolbert, M. A. Geophys. Res. Lett.
1997, 24, 1391.
(114) Delzeit, L.; Devlin, M. S.; Rowland, B.; Devlin, J. P.; Buch, V.
J. Phys. Chem. 1996, 100, 10076.
(115) Eyring, H. J. Chem. Phys. 1935, 3, 107.
(66) Ravishankara, A. R.; Hanson, D. R. J. Geophys. Res. 1996, 101,
3885.
(67) Moore, S. B.; Keyser, L. F.; Leu, M.-T.; Turco, R. P.; Smith, R.
H. Nature 1990, 345, 333.
(68) Leu, M.-T.; Moore, S. B.; Keyser, L. F. J. Phys. Chem. 1991, 95,
(116) Evans, G. M.; Polyani, M. Trans. Faraday Soc. 1935, 31, 875.
(117) Steinfeld, J. I.; Francisco, J. S.; Hase, W. L. Chemical Kinetics
and Dynamics; Prentice Hall: Eaglewood Cliffs, New Jersey, 1989.
(118) Graham, J. D.; Roberts, J. T. J. Phys. Chem. 1994, 98, 5974.
(119) Christmann, K. Introduction to Surface Physical Chemistry;
Springer-Verlag: New York, 1991.
(120) Atkins, P. W. Physikalische Chemie; VCH: Weinheim, 1990.
(121) CRC Handbook of Chemistry and Physics, 71st ed.; CRC Press:
Boston, 1990-1991.
(122) Lister, D. G.; Millen, D. J. Trans. Faraday Soc. 1971, 67, 601.
(123) Bianco, R., Hynes, J. T. J. Phys. Chem. A 1998 102, 309.
(124) Chance, K.; Traub, W. A.; Johnson, D. G.; Jucks, K. W.;
Ciarpallini, P.; A., S. R.; Salawitch, R. J.; Michelsen, H. A. J. Geophys.
Res. 1996, 101, 9031. Zander, R.; Mahieu, E.; Gunson, M. R.; Abrams, M.
C.; Chang, A. Y.; Abbas, M.; Aellig, C.; Engel, A.; Goldman, A.; Irion, F.
W.; Kampfer, N.; Michelsen, H. A.; Newchurch, M. J.; Rinsland, C. P.;
Salawitch, R. J.; Stiller, G. P.; Toon, G. C. Geophys. Res. Lett. 1996, 23,
2357.
7763.
(69) Quinlan, M. A.; Reihs, C. M.; Golden, D. M.; Tolbert, M. A. J.
Phys. Chem. 1990, 94, 3255.
(70) Donaldson, D. J.; Ravishankara, A. R.; Hanson, D. R. J. Phys.
Chem. 1997, 101, 4717.
(71) Hanson, D. R.; Lovejoy, E. R. J. Phys. Chem. 1996, 100, 6397.
(72) Hanson, D. R.; Ravishankara, A. R. J. Phys. Chem. 1993, 97, 12309.
(73) Zhang, R.; Leu, M.-L.; Keyser, L. F. J. Phys. Chem. 1994, 98,
13563.
(74) Zhang, X.; Mereand, E. L.; Castleman, A. W., Jr. J. Phys. Chem.
1994, 98, 3554.
(75) Molina, M. J.; Zhang, R.; Wooldridge, P. J.; McMahon, J. R.; Kim,
H. Y.; Chang, H. Y.; Beyeer, B. D. Science 1993, 261, 1418.
(76) Henson, B. F.; Wilson, K. R.; Robinson, J. M. Geophys. Res. Lett.
1996, 23, 1021.
(77) Tabazadeh, A.; Turco, R. P. J. Geophys. Res. 1993, 98, 12727.
(78) Geiger, F. M.; Hicks, J. M. Laser Techniques in Surface Science;
SPIE: Bellingham, WA, 1998; Vol. 3272, p 296.
(79) Wong, T. M.; Tridico, A. C.; Miller, L. M.; Hicks, J. M. In Physics
and Chemistry of Ice; Maeno, N., Hondoh, T., Eds.; Hokkaido University
Press: Sapporo, 1992; p 190.
(80) Tridico, A. C.; Lakin, M.; Hicks, J. M. Laser Techniques in Surface
Science; SPIE: Bellingham, WA, 1995; Vol. 2125, p 160.
(81) Tridico, A. C., Ph.D. Thesis, Georgetown University, 1995.
(82) Shen, Y. R. The Principles of Nonlinear Optics; John Wiley &
Sons: New York, 1984.
(125) Johnson, D. G.; Traub, W. A.; Chance, K. V.; Jucks, K. W.;
Stachnik, R. A. Geophys. Res. Lett. 1995, 22, 1869.
(126) Webster, C. R.; May, R. D.; Toohey, D. W.; Avallone, L. M.;
Anderson, J. G.; Newman, P.; Lait, L.; Schoeberl, M. R.; Elkins, J. W.;
Chan, K. R. Science 1993, 261.
(127) Finlayson-Pitts, B. J.; Pitts, J. N., Jr. Atmospheric Chemistry:
Fundamentals and Experimental Techniques; John Wiley & Sons: New
York, 1986.
(128) Buch, V.; Delzeit, L.; Blackledge, C.; Devlin, J. P. J. Phys. Chem.
1996, 100, 3732.
(129) Carslaw, K. S.; Wirth, M.; Tsias, A.; Luo, B. P.; Doernbrack, A.;
Leutbecher, M.; Volkert, H.; Renger, W.; Bacmeister, J. T.; Peter, T. J.
Geophys. Res. 1998, 103, 5785.
(83) Richmond, G. L.; Robinson, J. M.; Shannon, V. L. Prog. Surf. Sci.
1988, 28, 1.
(84) Heinz, T. F. In Nonlinear Surface Electromagnetic Phenomena;
Ponath, H. E., Stegeman, G. I., Eds.; Elsevier Science Publishers: New
York, 1991; p 353.
(130) Chemical Kinetics and Photochemical Data for Use In Strato-
spheric Modeling, EValuation no. 12; JPL Publication: Pasadena, California,
Jan 15, 1997.