Photodissociation of a surface-active species at a liquid surface: A study by time-of-flight spectroscopy

Article abstract of DOI:10.1021/jp984054v

The photochemistry at a gas-liquid interface was investigated by time-of-flight/quadrupole mass spectroscopy (TOF/QMS). A thin liquid film of 4-iodobenzoic acid (IBA), dissolved in glycerol, was irradiated with nanosecond laser pulses at 275 nm. Atomic and molecular iodine were the only photoproducts leaving the liquid after a low-fluence laser pulse (<5 mJ/cm²). The amount of released I atoms was 2 orders of magnitude larger than the amount of desorbed I₂. Model calculations were carried out that take into account laser photolysis of IBA, diffusion, and surface evaporation of I and I₂, and the condensed-phase kinetics of radical reactions. Ejection of hyperthermal I atoms by direct photodissociation of surface layer molecules is also considered. The quantitative analysis is restricted to low laser fluence conditions at which laser-induced heating of the illuminated liquid is negligible. The results of the model calculations were compared to previously obtained TOF data of an alkyl iodide (C₁₈H₃₇I) dissolved in the apolar liquid squalane (C₃₀H₆₂). The velocity distribution of the iodine atoms from the alkyl iodide solution corresponds to the temperature of the liquid (278 K). The contribution of I atoms from depths greater than 1 nm is large (>99%). In contrast, I atoms desorbing from IBA/glycerol are hyperthermal (T_trans=480 K) and originate almost exclusively from the topmost molecular layer (1 nm). TOF measurements with a fast chopper wheel in front of the surface provide the time-dependent desorption flux from the surface and confirm that the contribution from deeper layers in the alkyl iodide solution is much larger than in the aryl iodide solution. Model calculations predict the behavior of the two solutions correctly if differences in caging of the geminate pair, diffusion coefficients of the free radicals, and the set of bulk radical reactions in the two solutions are taken into account. The hyperthermal energies of the ejected I atoms from the IBA solution are discussed in terms of the surface orientation of excited IBA molecules. The dependence of the TOF spectra on laser power and IBA concentration is interpreted by a surface excess of IBA. The result is compared to temperature-dependent surface tension measurements of IBA solutions in glycerol and water. The response of the surface tension to an accumulation of IBA at the surface is very weak.

Full text of DOI:10.1021/jp984054v

1550
J. Phys. Chem. B 1999, 103, 1550-1557
Photodissociation of a Surface-Active Species at a Liquid Surface: A Study by
Time-of-Flight Spectroscopy
Alan Furlan†
Physikalisch-Chemisches Institut der UniVersita¨t Zu¨rich, Winterthurerstrasse 190, CH-8057 Zu¨rich, Switzerland
ReceiVed: October 13, 1998; In Final Form: January 8, 1999
The photochemistry at a gas-liquid interface was investigated by time-of-flight/quadrupole mass spectroscopy
(TOF/QMS). A thin liquid film of 4-iodobenzoic acid (IBA), dissolved in glycerol, was irradiated with
nanosecond laser pulses at 275 nm. Atomic and molecular iodine were the only photoproducts leaving the
liquid after a low-fluence laser pulse (<5 mJ/cm2). The amount of released I atoms was 2 orders of magnitude
larger than the amount of desorbed I2. Model calculations were carried out that take into account laser photolysis
of IBA, diffusion, and surface evaporation of I and I2, and the condensed-phase kinetics of radical reactions.
Ejection of hyperthermal I atoms by direct photodissociation of surface layer molecules is also considered.
The quantitative analysis is restricted to low laser fluence conditions at which laser-induced heating of the
illuminated liquid is negligible. The results of the model calculations were compared to previously obtained
TOF data of an alkyl iodide (C18H37I) dissolved in the apolar liquid squalane (C30H62). The velocity distribution
of the iodine atoms from the alkyl iodide solution corresponds to the temperature of the liquid (278 K). The
contribution of I atoms from depths greater than 1 nm is large (>99%). In contrast, I atoms desorbing from
IBA/glycerol are hyperthermal (Ttrans ) 480 K) and originate almost exclusively from the topmost molecular
layer (1 nm). TOF measurements with a fast chopper wheel in front of the surface provide the time-dependent
desorption flux from the surface and confirm that the contribution from deeper layers in the alkyl iodide
solution is much larger than in the aryl iodide solution. Model calculations predict the behavior of the two
solutions correctly if differences in caging of the geminate pair, diffusion coefficients of the free radicals,
and the set of bulk radical reactions in the two solutions are taken into account. The hyperthermal energies
of the ejected I atoms from the IBA solution are discussed in terms of the surface orientation of excited IBA
molecules. The dependence of the TOF spectra on laser power and IBA concentration is interpreted by a
surface excess of IBA. The result is compared to temperature-dependent surface tension measurements of
IBA solutions in glycerol and water. The response of the surface tension to an accumulation of IBA at the
surface is very weak.
1. Introduction
at liquid-gas interfaces, in part because of the technical
problems in studying clean liquid surfaces under clean, high-
vacuum conditions. Only in the past decade have new techniques
been developed, including sum-frequency mixing5 and molecular
beam surface scattering,6 which reveal molecular details of gas-
liquid interfaces such as surface coverage and orientation of
surface molecules in liquid mixtures.7-9 Another technique often
used to analyze photodissociation processes at solid surfaces is
the time-of-flight method combined with quadrupole mass
spectroscopy (TOF/QMS).10 Owing to the universal detection
method (ionization by electron bombardment), virtually any
photoproduct can be identified. In addition, a TOF spectrum of
the desorbed products reveals the history of the species between
the laser pulse and the time of desorption and thus indirectly
probes the structure and orientation of the surface. In our
laboratory the TOF/QMS method has been recently applied for
the first time to the study of photoproducts ejected from liquid
surfaces.11,12 The liquid film preparation technique developed
by Fenn et al.13 and Nathanson et al.6 was used for producing
clean, renewed liquid surfaces of long-chain alkyl iodides (1-
iodododecane11 and 1-iodooctadecane in squalane12) in high
vacuum. At high enough laser fluences (>10 mJ/cm2) the
photoexcitation pulse was found to be followed by explosive
desorption of the ejected species. At fluences less than 5 mJ/
cm2, the velocities of the photoproducts, atomic I, HI, and I2,
The photochemistry at gas-liquid interfaces is of great
theoretical and practical interest, in particular since it has been
realized that processes occurring on atmospheric cloud particles
and aerosols have a large impact on photochemical cycles
responsible for the destruction of ozone.1-3 Applications of
photochemistry at gas-liquid interfaces can be found in various
fields including photography, bleaching, analytical methods such
as matrix-assisted laser desorption and ionization, and photo-
induced reactions on biological tissues exposed to UV radiation.4
The fate of a molecule in the liquid that is excited to a repulsive
electronic state depends on the location and orientation of the
species with respect to the interface. If the excited species is
located in the topmost liquid layer, the bond cleavage may be
followed by direct ejection of the fragments into the gas phase.
If the axis of the breaking bond is oriented parallel to the surface,
the recoiling fragments will collide with surrounding molecules
and desorb with a reduced kinetic energy. Fragments originating
from chromophores in deeper layers of the liquid will mostly
recombine geminately because of the surrounding solvent cage,
and the energy will be dissipated as heat.
Compared to gas-solid interfaces, it has been difficult to
obtain structural and orientational information about molecules
† E-mail: furlan@pci.unizh.ch.
10.1021/jp984054v CCC: $18.00 © 1999 American Chemical Society
Published on Web 02/13/1999

Photodissociation at Liquid Surface
J. Phys. Chem. B, Vol. 103, No. 9, 1999 1551
were in full thermal equilibrium with the surrounding liquid.
The time-dependent desorption flux of these species was
obtained by using a fast chopper wheel in front of the surface
synchronized with the laser pulse. By comparison of the TOF
spectra with numerical simulations, it was demonstrated that a
considerable fraction of the ejected photoproducts originate from
deeper layers within the liquid. The absence of prompt hyper-
thermal I fragments was interpreted as a preferential orientation
of the alkyl iodide molecules in the topmost layer with the I
atom pointing into the liquid.
This study is a continuation of our previous work11,12 with
amphiphilic molecules. These molecules tend to accumulate at
the surface of a polar solvent. As a test system, 4-iodobenzoic
acid (IBA) was chosen as a chromophore, dissolved in glycerol.
The photodissociation mechanism of isolated IBA molecules
is expected to be similar to the mechanism of related compounds
such as iodobenzene and other aryl halides.14 In contrast to alkyl
iodides, the cleavage of the C-I bond in aryl iodides is an
indirect process.14 Excitation in the range 250 ( 30 nm promotes
the molecules to the (π,π*) state. After internal conversion to
the repulsive (n,σ*) state on a time scale of 0.5-1.0 ps, the
C-I bond breaks promptly.14-16 A fraction of 30-50% of the
excess energy is partitioned to translational recoil of the
fragments.15
used in this study. The liquid was cooled to 5 °C by Peltier
elements attached to the bottom of the reservoir and controlled
with a thermocouple. The pressure in the source chamber
surrounding the liquid container was about 5 × 10-6 mbar. The
insertion of a liquid-nitrogen-cooled trap with a coldfinger in
front of the liquid reservoir reduced the pressure to e1 × 10-6
mbar but had no effect on the TOF spectra.
Excitation laser pulses at 275 nm and pulse energies of 25
µJ to 12 mJ/pulse (τ ≈ 6 ns and ∆ν ≈ 0.15 cm-1) were obtained
by frequency-doubling the output of a Nd:YAG-pumped dye
laser.
The collimated laser beam was focused horizontally with a
700 mm cylindrical lens and directed onto the liquid surface at
a grazing angle of 87°. The illuminated liquid surface area was
determined to be 2 × 6 mm2 by measuring the spot area at
normal incidence with a CCD camera and considering the
expansion in one direction due to the grazing incidence angle.
The reflection loss of the s-polarized laser beam impinging at
87° is 82%. The resulting laser fluence of the beam refracted
into the liquid is in the range 0.05-20 mJ/cm2. The penetration
depth within which the intensity falls to 1/e is 60 mm for a 4
× 10-5 mol/L solution, much larger than the film thickness.
The reflected light was directed out of the chamber via a dichroic
mirror.
In our previous study of alkyl iodide dissolved in squalane
an estimate of the surface sensitivity was obtained based on
model calculations. For that system it was found that hyper-
thermal photofragments from the topmost liquid layer should
be detected if their fraction exceeded 1-10% of all photofrag-
ments formed in the surface layer.12 A goal of this study is to
test this prediction with a system for which ejection of hyper-
thermal fragments from the surface is expected. Since IBA is
an amphiphilic molecule, it may accumulate at the surface. The
surface orientation of IBA is unknown. However, it can be
speculated that the -COOH group points preferentially into the
bulk liquid and forms H-bonds with the polar solvent, whereas
the -I end is preferentially directed out of the liquid. If this
preferential orientation of IBA at the surface is correct, one
would expect that the fastest I fragments desorb without
collisions and exhibit a velocity distribution similar to that of
gas-phase photodissociation.17,18 Another aim of this work is
to determine the relative yield of hyperthermal fragments, which
represents a measure of the surface accumulation of the chro-
mophore. The partitioning of the chromophore between bulk
and surface as obtained from the TOF spectra will be compared
to surface tension measurements and to our previous measure-
ments with alkyl iodide.
The detector is contained in a bakeable ultrahigh vacuum
system consisting of three differentially pumped stages.19 After
a total flight distance of 374 mm the molecules were ionized
by impact with 130 eV electrons and mass-filtered in a QMS.
A time offset of 3.9 µs ((m/e)1/2) was subtracted from the
measured TOF spectra in order to correct for the transit time of
the ions through the mass filter.19 Some of the TOF measure-
ments were carried out with a chopper wheel inserted at a
distance of 34 mm from the surface. The wheel (80 mm
diameter, one 0.7 mm slit) spins at 200 Hz and is synchronized
with the laser, producing an open duration of about 15 µs at
selectable times following the laser-surface interaction.
The accumulation of surface contaminations during the
measurements was checked by measuring the surface tension
with a De Nouy ring attached to an electronic balance (Sigma
703).20 Unchanged values before and after measurements
indicated that the surface remained clean during the measure-
ment time. Temperature-dependent measurements of the surface
tension were carried out with a Wilhelmy plate.20 Peltier
elements attached to the liquid reservoir allowed the tempera-
ture to be varied in the range 0-100 °C. The tensiometer was
enclosed in a box purged with nitrogen to avoid contamination
of the liquid by water vapor and greasy particles. The temper-
ature was changed in steps of 10 K and left to equilibrate for
20 min. No hysteresis of the surface tension was observed
between warming and cooling phases. The values obtained in
several cycles were constant within (0.2 mN/m.
2. Experimental Section
The technique used to prepare clean and continuously
renewed liquid surfaces in a vacuum has been described
elsewhere.6,12 Briefly, a rotating aluminum wheel is submersed
in a reservoir filled with the liquid. The high viscosity of the
liquid studied here required a low rotation speed of 0.2 Hz. To
improve the wetting of the wheel, the surface was roughened
and patterned with spiraling grooves. No scraper was used to
allow for molecular ordering of the surface layer of the liquid.
Solutions of the photoabsorbing species 4-iodobenzoic acid
(>98%, Fluka) in glycerol (>99.5%, Fluka) were prepared at
concentrations of 4 × 10-5 and 2 × 10-6 mol/L. The
concentrations were determined spectrophotometrically. The
extinction coefficient ꢀ of IBA in methanol is 1.7 × 104 dm3
mol-1 cm-1 at the peak of the absorption band at 250 nm and
is 4 × 103 dm3 mol-1 cm-1 at 275 nm, the excitation wavelength
3. Results
TOF Spectra. Following irradiation of the IBA/glycerol film
with low-fluence pulses (e5 mJ/cm2) at 275 nm, a signal was
observed exclusively at the mass of atomic iodine, m/e ) 127,
and molecular iodine, m/e ) 254.21 The TOF spectrum recorded
at m/e ) 127 is shown in Figure 1a for an IBA concentration
+
of 4 × 10-5 mol/L. The TOF signal of I2 (not shown) was
weaker than the I+ signal by a factor of 80 (corrected for
ionization efficiencies and I2 cracking12). Only about 1% of the
observed m/e ) 127 signal originates in cracking of I2, as
determined in a separate experiment with a pure I2 source. The
shape of the I2+ profile is similar to the I2+ spectrum as displayed

1552 J. Phys. Chem. B, Vol. 103, No. 9, 1999
Furlan
Figure 1. TOF distributions of iodine atoms (m/e ) 127) following
irradiation of liquids with iodine-containing chromophores at 275 nm
and a laser fluence of 0.6 mJ/cm2. The chromophore in spectrum a
was 4-iodobenzoic acid (4 × 10-5 mol/L) dissolved in glycerol.
Spectrum b was obtained by irradiation of a 0.1 M 1-iodooctadecane
(C18H37I) solution in squalane (C30H62). The model simulations of the
spectra, shown as solid lines, are described in the text.
Figure 2. TOF distributions of I atoms (m/e ) 127) recorded at laser
fluences between 0.05 and 20 mJ/cm2. The relative scaling of the spectra
is arbitrary, since acquisition times were adjusted depending on the
signal level. The solid lines represent Maxwell-Boltzmann distributions
with a temperature of Ttrans ) 480 K. The dotted lines in curves c and
d are shifted Maxwell-Boltzmann distributions with Ttrans ) 1300 and
2500 K and stream velocities Vs ) 500 and 1000 m/s, respectively.
in Figure 2 of ref 12, with a late onset at about 1.5 ms and a
+
slow decay. However, the I2 signal was too weak to allow
measurements with a chopper wheel. The I2 profiles could
therefore not be used for the quantitative analysis that is
presented in the next section. No signal was found at the masses
of the parent molecule IBA, m/e ) 248, the benzoic acid
fragment, m/e ) 121, or HI, m/e ) 128. The limited resolution
of the QMS in the mass range >100 amu (fwhm ≈ 2 amu)
required a calibration measurement to distinguish between I and
HI.12 From the complete overlap of the nominal m/e ) 127 peak
observed in a pure I2 beam with the m/e ) 127 peak produced
with the irradiated liquid, it was concluded that no HI is ejected
off the irradiated liquid.
In Figure 1b the TOF profile of I atoms desorbed from a 0.1
M 1-iodooctadecane (C18H37I) solution in squalane (C30H62) is
shown. This TOF profile has a later onset and a slower decay
at long flight times compared to the spectrum of the IBA/
glycerol solution (Figure 1a). Both spectra shown in Figure 1
were recorded at a low laser fluence of 0.6 mJ/cm2. The goal
of the TOF measurements presented in the following is to
explore the different shapes of profiles in spectra a and b of
Figure 1. In general, the signal decay at long flight times is a
signature of species originating from deeper liquid layers. The
late arrival time at the detector is due to diffusion of these
photoproducts to the surface followed by delayed desorption.
On the other hand, the difference in signal onsets reflects a
different degree of thermalization of the nascent fragments
before they leave the surface. If the desorbing species exhibit a
highly hyperthermal velocity distribution similar to gas-phase
photofragments, the orientation of the excited molecule in the
surface layer must have allowed collision-free or nearly colli-
sion-free desorption. The signal onset also strongly depends on
laser fluence. A series of TOF spectra of IBA in glycerol at
different laser fluences are shown in Figure 2. In the range
0.05-5 mJ/cm2 the shape of the TOF profiles is unchanged
(curves a and b of Figure 2), indicating that at these low fluences
laser-induced heating of the illuminated liquid is negligible.
However, the TOF distributions change substantially at laser
fluences above approximately 10 mJ/cm2 and become bimodal
(curves c and d of Figure 2). A fast component grows quickly
and shifts toward shorter flight times with increasing laser
fluence. This fast part can be modeled by a stream velocity Vs
superimposed on a Maxwell-Boltzmann distribution. This type
of distribution is often found in ablation studies when the
desorption flux is large enough to allow gas-phase collisions.22,23
The collisions lead to a fast particle beam directed perpendicular
and away from the surface. At the low fluences (<5 mJ/cm2)
discussed in the following, however, gas-phase collisions have
a low probability.11
From a comparison of the onset of ablative desorption
occurring at ∼10 mJ/cm2 with the theoretically expected onset,
which will be discussed in the next section, it appears that IBA
tends to accumulate at the gas-liquid interface. To test this
hypothesis, a series of dilution measurements was performed.
As shown in Figure 3, a dilution of the IBA solution by a factor
of 20 does not affect the shape of the TOF profiles, and the
signal intensity is only reduced by a factor of 2-3. If the surface

Photodissociation at Liquid Surface
J. Phys. Chem. B, Vol. 103, No. 9, 1999 1553
Figure 3. TOF spectra of iodine atoms ejected from IBA/glycerol
solutions at two different IBA concentrations: (a) 4 × 10-5 mol/L; (b)
2 × 10-6 mol/L. Both spectra were recorded at a laser fluence of 0.6
mJ/cm2.
Figure 4. (a) Chopped TOF spectra at m/e ) 127 of IBA/glycerol
recorded at three different delays between the laser shot and the chopper
opening (100, 150, and 200 µs). (b) Simulations of the chopped IBA/
glycerol spectra with Ts ) 480 K. (c) Best-fit simulations of the chopped
iodooctadecane/squalane spectra with Ts ) 298 K. The chopped profiles
of IBA/glycerol are dominated by desorption of incompletely thermal-
ized I atoms from the topmost liquid layer, whereas the iodooctadecane/
squalane system exhibits a larger contribution from late desorption,
evident as a broad shoulder at short flight times of the chopped profiles.
concentration of IBA were equal to the bulk concentration, a
reduction proportional to the dilution factor would be expected.
No signal reduction should occur if the topmost liquid layers
consisted of pure IBA. The small signal reduction observed in
our measurements clearly confirms an accumulation of IBA at
the liquid surface.
As noted above, Figure 1 reveals, apart from a later onset,
also a much slower decay of the I signal in the C18H37I/C30H62
system (Figure 1b) compared to that in IBA/glycerol (Figure
1a). The slow decay is a consequence of I atoms originating
from deeper liquid layers and desorbing at long delay times
after the laser pulse. A detailed discussion of delayed desorption
in the system C18H37I/C30H62 is given in ref 12. The fast decay
of the I signal in the IBA/glycerol spectrum (Figure 1a) suggests
that the desorption flux at long delay times is comparatively
small. This result is confirmed by a series of measurements with
a chopper wheel synchronized to the laser pulses. Chopped
spectra of I atoms from IBA are shown in Figure 4 for three
different laser-chopper delays. As described in ref 12 late
desorption of species can be distinguished by a broad shoulder
in the chopped profiles at short flight times. This feature is
completely absent in the profiles displayed in Figure 4a,
confirming that desorbing I atoms originate almost exclusively
from depths less than 1 nm from the liquid-gas interface, which
corresponds approximately to the topmost molecular layer. The
best-fit simulations of the chopped TOF profiles are shown in
parts b and c of Figure 4 for IBA/glycerol and C18H37I/C30H62,
respectively. The shapes exhibit considerable differences for the
two systems. The chopped profiles in Figure 4b consist of a
single peak at all delay times. In contrast, the simulations in
Figure 4c exhibit a shoulder at short flight times, which grows
to a second peak at large laser-chopper delays. This second
feature indicates a signal contribution from depths greater than
1 nm, which is much larger in C18H37I/C30H62 because of the
larger cage escape probability and smaller viscosity of squalane.
It must be noted that the TOF spectra of iodooctadecane/
squalane recorded at m/e ) 127 contain a contribution from HI
molecules, which is not present in the IBA/glycerol system.
However, the I and HI profiles are very similar in shape. A
detailed discussion of the shapes of chopped profiles was given
in ref 12.
Surface Tension Measurements. Surface tension measure-
ments are often used to probe the accumulation of surface-active
species at the liquid surfaces. The surface tension of IBA in
glycerol was determined at different IBA concentrations between
1 × 10-6 and 1.3 × 10-4 mol/L. The upper value corresponds
to a saturated solution at 25 °C. A constant value σ ) 62.0 (
0.2 mN/m, equal to the surface tension of pure glycerol, was
obtained in the whole concentration range. The temperature
dependence of the surface tension, shown in Figure 5, was
determined between 5 and 95 °C for 4 × 10-5 mol/L IBA
solutions in glycerol and water. The values obtained for the
pure solvents are displayed by full dots. Within the experimental
error bounds of (0.2 mN/m no difference was observed between
the IBA/glycerol solution and pure glycerol. The precision of
the measurements is mainly limited by changes in wetting of
the Wilhelmy plate over longer measurement periods. The
surface tension of the aqueous IBA solution was found to be
on average 1.0 ( 0.2 mN/m lower than that of pure water, with
the difference decreasing from 1.7 mN/m at 5 K to 0.6 mN/m
at 90 K.

1554 J. Phys. Chem. B, Vol. 103, No. 9, 1999
Furlan
4. Quantitative Analysis of TOF Spectra
The model calculations used to analyze the TOF spectra have
been described in detail elsewhere.12 Briefly, the time-dependent
concentration profiles of I and I2 are computed by numerical
integration of one-dimensional coupled differential equations,
which govern the diffusion and surface evaporation of both
species, and the kinetics of radical combination. Absorption of
laser light by IBA molecules is followed by dissociation of the
excited species and production of benzoic acid radicals and
iodine radicals at the surface and within the liquid.
I-BA + hν f I + BA
(1)
I fragments formed at the surface can be directly ejected into
the gas phase:
I + BA f I(ejected) + BA
(2)
The desorption energy of the benzoic acid radical BA is
relatively large owing to hydrogen bonding to glycerol mol-
ecules. The desorption rate of BA at 278 K is therefore expected
to be very low, in agreement with experimental results, and is
neglected in the calculation.
Radicals formed in deeper layers of the liquid are surrounded
by a cage of glycerol molecules. These radicals will recombine
geminately (reaction 3), or escape from the cage (reaction 4).
Figure 5. Surface tensions of a 4 × 10-5 mol/L solution of
4-iodobenzoic acid in glycerol and in water as a function of temperature
(crosses). The surface tensions of pure glycerol and water are also
shown (full dots). The size of the markers indicates the error bounds.
{I + BA} f IBA
(3)
(4)
{Ι + ΒΑ} f I(free) + BA(free)
of the numerical calculation, the mean diffusion length of I2 in
glycerol is (2Dt)1/2 ≈ 0.7 nm, 14 times smaller than the mean
diffusion length of I2 in squalane.
The braces indicate the liquid cage surrounding the geminate
pair. The mobile radicals (reaction 4) diffuse freely in solution
and undergo combination reactions 5-7.
The initial depth-dependent concentrations of benzoic acid
and iodine radicals are calculated from the laser fluence, the
optical density of the sample, and the escape probability for
avoiding geminate recombination k4/(k3 + k4). The evaporative
escape of I and I2 is simulated with a rate proportional to the
concentration in the surface layer and the respective diffusion
constant. The time-dependent flux of I and I2 leaving the surface
is used to calculate the total TOF signals and the TOF signals
through the chopper wheel, assuming thermal velocity distribu-
tions. A contribution of prompt, hyperthermal I atoms from a
surface layer (reaction 2) can be modeled for a given transla-
tional energy distribution.
In Figure 6 the TOF spectrum of iodine atoms is shown
together with various model simulations. In this simple calcula-
tion the liquid is subdivided into a surface layer of variable
thickness L and the bulk at depths larger than L. All I atoms
originating from the surface layer are ejected with a velocity
distribution corresponding to a surface temperature Ts, whereas
the I atoms originating from the bulk are assumed to be ther-
mally equilibrated with the surrounding solvent molecules and
desorb with the temperature of the bulk liquid Tliq ) 278 K.
Profile a in Figure 6 is calculated under the assumption that
the IBA molecules in the surface layer of thickness L ) 1 nm
are oriented with the I end pointing out of the liquid. The
translational energy Etrans of the recoiling fragments I + BA is
estimated by a soft impulsive model. According to this model,
the initial repulsive force acts only between the I and the
attached C atom, channeling m(I)/m(C) ≈ 10% of the available
energy Eavl into recoil of the I atom. With Eavl ) hν - D0(I-
BA) and a bond dissociation energy D0(I-BA) ) 276 kJ/mol
(at 298 K),25 this corresponds to an average atomic iodine speed
of about 450 m/s or a translational temperature Ttrans ) 1100
K. Although the soft impulsive model gives a lower bound for
I + I f I2
(5)
Ι + ΒΑ f IBA
(6)
(7)
ΒΑ + ΒΑ f ABBA
The radical reactions 5-7 are exothermic, have no activation
barrier, and are modeled with diffusion-limited rates ap-
proximated by the form k ≈ 8RT/(3η) dm3 mol-1 s-1 24
.
All
steric factors were arbitrarily neglected for the sake of simplicity.
The reaction scheme previously used to model the bulk
photochemistry of C18H37I in squalane was equivalent to
reactions 1-7 presented above but included a few more
reactions that can be neglected in IBA/glycerol.12 Most impor-
tantly, the formation of HI according to
{I+C18H37} f HI + C18H36
(8)
does not occur with the reaction pair {I + BA}. In contrast to
the exothermic R-H abstraction of reaction 8, the â-H abstraction
from BA is endothermic. The simulation only considers diffu-
sion of I and Ι2 in the direction perpendicular to the surface,
since the illuminated surface area (2 × 6 mm2) is much larger
than the liquid depth relevant in the TOF experiment. The I
and Ι2 desorption flux from depths greater than or equal to 1
µm is negligible, since the large viscosity of glycerol (∼5000
cP at 5 °C) prevents larger diffusion lengths on the millisecond
time scale of the experiment. The diffusion constants of I and
I2 are estimated to be 1 × 10-12 and 2 × 10-13 m2/s,
respectively, on the basis of scaling of published values for I2
in hexane with viscosity of the solvent and on hydrodynamic
radii and mass factors of the solute.24 Within a 1 µs time step

Photodissociation at Liquid Surface
J. Phys. Chem. B, Vol. 103, No. 9, 1999 1555
Figure 6. Comparison of experimental TOF spectrum of iodine atoms
(4 × 10-5 mol/L IBA in glycerol, 0.6 mJ/cm2) with model simulations
for various temperatures of the surface layer Ts. A surface layer
thickness of 1 nm is assumed. (a) Ts ) 1100 K, corresponds to direct
ejection of I with energy partitioning according to the soft impulsive
model. (b) Ts ) 480 K yields the best agreement with the experiment.
(c) Isothermal desorption with Ts equal to the bulk temperature before
irradiation, Tliq ) 278 K. (d) The contribution to simulation b of I atoms
originating from depths greater than 1 nm (multiplied by 10).
Figure 7. Dependence of translational temperature of iodine fragments
on the laser fluence. Below ∼5 mJ/cm2, a constant temperature of 480
( 30 K (∆T ) 200 K) is found. At higher fluences the temperature
rises steeply, indicating explosive desorption. The temperature rise due
to laser-induced thermal desorption is calculated for (a) an ideal solution
of IBA in glycerol with no surface accumulation of IBA and (b) a
surface monolayer of IBA.
Etrans, the calculated TOF distribution displayed in Figure 6a
has a too early onset.
originating from depths greater than 1 nm. This contribution to
the total I yield is very small because of the low mobility of I
atoms in glycerol. However, pesc strongly influences the I/I2
signal ratio. For pesc ) 0 the yield of I2 is zero, since all I
fragments from the topmost layer are directly ejected into
vacuum and deeper layers give no contribution to the desorption
flux. For pesc ) 0.1 the I/I2 ratio is 5.8. Good agreement with
the experimental value of 80 ( 20 (corrected) is obtained with
pesc ) 0.005.
The TOF distribution in Figure 6c is obtained with a surface
temperature equal to the bulk liquid temperature before irradia-
tion, Ts ) Tliq ) 278 K. Such a distribution has been found for
the C18H37I/squalane system.12 The lack of any hyperthermal
component was interpreted by a preferential orientation of the
chromophores with the iodine end poining into the liquid.
Comparison of the measured spectrum in Figure 6a with the
simulation in Figure 6c shows that the experimental profile has
an earlier onset than predicted by the isothermal simulation at
278 K.
5. Discussion
The best agreement for the IBA/glycerol system was found
with a slightly hyperthermal distribution with a temperature for
surface layer I atoms Ts ) 480 ( 20 K (Figure 6b). The I atoms
from the surface layer are thus ejected before reaching full
thermalization with the surrounding liquid. The thickness of the
surface layer was varied between 0.1 and 10 nm. Within this
range the thickness has only a small effect on the TOF profiles,
since even a 0.2 nm surface layer contributes a larger fraction
to the total flux than the bulk. The surface layer thickness was
fixed at an estimated value of one molecular IBA monolayer
of L ) 1 nm. The contribution to the best-fit TOF spectrum
(Figure 6b) from depths greater than 1 nm is depicted by the
dashed area in Figure 6d. Since this diffusive contribution to
the total flux is negligibly small compared to the surface layer
contribution, the TOF profiles of I atoms from IBA/glycerol
can be fitted in a good approximation to a single Maxwell-
Boltzmann distribution at T ) 480 K.
Surface Partitioning of IBA. An aim of this study was to
determine the surface concentration of IBA in glycerol using
TOF spectroscopy. The TOF results are now compared to the
surface tension measurements to elucidate the surface sensitivity
of both methods. In an ideal solution the surface concentrations
of IBA and glycerol should exactly mirror the concentrations
in the bulk liquid. Our TOF results, however, cannot be
explained by a homogeneous solution but rather require a surface
excess of IBA. The translational temperatures obtained from
fits of the TOF distribution (Figure 2) allow an estimate of the
density of IBA molecules at the surface. Figure 7 shows the
translational temperatures obtained at different laser fluences
in the range 0.05-20 mJ/cm2. The temperature at fluences below
about 5 mJ/cm2 was found to be constant at 480 ( 20 K. Above
∼10 mJ/cm2 the temperature rises steeply. The velocity distribu-
tion changes from a Maxwell-Boltzmann to shifted Maxwell-
Boltzmann distribution with a stream velocity Vs, indicating
explosive desorption. The dashed lines in Figure 7 display the
expected translational temperature of I atoms, assuming that
the laser energy is immediately converted into heat within the
irradiated liquid volume. It is further assumed that the energy
The cage escape probability pesc was varied in the range
0-0.1. For the IBA/glycerol system caging was found to have
almost no effect on the shape of the calculated TOF profiles,
since pesc only affects the desorption yield of those I atoms

1556 J. Phys. Chem. B, Vol. 103, No. 9, 1999
Furlan
losses by thermal conductivity and photoejection are negligible
during the 6 ns irradiation time and that all I atoms desorb with
the temperature of the heated liquid. The temperature rise is
calculated by
the components are not sufficiently different. The TOF/QMS
method, in contrast, is highly sensitive to chromophoric species
in the surface layer. This is partly due to an efficient discrimina-
tion of the bulk contribution by caging and slow diffusion. A
similar difference in sensitivity was found in a comparison of
surface excess data from sum-frequency mixing and surface
tension measurements.7
hν
∆Ts ) ∆Tliq
)
(9)
cVV
Surface Orientation of IBA. According to the model
calculation shown in Figure 6, the velocity distribution of the
ejected I atoms can be described by a Maxwell-Boltzmann
distribution with a translational temperature Ttrans ≈ 500 Κ, 200
Κ above the temperature of the surrounding liquid and 600 K
colder than expected for direct ejection of I from the topmost
layer according to a soft impulsive model. The constant value
Ttrans ≈ 500 Κ found for laser fluences less than or equal to 5
mJ/cm2 deviates from the prediction for laser-induced thermal
desorption from an IBA-covered surface (see upper dashed line
in Figure 7). The temperature of the desorbate would be
expected to rise linearly with laser pulse energy in the case of
thermal desorption. The observed constant translational tem-
perature rather implies incomplete thermalization of the I atoms.
The soft impulsive model gives a lower bound for the recoil
energy Etrans of the fragments. Since the observed temperature
is below this lower bound, I atoms must lose part of their kinetic
energy before leaving the surface. This excludes a preferential
surface orientation of IBA molecules with the I atom sticking
out of the liquid, because such an orientation allows no energy
loss. An orientation of the C-I bond parallel to the surface
would ensure at least one collision of the I atoms before
desorption. Since the efficiency of translational energy transfer
is maximized when the masses are matched, the strongest
cooling of an I fragment would be achieved by a collision with
an I atom bound to a neighboring IBA molecule. An important
aspect of surface ordering may be the formation of IBA dimers
at the gas-liquid interface. The dimerization of benzoic acid
is known to be almost complete in organic solutions27 and in
the solid.28,29 By dimerization, the IBA molecules lose their
amphiphilic character, which diminishes the orientational forces
acting on the surface molecules. This may enhance the prefer-
ence for surface accumulation of the apolar dimers. Furthermore,
dimerization could modify the nature of the electronically
excited state and thereby open new channels for loss of
translational energy. Corrugation of the surface could also play
an important role in translational energy loss, if the gas-liquid
interfacial width is similar to or larger than the molecular
dimensions. TOF measurements at different temperatures are
under way to distinguish between the effect of surface roughness
and the average orientation of a single molecule at the surface.
The orientation could be changed by using benzoic acids
substituted with iodine at different ring positions.
where hν is the fraction of the pulse energy entering the liquid,
cV is the heat capacity of glycerol (3.0 J/K/cm3), and V is the
liquid volume absorbing 1/e of the pulse energy.26 A reflection
loss of 82% of the laser energy due to the grazing incidence
angle is taken into account. The lower line corresponds to an
ideal solution of IBA in glycerol with no surface accumulation
of IBA (4 × 10-5 mol/L), and the upper line corresponds to a
surface fully covered with IBA. The upper line is in good
agreement with the experimental data, implying a strong surface
excess of IBA in glycerol. The deviations between the upper
line and the data points are smaller than the error bounds in
laser fluence and temperature.
Similarly, the onset of explosive desorption can be estimated
from the product of the absorption cross section of IBA at 275
nm (1.5 × 10-17cm2), the laser fluence (photons/cm2), and the
surface concentration of IBA. Explosive desorption occurs when
a substantial fraction of a monolayer is desorbed per laser
pulse.22 The collisions of desorbing species in the gas phase
result in a shifted Maxwell-Boltzmann velocity distribution,
as observed at fluences above ∼10 mJ/cm2 (curves c and d of
Figure 2). In a perfectly mixed 4 × 10-5 mol/L IBA solution
at 10 mJ/cm2 only a fraction of 2 × 10-6 of a monolayer (ML)
desorbs per laser pulse, far too low for explosive desorption.
However, with a surface fully covered with IBA (equivalent to
a local concentration of 4 mol/L), 0.2 ML is expected to desorb
per laser pulse. This fraction is sufficient for ablative desorption,
in agreement with the observed onset of shifted Maxwell-
Boltzmann velocity distributions.
As noted above, further evidence for a surface excess of IBA
was obtained in the dilution study (Figure 3). The reduction of
the IBA concentration by a factor of 20 reduces the atomic
iodine signal only by a factor of 2-4. A quantitative analysis
of the surface accumulation is not attempted at this stage, since
the observed concentration dependence may be affected by the
film preparation technique (stress-induced fragmentation of the
monolayer into islands20) as well as limited reproducibility when
the liquid sample was exchanged.
In contrast to the TOF data, measurements of the surface
tension σ show no evidence of surface partitioning of IBA. At
IBA concentrations ranging from 10-6 to 10-4 mol/L and at
temperatures from 5 to 95 °C no deviation of σ was found from
the value of pure glycerol, as displayed in Figure 5. The surface
tension of water is slightly reduced by IBA (at a concentration
near saturation). Since glycerol is less polar than water, it is
reasonable to assume that the surface excess free energy ∆Gexcess
(the extra energy the system gains by partitioning IBA to the
solution interface) of IBA/water is larger than that of IBA/
glycerol. The surface excess of IBA in glycerol is thus expected
to be smaller than in water. Compared to σ values of typical
surface-active species, which are a factor of ∼3 lower than σ
of pure water, an IBA monolayer probably has a surface tension
similar to that of pure glycerol. Therefore, the effect of adding
10-4 mol/L IBA to glycerol may be below the detection limits
of (0.2 mN/m of our measurement. It is concluded that
tensiometry is a rather insensitive technique for detecting low
concentrations of dissolved species that are not strongly surface-
active. It is particularly insensitive if the surface tensions of
In summary, the TOF data presented here are insufficient to
determine the orientation of the surface molecules. Only if the
fragments have a velocity distribution similar to that in gas-
phase photodissociation, indicating that desorption occurs
without inelastic collisions, could one infer an orientation of
the excited molecules. This situation has been found in some
recent photochemical studies of solid films. Polanyi and co-
workers investigated the photolysis of HI adsorbed on LiF and
NaF substrates at submonolayer coverage.18 They observed the
ejection of H atoms with a kinetic energy distribution strongly
reminiscent of the gas-phase distribution and a background of
inelastically scattered H atoms. Methyl nitrite and tert-butyl
nitrite adsorbed on solid substrates have also been studied in
detail. The velocity distribution of the photodesorbed NO was

Photodissociation at Liquid Surface
J. Phys. Chem. B, Vol. 103, No. 9, 1999 1557
TABLE 1: Summary of Results from TOF Analysis for the
Systems IBA/Glycerol and C18H37I/Squalane
orientation with the I end pointing into the liquid. After a
diffusion length of ∼10 nm to reach the surface, these I atoms
are therefore fully thermalized.
IBA/glycerol
I, I2
C18H37I/squalane
I, HI, I2
The onset of explosive desorption from IBA/glycerol and the
concentration dependence of the TOF signal are interpreted by
a surface excess of IBA. The surface excess is not apparent
from tensiometric measurements. Further TOF experiments are
under way to determine the surface accumulation and orientation
of IBA molecules quantitatively. Second harmonic generation
or sum frequency mixing are also considered as additional
methods for measuring the surface orientation of IBA.
photoproducts detected
by TOF/QMS
product ratio (corrected)
80:1
480 K
5:10:1
298 K
T
trans of I atoms
(at <10 mJ/cm2)
cage escape probability
origin of 99% of I flux
(depth below surface)
concentration dependence
of m/e ) 127 signal
0.5%
<1 nm
2%
∼200 nm
strongly sublinear
linear
Acknowledgment. This work was supported by the Swiss
National Foundation. The author thanks Dr. G. E. Hall for his
help with the spectral simulations, Dr. R. T. Carter for carefully
reading the manuscript, and Professor J. R. Huber for his
generous support.
found to depend on the film thickness and the nature of the
substrate and varies from gas-phase behavior to much slower
velocity distributions.17,30,31 In contrast to these studies, the
nature of the substrate is unimportant in this work, since the
liquid film has a thickness of several thousand molecular layers.
References and Notes
6. Summary
(1) Crutzen, P. J.; Arnold, F. Nature 1986, 324, 651.
(2) Prather, M. J. Nature 1992, 355, 534.
Compared to solid surfaces, a rather limited number of
methods is available for the study of surfaces of liquids, despite
their importance in many natural reactions. This work reports
on an application of TOF/QMS to the study of a liquid surface
and its photochemistry in high vacuum. TOF spectra of
photofragments ejected from the liquid surface into the gas phase
are analyzed using model calculations, which take into account
photolysis of the solute, kinetics of the radicals in the liquid,
diffusion, and surface evaporation of the volatile photoproducts,
and allow for direct ejection of hyperthermal photofragments
into the gas phase. Table 1 summarizes the analysis of the TOF
results of IBA/glycerol, together with the previously obtained
data on C18H37I/squalane. In both systems the measurements
were taken under low-fluence conditions where laser-induced
heating is negligible.
(3) Rodriguez, J. M.; Ko, M. K. W.; Sze, N. D. Nature 1991, 352,
134.
(4) Turro, N. J. Modern molecular photochemistry; University Science
Books: Mill Valley, CA, 1991.
(5) Shen, Y. R. Nature 1989, 337, 519.
(6) Saecker, M. E.; Nathanson, G. M. J. Chem. Phys. 1993, 99, 7056.
(7) Baldelli, S.; Schnitzer, C.; Shultz, M. J.; Campbell, D. J. J. Phys.
Chem. B 1997, 101, 4607.
(8) Baldelli, S.; Schnitzer, C.; Shultz, M. J.; Campbell, D. J. J. Phys.
Chem. B 1997, 101, 10435.
(9) Radu¨ge, C.; Pflumio, V.; Shen, Y. R. Chem. Phys. Lett. 1997, 274,
140.
(10) Zhu, X.-Y. Annu. ReV. Phys. Chem. 1994, 45, 113 and references
therein.
(11) Furlan, A. Chem. Phys. Lett. 1997, 275, 239.
(12) Furlan, A.; Hall, G. E. J. Chem. Phys. 1998, 109, 10390.
(13) Lednovich, S. A.; Fenn, J. B. AIChE J. 1977, 23, 454.
(14) Dzvonik, M.; Yang, S.; Bersohn, R. J. Chem. Phys. 1974, 61, 4408.
(15) Kawasaki, M.; Lee, S. J.; Bersohn, R. J. Chem. Phys. 1977, 66,
2647.
The flux of desorbing species after a laser pulse consists of
both cases of atomic and molecular iodine, and HI is an addi-
tional photoproduct only produced in the alkyl iodide solution.
The different yields of the volatile photoproducts reflect
differences in the set of bulk radical reactions and the viscosity
of the solvent. The cage escape probabilities of 0.5% (I + BA)
and 2% (C18H37 + I) are in agreement with the data of Noyes
et al.32 who studied I2 photolysis in a varietiy of solvents and
found decreasing cage escape with increasing solvent viscosities.
Iodine atoms desorbing off IBA/glycerol originate almost
entirely from the topmost liquid layer (<1 nm), whereas the
TOF spectra of alkyl iodide solutions probe a depth of ∼100
nm. The high surface sensitivity of the TOF/QMS method in
the IBA/glycerol system is due to efficient caging and the low
mobility of photofragments in glycerol.
Velocity distributions obtained from the TOF spectra indicate
that the I atoms desorbing from IBA/glycerol are only partially
thermalized by collisions with neighboring surface molecules
(Ttrans ≈ 480 Κ), whereas the velocity distribution of I atoms
from C18H37I/squalane can be described by the temperature of
the liquid (Ttrans ) 298 K). This indicates that the IBA molecules
at the liquid surface are oriented in such a way that on average
the I fragments undergo only few collisions before leaving the
liquid, whereas C18H37I molecules have a preferential surface
(16) Cheng, P. Y.; Zhong, D.; Zewail, A. H. Chem. Phys. Lett. 1995,
237, 399.
(17) Simpson, C. J.; Griffiths, P. T.; Towrie, M. Chem. Phys. Lett. 1995,
234, 203.
(18) Barclay, V. J.; Hung, W.-H.; Keogh, W. J.; Ku¨hnemuth, R.; Polanyi,
J. C.; Jennison, D. R.; Li, Y. S. J. Chem. Phys. 1996, 105, 5005.
(19) Felder, P. Chem. Phys. 1990, 143, 141.
(20) Gaines, G. L. Insoluble Monolayers at the Liquid Gas Interface;
Wiley: New York, 1966.
(21) At high laser fluences (e5 mJ/cm2) a signal also appeared at m/e
) 121 (BA), 248 (IBA), and 92 (glycerol).
(22) Kelly, R. J. Chem. Phys. 1990, 92, 5047.
(23) Miller, D. R. In Atomic and Molecular Beam Methods; . Scoles,
G., Ed.; Oxford University Press: Oxford, 1988.
(24) Atkins, P. W. Physical Chemistry; Oxford University Press: Oxford,
1998.
(25) McMillen, D. F.; Golden, D. M. Annu. ReV. Phys. Chem. 1982,
33, 493.
(26) Volume is given by the spot size (0.12 cm2) and the 1/e penetration
depth d ) 1/(ꢀc), where ꢀ is the extinction coefficient and c the local IBA
concentration.
(27) Davis, J. C.; Pitzer, K. S. J. Phys. Chem. 1960, 64, 886.
(28) Nagaoka, S.; Terao, T.; Imashiro, F.; Saika, A.; Hirota, N.; Hayashi,
S. Chem. Phys. Lett. 1981, 80, 580.
(29) Neukomm, M. A. Ber. Bunsen-Ges. Phys. Chem. 1998, 102, 325.
(30) Fieberg, J. E.; Szulczewski, G. J.; White, J. M. Chem. Phys. Lett.
1998, 290, 268.
(31) Jenniskens, H. G.; Philippe, L.; Essenberg, W. v.; Kadodwala, M.;
Kleyn, A. W. J. Chem. Phys. 1998, 108, 1688.
(32) Booth, D.; Noyes, R. M. J. Am. Chem. Soc. 1960, 82, 1868.