Vibrationally mediated photolysis dynamics of H20 in the VOH=3 manifold Far off resonance photodissociation cross sections and OH product state distributions

Article abstract of DOI:10.1063/1.478764

Vibrationally mediated photodissociation dynamics of water on the first excited electronic state surface (<*>) has been studied with slit jet-cooled H20 prepared in the complete polyad of ν_OH=3 overtone stretch levels (<*>03⁺>, <*>12⁺>, <*>12^->, and <*>03^->). (Notationally, <*>n₁n_+/-²> refers to symmetric/antisymmetric combinations of local mode OH stretch excitation, roughly corresponding to n₁ and n₂ quanta in the spectator and photolyzed OH bond, respectively.) At 248 nm photolysis wavelength the Condon point for bond cleavage occurs in the classically forbidden region, primarily sampling highly asymmetric H+OH exit valley geometries of the potential energy surface. Rotational, vibrational, spin orbit, and lambda doublet distributions resulting from this far off resonance photodissociation process are probed via laser induced fluorescence, exploiting the high efficiency laser excitation and light collection properties of the slit jet expansion geometry. Only vibrationally unexcited OH products are observed for both <*>12^+/-> and <*>03^+/-> initial excitation of H20, despite different levels of vibration in the spectator OH bond. This is in contrast with near-resonance vibrationally mediated photolysis studies by Crim and co-workers in the <*>04^->and <*>13^-> manifold, but entirely consistent with theoretical predictions from a simple two-dimensional quantum model. Photolysis out of the rotational ground H20 state (i.e., J_KaKc=0₀₀) yields OH product state distribution that demonstrate remarkable insensitivity to the initial choice of H2O vibrational stretch state, in good agreement with rotational Franck-Condon models. However, this simple trend is not followed uniformly for rotationally excited H2O precursors, which indicates that these Franck-Condon models are insufficient and suggests that exit channel interactions do play a significant role in photodissociation dynamics of H2O at the fully state-t-state level.

Full text of DOI:10.1063/1.478764

Vibrationally mediated photolysis dynamics of H 2 O in the v OH =3 manifold: Far off
resonance photodissociation cross sections and OH product state distributions
Ondrej Votava, David F. Plusquellic, and David J. Nesbitt
Citation: The Journal of Chemical Physics 110, 8564 (1999); doi: 10.1063/1.478764
View online: http://dx.doi.org/10.1063/1.478764
View Table of Contents: http://scitation.aip.org/content/aip/journal/jcp/110/17?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Photodissociation of N 2 O: Energy partitioning
J. Chem. Phys. 135, 024311 (2011); 10.1063/1.3602324
Ab initio potential energy surfaces, total absorption cross sections, and product quantum state distributions for
the low-lying electronic states of N 2 O
J. Chem. Phys. 122, 054305 (2005); 10.1063/1.1830436
Dissociation dynamics of thiolactic acid at 193 nm: Detection of the nascent OH product by laser-induced
fluorescence
J. Chem. Phys. 120, 6964 (2004); 10.1063/1.1667878
Spectrum and vibrational predissociation of the HF dimer. II. Photodissociation cross sections and product state
distributions
J. Chem. Phys. 119, 286 (2003); 10.1063/1.1577112
Photodissociation dynamics of acetylacetone: The OH product state distribution
J. Chem. Phys. 110, 11850 (1999); 10.1063/1.479126
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.252.67.66 On: Tue, 23 Dec 2014 02:43:26

JOURNAL OF CHEMICAL PHYSICS
VOLUME 110, NUMBER 17
1 MAY 1999
Vibrationally mediated photolysis dynamics of H2O in the vOH؍3 manifold:
Far off resonance photodissociation cross sections and OH product
state distributions
Ondrej Votava, David F. Plusquellic,a) and David J. Nesbittb)
JILA, University of Colorado and National Institute of Standards and Technology,
and Department of Chemistry and Biochemistry, University of Colorado, Boulder, Colorado 80309-0440
͑Received 15 December 1998; accepted 2 February 1999͒
Vibrationally mediated photodissociation dynamics of water on the first excited electronic state
˜
surface (A) has been studied with slit jet-cooled H2O prepared in the complete polyad of
vOH
ϭ3 overtone stretch levels (
͉
03ϩ
,
͉
12ϩ
,
͉
12Ϫ , and
͉
03Ϫ ). ͑Notationally,
͉
n1nϮ refers to
͘
͘
͘
͘
͘
2
symmetric/antisymmetric combinations of local mode OH stretch excitation, roughly corresponding
to n1 and n2 quanta in the spectator and photolyzed OH bond, respectively.͒ At 248 nm photolysis
wavelength the Condon point for bond cleavage occurs in the classically forbidden region, primarily
sampling highly asymmetric HϩOH exit valley geometries of the potential energy surface.
Rotational, vibrational, spin orbit, and lambda doublet distributions resulting from this ‘‘far off
resonance’’ photodissociation process are probed via laser induced fluorescence, exploiting the high
efficiency laser excitation and light collection properties of the slit jet expansion geometry. Only
vibrationally unexcited OH products are observed for both
H2O, despite different levels of vibration in the spectator OH bond. This is in contrast with
‘‘near-resonance’’ vibrationally mediated photolysis studies by Crim and co-workers in the
04Ϫ
and
13Ϫ manifold, but entirely consistent with theoretical predictions from a simple
two-dimensional quantum model. Photolysis out of the rotational ground H2O state ͑i.e., JK K
͉ ͉
12Ϯ and 03Ϯ initial excitation of
͘ ͘
͉
͘
͉
͘
a
c
ϭ000) yields OH product state distributions that demonstrate remarkable insensitivity to the initial
choice of H2O vibrational stretch state, in good agreement with rotational Franck–Condon models.
However, this simple trend is not followed uniformly for rotationally excited H2O precursors, which
indicates that these Franck–Condon models are insufficient and suggests that exit channel
interactions do play a significant role in photodissociation dynamics of H2O at the fully state-to-state
level. © 1999 American Institute of Physics. ͓S0021-9606͑99͒00517-6͔
I. INTRODUCTION
H–OH bond breaking dynamics from fully labeled quantum
states and the subsequent evolution of excess energy into
translational, rotational, vibrational, and electronic degrees
of freedom of the H and OH products.
As one of the simplest paradigms of chemical bond
breaking, the dynamics of photodissociation has long been
an important focus of chemical physics research, both from
experimental and theoretical points of view.1–12 The photo-
dissociation of H2O in its first absorption band, a classic yet
tractible triatomic system, has received particularly extensive
attention over the past decade.1–12 Stimulated in large mea-
sure by close interaction between theory and experiment,
H2O has effectively become the benchmark system for direct
unimolecular dissociation on a single repulsive electronic po-
tential energy surface ͑PES͒. As one example of astrophysi-
cal significance, H2O photolysis has been speculated to be
responsible for generating ⌳-doublet population inversions
evident in interstellar OH maser emission phenomena.3,4 Of
special importance in promoting strong interaction between
theory and experiment has been the development of multiple
resonance methods3,5–12 to control the initial photolysis state
of H2O. Specifically, these methods permit investigation of
In the pioneering vibrationally mediated photolysis ex-
periments of Andresen and co-workers and Crim and co-
workers, a single quantum state of H2O is selected by near
infrared ͑IR͒ or visible overtone excitation prior to ultravio-
let ͑UV͒ dissociation and the distribution of final quantum
states is probed via laser induced fluorescence ͑LIF͒ on the
OH product.3,9–12 Experimental access to such levels of
state-to-state detail is greatly facilitated in H2O for several
reasons. First of all, the large energy spacing in the OH
stretch manifold in water (⌬␯Ϸ3700 cmϪ1) permits isolated
quantum states of H2O at chemically significant energies to
be accessed via low overtone vibrational pumping. In con-
junction with a 248 nm UV photolysis wavelength, this per-
mits rigorously background free cleavage of the H–OH bond
from a specific rovibrational state. Second, the first excited
electronic state of H2O is purely repulsive, therefore excita-
tion results in rapid dissociation into photoproducts formed
exclusively in the lowest electronic states, i.e., H(1S) and
OH(2⌸).4,13 Third, outside of fine structure interactions,
there are no curve crossings with this lowest surface to form
a͒
Present address: National Institute for Standards and Technology Gaithers-
burg, MD 20899.
b͒
Staff member, Quantum Physics Division, National Institute of Standards
and Technology; electronic mail: djn@jila.colorado.edu
0021-9606/99/110(17)/8564/13/$15.00
8564
© 1999 American Institute of Physics
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.252.67.66 On: Tue, 23 Dec 2014 02:43:26

J. Chem. Phys., Vol. 110, No. 17, 1 May 1999
Votava, Plusquellic, and Nesbitt
8565
˜
exit channel portion of the A PES. Alternatively stated, Fig.
1͑a͒ illustrates excitation in the Franck–Condon region, pre-
dominantly sampling the classically allowed portion of the
vibrational wave function, while Fig. 1͑b͒ illustrates non-
Franck–Condon excitation where the classically forbidden
wings of the vibrational wave function are primarily respon-
sible for the absorption event.
˜
Inspection of the A state PES reveals several features
that are unique to the H2O system and simplify the dissocia-
tion dynamics considerably. First of all, as the dissociating
O–H bond length increases, both the equilibrium bond dis-
tance and the diatomic force constant of the spectator OH
change only slightly. Consequently, to a first approximation,
the photodissociation process is anticipated to be vibra-
tionally adiabatic, i.e., conserving the number of quanta of
initial H2O vibrational excitation in the spectator OH bond.
Indeed, this is the major reason why H2O has proven to be
such an excellent system for mode selective photochemistry,
i.e., controlled UV cleavage of a specific molecular bond by
vibrational excitation. Classic studies by Crim and
co-workers,12 for example, have demonstrated that photoly-
FIG. 1. Pictoral representation of vibrationally mediated photodissociation
of H2O. The photodissociation event is initiated by UV laser excitation from
˜
a specific laser selected overtone level of ground electronic state X and the
˜
repulsive excited electronic state A. The energy of the photolysis photon
sis of the ͉
04Ϫ state of H O ͑i.e., the antisymmetric combi-
͘
2
˜
determines the regions of the A PES sampled in the photodissociation pro-
cess. At higher photon energies ͑a͒, regions near the saddle point are ener-
getically accessed, while at lower photon energies ͑b͒, only potential surface
regions far out in the exit channel are sampled significantly.
nation of local mode states with zero and four quanta of OH
stretching excitation, respectively͒ yields OH fragments pre-
dominantly in the OHϭ0 vibrational state, whereas mostly
v
OHϭ1 is produced from photolysis of the ͉
13Ϫ state.
v
͘
higher electronic states such as OH(2⌺) manifold, making
complications due to nonadiabatic processes unlikely. Fi-
nally, the nascent distributions of vibrational, rotational, spin
orbit, and lambda doublet states of OH can be detected with
high experimental sensitivity via LIF.14,15
These experiments indicate that the vibrationally tagged OH
bond with higher local mode vibrational excitation dissoci-
ates preferentially, while vibrational excitation in the remain-
ing OH bond is adiabatically conserved during the dissocia-
tion.
From a theoretical perspective, the electronic structure of
H2O is accessible to high level ab initio calculations and thus
a reliable PES for the repulsive electronic state is available.16
At the same time, dynamics capabilities have progressed to a
point where fully rigorous quantum dynamics of triatomic
photodissociation phenomena can now be calculated. Indeed,
quantum time independent17–19 and wave-packet propagation
calculations20–22 have been successfully performed on H2O
photolysis with few or no dynamical approximations. This
special combination of capabilities makes possible unprec-
edentedly refined investigations of the triatomic photodisso-
ciation event, providing explicit theoretical predictions for
experimental observables such as photolysis cross sections or
product state distributions.
The vibrationally mediated photodissociation process is
schematically illustrated in Fig. 1. An IR photon excites a
specific rovibrational state on the ground electronic PES, fol-
lowed by selective UV excitation to the excited electronic
state and prompt dissociation to products. As is especially
important in the present studies, the choice of IR and UV
The partial cross sections for different OH vibrational
states have been calculated for several initial vibrational
states of H2O including ͉ ͉
04Ϫ and 13Ϫ using both time
͘ ͘
independent19 and time dependent wave-packet propagation
methods,21,22 with results in nearly quantitative agreement
with experimental data. However, these calculations also in-
dicate that the simple physical picture described above for
preferential bond cleavage and vibrational adiabaticity is
valid only for a range of photolysis energies near the saddle
˜
point between the two H–OH and HO–H channels on the A
state PES. Thus, at low photolysis energies it has been
suggested21 that production of the ground vibrational states
of the OH should become the dominant channel regardless of
initial H2O vibrational state, even though higher vibrational
states of OH may still be energetically accessible. This can
be directly tested by vibrationally mediated photolysis of
H2O at lower overtone levels and longer photolysis wave-
lengths, which forms one focus of the present studies.
The rotational distributions of the OH fragment also pro-
vide detailed information on the photodissociation process
and opportunities for comparison between experiment and
theoretical predictions. From ab initio calculations, the equi-
˜
photon energies qualitatively controls the regions of the A
PES sampled in the photodissociation process. For example,
‘‘on-resonance’’ excitation ͓marked ͑a͒ in Fig. 1͔ reflects a
situation where the photons provide enough energy to reach
the saddle point region, whereas for ‘‘far off resonance’’
excitation at longer wavelengths ͓marked ͑b͒ in Fig. 1͔, the
initial H2O wave function is localized to the lower energy
˜
˜
librium H–O–H bond angle for the A state is similar to the X
state and does not change considerably as the dissociating
OH bond is stretched.16 As a result, the OH fragment is
expected to experience only minor torques while rolling out
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.252.67.66 On: Tue, 23 Dec 2014 02:43:26

8566
J. Chem. Phys., Vol. 110, No. 17, 1 May 1999
Votava, Plusquellic, and Nesbitt
the exit channel, i.e., qualitatively consistent with the rela-
tively cold distribution of rotational states experimentally ob-
served. It is worth noting that this is not simply due to ener-
getics; the rotational states populated reflect only Շ1%–2%
of the total excess energy available from photolysis. More
quantitatively, detailed theoretical studies of rotational distri-
butions have been carried out by Schinke, Engel, and
Staemmler,18 based on the rotational Franck–Condon model
originally developed by Balint-Kurti.23 In this model, all
such ‘‘exit channel interactions’’ are explicitly neglected, for
example, those arising from differences in ground and elec-
tronically excited state bending potentials, bend–stretch or
rotation–vibration coupling, as well as recoil torques be-
tween the receding H and OH fragments. In this Franck–
Condon approximation, the rotational distributions are pre-
dicted simply by expanding the initial rovibrational bending
wave function of H2O in a complete set of free rotational
states of the OH fragment. Despite the simplicity of these
assumptions, this model has yielded surprisingly reasonable
agreement with experiment in many previous studies.3,10,12
However, recent vibrationally mediated photodissociation
work of Brouard and Langford for both H2O and HOD has
begun to reveal significant discrepancies from these rota-
tional Franck–Condon predictions, which have therefore
been modified to incorporate exit channel interactions such
as dynamical torques on the OH fragment during the disso-
ciation event.6,7
FIG. 2. Schematic of the experimental apparatus. Jet-cooled H2O molecules
are introduced to the photolysis and probe region via slit supersonic expan-
sion. The molecular beam is intersected along the slit axis by a series of
three laser pulses. ͑1͒ A single rovibrational state is selected with a single
model OPO operating in the vOHϭ3←0 overtone region. ͑2͒ The selected
state is photolyzed by an UV pulse at 248 nm wavelength temporally coin-
cident with the IR pulse. ͑3͒ The OH products are probed via LIF at 308 nm
after a 200 ns time delay, with the LIF signal detected by a PMT normal to
the slit expansion plane.
interpreted by comparison with results from a simple two-
dimensional ͑2D͒ Franck–Condon analysis. In Sec. IV, we
present OH rotational, spin orbit and lambda doublet distri-
butions for photolysis at 248 nm and make comparisons with
the results of previous initial rotational quantum state re-
solved photolysis studies in the OHϭ1, 4, and 5 manifolds.
v
For the rotationless (000) state, these comparisons under-
score the remarkable insensitivity of OH rotational state dis-
tributions to initial H2O vibrational state. For higher initial
rotational states, however, these results reveal a more dra-
matic and selective dependence on the H2O vibrational state,
which suggest a more complicated role of exit channel inter-
actions in the photolysis event. The conclusions of the article
are summarized in Sec. V.
In this work we investigate the vibrationally mediated
photodissociation dynamics of OHϭ3 excited H2O at 248
v
nm. The combination of second overtone vibrational excita-
tion in water combined with a 248 nm photodissociation
wavelength yields a total photolysis energy that is 4500
cmϪ1 below the saddle point on the A PES, i.e., sampling the
˜
far off resonance regime. From a semiclassical Condon point
perspective, this corresponds to photolysis from a highly
asymmetrical H2O molecule, with one bond differentially
elongated by 0.4 Å and consequently strongly skewed to-
ward the exit channel region of the upper surface. One focus
of this work is to determine the OH product vibrational dis-
tributions as a function of the initial vibrational state of
II. EXPERIMENT
The experimental techniques used in the present study
for photolysis of OHϭ3 excited H2O are based on the vi-
v
brationally mediated photodissociation methods of Hausler,
Andresen, and Schinke3 and Crim and co-workers,10,12 modi-
fied for enhanced sensitivity in a low density supersonic jet
environment. The key components of the experimental appa-
ratus are summarized in Fig. 2. A slit supersonic expansion
is used to cool H2O molecules into the lowest rotational lev-
els allowed by nuclear spin statistics, i.e., JK K ϭ000 ͑para͒
H2O
(
͉
03ϩ
,
͉
03Ϫ
,
͉
12ϩ
,
͉
12Ϫ
)
as well as relative
͘
͘
͘
͘
photolysis cross sections out of these vibrationally mediated
target states. A secondary focus is to provide OH rota-
tional, spin orbit, and lambda doublet distributions for
248 nm photolysis from selected H2O intermediate rotational
a
c
(JK K ϭ000,101,110, 111, 202, 212, 220)
and vibrational
and 101 ͑ortho͒. The jet cooled H2O molecules are then
a
c
(
͉
03ϩ
,
͉
03Ϫ
,
͉
12ϩ
,
͉
02Ϫ ) states accessible via excitation
pumped into specific rovibrational states in the OHϭ3 OH
v
͘
͘
͘
͘
out of 000 ͑para͒ and 101 ͑ortho͒ lower state levels populated
under jet cooled conditions. This represents a study of vibra-
tionally mediated photolysis of H2O out of the OHϭ3
stretch manifold, which therefore nicely complements previ-
stretch overtone manifold by IR pump pulses from a single
mode, tunable optical parametric oscillator ͑OPO͒, which in
turn are selectively photodissociated by a KrF excimer laser
pulse at 248 nm. Finally, the nascent OH product distribu-
tions resulting from the state-selected photolysis events are
probed under collision free conditions by LIF methods opti-
mized for the slit jet configuration. Some relevant experi-
mental details are provided below.
v
ous experiments obtained for OHϭ1, 4, and 5 vibrational
v
excitation, both at total photolysis energies comparable to
our experiment6,7 and at higher photolysis energies.3,8,10,12
The rest of the article is organized as follows. Section II
provides the experimental details relevant to the present
work. The results of OH product vibrational distributions and
relative photolysis cross sections for the overtone manifold
The H2O supersonic expansion sample is introduced into
the vacuum chamber through
a pulsed slit nozzle
(4 cmϫ50 ␮m, 500 ␮s pulse duration͒, with typically
0.5%–1% of H2O seeded in a carrier gas ͑He, Ne, or Ar͒ at
of OHϭ3 initial H2O states are summarized in Sec. III and
v
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.252.67.66 On: Tue, 23 Dec 2014 02:43:26

J. Chem. Phys., Vol. 110, No. 17, 1 May 1999
Votava, Plusquellic, and Nesbitt
8567
TABLE I. Experimentally observed
ϭ1/ OHϭ0 branching ratios ob-
vOH
v
a total stagnation pressure of 200–600 Torr. Due to rota-
tional cooling in the supersonic expansion, the H2O popula-
tions are concentrated into the lowest ortho (101) and para
(000) rotational states in the 3:1 ratio dictated by the nuclear
spin statistics. As a result of the slower (1/r) drop off in gas
density, the slit expansion provides high concentrations of
H2O per quantum state in the interaction region and also
provides a long absorption path for enhancing weak overtone
transition probabilities. Furthermore, the slit geometry re-
sults in efficient collimation of velocities along the slit axis,
which leads to an order of magnitude narrowing of overtone
Doppler widths and thus a corresponding enhancement of
vibrationally mediated photodissociation signals for a sub-
Doppler pump laser. The LIF detection volume (1 mm
ϫ1 mmϫ4 cm) is spatially filtered and imaged onto the pho-
tomultiplier tube ͑PMT͒, using light collection assembly op-
tics described elsewhere for high scattered light rejection
(1:109) and high efficiency ͑Ϸ10%͒ collection along the slit
dimension.24
tained via vibrationally mediated photolysis of H2O pre-excited into
vOH
ϭ3 ͑this work͒ and OHϭ4 ͓Crim and co-workers ͑Ref. 10͔͒ overtone
v
manifolds. Note the strong decrease in
decreasing photolysis excess energy, i.e., exciting further out in the exit
channel of the potential energy surface.
ϭ1 vs OHϭ0 populations with
vOH
v
Photolysis wavelength
Excess energy
͑cmϪ1
H2O state
͑nm͒
͒
ϭ1/ OHϭ0
vOH
v
03Ϫ
04Ϫ
04Ϫ
12Ϫ
13Ϫ
13Ϫ
248
9000
13 600
17 600
9400
14 100
18 100
Ͻ1:200
1:100
1:10
Ͻ1:50
5.2:1
6:1
͉
͉
͉
͉
͉
͉
͘
͘
͘
͘
͘
͘
239.5
218.5
248
239.5
218.5
An excimer laser operating on KrF mix at 248 nm is
used as the photolysis source, typically scaled down to 20
mJ/pulse in the jet region. The measured UV photolysis
power dependence of the vibrationally mediated signals is
linear and indicates no signs of optical saturation. This is
consistent with the far off resonance nature of the vibra-
tionally mediated photolysis process, which in more favor-
able cases yields UV cross sections more than two orders of
magnitude smaller than peak values (␴Ϸ10Ϫ18 cm2) pre-
dicted in the Franck–Condon region. For typical UV pulse
energies ͑20 mJ/pulse͒ and beam areas ͑1 mm2͒, this trans-
lates into Ϸ4ϫ109 photolyzed H2O molecules in the detec-
tion volume.
A single mode injection seeded OPO of our own
design25 serves to select single rovibrational states in the
OHϭ3 stretching manifold. The OPO is based on a four
v
mirror ring cavity resonator, with a matched pair of beta
barium borate ͑BBO͒ crystals pumped at 355 nm from a
frequency tripled, single-mode, injection seeded neodymium:
yttrium–aluminum–garnet ͑Nd:YAG͒ laser to produce high
spectral brightness ͑5–10 mJ͒ signal and idler pulses in the
visible ͑570–650 nm͒ and near-IR ͑790–950 nm͒ spectral
region, respectively. By injection seeding the BBO resonator
with a single-mode continuous wave dye laser, a Fourier-
transform limited output of the ring OPO is achieved with a
spectral width of 160 MHz ͑Ϸ0.005 cmϪ1͒, averaged over
multiple pulses. With output pulse energies of у5 mJ/pulse,
this OPO light source provides sufficient spectral brightness
A conventional frequency doubled pulsed dye laser
source provides tunable radiation near 308 nm to probe the
2
2
OHϭ0←0 and 1←1 sub-bands of the ⌺← ⌸ electronic
v
transition in the OH product. To obtain reliable product state
distributions from the obtained LIF spectra, one must com-
pensate for LIF signal saturation effects. Specifically, at
these OH photofragment densities in the jet expansion envi-
ronment, it is difficult to obtain adequate signal-to-noise
͑S/N͒ for all OH transitions with sufficiently low probe pow-
ers to be rigorously in the linear regime. Thus, the UV probe
laser is operated at typical energies of 2–20 ␮J/pulse, a range
over which the LIF signals begin to show slight curvature
with increasing power due to partial saturation. This partial
saturation behavior is corrected by calibrating the OH signals
as a function of pulse energy against thermally equilibrated
300 K OH distributions, which are readily obtained in the
same detection geometry by 193 nm photolysis of static H2O
inert gas mixtures. As described elsewhere in detail, this pro-
cedure permits a quantitative mapping between observed sig-
nal strengths and OH population in given state, and allows
one to obtain product state distributions24 in a partially satu-
rated regime.
to partially saturate up to ⌬ ϭ3 OH stretch overtone rovi-
v
brational transitions. With efficient jet cooling into the low-
est two rotational levels, this permits significant optical
pumping ͑10%–20%͒ of the total H2O molecules in the
probe volume into a given rovibrational quantum state. For
typical stagnation pressures and distances downstream from
the slit nozzle, this translates into excited state H2O densities
in excess of 1013 molecules/cm3 and 4ϫ1011 total molecules
in the detection volume, respectively.
Due to high spectral resolution of the OPO pump source
and sub-Doppler linewidths in the slit expansion geometry,
the long term frequency stability of the OPO output is criti-
cal. Specifically, even a relatively small frequency drift
͑Ϸ100 MHz͒ during the course of an experiment would tune
the laser off the molecular transition and consequently lead
to loss of multiple-resonance signals. Active servo-loop
methods for OPO frequency stabilization are based on lock-
ing the stabilized optical transfer cavity onto the output of a
polarization stabilized He–Ne laser followed by locking both
͑i͒ the injection seed dye laser for the OPO and ͑ii͒ injection
seed diode laser for the Nd:YAG onto the same stabilized
transfer cavity. With this procedure, one can achieve fre-
quency drifts in the OPO output of less than Ϯ20 MHz over
a period of several hours,26 which proves entirely adequate
for present purposes.
III. RESULTS AND ANALYSIS
As summarized in Table I, single-mode near-IR OPO
excitation of a jet-cooled H2O sample is exploited to select a
series of intermediate rotational states (JK K ϭ000 , 101
,
a
c
111 , 110 , 202 , 212 , and 220) in all four vibrational levels of
the vOHϭ3, OH stretch overtone manifold. Subsequent pho-
tolysis at 248 nm leads to HϩOH products, preferentially
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.252.67.66 On: Tue, 23 Dec 2014 02:43:26

8568
J. Chem. Phys., Vol. 110, No. 17, 1 May 1999
Votava, Plusquellic, and Nesbitt
tional state is
͉
12Ϫ , i.e., the antisymmetric combination of
͘
two and one quanta of local mode stretching excitation lo-
calized in the two OH bonds, respectively.
The data reveal several interesting results. First, only the
OHϭ0 ground vibrational state is observed at 248 nm pho-
v
tolysis wavelength, with an upper limit of Ͻ0.05% OHϭ1
v
product within the S/N ͑note the 50-fold increase in scale
sensitivity͒. Second and more importantly, this vanishingly
small
ϭ0/ OHϭ1 ratio is observed for both ͉
03Ϫ and
vOH
͘
v
͉
12Ϫ excitation, i.e., independent of the intermediate vibra-
͘
tional state of H2O. It is important to note that this is not
simply due to energetics, which would make vibrational
states up to OH ( ϭ3) accessible. This is in contrast with
v
previous observations at higher total photolysis energies,10
which clearly demonstrate that vibrational excitation in the
nondissociating ‘‘spectator’’ OH bond tends to be adiabati-
cally conserved in the dissociation event. Specifically, a five-
fold excess of excited state OHϭ1 products were observed
v
when the
from the
ϭ0 products. By way of comparison, Table I compares the
present ϭ0/ OHϭ1 branching ratios with those previ-
͉
13Ϫ state was dissociated, while dissociation
͘
FIG. 3. Sample LIF spectra of OH products produced via vibrationally
mediated photodissociation of H2O. The OH products from
03Ϫ and 12Ϫ
initial vibrational states of H2O are reflected in the top and bottom panels,
respectively. OH in its ground ( ϭ0) and first excited ( OHϭ1) vibra-
͉
04Ϫ state led to predominantly ground state
vOH
͘
͉
͉
͘
͘
vOH
tional states are probed in the left hand and right hand panels, respectively.
Note the lack of any OH( ϭ1) nascent state population from photolysis of
v
vOH
ously reported. The final feature worth noting is the order of
magnitude increase in LIF signal intensity for
03Ϫ versus
12Ϫ vibrational excitation. Since the known overtone IR
v
v
͉
͘
either ͉ ͉
03Ϫ or 12Ϫ , despite the 50-fold increase in scale sensitivity.
͘ ͘
͉
͘
absorption cross sections are comparable for exciting these
two states, this reflects significantly different photolysis
cross sections for UV photodissociation out of these two
vibrational levels.
sampling photofragmentation dynamics in the exit valley
versus the inner Franck–Condon region ͑see Fig. 1͒. This
combination of IRϩUV photolysis wavelengths yields ex-
cess energies in the HϩOH product on the order of Ϸ9400
cmϪ1; this is appreciably smaller than sampled in the previ-
ous vibrationally mediated photolysis studies of Crim and
More quantitatively, we obtain the relative photodisso-
ciation cross sections for each of the four intermediate rovi-
brational states of H2O from the experimental data and com-
pare it with predictions of a 2D quantum model discussed in
Sec. III B. For extracting relative cross sections from the
measured OH populations, one must take into account ͑i͒ the
overtone transition probability to the intermediate state and
͑ii͒ the H2O concentration in the initial quantum state. As
discussed above, only the two lowest rotational states of
H2O, 000 , and 101 , are populated in the free jet expansion,
with a 1:3 population ratio dictated by nuclear spin statistics.
In the unsaturated limit, the optical transition probability is
proportional to the transition line strength, which is experi-
mentally known from the HITRAN database.27 Since the full
spectral brightness of the infrared OPO is capable of partially
saturating the second overtone vibrational transitions. The
pulse energies are attenuated until H2O photoacoustic signals
scale linearly with power and transition line strength, i.e.,
verifying that saturation effects are negligible. The 248 nm
photodissociation cross sections for photodissociation from
co-workers for OHϭ4 ͑between 14 100 and 18 100 cmϪ1 for
v
220–240 nm͒10,12 and of Hausler, Andresen, and Schinke for
OHϭ1 ͑Ϸ13 900 cmϪ1 for 193 nm͒.3 However, these excess
v
product state energies are more comparable to those investi-
gated by Brouard and co-workers,6,7 who used 282 nm light
as both the UV photolysis and OH probe laser to study vi-
brationally mediated photofragmentation dynamics of H2O
OHϭ4 and OHϭ5 overtones, yielding excess energies of
v
v
7700 and 10 700 cmϪ1, respectively.
A. OH vibrational distributions
The four panels in Fig. 3 represent LIF spectra probing
the OH product state distributions from vibrationally medi-
ated photodissociation of H2O. In the left hand set of panels,
the OH ground vibrational state population ( OHϭ0) is
v
sampled via excitation on the OHϭ0←0 sub-band of the
v
͉
03Ϫ 03ϩ 12Ϫ , and 12ϩ states are found to be
, ͉ , ͉ ͉
͘ ͘ ͘ ͘
2
2⌺← ⌸ electronic transition. In the right hand set of panels,
1:3.0͑15͒: 0.08͑3͒: 0.17͑5͒, respectively, with no dependence
on intermediate rotational state within experimental uncer-
tainty.
the corresponding OHϭ1←1 sub-band region of the same
v
electronic transition is presented, thus revealing the absence
of any nascent population in the OHϭ1 state. For the top
v
two panels, the OPO IR excitation laser is tuned to the 000
←101 rotational transition in the
͉
03Ϫ vibrational manifold
͘
B. 2D quantum theoretical model
of H2O, i.e., an antisymmetric state with nominally three and
zero quanta of local mode stretching vibration, respectively,
localized in one of the two equivalent OH bonds. Con-
versely, for the lower two panels, the intermediate vibra-
To interpret these results, we have carried out simplified
time independent quantum calculations of the final state re-
solved photodissociation cross sections. The model follows
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.252.67.66 On: Tue, 23 Dec 2014 02:43:26

J. Chem. Phys., Vol. 110, No. 17, 1 May 1999
Votava, Plusquellic, and Nesbitt
8569
the time independent approach by Engel, Schinke, and
Staemmler,19 which in turn is based on the Fermi golden rule
2
4␲
g
v ͑v
͒
2
␴
͒ϭ
͉
⌿
͉
␮
͉
⌿
͉
Љ
͘
.
͑1͒
͑v
Ј
͗
3បc
In Eq. ͑1͒, ␴(␯) is the UV absorption cross section, ⌿ and
Љ
⌿ are the ground and excited wave functions, respectively,
Ј
␯ is the frequency of the excitation photon, g(␯) is the den-
sity of states, and ␮ is the dipole moment operator. In the
context of the Born–Oppenheimer approximation, ⌿ and
Љ
⌿
represent bound and continuum wave functions for the
Ј
nuclear motion on the ground and excited PESs. Similarly, ␮
is the electronic transition dipole moment vector as a func-
tion of intramolecular coordinates, as taken from the work of
Staemmler and Palma.16 Throughout this analysis, the upper
state potential surface is also taken from the full three-
dimensional ͑3D͒ ab initio PES by Staemmler and Palma,16
while the potential surface of Sorbie and Murrell28 is used
for the ground state.
FIG. 4. Sample matrix element overlap between the ground state vibrational
wave function ⌿
03Ϫ and the scattering wave function ⌿ OH(
͉
Љ
Ј ͓
v
͘
ϭ0) . The resulting dipole moment integrand ⌿ ␮⌿ is strongly localized
͔
Ј
Љ
in the optimal overlap region between those two wave functions, which for
these excess energies leads to Franck–Condon factors sensitive to the out-
The following additional assumptions are implemented
to simplify the analysis. First, the bending and stretching
coordinates in H2O are assumed to be decoupled from each
other, with the bend angle fixed at 104° for both ground and
first excited electronic states. This approximation has been
commonly invoked in previous theoretical treatments, based
on small changes in the bending potentials between ground
and excited electronic states. The ground state wave func-
tions are then expanded in a direct product basis set of the
two OH stretching coordinates and solved numerically via
matrix diagonalization. Second, coupling between the two
OH stretch degrees of freedom in the exit channel is ne-
glected, i.e., the excited state potential surface is approxi-
mated by
going OH( ) state.
v
a given lower state potential and intermediate rovibrational
state, this value of Etot is supplied by the UV photolysis
energy, which uniquely determines ⑀ for each separate
OH( ) vibrational channel.
v
In this model, both ␾v(r) and ␺⑀(R) are 1D wave func-
tions that can be readily obtained by a number of numerical
methods. For calculational simplicity, the wave functions are
solved variationally in a basis set of harmonic oscillator
eigenfunctions for the lower state and a grid of distributed
Gaussians for the upper state followed by matrix diagonal-
ization. To make the wave function normalization in Eq. ͑1͒
consistent between lower ͑bound͒ and upper ͑continuum͒
states, the continuum solutions for Vrepulsive(R) are trans-
formed to a discrete but energetically closely spaced set of
bound-state solutions by adding a repulsive wall at large val-
ues of RϭRw . This procedure generates solutions only for
discrete values of the excess energy ⑀, but with a density of
states g(␯) that also depends on Rw . As a result, the transi-
V R,r͒ϭV
͑
R͒ϩV
r͒,
͑
OH
͑2͒
͑
repulsive
where r and R are Jacobi coordinates representing ͑i͒ the
nondissociating OH bond length, and ͑ii͒ the distance be-
tween the OH center of mass and departing H atom, respec-
tively. This assumption should be reasonably accurate for
photolysis far into the exit channel, where the coupling be-
tween OH vibration and H–OH translation is weak. Consis-
tent with this approximation, Vrepulsive(R) is determined from
the minimum energy of V(R,r) with respect to r, obtained
parametrically as a function of R, while VOH(r) is simply
taken to be V(R,r) at asymptotically large values of R.
Within the context of these two assumptions, the total
excited state wave function can be expressed as
tion dipole moment matrix element ⌿
͉
␮
͉
⌿
decreases
Љ
͘
Ј
͗
with increasing displacement of the outer wall, but is exactly
compensated by a corresponding increase g(␯) for suffi-
ciently large Rw . The basis set is then successively increased
in size until the calculated absorption cross sections defined
by Eq. ͑1͒ are converged to at least three significant figures.
⌿
R,r͒Ϸ␺ R͒␾ r͒,
͑3͒
͑
͑
͑
Ј
⑀
v
C. Discussion
where ␾ (r) is the th vibrational wave function of free OH
and ␺⑀(R) is the one-dimensional ͑1D͒ scattering wave func-
v
v
From Eq. ͑1͒, the predicted absorption cross sections are
obtained from integration of ⌿ ␮⌿ over all space; it is
Ј
Љ
¨
tion satisfying the following Schrodinger equation:
therefore instructive to observe the behavior of the dipole
moment integrand itself. This is demonstrated in Fig. 4 for
Ϫប2 ץ2
2␮ ץR2
͑
the
ing wave function corresponding to 248 nm photodissocia-
͉
03Ϫ ground vibrational wave function and the scatter-
͘
ϩVrepulsive R͒ ␺ ϭ⑀␺ .
͑4͒
ͩ
ͪ
⑀
⑀
In Eqs. ͑3͒ and ͑4͒, ⑀ is the asymptotic center of mass trans-
lational energy for a given scattering wave function, which
in conjunction with the internal vibrational energy of depart-
ing OH product requires a total energy of Etotϭ⑀ϩEOH . For
tion and OH production in OHϭ0. As anticipated, the inte-
v
grand is strongly localized in the exit channel region near the
outer turning point of the ground state wave function ⌿ ,
Љ
i.e., with the nonspectator OH bond highly extended. If one
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.252.67.66 On: Tue, 23 Dec 2014 02:43:26

8570
J. Chem. Phys., Vol. 110, No. 17, 1 May 1999
Votava, Plusquellic, and Nesbitt
FIG. 5. Dipole moment integrand for initial H2O states
͉
03Ϫ and 12Ϫ and final OH vibrational states
͉
vOH
͘ ͘
ϭ0 and OHϭ1.
v
neglects any coordinate dependence of the dipole moment
operator, this peaking of the dipole moment integrand occurs
semiclassically at the Condon point, i.e., the stationary point
for which the momentum of the particle is the same on both
the ground and excited PES.29 For higher photolysis energies
͑and thus higher values of Etot) the inner lobe of the scatter-
ing wave function shifts to progressively lower values of R,
exponentially increasing overlap with the ground state wave
function and therefore the amplitude of the dipole moment
integrand.
for these two pairs of intermediate levels, the
͉
12Ϯ states
͘
exhibit a nodal plane in the r coordinate, which in a local
mode picture corresponds to one quantum of vibrational ex-
citation in the spectator OH bond. When integrated over the
intermediate vibrational state wave function, this leads to
significant cancellation in the Franck–Condon overlap and
therefore predicts much lower UV photolysis cross sections
for ͉ ͉ .
12Ϯ than 03Ϯ
͘ ͘
A more quantitative comparison of calculated versus ob-
served cross sections into OHϭ0 for all four initial vibra-
v
This 2D quantum analysis provides a simple qualitative
tional states, ͉ ͉
12Ϯ and 03Ϯ is summarized in Fig. 6. Con-
͘ ͘
picture for the experimentally observed trends in OHϭ0 and
sidering the simplicity of the 2D model, the agreement is
v
OHϭ1 partial photodissociation cross sections. Figure 5
v
demonstrates this by plotting dipole moment integrands for
248 nm photolysis of the two H2O vibrational states
03Ϫ
and
12Ϫ , projected onto the asymptotic HϩOH continuum
͉
͘
͉
͘
wave functions for OHϭ0 and 1. This combination is cho-
v
sen explicitly for illustrative comparison with the experimen-
tal observations presented in Fig. 3. The integrated ampli-
tudes are significantly lower for the OHϭ1 channels for
v
both the
perimental observation that OH production in OHϭ1 is
͉ ͉
03Ϫ and 12Ϫ initial states, consistent with ex-
͘ ͘
v
down from OHϭ0 by at least 50- to 100-fold. The reason
v
for this is straightforward; for OHϭ1 the excess transla-
v
tional energy available to the fragments is reduced by an
additional Ϸ3700 cmϪ1. Thus, the inner turning point of the
scattering wave function is shifted to larger values of R,
which exponentially diminishes Franck–Condon overlap
with the ground state wave function and partial absorption
FIG. 6. Measured and calculated relative photodissociation cross sections at
248 nm, for each of the four initial H2O vibrational states probed in the
cross section into OHϭ1.
v
More subtly, these plots of dipole moment integrands
also explain the strong decrease in total absorption cross sec-
experiment (͉ ,͉ ,͉ ,͉
03Ϫ 03ϩ 12Ϫ 12ϩ ). Two different rotational states have
͘ ͘ ͘ ͘
been included for each vibrational state to demonstrate that photodissocia-
tion cross sections do not depend significantly on the initial rotational state,
within experimental uncertainty.
tion between the
lute values of the wave function amplitudes are comparable
͉ ͉
03Ϯ and 12Ϯ states. Although the abso-
͘ ͘
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.252.67.66 On: Tue, 23 Dec 2014 02:43:26

J. Chem. Phys., Vol. 110, No. 17, 1 May 1999
Votava, Plusquellic, and Nesbitt
8571
reasonable and captures several of the experimental trends
observed. Specifically, the 248 nm photolysis cross sections
are predicted to be comparable for each of the
͉
12ϩ ), one can monitor the complementary 110 , 111 , and
͘
212 rotational levels. Experimental data have been obtained
and analyzed on each of these intermediate states, the results
of which are summarized in Table II and graphically pre-
sented in Figs. 7͑a͒–7͑d͒.
The most dominant feature of the nascent OH rotational
distributions is the large, nonmonotonic behavior in the
populations. This effect is most clearly evident for low N
quantum states, where the populations alternate with
͉
03ϩ and
͘
͉
03Ϫ states, in good agreement with the 3.0͑15͒ ratio ob-
͘
served experimentally. Furthermore, each of the
͉
12ϩ and
͘
͉
12Ϫ states are predicted to have 10- to 100-fold diminished
͘
UV absorption cross sections with respect to
͉
03ϩ and
͘
͉
03Ϫ , due to quantum interference in the dipole moment
͘
integrand, again in good qualitative agreement with experi-
ment. In fact, the
͉
12Ϫ cross section is predicted to be al-
͘
OH( ,N) rotational level and occur with opposite phase for
v
ϩ
Ϫ
2
the two ⌳-doublet (2⌸3/2 and ⌸3/2) components. Such os-
cillations are due to quantum interference between the inter-
mediate H2O and final OH wave functions, where both
nuclear and electronic degrees of freedom are included. Con-
versely, the qualitative trends observed for the two spin-orbit
most an additional order of magnitude lower than observed
experimentally. However, this is a result of quantum inter-
ference effects, which for near perfect cancellation in the
dipole moment integral depend quite sensitively on both the
model and details of the upper/lower potential surfaces. It
would be of interest to calculate these relative absorption
cross sections via a full 3D quantum model, and thereby
permit a more rigorous test of the associated upper/lower
potential surfaces. Finally, it is worth noting that at shorter
wavelengths that access the Franck–Condon region directly,
this 2D model also predicts photolysis trends analogous to
ϩ
ϩ
(2⌸3/2 and ⌸1/2) distributions are very similar; in fact, the
two spin-orbit components at low N are often close to the 2:1
ratio predicted in the limit of a statistical dissociation. Both
the strongly nonstatistical out of phase oscillations between
⌳-doublet components and the nearly statistical ratios be-
tween spin-orbit manifolds have been observed in previous
vibrationally mediated photolysis studies of the other over-
tone manifolds. Furthermore, there has been reasonable suc-
cess at predicting many qualitative features of these highly
structured distributions by Hausler, Andresen, and Schinke3
based on rotational Franck–Condon models of Balint-
Kurti.23
2
those observed for the
cally, the OHϭ1 channel becomes dominant for
while for
channel.
͉ ͉
13Ϫ and 04Ϫ initial states. Specifi-
͘ ͘
͉
12Ϫ
͘
v
͉
03Ϫ , production of OHϭ0 remains the majority
v
͘
IV. OH PRODUCT STATE DISTRIBUTIONS
It has also been noted in previous studies of the
vOH
2
ϭ1, 4, and 5 vibrational manifolds that the OH rotational
and fine structure distributions are insensitive to the nature of
the intermediate vibrational state.3,10,12 As rather dramati-
The presence of an unpaired electron in the ⌸ state of
OH leads to spin and orbital angular momentum contribu-
tions to the total angular momentum. The two spin-orbit
2
2
cally shown in Fig. 8, the current study of the OHϭ3 over-
v
states ͑labeled ⌸1/2 and ⌸3/2) are further split by coupling
between electronic angular momentum and the end-over-end
rotation into two ⌳-doublet components, denoted by ϩ/Ϫ
superscripts according to the electronic wave function sym-
metry with respect to the OH end-over-end rotational plane.
As a result of optical selection rules between the fine struc-
ture components, the rotational state populations in these
four spin-orbit/⌳-doublet states can be extracted from Q and
P/R branch analysis of the LIF spectrum of the OH photoly-
sis product.15 In practice, quantitative extraction of ⌸Ϫ1/2
populations is not easily obtained due to overlapping Q
branch lines at low N. However, all three other fine structure
states (⌸ϩ3/2 , ⌸3Ϫ/2 , and ⌸1ϩ/2) can be readily resolved and
analyzed to yield relative populations of product OH.
tone manifold for ͉ ͉
03Ϯ and 12Ϯ provides strong support
͘ ͘
of this trend for the rotationless ground state ͑though specific
discrepancies in selected JϾ0 levels will be mentioned
later͒. Such behavior is of course consistent with the
Franck–Condon model, in which the predicted OH rotational
distributions solely reflect an expansion of the zero point
bending wave function for H2O in asymptotic free rotor OH
states. Specifically, in the absence of such exit channel inter-
actions as bend–stretch coupling, rotation–vibration mixing,
recoil torques, and other dynamical effects that could be in-
fluenced by pre-excitation of the OH bond, the model pre-
dictions should be completely insensitive to the initial over-
tone vibrational state. However, this experimental trend has
been argued by Brouard and co-workers6,7 not to be due
simply to a lack of such exit channel effects, but rather to a
fortuitous balancing between ͑i͒ softening of the HOH bend-
ing potential and ͑ii͒ increase in the moment of inertia due to
prestretching of the OH bond. If this is in fact the case, one
might anticipate the insensitivity to overtone vibrational state
for JK K ϭ000 displayed in Fig. 8 to be less predictively
The OH product state distributions have been deter-
mined for each rotational state accessible under jet cooled
conditions for the four stretching vibrational states in the
OHϭ3 overtone manifold. As a result of the low slit jet
v
temperatures and angular momentum considerations, the in-
termediate rotational states for vibrationally mediated pho-
tolysis are constrained to Jр2 for single photon transitions.
Furthermore, the ϩ/Ϫ symmetry of H2O overtone vibra-
tional levels permits only A-type ͑i.e., ⌬Kaϭeven, ⌬Kc
ϭodd) and B-type ͑i.e., ⌬Kaϭodd, ⌬Kcϭodd) transitions.
Consequently, for the antisymmetric vibrational states
a
c
reliable for rotationally excited H2O levels.
Indeed, the OH product state behavior for OHϭ3 pho-
v
todissociation of H2O molecules in JϾ0 is more compli-
cated. As demonstrated in Fig. 9 for JK K ϭ110 , the initial
a
c
(
0
͉
03Ϫ and
͉
12Ϫ ), vibrationally mediated photolysis out of
vibration appears to play only a modest role in determining
OH product state distributions, qualitatively similar to what
is observed for the photolysis of the rotationless ground state
͘
͘
00 , 101 , 202 , and 220 rotational levels can be investigated,
whereas for the symmetric vibrational states (
͉
03ϩ and
͘
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.252.67.66 On: Tue, 23 Dec 2014 02:43:26

8572
J. Chem. Phys., Vol. 110, No. 17, 1 May 1999
Votava, Plusquellic, and Nesbitt
FIG. 7. Rotational, spin-orbit, and ⌳-doublet distributions of the OH product for selected initial H2O vibrational and rotational states: ͑a͒
͑c͒
03ϩ , and ͑d͒ 12ϩ
͉ ͉ ,
03Ϫ , ͑b͒ 12Ϫ
͘ ͘
͉
͉
.
͘
͘
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.252.67.66 On: Tue, 23 Dec 2014 02:43:26

J. Chem. Phys., Vol. 110, No. 17, 1 May 1999
Votava, Plusquellic, and Nesbitt
8573
TABLE II. Fractional OH photofragment populations in rotational, spin-orbit, and lambda doublet states ob-
tained from vibrationally mediated photolysis of all intermediate rovibrational states in the OHϭ3 overtone
v
manifold optically accessible from JK ϭ000 ͑para͒ and 101 ͑ortho͒ H2O levels populated in the slit jet
K
a
c
expansion. The highly nonmonotonic quantum state dependent structure is all quite reproducible; experimental
uncertainties are Շ5% of the reported values.
Intermediate
H2O state
NOH
vibration
JK
OH state
1
2
3
4
5
6
K
a
c
ϩ
2
2
2
͉
03Ϫ
000
101
202
220
000
101
202
110
111
212
110
212
31
10
4.9
4.4
2.2
26
2.0
2.4
4.5
1.0
1.3
3.1
0.2
0.3
4.3
0.2
0.5
0.3
⌸
⌸
⌸
͘
3/2
ϩ
1/2
Ϫ
3/2
ϩ
2
2
2
9.4
3.4
21
15
7.7
12
2.5
2.2
14.5
1.0
1.4
5.2
0.4
0.40
2.5
0.0
0.2
1.8
⌸
⌸
⌸
3/2
ϩ
1/2
Ϫ
3/2
ϩ
2
2
2
5.1
2.1
23
15
8.4
9.8
4.2
2.9
14
0.9
1.3
7.1
0.6
0.9
1.8
0.1
0.1
2.2
⌸
⌸
⌸
3/2
ϩ
1/2
Ϫ
3/2
ϩ
2
2
2
16
6.0
5.3
4.2
2.7
18
12
9.8
7.2
2.9
2.9
5.9
0.4
0.6
3.8
0.3
0.9
0.4
⌸
⌸
⌸
3/2
ϩ
1/2
Ϫ
3/2
ϩ
2
2
2
͉
͉
͉
12Ϫ
03ϩ
12ϩ
29
9.5
2.5
2.4
0.9
23
3.6
4.7
2.1
0.6
2.2
9.4
1.1
1.4
3.5
1.0
0.5
1.8
⌸
⌸
⌸
͘
͘
͘
3/2
ϩ
1/2
Ϫ
3/2
ϩ
2
2
2
16.9
5.3
12
9.7
5.9
15
3.0
3.0
5.9
1.7
2.5
7.1
0.7
0.8
5.3
0.8
1.0
2.1
⌸
⌸
⌸
3/2
ϩ
1/2
Ϫ
3/2
ϩ
2
2
2
6.2
3.1
18
14
7.9
6.4
3.0
2.8
14.5
2.7
3.4
4.9
0.9
1.5
6.0
1.1
0.9
2.9
⌸
⌸
⌸
3/2
ϩ
1/2
Ϫ
3/2
ϩ
2
2
2
2.9
1.1
25
19
11
7.9
4.0
2.6
12
0.7
1.2
7.8
0.6
0.8
1.0
0.1
0.1
1.9
⌸
⌸
⌸
3/2
ϩ
1/2
Ϫ
3/2
ϩ
2
2
2
33
10.0
5.9
12
6.6
12
4.8
4.4
1.5
1.0
1.5
3.4
0.5
0.8
1.5
0.2
0.3
0.5
⌸
⌸
⌸
3/2
ϩ
1/2
Ϫ
3/2
ϩ
2
2
2
29
9.6
4.5
4.2
2.2
24
5.9
5.2
0.7
1.6
1.8
4.9
0.3
0.5
4.7
0.3
0.5
0.5
⌸
⌸
⌸
3/2
ϩ
1/2
Ϫ
3/2
ϩ
2
2
2
2.7
1.1
29
17
10
4.7
2.2
1.3
8.8
1.8
2.4
1.2
0.9
0.8
1.3
0.0
0.4
0.1
⌸
⌸
⌸
3/2
ϩ
1/2
Ϫ
3/2
ϩ
2
2
2
22
7.0
4.3
4.7
1.6
22
8.8
6.3
1.8
1.2
1.5
10
1.3
1.1
3.1
0.1
0.3
1.8
⌸
⌸
⌸
3/2
ϩ
1/2
Ϫ
3/2
000 . However, as also demonstrated in Fig. 9, strong quali-
tative differences are observed for some initial rotational lev-
els such as 101 , where the phase of the OH lambda doublet
It is worth noting that a strong dependence of OH prod-
uct distributions on initial vibrational state for a given inter-
mediate rotational level has been also observed in the vOH
oscillations clearly reverses between ͉ ͉
12Ϫ versus 03Ϫ
͘ ͘
ϭ4 and
5
manifold by Brouard, Langford, and
Manolopoulos.7 Specifically, photolysis of JK K ϭ404 H2O
states. Due to the finite number of levels sampled in the slit
jet, it is not clear if this effect is isolated to one or a few
rotational states, or whether similar discrepancies would also
be evident in higher rotational states. Nevertheless the data
in Fig. 9 indicate that quite different dynamical behavior can
be observed even for states of the same total angular momen-
tum.
a
c
results in dramatically different OH rotational distributions
depending on whether the intermediate vibrational level is
͉
04Ϫ and 05Ϫ . Similar to the present work, these studies
͉
͘ ͘
were performed at total energies that sampled initial H2O
vibrational states localized far out into the exit channel.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.252.67.66 On: Tue, 23 Dec 2014 02:43:26

8574
J. Chem. Phys., Vol. 110, No. 17, 1 May 1999
Votava, Plusquellic, and Nesbitt
FIG. 9. OH product rotational distributions for photodissociation of 101 and
110 states of H2O from all four intermediate vibrational states. The filled
circles represent the data from
the unfilled circles represent data for
rotational state, there are quite significant differences in distributions origi-
nating from
03Ϫ and 12Ϫ initial vibrational states, respectively.
͉ ͉
03Ϫ and 03ϩ initial H O vibration, while
͘ ͘
2
͉
12Ϫ and 12ϩ vibrations. For the 1
͉
͘ ͘
01
͉
͉
͘
͘
the effective 1D potential used to calculate the bending wave
functions. As mentioned above, it has been argued7 that con-
tributions from those two effects may fortuitously cancel for
the 000 state of H2O, but that this balance could be upset in
higher rotational states. This would suggest that sensitivity in
the photofragment distributions to initial vibrational state
would show some systematic increase with rotational excita-
tion. However, the current OHϭ3 overtone data tend not to
v
FIG. 8. OH rotational product state distributions obtained from photolysis of
H2O in the same 000 rotational state, but for a range of initial vibrational
states. Data for the
and 7, respectively. Note the remarkable insensitivity to initial vibrational
support this hypothesis, since, for example, much bigger ini-
tial vibrational state sensitivities are observed in the 101
state, than in, for example, either of the 202 and 212 states.
As a final note, we have attempted to model these prod-
uct state distributions for the 101 state in the context of the
modified rotational Franck–Condon model by changing the
Jacobi H–OH bond length ͑see Fig. 10͒. It would appear that
the modified Franck–Condon model is generally in better
͉ , ͉ ͉
01Ϫ 04Ϫ , and 05Ϫ states come from Refs. 3, 10,
͘ ͘ ͘
state and how well the current OHϭ3 data for 000 support the trend.
v
These behaviors have been rationalized by Brouard, Lang-
ford, and Manolopoulos7 in a physically plausible modifica-
tion of the Franck–Condon model by allowing the OH bond
lengths to vary with initial vibrational level. Interestingly,
they find that by fixing the OH spectator bond and varying
the fragmenting Jacobi coordinate H–OH bond length from
values corresponding to the ground state equilibrium separa-
tion to values corresponding to a vibrationally ‘‘pre-
stretched’’ OH bond brings the theoretical predictions into
better agreement with the experiment. Indeed, much of the
agreement with the
However, even for ⌬R up to 0.4 Å ͑i.e., the upper limit of
͉
03Ϫ data for increased values of R.
͘
what might be physically justified for OHϭ3 excitation͒, the
v
Franck–Condon model is not able to reproduce the experi-
mentally observed distributions for ͉
12Ϫ . Such anomalous
͘
photolysis behavior for isolated rovibrational states again
suggests the importance of specific resonances in the exit
channel between stretching, bending, and overall rotational
degrees of freedom, and therefore a much richer and more
complicated range of photofragmentation dynamics than can
be currently predicted by Franck–Condon models.
oscillatory rotational state dependent trends in the OHϭ4
v
and 5 data could be qualitatively reproduced by shifting the
Jacobi coordinate over a relatively small window of ⌬R
Ϸ0.25 Å.
There are two contributions to this effect in the context
of the modified rotational Franck–Condon model. First, the
rotational constants of the rigid bender/rotor change as the
initial Jacobi H–OH bond length is increased. Second, for
different values of R, a rigid bender model samples a differ-
ent part of the intramolecular PES, which therefore changes
V. SUMMARY
The photodissociation of state-selected H2O has been
studied in the second OH stretch overtone manifold by vi-
brationally mediated photolysis techniques at 248 nm. By
choice of far off resonance photolysis wavelength and low
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.252.67.66 On: Tue, 23 Dec 2014 02:43:26

J. Chem. Phys., Vol. 110, No. 17, 1 May 1999
Votava, Plusquellic, and Nesbitt
8575
Rotational, ⌳-doublet and spin-orbit state distributions
of the OH product have been obtained for the complete set of
initial rotational states accessible via dipole excitation from
jet-cooled 000 and 101 H2O. In agreement with previous
studies3,10,12 in other overtone manifolds, the results support
a general insensitivity of the product rotational distributions
to the initial vibration excited. This effect is most clearly
demonstrated for photodissociation out of the lowest H2O
rotational state 000 , where the rotational distributions are
nearly identical for levels of OH stretch excitation ranging
from OHϭ1 to 5 quanta. This is in good qualitative agree-
v
ment with Franck–Condon model of Balint-Kurti and co-
workers, which predicts final state distributions from project-
ing the initial H2O rigid bender wave function onto a
complete set of final OH rotor states. However, deviations
from this general trend are clearly observed for specific ini-
tial JϾ0 rotational states of H2O. This sensitivity to OH
stretch-vibrational state is not consistent with predictions of
the rotational Franck–Condon model, nor can the data be
accounted for by simple modifications of this model that
include pre-extension of the OH bond to be broken. This
complex photolysis behavior is likely to require more refined
models of the dissociation event, which, for example, include
exit channel interactions between stretching, bending and
overall molecular rotation in the outgoing products.
FIG. 10. Comparison of experimental OH distributions ͑filled symbols͒ with
the predictions of the rotational Franck–Condon model ͑open symbols͒,
with the results offset along the vertical axis for purposes of clarity. The
spectator OH bond length is held fixed at rϭ1.0 Å, while the initial H–OH
Jacobi coordinate has been adjusted by ⌬RϷ0.4 Å between Rϭ1.0 Å and
Rϭ1.4 Å without qualitatively reproducing the experimentally observed
differences.
ACKNOWLEDGMENTS
This work has been supported by grants from the Na-
tional Science Foundation. The authors would like to thank
to Professor Mark Brouard and Dr. Stephen Langford for
graciously providing us with the computer codes for rota-
tional Franck–Condon calculations and for many stimulating
discussions.
total excess energies, Franck–Condon dissociation from the
OHϭ3 manifold occurs via excitation significantly into the
v
exit channel on the upper potential surface. This has a sig-
nificant effect on OH product vibrational distributions; spe-
cifically, only the OHϭ0 product is observed for both the
v
12Ϫ and 03Ϫ initial vibrational states, with an upper limit
͉
͘ ͘
1 J. R. Huber and R. Schinke, J. Phys. Chem. 97, 3463 ͑1993͒.
2 R. N. Dixon, Chem. Soc. Rev. 23, 375 ͑1994͒.
3 D. Hausler, P. Andresen, and R. Schinke, J. Chem. Phys. 87, 3949 ͑1987͒.
4 P. Andresen, G. S. Ondrey, B. Titze, and E. W. Rothe, J. Chem. Phys. 80,
2548 ͑1984͒.
͉
on the ratio of OH( ϭ1)/OH( ϭ0) of Ͻ1%–2%. This is
v
v
in contrast to previous vibrationally mediated photolysis
studies in the OHϭ4 manifold at higher total excess ener-
v
gies accessing the Franck–Condon region, for which local
mode pre-excitation in the spectator OH bond was found to
be adiabatically conserved in the photolysis event.
5 I. Bar, D. David, and S. Rosenwaks, Chem. Phys. 187, 21 ͑1994͒.
6 M. Brouard and S. R. Langford, J. Chem. Phys. 106, 6354 ͑1997͒.
7 M. Brouard, S. R. Langford, and D. E. Manolopoulos, J. Chem. Phys. 101,
7458 ͑1994͒.
8 D. David, A. Strugano, I. Bar, and S. Rosenwaks, J. Chem. Phys. 98, 409
͑1993͒.
Relative UV photodissociation cross sections out of the
͉
03ϩ 03Ϫ 12ϩ , and 12Ϫ manifold of states are ex-
, ͉ , ͉ ͉
͘ ͘ ͘ ͘
9 R. L. Vander Wal, J. L. Scott, and F. F. Crim, J. Chem. Phys. 94, 3548
͑1991͒.
tracted by integrating over all final OH distributions and nor-
malizing to relative IR cross sections, laser power, etc. in the
overtone excitation step. Though not evident in the OH vi-
brational state distributions, the selectivity to initial H2O vi-
brational state is extremely high, for example, with more
than an order of magnitude differences observed between
10 R. L. Vander Wal, J. L. Scott, and F. F. Crim, J. Chem. Phys. 94, 1859
͑1991͒.
11 R. Schinke, R. L. Vander Wal, J. L. Scott, and F. F. Crim, J. Chem. Phys.
94, 283 ͑1991͒.
12 V. Engel, V. Staemmler, R. L. Vander Wal, F. F. Crim, R. J. Sension, B.
Hudson, P. Andresen, S. Hennig, K. Weide, and R. Schinke, J. Phys.
Chem. 96, 3201 ͑1992͒.
photolysis cross sections for ͉ ͉
03Ϫ and 12Ϫ levels. The
͘ ͘
13 H. T. Wang, W. S. Felps, and S. P. McGlynn, J. Chem. Phys. 67, 2614
͑1977͒.
results have been theoretically interpreted in terms of
Franck–Condon overlap between the vibrational wave func-
tions for H2O on the ground state surface, the transition di-
pole surface for electronic excitation, and the continuum
wave functions for HϩOH on the excited state potential sur-
face. 2D quantum calculations of these dipole transition ma-
trix elements yield results in qualitatively good agreement
with these experimental observations.
14 W. L. Dimpfl and J. L. Kinsey, J. Quant. Spectrosc. Radiat. Transf. 21,
233 ͑1979͒.
15 G. H. Dieke and H. M. Crosswhite, J. Quant. Spectrosc. Radiat. Transf. 2,
97 ͑1962͒.
16 V. Staemmler and A. Palma, Chem. Phys. 93, 63 ͑1985͒.
17 R. Schinke, V. Engel, and V. Staemmler, Chem. Phys. Lett. 116, 165
͑1985͒.
18 R. Schinke, V. Engel, and V. Staemmler, J. Chem. Phys. 83, 4522 ͑1985͒.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.252.67.66 On: Tue, 23 Dec 2014 02:43:26

8576
J. Chem. Phys., Vol. 110, No. 17, 1 May 1999
Votava, Plusquellic, and Nesbitt
19 V. Engel, R. Schinke, and V. Staemmler, J. Chem. Phys. 88, 129
͑1988͒.
25 O. Votava, J. R. Fair, David F. Plusquellic, E. Riedle, and D. J. Nesbitt, J.
Chem. Phys. 107, 8854 ͑1997͒.
20 J. Zhang and D. G. Imre, J. Chem. Phys. 90, 1666 ͑1989͒.
21 K. Weide, S. Hennig, and R. Schinke, J. Chem. Phys. 91, 7630 ͑1989͒.
22 M. V. Dirke and R. Schinke, Chem. Phys. Lett. 196, 51 ͑1992͒.
23 G. G. Balint-Kurti, J. Chem. Phys. 84, 4443 ͑1985͒.
24 D. F. Plusquellic, O. Votava, and D. J. Nesbitt, J. Chem. Phys. 107, 6123
͑1997͒.
26 D. F. Plusquellic, O. Votava, and D. J. Nesbitt, Appl. Opt. 35, 1464
͑1996͒.
27 L. S. Rothman, HITRAN database ͑ONTAR, 1996͒.
28 K. S. Sorbie and J. N. Murrell, Mol. Phys. 29, 1387 ͑1975͒.
29 M. S. Child, Semiclassical Mechanics with Molecular Applications, 1st
ed. ͑Oxford University Press, New York, 1991͒.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.252.67.66 On: Tue, 23 Dec 2014 02:43:26