Quantitative spectroscopic and theoretical study of the optical absorption spectra of H2O, HOD, and D2O in the 125-145 nm region

Article abstract of DOI:10.1063/1.1630304

Synchrotron radiation was used to determine the room temperature absorption spectra of water and its isotopomers D₂O and HOD in absolute cross section units in the 125 to 145 nm wavelength region. The cross sections at a temperature of 300K were calculated by applying a Monte Carlo sampling over the initial rotational states of the molecules. Two more resonances were found at low energy in the case of HOD. The width of the resonances of 0.04 eV is the result of overlapping and narrower resonances in the spectra of molecules differing in rotational ground state.

Full text of DOI:10.1063/1.1630304

Quantitative spectroscopic and theoretical study of the optical absorption spectra of H
2 O , HOD, and D 2 O in the 125–145 nm region
Bing-Ming Cheng, Chao-Yu Chung, Mohammed Bahou, Yuan-Pern Lee, L. C. Lee, Rob van Harrevelt, and Marc
C. van Hemert
Citation: The Journal of Chemical Physics 120, 224 (2004); doi: 10.1063/1.1630304
View online: http://dx.doi.org/10.1063/1.1630304
View Table of Contents: http://scitation.aip.org/content/aip/journal/jcp/120/1?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Communication: Quasiclassical trajectory calculations of correlated product-state distributions for the dissociation
of (H2O)2 and (D2O)2
J. Chem. Phys. 135, 151102 (2011); 10.1063/1.3655564
Theoretical study of excitations in furan: Spectra and molecular dynamics
J. Chem. Phys. 121, 4585 (2004); 10.1063/1.1780160
The bond-forming reaction between CF 2 2+ and H 2 O / D 2 O : A computational and experimental study
J. Chem. Phys. 119, 1421 (2003); 10.1063/1.1581852
Structures, energetics, and spectra of electron–water clusters, e − –( H 2 O ) 2–6 and e − – HOD ( D 2 O ) 1–5
J. Chem. Phys. 119, 187 (2003); 10.1063/1.1576757
High resolution vibrational overtone studies of HOD and H 2 O with single mode, injection seeded ring optical
parametric oscillators
J. Chem. Phys. 107, 8854 (1997); 10.1063/1.475177
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:
169.233.6.10 On: Wed, 03 Dec 2014 19:21:22

JOURNAL OF CHEMICAL PHYSICS
VOLUME 120, NUMBER 1
1 JANUARY 2004
Quantitative spectroscopic and theoretical study of the optical absorption
spectra of H2O, HOD, and D2O in the 125–145 nm region
Bing-Ming Cheng
National Synchrotron Radiation Research Center, Hsinchu Science-Based Industrial Park,
Hsinchu 30077, Taiwan
Chao-Yu Chung, Mohammed Bahou, and Yuan-Pern Lee
Department of Chemistry, National Tsing Hua University, Hsinchu, 30013, Taiwan
L. C. Lee
Department of Electrical and Computer Engineering, San Diego State University, San Diego,
California 92182
Rob van Harrevelt
Theoretische Chemie, Technische Universitaet Muenchen, 85747 Garching, Germany
Marc C. van Hemert
Leids Instituut voor Chemisch Onderzoek, Gorlaeus Laboratoria, Universiteit Leiden, Postbus 9502,
2300 RA Leiden, The Netherlands
͑Received 20 August 2003; accepted 8 October 2003͒
The room temperature absorption spectra of water and its isotopomers D2O and HOD have been
determined in absolute cross section units in the 125 to 145 nm wavelength region using synchrotron
˜
radiation. The experimental results for these B band spectra are compared with results from
quantum mechanical calculations using accurate diabatic ab initio potentials. A Monte Carlo
sampling over the initial rotational states of the molecules is applied in order to calculate the cross
sections at a temperature of 300 K. The overall rotation of the water molecule is treated exactly.
Both for the experimental and for the theoretical spectrum an analysis is made in terms of a
component attributed to rapid direct dissociation processes and a component attributed to
longer-lived resonances. The agreement between the results from experiment and theory is excellent
for H2O and D2O. In the case of HOD in the results of theory two more resonances are found at low
energy. It is demonstrated that the width of the resonances of 0.04 eV is the result of overlapping
and somewhat narrower resonances in the spectra of molecules differing in rotational ground state.
© 2004 American Institute of Physics. ͓DOI: 10.1063/1.1630304͔
I. INTRODUCTION
ion state that forms the limit of the Rydberg series. The mix-
ing of the 3s Rydberg orbital with the dissociative 4a1 va-
lence orbital was held responsible for the width of the lines.
In 1980 Segev and Shapiro2 performed the first full
three-dimensional ͑3D͒ quantum dynamics calculations for
the water photodissociation. In these calculations the two-
The optical absorption spectrum of water in the 8.6–9.9
˜
eV range ͑145 to 125 nm͒, the so-called B band, shows a
progression of lines superimposed on a broad background
continuum. The lines are approximately 0.04 eV wide and
0.1 eV apart. This structure has intrigued experimentalists
and theorists alike over the past 50 years. Among the first to
give a detailed analysis of their observed absorption spectra
of both H2O and D2O, were Wang, Felps, and McGlynn.1
Lacking detailed knowledge of the excited state potential
energy surface ͑PES͒ they assigned the bands to a progres-
sion in the bending vibration on the second excited state PES
that originates from the H–H bonding 3a1 orbital to a 3s
Rydberg orbital transition. The observed slight irregularities
in the spacing were interpreted as a signature of the occur-
rence of another excited state related to a nonbonding 1b1 to
3s transition. The spectra for H2O and D2O are largely simi-
lar, except for the difference in spacing between the bands;
that is, 0.1 eV on the average for H2O and 0.08 eV for D2O.
Support for the above assignments was found in the similar-
ity of the scaling with the hydrogen/deuterium mass ratio of
dimensional ͑2D͒ Flouquet–Horsley3 ab initio PES was used
1
˜
for the excited B state, of A1 symmetry. According to Segev
˜
and Shapiro, all members of the B-band progression could be
interpreted as resonances on one single PES, while the width
of the resonances was the result of excited state vibrations
related to a bending motion on the ridge of the PES embed-
ded in the dissociation continuum. Later calculations by Von
4
1
˜
Dirke et al., using an improved full 3D PES for the B A1
state,5 demonstrated that this single surface assignment could
not be correct. Their spectrum, calculated deliberately with
one single adiabatic PES, consisted of a dense manifold of
narrow resonances, rather different from the resonances cal-
culated by Segev and Shapiro and from the experimental
spectrum. Also experimental evidence had appeared to show
that the interpretation of Segev and Shapiro was too simple.
It was seen in photofragmentation studies6 that the OH frag-
2
the intervals in the bending progression of the excited A1
0021-9606/2004/120(1)/224/6/$22.00
224
© 2004 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:
169.233.6.10 On: Wed, 03 Dec 2014 19:21:22

225
J. Chem. Phys., Vol. 120, No. 1, 1 January 2004
Absorption spectra of H2O
ments were formed both in the ground state X 2⌸ and in the
excited state A 2⌺Ϫ, a phenomenon indicative of a dissocia-
tion process occurring on more than one PES. Finally in
2000 Van Harrevelt and Van Hemert7 showed that positions
and widths of the peaks in the progression could be repro-
duced in quantum dynamics calculations when nonadiabatic
monochromator. The wavelength of the monochromator was
calibrated against absorption features of CO ͑Refs. 15 and
16͒ and O2 ͑Ref. 17͒. A grating with 600 grooves/mm was
used to cover the spectral range 65–250 nm. The slit width
was typically 0.05 mm, corresponding to a spectral band-
width ϳ0.02 nm. The wavelength uncertainty is estimated to
be Ϯ0.02 nm, corresponding to one step of a monochromator
scan.
The absorption cross section was measured with a
double-beam apparatus.18 LiF windows transmitting light of
␭Ͼ105 nm served to seal the gas cell. The intensity of the
VUV light was monitored with light reflected from a LiF
window placed before the gas cell and at 45° from the beam
line; the VUV light was converted to visible light with so-
dium salicylate coated on a glass window, and subsequently
detected by a photomultiplier tube ͑Hamamatsu R943-02͒.
The VUV light transmitted through the gas cell ͑inner diam-
eter 39.5 mm and path length 8.9 cm͒ was detected similarly.
To avoid variation in pressure due to irradiation or interfer-
ence from surface absorption or desorption, a reservoir of
1382 cm3 in volume was connected to the absorption cell.
The absorption cross section ␴ is evaluated with the
equation,
1
interactions between the excited A1 state and the ground
state of the same symmetry were taken into account. In that
work it was made clear that the resonances in the calculated
spectrum correspond, at least in a 2D reduced dimensionality
analysis, to combined bend-stretch motions. Due to the com-
plexity of the calculations, initially only water in the lowest
rotational state, Jϭ0, was considered. When very recently
also the effect of room temperature of the sample was taken
into account, i.e., when rotational states with J and K values
up to 10 were considered, the amplitudes of the resonances
could be reproduced much better than in the Jϭ0 only
calculations.8
A further test of the interpretation of the spectrum, as
given by Van Harrevelt and Van Hemert,8 would be a com-
parison of the experimental and theoretical spectra for the
isotopomers. Such a check had been successfully performed
for the rovibrational distributions of the OH/OD fragments
˜
resulting in the case of H2O/D2O B-state photodissociation,
ln I /I͒ϭn␴lϩA ,
͑
0
0
both in the X 2⌸ ground and in the A 2⌺ϩ excited fragment
state.9 Here the experimental data to compare with were only
limited to about 10 different wavelengths in the 125–135 nm
interval. For the absorption spectra, the absolute absorption
cross sections for both H2O ͑Refs. 10 and 11͒ and D2O ͑Ref.
10͒ in the 125 to 145 nm wavelength region were known, but
there are as yet no data for HOD. In this paper we report the
absorption spectra of H2O, D2O, and HOD, in absolute cross
section units, as they have been obtained at the National
Synchrotron Radiation Research Center ͑NSRRC͒ in Taiwan.
We confront these new spectra with the results of quantum
dynamics calculations performed for these two isotopomers
along the same lines as for water as reported in Ref. 8.
In the language of classical trajectories, the broad back-
ground continuum can be thought of as resulting from trajec-
tories that lead to rapid, within about 10 femtoseconds ͑fs͒,
dissociation, while the resonances correspond to indirect tra-
jectories on the coupled surfaces.12 To visualize this bimodal
dissociation character, we transform the experimental spectra
to the time domain. In this manner the equivalent is obtained
from the autocorrelation function that is used in the theoret-
ical, time-dependent wave packet, formalism. Both the ex-
perimental and the theoretical autocorrelation functions are
then transformed back over a short time interval sufficient
for direct dissociation to obtain only the direct component of
the spectrum. The resonant part of the spectrum is found as
the difference between the direct spectrum and the full spec-
trum. In the theoretical studies the full spectrum is calculated
from the 100 fs long autocorrelation function.
in which I and I0 are intensities of transmitted and reflected
beams, respectively, n is the gas density, lϭ8.9 cm is the
path length of the absorption cell, and A0 is a proportionality
factor determined from I and I0 when the cell was evacuated
to less than 5ϫ10Ϫ7 Torr. At each wavelength, the absorp-
tion cross section was obtained on fitting with linear least-
squares 8–14 absorbance values measured at various pres-
sures; the uncertainty is estimated to be 5%. The pressure of
the absorbing gas was selected in the range 0.04–10.2 Torr,
so as to limit absorbance to be less than 2 to avoid saturation
effects. The gas density at room temperature was determined
from gas pressure, which was monitored with a MKS-
Baratron pressure meter.
Gas vapors were obtained from pure liquid samples of
H2O and D2O ͑Merck Sharp and Dome, isotopic purity
99.8%͒. HOD was prepared from a mixture of deionized
H2O and pure D2O with a molar ratio 1:2.806. For experi-
ments involving deuterated samples at least ten cycles of
passivation were carried out. Relative concentrations of
H2O, D2O, and HOD in the gas sample mixture were deter-
mined from the equilibrium constant for
H2O ϩD2O ꢀ2HOD
͑l͒
͑l͒
͑l͒
which is keqϭ3.74Ϯ0.02 ͑at 298 K͒.19,20 The molar fractions
of gaseous H2O, HOD, and D2O of this sample are deter-
mined to be 0.079, 0.395, 0.526. For samples of mixed iso-
topic species, absorption cross sections of H2O and D2O
measured in this work were subtracted from the observed
spectrum to derive a spectrum of HOD.
The cross section ␴ is decomposed into a diffuse back-
ground component, ␴dir , and a modulating indirect compo-
II. EXPERIMENTAL PROCEDURE
The experimental setup has been described
previously.13,14 In short, VUV light from a high-flux beam
line at the NSRRC in Taiwan was dispersed with a 6 m
nent, ␴indirϭ␴Ϫ␴dir . ␴ is determined by smoothing an
observed spectrum into a continuous profile; this smoothing
results from removing Fourier components with frequencies
dir
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:
169.233.6.10 On: Wed, 03 Dec 2014 19:21:22

226
J. Chem. Phys., Vol. 120, No. 1, 1 January 2004
Cheng et al.
greater than 1/(100 ⌬n), in which ⌬n is the abscissa spacing
of data points. The function used to eliminate the high-
frequency components is a parabola with its maximum at
zero frequency, decaying to zero at the cutoff frequency de-
fined above.
III. QUANTUM DYNAMICS CALCULATION
PROCEDURE
The procedure for calculating cross sections of water
with nonzero angular momentum has been described in de-
tail previously.8 Here we summarize the procedure for H2O.
With a Monte Carlo sampling a representative 300 K en-
semble of water molecules in different rotational states is
generated. The different statistical weights for ortho and para
species are taken into account. For each rotational state,
(J, K), where J is the total angular momentum and K labels
the 2Jϩ1 sublevels, a separate cross section calculation is
performed.
FIG. 1. Experimental absorption cross sections of H2O ͑full line͒, HOD
͑dotted line͒, and D2O ͑dashed line͒ in the spectral region 125–145 nm
͑8.6–9.9 eV͒.
IV. RESULTS AND DISCUSSION
The time-dependent wave packet formalism is used
where the wave function is expanded in rotoelectronic basis
functions.21 This is a convenient manner to account for cou-
pling between electronic states due to nonadiabatic interac-
tion and rotation of a molecule. The wave packets are propa-
gated on three coupled diabatic surfaces during a time span
of 76.8 fs. For HOD and D2O, the same procedure has been
applied. Also for D2O, the appropriate weight factors for
ortho and para levels have been taken into account. As re-
quired by the slower dynamics of the heavier D atoms, a
longer propagation time is applied ͑105.6 fs͒.
The potential energy surfaces and the couplings are cop-
ied from Ref. 22. The wave packets at time zero are obtained
by multiplying the wave function for the (J, K) ground state
with the appropriate electronic transition dipole moment
functions taken from Ref. 23. For D2O and HOD the grid
spacing has been decreased from 0.12 to 0.09 Bohr, and the
size of the basis set used in the rotational kinetic energy
operation has been increased from jmaxϭ60 to jmaxϭ80. For
each selected (J, K), separate calculations with ⌬JϭϪ1, 0,
and 1 are performed. For each (J, K, ⌬J), the cross section
is obtained from the Fourier transform of the autocorrelation
function of the time-dependent wave function. The full spec-
trum is obtained by averaging the (J, K, ⌬J) contributions.
In order to facilitate the study of the resonances, the
cross section ␴ is decomposed into a diffuse background
part, ␴dir , and an indirect part, ␴ind , as was done for the
experimental spectrum. The direct part is also computed
from the autocorrelation function, but now truncated before
the first recurrence time, i.e., at 12 fs. This decomposition
scheme is applied both for the full summed spectrum, and for
individual (J, K) contributions.
Experimental absorption cross sections of H2O, HOD,
and D2O in the spectral region 125–145 nm ͑8.6–9.9 eV͒ are
shown in Fig. 1. Cross sections at 1 nm intervals for each
species are listed in Table I; a complete listing of data at 0.02
nm intervals is available through internet at http://ams-
bmc.nsrrc.org.tw. Cross sections of H2O determined in this
work are in good agreement with those of Lee and Suto10
and Yoshino et al.;11 however, they are lower than those re-
ported by Chan et al.24 by about 30%. Cross sections of D2O
determined in this work are in good agreement with those of
Lee and Suto.10 No absorption cross sections of HOD in this
spectral region have been reported previously.
In Fig. 2 the indirect spectrum derived from the present
experimental spectrum is compared with the indirect spec-
TABLE I. Absorption cross sections ͑in units of 10Ϫ18 cm2) of water iso-
topomers in the 125–145 nm region.
Wavelength/nm
H2O
HOD
D2O
145
144
143
142
141
140
139
138
137
136
135
134
133
132
131
130
129
128
127
126
125
0.55
0.53
0.53
0.59
0.77
0.88
1.17
1.69
1.94
2.42
3.67
4.36
4.82
6.89
7.19
7.19
7.74
8.56
8.39
7.24
6.30
0.30
0.24
0.22
0.22
0.24
0.36
0.50
0.70
1.03
1.43
2.30
3.30
4.44
5.11
5.96
7.56
8.77
9.29
9.20
8.62
7.58
0.35
0.34
0.36
0.45
0.56
0.79
1.14
1.52
2.07
2.76
3.61
5.05
5.73
5.97
6.94
7.69
8.50
8.58
8.46
7.29
In our previous study for water we used 17 rotational
states in the Boltzmann average, and there the statistical error
in the cross section was always less than 5% and 2% on the
average, for the energy range considered ͑8.6–9.9 eV͒. It
appears that for 12 rotational states the average is hardly
different from the 17 states average. The statistical error is
then 5% on the average, and always smaller than 10%.
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:
169.233.6.10 On: Wed, 03 Dec 2014 19:21:22

227
J. Chem. Phys., Vol. 120, No. 1, 1 January 2004
Absorption spectra of H2O
also underlines that the main effect of the averaging process
is a washing out of the resonance structures, as has been
discussed before.
Next, in Figs. 4͑a͒–4͑c͒ the total cross sections from the
experiments are compared with those from theory. Both the
experimental and the theoretical spectra show that the gen-
eral structure of the spectrum is similar for the three isoto-
pomers. The most obvious difference concerns the width of
the band: going from H2O via HOD to D2O, the ground state
vibrational wave function becomes narrower, and this is re-
flected in the smaller bandwidth. The agreement between the
experimental and theoretical direct spectra ͑not shown͒ is
striking: the experimental values of the energy at half of the
maximum are 9.24 eV for H2O and 9.35 eV for D2O, while
the theoretical are 9.26 and 9.37 eV, respectively. The theo-
retical direct spectrum for HOD is again found at only 0.02
eV higher energy values than the experimental spectrum. The
good agreement between theoretical and experimental direct
cross sections implies that the potential energy surface is
very accurate in the Franck–Condon region. Also the abso-
lute values of the cross sections are well reproduced, which
implies that the calculated transition dipole moment surface
is accurate. At energies above 9.6 eV the direct theoretical
spectra start to deviate from the experimental spectra, be-
cause the calculated transition dipole moment also contains a
FIG. 2. The indirect spectrum of H2O derived from the present experimental
spectrum as compared with the indirect spectrum derived from the spectrum
of Yoshino et al. ͑Ref. 11͒.
trum derived from Yoshino’s spectrum.11 The agreement with
respect to resonance positions is excellent. The small differ-
ences in amplitudes might be due to experimental uncertain-
ties. Also the finer details that were sometimes attributed to
noise appear in both spectra.
In Fig. 3 the convergence of the theoretical cross sec-
tions with the number of rotational states present in the Bolt-
zmann averaging procedure is shown for the indirect part of
the cross section for each of the three isotopomers. Taking
more than 12 rotational states would probably change only
minor details in the spectra of HOD and D2O, thus we ex-
pect that the main features will remain the same. The con-
vergence is faster for H2O than for HOD and D2O. Figure 3
˜
contribution of the bound D state for geometries where the
potential energy is high.23
The indirect cross section depends on a range of the
potential energy surface much larger than for the direct cross
section. In Ref. 7, the nature of the resonances has been
investigated. Here, the results of that analysis are briefly
summarized. The main characteristics of the resonance struc-
ture in the spectrum can be explained by a simple two-
dimensional model, in which one OH distance is fixed, and
˜
in which only dynamics on the adiabatic B potential energy
surface is considered. This surface has two minima, one for
the collinear HOH geometry, and one for the collinear HHO
FIG. 3. The convergence of the cross sections with the number of rotational states, Nrot , present in the Boltzmann averaging procedure as shown for the
indirect part of the cross section for each of the three isotopomers. Thick line is the result for 12 rotational states. The thin lines are average cross sections for
Nrot values of 1 ͑dotted line͒, 4 ͑dashed line͒, and 8 ͑full line͒.
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:
169.233.6.10 On: Wed, 03 Dec 2014 19:21:22

228
J. Chem. Phys., Vol. 120, No. 1, 1 January 2004
Cheng et al.
FIG. 4. Comparison of the total cross sections from the experiments ͑drawn curves͒ with those from theory ͑dashed curves͒. Panel ͑a͒ for H2O, panel ͑b͒ for
D2O, and panel ͑c͒ for HOD.
geometry. For both minima the distance between OH and H
is large (3–4a0). The wave packet initially moves directly
to the HOH minimum. Here, the OH–H distance is so large
(3a0), that one can consider one OH bond to be almost
broken. The largest part of the wave packet moves further
towards the asymptotic OHϩH region, another part moves
towards the HHO geometry. The excitation of the bending
motion is so high that rotational barriers prevent a rapid dis-
sociation process for this part of the wave packet, which is
thus trapped in long-lived resonances. The bifurcation of the
wave packet has been described by Dixon et al.25 in a calcu-
lation of the photodissociation dynamics at 10.2 eV.
The nature of the resonances depends on energy. Below
9 eV, the resonances found in the spectrum correspond to
resonances in the HOH well, and can roughly be described
by bending and stretching quantum numbers. For energies
higher than the barrier between the HOH and HHO wells, the
resonances are very complicated, and show tunneling behav-
ior between the two minima. When the energy is signifi-
cantly above the barrier, the resonances look simple again. In
a classical description, the H atom makes a complete circle
around the OH molecule, while the distance to OH varies,
but is most of the time large (3–4a0). The spectrum shows
a rather regular progression over the entire energy range,
even though the nature of the resonances changes. The spec-
trum of D2O is found to be similar to that of H2O ͑Ref. 7͒
with a spacing of 0.08 eV instead of 0.10 eV. This is ex-
plained by the mass difference between D and H, the ratio is
slightly larger than ͱ2.
In Figs. 5͑a͒–5͑c͒, the new experimental and theoretical
indirect cross sections are shown for H2O, HOD and D2O.
All indirect cross sections show a regular progression with a
spacing of 0.10 for H2O, and 0.08 eV for HOD and D2O.
The agreement between experimental and theoretical spac-
ings is very good. When the spectra of H2O, HOD, and D2O
are overlaid, both the experimental and the computed ones, it
is seen that the HOD spectrum resembles the D2O spectrum
more than the H2O spectrum. This can be understood by
noting that in HOD the dissociation towards ODϩH is four
times as probable as the dissociation into OHϩD, because
the OH bond is weaker. What are seen in the spectrum of
HOD are sums of resonances in combined stretch-bend
modes of OD and departing H. Since the bending mode is
dominant, it can be expected that the spacing between the
resonances correlate with the reduced mass of the bending
mode. The reduced mass for the bending mode of the OD–H
complex with H far away is close to that of the OD–D com-
plex.
Also for the amplitudes of the resonances, the agreement
between theory and experiment is good, for all isotopomers.
A striking feature of the D2O spectrum is that the amplitude
of the modulation is almost a factor of 2 smaller than for
FIG. 5. Comparison of the indirect cross sections from the experiments ͑drawn curves͒ with those from theory ͑dashed curves͒. Panel ͑a͒ for H2O, panel ͑b͒
for D2O, and panel ͑c͒ for HOD.
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:
169.233.6.10 On: Wed, 03 Dec 2014 19:21:22

229
J. Chem. Phys., Vol. 120, No. 1, 1 January 2004
Absorption spectra of H2O
posed into components corresponding to a fast, direct disso-
ciation process and a process proceeding through resonances
by transforming to and from the time domain. In this way a
detailed comparison is made with the theoretical results
where an analogous separation into direct and resonant con-
tributions is made through manipulation of the autocorrela-
tion function. The direct experimental and theoretical spectra
are seen to be nearly identical, indicative of high accuracy in
the part of the potential energy surface related to direct dis-
sociation. Also the experimental and theoretical resonances
show nice agreement, both in position and in width, at least
for H2O and D2O. The width of the resonances, that are 0.10
eV apart for H2O and 0.08 eV for D2O, amounts to 0.04 eV
and is the consequence of overlapping resonances in the
spectra of molecules having different ground state rotational
levels. It is therefore expected that at still higher temperature
H2O. This is also the case for cross sections for individual
rotational states. An explanation is that the resonances in the
spectra are already partially overlapping. For D2O, for which
the resonances are closer to each other, the overlapping is
stronger than for H2O, which leads to a reduced indirect
cross section. But the same argument apparently does not
hold for HOD: For HOD the indirect cross section is even a
bit larger than that of H2O, both experimentally and theoreti-
cally.
There is only one feature for which experiment and
theory disagree: for HOD the onset of the resonances occurs
at lower energy in the theoretical spectrum than in the ex-
perimental spectrum. The experimentally observed trend in
the onsets of the resonances, first H2O, then D2O and HOD
last, is however reproduced by theory. For H2O and D2O the
onset of the resonances is seen to be hardly different in
theory and experiment. In the case of HOD theory predicts
two low energy resonances that are not present in the experi-
mental spectrum; also the amplitude of the next resonances is
higher in theory than in experiment.
In summary, good agreement between experiment and
theory is found for H2O and D2O, but for HOD a large
discrepancy in the onset of the resonances has been found.
Since agreement between the theoretical and experimental
direct spectra is excellent, one might jump to the conclusion
that the direct dissociation process is well described, but that
the indirect dissociation process where larger parts of the
potential energy surfaces are sampled still requires improve-
ment. Along what lines is as yet not clear. That the only
disagreement is found for HOD and not for H2O and D2O
suggests that the potential does not describe adequately sym-
metry effects. The energy of 9.3 eV, where the resonances
appear in the experimental HOD spectrum, corresponds
roughly to the energy of the barrier between the HOD and
HDO regions. Resonances below 9.3 eV correspond to large
amplitude stretch–bend vibrations, and those above 9.3 eV
to more complicated orbital involving a rotation of the de-
parting H around OD. It could be that especially the low
energy stretch–bend modes are sensitive to symmetry effects
that are not well described by the current potential energy
surfaces and couplings.
˜
the water B band absorption spectrum will be even more less
structured, while it should be possible to detect narrower
resonances in cold beams. For HOD the resonant component
of the theoretical spectrum shows two resonances at low en-
ergy that are not seen in the experimental spectrum. On the
origin of this unexpected discrepancy we can only speculate.
1 H. T. Wang, W. S. Felps, and S. P. Mc Glynn, J. Chem. Phys. 67, 2614
͑1977͒.
2 E. Segev and M. Shapiro, J. Chem. Phys. 77, 5604 ͑1982͒.
3 F. Flouquet and J. A. Horsley, J. Chem. Phys. 60, 3767 ͑1974͒.
4 M. Von Dirke, B. Heumann, K. Kuehl, T. Schroeder, and R. Schinke, J.
Chem. Phys. 101, 2051 ͑1994͒.
5 B. Heumann, K. Kuehl, K. Weide, R. Dueren, B. Hess, U. Meier, and S.
D. Peyerimhoff, Chem. Phys. Lett. 166, 385 ͑1990͒.
6 K. Mordaunt, M. N. R. Ashfold, and R. Dixon, J. Chem. Phys. 100, 7360
͑1994͒.
7 R. Van Harrevelt and M. C. Van Hemert, J. Chem. Phys. 112, 5787 ͑2000͒.
8 R. Van Harrevelt and M. C. Van Hemert, Chem. Phys. Lett. 371, 701
͑2003͒.
9 J. H. Fillion, R. van Harrevelt, J. Ruiz, N. Castillejo, A. H. Zanganeh, J. L.
Lemaire, M. C. van Hemert, and F. Rostas, J. Phys. Chem. A 105, 11414
͑2001͒.
10 L. C. Lee and M. Suto, Chem. Phys. 110, 161 ͑1986͒.
11 K. Yoshino, J. R. Esmond, W. H. Parkinson, K. Ito, and T. Matsui, Chem.
Phys. 211, 387 ͑1996͒.
12 R. Van Harrevelt, M. C. Van Hemert, and G. C. Schatz, J. Phys. Chem. A
105, 11480 ͑2001͒.
13 B.-M. Cheng, E. P. Chew, C.-P. Liu, M. Bahou, Y.-P. Lee, Y. L. Yung, and
M. F. Gerstell, Geophys. Res. Lett. 26, 3657 ͑1999͒.
14 B.-M. Cheng, C.-Y. Chung, M. Bahou, Y.-P. Lee, and L. C. Lee, J. Chem.
Phys. 117, 4293 ͑2002͒.
V. CONCLUDING REMARKS AND SUMMARY
15 S. G. Tilford, J. T. Vanderslice, and P. G. Wilkinson, Can. J. Phys. 43, 450
͑1965͒.
The photodissociation cross sections for H2O, D2O and
a mixture of H2O and D2O were measured in the 125–145
16 J. D. Simmons, A. M. Bass, and S. G. Tilford, Astrophys. J. 155, 345
͑1969͒.
˜
nm, B band, region. Synchrotron radiation was used. The
17 S. Ogawa and M. Ogawa, Can. J. Phys. 53, 1845 ͑1975͒.
18 C.-Y. Chung, E. P. Chew, B.-M. Cheng, M. Bahou, and Y.-P. Lee, Nucl.
Instrum. Methods Phys. Res. A 467–468, 1572 ͑2001͒.
19 J. W. Pyper, R. S. Newbury, and G. W. Barton, Jr., J. Chem. Phys. 46,
2253 ͑1967͒.
spectra of the pure isotopomers are in accord with the most
accurate previous measurements, the mixture enables the
derivation of the HOD cross section for the first time. Quan-
tum mechanical calculations are performed for all three iso-
topomers. In the wave packet calculations the effect of a
temperature of 300 K of the samples is accounted for by
explicitly including molecular rotation. The nonadiabatic
coupling between the second excited state and the ground
state is taken into account by the use of accurate ab initio
diabatic potentials. The experimental spectra are decom-
20 I. Kirshenbaum, in Physical Properties and Analysis of Heavy Water, ed-
ited by H. C. Urey and G. M. Murphy ͑McGraw-Hill, New York, 1951͒.
21 G. G. Balint-Kurti and M. Shapiro, Chem. Phys. 61, 137 ͑1981͒.
22 A. J. Dobbyn and P. J. Knowles ͑unpublished͒.
23 R. Van Harrevelt and M. C. Van Hemert, J. Chem. Phys. 112, 5777 ͑2000͒.
24 W. F. Chan, G. Cooper, and C. E. Brion, Chem. Phys. 178, 387 ͑1993͒.
25 R. N. Dixon, D. W. Huang, X. F. Yang, S. Harich, J. J. Lin, and X. Yang,
Science 285, 1249 ͑1999͒.
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:
169.233.6.10 On: Wed, 03 Dec 2014 19:21:22