Velocity distribution of CO desorbing from NiO(100)/Ni(100) after picosecond UV laser irradiation

Article abstract of DOI:10.1016/j.cplett.2005.12.057

The UV laser-induced desorption of CO from an epitaxial film of NiO(1 0 0) on Ni(1 0 0) yields a linear dependence of the desorption signal on the laser intensity with a cross-section of (8.3 ± 0.9) × 10^-18 cm². Rotational and vibrational temperatures of 870 K and 1340 K, respectively, are observed. State-resolved velocity distributions are observed, where one part can be described by a Maxwellian flux distribution with 〈E_kin〉/2k ranging from 1000 to 1800 K. In addition a faster component is present which gains in importance for increasing J″. A distinct rotation-translational coupling is found.

Full text of DOI:10.1016/j.cplett.2005.12.057

Chemical Physics Letters 420 (2006) 110–114
www.elsevier.com/locate/cplett
Velocity distribution of CO desorbing from NiO(100)/Ni(100)
after picosecond UV laser irradiation
a
b
c
d
d
c,
*
B. Redlich , A. Kirilyuk , T. Hoger , G. von Helden , G. Meijer , H. Zacharias
a
FOM Institute for Plasma Physics, ‘‘Rijnhuizen’’, Nieuwegein, 3439 MN, Netherlands
Research Institute for Materials, University of Nijmegen, 6525 ED Nijmegen, Netherlands
b
c
Physikalisches Institut, Westfa¨lische Wilhelms-Universita¨t, 48149 Mu¨nster, Germany
d
Fritz-Haber-Institut der Max-Planck-Gesellschaft, 14195 Berlin, Germany
Received 4 October 2005; in final form 14 December 2005
Available online 18 January 2006
Abstract
The UV laser-induced desorption of CO from an epitaxial film of NiO(100) on Ni(100) yields a linear dependence of the desorption
signal on the laser intensity with a cross-section of (8.3 0.9) · 10ꢀ18 cm2. Rotational and vibrational temperatures of 870 K and
1340 K, respectively, are observed. State-resolved velocity distributions are observed, where one part can be described by a Maxwellian
flux distribution with ÆEkinæ/2k ranging from 1000 to 1800 K. In addition a faster component is present which gains in importance for
increasing J00. A distinct rotation–translational coupling is found.
ꢀ 2005 Elsevier B.V. All rights reserved.
1. Introduction
adsorbed molecules. At photon energies above the band
gap, substrate electronic excitations to the conduction
band, thereby creating electron – hole pairs, followed by
rapid intraband relaxation to the bottom of this band con-
stitutes an excitation scheme similar to that proposed for
metal substrates. For physisorbed molecules direct excita-
tions like in the gas phase are possible, and for chemi-
sorbed adsorbates also direct optical transitions from the
substrate to electronically excited states of the adsorbate/
substrate complex may be considered.
On oxide surfaces earlier experiments using nanosecond
UV laser radiation have been carried out for NO desorp-
tion from NiO(100) [2,3] and Cr2O3(0001) [4]. Femtosec-
ond laser excitations have concentrated on NO desorption
from NiO(100) [5]. On these surfaces NO adsorbs in a bent
geometry [6,7], while in an electronically excited NOꢀ con-
figuration the equilibrium geometry is upright [8–10]. Car-
bon monoxide on the other hand adsorbes in the ground
state in nearly linear configuration (# ꢁ 7ꢁ) [11–14] and
for electronically excited CO a bent geometry is expected.
Experimental studies for laser-induced CO desorption have
been performed mostly for metal substrates [15–19] using
femtosecond laserpulses. Nanosecond UV laser desorption
The understanding of photochemical processes and
reactions at solid surfaces is of fundamental importance,
as well as of practical relevance for a variety of processes
like heterogeneous photocatalysis or photochemical pro-
cesses and for the micro and nano structuring of surfaces.
For molecules adsorbed on metallic substrates detailed
studies have been performed with a variety of adsorbates,
substrates and photon energies [1]. It is commonly accepted
that on these surfaces photoexcitation of the electronic sys-
tem of the metal constitutes the primary process. Besides
energy dissipation to the bulk scattering of the hot elec-
trons at the adsorbate generates vibrational and electronic
excitations which after coupling to the desorption coordi-
nate eventually lead to desorption of the adsorbate.
Similar studies on insulating substrates with large band
gaps remain comparatively sparse. In this case the creation
of a gas of hot electrons with a broad energy distribution
may no longer be the dominant excitation pathway of
*
Corresponding author.
E-mail address: hzach@uni-muenster.de (H. Zacharias).
0009-2614/$ - see front matter ꢀ 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.cplett.2005.12.057

B. Redlich et al. / Chemical Physics Letters 420 (2006) 110–114
111
of CO has been performed for Cr2O3(0001) [20–22] and
NiO(111) [23] substrates. The latter surface differs from
the one investigated in the present study in that the polar
NiO(111) surface exhibits terraces with either pure Ni or
pure O termination, while NiO(100) is terminated by a
neutral-Ni–O-configuration.
Ho¨nl–London factors have to be taken into account. How-
ever, since we are only detecting Q-branch transitions, the
1
Ho¨nl–London factors for a R–1R transition are indepen-
dent of the rotational quantum number [26]. By varying
the time delay between desorption and probe laser pulses
information about the kinetic energies of the desorbing
molecules is obtained.
2. Experimental
3. Results and discussion
The experiment is performed in an ultrahigh vacuum
(UHV) chamber with a base pressure of less than
2 · 10ꢀ10 mbar. Surface structure and cleanliness are con-
trolled with LEED and Auger electron spectroscopy.
Cleaning of the sample is accomplished by Ar ion bom-
bardment at low kinetic energy. The epitaxial film of
NiO(100) is prepared by repeated cycles of oxidation with
O2 at pressures up to 5 · 10ꢀ6 mbar and annealing at tem-
peratures between 570 and 700 K [7]. The sample can be
cooled down to 110 K and heated up to 1000 K. The initial
preparation of the CO adsorbate layer occurs via back-
ground dosing at 110 K. At this temperature about half a
monolayer of CO can be adsorbed on the NiO(100) thin
film [24]. For redosing between the desorption laser pulses
a pulsed valve is employed to keep a steady state coverage
of carbon monoxide.
The desorption of the CO molecules is initiated by the
radiation of a frequency quadrupled Nd:YAG laser at
k = 266 nm (hm = 4.66 eV) with a pulse duration of 80 ps
and operating at 5 Hz repetition rate. The p-polarized radi-
ation was incident under 45ꢁ with respect to the surface
normal. Pulse energies up to 1.9 mJ are applied to a spot
size of about 10 mm2 area, resulting in fluences of up to
19 mJ/cm2.
The desorption yield increases linearly with laser inten-
sity up to a maximum applied pulse energy of 1.9 mJ,
which corresponds to a fluence and intensity of 19 mJ/
cm2 and 2 · 1011 W/cm2, respectively. For such a linear
intensity dependence the desorption yield can be expressed
by an effective desorption cross-section rdes via
lnðN=N0Þ ¼ ꢀrdes ꢂ nph;
ð1Þ
where N0, N, nph denote the number of adsorbate mole-
cules before and after desorption and the photon flux,
respectively. The cross-section is related to the photon flux
applied to the CO/NiO(100) system. The desorption cross-
section has been determined on the rotational Q-branch
bandhead of the (v00 = 0) state at a probe laser delay time
of 25 ls, where the arrival time distribution peaks. Fig. 1
displays the decay of the CO desorption yield. Here the log-
arithm of the desorption signal is plotted versus the applied
photon flux. A strong decrease of the signal is observed up
to an accumulated flux of about 1.5 · 1017 photons/cm2,
followed by a more shallow diminution. From the initially
strong decrease an effective desorption cross-section of
r
des = (8.3 0.9) · 10ꢀ18 cm2 is derived. This channel ac-
counts for more than 80% of the desorption flux. The
slower decay at higher total photon flux may be described
by a cross-section of rdes = (1.8 0.2) · 10ꢀ18 cm2.
Such a bimodal decay of the coverage is also observed
for UV laser induced desorption from NiO(111) at 4 eV
The desorbed CO molecules are detected rovibrational
state selectively by resonance enhanced multiphoton ioni-
1
sation (2 + 1 REMPI) via the B R+(v0 = 0,1) X
1R+(v00 = 0,1) transition [25]. The generated photoions are
measured by microchannel plates. Although this detection
scheme suffers from an unresolved Q-branch at low J00 it
was chosen, because it avoids problems which might be
caused by a possible strongly varying rotational alignment
as was observed for desorption from Cr2O3(0001) [21].
The tuneable (k ꢁ 230 nm) probe laser radiation with a line
width of ꢁ0.21 cmꢀ1 is provided by a frequency doubled
dye laser pumped by the third harmonic of a Q-switched
Nd:YAG laser. Operating at 10 Hz repetition rate UV
pulse energies of 100 lJ with 8 ns duration are applied.
The probe laser beam has a diameter of 0.1 mm and a
two-photon Rayleigh length of 44 mm. It intersected the
desorption flux at a distance of 20 mm in front of the sur-
face. No attempt has been made to restrict the polar angle
of detection, exciting thus desorbing molecules in an angu-
lar range of 45ꢁ. Operating the electronics in a toggle
mode enabled an efficient registration of background CO
molecules from the gas phase between desorption laser
pulses. At the probe laser intensity applied the B–X
two-photon transition of CO is not saturated, thus the
Fig. 1. Semi-logarithmic plot of the desorption yield as a function of
applied photon flux.

112
B. Redlich et al. / Chemical Physics Letters 420 (2006) 110–114
photon energy [23]. In that experiment cross-sections of
3.3 · 10ꢀ18 cm2 and 4.5 · 10ꢀ19 cm2 for the fast and slow
decay, respectively, were measured. Since the ground state
binding energy of CO on nickel oxide is higher for the
(100) than for the (111) surface, the generally higher
cross-sections in the present study might be a first hint
to a more effective electronic excitation of the system. Sim-
ilary, the desorption of NO from NiO(100) shows a bimo-
dal decay [3,5]. While the initial cross-section for CO
desorption is lower by a factor of 2.3 than the correspond-
ing one for NO of 1.9 · 10ꢀ17 cm2, the slow decay shows a
nearly one order of magnitude higher cross-section (NO:
2.7 · 10ꢀ19 cm2). The slow decay might originate from
molecules adsorbed on step and defect sites of the
NiO(100) surface. The cross-section for CO desorption
from Cr2O3(0001) of 3.5 · 10ꢀ17 cm2 at k = 193 nm is sig-
nificantly larger [20]. According to mechanisms where
desorption is induced by an electronic transition to an
excited state (DIET) a large desorption cross-section
implies a comparatively long lifetime of the electronically
excited state, estimated here to be in the range of about
10–30 fs.
The vibrational population can be estimated from the
relative intensities of the B–X (0–0) and (1–1) bands. The
Franck–Condon factors for the two transitions are nearly
identical [30], and the ionisation cross-section can be
assumed to be also the same due to the Rydberg character
of the B state. In addition the photon energies to excite the
two bands and thus ionise the B state differ only by about
30 cmꢀ1. Then from the spectra a vibrational population of
about N(v00 = 1) = 10% can be deduced, corresponding to a
vibrational temperature of Tvib = 1340 K. This vibrational
temperature is within the experimental uncertainly identi-
cal to that observed for desorption from NiO(111) [23].
The population translates to an average vibrational energy
of about ÆEvibæ = 214 cmꢀ1. This value compares well with
the average vibrational energy in the (v00 = 1) state of NO
desorbing from NiO(100) using nanosecond [2] or femto-
second [5] laser pulses. Higher vibrationally excited states
could not be observed, probably due to the high predisso-
1
ciation rate of B R+(v0 P 2).
The main result of this letter concerns the velocity distri-
bution of the desorbing CO molecules. For this purpose the
probe laser pulse has been continuously delayed up to
220 ls with respect to the desorption laser pulse. At a flight
distance of 20 mm this corresponds to a slow limit of
90 msꢀ1 velocity which has been recorded. This arrival time
signal S(t) has to be transformed first into a flight-time dis-
tribution I(t) by
Desorbing CO molecules are rotational state selectively
detected in both the vibrational ground and the first vibra-
tionally excited state. With the given spectral resolution the
Q-branch heads at 230.106 nm and at 230.259 nm of the
(0–0) and (1–1) band, respectively, could not be resolved
into their rotational components, but for J00 P 12 single
rotational states could be addressed. In (v00 = 0) rotational
states up to J00 = 37 and in (v00 = 1) rotational states up to
J00 = 22 are observed. The former limit is given by the
threshold to predissociation in the B 1R+ (v0 = 0) state
Z
v¼ðLþlÞ=t
l
t
SðtÞ /
Iðv0Þdv0 ꢃ Iðv ¼ L=tÞ
ð2Þ
v¼L=t
from which directly follows that
) Iðv ¼ l=tÞ / tSðtÞ
ð3Þ
1
which begins at J0 = 38, while in the B R+ (v0 = 1) state
slow predissociation sets in at J0 = 17 [27,28]. An abrupt
decrease in intensity above the predissociation limits is
not observed, since the predissociation rate of about
1010 sꢀ1 is slow compared to the ionisation rate [28].
From the rotationally resolved part of spectra taken at
where L denotes the flight distance, l the diameter of the
probe laser and t is the arrival time.
Fig. 2 shows the results for various rotational states in
v00 = 0. The individual velocity distributions are normalized
different sensitivities a rotational temperature of Trot
=
(870 150) K can be deduced. Although this value should
be taken with some caution since the low-J00 part is not
resolved, a significant overpopulation of the low-J00 states
as observed for desorption from NiO(111) [23] is not
reflected in the spectra. For NiO(111) rotational tempera-
tures of Trot = 200 and 640 K have been observed [23]. The
higher rotational excitation for the NiO(100) substrate
supports the conclusion of a larger rotational corrugation
in the excited state potential [29]. The spectra are taken
at a delay time of 25 ls, corresponding to a velocity of
800 msꢀ1 for molecules desorbing in the direction of the
surface normal. Although a significant rotation-transla-
tional coupling is observed, see below, a resulting correc-
tion of the rotational population lies well within the
indicated uncertainty from different spectral scans. Also
the above mentioned uncertainty due to unresolved low-
J00 lines may account for a greater uncertainty in the aver-
age rotational energy.
Fig. 2. Velocity distributions of CO (v00 = 0) desorbing in different
rotational states. For J00 = 0ꢀ6 and J00 = 23 fits with Maxwellian flux
distributions of ÆEkin/2kæ = 950 K and 1200 K, respectively, are indicated.

B. Redlich et al. / Chemical Physics Letters 420 (2006) 110–114
113
to one at their maximum value and are displayed offset for
clarity. For low to medium J00 the distributions are very
similar and show a Maxwellian like velocity distribution.
The maxima of the distributions appear around
1100 msꢀ1. These distributions can approximately be
described by kinetic temperatures between about ÆEkin
/
2kæ = 950 K for J00 = 0ꢀ6 and 1200 K for J00 = 23. It is fur-
ther evident that in addition to the Maxwellian distribution
a component with velocities between 1500 and 3000 msꢀ1 is
present. Beginning at J00 = 27 and pronounced at J00 = 30
and 32, this high velocity component gains in relative
strength compared to the Maxwellian part mainly observed
for the low-J00 molecules. Also the maximum of the distri-
bution shifts to higher velocities around about 1500 msꢀ1
.
Fig. 3 shows corresponding results for CO molecules
desorbing in the vibrationally excited state, now for rota-
tional states up to J00 = 20. For low J00 the distributions
peak slightly below 1000 msꢀ1 and are approximately
described by ÆEkin/2kæ = 1000 to 1200 K. For higher J00 in
(v00 = 1), J00 = 17 and 20, the maximum shifts to about
1300 msꢀ1 and the best Maxwellian fits increase to 1400
and 1800 K, respectively. Again a fast velocity component
increases in relative strength with increasing J00.
Fig. 4. Average kinetic energy ÆEkinæ of desorbing CO as a function of
rotational quantum number; triangles: (v00 = 0), squares: (v00 = 1).
Table 1
Average energies for desorption from nickel oxide in [cmꢀ1
]
CO
NO
NiO(100)
NiO(111) [23]
NiO(100) [5]
ÆEkin
ÆErot
æ
æ
2550
605
1600
362
1774
308
409
Fig. 4 shows the average kinetic energy of the states
measured as a function of their rotational quantum num-
ber for CO molecules desorbing in both vibrational states.
The average energy increases from about 250 meV at low
J00 to more than 500 meV at J00 = 30. This increase is more
monotonic for the vibrationally excited state while in the
ground state the kinetic energy is approximately indepen-
dent of J00 up to J00 = 27 with a significant increase at higher
J00. It is therefore evident that a strong rotation–translation
coupling is observed for CO desorption from NiO(100).
Table 1 summarizes the average energies for CO and NO
desorption from NiO(100) and NiO(111).
ÆEvib
æ
214a
236a
a
For the vibrational energies the population in (v00 = 1) only has been
taken into account.
The observed velocity distributions differ significantly
from those observed for the NiO(111) substrate [23].
There velocity distributions which can completely be
described by Maxwellian flux distributions of ÆEkin
/
2kæ = (1150 150) K were found and a rotation–transla-
tion coupling could not be established up to J00 = 23. In
the present study for the NiO(100) substrate the majority
of the desorption flux up to medium J00 can also be
described by a kinetic temperature of about 1000 to
1200 K. However, even at low J00 a fast component is pres-
ent which increases in relative importance with J00. This
also clearly indicates that the electronically excited state
is rotationally more corrugated or has a significantly dif-
ferent equilibrium position in the angular coordinate than
in the CO/NiO(111) system.
The observed velocity distributions are also very differ-
ent from corresponding results for NO desorption from
NiO(100) [3,5]. There, a clear bimodal distribution was
observed with
a slow component peaking around
300 msꢀ1 and a fast one around 1100 msꢀ1 at low J00. The
distributions in this case were clearly non-Maxwellian,
especially the fast one which displayed a nearly Gaussian
distribution. This fast component showed a strong rota-
tion–translation coupling, as in the present case for CO
desorption.
Laser induced desorption of NO from metal surfaces is
generally discussed in terms of a temporary formation of a
NOꢀ state [1]. The dynamics then proceeds via an Anton-
Fig. 3. Same as Fig. 2 for CO (v00 = 1). Maxwellian velocity distributions
with ÆEkin/2kæ = 1000 K for J00 = 3ꢀ6 and ÆEkin/2kæ = 1200 K at J00 = 14
are shown as thin lines.

114
B. Redlich et al. / Chemical Physics Letters 420 (2006) 110–114
iewicz type mechanism [31], where this electronically
excited state is more strongly bound to the surface with a
shorter bond length than the ground state. For CO desorp-
tion from metal surfaces (Cu,Pt,Ru) [15–19] the unoccu-
pied 2p* level is of importance. After excitation with UV
radiation electrons at the Fermi level of the metal can
acquire sufficient energy to populate the antibonding
hybride level in the Blyholder model. For near-IR radia-
tion this is not longer the case. Beyond this one electron
Blyholder model Jennison et al. [32] took correlation effects
into account. In their adiabatic model the 2p* level devel-
ops two antibonding levels, one energetically close to that
of the Blyholder model and one at significantly lower ener-
gies. Loy and co-workers [33] conclude that this lower level
can probably account for CO desorption by 800 nm radia-
tion. This eventually leads to desorption of the CO mole-
cule from metal surfaces.
References
[1] see, e.g. H.L. Dai, W. Ho (Eds.), Laser Spectroscopy and Photo-
chemistry on Metal Surfaces, World Scientific, Singapore, 1995.
[2] Th. Mull, B. Baumeister, M. Menges, H.-J. Freund, D. Weide, C.
Fischer, P. Andresen, J. Chem. Phys. 96 (1992) 7108.
[3] M. Menges, B. Baumeister, K. Al-Shamery, H.-J. Freund, C. Fischer,
P. Andresen, J. Chem. Phys. 101 (1994) 3318.
[4] M. Wilde, O. Seifert, K. Al-Shamery, H.-J. Freund, J. Chem. Phys.
111 (1999) 1158.
[5] G. Eichhorn, M. Richter, K. Al-Shamery, H. Zacharias, J. Chem.
Phys. 111 (1999) 386.
[6] H. Kuhlenbeck, G. Odo¨rfer, R. Jaeger, G. Illing, M. Menges, Th.
Mull, H.-J. Freund, M. Po¨hlchen, V. Staemmler, S. Witzel, C.
Scharfschwerdt, K. Wennemann, T. Liedke, M. Neumann, Phys. Rev.
B 43 (1991) 1969.
[7] J.T. Hoeft, M. Kittel, M. Polcik, S. Bao, R.L. Toomes, J.H. Kang,
D.P. Woodruff, M. Pascal, C.L.A. Lamont, Phys. Rev. Lett. 87 (2001)
086101.
[8] T. Kluner, H.-J. Freund, F. Freitag, V. Staemmler, J. Chem. Phys.
¨
104 (1996) 10030.
For the desorption of NO from oxide surfaces direct
charge transfer excitations seem to play an important role,
as has been found theoretically [8–10] and is also confirmed
experimentally for NiO as substrate [34]. For CO most
experimental and theoretical information is available for
Cr2O3(0001) as substrate. Exciting this system with
6.4 eV radiation bimodal velocity distributions were
observed for CO (v00 = 0) with one very slow component
of only 300 msꢀ1 pointing towards a thermal origin of these
molecules and a fast one with about 1300 msꢀ1. For CO
(v00 = 1) only a fast channel with velocities around
900 msꢀ1 appears [20]. For the electronic dynamics direct
excitation of CO from the 5r to the antibonding 2p* state
was considered in a 3D calculation [35]. A strong rotational
corrugation of both potential energy surfaces was impor-
tant to explain essential experimental results for the CO/
Cr2O3(0001) system. It seems conceivable that also for
the CO/NiO(100) system a direct 5r ! 2p* excitation
within the CO adsorbate constitutes the first excitation
step. Whether the interaction of the excited electron with
the substrate is strong enough to cause the development
of two antibonding levels due to correlation effects,
remains open. In the present experiment the energy of the
exciting photon of 4.66 eV, however, is significantly lower
than the gas phase 5r ! 2p* (a3P) transition energy,
which is close to 6 eV. Since we observe a linear depen-
dence of the desorption yield on the laser intensity and find
a large desorption cross-section, the existence of an ener-
getically lower antibonding level is corroborated.
[9] T. Kluner, H.-J. Freund, V. Staemmler, R. Kosloff, Phys. Rev. Lett.
¨
80 (1998) 5208.
[10] C.P. Koch, T. Kluner, H.-J. Freund, R. Kosloff, J. Chem. Phys. 119
¨
(2003) 1750.
[11] D. Cappus, J. Klinkmann, H. Kuhlenbeck, H.-J. Freund, Surf. Sci.
325 (1995) L 421.
[12] M. Kittel, J.T. Hoeft, S. Bao, M. Polcik, R.L. Toomes, J.H. Kang,
D.P. Woodruff, M. Pascal, C.L.A. Lamont, Surf. Sci. 499 (2002) 1.
[13] G. Pacchioni, C. Di Valentin, D. Dominguez-Ariza, F. Illas, T. Bredow,
T. Kluner, V. Staemmler, J. Phys. Condens. Mater. 16 (2004) S2497.
¨
[14] A. Rohrbach, J. Hafner, G. Kresse, Phys. Rev. B 69 (2004) 075413.
[15] L.M. Struck, L.J. Richter, S.A. Buntin, R.R. Cavanagh, J.C.
Stephenson, Phys. Rev. Lett. 77 (1996) 4576.
[16] F.-J. Kao, D.G. Busch, D. Gomes da Costa, W. Ho, Phys. Rev. Lett.
70 (1993) 4098.
[17] S. Deliwala, R.J. Finlay, J.R. Goldman, T.H. Her, W.D. Mieher, E.
Mazur, Chem. Phys. Lett. 242 (1995) 617.
[18] S. Funk, M. Bonn, D.N. Denzler, Ch. Hess, M. Wolf, G. Ertl, J.
Chem. Phys. 112 (2000) 9888.
[19] L. Cai, X. Xiao, M.M.T. Loy, Surf. Sci. 464 (2000) L 727.
[20] K. Al-Shamery, I. Beauport, H.-J. Freund, H. Zacharias, Chem.
Phys. Lett. 222 (1994) 107.
[21] I. Beauport, K. Al-Shamery, H.-J. Freund, Chem. Phys. Lett. 256
(1996) 641.
[22] S. Borowski, T. Kluner, H.-J. Freund, I. Klinkmann, K. Al-Shamery,
¨
M. Pykavy, V. Staemmler, Appl. Phys. A 78 (2004) 223.
[23] M. Asscher, F.M. Zimmermann, L.L. Springsteen, P.L. Houston, W.
Ho, J. Chem. Phys. 96 (1992) 4808.
[24] R. Wichtendahl, M. Rodriguez-Rodrigo, U. Ha¨rtel, H. Kuhlenbeck,
H.-J. Freund, Surf. Sci. 423 (1999) 90.
[25] G.W. Loge, J.J. Tiee, F.B. Wampler, J. Chem. Phys. 79 (1983) 196.
[26] J.B. Halpern, H. Zacharias, R. Wallenstein, J. Mol. Spectrosc. 79
(1980) 1.
[27] S.G. Tilford, J.T. Vanderslice, J. Mol. Spectrosc. 26 (1968) 419.
[28] H. Rottke, H. Zacharias, Opt. Commun. 55 (1985) 87.
[29] E. Hasselbrink, Chem. Phys. Lett. 170 (1990) 329.
[30] T.A. Carlson, N. Duric, P. Erman, M. Larsson, Z. Phys. A 287 (1978)
123.
[31] P.R. Antoniewicz, Phys. Rev. B 21 (1980) 3811.
[32] D.R. Jennison, E.B. Stechel, A.R. Burns, J. Electr. Spectr. Rel. Phen.
72 (1995) 9.
[33] L. Cai, X. Xiao, M.M.T. Loy, Surf. Sci. 492 (2001) L 688.
[34] H. Zacharias, G. Eichhorn, R. Schliesing, K. Al-Shamery, Appl.
Phys. B 68 (1999) 605.
In conclusion, the high rotational excitation supports
the notion of a bent geometry for the electronically
excited CO adsorbate. This is also corroborated by the
observed rotation–translational coupling, which is a man-
ifestation of a Franck–Condon excitation from the
ground state at a different bond angle than the equilib-
rium geometry in the excited adsorbate. Further, the
strong vibrational excitation of the desorbing molecules
points to a bond length in the excited state different from
the ground state.
[35] S. Thiel, M. Pykavy, T. Kluner, H.-J. Freund, R. Kosloff, V.
¨
Staemmler, J. Chem. Phys. 116 (2002) 762.