Laser catalysis of acrylonitrile on copper surfaces

Article abstract of DOI:10.1016/S0009-2614(00)00344-4

The electron emission and CN^- yield produced by laser irradiation of acrylonitrile (ACN) adsorbed on a polycrystalline Cu surface have been investigated over the 930-953 cm^-1 wavenumber range. Both emissions show a clear wavelength dependence with a ca. 20 cm^-1 red shift with respect to the gas-phase ACN spectrum. The laser catalysis CN^- yield is discussed in light with both substrate and adsorbate-mediated mechanisms, which seem to be controlling this (charge-transfer) surface reaction.

Full text of DOI:10.1016/S0009-2614(00)00344-4

5
May 2000
Chemical Physics Letters 321 Ž2000. 349–355
www.elsevier.nlrlocatercplett
Laser catalysis of acrylonitrile on copper surfaces
J.O. C a´ ceres , J. Tornero L o´ pez, A. Gonz ´a lez Ure n˜ a)
1
Unidad de L a´ seres y Haces Moleculares, Instituto Pluridisciplinar, UniÕersidad Complutense de Madrid, Juan XXIII,
E-28040 Madrid, Spain
Received 12 November 1999; in final form 7 March 2000
Abstract
y
The electron emission and CN yield produced by laser irradiation of acrylonitrile ŽACN. adsorbed on a polycrystalline
y
1
Cu surface have been investigated over the 930–953 cm
dependence with a ca. 20 cm
wavenumber range. Both emissions show a clear wavelength
y
1
y
red shift with respect to the gas-phase ACN spectrum. The laser catalysis CN yield is
discussed in light with both substrate and adsorbate-mediated mechanisms, which seem to be controlling this Žcharge-trans-
fer. surface reaction. q 2000 Elsevier Science B.V. All rights reserved.
1
. Introduction
in which the thermal Žphotodesorption. mechanism,
due to resonant substrate heating w6–8x, seems to be
well accepted w8x. There is, however, an increasing
Photon beams, particularly intense lasers, have
been extensively used in recent years to induce or
enhance chemical interaction at gas–solid interfaces
w1,2x. Both the basic scientific interest and the tech-
nological application in material processing for new
materials drive the rapid expansion of the field,
which can be generally referred to as laser-induced
gas–surface chemistry w2–5x.
evidence that non-thermal photochemical processes
play an important role in certain gas–surface systems
when resonant IR radiation is used w9–11x.
Thus, the surface femtochemistry of O and CO
2
on PtŽ111. has been investigated with the finding
that non-thermal electron distributions play a signifi-
cant role in both O desorption and CO formation
2
2
Electronic relaxation on metal surfaces is ex-
tremely rapid and consequently, many of the surface
reactions promoted by laser irradiation on metal
substrates in contact with optically non-absorbing
gases are often interpreted as thermally activated
processes. This is the case of IR-induced desorption
w9x. More recently, the desorption and oxidation of
CO on Ru Ž0001. have been investigated with fem-
tosecond IR laser pulses w12x. This study led to the
interesting conclusion that while CO desorption is
caused by a phonon-mediated mechanism, the CO
oxidation to CO product was initiated by hot sub-
2
strate electrons.
In the present work, as the laser pulse duration
Žsee below. is of the order of microseconds both
electrons and phonons are in equilibrium and only
)
Corresponding author. Fax: q34-91-3943265; e-mail:
laseres@eucmax.sim.ucm.es
1
CONICET postdoctoral fellow.
thermal surface chemistry should be, in principle,
0
009-2614r00r$ - see front matter q 2000 Elsevier Science B.V. All rights reserved.
PII: S0009-2614Ž00.00344-4

3
50
J.O. C a´ ceres et al.rChemical Physics Letters 321 (2000) 349–355
expected w13x. Nevertheless, the possibility of some
2. Experimental
non-thermal effects, i.e. that other parameters, in
addition to the temperature, play a role in the surface
chemistry, cannot be ruled out. This could be the
case for substrate–adsorbate charge transfer reac-
tions in which the Žadsorbate. dissociative electron
attachment rate is fast enough to be competitive with
energy redistribution among the vibrational modes of
the adsorbate following laser excitation. Under these
conditions, in spite of the fast electron relaxation to a
thermal ŽFermi–Dirac. distribution and the adsor-
The experimental setup is schematically shown in
Fig. 1. The main parts of it are: a the line-tunable
Ž .
IR CO2 laser focused on the Cu surface, b the
argon ion gun for cleaning the surface and c the
linear time-of-flight mass spectrometer TOFMS .
Ž .
Ž .
Ž
.
Experiments were performed with an IR TEA–
CO laser Coherent Hull XL-370 TS , which can be
Ž
.
2
tuned over the 9–11 mm wavelength range. The
laser enters in an ultrahigh vacuum chamber UHVC
Ž
.
bate–substrate energy transfer, some vibrational se-
lectivity may be retained when local mode excitation
changes the adsorbate electron affinity, and so the
barrier for its electron attachment, at a rate faster or
comparable to that of adsorbate vibrational energy
redistribution. These conditions would then originate
a selective chemistry with a different reaction yield
from that of the non-selective case, in which only an
enhanced heating, upon vibrational excitation of the
adsorbate, would take place. The criterion that exci-
tation energy must remain relatively localized during
the reaction in order to achieve mode-selective
chemistry is well established for many fast direct
reactions w14x. We anticipate that this seems to be the
in which a mechanically polished polycrystalline
copper surface 99.9998% Aldrich Chemical was
Ž
.
allocated. The typical laser spot area on the surface
2
was 0.05 cm . A pure TEM mode was always used
oo
for all wavelengths. Under the present conditions,
the surface temperature, estimated with the model
developed by Burgess 17 never exceeded 810 K. In
addition, careful monitoring of both spatial distribu-
tion and temporal profile of laser pulses revealed the
w x
absence of hot spots. A spectrum analyzer Macken
Ž
Instrument model 16-A. was used to tune the laser
.
to the desired wavelength. An Ar ion gun VG
Microtech Instruments model AG5000, 5 keV ions
sputtering at 298 K cleaned the surface before each
experiment. The cleanliness of the Cu surface was
Ž
.
case for the laser-induced dissociative ionization of
acrylonitrile ŽC H N or vinylcyanide., denoted by
3
3
ACN hereafter, adsorbed on copper as investigated
here.
The present work reports on the investigation of
laser-induced ionisation processes at the gas–surface
interface using a TEA–CO laser for the excitation
2
of molecules. The surface used was polycrystalline
Cu, which has a rather low work function Ž4.64 eV.
w15x. The adsorbate was the acrylonitrile molecule
Želectron affinity 1.247 eV. w15x, which strongly
absorbs in the n13 band associated with the RCH–CN
bond resonant with transitions between 954.55 and
y1
9
1
31.00 cm
corresponding to the 10PŽ8. and
0PŽ34. lines of the CO laser w16x, respectively.
Our main motivation was to look for gas–surface
2
ionization processes induced by laser radiation. Thus,
the experiment is aimed at investigating the parent or
any fragment ion yield as a function of the laser
wavelength and power as well as other surface pa-
rameters, which characterize the gas–surface interac-
tion, in a search for vibrationally enhanced effects in
the acrylonitrile dissociative attachment.
Fig. 1. Schematic view of the experimental setup including the
CO2 laser, Ar ion gun and TOFMS.

J.O. C a´ ceres et al.rChemical Physics Letters 321 (2000) 349–355
351
Table 1
Experimental conditions
3
. Results
The appearance of a CNy peak in the TOF
Laser system
Wavelength range: 9–11 mm
spectrum when ACN is introduced in the vacuum
chamber and adsorbed on the surface is noticeable as
is displayed in Fig. 2. This spectrum persisted after
the ACN leak was closed and the background pres-
sure reached the low 10
figure shows the same spectrum obtained before the
ACN was added. Only the fast electron signal can be
seen. For a better illustration the inset shows a
typical TOF spectrum of the electron signal together
with that of the laser pulse.
2
Typical fluence: f4 Jrcm
Temporal profile per pulse: FWHN peak 100 ns Žtail 3 ms.
Pulse frequency: 2 Hz
Spot size: 0.05 cm
Vacuum
Background pressure: (3=10 Torr
Acrylonitrile range pressure: 2.0=10 to 2.0=10 Torr
Copper temperature: 300 K
2
y9
Torr range. The top
y9
y8
y7
Surface cleaning: 5 keV ion sputtering with Ar ion gun
y
Both CN and electron signals were monitored at
y1
different wavelengths between 953 and 930 cm at
constant laser fluence. Fig. 3 shows the wavelength
checked by detection of impurities in the TOF signal.
No negative or positive ion desorption was observed
from the clean Cu surface for laser fluences of the
2
order of 4 Jrcm , which was the typical value used.
The Cu surface can be heated resistively by a
nicrom wire up to 675 K. This heat treatment was
employed to eliminate surface impurities because
they may alter the electronic properties of the surface
and, consequently, the negative ion yield.
Acrylonitrile Ž99q% Aldrich Chemical. is a col-
orless liquid at room temperature and atmospheric
pressure, therefore volatile impurities were removed
from the samples through freeze and thaw cycles.
Gas chromatographic analyses and IR spectroscopy
showed the absence of impurities in the ACN sam-
ple. During the experiments the acrylonitrile pressure
y8
was kept at 5.0=10 Torr. Since the laser repeti-
tion rate was 2 Hz, the acrylonitrile exposure during
y6
the laser shot interval was 0.025 L Ž1 Ls1=10
Torr s..
A 50 cm linear TOFMS mounted with a dual
microchannel plate ŽMCP. detector ŽComstock CP-
6
25Cr50F. and optimized for negative ion detection
was used. The signal was then further amplified with
an oscilloscope Tektronix ŽTDS 540. and finally
stored and processed in a computer. When the sig-
nal-to-noise demanded better resolution or discrimi-
nation, a multichannel signal averager ŽStanford Re-
search System. was used to obtain the TOF spectra.
Typically the average of 100 TOF spectra resulted in
good statistics. Table 1 lists the most relevant experi-
mental parameters and conditions of the present work.
y9
Fig. 2. Negative ion TOF spectra obtained at 3=10
Torr of
background pressure and rest of conditions specified in Table 1
with a laser excitation wavenumber of 945.48 cm . Ža. Only the
electron signal is present with no ACN added. The inset shows a
typical electron TOFMS signal Žbottom. with the laser signal
y1
incorporated Žtop. for a better illustration. Žb. Same as in Ža. but
y8
with ACN added Ž pACN f10
Torr..

3
52
J.O. C a´ ceres et al.rChemical Physics Letters 321 (2000) 349–355
dependence of the integrated signals. It can be seen
y
ŽFig. 3a. that the maximum CN signal is obtained
y1
for the 10P32 laser line Ž933.7 cm . which also
corresponds to the maximum electron yield ŽFig. 3b.
when acrylonitrile is present. With no acrylonitrile, a
random dispersion around a constant value indicates
no wavelength dependence for the emitted electrons
as depicted in Fig. 3c.
Fig. 3b,c have the same arbitrary units for the
electron signal. The constant value Žwithin experi-
y
y
Fig. 4. Ža. Solid square: CN re signal ratio as a function of the
laser wavenumber. Open square. Same as before, but shifted 20.17
y1
cm
to the blue part of the spectrum. In both cases a solid
Ždashed. line was drawn through the points to guide the eye. Žb.
IR ACN gas-phase spectrum. Vertical dashed lines are drawn
through the peaks to reinforce the similarity between these two
spectra.
mental error. of the electron signal in the absence of
ACN Žsolid line of Fig. 3c. has also been marked by
a dashed-line in Fig. 3b. Thus, except for the red part
of the spectrum, a small decrease in the electron
yield can be noticed when ACN is added.
A closer look at the red part of the spectrum
y
suggests a correlation between both CN and elec-
y
tron signals. Indeed, measuring the total CN and
the electron signal dependence on laser fluence, a
linear correlation Žnot shown. between them was
always found for all wavelengths. Nevertheless, the
y
CN signal normalized by the electron yield, i.e. the
y
y
CN re ratio still retains a clear wavelength depen-
dence as displayed in the upper part of Fig. 4. On the
other hand, it is interesting to point out the close
y
y
resemblance between the present CN re spectrum
and the ACN gas-phase absorption spectrum shown
in the lower part of the same figure. Moreover, a red
y1
shift of ca. 20 cm , as indicated, should be re-
marked.
y
Fig. 3. Electron and CN signal wavelength dependence for a
4
. Discussion
y
fixed laser fluence: Ža. CN spectrum obtained at a pACN s3=
y8
1
0
Torr. Žb. Electron signal spectrum obtained at same condi-
The observation of a CNy peak in the TOF
tions as in Ža.. Žc. Same as in Žb. but with no ACN pressure
added: notice the clear wavelenth dependence in both top and
middle panels which, in turn, is absent in the bottom spectrum.
spectrum proves the occurrence of ACN dissociative
attachment induced by laser radiation. Furthermore,

J.O. C a´ ceres et al.rChemical Physics Letters 321 (2000) 349–355
353
y
y
the wavelength dependence, even in the CN re
Obviously, the observation of a much shorter TOF
for electrons Ž0.3 ms. and CN ions Ž10.5 ms.
precludes the gas–surface collision as the limiting
y
ratio, clearly indicates the signature of the ACN
resonant vibrational absorption. The 970 and 953
cm peaks of the ACN gas-phase absorption spec-
y1
step of the charge transfer process. In other words,
y
trum are assigned to the Ž5CHCN. and Ž5CH2 .
the observed electrons and CN must be produced in
wagging modes w16x, respectively. The red shift of
a much faster process, as the ACN surface dissocia-
tive attachment, i.e.
y1
y
y
ca. 20 cm
observed in the CN re spectrum
could be due either to the presence of gas-phase
multiphoton absorption leading to ACN dissociative
attachment or to the weakening of the molecular
bond typically associated to the bond formation be-
tween the adsorbate and the metal surface w18x.
ACN
Ž
ad
.
qCu
Ž
surface
.
CNy
Ž
g
.
qA
Ž
ad or g
.
which can take place via the following two modali-
ties.
1. Substrate-mediated absorption. Essentially the
laser radiation heats the electron in the substrate
pumping them out of the Fermi level and subse-
There are several arguments and observations that
run against the gas-phase dissociative attachment
y
y
mechanism, i.e. schematically: ACN†qe
Aq
quently they tunnel to the ACN potential which
leads to AqCN products. This process is
schematically shown in the top panel of Fig. 5 for
a better illustration. Obviously, this mechanism
would show a thermal character without signifi-
cant frequency selectivity.
y
y
CN . These are the following.
1
. Energetics: This process is very endoergic as
D H 8 is given by D H 8sD 8 ŽA–CN.–EAŽCN.
o
o
o
s5.66 eV ŽRef. w19x. y3.83 eV ŽRef. w15x.s
.83 eV. A simple estimate of the average kinetic
energy of the emitted electrons, E , gives E -10
1
2. Adsorbateymediated absorption. In this case,
the vibrational absorption of the ACN molecule is
predominant. This may have several conse-
quences: first an enhanced local heating of the
surface electrons and therefore an enhanced reali-
sation of the first mechanism with the correspond-
ing increase of the electron yield. Secondly, an
e
e
meV which by energy balance requires a mini-
mum ACN internal energy excitation of 1.82 eV
to overcome the dissociative attachment barrier.
This energy barrier would correspond to ca. 16
y1
photons of 933 cm . By contrast the measured
y
y
CN re signal versus laser fluence Žnot shown.
y
showed zero slope.
increase in the CN yield due to a lowering of
2
. Electron yield dependence on laser frequency:
The gas-phase attachment mechanism does not
explain the electron yield dependence on laser
frequency observed in Fig. 3 Žmiddle panel. which
the D E value Žsee bottom panel of Fig. 5.. Now
the value of D E could be lower than D H 8 since:
o
Ži. the ACN is adsorbed on the Cu surface and
presumably the CH C–CN bond is softer. Žii. the
2
was measured at constant laser fluence. Such a
resonant vibrational absorption increases the
molecular electron affinity and so it facilitates the
electron attachment.
‘
resonant’ enhancement of the electron yield nec-
essarily implies a gas–surface or adsorbate–sub-
strate interaction.
In line with above considerations a crucial ques-
In principle, due to the thermal mechanism ex-
pected upon the microsecond timescale of the laser
y
tion to be considered is whether the ACNŽg.qCuŽs.
excitation, the enhancement of the CN ion and
collisions is the limiting step of the observed elec-
electron yield would be caused by additional heating
of the system though resonant absorption of light by
excitation of the ACN vibration. In this view, the
local vibrational mode would only function as an
antenna for the gathering of heat. If we consider that
while free electron formation requires electrons with
energy above 4.6 eV Ži.e. just to surmount the work
y
tron and CN ions. Under the present pressure con-
y7
ditions Ž ps10
Torr. the number of gas–surface
1
1
y2
y1
collisions is about 2=10 collisions cm
Thus, considering a Cu density of 10 atoms cm
s
.
1
4
y2
every surface Cu atom collides with an ACN
molecule within
y
function barrier. and that CN formation only elec-
1
4
1
0
f5000 s collisiony1
trons with energy above 1.8 eV Ži.e. the work func-
tion barrier minus the CN electron affinity. the
ts 2
=10
.
1
1

3
54
J.O. C a´ ceres et al.rChemical Physics Letters 321 (2000) 349–355
Since the dissociative attachment of the Žad-
sorbed. ACN should also depend on its electron
y
y
affinity, the observed enhancement in the CN re
ratio could be due to the lowering of the dissociative
electron attachment barrier such that a vibrationally
selective chemistry cannot be completely ruled out.
In this picture, the ACN vibrational mode would
function not only as an antenna for increasing the
metal temperature, but it would also introduce some
selectivity by reducing the surface reaction barrier
due to the enhanced adsorbate electron affinity. In
this picture, the resonant excitation remains localized
during the very fast dissociative attachment process
such that subsequent energy transfer from the excited
adsorbate into dissipative channels is not comparable
y
to the rate of formation of the CN moiety.
In view of all the above arguments and the whole
body of data, both possibilities, i.e. substrate- and
adsorbate-mediated absorption may be present as the
laser radiation is focused onto the surface. Obvi-
ously, pump and probe femtosecond experiments
carried out on this system would provide an experi-
mental real-time confirmation of the suggested dy-
namical picture for the acrylonitrile catalysis on cop-
per investigated here.
Fig. 5. Top: idealized one-dimensional potential energy for the
ACNrCu system. Both neutral Cu . . . ACN and negative
5
. Conclusions
y
Cu...ACN potentials are displayed. Here W, EF and EA stand
for work function, Fermi energy level and electron affinity, re-
y
The main conclusions to be drawn from this
investigation are the following.
spectively. Dashed line represents the same Cu... ACN poten-
y
tial, but for an increased A...CN distance and so higher elec-
tron affinity ŽEAACN . at the asymptotic limit. The electron pumps
out of the Fermi level upon laser excitation and subsequent
tunnelling takes place. See text for comments. At the bottom part
of the figure the neutral covalent and ionic A...CN potentials are
qualitatively shown in which the asymptotic energy difference is
1. Important catalytic effects can be achieved by
laser-induced dissociative attachment of ACN on
y
Cu surfaces leading to CN as the main product.
2
. Both substrate and adsorbate mediated mecha-
nisms seem to be controlling the charge transfer
surface reactions. Indeed, a clear enhancement in
0
given by D Es D ŽA–CN.–EAŽCN.. See text for comments.
0
y
both electron and CN signals is obtained by
resonant excitation of the adsorbate vibrational
modes.
y
y
CN re branching ratio will then be proportional to
exp. ŽŽ4.6–1.8. eVrkT.. Here k is the Boltzmann
y
y
3. The wavelength dependence of the CN re
constant and T the temperature. This ratio decreases
with temperature contrary to that experimentally ob-
served. This discrepancy indicates that all observa-
tions cannot be explained assuming thermal equilib-
rium. In other words, other parameters, in addition to
the temperature, should play a role in the surface
chemistry.
branching ratio suggests the possibility of some
vibrational selectivity controlling the yield of this
surface electron attachment chemical reaction.
Finally, the present results provide a new insight
not only into the dynamics of gas–surface interac-
tions but also into new routes for industrial applica-
tions of chemical processes. With regard to the

J.O. C a´ ceres et al.rChemical Physics Letters 321 (2000) 349–355
355
w5x J. Casta n˜ o, V. Zapata, G. Makarov, A. Gonz a´ lez Ure n˜ a, J.
Phys. Chem. 99 Ž1995. 13659.
former it would be interesting to perform this type of
investigation using ultrafast femtosecond laser ex-
cited surface reactions. It could unambiguously dif-
ferentiate the thermal from the vibrationally selective
mechanisms that seem to be present in this surface
reaction.
w6x I. Hussla, H. Seki, T.J. Chuang, Z.W. Gortel, H.J. Kreuzer,
P. Piercy, Phys. Rev. B 32 Ž1985. 3498.
w7x A. Dereux, A. Peremans, J.-P. Vigneron, J. Darville, J.M.
Gilles, J. Electron Spectrosc. Relat. Phenom. 45 Ž1987. 261.
w8x A. Peremans, A. Dereux, F. Maseri, J. Darville, J.M. Gilles,
J.-P. Vigneron, Phys. Rev. B 45 Ž1992. 8598.
w9x M.E. Umstead, M.C. Lin, J. Phys. Chem. 82 Ž1978. 2047.
w10x I. Her, R.J. Finlay, C. Wu, E. Mazur, J. Chem. Phys. 108
Ž1998. 8595.
Acknowledgements
w11x S. Delivala, R.J. Finlay, J.R. Goldman, T.H. Her, W.D.
Mieher, E. Mazur, Chem. Phys. Lett. 242 Ž1995. 617.
w12x M. Bonn, S. Funk, Ch. Hess, D.N. Denzler, C. Stampfl, M.
Scheffler, M. Wolf, G. Ertl, Science 285 Ž1999. 1042.
Financial support from DGES of Spain Grant
PB97-0272 and Ramon Areces Foundation is grate-
fully acknowledged.
w13
x
J.W. Gadzuk, in: J. Manz, L. W o¨ ste Eds. , Femtosecond
Chemistry, Chap. 20, Vol. 2, VCH, Weinheim, 1995.
Ž
.
w14x R.N. Zare, Science 279 Ž1998. 1875.
w15x Handbook of Chemistry and Physics, 74rd edn., 1993–1994.
w16x M. Khlifi, M. Nollet, P. Paillous, P. Bruston, F. Raulin, Y.
B e´ nilan, R.K. Khanna, J. Mol. Spectrosc. 194 Ž1999. 206.
w17x D. Burgess Jr., P.C. Stair, E. Weitz, J. Vac. Sci. Technol. A
4 Ž1986. 1362.
References
w1x T.J. Chuang, Surf. Sci. Rep. 3 Ž1983. 105.
w2x T.J. Chang, Surf. Sci. 178 Ž1986. 763.
w3x W. Zhang, P.G. Smirniotis, J. Catal. 182 Ž1999. 70.
w4x J. Kubota, J. Kondo, K. Domen, C. Hirose, J. Elec. Spec-
trosc. 64r65 Ž1993. 163.
w18x J.K. Norskov, Rep. Prog. Phys. 53 Ž1990. 1253, See for
example.
w19x S.W. North, G.E. Hall, Chem. Phys. Lett. 263 Ž1996. 148.