Adsorbate ordering effects in the trimerization reaction of acetylene on Cu(100)

Article abstract of DOI:10.1021/jp981956n

The trimerization of acetylene to benzene on a Cu(100) surface is reported. The reaction occurs at both room temperature and at or below 200 K depending on the coverage. The reaction yield exhibits a strong coverage dependence, with the yield increasing as the initial acetylene coverage increases. The maximum yield for the room temperature reaction is 30%. The structural properties of the acetylene overlayer have been investigated by LEED, and the results correlated with the reaction yield observations. The results are discussed in light of the coverage dependence of the benzene yield for the same reaction on Cu(110) and Pd(111). It is found that the structural arrangement of the acetylene overlayer on the three different surfaces explains the different coverage dependences of the reaction yield. A unique feature is observed in the thermal desorption data for this system; the mass 26 signal mimics the trimerization reaction rate at high coverages. This is most likely due to the exothermicity of the benzene reaction, leading to the desorption of vibrationally hot benzene molecules, or the transfer of energy to nearby acetylene molecules, resulting in simultaneous desorption.

Full text of DOI:10.1021/jp981956n

J. Phys. Chem. B 1998, 102, 9443-9450
9443
Adsorbate Ordering Effects in the Trimerization Reaction of Acetylene on Cu(100)
Joseph Dvorak*,†
Department of Chemistry, UniVersity of PennsylVania, Philadelphia, PennsylVania 19104
Jan Hrbek
Chemistry Department, BrookhaVen National Laboratory, Upton, New York 11973
ReceiVed: April 22, 1998; In Final Form: August 31, 1998
The trimerization of acetylene to benzene on a Cu(100) surface is reported. The reaction occurs at both
room temperature and at or below 200 K depending on the coverage. The reaction yield exhibits a strong
coverage dependence, with the yield increasing as the initial acetylene coverage increases. The maximum
yield for the room temperature reaction is 30%. The structural properties of the acetylene overlayer have
been investigated by LEED, and the results correlated with the reaction yield observations. The results are
discussed in light of the coverage dependence of the benzene yield for the same reaction on Cu(110) and
Pd(111). It is found that the structural arrangement of the acetylene overlayer on the three different surfaces
explains the different coverage dependences of the reaction yield. A unique feature is observed in the thermal
desorption data for this system; the mass 26 signal mimics the trimerization reaction rate at high coverages.
This is most likely due to the exothermicity of the benzene reaction, leading to the desorption of vibrationally
hot benzene molecules, or the transfer of energy to nearby acetylene molecules, resulting in simultaneous
desorption.
1
. Introduction
being simple enough to study in detail and yet complex enough
to demonstrate interesting chemistry.
A general strategy for surface science is to study simplified
The behavior of acetylene under UHV conditions on single-
crystal substrates is, in general, complex. Acetylene adsorption
has been studied on over 14 low-index faces of transition and
noble metals. The potential for carbon-carbon and carbon-
hydrogen bond cleavage leads to a wide range of possible
products, several of which are typically observed at varying
temperatures/coverages on any given metal (for a review, see
ref 6). The trimerization reaction, however, has proven to be a
highly surface-sensitive reaction. It has been shown conclu-
sively to occur on only two well-defined elemental single-crystal
metal surfaces to date, namely Pd(111) and Cu(110). Recently,
several bimetallic model catalysts have been shown to catalyze
model systems that can be characterized in great detail by the
wide array of available experimental probes. Presumably the
detailed characterization of well-defined systems and the
determination of the underlying physics will lead to the
development of general principles that will aid in the under-
standing of more complex, “real world” systems. The goal of
understanding heterogeneously catalyzed reactions is perhaps
the most mentioned example. In this regard, the choice of model
system is then crucial to developing interesting insights into
more complex phenomena.
One such model reaction that has received a great deal of
attention is the trimerization reaction of acetylene to form
benzene on metal substrates (for a recent review, see ref 1).
This system is interesting for several reasons. The study of
the series of ethane, ethylene, and acetylene on metal single-
crystal surfaces provides insight into the chemistry of single,
double, and triple C-C bonds on surfaces. In addition, this
reaction is exciting in that rather than decomposition to simpler
fragments, the carbon backbone is increased in length by a factor
of 3. It is very rare to see the formation of more complex
molecules on metal surfaces under UHV conditions. Further-
more, on surfaces where this reaction occurs, the intermediate
7
the reaction. Under UHV conditions, the Au/Pd(111) system
8
and the Pd/Au(111) system under appropriate preparation
conditions can show enhanced benzene yield relative to the Pd-
(
111) surface. Sn/Pd-supported bimetallic catalysts can also
9
catalyze the reaction under high-pressure conditions. In
addition, the Sn/Pt(111) system has shown activity toward
benzene formation both under UHV and high-pressure condi-
tions, an interesting result as the Pt(111) surface is highly
selective against benzene formation. The trimerization of
acetylene and related alkynes has also been observed on TiO2
10
1
1
1
2,13
1
,2
under UHV conditions.
to benzene formation has been identified as a C4 species.
The results of over 10 years of investigation of this reaction
on Pd(111) by several research groups has been recently
reviewed by Lambert et. al. On Pd, the reaction is specific to
the (111) surface; it does not occur on the (110) or (100)
surfaces. Under UHV conditions, the reaction proceeds by the
association of two acetylene molecules to form a C4 intermedi-
ate, which then reacts with a third acetylene molecule to form
benzene. The reaction shows a pronounced coverage depen-
Because it is stable, it has been possible to study the intermediate
in great detail by conventional UHV techniques. Also, recent
STM investigations are currently yielding atomic-level detail
of the motions of individual acetylene molecules on metal
1
3-5
surfaces.
Thus, this reaction embodies the characteristics of
†
Current address: Brookhaven National Laboratory, Upton, NY 11973.
E-mail: dvorak@bnl.gov.
1
0.1021/jp981956n CCC: $15.00 © 1998 American Chemical Society
Published on Web 11/03/1998

9
444 J. Phys. Chem. B, Vol. 102, No. 47, 1998
Dvorak and Hrbek
dence, with a threshold coverage necessary for the onset of
benzene formation. Thermal desorption experiments have
shown that benzene is formed only for initial acetylene
coverages above 0.3 ML (monolayer) and reaches a maximum
yield of 25% for a saturated acetylene layer (0.46 ML). It was
also observed that for low-temperature adsorption (175 K),
acetylene forms an ordered overlayer at 0.3 ML, as evidenced
was differentially pumped and equipped with a nose cone with
an aperture of 1 mm. During a TDS experiment, the sample
was positioned within 2 mm of the nose cone. Furthermore, to
eliminate any possible electron-induced reaction, the nose cone
was biased at -45 V to block any electrons from the mass
spectrometer ionizer from reaching the sample. The coverage
was measured by a differential reflectivity technique. A visible
light source (λ ) 632.8 nm) was reflected off of the crystal
surface, and the reflectivity change for p-polarized light was
monitored as a function of acetylene exposure. For adsorption
of acetylene into the chemisorbed state, the reflectivity change
was found to correlate linearly with the coverage.17 This
allowed for an accurate correlation of the exposure to the
coverage (within 0.0l ML). For details of this aspect of the
experiment, see refs. 17 and 18. Absolute coverages are
reported relative to the surface Cu atom density.
The trimerization reaction channel on the Cu(100) surface
was identified by thermal desorption spectroscopy. Benzene
is identified via its parent ion at m/e 78. The acetylene is
identified by its parent ion at m/e 26. These studies were
performed by exposing the clean crystal to acetylene at 120 K
and subsequently performing a TDS run to 420 K. The benzene
and acetylene mass fragments were detected simultaneously.
At the completion of a single run, the crystal was cleaned by
ion bombardment. This procedure was repeated for several
different initial acetylene exposures.
by the observation of a (
x
3 x
x
3)R45° LEED pattern.14
Thus, the onset of the trimerization reaction and the formation
of long-range order in the acetylene overlayer coincide. Another
interesting observation was made. It was possible to force the
trimerization reaction to proceed at lower acetylene coverages
by the adsorption of a spectator molecule, NO. The NO formed
segregated islands, which then forced the acetylene into a higher
local coverage, allowing the trimerization reaction to proceed.
The trimerization of acetylene is also known to occur on the
Cu(110) surface. The reaction was first reported by Avery in
985,15 and has recently been studied in detail by Lomas et al.
2
1
Lomas et al. have shown that the reaction proceeds via an
associative mechanism to form a C4 intermediate, as on the Pd-
(111) surface. However, there is little coverage dependence of
the benzene yield, with nearly complete conversion of 85% of
the adsorbed acetylene to benzene in a single thermal desorption
cycle for initial coverages below 0.4 ML, with the yield falling
slightly at the saturation coverage of 0.5 ML. On Cu(110), a
c(2 × 2) pattern is observed for acetylene adsorption at 120 K
at coverages above 0.15 ML, which increases with quality until
saturation of the monolayer at 0.5 ML. The formation of a
LEED pattern corresponding to the maximum coverage but
observable at low coverage is indicative of attractive interactions
The benzene yield was measured by the following procedure.
The m/e 26 signal was integrated to give the total desorbed
acetylene. Benzene does have a mass fragment at m/e 26, but
based on the measured cracking pattern of benzene and the
amount of benzene detected, this leads in the worst case to an
error of only 2% in the acetylene desorption yield and we,
therefore, ignore it. The m/e 78 signal is integrated to give the
proportional amount of benzene desorbed. This number is then
corrected for the mass spectrometer sensitivity to benzene
relative to acetylene. The relative sensitivity of the mass
spectrometer to the two molecules is estimated from the
1
6
leading to islanding of the adsorbed acetylene molecules.
It appears that the ordering of the acetylene overlayer plays
a key role in determining the trimerization yield. The necessity
of bringing two acetylene molecules close enough to form the
intermediate is a key feature of the reaction. The probability
to form benzene increases the closer the acetylene molecules
are packed together on the surface. For the Pd(111) surface,
the LEED results indicate overall repulsive interactions, which
lead to no benzene formation at low coverages. For Cu(110),
the observed islanding of the acetylene at low coverages explains
both the high yield and the minor coverage dependence of the
yield on this surface.
These interesting results have prompted our investigation of
this reaction on the Cu(100) surface. We report the observation
of the trimerization reaction on the Cu(100) surface for the first
time. The structure of the acetylene overlayer is investigated
via LEED crystallography, and the coverage dependence of the
reaction is studied via thermal desorption spectroscopy. A clear
picture emerges where the coverage dependence of the acetylene
overlayer structure explains the probability of trimerization to
benzene.
1
9,20
ionization cross sections of the molecules,
the measured
cracking pattern, and the transmission and gain factors of the
spectrometer as a function of fragment mass. Using this
procedure, the percent yield of the reaction, defined as the
percentage of initially adsorbed acetylene that reacts to form
benzene, can be calculated. The major source of error in this
procedure is the estimate of the relative sensitivity of the mass
spectrometer to the two different species. Any error introduced
by the estimation of the relative sensitivities will be the same
for all runs, so that any observed trends will be real.
3
. Results
Molecular Acetylene: Thermal Desorption Results. The
adsorption and thermal evolution of acetylene on the Cu(100)
surface has been investigated by HREELS by Marinova et al.21
In this study it was observed that acetylene adsorbs molecularly
at low temperatures (140 K) but is strongly distorted with a
C-H stretch frequency indicating rehybridization to close to
2. Experimental Section
The experiments were performed in a stainless steel UHV
3
chamber equipped with a four-grid LEED apparatus, a quad-
rupole mass spectrometer, and an Auger spectrometer. The
sample, a commercially prepared Cu(100) single crystal, was
attached to a cryostat that allowed cooling to 90 K via liquid
nitrogen and resistive heating. The sample temperature was
measured by a type C thermocouple. Following initial crystal
cleaning, the sample was routinely cleaned by 10 min of Ar+
sputtering followed by annealing to 800 K for 5 min. The
experimental apparatus allowed for high-quality thermal de-
sorption spectra (TDS) to be obtained. The mass spectrometer
sp and a C-C stretch frequency indicative of a bond order of
1.5. The acetylene was stable on the surface to about 300 K,
at which temperature a new hydrocarbon species was detected.
This species was thought to be CCH, but CCH2 could not be
ruled out on the basis of the observed frequencies. Further
heating to 375 K resulted in simultaneous desorption of
molecular acetylene and decomposition to the new hydrocarbon
species.
We have studied this system utilizing thermal desorption
spectroscopy. Selected TDS spectra for increasing coverages

Trimerization Reaction of Acetylene
J. Phys. Chem. B, Vol. 102, No. 47, 1998 9445
2
a). Increasing the coverage to near 0.18 ML resulted in
1
1
1
increased intensity of the ( /2, /2) and (0, /2) spots but still a
generally diffuse pattern (Figure 2b). Cooling the sample down
to 90 K at 0.18 ML coverage resulted in a somewhat sharper
1
1
LEED pattern with spots in the ( /4, /2) positions becoming
discernible. Subsequent exposure increased the sharpness of
the observed pattern, and the pattern developed continuously
to that displayed in Figure 2c. The LEED pattern shown was
obtained at the saturation coverage of 0.25 ML and was the
sharpest one observed. This LEED pattern could also be
observed by dosing at 90 K and annealing the crystal to 275 K.
Heating the crystal to 400 K resulted in the return of a sharp
ap c( 1e t×yle 1n )e p. attern, indicative of desorption of the majority of the
Trimerization Reaction: Thermal Desorption Results.
2
1
The previously cited HREELS study did consider the pos-
sibility of acetylene trimerization, however, the authors were
unable to positively identify this reaction channel for two
reasons. First, for preadsorbed benzene, the observed vibrational
modes coincided with some of the modes of the acetylene
decomposition product, and second, preadsorbed benzene was
found to desorb below room temperature.
Figure 1. Acetylene thermal desorption spectra for increasing exposure.
Heating rate is 1.1 K/s. Acetylene is monitored via m/e 26. Successive
traces are offset vertically for clarity.
are shown in Figure 1. The spectra indicate that acetylene
adsorbs into three main energetically different states as a
function of coverage. At low exposure, the acetylene desorbs
from a chemisorbed state at 355 K. At low exposure, this peak
is well-described by first-order kinetics. The desorption energy
We have identified the presence of this reaction channel and
characterized it via TDS measurements. Figure 3a shows the
TDS spectra for simultaneously acquired m/e 26 (acetylene) and
m/e 78 (benzene) for varying initial acetylene exposures. An
interesting feature present in Figure 3a is the simultaneous
desorption of benzene and a correlated peak in the m/e 26
spectra. This correlation is quite clearly demonstrated for the
higher exposure runs. It cannot be accounted for by the ground-
state cracking pattern of benzene. The 26/78 cracking ratio for
gas-phase benzene is 0.05; this can only explain less than 5%
of the “extra” m/e 26 signal. There are two possible reasons
for this correlated feature in the m/e 26 signal. First, there could
be some factor that leads to both an increased probability of
benzene formation and simultaneous acetylene desorption. The
second possibility is that the reactively formed benzene desorbs
with a large amount of internal energy, which alters the cracking
pattern significantly. We can estimate from the TDS data that
the 26/78 cracking ratio must change by a factor of about 20 to
account for the observed m/e 26 signal.
is found to be 109.0(4) kJ/mol, and the desorption prefactor is
estimated to be 101
5.7(1.4)
. Increasing the exposure leads to
growth of this feature, but the peak desorption temperature
decreases, consistent with short-range repulsive interactions
among nearby molecules. Near saturation of the chemisorbed
feature there is an interesting development: a sharp spike grows
in at about 320 K, which is partially resolvable from the main
chemisorbed peak and correlates with the benzene desorption.
There are several possible explanations for this feature that will
be discussed later. At higher exposures, the chemisorbed state
saturates and a new state is occupied, the “weakly chemisorbed”
state, which desorbs at 240 K. The third major feature,
consisting of acetylene bound in multilayers, is found at 90 K.
It is known that one reaction channel available to acetylene
on the Cu(100) surface is a decomposition/isomerization leading
to a strongly bound surface species. We have measured the
amount of acetylene that decomposes on the surface rather than
desorbing. For repeated saturation exposures of acetylene and
subsequent thermal desorption without cleaning the crystal, the
amount of desorbed acetylene decreases for each subsequent
run. We estimate that about 5% of the acetylene overlayer
undergoes decomposition from these results.
To determine whether the desorption of reactively formed
benzene was desorption limited or reaction rate limited, TDS
experiments were carried out for benzene adsorbed directly onto
the clean Cu surface. It is found that pure benzene desorbs at
or below 246 K, depending on the coverage. Therefore, the
benzene evolution observed under reactive conditions is reaction
rate limited, and the m/e 78 curve directly reflects the kinetics
of the trimerization reaction.
Ordering of the Acetylene Overlayer: LEED Results. The
development of the LEED pattern is shown in Figure 2. The
acetylene coverages quoted are based on a calibration of the
relative coverage versus exposure using an optical reflection
technique and Auger spectroscopy and taking the saturation
coverage to be 0.25 ML on the basis of our LEED results. No
The reaction yield as a function of initial coverage is shown
in Figure 3b. The reaction yield was calculated as described
in the Experimental Section. From the figure it is clear that
there is a strong coverage dependence of the reaction yield on
the initial coverage, with the yield reaching a maximum value
of approximately 30% for the saturated acetylene overlayer. At
saturation coverage, we estimate that about 5% acetylene
undergoes decomposition to an irreversibly bound hydrocarbon
species. Thus, the benzene channel is the major reaction on
this surface.
pattern is observed for adsorption at 90 K, in accordance with
previously reported results for adsorption at 60 K.22 However,
dosing at 275 K resulted in the LEED patterns shown.
Adsorption at this temperature restricts acetylene to occupying
the chemisorbed state. This temperature was chosen specifically
to prevent adsorption into the weakly chemisorbed state and
yet to remain below the temperature where the decomposition/
isomerization channel opens. Thus, all ordered overlayers
observed here relate to acetylene situated in the chemisorption
sites. At coverages near 0.12 ML, an intensity pattern begins
to emerge with nearly radial symmetry that approximates a
doughnut with a discernible inner radius corresponding to one-
half the reciprocal lattice vector of the clean substrate (Figure
Low-Temperature Acetylene Trimerization. When the
crystal is held at temperatures below about 240 K, it becomes
possible to adsorb additional acetylene on the surface, as
evidenced by the “weakly chemisorbed” feature in the TDS
spectrum at 240 K. It is found that the total amount of additional

9446 J. Phys. Chem. B, Vol. 102, No. 47, 1998
Dvorak and Hrbek
(c)
(d)
Figure 2. Coverage development of ordering of acetylene on Cu(100) (a) 0.12 ML, (b) 0.18 ML, (c) 0.25 ML (saturated chemisorbed overlayer).
(d) Simulated radial dependence of S(k) for the hard-disk model. Incident beam energy ) 52.0 eV, Tads ) 275 K.
reflect the reaction kinetics. The benzene yield in the low-
temperature/high-coverage regime is also substantial. We find
that 36% of the acetylene adsorbed above 0.25 ML is converted
to benzene. The fact that the reactively formed benzene desorbs
at lower temperatures than from a clean surface is attributable
to steric effects which lower the effective desorption barrier.
One possible explanation is that with large amounts of acetylene
also present on the surface, the benzene is unable to lie
completely flat and maximize its adsorption energy. This effect
1
is seen also for benzene on Pd(111).
4
. Discussion
.1. Acetylene Overlayer Structure: Coverage Depen-
4
dence. Interpretation of Intermediate Coverage Diffuse
LEED Results. The diffuse LEED pattern observed at inter-
mediate coverages (Figure 2a) can be understood in a semi-
quantitative way by a simple hard-disk model. Consider an
isolated acetylene molecule on the surface with its ability to
scatter electrons contained in the form factor F(k), where ideally
this is calculated in a rigorous, multiple scattering formalism.
If the lateral potential of the surface is turned off, then F(k)
averaged over all orientations becomes a radially symmetric
function. In addition, the probability of finding a nearby
acetylene molecule is given by the pair correlation function g(r),
which in the absence of a lateral potential becomes a radially
symmetric function as well. In a simple model, the interactions
of acetylene with its neighbors is approximated as a hard disk
of diameter a. Under these approximations, the scattered
Figure 3. (a) Simultaneously acquired acetylene and benzene thermal
desorption spectra. Solid line: acetylene (m/e 26). Dotted line: benzene
(
m/e 78). Tads ) 120 K. Heating rate is 1.1 K/s. (b) Percent reaction
yield of benzene as a function of initial acetylene coverage for benzene
originating from the chemisorbed acetylene overlayer.
acetylene adsorbed is nearly one-fifth of the acetylene in the
0.25 ML structure. Some of the acetylene adsorbed in this
2
3
intensity can be written as
coverage regime is also capable of trimerizing to benzene. Since
benzene desorbs from the clean surface at or below 250 K, the
benzene evolution in this coverage regime does not necessarily
2
I(k) ) N|F(k)| S(k)

Trimerization Reaction of Acetylene
J. Phys. Chem. B, Vol. 102, No. 47, 1998 9447
where
∞
S(k) ) 1 + 2πF 0
∫
r(g(r) - 1)J (kr)dr
0
In eq 1 N is the total number of scatterers, S(k) is the structure
factor, F is the number of scatterers per unit area, and J0 is the
zero-order Bessel function. It is then necessary to deduce the
functional form of the pair correlation function. In the limit of
-
V(r)/kT
low density, g(r) becomes simply e
, where V(r) is the
intermolecular pair potential.24 For a hard disk, V(r) is ∞ for r
e a and 0 for r > a. This leads to g(r) being 0 for r e a and
Figure 4. Four possible saturated chemisorbed acetylene overlayer
structures. Structures a and b correspond to acetylene in the local C
binding site, and structures c and d correspond to acetylene in the local
2V binding site as described in the text.
s
1
for all other r. The physical interpretation of this is that the
acetylene molecules are situated randomly on the surface except
that they can never approach each other closer than a. This
leads to the structure factor being given by
C
occupy any randomly chosen position on the surface. However,
it is clear that for an isolated acetylene molecule, F(k) will not
be radially symmetric and each molecule will be strongly
localized to a specific bonding site. Considering this, it seems
surprising that a diffraction pattern with approximate radial
symmetry is observed at these intermediate coverages.
a
∫
S(k) ) 1 + 2πF 0
(-1)rJ (kr)dr
0
The intensity pattern calculated from eq 2 is then shown in
Figure 2d for a 0.12 ML coverage and a hard-disk diameter of
3.3 Å. It is interesting to note that in the above formula all the
As will be discussed in the next section, there is evidence
that the local binding site of an acetylene molecule is either the
low-symmetry site shown in Figure 4a,b or the higher symmetry
site shown in Figure 4c,d. At low coverages, an acetylene
molecule can orient itself in more than one way in any given
substrate primitive cell. This leads to some averaging of the
structure factor so that it can approach radial symmetry. Also,
as long as a second acetylene molecule is further apart than the
hard-disk diameter, there are more possible relative orientations
than the square lattice presented by the substrate would suggest.
These considerations clarify how such a simple model is able
to capture the qualitative aspects of the diffraction pattern at
relatively high densities and also points out the inadequacies.
The lack of considering any polar angle dependence of the form
factor F(k) is another simplification which limits the accuracy
of the calculated intensity.
structural information is contained in the integral. In the low-
coverage limit, a change in the coverage simply scales the
diffraction features while their positions in reciprocal space are
unaltered. A change in the hard-sphere diameter does lead to
a shifting of the diffraction features in reciprocal space, however.
We have numerically integrated eq 2 for various values of a
from 1 to 4 times the Cu(100) nn distance of 2.55 Å. We then
calculate the position of the first diffraction maximum as a
function of a. We find that to negligible error over the stated
range, the position of the first-order maximum k0 as a function
of the hard disk diameter is
2
a
π
k0 ) 0.817
Thus, the position of the first-order maximum is located at a
distance slightly less than the reciprocal lattice vector associated
with the distance a. The position of the first-order maximum
Long-Range Order Near Saturation Coverage. The satu-
ration pattern shown in Figure 2c leads to a simple and
unambiguous interpretation. The diffraction positions observed
in a LEED pattern can be simply interpreted to yield the
symmetry and dimensions of the overlayer unit cell. Note that
while the relative intensities of the observed spots vary as a
function of the incident electron energy, no additional spots are
observed over the energy range of 40-180 eV. The Cu(100)
substrate provides a square lattice. All the observed spots can
be accounted for by a p(2 × 4) overlayer lattice, allowing for
the presence of two domains oriented at right angles to each
other. Then it is necessary to account for the missing spots
that are predicted by such an overlayer but are not observed.
The answer lies in the presence of a glide plane lying along the
long real space axis. If the real space p(2 × 4) unit cell is
oriented with the long axis lying along the [01] direction, such
a glide plane results in every spot with indices (0,k), with k
-1
from Figure 2b is measured to be 1.55(16) Å , which according
to eq 3 yields a hard-disk diameter of 3.31(35) Å. Our
interpretation of the diffuse LEED pattern at low to intermediate
coverage is that the dominant forces ordering the structure of
acetylene are short-range repulsive interactions that restrict the
approach of the acetylene molecules to no closer than the hard-
disk diameter. For farther distances, the site occupation is
approximately random, leading to the radially symmetric pattern.
It will be seen later that the saturation structure leads to acetylene
molecules at 5.10 Å apart. Thus, while this simple hard-sphere
model can qualitatively account for the observed diffraction
pattern, the model underestimates the hard-disk diameter by
3
5%.
The underestimation of the hard-disk diameter is due to the
simplifications made to develop the model, and it is instructive
to consider these in more detail. One might anticipate that the
assumption of low coverage would lead to large inaccuracies
in the calculation of k0, but this is not the case. It has been
shown, e.g., in extensive calculations on liquids, that the position
of the first order maximum is dependent essentially upon the
repulsive part of the intermolecular potential and is very weakly
dependent upon the density,25 varying only 7% up to a density
of one-half of the close-packed density. We note that the
coverage in Figure 2b is about one-half of the saturated
coverage. In developing the model, it was assumed that the
form factor F(k) was radially symmetric and the acetylene could
2
6
odd to be missing, with the indices referring to the overlayer
reciprocal lattice points. Then the sum of such a pattern with
an equivalent domain at right angles accounts simply for the
observed pattern. The appearance of a glide plane in the space
symmetry is fortunate in that it allows the absolute coverage to
be determined with little doubt. The glide plane necessarily
restricts the number of acetylene molecules in the overlayer unit
cell to be an even number. This restriction requires that there
are at least two acetylene molecules in the unit cell.
To further narrow down the possible surface structure, we
consider two possibilities for the local binding site geometry

9448 J. Phys. Chem. B, Vol. 102, No. 47, 1998
Dvorak and Hrbek
TABLE 1: Intermolecular Distancesa
structure a structure b structure c structure d
for the acetylene molecule. Utilizing SEXAFS and NEXAFS,
Arvanitis et al.2
to three Cu surface atoms in a di-σ + π bonding arrangement
nomenclature from ref 6). The point group of acetylene in
2,27,28
have found the acetylene to be coordinated
carbon
hydrogen nn
acetylene nn
nn
4.22
3.44
5.10(2)
5.33(2)
6.22(2)
8.37(2)
4.22
3.72
5.10(2)
5.20(2)
6.54(2)
7.96(2)
4.22
3.94
5.10(2)
5.70(4)
9.20(4)
10.20(4)
4.10
3.00
5.10(4)
7.20(4)
10.20(4)
11.40(8)
(
this binding site is Cs. The C-C bond lies parallel to the surface
at 1.30 Å above the surface Cu plane. The bond length is 1.42
Å, indicating significant stretching of the gas-phase C-C bond
nnn
nn
4nn
3
29
length of 1.21 Å. The positions of the hydrogen atoms cannot
be determined by the above experimental techniques, however,
they can be estimated. The observed hydrogen stretch indicates
a
Inter-acetylene distances refer to the distance between molecules
measured from the center of the C-C Bond. Distances reported in Å.
3
21
a high degree of rehybridization, close to sp , which implies
significant bending of the hydrogen away from the surface. A
recent theoretical study optimizing the hydrogen position for
the experimentally determined binding site places the molecular
plane of acetylene at 60° to the surface with a C-C-H bond
dependence of the LEED pattern and the TDS data for repulsive
interactions at close distances for acetylene molecules. How-
ever, taking the van der Waals radius of a hydrogen atom at
3
2
1
.0 Å and noting that the closest approach of H atoms in both
angle of 120°, where a C-H bond length of 1.09 Å was
assumed.30 We then use these dimensions to fix the hydrogen
structures is significantly greater than 2.0 Å, it is not possible
to distinguish between the two structures based on steric
arguments.
positions.
A recent STM study has found a different local binding site
for acetylene. Stipe et al.4,5 observed the acetylene to be
centered above the 4-fold hollow site with the molecular plane
aligned across the diagonal of the square formed by four Cu
atoms. In this configuration, the acetylene is expected to be
coordinated to four Cu atoms. The point group symmetry of
acetylene in this site is C2V.
If the local binding site is the site with C2V symmetry, then
structures c and d are possible. Of these two sites, structure c
appears to be the most probable. This structure has farther nn
distances between both carbon and hydrogen than structure d.
It also has fewer nn molecules. Therefore, it is more consistent
with the observation of repulsive intermolecular interactions.
4
,5
Also, the STM data provides convincing evidence of a C2V
site symmetry. However, none of the structures can be
rigorously excluded on the basis of the LEED data.
Given the two proposed local binding sites of acetylene in
its chemisorbed state, it is highly unlikely that there are four
acetylene molecules per p(2 × 4) unit cell. Any such config-
uration would require some or all of the Cu atoms in the unit
cell to be coordinated to more that one acetylene molecule at
saturation coverage, which implies a significant change in the
bonding as a function of coverage. Our LEED results argue
strongly against such a possibility. There is no evidence in the
LEED pattern for a change in periodicity of the acetylene unit
cell or a change in symmetry of the p(2 × 4) unit cell for
coverages between 0.18 and 0.25 ML, corresponding to a 40%
increase of the coverage. It seems highly unlikely that a shift
from two to four molecules per unit cell is occurring and that
we have misassigned the absolute coverage. We, therefore, set
the number of acetylene molecules in the unit cell at 2, yielding
a saturation coverage of 0.25 ML. This result compares well
with the results of Marinova et al.31 Measuring the Auger
intensity for a saturated acetylene monolayer and comparing to
the Auger intensity for a saturated monolayer of oxygen for
which the coverage is known, they estimate the acetylene
saturation coverage at 0.24 ML.
4.2. The Trimerization Reaction and Its Coverage De-
pendence. In this section we will concentrate on the acetylene
trimerization in the chemisorbed layer. The correlation of the
LEED results with the reaction yield results provide a satisfying
picture that compares well with the yield vs coverage results
on Pd(111) and Cu(110). For the Cu(100) system, there is no
threshold for reaction. Short-range repulsive interactions lead
to a low probability of reaction at low coverages. The
probability is not zero, however, because the repulsion is only
short range. The partial ordering observed at lower coverage
in LEED indicates that it is possible (on average) for some
acetylene molecules to approach sites as close as about two
lattice units. Heating the surface gives some of these pairs
enough thermal energy to then approach nn sites and react.
Increasing the surface coverage increases these nearby “pairs”,
which can then go on to react.
The LEED results also shed light on the interesting correlation
between the benzene reaction and the simultaneous peak
observed in the m/e 26 TDS data. To discuss this, first we will
assume that the signal is due to simultaneously desorbing
acetylene. Under this assumption, we will consider two possible
scenarios. One possible explanation is that at the higher
coverages a new overlayer structure, which we will term a
“compression structure”, is formed, forcing the closer approach
of adsorbate molecules. It is well-known that nn repulsive
interaction can lead to two peaks in a TDS spectrum for
coverages above 0.5 ML on a square lattice, even though the
molecules remain bound in the same local site over the entire
The two proposed binding sites lead to four possible structures
for the saturated acetylene overlayer based on the symmetry
requirements imposed by the LEED data. These are shown in
Figure 4. Structures a and b belong to the space group pg,
structure c belongs to space group p2gg, and structure d belongs
to space group p2mg.
To further distinguish between these four structures, we have
calculated several intermolecular distances. These distances are
displayed in Table 1. If the local binding site is the site with
Cs symmetry, then structures a and b are possible. Both
structures are very similar in terms of intermolecular distances.
Both structures have identical nn C distances and nn “center of
molecule” distances. The hydrogens at closest approach are
3
3
coverage range. In our case, with a saturation coverage of
0.25 ML, addition of more molecules would force occupation
of nn or nnn lattice cells. At high coverages, the formation of
a “compression structure” would lead to a lower activation
barrier to desorption due to short-range repulsions and to a high
benzene reaction rate due to the closer approach of acetylene
pairs. However, our LEED results indicate that this is not the
case. The LEED results show that nn and nnn lattice cells are
8
% farther apart for structure b. For higher order distances
measured from the center of the C-C bond, it can be seen from
Table 1 that there is a compensation in the bond distances. That
is, the nnn distance is closer for structure b but the 3nn distance
is closer for structure a, etc. There is evidence in the coverage

Trimerization Reaction of Acetylene
J. Phys. Chem. B, Vol. 102, No. 47, 1998 9449
not occupied to any significant extent in the overlayer and the
saturation coverage is insufficient to force occupation of these
sites.
N2O from Si as well.37 Either one of these two possibilities
would explain the correlation between the m/e 78 and 26 signals
without the need to invoke an additional binding site. It should
be possible to gain further insight into these possibilities via
angle-resolved TDS measurements.
The second possibility is as follows. It is conceivable that
once the well-ordered chemisorbed acetylene layer is formed,
some acetylene adsorbs randomly into some weakly bound state
but does not alter the structure of the chemisorbed monolayer
and is not directly evidenced in the LEED data. Upon heating
this, acetylene could react with acetylene in the chemisorbed
layer to form benzene or desorb. Then, presumably the 320 K
peak could be due to this “weaker” bound state. From the TDS
measurements, the amount of acetylene that must exist in such
a state can be measured. It turns out that at saturation of the
chemisorbed layer, one-fifth of the total desorbed acetylene
comes at 320 K. If we attribute adsorption of acetylene into
this hypothetical state as being responsible for the increased
benzene yield as well, then at least one acetylene molecule for
each desorbed benzene must come from this state. This leads
to about 30% of the total adsorbed acetylene being in this state
prior to the TDS cycle. The bonding energy of acetylene in
this state can then be estimated at 85 kJ/mol by the common
5
. Conclusion
We conclude this discussion by summarizing the principle
findings of these experiments and the insight gained into this
reaction. The trimerization reaction of acetylene to benzene
has been observed on the Cu(100) surface, this being the third
elemental metal single-crystal face and the first (100) face
known to catalyze this reaction under UHV conditions.
Molecularly adsorbed acetylene can adsorb in three energeti-
cally different states, the chemisorbed state (340 K), the weakly
chemisorbed state (240 K), and the multilayer state (90 K). For
adsorption into the chemisorbed state, the acetylene forms a
well-ordered overlayer at coverages approaching 0.25 ML. The
structure of this overlayer has been determined by LEED. At
lower acetylene coverages, we have proposed a structural model
for the overlayer that is consistent with the diffuse LEED
patterns observed. At lower coverages, the overlayer shows
short-range order that is dominated by short-range repulsive
forces. The analysis of the diffuse LEED pattern suggests that
this distance is 3.3 Å, however, it may be closer to 5.1 Å as the
well-ordered overlayer results indicate. Beyond this distance
the distribution is at least approximately random. At coverages
approaching 0.25 ML, these repulsive interactions cause the
overlayer to adopt a well-ordered arrangement on the surface.
At saturation coverage, the repulsive interactions force the
overlayer to adopt a structure with long-range order; the LEED
data allows this structure to be identified.
The trimerization reaction yield shows a strong coverage
dependence but does not show a threshold as in the case of
Pd(111). We find that the coverage dependence of the reaction
yield can be correlated well with the LEED observations; the
repulsive interactions among neighboring acetylene molecules
leads to little probability of trimerization at low coverage, while
at higher coverages, the acetylene molecules are forced together
in a closer arrangement and the reaction yield increases. The
LEED observations also shed light on an interesting feature
observed in the TDS data for this system. At higher coverages
where the benzene yield is high, a feature is observed in the
m/e 26 signal with kinetics that mimic the trimerization reaction
rate. The most likely explanation for this phenomenon is that
the exothermicity of the trimerization reaction leads to direct
desorption of vibrationally excited benzene, with the possibility
that some of the energy is transferred from the benzene to nearby
acetylene monomers, resulting in their desorption.
34
13
Redhead analysis (prefactor ) 10 assumed). This bonding
strength is significant, being 70% of the bonding energy of the
acetylene desorbing at 350 K, and if this state indeed exists, it
indicates significant interaction with the substrate. Thus, given
the amount of acetylene that must be present in this hypothetical
state to explain the observations, one would expect to see an
altering of the overlayer structure at high coverages. Again,
the LEED data shows no evidence for this.
The lack of evidence for a structural explanation leads us to
consider an alternative mechanism. As already mentioned, the
correlation between the two features cannot be explained by
the cracking ratio of ground-state benzene. The correlation is
most likely due to the exothermicity of the trimerization reaction.
The energetics involved in the reaction can be estimated. The
∆
-
H of reaction from acetylene to benzene in the gas phase is
602 kJ/mol. The ∆H of acetylene adsorption is -109 kJ/
3
5
mol from our TDS results, and the ∆H of benzene adsorption
can be estimated to be approximately -65 kJ/mol using the
standard Redhead analysis of our TDS data. This means that
the formation of an adsorbed benzene molecule from adsorbed
acetylene releases 340 kJ/mol of energy. Desorption of the
benzene molecule requires 65 kJ/mol, leaving 275 kJ/mol of
energy to be dissipated. Therefore, it is reasonable that some
of this energy ends up as internal energy within the benzene
molecule. The correlation could then be due to vibrationally
hot benzene desorbing before it can equilibrate with the surface.
If the benzene desorbs with a large amount of internal energy,
this will alter the cracking ratio. It is known that reactively
formed surface species can desorb without equilibrating with
the surface; such an effect is seen in the angular dependence of
Acknowledgment. The authors thank Prof. Hai-Lung Dai
for critical reading of the manuscript. This research was carried
out at Brookhaven National Laboratory under Contract No. DE-
AC02-76CH00016 with the U.S. Department of Energy, Divi-
sion of Chemical Sciences, Office of Basic Energy Sciences.
J.D. acknowledges support from the National Science Founda-
tion MRSEC program under Grant No. DMR96-32598, as well
as the receipt of a Teagle Foundation Scholarship administered
by Exxon Corp.
36
reactively formed N2 desorbing from the NO/Pd(110) system.
However, if hot benzene were the only mechanism respon-
sible for the “extra” m/e 26 signal, then then the “extra” signal
should be proportional to the m/e 78 signal. Then it should be
possible to subtract out the correlated peak in the m/e 26
spectrum and get a smooth curve indicative of pure acetylene
desorption. However, we find that it is not possible to remove
the 320 K feature completely from the m/e 26 spectrum by
subtraction for any proportionality constant. We, therefore,
cannot rule out the possibility that some of the energy from the
reaction is transferred to nearby acetylene molecules, causing
them to desorb simultaneously. A reaction-assisted desorption
mechanism has recently been suggested for the desorption of
References and Notes
(1) Lambert, R. M.; Omerod, R. M. In Surface Reactions; Springer
Series in Surface Science; Madix, R. J., Ed.; Springer-Verlag: Berlin, 1994;
Vol. 34, Chapter 4.

9450 J. Phys. Chem. B, Vol. 102, No. 47, 1998
Dvorak and Hrbek
(
2) Lomas, J. R.; Baddeley, C. J.; Tikhov, M. S.; Lambert, R. M.
Langmuir 1995, 11, 3048.
3) Dunphy, J. C.; Rose, M.; Behler, S.; Ogletree, D. F.; Salneron,
(20) Bobeldijk, M.; van der Zande, W. J.; Kistemaker, P. G. Chem. Phys.
1994, 179, 125.
(21) Marinova, T. S.; Stefanov, P. K. Surf. Sci. 1987, 191, 66.
(
M.; Sautet, P. Phys. ReV. B 1998, 57, 12705.
(
22) Arvanitis, D.; D o¨ bler, U.; Wenzel, L.; Baberschke, K.; St o¨ hr, J.
Surf. Sci. 1986, 178, 686.
23) Hukins, D. W. X-ray Diffraction By Disordered and Ordered
Systems; Pergamon Press: Oxford, 1981; Chapter 9.
24) McQuarrie, D. A. Statistical Mechanics; Harper Collins, New York,
976.
(
(
4) Stipe, B. C.; Rezaei, M. A.; Ho, W. Science 1998, 280, 1732.
5) Stipe, B. C.; Rezaei, M. A.; Ho, W. Phys. ReV. Lett. 1998, 81,
(
1
263.
(
6) Sheppard, N. Annu. ReV. Phys. Chem. 1988, 39, 589.
(
(7) Baddeley, C. J.; Tikhov, M.; Hardacre, C.; Lomas, J. R.; Lambert,
1
R. M. J. Phys. Chem. 1996, 100, 2189.
8) Baddeley, C. J.; Ormerod, R. M.; Stephenson, A. W.; Lambert, R.
M. J. Phys. Chem. 1995, 99, 5146.
9) Lee, A. F.; Baddeley, C. J.; Hardacre, C.; Mooggridge, G. D.;
(
(
25) Verlet, L. Phys. ReV. 1968, 165, 201.
26) Clarke, L. J. Surface Crystallography; John Wiley and Sons:
(
Chichester, 1985; p 68.
(27) Arvanitis, D.; Baberschke, K.; Wenzel, L.; D o¨ bler, U. Phys. ReV.
Lett. 1986, 57, 3175.
(
Ormerod, R. M.; Lambert, R. M.; Candy, J. P.; Basset, J.-M. J. Phys. Chem.
B 1997, 101, 2797.
(
(
(
(
10) Xu, C.; Peck, J. W.; Koel, B. J. Am. Chem. Soc. 1993, 115, 751.
11) Szanyi, J.; Paffett, M. T. J. Am. Chem. Soc. 1995, 117, 1034.
12) Pierce, K. G.; Barteau, M. A. J. Phys. Chem. 1994, 98, 3882.
13) Lusvardi, V. S.; Pierce, K. G.; Barteau, M. A. J. Vac. Sci. Technol.
(28) Arvanitis, D.; Wenzel, L.; Baberschke, K. Phys. ReV. Lett. 1987,
59, 2435.
(29) Lafferty, W. J.; Thibault, R. J. J. Mol. Spectrosc. 1964, 14, 79.
(30) Wong, Y.-T.; Hoffmann, R. J. Chem. Soc., Faraday Trans. 1990,
86, 553.
A 1997, 15, 1586.
14) Tysoe, W. T.; Nyberg, G. L.; Lambert, R. M. Surf. Sci. 1983, 135,
28.
(
(31) Marinova, T. S.; Stefanov, P. Surf. Sci. 1989, 219, 490.
1
(32) Handbook of Biochemistry, 2nd ed.; Sober, H. A., Ed.; The
(
(
15) Avery, N. R. J. Am. Chem. Soc. 1985, 107, 6711.
16) Woodruff, D.; Wang, G. C.; Lu, T. M. In The Chemical Physics of
Chemical Rubber Company: Cleveland, OH, 1970.
33) Masel, R. I. Principles of Adsorption and Reaction on Solid
Surfaces; John Wiley and Sons: New York, 1996; Chapter 7.
(
Solid Surfaces and Heterogeneous Catalysis; King, D. A., Woodruff, D.
P., Eds.; Elsevier: New York, 1983; Vol. 2, Chapter 4, pp 327-331.
(34) Redhead, P. A. Vacuum 1962, 12, 203.
(17) Dvorak, J.; Dai, H.-L. SPIE Proc. 3272 (Int. Soc. Opt. Eng.,
(35) CRC Handbook of Chemistry and Physics, 76th ed.; Lide, D. R.,
Bellington, WA, 1998).
Ed.; CRC Press: New York, 1995.
(36) Ikai, M.; Tanaka, K. Surf. Sci. 1996, 781, 357-358.
(37) Kato, H.; Sawabe, K.; Matsumoto, Y. Surf. Sci. 1996, 351, 43.
(
(
18) Dvorak, J.; Borguet, E.; Dai, H.-L. Surf. Sci. 1996, 369, L122.
19) Zheng, S.-H.; Srivastava, S. K. J. Phys. B: At. Mol. Opt. Phys.
1
996, 29, 3235.