Adsorption and reaction of trans-1,2-dichloroethene on copper nanoclusters supported on thin Al2O3 films

Article abstract of DOI:10.1016/j.jcat.2004.04.036

The adsorption and reaction of trans-1,2-dichloroethene on Cu nanoclusters supported on a thin Al₂O₃ film grown on NiAl(110) have been studied as a function of temperature and Cu cluster size using molecular beams and temperature-programmed desorption. At 300 K, dechlorination is observed on clusters of all sizes; subsequent desorption yields benzene and acetylene as the main reaction products. The ratio of these is strongly dependent on cluster size; for small clusters only acetylene evolution is observed. With increasing cluster size the probability of trimerisation of acetylene to produce benzene increases, although the desorption of acetylene remains dominant. The adsorption and reaction are compared to previous work on a Cu(110) single crystal where benzene was the main desorption product after adsorption at 300 K. The cluster size-dependent yield of desorption products is attributed to the inability of small clusters to accommodate the dissociation products prior to the trimerisation reaction.

Full text of DOI:10.1016/j.jcat.2004.04.036

Journal of Catalysis 226 (2004) 1–8
www.elsevier.com/locate/jcat
Adsorption and reaction of trans-1,2-dichloroethene
on copper nanoclusters supported on thin Al2O3 films
S. Haq, C. Winkler, A. Carew, J. Ledieu, R. Raval ∗
The Surface Science Research Centre, University of Liverpool, Liverpool L69 3BX, UK
Received 10 August 2003; revised 29 April 2004; accepted 29 April 2004
Available online 5 June 2004
Abstract
The adsorption and reaction of trans-1,2-dichloroethene on Cu nanoclusters supported on a thin Al O film grown on NiAl(110) have
2
3
been studied as a function of temperature and Cu cluster size using molecular beams and temperature-programmed desorption. At 300 K,
dechlorination is observed on clusters of all sizes; subsequent desorption yields benzene and acetylene as the main reaction products. The
ratio of these is strongly dependent on cluster size; for small clusters only acetylene evolution is observed. With increasing cluster size the
probability of trimerisation of acetylene to produce benzene increases, although the desorption of acetylene remains dominant. The adsorption
and reaction are compared to previous work on a Cu(110) single crystal where benzene was the main desorption product after adsorption at
300 K. The cluster size-dependent yield of desorption products is attributed to the inability of small clusters to accommodate the dissociation
products prior to the trimerisation reaction.
2004 Elsevier Inc. All rights reserved.
Keywords: Molecular beam; TPD; trans-1,2-dichloroethene; Acetylene; Trimerisation; Benzene; Thin alumina film; Cu clusters
1. Introduction
catalytic activity, selectivity, and stability [3,4]. Therefore,
experiments following the reaction of unsaturated chlori-
nated molecules on Pt(111) and Pt(110) have been reported
[5,6], but the scale of the complexity of the reaction-phase
space has only recently been demonstrated by Bloxham et
al. [7] for trans-1,2-dichloroethene on Pd(110). The ad-
sorption of chloroethenes on Cu(100) has been reported
by Yang et al. [8] using near-edge X-ray absorption fine
structure spectroscopy, and on Cu(110) by Jugnet et al. [9]
using high-resolution electron energy loss spectroscopy.
We have previously reported on a comprehensive study of
the adsorption of trans-1,2-dichloroethene on Cu(110) us-
ing reflection absorption infrared spectroscopy (RAIRS),
temperature-programmed desorption (TPD), and molecu-
lar beam adsorption reaction spectroscopy (MBARS) [10].
Specifically, we have shown that the behaviour of trans-1,2-
dichloroethene on Cu(110) can be divided into three main
regimes: Regime 1, which occurs over 85–165 K in which
molecular adsorption occurs; Regime 2 in the 170–280 K
range in which a sequence of desorption/dechlorinationreac-
tions is triggered that leave adsorbed acetylene and Cl atoms
on the surface; and, finally, Regime 3 spanning the temper-
ature range 280–500 K is governed by the behaviour of the
The chemical reactions of chlorinated compounds are of
particular interest regarding environmental problems like air
and water pollution, as well as for industrial applications.
The latter include reactions with or on metallic surfaces, e.g.,
copper etching in semiconductor fabrication [1], or catalytic
processes like the production of the widely used polyvinyl
chloride [2]. Since the use of catalysts generally plays an
important role in the chemical industry, a good understand-
ing of chemical reactions between chlorinated compounds
and metal surfaces provides the basis for improving and de-
veloping efficient catalysts.
Due to their simplicity, chloroethenes are good model
systems for generating a better fundamental understanding
of the reactivity of unsaturated chlorinated compounds with
metal surfaces, and a small but growing literature database
now exists on these systems. With respect to hydrodechlori-
nation reactions, the group VIII metals, palladium, platinum,
and copper, appear to be the most attractive in terms of
*
Corresponding author.
E-mail address: R.Raval@liv.ac.uk (R. Raval).
0021-9517/$ – see front matter
doi:10.1016/j.jcat.2004.04.036
2004 Elsevier Inc. All rights reserved.

2
S. Haq et al. / Journal of Catalysis 226 (2004) 1–8
acetylene intermediate which, when created at low tempera-
tures, trimerises to form benzene that subsequently desorbs
at 370 K. However, if created at high temperatures, direct
desorption of acetylene competes with the trimerisation re-
action.
dent beam; this is due to adsorption onto the Cu evaporated
on the surroundings of the manipulator and chamber walls.
However, the beam flux remains constant, allowing compar-
ison between the different set of experiments to be valid.
A 5 Å Al2O3 film was created by exposing the clean sur-
face to oxygen and heating to 1200 K; this produced a sharp
LEED pattern characteristic of the alumina film [12]. The
film was then exposed to Cu vapour from the bead source;
the exposure determined the mean cluster size and ranged
from 60 to 480 s. This work was part of an STM study al-
ready reported [13] using the same bead source with as near
identical evaporation conditions as possible. In that study
it was shown that the growth of the Cu clusters proceeds
according to the Volmer–Weber mechanism, with the forma-
tion of 3-dimensional islands from the onset of nucleation.
The mean cluster size was found to change according to a
semilog scaling law with increasing Cu exposure. Represen-
tative examples of the size and distribution of the clusters
taken from our previous work are shown in Fig. 1 for 0.12
and 0.22 ML of Cu. The nucleation and growth of the clus-
ters occurs at the characteristic defect sites on the alumina
film, which can be step edges or reflectional and antiphase
domain boundaries. This results in large areas of the alu-
mina surface free of clusters. The clusters range typically
in size from 20 to 40 Å diameter and 6 to 8 Å high, and
are similar to those reported by others [14,15]. The diame-
ter is uncorrected for any effect that the tip apex may have in
measuring the particle shape and size, as it has been shown
that depending on the size and shape of the tip, the size of
the clusters can be overestimated with STM [16]. With in-
creasing exposure these grow in size and number density,
eventually coalescing into 3D chains of clusters. The STM
work of Worren et al. [14] has shown that when these clusters
are over 40 Å in diameter, the top facet has a (111) orienta-
tion.
In this report, we develop on our previous work and try to
bridge the “materials gap” that exists between real catalysts
and single crystals by studying the adsorption and reaction of
trans-1,2-dichloroethene on copper nanoclusters supported
on a thin Al2O3 film, using MBARS and TPD as a function
of temperature and mean cluster size. Although still under
idealised UHV conditions, this approach allows us to de-
termine the influence of the oxide support and whether the
cluster size influences the reaction mechanism, as often seen
for gas-phase clusters (an overview can be found in [11]).
Additionally, comparison with single crystal work is useful
for understanding any differences in reactivity due to the ma-
terials gap.
2. Experimental
A detailed description of the experimental procedure used
has been reported elsewhere [7,10]. Briefly, all experiments
were carried out using standard ultrahigh vacuum (UHV)
techniques and conditions. A NiAl(110) single crystal was
used to create a model Al2O3 film. Copper was evaporated
from a bead source made from melting high-purity Cu wire
on a tungsten filament. The purity of the liquid trans-1,2-
dichloroethene (Supelco, 99% purity) was checked by mass
spectrometry on admission into the UHV chamber. The ther-
mal molecular beam is generated by expanding the trans-
1,2-dichloroethene gas from a 0.2-mm quartz capillary and
collimated using a 0.3-mm skimmer. The data are shown
with the beam initially being reflected from an inert flag
in front of the crystal and then striking the sample at time
T = 0. The uptake is shown using the most intense cracking
fragment of trans-1,2-dichloroethene, mass 61, whose pro-
file is identical to that of the parent ion of mass 96 (35Cl iso-
tope). Mass 26 is also an intense cracking fragment of trans-
1,2-dichloroethene and for simple uptake, with no desorbing
products, the profile of this is identical to that of masses
61 and 96. However, at certain temperatures this correlation
breaks down due to the coincident evolution of acetylene
produced from the reaction of trans-1,2-dichloroethene with
the surface. Therefore, the data shown for mass 26 have been
normalised and subtracted accordingly to evaluate the pro-
duction of acetylene. Additionally, masses 2, 36, 54, and
104 were also measured for desorbing hydrogen, HCl, 1,2-
butadiene, and cyclooctatetraene, respectively; these prod-
ucts were not observed and hence not shown. However,
where evolution of the reaction product benzene (mass 78)
is observed it is shown accordingly. We note that evapora-
tion of Cu from the bead source in the UHV chamber is not
directional and leads to changes in the absolute intensity and
slope profiles of the mass spectrometer signals of the inci-
3. Results and discussion
3.1. Molecular beam data and sticking probabilities
The molecular beam experiments were initially con-
ducted on the clean alumina thin film at several different
temperatures. These showed no measurable sticking or up-
take of trans-1,2-dichloroethene between 300 and 450 K
(Fig. 2). This implies that the uptake and reaction prod-
ucts observed in the presence of Cu clusters can be directly
attributed to the Cu clusters and not the alumina support.
However, this does not necessarily imply that there is no in-
teraction between the incident molecules and the support,
only that within the limitations of the experimental tech-
nique being used it is not measurable. For comparison, also
shown in Fig. 1 is the uptake data measured on Cu(110) at
300 K.
Molecular beam experiments were also conducted on
three different Cu nanoclusters/Al2O3 systems, created by

S. Haq et al. / Journal of Catalysis 226 (2004) 1–8
3
(A)
(B)
Fig. 1. STM images of Cu clusters on the alumina thin film, (A) 0.12 ML Cu, 1300 × 1300 Å, and (B) 0.22 ML Cu, 1300 × 850 Å. I = 0.3 nA and V = 2 V.
t
t
varying the exposure of Cu, and correspond to approx-
imately 0.05, 0.22, and 0.5 ML Cu. The MBARS data
from these three systems are shown in Fig. 3. For each Cu
cluster sample, experiments were carried out over a range
of temperatures. At 300 K, simple uptake is observed for
all Cu cluster samples, with no desorbing reaction prod-
ucts. Furthermore, the uptake profiles are consistent with
precursor-mediated adsorption. The initial sticking proba-
bility, S0, changes from 0.34 to 0.56 as the number den-
sity and size of Cu clusters increase. The variation of S0
with temperature is shown in Fig. 4; for comparison the
data from Cu(110) are also included. On single crystals the
sticking probability is a measure of the product of conden-
sation into a physisorbed state and the probability of transfer
from this to the chemisorbed state, commonly referred to
as precursor-mediated adsorption, and the variation of the
sticking probability with temperature and coverage simply
reflects changes in these. Trapping into the physisorbed state

4
S. Haq et al. / Journal of Catalysis 226 (2004) 1–8
Fig. 4. The variation of the initial sticking coefficient S as a function of
0
temperature. Also included are data from Cu(110).
Fig. 2. Molecular beam data showing scattering of trans-1,2-dichloroethene
from the clean alumina film. The dotted line shows the uptake of
trans-1,2-dichloroethene on Cu(110) with a beam flux identical to that used
in these experiments. In both cases the beam strikes the sample at T = 0.
may itself be classified as being intrinsic or extrinsic, the dif-
ference being whether a molecule traps on the clean metal or
above an adsorbate-covered surface. For a surface support-
ing small metal clusters this trapping-adsorption mechanism
is equally applicable. However, in these cases, the experi-
mentally measured sticking coefficient is for the combined
system of clusters and support. Although the support may
be inert with respect to sticking and uptake, as is the case
here, accommodation of the incident molecules is still pos-
sible and this may be considered as an additional extrinsic
precursor state for the combined system. The importance
of this becomes apparent for supported clusters that show
relatively high sticking coefficients relative to their small
number densities. For example, the relatively high value for
the initial sticking probability for the lowest Cu cluster cov-
erages suggests that the incident molecules do interact with
the alumina film; i.e., the molecules are trapped into a weak
adsorption state in which they diffuse across the alumina
surface before either desorbing back into the gas phase or
finding a reaction site on a Cu cluster. Therefore, the ac-
tual sticking probability on these smaller clusters may in
fact be higher than that measured for the larger clusters or
the single crystal surface. It is interesting to note from Fig. 4
that smaller clusters do indeed show a different temperature-
dependent trend for the intial sticking coefficient than the
large clusters and Cu(110). This may be due either to the
lifetime in the precursor state on the alumina support be-
ing less temperature dependent than the metal surface due
to a relatively stronger interaction or to smaller clusters hav-
ing highly reactive sites which show a smaller temperature
dependence for adsorption. It is likely that the combination
of these factors contributes to the overall adsorption dy-
namics for the initial sticking, and as the cluster size and
number density increase the contributions change accord-
ingly.
3.2. Reaction pathways at the surface
Fig. 3. Molecular beam experiments showing the uptake and product evo-
lution from the reaction of trans-1,2-dichloroethene with Cu clusters sup-
ported on alumina as a function of temperature. From clusters formed with
(A) 0.05, (B) 0.22, and (C) 0.5 ML Cu. The beam strikes the sample at time
T = 0.
3.2.1. TPD data
The TPD data measured after the molecular beam experi-
ments at 300 K for the three different Cu nanoclusters/Al2O3

S. Haq et al. / Journal of Catalysis 226 (2004) 1–8
5
Fig. 6. Ratio of the mass 78 and mass 26 integrals from the desorption data
shown in Fig. 4.
almost completely absent for cluster formed from the lowest
Cu exposure.
3.2.2. Molecular beam data above 300 K
Molecular beam data in Fig. 3 show that above 330 K,
the uptake is accompanied by the simultaneous evolution of
gas-phase reaction products, benzene and acetylene. Inter-
estingly, the ratio of these products depends on both the Cu
cluster size and the temperature. For clusters formed from
the smallest Cu exposure, the formation of benzene as a re-
action product is not observed at any temperature in these ex-
periments. However, for clusters corresponding to 0.22 ML
Cu and above, benzene is clearly evolved, with a maximum
rate at about 353 K. In contrast, acetylene is always observed
as a reaction product, regardless of Cu cluster size or num-
ber density. On clusters where both products are observed,
the temperature-dependent ratio of these is very similar to
that previously reported for Cu(110). Above 400 K, the be-
haviour of all the systems converges in that only acetylene is
produced into the gas phase. This can be readily understood
since the surface is now well above the desorption tempera-
ture of acetylene, i.e., the acetylene immediately desorbs on
production as its lifetime on the surface is too short to allow
trimerisation reactions to occur.
Fig. 5. Desorption spectra showing evolution of benzene (mass 78) and
acetylene (mass 26) after adsorption of trans-1,2-dichloroethene at 300 K
on clusters formed with (A) 0.05, (B) 0.22, and (C) 0.5 ML Cu. For com-
parison data from Cu(110) are also shown (D).
systems are shown in Fig. 5. Essentially, the desorption
products can be associated with two main masses: 26 repre-
senting acetylene and 78 representing benzene. This clearly
shows that dechlorination of trans-1,2-dichloroethene to
form acetylene and the subsequent trimerisation of acety-
lene to form benzene are the major reaction pathways on the
Cu clusters. The main acetylene desorption peak shifts from
368 to 346 K with increasing cluster size and number den-
sity. Also shown in Fig. 5, for comparison, are the TPD data
from Cu(110) measured after exposure to a molecular beam
of trans-1,2-dichloroethene at 300 K. This latter reference
data are important, since on Cu(110) at 300 K, the acetylene
intermediate has been shown to almost exclusively trimerise
into benzene, which subsequently desorbs at 360 K. There-
fore, the sensitivity of our mass spectrometer to benzene and
acetylene, and the fraction of mass 26 which arises from
a cracking fragment of benzene in the mass spectrometer,
can be evaluated. A ratio of the mass 78 and mass 26 in-
tegrals is shown in Fig. 6. This clearly shows that whereas
the main reaction pathway on Cu(110) is the trimerisation
to benzene, the situation is quite different on the clusters
where the main reaction pathway is the desorption of acety-
lene after dechlorination. Some trimerisation into benzene
does occur for larger clusters and number densities, but is
3.3. Structure sensitivity in reaction pathways
Overall, it is clear from the TPD and molecular beam
data that the adsorption proceeds according to the mecha-
nism outlined below, and the overall behaviour is governed
by the relative rates of these reaction steps:
C2H2Cl2(ad) → C2H2Cl(ad) + Cl(ad)
C2H2Cl(ad) → C2H2(ad) + Cl(ad)
C2H2(ad) → C2H2(g)
2C2H2(ad) → C4H4(ad)
C2H2(ad) + C4H4(ad) → C6H6(ad)
C6H6(ad) → C6H26(g)
,
(1a)
(1b)
(2)
,
,
,
(3a)
(3b)
(4)
,
.

6
S. Haq et al. / Journal of Catalysis 226 (2004) 1–8
It would seem that process 1, dechlorination, occurs ef-
ficiently on all of the Cu cluster surfaces studied, i.e., it is
structure insensitive. However, the above data show distinct
differences in catalytic reaction processes (2) and (3) be-
tween different sizes of clusters, and between these and the
Cu(110) single crystal surface. Generally, the catalytic prop-
erties of clusters are related to either electronic or geometric
structure, although strictly these are interrelated. Geomet-
ric structure is defined by cluster size, shape, and surface
morphology. Additionally, even for a given morphology, in-
teratomic distances may vary as a function of cluster size,
and hence the electronic structure and catalytic activity. The
combination of these variables makes the process of relat-
ing the catalytic activity of clusters to a specific attribute
difficult, especially in this case where a range of cluster
sizes exists. However, the physical basis for the differences
observed in this work can be established by understand-
ing the reactivity of trans-1,2-dichloroethene and acetylene
on single-crystal surfaces and translating this to cluster sur-
faces.
(A)
The adsorption of acetylene has been studied on low in-
dex faces of Cu(100) [8], (111) [17], and (110) surfaces
[18,19]. On these surfaces, acetylene is believed to bond
with the surface in a di-σ/di-π coordination [20]. More
importantly for this study, acetylene trimerisation to form
benzene is very efficient on all three surfaces, on Cu(110)
with over 85% conversion [19]. This indicates that, inso-
far as the low index surfaces are concerned, the trimeri-
sation process is not too sensitive to the surface structure.
Furthermore, small Cu clusters supported on MgO(100)
also catalyze the trimerisation reaction of acetylene with a
high efficiency [21]. However, the trimerisation of acety-
lene formed from the reaction of trans-1,2-dichloroethene
on Cu single crystals does appear to be sensitive to sur-
face crystallography, as reports have shown for Cu(110) and
Cu(100) surfaces. On Cu(100), Yang et al. [8] report that
the acetylene predominantly desorbs without trimerisation
and attributes this to the presence of coadsorbed Cl poison-
ing this reaction pathway. We note that in our own study
on Cu(110), atomic Cl had no effect on the trimerisation
reaction, and the acetylene produced from the dechlorina-
tion reaction trimerised efficiently into benzene, regardless
of uptake. Furthermore an efficient trimerisation reaction
is observed with excess coadsorbed Cl, as observed from
the adsorption of trichloroethene on Cu(110) [22], where
the acetylene is formed from the disproportionation of an
ethynyl species. Clearly, the presence of the coadsorbed Cl
leads to significant reactivity differences on Cu(100) and
Cu(110) surfaces. A possible explanation for the difference
between these two surfaces may be due to segregation of
Cl into its own islands on Cu(110)—we observe a c(2 × 2)
LEED pattern at room temperature, which does not change
as the benzene desorbs from the surface. This segregation
allows the surface to retain Cl-free areas where the trimeri-
sation reaction can proceed. Therefore, the data from these
two Cu surfaces may provide a simple explanation for the
(B)
Fig. 7. Hard sphere models using Van der Waals radii showing surface cov-
erage for randomly adsorbed Cl and C H on the topmost facet of Cu
clusters with (A) 13 Å and (B) 20 Å diameters.
2
2
results from the Cu clusters supported on alumina; i.e., clus-
ters of different sizes expose different surface facets and
hence give rise to different catalytic activity. So, in the first
instance, clusters where benzene is not evolved may predom-
inantly expose facets that have (100) characteristics, while
on larger cluster sizes benzene may be produced from facets
that have (110) characteristics. To date, only the top (111)
facet for Cu clusters over 40 Å has been directly imaged
using STM [14]. However, it is not clear whether all clus-
ters over 40 Å have this surface morphology. Additionally,
no direct structural information is yet available for smaller
clusters. Furthermore, the vulnerability of small clusters to
adsorbate-induced reconstructions may make any compari-
son based on clean cluster morphology irrelevant.
However, the consideration of steric effects between
coadsorbed species leads to an alternative and simpler expla-
nation for the size-dependent catalytic activity for acetylene
desorption versus trimerisation reactions in this particu-
lar system. Unlike the simple adsorption and desorption
of molecules like CO, reactive adsorption of trans-1,2-
dichloroethene requires a critical number of empty sites, as
illustrated by Fig. 7, which shows surface coverage accord-

S. Haq et al. / Journal of Catalysis 226 (2004) 1–8
7
ing to the Van der Waals radii of the species on clusters
of 13 and 20 Å diameter, the latter being typical of the
smaller type of cluster we observe by STM. Therefore, when
lateral interactions between the resulting adsorbed species
are taken into account, a minimum size of cluster is nec-
essary for particular aspects of reactivity to be maintained.
Again, it is the influence of the coadsorbed Cl that is critical,
and there is evidence that it adversely affects both reaction
steps (2) and (3). Consider reaction step (2)—the stability of
acetylene is increased by the influence of coadsorbed Cl, as
evident from the TPD results where the desorption tempera-
ture is higher than for clean Cu cluster surfaces (330 K) [21].
For small clusters the desorption shifts to a higher tempera-
ture due to the forced proximity of these species increasing
the surface stability of acetylene and hence the barrier to its
desorption into the gas phase. However, the size of the clus-
ter and the influence of adsorbed Cl become even more crit-
ical when step (3), the trimerisation reaction, is considered.
Here, the equivalent of six adsorption sites on the cluster
is required for adsorbed Cl in addition to the sites required
for adsorption of the acetylene molecules prior to trimerisa-
tion. Furthermore, for those clusters that can accommodate
the minimum number of dissociated species involved in the
formation of benzene, there is the additional requirement of
sufficient physical segregation of Cl and acetylene to enable
trimerisation to occur. For small clusters, these requirements
are impossible to fulfils and reaction (3) is completely inhib-
ited, leaving reaction (2) as the only alternative. It is difficult
to predict what the critical cluster size needs to be to enable
the trimerisation reaction to occur efficiently. The detailed
mechanism for the trimerisation of acetylene to benzene has
been discussed previously for both Cu and Pd single crystal
and cluster surfaces [19,21,23–25]. Of particular relevance
to this study is the work of Lomas et al. [19] who, with the
careful use of isotope experiments, have suggested that the
most probable mechanism on Cu(110) consists of a two-step
process. The first step, which is rate limiting, involves the
reaction of two acetylene molecules to form a metallopen-
tacycle intermediate, C4H4M, step (3a). In the second step,
this intermediate reacts with a further acetylene molecule
to form benzene, step (3b). The work of Judai et al. [21]
has shown that this reaction mechanism also applies to sup-
ported Cu clusters, and, furthermore, they were able to detect
the desorption of a small quantity of a C4H6 species which
is produced from unreacted metallopentacylce. Application
of this mechanism to this work then sets a minimum size
of cluster, such that it allows two acetylene molecules to
be accommodated at positions that were not influenced by
the coadsorbed Cl to produce the C4H4 intermediate. This
requirement is obviously difficult for small clusters, and
even for the largest clusters studied here, the trimerisation
reaction is still highly unfavoured, indicating that accommo-
dation and segregation of the surface species are critical to
the overall surface chemistry.
Although the data and the above arguments clearly indicate
this, as otherwise acetylene trimerisation would readily oc-
cur as reported for clean Cu clusters [21]. We have additional
indirect evidence to support this from RAIRS experiments.
In these experiments, at low temperatures the intact mono-
layer adsorbed on the clusters is observed; however, at 300 K
no absorption peaks were observed. This may be due to the
sensitivity of RAIRS for this particular adsorption system,
as previously we have found that the IR absorption cross
sections for benzene and acetylene produced on Cu(110)
are extremely small [10]. However, a strong dipole-active
phonon mode of the Al2O3 thin film at 860 cm−1 has been
shown to be very sensitive to adsorption onto the film, as we
have previously reported [26] and as also discussed by Frank
et al. [27]. No change in this mode was observed on adsorp-
tion at 300 K, suggesting that spillover of species onto the
alumina is minimal.
Finally, we note that the main outcome of the simple con-
siderations laid out above naturally explains the different
structural sensitivities we report here compared to our previ-
ous work on the adsorption and reaction of NO on similarly
sized alumina-supported Cu clusters [26], where the parti-
cles shows almost identical reactivity characteristics to that
observed on Cu single crystals. This difference can simply
be attributed to the unique adsorption and reaction prop-
erties of these two systems. The adsorption of nitric oxide
shows a complex coverage-dependent reaction mechanism:
NO–NO interactions lead to the formation of a dinitrosyl
species, which is either stable on the surface or can decom-
pose into atomic O and N2O(ad) or (g), depending on surface
coverage. The latter can decompose further into O and N2(g)
depending on the availability of surface sites. Overall, the
key reaction is conversion of the monomer into the dimer,
which may be sensitive to particle surface crystallography
but is less demanding of cluster size, requiring at most two
to three adjacent sites. In contrast, the trimolecular reaction
for the conversion of acetylene to benzene, plus the steric in-
fluence of adsorbed Cl, requires a relatively more complex
arrangement of the adsorbed species, and therefore a larger
surface area for the reaction to proceed.
4. Conclusions
The adsorption and reaction dynamics of trans-1,2-di-
chloroethene on Cu clusters supported on a thin Al2O3 film
grown on NiAl(110) are found to be size dependent. Al-
though precursor-mediated adsorption is observed for all
cluster sizes and number densities, small clusters show sub-
tle differences that can be related to the interaction of the
incident molecules with the oxide support and a higher ad-
sorption probability on these clusters. At 300 K and above,
dechlorination is observed on clusters of all sizes with the
formation of benzene and acetylene as the main reaction
products. The ratio of these is strongly dependent on cluster
size; for small clusters only acetylene evolution is observed.
A basic assumption in the discussion above is that there
is no substantial spillover of atomic Cl onto the alumina.

8
S. Haq et al. / Journal of Catalysis 226 (2004) 1–8
With increasing cluster size the probability of trimerisation
of acetylene to produce benzene increases, although the des-
orption of acetylene remains dominant. The adsorption and
reaction are compared to previous work on a Cu(110) sin-
gle crystal where benzene was the only desorption product
after adsorption at 300 K. The cluster size-dependent yield
of the desorption products is attributed to the inability of
small clusters to accommodate and segregate the dissocia-
tion products in order for the trimerisation reaction to pro-
ceed.
[8] M.X. Yang, P.W. Kash, D.H. Sun, G.W. Flynn, B.E. Bent, M.T. Hol-
brook, S.R. Bare, A. Fischer, J.L. Gland, Surf. Sci. 380 (1997) 151.
[9] Y. Jugnet, N.S. Prakash, J.C. Bertolini, S.C. Laroze, R. Raval, Catal.
Lett. 56 (1998) 17.
[10] S. Haq, S.C. Laroze, C. Mitchell, N. Winterton, R. Raval, Surf.
Sci. 531 (2003) 145.
[11] A. Kaldor, D.M. Cox, Pure Appl. Chem. 62 (1990) 79.
[12] R.M. Jaeger, H. Kuhlenbeck, H.-J. Freund, M. Wuttig, W. Hoffmann,
R. Franchy, H. Ibach, Surf. Sci. 259 (1991) 235.
[13] C. Winkler, A. Carew, J. Ledieu, R. McGrath, R. Raval, Surf. Rev.
Lett. 8 (2001) 693.
[14] T. Worren, K.H. Hansen, E. Laegsgaard, F. Besenbacher, I. Stensgaard,
Surf. Sci. 477 (2001) 8.
[15] C.L. Pang, H. Raza, S.A. Haycock, G. Thornton, Surf. Sci. 460 (2000)
L510.
[16] D. Keller, Surf. Sci. 253 (1991) 353.
Acknowledgments
[17] R.L. Middleton, R.M. Lambert, Catal. Lett. 59 (1999) 15.
[18] N.R. Avery, J. Am. Chem. Soc. 107 (1985) 6711.
[19] J.R. Lomas, C.J. Baddeley, M.S. Tikhov, R.M. Lambert, Langmuir 11
(1995) 3048.
[20] N. Sheppard, Annu. Rev. Phys. Chem. 39 (1988) 589.
[21] K. Judai, S. Abbet, A.S. Worz, A.M. Ferrari, L. Giordano, G. Pacchio-
ni, U. Heiz, J. Mol. Catal. A: Chem. 199 (2003) 103.
[22] S.C. Laroze, S. Haq, R. Raval, Y. Jugnet, J.C. Bertolini, Surf. Sci. 433–
435 (1999) 193.
[23] R.M. Ormerod, R.M. Lambert, in: R.J. Madix (Ed.), Surface Reac-
tions, in: Springer Series in Surface Science, vol. 34, Springer, Berlin,
1994.
[24] C.H. Patterson, J.M. Mundenar, P.Y. Timbrell, A.J. Gellman, R.M.
Lambert, Surf. Sci. 208 (1989) 93.
We are grateful to the EPSRC for an equipment grant and
a research fellowship to C.W. We also thank EPSRC and
Daresbury Laboratory for a CASE studentship for A.C. We
also thank R. McGrath for use of STM facilities.
References
[1] J.H. Brannon, K.W. Brannon, J. Vac. Sci. Technol. B 7 (1989) 1275.
[2] B.C. Gates, Catalytic Chemistry, Wiley, New York, 1992.
[3] S.C. Fung, J.H. Sinfelt, J. Catal. 103 (1987) 220.
[4] L.J. Matheson, P.G. Tratneyk, Environ. Sci. Technol. 28 (1994) 2045.
[5] V.H. Grassian, G.C. Pimentel, J. Chem. Phys. 88 (1988) 4478.
[6] A. Cassuto, M.B. Hugenschmidt, P. Parent, C. Laffon, G. Tourillon,
Surf. Sci. 310 (1994) 390.
[25] I.M. Abdelrehim, N.A. Thornburg, J.T. Sloan, T.E. Cadwell, D.P.
Land, J. Am. Chem. Soc. 117 (1995) 9509.
[26] S. Haq, A. Carew, R. Raval, J. Catal. 221 (2004) 204.
[27] M. Frank, K. Wolter, N. Magg, M. Heemeier, K. Kuhnemuth, M. Bau-
mer, H.-J. Freund, Surf. Sci. 492 (2001) 270.
[7] L.H. Bloxham, S. Haq, C. Mitchell, R. Raval, Surf. Sci. 489 (2001) 1.