Structure-Reactivity Correlations in the Catalytic Coupling of Ethyne over Novel Bimetallic Pd/Sn Catalysts

Article abstract of DOI:10.1021/jp9638346

The structure, thermal stability, and catalytic behavior of a novel highly dispersed silica-supported Pd/Sn catalyst prepared by an organometallic route have been examined by X-ray photoelectron, X-ray diffraction, and X-ray absorption, fine structure spectroscopies, the latter two measurements being carried out with an in situ reaction cell.Additional reactor measurements were performed on a more Sn-rich catalyst and on a pure Pd catalyst.Varying the temperature of reduction induced large variations in catalytic performance toward ethyne-coupling reactions.These changes are understandable in terms of the destruction of SnO2-like structures surrounding the Pd core, yielding a skin of metallic Sn which subsequently undergoes intermixing with Pd.The overall thermal and catalytic behavior of these highly dispersed materials accords well with the analogous single-crystal model system.

Full text of DOI:10.1021/jp9638346

J. Phys. Chem. B 1997, 101, 2797-2805
2797
Structure-Reactivity Correlations in the Catalytic Coupling of Ethyne over Novel
Bimetallic Pd/Sn Catalysts
Adam F. Lee, Christopher J. Baddeley, Christopher Hardacre,† Geoffrey D. Moggridge,‡
R. Mark Ormerod,§ and Richard M. Lambert*
Department of Chemistry, UniVersity of Cambridge, Cambridge CB2 1EW, England
Jean P. Candy and Jean-M. Basset
Laboratoire de Chimie Organome´tallique de Surface, UMR CNRS-CPE 9986, CPE Villeurbanne, France
ReceiVed: NoVember 18, 1996; In Final Form: January 28, 1997X
The structure, thermal stability, and catalytic behavior of a novel highly dispersed silica-supported Pd/Sn
catalyst prepared by an organometallic route have been examined by X-ray photoelectron, X-ray diffraction,
and X-ray absorption, fine structure spectroscopies, the latter two measurements being carried out with an in
situ reaction cell. Additional reactor measurements were performed on a more Sn-rich catalyst and on a pure
Pd catalyst. Varying the temperature of reduction induced large variations in catalytic performance toward
ethyne-coupling reactions. These changes are understandable in terms of the destruction of SnO2-like structures
surrounding the Pd core, yielding a skin of metallic Sn which subsequently undergoes intermixing with Pd.
The overall thermal and catalytic behavior of these highly dispersed materials accords well with the analogous
single-crystal model system.
Introduction
monometallic catalysts were prepared by ion exchange of [PdCl-
(NH3)5]2+ onto a 250 m2 g-1 silica support. The resulting
The catalytic coupling of ethyne over Pd and Pd-derived
surfaces provides a relatively complex network of reactions
which exhibit very pronounced sensitivity to the structure1 and
composition2 of the metal surface. It is therefore a useful system
for examining mechanistic aspects of hydrocarbon conversion.
The application of alloy catalysts provides an additional
dimension and can yield results that are of wider significance
in fields such as catalytic hydrocarbon reforming.3,4
Our earlier work on Au{111}/Pd5 and Pd{111}/Au2 model
catalysts gave results that were of value in illuminating the
behavior of practical, dispersed Au/Pd core/shell catalysts.6
Recently, we reported on the structural and thermal properties
of the Pd{111}/Sn model planar system.7 Here we describe
corresponding structural and catalytic results obtained with a
novel type of a highly dispersed Pd/Sn catalyst in which a
coating of Sn is applied to a Pd core by an organometallic route.
For comparison purposes, additional catalytic measurements
were carried out on a more Sn-rich catalyst and on a pure Pd
catalyst; the Pd content was identical in the three cases. It is
found that large variations in catalytic perfomance are under-
standable in terms of the deliberately induced changes in surface
structure and composition. Furthermore, in this case too,
Pd{111}/Sn single-crystal data provide relevant insight. Com-
parisons are made with the related Pt{111}/Sn system.7
surface complex was calcined in air at 673 K and reduced at
673 K in H2 to yield 10-20 Å Pd particles with a dispersion of
∼0.75,9 and metal loading of 2 wt % Pd. The subsequent
surface reaction was performed in the liquid phase (n-hexane
solvent) in a stainless steel autoclave between the desired
amounts of tetra-n-butyl tin Sn(n-C4H9)4 and the monometallic
Pd catalyst by reduction at 473 K under 30 mbar hydrogen.
The resulting slurry was refluxed in boiling n-hexane under Ar
at 370 K for 24 h to remove unreacted Sn(n-C4H9)4 and was
dried in Ar, and the catalyst samples were sealed under an Ar
atmosphere for transportation. All measurements were per-
formed on deliberately air-exposed samples. The amounts of
monometallic Pd/SiO2 and Sn(n-C4H9)4 were chosen to give a
ratio of Sn to surface Pd atoms of Sn:Pds ) 0.5 and Sn:Pds )
0.66. Hereafter, these samples are designated Pd2Sn and Pd3-
Sn2 by reference to the idealized surface stoichiometries.
In situ X-ray diffraction (XRD) experiments were performed
at 300 K, in 1 bar He with a Siemens D500 diffractometer using
Cu KR radiation (154 pm), and with a specially constructed
silica reaction cell.10 This cell permits both XRD and X-ray
fine absorption structure (XAFS) studies to be carried out in
controlled, reactive atmospheres. Powdered catalyst samples,
typically ∼100 mg, were placed on a ceramic/silica sinter fitted
into a recess of the hollow silica sample holder. The XRD/
XAS (XAS ) X-ray absorption spectroscopy) cell was attached
to a Perkin-Elmer 700HWD GC (Poropak-N and 5X molecular
sieve), allowing precise monitoring of the catalyst environment.
The sample cell could also be operated as a single-pass flow
reactor, and the catalytic data presented here were obtained in
Experimental Section
Two silica-supported Pd/Sn bimetallic catalysts were prepared
by the controlled surface organometallic reaction (CSOR)
method developed by Basset and co-workers.8 The Pd/SiO2
this way. Reductions were performed in H2 (10 mL min-1
,
† School of Chemistry, Queen’s University, Belfast BT9 5AG, Northern
Ireland.
distillers 4.6, 20 K/min) at temperatures between 300-873 K
for 20 min, followed by purging and cooling in He. Data were
collected between 2θ ) 35-85° at 0.02° increments with
spectral acquisition times of 500-1000 min/scan.
XAFS measurements were performed as a function of catalyst
reduction temperature on Stations 9.2 and 9.3 of the Daresbury
‡ Department of Chemical Engineering, University of Cambridge,
Cambridge CB2 3RA, England.
§ Department of Chemistry, University of Keele, Keele, Staffs ST5 5BG,
England.
* Corresponding author.
X Abstract published in AdVance ACS Abstracts, March 1, 1997.
S1089-5647(96)03834-5 CCC: $14.00 © 1997 American Chemical Society

2798 J. Phys. Chem. B, Vol. 101, No. 15, 1997
Lee et al.
Figure 1. XRD pattern from Pd2Sn/SiO2 reduced in H2 at 873 K.
SRS facility using the in situ reaction cell. Both the Pd and Sn
K-edges (at 24.35 and 29.19 keV, respectively) were monitored
in fluorescence mode at 300 K in He, using a Canberra solid-
state multichannel detector, mounted in the horizontal plane.
Transmission spectra of 8 µm Pd and Sn foils, together with
SnO and SnO2 powder standards, were also measured. Signal
to noise ratio was optimized by acquiring multiple scans for
every spectrum. Data analysis employed the Daresbury
EXBROOK and EXCURVE92 packages for background sub-
traction and subsequent theoretical fitting, respectively. Fourier-
filtering procedures were not employed in this study, due to
the quality of background subtraction and absence of significant
artifacts adjacent to the principal fitted coordination shells.
Catalyst testing was performed under steady-state conditions
as a function of the H2:C2H2 reactant ratio and as a function of
the reaction or reduction temperature at 1 bar total reaction
pressure. Product yields stabilized after ∼45 min and remained
constant for >5 h. The errors in quoted yields and selectivities
are ∼(5% for the major (C6) products and (10% for the minor
(C4) products. The activity was calculated Via carbon mass-
balance, thus
Figure 2. (a) Pd 3d and (b) Sn 3d photoemission core-level spectra
from air-exposed Pd2Sn/SiO2.
of the 300-873 K reduced samples, the Scherrer equation yields
a constant fcc lattice parameter of ∼3.9 Å and corresponding
volume-averaged particle size of e35 Å for cubeoctahedral/
spherical particle morphologies. Note the errors in the XRD-
derived lattice parameter (∼(0.1%), and particle diameters
((20 Å) are significant, due to the low signal intensities and
limiting particle sizes. This lattice parameter and the additional
refelections match well to those of pure Pd. No reflections
characteristic of tin oxide or crystalline Pd/Sn alloys were
observed for any sample. Minority, intermediate PdxSn were
also undetected; note that no random substitutional Pd/Sn alloys
are reported in the literature.12
activity ) no. of moles of C2H2 consumed
)
(n/2) (no. of moles of Cn product)
∑
n)4,6
and product selectivity is defined as
moles of Cn product
X-ray Photoelectron Spectroscopy. Photoelectron spectra
of the as-prepared Pd2Sn/SiO2 sample were measured in order
to determine the electronic state of both metallic components.
The principal Pd and Sn 3d3/2,5/2 X-ray photoemission transitions
at ∼336 and ∼485 eV binding energy (BE), respectively, are
presented in Figure 2a,b. A single Pd 3d5/2 line is present at
∼336.3 eV BE (Figure 2a), with a shape characteristic of the
bulk material but shifted by ∼1 eV to higher BE.13 The Sn
3d5/2 line has a composite shape (Figure 2b), which standard
Shirley background subtraction and deconvolution procedures
reveal to comprise a majority component at ∼486.8 eV BE,
together with a weaker lower BE state at ∼484.5 eV. For
reference, the 3d5/2 line of bulk R-Sn occurs at 484 eV.14
X-ray Absorption Fine Structure Studies. The local
structural and electronic environment of each metal component
in the Pd2Sn sample were investigated by X-ray absorption
spectroscopy as a function of H2 prereduction temperature.
Additional measurements on the Pd3Sn2 sample gave quanti-
tatively similar results to those for Pd2Sn and are thus not
reported here. Results from the deliberately air-exposed and
control samples were in good quantitative agreement. Figure
3a,b show the respective raw Pd and Sn K-edge EXAFS
(extended X-ray absorption fine structure spectroscopy) data
Sproduct
)
moles of Cn product
∑
n)4,6
X-ray photoelectron spectra were recorded at 300 K in a VSW
ARIES system11 equipped with a HA100 spectrometer (Mg KR)
at 40 eV pass energy. The sample was prepared as a
homogeneous powder supported on adhesive tape, with energy
referencing made to the adventitious carbon line at 284.3 eV to
compensate for charging effects.
Results
X-ray Diffraction Studies. Very weak, common diffraction
features were observed for all PdxSn/SiO2 samples irrespective
of the reductive pretreatment or scan time used. Figure 1
illustrates the XRD pattern for a Pd2Sn/SiO2 sample prereduced
at 873 K in H2. Three weak, broad peaks associated with a
metal phase are visible at ∼40°, 46° and 68°, the corresponding
positions for bulk fcc Pd and R-Sn are shown for comparison.
The sloping background and poor signal to noise reflects the
low counts detected from the dilute samples and the absence
of large metal particles. From the 40° ([111]) diffraction peak

Catalytic Coupling of Ethyne
J. Phys. Chem. B, Vol. 101, No. 15, 1997 2799
Figure 3. Raw (a) Pd and (b) Sn K-edge EXAFS spectra from Pd2-
Sn/SiO2 as a function of H2 reduction temperature. Spectra from Pd
and Sn foils and SnO/SnO2 powder standards are shown for comparison.
Figure 4. Background-subtracted raw k3-weighted (a) Pd and (b) Sn
ø data from Pd2Sn/SiO2 as a function of H2 reduction temperature.
for Pd2Sn together with standard transmission spectra from 8
µm Pd and Sn foils and SnO/SnO2 powders for comparison.
The raw Pd-edge data for Pd2Sn display no clear variation with
increasing reduction temperature and closely resemble the
standard foil spectrum, hence only the as-received and 873 K
reduced spectra are discussed. In contrast, the raw Sn-edge
spectra display a dramatic decrease in the white line (edge-
step height) intensity and in the amplitude and periodicity of
the EXAFS oscillations, with increasing reduction temperature.
Comparison of the raw Pd2Sn data with the Sn standards
suggests a progressive transformation from a SnO2/SnO-like
phase at 300 K to a metallic phase after 873 K reduction.
The background-subtracted, k3-weighted Pd and Sn raw
EXAFS functions appear in Figure 4a,b, respectively, together
with their associated theoretical fits for various reduction
treatments. No variation is apparent between the Pd-edge
spectra, which are directly superimposable. In contrast, upon
reduction, the Sn-edge spectra exhibit gross changes. Again,
the as-received and 300 K reduced Sn-edge spectra resemble
that of SnO2, while the 873 K spectrum closely matches the Sn
foil EXAFS function. The structural transformations ac-
companying reduction are more apparent from the Pd- and
Sn-edge unfiltered RDFs, shown together with their theoretical
fit functions in Figure 5a,b, respectively. The fitted parameters
derived for these figures are quantified in Table 1.
Sn-C and oxidic Sn-O bond lengths are both ∼2 Å. In the
absence of surface hydrocarbon fragments9 this shell is assigned
to Sn-O scattering. An excellent fit to the unreduced Pd2Sn
data was achieved using bulk SnO2 fit parameters as well as a
constrained iteration of the Sn-O Debye-Waller and phase-
shift correction factors (σ1Sn-O and Ef). No fit could be
achieved using the parameters of SnO.
Increasing reduction temperatures induce a gradual decrease
in the number of nearest-neighbor O atoms from ∼6 at 300 K
to effectively zero following an 873 K reduction (Figure 5b).
A new sharp, high R scattering shell simultaneously emerges
at ∼2.6-2.7 Å. This shell could not be fitted using oxygen
scatterers, and in any case it does not correspond to any known
bulk oxide of Sn. This interatomic distance is thus assigned to
heavier, metallic scatterers, although distinguishing beween Sn
and Pd nearest-neighbor atoms is not trivial, due to their similar
scattering power. The Sn-Sn nearest-neighbor distance varies
from 3.02 to 2.8 Å between bulk R- and â-Sn structures, and
therefore complicates the accurate estimation of Sn-Pd bond
distances in fcc-like bimetallic particles. The bulk Pd-Pd
separation of 2.75 Å is closest to the observed 2.65 Å shell
distance (which may be contracted due to Sn charge-polariza-
tion). This shell is thus reasonably attributed to nearest-neighbor
Pd atoms, an assumption supported by the final “high”
coordination of ∼5 attained after an 873 K reduction. Sn-Sn
coordination of ∼5 is unlikely considering the low initial Sn:
Pd atom ratio and the resultant relatively low maximum possible
Sn surface concentration of ∼0.6616 Via seeding with bulky tert-
butyl ligands. The absence of Sn X-ray diffraction features is
The as-prepared (air-exposed) Sn-edge (RDF) of Pd2Sn
exhibits a single major peak at ∼2 Å, Figure 5b. The presence
of tert-butyl groups on unreduced NixSn/SiO2 catalysts yields
a similar peak attributed to a Sn-C shell distance.15 By
analogy, the present peak distance is too short to correspond to
a Sn-metal scattering coordination shell, while organometallic

2800 J. Phys. Chem. B, Vol. 101, No. 15, 1997
Lee et al.
Figure 6. Normalized, derivative (a) Pd and (b) Sn K-edge XANES
spectra from Pd2Sn/SiO2 as a function of reduction temperature. Spectra
from Pd and Sn foils and SnO/SnO2 powder standards are shown for
comparison.
and confirmed that Pd atoms reside in a truncated bulk Pd-like
environment. A weak shoulder to low R on the principal Pd
peak is a termination error resulting from Fourier transformation
of the EXAFS data over a finite k-range.17
Figure 5. Unfiltered (a) Pd and (b) Sn K-edge pseudo-radial-
distribution functions obtained from Pd2Sn/SiO2 as a function of H2
reduction temperature. Data from a pure SnO2 bulk standard are shown
for comparison.
Absorption features in the energy range ∼0-50 eV above
the edge jump can also provide information on the valence
electronic structure of the excited atom.18 Core-ionized pho-
toelectrons possessing a sufficiently low kinetic energy may
become trapped in bound or quasibound unoccupied states close
to the Fermi energy of a sample,19 thus modulating the smooth
atomic absorption function. The occurrence of such transitions
is controlled by the dipole-selection rule (∆l ) (1), and where
this is satisfied, their intensity reflects the local excited atom
density of states.20 The Sn K-edge X-ray absorption near edge
structure (XANES) is thus dominated at low energies by allowed
1s f 5p transitions (analogous to L1-edge spectra20) and can
yield information on the Sn 5p occupancy. In contrast, the Pd
valence band possesses primarily d-orbital character. Conse-
quently, 1s f 4d transitions are only weakly allowed (Via p-d
rehybridisation), and low-energy Pd K-edge XANES features
are thus weak and are less sensitive probes of the Pd d-band
occupancy.
TABLE 1
300/K
473/K
673/K
873/K
Sn K-edge
N1Sn-O
N2Sn-Pd
6
3
3.3
1
3.1
4.7
R1Sn-O/Å
R2Sn-Pd/Å
σ1Sn-O/Å2
σ2Sn-Pd/Å2
2.05
0.008
2.02
2.64
0.013
0.017
2.01
2.67
0.007
0.015
2.65
0.0174
Pd K-edge
N1Pd-Pd
N2Pd-Pd
8
4.7
8.2
4.7
R1Pd-Pd/Å
R2Pd-Pd/Å
σ1Pd-Pd/Å2
σ2Pd-Pd/Å2
2.77
3.87
0.023
0.039
2.77
3.94
0.022
0.034
Analysis of fine structure within the XANES region is usually
performed on the normalized, derivative spectra. Differences
in the relative intensities of absorption features are thus greatly
enhanced. Figure 6a,b show the resultant Sn and Pd derivative
XANES spectra as a function of reduction temperature. The
energy scale of each spectrum is referenced to the first maximum
(E0) to compensate for experimental variations in the raw data.
The Sn XANES exhibits a sharp maximum at E0 and a sharp
minimum at ∼+10 eV. The magnitude of both features for
the as-received sample closely matches both oxide standards
and decreases systematically with increasing reduction temper-
ature. (The difference between the oxidized and reduced cases
is indicated by the double-headed arrows in Figure 6b). The
in accord with this view. Discriminating between possible
Sn-Sn and Sn-Pd scattering pairs from the theoretical fits alone
proved inconclusive, although disorder factors were in better
agreement with those from Pd foil as opposed to Sn foil.
Reduction treatments thus result in the progressive elimination
of Sn-O nearest neighbors and their replacement by Sn-Pd
nearest neighbors. This process was essentially complete above
673 K.
The Pd-edge RDFs exhibit a single, major peak at ∼2.7-
2.8 Å for all reduction temperatures. This feature can be
accurately fitted using bulk Pd foil parameters and a reduced
Pd-Pd nearest-neighbor atom coordination of ∼8. The inclu-
sion of Pd-Sn scattering pairs served only to worsen the fit

Catalytic Coupling of Ethyne
J. Phys. Chem. B, Vol. 101, No. 15, 1997 2801
Figure 7. Activity of PdxSn/SiO2 catalysts as a function of reactant
mix and Sn loading. Treaction ) 540 K.
first peak maximum is assigned to 1s f 5s,p transitions,20 and
thus reflects the Sn valence band occupancy. It is apparent that,
following reduction at either 673 or 873 K, the features above
∼10 eV are significantly shifted. In contrast, the phase and
magnitude of the principal Pd XANES maxima and minima
are unperturbed by reduction, mirroring those of the Pd foil
standard.
Catalytic Measurements
The in-situ reaction data, which provide a complementary
surface sensitive probe to the preceding spectroscopic measure-
ments, reveal the exclusive formation of C4 and C6 cyclization
products at elevated temperature over all 3 PdxSn/SiO2 samples.
The principal reaction products observed under all reaction
conditions, for all sample pretreatments, in order of yield, were
benzene, n-hexane, isobutene, and trans- and cis-2-butene. Trace
cyclohexene and cyclohexane were also observed.
Figure 8. Selectivity toward (a) C6 and (b) C4 products of PdxSn/
SiO2 catalysts as a function of reactant mix and Sn loading. Treaction
)
540 K.
Influence of Reactant Gas Composition. Figure 7 shows
the catalytic activity at 540 K as a function of the loading ratios
of H2:C2H2 and Sn:Pd for samples prereduced at 540 K. The
steady-state activities of all three catalysts exhibit very similar
hydrogen partial pressure dependence. A rapid increase occurs
from essentially zero conversion in pure ethyne to a maximum
for H2:C2H2 ∼1-1.5, decaying slowly to low (nonzero) activi-
ties for ratios >5. The maximum activity increases with Pd:
Sn ratio, corresponding to 58, 86, and 95% conversion for the
Pd, Pd2Sn, and Pd3Sn2 samples, respectively. The peak Pd3-
Sn2 activity also occurs at a slightly lower H2:C2H2 ratio than
for the other samples. For H2:C2H2 ratios <1, all rates exhibit
positive order in H2, while for H2:C2H2 ratios >2, all rates show
negative order in H2.
The principal reaction products for all H2:C2H2, and Sn:Pd
ratios were benzene and n-hexane. The yields of these two
products exhibit strong anticorrelation as a function of the H2:
C2H2 ratio (Figure 8a). All samples exhibit similar dependencies
for increasing H2:C2H2 ratios between 0-1.2. Sbenzene falls from
∼90 to 50%, while Sn-hexane increases in the same manner from
∼0 to 50% over this regime. This transition, from essentially
pure benzene to predominantly n-hexane production, continues
for the Pd and Pd2Sn catalysts at higher H2 partial pressures,
and is complete for H2:C2H2 g1.75. In contrast, the trends in
the Pd3Sn2 catalyst selectivity are much less pronounced for
H2:C2H2 ratios g1.2, yielding limiting Sbenzene and Sn-hexane
values of ∼20% and 75%, respectively. Note that all the
minority C4 product selectivities track the variation in Sbenzene
with H2:C2H2 for the Pd and Pd2Sn samples. Figure 8b
illustrates the case for Strans-butene. The Pd3Sn2 C4 selectivities
display a similar dependence (tending to zero production), and
thus appear to deviate from Sbenzene for ratios g2. This
Figure 9. Activity of PdxSn/SiO2 catalysts as a function of H2
prereduction temperature and Sn loading. Treaction ) 540 K.
observation may reflect the very low C4 yields, and thus large
measurement error in these latter data points. The peak C4
selectivities decrease in the order Strans-butene (∼6%) > Sisobutene
(∼5%) > Scis-butene (∼3%).
Influence of Reduction Temperature. The activities of the
PdxSn catalysts as a function of prereduction temperature and
Sn:Pd loading are shown in Figure 9. The dependence of
activity on reduction temperature increases with Sn loading.
Thus the activity of the pure Pd sample was completely
insensitive to all reduction treatments employed, while the Pd2-
Sn and Pd3Sn2 activities decrease by ∼8 and 13%, respectively,
following reduction between 300-1073 K. The maximum
activities again equate to 58, 86, and 95% for the Pd, Pd2Sn,
and Pd3Sn2 samples, respectively. Figure 10 shows the corre-
sponding C6 selectivities of the Pd and Pd2Sn samples. Data
for the monometallic catalyst do not exhibit any significant

2802 J. Phys. Chem. B, Vol. 101, No. 15, 1997
Lee et al.
particle size of e30 Å. Pd K-edge EXAFS data (Figure 5a),
which indicate bulk Pd-Pd separations and the absence of Pd-
Sn nearest neighbors, confirm the majority of Pd atoms are
initially located in a “pure” Pd phase. An independent estimate
of the Pd cluster size may be obtained from the nearest-neighbor
coordination number. For average particle sizes below ∼100
Å, high-temperature (g300 K) EXAFS analysis yields artifically
reduced coordination numbers due to anharmonic atom vibra-
tions.21 Employing the correction factors determined by Clausen
et al., the average, “true” EXAFS-derived Pd particle size is
∼10-30 Å, in agreement with the XRD measurements. This
is in line with the observed high Pd 3d BE (∼336.3 eV), which
is shifted by ∼+1 eV relative to the bulk. For small metal
clusters supported on insulating substrates with which they
weakly interact (e.g. SiO2), such core-level shifts predominantly
reflect changes in final-state, core-hole screening processes
relative to the bulk environment.22 The magnitude of the shift
is thus a strong function of the cluster size, with which it is
inversely correlated; previous studies of Pd clusters deposited
on silica indicate that a 1 eV shift is associated with ∼10-20
Å particles.23
Figure 10. Selectivity toward C6 products of PdxSn/SiO2 catalysts as
a function of H2 prereduction temperature and Sn loading. Treaction
)
540 K.
Basset et al. have shown the CSOR method is highly
selective, localizing the SnR4 ligands exclusively on the metallic
clusterssinteraction with silanol groups on the support is
negligible.9 The absence of significant Sn-Pd scattering in the
as-prepared sample presumably reflects the effect which might
be expected to result in tin oxide formation at the cluster surface.
Indeed, studies of ultrathin Sn films grown on a Pd{111} single-
crystal demonstrate that the low-pressure oxidation of submono-
layer Sn deposits is extremely facile and proceeds rapidly at
300 K.24 These findings are in accord with our Sn K-edge
EXAFS data which indicate that the majority of Sn atoms are
initially present within a SnO2 film, Figure 5b. However, the
Sn 3d XPS spectra do provide clear evidence for the retention
of some Sn (∼10% of the grafted overlayer) within a metallic
environment. These Sn0 atoms may be either localized on the
surface of the Pd particles, as a partial encapsulating overlayer,
or present within a Pd/Sn surface alloy phase. In contrast with
unannealed Sn overlayers, single-crystal studies show that Sn
atoms incorporated within Pd/Sn surface alloys are highly
resistant to oxidation even at pressures of >50 Torr O2.24
Oxidation (g10-4 Torr O2) induces cracking of thin Sn films
supported on Pd{111}.24 It is thus unlikely that the SnO2 phase
formed during room temperature air-exposure constitutes a
continuous encapsulating layer around the Pd clusters. Exposure
of bare Pd sites following disruption of the Sn oxide overlayer
may also contribute to the room temperature reactivity of the
as-prepared Pd2Sn sample, discussed later. This cracked oxide
phase is XRD invisible, and given the absence of higher Sn-O
coordination shells in the Sn EXAFS, it could be little more
than a monolayer thick. EXAFS measurements on silica-
supported Ge/Ru bimetallic clusters prepared by CSOR25 also
indicate that air exposure yields a partially encapsulating GeOx
structure, similar to that proposed here. Note also that our
results eliminate the possibility that substantial Sn oxide
agglomerates were present at the cluster/silica interface.
Figure 11. Selectivity toward C6 products of a 540 K prereduced Pd2-
Sn/SiO2 catalyst as a function of reaction temperature.
trends. However, for Pd2Sn, a rise in Sbenzene and a correspond-
ing fall in Sn-hexane (both by ∼12%) with reduction temperature
are apparent. A similar, less pronounced transition between
benzene and n-hexane production also occurs for the Pd3Sn2
sample (not shown for clarity)sthese smaller changes may
reflect the initial high Sbenzene (∼85%) and low Sn-hexane (∼4%)
of the as-prepared samples. The C4 selectivities display no
systematic variations, and individual product selectivities remain
at levels <5%. In all cases the maximum total C4 selectivity
is <7%, and the measurement error for these small yields is
significant.
Influence of Reaction Temperature. The dependence of
C6 selectivities on reaction temperature is shown in Figure 11.
The selectivities toward all C4 products (not shown) exhibited
a slight decrease with increasing reaction temperature: Siso-butene
from 5% to 4%, Scis-butene from ∼5% to ∼2%, Strans-butene from
∼2% to ∼1%.
Discussion
Structural Characterization of the Pd2Sn Catalyst. Previ-
ous studies of the Rh,Ni/Sn systems prepared by CSOR routes
indicate that submonolayer quantities of Sn(n-C4H9)4 are
selectively anchored to, and undergo hydrogenolysis exclusively
at, the surface of small (∼10-15 Å) metallic Rh and Ni
particles. Low reduction temperatures stabilize the Ms[Sn(n-
C4H9)x]y surface organometallic fragments.9 In the present case,
the hydrogenolysis reaction was performed at 473 K, sufficient
to remove the (n-C4H9)4 ligands, yielding pure metallic Pd
particles partially encapsulated by a Sn overlayer.
Reduction temperatures above 300 K progressively destroy
the Sn oxide phase: Sn K-edge EXAFS shows that nearest-
neighbor oxygen scatterers around each Sn atom are progres-
sively replaced by Pd, Figure 5b. Although this process
becomes detectable after reduction at room temperature, sig-
nificant oxygen loss only occurs above 473 K, with an
accompanying reduction in the Sn-O coordination from 6 to
∼3. A scheme which is consistent with our results is illustrated
in Figure 12 which represents the conversion of the SnO2 Via
The XRD-derived lattice parameter for the Pd2Sn sample
indicates the presence of small Pd clusters with an average

Catalytic Coupling of Ethyne
J. Phys. Chem. B, Vol. 101, No. 15, 1997 2803
environment, in accord with the EXAFS data. However, the
873 K reduced Sn XANES still exhibit a significant shift relative
to the bulk Sn standard, suggesting incorporation of Sn within
an electronically distinct Pd-Sn surface alloy.
Our observations regarding the structural evolution of these
cluster-derived Pd/Sn catalysts are in general accord with studies
of supported Pt/Sn catalysts prepared by wet-impregnation/
coprecipitation.3,30,31 These report that low-temperature reduc-
tion yields Sn2+ and/or Sn4+, although there is disagreement as
to whether the Sn oxide is coordinated to the support material.
Reduction at >673 K results in loss of oxygen and induces
strong Sn-Pt interaction, although in contrast with the present
system, complete reduction of Sn to the metallic state is not
generally observed.30,31 Our CSOR method appears to produce
better defined precursor material in which the Sn is apparently
associated exclusively with the Pd clusters, resulting in efficient
and complete reduction to a true bimetallic system.
Figure 12. Schematic illustrating progressive destruction of surface
SnO2 particles and subsequent Pd/Sn surface alloying for PdxSn/SiO2
catalysts as a function of reduction in H2.
disordered/nonstoichiometric SnO to a bimetallic system. In
this connection, it is worth noting that ultrathin Sn oxide films
on Pd{111} also begin to decompose at ∼473 K.24 In contrast,
bulk SnO2 and SnO are stable up to 1000 K,26,27 illustrating the
much lower stability of the supported Sn oxide layer on both
clusters and extended Pd surfaces. This scheme is supported
by a recent EELS, XPS, and ISS study of Hoflund28 showing
that step 1, representing the direct reduction of SnO2 to SnO
without passage Via a transitional oxide phase, occurs for bulk
SnO2 samples following 473 K reduction in 40 Torr H2. The
subsequent high-temperature reduction of SnO to Sn0 incorpo-
rated within ordered Pd/Sn surface alloys, step 2 of the scheme,
is also observed during in Vacuo annealing of ultrathin,
stoichiometric SnO films supported on Pd{111}. This behavior
contrasts with that of bulk SnO, for which annealing treaments
between 473-873 K actually reoxidise the selvedge to SnO2
Via a transitional Sn2O3 phase.26
The short Sn-Pd distance found for the “473 K reduced”
sample (2.65 Å Versus >2.75 Å for a bulk substitutional Pd/Sn
alloy12) is consistent with stage 2 of the scheme shown in Figure
12 with interfacial Sn atoms residing between the Pd substrate
and an oxygen top layer. (Goursot et al.29 recently calculated
a similar metal-metal separation for Sn on a Ru9{111} cluster.)
Reduction in Sn radius due to Sn-O charge polarization (e.g.,
in SnO and SnO2: 0.93 and 0.73 Å, respectively27) would also
contribute to the contracted Pd-Sn separation.
It might be expected that Sn surface f bulk diffusion would
be thermodynamically disfavored for small Pd particles where
the low Sn:Pd atom ratio prohibits intermetallic compound
formation,12 the proposed driving force for Sn bulk diffusion
into Pt particles.30 This is what we find. Thus our EXAFS
data suggest that reduction of the surface Sn oxide phase is
complete at temperatures >673 K and that the total Sn nearest-
neighbor coordination remains low, even after reduction at the
highest temperature. This points to Sn atoms predominantly
occupying surface sites, as anticipated from the very large
difference in Pd and Sn surface free energies (0.71 and 2.1 J
m-2, respectively27). Correspondingly, the XRD results confirm
the absence of bulk Pd/Sn alloy phases.
Again, the behavior of the dispersed Pd/Sn system accords
well with that found for macroscopic model planar catalysts.
Heating of Sn monolayers deposited on Pd{111} under vacuum
conditions leads to surface alloy formation.7 A x3 (Pd2Sn)
surface structure is formed at ∼773-873 K, its thermal stability
increasing with initial Sn overlayer thickness. High temperature
reduction of supported Pd-Sn catalyst particles should therefore
induce Sn-Pd surface alloy formation and our catalytic
measurements (see below) do in fact indicate the continued
presence Sn atoms at the surface, even after reduction at 873
K.
Ethyne Coupling over PdxSn Catalysts. Both the pure Pd
and bimetallic PdxSn catalysts were active for ethyne coupling,
yielding only C4 and C6 products. Furthermore, the yields of
the C4 products and benzene from each sample were closely
correlated under all pretreatment or reaction conditions (Figure
8a-b) and C3 and C5 products were never observed. We have
already demonstrated that formation of a C4H4 intermediate is
the key reaction initiating step in the coupling chemistry of
ethyne on single-crystal Pd and Pd alloy surfaces under UHV
conditions.2,32 The present observations emphatically confirm
the view that the same C4H4 intermediate plays a crucial role
in catalysis by practical materials operated at atmospheric
pressure.
The absence of ethene, even at high H2:C2H2 ratios, at first
sight suprising, is not without precedent. Szanyi and Paffett33
report an almost identical range of products resulting from
ethyne coupling over extensively carbided Pt{111} and {x3
× x3}R30°-Sn/Pt{111} single-crystal surface alloys under
similar reaction conditions (400-550 K, 20-100 Torr, H2:C2H2
) 2.5-50). Furthermore, alkyne hydrogenation over supported
Pd is a strong function of particle size,34 e.g., rates of
vinylacetylene hydrogenation over 2 wt % Pd/SiO2 decrease
by an order of magnitude for Pd/SiO2 catalysts with decreasing
particle size from 40 to 10 Å.35 Note that under our conditions
(Treax g3 73 K and/or H2 partial pressures e1 bar), with small
small (e20 Å) Pd particles, complications due to phase
transformation to âPd-H may be discounted.
Influence of Reactant Mix. The activity dependence of all
three catalysts on the H2:C2H2 ratio (Figure 7) is readily
understood in terms of site-blocking effects. At low H2 partial
pressures, extensive decomposition of both ethyne and its
coupling products occurs over both Pd36 single-crystal surfaces
and Pd/SiO2 catalysts prepared by conventional wet-impregna-
tion.37 Carbiding of the active surface results in rapid catalyst
deactivation on a time scale of minutes. Increasing hydrogen
concentration inhibits dissociative chemisorption,1,38 and it
facilitates clean-off reactions. Beyond an optimum reactant
ratio, competitive chemisorption of ethyne results in a rate
decrease as would be expected under Langmuir-Hinshelwood
kinetics. The dependence of reaction orders of individual
products on hydrogen partial pressure (determined at <10%
ethyne conversion) is in line with this. Closely similar kinetic
behavior is found for ethyne coupling over the x3 Sn/Pd{111},-
Pt{111} alloy surfaces,33,38 confirming the close connection
between the model systems and supported PdxSn catalysts.
The Pd XANES results (Figure 6a) confirm that reduction
does not significantly perturb the average local electronic and
structural environment of Pd atoms within the Pd2Sn catalyst.
In contrast, the Sn XANES (Figure 6b) shows a progressive
transformation from an oxidic (electron deficient) to metallic
The samples used for testing contained, within the experi-
mental error, identical amounts of Pd. Activity differences
between the three catalysts cannot simply be ascribed to

2804 J. Phys. Chem. B, Vol. 101, No. 15, 1997
Lee et al.
differences in dispersion because the XRD results indicate
similar particle sizes in all three cases. Differences in maximum
activities (Figure 7) are attributed to the differing amounts of
Sn at the surface. The XAFS measurements show that Sn atoms
are predominantly in metallic form after reduction at >473 K,
the temperature regime in which stable Pd/Sn surface alloys
are formed on Pd{111}.7 The most Sn rich catalyst exhibits
the highest activity because Sn surface atoms act to inhibit the
decomposition of reactants and products as a result of C-C
and C-H cleavage,4,38-40 they also facilitate the desorption of
products by reducing chemisorption bond strengths.7,40,41 The
net result of these effects is to enhance the total activity of
Pd3Sn2 by ∼50% with respect to the pure Pd.
Sn is present on the surface of the Pd particles where it promotes
ethyne conversion, as discussed above. At higher reduction
temperatures (>673 K) the remaining surface tin oxide under-
goes progressive decomposition, liberating additional Sn which
spreads over the Pd, increasing selectivity to benzene (Figure
10) and decreasing overall activity (Figure 9). The small
PdxSn cluster size inhibits complete dissolution of Sn in the Pd
core (which does occur above 523 K for Sn overlayers on bulk
Pd7) even at reduction temperatures of 1100 K.
Influence of Reaction Temperature. The variation in C6
selectivities with reaction temperatures between 300-500 K
(Figure 11) cannot be attributed to changes in surface composi-
tion of the Pd2Sn catalyst. An (admittedly oversimplified)
rationalization in terms of the surface lifetime of reactively
formed benzene may be offered as follows. It may be
rationalized in terms of reduced surface lifetime of the reactively
formed benzene with increasing temperature (desorption is the
rate limiting step for appearance of benzene in the gas phase).
This should reduce the probability of further conversion to
hydrogenation and hydrogenolysis products, in keeping with
observation. Interestingly, the benzene selectivity increases
between 300 and ∼450 K, which is just the temperature interval
over which desorption of benzene form Pd{111} occurs.42 For
T > 450 K, the surface lifetime of reactively formed benzene
is very short and the yield of this product reaches a limiting
value.
Under all conditions examined, the product distributions from
the PdxSn catalysts were heavily weighted in favor of C6 over
C4 species; i.e., addition of a third ethyne molecule to the C4H4
intermediate is strongly favored over its hydrogenation, even
in hydrogen-rich atmospheres. This stands in interesting contrast
to measurements on Sn/Pt{111} single-crystal33 which showed
that C4 production always exceeded C6 production and that the
C4:C6 relative yield increased slightly with H2 pressure. It is
possible that this reflects the reduced hydrogenation activity of
Pd relative to Pt.
The benzene and n-hexane yields for all samples display a
strong inverse correlation, respectively decreasing and increasing
with rising hydrogen partial pressure (Figure 8). This suggests
that these two products derive from a common reaction
intermediate and are related sequentially, as follows:
Conclusions
Conclusions are as follows: 1. Reduction of Pd/Sn catalyst
precursors prepared by controlled surface organometallic reac-
tion generates highly dispersed (<30 Å), thermally stable
catalysts for ethyne coupling. These materials are resistant to
particle growth but show important changes in surface composi-
tion as the reduction temperature is increased from 473 to 873
K.
2. Increasingly vigorous reduction leads to progressive
destruction of the SnO2-like 3D structures that surround the Pd
core. Surface alloy formation ensues, followed eventually by
more extensive Pd/Sn intermixing. This results in destruction
of the critical ensembles necessary for structurally demanding
hydrogenolysis reactions; significant modification of surface
electronic properties also occurs.
3. The above changes are accompanied by pronounced
changes in catalytic activity and very large changes in selectivity.
Only C6 and C4 products were ever observed with benzene and
n-hexane being the major species formed. Increased levels of
surface Sn greatly increased benzene selectivity and inhibited
hydrogenolysis. In the case of “Pd2Sn”, the product distribution
could be varied from ∼100% benzene to ∼100% n-hexane. With
“Pd3Sn2”, benzene formation could never be totally quenched
because of the limited bulk solubility of Sn in the very small
particles.
4. The thermal properties and catalytic chemistry of these
highly dispersed materials are in good accord with those reported
for Pd{111}/Sn model planar catalysts, suggesting that the latter
do provide a useful means of isolating and mimicking important
aspects of the behavior of practical systems.
The observed formation of cyclohexane and cyclohexene
(intermediates in benzene hydrogenolysis to n-hexane33) under
all reaction conditions is in accord with the above scheme.
In the absence of hydrogen, chemisorbed ethyne reacts to
form benzene almost exclusively. Increasing hydrogen partial
pressures progressively favors hydrogenation and the increasing
production of cyclohexane and its hydrogenolysis product
n-hexane at the expense of benzene formation. With the pure
Pd and Pd2Sn catalysts, almost 100% conversion to n-hexane
occurs for H2:C2H2 ratios >1.75 (Figure 7). In the case of Pd3-
Sn2, benzene production is not fully quenched, even at high
hydrogen partial pressures, reflecting the effect of increased
surface Sn concentration in limiting the number of Pd ensembles
effective for hydrogenation/hydrogenolysis. This is consistent
with a study by Xu and Koel40 who showed that surface Sn
atoms destabilise cyclohexene chemisorption on Pt{111}, induc-
ing a transition from a di σ- to H-bonded adsorption geometry,
resulting in increased benzene production.
Influence of Reduction Temperature. Ethyne coupling on
Pd and Pd-based surfaces is a very structure-sensitive reaction:
{111}-oriented facets are critically important.5 It is apparent
that ethyne conversion over the Pd, Pd2Sn, and Pd3Sn2 samples
is relatively insensitive to reduction temperature, even up to
∼1100 K (Figure 9). This indicates the particles are rather stable
with respect to sintering, a conclusion that is supported by the
corresponding XRD and Pd EXAFS. The activities of both
bimetallic catalysts do however show some decrease with
increasing reduction temperaturesthe magnitude of this decrease
increasing with Sn content. This is understandable as follows.
EXAFS shows that after low temperature reduction, much
of the Sn is present within 3D oxide structures; some metallic
Acknowledgment. Financial support from the UK Engineer-
ing and Physical Sciences Research Council under Grant GR/
K45562 is gratefully acknowledged. We thank Johnson Matthey
PLC for a loan of precious metals.
References and Notes
(1) Ormerod, R. M.; Lambert, R. M. J. Chem. Soc., Chem. Comm.
1990, 1421. Yoshinabu, J.; Seikitani, T.; Onchi, M.; Nishijima, M. J. Phys.
Chem. 1990, 94, 4269.

Catalytic Coupling of Ethyne
J. Phys. Chem. B, Vol. 101, No. 15, 1997 2805
(2) Baddeley, C. J.; Hardacre, C.; Tikhov, M. S.; Lomas, J. R.; Lambert,
R. M. J. Phys. Chem. 1996, 100, 2189.
(3) Srinavasan, R.; Davis, B. H. Platinum Met. ReV. 1992, 36, 151.
(4) Xu, C.; Tsai, Y. L.; Koel, B. E. J. Phys.Chem. 1994, 98, 585.
(5) Baddeley, C. J.; Ormerod, R. M.; Stephenson, A. W.; Lambert, R.
M. J. Phys. Chem. 1995, 99, 5146.
(6) Lee, A. F.; Baddeley, C. J.; Hardacre, C.; Ormerod, R. M.; Lambert,
R. M.; Schmid, G.; West, H. J. Phys. Chem. 1995, 99, 6096.
(7) Lee, A. F.; Baddeley, C. J.; Tikhov, M. S.; Lambert, R. M. Surf.
Sci. In press.
(8) Didillon, B.; El Mansour, A.; Candy, J. P.; Bournonville, J. P.;
Basset, J. M. In Heterogeneous Catalysis and Fine Chemicals II; Guisnet
M., et al., Eds.; Elsevier: New York, 1991; 137.
(9) Didillon, B.; Houtman, C.; Shay, T.; Candy, J. P.; Basset, J. M. J.
Am. Chem. Soc. 1993, 115, 9380.
(10) Moggridge, G. D; Rayment, T.; Lambert, R. M. J. Catal. 1992,
134, 242.
(11) Horton J. H.; G. D. Moggridge; Ormerod R. M.; Kolobov, A. V.;
Lambert, R. M. Thin Solid Films 1994, 237, 134.
(12) Hansen, M. In Constitution of Binary Alloys; McGraw-Hill: New
York, 1958; 1126.
(13) Anderson, J. N.; Hennig, D.; Lundgren, E.; Methfessel, M.; Nyholm,
R.; Scheffler, M. Phys. ReV. B 1994, 50, 17525.
(14) Cheung, T. T. P. Chem. Phys. Lett. 1984, 110, 219.
(15) Lesage, P.; Clause, O.; Moral, P.; Didillon, B.; Candy, J. P.; Basset,
J. M. J. Catal. 1995, 155, 238.
(16) Candy, J. P.; Ferretti, O. A.; El Mansour, A.; Mabilon, G.;
Bournonville, J. P.; Basset, J. M.; Martino, G. J. Catal. 1988, 112, 201.
(17) Sinfelt, J. H.; Via, G. H.; Lytle, F. W. J. Chem. Phys. 1980, 72,
4832.
(18) Bianconi, A. In X-Ray Absorption; Koningsberger, D. C., Prins,
R., Eds.; Wiley: New York, 1988; 573.
(21) Clausen, B. S.; Grabaek, L.; Topsoe, H.; Hansen, L. B.; Stoltze,
P.; Norskov, J. K.; Nielsen, O.-H. J. Catal. 1993, 141, 368.
(22) Cheung, T. T. P. Surf. Sci. 1984, 140, 151.
(23) Nosova, L. V.; Stenin, M. V.; Nogin, Y. N.; Ryndin, Y. A. Appl.
Surf. Sci. 1992, 55, 43.
(24) Lee, A. F.; Lambert, R. M. Manuscript in preparation.
(25) Coq, B.; Dutartre, R.; Bond, G. C.; Slaa, J. C.; Galvango, S.;
Mercandante, L.; Garcia Ruiz, J.; Sanchez Sierra, M. C. In Europacat II
Conference Abstracts; Maastricht 1995, 431.
(26) Moreno, M. S.; Mercader, R. C.; Bilboni, A. G. J. Phys. C 1992,
4, 351.
(27) Weast, R. C., Ed. CRC Handbook of Chemistry and Physics, 69th
ed.; The Chemical Rubber Company: Cleveland, 1988.
(28) Hoflund, G. B. Chem. Mater. 1994, 6, 562.
(29) Goursot, A.; Pedocchi, L.; Coq, B. J. Phys. Chem. 1995, 99, 12718.
(30) Balakrishnan, K.; Schwank, J. J. Catal. 1991, 127, 287.
(31) Burch, R. J. Catal. 1981, 71, 348.
(32) Patterson, C. H.; Lambert, R. M. J. Am. Chem. Soc. 1988, 110,
6871. Patterson, C. H.; Lambert, R. M. J. Phys. Chem. 1988, 92, 1266.
(33) Szanyi, J.; Paffett, M. T. J. Am. Chem. Soc. 1995, 117, 1034.
(34) Karpkinski, Z. AdV. Catal. 1990, 37, 45.
(35) Ryndin, Y. A.; Nosova, L. V.; Boronin, A. L.; Chuvilin, A. L. Appl.
Catal. 1988, 42, 131.
(36) Rucker, T. G.; Logan, M. A.; Gentle, T. M.; Mutterties, E. L.;
Somorjai, G. A. J. Phys. Chem. 1986, 90, 2703.
(37) Ormerod, R. M.; Lambert, R. M. J. Chem. Soc., Chem. Comm.
1990, 1492.
(38) Lee, A. F.; Lambert, R. M. Manuscript in preparation.
(39) Xu, C.; Peck, J. W.; Koel, B. E. J. Am. Chem. Soc. 1993, 115,
751.
(40) Xu, C.; Koel, B. E. Surf. Sci. 1994, 304, 249.
(41) Xu, C.; Koel, B. E. Surf. Sci. 1994, 310, 198.
(42) Tysoe, W. T.; Nyberg, G. L.; Lambert, R. M. Surf. Sci. 1983, 135,
128.
(19) Sham, T. K. Phys. ReV. B 1985, 31, 1888.
(20) Kostroun, V. O.; Fairchild, R. W.; Kukkonen, C. A.; Wilkins, J.
W. Phys. ReV. B 1976, 13, 3268.