Coordination of palladium on carbon fibrils as determined by XAFS spectroscopy

Article abstract of DOI:10.1039/a704989g

To investigate the anchoring of precious metals onto carbon fibrils Pd/C-fibril samples were investigated in both the precursor and the reduced state by X-ray absorption fine structure (XAFS) spectroscopy. Analysis of the XAFS data of the freshly prepared catalyst showed a palladium-tetraamine complex in interaction with carbon. The proposed model involves [Pd(NH₃)₄]²⁺ coordinated to the carbon fibril surface, most probably stabilised by the carboxylic groups (through O - H bridging and charge balance) and the π-electron system of the support. After in situ reduction, analysis pointed to small palladium particles of ca. 15 A present on the fibrils with a significant Pd-C interaction.

Full text of DOI:10.1039/a704989g

View Article Online / Journal Homepage / Table of Contents for this issue
Coordination of palladium on carbon Ðbrils as determined by XAFS
spectroscopy
Barbara L. Mojet,* Marco S. Hoogenraad, Adrianus J. van Dillen, John W. Geus and
Diek C. Koningsberger
Debye Institute, Department of Inorganic Chemistry, Utrecht University, P.O. Box 80083,
3502 T B Utrecht, T he Netherlands
To investigate the anchoring of precious metals onto carbon Ðbrils Pd/C-Ðbril samples were investigated in both the precursor
and the reduced state by X-ray absorption Ðne structure (XAFS) spectroscopy. Analysis of the XAFS data of the freshly prepared
catalyst showed a palladiumÈtetraamine complex in interaction with carbon. The proposed model involves [Pd(NH ) ]2` coor-
3 4
dinated to the carbon Ðbril surface, most probably stabilised by the carboxylic groups (through OwH bridging and charge
balance) and the p-electron system of the support. After in situ reduction, analysis pointed to small palladium particles of ca. 15 Ó
present on the Ðbrils with a signiÐcant PdÈC interaction.
Introduction
Experimental
Catalyst preparation
Activated carbon is widely used as an adsorbent and as a
support material for catalytically active components in liquid-
phase processes because of its high resistance towards strong
acidic and alkaline solutions and its high speciÐc surface area.
Less favourable are the presence of micropores, which can
lead to severe transport limitations, and the liability to attri-
tion; the resulting Ðnes are difficult to separate from the
liquid. Carbon Ðbrils, on the other hand, exhibit a high
mechanical strength, contain no micropores, and display a
considerable speciÐc surface area and a high pore volume, as
determined with nitrogen physisorption.1,2 It has been shown
that precious metals supported on carbon Ðbrils provide
highly active and selective catalysts in hydrogenation reac-
tions.3 Anchoring on the carbon surface of precious metal pre-
cursor compounds as well as precious metal particles obtained
by subsequent reduction is difficult owing to the chemical
inertness of the surface of carbon Ðbrils and activated carbon.
Drastic pretreatment procedures are therefore required.
Studies on amorphous carbon have accordingly revealed that
catalyst preparation method and carbon surface chemistry
determine the Ðnal dispersion of supported precious
metals.4h7 Platinum supported on amorphous carbon was
investigated with TEM, XAFS and BET.4 It was concluded
that negatively charged precursor ions do not attach well to
an oxidised, negatively charged carbon support surface;
however, positively charged complexions can be attached
easily.
The observation that the degree of oxidation of the carbon
support determines the Ðnal metal loading3 suggests that the
number of carboxylic groups is limiting the amount of metal
ions that can be attached to the Ðbrils. Information about the
interaction between precious metal compounds and carbon
Ðbrils is therefore very useful for the development of immobil-
ised homogeneous catalysts on carbon.
To investigate the nature of anchoring of palladium precur-
sor ions and palladium metal particles onto carbon Ðbrils,
XAFS experiments were performed on the freshly prepared as
well as the reduced catalyst. The results indicate a signiÐcant
metalÈÐbril interaction, which accounts for the stabilisation of
the precursor complex and the small palladium metal particles
obtained after reduction.
Carbon Ðbrils with the graphite layers parallel to the Ðbre axis
were grown in a stream of CO and H out of a 20 wt.%
2
Fe/Al O catalyst at 570 ¡C.2 Details about the graphitic mor-
2
3
phology of the Ðbrils can be found in the literature.2,3,8 Reac-
tive species were applied on the Ðbrils, by treatment with
boiling concentrated HNO (65%). The treated Ðbrils were Ðl-
tered, washed and dried. Subsequently, the Ðbrils were (under
nitrogen) ion-exchanged with [Pd(NH ) ](NO ) and dried at
80 ¡C. This procedure resulted in a palladium loading of 3
wt.%.
3
3 4
3 2
XAFS experiments
The X-ray absorption Ðne structure (XAFS) spectra were
recorded for four samples: Pd-foil, [Pd(NH ) ](NO ) , ion-
3 4
3 2
exchanged carbon Ðbrils before reduction Mdenoted by
[Pd(NH ) ]/C-ÐbrilN, and ion-exchanged carbon Ðbrils
3 4
after in situ reduction (denoted by Pd/C-Ðbril). The XAFS
experiments were done at the Pd K-edge (24 350 eV) on
Wiggler station 9.2 at the SRS Daresbury (UK) using an
Si(220) double crystal monochromator MPd-foil and
[Pd(NH ) ](NO ) N or an Si(220) channel cut monochro-
3 4
3 2
mator M[Pd(NH ) ]/C-Ðbril and Pd/C-ÐbrilN. The double
3 4
crystal monochromator was detuned to 80% of the maximum
intensity at the Pd K-edge. The measurements were performed
in transmission mode using ion-chambers Ðlled with krypton
to have an absorbance (lx) of 20% in the Ðrst and a lx of
80% in the second chamber. All XAFS data were taken at
liquid-nitrogen temperature to optimise the signal-to-noise
ratio of the spectra.
Sample treatment
All gases were puriÐed and dried before use. The Pd-foil (20
lm) was mounted into a sample holder and placed in an in
situ cell.9 [Pd(NH ) ](NO ) was thoroughly mixed with BN
3 4
3 2
and an amount with a calculated *lx \ 1 at the Pd K-edge
(total absorbance \2.5) was pressed in a self-supporting wafer
and placed in an in situ cell. Both Pd-foil and
[Pd(NH ) ](NO ) were Ñushed with He at room temperature
3 4
3 2
for 15 min and cooled down to liquid-nitrogen temperature.
J. Chem. Soc., Faraday T rans., 1997, 93(24), 4371È4375 4371

View Article Online
Table 1 Parameters of the experimental and theoretical references
k-range
/Ó~1
R-range
/Ó
backscatterer
reference source
N
R/Ó
Pd
N
C
Pd-foil (20 lm)
12 2.75 2.6È21.1 1.99È2.92
[Pd(NH ) ](NO )
4
6
2.05 2.8È1.81 0.19È2.54
3 4
3 2
FEFF-3.1a
4.00
a S 2 \ 0.9, *p2 \ 0.000 Ó2.
0
[Pd(NH ) ]/C-Ðbril was pressed into a self-supporting
3 4
wafer (with a calculated absorbance of 2.5, *lx \ 0.4) and
placed in an in situ cell. The wafer was Ñushed with He
(puriÐed and dried) at room temperature for 15 min and
cooled to liquid nitrogen temperature, where XAFS spectra
were collected. The sample was next allowed to warm up to
room temperature, and subsequently reduced in situ in a Ñow
of 10% H in He at 200 ¡C (heating rate 2 ¡C min~1, kept at
2
200 ¡C for 1 h) and cooled under He to avoid the formation of
palladium hydrides at low temperatures.
XAFS data analysis
Fig. 1 (a) Raw EXAFS data of [Pd(NH ) ]/C-Ðbril. (b) Fourier
3 4
transform (k1, *k: 3.0È14.5 Ó~1) of EXAFS data of [Pd(NH ) ]/C-
3 4
The pre-edge was approximated by a modiÐed Victoreen
curve10 and the background was subtracted using cubic spline
routines.11 Spectra were normalised by dividing the absorp-
tion intensity by the height of the absorption edge at 50 eV
above the edge. The Ðnal EXAFS function was obtained by
averaging four individual background-subtracted and normal-
ised scans. The XAFS data were analysed in R-space12 by
multiple shell Ðtting with the data analysis program XDAP.13
Data for phase shifts and backscattering amplitudes were
obtained from reference compounds and FEFF 3.1,14 the
parameters are given in Table 1. Variances for imaginary and
absolute parts of the model spectra are calculated by:
Ðbril (solid line) and [Pd(NH ) ](NO ) (dashed line).
3 4
3 2
Ðrst shell was investigated to check for the possible existence
of a PdwO interaction in the Ðrst coordination shell from the
carboxylic groups of the carbon Ðbril surface. However, these
Ðts did not display the low variances found with only nitrogen
as a scatterer. It has been reported previously that it is pos-
Table 2 EXAFS analysis parameters and estimated errors for
[Pd(NH ) ]/C-Ðbril, model Ia
3 4
P
Rmax[kn(FT
(R) [ FT (R))]2
backscatterer
N
distance/Ó
*p2/10~3 Ó2
*E /eV
model
exp
0
Rmin
variance \ 100%
P
Rmax[knFT (R)2]
N
C
4.4 ^ 0.4 2.04 ^ 0.02 [ 0.12 ^ 0.01 2.2 ^ 0.2
11.5 ^ 1.0 3.95 ^ 0.04 17 ^ 1 7.7 ^ 0.7
exp
Rmin
a R-space Ðt, k1 weighting, *k: 3.00È14.50 Ó~1, *r: 0.50È5.00 Ó.
The errors in the calculated parameters were estimated to be
10% in coordination number N, 1% in distance R, 10% in
DebyeÈWaller factor *p2 and 10% in the inner potential shift
Variance imaginary part: 0.36%; variance magnitude: 0.14%.
*E .
0
Results
[Pd(NH ) ]/C-Ðbril
3 4
The XAFS data of [Pd(NH ) ]/C-Ðbril are shown in Fig. 1(a).
3 4
The data exhibit an excellent signal-to-noise ratio, which
makes data analysis possible up to 14.5 Ó~1. Examination of
the Fourier transforms of [Pd(NH ) ]/C-Ðbril and
3 4
[Pd(NH ) ](NO ) [Fig. 1(b), *k: 3.0È14.5 Ó~1, k~1-
3 4
3 2
weighted] reveals di†erences for both magnitude and imagin-
ary parts of the spectra. In comparison with the spectrum of
[Pd(NH ) ](NO ) , the Fourier transform of [Pd(NH ) ]/C-
Ðbril shows a somewhat higher intensity and a small shift in
imaginary part in the Ðrst shell at 1.5 Ó, and a signiÐcantly
enhanced intensity together with a di†erent imaginary part
between 3 and 4 Ó. Furthermore, the higher shell contribution
present in the precursor complex (at 5.4 Ó) is absent in
3 4
3 2
3 4
[Pd(NH ) ]/C-Ðbril. Both observations indicate a well dis-
3 4
persed precursor complex in [Pd(NH ) ]/C-Ðbril with most
3 4
probably additional scatterers present, arising from the
carbon support surface.
Fig. 2 Comparison of EXAFS data of [Pd(NH ) ]/C-Ðbril and cal-
Analysis of the Ðrst shell data in R-space [Table 2, Fig. 2(a)]
shows a PdwN contribution at 2.04 Ó, which is in accordance
with coordination of palladium by four nitrogen atoms as in
3 4
culated Ðrst shell (PdwN). (a) Fourier transform (k1-weighted, *k:
3.0È14.5 Ó~1) of EXAFS data (solid line) and calculated PdwN con-
tribution (dashed line). (b) EXAFS data minus calculated PdwN con-
tribution (solid line) and standard deviation (dashed line).
[Pd(NH ) ]2`. Inclusion of one or more oxygen atoms in the
3 4
4372
J. Chem. Soc., Faraday T rans., 1997, V ol. 93

View Article Online
respectively. The total variances for both models are below
1% as can be seen in Tables 2 and 3. F-tests16,17 of the Ðts
(including the PdwN contribution) indicate that both models
are equally signiÐcant and 90% signiÐcant as compared to the
model with only PdwN scattering. The Fourier transforms of
the data of [Pd(NH ) ]/C-Ðbril and model II are represented
3 4
in Fig. 3(c). Both magnitude and imaginary parts of experi-
mental data and calculated spectrum are in good agreement.
Pd/C-Ðbril
Fig. 4(a) shows the Fourier transforms of the XAFS data (*k:
3.2È14.5 Ó~1, k1-weighted) for [Pd(NH ) ]/C-Ðbril and Pd/C-
3 4
Ðbril. It is clear that the sample has changed due to reduction
in H . The contribution at low distance present in the Fourier
2
transform of [Pd(NH ) ]/C-Ðbril has vanished and a new
3 4
peak has appeared around 2.5 Ó in the spectrum of Pd/C-
Ðbril. Fig. 4(b) presents the Fourier transforms (*k: 3.2È15.0
Ó~1, k1-weighted, scaled to 1 on the Ðrst PdwPd peak) of
Pd/C-Ðbril and Pd-foil. Hardly any higher PdwPd shells are
visible in the spectra of the catalyst. Di†erences in imaginary
part and magnitude between 1 and 2 Ó are clearly indicative
for additional contributions. Analysis in R-space revealed
besides a PdwPd contribution at 2.76 Ó also PdwC scat-
tering at 2.60 Ó. Model parameters are presented in Table 4.
Comparison of the model spectrum with the experimental
data [Fig. 5(a)] and the separate contributions with their dif-
ference Ðles [PdwPd: Fig. 5(b), PdwC: Fig. 5(c)] exhibit
good agreement for model and experiment in the range 0.5È
3.10 Ó. At higher distance (R [ 3.10 Ó) additional contribu-
tions are left that are due to higher PdwPd and PdwC shells.
Fig. 3 Fourier transform (k~1-weighted, *k: 3.0È14.5 Ó~1) of
Discussion
EXAFS data of [Pd(NH ) ]/C-Ðbril and calculated spectra. (a) Model
3 4
I: FT of EXAFS data minus calculated PdwN contribution (solid
[Pd(NH ) ]/C-Ðbril
line) and calculated PdwC contribution (dashed line). (b) Model II:
FT of EXAFS data minus calculated PdwN contribution (solid line)
and calculated PdwC contributions (dashed line). (c) FT of EXAFS
3 4
For Pt/C catalysts it has been reported that a signiÐcant inter-
action exists between surface acidic groups and the precursor
complex.4 The authors used XAFS spectroscopy and drew
their conclusions by comparing Fourier transforms, without
analysing the XAFS data. Since DebyeÈWaller factors, di†er-
ent types of backscattering neighbours, and coordination
numbers determine the shape and intensity of the Fourier
transforms, it is difficult to draw conclusions about the coordi-
nation sphere of the absorber atoms only from comparison.
Analysis of the data analysis is required. In our XAFS data
data of [Pd(NH ) ]/C-Ðbril (solid line) and calculated spectrum with
3 4
parameters given in Table 3 (dashed line).
sible indeed to identify N, O and C as di†erent backscatterers,
despite the fact that they are neighbouring elements in the
Periodic Table.15
The di†erence spectrum of the experimental data of
[Pd(NH ) ]/C-Ðbril and the calculated PdwN contribution
3 4
analysis of [Pd(NH ) ]/C-Ðbril we found no signiÐcant
oxygen contribution in the Ðrst coordination shell of palla-
[Fig. 2(b)] shows that there are signiÐcant contributions left
3 4
above the noise level. This remaining scattering could be suc-
cessfully Ðtted with PdwC scatterer pairs, but determination
of coordination numbers turned out to be not straightforward.
Fig. 3(a) and (b) show two possibilities (model I and II) for
Ðtting the signal in the Fourier transform after subtraction of
the PdwN contribution from the experimental data. The Ðt
parameters were optimised in a two-shell (model I, 8 free
parameters) and three-shell Ðt (model II, 12 free parameters)
and are given in Tables 2 and 3. Model I consists of about
twelve carbon atoms at 3.95 Ó and model II is the result after
reÐnement with two carbon contributions at 3.57 and 4.30 Ó,
dium. Furthermore, an additional scattering between 3 and 4
Ó was observed, which was not reported before by others.
This additional scattering could successfully be Ðtted with
PdwC contributions, although precise coordination numbers
were difficult to determine. The uncertainty of the number of
coordination shells and neighbours can be attributed to the
fact that the carbon support exhibits a very symmetric long
range order with multiple carbonÈcarbon distances. Fig. 6(a)
shows the distances in graphite as obtained from literature,18
that are helpful in the determination of the coordination of
the precursor complex to the Ðbril carbon atoms. Model I
(Table 2) corresponds to a [Pd(NH ) ]2` complex situated at
3 4
3.7 Ó above the support with scattering from twelve carbon
Table 3 EXAFS analysis parameters and estimated errors for
[Pd(NH ) ]/C-Ðbril, model II
3 4
Table 4 EXAFS analysis parameters and estimated errors for Pd/C-
back-
scatterer
Ðbril
N
distance/Ó
*p2/10~3 Ó2
*E /eV
0
backscatterer
N
distance/Ó
*p2/10~3 Ó2
*E /eV
0
N
C
C
4.3 ^ 0.4 2.04 ^ 0.02
[ 0.34 ^ 0.03
23 ^ 2
1.5 ^ 0.1
[ 7.2 ^ 0.7
[ 8.5 ^ 0.8
5.9 ^ 0.6
6.3 ^ 0.6
3.57 ^ 0.03
4.30 ^ 0.04
Pd
C
7.9 ^ 0.8 2.76 ^ 0.02
7.3 ^ 0.7 2.60 ^ 0.02
3.6 ^ 0.3
10 ^ 1
1.5 ^ 0.1
20.5 ^ 2
31 ^ 3
R-space Ðt, k1 weighting, *k: 3.00È14.50 Ó~1, *r: 0.50È5.00 Ó.
R-space Ðt, k1 weighting, *k: 3.20È15.00 Ó~1, *r: 0.50È3.10 Ó.
Variance imaginary part: 0.35%; variance magnitude: 0.13%.
Variance imaginary part: 0.16%; variance magnitude: 0.08%.
J. Chem. Soc., Faraday T rans., 1997, V ol. 93
4373

View Article Online
Fig. 4 (a) Fourier transform (k~1, *k: 3.2È14.5 Ó~1) of EXAFS data
of Pd/C-Ðbril (solid line) and [Pd(NH ) ]/C-Ðbril (dashed line). (b)
3 4
Fourier transform (k1, *k: 3.2È15.0 Ó~1, scaled to 1 on main peak) of
EXAFS data of Pd/C-Ðbril (solid line) and Pd-foil (dashed line).
Fig. 6 (a) Interatomic distances in graphite as taken from ref. 18. (b)
First and second shell PdwC distances for a Pd atom located at
height
H
above centre of graphite-ring. R1 \ [H2 ] (2.84/
2)2]1@2 \ 3.57 Ó for H \ 3.27 Ó, R2 \ [H2 ] (2.84)2]1@2 \ 4.33 Ó for
atoms at an average distance of 3.95 Ó. Since the distance
between the graphite layers is 3.35 Ó,18 which is about the
same in Ðbrils, it is unlikely that the precursor complex is
squeezed between two graphite layers, which could have
accounted for a carbon coordination number of twelve. The
high coordination number might indicate the presence of
several carbon interactions relatively close to each other.
H \ 3.27 Ó.
Model II (Table 3) consists of a [Pd(NH ) ]2` complex at
3 4
3.27 Ó above the Ðbril, interacting with the Ðrst and second
six-ring of the carbon system. The PdwC distances of 3.57
and 4.30 Ó are in very good agreement with the distances in
graphite [Fig. 6(b)]. In addition, location of the palladium
complex at 3.27 Ó above the Ðbril is more likely to give rise to
signiÐcant support scattering than when it would be at a dis-
tance of 3.7 Ó which is found in model I. The DebyeÈWaller
factors for the carbon scattering are quite high, but this can be
explained by the fact that they are determined relative to a
theoretical reference in which *p2 was set to be zero. Further-
more, the precursor complex is located at a relatively large
distance from the Ðbril, which also introduces more disorder
as a strong chemical bond is missing. The second PdwC shell
exhibits larger disorder than the Ðrst PdwC shell, which is
consistent with higher disorder at larger coordination dis-
tances. It was found that models I and II, including PdwC
scattering, result in mathematically good models for the
experimental data. Although XAFS spectroscopy alone
cannot give unique coordination numbers and distances for
the PdwC interactions in [Pd(NH ) ]/C-Ðbril, it shows
3 4
unambiguously a supportÈcarbon backscattering contribu-
tion. Combining mathematics and chemistry suggests model
II as the best representative of the geometrical structure of
[Pd(NH ) ]/C-Ðbril.
3 4
In organometallic complexes that contain benzene as a
ligand, the metalÈcarbon distance is around 2.2 Ó.19 Since the
observed palladiumÈcarbon distance in this study is much
larger, we conclude that no chemical bond exists between pal-
ladium and the Ðbril, although the p-system is likely to have a
stabilising e†ect on the complex.
Although a PdwO contribution could not be included with
statistical signiÐcance, oxygen atoms might be present at a
longer distance. The immense sea of carbon contributions to
the XAFS data might bring about that a few oxygen atoms in
a higher coordination sphere of palladium cannot be appar-
ent. As the amount of surface carboxylic groups was found to
be important in the preparation of the catalyst,3,4 but no
oxygen could be detected at a short distance from palladium,
we suggest a charge stabilising e†ect of the acidic groups
Fig. 5 Fourier transform (k1 weighted, *k: 3.2È15.0 Ó~1) of EXAFS
data of Pd/C-Ðbril and calculated spectra with parameters given in
Table 4. (a) FT of EXAFS data (solid line) and total calculated spec-
trum (dashed line). (b) FT of EXAFS data minus calculated PdwC
contribution (solid line) and calculated PdwPd contribution (dashed
line). (c) FT of EXAFS data minus calculated PdwPd contribution
(solid line) and calculated PdwC contribution (dashed line).
4374
J. Chem. Soc., Faraday T rans., 1997, V ol. 93

View Article Online
together with OwH bridging with hydrogen of the amine
groups.
tances around 3 Ó and higher indicate the presence of
interaction with second carbon shells of the Ðbrils. Both
PdwPd and PdwC coordination numbers and distances as
given in Table 4 are consistent with a model of a hemispheri-
cal Pd particle of ca. 15 Ó supported on the Ðbril. The large
p-system of the support and possibly dimples in the carbon
surface, due to the oxidation treatment of the Ðbrils, are likely
to account for the stabilisation of the small metal particles
obtained after reduction.
The observation that higher PdwPd backscattering is
absent indicates the presence of monodispersed [Pd(NH ) ]2`
3 4
on the Ðbrils. Taken all results into account, we propose a
model of [Pd(NH ) ]2` above the Ðbril as schematically rep-
3 4
resented in Fig. 7.
Pd/C-Ðbril
Reduction of [Pd(NH ) ]/C-Ðbril in H led to signiÐcant
3 4
2
changes in the XAFS data of the sample. As data analysis
revealed, the four-coordinate palladiumÈtetraamine complex
was reduced to palladium metal particles with an average
coordination number of eight, corresponding to a particle size
of ca. 15 Ó. The PdwPd distance of 2.76 Ó agrees very well
with the bond length in Pd-foil (2.75 Ó). The reason that the
palladium particles after reduction are relatively small can
presumably be explained by the fact that the precursor
complex is monoatomically dispersed on the support after ion
exchange. During careful reduction the resulting metal atoms
migrate to each other, but the thus formed metal particles stay
small due to stabilisation by the carbon Ðbril surface.
Conclusion
This study showed that no oxygen could be detected in the
Ðrst coordination shell of palladium after ion-exchange of acti-
vated carbon Ðbrils with [Pd(NH ) ](NO ) . Furthermore,
analysis of the XAFS data revealed a signiÐcant metalÈÐbril
interaction for ion-exchanged Ðbrils both before and after
3 4
3 2
reduction in H . The results suggest that a well dispersed pre-
cursor complex is stabilised by carboxylic groups on the
2
surface and the p-system of the support. Reduction of this
sample results in very small palladium particles in close
contact with the support, pointing to an important metalÈ
carbon interaction responsible for anchoring the metal par-
ticles to the carbon Ðbrils.
Modelling of the Ðrst palladium coordination shell revealed
a PdwC scattering around 2.60 Ó. As can be seen from Table
4, this contribution exhibits a large inner potential correction
that came out to be independent of the PdwC distance in the
reference Ðle made with FEFF-3.1. As in the case of
The authors would like to thank Michel Onwezen and Gijs
van Breda Vriesman (Utrecht University) for the preparation
of the catalyst, Bob Leliveld (Utrecht University) and Gert
van Dorssen (Daresbury Laboratory) for their company and
assistance at Daresbury Laboratory.
[Pd(NH ) ]/C-Ðbril, this additional contribution in Pd/C-
3 4
Ðbril could also be Ðtted with two or three PdwC interactions
showing much smaller inner potential shifts. However, the sta-
tistical signiÐcance for division of the model into several shells
was only 60%.
References
When the surface of carbon is studied more precisely [Fig.
6(a)], it can be seen that the PdwPd distance of 2.76 Ó dis-
played by the reduced catalyst cannot be found in the highly
symmetric structure of carbon. This suggests that when a
metal particle is attached to a Ðbril, multiple PdwC distances
at the interface of palladium and carbon exist close to each
other, all contributing to the XAFS signal. This can explain
the high inner potential shift for the carbon scattering when
Ðtted with only one shell. The average distance of 2.60 Ó for
the PdwC interactions points to palladium particles at about
2.2È2.4 Ó above the Ðbril. A similar metalÈsupport distance
was also found for platinum metal particles supported on
carbon electrodes.20 The average distance of 2.60 Ó is, further-
more, consistent with metalÈsupport distances found in other
systems like oxide supported catalysts.21h23 Fig. 5(b) shows
that also scattering at a larger distance (3È3.5 Ó) is present.
These contributions could be Ðtted with a PdwPd second
shell contribution and additional PdwC scattering, exhibiting
large inner potential shifts when Ðtted with only one contribu-
tion. If the metal particles are supported on the Ðbril, dis-
1
N. M. Rodriguez, M. S. Kim and R. T. K. Baker, J. Phys. Chem.,
1994, 98, 13108.
M. S. Hoogenraad, R. A. G. M. M. van Leeuwarden, G. J. B. van
Breda Vriesman, A. Broersma, A. J. van Dillen and J. W. Geus,
Stud. Surf. Sci. Catal., 1995, 91, 263.
2
3
4
M. S. Hoogenraad, PhD Thesis, Utrecht University, 1996.
M. C. Roman-Martinez, D. Cazorla-Amoros, A. Linares-Solano
and C. Salinas-Mart•nez de Lecea, Carbon, 1995, 33, 3.
J. S. Noh and J. A. Schwartz, Carbon, 1990, 28, 675.
H. E. van Dan and H. van Bekkum, J. Catal., 1991, 131, 335.
D. J. Suh, T. J. Park and S. K. Ihm, Carbon, 1993, 31, 427.
E. Boellaard, P. K. de Bokx, A. J. H. M. Kock and J. W. Geus, J.
Catal., 1985, 96, 481.
5
6
7
8
9
M. Vaarkamp, B. L. Mojet, M. J. Kappers, J. T. Miller and D. C.
Koningsberger, J. Phys. Chem., 1995, 99, 16067.
10 M. Vaarkamp, I. Dring, R. J. Oldman, E. A. Stern and D. C.
Koningsberger, Phys. Rev. B, 1994, 50, 7872.
11 J. W. Cook, Jr. and D. E. Sayers, J. Appl. Phys., 1981, 52, 5024.
12 K. R. Bauchspiess, Jpn. J. Appl. Phys., 1993, 32, 131.
13 M. Vaarkamp, J. C. Linders and D. C. Koningsberger, Physica B,
1995, 208–209, 159.
14 J. Mustre de Leon, J. J. Rehr, S. I. Zabinsky and R. C. Albers,
Phys. Rev. B, 1991, 44, 4146.
15 K. I. Pandya and D. C. Koningsberger, Physica B, 1989, 158, 386.
16 M. Vaarkamp, PhD Thesis, Eindhoven University of Tech-
nology, 1993.
17 D. M. Bates and D. G. Watts, in Nonlinear Regression Analysis
and its Applications, John Wiley & Sons, New York, 1988.
18 C. L. Mantell, in Carbon and Graphite Handbook, John Wiley &
Sons, New York, 1968.
19 Ch. Elschenbroich and A. Salzer, in OrganometallicsÈA Concise
Introduction, VCH, Weinheim, 1989.
20 W. E. OÏGrady and D. C. Koningsberger, J. Phys. Chem., in
press.
21 M. Vaarkamp, F. S. Modica, J. T. Miller and D. C. Koni-
ngsberger, J. Catal., 1993, 144, 611.
22 D. C. Koningsberger and M. Vaarkamp, Physica B, 1995, 209,
633.
23 M. S. Tzou, B. K. Teo and W. M. H. Sachtler, J. Catal., 1988,
113, 220.
Fig. 7 Schematic representation of proposed structure of
[Pd(NH ) ]/C-Ðbril
Paper 7/04989G; Received 11th July, 1997
3 4
J. Chem. Soc., Faraday T rans., 1997, V ol. 93
4375