Understanding palladium hydrogenation catalysts: When the nature of the reactive molecule controls the nature of the catalyst active phase

Article abstract of DOI:10.1002/anie.200802134

(Graph Presented) Alkynes of everything: Fundamental differences in the palladium-catalyzed hydrogenation of double and triple C-C bonds arise from marked differences in the composition of the catalyst surface. In situ X-ray photoelectron spectroscopy of the near-surface region of active palladium catalysts uncovers strong links between the chemical nature of the (alkyne/alkene) reactive molecules and the subsurface state of the catalyst (I-IV).

Full text of DOI:10.1002/anie.200802134

Communications
DOI: 10.1002/anie.200802134
Heterogeneous Catalysis
Understanding Palladium Hydrogenation Catalysts: When the Nature
of the Reactive Molecule Controls the Nature of the Catalyst Active
Phase**
Detre Teschner,* Zsolt Rꢀvay, Jꢁnos Borsodi, Michael Hꢂvecker, Axel Knop-Gericke,
Robert Schlꢃgl, David Milroy, S. David Jackson, Daniel Torres, and Philippe Sautet
Dedicated to the Catalysis Society of Japan on the occasion of the 50th anniversary
Catalytic hydrogenations are one of the most important
processes in the chemical industry and selectivity is a timely
issue. The presence of triple-bonded compounds in the alkene
stream is undesirable in both chemical and polymer-grade
propylene and ethylene, as they poison the polymerization
catalysts.[1] Starting with a mixture of alkynes and alkenes,
palladium-based catalysts are known to be capable of
selectively hydrogenating triple bonds, leaving the olefinic
function intact. The presence of carbonaceous deposits was
assumed to play a key role, influencing the performance of the
catalyst.[2] Although the origin of this behavior is unknown,
very recently a relation between selectivity and subsurface
chemistry has been established, showing that the population,
by either hydrogen or carbon, of subsurface sites of palladium
governs selective alkyne hydrogenation.[3] Hydrogen in the
subsurface is known to play a crucial role in hydrogenating
surface species effectively.[4,5]
of the Pd/C surface phase is directly related to the nature of
the reactive molecule, being a general pattern for alkyne
hydrogenation and absent for alkenes. We show in addition
that the presence of this Pd/C phase affects the surface
chemistry by decreasing the amount of activated subsurface
hydrogen compared to the more selective surface hydrogen,
hence controlling the state of the Pd catalyst surface under
reaction conditions.
To investigate the nature of the active Pd surface, we have
studied, by in situ X-ray Photoelectron Spectroscopy (XPS),
the near-surface region of palladium under various hydro-
genation conditions for various reacting alkynes and alkenes.
Under reduced pressure conditions (1 mbar), alkynes are
hydrogenated to alkenes while alkenes are transformed to
alkanes, just as at more realistic, atmospheric conditions.
Figure 1 summarizes the state of palladium under such
hydrogenation conditions. The palladium 3d core level was
found to be the best indicator of the near-surface state.[3]
While the metallic bulk peak is observed at 335 eV, a strong
new component (filled peak in Figure 1) appears when
alkynes are hydrogenated selectively. This peak, at 335.6 eV,
is characteristic of the carbon-modified, Pd/C surface phase.
Note that adsorbates give rise to a surface state at similar
energy (adsorbate-induced surface core-level shift). How-
ever, at the applied 720 eVexcitation energy, the contribution
from this state should be less than 25% of the whole peak
intensity.
Clearly, all the investigated alkynes induce the formation
of a Pd/C phase,[3] since the higher binding energy component
dominates the spectra. The only difference seems to be in the
amount of carbon dissolved in the Pd lattice: for lower chain
alkynes more carbon is dissolved in Pd, thickening the Pd/C
phase. In contrast, the hydrogenation of alkenes generates a
much weaker signal at the higher binding energy side of bulk
Pd. According to our fitting procedure, this peak corresponds
to roughly one third of the whole signal intensity, which
indicates that some carbon is indeed dissolved in the upper
part of the metal lattice. However, we attribute the lack of a
clearly distinguishable 335.6 eV component to the absence of
Pd/C. These results show the widely different nature of the
surface under alkyne or alkene feed.
Herein, we extend our previous studies addressing the
fundamental differences of carbon–carbon double or triple
bond hydrogenation over Pd. The present results provide
solid evidence that differences in the hydrogenation of
alkynes and alkenes results from the markedly different
state of the near-surface regions. We show that the formation
[*] Dr. D. Teschner, J. Borsodi, Dr. M. Hꢀvecker, Dr. A. Knop-Gericke,
Prof. Dr. R. Schlꢁgl
Fritz-Haber-Institut der Max-Planck-Gesellschaft
Faradayweg 4-6, 14195 Berlin (Germany)
E-mail: teschner@fhi-berlin.mpg.de
Dr. Z. Rꢂvay, J. Borsodi
Institute of Isotopes, Hungarian Academy of Sciences, Budapest
(Hungary)
Dr. D. Milroy, Prof. Dr. S. D. Jackson
WestCHEM, Department of Chemistry, University of Glasgow (UK)
Dr. D. Torres, Dr. P. Sautet
Laboratoire de Chimie, Universitꢂ de Lyon, CNRS (France)
[**] Support for this work was provided through the ATHENA project,
funded by the Engineering and Physical Sciences Research Council
(UK) and Johnson Matthey plc. The authors acknowledge the co-
operation project between the Fritz-Haber Institute and the Institute
of Isotopes founded by the Max-Plank Gesellschaft. D. Torres
acknowledges support by the Generalitat de Catalunya, under Grant
No. 2006 BP-A 10128. We thank BESSY and the Budapest Neutron
Centre for beamtimes provided.
Since palladium is used in industry to hydrogenate alkynes
from a mixed alkyne/alkene feed, we have investigated the
temporal evolution of the surface using time-resolved in situ
XPS, when co-dosing propyne to a propene/H2 feed. As
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.200802134.
9274
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 9274 –9278

Angewandte
Chemie
components was inverted, with Pd/C reaching
two thirds of the whole signal intensity. The
detected temporal profile indicates that the
carbon uptake can be divided into two
processes: an initial fast step and a second
slower step. Since carbon energetically pre-
fers to occupy the first subsurface octahedral
sites, we assign the fast step to the occupation
of these sites. The very first propyne (alkyne)
molecules chemisorbing on Pd undergo frag-
mentation and the fragments, likely individual
carbon atoms, penetrate the surface. Since the
second step of Pd/C phase formation is much
slower, the build up of additional mixed Pd/C
layers should be energetically less favorable.
The calculated[6] growth rate of the Pd/C layer
is approximately 7 ꢀhꢀ1, which is equivalent
to a diffusion coefficient of carbon (Dc) of
approximately 10ꢀ22 m2 s, in good agreement
Figure 1. Comparison of in situ Pd 3d5/2 spectra of Pd foil under alkene and alkyne
hydrogenation at 1 mbar (H2/CxHy: 9/1) and 343–353 K. A: 1-pentene; B: propene; C:
ethylene; D: 1-pentyne; E: propyne; F: acetylene.
expected from the previous results, the Pd 3d core level was
dominated by the bulk component in the propene + H2 feed
(see Supporting Information, Figure S1). It was however
strongly attenuated in the first few minutes after propyne was
introduced, and the proportion of the higher binding-energy
component slowly increased further in the mixed feed
(Figure 2A). After 30 minutes, the ratio of the two Pd
with the results of Yokoyama et al.[7] (10ꢀ23 m2 s) extrapolated
to our temperature.
Similar behavior in carbon uptake is also seen over a Pd
catalyst under industrially relevant conditions. Figure 2B
shows the conversion and selectivity obtained from acetylene
hydrogenation. Over the fresh catalyst there is poor alkene
selectivity but with time this selectivity increases from 27% to
95%. Over the same period the mass balance increases from
96.5 to 100% showing that as carbon is deposited the
selectivity increases. Note that there is little change in activity.
To hydrogenate a given chemical functionality, the
catalyst clearly needs hydrogen. For palladium, its propensity
to form subsurface- or bulk-dissolved hydrogen (e.g. b-
hydride) makes the assignment of the active hydrogen non-
trivial. Both, experimental and theoretical studies clearly
indicated that hydrogen, when emerging from the subsurface/
bulk to the surface, could very effectively hydrogenate
different adsorbates.[4,5] If subsurface hydrogen plays an
important role in our processes, the occupation of the first
subsurface layer with carbon would be critical in the hydro-
genation path. To be able to measure and quantify hydrogen
dissolved in palladium we have developed prompt gamma
activation analysis (PGAA)[8] as an in situ technique.[9] In the
PGAA process, the nuclei of chemical elements in the
investigated specimen capture neutrons and, during the de-
excitation, they emit characteristic prompt gamma photons,
which can be detected using an appropriate spectrometer.
Hydrogen has a medium neutron-capture cross-section, but
gives a strong signal in the gamma spectrum, thus hydrogen
atoms dissolved in palladium can be analyzed by PGAA with
high sensitivity.
Table 1 compares the hydrogen content of palladium
black during 1-pentyne and 1-pentene hydrogenation. When
fresh samples are introduced into flowing hydrogen, the
hydrogen content matched perfectly the values from the Pd/H
phase diagram.[10] Thereafter the H/Pd ratio decreased
significantly when selective hydrogenation of 1-pentyne was
carried out using a feed ratio (H2/C5H8) of 2.5. The hydrogen
content in pure hydrogen of the sample after 1-pentyne
hydrogenation (Table 1, H2 (3rd)) was slightly higher than
Figure 2. A: Temporal evolution of the proportion of the higher binding
energy Pd 3d5/2 component at approximately 335.6 eV after adding
0.1 mbar propyne in the feed of 0.9 mbar H2 and 0.1 mbar propene
according to time-resolved in situ XPS. T: 343 K. B: Change in selectiv-
ity of acetylene hydrogenation at 20 bar and 323 K. Catalyst Pd/
alumina, Gas Hourly Space Velocity (GHSV): 5000, hydrogen:ethyle-
ne:acetylene ratio, 25:25:1.
Angew. Chem. Int. Ed. 2008, 47, 9274 –9278
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
9275

Communications
Table 1: Conversion, product selectivities and atomic H/Pd ratios during
in situ PGAA experiments.[a]
Experiments
Conversion [%]
Products [%]
H/Pd
H2[b] (1st)
1-pentyne (1st)
–
21
–
0.75
0.27
97.7 1-pentene
2.3 pentane
–
98.1 1-pentene
1.9 pentane
–
H2[b] (2nd)
1-pentyne (2nd)
–
44
0.77
0.01
H2 (3rd)
1-pentene (3rd)
–
99.8
0.88
0.77
98.6 pentane
1.4 2-pentenes
[a] Sample: 7 mg Pd black; in flowing hydrogen, under 1-pentyne or 1-
pentene hydrogenation (after H2 pretreatment); at 1 bar and room
temperature. Gas flow during hydrogenation: 4 mLminꢀ1 H2 and
1.6 mLminꢀ1 C5Hx. 1-pentyne hydrogenation was repeated. [b] Fresh Pd
black sample.
with fresh sample, suggesting that the surface incorporates
carbonaceous deposits containing additional hydrogen. After
1-pentene hydrogenation, the hydrogen content of the sample
was even higher than with the equivalent alkyne. This
difference can be related to the XPS data (see below).
To achieve an atomistic understanding for the formation
of the Pd/C phase, we first addressed the incorporation of C
into the Pd substrate. We performed self-consistent density
functional theory (DFT) calculations using the VASP code[11]
Figure 3. Top: Contour maps showing the average binding energy
(color coded, in eV) of carbon as a function of total C coverage
(X axis) and the distribution of carbon atoms between interlayers
(Y axis) using a Pd(111) substrate. C atoms are distributed between
surface and 1st interlayer sites (left) or between 1st and 2nd inter-
layer (right). Bottom: Contour maps showing the average binding
energy (colour coded, in eV) of hydrogen as a function of total H
coverage (X axis) on Pd(111) (left) or on the model Pd/C phase (right).
H atoms are distributed between surface and 1st interlayer sites.
[
]
*
and a periodic Pd(111) model surface. The results of our
calculations are shown in Figure 3 (top panels). We deter-
mined the average binding energy of carbon (ECb ) as a
function of total C coverage (VC) with different distributions
of C between surface, subsurface, and second interlayer. After
exploring all possible adsorption site combinations, the most
stable situation is shown.
To describe the XPS experiments herein, at constant
pressure and temperature, the appropriate thermodynamic
potential to consider is the Gibbs free surface energy g(T,P).
This scheme has been successfully applied to various sys-
tems,[13] and we will use it to determine the thermodynami-
cally stable composition of the surface in equilibrium with a
reactive atmosphere, described by acetylene and hydrogen
partial pressures (pC H and pH respectively) at a fixed
The initial stages in the formation of the Pd/C phase
involve the incorporation of carbon from the surface to the
first interlayer. Independently of VC, structures involving C
atoms adsorbed only in the subsurface are energetically
favored (Figure 3, top left panel) with respect to structures
involving C adatoms on the surface or on both the surface and
2
2
2
temperature T. Assuming that thermodynamic equilibrium
applies, the environment then acts as a C adatoms reservoir,
as it can give or take C atoms from the decomposition or
formation of carbon-containing molecules (H atoms concert-
edly being evolved as H2 to the gas phase). The calculated free
surface energy plot as a function of the chemical potential of
C (mC) is shown in Figure 4. For low values of mC, molecule
decomposition and C deposition onto the surface is not
favored. Only once the mC has reached the critical value of
ꢀ8.8 eV (mCCrit), does the carbon deposition start.
subsurface. The most stable conformation corresponds to a
pffiffi pffiffi
3 ꢁ 3 distribution of C in the first interlayer.[12] Regarding
the C distribution between the first and second interlayer, the
situation changes. Placing the C atoms in the second
interlayer is roughly as favorable as placing them in the
first. For coverage higher than approximately 0.3 monolayer
(ML), distribution of the C atoms on both first and second
interlayer is energetically more favored than the population
of a single interlayer, creating the thermodynamic driving
force for the growth of the Pd/C phase. We considered C
incorporation into deeper interlayers and, as a general rule,
we detected a significant weakening of the average C binding
when an increasing number of interlayers were populated;
hence this will limit deeper extension of the Pd/C phase.
When acetylene and hydrogen are dosed onto the catalyst
surface, reaction starts and mC varies within certain well-
defined boundaries. While the upper boundary (mMC ax) is given
by the chemical potential of carbon in the acetylene molecule,
the lower boundary is given by the chemical potential of
carbon in ethylene, at the given H2 pressure. The formation of
the Pd/C phase will be thermodynamically favored under
reaction conditions only when mC is above the critical
equilibrium mCCrit value for the migration of carbon to the
subsurface. From Figure 4 it is clearly the case for acetylene
hydrogenation, but not for ethylene hydrogenation. The
stability of this phase for acetylene is dependent on the H2
[*] Note, the Pd(111) surface was used as a model aiming only to extract
the main phenomena for C penetration and Pd/C phase formation.
Hence DFTwas used to qualitatively (and not quantitatively) describe
the experiments.
9276 www.angewandte.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 9274 –9278

Angewandte
Chemie
clean Pd(111) surface, the binding energy does not strongly
depend on the coverage (Figure 3, bottom left panel). H is
most stable on the surface, but penetration into the subsurface
is more favored than desorption.[14] In contrast, the bonding
properties of H are strongly modified for Pd/C (Figure 3,
bottom right panel). Adsorption on the surface is weakened
and, most importantly, the accumulation of H into the
subsurface is thermodynamically disfavored. Thus one role
of the Pd/C phase is to hinder the migration of H to the
subsurface, hence decreasing the Hsub/Hon ratio in the sample.
The Pd/C phase will, in addition, prevent the migration of
bulk H toward the surface. Clearly, DFT is in accord with the
PGAA results because, once the b-hydride has been depleted,
hydrogen is hindered in replenishing the bulk, and the H/Pd
ratio will be low. Hence, alkynes are hydrogenated selectively
by surface hydrogen, since hydrogen cannot emerge from the
bulk, if it is present at all. On the other hand, alkene
hydrogenation occurs using subsurface hydrogen, as its
concentration is high and no energetic barrier is built up by
a subsurface carbon population.
These data clearly indicate that catalytic palladium
materials are highly versatile, and their actual (near-)surface
compositions are a strong function of the experimental
hydrogenation conditions. Understanding such phenomena
will allow the design of heterogeneous catalysts to be tailored
to a desired reaction. The present study clearly demonstrates
the significance of combining theory and experimentation in
bringing understanding to a new level.
Figure 4. Surface free energy g (meVꢃꢀ2) of the most stable Pd/C
structures as a function of carbon chemical potential mC. Structure I is
the pure Pd(111) surface, II has a C content of 1/3 ML in the first
interlayer, III a C content of 2/3 ML distributed among first and
second interlayers, and IV a C content of 1 ML distributed in the first
three interlayers. Vertical blue lines indicate the mC value for the
dissociation of acetylene (right line) and ethylene (left line) into C and
H2, at pressure conditions employed in XPS (10ꢀ4, and 10ꢀ3 atm for
pC H and pH ). mC is displayed in a logarithmic pressure scale at a fixed
2
x
2
temperature of 400 K. Acetylene and ethylene pressures with respect to
p0 (1 atm) are, respectively, shown in the bottom and top X axes.
pressure, because the effective chemical potential of carbon
from the molecular reservoir depends on H2 pressure. At high
H2 pressure, formation of the phase becomes thermodynami-
cally inhibited.
Experimental Section
The difference (Dgf) between the surface free energy of
the most stable structure and that of the Pd(111) surface, for
the mC associated to the reactant, can be considered as a
descriptor of the thermodynamic driving force for the Pd/C
phase formation. For the hydrogenation of acetylene, at
typical pressure conditions during in situ XPS (10ꢀ4 and
10ꢀ3 atm for pC H and pH ), and at 400 K, the formation of the
In situ X-ray photoelectron spectroscopy (XPS) experiments were
performed at beamlines U49/2-PGM1 and PGM2 at BESSY, Berlin.
Pd 3d core levels of a Pd foil sample were recorded at normal
emission with 720 eV excitation, corresponding to an inelastic mean
free path of approximately 9 ꢀ.[15]
In situ prompt gamma activation analysis (PGAA) was carried
out at the cold neutron beam of the Budapest Neutron Centre,
Budapest, Hungary.[16] H/Pd molar ratios of 7 mg Pd black were
determined under different conditions from the characteristic peak
areas corrected by the detector efficiency and the nuclear data of the
detected elements.[8]
2
2
2
Pd/C phase resulting from the decomposition of acetylene is
thermodynamically favored, with Dgf ꢁ ꢀ25 meVꢀꢀ2. In
contrast, the formation of the phase in the presence of
ethene is disfavored, with positive Dgf. To get more insight
into the relation of the Pd/C phase formation with respect to
the nature of the reacting molecule we calculated Dgf for a set
of reactive molecules of the type HC ꢂ CX, with X = CH3, H,
F, Cl (see Supporting Information, Figure S2). The formation
of the Pd/C phase is favored for lower-chain alkyne hydro-
genation or when electronegative species are attached to the
triple bond, qualitatively explaining the thicker Pd/C phase
obtained experimentally when going from 1-pentyne, to
propyne, to acetylene.
The 20 bar acetylene hydrogenation was carried out in
a
continuous-flow microreactor with on-line GC using a 0.02% Pd/
alumina catalyst.
Density functional theory-based calculations on slab models have
been carried out to investigate the accumulation of carbon on and in
Pd(111). The adsorption energy was evaluated using the PW91
function.[17] The PAW method[18] was used to represent the inner cores
and one electron states were expanded in a plane wave basis with a
kinetic cut-off energy of 400 eV. A Monkhorst-Pack mesh with 11 ꢁ
11 ꢁ 1 and 9 ꢁ 9 ꢁ 1 k-points was used for small and large cells.[19]
For more experimental details, please consult the Supporting
Information available online.
Finally, the presence of C in the subsurface affects the
properties of surface hydrogen. To analyze this influence, we
compared the binding energy of H on Pd(111) and on a Pd/C
phase model. To match the experiments best, we chose
structure IV of Figure 4, with the first three interlayers
occupied with one third C coverage. The average binding
energy of H (EHb ) as a function of total H coverage (VH) was
calculated with respect to the gas phase H2 molecule. On the
Received: May 7, 2008
Published online: August 13, 2008
Keywords: adsorption · density functional calculations ·
.
hydrogenation · palladium · photoelectron spectroscopy
Angew. Chem. Int. Ed. 2008, 47, 9274 –9278
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
9277

Communications
[7] H. Yokoyama, H. Numakura, M. Koiwa, Acta mater. 1998, 46,
2823.
[8] Handbook of Prompt Gamma Activation Analysis with Neutron
Beams (Ed.: G. L. Molnꢆr), Kluwer Academic Publishers,
Dordrecht, 2004.
[9] Zs. Rꢃvay, T. Belgya, L. Szentmiklꢇsi, Z. Kis, A. Wootsch, D.
Teschner, M. Swoboda, R. Schlꢅgl, J. Borsodi, R. Zepernick,
Anal. Chem. 2008, 80, 6066.
[1] a) H. Arnold, F. Dobert, J. Gaube in Hydrogenation Reaction in
Handbook of Heterogeneous Catalysis, Vol. 5 (Eds.: G. Ertl, H.
Knozinger, J. Weitkamp), Wiley–VCH, New York, 1997, p. 2165;
b) A. Molnar, A. Sarkany, M. Varga, J. Mol. Catal. A 2001, 173,
185.
[2] A. Borodziꢂski, G. C. Bond, Catal. Rev. 2006, 48, 91, and
references therein.
[3] a) D. Teschner, J. Borsodi, A. Wootsch, Zs. Rꢃvay, M. Hꢄvecker,
A. Knop-Gericke, S. D. Jackson, R. Schlꢅgl, Science 2008, 320,
86; b) D. Teschner, E. M. Vass, M. Hꢄvecker, S. Zafeiratos, P.
Schnꢅrch, H. Sauer, A. Knop-Gericke, R. Schlꢅgl, M. Chamam,
A. Wootsch, A. S. Canning, J. J. Gamman, S. D. Jackson, J.
McGregor, L. F. Gladden, J. Catal. 2006, 242, 16.
[4] A. D. Johnson, S. P. Daley, A. L. Utz, S. T. Ceyer, Science 1992,
257, 223.
[10] H. Frieske, E. Wicke, Ber. Bunsen-Ges. 1973, 77, 48.
[11] G. Kresse, J. Hafner, Phys. Rev. B 1993, 47, 558.
[12] L. Gracia, M. Calatayud, J. Andrꢃs, C. Minot, M. Salmeron,
Phys. Rev. B 2005, 71, 033407.
[13] K. Reuter, M. Scheffler, Phys. Rev. B 2003, 68, 045407.
[14] a) W. Dong, V. Ledentu, P. Sautet, A. Eichler, J. Hafner, Surf.
Sci. 1998, 411, 123; b) J.-F. Paul, P. Sautet, Phys. Rev. B 1996, 53,
8015.
[5] a) A. Michaelides, P. Hu, A. Alavi, J. Chem. Phys. 1999, 111,
1343; b) V. Ledentu, W. Dong, P. Sautet, J. Am. Chem. Soc. 2000,
122, 1796.
[15] S. Tanuma, C. J. Powell, D. R. Penn, Surf. Interface Anal. 1991,
17, 911.
[16] Zs. Rꢃvay, T. Belgya, Zs. Kasztovszky, J. L. Weil, G. L. Molnꢆr,
Nucl. Instrum. Methods Phys. Res. Sect. B 2004, 213, 385.
[17] J. P. Perdew, J. A. Chevary, S. H. Vosko, K. A. Jackson, M. R.
Pederson, D. J. Singh, C. Fiolhais, Phys. Rev. B 1992, 46, 6671.
[18] P. E. Blꢅchl, Phys. Rev. B 1994, 50, 17953.
[19] H. J. Monkhorst, J. D. Pack, Phys. Rev. B 1976, 13, 5188.
[6] Thickness of the Pd/C phase was calculated, assuming only
inelastic scattering, using the formula
:
d = (IMFP/
cosE)*(1+I335.6 eV/I335 eV), where I is the corresponding peak
`
`
area, E the takeoff angle, and IMFP the inelastic mean free path.
IMFP was 9 ꢀ at 720 eV. The growth of the Pd/C phase was
determined from the temporal evolution of the thickness.
9278 www.angewandte.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 9274 –9278