Selective hydrogenation of 1,3-butadiene on PdNi bimetallic catalyst: From model surfaces to supported catalysts

Article abstract of DOI:10.1016/j.jcat.2014.04.015

The selective hydrogenation of 1,3-butadiene serves as a means to purify the butene stream generated from cracking naphtha or gas oil. To identify selective hydrogenation catalysts, 1,3-butadiene was studied on single crystal Ni/Pd(1 1 1) bimetallic surfaces, utilizing density functional theory (DFT) calculations and temperature-programmed desorption (TPD). DFT calculations predicted that the Pd-terminated bimetallic surface should be more active and selective to produce 1-butene, which were verified experimentally using TPD. The promising results on model surfaces were extended to γ-Al ₂O₃-supported catalysts using both batch and flow reactors. Extended X-ray absorption fine structure (EXAFS) and transmission electron microscopy (TEM) confirmed the formation of bimetallic nanoparticles. The PdNi bimetallic catalyst showed higher hydrogenation activity and 1-butene selectivity than the monometallic catalysts. The excellent correlation between model surfaces and supported catalysts demonstrates the feasibility of designing effective bimetallic catalysts for selective hydrogenation reactions.

Full text of DOI:10.1016/j.jcat.2014.04.015

Journal of Catalysis 316 (2014) 1–10
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Selective hydrogenation of 1,3-butadiene on PdANi bimetallic catalyst:
From model surfaces to supported catalysts
a,
Ruijun Hou a,b, Weiting Yu b, Marc D. Porosoff c, Jingguang G. Chen b, , Tiefeng Wang
⇑
⇑
a Beijing Key Laboratory of Green Reaction Engineering and Technology, Department of Chemical Engineering, Tsinghua University, Beijing 100084, China
b Department of Chemical Engineering, Columbia University, New York, NY 10027, United States
c Department of Chemical and Biomolecular Engineering, University of Delaware, Newark, DE 19716, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
The selective hydrogenation of 1,3-butadiene serves as a means to purify the butene stream generated
from cracking naphtha or gas oil. To identify selective hydrogenation catalysts, 1,3-butadiene was studied
on single crystal Ni/Pd(111) bimetallic surfaces, utilizing density functional theory (DFT) calculations
and temperature-programmed desorption (TPD). DFT calculations predicted that the Pd-terminated
bimetallic surface should be more active and selective to produce 1-butene, which were verified exper-
Received 18 December 2013
Revised 10 April 2014
Accepted 23 April 2014
imentally using TPD. The promising results on model surfaces were extended to c-Al2O3-supported cat-
Keywords:
alysts using both batch and flow reactors. Extended X-ray absorption fine structure (EXAFS) and
transmission electron microscopy (TEM) confirmed the formation of bimetallic nanoparticles. The PdANi
bimetallic catalyst showed higher hydrogenation activity and 1-butene selectivity than the monometallic
catalysts. The excellent correlation between model surfaces and supported catalysts demonstrates the
feasibility of designing effective bimetallic catalysts for selective hydrogenation reactions.
Ó 2014 Elsevier Inc. All rights reserved.
Ni/Pd(111) surfaces
PdANi bimetallic catalysts
Selective hydrogenation
1,3-Butadiene
1. Introduction
lower hydrogenation activity [12]. The effect of bimetallic forma-
tion in PdAAu appeared to be dependent on the reaction tempera-
Selective hydrogenation is one of the critical reactions to
remove alkynes and dienes in polymer synthesis. The hydrocarbon
streams from cracking units typically contain ꢀ1% alkynes or
dienes, which must be removed below 10 ppm for polymerization
processes because they poison the catalysts and degrade the prod-
uct quality. For example, the hydrogenation of 1,3-butadiene is
used for the purification of butene streams. Among the products
of 1,3-butadiene hydrogenation, 1-butene is the desired product
for manufacturing polybutene and polyethylene. In this paper,
1,3-butadiene was chosen as a probe molecule to identify catalysts
for the selective hydrogenation of one C@C bond in dienes.
Pd- [1–5] and Pt- [1,6–9] based catalysts are commonly studied
for the selective hydrogenation of unsaturated CAC bonds. For the
selective hydrogenation of dienes, Pd-based catalysts are more
active and selective than their Pt-based counterparts [10]. The
Pt- and Pd-based bimetallic catalysts typically display higher selec-
tivity than the corresponding monometallic catalysts [9,11–24].
For example, PdAAg showed higher selectivity to butenes, but
ture and 1,3-butadiene conversion. At low temperature, 294 K,
PdAAu was reported to have a lower activation barrier for 1,3-
butadiene hydrogenation. The PdAAu bimetallic catalysts showed
higher selectivity to 1-butene than Pd at 1,3-butadiene conversions
higher than 30%. However, at conversions lower than 30%, the
PdAAu catalyst was less selective than Pd to 1-butene formation
[15]. The PdACu catalyst had higher 1-butene selectivity, but
maintained similar activity to that of Pd [18]. PdASn was studied
on different alumina supports, which showed that the addition of
Sn to Pd increased the butene selectivity, but had an insignificant
effect on the catalyst activity [22]. The inclusion of Mo in Pd sup-
ported on zeolites decreased the initial catalytic activity of 1,3-
butadiene hydrogenation, while the selectivity to butene was
improved [23]. Overall, most of the Pd-based bimetallic catalysts
reported in the literature showed improved butene selectivity,
but often at the expense of activity. Therefore, it is important to
identify bimetallic catalysts that increase both selectivity and
activity.
Previous studies on model surfaces of Pd5Ni95 [25] and Pd8Ni92
[26] showed higher activity for 1,3-butadiene hydrogenation
(based on surface Pd atoms) than Pd(111) and Pd(110) surfaces,
and the selectivities to butene on these Pd-Ni surfaces were the
same as on Pd surfaces. Furthermore, when the Ni surface was
⇑
Corresponding authors. Fax: +86 10 62772051 (T.F. Wang); +1 212 854 3054
(J.G. Chen).
E-mail addresses: jgchen@columbia.edu (J.G. Chen), wangtf@tsinghua.edu.cn
(T. Wang).
http://dx.doi.org/10.1016/j.jcat.2014.04.015
0021-9517/Ó 2014 Elsevier Inc. All rights reserved.

2
R. Hou et al. / Journal of Catalysis 316 (2014) 1–10
modified with 0.5 ML Pd, the catalytic activity for 1,3-butadiene
hydrogenation increased by an order of magnitude [27]. These sur-
face science studies suggest that PdANi bimetallic catalysts have
higher selectivity for 1-butene production and higher activity for
1,3-butadiene hydrogenation.
The objective of this work was to correlate PdANi bimetallic
surfaces with supported catalysts for 1,3-butadiene hydrogenation.
Surface science experiments and density functional theory (DFT)
calculations were performed on Ni-modified Pd(111) surfaces to
determine the effect of PdANi bimetallic bonds on the hydrogena-
tion pathway of 1,3-butadiene. In parallel, supported PdANi cata-
lysts were characterized and evaluated in both batch and flow
reactors to verify the predictions from the model surfaces.
0.20
0.16
0.12
0.08
0.04
0.00
(a)
B
A
0
2
4
6
8
10
2. Theoretical and experimental methods
Deposition time (min)
2.1. DFT calculations
1.0
0.8
0.6
0.4
0.2
0.0
All DFT calculations of binding energies were performed with
the Vienna ab initio Simulation Package (VASP) [28–31]. The PW
91 functional [32] was used in the generalized gradient approxi-
mation (GGA) [33] calculation, and a kinetic cutoff energy of
396 eV was chosen for the plane wave truncation. The surfaces
were modeled using a periodic 4 ꢁ 4 unit cell and the calculations
used a 3 ꢁ 3 ꢁ 1 Monkhorst–Pack k-point grid. The clean Pd(111)
surface was modeled by adding six equivalent layers of vacuum
onto four Pd layers, in which the two bottom layers were frozen
at metal distances of 2.80 ÅA in each layer and 2.3 ÅA between the
layers, while the top two layers were allowed to relax to reach
the lowest energy configuration. The Pd-terminated bimetallic sur-
face, designated as PdNiPd(111), was derived by replacing the sec-
ond layer of Pd atoms with Ni atoms. The binding energy of
adsorbed molecules on the surface was calculated by subtracting
the energies of the bare slab and of the free molecule from the total
energy of the slab plus the adsorbed molecule. The configurations
from VASP calculations were further optimized using the Cam-
bridge Serial Total Energy Package (CASTEP) suite of programs to
identify the transition state and calculate the activation energy
for each reaction step.
(b)
A
C
D
0
0
Regime I
300
Regime III
Regime II
400
500
600
700
800
900
Temperature (K)
Fig. 1. (a) Deposition time vs AES Ni/Pd peak ratio and (b) thermal stability of the
monolayer Ni surface (Regime I: NiPdPd(111) configuration; Regime II: transition
from NiPdPd(111) to PdNiPd(111) configuration; Regime III: diffusion of Ni into
bulk phase).
of the NiPdPd(111) surface, it was heated to higher temperatures,
and Auger measurements [36] were performed every 50 K. The
change in the Ni/Pd Auger peak ratio was plotted as a function of
the surface temperature in Fig. 1b. The slope of the INi/IPd curve
changed when the surface was heated between 550 K (point C)
and 650 K (point D), which indicated that Ni atoms started to dif-
fuse into the Pd substrate [37,38]. Based on the previous studies of
the thermal behavior of Ni/Pt(111), heating to approximately
600 K led to diffusion of Ni between the first and second layers
of Pt to produce a subsurface PtNiPt(111) structure, followed by
diffusion of Ni further into the bulk at higher temperatures [39].
The observation of two distinct slopes in Fig. 1b suggested that
two thermally induced diffusion regimes, subsurface and bulk dif-
fusion, might exist for the Ni/Pd(111) surface. In the current paper,
the NiPd surface annealed to 600 K is referred to as the subsurface
PdNiPd(111) structure in DFT calculations and TPD measurements.
However, the slope of Ni subsurface diffusion is greater than that of
the Ni bulk diffusion, which is uncommon compared with the pre-
vious studies of metal/Pt(111) surfaces [40]. The only plausible
explanation is that the decrease in the Ni AES signal is related to
the term of exp[ꢂd/k], where d is the thickness of Pd layers screen-
ing the Ni atoms and k the mean free path of Ni Auger electrons.
Because the mean free path of the Ni(LMM) electrons is relatively
large, it is possible that Ni can still be detected by AES, although at
a reduced intensity of exp[ꢂd/k], even after Ni atoms diffuse sev-
eral layers deep in the bulk. However, it is difficult to quantify
2.2. Measurements of model surfaces
TPD measurements were performed in an ultrahigh vacuum
(UHV) chamber with a base pressure of 1 ꢁ 10ꢂ10 Torr, equipped
with an Auger electron spectroscopy (AES), a mass spectrometer,
a sputter gun, and a Ni metal source, as described previously
[34]. A Pd(111) single crystal (Princeton Scientific, 99.99%, 2 mm
thick and 10 mm in diameter) was placed at the center of the
UHV chamber by directly spot-welding the crystal to two tantalum
posts, for resistive heating and cooling with liquid nitrogen. A chro-
mel–alumel K type thermocouple was welded to the back of the Pd
sample for temperature measurements. The Pd(111) surface was
cleaned using sputtering–annealing cycles.
The
Ni-terminated,
NiPdPd(111)
and
Pd-terminated,
PdNiPd(111) bimetallic surfaces were prepared by controlling
the current of the Ni source, the temperature of the Pd(111) sur-
face, and the deposition time. The NiPdPd(111) surface was pre-
pared by depositing Ni onto the Pd(111) surface, while the
surface was maintained at 300 K. The ratio of Ni and Pd peak inten-
sities (INi/IPd) from Auger measurements is plotted in Fig. 1a as a
function of deposition time. Breaks were found at 4 min (point A)
and 8 min (point B), suggesting that the deposition of the Ni atoms
on the Pd(111) surface was in a layer-by-layer [35] growth mode.
At point A, one layer of Ni atoms was deposited on the Pd(111)
surface, referred to as the NiPdPd(111) surface. After preparation

R. Hou et al. / Journal of Catalysis 316 (2014) 1–10
3
the slope without knowing the rate and depth of Ni diffusion into
the bulk Pd(111). The surface structure and elemental composition
of the 300 K and 600 K Ni/Pd(111) surfaces have been character-
ized using scanning tunneling microscopy (STM) and low-energy
ion scattering (LEIS) [41]. The LEIS results confirmed the thermally
induced diffusion of Ni, leading to a Pd-terminated surface after
annealing to 600 K.
images were taken in scanning (STEM) mode with a 12-nm camera
length and a 1-nm diameter nanoprobe. The TEM samples were
prepared by first reducing the catalyst samples, then grinding the
reduced sample into fine powders, and suspending them in ace-
tone. Droplets of the suspension were placed onto a carbon coated
copper grid. The grid was dried overnight before loading the sam-
ple into the TEM.
All gases used in this work, 1,3-butadiene, hydrogen, and neon,
were of research purity and were used without further purification.
The reagents were dosed into the UHV chamber through a stainless
steel tube, and the purity was verified with mass spectrometry
before UHV experiments.
2.4.3. Extended X-ray absorption fine structure (EXAFS)
EXAFS measurements were used to confirm the presence of
PdANi bimetallic bonds. Measurements of the Pd K-edge were per-
formed on the X18B beamline at the National Synchrotron Light
Source (NSLS), Brookhaven National Laboratory. The catalyst sam-
ples were pressed into a pellet with a thickness twice the absorp-
tion length to maximize the signal to noise ratio. The catalysts
were reduced in situ in 40 sccm of 50% H2 in He at 723 K for
1.5 h, and the EXAFS spectra were collected at room temperature.
The EXAFS spectra were calibrated to the Pd K-edge energy from
a Pd reference foil.
The X-ray signal was analyzed using the IFEFFIT 1.2.11 data
analysis package (Athena, Aretmis, Atoms, and FEFF6) [42,43].
The structural information was obtained by Fourier-transforming
the absorbance signal into R-space and then fitting each data set
to the theoretical standards generated in FEFF6 [44]. The theoreti-
cal monometallic photoelectron amplitude and phase were calcu-
lated for the bulk Pd fcc structure. The passive electron reduction
factor S20 was found to be 0.878 from fitting the Pd foil data. The
seven parameters used in the fitting procedure were the correction
to the edge energy, the coordination numbers of the PdAPd and
PdANi bonds, corrections to their model interatomic distances,
and the mean square deviations in interatomic distances (EXAFS
Debye–Waller factors).
2.3. Preparation of supported catalysts
The
c
-Al2O3 (Alfa Aesar, 244 m2/g) supported catalysts were
synthesized using incipient wetness impregnation. The precursor
solutions were prepared by dissolving Pd(NO3)2ꢃ2H2O (Alfa Aesar)
and/or Ni(NO3)2ꢃ6H2O (Alfa Aesar) in an amount of water just suf-
ficient to fill the pores of 3 g of the support. The precursor solution
was then added to the support by dropwise addition and was stir-
red thoroughly between droplets. The catalysts were dried at 373 K
for 10 h and calcined at 563 K for 2 h. A ramp rate of 0.4 K/min was
used up to 373 K and 0.8 K/min up to 563 K. A co-impregnation
synthesis procedure was used for the bimetallic catalysts to pro-
duce the greatest extent of bimetallic bond formation, as verified
previously for supported PtANi catalysts [24]. The corresponding
monometallic Ni and Pd catalysts were prepared to serve as control
samples. The compositions of the synthesized catalysts are as fol-
lows: 1.51 wt% Ni/c-Al2O3, 0.91 wt% Pd/c-Al2O3, and 1.51 wt%
Ni–0.91 wt% Pd/c-Al2O3, corresponding to a Ni:Pd atomic ratio of
3:1.
2.4. Catalyst characterization
2.5. Catalytic evaluation of supported catalysts
2.4.1. Pulse CO chemisorption
2.5.1. Batch reactor
The number of active sites on each catalyst was measured
through CO uptake with an AMI-200ip (Altamira). A quartz reactor
was loaded with 100 mg of catalyst and reduced in 40 sccm of 50%
H2 in He at 723 K for 1 h. After reduction, the catalyst was cooled to
room temperature in He before pulsing CO. The amount of CO flow-
ing out of the reactor was measured with a thermal conductivity
detector. Each adsorbed CO molecule was assumed to correspond
to one active site, which provided a means to quantitatively com-
pare the number of active sites between catalysts.
Fourier transform infrared (FT-IR) spectroscopy was used to
monitor the gas-phase concentrations of reactants and products
during the hydrogenation 1,3-butadiene. The FT-IR spectra were
recorded with 4-cmꢂ1 resolution using a Thermo Nicolet Nexus
470 spectrometer equipped with a mercury cadmium telluride
(MCT-A) detector. The details of the batch reactor cell and the sam-
ple preparation procedure have been previously reported [24]. For
all FT-IR experiments, ꢀ13.6 mg of supported catalyst was loaded
into the IR cell, which was then evacuated for 12 h, until the sys-
tem reached a pressure below 10ꢂ6 Torr. The catalysts were
reduced at 723 K in 30 Torr hydrogen for 30 min, the cell was then
evacuated, and a high-temperature flash (723 K) was performed to
remove any surface species generated during the reduction period.
2.4.2. Transmission electron microscopy (TEM)
TEM measurements were performed using
a JEOL2010F
equipped with a Schottky field emission gun at 200 keV. All TEM
Fig. 2. Top and side views of optimized adsorption geometry of C4H6, C4H7, C4H8, C4H9, and C4H10 species on PdNiPd(111) surfaces.

4
R. Hou et al. / Journal of Catalysis 316 (2014) 1–10
Scheme 1. The reaction pathways for 1,3-butadiene hydrogenation on Pd(111) and PdNiPd(111) surfaces. The activation barriers are in eV.
Table 1
The binding energies on Pd(111) and PdNiPd(111) surfaces.
of the reactant and products were estimated using the absor-
bance intensities of their characteristic vibration modes as
Surface
Pd(111)
PdNiPd(111)
follows:
m
(C@CAC@C) at 1586 cmꢂ1 for 1,3-butadiene,
m
(C@C) at
d-Band center (eV)
Binding energy (eV)
ꢂ1.9
ꢂ2.25
1655 cmꢂ1 for 1-butene, and
m
(ACH3) at 1466 cmꢂ1 for n-butane.
No 2-butene characteristic vibration modes (1697 cmꢂ1 and
1540 cmꢂ1) were found in the gas-phase products. Details of the
calculation of the gas-phase concentrations of molecules from
1,3-butadiene hydrogenation were reported previously [24]. All
carbon atoms were accounted for by performing a carbon balance,
confirming the absence of 2-butenes and any other hydrocarbon
fragments. At present, we do not understand why 2-butene was
not detected in the batch reactor. One potential explanation is that,
in the batch environment, gas-phase products constantly undergo
re-adsorption and reaction. More detailed study is needed to
understand such phenomena; however, because our batch reactor
results are primarily used for activity comparison, the absence of
2-butene does not affect our batch reactor conclusions of trends
in hydrogenation activity.
C4H6
C4H7
C4H8
C4H9
C4H10
ꢂ1.49
ꢂ1.62
ꢂ0.55
ꢂ1.58
ꢂ0.11
ꢂ0.83
ꢂ0.58
ꢂ0.13
ꢂ1.29
ꢂ0.11
The reduction cycle was repeated three times before performing
1,3-butadiene hydrogenation experiments.
After reducing the catalyst, 7.8 Torr C4H6, 17.2 Torr H2, and
25 Torr He were introduced into the reaction vessel simulta-
neously, with a 1,3-butadiene to hydrogen ratio of 1:2.2. The reac-
tion proceeded at a catalyst temperature of 308 K. During the
reaction, the gas-phase reactants and products were monitored
by recording IR spectra (32 scans) every 30 s. The concentrations
6e6
1.2e5
4e4
(41 amu)
1-butene
butane (43 amu)
NiPdPd(111)
(2 amu)
H2
(c)
(b)
(a)
NiPdPd(111)
Pd(111)
NiPdPd(111)
Pd(111)
Pd(111)
PdNiPd(111)
PdNiPd(111)
PdNiPd(111)
200 300 400 500 600 700
T (K)
200 300 400 500 600 700
T (K)
200 300 400 500 600 700
T (K)
Fig. 3. TPD spectra of (a) H2 (b) 1-butene, and (c) butane following 2.0 L predosed H2 and 2.0 L 1,3-butadiene on NiPdPd(111), Pd(111) and PdNiPd(111) surfaces.
Table 2
CO chemisorptions results and first-order consumption rate constants for the hydrogenation of 1,3-butadiene over the monometallic and bimetallic catalysts.
Catalysts
Average particle size
(nm)
CO uptake
mol gꢂ1 cat)
Normalized kr
(by weight)
Normalized kr
Normalized kr
(
l
(by CO uptake)
(by total metal loading)
(minꢂ1 gꢂ1 cat)
(10ꢂ2 minꢂ1 mol COꢂ1
l )
(10ꢂ3 minꢂ1 gꢂ1 metal)
PdNi/
c
-Al2O3
5.9
5.1
5.2
29.3
15.2
12.3
2.32
1.08
0.038
7.92
7.11
0.31
9.59
11.87
0.25
Pd/c
-Al2O3
Ni/c
-Al2O3

R. Hou et al. / Journal of Catalysis 316 (2014) 1–10
5
2.5.2. Flow reactor
observed from NiPdPd(111). In Fig. 3c, no butane desorption peak
is observed on the three surfaces, suggesting that there is no
butane formed during the TPD experiments. The TPD results dem-
onstrate that the PdNiPd(111) bimetallic surface shows the high-
est activity for 1,3-butadiene hydrogenation, while the Pd(111)
and NiPdPd(111) surfaces are not active. Furthermore, no butane
formation is observed on the PdNiPd(111) surface, suggesting that
the Pd-terminated bimetallic structure should have both high
activity and selectivity toward 1-butene.
Previous studies have revealed the potential effect of hydrides
on hydrogenation, as demonstrated previously on the Pd8Ni92 sur-
face. For example, Valcarcel et al. [47] have demonstrated that
hydrogen dissolution is a major reason for lower activity on Pd sur-
faces in a gas-phase static reactor. Additionally, according to Rup-
prechter and Somorjai [48], although the dissolved hydrogen can
be available for surface hydrogenation, the reaction rate with dis-
solved hydrogen is much lower than with the gas-phase hydrogen.
The TPD spectra in Fig. 3 suggest that hydride formation might not
be the reason for the lack of 1,3-butadiene hydrogenation on
Pd(111).
As shown in Fig. 3a, two H2 desorption peaks are observed at
ꢀ300 K and ꢀ400 K from both Pd(111) and PdNiPd(111) surfaces,
which correspond to the subsurface hydrogen and surface hydro-
gen, respectively [49,50]. The onset of H2 desorption occurred at
around 200 K from both Pd(111) and PdNiPd(111) surfaces, which
was in the temperature range of 1,3-butadiene hydrogenation on
PdNiPd(111). The desorption temperatures for both the subsurface
(ꢀ300 K) and surface (ꢀ400 K) remain similar on the two surfaces.
The peak area of the subsurface hydrogen on PdNiPd(111) is
The catalyst was mixed with quartz particles in a ratio of 1:29
(100 mg catalyst and 2.9 g quartz particles) in a fixed bed reactor.
Prior to the reaction, the catalysts were reduced in pure hydrogen
at 723 K for 3 h and then cooled to reaction temperature. The total
flow rate of 1,3-butadiene and hydrogen was 9.6 mL/min, and the
1,3-butadiene to hydrogen ratio was 1:2.2 or 1:4. The reaction
temperatures studied were from 313 to 373 K at intervals of
ꢀ20 K. The catalyst was left on stream for 1.5 h at each tempera-
ture to achieve steady state. The outlet stream was analyzed online
by a gas chromatograph equipped with a FID detector (GC 7900,
Techcomp Ltd.) and an HP-Al/S PLOT column (Agilent,
30 m ꢁ 0.53
lm). Only the C4 species (1,3-butadiene, 1-butene,
trans-2-butene, cis-2-butene, and butane) were observed in the
outlet, with the gas-phase carbon balance above 95%. The selectiv-
ity of each product was calculated by dividing its concentration by
the total products. The total butene selectivity was obtained by
summing the selectivities of all butene species.
3. Results
3.1. Model surface results
3.1.1. DFT calculations
DFT calculations were performed on Pd(111) and PdNiPd(111)
surfaces to compare the adsorption and reaction network of 1,3-
butadiene. The adsorption geometries of 1,3-butadiene and its
hydrogenation products used in this work are the same as those
reported on Pd(111) in the literature [45], and the geometries on
PdNiPd(111) are shown in Fig. 2. As shown in the activation barrier
of each step (Scheme 1), the hydrogenation of 1,3-butadiene to 1-
butene on Pd(111) needs to overcome a relatively high activation
barrier of 1.04 eV for the first step and 0.92 eV for the second step.
In comparison, lower reaction barriers are required on the
PdNiPd(111) surface, with barriers for the first and second steps
being 0.68 eV and 0.88 eV, respectively. The lower activation barri-
ers on PdNiPd(111) suggest higher hydrogenation activity on the
bimetallic surface.
The binding energies of 1,3-butadiene and its hydrogenation
products are summarized in Table 1. The PdNiPd(111) surface
exhibits a surface d-band center further away from the Fermi level
and correspondingly lower binding energies for the adsorbed mol-
ecules. On the PdNiPd(111) surface, 1-butene shows a very low
binding energy (ꢂ0.13 eV), indicating that 1-butene should readily
desorb from the surface after its formation rather than undergo
further hydrogenation. This suggests that 1,3-butadiene should
be selectively hydrogenated to 1-butene rather than to butane.
Using DFT calculations, Gomez et al. found that the binding energy
of 1-butene was significantly lower on Pd1Ni3(111) than on
Pd(111) and Pd3Ni1(111) surfaces [46]. They also proposed that
the Pd1Ni3(111) structure should give higher 1-butene selectivity.
3.1.2. TPD results
To verify the DFT results, TPD experiments of 1,3-butediene
hydrogenation were performed on the Pd(111) and PdNiPd(111)
surfaces. For comparison, the Ni-terminated, NiPdPd(111) surface
was also studied. Fig. 3 shows the TPD results following the reac-
tion of 1,3-butadiene with pre-adsorbed atomic hydrogen on the
three surfaces. Fig. 3a shows the TPD spectra of hydrogen on the
three surfaces. The hydrogen desorption peak areas indicate the
amount of unreacted hydrogen on each surface. In Fig. 3b, a
desorption peak of 1-butene is observed around 230 K on the
PdNiPd(111) surface, indicating the formation of 1-butene from
the hydrogenation of 1,3-butadiene. In comparison, only a small
peak around 230 K is observed from Pd(111), and no peak is
Fig. 4. HAADF TEM micrographs and particle size distributions for the catalysts: (a)
PdNi/c-Al2O3, (b) Pd/c-Al2O3, (c) Ni/c-Al2O3.

6
R. Hou et al. / Journal of Catalysis 316 (2014) 1–10
5
Fresh
(a)
After H2 Treatment
PdNi/γ-Al2O3
4
3
Pd/γ-Al2O3
2
Pd Foil
1
0
24200
24400
24600
24800
25000
Energy (eV)
5
4
3
2
1
0
5
(b)
(c)
Fresh
Fourier Transform
First Shell Fit
After H2 Treatment
4
3
2
1
0
PdNi/γ−Al2O3
PdNi/γ-Al2O3
Pd/γ−Al2O3
Pd/γ-Al2O3
0
1
2
3
4
5
6
0
1
2
3
4
5
6
R (Å)
R (Å)
Fig. 5. Pd K-edge XANES spectra before and after reduction (a), Fourier transformed (magnitude) k2 weighted EXAFS function (
v(k)) before and after reduction (b), Pd K-edge
Fourier transformed (magnitude) k2 weighted EXAFS function (
v(k)) after reduction with first shell fit (c). The Pd foil is included in (a) to serve as a reference for metallic Pd.
Table 3
alkene hydrogenation [51]. Similar strain effects should also play
a role in the higher hydrogenation activity on PdNiPd(111).
Summary of EXAFS analysis of Pd K-edge catalysts supported on
c
-Al2O3.
PdNi/
Catalysts
Pd/
c
-Al2O3
c-Al2O3
N(PdAPd)
N(PdANi)
9.77 0.38
8.34 0.74
1.44 0.50
2.79 0.01
2.64 0.01
0.010 0.001
0.016 0.004
3.2. Characterization of supported catalysts
/
R(PdAPd) (Å)
R(PdANi) (Å)
2.83 0.002
The CO uptake values are 29.3, 15.2, and 12.3
PdNi/ -Al2O3, Pd/ -Al2O3, and Ni/ -Al2O3, respectively, as shown
l
mol gꢂ1 for
/
r
r
2(PdAPd) (Å2)
2(PdANi) (Å2)
0.008 0.001
/
c
c
c
in Table 2. The high-angle annular dark field (HAADF) TEM images
of the catalysts are shown in Fig. 4. The particle size distributions
were calculated by measuring horizontal particle diameters in sev-
eral different images for each catalyst. The three catalysts have
similar particle size distributions, as shown in Fig. 4 and Table 2.
Fig. 5a shows the Pd K-edge X-ray absorption near-edge struc-
reduced by only about 25% from that of Pd(111), indicating that
the formation of bimetallic surface does not significantly suppress
hydride formation compared to Pd(111). Significant amounts of
subsurface and surface hydrogen are present on Pd(111) and are
available to react with 1,3-butadiene if Pd(111) has comparable
activity with PdNiPd(111). These results suggest that the lower
hydrogenation activity on the Pd(111) surface is not entirely due
to hydride formation.
ture (XANES) spectra of the Pd/c-Al2O3 and PdNi/c-Al2O3 catalysts
before and after reduction, along with the spectra of the Pd foil for
reference. The background-subtracted, edge-step normalized and
k2-weighted Pd K-edge EXAFS data (
v(k)) are shown in R-space
Moreover, as pointed out in the previous studies by Filhol et al.,
strain relaxation effects in the PdANi surfaces also contributed to
in Fig. 5b, along with the fits obtained using FEFF6 theory shown
in Fig. 5c.

R. Hou et al. / Journal of Catalysis 316 (2014) 1–10
7
1.4
(a)
1.2
1.51% Ni/γ−Al2O3
1.0
0.8
0.91% Pd/γ−Al2O3
0.6
0.4
0.91% Pd 1.51% Ni/γ−Al2O3
0.2
0.0
0
20
40
60
80 100 120
Time, min
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
1.4
(c)
(b)
1.2
0.91% Pd 1.51% Ni/γ−Al2O3
1.0
0.8
0.6
0.4
0.2
0.0
0.91% Pd 1.51% Ni/γ−Al2O3
0.91% Pd /γ−Al2O3
0.91% Pd/γ−Al2O3
1.51% Ni/γ−Al2O3
1.51% Ni/γ−Al2O3
0
20
40
60
80 100 120
0
20
40
60
80 100 120
Time, min
Time, min
Fig. 6. The consumption of (a) 1,3-butadiene, production of (b) 1-butene, and production of (c) n-butane during the hydrogenation of 1,3-butadiene at 308 K.
Careful examination of the Pd K-edge XANES and EXAFS reveals
information regarding the extent of oxidation of the catalysts. Rel-
atively intense peaks are present in Fig. 5a at 24,350 eV prior to
reduction and are attributed to the PdAO bond, which is confirmed
by the peaks at low radial distribution in Fig. 5b. After reduction in
H2, the spectra are more similar to that of the Pd foil in Fig. 5a, and
the PdAO peaks are not present in Fig. 5b. The PdAO peaks are
replaced by PdANi peaks at slightly larger values of R. Together,
these changes in the spectra indicate that Pd is in a metallic state
after reduction in H2.
Fig. 5c presents the Fourier transformed (R-space) data of the
reduced catalysts and their associated fits from FEFF6. Fits were
obtained by only including PdAPd and PdANi contributions. The
results from the fitting procedure are shown in Table 3. For both
the Pd/c-Al2O3 and PdNi/c-Al2O3 catalysts, the PdAPd distances
(2.83 Å and 2.79 Å, respectively) are larger than the bulk PdAPd
related to the strain of the larger Pd atoms residing on top of smal-
ler Ni atoms. A similar effect was observed previously from EXAFS
studies of the PtNi/c-Al2O3 catalyst [24]. The EXAFS results confirm
that PdANi bimetallic bonds are formed in the catalysts because
the bimetallic bond lengths are between that of either parent
metal. In addition to the observed attenuation in bond lengths,
the CN of PdANi is 1.44, which indicates that bimetallic bonds
are present in the PdANi/c-Al2O3 catalyst. During data fitting, good
fits to the Pd K-edge experimental data could only be obtained by
including both PdAPd and PdANi contributions in the model,
which further suggest that bimetallic bonds are present.
The overall coordination number is 9.77 on Pd/
c-Al2O3 and 9.78
on PdNi/ -Al2O3 (N(PdAPd) + N(PdANi)). The similarity between
c
the coordination numbers indicates that the metallic particles on
each catalyst should be of similar size, consistent with the particle
size statistics from TEM images as listed in Table 2. In the bimetal-
lic catalyst, the ratio of PdAPd to PdANi coordination numbers
(CN(PdAPd):CN(PdANi) = 8.34:1.44 = 5.79:1) is much larger than
the atomic ratios of Pd:Ni (1:3). This implies that a considerable
amount of remaining Ni atoms is located as monometallic Ni nano-
distance (2.75 Å), possibly due to the formation of Pd hydride
and the expansion of PdAPd bond [52,53]. For the PdNi/c-Al2O3
catalyst, the PdAPd distance is 2.79 Å, which is 0.04 Å smaller than
the monometallic PdAPd distance (2.83 Å), while the PdANi dis-
tance is 2.64 Å, which is 0.15 Å larger than the metallic NiANi dis-
tance (2.49 Å). These changes in bond distance are most likely
particles, dispersed across the
form of NiAl2O4.
c-Al2O3 support, or present in the

8
R. Hou et al. / Journal of Catalysis 316 (2014) 1–10
3.3. Catalytic evaluation of supported catalysts
Fig. 8. The total butene selectivity is 100% at conversions below
60% over the three catalysts.
3.3.1. Batch reactor
Moreover, the selectivities to 1-butene, trans-2-butene, and cis-
Fig. 6 shows the changes in gas-phase concentrations of 1,3-
butadiene, 1-butene, and n-butane with reaction time over the
PdNi/c-Al2O3, Pd/c-Al2O3, and Ni/c-Al2O3 catalysts. The PdANi
bimetallic catalyst exhibits a higher hydrogenation activity than
both the Pd and Ni monometallic catalysts, as illustrated in the
1,3-butadiene consumption rates in Fig. 6a, which is consistent
with the model surface studies. After two hours of reaction, 1,3-
2-butene are different over the three catalysts. The PdNi/c-Al2O3
catalyst is ꢀ20% more selective to 1-butene at conversions below
60% than its monometallic counterparts. As shown in Fig. 8, the
steady-state product distribution showed the following trend:
PdNi/
Pd/ -Al2O3 > Ni/
Ni/ -Al2O3 > Pd/
selectivity for 1-butene formation over PdNi/c
c
-Al2O3 > Pd/
-Al2O3 > PdNi/
-Al2O3 ꢀ PdNi/ -Al2O3 for cis-2-butene. The enhanced
-Al2O3 is consistent
c
-Al2O3 ꢀ Ni/
c
-Al2O3 for 1-butene production,
c
c
c
c
c-Al2O3 for trans-2-butene, and
c
butadiene is completely converted over PdNi/
c
-Al2O3, ꢀ80% over
Pd/ -Al2O3. Fig. 6b and c show once
c
-Al2O3, and only ꢀ7% over Ni/
c
with the DFT and TPD results that the bimetallic structure
promotes the production and subsequent desorption of 1-butene.
However, more studies are needed to understand the trends for
the cis- and trans-2-butene production.
again that the bimetallic catalyst outperforms the monometallic Pd
and Ni catalysts. A broad peak of 1-butene appears between 40 and
60 min over PdNi/c-Al2O3, which indicates the possibility of pro-
ducing the maximum amount of 1-butene within short contact
As 1-butene is the most desirable product from 1,3-butadiene
time on the bimetallic catalyst.
hydrogenation, PdNi/c-Al2O3 is superior than Pd/c-Al2O3 because
To make a quantitative analysis, the hydrogenation reaction
rates are estimated by a first-order rate law for the consumption
of 1,3-butadiene and are normalized by catalyst weight, by CO
uptake, and by metal loading, as shown in Table 2. The rate con-
it has higher hydrogenation activity and better performance in
suppression of both complete hydrogenation and isomerization
reactions.
stant of PdNi/
-Al2O3, while that of Ni/
nitude. This suggests that Ni/
butadiene hydrogenation. The rate constants normalized by CO
uptake follow the same trend of PdNi/ -Al2O3 > Pd/ -Al2O3 > Ni/
-Al2O3. The rate constant of PdNi/ -Al2O3 normalized by CO
uptake is only slightly larger than that of Pd/ -Al2O3 and that of
Ni/ -Al2O3 is still smaller by orders of magnitude. The rate con-
stant of PdNi/ -Al2O3 normalized by metal loading is even smaller
than that of Pd/ -Al2O3. However, the normalizations by CO uptake
c
-Al2O3, normalized by weight, is twice that of Pd/
-Al2O3 is smaller by two orders of mag-
-Al2O3 is almost inactive for 1,3-
4. Discussion
c
c
c
Several parallels between model surfaces and supported cata-
lysts can be drawn based on the results presented above. In the
c
c
c
c
c
100
(a)
c
0.91% Pd 1.51% Ni/γ−Al2O3
c
0.91% Pd/γ−Al2O3
c
80
or metal loading are not very suitable in the current case. Accord-
ing to the EXAFS results, not all Ni atoms are alloyed with Pd on the
bimetallic catalyst. Some Ni particles are isolated on the support.
These Ni particles have CO uptake values but barely contribute to
hydrogenation activity. Therefore, the rate constants normalized
by weight are more suitable for the activity comparison among
1.51% Ni/γ−Al2O3
60
40
20
0
the different catalysts. The higher rate constant of PdNi/c-Al2O3
verifies the prediction of higher hydrogenation activity in the
TPD results and the lower activation barrier in the DFT calculations
on the Pd-terminated PdNiPd(111) surface. Previous DFT calcula-
tions of PdANi bimetallic systems have indicated that, in the pres-
ence of adsorbed hydrogen, the Pd-terminated surface is the
thermodynamically stable structure [40,54], which contributes to
the good correlation between the supported catalysts and model
surfaces.
315
330
345
360
375
Temperature (K)
100
80
60
40
20
0
(b)
3.3.2. Flow reactor
0.91% Pd 1.51% Ni/γ−Al2O3
0.91% Pd/γ−Al2O3
To verify the batch reactor results and to test the catalytic selec-
tivity under steady-state conditions, 1,3-butadiene hydrogenation
was performed in a flow reactor at different temperatures from
315 to 375 K. Higher temperature evaluations of Ni/
measured to obtain selectivity over a broad range of conversions
to compare selectivity with Pd/ -Al2O3 and PdNi/ -Al2O3. As
1.51% Ni/γ−Al2O3
c-Al2O3 were
c
c
shown in Fig. 7, the trend in the conversion of 1,3-butadiene at dif-
ferent temperatures is consistent with the batch reactor results.
For both H2/C4H6 ratios, PdNi/
Pd/ -Al2O3 at all temperatures, while Ni/
genation activity below 373 K.
c
-Al2O3 exhibits higher activity than
c
c
-Al2O3 shows no hydro-
To further evaluate the selective hydrogenation of 1,3-butadi-
ene on different catalysts, the selectivities to total butene,
1-butene, trans-2-butene, and cis-2-butene were plotted at
conversions of 10% and 60% in Fig. 8. The butene selectivities are
independent of the H2/C4H6 ratios. Similar product distribution
and product selectivities are obtained for H2/C4H6 = 2.2 and for
H2/C4H6 = 4, so only selectivities at H2/C4H6 = 2.2 are shown in
310
320
330
340
350
360
Temperature (K)
Fig. 7. Conversion of 1,3-butadiene in flow reactor at different temperatures. (a)
H2:C4H6 = 2.2:1; (b) H2:C4H6 = 4:1.

R. Hou et al. / Journal of Catalysis 316 (2014) 1–10
9
of Ni is not as significant in supported catalysts as suggested from
surface science studies. This is most likely due to the relatively low
degree of bonding between Ni and Pd in the supported catalysts.
According to the EXAFS results, the coordination number of PdANi
is only 1.44, which is much lower than common bimetallic cata-
lysts [55]. A similar coordination number of 1.35 is found on
(a)
100
80
60
40
20
0
Total Butene
1-butene
trans-2-butene
cis-2-butene
0.18%Pd–1.5%Ni/c-Al2O3 (atomic ratio 1:15; EXAFS results not
shown), indicating the difficulty of Ni insertion into Pd nanoparti-
cles. To better understand the bimetallic effect and to promote the
formation of PdANi bimetallic bonds, PdANi nanoparticle struc-
tures on other oxide supports should be further explored.
5. Conclusions
Ni
Pd
PdNi
Hydrogenation of 1,3-butadiene has been studied on PdANi
bimetallic catalysts using a combination of DFT calculations and
TPD experiments on model surfaces and reactor evaluation over
supported catalysts. The results show that the PdANi bimetallic
structure has higher hydrogenation activity than its monometallic
counterparts. The activity of the catalysts follows the trend of
PdANi > Pd > Ni. The selectivity is evaluated with supported cata-
lysts in both batch and flow reactors. The PdANi, Pd and Ni cata-
(b)
100
80
60
40
20
0
Total butene
1-butene
trans-2-butene
cis-2-butene
lysts supported on
butenes, while PdANi/
ity at similar conversions. The high hydrogenation activity and 1-
butene selectivity make PdNi/ -Al2O3 a better catalyst for 1,3-
c-Al2O3 show similar total selectivity to
c-Al2O3 gives the highest 1-butene selectiv-
c
butadiene removal than the Pd monometallic catalyst. The good
correlation between model surfaces and supported catalysts dem-
onstrates the feasibility of designing selective hydrogenation cata-
lysts using well-defined single crystal surfaces.
Ni
Pd
PdNi
Fig. 8. Selectivities of total butene, 1-butene, trans-2-butene, and cis-2-butene in
flow reactor at conversion of (a) 10% (b) 60%. (H2:C4H6 = 2.2:1, similar product
distribution is obtained for H2:C4H6 = 4:1).
Acknowledgments
The work carried out at Brookhaven National Lab was spon-
sored under Contract No. DE-AC02-98CH10886 with the U.S.
Department of Energy, Office of Science, Division of Chemical Sci-
ences. Computational resources from the Center for Functional
Nanomaterials at Brookhaven National Lab were used for DFT cal-
culations. Ruijun Hou was sponsored by China Scholarship Council.
T.F. Wang was supported by Program for New Century Excellent
Talents in University of China (NCET-12-0297).
TPD experiments of model surfaces, the Pd-terminated
PdNiPd(111) surface shows much higher activity for 1,3-butadiene
hydrogenation than Pd(111) and NiPdPd(111) surfaces. This trend
is consistent with DFT calculations, which show that the activation
barrier for each hydrogenation step is lower on PdNiPd(111) than
on Pd(111). The catalytic evaluation of the supported catalysts in
both batch and flow reactors shows that PdNi/c-Al2O3 exhibits
higher activity than its monometallic catalysts, demonstrating an
excellent correlation between model surfaces and supported cata-
lysts. The selectivity also shows similarities between model sur-
faces and supported catalysts. The PdNiPd(111) surface does not
produce butane from 1-butene under the UHV-TPD experimental
conditions. DFT calculations show that 1-butene has much lower
binding energy on PdNiPd(111) than on Pd(111), indicating a
higher 1-butene selectivity due to its facile desorption from the
PdNiPd(111) surface. Both DFT calculations and TPD experiments
predict that the Pd-terminated bimetallic structure should give
higher 1-butene selectivity than the monometallic catalysts. This
is validated by the evaluation of supported catalysts using the flow
References
[1] A. Valcarcel, A. Clotet, J.M. Ricart, F. Delbecq, P. Sautet, J. Phys. Chem. B 109
(2005) 14175.
[2] B.C. Khanra, Indian J. Chem. A 35 (1996) 174.
[3] P. Kripylo, F. Turek, K.D. Hempe, H. Kirmse, Chem. Tech. (Leipzig) 27 (1975)
675.
[4] J.C. Bertolini, J. Massardier, B. Tardy, Abstr. Pap. Am. Chem. S. 199 (1990) 150.
[5] P.R.N. Souza, M.M. Pereira, O.A.C. Antunes, D.A.G. Aranda, J.W.M. Carneiro,
Braz. J. Chem. Eng. 19 (2002) 187.
[6] N. Kubota, T. Masui, H. Kita, Electrochim. Acta 28 (1983) 1663.
[7] A. Sarkany, G. Stefler, J.W. Hightower, Appl. Catal. A: Gen. 127 (1995) 77.
[8] M. Komiyama, K. Ohashi, Y. Morioka, J. Kobayashi, Bull. Chem. Soc. Jpn. 70
(1997) 1009.
[9] W.W. Lonergan, X.J. Xing, R.Y. Zheng, S.T. Qi, B. Huang, J.G. Chen, Catal. Today
160 (2011) 61.
[10] J.C. Bertolini, A. Cassuto, Y. Jugnet, J. Massardier, B. Tardy, G. Tourillon, Surf. Sci.
349 (1996) 88.
[11] A. Sarkany, Appl. Catal. A: Gen. 165 (1997) 87.
[12] J.W. Hightower, B. Furlong, A. Sarkany, L. Guczi, Stud. Surf. Sci. Catal. 75 (1993)
2305.
[13] H. Miura, M. Terasaka, K. Oki, T. Matsuda, Stud. Surf. Sci. Catal. 75 (1993) 2379.
[14] N. El Kolli, L. Delannoy, C. Louis, J. Catal. 297 (2013) 79.
[15] A.C. Krauth, G.H. Bernstein, E.E. Wolf, Catal. Lett. 45 (1997) 177.
[16] L. Piccolo, A. Piednoir, J.C. Bertolini, Surf. Sci. 592 (2005) 169.
[17] L. Constant, P. Ruiz, M. Abel, Y. Robach, L. Porte, J.C. Bertolini, Top. Catal. 14
(2001) 125.
[18] B.K. Furlong, J.W. Hightower, T.Y.L. Chan, A. Sarkany, L. Guczi, Appl. Catal. A:
Gen. 117 (1994) 41.
[19] M. Kang, M.W. Song, K.L. Kim, React. Kinet. Catal. Lett. 75 (2002) 177.
reactor, where PdNi/c-Al2O3 shows a higher 1-butene selectivity
over Pd/ -Al2O3 at similar 1,3-butadiene conversions.
c
The results in the present work successfully bridge the ‘‘pres-
sure gap’’ and ‘‘materials gap’’ between surface science studies
under UHV conditions and reactor evaluations of supported cata-
lysts under catalytic conditions. Equally important, this paper
reports a more effective catalyst for the selective hydrogenation
of 1,3-butadiene. The PdNi/c-Al2O3 catalyst has both higher hydro-
genation activity and higher selectivity to 1-butene, which is the
most desirable product in the purification of butane streams.
However, an inconsistency has been found between model sur-
faces and supported catalysts. The activity increase by the addition

10
R. Hou et al. / Journal of Catalysis 316 (2014) 1–10
[20] M.M. Pereira, F.B. Noronha, M. Schmal, Catal. Today 16 (1993) 407.
[21] C. Breinlich, J. Haubrich, C. Becker, A. Valcarcel, F. Delbecq, K. Wandelt, J. Catal.
251 (2007) 123.
[22] K. Pattamakomsan, E. Ehret, F. Morfin, P. Gelin, Y. Jugnet, S. Prakash, J.C.
Bertolini, J. Panpranot, F.J.C.S. Aires, Catal. Today 164 (2011) 28.
[23] M.S. Zina, A. Ghorbel, in: E. VanSteen, M. Claeys, L.H. Callanan (Eds.), Recent
Advances in the Science and Technology of Zeolites and Related Materials, Pts
A–C, 2004, p. 2364.
[37] M. Carazzolle, S. Maluf, A. de Siervo, P. Nascente, R. Landers, G. Kleiman, J.
Electron Spectrosc. Relat. Phenom. 156 (2007) 405.
[38] A.C. Michel, L. Lianos, J.L. Rousset, P. Delichere, N.S. Prakash, J. Massardier, Y.
Jugnet, J.C. Bertolini, Surf. Sci. 416 (1998) 288.
[39] J.R. Kitchin, N.A. Khan, M.A. Barteau, J.G. Chen, B. Yakshinskiy, T.E. Madey, Surf.
Sci. 544 (2003) 295.
[40] W. Yu, M.D. Porosoff, J.G. Chen, Chem. Rev. 112 (2012) 5780.
[41] J. Fu, X. Yang, B.E. Koel, Unpublished Results.
[24] W.W. Lonergan, D.G. Vlachos, J.G. Chen, J. Catal. 271 (2010) 239.
[25] P. Miegge, J.L. Rousset, B. Tardy, J. Massardier, J.C. Bertolini, J. Catal. 149 (1994)
404.
[26] A.C. Michel, L. Lianos, J.L. Rousset, P. Delichere, N.S. Prakash, J. Massardier, Y.
Jugnet, J.C. Bertolini, Surf. Sci. 416 (1998) 288.
[42] M. Newville, J. Synchrot. Radiat. 8 (2001) 322.
[43] B. Ravel, M. Newville, J. Synchrot. Radiat. 12 (2005) 537.
[44] J.J. Rehr, R.C. Albers, Rev. Mod. Phys. 72 (2000) 621.
[45] A. Valcárcel, A. Clotet, J.M. Ricart, F. Delbecq, P. Sautet, J. Phys. Chem. B 109
(2005) 14175.
[27] P. Hermann, B. Tardy, Y. Jugnet, D. Simon, J.C. Bertolini, Catal. Lett. 36 (1996) 9.
[28] C.M. Ammann, G.A. Meehl, W.M. Washington, C.S. Zender, Geophys. Res. Lett.
30 (2003) 1657.
[29] G. Kresse, J. Furthmuller, Phys. Rev. B 54 (1996) 11169.
[30] G. Kresse, J. Furthmuller, Comput. Mater. Sci. 6 (1996) 15.
[31] G. Kresse, J. Hafner, Phys. Rev. B 47 (1993) 558.
[46] G. Gomez, P.G. Belelli, G.F. Cabeza, N.J. Castellani, J. Solid State Chem. 183
(2010) 3086.
[47] A. Valcarcel, F. Morfin, L. Piccolo, J. Catal. 263 (2009) 315.
[48] G. Rupprechter, G.A. Somorjai, Catal. Lett. 48 (1997) 17.
[49] S. Shaikhutdinov, M. Heemeier, M. Bäumer, T. Lear, D. Lennon, R. Oldman, S.
Jackson, H. Freund, J. Catal. 200 (2001) 330.
[32] J.P. Perdew, J.A. Chevary, S.H. Vosko, K.A. Jackson, M.R. Pederson, D.J. Singh, C.
Fiolhais, Phys. Rev. B 46 (1992) 6671.
[33] M.P. Teter, M.C. Payne, D.C. Allan, Phys. Rev. B 40 (1989) 12255.
[34] W.T. Yu, M.A. Barteau, J.G. Chen, J. Am. Chem. Soc. 133 (2011) 20528.
[35] P.A. Nascente, M. Carazzolle, A. de Siervo, S. Maluf, R. Landers, G. Kleiman, J.
Mol. Catal. A: Chem. 281 (2008) 3.
[50] D. Farias, M. Patting, K. Rieder, Phys. Status Solidi A 159 (1997) 255.
[51] J.S. Filhol, D. Simon, P. Sautet, J. Am. Chem. Soc. 126 (2004) 3228.
[52] M.W. Tew, M. Janousch, T. Huthwelker, J.A. van Bokhoven, J. Catal. 283 (2011)
45.
[53] R. Hou, T. Wang, X. Lan, Ind. Eng. Chem. Res. 52 (2013) 13305.
[54] C.A. Menning, J.G. Chen, J. Chem. Phys. 130 (2009).
[55] M.D. Porosoff, J.G. Chen, J. Catal. 301 (2013) 30.
[36] P.J. Cumpson, M.P. Seah, Surf. Interface Anal. 25 (1997) 430.