Effect of oxide supports on Pd-Ni bimetallic catalysts for 1,3-butadiene hydrogenation

Article abstract of DOI:10.1016/j.apcata.2014.11.001

Our previous work showed that the Pd-Ni bimetallic catalyst had better hydrogenation activity and 1-butene selectivity in the selective hydrogenation of 1,3-butadiene. In the present work, the effect of oxide supports on the hydrogenation of 1,3-butadiene was studied over Pd-Ni bimetallic catalysts supported on γ-Al₂ O₃, SiO₂, CeO₂, ZrO₂ and TiO₂. Transmission electron microscopy (TEM) and extended X-ray absorption fine structure (EXAFS) were used to characterize the particle size and the extent of Pd-Ni bimetallic bond formation, respectively. The support effect affected the bimetallic particle structure, but appeared to have no apparent influence in the hydrogenation activity. All the bimetallic catalysts showed ～100% selectivity to butenes; however, the supports affected the 1-butene selectivity. PdNi/γ-Al2O3 showed the highest 1-butene selectivity of ～80% at 1,3-butadiene conversions lower than 50% but decreased rapidly at higher conversion. The different performance of the supports in 1-butene selectivity might be attributed to strong metal-support interaction (SMSI), the oxygen defects and the geometric/electronic effect on the catalysts.

Full text of DOI:10.1016/j.apcata.2014.11.001

Accepted Manuscript
Title: Effect of oxide supports on Pd–Ni bimetallic catalysts
for 1,3-butadiene hydrogenation
Author: Ruijun Hou Marc D. Porosoff Jingguang G. Chen
Tiefeng Wang
PII:
S0926-860X(14)00685-1
DOI:
Reference:
http://dx.doi.org/doi:10.1016/j.apcata.2014.11.001
APCATA 15094
To appear in:
Applied Catalysis A: General
Received date:
Revised date:
Accepted date:
26-8-2014
30-10-2014
1-11-2014
Please cite this article as: R. Hou, M.D. Porosoff, J.G. Chen, T. Wang, Effect of oxide
supports on PdndashNi bimetallic catalysts for 1,3-butadiene hydrogenation, Applied
Catalysis A, General (2014), http://dx.doi.org/10.1016/j.apcata.2014.11.001
This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.

100
100
TOF
1-Butene Selectivity
80
Al O3
60
2
1
0
1
SiO2
4
0
0
0
ZrO2
CeO2
TiO2
γ−Al O
SiO2
2
3
ZrO2
TiO2
CeO2
2
0.0028
0.0030
0.0032
0.0034
0
20
40
60
80
100
1/(T+273.14)
Conversion, %
Graphical abstract
1
Page 1 of 29

ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
Highlights
Pd-Ni is more selective for 1,3-butadiene hydrogenation to 1-butene than Pd
Support nature does not affect 1,3-butadiene hydrogenation activity
2 3
PdNi/γ-Al O is most selective to 1-butene at low 1,3-butadiene conversions
Support reducibility/oxide defect affect 1-butene selectivity
2
Page 2 of 29

Effect of oxide supports on Pd–Ni bimetallic catalysts for
,3-butadiene hydrogenation
1
a,b
b
b*
a*
Ruijun Hou , Marc D. Porosoff , Jingguang G. Chen , 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,
USA
*
Corresponding Authors: jgchen@columbia.edu (J. G. Chen);
wangtf@tsinghua.edu.cn (T. F. Wang).
3
Page 3 of 29

Abstract
Our previous work showed that the Pd-Ni bimetallic catalyst had better
hydrogenation activity and 1-butene selectivity in the selective hydrogenation of 1,3-
butadiene. In the present work, the effect of oxide supports on the hydrogenation of
1
,3-butadiene was studied over Pd-Ni bimetallic catalysts supported on γ-Al O , SiO ,
2
3
2
CeO , ZrO and TiO . Transmission electron microscopy (TEM) and extended X-ray
2
2
2
absorption fine structure (EXAFS) were used to characterize the particle size and the
extent of Pd-Ni bimetallic bond formation, respectively. The support effect affected
the bimetallic particle structure, but appeared to have no apparent influence in the
hydrogenation activity. All the bimetallic catalysts showed ~100% selectivity to
butenes; however the supports affected the 1-butene selectivity. PdNi/γ-Al O showed
2
3
the highest 1-butene selectivity of ~80% at 1,3-butadiene conversions lower than 50%
but decreased rapidly at higher conversion. The different performance of the supports
in 1-butene selectivity might be attributed to strong metal-support interaction (SMSI),
the oxygen defects and the geometric/electronic effect on the catalysts.
Keywords: Support effect; Pd-Ni bimetallic catalysts; Selective hydrogenation;
1
,3-butadiene.
4
Page 4 of 29

1
. Introduction
Selective hydrogenation is one of the critical reactions to remove alkynes and
dienes in the feedstock for polymer synthesis. The hydrocarbon streams from cracking
units typically contain ~1% alkynes or dienes, which must be reduced to below 10
ppm for polymerization processes because they poison the catalysts and degrade the
product quality [1] [2]. For example, 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. In this paper, 1,3-
butadiene was chosen as a probe molecule to identify catalysts for the selective
hydrogenation of one of the C=C bonds in dienes.
In our previous work on single crystal surfaces, the Pd-terminated, PdNiPd(111)
surface was demonstrated to be more active and more selective to 1-butene than the
Pd(111) surface for the hydrogenation of 1,3-butadiene using ultra-high vacuum
(
UHV) temperature programmed desorption (UHV-TPD) and density functional
theory (DFT) calculations. The better performance of the Pd-Ni bimetallic surface
was verified on γ-Al O supported catalysts [3]. However, the extent of Pd-Ni
2
3
bimetallic bond formation in PdNi/γ-Al O was relatively small. The Pd-Ni
2
3
coordination number was only 1.4 because most Ni atoms existed either in the form of
NiAl O or as Ni monometallic nanoparticles. It is very likely that the Pd-Ni
2
4
bimetallic effect would be more pronounced by increasing the extent of Pd-Ni
bimetallic bond formation. One possible way to increase the Pd-Ni coordination
2 3
number is using oxide supports other than γ-Al O on which the active metal would
5
Page 5 of 29

not form the spinel species.
Another motivation to study the support effect for the Pd-Ni bimetallic catalyst is
the importance to identify a proper support for industrial application, with respect to
the catalytic activity and selectivity. The interaction between the oxide supports and
metal particles was previously shown to have a significant effect on the catalyst
activity for hydrogenation reactions. For the hydrogenation of acetone over the Pt-Ni
bimetallic catalysts, Qi et al. [4] reported that PtNi/SiO had much higher activity than
2
PtNi/γ-Al O and PtNi/TiO . For the hydrogenation of 1,3-butadiene, Lonergan et al.
2
3
2
[5] reported that the activity of Pt-Ni catalysts were similar on γ-Al
2
O
3
2 3
, α-Al O and
low surface area (LSA)-ZrO , while Pt-Ni supported on high surface area (HSA)-ZrO2
2
showed much lower activity due to the strong metal-support interaction (SMSI). It
was also reported that Pt-Ni/TiO showed much lower activity than Pt-Ni/γ-Al O due
2
2
3
to the SMSI between the Pt particles and TiO [6].
2
According to the literature above, the SMSI effect inhibits the hydrogenation
activity of the Pt-based catalysts. In this work, the Pd-Ni bimetallic catalysts on
several commonly used oxide supports were synthesized for the hydrogenation of 1,3-
butadiene to study the support effect on the hydrogenation activity and selectivity.
Transmission electron microscopy (TEM), CO chemisorption and extended X-ray
absorption fine structure (EXAFS) were used to characterize the catalysts. The
bimetallic catalysts on different supports were further evaluated in a flow reactor
equipped with a gas chromatograph (GC).
2. Experimental
6
Page 6 of 29

2
.1. Preparation of supported catalysts
The Pd-Ni bimetallic catalysts were synthesized by incipient wetness
2
impregnation on five oxide supports: γ-Al
2
O
3
(Alfa Aesar, 255 m /g), SiO
2
(Alfa
2
2
Aesar, 120 m /g), ZrO (Alfa Aesar, 90 m /g), CeO (Sigma Aldrich, nanopowder,
2
2
2
particle size < 25 nm), and TiO (Alfa Aesar, Anatase, 15 nm APS Powder, 240 m /g).
2
The precursor solutions were prepared by dissolving Pd(NO ) ·2H O (Alfa Aesar)
3
2
2
3 2 2
and/or Ni(NO ) ·6H O (Alfa Aesar) in an amount of water just sufficient to fill the
pores of 3 g of the support. The precursor solution was then added to the support by
dropwise addition and was mixed thoroughly between droplets. The catalysts were
o
o
dried at 100 C for 10 h and calcined at 290 C for 2 h. The sample was heated to 100
o
o
o
o
C at 0.4 C/min and then to 290 C at 0.8 C/min. Co-impregnation was used for the
bimetallic catalysts to produce greater extent of bimetallic bond formation, as verified
previously for supported Pt-Ni catalysts [7]. The composition of the synthesized
catalysts was 1.51 wt% Ni–0.91 wt% Pd/γ-Al O , corresponding to a Ni:Pd atomic
2
3
ratio of 3:1. The monometallic Pd catalysts (0.91 wt%) were synthesized on γ-Al O3
2
and SiO to study the Pd-Ni bimetallic effect.
2
2
.2. Catalyst characterization
2.2.1. Transmission electron microscopy (TEM)
TEM measurements were performed using a JEOL2010F equipped with a
Schottky field emission gun at 200 keV. All TEM images were taken in the scanning
(
STEM) mode with a 12 nm camera length and a 1 nm diameter nanoprobe. The TEM
o
samples were prepared by first reducing the catalysts at 450 C for 1 hour, then
7
Page 7 of 29

grinding them into fine powders and suspending them in acetone. In order to fully
reduce the Pd-Ni bimetallic particles and to avoid the formation of larger metal
particles at higher temperatures, the reduction temperature was chosen based on the
TPR results (provided in Fig. R1 in the Supplementary Information). Droplets of the
suspension were placed onto a carbon coated copper grid. The grid was dried
overnight before loading the sample into the TEM. More than 100 particles were
collected for the analysis of particle size distribution.
2.2.2. Pulse CO chemisorption
The number of active sites on each bimetallic catalyst was measured through CO
uptake with a ChemBET (QuantaChrome). A quartz reactor was loaded with ~200 mg
o
of catalyst and reduced in 80 sccm of 10% H in He at 450 C for 1 h. After reduction,
2
o
the catalyst was held in He at 450 C for 30 min to remove the remaining H on the
2
surface of the catalysts. Then the catalyst was cooled to room temperature before
pulsing CO. The amount of CO flowing out of the reactor was measured with a
thermal conductivity detector. Each adsorbed CO molecule was assumed to
correspond to one active site, i.e. each Pd or Ni active site was hypothesized to absorb
one CO molecule, which provided a means to compare the number of active sites
among the various oxide-supported catalysts.
2.2.3. Extended X-ray absorption fine structure (EXAFS)
EXAFS measurements were used to confirm the presence of Pd-Ni bimetallic
bonds. Measurements of the Pd K-edge were performed on the X18B beamline at the
National Synchrotron Light Source (NSLS), Brookhaven National Laboratory. The
8
Page 8 of 29

catalyst samples were pressed into a pellet with a thickness twice the absorption
length to maximize the signal to noise ratio. The catalysts were reduced in-situ in 40
o
sccm of 50% H
2
in He at 450 C 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) [8, 9]. 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 [10]. The theoretical monometallic
photoelectron amplitude and phase were calculated for the bulk Pd fcc structure. The
2
passive electron reduction factor S was found to be 0.878 from fitting the Pd-foil
0
data. The seven parameters used in the fitting procedure were the correction to the
edge energy, the coordination numbers of the Pd-Pd and Pd-Ni bonds, corrections to
their model interatomic distances, and the mean square deviations in interatomic
distances (EXAFS Debye-Waller factors).
2
.3. Catalyst evaluation
Flow reactor experiments were performed to evaluate the hydrogenation activity
and selectivity of the supported catalysts for 1,3-butadiene hydrogenation. 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
o
reduced in pure hydrogen at 450 C 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-
9
Page 9 of 29

o
butadiene to hydrogen ratio was 1:4. The reaction temperature range was 30 – 110 C.
The catalyst was left on stream for 2 h at each temperature 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 μm). Only the C species (1,3-butadiene, 1-butene, trans-2-butene, cis-2-butene
4
and butane) were observed in the outlet, with the gas phase carbon balance above
5%. The selectivity of each product was calculated by dividing its concentration by
9
that of the total products. The butenes selectivity was obtained by summing the
selectivities of all butene species.
3. Results
3
.1. Catalyst characterization
High angle annular dark field (HAADF) TEM images are presented in Fig. 1 for
the Pd-Ni catalysts. The average particle sizes and the CO uptake values are
summarized in Table 1. Except Pd-Ni/CeO , all catalysts show similar average
particle sizes and broad particle size distributions, ranging from ~1 nm up to at least
5 nm; while PdNi/CeO has a much narrower particle size distribution and smaller
2
1
2
average particle size of ~3 nm, likely due to the partial alloying of Pd with Ce which
has been reported by Bernal et al.[11]. The CO uptake values are consistent with the
TEM results. The Pd-Ni catalysts supported on the investigated supports exhibit
similar CO uptake values, except that PdNi/CeO
2
has a much higher CO uptake.
_
_
Combining the TEM and CO uptake results, the values of CO uptake/D , where D is
p
p
the average particle size, are summarized in Table R1 in the Supplementary
10
Page 10 of 29

_
Information. As shown in the table, the CO uptake/D
p
values are similar on Al
2
O
3
,
SiO , TiO and ZrO , (1.4-2.7), while that on CeO is much higher (11.9).
2
2
2
2
Fig. 2(a) shows the Pd K-edge X-ray absorption near-edge structure (XANES)
spectra of the PdNi catalysts supported on γ-Al O , SiO , TiO and CeO [12] before
2
3
2
2
2
and after reduction, along with the spectra of the Pd foil for reference. The
2
background-subtracted, edge-step normalized and k -weighted Pd K-edge EXAFS
data (χ(k)) are shown in R-space in Fig. 2(b), along with the fits obtained using
FEFF6 theory shown in Fig. 2(c). The ZrO supported catalyst is not included because
2
of the poor quality of EXAFS data, likely due to the high background absorbance of
Zr (17998 eV) in the range of Pd (24350 eV).
Examination of the Pd K-edge XANES and EXAFS reveals information
regarding the extent of oxidation of the samples. Relatively intense peaks are present
in Fig. 2(a) at 24350 eV prior to reduction and are attributed to the Pd–O bond, which
is confirmed by the peaks at low radial distribution in Fig. 2(b). After reduction in H2,
the spectra are similar to that of the Pd foil in Fig. 2(a), and the Pd–O peaks are no
longer present in Fig. 2(b). The Pd–O peaks are replaced by Pd–Ni peaks at slightly
larger values of R. These changes in the spectra indicate that Pd is in a metallic state
after reduction in H2.
Fig. 2(c) presents the Fourier transformed (R-space) data of the reduced catalysts
and the associated fits from FEFF6 by including only Pd–Pd and Pd–Ni contributions.
The results from the fitting are shown in Table 2. The coordination numbers reveal
that bimetallic Pd-Ni bonds are formed in all four catalysts. Table 2 also compares the
11
Page 11 of 29

value of N(Pd-Ni)/[(N(Pd-Ni) + N(Pd-Pd)], which can be used to compare the general
trend in the extent of Pd-Ni bimetallic formation. The values are similar on
2 2 2 2 3
PdNi/CeO , PdNi/TiO and PdNi/SiO , with PdNi/γ-Al O showing a significantly
lower value. Such comparison indicates that the extent of Pd-Ni bimetallic formation
is much lower on γ-Al O than on the other three supports. The relatively small
2
3
amount of bimetallic bond formation on γ-Al O is because a large fraction of Ni
2
3
2 4
atoms form NiAl O with the supports. For all Pd-Ni catalysts, the EXAFS results
show that the Pd-Pd distances are smaller than the Pd-Pd distance on the
monometallic catalyst (2.83 Å), while the Pd-Ni distances are larger than the metallic
Ni-Ni distance (2.49 Å). These results suggest that Pd-Ni bimetallic bonds are formed
on the catalysts because the bimetallic bond lengths are between that of either parent
metal. In addition, good fits to the Pd K-edge experimental data could only be
obtained by including both Pd-Pd and Pd-Ni contributions in the model, strongly
suggesting that bimetallic bonds are present.
In principle, the total coordination numbers from the EXAFS measurements, such
as the sum of N(Pd-Pd) and N(Pd-Ni), should correlate with the TEM analysis of
particle size. However, quantitatively such correlation is valid only when the particles
are uniform, both in physical size and chemical composition. The TEM images reveal
that the metals particles have a relatively wide range of size distribution. Therefore
TEM results are more reliable and more direct than EXAFS for particle size
estimation. The TEM results are also consistent with the trend in the metal dispersion
from the CO chemisorption measurements. For example, among the bimetallic
12
Page 12 of 29

2
catalysts on the four oxide supports, PdNi/CeO , is characterized by the smallest
particle size in TEM images and the highest metal dispersion in CO chemisorption
measurements.
3
.2. Pd-Ni bimetallic effect
2 3 2
The Pd-Ni bimetallic effect was investigated on γ-Al O - and SiO -supported
monometallic and bimetallic catalysts. The Pd catalysts and Pd-Ni bimetallic catalysts
were evaluated for the selective hydrogenation of 1,3-butadiene at different
temperatures and the results are shown in Fig. 3. The conversions of 1,3-butadiene are
plotted as a function of temperature in Figs. 3(a) and (b), and the selectivities to
butenes and 1-butene are plotted as a function of conversion in Figs. 3(c) and (d),
respectively. The Pd-Ni bimetallic catalysts on both supports show higher conversions
and higher 1-butene selectivity than the Pd monometallic catalysts. The selectivity to
1
-butene is generally higher on the Pd-Ni bimetallic catalysts than on the Pd catalysts,
indicating that the formation of the Pd-Ni bimetallic bond plays a major role in the
enhancement of 1-butene selectivity. At conversions below 50%, the selectivity to 1-
butene is ~20% higher on PdNi/γ-Al O than on Pd/γ-Al O . Note that on PdNi/γ-
2
3
2
3
Al O the 1-butene selectivity decreases rapidly with increasing 1,3-butadiene
2
3
conversion above 50%, while on Pd/γ-Al O it decreases more slowly. Nevertheless
2
3
the 1-butene selectivity on PdNi/γ-Al O remains higher than that on Pd/γ-Al O at
2
3
2
3
2 3
conversions below 90% and is similar to that on Pd/γ-Al O at conversion above 90%.
The possible explanation could attribute to the local concentration of the C4 species.
At conversions above 90%, the concentration of 1-butene on PdNi/γ-Al O becomes
2
3
13
Page 13 of 29

very high and the 1,3-butadiene concentration becomes very low. The high
concentration ratio of 1-butene and 1,3-butadiene likely enhances the coverage of 1-
butene on the surface of Pd-Ni, leading to further hydrogenation and isomerization
reactions. As a result, the 1-butene selectivity decreases to the same value with Pd/γ-
Al O at high 1,3-butadiene conversions. Moreover, the difference between the Pd-Ni
2
3
and Pd catalysts is larger on SiO than on γ-Al O . The greater enhancement is most
2
2
3
likely from the different extent of Pd-Ni bimetallic bonding. As summarized in Table
, the coordination number of Pd-Ni on SiO is 3.5, which is about 2.5 times larger
2
2
than that on γ-Al O .
2
3
3
.3. Support effect
3.3.1. Hydrogenation activity
The Pd-Ni bimetallic catalysts on different supports were evaluated and the
activity results are summarized in Fig. 4. Fig. 4(a) shows the 1,3-butadiene conversion
on different catalysts as a function of reaction temperature. Based on catalyst weight,
the activity of the Pd-Ni bimetallic catalysts follows the trend of CeO > γ-Al O ~
2
2
3
SiO > ZrO > TiO . However, after normalizing by CO uptake, the turnover
2
2
2
frequencies (TOFs, calculated using the data with conversions below 20%) are similar
among the five catalysts and the logarithm of TOFs show a linear relationship with
the offset reciprocal of the reaction temperatures, except that the TOF value of
2
PdNi/CeO is lower than the average value. The results demonstrate that the
bimetallic catalysts on different supports display similar apparent hydrogenation
activity regardless of the support nature, suggesting the presence of similar bimetallic
14
Page 14 of 29

active sites on the different oxide supports.
3.3.2. Selectivity
The selectivities to different butene products on the Pd-Ni bimetallic catalysts are
shown in Fig. 5. Fig. 5(a) compares the selectivities to butenes on different supports.
The butenes selectivities are all close to 100% at conversions lower than 80%; when
the conversion approaches 100%, the selectivity decreases slightly over SiO , CeO ,
2
2
ZrO
2
and TiO
The selectivities to 1-butene, trans-2-butene and cis-2-butene are plotted as a
function of conversion in Figs. 5(b), (c) and (d), respectively. No isobutene was
detected during the reaction. Among all the catalysts, the γ-Al O supported catalyst
2 2 3
, but to a greater extent over PdNi/γ-Al O .
2
3
gives the highest 1-butene selectivity. At conversions lower than 25%, the selectivity
to 1-butene over the Pd-Ni bimetallic catalysts on different supports follows the trend
of γ-Al O > SiO > CeO ~ ZrO > TiO . With increasing conversion, the general
2
3
2
2
2
2
trend changes. The 1-butene selectivity on PdNi/ZrO decreases slightly when the
2
2
conversion increases to above 25% and it falls between that on PdNi/CeO and
PdNi/TiO . When the 1,3-butadiene conversion further increases to 100%, the
2
selectivity to 1-butene on γ-Al O supported catalysts decreases rapidly, while that on
2
3
SiO supported catalyst decreases only slightly.
2
The selectivity to trans-2-butene shown in Fig. 5(c) follows the trend of TiO >
2
2 2 2 2 3
CeO ~ ZrO > SiO > γ-Al O , which is opposite to that of the selectivity to 1-butene.
The selectivity to cis-2-butene is similar on different supports, as shown in Fig. 5(d).
Because the hydrogenation of 1,3-butadiene only produces 1-butene in a batch reactor
15
Page 15 of 29

as reported before [3], it is very likely that 1,3-butadiene is firstly hydrogenated to 1-
butene, and then 1-butene isomerizes to trans- and cis-2-butenes in the flow reactor.
The opposite trend in the selectivities to 1-butene and trans-2-butene suggests that
most 1-butene isomerizes into trans-2-butene. Based on the conclusion above, the
oxide support appears to play an important role in the isomerization of 1-butene to
trans-2-butene.
4. Discussion
4.1. Bimetallic effect
The better performance of the Pd-Ni bimetallic catalysts has been previously
demonstrated on single crystal surfaces by both density functional theory (DFT)
calculations and surface science experiments, and further verified on the γ-Al O
3
2
supported catalysts using both batch and flow reactors [3]. The present study further
confirms the trend by supporting Pd-Ni and Pd onto γ-Al O and SiO . The bimetallic
2
3
2
catalysts show enhancement in both hydrogenation activity and 1-butene selectivity.
Moreover, the larger coordination number of Pd-Ni on SiO indicates a greater extent
2
of Pd-Ni bond formation, which is consistent with the greater activity enhancement of
PdNi/SiO than PdNi/γ-Al O compared with their Pd monometallic counterparts.
2
2
3
4
.2. Support effect
4
.2.1. Activity
The characterization results from CO chemisorption and TEM are consistent in
that PdNi/CeO has higher CO uptake and more uniform small metal particles. The
2
other Pd-Ni supported catalysts show similar particle size distributions and similar
16
Page 16 of 29

CO uptake values. Normalized by CO uptake, the Pd-Ni bimetallic catalysts show
similar activity regardless of the support types except that PdNi/CeO shows a slightly
2
lower activity.
Boitiaux et al. [13] reported that the 1,3-butadiene hydrogenation activity
strongly depends on the metal dispersion on the Pd monometallic catalysts. The
hydrogenation activity is similar when the Pd metal dispersion is lower than 20% and
decreases with increasing metal dispersion at above 20%, regardless of the support
types (γ-Al O , α-Al O and SiO ). Goetz et al. [14] also reported that large Pd
2
3
2
3
2
particles (~6-7 nm) exhibits higher TOF than small Pd particles (~3 nm). These
conclusions are consistent with the results in the current study for the Pd-Ni bimetallic
catalysts. The investigated Pd-Ni bimetallic catalysts exhibit metal dispersions lower
than 20% except PdNi/CeO , and they display similar TOFs regardless of the support
2
types. The higher metal dispersion on PdNi/CeO results in slightly lower TOF, which
2
is likely due to the exposure of less active sites on small bimetallic nanoparticles [13].
The electronic effect of the small metal particles makes 1,3-butadiene bind more
strongly on the catalyst and accordingly decreases the hydrogenation activity [15].
However, the observation of similar TOF values on different oxide supports is
different from the previous study on Pt-Ni bimetallic catalysts [5, 6]. It has been
reported that the activity of Pt-Ni bimetallic catalysts normalized by CO uptake is
higher on γ-Al O than on ZrO [5] and TiO [6] for the hydrogenation of 1,3-
2
3
2
2
butadiene in a batch reactor. Moreover, Boitiaux et al. [16] and Primet et al. [17] have
demonstrated that the hydrogenation activity of 1,3-butadiene is insensitive to the
17
Page 17 of 29

metal dispersions for Pt catalysts. Because the different activities among the Pt-Ni
bimetallic catalysts [5, 6] are not from the metal dispersion[16, 17], they are possibly
from the strong metal-support interaction (SMSI) between the small Pt-Ni
nanoparticles and the oxide supports. Even though Pd-Ni might also have SMSI with
the corresponding oxide supports, the interaction of Pd with ZrO and TiO might be
2
2
different from that of Pt with ZrO and TiO .
2
2
4
.2.2. Selectivity
The selectivity to 1-butene over the Pd-Ni bimetallic catalysts on different
supports follows the trend of γ-Al O > SiO > CeO ~ ZrO > TiO at low conversion
2
3
2
2
2
2
of 1,3-butadiene. The different performance reflects the different support modification
on Pd-Ni bimetallic catalysts. However, the selectivity performance varies from low
to high 1,3-butadiene conversions.
It has been reported that the selectivities to butene species are independent of the
metal dispersion [13], therefore the nature of supports, as well as the interaction
between metal and support, should be responsible for the different performance in 1-
butene selectivity among Pd-Ni catalysts on different supports. Several support effects
could affect the selectivity of 1,3-butadiene hydrogenation, i.e., metal-support
interaction, oxide defect, and geometric/electronic effect.
SMSI has been reported to exist on the TiO , CeO and ZrO supported catalysts
2
2
2
o
if reduced at temperatures higher than 500 C [11, 18, 19], and that the SiO
2
and
Al O supported catalysts exhibit weaker metal-support interactions [19-21]. By
2
3
forming bimetallic metal bond, the required reduction temperature for SMSI could be
18
Page 18 of 29

reduced (see Fig. 3 in Ref. [22]). As a result, the Pd-Ni bimetallic catalysts on TiO
2
,
CeO and ZrO probably present SMSI. The results show that the catalysts on the
2
2
oxide supports with possible SMSI or possible oxide defects generally give lower 1-
buene selectivity. The trend of 1-butene selectivity indicates that the possible SMSI
effect decreases the 1-butene selectivity for the Pd-Ni bimetallic catalysts. The
difference in 1-butene selectivity might also come from the oxide defects. TiO and
2
2 2
CeO are reducible supports and high-surface area ZrO may have oxygen defects
during reaction. The apparent correlation between the oxygen defects and the lower 1-
butene selectivity suggests that the oxygen defects are also a possible reason for the
lower 1-butene selectivity.
Different support might also affect the geometric and electronic properties on the
Pd-Ni bimetallic catalysts. [23] According to the EXAFS results, the coordination
numbers of Pd-Ni on different supports are different, leading to different particle
composition on Pd-Ni bimetallic particles. On the other hand, TEM results revealed
that the particle sizes were similar on all supports except on CeO . The EXAFS and
2
TEM results suggest that the bimetallic composition, not the physical size of the
particles, plays a significant role for the selectivity in 1,3-butadiene hydrogenation.
5
. Conclusions
Hydrogenation of 1,3-butadiene has been studied on the Pd-Ni bimetallic
catalysts supported on γ-Al O , SiO , CeO , ZrO and TiO using a flow reactor. The
2
3
2
2
2
2
Pt-Ni bimetallic catalysts on γ-Al O and SiO are compared with the Pd catalysts to
2
3
2
study the bimetallic synergy effect. The results show that the Pd-Ni bimetallic
19
Page 19 of 29

catalysts have both higher hydrogenation activity and higher 1-butene selectivity than
the Pd catalysts. The bimetallic synergy effect is more significant on SiO than on γ-
2
Al
2
O
3
, i.e. the enhancement by Ni addition is greater on SiO
2
2 3
than on γ-Al O , which
is consistent with the greater extent of Pd-Ni bimetallic bond formation on SiO2.
The chemical nature of the supports influences the bimetallic metal particle
structures, but does not affect the hydrogenation activity. Only the metal particle size
affects the turnover frequency. On the other hand, the 1-butene selectivity depends on
the nature of supports. The possible SMSI effect and oxygen defects are possible
reasons for the lower 1-butene selectivity on PdNi/CeO , PdNi/ZrO and PdNi/TiO .
2
2
2
Acknowledgements
The work at Columbia University and Brookhaven National Laboratory was carried
out under Contract No. DE-AC02-98CH10886 with the U.S. Department of
Energy, Office of Basic Energy Sciences. R. Hou was sponsored by the China
Scholarship Council. T.F. Wang was supported by Program for New Century
Excellent Talents in University of China (NCET-12-0297).
20
Page 20 of 29

References
[
1] A. Hugon, L. Delannoy, J.M. Krafft, C. Louis, J. Phys. Chem. C 114 (2010)
0823-10835.
2] J.A. Alves, S.P. Bressa, O.M. Martinez, G.F. Barreto, Chem. Eng. J. 99 (2004) 45-
1.
1
[
5
[
3] R. Hou, W. Yu, M.D. Porosoff, J.G. Chen, T. Wang, J. Catal. 316 (2014) 1-10.
4] S. Qi, B.A. Cheney, R. Zheng, W.W. Lonergan, W. Yu, J.G. Chen, Appl. Catal. A
[
3
93 (2011) 44-49.
5] W.W. Lonergan, T. Wang, D.G. Vlachos, J.G. Chen, Appl. Catal. A 408 (2011) 87-
5.
6] T. Wang, G. Mpourmpakis, W.W. Lonergan, D.G. Vlachos, J.G. Chen, Phys.
Chem. Chem. Phys. 15 (2013) 12156-12164.
[
9
[
[
[
[
[
[
7] W.W. Lonergan, D.G. Vlachos, J.G. Chen, J. Catal. 271 (2010) 239-250.
8] M. Newville, J. Synchrot. Radiat. 8 (2001) 322-324.
9] B. Ravel, M. Newville, J. Synchrot. Radiat. 12 (2005) 537-541.
10]J.J. Rehr, R.C. Albers, Rev. Mod. Phys. 72 (2000) 621-654.
11]S. Bernal, J. Calvino, M. Cauqui, J. Gatica, C. Larese, J. Pérez Omil, J. Pintado,
Catal. Today 50 (1999) 175-206.
[
[
[
[
(
[
[
[
[
12]M.D. Porosoff, J.G. Chen, J. Catal. 301 (2013) 30-37.
13]J.P. Boitiaux, J. Cosyns, S. Vasudevan, Appl. Catal. 6 (1983) 41-51.
14]J. Goetz, M.A. Volpe, R. Touroude, J. Catal. 164 (1996) 369-377.
15]W.W. Lonergan, X. Xing, R. Zheng, S. Qi, B. Huang, J.G. Chen, Catal. Today 160
2011) 61-69.
16]J.P. Boitiaux, J. Cosyns, E. Robert, Appl. Catal. 32 (1987) 145-168.
17]M. Primet, M. Elazhar, M. Guenin, Appl. Catal. 58 (1990) 241-253.
18]S. Tauster, S. Fung, R. Garten, J. Am. Chem. Soc. 100 (1978) 170-175.
19]E.V. Benvenutti, L. Franken, C.C. Moro, C.U. Davanzo, Langmuir 15 (1999)
8140-8146.
[20]J.H. Kang, E.W. Shin, W.J. Kim, J.D. Park, S.H. Moon, J. Catal. 208 (2002) 310-
21
Page 21 of 29

320.
[
[
[
21]S.J. Tauster, S.C. Fung, J. Catal. 55 (1978) 29-35.
22]M. Kang, M.W. Song, K.L. Kim, React. Kinet. Catal. Lett. 75 (2002) 177-183.
23]G.L. Haller, D.E. Resasco, Advances in Catalysis 36 (1989) 173-235.
22
Page 22 of 29

Table 1. Summary of average particle sizes and CO uptake values for supported Pd‐Ni
catalysts
Catalyst
Al
Average Particle Size (nm)
CO‐uptake (μmol/g)
2
O
3
5.9
6.3
3.1
6.7
5.9
15.7
10.1
36.8
9.3
SiO
2
CeO
2
TiO
2
ZrO
2
16.1
Table 2. Summary of EXAFS analysis of Pd K‐edge for supported Pd‐Ni catalysts
Catalysts
N(Pd‐Pd)
CeO
2
TiO
2
SiO
2
γ‐Al
2 3
O
4.6±0.4
2.3±0.3
5.1±0.4
2.6±0.7
5.8±0.5
3.5±0.6
8.3±0.7
1.4±0.5
N(Pd‐Ni)
R(Pd‐Pd) (Å)
R(Pd‐Ni) (Å)
2.71±0.01
2.59±0.01
2.71±0.01
2.58±0.01
2.70±0.01
2.59±0.01
2.79±0.01
2.64±0.01
2
2
σ (Pd‐Pd) (Å )
0.008±0.001 0.008±0.003 0.007±0.001 0.010±0.001
0.007±0.001 0.009±0.001 0.008±0.002 0.016±0.004
2
2
σ (Pd‐Ni) (Å )
N(Pd‐Ni) + N(Pd‐Pd)
6.9
7.7
9.3
9.7
N(Pd‐Ni)/[N(Pd‐Ni) +
N(Pd‐Pd)]
0.33
0.34
0.38
0.14
23
Page 23 of 29

Fig. 1. HAADF TEM images and particle size distributions of (a) γ‐Al
ZrO , (d) CeO , and (e) TiO
2
O
3
, (b) SiO , (c)
2
2
2
2
Fig. 2. (a) Pd K-edge XANES spectra before and after reduction; (b) Fourier
2
transformed (magnitude) k weighted EXAFS function (χ(k)) before and after
2
reduction; (c) Pd K-edge Fourier transformed (magnitude) k weighted EXAFS
function (χ(k)) after reduction with first shell fit
Fig. 3. Conversions of 1,3-butadiene in flow reactor as a function of temperatures for
the comparison of Pd and Pd-Ni catalysts on (a) γ-Al O and (b) SiO ; Selectivities to
2
3
2
(
(
c) butenes and (d) 1-butene as a function of conversion of Pd-Ni (solid) and Pd
hollow)
Fig. 4. (a) Conversions and (b) TOFs of 1,3-butadiene in flow reactor at different
temperatures for the comparison of Pd-Ni bimetallic catalysts on different supports
Fig. 5. Selectivities to (a) butenes, (b) 1-butene, (c) trans-2-butene and (d) cis-2-
butene over Pd-Ni bimetallic catalysts on different supports
24
Page 24 of 29

Figure 1
30
20
10
0
0
2
2
2
2
2
4
6
8 10 12 14 16
Diameter, nm
(
a)
3
2
1
0
0
0
0
0
4
6
8 10 12 14 16
Diameter, nm
(b)
30
20
10
0
0
4
6
8 10 12 14 16
Diameter, nm
(c)
5
0
4
3
2
1
0
0
0
0
0
0
4
6
8 10 12 14 16
Diameter, nm
(
d)
3
2
1
0
0
0
0
0
4
6
8 10 12 14 16
Diameter, nm
(
e)
Fig. 1. HAADF TEM images and particle size distributions of (a) -Al O , (b) SiO ,
2
3
2
(
c) ZrO , (d) CeO , and (e) TiO
2 2 2
Page 25 of 29

Figure 2
(
a)
Fresh
After H Treatment
(b)
Fresh
After H Treatment
(c)
Fourier Transform
First Shell Fit
2
2
6
5
4
3
2
1
0
6
3
0

-Al O
2
3
6
Al O3
-Al O
2
2
3
SiO2
TiO2
SiO2
SiO2
3
CeO2
TiO2
TiO2
Pd Foil
CeO2
CeO2
0
2
4300
24400
24500
0
1
2
3
4
5
6
0
2
4
6
Energy (eV)
R (Å)
R (Å)
Fig. 2. (a) Pd K-edge XANES spectra before and after reduction; (b) Fourier
2
transformed (magnitude) k weighted EXAFS function (χ(k)) before and after
2
reduction; (c) Pd K-edge Fourier transformed (magnitude) k weighted EXAFS
function (χ(k)) after reduction with first shell fit. (The PdNi/CeO data was obtained
2
from literature [12])
Page 26 of 29

Figure 3
1
00
100
80
60
40
20
0
(
a) Al O
(b) SiO2
2
3
8
6
4
2
0
0
0
0
0
Pd
PdNi
Pd
PdNi
40
50
60
70
4
0
50
60
Temperature, C
70
o
o
Temperature, C
(c)
80
(d)
1
00
60
40
20
0
8
6
4
2
0
0
0
0
PdNi Pd
PdNi Pd
-Al O3
2

-Al O3
2
SiO2
SiO2
0
20
40
60
80
100
0
20
40
60
80
100
Conversion, %
Conversion, %
Fig. 3. Conversions of 1,3-butadiene in flow reactor as a function of temperatures for
the comparison of Pd and Pd-Ni catalysts on (a) γ-Al O and (b) SiO ; Selectivities to
2
3
2
(
c) butenes and (d) 1-butene as a function of conversion of Pd-Ni (solid) and Pd
hollow)
(
Page 27 of 29

Figure 4
(
a)
100

Al O3
2
80
60
40
20
0
SiO2
ZrO2
CeO2
TiO2
20
40
60
80 o
100
120
Temperature, C
100
(b)
Al O3
2
10
SiO2
ZrO2
CeO2
TiO2
1
0.0028
0.0030
0.0032
0.0034
1
/(T+273.14)
Fig. 4. (a) Conversions and (b) TOFs of 1,3-butadiene in flow reactor at different
temperatures for the comparison of Pd-Ni bimetallic catalysts on different supports
Page 28 of 29

Figure 5
1
00
(a) Butenes
1
00
(b) 1-Butene
8
6
0
0
8
6
4
2
0
0
0
0

Al O
SiO2
2
3
40
20
0

Al O
SiO2
ZrO2
TiO2
CeO2
2
3
ZrO2
TiO2
CeO2
0
20
40
60
80
100
0
20
40
60
80
100
Conversion, %
Conversion, %
1
00
100
80
60
40
20
0
(
c) Trans-2-Butene
(d) Cis-2-Butene
80
60
40
20
0

Al O3
SiO2
2
Al O3
SiO2
2
ZrO2
TiO2
CeO2
ZrO2
TiO2
CeO2
0
20
40
60
80
100
0
20
40
60
80
100
Conversion, %
Conversion, %
Fig. 5. Selectivities to (a) butenes, (b) 1-butene, (c) trans-2-butene and (d)
cis-2-butene over Pd-Ni bimetallic catalysts on different supports
Page 29 of 29