Catalyst performance changes induced by palladium phase transformation in the hydrogenation of benzonitrile

Article abstract of DOI:10.1016/j.jcat.2010.06.013

The influence of hydrogen pressure on the performance of a γ-alumina-supported palladium catalyst was studied for the multiphase selective hydrogenation of benzonitrile to benzylamine and byproducts. Semi-batch experiments of benzonitrile hydrogenation in 2-propanol were performed with hydrogen pressures between 2.5 and 30 bar, at a constant temperature of 80 °C. The intrinsic property of palladium to absorb hydrogen into its lattice structure has a strong influence on activity and selectivity. The transformation to stable palladium β-hydride above a threshold hydrogen pressure of 10 bar induces a persistent change in turnover frequency and byproduct selectivity. The turnover frequency increases from 0.32 s^-1 to a maximum of 0.75 s^-1 at this threshold pressure and decreases to 0.25 s^-1 with increasing hydrogen pressure. The palladium β-hydride phase suppresses the hydrogenolysis to toluene changing the selectivity from 6.5% to 2.0% and increasing the selectivity of the condensation to dibenzylamine from 1.6% to 2.7%, attributed to modified electronic interactions between catalyst and substrates. The selectivity to the desired product benzylamine is always high and increases with hydrogen pressure from 92.7% to 95.3%. The palladium catalyst state is mainly determined by the activation or operational hydrogen pressure, whichever was the highest, if the activation pressure was above 10 bar.

Full text of DOI:10.1016/j.jcat.2010.06.013

Journal of Catalysis 274 (2010) 176–191
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Catalyst performance changes induced by palladium phase transformation in the
hydrogenation of benzonitrile
a,*
a
b
a
a
Jasper J.W. Bakker , Anne Geert van der Neut , Michiel T. Kreutzer , Jacob A. Moulijn , Freek Kapteijn
a
Catalysis Engineering, Department of Chemical Engineering, Delft University of Technology, Julianalaan 136, 2628 BL Delft, The Netherlands
Product and Process Engineering, Department of Chemical Engineering, Delft University of Technology, Julianalaan 136, 2628 BL Delft, The Netherlands
b
a r t i c l e i n f o
a b s t r a c t
Article history:
The influence of hydrogen pressure on the performance of a
c
-alumina-supported palladium catalyst was
Received 18 April 2010
Revised 20 June 2010
Accepted 26 June 2010
Available online 31 July 2010
studied for the multiphase selective hydrogenation of benzonitrile to benzylamine and byproducts. Semi-
batch experiments of benzonitrile hydrogenation in 2-propanol were performed with hydrogen pressures
between 2.5 and 30 bar, at a constant temperature of 80 °C. The intrinsic property of palladium to absorb
hydrogen into its lattice structure has a strong influence on activity and selectivity. The transformation to
stable palladium b-hydride above a threshold hydrogen pressure of 10 bar induces a persistent change in
Keywords:
Palladium
Palladium b-hydride
Benzylamine
Benzonitrile
ꢀ1
turnover frequency and byproduct selectivity. The turnover frequency increases from 0.32 s to a max-
ꢀ1
ꢀ1
imum of 0.75 s at this threshold pressure and decreases to 0.25 s with increasing hydrogen pressure.
The palladium b-hydride phase suppresses the hydrogenolysis to toluene changing the selectivity from
6.5% to 2.0% and increasing the selectivity of the condensation to dibenzylamine from 1.6% to 2.7%, attrib-
uted to modified electronic interactions between catalyst and substrates. The selectivity to the desired
product benzylamine is always high and increases with hydrogen pressure from 92.7% to 95.3%. The pal-
ladium catalyst state is mainly determined by the activation or operational hydrogen pressure, whichever
was the highest, if the activation pressure was above 10 bar.
Toluene
Selective hydrogenation
Hydrogenolysis
Condensation
Adsorption modes
Surface intermediates
Ó 2010 Elsevier Inc. All rights reserved.
1
. Introduction
of solvent(s), reaction conditions and additives (e.g. ammonia) also
play an important role [2–4].
The topic of this paper is structure sensitivity of palladium (Pd)
catalysts, in particular the effect of structural changes due to Pd b-
hydride (b-PdH) formation at higher hydrogen pressure (pH ). The
reaction of interest is the hydrogenation of aromatic nitriles to the
corresponding primary amines. Here, controlling the selectivity is a
crucial demand and insight into the reaction mechanisms of
desired and undesired reactions is very valuable.
A scheme that depicts the reaction pathways during the hydro-
genation of benzonitrile (BN) to benzylamine (BA) and the main
byproducts can be found in Fig. 1. The reaction network is a
combination of hydrogenation, condensation and hydrogenolysis
reactions that can yield BA, N-benzylidenebenzylamine (DBI),
dibenzylamine (DBA), and toluene (TOL), respectively. Ammonia
2
3
(NH ) is formed during condensation reactions and hydrogenolysis
Primary aromatic amines are important chemical compounds
that have found widespread applications as chemical and pharma-
ceutical intermediates, solvents, paints, herbicides and synthetic
textiles. A major route to produce primary aromatic amines is
the heterogeneously catalyzed hydrogenation of aromatic nitriles
to TOL. The reaction mechanism is generally explained by the pre-
mise of benzylideneimine (BI) or other semi-hydrogenated inter-
mediates and a surface aminal (i.e. a-aminodialkylamine (BIBA)).
Several authors contributed to the elucidation of the reaction
mechanism since the beginning of the 20th century by proposing
that imines (ACH@NH), Schiff bases (AC@NAC), and enamines
(ANAC@CA) are the intermediates in the formation of primary,
secondary, and tertiary amines [5–10]. However, this mechanism
is based on a number of unproven assumptions that are corrobo-
rated only by indirect evidence: e.g. Schiff bases have been identi-
fied in solution, imines were detected on the catalyst surface by IR
spectroscopy, and the absence of tertiary amine formation in the
hydrogenation of BN has been observed. The inability to detect
imine and enamine intermediates is attributed to their high
reactivity.
[
1–4]. However, during the hydrogenation of aromatic nitriles to-
ward the aromatic amines, several byproducts are usually formed,
resulting in a loss of yield [1–4]. Activity and selectivity are mainly
determined by the amount, type and mobility of surface interme-
diates and transition states which are controlled by the catalyst.
Further, it should be noted that the reactant concentration, type
*
Corresponding author. Fax: +31 (0) 152785006.
E-mail address: j.j.w.bakker@tudelft.nl (J.J.W. Bakker).
0
021-9517/$ - see front matter Ó 2010 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2010.06.013

J.J.W. Bakker et al. / Journal of Catalysis 274 (2010) 176–191
177
and low pH
the lattice Pd expands (lattice constant increases) when the solid
solution phase ( -PdH) transforms into the b-PdH phase [28].
Moreover, the character of the electronic d-band of Pd changes
27,29–32]. The amount of H absorption in Pd depends on the
concentration in the liquid, temperature, Pd particle size, sur-
face topography, support interactions, precursor, and pretreat-
ment steps [28,33,34]. Structural changes upon transformation
to b-PdH can have a profound effect on the adsorption strength,
type, and amount of surface species, and thus on the activity
and selectivity [28,35–42].
2
[25–27]. Upon absorbing large amounts of hydrogen,
BN
BI
BA
H2
H2
N
a
NH
NH2
H -NH
[
H
BA
2
2
3
H
BIBA
TOL
N
NH2
In this paper, we demonstrate that in
alyzed BN hydrogenation, the b-PdH phase formed at increased
pH is responsible for the activity and selectivity effects. The pH
2 3
c-Al O -supported Pd-cat-
H2 -NH3
2
2
H2 -NH3
-
NH3
-BA
and the entailing transformation to b-PdH have a profound influ-
ence on the catalyst performance. Further, based on analysis of
the reaction mixture and characterization of fresh, activated, and
spent catalysts, a scheme of the reaction mechanism, modes of sur-
face adsorption, and surface intermediates are presented. By com-
bining d-band theory with simple frontier molecular orbital theory,
we explain the observed results.
H
N
H2
N
DBI
DBA
Fig. 1. Reaction pathways to different products in the hydrogenation of benzoni-
trile (BN): hydrogenation of BN to benzylamine (BA) via benzylideneimine (BI).
Condensation of BA and BI via
a-aminodialkylamine (BIBA) and N-benzylideneb-
enzylamine (DBI) to dibenzylamine (DBA). Hydrogenolysis to toluene (TOL) of DBA
and BA. Direct hydrogenolysis of BIBA to DBA is possible. Species in between
straight brackets are surface intermediates never detected in solution.
2
. Experimental
2.1. Catalyst and materials
Unless explicitly stated, the experiments were carried out using
From several BN hydrogenation studies and our own findings, it
can be concluded that the use of supported Pd catalysts results in
high activity and, more important, a high selectivity toward BA
compared to other platinum-group metal catalysts [11–15]. An
overview of different catalysts used under many different condi-
tions in the heterogeneous hydrogenation of BN is shown in Table
reduced 5 wt% Pd/
ments were performed using
sar. The catalysts we studied and reported in Table 1 (5 wt% Pd/C,
wt% Pd/BaCO , 5 wt% Ru/ -Al , 5 wt% Pt/C) were supplied by
c-Al
2
O
3
supplied by Alfa Aesar. Blank experi-
c
-Al (99.97%) supplied by Alfa Ae-
2 3
O
5
3
c
2 3
O
Aldrich. The other chemicals used in this study were purchased
from commercial suppliers and used as received (benzonitrile,
9%, Alfa Aesar; benzylamine, >98%, Alfa Aesar; dibenzylamine,
1
. From our own investigations, it appeared that a commercially
available -Al -supported Pd catalyst exhibited the highest
selectivity and was therefore used in this study.
While studying the effect of pH on activity and selectivity over
this catalyst, we encountered an unexpected anomaly: we found a
maximum in catalyst activity with increasing pH , accompanied by
9
9
9
c
2 3
O
8%, Alfa Aesar; toluene, 99.5%, J.T. Baker; n-isopropylbenzylamine,
8%, Alfa Aesar; 2-propanol, 99%, Aldrich;
2 3
c-Al O , 99.97%, Alfa
2
Aesar).
2
a sudden selectivity shift in byproducts leading to an increased
yield of BA. The literature does not report such an effect during
Pd-catalyzed nitrile hydrogenation. However, in the hydrogenation
and hydrogenolysis of hydroxymatairesinol using a Pd catalyst,
2.2. Selective hydrogenation
The selective hydrogenation experiments were carried out in a
semi-batch mode in a high-pressure stainless steel autoclave
(Medimex Reactor CH-2543) equipped with baffles and a gas-in-
Bernas et al. also encountered a negative dependence on the pH
above a certain pressure [16]. They explained this effect as a result
of competitive adsorption of H and the reactant. Dubois et al.
2
2
duced stirrer. H was continuously fed to the reactor to maintain
2
a constant pH (units: bar gauge further denoted as bar) during a
2
studied the hydrogenation combined with hydrogenolysis of 2-
methyl-2-nitropropane over a supported Pd catalyst and also
encountered an activity maximum; they gave two possible expla-
hydrogenation experiment. We used a simple alcoholic solvent,
2-propanol, and did not use any additives.
In a typical experiment, 0.5 g of 5 wt% Pd/
150 ml of 2-propanol and subsequently fed to the reactor. Different
cycles of pressurizing and depressurizing with N were applied to
2 3
c-Al O was added to
2
nations for this effect: more favorable H adsorption for the same
active sites or the formation of a new, less active, Pd phase [17].
Skakunova et al. hypothesized that the negative-order effect in
the hydrogenolysis of propane over a Pd–Ru catalyst was caused
2
remove the air from the reactor. Before a hydrogenation experi-
ment, the catalyst was activated in the reactor in 2-propanol under
by competitive adsorption of H
2
[18]. In general, hydrogenolysis
2 2
elevated pH at 80 °C for 1 h. The typical pH used for the standard
reactions catalyzed by different supported metals are structure
sensitive and often have a negative reaction order with respect
activation was 10 bar. In addition, the influence of activation pH
(2.5, 10, and 30 bar) on the catalyst performance was also
investigated.
After the in situ activation of the catalyst, the reactor was
brought to reaction conditions, i.e. a temperature (T) of 80 °C and
2
to pH
We demonstrate that in BN hydrogenation, competitive
adsorption between H and BN is not the explanation of the
anomaly observed. Rather, our hypothesis is that structural
changes of the Pd crystallites at higher pH explain the observed
2
up to ꢀ2.5 [19–23].
2
2
a pH between 2 and 30 bar for the different experiments. Then,
2
50 ml 2-propanol, adding up to 200 ml, with the appropriate
amount of BN, was added to the autoclave via a pressurized filling
phenomena. A well-known property of Pd, discovered by Graham
in 1866 [24] that differentiates it from all other metals, is the
easy absorption of atomic hydrogen in large quantities in the
Pd lattice, occupying the octahedral interstices of its face-cen-
tered cubic Pd lattice structure already at ambient temperature
3
tank (CBN,0 = 520 mol/m ). Reference hydrogenation experiments
3
(CBN,0 = 520 mol/m , T = 80 °C, pH
2
= 10 and 30 bar) were performed
with a
2 3
c-Al O support without Pd to determine its effect in the
hydrogenation of BN. In addition, hydrogenation experiments with

1
78
J.J.W. Bakker et al. / Journal of Catalysis 274 (2010) 176–191
Table 1
Hydrogenation of BN: comparison of BN conversion (XBN), BA selectivity (SBA), and yield (YBA) of different catalysts used in the literature and this study.
3
Catalyst
C
BN,0 (mol/m ), solvent T (°C), pH
2
(bar)
X
BN (%)
S
BA (%)
YBA (%) Remarks
5
5
5
% Pt/C
% Pd/C
% Rh/C
–
100, 35
100, 35
100, 83
100
100
100
0
19
22
0
19
22
Pt: major product was DBA
[9,43]
2
H O added
5% Rh/C
5% Pt/C
5% Pd/C
5% Pd/C
5% Pd/C
5% Ru/C
5% Pt/C
5% Pd/C
5% Rh/C
2000, octane
2000, ethanol
2000, benzene
2000, hexane
25, 3.5
25, 3.5
25, 3.5
25, 69
100
100
100
100
100
0
38
100
100
0
0
63
59
34
0
7
55
33
0
0
Mostly DBA was formed for Pt and Rh catalysts
[11–13]
63
59
34
0
18
55
33
Adding diethylamine increased YBA to 100% for Pd catalysts
Raney Ni
485, methanol
220, ethanol
1000, methanol
100, 40
100, 21
120, 10
n.a.
20
n.a.
76
n.a.
n.a.
n.a.
n.a.
96
Several solvents used
[44]
[45]
[46]
2
6
.5% Ir/Al
5% Ni/SiO
Al
0% Ni/SiO
2
O
3
YDBA was 36%
2
/
NH /CH OH added BN/NH ratio = 2
3
3
3
2
O
3
6
2
n.a.
100
88
n.a.
26
91
99
26
80
Pd/pyridyl
Raney Ni
–
100, 36
75, 40
32% TOL, 37% DBI
[47]
1530, methanol
Ni–Mo (CH
2
O)
485, methanol
323,
105, 41
30, 6
100
95
95
71
48
99
95
94
94
90
61
99
90
89
67
43
61
32% NH
NaH PO
3
/H
2
O added
2
O added
[48,49]
[15]
1
5
5
5
0% Pd/C
% Pd/C
% Pd/TiO
2
4
ꢁH
H
2
O/di-chloro-
2
methane
% Pd/Al
2
O
3
Ni catalyst
n.a., ethanol
20, n.a.
85, 15
100
Thiophene added
[50]
[14]
5
5
5
5
% Pd/C
% Pd/BaSO
% Pd/CaCO
% Pt/C
780,
2-propanol
100
100
100
100
100
21
24
15
10
36
21
24
15
10
36
Higher SBA achieved by adding NH
3
4
3
Raney Ni
Pt–Sn/nylon
Raney Ni
192, ethanol
60, n.a.
100, 15
100, 7
60, 4
145, 8.3
100, 15
100
100
99
100
46
15
95
0.1
20
n.a.
40
15
95
0.1
20
n.a.
16
Mostly DBA formed
Cont. counterc. reactor NH accumulation
3
[51]
[52]
1
% Pt/Al
2
O
3
–
Sn–Pt/SiO
2
390, ethanol
–
1000, methanol
DBA major product
Only XBN reported
Dibenzylhydrazine
[53]
[54]
[55]
0
5
.1% Pd/SiO
% Pt/Al
2
2
O
3
40
a
b
5
5
5
5
5
% Pd/Al
% Pt/C
% Pd/C
% Pd/BaCO
% Ru/Al
2
O
3
520, 2-propanol
80, 10
50
50
50
50
0
94
1
81
86
3
61
Main product for Pt/C was DBA
This study
a
b
a
b
Main byproduct for Pd catalysts was TOL
a
b
3
34
29
a
b
2
O
3
0
0
n.a. – not available.
a
S
Y
BA at XBN = 50%.
BA is maximum BA yield.
b
BA or mixtures of BN and BA were performed to get more insight
into the mechanism. Our results reported in Table 1 (bottom row)
No external and internal mass transfer limitations disguised our
rate data. See Supplementary material for the details about the
measurements and calculations performed.
were achieved under standard hydrogenation conditions
3
(C
BN,0 = 520 mol/m , T = 80 °C, pH
2
= 10 bar). The supported Pd cat-
Density functional theory (DFT) calculations were executed
*
alysts were activated for 1 h in situ in 2-propanol at 10 bar and
0 °C. The supported Pt and Ru catalysts were activated under H
flow for 1 h ex situ at 350 and 250 °C, respectively.
using the Becke 3LYP 6-31G method to calculate the frontier orbi-
8
2
tal energy levels of the highest occupied molecular orbital (HOMO)
and of the lowest unoccupied molecular orbital (LUMO) of the
compounds involved in the hydrogenation of BN. These calcula-
tions were performed with the Spartan’06 package version 1.1.2
(a program for quantum mechanical computations) from Wave-
function, Inc. (http://www.wavefun.com).
Samples were withdrawn from the reaction mixture during an
experiment at selected intervals to determine the composition of
the reaction mixture. Analysis of these samples was performed
on a Chrompack CP9001 gas chromatograph equipped with FID
detector and Chrompack Liquid Sampler 9050 autosampler fitted
with a CP-SIL 8 column (length 50 m, 0.25 mm ID). The GC temper-
ature program ranged from 50 to 250 °C at a heating rate of 10 °C/
min. The standard deviation of the integrated peaks of the FID sig-
The initial rate of consumption of BN is given as a turnover
frequency (TOF) and is defined as
r
TOF ¼
ꢁ N
A
ð1Þ
nal of the GC was 2.5%. GC data were checked by comparing H
flow into the autoclave with H consumption as calculated on
2
q
sites ꢁ aPd
2
2
the basis of reaction stoichiometry and GC data. A normalized car-
bon balance was performed to check whether no other compounds
were formed undetected by the GC analysis. The standard devia-
tion of the balance was 2.4%.
where aPd is the specific surface area of Pd in m /gPd as determined
2
A
by CO chemisorption, qsites is the site density of Pd in sites/m , N is
ꢀ1
the number of Avogadro in mol , and r is the catalyst activity in
mol/gPd/s and is defined as

J.J.W. Bakker et al. / Journal of Catalysis 274 (2010) 176–191
179
r ¼ dC ꢁ W
BN
1
pH = 10 bar and T = 80 °C (standard activation procedure used)
are shown in Fig. 2c.
2
ð2Þ
dt
Pd
The main product is benzylamine (BA) at the investigated pH
range. The major byproducts are toluene (TOL) and dibenzylamine
DBA). Together with BN, these three compounds make up more
2
where WPd is the concentration of Pd per liquid volume in
gPd=m 3l iquid. The activity is determined at the start of a hydrogenation
experiment up to 10% conversion of BN.
(
than 99% of the carbon mass balance. We did not find half-hydro-
genated species, e.g. benzylideneimine (BI) or N-benzylideneben-
zylamine (DBI), in the reaction mixture. Also, no DBI was found
in solution, which is supposed to be a relatively stable imine [56]
and has been detected by several researchers in the hydrogenation
of BN [57]. A minor byproduct detected was N-isopropylbenzyl-
amine (<1%) which is a product of the reaction between BA and
the solvent (2-propanol).
The initial concentration–time profile of DBA in Fig. 2a is an s-
shaped curve, whereas the formation of TOL starts from t = 0 min
onwards, and a straight line can be fitted through the data points.
This is an indication that TOL is also formed from BN directly, via
surface intermediates, without desorption and readsorption of
BA. After the maximum in CBA (i.e. YBA), when BN was mostly de-
pleted, hydrogenolysis of BA to TOL takes off (Fig. 2). The rate of
TOL formation and BA depletion are equal after full conversion of
BN.
The conversion of BN (XBN) is defined as
C
BN;0 ꢀ CBN
X
BN
¼
ꢂ 100%
ð3Þ
C
BN;0
where CBN,0 is the concentration of BN at t = 0 min and CBN the ac-
3
tual concentration of BN both in mol/m .
x
All selectivities (S ) are determined at XBN = 50% and are defined
as
C
x
S
x
¼ C
ꢀ CBN ꢂ 100%
ð4Þ
BN;0
where x is BA, TOL, or DBA.
The desired product is BA and the maximum yield of BA (YBA) is
defined as
C
C
BA;max
Y
BA
¼
ꢂ 100%
ð5Þ
BN;0
2
TOFs, selectivities, and yield versus pH are given in Fig. 3. The
3
where CBA,max is the maximum concentration of BA in mol/m3.
The TOL yield (YTOL) and DBA yield (YDBA) are defined as CTOL
standard condition is CBN,0 = 520 mol/m , but also experiments
3
were conducted at CBN,0 = 300 and 1000 mol/m , see Fig. 3a.
/
Clearly, for these experiments, the TOFs are the same as for the
standard conditions.
C
BN,0 and CDBA/CBN,0, respectively, at YBA.
Concentration–time profiles with variation of activation and
hydrogenation pressure are shown in Fig. 4a and b. Fig. 4a shows
two concentration–time profiles: one with activation at 10 bar fol-
lowed by hydrogenation at 30 bar, and one with 30-bar activation
followed by 10-bar hydrogenation. These two configurations give
equal results. Fig. 4b shows three concentration–time profiles:
2
.3. Catalyst characterization
Prior to analysis, fresh, activated, and spent catalyst samples
were dried in an oven (static air atmosphere) that was ramped
up from room temperature at a heating rate of 0.5 °C/min up to
1
20 °C, followed by 2 h at 120 °C.
The catalysts were characterized using X-ray powder diffraction
1
0-bar activation followed by 2.5-bar hydrogenation, 2.5-bar acti-
vation followed by 10-bar hydrogenation, and 10-bar activation
followed by 10-bar hydrogenation. The activation at 2.5 bar fol-
lowed by 10-bar hydrogenation displays similar results as 10-bar
activation followed by 10-bar hydrogenation, but 10-bar activation
followed by 2.5-bar hydrogenation yields different results. The re-
sults for 30-bar activation followed by 10-bar hydrogenation
(
XRD), temperature-programmed reduction (TPR), thermal gravi-
metric analysis (TGA), inductively coupled plasma optical emission
spectroscopy (ICP-OES), diffuse reflectance infrared Fourier trans-
form (DRIFT), CO chemisorption, N physisorption, transmission
2
electron microscopy (TEM), and laser diffraction. The details
regarding these characterization methods can be found in the Sup-
plementary material. Note that XRD and TPR were only performed
on fresh and activated Pd/
for TGA and DRIFT, fresh Pd/
BN, BA, and DBA for 4 h at 80 °C. Fresh Pd/
also investigated with TGA and DRIFT. Characterization of spent
Pd/ -Al catalysts using TGA and DRIFT was performed after
hydrogenation experiments that were stopped at YBA. Character-
ization of spent Pd/ -Al catalysts using ICP-OES, CO chemisorp-
tion, N physisorption, TEM, and laser diffraction was performed
after hydrogenation experiments stopped at XBA = 100%.
(
(
Fig. 4a) and 10-bar activation followed by 10-bar hydrogenation
Fig. 4b) are also dissimilar from each other.
c
-Al
2
O
O
2 3
3
catalysts. To obtain references
catalyst was soaked in pure
-Al catalysts were
c-Al
3.2. Catalyst characterization
c
2 3
O
An overview of the characterization of the fresh, activated, and
spent Pd catalysts is shown in Table 2. ICP-OES analysis showed a
Pd content of 4.22 wt% in the fresh catalyst. This value did not
c
2 3
O
c
2 3
O
change during activation or hydrogenation at different pH
2
, so,
2
no leaching occurred during any of the activation steps or hydro-
2
genation. The SBET of the
c
-Al
pore volume of 0.47 cm /g and a mean pore diameter of 12 nm.
The mean particle size of the -Al support was 37 m and
2 3
O support was 154 m /g with a
3
3
. Results
c
2
O
3
l
did not undergo any significant change during activation or
3.1. Hydrogenation of benzonitrile
hydrogenation.
The average size of the Pd crystallites of the fresh catalyst was
5.6 nm (dispersion = 20.0%) and 4.2 nm (dispersion = 26.7%) deter-
mined by CO chemisorption and TEM, respectively. The Pd crys-
Typical concentration–time profiles of the various components
involved in the semi-batch hydrogenation of benzonitrile (BN)
using a Pd/ -Al catalyst, activated at 10 bar, at 80 °C with a
pH of 5.3 and 15 bar are shown in Fig. 2a and b. Fig. 2a also shows
c
2
O
3
2 3
tallites were uniformly distributed over the c-Al O support.
2
The discrepancy between the Pd crystallite sizes can be explained
by the error we introduce by assuming that the CO/Pd ratio is
unity in the calculation of the Pd crystallite size from CO chemi-
sorption measurements. CO is known to adsorb on Pd not only in
an end-on configuration (CO/Pd = 1) but also in a bridged config-
uration (CO/Pd = 0.5) [58–60]. After activation at 10 and 30 bar,
no changes to the average Pd crystallite size was observed,
where the catalyst performance parameters, i.e., turnover frequen-
cies, selectivities, and yields, were determined in the concentra-
tion–time profiles by using formulas (1)–(5) defined in the
experimental part. Concentration–time profiles of the hydrogena-
3
tion of BA (CBA,0 = 520 mol/m ) and of a mixture of equimolar
3
amounts of BN and BA (CBN,0 = CBA,0 = 300 mol/m ) performed at

1
80
J.J.W. Bakker et al. / Journal of Catalysis 274 (2010) 176–191
3
3
2 3 2
Fig. 2. Concentration (mol/m ) vs. time (min) profiles for BN hydrogenation at 80 °C over a Pd/c-Al O catalyst with CBN,0 = 520 mol/m performed at 5.3 (a) and 15 (b) bar H ;
3
3
(
C
c) concentration (mol/m ) vs. time (min) profiles for BA hydrogenation with CBA,0 = 520 mol/m (gray symbols) and hydrogenation of a mixture of BN and BA with
BN,0 = 300 mol/m3 and CBA,0 = 300 mol/m (black symbols) at 80 °C over a Pd/
3
c-Al
catalyst at 10-bar hydrogenation. All catalysts were activated at 10 bar. Left y-axis
2
O
3
shows the concentration of BN, BA and TOL. The right y-axis shows the DBA concentration. Key: ꢀ = CBN; j = CBA; d = CDBA; N = CTOL. The dashed lines are guidance for the
eyes.

J.J.W. Bakker et al. / Journal of Catalysis 274 (2010) 176–191
181
3
Fig. 3. Catalyst performance vs. pH
The diamonds indicated by the arrows at 20 and 30 bar pH
2
(bar) for BN hydrogenation over a Pd/
2 3
c-Al O catalyst at 80 °C, CBN,0 = 520 mol/m , and 10-bar activation: (a) turnover frequencies (TOF).
3
represent hydrogenations performed at CBN,0 = 1000 mol/m3 and 300 mol/m ; (b) maximum BA yield (YBA); (c)
2
Y
TOL (left y-axis) and YDBA (right y-axis) at YBA; (d) SBA at XBN = 50%; (e) STOL (left y-axis) and SDBA (right y-axis) at XBN = 50%. Key: ꢀ = TOF; j = SBA or YBA; d = SDBA or YDBA
;
N = STOL or YTOL. The dashed lines are guidance for the eyes.
whereas the spent catalysts (hydrogenations performed at 10 and
Ex situ X-ray diffraction patterns of four Pd/
activated at different pH
reflections region is presented because there is no interference
of -Al support peaks in this region. Note that the Pd(1 1 1)
reflection showed the same trends but was heavily obscured
by -Al support reflections. The reflection at 2h = 84.6° is a
reflection of the crystalline part of the -Al support and is
shown in Fig. 5 as a reference. The observed Bragg diffraction
reflections correspond to Pd(3 1 1) and b-PdH (3 1 1) reflections.
c
-Al
2
O
3
catalysts
3
5
0 bar) showed an increase in the average Pd crystallite size from
.6 to 6.2 nm. However, TEM analysis did not show any growth of
2
are shown in Fig. 5. The Pd(3 1 1)
the Pd crystallite size: the average Pd crystallite size of the fresh
and spent catalyst was 4.2 nm. The increased average Pd crystal-
lite size of the spent catalyst (i.e. lower Pd surface area) as mea-
sured with CO chemisorption is probably caused by adsorbed
species on the Pd surface that decrease the area that is available
for chemisorption of CO.
c
2 3
O
c
2 3
O
c
2 3
O
x

1
82
J.J.W. Bakker et al. / Journal of Catalysis 274 (2010) 176–191
3
Fig. 4. Concentration (mol/m ) vs. time (min) profiles of BN hydrogenation over a Pd/
C
c
-Al
2 3
O catalyst at different activation and hydrogenation pressures at 80 °C and
3
BA,0 = 520 mol/m : (a) 30-bar activation and 10-bar hydrogenation (open markers) and 10-bar activation and 30-bar hydrogenation (closed markers); (b) 10-bar activation
and 10-bar hydrogenation (gray markers), 10-bar activation and 2.5-bar hydrogenation (open markers), and 2.5-bar activation and 10-bar hydrogenation (closed markers).
Left y-axis shows the concentration of BN, BA, and TOL. The right y-axis shows the DBA concentration. Key: ꢀ = CBN; j = CBA; d = CDBA; N = CTOL. The dashed lines are guidance
for the eyes.
Table 2
2 3
Summary of characterization results of fresh, activated, and spent 5 wt% Pd/c-Al O catalyst.
2
D < 50%a
(lm)
Catalyst
Average Pd crystallite size (nm)
CO chemisorptionb
Pd loading (wt%)
ICP-OES
S
BET (m /g)
TEM
N
2
physisorption
Laser diffraction
Fresh
5.6
5.6
5.6
6.2
6.2
4.2 ± 0.6
n.m.
n.m.
4.2 ± 0.9
n.m.
4.22
4.22
4.21
4.23
4.22
154
n.m.
154
n.m.
150
37
Activated at 10 bar
Activated at 30 bar
Spent 10 bar
Spent 30 bar
n.m.
n.m.
37
n.m.
n.m. – not measured.
a
D < 50% is the diameter below which 50% of the particle volume is comprised.
Assumptions: CO:Pd = 1 and spherically shaped crystallites.
b
When H
2
is absorbed in the Pd lattice structure and b-PdH is
79.7° (indicated with the arrow in Fig. 5) due to expansion of the
Pd lattice. Reflections at 81.6° with a calculated lattice constant
formed, the Bragg reflection shifts to smaller 2h angles of around

J.J.W. Bakker et al. / Journal of Catalysis 274 (2010) 176–191
183
2 3 2
Fig. 5. Four ex situ X-ray diffraction patterns of Pd/c-Al O catalysts activated at different pH at 80 °C in 2-propanol prior to XRD analysis. Key: (a) = not activated;
(
b) = activated at 2.5 bar; (c) = activated at 10 bar; (d) = activated at 30 bar.
of 3.91 Å and 79.7° with a calculated lattice constant of 3.99 Å are
from Pd(3 1 1) and b-PdH (3 1 1), respectively. So, the lattice con-
stant of Pd increases 2.0% when a pH of 30 bar is applied during
the activation step (in 2-propanol at 80 °C for 1 h). The amount
of hydrogen absorbed (represented by x in b-PdH ) in the Pd crys-
x
2
x
tallites correlates with the location of the Bragg reflection and is
determined by the Pd crystallite size: smaller Pd crystallites absorb
less hydrogen in the b-PdH phase and more in the
a-PdH phase
[
33,34,61]. We determined x for the -PdH phase using Vegard’s
law and lattice constants and x-values for similarly sized nanoscale
particles from the literature as x = 0.43 [62]. The lattice constant for
Pd(3 1 1) is reported to be 3.89 Å [63]. Compared to our lattice con-
stant of 3.91 Å, it shows that the fresh catalyst and those activated
at 2.5 and 10 bar experienced minor lattice expansion, obviously
because of the pretreatment steps and x being 0.06. Note that the
fresh catalyst had already been reduced by the supplier. The cata-
lysts activated at 2.5 and 10 bar also showed a small reflection at
7
9.8° but the main reflection was from Pd(3 1 1) at 81.5°.
The catalyst activated at 10 and 30 bar showed a strong increase
in intensity of the Pd(3 1 1) and b-PdH0.43(3 1 1) reflections,
respectively. Note that the 10-bar experiment did not shift the
3
1 1 reflection to smaller 2h angles. The increase in intensity could
be caused by a rise in the average length of an ordered row of
atoms in the 3 1 1 crystallographic direction caused by structure
rearrangement: exposure to H can result in ordering of Pd crystal-
2
lites [64]. Although XRD is considered as a bulk characterization
technique, surface topography changes can also enhance the inten-
sity of a reflection since for nanoscale particles a substantial
amount of the atoms is located at the surface. The observed broad-
ening of the b-PdH0.43(3 1 1) reflection (Fig. 5) is attributed to
microstrain (e.g. from non-uniform lattice distortion) and solid
solution inhomogeneity.
Fig. 6. H
different pH
activation; (c) = not activated. The dashed lines show the baselines for the
determination of the relative amounts of H release.
2
–TPR profiles of three Pd/
2 3 2
c-Al O catalysts activated prior to H –TPR at
2
in 2-propanol for 1 h at 80 °C. Key: (a) = 30-bar activation; (b) = 10-bar
Three H
activated at 10 and 30 bar H
a fresh Pd/ -Al catalyst, are shown in Fig. 6. The H
all have a negative H peak around 83 °C, which is characteristic of
Pd catalysts with Pd crystallite sizes larger than 2–3 nm. This peak
is attributed to the H desorption from the decomposition of the
b-PdH formed by the uptake of H at room temperature (5 min)
during the TPR experiment [65–68]. The area of the b-PdH decom-
position peak was determined in the temperature range 60–100 °C
60,69–72]. The b-PdH decomposition peak area decreased by 20%
2
–TPR curves, two of Pd/
(in 2-propanol at 80 °C for 1 h) and
–TPR curves
2 3
c-Al O catalysts that were
2
2
c
2
O
3
2
2
and 28% (corrected by the weight of the samples) if the Pd/
2 3
c-Al O
2
prior to H –TPR was activated at 10 and 30 bar, respectively.
2
x
2
The overall DRIFT spectra indicated that BA and BN were ad-
sorbed in excess on the catalyst surface when comparing these
spectra with reference spectra (not shown). An overall DRIFT spec-
x
[
x
2 3
trum of a spent Pd/c-Al O catalyst used in hydrogenation at 20 bar

1
84
J.J.W. Bakker et al. / Journal of Catalysis 274 (2010) 176–191
is shown in Fig. 7a. The hydrogenations were stopped directly after
BA was reached. DRIFT spectra of the (C„N) stretching region of
two spent Pd/ -Al catalysts from a 10- and 20-bar hydrogena-
tion experiment are shown in Fig. 7b. The C„N stretches are very
valuable characteristics because they fall in a practically unpopu-
total weight loss (discarding water and catalyst peaks) was
3.1 ± 0.1 and 4.3 ± 0.2 wt% for hydrogenations at 10 and 20 bar,
respectively. The weight loss of 2 ± 0.1 wt% at 75 °C (endothermic)
is due to desorption of physisorbed water. The weight loss at 190
and 215 °C (both exothermic) for the 10- and 20-bar experiment,
respectively, originates from decomposition of adsorbed species.
From reference experiments (not shown), it is concluded that this
weight loss (at 190 and 215 °C) is from decomposition of BA. The
catalyst used in hydrogenation at 20 bar showed more strongly ad-
sorbed and larger amounts of BA on the catalyst surface: 0.7 ± 0.1
and 1.3 ± 0.1 wt% for hydrogenation at 10 and 20 bar, respectively.
A weight loss of 0.8 ± 0.1 wt% at 260 °C (for hydrogenation at 10
and 20 bar) originates from desorption of the Pd precursor (endo-
thermic), because this weight loss also occurred when TGA is per-
formed using a fresh catalyst (not shown). The weight loss around
285 °C (exothermic) observed in the thermographs of both spent
catalysts was not detected in the thermographs of a fresh catalyst
and catalysts used in the reference experiments. This weight loss
was ascribed to decomposing semi-hydrogenated species that
were chemisorbed on the Pd crystallites. The weight losses at
340 and 440 °C (both exothermic) are from combustion of basic
Y
v
c
2 3
O
lated spectral region [73]. A DRIFT spectrum of a fresh Pd/
2 3
c-Al O
catalyst did not show any peaks in this region (not shown). Free
ꢀ1
BN shows a band at 2231.7 cm in 2-propanol [74]. Interactions
3
þ
3þ
of
AlAOH groups) with BN are found above 2240 cm
interactions of the acidic -Al groups will be with NH
amine species [75–77]. The interaction at 2245 cm is attributed
to end-on adsorption via an acidic AlAOH of -Al . Several con-
2 3
c-Al O surface groups (e.g. acidic AlT and AlO , basic and acidic
ꢀ1
but most
c
2
O
3
3
and
ꢀ1
c
2 3
O
figurations of BN adsorption are possible on Pd, i.e., end-on and
side-on. After hydrogenations at 10 and 20 bar, both adsorption
configurations were observed (Fig. 7b): side-on adsorption at
ꢀ
1
ꢀ1
ꢃ2165 cm and end-on at ꢃ2229 cm . Strikingly, there is a dis-
tinct difference between the BN adsorption configuration ratios
(
side-on/end-on) for both spent catalysts: this ratio is much larger
for hydrogenation at 10 bar than at 20 bar.
Thermogravimetric analyses of two spent Pd/
2 3
c-Al O catalysts
used in hydrogenations at 10 and 20 bar are compared in Fig. 8.
It is also shown in Fig. 8 whether exothermic combustion or endo-
thermic desorption occurred, as determined by single-point differ-
ential thermal analysis (STDA) present in the thermobalance. The
species (e.g. BA, DBA, and NH
3
) deposited near Pd crystallites and
on the -Al support, respectively [78–81]. A catalyst used in
c
2 3
O
hydrogenation at 10 bar contains less of these deposits on the sur-
face: 2.0 ± 0.1 and 3.0 ± 0.1 wt% for hydrogenation at 10 and 20 bar,
respectively.
The weight loss at 810 °C (exothermic) is also seen in the TGA of
a soaked catalyst in BN but was absent for the catalysts soaked in
the other compounds. The weight losses were 0.4 ± 0.02 and
0
.1 ± 0.01 wt% for the 10 and 20 bar experiment, respectively. The
weight loss of 0.5 ± 0.02 wt% at 860 °C (exothermic) is from the
phase transformation of -Al to d-Al [82] and was also visi-
ble in the TGA of a fresh catalyst sample.
c
2
O
3
2 3
O
3.3. Molecular orbital energy calculations
The calculated values of HOMO and LUMO energy levels are
shown in Table 3. The LUMO level represents the ease to accept
electrons: the lower the LUMO the easier it accepts electrons.
The lowest calculated LUMO level is the
p C„N orbital of BN at
ꢀ
1.41 eV. The HOMO level represents the ease to donate electrons:
the higher the HOMO level the better it donates electrons. BIBA,
DBA, and BA show the highest HOMO levels at ꢀ6.11, ꢀ6.09, and
ꢀ
6.18, respectively. Note that the orbital energy levels were calcu-
lated without any interaction with the Pd surface or solvent effects
taken into account. The orbital levels for the intermediates are
thereby speculative: first, because the intermediates were not ob-
served during reaction and secondly, because interaction with the
surface might significantly alter the levels of these short-lived
intermediates.
4
. Discussion
Some striking peculiarities related to the activity and selectivity
of the hydrogenation of BN over Pd/
Upon increasing pH , the activity of the catalyst goes through a
maximum around 10 bar, while the product selectivities show a
steep change around this pH . Moreover, the turnover frequency
2 3
c-Al O have been observed.
2
2
and selectivities depend on the activation pH
2
prior to
hydrogenation.
2 3
Fig. 7. DRIFT spectra of spent Pd/c-Al O catalysts after BN hydrogenations
3
performed at 10 and 20 bar at 80 °C, CBN,0 = 520 mol/m , and 10-bar activation.
The reactions were stopped at YBA: (a) full spectrum for hydrogenation performed
4.1. Activity profile vs. pH
2
2
0 bar showing the C„N stretching region indicated by the square; (b) enlarged
C„N stretching region for hydrogenation performed at 10 (dashed line) and 20 bar
full line).
The TOF of BN hydrogenation increases with increasing pH
to 10 bar, where it reached its maximum (Fig. 3a). Above the
2
, up
(

J.J.W. Bakker et al. / Journal of Catalysis 274 (2010) 176–191
185
Fig. 8. Thermogravimetric (TG) and differential thermogravimetric (DTG) profiles in air of two spent Pd/
c
-Al
2
O
3
catalysts after hydrogenation performed at 10 bar (a) and
3
2
0 bar (b) at 80 °C, CBN,0 = 520 mol/m , and activated at 10 bar. The reactions were stopped at YBA. The left y-axis represents the weight loss and the right y-axis represents the
differential weight loss. Key: endo = endothermal desorption process; exo = exothermal decomposition process; as determined by the single-point differential thermal
analysis (SDTA) present in the thermobalance.
Table 3
HOMO–LUMO levels.
4.2. The b-PdH phase
Species
HOMOa (eV)
LUMOb (eV)
2
The sudden drop in activity around a pH of 10 bar might be
caused by the well-known formation of b-PdH. Ex situ XRD analysis
of activated and non-activated catalysts showed remarkable
changes of the Pd crystallite lattice parameters after activation at
BN
ꢀ7.26
ꢀ6.61
ꢀ6.84
ꢀ6.18
ꢀ6.40
ꢀ6.87
ꢀ6.11
ꢀ6.43
ꢀ6.44
ꢀ6.09
ꢀ7.10
ꢀ1.41
ꢀ1.19
ꢀ1.33
0.05
0.14
2.14
BI (Z)
BI (E)
BA
2
increased pH (Fig. 5). The lattice constant of Pd increases by
TOL
2
.0% after activation at 30 bar, which is attributed to the formation
NH3
BIBA
DBI (E)
DBI (Z)
DBA
of the b-PdH phase. This observation is remarkable because the
XRD patterns were measured ex situ, and probably all the hydrogen
was already desorbed from the catalyst samples.
0.10
ꢀ1.18
ꢀ1.07
0.08
The conditions for a transformation of nanoscale Pd crystallites
2
-Propanol
1.96
into b-PdH have been studied extensively. Typically, ꢃ1 bar H
2
is
ꢄ
Calculated using Becke 3LYP 6-31G method.
sufficient in the gas phase to transform supported nanoscale Pd
a
HOMO = highest occupied molecular orbital.
LUMO = lowest unoccupied molecular orbital.
crystallites into b-PdH at around 80 °C [30,88,89]. In the liquid
b
phase, the critical value for pH
cause the H solubility in a liquid phase is an order of magnitude
smaller than the H concentration in the gas phase [35,41,90]. This
corresponds with a pH
of ꢃ10 bar for liquid phase processes, ni-
cely matching the pH of 10 bar for the maximum in Fig. 3a. Note
2
is expected to be much higher be-
2
threshold pH
2
of 10 bar, further increasing pH
2
showed a strong in-
2
verse effect on the reaction rates and the order changed to ꢃꢀ1.
This maximum in activity was not expected, and no Pd-catalyzed
nitrile hydrogenation results are reported in the literature showing
this behavior. However, it is observed frequently in other hydroge-
nation and also in hydrogenolysis reactions using supported Pd
catalysts where it is attributed to competitive adsorption [16–
8,83–86]. Hydrogenolysis of simple alkanes over different sup-
ported catalysts all exhibited a negative order in H , and this was
explained by the decreasing concentration of the dehydrogenated
intermediate that undergoes C–C bond scission [20,87], although
2
2
that the transformation of Pd into b-PdH is accelerated when the
solvent is an alcohol compared to non-alcoholic solvents and water
[91].
Under TPR conditions, the b-PdH phase is unstable above 80 °C
as can be seen in Fig. 6, but it is striking that the expansion/distor-
tion of the Pd lattice was still retained, as shown by ex situ XRD (see
Fig. 5), after activation at 30 bar and a drying step performed at
120 °C in air. Several researchers observed the lattice distortion
1
2
it was also speculated that competitive adsorption between H
2
2
after the H had been removed: only after annealing at 450 °C,
and the hydrocarbons could play a role. To check whether compet-
itive adsorption could indeed be the reason for our observation, we
performed hydrogenation experiments at 20 and 30 bar with other
the Pd atoms could be returned to their original position [25,92].
This also implies that the persistent expansion of the Pd lattice
after activation of 30 bar could still have an effect on the catalyst
performance during a hydrogenation experiment even if the b-
PdH phase is somehow eliminated in between activation and
hydrogenation.
3
3
C
BN,0 concentrations, viz., 1000 mol/m and 300 mol/m . However,
increasing or decreasing CBN,0 did not have any influence on the
TOF (Fig. 3a) and, as a consequence, this explanation is rejected.

1
86
J.J.W. Bakker et al. / Journal of Catalysis 274 (2010) 176–191
When the catalyst precursor contains carbonaceous material or
2
step, i.e., at pH of 10 bar or lower, then the applied hydrogenation
when a carbon support is used, carbon might incorporate in the Pd
lattice and yield a solid solution in Pd (PdC) which could give sim-
ilar shifts in reflections as b-PdH [26,93,94]. In our case, a PdC
phase might be formed with the solvent (i.e. 2-propanol) as a car-
bon source. Formation of a PdC phase prohibits the formation of a
b-PdH phase [65,93]. TPR is a useful technique for elucidating this
point. If no b-PdH decomposition peak would be observed in the
TPR profiles of the 10- and 30-bar activated catalysts then, the cat-
alysts could have been contaminated by carbon soluble in the Pd
lattice. However, we do see b-PdH decomposition peaks (Fig. 6)
after the activation performed in 2-propanol, so it is concluded that
no extensive PdC phase has been formed during the activation step.
It should be noted that Teschner et al. observed that also a meta-
stable PdC phase in the top few layers can be formed in the early
stage of the reaction. This phase was only stable under reaction
conditions during the gas-phase hydrogenation of acetylene at
pH determines the catalyst performance (Fig. 4b). So, a higher BA
2
productivity can be achieved by applying an activation step above
10 bar prior to hydrogenation of BN at 10 bar compared to a stan-
dard experiment of both activation and hydrogenation at 10 bar.
4.3.2. Electronic interactions
The lattice expands and changes to the d-band of Pd occur upon
transformation into the b-PdH phase [27,29–32,40,101–105]:
– Center of the d-band (
its Fermi level (E ). The energy level of E
icantly when x (in b-PdH
) is below 0.6 and is ꢀ5.22 eV for poly-
crystalline Pd. for Pd is 1.83 eV below E for b-PdH0.4 is
ꢃ2.2 eV below E . Catalytic activity is often related with the dif-
ference between and E : for a larger difference, the transition
e
d
) of b-PdH is shifted down compared to
f
f
does not change signif-
x
e
d
f d
. e
f
e
d
f
metal becomes less reactive [106–109].
very low pH
2
(<0.13 bar) [95,96]. If the pH
2
is raised above a certain
– Width of the d-band of b-PdH0.4 is ꢃ15% narrower compared to
Pd. A narrower d-band results in a decrease in Pauli repulsion.
Weaker repulsive interactions increase the binding energy of
adsorption with HOMO orbitals [110]. However, a spatially
more extended d-band increases the propensity of rehybridiza-
pressure, then this PdC surface phase disappears and a b-PdH
phase is formed. Moreover, when trans-2-pentene is chosen as
the reactant under similar reaction conditions, then no PdC phase
is formed. Obviously, double bonds do not strongly fragment on
the surface of Pd; hence, the C„C triple bond is essential in form-
ing a PdC surface phase. This is supported by DFT calculations,
which shows that only for certain reactants, in combination with
a Pd surface, a PdC surface phase can be formed [97–99]. Note that
the morphology of the Pd catalyst is also important: decomposition
of organic species takes place more efficiently on stepped than on
flat surfaces [100]. Since our experiments were performed under
completely different conditions and with other reactants than the
ones used by Teschner et al. [95], we would argue against the for-
mation of a metastable PdC surface phase in our experiments espe-
tion from
p to di-r coordination to overcome the larger repul-
sive interaction [111].
Thus, the lower BN hydrogenation activity above the threshold
2 d f
pH of 10 bar can be caused by the lower e (compared to E ) and
reduced width of the d-band of b-PdH compared to Pd.
Interaction of occupied (e.g. HOMO) and unoccupied molecular
orbitals (e.g. LUMO) of an adsorbate with the broad sp-band of Pd is
always attractive and causes broadening into resonances and a
down shift in energy of these orbitals. Subsequently, hybridization
between the d-band of Pd and the broadened molecular orbitals re-
sults in splitting of these orbitals, thereby forming bonding contri-
butions at lower energies and antibonding ones at higher energies.
The primary interactions during adsorption of BN are electron
cially in view of the used pH
2
.
4.3. Influence of b-PdH phase transformation on selectivity
The hydrogenation of BN over Pd/
in three products: BA, TOL, and DBA, at the investigated pH
from 2.5 to 30 bar (Fig. 2). The production of BA dominated over
the formation of TOL and DBA up to nearly full conversion of BN
c
-Al
2
O
3
in 2-propanol resulted
range
donation from occupied donative molecular orbitals of BN (
C„N is the donating HOMO but also -donation from the lone pair
on nitrogen is possible) to empty orbitals of Pd (e.g. d 2 and s) and
backdonation from occupied d-orbitals of Pd (e.g. d
p
2
r
z
x
2
ꢀy2) to the
ꢄ
(
X
BN ꢃ 95%).
unoccupied acceptor molecular orbitals of BN (LUMO is
p
C„N),
The highest YBA and SBA achieved were above 90% and 95%,
similar to the Blyholder model for CO adsorption on transition
metals [112]. However, donative interactions are mainly repulsive
(Pauli repulsion) due to orthogonalization between occupied orbi-
tals of BN and occupied d-orbitals of Pd [113]. These can only be-
come attractive when donative orbital-derived antibonding
2
respectively, when a pH was applied above 20 bar. This is an
excellent result compared to other Pt-group catalysts and a good
result compared to Ni catalysts (Table 1), considering that in this
study no additives (e.g. NH
The maximum in catalyst activity at the threshold pH
is accompanied by sharp change in selectivities of TOL and DBA
Fig. 3e). As competitive adsorption is ruled out as explanation
for the activity maximum at this pH , this cannot be the cause of
3
) were used.
2
of 10 bar
f
orbitals are shifted up through E and become empty (‘relieved
repulsion’) [114]: a higher energy level of the donative orbital re-
sults in more attraction and a stronger bond to the Pd surface.
Backdonation is attractive if the energy level of the acceptor orbital
(LUMO) is low enough in energy to interact with the d-band of Pd.
Backdonation is always attractive because the antibonding part of
the LUMO is too high in energy to be filled.
(
2
this of this sharp selectivity change. Since the trends in selectivity
and yield are similar, we will concentrate on the difference in
selectivity in the following.
BN is the best electron acceptor of all the compounds involved
in the hydrogenation of BN since it has the lowest LUMO (Table 3).
The interaction of BN with Pd upon adsorption is mainly governed
by electron backdonation of Pd to the LUMO of BN. This interaction
4
.3.1. Influence of activation pressure
The performance of a catalyst activated at 10 bar and applied in
a BN hydrogenation experiment at 30 bar resulted in the same per-
formance as one activated at 30 bar and hydrogenation applied at
d
is higher when the energy level of e is higher and the d-band is
1
0 bar (Fig. 4a). Clearly, at 30 bar the b-PdH phase was formed, and
wider as is the case for Pd compared to b-PdH. However, not only
the strength of adsorption increases when increased backdonation
occurs but the LUMO of BN is of antibonding character with respect
to the C„N bond and thus weakens this intramolecular bond and
increases the reactivity of BN [114]. Additionally, the orientation
mode changes from a more end-on to a more side-on adsorption
orientation. Interaction of the d-band with the HOMO of BN will
be mostly repulsive because this orbital is low in energy, thus
this phase is stable during hydrogenation at lower pH . In agree-
ment with this conclusion, activation at 2.5 bar and hydrogenation
at 10 bar gave similar results as activation and hydrogenation at
2
1
2
0 bar. Apparently, if the pH is high enough, then the Pd crystal-
lites transform into a stable b-PdH phase that still determines
the catalyst performance, even if the hydrogenation is carried out
at a lower pH . If no stable b-PdH is formed during the activation
2

J.J.W. Bakker et al. / Journal of Catalysis 274 (2010) 176–191
187
the bonding and antibonding orbitals from coupling to the d-band
will mostly fall below E
(a)
f
.
DBA and BA (and BIBA) are the best electron donors having the
highest HOMOs (Table 3). In contrast to BN, the adsorption of BA
and DBA on Pd is mainly governed by electron donation from their
HOMO orbitals to unoccupied d-orbitals of Pd since the LUMOs of
these amines have unfavorably high energy to sufficiently couple
with the d-band of Pd. The HOMO of BA and DBA is localized on
the lone pair of the nitrogen atom. The interaction is attractive
since the HOMO of BA and DBA are sufficiently high-lying. This en-
sures that the antibonding orbitals originating from coupling of the
C
N
C
C
N
N
C
N
C
N
Pd
I
Pd
II
Pd
Pd Pd Pd Pd Pd
III
IV
V
(b)
f
d-band to the HOMO of BA (or DBA) are pushed above E and be-
come empty. The interaction of the d-band of Pd with the HOMO
of BA and DBA is increased upon a decrease in the width of the
CH
H
2
C
N
NH
2
NH
H
C
HC
N
NH
C
Pd
Pd
σ
Pd
di-
d-band (weaker repulsion) and a decrease in
b-PdH. This is confirmed by TGA (Fig. 8). Operation above the
threshold pH results in more amine surface species than operation
at below the threshold pH . Interaction of the d-band of Pd with
e
d
, as is the case for
Pd
Pd
Nitrene
Pd
π
σ
Carbene
Imines
2
2
Fig. 10. (a) Possible adsorption modes of BN on Pd. (I)
side-on adsorption ( ); (III) di- rehybridized; (IV) further rehybridized; (V) p-
interactions with phenyl and/or C„N group. (b) Possible intermediate surface
species after addition of two hydrogen atoms: nitrene, carbene, and three
r end-on adsorption; (II)
r
+
p
r
the LUMO of the amines will be much less effective than with
the LUMO of BN since the LUMO of the amines is significantly high-
er in energy (Table 3).
differently coordinated imines: r, p, and di-r.
d
In conclusion, a lowering of e and decrease in the width of the
d-band upon the transformation into b-PdH will not only change
the strength and orientation of adsorption to the surface of Pd
but it will have different effects on BN and BA, thereby changing
the adsorption ratio of BN/BA. The phase transformation of Pd re-
sults in a change in electronic and geometrical interactions with
interacting species as schematically depicted in Fig. 9.
the C„N absorbance at wavenumbers below 2231.7 cm 1 are
attributed to adsorption of BN on the Pd crystallites. However, both
electron donation and backdonation occur when BN adsorbs on Pd
crystallites, which causes a red shift of the C„N stretching: elec-
tron backdonation results in a more side-on adsorption mode
ꢀ
ꢀ1
and a weaker C„N bond [122]. The absorbance at 2230 cm is
attributed to an adsorption configuration, dominated by the nitro-
gen lone pair donation bond which is more end-on with relatively
less backdonation (I), and the peak at 2165 cm is attributed to a
more side-on adsorption mode (II) with relatively more backdona-
tion [117]. The differences in the 10- and 20-bar spectrum nicely
supports the above reasoning: the absorbance intensity ratio of
230–2165 cm is much smaller for 10-bar hydrogenation than
for 20-bar hydrogenation because of the transformation of Pd to
b-PdH going from 10 to 20 bar. The larger electron backdonation
of Pd compared to b-PdH results in a more side-on adsorption
and a weaker C„N bond.
4.3.3. Adsorption of BN
The possible coordination states of adsorption of BN on Pd are
ꢀ1
shown in Fig. 10a [115–118]: end-on via
via -donation and -backdonation (II), di-
the C„N group to sp and sp configuration (III, IV) and
r
-donation (I), side-on
rehybridization of
-interac-
r
p
r
2
3
p
tions with the phenyl and/or C„N group (V). The extent of rehy-
bridization is increased when backdonation increases and when
the d-band is spatially more extended, as is the case for Pd com-
pared to b-PdH. The DRIFT spectra shown in Fig. 7 give direct infor-
mation on the modes of coordination of BN to Pd in catalysts used
ꢀ1
2
in hydrogenation at 10 and 20 bar. A blue shift in v(C„N) stretch-
ꢀ1
ing compared to 2231.7 cm of free BN implies strengthening of
the C„N bond and is mostly explained as end-on adsorption via
electron donation from the nitrogen lone pair to acid sites on the
4.3.4. The semi-hydrogenated intermediates
The most abundant semi-hydrogenated intermediate is BI,
although it is not detected in the reaction mixture, because it re-
acts at the surface to BA without desorbing first [49]. As direct
c
2 3
-Al O support [119]. Side-on adsorption, governed by electron
backdonation, is not observed on Al
2
O
3
supports [120,121]. So,
LUMO
HOMO
backdonation
interaction
π-interaction
donation
increases
interaction
decreases
2
γ -Al O3
γ -Al O3
2
2
Fig. 9. Representation of major electron donation and electron withdrawing interactions between Pd, BN, and BA. Note the changing strength of interaction indicated by the
arrow size, the changing orientation to the surface of BN, and the changing ratio of adsorption strengths BN/BA when Pd is transformed into b-PdH above the threshold pH of
0 bar.
2
1

1
88
J.J.W. Bakker et al. / Journal of Catalysis 274 (2010) 176–191
experimental evidence of BI is not reported in the literature, other
possible surface intermediates might be present after partial
hydrogenation of BN with two H atoms, such as nitrenes, carbenes,
We speculate that a multicoordinated side-on adsorbed inter-
mediate (i.e. -adsorbed imine), with interaction between the
delocalized -electrons of the phenyl ring and the Pd surface, is
r/p
p
end-on,
28]. Recently, Schärringer et al. found experimental evidence of
nitrene intermediates in the hydrogenation of CD C„N over
Raney-Co [129]. In addition, interactions of the phenyl group of
any of the intermediates with the Pd surface are possible [118].
For instance, the most stable calculated adsorption mode for BI
on Pd(1 1 1) is via a combination of the C@N and phenyl group
p
-adsorbed and rehybridized imines (Fig. 10b) [2,123–
responsible for the direct hydrogenolysis of BN to TOL (11). The
phenyl group in the vicinity of the C„N group effectively lowers
the activation energy toward TOL by stabilizing this adsorption
mode internally by conjugation and by coordinating the phenyl
group to the Pd surface [118]. This is seen more often in the litera-
ture: the hydrogenation of C@C and C@O bonds of aromatic mole-
cules over Pd catalysts are activated by the presence of a phenyl
group. This phenyl activation implies that the phenyl group adsorbs
on active sites [131,132]. Carbene and nitrene species do not have a
1
3
[
130]. A priori, it is not clear which of the intermediates forms
BA, DBA, and TOL, and speculations are based on work with ali-
phatic nitrile hydrogenation without considering hydrogenolysis.
p-system, so the adsorbate will be more perpendicular orientated
For the nitrene intermediate, the
a
-carbon is already saturated
to the surface and hydrogenolysis to TOL is less probable.
Hydrogenolysis of BN to TOL is much more facile than hydrog-
enolysis of BA to TOL over the same supported Pd catalysts and
with hydrogen and thus not likely to be involved in one of the
above-mentioned mechanisms, so nitrene intermediates will only
form BA. In principle, the other intermediates can form BA, too,
but can also undergo hydrogenolysis or condensation reactions.
The most probable reaction pathways to BA, TOL, and DBA are
shown in Fig. 11.
conditions [133]. The rate of hydrogenolysis of BA (CBA,0 = 520 -
3
mol/m , pH
2
= 10 bar, and T = 80 °C) was 2.6 times lower than for
BN, and the overall hydrogenation activity was 42 times lower.
So, even though BA is an intermediate in the hydrogenolysis of
BN to TOL, there are differences between the mechanism of BN
and BA hydrogenolysis. BA will preferentially adsorb end-on with
the lone pair of the nitrogen atom (highest HOMO). This explains
the lower hydrogenolysis rate of BA compared to BN, since for
hydrogenolysis of BA interaction is needed with both the phenyl
4.3.5. Hydrogenolysis to TOL
The main byproduct in the hydrogenation of BN was TOL, but
above the threshold pH a sharp decrease in STOL was observed as
2
shown in Fig. 3e. From the discussion above, this can be under-
stood from the formation of b-PdH with its relatively low backdo-
nation to the adsorbates, thereby reducing the activation of the
C„N bond.
and NH
Fig. 11, we propose a
2
group or with the phenyl ring only [134–137]. In
-benzyl complex as intermediate toward
p
breakage of the C–N bond in BA. An alternative mechanism is that
H
2
C
NH2
H2
Pd
Pd
Pd
Pd
Pd
Pd
H2
H2
C
N
CH3
H2
2H2
Pd
Pd
Pd
NH3
H2
H2
Pd
Pd
Pd
H
2
Pd
C
NH2
H
2
H
N
C
NH2
C
H
C
H
2
2
H N
H2
Pd
Pd
Pd
NH3
H
N
C
C
H
H
2
2
Fig. 11. Reaction scheme of the hydrogenation of BN showing the reaction pathways toward the products BA, DBA, and TOL. This scheme shows which semi-hydrogenated
intermediates are most probable to form BA, DBA, and TOL and shows the multiple pathways to TOL: via BN (direct hydrogenolysis) and via readsorbed BA and DBA (BA and
DBA hydrogenolysis). Note that all semi-hydrogenated intermediates (carbene, imine and nitrene) can form BA (not indicated). The transformation of Pd to b-PdH above the
2
threshold pH decreases the hydrogenolysis to TOL and increases the condensation to DBA.

J.J.W. Bakker et al. / Journal of Catalysis 274 (2010) 176–191
189
BA first partially dehydrogenates to a hydrogen-deficient interme-
diate (i.e. di- rehybridized semi-hydrogenated intermediate) fol-
palladium. The changed electronic properties of palladium may
result in drastic changes in adsorption strengths and modes and
consequently in activity and selectivity.
r
lowed by the breakage of the C–N bond, similar to that proposed
for the hydrogenolysis of methylamines and alkanes [86,138,139].
Note that we propose that DBA forms TOL and BA via a similar
mechanism. Furthermore, TOL is formed immediately after the
reaction was started (e.g. Fig. 2a), which is additional evidence
for a direct hydrogenolysis route of BN to TOL via surface interme-
diates without intermediates (i.e. BA and DBA) desorption
– The main product was benzylamine over the whole investigated
pressure range. Byproducts were toluene and dibenzylamine. A
yield of benzylamine up to 90% was reached without any
additives.
– Palladium completely transforms into palladium b-hydride
above the threshold hydrogen pressure of 10 bar, identified by
ex situ XRD. The expansion of the palladium lattice caused by
hydrogen incorporation was irreversible at the applied
conditions.
– The activation pressure influences the performance of the cata-
lyst if the pressure is high enough to transform palladium into
palladium b-hydride during activation. This palladium b-
hydride, which is formed during activation, is then stable during
hydrogenation at lower hydrogen pressure and results in equal
activity and selectivities as hydrogenation at higher hydrogen
pressures.
[
140,141]. Hydrogenation of an equimolar mixture of BN and BA
(
pH = 10 bar and T = 80 °C) gave 1.7 times lower overall hydroge-
2
nation activity and 3.4 times lower hydrogenolysis rate resulting in
a higher SBA. This shows that BN and BA adsorb competitively on
the same active sites but BN adsorbs stronger than BA. Still, BA
blocks active sites for BN hydrogenolysis by reducing the probabil-
ity for multicoordinated side-on adsorption. This, together with the
slower BA hydrogenolysis rate compared to BN hydrogenolysis, ac-
counts for the lower STOL
.
4.3.6. Condensation to DBA
–
The palladium b-hydride phase has a lower benzonitrile hydro-
genation activity attributed to electronic modifications of the d-
band of palladium upon the transformation to palladium that
decreases the electron backdonation to benzonitrile, resulting
in a less-activated nitrile group.
Secondary amine formation is usually explained by the nucleo-
philic attack of a primary amine with the lone pair on the nitrogen
atom on an electrophilic carbon atom of semi-hydrogenated inter-
mediate (i.e. imine, carbene, and rehybridized species). The initially
s-shaped curve of DBA formation (Fig. 2a) is typical for a consecutive
reaction, i.e., the bimolecular reaction between a semi-hydroge-
nated intermediate and BA. Carbene and to a lesser extent imine
intermediates have the highest probability to form DBA because
of the unsaturated electrophilic carbon atom (Fig. 11). BIBA is as-
sumed to be the intermediate, although it has never been observed
–
The changed electronic properties of palladium upon the forma-
tion of the palladium b-hydride modify the modes and strength
of benzonitrile adsorption, thereby influencing the type and
amount of intermediates, resulting in a changed selectivity:
o Hydrogenolysis to toluene is decreased above the thresh-
old pressure.
[
142].
The low SDBA, over the whole investigated pH
o Condensation to dibenzylamine is increased above the
threshold pressure.
2
range, can be
explained by the low nucleophilicity of BA, caused by conjugation
effects that delocalize the electron density of nitrogen. It is striking
that DBA is the main product when BN is hydrogenated over sup-
ported Pt catalysts (Table 1). Clearly, the catalyst plays a crucial
role in determining selectivity. The difference between Pt and Pd
–
Toluene can be formed directly from benzonitrile and indirectly
from readsorbed benzylamine and dibenzylamine. Both direct
and indirect hydrogenolysis reactions proceed via different
pathways: the rate of hydrogenolysis of benzylamine is slower
than that of benzonitrile. Interaction via the p-electrons of the
could be explained as follows. Compared to Pt (with a lower E
f
, a
aromatic ring is a prerequisite for facile hydrogenolysis. For pal-
ladium b-hydride, this interaction is less than for palladium.
The formation of palladium b-hydride resulted in a higher
amount of benzylamine on the surface due to the higher adsorp-
tion strength and a decreased benzylamine hydrogenolysis,
explaining a higher degree of condensation to dibenzylamine.
The condensation to dibenzylamine is further enhanced by the
lower production of inhibiting ammonia and by the decreased
electron backdonation, yielding surface intermediates that are
more susceptible to an attack from nucleophiles.
lower , and a more extended d-band compared to Pd), the Pd sur-
e
d
face is populated with semi-hydrogenated intermediates that are
much less susceptible to condensation, i.e., nitrenes, and the subse-
quent desorption of BA is easier. Furthermore, the fast rate of
hydrogenation to BA of these intermediates compared to the slow
formation from BN result in a low surface coverage.
–
–
TOL is the main byproduct: YTOL is always higher than YDBA
2
(Fig. 3c). However, above the threshold pH , SDBA becomes higher
than STOL (Fig. 3e). This is attributed to an increased amount of BA
on the surface of b-PdH compared to Pd because of the decreased
Pauli repulsion and the lower
d
e . The decreased STOL implies also a
lower NH content (NH suppresses the condensation to DBA [4]).
3
3
Acknowledgments
Moreover, the decreased electron backdonation of b-PdH to an
intermediate results in a more electropositive carbon atom, which
will be more susceptible to a nucleophilic attack by BA to DBA.
Ferdinand C. Grozema of the Opto-electronic Materials section
of the Delft University of Technology is acknowledged for fruitful
discussions.
5
. Conclusions
Appendix A. Supplementary material
The influence of hydrogen pressure on the semi-batch hydro-
genation of benzonitrile to benzylamine over a
c-alumina-sup-
Supplementary material regarding mass transport limitations
and details of the used characterization techniques can be found,
in the online version, at doi:10.1016/j.jcat.2010.06.013.
ported palladium catalyst in 2-propanol was investigated to
explain the maximum in activity and the selectivity change
around a certain hydrogen threshold pressure. An important mes-
sage of this study for practical operation of selective hydrogena-
tions over palladium catalysts is the careful selection of the
operational and activation pressure. An optimum must be sought
between activity and selectivity and the active-phase structure of
References
[
1] S. Nishimura, Handbook of Heterogeneous Catalytic Hydrogenation for
Organic Synthesis, Wiley-VCH, New York, 2001. pp. 254–285.

1
90
J.J.W. Bakker et al. / Journal of Catalysis 274 (2010) 176–191
[
[
2] C. De Bellefon, P. Fouilloux, Catal. Rev. Sci. Eng. 36 (1994) 459–506. [61] S. Huang, C. Huang, B. Chang, C. Yeh, J. Phys. Chem. B 110 (2006) 21783–
3] J. Volf, J. Pasek, in: L. Cerveny (Ed.), Catalytic Hydrogenation, Elsevier,
Amsterdam, 1986, p. 105.
21787.
[62] A.R. Denton, N.W. Ashcroft, Phys. Rev. A 43 (1991) 3161–3164.
[63] E.F. Skelton, P.L. Hagans, S.B. Qadri, D.D. Dominguez, A.C. Ehrlich, J.Z. Hu, Phys.
Rev. B 58 (1998) 14775–14779.
[
4] S. Gomez, J.A. Peters, T. Maschmeyer, Adv. Synth. Catal. 344 (2002) 1037–
1
057.
[
[
[
[
[
5] P. Sabatier, J.R. Senderens, C.R. Hebd, Seances Acad. Sci. 140 (1905) 482.
6] G. Mignonac, C.R. Hebd, Seances Acad. Sci. 171 (1920) 114–117.
7] J. von Braun, G. Blessing, F. Zobel, Chem. Ber. 36 (1923) 1988–2001.
8] K. Kindler, F. Hesse, Arch. Pharm. 27 (1933) 439–445.
[64] Z. Kaszkur, J. Appl. Crystallogr. 33 (2000) 1262–1270.
[65] C.M. Mendez, H. Olivero, D.E. Damiani, M.A. Volpe, Appl. Catal. B 84 (2008)
156–161.
[66] N.S. Babu, N. Lingaiah, R. Gopinath, P.S. Sankar Reddy, P.S. Sai Prasad, J. Phys.
Chem. C 111 (2007) 6447–6453.
9] H. Greenfield, Ind. Eng. Chem. Prod. Res. Dev. 15 (1976) 156–158.
[
[
[
10] Y. Huang, W.M.H. Sachtler, J. Catal. 184 (1999) 247–261.
11] P. Rylander, Catalytic Hydrogenation in Organic Syntheses, 1979, p. 138.
12] P.N. Rylander, L. Hasbrouck, I. Karpenko, Ann. NY Acad. Sci. 214 (1973) 100–
[67] J. Panpranot, K. Pattamakomsan, P. Praserthdam, J.G. Goodwin, Ind. Eng.
Chem. Res. 43 (2004) 6014–6020.
[68] J. Panpranot, O. Tangjitwattakorn, P. Praserthdam, J.G. Goodwin Jr., Appl.
Catal. A 292 (2005) 322–327.
1
09.
[
[
[
[
13] P.N. Rylander, J.G. Kaplan, US Patent 3117,162, 1964.
14] S.P. Bawane, S.B. Sawant, Chem. Eng. J. (Lausanne) 103 (2004) 13–19.
15] L. Hegedus, T. Máthé, Appl. Catal. A 296 (2005) 209–215.
[69] H. Lieske, J. Voelter, J. Phys. Chem. 89 (1985) 1841–1842.
[70] M. Gurrath, T. Kuretzky, H.P. Boehm, L.B. Okhlopkova, A.S. Lisitsyn, V.A.
Likholobov, Carbon 38 (2000) 1241–1255.
[71] B. Wen, Q. Sun, W.M.H. Sachtler, J. Catal. 204 (2001) 314–323.
[72] M. Bonarowska, J. Pielaszek, W. Juszczyk, Z. Karpinski, J. Catal. 195 (2000)
304–315.
[73] H. Knoezinger, H. Krietenbrink, J. Chem. Soc. Faraday Trans. 71 (1975) 2421–
2430.
[74] R.A. Nyquist, Appl. Spectrosc. 44 (1990) 1405–1407.
[75] H. Sato, Y. Kusumoto, S. Arase, M. Suenaga, S. Kammura, J. Phys. Chem. 82
(1978) 66–68.
[76] P. Hirva, T.A. Pakkanen, Surf. Sci. 277 (1992) 389–394.
[77] M. Skotak, Z. Karpinski, W. Juszczyk, J. Pielaszek, L. Kepinski, D.V. Kazachkin,
V.I. Kovalchuk, J.L. d’Itri, J. Catal. 227 (2004) 11–25.
[78] A. Quintanilla, J.J.W. Bakker, M.T. Kreutzer, J.A. Moulijn, F. Kapteijn, J. Catal.
257 (2008) 55–63.
16] H. Bernas, A. Taskinen, J. Wärnå, D.Y. Murzin, J. Mol. Catal. A: Chem. 306
(
2009) 33–39.
17] V. Dubois, G. Jannes, J.L. Dallons, A. Van Gysel, in: J.R. Kosak, T.A. Johnson
Eds.), Catalysis of Organic Reactions, Marcel Dekker Inc., 1994, p. 1.
[
[
(
18] E.V. Skakunova, M.M. Ermilova, V.M. Gryaznov, Izv. Akad. Nauk Gruz. SSR, Ser.
Khim (1988) 986–991.
19] J.H. Sinfelt, Catal. Lett. 9 (1991) 159–171.
20] J.H. Sinfelt, Cat. Rev. 3 (1969) 175–205.
21] J.H. Sinfelt, Adv. Catal. 23 (1973) 91–119.
22] J.H. Sinfelt, J.L. Carter, D.J.C. Yates, J. Catal. 24 (1972) 283–296.
23] J.H. Sinfelt, D.J.C. Yates, J. Catal. 8 (1967) 82–90.
24] T. Graham, Philos. Trans. Roy. Soc. 156 (1866) 399–439.
25] L.L. Jewell, B.H. Davis, Appl. Catal. A 310 (2006) 1–15.
26] N.K. Nag, J. Phys. Chem. B 105 (2001) 5945–5949.
27] R.J. Davis, S.M. Landry, J.A. Horsley, M. Boudart, Phys. Rev. B 39 (1989) 10580–
[
[
[
[
[
[
[
[
[
[79] W.J. Kim, E.W. Shin, J.H. Kang, S.H. Moon, Appl. Catal. A 251 (2003) 305–
313.
1
0583.
[80] P. Marécot, A. Akhachane, J. Barbier, Catal. Lett. 36 (1996) 37–39.
[81] J.-W. Park, Y.-M. Chung, Y.-W. Suh, H.-K. Rhee, Catal. Today 93–95 (2004)
445–450.
[
28] K. Christmann, in: Z. Paal, P.G. Menon (Eds.), Hydrogen Effects in Catalysis.
Fundamentals and Practical Applications, Marcel Dekker Inc., New York,
1
988, p. 44.
29] I.P. Chernov, Y.M. Koroteev, V.M. Silkin, Y.I. Tyurin, Dokl. Phys. 53 (2008) 318–
22.
30] M. Yamauchi, R. Ikeda, H. Kitagawa, M. Takata, J. Phys. Chem. C 112 (2008)
294–3299.
[82] M.R. Othman, I.S. Sahadan, Micropor. Mesopor. Mater. 91 (2006) 145–150.
[83] F. Figueras, B. Coq, J. Mol. Catal. A: Chem. 173 (2001) 223–230.
[84] G. Meitzner, W.J. Mykytka, J.H. Sinfelt, J. Catal. 98 (1986) 513–521.
[85] G. Meitzner, W.J. Mykytka, J.H. Sinfelt, Catal. Lett. 37 (1996) 137–141.
[86] G. Meitzner, W.J. Mykytka, J.H. Sinfelt, Catal. Lett. 32 (1995) 335–344.
[87] L. Guczi, A. Sarkany, P. Tetenyi, J. Chem. Soc. Faraday Trans. 70 (1974) 1971–
1981.
[
[
3
3
[
[
[
31] C.T. Chan, S.G. Louie, Phys. Rev. B 27 (1983) 3325–3337.
32] J.S. Faulkner, Phys. Rev. B 13 (1976) 2391–2397.
33] F. Pinna, M. Signoretto, G. Strukul, S. Polizzi, N. Pernicone, React. Kinet. Catal.
Lett. 60 (1997) 9–13.
34] M. Boudart, H.S. Hwang, J. Catal. 39 (1975) 44–52.
35] A.J. Bird, D.T. Thompson, in: W.H. Jones (Ed.), Catalysis in Organic Syntheses,
Academic Press, New York, 1980, p. 61.
36] A. Valcarcel, F. Morfin, L. Piccolo, J. Catal. 263 (2009) 315–320.
37] D. Teschner, J. Borsodi, A. Wootsch, Z. Revay, M. Havecker, A. Knop-Gericke,
S.D. Jackson, R. Schlogl, Science 320 (2008) 86–89.
[88] A. Pundt, R. Kirchheim, Annu. Rev. Mater. Res. 36 (2006) 555–608.
[89] A. Pundt, M. Suleiman, C. Bähtz, M.T. Reetz, R. Kirchheim, N.M. Jisrawi, Mater.
Sci. Eng. B 108 (2004) 19–23.
[
[
[90] P.C. Aben, J. Catal. 10 (1968) 224–229.
[91] I.T. Caga, E. Shutt, J.M. Winterbottom, J. Catal. 44 (1976) 271–280.
[92] E.A. Owen, J.I. Jones, Proc. Phys. Soc. 49 (1937) 603–610.
[93] S.B. Ziemecki, G.A. Jones, J. Catal. 95 (1985) 621–622.
[94] E.J.A.X. van de Sandt, A. Wiersma, M. Makkee, H. van Bekkum, J.A. Moulijn,
Appl. Catal. A 155 (1997) 59–73.
[95] D. Teschner, E. Vass, M. Havecker, S. Zafeiratos, P. Schnorch, H. Sauer, A. Knop-
Gericke, R. Schloegl, M. Chamam, A. Wootsch, A.S. Canning, J.J. Gamman, S.D.
Jackson, J. McGregor, L.F. Gladden, J. Catal. 242 (2006) 26–37.
[96] D. Teschner, J. Borsodi, Z. Kis, L. Szentmiklosi, Z. Revay, A. Knop-Gericke, R.
Schlogl, D. Torres, P. Sautet, J. Phys. Chem. C 114 (2010) 2293–2299.
[97] N. Seriani, F. Mittendorfer, G. Kresse, J. Chem. Phys. 132 (2010) 1–8.
[98] P. Sautet, F. Cinquini, ChemCatChem 2 (2010) 636–639.
[99] M. García-Mota, B. Bridier, J. Pérez-Ramírez, N. López, J. Catal. (2010),
doi:10.1016/j.jcat.2010.1004.1018.
[100] J. Andersin, N. Lopez, K. Honkala, J. Phys. Chem. C 113 (2009) 8278–8286.
[101] D.A. Papaconstantopoulos, B.M. Klein, J.S. Faulkner, L.L. Boyer, Phys. Rev. B 18
(1978) 2784–2791.
[102] S. Mizusaki, N. Hiraoka, I. Yamamoto, M. Itou, Y. Sakurai, M. Yamaguchi, J.
Phys. Soc. Jpn. 72 (2003) 1145–1151.
[103] D.E. Eastman, J.K. Cashion, A.C. Switendick, Phys. Rev. Lett. 27 (1971) 35.
[104] W. Jaworski, J. Phys. F: Met. Phys. 17 (1987) 373–381.
[105] M. Gupta, A.J. Freeman, Phys. Rev. B 17 (1978) 3029.
[106] B. Hammer, J.K. Nørskov, Surf. Sci. 343 (1995) 211–220.
[107] B. Hammer, J.K. Nørskov, Nature 376 (1995) 238–240.
[108] A. Ruban, B. Hammer, P. Stoltze, H.L. Skriver, J.K. Nørskov, J. Mol. Catal. A:
Chem. 115 (1997) 421–429.
[109] P. Liu, J.K. Nørskov, Phys. Chem. Chem. Phys. 3 (2001) 3814–3818.
[110] P. Gallezot, D. Richard, Catal. Rev. Sci. Eng. 40 (1998) 81–126.
[111] B.A. Averill, I.M.C.M. Rietjens, P.W.N.M. Van Leeuwen, R.A. Van Santen, Stud.
Surf. Sci. Catal. 123 (1999) 109–208.
[
[
[
[
38] D. Teschner, Z. Révay, J. Borsodi, M. Hävecker, A. Knop-Gericke, R. Schlögl, D.
Milroy, S.D. Jackson, D. Torres, P. Sautet, Angew. Chem. 120 (2008) 9414–
9
418.
39] F. Studt, F. Abild-Pedersen, T. Bligaard, Rasmus, Z. Sørensen, Claus, H.
Christensen, Jens K. Nørskov, Angew. Chem. Int. Ed. 47 (2008) 9299–9302.
40] M.P. Jigato, B. Coussens, D.A. King, J. Chem. Phys. 118 (2003) 5623–5634.
41] Á. Mastalir, Z. Király, F. Berger, Appl. Catal. A 269 (2004) 161–168.
42] G. Carturan, G. Facchin, G. Cocco, S. Enzo, G. Navazio, J. Catal. 76 (1982) 405–
[
[
[
4
17.
[
[
43] H. Greenfield, and R.S. Sekellick, US Patent 3923,891, 1975.
44] O.G. Degischer, F. Roessler, P. Rys, Chem. Ind.: Catal Org. React. 82 (2001)
2
41–254.
[
45] Y.M. Lopez-De Jesus, A. Vicente, G. Lafaye, P. Marecot, C.T. Williams, J. Phys.
Chem. C 112 (2008) 13837–13845.
46] T. Fuchigami, S. Takamizawa, and N. Wakasa, US Patent 6476,267 B1, 2002.
47] M.B. Dines, P.M. DiGiacomo, and K.P. Callahan, US Patent 4384,981, 1983.
48] D. Ostgard, M. Berweiler, and S. Roder, US Patent 0173676 A1, 2002.
49] D.J. Ostgard, Spec. Chem. Mag. 28 (2008) 28–31.
[
[
[
[
[
[
50] K. Hata, K. Watanabe, Bull. Chem. Soc. Jpn. 32 (1959) 861–867.
51] S. Galvagno, A. Donato, G. Neri, R. Pietropaolo, J. Mol. Catal. 58 (1990) 215–
2
25.
[
[
52] R.A. Plunkett, J.L. Neff, and T.A. Bemish, US Patent 4163,025, 1979.
53] S. Gobolos, N. Mahata, I. Borbath, M. Hegedus, J.L. Margitfalvi, React. Kinet.
Catal. Lett. 74 (2001) 345–352.
54] O. Domínguez-Quintero, S. Martínez, Y. Henríquez, L. D’Ornelas, H. Krentzien,
J. Osuna, J. Mol. Catal. A: Chem. 197 (2003) 185–191.
[
[
55] H. Paul, S. Basu, S. Bhaduri, G.K. Lahiri, J. Organomet. Chem. 689 (2004) 309–
[112] G. Blyholder, J. Phys. Chem. 68 (1964) 2772–2777.
[113] T. Bligaard, J.K. Nørskov, Chemical Bonding at Surfaces and Interfaces,
Elsevier, Amsterdam, 2008. p. 255.
[114] I. Chorkendorff, J.W. Niemantsverdriet, Concepts of Modern Catalysis and
Kinetics, WILEY-VCH, Weinheim, 2003. p. 215.
3
16.
[
[
56] A.W. Heinen, J.A. Peters, H. Van Bekkum, Eur. J. Org. Chem. (2000) 2501–2506.
57] S. Enthaler, K. Junge, D. Addis, G. Erre, M. Beller, ChemSusChem 1 (2008)
1
006–1010.
[
[
58] C. Amorim, M.A. Keane, J. Colloid Interface Sci. 322 (2008) 196–208.
59] G. Fagherazzi, A. Benedetti, S. Polizzi, A. Mario, F. Pinna, M. Signoretto, N.
Pernicone, Catal. Lett. 32 (1995) 293–303.
[115] K. Kishi, F. Kikui, S. Ikeda, Surf. Sci. 99 (1980) 405–418.
[116] T. Nakayama, K. Inamura, Y. Inoue, S. Ikeda, K. Kishi, Surf. Sci. 179 (1987) 47–
58.
[
60] J. Sa, G.D. Arteaga, R.A. Daley, J. Bernardi, J.A. Anderson, J. Phys. Chem. B 110
[117] O. Oranskaya, I. Semenskaya, V. Filimonov, React. Kinet. Catal. Lett. 5 (1976)
135–139.
(
2006) 17090–17095.

J.J.W. Bakker et al. / Journal of Catalysis 274 (2010) 176–191
191
[
118] S. Zou, C.T. Williams, E.K.Y. Chen, M.J. Weaver, J. Phys. Chem. B 102 (1998)
039–9049.
[130] L. Hegedus, T. Máthé, T. Kárpáti, Appl. Catal. A 349 (2008) 40–45.
[131] J. Jenck, J.-E. Germain, J. Catal. 65 (1980) 141–149.
9
[
[
[
119] K.F. Purcell, R.S. Drago, J. Am. Chem. Soc. 88 (1966) 919–924.
120] E. Paukstis, E.N. Yurchenko, Usp. Khim. 52 (1983) 426–454.
121] O.M. Oranskaya, V.N. Filimonov, E.Y. Shmulyakovskii, Vopr. Mol. Spektrosk.
[132] J.L. Dallons, G. Jannes, B. Delmon, Catal. Today 5 (1989) 257–264.
[133] W.F. Maier, P. Grubmüller, I. Thies, P.M. Stein, M.A. McKervey, P.v.R. Schleyer,
Angew. Chem. Int. Ed. 18 (1979) 939–940.
(
1974) 195–198.
[134] S. Mitsui, Y. Sugi, Tetrahedron Lett. 10 (1969) 1291–1294.
[135] Y. Sugi, S. Mitsui, Tetrahedron 29 (1973) 2041–2045.
[136] C.Ó. Murchú, Tetrahedron Lett. 10 (1969) 3231–3234.
[137] K. Ikedate, S. Suzuki, Nippon Kagaku Kaishi (1975) 234–241.
[138] D.F. Johnson, Y. Wang, J.E. Parmeter, M.M. Hills, W.H. Weinberg, J. Am. Chem.
Soc. 114 (1992) 4279–4290.
[
[
122] J. Raskó, J. Kiss, Appl. Catal. A. 298 (2006) 115–126.
123] A. Chojecki, H. Jobic, A. Jentys, T.E. Müller, J.A. Lercher, Catal. Lett. 97 (2004)
1
55–162.
[
124] A. Chojecki, M. Veprek-Heijman, T.E. Müller, P. Schärringer, S. Veprek, J.A.
Lercher, J. Catal. 245 (2007) 237–248.
[
[
[
[
[
125] B. Coq, D. Tichit, S. Ribet, J. Catal. 189 (2000) 117–128.
[139] J.R. Anderson, N.J. Clark, J. Catal. 5 (1966) 250–263.
[140] M. Jian, F. Kapteijn, R. Prins, J. Catal. 168 (1997) 491–500.
[141] T. Vergunst, F. Kapteijn, J.A. Moulijn, Catal. Today 66 (2001) 381–387.
[142] Y. Huang, W.M.H. Sachtler, Appl. Catal. A 182 (1999) 365–378.
126] B. Bigot, F. Delbecq, A. Milet, V.-H. Peuch, J. Catal. 159 (1996) 383–393.
127] P. Schärringer, T.E. Müller, J.A. Lercher, J. Catal. 253 (2008) 167–179.
128] J. Krupka, J. Patera, Appl. Catal. A 330 (2007) 96–107.
129] P. Schärringer, T.E. Müller, A. Jentys, J.A. Lercher, J. Catal. 263 (2009) 34–41.