Catalytic hydrogenation of tertiary amides at low temperatures and pressures using bimetallic Pt/Re-based catalysts

Article abstract of DOI:10.1016/j.jcat.2011.07.007

Hydrogenation of tertiary amides, in particular, N-methylpyrrolidin-2-one, can be efficiently facilitated by a TiO₂-supported bimetallic Pt/Re catalyst at low temperatures and pressures. Characterisation of the catalysts and kinetic tests have shown that the close interaction between the Re and Pt is crucial to the high activity observed. DFT calculations were used to examine a range of metal combinations and show that the role of the uncoordinated Re is to activate the CO and that of the Pt is as a hydrogenation catalyst, removing intermediates from the catalyst surface. The rate enhancement observed on the TiO₂ support is thought to be due to the presence of oxygen vacancies allowing adsorption and weakening of the CO bond.

Full text of DOI:10.1016/j.jcat.2011.07.007

Journal of Catalysis 283 (2011) 89–97
Contents lists available at SciVerse ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Catalytic hydrogenation of tertiary amides at low temperatures and pressures
using bimetallic Pt/Re-based catalysts
⇑
R. Burch, C. Paun, X.-M. Cao, P. Crawford, P. Goodrich, C. Hardacre, P. Hu , L. McLaughlin, J. Sá,
J.M. Thompson
⇑
School of Chemistry and Chemical Engineering, Queen’s University, Stranmillis Road, Belfast, Northern Ireland BT9 5AG, UK
a r t i c l e i n f o
a b s t r a c t
Article history:
Hydrogenation of tertiary amides, in particular, N-methylpyrrolidin-2-one, can be efficiently facilitated
by a TiO2-supported bimetallic Pt/Re catalyst at low temperatures and pressures. Characterisation of
the catalysts and kinetic tests have shown that the close interaction between the Re and Pt is crucial
to the high activity observed. DFT calculations were used to examine a range of metal combinations
and show that the role of the uncoordinated Re is to activate the C@O and that of the Pt is as a hydroge-
nation catalyst, removing intermediates from the catalyst surface. The rate enhancement observed on the
TiO2 support is thought to be due to the presence of oxygen vacancies allowing adsorption and weaken-
ing of the C@O bond.
Received 1 May 2011
Revised 19 July 2011
Accepted 20 July 2011
Available online 27 August 2011
Keywords:
Platinum
Rhenium
Bimetallic
Titania
Ó 2011 Elsevier Inc. All rights reserved.
Amide
DFT
1. Introduction
amides using hydrogen gas in the presence of homogeneous or het-
erogeneous catalysts including copper chromite [6], rhenium metal
Amines are an important class of organic compounds, with a
wide range of applications including as dyes, solvents, additives,
anti-foam agents, corrosion inhibitors, detergents and drugs.
Tertiary amines have found applications as catalysts for use in
the urethane industry, e.g. N,N-dimethylcyclohexylamine or alkyl-
dimethylamine with typical applications as quaternary derivatives,
amine oxides and betaines, which can be further used in the man-
ufacture of disinfectants, wood treatment, personal care, oilfields
and water treatment. As a result of the large number of applica-
tions, the development of new methods for the production of
amines has been the focus of a large number of research studies.
A potentially simple method of forming amines is via the reduc-
tion of amides, which can be easily obtained from the reaction of
carboxylic acids, acid anhydrides or acid chlorides with primary
or secondary amines. However, amide functional groups are con-
sidered difficult to reduce [1] and, to date, catalytic reduction of
amides has been problematic. Commonly, the reduction of amides
is achieved using stoichiometric reducing agents such as lithium
aluminium hydride, for primary and secondary amides [2], or bor-
ane [3–5], for secondary and tertiary amides, resulting in signifi-
cant amounts of waste and issues with the workup. Reduction of
[7] and group 8–10 transition metals [8,9] has been shown to be
successful, but typically has required high temperatures of up to
260 °C and high pressures of 100–300 bar. In these cases, the rate
and selectivity has been found to be strongly dependent on the
structure of the amide, the metal catalyst and reduction conditions.
Dobson [10] reported the reduction of amides to form the cor-
responding amine with hydrogen at elevated temperature (175–
225 °C) and moderate to high pressures (27.6–137.9 bar) using
group 8/Re bimetallic catalysts on a range of carbon and silica sup-
ports. Although some good conversions were found, the selectivi-
ties were variable. Hirosawa et al. [11] also showed that the
synergistic effects of rhenium, molybdenum or tungsten with rho-
dium or ruthenium facilitated amide reduction in both homoge-
neous and heterogeneous catalytic systems. In this case, reaction
conditions of >150 °C and 100 bar hydrogen were employed with
high selectivities obtained. Smith et al. [12] compared a range of
bi- and trimetallic catalysts (consisting of combinations of A, B, C
where A is chosen from Co, Fe, Ir, Pt, Rh or Ru; B from Cr, Mo, Re
and V; and C from Cu, In and Zn) at temperatures below 200 °C
and at pressures below 50 bar hydrogen (6–10 bar). Although some
of the catalysts provided good selectivity, low conversions were
found if the temperature was below 100 °C.
Other examples of amide hydrogenation include the use of cop-
per chromite with a nucleophilic reagent [13] in the preparation of
primary, secondary and tertiary amines from fatty amides. Using a
⇑
Corresponding authors. Fax: +44 (0)28 9097 4687.
E-mail addresses: p.hu@qub.ac.uk (P. Hu), jillian.thompson@qub.ac.uk
(J.M. Thompson).
0021-9517/$ - see front matter Ó 2011 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2011.07.007

90
R. Burch et al. / Journal of Catalysis 283 (2011) 89–97
2:1 ratio of CuO:Cr2O3 along with sodium methoxide, the authors
obtained91–99%yieldand91–98%selectivityathydrogenpressures
of 10–27 bar and 250 °C. Very mild reaction conditions can be em-
ployed for amide hydrogenation using PtO2 in dilute HCl, with
hydrogenation of N-methylpyrrolidin-2-one occurring at room tem-
perature. However, a large amount of catalyst is needed, and the cat-
alyst cannot be easily recycled due to the formation of Magnus’ salt
([Pt(NH3)4][PtCl4]) [9]. Recently, Beamson et al. [14] have reported
the use of a range of heterogeneous Rh/Mo, Ru/Re and Rh/Re bime-
tallic catalysts, derived from zero-valent metal carbonyls for the
selective hydrogenation of primary amides such as cyclohexane car-
boxamide. Therein, synergistic behaviour of the metals provided
activity for the hydrogenation; however, pressures below 50 bar
H2 and temperatures 6120 °C resulted in low conversions and selec-
tivities. In addition, hydrogenation of amides to form alcohols and
amines has recently been reported using homogeneous ruthenium
complexes at 10 bar H2 and 110 °C in THF [15].
Here, we report the use of bimetallic Pt–Re catalysts for the
hydrogenation of the amide N-methylpyrrolidinone Scheme 1 un-
der mild reaction conditions (120–160 °C and 5–30 bar hydrogen).
The relationship between the structural characterisation of these
catalysts and kinetic experimental results is presented along with
DFT calculations providing a rationale for the high activity ob-
served with this bimetallic catalyst.
chased from Aldrich in P97% purity and were used without
further purification. The gases, H2, 5% H2 in Ar and N2, used in this
work were purchased from BOC as >99% pure gases. Platinum ni-
trate (assay 15.14%) and perrhenic acid (assay 39.4%) solutions
were supplied by Johnson Matthey. Alumina (c type), ceria-zirco-
nia (zirconium doped cubic ceria) and titania P90 were used as re-
ceived from Ketjen Catalysts, Johnson Matthey and Nippon Aerosil,
respectively.
The monometallic Pt and Re catalysts were prepared by incipi-
ent wetness impregnation. An appropriate mass of the required
precursor, perrhenic acid solution or platinum nitrate solution, to
give the desired wt% metal loading was diluted with sufficient dou-
bly distilled deionised water (18 MX) to give a volume equal to the
pore volume of the support used. The solution was added to the
support in three portions of equal volume with stirring after each
addition until thoroughly mixed. The product was then dried at
120 °C for 12 h followed by calcination at 500 °C for 4 h. The bime-
tallic catalysts were prepared by sequential incipient wetness
impregnation (S.I.) or co-impregnation by incipient wetness (C.I.).
In the case of sequential impregnation, following deposition of
one metal, the catalyst was dried and calcined before the process
was repeated for the second metal. After this deposition, the cata-
lyst was again dried and calcined.
2.2. Characterisation
2. Experimental
The surface area and pore volumes of the catalysts obtained by
BET N2 isotherm (Micromeritics ASAP 2010) are shown in Table 1.
The metal loading of all the catalysts prepared (with the nominal
wt% loadings given in the catalyst descriptor) was determined by
ICP–OES (Perkin–Elmer Optima 4300), and the results are
2.1. Catalysts and materials
All the chemicals, N-methylpyrrolidinone, hexane, tetrahydro-
furan, diethylether, methanol, methyl tert-butyl ether, were pur-
O
N
C
O
+
N
N-methylpyrrolidin-2-ylidene
N-methylpyrrolidin-2-one
(MPDYE)
(MPDO)
+H
+H
N
2H2
OH
CH
N-methylpyrrolidinyl
(MPDY)
+H
+H
+
H2O
N
N-methylpyrrolidine
(MPD)
Scheme 1. Schematic reaction pathway for the catalytic hydrogenation of N-methylpyrrolidin-2-one (MPDO) to N-methylpyrrolidine (MPD) and water via an initial C@O
cleavage followed by hydrogenation. (Possible intermediates: N-methylpyrrolidin-2-ylidene (MPDYE), N-methylpyrrolidinyl (MPDY).)

R. Burch et al. / Journal of Catalysis 283 (2011) 89–97
91
Table 1
Physical properties of the supports used and catalysts prepared.
Catalyst
Prep. methoda
BET surface area (m2 gꢀ1
)
Pore volume (cm3 gꢀ1
)
Pt%b
Re%b
Al2O3
CeZrO4
TiO2
–
–
–
238
94
108
226
236
70
65
53
60
55
0.17
0.17
0.18
0.17
0.17
0.17
0.17
0.16
0.15
0.17
0.14
0.22
0.20
0.22
–
–
–
–
–
–
4 wt%Pt–4 wt%Re/Al2O3
S.I.
C.I.
S.I.
C.I.
S.I.
C.I.
S.I.
S.I.
S.I.
–
3.8
3.7
3.8
3.2
4.3
4.3
4.3
4.6
5.4
4
4.3
3.9
3.9
4.1
3.5
4
3.7
1.8
0.5
–
4 wt%Pt–4 wt%Re/CeZrO4
4 wt%Pt–4 wt%Re/TiO2
4 wt%Re–4 wt%Pt/TiO2
4 wt%Pt–1 wt%Re/TiO2
4 wt%Pt–0.25 wt%Re/TiO2
4 wt%Pt/TiO2
38
65
83
63
4 wt%Re/TiO2
–
–
3.8
a
S.I. = sequential impregnation. C.I. = co-impregnation.
Weight% loading as determined by ICP–OES.
b
presented in Table 1. As expected, the results show a decrease in
the surface area of the catalyst after metal deposition.
All the models were simulated as a periodic 4-layer slab with a
ꢁ12 Å vacuum separating neighbouring slabs in the Z direction, in
which adsorbates and top two layers were relaxed and the others
were fixed. The energies of all the intermediates involving two spe-
cies were calculated in two separate unit cells. Forces on relaxed
atoms were less than 0.05 eV Åꢀ1, after the geometry optimisation.
The Re flat surface was modelled by a close-packed (0001) facet. A
p(3 ꢂ 3) supercell was used with corresponding 3 ꢂ 3 ꢂ 1 Monk-
horst–Pack k-point mesh for Brillouin zone sampling. The stepped
surfaces of Pt and Rh were modelled by a (211) facet. A p(1 ꢂ 3)
unit cell was used with a 4 ꢂ 3 ꢂ 1 Monkhorst–Pack k-point mesh.
The stepped surfaces of Re and Ru were modelled by removing two
neighbouring rows of atoms on the top layer of (0001) facets. A
p(5 ꢂ 3) unit cell was used with 2 ꢂ 3 ꢂ 1 Monkhorst–Pack k-point
mesh. The two-phase Re/X (X: Pt, Pd, Rh or Ru) was modelled with
a p(3 ꢂ 3) Re monolayer terrace on a p(5 ꢂ 3) 3-layer X-metal flat
surface substrate. The Re terrace was covered with 1 monolayer
(ML) oxygen excluding the step edge sites in the Re/X interface.
The Brillouin zone was sampled with 2 ꢂ 3 ꢂ 1 Monkhorst–Pack
k-point mesh. The five lowest energy structures from the results
of Nose thermostat molecular dynamics simulation (T = 120 °C,
2 fs per step, 1000 steps) were selected to be further optimised
by conjugate gradient methods for Re/X systems. The lowest en-
ergy structure was then utilised for all subsequent studies. Transi-
tion states (TSs) of reactions were located using the constrained
minimisation method [20–22].
The temperature-programmed reduction (TPR) (Micromeritics
Autochem 2910) was measured by placing approximately 0.1 g of
catalyst in a U-shaped tube, which was cooled to ꢀ50 °C in argon.
The catalyst was reduced using 5% H2 in Ar with the temperature
being ramped from ꢀ50 to 800 °C at a rate of 5 °C minꢀ1 and the
hydrogen uptake monitored by a thermal conductivity detector
(TCD). Powder XRD measurements were obtained using Cu K
radiation (PANalytical X’PERT PRO MPD X-ray diffractometer).
a
2.3. Typical reaction procedure
The hydrogenation reactions were performed in an Autoclave
Engineers 200 cm3 stainless-steel stirred-tank reactor. The reactor
was charged with catalyst (typically 0.2 g – sieved before reaction
to between 50 and 250 l
m) and solvent (15 cm3), and the vessel
purged three times with nitrogen before the reaction mixture
was heated to the reduction temperature, 120 °C, with stirring
(1500 rpm). The autoclave was charged with 20 bar hydrogen pres-
sure, and the reaction mixture was left to stir for 1 h to pre-reduce
the catalyst. Thereafter, the reactor was cooled to room tempera-
ture, depressurised and charged with amide (0.0052 mol), decane
as an internal standard (0.001 mol) together with an additional
15 cm3 of solvent. The reactor was heated under a nitrogen atmo-
sphere to the desired reaction temperature (typically 120 °C) and
pressurised to reaction pressure (typically 20 bar) at which point
the reaction was started by starting the stirrer (typically
1500 rpm) and taking a sample of the liquid phase corresponding
to t = 0. Hydrogen was supplied to the reactor by means of a deliv-
ery system connected to the reactor, thus guaranteeing a constant
pressure of gas in the autoclave throughout the course of the reac-
tion. Samples were taken at regular intervals, and analysed by an
Agilent GC-FID using a DB-1 capillary column and GC–MS (Agilent)
using a RTX-5 capillary column. Unless otherwise stated, the cata-
lysts used/characterised were those prepared by sequential
impregnation using platinum deposited on a rhenium-based cata-
lyst with the nominal metal loading of each metal being 4 wt%.
The reaction free energy was calculated as follows:
D
G ¼
D
E ꢀ T
DS
where
D
E denotes the reaction energy from DFT, and
DS is the en-
tropy change from the gas or liquid phase to the adsorption on the
surface. For a gas-phase molecule, its entropy is directly calculated
in terms of the statistical thermodynamics formulae of a classical
ideal gas [23], and for a liquid molecule, its entropy Sl is calculated
by correcting its gaseous entropy Sg in the standard pressure Po with
the gas vapour Pꢃ in the ideal solution via:
Sl ¼ Sg þ RT lnðPꢃ=PoÞ
2.4. Density Functional Theory (DFT) calculations
3. Results
All periodic DFT calculations with Perdew–Burke–Ernzerhof
(PBE) generalised gradient approximation (GGA) functional [16]
were performed using Vienna ab initio simulation program (VASP)
[17,18]. The projector-augmented-wave (PAW) pseudopotentials
[19] were utilised to describe the core electron interactions with
a 400 eV cut-off energy of plane-wave basis set.
3.1. Catalyst testing
A range of bimetallic Pt/Re catalysts supported on CeZrO4, TiO2
and Al2O3 were prepared and tested for the selective hydrogena-
tion of N-methylpyrrolidin-2-one to N-methylpyrrolidine. Fig. 1
shows the conversion with respect to time at 120 °C and 20 bar

92
R. Burch et al. / Journal of Catalysis 283 (2011) 89–97
100
80
60
40
20
0
100
80
60
40
20
0
0
5
10
15
20
25
0
5
10
15
20
25
Time / h
Time / h
Fig. 1. Variation of amide conversion as a function of time for the hydrogenation of
N-methylpyrrolidin-2-one at 120 °C, 20 bar H2, stirred at 1500 rpm in hexane using
(h) 4 wt%Pt–4 wt%Re/TiO2 C.I., (j) 4 wt%Pt–4 wt%Re/TiO2 S.I., (s) 4 wt%Pt–
Fig. 2. Variation of amide conversion as a function of time for the hydrogenation of
N-methylpyrrolidin-2-one at 120 °C, 20 bar H2, stirred at 1500 rpm in hexane using
(j) 4 wt%Pt–4 wt%Re/TiO2 S.I., (d) 4 wt%Re–4 wt%Pt/TiO2 S.I., (N) 4 wt%Pt/TiO2 and
(ꢀ) 4 wt%Re/TiO2.
4 wt%Re/Al2O3 C.I., (d) 4 wt%Pt–4 wt%Re/Al2O3 S.I., (D) 4 wt%Pt–4 wt%Re/CeZrO4
C.I. and (N) 4 wt%Pt–4 wt%Re/CeZrO4 S.I.
100
80
60
40
20
0
H2 pressure using hexane as the solvent. For the supported 4 wt%
Pt–4 wt% Re catalysts some activity was observed in all cases,
but the TiO2-supported catalyst showed almost complete conver-
sion after 24 h. Significantly lower activity was observed for both
CeZrO4 and Al2O3 supports. Neither the surface areas or pore vol-
umes of the catalysts examined (Table 1) can be correlated with
their activity showing that these are not fundamental factors in
achieving high conversions and selectivities. A 100% selectivity to
N-methylpyrrolidine was found for all platinum–rhenium sup-
ported catalysts. Little activity for the bare supports was observed
with the exception of the CeZrO4 support, where 10% conversion
was found after 20 h and, in this case, 49% selectivity to the amine
was observed with the balance found to be 1-butanol. Some varia-
tion was found with the method of preparation between co-
impregnation and sequential impregnation, with the latter show-
ing higher activity, in general; however, this variation is minor
compared with the effect of the support. In contrast to the bimetal-
lic catalysts, the monometallic TiO2-supported catalysts showed
low activity, Fig. 2. Both 4 wt%Pt/TiO2 and 4 wt%Re/TiO2 catalysts
showed little conversion after 7 h. In the case of the platinum-
based catalyst, after 24 h, ꢁ15% conversion was achieved, again
with 100% selectivity to the amine.
0
5
10
15
20
25
Time / h
Fig. 3. Variation of amide conversion as a function of time for the hydrogenation of
N-methylpyrrolidin-2-one at 120 °C, 20 bar H2, stirred at 1500 rpm in hexane using
(j) 4 wt%Pt–4 wt%Re/TiO2, (N) 4 wt%Pt–1 wt%Re/TiO2, (.) 4 wt%Pt–0.5 wt%Re/
TiO2, (d) 4 wt%Pt–0.25 wt%Re/TiO2, (ꢀ) 1 wt%Pt–0.25 wt%Re/TiO2 and (e)
4 ꢂ 1 wt%Pt–0.25 wt%Re/TiO2.
The effect of the impregnation sequence of Pt and Re on the
conversion and selectivity was also studied. Fig. 2 shows the ki-
netic profiles obtained when using catalysts, where Re was sequen-
tially impregnated onto a Pt/TiO2 catalyst (denoted Re–Pt/TiO2)
and where Pt was sequentially impregnated onto a Re/TiO2 catalyst
(denoted Pt–Re/TiO2); for each metal, the loading used was 4 wt%.
Both preparation sequences produced highly active catalysts.
Although platinum deposition on a rhenium-based catalyst re-
sulted in a higher final conversion compared with the reverse se-
quence of impregnation, the difference is due to the increased
activity in the first 3 h of reaction, with both catalysts showing
similar activity after this stage. In each case, >80% conversion
was found after 24 h, and no change in the selectivity was
observed.
with the initial rates increasing from 0:096 mmol hꢀ1
g
to
ꢀ1
cat
3:2 mmol hꢀ1
g
; however, no further rate enhancement was
found on increasing the Re loading to 4 wt% (ICP 3.5wt%), where
ꢀ1
cat
the initial rate was 3:1 mmol hꢀ1
g
. To determine the effect of
ꢀ1
cat
dispersion, a catalyst was prepared with lower metal loadings,
but maintaining the ratio Pt:Re at 4:1. 1 wt%Pt–0.25 wt%Re/TiO2
was not found to be very active even on increasing the overall cat-
alyst mass used from 0.2 g to 0.8 g to ensure the same amount of
metal present in the reaction mixture compared with the reaction
catalysed by 4 wt%Pt–1 wt%Re/TiO2, Fig. 3. This implies that the
lower activity of the low-loading catalyst may be due to insuffi-
cient contact between the Re and Pt due to the higher dispersion
compared with the higher loaded catalysts.
The catalytic testing has shown that although the critical inter-
action required for the amide hydrogenation is between the plati-
num and the rhenium, the presence of titania can promote the
activity of the catalyst. Catalysts based on metals supported on
Clearly, the presence of Re and Pt together is critical in provid-
ing an active catalyst. Fig. 3 shows the effect of varying the Re load-
ing, whilst maintaining the loading of Pt at 4 wt%. A significant
increase in the activity of the catalyst was observed with Re load-
ing from a nominal 0.25 wt% (ICP 0.55 wt%) to 1 wt% (ICP 1.8 wt%),

R. Burch et al. / Journal of Catalysis 283 (2011) 89–97
93
reducible oxides, particularly TiO2, have been studied extensively
due to the ability of the metal to induce facile reduction of the
oxide support and form oxygen vacancies. For example, in the cat-
alysts studied herein, the Ti4+ is reduced to Ti3+ under hydrogen
even at low temperature. This activates the support and allows
both the metallic and oxidic phases to participate in C@O bond
hydrogenation reactions, whereby the C@O bonds are thought to
interact with the exposed Ti3+ sites on the support or oxygen
vacancies [24–26]. A similar mechanism is thought to be in opera-
tion for the amide hydrogenation.
(e)
(d)
3.2. Reaction kinetics
(c)
(b)
To ensure that the kinetics obtained were in the intrinsic kinetic
regime, a study into mass transfer limitations using the 4 wt%Pt–
4 wt%Re/TiO2 catalyst prepared by sequential impregnation was
carried out in hexane for the hydrogenation of N-methylpyrroli-
din-2-one. The speed of agitation was varied between 800 and
2000 rpm with no change in the reaction rate observed above
1500 rpm, Fig. S1, indicating the absence of gas to liquid mass
transfer at the rotation speed of 1500 rpm used in these studies.
The kinetics showed a linear dependence of the initial rate of reac-
tion with the catalyst mass, Fig. S2, indicating that liquid to solid
mass transfer limitations are negligible at the catalyst mass of
0.2 g typically used. Fig. S3 shows the Arrhenius plot between
120 °C and 160 °C, below which the amide solubility was limited
in the hexane solvent, and an apparent activation energy of
56 kJ molꢀ1 was determined indicating that the rates obtained
are predominantly kinetically controlled. It should be noted that
no change in the selectivity was observed over these experimental
conditions, with the only product observed being N-methylpyrrol-
idine. At the amide concentrations studied herein (0.017–0.183 M),
the reaction was found to be zero order in amide, Fig. S4, and first
order in hydrogen, Fig. S5. Above 0.183 M, the rate of reaction
dropped dramatically due to the maximum solubility of the amide
in hexane being exceeded. Under these experimental conditions,
the catalyst aggregates in the polar amide phase and was no longer
found to be evenly distributed throughout the liquid phase even on
strong mixing.
(a)
0
100 200 300 400 500 600 700 800
Temperature /οC
Fig. 4. Temperature-programmed reduction profiles for sequentially impregnated
catalysts, (a) 4 wt%Pt–4 wt%Re/CeZrO4, (b) 4 wt%Pt–4 wt%Re/Al2O3, (c) 4 wt%Pt–
4 wt%Re/TiO2, (d) 4 wt%Pt/TiO2 and (e) 4 wt%Re/TiO2.
shown in Fig. S6, which is likely due to the metal particle size being
extremely small.
3.4. Solvent effects
Fig. 5 shows the influence of solvent on the reaction time profile
for the hydrogenation of N-methylpyrrolidin-2-one. There is a
large solvent effect with the following rate order: hexane > tetra-
hydrofuran ꢁ diethyl ether > methanol > methyl tert-butyl ether.
In all cases, the amine is the only product observed. There are a
3.3. Catalyst characterisation
Fig. 4 shows the TPR characterisation of the 4 wt%Pt–4 wt%Re
catalysts supported on titania, alumina and ceria-zirconia and,
for comparison, the TPR of 4 wt%Pt/TiO2 and 4 wt%Re/TiO2. The
TPR analysis of the 4 wt%Pt–4 wt%Re/TiO2 catalyst shows three
peaks at 3 °C, 40 °C, and 400–600 °C. The first two features are
attributed to reduction of Pt only, in agreement with the TPR of
the 4 wt%Pt/TiO2 catalyst, and oxygen at the interface between Pt
and Re, which would indicate a significant interaction between
the Pt and Re. Similar high temperature features, albeit showing
a maximum reduction at a lower temperature of 315 °C and a sec-
ond peak at 380 °C, is found with 4 wt%Re/TiO2. These high tem-
perature peaks could be due to the reduction of ReOx [27] and/or
reduction of the bulk titania support. The ceria-zirconia-supported
bimetallic catalyst shows one peak at ꢁ65 °C, which could be
attributed to reduction of Pt only [28] or, as found for the TiO2-
based catalyst, the interfacial oxygen between Pt and Re. A similar
assignment of the low-temperature peaks between 50 and 125 °C
for the alumina supported catalyst is likely [29]. The TPR is in good
agreement with recently reported EXAFS results on these systems
[30], which shows that under both gas phase and liquid phase
reduction of the catalysts, the predominant state of the both Pt
and Re are zero valent, irrespective of the support used. There is lit-
tle evidence of Re or Pt in the XRD patterns of any of the catalysts
100
80
60
40
20
0
0
5
10
15
20
25
Time / h
Fig. 5. Variation of amide conversion as a function of time for the hydrogenation of
N-methylpyrrolidin-2-one at 120 °C, 20 bar H2, stirred at 1500 rpm using 4 wt%Pt–
4 wt%Re/TiO2 in (j) hexane, (N) tetrahydrofuran, (d) diethyl ether, (ꢀ) methanol
and (.) methyl tert-butyl ether.

94
R. Burch et al. / Journal of Catalysis 283 (2011) 89–97
number of possible explanations for these solvent effects; for
example the rate could be dependent on the concentration of
hydrogen dissolved in each solvent or on the solubility of the
amide in each solvent.
run and to only 6% on the fifth run, when simply washed with sol-
vent between each runs. Although drying and recalcination went
someway to restore the initial activity with conversions increasing
again to 53% and 30% after each treatment, respectively, the initial
high activity was not completely regained. Any carbon present on
the surface of the catalyst will have been removed by recalcina-
tion; therefore, the loss of activity may be attributed to decreased
interaction between the platinum and rhenium.
Table 2 shows the conversion after 24 h reaction time together
with the hydrogen solubility [31] in each solvent and the ENT [32]
solubility parameter, which provides a measure of the polarity of
the solvent. From the values available, the rate of reaction is found
to increase with the solubility of hydrogen in the solvent. Con-
versely, with the exception of the ether solvents, as the polarity
of the solvent increases, the rate of reaction generally decreases.
Although hydrogen solubility data are not available for the ether
solvents, the Hildebrand solubility parameter, the square of which
has been shown to correlate with the log of the hydrogen solubil-
ity, for diethyl ether is almost the same as hexane, 15.1 and 14.9,
respectively [33], indicating that the change in rate is unlikely to
be simply explained by either the hydrogen or amide solubility.
However, it is interesting to note that the amide is less soluble in
the non-polar hexane solvent and can adsorb strongly to the cata-
lyst surface, whereas in polar solvents, partitioning between the
surface and the solvent will mean that the concentration of the
amide on the catalyst surface will be less resulting in a lower rate
of reaction.
The strong adsorption of the amide on the surface in a non-po-
lar solvent was demonstrated by contacting a pre-reduced catalyst
with a hexane solution of amide in a nitrogen atmosphere for 3 h
before reaction. Fig. S7 shows a decrease in the conversion from
92% after 24 h when H2 was introduced at the start of the reaction
to 30% when the catalyst was pre-treated with the substrate. This
effect is attributed to the high surface coverage set up under the
nitrogen atmosphere, thus inhibiting the access of the hydrogen
to the surface. Conversely, when a similar experiment was carried
out with the more polar THF, in which the amide is more soluble,
only a small drop in conversion was observed, from 35% to 30%.
The formation of a mole of water for each mole of product
formed could also contribute to the solvent effect observed, partic-
ularly in hydrophobic solvents such as hexane. Magnesium sul-
phate (58 mg) was, therefore, added to a reaction carried out in
hexane using 4 wt%Pt–4 wt%Re catalysts supported on alumina,
ceria-zirconia and titania, and small changes in the reaction rate
were observed. In all cases, a slight increase in the reaction rate
was found, albeit small; however, a detailed study needs to be
undertaken to fully assess the effect of water on the rate. No
change in the selectivity was observed following the addition of
the drying agent.
3.6. Density Functional Theory (DFT) calculations
To obtain a general understanding of the trends of hydrogena-
tion of N-methylpyrrolidin-2-one (MPDO), Scheme 1, DFT calcula-
tions were used to investigate the reaction mechanism on four
transition metal surfaces (Re, Ru, Rh and Pt) with monatomic steps.
Amide hydrogenations can be initiated via one of three possible
activation processes of the carbonyl bond, namely direct carbonyl
bond cleavage, the cleavage with the assistance of the first hydro-
genation of the carbonyl carbon or the first hydrogenation of the
carbonyl oxygen. The subsequent step-wise hydrogenations of
the intermediates after the carbonyl bond cleavage give rise to
the two products, water and N-methylpyrrolidine (MPD). The
free-energy profile for each metal under the reaction conditions
(T = 120 °C, pH2 = 20 bar, pMPDO = 4.14 ꢂ 10ꢀ3 bar being the vapour
pressure of MPDO from an ideal solution) shows two striking fea-
tures, as illustrated in Fig. 6. Firstly, the free-energy barrier of car-
3.5. Catalyst recycle
Although the reactions proceeded to completion under mild
conditions over the fresh catalyst, with conversions of 92% being
obtained in 24 h, the conversion decreased to 67% on the second
Table 2
Comparison of the
% conversion of N-methylpyrrolidin-2-one after 24 h using
4 wt%Pt–4 wt%Re/TiO2 at 20 bar and 120 °C as a function of the solvent used and
the hydrogen solubility and polarity parameter ETN for each solvent.
Fig. 6. Free-energy profiles of all intermediates and transition states in the
hydrogenation of N-methylpyrrolidin-2-one (MPDO) to N-methylpyrrolidine
(MPD) over different surfaces initiated by carbonyl dissocaition under realistic
reaction conditions (T = 120 °C, pH2 = 20 bar, pMPDO = 4.14 ꢂ 10ꢀ3 bar, 94% conver-
sion). (a) Reaction over pure stepped Re, Ru, Rh and Pt sufaces, compared with a Re/
Pt bimetallic system. (b) Reactions over bimetallic systems with an oxidised Re
terrace on the substrate Pt, Pd, Rh and Ru. The preferential carbonyl activation
mechanism on stepped Rh and Pt surfaces is via the first hydrogenation on carbonyl
O, which is different from the others. However, as the difference in the C@O
activation mechanism does not change the reactivity trend of the metal group, the
data is plotted in terms of the C@O direct cleavage mechanism.
Solvent
%Conversion 103 ꢂ H2 at 20 bar and Polarity
parameter ENT
[32]
25 °C [31]
Hexane
Tetrahydrofuran 35
Diethyl ether
Methanol
Methyl tert-
butyl ether
94
13.4
5.4
–
0.009
0.207
0.117
0.762
0.124
30
15
0
3.2
–

R. Burch et al. / Journal of Catalysis 283 (2011) 89–97
95
bonyl activation increases, whilst the overall barriers of the subse-
quent hydrogenations decrease in the order of the metal reactivi-
ties: Re, Ru, Rh and Pt. Secondly, the rate-determining step
changes from carbonyl activation on Ru, Rh and Pt to the reduction
of oxygen to water on Re. On the stepped Re surface, the overall
activation barrier was found to be 2.74 eV, i.e. substantially higher
than that found for the C@O bond breaking over the same surface
(0.91 eV), implying that the hydrogenation of oxygen is the rate-
determining step, Table S1. On all other metals examined, the acti-
vation of the C@O bond was found to be the most energy-demand-
ing step. The free-energy barrier of direct C@O dissociation on Ru is
1.60 eV, which is sufficiently high to make activation under the
reaction conditions very difficult, and this value was found to in-
crease to 2.22 eV and 2.65 eV over the more inert Rh and Pt sur-
faces, respectively. It should be pointed out that the preferential
carbonyl activation pathway on stepped Rh and Pt surfaces is via
the first hydrogenation of the carbonyl oxygen followed by the
CAOH cleavage instead of the direct C@O cleavage, which is pre-
ferred on stepped Re and Ru surfaces. This is in agreement with
the DFT results from other groups and experimental studies, which
show that on inert metal surfaces such as Pt and Pt-based bimetal-
lic alloys, hydrogenation occurs readily to convert C@O to CAOH
[34]. However, even with the assistance of hydrogen, the carbonyl
activation barriers are still as high as 2.04 and 2.07 eV over stepped
Rh and Pt surfaces, respectively, indicating that the rate-determin-
ing step on Rh and Pt is still the carbonyl activation.
To further understand the system, the energy barriers over a flat
Re(0001) surface were compared with those over a stepped Re
surface. Table S1 shows that on the flat Re(0001) surface, the
C@O dissociation energy barrier is 2.11 eV, i.e. significantly higher
than that on the step site. These results indicate that only low-
coordination Re sites are capable of activating the C@O bond in
MPDO. In addition, the overall barriers of subsequent hydrogena-
tions on the flat Re surface were found to be as high as those on
the stepped surface, implying that Re is a poor catalyst for the
hydrogenation of the intermediates to products.
It is clear from these results why it is very difficult for pure met-
als to efficiently catalyse the hydrogenation of MPDO due to the
balance between having sufficient oxophilicity to activate the
C@O bond, whilst also enabling facile oxygen hydrogenation. In
contrast, bimetallic catalysts can provide the necessary bifunc-
tional sites to enable both the cleavage and the oxygen reduction,
as shown for the Pt–Re catalytic system reported in this paper. This
catalyst was modelled by a layer of Re with a size of (3 ꢂ 3) lying
on a flat Pt(111) with p(5 ꢂ 3) unit cell as shown in Fig. 7, where
Re was covered with 1 monolayer of oxygen, excluding the step
edge sites at the Re/Pt interface. According to the calculated overall
barrier (2.88 eV, see Table S1), the adsorbed oxygen is very difficult
IS:E=-1.25 eV
TS1: Ea=0.59 eV
MS1: E=-1.13 eV
TS2: Ea=0.87 eV MS2: E=0.02 eV
TS3: Ea=1.25 eV MS3: E=0.82 eV
TS4: Ea=0.66 eV MS4: E=-0.22 eV TS5: Ea=0.80 eV MS5: E=0.26 eV
Fig. 7. Schematic representation of the side and top view of the initial states (IS), transition states (TS) and Intermediates (MS) in the hydrogenation of N-methylpyrrolidin-2-
one (MPDO) on a Re/Pt surface and their corresponding energy barrier or reaction energy in each elementary step. The top, middle and bottom lines illustrate C@O activation,
oxygen and N-methylpyrrolidin-2-ylidene (MPDYE) step-wise hydrogenation processes, respectively. The white, grey, blue and red balls and stick style indicate H, C, N and O
atoms, respectively. The green and blue balls in CPK style are Re and Pt, respectively. (For interpretation of the references to colour in this figure legend, the reader is referred
to the web version of this article.)

96
R. Burch et al. / Journal of Catalysis 283 (2011) 89–97
to hydrogenate to water on a flat Re surface under mild reaction
conditions used here. In contrast, the adsorbed oxygen on Re step
edge sites at the Re/Pt interface can be readily removed either via
diffusion to the neighbouring terrace Re sites or by being hydroge-
nated to water. This is in agreement with the TPR results, which
show a significant decrease in the temperature of ReOx reduction
in the presence of Pt, which is almost certainly due to strong elec-
tronic perturbation of the Re by the Pt. Similar effects have been re-
ported previously and have recently been reviewed [35].
Fig. 7 shows that MPDO initially adsorbs atop on the step edge
site of oxygen-covered Re with a binding energy of ꢀ1.25 eV (the
corresponding free adsorption energy is 0.11 eV). The C@O bond
was found to be easily split to yield adsorbed O and 1-methylpyrr-
olidin-2-ylidene (MPDYE) via TS1 with a low barrier of 0.59 eV (the
corresponding free energy barrier of C@O activation with respect
to liquid MPDO is 0.70 eV, Table S1), which is 0.21 eV lower than
that on the Re stepped surface (0.91 eV on Re stepped surface).
The adsorbed hydrogen for subsequent hydrogenation steps
was assumed to come mainly from the Pt lower terrace surface
for the following reasons. Firstly, the C@O dissociation barrier on
Re step edge sites on Re/Pt was found to be moderate, and the dis-
sociation was strongly exothermic, leading to high coverage of oxy-
gen and MPDYE, and very low coverage of adsorbed hydrogen.
Secondly, MPDO does not adsorb onto a Pt terrace according to
the calculated results on Pt(111), meaning that the Pt surface
should be almost covered with adsorbed hydrogen for the subse-
quent hydrogenation steps. There are two possible pathways for
the hydrogenation processes over the two-phase catalyst system:
is in the order Pt > Pd > Rh ꢁ Ru. This is in good agreement with
previously reported results on Ru/Re and Rh/Re bimetallic catalyst
systems, which showed that, although the reduction of N-acetylpi-
peridine could be achieved, the reaction required 100 bar H2 pres-
sure and >150 °C [11].
4. Conclusions
Pt–Re bimetallic catalysts have been shown to be able to reduce
N-methylpyrrolidin-2-one to N-methylpyrrolidine at low tempera-
tures and moderate pressures (120 °C and 20 bar H2 pressure) for
the first time. The interaction between the Pt and Re was found
to be critical in forming active amide hydrogenation catalysts.
Although a range of supports was found to be effective for the
hydrogenation, the use of titania was found to promote the activity
of the bimetallic Pt/Re catalyst resulting in over 90% conversion
and >99% selectivity to the corresponding amine after 24 h. Density
functional theory calculations clearly demonstrated that the role of
the Re was to activate the C@O bond, whereas the Pt aided the
reduction of the ReAO and enabled the formation of water and
the amine. Monometallic systems were shown not to provide the
appropriate bifunctional capability.
Acknowledgments
The authors thank the EPSRC and Johnson Matthey plc for fund-
ing this work as part of the CARMAC (GR/S43702/01) and CASTech
(EP/G012156/1) projects. LMcL acknowledges studentship funding
from DELNI.
(i) the hydrogen from Pt directly attacks the intermediates on
the Re step edge sites via the Re–Pt interface; or
(ii) the adsorbed intermediates diffuse from the Re sites to Pt
sites and are then reduced.
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.jcat.2011.07.007.
Mechanism (i) is favoured, as the overall barriers of hydrogena-
tion of oxygen and MPDYE decrease to 1.27 and 0.66 eV, respec-
tively, Table S1, compared with 2.74 eV and 2.00 eV, respectively,
for pure Re step sites. Furthermore, higher hydrogenation cover-
ages further promote the hydrogenation reaction rate. In mecha-
nism (ii), the chemisorption energies of intermediates on the Pt
terrace are much less favourable than the high chemisorption
energies of intermediates on the step edge sites of Re on Re/Pt,
Table S2. For the intermediates to migrate from Re to Pt, the diffu-
sion barrier would have to be larger than the chemisorption energy
difference between the two sites. As the diffusion barrier of the
intermediates from Re to Pt (mechanism (ii)) is larger than the
overall hydrogenation barrier in mechanism (i), mechanism (i)
would be favoured in this system.
It is clear that the introduction of Pt as a second component
forming a two-phase catalyst greatly decreases the hydrogenation
barriers compared with those on pure Re surfaces. However, the
C@O activation energy barrier is also slightly lowered on Re/Pt rel-
ative to that found on the Re stepped surface, further contributing
to the ease with which this reaction can take place on this two-
phase surface as compared with the pure metal catalysts. Similar
analysis using Pd, Rh and Ru in combination with Re showed that
the surfaces were less active than found with Pt. For example,
although the combination of Re/Pd results in a C@O activation of
free-energy barrier similar to that found for Re/Pt, Table S1, the
overall barriers for the subsequent hydrogenation of oxygen and
MPDYE were 0.33 and 0.26 eV higher than those found on Re/Pt.
In the case of Re/Rh and Re/Ru, the free-energy barriers for C@O
dissociation were found to be about 0.3–0.5 eV higher than those
on Re/Pt and Re/Pd, whilst the overall barriers of subsequent
hydrogenation reactions were close to those on Re/Pd. It can be
concluded that the effectiveness of the second component metal
References
[1] G. Smith, F. Notheisz, Heterogeneous Catalysis in Organic Chemistry, first ed.,
Academic Press, 1999.
[2] J. March, Advanced Organic Chemistry, fourth ed., Wiley, New York, 1992. 1212.
[3] ] H.C. Brown, S. Narasimhan, Y.O.N.G.M. Choi, Synthesis (1981) 441–442.
[4] H.C. Brown, P. Heim, J. Org. Chem. 38 (1973) 912–916.
[5] Z.B. Papanastassiou, J.B. Robert, J. Org. Chem. 29 (1964) 2870–2872.
[6] (a) B. Wojcik, H. Adkins, J. Am. Chem. Soc. 56 (1934) 247–248;
(b) J.H. Paden, H. Adkins, J. Am. Chem. Soc. 58 (1926) 2487–2499;
(c) J.C. Sauer, H. Adkins, J. Am. Chem. Soc. 60 (1938) 402–406.
[7] H.S. Broadbent, W.J. Bartley, J. Org. Chem. 28 (1963) 2345–2347.
[8] A. Guyer, A. Bieler, G. Gerliczy, Helv. Chim. Acta 38 (1955) 1649–1654.
[9] F. Galinovsky, E. Stern, Chem. Ber. 76 (1943) 1034–1038.
[10] I.D. Dobson, Eur. Patent, 286 280, 1988 (British Petroleum).
[11] C. Hirosawa, N. Wakasa, T. Fuchikami, Tetrahedron Lett. 37 (1996) 6749–6752.
[12] A.A. Smith, P. Dani, P.D. Higginson, A.J. Pettman, International patent WO2005/
066112, 2005 (Avantium International).
[13] M.J. Nepras, R.J. Bernhardt, C.J. Sporer, United States Patent 5 840,985, 1998.
[14] (a) G. Beamson, A.J. Papworth, C. Philipps, A.M. Smith, R. Whyman, J. Catal.
269 (2010) 93–102;
(b) G. Beamson, A.J. Papworth, C. Philipps, A.M. Smith, R. Whyman, J. Catal.
278 (2011) 228–238.
[15] E. Balaraman, B. Gnanaprakasam, L.J.W. Shimon, D. Milstein, J. Am. Chem. Soc.
132 (2010) 16756–16758.
[16] J.P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 77 (1996) 3865–3868.
[17] G. Kresse, J. Hafner, Phys. Rev. B 49 (1994) 14251–14269.
[18] G. Kresse, J. Furthmüller, Comput. Mater. Sci. 6 (1996) 15–50.
[19] G. Kresse, D. Joubert, Phys. Rev. B 59 (1999) 1758–1775.
[20] A. Alavi, P. Hu, T. Deutsch, P.L. Silvestrelli, J. Hutter, Phys. Rev. Lett. 80 (1998)
3650–3653.
[21] A. Michaelides, P. Hu, J. Am. Chem. Soc. 123 (2001) 4235–4242.
[22] Z.-P. Liu, P. Hu, J. Am. Chem. Soc. 125 (2003) 1958–1967.
[23] P.W. Atkins, Physical Chemistry, Oxford University Press, Oxford, UK, 1998.
[24] M.J. Mendes, O.A.A. Santos, E. Jordão, A. Silva, Appl. Catal. A 217 (2001) 253–262.
[25] W. Rachmady, M.A. Vannice, J. Catal. 192 (2000) 322–334.
[26] T. Huizinga, J. Van Grondelle, R. Prins, Appl. Catal. 10 (1984) 199–213.
[27] K.G. Azzam, I.V. Babich, K. Seshan, L. Lefferts, J. Catal. 251 (2007) 163–171.
[28] J. Fan, X. Wu, R. Ran, D. Weng, Appl. Surf. Sci. 245 (2005) 162–171.

R. Burch et al. / Journal of Catalysis 283 (2011) 89–97
97
[29] S.M. Augustine, M.S. Nache, C.M. Tsang, J.B. Butt, W.M.H. Sachtler, in: M.J.
Phillips, M. Ternan (Eds.), Proc. 9th Int. Congr. Catal., Chemical Institute of
Canada, Calgary, 1988.
[30] J. Sá, C. Kartusch, M. Makosch, C. Paun, J.A. van Bokhoven, E. Kleymenov, J.
Szlachetko, M. Nachtegaal, H. Manyar, C. Hardacre, Chem. Commun. 47 (2011)
6590–6592.
[32] C. Reichardt, Solvents and Solvent Effects in Organic Chemistry, third ed.,
Wiley-VCH, Weinheim, 2003. 472.
[33] A.F.M. Barton, Handbook of Solubility Parameters and Other Cohesion
Parameters, second ed., CRC Press Inc., Florida, 1991. 250, 267.
[34] (a) D. Loffreda, F. Delbecq, F. Vigné, P. Sautet, J. Am. Chem. Soc 128 (2006)
1316–1323;
[31] (a) E. Brunner, J. Chem. Eng. Data 30 (1985) 269–273;
(b) T. Katayama, T. Nitta, J. Chem. Eng. Data 21 (1976) 194–196.
(b) L.E. Murillo, A.M. Goda, J.G. Chen, J. Am. Chem. Soc 129 (2007) 7101–7105.
[35] J.G. Chen, C.A. Menning, M.B. Zellner, Surf. Sci. Reports 63 (2008) 201–254.