Selective hydrogenation of amides using bimetallic Ru/Re and Rh/Re catalysts

Article abstract of DOI:10.1016/j.jcat.2010.12.009

Heterogeneous Ru/Re and Rh/Re catalysts, formed in situ from Ru ₃(CO)₁₂/Re₂(CO)₁₀ and Rh ₆(CO)₁₆/Re₂(CO)₁₀ respectively, are effective for the liquid phase hydrogenation of cyclohexanecarboxamide (CyCONH₂) to CyCH₂NH₂ in up to 95% selectivity without the requirement for ammonia to inhibit secondary and tertiary amine formation. Good amide conversions are noted within the reaction condition regimes 50-100 bar H₂ and ≥150 (Rh) - 160 °C (Ru). Variations in Ru:Re and Rh:Re composition result in only minor changes in product selectivity with no evidence of catalyst deactivation at higher levels of Re. In situ HP-FTIR spectroscopy has shown that catalyst genesis occurs via decomposition of the metal carbonyl precursors. Ex situ characterization, using XRD, XPS and EDX-STEM, has provided evidence for the active components of these catalysts containing bimetallic Ru/Re and Rh/Re nanoclusters, the surfaces of which become significantly oxidized after use in amide reduction. Potential mechanistic pathways for amide hydrogenation are discussed, including initial dehydration to nitrile, a pathway potentially specifically accessible to primary amides, and evidence for often postulated imine intermediates.

Full text of DOI:10.1016/j.jcat.2010.12.009

Journal of Catalysis 278 (2011) 228–238
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Selective hydrogenation of amides using bimetallic Ru/Re and Rh/Re catalysts
b
c
a
a
a,
⇑
Graham Beamson , Adam J. Papworth , Charles Philipps , Andrew M. Smith , Robin Whyman
a
Department of Chemistry, Donnan and Robert Robinson Laboratories, University of Liverpool, Liverpool L69 7ZD, UK
National Centre for Electron Spectroscopy and Surface Analysis, STFC Daresbury Laboratory, Warrington, Cheshire WA4 4AD, UK
Materials Science and Engineering, Department of Engineering, University of Liverpool, Liverpool L69 3GH, UK
b
c
a r t i c l e i n f o
a b s t r a c t
Article history:
3 2 10 6
Heterogeneous Ru/Re and Rh/Re catalysts, formed in situ from Ru (CO)12/Re (CO) and Rh (CO)16/
Received 2 November 2010
Revised 14 December 2010
Accepted 15 December 2010
Available online 26 January 2011
Re
CyCONH
ary and tertiary amine formation. Good amide conversions are noted within the reaction condition
regimes 50–100 bar H and P150 (Rh) – 160 °C (Ru). Variations in Ru:Re and Rh:Re composition result
2
(CO)10 respectively, are effective for the liquid phase hydrogenation of cyclohexanecarboxamide
(
2
) to CyCH NH in up to 95% selectivity without the requirement for ammonia to inhibit second-
2
2
2
in only minor changes in product selectivity with no evidence of catalyst deactivation at higher levels
of Re. In situ HP-FTIR spectroscopy has shown that catalyst genesis occurs via decomposition of the metal
carbonyl precursors. Ex situ characterization, using XRD, XPS and EDX-STEM, has provided evidence for
the active components of these catalysts containing bimetallic Ru/Re and Rh/Re nanoclusters, the sur-
faces of which become significantly oxidized after use in amide reduction. Potential mechanistic path-
ways for amide hydrogenation are discussed, including initial dehydration to nitrile, a pathway
potentially specifically accessible to primary amides, and evidence for often postulated imine
intermediates.
Keywords:
Heterogeneous catalysis
Ruthenium
Rhenium
Rhodium
Bimetallic
Hydrogenation
Amide
Primary amine
Characterization
Reaction mechanism
Ó 2010 Elsevier Inc. All rights reserved.
1
. Introduction
The generation, characterization and use of bimetallic heteroge-
Yoshino et al. [5] with a report of the use of promoted bimetallic
Re/Os catalysts under considerably milder reaction conditions
(25–100 bar H , 100–120 °C, 6 h), with highest alcohol selectivities
2
neous Rh/Mo and Ru/Mo catalysts for the selective hydrogenation
of amides to amines in the liquid phase has recently been described
being favoured at the highest pressure. A subsequent very recent
development is the use of titania-supported Pt (and Pt/Re) cata-
[
1,2]. Key features of this work include both high catalyst selectivity
lysts with optimum reaction conditions at 20 bar H
although very low reaction rates at as low as 5 bar H
are also quoted [6]. The first account of the use of Re in amide
reduction was also provided by Broadbent et al. [7], who described
the use of Re(VI) oxide for the selective hydrogenation of benzam-
2
and 130 °C,
and 60 °C
for the reduction of primary amides to the respective primary
amines, without the necessity for the addition of ammonia and/or
amines to inhibit side reactions leading to secondary and tertiary
amine by-products, and considerably milder reaction conditions
relative to those required by standard first generation copper chro-
mite catalysts [3]. Catalyst performance was shown to be crucially
dependent on Mo:Rh and Mo:Ru composition, respective ratios of
2
2
ide to benzylamine (205 bar H , 220 °C, 49 h, ethanol solvent) in
69% yield, with toluene as the only by-product. Surprisingly, the
formation of neither cyclohexylamine nor N-ethyl-substituted sec-
ondary or tertiary amine derivatives were reported [cf. Ref. [2]].
Subsequently, a BP patent claimed the use of a Pd/Re/high surface
area graphite/zeolite 4A combination, dispersed in a solvent such
>
2 and >1 leading to either significant, or complete, inhibition of
catalysis.
Rhenium-based catalysts have received greater attention than
Rh, Ru or Mo for the reduction of ‘difficult’ functional groups such
as amides, and particularly carboxylic acids and esters. Broadbent
2
as 1,4-dioxane, for the reduction of amides at 130 bar H and
200 °C [8]. Using propionamide as substrate, the product distribu-
tion comprised a mixture of primary, secondary and tertiary
amines. Fuchikami et al. [9] briefly described the behaviour of
Rh/Re catalysts for the reduction of a range of amides. In the one
example of a primary amide tested, n-hexanamide required the
co-addition of an amine (diethylamine) to induce good selectivity
et al. [4] described the use of Re
the hydrogenation of carboxylic acids under the severe reaction
conditions (312 bar H , 217 °C, 6 h) typically required by copper
chromite catalysts. A step change was described in 1990 by
2 7
O as a catalyst precursor for
2
to n-hexylamine under reaction conditions of 100 bar H
2
and
⇑
Corresponding author. Fax: +44 (0) 151 794 3588.
E-mail address: whyman@liv.ac.uk (R. Whyman).
1
80 °C. Otherwise di(n-hexyl)amine comprised the predominant
0
021-9517/$ - see front matter Ó 2010 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2010.12.009

G. Beamson et al. / Journal of Catalysis 278 (2011) 228–238
229
product, in complete contrast to our recent findings with Rh/Mo
and Ru/Mo catalysts using cyclohexanecarboxamide (CyCONH
as substrate [1,2]. Here, the genesis and performance of both Ru/
Re and Rh/Re catalysts in the hydrogenation of CyCONH (and N-
acetylpiperidine) are reported, together with ex situ characteriza-
tion using microanalysis, XRD, XPS and EDX-STEM.
tively Coupled Plasma (ICP) source linked to Atomic Emission Spec-
troscopy (AES). Sample digestion was accomplished using a matrix
combination containing aqua regia (2 eq. HCl: 1 eq. HNO ) (5 mL)
3
2
)
2
and HF (0.5 mL), followed by microwave treatment in a CEM
MARS5 oven using the following programme conditions; heating
ꢀ1
to 220 °C at 20 °C min , holding at 220 °C for 20 min followed
by a cooling period of 20 min. The resultant solutions were diluted
to 100 mL in distilled water and referenced against Ru, Rh and Re
2
. Materials and methods
3
standards made up in similar HCl/HNO /HF matrix solutions.
Application of this procedure overcame the well-known resistance
of Ru in particular to digestion [10] (see Supplementary material
S2 for further details).
With the exception of the additional details described below,
experimental and analytical procedures are as outlined in Refs.
1,2].
[
Standard C, H and N determinations were carried out by com-
bustion analysis using either a Flash EA-1112 or a Model 1106
Carlo Erba thermal elemental analyzer at ꢁ2000 °C. For the bime-
2
.1. Reagents
tallic catalyst samples, a V
combustion.
2 5
O catalyst was required to aid
2
Re (CO)10 (98%) was purchased from Strem Chemicals and the
XPS standard, Re foil (0.025 mm thickness, 99.98%), from Aldrich
Chemicals.
3
. Results and discussion
2.2. Catalytic procedures
3.1. N-acetylpiperidine hydrogenation using Ru/Re catalysts
A typical ‘single-pot’ batch Ru/Re catalyst preparation, and eval-
uation in CyCONH
nominal Re:Ru atomic composition of 0.25. [Ru
.103 mmol Ru), [Re (CO)10] (8.4 mg, 0.026 mmol Re) and cycloh-
exanecarboxamide (0.235 g, 1.85 mmol) contained in a glass liner
were dissolved in 1,2-dimethoxyethane (DME) (30 mL) and n-oc-
tane (0.100 g) added as an internal standard for GC analysis. The
liner was placed in a ca. 300-mL capacity pressure vessel and the
2
reduction, was carried out as follows, for a
A hydrogenation catalyst derived from Ru
Re:Ru = 1.0, 1 mol% Ru) was reported by Fuchikami et al. [9] to
give high conversions of N-acetylpiperidine into N-ethylpiperidine
96% yield) under standard reaction conditions of 100 bar H and
60 °C. Ru alone was essentially inactive (1% conversion and amine
yield) and Re showed low activity (14% conversion) with only 7%
yield of the desired amine. Repetition of this work using Re (CO)10
has confirmed reproducibility (16% conversion, 53% amine selec-
tivity) and revealed that competing CAN bond hydrogenolysis of
N-acetylpiperidine (to ethanol and piperidine) accounts for the
additional products. No attempts to use Ru/Re catalysts for the
reduction of primary amides such as cyclohexanecarboxamide, Cy-
3 2
(CO)12 and Re (CO)10
3
(CO)12] (22 mg,
(
0
2
(
1
2
2
reaction mixture, under agitation, purged three times with N
2
(at
(5 bar). The autoclave was then
2
and heated to 160 °C for 16 h. A very
5
bar), and three times with H
2
pressurized to 100 bar H
dark coloured liquid was initially recovered after cooling and
venting and from this a black residue slowly settled, leaving a col-
ourless solution. The residue was separated by centrifugation
2
CONH , have been reported and this has provided the focus for our
(
2000 rpm, 20 min) and the supernatant liquid containing reaction
work.
products removed. After several washings with DME (10 mL), the
catalyst residue (ca 15 mg) was dried and the resultant fine black
powder either recycled using the same quantities of fresh substrate
and solvent, or characterized using the techniques described be-
low. Product solutions were analysed by GC as previously de-
scribed [1].
In addition to ‘batch’ experiments, ‘parallel’ catalyst evaluations
were also conducted. A typical ‘parallel’ catalyst evaluation was
carried out using the standard autoclave used for batch runs, mod-
ified to accommodate four vertically mounted modified test tubes
of ca. 20 mL capacity, all containing mini-magnetic followers to en-
sure effective stirring. Using a 10 mL volume of DME solvent, the
scale of substrate and catalyst precursors (4 mol% Ru) were re-
duced accordingly.
3
2
.2. CyCONH hydrogenation using Ru/Re catalysts
3
.2.1. Variation of Re:Ru composition
Using a series of parallel experiments (10 mL scale and 4 mol%
catalyst, based on Ru), the results of variation in Re:Ru ratio are
shown in Fig. 1. From this, it is evident that the addition of only
a small concentration of Re (CO)10 to Ru (CO)12, comparable to
2 3
that observed with Ru/Mo (and Rh/Mo) catalysts, is necessary to
initiate analogous overall conversion/product selectivity synergy.
In contrast however there is no evidence of significant reduction
in conversion with increasing Re content, for up to Re:Ru = 1.8.
Moreover, product selectivity remains essentially constant be-
tween Re:Ru values of 0.25 and 1.1, with uniformly trace amounts
of (CyCH ) NH present across the range Re:Ru = 0.3–1.8, and only
2
2
.3. Ex situ characterization
2
2
at Re:Ru = 1.8 are slight divergences of product distribution in fa-
vour of CyCH OH apparent. A Re:Ru composition of ca. 0.3 appears
.3.1. EDX-STEM
2
Sample quantification was determined using the most intense
optimum for a combination of conversion and selective formation
‘
pure’ Ru K 1
a
(19.279 keV) and Re L 1
a
(8.654 keV) lines that
of CyCH NH (both >90%). The conversion and selectivity recorded
2
2
are largely free from interference by signals from other elements.
Even so, both trace O and Si were also detected during the analysis
and deconvolution of the Si K and Re L lines became necessary for
accurate quantification. Also, since Ru K and Re L lines are under
comparison, the quoted values are subject to an error of ±5%.
for the Re:Ru = 0.34 catalyst in Fig. 1 are those obtained from the
final reaction solution from the in situ HP-FTIR experiment using
the complete CyCONH2 reduction system, described in Section
3.6.1, thus confirming internal consistency between the two sets
of experimental data.
Furthermore, for the purposes of checking reproducibility be-
tween ‘parallel’ and ‘batch’ processing, three Ru/Re catalysts were
examined in batch experiments, using 5 mol% Ru. The results sum-
marized in Table 1 provide confirmation, with essentially the same
reaction product profile and only a marginal improvement in
2
.3.2. Microanalyses
All elemental analyses were performed by Mr. S.G. Apter,
Department of Chemistry, University of Liverpool. Transition metal
concentrations were determined using a Spectro Ciros CCD Induc-

2
30
G. Beamson et al. / Journal of Catalysis 278 (2011) 228–238
Table 2
2
CyCONH hydrogenation. Dependence of conversion and product distribution on
catalyst concentration.
Catalyst concentration Conversion Product selectivity (%)
(
mol%)
(%)
CyCH
2
NH
2
CyCH
2 2 2
OH (CyCH ) NH
4
2
1
.0
.2
.2
90
50
32
95
93
96
3
6
3
2
1
<1
2
Reaction conditions: 100 bar H , 160 °C, 16 h, Re:Ru = 0.25 catalyst.
Table 3
2
CyCONH hydrogenation: conversion and product selectivity vs. pressure and
temperature.
Entry
T
P
H2
Conversion
(%)
Product selectivity (%)
(
°C)
(bar)
CyCH
2
NH
2
CyCH
2 2 2
OH (CyCH ) NH
1
2
3
4
5
160
160
160
150
140
100
50
20
100
100
95
98
60
20
0
91
81
77
80
–
8
16
13
17
–
1
3
9
2
–
Fig. 1. CyCONH
vs. Re:Ru composition (100 bar H
CyCH OH.
NH, 4 CyCH
2
hydrogenation: (a) overall conversion and (b) product selectivity
2
, 160 °C, 16 h). Key: j Conversion, } CyCH NH , +
2
2
(
)
2 2
2
2 2
selectivity towards CyCH NH . No secondary amine production is
evident with Re:Ru = 0.34 and 0.54 catalysts (Entries 2 and 3)
and only traces at Re:Ru = 0.15 (Entry 1), in which the
Reaction time:16 h, Re:Ru = 0.25 catalyst.
available to Ru/Mo catalysts [2]. Any potential explanation in
terms of decreased availability of H in DME solution at the lower
pressures using Ru/Re catalysts seems therefore to be unlikely (for
further information on H solubility in DME, see Supplementary
2 2 2
CyCH NH /CyCH OH product distribution is a close parallel with
that observed with Ru/Mo catalysts of equivalent composition [2].
2
2
3.2.2. Variation of total catalyst concentration
material S1, Table S1 and Figs. S2 and S3).
A preformed Ru/Re (Re:Ru = 0.25) catalyst previously prepared
in the absence of the amide substrate was used, in batch experi-
ments, to establish a conversion/selectivity profile for CyCONH
reduction as a function of total catalyst concentration. The results
Table 2 and Supplementary material Fig. S1) show that CyCH NH
selectivity (93–96%) is essentially unaffected and that amide con-
version displays an approximately first-order dependence on cata-
lyst concentration. Hence, an apparent discrepancy between the
incomplete amide conversions evident throughout the ‘parallel’
experiments shown in Fig. 1, and the 100% conversions noted in
Table 1, may be rationalized in terms of the slightly higher (5 vs.
3.3. N-acetylpiperidine hydrogenation using Rh/Re catalysts
2
(
2
2
In a preliminary single-pot reaction using N-acetylpiperidine as
substrate, a Rh/Re catalyst (Re:Rh = 1.0, 1 mol% Rh) derived from
Rh
dine in >85% selectivity, using the standard reaction conditions of
100 bar H , 160 °C and 16 h, in confirmation of the report of Fuchi-
6 2
(CO)16 and Re (CO)10 gave >80% conversion to N-ethylpiperi-
2
kami et al. [9]. As found with the Rh/Mo [1] and Ru/Mo [2] cata-
lysts, the solid material generated during this single-pot reaction
could be recovered most easily from the product solution by cen-
trifugation, repeated washing with DME solvent, and subsequent
drying. Following re-charging with N-acetylpiperidine after this
treatment catalyst recycle could be conveniently monitored using
HP-FTIR spectroscopy, by following the decay of the amide car-
4
mol%) Ru catalyst concentration used in the batch experiments.
3.2.3. Variation of pressure and temperature
The effects of variation in pressure and temperature on Cy-
CONH
in Table 3. From this data, it is clear that the standard reaction con-
ditions of 100 bar H and 160 °C are necessary for highest
CyCH NH selectivity, which undergoes a significant reduction at
0 bar H , notwithstanding essentially complete CyCONH conver-
sion (cf. Entries 1 and 2). Reduction in reaction pressure to 20 bar
(Entry 3) results in much lower amide conversion, a further
slight reduction in selectivity towards CyCH NH , and increased
CyCH NH production. Furthermore, reaction temperatures be-
2
hydrogenation during batch experiments are summarized
–
1
bonyl absorption band at 1658 cm as a function of time (see Sup-
plementary material, Fig. S4), from which it is clear that initiation
of hydrogenation occurs without an induction period on reaching
the operational temperature of 160 °C (at time = 0 min).
2
2
2
5
2
2
H
2
2
3.4. CyCONH hydrogenation using Rh/Re catalysts
2
2
(
2
)
2
In Fuchikami’s work [9], the one experiment reporting the use
of a primary amide (n-hexanamide), at a Re:Rh composition of 2,
required the addition of an amine (diethylamine) to induce good
selectivity to the primary amine. This contrasts with our more re-
cent results using Rh/Mo, Ru/Mo and Ru/Re catalysts prepared in a
similar manner, in which the addition of neither ammonia nor
amine is required for high primary amine selectivity. Our standard
low 160 °C are also clearly deleterious to catalytic performance,
with 20% conversion at 150 °C (Entry 4) and zero activity evident
at 140 °C (Entry 5). This behaviour contrasts markedly with the
2 2
limiting parameters of 20 bar H /160 °C and/or 100 bar H /130 °C
Table 1
CyCONH
2
hydrogenation. Dependence of conversion and product distribution on
2
substrate CyCONH was therefore used to determine whether Rh/
Re:Ru composition in batch experiments.
Re catalysts do indeed exhibit fundamentally different behaviour
towards the reduction of primary amides.
Entry
Re:Ru
Conversion (%)
Product selectivity (%)
CyCH
2
NH
2
CyCH
2
OH
2 2
(CyCH ) NH
1
2
3
0.15
0.34
0.54
100
100
100
86
92
93
12
7
6
Trace
0
0
3
.4.1. Variation of Re:Rh composition
Results of variation in Re:Rh composition on CyCONH hydroge-
2
nation are shown in Fig. 2, from which it is evident that in contrast
to Ru/Re catalysts (also Rh/Mo and Ru/Mo), the sequential addition
2
Reaction conditions: 100 bar H , 160 °C, 16 h, 5 mol% Ru.

G. Beamson et al. / Journal of Catalysis 278 (2011) 228–238
231
of Re to Rh only leads to a gradual increase in conversion over the
Re:Rh range 0–0.67, and quite possibly beyond. Consequently, a
more substantial concentration of Re (ca. Re:Rh = 0.7) is required
on CyCONH
high amide conversions are maintained at 50 bar H
nificant reduction of primary amine selectivity in favour of
(CyCH NH, relative to that noted at 100 bar (Entry 1) is evident,
CyCH OH remaining unchanged. A considerable reduction in con-
version is noted at 20 bar (Entry 3), with selectivity towards Cy-
CH OH significantly increased in relation to that at 50 bar,
notwithstanding a minor increase in (CyCH NH formation. Inspec-
tion of Entries 4–6 shows that any reduction in reaction temperature
below 150 °C is deleterious to catalyst performance, with no reac-
tion evident at 130 °C during the standard 16 h reaction time. Nev-
ertheless, slow reduction does occur during a considerably extended
reaction time (Entry 7). This could possibly be a consequence of low-
2
hydrogenation are summarized in Table 5. Although
2
(Entry 2), a sig-
for 100% CyCONH
2
conversion. Nevertheless, CyCH
2
NH
2
is still
2 2
)
the major product throughout, formed in 90% selectivity using a
Re:Rh = 0.80 catalyst composition. These results provide a distinct
contrast with that of Fuchikami et al. [9] and confirm our belief
that high primary amine selectivities comprise an intrinsic charac-
teristic of all the bimetallic systems investigated (Ru/Mo, Ru/Re,
2
2
2 2
)
2
Rh/Mo and Rh/Re), at least when using CyCONH as a primary
amide substrate. Nevertheless, in displaying a gradual increase in
conversion as a function of Re content, the Rh/Re catalysts clearly
differ in relation to the degree of synergism that was so clearly evi-
dent, and necessary, for the high amide conversions and primary
amine selectivities associated with the first three members of the
series.
er rates of Re
2
(CO)10 decomposition, and active catalyst formation,
at 130 °C. Overall, reaction conditions of 100 bar H
2
and 160 °C (cf.
Entry 1) do appear necessary to give a combination of the highest
amide conversions and primary amine selectivities.
3
.4.2. Variation of amide concentration
Using batch experiments, the dependence of amide concentra-
3.5. Similarities and differences in catalyst behaviour
tion on reaction selectivity, using a Re:Rh = 0.80 catalyst composi-
tion (5 mol% Rh), are shown in Table 4, from which it is clear that
high dilution is beneficial, with the highest selectivity towards
From the results described earlier, it appears that Ru/Re and Rh/
Re catalysts both provide excellent intrinsic selectivities for the
reduction of CyCONH to CyCH NH . Although they are actually
2 2
CyCH NH occurring at the lowest amide concentration (Entry 1),
2
2
2
superior in terms of primary amine selectivity, the minimum pres-
sure and temperature requirements (cf. Tables 3 and 5, Entry 1)
dictate the use of 100 bar H and 160–150 °C, respectively, for good
2
the standard concentration used in all the Re:Rh, pressure and
temperature variation experiments (see Section 3.4.3). The in-
creased amount of secondary amine formation in solutions
containing higher amide concentrations is presumably a conse-
quence of the greater probability of condensation reactions occur-
amide conversions, and in this respect, they are of inferior perfor-
mance to their Ru/Mo and Rh/Mo counterparts. With Ru/Re cata-
lysts, an increase in the concentration of Re above the threshold
value of ca. Re:Ru = 0.15 has little effect on catalyst selectivity,
although there is a general trend of decreasing conversion and in-
creased secondary amine production at higher Re:Ru catalyst load-
ings. With the Rh/Re systems, low initial concentrations of Re
enhance the high primary amine selectivity above that observed
with the base Rh case (70%, Fig. 2), but additional Re content is nec-
2
ring on the catalyst surface. In contrast, CyCH OH formation
remains largely constant throughout.
3
.4.3. Variation of pressure and temperature
Using a Re:Rh = 0.80 catalyst and standard batch experiments,
the results of the effects of variation in pressure and temperature
2
essary to allow 100% CyCONH conversion under the standard
reaction conditions, cf. Fig. 2.
3
3
1
.6. Catalyst genesis using in situ HP-FTIR spectroscopy
.6.1. Ru/Re catalysts
Inspection of absorbances of the amide carbonyl band at
–
1
693 cm in spectra of the complete CyCONH
(CO)12 and Re (CO)10 (Re:Ru = 0.34) as catalyst
precursors, and a standard heating time of ca. 15 min to 160 °C
see Supplementary material Fig. S5, first entry at 160 °C shown
2
amide reduction
system, using Ru
3
2
(
as time = 0 min), reveals that amide reduction commences imme-
diately on reaching the operational reaction temperature. Calcu-
lated as previously described [2], the amide decay curve
corresponds to an initial TON (mmol. CyCONH
2
consumed vs.
mmol. Ru) of 1.9 h . Thus, in sharp contrast to Rh/Re catalysts
see Section 3.6.2) and to the Rh/Mo and Ru/Mo catalyst systems
described previously [1,2], no catalyst induction period appears
necessary for this combination. All evidence of Ru (CO)12 disap-
pears during the initial heating period, and the residual (CO)
(CO)10, slowly
–
1
(
Fig. 2. CyCONH
Re:Rh composition (100 bar H
CyCH OH.
NH, 4 CyCH
2
hydrogenation: (a) conversion and (b) product selectivity vs.
2
, 160 °C, 16 h). Key: j Conversion, } CyCH NH , +
2
2
3
(
)
2 2
2
m
absorptions, corresponding with the spectrum of Re
degrade during the subsequent 2 h.
2
Table 4
CyCONH
concentration.
2
hydrogenation. Variation of conversion and product selectivity with amide
A control experiment, also using a 0.34 Re:Ru precursor ratio
but with stepped, rather than continuous heating (see Fig. 3), pro-
vides some amplification of the initial stages of catalyst genesis. At
room temperature, both Ru and Re precursors are clearly identifi-
able by comparison with dominant features of reference spectra,
Entry Amide (mol/ Conversion
Product selectivity
L)
(%)
CyCH
2
NH
2
CyCH
2 2 2
OH (CyCH ) NH
1
2
3
4
0.072
0.112
0.147
0.181
97
89
80
82
74
3
6
6
6
7
11
11
18
ꢀ1
–1
100
100
93
cf. Ru
(CO)12 (2062 cm ), Re
(CO)10 (2069, 2014 and 1971 cm ).
3
2
ꢀ1
On heating to 80 °C, the 2062-cm band has been displaced by
absorptions at 2036 and 2000 cm , consequent upon the forma-
–1
ꢀ
Reaction conditions: 100 bar H
2
, 160 °C, 16 h, Re:Rh = 0.80 catalyst.
tion of [H
3
Ru
4
(CO)12] , as found in the Ru/Mo systems [2]; the

2
32
G. Beamson et al. / Journal of Catalysis 278 (2011) 228–238
Table 5
CyCONH
standard amide reduction conditions, are also revealing. Here,
full decomposition of the organometallic molecule to metallic
Re does not occur during the standard 16-h reaction period, in
2
hydrogenation: conversion and product selectivity vs. pressure and
temperature (Re:Rh = 0.80 catalyst).
Entry
T
P
H2
Conversion
(%)
Product selectivity (%)
which Re
(CO)10 is converted, possibly via H
Re
(CO)12
(
m
(CO)
(OH) (-
CO)12 (m(CO) 2027s and 1923br cm ) [12], which could be recov-
2
4
4
(°C)
(bar)
–1
CyCH
2
NH
2
CyCH
2
OH (CyCH
2 2
) NH
2042s, 1990 m cm ) [11] in the O-donor solvent, into Re
4
4
–
1
1
2
3
4
5
6
160
160
160
150
140
130
130
100
50
20
100
100
100
100
98
98
61
97
64
0
90
78
69
84
83
–
6
7
14
8
8
–
3
13
17
6
7
–
ered in the form of a clear pale yellow solution after cooling and
depressurization. Furthermore, from an inspection of the relative
band intensity ratios associated with the spectrum of Re
2
(CO)10,
there is no evidence for the intermediate formation of significant
–
1
a
amounts of HRe(CO)
the reaction of H with Re
In summary therefore, the presence of both Ru
(
m
(CO) 2015s and 2006 m cm ) [13] during
(CO)10 at 100 bar H and 160 °C.
(CO)12 and sub-
appear to initiate
(CO)10 during the initial
7
60
81
9
6
5
2
2
2
Reaction time: 16 h, Re:Rh = 0.80 catalyst.
a
3
Reaction time 68 h.
–
3 4
sequent reaction products, e.g., [H Ru (CO)12]
and enhance partial decomposition of Re
2
additional absorption expected at 2018 cm–1 being masked by the
intense band due to Re (CO)10. All Ru-containing species decom-
pose before 160 °C is reached, at which point ca. Thirty percentage
of the Re (CO)10 initially charged remains in solution. A combina-
tion of 70% of the Re (CO)10 initially present with complete decom-
position of Ru (CO)12 would in fact correspond to a catalyst
comprising the optimum Re:Ru value of 0.25 (cf. Fig. 1). Conse-
quently, decomposition of the remaining 30% of Re (CO)10 during
the subsequent 2-h period shown in Fig. 3, throughout which slow
reduction of CyCONH is taking place, may be largely superfluous
to overall catalyst performance. It may indeed be deleterious, since
inspection of Fig. 1 reveals a slight decrease in conversion with
increasing Re content, which may be a reflection of the onset of
the considerably more extensive catalyst poisoning previously
encountered with the Ru/Mo and Rh/Mo catalyst systems. The
stages of catalyst genesis. Onset of catalytic activity appears to
be dependent only on complete decomposition of the Ru carbonyl
intermediates, which occurs very rapidly, and is largely indepen-
2
2
dent of the overall rate of decomposition of Re (CO)10.
2
2
3
3.6.2. Rh/Re catalysts
In situ HP-FTIR spectra of the Rh/Re systems (Fig. 4) were less
conclusive than those of Ru/Re in terms of identifying a parallel ef-
fect, namely the role of Rh carbonyl decomposition in initiating
2
2
and enhancing decomposition of Re
simply because the decay of (CO) absorption bands associated
with Rh (CO)16 could not be readily identified due to its limited
2
(CO)10 during catalyst genesis,
m
6
–
1
solubility in DME. Nevertheless, a peak at 1755 cm may corre-
spond with the formation and presence of either polynuclear Rh
carbonyl anions such as [Rh
2
–
n–
greater stability of Re
to decomposition is entirely consistent with the position of Re
comprising a third row transition element) in the Periodic Table.
Thus, the situation is not directly comparable to that in which
Mo(CO) was used as co-catalyst precursor with Ru (CO)12, and
2
(CO)10, relative to Ru
3
(CO)12, with respect
6
(CO)15
]
x
[14], or [Rh13(CO)24H ]
(x = 2, n = 3; x = 3, n = 2, etc.), the carbonyl hydride-containing fam-
ily proposed in the Rh/Mo catalyst precursor systems [1]. Signifi-
cant differences from the Ru/Re systems are however readily
evident. Genesis of a Rh/Re (Re:Rh = 1) catalyst, chosen on the basis
of the information in Fig. 2 to be above the threshold of 0.8 for
complete amide hydrogenation under the standard reaction condi-
tions, requires an induction period of ca. 500 min, with initiation
(
6
3
where the former species was found to be the first component to
undergo rapid decomposition [2].
The results of a further control experiment, namely the reac-
tion of H
2
with a DME solution of Re
2
(CO)10 alone under the
apparently dependent on complete decomposition of Re
2
(CO)10, a
situation in close parallel to that observed with Rh/Mo catalysts
[
1
1] (cf. Fig. 4, in which
971 cm correspond to the presence of Re (CO)10). Internal con-
m
(CO) absorptions at 2069, 2014 and
–
1
2
1
60 °C
40 °C
sistency with catalyst performance as a function of Re:Rh (cf.
Fig. 2) is also evident.
0.2
1
2
120 °C
1.6
1.2
0.8
0.4
0
1
00 °C
2014
2000
2
036
1971
2069
80 °C
6
0 °C
2062
2030
22 °C
0
200
400
time (min)
600
800
1000
2
090
2070
2050
2030
2010
1990
1970
1950
-
1
(cm )
Fig. 4. In situ HP-FTIR spectra (2100–1600 cm–1) showing decay and disappearance
of (CO) absorptions during catalyst genesis followed by initiation of hydrogena-
Fig. 3. In situ HP-FTIR spectra (2100–1950 cm–1) during initial stages of catalyst
m
genesis. Reaction conditions: 100 bar H
solvent.
2
, 22–160 °C, Re:Ru = 0.34 catalyst, DME
tion. Reaction conditions: 100 bar H and 160 °C, Re:Rh = 1 catalyst. Key: j
2
(1693 cm–1), } 2014 cm , 4 2069 cm , s 1971 cm , + 1755 cm
–1
–1 –1 –1
.
CyCONH
2

G. Beamson et al. / Journal of Catalysis 278 (2011) 228–238
233
3.7. Ex situ catalyst characterization
500
4
4
3
50
00
50
3
3
.7.1. XRD
.7.1.1. Ru/Re catalyst. The XRD profile of a Ru/Re (Re:Ru = 0.23)
composition, representative of an active and highly selective cata-
lyst, is shown in Fig. 5. As might be expected, given the above stoi-
chiometry, the diffractogram appears dominated by the profile of
metallic Ru. Unfortunately, the most intense 2h = 43.98, 47.35
and 50.29° lines expected for metallic Re [15] are largely obscured
by the Ru diffraction pattern. However, by analogy with the Ru/Mo
and Rh/Mo catalysts, where the bulk of the Mo was found to be
amorphous, and in the oxidized form, it is possible that a similar
situation may arise with the Re component of the Ru/Re catalysts.
The most intense ‘Ru’ (1 0 1) line at 2h = 51.54° corresponds to a d-
spacing of 2.059 Å (cf. 2.056 Å for Ru metal), and thus any signifi-
cant lattice expansion associated with incorporation of Re into
the Ru lattice, as observed and tentatively suggested for Mo in
the Ru/Mo systems [2], is not evident here. Moreover, application
of the Scherrer equation to the Ru 2h lines at 51.54 and 94.35°
Rh
Re
300
250
200
150
100
5
0
0
10
20
30
40
50
60
70
80
90
100
Fig. 6. XRD profile of a fresh Rh/Re catalyst (Re:Rh = 0.78), including reference Rh
and Re 2h metal data points.
estimated to lie within the range ca. 2–4 nm. Individual Ru K and
Re L maps are shown in Fig. 8a and b, respectively; their compar-
ison showing clearly that wherever Ru is detected, Re is also pres-
ent, thus confirming the close association between Ru and Re; Re
alone is also evident, in much lower concentrations, distributed
across other areas of the sample, the homogeneous mixing provid-
ing a close parallel with the Ru/Mo and Rh/Mo catalysts. Quantifi-
cation of the constituent elements, selected by particle density,
over areas of the micrograph highlighted in green, Fig. 8c–e shows
that the Re:Ru atomic composition is largely homogeneous (ca.
[
Ru (1 0 3)], respectively, of which the latter is most free of inter-
ference from Re, gives average crystallite sizes of 8.4 and 8.5 nm.
These values are closely comparable with that (8.2 nm) found pre-
viously for metallic Ru derived from Ru
which proved to be a very poor catalyst for CyCONH
tion [2]. These results are however consistent with in situ HP-FTIR
observations during catalyst genesis, where decomposition of the
3
(CO)12 alone, material
hydrogena-
2
Ru
3
(CO)12 precursor was found to occur largely prior to, and par-
(CO)10
tially independently of, Re
2
.
1
9:81 at.%) throughout, notwithstanding the greater error inherent
3.7.1.2. Rh/Re catalyst. The diffractogram of a Re:Rh = 0.78 catalyst,
in the smaller area of analysis (e) which compares reasonably well
with the expected Re/Ru at.% composition of 15/85 based on
microanalysis. For a rationalization of the apparent discrepancy be-
tween the XRD data (Section 3.7.1.1) and these EDX-STEM results,
see Section 4.
shown in Fig. 6, appears consistent with the presence of both
metallic Rh and Re, the latter comprising a minor component
which again may be partly in the amorphous, presumably oxidized
form (see Section 3.7.1.1). The profile is significantly broader than
6
that of metallic Rh derived from Rh (CO)16 alone, for which a mean
crystallite size of 10.6 nm was estimated [1]. Application of the
Scherrer equation to the Rh (2 2 0) 2h line at 83.37° gives a mean
crystallite size of 3.8 nm, a similar value to those found for Rh/
Mo (and Ru/Mo) catalysts and apparently in sharp contrast to the
Ru/Re system.
3.7.3. XPS
3.7.3.1. Ru/Re catalysts. XPS data were recorded for two catalysts,
Re:Ru = 0.12 and 0.54, both before and after Ar sputter etching,
the former after use in cyclohexene hydrogenation, and the latter
+
2
after recycle following a single-pot CyCONH hydrogenation.
3
.7.2. EDX-STEM of Ru/Re catalyst (Re:Ru = 0.15)
A fresh catalyst of composition Re:Ru = 0.15 was examined
Assignment of the spectra was complicated by both the superim-
position of C1s and Ru 3d3/2 lines at ca. 280 eV, and the ‘Re 4f’ spec-
tral envelope comprising a composite of Re 4f (major) and Ru 4p
(minor) components. Both catalysts showed a dominant Ru (3d5/
using EDX-STEM, with an EDX spectrum to confirm the purity of
the sample, and several different areas were mapped. Fig. 7 shows,
in high magnification, the bright field image of an aggregate of
particles in one typical area, in which individual particle sizes are
2
) peak at 280.1 eV, consistent with the metallic Ru(0) state, to-
gether with a slight asymmetry on the high energy side that re-
+
duced during Ar sputter etching. The asymmetry could be
attributed to the presence of either RuO
2
(280.7 eV) and/or C
600
500
400
300
200
100
0
(
280.4 eV). Application of curve fitting procedures (details of which
for both these and Rh/Re catalysts, see Section 3.7.3.2, are de-
scribed in Supplementary material S3), to the ‘Re 4f’ envelopes
(Figs. 9 and 10), allowed the Re and Ru components to be resolved;
quantification data for M:Re (M = Ru or Rh) and Re valence state
distributions are summarized in Table 6.
Ru
Re
After curve fitting, the composite spectrum of the Re:Ru = 0.12
catalyst (Fig. 9) was found to be consistent with three components,
namely two distinct Re (4f7/2) lines at 40.3 and 41.4 eV (1 and 3
respectively), of which the former may be readily assigned to Re
metal by comparison with reference standards [16] and the third,
a Ru (4p3/2) line (5) at 43.1 eV, also assigned to the metallic state.
The second Re (4f7/2) component at 41.4 eV clearly corresponds
to an oxidized valence state of Re, and chemically, the sesquioxide
10
20
30
40
50
60
70
80
90
100
2θ
2 3
Re O [17] appears a likely possibility, although no XPS measure-
ments are available for this material. Support for the suggestion
of a Re(III) valence state is however provided by a Re (4f7/2) binding
Fig. 5. XRD profile of a fresh Ru/Re catalyst (Re:Ru = 0.23), including reference Ru
and Re metal 2h data points.

2
34
G. Beamson et al. / Journal of Catalysis 278 (2011) 228–238
Fig. 7. EDX-STEM analysis: bright field image of typical area of Re:Ru = 0.15 catalyst.
energy of 41.8 eV reported for the complex [ReCl
3
(PMe
2
Ph)
3
] [18].
Fig. S6 and Table 6) is consistent with the Re surface initially com-
prising three components, mainly Re(VI), together with Re(III) and
Re(0), both oxidized states being removed completely by Ar sput-
For comparison, ReO , ReO and Re have (4f7/2) binding energies
2
3
2 7
O
+
of 42.3, 44.3 and 46.7 eV, respectively [16].
The result of curve fitting of the spectrum of the Re:Ru = 0.54
catalyst (Fig. 10) provides a distinct contrast, revealing no fewer
than five Re valence states [Re (4f7/2) lines 1, 3, 5, 7 and 9, respec-
tively], comprising Re(0) and Re (III, IV, VI, and VII) (see Table 6).
Because of the higher Re content in this sample, the Ru 4p compo-
nents were considered to be too small to be meaningful. The oxi-
dized valence states, most notably Re(VII), also appeared
ter etching for 1 min, at which point the surface Re:Rh composition
is in close correspondence with the original stoichiometry used in
catalyst preparation. After etching, no Rh 4p (4p3/2 B.E. ꢁ48 eV)
intensity is evident, presumably due to the low sensitivity of the
4p signal.
XPS data on both Ru/Re and Rh/Re catalysts reveal that the Ru
and Rh surfaces are essentially metallic, in sharp contrast to the
Re components of the catalysts used in amide hydrogenation. In
addition to Re(0), these contain significant amounts of higher va-
lence states of Re, in the range Re(VII–III) with Ru, (VI and III) with
Rh, and which, in the case of Ru, are resistant towards complete re-
+
resistant to removal by Ar sputter etching (but in marked contrast
to Re(VI)); greater than 60% of the total oxidized component re-
mained after 5 min bombardment, during which the surface
Re(0) composition increased from 16.5% to 36.2%. In further con-
trast to the Re:Ru = 0.12 sample, the surface Re:Ru content of
+
moval by Ar bombardment, differences that are possibly attribut-
0
(
8
.43 is at greater variance with the initial stoichiometry used
0.54), which may be a consequence of a bimodal distribution of
nm and 2–4 nm Ru/Re particles determined for the system. The
able to the higher oxophilicity of Ru relative to Rh. An overlay of
the Re 4f spectra of the three catalysts (see Supplementary mate-
rial, Fig. S7) provides a direct comparison between the Ru/Re cata-
lyst used for cyclohexene hydrogenation and the two used in
amide reduction, clearly emphasizing the high levels of oxidized
Re on the latter two, which may be indicative of an active role
for such materials during amide reduction. In this context, reports
of the use of Re-oxo complexes such as [(OH)ReO ] as catalysts for
3
the dehydration of amides and aldoximes [19] could be significant
in mechanistic terms (see Section 6).
contrast between the final states of the two Ru/Re catalyst samples
may be readily rationalized by a consideration of the different
reactions involved in their genesis. Whereas cyclohexene hydroge-
nation over the Re:Ru = 0.12 catalyst occurs under essentially
anhydrous conditions, amide hydrogenation with the Re:Ru = 0.54
catalyst requires the elimination of a stoichiometric amount of
water during reduction, the presence of which presumably leads
to the observed extensive oxidative hydrolysis of the Re surface.
3
.7.3.2. Rh/Re catalysts. XPS data for the Re:Rh = 1.0 catalyst, both
4. Nature of the working catalysts
+
in the as-prepared form and after Ar sputter etching, show that
Rh is predominantly in the metallic state throughout, with a 3d5/
With Re:Ru and Re:Rh values obtained by ICP analysis in good
agreement with those expected from the nominal precursor ratios,
the Ru/Re- and Rh/Re-containing materials seem better defined
2
binding energy of 307.2 eV, and no evidence of Rh
2 3
O (308.5 eV
[
1]). Curve fitting of the Re 4f data (see Supplementary material,

G. Beamson et al. / Journal of Catalysis 278 (2011) 228–238
235
Bright field image
(a) Ru K
(b) Re L
(
c)
(d)
(e)
Ru = 81.3, Re = 18.7 at%
Ru = 81.2, Re = 18.8 at%
Ru = 79.4, Re = 20.6 at%
Fig. 8. EDX-STEM analysis of Re:Ru = 0.15 catalyst. (a) Ru K (19.279 keV) and (b) Re L (8.654 keV) maps. (c–e) Ru, Re quantification, areas of analysis depicted in light green.
For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
(
than their Ru/Mo and Rh/Mo counterparts in terms of elemental
composition. From XRD measurements, the Rh/Re catalysts contain
ca. 3–4 nm aggregates, considerably smaller than those obtained
2 x 103
1
10
6
using Rh (CO)16 alone (as also found with Rh/Mo catalysts). In con-
trast, based on XRD evidence alone, the Ru/Re catalysts appear to
comprise ca. 8-nm aggregates, as also found for Ru particles de-
rived from Ru (CO)12 alone. However, such Ru aggregates behave
3
very poorly as catalysts for amide hydrogenation [2] and are there-
fore unlikely to be responsible for the highly selective reduction of
2 2 2
CyCONH to CyCH NH . We presume therefore that the dominant
XRD pattern, interpreted in terms of 8 nm particles, must mask
the smaller homogeneously mixed particles of Ru and Re revealed
by the very much more sensitive EDX-STEM technique. As demon-
8
6
4
2
0
2
1
5
7
6
4
3
strated by in situ HP-FTIR, the rapid degradation of Ru
3
(CO)12 cou-
60
50
40
30
pled with the much slower decomposition of Re (CO)10 must
2
Binding Energy (eV)
account for the co-formation of a mixture of the larger Ru particles,
in tandem with catalytically active Ru/Re nanocluster generation.
Fig. 9. XPS data: Curve fit to Re 4f region of Re:Ru = 0.12. Components 1, 2: Re(0)
4
f
7/2
,
5/2; components 3, 4: Re(III) 4f7/2
,
1/2
5/2; components 5, 6: Ru(0) 4p3/2, ;
component 7: Ru 4p satellite.
5
. Comparison between Ru/Mo, Ru/Re, Rh/Mo and Rh/Re
catalysts
x 102
0
7
A comparison between the behaviour of Ru/Mo and Rh/Mo cat-
60
alysts for amide reduction has been reported previously [2]. Substi-
tution of Mo by Re has revealed a further dimension in terms of
variations, and the key features of all the catalyst combinations
investigated during this work are summarized in Table 7. Factors
that appear common to all catalysts are (i) the synergistic behav-
iour between each pair of elements in promoting amide reduction
under reaction conditions which encompass the ranges between
5
0
1
40
2
3
4
5
30
6
7
8
9
20
1
0
2
0 and 100 bar H
ties of these catalysts towards hydrogenation of the primary amide
CyCONH to the primary amine CyCH NH , and (iii) their recycla-
2
and 130 and 160 °C, (ii) intrinsic high selectivi-
10
2
2
2
0
bility. Whereas Ru/Mo, Rh/Mo and Rh/Re are all comparable in
exhibiting significant induction periods prior to initiation of catal-
ysis, Ru/Re only requires heating to reaction temperature. A conse-
quence is that the extremely rapid rate of decomposition of
6
0
50
40
30
Binding Energy (eV)
Fig. 10. XPS data: Curve fit to Re 4f region of Re:Ru = 0.54; components 1, 2: Re(0)
5/2; components 3, 4: Re(III) 4f7/2 5/2; components 5, 6: Re(IV) 4f7/2
components 7, 8: Re(VI) 4f7/2 5/2; components 9, 10: Re(VII) 4f7/2
3 2
Ru (CO)12, relative to Re (CO)10, leads to a bimodal distribution
4
f
7/2
,
,
5/2
, ;
,
5/2
, .
of ca. 8-nm Ru crystallites and ca. 2–4-nm Ru/Re nanoclusters.

2
36
G. Beamson et al. / Journal of Catalysis 278 (2011) 228–238
Table 6
XPS data. Quantification of Re:Ru = 0.12, Re:Ru = 0.54 and Re:Rh = 1.0 catalysts (M vs. Re%), and Re valence state distribution (%) as a function of Ar+
sputter etching time at 3 keV.
Catalyst
Ar+ sputter
M total (%)
Re total (%)
Re(0)
Re(III)
Re(IV)
Re(VI)
Re(VII)
etch time (min)
Re:Ru = 0.12
0
1
5
87.8
87.0
87.9
12.2
13.0
12.1
86.9
94.6
92.6
13.1
5.4
7.4
–
–
–
–
–
–
–
–
–
Re:Ru = 0.54
Re:Rh = 1.0
0
1
5
70.0
71.0
70.0
30.0
29.0
30.0
16.5
27.0
36.2
26.9
24.2
23.2
29.0
25.6
22.6
18.5
14.4
8.8
9.2
8.8
9.1
0
1
3
27.7
48.6
40.6
72.3
51.4
59.4
27.1
100
100
31.4
–
–
–
–
–
41.4
–
–
–
–
–
Table 7
Comparison between bimetallic catalysts for amide hydrogenation.
Ru/Mo
Rh/Mo
Ru/Re
Rh/Re
0
0
Optimum Mo:M /Re:M
CyCH NH selectivity (%)
pressure range (bar)
0.5
85
20–100
140
0.55
85
20–100
130
0.3–1.5
95
50–100
160
0.8
90
50–100
150
2
2
H
2
Minimum temperature (°C)
Nature of active catalyst
Ru/Mo and Mo oxides
Rh and Mo oxides
Ru/Re and Re oxides
Rh/Re and Re oxides
Ru/Mo, Ru/Re, Rh/Mo and Rh/Re catalysts are comparable in
requiring only small amounts of the second element i.e., Re and
Mo, for high primary amine selectivities, although Rh/Re catalysts
do require higher Re:Rh compositions to allow full amide conver-
sion. Contrasting features that emerge from the substitution of
Re for Mo are (i) little significant catalyst deactivation at high Re
levels, possibly because oxides of Re are much more readily re-
duced to the metallic state than Mo oxides and (ii) the requirement
this limitation did not necessarily extend to Ru or Rh, where the
Ru(III) precursor Ru(acac) [and Rh(I) precursor Rh(CO) (acac)]
(acac = acetylacetonato-, [CH ] ) were also shown
to be effective. Oxides of rhenium are much more readily reduced
to the metallic state than Mo oxides, cf. the in situ reduction of
3
2
–
C(O)CHC(O)CH
3 3
Re
hydrogenation of carboxylic acids [5,20]. A catalyst prepared using
Ru (CO)12 and Re was found to give CyCH NH in 90% selectiv-
ity, at 100% conversion of CyCONH [21], suggesting that in further
work Re might represent a more convenient precursor than
Re (CO)10 to Re-containing catalysts for amide hydrogenation.
2 7 4
O and OsO to give the bimetallic Re/Os catalysts used in the
3
2
O
7
2
2
for more severe reaction conditions, ca. 50–100 bar H
1
1
tions required by Ru/Mo and Ru/Re catalysts are 20/50 bar H
1
2
and 150–
60 °C with Ru/Re and Rh/Re-containing catalysts (cf. 20 bar H
50 °C with Ru/Mo catalysts). Minimum acceptable reaction condi-
and
60 °C, respectively, with almost identical (84% vs. 81%) primary
2
2
,
2 7
O
2
2
6
. Comments on mechanistic steps involved in the reduction of
amine selectivities, and thus Ru/Mo catalysts appear preferable
to their Ru/Re analogues, particularly in the light of a higher TON
amides
of 4.7 h– for Ru/Mo [2] vs. 1.9 h with Ru/Re (see Section 3.6.1),
in comparable experiments. Nevertheless, the real TON associated
with Ru/Re nanoclusters may in fact be significantly higher when
the co-presence of largely inactive 8-nm Ru particles is taken into
account.
1
–1
Notwithstanding extensive searching, we note an extreme pau-
city of information concerning mechanistic steps associated with
the reduction of amide functional groups.
0
00
0
00
RCONR R þ 2H
2
! RCH
2
NR R þ H
2
O
0
00
None of these catalysts require the customary co-addition of
ammonia or amines to inhibit the formation of secondary and ter-
tiary amine by-products. We previously commented on the forma-
tion of cyclohexanemethanol (in ca.14% selectivity) as the only
R;R ; R ¼ H; alkyl or aryl; etc:
ð1Þ
RCONH2
2
detectable by-product from CyCONH hydrogenation over Ru/Mo
catalysts [2]. Slightly less (ca. 10%) appears to be the norm with
the Re-containing systems reported here (with correspondingly
H2
-H O
2
H2
RCH OH
RCH(OH)NH2
RCN
2
-
NH3
H2
-
H O
2
slightly higher selectivities to CyCH
CH OH, presumably consequent upon CAN rather than CAO bond
scission, requires the elimination of one equivalent of NH per
equivalent of CyCONH consumed via this pathway (see Scheme 1),
leading to the possibility that this internal generation of NH might
be sufficient to suppress (CyCH NH formation.
Although the bulk of this work has been undertaken using the
zerovalent metal carbonyls Ru (CO)12, Rh (CO)16 and Re (CO)10
2
NH
2
). The formation of Cy-
2
RCH NHCH R
RCH=NH
H2
2
2
3
secondary RCH NH(C=O)R
products
2
2
3
2 2
)
RCH NH2
2
3
6
2
H O
2
2H2
-H O
RCONH2
RCO H
2
RCH OH
2
as catalyst precursors, other more readily available sources of Ru,
Rh and Re may also suffice. For Ru/Mo and Rh/Mo catalysts, the
-
NH3
2
6
Mo(0) precursor Mo(CO) appeared essential for both synergistic
Scheme 1. Postulated alternative reaction pathways, and intermediates, in the
amide conversion and primary amine selectivity [1,2]. However,
reduction of primary amides.

G. Beamson et al. / Journal of Catalysis 278 (2011) 228–238
237
Two principal pathways appear to suggest themselves
Scheme 1). First, addition of H across the C@N bond of the ther-
modynamically strongly disfavoured iminol tautomer, which may
nevertheless be kinetically significant, to give the carbinolamine
respectively. Concerted, or stepwise addition of H
the intermediacy of adsorbed CyCH(O)NH
CyCH NH formation and desorption. Removal of the co-formed
chemisorbed O, as H O, by reaction with the second equivalent
of H , would not only lead to catalyst regeneration but also invoke
2
, the latter via
2
, would then lead to
(
2
2
2
2
(
hemiaminal) would intuitively seem to be more favoured than di-
2
rect addition of H across the strongly deactivated C@O group of
2
a role for redox behaviour on the surface during amide reduction.
The contrasting nature of surface species associated with Ru/Re
catalysts used for cyclohexene hydrogenation and amide reduc-
tion, as derived from XPS measurements (cf. Section 3.7.3), pro-
vides supporting circumstantial evidence for this possibility.
Moreover, such behaviour may also account for the lack of a close
correlation between amide and nitrile hydrogenation (see, Supple-
mentary material S4) with similar catalysts. Here, very high pri-
mary amine selectivities were a feature of both monometallic and
bimetallic CyCN hydrogenation catalysts examined (Ru, Ru/Re, Ru/
Mo, Rh and Rh/Mo), in marked contrast to the behaviour of mono-
metallic amide reduction catalysts. Also, the variable synergism
observed with nitrile hydrogenation catalysts following the intro-
duction of Re and Mo components may be a direct consequence
of the ‘anhydrous’ nature of these systems and the absence of re-
dox catalysis, thus rendering the amide and nitrile reduction sys-
tems incomparable in direct terms.
the amide. Addition of a second equivalent of hydrogen with con-
certed elimination of water would complete the reduction se-
quence. Alternatively, initial dehydration of the carbinolamine to
the imine, followed by direct addition of hydrogen could also allow
completion of the reduction sequence. A second ‘reverse’ pathway,
available to primary amides, concerns initial dehydration of the
amide to the corresponding nitrile, followed by sequential hydro-
genation with two equivalents of hydrogen.
ꢀ
H
2
O
H
2
H
2
RCONH
2
! RCN ! RCH ¼ NH ! RCH
2
NH
2
ð2Þ
Free energy changes for potential intermediates in the two
alternative reduction pathways, for R = Cy, cyclohexyl, have been
estimated using group additivity methods [22,23] and the results
ꢂ
2
are summarized in Table 8.
carbinolamine CyCH(OH)NH
1
DG
for the formation of the
98:15
2
from CyCONH
2
is estimated to be
ꢀ1
04.8 kJ mol (Entry 2) [23], a value far in excess of the amide
Finally, transient imine intermediates may be common to all
mechanistic pathways, and some consideration of the evidence
for their existence is appropriate. Imines have frequently been pos-
tulated, particularly in the context of intermediates in the reduc-
tion of nitriles [24,25], and the apparent lack of direct evidence
for their existence attributed to high reactivity. Nevertheless, it is
pertinent to note that coordinated imines have in fact been isolated
and characterized during model studies in organometallic chemis-
try directed towards the definition of intermediate stages in the
reduction of nitriles [26,27]. Moreover, using NMR spectroscopy,
Fryzuk et al. [28] demonstrated the formation of bridged imines
and their conversion to coordinated secondary amines during stoi-
chiometric reactions of nitriles with Rh hydride-containing com-
plexes. Here, temperatures as low as 77 K were required to
monitor their reactions because of the extremely high lability of
even these organometallic systems, in which the imine would be
expected to be ‘stabilized’ by coordination, thus providing
unequivocal confirmation of their extremely high reactivity. Thus,
although there is little direct evidence for the involvement of imine
intermediates, postulation of their transient existence certainly
conveniently accounts for the formation of secondary and tertiary
amine by-products commonly formed during the reduction of both
amides (Scheme 1) and nitriles. Furthermore, the detection and
unequivocal characterization of the secondary amide Cy-
dehydration barrier (Entry 4), suggesting that amide dehydration
might actually provide a more favourable reaction pathway for
the reduction of primary amides. The carbinolamine actually
proves thermodynamically unstable with respect to imine forma-
tion and elimination of water, cf. Entries 2 and 3, thereby favouring
this option over the alternative addition of H
ination of water referred to above.
2
and concerted elim-
The operation of an amide dehydration pathway may clearly
play a role in controlling the unusually high selectivity towards
CyCH
bit CyCH
2
NH
2
, because removal of water in the first step should inhi-
OH formation. Also, if dehydration to the nitrile were to
2
be rate-limiting (considering the higher reaction temperatures,
i.e.130–160 °C, required relative to those required by standard ni-
trile hydrogenation catalysts [24,25]), the low standing concentra-
tion of nitrile, and presumably very rapid subsequent reduction to
amine, could reasonably be expected to lead to a limitation of sec-
ondary amine by-product formation, as observed experimentally.
Clearly both postulated mechanistic pathways may operate
concurrently, but the potential availability of an additional route
for primary amide reduction may perhaps also go some way
towards providing an explanation for the demonstrated difference
in reactivity of e.g., Ru/Mo catalysts towards primary, secondary
and tertiary amide reduction, in the order CyCONH
2
> CyC-
ONMe
2
ꢃ CyCONHMe [2].
2
C(O)N(CH Cy)H as a minor by-product during the reduction of Cy-
Whilst providing useful background information, the free en-
ergy estimates described above only give an idealistic representa-
tion of the reaction pathways because they do not take account of
the effect of the requirement for adsorption of the reactants and
intermediates (i.e. carbinolamine, imine) on the catalyst surface.
In this context, it seems probable that a preferred orientation of
the substrate on the catalyst surface would require adsorption of
the amide N and carbonyl O at Ru/Rh and Re/Re oxide interfaces
CONH over Rh/Mo catalysts [1] provides further circumstantial
2
evidence in support of the transient existence of CyCH = NH.
7. Summary and conclusions
The use of recyclable, bimetallic Re-containing heterogeneous
catalysts for highly selective reduction of the primary amide
Table 8
Thermodynamic parameters relevant to amide hydrogenation.
ꢀ
1
ꢀ1 ꢀ1
ꢂ
2
ꢀ1
Entry
Reaction
D
H° kJ mol
D
S° Jmol
K
DG
kJ mol
98:15
1
2
3
4
5
6
7
CyCONH
CyCONH
CyCONH
CyCONH
2
2
2
2
+ 2H
2
? CyCH
? CyCH(OH)NH
? CyCH = NH + H
2
NH
2
+ H
2
O
ꢀ49.8
51.5
63.2
ꢀ72.8
ꢀ178.7
ꢀ21.1
ꢀ28.1
104.8
69.5
26.5
43.0
+ H
+ H
2
2
2
2
O
? CyCN + H
? CyCH = NH
CyCH = NH + H ? CyCH
CyCN + 2H ? CyCH NH
2
O
73.2
156.5
CyCN + H
2
ꢀ10.0
ꢀ113.0
ꢀ123.0
ꢀ177.6
2
2
NH
2
ꢀ51.7
ꢀ97.6
2
2
2
ꢀ229.7
ꢀ54.6

2
38
G. Beamson et al. / Journal of Catalysis 278 (2011) 228–238
CyCONH
amine, CyCH
2
, in the liquid phase, to the corresponding primary
NH , has been demonstrated for the first time.
Appendix A. Supplementary material
2
2
These Ru/Re and Rh/Re catalysts are effective under considerably
milder reaction conditions than those required by earlier gener-
ation catalysts such as copper chromite. The new nanocluster
catalysts, prepared in situ from the parent metal carbonyls, have
only been investigated in the unsupported state, whereas sup-
ported catalysts would clearly be preferable for larger scale
use. Our preliminary attempts to support related Rh/Mo catalysts
proved largely unsuccessful, notwithstanding their displaying
higher initial TONs in amide reduction than the unsupported
materials [1]. This was a consequence of accumulation of C, H
and N residues on the silica support used during recycling,
resulting in very significant deactivation. Future work in this
area should be directed towards this aspect, with careful atten-
tion to the selection and use of alternative supports, e.g. high
surface area carbon, or possibly other reducible oxides such as
titania [6]. The use of more readily available catalyst precursors
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.jcat.2010.12.009.
References
[
[
[
[
[
[
[
1] G. Beamson, A.J. Papworth, C. Philipps, A.M. Smith, R. Whyman, J. Catal. 269
(2010) 93–102.
2] G. Beamson, A.J. Papworth, C. Philipps, A.M. Smith, R. Whyman, Adv. Synth.
Catal. 352 (2010) 869–883.
3] B. Wojcik, H. Adkins, J. Am. Chem. Soc. 56 (1934) 247;
2419–2424.
4] H.S. Broadbent, G.C. Campbell, W.J. Bartley, J.H. Johnson, J. Org. Chem. 24
(
1959) 1847–1854.
5] K. Yoshino, Y. Kajiwara, N. Takaishi, Y. Inamoto, J. Tsuji, J. Am. Oil Chem. Soc. 67
(1990) 21–24.
6] H.G. Manyar, C. Paun, R. Pilus, D.W. Rooney, J.M. Thompson, C. Hardacre, Chem.
Commun. 46 (2010) 6279–6281.
7] H.S. Broadbent, W.J. Bartley, J. Org. Chem. 28 (1963) 2345–2347.
[8] I.D. Dobson, Eur. Patent, 286 280, 1988 (to BP).
than the metal carbonyls, e.g. [Ru(acac)
3
], Re
2
O
7
and [Rh(CO)
2
(a-
[9] C. Hirosawa, N. Wakasa, T. Fuchikami, Tet. Lett. 37 (1996) 6749–6752.
10] M. Balcerzak, Crit. Rev. Anal. Chem. 32 (2002) 181–226.
11] J.R. Johnson, H.D. Kaesz, Inorg. Synth. 18 (1978) 60–62.
12] M. Herberhold, G. Süss, J. Ellerman, H. Gäblein, Chem. Ber. 111 (1978) 2931–
2941.
[
[
[
cac)], may also prove advantageous. Furthermore, the general ap-
proach should be extended towards secondary amides, the
hydrogenation of which appears to be considerably more
difficult than primary amides in both catalytic and stoichiome-
[
[
[
13] A. Davison, J.A. McCleverty, G. Wilkinson, J. Chem. Soc. (1963) 1133–1138.
14] S. Martinengo, P. Chini, Gazz. Chim. Ital. 102 (1972) 344–354.
15] H.E. Swanson, R.K. Fuyat, Natl. Bur. Stand. (US) Circ 539, vol. II, 1953, p. 13.
tric reactions (the latter using hydride reagents such as LiAlH
2], and references therein). In contrast, reduction of the tertiary
4
[
[16] A. Cimino, B.A. De Angelis, D. Gazzoli, M. Valegi, Z. Anorg. Allg. Chem. 460
1980) 86–98.
(
amides that are commonly used as models of amide hydrogena-
tion, e.g. N-acetylpiperidine and N-acetylpyrrolidine, is generally
trivial, principally because amine product selectivity only poses a
minor problem, certainly with the bimetallic catalysts described
here. Other functional groups that are well known for their dif-
ficult reducibility include carboxylic acids and esters, and these
may also be accessible to these catalysts. Finally, significant re-
cent reports of the reduction of amides to the corresponding
amines (and alcohols) under mild reaction conditions include
the use of homogeneous bipyridyl-based Ru [29], Fe and Zn/tri-
ethoxysilane-mediated [30] catalysts, indicative of a resurgence
of interest in this challenging area.
[
[
17] W. Geilmann, F.W. Wrigge, Z. Anorg. Chem. 214 (1933) 239–243.
18] G.J. Leigh, W. Bremser, J. Chem. Soc. Dalton Trans. (1972) 1216–1219.
[19] K. Ishihara, Y. Furuya, H. Yamamoto, Angew. Chem. Internat. Edit. 16 (2002)
983–2986.
[
[
[22] S.E. Stein, R.L. Brown, ‘‘Structures and Properties Group Additivity Model’’ in
NIST chemistry WebBook, NIST standard reference database number 69, in: P.J.
Linstrom, W.G. Mallard (Eds.), March 2003, National Institute of Standards and
Technology, Gaithersburg, MD, 20899. http://webbook.nist.gov.
[23] S.W. Benson, Thermochemical Kinetics: Methods for the Estimation of
Thermochemical Data and Rate Parameters, John Wiley & Sons, New York,
2
20] C. Philipps, Ph.D. Thesis, University of Liverpool, 2004.
21] A.M. Smith, Ph.D. Thesis, University of Liverpool, 2006.
1968.
[
24] S. Nishimura, Handbook of Heterogeneous Catalytic Hydrogenation for
Organic Synthesis, John Wiley and Sons, New York, 2001 (and references
therein).
[
[
25] J. Volf, J. Pašek, Catalytic hydrogenation, in: L. Cerverny (Ed.), Elsevier Science,
Amsterdam, 1986, pp. 105–145.
26] M.A. Andrews, H.D. Kaesz, J. Am. Chem. Soc. 101 (1979) 7255–7259.
Acknowledgments
The authors acknowledge helpful discussions with Dr. J.
Claridge and Dr. J.A. Iggo (Department of Chemistry) and Mr. A.J.
Pettman (Pfizer Ltd.), EPSRC for financial support of NCESS, Dares-
bury, under grant GR/S14252/01, and the Leverhulme Fine
Chemicals Forum for financial support (to C.P. and A.M.S.).
[27] Z. Dawoodi, M.J. Mays, K. Henrick, J. Chem. Soc. Dalton Trans. (1984) 433–440.
[
[
28] M.D. Fryzuk, W.E. Piers, S.J. Rettig, Can. J. Chem. 70 (1992) 2381–2389.
29] E. Balaraman, B. Gnanaprakasam, L.J.W. Shimon, D. Milstein, J. Am. Chem. Soc.
1
32 (2010) 16756–16758.
[30] S. Das, S. Zhou, D. Addis, S. Enthaler, K. Junge, M. Beller, Top. Catal. 53 (2010)
79–984.
9