Selective hydrogenation of amides using ruthenium/ molybdenum catalysts

Article abstract of DOI:10.1002/adsc.200900824

Recyclable, heterogeneous bimetallic ruthenium/molybdenum catalysts, formed in situ from triruthenium dodecacarbonyl [Ru₃(CO)₁₂] and molybdenum hexacarbonyl [Mo(CO)₆], are effective for the selective liquid phase hydrogenation of cyclohexylcarboxamide (CyCONH₂) to cyclohexanemethylamine (CyCH₂NH₂), with no secondary or tertiary amine by-product formation. Variation of Mo:Ru composition reveals both synergistic and poisoning effects, with the optimum combination of conversion and selectivity at ca. 0.5, and total inhibition of catalysis evident at ≥1. Good amide conversions are noted within the reaction condition regimes 20100 bar hydrogen and 145-160°C. The order of reactivity of these catalysts towards reduction of different amide functional groups is primary > tertiary ? secondary. In situ HP-FT-IR spectroscopy confirms that catalyst genesis occurs during an induction period associated with decomposition of the organometallic precursors. Ex situ characterisation, using XRD, XPS and EDX-STEM, for active Mo:Ru compositions, has provided evidence for intimately mixed ca. 2.5-4 nm particles that contain metallic ruthenium, and molybdenum (in several oxidation states, including zero).

Full text of DOI:10.1002/adsc.200900824

FULL PAPERS
DOI: 10.1002/adsc.200900824
Selective Hydrogenation of Amides using Ruthenium/
Molybdenum Catalysts
b
c
a
a
Graham Beamson, Adam J Papworth, Charles Philipps, Andrew M Smith,
a,
and Robin Whyman *
Department of Chemistry, Donnan and Robert Robinson Laboratories, University of Liverpool, Liverpool L69 7ZD, U.K.
Fax : (+44)-151-794-3588; phone: (+44)-151-794-3515; e-mail: whyman@liv.ac.uk
National Centre for Electron Spectroscopy and Surface Analysis, STFC, Daresbury Laboratory, Warrington, Cheshire
WA4 4AD, U.K.
Materials Science and Engineering, Department of Engineering, University of Liverpool, Liverpool L69 3GH, U.K.
a
b
c
Received: November 26, 2009; Revised: February 18, 2010; Published online: March 12, 2010
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/adsc.200900824.
Abstract: Recyclable, heterogeneous bimetallic amide functional groups is primary>tertiary@sec-
ruthenium/molybdenum catalysts, formed in situ ondary. In situ HP-FT-IR spectroscopy confirms that
from triruthenium dodecacarbonyl [Ru (CO) ] and catalyst genesis occurs during an induction period as-
3
12
molybdenum hexacarbonyl [Mo(CO) ], are effective sociated with decomposition of the organometallic
6
for the selective liquid phase hydrogenation of cyclo- precursors. Ex situ characterisation, using XRD, XPS
hexylcarboxamide (CyCONH ) to cyclohexaneme- and EDX-STEM, for active Mo:Ru compositions,
2
thylamine (CyCH NH ), with no secondary or terti- has provided evidence for intimately mixed ca. 2.5–
2
2
ary amine by-product formation. Variation of Mo:Ru 4 nm particles that contain metallic ruthenium, and
composition reveals both synergistic and poisoning molybdenum (in several oxidation states, including
effects, with the optimum combination of conversion zero).
and selectivity at ca. 0.5, and total inhibition of catal-
ysis evident at ꢀ1. Good amide conversions are
noted within the reaction condition regimes 20– Keywords: amide hydrogenation; heterogeneous bi-
100 bar hydrogen and 145–1608C. The order of reac- metallic catalysis; high selectivity; molybdenum;
tivity of these catalysts towards reduction of different ruthenium
Introduction
groups. The discovery and development of catalytic
methods that are effective under mild reaction condi-
The discovery and development of methods for the tions, ideally ꢁ30 bar H and 708C, represents a chal-
2
selective reduction of ꢀdifficultꢁ functional groups lenge that is of particular significance to the fine
such as amides, carboxylic acids and esters has been chemicals industry. Recently a step change in this di-
[
1]
recognised for many years as a formidable problem,
rection has been provided by reports of a new range
within which grouping amides are acknowledged to of ꢀbimetallicꢁ catalysts, particularly those derived
represent the most extreme difficulty. The current from organometallic precursors such as Rh and Mo
[
3–5]
state of the art requires the use of either expensive re- carbonyls, initially described as homogeneous,
agents, such as LiAlH , in stoichiometric quantities typically exemplified by the reduction of the tertiary
and
4
(
with associated environmental clean-up penalties), or amide N-acetylpiperidine to N-ethylpiperidine at
[
4]
traditional copper chromite-based catalysts which, al- 100 bar H and 1608C. We have described details of
2
though effective, require very severe reaction condi- the genesis and characterisation of such catalysts,
tions, e.g., a minimum of 200 bar H and 2508C, and formed in situ from Rh (CO) and Mo(CO) , and un-
2
6
16
6
[
2]
high catalyst loadings (ca. 20 wt%). These are clear- equivocally demonstrated that they are in fact hetero-
[
6]
ly unsuitable for the production of intermediates in geneous. Their performance in the selective hydro-
the manufacture of, for example, pharmaceuticals, genation of a representative primary amide such as
which frequently contain multiple sensitive functional cyclohexylcarboxamide to the corresponding primary
Adv. Synth. Catal. 2010, 352, 869 – 883
ꢂ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
869

Graham Beamson et al.
FULL PAPERS
amine, cyclohexanemethylamine [(Eq. (1)] in up to
Bimetallic Ru-containing catalysts are well estab-
8
7% selectivity, was found to be critically dependent lished for carbonyl group reduction, see for exam-
[7,8]
on the Mo:Rh composition. The only by-products de- ple,
and ruthenium should therefore also represent
a strong candidate for amide hydrogenation, a view
that is reinforced by the knowledge that Ru is also an
effective catalyst for the selective reduction of nitriles
[
9]
to amines. Here we report the genesis of a family of
Ru/Mo catalysts, prepared in situ in an analogous
manner to the Rh/Mo systems, and their performance
in the reduction of a representative selection of pri-
tected in significant amounts were the secondary mary, secondary and tertiary amides. An analogous
amine
N-(cyclohexylmethyl)-cyclohexanemethane- terminology to that used in the Rh/Mo work has been
ACHTUNGTRENNUNGa mine, (CyCH ) NH and cyclohexanemethanol, employed, namely Ru/Mo, except where reference
2
2
CyCH OH (cf. proposed reaction pathways, summar- has been made to variations in Ru/Mo content and
2
ised in Scheme 1), together with traces (typically where the reciprocal terminology Mo:Ru most effec-
<
1%) of methylcyclohexane.
tively highlights the very significant effects resulting
from addition of minor quantities of Mo to poorly
performing Ru catalysts.
Results and Discussion
CyCONH Hydrogenation using Ru/Mo Catalysts
2
The results of preliminary experiments with
CyCONH using various Ru catalyst precursors, and
2
two to which Mo(CO) have been added, are summar-
6
ised in Table 1.
With Ru alone (entries 1 and 2) no primary amine
was formed, and the product distribution approximat-
ed to a 2:1 mixture of CyCH OH and (CyCH ) NH at
2
2 2
similar CyCONH conversion levels of ca. 35%. The
2
co-addition of Mo(CO) at an Mo:Ru atomic ratio of
6
0.50 (entry 3) resulted in a profound effect on catalyt-
ic behaviour, with both considerably improved amide
conversion but far more importantly product selectivi-
ty, namely a complete switch in product distribution
to a ca. 6:1 mixture of the desired primary amine
CyCH NH and CyCH OH, with total absence of sec-
Scheme 1. Primary amide hydrogenation: reaction pathways
to by-products, highlighting the significance of the postulat-
†
2
2
2
ed imine intermediate. There is a paucity of information
ondary amine formation, thus representing a highly
concerning mechanistic reaction steps involved in amide hy-
drogenation. Although there is no direct evidence for the in-
[2]
unusual feature of amide reduction chemistry. In
volvement of imine intermediates their transient presence contrast, reference Ru/MoO and Ru/Mo powder cat-
3
does conveniently account for secondary and tertiary amine alysts prepared using Ru (CO) (entries 5 and 6)
3
12
by-product formation in the reduction of amides (and ni- both showed behaviour that was typical of Ru alone.
triles).
[a]
Table 1. CyCONH hydrogenation: conversion and product selectivity using bimetallic Ru/Mo and Ru catalyst precursors.
2
Entry
Catalyst precursor
Ru (CO)
Conversion [%]
Product Selectivity [%]
CyCH NH
CyCH OH
ACHTUNGTNERUNNG( CyCH ) NH
2 2
2
2
2
1
2
3
4
5
6
35
32
100
100
35
0
0
85
85
0
62
68
14
10
66
64
36
31
0
4
29
33
3
12
Ru/C
Ru (CO) /Mo(CO)
3
12
6
Ru
Ru (CO) /MoO
3
A
C
H
T
U
N
G
T
R
E
N
N
U
N
G
(acac) /Mo(CO)
3
6
3
12
Ru (CO) /Mo powder
36
0
3
12
[
a]
Catalyst concentration: 5 mol% Ru; Mo:Ru=0.50; reaction conditions: 100 bar H , 1608C, 16 h.
2
870
asc.wiley-vch.de
ꢂ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2010, 352, 869 – 883

Selective Hydrogenation of Amides using Ruthenium/Molybdenum Catalysts
Figure 2. CyCONH
1693 cm amide carbonyl absorption band monitored by
HP-FT-IR using an Mo:Ru=0.50 catalyst at 100 bar H and
1608C.
hydrogenation: decay profile of
2
À1
À1
Figure 1. In situ HP-FT-IR spectra (2100–1950 cm ) during
catalyst genesis using Ru (CO) /Mo(CO) in DME at
3
12
6
2
1
00 bar H , 1608C.
2
In situ HP-FT-IR spectroscopy (in the range 2100–
À1
1
950 cm ) was used to monitor changes that occurred
as a function of time using the catalyst composition
corresponding to Table 1, entry 3. Figure 1 shows that
loss of all Ru/Mo metal n(CO) absorptions was com-
plete after the first 130 min of reaction at 1608C.
CyCONH reduction was initiated after this induction
2
period, as shown by the decay profile of the amide
À1
carbonyl absorption at 1693 cm (Figure 2). Analysis
of this curve, using the first four points from the end
of the induction period, i.e., over the time period
1
00–170 min, gave an estimated initial TON (mmol
À1
CyCONH consumed vs. mmol Ru) of 4.7 h . After
2
reaction at 100 bar H and 1608C for 16 h the product
2
solutions appeared colloidal [see Supporting Informa-
tion, Figure S1 (b)]. Settlement occurred slowly on
standing, with the deposition of black residues, leav-
ing a clear supernatant product solution [see Support-
ing Information, Figure S1 (c)], suggesting that the
real catalysts were heterogeneous. Analysis of spec-
troscopic changes occurring during the initiation
period, together with results of control experiments
Figure 3. CyCONH hydrogenation: conversion and product
selectivity vs. variation in Mo:Ru catalyst composition
2
(
100 bar H , 1608C and 16 h). Key:
&
conversion,
*
2
CyCH NH +(CyCH ) NH, ~ CyCH OH.
2
2
2
2
2
using, e.g., Ru (CO) alone, have enabled the assem- dent that only very minor quantities of Mo(CO)6
3
12
bly of a complete picture of the molecular species (e.g., Mo:Ru=0.06) were necessary to initiate syner-
present in solution during catalyst genesis (see gistic effects. The optimum combination of amide
below).
conversion and primary amine selectivity (typically
5% CyCH NH and 14% CyCH OH) was found
8
2
2
2
with Mo:Ru values of ca. 0.5; furthermore, complete
catalyst deactivation was evident at Mo:Ru ꢀ1. With
increasing Mo content over the composition range
Variation of Mo:Ru Composition
Figure 3 shows the effects, on conversion and product Mo:Ru=0–0.5 selectivity towards the primary amine
selectivity, of variation of Mo:Ru, from which it is evi- increased as that towards CyCH OH decreased, with
2
Adv. Synth. Catal. 2010, 352, 869 – 883
ꢂ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
871

Graham Beamson et al.
FULL PAPERS
the converse applying over the range Mo:Ru=0.5–
The explanation for the limiting temperature de-
0.8. Significant secondary amine formation (36% se- pendence on conversion seems most likely to be relat-
lectivity at 35% conversion) was only evident in the ed to the balance between the adsorption/desorption
absence of Mo.
behaviour of reactants and products. In this respect
amide adsorption would be expected to be more sen-
sitive than amine desorption. Amide adsorption
would become less favoured with increasing tempera-
ture, whereas amine desorption should be less sensi-
Variation of Pressure and Temperature
The effects of pressure and temperature variations on tive and possibly largely independent of temperature.
CyCONH reduction, using the optimum Mo:Ru= It is therefore possible that, at temperatures below
2
0.50 catalyst precursor, are summarised in Table 2, the threshold for catalytic reduction, the amide sub-
from which it is clear that primary amine selectivity strate is simply too strongly chemisorbed on the Ru/
was essentially independent of hydrogen pressure Mo surface. Such temperature limitations will be sen-
over the range 100–10 bar; moreover a significant de- sitive to the intimate nature of the catalyst and, for
crease in conversion was only detected between 20 example, the recently reported Rh/Mo catalysts ex-
and 10 bar (entries 1–4). In contrast, conversions ap- hibited high CyCONH conversions at 1308C (and
2
peared extremely temperature dependent, showing a 100 bar H ) but with significantly reduced primary
2
very marked decline between 145 and 1408C (cf. en- amine selectivity (53%), and both (CyCH ) NH
2
2
[
6]
tries 1, 5–7), also accompanied by a minor decrease of (23%) and CyCH OH (19%) formation.
2
primary amine selectivity, and zero conversion at
1308C. These temperatures represent a limiting re-
quirement for reactions in which the catalyst is Additional Reaction Variables
formed in situ. It is noteworthy that no secondary
amine was formed at 100 bar H throughout the tem- The results of variation of other reaction parameters
2
perature range 160–1408C, although traces were de- such as solvent, exposure of the systems to CO and
tected at 50 bar H and below. Temperature would be air, and attempted in situ removal of water during
2
expected to exert a significant influence on the length Ru/Mo-catalysed CyCONH reduction, are detailed
2
of the induction period required for full decomposi- in the Supporting Information S1.
tion of the metal carbonyl precursors, with lower tem-
peratures likely to necessitate longer times. Thus,
slower/incomplete decomposition might account for Catalyst Separation and Recycle
the reduced conversions noted at lower temperatures.
This limitation should not necessarily apply to pre- After all ꢀsingle-potꢁ reactions, complete settlement of
formed catalysts, although even here (entries 8 and 9) the fine particulates from DME solution could re-
it is quite evident that conversions were substantially quire an extended period of time (see Supporting In-
reduced at lower temperatures, notwithstanding con- formation, Figure S1) and this might account for ear-
siderably extended reaction times. Reaction condition lier suggestions that these were homogeneous cata-
[
3–5]
combinations of either 20 bar H /1608C or 100 bar H / lysts.
1
Nevertheless, used catalysts required no fur-
458C appear to be the lower limiting operating pa- ther treatment, other than separation and washing,
rameters for reduction of CyCONH2. and could be readily recycled under the standard
2
2
Table 2. CyCONH hydrogenation: conversion and product selectivity vs. pressure and temperature (Mo:Ru=0.50 catalyst).
2
Entry
T [8C]
PH2 [bar]
Conversion [%]
Product Selectivity [%]
CyCH NH
CyCH OH
ACHTUNGTNERUNNG( CyCH ) NH
2 2
2
2
2
1
2
3
4
5
6
7
8
9
160
160
160
160
150
145
140[
100
50
20
100
99
97
63
100
99
44
70
10
85
84
85
87
83
83
80
77
14
12
13
11
16
17
20
23
–
0
3
2
1
0
0
0
0
–
10
100
100
100
100
100
a]
130[
b]
120
major product
[
[
a]
b]
Using pre-formed catalyst; reaction time: 68 h.
Using pre-formed catalyst; reaction time: 96 h.
872
asc.wiley-vch.de
ꢂ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2010, 352, 869 – 883

Selective Hydrogenation of Amides using Ruthenium/Molybdenum Catalysts
cyclohexanecarboxamide, was evaluated and the re-
sults are summarised in Table 3. These substantiate
the general applicability of Ru/Mo catalysts towards
amide reduction and provide a qualitative evaluation
of the relative ease of hydrogenation in the order pri-
mary>tertiary@secondary.
Amongst primary amides, benzamide (Table 3,
entry 1), the aromatic analogue of CyCONH , which
2
should be more activated towards reduction because
of the electron-withdrawing nature of the aromatic
ring, did indeed display extremely similar behaviour
in respect of both conversion and selectivity (cf.
Table 1, entry 3). This is presumably a reflection and
confirmation that, here at least, hydrogenation of the
aromatic ring is facile in relation to the far more de-
manding nature of amide functional group reduction,
and there is thus no suggestion of catalyst inhibition
by the presence of the aromatic ring (see below). The
aliphatic primary amide n-butyramide (Table 3,
entry 2) afforded n-butylamine in high selectivity with
Figure 4. CyCONH2 hydrogenation: product selectivity vs.
recycle using a Mo:Ru=0.57 catalyst. Key: & conversion,
&
CyCH NH , & CyCH OH, & (CyCH ) NH. Cycle 6 with
2 2 2 2 2
addition of LiOH.
processing conditions. The behaviour of a representa- no detectable secondary amine formation, whereas
tive optimum catalyst (Mo:Ru=0.57) towards recycle the more sterically hindered trimethylacetamide
is indicated in Figure 4.
(entry 3) yielded predominantly 2,2-dimethylpropan-
Complete amide conversions are evident through- 1-ol and only 40% selectivity towards the expected
out and CyCH OH formation remains essentially con- product, 2,2-dimethylpropanamine. The secondary
2
stant, although primary amine selectivity decreases amides N-methylbenzamide (entry 4) and N-methyl-
slightly during cycles 3–5, in which the first appear- cyclohexanecarboxamide (entry 5) proved significant-
ance and increase in secondary amine formation is ly more resistant to reduction. Although the aromatic
evident. This variation in selectivity seems most likely ring was readily reduced, only minor overall conver-
to be a consequence of sequential retention and accu- sions to the secondary amine were evident. Reaction
mulation of minor amounts of insoluble C-, H-, and of the secondary amide N-methylformamide,
N-containing residues on the catalyst surface. Such HCONHCH , was also followed by in situ HP-FT-IR
3
material would be resistant to removal during catalyst spectroscopy. Some reduction was evident from a sig-
separation and centrifugation and may lead to more nificant decrease in intensity of the n(CO) amide ab-
À1
prolonged residence times of reactants and intermedi- sorption at 1694 cm , but it proved impractical to
ates on the catalyst surface, thus favouring condensa- quantify the product distribution by GC analysis be-
tion reactions with the postulated imine intermediate cause of the high volatility of the principal reaction
(
(
cf. Scheme 1), and
a
resultant increase in product, dimethylamine, detected by its odour in the
CyCH ) NH production, as observed. However, addi- off-gases during depressurisation. Tertiary amides
2
2
tion of trace quantities of a strong base in the form of (Table 3, entries 6–9) were more readily reduced than
LiOH·H O (1 mg), in accordance with a literature secondary amides (cf. entries 4 and 5), although with
2
[
10]
precedent for the hydrogenation of nitriles, prior to wider variation in product selectivity. N,N-Dimethyl-
cycle 6 resulted in complete suppression of secondary benzamide (entry 6) showed similar behaviour to that
amine formation, and restoration of the initial pri- of benzamide (entry 1) in terms of preferred aromatic
mary amine selectivity to 90% (cf. cycle 2). The role ring reduction, but the high residual CyCON
ACHTUNGTRENNUNG( CH )
3 2
of LiOH addition has previously been attributed to content and only 30% overall conversion to CyCH N-
2
favouring rapid desorption of the amine, thus leading
to inhibition of side reactions.
ACHTUNGTNERNUN(G CH ) suggested significant catalyst inhibition. Never-
3 2
thless, a similar 30% overall amide to amine conver-
[
9]
sion was evident with N,N-dimethylcyclohexane-car-
boxamide
(entry 7).
N,N-Diethylpropionamide
Variation of Amide Substrates
(entry 8), an example of a conventional aliphatic terti-
ary amide, yielded N,N-diethylpropylamine as the
The behaviour of the optimum (Mo:Ru=0.50) cata- major product. N-Acetylpiperidine (NAP; entry 9)
lyst composition for CyCONH reduction towards a underwent complete conversion to N-ethylpiperidine
2
representative selection of aliphatic, cyclo-aliphatic in 90% selectivity, behaviour which is very similar to
[
6]
and aromatic primary, secondary and tertiary amide that observed with Rh/Mo catalysts. Although used
[
4]
substrates, including in particular, close derivatives of as a standard substrate by Fuchikami et al., it is im-
Adv. Synth. Catal. 2010, 352, 869 – 883
ꢂ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
873

Graham Beamson et al.
FULL PAPERS
[a]
Table 3. Products of amide hydrogenation (Mo:Ru=0.50 catalyst).
Entry
Substrate
Conversion [%]
Products
Selectivity [%]
83/16
1
100
CyCH NH /CyCH OH
2
2
2
2
3
100
100
n-butylamine/no (C H ) NH/n-butanol
77/22
4
9 2
2,2-dimethylpropanamine/2,2-dimethylpropan-1-ol
40/60
8/85
[
b]
4
5
6
7
8
100
5
CyCH NHCH /CyCONHCH
2
3
3
CyCH NHCH (traces)
2
3
[b]
100
30
CyCH N
A
H
U
G
R
N
U
G
A
T
N
T
E
N
N
30/5/65
73/27
2
3
2
2
CyCH N
A
H
U
G
R
N
U
G
2
3
2
2
100
C H N ACHTUNGTRNNEUG( C H ) (major product)
3 7 2 5 2
9
100
N-ethylpiperidine/piperidine
90/10
[
[
a]
b]
Reaction conditions: 100 bar H , 1608C, 16 h.
NB: 100% aromatic ring reduction, but considerable unreacted saturated amide.
2
portant to recognise that NAP (and N-acetylpyrroli- weaker CÀO bonds, thus assisting addition of H in a
2
dine), whilst providing convenient model amides in manner that is less accessible to more typical tertiary
terms of availability and limited number of potential amides.
reduction products, are not representative of typical
A possible explanation for the observed order of
tertiary amide structures because the amide N-atoms reactivity primary>tertiary@secondary amide is that
are integral components of six (and five)-membered the greater degree of steric hindrance presented by
[6]
rings. Amide reduction may be facilitated by the for- N-methyl substituents acts to hinder adsorption. In
mation of ring-stabilised zwitterions (cf. Scheme 2), simplistic terms tertiary amides would be expected to
with the charge separation between O and N atoms present even greater resistance to reduction but an
resulting in stronger CÀN bonds, and correspondingly additional important factor for consideration is the re-
spective orientation of each amide on the catalyst sur-
face. An alternative explanation for the apparent re-
sistance of secondary amides to reduction concerns
stabilisation as a consequence of their propensity to
form exceptionally strongly hydrogen-bonded dimers
[11]
in solution, as exemplified in peptide chemistry, al-
Scheme 2. Neutral and zwitterionic forms of N-acetylpiperi- though this does perhaps seem unlikely to be domi-
dine.
nant at the temperatures used during catalysis. The
874
asc.wiley-vch.de
ꢂ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2010, 352, 869 – 883

Selective Hydrogenation of Amides using Ruthenium/Molybdenum Catalysts
above ordering of reactivity parallels the trend noted CyCH OACHTUNGTRENNUNG
2
[
12]
ods with, for example, LiAlH4. Reports of the use
Reference to the proposed reaction pathways for
of the less aggressive reducing agent NaBH towards amide reduction (Scheme 1) suggests that CyCH OH
4
2
[
13,14]
reduction of amides at high temperatures
are also formation may occur via competitive hydrogenolysis
instructive. Here, secondary amides are unreactive, of the intermediate aminocarbinol with the loss of
tertiary amides give the corresponding amine in mod- ammonia (hence the reasoning for the conventional
erate yield, whereas primary amides undergo initial addition of NH to suppress secondary amine forma-
3
dehydration to the nitrile followed by slow reduction tion). Since water is released during conversion of
to the amine, a point that may be of significance in aminocarbinol to the postulated imine, accumulation
mechanistic terms. Ru catalysts are well known to be of additional water would presumably serve to inhibit
effective for selective reduction of nitriles to primary this step, leading to a higher standing aminocarbinol
[9]
amines, and such a demonstration of an amide dehy- concentration, thereby favouring CyCH OH forma-
2
dration step to the nitrile could conceivably provide tion. Although it appears rather unlikely at very low
an initial mechanistic pathway of lower energy that is water levels, an alternative potential competing reac-
accessible only to primary amides. The operation of tion pathway that requires consideration is initial,
[
12]
such a pathway might also explain the switch in prod- rather demanding hydrolysis of CyCONH to cyclo-
2
uct distribution with 2,2-dimethylpropionamide hexanecarboxylic acid, followed by reduction. Using
(
Table 3, entry 3) where dehydration to the nitrile the optimum Mo:Ru catalyst composition (0.50) and
might be expected to be more difficult than with n- standard reaction conditions (100 bar H , 1608C,
2
butyramide.
16 h), CyCO H underwent 85% conversion (cf. com-
2
plete reduction of CyCONH , Table 1) to CyCH OH
2
2
and methylcyclohexane (in 75 and 18% selectivity, re-
spectively), with additional traces of coupled prod-
ucts. This demonstrates not only the viability of
Controlled Addition of Water
Since water is a primary product of amide hydrogena- Ru/Mo catalysts in the second step of a potential al-
tion [Eq. (1)], and at the end of these reactions ternative pathway from CyCONH to CyCH OH, but
2
2
should therefore be present in equivalent amounts to also an indication of their applicability towards reduc-
the amine(s) formed, it was appropriate to determine tion of carboxylic acids in general.
whether the presence of additional water at the outset
would be beneficial, or otherwise, to reaction rate and
selectivity. It is also relevant to note that small Catalyst Genesis using HP-FT-IR Spectroscopy
amounts of water are commonly recognised to act as
promoters of many chemical processes, e.g., nitrile hy- In situ studies of the complete Ru/Mo precursor cata-
drogenation over unsupported Co catalysts, where the lyst system (Figure 1), together with control experi-
addition of H O leads to enhanced yields of primary ments using Ru and Mo carbonyls, e.g., Ru (CO) in
2
3
12
[9]
amine. Table 4, entries 2 and 3, show the effects of DME alone, have enabled a full description of the
the addition of 1 and 5 equivalents of water (relative molecular transformations involved during catalyst
to amide) to a Mo:Ru=0.50 catalyst at the start of evolution.
the reaction, clearly highlighting the incremental re-
For Ru alone, spectroscopic changes occurring
duction in selectivity to primary amine in favour of during slow incremental heating over the range 25–
CyCH OH. In a further experiment, Table 4, entry 4, 1608C (Figure 5) are consistent with the initial trans-
2
in which D O was used for putative labelling studies, formation of Ru (CO) into H Ru (CO) in accord-
2
3
12
4
4
12
15]
[
the addition of 30 equivalents of D O resulted not ance with literature precedent [Eq. (2)].
2
only in a drastic reduction in conversion but also a
complete switch of selectivity in favour of
Table 4. CyCONH hydrogenation: conversion and product selectivity vs. incremental addition of water (Mo:Ru=0.50 cata-
2
lyst).
Entry
H O/D O
Conversion [%]
Product selectivity [%]
2
2
CyCH NH
CyCH OH
2
2
2
1
2
3
4
none
100
100
100
30
85
82
77
26
14
16
22
74
H O (1 equiv.)
2
H O (5 equiv.)
2
D O (30 equiv.)
2
Adv. Synth. Catal. 2010, 352, 869 – 883
ꢂ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
875

Graham Beamson et al.
FULL PAPERS
stages, the anion appears to catalyse the rapid decom-
position of Mo(CO)6 at 1608C (Figure 1), the
À1
1
986 cm absorption decreasing in intensity and dis-
appearing completely after ca. 30 min. This behaviour
is in marked contrast to the extended stability (in
excess of 16 h) of Mo(CO) in DME alone under the
6
[
6]
same reaction conditions. It appears therefore that
Mo(CO) may provide a convenient source of CO
6
that is scavenged by Ru to sustain the stability of
À
[
H Ru (CO) ] during the catalyst induction period.
3
4
12
A precedent for this suggestion is available from the
demonstration of a similar, but reversible, reaction in
which ionic Ru and neutral Ir carbonyl intermediates
exchange CO in the proposed mechanism for Ir-cata-
lysed methanol carbonylation in the presence of Ru
[
18]
promoters. It is also relevant to note that the bond
enthalpy of the MoÀC bond in Mo(CO)6
À1 [19]
(
(
151 kJmol ) is lower than that of the RuÀC bond
À1 [20]
164 kJmol ) in Ru (CO) .
3
12
Once the available CO from Mo(CO) has been
6
À
consumed, [H Ru (CO) ] itself slowly degrades,
3
4
12
À1
Figure 5. In situ HP-FT-IR spectra (2150–1950 cm ) show-
total loss of n(CO) absorptions coinciding with the
onset of catalytic activity. Although no supporting
spectroscopic evidence has been obtained for the
presence of mixed Ru-Mo carbonyl species in solution
ing the transformations of Ru (CO) in DME at 100 bar H2
3
12
and 25–1608C.
(
it is significant to note that no equivalent binary car-
Further reaction occurs at 1008C with the forma- bonyls have been reported in the literature, and the
À[16]
12
tion of [H Ru (CO) ]
an anion that has been re- few examples of mixed Ru-Mo complexes that are
known contain either cyclopentadienyl or tertiary
3
4
ported to be a deprotonation product of H Ru (CO)
and, because of the absence of phosphine ligands), the accelerated decomposition of
Mo(CO)6 [with intense n(CO) at 1986 cm , cf. Mo(CO) in the presence of Ru carbonyl seems indi-
4
4
12
[17]
in methanol
À1
6
Figure 1], the additional characteristic absorption of cative of an intimate association between the two ele-
À1
this anion at 1978 cm is also clearly visible (for band mental components during controlled catalyst genesis.
frequencies see Supporting Information Table S1).
In addition to the Ru and Mo carbonyl compo-
The driving force for transformation from neutral to nents, the nature of the amide substrates themselves
anionic Ru seems likely to comprise reaction with appears to exert an influence on the catalyst activa-
water [Eq. (3)] formed by methanation of the CO re- tion process, as variations in the length of induction
period with different amides bear witness. In contrast
to the ca. 2 h required by CyCONH , induction peri-
2
ods of 6 and greater than 8 h, respectively, were evi-
dent with the secondary amides CyCONHCH and
3
leased according to Eq. (2). Evidence in support of HCONHCH ; throughout which the presence of
3
À
water formation has been provided by the appearance [H Ru (CO) ] was retained. There is ample prece-
3
4
12
and growth of n
A
C
H
T
U
N
G
T
R
E
N
N
U
N
G
(OÀH) absorption bands at 3590 and dence for the formation of amide-substituted metal
À1
[21]
3518 cm .
carbonyls, particularly with Mo, although the situa-
Traces of H Ru (CO) are identifiable up to tion with respect to Ru is less clear. Both the stoichi-
4
4
12
1408C, a further increase in temperature to 1608C re- ometry and stability of such products are likely to be
sulting in complete disappearance of all n(CO) ab- dictated by the nature of the individual amide, which
sorptions. can therefore reasonably be expected to lead to varia-
In the full Ru/Mo catalyst system (Figure 1) the tions in the degree of inhibition towards complete de-
spectrum initially recorded after rapid heating to composition. Hence the most likely explanation for
1608C (at 0 min) corresponds to a mixture of the variation in length of the induction periods is an
H Ru (CO) and Mo(CO) . The intense absorption amide-dependent stabilisation of Mo(CO) (amide)
AHCTUNGTRENNUNG
4
4
12
À1
6
6Àx
x
[21]
at 1986 cm due to the latter disappears after 30 min, (x=1–3), which in turn limits the rate of total de-
À
leaving [H Ru (CO) ] in solution, which survives for composition at 1608C and allows the controlled as-
3
4
12
over 100 min (i.e., approaching the length of the ob- sembly of the respective Ru/Mo catalysts.
served induction period). However, in the initial
876
asc.wiley-vch.de
ꢂ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2010, 352, 869 – 883

Selective Hydrogenation of Amides using Ruthenium/Molybdenum Catalysts
Ex situ Catalyst Characterisation
3 h results in a considerably sharper XRD profile
Figure 6) with parameters (Table 5, entry 5) that cor-
XRD profiles and parameters, including mean crystal- respond closely with the reference values for bulk
(
[
22]
lite sizes (estimated using the Scherrer equation ap- metallic Ru, thus reflecting the extensive sintering
plied to the most intense (101) Ru Bragg reflection that occurs during treatment at considerably higher
at 2q=ca. 51.58), of materials containing a range of temperatures than the maximum of 1608C used for
Mo:Ru values are shown in Figure 6 and Table 5, re- catalytic amide reduction. Comments on additional
spectively. Dominant features of the profiles are con- minor features of XRD data are included in the Sup-
sistent with the diffraction pattern expected for hex- porting Information, S2.
agonal Ru with no evidence of any discrete crystalline
The optimum Ru/Mo (Mo:Ru=0.50) amide reduc-
Mo-containing phase(s). Broad profiles are observed tion catalyst and reference material Ru/MoO [ex-
3
for the bimetallic Ru/Mo samples, indicative of either Ru (CO) ] (Mo:Ru=0.53), were examined by EDX-
3
12
small particle sizes and/or low crystallinity. For that STEM microscopy. Nine different areas of the former,
with the lowest nominal Mo content, Mo:Ru=0.52, all of which exhibited closely similar features, were
representative of the most active and selective cata- mapped by STEM and analysed. Figure 7 illustrates,
lyst for CyCONH reduction (Table 5, entry 2), the in high magnification, the bright field image of one
2
profile is particularly broad, equating to a mean Ru typical area, examination of which reveals several ag-
crystallite size of ca. 2.6 nm. The catalytically inactive glomerates in the size range 30–50 nm; localised
materials, with successively higher Mo:Ru ratios (en- within these agglomerates, smaller areas (ca. 2.5–
tries 3 and 4), contain larger (6–7 nm) aggregates 4 nm, cf. XRD data, Table 5) of higher contrast are
which are nevertheless still smaller than the value of apparent. Individual Mo and Ru L maps are shown in
8.2 nm obtained from the reference Ru sample pre-
pared from the reaction of Ru (CO)12 alone (entry 1).
3
Calcination of the Mo:Ru=0.52 catalyst at 3008C for
Figure 6. XRD profiles of Ru/Mo catalysts. 8 Parameters for
metallic Ru.
Figure 7. EDX-STEM analysis: bright field image of typical
area of Mo:Ru=0.50 catalyst.
[22]
Table 5. XRD data: 2q values, d-spacings and mean particle sizes of Ru:Mo catalysts.
Entry
Sample
Ru (101) 2q value [8]
d-spacing [ꢃ]
Mean crystallite size [nm]
1
2
3
4
5
6
Ru
51.56
51.36
51.36
51.26
51.65
51.62
2.058
2.066
2.066
2.069
2.055
2.056
8.2
2.6
6.4
6.9
26.5
–
Mo:Ru=0.52
Mo:Ru=0.99
Mo:Ru=1.36
Mo:Ru=0.52 calcined 3008C
[22]
Ru reference
Adv. Synth. Catal. 2010, 352, 869 – 883
ꢂ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
877

Graham Beamson et al.
FULL PAPERS
Figure 8. EDX-STEM analysis of Mo:Ru=0.50 catalyst: (a) Ru L(2.559 keV) and (b) Mo L(2.293 keV) maps, (c–e) Ru, Mo
quantification, areas of analysis depicted in light green.
Figure 8 (a) and (b), their comparison showing clearly tentatively associated with some substitution by Mo.
[
23]
that wherever Ru is detected Mo is also present, thus Ruban et al. have reported surface segregation en-
confirming a close association between Ru and Mo; ergies for binary combinations of the transition
Mo alone is also evident, in lower concentrations, dis- metals. The Ru lattice is considered to be a poor host
tributed across other areas of the sample. Quantifica- for most second row metals where either ꢀstrong, or
tion of the constituent elements, selected by particle very strong, segregationꢁ predominates. However, an
density, over areas of the micrograph highlighted in exception is noted when Mo comprises the solute and
green, Figure 8 (c)–(e), show a progressive increase in here ꢀmoderate antisegregationꢁ is said to occur, with
Mo content, from 37% in the small areas of highest positive surface segregation energies Eseg =0.05–
contrast (c) to ca. 50% over the larger, broader areas 0.3 eV. Calcination and sintering of the Mo:Ru=0.52
of lower contrast (d), (e). These materials of ca. 2.4– sample at 3008C presumably results in expulsion of
4
nm diameter appear extremely well intermixed and any postulated interstitial Mo from the Ru lattice (cf.
homogeneous in composition. In complete contrast, Table 5, entry 5).
Figure 9 reveals that the reference Ru/MoO3 [ex-
XPS data were recorded for three Ru/Mo catalysts
Ru (CO) ] sample, which exhibited typical behaviour [Mo:Ru=0.19, 0.57 (recycled) and 1.51], and the ref-
3
12
of a standard supported Ru catalyst, e.g., Ru/C erence material, Ru/MoO [ex-Ru (CO) ] (Mo:Ru=
3
3
12
(
Table 1, entries 2 and 5) towards CyCONH reduc- 0.53). All exhibited a dominant Ru 3d5/2 line at ca.
2
+
tion, contains discrete Ru particles of mean diameter 280.1 eV (which sharpened slightly on Ar sputter
nm (cf. Table 5, entry 1) distributed around the etching), consistent with the metallic Ru(0) state. Mo
edges of platelets of MoO , with very clear segrega- 3d spectra were considerably more complex, with sev-
8
3
tion between Ru and Mo components.
eral different Mo environments evident in all exam-
Close inspection of the XRD parameters of three ples examined. A particularly clear profile for the cat-
samples with successively higher nominal Mo to Ru alytically active and selective Mo:Ru=0.19 composi-
ratios (Table 5, entries 2–4) reveals a consistent slight tion (cf. Figure 3) is shown in Figure 10.
increase in d-spacing relative to metallic Ru (entry 5).
Use of curve fitting techniques reveals the presence
The origin of this hcp Ru lattice expansion may be of four distinct components with 3d5/2 lines centred at
878
asc.wiley-vch.de
ꢂ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2010, 352, 869 – 883

Selective Hydrogenation of Amides using Ruthenium/Molybdenum Catalysts
Figure 9. EDX-STEM analysis: (a) bright field image, (b) Ru (L), (c) O (K) and (d) Mo (L), maps of Ru/MoO [ex-
3
Ru (CO) ], Mo:Ru=0.53.
3
12
2
32.4, 231.3, 228.7 and 228.0 eV, and which, by com-
[
6]
parison with known standards may be assigned to
Mo(VI), Mo(V), Mo(IV) and, most notably zerova-
lent Mo environments, respectively. Significant, but
different, amounts of each of these states were detect-
ed in all the Ru/Mo catalysts derived from both metal
carbonyls. In contrast, XPS spectra of the Ru/MoO3
[
ex-Ru (CO) ] sample revealed a much simpler Mo
3 12
3d5/2 profile (Figure 11), with Mo(VI) (62.2%) and
Mo(IV) (29.3%) as major components, a minor
amount of Mo(V) (8.5%), and most significantly, no
zerovalent Mo.
In the as-prepared state the Mo(VI), Mo(V),
Mo(IV) and Mo(0) distribution for the Mo:Ru=0.19
catalyst comprises 52, 24, 11, and 13 at%, respectively
(
Table 6), reference to which shows that the surface
concentration of Mo decreases slightly, with a con-
+
comitant increase in Ru, following Ar sputter etch- Figure 10. XPS data: curve fitting of Mo 3d region of
ing. The concentration of Mo(VI), as a fraction of Mo:Ru=0.19 catalyst.
Adv. Synth. Catal. 2010, 352, 869 – 883
ꢂ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
879

Graham Beamson et al.
FULL PAPERS
nificantly higher temperatures are generally required
for the gas phase hydrogen reduction of MoO , e.g.,
3
the formation of Mo O and Mo O as intermediates
4
11
2
5
[24]
in the reduction of MoO to MoO at 3008C ); (b)
3
2
partial reaction of Mo(CO) in DME under the stan-
6
dard reaction conditions yielded MoO (Mo content
3
6
6.7 wt% by ICP analysis) as the only detectable
[
6]
product, and (c) addition of Mo powder (which is
itself inactive for amide hydrogenation) to
Ru (CO) precursor in a control experiment (cf.
a
3
12
Table 1, entry 6) simply maintains catalytic perfor-
mance that is typical of Ru alone, indicating that bulk
Mo(0) is not responsible for the synergy. Thus the
presence of Ru appears crucial to the stabilisation of
Mo(0), introduced as Mo(CO) , in the active and se-
6
lective catalysts for amide reduction, most probably
Figure 11. XPS data: curve fitting of Mo 3d region of Ru/ as a consequence of reactions between the organome-
MoO [ex-Ru (CO) ], Mo:Ru=0.53.
3
3
12
tallic precursors during catalyst genesis.
+
total Mo, remains essentially unchanged, but Ar in-
duced reduction of Mo(V) to Mo(IV), together with Catalyst Model
an increase in Mo(0) surface concentration after 8 min
sputter etching, is clearly evident. Similar extended Addition of Mo(CO) to the Ru (CO) catalyst pre-
6
3
12
etching (8, 12 min) of MoO alone also led to the de- cursor at Mo:Ru compositions of ꢁ1 clearly has a
3
[
6]
tection of Mo(IV), but without evidence of Mo(V).
profound effect on amide conversion and product se-
With increasing Mo content (Mo:Ru=0.57 and 1.51) lectivity (cf. Table 1, entries 1 and 3), and on the ulti-
progressively greater surface concentrations of MoO3 mate Ru particle sizes of the active materials (cf.
are evident, as might be expected. Highest concentra- Table 5, entries 1 and 2). A combination of HP-FT-IR
tions of zerovalent Mo (also present after recycling), results during catalyst genesis and ex situ catalyst
were detected on the most active amide hydrogenation characterisation data, for the most active and selec-
catalysts of nominally low Mo content (Mo:Ru=0.19, tive catalysts, seem best interpreted in terms of 2–
0.57), a possible implication being that Ru is more ef- 4 nm aggregates containing Ru(0) intimately associat-
fective in the retention of reduced Mo oxidation states ed with Mo in oxidation states ranging between (0),
when initial concentrations of Mo are low. Finally, it is (IV), (V) and (VI). Substitution of minor amounts of
relevant to note that neither XRD nor XPS have pro- Mo(0) in the Ru lattice would be consistent with the
vided any evidence in support of RuC formation.
slight lattice expansion evident from XRD, the detec-
tion of Mo(0) by XPS and the homogeneous distribu-
tion of Ru and Mo as revealed by EDX-STEM. The
chemical states of Mo would thus range from close to
metallic in areas intimately associated with the sur-
face of Ru, to increasingly higher oxidation states, fi-
Results of Control Experiments using Mo(CO)6
Alone
Further evidence in support of the role of Ru in re- nally Mo(VI), towards the exterior surfaces of the ag-
tention of reduced oxidation states of Mo, particularly gregates. From XPS data, at and above the threshold
Mo(0), has been provided by the results of the follow- Mo:Ru composition of 1, the assemblies contain pro-
ing control experiments: (a) no detectable reaction gressively higher surface concentrations of Mo(VI),
occurred during treatment of DME suspensions of e.g., for Mo:Ru=1.51, Mo(VI):Mo(IV)=97:3%,
MoO at 100 bar H and 1608C for 16 h (normally sig- which presumably serve to block active sites, with
3
2
+
Table 6. XPS data: quantification of Ru and Mo, and relative composition of Mo oxidation states vs. Ar sputter etching
[a]
time (Mo:Ru=0.19 catalyst).
Sputter etching time (3 keV)
Composition [at%]
Ru
Mo total
Mo(VI)
Mo(V)
Mo(IV)
Mo(0)
none
87.7
89.2
90.5
12.3
10.8
9.5
6.4 (51.6)
5.6 (52.8)
4.7 (51.1)
3.0 (24.2)
2.0 (18.9)
0.4 (4.3)
1.4 (11.3)
1.7 (16.0)
2.3 (25.9)
1.6 (12.9)
1.3 (12.3)
1.8 (19.6)
1
8
min
min
[
a]
Atomic% distribution of Mo oxidation states in parentheses.
880
asc.wiley-vch.de
ꢂ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2010, 352, 869 – 883

Selective Hydrogenation of Amides using Ruthenium/Molybdenum Catalysts
consequent inhibition of all amide reduction activity. tions, no secondary or tertiary amine formation, ap-
Finally, under the strongly reducing reaction condi- pears unique in the history of amide hydrogenation.
tions required for amide hydrogenation, the Ru/Mo Consequently, the traditional requirement for the de-
[
2,25]
aggregates may also conceivably exist in a form of dy- liberate addition of excess ammonia
to suppress
namic equilibrium, with reversible formation of re- the postulated imine-amine coupling reactions respon-
duced oxidation states of Mo.
sible for by-product formation (cf. Scheme 1) is ren-
dered unnecessary. It is however recognised that
during the formation of CyCH OH, the only signifi-
2
Comparison with Rh/Mo Catalysts
cant by-product detected (in ca. 14% selectivity) over
these Ru/Mo catalysts under optimum reaction condi-
There are both similarities and significant differences tions, one equivalent of NH is liberated per equiva-
3
between the Ru/Mo catalysts described here and the lent of CyCONH consumed via this pathway, leading
2
Rh/Mo systems recently reported for CyCONH re- to the possibility that this in situ generation of ammo-
2
[6]
duction. Similarities include (i) the heterogeneous nia may be sufficient to inhibit (CyCH ) NH forma-
2
2
nature of the catalysts, (ii) reduction of CyCONH to tion. Additional factors are also likely to be impor-
2
CyCH NH in good selectivity, (iii) no requirement tant, including, e.g., (i), reduced reaction tempera-
2
2
for addition of ammonia or amines to suppress secon- tures in relation to those required by copper chro-
dary amine formation, although minor amounts of mite-based catalysts, and (ii), the small (2–4 nm) sizes
secondary amines were always evident with Rh/Mo, of the well dispersed nanocomposite aggregates that
(
iv) critical dependence of catalyst performance on may facilitate rapid, solvent-assisted, desorption of
Mo:Rh composition, (v) reduction under milder reac- primary amine into the liquid phase immediately after
tion conditions than those commonly required by, formation on the catalyst surface. A further point of
e.g., copper chromite, (vi) a significant decrease in significance is a potential role for reduced oxidation
amide conversions below a limiting temperature and states of Mo during amide reduction, as revealed by a
pressure of 1308C and 100 bar H , respectively (cf. consideration of reports of the MoO -catalysed iso-
2
2
[
26]
1
458C and 20 bar H with Ru/Mo), and (vii) catalysts merisation of alkanes. Here the existence of MoO3
2
that comprise intimate mixtures of metallic Rh and and Mo O on the surface of MoO (cf. Figure 10)
2
5
2
amorphous Mo (with excess Mo present in the form was demonstrated and considered to be the origin of
of MoO ). Differences include (i) the much shorter in- both metallic and acidic functions that were responsi-
3
duction period required for catalyst genesis (450 min ble for unusual activity associated with this ꢀbifunc-
vs. 130 min) with Ru/Mo for optimum catalyst compo- tionalꢁ material. Analogous behaviour could equally
sitions, which is presumably a reflection of the rela- well be of relevance to amide reduction, with the
tive (thermal) stabilities of the penultimate Ru and acidic function responsible for enhancement of strong
Rh molecular precursors to the heterogeneous cata- initial adsorption of the amide carbonyl group, lead-
lysts, (ii) generally slightly lower primary amine selec- ing to synergistic hydrogen reduction at the Ru/Mo
tivities and greater variation in product selectivities interface, with a consequent enhancement of overall
throughout, with (iii) secondary amine formation in conversion, as observed experimentally. An additional
particular observed to increase significantly at low consequence of the acidic function may be protona-
pressures, (iv) good conversions maintained up to a tion of CyCH NH immediately on formation, leading
2
2
limiting composition threshold of Mo:Rh=2, above to suppression of secondary reactions.
which no primary amine was formed, with a complete
switch of product distribution in favour of
(
CyCH ) NH and CyCH OH, at much lower overall Conclusions
2
2
2
conversions, and neither (v) complete catalyst deacti-
vation evident at least as far as Mo:Rh=4.5, nor (vi) A family of recyclable heterogeneous Ru/Mo cata-
evidence of the presence of Mo(0) (XPS) or Rh lat- lysts, derived from zerovalent metal carbonyls
tice expansion (XRD). Overall the behaviour of the Ru (CO) and Mo(CO) , have proved effective for
3
12
6
Ru/Mo catalysts appears ꢀcleanerꢁ and more readily the selective hydrogenation of a range of amides to
rationalised than that of their Rh/Mo counterparts.
the corresponding amines under significantly milder
reaction conditions than those reported previously for
these difficult transformations. They are particularly
useful for the highly selective reduction of primary
amides such as CyCONH to CyCH NH , without the
Origin of High Selectivity towards Primary Amide
Reduction
2
2
2
need for the co-addition of excess ammonia and/or
The highly selective behaviour of these Ru/Mo cata- amines to inhibit side reactions leading to secondary
lysts towards reduction of a primary amide such as product formation, hitherto a universal feature of
CyCONH2 with, under optimum catalyst composi- both amide and nitrile functional group reduction.
Adv. Synth. Catal. 2010, 352, 869 – 883
ꢂ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
881

Graham Beamson et al.
FULL PAPERS
These catalysts have been shown to be effective for Experimental Section
amide reduction at pressures as low as 20 bar H , but
2
the current limiting effective operational minimum With the exception of the additional details provided below,
experimental and analytical procedures are as outlined in
ref.
temperature of 1458C represents a further obstacle to
[6]
be surmounted if they are to become attractive for the
production of intermediates in the manufacture of fine
chemicals and pharmaceuticals. A key factor associated
with the performance of these Ru/Mo catalysts is clear-
ly the incorporation of a formally zerovalent Mo pre-
Safety warning. Experiments involving pressurised gasses
can clearly be hazardous and must be conducted with suita-
ble equipment following appropriate safety procedures.
cursor, namely Mo(CO) . This molecule, together with Reagents
6
CO ligand-substituted derivatives, e.g., Mo(CO)6 Ln
Àn
Molybdenum hexacarbonyl, Mo(CO) , was supplied by Alfa
6
(
n=1–3; L=N, O, P donor ligands, and unsaturated
Organics, RuCl ·xH O (99.9%) by Strem, Mo powder
3
2
hydrocarbons such as arenes, dienes and trienes, e.g.,
(
99.8%) by Newmet, 5 wt% Ru on carbon by Aldrich
bicyclohepta-2,5-diene and 1,3,5-cycloheptatriene), Chemicals, and Ru
AHCTUNGTRENNUNG
(acac) by Johnson Matthey. Trirutheni-
3
comprise a group of organometallic Mo(0) complexes, um dodecacarbonyl, Ru (CO) , was prepared from RuCl
3
12
3
[27]
of wide ranging stability with respect to degradation, hydrate by the method of Johnson et al. Except where in-
dicated otherwise the amide substrates, potential amine
available for catalyst tuning purposes. Application of
products and GC standards (98–99% purity) were obtained
from Aldrich Chemicals. N-Ethylpiperidine (99%), piperi-
dine (99%), methylcyclohexane (99%) and cyclohexanecar-
boxylic acid (97%) were obtained from Lancaster, and ben-
high throughput methods, using such ligand-substituted
Mo(0) derivatives as precursors, may thus provide op-
portunities for the discovery of catalysts that are effec-
tive at lower reaction temperatures, ideally ca. 708C.
zylamine (99%) from Acros Organics.
Although an Mo(0) catalyst precursor appears essen-
tial, this does not necessarily apply to ruthenium, as
N,N-Dimethylcyclohexanecarboxamide,
the Ru
to be an acceptable precursor (cf. Table 1, entries 3
and 4), thus allowing scope for the use of more readily White crystals of the title compound were prepared by hy-
available Ru sources than Ru (CO) .
drogenation of N,N-dimethylbenzamide over Pd/C
0.5 mol%) at 1608C for 16 h; yield: 87%; mp 84–878C;
anal. calcd. C H NO: C 69.63, H 11.04, N 9.02; found: C
ACHTUNGTRENNUNG( III) comples [Ru ACTHUNGTRENUNNG( acac) ] has also been shown
3
CyCON ACHTUGNTRNEUNG( CH )
3 2
3
12
(
Our previous paper reported preliminary attempts,
using silica as a ꢀneutralꢁ support, to prepare, charac-
terise and evaluate supported Rh/Mo amide reduction
9
17
1
6
8.82, H 10.95, N 8.48; H NMR (200 MHz, CDCl ): d=
3
[6]
1.15–1.75 (m, 10), 2.43 (tt, 1), 2.86 (s, 3), 2.98 (s, 3); CI-MS:
catalysts. Although these proved to be initially more
active than their unsupported counterparts towards
the reduction of N-acetylpiperidine, they performed
poorly at <1508C and progressively deactivated
during recycle, as a consequence of accumulation of
+
m/z=156 [M] .
Catalytic Procedures
A typical ꢀsingle-potꢁ Ru/Mo catalyst preparation, and eval-
C-, H-, and N-containing residues on the support. The uation in CyCONH reduction, was carried out as follows,
2
poor performance with respect to reduced tempera- using
a
nominal Mo:Ru atomic composition of 0.50.
tures using fresh catalysts was attributed to tempera- [Ru
(CO)12] (0.022 g, 0.103 mmol Ru), freshly sublimed
3
Mo(CO) (0.0135 g, 0.0511 mmol Mo) and cyclohexanecar-
ture-dependent limitations to substrate adsorption/
product desorption onto/from the catalyst surface (or
possibly, in light of this work, strong amide chemi-
sorption and even oligomerisation on the silica sup-
port). Equivalent Ru/Mo catalysts have not therefore
been examined. Nevertheless carbon, a standard sup-
port commonly used for metal-catalysed reductions of
6
boxamide (0.235 g, 1.85 mmol) contained in a glass liner
were dissolved in 1,2-dimethoxyethane (DME) (30 mL), af-
fording a yellow solution [see Supporting Information, Fig-
ure S1(a)]; n-octane (0.100 g) was added as internal stan-
dard for GC analysis. The liner was placed in a ca. 300 mL
capacity pressure vessel, origin of manufacture ICI, and the
reaction mixture, under agitation, purged 3 times with N2
organic molecules may, particularly in the high sur- (to 5 bar), and 3 times with H (5 bar). The autoclave was
2
face area form, actually be more appropriate than then pressurised to 100 bar H and heated to 1608C for 16 h.
2
silica, and certainly merits future consideration.
In the broader context, the approach described in
this and our previous paper may also lead to the dis-
covery and development of catalysts for the reduction
of other ꢀdifficultꢁ organic carbonyl-containing func-
tional groups, e.g., carboxylic acids, esters and anhy-
drides, the majority of which still rely on stoichiomet-
After cooling to room temperature, the autoclave was
vented and opened, revealing a very dark coloured liquid
[see Supporting Information Figure S1(b)] from which a
black residue slowly settled, leaving a colourless solution
[
6]
[
see Supporting Information, Figure S1(c)]. The residue was
separated by centrifugation (2000 rpm, 20 min) and the su-
pernatant liquid containing reaction products removed.
After several washings with DME (10 mL) the catalyst resi-
due (ca. 15 mg) was dried, and the resultant fine black
ric hydridic reagents, e.g., LiAlH and derivatives,
4
rather than intrinsically more efficient catalytic routes powder either recycled using the same quantities of fresh
using molecular hydrogen. substrate and solvent, or characterised using the techniques
882
asc.wiley-vch.de
ꢂ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2010, 352, 869 – 883

Selective Hydrogenation of Amides using Ruthenium/Molybdenum Catalysts
described below. Product solutions were analysed by GC as
previously described.
[6] G. Beamson, A. J. Papworth, C. Philipps, A. M. Smith,
[6]
R. Whyman, J. Catal. 2010, 103, 93–102.
7] P. Kluson, L. Cerveny, Appl. Catal. A: Gen. 1995, 128,
13–31.
8] J. P. Breen, R. Burch, K. Griffin, C. Hardacre, M.
Hayes, X. Huang, S. D. OꢁBrien, J. Catal. 2005, 236,
[
[
Preparation of Ru/MoO from Ru (CO) and MoO
3
3
3
12
This material was obtained using the procedure described
above, with [Ru (CO) ] (0.022 g) and a suspension of the
3
12
2
70–281.
requisite amount of MoO3 (0.0075 g) to afford a Mo:Ru
composition of ca. 0.5, in a DME (30 mL) solution contain-
[
9] J. Volf, J. Pa sˇ ek, in: Catalytic Hydrogenation, (Ed.: L.
Cerveny), Elsevier Science, Amsterdam, 1986, pp 105–
ing CyCONH (0.235 g).
2
1
45.
[
10] S. Nishimura, Handbook of Heterogeneous Catalytic
Hydrogenation for Organic Synthesis, John Wiley and
Sons, Inc., New York, 2001, and references cited there-
in.
Ex situ Characterisation
XRD: A monochromated Co K (l=1.7902 ꢃ) X-ray source
was used.
XPS: Representative examples of both freshly prepared,
and used, catalysts, together with a series of primary stand-
[11] H. Sigel, R. B. Martin, Chem. Rev. 1982, 82, 385–426.
[12] See, for example, J. Clayden, N. Greeves, S. Warren, P.
Wothers, Organic Chemistry, Oxford University Press,
2001.
ards (Mo foil, MoO , Ru pellet and Ru powder) were ana-
3
lysed using a Scienta ESCA 300 XPS spectrometer at
NCESS, Daresbury. The curve fitting procedure for the Mo
[13] Y. Kikugawa, S. Ikegami, S. I. Yamada, Chem. Pharm.
3
d spectra involved fixing the areas of the Mo 3d5/2 and 3d3/2
Bull. 1969, 17, 98–104.
lines at the theoretical ratio of 60:40, and using a DE value
of 3.2 eV in accordance with the known separation of the 3d
doublet in Mo and MoO3.
EDX-STEM: Elemental compositions of the metallic
components were measured using the most intense Ru and
Mo L 1a lines at 2.559 and 2.293 keV, respectively; Ru, Mo
and O K 1a lines at 19.279, 17.479 and 0.529 keV, respec-
[14] H.-J. Zhu, K.-T. Lu, G.-R. Sun, J.-B. He, H.-Q. Li, C. U.
Pittman, Jr, New J. Chem. 2003, 27, 409–413.
[15] F. Piacenti, M. Bianchi, P. Frediani, E. Benedetti,
Inorg. Chem. 1971, 10, 2759–2763.
[28,29]
[16] J. W. Koepke, J. R. Johnson, S. A. R. Knox, H. D.
Kaesz, J. Am. Chem. Soc. 1975, 97, 3947–3952.
[
[
[
[
[
[
[
[
[
[
17] H. W. Walker, C. T. Kresge, P. C. Ford, R. G. Pearson,
tively, were also used for Ru/MoO sample.
3
J. Am. Chem. Soc. 1979, 101, 7428–7429.
18] R. Whyman, A. P. Wright, J. A. Iggo, B. T. Heaton, J.
Chem. Soc. Dalton Trans. 2002, 771–777.
19] F. A. Cotton, A. K. Fischer, G. Wilkinson, J. Am.
Chem. Soc. 1959, 81, 800–803.
20] R. Koelliker, G. Bor, J. Organomet. Chem. 1991, 417,
Microanalysis: The term ꢀnominal compositionꢁ used in
the text refers to the relative quantities of Ru and Mo used
in the initial catalyst preparations. Unfortunately, despite
numerous attempts, it proved impossible to confirm the
actual compositions in the catalytically active materials by
ICP-AES analysis, principally because of the well known in-
4
39–451.
21] I. W. Stolz, G. R. Dobson, R. K. Sheline, Inorg. Chem.
963, 2, 322–326.
[30]
tractable nature of Ru towards acid digestion.
1
22] H. E. Swanson, R. K. Fuyat, G. M. Ugrinic, Natl. Bur.
Stand. (U.S.) 1955, Circ. 539, Vol. IV.
23] A. V. Ruban, H. L. Skriver, J. S. Norskov, Phys. Rev. B
Acknowledgements
1
999, 59, 15990–16000.
24] R. Burch, J. Chem. Soc. Faraday Trans. 1 1978, 74,
982–2990.
The authors acknowledge the assistance of Mr. S. Apter and
Mr. A. Mills for microanalyses and mass spectrometry, re-
spectively, Dr. J. Claridge and Dr. J. A. Iggo (Department of
Chemistry) and Mr. A. J. Pettman (Pfizer Ltd) for discus-
sions, EPSRC for financial support of NCESS, Daresbury,
under grant GR/S14252/01, and the Leverhulme Fine Chemi-
cals Forum for financial support (to CP and AMS).
2
25] A. A. N. Magro, G. R. Eastham, D. J. Cole-Hamilton,
Chem. Commun. 2007, 3154–3156.
26] A. Katrib, J. W. Sobczak, M. Krawczyk, L. Zommer, A.
Benadda, A. Jablonski, G. Maire, Surf. Interface Anal.
2
002, 34, 225–229.
[
27] C. R. Eady, P. F. Jackson, B. F. G. Johnson, J. Lewis,
M. C. Malatesta, J. Chem. Soc. Dalton Trans. 1980,
References
3
83–389.
[
28] C. D. Wagner, W. M. Riggs, L. E. Davis, J. F. Moulder,
G. E. Muilenberg, Handbook of X-ray Photoelectron
Spectroscopy, Perkin–Elmer Corporation, Physical Elec-
tronics Division, Eden Prairie, Minn. 55344, USA, 1979.
29] D. Briggs, M. P. Seah, Practical Surface Analysis, Vol. 1,
Auger and X-ray Photoelectron Spectroscopy, 2nd edn.,
Wiley, New York, 1990, and references cited therein.
30] M. Balcerzak, Crit. Rev. Anal. Chem. 2002, 32, 181–
[
[
[
[
[
1] P. N. Rylander, Hydrogenation Methods, Academic
Press, 1985, and references therein.
2] B. Wojcik, H. Adkins, J. Am. Chem. Soc. 1934, 56, 247;
2
419–2424.
3] D.-H. He, N. Wasaka, T. Fuchikami, Tetrahedron Lett.
995, 36, 1059–1062.
[
[
1
4] T. Fuchikami, C. Hirosawa, N. Wasaka, Tetrahedron
Lett. 1996, 37, 6749–6752.
5] A. Behr, V. A. Brehme, Adv. Synth. Catal. 2002, 344,
2
26.
5
25–532.
Adv. Synth. Catal. 2010, 352, 869 – 883
ꢂ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
883