G. Beamson et al. / Journal of Catalysis 269 (2010) 93–102
101
examined, could be incomplete decomposition of the metal car-
bonyl precursors during genesis of the active catalyst. However,
recycled catalysts that had been pre-formed at 160 °C also exhib-
ited a similar behaviour at low temperatures, and a more probable
explanation seems to be associated with temperature dependent
adsorption/desorption of reactants and products on the catalyst
surface. Longer residence times of more strongly adsorbed reaction
intermediates at 120 °C might also account for the observed switch
decreased from 2.17 through 1.10, 0.95, 0.67 to 0.62, respectively,
the last two values comparing favourably with the nominal value
of 0.67. This behaviour is consistent with higher Mo content in
the outer layer of the catalyst in the ‘as-prepared’ state; moreover
the respective surface concentrations of Mo(3d5/2) and O(1s) of
3
11.7 and 32.7 at% are close to the ratio expected for MoO . Spectra
of silica-supported Rh/Mo catalysts showed very similar features to
those obtained from the unsupported materials, but because of the
dilution factor the spectra exhibited much higher signal to noise
and were therefore of lower diagnostic value. XPS spectra of the
2
in product selectivity in favour of CyCH OH (Fig. 5), the preference
for which indicates that alcohol formation, presumably via CꢁN
bond hydrogenolysis, has a lower activation energy than CꢁO bond
hydrogenolysis (cf. Scheme 1).
3
reference Rh/MoO catalyst were considerably different (see Sup-
plementary material S3), consistent with the preparative method
used, but showed no indication of an intermediate Mo(V) state
(cf. Entry 12 with 13–15).
4.3. Catalyst stability
The unsupported catalysts are robust. They may be recovered
4.5. Nature of the working catalysts and role of Molybdenum
and recycled with only minor loss of activity, requiring no reactiva-
tion procedure other than the application of the standard reaction
conditions. Notwithstanding their stability, exposure to CO com-
pletely, but reversibly, inhibits all activity (see Supplementary
material S2), presumably via either/or both (i) stabilization of the
metal carbonyl precursors and prevention of active catalyst forma-
tion in the case of the ‘one-pot’ preparations, and (ii) preferential
adsorption of CO on active sites on the preformed catalysts.
Consideration has been given to the formation and possible role
of Mo carbides but neither XPS nor XRD has provided any support-
ing evidence (see Supplementary material S4). Similarly any role
for Mo hydrogen bronzes in this chemistry can be discounted as
a test Rh/H-Mo bronze catalyst proved inactive towards amide
hydrogenation (Supplementary material S5).
3
The formation of MoO in these Rh/Mo catalysts prepared from
zerovalent metal carbonyls under strongly reducing conditions re-
quires comment. Since the CO ligands in the starting metal carbon-
yls are not apparently involved in the formation of carbidic
material, it appears that the final stage in the genesis of these cat-
alysts must involve loss of carbonyl groups as CO, which undergo
methanation, during which the total amount of water generated
(6 equivalents per Mo atom, and 22/3 per Rh atom) is sufficient to
cause the oxidative hydrolysis of, in particular, Mo(0) to the higher
oxidation states that are evident from ex situ measurements. In
4
.4. Catalyst characterization
EDX-STEM results provide clear evidence that the Rh and Mo
components of these catalysts are intimately associated into parti-
cles within the size range 2–4 nm. In addition much larger agglom-
erates, too impervious to the electron beam to permit detailed
analysis, were detected in other areas of the specimen; the latter
3
contained Mo, presumably in the form MoO . XRD provided sup-
porting evidence for the particle size range 2–4 nm although Rh
was the only bulk crystalline phase that proved possible to charac-
support of this suggestion, partial reaction of Mo(CO)
6
alone in
DME at 160 °C and 100 bar H for 16 h resulted in the formation
2
terize. Reaction of Rh
led to the formation of much larger (ca. 10–11 nm) Rh particles
Table 3), the co-presence of Mo thus exerting a very significant
moderation in the overall aggregate size, also consistent with inti-
mate association between the two components.
6
(CO)16 alone under the standard conditions
of only Mo(VI) oxide, as confirmed by ICP analysis. Water forma-
tion during the initial stages of catalyst genesis is also likely to
+
(
act as a source of the stabilizing countercation [H
3
O] for the
2ꢁ
intermediate anion [Rh13
Scheme S1).
H (CO)24
3
]
(Supplementary material,
XPS measurements on a Mo:Rh = 0.67 catalyst, in both the ‘as-
prepared’ and recycled forms, provide confirmation that Rh was
in the zerovalent state throughout and that Mo appeared to be pre-
dominantly in the form of the trioxide (Mo 3d5/2, 233.2 eV). Never-
Ex- situ characterization evidence, taken together with the syn-
ergistic behaviour between Rh and Mo apparent during genesis
and in catalysis, is consistent with an intimate association between
Rh and Mo. Although the metallic state of Rh is not in doubt, the
nature of Mo, particularly under reaction conditions is less clear.
Although surface Mo is predominantly in the form Mo(VI), under-
lying layers containing reduced oxidation states such as Mo(V) and
Mo(IV) (possibly directly associated with Rh) are also present. Such
+
theless, evidence of reduced oxidation states, particularly after Ar
sputter etching, was provided by resolution of composite Mo dou-
+
blet patterns using curve fitting techniques (Fig. 7). After Ar treat-
ment for 2 min a Mo 3d5/2 line at 228.6 eV became apparent, and
this increased in intensity with further sputter etching (up to a to-
tal of 8 min), when the second (3d3/2) component of the expected
2
reduced oxides exhibit both acidic, and for MoO metallic, proper-
ties [21], both of which appear likely to be of significance to the
‘Mo effect’ in amide reduction. Each could quite reasonably consti-
tute active sites for initial strong adsorption of the amide carbonyl
group, which could in turn lead to the promotion of overall rates of
reduction at adjacent Rh centres. At high Mo:Rh levels excess
molybdena has the effect of coating the Rh/Mo ensemble, blocking
active sites, and thus leading to the observed reduction/poisoning
of catalytic activity, and possibly influencing reaction selectivity
via longer residence times.
In light of the above-mentioned evidence we suggest that the
catalytically active materials are 2–4 nm-sized composites con-
taining Rh cores encapsulated by surface layers containing oxi-
dized states of Mo, particularly Mo(IV) and Mo(V), and any Mo in
excess ‘to catalytic requirements’ in the form of the trioxide. Such
ensembles may exist in dynamic equilibrium under the strongly
reducing reaction conditions required for amide hydrogenation,
with the possibility of reversible formation of reduced oxidation
doublet for MoO
At this stage there was also an indication of a minor component
centred at ca. 231 eV, intermediate between MoO and MoO
2
at ca. 232.2 eV was also clearly resolved [20].
3
2
,
which has been tentatively attributed to a Mo(V)-containing envi-
ronment, recently identified by Katrib et al. [21]. A concomitant de-
+
crease in O (1s) concentration as a function of Ar sputter etching
3
time was also evident. Analogous reference spectra of MoO alone
+
revealed only minor surface reduction even after extended Ar
treatment (Table 3, Entries 6ꢁ8).
Quantification of XPS profiles of the fresh Mo:Rh = 0.67 catalyst,
before and after Ar sputter etching at various time intervals (1, 2, 4
+
and 8 min), has enabled changes in atomic concentration profiles to
be determined. In the ‘as-prepared’ state, the surface concentration
of Mo was much higher (Mo:Rh = 2.17) than that expected from the
nominal precursor ratio used in catalyst preparation. However, dur-
ing Ar+ sputter etching the surface Mo:Rh composition gradually