Supported nickel catalysts: Preparation and characterisation of alumina-, molybdena-, and silica-supported nickel, and the identification of reactive oxygen on these catalysts by exchange with isotopically labelled carbon dioxide

Article abstract of DOI:10.1039/a809293a

Nickel catalysts, supported on alumina, silica, and molybdena, have been prepared by impregnation and co-crystallization. In the precursor state the catalysts were characterised by UV-visible spectroscopy, thermogravimetric analysis/differential thermal analysis (TGA/DTA), and X-ray photoelectron spectroscopy (XPS). The nickel was principally in the 2+ oxidation state with an octahedral coordination. However, the ligand sphere surrounding the nickel ion was sensitive to the support, indicating that the species on the different supports were not identical thus suggesting a metal complex-support interaction. Reduction was followed by temperature programmed reduction (TPR) and TGA, the results of which indicated that reduction and decomposition of nickel nitrate occurred simultaneously. X-ray diffraction (XRD) analysis revealed that, with the Ni/MoO₃ sample, no hydrogen bronze was formed on reduction. The reduced catalysts were characterised by carbon monoxide chemisorption, carbon dioxide chemisorption, and by reaction of buta-1,3-diene with dihydrogen. In the absence of a dihydrogen stream it was found that the catalysts adsorbed no carbon monoxide due the presence of sub-monolayer quantities of surface oxygen. The extent of the oxygen was quantified by exchange with isotopically labelled carbon dioxide. Differences in the electronic nature of the nickel between the Ni/MoO₃ sample and the other catalysts were revealed by their behaviour towards buta-1,3-diene hydrogenation.

Full text of DOI:10.1039/a809293a

View Article Online / Journal Homepage / Table of Contents for this issue
Supported nickel catalysts: Preparation and characterisation of
alumina-, molybdena-, and silica-supported nickel, and the
identiÐcation of reactive oxygen on these catalysts by exchange with
isotopically labelled carbon dioxide
S. David Jackson,*a Janice Willis,a Gordon J. Kelly,b¤ Gavin D. McLellan,b Geo† Webb,*b
Sue Mather,c Richard B. Moyes,c Sydney Simpson,c Peter B. Wells*c and Robin Whyman*d”
a Synetix, Research, T echnology and Engineering Group, PO Box 1, Belasis Ave., Billingham,
Cleveland, UK T S23 1L B
b Department of Chemistry, University of Glasgow, Glasgow, UK G12 8QQ
c Department of Chemistry, University of Hull, Hull, UK HU6 7RX
d ICI Chemicals and Polymers L td, Research and T echnology Dept, PO Box 8, T he Heath,
Runcorn, Cheshire, UK W A7 4QE
Received 27th November 1998, Accepted 25th March 1999
Nickel catalysts, supported on alumina, silica, and molybdena, have been prepared by impregnation and
co-crystallization. In the precursor state the catalysts were characterised by UVÈvisible spectroscopy,
thermogravimetric analysis/di†erential thermal analysis (TGA/DTA), and X-ray photoelectron spectroscopy
(
XPS). The nickel was principally in the 2] oxidation state with an octahedral coordination. However, the
ligand sphere surrounding the nickel ion was sensitive to the support, indicating that the species on the
di†erent supports were not identical thus suggesting a metal complexÈsupport interaction. Reduction was
followed by temperature programmed reduction (TPR) and TGA, the results of which indicated that reduction
and decomposition of nickel nitrate occurred simultaneously. X-ray di†raction (XRD) analysis revealed that,
with the Ni/MoO sample, no hydrogen bronze was formed on reduction. The reduced catalysts were
3
characterised by carbon monoxide chemisorption, carbon dioxide chemisorption, and by reaction of buta-1,3-
diene with dihydrogen. In the absence of a dihydrogen stream it was found that the catalysts adsorbed no
carbon monoxide due the presence of sub-monolayer quantities of surface oxygen. The extent of the oxygen
was quantiÐed by exchange with isotopically labelled carbon dioxide. Di†erences in the electronic nature of the
nickel between the Ni/MoO sample and the other catalysts were revealed by their behaviour towards buta-1,3-
3
diene hydrogenation.
low nickel loadings. The reason for this was to amplify any
Introduction
support or preparation e†ects, while maintaining a sufficient
loading to give satisfactory analysis sensitivity, and to retain
parity with the platinum systems.
The central objective of this collaborative investigation has
been to establish a framework of theory and practice within
which the performance (i.e. activity, selectivity and lifetime
characteristics) of supported metal catalysts (particularly those
containing Pt and Ni) can be evaluated and correlated with (i)
the di†erent preparative procedures used, (ii) the physical
properties, and (iii) the chemical properties, both of the pre-
cursors and of the activated materials. Some of our work on
supported platinum catalysts has been reported1h4 and this
paper is the Ðrst of a series in which parallel work on sup-
ported nickel catalysts is described. It reports the preparation,
physical and chemical characterisation of four materials, three
of which were prepared by conventional impregnation
methods, using each of the title supports, together with a Ni/
alumina which was prepared by co-crystallization (as deÐned
in the Experimental section). In contrast to much previous
work5h7 the catalysts used in this study were chosen to have
The physical states of the catalyst precursors have been
characterised by X-ray photoelectron spectroscopy (XPS),
X-ray di†raction (XRD), thermogravimetric analysis (TGA),
temperature programmed reduction (TPR), UVÈvisible spec-
troscopy, and microanalysis. The chemical states of the active
catalysts have been characterised by pulse and static chemi-
sorption measurements, incorporating isotope tracer tech-
niques. Since it is well established that the product
composition from buta-1,3-diene hydrogenation is a sensitive
indicator of the presence of electronegative impurities
(especially S, but also P, As, Se, Cl, and Br, but not O) at
catalytically active metal surfaces8,9 this reaction was utilised
to provide further chemical characterisation of the supported
nickel catalysts.
Throughout this series of publications1h4 a standard
nomenclature has been adopted to designate the catalyst prep-
aration method. Catalysts designated (I) have been prepared
by impregnation and catalysts designated (C) have been pre-
pared by co-crystallization (see below). The units used
throughout this paper are molecules g~1, where the weight
¤
Present address: Synetix, Research, Technology and Engineering
Group, PO Box 1, Belasis Ave., Billingham, Cleveland, UK TS23 1LB.
Present address: Department of Chemistry, University of Liver-
pool, PO Box 147, Liverpool, UK L69 7ZD
”
Phys. Chem. Chem. Phys., 1999, 1, 2573È2580
2573

View Article Online
refers to the total weight of the catalyst as distinct from the
weight of metal. With respect to ratios of nickel atoms, a sub-
script ““sÏÏ will be used to denote when only surface atoms are
being considered.
In the course of this study a Ni/tungstia catalyst was also
prepared, however, it was found to have very low activity for
any of the catalytic tests and so we have not reported any of
the characterisation relating to this sample. If a reader has an
interest in the data we would be happy to enter into corre-
spondence.
also analysed to provide the background information neces-
sary for the interpretation on the spectra of the catalysts.
X-ray di†raction measurements were carried out on a
Siemens automatic powder di†raction system which incorpor-
ated in situ reduction facilities. Cu-Ka radiation was used.
Thermogravimetric analyses (TGA) were carried out using a
du Pont Model 1090 thermogravimetric analyser and a
heating rate of 10 K min~1 in either Ñowing dinitrogen or
Ñowing 5% H ÈN .
2
2
Temperature programmed reduction (TPR) was performed
using a pulse-Ñow microreactor system. Using this system the
catalysts were reduced in situ in Ñowing 5% H ÈN by heating
2
2
Experimental
to 873 K. The TPRs were monitored using a thermal conduc-
tivity detector (TCD) gas chromatograph coupled to a mass
Preparation of the catalyst precursors
spectrometer (Spectramass SM100D, Ðtted with
resolution RF head).
a high
Two methods of catalyst preparation, namely impregnation (I)
and co-crystallization (C), were used in the production of the
four supported nickel catalysts.
Adsorption studies
Three catalysts were prepared by impregnation, namely Ni/
silica, Ni/alumina, and Ni/molybdena. The supports used were
silica (M5 Cab-O-Sil), alumina (Degussa Aluminium Oxid C),
and molybdena (BDH Analar grade, 99.5%) of surface areas
Chemisorption studies were performed in a dynamic mode
using a pulse-Ñow microreactor system in which the catalyst
sample was placed on a sintered glass disc in a vertical tube (8
mm id, down Ñow) inside a furnace. The reactant pulses were
introduced into the gas stream immediately above the catalyst
bed using a sample loop of identical dimensions to the reactor.
Using this system the catalysts (typically 0.2È1.0 g) were
reduced in situ in Ñowing 5% dihydrogen in dinitrogen, or
dihydrogen by heating to 723 K at 7 K min~1 and holding at
this temperature for 2 h. When reduction had ceased, as mea-
sured by the absence of any dihydrogen up-take, the Ñow was
changed to helium (60 cm3 min~1) and the catalyst held at
2
03, 102, and 2.1 m2 g~1 respectively, and zero pore volume.
Aqueous solutions of nickel(II) nitrate hexahydrate were
added to each support in appropriate amounts to achieve
weight loadings of ca. 1.0% for silica and alumina and 0.5%
for molybdena respectively. The former were prepared on the
500 g scale and the latter 250 g. Catalyst precursor slurries
were transferred to a Buchi rotary evaporator and the water
slowly removed by maintaining the contents at 353 K under a
partial pressure of dry dinitrogen. After 48 h pale green free-
Ñowing powders were obtained. The nickel contents (wt.%) of
the supported catalysts in the pre-reduction state were deter-
mined by inductively coupled plasma mass spectrometry
723 K until no dihydrogen could be detected in the effluent.
The catalyst was cooled to the desired temperature, under
Ñowing helium and the adsorbate gas admitted by injecting
pulses of known size (typically 11 lmol) into the helium
carrier-gas stream and hence onto the catalyst. In all cases the
whole pulse was analysed by on-line GC-MS. The amount of
gas adsorbed, from any pulse, was determined from the di†er-
ence between a calibration peak area and the peak area
obtained following the injection of pulses of comparable size
onto the catalyst. The detection limit for adsorption is
(
ICP-MS) to be 0.97, 0.95, and 0.54 for Ni/silica, Ni/alumina,
and Ni/molybdena respectively.
One catalyst, Ni/alumina (C) was prepared by co-
crystallization using the following procedure. Nickel(II) nitrate
hexahydrate (24.8 g) in de-ionised water (250 ml) was added,
with stirring, to a gelatinous solution of alumina in nitric acid
previously prepared by the slow (16 h) dissolution of alumina
2
] 1017 molecules g~1. Desorptions were performed by
(
495.0 g) in concentrated nitric acid (3.47 l). The resultant solu-
heating the catalyst to 723 K in Ñowing helium. After thermal
desorption the catalyst was cooled in Ñowing helium. Adsorp-
tions and desorptions were followed using a gas chromato-
graph Ðtted with a thermal conductivity detector and Porapak
Q column coupled to a mass spectrometer (Spectra mass
SM100D).
tion was neutralised (Ðnal pH 7), with continuous stirring, by
the careful drop-wise addition of 0.880 ammonia (3.89 l),
during which the temperature rose to 323È333 K. After stand-
ing overnight water was slowly removed (48 h) on a rotary
evaporator and the resulting solid dried in a vacuum oven
(
353 K, 48 h). Batches (50 g) of this material were then Ðred in
The apparatus and general methods for the study of buta-1,3-
diene hydrogenation have been described previously.10
a tube furnace at 523 K using the following reaction condi-
tions: heating from ambient to 523 K over 1.5 h, maintaining
the temperature for 2 h followed by cooling slowly to room
temperature over several hours. An air Ñow of 50 ml min~1
was maintained throughout. During this process a weight loss
of ca. 88% was noted. The resultant pale green solid, of nickel
content 0.85 wt.%, was used as the catalyst precursor.
Materials
Both the helium (Air Products, 99.998%) and the 5% dihydro-
gen in dinitrogen (ICI) were further puriÐed by passing
through activated palladium bronze to remove any dioxygen
impurity, and a bed of Carbosorb AS (BDH) to remove any
water impurity. [18O] carbon dioxide and [13C] carbon
dioxide were supplied by Amersham International (both [99
atom%) and were used without further puriÐcation.
Methods of characterisation
UVÈvisible spectra of the solid catalyst precursors were
recorded over the range 190È800 nm using a Philips 8800
UVÈvisible spectrometer with a di†use reÑectance attachment.
Spectra of solutions were recorded using standard 1 cm cells.
X-ray photoelectron spectroscopy (XPS) spectra were col-
lected using either a Kratos XSAM 800 electron spectrometer
or a VG ESCALAB-200D instrument. The pressure was
maintained at +10~10 mbar during analyses. Spectra were
initially calibrated with reference to the C(1s) line (285.0 eV)
and nickel binding energies normalised against the Si(2p)
Results
Characterisation of supported materials before/after reduction
Spectra of the supported materials, after initial drying, but
before calcination and reduction, were measured by di†use
reÑectance UVÈvisible spectroscopy and their absorption
maxima and assignments are recorded in Table 1.
(
104.1 eV), A1(2p) (74.6 eV), and Mo(3d) (233.3 eV) values of
XPS data were recorded for the impregnated catalysts in
their pre-reduction states only. The quantitative results of the
the respective supports. A number of reference samples were
2574
Phys. Chem. Chem. Phys., 1999, 1, 2573È2580

View Article Online
Table 1 Di†use reÑectance spectra (nm) of supported nickel material in the pre-reduction state
Band positions and assignments
Catalyst
3A
] 3T1g(F)
3A ] 3T
Nitrate
Nitrate
2g
2g
1g
Ni/silica (I)
740
740
740
750
708
728
686 sh
698
642
662 sh
650
656
400
372
384
290
296
300
206
208
206
Ni/alumina (I)
Ni/alumina (C)
Ni/molybdena (I)
Ni(NO ) É 6H O
Obscured by MoO
386
393
3
296
300
3
2
2
Ni(NO ) aqueous solution
3
2
Table 2 XP spectra: Elemental concentrations (atom%) in the pre-reduction state
Catalyst
C(1s)
Ni(2p)
O(1s)
Si(2p)
Al(2p)
Mo(3d)
Ni/silica (I)
8.3
10.2
35.3
36.8
0.4
1.5
5.6
È
62.5
60.9
48.8
46.7
28.9
È
È
È
27.5
È
È
È
10.2
16.6
Ni/alumina (I)
Ni/molybdena (I)
MoO3
È
È
Table 3 XP Spectra: Binding energies (eV) in the pre-reduction state
Catalyst
C(1s)
Ni(2p)
O(1s)
Si(2p)
Al(2p)
Mo(3d5@2)
Ni/silica (I)
285.1
285.0
285.1
285.1
854.9
854.0
854.5
È
532.3
530.3
530.2
530.5
104.4
È
È
È
74.8
È
È
È
233.1
233.3
Ni/alumina (I)
Ni/molybdena (I)
MoO3
È
È
analyses are summarised in Table 2, in which all elements
detected are expressed as relative atomic percentages. Binding
energy data are summarised in Table 3.
TPR measurements were checked at three centres
(Billingham, Glasgow and Hull) and, although some minor
di†erences in detail were noted, the results were in general
agreement. Reduction maxima observed over the range 273È
873 K are summarised in Table 5, and a typical TPR proÐle is
shown in Fig. 2.
XRD measurements were carried out on Ni/molybdena (I)
both in the as prepared state and in situ during reduction on a
heated stage. The results were consistent with the presence of
The impregnated and co-crystallized catalysts were each
examined by TGA in both Ñowing dinitrogen and 3.5%
H ÈN . Typical TGA proÐles, for Ni/silica (I), are illustrated
2
2
in Fig. 1 and temperatures associated with the various
decomposition/reduction transitions are summarised in Table
4.
MoO and, in contrast to the situation with Pt/molybdena
3
(
I),1 provided no evidence of bronze formation.
Adsorption studies
Adsorption of carbon monoxide at 273 K was investigated on
freshly reduced catalysts using a helium carrier gas. No
adsorption could be detected on any of the catalysts.
Dioxygen contamination of the helium supply was initially
suspected to be the source of this unexpected behaviour.
However, tests using the same gas supply with other catalysts
Table 4 Thermogravimetric analyses: Temperature (K) of weight
loss maxima
Catalyst
Under N2
Under 5% H ÈN
2
2
Ni/silica (I)
568
563,713
565
493 (718), [973
553
613
560
Ni/alumina (I)
Ni/alumina (C)
Ni/molybdena (I)
483, 558, [973
Table 5 TPR data: Temperature (K) of reduction maxima
Catalyst
Temperature/K
Ni/silica (I)
578
573
556
560
Ni/alumina (I)
Ni/alumina (C)
Ni/molybdena (I)
Fig. 1 (a) TGA of nickel nitrate/silica under dinitrogen. (b) TGA of
nickel nitrate/silica under dihydrogen/dinitrogen.
Phys. Chem. Chem. Phys., 1999, 1, 2573È2580
2575

View Article Online
Fig. 3 Variation in carbon dioxide isotope proÐle. Catalyst: Ni/
molybdena (I)
Fig. 2 TPR proÐle of nickel nitrate/silica.
had altered between inlet (1.02) and exit (1.27) such that there
was an excess of [16O] oxygen. The amount of [16O] oxygen
incorporated into the carbon dioxide was calculated to be
2.03 ] 1018 atoms g~1. No gaseous material was desorbed
from the catalyst during heating to 623 K and no carbon
monoxide was adsorbed, or underwent exchange, after the
sample had been cooled to 373 K.
showed no evidence of contamination by dioxygen. The
experiment was also repeated in two of the participating
laboratories using commercial as well as home-made dioxygen
getter units; identical behaviour was observed, i.e. no carbon
monoxide was adsorbed. The experiments were repeated using
dihydrogen as the carrier gas. In this case adsorption was
observed. The results are presented in Table 6 along with
derivative calculations of dispersion and mean particle size. A
CO : Ni ratio of 1 : 2 was used to calculate the dispersion of
the metal. However we are conscious that the ratio of CO : Ni
achieved in a dihydrogen Ñow, where competitive adsorption
can occur, may not be equivalent to that obtained in an inert
gas Ñow. There was also concern that some reaction may take
place, however analysis of the exit gas, by GC-MS, with dihy-
drogen as carrier, conÐrmed that no carbon dioxide or
methane had been produced, indicating that no obvious reac-
tion was occurring.
The adsorption of carbon dioxide, as a mixture of [13C]
carbon dioxide and carbon [18O] dioxide, was investigated on
all catalyst samples. By use of such an isotopic mix it is pos-
sible to investigate both adsorption and exchange processes.
Two types of isotope exchange are detectable using this
experimental procedure: (i) the exchange of an atom(s)
between an adsorbed molecule and the surface, and (ii) the
transfer of an atom(s) between two adsorbed molecules. In
order to establish the most suitable temperature at which to
perform the experiments, adsorption and exchange processes
on Ni/silica (I) were investigated at 273, 373 and 473 K using
carbon [18O] dioxide. The most e†ective exchange was
observed at a temperature of 373 K, hence this temperature
was selected for all subsequent carbon dioxide experiments.
Ni/silica (I)
On pulsing aliquots of 1 : 1 [13C] carbon dioxideÈcarbon
[
18O] dioxide over freshly reduced Ni/silica, at 373 K, adsorp-
tion was detected. The amount of carbon dioxide adsorbed
was 0.81 ] 1018 molecules g~1. That carbon dioxide which
did not adsorb was shown by mass spectrometric analysis
(
Fig. 4) to have undergone exchange with a source of [16O]
oxygen (inlet [16O] : [18O] ratio 0.94, exit [16O] : [18O] ratio
for pulse 1, 6.35). The total amount of excess [16O] oxygen
incorporated into the carbon dioxide over the ten pulses was
2
.069 ] 1019 atoms g~1. The [12C] : [13C] ratio was identical
for both inlet and outlet gas (0.98). When the experiment was
repeated using a sample of the support material alone, which
had been treated in an identical manner to that of the catalyst
sample, no adsorption or exchange was observed.
Ni/alumina (I)
The amount of carbon dioxide adsorbed at 373 K on a freshly
reduced catalyst was 3.53 ] 1018 molecules g~1. There was
also signiÐcant incorporation of [16O] oxygen into the carbon
dioxide (Fig. 5). However when the experiment was repeated
using the alumina support material alone, after identical treat-
ment to that of the catalyst, all the adsorptive capacity
observed with Ni/alumina (I) could be accounted for by
adsorption on the support, and a majority of the exchange
also occurred on the support. Nevertheless approximately
Ni/molybdena (I)
4
] 1018 atoms g~1 of [16O] oxygen were not accounted for
No adsorption was detected when ten aliquots of 1 : 1 [13C]
carbon dioxideÈcarbon [18O] dioxide were pulsed over a
freshly reduced catalyst at 373 K. However, when the isotope
pattern of the exit carbon dioxide was analysed (Fig. 3), it was
found that whereas the [12C] : [13C] ratio was the same for
both inlet (0.92) and outlet gas (0.98), the [16O] : [18O] ratio
by exchange on the support. The fact that the majority of the
adsorption and exchange processes were associated with the
support in Ni/alumina (I) made it inappropriate to examine
Table 6 Carbon monoxide adsorptiona
CO adsorbedb
(]1019)
Dispersionc
(%)
SdTd
Catalyst
/nm
Ni/silica (I)
Ni/alumina (I)
Ni/molybdena (I)
1.87 ^ 0.40
0.50 ^ 0.05
1.31 ^ 0.01
38
10
47
2.7
10.1
2.1
a Conditions: carrier gas, dihydrogen, 60 cm3 min~1; temperature 273
K. b Units: molecules g~1. c Calculated assuming 1 CO : 2 Ni.
d SdT \ mean surface average diameter, calculated using SdT \ 101/d,
where d \ dispersion.
Fig. 4 Variation in carbon dioxide isotope proÐle. Catalyst: Ni/silica
(
I)
2576
Phys. Chem. Chem. Phys., 1999, 1, 2573È2580

View Article Online
The TGA traces of the impregnated materials were all basi-
cally similar, comprising a peak in the region 333È358 K, con-
sistent with loss of water, and one or two peaks at higher
temperatures, which could be attributed to the decomposition
of nickel nitrate to nickel oxide (or, in the presence of dihydro-
gen, reduction to metallic nickel). For example, in the case of
Ni/SiO (I) (Fig. 1), examination of the two derivative peaks
2
for the weight loss at 568/533 K indicates that the presence of
hydrogen results in a broader peak which is shifted slightly to
a lower temperature. That fact, taken in conjunction with the
TPR data (below) suggests that decomposition of nickel
nitrate to nickel oxide and reduction to the metal occur at
essentially coincident temperatures of ca. 560 K.
Fig. 5 Variation in carbon dioxide isotope proÐle. Catalyst: Ni/
alumina (I)
After TGA under dinitrogen the Ni/SiO (I) and Ni/Al O
2
2 3
(
I) residues were found to be 0.31 and 0.34% heavier than
those after treatment with 3.5% H ÈN . These Ðgures are in
good agreement with the expected di†erences between NiO
and Ni in the respective samples.
Ni/alumina (C), since no blank support material was available
for comparison.
2
2
From the TGA traces for Ni/MoO (I) it was difficult to
assign unambiguously a speciÐc weight loss to the reduction
Buta-1,3-diene hydrogenation
3
All catalysts were tested for the hydrogenation of buta-1,3-
diene. With the exception of Ni/silica (I) only low catalytic
activities were observed following reduction using the pre-
viously determined standard procedures. The alumina- and
molybdena-supported catalysts were therefore reduced at the
higher temperatures of 923 and 693 K respectively. Observed
product distributions and selectivities are summarised in
Table 7.
of nickel. The most probable values are indicated in Table 5.
The catastrophic weight losses observed above 923 K are con-
sistent with the decomposition/reduction of the support as
was observed with the analogous Pt/MoO (I) system.1 In
3
contrast to the Pt/MoO (I) system no small weight increases
3
were observed and there is thus, as expected, no evidence for
bronze formation with nickel. This was conÐrmed by XRD
measurements.
In the case of the Ni(NO ) /Al(NO ) /NH NO sample
3
2
3 3
4
3
prepared by co-crystallization, both TGA and di†erential
scanning calorimetry (DSC) were used initially to establish the
most appropriate temperature at which to calcine the material
prior to reduction and generation of Ni/alumina (C). Broad
endothermic proÐles, which corresponded to signiÐcant
weight losses, were observed in the ranges 300È430 K and
473È523 K. Unfortunately these broad features masked the
much sharper proÐles exhibited by a sample of pure ammon-
ium nitrate which was measured for comparison. The large
endothermic weight losses correspond to the removal of water
and decomposition of aluminium nitrate and ammonium
nitrate, respectively. At temperatures above 523 K a sharp
exotherm was observed at ca. 560 K in both TGA and DSC
traces. The bulk of the sample was therefore calcined at 523 K
and a further TGA carried out on the resultant material. This
Discussion
The di†use reÑectance spectral proÐles of the materials pre-
pared by conventional aqueous impregnation are all very
similar, closely resembling the spectra of nickel nitrate hexa-
hydrate, and are thus consistent with the presence of octa-
hedrally co-ordinated Ni(II) species prior to calcination and
reduction. There is evidence for a slight shift to longer wave-
length for the 3A ] 3T band suggesting the presence of a
ligand of weaker Ðeld strength than water (possibly OSi).
Absorption maxima at ca. 295 and 205 nm are characteristic
of the spectrum of nitrate ion. These were not measured in
every spectrum because of the cut-o† due to high absorbance
values.
From the XP spectra, taking into consideration the di†erent
metal loadings, the observed nickel levels appear to bear an
approximately inverse relationship to the surface areas of the
supports. The Ni(2p) binding energies are essentially consis-
tent with the presence of Ni(II) species in the pre-reduction
state, as would be expected. Taken together with the UVÈ
visible di†use reÑectance spectra, both techniques indicate
that the nickel is in a 2] oxidation state with octahedral
coordination. However they also indicate that the species on
each support is not identical, but has a slightly di†erent ligand
coordination sphere suggesting a metal complexÈsupport
interaction in the catalyst precursors.
2g
lg
showed
a
very simple trace with
a
single, sharp
decomposition/reduction peak. From TPR measurements all
samples gave reduction peaks in the temperature range 550È
580 K which, taken in conjunction with the TGA results,
could be assigned to the reduction of nickel to the metallic
state. Additional maxima were observed at higher tem-
peratures, e.g. 668 K for Ni/SiO (I), 636 and 656 K with
Ni/Al O (I) and above 700 K for Ni/MoO (I). Other
studies11 on uncalcined and calcined Ni/SiO catalysts pre-
pared from nickel(II) nitrate clearly show that a peak at ca.
2
2
3
3
2
580 K is due to the reduction of the nitrate precursor rather
Table 7 Product distributions and selectivities in the catalytic hydrogenation of buta-1,3-diene
Catalyst
Ni/silica (I)
Ni/alumina (I)
Ni/alumina (C)
Ni/molydbena (I)
Reduction conditions
(i) 623 [2] (]2)
(ii) 693 [1.5]
80
/K [h]
623 [2]
296
923 [3]
353
923 [2]
80
Reaction temperature/K
(
65
20
15
1.3
0.80
i)
(ii)
51
32
16
2.0
0.93
b-1 61
t-b-2 25
c-b-2 14
1.8
64
25
11
2.3
0.95
62
30
8
3.7
0.92
Products
T rans/cis
Selectivitya
0.96
a Selectivity deÐned as butene/(butenes ] butane).
Phys. Chem. Chem. Phys., 1999, 1, 2573È2580
2577

View Article Online
than the oxide precursor; reduction of the latter occurs at ca.
amounts of retained oxygen. The interaction is not simple and
the support plays a signiÐcant part in determining the behav-
ior of the system. A study of isotope exchange over platinum
catalysts12 conÐrmed that this technique allows a determi-
nation of surface oxygen concentration.
With the Ni/molybdena catalyst, no retention of any carbon
dioxide was observed but adsorption/desorption processes
had occurred. Analysis of the isotopic composition also indi-
cated that exchange with residual oxygen from the catalyst
surface was occurring, i.e. the [16O] : [18O] ratio increased.
This excess [16O] oxygen was not large (2.03 ] 1018 atoms
g~1). The following outlines the exchange process;
6
50 K.11 Therefore in the cases of Ni/SiO (I) and Ni/Al O
2
2 3
(
I) the additional maxima may be assigned to the reduction of
small amounts of nickel(II) oxide.
The maxima observed at temperatures above 700 K appear
to be associated with the reduction of the supports. For
Ni/MoO3 (I) the support reduction can be attributed to
hydrogen spillover from nickel (without bronze formation).
An indication of the extent of reduction, as measured by
hydrogen uptake, was obtained for each of the impregnated
catalysts by measuring the peak areas as a function of the
nickel content. The H : Ni ratios were 19, 14, and 12 for
(I), Ni/Al O (I), and Ni/MoO (I) respectively,
Ni/SiO2
2 3
3
12C18O H 12C18O(ads.) ] 18O(ads.)
(1)
(2)
(3)
(4)
largely consistent with the stoichiometry required for the
2
reduction of nickel nitrate, viz. Ni(NO ) ]9H ]Ni]2NH
13C16O H 13C16O(ads.) ] 16O(ads.)
3
2
2
3
2
]
6H O. Thus the H : Ni ratio showed a range between 18
2
1
1
2C18O(ads.) ] 16O(ads.) H 12C18O16O
and 2 depending upon whether the major process is the direct
reduction of nickel(II) nitrate to metallic nickel or initial
thermal decomposition of the nitrate to oxide followed by
reduction of NiO to the metal. The results suggest that the
former situation predominates. In support of this an evolved
gas analysis recorded during the reduction of Ni/Al O (I)
3C16O(ads.) ] 18O(ads.) H 12C16O18O
Such exchange is not unexpected as the literature13h15 pro-
vides evidence that adsorption of carbon dioxide on a nickel
surface is dissociative towards CO and O. From Table 6 it can
be calculated that the number of surface nickel atoms g~1 in
2
3
showed the formation of NO . The evolution reached a
x
maximum at ca. 565 K coincident, within the accuracy of the
the Ni/MoO catalyst is 2.62 ] 1019. Therefore the residual
3
experiment, with the TPR maximum (see Table 5). In contrast
to the behaviour of the analogous platinum catalysts1 there
was no evidence for signiÐcant bronze formation with the
molybdena support.
surface oxygen represents 8% of a monolayer (O : Ni ratio of
1 : 27). The degree of incorporation into the gas phase decayed
as the [18O] oxygen, exchanged from the carbon dioxide,
diluted the [16O] oxygen reservoir on the surface. In the Ðrst
pulse 10% of the carbon dioxide was exchanged, by the tenth
pulse only 3% of the carbon dioxide contained excess [16O]
oxygen. The processes occurring are as outlined in reactions
(1)È(4) but with an imbalanced population of [16O] oxygen
and [18O] oxygen such that reactions (2) and (3) are favoured.
When carbon dioxide was pulsed over the Ni/silica catalyst
considerable exchange occurred. The O : Ni ratio for the
amount of oxygen exchanged (2.276 ] 1019 atom g~1) over
the ten pulses was 0.23 : 1 or 1 : 4.4. This represents, in terms
of the amount of surface nickel, the equivalent of 61% of a
monolayer. By the tenth pulse the isotope ratio of the oxygen
in the carbon dioxide was normal, but the gas was isotopically
scrambled (Fig. 4).
By comparing the carbon dioxide exchange results from the
Ni/molybdena and Ni/silica samples with their particle sizes
(Table 6) it is apparently not a site density di†erence that
accounts for the di†erence in isotope exchange. There was
however, a larger amount of residual surface oxygen present
with the Ni/silica than with the Ni/molybdena. If carbon
dioxide was required only to adsorb where there was no
adsorbed oxygen, then given the available surface area on the
Ni/silica sample, only a low rate of exchange would be
expected. However, if the carbon dioxide can adsorb on the
oxygen pre-covered sites, an alternative exchange mechanism
is possible. When the concentration of surface oxygen is low
the exchange reaction can be rationalised by dissociative/
recombination mechanism where, in the presence of a large
concentration of surface oxygen, the mechanism of
adsorption/exchange can change from dissociative to associa-
tive, i.e.
Chemisorption
The absence of carbon monoxide adsorption when helium was
used as the carrier was surprising in the light of the TPR
experiments and a range of other physical techniques which
indicated the presence of nickel metal. This lack of carbon
monoxide adsorption was conÐrmed by independent studies
in two centres (Glasgow and Billingham). Catalysts were
reduced in dihydrogen/dinitrogen and the reduction was fol-
lowed by measuring the up-take of dihydrogen. The samples
were heated to a temperature (723 K) well above that where
the maximum rate of hydrogen abstraction was observed and
the catalysts were held at this temperature until no further
hydrogen up-take was observed. Had it not been for the fact
that the same result was obtained in more than one of the
co-operating laboratories, contamination of the samples
would almost certainly have been assumed, and indeed con-
siderable care was taken in all of the laboratories to ensure no
contamination to the catalysts occurred post-reduction. The
reason for the absence of carbon monoxide adsorptive capac-
ity was, as conÐrmed by carbon dioxide chemisorption (as dis-
cussed below), the presence of surface oxygen. These results
suggest that removal of monolayer and sub-monolayer levels
of oxygen from nickel catalysts is kinetically slow and that in
the absence of a reducing atmosphere will block the surface to
carbon monoxide adsorption. Interestingly the use of dihydro-
gen as a carrier gas was sufficient to allow adsorption of
carbon monoxide. No evidence was obtained to suggest that
dihydrogen removed the surface oxygen at 298 K. Two expla-
nations can be advanced although neither is entirely satisfac-
tory. If the oxygen was randomly distributed over the surface
under helium and the action of dihydrogen was to cause
agglomeration into islands then a signiÐcant proportion of the
adsorptive capacity could be reclaimed. Alternatively if the
surface oxygen was to di†use into bulk or be re-dispersed
throughout the crystallite, then adsorptive capacity would
also be regained. A deÐnitive explanation must however await
further study.
CO ] O(ads.) H CO (ads.)
2
3
In the form CO (ads.) all oxygen atoms are equivalent and
3
exchange can be rapid as has been observed for carbon
dioxide adsorption and exchange over copper catalysts.16,17
Moreover such species have been identiÐed, at 300 K, in a
UPS (ultraviolet photoelectron spectroscopy)/XPS/TPR study
by Behm and Brundle,18 when carbon dioxide was adsorbed
on a nickel (100) surface containing 50% of a pre-adsorbed
monolayer of oxygen. However at less than 25% of a mono-
The results obtained from the carbon dioxide adsorption
conÐrm the sensitivity of isotopically labelled carbon dioxide
as a probe molecule for nickel surfaces containing various
layer of oxygen pre-adsorbed, no CO (ads.) species were
3
detected. This is in agreement with results from the Ni/
2578
Phys. Chem. Chem. Phys., 1999, 1, 2573È2580

View Article Online
molybdena catalyst in which with only 8% of a monolayer of
surface oxygen, single atom dissociative exchange only was
observed.
reduction at 693 K, the but-1-ene yield was depressed but the
trans/cis ratio and the butane yield both showed regular
values. This is not consistent with perturbation by an electro-
negative impurity but it might indicate that either molyb-
denum has become incorporated into the metallic active
phase, or that the metallic phase has been incorporated into
the support. There is some support for the latter interpreta-
tion from high resolution electron microscopy where local
areas of distortion were found in the molybdenum oxide
lattice.21 Such distortion may be caused by the presence of
nickel insertion in the lattice. Buta-1,3-diene hydrogenation at
423 K over evaporated molybdenum Ðlms gives S \ 0.83, and
but-1-ene yields in the range 56È62% (trans/cis \ 1.7), all
values being lower than for nickel Ðlms.19,22 However we are
unaware of any literature on buta-1,3-diene hydrogenation
over Ni/Mo bimetallic surfaces and so the conclusion must
remain provisional at this stage.
Although massive oxygen exchange occurred when carbon
dioxide was pulsed over the Ni/alumina (I) catalyst, the
majority was shown to be due to interaction between the
support and the carbon dioxide (Fig. 5). Similarly the carbon
dioxide adsorption could also be shown to be due to adsorp-
tion on the support. However, in respect of the exchange, the
e†ect of the support can be subtracted out by reference to the
results from the support alone. Following this procedure a
Ðgure of 40% of a monolayer of oxygen is calculated as the
amount of residual oxygen present at the metal surface.
Let us now consider some general aspects of the interaction
of carbon dioxide with surface oxygen. For a catalyst where
the amount of [18O] oxygen in one pulse is large (or at a
similar level) relative to the amount of surface oxygen, then,
with the assumption that all gas phase molecules can interact
with the surface, an exponential decay would be observed. As
is evident from Fig. 6, such an exponential decay was
observed for the Ni/silica (I) and Ni/molybdena (I) catalysts.
Although we cannot be deÐnitive at this stage, due to the
interference by the support, it appears that the Ni/alumina (I)
catalyst behaved in a similar manner.
In conclusion, it has been demonstrated that isotopically
labelled carbon dioxide is an informative probe of nickel sur-
faces which retain sub-monolayers of oxygen. Adsorption,
desorption and exchange processes have been followed and
the change in the degree of exchange and the mode of carbon
dioxide bonding has been shown to be dependent on the
amount of retained oxygen. The amount of retained oxygen
was support dependent, with the Ni/silica catalyst having the
greatest surface coverage.
Conclusions
In this paper we have reported the preparation and character-
isation of a range of nickel catalysts. We have demonstrated a
high degree of internal consistency between (i) results obtained
using duplicate techniques at di†erent centres and, (ii) results
obtained using a variety of di†erent physical techniques.
Using a combination of physical and chemical probes we have
shown that in the as-prepared materials the nickel is in a 2]
oxidation state but that the ligand sphere surrounding the
metal is inÑuenced by the di†erent supports indicating a metal
complexÈsupport interaction. The reduction of the surface
species was shown to occur in a single process involving
simultaneous decomposition/reduction of nickel nitrate. The
adsorption studies revealed that with each catalyst the surface
after reduction contained sub-monolayers of oxygen, that
blocked carbon monoxide adsorption in an inert gas stream.
Carbon monoxide adsorption was possible when the carrier
gas was dihydrogen. The extent of the surface oxygen was
determined by exchange with isotopically labelled carbon
dioxide. The electronic structure of the reduced catalysts was
probed by buta-1,3-diene hydrogenation, only the Ni/
molybdena (I) showed an unusual pattern to the product dis-
tribution suggesting an interaction with the support.
Therefore having now characterised the systems it is our
intention to undertake catalytic studies with a view to inter-
preting the results with respect to the physico-chemical char-
acterisation.
Catalytic hydrogenation of buta-1,3-diene
The characterisation of the catalysts with respect to their per-
formance in buta-1,3-diene hydrogenation is reported in Table
7. Clean evaporated nickel Ðlms catalyse buta-1,3-diene
hydrogenation at about 363 K giving typically 99% butene
and 1% butane (selectivity, S \ 0.99), and a butene composi-
tion of b-1 \ 70%, t-b-2 \ 22%, c-b-2 \ 8% (trans/
cis \ 2.8).19 In comparison, highly dispersed nickel in a range
of Ni/silicas gives S \ 0.92È0.98 and butene compositions of
b-1 \ 60 ^ 4%, t-b-2 \ 28 ^ 3%, c-b-2 \ 12 ^ 2% (trans/
cis \ 2.3).20 Such butene distributions are strictly independent
of dihydrogen pressure and only mildly dependent upon tem-
perature and catalyst preparation conditions. Therefore Ni/
silica (I) and Ni/alumina (I) clearly conform to the criteria for
a surface free of electronegative impurities. Ni/alumina (C)
also obeys the criteria except that the trans/cis ratio, at 3.7, is
higher than expected. The high trans/cis ratio on its own is
not sufficient to suggest the presence of electronegative impu-
rities. The behaviour of the Ni/molybdena (I) is problematical.
When reduced at 623 K the catalyst gave twice the expected
yield of butane but a regular butene composition whereas, on
Acknowledgements
The authors would like to thank the technique specialists at
ICI plc, the University of Glasgow, and the University of Hull,
who helped obtain and interpret some of the results presented
in this paper. We are grateful to the SERC (now EPSRC) and
ICI Chemicals & Polymers Ltd., for their generous Ðnancial
support of the work described in this paper.
References
1
2
3
4
5
S. D. Jackson, J. Willis, R. Whyman, G. Webb, G. D. McLellan,
M. B. T. Keegan, R. B. Moyes, S. Simpson and P. B. Wells, J.
Catal., 1993, 139, 191.
S. D. Jackson, J. Willis, R. Whyman, G. Webb, G. D. McLellan,
M. B. T. Keegan, R. B. Moyes, S. Simpson and P. B. Wells, J.
Catal., 1993, 139, 207.
S. D. Jackson, J. Willis, R. Whyman, G. Webb, G. D. McLellan,
M. B. T. Keegan, R. B. Moyes, S. Simpson and P. B. Wells, J.
Catal., 1993, 139, 221.
S. D. Jackson, G. Webb, P. B. Wells, R. Whyman, G. D. McLel-
lan, D. A. Whan, S. Mather, S. Simpson, L. Conyers and M. B. T.
Keegan, J. Catal., 1996, 162, 10.
L. A. M. Hermans and J. W. Geus, in Preparation of Catalysts, II,
ed. B. Delmon, P. Grange, P. A. Jacobs and G. Poncelet, Studies
Fig. 6 Variation in oxygen exchange with pulse number.
Phys. Chem. Chem. Phys., 1999, 1, 2573È2580
2579

View Article Online
in Surface Science and Catalysis, Elsevier, Amsterdam, 1979, vol.
, p. 113.
J. W. E. Coenen and P. B. Wells, in Preparation of Catalysts, III,
ed. P. Grange, P. A. Jacobs and G. Poncelet, Studies in Surface
Science and Catalysis, Elsevier, Amsterdam, 1983, vol. 16, p. 801.
J. W. E. Coenen, Appl. Catal., 1991, 75, 193.
M. George, R. B. Moyes, D. Ramanarao and P. B. Wells, J.
Catal., 1978, 52, 486.
J. J. Phillipson, P. B. Wells and G. R. Wilson, J. Chem. Soc., A,
15 F. Solymosi, J. Mol. Catal., 1991 65, 337, and references therein.
16 S. D. Jackson, J. Catal., 1987, 108, 250; G. C. Chinchen, M. S.
Spencer, K. C. Waugh and D. A. Whan, J. Chem. Soc., Faraday
T rans 1, 1987, 83, 2193; S. Kinnaird, G. Webb and G. C. Chin-
chen, J. Chem. Soc., Faraday T rans. 1, 1988, 84, 2135.
3
6
7
8
17 S. D. Jackson, J. Catal.,1989, 115, 247.
18 R. J. Behm and C. R. Brundle, Surf. Sci., 1991, 225, 327.
19 A. G. Burden, J. Grant, J. Martos, R. B. Moyes and P. B. Wells,
Discuss. Faraday Soc., 1981, 72, 95.
20 R. G. Oliver, P. B. Wells and J. Grant, in Proceedings, 5th Inter-
national Congress of Catalysts, ed. J. W. Hightower, North
Holland, Amsterdam, 1973, vol. I, p. 659.
9
0
1
2
1
969, 1351.
1
1
1
G. C. Bond, G. Webb, P. B. Wells and J. M. Winterbottom, J.
Chem. Soc., 1965, 3218, and references therein.
S. D. Jackson, G. Webb, J. R. Jones, L. F. Gladden and A. P.
Sharratt, unpublished results.
V. Ducarme and J. C. Vedrine, J. Chem. Soc., Faraday T rans. 1,
21 T. Baird, J. Fryer, G. J. Kelly and G. Webb, unpublished results.
22 J. Grant, R. B. Moyes and P. B. Wells, J. Catal., 1978, 51, 355.
1
977,75, 506.
1
1
3
4
I. McCarty, J. Falconer and R. J. Madix, J. Catal., 1973, 30, 235.
J. B. Benzinger and R. J. Madix, Surf. Sci., 1979, 79, 394.
Paper 8/09293A
2580
Phys. Chem. Chem. Phys., 1999, 1, 2573È2580