CuAu/SiO2 catalysts for the selective oxidation of propene to acrolein: The impact of catalyst preparation variables on material structure and catalytic performance

Article abstract of DOI:10.1039/c3cy00254c

CuAu/SiO₂ catalysts are active and selective for the synthesis of acrolein from propene by selective oxidation using a mixture of H₂ and O₂ as oxidant. The catalyst synthesis method (impregnation or deposition) and the activation procedure (reduction, calcination or reduction followed by calcination and the choice of reducing agent) all affect the structure of the catalyst and therefore the performance in catalysis. For a material prepared by coimpregnation of HAuCl₄ and Cu(NO ₃)₂, reduction with hydrogen leads to a catalyst with a significant level of CuAu alloy, whilst reduction with NaBH₄ solution leads to small particles but with negligible interaction between copper and gold. Subsequent high temperature calcination of these materials destroys any CuAu phase but retains a Cu-Au interaction as observed by Transmission Electron Microscopy. Preparation of the catalyst by a two-step deposition method gives smaller particles than coimpregnation when reduced by H₂ but a different alloy phase is observed by EXAFS. Calcination of this material gives a catalyst which combines high activity with excellent selectivity to acrolein. Characterisation of the materials using various techniques allows the nature of the active sites to be understood.

Full text of DOI:10.1039/c3cy00254c

Catalysis
Science & Technology
View Article Online
View Journal
PAPER
CuAu/SiO catalysts for the selective oxidation of
2
propene to acrolein: the impact of catalyst preparation
variables on material structure and catalytic
performance†
Cite this: DOI: 10.1039/c3cy00254c
a
b
a
c
St e´ phanie Belin, Charlotte L. Bracey, Val e´ rie Briois, Peter R. Ellis,*
b
c
cd
Graham J. Hutchings, Timothy I. Hyde and Gopinathan Sankar
CuAu/SiO
2
catalysts are active and selective for the synthesis of acrolein from propene by selective oxidation
and O as oxidant. The catalyst synthesis method (impregnation or deposition) and the
using a mixture of H
2
2
activation procedure (reduction, calcination or reduction followed by calcination and the choice of reducing
agent) all affect the structure of the catalyst and therefore the performance in catalysis. For a material
prepared by coimpregnation of HAuCl
4
and Cu(NO
3
)
2
, reduction with hydrogen leads to a catalyst with a
solution leads to small particles but with
significant level of CuAu alloy, whilst reduction with NaBH
4
negligible interaction between copper and gold. Subsequent high temperature calcination of these materials
destroys any CuAu phase but retains a Cu–Au interaction as observed by Transmission Electron Microscopy.
Preparation of the catalyst by a two-step deposition method gives smaller particles than coimpregnation
Received 15th April 2013,
Accepted 6th June 2013
2
when reduced by H but a different alloy phase is observed by EXAFS. Calcination of this material gives a
DOI: 10.1039/c3cy00254c
catalyst which combines high activity with excellent selectivity to acrolein. Characterisation of the materials
using various techniques allows the nature of the active sites to be understood.
www.rsc.org/catalysis
paper we address this topic for the selective oxidation of propene
Introduction
using bimetallic CuAu/SiO catalysts.
2
Designing catalysts for selective oxidation reactions poses a
significant challenge. With the notable exception of CO oxida- oxidation reactions, including the selective oxidation of CO,
CuAu catalysts have been found to be active for a number of
1
2
3
–6
4
7–9
tion, the mechanism of the reaction and the structure of the propene, toluene and alcohols. These materials are complex,
catalysts needed to give good performance are not generally with many factors having an influence on the structure and catalytic
well understood. Yet the controlled introduction of oxygenated performance of the final material. These include preparation
functionality into molecules is desirable for the sustainable method, which in turn determines the final oxidation states of
production of a wide range of chemical products. To success- copper and gold on the catalyst and whether they are present as a
fully synthesise catalysts for these reactions needs a good bimetallic phase or in a segregated form. The catalytic performance
understanding of structure–function relationships; that is, is further influenced by reaction conditions such as temperature,
how changes to the structure of the catalyst affect the activity, space velocity and the composition of the gas mixture – which
selectivity and stability of the catalyst in the reaction. In this contains both oxidising and reducing components.
Understanding the role of CuAu catalysts in propene oxida-
tion presents a particular challenge since both metals are active
catalysts in their own right. Copper catalysts on a range of supports
a
Synchrotron Soleil, L’Orme des Merisiers, Saint-Aubin, BP 48 91192 Gif-sur-Yvette
CEDEX, France
b
c
have been used in the selective oxidation of propene. The major
product from this oxidation pathway is typically acrolein. There is
Cardiff Catalysis Institute, School of Chemistry, Cardiff University, Park Place,
Cardiff CF10 3AT, UK
10
11
12
Johnson Matthey Technology Centre, Blount’s Court, Sonning Common,
Reading RG4 9NH, UK. E-mail: ellispr@matthey.com
Department of Chemistry, University College London, 20 Gordon Street,
London WC1H 0AJ, UK
evidence that copper metal, Cu O and CuO all possess some
2
activity in alkene selective oxidation. Cu O in particular is reported
d
2
as being a highly selective phase in propene oxidation whilst CuO
11
leads to combustion. In addition, copper catalysts are used in the
†
Electronic supplementary information (ESI) available. See DOI: 10.1039/
13
c3cy00254c
total oxidation of organic compounds – including propene – and
This journal is c The Royal Society of Chemistry 2013
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
the selective reduction of nitrogen oxides using propene as
Impregnation synthesis. The pore volume of the silica was
14
reducing agent. Gold catalysts, meanwhile have been shown to measured by adding water until the incipient wetness point was
15
be active for the selective oxidation of propene to propene oxide.
reached, and following the change in mass during the addition.
Here, a catalyst containing gold nanoparticles supported on A solution of copper nitrate hemipentahydrate (0.89 g, 38 mmol)
reducible oxide supports such as TiO , titanosilicates or iron oxide and tetrachloroauric acid (1.83 g solution, 38 mmol) was prepared
2
1
6
is preferred as the interfacial site is thought to be important.
using the appropriate volume of water to fill the pores of the
In contrast to copper, gold catalysts for propene oxidation are support and to give the desired concentration and ratio of
typically used with hydrogen as a co-reductant, forming a metals in the final product. The impregnated material was
hydroperoxy-type intermediate.
dried at 105 1C for three hours.
CuAu catalysts are known to be active for the selective oxidation
Deposition synthesis. In this method, the copper and gold
of propene, and the structure of the catalyst and reaction conditions are added sequentially, the copper being added first. A solution
determine the spectrum of products and propene conversion of basic copper carbonate (0.76 g, 3.4 mmol salt, 6.9 mmol Cu)
4
observed. Sinfelt and Barnett used a coimpregnated CuAu/SiO
and ammonium carbonate (7.5 g, 85 mmol) was prepared in a
2
catalyst to oxidise propene to acrolein at temperatures of 265–305 1C mixture of ammonium hydroxide solution (85 ml) and water
with O as oxidant. They found that reduction of the catalyst with (75 ml). To this was added the silica support (36 g), and the
2
hydrogen followed by high temperature calcination gave a material suspension heated to boiling. Ammonia was removed by distilla-
which could convert up to 40% of the propene present with tion causing the copper to deposit onto the support. The product
acrolein selectivities of 50–70%. We also studied impregnated of the reaction was collected by filtration, washed with water and
3
CuAu/SiO catalysts and found that hydrogen co-feeding generally dried at 105 1C for three hours. The dried Cu/SiO (10 g) was
2
2
increased activity. The performance of reduced and reduced- re-suspended in water (400 ml) at 65 1C and gold was added by
calcined catalysts was compared; reduced catalysts contained a deposition precipitation using sodium hydroxide (0.1 M solution)
number of different active sites and gave a mixed selectivity profile, as base and tetrachloroauric acid (0.89 g solution, 1.9 mmol) as
whilst reduced-calcined catalysts were generally more selective to gold precursor. The pH of the reaction was maintained around
acrolein. Llorca et al. have studied CuAu/TiO
2
catalysts prepared nine by appropriate addition rates of gold solution and base. The
O as oxidant in their reaction product was isolated by filtration, washed with water and
. At 300 1C, a range of products then dried at 105 1C for three hours.
were observed: propene oxide, propanal, acetone, acrolein and CO2. Hydrogen reduction. Materials were reduced in 5% H /N in a
The catalysts were shown by XRD and TEM to contain significant tube furnace. The temperature was increased to 40 1C at 2 1C min
quantities of CuAu alloy phases. Catalysts which contained pre- and held for 30 minutes to purge the tube. It was then further
5
using coimpregnated precursors with N
studies rather than O or H /O
2
2
2
2
2
2
À1
6
À1
formed thiol-stabilised CuAu nanoparticles were more active and increased to 315 1C at 10 1C min , holding at 315 1C for two
gave up to 40% selectivity to propene oxide, although the other hours then cooling to room temperature.
products were still observed. There is therefore a consensus that
catalysts which contain CuAu alloys are not selective to a single by co-impregnation was impregnated for a second time with an
product in propene oxidation. Au/CuO/Al catalysts have been aqueous solution of sodium borohydride. The reaction between the
found to be active for the total oxidation of propene to carbon catalyst precursor and the sodium borohydride solution was strongly
Sodium borohydride reduction. The dried precursor prepared
2 3
O
17
dioxide and water at around 200 1C using an excess of oxygen exothermic. The material was aged for one hour before being
propene : oxygen = 1 : 9). The catalysts were reduced at 300 1C prior washed with water, filtered and dried at 105 1C for three hours.
to catalysis, although alloying of gold and copper was not observed. Direct calcination. The impregnated catalyst was calcined
(
In this paper we explore the impact of catalyst synthesis directly without reduction. This was performed in a static air
À1
method – either by impregnation using copper nitrate and furnace by heating to 400 1C at 10 1C min , holding for two hours
tetrachloroauric acid, or by a two-step deposition method – and cooling to room temperature.
and the effect of different activating treatments – reduction
High temperature calcination. Selected reduced catalysts
with hydrogen or NaBH solution followed optionally by a high were subsequently calcined at high temperature. The calcination
4
temperature calcination – on the structure of the catalyst was performed in a static air furnace. The temperature was
À1
produced and the performance in the selective oxidation of increased to 675 1C at 10 1C min , holding at 675 1C for sixteen
propene using a mixture of H
2
and O
2
as oxidant.
hours then cooling to room temperature.
Characterisation
Experimental
Catalyst preparation
The metal contents of the catalysts were determined by Inductively-
coupled plasma electronic spectroscopy (ICP-ES) following
Materials. Copper nitrate hemipentahydrate, sodium boro- digestion in aqua regia. The samples were analysed using a
hydride, basic copper carbonate, sodium hydroxide, ammonium Perkin Elmer Optima instrument. XRD patterns were recorded
carbonate (Alfa Aesar), ammonium hydroxide solution (35%, on a Bruker D8 instrument between 10 and 1301 2Y using Ni
Fisher) tetrachloroauric acid (41.63% Au, Johnson Matthey) and filtered Cu radiation. Crystallite sizes and lattice parameters
silica (Grace P432, reference SP550-SP10022, BET surface area were determined by Rietveld refinement using Bruker Topas V4
2
À1
3
40 m g ) were all used as received.
analysis software. Lattice parameter data was taken from the
Catal. Sci. Technol.
This journal is c The Royal Society of Chemistry 2013

View Article Online
Catalysis Science & Technology
Paper
ICDD Database, PDF-4 release 2012. The lattice parameters for The catalyst (0.2 g) was heated in a flow of reactant gases with a
À1
copper (PDF 00-004-0836) and gold (00-004-0784) were 3.615 Å total space velocity of 22500 h . The gas mixture used was
20
and 4.079 Å respectively. The lattice parameters for the mono- 1:1:1:7 H :O :C H :He. The high space velocity used makes
2
2
3 6
clinic copper oxide (tenorite, PDF 00-048-1548) were a = 4.684 Å, the propene conversion values reported low. Reactions were started
b = 3.425 Å, c = 5.129 Å and b = 99.471. BET analyses were at 200 1C and increased in steps of 20 1C to a final temperature
performed using a five-point method on a Quantachrome of 300 1C or 320 1C before being cooled back to 200 1C in 20 1C
Autosorb instrument. TEM images were acquired using a steps. The reaction was dwelled at each temperature for
Tecnai-F20 instrument. Powder samples were set in resin and thirty minutes to allow steady-state data to be recorded. The
sectioned for analysis, or the powder was dusted onto a holey hysteresis in reaction performance observed with temperature
carbon grid. Nickel TEM grids were used to avoid interference allowed changes in the catalyst during use to be assessed.
with the copper in the samples. Particle size was determined by Changes in activity and selectivity can be ascribed to different
an image contrast method. Diffuse Reflectance Visible Spectro- effects (sintering, poisoning, dealloying etc.) and so this is a
scopy was performed on solid samples using an Agilent Cary sensitive probe to the structure of the catalyst.
Series Spectrometer with a Praying Mantis attachment. Samples
were diluted with barium sulphate to increase the signal to
noise ratio. The temperature-programmed reduction study in
Results
2
Fig. S7 (ESI†) was performed using a 5% Cu/SiO catalyst A series of seven catalysts were synthesised as described above.
containing copper(II) oxide. The catalyst was prepared by inci- Their characterisation data are presented in Tables 1–5 and
pient wetness impregnation using copper nitrate and was Fig. 1–6. Their catalytic performance is summarised in Table 6.
calcined in air at 400 1C prior to the experiment. The TPR Further details on the catalytic performance and XAS data
spectrum was obtained using an Altamira AMI-200 catalyst analysis is available in the ESI.†
À1
characterisation unit. The gas used was 30 ml min of 10%
À1
Impregnated catalyst, calcined only
H
2
/Ar and the heating rate was 10 1C min
.
X-Ray Absorption Spectroscopy was performed on the SAMBA Calcination of the precursor leads to a material with very little
beamline at Synchrotron Soleil, Paris, France. Data were collected interaction between copper and gold. The copper is present as
at both the Cu K (8900–13 400 eV) and Au L3 (11 165–13 400 eV) copper(II) oxide particles on the silica support, whilst the gold is
18
edges in QuickEXAFS mode. Each scan lasted for 1 s. Typically present in large (micron-sized) clusters which are only weakly
around 200 spectra were combined to give good quality data for bound to the surface (Fig. 5). XAS analysis confirms the presence
analysis. Spectra were referenced to Cu foil or Au foil. The data were of gold metal and copper(II) species (Fig. 2–4) with values close to
19
analysed using the Athena software for data handling and XANES bulk materials (Table 4). The catalyst is poorly active and produces
fitting. XANES spectra were fitted from À50 eV to +100 eV relative to a mixture of acrolein, carbon dioxide and propene oxide (Table 6).
the edge energy using normalised data. The fit quality values were
Impregnated catalyst, hydrogen reduction
used to rank the fits generated from different combinations of
reference materials. Copper(II) oxide, copper(I) oxide, copper(II) Reducing the impregnated catalyst precursor with H (315 1C,
2
nitrate hemipentahydrate and copper foil were used as standards 2 h) gives a material with a stronger interaction between copper
at the copper edge, whilst sodium gold(III) chloride, sodium gold(I) and gold, as indicated by the shift from 600 nm to 620 nm in the
thiosulphate and gold foil were the standards used at the gold edge. plasmon band in DR-visible spectroscopy (Fig. 6) compared with
All of these standards were used in XANES fitting, although some that of the calcined material. It is also accompanied by
were found not to be relevant. EXAFS data analysis was performed some quenching of the peak. XRD (Fig. 1) also indicates that
19
using Artemis in multi-edge mode fitting both Cu K and Au L3 CuAu species are present, but they are poorly defined and more
edge data in the k-range from 2.5 to 12.
than one environment may exist. EXAFS (Fig. 4 and Table 4),
on the other hand, identifies the presence of copper–copper
interactions as well as those of copper–oxygen, copper–gold and
Catalyst testing
Propene oxidation was performed using a flow microreactor and gold–gold. TEM (Fig. 5) detected large particles in the sample
the products were analysed using on-line gas chromatography. which contained both copper and gold. These will also be
Table 1 Summary of the catalysts synthesised and the copper and gold species present
Phases observed
Particle size
Preparation method
Impregnation
Pretreatment
Cu(II) oxide
Cu(I) oxide
Cu metal
Au metal
CuAu alloy
range
Calcination
Reduction in H
Reduction in H
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
Large
Large
Large
Large
Large
Small
Small
2
ꢀ
ꢀ
2
and calcinations
NaBH
NaBH
4
reduction
reduction and calcination
4
Deposition
Reduction in H
Reduction in H
2
ꢀ
2
and calcination
This journal is c The Royal Society of Chemistry 2013
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
Table 2 ICP analysis, XRD average crystallite size and TEM determined particle size
ICP data
Preparation
XRD average crystallite TEM particle size/nm
method
Pretreatment
Cu/wt% Au/wt% Cu/Au molar ratio size/nm (phase)
(particles counted)
a
b
b
Impregnation Calcination
1.1
1.35
4.8
0.71
1.28
54 (Au)
14 (CuAu(1))
Reduction in H
Reduction in H
2
2
3.28
9
(CuAu(2))
54 (Au)
5 (Au)
and calcination
1.22
0.63
3.98
2.98
3.39
3.77
3.97
0.95
0.66
0.60
1.01
0.79
7 (184)
24 (187)
24 (35)
8 (191)
6 (39)
NaBH
NaBH
Reduction in H
Reduction in H
4
reduction
reduction and calcination 0.66
4
11 (Au)
Deposition
2
1.23
1.01
o2 (CuAu)
2
and calcination
15 (Au)
69 (CuO)
a
b
Heterogeneity observed in these samples, the ICP value presented is an average of a number of replicates. Not determined.
Table 3 XRD characterisation of catalysts
Phases observed – crystallite size/nm (lattice parameter/Å)
c
Preparation method
CuAu/SiO catalysts
Pretreatment
Cu(II) oxide
Au metal
CuAu alloy
2
Impregnation
Calcination
Reduction in H
55 (4.07935(7))
2
2
14 (4.0460)
b
9
(4.0069)
Reduction in H
and calcination
20
a = 4.674(5)
54 (4.07954(4))
(
b = 3.427(3)
c = 5.139(6)
b = 99.54(7)1)
NaBH
NaBH
Reduction in H
Reduction in H
4
reduction
reduction and calcination
5 (4.0752)
a
4
11 (4.0791)
o2 (4.0792)
16 (4.0794(1))
Deposition
2
2
and calcination
50
(
a = 4.689(2)
b = 3.427(3)
c = 5.139(6)
b = 99.54(7)1)
Cu/SiO
Impregnation
2
catalyst
Calcination
Calcination
19
(
a = 4.689(1)
b = 3.431(1)
c = 5.134(1)
b = 99.46(1)1)
2
Au/SiO catalyst
Impregnation
39 (4.0815)
4.079
Literature values
a = 4.684
b = 3.425
c = 5.129
b = 99.471
a
b
c
Observed at low levels, no reliable crystallite size or lattice parameter data could be obtained. Two distinct phases observed. Tenorite has a
monoclinic structure and hence is described by three lattice parameters and one angle (b). In this case a = g = 901.
detected by XRD and EXAFS but, at particle sizes of 100 nm and removal from the tube furnace in our study. Meitzner et al. also
above, they are unlikely to impact significantly on catalysis. TEM report the ratio of Cu–Cu and Cu–Au coordination at the
also observes smaller particles on the catalyst (Fig. 5) but these Cu edge as 2.7. In our study we find this ratio to be only 0.45.
are poorly defined and it is difficult to completely understand If the Cu–O interactions are considered to come from the
their nature.
oxidation of Cu–Cu interactions, then this increases to 1.9
The EXAFS results reported here are similar to those of (assuming that one Cu–Cu interaction yields two Cu–O inter-
2
1
Meitzner et al. Both studies observe Cu–Au and Cu–Cu inter- actions). At the gold edge, the reported ratio of Au–Cu to Au–Au
actions. The main difference is that in our sample we also is 0.41, compared with 0.28 in this study. Their Cu–Au bond
observe Cu–O interactions. This may result from either the length (2.68 Æ 0.02 Å) is very close to the one reported in this
storage of the sample in air or the more rapid passivation on work, 2.71 Æ 0.02 Å.
Catal. Sci. Technol.
This journal is c The Royal Society of Chemistry 2013

View Article Online
Catalysis Science & Technology
Paper
2
Table 4 (A) Cu K edge EXAFS fit parameters. R = bond distance (Å), N = coordination number, s = Debye–Waller factor. The measured fit index values were from 0.01 to
2
0
.06. (B) Au L3 edge EXAFS fit parameters. R = bond distance (Å), N = coordination number, s = Debye–Waller factor. The measured fit index values were 0.005 to 0.06
(A)
Cu–O
Cu–Cu
Cu–Au
2
2
2
2
2
2
Preparation method
Impregnation
Pretreatment
R/Å
N
s /Å
R/Å
N
s /Å
R/Å
N
s /Å
Calcination
Reduction in H
Reduction in H
1.95
1.92
1.97
1.97
1.95
1.93
1.96
4.3
2.2
3.1
4.3
3.3
2.5
3.1
0.008
0.008
0.005
0.008
0.006
0.006
0.006
—
2.55
—
—
—
—
—
—
1.4
—
—
—
—
—
—
0.006
—
—
—
—
—
—
2.75
—
—
—
2.68
—
—
3.1
—
—
—
0.3
—
—
0.014
—
—
—
0.001
—
2
2
and calcination
NaBH
NaBH
4
reduction
reduction and calcination
4
Deposition
Reduction in H
Reduction in H
2
2
and calcination
(B)
Au–Cu
Au–Au
2
2
2
2
Preparation method
Impregnation
Pretreatment
Calcination
Reduction in H
Reduction in H
R/Å
N
s /Å
R/Å
N
s /Å
—
2.75
—
—
—
2.68
—
—
2.0
—
—
—
0.3
—
—
0.014
—
—
—
0.001
—
2.88
2.86
2.88
2.87
2.87
2.84
2.88
7.8
7.2
11.0
9.6
9.6
6.7
0.008
0.008
0.008
0.009
0.009
0.009
0.008
2
2
and calcination
NaBH
NaBH
4
reduction
reduction and calcination
4
Deposition
Reduction in H
Reduction in H
2
2
and calcination
10.3
Table 5 (A) Best fits of XANES data from Cu K edge spectra. The standards used in fitting were copper foil for Cu(0), Cu
.5H O (B) were applied. ‘—‘ means not included in fit. Fit quality values are reported in Table S2 (ESI). (B) Best fits of XANES data from Au L3 edge spectra. The standards
used in fitting were gold foil for Au(0), Na
2
O for Cu(I) and for Cu(II) both CuO (A) and Cu(NO
3
)
2
Á
2
2
3
Au(S
2
O
)
3 2
for Au(I) and NaAuCl
4
for Au(III). ‘—’ means not included in fit. Fit quality values are reported in Table S3 (ESI)
(A)
Fit/%
Preparation method
Impregnation
Pretreatment
Cu(II) (A)
Cu(II) (B)
Cu(I)
Cu(0)
Calcination
Reduction in H
Reduction in H
51.9
36.1
61.6
50.8
54.6
41.7
62.3
48.1
53.2
38.4
49.2
45.4
22.5
37.7
—
8.5
—
—
—
17.1
—
—
32.2
—
—
—
18.7
—
2
2
and calcination
NaBH
NaBH
4
reduction
reduction and calcination
4
Deposition
Reduction in H
Reduction in H
2
2
and calcination
(B)
Fit/%
Au(0)
Preparation method
Impregnation
Pretreatment
Calcination
Reduction in H
Reduction in H
Au(I)
Au(III)
100
60.4
100
100
100
74.4
100
—
39.6
—
—
—
25.6
—
—
—
—
—
—
—
—
2
2
and calcination
NaBH
NaBH
4
reduction
reduction and calcination
4
Deposition
Reduction in H
Reduction in H
2
2
and calcination
This catalyst also demonstrates poor catalytic activity, although impregnated catalyst, the selectivity to acrolein is significantly
it is approximately twice as active as the calcined impregnated higher, and other oxygenated products are no longer produced
catalyst. The catalyst produces a mixture of products which (Table 6). XRD (Fig. 1) and EXAFS (Fig. 4 and Table 4) show that
suggests that multiple active sites are present (Table 6).
there are no crystalline CuAu alloy species in this catalyst,
rather the copper is present as copper(II) oxide and the gold
as gold metal. There is some evidence from TEM (Fig. 5)
Impregnated catalyst, hydrogen reduced and calcined
4
This is the material described by Sinfelt and Barnett and that the copper and gold species are in close proximity to
it performs much better than the reduced or calcined impreg- each other, unlike in the calcined material where they are
nated materials. Although it is less active than the reduced physically remote.
This journal is c The Royal Society of Chemistry 2013
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
Fig. 1 (A) XRD patterns of CuAu/SiO
2
prepared by coimpregnation and (a) calcined; (b) reduced in H
prepared by deposition and (f) reduced in H
the expected positions of reflections due to Au metal, and the blue bars to the expected positions of peaks due to CuO (tenorite). (B) Expansion of XRD patterns of
CuAu/SiO prepared by coimpregnation and (a) calcined; (b) reduced in H ; (c) reduced in H and calcined; (d) reduced using NaBH solution; (e) reduced using NaBH
solution and calcined; CuAu/SiO prepared by deposition and (f) reduced in H ; (g) reduced in H and calcined; the gold bars refer to the expected positions of
reflections due to Au metal, and the blue bars to the expected positions of peaks due to CuO (tenorite).
2
; (c) reduced in H
2
and calcined; (d) reduced using NaBH
4
solution; (e) reduced using NaBH
4
solution and calcined; CuAu/SiO
2
2
; (g) reduced in H
2
and calcined; the gold bars refer to
2
2
2
4
4
2
2
2
(
Fig. 4 and Table 4) nor XRD (Fig. 1) detect a CuAu phase,
Impregnated catalyst, sodium borohydride reduction
Reduction of the catalyst with NaBH was highly exothermic. It
is also expected to be a more rapid reduction than using
instead showing that gold is present as metallic nanoparticles
and copper as copper(II) oxide. Given that the structure of this
material is quite different to that reduced using hydrogen, it is
4
hydrogen. TEM shows that the particle size is much smaller not surprising that the catalytic performance is also different.
than for the hydrogen-reduced material (Fig. 5), but never- The only products observed are acrolein and carbon dioxide, and
theless the catalyst activity is similar. The NaBH reduction the acrolein selectivity is good at 84% (Table 6). The product
4
does not lead to the formation of an appreciable amount of distribution observed is very similar to that of the reduced and
CuAu alloy phase. EDX suggests that the particles contain both calcined material described above, which is a consequence of the
copper and gold, but are very gold rich. However, neither EXAFS catalysts containing the same metal species.
Catal. Sci. Technol.
This journal is c The Royal Society of Chemistry 2013

View Article Online
Catalysis Science & Technology
Paper
Fig. 2 (A) Cu-K edge XANES data for CuAu/SiO
2
prepared by coimpregnation and (a) calcined; (b) reduced in H
prepared by deposition and (f) reduced in H
O. (B) Au-L3 edge XANES data for CuAu/SiO prepared by coimpregnation and (a) calcined; (b) reduced in H
solution; (e) reduced using NaBH solution and calcined; CuAu/SiO prepared by deposition and (f) reduced in H
Au(S and (j) NaAuCl
2
; (c) reduced in H
2
and calcined; (d) reduced using
NaBH
i) Cu
and calcined; (d) reduced using NaBH
in H and calcined; (h) Au foil; (i) Na
4
solution; (e) reduced using NaBH
4
solution and calcined; CuAu/SiO
2
2
; (g) reduced in H
2
and calcined; (h) Cu foil;
; (c) reduced in H
; (g) reduced
(
2
O; (j) CuO and (k) Cu(NO
3
)
2
Á2.5H
2
2
2
2
4
4
2
2
2
3
2
O
3
)
2
4
.
Sodium borohydride reduction has previously be applied to TEM (Fig. 5) does not show the large copper–gold particles
7
CuAu catalyst preparation. In this case the metals were added found in the impregnated sample. DR-visible spectroscopy
4
sequentially with NaBH treatment after each addition, followed (Fig. 6) suggests that a CuAu phase is present in at least part
by calcination at 250 1C. As with our experience, no CuAu alloy of the sample, as the plasmon peak is shifted from the value
phase was observed on the catalyst.
expected for monometallic gold, and indeed EXAFS (Fig. 4 and
Table 4) also finds a small amount of gold alloy (Cu–Au
coordination number 0.3). However, most of the gold is present
as gold metal and the copper as copper(II) species.
Impregnated catalyst, sodium borohydride reduced and
calcined
Calcination of the NaBH -reduced material sinters the gold and
This material performs well as a catalyst, with a much higher
4
copper oxide particles, as shown by both XRD (Fig. 1) and activity than any of the impregnated catalysts (Table 6). The
EXAFS (Fig. 4 and Table 4). This makes the catalyst less active; selectivity to acrolein is also high, indicating that only one
indeed, it is the least active of all the catalysts reported in major type of active site is present in the material.
this paper, although the selectivity to acrolein is high (Table 6).
TEM shows that the small particles on the catalyst have an
interesting structure (Fig. 5) where the gold and copper oxide
Deposition catalyst, hydrogen reduced and calcined
particles are next to each other or even in contact on the High temperature calcination of the hydrogen-reduced material
support, suggesting that they have been dealloyed and sintered. leads to the destruction of all CuAu alloy species on the catalyst,
This is not observed in the NaBH -reduced material, so is likely as shown by DR-visible spectroscopy (Fig. 6), XRD (Fig. 1) and
4
to have occurred during the calcination step.
XAS (Fig. 2–4 and Table 4). The gold and copper oxide particles
This catalyst is the least active of all the materials studied, are still quite small (5–10 nm by TEM (Fig. 5)) despite the high
although in common with other reduced-calcined catalysts it temperature treatment. Unlike the impregnated catalysts, high
shows good selectivity to acrolein, with no oxygenates being temperature calcination of the reduced catalyst increases its
produced (Table 6).
activity markedly; this is by far the most active of the catalysts
reported in this paper. Selectivity to acrolein is also good.
The performance of this material compared with the impreg-
Deposition catalyst, hydrogen reduction
The first important difference between the deposition and nated catalyst hydrogen reduced and calcined is presented in
impregnation routes is in dispersion. Here there are many Fig. 7. Here, at 320 1C the propene conversion is just over 10%
fewer large particles, so the peaks in the XRD spectrum (Fig. 1) and the selectivity to acrolein 89% with small amounts of
are broad and weak, in contrast to the impregnated catalyst. ethanal and carbon dioxide observed as by-products.
This journal is c The Royal Society of Chemistry 2013
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
3
Fig. 3 (A) Cu-K edge EXAFS data (as k ) for CuAu/SiO
2
prepared by coimpregnation and (a) calcined; (b) reduced in H
solution and calcined; CuAu/SiO prepared by deposition and (f) reduced in H
prepared by coimpregnation and (a) calcined; (b) reduced in H ; (c) reduced in H
solution and calcined; CuAu/SiO prepared by deposition and (f) reduced in H ; (g) reduced in H
2
2
; (c) reduced in H
; (g) reduced in H
and calcined; (d) reduced using
and calcined.
2
and calcined; (d) reduced
using NaBH
4
solution; (e) reduced using NaBH
4
2
2
2
and calcined.
3
(
B) Au-L3 edge EXAFS data (as k ) for CuAu/SiO
2
2
2
NaBH
4
solution; (e) reduced using NaBH
4
2
2
standard, copper(II) nitrate hemipentahydrate. The materials
which had been calcined or reduced and calcined could be
fitted well with a mixture of copper(II) oxide and copper(II)
Discussion
XAS data analysis
XAS analysis is an excellent technique for understanding the nitrate hemipentahydrate with no contribution from either
structure of bimetallic catalysts, which tend to be poorly crystal- copper(I) oxide or copper metal foil (Table 5). The materials
line and contain many metal species: oxidised and reduced, do not of course contain copper nitrate as they have all been
bimetallic and monometallic. We have analysed both the thermally treated at temperatures where copper nitrate would
XANES and EXAFS parts of the data to help understand the be decomposed, or reacted with sodium borohydride which
structure of the catalysts tested.
would have the same effect. Copper(II) oxide and copper(II)
Initial attempts to fit the Cu edge XANES data were made nitrate have different co-ordination environments – copper(II)
using three standards: CuO, Cu O and Cu metal foil. However, oxide has four nearest neighbour oxygen atoms at 1.941 Å
2
2
2
for most of the catalysts this did not give a satisfactory fit distance in an approximately square planar arrangement
see ESI†). To improve the fit required the addition of a new whilst copper(II) nitrate contains a distorted octahedron of
(
Catal. Sci. Technol.
This journal is c The Royal Society of Chemistry 2013

View Article Online
Catalysis Science & Technology
Paper
Fig. 4 (A) Fourier-transform Au-L3 edge EXAFS data for CuAu/SiO
2
prepared by coimpregnation and (a) calcined; (b) reduced in H
prepared by deposition and (f) reduced in H
prepared by coimpregnation and (a) calcined; (b) reduced in H ; (c) reduced in H
solution and calcined; CuAu/SiO prepared by deposition and (f) reduced in H ; (g) reduced in H
2
2
; (c) reduced in H
; (g) reduced in H
and calcined;
and
2
and calcined;
(
d) reduced using NaBH
calcined. (B) Fourier-transform Au-L3 edge EXAFS data for CuAu/SiO
d) reduced using NaBH solution; (e) reduced using NaBH
calcined. Data is not phase corrected.
4
solution; (e) reduced using NaBH
4
solution and calcined; CuAu/SiO
2
2
2
and
2
2
2
(
4
4
2
2
Cu–O distances due to the Jahn–Teller effect: four shorter environment which is intermediate between CuO (square
bonds of 1.959 Å and two longer bonds at 1.99 Å. A seventh planar) and Cu(NO O, which is a distorted octahedron.
Á2.5H
Cu–O interaction with a neighbouring unit cell at 2.39 Å This is supported by the absence of crystalline CuO from the
3
)
2
2
2
3
completes the coordination environment in the solid state.
XRD of many of the samples. It is interesting in light of the
1
1
One interpretation of the Cu edge XANES data is the character- reported good selectivity of Cu O in propene oxidation that
2
isation of ‘bulk-like’ copper oxide, represented by CuO, and no evidence for this oxidation state was found.
‘
well-dispersed’ CuO species, represented by Cu(NO ) Á2.5H O.
Fitting the two hydrogen-reduced samples was hampered by
3
2
2
The ratio of these two species is similar across all the calcined the lack of a suitable standard material for CuAu alloy phases.
and reduced-calcined samples, with the CuO contribution The impregnated catalyst reduced with hydrogen was fitted
varying from 50% to 63%. An alternative explanation for the with a combination of four standards whilst the deposition
Cu edge XANES data is that the spectrum shows a single copper catalyst reduced with hydrogen was fitted with three (Table 5).
This journal is c The Royal Society of Chemistry 2013
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
Fig. 6 Diffuse Reflectance Visible Spectra of CuAu/SiO
tion and (a) calcined; (b) reduced in H ; (c) reduced in H
using NaBH solution; (e) reduced using NaBH
prepared by deposition and (f) reduced in H ; (g) reduced in H
h) a 5% Au/SiO reference material prepared by HAuCl impregnation.
2
prepared by coimpregna-
2
2
and calcined; (d) reduced
4
4
solution and calcined; CuAu/SiO
and calcined; and
2
2
2
(
2
4
The Au L3 edge XANES data, meanwhile, was initially fitted using
three standards, one for each of the commonly-observed gold
oxidation states. Gold metal foil was used to represent Au(0),
3 2 3 2 4
Na Au(S O ) for Au(I) and NaAuCl for Au(III). Whilst all the calcined
and reduced-calcined materials could be fitted with a combination
of the three standards, fitting using gold metal only is considered
the best solution as the quality of the fits were similar in both cases
(see ESI†). The two hydrogen-reduced samples were again different
and were best fitted using a combination of gold metal foil and
24
sodium gold thiosulphate. Again this does not suggest the
presence of sulphur in the materials, but is probably a mimic for
the unreferenced CuAu phase. Fitting using a Au(I) standard may
also show that the gold is slightly electron deficient in these
materials, with electron density being transferred to copper.
Consideration of the EXAFS data confirms the XANES results
above. All the calcined and reduced-calcined catalysts are found
to contain Cu–O and Au–Au bonding only, with no Cu–Au
interaction observed. The Cu–O bond lengths are in the range
Fig. 5 TEM data for CuAu/SiO
2
prepared by coimpregnation and (a) calcined;
and calcined; (d) reduced using NaBH
solution and calcined; CuAu/SiO prepared by
; (g) reduced in H and calcined.
(
b) reduced in H
solution; (e) reduced using NaBH
deposition and (f) reduced in H
2
; (c) reduced in H
2
4
4
2
2
2
1
.92–1.97 Å which is typical for Cu(II)–oxygen interactions such
2
2
25
23
as CuO (1.94 Å), copper oxalate (1.98 Å) or Cu(II) nitrate
However, these fits may not be meaningful in the absence of a (1.96 Å, 1.99 Å). The Cu–O distance in Cu
2
O is significantly
2
6
copper–gold reference material.
shorter (1.83 Å) and there is no evidence for it being present
Catal. Sci. Technol.
This journal is c The Royal Society of Chemistry 2013

View Article Online
Catalysis Science & Technology
Paper
Table 6 Catalyst test data: performance summary at 300 1C
Selectivity/%
Acrolein
a
Preparation method
CuAu/SiO catalysts
Pretreatment
Propene conversion/%
Carbon dioxide
Oxygenates
2
Impregnation
Calcination
Reduction in H
Reduction in H
0.14
0.28
0.20
0.26
0.20
2.0
45
41
83
84
82
93
84
21
36
17
16
18
6
34
23
0
0
0
1
9
2
2
and calcination
NaBH
NaBH
4
reduction
reduction and calcination
4
Deposition
Reduction in H
Reduction in H
2
2
and calcination
5.6
7
Cu/SiO
Impregnation
2
catalyst
Calcination
0.22
0.09
85
74
15
26
0
0
2
Au/SiO catalyst
Impregnation
Calcination
a
Acetone, propene oxide and ethanal.
2
Fig. 7 Catalytic performance of CuAu/SiO prepared by deposition followed by a reduction–calcination activation step. (solid line, square markers) Compared with
catalysts prepared by impregnation (dashed lines) calcined only (circles) and reduced and calcined (triangles).
in these materials. The shortest Cu–O distances belong to the average composition of Au Cu for the impregnated catalyst
6
3
27
two hydrogen-reduced materials (1.92 and 1.93 Å) which could and Au Cu for the deposition catalyst. XRD, for comparison,
3
0
70
suggest a contribution from Cu O. However, the contribution determined two CuAu phases in the impregnated catalyst:
2
7
would be very small as the bond distances are still close to that Au93Cu and Au84Cu16 whilst in the deposition catalyst the lattice
typical for CuO. The Au–Au distances are in the range 2.84–2.88 Å parameter was the same as for monometallic gold catalysts
27
which are typical of gold metal (2.86 Å). The two hydrogen-reduced (Table 3). The difference between XRD and EXAFS results
samples contain a mixture of interatomic interactions: Cu–O, Cu–Au suggests that for the impregnated catalyst, the most crystalline
and Au–Au whilst the catalyst prepared by impregnation also has part of the sample (the largest particles) is gold rich whilst there
21
Cu–Cu co-ordination, in agreement with Meitzner et al. The Cu–Au are also copper-rich smaller nanoparticles. For the deposition
distances calculated for the two hydrogen reduced materials are sample, the CuAu phase is likely to be found in the smaller
different (2.75 Å for the impregnated catalyst and 2.65 Å for the nanoparticles as it is not clearly observed by XRD.
deposition catalyst) but both lie between the metal–metal distances
28
27
Impact of preparation method
for Cu–Cu (2.56 Å) and Au–Au (2.86 Å). The difference is likely to
reflect differences in the composition of the alloys. Taking a Vegard’s Changing the catalyst preparation method from impregnation
29
law approach to the bond distances measured by EXAFS gives an to deposition has been shown to have a significant impact on
This journal is c The Royal Society of Chemistry 2013
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
Table 7 Illustrative examples of copper and gold catalysts in propene oxidation
catÀ1 À1
Catalyst
Reaction conditions GHSV/l kg
h
T/1C Propene conversion/% Major product (selectivity/%)
Ref.
0
0
0
1
1
.06% CuO/SiO
2
C
C
C
C
3
H
3
H
3
H
3
H
3
H
3
H
3
H
3
H
6
6
6
6
6
6
6
6
/O
2
= 1/5
10.2
4000
4000
4000
475 40
50 0.5
Acrolein (10)
Propene oxide (95)
Propanal (74)
Acrolein, propanal, acetone, ethanal (NR) 33
Propene oxide (50)
Acrolein (35)
32
15
15
.67% Au/TiO
.2% Au/TS-1
2
/H
/H
/H
2
2
2
/O
/O
/O
2
2
2
= 1/1/1
= 1/1/1
= 1/1/1
200 0.6
>200 o1
225 0.3
360 3.3
225 0.1
350 12.5
% Au/SiO
% Cu/SiO
2
2
small particles C
/O
2
/O
2
/O
2
/O
2
/He = 1/1/18 30 000
/He = 1/1/18 30 000
/He = 1/1/18 30 000
/He = 1/1/18 30 000
10
10
10
10
C
larger particles C
C
5
% Cu/SiO
2
Acrolein (40)
Acrolein (40)
NR = not reported.
catalyst performance. For example, considering the hydrogen- crystalline CuO is identified at 361 2y (Fig. 1A) in the hydrogen-
reduced and calcined catalysts, at 300 1C the rate of propene reduced and calcined material, but is absent in the NaBH -reduced
4
conversion is increased by a factor of almost thirty. A lot of this material. Of note is the gold peaks in the hydrogen-reduced and
is likely to be a consequence of the smaller particle size of the calcined sample which are very intense, making it the most
catalysts prepared by deposition (around 5–10 nm) compared crystalline gold of all the samples investigated. However, this is
with those prepared by impregnation which can be much larger not detrimental to activity as the least active sample, the
(upwards of 20 nm) and which TEM shows to contain larger impregnated precursor reduced with sodium borohydride and
metal clusters (Fig. 5) as well as small particles. calcined, has broader XRD lines than the equivalent material
The other major impact of changing from an impregnation reduced using hydrogen, despite being only one-fifth as active.
to a deposition method is to change the speciation of the metals This shows that, as is often the case, not all the gold in the
3
1
on the catalyst. Comparing the hydrogen-reduced catalysts shows catalyst is relevant to catalysis.
that synthesis by deposition gives rise to only a small amount of
copper–gold alloy, whilst impregnation gives much more. How-
ever, given that the catalyst prepared by deposition is much more
active, it can be assumed that the impact of the smaller particle Probing the nature of the active site is a difficult task in
Comments on the nature of the active site
size dominates any effect of the copper–gold alloy content.
heterogeneous catalysis due to the large number of different
species present as we have shown in this catalyst system. In this
case it is made more difficult since both copper and gold
Impact of reducing agent
Comparison of hydrogen reduction with chemical reduction is catalysts are active in propene oxidation. For copper catalysis,
of interest as they are both industrially relevant and yet give quite Cu/SiO catalysts are known for the conversion of propene into
different results. Hydrogen reduction is usually performed at acrolein. This reaction is usually run without the addition of
2
3
2
elevated temperature (315 1C in this work) and is often used as hydrogen in the gas stream. We have previously reported the
it is ‘clean’, leaving no residues apart from water. Sodium effect of hydrogen co-feeding in propene oxidation for CuAu/
3
borohydride reduction, in comparison, is a low temperature SiO
2
catalysts. The presence of Cu
2
O on the catalyst is reported
1
1
process although it is highly exothermic and difficult to control to be beneficial, whilst CuO is reported to be responsible for
at larger scales. A washing step is also needed to remove the large complete combustion of propene to CO . Gold catalysts for
amounts of boron and sodium which are present in NaBH
propene oxidation are generally not supported on silica, with
In this study the two reduction methods gave quite different titania and titania-containing materials generally preferred.
2
4
.
3
3
products. Hydrogen reduction leads to large particles of CuAu Gold-on-silica catalysts have been reported to be poorly active,
alloy and a material which was relatively poorly active. Sodium certainly when compared to equivalent materials on other,
borohydride reduction led to smaller particles which were not reducible, supports. Gold catalysts also tend to make propene
alloyed and also some larger clusters, which appear to contain oxide in propene oxidation with hydrogen co-feeding. Copper
both copper and gold, on the surface of the catalyst particles and gold catalysts tend to be used at different temperatures.
(
(
Fig. 5). These could be caused by the rapid evolution of gases The best results with gold catalysts are reported at low temperatures
such as H or NO
) in the catalyst pores, ejecting solution and (o100 1C) whilst copper catalysts retain selectivity to acrolein
2
x
metal salts to the exterior of the particles (an effect which has even above 400 1C. The temperatures investigated in this
3
0
been observed before ).
study are therefore intermediate between the two. Illustrative
Interestingly, the structure of the borohydride-reduced examples of these catalysts are summarised in Table 7; this is
catalyst was very similar to the hydrogen-reduced and calcined not an exhaustive list.
material, giving a rapid, one-step synthesis of this material. In
Given the tendency for copper catalysts to make acrolein
both cases EXAFS analysis determined the presence of copper whilst gold catalysts make propene oxide on the oxidation of
oxide and gold metal on the catalyst. Their selectivity profiles are propene, it is reasonable to assume that the main active site
almost identical; however, the borohydride-reduced material in these catalysts is copper-based, since for the majority of
was more active. This is reflected in the XRD pattern where materials the major product is acrolein. This prompts two
Catal. Sci. Technol.
This journal is c The Royal Society of Chemistry 2013

View Article Online
Catalysis Science & Technology
Paper
further questions – what is the nature of copper in the active small particles. However, comparison of the data in Tables 2 and
site, and what is the role of the gold? 6 shows that there are other catalysts which contain small
The fresh catalysts were characterised using EXAFS, which particles which are much less active. Clearly other factors, such
determined the presence of oxidic copper phases in many of the as speciation, also have an impact. Also, for catalysts such as
materials. The range of bond lengths determined is 1.93–1.97 Å, these which contain a range of metal species, measuring an
which shows the presence of a CuO-like phase. The Cu–O accurate particle size is complicated by sample size and the
2
6
2
distance in Cu O is 1.83 Å which is significantly shorter than laborious nature of particle counting (TEM) and the presence of
2
2
the distance observed, which is more typical of CuO (1.94 Å ). large particles which do not participate in catalysis (XRD).
Also XANES fitting did not find a significant contribution of
Cu O to any of the catalysts. The XANES spectrum of Cu O has a
2 2
Effect of reaction temperature
distinctive, sharp pre-edge which is not observed in any of the We have studied the catalysts in propene oxidation in the
catalysts or precursors (Fig. 2A). Therefore the catalysts contain temperature range 200–300 1C following a strict temperature
significant amounts of copper(II) oxide as the active phase. protocol: the reaction was started at 200 1C and then increased
However, the performance of these catalysts is not the same as in steps of 20 1C to 300 1C and then reduced in steps back to
Cu/SiO
2
catalysts tested under the same regime – although 220 1C. This allows us to understand changes in catalyst
performance at 300 1C is similar (Table 5), at lower temperatures performance during catalysis (see ESI†). The stability of the
the copper-only catalyst produces significant amounts of oxygenates catalysts varies greatly; some barely change whilst others
(
ESI†), which is not observed with the CuAu catalysts. This suggests change markedly over the temperature range. These changes
that the presence of gold has an influence on the copper, but not by are seen in both selectivity and activity. As steady-state data
directly binding to it. This could be longer range interactions – such were recorded in each case, the changes must be caused by
as those suggested by the TEM images of the borohydride-reduced variations in the structure or performance of the active sites on
and calcined catalyst prepared by impregnation (Fig. 5e). However, the catalyst. Generally, the catalysts which had been reduced
TEM images of other samples do not have the same conclusive and calcined were the most stable over the temperature range,
evidence for this arrangement of copper and gold, and furthermore whilst the reduced-only materials tended to change more. The
this is the least active sample of all those studied. Another possibility stability of the reduced-calcined catalyst might be expected; a
is that synthesis via an alloy leads to a well dispersed copper oxide catalyst which has been calcined at 675 1C could be expected to
phase. This is not interacting directly with the gold but instead it is be stable at 300 1C. The reduced-only catalysts have not had this
dispersed onto the support. XRD shows that the copper oxide in high temperature treatment and therefore might be expected to
these materials is well-dispersed, even the high temperature calcined change more. Additionally, these materials have the greatest
3
materials only showing low intensity CuO reflections, when com- number of active sites and so the selectivity will change most
pared with the nearby gold metal reflection. Interestingly there is a with temperature. Indeed, the sodium borohydride-reduced
sharp reflection at 36.21 2y in the most active catalyst, close to the catalyst displayed the greatest change with temperature. The
main intensity reflection for CuO although slightly shifted to higher nature of these changes is likely to be complex, and include
2y. The origin of this peak is not clear.
effects such as sintering, carbon laydown, alloying or de-alloying.
As we conclude that these catalysts do not contain Cu(I), this This is the subject of a further study.
puts us at variance with the literature on copper-catalysed
Another general feature of the catalysts investigated is
propene oxidation, where Cu(I) is generally proposed as an active that selectivity to oxygenates tended to be observed at lower
1
1
species. However, as described above we find no evidence for temperature (close to 200 1C), whilst at higher temperature
Cu(I) in any of our analyses. It is, of course, possible that Cu(I) is (nearer 300 1C) acrolein and carbon dioxide were often the only
made during the reaction, by reduction of Cu(II) by either two products observed. This is consistent with the expected
propene or hydrogen. Temperature-programmed reduction behaviour of copper, gold and copper–gold catalysts (Table 7).
(
TPR) analysis of a CuO/SiO
2
catalyst showed that the metal At lower temperatures (r200 1C), propene oxide is produced
5,10,15
was reduced between 160–260 1C, with a very small fraction over all three systems. For gold catalysts, propanal is
reduced at 260–340 1C (Fig. S7, ESI†). Hence reduction of also produced. However, at higher temperatures (Z300 1C),
1
5
1
0
5
copper(II) at reaction temperatures is feasible.
acrolein is produced over copper and copper–gold catalysts.
catalyst, despite being poorly active, also made
The exception to this discussion is the catalyst prepared by Our Au/SiO
2
impregnation reduced with hydrogen which EXAFS shows to acrolein at temperatures above 260 1C (Table S1, ESI†). Making
contain Cu–Au and Cu–Cu phases as well as gold metal and acrolein requires a different part of the propene molecule to be
copper oxide. All of these phases are active catalysts in propene activated (CH
3
group) compared with making propene oxide
5
,6,9
oxidation
although the particle size is large, meaning that (C–C double bond). It is possible, therefore, that a different
catalytic activity is low. The higher number of active sites adsorption geometry is required in each case.
contributes to the high selectivity to oxygenates (in this case,
3
propene oxide), even at 300 1C.
Conclusions
Good dispersion of the metal particles on the catalyst support
(i.e. a small particle size) seems to be important in achieving This study has demonstrated that control of the properties of
good propene conversion: both of the most active catalysts have CuAu/SiO
2
catalysts through careful choice of synthesis method
This journal is c The Royal Society of Chemistry 2013
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
and activation treatment can be used to tune their catalytic 10 O. P. H. Vaughan, G. Kyriakou, N. Macleod, M. Tikhov and
performance. Catalysts prepared by a two-step sequential deposi-
R. M. Lambert, J. Catal., 2005, 236, 401–404.
tion method showed greater activity than those prepared by 11 J. B. Rietz and E. I. Solomon, J. Am. Chem. Soc., 1998, 120,
impregnation, which is a consequence of their smaller particle
11487–11488.
size and therefore greater reactive surface area. The majority of 12 N. Scotti, N. Ravasio, F. Zaccherra, R. Psaro and
the catalysts studied contain gold metal and well-dispersed copper
C. Evangelista, Chem. Commun., 2013, 49, 1957–1959.
oxide, and it is this latter phase which is thought to be the origin 13 P. M. Heyndrickx, J. W. Thybaut, H. Poelman, D. Poelman
of the high selectivity to acrolein observed at higher temperature.
and G. B. Marin, J. Catal., 2010, 272, 109–120.
The gold in the catalysts is only active at lower temperature and 14 R. Zhang, D. Shi, Y. Zhao, B. Chen, J. Xue, X. Liang and
generally makes oxygenated products such as ethanal or acetone.
Z. Lei, Catal. Today, 2011, 175, 26–33.
The role of gold in the catalysts seems to be to mediate the 15 B. S. Uphade, S. Tsubota, T. Hayashi and M. Haruta, Chem.
synthesis of the well-dispersed copper oxide rather than to inter-
Lett., 1998, 1277–1278.
act directly with the copper itself. Detailed catalytic studies 16 J. Chen, S. J. A. Halin, E. A. Pidko, M. W. G. M. Verhoeven,
showed that hysteresis is observed with some materials on cycling
the temperature between 200 1C and 300 1C, and this is the
subject of a further investigation.
D. M. P. Ferrandez, E. J. M. Hensen, J. C. Schouten and
T. A. Nijhuis, ChemCatChem, 2013, 5, 467–478.
17 A. C. Gluhoi, N. Bogdanchikova and B. E. Nieuwenhuys,
Catal. Today, 2006, 113, 178–181.
1
8 E. Fonda, A. Rochet, M. Ribbens, L. Barthe, S. Belin and
V. Briois, J. Synchrotron Radiat., 2012, 19, 417–424.
Acknowledgements
The authors wish to thank Lubomira Duhackova, Godson Nnorom- 19 B. Ravel and M. Newville, J. Synchrotron Radiat., 2005, 12,
Junior, James McNaught and Laura Stead for ICP analysis, Dr Kerry
537–541.
Simmance, Hoi Jobson, James McNaught and Dr Edward Bilb ´e 20 T. Hayashi, K. Tanaka and M. Haruta, J. Catal., 1998, 178,
for XRD analysis, Dr Jingshan Dong, Dr Gregory Goodlet and
566–575.
Dr Sarennah Longworth-Cook for TEM analysis and Dr Crina 21 G. Meitzner, G. H. Via, F. W. Lytle and J. H. Sinfelt, J. Chem.
Corbos and Dr Jose Antonio Lopez Sanchez for help collecting
Phys., 1985, 83, 4793–4799.
the XAS data. CLB is grateful to Johnson Matthey for a 22 H. Irie, K. Kamiya, T. Shibanuma, S. Mirura, D. A. Tryk,
PhD studentship. We acknowledge financial support from the
European Union via the FP7 ELISA project for travel and
sustenance for the collection of the XAS data.
T. Yukoyama and K. Hashimoto, J. Phys. Chem. C, 2009, 113,
10761–10766.
23 B. Morosin, Acta Crystallogr., 1970, B26, 1203–1208.
24 H. Ruben, A. Zalkin, M. O. Faltens and D. H. Templeton,
Inorg. Chem., 1974, 13, 1836–1839.
References
2
5 A. Micalowicz, J. J. Girerd and J. Goulon, Inorg. Chem., 1979,
18, 3004–3010.
1
2
3
C. L. Bracey, P. R. Ellis and G. J. Hutchings, Chem. Soc. Rev.,
009, 38, 2231–2243.
X. Liu, A. Wang, X. Wang, C.-Y. Mon and T. Zhang, Chem.
Commun., 2008, 487–489.
C. L. Bracey, A. F. Carley, J. K. Edwards, P. R. Ellis and
G. J. Hutchings, Catal. Sci. Technol., 2011, 1, 76–85.
J. H. Sinfelt and A. E. Barnett, U.S. Pat., 3989674, 1976.
R. J. Chiment ˜a o, F. Medina, J. L. G. Fierro, J. Llorca,
2
26 J. Lee, T. Yano, S. Shibata, A. Nukui and M. Yamane,
J. Non-Cryst. Solids, 2000, 277, 155–161.
27 R. E. Benfield, D. Grandjean, M. Kr o¨ ll, R. Pugin,
T. Sawitowski and G. Schmid, J. Phys. Chem., 2001, 105,
1961–1970.
4
5
28 L. X. Chen, T. Rajh, Z. Wang and M. C. Thurnaner, J. Phys.
Chem. B, 1997, 101, 10688–10697.
J. E. Sueras, Y. Cesteros and P. Salagre, J. Mol. Catal. A: 29 S. N. Reifsnyder and H. H. Lamb, J. Phys. Chem. B, 1999, 103,
Chem., 2007, 274, 159–168. 321–329.
J. Llorca, M. Dominguez, C. Ledesma, R. J. Chiment ˜a o, 30 P. R. Ellis, D. James, P. T. Bishop, J. L. Casci, C. M. Lok and
6
F. Medina, J. Sueiras, I. Angurell, M. Seco and O. Rossell,
J. Catal., 2008, 258, 187–198.
C. D. Pina, E. Falletta and M. Rossi, J. Catal., 2008, 260, 384–386.
G. J. Kelly, in Advances in Fischer Tropsch Synthesis, Catalysts
and Catalysis, ed. B. H. Davis and M. L. Occelli, CRC Press,
2010, pp. 1–16.
7
8
T. Pasini, M. Piccinini, M. Blosi, R. Bonelli, S. Albonetti, 31 Q. Fu, H. Saltsburg and M. Flytzani-Stephanopoulos,
N. Dimitratos, J. A. Lopez-Sanchez, M. Sankar, Q. He, Science, 2003, 301, 935–938.
C. J. Kiely, G. J. Hutchings and F. Cavani, Green Chem., 32 H. T u¨ ys u¨ z, J. L. Galilea and F. Sch u¨ th, Catal. Lett., 2009, 131,
011, 13, 2091–2099. 49–53.
T.-C. Ou, F.-W. Chang and L. S. Roselin, J. Mol. Catal. A: 33 T. A. Nijhuis, B. J. Huizinga, M. Makkee and J. A. Moulijn,
Chem., 2008, 293, 8–16. Ind. Eng. Chem. Res., 1999, 38, 884–891.
2
9
Catal. Sci. Technol.
This journal is c The Royal Society of Chemistry 2013