Ex Situ and Operando Studies on the Role of Copper in Cu-Promoted SiO2-MgO Catalysts for the Lebedev Ethanol-to-Butadiene Process

Article abstract of DOI:10.1021/acscatal.5b00755

Dehydrogenation promoters greatly enhance the performance of SiO₂-MgO catalysts in the Lebedev process. Here, the effect of preparation method and order of addition of Cu on the structure and performance of Cu-promoted SiO₂-MgO materials is detailed. Addition of Cu to MgO via incipient wetness impregnation (IWI) or coprecipitation (CP) prior to wet-kneading with SiO₂ gave similar butadiene yields (～40%) as when Cu was added to the already wet-kneaded catalyst. In contrast, the catalyst prepared by impregnation of Cu on SiO₂ first proved to be the worst catalyst of the series. TEM, XRD, and XPS analyses suggested that, for all catalyst materials, Cu²⁺ forms a solid solution with MgO. This was confirmed by UV-vis, XANES, and EXAFS data, with Cu being found in a distorted octahedral geometry. As a result, the acid-base properties, as determined by Pyridine- and CDCl₃-IR as well as NH₃-TPD, are modified, contributing to the improved performance. Operando XANES and EXAFS studies of the evolution of the copper species showed that Cu²⁺, the only species initially present, is extensively reduced to a mixture of Cu⁰ and Cu⁺, leaving only a limited amount of unreduced Cu²⁺. This formation of Cu⁰ is the result of the reducing environment of the Lebedev process and is thought to be mainly responsible for the improved performance of the Cu-promoted catalysts.

Full text of DOI:10.1021/acscatal.5b00755

Research Article
pubs.acs.org/acscatalysis
Ex Situ and Operando Studies on the Role of Copper in Cu-Promoted
SiO2−MgO Catalysts for the Lebedev Ethanol-to-Butadiene Process
Carlo Angelici,† Florian Meirer,† Ad M. J. van der Eerden,† Herrick L. Schaink,† Andrey Goryachev,‡
Jan P. Hofmann,‡ Emiel J. M. Hensen,‡ Bert M. Weckhuysen,*,† and Pieter C. A. Bruijnincx*,†
†Inorganic Chemistry and Catalysis, Debye Institute for Nanomaterials Science, Utrecht University, Universiteitsweg 99, 3584 CG
Utrecht, The Netherlands
‡Laboratory of Inorganic Materials Chemistry, Schuit Institute of Catalysis, Department of Chemical Engineering and Chemistry,
Eindhoven University of Technology, P.O. Box 513, 5600 MB Eindhoven, The Netherlands
S
* Supporting Information
ABSTRACT: Dehydrogenation promoters greatly enhance
the performance of SiO2−MgO catalysts in the Lebedev
process. Here, the effect of preparation method and order of
addition of Cu on the structure and performance of Cu-
promoted SiO2−MgO materials is detailed. Addition of Cu to
MgO via incipient wetness impregnation (IWI) or coprecipi-
tation (CP) prior to wet-kneading with SiO2 gave similar
butadiene yields (∼40%) as when Cu was added to the already
wet-kneaded catalyst. In contrast, the catalyst prepared by
impregnation of Cu on SiO2 first proved to be the worst
catalyst of the series. TEM, XRD, and XPS analyses suggested
that, for all catalyst materials, Cu2+ forms a solid solution with MgO. This was confirmed by UV−vis, XANES, and EXAFS data,
with Cu being found in a distorted octahedral geometry. As a result, the acid−base properties, as determined by Pyridine- and
CDCl3−IR as well as NH3-TPD, are modified, contributing to the improved performance. Operando XANES and EXAFS studies
of the evolution of the copper species showed that Cu2+, the only species initially present, is extensively reduced to a mixture of
Cu0 and Cu+, leaving only a limited amount of unreduced Cu2+. This formation of Cu0 is the result of the reducing environment
of the Lebedev process and is thought to be mainly responsible for the improved performance of the Cu-promoted catalysts.
KEYWORDS: ethanol, operando, solid solution, SiO2−MgO catalysts, butadiene
again, this was tentatively attributed to the formation of mixed
Zn, Mg silicates. The latter example furthermore emphasizes
the importance of preparation method and order of component
addition, when using transition metal oxides to promote the
activity of SiO2−MgO catalysts. Finally, Kitayama et al. also
proposed NiO on SiO2−MgO materials to actually consist of
mixed Ni, Mg silicates and Ni dispersed in MgO.5 In their case,
the formation of such a mixed metal silicate was in fact
considered one of the prime reasons for its good catalytic
activity.
Parameters such as promoter oxidation state, loading and
order of addition, will all affect the final catalyst structure and
can thus be optimized with regard to their effect on catalytic
performance. Indeed, as illustrated below and suggested before,
promotion of these MgO-containing materials will affect the
acid−base properties of the catalysts,7 the balance of which is
key to butadiene production. Studies on these dehydrogenation
promoters typically report limited characterization data on both
1. INTRODUCTION
SiO2−MgO materials have been extensively explored for the
ethanol-to-butadiene Lebedev process.1,2 The performance of
these catalysts can be greatly enhanced by the addition of a
third component, typically a transition metal or metal oxide.
These promoters generally improved butadiene yields, by
increasing the amount of acetaldehyde produced via ethanol
dehydrogenation.3 The addition of Cr,3,4 Ni,5 Zn,3 Cu3 oxides
and metallic Ag6 has, for example, been studied in this respect.
That addition of such metal oxides is not always beneficial was
shown by Natta and Rigamonti, who prepared ZnO- and CuO-
promoted SiO2−MgO materials by coprecipitation (CP) of the
carbonates of Mg with Zn and Cu, respectively.3 The reduced
performance shown by these catalysts was attributed to the
formation of an MxMg1−xO solid solution (with M = Cu or
Zn), resulting in reduced dehydrogenation and condensation
activity. No experimental evidence for this proposed structur-
e−activity relation was reported, however. Moreover, Makshina
et al. showed that simultaneous impregnation of Zn and Mg
nitrates on SiO2 resulted in a catalyst giving a lower butadiene
yield than one synthesized by adding Zn(NO3)2 to a SiO2−
MgO material that was obtained by a wet-kneading procedure;6
Received: April 10, 2015
Revised: August 23, 2015
© XXXX American Chemical Society
6005
DOI: 10.1021/acscatal.5b00755
ACS Catal. 2015, 5, 6005−6015

ACS Catalysis
Research Article
Table 1. Overview of the Different Preparation Methods for the Various Catalyst Materials Studied Here for the Lebedev
Ethanol-to-Butadiene Process
sample name
first step
second step
CuO/SiO2−MgO (III)
CuO/SiO2
WK SiO2, MgO
IWI Cu(NO3)2 on SiO2−MgO (III)
a
IWI Cu(NO3)2 on SiO2
IWI Cu(NO3)2 on MgO
IWI Cu(NO3)2 on SiO2
IWI Cu(NO3)2 on Mg(OH)2
CP Cu(NO3)2, Mg(NO3)2
-
a
CuO/MgO
-
CuO/SiO2(IWI)+MgO
CuO/MgO(IWI)+SiO2
CuO/MgO(CP)+SiO2
WK with Mg(OH)2
WK with SiO2
WK with SiO2
a
Not applicable. WK, wet-kneading; IWI, incipient wetness impregnation; CP, coprecipitation.
the acid−base and structural properties, making the explan-
ations for the observed beneficial or detrimental catalytic effects
often somewhat speculative.
containing MgO over MgO in the dehydrogenation of 2-
propanol. Cu promotion can thus be expected to affect the
performance of Lebedev catalysts not only by introducing a
redox functionality that aids in acetaldehyde formation but also
by modifying their acid−base properties.
We previously showed that the performance of wet-kneaded
SiO2−MgO catalysts could be significantly improved by the
addition of CuO (1 wt %) via incipient wetness impregnation
(IWI, Figure S1A), followed by calcination in stagnant air.8 As
Cu-promotion resulted in a significantly increased acetaldehyde
yield, it was proposed that upon addition of Cu the aldol
condensation (see Scheme S2 for the commonly accepted
mechanism) rather than acetaldehyde formation becomes rate-
determining. On the basis of the drop in ethylene and diethyl
ether also observed upon Cu promotion, we also proposed
copper to poison the most acidic sites of the SiO2−MgO
materials. UV−vis results provided some insight into how the
type of Cu species present depended on preparation and
correlated with catalyst performance.8 In addition to isolated
CuO species supported on either SiO2 or MgO,9−15
subnanometric cluster-like (CuO)x species were, for example,
identified, and the extent of their presence was found to
coincide with improvements in butadiene yield16 (Figure S1);
these (CuO)x clusters were suggested to enhance the rate-
determining acetaldehyde condensation step.17,18 Transmission
electron microscopy (TEM) showed the morphology of wet
kneaded SiO2−MgO materials not to be modified upon
subsequent Cu-promotion,8 nor were any copper nanoparticles
(NP) detected. Energy-dispersive X-ray spectroscopy (EDX)
did show that Cu is present on both Si- and Mg-rich phases of
Knowledge of the structural and morphological properties of
Cu in CuO/SiO2−MgO catalysts is thus still very limited.32,33
More insight into the nature, dispersion, and role of CuO, and
in fact metal oxide promoters in general, in the catalysts for the
Lebedev process is therefore highly desired. Furthermore, it is
essential to get insight into any evolution in the nature of the
metal (oxide) promoter under Lebedev reaction conditions;
this has not, however, been studied yet. Indeed, given the
reducing environment of the ethanol-to-butadiene process,
changes in oxidation state of transition metal oxides might be
expected. Ethanol, for instance, is able to reduce both noble34
and non-noble transition metal oxides,35 with CuO reduction
by ethanol having been reported at temperatures of ≥453 K
(i.e., significantly lower than our reaction temperature of 698
K).36 H2 is furthermore formed during reaction and could also
cause reduction of CuO. The observation that SiO2−MgO
materials promoted with NiO, ZnO, CuO, or metallic Ag all
performed remarkably similarly in catalysis6 suggests that
similar species might be formed under reaction conditions.
Different materials were synthesized with the expectation
that the order of component addition will impact catalyst
performance by altering the location and nature of Cu species
formed, the (extent of) solid solution formation, and the acid/
base properties of the materials. To this extent, catalysts
prepared supporting Cu on either SiO2 or MgO prior to the
addition of the other component during wet-kneading are
compared with a sample synthesized by incipient wetness
impregnation (IWI) of copper on already wet-kneaded SiO2−
MgO (CuO/SiO2−MgO (III)). The as-synthesized materials
were extensively characterized with Fourier transform infrared
spectroscopy after pyridine and deuterated chloroform
adsorption (Pyridine− and CDCl3−IR), temperature-pro-
grammed desorption of ammonia (NH3−TPD), TEM, XRD,
UV−vis, and X-ray photoelectron spectroscopy (XPS), X-ray
absorption near edge structure (XANES), and extended X-ray
absorption fine structure (EXAFS) analyses. These techniques
provided information into the speciation, oxidation state, and
geometry of Cu species in the various samples and into changes
in the acid/base properties upon copper addition. The results
show that the order of addition indeed affects the presence of
(CuO)x clusters and thus catalyst performance.8 In addition, a
large portion of copper was found to be present in a
CuxMg1−xO solid solution, which again affects the acid−base
properties. Furthermore, one selected CuO/SiO2−MgO
catalyst was studied with X-ray Absorption Spectroscopy
(XAS) under operando conditions to investigate the nature
the samples, however.8 Although TEM is often used to study
19−22
Cu(O) NP on SiO2
and metallic copper on MgO,23−25
reports on TEM imaging of CuO/MgO samples are very
limited. This might be due to lack of contrast between CuO
and MgO or, alternatively, to a very high CuO dispersion.26 El-
Shobaky et al., for instance, did not observe any CuO
reflections by X-ray diffraction (XRD) for CuO/MgO with
up to 23 wt % of copper oxide,27 concluding that a solid
solution is formed for samples treated at 673 K (our samples
were calcined at 773 K, see below). Other have also observed
the formation of CuxMg1−xO solid solutions.28,29 For our
materials, XRD analysis also did not show any CuO reflections,
perhaps because of low loading or as a result of solid solution
formation. Grirrane et al. suggested that CuO NP can be
formed on MgO and visualized these structures with TEM.30
While the NP are difficult to identify in the TEM images, their
formation might be due to the particular low-temperature
preparation method used, in which solid solution formation
might be prevented. Finally, Ueda et al. proposed that the
insertion of transition metal cations with a slightly larger ionic
radius than Mg2+, e.g., Cu2+ (0.65 and 0.69 Å, respectively),31 in
MgO increases its basicity.7 This was thought to be the reason
for the improved activity and selectivity shown by Cu-
6006
DOI: 10.1021/acscatal.5b00755
ACS Catal. 2015, 5, 6005−6015

ACS Catalysis
Research Article
Figure 1. Ethanol conversion (left), butadiene yield (middle), and conversion and yield of the main (by-) products for CuO/SiO2−MgO (III) and
the catalyst materials prepared varying the order of addition of CuO.
Figure 2. (a) Pyridine-IR spectra at 423 K in the region 1650−1400 cm−1 and (b) CDCl3−IR spectra at 323 K in the region 2300−2050 cm−1 for
SiO2−MgO (III) and CuO/SiO2−MgO (III).
of the active copper species during ethanol-to-butadiene
conversion. This way, it could be shown that the reducing
atmosphere of the Lebedev process indeed causes reduction of
the Cu2+ species originally found in the CuxMg1−xO solid
solution to a mixture of unreduced Cu2+, Cu+ and large
amounts of metallic copper, the latter of which is held
responsible for the improved ethanol dehydrogenation activity.
Ethanol conversions, determined at 698 K for 4 h time on
stream (Figure 1), are quite similar for all catalysts, with the
exception of CuO/SiO2(IWI)+MgO, which proved less active.
The latter sample also gave a significantly lower butadiene yield,
showing that deposition of copper on silica first is undesired.
Overall, quite similar product distributions (see the SI for a full
description of products observed) were obtained with very
small variations in acetaldehyde formation and a slightly higher
ethylene yield for CuO/MgO(CP)+SiO2.
2. RESULTS AND DISCUSSION
CuO/SiO2 and CuO/MgO, prepared as reference samples
for the characterization studies, both showed much lower
ethanol conversion levels (Figure S2) with no or very little
(1%) butadiene formation, respectively. Moreover, CuO/SiO2
showed the highest amount of acetaldehyde, confirming that
MgO or MgO-containing catalysts are required for the aldol
condensation step in the mechanism. Previously, NiO/SiO2 and
NiO/MgO were also found to be essentially inactive toward
butadiene formation.5 These results further support the notion
that both components are required, with the presence of both
MgO and the magnesium silicates that formed upon wet-
kneading being essential for catalysis,37 also in the case of Cu-
promoted materials.
2.1. Performance of the Catalysts. To study the effect of
copper addition order, Cu(NO3)2 was supported on either SiO2
or MgO prior to wet-kneading with the other oxide and
performance compared to the benchmark CuO/SiO2−MgO
(III) catalyst, prepared by copper addition after wet-kneading.
CuO/SiO2(IWI)+MgO was thus prepared by supporting
copper on SiO2 by IWI, before wet-kneading with Mg(OH)2
(Table 1). Two other samples were prepared by adding CuO to
MgO with two different techniques: CuO/MgO(IWI)+SiO2
was prepared by supporting Cu(NO3)2 on Mg(OH)2 via IWI,
while for CuO/MgO(CP)+SiO2, Cu(NO3)2 and Mg(NO3)2
were coprecipitated with TMAOH; in both cases, the resulting
solid was then wet-kneaded with SiO2. Lastly, all catalysts were
calcined in stagnant air at 773 K for 5 h (see Supporting
Information (SI) for experimental details).
2.2. Acid−Base Properties of the Catalysts. Different
characterization techniques were used to compare the acid−
6007
DOI: 10.1021/acscatal.5b00755
ACS Catal. 2015, 5, 6005−6015

ACS Catalysis
Research Article
base properties of one selected Cu-containing sample, i.e.,
CuO/SiO2−MgO (III), with the corresponding, unpromoted
material. Pyridine-IR results reported in Figure 2a show that
only Lewis acid sites are present in both materials;38 Larina et
al., on the other hand, recently reported both Lewis and
Brønsted acid sites to be present on SiO2−MgO catalysts, prior
and after promotion with ZnO.39 Although the similar
intensities observed by Pyridine-IR suggest similar amounts
of acidic sites, NH3-TPD shows the Cu-promoted sample to
have less acidic sites (Table 2). We previously already observed
addition to also poison the strong basic sites in the Cu-
containing material.
The much lower yields of ethylene generally seen upon Cu-
promotion (Table 2),8 are in line with the lower amount of
acidic and strong basic sites observed for CuO/SiO2−MgO
(III). The significantly higher butadiene yield observed for the
promoted sample might be not only the result of the
dehydrogenation activity of Cu but also the result of the
increase in relative strength and amount of weak basic sites that
are beneficial for the aldol condensation step. To better
understand the reasons behind the modified acid−base
properties and the effect of the different preparation on the
nature of Cu species, the set of samples was extensively
characterized, as detailed below.
Table 2. Amount of Acidic Sites Obtained by NH3-TPD
Analysis, Butadiene and Ethylene Yields for SiO2−MgO (III)
and CuO/SiO2−MgO (III)
2.3. Catalyst Morphology, Structure, and Copper
Geometry. Figure 3 shows the TEM images of CuO/SiO2−
MgO (III), CuO/MgO and CuO/SiO2. No copper species can
be seen on CuO/MgO and CuO/SiO2−MgO (III). On the
contrary, well-defined, quite monodisperse (∼2−3 nm) CuO
NP are present on CuO/SiO2.
number of acidic sites
butadiene
yield (%)
ethylene
sample
(mmol g−1
)
yield (%)
SiO2−MgO (III)
0.219
0.180
16
37
12
4
CuO/SiO2−
MgO (III)
XRD analyses (Figure 4) show that the intensity of the peaks
associated with the crystalline periclase MgO phase varies for
a similar discrepancy between these techniques and attributed
this to the difference in probe molecule size, with NH3 able to
probe a higher and more realistic amount of acidic sites.40 The
results thus show that some acidic sites of the original SiO2−
MgO material are poisoned upon Cu addition.
Figure 2b shows the remarkable differences between basic
sites found on the promoted and unpromoted SiO2−MgO (III)
as probed by CDCl3−IR (see ref 40 for detailed assignments).
In particular, the FT-IR peak due to weak basic sites shifted to
lower wavenumbers for the Cu-containing sample (i.e., from
2256 to 2252 cm−1); such a shift indicates that the relative
strength of these sites is higher in CuO/SiO2−MgO (III).41
Moreover, a slightly larger amount of weak basic sites is seen
for the promoted sample.
The increase in amount and relative strength of weak basic
sites might be the result of solid solution formation between
CuO and MgO, as discussed below. It is indeed in line with the
proposal by Ueda et al. that solid solution formation causes a
distortion in the MgO crystal lattice resulting in increased
basicity.7 The FT-IR peak assigned to medium basic sites at
2213 cm−1 is similar in both position and intensity in the two
samples. The intensity of the two FT-IR peaks assigned to
strong basic sites of the Mg−O(H)-Si type,40 i.e., those at
∼2139 and 2089 cm−1, is significantly lower for the CuO/
SiO2−MgO (III) sample, however. This suggests copper
Figure 4. XRD patterns of (a) CuO/MgO, (b) CuO/MgO-
(IWI)+SiO2, (c) CuO/MgO(CP)+SiO2, (d) CuO/SiO2−MgO (III),
(e) CuO/SiO2(IWI)+MgO, and (f) CuO/SiO2.
the MgO-containing samples. For all wet-kneaded samples, the
broadening (with reference to CuO/MgO) and reduced
intensity suggests significant chemical changes in the MgO
component. Moreover, the two low-intensity peaks at ∼41 and
45 2θ degrees seen for CuO/SiO2 and assigned to a tenorite
Figure 3. TEM images of (a) CuO/MgO, (b) CuO/SiO2, and (c) CuO/SiO2−MgO (III).
6008
DOI: 10.1021/acscatal.5b00755
ACS Catal. 2015, 5, 6005−6015

ACS Catalysis
Research Article
phase,42 i.e., crystalline CuO NP, show that they can indeed be
detected with XRD even at these low Cu loadings of 1 wt %. It
is therefore telling that all MgO-containing samples show no
such peaks; given that all materials were thermally treated in the
same way, this clearly shows that the reason CuO is not seen by
TEM in the other samples is not because of lack of contrast, but
is rather the consequence of improved copper dispersion.
The UV−vis spectra of these materials proved quite complex,
with a number of features contributing to the observed intensity
(Figure 5). For the CuO/SiO2 material, a very broad d−d
this and provide additional insight into nature, oxidation state
and geometry of Cu species in the various samples.
A selected number of samples were studied with XPS and
binding energies (BE) of the Cu 2p3/2, Si 2p and Mg 2p signals
are reported in Table 3. In line with the TEM, XRD, and UV−
Table 3. Binding Energies of Cu 2p3/2, Si 2p, and Mg 2p
Levels in the Various Samples
Cu 2p3/2 BE
(eV)
Si 2p BE
(eV)
Mg 2p BE
(eV)
sample
CuO/SiO2
a
934.8
933.5
934.0
933.9
933.7
104.2
-
a
CuO/MgO
-
49.7
50.3
50.0
49.8
CuO/SiO2−MgO (III)
CuO/SiO2(IWI)+MgO
CuO/MgO(IWI)+SiO2
102.8
103.0
102.5
a
The element is not present in the analyzed sample.
vis results, the BE of in CuO/SiO2 is in good agreement with
the one observed for pure SiO2.45 Interestingly, the Si 2p values
for the CuO/SiO2−MgO samples are significantly lower,
however, and correspond well to the range Si 2p BE reported
for a number of (magnesium) silicates by Okada et al.46 The BE
of Si 2p of 102.6 eV reported for talc (a magnesium silicate that
can be formed upon wet-kneading of SiO2 and MgO)37 is
similar to the BE of Si observed for the various CuO/SiO2−
MgO samples.
The XPS data also showed all copper to be Cu2+ in all
samples, based on the characteristic shakeup satellite peaks47 at
approximately 943 eV alongside a BE of the Cu 2p3/2 core level
(Table 3) in the range of 933−935 eV; indeed, the BE of bulk
CuO is reported in the range of 933.2−934.2 eV,48,49 whereas
Cu0 and Cu+ species are generally found at ∼932.0−932.5 eV.50
The low amount of Cu detected at the surface of all MgO-
containing samples hinders the deconvolution of the signal for
Cu 2p3/2 into contributions of different copper species. The Cu
2p3/2 region can thus only be used for semiquantitative analysis
of Cu abundance and to assess the variation in BE position
among the different samples. Moreover, the high-resolution Cu
2p3/2 spectra (Figure S4) show the surface Cu amounts to vary
widely, being the highest on CuO/SiO2. For CuO/MgO, it is
below the quantification limit, which implies that the copper
must be distributed throughout the bulk of the sample (see
UV−vis and XAS below). Indeed, this provides further
evidence for solid solution formation and inclusion of copper
in the crystal lattice of MgO. Finally, for all CuO/SiO2−MgO
samples a surface copper amount intermediate between that of
CuO/SiO2 and that of CuO/MgO was observed; this might be
due to Cu being present on both oxides, as well as on the
magnesium silicates formed during wet-kneading. The BE of
Cu 2p3/2 was observed to vary as well and follows the order:
CuO/SiO2 > CuO/SiO2−MgO (III) ≈ CuO/SiO2(IWI)+MgO
≈ CuO/MgO(IWI)+SiO2 > CuO/MgO. Bennici et al.
proposed shifts to higher binding energy to correlate with the
acidity of the support for CuO supported on SiO2−Al2O3.47 In
our case, the order listed above also seems to be in line with the
overall acidity/basicity of the supports.
Figure 5. UV−vis spectra of (a) CuO/MgO, (b) CuO/MgO-
(CP)+SiO2, (c) CuO/MgO(IWI)+SiO2, (d) CuO/SiO2−MgO (III),
(e) CuO/SiO2(IWI)+MgO, and (f) CuO/SiO2. The spectra are offset
for clarity. i, bands assigned to LMCT transitions; ii, bands assigned to
d−d transitions.
transition is detected at ∼625 nm, well in line with the 636 nm
reported for bulk CuO (see Figure S3).43,44 The UV−vis
spectra of all ternary oxides and CuO/MgO are, however, very
different, suggesting copper not to be in the square planar
geometry of (bulk) CuO. Indeed, the absorption bands
observed in the 200−300 nm region can be assigned to
ligand-to-metal charge transfer (LMCT) bands of isolated CuO
species on different supports.9,10,15 For example, Bravo-Suar
́
ez
et al. attributed a band at ∼260 nm detected for CuMgAlOx to
an LMCT of Cu2+ in an octahedral field.16 Moreover, the d−d
transitions seen at 700−800 nm are also indicative of Cu2+
species in a (distorted) octahedral field.
The 300−500 nm absorption band, previously assigned to
LMCT transitions of subnanometric (CuO)x clusters,10,11,16
has a similar intensity for the three catalyst materials that show
a similar, superior performance, i.e., CuO/MgO(CP)+SiO2,
CuO/MgO(IWI)+SiO2 and CuO/SiO2−MgO (III); the same
feature is also observed in CuO/MgO. On the other hand, for
the worst catalyst, CuO/SiO2(IWI)+MgO, the 300−500 nm
absorption band is negligible in intensity. With this shoulder
also being absent in CuO/SiO2, this means that the formation
of small (CuO)x clusters is favorable on MgO-containing
materials but not on SiO2.
TEM and XRD results thus show CuO NP to be present on
CuO/SiO2 but not on any of the MgO-containing materials.
Moreover, the UV−vis results suggest the copper species in all
MgO-containing samples to have a (distorted) octahedral
geometry. All of this points at solid solution formation with the
Cu2+ atoms being located at crystal lattice sites of MgO. XPS,
XANES, and EXAFS analyses, discussed below, further confirm
Ex situ XAS analysis provided further information on Cu
speciation, coordination, and chemical environment in the as-
synthesized samples. The XANES spectra show two pre-edge
features (labeled 1 and 2 in Figure 6). This feature at ∼8979
eV, originating from the dipole forbidden 1s → 3d transition, is
considered a signature for Cu2+ species.51 This, together with
6009
DOI: 10.1021/acscatal.5b00755
ACS Catal. 2015, 5, 6005−6015

ACS Catalysis
Research Article
lattice.57 The nature and geometry of Cu species in our
materials is expected to be further complicated, however, by the
third SiO2 component and by the magnesium silicates. EXAFS
analyses nonetheless provided more insight into the geometry
and bond distances of Cu present in our samples (see Table 4
and Figure S5).
Expectedly, the EXAFS data for CuO/SiO2 and reference
bulk CuO were seen to be quite similar (Table 4 and Table S1),
in line with the XRD, TEM, XPS, and XANES results. A very
different situation is observed for CuO/SiO2−MgO (III),
CuO/MgO(IWI)+SiO2, and CuO/MgO (see Table 4).
Importantly, no Cu−Cu contributions could be found for
these samples; instead, Cu−Mg bond distances fitted for these
samples are in the range 2.98−3.02 Å, i.e., similar to the Mg−
Mg distance of 2.97 Å observed in periclase,58 corroborating
the presence of copper in the crystal lattice of MgO. An initial
fit of the EXAFS data showed Cu to coordinate to four oxygen
atoms at a bond distance of 1.92−1.94 Å (Table S2 and Figure
S6), which is smaller than the Mg−O distance of ∼2.10 Å in
MgO59 and reflects the significant distortion of the crystal
lattice caused by copper atom insertion; this distortion can
contribute to the alteration in acid−base properties of our Cu-
promoted SiO2−MgO materials.7 The presence of only four
Cu−O contributions is not in line, however, with the UV−vis
and XANES results that show Cu to be in a distorted
octahedral geometry in CuO/MgO, CuO/SiO2−MgO (III)
and CuO/MgO(IWI)+SiO2. Interestingly, Asakura et al. also
observed Cu in an octahedrally distorted geometry to be
present in a MgO crystal lattice to have an oxygen CN of less
than six.60 Given the strong indications for octahedral
coordination, the EXAFS spectra were again analyzed with
the assumption that two additional O atoms could be present.
The latter fit gave results with a similar goodness-of-fit as the
initial one, with two additional oxygen atoms now located at a
distance of ∼2.40 Å. This corresponds very well with the
average Cu−Oapical distances of 300 (well-defined) CuO6
complexes (Cambridge Structural Database).61 Specifically,
the structure search showed these CuO6 coordination spheres
to have a distorted octahedral geometry with bond distances in
the ranges 1.88−2.00 and 2.30−2.50 Å for equatorial and apical
oxygen atoms, respectively. The values obtained by EXAFS
analysis of 1.94 and 2.40 Å for CuO/MgO (Table 4) can thus
Figure 6. Normalized XANES spectra of (a) CuO, (b) CuO/SiO2, (c)
CuO/SiO2(IWI)+MgO, (d) CuO/SiO2−MgO (III), (e) CuO/
MgO(IWI)+SiO2, and (f) CuO/MgO.
the absence of a feature at 8982 eV characteristic of Cu+
confirms all copper to be present as Cu2+ in all samples. Pre-
edge feature 2, due to a 1s → 4p transition, is again
characteristic of Cu2+.52 The position of this feature depends
on local copper symmetry, in addition to its charge.53 Feature 2
is centered at ∼8987 eV for CuO and CuO/SiO2, implying it to
be very similar to bulk CuO (as confirmed by EXAFS analyses,
see below). Cu2+ in tenorite has been described as square
planar (i.e., with four oxygen atoms at ∼1.95 Å),54 but can also
be considered as strongly distorted octahedral when taking into
account the two oxygen atoms at 2.78 Å (see Table S2).
Notably, feature 2 is found at a different energy for all MgO-
containing samples, i.e., at ∼8989 eV, and shows Cu to be
predominantly present in an octahedral environment. This pre-
edge feature 2 is known to progressively shift to higher energy
with decreasing octahedral distortion.55 Also in line with our
results, Hilbrandt et al. noted a shift to higher energy for feature
2 for CuO/MgO compared to bulk CuO and attributed this to
copper in CuO/MgO being in an octahedral or distorted
octahedral geometry, occupying Mg lattice sites in a solid
solution.56
It has been reported that electron spin resonance (ESR)
measurements can distinguish between two types of copper, i.e.,
Cu2+, in an octahedral field in the bulk of MgO and Cu2+ in an
unclear geometry close to the surface, in such a MgO crystal
Table 4. Parameters Obtained from Analysis of Cu K-Edge k3-Weighted EXAFS of the Different Samples
c
d
e
f
sample
CuO/SiO2
CuO/SiO2(IWI)+MgO
shell
CN
r (Å)
DW
Ef
Cu−O
Cu−O (2)
Cu−Cu
Cu−O
4.0
2.9
3.2
2.8
4.0
1.5
4.6
4.0
1.1
3.7
4.1
1.5
4.7
1.94
2.79
2.86
1.93
1.94
2.48
3.02
1.91
2.39
3.00
1.94
2.40
2.98
0.006
0.012
0.012
0.005
0.009
0.149
0.019
0.009
0.013
0.016
0.009
0.013
0.008
−1.0
3.6
14.2
0.9
a
b
CuO/SiO2−MgO (III)
Cu−O
4.7
Cu−O (2)
Cu−Mg
Cu−O
Cu−O (2)
Cu−Mg
Cu−O
0.3
2.9
b
CuO/MgO(IWI)+SiO2
8.5
12.5
−0.9
7.7
b
CuO/MgO
Cu−O (2)
Cu−Mg
0.1
3.0
a
b
The shells higher than the first one could not be fitted due to the presence of at least two contributions in antiphase. The values for the parameters
c
d
e
in the original fit of these samples are reported in the SI Table S2. Coordination number. Distance averaged over the shell. Debye−Waller factor.
f
Fermi energy (edge position).
6010
DOI: 10.1021/acscatal.5b00755
ACS Catal. 2015, 5, 6005−6015

ACS Catalysis
Research Article
be considered common for a Jahn−Teller distorted Cu2+ ion in
a CuO6 environment.
Differently from all other samples, EXAFS analysis did not
find one specific element in the second Cu shell for CuO/
SiO2(IWI)+MgO due to two or more contributions in
antiphase in the Fourier transform (FT). This is surprising
considering that the XANES data showed features quite similar
to the other samples containing both CuO and MgO. CuO/
SiO2(IWI)+MgO showed different UV−vis features and
catalytic performance, though, indeed suggesting differences
in Cu speciation. As EXAFS measures an average for the
contribution of the various Cu species present, those present in
small amounts can be difficult to discern with this technique.
The EXAFS results showing most of the Cu in a solid solution
with MgO thus do not exclude the presence of minor amounts
of other species ((CuO)x clusters, Cu on SiO2 or Cu present in
magnesium silicates), as suggested by other techniques.
Summarizing the ex situ characterization results, it can first
be noted that the TEM and XPS results indeed support the
formation of surface magnesium silicates during wet-kneading.
The XAS spectra show CuO/SiO2−MgO (III), CuO/MgO-
(IWI)+SiO2, and CuO/MgO to be similar and have copper
mainly present in the MgO crystal lattice sites. The solid
solution formation is reflected by the distorted octahedral
geometry seen for Cu2+. Although the XRD and XANES data
suggests that CuO/SiO2(IWI)+MgO also can be best described
as a solid solution, the UV−vis and the EXAFS data show that
this material is structurally somewhat different from a copper
perspective. Finally, all copper was observed to be present with
oxidation state +2 only.
The results thus show that the samples prepared supporting
Cu(NO3)2 on MgO or coprecipitating the nitrates of Cu and
Mg have remarkably similar features, implying a similar end-
state for both samples. In particular, the use of a CP method for
CuO/MgO(CP)+SiO2 might allow for the direct formation of a
solid solution, which is also obtained eventually with the other
sample where Cu is supported on MgO by IWI, by
redistribution of Cu species either during wet-kneading or the
thermal treatment. Additionally, these samples possess similar
crystallinity and nature of copper species as compared with
CuO/SiO2−MgO (III), the benchmark catalyst of this study.
Remarkably, the method chosen for copper addition, be it to
MgO or SiO2−MgO, then only affects the final structure to a
very limited extent. On the other hand, the sample prepared by
impregnating Cu(NO3)2 on SiO2 prior to wet-kneading with
MgO, i.e., CuO/SiO2(IWI)+MgO, does show a number of
structural differences such as a highly reduced crystallinity of
the periclase phase, the lack of CuO subnanometric clusters and
copper location and geometry.
Figure 7. Graphical summary of the two experiments studying the
evolution of copper in CuO/SiO2−MgO (III) under operando
conditions, including the labels of the different steps.
The capillary reactor was heated to 698 K over 1 h in N2 flow
before allowing ethanol onto the catalyst, after which XAS data
was collected for 4 h time on stream. In the second experiment,
the sample was prereduced by allowing pure H2 over the
catalyst for 2 h at 698 K. After reduction, the system was
flushed with pure N2 for 4 h to ensure all H2 was removed
before switching to EtOH/N2 and again monitoring the
reaction for 4 h.
To ensure operando conditions, EtOH consumption and
formation of the main products were monitored by online MS.
Butadiene is indeed formed, as shown by the product
composition of step 1b (Figure 8); in fact, after an initial
induction period ethanol conversion, as well as butadiene and
acetaldehyde yields, were all found to be stable for 4 h time on
stream. The product compositions observed during steps 1b
and 2d are in fact very similar. The slightly higher amounts seen
for all compounds during experiment 2 should be interpreted
with caution, as these can be due to small variations in the
amount of ethanol fed or to small changes in flow as a result of
packing of the catalyst bed. The time required to reach a steady
state in terms of butadiene production is different for the two
experiments, though, i.e., 1 h for 1b and 3 h for 2d.
XANES analysis provided insight into changes in Cu
oxidation state during reaction. The X-ray absorption spectra,
recorded over an energy range of 8965−9055 eV, were fitted
with four reference samples: the spectrum of CuO/SiO2−MgO
(III) was used as reference to account for Cu species that
remained unchanged; Cu foil and Cu2O were used to assess any
changes in oxidation state of Cu occurring during reaction;
finally, CuO/SiO2 was included to account for the possibility of
CuO NP formation.
The fitting of the XANES recorded during all experimental
steps (10−1b, 2a−2d) resulted in satisfactory R-factors of
<0.002. Indeed, the good match between the experimental and
fitted data for experiments 1 and 2 (Figure S7) shows that the
deconvolution procedure captures the chemical evolution of
copper species well. The XANES obtained during step 10 (RT,
nitrogen flow) should be identical to the observed for CuO/
SiO2−MgO (III) XANES measured ex situ (see above). This
thus allowed us to estimate the experimental uncertainty of our
fitting calculations to be ∼10% (see experimental section). As
compared with step 10, no changes are again observed in the
XANES upon heating the sample under nitrogen to 698 K
(step 1a, data not shown). Only upon introduction of ethanol,
Copper addition was furthermore shown to alter the acid−
base properties of the samples (shown here for (CuO/)SiO2−
MgO (III)). The reduction in amount of acidic and strong basic
sites explains the lower amounts of unwanted dehydration
products that are formed. At the same time, Cu-promotion
brings about a slight increase in the amount of weak basic sites,
thought to be beneficial for the aldol condensation step.
2.4. Operando XAS Measurements of the Evolution in
Copper Speciation. Changes in the nature of the Cu species
in CuO/SiO2−MgO (III) were studied by XAS under
operando conditions following two different approaches
(Figure 7). First, the effect of the chemical potential of the
Lebedev process on the nature of copper was assessed under
standard reaction and pretreatment conditions (experiment 1).
6011
DOI: 10.1021/acscatal.5b00755
ACS Catal. 2015, 5, 6005−6015

ACS Catalysis
Research Article
features associated with reduced copper species appear almost
instantaneously upon introduction of ethanol and the final
copper composition is reached within 1 h. Importantly, no
further changes are detected for the remaining 3 h time on
stream.
The XAS data obtained for step 2a, again showed the nature
of Cu to be very similar to the ex situ XAS analysis and to the
sample measured at 298 K (i.e., step 10). Indeed, the inclusion
of CuO NP (10) and reduced copper species (2a) in the fit
again reflects the experimental uncertainty (∼10%, see
experimental section). In step 2b, pure H2 is allowed in the
reactor and reduction of copper is observed within the first 0.5
h after which the XANES spectra do not change any more.
Quantification of copper species present after prereduction (i.e.,
during step 2c) again shows Cu0 to be predominant; in fact, the
fits obtained for step 2c and 1b are, given the experimental
uncertainty, very similar (Figure 9). Finally, switching to
Lebedev conditions in step 2d, only slightly changed the
composition to give ∼65% Cu0, 16% Cu+ and 18% Cu2+. This
shows that the same ratio of copper oxidation states is obtained,
whether the sample was subjected to a prereduction step or not.
The incomplete reduction of Cu2+ could be due to the
presence of unreducible copper species or to slow reduction
kinetics of some species with the 2 h reduction step being too
short. The former is more likely, given the lack of change in the
XANES spectra at the end of the reduction step 2b. Indeed,
given that the different steps taken to reach the operando
conditions of 1b and 2d essentially give the same outcome as
far as copper is concerned, it seems that a quasi-steady-state
situation is reached. It is difficult to establish which of the Cu2+
species do not undergo reduction, but for similar systems, it has
been suggested that some of the Cu2+ in a solid solution can
retain its original oxidation state under reducing atmosphere.
Although no examples could be found for copper reducibility in
a CuxMg1−xO solid solution, Ni2+ species in NixMg1−xO have
often been observed to undergo reduction at significantly
higher temperatures as compared with other Ni species.62,63
Similarly, Cu2+ in a solid solution with CeO2 was also shown to
be less reducible than bulk CuO.64,65
Figure 8. Online MS analysis of the product composition during
operando XAS analysis of steps 1b (top) and 2d (bottom) as
described in Figure 7.
the spectra show a remarkable change in copper speciation. In
particular, the XANES data at steady state of step 1b shows
severe reduction of copper to give a composition of ∼60% Cu0,
20% Cu+, and 20% of the Cu2+ species found in the initial state
of CuO/SiO2−MgO (III). It is difficult to say if the observed
reduction is due to ethanol or to the H2 formed, both being
competent reductants under the applied conditions.36 The
Figure 9. Amount of the different species estimated by fitting the different contributions in the XANES region and R-factor expressing the goodness-
of-fit for each experimental step. Note that as steps 1a and 2a are identical, only the latter is included here.
6012
DOI: 10.1021/acscatal.5b00755
ACS Catal. 2015, 5, 6005−6015

ACS Catalysis
Research Article
Operando EXAFS data again provided further insight into
the nature of copper with regards to coordination number
(CN) and nearest-neighbor contributions (Table 5 and Figure
for 2-propanol dehydrogenation to reduction of CuO to
metallic Cu.67 Finally, Carotenuto et al. used three commercial
Cu-based catalysts for ethanol dehydrogenation with operando
XANES,68 proposing Cu0 species formed in situ to be
responsible for ethanol adsorption and dehydrogenation. That
Cu0 generated in situ is responsible for increased dehydrogen-
ation activity could then also explain the similar performance
reported for SiO2−MgO Lebedev catalyst promoted with
metallic Ag and CuO.6
Table 5. Parameters Obtained from Analysis of Cu K-Edge
a
k3-Weighted Operando EXAFS Data
b
c
d
e
experiment step
shell
CN
r (Å)
DW
Ef
10
Cu−O
Cu−O (2)
Cu−Mg
Cu−O
Cu−Cu
Cu−Mg
Cu−O
Cu−O (2)
Cu−Mg
Cu−O
Cu−Cu
Cu−Mg
Cu−O
4.4
2.0
3.7
2.2
3.6
3.3
3.3
1.8
4.0
1.4
3.6
3.3
1.1
3.6
3.3
1.94
2.42
2.87
1.91
2.54
2.91
1.94
2.42
2.92
1.93
2.54
2.91
1.93
2.54
2.91
0.009
0.020
0.015
0.020
0.008
0.002
0.011
0.020
0.020
0.134
0.008
0.003
0.011
0.008
0.002
3.5
−8.8
8.8
We previously proposed that acetaldehyde self-condensation
is rate-determining for the Cu-promoted materials,8 whereas
acetaldehyde formation itself is thought to be rate-determining
for the unpromoted SiO2−MgO catalysts.69 This is in
agreement with Makshina et al.,6 who also observed
acetaldehyde accumulation for Ag-promoted SiO2−MgO
materials and concluded that more MgO was required to
perform the subsequent aldol condensation step. This shift in
rate-determining step then also explains why the amount of
dehydrogenation promoter introduced does not affect the
overall butadiene yield for our Cu-promoted materials (in the
range of 0.5−2 wt %, see ref 8), as also previously reported for
Ag- and Zn-promoted ones.6 Apparently, a small amount of
promoter is already sufficient to produce more acetaldehyde
than can be converted by the sites responsible for aldol
condensation. While solid solution formation has been
previously suggested for CuO-, ZnO-promoted SiO2−MgO
Lebedev catalysts3 and the formation of mixed Ni, Mg silicates
have been proposed, our operando results show for the first
time that the oxidation state and location of Cu changes during
reaction; such an in situ reduction might also occur for
promoters which are chemically similar to Cu (e.g., Ni). Finally,
it should be noted here that a considerable amount of copper
(i.e., ∼ 40%) is present with oxidation state +1/+2, Cu species
that are still expected to also contribute to the different steps of
the Lebedev process. These CuxO species are for instance
believed to poison those sites responsible for ethanol
dehydration, given the significantly lower ethylene yield with
promoted SiO2−MgO materials.
Finally, the operando XAS studies shed light on the
deactivation shown by the Cu-promoted catalysts during 24 h
time on stream (Figure S9). UV−vis spectra of the spent CuO/
SiO2−MgO samples showed considerably more carbonaceous
materials to be present on these samples than on the
unpromoted catalysts (cf. Figures S10 and S11), suggesting
that deactivation should be attributed to blockage of the active
species. Alternatively, deactivation might be the result of
sintering of the Cu0 particles formed during reaction. Marchi et
al., for instance, held sintering of metallic copper formed in situ
on Cu/SiO2 catalysts responsible for the drop seen in 2-
propanol dehydrogenation activity.67 The fact that Cu−Cu CN
is the same in 1b, 2c, and 2d, together with the observation that
the Cu−Cu CN calculated at different times of step 2d (not
shown) does not vary during reaction, seems to exclude any
sintering of Cu0. Also taking into account that only a small
amount of Cu0 is needed to shift the rate-determining step, loss
of active Cu0 dehydrogenation sites by blockage rather than
sintering is thought to be the main reason for the slight, yet
gradual deactivation that is observed.
1b
2a
2c
2d
15.0
−2.3
9.5
8.4
10.0
0.5
13.8
−2.3
9.5
13.8
−2.3
9.5
Cu−Cu
Cu−Mg
a
The experiment steps are shown in Figure 7. 10 refers to the EXAFS
spectrum taken at RT prior to thermal treatment, i.e., before 1a.
b
c
d
Coordination number. Distance averaged over the shell. Debye−
e
Waller factor. Fermi energy (edge position).
S8). For step 10, the Cu−O and Cu−Mg distances are, as
expected, very similar to those observed for the sample
analyzed ex situ, no Cu−Cu contributions are seen, and the
data fits to Cu found in a distorted octahedral geometry at
MgO crystal lattice sites. The EXAFS data for step 2a, i.e., at
698 K, essentially shows the same situation, with bond
distances being quite similar to 10, whereas CN are slightly
different. For all steps, the EXAFS results are thus in good
agreement with the XANES data. Notably, EXAFS shows the
appearance of Cu−Cu contributions in step 1b, with distances
being similar to metallic Cu (i.e., 2.55 Å; see Table S2), thus
confirming that a large percentage is now completely reduced
and that Cu0 particles are formed. As a result of Cu0 and Cu+
formation, the number of O and Mg atoms coordinated by Cu
is lower. A similar trend is observed in 2c and 2d, the main
difference with 1b being the slightly lower number of O atoms
coordinating to Cu.
The EXAFS data thus also shows that in all cases (i.e., 1b, 2c
and 2d) a quasi-steady-state is reached in terms of Cu
speciation, as indicated by same number of Cu and Mg
scatterers found in the first shell, for example. While the larger
amounts of Cu0 fitted to the XANES of steps 2c and 2d are
within the experimental error, EXAFS analysis does show a
reduction in the average number of O atoms around Cu in 2c
and 2d, thus suggesting that these might indeed be slightly
more reduced than 1b.
The time required for the measured butadiene yields to
stabilize under the operando XAS conditions seems to coincide
with the time required to reach a steady state in copper
oxidation state. It is likely that the Cu0 species formed are
mainly responsible for the boost in butadiene yield seen upon
Cu-promotion. Indeed, Iwasa and Takezawa showed ethanol
dehydrogenation to occur rapidly on prereduced but not on
unreduced Cu-based catalysts.66 Similarly, Marchi et al.
attributed the initial increase in activity of a Cu/SiO2 catalyst
3. CONCLUSION
The higher butadiene yields obtained with Cu-promoted SiO2−
MgO catalysts are attributed to improved dehydrogenation
activity and modified acid−base properties as compared with
6013
DOI: 10.1021/acscatal.5b00755
ACS Catal. 2015, 5, 6005−6015

ACS Catalysis
Research Article
unpromoted SiO2−MgO catalysts. Here, we show how
different preparation methods and addition orders of Cu affect
both the structure and performance of Cu-promoted SiO2−
MgO catalysts. In particular, the two materials where Cu was
supported on MgO prior to wet-kneading with SiO2 showed
overall performances very similar to the benchmark catalyst
(prepared by impregnation of copper on SiO2−MgO). On the
other hand, the catalyst material prepared by adding copper to
SiO2 prior to wet-kneading with MgO results in significantly
worse performance. The reduced performance of the latter
sample coincides with the absence of CuO subnanometric
clusters, proposed to favor the aldol condensation of
acetaldehyde. Extensive characterization shows all copper to
be Cu2+ and in the MgO-containing samples, to be present in
an octahedrally distorted geometry in a CuxMg1−xO solid
solution. The presence of Cu2+ in the crystal lattice of MgO
also contributes to the modification of the acid−base properties
of Cu-containing SiO2−MgO materials as compared with
unpromoted ones.
AUTHOR INFORMATION
■
Corresponding Authors
*(B.M.W.) E-mail: b.m.weckhuysen@uu.nl.
*(P.C.A.B.) E-mail: p.c.a.bruijnincx@uu.nl.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
This research has been performed within the framework of the
CatchBio program. The authors gratefully acknowledge the
support of the Smart Mix Program of The Netherlands
Ministry of Economic Affairs and The Netherlands Ministry of
Education, Culture, and Science. M.E.V. Velthoen and R. Oord
are thanked for the acid−base characterization of the catalysts.
The scientific staff of the DUBBLE beamline at ESRF (France)
is acknowledged. S. Kalirai is thanked for his help and valuable
discussions during the measurement at the DUBBLE beamline
at ESRF (France).
Operando XAS measurements provided valuable insights
into the dynamics of the copper species and on their catalytic
role. In fact, the majority of Cu2+ is rapidly reduced to Cu0
under reaction conditions. Furthermore, a quasi-steady-state
situation in terms of ratio of Cu0, Cu+, and Cu2+ species is
obtained regardless of the pretreatment history. With Cu0 being
the dominant species under reaction conditions and given the
similar performance reported for Cu- and Ag-promoted SiO2−
MgO catalyst materials, it is proposed that metallic copper is in
fact mainly responsible for the increased dehydrogenation
activity and, as a result, the observed butadiene yields. Finally,
deactivation of the copper-containing catalysts was ascribed to
carbonaceous deposits on the active sites rather than to
sintering of the promoter.
REFERENCES
■
(1) Angelici, C.; Weckhuysen, B. M.; Bruijnincx, P. C. A.
ChemSusChem 2013, 6, 1595−1614.
̀
(2) Makshina, E. V.; Dusselier, M.; Janssens, W.; Degreve, J.; Jacobs,
P. A.; Sels, B. F. Chem. Soc. Rev. 2014, 43, 7917−7953.
(3) Natta, G.; Rigamonti, R. Chim. Ind. 1947, 29, 195−201.
(4) Corson, B. B.; Jones, H. E.; Welling, C. E.; Hinckley, J. A.; Stahly,
E. E. Ind. Eng. Chem. 1950, 42, 359−373.
(5) Kitayama, Y.; Satoh, M.; Kodama, T. Catal. Lett. 1996, 36, 95−
97.
(6) Makshina, E. V.; Janssens, W.; Sels, B. F.; Jacobs, P. A. Catal.
Today 2012, 198, 338−344.
(7) Ueda, W.; Yokoyama, T.; Moro-oka, Y.; Ikawa, T. Chem. Lett.
1985, 14, 1059−1062.
(8) Angelici, C.; Velthoen, M. E. Z.; Weckhuysen, B. M.; Bruijnincx,
P. C. A. ChemSusChem 2014, 7, 2505−2515.
(9) Chary, K. V. R.; Seela, K. K.; Sagar, G. V.; Sreedhar, B. J. Phys.
Chem. B 2004, 108, 658−663.
ASSOCIATED CONTENT
■
(10) Shimokawabe, M.; Takezawa, N.; Kobayashi, H. Appl. Catal.
1982, 2, 379−387.
S
* Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acscatal.5b00755.
(11) Velu, S.; Suzuki, K.; Okazaki, M.; Kapoor, M. P.; Osaki, T.;
Ohashi, F. J. Catal. 2000, 194, 373−384.
(12) Praliaud, H.; Mikhailenko, S.; Chajar, Z.; Primet, M. Appl. Catal.,
B 1998, 16, 359−374.
Full experimental details; a comparison between the
butadiene yield and UV−vis spectra of CuO/SiO2−MgO
(III) and SiO2−MgO (III); the mechanism for the
Lebedev process; catalytic testing of CuO/MgO and
CuO/SiO2; UV−vis spectra of CuO, CuO/MgO, and
CuO/SiO2; XPS spectra of CuO/MgO, CuO/SiO2, and
CuO/SiO2−MgO (III); table containing the parameters
calculated for the references Cu foil, Cu2O, and CuO;
additional EXAFS data; fitting of the XANES spectra
obtained under operando conditions for selected steps;
comparison between the butadiene yield and the UV−vis
spectra (after reaction) of Cu-promoted and unpromoted
SiO2−MgO catalysts; schematic representation of the
setup employed for the operando XAS measurements;
comparison between the XANES spectra of SiO2−MgO
(III) promoted with 1 and 2% CuO; table containing the
fragments observed with MS under operando conditions
for the main products of the Lebedev process and
detailed description of the experimental procedure used
for the preparation of SiO2−MgO (III) and CuO/
SiO2MgO (III) (PDF)
(13) Wang, Z.; Wan, H.; Liu, B.; Zhao, X.; Li, X.; Zhu, H.; Xu, X.; Ji,
F.; Sun, K.; Dong, L.; Chen, Y. J. Colloid Interface Sci. 2008, 320, 520−
526.
(14) Kong, A.; Wang, H.; Yang, X.; Hou, Y.; Shan, Y. Microporous
Mesoporous Mater. 2009, 118, 348−353.
̀
(15) Derrien, G.; Charnay, C.; Zajac, J.; Jones, D. J.; Roziere, J. Chem.
Commun. 2008, 3118−3120.
(16) Bravo-Suarez, J. J.; Subramaniam, B.; Chaudhari, R. V. J. Phys.
́
Chem. C 2012, 116, 18207−18221.
(17) Kvisle, S.; Aguero, A.; Sneeden, R. P. A. Appl. Catal. 1988, 43,
117−131.
(18) Quattlebaum, W. M.; Toussaint, W. J.; Dunn, J. T. J. Am. Chem.
Soc. 1947, 69, 593−599.
(19) Prasetyanto, E. A.; Lee, S.-C.; Park, S.-E. Bull. Korean Chem. Soc.
2007, 28, 2359−2362.
(20) Munnik, P.; Wolters, M.; Gabrielsson, A.; Pollington, S. D.;
Headdock, G.; Bitter, J. H.; de Jongh, P. E.; de Jong, K. P. J. Phys.
Chem. C 2011, 115, 14698−14706.
(21) Gong, J.; Yue, H.; Zhao, Y.; Zhao, S.; Zhao, L.; Lv, J.; Wang, S.;
Ma, X. J. Am. Chem. Soc. 2012, 134, 13922−13925.
(22) Zhu, Y.-Y.; Wang, S.-R.; Zhu, L.-J.; Ge, X.-L.; Li, X.-B.; Luo, Z.-
Y. Catal. Lett. 2010, 135, 275−281.
6014
DOI: 10.1021/acscatal.5b00755
ACS Catal. 2015, 5, 6005−6015

ACS Catalysis
Research Article
(23) Marella, R. K.; Neeli, C. K. P.; Kamaraju, S. R. R.; Burri, D. R.
Catal. Sci. Technol. 2012, 2, 1833−1838.
(55) Lytle, F. W.; Greegor, R. B.; Panson, A. J. Phys. Rev. B: Condens.
Matter Mater. Phys. 1988, 37, 1550−1563.
(56) Hilbrandt, N.; Martin, M. J. Phys. Chem. B 1999, 103, 4797−
4802.
(24) Shi, R.; Wang, F.; Mu, X.; Li, Y.; Huang, X.; Shen, W. Catal.
Commun. 2009, 11, 306−309.
(57) Cordischi, D.; Pepe, F.; Schiavello, M. J. Phys. Chem. 1973, 77,
1240−1245.
(58) Yamaguchi, K.; Ebitani, K.; Yoshida, T.; Yoshida, H.; Kaneda, K.
J. Am. Chem. Soc. 1999, 121, 4526−4527.
(59) Hazen, R. M. Am. Mineral. 1976, 61, 1280−1293.
(60) Asakura, K.; Iwasawa, K. Mater. Chem. Phys. 1988, 18, 499−512.
(61) Cambridge Structural Database. http://webcsd.ccdc.cam.ac.uk/
(accessed August 27, 2015).
(62) Hu, Y. H.; Ruckenstein, E. Catal. Lett. 1996, 36, 145−149.
(63) Ruckenstein, E.; Hu, Y. H. Ind. Eng. Chem. Res. 1998, 37, 1744−
1747.
(25) Reddy, K. H. P.; Anand, N.; Venkateswarlu, V.; Rao, K. S. R.;
Burri, D. R. J. Mol. Catal. A: Chem. 2012, 355, 180−185.
(26) Wang, D.; Wang, X.; Xu, R.; Li, Y. Nanosci. Nanotechnol. Lett.
2007, 7, 1−5.
(27) El-Shobaky, G. A.; Hamed, M. N.; Abdalla, F. F.; El-Molla, S. A.
Colloids Surf., A 2002, 207, 293−301.
(28) Paranthaman, M.; David, K. A.; Lindemer, T. B. Mater. Res. Bull.
1997, 32, 165−173.
(29) Oku, M.; Hirokawa, K. J. Electron Spectrosc. Relat. Phenom. 1977,
10, 103−110.
(30) Grirrane, A.; Corma, A.; Garcia, H. Chem. - Eur. J. 2011, 17,
2467−2478.
(64) Chen, J.; Zhu, J.; Chen, C.; Zhan, Y.; Cao, Y.; Lin, X.; Zheng, Q.
Catal. Lett. 2009, 130, 254−260.
(65) Wang, X.; Rodriguez, A.; Hanson, J. C.; Gamarra, D.; Martínez-
(31) Raj, G. Advanced Inorganic Chemistry Vol. II; Krishna Prakashan
Media: Uttar Pradesh, India, 2014.
Arias, A.; Fernan
19603.
́
dez-García, M. J. Phys. Chem. B 2005, 109, 19595−
(32) Claus, P.; Lucas, M.; Lucke, B.; Berndt, T.; Birke, P. Appl. Catal.,
̈
A 1991, 79, 1−18.
(66) Iwasa, N.; Takezawa, N. Bull. Chem. Soc. Jpn. 1991, 64, 2619−
2623.
(33) Lopez, T.; Manríquez, M. E.; Gomez, R.; Campero, A.; Llanos,
M. E. Mater. Lett. 2000, 46, 21−29.
́
(67) Marchi, A. J.; Fierro, J. L. G.; Santamaría, J.; Monzon, A. Appl.
(34) Ron, H.; Rubinstein, I. J. Am. Chem. Soc. 1998, 120, 13444−
13452.
Catal., A 1996, 142, 375−386.
(68) Carotenuto, G.; Tesser, R.; Di Serio, M.; Santacesaria, E.
Biomass Convers. Biorefin. 2013, 3, 55−67.
(35) Rosmaninho, M. G.; Moura, F. C. C.; Souza, L. R.; Nogueira, R.
K.; Gomes, G. M.; Nascimento, J. S.; Pereira, M. C.; Fabris, J. D.;
(69) Niiyama, H.; Morii, S.; Echigoya, E. Bull. Chem. Soc. Jpn. 1972,
45, 655−659.
Ardisson, J. D.; Nazzarro, M. S.; Sapag, K.; Araujo, M. H.; Lago, R. M.
́
Appl. Catal., B 2012, 115−116, 45−52.
(36) Hong, Z.; Cao, Y.; Deng, J. Mater. Lett. 2002, 52, 34−38.
(37) Angelici, C. Ethanol-to-Butadiene Conversion over SiO2-MgO
Catalysts: Synthesis-Structure-Performance Relationships. PhD Thesis,
Utrecht University, 2015.
(38) Travert, A.; Vimont, A.; Sahibed-Dine, A.; Daturi, M.; Lavalley,
J.-C. Appl. Catal., A 2006, 307, 98−107.
(39) Larina, O. V.; Kyriienko, P. I.; Soloviev, S. O. Catal. Lett. 2015,
145, 1162−1168.
(40) Angelici, C.; Velthoen, M. E. Z.; Weckhuysen, B. M.; Bruijnincx,
P. C. A. Catal. Sci. Technol. 2015, 5, 2869−2879.
(41) Ono, Y.; Hattori, H. Solid Base Catalysis; Springer-Verlag and
Tokyo Institute of Technology Press: Berlin, Heidelberg, Tokyo, 2010.
́
́
(42) Aguila, G.; Gracia, F.; Cortes, J.; Araya, P. Appl. Catal., B 2008,
77, 325−338.
(43) Shimizu, K.; Maeshima, H.; Yoshida, H.; Satsuma, A.; Hattori,
T. Phys. Chem. Chem. Phys. 2000, 2, 2435−2439.
(44) Velu, S.; Wang, L.; Okazaki, M.; Suzuki, K.; Tomura, S.
Microporous Mesoporous Mater. 2002, 54, 113−126.
(45) Stakheev, A. Y.; Shpiro, E. S.; Apijok, J. J. Phys. Chem. 1993, 97,
5668−5672.
(46) Okada, K.; Kameshima, Y.; Yasumori, A. J. Am. Ceram. Soc.
1998, 81, 1970−1972.
(47) Bennici, S.; Gervasini, A.; Ravasio, N.; Zaccheria, F. J. Phys.
Chem. B 2003, 107, 5168−5176.
(48) Sagata, K.; Imazu, N.; Yahiro, H. Catal. Today 2013, 201, 145−
150.
(49) Francisco, M. S. P.; Mastelaro, V. R.; Nascente, P. A. P.;
Florentino, A. O. J. Phys. Chem. B 2001, 105, 10515−10522.
(50) Ghijsen, J.; Tjeng, L. H.; van Elp, J.; Eskes, H.; Westerink, J.;
Sawatzky, G. A.; Czyzyk, M. T. Phys. Rev. B: Condens. Matter Mater.
Phys. 1988, 38, 11322−11330.
(51) Deka, U.; Juhin, A.; Eilertsen, E. A.; Emerich, H.; Green, M. A.;
Korhonen, S. T.; Weckhuysen, B. M.; Beale, A. M. J. Phys. Chem. C
2012, 116, 4809−4818.
(52) Huang, H.-L.; Wang, H. P.; Wei, G.-T.; Sun, I.-W.; Huang, J.-F.;
Yang, Y. W. Environ. Sci. Technol. 2006, 40, 4761−4764.
(53) Choy, J.; Kim, D.; Hwang, S.; Demazeau, G. Phys. Rev. B:
Condens. Matter Mater. Phys. 1994, 50, 16631−16639.
(54) Acharya, A.; Behera, R.; Roy, G. S. Lat. Am. J. Phys. Educ. 2012,
6, 402−406.
6015
DOI: 10.1021/acscatal.5b00755
ACS Catal. 2015, 5, 6005−6015