G Model

CATTOD-10171; No. of Pages 10 ARTICLE IN PRESS

Catalysis Today xxx (2016) xxxxxx

Contents lists available at ScienceDirect

Catalysis Today
journal homepage: www.elsevier.com/locate/cattod

On the nature of the active sites in the selective oxidative

esterication of furfural on Au/ZrO2 catalysts
C. Ampelli a , K. Barbera a , G. Centi a , C. Genovese a , G. Papanikolaou a , S. Perathoner a, ,
K.J. Schouten b , J.K. van der Waal b
a
Universit di Messina, ERIC aisbl and CASPE/INSTM, Sect. Industrial Chemistry, V.le F. Stagno DAlcontres 31, 98166 Messina, Italy
b
Avantium Chemicals, Zekeringstraat 29, 1014 BV Amsterdam, The Netherlands

a r t i c l e i n f o a b s t r a c t

Article history: Two catalysts, based on gold on zirconia and prepared in the same way, except for the different thermal
Received 30 December 2015 treatment in air after the addition of gold (0.4% wt), show a quite different behavior in furfural oxidative
Received in revised form 25 March 2016 esterication, particularly in the selectivity to 2-MF (methyl 2-furoate). The analysis of their catalytic
Accepted 19 April 2016
behavior, in comparison with the ZrO2 support and a reference Au/TiO2 catalyst (AUROliteTM ), provides
Available online xxx
a series of indications on the reaction mechanism and the nature of the sites. It is evidenced in particular
that i) the addition of gold enhances the activity of the support in furfural conversion by a factor of
Keywords:
about ten, ii) the creation of active sites for the further conversion of the acetal intermediate to 2-MF
Selective oxidative esterication
Furfural
increases the reaction rate of furfural conversion, mainly by shifting the equilibrium of the furfural to
Methyl 2-furoate acetal reaction to right, iii) the precursor sites for the acetal to 2-MF oxidative conversion form during the
Gold supported catalysts thermal treatment in air from 200 C to 400 C, due to a reduction of gold occurring during this treatment
Zirconia with partial redispersion of gold and formation of very small Au NPs (nanoparticles) with average size of
0.7 nm, iv) the active sites for the acetal to 2-MF oxidative conversion are associated likely to these small
Au NPs, but they form in-situ at the initial stage of reaction.
2016 Elsevier B.V. All rights reserved.

1. Introduction to high-added-value products. For example, the oxidative esteri-

cation of furfural
Sugar chemistry is at the core of biorenery/biofactories [14],
because sugars are the units of polysaccharides (among which cel-
lulose and hemi-cellulose, starch, glycogen), besides to be present
as mono-, di- or oligo-saccharides in many biomass varieties. While forms methyl 2-furoate, which is a specialty fragrance, but it is
the manufacture of energy vectors may rise various objections, the also used for different purposes in ne chemical industry. On the
production of chemicals and especially high-added-value products other hand, HMF can be converted to the dimethyl ester of FDCA
is, and will be in the future, an important element of the bioren- (2,5-furandicarboxylic acid), a suitable replacement for tereph-
ery/biofactories scenario [46]. Monosaccharides can be classied thalic acid (TPA) and isophthalic acid [7,8].
in two main classes: aldoses, where an aldeidic group is present Oxidative esterication is realized currently using mostly
or ketoses, where a ketonic group is present. Isomerization or transition-metal or organo homogeneous catalysts [9,10], but the
dehydration of the latter may also lead to aldehydes, for example use of heterogeneous catalysts is preferable to intensify the process
HMF (5-hydroxymethyl furfural) can be produced by dehydration and reduce costs as well as the environmental impact. Christensen
of fructose and furfural by dehydration of xylose. Moreover, the and coworkers [11] rst introduced the use of supported gold cata-
conversion of the aldehydic group to an ester via oxidative esteri- lysts for the furfural and HMF oxidative conversion. Later, Casanova
cation is a relevant possibility for the conversion of these chemicals et al. [12] reported the superior performances of gold on ceria for
the conversion of HMF in methanol. Pinna and coworkers [1318]
used catalysts based on gold supported on various oxides (ceria,
titania, zirconia) for both furfural and HMF oxidative esterica-
tion. Li et al. [19] used gold supported on Ce-Zr in the oxidative
Corresponding author esterication of different benzylic aldehydes, indicating that the
E-mail address: [email protected] (S. Perathoner).

http://dx.doi.org/10.1016/j.cattod.2016.04.023
0920-5861/ 2016 Elsevier B.V. All rights reserved.

Please cite this article in press as: C. Ampelli, et al., On the nature of the active sites in the selective oxidative esterication of furfural
on Au/ZrO2 catalysts, Catal. Today (2016), http://dx.doi.org/10.1016/j.cattod.2016.04.023
G Model
CATTOD-10171; No. of Pages 10 ARTICLE IN PRESS
2 C. Ampelli et al. / Catalysis Today xxx (2016) xxxxxx

incorporation of Zr4+ into the ceria lattice improves the activa- ammonia solution was added under vigorous stirring until a pH
tion of methanol (to form methoxy species) and facilitates the -H of 8.6 was reached. After the complete precipitation, the hydroxide
elimination of hemiacetal, an intermediate in the reaction. Gold- suspension was aged for 20 h at 90 C. The aged hydroxide was l-
graphene nanocomposite catalysts were also reported recently to tered and washed with warm water until removal of chloride ions
be highly efcient in the oxidative esterication [20], although in (conrmed by AgNO3 test). The solid was dried at 110 C overnight
contrast with the role of the support indicated by Li et al. [19]. and calcined at 650 C for 3 h.
The optimal type of support in the reaction is also unclear. About 0.4 wt.% of gold was added by deposition-precipitation
Menegazzo et al. [15] indicated zirconia as the best support for (DP) at pH = 8.6. In a typical DP technique, the oxide support (2 g)
gold nanoparticles by comparing the performances of ceria, tita- was suspended in 80 mL of an aqueous solution containing the gold
nia and zirconia based catalysts. However, later the same authors precursor (HAuCl4 3H2 O, in a concentration corresponding to a the-
[16] remarked the use of Au-CeO2 catalysts and Au supported on oretical Au loading of 1% wt) for 3 h and the pH was adjusted by the
sulphated ZrO2 catalysts [17]. Au nanoparticles supported on ceria- addition of aqueous NaOH (0.5 M). After deposition of gold onto the
alumina mixed oxides were instead suggested recently as active support (in the form of gold-hydroxide), the solid was washed, l-
catalysts by Smolentseva et al. [21], indicating that the oxygen stor- tered and dried at room temperature for 12 h and nally annealed
age capacity of the support (possible only in ceria-based catalysts) at two different temperatures (200 C, 400 C) in air. These catalysts
is the key element for the catalytic performances. In contrast with are indicated hereafter as AuZ200 and AZ400, respectively.
this conclusion, Wang et al. [22] showed that a support like MgO for
gold nanoparticles, thus without oxygen storage capacity and acid 2.2. Characterization of the catalysts
sites, has also excellent performances in the oxidative esterication
of aldehydes. Suzuki et al. [23] showed that supported gold-nickel Atomic absorption spectroscopy (AAnalyst 200 Perkin Elmer)
oxide nanoparticles have excellent performances in the oxidative was used to determine the gold amount after dispersion of the sam-
esterication. In this case, a core-shell structure is present, with ples in HF by ultrasounds. The phase composition and the degree
the Au nanoparticles at the core and the surface covered by highly of crystallinity were analysed by X-ray diffraction analysis with a
oxidized NiOx. This result may thus question the idea proposed by ADP 2000 diffractometer using a Cu-K radiation. Data were col-
many authors [24] that sites at the gold nanoparticle-support inter- lected at a scanning rate of 0.025 s1 in a 2 range from 15 to 80 .
face are those active in selective oxidation reactions. Mononuclear Diffraction peak identication was made on the basis of the JCPDS
gold species present in the solution, deriving from the dissolution database of reference compounds.
of gold nanoparticles, are instead proposed by Hashmi et al. [25] as Transmission electron microscopy (TEM) images were acquired
the active species in the oxidative esterication. by using a Philips CM12 microscope (resolution 0.2 nm) with
There is thus a relevant interest on this reaction, although still an accelerating voltage of 120 kV, while the elemental analysis
aspects are under discussion related to the exact nature of the active was investigated by using a Philips XL-30-FEG scanning electron
sites and the reaction mechanism. In a high-throughput study on microscope equipped with an energy dispersive X-ray analyser.
this reaction [26] we observed that gold on zirconia is effectively Ultraviolet visible diffuse reectance spectra were recorded by a
the catalyst giving the best performances under diluted conditions, Jasco V570 spectrometer equipped with an integrating sphere for
but with the performances strongly depending on the modalities solid samples using BaSO4 as reference. BET surface area and pore
of preparation of the zirconia support. In addition, tests using a size distributions were determined using N2 adsorption/desorption
much higher furfural to methanol ratio (no solvent and bases), as isotherm at 77 K with an Autosorb iQ (Quantachrome Instruments)
necessary for the industrial development of the process, resulted system.
in a quite different ranking of the catalysts. XPS measurements were performed on a PHI Quantera (Physical
The aim of this work is to investigate in detail two Au-ZrO2 cat- Electronics). The survey and high resolution spectra were taken
alysts, with the second derived from the rst, calcined at 400 C using the large area analysis which is a 100 W 100 m diameter
rather than at 200 C. Giving these two catalysts a quite different X-ray beam scanned over a 1300 100 m area at high speed. High
behavior in the oxidative esterication of furfural, the comparison resolution spectra were obtained with a pass energy of 26 eV. The
of their physico-chemical properties provides an excellent basis to energy scale for all samples were adjusted by placing the C 1 s peak
derive indications on the nature of the active sites in this reaction, from adventitious carbon at 284.8 eV. The spectra were tted by
and about the possible role of the support as well as of the reac- software and the resulting peak position, FWHM and percentage of
tion conditions. In order to allow a more reliable comparison of the total area are given.
the catalytic performances of these samples, the reactivity of the FTIR measurements were carried out on self-supporting wafers
reference AUROliteTM Au/TiO2 catalyst (a commercially available and the spectra were collected at a resolution of 4 cm1 with a
standard reference gold-based catalyst) is also reported. Bruker IF55 FTIR spectrophotometer equipped with a MCT detec-
tor. The self-supporting wafers were prepared and activated under
vacuum (104 Torr) for 2 h at 200 C in an IR cell allowing in situ
2. Experimental thermal treatments, and pyridine dosage. Lewis and Brnsted acidic
sites quantication was performed by integrating the area under-
2.1. Preparation of the catalysts neath the bands at 1450 and 1545 cm1 . The integrated molar
extinction coefcients (IMEC) were calculated, according to the
AUROliteTM Au/TiO2 (1 wt.% Au) was a commercial sample pro- Beers law, by dosing pyridine (Py) at increasing concentrations
vided by Strem Chemicals Inc. in the form of small extrudates. over each sample under study, after evacuation at rt under high
Before testing the catalytic activity, the pellets were crushed in vacuum: A = cD, where A = absorbance, = molar extinction coef-
a mortar until a very ne powder was obtained and then sieved cient, c = concentration and D = path length. The value was
to particle size distribution of 120250 mesh, i.e. 0.0630.125 mm. obtained by plotting A vs c (mol Py), assuming a constant D (since
This particle size distribution was used also for the other catalysts. the sample is in a pellet form and the thickness is always constant).
All the reagents were of analytical grade and bought from Sigma Hence, Brnsted and Lewis acid sites were quantied using the
Aldrich. ZrO2 was prepared by chemical precipitation method. following equation:
Zirconium(IV) oxychloride octahydrate (ZrOCl2 8H2 O), used as zir-
conia source, was dissolved in distilled water, then a diluted CPy onXsites = (AreapelletIA(X))/(IMEC(X)Weightpellet) (1)

Please cite this article in press as: C. Ampelli, et al., On the nature of the active sites in the selective oxidative esterication of furfural
on Au/ZrO2 catalysts, Catal. Today (2016), http://dx.doi.org/10.1016/j.cattod.2016.04.023
G Model
CATTOD-10171; No. of Pages 10 ARTICLE IN PRESS
C. Ampelli et al. / Catalysis Today xxx (2016) xxxxxx 3

Table 1 AuZ400 catalysts, in comparison with the AUROliteTM reference

Reaction conditions used for tests in oxidative esterication reaction in a 250 mL
catalyst and zirconia alone (i.e. without gold) support.
Parr stainless steel autoclave (teon-lined).
Zirconia alone shows some activity, although low and with low
Solvent 150 mL MeOH selectivity to 2-MF. Nearly zero selectivity was instead observed
Catalyst loading 100 mg without the catalyst, thus indicating that 2-MF does not derive
Gas 6 bar O2 from auto-oxidation reactions, always possible in oxidation reac-
Temperature 120 C tions with aldehydes. The AuZ200 catalyst shows a signicantly
Furfural concentrations 0.02 M
Time of reaction 06 h
higher conversion of furfural, although a similar low selectivity to
2-MF. The main side product observed is the acetal (2-furaldehyde-
dimethyl-acetal). The pseudo-rst order rate constant of furfural
conversion is about 10 times higher for AuZ200 with respect to
where C = mol gcat 1 and IA(X) = integrated absorbance of the
ZrO2 . AuZ400 is more active than AuZ200, with the pseudo-rst
peak of the species X.
order rate constant of furfural depletion about ve times higher,
and about fty times higher than that of ZrO2 . However, the most
2.3. Testing of the catalysts relevant difference is that the selectivity to 2-MF is denitively
higher, with a correspondent decrease in the acetal formation. Ref-
The catalytic tests were performed in a 250 mL Parr stain- erence AUROliteTM catalyst shows a furfural conversion in between
less steel autoclave (Teon-lined). As a typical procedure, furfural that shown by AuZ200 and AuZ400 catalysts, but the selectivity
(0.02 M) was dissolved in 150 mL methanol under 6 bar of O2 , at to 2-MF is only slightly higher than that of AuZ200. Note that the
120 C and with 1000 rpm of stirring. The progress of the reaction amount of gold in AUROliteTM is about 2.5 times higher than in our
was monitored for 6 h by collecting samples as a function of the gold on zirconia samples, the latter having an Au loading three-four
time of reaction. Reaction conditions used for tests in oxidative times lower than that used typically in literature. Acetal is also the
esterication reaction in this lab-scale apparatus are summarized main product observed for AUROliteTM .
in Table 1. Analysis was performed by a GC-FID (Finningan Trace Acetal (geminal-diether derivative) stems from the addition of
GC Ultra, with a Restek Rx-i-5 column) and by a GCMS (Thermo a methoxy group to C O group in furfural converting thus the alde-
Scientic GC Trace 1310 ISQ MS with a Restek Stabilwax column). hyde (-CHO) group in furfural to an acetal group ((OCH3 )2 ). This
is an acid-catalyzed reaction, which can occur when the C O in the
3. Results aldehyde is activated, for example by coordination with Lewis acid
sites. The carbon atom in the activated C O bond is then susceptible
As part of a broader study on the development of an industrial for nucleophilic attack by the alcohol, followed by deprotonation
process for the synthesis of methyl 2-furoate from furfural, some to form a hemiacetal, protonation of the alcohol, dehydration and
results of which were reported elsewhere [26], we found that the further nucleophilic attack by the alcohol and deprotonation. It is
two Au/ZrO2 catalysts, prepared in the same way, except for a nal thus reasonable that zirconia alone is able to form the acetal, but
stage of calcination at 200 C or 400 C after the addition of gold when gold nanoparticles are present, Lewis acid sites likely at the
(about 0.4% wt.), resulted in a quite different catalytic behavior. gold-support interface (metallic gold nanoparticles may attract OH
These two samples (AZ200 and AZ400) thus offer a unique oppor- groups from Zr-OH sites at the interface, thus creating a Lewis acid
tunity to understand the possible nature of the active sites, with site), can activate the C O bond and thus promote the furfural con-
respect to other studies where samples with quite different char- version. The main difference of the AuZ400 sample with respect
acteristics, composition, and preparation method are compared. to the other catalysts is the presence of additional sites able to
In addition, we report here the catalytic behavior of a standard catalyse the oxidation of acetal to 2-MF. These sites are not related
reference gold-based catalyst (AUROliteTM ). In fact, the catalytic to the generic presence of gold NPs (nanoparticles), present also
performances of gold-based catalysts depend signicantly on the in AuZ200 and AUROliteTM (2.5 times higher in the latter), but to
experimental conditions and reproducibility of the results may be unique sites which generate in AuZ400 during thermal treatment
sometimes difcult. This AUROliteTM reference (and commercially from 200 C to 400 C.
available) sample allow to check the performances and understand The TON (turnover number, in h1 ) which can be estimated,
eventual deviations related to the use of different reaction condi- assuming all atoms in gold nanoparticles active in the conversion,
tions. is reported in Table 3. Considering the Au nanoparticles spherical,
and the average particles size determined by TEM (transmission
electron microscopy, see later), it is possible to determine the aver-
3.1. Catalytic performances age surface to volume ratio (S/V) and circumference to volume ratio
(C/V). These corrective factors may be considered when only sur-
The catalytic tests were made in a well-stirred autoclave under face sites (S/V) or at the perimetral edge between Au nanoparticles
reaction conditions reported in Table 1, which are close to those and the support (C/V) are active in the reaction. However, being
typically used in literature [1119], in order to have comparable likely only a fraction of the gold atoms active in the reaction, these
indications. Table 2 instead reports the results in the oxidative TON estimations may be relevant only for comparison with other
esterication of furfural in terms of conversion (after 6 h of reac- catalysts, while effective TON values will be certainly higher. Data
tion) and selectivity to 2-MF (methyl 2-furoate) for AuZ200 and

Table 2
Catalytic performances of catalysts in the oxidative esterication of furfural. Reactions conditions as in Table 1, time of reaction 6 h.

Catalyst Thermal treatment, C (a ) Au loading, wt.% Conversion of furfural, % Selectivity to 2-MF, %

ZrO2 5 10
AuZ200 200 0.4 0.1 63 10
AuZ400 400 0.4 0.1 99 95
AUROliteTM 250 1.0 0.1 73 31
a
In air, after addition of gold. For AUROliteTM refers to the calcination before to the catalytic tests.

Please cite this article in press as: C. Ampelli, et al., On the nature of the active sites in the selective oxidative esterication of furfural
on Au/ZrO2 catalysts, Catal. Today (2016), http://dx.doi.org/10.1016/j.cattod.2016.04.023
G Model
CATTOD-10171; No. of Pages 10 ARTICLE IN PRESS
4 C. Ampelli et al. / Catalysis Today xxx (2016) xxxxxx

Table 3
Turnover number (TON) in gold catalysts, determined considering all atoms in gold
nanoparticles active in the conversion (reactions conditions as in Table 1). The
100
AuZ200
AuZ400
a
average surface to volume ratio (S/V) and circumference to volume ratio (C/V) AUROlite
of gold particles (supposed spherical, and with average diameter given by TEM 80
measurements).

conversion, %
Catalyst TON in furfural TON in 2-MF S/V, nm1 C/V, nm2
60
conversion, h1 formation, h1

AuZ200 155.1 15.5 1.25 0.26

AuZ400 243.7 231.6 2.73 1.24 40
AUROliteTM 71.9 22.3 1.30 0.28

20
25

0
20 0 100 200 300 400
2-MF time, min
concentration, [mM]

100
15 AZ200
AZ400 b
AUROlite
80
10

yield 2-MF, %
60
5
furfural
40

0
0 50 100 150 200 250 300
20
time, min

Fig. 1. Concentration of furfural (2-FA) and methyl-2-furoate (2-MF) vs time of 0

reaction during the catalytic test of AuZ400. Reaction conditions as in Table 1. 0 100 200 300 400

time, min
in Table 3 evidence that catalysts based on zirconia supports are
Fig. 2. Conversion (a) and yield to 2-MF (b) vs time of reaction in the oxidative esteri-
more active in furfural conversion than the AUROliteTM reference cation of furfural on AuZ200, AuZ400 and AUROliteTM catalysts. Reaction conditions
per g of Au, but especially AuZ400 shows about one order of mag- as in Table 1.
nitude higher TON in 2-MF formation. This higher activity remains
when surface or perimetral edge sites are considered.
The change in the concentration of furfural and 2-MF with the drastic increase in the rate of conversion of furfural is observed
time of reaction for the AuZ400 catalyst is reported in Fig. 1. There particularly for the AuZ400 catalyst. This indication well supports
is a progressive conversion of furfural, with a parallel increase in the idea that the sites responsible for the conversion of furfural to
2-MF formation. To note that the formation of acetal or other inter- the intermediate acetal are nearly the same on the three catalysts,
mediates is not observed, if not in minimal part, and the selectivity but during reaction a unique type of site (able to convert acetal to
to 2-MF does not change signicantly with time on stream. This 2-MF) generates only on AuZ400 catalyst. The generation of these
indicates that the conversion of acetal to 2-MF is faster than acetal sites accelerate the rate of reaction by shifting the furfural-acetal
formation, which is the rate limiting step. The higher rate of con- equilibrium to the right. The results in terms of yield of 2-MF versus
version of furfural on AuZ400 with respect to AuZ200 could be time of reaction (Fig. 2b) are in good agreement with the above
explained considering reversible the acetal formation. By catalysing interpretation.
its further conversion to 2-MF the equilibrium is thus shifted to the
right and this explains the higher rate of furfural conversion on 3.2. Characterization
AuZ400.
In Fig. 1 it should be noted the presence of an apparent initial The ZrO2 sample prepared by precipitation shows relatively
induction time at the start of the reaction, particularly evident for high surface area (BET, 95.6 m2 g1 ) with a pore volume of
2-MF formation. In batch-type reactors, an initial induction time 0.25 cm3 g1 . No signicant changes in the textural aspects were
(necessary to establish a well-mixing in the system) is typically present after addition of gold and further thermal treatment
observed. In our reactor, this induction times should be less than (AuZ200 and AuZ400 samples). For comparison, the AUROliteTM
about 5 min, thus shorter than the effect observed in Fig. 1. It is Au-TiO2 sample has a surface area of about 50 m2 g1 .
thus likely that the induction time observed in Fig. 1 is related to The phase composition of AuZ samples was investigated by X-
an in-situ (during the reaction) modication to generate the active ray diffraction and the obtained patterns are shown in Fig. 3 in the
sites for 2-MF formation. 2 region between 15 and 80 .
This interpretation is supported from the comparison of the The spectra are very similar corresponding to tetragonal zir-
curves of conversion of furfural versus the time of reaction on conia, without apparent presence of Au (expected at about 38.0
AuZ200, AuZ400 and AUROliteTM catalysts (Fig. 2a). It is evident and 44.3 for most intense (111) and (200) planes of metallic gold
that the rates of furfural conversion are nearly the same on the respectively). The absence of detection of Au is likely due to the very
three catalysts during about the rst hour of reaction, and then a small amount of gold (0.4% wt) and presence of small nanocrys-

Please cite this article in press as: C. Ampelli, et al., On the nature of the active sites in the selective oxidative esterication of furfural
on Au/ZrO2 catalysts, Catal. Today (2016), http://dx.doi.org/10.1016/j.cattod.2016.04.023
G Model
CATTOD-10171; No. of Pages 10 ARTICLE IN PRESS
C. Ampelli et al. / Catalysis Today xxx (2016) xxxxxx 5

our zirconia preparation, giving only tetragonal phase, is different

from that earlier reported in literature.
TEM (transmission electron microscopy) analysis was used to
investigate the morphology of AuZ samples and to analyse the effect
of the calcination temperature after the gold deposition. Selected
results are reported in Fig. 4.
Round-like shaped gold nanoparticles (NPs) are present in both
AuZ200 (see Fig. 4a and b) and AuZ400 (see Fig. 4c and d) samples,
but most of the Au NPs are slightly larger in AuZ200 with respect
to AuZ400. Higher magnication images of a Au NP in AuZ400 well
evidence the lattice planes (111) of a crystalline gold particle (dia.
6 nm), with a lattice spacing of 2.35 (see Fig. 4d) [28]. There
are no evidences of a specic modication by interaction with the
support for Au NPs above about 23 nm in both AuZ200 and Au400
samples, neither specic evidences of a difference in these Au NPs
(i.e. larger than 23 nm) between these two samples, except for
slightly larger sizes in those in AuZ200 sample.
However, a more careful analysis of the TEM images reveals
the presence of smaller Au particles (in the range 0.31.5 nm, as
Fig. 3. XRD patterns of Au/ZrO2 samples thermally treated at 200 and 400 C. shown in Fig. 5) in the AuZ400 samples. The distribution of gold
particle sizes is relatively narrow and the average diameter was
tals. The peaks observed at 2 = 30.4 , 35.3 , 50.6 , 60.5 , 63.0 and estimated to be 0.7 nm. These small Au nanoparticles have a sim-
74.8 can be assigned to (011), (110), (112), (202), (121) and (220) ilar size distribution in AuZ200 sample (see Fig. 1S), but they are
planes, respectively, of tetragonal ZrO2 (JCPDS le No. 50-1089). in a much smaller amount than for AuZ400. The average gold crys-
Normally, the monoclinic zirconia phase is obtained by calcina- tallite size in AUROliteTM is instead 23 nm, as also reported by
tion of the tetragonal phase at high temperatures, with complete Strem Chemicals. These high-resolution TEM images instead do not
transformation above about 700 C [27]. In our case (calcination at reveal differences in the zirconia support showing in both cases
650 C for 3 h) we do not observe monoclinic phases, as evidenced well-crystallized particles without relevant presence of amorphous
by XRD data (Fig. 3). Pinna and co-workers [14,15,17,18] reported phases and crystal sizes of about 1020 nanometers, in agreement
only in the more recent papers [18] indications about the differ- with XRD indications.
ent zirconia phases present in their samples. They indicated that Fig. 6 reports the visible diffuse reectance spectra of ZrO2 ,
both tetragonal and monocline phases are present, with the ratio AuZ200 and AuZ400 samples. ZrO2 , as expected, does not show
tetragonal/monocline phases of zirconia increasing with a sulpha- bands in the visible region, only the tail end (below about 350 nm)
tion treatment, being the presence of sulphate ions on the support of the metal-to-oxygen charge transfer (CT) bands present in UV
favouring the stabilization of the metastable tetragonal phase of region. An intense absorption is instead present in both AuZ200
zirconia during the calcination step. We could thus indicate that and AuZ400 samples, centred at 549 and 541 nm, respectively. It is

Fig. 4. Selected TEM images of AuZ200 sample (a and b) and AuZ400 sample (c and d). Fig. 4d shows the (111) lattice planes of a crystalline Au particle.

Please cite this article in press as: C. Ampelli, et al., On the nature of the active sites in the selective oxidative esterication of furfural
on Au/ZrO2 catalysts, Catal. Today (2016), http://dx.doi.org/10.1016/j.cattod.2016.04.023
G Model
CATTOD-10171; No. of Pages 10 ARTICLE IN PRESS
6 C. Ampelli et al. / Catalysis Today xxx (2016) xxxxxx

Fig. 5. Selected TEM images of AuZ200 and AuZ400 samples, the latter evidencing the presence of Au NPs below 2 nm. The estimated particle distribution of these Au NPs is
reported on the right.

2,0 is expected for Au NPs with smaller sizes [29,30]. The SPR band near
540550 nm is consistent with particles of around 24 nm [32], as
indicated by TEM data. The slight blue shift in AuZ400 with respect
to AuZ200 is consistent with the indication of slightly smaller par-
1,5
ticles in the latter and/or a slight change in the interaction of Au
absorbance, a. u.

541 nm NPs with the zirconia support. AuZ400 shows, in addition, a band
400 nm centred near 400 nm, not well resolved due to the tails of the CT
1,0 bands of the support and the SPR band. However, the presence of
AuZ400 this band can be clearly evidenced from the comparison between
549 nm AuZ200 and AuZ400 spectra. The presence of this further band in
0,5 AuZ400 with respect to AuZ200 is in agreement with the additional
AuZ200 presence of very small Au NPs in the former.
X-ray photoelectron spectroscopy (XPS) measurements were
ZrO2
0,0 made in order to analyse the surface characteristics of the AuZ200
and AuZ400 samples. The results for Au 4f spectral region are
300 400 500 600 700 800
reported in Fig. 7. While AuZ samples shows the typical bands at
wavelenght, nm 83.6 and 87.3 eV for Au04 f7/2 and 4 f5/2 core levels, additional relative
intense components at 85.0 and 88.7 eV could be observed in the
Fig. 6. Visible absorption diffuse reectance spectra of ZrO2 , AuZ200 and AuZ400 deconvoluted spectra of AuZ200 sample. Intensity of these compo-
samples.
nents is around half of that of Au0 species, and they are indicative
of a Au+ type of species [33,34].
The XPS spectra in the Zr 3 d and O 1 s region for AuZ200 and
well-known that Au NPs on a dieletric support give rise to surface
AuZ400 samples are shown in Fig. 8, together with their deconvo-
plasmon resonance (SPR) in the visible region [29]. Au NPs strongly
lution. The two samples are nearly indistinguishable both in Zr 3 d
interact with the incident light, owing to the localised surface
and in O 1 s regions. The position of Zr 3 d5/2 and 3 d3/2 bands are well
plasmon resonance (LSPR) effect, which represents the strong oscil-
in agreement with those expected for Zr4+ and there is no indication
lation of the metals free electrons in phase with the varying electric
of the presence of Zr3+ or other species. The spectra in O 1 s region,
eld of the incident light. The optical absorption properties in the
evidences the presence of OH species at 531.5 eV together with
visible range for Au nanoparticles are represented by intraband
the main ZrO2 lattice oxygen at 530.1 eV. The small peak at 532.7 eV
LSPR (Localized Surface Plasmon Resonance) due to the interband
is related to oxygen in some contaminating organic species. It may
d sp transition. The LSPR band intensity and wavelength depends
be noted that relative intensities of lattice to hydroxyl O 1 s bands
on the factors affecting the electron charge density on the particle
are the same in AuZ200 and AuZ400 samples, and similar to that
surface such as the metal type, particle size, shape, structure, com-
present in ZrO2 alone (not reported here). This indicates that the
position and the dielectric constant of the surrounding medium
addition of Au and the subsequent thermal treatment does not
[30]. Au NPs in the 515 nm range on zirconia (prepared by a similar
induce changes in the surface acidity, neither the formation of
method of that used in our case) give rise to a SPR band moving from
reduced zirconia species, at least detectable by XPS.
about 570595 nm on increasing the Au loading [31], but a blue shift

Please cite this article in press as: C. Ampelli, et al., On the nature of the active sites in the selective oxidative esterication of furfural
on Au/ZrO2 catalysts, Catal. Today (2016), http://dx.doi.org/10.1016/j.cattod.2016.04.023
G Model
CATTOD-10171; No. of Pages 10 ARTICLE IN PRESS
C. Ampelli et al. / Catalysis Today xxx (2016) xxxxxx 7

Fig. 7. XPS spectra of AuZ200 (a) and AuZ400 (b) samples in the Au 4 f region, with indication of the deconvoluted spectra.

Fig. 8. XPS spectra of AuZ200 and AuZ400 samples in the Zr 3 d and oxygen 1 s regions, with indication of the deconvoluted spectra.

Please cite this article in press as: C. Ampelli, et al., On the nature of the active sites in the selective oxidative esterication of furfural
on Au/ZrO2 catalysts, Catal. Today (2016), http://dx.doi.org/10.1016/j.cattod.2016.04.023
G Model
CATTOD-10171; No. of Pages 10 ARTICLE IN PRESS
8 C. Ampelli et al. / Catalysis Today xxx (2016) xxxxxx

Table 4 1.25
Amount of Lewis and Brnsted acid sites of ZrO2 prepared by precipitation, after
recording the adsorption of pyridine and subsequent evacuation at room tempera-
AZ_200
ture and 150 C. AZ_400

Evaluation at Lewis acid sites (mol/gcat ) Brnsted acid sites (mol/gcat )

1922
RT 254 126
1.20

Adsorbance (a.u.)
150 C 202 133
1870
1983
1.8 1960
AZ_400
ZrO2
AZ-200 1.15
AZ200
1.5
1445 2130
Adsorbance (a.u.)

AZ400
1.2
1.10
1613 1425 2150 2100 2050 2000 1950 1900 1850 1800
-1

1445
AZ400 Wavenumber (cm )
0.9 1545
1575 Fig. 10. FT-IR spectra of AuZ200 and AuZ400 in the 21501800 cm1 spectral region.
AZ200
fundamental vibrations [39], which may be enhanced in the pres-
0.6
ZrO2 ence of -back bonding between pyridine and Au+ ions [40,41].
It is thus reasonable that their detection only in AuZ200 may be
1700 1650 1600 1550 1500 1450 1400 1350
related to the presence of cationic gold species, evidenced also to
-1
Wavenumber (cm ) be present by XPS only in AuZ200.

Fig. 9. FT-IR spectra of ZrO2 , AuZ200 and AuZ400 in the 13501700 cm1 spectral
region. 4. Discussion

The catalytic results of ZrO2 , AuZ200, AuZ400 and Au/TiO2

The acid properties were studied by Fourier-Transformed (AUROliteTM ) catalysts evidence that the presence of gold nanopar-
InfraRed (FTIR) spectroscopy using pyridine as probe molecule. ticles enhances the rate of furfural conversion to the acetal
Table 4 reports the quantication of the amount of Lewis and intermediate with respect to zirconia alone. The acetal derives from
Brnsted acid sites of ZrO2 , after recording the adsorption of pyri- the addition of methanol to carbon atoms in the C O aldehydic
dine and subsequent evacuation at room temperature and 150 C. group, activated by interaction likely with surface Lewis acid sites.
Quantication was performed by integrating the areas underneath These sites are present also in ZrO2 , as shown by IR data with
bands at 1545 and 1455 cm1 , assigned to the pyridinium ion pyridine chemisorption (Table 4), but their amount is increased
(PyH+ ) formed on the Brnsted acid site and pyridine coordinated after addition of gold nanoparticles, and thus likely at the bound-
to Lewis acid sites, respectively. The normalization has been per- ary between Au NPs and the support, due to effect of metallic gold
formed according to the density of the pellet (dimension/weight). NPs (related to Fermi level) in attracting OH groups in nearlying
The FTIR background spectrum of ZrO2 (Fig. 9) is characterized Zr-OH sites.
by a sharp band at 1613 cm1 (O H bending vibration of absorbed While general characteristics of AuZ200 and AuZ400 samples
water molecules), a sharp peak at 1445 cm1 rising from a broad are quite similar, only AuZ400 shows the presence of additional
peak in the 14251375 cm1 region and a broad peak centred at active sites, likely formed in-situ at the beginning of the catalytic
1555 cm1 . The latter peaks cannot be attributed to contamination reaction, able to oxidize acetal to 2-MF. By shifting the furfural
by organic species, being the sample calcined at 650 C, but can to acetal equilibrium, AuZ400 catalysts results also more active,
be assigned tentatively to overtones and combination bands of the besides to be largely more selective to 2-MF than AuZ200 and
fundamental vibration modes of ZrO2 [3537], even if not stud- AUROliteTM reference catalyst.
ied in detail. When gold nanoparticles are added to ZrO2 (samples To note that various authors [12,14] reported that the ester-
AuZ200 and AuZ400), the intensity of these bands increases, with ication reaction passes through the formation of a hemiacetal
intensication also of components at 1425, 1460 and 1575 cm1 intermediate, and then, the reaction can follow two different
(Fig. 9). This is likely due to the plasmonic enhancement of vibra- routes: a direct conversion to 2-MF, or the rst transformation
tional excitations in the infrared [38]. This effect, however, prevents of hemiacetal in the corresponding 2-furaldehyde-dimethyl-acetal
to quantify the acidity by pyridine adsorption, due to the overlap (acetal) with the further conversion to 2-MF. We do not observe the
of these bands. The effect is indicative of the coupling between Au presence of hemiacetal, but only of the acetal, thus in agreement
NPs and the support. The similar effect in both AuZ200 and AuZ400 with the second pathway, but with the indication that the rst step
samples indicates that this coupling is not signicantly different in to acetal is reversible.
the two samples. There are discording indications on the nature of the active
While pyridine adsorption cannot be monitored in the sites responsible for these steps in the reaction mechanism. Pinna
14001600 cm1 region for AuZ200 and AuZ400 samples, as com- and co-workers [1318] reported that gold is necessary for the
mented above, new weak bands can be observed between 2100 formation of 2-MF, but indicating that gold nanoparticles below
and 1800 cm1 (Fig. 10), but only for AuZ200. This is not a spec- 1.52.0 nm are necessary for both activity and selectivity [13,14].
tral region typically analysed for pyridine adsorption, but pyridine Later the same authors showed that i) larger particles (58 nm)
shows a triplet of weak bands in this region due to a combination of are instead more active on Au/ceria catalysts, but giving similar

Please cite this article in press as: C. Ampelli, et al., On the nature of the active sites in the selective oxidative esterication of furfural
on Au/ZrO2 catalysts, Catal. Today (2016), http://dx.doi.org/10.1016/j.cattod.2016.04.023
G Model
CATTOD-10171; No. of Pages 10 ARTICLE IN PRESS
C. Ampelli et al. / Catalysis Today xxx (2016) xxxxxx 9

selectivity [16], or ii) roundish nanoparticles with average size of different behaviour in furfural oxidative esterication, particularly
2.5 nm are preferable [17], while irregular plate-like nanoparticles in the selectivity to 2-MF. The analysis of their catalytic behaviour,
with diameter above 3 nm have similar activity, but rather lower in comparison with that shown by the ZrO2 support and a reference
selectivity (on Au/sulphated ZrO2 ). Au/TiO2 catalyst (AUROliteTM ), evidences the following aspects:
It is clear that the identication of the possible active sites is
rather difcult and it is not simple the identication when samples i) the ZrO2 support alone is active in furfural conversion, but to
having different characteristics are compared. For this reason, we form only the acetal intermediate (selectivity to 2-MF about
have selected two samples giving quite different activity and selec- 10%);
tivity, but having quite similar characteristics, being different only ii) the addition of gold enhances the activity in furfural conversion
for the temperature of calcination after addition of gold. They can be by a factor about ten, although without an apparent change in
thus considered a good model system to understand the nature of other properties.
the active sites, although limited to the comparison of two samples. iii) the creation of active sites for the further conversion of the acetal
The rst observation is that the presence of gold nanoparticles intermediate to 2-MF increases the reaction rate of furfural con-
enhances the rate of furfural conversion with respect to that shown version, mainly by shifting the equilibrium to right of the furfural
by the pure support. However, the intermediate acetal is converted to acetal reaction.
only when a specic type of gold species is present. It may be argued iv) the precursor sites for the acetal to 2-MF oxidative conversion
that cationic gold may be implied in the selective oxidation step forms during the thermal treatment in air from 200 C to 400 C;
(from acetal to 2-MF), but XPS data indicate that this species is characterization data evidences that a reduction of gold occurs
abundant in the less active catalyst (AuZ200) rather than in the during this treatment and partial redispersion of gold, with for-
most selective (AuZ400) where cationic gold is virtual absent. The mation of very small Au NPs with average size of 0.7 nm
calcination at 400 C (AuZ400) rather than a 200 C (AuZ200) after v) the active sites for the acetal to 2-MF oxidative conversion are
gold addition by precipitation-deposition induces a reduction of associated likely to these small Au NPs, but they form in-situ at
the Au, as shown by XPS, and the generation of rather small (below the initial stage of reaction.
1.0 nm, average diameter 0.7 nm) gold NPs. These species are much
less in AuZ200. As commented above, however, the active sites
It is possible that during this initial in-situ activation, the par-
for the oxidation form during the catalytic reaction, as shown by
tial oxidation of gold sites at the interface with the support occurs.
the presence of an initial induction time. It is thus possible that
However, it should be remarked that the characteristics of these
perimetral sites at the edge between very small Au NPs and zir-
eventually oxidized sites are different from the cationic gold sites
conia may become partially oxidized during this induction time.
present in large amount in the catalyst treated at 200 C (AuZ200),
However, these sites are different from the Au+ sites present in
which lack, however, of activity in the acetal to 2-MF oxidative
the AuZ200 sample. Although not characterized here, also refer-
conversion.
ence AUROliteTM sample presents in minimal amount these Au NPs
The comparison of these two catalysts thus provide useful indi-
sites, in agreement with its relatively poor performances. Even if we
cations on the nature of the active catalysts for furfural oxidative
do not know the thermal history of AUROliteTM (as it is a trademark
esterication, although it must be commented that these indi-
product), we tried to calcine the AUROliteTM sample under the same
cations should be supported further by more detailed studies
conditions as AuZ400. Results (reported in the Supporting infor-
extending the analysis to a broader range of supported gold cata-
mation, see Fig. 2S), showed a comparable conversion prole with
lysts and to in-situ or operando studies. These results should be thus
respect to the not-thermally treated AUROliteTM , but an increase in
considered an indication (going beyond literature) on the possible
the yield to 2-MF from about 2040% was observed, conrming that
nature of the active sites in this interesting reaction of relevance
calcination at 400 C leads to the formation of more active species
for the future of bioreneries rather than a denitive proof of the
(reduced and very small Au NPs).
mechanism requiring further studies.
Infrared and visible diffuse reectance data further support the
The results, however, provide some clue to understand the two
presence of these very small gold particles in AuZ400, and the pres-
questions raised in the introduction, regarding the role of the sup-
ence of cationic gold species in AuZ200, while these and other
port and of the reaction conditions. The results presented here
experimental results (XPS, TEM, XRD) do not evidence a difference
evidence that the key steps in the formation of the active sites for
in the characteristics of the zirconia support in the two samples.
oxidative conversion of the acetal intermediate are associated to i)
Although the support may have a role in the rst step of furfural
the mechanism of self-reduction of Au NPs during thermal treat-
conversion, the presence of gold induces an enhancement of the
ment in air and the formation of very small (below 1 nm) Au NPs
activity by one order of magnitude. It is likely that the true rate con-
during this process and ii) the in-situ transformation of these very
stant of furfural conversion is similar between AuZ200 and AuZ400,
small Au NPs during the initial stage of the catalytic reaction. It is
being the apparent higher rate of AuZ400 due to the shifting of
very likely that the rst process is largely depending on the specic
the equilibrium in furfural to acetal reaction associated to the con-
characteristics of the support, while the second to the feed compo-
version of the latter to 2-MF. The main role of the support is thus
sition and reaction conditions. As comment above, further studies
in inducing the formation of very small gold nanoparticles (below
are necessary to understand better these aspects, but present data
about 1 nm) during the thermal treatment after gold deposition.
provide an indication on the aspects to investigate and a tentative
Likely the characteristics of the support, in addition to the specic
hypothesis to explain the rather large differences in the behaviour
procedure of catalyst activation after gold deposition, determine
of this class of supported Au catalysts.
the dispersion of gold and the possibility to have active and selec-
tive catalysts.
Acknowledgements

5. Conclusions This work was realized in the frame of the EU project

Eco-friendly biorenery ne chemicals from CO2 photocat-
Two catalysts, based on gold on zirconia prepared by precipita- alytic reduction (Eco2 CO2 ) (Project 309701), which is gratefully
tion and prepared in the same way, except for the different thermal acknowledged. The EU Marie Curie IAPP (Industry-Academia Part-
treatment in air after the addition of gold (0.4% wt), show a quite nerships and Pathways) project Biopolymers and Biofuels from

Please cite this article in press as: C. Ampelli, et al., On the nature of the active sites in the selective oxidative esterication of furfural
on Au/ZrO2 catalysts, Catal. Today (2016), http://dx.doi.org/10.1016/j.cattod.2016.04.023
G Model
CATTOD-10171; No. of Pages 10 ARTICLE IN PRESS
10 C. Ampelli et al. / Catalysis Today xxx (2016) xxxxxx

Furan based Building Blocks (BIOFUR, project 324292) of collab- [17] F. Menegazzo, M. Signoretto, D. Marchese, F. Pinna, M. Manzoli, J. Catal. 326
oration between Avantium and the University of Messina is also (2015) 1.
[18] M. Signoretto, F. Menegazzo, L. Contessotto, F. Pinna, M. Manzoli, B. Boccuzzi,
gratefully acknowledged. Appl. Catal. B Environ. 129 (2013) 287.
[19] Y. Li, L. Wang, R. Yan, J. Han, S. Zhang, Catal. Sci. Technol. 5 (2015) 3682.
Appendix A. Supplementary data [20] P. Mondal, N. Salam, A. Mondal, K. Ghosh, K. Tuhina, M. Sk. Islam, J. Colloid
Interface Sci. 459 (2015) 97.
[21] E. Smolentseva, V.V. Costa, R.F. Cotta, O. Simakova, S. Beloshapkin, E.V.
Supplementary data associated with this article can be found, Gusevskaya, A. Simakov, ChemCatChem 7 (2015) 1011.
in the online version, at http://dx.doi.org/10.1016/j.cattod.2016.04. [22] X. Wan, W. Deng, Q. Zhang, Y. Wang, Catal. Today 233 (2014) 147.
[23] K. Suzuki, T. Yamaguchi, K. Matsushita, C. Iitsuka, J. Miura, T. Akaogi, H. Ishida,
023. ACS Catal. 3 (2013) 1845.
[24] L. Prati, A. Villa, Gold Catalysis: Preparation, Characterization, and
References Applications, CRC Press, Boca Raton, FL, 2015.
[25] A.S.K. Hashmi, C. Lothschuetz, M. Ackermann, R. Doepp, S. Anantharaman, B.
Marchetti, H. Bertagnolli, F. Rominger, Chem. A Eur. J. 16 (2010) 8012.
[1] H. Kobayashi, A. Fukuoka, Green Chem. 15 (2013) 1740.
[26] C. Ampelli, G. Centi, C. Genovese, G. Papanikolaou, S. Perathoner, K.J.
[2] S.K. Maity, Renew. Sustain. Energy Rev. 43 (2015) 1427.
Schouten, J.K. van der Waal, Top. Catal. (2015), in press.
[3] S. Abate, P. Lanzafame, S. Perathoner, G. Centi, ChemSusChem 8 (2015) 2854.
[27] J. Matta, J.-F. Lamonier, E. Abi-Aad, E.A. Zhilinskaya, A. Aboukas, Phys. Chem.
[4] P. Lanzafame, G. Centi, S. Perathoner, Catal. Today 234 (2014) 2.
Chem. Phys. 1 (1999) 4975.
[5] R. Luque, Curr. Green Chem. 2 (2015) 90.
[28] Y.Q. Wang, W.S. Liang, C.Y. Geng, Nanoscale Res. Lett. 4 (2009) 684.
[6] K. Sriroth, K. Piyachomkwan, Bioprocessing Technologies in Biorenery for
[29] C. Wang, D. Astruc, Chem. Soc. Rev. 43 (2014) 7188.
Sustainable Production of Fuels, Chemicals, and Polymers, in: S.-T. Yang, H.A.
[30] X. Zhang, Y.L. Chen, R.-S. Liu, D.P. Tsai, Rep. Prog. Phys. 76 (2013) 046401.
El-Enshasy, N. Thongchul (Eds.), Wiley & Sons Pub., NY, 2013, p. 27.
[31] W. Huang, J. Shi, H. Arizpe-Chavez, E. Zayas, G. Ma, F.J. Espinoza-Bentrn,
[7] R.-J. van Putten, J.C. van der Waal, E. de Jong, C.B. Rasrendra, H.J. Heeres, J.G.
Supercies y Vaco 17 (2004) 13.
de Vries, Chem. Rev. 113 (2013) 149.
[32] Y. Zhang, Y. Zhan, C. Chen, Y. Cao, X. Lin, Q. Zheng, Int. J. Hydrogen Energy 37
[8]
I. Delidovich, P.J.C. Hausoul, L. Deng, R. Pfutzenreuter, M. Rose, R. Palkovits,
(2012) 12292.
Chem. Rev. 116 (3) (2016) 1540, http://dx.doi.org/10.1021/acs.chemrev.
[33] M.P. Casaletto, A. Longo, A. Martorana, A. Prestianni, A.M. Venezia, Surf.
5b00354.
Interface Anal. 38 (2006) 215.
[9] S. Tang, J. Yuan, C. Liu, A. Lei, Dalton Trans. 43 (2014) 13460.
[34] Q. Fu, H. Saltsburg, M. Stephanopoulus, Science 301 (2003) 935.
[10] F. Luo, C. Pan, J. Cheng, Synlett 23 (2012) 357.
[35] G.G. Siu, M.J. Stokes, Y. Liu, Phys. Rev. B 59 (1999) 3173.
[11] E. Taarning, I.S. Nielsen, K. Egeblad, R. Madsen, C.H. Christensen,
[36] A.P. Naumenko, N.I. Berezovska, M.M. Biliy, O.V. Shevchenko, Phys. Chem.
ChemSusChem 1 (2008) 75.
Solid State 9 (2008) 121.
[12] O. Casanova, S. Iborra, A. Corma, J. Catal. 265 (2009) 109.
[37] X. Zhao, D. Vanderbilt, Phys. Rev. B 65 (2002) 075105-1.
[13] F. Pinna, A. Olivo, V. Trevisan, F. Menegazzo, M. Signoretto, M. Manzoli, B.
[38] F. Neubrech, A. Pucci, IEEE J. 19 (2013) 4600809.
Boccuzzi, Catal. Today 203 (2013) 196.
[39] J.K. Wlshurst, H.J. Berbstein, Can. J. Chem. 35 (1957) 1182.
[14] F. Menegazzo, T. Fantinel, M. Signoretto, F. Pinna, M. Manzoli, J. Catal. 319
[40] N. Perret, X. Wang, T. Onfroy, C. Calers, M.A. Keane, J. Catal. 309 (2014) 333.
(2014) 61.
[41] D.-Y. Wu, B. Ren, Y.-X. Jiang, X. Xu, Z.-Q. Tian, J. Phys. Chem. A 106 (2002) 9042.
[15] F. Menegazzo, M. Signoretto, F. Pinna, M. Manzoli, V. Aina, G. Cerrato, F.
Boccuzzi, J. Catal. 309 (2014) 241.
[16] M. Manzoli, F. Menegazzo, M. Signoretto, G. Cruciani, F. Pinna, J. Catal. 330
(2015) 465.

Please cite this article in press as: C. Ampelli, et al., On the nature of the active sites in the selective oxidative esterication of furfural
on Au/ZrO2 catalysts, Catal. Today (2016), http://dx.doi.org/10.1016/j.cattod.2016.04.023