Selective oxidation of 5-hydroxymethyl-2-furfural over TiO 2-supported gold-copper catalysts prepared from preformed nanoparticles: Effect of Au/Cu ratio

Article abstract of DOI:10.1016/j.cattod.2012.05.039

5-Hydroxymethyl-2-furfural (HMF) oxidation to furandicarboxylic acid (FDCA) was performed under mild reaction conditions using TiO₂-supported Au and Au-Cu catalysts synthesized from pre-formed nanoparticles of different composition. Catalysts were characterized by BET, XRD and XPS. The Au ₃Cu₁/TiO₂ catalyst exhibited the best catalytic performance for FDCA yield. Moreover, after reaction, bimetallic Au-Cu catalysts with high gold content can be recovered by filtration and reused without significant loss of activity and selectivity; whereas, the monometallic gold materials are not stable.

Full text of DOI:10.1016/j.cattod.2012.05.039

Catalysis Today 195 (2012) 120–126
Contents lists available at SciVerse ScienceDirect
Catalysis Today
journal homepage: www.elsevier.com/locate/cattod
Selective oxidation of 5-hydroxymethyl-2-furfural over TiO -supported
2
gold–copper catalysts prepared from preformed nanoparticles: Effect of Au/Cu
ratio
a,∗
a
a
c
b
b
Stefania Albonetti , Thomas Pasini , Alice Lolli , Magda Blosi , Marco Piccinini , Nikolaos Dimitratos ,
b,1
b
b
b
a
Jose A. Lopez-Sanchez , David J. Morgan , Albert F. Carley , Graham J. Hutchings , Fabrizio Cavani
Dip. Chimica Industriale e dei Materiali, v.le Risorgimento 4, 40136 Bologna (BO), Italy
Cardiff Catalysis Institute, Cardiff University, Main Building, Park Place, Cardiff, CF10 3AT, UK
ISTEC-CNR, Via Granarolo 64, 48018 Faenza, Italy
a
b
c
a r t i c l e i n f o
a b s t r a c t
Article history:
5-Hydroxymethyl-2-furfural (HMF) oxidation to furandicarboxylic acid (FDCA) was performed under
mild reaction conditions using TiO2-supported Au and Au–Cu catalysts synthesized from pre-formed
nanoparticles of different composition. Catalysts were characterized by BET, XRD and XPS. The
Au3Cu1/TiO2 catalyst exhibited the best catalytic performance for FDCA yield. Moreover, after reaction,
bimetallic Au–Cu catalysts with high gold content can be recovered by filtration and reused without
significant loss of activity and selectivity; whereas, the monometallic gold materials are not stable.
Received 12 January 2012
Received in revised form 29 May 2012
Accepted 31 May 2012
Available online 26 June 2012
Keywords:
Selective oxidation of HMF
©
2012 Elsevier B.V. All rights reserved.
2
,5-Furandicarboxylic acid
Gold–copper catalysts
Preformed nanoparticles
1
. Introduction
[5–10]. Nevertheless, despite this activity, process productivity and,
in particular, catalyst stability remains very low.
One of the most important sources of biomass is represented
Recently, we demonstrated that bimetallic gold–copper
nanoparticles supported on titania were excellent catalysts for the
oxidation of 5-hydroxymethyl-2-furfural to furandicarboxylic acid
[11]. A strong synergistic effect was evident with the addition of
Cu to Au, especially in terms of sample stability and resistance to
poisoning. Moreover, obtained results emphasize the importance
of the bimetallic catalyst synthesis procedure in order to achieve
the desired Cu promotional effect on Au.
by sugars, which are widely available and easily transformed.
Homogeneous dehydration of glucose and/or fructose leads to the
formation of 5-hydroxymethyl-2-furfural (HMF) which is a key
precursor for the synthesis of chemicals that have applications
within the pharmaceutical and polymer industries [1,2]. More-
over, HMF can also be oxidized to 2,5-furandicarboxylic (FDCA)
acid, which has been recently proposed as a possible surrogate for
terephthalate acid [3,4]; the monomer used for the production of
terephthalate plastic (PET).
Herein, we report a detailed investigation of the selective
oxidation of 5-hydroxymethyl-2-furfural over TiO -supported
2
The synthesis of FDCA from HMF has been widely studied in the
last two decades using different catalysts and reaction conditions.
In particular, supported Au nanoparticles have been found to be
very active catalysts for FDCA synthesis and many researchers have
focused their attention in searching for the best supports, reaction
conditions and mechanistic studies to improve the yield of the FDCA
gold–copper catalysts prepared from preformed nanoparticles. In
particular, we investigate the effect of the copper–gold ratio on the
structure and activity of the prepared catalysts.
2. Experimental
2.1. Catalyst preparation
Au/TiO , Cu/TiO2 and Au–Cu/TiO2 catalysts were prepared by
2
immobilization on the TiO2 surface of the preformed monometal-
lic and bimetallic colloids. Mono and bimetallic nanoparticles were
prepared using the method previously developed [12–14]. In brief,
to a solution of ␤-d-glucose in water was added the necessary
∗ Corresponding author.
E-mail address: stefania.albonetti@unibo.it (S. Albonetti).
1
Present address: Stephenson Institute for Renewable Energy, Department of
Chemistry, University of Liverpool, Liverpool, L69 7ZD, UK.
0
920-5861/$ – see front matter © 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.cattod.2012.05.039

S. Albonetti et al. / Catalysis Today 195 (2012) 120–126
121
Table 1
2
.3. Analytical methods
Structural parameters and chemical composition of the Au/TiO2 and Au–Cu/TiO2
prepared catalysts.
XRD measurements were carried out at room temperature with
Catalyst
Total metal loading
%, wt/wt)
Surface area
(m /g)
Au crystallite size
(nm)a
a Bragg/Brentano diffractometer (X’pertPro PANalytical) equipped
with a fast X’Celerator detector, using a Cu anode as the X-ray
source (K˛, ꢀ = 1.5418 A˚ ). For all sols and catalysts, the complete
diffractogram was collected over the 10–80 2ꢁ range, counting for
2
lite size, a second acquisition was performed in the 2ꢁ range 34–47
2
2
(
TiO2
0
83
78
74
73
–
0.5Au–Ti
1.0Au–Ti
2.0Au–Ti
0.5
1.0
2.0
n.d.
6.0
6.5
◦
◦
0 s at each 0.05 step. However, for evaluation of the metal crystal-
◦
0.5Au1Cu1–Ti
1.0Au1Cu1–Ti
2.0Au1Cu1–Ti
0.5
1.0
2.0
74
73
58
n.d.
4.0
5.0
ꢁ range, counting for 1500 s at each 0.08 step. In fact, the coher-
ence length of the Au crystalline domains was evaluated through
◦
◦
single line profile fitting of the reflection at 2ꢁ 38.2 or 44.3 , since
at this last angle no overlap with the anatase pattern of the sup-
port was observed. Crystallite size values were calculated using the
Scherrer equation from the full width at half maximum intensity
measurements.
1
1
1
1
.5Au–Ti
1.5
1.5
1.5
1.5
74
72
64
60
6.5
5.0
5.0
5.0
.5Au3Cu1–Ti
.5Au1Cu1–Ti
.5Au1Cu3–Ti
n.d. = not detectable.
a
Particle size distribution, based on hydrodynamic diameter, was
Estimated from XRD.
◦
evaluated at 25 C by dynamic light scattering by using a Zetasizer
Nano ZS (Malvern Instrument, UK) equipped with a He–Ne laser
(
wavelength 633 nm). Solution were filtered through nylon filters
quantity of PVP used as nanoparticle stabilizer. The solution was
then heated to 95 C. At this temperature NaOH and an aqueous
◦
with a pore size of 0.45 ␮m (Minisart NY). Hydrodynamic diameter
includes both the coordination sphere and the species adsorbed
on particle surface, such as stabilizers, surfactants and so forth.
DLS analysis provides also a polydispersion index parameter (PDI),
ranging from 0 to 1, quantifying of the colloidal dispersion degree.
solution containing the metal precursors (HAuCl and CuSO ·5H O)
4
4
2
in the desired ratio were added and stirred for 2.5 min. The ratio
among PVP, ␤-d-glucose, NaOH and metals was optimized for each
gold and copper content [11,15].
Catalyst surface areas were measured by N physisorption appa-
Before use, the as-prepared sols were concentrated and washed
with distilled water using 50 kDa Amicon Ultra filters (Millipore)
to eliminate the excess PVP and other reagents dissolved in the
aqueous media. The Au and Au–Cu colloids were then impregnated
2
ratus (Sorpty 1750 CE instruments) and single point BET analysis
methods, in which samples were pre-treated under vacuum at
1
◦
00 C.
X-ray photoelectron spectroscopy (XPS) measurements were
into TiO . Two series of bimetallic samples were prepared. In one
2
performed on a Kratos Axis Ultra DLD spectrometer. Data were
acquired using a monochromatic Al source, operating at 120 W.
All spectra were acquired using the Kratos immersion lens sys-
tem for charge compensation, and the hybrid spectroscopy mode
at pass energies of 40 and 160 eV for high resolution and survey
scans respectively. Data were calibrated to the C 1s line at 284.8 eV,
of them the nominal ratio Au:Cu was maintained at 1:1 and the
total metal loading was varied from 0.5 to 2.0 wt%. In the other
series, the total metal loading was maintained at 1.5 wt% while
the ratio Au:Cu was varied from 3:1 to 1:6 on a molar basis. For
all samples the solvent was evaporated by thermal treatment at
◦
1
20 C. The catalyst samples are denoted as zAu–Ti, zCu–Ti and
TM
attributable to adventitious carbon and quantified using CasaXPS
zAuxCuy–Ti, where z refers to the total metal content in the material
and x and y refer to the Au:Cu molar ratio (i.e., 1.5Au Cu –Ti indi-
v2.3.15, using sensitivity factors supplied by the manufacturer.
To verify the behavior of different samples under thermogravi-
metric analysis TGA was obtained using a Rheometric Scientific
1
1
cates a TiO supported sample synthesized with 1.5 wt% total metal
2
with a Au:Cu molar ratio of 1). The catalysts were characterized
in detail by XRD, TGA, BET and XPS analysis. The correspond-
ing precursor nanoparticles were also studied. Table 1 reports
the composition and characteristic of catalysts investigated in this
work.
◦
STA1500 analyzer while heating the sample in air from 25 C to
◦
6
00 C.
3. Results and discussion
2.2. Oxidation reactions
3.1. Colloidal characterization of preformed Au and Au/Cu
nanoparticles
The oxidation of 5-hydroxymethyl-2-furfural (HMF) was car-
ried out using an autoclave (Parr Instruments) reactor of 300 mL
capacity and equipped with a mechanical stirrer (0–1200 rpm) and
provision for measurement of temperature and pressure. The reac-
tor was charged with an aqueous solution (25 mL distilled water)
containing the appropriate amount of 5-hydroxymethyl-2-furfural,
base (NaOH) and catalyst (HMF/metal molar ratio = 100). The auto-
clave was purged 3 times with O2 (5 bar) and then pressurized at
The preparation of gold and gold–copper materials by the
reduction of HAuCl4 and CuSO ·5H O with ␤-d-glucose yielded
4
2
nanoscale particles suspended in water. The resulting colloidal
suspensions were stable for several weeks. Laser diffraction inves-
tigation (DLS) was applied to examine the particle diameter and
the particle size distribution in suspension. These data can strongly
differ from that obtained by XRD or TEM due to the different analyt-
ical techniques used, which provide information on the size of the
diffracting nanocrystals (XRD) and on the hydrodynamic diameter
of the scattering object, including the eventual amorphous surface
layer and the solvation sphere (DLS) [16].
1
0 bar. If not differently indicated, the temperature was increased
◦
to 60 C and the reaction mixture was stirred at ca. 1000 rpm for
4
h. At the end of the reaction, the reactor was cooled to room tem-
perature and the solution was filtered. Then, 4 mL of water was
added to an aliquot of the reaction solution (1 mL) before analy-
sis with an Agilent Infinity 1200 liquid chromatograph equipped
with a Aminex HPX 87-H 300 mm × 7.8 mm column using a
As an example of our results, Fig. 1 reports DLS analysis for
Au/Cu samples at different Au:Cu atomic ratio. For gold and
gold/copper suspensions, up to Au:Cu atomic ratio 1:1, a relatively
monodispersed signal, with an averaged diameter of 17–20 nm,
was detected. Increasing the copper content leads to polydispersed
particles with larger dimensions. Table 2 summarizes DLS data on
0
.005 M H SO solution as the mobile phase. Identification of com-
2 4
pounds was achieved by calibration using reference commercial
samples.

1
22
S. Albonetti et al. / Catalysis Today 195 (2012) 120–126
1
4
2
0
Cu0
3
8.2
1
1
a)
8
6
4
2
0
b)
c)
d)
1
10
100
1000
e)
Mean particles size (nm)
Fig. 1. Particle size distribution (intensity %) of samples at different Au/Cu atomic
ratio in water. Legend: (—) Au sol; (
) Au1Cu3 sol; (
) Au3Cu1 sol; (
) Au1Cu6 sol.
) Au1Cu1 sol;
(
34
40
44
Position[2theta]
synthesized samples and confirms that the average particle diam-
eter of suspensions mainly depends on Cu content. Moreover, as
indicated by the polydispersity index (PDI), the size distribution
is closely related to the average particle diameter: the larger the
particle diameter, the broader the particle size distribution. Actu-
ally these results seems to indicate that aqueous sols, up to Au:Cu
atomic ratio 1:1, are predominantly constituted by homogeneous
bimetallic nanoparticles with an unimodal particle size distribution
while increasing the copper content in the sols causes the forma-
tion of aggregates or nanoclusters with inhomogeneous size and/or
composition.
Fig. 2. XRD data for all Au and Au/Cu preformed nanoparticles. The vertical lines
correspond to the (1 1 1) peak position for Au. Legend: (a) Au sol; (b) Au3Cu1 sol; (c)
Au1Cu1 sol; (d) Au1Cu3 sol; (e) Au1Cu6 sol.
3
.2. Catalyst characterization
XRD, BET, TGA and XPS have been used to determine the struc-
ture and morphology of the gold and gold/copper nanoparticles
deposited on TiO . The structural properties of the Au/TiO2 and
2
Au–Cu/TiO2 catalysts are shown in Table 1. In the case of Au/TiO2,
the surface area values are very similar to the TiO2 support, indi-
cating that deposition of the gold sol did not produce a significant
change in the total porosity. On the other hand, immobilization of
the bimetallic colloid induces a more significant decrease in catalyst
surface area. The effect of active phase content on this parameter is
evident and the BET surface area data are linearly correlated with
the total metals loading and on copper content in the active phase,
showing a significant decrease for samples with higher metals con-
tent.
XRD patterns for Au/Cu dried nanoparticles with different Au:Cu
atomic ratio (Fig. 2) show the characteristic peaks of gold as well
as the peak broadening typical of nano-sized crystallites.
These patterns, enlarged to highlight the 1 1 1 reflections of gold
◦
at 2ꢁ = 38.2 , are consistent with single-phase fcc structure, with
no observable impurities of CuOx of phase-segregated metals up to
Au Cu sample. For the sample with the higher Cu content (Au Cu
material) the presence of 2ꢁ = 43.1 reflection suggests the forma-
tion of segregated metallic copper.
1
3
1
6
◦
Compared to the monometallic gold sample, bimetallic
nanoparticles show a shift of the reflections toward higher 2ꢁ val-
ues with increasing the Cu content. This shift is consistent with
the alloying of gold and copper. Moreover, the bimetallic materials
give broader XRD peaks, suggesting that the Au/Cu system has a
smaller crystallite size than the monometallic one. In effect, crystal
size values (Table 2) estimated using the Scherrer equation indi-
cated ∼4 nm for gold and ∼3–3.5 nm for Au/Cu sols, confirming the
coarser size of the monometallic system. These results were also
confirmed by TEM analysis on supported samples [11] that indi-
cated a particle size decreases when introducing Cu into the Au
nanoparticles.
Fig. 3 compares the weight loss analysis for prepared samples.
The minor weight loss at temperatures lower that 100 C corre-
sponds to the release of adsorbed water while the more significant
◦
1
00
98
95
1
2
3
9
9
3
0
Table 2
Particles size distribution estimated by DLS (d-DLS) and XRD (d-XRD) analysis on
prepared sols.
88
4
Sol
d-DLS (nm)
PDI
d-XRD (nm)
85
Au
21
17
18
20
120
0.20
0.22
0.27
0.37
0.45
4.0
4.0
3.5
3.0
n.d.
0
100
200
300
400
500
600
Au3Cu1
Au1Cu1
Au1Cu3
Au1Cu6
Temperature ( C)
Fig. 3. Thermogravimetric analysis in air of TiO2 supported catalysts with dif-
ferent Cu:Au atomic ratio. (1) Au–TiO2; (2) 1.5Au3Cu1–Ti; (3) 1.5Au1Cu1–Ti; (4)
1.5Au Cu –Ti.
n.d. = not detectable.
1
3

S. Albonetti et al. / Catalysis Today 195 (2012) 120–126
123
a)
b)
c)
d)
2
0
30
40
50
60
70
Position [2θ]
Fig. 4. XRD patterns of the TiO2 supported catalysts with different gold:copper
atomic ratio. (a) 1.5Au–Ti; (b) 1.5Au3Cu1–Ti; (c) 1.5Au1Cu1–Ti; (d) 1.5Au1Cu3–Ti.
◦
In the insert: magnification of the reflection at 2ꢁ 44.3 ; utilized to evaluate the Au
crystal size.
◦
◦
weight loss observed between 200 C and 400 C can be assigned to
the decomposition of the organic capping agent (PVP) and reduc-
ing agent (glucose) chemisorbed onto the catalysts surface [17–19].
Weight loss significantly increases by increasing the Cu content in
the active phase and can be related to the higher amounts of sta-
bilizer and glucose necessary to synthesize sols at higher Cu:Au
atomic ratio.
Fig. 5. Au 4f XPS spectra of the monometallic Au/TiO2 catalysts (normalized to the
same intensity). (a) 0.5Au–Ti, (b) 1.0Au/Ti, (c) 1.5Au/Ti, (d) 2.0Au–Ti.
A typical series of XRD patterns for the prepared systems is
shown in Fig. 4 that represents the XRD pattern for sample at
process, leads to a slightly higher mean particle size in comparison
to preformed nanoparticles.
1
.5 wt% total metal content and increasing Au/Cu ratio. Comparison
Results of the XPS characterization of the titania-supported
nanoparticles are presented in Table 3 and Figs. 5–7. Analysis of the
Au 4f region for both monometallic and bimetallic samples encom-
passed the range 83.2–83.8 eV, this range is consistent with the
Au(0) oxidation state [20,21]. The majority of the reported energies
(Table 3) are lower than that for bulk gold (ca. 84 eV), and while
a shift lower (ca. −0.3 eV) is indicative of the lower coordination
of gold atoms, the lower bound of this range is indicative of the
rounded nature of the metallic Au nanoparticles [22], and consis-
tent with the microscopy data presented in our previous work [11].
The lower binding energy observed for the Au 4f7/2 for the Au–Cu
supported materials could be due to the following reasons: (i) the
with the pattern of pure titania reveals that most of the reflections
stem from the support (anatase TiO ). For all catalysts, a high metal
dispersion was observed since, in addition to the titania reflec-
tions, only weak and broad peaks corresponding to the most intense
metallic gold reflection were detected at 44.3 . This reflection can
be assigned to the (2 0 0) plane of fcc gold. The average size of gold
nanoparticles, as calculated from this reflection using the Scher-
rer equation, is given in Table 1. As a general observation, these
results indicated that the impregnation of the Au and Au/Cu sols
on TiO2 support generates a high metal dispersion despite of the
fact that partial aggregation, difficult to avoid during the deposition
2
◦
Fig. 6. Cu 2p3/2 XPS spectra of the bimetallic Au–Cu/TiO2 catalysts (normalized to the same intensity). (a) 0.5Au1Cu1–Ti, (b) 1.0Au1Cu1–Ti, (c) 1.5Au1Cu1–Ti, (d) 2.0Au1Cu1–Ti.

1
24
S. Albonetti et al. / Catalysis Today 195 (2012) 120–126
Table 3
XPS binding energies of Au/TiO2 and Au-Cu/TiO2 catalysts.
Samples
BE Au 4f7/2
Oxidation state
BE Cu 2p3/2
Oxidation state
%Au
%Cu
Au/Cu
0
1
1
2
0
1
1
2
.5%Au/TiO2
%Au/TiO2
.5%Au/TiO2
%Au/TiO2
.5%(AuCu)/TiO2
%(AuCu)/TiO2
.5%(AuCu)/TiO2
%(AuCu)/TiO2
83.6
83.7
83.8
83.7
83.2
83.3
83.6.0
3.7
Au(0)
Au(0)
Au(0)
Au(0)
Au(0)
Au(0)
Au(0)
Au(0)
0
0
0
0
1.25
2.44
4.08
5.18
0.78
1.44
1.63
2.07
0
0
0
0
0.57
0.84
1.63
1.16
+
931.9 (82), 932.6 (18)
931.6 (67.2), 932.9 (32.8)
931.7 (66.7), 932.4 (33.3)
931.9(67.0),933.2 (33.0)
Cu(0) Cu(ı )
1.36
1.71
1.0
+
Cu(0) Cu(ı )
+
Cu(0) Cu(ı )
+
Cu(0) Cu(ı )
1.78
smaller particle size of the Au–Cu particles as it has been observed
by Makee and co-workers [20] and (ii) electron transfer from copper
to gold atoms. The later can be explained as gold is more elec-
tronegative than copper according to Pauling’s electronegativity
table therefore we could expect some electron transfer shifting of
gold core levels toward lower binding energies [23,24]. However,
the Cu 2p peak centered at ca. 932 eV is lower than the expected
bulk value (932.7 eV) and that observed for smaller clusters of Cu
[
25] and this contradicts the explanation of an electron transfer
from copper to gold. In view of this, we suggest that the domi-
nant factor for the lower core level observed for Au 4f7/2 is mainly
due to the smaller particle size of the Au–Cu particle. While the
FWHM of the Cu 2p3/2 peak is almost twice that of, for example the
corresponding O 1s, Ti 2p and Au 4f (FHWM ∼ 1.1 eV), this could
possibly in part be attributed to broadening of the Cu signal within
the Au–Cu intermetallic [26], coupled with possible oxide forma-
tion which would present as a shoulder to the higher binding energy
side, suggesting that the major component is metallic Cu within a
Cu–Au alloy, whereby the shift is a result of charge transfer between
the Cu and Au atoms within the alloy [27]. This is consistent with
HRTEM data wherein a Cu/Au alloy in bimetallic samples was con-
firmed [11]. The weak Cu 2p signal precludes identification of any
Cu(II) species from shake-up structure and similarly a direct iden-
tification of the oxidation state using the Auger parameter while in
principle is possible, in practice for low-loading supported catalysts
it is difficult due to the very weak Auger LMM signal and possible
X-ray induced reduction of the ionic species.
Fig. 7. Au 4f XPS spectra of the bimetallic Au–Cu/TiO2 catalysts (normalized to
the same intensity). (a) 0.5Au1Cu1–Ti, (b) 1.0Au1Cu1–Ti, (c) 1.5Au1Cu1–Ti, (d)
2.0Au1Cu1-Ti.
Fig. 8 shows the comparison in catalytic activity of monometallic
and bimetallic samples, containing 0.5–1.5 wt% of total metal con-
tent, at different active phase loadings. The reaction was performed
◦
3.3. Catalytic tests
under optimized conditions: 60 C, 240 min, 10 bar of oxygen pres-
sure, HMF:metal loading:NaOH molar ratio 1:0.01:4 [11].
The partial oxidation of 5-hydroxymethyl-2-furfural (HMF) over
It is worth noting that at the end of 4 h reaction time, total
conversion of HMF was obtained with all catalysts. Nevertheless,
from reported results, it is evident that bimetallic Au–Cu materials
display improved yield to FDCA with respect to their correspond-
ing monometallic catalysts. Moreover, the optimum Au–Cu weight
loading for the highest FDCA yield was 1.5% metal on the titania
support. Taking into account crystallite size data for the bimetallic
samples, an increase of metal loading did not affect significantly the
mean metal particle size. The mean particle size for the 0.5–2 wt%
Au–Cu/TiO2 catalyst is in the range 4–5 nm. Hence, considering
that the metal particle size is similar the observed difference in
activity, should be more related to either the Au/Cu surface com-
position or the oxidation state of Au and Cu. XPS data revealed
the Au–Cu nanoparticles supported on TiO2 resulted in the for-
mation of 5-hydroxymethyl-2-furancarboxylic acid (HMFCA) and
furandicarboxylic acid (FDCA) (Scheme 1). In some reactions, by-
products derived from HMF degradation were observed, this being
an indication of the low activity of catalysts for HMF oxidation. In
addition, the repeated absence of the 5-formyl-2-furancarboxylic
acid intermediate (FFCA) in our experiments seems to confirm that
the oxidation of the hydroxyl group in 5-hydroxymethyl-2-furfural
is the rate limiting step of the reaction while FFCA, if formed, is
quickly transformed to FDCA.
In our studies catalytic tests over 1.5Au–TiO2 and
1
.5Au Cu –TiO samples at different metal loadings were firstly
1 1 2
carried out. The next step was the catalytic study of bimetallic
samples with different Au–Cu contents and the same total metal
loading (1.5%), in order to evaluate the effect of nanoparticles
composition on the reaction.
that the most active bimetallic catalyst (1.5 wt%Au–Cu/TiO ) was
2
characterized by a Au/Cu = 1 surface composition, with gold in
metallic state and the majority of Cu species in the presence of
+
Cu(0) and Cu(ı ), whereas the surface composition of the 0.5 wt%,
HO
O
HO
O
O
O
O
O
Fast
Slow
Fast
O
O
O
O
OH
OH
HO
OH
HMF
HMFCA
FFCA
FDCA
Scheme 1.

S. Albonetti et al. / Catalysis Today 195 (2012) 120–126
125
5
4
3
2
1
0
0
0
0
0
0
100
100
80
60
40
20
0
80
60
40
20
0
50
60
70
80
90
100
Temperature (°C)
0.5
1.0
1.5
2.0
Fig. 10. FDCA (solid line) and HMFCA (dotted line) selectivities as a function of
reaction temperature over 1.5Au1Cu1–Ti (ꢂ , ꢀ) and 1.5Au3Cu1–Ti (ꢀ , ꢁ ) catalysts.
For all reaction the HMF conversion was total. Reaction conditions: 240 min, 10 bar
of oxygen pressure, HMF:metal loading:NaOH molar ratio 1:0.01:4.
Total metals loading (%)
Fig. 8. Catalytic activity for the synthesis of FDCA on supported Au (ꢀ ) and Au/Cu
◦
(
1
ꢁ ) catalysts with different metals loading. Reaction conditions: 60 C, 240 min,
0 bar of oxygen pressure, HMF:metal loading:NaOH molar ratio 1:0.01:4.
1
.5 wt% and 2.0 wt% of Au–Cu/TiO2 catalysts are enriched in gold
content samples, suggest the prevalent degradation of HMF over
these materials.
(
Au/Cu = 1.3–1.8). These results indicate that for obtaining the high-
est activity for HMF transformation there is an optimum surface
composition (Au/Cu = 1) and the presence of Au in metallic state
and Cu species in (0, ı ) oxidation state favor selectivity toward
As indicated from characterization, gold nanoparticles on tita-
nia tend to be slightly larger than Au/Cu supported nanoparticles,
and this may be one of the factors controlling the reactivity of
these materials. Nevertheless, 1.5Au Cu –TiO , 1.5Au Cu –TiO
+
FDCA.
3
1
2
1
1
2
Taking into account that the most active catalyst was based on
and 1.5Au Cu –TiO have very similar mean crystallite sizes,
1 3 2
the 1.5 wt% Au–Cu/TiO catalyst with 1/1 nominal atomic ratio, the
hence the effect of active phase composition must account for the
lower activity observed with the catalysts containing higher than
Cu:Au = 1 atomic ratio. As a consequence, the inferior selectivity to
oxidized products obtained with catalysts at high copper contents
emphasises the importance of the Au:Cu ratio to achieve copper
promotion on gold and suggests that Cu acts as a gold promoter
and/or dispersing agent [24] only if present in a Au:Cu ratio higher
than 1:1.
2
effect of Au:Cu ratios was systematically studied. Fig. 9 shows the
performance of catalysts containing different Au:Cu ratios, for a
metal loading of 1.5%. Also in this case, for all studied samples, after
the 4 h of reaction, the HMF conversion was always complete but,
among these samples, strong differences appear in terms of product
selectivity. In particular, the selectivity toward the desired prod-
uct, the FDCA, increased significantly for 1.5Au Cu and 1.5Au Cu
3
1
1
1
samples with respect to monometallic systems and afterwards
strongly decreased as the Cu content in the catalyst formulation
further increased. TOF data, calculated as moles of formed FDCA
divided by reaction time (4 h) and moles of Au, resulted to be
Catalytic experiments were also performed over the more selec-
tive catalysts (1.5Au Cu –TiO and 1.5Au Cu –TiO ) by varying
3
1
2
1
1
2
reaction temperature (Fig. 10). A strong effect of this parameter
was observed on products distribution. In effect, obtained results
confirmed that studied catalysts display a very high activity for the
oxidation of the aldehydic functionality of HMF molecule, forming
high quantity of HMFCA also at low temperature. Nevertheless, the
oxidation of the primary alcohol side-chain, which is necessary to
form the FDCA, presumably via the corresponding aldehyde (FFCA),
is more demanding and proceeds with a significant rate at temper-
− −1 −1
.5 h , 14.5 h and 14 h for 1.5Au, 1.5Au Cu1 and 1.5Au Cu1
3 1
1
5
respectively. The lack of FDCA, but also the limited formation of
HMFCA and the presence of high amount of by-products (mainly
consisting of oligomers) in the experiments with high copper
1
00
◦
ature higher than 80 C. On both catalysts, after 4 h of reaction at
◦
9
5 C, a yield higher than 90% of FDCA was achieved. By-products
were not formed in any of the studied conditions and HMF was
exclusively oxidized to HMFCA and subsequently to FDCA.
8
6
4
2
0
0
0
0
0
Stability of the catalyst under operative conditions was eval-
uated by re-using it in consecutive runs. The catalyst was
filtered off, washed with water and dried before the new run
using fresh HMF/NaOH solution. FDCA yields during 3 runs for
monometallic 1.5Au–Ti sample and bimetallic 1.5Au Cu –TiO and
3
1
2
1
.5Au Cu –TiO catalysts after 4 h are shown in Fig. 11. Significant
1 1 2
deactivation was observed on repeated tests over Au/TiO2 cata-
lyst. According to previous studies, deactivation mainly occurs due
to the active phase being blocked by competitive adsorption [6].
However, this deactivation was not so evident for bimetallic sam-
ples and the stability increases, by increasing the Cu content in the
Au
Au3Cu1 Au1Cu1 Au1Cu3 Au1Cu6
Cu
active phase. The 1.5Au Cu –TiO catalyst only presents a slight
1
1
2
Fig. 9. Products selectivities on catalysts at different Au:Cu atomic ratios. Results are
given at total conversion of HMF. Reaction conditions: 70 C, 240 min, 10 bar of oxy-
decrease of FDCA yield confirming the importance of Au site isola-
tion in the bimetallic systems due to the alloying with Cu present
in this material.
◦
gen pressure, HMF:metal loading:NaOH molar ratio 1:0.01:4. Legend: (ꢁ) HMFCA,
(ꢀ) FDCA, ( ) by-products.

1
26
S. Albonetti et al. / Catalysis Today 195 (2012) 120–126
50
40
30
20
10
0
Acknowledgements
This work has been partially financed with founds of MIUR
Rome) through the PRIN national project “Catalytic upgrading of
(
the C5 fraction in ligno-cellulosic biorefineries”. INSTM is acknowl-
edged for co-financing the PhD project of T.P.
References
[
[
1] J.N. Chheda, G.W. Huber, J.A. Dumesic, Angewandte Chemie International Edi-
tion 46 (2007) 7164.
2] A.A. Rosatella, S.P. Simeonov, R.F.M. Frade, C.A.M. Afonso, Green Chemistry 13
(
2011) 754.
[
[
3] C. Moreau, M.N. Belgacem, A. Gandini, Topics in Catalysis 27 (2004) 11.
4] A. Gandini, Green Chemistry 13 (2011) 1061.
[5] O. Casanova, S. Iborra, A. Corma, Journal of Catalysis 265 (2009) 109.
[6] O. Casanova, S. Iborra, A. Corma, ChemSusChem 2 (2009) 1138.
Au-Ti
Au3Cu1-Ti
Au1Cu1-Ti
[
[
[
7] Y.Y. Gorbanev, S.K. Klitgaard, J.M. Woodley, C.H. Christensen, A. Riisager, Chem-
SusChem 2 (2009) 672.
8] E. Taarning, I.S. Nielsen, K. Egeblad, R. Madsen, C.H. Christensen, ChemSusChem
Catalysts
Fig. 11. Reusability study for the oxidation of HMF using 1.5Au–Ti, 1.5Au3Cu1–Ti
and 1.5Au1Cu1–Ti as catalysts. FDCA yield is given at total conversion of HMF. Reac-
tion conditions: 240 min, 60 C, 10 bar of oxygen pressure, HMF:metal loading:NaOH
molar ratio 1:0.01:4. Legend: FDCA yield during (ꢀ) 1 use, (ꢁ) 2 use and ( ) 3
1
(2008) 75.
9] S.E. Davis, L.R. Houk, E.C. Tamargo, A.K. Datye, R.J. Davis, Catalysis Today 160
2011) 55.
◦
(
◦
◦
◦
[10] S.E. Davis, B.N. Zope, R.J. Davis, Green Chemistry 14 (2012) 143.
[11] T. Pasini, M. Piccinini, M. Blosi, R. Bonelli, S. Albonetti, N. Dimitratos, J.A. Lopez-
Sanchez, M. Sankar, Q. He, C.J. Kiely, G.J. Hutchings, F. Cavani, Green Chemistry
use.
1
3 (2011) 2091.
4
. Conclusions
We have characterized and studied the activity for HMF oxi-
[
[
12] P. Raveendran, J. Fu, L. Wallen, Green Chemistry 8 (2006) 34.
13] J. Liu, G. Qin, P. Raveendran, Y. Ikushima, Chemistry – An European Journal 12
(
2005) 2131.
14] M. Blosi, S. Albonetti, M. Dondi, G. Baldi, A. Barzanti, PCT/EP2010/052534,
010.
[
dation to FDCA of TiO -supported Au and Au–Cu catalysts. The
2
2
catalysts were synthesized from pre-formed nanoparticles via col-
loidal methods of different compositions. The presence of Cu, which
by itself was not active in the desired reaction, allowed a remark-
able increase of the catalytic performance with respect to the
Au-only catalyst. This could be attributed to the generation of
alloyed bimetallic nanoparticles in the range 2–6 nm. The cata-
lyst containing 1.5 wt% of total metal and an Au/Cu atomic ratio of
[15] S. Albonetti, M. Blosi, F. Gatti, A. Migliori, L. Ortolani, V. Morandi, G.M. Baldi,
Studies in Surface Science and Catalysis 175 (2010) 621.
[16] A.M. Keene, K.M. Tyner, Journal of Nanoparticle Research 13 (2011) 3465.
[17] A.E. Jablonski, A.J. Lang, S. Vyazovkin, Thermochimica Acta 474 (2008) 78.
[18] M.P. Zheng, Y.-P. Jin, G.-L. Jin, M.-Y. Gu, Journal of Materials Science Letters 19
(2000) 433.
[
[
[
[
[
[
[
[
19] W. Yan, G.J. Suppes, Journal of Chemical and Engineering Data 53 (9) (2008)
2033.
20] A. Zwijnenburg, A. Goosens, W.G. Sloof, M.W.J. Graje, A.M. Kraan, L.J. Jongth, M.
Makee, J.A. Moulijn, Journal of Physical Chemistry B 106 (2002) 9853.
21] N. Dimitratos, A. Villa, C.L. Bianchi, L. Prati, M. Makkee, Applied Catalysis A 311
3
/1 presented the best catalytic activity, which allowed an efficient
transformation of the intermediate HMFCA into FDCA, which is the
rate limiting step of the process. Additionally, the Cu/Au mixed sys-
tems improve reusability of the catalyst recovered by filtration; in
fact, the loss of activity of the reused Au–Cu catalysts was much
lower than that shown by the monometallic Au material. In addi-
tion, the stabilization effect due to Cu presence in the active phase
seems significantly connected with the Cu amount, the bimetallic
system with Au/Cu atomic ratio of 1 showing the highest level of
reusability.
(
2006) 185.
22] J. Radnik, C. Mohr, P. Claus, Physical Chemistry Chemical Physics 5 (2003)
172.
23] R.J. Chimentão, F. Medina, J.L.G. Fierro, J. Llorca, J.E. Sueiras, Y. Cesteros, P.
Salagre, Journal of Molecular Catalysis A: Chemical 274 (2007) 159.
24] J. Liorca, M. Domínguez, C. Ledesma, R.J. Chimentão, F. Medina, J. Sueiras, I.
Angurell, M. Seco, O. Rossell, Journal of Catalysis 258 (2008) 187.
25] A.F. Carley, L.A. Dollard, P.R. Norman, C. Pottage, M.W. Roberts, Journal of Elec-
tron Spectroscopy and Related Phenomena 98–99 (1999) 223.
26] R.J. Cole, N.J. Brooks, P. Weightman, Physical Review Letters 78 (1997) 3777.
[27] M. Kuhn, T.K. Sham, Physical Review B 49 (1994) 1647.