Pd-modified Au on carbon as an effective and durable catalyst for the direct oxidation of HMF to 2,5-furandicarboxylic acid

Article abstract of DOI:10.1002/cssc.201200778

Mixed noblility: We show that the modification of a gold/carbon catalyst with platinum or palladium produces stable and recyclable catalysts for the selective oxidation of 5-hydroxymethylfurfural (HMF) to 2,5-furandicarboxylic acid (FDCA): the support and nanoparticle chemistry directly mediate the selective oxidation of terminal hydroxyl groups in bio-derived HMF. This finding is a significant advance over current conversion technology because of the technological importance of FDCA.

Full text of DOI:10.1002/cssc.201200778

CHEMSUSCHEM
COMMUNICATIONS
DOI: 10.1002/cssc.201200778
Pd-modified Au on Carbon as an Effective and Durable
Catalyst for the Direct Oxidation of HMF to 2,5-
Furandicarboxylic Acid
Alberto Villa,[a] Marco Schiavoni,[a] Sebastiano Campisi,[a] Gabriel M. Veith,[b] and
Laura Prati*[a]
We show that the modification
of a Au/C catalyst with Pt or Pd
produces stable and recyclable
catalysts for the selective oxida-
tion of 5-hydroxymethylfurfural
Scheme 1. Reaction pathway for aqueous HMF oxidation.
(HMF) to 2,5-furandicarboxylic
acid (FDCA). This finding is a sig-
nificant advance over current conversion technology because
of the technological importance of FDCA. Indeed, FDCA has
been identified as one of twelve potential building blocks for
the production of value-added chemicals derived from
biosources.[1] FDCA is a potential replacement source of
terephthalic acid; the monomer is currently used for the pro-
duction of polyethylene terephthalate (PET) and derived from
hydrocarbon sources.[2]
performed to increase the yield to FDCA by optimising the re-
action conditions (especially temperature), HMF/metal or HMF/
base ratios and support selection, that is, CeO2, TiO2 and
AC.[9–11] Studies by Davis et al. showed that Au catalysts were
more active than Pt in the first step of the oxidation reaction
(HMF to HFCA),[11] but they exhibited a lower activity for the
successive conversion of HFCA to FDCA. Interestingly, the use
of basic supports such as hydrotalcite has been reported to be
beneficial in enhancing the formation of FDCA even in the ab-
sence of a free base.[12] It should be noted that the role of
a basic environment can be beneficial from different points of
view as it reduces degradation of HMF (at 1008C in water HMF
degrades by 30% in 2 h), but also enhances the activity of the
Au catalysts.
FDCA is commonly synthesized through oxidation of HMF,
which is produced through cellulose depolymerisation and
subsequential dehydration.[3] The synthesis of FDCA from HMF
has been reported using stoichiometric amounts of oxidants,
such as KMnO4, and homogeneous metal salts (Co/Mn).[4,5]
A green alternative to KMnO4 involves the use of molecular
oxygen, in the presence of a heterogeneous catalyst, with
water as the solvent. Different systems using the classic Pt-,
Pd- or Ru-based heterogeneous catalysts have been reported
for the selective oxidation of HMF to FDCA,[6,7] but these sys-
tems are generally lacking stability or selectivity. In addition to
catalyst type it has been demonstrated by Verdeguer et al. that
basicity has a major influence on catalytic activity (and selectiv-
ity) of Pt/Pb-supported catalysts.[8] Furthermore, these authors
showed that the production of FDCA occurred in two steps.
In the first step, 5-hydroxymethyl-2-furancarboxylic acid (HFCA)
is formed by oxidation of the aldehyde group. Subsequent oxi-
dation of the hydroxyl group forms FDCA (Scheme 1).
However, although the activity of the Au catalyst is good,
they deactivate rapidly.[9] This deactivation is attributed to the
irreversible absorption of intermediates.[11,13] Recently, Pasini
et al. showed that alloying Cu with Au/TiO2 resulted in a cata-
lyst with significantly higher activity and drastically enhanced
stability.[14] However, in these stability studies the final product
was mainly HFCA instead of the desired FDCA.
The catalysts presented here overcome these limitations,
making them suitable for industrial-scale applications. Indeed,
we show that Au nanoparticles supported on activated carbon
(AC) under realistic reaction conditions are able to catalyze the
oxidation of HMF to FDCA. Similarly prepared Pd and Pt cata-
lysts showed high selectivity to the intermediate HFCA and
not FDCA. Moreover, by modifying Au/AC with Pd or Pt we ob-
tained catalysts that catalyzed both steps of the reaction (HMF
to HFCA and HFCA to FDCA), showing an excellent stability
during the recycling test.
Recently, a supported Au catalyst has been found to be
active in the oxidation of HMF, and many studies have been
[a] Dr. A. Villa, M. Schiavoni, S. Campisi, Prof. L. Prati
Department of Chemistry
Universitꢀ degli Studi di Milano
via Golgi 19—20133 Milano (Italy)
Fax: (+39)02503-14405
Monometallic polyvinyl alcohol (PVA)-protected Au, Pd, and
Pt nanoparticles were prepared on AC supports by impregna-
tion following a previously reported procedure.[15] All the cata-
lysts showed a similar mean particle diameter of 2.9–3.9 nm as
estimated from TEM studies (Figures S1 and S2 in the Support-
ing Information). The activity of these catalysts was evaluated
by oxidizing 0.15m of HMF in water with 2 equivalents of
NaOH, at 608C and 3 bar O2 (HMF/metal=200 mol/mol). The
E-mail: laura.prati@unimi.it
[b] Dr. G. M. Veith
Materials Science and Technology Division
Oak Ridge National Laboratory
Oak Ridge, TN, 37831 (United States)
Supporting Information for this article is available on the WWW under
http://dx.doi.org/10.1002/cssc.201200778.
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2013, 6, 609 – 612 609

CHEMSUSCHEM
COMMUNICATIONS
www.chemsuschem.org
conditions were the same as those reported in the literature to
allow for an easier comparison.[11,14] Profiles showing the per-
cent fraction of HMF, HFCA and FDCA as a function of time are
shown in Figure 1. From these results it is clear that the Au/AC
catalyst showed a higher selectivity to FDCA, converting nearly
all the intermediate HFCA at 80% completion. In contrast, the
Pd and Pt catalysts only converedt 9 and 20% of the HMF to
FDCA, respectively. The majority of the products are trapped in
the form of the HFCA intermediate, indicating that the Pd and
Pt materials was deactivated.
To quantify the effect of support on the reaction, different
carbon-based supports were investigated: AC, graphite
(Graph), carbon nanofibers (CNFs) and carbon nanotubes
(CNTs). The supports were impregnated as before by using
a AuPVA solution. There are numerous differences between the
various forms of carbon. Among the most important variables
for catalysis are the graphitic content and the surface function-
ality of the carbon. Comparing the D/G band ratios using
Raman spectroscopy confirmed the predicted graphitization
trends based on graphite form, that is, Graph>CNT>AC>
CNF (Table S3, Supporting Information).[16]
Analysis of X-ray photoelectron spectroscopy (XPS) data re-
vealed significant differences in oxygen concentrations on the
carbon surfaces. AC (8 at%), CNF (ꢀ7 at%)>CNT (ꢀ2 at%)@
Graph (0 at%).[17,18] The AC support did contain approximately
1—2 at% K on the surface of the carbon, which is a by-prod-
uct of the carbon source. Analysis of the Au particles sizes,
through TEM studies (Figure 2), revealed that the average par-
ticle sizes increased with decreasing oxygen content and disor-
Figure 2. TEM images of a) Au/AC, b) Au/CNFs, c) Au/CNTs and d) Au/Graph.
der: Au/AC (2.9 nm)>Au/CNFs (3.8 nm)>Au/CNTs (4.6 nm)>
Au/graph (5.4 nm) (Figure S1, Supporting Information). This is
not surprising as higher disorder and oxygen groups provide
nucleation sites to trap and stabilize the supported Au clusters.
However, despite the variation in size, the absolute magnitude
of variation was quite small indicating that in all cases excel-
lent particle distributions were obtained. It should be noted
that in the case of the CNTs some empty tubes and Au aggre-
gates were observed.
Figure 1. Product distribution (& HMF, * HFCA, ~ FDCA), for a) Pt/AC,
b) Pd/AC and c) Au/AC.
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2013, 6, 609 – 612 610

CHEMSUSCHEM
COMMUNICATIONS
www.chemsuschem.org
fore, support chemistry is likely the dominant factor in media-
ting activity. We note that AC did have a small concentration
of residual K on the surface, which may influence the basicity
of the catalyst in a manner similar to the work reported on hy-
drotalcite.[12] For determining the possible active role of K con-
tent, we doped the CNF support with K following the proce-
dure reported elsewhere.[19] Therefore, 2 at% K was added to
CNF and K-CNF was used as the support for AuPVA nanoparti-
cles. Alternatively, we also impregnated Au/CNF with a solution
of KBr (1 wt%). The catalytic tests, however, revealed in both
cases similar activity and selectivity of Au/CNF and Au/K-CNF
(Table 1), thus excluding a positive effect of K.
Table 1. HMF oxidation using Au based catalysts.
Catalyst[a]
Time of reaction Conversion
Selectivity [%]
[h]
[%]
HFCA
FDCA
monometallic
Au/AC
2
6
2
6
2
6
2
6
2
6
2
6
2
6
>99
>99
85
>99
88
>99
>99
>99
>99
>99
>99
>99
>99
>99
64
20
92
91
80
80
98
94
95
98
81
61
87
74
36
80
8
Pd/AC
9
Pt/AC
20
20
2
6
5
Au/CNFs
Au/K-CNFs
Au/CNTs
Au/Graph
Further efforts focused on the recyclability and stability of
the Au/AC catalyst. The catalyst was allowed to react for 6 h
each run and was then filtered and reused without further
treatment in the following run (Figure 3). The catalyst showed
good product selectivity, but it underwent deactivation, losing
20% of conversion efficiency after the fifth run. No Au leaching
from the catalyst was detected, as reported for other Au cata-
lysts; thus, this deactivation was possibly ascribed to irreversi-
2
19
39
13
26
bimetallic
Au6-Pd4/AC
2
4
2
4
2
4
2
4
>99
>99
>99
>99
>99
>99
>99
>99
5
–
1
95
>99
99
Au8-Pd2/AC
Au6-Pt4/AC
Au8-Pt2/AC
–
38
6
>99
62
94
ble
adsorption
of
intermediates
or
Au
particle
agglomeration.[9,13]
24
3
76
97
For the liquid-phase alcohol oxidation, we recently found
that it is possible to increase the activity and the stability of
the Au catalysts by alloying Au with Pd or Pt.[20,21] Moreover,
we found that varying the Au/precious metal ratio, the catalyst
activities varied depending on the nature of the alcohol.[22–24]
We hypothesized that these findings could also be applied to
HMF oxidation. Au bimetallic catalysts, with different Au/Pd (or
Pt) molar ratios (6:4 and 8:2) were prepared following a previ-
ously reported procedure, which ensured formation of alloyed
bimetallic nanoparticles.[23] Characterization of these catalysts
revealed similar metallic-particle dimensions (3.1–3.5 nm, Figur-
es S1 and S2 in the Supporting Information).
base-free experiments
Au/AC
Au8-Pd2/AC
6
6
5
3
–
–
–
–
[a] Reaction conditions: 0.15m, pO =3 bar, stirring at 1250 rpm, 608C,
2
HMF/metal=200 mol/mol, 2 eq NaOH.
Evaluation of the catalysts supported on different carbons
revealed significant variation in activity compared to the Au/
AC samples, Table 1. The Au particles on CNTs, Graph, and
CNFs are as active as Au on AC in the first step of the oxida-
tion to produce HFCA.
Comparing the activity of these catalysts with the pure Au
materials, we found that all the bimetallic catalysts showed an
However, the catalysts are extremely poorly active for the
subsequent oxidation of HFCA
to FDCA. The selectivity to FDCA
after 6 h was 80% for Au/AC,
6% for Au/CNFs, 39% for Au/
CNTs and 26% for Au/Graph, re-
spectively (Table 1). Clearly, Au/
AC was more active in the oxida-
tion of the hydroxyl group of
HFCA, producing more FDCA
compared to the other catalysts.
This result indicates the impor-
tance of the support in media-
ting the overall activity and se-
lectivity towards Au particle size.
However, we believe the differ-
ence in particle size plays
a minor role as Au/CNF, which
has a mean particle size very
similar to Au/AC (3.8 vs. 2.9 nm),
completely differs in activity for
the oxidation of HFCA. There- Figure 3. Stability tests for Au-based catalysts.
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2013, 6, 609 – 612 611

CHEMSUSCHEM
COMMUNICATIONS
www.chemsuschem.org
enhanced activity with respect to the corresponding mono-
metallic counterpart (Table 1). AuPd catalysts appeared more
active than AuPt ones, reaching almost full conversion to
FDCA after only 2 h, whereas AuPt needed 4 h for reaching
almost the same results. In both cases it was observed that the
activity increased by increasing the Au content, Au6M4/AC
being less active than Au8M2/AC, which is in agreement with
the trend observed in glycerol or alcohols oxidations.[22–24]
In addition, these catalysts were fully characterized elsewhere;
they all presented randomly alloyed phases without any pecu-
liar differences from a structural point of view.[23] Therefore,
a possible explanation for the different activity of these two
catalysts is likely attributable to different electronic properties
derived from different isolation of active sites.
a high stability for the production of FDCA from HMF. There-
fore, this latter catalyst could be proposed as a good candidate
for the large-scale production of FDCA from HMF.
Acknowledgements
A portion of this work (G.M.V.) was supported by the US Depart-
ment of Energy’s Office of Basic Energy Science, Division of Mate-
rials Sciences and Engineering.
Keywords: alloys · gold · hydroxymethylfurfural · oxidation ·
palladium
As our main concern was the durability of the catalysts, we
carried out stability tests similar to those performed with
Au/AC using the AuPd catalysts. In Figure 3, the conversions
after 6 h reactions for Au/AC and 2 h reactions for Au6Pd4/AC
and Au8Pd2/AC are compared. Au6Pd4/AC followed the same
deactivation profile as observed for the Au/AC samples.
Furthermore, as in the case of Au/AC, inductively coupled
plasma (ICP) measurements did not reveal any leaching of Au
or Pd. However, a different behaviour was observed for the
Au8Pd2/AC catalyst. This catalyst was able to completely con-
vert HMF at the same rate over all five runs, maintaining 99%
selectivity to FDCA during the whole experiment. Thus, alloy-
ing Au with Pd in a 8:2 atomic ratio produced an active and
stable catalyst for FDCA production from HMF, eliminating the
deactivating phenomena shown by Au/AC catalysts.
[1] T. Werby, G. Petersen, Top Value-Added Chemicals from Biomass, Vol. 1,
Pacific Northwest National Laboratory 2004, 27.
[2] A. Gandini, A. J. D. Silvestre, C. Pascoal Neto, A. F. Sousa, M. Gomes, J.
Polym. Sci. Part A 2009, 47, 295.
[3] J. N. Chheda, G. W. Huber, J. A. Dumesic, Angew. Chem. 2007, 119, 7298;
Angew. Chem. Int. Ed. 2007, 46, 7164.
[4] T. Miura, H. Kakinuma, T. Kawano, H. Matshisa, US7411078, 2008.
[5] W. Partenheimer, V. V. Grushin, Adv. Synth. Catal. 2001, 343, 102.
[6] P. Vinke, W. Van der Poel, H. van Bekkum, Stud. Surf. Sci. Catal. 1991, 59,
385.
[7] M. Krçger, U. Prusse, K.-D. Vorlop, Top. Catal. 2000, 13, 237.
[8] P. Verdeguer, N. Merat, A. Gaset, J. Mol. Catal. 1993, 85, 327.
[9] O. Casanova, S. Iborra, A. Corma, ChemSusChem 2009, 2, 1138.
[10] Y. Y. Gorbanev, S. K. Klitgaard, J. M. Woodley, C. H. Christensen, A. Riisag-
er, ChemSusChem 2009, 2, 672.
[11] S. E. Davis, L. R. Houk, E. C. Tamargo, A. K. Datye, R. J. Davis, Catal. Today
2011, 160, 55.
[12] N. K. Gupta, S. Nishimura, A. Takagaki, K. Ebitani, Green Chem. 2011, 13,
824.
[13] B. N. Zope, R. J. Davis, Green Chem. 2011, 13, 3484.
[14] 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 Chem. 2011, 13, 2091.
We also tested Au8Pd2/AC and Au6Pd4/AC in the absence of
a base. They were active also under these conditions, but the
yield of FDCA and also of HFCA was very low (<10%). These
results did not surprise us as it was reported elsewhere that,
by avoiding a basic environment, HMF rapidly degrades to lev-
ulinic and formic acids and FDCA to glycolic and formic acid.[14]
Moreover, under such conditions also the stability of the cata-
lysts was reduced.
[15] L. Prati, G. Martra, Gold Bull. 1999, 32, 96.
[16] R. Al-Jishi, G. Dresselhaus, Phys. Rev. B 1982, 26, 4514.
[17] L. Prati, A. Villa, A. R. Lupini, G. M. Veith, Phys. Chem. Chem. Phys. 2012,
14, 2969.
[18] J. P. Tessonnier, D. Rosenthal, T. W. Hansen, C. Hess, M. E. Schuster, R.
Blume, F. Girgsdies, N. Pfaender, O. Timpe, D. S. Su, R. Schlçgl, Carbon
2009, 47, 1779.
[19] Y. Zhao, X. Liu, K. Xin Yao, L. Zhao, Y. Han, Chem. Mater. 2012, 24, 4725–
4734.
[20] D. Wang, A. Villa, F. Porta, D. Su, L. Prati, Chem. Commun. 2006, 1956.
[21] N. Dimitratos, A. Villa, D. Wang, F. Porta, D. Su, L. Prati, J. Catal. 2006,
244, 113.
[22] A. Villa, C. Campione, L. Prati, Catal. Lett. 2007, 115, 133.
[23] D. Wang, A. Villa, F. Porta, L. Prati, D. Su, J. Phys. Chem. C 2008, 112,
8617.
[24] A. Villa, N. Janjic, P. Spontoni, D. Wang, D. S. Su, L. Prati, Appl. Catal. A
2009, 364, 221.
In summary, mono-metallic (Au, Pd, Pt) and bi-metallic cata-
lysts have been prepared by solution immobilization using PVA
as a protective agent and carbon-based supports. Au/AC can
effectively oxidize HMF to FDCA, whereas Pd and Pt rapidly de-
activate during the oxidation of the intermediate HFCA to
FDCA. The key factor influencing the activity of the Au-based
catalyst seems to be the support. Indeed, when comparing Au
supported on various carbon supports, only Au on AC effec-
tively oxidized HMF to FDCA. Au/AC was not stable in recy-
cling experiments. However, its stability (and activity) was in-
creased extraordinarily by alloying Au with Pd. In particular, we
showed that Au8Pd2/AC, where Au and Pd were present at
a 8:2 ratio, presented a high activity and most importantly
Received: October 18, 2012
Revised: December 18, 2012
Published online on March 11, 2013
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2013, 6, 609 – 612 612