Titania-supported catalysts for levulinic acid hydrogenation: Influence of support and its impact on γ-valerolactone yield

Article abstract of DOI:10.1002/cssc.201403332

A series of titania-supported ruthenium and platinum catalysts was investigated in the levulinic acid hydrogenation towards γ-valerolactone, a key reaction for the catalytic transformation of biomass. It was shown that various morphologies and phases of titania strongly influence the physicochemical and catalytic properties of supported Ru and Pt catalysts in different ways. In the case of the catalyst supported on mixed TiO₂ phases, Ru particles are exclusively located on the minority rutile crystallites, whereas such an effect was not observed for platinum. The platinum catalyst activity could be increased when the metal was dispersed on the large surface-area anatase, which was not the case for ruthenium as a result of its agglomeration on this support. The activity of ruthenium on anatase could be increased in two ways: a) when RuO₂ formation during catalyst preparation was avoided; b) when pure anatase support material was modified so that it exhibited no microporosity. The obtained results allow a better understanding of the role of the support for Ru and Pt catalysts.

Full text of DOI:10.1002/cssc.201403332

DOI: 10.1002/cssc.201403332
Full Papers
Titania-Supported Catalysts for Levulinic Acid
Hydrogenation: Influence of Support and its Impact on
g-Valerolactone Yield
[a]
[a]
[a]
[a]
[b]
[a]
A. M. Ruppert,* J. Grams, M. J e˛ drzejczyk , J. Matras-Michalska, N. Keller, K. Ostojska,
[c]
and P. Sautet
A series of titania-supported ruthenium and platinum catalysts
was investigated in the levulinic acid hydrogenation towards
g-valerolactone, a key reaction for the catalytic transformation
of biomass. It was shown that various morphologies and
phases of titania strongly influence the physicochemical and
catalytic properties of supported Ru and Pt catalysts in differ-
ent ways. In the case of the catalyst supported on mixed TiO2
phases, Ru particles are exclusively located on the minority
rutile crystallites, whereas such an effect was not observed for
platinum. The platinum catalyst activity could be increased
when the metal was dispersed on the large surface-area ana-
tase, which was not the case for ruthenium as a result of its
agglomeration on this support. The activity of ruthenium on
anatase could be increased in two ways: a) when RuO forma-
2
tion during catalyst preparation was avoided; b) when pure
anatase support material was modified so that it exhibited no
microporosity. The obtained results allow a better understand-
ing of the role of the support for Ru and Pt catalysts.
Introduction
Production of currently used chemicals, materials, and energy
is dependent on fossil fuel reserves. Their rapid depletion, to-
considered a platform molecule for conversion to many useful
chemicals such as monomers for the synthesis of nylon, poly-
mers of high thermal stability. As recently shown by Lange
et al., it can be transformed into valeric (pentanoic) liquid bio-
gether with the necessity of limiting the growing CO emis-
2
[1–7]
sions, create the need to develop alternative value chains.
[14]
This topic has recently been addressed in several excellent re-
views showing new perspectives for transforming biomass into
fuels by its conversion via valeric acid to its esters. GVL can
be obtained from cellulose or sugars in several steps that can
be performed via cellulose hydrolysis and glucose dehydration
to hydroxymethylfurfural (HMF) and then to levulinic acid (LA)
in the presence of acidic catalysts. In this reaction, formic acid
is formed in equimolar amounts to LA and interesting exam-
[
8]
biofuels, stressing the difficulties related to the recalcitrant
[
9]
nature of cellulose or emphasizing the biomass transforma-
[
10,11]
tion to chemicals.
In this respect, g-valerolactone (GVL),
which can be derived from biomass, has drawn an increasing
attention. Thanks to its low volatility, minimum toxicity, and
ples of its applications as an internal hydrogen source can be
[
12,13]
[15,16]
good stability,
it has potential versatile applications and is
found in the literature.
Catalytic hydrogenation of the
formed LA is performed with molecular hydrogen in the pres-
ence of a hydrogenation catalyst. According to the literature,
[
a] Dr. A. M. Ruppert, Dr. J. Grams, Dr. M. J e˛ drzejczyk , J. Matras-Michalska,
K. Ostojska
Institute of General and Ecological Chemistry
Faculty of Chemistry
Łꢀd z´ University of Technology
ul. Z˙ eromskiego 116
[17–19]
in mild reaction conditions and in protic solvent,
Scheme 1 illustrates a plausible mechanism that can match the
overall reaction pathway, with the hydrogenation of the levu-
linic acid carbonyl group to form the 4-hydroxypentanoic acid
9
0-924 Łꢀd z´ (Poland)
(
HPA) intermediate that subsequently undergoes intramolecu-
E-mail: agnieszka.ruppert@p.lodz.pl
lar esterification yielding g-valerolactone.
[b] Dr. N. Keller
Ruthenium and platinum are considered as catalysts of
choice for this process. In the literature there are several suc-
cessful examples in which both Pt and Ru, mainly supported
on carbon, were used in the hydrogenation of levulinic acid to-
Institut de Chimie et Procꢁdꢁs pour l’Energie
l’Environnement et la Santꢁ (ICPEES)
CNRS, University of Strasbourg
2
6
5 rue Becquerel
7087, Strasbourg (France)
[17,20–22]
wards g-valerolactone.
Also bimetallic systems like Ru–
[
c] Dr. P. Sautet
Universitꢁ de Lyon
CNRS, Ecole normale superieure de Lyon
Institut de Chimie de Lyon
Laboratoire de Chimie
Re were investigated, in which the addition of a second metal
[23,24]
was beneficial for increasing the activity.
The groups of
[25]
[26]
Weckhuysen and Palkovits demonstrated that the solvent
choice can have a strong influence on the ruthenium catalytic
performance and reaction yield as well. The influence of the
metal support on the catalytic performance has been a topic
of several recent studies and also proved to be a crucial factor.
4
6 allꢁe d’Italie
6
9364 Lyon Cedex 07 (France)
Supporting Information for this article is available on the WWW under
http://dx.doi.org/10.1002/cssc.201403332.
ChemSusChem 0000, 00, 0 – 0
1
ꢀ 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
screened titanias of different
crystalline phases (rutile and
anatase) with both low and high
surface area, as well as anatase–
rutile mixed phase TiO . All ma-
2
terials were tested after impreg-
nation with 1 wt% of Ru or Pt,
which resulted in different types
of dispersion of the metal parti-
cles and in metal–support inter-
actions. Different preparation
conditions and metal precursors
were evaluated as well. More-
Scheme 1. Plausible mechanism of the levulinic acid hydrogenation.
over, we used very mild reaction
conditions and water as an envi-
ronmentally friendly solvent,
The group of Weckhuysen showed that the support acidity can
since GVL does not form an azeotrope with water and can be
[
24]
[12]
enhance the conversion of levulinic acid. It was also shown
that strong acidic sites present in zeolite-supported catalysts
can even push the reaction course towards pentanoic acid. In
removed easily by distillation.
the same study, a titania-supported ruthenium catalyst was Results and Discussion
proved to be remarkably stable and selective for GVL forma-
Characterization of titania supports
tion. Furthermore, Lange in his earlier study showed that,
among many tested materials, the best performance in levulin-
ic acid hydrogenation was shown by Pt/TiO and Pt/ZrO espe-
The properties of titanias used as supports are listed in Table 1.
The materials exhibit different phase compositions, from full
2
2
cially with respect to the stability
of the support during extended
[
14]
reaction times.
For example,
Table 1. Physicochemical characterization of different titania.
acidic or basic oxide supports
such as alumina, silica–alumina,
and magnesium oxide suffer
from strong leaching or even
dissolution during the process,
whereas carbon, which is often
Property
Sample
TiO
2
(R100) TiO
2
(R20) TiO
2
(R10) TiO
2
(A-1) TiO
2
(A-2) TiO
2
(A-3) TiO
2
(A-4)
[
a]
composition [%]
R:100
17
R:20
A:80
R:32
A:22
55
R:10
A:90
R:23
A:13
90
A:100
A:100
9
A:100
9
A:100
7
[
a]
average crystallite size [nm]
14
2
ꢀ1
[
20,21,27,28]
BET [m g
]
26
0
83
0
0.16
5.7
312
91
0.59
4.4
330
243
0.22
5.5
336
222
0,21
5.25
used in this process
and
2
ꢀ1
microporous surface area [m g
pore volume [cm g
isoelectric point
]
0
0
takes advantage of a high resist-
ance to aggressive reaction con-
ditions, can burn off during the
regeneration step. On the other
hand, Palkovits et al. in their
work concerning the influence
3
ꢀ1
]
0.19
3.6
0.12
6.25
0.24
5.5
[a] A: anatase or R: rutile, derived from XRD measurements.
of the solvent and support on the catalytic performance of in-
vestigated materials for the hydrogenation of levulinic acid,
also indicated differences in the catalytic behavior of two dif-
rutile phase for TiO (R100) to anatase only for samples labelled
2
from TiO (A-1) to TiO (A-4), and high and low specific surface-
2
2
area titanias were investigated. TiO (A-1) sample was prepared
2
[
25]
ferent TiO materials impregnated with Ru.
by calcining TiO (A-2) at 5008C to get micropore-free pure ana-
2
2
Therefore, it appears of high interest to obtain a deeper in-
sight on the influence of the support character and method of
tase TiO with relatively low surface area, comparable to rutile-
2
containing samples such as TiO (R20) and TiO (R10). TiO (R20)
2
2
2
[25]
preparation on the catalytic behavior of Ru/titania catalysts.
and TiO (R10) mixed phase samples displayed the largest ana-
2
In order to be able to clearly present many aspects concerning
tase and rutile mean crystallite size, whereas the smallest ana-
tase mean crystallite sizes were identified for high surface-area
TiO (A-2), TiO (A-3), and TiO (A-4) pure anatase samples, and
the Ru/TiO catalysts, we decided to focus first on simplified
2
conditions using LA hydrogenation in the presence of molecu-
lar hydrogen as a model reaction of biomass-derived mole-
cules. To the best of our knowledge, there is no detailed study
concerning the influence of the properties of titanias on LA hy-
drogenation. Therefore, we decided to explore this topic and
evaluate which features of the titania support can influence
the activity of levulinic acid hydrogenation catalysts. We
2
2
2
the smallest rutile mean crystallite size was found for pure
rutile TiO (R100) sample.
2
In terms of isoelectric point, the lowest values were record-
ed for TiO (R100) and TiO (A-2), at 3.6 and 4.4, respectively, and
2
2
the highest one for TiO (R20) at 6.25, whereas all other samples
2
had similar values within the 5.25–5.7 range.
&
ChemSusChem 0000, 00, 0 – 0
www.chemsuschem.org
2
ꢀ 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
TEM and chemisorption measurements of catalysts
Table 2. Average metal particle size (nm) for the supported catalysts de-
rived from TEM and from chemisorption.
TEM as well as hydrogen and CO chemisorption studies were
used for measuring the metal-particle size distribution for Pt
Catalyst
Particle size [nm]
chemisorption
TEM
and Ru supported on the various TiO materials (Table 2). Eval-
2
Pt/TiO
Pt/TiO
Pt/TiO
Pt/TiO
Pt/C
Ru/TiO
Ru/TiO
Ru/TiO
Ru/TiO (A-4)
Ru/TiO (A-3)
Ru/TiO
Ru/TiO
Ru/C
2
2
2
2
(R20)
(R10)
(A-2)
(A-3)
3.2
–
2.7
3.3
2.2
3.7
–
uation of the metal particle size as a function of the support
surface area is not trivial. Judging from the TEM measure-
ments, mean particle size generally counter-intuitively increas-
es with increasing titania surface area. For the high surface-
area TiO (A-2-) and TiO (A-3) anatase samples, despite the pres-
4.5
5.4
4.4
2.1
2.9
2
(R20)
(R10)
(A-2)
3.3
–
2
2
2
2
ence of micropores, TEM did not evidence the formation of
smaller Pt particles in comparison to lower surface-area TiO2
[
a]
2.2
3.2
2.5
8.6
–
–
2.3
3.4
2
2
–
[
a]
supports. Indeed, on TiO (R20) support, the average Pt particle
2
(A-1)
(R100)
4.5
2.2
3.1
3.9
–
size obtained from TEM was about 3.2 nm with rather homo-
geneous distribution, whereas Pt particle clusters were ob-
2
2
served on the much higher surface-area TiO (A-2), with a mean
[b]
2
Ru/TiO (A-3)
Ru/TiO (A-2)
2
[
b]
Pt particle size of 5.4 nm (Figure 1). Such an agglomeration
tendency of platinum on titania support has been reported
earlier and it was associated with pH change during impregna-
tion; it can occur when the values of isoelectrical point and pH
2
[
a] It was not possible to measure the average metal particle size by TEM
in some cases due to the presence of large agglomerates and the limited
number of crystallites detectable by microscopy. All the Pt and Ru cata-
lysts were prepared following the incipient wetness impregnation
method with final thermal reduction treatment except: [b] Catalysts were
prepared by liquid phase direct chemical reduction. In that case the CO
chemisorption was performed.
[
29]
of the impregnation solution are too close.
On the other hand, hydrogen chemisorption gave contrast-
ing results (Table 2.), with an average Pt particle size of 2.2 nm
for the large surface-area Pt/TiO (A-2) material. This
2
might indicate that besides the large particles ob-
served by TEM, very small ones may also be present,
which could be localized in the micropores and
could not be visible by microscopy, probably due to
both the TEM detection limit and contrast limitation.
For a comparison, the platinum particles in the
carbon-supported catalyst were distributed homoge-
neously, with an average size of 4.4 nm (Figure 1).
Chemisorption TEM studies were also performed for
supported ruthenium catalysts. For Ru/TiO (R20),
2
a high metal dispersion was observed with an aver-
age particle size of 2.1 nm, but it was noticed that
the location of metal particles did not appear to be
homogeneous within the TiO sample; not all titania
2
crystallites were fully decorated by metal particles
(Figure 2a). A similar behavior was observed in the
case of Ru/TiO (R10), with the presence of bare tita-
2
nia crystallites beside crystallites supporting rutheni-
um particles. This was investigated further and,
based on the measurements of lattice fringe (Fig-
ure 3a), it was possible to identify that ruthenium
particles were mainly decorating the rutile phase of
titania, and anatase crystallites remained naked. Pos-
sibly due to the location only on rutile crystallites
and to the half rutile content of Ru/TiO (R10) com-
2
pared to that of Ru/TiO (R20) (Table 1), the average
2
metal particle size was slightly larger (2.9 nm) on the
TiO (R10) support. The TiO (R100) support solely con-
2
2
sisting of rutile phase, although with a reduced sur-
face area, exhibited a good Ru dispersion with
a small average particle size of 2.2 nm (Figure 2c),
similar to that obtained on TiO (R20) with anatase–
2
Figure 1. TEM images and the corresponding metallic particle size distribution histo-
2 2
rutile mixed phase and a twice higher surface area grams for selected platinum catalysts : a) 1%Pt/TiO (R20); b) 1%Pt/TiO (A-3); c) 1%Pt/C.
ChemSusChem 0000, 00, 0 – 0
www.chemsuschem.org
3
ꢀ 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
(
Figure 2a). On the other hand,
the impregnation of ruthenium
on the large surface-area TiO (A-
2
2
), TiO (A-3), and TiO (A-4) ana-
2
2
tase resulted in a largely inho-
mogeneous metal distribution
and very large Ru particles were
observed (Figure 3b). This sug-
gests that the adhesion of Ru
particles on the anatase surfaces
is very low, resulting in a poor
wetting. This was previously de-
[30]
scribed in the literature. If only
anatase is present, high disper-
sion of this metal therefore
seems unlikely. However, again,
hydrogen chemisorption gives
contrasting results, indicating
that Ru dispersion is high with
a calculated average particle size
in the 2.2–3.2 nm range. Similarly
to Pt/TiO (A-2) catalyst, this con-
2
firms that small Ru particles not
detectable by TEM could be
formed in the micropores only,
in addition to the presence of
large agglomerates observed in
TEM images, so that the mean
particle size derived from H2
chemisorption was averaged.
This indicates that anatase is
able to host small Ru particles in
the micropores, while the exter-
nal surfaces of the grains show
a weak adhesion with the Ru
nanoparticles. In order to ana-
lyze how the presence of micro-
pores can influence the metal
distribution, an anatase sample
without micropores, that is,
TiO (A-1), was prepared by high-
2
Figure 2. TEM images and the corresponding metallic particle size distribution histograms for selected ruthenium
catalysts: a) 1%Ru/TiO (R20); b) 1%Ru/TiO (R10); c) 1%Ru/TiO (A-4); d) 1%Ru/TiO (A-1).
temperature
008C of micropore-containing
high surface-area anatase, that
calcination
at
2
2
2
2
5
is, TiO (A-2), and ruthenium was further impregnated in the
2
same way. In that case, Ru particle agglomerates were no
longer observed and rather homogenous distribution was ob-
tained with an average particle size of 4.5 nm (Figure 2d).
Thus, the crystalline phase is not the only factor determining
the metal distribution.
Another attempt was made to achieve a homogeneous
metal distribution on large surface-area microporous anatase.
It is generally assumed that, upon impregnation, the fixation/
deposition of Ru(acac) occurs due to the exchange of acetyla-
3
Figure 3. TEM images of a) Ru/TiO
between anatase and rutile crystalline phases of the TiO
performed by assigning the 2.49 ꢁ and 3.52 ꢁ interplanar distances mea-
2
(R10); b) Ru/TiO
2
(A-3). The discrimination
[32,33]
cetonate ligands with hydroxyl groups of the support.
The
2
(R10) support was
removal of acetylacetonate ligands is accompanied by their de-
struction during oxidation at high temperature in air or
[
31]
sured to the (101) and (101) planes of rutile and anatase, respectively.
&
ChemSusChem 0000, 00, 0 – 0
www.chemsuschem.org
4
ꢀ 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
oxygen; during the subsequent reduction step the ruthenium
particles are formed. Assuming that model, the migration of
particles should occur during the oxidation and/or reduction
step during thermal treatment. Therefore, a series of catalysts
was prepared via chemical reduction so as not to subject the
catalyst to any thermal treatment, in order to evidence how
the metal distribution can be influenced by temperature and
how this would modify the catalytic activity.
1008C for Ru/TiO (A-3) and Ru/TiO (A-2), respectively, and
2
2
could be linked to the reduction of small RuO crystallites that
x
could be localized in micropores in interaction with titania,
[34]
which is in agreement with the literature. By contrast, the re-
duction profiles of rutile-containing samples strongly differed
to those of pure anatase samples, with the observation of
both low (70–908C) and higher temperature signals (130–
1708C) (Figure 4d–f). The former are most probably related to
amorphous RuO and well dispersed amorphous RuO , while
The chemical reduction approach allowed the preparation of
pure anatase-based catalysts with a relatively high metal dis-
persion and a smaller particle size, in comparison to analogous
catalysts prepared through the impregnation method with
thermal treatment (e.g., with a high mean metal-particle size
2
x
the latter can be assigned to crystalline RuO in strong interac-
x
[35]
tion with titania.
Taking into account the TEM images
(Figure 2), there is a clear tendency of deposition of Ru on
rutile, so we can assume the presence of a strong metal–rutile
interaction. Therefore, the higher the rutile contribution, the
stronger the high-temperature effect observed (Figure 4). A dif-
ferent reduction profile was obtained for the low surface-area
of 3.9 nm, derived from TEM for Ru/TiO (A-3), Table 2). The re-
2
sults of chemisorption studies confirm this distribution.
anatase sample, that is, Ru/TiO (A-1), with only a very narrow
2
Temperature programed reduction measurements
well-defined peak (sharp signal) centered at 1538C and a small
The reducibility of Ru/TiO catalysts was studied by tempera-
shoulder. It probably corresponds to the reduction of RuO
x
2
ture programed reduction (TPR) measurements (Figure 4). The
that weakly interact with titania. The TEM image also signifi-
cantly differs from the aforementioned cases, featuring larger
metal particles randomly distributed on anatase crystallites.
Time-of-flight secondary ion mass spectrometry analysis
Time-of-flight secondary ion mass spectrometry (TOF-SIMS)
analysis of the surface of selected catalytic systems was per-
formed. The aim was to understand the effect of the catalyst
preparation method of ruthenium-based catalysts and how it
can affect the large surface-area titania catalysts. The Ru/
TiO (R20) catalyst was examined as a reference, and did not ex-
2
hibit much difference between the samples subjected to calci-
nation, thermal reduction, and chemical reduction. Similar in-
tensities of Ru signal were observed with a slightly higher
value after chemical reduction (Table 3), probably resulting
from a partial coverage of metal particles by titania migrating
[36]
during the thermal treatment, so that Ru gives lower intensi-
ty on the surface.
More visible differences were observed for high-surface-area
Ru/TiO (A-2) and Ru/TiO (A-3) catalysts. Both samples after cal-
Figure 4. Temperature programmed reduction profiles for ruthenium cata-
lysts a) Ru/TiO
2
(A-3); b) Ru/TiO
2
(A-2); c) Ru/TiO
(A-1).
2 2
(A-4); d) Ru/TiO (R10); e) Ru/
2
2
TiO (R20); f) Ru/TiO
2
2
(R100); g) Ru/TiO
2
cination displayed lower Ru signals, probably caused by the
presence of large RuO agglomerates. A small Ru intensity in-
2
crease was observed after thermal reduction, probably caused
by the absence of interaction between such large agglomer-
ates and the support, so that during the high temperature hy-
reason for performing TPR measurements was to understand
what kind of interactions are present in different catalysts, and
in which way the particles of different sizes can interact with
the support in order to relate those observations to catalytic
activity. For all high surface-area anatase-supported samples,
the observed reduction profiles were very similar. The main re-
duction peak was recorded at very low temperatures (43, 53,
and 638C) for Ru/TiO (A-3), Ru/TiO (A-2), and Ru/TiO (A-4) sam-
[37]
drogen treatment, partial redispersion occurred.
+
Table 3. Normalized intensity of Ru ion calculated on the basis of TOF-
SIMS spectra collected from the surface of selected catalysts after differ-
ent treatments.
2
2
2
+
ples, respectively (Figure 4a–c), and could be assigned to the
Catalyst
Intensity of Ru ion
calcination thermal reduction chemical reduction
reduction of RuO that have no interaction with the support.
2
ꢀ
ꢀ
ꢀ
ꢀ
3
4
4
4
ꢀ3
ꢀ4
ꢀ4
ꢀ4
ꢀ3
ꢀ3
ꢀ3
The broad reduction profiles probably resulted from very
big crystallites/species, as confirmed by the observation of
large particles in TEM images (Figure 3a). Small and broad
peaks with no clear maximum were also observed at 85 and
1%Ru/TiO (R20)
1.3ꢂ10
1.8ꢂ10
2.0ꢂ10
5.2ꢂ10
1.0ꢂ10
2.4ꢂ10
2.4ꢂ10
4.3ꢂ10
1.4ꢂ10
1.0ꢂ10
1.3ꢂ10
–
2
1%Ru/TiO
2
2
2
(A-2)
(A-3)
(A-1)
1
1
%Ru/TiO
%Ru/TiO
ChemSusChem 0000, 00, 0 – 0
www.chemsuschem.org
5
ꢀ 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
By contrast, a substantial difference was visible for both cat-
alysts prepared by chemical reduction, for which the Ru signal
was considerably higher and in the same range than for the
Table 5. Catalytic results for different titania-supported catalysts at 308C
or 708C.
Catalyst
GVL yield LA conversion GVL yield LA conversion
Ru/TiO (R20) catalyst. This is in agreement with the TEM obser-
2
[
%]
[%]
[%]
[%]
vations, in which more Ru particles were observed on the sur-
Reaction temper- 308C
ature
708C
face of Ru/TiO (A-3). It was also interesting to note that TOF-
2
SIMS signals of ruthenium for Ru/TiO (A-2) and Ru/TiO (A-3)
2
2
Pt/TiO
Pt/TiO
2
(R20)
(R10)
–
–
–
–
–
–
23
18
50
41
95
100
31
7
0,1
93
83
82
81
92
27
22
54
42
99
100
38
7
samples after chemical reduction were similar.
2
It was also observed that Ru/TiO (A-1) sample with low sur-
2
Pt/TiO (A-2)
2
2
Pt/TiO (A-3)
face-area anatase exhibited a much higher ruthenium signal
than in the case of high surface-area anatase materials, but still
Ru/TiO
Ru/TiO
Ru/TiO (A-2)
2
(R20)
(R10)
62
60
4
64
60
8
3
5
54
40
43
58
54
2
lower than for Ru/TiO (R20). This is probably due to different
2
2
Ru particle distributions on the surface of anatase and to the
presence of larger metal particles.
Ru/TiO (A-4)
0.2
0.3
48
40
38
47
50
2
Ru/TiO
Ru/TiO
Ru/TiO
Ru/TiO
Ru/TiO
Ru/TiO
2
2
2
2
2
2
(A.-3)
(A-1)
(R100)
(R20)
(A-3)
(A-2)
2
99
95
90
88
99
[
a]
[
[
a]
a]
Catalytic activity
In general, no matter whether the reaction was carried out at
All the catalysts were prepared following the incipient wetness impregna-
tion method with final thermal reduction treatment: [a] Catalysts were
prepared by liquid phase direct chemical reduction.
308C or 708C, ruthenium catalysts showed better performan-
ces in the hydrogenation of levulinic acid towards GVL than
platinum catalysts, both in terms of LA conversion and GVL
[
23]
yield (Table 4), in agreement with the literature
comes to liquid aqueous phase reactions.
when it
(
Table 5). Significantly higher yields were obtained with Pt/
TiO (A-2)- and Pt/TiO (A-3)-supported catalysts.
2
2
Table 4. Comparison of g-valerolactone (GVL) yield and levulinic acid (LA)
Those differences do not correlate directly with the titania
2
conversion for platinum and ruthenium catalysts supported on TiO (R20)
phase composition, as in TiO (R10) the anatase content was
2
or carbon at reaction temperature of 308C or 708C.
higher than in TiO (R20), that is, 90% vs. 80%, and the GVL
2
Catalyst Reaction temperature [8C] GVL yield [%] LA conversion [%]
yield was not significantly different.
The highest GVL yields were obtained for Pt/TiO (A-2) in
2
Ru/TiO
Ru/TiO
Ru/C
2
2
30
70
30
70
30
70
30
70
62
95
50
75
13
23
0,8
0,8
64
99
61
93
19
27
18
17
which moderately sized platinum particles were observed
(Table 2), although some appeared in clusters, together with
small particles presumably located in the micropores.
Ru/C
Pt/TiO
Pt/TiO
Pt/C
2
Ruthenium catalysts behaved in a different way than their
platinum counterparts. At both reaction temperatures, the
highest yield was observed for anatase–rutile mixed phase
2
Pt/C
TiO (R10) and TiO (R20). Indeed, Ru catalyst supported by pure
1
g of levulinic acid, 0.3 g of catalyst, and 30 mL of water were combined
2
2
in a stainless steel autoclave from Berghof, equipped with a Teflon insert
allowing a reaction volume of 70 mL. The reactor was pressurized with
hydrogen to 50 bar and the temperature was maintained at 30–708C for
rutile TiO (R100) led also to high conversion and high yield,
2
even if slightly lower than on TiO (R10), with a GVL yield of
2
8
3% vs. 100% on TiO (R10) (Table 5). By contrast, largely lower
2
1
h.
performances were observed on Ru supported on pure ana-
tase TiO catalysts, with a moderate activity for Ru/TiO (A-2)—
2
2
Interestingly, a higher GVL yield was observed for platinum
in contrast with the case of Pt above—and a significant drop
in both conversion and yield for other ruthenium-on-anatase
catalysts, that is, Ru/TiO (A-4) and Ru/TiO (A-3).
and ruthenium catalysts supported on titania in comparison to
carbon-supported ones (Table 4).
2
2
To understand the reasons for this behavior, we examined as
supports for ruthenium and platinum particles the previously
presented different titania, varying mainly in terms of phase
compositions (rutile and/or anatase), specific surface areas, and
isoelectric points (Table 1).
The latter three pure anatase samples had a large surface
area with the presence of micropores and large “islands” of
ruthenium which were especially visible in the case of Ru/
TiO (A-3) and Ru/TiO (A-4) catalysts in which the highest micro-
2
2
porosity was observed. Therefore, pure anatase TiO sample
2
In the case of platinum catalysts, activities at the lower reac-
tion temperature (308C) were in general very low (a few per-
cent), verging on the borders of the measurement accuracy,
and any conclusions based on that could be misleading. There-
fore, only the results obtained at 708C are shown (Table 5).
With Pt supported on TiO (R20) and TiO (R10), similar GVL
with no microporosity, that is, TiO (A-1), was prepared by cal-
2
cining the anatase TiO (A-2). Surprisingly the activity shown by
2
the Ru/TiO (A-1) catalyst was much higher than that of the Ru/
2
TiO (A-2) analogue, reaching values close to those for Ru/
2
TiO (R10) and Ru/TiO (R20) catalysts (Table 5).
2
2
2
2
yields were achieved, with a slightly lower value for the latter
&
ChemSusChem 0000, 00, 0 – 0
www.chemsuschem.org
6
ꢀ 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
case, it was important to combine TEM and H or CO chemi-
Discussion
2
sorption, as the latter technique was able to reveal the pres-
ence of very small particles (2 nm) that we were unable to
image with TEM. This can be explained by the limited contrast
We investigated the influence of different properties of titania
support on the activity of platinum and ruthenium catalysts for
the hydrogenation of levulinic acid into g-valerolactone at low
temperatures in water. Generally, ruthenium catalysts showed
better activity, and this superior activity was recently explained
by a specific assistance of water for ketone hydrogenation in
for the metal particles on the TiO support, especially for Ru. In
2
pure anatase samples with the presence of microporosity and
larger surface area, TEM only imaged a very small number of
large Ru agglomerates, consistent with the previous discussion
on the weak adhesion of Ru on anatase and with previous lit-
[
22]
the case of Ru. However, the nature of titania exerts a strong
influence on the activity of both metal catalysts, but surprisingly
in a very different way. In addition to the type of titania crystal-
line phase, which is shown to be a crucial factor, not only the
support morphology but also its chemical properties and its in-
teraction with the metal are found to be of key importance.
[29]
erature. The discrepancy with the evidence of small Ru parti-
cles through H chemisorption led us to hypothesize that such
2
very small Ru particles were located in the micropores of the
high surface-area anatase samples. This was confirmed by hy-
drogen chemisorption performed on the anatase sample with
no microporosity and lower surface area that clearly revealed
the absence of any small Ru particles in the sample. The situa-
tion was somewhat different for Pt, since well defined particles
with a mean size of 4.5 nm are seen with TEM, one similarity
though being that chemisorption revealed a smaller average
size, and hence the presence of very small particles.
The influence of the phase of the TiO support (rutile or ana-
2
tase) on activity is indeed very different for Pt and Ru catalysts.
Pt catalysts are more active on the anatase phase. In contrast,
Ru is more active on the anatase–rutile mixed phase support,
and its activity can be dramatically decreased on pure anatase
support. Overall, the best Ru catalyst, that is, on TiO (R10),
2
gives a twice larger conversion that the best Pt catalyst (on
anatase).
If we now come back to the reactivity comparison, the best
activity was obtained for Ru on TiO (R10), which contains only
2
These differences of activity depending on the support can
be traced back to two key properties of the chosen phase: its
electronic properties, which govern the adhesion of the metal
particle and its dispersion on the oxide surface, and the oxide
morphology, which impacts the surface area and the porosity.
10% of rutile and shows no microporosity. Our analysis howev-
er clearly shows that in that case, the metal particles all gather
on rutile and come close to one to the other. Switching to
pure rutile TiO (R100) slightly decreases the Ru particle size
2
and enlarges the distance between particles. This results in
a slight decrease of activity, mostly seen at low temperature at
which the conversion remains partial. However the most dras-
tic change and loss of activity is seen on the anatase support
with high surface area and the presence of micropores. This is
explained by the strongly bimodal distribution of Ru particles,
with large agglomerates on one hand together with very small
particles in the micropores. One can hence conclude that small
particles of Ru are not active for levulinic acid hydrogenation,
and this could be explained by the more difficult reduction of
Mixed anatase and rutile samples, namely TiO (R20) and
2
TiO (R10), clearly illustrate the different properties of the inter-
2
face between the oxide and the metal for rutile and anatase in
the case of supported ruthenium catalysts. The microscopic
images showed that Ru particles strongly prefer to grow on
the surface of rutile grains and that no particle was imaged on
the anatase part of the support. Such segregation, however,
was not observed for Pt particles, which are randomly distrib-
uted on the two phases of the support.
[42]
Rutile and anatase TiO solids show different electronic prop-
small particles compared to large ones or by diffusion limita-
tions for the reactants and products in the micropores.
2
erties, with a larger band gap (3.2 eV vs. 3.0 eV) and a conduc-
[38]
tion band edge position higher in energy for anatase. This
There are two ways however to restore a high reactivity in
the case of anatase-supported Ru catalysts. The first one is to
calcine the anatase so that the microporosity is removed and
the surface area is lowered. In that case, small particles are not
[39]
results in a more difficult reduction process for anatase and
a much reduced number of O vacancies. These electronic dif-
ferences can explain the weaker adhesion of Ru on anatase by
a difficult electron transfer from the metal to the semiconduc-
tor or due to the absence of anchoring O defects. Pt shows
present anymore (H chemisorption) and the metal dispersion
2
on the anatase surface is improved, yielding medium-sized par-
ticles. Another attempt was made in order to achieve a homo-
geneous metal distribution on large surface-area anatase sam-
ples with microporosity, consisting of the modification of the
nanoparticle preparation method in order to avoid any final
temperature treatment. During impregnation, the fixation/dep-
[
40]
a d band energy center 0.8 eV lower than that of Ru so that
its d states lie significantly lower in energy and are less affect-
ed by the higher position of the conduction band in anatase.
This explains why Pt shows no strong preference between the
two titania phases.
Another explanation proposed in the literature is the similar-
ity of the crystal structure for rutile and ruthenium oxide,
which consequently favors the growth of RuO2 particles on
this surface. This hypothesis is also consistent with the differ-
osition of Ru(acac) occurs through the exchange of acetylacet-
3
[32,33]
onate ligands with hydroxyl groups of the support.
Their
decomposition and the formation of RuO occur during the
2
temperature treatment, so that a redispersion phenomenon
might occur, with consequent formation of large agglomerates.
A chemical reduction method enabled us to bypass the tem-
ent behavior of Pt since the orthorhombic (CaCl type) struc-
2
[
41]
ture is more stable that the rutile one for PtO2.
Pure anatase phases show however a different behavior
mainly because they present a different morphology. In that
perature treatment and the formation of RuO , so that
2
medium-sized particles with an average size of 3–4 nm could
ChemSusChem 0000, 00, 0 – 0
www.chemsuschem.org
7
ꢀ 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
be prepared on anatase TiO The resulting catalysts displayed
Ruthenium exhibits a stronger affinity to rutile titania, due
to more adequate electronic and crystal properties. Metal parti-
cles of optimum size for catalytic performance were found to
be formed on a support containing 10% rutile.
2.
a high activity, close to that recorded on Ru/TiO (R10) prepared
2
via the thermal reduction method. However, in the case of
TiO (R20), the activity after direct chemical reduction treatment
2
was lower, which suggests that, besides good dispersion,
strong interactions with the titania surface, which can be in-
duced only at the high temperature, are also beneficial.
Conclusions
The case of supported Pt particles is the opposite to that of
Ru. A moderate activity is obtained on mixed phase anatase–
rutile TiO (R20) and TiO (R10) catalysts, despite the correct
The influence of different types of titania supports for platinum
and ruthenium catalysts on the hydrogenation of levulinic acid
towards g-valerolactone was investigated in mild conditions in
water. Different phases of titania (rutile, anatase, mixed phase)
and different morphologies (surface area, microporosity) were
considered. A strong influence of the choice of the support on
the catalyst activity in the reaction was found. Surprisingly, the
influence of support was different for Ru and Pt catalysts.
2
2
metal dispersion on both rutile and anatase phases. The con-
version is doubled when Pt is deposited on the pure anatase
supports, although it remains significantly smaller than for the
best Ru catalyst. In contrast to Ru, as explained earlier, the ad-
hesion between anatase and Pt is strong enough to maintain
particles of reasonable size (4.5 nm) and to avoid the formation
of large agglomerates. Nevertheless, some clusters were ob-
served. Chemisorption shows that smaller particles are present,
but it is not possible to determine whether these, in contrast
to the Ru case, are active for GVL formation.
The most active catalyst is Ru on the mixed phase TiO (R10)
2
and TiO (R20) supports, consisting of 10% rutile and 90% ana-
2
tase, and 20% rutile and 80% anatase, respectively. It is inter-
esting to note that preliminary results highlighted that the
most active catalyst in LA hydrogenation, that is, Ru/TiO (R20),
2
When it comes to the selectivity, in the case of titania-sup-
ported materials, there were no other products observed (like
possible GVL decomposition products, for example, 1,4-penta-
nediol or 2-methyltetrahydrofuran). The reaction also does not
proceed further towards pentanoic acid, which could occur in
is also active in real conditions when cellulose is used as a feed-
stock (Supporting Information, SI1). Microscopy showed in
that case that the Ru particles are exclusively located on the
minority rutile crystallites. This was explained by a higher ad-
hesion of Ru on rutile in comparison to anatase. The activity of
Ru on pure rutile was globally similar to that observed fro the
rutile-rich catalyst. In contrast, pure anatase initially gave very
poorly active Ru catalysts, despite a much larger surface area.
The weak adhesion with the anatase support resulted in
strong sintering during the reduction process, and in the for-
mation of very large Ru particles and agglomerates. CO chemi-
sorption showed that small Ru particles were also present,
likely inside the micropores of the support. However these
small Ru particles were not active, presumably since they are
difficult to completely reduce.
[
43]
the presence of a strong acid.
The differences between yield and conversion visible for
carbon-supported materials are related to LA adsorption on
their surface, which was confirmed by the reaction with the
support. In a recent paper by B. Weckhuysen et al., it was also
mentioned that LA hydrogenation can proceed directly via 4-
hydroxypentanoic acid (HPA) towards pentanoic acid (via dehy-
dration and subsequent hydrogenation), but this path is forced
by the presence of a strong acid, which is not present with ti-
tania-supported materials.
In our case, the presence of metallic ruthenium sites seems
to be more important as the formation of HPA is most proba-
bly the rate-limiting step and the closure of the GVL ring pro-
Two methods were implemented to render the anatase-sup-
ported Ru catalyst active. First, the modification of the support
by calcination decreased the surface area and eliminated the
microporosity, and resulted in an improved dispersion of
medium-sized, active, Ru particles. Another method is to sub-
stitute the high temperature treatment that led to the forma-
[
22]
ceeds more easily.
Due to the fact that no other reaction products were ob-
served in any case, it seems that the role of the titania support
in the reaction pathway is limited to its interactions with ruthe-
nium and influencing the hydrogenation active sites on its sur-
face.
tion of RuO and subsequent sintering of the supported phase,
2
by a direct chemical reduction, avoiding the excessive mobility
of Ru on anatase, and thus maintaining a good dispersion.
The behavior of Pt is different, as it also gives a good adhe-
sion on anatase, and provides particles on both the rutile and
the anatase phase for mixed-phase supports. As a result, the
behavior of the platinum catalyst is more classical with an im-
proved activity on the larger surface-area anatase support. The
activity however remained significantly smaller than for the Ru
catalyst.
A common feature is that metal nanoparticles strongly inter-
acting with the support seem to be beneficial for the reac-
[
6]
tion. The reduction of the carbonyl group may even take
place at the metal–support interface where titania Lewis acid
sites could coordinate the oxygen of the ketones group, weak-
ening the C=O bond and rendering it more susceptible to hy-
drogenation. This synergistic effect is known to be especially
strong when a catalyst is activated by high-temperature reduc-
tion to form partially reduced titania species, which is more
likely on rutile than on anatase.
The results allow a better understanding of the role of the
support nature for Ru and Pt catalysts. They illustrate how the
different electronic and surface properties of rutile and anatase
can strongly affect the morphology and the activity of the cat-
alysts. Forthcoming works will investigate how the change
from model reaction systems to more complex systems,
The above seems to be connected with our observation that
the conversion of levulinic acid is strongly influenced by the
choice of the TiO support.
2
&
ChemSusChem 0000, 00, 0 – 0
www.chemsuschem.org
8
ꢀ 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
namely using cellulose or sugars as reactants and formic acid
XRD measurements were carried out on a D8 Advance Bruker dif-
fractometer in a q/q mode and using the Ka1 radiation of Cu at
as hydrogen source, impacts the catalytic behavior of TiO -sup-
2
1
.5406 ꢁ. The mean TiO crystallite size, that is, the average size of
2
ported Pt and Ru catalysts. In addition, TiO is an important
2
the coherent diffracting domains, was determined from the Scher-
rer equation with the usual assumption of spherical crystallites.
The surface area and porosimetry measurements were carried out
support in catalysis, and the results could also provide helpful
insights for other reactions, like the Fischer–Tropsch reaction in
the case of Ru catalysts.
on a ASAP2010 Micromeritics using N as adsorbant at ꢀ1968C,
2
with a prior outgassing at 2008C for 3 h in order to desorb the im-
purities or moisture. The Brunauer–Emmett–Teller (BET) specific
Experimental Section
surface area was calculated from the N adsorption isotherm; the
2
micropore surface area was derived using the t-plot method.
Isoelectrical point (IEP) measurements were carried out on a Mal-
vern ZetaSizer with automatic titration.
TiO supports and catalysts preparation
2
Several TiO supports were used: anatase–rutile mixed phase Aer-
2
TEM was employed to measure particle sizes with TEM JEOL 2010
and TEM JEOL 2100 microscope operating at a 200 kV equipped
with an energy dispersive X-ray detector (EDX).
For the platinum catalyst, the hydrogen chemisorption procedure
was as follows: The samples were calcined at 2008C for 1 h, cooled
down to room temperature and hydrogen was chemisorbed at
oxide TiO P25 and P90 were delivered by Evonik-Degussa (Germa-
2
ny), and further denoted as TiO (R20) and TiO (R10), respectively,
2
2
depending on the content of rutile phase within the sample. Hom-
bikat UV100, CristalACTIV PC500 and ST-01 supports consisting in
pure anatase TiO2 were delivered by Sachtleben Chemie GmbH
(
(
Germany), Millenium-Crystal (France), and Ishihara Sangyo Kaisha
ISK, Japan), and further denoted as TiO (A-3), TiO (A-4) and TiO (A-
2
008C for 2 h. The samples were subsequently cooled down in the
2
2
2
flow of argon and subjected to pulses of oxygen in order to mea-
sure the amount of chemisorbed hydrogen. For the ruthenium cat-
alyst, hydrogen chemisorption studies were carried out in the Mi-
cromeritics ASAP 2020 system.
2
), respectively, TiO (A-1) was obtained by calcination of TiO (A-2)
2 2
at 5008C in air for 3 h. 100% anatase TiO samples were labelled
from A-1 to A-4 in increasing order regarding the specific surface
area.
2
In each analysis, 1 g of the reduced TiO or the Ru/TiO sample was
The synthesis of rutile TiO , further denoted as R100, was carried
2
2
2
used. The sample was first reduced with H at 2008C for 1 h. After
out as detailed in [44], and consisted first in a sol–gel synthesis of
an anatase hydrosol by the hydrolysis of titanium isopropoxide(Al-
drich, 97% purity) in a hydrochloric acid aqueous solution (HCl,
Sigma–Aldrich, puriss p.a. ACS reagent, ꢁ37%) in the presence of
polyethylenglycol (PEG-4000) porogen. The anatase hydrosol was
then aged for 48 h at room temperature, and the resulting mixture
was dried first at room temperature and then overnight at 1008C,
before finally calcined at 5008C in flowing air.
The main description of the synthesized and commercial TiO ma-
terials is given in Table 1. C-AG5 carbon support with a high specif-
ic surface area of 960 m g was supplied by HPSDD, Poland.
Platinum and ruthenium catalysts with 1% loading were prepared
by incipient wet impregnation from Pt(acac)2 and Ru(acac)3
2
reduction, the sample was evacuated at 2008C for 60 min to
remove any residual H and cooled to the desired temperature for
2
analyses. H adsorption measurements were taken at 758C. To de-
2
termine the average particle sizes of supported metals from H2
chemisorption, monolayer coverage (complete surface coverage)
of H on metals was used. In the case of samples prepared by
chemical reduction, in order to avoid any temperature treatment,
samples were evacuated at 1208C for 1 h and then CO adsorption
was taken at 758C in the Micromeritics ASAP 2020 system, as hy-
drogen chemisorption in those conditions could give misleading
results.
2
2
ꢀ1
(
Sigma–Aldrich, 97%) methanolic solutions respectively on differ-
Catalytic tests
ent TiO . Catalysts were calcined in air at 2008C and reduced at the
2
same temperature for 1 h in hydrogen. Ru/TiO catalysts were also
2
In a typical levulinic acid (LA) hydrogenation experiment, 1 g of
levulinic acid, 0.3 g of a catalyst and 30 mL of water were com-
bined in a stainless steel autoclave from Berghof, equipped with
a Teflon insert, allowing a reaction volume of 45 mL. The reactor
was pressurized with hydrogen to 50 bar and the temperature was
maintained at 30–708C for 1 h. After the end of the reaction, the
reactor was cooled down, the remaining pressure was released
and the reaction mixture was centrifuged to separate the solid cat-
alyst and the product solution. The products were analyzed by
HPLC equipped with RID detector and Rezex ROA column, 0.005n
H SO was used as an eluent. The liquid after reaction was ana-
prepared by chemical reduction of ruthenium(III) nitrosyl nitrate(Al-
drich, 99.99%) using NaBH4 (Sigma-Aldrich 99%) as a reducing
agent. Firstly the aqueous titania suspension in the solution of
ruthenium(III) nitrosyl nitrate was stirred for 1 h, then a threefold
excess of sodium borohydride was added and reduction was car-
ried out for 15 min. The catalyst was then washed with deionized
water and dried at 1208C for 2 h.
Characterization techniques
2
4
lyzed by inductively coupled plasma mass spectrometry (ICP-MS)
in order to check whether leaching of active phase was taking
place. In all cases, it was proved that catalysts showed good stabili-
ty during reaction conditions.
TPR was performed on AMI1 system (Altamira Instruments)
equipped with a thermal conductivity detector (TCD) and was
used for examining the catalyst reducibility. The catalysts were sub-
mitted to H /Ar (5:95 v/v) flow at a weight hourly space velocity
2
ꢀ1
ꢀ1
(
WHSV) of 0.161 h and using a heating rate of 108Cmin . Before
the measurement samples were initially treated with 2 vol% O2
and 98 vol% Ar for 30 min at 2008C. Secondary ion mass spectra
were recorded with a time-of-flight secondary ion mass spectrome-
Acknowledgements
+
The authors gratefully acknowledge that this work was financial-
ly supported by a grant from the National Center of Science
(NCN) in Krakow (Poland) (NN 209 75 83 40). P.S. acknowledges
the support of the Polonium project “One-pot GVL production
ter (TOF-SIMS IV) manufactured by ION-TOF GmbH, Germany. Bi3
was used as primary ion source. The analyzed area of sample sur-
face was 500 mmꢂ500 mm. During analysis, a pulsed low-energy
electron flood gun was used for charge neutralization.
ChemSusChem 0000, 00, 0 – 0
www.chemsuschem.org
9
ꢀ 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
from sugars”. Mimoun Aouine and Laurence Burel from IRCELYON
CNRS) are acknowledged for TEM measurements.
[20] P. P. Upare, J.-M. Lee, D. W. Hwang, S. B. Halligudi, Y. K. Hwan, J.-S.
Chang, J. Ind. Eng. Chem. 2011, 17, 287.
(
[
[
21] Z. P. Yan, L. Lin, S. Liu, Energy Fuels 2009, 23, 3853.
22] C. Michel, J. Zaffran, A. M. Ruppert, J. Matras-Michalska, M. Jedrzejczyk,
J. Grams, P. Sautet, Chem. Commun. 2014, 50, 12450–12453.
23] D. J. Braden, C. A. Henao, J. Heltzel, C. T. Maravelias, J. A. Dumesic, Green
Chem. 2011, 13, 1755.
Keywords: biomass · levulinic acid · supported catalysts ·
titania · g-valerolactone
[
[
[
24] L. Corbel-Demailly, B.-K. Ly, D.-P. Minh, B. Tapin, C. Especel, F. Epron, A.
Cabiac, E. Guillon, M. Besson, C. Pinel, ChemSusChem 2013, 6, 2388.
25] W. Luo, U. Deka, A. M. Beale, E. R. H. van Eck, P. C. A. Bruijnincx, B. M.
Weckhuysen, J. Catal. 2013, 301, 175.
[
[
[
[
[
[
[
1] J. Q. Bond, D. M. Alonso, D. Wang, R. M. West, J. A. Dumesic, Science
010, 327, 1110.
2] C.-H. Zhou, X. Xia, Ch.-X. Lin, D.-S. Tong, J. Beltramini, Chem. Soc. Rev.
011, 40, 5588.
3] J. N. Chheda, G. W. Huber, J. A. Dumesic, Angew. Chem. Int. Ed. 2007, 46,
164; Angew. Chem. 2007, 119, 7298.
4] E. L. Kunkes, D. A. Simonetti, R. M. West, J. C. Serrano-Ruiz, C. A. Gartner,
J. A. Dumesic, Science 2008, 322, 417.
5] Y. Romꢃn-Leshkov, C. J. Barrett, Z. Y. Liu, J. A. Dumesic, Nature 2007, 447,
2
[
[
[
[
26] M. G. Al-Shaal, W. R. H. Wright, R. Palkovits, Green Chem. 2012, 14, 1260.
27] O. A. Abdelrahman, A. Heyden, J. Q. Bond, ACS Catalysis 2014, 4, 1171.
28] Y. Gong, L. Lin, Z. Yan, Bioresources 2011, 6, 686.
29] P. Panagiotopoulou, A. Christodoulakis, D. I. Kondarides, S. Boghosian, J.
Catal. 2006, 240, 114.
2
7
[
[
[
30] J. M. G. Carballo, E. Finocchio, S. Garcꢄa, S. Rojas, M. Ojeda, G. Busca,
J. L. G. Fierro, Catal. Sci. Technol. 2011, 1, 1013.
9
82.
6] A. M. Ruppert, K. Weinberg, R. Palkovits, Angew. Chem. Int. Ed. 2012, 51,
564; Angew. Chem. 2012, 124, 2614.
31] The planes were assigned according to the ¼
6
21–1276 and ¼
6
21–1272
JCPDS cards.
2
32] Y. V. Plyuto, I. V. Babich, L. F. Sharanda, A. M. de Wit, J. C. Mol, Thermo-
chim. Acta 1999, 335, 87.
7] J. Q. Bond, A. A. Upadhye, H. Olcay, G. A. Tompsett, J. Jae, R. Xing, D. M.
Alonso, D. Wang, T. Zhang, R. Kumar, A. Foster, S. M. Sen, C. T. Marave-
lias, R. Malina, S. R. H. Barrett, R. Lobo, C. E. Wyman, J. A. Dumesic, G. W.
Huber, Energy Environ. Sci. 2014, 7, 1500.
[
[
33] B. Coq, A. Bittar, F. Figueras, Appl. Catal. 1990, 59, 103.
34] K. Hadjiivanov, J.-C. Lavalley, J. Lamotte, F. Mauge, J. Saint-Just, M. Che,
J. Catal. 1998, 176, 415.
[
[
8] D. M. Alonso, J. Q. Bond, J. A. Dumesic, Green Chem. 2010, 12, 1493.
9] S. Van de Vyver, J. Geboers, P. A. Jacobs, B. F. Sels, ChemCatChem 2011,
[
[
[
35] L. Li, L. Qu, J. Cheng, J. Li, Z. Hao, Appl. Catal. B 2009, 88, 224.
36] A. M. Ruppert, T. Paryjczak, Appl. Catal. A 2007, 320, 80.
37] N. Kamiuchi, T. Mitsui, H. Muroyama, T. Matsui, R. Kikuchi, K. Eguchi, L.
Corbel-Demailly, B.-K. Ly, D.-P. Minh, B. Tapin, C. Especel, F. Epron, A.
Cabiac, E. Guillon, M. Besson, Appl. Catal. B 2010, 97, 120.
38] L. Kavan, M. Grꢅtzel, S. E. Gilbert, C. Klemenz, H. J. Scheel, J. Am. Chem.
Soc. 1996, 118, 6716.
3, 82.
[
[
[
10] P. Gallezot, Chem. Soc. Rev. 2012, 41, 1538.
11] M. Besson, P. Gallezot, C. Pinel, Chem. Rev. 2014, 114, 1827.
12] I. T. Horvꢃth, H. Mehdi, V. Fꢃbos, L. Boda, L. T. Mika, Green Chem. 2008,
[
[
10, 238.
[
[
13] D. M. Alonso, S. G. Wettstein, J. A. Dumesic, Green Chem. 2013, 15, 584.
14] J.-P. Lange, R. Price, P. M. Ayoub, J. Louis, L. Petrus, L. Clarke, H. Gosse-
link, Angew. Chem. Int. Ed. 2010, 49, 4479; Angew. Chem. 2010, 122,
39] A. Bouzoubaa, A. Markovits, M. Calatayud, C. Minot, Surf. Sci. 2005, 583,
107.
[
[
40] B. Hammer, J. K. Nørskov, Adv. Catal. 2000, 45, 71.
41] R. K. Nomiyama, M. J. Piotrowski, J. L. F. Da Silva, Phys. Rev. B 2011, 84,
4581.
[15] P. A. Son, S. Nishimura, K. Ebitani, RSC Adv. 2014, 4, 10525.
100101.
[16] L. Deng, J. Li, D. M. Lai, Y. Fu, Q. X. Guo, Angew. Chem. Int. Ed. 2009, 48,
[
42] H. Wang, Y. Wang, Z. Zhu, A. Sapi, K. An, G. Kennedy, W. D. Michalak,
G. A. Somorjai, Nano Lett. 2013, 13, 2976.
6
529; Angew. Chem. 2009, 121, 6651.
17] A. M. R. Galletti, C. Antonetti, V. De Luise, M. Martinelli, Green Chem.
012, 14, 688.
18] A. M. Raspolli Galletti, C. Antonetti, E. Ribechini, M. P. Colombini, N. N. O.
Di Nasso, E. Bonari, Appl. Energy 2013, 102, 157.
[
[
[
[
[
43] W. Luo, P. C. A. Bruijnincx, B. M. Weckhuysen, J. Catal. 2014, 320, 33.
44] F. Bosc, A. Ayral, N. Keller, V. Keller, Appl. Catal. B 2007, 69, 133.
2
19] M. Selva, M. Gottardo, A. Perosa, ACS Sustainable Chem. Eng. 2013, 1,
Received: December 8, 2014
180.
Published online on && &&, 0000
&
ChemSusChem 0000, 00, 0 – 0
www.chemsuschem.org
10
ꢀ 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

FULL PAPERS
What lies beneath: A number of differ-
ent titanias are implemented as sup-
ports for Ru and Pt catalysts. It is shown
that various morphologies and phases
of titania influence the physicochemical
properties of the studied catalysts in dif-
ferent ways. The best catalytic per-
formance in levulinic acid hydrogena-
tion is achieved in the case of mixed
rutile/anatase-supported ruthenium
catalyst.
A. M. Ruppert,* J. Grams, M. J e˛ drzejczyk ,
J. Matras-Michalska, N. Keller, K. Ostojska,
P. Sautet
&& – &&
Titania-Supported Catalysts for
Levulinic Acid Hydrogenation:
Influence of Support and its Impact on
g-Valerolactone Yield
ChemSusChem 0000, 00, 0 – 0
www.chemsuschem.org
11
ꢀ 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ