Improvement on the catalytic performance of Mg-Zr mixed oxides for furfural-acetone aldol condensation by supporting on mesoporous carbons

Article abstract of DOI:10.1002/cssc.201200710

A new procedure for improving the performance of the most common catalysts used in aqueous-phase aldol condensation (Mg-Zr mixed oxides) reactions is presented. This reaction is of interest for upgrading carbohydrate feedstocks. The procedure involves supporting Mg-Zr oxides on non-microporous carbonaceous materials, such as carbon nanofibers (CNFs) or high-surface-area graphites (HSAGs), using either incipient wetness or coprecipitation procedures. The use of HSAGs together with the coprecipitation method provides the best performance. Results obtained for the cross-condensation of acetone and furfural at 323K reveal that the catalyst performance is greatly improved compared to the bulk oxides (96.5 % conversion vs. 81.4 % with the bulk oxide; 87.8 % selectivity for C₁₃ and C₈ adducts vs. 76.2 % with the bulk oxide). This difference is even more prominent in terms of rates per catalytically active basic site (four and seven times greater for C₈ and C₁₃ adducts, respectively). The improved performance is explained in terms of a more appropriate basic site distribution and by greater interaction of the reactants with the carbon surface. In addition, deactivation behavior of the catalyst is improved by tuning the morphology of the carbonaceous support. An important enhancement of the catalytic stability can be obtained selecting a HSAG with an appropriate pore diameter. With HSAG100 the activity decreased by less than 20 % between successive reaction cycles and the selectivity for the condensation products remained almost unaltered. The decrease is greater than 80 % for the bulk oxides tested at these conditions, with important increases in the selectivity for by-product formation. Copyright

Full text of DOI:10.1002/cssc.201200710

CHEMSUSCHEM
FULL PAPERS
DOI: 10.1002/cssc.201200710
Improvement on the Catalytic Performance of Mg–Zr
Mixed Oxides for Furfural–Acetone Aldol Condensation by
Supporting on Mesoporous Carbons
Laura Faba, Eva Dꢀaz, and Salvador OrdꢁÇez[a]
A new procedure for improving the performance of the most
common catalysts used in aqueous-phase aldol condensation
(Mg–Zr mixed oxides) reactions is presented. This reaction is of
interest for upgrading carbohydrate feedstocks. The procedure
involves supporting Mg–Zr oxides on non-microporous carbo-
naceous materials, such as carbon nanofibers (CNFs) or high-
surface-area graphites (HSAGs), using either incipient wetness
or coprecipitation procedures. The use of HSAGs together with
the coprecipitation method provides the best performance. Re-
sults obtained for the cross-condensation of acetone and furfu-
ral at 323 K reveal that the catalyst performance is greatly im-
proved compared to the bulk oxides (96.5% conversion vs.
81.4% with the bulk oxide; 87.8% selectivity for C13 and C8 ad-
ducts vs. 76.2% with the bulk oxide). This difference is even
more prominent in terms of rates per catalytically active basic
site (four and seven times greater for C8 and C13 adducts, re-
spectively). The improved performance is explained in terms of
a more appropriate basic site distribution and by greater inter-
action of the reactants with the carbon surface. In addition, de-
activation behavior of the catalyst is improved by tuning the
morphology of the carbonaceous support. An important en-
hancement of the catalytic stability can be obtained selecting
a HSAG with an appropriate pore diameter. With HSAG100 the
activity decreased by less than 20% between successive reac-
tion cycles and the selectivity for the condensation products
remained almost unaltered. The decrease is greater than 80%
for the bulk oxides tested at these conditions, with important
increases in the selectivity for by-product formation.
Introduction
Processes based on using biomass as a raw material for the
synthesis of fuels and chemicals are considered among the
most promising technologies for the mitigation of fossil-fuel
dependency.[1,2] The main difficulty for transforming biomass
into biofuels is that it relies on the differences in their molecu-
lar composition: biofuels are hydrophobic whereas carbohy-
drates are hydrophilic and have high oxygen/carbon ratios.[3]
Therefore, new technologies are required for obtaining bio-
fuels from this raw material.[3–5] The main alternative consists of
low-temperature acid hydrolysis of wooden feedstocks,[6] a pro-
cess in which the formation of sugar derivatives (furfural, hy-
droxymethylfurfural, etc.) is catalyzed by homogeneous or het-
erogeneous acids.[7] These products can be used as biofuels
precursors; however, their short carbon-chain lengths prevents
their use as diesel fuel precursor, and their non-branched char-
acter leads (upon hydrogenation) to poor quality gasoline
The formation of CÀC bonds between precursor molecules
leads to C8–C15 adducts, which are platform molecules for
diesel-fuel synthesis. One approach for obtaining these larger
n-alkanes involves the aldol condensation of aldehydes or ke-
tones to form C8 and C13 adducts, which can yield the desired
alkanes after hydrogenation and deoxygenation reactions,[8–10]
as depicted in Scheme 1.
Scheme 1. Formation of C8 and C13 alkenes by aldol condensation of furfural
and acetone, followed by hydrogenation-dehydration reactions.
Different heterogeneous catalysts have been studied either
in the aqueous or in the gas phase,[11–13] showing that the con-
centration of basic sites and the distribution of basicity
strength are key parameters in CÀC bond formation reac-
tions.[8] We recently reported the study of furfural and acetone
aldol cross-condensation using Mg–Zr, Mg–Al, and Ca–Zr cata-
lysts. High selectivities for both biofuel precursors were ob-
tained and the Mg–Zr catalyst showed the highest activity
(61.5% of C13 final selectivity) with more than 85% of furfural
conversion and low reactants losses from side reactions or
solid-deposit formation.[14] Activity results were compared to
the surface properties of the tested mixed oxides, showing
that the formation of C8 and C13 adducts are catalyzed by
medium-strength basic sites, whereas the most basic sites lead
to the formation of heavy side-products. Analysis of the coke
deposits revealed that there was a high contribution of the C13
adduct, its low solubility favoring deposition onto the surface,
[a] L. Faba, Dr. E. Dꢀaz, Prof. S. OrdꢁÇez
Department of Chemical and Environmental Engineering
University of Oviedo, C/Juliꢂn Claverꢀa
s/n - 33006 Oviedo Asturias (Spain)
Fax: (+34)985103434
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2013, 6, 463 – 473 463

CHEMSUSCHEM
FULL PAPERS
www.chemsuschem.org
blocking the basic sites, and limiting the reaction evolution.
This disadvantage can be solved by integrating condensation
and partial hydrogenation into one single step. For this pur-
pose, the mixed oxide must be impregnated with palladium.
Bi-functional catalysts (Pd/Mg–Zr, Pd/Mg–Al and Pd/Ca–Zr)
were studied and the best results were reached with Pd/Mg–Zr
in the presence of H2.[15]
areas. This study includes the role of the preparation proce-
dure, the correlation of activity trends with the physicochemi-
cal characterization of the materials, and the determination of
the deactivation behavior of the catalysts.
Results and Discussion
The reaction was catalyzed by medium-strength basic sites;
therefore, strategies for increasing the concentration of basic
sites led to higher condensation rates. Dispersion of the mixed
oxides onto an inert support led to the presence of structural
defects creating new basic sites, as reported by Winter et al.
for the dispersion of hydrotalcites onto carbon nanofibers.[16,17]
However, there are several points to note: firstly, most of the
inorganic supports have a distribution of acidic/basic sites that
can interfere with the corresponding active phase and catalyze
undesired side reactions. Secondly, activated carbons are not
appropriate because of their microporous character, hindering
both the homogeneous dispersion of the active phase and the
diffusion of the reactants in the porous structure. Thus, graph-
itic supports, such as carbon nanotubes (CNTs), have good
properties for the dispersion of active phases without altering
the chemical behavior or introducing mass-transfer effects.[18]
The main disadvantages of these materials are their low sur-
Characterization
The main morphological properties of the studied catalysts are
shown in Table 1. The adsorption–desorption isotherms ob-
tained in all cases are type IV (BDDT Classification) and show
the characteristic pattern for mesoporous solids. The data ob-
tained were consistent with that expected, showing an in-
crease in the surface area for those materials in which the orig-
inal surface area was low (CNFs and HSAG100), which was
caused by a swelling of the original structures upon dispersion
of the active phase.[21] Supports with the highest surface area
(HSAG300 and HSAG500) presented a decrease in their surface
area after impregnation with the mixed oxides, because of the
partial blockage of the pore structure. Supported catalysts
showed an increase in the pore diameter, which is related to
Mg–Zr deposited onto the surface blocking the smaller pores
of the carbon material. Pore volumes show slight variations, at-
tributed to modifications of the morphology of the supports
during the preparation process.
face area (same order of magnitude as the geometric area[19]
)
and the low concentration of anchoring sites for active-phase
dispersion.
Catalyst crystallography was evaluated by XRD analysis; dif-
fraction patterns are given in Figure 1. These patterns are char-
acterized by the presence of a peak at 2q=268 (not shown),
characteristic of a highly graphitic carbon structure (002), with
similar intensity in all cases. A crystallinity loss (peak widening)
is observed when the mixed oxide is supported on the carbo-
naceous material, in good agreement with its decrease in crys-
tal size. This was also seen in other works with similar carbon
materials.[13] Comparing the diffractograms for the HSAG-based
catalysts, Mg–Zr/HSAG300* (the only catalyst prepared by
using incipient wetness impregnation) showed the maximum
loss, whereas catalysts prepared by coprecipitation retained
the crystallographic structure of the parent oxide. The CNF-
based catalyst loses almost all of the crystallinity of the parent
mixed oxide, although the catalyst was prepared by coprecipi-
tation. In general, there was not significant displacement be-
Other carbon materials such as high-surface-area graphites
(HSAGs) can be considered as an alternative support. HSAGs
are prepared by the mechanical modification of synthetic
graphites. This leads to disordered graphitic layers, resulting in
marked mesoporous character and moderate surface areas
(100–500 m2 gÀ1). To the best of our knowledge, these materi-
als have not been previously used as supports for this type of
catalyst, despite their interesting properties. Furthermore, the
unsaturated valences at the edges of the graphitic layers lead
to the formation of oxygenated functional groups, which act
as anchoring sites during the active-phase addition.[20]
We report in this article a study of the improvement of the
catalytic performance of a Mg–Zr mixed oxide for the furfural–
acetone aldol condensation by supporting this active phase on
carbon nanofibers (CNFs) and HSAGs with different surface
Table 1. Morphological (nitrogen physisorption) and acid–base properties of the catalysts and supports used in this work.
Catalyst
Morphological properties
Basic sites [mmolgÀ1], (CO2-TPD, [K])
Acidic sites [mmolgÀ1], (NH3-TPD, [K])
S [m2 gÀ1
]
Dp [ꢃ]
Vp [cm3 gÀ1
]
bicarb.
bident.
monodent.
weak
medium
strong
Mg-Zr
HSAG100
78
107
110
321
270
243
580
337
32
342
107
167
58
83
80
75
79
160
184
0.8
–
120 (427)
–
5.5 (513)
–
13 (585)
–
32 (776)
–
19 (716)
19 (841)
–
62 (425)
239 (550)
158 (863)
0.26
0.49
0.47
0.47
0.39
0.75
0.61
0.17
0.20
0.02(377)
11 (400)
0.01 (398)
19 (386)
10 (394)
–
MgZr/HSAG100
HSAG300
MgZr/HSAG300*
MgZr/HSAG300
HSAG500
MgZr/HSAG500
CNF
MgZr/CNF
117 (411)
–
85 (943)
–
448 (410)
984 (389)
–
–
–
17 (506)
0.1 (432)
13 (453)
85 (927)
10 (338)
20 (730)
489 (425)
12 (475)
–
24 (923)
41
2.1 (369)
1.6 (466)
38 (702)
197 (595)
114 (984)
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2013, 6, 463 – 473 464

CHEMSUSCHEM
FULL PAPERS
www.chemsuschem.org
tween the phases analyzed in the bulk material and those ob-
tained in the supported materials; however, there was a slight
shift to lower diffraction angles observed. The main peaks re-
lated to the mixed oxide are those associated with periclase
(MgO), located at 2q=378, 438, 628, and 788.[22] The peak locat-
ed at 2q=438 could also be associated to (101) graphene
layers. However, the proportionality between peaks at 2q=
438, 628, and 788 was the same as in the bulk material; there-
fore this peak was also attributed to periclase. Crystalline
phases related to tetragonal ZrO2 were detected with only
small peaks at 2q=308 in Mg–Zr/HSAG100 and Mg–Zr/
HSAG500; therefore, this metal was more affected by the im-
pregnation method and is present mainly as an amorphous
phase. No phases related to the mixed oxide were found. This
was previously observed in other works with Mg–Zr in which
only periclase and low-intensity peaks of tetragonal zirconia
are observed.[23] Only in the work of Sꢄdaba et al.[24] is the pres-
ence of a mixed oxide reported. However, these authors state
that the peaks are only observed at Mg/Zr ratios lower than
2.5, whereas this ratio in our study was greater than four.
Mixed oxide particle sizes and dispersions were studied by
TEM and the micrographs are shown in Figure 2. In micro-
graphs corresponding to Mg–Zr/HSAG100, Mg–Zr/HSAG300,
and Mg–Zr/HSAG500, the particles present a regular hemi-
spherical geometry and the crystallite sizes have been deter-
mined by counting 100 particles of different micrographs (his-
tograms are also depicted in Figure 2). The average crystallite
diameters are shown in Table 2. Micrographs of Mg–Zr/
HSAG300* and Mg–Zr/CNF display irregular deposits of mixed
oxides on the supports and the crystallite size could not be ac-
Figure 1. XRD diffractograms of the fresh catalysts used: (a) Mg–Zr; (b) Mg–
Zr/HSAG100; (c) Mg–Zr/HSAG300*; (d) Mg–Zr/HSAG300; (e) Mg–Zr/HSAG500;
(f) Mg–Zr/CNF. Marked peaks correspond to periclase phases.
Figure 2. TEM micrographs: (a) Mg–Zr/HSAG100; (b) Mg–Zr/HSAG300*; (c) Mg–Zr/HSAG300; (d) Mg–Zr/HSAG500, and (e) Mg–Zr/CNF.
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2013, 6, 463 – 473 465

CHEMSUSCHEM
FULL PAPERS
www.chemsuschem.org
value is justified when consider-
ing the low solubility of the zir-
conium precursor that limits the
amount of mixed oxide that can
be synthesized by using the dry-
impregnation method. For the
nanofibers, in addition to the
low concentration of both spe-
cies, the Mg/Zr ratio is complete-
ly different to the expected ratio,
suggesting a preferential adsorp-
tion of the zirconium precursor
onto the carbon surface. This is
Table 2. Particle size (TEM), XPS O 1s binding energy (B.E.), surface composition, and bulk metal loadings (ICP–
OES) for the catalysts used in this work.
Catalyst
TEM Dc
[nm]
XPS O 1s
B.E. [eV]
XPS superficial composition [%]
ICP–OES results
Mg/Zr
Mg
Zr
Mg/Zr
Mg-Zr
–
12
–
11
8
529.0
530.1
530.6
529.6
530.1
532.0
41.9
37.8
16.5
38.2
42.0
14.0
3.9
3.1
1.0
3.6
3.0
8.0
10.7
12.7
16.5
10.6
14.0
1.8
4.4
5.1
4.0
4.5
5.3
0.34
MgZr/HSAG100
MgZr/HSAG300*
MgZr/HSAG300
MgZr/HSAG500
MgZr/CNF
–
curately estimated. Mg–Zr/CNF micrographs were similar to
those reported by Winter et al;[16] these results were consistent
with XRD results, showing that incipient wetness impregnation
is not a good method for supporting these oxides. The results
also showed that the use of HSAGs as supports led to more
crystalline and dispersed particles in the active phase com-
pared to the use of CNFs.
not only related to the concentration of anchoring sites (ex-
pected to be lower in the case of CNFs), but also to the elec-
trostatic interactions of the ionic precursors with the surface.
Thus, the zero-point charge (ZPC) for the HSAG is between 3
and 3.5, whereas it is higher than 5 for the CNFs. As the pH
value of the precursor solution is approximately 6.4 and ZrO2+
is a more voluminous cation than Mg2+, this led to a higher Zr
surface concentration observed for CNFs. Comparison of XPS
and ICP–OES results showed a significant surface increase of
the Mg concentration, which can justify the subsequent modi-
fication of the basic properties.
SEM was used to analyze the surface morphology and to
confirm the homogeneous dispersion of Mg and Zr over all of
the carbon surfaces. The micrographs obtained (not shown
here) revealed that the supported materials maintained the
morphological aspect of the pattern supports, as previously re-
ported.[19] The CNF material revealed a fibrous morphology
with the mixed oxide particles deposited over it, whereas the
HSAG supports showed a granular and spongy appearance,
with the mixed oxides more homogeneously dispersed. The
surface composition was analyzed by using energy dispersive
X-ray (EDX) spectroscopy. Five different and random points
were chosen to evaluate the atomic disposition of Mg and Zr.
It was concluded that in all of the supported materials, Mg
and Zr were present as mixed oxides, without segregation be-
tween both oxides. Consequently, the supported materials
have the same active phases as the bulk material and their cat-
alytic activities could be compared with the unsupported
mixed oxide.
XPS analysis was also used to qualitatively evaluate the ba-
sicity of the materials, as the position of the O1s peak provides
insight into the basicity of crystalline solids; higher binding en-
ergies suggest lower basic strengths.[25] The values of the O1s
binding energy are summarized in Table 2. Comparing these
results with the value obtained for Mg–Zr (529.0 eV), the basic
sites of the bulk materials are stronger than the corresponding
supported catalysts; Mg–Zr/HSAG300 is the most similar in
terms of the global basicity. Concerning the deconvolution of
the O1s peaks (not presented here), only the bulk material
presents different types of contributions, with a very intense
peak at 529.1 eV from the O2À species in MgO[26] and a less in-
tense signal at 532.2 eV, corresponding to the presence of hy-
droxyl groups on the surface of MgO and/or ZrO2.[27] The sup-
ported materials only show one peak, corresponding to the
strongest groups. These results were checked by using CO2-
temperature-programmed desorption (CO2–TPD) analysis.
CO2–TPD results are detailed in Table 1, expressed as con-
centration and strength distribution of the basic centers on
the surface. The original graphites and nanofibers were treated
by using the same temperature program as that of the sup-
ported materials, with a maximum over 850 K. With this treat-
ment, the CO2 signal from carbon pyrolysis could be discarded.
The analysis of the bulk mixed oxide is also shown. Mg–Zr
showed the highest concentration of basic sites
(133.4 mmolgÀ1), distributed as both bidentate and monoden-
tate sites. All of the supported catalysts showed, in addition to
bidentate and monodentate centers, weaker sites assimilated
to bicarbonates. Analyzing the global results, all of the sup-
ported materials had similar basicity, with values between 38
(Mg–Zr/HSAG300*) and 48 mmolgÀ1 (Mg–Zr/HSAG100). Howev-
er, the strength distribution obtained for Mg–Zr/CNF was very
different, that is mainly monodentate centers, whereas the
Chemical analysis of the surface composition was performed
by using X-ray photoelectron spectroscopy (XPS) and the bulk
composition was analyzed by using inductively coupled
plasma optical emission spectroscopy (ICP–OES), after leaching
of the samples in a hydrochloric–nitric acid mixture (1:30).
As ICP–OES relies on the efficiency of the leaching of the solid,
only accurate data for Mg/Zr ratio could be obtained. The con-
centration of Mg and Zr at the catalyst surface, as well as the
Mg/Zr ratios in the surface and in the bulk material, are report-
ed in Table 2. In addition, XPS analysis of the samples revealed
the presence of Na in very low quantities (<1%), as a residue
from NaOH used in the preparation process.
Homogeneous behavior has been observed for the three
catalysts prepared by coprecipitation using different HSAGs as
supports. However, in the case of the catalyst prepared by in-
cipient wetness impregnation (Mg–Zr/HSAG300*) and the cata-
lyst prepared using CNFs as the support, the behavior is clearly
different. For Mg–Zr/HSAG300*, Mg and Zr concentrations are
markedly lower but retain the stoichiometric ratio. This low
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2013, 6, 463 – 473 466

CHEMSUSCHEM
FULL PAPERS
www.chemsuschem.org
graphitic materials had a more homogeneous distribution
among weak, medium, and strong basic sites. Basicity data ob-
tained were in good agreement with binding-energy values
obtained by XPS.
The acidity of the catalysts was analyzed in terms of concen-
tration and strength distribution of the acidic sites by using
NH3–TPD; the data obtained are detailed in Table 1. The rela-
tionship between the temperature of the desorption peaks
and the type of acidity was obtained by using the method pro-
posed by Arena and co-workers: desorption peaks with a maxi-
mum in the ranges 453–523, 553–603, and 653–773 K are cur-
rently attributed to NH3 chemisorbed on weak, medium, and
strong acid sites, respectively. It is not possible to discriminate
between Brønsted and Lewis acidity.[28] Supported catalysts
had a desorption peak at temperatures over 800 K, not present
in the bulk material, that was associated with functional
groups of the graphite and nanofiber supports.[29] These peaks,
which are associated to carbonaceous supports are not taken
into account in the general discussion as it was concluded in
previous work that the acidic sites that influence the reaction
are of weak and medium strength.[14] The acidity of the cata-
lysts supported on graphite showed similar trends, with differ-
ent concentrations but similar strength distribution. In all of
these materials medium-strength acidic sites disappeared, but
their contribution to the global acidity for the bulk material
and the CNFs is important. The weakest acidity increased in
supported materials, mainly in Mg–Zr/HSAG300.
Figure 3. Activity results obtained after 24 h for the aldol condensation of
furfural and acetone. Results analyzed in terms of acetone (C3) and furfural
(C5) conversion, C8 and C13 selectivities, and carbon balance (C.B.): Mg–Zr
(white); Mg–Zr/HSAG300* (light grey); Mg–Zr/HSAG300 (dark grey) and, Mg–
Zr/CNF (black).
Catalytic activity
Aqueous phase aldol condensation of acetone and furfural was
previously performed at 323 K over the bulk Mg–Zr oxide, after
24 h reaction time selectivities of 61.5 (C13) and 14.7% (C8)
were obtained with a carbon balance closure of 98.4%.[14]
As previously mentioned, the surface chemistry of mixed
oxides (especially basic site distribution) largely changes after
deposition onto a carbonaceous support. As aldol condensa-
tion reactions are catalyzed by basic sites, catalytic activities
are expected to change accordingly. To study this, CNFs and
HSAGs (taking HSAG300 as representative) were tested as sup-
ports for the active phase for the formation of C8 and C13 at
323 K. The final results (after 24 h reaction time) are shown in
Figure 3. Selectivities are defined as carbon atoms in each con-
sidered product (C8 or C13) divided by acetone and furfural
atoms in the feedstock.
When comparing reactant conversions and products selec-
tivities, it was observed that catalysts with poor active-phase
dispersions (Mg–Zr/HSAG300* and Mg–Zr/CNF) only yield the
C8 adduct, but never more than 10%, with negligible produc-
tion of the C13 adduct. In contrast, the Mg–Zr/HSAG300 catalyst
yielded higher reactant conversions and C8 selectivity com-
pared to the bulk oxide (the C13 selectivity and the carbon bal-
ance closure are very similar). The lower concentration of the
active phase (in all the cases, 0.5 g of catalyst were used) sug-
gested that this type of catalyst presents enhanced activity
compared to the bulk oxides.
Figure 4. Initial rates of C8 and C13 formation. Values obtained with Mg–Zr,
Mg–Zr/HSAG300*, Mg–Zr/HSAG300 and Mg–Zr/CNF considering: (a) the
total basicity and, (b) the medium-strength basic site concentration.
To compare the final results at the same conditions, product
concentrations were normalized considering the total concen-
tration of basic sites; results are shown in Figure 4. Analyzing
the C8 and C13 adduct-formation rates and considering the
total basic site concentration (Figure 4a), it was observed that
all supported catalysts showed a lower C8 formation rate than
the bulk oxide (less noticeable in the case of Mg–Zr/HSAG300).
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2013, 6, 463 – 473 467

CHEMSUSCHEM
FULL PAPERS
www.chemsuschem.org
Concerning the C13 formation rate, different behaviors were
found: Mg–Zr/HSAG300* and Mg–Zr/CNF showed a very low
formation rate, whereas Mg–Zr/HSAG300 showed the highest
rate.
It has been demonstrated in previous works that medium-
strength basic sites (corresponding to bidentate adsorption
modes of CO2 on the surface) are catalytically active for aldol
condensation.[14] For the first condensation reaction (formation
of C8 adduct), similar behavior was observed for both the bulk
mixed oxide and for the CNF-supported catalyst, whereas the
catalysts prepared by incipient wetness impregnation did not
show this type of basic site. However, the HSAG-supported cat-
alyst prepared by coprecipitation was four times more active
for this reaction than the other catalysts. This trend was sharp-
ly augmented for C13 adduct formation, with the reaction rate
per unit of active site seven times higher than for those corre-
sponding to the bulk oxide.
Differences between the catalysts were more marked when
the reaction rate was defined per medium-strength basic site,
which are considered the active sites for aldol condensation
(Figure 4b). Mg–Zr/HSAG300 showed the highest rate for C8
and C13 production, which was more significant for C13 as it
was six-times higher than that of the unsupported catalyst.
The results obtained with Mg–Zr/HSAG300* and Mg–Zr/CNF
denoted a slight increase in the C8 formation rate but lower
values for the C13 adduct formation. The same trends were ob-
served for the reactant conversions: Mg–Zr/HSAG300 showed
the highest conversion with similar conversion of furfural
(close to 80% in both cases, supported and unsupported) and
with increased acetone reduction compared to the bulk mate-
rial (63.2%). Mg–Zr/HSAG300* and Mg–Zr/CNF showed the
poorest performance with less than 45% acetone and 60% fur-
fural conversion in both cases. In addition, carbon balance clo-
sures (59.8% with the graphite and 58.3% with the nanofiber)
were also low for these two catalysts, suggesting the presence
of side reactions. The carbon balance obtained with Mg–Zr/
HSAG300 reached almost 90%, very similar to the bulk oxide.
Acetone self-condensation was previously reported as the
main side reaction involving the acetone and the mesityl oxide
formation competing with the aldol cross-condensation of fur-
fural and acetone. In the bulk material, the global C6 selectivity
(diacetone alcohol and mesityl oxide) was 4.3% after 24 h reac-
tion time.[14] For the supported catalyst, the formation of C6
compounds decreased to values lower than 2% (0.4% with
Mg–Zr/HSAG300*, 1.7% with Mg–Zr/HSAG300, and 1.2% with
Mg–Zr/CNF). This result is interesting as both reactions are cat-
alyzed by using the same type of active site. It suggests the
presence of a synergic effect between the active phase and
the carbonaceous support, enhancing those reactions in which
furfural and condensation adducts participate. The ratio be-
tween mesityl oxide and diacetone alcohol was higher for the
supported materials compared to the bulk material. This can
be related to the high concentration of weakly acidic sites
present on the carbon materials that are needed for catalyzing
alcohol dehydration.
A synergic effect of the carbonaceous materials can be in-
ferred from this and the behavior observed for acetone self-
condensation. Carbon surfaces adsorb organic compounds, in-
creasing the strength of this interaction as molecular weight,
presence of heteroatoms, and the number of multiple bonds
increases.[29] Therefore, this interaction is expected to be
weaker for acetone than for furfural and even more for the
condensation adducts. If the active phase is well dispersed (as
in the case of Mg–Zr/HSAG300), it will lead to local increases in
the reactant surface concentrations in the vicinity of the active
phase, thus increasing the reaction rate. In contrast, if the
active phase is poorly dispersed, this effect will be negligible
(as in the case of the other supports). This effect also justifies
the lower conversions obtained for acetone self-condensation
observed with the supported catalyst, despite this reaction
being catalyzed by the same active sites.
Considering the effect of metal loading, good distribution
and concentration of the basicity, and high dispersion of the
mixed oxides, the best results were obtained with Mg–Zr/
HSAG300. To study the role of the morphological properties of
the graphites on their performance, HSAG materials with simi-
lar chemical properties but different morphologies were
tested. Mg–Zr was supported on HSAG100 and HSAG500 and
results obtained were compared with those obtained with
HSAG300, summarized in Figure 5.
To analyze the activity results, three main parameters were
considered: the active phase loading, the distribution and con-
centration of the basic sites, and the role of the support on
the adsorption of reactants and products. Concerning the
metal loading, analytical results demonstrate poor impregna-
tion over the nanofibers and the graphites prepared by incipi-
ent wetness impregnation (Mg–Zr/HSAG300*). The low con-
centration of Mg–Zr on the catalytic surface caused the reac-
tion development to be difficult; therefore very low amounts
of C8 and C13 adducts could be formed. Concerning the basici-
ty, similar distributions of basic sites were obtained for all of
the supported materials, except in the case of HSAG300* in
which no bidentate sites were observed (corresponding to the
catalytically active sites for aldol condensation). Consequently,
the absence of these sites hindered the aldol condensation.
Figure 5. Activity results obtained in the aldol condensation catalyzed by
different graphites (after 24 h reaction time): Mg–Zr/HSAG100 (white), Mg–
Zr/HSAG300 (light grey); Mg–Zr/HSAG500 (dark grey). Results analyzed in
terms of acetone (C3) and furfural (C5) conversion, C8 and C13 selectivities,
and carbon balance (C.B.).
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2013, 6, 463 – 473 468

CHEMSUSCHEM
FULL PAPERS
www.chemsuschem.org
Concerning the reactant conversion, similar results were ob-
tained in all cases, with less than 10% difference between
them. Although metal-oxide dispersion increases as the surface
area increases (Table 2), the final conversion does not show
any systematic trend, suggesting again that the support affects
catalyst performance. The product selectivity results were also
very similar in the three cases, as well as the carbon balance
closure. The differences between them were less than 5% if
both products are considered together, which is consistent
with the similar medium-strength basicity. Considering each
product independently, higher C13 selectivity was obtained
with Mg–Zr/HSAG500 (55%), whereas Mg–Zr/HSAG300 provid-
ed higher C8 selectivities.
reaction. The similarity of the reaction patterns suggests the
absence of mass-transfer effects (as pore diameters vary from 8
to 17 nm) and that the only role of the dispersion of the active
phase is the formation of new active basic sites.
Catalytic stability
The stability of the supported catalysts was studied by per-
forming a second reaction cycle with the used catalyst. Cata-
lysts used in the first reaction cycle were filtered and dried at
373 K before being used in a second reaction cycle. This
second reaction cycle was performed under the same condi-
tions and a constant reactant/catalyst ratio.
The results were analyzed in terms of the initial reaction rate
expressed by the total concentration of both basic sites (Fig-
ure 6a) and the concentration of medium-strength basic sites
The obtained selectivities after 24 h on stream are summar-
ized in Figure 7. Mg–Zr/HSAG100 and Mg–Zr/HSAG500 present
Figure 7. C8 and C13 selectivities obtained after two reaction cycles (24 h
each cycle) with Mg–Zr/HSAG100, Mg–Zr/HSAG300 and Mg–Zr/HSAG500.
Results after the first cycle (light colors), and second cycle (dark colors).
higher stability and less than 30% difference between the first
and second cycle for C13 selectivity. The same experiment is re-
ported in a previous study for bulk Mg–Zr, showing that C13 se-
lectivity decreases by more than 85%.[14]
Mg–Zr/HSAG300 showed the poorest results, with 65 and
76% loss of selectivity towards C8 and C13, respectively. These
differences are consistent with the temperature-programmed
oxidation (TPO) analysis of the spent catalysts, shown in
Figure 8, which depicts the CO2 released during the tempera-
ture-programmed treatment of the spent catalysts in an oxidiz-
ing atmosphere. TPO of the fresh materials are also shown by
the dashed lines, coke deposits are identified as the difference
between the signals corresponding to used and fresh samples.
The coke blocks most of the active centers, as confirmed by
Brunauer–Emmett–Teller (BET) analysis: N2 physisorption onto
the Mg–Zr/HSAG300 spent catalysts indicated less than 20% of
the original surface area: 45.1 m2 gÀ1. The blockage of pores in
the other catalysts was less noticeable, with spent catalyst
areas of 72 m2 gÀ1, in the case of Mg–Zr/HSAG100 (66% of the
original area), and 270 m2 gÀ1 in the case of Mg–Zr/HSAG500
(80% of the original value).
Figure 6. C8 and C13 formation rates. Values obtained with Mg–Zr/HSAG100,
Mg–Zr/HSAG300 and Mg–Zr/HSAG500 considering: (a) the total basicity, and
(b) the medium-strength basic site concentration.
(Figure 6b). The differences between the catalysts are much
lower than those found for the first set of experiments. The re-
activity trends of HSAG100 and HSAG500 can be explained in
terms of the concentration of the medium-strength basic sites,
HSAG300 showed slightly lower activities than expected, espe-
cially in the case of C8 formation. The higher acidity of Mg–Zr/
HSAG300 (compared with the other HSAG) can explain the low
C8 formation rate, considering that aldol condensation is a re-
versible process and that the acidic sites catalyze the reverse
The observed deactivation pattern is explained considering
both the surface chemistry and the morphology of the sup-
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2013, 6, 463 – 473 469

CHEMSUSCHEM
FULL PAPERS
www.chemsuschem.org
Molecular diameters of furfural and condensation adducts
were estimated by Shen et al.,[30] providing values from 0.57
(furfural) to 0.74 nm (C13 adduct). However, the same authors
remark that these molecules (especially the adducts) are highly
non-spherical, with critical sizes (length) of up to 1.4 nm (C13
adduct). Therefore, the ratio between the molecule and pore
diameters is close to 0.1 in the least favorable case (C13 adduct
and narrowest pore). If it is considered that heavier condensa-
tion products present both larger molecular diameters and
more non-spherical character, shape selectivity effects can tune
the formation of these carbonaceous deposits. The HSAG300-
supported catalyst presents the worst behavior because of the
lowest pore and crystallite ratio, suggesting that an important
fraction of the active phase is not in the porous structure,
therefore not achieving the shape limitations for coke
formation.
Reaction mechanism and kinetic model
The presence of mass-transfer effects has been ruled out by
ensuring that both particle size and stirring the reaction mix-
ture do not affect obtained results and also by theoretical con-
siderations.[31] The study of liquid–solid mass transfer and the
Thiele-modulus-based efficiency factor for internal diffusion
demonstrated that external mass transfer is largely faster than
intrinsic reaction kinetics and the efficiency factor is close to
one.[32,33]
Figure 8. TPO of the catalyst after 25 h reaction time: (a) Mg–Zr/HSAG100;
(b) Mg–Zr/HSAG300; (c) Mg–Zr/HSAG500. Broken lines are profiles corre-
sponding to the same fresh catalyst.
The reaction mechanism of the aldol condensation catalyzed
by heterogeneous catalysts (mixed oxides) was discussed in
previous works,[14] concluding that the rate-determining step is
the abstraction of the a-proton on the acetone that was previ-
ously adsorbed over the catalytic surface by medium-strength
basic sites (bidentate centers). Once the enolate is formed, the
reaction can proceed via reaction with another acetone mole-
cule, yielding the C6 by-product, or with a furfural molecule,
yielding the C8 adduct. This adduct also has an a-proton that
can be abstracted to provide a new enolate able to react with
another furfural molecule yielding the C13 adduct.
ports. The deactivation of the bulk mixed oxide is attributed to
the formation of insoluble condensation products, also cata-
lyzed by basic sites, covering the active sites of the catalyst
(leaching effects were not observed under the operating con-
ditions for bulk or for supported catalysts).[14] Similar deactiva-
tion patterns were expected for the supported catalysts, as re-
actions leading to the formation of these insoluble products
are catalyzed by the same sites as the desired reaction. Howev-
er, the deactivation is markedly lower for the supported cata-
lysts, suggesting that the support plays an important role in al-
tering the reactions involved. When the size of the mixed
oxide particles (Table 2) and the average pore diameter
(Table 1) were compared, it was observed that the HSAG100
support was most likely to result in active-phase crystallites
within the pore. This support showed the highest reusability
ratio and this is attributable to its larger pore diameter, which
increases further upon impregnation. This suggests that shape
selectivity effects hinder the formation of carbonaceous depos-
its. Shen et al.[30] suggested using modified Y-zeolites as cata-
lysts for this reaction (pore diameter about 0.7 nm) to intro-
duce important shape selectivity effects, favoring the forma-
tion of C8 versus C13. In the case of HSAG-supported catalysts,
pore diameters are higher and do not affect C13 formation
rates, but they do hinder the formation of heavier condensa-
tion products.
According to this mechanism a single power law kinetic
model was developed, in which both C8 and C13 formation
rates follow first-order dependence on the acetone and C8 con-
centrations (enolizable species) and zero-order dependence on
furfural concentration. The kinetic data are summarized in
Table 3, and the ability of the model to predict the evolution
Table 3. Kinetic constants (minÀ1) for the fitting of the experimental re-
sults to the proposed kinetic model.[a]
Catalyst
k1
k2
k
k
r2 [b]
À1
À2
MgZr
0.0021
0.0069
0.0065
0.0070
0.006
0.004
0.076
0.97
0.94
0.96
0.96
MgZr/HSAG100
MgZr/HSAG300
MgZr/HSAG500
0.0018
0.0034
0.0042
0.0014
0.0019
0.0016
0.0025
0.0024
0.0028
[a] Reaction 1: condensation of acetone and furfural yielding C8 units; Re-
action 2: condensation of C8 and furfural yielding C13 adduct. [b] Good-
ness of fit.
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2013, 6, 463 – 473 470

CHEMSUSCHEM
FULL PAPERS
www.chemsuschem.org
of basic sites exposed to the reactants increases and
an increased activity is achieved.
The lack of anchoring sites for mixed oxides on the
carbon nanofibers hinders the depostion of the
active phase over this surface, and poor results were
obtained for this catalyst. Concerning the graphites,
the coprecipitation method leads to the most regular
disperson of the active phase on the support.
Mg–Zr/HSAG catalysts achieve higher conversion
(96.5% for furfural for the HSAG100 support) and
higher C8 and C13 selectivity (80–90%) than the bulk
oxide (considered as the most active catalyst for this
reaction). These activity trends are related to the
more favorable basic strength distribution as well as
to the interactions of the reactant with the carbon
surface.
It has been demonstrated that it is possible to min-
imize the deactivation caused by the deposition of
heavy condensation products by tuning the morphol-
ogy of the carbonaceous support.
Experimental Section
Catalysts preparation
HSAG100, HSAG300 and HSAG500 were kindly supplied
by Timcal S.A. (Switzerland). PR-24-HHT carbon nanofib-
ers (<0.2% Fe) were supplied by Applied Sciences (Ce-
darville, OH). These materials were used as supports to
increase the activity of the oxides by increasing the dis-
persion, trying to keep the oxide loading constant and
the same Mg/Zr ratio as in the bulk material (Mg/Zr =4).
Supported materials were prepared by using two differ-
ent methods: incipient wetness impregnation and copre-
cipitation. Mg–Zr/HSAG300* was synthesized by incipient
wetness impregnation of the mixed oxide precursor so-
Figure 9. Comparison between experimental and kinetic model fitting concentrations of
^
furfural (&), acetone ( ), C8 (*), and C13 (ꢅ) for the catalysts: (a) Mg–Zr/HSAG100,
(b) Mg–Zr/HSAG300, and (c) Mg–Zr/HSAG500.
of reactants and products is shown in Figure 9. In Table 3,
lution using the carbon support (5 g). Aqueous solution (2.35 cm3)
was prepared with magnesium nitrate hexahydrate (0.455 g) and
zirconyl nitrate (0.037 g). The metal oxide loading was limited by
the low solubility of the zirconyl precursor. The impregnated sup-
port was stirred in NaOH solution (500 mL, pH 10) for 1 h. The ma-
terial was then filtered, washed until it reached pH 7, and dried at
373 K. Finally, it was treated using a He flow at a heating rate of
3 KminÀ1 until the temperature reached 873 K. Mg–Zr/CNF, Mg–Zr/
HSAG100, Mg–Zr/HSAG300, and Mg–Zr/HSAG500 were prepared
using the coprecipitation method according to the procedure de-
scribed by Winter et al.[16] Supports (5 g) were suspended by con-
stant stirring in 500 mL of a solution containing 25.45 g of the
magnesium salt and 2.02 g of the zirconyl precursor. 25 wt% NaOH
solution was added until the pH value was 10 and the final solu-
tions were kept stirring for 1 h. They were filtered, washed, and
dried at 383 K in an oven before being treated using a He flow
with a heating rate of 3 KminÀ1 until 873 K.
After analyzing their morphological and physical–chemical proper-
ties, as well as their behavior as catalysts in the furfural–acetone
cross condensation, the coprecipitation method was chosen as the
preparation technique.
The bulk material, Mg–Zr oxide, was chosen because of its higher
activity demonstrated in our previous study.[14] It was synthesized
using the sol–gel technique described by Aramendꢀa et al.,[34] pre-
pared by dissolving 50.9 g of magnesium nitrate hexahydrate
values of Mg–Zr (bulk material) were also shown to compare
the data obtained.
The kinetic constant values were very similar in the three
cases, which is in accordance with the similarity of the cata-
lysts. Data obtained are consistent with the activity results
such that supported catalysts have higher values of k1 and
lower values of k compared to the bulk material, which im-
À1
plies greater final selectivity of the C8 adduct. Mg–Zr/HSAG300
shows the highest value of k of the three supported cata-
À1
lysts, consistent with the hypothesis of a higher influence of
retro-aldolization. The k2 values of the supported materials
were all lower than those obtained with the unsupported ma-
terials; however, in the case of Mg–Zr/HSAG500, the k2 value
was twice that obtained for Mg–Zr/HSAG100 and 20% higher
than the corresponding Mg–Zr/HSAG300 value. This result is in
good agreement with the C13 selectivities obtained.
Conclusions
Mg–Zr mixed oxides supported on mesoporous carbon materi-
als are very promising catalysts for the furfural–acetone aldol
condensation. Supporting the active phase, the concentration
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2013, 6, 463 – 473 471

CHEMSUSCHEM
FULL PAPERS
www.chemsuschem.org
(Fluka, >99.0%) and 4.04 g of ziconyl nitrate (Aldrich, hydrated) in
1 L of deionized water. The mixture was stirred at room tempera-
ture, and aqueous NaOH (Prolabo, 98%) solution was added until
the pH value was equal to 10. The gel was aged for 72 h, filtered,
and washed with deionized water until pH 7. It was dried at 393 K
over 24 h and finally calcined in O2 (100 cm3 minÀ1) at 873 K for 3 h
with N2, heated to reaction temperature, and stirred at 1000 rpm
for 24 h. Samples (12–15 samples of 1 mL) were withdrawn from
the sampling port during the condensation reaction. Samples were
filtered, extracted with ethyl acetate, and analyzed by performing
gas chromatography with a Shimadzu GC-2010 equipped with
a flame ionization detector (FID) and using a 15 m long CP-Sil 5 CB
capillary column as the stationary phase. Peak assignment was per-
formed by fractionation with HPLC and analysis by GC–MS using
standard calibration mixtures, following the same procedure as in
our previous work.[15]
at a heating rate of 5 KminÀ1
.
Catalyst characterization
Deactivation studies were carried out by collecting the solution ob-
tained after the first reaction with the catalyst in suspension and
filtering with a 2-micron mesh. The catalyst obtained was then
used in a new reaction with fresh reagents, keeping the same con-
ditions of temperature, reagent concentration, pressure, stirring,
and organic/catalyst ratio. Samples obtained were analyzed follow-
ing the same protocol as in the first cycle.
The specific surface area and pore volume were analyzed by N2 ad-
sorption at 77 K in a Micromeritics ASAP 2020 by using the BET
method (to calculate the surface area) and the Barrett–Joyner–Ha-
lenda (BJH) approach (to determine pore volume and diameter).
The crystallographic structures of the supported materials were de-
termined by XRD using a Philips PW 1710 diffractometer, working
with a CuKa line (l =0.154 nm) in the 2q range between 58 and
858, at a scanning rate of 28 per min. The active-phase loading of
the catalysts was determined by ICP–OES using SPECTRO-CIR-
OSCCD ICP-spectrometer. Approximately 50 mg of the sample was
inserted into a Teflon bomb; HF (4 mL), HCl (1 mL), and HNO3
(0.5 mL) were added. The sample was dissolved in a microwave
oven, diluted with deionized water, and analyzed in the spectrome-
ter. The oxides particle morphology and size distributions were de-
termined by TEM by using a MET JEOL-2000 EX-II microscope, and
the surface ratio Mg/Zr was investigated by SEM. The analysis was
conducted in a MEB JEOL-6100 with EDX analyzer.
Acknowledgements
This work was supported by the Spanish Government (contracts
CTQ2008-06839-C03-02 and CTQ2011-29272-C04-02). L.F thanks
the Government of the Principality of Asturias for a Ph.D. fellow-
ship (Severo Ochoa Program). TIMCAL is acknowledged for sup-
plying HSAG samples.
The concentration and strength distribution of the basic/acidic
sites were determined by TPD of preadsorbed CO2 or NH3 in a Mi-
cromeritics TPD/TPR 2900. Samples (10 mg) were treated with He
at 723 K for 2.5 h and then exposed to a CO2 or NH3 (2.5% NH3 in
He) stream at 323 K until saturation coverage was reached. Weakly
adsorbed CO2 or NH3 was removed by flushing with He at the
same temperature for about 1 h 30 min. The temperature was then
increased at a heating rate of 5 KminÀ1 from 293 to 923 K, and
CO2 or NH3 evolution was monitored by mass spectrometry. After
the reaction, carbonaceous deposits were characterized by using
TPO, which used the same apparatus as the TPD experiments.
A carbon sample (10 mg) was maintained in a stream of O2/He
(2%:98%) at 323 K for 30 min before being heated to 1200 K at
5 K per min.
The surface composition of the mixed oxides was measured by
XPS, using a SPECS system equipped with a Hemispherical Phoibos
detector operating in a constant pass energy, using MgKa radiation
(hu=1253.6 eV). The samples were fixed to the sample holder
using a carbon adhesive tape. The background pressure in the
analysis chamber was kept below 4·10À9 mbar during data acquisi-
tion. As samples are non-conducting, surface neutralization during
measurements was required.
Keywords: biomass · carbon · cÀc coupling · heterogeneous
catalysis · graphite
[1] D. L. Klass, Biomass for Renewable Energy, Fuels and Chemicals, Academic
Press, San Diego, 1998.
[2] D. M. Alonso, S. G. Wettstein, J. Q. Bond, T. W. Root, J. A. Dumesic, Chem-
SusChem 2011, 4, 1078–1081.
[3] A. Corma, O. de la Torre, M. Renz, Energy Environ. Sci. 2012, 5, 6328–
6344.
[4] J. C. Serrano-Ruiz, R. Luque, A. Sepffllveda-Escribano, Chem. Soc. Rev.
2011, 40, 5266–5281.
[5] S. De, S. Dutta, B. Saha, ChemSusChem 2012, 5, 1826–1833.
[6] E. L. Kunkes, D. A. Simonetti, R. M. West, J. C. Serrano-Ruiz, C. A. Gärtner,
J. A. Dumesic, Science 2008, 322, 417–421.
[7] K. Karimi, S. Kheradmandinia, M. J. Taherzadeh, Biomass Bioenergy 2006,
30, 247–253.
[8] J. Barrett, J. N. Chheeda, G. W. Huber, J. A. Dumesic, Appl. Catal. B 2006,
66, 111–118.
[9] J. C. Serrano-Ruiz, D. J. Braden, R. M. West, J. A. Dumesic, Appl. Catal. B
2010, 100, 184–189.
[10] D. A. Simonetti, J. A. Dumesic, ChemSusChem 2008, 1, 725–733.
[11] J. I. Di Cosimo, V. K. Dꢀez, C. R. Apesteguꢀa, Appl. Catal. A 1996, 137,
149–166.
[12] R. Zeng, X. Fu, C. Gong, Y. Sui, X. Me, X. Yang, J. Mol. Catal. A: Chem.
2005, 229, 1–5.
Reaction studies
[13] M. Choudary, M. L. Kantam, P. Sreekanth, T. Bandopadhyay, F. Figueras,
A. Tuel, J. Mol. Catal. A: Chem. 1999, 142, 361–365.
[14] L. Faba, E. Dꢀaz, S. OrdꢁÇez, Appl. Catal. B 2012, 113 – 114, 201–211.
[15] L. Faba, E. Dꢀaz, S. OrdꢁÇez, Catal. Today 2011, 164, 451–456.
[16] F. Winter, V. Koot, A. J. Dillen, J. W. Geus, K. P. deJong, J. Catal. 2005,
236, 91–100.
[17] N. A. Niels, H. Meis, J. H. Bitter, K. P. de Jong, Ind. Eng. Chem. Res. 2010,
49, 1229–1235.
[18] S. OrdꢁÇez, E. Dꢀaz, R. F. Bueres, E. Asedegbega-Nieto, H. Sastre, J. Catal.
2010, 272, 158–168.
The furfural and acetone aldol cross-condensation was carried out
in a stirred batch autoclave reactor (0.5 L, Autoclave Engnieers EZE
Seal, 500 mL) equipped with a proportional-integral-derivative
(PID) temperature controller and a back-pressure regulator. The re-
actor was loaded with 0.25 L of an aqueous solution of furfural
(Panreac, 98%) and the catalyst (500 mg, with an average particle
diameter of 50–80 nm). Air was purged by adding N2, up to 15 bar,
three times before starting the condensation reaction. Acetone
(Panreac, 99.5%) was introduced to start the reaction, resulting in
5% of organic compounds in aqueous solution (furfural/acetone
molar ratio of 1:1). Then, the reactor was pressurized to 10 bar
[19] E. Dꢀaz, S. OrdꢁÇez, A. Vega, J. Colloid Interface Sci. 2007, 305, 7–16.
[20] E. Dꢀaz, S. OrdꢁÇez, R. F. Bueres, E. Asedegbega-Nieto, H. Sastre, Appl.
Catal. B 2010, 99, 181–190.
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2013, 6, 463 – 473 472

CHEMSUSCHEM
FULL PAPERS
www.chemsuschem.org
[21] X. Chakraborty, D. Deva, A. Sharma, N. Verma, J. Colloid Interface Sci.
2011, 359, 228–229.
[22] D. Gulkovꢄ, O. Solcovꢄ, M. Zdrezil, Microporous Mesoporous Mater. 2004,
76, 137–149.
[30] W. Shen, G. A. Tompsett, K. D. Hammond, R. Xing, F. Dogan, C. P. Grey,
W. C. Conner Jr., S. M. Auerbach, G. W. Huber, Appl. Catal. A 2011, 392,
57–68.
[31] S. OrdꢁÇez, B. P. Vivas, F. V. Dꢀez, Appl. Catal. B 2010, 95, 288–296.
[32] F. Rodrꢀguez-Reinoso, M. Molina-Sabio, Adv. Colloid Interface Sci. 1998,
76–77, 271–294.
[23] M. Malyaadri, K. Jagadeeswaraiah, P. S. Sai Prasad, N. Lingaiah, Appl.
Catal. A 2011, 401, 153–157.
[33] H. S. Fogler, M. N. Gürmen in Elements of Chemical Reaction Engineering,
4th ed., Pearson, 2008, chap. 11.
[24] I. Sꢄdaba, M. Ojeda, R. Mariscal, J. L. G. Fierro, M. Lꢁpez Granados, Appl.
Catal. B 2011, 101, 638–648.
[34] M. A. Aramendꢀa, V. Borau, C. JimØnez, A. Marinas, J. M. Marinas, J. R.
Ruiz, F. J. Urbano, J. Mol. Catal. A: Chem. 2004, 218, 81–90.
[25] G. Busca, Chem. Rev. 2010, 110, 2217–2249.
[26] V. Fournier, P. Marcus, I. Olefjord, Surf. Interface Anal. 2002, 34, 494–497.
[27] Y. H. Hu, E. Ruckenstein, Catal. Lett. 1997, 43, 62–71.
[28] F. Arena, R. Dario, A. Parmaliana, Appl. Catal. A 1998, 170, 127–137.
[29] M. Paulis, M. Martı’n, D. B. Soria, A. Dꢀaz, J. A. Odriozola, M. Montes,
Appl. Catal. A 1999, 180, 411–420.
Received: September 20, 2012
Revised: November 29, 2012
Published online on January 29, 2013
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2013, 6, 463 – 473 473