Exploitment of niobium oxide effective acidity for xylose dehydration to furfural

Article abstract of DOI:10.1016/j.cattod.2015.01.018

Xylose together with other pentose and hexose sugars can be dehydrated to produce interesting platform chemical compounds, like 5-hydroxyfurfural (HMF) and furfural. This study continues our investigation on the niobium oxide based catalysts in connection with the research of adequate solvents systems to minimize catalyst deactivation and increase catalyst stability and durability during the dehydration of sugars. Silica-zirconia supported niobia samples (10 wt.% of Nb) prepared by impregnation or sol-gel in comparison with pure niobic acid are here presented for xylose dehydration. The reactions have been studied at different temperatures (130-180 °C) in batch or fixed bed continuous catalytic reactors in various solvents. Green solvents soluble in the aqueous solution of xylose or biphasic systems have been taken into account: water, water-isopropanol mixtures, water-γ-valerolactone, and water-cyclopentylmethyl ether. The surface acidities of the catalysts have been measured in cyclohexane (intrinsic acidity) and also in water to determine the effective catalyst acidity. The continuous tests and batch-recycling tests showed that the supported Nb-catalysts, even if initially less active, are more stable than niobic acid. The presence of isopropanol in water improves both the activity and stability of the catalysts in comparison with water and the use of cyclopentylmethyl ether gave the most interesting selectivity to furfural preserving the catalyst stability.

Full text of DOI:10.1016/j.cattod.2015.01.018

Catalysis Today 254 (2015) 90–98
Contents lists available at ScienceDirect
Catalysis Today
journal homepage: www.elsevier.com/locate/cattod
Exploitment of niobium oxide effective acidity for xylose dehydration
to furfural
Maria José Campos Molinaa, Manuel López Granadosa, Antonella Gervasinib,
Paolo Carnitib,∗
a Institute of Catalysis and Petrochemistry (CSIC), C/Marie Curie 2, Campus de Cantoblanco, 28049 Madrid, Spain
b Dipartimento di Chimica, Università degli Studi di Milano, via Golgi 19, 20133 Milano, Italy
a r t i c l e i n f o
a b s t r a c t
Article history:
Xylose together with other pentose and hexose sugars can be dehydrated to produce interesting platform
chemical compounds, like 5-hydroxyfurfural (HMF) and furfural. This study continues our investigation
on the niobium oxide based catalysts in connection with the research of adequate solvents systems to
minimize catalyst deactivation and increase catalyst stability and durability during the dehydration of
sugars. Silica-zirconia supported niobia samples (10 wt.% of Nb) prepared by impregnation or sol–gel in
comparison with pure niobic acid are here presented for xylose dehydration. The reactions have been
studied at different temperatures (130–180 ◦C) in batch or fixed bed continuous catalytic reactors in vari-
ous solvents. Green solvents soluble in the aqueous solution of xylose or biphasic systems have been taken
into account: water, water–isopropanol mixtures, water-␥-valerolactone, and water-cyclopentylmethyl
ether. The surface acidities of the catalysts have been measured in cyclohexane (intrinsic acidity) and also
in water to determine the effective catalyst acidity. The continuous tests and batch-recycling tests showed
that the supported Nb-catalysts, even if initially less active, are more stable than niobic acid. The presence
of isopropanol in water improves both the activity and stability of the catalysts in comparison with water
and the use of cyclopentylmethyl ether gave the most interesting selectivity to furfural preserving the
catalyst stability.
Received 5 September 2014
Received in revised form 14 January 2015
Accepted 16 January 2015
Available online 9 March 2015
Keywords:
Xylose dehydration
Furfural synthesis
Green solvents
Niobium oxide
Solid acid catalysts
Catalyst intrinsic and effective acidities
© 2015 Elsevier B.V. All rights reserved.
1. Introduction
solid acids containing niobium [12] has opened the possibility to
use such materials as efficient catalysts in reactions where water
Furfural is an important chemical species produced from
pentosan-rich biomass, in particular by xylose dehydration. Cur-
rently, conventional processes of furfural production utilize
mineral acids as catalysts but the development of more friendly acid
solid catalysts with separable and reusable properties is required
to prevent the environmental and economic drawbacks associated
to homogeneous catalysis such as extreme corrosion, high toxicity,
and excessive waste disposal.
In the past years, many solid acid catalysts have been devel-
oped and successfully applied in this reaction [1–11]. A problem
arrives when water is concerned as reaction solvent; in water, or
in general in highly protic and polar solvents, only very few solid
acids can maintain the desirable acidity due to the solvent–surface
interactions by solvation and coordination abilities of such sol-
vents. The discovery of the water-tolerant properties of several
is concerned as reactant, product, or solvent. Thus, different acid-
catalyzed reactions, such as hydrolysis, dehydration, condensation,
and esterification, among others, have been performed using nio-
bium based catalysts [9,13,14].
In particular, the hydrated niobium pentoxide, Nb2O5·nH2O
(NBO), which is usually called niobic acid, is considered one of
the most promising water-tolerant solid acid catalyst [15,16]. The
lively acid properties of niobic acid, which can be maintained also in
water, have been exploited in several reactions of biomass transfor-
mation as dehydration of pentoses and hexoses to obtain platform
compounds such as furfural and 5-hydroxymethylfurfural (HMF),
respectively [9,17,18].
The catalytic dehydration of pentoses and hexoses in water with
niobic acid presents different disadvantages to overcome for prac-
tical application; in particular, low selectivity and yield to the main
products (furfural and HMF) that are mainly associated to the for-
mation of several side-products, as humins and other insoluble
polymeric condensation by-products. These insoluble by-products
cause activity declining and surface deactivation, as they deposit
∗
Corresponding author. Tel.: +39 0250314261; fax: +39 0250314300.
E-mail address: paolo.carniti@unimi.it (P. Carniti).
http://dx.doi.org/10.1016/j.cattod.2015.01.018
0920-5861/© 2015 Elsevier B.V. All rights reserved.

M.J.C. Molina et al. / Catalysis Today 254 (2015) 90–98
91
on the catalyst surface. Neither the routes through which humins
are formed nor their molecular structure have yet been unequiv-
ocally established. Infrared spectroscopic experiments suggest
that humins are formed via aldol addition/condensation involv-
ing the 2,5-dioxo-6-hydroxy-hexanal (DHH) species. From DHH,
humins can grow as it is highly reactive and undergoes aldol addi-
tion/condensation with available aldehydes and ketones [19–21].
Both Lewis (LAS) and Brønsted (BAS) acid sites are involved in the
mechanism of humin formation, and it seems that the strong acid
site density of niobic acid is responsible for the fast decrease of
activity observed [14,22]. In order to decrease the acid site density
of the catalyst surfaces and to enhance the catalyst stability, sup-
ported or dispersed niobic acid on/in high surface area oxides have
been already developed [11,23,24]. Good catalyst stability during
long-term activity in the fructose dehydration has been observed
on niobia deposited on silica [23]; unfortunately, fructose conver-
sion was low likely due to the chemical inertness of silica. The
choice of a more adequate support for niobia could improve catalyst
activity in sugar dehydration reactions.
Moreover, the choice of the reaction solvent for the acid-
catalyzed conversion of carbohydrate biomass is of high impor-
tance because it is desirable that the formed by-products are
soluble, to improve the catalyst stability and durability. If water
is the most frequently used solvent, due to its excellent substrate
solubilization properties and to low cost, other friendly solvents
are searched for improving the catalyst stability, in particular. Many
different liquid solvents have been investigated as reaction medium
instead of pure water, such as ionic liquid solvents [25], organic sol-
vents [26,27] and water/organic solvent biphasic systems [28,29].
Different anhydrous solvents such as toluene [3], DMSO [1], methyl
isobutyl ketone (MIBK) [30], or cyclopentyl methyl ether (CPME)
[28] have been used as extracting solvent leading to furfural yields
generally higher than the use of water. The use of solvents in homo-
geneous aqueous solution, such as water/1-butanol system, has
been also suggested in different carbohydrate dehydration reac-
tions in order to improve the selectivity toward the target product
[2,27,31].
All the materials were used without further purification. Milli-Q
H2O was used for preparation of all aqueous solutions.
2.2. Catalyst preparation and characterization
Silica-zirconia (SZ) has been utilized as support of the NbOx
active phase, that has been deposited at 10 wt.% of Nb by classi-
cal wetness impregnation (Nb/SZi) and by a sol–gel like method
(Nb/SZsg) modifying the procedures described in Ref. [32]. Ammo-
nium niobium oxalate complex (NH4) NbO(C2O4)2 nH2O (ANBO)
and niobium (V) ethoxide (Nb(OCH2CH3)5 (NBE)) were used as Nb
sources, respectively.
For Nb/SZi preparation, a finite amount of SZ (ca. 15 g) was first
dried at 120 ◦C for 4 h. The adequate amount of ANBO was dissolved
in ca. 70 mL of water and it was added to SZ. The aqueous suspension
of SZ in the presence of ANBO was kept overnight under vigorous
stirring at room temperature. After 16 h of contact, water was mild
evaporated in a rotavapor under vacuum between 40 and 50 ◦C for
5 h and eventually the solid obtained was dried in the oven at 120 ◦C
overnight and calcined at 550 ◦C for 8 h.
For Nb/SZsg preparation, ca. 18 g of SZ was first dried at 120 ◦C
for 4 h, then an amount of ca. 80 mL of 1-propanol was added under
stirring, afterwards several drops of HCl 37% were added to the
suspension (pH 1–2). After ca. 2 h, NBE (5 mL dissolved in ca. 10 mL
of 1-propanol) was added to the slurry keeping it under stirring for
other 2 h. Then, ammonium hydroxide (NH4OH) solution (∼20 mL)
was added dropwise to obtain gelation. The unripe solid was aged
at room temperature for two days, then it was dried in a rotavapor
at r.t. for several hours, to eliminate the excess of propanol, dried
at 120 ◦C overnight. Eventually, the solid was calcined at 550 ◦C for
8 h.
Thermal gravimetric analyses (TGA) were performed on the
dried samples in a TGA analyzer from PerkinElmer (TGA7), with a
scan of 10 ◦C min−1, from 25 to 800 ◦C under flowing air. For better
evidencing of the thermal events, differential thermogravimetric
curves (DTGA) were also calculated from the parent TGA profiles.
Microstructure analysis of the samples have been determined
by adsorption and desorption of nitrogen at −196 ◦C (Sorptomatic
1900 instrument) and successive numerical interpretation of the
collected isotherms by BET and BJH models for the specific surface
area and pore size distribution, respectively. Prior to measurement,
the sample (ca. 0.1–0.3 g) crushed and sieved as 45–60 mesh parti-
cles was introduced in the sample holder and thermally activated
at 350 ◦C for 16 h under vacuum.
Therefore, the present study continues our investigation on
hydrated niobium pentoxide and niobia-based catalysts in connec-
tion with the most exhaustive research of adequate monophase
or biphase systems to minimize catalyst deactivation and increase
catalyst stability and durability. Furthermore, silica-zirconia sup-
porting niobia samples prepared by impregnation or sol–gel
process, respectively, in comparison with pure NBO, are here
presented for xylose dehydration studied at different tempera-
tures (130–180 ◦C) in batch and fixed bed continuous reactors (to
approach pilot plan conditions) using various solvents, to point out
catalyst performances and limitations.
Powder X-ray diffraction (XRD) patterns were recorded in the
15–80◦ 2ꢀ range in scan mode (0.02◦ step, 1 s) using a X’Pert
Pro PANalyticaldiffractometer with CuK␣ radiation of 0.154046 nm.
Diffractograms were analyzed with the X’PertHighScore Plus soft-
ware.
Scanning electron micrographs (SEM) were collected by a LEO-
1430 coupled with energy-dispersive X-ray spectroscopy (EDX)
working with an accelerating voltage of 20 kV.
2. Experimental
2.1. Materials
The acid titrations with PEA have been carried out at 30 ◦C in
a recirculation chromatographic line (HPLC), comprising a pump
(Waters 515) and a monochromatic UV detector (Waters, model
2487, ꢁ = 254 nm) [23]. Successive dosed amounts of PEA solution
in cyclohexane or in water were injected into the line in which
cyclohexane or water continuously circulated. The attainment of
the adsorption equilibrium was revealed by the attainment of sta-
ble UV-detector signal. The sample (ca. 0.1 g crushed and sieved as
80–200 mesh particles) was placed in a sample holder (stainless
steel tube, 4 mm i.d. and 8 cm of length) between two sand pillows.
Prior to the measurement, the sample was activated at 350 ◦C for
4 h in flowing air (8 mL min−1) and then filled with the liquid.
After the collection of the first adsorption isotherm of PEA on the
fresh sample (I run), pure solvent was allowed to flow for 30 min
d-(+)-Xylose (≥99%), anhydrous cyclopentyl methyl ether
(CPME) (≥99.9%), octanoic acid (98%), ␥-valerolactone (99%), 2-
propanol (≥99.5%), 1-propanol (≥99.5%), niobium(V) ethoxide
(Nb(OCH2CH3)5, 99.95%, NBE), 2-phenylethylamine (PEA), and
hydrochloric acid (37%) were purchased from Sigma–Aldrich.
Ammonium hydroxide (NH4OH) solution (purum, ∼28% in water)
was purchased from Fluka. Both hydrated niobium oxide (NBO) in
pellets (with graphite) and in powder form and ammonium nio-
bium oxalate complex (NH4) NbO(C2O4)2 nH2O, ANBO) were kindly
furnished from Companhia Brasileira de Metalurgia e Minerac¸ ao
(CBMM, Brazil). Silica-zirconia (5 wt.% ZrO2)was supplied from
Grace Company.

92
M.J.C. Molina et al. / Catalysis Today 254 (2015) 90–98
through the PEA-saturated sample, then a new adsorption of PEA
was repeated (II run). The collected isotherms were interpreted
following the Langmuir equation:
and internal standards for aqueous and organic phase were added
as above described before the analysis.
After the first reaction cycle, the catalyst inside the reactor was
washed several times with the reaction solvent to eliminate the
rests of xylose and organics weakly retained by the solid; the liquid
was then removed from the reactor and the catalyst was left inside
the reactor to dry for the next run.
bads[PEA]eq
PEAads
PEAads,max
=
(1)
(1 + bads[PEA]eq
)
From the conventional linearized equation, reporting
[PEA]eq/PEAads vs. [PEA]eq, the values of PEAads,max could be
obtained. Assuming a 1:1 stoichiometry for the PEA adsorption
on the acid site, the value of PEAads,max obtained from the I run
isotherm corresponded to the number of total acidic sites, while,
the value of PEAads,max obtained from the II run isotherm corre-
sponded to the number of weak acidic sites. The number of strong
acid sites was obtained as the difference between the number of
total and of weak sites.
Catalyst stability tests for long times on stream (up to 100 h) in
water and water/isopropanol solution (20% v/v isopropanol) were
carried out in a continuous reaction line equipped with a tubular
catalytic reactor and a dosing pump (HPLC pump, Waters 501). The
reactor and pre-heater were assembled in an oven with forced cir-
culation of hot air in order to keep a constant temperature (130 ◦C).
The catalyst sample (1 g), previously sieved to 25–45 mesh, was
held in the middle of the reactor, between two sands beds (0.5 g,
45–60 mesh). The aqueous xylose solution (ca. 0.3 M) was continu-
ously fed into the catalytic bed reactor. The feed flow rate was kept
constant at 0.1 mL min−1 obtaining a contact time of 10 min g mL−1
.
2.3. Catalytic tests of xylose dehydration
After starting the reaction, at least 100 mL of solution was left to
flow before collecting samples for the analysis, so ensuring the sta-
tionary conditions to be obtained. The pressure in the reactor was
kept between 103 and 2 × 103 kPa by means of a micrometric valve
at the end of the reaction line. The products were analyzed in a
liquid-chromatography apparatus (HPLC), consisting of a manual
injector (Waters U6K), pump (Waters 510), heater (Waters CHM)
for the column and refractive index detector (Waters 410). A sugar
Pack I column operating at 90 ◦C and eluted with an aqueous solu-
tion of Ca-EDTA (10−4 M) was used.
2.3.1. Catalytic tests of xylose dehydration with different solvents
All catalysts were tested in the dehydration of xylose to furfural
in batch conditions with the different selected reaction solvents at
lower (130 ◦C) and higher (160–180 ◦C) temperature.
Screening tests in different solvents were performed in amag-
netically stirred (1500 rpm) Ace sealed pressure glass reactor
(15 mL of volume) placed in a preheated oil bath at 130 ◦C for
6 h. Typical conditions used were: 4.5 wt.% of xylose and 3 wt.%
of powder catalyst (catalyst/xylose wt. ratio, 0.67). The mass
ratios of the solvents used were: for the biphasic water/CPME
system, 3:7, for the monophasic water/␥-valerolactone, 1:9, and
water/isopropanol (20% v/v), 8:2.
3. Results and discussion
Kinetic tests in water/␥-valerolactone at 160 ◦C and in
water/CPME at 180 ◦C were carried out, with different amounts of
catalyst and xylose respect to those used for the screening tests,
following the reaction for 4 h.
3.1. Characterization of materials
The two catalysts prepared by deposition of niobia over silica-
zirconia (Nb/SZi and Nb/SZsg) were prepared starting from two
different Nb precursors (ANBO and NBE) and employing two dif-
ferent preparation routes (impregnation and sol–gel, respectively).
In both cases, calcination at 550 ◦C ensured the formation of a
oxide sample. The calcination temperature was chosen based on the
results obtained from the thermogravimetric analysis performed
on the dried samples (Figure 1S, Supporting Information). Both the
thermograms show at about 100 ◦C losses of mass associated to the
physical desorption of water. The most intense losses of mass cen-
tered around 250–300 ◦C, could be attributed to the decomposition
of ANBO and NBE. Quantitative evaluation of the mass losses were
in agreement with the calculated amounts of carbon and nitrogen
of the Nb-precursor used.
X-ray powder diffraction (P-XRD) of the supported catalysts on
SZ support (Nb/SZi and Nb/SZsg) showed an amorphous halo cen-
tered at 2ꢀ = 22◦, typical of amorphous silica, indicating their main
amorphous nature. Peaks related to the presence of Nb2O5 crys-
talline aggregates are detectable only for Nb/SZi (Fig. 1, top). It
could be then inferred a higher Nb-dispersion on Nb/SZsg than on
Nb/SZi. P-XRD of pure NBO (Figure 3S, Supporting Information)
showed the typical crystalline pattern in which monoclinic H-
Nb2O5 and T-orthorhombic phases can be distinguished. On Nb/SZi
catalyst, the SEM-EDX analyses detected surface amounts of Nb2O5
of 17 5 wt.% with the different areal zones more or less covered
by Nb (Figure 2S, Supporting Information).
The reaction started when the charged reactor was immersed
in the oil bath (130 ◦C) and stopped by removing the reactor from
the oil bath and rapidly cooled down by immersion in water at
room temperature. Then, in case of biphasic systems, internal stan-
dards for organic phase (octanoic acid) and for aqueous phase (d
(+)-glucose) were added to the quenched reaction mixture and after
gently agitation for several minutes, ca. 2 mL aliquot was taken from
the each phase for the analysis after filtration (polyethersulfone
Millipore filter 0.22 ␮m).
Aqueous aliquots were analyzed with a HPLC Agilent 1200
series chromatograph equipped with a refraction index (RI) detec-
tor and a Bio-Rad Aminex HPX-87H column (300 × 7.8 mm) for
analysis of xylose and furfural in aqueous phase. A 0.005 M H2SO4
mobile phase was employed as eluent with 0.4 mL min−1 flow
rate and at 55 ◦C. In the case of organic aliquots, analysis of
furfural in this phase was conducted by gas chromatography
(CG) (Varian CP-3800) equipped with a ZB-WAX-Plus column
(30 m × 0.32 mm × 0.25 ␮m) and a flame ionization detector (FID).
2.3.2. Catalyst stability tests
Recycling experiments of the catalysts were performed in a
100 mL Parr stainless steel reactor mechanically stirred at high
temperature (180 ◦C) in water/CPME, 3:7 mass ratio, and 10 wt.%
xylose and 5 wt.% catalyst. The reactor was first loaded with xylose
and catalyst powder and the corresponding amount of solvent.
After purging with N2, the reactor was pressurized (500 kPa) and
temperature increased to reach 180 ◦C without stirring. Once
reached the stable reaction temperature, the mixture started
reacting by stirring at 1000 rpm (zero time). The reaction was
halted by stopping stirring and then the reactor was quenched to
room temperature. The solution was taken out from the reactor
The main textural properties of all catalysts and support have
been studied. Fig. 2 shows the N2 adsorption–desorption isotherms
of the two supported Nb-catalysts and bulk Nb2O5 (with graphite)
and the silica-zirconia support. In Table 1, a summary of the
main textural properties obtained is reported. SZ support shows
type IV isotherm typical of the mesoporous solids: high specific

M.J.C. Molina et al. / Catalysis Today 254 (2015) 90–98
93
because the reaction takes place in pure water and in some aque-
ous solutions (with ␥-valerolactone and isopropanol). Also when
the biphasic system water/CPME has been used, water, which has
hydrophilic characteristics, is in contact with the catalyst oxide
surfaces.
Because it is known that water with its polar, protic, and sol-
vating properties can interact with acid surfaces, modifying the
number and strength of the acid sites (in general decreasing it).
Therefore, a given acid surface can reconstruct itself in the pres-
ence of water. In order to find sound relations between the acid
properties of the catalysts and their activity, the knowledge of the
effective acidity seems more appropriate.
The intrinsic acidity of SZ is very high, but only the half of the
sites are strong acid sites (Table 2 and Fig. 3). By covering part of SZ
support with niobia phase, a decrease of surface acidity of the cat-
alysts (Nb/SZsg and Nb/SZi) has been obtained. This was expected
due to the lower surface area values of Nb/SZ compared with the
SZ support (Table 1). Interestingly, the percent of the strong acid
sites of Nb/SZsg and Nb/SZi increases compared with SZ, likely due
to the dispersed Nb-centers which can be associated to new LAS
sites created on the catalyst surfaces [33].
The determination of the effective acidity, measured in water,
shows a different scenario (Table 2 and Fig. 3). The catalyst prepared
by sol–gel (Nb/SZsg) shows the highest amount of effective acid sites
and it maintains in water ca. 65% of its intrinsic acid sites. In water,
the acidity of SZ deeply decreases as well as the acidity of Nb/SZi
(only about 35% of acidity is retained in water for both the samples).
In general, as known, the acid strength of the effective acid sites
in water is low, the highest acid strength is associated to Nb/SZsg
surface (0.042 mequiv./g corresponding to 14% of the titrated acid
sites in water).
The acidity determination of NBO confirmed the already known
water-tolerant acid properties; ca. 90% of the acid sites titrated in
cyclohexane are maintained in water and high percent (66%) of the
effective acid sites are strong. Considering the lower surface area of
NBO than that of Nb/SZsg and Nb/SZi, the acid site density of NBO
in water is very higher than that of the Nb/SZ catalysts.
Fig. 1. P-XRD of Nb/SZi (top) and Nb/SZsg (bottom) (P-XRD of NBO is shown in
Supporting Information section, Figure 3S).
surface area and pore volume with large pore diameter. The N2
adsorption/desorption isotherms of Nb/SZsg are also typical of a
mesoporous solid. Any significant variation of pore size compared
with the SZ support has not been observed, suggesting a uni-
form Nb-distribution on the support matrix. The same happens
with Nb/SZi, which has high surface area and pore volume but a
significant decrease of pore size compared with SZ. The results sug-
gest that niobium has been deposited at the external and internal
surface of SZ. Concerning Nb2O5, the results confirm the already
reported morphologic features with the presence of pores with
small size.
The surface acidity developed by the support and the prepared
catalysts is expected to direct the activity and selectivity in the
dehydration reaction of xylose. In view of the importance of the
acidity of the catalysts manifested in the reaction medium, we have
measured the intrinsic and effective acidities [23] of the catalyst
samples, by determining the amount and strength of the acid sites.
Starting from the determination of the intrinsic acidity of the sam-
ples, determined in cyclohexane (an aprotic and apolar liquid), we
moved to the effective acidity determined in water (a protic and
polar solvent with high solvating ability). We have chosen water
3.2. Activity of xylose dehydration to furfural
Despite the apparent simplicity of the synthesis of furfural from
xylose that consists only in a deep dehydration of the pentose, the
reaction catalyzed by solid acids is far to be simple. Studies from
literature have demonstrated that xylulose is a key intermediate
that is formed from the xylose isomerization, by the catalytic action
of LAS, and then it is dehydrated to furfural by BAS action [10].
However, excess of LAS on the catalyst surface causes side reactions
to produce humins that decrease the furfural selectivity and yield
and cause catalyst deactivation.
Our previous works [23,32] have shown that dispersed niobia
systems on silica are more stable than bulk niobia in the dehydra-
tion of fructose to HMF, but conversion was low due to the inertness
of silica surface. Taking into account these findings, for the xylose
dehydration we have studied acid solid catalysts consisting of dis-
persed niobia phase onto an acidic oxide support. These systems
might guarantee a convenient amount of acid sites with a balanced
concentration of LAS and BAS sites [32,34] for a high xylose con-
version and its selective transformation to furfural, according with
the mechanistic findings of the literature [10].
Table 1
Main textural properties of the studied samples.
Catalyst
BET surface
area (m2 g−1
Pore volume
Average pore
diameter (nm)
)
(cm3 g−1
)
3.2.1. Catalytic tests of xylose dehydration at low temperatures
with different solvents
First of all, all the catalysts were tested in batch experiments
for dehydration of xylose to furfural in different reaction sol-
vents at low temperature (130 ◦C). Besides water, abiphasic system,
SZ
294
217
262
108
1.79
1.21
0.95
0.52
19.7
18.1
13.6
3.4
Nb/SZsg
Nb/SZi
NBOa
a
Niobia in pellets.

94
M.J.C. Molina et al. / Catalysis Today 254 (2015) 90–98
1400
1200
1000
800
600
400
200
0
0.20
0.16
0.12
0.08
0.04
0.00
2.0
1.6
1.2
0.8
0.4
0.0
SZ
0
0.2
0.4
0.6
0.8
1
10
100
1000
P/P0
pore radius / Ǻ
1400
1200
1000
800
0.01
0.008
0.006
0.004
0.002
2.0
NBO
1.6
1.2
0.8
0.4
0.0
600
400
200
0
0
10
100
1000
0
0.2
0.4
P/P0
0.6
0.8
1
pore radius / Ǻ
0.035
2.0
1.6
1.2
0.8
0.4
0.0
1400
1200
1000
800
600
400
200
0
0.03
0.025
0.02
0.015
0.01
Nb/SZsg
0.005
0
0
0.2
0.4
0.6
0.8
1
10
100
1000
pore radius / Ǻ
P/P0
1400
1200
1000
800
600
400
200
0
0.03
0.02
0.01
0.00
2.0
1.6
1.2
0.8
0.4
0.0
Nb/SZi
0
0.2
0.4
0.6
0.8
1
10
100
1000
P/P0
pore radius / Ǻ
Fig. 2. N2 adsorption/desorption isotherms (left side) and BJH pore volume distribution (right side) of all the catalyst samples.
water/CPME, and two aqueous solutions, water/␥-valerolatone and
water/isopropanol, were employed. The goodness of CPME, a highly
hydrophobic ether, as green solvent for carbohydrate biomass con-
version has been already recognized [28]. ␥-Valerolactone, too, is
an interesting solvent as it can be produced from lignocellulose,
it has been also used in dehydration of sugars with interesting
performances [26,27,31,35]. Also aqueous mixtures containing iso-
propanol could improve the stability of the catalyst by improving
solubility of some of the condensation products formed during the
reaction.
Table 2
Summary of the intrinsic and effective acidities of the Nb-catalysts measured by PEA titration in cyclohexane (I.A.) and in water (E.A.), at 30 ◦C.a
Catalyst
Intrinsic acidity (mequiv./g) in cyclohexane
Effective acidity (mequiv./g) in water
Total sites
(mequiv./g)
Weak sites
(mequiv./g)
Strong sites
(mequiv./g)
Strong sites (%)
Total sites
(mequiv./g)
Weak sites
(mequiv./g)
Strong sites
(mequiv./g)
Strong sites
(%)
SZ
0.76
0.47
0.59
0.21
0.40
0.16
0.12
0.07
0.36
0.31
0.47
0.14
47
66
80
66
0.27
0.30
0.22
0.19
0.26
0.25
0.20
0.07
0.005
0.042
0.013
0.12
2
14
6
Nb/SZsg
Nb/SZi
NBO
64
a
Total acid sites (determined from the I run isotherm), weak acid sites (determined from the II run isotherm) and strong acid sites (determined by difference); see
Experimental.

M.J.C. Molina et al. / Catalysis Today 254 (2015) 90–98
95
Fig. 3. Intrinsic (in cyclohexane) and effective (in water) acidities of the samples: adsorption isotherms of PEA at 30 ◦C. I◦ run collected on fresh and activated sample and II◦
run collected on the previously sample saturated by PEA and eluted with solvent.
Fig. 4 shows the results of xylose conversion and furfural yield
water/
isopropanol
obtained for both NBO and the two supported Nb-catalysts (Nb/SZi
and Nb/SZsg). In all the solvents, NBO was more active than the
other supported niobia catalysts; this was expected due to the
lower amount of niobia on the supported catalysts and to the higher
acid site density of NBO in comparison with the diluted Nb/SZ cat-
alysts [23]. However, Nb/SZi and Nb/SZsg show similar catalytic
behavior, disregarding the solvent used, it seems that the method
of preparation does not have a crucial influence in the final cat-
alytic activity but the solvent nature directs both xylose conversion
and yield to furfural obtained. Comparing the results obtained in
the different solvents, both xylose conversion and furfural yield
are higher than those obtained when pure water was used, for
all the catalysts. The biphasic system water/CPME appeared the
best solvent by balancing the results obtained in terms of xylose
conversion and selectivity. It is also worthy to mention that the sup-
ported niobia catalysts show the highest conversion in presence of
␥-valerolactone compared with the other solvents.
water/GVL
water/CPME
water
100
90
80
70
60
50
40
30
20
10
0
100
90
80
70
60
50
40
30
20
10
0
Continuous catalytic tests in a flow fixed bed reactor have been
made aimed to study the stability of the catalysts for long time on
stream; we have chosen to compare water and isopropanol/water
as solvents (Fig. 5). The continuous tests showed that NBO was
more active than the supported niobia catalysts in both the solvents
tested, confirming what has been observed in batch reactor tests.
Fig. 4. Xylose conversion (dashed bars) and furfural yield (filled bars) for the
NBO and Nb/SZ catalysts in different reaction solvents. Reaction conditions:
batch reactor; 4.5 wt.% xylose, 3 wt.% catalyst; 5 g total solution; ␥-valerolactone
(GVL)/aqueous mass ratio, 9;CPME/aqueous mass ratio, 2.33; water/isopropanol 20%
v/v; reaction temperature, 130 ◦C; tos, 360 min.

96
M.J.C. Molina et al. / Catalysis Today 254 (2015) 90–98
100
NbO
Nb/Szsg
90
80
70
60
50
40
30
20
10
0
without catalyst
0
20 40 60 80 100 120 140 160 180 200 220 240
time (min)
Fig. 5. Xylose conversion for different time on stream for the NBO and Nb/SZ cata-
lysts. Reaction conditions: fixed bed reactor; 4.5 wt.% xylose, 3 wt.% catalyst; contact
time, 6–10 min g mL−1; reaction temperature, 130 ◦C.
Fig. 7. Comparison of the xylose conversion (filled markers) and furfural yield
(empty markers) for NBO and Nb/SZsg catalysts and without catalyst (blank test)
vs. reaction time. Reaction conditions: 10 wt.% of xylose and 5 wt.% of catalyst in the
aqueous phase, CPME/aqueous phase mass ratio, 2.33; 50 g total solution; reaction
temperature, 180 ◦C.
Concerning the supported catalysts, even if initially less active, they
are more stable; only a little loss of activity was observed up to
60 h of activity. In addition, the presence of isopropanol in water
improved the activity of the catalyst, even if a certain degree of
activity loss was still present.
3.2.2. Catalytic tests of xylose dehydration at high temperatures
with selected reaction solvents
the ranking of activity observed at low reaction temperature.
Moreover, on all the catalysts xylose conversion and furfural yield
can be described by increasing curves, suggesting that the activity
loss is weak. Moreover, when plotting the selectivity to furfural
against xylose conversion, exponential curves could be observed
on all the catalysts, indicating that under the used experimental
conditions, furfural is a final and stable product of reaction (Figure
4S, Supporting Information). The most interesting results (both
conversion and furfural yield) have been observed in water/CPME
system (Fig. 7). In this solvent, Nb/SZsg had quite complete xylose
conversion while yield to furfural was not higher than 40% after
240 min of reaction, with a poor increasing trend with reaction
time. At this high temperature (180 ◦C), resinification reactions
could be very fast with high amount of by-product formation.
In water/CPME biphasic system, besides catalytic activity, also
the stability of the NBO and Nb/SZ catalysts has been studied by per-
forming recycling experiments at 180 ◦C for various reaction times
(60 min, 120 min, and 240 min of reaction time for NBO and 240 min
for Nb/SZ catalysts). The catalytic runs were carried out for short
periods of time to facilitate the observation of any activity decline.
Fig. 8 shows the xylose conversion and furfural yield for sev-
eral consecutive runs for each catalyst tested. Nb/SZsg and Nb/SZi
have a quite stable production of furfural with yield around 45–50%
and selectivity of almost 80% after 7 runs, while xylose conversion
is roughly constant. This means that the presence of by-products
also deposited on the catalyst surface do not completely deactivate
the catalyst surfaces. On the opposite, a regular deactivate trend
of xylose conversion was observed on NBO, while once again, fur-
fural yield production is more stable. By increasing the reaction
time (recycling tests each 120 min and 240 min), the conversion
and furfural yield increased suggesting that the NBO surface is not
completely deactivated.
On the basis of the results obtained from the tests of xylose dehy-
dration at low temperatures (130 ◦C), emerged that the biphasic
system water/CPME and the monophasic ␥-valerolatone in mix-
ture with water seemed to be the most interesting solvents for the
studied reaction. The key role of the solvent in the conversion of
carbohydrate biomass has been recently pointed out also by Perez
and Fraga [36]. Therefore, all the catalysts were tested again in the
two selected solvents at higher temperatures (160 ◦C for water/␥-
valerolatone and 180 ◦C for water/CPME) and following the reaction
as a function of time.
Fig. 6 shows the results obtained in ␥-valerolactone as solvent
and Fig. 7 those obtained in water/CPME biphasic system, in
terms of xylose conversion and furfural yield, in comparison with
the blank test carried out without catalyst. The results confirm
100
NBO
90
80
70
60
50
40
30
20
10
0
Nb/SZsg
Nb/SZi
without catalyst
These results may suggest that the selective acid sites for the
furfural formation, still lively in water and associated with the
dispersed niobia phase, are not deactivated during the course of
reaction, while the rest of the acid surface is mainly involved in the
unselective conversion of xylose. Concerning the role of the reac-
tion solvent, it clearly appears that water and water solutions can
modulate the catalyst acidity decreasing the acid site density of the
surfaces. This leads to a positive effect on the catalyst stability.
0
20
40
60
80 100 120 140 160 180 200 220 240
time (min)
Fig. 6. Comparison of the xylose conversion (filled markers) and furfural yield
(empty markers) with reaction time on NBO and Nb/SZ catalysts and without cat-
alyst (blank test). Reaction conditions: batch reactor; 2 wt.% xylose and 0.2 wt.% of
catalyst; ␥-valerolactone/aqueous mass ratio, 9; 50 g total solution; reaction tem-
perature,160 ◦C.

M.J.C. Molina et al. / Catalysis Today 254 (2015) 90–98
97
Xylose conversion
The niobia supported catalysts have lower activity than bulk nio-
bia but higher stability during the reaction course, as proved by the
recycling tests and continuous catalytic tests. The catalyst with the
highest niobia dispersion (prepared by sol–gel) showed the most
interesting performances, which can be associated with the highest
amount of strong effective acid sites. Moreover, a crucial role of the
reaction solvent has been enlightened; the wise choice of solvent
prevents a severe catalyst deactivation and maintains the selective
conversion of xylose to furfural. By using a biphasic system, it is
possible extracting the final furfural product from aqueous phase
and avoid side-reactions or/and lowering the formation of humins.
It is also worth noticing that by using aqueous solutions (with alco-
hol or lactone), catalyst stability improves thanks to an increase of
solubility of the formed humins, so avoiding that they can deposit
on the catalyst surface and deactivate it.
Furfural yield
100
90
80
70
60
50
40
30
20
10
0
a)
1
2
3
4
5
6
7
Run
Our future activity will be directed toward the development of
an optimized catalyst plus solvent system, with catalyst acid sites
of adequate nature for isomerize and dehydrate monosaccharides
and with more solubility of the formed by-products in the selected
solvent. These are the two key factors to obtain a successful catalytic
process in the sugar conversion.
Xylose conversion
Furfural yield
100
b)
90
80
70
60
50
40
30
20
10
0
Supplementary information
SEM image of Nb/SZi showing a morphology characterized by
polyhedra of irregular shape; thermogravimetric profiles of the
dried catalysts (Nb/SZ); XRD patterns of fresh Nb2O5; trend of
the selectivity to furfural vs. xylose conversion on the differ-
ent Nb-catalysts, reaction carried out in ␥-valerolactone:water
monophasic system at 160 ◦C in the batch reactor; explicative note
on the properties of 2-phenylethyl amine used as basic probe in the
acidity measurements and on the method of acid–base titration in
liquid.
1
2
3
4
5
6
7
8
9
Run
Xylose conversion
Furfural yield
Acknowledgement
100
90
80
70
60
50
40
30
20
10
0
Tª= 190ºC Tª= 190ºC
Tª= 180ºC
4h
Tª= 180ºC,
2h
Tª= 180ºC,1h
c)
2h
4h
Financial support from the Spanish Ministry of Economy and
Competitiveness is gratefully acknowledged (project CTQ2012-
38204-C03-01 and predoctoral grant for M.J. Campos Molina)
Appendix A. Supplementary data
Supplementary data associated with this article can be found,
in the online version, at http://dx.doi.org/10.1016/j.cattod.2015.
01.018.
References
[1] A.S. Dias, M. Pillinger, A.A. Valente, J. Catal. 229 (2005) 414.
[2] J. Zhang, J. Zhuang, L. Lin, S. Liu, Z. Zhang, Biomass Bioenerg. 39 (2012) 73.
[3] S. Lima, A. Fernandes, M.M. Antunes, M. Pillinger, F. Ribeiro, A.A. Valente, Catal.
Lett. 135 (2010) 41.
[4] X. Wang, S. Cheng, J.C.C. Chan, J. Phys. Chem. C 111 (2007) 2156.
[5] M. López Granados, A.C. Alba-Rubio, I. Sádaba, R. Mariscal, I. Mateos-Aparicio,
A. Heras, Green Chem. 13 (2011) 3203.
[6] I. Sádaba, S. Lima, A.A. Valente, M. López Granados, Carbohydr. Res. 346 (2011)
2785.
[7] I. Sádaba, M. Ojeda, R. Mariscal, R. Richards, M. López Granados, Catal. Today
167 (2011) 77.
0
1
2
3
4
5
6
7
8
9 10 11 12 13 14 15 16
Run
Fig. 8. Recycling tests on NBO and Nb/SZ catalysts: a) Nb/SZsg, b) Nb/SZi, c) NBO.
Reaction conditions: 10 wt.% xylose and 5 wt.% catalyst in the aqueous phase,
CPME/aqueous phase mass ratio, 2.33; 50 g total solution, reaction temperature,
180 ◦C, time of reaction, 240 min for the supported catalysts.
[8] I. Agirrezabal-Telleria, J. Requies, M.B. Güemez, P.L. Arias, Appl. Catal. B 115–116
(2012) 169.
[9] P. Carniti, A. Gervasini, S. Biella, A. Auroux, Catal. Today 118 (2006) 373.
[10] B. Pholjaroen, N. Li, Z. Wang, A. Wang, T. Zhang, J. Energy Chem. 22 (2013) 826.
[11] C. García-Sancho, J.M. Rubio-Caballero, J.M. Mérida-Robles, R. Moreno-Tost, J.
Santamaría-González, P. Maireles-Torres, Catal. Today 234 (2014) 119.
[12] T. Okuara, Chem. Rev. 102 (2002) 3641.
[13] V.S. Braga, I.C.L. Barros, F.A.C. Garcia, S.C.L. Dias, J.A. Dias, Catal. Today 133–135
(2008) 106.
[14] K. Tanabe, Catal. Today 78 (2003) 65.
4. Conclusive remarks
In this work, we have shown that niobia phase dispersed over
an acid oxide support, like silica-zirconia, is effective for the acid
conversion of xylose to furfural, due to the water-tolerant acid prop-
erties of the niobia phase, as proved by our acidity measurements
carried out in water.
[15] I. Novak, M. Ziolek, Chem. Rev. 99 (1999) 3603.

98
M.J.C. Molina et al. / Catalysis Today 254 (2015) 90–98
[16] K. Nakajima, Y. Baba, R. Noma, M. Kitano, J.N. Kondo, S. Hayashi, M. Hara, J. Am.
Chem. Soc. 133 (2011) 4224.
[17] C. Carlini, M. Giuttari, A.M. Raspolli Galletti, G. Sbrana, T. Armaroli, G. Busca,
Appl. Catal. A 183 (1999) 295.
[18] F. Yang, Q. Liu, X. Bai, Y. Du, Bioresour. Technol. 102 (2011) 3424.
[19] S.K.R. Patil, C.R.F. Lund, Energy Fuels 25 (2011) 4745.
[20] S.K.R. Patil, J. Heltzel, C.R.F. Lund, Energy Fuels 26 (2012) 5281.
[21] I. van Zandvoort, Y. Wang, C.B. Rasrendra, E.R.H. van Eck, P.C.A. Bruijnincx, H.J.
Heeres,.B.M. Weckhuysen, ChemSusChem 6 (2013) 1745.
[22] M. Marzo, A. Gervasini, P. Carniti, Catal. Today 192 (2012) 89.
[23] P. Carniti, A. Gervasini, M. Marzo, Catal. Today 152 (2010) 42.
[24] C. García-Sancho, I. Agirrezabal-Telleria, M.B. Güemez, P. Maireles-Torres, Appl.
Catal. B 152–153 (2014) 1.
[26] L. Zhang, H. Yu, P. Wang, Y. Li, Bioresour Technol. 151 (2014) 355.
[27] E.I. Gürbüz, J.M.R. Gallo, D.M. Alonso, S.G. Wettstein, W.Y. Lim, J.A. Dumesic,
Angew. Chem. Int. Ed. 52 (2013) 1270.
[28] M.J. Campos Molina, R. Mariscal, M. Ojeda, M. López Granados, Bioresour. Tech-
nol. 126 (2012) 321.
[29] R. Xing, W. Qi, G.W. Huber, Energy Environ. Sci. 4 (2011) 2193.
[30] R. Weingarten, J. Cho, W.C. Conner Jr., G.W. Huber, Green Chem. 12 (2010) 1423.
[31] D.M. Alonso, S.G. Wettstein, M.A. Mellmer, E.I. Gurbuz, J.A. Dumesic, Energy
Environ. Sci. 6 (2012) 76.
[32] P. Carniti, A. Gervasini, M. Marzo, J. Phys. Chem. C 112 (2008) 14064.
[33] T. Onfroy, G. Clet, M. Houalla, J. Phys. Chem. B 109 (2005) 14588.
[34] A. Gervasini, C. Messi, D. Flahaut, C. Guimon, Appl. Catal. A 367 (2009) 113.
[35] D.M. Alonso, J.M.R. Gallo, M.A. Mellmer, S.G. Wettstein, J.A. Dumesic, Catal. Sci.
Technol. 3 (2012) 927.
[25] C. Sievers, I. Musin, T. Marzialetti, M.B.V. Olarte, P.K. Agrawal, C.W. Jones, Chem-
SusChem 2 (2009) 665.
[36] R.F. Perez, M.A. Fraga, Green. Chem. 16 (2014) 3942.