Catalytic dehydration of xylose to furfural: Vanadyl pyrophosphate as source of active soluble species

Article abstract of DOI:10.1016/j.carres.2011.10.001

The acid-catalysed, aqueous phase dehydration of xylose (a monosaccharide obtainable from hemicelluloses, e.g., xylan) to furfural was investigated using vanadium phosphates (VPO) as catalysts: the precursors, VOPO₄· 2H₂O, VOHPO₄·0.5H₂O and VO(H ₂PO₄)₂, and the materials prepared by calcination of these precursors, that is, γ-VOPO₄, (VO) ₂P₂O₇ and VO(PO₃)₂, respectively. The VPO precursors were completely soluble in the reaction medium. In contrast, the orthorhombic vanadyl pyrophosphate (VO)₂P ₂O₇, prepared by calcination of VOHPO₄· 0.5H₂O at 550 °C/2 h, could be recycled by simply separating the solid acid from the reaction mixture by centrifugation, and no drop in catalytic activity and furfural yields was observed in consecutive 4 h-batch runs (ca. 53% furfural yield, at 170 °C). However, detailed catalytic/characterisation studies revealed that the vanadyl pyrophosphate acts as a source of active water-soluble species in this reaction. For a concentration of (VO) ₂P₂O₇ as low as 5 mM, the catalytic reaction of xylose (ca. 0.67 M xylose in water, and toluene as solvent for the in situ extraction of furfural) gave ca. 56% furfural yield, at 170 °C/6 h reaction.

Full text of DOI:10.1016/j.carres.2011.10.001

Carbohydrate Research 346 (2011) 2785–2791
Contents lists available at SciVerse ScienceDirect
Carbohydrate Research
journal homepage: www.elsevier.com/locate/carres
Catalytic dehydration of xylose to furfural: vanadyl pyrophosphate as source
of active soluble species
Irantzu Sádaba a, Sérgio Lima b, Anabela A. Valente b, , Manuel López Granados a,
⇑
⇑
a Institute of Catalysis and Petrochemsitry (CSIC), C/Marie Curie 2, Campus de Cantoblanco, 28049 Madrid, Spain
b Department of Chemistry, CICECO, University of Aveiro, Campus de Santiago, 3810-193 Aveiro, Portugal
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 21 June 2011
Received in revised form 29 September 2011
Accepted 3 October 2011
Available online 8 October 2011
The acid-catalysed, aqueous phase dehydration of xylose (a monosaccharide obtainable from hemicellu-
loses, e.g., xylan) to furfural was investigated using vanadium phosphates (VPO) as catalysts: the precur-
sors, VOPO4ꢀ2H2O, VOHPO4ꢀ0.5H2O and VO(H2PO4)2, and the materials prepared by calcination of these
precursors, that is, c-VOPO4, (VO)2P2O7 and VO(PO3)2, respectively. The VPO precursors were completely
soluble in the reaction medium. In contrast, the orthorhombic vanadyl pyrophosphate (VO)2P2O7, pre-
pared by calcination of VOHPO4ꢀ0.5H2O at 550 °C/2 h, could be recycled by simply separating the solid
acid from the reaction mixture by centrifugation, and no drop in catalytic activity and furfural yields
was observed in consecutive 4 h-batch runs (ca. 53% furfural yield, at 170 °C). However, detailed cata-
lytic/characterisation studies revealed that the vanadyl pyrophosphate acts as a source of active water-
soluble species in this reaction. For a concentration of (VO)2P2O7 as low as 5 mM, the catalytic reaction
of xylose (ca. 0.67 M xylose in water, and toluene as solvent for the in situ extraction of furfural) gave
ca. 56% furfural yield, at 170 °C/6 h reaction.
Keywords:
Xylose
Furfural
Acid catalysis
Vanadyl pyrophosphate
Homogeneous catalysis
Ó 2011 Elsevier Ltd. All rights reserved.
1. Introduction
supported acids,8–10 Keggin-type heteropolyacids,11 titanates and
niobates12 and metal oxides.13 These catalysts are solids and may
The depletion of fossil resources is promoting the search of
alternative sources of energy and chemicals. Lignocellulose is
appearing as a very attractive raw material, because it is abundant
and inexpensive. In the context of the lignocellulosic-based biore-
fineries, furfural is acquiring increasing importance as one of the
most outstanding building blocks.1 It is one of the compounds in
the ‘top 10+4’ revised list of bio-based product opportunities from
carbohydrates.2 Furfural is obtained via dehydration of xylose,
which in turn is obtained from the hemicelluloses (polysaccha-
rides) component of plant biomass (Scheme 1).3 This reaction is
industrially carried out in water as solvent, using liquid acids as
catalysts such as, sulfuric acid which causes corrosion hazards,
the formation of sulfur containing by-product, and enhanced waste
production (e.g. from neutralisation processes). The development
of less hazardous, ‘greener’ catalysts to perform this reaction in
water is thus of important relevance to avoid the above mentioned
drawbacks.
therefore be easier and safer to handle in relation to liquid acids. Car-
lini et al. reported very interesting results for vanadium phosphate
oxides (hereinafter referred to as VPO) used as acid catalysts in a
similar reaction, that is, the dehydration of fructose to 5-hydroxy-
methyl-furfural (HMF),14 using solely water as solvent, under rela-
tively moderate reaction conditions. Recently, Wang et al. used
different types of VPO acid catalysts in the dehydration of glycerol
to acrolein.15,16 To the best of our knowledge, VPO catalysts have
not been investigated in the dehydration of xylose to furfural.
The crystalline structure and the surface morphology of VPO
materials are dependent upon the preparation method of the
precursor and the conditions of the calcination treatment (time,
temperature, atmosphere).16–19 Among the precursors the most
common phases are those with a P/V atomic ratio of 1, namely, van-
adyl phosphate dihydrate (VOPO4ꢀ2H2O) and vanadyl hydrogen
phosphate hemihydrate (VOHPO4ꢀ0.5H2O). When VOHPO4ꢀ0.5H2O
is calcined it may undergo a topotactic dehydration to form the
vanadyl pyrophosphate (VO)2P2O7, which is the typical industrial
catalyst for the production of maleic anhydride from n-butane.17
The thermal treatment of VOPO4ꢀ2H2O leads to different anhydrous
Several solid acids have been investigated as catalysts in the
batchwise conversion of xylose into furfural such as, microporous
zeolites and related materials,4–7 micro-mesoporous silica-
phases like
aI-VOPO4, aII-VOPO4, b-VOPO4 and c-VOPO4 among
others.17 These materials have been widely studied as redox cata-
lysts,20,21 and to a much smaller extent as acid catalysts, even though
VPO solids are known to possess acid properties.19,22 Kamiya et al.
reported that the pyrophosphate phase presents both Brønsted
⇑
Corresponding authors. Tel.: +351 234370603; fax: +351 234401470 (A.A.V.);
tel.: +34 915854937; fax: +34 915854760 (M.L.G.).
E-mail addresses: atav@ua.pt (A.A. Valente), mlgranados@icp.csic.es (M. López
Granados).
0008-6215/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.carres.2011.10.001

2786
I. Sádaba et al. / Carbohydrate Research 346 (2011) 2785–2791
temperature and time applied to the VP2-p precursor, was investi-
gated. These samples are denoted as VP2-X-Yh, where X refers to
the calcination temperature (°C) and Y to the calcination time
(h). VO(H2PO4)2 (VP3-p) was calcined at 500 °C in N2 flow (same
conditions as for VP2) during 16 h to obtain a solid denoted VP3-
c (VO(PO3)2).29
Scheme 1. Simplified reaction scheme of the dehydration of xylose to furfural.
2.2. Catalyst characterisation
and Lewis sites, and the acidic properties depended on the adopted
preparation method.23
In the present work, the dehydration of D-xylose to furfural is
investigated in the presence of vanadium phosphates as catalysts:
Powder XRD data were collected at room temperature on a
Philips X’pert Pro PANalytical diffractometer with Cu-Ka1
(k = 0.154046 nm) and Cu-Ka2 (k = 0.154443 nm) radiations (Ka1/
Ka
2 = 0.5), in a Bragg–Brentano para-focusing optics configuration
VOPO4ꢀ2H2O, VOHPO4ꢀ0.5H2O and VO(H2PO4)2, and of the materials
(40 kV, 50 mA). Samples were step-scanned in 0.02° steps with a
counting time of 50 s per step. Diffractograms were analysed with
the X’Pert HighScore Plus software. N2 adsorption–desorption iso-
therms were recorded at ꢁ196 °C using a Micromeritics TRISTAR
3000 apparatus; the samples were degassed at 120 °C for 12 h prior
analysis. Thermogravimetric (TGA) and differential scanning calo-
rimetry (DSC) analyses were performed with Shimadzu TGA-50
and DSC-50 systems (heating rate of 10 °C minꢁ1 under air), respec-
tively. An inductively coupled plasma mass spectrometer (ICP-MS)
Elan 6000 Perkin-Elmer Sciex equipped with an AS 91 autosampler,
was used for quantification of V and P.
obtained by calcination of these precursors,
c-VOPO4, (VO)2P2O7
and VO(PO3)2, respectively. Special attention has been drawn to
the assessment of the homo/heterogeneous catalytic contributions
for the overall reaction of xylose; this issue was not previously
addressed for VPO catalysts used in the dehydration of fructose
to HMF.14 Detailed characterisation and catalytic studies were car-
ried out for the orthorhombic vanadyl pyrophosphate (VO)2P2O7,
prepared by calcination of VOHPO4ꢀ0.5H2O at 550 °C (most promis-
ing catalyst, based on catalytic screening tests).
2. Experimental
Diffuse Reflectance Infrared Fourier Transform Spectroscopy
(DRIFTS) of the NH3 adsorption was carried out to evaluate the acid
properties of the solids. Spectra were collected on a Nicolet 5700 FT
spectrophotometer equipped with a high sensitivity Hg–Cd–Te
detector and a Praying Mantis (Harrick Scientific Co.) optical acces-
sory to focus the IR beam on the sample. A high temperature in situ
chamber (Harrick Scientific Co.) with Zn Se windows was used for
thermal treatment of the samples before recording the DRIFT spec-
tra. Spectra were obtained at a resolution of 4 cmꢁ1 with an accu-
mulation of 256 scans and presented in Kubelka-Munk mode.
Typically, the finely ground samples (ca. 50 mg) were placed in
the cup of the in situ DRIFT chamber. The calcined sample was pre-
treated under Ar flow (ca. 50 mL minꢁ1, Air Liquide) at 150 °C for
2.1. Catalyst preparation
Vanadyl phosphate dihydrate (VOPO4ꢀ2H2O, denoted VP1-p)
was synthesised in water according to the procedure described
by Johnson et al.24 Vanadium (V) pentoxide, V2O5 (5.0 g, 99.6%
from Aldrich) was mixed with aqueous 85% ortho-phosphoric acid
(30.0 mL, from Sigma-Aldrich) and deionised H2O (120 mL). The
mixture was refluxed while stirring for 24 h and a homogeneous
yellow precipitate was obtained. The resulting precipitate was sep-
arated by filtration, and washed with hot distilled water (100 mL)
and acetone (100 mL).25 The solid was dried at 110 °C in an oven
overnight, obtaining a greenish yellow powder.
60 min and cooled down at 30 °C. Then, NH3 (50 mL minꢁ1
,
Vanadyl hydrogen phosphate hemihydrate (VOHPO4ꢀ0.5H2O,
denoted VP2-p) was prepared by suspending V2O5 (15 g) in
90 mL of isobutyl alcohol (99%, Sigma-Aldrich) and 60 mL of benzyl
alcohol (99%, Sigma-Aldrich). The suspension was stirred continu-
ously under reflux for 3 h, cooled to room temperature, followed by
the additions of o-H3PO4 (16.2 g, 99% from Sigma-Aldrich). The
slurry was heated under reflux with constant stirring for 2 h, and
subsequently cooled, filtered, washed with isobutyl alcohol and
dried at 110 °C overnight to give a grey-blue powdered solid.26
The precursor VO(H2PO4)2 (denoted VP3-p) was synthesised
adapting the method proposed by Hannour et al.27 V2O5 (18 g),
H3PO4 (58 g) and water (400 ml) were mixed and heated in the
presence of an excess of oxalic acid (10.1 g) at 90 °C for 48 h. The
resulting blue solution was placed in a crystallizer and evaporated
overnight. The resultant powdered solid was washed with acetone
and dried.
Different calcination procedures were applied to the VP1-p,
VP2-p and VP3-p precursors. The calcinations were conducted un-
der static air in a muffle furnace or in N2 flow using a Pyrex glass
tube (internal diameter of 20 mm). The calcination conditions are
summarised in Table 1. Following a procedure described by Zhu
et al.,22 VP1-c was prepared by calcination of VOPO4ꢀ2H2O (VP1-
p), at 750 °C for 16 h, under air atmosphere to fully dehydrate
the precursor. With the aim of preparing vanadyl pyrophosphate
((VO)2P2O7), 0.5 g of VP2-p powdered solid was heated at 550 °C
(10 °C minꢁ1) in N2 flow (60 mL minꢁl) during 2 h (giving a solid
denoted VP2-550-2h).28 Since VP2-550-2h gave promising cata-
lytic results (discussed ahead), the influence of the calcination
900 ppm in Ar, Air Liquide) was passed through the sample for
30 min at that temperature. The sample was subsequently evacu-
ated under Ar flow for 30 min at 30, 100, 200, 300 and 400 °C be-
fore recording the IR spectra. For each temperature, the spectra
of the sample without exposure to NH3 were recorded and sub-
tracted from the spectrum of the same sample exposed to NH3.
2.3. Catalytic reactions
Batch catalytic experiments were performed in glass micro-
reactors equipped with magnetic stirring bars. The micro-reactors
were always purged with nitrogen prior to heating. In a typical pro-
cedure, 30 mg of D-xylose (SigmaUltra, >99%), 20 mg of catalyst and
H2O (0.3 mL, Milli-Q) and toluene (0.7 mL, Chromasolv Plus,
>99.9%) were poured into the reactor. The reaction mixture was
heated with a thermostatically controlled oil bath and stirred mag-
netically at 800 rpm. Zero time was taken when the micro-reactor
was immersed in the oil bath. In the catalyst recycling tests, the
solid was separated from the reaction mixture by centrifugation
at the selected times, washed and sonicated for 5 min in water,
methanol and acetone, dried at 50 °C overnight (denoted W-
treatment). For comparison, the washed solid was subsequently
calcined at 400 °C (heating rate of 10 °C minꢁ1, static air) for 2 h
(denoted WC-treatment). The analysis of D-xylose and furfural in
the aqueous phase was quantitatively determined using an HPLC
equipped with a Knauer K-1001 HPLC pump and a PL Hi-Plex H
300 ꢂ 7.7 (i.d.) mm ion-exchange column (Polymer Laboratories
Ltd., UK) coupled to a Knauer K-2401 differential refractive index

I. Sádaba et al. / Carbohydrate Research 346 (2011) 2785–2791
2787
Table 1
Synthetic conditions and physical properties of the prepared VPO solids
a
Sample
Crystalline Phase
Synthetic precursor/calcination conditions
Colour
SBET (m2 gꢁ1
)
VP1-p
VP2-p
VP3-p
VP1-c
VP2-550-2h
VP3-c
VOPO4ꢀ2H2Ob
VOHPO4ꢀ0.5H2O
VO(H2PO4)2
—
—
—
Green
0.3
36.2
0.2
Blue-grey
Turquoise
Yellow
Dark brown
Turquoise
c
-VOPO4
VP1/750 °C, 16 h, air
VP2/550 °C, 2 h, N2
VP3/500 °C, 16 h, N2
5.1
c
(VO)2P2O7
37.2
VO(PO3)2
Nme
d
a
b
c
Brunauer–Emmett–Teller (BET) specific surface area.
Contains (VO)(VO2)2H4(PO4)2(P2O7)0.5ꢀH2O.
Contains VO(H2PO4)2 impurity.
Contains V(PO3)3 impurity.
d
e
Not measurable.
detector (for xylose) and a Knauer K-2600 UV detector (280 nm, for
furfural). The mobile phase was water acidified to pH 3.05 with
H2SO4. The flow rate was 0.6 mL minꢁ1 and the column tempera-
ture was maintained at 65 °C. The furfural present in the organic
phase was quantified with a Gilson 306 HPLC pump and a Spheri-
sorb ODS S10 C18 column, coupled to a Gilson 118 UV/Vis detector
(280 nm). The mobile phase consisted of 40% v/v methanol in water
(flow rate 0.7 mL minꢁ1).
The reaction of xylose was investigated using the VPO precur-
sors as catalysts, which were completely soluble in the reaction
medium. Carlini and co-workers reported interesting results when
using modified vanadyl phosphate dihydrate (VOPO4ꢀ2H2O) as cat-
alyst in the dehydration of fructose to HMF, at 80 °C, using solely
water as solvent14: 79% HMF selectivity at 50% conversion within
1 h of reaction. Under these conditions, the reaction of xylose in
the presence of VP1-p (VOPO4ꢀ2H2O) does not take place. Possibly,
more demanding acid conditions are required for the dehydration
of xylose (aldose, pentose) than those for fructose (ketose, hexose).
The reaction of xylose in the presence of VP2-p, at 170 °C, using a
biphasic water–toluene solvent system, (which is favourable for
enhancing furfural selectivity, as discussed ahead), is quite fast giv-
ing 100% conversion in less than 2 h reaction, albeit the selectivity
towards furfural being always less than 50%. The VP3-p precursor
was poorly active in this reaction.
3. Results and discussion
3.1. Characterisation and screening catalytic tests of the VPO
precursors
The colours of the VPO precursors (VP1-p, VP2-p and VP3-p) are
given in Table 1, and are indicative of the oxidation state of the vana-
dium (blue and yellow is characteristic of V4+ and V5+ ions, respec-
3.2. Characterisation and screening catalytic tests of the
calcined solids
tively). The powder XRD pattern of VP1-p shows
a well-
crystallised dihydrate phase VOPO4ꢀ2H2O (1, PDF Reference code
01-084-0111) containing (VO)(VO2)2H4(PO4)2(P2O7)0.5ꢀH2O (PDF
Reference code 00-047-0967) as an impurity (Fig. 1). The VP2-p solid
consists of the hemihydrate phase VOPO4ꢀ2H2O (2, PDF Reference
code 00-038-0291), and the VP3-p pattern is assigned to the tetrag-
onal VO(H2PO4)2 phase (3, PDF Reference code 00-047-0953). In
general, the BET specific surface areas of the samples are quite
low, and the highest value was obtained for VP2-p (36 m2 gꢁ1),
which parallels that observed for the respective calcined solids (dis-
cussed ahead), Table 1; these results are in agreement with the liter-
ature data.20,23,30
The calcination of the VPO precursors (VP1-p, VP2-p and VP3-p)
under appropriate conditions gave the solids VP1-c, VP2-550-2h
and VP3-c, respectively. According to the powder XRD data, the
VP1-c solid consists of a monoclinic c-VOPO4 phase, according to
Bordes17 (1, PDF Reference code 00-047-0950), Figure 2. Although
the applied calcination treatment was similar to that reported pre-
viously by Zhu et al.22 to obtain tetragonal
aII-VOPO4, the latter
phase was not detected in the case of VP1-c. The VP2-550-2h solid
consists of an orthorhombic phase of vanadyl pyrophosphate
Figure 1. Powder X-ray diffraction patterns of the precursors. The data labels
correspond to the following crystalline phases: 1—VOPO4ꢀ2H2O, 2—VOH-
PO4ꢀ0.5H2O, 3—VO(H2PO4)2, 4—(VO)(VO2)2H4(PO4)2(P2O7)0.5ꢀH2O.
Figure 2. Powder X-ray diffraction patterns of the calcined solids. The data labels
correspond to the following crystalline phases: 1—
VO(H2PO4)2, 4—VO(PO3)2, 5—V(PO3)3.
c-VOPO4, 2—(VO)2P2O7, 3—

2788
I. Sádaba et al. / Carbohydrate Research 346 (2011) 2785–2791
Table 2
(VO)2P2O7 (2, PDF Reference code 00-041-0698) containing VO(H2-
PO4)2 as a minor impurity (3, PDF Reference code 00-047-0953).
The XRD pattern of this solid matches that reported by other
authors using the same calcination conditions.16 The calcination
of VP3-p produced a well crystallised VO(PO3)2 phase (4, PDF Ref-
erence code 00-033-1433), containing V(PO3)3 as a minor impurity
(5, PDF Reference code 00-033-1442). In general, the BET specific
surface areas of the samples are quite low, and the highest value
was obtained for VP2-550-2h (ca. 37 m2 gꢁ1), which is similar to
that observed for its precursor VP2-p (Table 1).
Crystallite sizes of the calcined solids
Catalyst
Crystallite sizea (nm)
020b
204c
VP2-550-2h
VP2-550-2h-used
VP2-550-8h
VP2-550-16h
VP2-750-2h
4.5
4.9
6.9
8.2
11.6
16.7
16.2
17.5
21.7
19.5
a
Calculated by the Debye–Scherrer equation.
Crystallite sizes calculated using the full width at half-maximum of (020)
b
Based on preliminary catalytic tests of the investigated materi-
als (discussed ahead), the VP-2, which was apparently insoluble in
the reaction medium, seemed a quite promising catalyst. Hence,
the effect of the calcination conditions on the crystalline structure
and the surface acid properties of VP-2 were further investigated.
The calcination time and temperature of VP2-p caused structural
modifications. The colours of the VP2-550-Yh samples are differ-
ent: dark brown, grey and greenish for calcination time of 2, 8
and 16 h, respectively. For longer calcination time (16 h) new
reflections appeared at ca. 15° 2h, assigned to the monoclinic phase
of vanadyl pyrophosphate (2, PDF Reference code 01-085-2281),
Figure 3. In general, these solids are nanocrystalline and the crys-
tallite size tends to increase slightly with increasing calcination
time at 550 °C (Table 2); concomitant narrowing of the peaks at
2h = 22.6° and 28.2° is observed. Apparently the crystallinity in-
creases faster on the perpendicular direction of the (020) planes.
Calcination at a higher temperature of 750 °C (N2 atmosphere) re-
sulted in a less crystalline orthorhombic (VO)2P2O7 phase and a
new phase appeared as orthorhombic VPO4, indicating the partial
reduction of vanadium, under these conditions.
reflection at ca. 22.6°.
Crystallite sizes calculated using the full width at half-maximum of (204)
reflection at ca. 28.2°.
c
The effect of the calcination treatment on the acid properties of
the VP2-derived catalysts was investigated by DRIFT analysis of
NH3 adsorbed on Lewis (LAS) and Brønsted (BAS) acid sites (bands
appearing in the region 2000–1200 cmꢁ1). The NH3 molecule chem-
isorbed on LAS displays a das (H–N–H) band at ca. 1650 cmꢁ1
,
+
whereas NH4 formed over BAS shows another band at ca.
1425 cmꢁ1, due to das (H–N–H).31–34 According to the literature,
the ratio of the absorption coefficients of the bands 1425 and
1650 cmꢁ1 (NH4+/NH3) is close to 7, which means that the band as-
signed to LAS is usually much weaker.35 Busca et al.36 assigned the
BAS to surface P–OH groups and the LAS to coordinatively
unsaturated V(IV) ions exposed on the surface. Figure 4 shows the
+
das (H–N–H) bands of NH3 and NH4 adsorbed on the catalysts
Figure 4. DRIFT spectra of the NH3 adsorption over (top) VP2-550-2h and (bottom)
VP2-750-2h for the indicated outgassing temperatures.
VP2-550-2h and VP2-750-2h, after different outgassing tempera-
tures. The most intense band was the one at 1425 cmꢁ1 indicating
that in these catalysts BAS predominate over LAS. The band assigned
to the asymmetric deformation modes of chemisorbed NH3 at ca.
1650 cmꢁ1 was vaguely detected in the case of the VP2-550-2h. Pos-
sibly, the activation temperature used in this experiment (150 °C)
was not high enough to completely remove the chemisorbed water
and form the LAS. Negative bands in the 3600–3500 cmꢁ1 region
(not shown) were due to the disappearance of the surface hydroxyl
groups.
The band assigned to BAS becomes weaker with increasing out-
gassing temperature (Fig. 4). In the case of VP2-550-2h, the presence
of a strong band at 1425 cmꢁ1 even after evacuation at 400 °C, which
was not observed for VP2-750-2h, indicates that the BAS of VP2-
Figure 3. Powder XRD patterns of the vanadyl pyrophosphate materials prepared
using different calcination temperatures and times and of VP2-550-2h after the
catalytic reaction (VP2-550-2h used). 1—orthorhombic (VO)2P2O7, 2—monoclinic
(VO)2P2O7, 3—VO(H2PO4)2, 4—VPO4.

I. Sádaba et al. / Carbohydrate Research 346 (2011) 2785–2791
2789
Table 3
the polar surface of the catalyst becomes less favourable than
without the organic co-solvent, avoiding its degradation by acid
species. A comparison of the results in Table 4 (entries 4 and 5)
for the reaction of xylose at 170 °C in the presence of VP2-550-
2h shows the beneficial effect of using toluene as a co-solvent on
furfural yield at 4 h (ca. two-fold increase).
Integrated area (IA) of the infrared band at 1425 cmꢁ1 (assigned to BAS, Kubelka-
Munk arbitrary units) as a function of the outgassing temperature of the VP2-X-Yh
solids
Temperature (°C)
VP2-550-2h
VP2-550-16h
VP2-750-2h
30
100
200
300
400
1499
1044
675
539
519
1376
1170
590
585
437
989
715
271
252
—
The effect of calcination time and temperature on the catalytic
performance of VP2-p was investigated in the reaction of xylose
at 170 °C, under biphasic solvent conditions. In all the cases, at
least 90% conversion was reached within 6 h of reaction (Fig. 5).
The precursor VP2-p was much more active than the VP2-X-Yh
counterparts (Fig. 5a), but the selectivity to furfural was much low-
er (Fig. 5b). The kinetic profiles and the curves of the dependency
of furfural selectivity on conversion are fairly coincident for all
the VP2-550-Xh samples, indicating that the influence calcination
time on the catalytic performance was not significant. In principle,
these results seem to correlate very well with the similar acid
properties of these catalysts (Table 3). In comparison to VP2-550-
2h, the VP2-750-2h sample possesses lower catalytic activity
(60% compared to 75% conversion at 2 h for VP2-550-2h, respec-
tively, Fig. 5a). Despite the higher reaction times required to reach
ca. 90% conversion in the case of VP2-750-2 h, somewhat higher
furfural selectivity was reached (Fig. 5b). These results may be
due to the weaker acidity of this latter sample. It is interesting to
stress that furfural selectivity increases with conversion possibly
because the primary step of conversion of xylose is faster than
one of the subsequent steps, leading to relatively slow increase
in concentration of furfural in the bulk (Fig. 5b).6
The stability of the VP2-550-2h catalyst was investigated by per-
forming consecutive 4 h-batch runs for the reaction of xylose at
170 °C, under biphasic solvent conditions. Prior to catalyst recycling,
the solid was separated from the reaction medium by centrifugation,
followed by a washing treatment (denoted W-treatment) or
washing plus calcination treatment (denoted WC-treatment);
experimental details are given in Section 2.3. The conversions of
xylose were somewhat comparable in the three consecutive runs
(90%, 80% and 86% for runs 1, 2 and 3, respectively). Figure 6 shows
that furfural yields in the first couple of runs are comparable for the
two catalyst treatments and remains fairly constant in the third run
by simply using the W-treatment. These results contrast with those
reported in the literature for some families of solid acid catalysts
investigated in the reaction of xylose, which required thermal treat-
ments for the recovery of the initial catalytic performance.6–8,10,38
These contrasting results may be due to differences in the homo/het-
erogeneous nature of the catalytic reaction (discussed ahead).
The powder XRD patterns of the fresh and washed catalysts
(Fig. 3) were quite similar and the crystallite sizes are comparable
(Table 2), suggesting that the structure of the solid can be relatively
stable under the applied hydrothermal conditions. Elemental anal-
ysis revealed that the VP2-550-2h solid recovered by the W-treat-
ment contained 7 wt % of carbon. A comparative study of the TGA
and DSC analyses (Fig. 7) for the fresh and used catalysts shows
that the used catalyst undergoes a different exothermic process
in the temperature range 250–400 °C, which may be related to
550-2h are stronger. Table 3 shows the integrated areas (IAs) of the
band at 1425 cmꢁ1 for the spectra given in Figure 4, and for the VP2-
550-16h sample. The IAs of the bands of VP2-550-2h and VP2-550-
16h samples were quite similar for all outgassing temperatures,
suggesting that calcination time at 550 °C does not affect signifi-
cantly the acid properties. The VP2-750-2h sample possesses a
lower amount of acid sites (based on the lower values of IAs) than
the VP2-550-Xh solids, which can be related to the comparatively
lower crystallinity of this sample and/or to the presence of the
VPO4 impurity phase (Fig. 3).
The catalytic performance of the calcined solids was investi-
gated preliminarily using solely water as solvent, at 170 °C (Table
4, entries 1, 4 and 6). For all materials, excluding VP3-c, the conver-
sion of xylose at 4 h reaction is quite high (70–80%), but the furfu-
ral yields are rather low (<25%). The liquid phase obtained after the
catalytic tests was coloured for the VP1-c (colour changed in-
stantly) and VP3-c solids, indicating the dissolution of metal spe-
cies. The same was not observed for VP2-c, which gave the
highest furfural yield. Hence, the reaction of xylose was further
investigated in the presence of VP2-c.
3.3. Reaction of xylose in the presence of VP2-X-Yh solid acids
In the case of VP2-550-2h as catalyst, decreasing the tempera-
ture of the reaction of xylose from 170 to 140 °C leads to a consid-
erable decrease in the conversion and the furfural yield, which are
even lower at 120 °C (Table 4, entries 2–4). The mechanism of the
dehydration of xylose may take place as an acid-catalysed series of
elementary steps proceeding through a 2,5-anhydride intermedi-
ate.37 In the present work, by-products were not detected by HPLC
analysis. However, according to the literature, furfural can further
react with itself, ‘furfural resinification’, or with intermediates of
the pentose-to-furfural conversion, ‘furfural condensation’, leading
to higher molecular weight products.3 On the other hand, fragmen-
tation reactions of xylose can take place in parallel to furfural
formation.37
In order to improve the selectivity to furfural, toluene was used
as a co-solvent (forming a biphasic solvent system) for extracting
furfural from the aqueous phase. Since the catalysts’ surface is po-
lar (e.g. possess surface P–OH groups), the solid particles dispersed
preferably in the aqueous phase. Once furfural is formed and trans-
ferred into the organic phase one may expect that its adsorption on
Table 4
Catalytic results for the calcined VPO catalystsa
Entry
Sample
Temperature (°C)
Xylose conversion (%)
Furfural yield (%)
Furfural selectivity (%)
1
2
3
4
5
6
VP1-c
170
120
140
170
170
170
71
21
33
80
91
13.4
22
7
8
24
53
3.8
31
33
23
29
58
28
VP2-550-2h
VP2-550-2h
VP2-550-2h
VP2-550-2hb
VP3-c
a
Reaction conditions: 30 mg of xylose, substrate/catalyst mass ratio = 1.5, 1 mL H2O, 4 h.
1 mL water–toluene solvent mixture (3:7 volume ratio).
b

2790
I. Sádaba et al. / Carbohydrate Research 346 (2011) 2785–2791
Figure 7. TGA and DSC curves of fresh (solid line) and used (dashes) VP2-550-2h
catalyst. Reaction conditions as in Figure 5.
the combustion of organic matter (a small endothermic band be-
low 120 °C is associated to the desorption of physisorbed water/
volatiles). Comparing the mass losses between the fresh and used
catalysts, one may deduce that the amount of organic matter is
ca. 9 wt % of the VP2-550-2h solid, which is consistent with the
elemental analysis result. The amount of organic matter in the
recovered solid is significantly lower than that reported in the lit-
erature for other solid acids tested as catalysts in the same reac-
tion, under similar conditions in which xylose conversion
reached at least 90%.7 Apparently, the hydrothermal stability of
the (VO)2P2O7 phase and the small amount of organic deposits
would explain the steadiness of the catalytic properties in recy-
cling runs. However ICP-MS analyses of the aqueous phase after
the reaction showed that the V and P concentrations in the solution
after the first run were 527 and 320 ppm respectively. These are
minor amounts and are equivalent to ca. 5 mmolðVOÞ
2.3 mol % of the initial catalyst loading of 217 mmolðVOÞ
der typical reaction conditions).
/L, or ca.
/L (un-
2P2O7
2P2O7
Figure 5. Kinetic profiles of the xylose conversion (a) and dependence of furfural
selectivity on xylose conversion (b), for the VP2-p precursor and the respective
calcined catalysts. Reaction conditions: 30 mg of xylose, substrate/catalyst mass
ratio = 1.5, 170 °C, 1 mL water–toluene mixture (3:7 volume ratio).
In order to get insight on the homogeneous contribution of the
solubilised species, a separate experiment was performed in which
the catalyst was put into contact with water (without xylose) at
170 °C, during 4 h; subsequently the reaction mixture was cooled
to room temperature, the solid acid was filtered through a 0.2 lm
PVDF w/GMF Whatman membrane, and xylose and toluene were
added to the resulting aqueous phase, which was then heated to
170 °C. After 4 h reaction, the xylose conversion and furfural yield
were similar to those observed for the catalytic test using 20 mg of
solid catalyst, indicating the existence of a homogeneous catalytic
contribution. In order to assess the extent of the homogeneous cat-
alytic contribution, a separate catalytic test was carried out using
VP2-550-2h in an amount equivalent to that measured by ICP-MS
(5 mM (VO)2P2O7): the conversion and furfural yields at 2 h/6 h
reaction were similar (differences within the experimental range
of order of ca. 10%) to those observed for the catalytic reaction using
20 mg of VP2-550-2h catalyst (equivalent to 217 mM (VO)2P2O7 in
the aqueous phase). These results suggest that the catalytic reaction
takes place essentially in the homogeneous phase and that the VP2-
550-2h solid acts as a supplier of very active soluble species. These
results can partly explain (i) the ‘steady’ catalytic performance of
VP2-550-2h in consecutive batch runs since an excess amount of
catalyst was used (Fig. 6), and (ii) the fact that the type of treatment
(W or WC) applied to the separated solid acid, on the catalytic reac-
tion of xylose, was minor.
Figure 6. Furfural yields in consecutive 4 h-batch runs, at 170 °C, using VP2-550-2h
as catalyst, and applying the W-treatment (runs 2 and 3 washed) or the WC-
treatment (run 2 calcined) to the separated solid prior to its reuse. Reaction
conditions as in Figure 5.
For comparative purposes, the water-soluble salts NaHPO4
(10 mM), VOSO4 (10 mM), Na4P2O7 (5 mM), and a mixture of
VOSO4 (10 mM) and Na4P2O7 (5 mM) were tested as homogeneous

I. Sádaba et al. / Carbohydrate Research 346 (2011) 2785–2791
2791
containing the aqueous phase with the soluble catalyst for initial-
ising a subsequent run.
Acknowledgements
This work was partially funded by the Spanish Ministry of
Science and Innovation (MICINN: ENE-2009-12743-C04-01), ‘Con-
sejería de Educación’ of the Autonomous Government of Madrid
(CAM: CARDENER-CM) and the Portuguese Fundação para a Ciên-
cia e a Tecnologia (FCT), POCI and FEDER (project POCI/QUI/
56112/2004). I. Sádaba is grateful to the Spanish National Research
Council (CSIC) for a JAE-Predoc and for the financial support for the
stay at CICECO, University of Aveiro. S. Lima is grateful to the FCT
for post-doctoral grant. The authors wish to express their gratitude
to Dr. F. Domingues (Department of Chemistry, University of
Aveiro) for access to HPLC equipment and to Doctor Martyn
Pillinger (CICECO, University of Aveiro) for helpful discussions.
Figure 8. Furfural yield versus reaction time, using as catalysts the (soluble) salts,
NaHPO4 (10 mM), VOSO4 (10 mM), Na4P2O7 (5 mM) or a mixture of VOSO4 (10 mM)
and Na4P2O7 (5 mM); the data for VP2-550-2h are given for comparison. Reaction
conditions: 30 mg of xylose, 170 °C, 1 mL water–toluene mixture (3:7 volume
ratio).
References
1. Biorefineries—Industrial Processes and Products Status Quo and Future Directions;
Kamm, B., Gruber, P. R., Kamm, M., Eds.; Wiley-VCH: Weinheim, 2005.
2. Bozell, J. J.; Petersen, G. R. Green Chem. 2010, 12, 539–554.
3. Zeitsch, K. J. The Chemistry and Technology of Furfural and Its Many By-
Products, 1st ed. In Sugar Series; Elsevier: The Netherlands, 2000; Vol. 13,.
4. Moreau, C.; Durand, R.; Peyron, D.; Duhamet, J.; Rivalier, P. Ind. Crops Prod.
1998, 7, 95–99.
5. O’Neill, R.; Ahmad, M.; Vanoye, L.; Aiouache, F. Ind. Eng. Chem. Res. 2009, 48,
4300–4306.
6. Lima, S.; Fernandes, A.; Antunes, M. M.; Pillinger, M.; Ribeiro, F.; Valente, A. A.
Catal. Lett. 2010, 135, 41–47.
7. Lima, S.; Antunes, M. M.; Fernandes, A.; Pillinger, M.; Ribeiro, F.; Valente, A. A.
Appl. Catal., A 2010, 388, 141–148.
8. Dias, A. S.; Pillinger, M.; Valente, A. A. J. Catal. 2005, 229, 414–423.
9. Dias, A. S.; Pillinger, M.; Valente, A. A. Microporous Mesoporous Mater. 2006, 94,
214–225.
catalysts (used in amounts equivalent to the approximate leached
amounts of V and P observed for VP2-550-2h), in the reaction of
xylose, at 170 °C. The furfural yield reached at 6 h reaction follows
the order, (Na4P2O7+VOSO4) (17%) < Na4P2O7 (24%) < NaHPO4
(34%) < VOSO4 (42%) < VP2-550-2h (56%), Figure 8. Based on these
results, the catalytic performance of the VP2-550-2h catalyst is dif-
ferent from those of the tested salts (containing similar ions), lead-
ing to higher furfural yields, under similar reaction conditions.
10. Shi, X.; Wu, Y.; Li, P.; Yi, H.; Yang, M.; Wang, G. Carbohydr. Res. 2011, 346, 480–
487.
4. Conclusions
11. Dias, A. S.; Pillinger, M.; Valente, A. A. Appl. Catal., A 2005, 285, 126–131.
12. Dias, A. S.; Lima, S.; Carriazo, D.; Rives, V.; Pillinger, M.; Valente, A. A. J. Catal.
2006, 244, 230–237.
13. Chareonlimkun, A.; Champreda, V.; Shotipruk, A.; Laosiripojana, N. Bioresour.
Technol. 2010, 101, 4179–4186.
14. Carlini, C.; Patrono, P.; Galletti, A. M. R.; Sbrana, G. Appl. Catal., A 2004, 275,
111–118.
15. Wang, F.; Dubois, J.-L.; Ueda, W. J. Catal. 2009, 268, 260–267.
16. Wang, F.; Dubois, J.-L.; Ueda, W. Appl. Catal., A 2010, 376, 25–32.
17. Bordes, E. Catal. Today 1987, 1, 499–526.
18. Taufiq-Yap, Y. H.; Saw, C. S. Catal. Today 2008, 131, 285–291.
19. Landi, G.; Lisi, L.; Volta, J. C. J. Mol. Catal. A: Chem. 2004, 222, 175–181.
20. Hutchings, G. J. J. Mater. Chem. 2004, 14, 3385–3395.
21. Cavani, F.; De Santi, D.; Luciani, S.; Löfberg, A.; Bordes-Richard, E.; Cortelli, C.;
Leanza, R. Appl. Catal., A 2010, 376, 66–75.
22. Zhu, Y. J.; Li, J.; Xie, X. F.; Yang, X. G.; Wu, Y. J. Mol. Catal. A: Chem. 2006, 246,
185–189.
23. Kamiya, Y. J. Jpn. Pet. Inst. 2003, 46, 62–68.
24. Johnson, J. W.; Johnston, D. C.; Jacobson, A. J.; Brody, J. F. J. Am. Chem. Soc. 1984,
106, 8123–8128.
25. Griesel, L.; Bartley, J. K.; Wells, R. P. K.; Hutchings, G. J. J. Mol. Catal. A: Chem.
2004, 220, 113–119.
26. Albonetti, S.; Cavani, F.; Trifirò, F.; Venturoli, P.; Calestani, G.; López Granados,
M.; Fierro, J. L. G. J. Catal. 1996, 160, 52–64.
27. Hannour, F. K.; Martin, A.; Kubias, B.; Lücke, B.; Bordes, E.; Courtine, P. Catal.
Today 1998, 40, 263–272.
28. Igarashi, H.; Tsuji, K.; Okuhara, T.; Misono, M. J. Phys. Chem. 1993, 97, 7065–
7071.
29. Sananes, M. T.; Hutchings, G. J.; Volta, J. C. J. Catal. 1995, 154, 253–260.
30. Rownaghi, A. A.; Taufiq-Yap, Y. H.; Rezaei, F. Ind. Eng. Chem. Res. 2009, 48,
7517–7528.
31. Cornaglia, L. M.; Lombardo, E. A.; Andersen, J. A.; Fierro, J. L. G. Appl. Catal., A
1993, 100, 37–50.
32. Jung, S. M.; Grange, P. Appl. Catal., B 2002, 36, 325–332.
33. Centeno, M. A.; Carrizosa, I.; Odriozola, J. A. J. Alloys Compd. 2001, 323–324,
597–600.
The dehydration of xylose can be effectively carried out in the presence
of vanadium phosphates as catalysts. The (VO)2P2O7 material (denoted
VP2-550-2h), prepared by calcination of the precursor VOHPO4ꢀ0.5H2O
at 550 °C/2 h, exhibited superior catalytic performance amongst the
investigated materials. The VP2-550-2h solid was essentially insoluble
in the reaction medium, and catalytic tests and solid state characterisation
studies of the solid separated after the catalytic reaction, revealed that (i)
the catalytic results correlated with the surface acid properties; (ii) the so-
lid could be recycled without drop in catalytic activity and furfural yields
(ca. 53% at 4 h reaction); (iii) no significant structural modifications of the
recovered solid were observed. Based on these results, one could conclude
that the catalytic reaction of xylose in the presence of VP2-550-2h is het-
erogeneous in nature. However, detailed studies based on ICP-MS analy-
ses coupled with catalytic tests (directed towards the assessment of
homo/heterogeneous nature of the catalytic reaction) indicated that the
VP2-550-2h solid actually acts as a sourceofveryactive(presentinminor
amounts) water-soluble species, which are responsible for the observed
catalytic activity. The VP2-550-2h catalyst leads to higher furfural yields
than for the soluble salts (containing similar ions), VOSO4 and Na4P2O7
(or their mixture), under similar reaction conditions. A concentration of
(VO)2P2O7 as low as 5 mM, led to ca. 56% furfural yield, at 170 °C/6 h reac-
tion. It is worth stressing the importance of conducting experiments in or-
der to unveil the effective homo/heterogeneous nature of the catalytic
reaction because soluble species can be extremely active, requiring minor
amounts for performing effectively the catalytic reaction.
In overall, furfural can be effectively synthesised via the dehy-
dration of xylose in the presence of minor amounts of (VO)2P2O7
and, in this case, the catalyst recycling could involve the separation
of the toluene phase from the aqueous one (decantation) after the
reaction of xylose, the recovery of furfural by distillation, and the
used toluene extracting solvent could be recharged to the reactor
34. Sun, D.; Liu, Q.; Liu, Z.; Gui, G.; Huang, Z. Appl. Catal., B 2009, 92, 462–467.
35. Van Tongelen, M. J. Catal. 1966, 5, 535–537.
36. Busca, G.; Centi, G.; Trifiró, F.; Lorenzelli, V. J. Phys. Chem. 1986, 90, 1337–1344.
37. Antal, M. J.; Leesomboon, T.; Mok, W. S. Carbohydr. Res. 1991, 217, 71–85.
38. Dias, A. S.; Lima, S.; Pillinger, M.; Valente, A. A. Catal. Lett. 2007, 114, 151–160.