Niobium phosphates as bifunctional catalysts for the conversion of biomass-derived monosaccharides

Article abstract of DOI:10.1016/j.apcata.2021.118099

The direct conversion of glucose and xylose into HMF and furfural is more efficient using a combination of Lewis and Br?nsted acids. Herein, niobium phosphates were prepared with different compositions, which affect its surface acidity and, consequently, the catalytic activity. The catalysts with a high P/Nb molar ratio presented a low molar ratio between Lewis and Br?nsted acid sites (L/B ratio), while those with low P/Nb molar ratio displayed a high L/B ratio. NbP-2, the sample with the highest L/B ratio, showed the highest reaction rate for both HMF and furfural formation. It was found, indeed, that the reaction rate for monosaccharides conversion and furans formation correlate linearly with the L/B ratio. The results presented not only introduce niobium phosphates with a high L/B molar ratio as promising catalysts for HMF and furfural production but also provide fundamental knowledge that will guide the design of other bifunctional heterogeneous catalysts.

Full text of DOI:10.1016/j.apcata.2021.118099

Applied Catalysis A, General 617 (2021) 118099
Contents lists available at ScienceDirect
Applied Catalysis A, General
journal homepage: www.elsevier.com/locate/apcata
Niobium phosphates as bifunctional catalysts for the conversion of
biomass-derived monosaccharides
a
b
a,
c
a
c
´
´
Jose Lucas Vieira , Geo Paul , Gustavo D. Iga , Natalia M. Cabral , Jose Maria C. Bueno ,
,
,
Chiara Bisiob d, Jean Marcel R. Galloa *
a
˜
Group of Renewable Energy, Nanotechnology, and Catalysis (GreenCat), (www.greencat.ufscar.br), Department of Chemistry, Federal University of Sao Carlos, Rod.
˜
Washington Luis, KM 235, CEP 13565-905, PO Box 676, Sao Carlos, SP, Brazil
b Department of Sciences and Technological Innovation and Interdisciplinary Nano-SiSTeMI Centre, University of Eastern Piedmont A. Avogadro, 15121 Alessandria,
Italy
c
˜
˜
Department of Chemical Engineering, Federal University of Sao Carlos, Rod. Washington Luis, KM 235, CEP 13565-905, PO Box 676, Sao Carlos, SP, Brazil
d CNR-SCITEC Institute of Science and Molecular Technologies “G. Natta” via C. Golgi 19, 20133 Milano, Italy
A R T I C L E I N F O
A B S T R A C T
Keywords:
The direct conversion of glucose and xylose into HMF and furfural is more efficient using a combination of Lewis
and Brønsted acids. Herein, niobium phosphates were prepared with different compositions, which affect its
surface acidity and, consequently, the catalytic activity. The catalysts with a high P/Nb molar ratio presented a
low molar ratio between Lewis and Brønsted acid sites (L/B ratio), while those with low P/Nb molar ratio dis-
played a high L/B ratio. NbP-2, the sample with the highest L/B ratio, showed the highest reaction rate for both
HMF and furfural formation. It was found, indeed, that the reaction rate for monosaccharides conversion and
furans formation correlate linearly with the L/B ratio. The results presented not only introduce niobium phos-
phates with a high L/B molar ratio as promising catalysts for HMF and furfural production but also provide
fundamental knowledge that will guide the design of other bifunctional heterogeneous catalysts.
Bifunctional heterogeneous catalyst
Biomass conversion
Acid catalysis
Furfural
5-Hydroxymethylfurfural
1. Introduction
As for the chemical conversion, glucose and xylose can be converted
by acid catalysis to 5-hydroxymethylfurfural (HMF) and furfural,
Petroleum is currently the main carbon source to produce fuels and
chemical compounds. However, as a non-renewable feedstock, the
global reserves are expected to last 70–130 years [1]. Also, the oil-based
industry plays a big role in the production of greenhouse gases (GHG).
For instance, the petroleum well-to-refinery GHG emissions were esti-
mated at approximately 1.7 Gt of CO2 eq. in 2015, 42 % higher than the
previously anticipated by the International Association of Oil and Gas
Producers [2].
respectively. These compounds, indicated among the most important
biomass-derived molecules [3–5], are key intermediates in a biorefinery
to obtain a broad range of chemicals, monomers to polymers, and fuels
[1].
The direct conversion of glucose to HMF in high yields depends on a
combination of a Lewis acid catalyst and a Brønsted acid catalyst [6–9].
The former is responsible for isomerizing glucose to fructose, which is
then efficiently dehydrated to HMF catalyzed by the Brønsted acid.
Furfural can be directly obtained from xylose with high yields using only
a mineral acid or zeolite [10], however, it was shown that using a
combination of Lewis and Brønsted acid catalysts, the product formation
TOF increases by 5–8 times [11].
Hence, lignocellulosic biomass appears as an alternative and
renewable source of carbon, capable of delivering fuels and chemicals
through processes with reduced CO2 emission [1,3]. This source of
biomass is composed of approximately 40–50 % of cellulose, a crystal-
line polymer of the monosaccharide glucose; 25–30 % of hemicellulose,
a polymer formed mainly by the monosaccharide xylose, and 15–30 % of
lignin, a polymer composed of phenolic units [1]. Therefore, mono-
saccharides account for 60–80 % of the lignocellulosic composition, and
their use is strategic for the success of the potential biorefinery.
H-Beta zeolite [11], CrCl3/HCl [12], and Nb2O5/Amberlyst have
been studied as bifunctional systems for the conversion of xylose to
furfural [13,14]. As for the direct conversion of glucose to HMF, cata-
lytic systems based on AlCl3/HCl [9], (Sn-Beta or Sn-SBA-15)/HCl [7,
15], and (Sn-Beta or Sn-SBA-15)/Amberlyst [6,15] have been used. It is
* Corresponding author.
E-mail address: jean@ufscar.br (J.M.R. Gallo).
https://doi.org/10.1016/j.apcata.2021.118099
Received 24 November 2020; Received in revised form 16 February 2021; Accepted 10 March 2021
Available online 15 March 2021
0926-860X/© 2021 Elsevier B.V. All rights reserved.

J.L. Vieira et al.
Applied Catalysis A, General 617 (2021) 118099
interesting to note that bifunctional catalytic systems that are efficient
for the conversion of one of the monosaccharides, might not be to the
other. For instance, Sn-Beta/Amberlyst reached high product yields only
for glucose conversion [6,16], while H-Beta only for xylose conversion
[11].
studied (sample hereafter coded NbP-C).
The NbP-1 sample was prepared by the addition of 50.0 mL of
deionized water in 2.73 g of NbCl5 under stirring, then 2.30 g of H3PO4
(85 wt% in water) and 50.0 mL of deionized water were added under
vigorous stirring. The mixture was stirred for 30 min, followed by
adjusting the pH to 2.60 with NH4OH (27 wt% in water). After, stirring
for 5 more minutes, the white precipitate formed was separated by
filtration, washed with 1.5 L of deionized water, and transferred (still
wet due to the filtration process) to a 50 mL Beaker containing a solution
of 2.90 g of hexadecylamine in 20.0 mL of deionized water. The
dispersion was stirred for 30 min, the pH adjusted to 4.00 with
concentrated H3PO4 and the solution stirred for another 30 min. The
The direct conversion of glucose to HMF mainly relies on the com-
bination of two catalysts, however, using a single bifunctional catalyst
would be ideal since lower diffusion issues and higher reaction rates are
expected when both catalytic sites are nearby in the surface. In this
respect, developing a bifunctional heterogeneous Lewis and Brønsted
acid catalyst is one of the major challenges in the conversion of biomass-
derived saccharides into platform molecules [1].
◦
Due to their inherently bifunctional nature [17], niobium oxide has
been applied as catalysts for the transformation of saccharides in recent
studies [18–26]. For the conversion of xylose, Nb2O5 acts as an efficient
bifunctional catalyst using as solvent both THF (with 20 wt. % of water)
or pure water. The catalyst was shown to promote the isomerization of
xylose to xylulose followed by dehydration to furfural, reaching selec-
tivities between 42 and 47 %.
mixture was transferred to an autoclave and aged for 48 h at 65 C.
Finally, the white solid was filtered out and washed with 2.0 L of
deionized water, dried overnight at room temperature, and calcined
under air flow for 6 h at 550 ◦C (1 ◦C minꢀ 1).
The NbP-2 sample was synthesized as described for NbP-1 except
that sodium dodecyl sulfate (SDS) was used instead of hexadecylamine.
Furthermore, before the calcination process, the sample underwent a
Soxhlet extraction process with a 0.10 mol Lꢀ 1 HCl ethanolic solution.
The synthesis of NbP-3 was similar to the one described for the NbP-1
sample, with two modifications: (i) niobium chloride was added to a
mixture of 10.0 mL of deionized water and 1.45 g of hexadecylamine,
instead of 2.90 g of hexadecylamine and 20.0 mL of deionized water; (ii)
the aging process in an autoclave was performed at 90 ◦C, instead of 65
◦C.
As for the conversion of glucose in THF/H2O, Nb2O5 promoted
efficiently the Lewis acid-catalyzed isomerization to fructose, whereas it
displayed very low activity towards its subsequent dehydration, indi-
cating that the Brønsted acid sites are not effective for this reaction.
Hence, for glucose conversion, niobium oxide should be treated as a
Lewis acid catalyst, instead of a bifunctional catalyst [13,14].
In this context, niobium oxyphosphate (also commonly referred to as
niobium phosphates and herein named NbP) appears as a water-tolerant
bifunctional catalyst [27] and, hence, an alternative to niobium oxide.
The surface phosphate groups should be responsible for providing the
Brønsted acidity, while the Lewis acid sites are still on the top of pen-
tacoordinated Nb sites [14,28].
2.2. Catalyst characterization
The powder X-ray diffraction (XRD) analyses were performed using a
Shimadzu XRD 6000, with CuKα radiation, a voltage of 30 kV, and a
current of 30 mA using a scan speed of 1.5◦minꢀ 1. X-ray fluorescence
(XRF) analyses were carried out in a Thermo Scientific ARL PER-
FORM’X, with a rhodium tube source.
Recently, niobium phosphate has been suggested as a catalyst for the
conversion of monosaccharides. For instance, commercial niobium
phosphate (NbP-C) was combined with Nb2O5 (both from CBMM-Brazil)
for the direct conversion of glucose to HMF, in which the former acts as
Brønsted acid and the latter as Lewis acid [19,29]. In these catalytic
systems, moderate yields for HMF (between 30–35 %) were obtained.
Comparable yields were also obtained in a biphasic solvent system using
a laboratory prepared NbP [30] and in water using a mesoporous
niobium phosphate [31]. NbP-C was also studied for the conversion of
cellobiose and glucose was obtained as a major product [27]. Low
selectivity to fructose was obtained (<3 %) suggesting the inefficiency of
the Lewis acidity, and consequently, HMF selectivity was also low (< 10
%) [27]. As for the conversion of xylose to furfural, NbP led to ca. 45 %
yield in a biphasic solvent system [32].
31P solid-state nuclear magnetic resonance (31P ssNMR) spectra were
obtained on a Bruker Avance III 500 spectrometer with a wide-bore
magnet of 11.75 T using a 31P operating frequency of 202.45 MHz.
Before the analyses, the samples were pretreated under vacuum (1 ×
10ꢀ 4 mbar) at 150 ◦C for 2 h. A triple resonance 4 mm probe, in double
resonance mode, with magic angle spinning (MAS) was employed in all
experiments and the samples were placed in a Zirconia rotor and spun at
a rate of 15 kHz. The magnitude of the 31P RF field (μrf) used was 83 kHz
and the repetition time between the accumulations was 30 s. During the
acquisition, high-power proton decoupling was applied while no
decoupling was done on 93Nb. Chemical shifts are reported on δ scale
and were externally referenced to H3PO4 (85 %). 31P MAS NMR spectra
were fitted with DMFIT functions for quantitative deconvolution of the
overlapping peaks [37].
In this work, niobium phosphates were prepared by different simple
methodologies to obtain efficient bifunctional catalysts for direct con-
version of glucose and xylose to HMF and furfural using monophasic
solvent systems, which are more suitable for industrial application [6,
33]. Structure and surface properties of the most promising catalysts
were also studies seeking to correlate their properties and catalytic
activities.
Nitrogen physisorption at ꢀ 196 ◦C was carried out in an Autosorb 1.
The surface area was calculated by the BET equation and the pore size
distribution by DFT method using the adsorption branch and the plot
went through smoothing. Pore volume was calculated at PPꢀ0 1 of 0.9.
The presence of micropores was ruled out according to the t-plot
method.
2. Materials and methods
2.1. Synthesis of the catalyst
2.3. Study of the surface properties of solids by adsorption of probe
molecules followed by FTIR spectroscopy
Neutral and anionic surfactants have been previously used for pro-
ducing porous niobium oxyphosphates and [34–36], herein, we adapted
these methodologies to obtain NbP by eight different methodologies, as
described in the Supporting Information (SI). The NbP samples were
used in preliminary catalytic activity studies for the conversion of
glucose (as discussed in Section 3.1) and the most relevant catalysts
(hereafter named NbP-1, NbP-2, and NbP-3) were characterized and
further studied as for their catalytic activity. For the sake of comparison,
a commercial niobium phosphate provided by CBMM-Brazil was also
FTIR analyses were performed using a Bruker Equinox 55 spectro-
photometer equipped with a pyroelectric detector (DTGS) using a res-
olution of 4 cmꢀ 1. The samples, reduced under the form of self-
supported pellets (pressed at 3 Ton cm-2) with density between 4 and
8 mg cm-2, were placed inside of an infrared cell equipped with KBr
windows permanently connected to a vacuum line. Before the adsorp-
◦
◦
tion/desorption experiments, the samples were dried at 150 C (5 C
minꢀ 1) under vacuum (1 × 10-4 mbar) for 2 h.
2

J.L. Vieira et al.
Applied Catalysis A, General 617 (2021) 118099
Table 1
3. Results and discussion
Quantification of niobium and phosphorus of the NbP by X-ray fluorescence.
Samples
P/Nb
Formulaa
3.1. Catalyst screening
Molar Ratio
NbP-1
NbP-2
0.9
0.5
1.0
0.6
0.8
0.7
1.0
0.9
0.4
0.5
(NbO1.15)10(PO4)9
The chemical composition of the niobium oxyphosphate samples
(NbP) was determined by X-ray fluorescence (XRF) analyses, revealing
that the preparation method affects the phosphorous loading and,
consequently, the sample composition (Table 1). A rationalization of the
chemical composition and the synthesis method is present in the Sup-
porting Information. The stoichiometric niobium oxyphosphate [NbO
(PO4)] is expected to have a P/Nb molar ratio of 1.0, whereas, according
to the literature, this class of compounds is typically formed with P/Nb
molar ratios between 0.80 and 1.00 [40–42]. Hence, NbP-1, NbP-3,
NbP-5, NbP-7, NbP-8 obtained a composition similar to the expected for
niobium oxyphosphate (Table 1). Lower P/Nb ratios indicate the pres-
ence of niobium oxide and, indeed, the P/Nb molar ratio for a phosph-
ated niobium oxide was found to be 0.5 (Nb2O5-PO4, Table 1).
Interestingly, NbP-2, NbP-4, and NbP-C displayed P/Nb similar to the
phosphated niobium oxide (Table 1), suggesting that these syntheses
methodologies led to a high degree of Nb-O-Nb condensation. Based on
the P and Nb loading, a molecular formula was proposed for the NbP
samples (Table 1).
(NbO1.75)2(PO4)
NbO(PO4)
NbP-3
NbP-4
(NbO1.60)5(PO4)3
(NbO1.30)5(PO4)4
NbP-5
NbP-6
(NbO1.45)10(PO4)7
NbP-7
NbO(PO4)
NbP-8
(NbO1.15
)
10(PO4)9
NbP-C
Nb2O5-PO4
(NbO1.90)5(PO4)2
(NbO1.75)2(PO4)
a
Proposed based on the P/Nb molar ratio, considering Nb(IV) and that P is
present as phosphate species.
Ammonia, pyridine, or acetonitrile was adsorbed on the catalysts at
room temperature with a maximum pressure of 50, 25, and 130 mbar,
respectively, for 1 h to warrant the saturation of adsorbed molecules on
the catalyst surface. The desorption of the probe molecules was per-
formed under vacuum conditions (1 × 10ꢀ 4 mbar) at 100 ◦C for pyridine
and acetonitrile and room temperature for ammonia.
Analyses of ammonia and pyridine were used for quantification of
the Lewis and Brønsted acid sites using the methodology reported
elsewhere [38,39]. The spectra were normalized as described in these
previous works.
To identify if the catalysts were active in the direct conversion of
glucose to HMF, a first screening was performed and as shown in Fig. 1A,
as glucose is converted, both fructose and 5-hydroxymethylfurfural
(HMF) are formed. Scheme 1 shows a typical reaction roadmap for
glucose conversion in the presence of a combination of Brønsted and
Lewis acids. Fructose is formed from the Lewis acid-catalyzed isomeri-
zation of glucose, while HMF is obtained from the Brønsted acid-
catalyzed dehydration of fructose [43,44]. These results suggest that
the NbP catalysts are indeed bifunctional.
2.4. Catalytic studies
The catalytic reactions were performed in thick-glass reactors at 130
◦C. The reactors were loaded with 2.0 g of a solution of 2 wt. % of
glucose or xylose in tetrahydrofuran (THF) containing 20 wt. % of water
(THF/H2O = 4/1) or pure water. For reactions using HCl as co-catalyst,
the acid was introduced in the reaction mixture with a concentration of
0.18 mol Lꢀ 1. The reactions were stopped at different reaction times by
cooling down the reaction in an ice bath. Identification and quantifi-
cation of the products were performed by High-Performance Liquid
Chromatography (HPLC) using standards. For the analyses, a Shimadzu
HPLC LC-10AD was coupled to a Bio-rad Aminex® HPX-87H column of
300.0 mm x 7.8 mm (0.005 mol Lꢀ 1 of aqueous H2SO4 was used as
mobile phase) and to detectors of refraction index (Shimadzu RID-20A)
and diode array (Shimadzu SPD-10A).
The catalytic performance among the NbPs is significantly different
which could be associated with their composition. Indeed, by plotting
the P/Nb ratio of the different materials as a function of the reaction rate
value obtained from the data in Fig. 1A, a clear trend is observed: re-
action rate for HMF formation increases as the P/Nb ratio decreases
(Fig. 1B). In particular, NbP-2 and NbP-C, the catalysts with the lowest
P/Nb ratio, displayed the highest reaction rate for HMF formation.
NbP-2 and NbP-C samples were selected for further characterization
and catalytic studies since they are the most promising samples for the
direct conversion of glucose to HMF. For sake of comparison, NbPs with
different compositions, such as NbP-1 and NbP-3, were also chosen for
further studies.
The reaction rate was calculated using the equation below:
(initial mol Glucose) ꢀ (final mol Glucose)
reaction rateGlucose consumed
=
(catalyst weight)x(reaction time)
3.2. Physicochemical characterization of selected solids
(mol HMF formed)
reaction rateHMF formation
=
X-ray diffraction analyses (Figure S1) suggest that the prepared
samples and the commercial NbP-C are amorphous. Nevertheless,
(catalyst weight)x(reaction time)
Fig. 1. (A) Catalytic tests of the NbP samples (0.020 g) in THF/H2O = 4/1 for 1 h at 130 ◦C and (B) Correlation between P/Nb molar ratio and reaction rate for HMF
formation at 1 h of reaction.
3

J.L. Vieira et al.
Applied Catalysis A, General 617 (2021) 118099
Scheme 1. Conversion of glucose with a combination of Brønsted and Lewis acids. Thicker arrows indicate the preferential routes. Green and blue arrows indicate
the reaction steps catalyzed by Lewis and Brønsted acid sites, respectively.
Fig. 2. (A) Nitrogen physisorption isotherm at ꢀ 196 ◦C and (B) Smoothed pore width distribution obtained by the DFT method.
larger pores, with a maximum at 4.7 nm.
Table 2
NbP-C is the sample with the highest surface area (366 m2 gꢀ 1),
Textural properties obtained by N2 physisorption at ꢀ 196 ◦C and the relative
followed by NbP-2 and NbP-3, both with similar surface areas (231 and
proportion of the population distribution of phosphorus species in the NbP
244 m2 gꢀ 1, respectively) (Table 2). NbP-1 presented a surface area
samples as estimated by 31P MAS NMR analysis.
significantly lower than the other samples (44 m2 gꢀ 1), as a consequence
Textural properties
31P NMR relative contribution / wt. %
of the low porosity. The materials did not present a microporous surface
area according to the t-plot. The total pore volume was calculated at a
partial pressure of 0.9 and followed the same tendency of the surface
area.
a
c
SBET
m
/
Dpb
V0.9
/
2
ꢀ 7
ꢀ 14
ꢀ 21
Samples
/
cm
ppm
ppm
ppm
ppm
2 gꢀ 1
nm
3 gꢀ 1
NbP-1
NbP-2
NbP-3
NbP-C
44
3.8
4.7
3.9
3.6
0.045
0.23
0.27
0.44
1
1
1
2
27
34
21
36
29
43
Since XRF analyses indicate the different compositions of the sam-
ples, solid-state nuclear magnetic resonance analyses of 31P (31P ssNMR)
were carried out seeking an understanding of the P species formed
(Fig. 3). As a general feature, NMR bands in the ranges -18 to -27, -11 to
231
244
366
38
34
42
27
44
20
a
b
c
Specific surface area calculated by the BET equation.
pore size distribution calculated by DFT method.
total pore volume calculated at a partial pressure of 0.9.
-17, and -4 to -10 ppm are due to, respectively, phosphate groups (PO43ꢀ
)
in the bulk of the material, surface hydrogenphosphate (HPO-42), and
surface dihydrogenphosphate (H2PO4-) [45,46]. Moreover, a very weak
band at 2 ꢀ 0 ppm is also visible and can be attributed to hydrolyzed
phosphate groups [42].
analyses performed at a low angle (2θ < 3◦), displays poorly defined
diffraction peaks, which could indicate the formation of mesopores, for
both NbP-2 and NbP-C samples (Figure S1B).
The relative area of the bands observed for the different samples is
shown in Table 2. NbP-1 and NbP-3 are the samples with the highest
contribution of PO34ꢀ (and similar between the two materials), in
accordance with the XRF results shown in Table 1. These samples,
however, differ on the composition of the surface species, i.e, NbP-3 is
richer HPO-42, while NbP-1 surface displays a larger contribution of
H2PO4- species (Table 2). Interestingly, in the 31P ssNMR spectra of NbP-
1 and NbP-3, an up-field shift of the resonance peaks towards more
negative chemical shift values has been observed. This shift might
indicate the presence of more internal phosphate groups that do not
interact with surface OH groups or adsorbed water [47,48]. NbP-2 and
NbP-C are characterized by a large predominance of surface HPOꢀ4 2 and
H2POꢀ4 and the relative concentration of these species is similar in both
samples. NbP-2, however, displays a slightly higher concentration of
bulk PO34ꢀ (27 % against 20 % for the NbP-C). It is outstanding to
The textural properties, such as specific surface area and porosity,
were determined by nitrogen physisorption at ꢀ 196 ◦C and the results
are shown in Fig. 2 and Table 2. NbP-2, NbP-3, and NbP-C present Type-
IV isotherms (Fig. 2A), typical for mesoporous materials, endorsing that
the XRD pattern observed at a low angle (Figure S1B) could be due to
poorly organized mesopores. Isotherm for NbP-1 presents low adsorp-
tion, typical of non-porous material, while DFT (Fig. 2B) reveals the
presence of mesopores. Hence, this isotherm type could be an interme-
diate between Type II and Type IV, i.e, a predominantly non-porous
material with some contribution of mesopores. The pore size distribu-
tion (Fig. 2B) reveals well-defined families of mesopores with a broad
distribution. In particular, NbP-1 and NbP-3, both synthesized with
hexadecylamine, presented a maximum in the pore size distribution at
3.9 nm, which was similar to the commercial NbP-C. NbP-2, prepared
with sodium dodecyl sulfate as a structure-directing agent presented
observe that the relative concentration of the P species obtained by 31
P
4

J.L. Vieira et al.
Applied Catalysis A, General 617 (2021) 118099
(FTIR) spectroscopy of adsorbed pyridine and ammonia. The use of
pyridine is well-established in the literature for the quantification of acid
sites in different types of materials, while ammonia is considerably less
employed due to a lack of consensus on the extinction coefficient.
Furthermore, NH3 interacting with Lewis acid sites have a low FTIR
extinction coefficient which can lead to analytical error [49]. Hence,
herein, quantification of the acid sites was performed using pyridine as a
probe molecule, while the results obtained with ammonia adoption
should be taken with parsimony and were used for the calculation of the
L/B molar ratio for sake of comparison.
Lewis acidity in niobium oxyphosphates and phosphate niobium
oxide arise from low coordination surface niobium sites, similarly to
found for niobium oxide [50]. The Brønsted acid sites, as previously
discussed, are provided by the hydrogenphosphate and dihy-
drogenphosphate groups already characterized by 31P ssNMR (Fig. 3).
Upon adsorption of pyridine in the NbP samples (Fig. 4A), eight
bands are present in the 1700ꢀ 1400 cmꢀ 1 range. According to the
literature, the bands at 1637 and 1540 cmꢀ 1 are due to the ring vibration
of pyridinium ion formed by the interaction of pyridine and the Brønsted
acid sites, the bands at 1598 and 1437 cmꢀ 1 related to pyridine
hydrogen bond to terminal hydroxyls, and the absorptions at 1609, 1580
and 1446 cmꢀ 1 are due to molecular pyridine bond to Lewis acid sites
[27,51–53]. Finally, the formation of the band at 1489 cmꢀ 1 is assigned
to both pyridine and pyridinium species. The area of bands at 1446 and
1540 cmꢀ 1 were used to quantify the concentrations of both Lewis and
Brønsted acid sites (Table 3), respectively, using the molar absorption
coefficients and according to the methodology described elsewhere
[38]. A schematic view of the interaction between pyridine with both
Lewis and Brønsted acid sites on the NbPs surface is reported in Fig. 4A.
As for adsorption of ammonia monitored by FTIR (Fig. 4B), in the
range between 1800 and 1300 cmꢀ 1, it is possible to observe the for-
mation of three bands located at 1678, 1608, and 1436 cmꢀ 1. As it is
well known, the absorptions at 1678 and 1436 cmꢀ 1 are characteristics
of the asymmetric and symmetric bending modes, respectively, of the
ammonium ion (NH+4 ) formed in the presence of surface Brønsted acid
sites [54]. The band at 1608 cmꢀ 1 is related to the asymmetric bending
mode of the ammonia coordinated with Lewis acid sites [54]. From the
Fig. 3. 31P MAS NMR spectra of the NbP samples at 11.74 T after thermal
treatment at 150 ◦C for 2 h. A MAS rate of 15 kHz and high-power proton
decoupling was applied during the acquisition.
Table 3
Quantification of the Lewis and Brønsted acid sites by ammonia and pyridine
adsorption monitored by FTIR spectroscopy.
ssNMR and the P/Nb molar ratio calculated from the XRF appear to
correlate, as shown in Figure S2.
Ammonia
Pyridine
Concentration / μ
mol gꢀ 1
Sample
L/B Ratio
L/B Ratio
3.3. Surface characterization by FTIR spectroscopy
Brønsted
Lewis
NbP-1
NbP-2
NbP-3
NbP-C
0.15
0.38
0.26
0.34
182
172
134
484
31
0.17
0.90
0.40
0.65
Different surface acidity of the samples can be expected based on the
different structural and chemical compositions of the NbP samples.
Therefore, the overall concentration of Brønsted and Lewis acid species
was determined by using Fourier-transform infrared spectroscopy
155
54
314
Fig. 4. Normalized FTIR spectra of the samples adsorbing (A) pyridine and (B) ammonia after desorption (10ꢀ 4 mbar) at 100 ◦C and room temperature, respectively.
5

J.L. Vieira et al.
Applied Catalysis A, General 617 (2021) 118099
Fig. 5. Normalized FTIR of the samples (A) after adsorption of 130 mbar of acetonitrile at room temperature and (B) after subsequent desorption at 100 ◦C under
vacuum (10ꢀ 4 mbar).
intensity of the IR bands formed contacting the NbP samples with NH3
and then evacuating the probe molecule at room temperature [49], it is
possible to notice that the NbP samples contain a different amount of
both Brønsted and Lewis acid sites. The integrated intensity of the bands
at 1608 and 1436 cmꢀ 1 (related to the presence of Lewis and Bronsted
acid sites, respectively) and an extinction coefficient previously reported
[39] were used to calculate the L/B molar ratio (Table 3).
NbP-C) are the ones previously shown to have predominantly surface
HPOꢀ4 2 and H2PO-4 groups (Table 2). Contrarily, for NbP-1 and NbP-3,
structures with nearly stoichiometric niobium oxyphosphate composi-
tions display a higher relative concentration of Brønsted acid sites (lower
L/B ratio), suggesting that the stoichiometric niobium oxophosphate
structure [(NbO)(PO4)] shall present a low loading of low coordination
Nb sites.
As shown in Table 3, NbP-C is the sample with the largest overall
loading of Lewis and Brønsted acid sites, followed by NbP-2. Interest-
ingly, although NbP-1 presents a specific surface area significantly lower
than the other NbPs, the concentration of acid sites is comparable, thus
indicating a higher acid site concentration by the unit of area. The L/B
molar ratios obtained by pyridine or ammonia adsorption differ in ab-
solute value, however, there is a fairly good linear correlation between
them (Figure S4). This observation is extremely important since it sug-
gests that although the absolute values may differ, the trend is validated
by using the two probe molecules.
Since the presence of Lewis acidity is crucial for an efficient con-
version of glucose, these sites were further studied by acetonitrile
adsorption monitored by FTIR spectroscopy (Fig. 5). After adsorption of
130 mbar (Fig. 5A), the following bands are observed in the 2450ꢀ 2100
cmꢀ 1 range: (i) a triplet at 2315, 2294, and 2269 cm-1 resulting from the
overlap of two doublets related to the Fermi resonance between the
–
–
–
stretching of the CN and the combination mode υ(C C) + δ(CH3),
–
wherein the doublet with lower wavenumbers (2269 and 2294 cm-1) is
related to weak hydrogen bonds between the acetonitrile and terminal
hydroxyls and the doublet with higher wavenumbers (2294 and 2315
cm-1), related to the interaction of acetonitrile with Lewis acid sites with
Interestingly, the two samples with the highest L/B ratio (NbP-2 and
Fig. 6. Conversion of glucose catalyzed by NbPs as a function of reaction time at 130 ◦C using THF/H2O = 4/1 as a solvent system. For NbP-C and NbP-2, it was used
0.080 g of catalyst and for the others, 0.160 g. Mass balance is shown in Figure S5.
6

J.L. Vieira et al.
Applied Catalysis A, General 617 (2021) 118099
Fig. 7. Correlation between Lewis/Brønsted acid sites ratio and reaction rate for (A) Glucose conversion and (B) HMF formation. Reaction rate calculated at ca. 50 %
conversion from the data in Fig. 6.
medium strength; (ii) a doublet at 2294 and 2254 cm-1 related to
acetonitrile weakly physisorbed on the surface of the catalysts [28].
Upon desorption of the excess of acetonitrile by evacuation at 100 ◦C
(Fig. 5B), the bands related to physisorbed molecules disappear and the
other bands decrease in intensity. When acetonitrile interacts with the
Lewis acid sites, a blueshift is observed in respect to its original band,
due to the interaction with electron-withdrawing surface centers. Hence,
the higher the Lewis acidic strength, the greater the shift. The analysis of
the IR bands (Fig. 5B) reveals the presence of two families of strong
Lewis acid sites due to the bands at 2315 and 2340 cmꢀ 1, representing
shifts of ca. 20 and 45 cmꢀ 1, respectively. The former is present for all
samples, while the latter, a stronger site [28], is present for NbP-1
(Fig. 5B) [55]. The presence of stronger Lewis acid sites is not neces-
sarily an advantage since these sites not only catalyze the isomerization
of glucose but can also promote the decomposition of fructose [56].
The mass balance for the reactions is shown in Figure S5. The acid-
catalyzed conversion of glucose or xylose commonly leads to the low
mass balance due to the formation of humins [59], condensation re-
actions, degradation, and/or decomposition [14]. Importantly, the
Lewis acid sites necessary to isomerize glucose are also active in the
decomposition of the reaction products [56]. Overall, the NbPs pre-
sented a similar mass balance, except for the NbP-1 when used to
convert glucose (Figure S5). Hence, developing bifunctional catalysts for
the conversion of glucose (and xylose) which leads to high furan yield
and high carbon balance in water or aprotic solvents is an important
challenge in the field.
NbP-1 is noticeably the catalyst that leads to the lowest HMF selec-
tivity and yield. According to the acetonitrile adsorption analysis by
FTIR, this sample was the only one to display an important contribution
of strong Lewis acid sites (Fig. 5B), which could be responsible for the
poor performance of this catalyst. It is known that for some catalysts the
Lewis acid sites can also catalyze the decomposition of fructose, as
observed for niobium oxide and other systems [14,56]. It is rather
difficult to correlate the concentration of the active sites and the activity
for Lewis/Brønsted bifunctional catalysts because the nature of these
sites can be, sometimes, more important than the concentration. For
instance, the Brønsted acid sites on niobium oxide showed to be effective
in the dehydration of xylose but not for fructose [14], while the Lewis
acid sites on Al-Beta Zeolite are active in isomerizing xylose, but not for
glucose isomerization [11].
3.4. Glucose direct conversion to HMF
As mentioned before, the motivation for developing niobium oxo-
phosphate catalysis was based on our previous work [14], in which we
report that niobium oxide Brønsted acid poor performance in dehy-
drating fructose. Hence, in NbP, phosphates would provide the Brønsted
acidity necessary for the reaction [6,14]. Indeed, previous studies show
that phosphates provide stronger Brønsted acids than the ones found in
niobium oxide [57]. As for the reaction media, it has been already shown
that pure water leads to a high decomposition of the formed HMF [14,
58], hence, monophasic solvent systems composed of THF or GVL con-
taining 10ꢀ 20 wt. % of water as co-solvent has gained attention due to
its compatibility with industry requirements [6,14,33,58].
Nonetheless, previous studies for homogeneous catalysts have shown
the importance of controlling the ratio of Lewis and Brønsted acid sites
for glucose conversion to HMF [56]. It is shown that if the Lewis acid
concentration is too high compared to the concentration of the Brønsted
acid, fructose degradation to humins can take place, while if it is too low,
isomerization becomes the rate-limiting step [56].
Therefore, using the reaction conditions previously established by
our and other groups [6,14,33], NbP-1, NbP-2, NbP-3, and NbP-C were
studied, as a function of time, for direct conversion of glucose to HMF
using THF containing 20 wt.% of water (THF/H2O = 4/1), as shown in
Fig. 6. It is important to note that NbP-1 and NbP-3 samples are
significantly less active, and hence, for the reaction, it was used twice
the amount of catalyst relative to NbP-2 and NbP-C.
Indeed, for the NbP catalysts, it appears that the catalytic activity is
directly controlled by the Lewis/Brønsted acid sites (L/B) ratio and as
shown in Fig. 7, the reaction rate for both glucose conversion and HMF
formation increases along with the L/B ratio. Importantly, this trend is
valid for niobium phosphate-type catalysts. Furthermore, it cannot be
extrapolated, otherwise one could conclude that the best activity would
be reached when only Lewis acid sites are present. Herein, we used
catalysts with borderline compositions, i.e, stoichiometric niobium
oxophosphate (NbP-3) and (phosphated niobium oxide)-like (NbP-C) as
well as catalysts with an intermediate composition (NbP-1 and NbP-2),
and the correlation observed in Fig. 7 is valid for this range.
As expected, the glucose conversion increases along with the reac-
tion time, except for NbP-1 which deactivates after the second hour of
reaction. In the first hour of reaction, the formation of fructose by the
isomerization of glucose promoted by Lewis acids is observed. The
concentration of fructose along the reaction is higher for reaction pro-
moted by the catalysts with a higher Lewis/Brønsted sites ratio
(Table 3), i.e. NbP-2 and NbP-C. All the catalysts were able to convert
fructose to HMF and after 5 h of reaction, the selectivities for HMF was
ca. 40 % for NbP-1 and ca. 50 % for NbP-2, NbP-3, and NbP-C. The
highest HMF yield was obtained for NbP-2 (42.5 %), followed by NbP-3
and NbP-C (ca. 38 %), and NbP-1 reached the lowest catalyst yield (21.6
%).
The stability of NbP-2 and NbP-C, the catalysts with the best per-
formance for HMF production, were studied by reusing the catalysts, as
shown in Figure S6. NbP-2 is stable for reuse but after the 4th use, the
selectivity to HMF reduces. NbP-C, instead, was shown to be stable for
all five uses. It is interesting to mention that the activity of NbP-C in-
creases along with the number of uses, although the standard deviation
7

J.L. Vieira et al.
Applied Catalysis A, General 617 (2021) 118099
-14 ppm due to hydrogenphosphate is observed. Although phosphate
leaching is observed, it appears to be small enough not to affect signif-
icantly the catalytic activity of the catalysts since both NbP-2 and NbP-C
were shown to be successfully reused (Figure S6).
Previous studies in the literature report that combining Nb2O5 with
an additional Brønsted acid catalyst can increase the HMF productivity
[14]. In order to compare NbP with the Nb2O5-HCl catalytic system,
experiments combining NbP and HCl (0.18 mol Lꢀ 1) were performed
(Figure S9), however, it was observed a decrease in the HMF yield
compared to the reaction catalyzed by neat NbP (ca. 29, 39, 28, and 31
%, respectively, for NbP-1, NbP-2, NbP-3, and NbP-C). Therefore, NbP-2
and NP-C were shown to be efficient bifunctional catalysts able to ach-
ieve HMF yields comparable to the combination of Nb2O5/HCl. NbPs,
however, are purely heterogeneous and easy to prepare, in accordance
with industrial requirements.
Scheme 2. Conversion of xylose with a combination of Brønsted and Lewis
acids. The thicker arrow indicates the preferential route. Green and blue arrows
indicate the reaction steps catalyzed by Lewis and Brønsted acid sites,
respectively.
3.5. Xylose conversion to furfural
Xylose conversion to furfural was performed in THF/H2O = 4/1 as
the solvent system and revealed that xylose is first isomerized to xylulose
and then dehydrated to furfural (Scheme 2). In this case, there is a
competitive dehydration between xylose and xylulose, but the reaction
is faster for xylulose [11,12,14,56]. NbP-2 is the catalyst with the
highest reaction rate and the reaction is nearly completed after 2 h
(Fig. 8). The highest furfural yields obtained after 5 h of the reaction
were ca. 28, 49, 37, and 50 %, respectively, for NbP-1, NbP-2, NbP-3,
and NbP-C. Interestingly, NbP-2 presented product yield comparable
to NbP-C, whereas, with a higher reaction rate. Compared with previous
works, it seems that Nb2O5 is not less efficient in producing furfural from
xylose through the isomerization/dehydration route [14], but NbP-C,
NbP-2 appears as valid alternative.
for the conversion is pronounced and one could argue that the activities
are comparable. After the first use, NbP-2 and NbP-C were characterized
by XRD (Figure S7), XRF, and 31 P ss-NMR (Figure S8). The spent
samples presented a similar XRD pattern compared to the fresh ones,
confirming that no crystallization or phase transformation took place.
XRF analyses revealed that the spend samples presented a reduction of
1.87 and 4.07 mol % of the phosphorous content (based on the total
amount of P and Nb), respectively. The phosphate leaching was also
confirmed by 31P ss-NMR, which identified leaching of the dihy-
drogenphosphate (H2PO4ꢀ ) species (Figure S8). The NMR spectra of the
spent NbP-2 and NbP-C present a slight decrease in the band due to
As for the effect of the surface properties in the reaction rate, a
similar trend to the observed for glucose is also present in the conversion
of xylose. The reaction rate for xylose conversion (Fig. 9A) and furfural
formation (Fig. 9B) increases along with the Lewis/Brønsted acid sites
dihydrogenphosphates species (ca.
2 and -7 ppm, respectively)
compared to the fresh samples. For NbP-2, these species could also be
partially condensing with the structure, since an increase of the band at
Fig. 8. Conversion of xylose catalyzed by NbPs as a function of reaction time at 130 ◦C using THF/H2O = 4/1 as a solvent system. It was used 0.080 g of catalyst,
except for NbP-1 (0.160 g). Mass balance is shown in Figure S5.
8

J.L. Vieira et al.
Applied Catalysis A, General 617 (2021) 118099
Fig. 9. Correlation between Lewis/Brønsted acid sites ratio (obtained by pyridine adsorption) and reaction rate for (A) xylose conversion and (B) furfural formation.
Reaction rate calculated at ca. 50 % conversion from the data in Figs. 8 and 10.
Fig. 10. Conversion of xylose catalyzed by NbPs as a function of reaction time at 130 ◦C using pure water as the solvent system. It was used 0,160 g of NbP-C or NbP-
2 and 0.250 g of NbP-1 or NbP-3. Mass balance is shown in Figure S5.
ratio acids [11]. This confirms the previous findings that state that the
reaction rate for xylose conversion to furfural increases when combining
Lewis and Brønsted. Interestingly, although NbP-2 and NbP-C have L/B
molar ratios significantly different, they display similar reaction rates
for furfural formation (Fig. 9B). To verify if the strength of the Brønsted
acids is responsible for this behavior, the samples were characterized by
CO adsorption at -176 ◦C monitored by FTIR. As shown in Figure S10,
upon adsorption of CO in the samples (followed by subtraction of the
ΔνOH for a weak acid such as silica and a medium strength acid such as
Al-SBA-15 is higher than 100 cmꢀ 1 and the difference in ΔνOH for
Al-SBA-15 and a strong Bronsted acid, such as a Zeolite, is of at least 70
cm-1 [60]. Hence, due to the small difference between the acid strength
of NbP-2 and NbP-C, we were unable to confirm if this parameter is
responsible for the results observed in Fig. 9B, yet, this cannot be
completely ruled out.
Different from glucose, xylose can be efficiently converted to furfural
using pure water as a solvent for niobium-based catalysts [14]. For
Nb2O5, for instance, the product yields in water tend to be slightly lower
than in THF/H2O, but still comparable [14]. Pure water also tends to be
inadequate for phosphate and sulfates catalysts due to hydrolysis and
leaching of the inorganic anions, however, seeking a comparison to the
previously reported niobium oxide system [14], NbPs performance in
xylose conversion as a function of reaction time were also studies in pure
water (Fig. 10). NbP-C and NbP-3 appear to lose activity after a few
hours of reaction, which was not observed in THF:H2O = 4/1. NbP-2,
spectra before adsorption) it is observed a negative band at 3665 cmꢀ 1
,
–
corresponding to the O H stretching of the hydrogen phosphate ter-
minals group (POH). The negative band indicates that this species was
consumed due to its interaction with CO, leading to a broad band be-
tween 3600 and 3400 cmꢀ 1. The relative strength of the Bronsted acid
site is proportional to the shift of this band relative to the original band
–
of P OH (at 3665 cmꢀ 1) [60]. The shifts (Δ
νOH) observed for NbP-2
and NbP-C were 175 and 198 cmꢀ 1, respectively, suggesting compara-
ble acid site strength [60]. For sake of comparison, the difference of
9

J.L. Vieira et al.
Applied Catalysis A, General 617 (2021) 118099
once again, displayed the highest reaction rate (although lower than in
THF/H2O = 4/1) since the reaction was nearly completed after 5 h.
Fig. 9 shows that, overall, the reaction rates are significantly lower in
water than in THF:H2O = 4/1, however, the trend of activity per sample,
as well as the effect of Lewis/Brønsted acid sites in the catalytic per-
formance is similar.
for the helpful discussions. The authors thank CBMM Brazil for
providing the samples of niobium phosphate among others.
Appendix A. Supplementary data
Supplementary material related to this article can be found, in the
online version, at doi:https://doi.org/10.1016/j.apcata.2021.118099.
4. Conclusions
References
By modifying synthesis parameters such as type and concentration of
the surfactant, temperature, and post-reaction treatment, niobium
phosphates were prepared with different compositions. It was obtained
from materials with the composition of stoichiometric niobium oxy-
phosphate [(NbO)(PO4)] until those that resemble a phosphated
niobium oxides. The overall composition affected the surface composi-
tion and, consequently, the surface acidity.
[1] J.M.R. Gallo, M.A. Trapp, J. Braz. Chem. Soc. 28 (2017) 1586–1607.
[2] M.S. Masnadi, H.M. El-Houjeiri, D. Schunack, Y. Li, J.G. Englander, A. Badahdah,
J.-C. Monfort, J.E. Anderson, T.J. Wallington, J.A. Bergerson, D. Gordon,
J. Koomey, S. Przesmitzki, I.L. Azevedo, X.T. Bi, J.E. Duffy, G.A. Heath, G.
A. Keoleian, C. McGlade, D.N. Meehan, S. Yeh, F. You, M. Wang, A.R. Brandt,
Science (80-.) 361 (2018) 851–853.
[3] E4tech, Re-Cord, Wur, From the Sugar Platform to Biofuels and Biochemicals,
contract No. ENER/C2/423-2012/SI2.673791, 2015.
[4] J.J. Bozell, G.R. Petersen, Green Chem. 12 (2010) 539–555.
[5] G. Gibson, The Synthesis and Evolution of a Supermodel, 2005.
[6] J.M.R. Gallo, D.M. Alonso, M.A. Mellmer, J.A. Dumesic, Green Chem. 15 (2013)
85–90.
The catalysts with a high P/Nb molar ratio (composition similar to
niobium oxyphosphate) presented a low molar ratio between Lewis and
Brønsted acid sites (L/B ratio), while those with low P/Nb molar ratio
displayed a high L/B ratio. The trend of the L/B ratio obtained from
adsorption of ammonia or pyridine monitored by FTIR was NbP-1 <
NbP-3 < NbP-C < NbP-2. Acetonitrile adsorption monitored by FTIR
revealed that the samples have predominantly the same strength of
Lewis acidity, however, NbP-1 and NbP-3 also present a small contri-
bution of a family of stronger Lewis acid sites.
´
[7] E. Nikolla, Y. Roman-Leshkov, M. Moliner, M.E. Davis, ACS Catal. 1 (2011)
408–410.
[8] M. Moliner, Y. Roman-Leshkov, M.E. Davis, Proc. Natl. Acad. Sci. 107 (2010)
6164–6168.
´
[9] Y.J. Pagan-Torres, T. Wang, J.M.R. Gallo, B.H. Shanks, J.A. Dumesic, ACS Catal. 2
(2012) 930–934.
[10] E.I. Gürbüz, J.M.R. Gallo, D.M. Alonso, S.G. Wettstein, W.Y. Lim, J.A. Dumesic,
Angew. Chemie Int. Ed. 52 (2013) 1270–1274.
[11] J.M.R. Gallo, D.M. Alonso, M.A. Mellmer, J.H. Yeap, H.C. Wong, J.A. Dumesic,
Top. Catal. 56 (2013) 1775–1781.
NbP-2, the sample with the highest L/B ratio, presented the highest
reaction rate for both HMF and furfural formation. It was found, indeed,
that the reaction rate for monosaccharides conversion and for furans
formation correlate linearly with the L/B ratio. The importance of the L/
B ratio has been previously observed for homogeneous catalysts,
whereas, to the best of our knowledge, it is the first time that a direct and
linear correlation between the L/B ratio and the reaction rate is found.
The results presented herein not only introduce niobium phosphates
with a low L/B molar ratio as highly promising catalysts for HMF and
furfural production but also provides fundamental knowledge that will
guide the design of other bifunctional heterogeneous catalysts.
[12] V. Choudhary, S.I. Sandler, D.G. Vlachos, ACS Catal. 2 (2012) 2022–2028.
[13] I. Agirrezabal-Telleria, C. García-Sancho, P. Maireles-Torres, P.L. Arias,
Cuihua Xuebao, Chinese J. Catal. 34 (2013) 1402–1406.
¨
[14] J.L. Vieira, M. Almeida-Trapp, A. Mithofer, W. Plass, J.M.R. Gallo, Catal. Today
344 (2020) 92–101.
[15] J.P. Lorenti, E. Scolari, E.M. Albuquerque, M.A. Fraga, J.M.R. Gallo, Appl. Catal. A
Gen. 581 (2019) 37–42.
[16] V. Choudhary, A.B. Pinar, S.I. Sandler, D.G. Vlachos, R.F. Lobo, ACS Catal. 1
(2011) 1724–1728.
[17] K. Nakajima, Y. Baba, R. Noma, M. Kitano, J.N. Kondo, S. Hayashi, M. Hara, J. Am.
Chem. Soc. 133 (2011) 4224–4227.
[18] N.V. Gromov, O.P. Taran, V.S. Semeykina, I.G. Danilova, A.V. Pestunov, E.
V. Parkhomchuk, V.N. Parmon, Catal. Lett. 147 (2017) 1485–1495.
[19] M.N. Catrinck, E.S. Ribeiro, R.S. Monteiro, R.M. Ribas, M.H.P. Barbosa, R.
CRediT authorship contribution statement
´
F. Teofilo, Fuel 210 (2017) 67–74.
[20] C. Fang, W. Wu, H. Li, T. Yang, W. Zhao, Z. Wang, S. Yang, Energy Sources Part A
Recover. Util. Environ. Eff. 39 (2017) 2072–2077.
´
Jose Lucas Vieira: Writing - original draft, Investigation, Formal
[21] X. Li, K. Peng, X. Liu, Q. Xia, Y. Wang, ChemCatChem 9 (2017) 2739–2746.
[22] N.K. Gupta, A. Fukuoka, K. Nakajima, ACS Catal. 7 (2017) 2430–2436.
[23] A. Aronne, M. Di Serio, R. Vitiello, N.J. Clayden, L. Minieri, C. Imparato,
A. Piccolo, P. Pernice, P. Carniti, A. Gervasini, J. Phys. Chem. C 121 (2017)
17378–17389.
analysis, Validation, Methodology. Geo Paul: Investigation. Gustavo D.
´
Iga: Investigation, Validation. Natalia M. Cabral: Investigation. Jose
Maria C. Bueno: Conceptualization, Funding acquisition, Supervision.
Chiara Bisio: Conceptualization, Supervision, Writing - review & edit-
ing, Resources. Jean Marcel R. Gallo: Conceptualization, Funding
acquisition, Supervision, Writing - review & editing, Resources,
Methodology.
[24] B. Guo, L. Ye, G. Tang, L. Zhang, B. Yue, S.C.E. Tsang, H. He, Chinese J. Chem. 35
(2017) 1529–1539.
[25] N.T. do Prado, T.E. Souza, A.R.T. Machado, P.P. Souza, R.S. Monteiro, L.C.
A. Oliveira, J. Mol. Catal. A Chem. 422 (2016) 23–34.
[26] H.T. Kreissl, K. Nakagawa, Y.K. Peng, Y. Koito, J. Zheng, S.C.E. Tsang, J. Catal. 338
(2016) 329–339.
[27] P. Carniti, A. Gervasini, F. Bossola, V. Dal Santo, Appl. Catal. B Environ. 193 (2016)
93–102.
Declaration of Competing Interest
[28] H.G. Bernal, A.M.R. Galletti, G. Garbarino, G. Busca, E. Finocchio, Appl. Catal. A
Gen. 502 (2015) 388–398.
The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influence
the work reported in this paper.
[29] R.S. de Carvalho, F. de A. Rodrigues, R.S. Monteiro, W.L. da Silva Faria, Waste
Biomass Valorization 10 (2019) 2673–2680.
[30] V.V. Ordomsky, V.L. Sushkevich, J.C. Schouten, J. Van Der Schaaf, T.A. Nijhuis,
J. Catal. 300 (2013) 37–46.
[31] Y. Zhang, J. Wang, X. Li, X. Liu, Y. Xia, B. Hu, G. Lu, Y. Wang, Fuel 139 (2015)
301–307.
Acknowledgments
[32] B. Pholjaroen, N. Li, Z. Wang, A. Wang, T. Zhang, J. Energy Chem. 22 (2013)
826–832.
This work has been funded by the FAPESP Young Research Award
(Process # 2016/02128-2). J.M.R.G thanks the Alexander von Hum-
boldt Foundation for Return Fellowship and the Conselho Nacional de
[33] J.A. Dumesic, J.M.R. Gallo, D. Alonso, Method to Convert Monosaccharides to 5-
(Hydroxymethyl) Furfural (Hmf) Using Biomass-Derived Solvents, US8,772,515B2,
2014.
´
Desenvolvimento Científico e Tecnologico (CNPq) for the productivity
[34] N.K. Mal, M. Fujiwara, Chem. Commun. 2 (2002) 2702–2703.
[35] N.K. Mal, A. Bhaumik, M. Fujiwara, M. Matsukata, Microporous Mesoporous
Mater. 93 (2006) 40–45.
fellowship (303640/2019-7). J.M.R.G also thanks to the Department of
Sciences and Technological Innovation at the University of Eastern
Piedmont for the visiting professorship. J.M.C.B thanks FAPESP for the
funding (2018/01258-5). G.D.I thanks FAPESP for the fellowship
(2020/15949-0). J.L.V thanks FAPESP for the fellowship (2017/17042-
9, 2019/02132-8, and 2020/04109-0) and Dr. Diana F. Olivas Olivera
[36] N.K. Mal, A. Bhaumik, P. Kumar, M. Fujiwara, Chem. Commun. 3 (2003) 872–873.
´
[37] D. Massiot, F. Fayon, M. Capron, I. King, S. Le Calve, B. Alonso, J.O. Durand,
B. Bujoli, Z. Gan, G. Hoatson, Magn. Reson. Chem. 40 (2002) 70–76.
[38] C.A. Emeis, J. Catal. 141 (1993) 347–354.
10

J.L. Vieira et al.
Applied Catalysis A, General 617 (2021) 118099
[39] A. Taouli, W. Reschetilowski, Stud. Surf. Sci. Catal., Elsevier Inc., 2002,
[50] K. Nakajima, Y. Baba, R. Noma, M. Kitano, J.N. Kondo, S. Hayashi, M. Hara, J. Am.
Chem. Soc. 133 (2011) 4224–4227.
pp. 1315–1322.
[40] Q. Sun, A. Auroux, J. Shen, J. Catal. 244 (2006) 1–9.
[51] A. Ramírez, B.L. Lopez, L. Sierra, J. Phys. Chem. B 107 (2003) 9275–9280.
[52] B. Chakraborty, B. Viswanathan, Catal. Today 49 (1999) 253–260.
[53] T. Barzetti, E. Selli, D. Moscotti, L. Forni, J. Chem. Soc. Faraday Trans. 92 (1996)
1401–1407.
[41] V.L. Sushkevich, V.V. Ordomsky, I.I. Ivanova, Appl. Catal. A Gen. 441–442 (2012)
21–29.
[42] Y. Zhang, J. Wang, J. Ren, X. Liu, X. Li, Y. Xia, G. Lu, Y. Wang, Catal. Sci. Technol.
2 (2012) 2485–2491.
[54] Q. Sun, Y. Fu, H. Yang, A. Auroux, J. Shen, J. Mol. Catal. A Chem. 275 (2007)
183–193.
[43] J.M.R. Gallo, D.M. Alonso, M.A. Mellmer, J.A. Dumesic, Green Chem. 15 (2013)
85–90.
[55] A. Travert, A. Vimont, J.C. Lavalley, V. Montouillout, M. Rodríguez Delgado, J.
´
´
[44] Y.J. Pagan-Torres, T. Wang, J.M.R. Gallo, B.H. Shanks, J.A. Dumesic, ACS Catal. 2
J. Cuart Pascual, C. Otero Arean, J. Phys. Chem. B 108 (2004) 16499–16507.
(2012) 930–934.
[56] T. Dallas Swift, H. Nguyen, A. Anderko, V. Nikolakis, D.G. Vlachos, Green Chem.
17 (2015) 4725–4735.
[45] A. Sarkar, P. Pramanik, Microporous Mesoporous Mater. 117 (2009) 580–585.
[46] G. Paul, C. Bisio, I. Braschi, M. Cossi, G. Gatti, E. Gianotti, L. Marchese, Chem. Soc.
Rev. 47 (2018) 5684–5739.
[57] S. Impellizzeri, S. Simoncelli, C. Fasciani, M.L. Marin, G.L. Hallett-Tapley, G.
K. Hodgson, J.C. Scaiano, Catal. Sci. Technol. 5 (2015) 169–175.
[58] M.A. Mellmer, C. Sanpitakseree, B. Demir, P. Bai, K. Ma, M. Neurock, J.A. Dumesic,
Nat. Catal. 1 (2018) 199–207.
[47] J.P. Lourenço, M.F. Ribeiro, F.R. Ribeiro, J. Rocha, Z. Gabelica, J. Chem. Soc.
Faraday Trans. 91 (1995) 2213–2215.
[48] Y. Watanabe, A. Koiwai, H. Takeuchi, S.A. Hyodo, S. Noda, J. Catal. 143 (1993)
430–436.
[59] Z. Xu, Y. Yang, P. Yan, Z. Xia, X. Liu, Z.C. Zhang, RSC Adv. 10 (2020)
34732–34737.
[49] J.M.R. Gallo, C. Bisio, G. Gatti, L. Marchese, H.O. Pastore, Langmuir 26 (2010)
5791–5800.
[60] K.I. Hadjiivanov, G.N. Vayssilov, Adv. Catal., Academic Press, Cambridge, USA,
2002, pp. 307–511.
11