Development of a continuous-flow tubular reactor for synthesis of 5-hydroxymethylfurfural from fructose using heterogeneous solid acid catalysts in biphasic reaction medium

Article abstract of DOI:10.1039/d0nj05978a

5-Hydroxymethylfurfural (5-HMF) is an important biomass-derived platform chemical used to produce polymers, biofuels, and other valuable industrial chemicals. In this work, 5-HMF was synthesized from biomass-derived fructose through a continuous flow process using heterogeneous solid acid catalysts. Different solid acid catalysts, including niobium-based catalysts, Amberlyst 15, and Amberlyst 36, were tested for selective dehydration of fructose to 5-HMF in a biphasic (H2O/MIBK) continuous-flow tubular reactor. The catalysts were characterized using complementary techniques, including BET surface area, XRD, TGA, NH3-TPD, FT-IR, and pyridine-FT-IR. We also studied the effects of different reaction parameters such as the initial fructose concentration, reaction temperature, feeding flow rate, and aqueous-to-organic phase ratio. The optimal conditions were determined to be 150 °C temperature, a 0.25 ml min-1 feeding flow rate, 200 mg ml-1 NaCl concentration, 200 and 400 mg ml-1 fructose concentrations, and aqueous-to-organic phase ratios of 1?:?5 and 1?:?10. In addition, niobium phosphate (NbP), synthetic sulphated niobia (NbS) and Amberlyst 36 (Amb. 36) were active and selective, leading to 5-HMF yields in the range of 54-60% under the optimal operating conditions. Meanwhile, the Amb. 36 catalyst exhibited a 5-HMF selectivity of 70% at 150 °C, and therefore it was selected as the catalyst for the fructose dehydration reaction. Additionally, the Amb. 36 catalyst showed consistent catalytic activity and selectivity during a time-on-stream of 8 h. Furthermore, a reusability test with the used catalyst demonstrated that this catalyst can be recycled and reused without losing its catalytic activity.

Full text of DOI:10.1039/d0nj05978a

NJC
View Article Online
View Journal
PAPER
Development of a continuous-flow tubular
reactor for synthesis of 5-hydroxymethylfurfural
from fructose using heterogeneous solid acid
catalysts in biphasic reaction medium†
Cite this: DOI: 10.1039/d0nj05978a
Sadra Souzanchi,a Laleh Nazari,a Kasanneni Tirumala Venkateswara Rao,*a
a
Zhongshun Yuan,a Zhongchao Tan*b and Chunbao Charles Xu
*
5-Hydroxymethylfurfural (5-HMF) is an important biomass-derived platform chemical used to produce
polymers, biofuels, and other valuable industrial chemicals. In this work, 5-HMF was synthesized from
biomass-derived fructose through a continuous flow process using heterogeneous solid acid catalysts.
Different solid acid catalysts, including niobium-based catalysts, Amberlyst 15, and Amberlyst 36, were
tested for selective dehydration of fructose to 5-HMF in a biphasic (H2O/MIBK) continuous-flow tubular
reactor. The catalysts were characterized using complementary techniques, including BET surface area,
XRD, TGA, NH3-TPD, FT-IR, and pyridine-FT-IR. We also studied the effects of different reaction
parameters such as the initial fructose concentration, reaction temperature, feeding flow rate, and
aqueous-to-organic phase ratio. The optimal conditions were determined to be 150 1C temperature, a
0.25 ml minꢀ1 feeding flow rate, 200 mg mlꢀ1 NaCl concentration, 200 and 400 mg mlꢀ1 fructose
concentrations, and aqueous-to-organic phase ratios of 1 : 5 and 1 : 10. In addition, niobium phosphate
(NbP), synthetic sulphated niobia (NbS) and Amberlyst 36 (Amb. 36) were active and selective, leading to
5-HMF yields in the range of 54–60% under the optimal operating conditions. Meanwhile, the Amb. 36
catalyst exhibited a 5-HMF selectivity of 70% at 150 1C, and therefore it was selected as the catalyst for
the fructose dehydration reaction. Additionally, the Amb. 36 catalyst showed consistent catalytic activity
and selectivity during a time-on-stream of 8 h. Furthermore, a reusability test with the used catalyst
demonstrated that this catalyst can be recycled and reused without losing its catalytic activity.
Received 7th December 2020,
Accepted 30th March 2021
DOI: 10.1039/d0nj05978a
rsc.li/njc
petroleum-based terephthalic acid in the production of poly-
ethylene furanoate (PEF) polymer for bottle and packaging
1. Introduction
Over the past decade, the production of chemicals and fuels applications.14,15 In addition, 5-HMF also can be used in the
from alternative resources has attracted much attention because preparation of formaldehyde-free phenol-5-hydroxymethylfurfural
of the fast dwindling of fossil deposits and environmental issues (PHMF) resin, which is an alternative to the conventional Novolac
associated with the continuous use of petroleum reserves.1,2 PF resin.16,17
5-Hydroxymethylfurfural (5-HMF) is one of the top-value-added
Although the most economical hexose is glucose as a key
bio-based intermediate platform chemicals that can be directly component of cellulosic biomass, fructose has proven to be a more
produced from lignocellulosic biomass by depolymerization, favorable feedstock than glucose for 5-HMF production as it gives
followed by a dehydration reaction.3 5-HMF can be used to higher yields of 5-HMF.18 On the one hand, glucose has a more
synthesize various industrially useful chemicals and fuels by stable ring structure than fructose does, and the formation of 5-HMF
oxidation,4 amination,5,6 aldol condensation,7,8 and hydrogenation from hexoses proceeds through an open-chain mechanism. A low
reactions.9,10 Particularly, 5-HMF can be used to synthesize fraction of open chain that is formed in solution when glucose is
2,5-furandicarboxylic acid (FDCA)11–13 which can replace the used as a feedstock can result in low enolization rates.18,19 Glucose
first isomerizes to fructose before its conversion to 5-HMF. As the
a Department of Chemical and Biochemical Engineering, Western University,
isomerization process is via alkali-catalyzed reactions, high yields of
5-HMF from glucose in acidic media are not expected.18
London, Ontario, N6A 5B9, Canada. E-mail: venkat.kasanneni@gmail.com,
cxu6@uwo.ca
Several studies have reported the production of 5-HMF from
fructose and glucose in various solvents with different acidic
b Department of Mechanical & Mechatronics Engineering, University of Waterloo,
Waterloo, ON, N2L 3G1, Canada. E-mail: tanz@uwaterloo.ca
† Electronic supplementary information (ESI) available. See DOI: 10.1039/d0nj05978a catalysts, either homogeneous or heterogeneous.20–25 Among the
This journal is © The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2021
New J. Chem.

View Article Online
Paper
NJC
possible reaction media, water is considered the most economical All these catalysts exhibit moderate to high 5-HMF yields under
and environmentally-friendly. The application of heterogeneous various reaction conditions using different reaction media.
catalysts is also favourable because of their recyclability, thus
Furthermore, polystyrene sulfonic acid resins act as solid
facilitating an environmentally-friendly and cost-effective con- Brønsted acids and these catalysts can also be used for the
version. Moreover, heterogeneous catalysts have shown superior dehydration of fructose to 5-HMF. Simeonov et al. reported
5-HMF selectivity.26 However, most of the solid acid catalysts Amberlyst-15 catalyst for the dehydration of fructose in a
cannot maintain their acidity in water without deactivation of tetraethylammonium bromide reaction medium at 100 1C for
their acid sites, and they have low 5-HMF yields even at high 15 min and obtained 97% isolated yield of 5-HMF with 99%
reaction temperatures in addition to low selectivity because of purity.45 In other work, Antonetti et al. use an Amberlyst-70
the uncontrolled rehydration of 5-HMF to levulinic and formic catalyst for the production of 5-HMF from a highly concen-
acids and/or the self-polymerization of 5-HMF leads to the trated solution of fructose (10 and 20 wt%) and achieve a more
formation of soluble and insoluble polymeric substances called than 46% molar yield of 5-HMF at 180 1C.46 Qi et al. reported
humins.27–29 One of the proposed solutions to this problem is Amberlyst-15 catalyst for fructose dehydration into 5-HMF in
the in situ extraction of 5-HMF from the aqueous reaction media ionic liquid 1-butyl-3-methyl imidazolium chloride. A fructose
using an organic extractive solvent, which is immiscible in water conversion of 98.6% and a 5-HMF yield of 82.2% were achieved
to avoid its successive transformation.18,29 Efforts have been at 80 1C.24 Although these catalysts are highly selective in batch
made to improve the 5-HMF yield in the dehydration of fructose/ processes, it is necessary to evaluate their performance in a
glucose with a water-immiscible low-boiling organic solvent to continuous mode of fructose dehydration.
extract the 5-HMF formed from the reacting phase. Different
So far, most 5-HMF production reactions have been per-
organic solvents such as methyl isobutyl ketone (MIBK), tetra- formed in batch reactors26,28,29,47 or continuous stirred tank
hydrofuran (THF), n-butanol, g-valerolactone (GVL), and 2-methyl reactors (CSTR).48–50 Continuous plug flow reactors (PFR) can offer
tetrahydrofuran (2-MeTHF) have been used to improve the lower operating costs and lower environmental impacts of
5-HMF yield.30–33 Furthermore, addition of NaCl salt to the chemical production compared with batch reactors.51 They also
biphasic reaction medium results in the salting-out effect, which can vary the reaction time and alter the product properties by
further enhances the 5-HMF yield.31
changing the feeding flow rate and/or catalyst loading. In addition,
Both niobic acid (Nb2O5ꢁnH2O) and niobium phosphate they could be a better mode for large-scale operations. There are
(NbOPO4) contain Brønsted and Lewis acid sites and are well few studies on the continuous flow dehydration of hexoses into
known as water-tolerant bifunctional heterogeneous solid acid 5-HMF using plug flow reactors in single-phase52 or biphasic media
catalysts for different catalytic reactions, especially for the using a microscale reactor with a homogeneous catalyst,51,53–55 and
conversion of sugars to 5-HMF and furfural.28,34,35 Zhang et al. metal oxide particles, metal phosphates, ion exchange resins and
have studied the performance of a series of porous niobium mesoporous organosilicas as heterogeneous catalysts.56–59 Since
phosphate solid acid catalysts for the conversion of glucose or continuous-flow reactors are more desirable for industrial and
glucose-united carbohydrates into 5-HMF in pure water and water/ commercial production, more comprehensive studies are needed
MIBK solvent mixtures.36 They find that the niobium phosphate to optimize the operating conditions for the continuous production
catalyst contains both Lewis acid and Brønsted acid sites, which of 5-HMF.
are effective in isomerizing glucose to fructose and dehydrating
In this study, we report a home-made continuous-flow tubular
fructose to 5-HMF, respectively. Under the optimum operating reactor for the biphasic dehydration of fructose to 5-HMF using
conditions and in the presence of a niobium phosphate catalyst, niobium compounds and ion exchange resins as solid acid
33.2% and 39.3% 5-HMF yields were obtained from glucose in catalysts. To the best of our knowledge, these catalysts have
pure water and water/MIBK solvents, respectively.36 In another not been tested for the dehydration of fructose to 5-HMF in a
study by Carlini et al., the dehydration of fructose, sucrose and biphasic continuous-flow tubular reactor. The catalysts were
inulin to 5-HMF is studied by using niobium-based catalysts in an characterized using complementary technologies, including
aqueous medium. All the examined niobium catalysts display TGA, XRD, TPD-NH3, BET/PSD, FT-IR and Py-FTIR analyses to
higher activities than other heterogeneous systems do under examine their fructose conversion, 5-HMF selectivity, and
similar operating conditions.37 H3PO4-Treated niobic acid has a 5-HMF yield. The effects of different operating conditions, such
lower selectivity at a higher substrate conversion than niobium as the reaction temperature, feeding flow rate, initial fructose
phosphate catalysts do, which results from the slightly higher concentration, aqueous-to-organic phase ratio (A/O), and phase
Lewis and Brønsted acid sites for the niobium phosphate. transfer catalyst (PTC), on different catalysts are studied to
Improvement in the activity and selectivity of the catalysts was determine the optimal conditions for 5-HMF production.
observed when 5-HMF was extracted with MIBK from the
aqueous media.37 Apart from the niobium based catalysts, various
other heterogeneous catalysts such as vanadyl pyrophosphate,38
2. Materials and methods
2.1 Materials
TPA-AC,39 SO42ꢀ/TiO2,40 mesoporous ZrO2 nanopowder,41 arene
sulfonic acid-functionalized metal–organic framework (MOF),42
functionalized nano graphitic carbon,43 and functionalized zeolite D-Fructose (499%), Amberlyst 15, Amberlyst 36, triethylamine
catalyst44 have been tested for the conversion of fructose to 5-HMF. (TEA), sodium chloride (NaCl) and 5-hydroxymethylfurfural
New J. Chem.
This journal is © The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2021

View Article Online
NJC
Paper
(99%, for preparing HPLC standard solution) were purchased sulphated niobia (named as NbS-syn) were prepared by wet
from Sigma-Aldrich. Niobic acid (Nb2O5ꢁnH2O), and niobium impregnation of 100 g niobic acid powder in 1 litre of phosphoric
phosphate hydrate (NbOPO4ꢁnH2O) were supplied by CBMM acid and sulfuric acid solutions in water (1 M), respectively. The
˜
(Companhia Brasileira de Metalurgia e Mineraça). HPLC grade solutions were first mixed using a magnetic stirrer for 24 hours,
water and acetonitrile for preparing the mobile phase for HPLC then filtered, and finally dried overnight at 105 1C in an oven. The
analysis as well as methyl isobutyl ketone (MIBK) were purchased dried powders were then pelletized following the same procedure
from Caledon Laboratory Chemicals and used as received.
as that for NbA and NbP. Amberlyst 15 (hereafter named as
Amb. 15) and Amberlyst 36 beads (hereafter named as Amb. 36)
were used as received.
2.2 Catalyst preparation
The powdered catalysts were pelletized for using powdered
catalysts in a plug flow reactor as a packed catalytic bed.
2.3 Continuous-flow reactor setup and experimental procedure
Niobium phosphate (hereafter named as NbP) and niobic acid The catalytic conversion of fructose to 5-HMF was performed in
(hereafter named as NbA) powders were humidified overnight a novel biphasic continuous-flow tubular reactor, which is
and then pressed in a pellet die using a hydraulic press at presented in Fig. 1. The lab-scale tubular reactor was home-
10 tonnes per cm2 of pressure to create pellets. The pellets were made for testing different heterogeneous solid catalysts as a
then crushed using a Wiley Mill and sieved; particles of sizes fixed bed within the tubular reactor for the conversion of sugars
between 420 and 840 mm (mesh no. 40 to mesh no. 20) were to 5-HMF. For a typical test, pure fructose solution in an aqueous
collected and used for the experiments. Synthesized phosphated medium was used as the feedstock, and MIBK was used as the
niobia (named as NbP-syn in this study) and synthesized extracting organic solvent. MIBK could be used for the continuous
Fig. 1 Schematic diagram of the biphasic continuous-flow tubular reactor system.
This journal is © The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2021
New J. Chem.

View Article Online
Paper
NJC
in-situ extraction of the produced 5-HMF from the aqueous medium temperature at the set-point and steady-state conditions (depending
inside the fixed-bed catalytic reactor to enhance the 5-HMF selectivity on the feeding flow rate), samples are taken every hour, and each
and yield by suppressing the side reactions of 5-HMF in water. In phase (aqueous or organic) fractionated from the sample is
addition, sodium chloride (NaCl) was added to the aqueous phase of separately analyzed using HPLC. The concentrations of fructose
the feedstock to enhance the partition coefficient of 5-HMF towards and 5-HMF in each phase are calculated using data from the HPLC
the organic phase (via the salting-out effect).
chromatographs. The time on stream (TOS) for all experiments is 8
This reactor setup mainly consists of a vertical tubular hours, and no detectable soluble by-product or any other inter-
reactor (SS-316 1/200 or 5/800 tubes, 30 cm long) seamlessly fitted mediate is found in the HPLC chromatographs.
inside a custom-manufactured bipartite solid aluminum column
2.4 Product analyses
envelope connected to a heat generator surface inside a column
heater (Eppendorf CH-30). The column envelope is made of
aluminum because of its high thermal conductivity for uniform
heat transfer from the heater to the tubular reactor. In the column
heater, the temperature of the heat generator surface is adjusted
and controlled using a temperature controller (Eppendorf TC-50);
hence the aluminum column envelope can keep the temperature
constant and uniform along the tubular reactor.
Each phase (organic and aqueous) of the product samples
collected from the experiments is separately analyzed using an
HPLC instrument (Waters 2690 Separation Module) equipped
with both an RI detector (Waters 410 Differential Refractometer)
with an internal detector temperature of 35 1C and a UV detector
(Waters 484 Tunable Absorbance Detector) set at 284 nm to
determine the amount of feedstock (fructose) consumed and
the amount of product (5-HMF) produced, respectively. A Waters
XBridge Amide column (3.5 mm, 4.6 ꢂ 250 mm) maintained at
35 1C is used, and the mobile phase is 75/25 acetonitrile/water (v/v)
with 0.2 v% triethylamine (TEA) at a flow rate of 0.6 ml minꢀ1. The
results for all experiments are analyzed by external calibration
curves generated for fructose and 5-HMF separately using standard
solutions of fructose and 5-HMF of known concentrations. The
results are reported in terms of the conversion, selectivity, and
yield, which are defined and calculated as follows:
In a typical operation, an appropriate amount of heterogeneous
solid catalyst (with particle sizes between 420 and 840 mm) is
preloaded and supported inside the tubular reactor as a packed
bed between two quartz wool plugs at the upper 2/3 length of the
reactor (20 cm, reaction zone); the lower 1/3 length of the reactor
(10 cm, pre-heating zone) remains empty to pre-heat the flowing
biphasic media to the predetermined reaction temperature before
entering the reaction zone. Two HPLC feeding pumps (SSI Mighty
Mini Pump) connected to the bottom of the tubular reactor
provide independent and adjustable flow rates of the aqueous
feedstock solutions (containing substrate sugar, NaCl and water)
and extracting organic solvent (pure MIBK) and then two phases
are mixed in a tee union to provide a uniform upward rising flow
of biphasic media through the tubular reactor. Meanwhile, the
temperature of the flowing biphasic media before and after
passing the catalyst bed is monitored using two thermocouples
(Omega 1/800 K-type) located at upstream and downstream of the
reaction zone (catalyst bed) inside the tubular reactor; the thermo-
couples are connected to a digital thermometer. The pressure of
the flowing media inside the reactor is adjusted and controlled
using a back pressure regulator valve (Swagelok KBP Series) located
on the exit line of the reactor, and the pressure of the system is
monitored using a pressure gauge.
In a typical run, after a specific amount of the heterogeneous
solid catalyst particles (depending on the bulk density of the
catalyst needed to completely fill the reaction zone) had been
preloaded inside the tubular reactor as a packed bed and the
reactor had been assembled inside the column heater within the
aluminum column envelope, the aqueous feedstock solution was
pumped into the reactor using the dedicated feeding pump at a
specific flow rate. When the reactor is filled up with feedstock
solution, the extracting organic solvent (MIBK) is concurrently
pumped at a specific flow rate to the reactor. Then, the pressure
inside the reactor is raised to the desired pressure (typically 10 bar)
using the backpressure regulator valve to avoid boiling of the water
at the reaction temperatures above 100 1C and the formation
of vapour bubbles within the reactor system. The reactor is then
heated to the desired temperature after insulating the column
heater and tubular reactor. After the reactor reaches a stable
Fructose conversion ð%Þ
hꢀ
ꢁ
ꢀ
ꢁi.
aqu;F
Fru
aqu;P
Fru
C
ꢂ Qaqu
ꢀ
C
ꢂ Qaqu
MFru
ꢀ
ꢁ.
¼
¼
ꢂ 100%
aqu;F
Fru
C
ꢂ Qaqu MFru
aqu;F
Fru
aqu;P
Fru
C
ꢀ C
Fru
ꢂ 100%
aqu;F
C
(5.1)
HMF selectivity ð%Þ
hꢀ
ꢁ
ꢀ
ꢁi.
aqu;P
HMF
org;P
HMF
C
ꢂ Qaqu
ꢂ Qaqu
þ
C
ꢂ Qorg
ꢂ Qaqu
MHMF
hꢀ
ꢁ
ꢀ
ꢁi.
¼
ꢂ 100%
aqu;F
Fru
aqu;P
Fru
C
ꢀ
C
MFru
ꢄ
ꢂ
ꢃꢅꢆ
Qorg
Qaqu
aqu;P
HMF
org;P
HMF
C
þ
C
ꢂ
MHMF
ꢀ
ꢁ.
¼
ꢂ 100%:
aqu;F
Fru
aqu;P
Fru
C
ꢀ C
MFru
(5.2)
Total HMF yield ð%Þ
hꢀ
ꢁ
ꢀ
ꢁi.
aqu;P
HMF
org;P
HMF
C
ꢂ Qaqu
þ
C
ꢂ Qorg
MHMF
ꢀ
ꢁ.
¼
ꢂ 100%
aqu;F
Fru
C
ꢂ Qaqu MFru
ꢄ
ꢂ
ꢃꢅꢆ
Qorg
Qaqu
aqu;P
HMF
org;P
HMF
C
þ
C
ꢂ
MHMF
.
¼
ꢂ 100%
aqu;F
Fru
C
MFru
New J. Chem.
This journal is © The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2021

View Article Online
NJC
Paper
where, CaFqruu,F is the mass concentration of fructose in the aqueous nitrogen atmosphere. The samples are heated in a nitrogen
aqu,P
feedstock solution (mg mlꢀ1), C
is the mass concentration of flow at 20 ml minꢀ1 from 40 1C to 100 1C at 10 1C minꢀ1 and
Fru
fructose in the aqueous phase of the product sample (mg mlꢀ1), then kept at 100 1C for 10 min to remove the adsorbed moisture
CaHqMu,FP is the mass concentration of HMF in the aqueous phase of and volatile compounds. Then they are heated to 700 1C at a
the product sample (mg mlꢀ1), CHorMg,PF is the mass concentration of heating rate of 10 1C minꢀ1, and the change in the sample
HMF in the organic phase of the product sample (mg mlꢀ1), weight by temperature is recorded. For the spent catalyst, an
Qaqu is the volumetric flow rate of the aqueous feedstock oxygen flow is used to burn the humins deposited on the
solution (ml minꢀ1), Qorg is the volumetric flow rate of the surface of the catalysts. Derivative thermogravimetric (DTG)
extracting organic solvent (ml minꢀ1), MFru is the molar mass of graphs are obtained from the first derivative of the TGA results
fructose (= 180.16 g molꢀ1), and MHMF is the molar mass of with respect to time or temperature.
HMF (= 126.11 g molꢀ1).
2.5 Catalyst characterization methods
3. Results and discussion
The Brunauer–Emmett–Teller (BET) surface area, as well as the
pore volume and pore size distribution (PSD) measurements,
are measured using a Micrometrics Tristar II 3020 series
instrument. The samples are initially degassed under nitrogen
flow for 8 hours at 110 1C. The crystalline phases and structure
of the catalyst samples were investigated by X-ray diffraction
(XRD) on a PANalytical X’Pert Pro diffractometer using Cu-Ka
radiation.
3.1 Characterization of fresh catalysts
3.1.1 BET. The textural properties of the catalyst play an
important role in the catalytic dehydration of sugars.61 The BET
surface area and pore size distribution of both niobium-
and Amberlyst-based heterogeneous solid acid catalysts were
measured, and these values are presented in Table 1. As shown
in Table 1, both commercially available NbP and NbA exhibit a
high BET surface area compared with synthetic NbS-syn and
NbP-syn catalysts. However, the average pore diameter of these
catalysts is nearly the same. Although the Amb. 15 and Amb. 16
catalysts display lower BET surface area values compared with
niobium-based catalysts, they have larger pores.
3.1.2 XRD. Furthermore, XRD is used to characterize the
crystalline structure of these catalyst materials. No diffraction
peaks are observed in the XRD pattern, suggesting that all fresh
catalysts used in this work are amorphous; this finding is
compatible with those in previous studies.36,62,63
3.1.3 NH3-TPD. The acidity of catalyst plays a key role in
the catalytic dehydration of fructose to 5-HMF.64 In this study,
the acidity of niobium based solid acid catalysts was measured by
TPD-NH3 (Fig. 2). However, the Amb-15 and Amb. 36 catalysts
decompose under NH3-TPD conditions, which results in inaccurate
results and hence we did not perform the TPD-NH3 analysis for
these.65 The decomposition of Amb. 15 and Amb. 36 catalysts at
higher temperatures is confirmed by TGA analysis, which will be
discussed in Section 3.1.5. The ammonia desorption profile of the
catalysts can be divided into three regions: weak (200–350 1C),
intermediate or medium (350–450 1C), and strong (450–700 1C) acid
sites. As shown in Fig. 2, all catalysts show these three types of
acid sites; however, differences between peak intensities exist.
The commercial NbP catalyst shows a broad distribution of acid
sites and exhibits a high-intensity peak from low temperature
The total acidity of the catalysts is measured by temperature-
programmed desorption of ammonia (TPD-NH3). TPD-NH3 is
analyzed using a ChemBET Pulsar TPR/TPD automated chemi-
sorption analyzer. In a typical experiment, about 1 mg of the
catalyst sample is pretreated in a quartz reactor at 300 1C for 1 hour
with flowing helium (99.9%, 120 ml minꢀ1) to remove the physically
adsorbed substances. After pretreatment, the sample is saturated
with anhydrous ammonia at 100 1C for 10 minutes, and sub-
sequently it is flushed with helium at the same temperature
to remove any physisorbed ammonia. Then, TPD analysis is
carried out by heating the catalyst sample from ambient tem-
perature to 800 1C with a heating rate of 10 1C minꢀ1, during
which the desorbed ammonia is measured by thermal conductivity
detector (TCD) within the temperature range of 100–800 1C.
Fourier transform infrared (FT-IR) spectrometry analyses of
the catalyst samples are conducted using a PerkinElmer FT-IR
spectrometer, and the spectra are recorded in the region of
4000–550 cmꢀ1. The FT-IR spectra with adsorbed pyridine
(Py-FTIR) are in the range of 1600–1400 cmꢀ1. Before pyridine
adsorption, 1 g of each catalyst sample is heated in a vacuum
oven at 150 1C overnight to be degassed and then cooled to
room temperature in a desiccator. 500 mg of each degassed
sample is separated at this stage and is used as the background
reference for the corresponding pyridine adsorbed sample.
Another 500 mg of each sample is then treated with 500 ml of
pyridine and degassed again in a vacuum oven at 150 1C for
4 hours to remove physically adsorbed pyridine. The spectra are
recorded in the region of 1600–1400 cmꢀ1 after the samples
have been cooled to ambient temperature. The Brønsted acid to
Lewis acid site ratio (B/L) for each catalyst is evaluated by the Catalyst
Table 1 Physicochemical properties of different niobium- and Amberlyst-
based catalysts
BET surface
area (m2 gꢀ1
Average pore
diameter (nm)
Total acidity
B/L ratio (mmol gꢀ1
)
)
ratio of the peak areas at 1540 cmꢀ1 and 1445 cmꢀ1, which are
NbA
155
108
101
246
51
4
5
5
5
29
22
0.24
0.24
0.78
0.70
N/A
N/A
0.88
0.86
1.01
2.09
—
attributed to the characteristic peaks for pyridine adsorbed on NbS-syn
Brønsted acid and Lewis acid sites, respectively.60
NbP-syn
NbP
Amb. 15
Thermogravimetric analysis (TGA) of the fresh catalysts is
conducted using a PerkinElmer Pyris 1 TGA instrument in a Amb. 36
30
—
This journal is © The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2021
New J. Chem.

View Article Online
Paper
NJC
NbS-syn (0.86 mmol gꢀ1) catalysts, and hence the NbP catalyst
exhibits more acidity/activity per unit mass in comparison with
the others.
3.1.4 Pyridine FT-IR. Furthermore, these catalysts were
characterized by pyridine FT-IR analysis to distinguish the
nature of the acid sites on the catalyst surface. The FT-IR
spectra of the saturated pyridine samples are shown in Fig. 3.
All niobium-based catalysts display three different types of
pyridine FT-IR peaks. The first peak at 1445 cmꢀ1 characterizes
the Lewis acid sites, and the second peak at 1540 cmꢀ1 is
allocated to the Brønsted acid sites. The third pyridine FT-IR
peak appears at 1495 cmꢀ1, and this peak characterizes the
presence of both Brønsted and Lewis acid sites on the catalyst
surface.66 The B/L ratio is calculated based on the intensity of
the Brønsted and Lewis acid site peaks at 1540 and 1445 cmꢀ1
in the pyridine FT-IR spectra, and these values are presented in
Table 1. The B/L ratio of these catalysts are in the sequence of
NbP-syn 4 NbP 4 NbA = NbS-syn. However, the Amberlyst
Fig. 2 The TPD-NH3 profile for fresh NbP (a), NbP-syn (b), NbA (c) and
NbS-syn (d).
through to the high-temperature region compared with a catalysts show only two peaks at 1540 and 1495 cmꢀ1, corres-
similar catalyst prepared by the phosphorylation of niobic acid ponding to pure Brønsted acid sites. No peak was detected at
(NbP-syn). On the other hand, NbA and NbS-syn catalysts exhibit 1445 cmꢀ1 relating to the Lewis acid sites, indicating that the
similar intensity peaks, except that one extra shoulder peak Amberlyst catalysts do not have pure Lewis acid sites.67
appears in the high-temperature region in the NbS-syn catalyst.
3.1.5 TGA/DTG. The thermal stabilities of both niobium-
However, the intensities of these peaks are low compared with and Amberlyst-based catalysts were characterized by TGA/DTG
the NbP catalyst. The total acidities of all these catalysts are analysis, and the experimental results are shown in Fig. 4 and 5.
calculated, and these values are presented in Table 1. The total Fig. 4 shows that the Amb. 15 catalyst exhibits a slightly larger
acidity of the NbP catalyst (2.09 mmol gꢀ1) is much higher than total mass loss (66%), compared with Amb. 36 (61%). The mass
those of NbP-syn (1.01 mmol gꢀ1), NbA (0.88 mmol gꢀ1) and loss of the Amberlyst samples consists of two steps for Amb. 36
Fig. 3 Py-FTIR spectra of fresh NbA (a), NbS-syn (b), NbP-syn (c), NbP (d), Amb. 36 (e) and Amb. 15 (f).
Fig. 4 TGA/DTG graphs for fresh Amberlyst 15 (a) and Amberlyst 36 (b).
New J. Chem.
This journal is © The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2021

View Article Online
NJC
Paper
Fig. 5 TGA/DTG graphs for the NbA (a), NbP (b), NbP-syn (c) and NbS-syn (d).
and three steps for Amb. 15, which can be verified by the DTG
curves of these two samples. A small peak can be found in the
DTG of both Amberlyst catalysts at temperatures up to 200 1C,
which is likely associated with the removal of physically
adsorbed water. The second mass loss peak in the range of
200–350 1C may be due to the depolymerization of polystyrene
chains and decomposition of sulfonic groups. The mass loss
peak above 350 1C, which is only seen in Amb. 15, can be
attributed to the breakdown of the polymer backbone and the
degradation of divinylbenzene.68
On the other hand, among the TGA curves of niobium-based
catalysts, NbP has the largest total mass loss (12.1%) compared
with NbA (11.6%), NbP-syn (7.1%) and NbS-syn (6.7%) (Fig. 5).
Whereas a single mass loss peak is found within the temperature
range of 100–200 1C in the DTG analysis of all niobium catalysts,
indicating the removal of water molecules coordinated to niobium
atoms.69,70
3.1.6 FT-IR. Additionally, FT-IR analysis identified different
functional groups in the catalyst samples, and the results are
presented in Fig. 6. All niobium-based catalysts show a broad
O–H band absorption centred at 3400 cmꢀ1 and a weak O–H
band at 1620 cmꢀ1, which can be assigned to the absorption of
water. The commercial NbP and NbP-Syn catalysts have stretch-
ing vibrations at around 1000 cmꢀ1 that can be attributed to the
stretching vibration of the Nb–O–P bond.71 As expected, this
band is absent in NbA and NbS-syn catalysts. Furthermore, the
intensity of this peak is greater in NbP, suggesting a higher P
content in this catalyst.63 The Amberlyst catalysts have a broad
absorption centred at 3440 cmꢀ1 because of the overlapping of
O–H stretching vibration of water molecules and the N–H stretch of
Fig. 6 FT-IR spectra of fresh catalysts.
amines. The FT-IR absorption band at around 1700 cmꢀ1 represents
the CQO stretching vibration. The asymmetric vibration of the
C–O–C aliphatic ether and the stretching vibration of the –CH2OH the effects of the feeding flow rate and temperature on the
groups can be observed at 1120 cmꢀ1 and 1020 cmꢀ1, respectively.72 fructose conversion, 5-HMF selectivity and yield in a single
The FT-IR peaks observed between 670 cmꢀ1 and 830 cmꢀ1 are aqueous phase reaction medium. It shows very low 5-HMF yields
ascribed to the out-of-plane bending vibration of C–H.
in the pure water solvent at both feeding flow rates because of the
formation of byproducts (as indicated by the high conversion but
low selectivity/yield). The main secondary reactions in aqueous
3.2 Continuous flow catalytic dehydration of fructose to 5-HMF
3.2.1 Effects of feeding flow rate and temperature in a single media are the rehydration of 5-HMF to levulinic and formic acids
aqueous phase. The continuous-flow selective dehydration of and the polymerization of fructose and/or 5-HMF to humins.73
fructose is first performed in a pure water solvent in the presence With a constant flow rate (i.e., fixed WHSV), increasing the
of the NbA catalyst (18 g catalyst loading). Table S1 (ESI†) shows temperature from 80 1C to 120 1C increases the conversion,
This journal is © The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2021
New J. Chem.

View Article Online
Paper
NJC
selectivity, and yield of fructose. At a fixed reaction temperature, increasing the feed concentration from 10 wt% to 30 wt% and
both fructose conversion and the 5-HMF yield are slightly then to 50 wt%.26
enhanced by decreasing the flow rate from 1.0 to 0.5 ml minꢀ1
The residence time of the reaction feed in the catalytic bed
(or reducing the WHSV from 0.333 hꢀ1 to 0.166 hꢀ1). This is due can be estimated by the reciprocal of the WHSV. A longer
to the increased residence time for fructose inside the reactor. retention time has a positive effect on fructose dehydration to
It has been reported that fructose dehydration increases with 5-HMF; however, overly long retention times might also cause
increasing reaction temperature and residence time.26,74 Thus, the decomposition of 5-HMF to other by-products, subse-
higher reaction temperatures (4100 1C) and lower feed flow rates quently decreasing the 5-HMF selectivity and yield.28,76 In
(o1.0 ml minꢀ1) are further investigated in the rest of the study. the abovementioned experiments at 130 1C, the feedstock
3.2.2 Effects of feeding flow rate and temperature in a concentration is increased four times (from 100 mg mlꢀ1 to
biphasic reaction medium. The performance of the biphasic 400 mg mlꢀ1), and the aqueous feeding flow rate is reduced to
medium on fructose dehydration is studied with MIBK as the half of the initial flow (from 0.50 ml minꢀ1 to 0.25 ml minꢀ1).
extracting organic phase in the presence of the NbA catalyst The fructose conversion decreases from 93.5 to 64.8%, while
(18 g catalyst loading, the same as the reference tests in the the 5-HMF selectivity and yield change to 35% and 23%, respec-
single aqueous phase as discussed above). The results are tively. Other researchers have reported similar observations dur-
presented in Table S2 (ESI†). First, the MIBK solvent flow rate ing the catalytic dehydration of fructose to 5-HMF over different
is adjusted at 1.0 ml minꢀ1, while the aqueous feedstock flow solid acid catalysts.21,22 It should be noted that using more
rate is tested at 0.5 ml minꢀ1 (i.e., an aqueous-to-organic extracting organic solvent by decreasing the A/O ratio from 1 : 2
phase ratio of 1 : 2 (v/v)) containing 100 mg mlꢀ1 fructose and to 1: 5 (v/v) and using a higher concentration of phase transfer
100 mg mlꢀ1 NaCl. Comparing the experimental results at 120 1C catalyst (NaCl) could also enhance the 5-HMF selectivity and yield
and the 0.5 ml minꢀ1 feeding flow rate (WHSV = 0.166 hꢀ1) from by suppressing the side reactions (e.g., polymerization and rehy-
Tables S1 and S2 (ESI†) shows that the fructose conversion and dration of 5-HMF) and the formation of humins.24,50,77
5-HMF yield in the biphasic system are all higher than those in
Based on the preceding experimental results, a high fructose
the single aqueous phase. The fructose conversion and 5-HMF feedstock solution (400 mg mlꢀ1 or B40 wt%) and an A/O of
yield further increase with an increase in the reaction temperature 1 : 5 (v/v) (a feeding flow rate of 0.25 ml minꢀ1 and an MIBK
to 130 1C, and other reaction conditions are kept the same flow rate of 1.25 ml minꢀ1) are used for the rest of the study,
(0.5 ml minꢀ1 of aqueous feeding flow rate containing unless stated otherwise.
100 mg mlꢀ1 fructose and 100 mg mlꢀ1 NaCl, and an aqueous-
3.2.3 Catalytic performance of different solid acid catalysts
to-organic phase ratio of 1: 2 (v/v)). Therefore, these results for continuous dehydration of fructose to 5-HMF. Based on the
demonstrate the promoting effects of MIBK as an extracting results of the NbA catalyst, we compare various heterogeneous
organic solvent on the conversion of fructose to 5-HMF. It has solid acid catalysts, including NbA, NbS-syn, NbP-syn, NbP,
been widely believed that the presence of an organic solvent can Amb. 15 and Amb. 36, for selective dehydration of fructose to
promote the dehydration reaction by extracting the 5-HMF into 5-HMF at 130 1C under the above experimental reaction conditions,
the organic phase immediately and suppressing the unwanted except for Amb. 15 (tested at 110 1C since the maximum recom-
side reactions.26,75
The effects of the biphasic reaction medium are further catalytic results are presented in Table 2.
mended operating temperature for this catalyst is 110 1C). The
studied at 130 1C and decreasing the aqueous-to-organic phase
As shown in Table 2, both NbP and NbP-syn catalysts exhibit
ratio (A/O) to 1 : 5 (v/v) through reducing the aqueous feeding a high fructose conversion (79–91%) and 5-HMF yield (34%),
flow rate to 0.25 ml minꢀ1 and increasing the MIBK flow rate to whereas the NbA and NbS-syn catalysts produce only 63–64%
1.25 ml minꢀ1. The fructose and NaCl concentrations in the fructose conversion and 22% 5-HMF yield, respectively. It was
aqueous feed are also increased to 400 and 200 mg mlꢀ1
,
mentioned that under all the experimental reaction conditions
respectively. The results presented in Table S2 (ESI†) show that
using more extracting organic solvent (decreasing the A/O ratio),
more phase transfer catalyst, and a more concentrated feedstock
has a positive effect on the fructose dehydration reaction, resulting
in a significantly increased 5-HMF yield (23%), although a lower
fructose conversion (64.8%) is observed likely due to the decreased
retention time (with a higher WHSV = 0.333 hꢀ1). For economic
considerations, the feedstock solution should be as concentrated
as possible. However, in practice, a concentrated feedstock
solution might lower the 5-HMF selectivity because of a higher
rate of cross-polymerization and formation of humins.19,73 For
instance, Fan et al. observed an initial rise followed by a drop in
5-HMF selectivity when the initial fructose concentration
was increased from 10 wt% to 30 wt% and then to 50 wt%,
where a continuous drop in conversion was observed when
Table 2 Performance of different solid catalysts in fructose dehydration
at 130 1Ca
Fructose
conversion (%)
5-HMF
selectivity (%)
5-HMF
yield (%)
Catalyst
NbA
64.8 ꢃ 0.8
63.4 ꢃ 1.1
79.8 ꢃ 1.2
91.8 ꢃ 1.5
35.1 ꢃ 0.4
40.2 ꢃ 1.0
34.8 ꢃ 2.1
35.0 ꢃ 1.4
43.5 ꢃ 0.5
37.6 ꢃ 0.6
4.9 ꢃ 1.8
22.6 ꢃ 2.8
22.2 ꢃ 1.9
34.7 ꢃ 1.3
34.5 ꢃ 1.6
1.7 ꢃ 2.5
4.2 ꢃ 2.6
NbS-syn
NbP-syn
NbP
Amb. 15b
Amb. 36
10.1 ꢃ 1.6
a
Initial fructose concentration 400 mg mlꢀ1 (B40 wt%), NaCl concentration
200 mg mlꢀ1, feeding flow rate 0.25 ml minꢀ1, MIBK flow rate
1.25 ml minꢀ1, A/O = 1 : 5 (v/v). The operating temperature for this
b
catalyst is 110 1C.
New J. Chem.
This journal is © The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2021

View Article Online
NJC
Paper
over these catalysts, no levulinic acid and formic acid were
detected in the product mixture. The superior catalytic activity
of the niobium phosphate catalyst compared with niobic acid
has also been reported by Carniti et al. for the dehydration of
fructose to 5-HMF in an aqueous medium.74 This could be
related to the effective acidity in terms of the total number of
acid sites as well as the nature of the acid sites (the ratio of
Brønsted to Lewis acid sites [B/L]) on the catalyst surface. As
seen in Table 1, the total number of acid sites on both NbP
(2.09 mmol gꢀ1) and NbP-syn (1.01 mmol gꢀ1) catalysts is much
higher than that of both NbA and NbS-syn (0.86–0.88 mmol gꢀ1
)
catalysts. In addition, they also exhibit higher B/L ratios (0.70–0.78)
than NbA and NbS-syn (0.24). It is well known that the presence of
both Lewis and Brønsted acid sites should be responsible for
catalyzing the dehydration of carbohydrates. Weingarten et al.
studied the dehydration of xylose over solid acids, where the
furfural selectivity was reported to depend on the nature of the
acid sites on the catalyst surface. Brønsted acid sites were shown to
be more selective towards furfural production than Lewis acid
sites.78
In this study, both NbP and NbP-syn catalysts show almost
the same activity in terms of producing the same quantity of
5-HMF, which can be attributed to the high B/L ratio of these
catalysts compared with the other catalysts examined in this
work. However, the fructose conversion is higher in the
presence of the NbP catalyst. This might be due to its high
BET surface area (246 m2 gꢀ1); in addition, more total acid sites
(2.09 mmol gꢀ1) are available on the catalyst surface than those
of any other catalysts being examined. On the other hand, both
Amb. 15 and Amb. 36 catalysts give very low fructose conver-
sions (35–40%) and poor 5-HMF yields (2–4%) under the above-
stated reaction conditions.
3.2.4 Effects of reaction temperature. The reaction tem-
perature plays a major role in the selective synthesis of 5-HMF
from carbohydrates. In this study, the effects of reaction
temperature (ranging from 110 to 150 1C) on the selective
dehydration of fructose to 5-HMF over three different catalysts
(NbP, NbS-syn and Amb. 36) were examined under the following
reaction conditions: a feed flow rate of 0.25 ml minꢀ1 (WHSV =
0.428 hꢀ1), an initial fructose concentration of 400 mg mlꢀ1
Fig. 7 Effects of reaction temperature on the activity of NbP (a), NbS-syn
(b) and Amb. 36 (c). (feeding flow rate 0.25 ml minꢀ1 (WHSV = 0.428 hꢀ1),
initial fructose concentration 400 mg mlꢀ1 (B40 wt%), NaCl concentration
200 mg mlꢀ1, and A/O = 1 : 5 (v/v).)
(B40 wt%), NaCl concentration of 200 mg mlꢀ1, and A/O = NbS-syn catalysts, the 5-HMF yields for these catalysts are nearly
1: 5 (v/v). The catalytic activity results are shown in Fig. 7. equal. Therefore, from economic point of view, the Amb. 36
The results show that all catalysts can improve both fructose catalyst is more favourable than the NbP and NbS-syn catalysts
conversion and 5-HMF yield by increasing the reaction tem- for the dehydration of fructose to 5-HMF.
perature. As expected, these results agree with those reported
3.2.5 Effects of NaCl on 5-HMF selectivity. Many studies
by other researchers.24,26,74 However, the Amb. 36 catalyst have shown that the addition of NaCl to the biphasic reaction
exhibits drastic improvement in both fructose conversion and medium (H2O/MIBK) improves the 5-HMF selectivity.30,79
5-HMF selectivity upon increasing the reaction temperature The reason is that addition of NaCl to the reaction medium
from 110 1C to 150 1C. For instance, the fructose conversion modifies the ionic strength of water, which improves the
increases from 30.3% at 110 1C to 77.5% at 150 1C. Overall, a partitioning of 5-HMF into MIBK. In this study, we also examine
maximum of 78–99% fructose conversion and 54–57% of the presence and the absence of NaCl in the feedstock solution
5-HMF yields are obtained at 150 1C for these three catalysts. for the continuous synthesis of 5-HMF over the NbP catalyst at
Especially, the performance of the Amb. 36 is exceptionally two different reaction temperatures (130 1C and 150 1C) under
good with respect to 5-HMF selectivity, as high as 70.1% was the following reaction conditions: 400 mg mlꢀ1 fructose concen-
obtained. Despite the slight decrease in fructose conversion tration, 0.25 ml minꢀ1 feed flow rate (WHSV = 0.428 hꢀ1), and
obtained with the Amb. 36 catalyst compared with NbP and A/O of 1 : 5 (v/v).
This journal is © The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2021
New J. Chem.

View Article Online
Paper
NJC
by using 400 mg mlꢀ1 of initial fructose concentration (while the
WHSV is doubled or the residence time is reduced to half). This
signifies that there are sufficient catalytic sites for the conversion
of fructose even at a low catalyst dosage of 7 g. However, a high
catalyst dosage of 14 g does not show any detrimental effects on
the product selectivity and yield. Although with a fixed amount
of catalyst (7 g) and the initial fructose concentration in the
aqueous feedstock solution being reduced from 400 mg mlꢀ1 to
200 mg mlꢀ1 (i.e., reducing the WHSV to half or doubling the
residence time), this favors the side reaction leading to the
5-HMF yield dropping from 55% to 45%.
On the other hand, with a fixed amount of Amb. 36 catalyst
(14 g) and reducing the initial fructose concentration from
400 mg mlꢀ1 to 200 mg mlꢀ1 via the feeding flow rate (i.e.,
reducing the WHSV to half or doubling the residence time)
even though the conversion of fructose is more or less the
same, the 5-HMF yield is significantly decreased from 54% to
33% (Table 4). At a fixed fructose concentration (200 mg mlꢀ1),
when the catalyst loading is increased from 14 g to 24 g (i.e.,
reducing the WHSV by around 40% or increasing the residence
time by around 40%), the fructose conversion increases from
73.6 to 91.9%, while the 5-HMF yield drastically increases from
33% to 60%. Therefore, for Amb. 36, simply increasing the
catalyst loading can effectively enhance the performance of the
catalyst in the catalytic dehydration of fructose to 5-HMF.
Based on the preceding results and discussion, NbP, NbS-syn
and Amb. 36 are the most active catalysts among all catalysts
tested in this work, and the optimal conditions for the catalytic
dehydration of fructose to 5-HMF in the continuous-flow reactor
are summarized as follows: a temperature of 150 1C, aqueous feed
Fig. 8 Effects of NaCl on the activity of the NbP catalyst at 130 1C and
150 1C (feeding flow rate 0.25 ml minꢀ1 (WHSV = 0.428 hꢀ1), initial fructose
concentration 400 mg mlꢀ1 (B40 wt%), A/O = 1 : 5 (v/v)).
As shown in Fig. 8, the fructose conversion with and without
phase transfer catalyst remains almost constant (490%). How-
ever, both 5-HMF selectivity and yield increased when NaCl is
present in the aqueous feedstock solution, and this increase
becomes more evident at a higher reaction temperature
(150 1C). These results demonstrate that the presence of NaCl
in the aqueous feed contribute to the in situ extraction of
5-HMF from the aqueous phase to the organic phase, which
then suppresses the side reactions and hence increases the
5-HMF selectivity and yield. With the NbP catalyst and
200 mg mlꢀ1 NaCl, fructose dehydration achieves the highest
selectivity of 58.5% and yield of 56.7% at 150 1C.
3.2.6 Effects of catalyst loading and initial fructose concen-
tration. Previous studies have shown that increasing the catalyst
dosage could result in a higher conversion because of the
increase in residence time and catalyst active sites.22 However,
it can also result in reduced 5-HMF selectivity and yield due to
the longer residence time that favors the side reactions, for
example, rehydration and polymerization of 5-HMF and inter-
mediates into levulinic acid and other by-products.26,28 In our
experiments, the effects of catalyst loading and initial fructose
concentration on the feedstock conversion and product selectivity
are studied by changing the amount of NbP and Amb. 36 catalysts
and varying the initial fructose concentration in the aqueous
feedstock solution. The results for the NbP catalyst are presented
in Table 3. No significant change in the catalytic results is
observed when the catalyst amount is reduced from 14 g to 7 g
flow rate 0.25 ml minꢀ1, organic (MIBK) flow rate 1.25 ml minꢀ1
,
fructose concentration in the aqueous feedstock solution 200 or
400 mg mlꢀ1, and NaCl concentration in the aqueous feedstock
solution 200 mg mlꢀ1. Table 5 summarizes the activity of the
three catalysts under the optimal conditions for the production
of 5-HMF from fructose in the continuous-flow reactor. With
these three catalysts (NbP, NbS-syn and Amb. 36) under the
optimal conditions, fructose dehydration in the continuous-flow
reactor produces 5-HMF with both high selectivity (55–70%) and
high yield (54–60%).
3.2.7 Catalyst reusability study. The possibility of reusing
and recycling the catalysts was studied by testing the used Amb.
36 and NbS-syn under the optimal conditions. After running
the experiments for 8 hours on stream, the system is cooled
Table 4 Effects of catalyst loading and initial fructose concentration on
activity of Amb. 36 at 150 1Ca
Table 3 Effects of catalyst loading and initial fructose concentration on
the activity of NbP at 150 1Ca
Catalyst loading (g)
Fructose concentration (mg mlꢀ1
14 g
400b
24 g
200d
Catalyst loading (g)
Fructose concentration (mg mlꢀ1
14 g
400b
7 g
400c
)
200c
)
200b
Conversion (%)
96.9 ꢃ 2.2 98.9 ꢃ 1.9 99.7 ꢃ 1.6 Selectivity (%)
58.5 ꢃ 1.7 54.9 ꢃ 3.1 45.1 ꢃ 0.9 Yield (%)
77.5 ꢃ 2.2 73.6 ꢃ 1.5 91.9 ꢃ 0.8
70.1 ꢃ 1.8 44.5 ꢃ 2.7 64.8 ꢃ 1.7
54.3 ꢃ 3.1 32.8 ꢃ 3.3 59.6 ꢃ 1.9
Conversion (%)
Selectivity (%)
Yield (%)
56.7 ꢃ 3.5 54.3 ꢃ 3.5 44.9 ꢃ 1.8
a
Feeding flow rate 0.25 ml minꢀ1, NaCl concentration 200 mg mlꢀ1
,
Feeding flow rate 0.25 ml minꢀ1, NaCl concentration 200 mg mlꢀ1
,
A/O = 1 : 5 (v/v). WHSV = 0.428 hꢀ1
.
WHSV = 0.214 hꢀ1
.
WHSV =
a
b
c
d
A/O = 1 : 5 (v/v), 150 1C. WHSV = 0.428 hꢀ1
.
WHSV = 0.856 hꢀ1
.
0.125 hꢀ1
.
b
c
New J. Chem.
This journal is © The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2021

View Article Online
NJC
Paper
Table 5 Catalytic activity of NbP, NbS-syn and Amb. 36 under optimal
conditions for the production of 5-HMF from fructose in a continuous-
flow reactor
catalysts (see Table 6). The catalytic activity of the used NbS-syn
is reduced compared with the fresh catalyst, resulting in a lower
fructose conversion as well as a decreased selectivity and yield.
This is likely due to the deposition of humins or other organic
residues on the catalyst surface, which hinders access to the
catalytic active sites. In contrast, the Amb. 36 catalyst shows a
superb stability after 8 hours of on-stream analysis, with an
even greater selectivity and only a slight decrease in fructose
conversion and 5-HMF yield. Therefore, Amb. 36 is expected to
NbP
7
NbS-syn Amb. 36
Catalyst
loading (g)
14
14
14
24
Fructose
concentration
(mg mlꢀ1
)
400a
400b
400b
400b
200c
Conversion (%) 98.9 ꢃ 1.9 96.9 ꢃ 2.2 97.8
Selectivity (%) 54.9 ꢃ 3.1 58.5 ꢃ 1.7 59.1
77.5 ꢃ 2.2 91.9 ꢃ 0.8 have a longer lifetime in real applications.
70.1 ꢃ 1.8 64.8 ꢃ 1.7
3.2.8 Reaction mechanism. Based on the experimental
Yield (%)
54.3 ꢃ 3.5 56.7 ꢃ 3.5 57.8
54.3 ꢃ 3.1 59.6 ꢃ 1.9
results and the characteristic properties of the catalyst, a
plausible reaction mechanism for the dehydration of fructose
to 5-HMF over a highly active and selective NbP or Amb. 36
catalyst is proposed as follows. Generally, the conversion of
fructose to 5-HMF should require Brønsted acid sites.3,80 The
pyridine FT-IR results show that both NbP and Amb. 36 catalysts
contain Brønsted acid sites on the catalyst surface and these are
strong enough to produce 5-HMF from fructose. Fig. 9 shows the
dehydration mechanism of fructose to 5-HMF over NbP or Amb.
36 catalysts. Fructose can exist in the cyclic furanose structure,
which undergoes three consecutive dehydration reactions with
the aid of Brønsted acid sites on the catalyst surface, leading to
the formation of 5-HMF. However, the experimental results show
that the 5-HMF selectivity over the NbP catalyst is slightly lower
than for Amb. 36. This might be due to the Lewis acid sites on the
NbP catalyst that can initiate secondary reactions between the
formed 5-HMF and the unreacted sugar or self-polymerization of
5-HMF to produce a polymeric product called humins. Humin
deposition is observed over the spent NbP catalyst. Experimental
results show that fructose conversion on the niobium catalysts is
significantly higher than for the Brønsted acid-based Amb. 36 catalyst.
a
b
c
WHSV = 0.856 hꢀ1
.
WHSV = 0.428 hꢀ1 WHSV = 0.125 hꢀ1
. .
down and washed by pumping distilled water through the
reactor with the used catalysts packed inside. Left overnight,
the reactor is tested for another 8 hours on stream again with
fresh feedstock and the used catalyst left inside the reactor, and
the results are compared with those achieved with the fresh
Table 6 Performance of NbS-syn and Amb. 36 used catalysts at 150 1Ca
NbS-synb
Amb. 36c
Fresh
catalyst
Used
catalyst
Fresh
catalyst
Used
catalyst
Conversion (%)
Selectivity (%)
Yield (%)
97.8
59.1
57.8
87.7 ꢃ 2.1
36.7 ꢃ 1.8
32.2 ꢃ 2.9
91.9 ꢃ 0.8
64.8 ꢃ 1.7
59.6 ꢃ 1.9
81.7 ꢃ 2.2
66.3 ꢃ 0.8
54.2 ꢃ 2.5
a
Feeding flow rate 0.25 ml minꢀ1, NaCl concentration 200 mg mlꢀ1
,
b
A/O = 1 : 5 (v/v). Catalyst dosage 14 g, initial fructose concentration
400 mg mlꢀ1 (WHSV = 0.428 hꢀ1). Catalyst dosage 24 g, initial fructose
c
concentration 200 mg mlꢀ1 (WHSV = 0.125 hꢀ1).
Fig. 9 Proposed reaction mechanism for the selective dehydration of fructose to 5-HMF.
This journal is © The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2021
New J. Chem.

View Article Online
Paper
NJC
4 P. Pal and S. Saravanamurugan, ChemSusChem, 2019, 12,
145–163.
5 Z. Xu, P. Yan, K. Liu, L. Wan, W. Xu, H. Li, X. Liu and
Z. C. Zhang, ChemSusChem, 2016, 9, 1255–1258.
The Lewis acidity on the niobium-based catalysts certainly helps to
increase the fructose conversion but decreases the 5-HMF selectiv-
ity slightly. No formation of levulinic acid or formic acid is detected
in the HPLC analysis of the reaction mixture, which indicates the
absence of rehydration of 5-HMF over these catalysts under the
present experimental conditions.
´
´
6 A. Garcıa-Ortiz, J. D. Vidal, M. J. Climent, P. Concepcion,
A. Corma and S. Iborra, ACS Sustainable Chem. Eng., 2019, 7,
6243–6250.
7 J. N. Chheda and J. A. Dumesic, Catal. Today, 2007, 123, 59–70.
8 C. J. Barrett, J. N. Chheda, G. W. Huber and J. A. Dumesic,
Appl. Catal., B, 2006, 66, 111–118.
9 L. Hu, J. Xu, S. Zhou, A. He, X. Tang, L. Lin, J. Xu and
Y. Zhao, ACS Catal., 2018, 8, 2959–2980.
10 L. Hu, L. Lin, Z. Wu, S. Zhou and S. Liu, Renewable
Sustainable Energy Rev., 2017, 230–257.
11 K. T. V. Rao, J. L. Rogers, S. Souzanchi, L. Dessbesell, M. B.
Ray and C. (Charles) Xu, ChemSusChem, 2018, 11, 3323–3334.
12 J. Zhang, J. Li, Y. Tang, L. Lin and M. Long, Carbohydr.
Polym., 2015, 420–428.
13 K. T. V. Rao, Y. Hu, Z. Yuan, Y. Zhang and C. (Charles) Xu,
Chem. Eng. J., 2021, 404, 127063.
14 A. J. J. E. Eerhart, A. P. C. Faaij and M. K. Patel, Energy
Environ. Sci., 2012, 5, 6407–6422.
15 R. J. I. Knoop, W. Vogelzang, J. Van Haveren and D. S. Van
Es, J. Polym. Sci., Part A: Polym. Chem., 2013, 51, 4191–4199.
16 Z. Yuan, Y. Zhang and C. Xu, RSC Adv., 2014, 4, 31829–31835.
17 Y. Zhang, Z. Yuan and C. Xu, AIChE J., 2015, 61, 1275–1283.
18 A. I. Torres, P. Daoutidis and M. Tsapatsis, Energy Environ.
Sci., 2010, 3, 1560–1572.
19 B. F. M. Kuster, Starch/Staerke, 1990, 42, 314–321.
20 T. Ståhlberg, M. G. Sørensen and A. Riisager, Green Chem.,
2010, 12, 321–325.
4. Conclusions
In summary, the continuous synthesis of 5-HMF from fructose
was investigated in a biphasic continuous-flow tubular reactor
using different solid acid catalysts including niobium phos-
phate (NbP), niobic acid (NbA), phosphated niobia (NbP-syn),
sulfated niobia (NbS-syn), Amberlyst 15 (Amb. 15) and Amberlyst
36 (Amb. 36). Among these catalysts, Amb. 36, NbP and NbP-syn
exhibit superior catalytic activity and selectivity. Various reaction
parameters such as the effects of the initial fructose concentration,
catalyst loading, reaction temperature, aqueous-to-organic solvent
flow ratio, phase transfer catalyst and the feed flow rate were
investigated to obtain a high fructose conversion and 5-HMF yield.
Under the optimal conditions of 150 1C, 0.25 ml minꢀ1 aqueous
feed flow, 1.25 ml minꢀ1 organic (MIBK) solvent flow, 200 or
400 mg mlꢀ1 fructose feed flow and 200 mg mlꢀ1 of NaCl in the
aqueous feedstock solution, 78–99% fructose conversion, and
54–60% 5-HMF yield were obtained over the three different solid
acid catalysts (NbP, NbP-syn and Amb. 36). Moreover, NbP and
NbP-syn catalysts showed considerably high catalytic activity
and selectivity even at lower reaction temperatures (110 1C and
130 1C) compared with the other catalysts being studied, which
could be attributed to the high acidity and high B/L ratio. In
addition, the Amb. 36 catalyst showed superb stability even
after an 8 h period of on stream analysis, with better 5-HMF
selectivity and only a slight decrease in fructose conversion and
5-HMF yield. Therefore, Amb. 36 is expected to have a longer
lifetime in real applications.
21 F. S. Asghari and H. Yoshida, Carbohydr. Res., 2006, 341,
2379–2387.
22 N. Lucas, G. Kokate, A. Nagpure and S. Chilukuri, Micro-
porous Mesoporous Mater., 2013, 181, 38–46.
23 S. Caratzoulas and D. G. Vlachos, Carbohydr. Res., 2011, 346,
664–672.
24 X. Qi, M. Watanabe, T. M. Aida and R. L. Smith Jr, Green
Chem., 2009, 11, 1327–1331.
Conflicts of interest
25 G. Raveendra, M. Srinivas, N. Pasha, A. V. Prasada Rao,
P. S. Sai Prasad and N. Lingaiah, React. Kinet., Mech. Catal.,
2015, 115, 663–678.
There are no conflicts to declare.
Acknowledgements
26 C. Fan, H. Guan, H. Zhang, J. Wang, S. Wang and X. Wang,
Biomass Bioenergy, 2011, 35, 2659–2665.
27 P. Carniti, A. Gervasini and M. Marzo, Catal. Commun.,
2011, 12, 1122–1126.
28 F. Yang, Q. Liu, X. Bai and Y. Du, Bioresour. Technol., 2011,
102, 3424–3429.
The authors acknowledge the financial support from NSERC
Discovery Program, Ontario government via the NSERC/FPIn-
novations Industrial Research Chair Program and the ORF-RE
Program in Forest Biorefinery.
29 R. L. de Souza, H. Yu, F. Rataboul and N. Essayem, Challenges,
2012, 3, 212–232.
References
´
30 Y. Roman-Leshkov and J. a. Dumesic, Top. Catal., 2009, 52,
297–303.
1 D. M. Alonso, S. G. Wettstein and J. A. Dumesic, Chem. Soc.
Rev., 2012, 41, 8075–8098.
31 J. N. Chheda and J. a Dumesic, Science, 2006, 312, 1933.
´
32 L. Qi and I. T. Horvath, ACS Catal., 2012, 2, 2247–2249.
2 C. H. Zhou, X. Xia, C. X. Lin, D. S. Tong and J. Beltramini,
Chem. Soc. Rev., 2011, 40, 5588–5617.
33 A. H. Motagamwala, W. Won, C. Sener, D. M. Alonso,
C. T. Maravelias and J. A. Dumesic, Sci. Adv., 2018, 4, 1–9.
3 B. Saha and M. M. Abu-omar, Green Chem., 2014, 16, 24–38.
New J. Chem.
This journal is © The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2021

View Article Online
NJC
Paper
34 M. T. Reche, A. Osatiashtiani, L. J. Durndell, M. A. Isaacs, 57 C. Zhu, C. Cai, Q. Liu, W. Li, J. Tan, C. Wang, L. Chen,
ˆ
A. Silva, A. F. Lee and K. Wilson, Catal. Sci. Technol., 2016, 6,
Q. Zhang and L. Ma, ChemistrySelect, 2018, 3, 10983–10990.
58 C. Sonsiam, A. Kaewchada, S. Pumrod and A. Jaree, Chem.
Eng. Process., 2019, 138, 65–72.
59 M. H. Tucker, A. J. Crisci, B. N. Wigington, N. Phadke,
R. Alamillo, J. Zhang, S. L. Scott and J. A. Dumesic, ACS
Catal., 2012, 2, 1865–1876.
7334–7341.
35 N. K. Gupta, A. Fukuoka and K. Nakajima, ACS Catal., 2017,
7, 2430–2436.
36 Y. Zhang, J. Wang, X. Li, X. Liu, Y. Xia, B. Hu, G. Lu and
Y. Wang, Fuel, 2015, 139, 301–307.
37 C. Carlini, M. Giuttari, A. Maria Raspolli Galletti, G. Sbrana, 60 B. Pholjaroen, N. Li, Z. Wang, A. Wang and T. Zhang,
T. Armaroli and G. Busca, Appl. Catal., A, 1999, 183, 295–302. J. Energy Chem., 2013, 22, 826–832.
38 J. Najafi Sarpiri, A. Najafi Chermahini, M. Saraji and 61 K. T. V. Rao, S. Souzanchi, Z. Yuan and C. Xu, New J. Chem.,
A. Shahvar, Renewable Energy, 2021, 164, 11–22. 2019, 43, 12483–12493.
39 N. Lopes da Costa, L. Guedes Pereira, J. V. Mendes Resende, 62 D. Prasetyoko, Z. Ramli, S. Endud and H. Nur, Adv. Mater.
C. A. Diaz Mendoza, K. Kaiser Ferreira, C. Detoni, M. M. V. M. Sci. Eng., 2008, 2008, 1–12.
Souza and F. N. D. C. Gomes, Mol. Catal., 2021, 500, DOI: 63 Y. Zhang, J. Wang, J. Ren, X. Liu, X. Li, Y. Xia, G. Lu and
10.1016/j.mcat.2020.111334. Y. Wang, Catal. Sci. Technol., 2012, 2, 2485–2491.
40 R. Tomer and P. Biswas, New J. Chem., 2020, 44, 20734–20750. 64 C. Bispo, K. De Oliveira Vigier, M. Sardo, N. Bion, L. Mafra,
´ ˆ
41 Y. Zhou, L. Zhang and S. Tao, ACS Appl. Nano Mater., 2019,
2, 5125–5131.
P. Ferreira and F. Jerome, Catal. Sci. Technol., 2014, 4,
2235–2240.
42 Y. Zhong, Q. Yao, P. Zhang, H. Li, Q. Deng, J. Wang, Z. Zeng 65 P. F. Siril, H. E. Cross and D. R. Brown, J. Mol. Catal. A:
and S. Deng, Ind. Eng. Chem. Res., 2020, 59, 22068–22078. Chem., 2008, 279, 63–68.
43 G. Delgado Martin, C. E. Bounoukta, F. Ammari, M. I. 66 V. V. Ordomsky, V. L. Sushkevich, J. C. Schouten, J. Van Der
´
´
´
Domınguez, A. Monzon, S. Ivanova and M. A. Centeno, Catal.
Today, 2021, 366, 68–76.
Schaaf and T. A. Nijhuis, J. Catal., 2013, 300, 37–46.
67 B. Ikizer, N. Oktar and T. Dogu, Fuel Process. Technol., 2015,
138, 570–577.
68 G. Fan, C. Liao, T. Fang, S. Luo and G. Song, Carbohydr.
Polym., 2014, 112, 203–209.
69 C. C. M. Pereira and E. R. Lachter, Appl. Catal., A, 2004, 266,
67–72.
70 M. H. C. de la Cruz, J. F. C. da Silva and E. R. Lachter, Catal.
Today, 2006, 118, 379–384.
44 X. Jia, I. K. M. Yu, D. C. W. Tsang and A. C. K. Yip,
Microporous Mesoporous Mater., 2019, 284, 43–52.
45 S. P. Simeonov, J. A. S. Coelho and C. A. M. Afonso, Chem-
SusChem, 2012, 5, 1388–1391.
46 C. Antonetti, A. M. Raspolli Galletti, S. Fulignati and
D. Licursi, Catal. Commun., 2017, 97, 146–150.
47 P. Daorattanachai, S. Namuangruk, N. Viriya-empikul,
N. Laosiripojana and K. Faungnawakij, J. Ind. Eng. Chem., 71 K. T. V. Rao, S. Souzanchi, Z. Yuan, M. B. Ray and C. Xu, RSC
2012, 18, 1893–1901. Adv., 2017, 7, 48501–48511.
48 Y. Roman-Leshkov, J. N. Chheda and J. A. Dumesic, Science, 72 A. Wolowicz and Z. Hubicki, Chem. Eng. J., 2010, 160,
2006, 312, 1933–1937. 660–670.
49 J. Y. G. Chan and Y. Zhang, ChemSusChem, 2009, 2, 731–734. 73 F. Salak Asghari and H. Yoshida, Ind. Eng. Chem. Res., 2006,
´
50 B. F. M. Kuster and H. J. C. van der Steen, Starch/Staerke,
1977, 29, 99–103.
45, 2163–2173.
74 P. Carniti, A. Gervasini, S. Biella and A. Auroux, Catal.
Today, 2006, 118, 373–378.
¨
51 M. Brasholz, K. von Kanel, C. H. Hornung, S. Saubern and
´ ´ ´ ´
75 I. Jimenez-Morales, J. Santamarıa-Gonzalez, A. Jimenez-
J. Tsanaktsidis, Green Chem., 2011, 13, 1114–1117.
´
Lopez and P. Maireles-Torres, Fuel, 2014, 118, 265–271.
52 T. M. Aida, Y. Sato, M. Watanabe, K. Tajima, T. Nonaka,
H. Hattori and K. Arai, J. Supercrit. Fluids, 2007, 40, 381–388. 76 W. H. Hsu, Y. Y. Lee, W. H. Peng and K. C. W. Wu, Catal.
53 W. Guo, H. J. Heeres and J. Yue, Chem. Eng. J., 2020,
381, 122754.
54 Y. Muranaka, K. Matsubara, T. Maki, S. Asano, H. Nakagawa
and K. Mae, ACS Omega, 2020, 5, 9384–9390.
55 M. Sayed, N. Warlin, C. Hulteberg, I. Munslow, S. Lundmark,
Today, 2011, 174, 65–69.
77 R. Van Putten, J. C. Van Der Waal, E. De Jong, C. B. Rasrendra,
H. J. Heeres and J. G. De Vries, Chem. Rev., 2013, 113, 1499–1597.
78 R. Weingarten, G. A. Tompsett, W. C. Conner and G. W. Huber,
J. Catal., 2011, 279, 174–182.
O. Pajalic, P. Tunå, B. Zhang, S.-H. Pyo and R. Hatti-Kaul, 79 E. Nikolla, Y. Roman-Leshkov, M. Moliner and M. E. Davis,
Green Chem., 2020, 22, 5402–5413. ACS Catal., 2011, 1, 408–410.
56 C. V. McNeff, D. T. Nowlan, L. C. McNeff, B. Yan and 80 Q. Hou, X. Qi, M. Zhen, H. Qian, Y. Nie, C. Bai, S. Zhang,
R. L. Fedie, Appl. Catal., A, 2010, 384, 65–69. X. Bai and M. Ju, Green Chem., 2021, 23, 119–231.
This journal is © The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2021
New J. Chem.