Synthesis of HMF from fructose using Purolite strong acid catalyst: Comparison between BTR and PBR reactor type for kinetics data acquisition

Article abstract of DOI:10.1016/j.mcat.2017.12.021

5-hydroxymethyl-2-furfural (HMF) is an important furanic compound that can be obtained from renewable raw materials. HMF has wide application in the industry, acting as an intermediate in the production of fine chemicals, polymer, biofuels, and pharmaceuticals. In this research, HMF was obtained from the catalytic fructose dehydration. Purolite SGC650H, a gel-type strongly acidic resin, was applied as a catalyst using dimethylsulfoxide (DMSO) as a solvent. All experiments were carried out in a batch reactor (BTR) and a continuous packed bed reactor (PBR). The reactor performance data showed that increase in temperature resulted in higher HMF yield. In contrast, the higher the fructose concentration, the lower the HMF conversion. The HMF production rates in the PBR were higher because of the concentration of the catalyst. The rate constants were calculated by means of a pseudo-first-order reaction model, and the Arrhenius kinetic parameters obtained changed with regards to the reactants concentrations. The kinetic parameters displayed linear correlation in the Cremer-Constable diagram, which means that the system was under a kinetic compensation effect (KCE); therefore, the model was not suitable to represent the reaction. Langmuir-Hinshelwood (LHHW) models for rate equation were also tested. However all of them failed to adjust to the data, and thus a rate law was not determined. Due to the observed KCE, it was believed that there is at least a set of parallel dehydration reactions initiated from the different fructose isomers which has an influence on kinetic parameters.

Full text of DOI:10.1016/j.mcat.2017.12.021

G Model
MCAT-428; No. of Pages9
ARTICLE IN PRESS
Molecular Catalysis xxx (2018) xxx–xxx
Contents lists available at ScienceDirect
Molecular Catalysis
journal homepage: www.elsevier.com/locate/mcat
®
Synthesis of HMF from fructose using Purolite strong acid catalyst:
Comparison between BTR and PBR reactor type for kinetics data
acquisition
a
b
a,∗
Juliana Tacacima , Silas Derenzo , Joao Guilherme Rocha Poco
a
Department of Chemical Engineering, FEI University, CEP 09850-901, Sao Bernardo do Campo, SP, Brazil
Laboratory of Chemical Processes and Particle Technology, Institute for Technological Research, CEP 05508-901, Sao Paulo, SP, Brazil
b
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 29 July 2017
Received in revised form
5-hydroxymethyl-2-furfural (HMF) is an important furanic compound that can be obtained from renew-
able raw materials. HMF has wide application in the industry, acting as an intermediate in the production
of fine chemicals, polymer, biofuels, and pharmaceuticals. In this research, HMF was obtained from the
catalytic fructose dehydration. Purolite SGC650H, a gel-type strongly acidic resin, was applied as a cat-
alyst using dimethylsulfoxide (DMSO) as a solvent. All experiments were carried out in a batch reactor
1
3 December 2017
®
Accepted 16 December 2017
Available online xxx
(BTR) and a continuous packed bed reactor (PBR). The reactor performance data showed that increase in
temperature resulted in higher HMF yield. In contrast, the higher the fructose concentration, the lower
the HMF conversion. The HMF production rates in the PBR were higher because of the concentration of
the catalyst. The rate constants were calculated by means of a pseudo-first-order reaction model, and
the Arrhenius kinetic parameters obtained changed with regards to the reactants concentrations. The
kinetic parameters displayed linear correlation in the Cremer-Constable diagram, which means that the
system was under a kinetic compensation effect (KCE); therefore, the model was not suitable to represent
the reaction. Langmuir-Hinshelwood (LHHW) models for rate equation were also tested. However all of
them failed to adjust to the data, and thus a rate law was not determined. Due to the observed KCE, it was
believed that there is at least a set of parallel dehydration reactions initiated from the different fructose
isomers which has an influence on kinetic parameters.
Keywords:
5
-Hydroxymethyl-2-furfural
Fructose dehydration
Heterogeneous catalysis
Batch reactor
Continuous reactor
©
2018 Elsevier B.V. All rights reserved.
1
. Introduction
charide dehydration follows the removal of up to three molecules
of water in an aqueous or organic medium through acid catalysis.
However, the acid catalysis also fosters the following, hydrolysis,
generating the levulinic and formic acids, or even the formation
of condensation by-products, oligomerization, and polymerization,
such as soluble polymers and insoluble humins (Fig. 1). These can
be formed from the reaction between the reagent and products
from the dehydration (HMF), which contribute to decreasing the
selectivity of HMF [4,34].
Glucose as a substrate presents lower productive efficiency
when compared to fructose during the formation of HMF [5,44].
This occurs due to the stability of the cyclical structure which is
made up of six atoms in glucose, namely pyranose, when compared
to the cycle of five atoms in fructose (furanose), which results in a
lower degree of enolization responsible for the generation of HMF
The increase in demand for fossil fuels, the forecasts for their
supplies depletion, the rise in their extraction costs as well as
growing concern about environmental issues are the main fac-
tors necessary in the search for technologies to produce energy
and chemical inputs, in order to reduce the dependence on these
non-renewable resources [1,2].
5
-hydroxymethyl-2-furfural (HMF) is an essential building
block due to its industrial potential to produce biofuels, solvents
2,5-dimethylfuran, ␥-valerolactone), drugs, resins or monomers
(
for biopolymers (2,5-furan dicarboxylic acid) and other platform
substances (levulinic acid) [3].
Due to its potential as a chemical platform, the synthesis of HMF
with high yield and selectivity is desirable. The general route of sac-
[6,7]. Therefore, the initial isomerization from glucose to fructose is
an important reaction to be considered to improve the generation
of HMF, and it occurs using Lewis acids (such as ZnCl , AlCl [45,46]
2
3
∗ Corresponding author.
E-mail address: jgrpoco@fei.edu.br (J.G.R. Poco).
and BF ) as catalysts, while Brönsted acids (e.g., niobium pentoxide
3
https://doi.org/10.1016/j.mcat.2017.12.021
2
468-8231/© 2018 Elsevier B.V. All rights reserved.
®
Please cite this article in press as: J. Tacacima, et al., Synthesis of HMF from fructose using Purolite strong acid catalyst: Comparison
between BTR and PBR reactor type for kinetics data acquisition, Mol. Catal. (2018), https://doi.org/10.1016/j.mcat.2017.12.021

G Model
MCAT-428; No. of Pages9
ARTICLE IN PRESS
2
J. Tacacima et al. / Molecular Catalysis xxx (2018) xxx–xxx
Fig. 1. Formation of HMF with possible by-products (adapted from Jiang et al. [12]).
Table 1
and phosphate) are responsible for the reaction of fructose dehy-
dration [8,9]. In the cases when disaccharides and polysaccharides
are used (e.g., sucrose, inulin, and cellulose) as raw materials, it is
necessary to carry out an initial depolymerization through hydrol-
ysis [10,11].
The main catalysts used are mineral acids, organic acids, ionic
liquids, Lewis acids, metallic chlorides, oxides, phosphates, het-
eropoly acid compounds, zeolites, mesoporous solids and ion
exchange resins [9]. The acid catalysis performed to obtain HMF
may be carried out with homogeneous [32,37,38] and heteroge-
neous [36,39–43] catalysts. Nonetheless, heterogeneous catalysts
present higher selectivities when compared to homogeneous ones,
in addition to being less toxic, less corrosive, and presenting an
easier separation, which enables the recycling and adaptation for
large-scale continuous systems.
Resins are made up of reticulated polystyrene containing sul-
fonic groups and are usually presented in the form of a gel. They
are insoluble and are not attacked by solvents; however, they may
expand and swell due to the elasticity of the matrix, which makes
them different from the rigid structure of zeolites. The catalytic
reaction of resins in liquid mediums is analogous to the homoge-
neous catalysis by dissolved electrolytes, and the advantages of its
use are the separation by decantation or filtration and the possi-
bility of continuous operation in columns [13]. Another polymeric
heterogeneous catalyst has been studied such as polyaniline [43],
or its sulfonated form [42] or polymers designed to mimic the co-
catalyst effect of DMSO [40].
®
Characteristics of SGC650H (Purolite ).
Average diameter (␮m)
Concentration of acid sites (mmol/g)
Type of acid site
Temperature limit ( C)
Surface area (m /g)
650 ± 50
5.0 (calculated on dry basis)
−
+
R-SO3
120
(c)
H (a,b)
◦
2
a – Brönsted acid type.
b – pKa = –0,56 ± 0,5 (Estimated using ACD/Labs Software V11.02 based on the
monomer) [49].
c – Surface area (BET) is not measurable by standard means since it is the gel form
without porosity.
Regardless of the type of catalyst or solvent used, most of
the studies previously carried out were carried out in batch type
reactors [3–5,8,9,12,14,15,18–21,27,28,30–33,36–44]. The use of
continuous reactors was less explored [10,47,48]. One study carried
out with both batch and continuous reactors [11] was performed
with a didactic purpose.
This study presents the results of a research comparing two
reactor systems; batch reactor (BTR) and a continuous packed bed
reactor (PBR), for the reaction of fructose dehydration to form HMF
using a strongly acidic resin as catalyst and DMSO as a solvent. The
objective of the study was to evaluate the kinetics of the reaction
that fit the data and explain the mechanism and the performance
of the reactors.
2. Experimental design
According to Kuster [6], by using ionic exchange resins as cata-
lysts, the formation of insoluble humins must be avoided. To that
end, the use of dimethyl sulfoxide (DMSO) is recommended with
a solvent or biphasic system for the continuous extraction of HMF
from the reaction medium. The solvent is a determining factor for
the conversion reaction from fructose to HMF, as it promotes fluid-
ity and contact between the reagent and the catalyst. In choosing
the choice of the reaction medium, the solvation properties must
be carefully considered because it is economically recommended
to work with high concentrations of the solute, which may increase
the probability of humin formation, depending on the solvent [6].
The commonly applied solvents are water, methanol, DMSO,
organic/aqueous mixtures, ionic liquids and biphasic systems [14].
The use of an aqueous solvent is most advantageous from an eco-
nomic point of view – because it is less toxic and non-inflammable;
however, most solid acids lose their catalytic activity and stability
in an aqueous medium [6]. Additionally, anhydrous solvents such
as DMSO and ionic liquids foster higher yields and lower forma-
tions of humins and products from hydrolysis, such as formic and
levulinic acids [4].
d-Fructose was used as a reagent and dimethyl sulfoxide
(
DMSO) was used as a solvent in this study. Both chemicals were
®
purchased from Synth . The strong acid cationic gel resin obtained
®
from Purolite , SGC650H, which is a Brönsted acid type (character-
istics presented in Table 1) was used as a catalyst. Nitrogen from
®
Air Liquide was used for the inertization of the reaction medium®.
Acetonitrile (HPLC level) and glacial acetic acid (100%) from Merck
were used in the characterization by HPLC. HMF (99%) was obtained
®
from Sigma Aldrich to prepare the patterns and calibration curve.
2
.1. Batch reaction
Based on the methodology applied by Kimura et al. [15], fruc-
tose, DMSO (300 mL) and SGC650H resin (30 g) were added to a
stainless-steel Parr autoclave reactor with capacity for 1 L (Fig. 2).
The autoclave was equipped with a built-in heater, a stirrer
adjusted for 700 rpm and a sampling system equipped with a filter
steel screen) to prevent the loss of the catalyst. After its proper
closure, the reactor was cleansed with nitrogen to remove the oxy-
(
®
Please cite this article in press as: J. Tacacima, et al., Synthesis of HMF from fructose using Purolite strong acid catalyst: Comparison
between BTR and PBR reactor type for kinetics data acquisition, Mol. Catal. (2018), https://doi.org/10.1016/j.mcat.2017.12.021

G Model
MCAT-428; No. of Pages9
ARTICLE IN PRESS
J. Tacacima et al. / Molecular Catalysis xxx (2018) xxx–xxx
3
Fig. 2. BTR system.
Legend: 1–N2 purge; 2–Autoclave; 3–Controler; 4–Trap; 5–Bubblemeter.
gen from its interior. Afterward, the reactor was pressurized at
0 psig with nitrogen in order to facilitate subsequent samples, and
The concentration of HMF was based on the calibration curve.
Thus, their yield and conversion were defined as:
1
the heating up to the desired temperature was conducted, which
was maintained by the controller. The heating time of the tests
varied from 20 to 45 min until the reaction medium reached the
pre-defined temperature. At that stage, it was considered as the
initial reaction time. Ten-milliliter (10 mL) samples were removed
in the course of the reaction and cooled for further analysis.
(
HMF molar concentration)
Fructose molar concentration feed)
Y (%) =
∗ 100
(1)
(2)
(
ꢀ
ꢁ
(
Fructose molar concentration)
(Fructose molar concentration feed)
X (%) = 1 −
∗ 100
2
.4. Kinetic modeling
2.2. Continuous reaction
In this calculation method, a similar approach to Lee & Wu [31]
was used, who also considered the high selectivity of the trans-
formation of fructose to HMF using DMSO as a solvent in the
presence of sulfonic Brönsted acid catalysts [21]. The reaction was
initially considered as pseudo-first-order-homogeneous and inte-
grating the equation regarding the conversion. The rate law of the
reaction is defined by Eq. (3)
The continuous tests were carried out in a stainless-steel tube
prepared to behave like a Packed Bed Reactor. The tube was 82 mm
ꢀ
ꢀ
long, with an external diameter of ½ and an internal diameter of
0 mm into the reactor. 5.3 g of the raw SGC650H resin was placed
1
inside the reactor by the use of steel screens and glass wool on
the edges of the tube (Fig. 3). It was submerged in an oil bath to
control the desired temperature, and the raw solvent was initially
pumped into the reactor at 1 mL/min for approximately 20 min.
After the change for the solution of fructose feed, the beginning
of each test was considered after approximately three residence
times to ensure that the system was steady. The pump was deacti-
vated after completing each sampling process, and its outflow was
reset to start another test condition, using the same temperature
of reaction and feed concentration, but with a different flow rate.
After experimenting with each sample using different flow rate,
the entire system together with the pump and raw solvent was
cleaned up again before moving on the next experiment. The same
procedure was carried out while varying the parameters regarding
temperature and feed solution concentration.
n
rF = −k · CF
(3)
In which rF represents the index of the fructose reaction; k, the rate
constant; CF , the molar concentration of fructose in mol/L and n,
the reaction order (n = 1).
The molar balances of the limiting reagent in a BTR considering
a perfect mixture and in a PRF are, respectively:
dNF
=
rF · V
(4)
(5)
dt
dX
dV
FF 0
·
= −rF
In which NF represents the molar quantity of the limiting reagent
in this case, fructose) at the time t of the reaction; V, the reactional
(
Due to the high boiling temperature of the DMSO, its use as a
solvent allowed us to carry out tests at atmospheric pressure in a
volume; FF0, the initial molar outflow of the limiting reagent in
mol/h; and X, the reagent conversion.
Combining the rate laws with the molar balances, we can find
three first-class ordinary differential equations which, when inte-
grated, respectively result in:
◦
◦
continuous reactor at temperatures ranging from 90 C to 110 C.
2
.3. Products characterization
The samples were diluted in deionized water with factors vary-
− ln (1 − X) = k · t
− ln (1 − X) = k · ꢀ
In which ␶ means the average residence time or stay at the continu-
ous reactor, taking the volumetric outflow v0 constant throughout
the reactor’s volume, being defined by ␶ = V/v0.
(6)
(7)
ing from 1 to 400, for the characterization of HMF. The HPLC column
®
was a Shim-pack CLC-ODS (M) reversed-phase column operating
◦
at 30 C, binary eluent with a flow rate at 1.2 mL/min of acetonitrile
and acetic acid 1% (m/m) in the ratio of 20:80 and UV–vis detector
◦
at 266 nm at 40 C.
®
Please cite this article in press as: J. Tacacima, et al., Synthesis of HMF from fructose using Purolite strong acid catalyst: Comparison
between BTR and PBR reactor type for kinetics data acquisition, Mol. Catal. (2018), https://doi.org/10.1016/j.mcat.2017.12.021

G Model
MCAT-428; No. of Pages9
ARTICLE IN PRESS
4
J. Tacacima et al. / Molecular Catalysis xxx (2018) xxx–xxx
Fig. 3. PBR system.
Legend: 1–feed solution; 2–continuous pump; 3–thermostatic bath; 4–product sampling; 5–PBR reactor.
In the calculation of initial rates [16], the variances for high res-
idence times in the reactor which did not adjust to the model was
ignored and only the initial data that presented linear trend esti-
mation was considered, in a way that the inclination at the initial
stage corresponds to the kinetical constant k. In this study, this
inclination was calculated by the derivation of the equation of the
linear trend estimation at the initial stage (t = 0 h or ꢀ = 0 min). As
the system is heterogeneous, the molar balances and rate law can
be written based on the catalyst’s mass (W), in a way that the reac-
tion’s rates and rate constants are correlated by the concentration
of the catalyst according to:
Still considering a pseudo-first-order homogeneous reaction,
the reaction rate in which the variables were a function of time
was:
dCF (t)
dt
ꢀ
rF (t) =
= −k (t) · Ccat (t) · CF (t)
(11)
The reagent’s concentration was only known in the reactor’s
feed, in a way that it was simulated on an electronic spreadsheet
by optimization, stipulating a consistent ꢁt along the reaction.
ꢂ
ꢃ
ꢀ
(
CF )
= (CF ) − kt
· (Ccat) · (CF ) · ꢂt
(12)
t+ꢁt
t
t
t
t
ꢀ
rF = rF · Ccat
(8)
(9)
The values obtained (E and k0
ꢀ
) were adjusted to the experi-
a
mental data through the method of the least squares, in which the
ꢀ
k = k · Ccat
sum of the deviations to the power of two is minimized through
®
ꢀ
ꢀ
in which r F and k are based on the dry catalyst’s mass; Ccat is the
catalyst’s concentration which, in the BTR, was calculated at each
point of the sample, correcting the reactor’s volume after the sam-
pling rate without a catalyst. It was suppose that the density of the
reaction medium is approximately equal to the solvent’s (DMSO)
and that the solutes do not significantly interfere in this parameter.
In order to simplify the calculations of the differential equations for
the BTR, it was proposed that the reactional volume be constant and
that the catalyst’s concentration increase along the time, similarly
to a semi-batch reactor with catalyst feeding flow.
the Solver tool from Excel by the Non-linear GRG method when
varying the parameters studied.
2.4.2. LHHW method
The heterogeneous kinetic catalysis uses the same concentra-
tion and temperature principles; however, other factors are also
essential and must be added to the equation, such as the formation
of a complex on the surface and its formation rates.
Following the described methodology elsewhere [17], the deter-
mination of the rate equation by the Langmuir-Hinshelwood &
Hougen-Watson (LHHW) approach may be carried out by obtaining
data under conditions in which diffusion does not limit the reac-
tion. Moreover, by observing the behavior of the initial reaction
rate graph in function of the reagent’s initial concentration, which
may show indications regarding the limiting stage of the reaction:
adsorption, desorption or reaction on the surface.
With the values of the kinetic constants, it was possible to cal-
culate the reaction’s activation energy (E ) through the Arrhenius
A
equation:
ꢀ
ꢁ
EA
’
k’ = k · exp
−
(10)
0
R · T
ꢀ
in which k0 is the pre-exponential constant; T, the reaction’s abso-
The development of a rate law for the global reaction involves a
selection of a mechanism that describes each stage of the reaction,
which rate laws are treated as elementary. Considering that the one
from these steps is limiting, the concentrations of the adsorbed
species may be obtained by manipulating the expressions of the
non-limiting stages. Combining them with the expression of the
limiting stage and the balance of sites on the catalyst’s surface, a
possible rate equation for the global reaction can be obtained.
Several kinetic equations may be found, depending on the
accepted hypothesis, but always following:
lute temperature; and R, the universal constant of ideal gases
−1
(
83,145 J/mol K ).
The comparisons of the rate equations and the equation
obtained through the full method of each type of reactor are pre-
sented in Table 2, in which the ratio PBR:BTR was obtained at stages
with the same fructose conversion and concentration (in both reac-
BTR
PBR
BTR
PBR
tors), i.e., X
= X
and CF
= CF
.
2.4.1. Iterative method
One of the disadvantages of using the Parr reactor was the
temperature variation during the heating until it reached reaction
temperature, which hampered the kinetic analysis. For this rea-
son, an iterative optimization method was used to calculate the
Arrhenius parameters, applying the ones obtained through the full
method as an initial value.
reaction rate) = (
kinetic term) · (driving force or balance shift)
(
(
adsorption term)
(
13)
®
Please cite this article in press as: J. Tacacima, et al., Synthesis of HMF from fructose using Purolite strong acid catalyst: Comparison
between BTR and PBR reactor type for kinetics data acquisition, Mol. Catal. (2018), https://doi.org/10.1016/j.mcat.2017.12.021

G Model
MCAT-428; No. of Pages9
ARTICLE IN PRESS
J. Tacacima et al. / Molecular Catalysis xxx (2018) xxx–xxx
5
Table 2
Equations for comparison between batch and continuous reactors.
ꢂ
ꢃ
ꢃ
ꢀ
n
n
BTR
PBR
ꢀ
Batch reactor:
(rF )BTR = −k · (Ccat )
· (CF )BTR
· (CF )PBR
− ln 1 − X
= k · (Ccat )
· tBTR
· ꢀPBR
BTR
PBR
BTR
PBR
ꢂ
ꢀ
ꢀ
Continuous reactor:
Ratio PBR:BTR
(rF )PBR = −k · (Ccat )
− ln 1 − X
= k · (Ccat )
(
Ccat )PBR
(
Ccat )BTR
(rF )PBR
(rF )BTR
(Ccat )PBR
(Ccat )BTR
tBTR
ꢀPBR
=
=
Note: Overwritten data are referring to the variables in the respective reactor, BTR for the batch reactor and PBR for the continuous reactor.
If the rate laws linearly adjust the experimental data, the con-
stants may be determined through multiple linear regression.
Statistical interpretations were necessary to determine whether
the adjustments are significant or not. The coefficients obtained
may also be evaluated if the data are adjusted on characteristic
straight lines in the Arrhenius graphs for the rate coefficients and
in the Van’t Hoff graphs for the adsorption coefficients [17].
3
. Results and discussion
The high selectivity dehydration of fructose to HMF catalyzed
by the SGC650H acid resin in DMSO proved to be influenced by
the catalyst’s concentration, residence time, temperature and the
reagent’s initial concentration, among other factors. The main fac-
tor seems to be the absence of a porous system.
A blank test was carried out to verify if the DMSO solvent pre-
◦
sented any catalytical activity, in the batch reactor at 110 C with a
feed of 4.5% (m/m) of fructose without resin. After 2 h of reaction, it
did not present any apparent formation of HMF in the HPLC analy-
◦
ses. This fact is in accordance with a published study made at 130 C
[33]. Therefore, it was considered that the product generated was
catalyzed mainly by the resin under the experimental conditions
applied.
3.1. Reaction temperature effect
High yield from HMF was obtained by using fructose as a
reagent, SGC650H as catalyst and DMSO as a solvent in the BTR,
◦
which virtually reached 100% after 3 h of reaction at 100 C. The
use of the DMSO anhydrous solvent enabled us to keep the water
concentration low in the reaction medium, reducing the formation
of hydrolysis products and humin.
Fig. 4. Temperature effect on the reaction with 4.5 wt% fructose using SGC650H,
DMSO.
In Fig. 4, it is observed that the reaction was favored by the
rise in temperature, once the increase in HMF yield reveals a faster
consumption of the reagent, requiring a shorter reaction time. The
same effect was observed in different systems [36–41] which are in
accordance with that observed by Kuster [6]; and justifies that the
temperature rise effect favoring the reaction, is due to the activation
energy of the HMF formation being higher than the HMF consump-
tion. Besides, the enolization rate is higher at elevated temperatures
in a way that the acyclical forms and furanose of the fructose are
higher.
Low temperatures lead to low HMF yield, while elevated tem-
peratures foster the formation of by-products [7]. This fact can
be explained because the reaction is sensitive to the temperature,
significantly increasing the reaction rate due to the endothermic
nature of the fructose dehydration reaction. However, the HMF
selectivity decreases due to the formation of humins and soluble
polymers. Additionally, longer reaction times increase the water
formation, and they consequently accelerate the reaction of lev-
ulinic acid formation, as already observed [8,18–21].
Besides, it was possible to realize that its reaction rates in the PBR
were higher than in the BTR. The time needed to reach high yield
values were shorter, with values above 74% in 20 min at 90 C, in
◦
contrast to the values below 55% at an equivalent time for the batch
◦
reactor at 110 C (Fig. 4).
3.2. Fructose concentration feed effect
The use of high concentrations of substrate feed is desirable for
economic reasons [6]. However, this increases the possibility of col-
lisions among the reactive compounds, favoring the formation of
humins and polymers, thus decreasing selectivity. This effect was
observed through the fastest reaction tendency for the tests with
the lowest substrate concentrations, in order to demand shorter
reaction times to achieve the same yield (Fig. 4). In the experi-
ments with higher fructose feed concentration (13.5% m/m), the
HMF formation curve was in an inferior position in relation to the
other curves. This fact could be explained due to the increase or
appearance of limitations by transport phenomena or the increase
in the adsorption of reagents, solvent or products in the catalyst.
Furthermore, polymerization and formation of humins is more
probable to occur, thus decreasing the HMF selectivity, as observed
by [18,19,21,22] and these humins present a color ranging from
brown to black, which can be visually verified when comparing the
samples [6].
In the HMF yield profiles in the BTR, the differences in the
concentrations of the samples gotten at the time 0 h of the reac-
tion were noticeable, in which the reaction medium reached the
reaction temperature under study. These differences were more
◦
evident from the curves at 110 C, showing that the HMF product
was formed during the heating before the reaction’s initial time.
®
Please cite this article in press as: J. Tacacima, et al., Synthesis of HMF from fructose using Purolite strong acid catalyst: Comparison
between BTR and PBR reactor type for kinetics data acquisition, Mol. Catal. (2018), https://doi.org/10.1016/j.mcat.2017.12.021

G Model
MCAT-428; No. of Pages9
ARTICLE IN PRESS
J. Tacacima et al. / Molecular Catalysis xxx (2018) xxx–xxx
6
Table 3
Rate constants and Arrhenius parameters of the global reaction of HMF formation
obtained from the integral method.
4
−1
−1
−1
ꢀ
Reactor
CF0 (% m/m)
k’·10 (L gcat min
)
Ea (kJ mol
)
ln(k0 )
◦
◦
◦
110 C
90
C
100
C
BTR
BTR
BTR
BTR
PBR
PBR
1.5
3.0
4.5
13.5
4.5
1.180
1.069
1.024
0.917
1.800
1.575
2.043
1.969
1.961
1.632
3.767
5.757
5.095
3.335
2.543
5.391
91.41
90.14
68.35
59.07
59.96
21.14
20.65
13.46
10.28
11.33
13.5
The dry-based catalyst concentration was 450.0 g/L in
the PBR and approximately 56.5 g/L in the BTR, therefore
(
Ccat)PBR/(Ccat)BTR = 8.0 (Table 2). Which means that, under the
same temperature conditions and feed concentration, the average
residence time needed to reach the same conversion is 8.0 times
higher in the batch reactor, while the reaction rate is 8.0 times
higher in the continuous reactor. Experimentally, these behaviors
were confirmed with variations smaller than 5%.
An essential outcome of the catalyst concentration being higher
in the continuous reactor (PBR) is the lower probability of occur-
ring parallel or consecutive homogeneous reactions (at the liquid
phase), once the residence time in the PBR is shorter to reach the
same conversion values.
3.3. Kinetic modeling: integral method
Even without the conversion data, it is known that every HMF
molecule from the reaction medium is generated from fructose
molecule. Thus, the conversion must be equal, if there were no par-
allel reactions. Alternatively, conversion would be higher than the
HMF yield when by-products were generated. The conversion was
calculated based on this reaction stoichiometry and considering
that there were no secondary reactions from fructose, in a way that
the HMF selectivity was admitted as 100% [21,31].
Fig. 5. Fructose concentration feed effect on the reaction at 90 ◦C using SGC650H,
DMSO.
The values of the constants of the reaction rate in relation to
the catalyst mass presented in Table 3 correspond to the graphic
ꢀ
qualitative analyses obtained from Figs. 4 and 5: higher k for higher
reaction temperatures and lower for higher reagent concentrations.
The Arrhenius parameters presented in Table 3 were obtained
from the rate constants of the reaction in terms of catalyst mass
through linear Arrhenius graphs illustrated in Fig. 6.
It is observed that the curves presented in Fig. 6 tend to present
an intersection at a specific temperature. This tendency is known
as the isokinetic phenomenon of compensation or kinetic compen-
sation effect (KCE), which is confirmed by the Cremer-Constable
Diagram (Fig. 7) and shows a linear tendency between the values
Fig. 6. Arrhenius plot for different fructose concentration feed in the BTR.
ꢀ
obtained from ln(k0 ) and Ea. The variation of the activation energy
values observed frequently occurs in heterogeneous catalysis when
modifying controlled variables, in this case, the reagent concentra-
tion. If the reaction presents the compensation effect, the curves
of the Arrhenius diagram for different data sets cross at the same
point called isokinetic temperature [23]. In this study, the isokinetic
◦
temperature calculated is between 80 and 90 C.
3.4. Kinetic modeling: iterative method
The integral method calculates the kinetic constant at the begin-
ning of the reaction, presuming that the reaction’s temperature is
constant; however, in the BTR experiments, this temperature var-
ied during the heating so that the product had already been formed
before the time t = 0 h. The iterative method, by optimization,
treated the temperature as a variable to simulate the concentra-
Fig. 7. Cremer-Constable Diagram showing the compensation effect from the
results obtained for the BTR.
®
Please cite this article in press as: J. Tacacima, et al., Synthesis of HMF from fructose using Purolite strong acid catalyst: Comparison
between BTR and PBR reactor type for kinetics data acquisition, Mol. Catal. (2018), https://doi.org/10.1016/j.mcat.2017.12.021

G Model
MCAT-428; No. of Pages9
ARTICLE IN PRESS
J. Tacacima et al. / Molecular Catalysis xxx (2018) xxx–xxx
7
Table 4
Arrhenius parameters for the pseudo-first-order reaction obtained by the iterative
method.
ꢀ
Reactor
CF0 (%m/m)
Ea (kJ/mol)
ln (k0 )
BTR
BTR
BTR
BTR
1.5
3.0
4.5
13.5
91.09
89.66
67.80
58.84
21.14
20.65
13.46
10.28
tion of the reagent and calculated the Arrhenius parameters for the
pseudo-first-order reaction, which values are presented in Table 4.
Once again, a variation in the linear activation energy values for
different feed concentrations was observed, so that a compensation
ꢀ
effect of the ln(k0 ) pre-exponential term occurs, as presented in the
Cremer-Constable graph (Fig. 7).
Fig. 8. Graph of the initial reaction rate according to the initial fructose concentra-
tion.
Note: Tests carried out in batch with fructose feed, DMSO, and SGC650H Resin as a
catalyst.
Among the several probable causes for the compensation effect,
we can mention the reaction rates on the catalyst surfaces or active
sites varying in a different way since the mechanism involves the
formation of intermediates; variation of the reagent’s concentra-
tion, which is not the same as fed; the existence of parallel reactions
so that the same reaction occurs at different sites of the catalyst, or
between different sites of the same molecule [24–26].
The results for the activation energy obtained in this study are
similar to those obtained by Lee & Wu (67.5 and 80.05 kJ/mol)
with sulfonic ionic liquid functionalized mesoporous silica [31].
However, they were smaller than the ones obtained by Moreau
et al. [27,28], which is 141 kJ/mol for the reaction catalyzed by
H-mordenite zeolite in the biphasic solvent water-MIBK, and 143
occurrence of the compensation effect (KCE) showed previously
could indicate the existence of parallel processes which would
hamper the development of a representative kinetic model for the
reaction mechanism.
Studies using RMN [14,29], revealed not only the existence of
intermediates in the HMF formation reaction from fructose using
DMSO as a solvent but also showed the presence of five isomeric
species of fructose. The quantity of them varies according to the
temperature, and include an open chain form, ␣ and ␤ pyranoses,
␣ and ␤ furanoses. Among the later, ␤-d-fructofuranose shows
the higher concentration in reaction media. More recently, some
insights over the role of DMSO in fructose dehydration in the pres-
ence of Brönsted acid arise from new findings based on theoretical
calculations [35]. Besides the tautomeric forms of fructose, DMSO
molecules also could be protonated and has activity in the first
and third dehydration steps [35]. This way, it is possible that both
phenomena are responsible for the KCE observed. Therefore, a rep-
resentative model may take into account the balance between the
reactivity of different isomers of fructose and the role played by
the catalyst and DMSO as co-catalyst in the different steps of the
reaction.
+
−
ꢀ
kJ/mol for the ionic liquid HMIM Cl and ln(k0 ) = 39. The appar-
ent activation energy of the order of 53 kJ/mol was reported using
mesoporous zirconium phosphate catalysts in DMSO, presenting
higher reaction rates at a specific temperature [21]. As the data in
these study presents a compensation effect (KCE), any comparison
of the parameters obtained from catalysts used in different sys-
tems is impaired. Moreover both Arrhenius parameters are needed
for comparisons among different studies.
ꢀ
The Ea and ln(k0 ) data presented by Moureu et al. [28] are also
aligned in the Cremer-Constable Diagram with the data obtained
in this study, which may indicate a possible similarity regarding
reaction mechanisms. One possible explanation for this similarity
would be the reactivity difference among isomeric species of fruc-
tose which vary with temperature [14] and may vary with fructose
concentration.
4. Conclusion
The use of fructose as a reagent, SGC650H (gel-type strongly
3
.5. LHHW kinetic model
acidic resin) as catalyst and DMSO as solvent allowed for the obtain-
ing of high HMF selectivity and yield. It was observed that the
behavior of the fructose dehydration reaction for the formation
of HMF is favored by increase in temperature, reduction in feed
substrate concentrations and when a continuous reactor (PBR) was
used instead of a batch reactor (BTR).
The data from the experiments with the BTR obtained through
the integral method allowed us to plot the initial reaction rate
graph according to the reagent’s initial concentration (Fig. 8). From
the graph, it was possible to conclude that the desorption of the
product is not the limiting stage, as it did not present independent
CF0 profiles; however, it could not be deduced if the adsorption or
the reaction on the surface was the rate-limiting step, and there-
fore both hypotheses were tested. Evaluating several models using
both hypotheses, we noticed that the Van’t Hoff graphs obtained
incoherent signals for angular coefficients, allowing us to conclude
that no model used was correct. For this reason, it was concluded
that the mechanism of the fructose dehydration reaction is much
complicated.
A more detailed study should be carried out in the future to
obtain a rate law which applies to any temperature condition and
substrate concentration.The mechanisms proposed in the litera-
ture [14,29,30] showed that the reaction occurs with the formation
of intermediates so that the rate law might be obtained by using
the pseudo-stationary state phase approach (HEPE). However, the
The reaction’s rate constants obtained through the integral
method for the continuous packed bed reactor (PBR) were higher
than the ones for the batch reactor (BTR). This fact can partially be
explained by the calculation method applied, which determines the
rate constant at the reaction’s initial stage. Although it was possi-
ble to obtain important conclusions from the BTR results, the PBR
reactor proved to be more adequate for the kinetic study of hetero-
geneous catalysts for this system. Since the time needed to reach a
given conversion is shorter in the PBR than in the BTR, less inter-
ference of series or parallel homogeneous reactions is expected.
The Arrhenius parameters calculated for different concentra-
tions of fructose presented a characteristic variation known as
kinetic compensation effect (KCE). This variation of the parame-
ters obtained indicated that the calculated values are apparent and
that the pseudo-first-order pseudo-homogeneous kinetic model
®
Please cite this article in press as: J. Tacacima, et al., Synthesis of HMF from fructose using Purolite strong acid catalyst: Comparison
between BTR and PBR reactor type for kinetics data acquisition, Mol. Catal. (2018), https://doi.org/10.1016/j.mcat.2017.12.021

G Model
MCAT-428; No. of Pages9
ARTICLE IN PRESS
J. Tacacima et al. / Molecular Catalysis xxx (2018) xxx–xxx
8
adopted is insufficient to describe the reaction, as there would be
other relevant factors that should be considered in the rate law.
By applying the rate-limiting step (RLS) method and considering
the Langmuir-Hinshelwood (LHHW) approach with the experi-
mental data obtained in the BTR, it was not possible to define a
feasible model with the data available. Therefore, a kinetic model
with significant values for the heterogeneous reaction through the
methodology applied, was not obtained.
As a mechanism for the heterogeneous catalytic reaction could
not be determined, another method, i.e, the pseudo-stationary
phase hypothesis (HEPE) could be applied in the future to obtain
the rate law when considering series reactions. This model should
regard the existence of previously identified intermediates species
and the reactions that occurs in parallel, examining the existence
of fructose isomers, the chemical balance among them and the
interference of DMSO as co-catalyst.
[18] A. Chinnappan, A.H. Jadhav, H. Kim, W.-J. Chung, Ionic liquid with metal
complexes: an efficient catalyst for selective dehydration of fructose to
5-hydroxymethylfurfural, Chem. Eng. J. 237 (2014) 95–100.
[
19] H. Liu, C. Hua, C. Song, S. Dai, H. Wang, W. Zhu, H. Li, Commercially available
ammonium salt-catalyzed efficient dehydration of fructose to
5
-hydroxymethylfurfural in ionic liquid, Inorg. Chim. Acta 428 (2015) 32–36.
20] N. Lucas, G. Kokate, A. Nagpure, S. Chilukuri, Dehydration of fructose to
-hydroxymethyl furfural over ordered AlSBA-15 catalysts, Microporous
Mesoporous Mater. 181 (2013) 38–46.
[
5
[21] H. Xu, Z. Miao, H. Zhao, J. Yang, J. Zhao, H. Song, N. Liang, L. Chou, Dehydration
of fructose into 5-hydroxymethylfurfural by high stable ordered mesoporous
zirconium phosphate, Fuel 145 (2015) 234–240.
[22] K.-I. Shimizu, R. Uozumi, A. Satsuma, Enhanced production of
hydroxymethylfurfural from fructose with solid acid catalysts by simple
water removal methods, Catal. Commun. 10 (2009) 1849–1853.
[
23] G.C. Bond, M.A. Keane, H. Kral, J.A. Lercher, Compensation phenomena in
heterogeneous catalysis: general principles and a possible explanation, Catal.
Rev.: Sci. Eng. 42 (2000) 323–383.
[
24] P.E. Ferreira, J.G.R. Po c¸ o, Efeito Cinético de Compensa c¸ ão Observado na Rea c¸ ão
de Cura de Resina Epóxi (DGEBA) com Dicianodiamida, XX Congresso
Brasileiro de Engenharia Química, Blucher Chem. Eng. Proc. 1 (2015)
9852–9859.
[
[
25] J.G.R. Po c¸ o, H. Furlan, R. Giudici, A discussion on kinetic compensation effect
and anisotropy, J. Phys. Chem. B. 106 (2002) 4873–4877.
26] A.K. Galwey, Compensation behavior recognized in literature reports of
selected heterogeneous catalytic reactions: aspects of the comparative
analyses and significance of published kinetic data, Thermochim. Acta 294
Acknowledgements
We would like to thank CAPES (Coordena c¸ ão de
Aperfei c¸ oamento de Pessoal de Nível Superior) for the finan-
cial support, Centro Universitário da FEI and IPT (Instituto de
Pesquisas Tecnológicas do Estado de São Paulo) for providing us
with the essential resources to develop this study, and the Purolite
company for supplying the catalyst used in the experiments.
(
1997) 205–219.
[27] C. Moreau, R. Durand, S. Razigade, J. Duhamet, P. Faugeras, P. Rivalier, P. Ros,
G. Avignon, Dehydration of fructose to 5-hydroxymethylfurfural over
H-mordenites, Appl. Catal. A: Gen. 145 (1996) 211–224.
[
28] C. Moreau, A. Finiels, L. Vanoye, Dehydration of fructose and sucrose into
5-hydroxymethylfurfural in the presence of 1-H-3-methyl imidazolium
chloride acting both as solvent and catalyst, J. Mol. Catal. A: Chem. 253 (2006)
1
65–169.
References
[29] A.S. Amarasekara, L.D. Williams, C.C. Ebede, Mechanism of the dehydration of
◦
D-fructose to 5-hydroxymethylfurfural in dimethyl sulfoxide at 150 C: an
[
[
[
[
[
1] R. van Putten, J.C. van der Waal, E. de Jong, C.B. Rasrendra, H.J. Heeres, J.G.
Vries, Hydroxymethylfurfural, a versatile platform chemical made from
renewable resources, Chem. Rev. 113 (2013) 1499.
2] M. Kunz, Hydroxymethylfurfural, a possible basic chemical for industrial
intermediates, Studies in Plant Science, Inulin Inulin-Contain. Crops 3 (1993)
NMR study, Carbohydr. Res. 343 (2008) 3021–3024.
[30] J. Zhang, A. Das, R.S. Assary, L.A. Curtiss, E. Weitz, A combined experimental
and computational study of the mechanism of fructose dehydration to
5-hydroxymethylfurfural in dimethylsulfoxide using Amberlyst 70
3
−
PO4 /niobic acid, or sulfuric acid catalysts, Appl. Catal. B: Environ. 181
(2016) 874–887.
1
49–160.
3] A.J. Crisci, M.H. Tucker, J.A. Dumesic, S.L. Scott, Bifunctional solid catalysts for
the selective conversion of fructose to 5-Hydroxymethylfurfural, Top. Catal.
[31] Y.-Y. Lee, K.C.-W. Wu, Conversion and kinetics study of
fructose-to-5-hydroxymethylfurfural (HMF) using sulfonic acid and ionic
liquid groups bi-bifunctionalized mesoporous silica nanoparticles as
recyclable solid catalysts in DMSO systems, Phys. Chem. Chem. Phys. 14
(2012) 13914–13917.
[32] G. Tian, X. Tong, Y. Wang, Y. Yan, S. Xue, Highly efficient and
N-bromossucinamide mediated conversion of carbohydrates to
5-hydroxymethylfufural under mild conditions, Res. Chem. Intermed. 39
(2013) 3255–3263.
5
3 (2010) 1185–1192.
4] C. Carlini, M. Giuttari, A.M.R. Galletti, G. Sbrana, T. Armaroli, G. Busca,
Selective saccharides dehydration to 5-hydroxymethyl-2-furaldehyde by
heterogeneous niobium catalysts, Appl. Catal. A: Gen. 183 (1999) 295–302.
5] A. Takagaki, S. Nishimura, K. Ebitani, Catalytic transformations of
biomass-derived materials into value-added chemicals, Catal. Surv. Asia 16
(
2012) 164–182.
[
[
6] B.F.M. Kuster, 5-Hydroxymethylfurfural (HMF) – a review focusing on its
manufacture, Starch 42 (1990) 314–321.
7] M.E. Zakrzewska, E. Bogel-Łukasik, R. Bogel-Łukasik, Ionic liquid-mediated
formation of 5-hydroxymethylfurfural – a promising biomass-derived
building block, Chem. Rev. 111 (2011) 397–417.
8] A. Jain, A.M. Shore, S.C. Jonnalagadda, K.V. Ramanujachary, A. Mugweru,
Conversion of fructose, glucose and sucrose to 5-hydroxymethyl-2-furfural
over mesoporous zirconium phosphate catalyst, Appl. Catal. A: Gen. 489
[33] K.B. Sidhpuria, A.L. Daniel da Silva, T. Trindade, J.A. Coutinho, Supported ionic
liquid silica nanoparticles (SILnPs) as na efficient and recyclabe
heterogeneous catalyst for the dehydration of fructose to
5-hydroxymethylfurfural, Green Chem. 13 (2011) 340–348.
[34] S.K.R. Patil, J. Heltzel, C.R.F. Lund, Comparison of structural features of humins
formed catalytically from glucose, fructose, and
5-hydroxymethylfurfuraldehyde, Energy Fuels 26 (2012) 5281–5293.
[35] L.-K. Ren, L.-F. Zhu, T. Qi, J.-Q. Tang, H.-Q. Yang, C.-W. Hu, Performance of
dimethyl sulfoxide and Brönsted acid catalysts in fructose conversion to
5-hydroxymethylfurfural, ACS. Catal. 7 (2017) 2199–2212.
[36] G. Nie, X. Tong, Y. Zhang, Efficient production of 5-hydroxymethylfurfural
(HMF) from D-fructose and inulin with graphite derivatives as the catalysts,
Catal. Lett. 144 (2014) 1759–1765.
[37] X. Tong, Y. Li, Efficient and selective dehydration of fructose to
5-hydroxymethylfurfural catalyzed by Brönsted acidic ionic liquids, Chem.
Sus. Chem. 3 (2010) 350–355.
[
[
(
2015) 72–76.
9] F.S. Asghari, H. Yoshida, Dehydration of fructose to 5-hydroxymethylfurfural
in sub-critical water over heterogeneous zirconium phosphate catalysts,
Carbohydr. Res. 341 (2006) 2379–2387.
[
[
10] M.J. Antal, W.S.L. Mok Jr., G. Richards, Mechanism of formation of
-(hydroxymethyl)-2-furaldehyde from D-fructose and sucrose, Carbohydr.
5
Res. 199 (1990) 91–109.
11] S.P. Simeonov, C.A.M. Afonso, Batch and flow synthesis of
5
-hydroxymethylfurfural (HMF) from fructose as a bioplatform intermediate:
an experiment for the organic or analytical laboratory, J. Chem. Educ. 90
2013) 1373–1375.
[38] G. Tian, X. Tong, Y. Cheng, S. Xue, Tin-catalyzed efficient conversion or
carbohydrates for the production of 5-hydroxymethylfurfural in the presence
of quaternary ammonium salts, Carbohydr. Res. 370 (2013) 33–37.
[39] Y. Wang, X. Tong, Y. Yan, S. Xue, Y. Zhang, Efficient and selective conversion of
hexose to 5-hydroxymethylfurfural with tin-zirconium-containing
heterogeneous catalysts, Catal. Commun. 50 (2014) 38–43.
[40] X. Guo, J. Tang, B. Xiang, L. Zhu, H. Yang, C. Hu, Catalytic dehydration of
fructose into 5-hydroxymethylfurfural by a DMSO-like polymeric solid
organocatalyst, Chem. Cat. Chem. 9 (2017) 3218–3225.
[41] X. Tong, Y. Wang, G. Nie, Y. Yan, Selective dehydration of fructose and sucrose
to 5-hydroxymethyl-2-furfural with heterogeneous Ge(IV) catalysis, Environ.
Progress Sustein. Energy 34 (2015) 207–210.
[42] J. Dai, L. Zhu, D. Tang, X. Fu, J. Tang, X. Guo, C. Hu, Sulfonated polyaniline as
solid organocatalyst for dehydration of fructose into
(
[
12] N. Jiang, R. Huang, W. Qi, R. Su, Z. He, Effect of formic acid on conversion of
fructose to 5-hydroxymethylfurfural in aqueous/butanol media, Bioenergy
Res. 5 (2012) 380–386.
[
[
13] F.G. Helfferich, Ion Exchange, McGraw-Hill, New York, 1989.
14] H. Kimura, M. Nakahara, N. Matubayasi, Solvent effect on pathways and
mechanisms for D-fructose conversion to 5-Hydroxymethyl-2-furaldehyde:
in situ 13C NMR Study, J. Phys. Chem. A 117 (2013) 2102–2113.
[
15] V.T. Kimura, J.G.R. Po c¸ o, S. Derenzo, R.M.S. Matsubara, D.Z. Gomes, A.
Marcante, Obten c¸ ão de ácido levulínico e outros produtos a partir de a c¸ úcares
usandocatalisadores heterogêneos, XX Congresso Brasileiro de Engenharia
Química, Blücher Chem. Eng. Proc. 1 (2015) 9971–9978.
[
[
16] O. Levenspiel, Chemical Reaction Engineering, 3rd ed., John Wiley & Sons,
New York, 1999.
17] G.F. Froment, K.B. Bischoff, J. de Wilde, Chemical Reactor Analysis and Design,
5-hydroxymethylfurfural, Green Chem. 19 (2017) 1932–1939.
[43] L. Zhu, J. Dai, M. Liu, D. Tang, S. Liu, C. Hu, Formyl modified polyaniline for the
catalytic dehydration of fructose to 5- hydrosmethylfurfural, ChemSusChem 9
(2016) 2174–2181.
3
rd ed., John Wiley & Sons, New York, 2011.
®
Please cite this article in press as: J. Tacacima, et al., Synthesis of HMF from fructose using Purolite strong acid catalyst: Comparison
between BTR and PBR reactor type for kinetics data acquisition, Mol. Catal. (2018), https://doi.org/10.1016/j.mcat.2017.12.021

G Model
MCAT-428; No. of Pages9
ARTICLE IN PRESS
J. Tacacima et al. / Molecular Catalysis xxx (2018) xxx–xxx
9
[
[
[
44] H. Yan, Y. Yang, D. Tong, X. Xiang, C. Hu, Catalytic conversion glucose to
[47] P. Rivalier, J. Duhamet, C. Moreau, R. Durand, Development of a continuous
caalytic heterogeneous colum reactor with simultaneous extraction of an
intermediate product by an organic solvente circulating in countercurrent
manner with the aqueous phase, Catal. Today 24 (1995) 165–171.
[48] C.V. McNeff, D.T. Nowlan, L.C. McNeff, B. Yan, R.L. Fedie, Continuous
production of 5-hydroxymethylfufural from simple and complex
carbohydrates, Appl. Catal. A: Gen. 384 (2010) 65–69.
5
-hydroxymethylfurfural over SO42−/ZrO2 and SO4 /ZrO2-Al2O3 solid acid
2−
catalysts, Catal. Commun. 10 (2009) 1558–1563.
45] J. Tang, L. Zhu, X. Fu, J. Dai, X. Guo, C. Hu, Insights into the kinetics and
reaction network of aluminum chloride-catalyzed conversion of glucose in
NaCl-H2O/THF biphasic system, ACS Catal. 7 (2017) 256–266.
46] J. Tang, X. Guo, L. Zhu, C. Hu, Mechanistic study of glucose-to-fructose
+
isomerization in water catalyzeTd by [Al(OH)2(aq)] , ACS Catal. 5 (2015)
[49] Advanced Chemistry Development (ACD/Labs), Software V11.02 (©
1994–2017 ACD/Labs).
5
097–5103.
®
Please cite this article in press as: J. Tacacima, et al., Synthesis of HMF from fructose using Purolite strong acid catalyst: Comparison
between BTR and PBR reactor type for kinetics data acquisition, Mol. Catal. (2018), https://doi.org/10.1016/j.mcat.2017.12.021