Catalytic conversion of glucose to 5-hydroxymethylfurfural over biomass-based activated carbon catalyst

Article abstract of DOI:10.1016/j.cattod.2019.02.040

Selective and efficient dehydration of glucose to 5-hydroxymethylfurfural (HMF) has been widely explored research problem recently, especially from the perspective of more sustainable heterogeneous catalysts. In this study, activated carbon was first produced from a lignocellulosic waste material, birch sawdust. Novel heterogeneous catalysts were then prepared from activated carbon by adding Lewis or Br?nsted acid sites on the carbon surface. Prepared catalysts were used to convert glucose to HMF in biphasic water:THF system at 160 °C. The highest HMF yield and selectivity, 51% and 78%, respectively, were obtained in 8 h with a catalytic mixture containing both Lewis and Br?nsted acid sites. Also, preliminary recycling experiments were performed. Based on this study, biomass-based activated carbon catalysts show promise for the conversion of glucose to HMF.

Full text of DOI:10.1016/j.cattod.2019.02.040

Accepted Manuscript
Title: Catalytic conversion of glucose to
5-hydroxymethylfurfural over biomass-based activated carbon
catalyst
Authors: Annu Rusanen, Riikka Lahti, Katja Lappalainen,
Johanna K a¨ rkk a¨ inen, Tao Hu, Henrik Romar, Ulla Lassi
PII:
S0920-5861(18)31324-5
DOI:
Reference:
https://doi.org/10.1016/j.cattod.2019.02.040
CATTOD 11992
To appear in:
Catalysis Today
Received date:
Revised date:
Accepted date:
27 September 2018
10 January 2019
16 February 2019
Please cite this article as: Rusanen A, Lahti R, Lappalainen K, K a¨ rkk a¨ inen J, Hu T,
Romar H, Lassi U, Catalytic conversion of glucose to 5-hydroxymethylfurfural
over biomass-based activated carbon catalyst, Catalysis Today (2019),
https://doi.org/10.1016/j.cattod.2019.02.040
This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.

Catalytic conversion of glucose to 5-hydroxymethylfurfural over biomass-based
activated carbon catalyst
a
a,b
a,b
a
a
a
Annu Rusanen , Riikka Lahti , Katja Lappalainen , Johanna Kärkkäinen , Tao Hu , Henrik Romar , Ulla
Lassi*
,a,b
a
University of Oulu, Research Unit of Sustainable Chemistry, P.O. Box 4300, FIN-90014 Oulu, Finland
University of Jyväskylä, Kokkola University Consortium Chydenius, Talonpojankatu 2B, FIN-67100 Kokkola, Finland
b
*Corresponding author: ulla.lassi@oulu.fi
Graphical abstract
Highlights

Biomass-based activated carbon catalysts modified with Lewis or Brønsted acid sites were
prepared


Catalysts were used to convert glucose to HMF in biphasic water:THF system
51% HMF yield was obtained with catalytic mixture containing both Lewis and Brønsted acid
sites

The water phase containing the catalyst was recycled successfully
Abstract
Selective and efficient dehydration of glucose to 5-hydroxymethylfurfural (HMF) has been widely explored
research problem recently, especially from the perspective of more sustainable heterogeneous catalysts. In
this study, activated carbon was first produced from a lignocellulosic waste material, birch sawdust. Novel
heterogeneous catalysts were then prepared from activated carbon by adding Lewis or Brønsted acid sites
on the carbon surface. Prepared catalysts were used to convert glucose to HMF in biphasic water:THF
1

system at 160 °C. The highest HMF yield and selectivity, 51% and 78%, respectively, were obtained in 8
hours with a catalytic mixture containing both Lewis and Brønsted acid sites. Also, preliminary recycling
experiments were performed. Based on this study, biomass-based activated carbon catalysts show promise
for the conversion of glucose to HMF.
Keywords
Activated carbon; Catalyst; 5-hydroxymethylfurfural; Glucose conversion
1. Introduction
The limitations of fossil resources have driven global society toward bioeconomy, in which bioresources are
utilized in the production of biofuels and biochemicals. 5-hydroxymethylfurfural (HMF), a top value-added
biomass-derived chemical, has emerged as an important target product, since it represents a potential
substitute for petroleum-based monomers of various polymers and can be used as a starting material for
biofuels, solvents, and pharmaceuticals [1, 2]. HMF is a heterocyclic furanic molecule substituted in 2–5
positions with hydroxide and aldehyde functionalities (Figure 1), making it capable oxidizing into a
dicarboxylic acid or reducing to a diol, both of which can be used for the synthesis of polymers. In addition,
as a relatively unsaturated aromatic compound, HMF can be upgraded to a fuel via hydrogenation, and the
heterocyclic structure of furans can be found in an array of biologically active molecules with
pharmaceutical applications. [3]
HMF can be produced by the dehydration of C6 carbohydrates, including monomeric and polymeric
carbohydrates, such as fructose, glucose, sucrose, starch, cellulose, and raw biomass. The mechanistic
pathway of transformation is not yet clear, although a few possible alternatives have been proposed [4].
The most widely discussed route, when using fructose as feedstock, is direct dehydration of hexoses into
HMF, via either an acyclic or cyclic intermediate [5, 6]. When using hexoses other than fructose,
isomerization has been proposed as an initial step prior to dehydration (Figure 1) [4]. Because of this extra
step, the typical yields of HMF from fructose are superior to those obtained from glucose or other hexoses
under the same reaction conditions. Up to 99% yields have already been achieved in fructose conversion to
HMF [7]. Still, it should be emphasized that fructose is not an ideal feedstock because of its high price and
low abundance in nature [8]. It exists only in food biomass, such as in sugar cane or corn; in contrast,
glucose is the cheapest hexose and the most abundant monosaccharide, and is also available in non-food
lignocellulosic biomass. Therefore, the conversion of glucose to HMF is an important challenge worthy of
study.
There are many ways to produce HMF from glucose—for example, aqueous, organic, biphasic and
ionic liquid systems have all been utilized recently [4]. When compared to fructose conversion systems, the
challenge with glucose conversion is to find media wherein a rather stable glucose pyranose can transfer
effectively into a furanose form. Also, the low selectivity of the dehydration reaction, which is caused by
multiple side reactions, such as the formation of undesired byproducts, lowers the yield [9]. Formic acid
and levulinic acid can be formed via the rehydration of HMF [10]. Also, a cross-polymerization reaction
leads to the formation of soluble polymers and insoluble brown humins [11]. Biphasic systems are
promising media for conversion since the separation of HMF is simple and the system is recyclable, while
yields remain good (up to 81% in THF/water/NaCl system) [12, 13]. The most remarkable advantage of the
biphasic system is the continuous removal of HMF from the reaction mixture, which prevents rehydration
2

reactions and increases yields. DMSO:water, MIBK:water, and THF:water are the most-used biphasic
systems according to the literature [4, 14].
In addition to reaction media, the catalyst also plays an important role in reaction selectivity and
the dehydration rate. The dehydration step is generally catalyzed by mineral or organic acid. The catalyst
should either possess a proton or be a Lewis acid [15, 16]. The most commonly used Brønsted acids are
sulfuric acid, phosphoric acid, hydrochloric acid, oxalic acid, levulinic acid, and p-toluenesulfonic acid [4].
The isomerization step is catalyzed by Brønsted bases or Lewis acids [17]. However, the basic catalyst
typically leads to side reactions, so Lewis acids, such as metal chlorides, are favored [18]. Homogeneous
metal chlorides are capable of producing good HMF yields from glucose, e.g., AlCl
3
produced a 61% yield in
the biphasic system at 160 °C [19]. Also, tandem systems, in which the presence of both Lewis and
Brønsted acidity are combined, have been used, e.g., AlCl
yield in the biphasic system at 170 °C [20].
3
and HCl as a tandem system produced a 62%
However, traditional homogeneous catalysts are not easily recycled, they cause corrosion and incur
significant costs in separation and waste treatment processes. Heterogeneous catalysts, on the other hand,
are easily separated from the liquid reaction mixture after the reaction, enabling their reuse. Therefore, the
replacement of homogeneous catalysts with heterogeneous catalysts is preferred and is currently drawing
attention in the research field [21]. The most commonly used heterogeneous acid catalysts include metal
oxides, functional polymers, ion exchange resins, and zeolites [22-25]. As an alternative to them, novel
heterogeneous catalysts can be prepared from activated carbons (AC). ACs are low-cost materials
possessing large specific surface areas, well-developed, highly porous structures, chemical and physical
stability, and surface functionalities influencing the surface characteristics and adsorption behavior. ACs
can be inert or active in reactions. The carbon materials exhibit an acid–base character by containing
heteroatom, such as oxygen-bearing surface groups, which can have an effect on its properties. [26-29]
Furthermore, AC can be used as a catalyst support and be chemically functionalized and/or impregnated
with metals to improve catalytic activity [30-32]. ACs can be prepared from any carbon-containing raw
material, including bituminous coal and lignite, from which a large part of the ACs on the market is
produced [33]. Biomass-based ACs [34-36] are becoming more attractive, since they can be prepared from
lignocellulosic second-generation biomasses, side streams, and waste materials as a renewable raw
material, and as such contribute to carbon neutrality in the face of climate change.
Recently, a few studies have been published about fructose conversion to HMF by using modified
carbon catalysts: Deng et. al. used sodium ligninsulfonate derived AC as a catalyst support functionalized
with phosphoric acid, whereas Xiong et. al. used wood-based forestry biochar sulfonated with sulfuric acid
[
37, 38]. Both studies generated good HMF yields, —55.6% and 42.3%, respectively—using water as a
conversion solvent. In addition, other studies have used glucose as a feedstock: Zou et al. used bagasse-
based activated carbon and sulfonated it with sulfuric acid, while Villanueva et al. used carboxylated
activated carbon [39, 40]. However, both studies produced low HMF yields: 10%, at 130 °C in 8 hours; and <
5
%, at 125 °C in 3 hours, respectively. Meanwhile, Tyagi et al. combined Brønsted and Lewis acid sites in the
same carbon catalyst by impregnating metals into sulfuric acid-treated activated carbon [41]. These
researchers were able to produce a 49% HMF yield using cellulose as feedstock and an ionic liquid as a
reaction medium. In this work, AC produced from birch sawdust was modified with ZnCl
2
or H
2
SO to create
4
Lewis or Brønsted acid sites, respectively, on the AC surface. The objective was then to study the
effectiveness of prepared heterogeneous AC catalysts in the conversion of glucose to HMF in biphasic
water:THF system. The reaction conditions were studied in detail to find optimal catalyst loading and
reaction time. Furthermore, the recyclability of the catalyst containing water phase was studied with some
preliminary experiments.
3

Figure 1. General reaction scheme of HMF formation from glucose. First, glucose is isomerized into fructose
over a Lewis acid catalyst, and then fructose is dehydrated into HMF over a Brønsted acid catalyst.
2. Experimental
2
.1 Materials
All chemicals in this work were commercially available and were used without further purification. Birch
sawdust (Betula pendula) was received from a sawmill in Northern Sweden.
2
.2 Activated carbon support and catalyst preparation
Activated carbon was produced from birch sawdust. The sawdust was dried, carbonized and steam-
activated in a one-step process in a rotating quartz reactor (Nabertherm GmbH RSRB 80). The thermal
profile during the process was divided into two parts. In the carbonization step, which was performed
under an N
activation, the temperature was kept at 800 °C for 120 minutes with a stream of water steam and N
120 g/h at 140 °C and 200 ml/min, respectively). The resulting activated carbon (AC, Figure 2) was crushed
2
flow (200 ml/min), the temperature was raised to 800 °C with a ramp of 6.7 °C/min. During the
2
gas
(
and sieved to a fraction size of < 150 µm. Finally, it was either washed with hot water (ACW, Figure 2) or
used as such for further modification.
Modified solid acid catalysts were prepared by the reflux method from AC or the impregnation
method from ACW. To introduce Brønsted acid sites on the activated carbon surface, untreated AC was
heated in reflux system at 80 °C for 2 hours with 9 M or 18 M H
2
SO (ACB or ACB2, Figure 2). The amount of
4
acid was 10 mL per g of AC. After acid treatment, ACBs were filtered and washed with distilled water until
the pH of the filtrate was neutral; lastly, the ACBs were dried overnight at 105 ˚C. Dried ACBs were
extracted in boiling toluene for 2 hours in order to remove unreacted sulfuric acid, after which they were
water-washed and then dried overnight at 105 ˚C [42]. To introduce Lewis acid sites on the ACW surface,
catalysts were prepared by incipient-wetness impregnation method [43-44] of a precursor salt, ZnCl , and
2
selected to contain nominal 5 wt.% or 15 wt.% of zinc (ACL and ACL2, respectively, Figure 2). Impregnation
was performed overnight in a rotating mixer (Rotavapor) at room temperature. After impregnation, the
catalysts were dried overnight at 105 °C. The thermal calcination treatment was performed in a chemical
vapor deposition (CVD) oven at 550 °C for 2.5 hours with an N flow (240 ml/h per g of catalyst). In addition,
2
two catalytic mixtures, ACBL and ACBL2, were made by combining equal amounts of ACB and ACL or ACB2
and ACL2, respectively (Figure 2).
Figure 2. Flow chart for catalyst preparation: a) carbonization and activation step b) acid modification with
9
2
M H SO
4
c) acid modification with 18 M H
2
SO
4
d) washing with H
2
O e) impregnation with ZnCl
2
(5 wt.% Zn)
f) impregnation with ZnCl
2
(15 wt.% Zn) g) mixing in 1:1 ratio.
4

2
.3 Characterization of catalysts
The morphology of the catalyst particles was studied using a JEOL JEM-2200FS energy filtered transmission
electron microscope equipped with a scan generator (EFTEM/STEM). The catalyst samples were dispersed
in pure ethanol and pretreated in an ultrasonic bath for several minutes to create a microemulsion. A small
drop of the microemulsion was deposited on a copper grid pre-coated with carbon (Lacey/Carbon 200
Mesh Copper) and evaporated in air at room temperature. The accelerating voltage in the measurements
was 200 kV, while the resolution of the STEM image was 0.2 nm. The metal particle sizes were estimated
from high-resolution STEM images of each sample.
Specific surface areas were calculated from adsorption isotherms of N at isothermal conditions in
2
liquid nitrogen according to the Brunauer–Emmett–Teller (BET) theory [45]. Pore distribution was
calculated from the adsorption isotherms using the density functional theory (DFT) model [46]. The fresh
catalyst samples (about 100 mg) were weighted in a quartz tube. Samples were evacuated and heated to
1
40 °C to remove any adsorbed components as well as to reduce moisture. The measurements were
performed by a Micromeritics ASAP 2020 equipment.
The Zn, Na, K, Ca, and S contents of prepared AC and catalysts were measured by ICP-OES using a
Perkin Elmer Optima 5300 DV instrument. Samples of 0.1-0.2 g were first digested in a microwave oven
(
MARS, CEM Corporation) with 9 ml of HNO
mixture was digested at 200 °C for 10 minutes. Finally, 1 ml of HF was added, and the mixture was again
digested at 200 °C for 10 minutes. Excess HF was neutralized with H BO by heating at 170 °C for 10
3
at 200 °C for 10 minutes. Then, 3 ml of HCl was added, and the
3
3
minutes. Afterwards, the solution was diluted to 50 ml with water, and the former elements were analyzed
by the ICP-OES.
X-ray diffractograms were recorded by Rigaku SmartLab 9 kW X-ray diffraction (XRD) equipment
using Cu K
α
radiation (K filtered) at 45 kV and 200 mA. Diffractograms were collected in the 2θ range of 5–
β
1
00°, with a step size of 0.02° and a scan speed of 4.06 degree/min. The crystalline phases and structures
+
were analyzed by PDXL2 program and PDF-4 2018.
X-ray photoelectron spectroscopy (XPS) analyses were performed using the Thermo Fisher
Scientific ESCALAB 250Xi XPS System. With a pass energy of 20 eV and a spot size of 900 µm, the accuracy
of the reported binding energies (BEs) was ±0.2 eV. The Zn, C, O, S, and N were measured for all samples.
The measurement data were analyzed by Avantage software. The monochromatic AlKα radiation (1486.6
eV) was operated at 20 mA and 15 kV. Charge compensation of the BEs was performed by applying the C1s
line at 284.8 eV as a reference.
Temperature programmed desorption (TPD) of NH
system. Prior to the NH -TPD analysis, the samples (about 100 mg) were pre-treated in an He flow of 50
ml/min at 600 °C for 30 minutes. Afterward, the samples were cooled to 100 °C, and the adsorption of
ammonia (50 ml/min of 15% NH /He at 100 °C) was continued for 60 minutes. Prior to the desorption, the
samples were flushed in an He flow of 50 ml/min for 30 minutes to remove all reversibly adsorbed NH . The
NH desorption was carried out from 100 to 800 °C and left for 10 minutes at this temperature. During the
3
was performed by an AutoChem II 2920
3
3
3
3
analysis, a temperature ramp of 10 °C/min and an He flow rate of 50 ml/min were used. The total amount
of acid sites (mmol/g) were calculated.
2
.4 Conversion of glucose into HMF
5

In a typical reaction, 45 mg (0.25 mmol) of glucose, 0.35 g of NaCl, 1.5-18 mg of catalyst (corresponding to
.4-5.0 µmol of Zn and 0.7-23 mM H SO ), and a magnetic stirring bar were placed into a 5 ml reaction
0
2
4
tube. Water (1 ml) and THF (3 ml) were added, and the tube was sealed. The reaction was carried out in a
Biotage Initiator microwave reactor at 160 °C for 30 minutes to 8 hours. After the reaction, a sample was
taken from both phases, which were filtered through a 0.45 µm PTFE filter for further analysis. The pH of
the water phase was checked before and after the reaction. In the recycling experiments, the organic phase
was removed after the reaction, and a new portion of glucose and THF was added to a recycled water
phase, which included the catalyst. Recycling continued for five runs. All reactions were duplicated.
2
.5 Analytical methods
High-performance liquid chromatography (HPLC) and gas chromatography (GC) were used as analytical
methods to determine the amount of HMF in the reaction solutions. The water phase was analyzed with
HPLC, while the organic phase was analyzed with HPLC and GC. HPLC analysis was carried out using a
Waters 2695 separation module fitted with an Atlantis T3 (3 µm, 4.6 x 150 mm) column and a Waters 996
photodiode array (PDA) detector. A water:methanol (90:10) mixture was used as the mobile phase, with a
flow rate of 1 ml/min. The injection volume was 4 μl. The column temperature was kept constant at 30 °C,
and the UV detection for HMF was performed at 284 nm. GC analysis was carried out using an HP 6890
Series GC system fitted with a Sulpelco SPB-1701 (0.25 µm, 15 m x 0.25 mm) column and an HP 5973 mass
selective detector. The oven temperature program was as follows: from 60 °C (hold 1 min) at 20 °C/min to
2
50 °C (hold 2 min), with the injection and detection temperatures set at 250 °C. Helium was used as a
carrier gas, with a flow of 1.5 ml/min. GC-MS was also used to qualitatively determine the byproducts of
the conversion reaction. Amounts of byproducts other than levulinic acid were not determined. HMF and
LA calibrations were performed using analytical standards (Sigma Aldrich).
The yield of HMF was calculated from the equation:
Y
HMF (%) = [concentration of HMF in the sample / theorethical maximum concentration of HMF in the
sample] x 100%
The yield of LA was calculated using the same principle.
To determine the reaction selectivity, the amount of glucose from water phase was measured after the
conversion reaction. Glucose was measured with YSI 2700 Select Biochemistry Analyzer with glucose
solution (2.5 g/l) as a standard. The selectivity of HMF production was then calculated using the following
equation:
Selectivity (%) = [moles of HMF produced / (moles of glucose at the beginning - moles of glucose after the
reaction)] x 100%
3. Results and discussion
3
.1 Characterization of catalysts
AC catalysts prepared in this work were ACW, two Lewis acid catalysts (ACL and ACL2), and two Brønsted
acid catalysts (ACB and ACB2). Prepared catalysts were characterized with multiple characterization
6

techniques to verify their catalytic composition, morphology, and surface properties, e.g., surface areas,
active metal phases, and functional groups.
Catalyst morphology was observed with transmission electron microscope (EFTEM/STEM). From
STEM images (Fig. S1), no major differences in the surface composition were detected when the ACW
surface was compared to the surfaces of ACB and ACB2. Instead, on Lewis acid catalysts (ACL and ACL2)
prepared from ZnCl , metal particles were clearly seen and evenly distributed on the surface of the catalyst.
2
Metal particle sizes in both ACL and ACL2 catalyst were ca. 10–50 nm; however, some metal aggregations
around 100 nm were also detected on both catalysts.
Specific BET surface areas, average pore volumes, and pore size distributions of the prepared AC
support and catalysts were calculated from nitrogen adsorption isotherms by the BET and DFT methods
2
3
(
Table S1). AC had a BET surface area of 800 m /g. The pore volumes were ca. 0.5 cm /g with mesopore (2–
5
0 nm) volumes of 52% according to the DFT model. Moreover, all catalysts had high BET surface areas,
2
from 560 to 860 m /g, with high mesoporous volumes, which are usually preferred in catalytic systems. The
mesoporous structure was preferred, since smaller pores can easily be blocked, especially in the liquid
phase. Chemical modification of the AC support with concentrated sulfuric acid seemed to have an impact
on the surface pore composition. The treatment opened mesoporous structure of the AC surface.
Therefore, there was a 20% increase in the mesopore volumes and a decrease in micropore volumes in
2
ACB2 compared to AC. As the result, the surface area decreased to 700 m /g (Table S1). When AC was
treated with 9 M H
2
SO , both surface area and meso and micropore volumes decreased (ACB, Table S1).
4
This might be due to some pore blocking or the addition of functional groups into the pores. To prepare
ACW catalyst AC was washed with water, which did not alter the surface area or the meso and micropore
volumes considerably (Table S1). However, when ACW was further impregnated with zinc precursor, the
surface areas and pore volumes decreased slightly, ca. 3-5% compared to ACW (ACL and ACL2, Table S1).
This indicated that metal was present in the pores and surfaces of the ACL and ACL2 catalysts.
The Zn and S contents of the ACW, ACB, ACB2, ACL and ACL2 catalysts were measured with ICP-
OES. The nominal active metal contents in ACL and ACL2 were 5 or 15 wt.% of Zn, respectively. However,
their Zn contents measured by ICP-OES were similar, 1.7 wt.% for ACL and 1.8 wt.% for ACL2. Therefore, the
detected zinc concentrations in catalysts ACL and ACL2 were lower than nominal. It has been proposed in
literature that during the heat treatment process, in the temperature range between 400-600 °C, the ZnCl
2
dissociates to Zn and Cl
temperature was below the boiling point of ZnCl
presence of oxygen can occur [47, 48]. Since no O
2
. Therefore, the evaporation of Zn and Cl
2
could have occurred, though the
2
(b.p.732°C) [47]. Further, reaction of ZnCl
2
into ZnO in the
) of calcination
2
was present in the inert atmosphere (N
2
process, it is possible that Zn was bonded only with the oxygen atoms on the surface of AC. No zinc was
detected in ACB and ACB2 catalysts, however, sulfur was present 0.37 wt.% (ACB) or 0.24 wt.% (ACB2)
indicating the addition of sulfur during the H
2
SO treatment. In all catalysts, other impurity metals, e.g., K,
4
Na, were present in negligible amounts (< 0.2 wt.%). Although some calcium was detected from the ACL
catalysts (< 1 wt.%), in ACBs it was present in negligible amounts (< 0.2 wt.%).
Metal phases in the catalysts were observed with XRD (Fig. S2). In all catalysts, the assignments
from carbon (JCPDS file no. 01-082-9929) were detected at 2θ=24.6° and 2θ=43.7°, with broad peaks
indicating the amorphous phase of the carbon. For AC, ACW, ACB, ACL, and ACL2, phase identification was
detected for polymorphic silica (tridymite, SiO
θ=20.5, 21.6, 23.2, 29.9, and 35.8°. In ACB2, no silica was detected with XRD analysis. This indicates that
the treatment with 18 M H SO removes most of the metal impurities (e.g., water-soluble metals and silica
2
) according to JCPDS file no. 04-012-1133, with peaks at
2
2
4
originating from biomass) from AC, since no peaks other than the one for carbon were detected from the
ACB2. This was also verified with ICP-OES analysis. For the Lewis acid catalysts, ACL and ACL2, JCPDS file no.
7

0
4-004-4531 presented zinc oxide corresponding to ZnO(100), (002), (101), (102), (110), (103), and (112),
0
with peaks at 2θ=31.7, 34.4, 36.2, 47.5, 56.6, 62.8, and 68.0°, respectively. No ZnCl
2
or metallic Zn was
detected, which confirms that ZnCl
catalyst surface.
2
was decomposed or bonded to the oxygen functionalities on the
XPS was performed to determine the presence and abundance of oxygen- and sulfur-containing
functionalities on the catalyst surface (Table 1). Also, information about the presence and relative oxidation
state of the active metal was collected. For all catalysts, the XPS C1s spectrum revealed a high amount of
carbon functionalities (total C-% from 87 to 96%, see Table 1) on the surface, consisting mainly of carbon-
carbon-type bonds and some carbon-oxygen functionalities. From the O1s pattern (Table 1, Fig. S3), both
ACB and ACB2 catalysts were showing higher atomic percent for oxygen functionalities on the surface
indicating surface oxidation. For ACB2 the oxygen content, ca. 12%, was higher than for ACB (ca. 6%, Table
1
). From the O1s pattern, the signal at 531 eV can be assigned to C=O double bonds in carbonyl groups as
well as to S=O double bonds in sulfonic acid groups. The peak at 533 eV can correspond to the single-
bonded oxygen atoms, e.g., from hydroxyl groups of phenol type, or to S-O bonds e.g., in sulfonic acid
groups [49-51]. In addition, the peak at 168 eV in the XPS spectra for S2p (Fig. S4) indicated the presence of
sulfonic acid groups in both ACB and ACB2 [52]. The high-resolution XPS spectra of ACL and ACL2 (Fig. S4)
for Zn2p showed peaks at 1022 eV, corresponding to Zn2p3/2, and at 1045 eV assigned to Zn2p1/2. This
2+
indicated the presence of zinc oxide (Zn ) in the catalysts [53], which was also in accordance with the XRD
analysis. For ACW, the total amount of oxygen groups detected on the surface with XPS O1s scan was about
4
%. This indicates the presence of for example carboxylic acid groups, which are common in the surfaces of
activated carbons [27].
Table 1. Atomic percentages (atom-%) of surface groups based on the XPS analysis from catalysts obtained
via binding energies of the C1s, O1s, S2p and Zn2p photoelectrons [32-35].
Sample
ACW
%
ACB
%
ACB2
%
ACL
%
ACL2
%
BE (eV)a
b
b
b
b
b
Total C-%
From C1s
(284.8 eV)
(285.7 eV)
(288.3 eV)
(290.9 eV)
(293.4 eV)
95.7
52.2
25.2
7.7
8.1
2.6
93.4
49.7
25.8
7.8
7.7
2.5
87.4
43.0
26.5
9.7
6.3
2.0
95.4
51.9
25.6
7.4
8.0
2.5
94.5
50.6
25.7
7.5
8.3
2.5
3
C- (sp )
2
C= (sp )
C-O
C=O
π–π* in
aromatic
ring
Total O-%
O=
O-
From O1s
(531 eV)
(533 eV)
(536 eV)
(168 eV)
(1022 eV)
4.3
2.2
2.1
n.d.
n.d.
n.d.
6.3
1.3
4.3
0.6
0.2
n.d.
12.4
3.7
7.6
1.2
0.1
3.9
2.4
1.2
0.3
n.d.
0.4
4.6
3.2
1.0
0.4
n.d.
0.6
2
H O (ads.)
Total S-%
Total Zn-%
n.d.
a
b
binding energy ± 0.2 eV
is the relative amount from atom percent of the total sample
%
The total amount of acid sites on the surfaces of ACW, ACB, ACB2, ACL and ACL2 was determined
by temperature-programmed desorption of ammonia (NH -TPD). The position and the area of the
3
8

desorption peak correlates directly with the acid strength and the acid amount, respectively. The
desorption process of ammonia was measured up to 800 °C. From the TPD profiles (Fig. S5) desorption
peaks at 700 °C for ACB2 and at 750 °C for ACL and ACL2 were detected. These high-temperature peaks can
be attributed to the desorption of NH
peak between 500 and 700 °C [54] and from more thermally stable strong Lewis acid sites at higher
temperature, respectively [50, 55]. According to NH -TPD-performed analysis, ACW might contain some
3
from strong Brønsted acid sites with strong and broad desorption
3
amount of acidic surface groups, up to a total of 0.45 mmol/g, which could originate from oxygen-bearing
functionalities, e.g. carboxylic acids, anhydrides, lactones, lactols or phenols which are typical acidic groups
on the activated carbon surface [27]. The number of total acid sites introduced into the catalysts was, at
highest, 1.32 mmol/g when the AC was modified with 18 M H
for ACW indicating addition of acidic functionalities on the surface. As expected, treatment with 9 M H
2
SO
4
(ACB2) and was three times higher than
SO
2
4
did not add acidity as much as the treatment with strong acid. In fact, the acidity of ACB was 0.45 mmol/g,
the same as for ACW. It is very likely, that detected sulfur compounds on ACB (see ICP-OES and XPS) were
composed from other type of sulfur groups than from acidic sulfonic groups. According to literature, a large
number of oxidized functionalities such as phenolic and carboxylic acid groups can be created on the
carbon surface during the sulfuric acid treatment [56-59]. This seems to be in line with the results in this
study, since measured sulfur content seemed to be relatively low, but acidity was increased (ACB2). I.e. not
only sulfonic acids were formed during the H
2
SO treatment but also other oxidized functionalities such as
4
acidic phenolic and carboxylic groups, for example, were formed. Treatment with zinc chloride seemed to
increase acidity slightly in the ACLs, since metal oxide (ZnO) was present as a Lewis acid site in the catalyst.
The acidity determined by NH -TPD was slightly higher in ACL (0.75 mmol/g) than in the ACL2 catalyst (0.62
3
mmol/g). However, no significant differences were detected with acid sites in ACL and ACL2, confirming
that the catalysts were almost identical.
To conclude the catalyst characterization results, ACB and ACB2 had sulfur-containing
functionalities on the catalyst surface. Based on the NH -TPD analysis, sulfuric acid treatment with
3
concentrated acid increased the acidity of the ACB2 catalyst. However, also other oxidized functionalities
than sulfonic acids seemed to be present such as phenol or carboxylic acid groups increasing the acidic sites
on the surface. With the characterization of ACL and ACL2, the addition of zinc was detected with STEM and
ICP-OES analysis. Further, the metal phase as metal oxide (ZnO) was verified with XRD and XPS. Overall,
both Lewis acid catalysts were almost identical, even though, the nominal metal content was different,
indicating ZnCl impregnation was not successful at higher loading. In the future, optimization of the
2
preparation method of activated carbon catalysts containing Lewis and Brønsted acid sites still needs work
to obtain higher acidic site loading on the catalyst.
3
.2 Conversion of glucose into HMF
Prepared heterogeneous catalysts (ACW, ACB, ACB2, ACL, ACL2, ACBL and ACBL2) were used to convert
glucose to HMF in biphasic media consisting of water and THF. A reaction temperature of 160 °C and a time
of 8 hours, were chosen based on some preliminary experiments and were kept constant (Table 2). The first
reaction was carried out without catalyst and only a 1% HMF yield was obtained, which indicated that the
THF/water mixture could not catalyze the reaction by itself and acted only as a solvent (entry 1). Since THF
and water are mutually miscible, NaCl was added to the reaction vessel in order to saturate the water
phase and separate the layers. In the biphasic system, the produced HMF was extracted into THF in situ,
hindering contact between HMF and water. This prevented HMF from reacting further into undesired
byproducts. However, NaCl has been shown to catalyze HMF production, since the chloride ions enhance
9

the glucose to fructose isomerization step as well as the fructose to HMF dehydration step [60]. Therefore,
another reference reaction with NaCl but without any other catalyst was performed, and a HMF yield of
3
5% was achieved (entry 2). The reaction selectivity was only 44%, though. When prepared catalysts were
introduced to the reaction system, the HMF yield and the reaction selectivity increased. With ACW the yield
increased by 9 percentage units to 44% (entry 3), which was probably due to the acidity of oxygen-
containing groups on the surface of the ACW (XPS, Table 1). The reaction selectivity increased to 57%.
When ACB, ACB2, ACL or ACL2 were used as the catalyst, the HMF yield increased to 46–49% (entries 4-7),
indicating a slight increase compared to NaCl- and ACW-catalyzed reactions. ACB2 and ACL2 produced
slightly higher HMF yields than ACB and ACL, but the difference was only 2 percentage units in both cases
(entries 4-5 and 6-7). Based on catalyst characterization results, similar HMF yields between ACL and ACL2
were reasonable since their metal contents differed only by 0.1 wt.% according to ICP-OES, and 0.2 atom-%
according to XPS. Also, similar HMF yields, 46% and 48%, achieved with ACB and ACB2 catalysts,
respectively, were supported by the characterization results with XPS. HMF yields achieved with ACB and
ACB2 were excellent compared to those described in the literature, where the isomerization step is not
usually catalyzed successfully by acid-treated carbon catalysts [61]. It must be noted, though that NaCl was
present in the reaction system and most likely took part in the isomerization of glucose to fructose [60].
Similar yields between ACB and ACL as well as between ACB2 and ACL2 (difference of only 1 percentage
unit, entries 4–7) can also be explained by the effect of NaCl. Since the HMF yields achieved with ACL, ACL2,
ACB and ACB2 catalysts were similar, the reaction selectivity was determined only for the reactions
performed with ACB2 and ACL2 catalysts. Both catalysts functioned equally resulting in reaction selectivity
of 67% (entries 5 and 7), which was 10 percentage units higher than with ACW. This indicated that even if
the catalyst modification did not have a strong effect on the HMF yield, it increased the reaction selectivity
considerably. However, the selectivity was not affected by whether the catalyst was modified with Lewis or
Brønsted acid.
Creating tandem sites in the same AC catalyst is problematic because, after the sulfonation of AC,
impregnating zinc in the catalyst without destroying the oxygen-bearing sulfonic acid groups is challenging,
since high temperatures (> 250 °C) during the calcination process can destroy these sulfonic groups [62].
Therefore, two catalytic mixtures, ACBL and ACBL2, were prepared and used in the conversion reaction to
introduce Brønsted and Lewis acid functionalities in the reaction system simultaneously. With ACBL as the
catalyst, similar HMF yields (48%, entry 8) as those achieved with the ACB, ACB2, ACL, and ACL2 catalysts
were generated. However, with ACBL2 as the catalyst, the HMF yield increased to 51% (entry 9). Also, the
reaction selectivity increased further to 78% when compared to single acid catalysts ACB2 or ACL2.
Therefore, based on the results, it seems that both Brønsted and Lewis acid sites not only accelerate the
conversion reaction but also affect the reaction selectivity. The increase in HMF yield was not as high as
expected (only 2 percentage units compared to ACL2) but the increase in reaction selectivity was significant
(
6
11 percentage units). In the literature, HMF yields of 63%, 53% and 23.1% and reaction selectivities of 63%,
0% and 24%, respectively, have been achieved in similar reaction media (water/THF/NaCl) with solid TiO
ZrO +Amberlyst 70, Sn-Beta with NH F and FePO catalysts, respectively [22, 25, 63]. The HMF yield of 51%
achieved in this study is comparable to those obtained in above mentioned studies and HMF selectivity of
8% is notably better compared to those. Also, the studied catalytic mixture was prepared using more
2
-
2
4
4
7
sustainable lignocellulosic waste as a raw material for the catalysts support, thereby increasing the value of
the obtained results.
The effect of solvent system was studied with the ACBL2 catalyst, which produced the highest HMF
yield in previous reactions (Table 2). When the reaction was carried out in water, the HMF yield was low:
1
5% (entry 10). This indicates that although the biphasic system was important the lack of NaCl could have
1
0

attributed to the low HMF yield. Therefore, the reaction was repeated using NaCl. However, the HMF yield
was even lower—only 10% (entry 11). Based on GC-MS spectra, it was concluded that notable amount of
levulinic acid (14%) was formed when NaCl was introduced into the reaction and only water was used as
the solvent. In the biphasic reactions, only trace amounts (< 4%) of levulinic acid were detected.
Finally, the studied ACB2, ACL2, and ACBL2 catalysts were compared to homogeneous catalysts, in
which the amounts of H
2
SO and Zn were similar to those in heterogeneous catalysts. With homogeneous
4
catalysts the HMF yields as well as the reaction selectivities were lower than with the studied AC-based
catalysts, only 33–37% and 43-47%, respectively (entries 12–14). Low selectivities are consistent with
literature since homogeneous catalysts and NaCl increase the production of soluble humins and humin
precursors [64, 65, 50]. Low HMF yields are also reasonable since the amounts of H
very low in homogeneous systems.
2
SO
4
and/or Zn were
Table 2. Results for conversion of glucose to HMF using various catalysts and solutions.
Entry System
Catalyst
-
-
ACW
ACB
ACB2
ACL
ACL2
ACBL
ACBL2
ACBL2
HMF yield (%) Selectivity (%)
1
2
3
4
5
6
7
8
9
1
1
1
1
1
water/THF
1
-
water/THF/NaCl
water/THF/NaCl
water/THF/NaCl
water/THF/NaCl
water/THF/NaCl
water/THF/NaCl
water/THF/NaCl
water/THF/NaCl
water
35
44
46
48
47
49
48
51
15
10
37
35
33
44
57
-
67
-
67
-
78
-
0
1
2
3
4
water/NaCl
ACBL2
-
a
water/THF/NaCl
water/THF/NaCl
water/THF/NaCl
H
ZnCl
2
SO
4
b
47
43
45
2
H
2
SO4 + ZnCl c
2
a
Reaction conditions: 3 mg catalyst, 45 mg glucose, 0.35 g NaCl, 1 ml water, 3 ml THF, 160 °C, 8 h. 1.944
b
c
mM H
SO
2
SO
4
(corresponding to 3 mg ACB2), 0.1158 g/l ZnCl
(corresponding to 3 mg ACBL2).
2
(corresponding to 3 mg ACL2), 0.972 mM
H
2
4
and 0.0579 g/l ZnCl
2
The results from the glucose to HMF conversion reactions suggested that activated carbon itself
had an important role as a catalyst during the reaction. Based on the catalyst characterizations the
amounts of Lewis and Brønsted acid sites were low in heterogeneous catalysts, so it is likely that activated
carbon’s high specific surface area and oxygen-containing surface groups had an impact on its catalytic
activity. These groups might participate in the dehydration reaction of fructose to HMF, while the large
surface area of activated carbon may promote the conversion of glucose by increasing contact areas
between reactants [26, 29]. Besides this study, there are also several examples in the literature of how the
surface of activated carbon plays an important role in catalytic reactions, such as alcohol dehydration [27,
6
6, 67]. However, from the reaction selectivity point of view the values achieved in this study with plain
water washed activated carbon were considerably lower than those with acid treated carbon catalysts.
3
.2.1 Optimization of reaction time and catalyst loading
1
1

ACBL2 was chosen for optimization experiments because it produced the highest HMF yield compared to
the other catalysts (Table 2). The same reaction temperature, 160 °C, and a biphasic water:THF system
were used as in previous reactions, but the reaction time and catalyst loading were varied in order to find
the optimal reaction conditions for the conversion of glucose to HMF. As shown in Figure 3a, the HMF yield
increased from 8% to 51% when the reaction time increased from 1 to 8 hours, respectively. With a longer
reaction time, the HMF yield started to decrease, indicating that 8 hours was the optimal time to maximize
the yield. In addition, the amount of 1,4-butanediol, which is the degradation product of THF, was observed
to increase with the increasing reaction time. However, the most significant increase in the amount of 1,4-
butanediol occurred when the reaction time was 10 hours, implying that 8 hours may be the optimal
reaction time also regarding the stability of the biphasic system. The amount of levulinic acid did not seem
to increase with increasing time, which implied that the biphasic system worked successfully and prevented
rehydration of HMF in the organic phase.
The amount of catalyst was varied in the range of 1.5–18 mg to determine the optimal catalyst
loading. As shown in Figure 3b, the catalyst amount of 1.5 mg was clearly too low to accelerate the
conversion reaction, but when the amount was increased to 3 mg, the HMF yield improved significantly,
from 45 to 51%. When the catalyst loading was increased to 9 mg, the increase in the HMF yield was only 2
percentage units. A further increase in the catalyst amount caused the HMF yield to decrease (Figure 3b),
which may be due to the higher amount of Brønsted acid or metal sites in the reaction system, which
accelerated side reactions. This was confirmed by measuring levulinic acid concentrations, which increased
together with the catalyst loading (Figure 3b). It was also noticed that the formation of 1,4-butanediol in
the THF phase was increased with increasing amounts of catalyst. In addition to levulinic acid and 1,4-
butanediol, some other unidentified byproducts were formed when the catalyst loading was increased.
Thus, 3 mg was selected as the optimal amount of catalyst for the conversion reaction, providing a high
yield of HMF with a low amount of byproducts.
Figure 3. (a) Effect of reaction time on HMF yield. (b) Effect of catalyst loading on HMF and levulinic acid
(LA) yield.
3
.2.3 Preliminary recycling experiments
The recyclability of the catalyst is of great importance in the practical production of HMF. In this study, the
catalyst was recycled together with the whole water phase, because the separation of the catalyst would
have significantly reduced the amount of catalyst. The analysis of water phases from previous reactions
revealed that the water phase was suitable for recycling since it did not contain significant amounts of
byproducts: only HMF (yield 5–7%), gamma-butyrolactone and 1,4-butanediol were detected. Also,
unreacted glucose (ca. 34%) was present in the water phase after the conversion reactions, which was
taken into account during the subsequent runs.
Recycling was accomplished with an 8-hour reaction time and a 3-mg dosage of ACBL2 catalyst.
After the first reaction the organic phase was removed and analyzed for the HMF yield and a fresh portion
of glucose and THF was added among the water phase for the next run. The results of the recycling
experiments are shown in Figure 4 and based on them the HMF yield decreased after each consecutive run.
However, the decrease in the yield was not significant until after the third run, when it was over 20%. After
that the yield seemed to balance and decreased only slightly between fourth and fifth run. The decrease in
1
2

the yield during the recycling experiments could have been caused by adsorption and accumulation of
polymeric side products or humins on the acid sites in the porous catalyst [68]. The composition of the THF
layer did not change notably during the first four runs. After fifth run byproducts such as levulinic acid and
2
,3-dihydro-3,5-dihydroxy-6-methyl-4(H)-pyran-4-one [69] were detected with GC-MS. This indicates that
HMF was successfully isolated along with the organic phase between the runs.
Figure 4. The normalized HMF yield after each recycling run in relation to the first run.
4. Conclusions
Lignocellulosic waste material, birch sawdust, was used as a raw material for the preparation of activated
carbon based catalysts. The prepared novel catalysts, ACB, ACB2, ACL, and ACL2, were characterized
extensively and found to contain Lewis or Brønsted acid sites on the catalyst surface. Catalysts were used in
water:THF biphasic system to convert glucose to HMF. The highest HMF yield and selectivity, 51% and 78%,
respectively, were obtained with the mixture of ACB2 and ACL2. Also, ACW produced good HMF yield, 44%,
but the selectivity was only 57%. The prepared heterogeneous catalysts were found to be more effective
than homogeneous sulfuric acid or ZnCl , indicating that activated carbon based catalysts are promising for
2
HMF production, especially in terms of reaction selectivity.
Acknowledgement
The authors acknowledge Davide Bergna for preparing the activated carbon, Seija Liikanen for ICP-OES
measurements and Zouhair El Assal for TPD measurements. Also, the Center of Microscopy and
Nanotechnology at the University of Oulu is acknowledged for their facilities for STEM/EFTEM imaging and
XPS and XRD measurements.
Funding
This work was supported by two projects of the EU/European Regional Development Fund, leveraged from
the EU program: project Biomass value chains (no. A71029) and project PreBio (no. A70594). The financial
support of EU/Interreg Botnia-Atlantica for project Green Bioraff Solutions (no. 20201508) and Fortum
foundation (201700072) is also greatly acknowledged.
Declarations of interest: none.
References
[
[
[
[
[
1] S.P. Teong, G. Yi, Y. Zhang, Green Chem. 16 (2014) 2015-2026.
2] I.K.M. Yu, D.C.W. Tsang, Bioresour. Technol. 238 (2017) 716-732.
3] T. Wang, M.W. Nolte, B.H. Shanks, Green Chem. 16 (2014) 548-572.
4] L.T. Mika, E. Cséfalvay, Á Németh, Chem. Rev. 118 (2018) 505-613.
5] C. Moreau, R. Durand, S. Razigade, J. Duhamet, P. Faugeras, P. Rivalier, R. Pierre, G. Avignon, Appl. Catal.
A Gen. 145 (1996) 211-224.
1
3

[
[
[
[
[
[
[
[
[
6] M.J. Antal, W.S. Mok, G.N. Richards, Carbohydr. Res. 199 (1990) 91-109.
7] Y. Qu, C. Huang, J. Zhang, B. Chen, Bioresour. Technol. 106 (2012) 170-172.
8] F.K. Kazi, A.D. Patel, J.C. Serrano-Ruiz, J.A. Dumesic, R.P. Anex, Chem. Eng. J. 169 (2011) 329-338.
9] Y. Román-Leshkov, J.N. Chheda, J.A. Dumesic, Science 312 (2006) 1933-1937.
10] J. Zhang, E. Weitz, ACS Catal. 2 (2012) 1211-1218.
11] J. Lewkowski, ARKIVOC 2001 (2005) 17-54.
12] B. Saha, M.M. Abu-Omar, Green Chem. 16 (2014) 24-38.
13] D. Chen, F. Liang, D. Feng, M. Xian, H. Zhang, H. Liu, F. Du, Chem. Eng. J. 300 (2016) 177-184.
14] R. van Putten, van der Waal, Jan C, E. de Jong, C.B. Rasrendra, H.J. Heeres, J.G. de Vries, Chem. Rev.
13 (2013) 1499-1597.
1
[
[
[
[
[
[
[
[
15] C.J. Moye, Rev. Pure Appl. Chem. 14 (1964) 161-170.
16] M.S. Feather, J.F. Harris, Adv. Carbohydr. Chem. 28 (1973) 161-224.
17] P. Bhaumik, P.L. Dhepe, Catal. Rev. 58 (2016) 36-112.
18] B.Y. Yang, R. Montgomery, Carbohydr. Res. 280 (1996) 27-45.
19] Y. Yang, C. Hu, M.M. Abu-Omar, Green Chem. 14 (2012) 509-513.
20] Y.J. Pagán-Torres, T. Wang, J.M.R. Gallo, B.H. Shanks, J.A. Dumesic, ACS Catal. 2 (2012) 930-934.
21] P.Y. Dapsens, C. Mondelli, J. Pérez-Ramírez, ACS Catal. 2 (2012) 1487-1499.
22] L. Atanda, A. Silahua, S. Mukundan, A. Shrotri, G. Torres-Torres, J. Beltramini, RSC Adv. 5 (2015) 80346-
8
0352.
[
[
[
[
[
[
23] X. Wang, H. Zhang, J. Ma, Z. Ma, RSC Adv. 6 (2016) 43152-43158.
24] J.M.R. Gallo, D.M. Alonso, M.A. Mellmer, J.A. Dumesic, Green Chem. 15 (2012) 85-90.
25] G. Yang, C. Wang, G. Lyu, L.A. Lucia, J. Chen, BioResources 10 (2015) 5863-5875.
26] F. Rodríguez-Reinoso, Carbon 36 (1998) 159-175.
27] J.L. Figueiredo, M.F.R. Pereira, Catal Today 150 (2010) 2-7.
28] T.J. Bandosz, Surface Chemistry of Carbon Materials, in: P. Serp, J.L. Figueiredo (Eds.), Carbon Materials
for Catalysis, John Wiley & Sons, 2008, pp. 45-92.
[
29] E. Lam, J.H.T. Luong, ACS Catal. 4 (2014) 3393-3410.
1
4

[
[
[
[
30] J.L. Figueiredo, M.F.R. Pereira, M.M.A. Freitas, J.J.M. Órfão, Carbon 37 (1999) 1379-1389.
31] C. Moreno-Castilla, M.V. López-Ramón, F. Carrasco-Marın, Carbon 38 (2000) 1995-2001.
́
32] Y. Yang, K. Chiang, N. Burke, Catal. Today 178 (2011) 197-205.
33] Anon., Calgon Carbon Corporation Investor Presentation, 2011, http:// phx.corporate-
ir.net/External.File?item=UGFyZW50SUQ9ODc0MjV8Q2hpbGRJRD0tMXxUeXBlPTM=& t=1, accessed 19
September 2018.
[
34] M. Danish, T. Ahmad, Renew. Sust. Energ. Rev. 87 (2018) 1-21.
[
2
35] L. Prati, D. Bergna, A. Villa, P. Spontoni, C.L. Bianchi, T. Hu, H. Romar, U. Lassi, Catal. Today 301 (2018)
39-243.
[
36] P. González-García, Renew. Sust. Energ. Rev. 82 (2018) 1393-1414.
[
3
37] T. Deng, J. Li, Q. Yang, Y. Yang, G. Lv, Y. Yao, L. Qin, X. Zhao, X. Cui, X. Hou, RSC Adv. 6 (2016) 30160-
0165.
[
38] X. Xiong, I.K.M. Yu, S.S. Chen, D.C.W. Tsang, L. Cao, H. Song, E.E. Kwon, Y.S. Ok, S. Zhang, C.S. Poon,
Catal. Today 314 (2018) 52-61.
[
[
[
[
39] B. Zou, X. Chen, C. Zhou, X. Yu, H. Ma, J. Zhao, X. Bao, Can. J. Chem. Eng. 96 (2018) 1337-1344.
40] N.I. Villanueva, T.G. Marzialetti, Catal. Today 302 (2018) 100-107.
41] U. Tyagi, N. Anand, D. Kumar, Bioresour. Technol. 267 (2018) 326-332.
42] J.H. Clark, V. Budarin, T. Dugmore, R. Luque, D.J. Macquarrie, V. Strelko, Catal. Commun. 9 (2008)
709-1714.
1
[
(
43] E. Mäkelä, R. Lahti, S. Jaatinen, H. Romar, T. Hu, R.L. Puurunen, U. Lassi, R. Karinen, ChemCatChem 10
2018) 3269-3283.
[
1
44] R. Lahti, D. Bergna, H. Romar, T. Hu, A. Comazzi, C. Pirola, C.L. Bianchi, U. Lassi, Top. Catal. 60 (2017)
415-1428.
[
[
[
[
45] S. Brunauer, P.H. Emmett, E. Teller, J. Am. Chem. Soc. 60 (1938) 309-319.
46] N.A. Seaton, Walton, J. P. R. B., N. Quirke, Carbon 27 (1989) 853-861.
47] W.T. Tsai, C.Y. Chang, S.L. Lee, S.Y. Wang, J. Therm. Anal. Cal. 63 (2001) 351–357.
48] S.A. Schmidt, N. Kumar, A. Shchukarev, K. Eränen, J-P. Mikkola, D.Y. Murzin, T. Salmi, Appl. Catal., A
68 (2013) 120-134.
4
[
49] P. Rechnia-Goracy, A. Malaika, M. Kozlowski, Diamond Relat. Mat. 87 (2018) 124-133.
[
(
50] P.P. Upare, J. Yoon, M.Y. Kim, H. Kang, D.W. Hwang, Y.K. Hwang, H.H. Kung, J. Chang, Green Chem. 15
2013) 2935-2943.
1
5

[
[
[
[
[
[
[
[
51] P. Burg, P. Fydrych, D. Cagniant, G. Nanse, J. Bimer, A. Jankowska, Carbon 40 (2002) 1521-1531.
52] Q. Shu, J. Gao, Z. Nawaz, Y. Liao, D. Wang, J. Wang, Appl. Energy 87 (2010) 2589-2596.
53] C.H. Kim, B.H. Kim, J. Power Sources. 274 (2015) 512-520.
54] M. Farabi, M. Ibrahim, U. Rashid, Y. Taufiq-Yap, Energy Convers. Manag. 181 (2019) 562-570.
55] F. Lónyi, J. Valyon, Micropor. Mesopor. Mater. 47 (2001) 293-301.
56] J. Wang, W. Xu, J. Ren, X. Liu, G. Lua, Y. Wang, Green Chem. 13 (2011) 2678-2681.
57] S. Kang, J. Ye, Y. Zhanga, J. Chang, RSC Adv. 3 (2013) 7360-7366.
58] H.T. Gomes, S.M. Miranda, M.J. Sampaio, J.L. Figueiredo, A.M.T. Silva, J.L. Faria, Appl. Catal. B: Environ.,
06 (2011) 390-397.
1
[
59] X. Xiong, I.K.M. Yu, S.S. Chen, D.C.W. Tsang, L. Cao, H. Song, E.E. Kwon, Y.S. Ok, S. Zhang, C.S. Poon,
Catal. Today, 314 (2018) 52-61.
[
60] X. Li, Y. Zhang, Q. Xia, X. Liu, K. Peng, S. Yang, Y. Wang, Ind. Eng. Chem. Res. 57 (2018) 3545-3553.
61] W. Daengprasert, P. Boonnoun, N. Laosiripojana, M. Goto, A. Shotipruk, Ind. Eng. Chem. Res. 50 (2011)
[
7
903-7910.
[
[
[
62] Q. Pang, L. Wang, H. Yang, L. Jia, X. Pan, C. Qiu, RSC Adv. 4 (2014) 41212-41218.
63] L. Yang, X. Yan, S. Xu, H. Chen, H. Xia, S. Zuo, RSC Adv. 5 (2015) 19900-19906.
64] V. Maruani, S. Narayanin-Richenapin, E. Framery, B. Andrioletti, ACS Sustainable Chem. Eng. 6 (2018)
3487-13493.
1
[
[
[
[
[
65] I.K.M Yu, D.C.W Tsang, Bioresour. Technol. 238 (2017) 716-732.
66] Q.X. Lin, C.H. Zhang, X.H. Wang, B.G. Cheng, N. Mai, J.L. Ren, Catal. Today (2018) Article in press.
67] S.S. Chen, T. Maneerung, D.C.W. Tsang, Y.S. Ok, C.H. Wang, Chem. Eng. J. 328 (2017) 246-273.
68] A. Deng, Q. Lin, Y. Yan, H. Li, J. Ren, C. Liu, R. Sun, Bioresour. Technol. 216 (2016) 754-760.
69] M. Kim, W. Baltes, J. Agric. Food Chem. 44 (1996) 282-289.
Figures
1
6

Figure 1. General reaction scheme of HMF formation from glucose. First glucose is isomerized into fructose
over Lewis acid catalyst and then fructose is dehydrated into HMF over Brønsted acid catalyst.
Figure 2. Flow chart for catalyst preparation: a) carbonization and activation step b) acid modification with
9
M H
2
SO
4
c) acid modification with 18 M H
2
SO
4
d) washing with H
2
O e) impregnation with ZnCl (5 wt.% Zn)
2
f) impregnation with ZnCl
2
(15 wt.% Zn) g) mixing in 1:1 ratio.
Figure 3. a) Effect of reaction time on HMF yield. b) Effect of catalyst loading on HMF and levulinic acid (LA)
yield.
1
7

Figure 4. The normalized HMF yield after each recycling run in relation to the first run.
1
8