Assessment of ion exchange resins as catalysts for the direct transformation of fructose into butyl levulinate

Article abstract of DOI:10.1016/j.apcata.2021.117988

The transformation of fructose into butyl levulinate in aqueous 1-butanol (initial molar ratio 1-butanol/fructose 79, and butanol/water 1.19) has been studied in a discontinuous reactor at 80?120 °C and 2.0 MPa over 8 sulfonic polystyrene-DVB ion exchange resins as catalysts (catalyst loading 0.85–3.4 %). Resins swell greatly in the reaction medium and the reaction takes place mainly in the swollen gel-phase. Swollen resins in water have been characterized by analysis of ISEC data, and spaces originated in the gel phase upon swelling are described in terms of zones of different polymer density. A relationship has been found between the morphology of swollen resins and ester production. Swollen resins with low polymer density show the highest butyl levulinate yield. Dowex 50Wx2 was the most effective because it creates the largest and widest spaces in the gel-phase when swelling. Consequently, it better accommodates the proton-transfer-reaction mechanisms.

Full text of DOI:10.1016/j.apcata.2021.117988

Applied Catalysis A, General 612 (2021) 117988
Contents lists available at ScienceDirect
Applied Catalysis A, General
journal homepage: www.elsevier.com/locate/apcata
Assessment of ion exchange resins as catalysts for the direct transformation
of fructose into butyl levulinate
Eliana Ramírez, Roger Bringu ´e , Carles Fit ´e , Montserrat Iborra, Javier Tejero*, Fidel Cunill
Department of Chemical Engineering and Analytical Chemistry, Faculty of Chemistry, University of Barcelona, Martí i Franqu `e s 1, 08028, Barcelona, Spain
A R T I C L E I N F O
A B S T R A C T
Keywords:
The transformation of fructose into butyl levulinate in aqueous 1-butanol (initial molar ratio 1-butanol/fructose
Butyl levulinate
Fructose
9, and butanol/water 1.19) has been studied in a discontinuous reactor at 80ꢀ 120 ◦C and 2.0 MPa over 8
7
sulfonic polystyrene-DVB ion exchange resins as catalysts (catalyst loading 0.85–3.4 %). Resins swell greatly in
the reaction medium and the reaction takes place mainly in the swollen gel-phase. Swollen resins in water have
been characterized by analysis of ISEC data, and spaces originated in the gel phase upon swelling are described in
terms of zones of different polymer density. A relationship has been found between the morphology of swollen
resins and ester production. Swollen resins with low polymer density show the highest butyl levulinate yield.
Dowex 50Wx2 was the most effective because it creates the largest and widest spaces in the gel-phase when
swelling. Consequently, it better accommodates the proton-transfer-reaction mechanisms.
5
-Hydroxymethylfurfural
Ion-exchange resin catalysts
fermentation alcohols (ethanol, butanol) are 100 % biofuels and,
therefore, their use is very attractive. Ethyl and butyl levulinate are the
most studied ALs as fuel enhancers. They have good blending octane
numbers (107 and 97, respectively [11,12]), but because of their
1
. Introduction
The transformation of lignocellulose-derived sugars gives rise to
physical properties, they are more suitable for blending with commer-
cial diesel, mostly as cold-flow enhancers. Compared to ethyl levulinate,
butyl levulinate (BL) is more promising as a diesel blend component. It
has a blending cetane number of about 46, close to that of commercial
diesel, and it is much less soluble in water, mitigating the risk of ester
separation from diesel fuel at low temperatures as an aqueous liquid
phase [13,14].
valuable compounds for the refining and chemical industries called
platform chemicals, such as levulinic acid (LA). The US Department of
Energy has highlighted levulinic acid as a very promising platform
chemical because of the number of industrially interesting LA-derived
compounds [1,2]. LA production costs are expected to drop to 1 $/kg
after new upgraded biomass conversion technologies are successfully
commercialized [3,4].
Apart from a few exceptions, raw biomass or biomass-derived re-
actants are the starting materials to synthesize alkyl levulinates. In
general, syntheses involve the treatment of reactants in alcohol and acid
catalysis [15]. Synthesis of BL was undertaken for the first time in the
The most notable LA-derived compounds are alkyl levulinates (ALs).
They have commercial use in mineral oil refining [5], flavoring prepa-
rations [6], latex coating compositions [7], as well as additives to
polymers and perfumes [8]. Nowadays, besides its potential use as green
solvents [9], levulinates are interesting as biofuels for blending with
commercial gasoline or diesel to fit the quality standards settled by the
Directive 2009/30/EC, a demanding European regulation on pollutant
levels in exhaust combustion gases and fuels composition [10]. This
Directive sets a minimum oxygen content from biomass origin of 15 %
by December 31, 2020, to mitigate the greenhouse effect of vehicle
exhaust emissions. Faced with the challenge of reformulating commer-
cial fuels to meet such demanding regulations, the refining industry has
realized the ALs potential for blending to commercial fuels because of
their physicochemical properties and low toxicity. The levulinates of
3
0s by the reaction between levulinic acid and 1-butanol (BuOH) in
excess using HCl as the catalyst, although reported yields were low
40–65 %) [16–18]. Much later, in the 90s, Bart et al. proposed a kinetic
model for the esterification of LA with BuOH in the presence of H SO
19].
Acidic homogeneous catalysts present corrosion and separation
(
2
4
[
problems and the industry tends to replace them with solid ones. Het-
erogeneously catalyzed esterification of LA with BuOH has been
attempted recently, using often solid Br o¨ nsted acids: zeolites [20,21],
Zr-containing MOFs [22], HPA supported on acid-treated clay
*
Corresponding author.
E-mail address: jtejero@ub.edu (J. Tejero).
https://doi.org/10.1016/j.apcata.2021.117988
Received 5 August 2020; Received in revised form 30 December 2020; Accepted 2 January 2021
Available online 7 January 2021
0
926-860X/© 2021 Elsevier B.V. All rights reserved.

E. Ramírez et al.
Applied Catalysis A, General 612 (2021) 117988
Nomenclature
liquid and the resin surface, m/h
Le
Effective diffusion-path-length parameter, m
levulinic acid
Als
am
Alkyl levulinates
LA
2
External surface area per unit mass of catalyst (m /g)
m
F
fructose mass, g
BF
Butyl formate
PS-DVB Polystyrene-divinylbenzene
BL
Butyl levulinate
5-Butoxymethylfurfural
1-Butanol
R
BuOH/W Initial mole ratio alcohol/water
BMF
BuOH
C
R
BuOH/F Initial mole ratio 1-butanol/fructose
Initial reaction rate of fructose consumption, mol/h⋅g
0
F
r
3
2
Polymer chain density in the swollen state, nm/nm .
Sg
BET Surface area, m /g dry catalyst
2
CF,b
CF,S
Fructose concentration in the bulk liquid, mol/L
Spore
t
Surface area of resin in the swollen state, m /g dry catalyst
Fructose concentration in the external catalyst surface,
time, h
3
mol/L
V
V
V
g
pore volume by N
dry catalyst
2
adsorption-desorption at 77K, cm /g
DBE
Di-n-butyl ether
2
3
De,F
Fructose effective diffusivity in the resin particle, m /h
pore
sp
Pore volume of resin in the swollen state, cm /g dry
catalyst
d
d
C
diameter of polymer chains, nm
3
m
molecule diameter assumed spherical, nm
Specific volume of the swollen polymer, cm /g dry catalyst
dP
Mean particle diameter of swelling particle, m
mean pore diameter, nm
W
mass of dry catalyst, g
dpore
F
X
Y
θ
F
J
F
Fructose conversion, dimensionless
Fructose
Yield of fructose towards the compound j, dimensionless
Porosity of the swollen catalyst, dimensionless
FA
Formic acid
3
HMF
5-Hidroxymethylfurfural
ρ
P
Apparent density of the catalyst, kg/m .
3
K
0
Ogston distribution coefficient
Mass-transfer coefficient of fructose between the bulk
ρ
S
Skeletal density of the catalyst, kg/m .
km,F
Φ
Weisz-Prater Modulus, dimensionless
montmorillonite K10 [23] and silicalite-1 [24], sulfated ZrO
mesoporous silica [25], ZrO -based hybrid catalysts functionalized by
both organosilica and HPA [26], HPA and ZrO bi-functionalized orga-
nosilica nanotubes [27], WOx/mesoporous-ZrO [28], ammonium and
2
over
macroreticular resin specially designed to work in low polar organic
media. However, it is often not the most suitable to work in polar en-
vironments, for example in reactions where water is produced. This resin
is too stiff because of its high DVB content (20 %) and swells only
moderately in polar media. In alcoholic media containing water, gel-
type resins are usually more active and selective catalysts than macro-
reticular ones with high DVB content [31,42–44]. Therefore, this paper
aims to study the liquid-phase synthesis of butyl levulinate from
1-butanol and fructose over both gel-type and macroreticular sulfonic
PS-DVB resins. Since the catalytic behavior of the resins is highly
dependent on their morphology, tested resins have been characterized in
the swollen state, and the influence of resins morphology on their cat-
alytic activity is discussed.
2
2
2
silver co-doped phosphotungstic acid [29], Al-containing MCM-41 [30],
and ion exchange resins [31,32]. In general, high ester yields are ob-
tained using 1-butanol in excess (BuOH/LA molar ratios between 5 and
1
0), and moderate temperatures (60ꢀ 90 ◦C). Finally, successful at-
tempts at BL production via immobilized enzymes are also reported
[
33].
As levulinic acid is quite an expensive bio-based feedstock today,
some attempts have been made to produce BL directly from cellulose
using both homogeneous and heterogeneous acid catalysts. Cellulose is
2 4
hardly soluble in 1-butanol, and H SO is more effective to catalyze the
alcoholysis of cellulose [34,35] than sulfonated hyperbranched poly-
2. Experimental
mers [36] or metal sulfates [37,38]. However, temperatures between
◦
1
60 and 200 C and a very large excess of BuOH are necessary to obtain
2.1. Chemicals
moderate ester yields (30–62 %). The main drawback is the formation of
a noticeably amount of soluble and insoluble polymers, called humins,
because of cellulose degradation, hindering the selectivity to BL as well
as the production of di-n-butyl ether (DBE) by BuOH dehydration.
1-Butanol and D-Fructose (≥99.5 %, Across Organics) were used
without further purification. Butyl levulinate (≥98 %, Sigma Aldrich),
formic acid (≥98 %, Labkem), 5-hydroxymethylfurfural, levulinic acid,
butyl formate and di-n-butyl ether (≥98 %, Across Organics), as well as
distilled water, were used for analysis.
To the best of our knowledge, attempts to produce BL from C
bohydrates are scarce. Among the most studied monosaccharides
glucose and fructose (F)), alkyl levulinate yields are significantly higher
6
car-
(
when starting from fructose [15]. An et al. studied fructose butanolysis
2.2. Catalysts
◦
over metal sulfates at 190 C obtaining a BL yield of 62.8 % after 2 h
◦
[
37]. Kuo et al. used TiO
2
nanoparticles at 150 C and reported a BL
Eight sulfonic PS-DVB ion-exchange resins were used as catalysts.
Selected resins have similar acid capacity but different morphology,
including both macroreticular and gel-type ones. These resins are
conventionally sulfonated (monosulfonated, having an average con-
centration of about a sulfonic group per styrene ring). Rohm and Haas
(Amberlyst 15, Amberlyst 16, Amberlyst 39 and Amberlyst 31), Aldrich
(Dowex 50Wx2, Dowex 50Wx4, and Dowex 50Wx8), and Purolite
(CT124) supplied tested catalysts. Table 1 shows the properties of these
resins.
yield of 75 % after 3 h [39]. Over sulfonated materials, Balakrishnan
◦
et al. reported a BL yield of 16 % on Amberlyst 15 at 110 C after 30 h
[
40]. Interestingly Liu et al. obtained BL yields of 87 and 89 % on
CNT-PSSA (poly (p-styrene sulfonic acid)-grafted carbon nanotubes) and
◦
Amberlyst 15, respectively, at 120 C after 24 h [41]. In the mentioned
works fructose conversion is high (97–100 %), and the differences in
yield are probably due to the presence of side reactions, as humins
formation. The work of Liu et al. [41] shows that acidic ion-exchange
resins are interesting catalysts. They are selective and at a moderate
◦
temperature (120 C) give better yields than TiO
2
nanoparticles at 150
2.3. Apparatus and analysis
◦
◦
C [39], and metal sulfates at 190 C [37].
Amberlyst 15 is an acidic polystyrene-divinylbenzene (PS-DVB)
Experiments were carried out in a 100-mL stainless steel reactor
2

E. Ramírez et al.
Applied Catalysis A, General 612 (2021) 117988
◦
Table 1
described elsewhere [46]. Solubility at 25 C is very low (2.60 g/L).
◦
Properties of the resin catalysts used.
Solubility increases highly with temperature and it is 37.7 g/L at 80 C.
◦
a
At 100 C solubility is about 270 g/L, but liquid becomes pale yellow
Catalyst
Acronym
Morphology
DVB
Acid
Native
particle
size
T
max
b
◦
e
%
capacity
( C)
probably due to the thermal decomposition of fructose.
(
mmol
◦
Experiments were performed at 80ꢀ 120 C. In a typical experiment,
+
c
H /g)
range
a mixture of 1-butanol (60 mL) and water (10 mL) containing 1.5 g of
fructose was charged into the reactor (initial molar ratio 1-butanol/fruc-
tose, RBuOH/F = 79; fructose concentration 0.142 mol/kg). Water and 1-
butanol are partly miscible. Solubility of water in the butanol-rich re-
d
(
mm)
Dowex
DOW2
DOW4
A31
G
G
G
G
G
M
M
M
2
4.98
4.95
4.84
5.00
4.83
4.82
4.80
4.81
0.074 –
0.149
150
150
130
130
150
130
130
120
5
0Wx2
Dowex
4
0.074 –
0.149
◦
◦
gion is 20.4 wt.% at 25 C [47], 30.6 wt.% at 90 C [47], and 50 wt.% at
5
0Wx4
◦
Amberlyst
4
0.309 –
1.143
120 C [48]. As the liquid mixture contained 17.2 wt.% of water
3
1
(
1-butanol/water molar ratio, RBuOH/W = 1.19), it was not separated into
two phases. Furthermore, despite fructose reduces water solubility in
-butanol, a 0.142 mol/kg solution of fructose in water-saturated
CT124
CT124
DOW8
A39
4
0.425 –
1
.200
1
Dowex
8
0.074 –
0.149
1-butanol does not form clouds in the range 25ꢀ 80 ◦C [49]. As the
5
0Wx8
Amberlyst
8
0.410 –
1.141
water solubility increases with temperature it is assured that the solution
3
9
◦
is homogeneous up to 120 C.
Amberlyst
A16
12
20
0.410 –
0.948
The reactor was stirred at 500 rpm, set at 2 MPa to maintain the
liquid phase over the reaction, and heated. When the liquid reached the
desired temperature, dried catalyst (0.5ꢀ 2 g; catalyst loading 0.85–3.4
%) was injected into the reactor. Catalyst injection was taken as time
zero. 8 h-experiments were carried out.
1
6
Amberlyst
A15
0.310 –
1.041
1
5
a
b
c
Macroreticular (M) and gel-type (G).
From Bringu ´e et al. [42].
F
In each experiment, fructose conversion (X ), and yield of fructose
Titration against standard base following the procedure described by Fischer
j
F
and Kunin [45].
towards the product j (Y ) were estimated by Eqs. 1 and 2, respectively
d
Particle size distribution of the commercial sample (1–99 % volume
[
]
mole of F reacted
mole of F initially
fraction).
e
X
Y
F
=
x 100 %, mol/mol
(1)
(2)
Maximum operating temperature. Information supplied by the
manufacturer.
[
]
mole of F reacted to form j
mole of F initially
j
F
=
x100 %, mol/mol
(
Magnedrive II, Autoclave Engineers). The reactor was heated by an
◦
electric oven TC 22 Pro 9, and its temperature was controlled to ±0.1 C
by a TIC controller TOHO TIM-125. The reaction medium was stirred by
a magnetic drive turbine equipped with a 4-blade axial-up mixer. The
dried catalyst was injected into the reactor from an external cylinder by
Replicated runs were made for some experiments and the repro-
ducibility of results was found to be reliable. Fructose conversions were
accurate within ± 1%. On the other hand, BL yields were accurate within
±
5% and those of HMF, BMF, and LA within ± 5–8 %
2
shifting with N .
Liquid samples were taken hourly out of the reactor and analyzed.
Butyl levulinate, butyl formate (BF), 5-butoxymethylfurfural (BMF), di-
n-butyl ether, and water were determined by Gas-Liquid Chromatog-
raphy. An Agilent HP6890 GLC apparatus equipped with a TCD detector
and an HP Pona methyl silicone capillary column HP190195-001 (50 m
2
.5. Reused catalyst
The recovered resin was washed in a 25-mL column by percolation
with deionized water to remove solid materials present in the used
catalyst. To assure that the resins were completely swollen, the column
was filled with water, and 2-BV (bed volume) of water was poured into
the column at 4 BV/h. This procedure was repeated until the poured
water was colorless. After that, the resin was downwash with methanol
three times to remove all the water. Finally, the catalyst was dried as
explained before. Its final water content was less than 3 wt.% (Karl
Fischer titration).
x 0.2 mm x0.5
μ
m) analyzed 0.2
μ
L liquid aliquots. A temperature ramp
◦
◦
◦
of 10 C/min from 50 C up to 250 C was initially programmed and then
held for 6 min. He (≥ 99.998 %, Linde) was the carrier gas at a column
flow rate of 1 mL/min. Fructose, formic acid (FA), levulinic acid, and 5-
hydroxymethylfurfural (HMF) were determined by High-Performance
Liquid Chromatography. An Agilent 1200 Infinity II HPLC equipped
with a Refractive Index (RI) detector and Agilent Hi-Plex H column (300
mm x 7.7 mm) analyzed 100
μ
L liquid aliquots. The mobile phase was a
3
. Results and discussion
0
.005 N aqueous solution of H
2
SO at a flow rate of 0.6 mL/min and the
4
◦
column temperature 50 C. Reaction products were identified by an
Agilent GC/MS (6890A series GLC with an Agilent GC/MS 5973 detec-
tor, and chemical database software).
3
.1. Blank experiments
Blank experiments with a 1-butanol/water mixture (RBuOH/W = 1.19)
◦
containing 1.5 g of fructose were carried out at 120 C without a cata-
2
.4. Methodology
lyst. Fructose undergoes thermal decomposition, with X
F
≈ 20 % after 8
h. HMF was the only product observed with a yield of 8.8 %. The dif-
ference between fructose conversion and HMF yield suggests the for-
mation of a small number of byproducts by thermal decomposition of
fructose, although below the detection limits of HPLC and GLC analyses.
This fact is in line with blank experiments with fructose (9 wt. %) in 0.5
mL of water/GVL mixture (50:50 v/v) performed by Thapa et al., who
Dowex 50Wx2, Dowex 50Wx4, and Dowex 50Wx8 resins were used
as 100–200 mesh beads (0.074ꢀ 0.149 mm) as shipped. Commercial
beads of Amberlyst 15, Amberlyst 16, Amberlyst 39, Amberlyst 31, and
CT124 were crushed and sieved, and 0.08ꢀ 0.1 mm particles were
◦
selected. Catalysts were dried at 110 C for 2 h at atmospheric pressure
HMF
◦
and then at 10 mbar overnight. The residual water of the resins, deter-
mined with a Volumetric Karl Fischer Titrator Orion AF8 (Thermo
Electron Corporation), was less than 3 wt.%.
quoted X
F
≈ 20 % and Y
F
= 7% after 24 h at 120 C [50]. In contrast,
1
-butanol does not undergo reaction under the specified reaction con-
ditions since neither DBE nor butenes were detected.
Since the solubility of fructose in alcohol is limited [15], the solu-
◦
bility of fructose in 1-butanol has been determined at 25, 80, and 100 C
by
a gravimetric method following the experimental procedure
3

E. Ramírez et al.
Applied Catalysis A, General 612 (2021) 117988
3
.2. Analysis of a model experiment with Dowex 50Wx2
noted that compounds such as pyruvaldehyde, furfuryl alcohol, or
furfural, are unstable and react easily forming polymers or humins. The
pathways to produce humins from HMF are not so clear. It has been
hypothesized that soluble humins are formed by con-
densation/etherification of HMF, and insoluble humins by HMF addition
(polymerization) [56]. Solid humins are retained in the catalyst, and
soluble ones remain in the solution.
It is found in the open literature that in alcoholic media containing
water, gel-type resins with low DVB content are usually effective cata-
lysts [31,42–44], thus Dowex 50Wx2 resin (gel-type, with 2% DVB) was
selected. Fig. 1 shows the variation in the mole number of fructose and
◦
reaction products over time in an experiment performed at 120 C. It is
seen that fructose decomposes readily and its conversion is almost
complete after 4 h. HMF forms swiftly from the beginning of the reaction
reaching a maximum at about 1 h. Then the HMF amount decreases
continuously and the HMF yield is only about 3.8 % after 8 h. The yield
A large amount of BMF remains in the liquid phase after 8 h of re-
BMF
action (Y
F
= 29.3 %). Since 1-butanol is in large excess, BL yield is
expected to increase substantially over time as BMF alcoholysis and LA
esterification occur.
of BMF, LA, BL, and BF increases steadily over time, although that of
The molar balance of 1-butanol was fulfilled within ± 0.5 %. No DBE
was detected despite the large excess of BuOH in the reaction medium,
in agreement with the study of 1-butanol dehydration to di-n-butyl ether
on ion-exchange resins by P ´e rez et al. [44], who showed the low cata-
lytic activity of acidic resins in the dehydration of 1-butanol to DBE at
BMF
BMF seems to reach a smooth maximum at about 7 h (Y
The yield of butyl levulinate at 8 h is 32.3 %.
F
= 32.6 %).
The process for producing butyl levulinate from fructose takes place
through a complex series-parallel reaction scheme (Fig. 2) where [15,
◦
5
1]:
120 C in the presence of water.
1
2
) Fructose dehydrates firstly to HMF
3.3. Effect of catalyst mass of Dowex 50Wx2 and temperature
) HMF reacts with 1-butanol to give BMF. The subsequent alcoholysis
of BMF yields BL and BF
The effect of the catalyst mass of Dowex 50Wx2 was checked at 120
◦
3
) HMF rehydrates to LA giving place to FA as a coproduct. Subse-
quently, both acids esterify to BL and BF, respectively.
C by changing the mass of dried resin between 0.5 and 2 g. Fig. 3 shows
the curves of fructose conversion and yield of HMF, LA, BMF, and BL as a
function of contact time (t⋅W/n◦
F F
). As can be seen, X and yield curves
According to the stoichiometry of the involved reactions, the sum of
obtained with different catalyst mass practically overlap, which con-
firms the reliability of data and that the reactor was not saturated with
the solid catalyst. As the initial composition of the reactor is the same,
and the reaction scheme is a series-parallel one, the contact time in-
creases on increasing the catalyst mass, and consequently, the system
moves to the production of butyl levulinate. Fig. 3E shows that BL yield
FA and BF moles (reaction coproducts) should be the same as that of LA
and BL, within the limits of the experimental error. However, as seen in
Fig. 1, the sum of moles of FA and BF is higher by 1.49 times at the end of
the experiment. The fact that solvolysis of carbohydrates gives place to a
molar excess of FA (plus BF) over LA (plus BL) is often reported in the
presence of acid catalysts [50,52–55] although it is not well understood.
At least four potential pathways, depending on reaction conditions, have
been quoted as responsible for the formation of the FA excess in the
hydrolysis of biomass-derived hexoses: through D-erythrose, furfuryl
alcohol, furfural, or pyruvaldehyde formation [52]. FA is formed as well
as D-erythrose, furfuryl alcohol, furfural, or pyruvaldehyde in all four
pathways. However, we do not have detected any of those substances,
probably because they were below the detection threshold of the GLC
and HPLC apparatus due to the small initial amount of fructose.
Furthermore, they are unstable and could probably react quickly giving
place to polymers contributing this way to the formation of humins.
At the end of the experiment, the liquid is brown. The molar balance
of fructose shows that about 80 % of carbohydrate has been converted
into HMF, BMF, LA, and BL, which implies that some fructose de-
composes forming humins. Correspondingly, the balance of fructose
carbon atoms shows that about 17 % are lost. Humins formation is
currently attributed to degradation reactions of fructose [52], HMF [56],
LA [50], and cross-reaction between fructose and HMF [57]. It is to be
rises to 43.5 % when contact time is about 1900 g⋅h/mol. At such a long
contact time, HMF is almost consumed (Y
HMF
F
≈ 1.5 %, Fig. 3B). As seen
in Fig. 3D, BMF yield over contact time has a maximum according to its
role as an intermediate compound in the reaction scheme. However, the
amount of BMF at 1900 g.h/mol is still considerable and BL production
is expected to increase even more as BMF is consumed. The production
of LA (Fig. 3C), FA, and BF (not shown), like that of BL, rise moderately.
Finally, it is to be noted that DBE was detected only in the experiment
with 2 g of catalyst at very long contact time.
Fig. 4 shows fructose conversion and yield of HMF, LA, BMF, and BL
◦
over time as a function of temperature in the range 80ꢀ 120 C. As ex-
◦
pected, X increases with temperature (Fig. 4A). At 100 C it is higher
F
◦
than 99.5 % at 8 h reaction, and at 110 C at 4 h. Accordingly, product
formation accelerates with temperature. Regarding HMF formation,
HMF
◦
Y
F
increases continuously at 80 and 90 C whereas yield curves at
◦
100ꢀ 120 C show a maximum (Fig. 4B). The maximum value is smaller,
and it appears at shorter times as temperature increases, since HMF
◦
transformation into products is faster. At 80 C, decomposition of HMF is
very slow, and only LA (Fig. 4C) and FA (not shown) are formed.
◦
Therefore, the hydrolysis of HMF takes place already at 80 C. The
◦
formation of BMF is observed at 90 C after 6 h (Fig. 4D). BMF yield
◦
increases with temperature, and it shows a maximum at 120 C, ac-
cording to its role as an intermediate compound in the series-parallel
reaction scheme (Fig. 2). Finally, BL (Fig. 4E) and BF (not shown) are
◦
observed at 100 C and 3 h. The alcoholysis of BMF is, therefore, active
◦
at 100 C. The production of BL, BF, LA, and FA increases with tem-
perature, and the esters appear at shorter reaction times. By comparing
the formation of LA and BMF it can be inferred that the hydrolysis
mechanism predominates at the lower temperature of the range
◦
◦
explored (80 C), and alcoholysis at the highest one (120 C).
3
.4. Assessment of the resistance to mass transfer in Dowex 50Wx2 resin
The influence of external and intra-particle mass transfer was
Fig. 1. Moles of fructose, HMF, BMF, LA, BL, FA, and BF vs. time on Dowex
0Wx2 (T = 120 ◦C; RBuOH/F = 79; RBuOH/W = 1.19; 1 g dried catalyst; catalyst
◦
5
checked in the experiment performed at 120 C and 1 g of dried Dowex
50Wx2. As a general rule, mass transfer resistance is reduced to negli-
loading 1.7 wt. %).
4

E. Ramírez et al.
Applied Catalysis A, General 612 (2021) 117988
Fig. 2. Reaction scheme [15,51].
◦
Fig. 3. Fructose conversion (A) and yield of HMF (B), LA (C), BMF (D), and BL (E) over contact time as a function of catalyst mass (Dowex 50Wx2; T = 120 C; RBuOH/
F
= 79; RBuOH/W = 1.19).
gible proportions at moderate temperatures if sufficiently small catalyst
particles are used and the reaction rate is not high. Since the reaction
network is complex, resistance to mass transfer was checked at zero time
examining the rate of fructose consumption. Assuming steady state, the
0
F
rate of fructose consumption in the resin bead, ꢀ r , may be written in
5

E. Ramírez et al.
Applied Catalysis A, General 612 (2021) 117988
Fig. 4. Fructose conversion (A) and yield of HMF (B), LA (C), BMF (D), and BL (E) vs. time as a function of temperature (Dowex 50Wx2; RBuOH/F = 79; RBuOH/W
.19; 1 g dried catalyst; catalyst loading 1.7 wt. %).
=
1
terms of the diffusion rate of fructose from the bulk liquid to the surface
as [58]:
where Le is the effective diffusion-path-length parameter (a third of the
radius for spherical beads), ρP = is the apparent density of the particle,
and De,F is the effective diffusivity of fructose in catalyst pores. As seen in
Table 2, Φ = 0.002. Consequently, the reaction rate is also free of intra-
particle mass transfer resistance. The detailed procedure followed to
estimate both the fractional concentration difference and the Weisz-
Prate modulus is described in the SI section.
ꢀ
)
0
ꢀ
r
= km,F
a
m
C
F,b ꢀ CF,s
(3)
F
where km,F is the fructose mass-transfer coefficient between bulk liquid
and the solid external surface, am is the external surface area per unit
mass of the particle, CF,b is the fructose concentration in the bulk liquid,
and CF,s the fructose concentration in the resin external surface,
respectively. As can be seen in Table 2, the fractional concentration
3
.5. Catalyst screening
A series of experiments was performed using 8 acidic resins at 120 C
ꢀ
)/
difference CF,b ꢀ CF,s CF.b is less than 0.2 %. Consequently, the resis-
tance to external mass transfer is negligible.
◦
Since the influence of external mass transfer is negligible, the crite-
rion for observing chemically controlled reaction rates is that the Weisz-
Prater modulus, in terms of measurable quantities, should be [59]:
with a 1-butanol/water mixture (RBuOH/W = 1.19) containing 1.5 g of
fructose. In all the experiments 1 g of dried catalyst was used. Tested
resins were gel-type (Dowex 50Wx2, Dowex 50Wx4, Amberlyst 31,
CT124, and Dowex 50Wx8) and macroreticular (Amberlyst 39,
Amberlyst 16, and Amberlyst 15). All the resins have a very similar acid
capacity. As described in Section 2.4, catalyst particle size distribution
ꢀ
)
0
2
ꢀ
r
ρP
L
F
e
Φ =
<< 1
(4)
D
e,F
C
F,S
6

E. Ramírez et al.
Table 2
Applied Catalysis A, General 612 (2021) 117988
is higher than those of Amberlyst 39 and Amberlyst 15 (Fig. 5D). It is to
be noted that BMF is detected at 2 h of reaction on Amberlyst 15,
whereas it is formed at the beginning of the reaction on the other two
catalysts
Fructose fractional concentration difference between
bulk liquid and catalyst surface and Weisz-Prater
◦
modulus (T = 120 C; RBuOH/F = 79; RBuOH/W = 1.19;
1
g dried Dowex 50Wx2; 500 rpm).
Parameter
Butyl levulinate is formed by the esterification of LA and by the
etherification of BMF. As can be seen in Fig. 5E, whereas Dowex 50Wx2
and Amberlyst 39 have similar BL yields, and even show very close
values at 8 h (about 33 %), BL is detected on Amberlyst 15 at 3 h, and the
BL yield is only of 15 % at 8 h
Value
0
a
5.0⋅10ꢀ 3
ꢀ
r (mol/h g)
F
2
b
am (m /g)
0.0243
c
km,F (m/h)
1.2
ꢀ
)/
Finally, Fig. 5F and G show the yield of FA and BF. As seen in Fig. 5F,
FA yields differ at the beginning of the reaction, but they are very similar
in the three resins from 4 h. In contrast, BF formation shows a similar
picture to the BMF yield. Dowex 50Wx2 and Amberlyst 39 have similar
BF yields at 8 h (about 25 %), and on Amberlyst 15 it is detected at 3 h
with a yield of 15 % at 8 h. It is to be noted that DBE was not detected in
any experiment of this series.
CF,b ꢀ CF,s CF.b
0.0017
3
d
ρ
(kg/m )
374
2.88⋅10ꢀ
2.03⋅10
103
P
5
Le (m)
2
e
ꢀ 9
De,F (m /s)
3
CF,S (mol/m )
Φ
a
0.0020
Obtained from the function of variation of fructose
It is seen that although fructose conversion is fast, the yield of HMF,
BMF, BL, and BF is different on Dowex 50Wx2, Amberlyst 39, and
Amberlyst 15. The formation of BMF, BL, and BF is lower over Amberlyst
conversion against time.
b
For an spherical particle, am = 6/ρPdP, with ρP
=
apparent density of the resins and dP = mean particle
1
5, the resin with the higher DVB content and consequently the stiffest
diameter.
c
one. At the same time, HMF yield shows the opposite trend: the higher
yield at 8 h is found on Amberlyst 15. Although LA and FA yield is of the
same order on the three catalysts, it can be inferred that the actual
morphology of resins in the reaction medium plays an important role in
butyl levulinate production.
Estimated by using the correlation of Sano et al.
[
60].
d
ρP
= ρS(1 ꢀ θ). ρS is the skeletal density [42], and θ
the porosity of the catalyst.
e
Estimated by a modified Wilke-Chang equation
[
61].
3
.6. Influence of the acid site density of swollen resins on catalytic activity
was 0.074 – 0.149 mm (average 0.105 mm) for Dowex resins and
Table 3 shows the yield of butyl levulinate and the other reaction
0
.08ꢀ 0.1 mm (average 0.089 mm) for the other. The stirring rate was
products at 6 h. As seen in Fig. 5, fructose conversion is higher than 99.5
on all the catalysts. The highest BL yield is obtained on Dowex 50Wx2
fixed at 500 rpm. Under such operating conditions and small particles,
the influence of mass transfer was negligible for all the resins as shown
in detail in the SI section.
%
and the lowest on Amberlyst 15. As a whole, BL and BMF yield decrease
in the order: gel-type resins with 2% DVB (Dowex 50Wx2), 4% DVB
All catalysts were active producing BL from fructose but important
differences in their catalytic behavior are seen depending on the resin
structure (gel-type or macroreticular) and the crosslinking degree
(
Dowex 50Wx4, CT124, Amberlyst 31) and 8% DVB (Dowex 50Wx8),
and macroreticular ones with 8% DVB (Amberlyst 39), 12 % DVB
Amberlyst 16) and 20 % DVB (Amberlyst 15). On the contrary, HMF
(
(
percentage by weight of DVB in the mixture of styrene and DVB co-
yield follows the opposite trend, since the highest HMF yield is obtained
on Amberlyst 15, and the lowest on Dowex 50Wx2. BF yield is
approximately the same on all catalysts, within the limits of the exper-
imental error, but Amberlyst 16 and Amberlyst 15, which show lower
values. As for LA and FA, the yield is similar over all the resins; FA yield
being much higher than that of LA. The fact that the formation of the
bulkiest molecules of the reaction system was related to the resin type
and their crosslinking degree suggests that the reactions take place
mainly within the resin gel-phase.
polymers that gives rise to the resin). Fig. 5 shows the conversion of
fructose and the yield of HMF, LA, BMF, BL, FA, and BF over time. For
the sake of clarity, only the results of three catalysts are presented in
Fig. 5: the gel-type resin Dowex 50Wx2 (2%DVB), the low-crosslinked
macroreticular resin Amberlyst 39 (8%DVB), and the high-crosslinked
macroreticular resin Amberlyst 15 (20 %DVB).
As Fig. 5A shows, fructose reacts completely irrespective of the
catalyst used. Although some fructose is decomposed thermally as seen
in section 3.1, the presence of a catalyst accelerates the conversion of the
monosaccharide into HMF and humins. It is seen that fructose conver-
sion on Dowex 50Wx2 and Amberlyst 39 is similar at any time, but it is
initially slower on Amberlyst 15.
Ion exchange resins swell in polar media. Table 4 shows the mean
particle size of tested resins determined by laser measurements in air,
water, and 1-butanol with a Beckman Coulter LS Particle Size Analyzer.
As seen, resins greatly swell both in water and in 1-butanol, although
they swell more in water. In general, swelling decreases in the order: gel-
type resins with 2, 4, and 8% DVB, and macroreticular ones with 8, 12,
and 20 % DVB.
Fig. 5B shows the variation of HMF yield over time. HMF forms fast,
and subsequently, it rehydrates giving place to LA and etherifies to BMF.
Curves of HMF yield show a maximum. Dowex 50Wx2 reaches that
maximum at about 1 h, Amberlyst 39 does it between 1 and 2 h, and
Amberlyst 15 at 2 h. Clear differences are seen at the end of the runs:
HMF yield at 8 h is 3.8 % on Dowex 50Wx2, about 10 % on Amberlyst
Resins undergo a series of morphological changes on swelling, by
which spaces that did not exist in the dry state appear (Fig. 6). Dried gel-
type resins contain only micropores (spaces between the polymer
chains). Mesopores appear when they swell in polar media (water,
alcohol… ), and disappear on shrinking in non-polar media. Macro-
reticular resins can be described as a collection of polymeric gel-phase
microspheres interspersed by permanent pores. In a non-polar me-
dium, these resins have macro and mesopores (between the aggregates
of microspheres) and micropores (inside gel-type microspheres). As a
result of swelling, an additional number of macro and mesopores
appear, spaces that existed in the gel-phase in the range of micropores
slightly increase their size, and at the same time, new spaces are formed
within the gel phase.
3
9, and 36 % over Amberlyst 15, respectively. However, interpretation
of HMF yield curves is not easy, since HMF also forms humins through
several mechanisms, and humins formation from HMF is more notice-
able at the higher HMF concentration [57].
HMF produces butyl levulinate via the hydrolysis and alcoholysis
pathways. Hydrolysis of HMF gives rise firstly to LA, whereas alcoholysis
requires the etherification of HMF to BMF. As seen in Fig. 5C and D, LA
yield is lower than that of BMF. So, it can be inferred that etherification
to BMF is a faster reaction and the alcoholysis pathway predominates
◦
over hydrolysis at 120 C with all catalysts. As seen in Fig. 5C, LA yield is
similar over the three resins. However, the BMF yield on Dowex 50Wx2
The spaces formed on swelling are difficult to characterize because
7

E. Ramírez et al.
Applied Catalysis A, General 612 (2021) 117988
◦
Fig. 5. Fructose conversion (A) and yield of HMF (B), LA (C), BMF (D), BL (E), FA (F), and BF (G) vs. time on DOW2, A39, and A15 (T = 120 C, RBuOH/F = 79; RBuOH/
W
= 1.19; 1 g dried catalyst; catalyst loading 1.7 wt. %).
they are highly dependent on the medium polarity. An added difficulty
is a consequence of the non-homogeneity of the gel-phase. However, a
good picture of the spaces formed on swelling can be obtained from the
analysis of elution data of solutes of different molecular weight in an
aqueous solution using the ISEC (Inversion Steric Exclusion Chroma-
tography) technique, as proposed by Jerabek [62–64]. He modeled the
porous resin as a set of discrete fractions, each composed of pores having
simple geometry and uniform sizes. Macro and mesopores are described
8

E. Ramírez et al.
Applied Catalysis A, General 612 (2021) 117988
Table 3
Table 5
Morphology of tested resins in the dry state and swollen in water.
◦
Yield of HMF, LA, BMF, BL, FA, and BF on tested catalysts at 6 h (T = 120 C,
BuOH/F = 79; RBuOH/W = 1.19; 1 g dry catalyst; catalyst loading =1.7 wt. %). X
99.5 % on all resins.
R
F
Dry state Swollen in water (from ISEC data)
>
True pores
Gel
HMF
LA
BMF
BL
F
FA
F
BF
Catalyst
DVB
%)
Y
F
Y
F
Y
F
Y
Y
Y
F
phase
(
(%)
(%)
(%)
(%)
(%)
(%)
Catalyst
S
g
BET
V
g
BET
d
pore
ΣSpore
ΣVpore
d
pore
ΣVsp
Dowex
2
4
9.49 ±
0.45
10.0
31.8
28.4
35.7
22.8
± 1.2
22.7
±
2
3
c
2
3
c
3
(
m /
(cm /
(nm)
(m /
(cm /
(nm)
(cm /
5
0Wx2
± 0.5
11.1
± 1.4
27.6
± 1.2
27.1
± 1.6
36.2
a
a
b
b
g)
g)
g)
g)
g)
Dowex
12.7 ±
0.3
5
0Wx4
± 0.2
± 1.5
± 1.4
± 1.5
Dowex
50Wx2
Dowex
50Wx4
Amberlyst
31
2.677
1
.45
Amberlyst
4
15.3
13.5
22.3
22.8
34.2
20.8
2.061
2.096
3
1
CT124
Dowex
4
8
10.6
19.3
5.45
12.3
23.2
20.8
23.1
24.2
31.4
36.3
20.1
22.5
5
0Wx8
CT124
Dowex
50Wx8
Amberlyst
39
2.006
1.404
Amberlyst
8
20.5
34.5
43.6
10.1
13.5
10.5
24.4
15.1
13.7
25.3
17.7
13.4
32.6
35.8
34.5
22.0
18.8
14.4
3
9
Amberlyst
12
20
0.09
1.69
42.0
3 ×
17.6
29.7
31.8
56
0.155
0.188
0.616
15.0
15.5
12.4
1.643
1.136
0.622
ꢀ
4
1
6
10
Amberlyst
Amberlyst
16
0.013
46
1
5
Amberlyst
0.328
192
1
5
Note:
The
porosity
of
the
swollen
resins
is.θ =
= 0.99
ꢀ
)/ꢀ
)
Table 4
ΣVpore + ΣVsp ꢀ 1/
ρ
S
ΣVpore + ΣVsp
a
BET
g
Mean particle size of commercial catalysts determined by laser measurements in
Surface area from BET method (S
), and pore volume at P/P
0
BET
g
b
air, water, and 1-butanol and swelling in water and 1-butanol.
(V
).
Surface (ΣSpore) and volume (ΣVpore) of mesopores in the swollen state.
Mean pore diameter.
a
Mean particle diameter
m)
Swelling ratio
(%)
c
(μ
Native particle
size range (mm)
Catalyst
Air
Water
1-
Water
1-
by the cylindrical pore model in terms of pore surface and volume [63].
Micropores are described in terms of the Ogston model as spaces be-
tween randomly oriented rigid rods [65]. The relevant parameter of the
model is Vsp, the specific volume of swollen resins (pores + polymer
skeleton). The Ogston model also allows distinguishing between zones of
different polymer density in the swollen gel phase (in terms of total rod
Butanol
Butanol
Dowex
0.147 – 0.296
0.074 – 0.149
0.309 – 1.143
252
118
507
451
163
720
383
154
660
473
165
186
250
122
121
5
0Wx2
Dowex
5
0Wx4
Amberlyst
3
1
3
length per volume unit of the swollen polymer, nm/nm ).
CT124
Dowex
0.425 – 1.200
0.074 – 0.149
758
167
1194
235
1110
212
291
178
214
104
As Table 5 shows, pore surface and volume in macroreticular resins
increase on swelling, and mean mesopore diameter is in the range
12ꢀ 15 nm, large enough to make easier the access to the gel-phase
surface. Gel-type resins, which do not have mesopores in the dried
state, do not generate spaces that can be characterizable with the cy-
lindrical pore model. However, new spaces are formed within the gel-
phase of the macroreticular and gel-type resins when swelling in
water. Table 5 shows that the total specific volume of swollen resins
5
0Wx8
Amberlyst
0.410 – 1.141
0.410 – 0.948
0.310 – 1.041
540
562
650
768
690
741
724
630
729
188
87
141
40
3
9
Amberlyst
1
6
Amberlyst
48
21
1
5
a
Swelling ratio: the quotient of the increase of the volume of a particle in
water or 1-butanol respect the volume of the dry particle.
(
ΣVsp) highly varies from Dowex 50Wx2 to Amberlyst 15. Typically, gel-
type resins show higher ΣVsp values than macroreticular ones and,
among each resins group, ΣVsp decreases as DVB% increases. The dis-
tribution of zones of different polymer density of the tested resins is seen
Fig. 6. Morphological changes of gel-type and macroreticular resin particles due to swelling in polar media.
9

E. Ramírez et al.
Applied Catalysis A, General 612 (2021) 117988
Fig. 7. Distribution of zones of different polymer density of swollen PS-DVB catalysts determined from ISEC data analysis in aqueous solution.
◦
in Fig. 7. Gel-type resins with 2–4 % DVB and macroreticular ones with
were performed at 120 C. Table 6 shows the average values computed
8
% DVB show chain concentrations in swollen gel-phase of 0.4ꢀ 0.8 nm/
for d
and BMF. As seen, in each polymer zone K
molecules are bulkier. According to K values, the polymer zone of chain
m
. The smallest molecule is water, and the largest ones are BL, DBE,
3
nm , typical of slightly dense polymer mass. Gel-type resins with 8%
DVB and medium and highly crosslinked macroreticular ones show
0
values decrease as the
0
3
3
polymer densities in the range 0.8–1.5 nm/nm , typical of highly dense
density 0.2 nm/nm is accessible to all reactants and reaction products.
3
polymer mass. Equivalent pore diameter of spaces in polymer zones with
In the zone of density 0.4 nm/nm , fructose, 1-butanol, HMF, LA, and BF
3
3
3
3
chain density 0.1 nm/nm , 0.2 nm/nm , 0.4 nm/nm , 0.8 nm/nm and
have a distribution coefficient relatively high, but the accessibility of the
bulkiest molecules (BL, DBE, and BMF) is limited. Finally, the stiffer
3
1
.5 nm/nm are 9.3 nm, 4.8 nm, 2.6 nm, 1.5 nm and 1.0 nm, respectively
3
[
66]. Highly swollen resins, e.g. Dowex 50Wx2, have a very high ΣVsp
zones of resins (chain density 0.8 and 1.5 nm/nm ) are moderately
value (large pore volume within gel-phase) and low polymer density. As
a result, such resins have a very flexible polymer network and can
accommodate a greater number of bulky molecules. On the contrary,
low swollen resins, e.g, Amberlyst 15, have low ΣVsp value (small pore
volume within the gel-phase) and a high polymer density. Therefore,
they have a stiffer polymer network, contain fewer molecules in the
open spaces and bulky molecules tend to be excluded
accessible to fructose, 1-butanol, LA, BF, and HMF. However, BL, DBE,
and BMF are practically excluded. Water penetrates easily into all the
polymer zones, even the more impervious, favoring resin swelling.
Therefore, the accessibility of the compounds to the different polymer
zones of the swollen resin could explain the different activity shown by
tested catalysts. Like water, formic acid penetrates easily in all the resin
zones.
An indication that a resin zone with polymer chains concentration C
Tested resins have a similar acid capacity (Table 1) and acid site
strength [31], therefore their different catalytic behavior in 1-butanol/-
water solution can be attributed mainly to their morphology in a swollen
state. As Table 5 and Fig. 7 show, ΣVsp and the spaces that appear when
resins swell are very different and give rise to a distinct reaction arena in
each catalyst. Dowex 50Wx2 has the highest ΣVsp and the widest spaces
(
length of rods per total volume unit, pores + polymer skeleton) is
accessible to spherical molecules of diameter d
m
is given by the Ogston
distribution coefficient, K
0
, defined as [64]:
ꢀ
ꢀ
) )
K
0
= exp ꢀ 0.25
π
C d2 + d2
(5)
m
C
3
open on swelling (polymer density 0.2ꢀ 0.4 nm/nm ; equivalent to
The diameter of randomly distributed rigid rods representing poly-
2
.6–4.8 nm pores [66]), whereas Amberlyst 15 has the lowest ΣVsp and
C 0
mer chains is d . K ranges from 1 (the concentration of a given com-
the open spaces on swelling are the narrowest (polymer density 1.5
pound is the same as outside the gel-phase) to 0 (the compound is
excluded from the gel phase). Eq. 5 considers spherical molecules,
3
nm/nm ; equivalent to 1-nm pores [66]). Consequently, the volume of
liquid inside the Dowex resin is greater than in the Amberlyst 15, and in
the latter, the open spaces are much narrower. In the case of Dowex
m
consequently, the length is a good estimate of their effective size, d .
Molecular lengths were computed with Chem3D 18.0 software (Che-
mOffice) using the MM2 computational engine. First, the energy of the
structure was minimized, and then molecular dynamics calculations
5
0Wx2, the molecules of all compounds are considerably smaller than
the equivalent pore diameter of gel-phase spaces (4.8ꢀ 2.6 nm), and the
values ensure that their concentrations in the gel-phase are similar to
K
0
the bulk environment. In the case of Amberlyst 15, the size of the
bulkiest compounds is of the same order as spaces within the gel-phase
Table 6
Equivalent pore diameter and Ogston distribution coefficients, K
0
, in the
(
0
1.0 nm) and the low K values suggest that they tend to be excluded
different density of gel-phase zones of swollen resins. The diameter of randomly
from the gel phase. Fructose initially would be present in the gel-phase
of Dowex 50Wx2 in concentrations close to the bulk phase, producing
distributed rigid rods representing polymer chains is d
C
= 0.4 nm.
3
Polymer chain density (nm/nm )
Equivalent pore diameter (nm)
0.1
9.3
0.2
4.8
0.4
0.8
1.5
1.0
readily HMF. Since K
0
for HMF (0.85ꢀ 073) is high, it tends to remain in
2.6
1.5
the gel phase and reacts producing LA, BMF, and BL according to the
reaction scheme. In the case of Amberlyst 15, fructose has a very low
Compound
d
m
(nm)
K
0
Water
0.150
0.305
0.602
0.636
0.679
0.772
0.772
1.006
1.131
1.175
0.98
0.96
0.92
0.92
0.91
0.90
0.90
0.86
0.83
0.82
0.95
0.92
0.85
0.84
0.83
0.81
0.81
0.73
0.69
0.68
0.91
0.86
0.73
0.71
0.69
0.65
0.65
0.54
0.48
0.46
0.83
0.73
0.53
0.51
0.48
0.42
0.42
0.29
0.23
0.21
0.70
0.56
0.31
0.28
0.25
0.20
0.20
0.10
0.06
0.05
0
value of K (0.20). It can be inferred that although fructose penetrates
Formic acid
HMF
inside the resin, the amount of the monosaccharide in the gel-phase is
lower than in Dowex 50Wx2, and as Fig. 5A shows, it reacts slowly. The
distribution coefficient of HMF (0.31) is lower than in the gel-type resin,
and HMF tends to go out to the bulk liquid rather than remain in the
pores and continue the reaction. The size of LA and BMF is closer to the
1
-Butanol
Levulinic acid
Butyl formate
Fructose
Butyl levulinate
DBE
width of spaces open. The low K
0
values for both LA and BMF suggest
that they tend to be excluded from the gel phase, producing a small BL
BMF
1
0

E. Ramírez et al.
Applied Catalysis A, General 612 (2021) 117988
amount. Amberlyst 39 in a swollen state shows a noticeable ΣVsp with a
low acid site density perform better to synthesize BMF and BL. The best
values are found on Dowex 50Wx2 and Dowex 50Wx4 whereas the
worst one was obtained on Amberlyst 15. Correspondingly, HMF yield
3
polymer zone of density 0.4 nm/nm . The production of most of the
compounds of the reaction scheme is like that of Dowex 50Wx2, except
+
in the case of BMF. Amberlyst 39 also have an impervious zone of
increases with [H ]/ΣVsp showing that formation and decomposition of
3
density 1.5 nm/nm . The very low K
0
value of BMF in this polymer
HMF require wide spaces in the swollen resins and a low density of acid
sites (Fig. 8A). BF production is favored in low acid density resins
(Fig. 8F), whereas that of LA (Fig. 8B) and FA (Fig. 8E) is almost inde-
fraction (0.05) may explain the lower BMF yield observed. Therefore,
the different yields of Table 3 are related to the morphology of resins
swollen in the reaction medium: resins able to highly swell are more
effective in producing butyl levulinate.
+
pendent on [H ]/ΣVsp. It can be hypothesized that HMF formation and
conversion to BMF, BF, and BL might probably require the contribution
of at least two acid sites [67]. The reaction mechanisms to produce these
substances require adequate distances between acid centers which are
more easily found in highly swollen resins with a low density of acid
sites due to their more flexible morphology. Therefore, acid site density
has a relevant role in this reaction network. Since HMF, BMF, BL, and BF
are relatively bulky, swollen resins with lower acid site density perform
better. The best catalyst (Dowex 50Wx2) has the maximum ΣVsp value
In water/alcohol media, where resins largely swell by the interaction
with the polar liquid, the density of acid sites in the swollen polymer
+
(
[H ]/ΣVsp, a measure of the mean density of acid sites in the swollen
resin) illustrates better the influence of morphology than ΣVsp [31,42,
3]. Fig. 8 shows the yield of products and byproducts at 6 and 8 h of
4
reaction. There is an almost linear relationship between the yield of BMF
+
(
Fig. 8C) and BL (Fig. 8D) against [H ]/ΣVsp showing that resins with
Fig. 8. HMF (A), LA (B), BMF (C), BL (D), FA (E), and BF (F) yield as a function of [H ]/ΣVsp at 6 and 8 h. T = 120 ◦C, RBuOH/F = 79; RBuOH/W = 1.19; 1 g dried
+
catalyst; catalyst loading 1.7 wt. %.
1
1

E. Ramírez et al.
Applied Catalysis A, General 612 (2021) 117988
+
and therefore the lowest [H ]/ΣVsp one. On the contrary, the formation
described in section 2.5. Results with fresh and reused catalysts are
shown as a function of contact time since in the second reusing cycle the
experiment was performed with a catalyst mass significantly lower. As
seen in Fig. 9A, fructose readily reacts on reused resin and the conver-
sion of fructose is almost complete at a contact time higher than 400 h⋅g/
mol of fructose. Interestingly, reused catalyst gives rise to lower BL and
BMF formation, but higher HMF formation. HMF forms and decomposes
on reused resin at a lower reaction rate and the yield curves are above
that of fresh resin (Fig. 9B). It is also seen that the maximum yield of
HMF on reused resin appears at a longer contact time. As for BL, the
yield is about 25 % lower over reused resin, although the curves for first
and second reuse overlap (Fig. 9E). The yields of BMF (Fig. 9D) and BF
+
of FA and LA is almost independent on [H ]/ΣVsp
.
3
.7. Humins formation
The reaction medium is dark brown at the end of the experiments
showing the formation of soluble and insoluble polymers, called humins,
because of the degradation reactions of fructose [52], HMF [56], and/or
cross-reaction between fructose and HMF [57]. Such compounds are
difficult to quantify by analytical methods because of the number of
polymers that are formed. Thus the formation of humins may be
computed in terms of the decrease in the carbon balance, assuming that
all the compounds of the main pathway are identified and all the lack of
carbon is due to the formation of these polymers [68]. Both the mole
balance of fructose and the carbon atom balance of fructose at 6 h are
shown in Table 7. In general, fructose loss is 3% higher than fructose
carbon atoms, which is consistent with the stoichiometric excess of FA
(
not shown) in the second reuse are intermediate between those of fresh
and first reuse cycles. The yield of LA on the fresh and reused catalyst are
similar within the limits of the experimental error, although values on
the reused resin are a bit higher (Fig. 9C). Formation of FA over the
reused resin (not shown) is slower at the beginning of the reaction, but at
long reaction time, FA yield over fresh and reused catalyst tends to the
same value within the limits of the experimental error. Although more
cycles with reused catalysts are necessary to evaluate the catalyst span,
it is a promising fact that runs carried out with reused resin give similar
results. Finally, humins formation on fresh and reused resins is also of
the same order (Table 7, entries 5, 15, and 16).
(
plus BF) observed in most experiments. It was observed that fructose
loss is remarkable in the two first hours of the experiment and then in-
creases slowly with time.
Entries 1–5 show that humins formation increases with temperature.
On Dowex 50Wx2, fructose loss is negligible at 80 and 90 C: the liquid is
◦
only slightly yellow after runs. Some degradation to humins is observed
◦
The behavior of reused catalysts might be due to both the loss of
active sites and the formation of polymers (solid humins) covering the
resin during the reaction. To ascertain the cause of such behavior, fresh
and reused Dowex 50Wx2 were titrated against standard base after each
run, and S distribution was studied with SEM/EDS. As Table 8 shows,
after the second reuse cycle, the acid capacity of resin as determined by
titration decreases 47 % compared to fresh resin. EDS analyses show that
the percentage of S atoms decreases on reusing the resin, whereas that of
O atoms increases. The percentage of C atoms decreases only slightly.
Since the loss of S atoms after the second reuse cycle as determined by
EDS is only of 25 % compared to the fresh catalyst, it can be inferred that
in addition to deactivation for loss of sulfonic groups, the reaction rate is
limited by organic polymers covering the resin and thus several sulfonic
groups. The presence of polymers agrees with the fact that the per-
centage of O atoms in the reused catalyst increases, as measured by EDS
since humins covering the resin consist of a furan-rich polymer network
containing different oxygen functional groups [56]. Therefore, both
desulfonation and fouling by humins might explain the loss of activity
on a catalyst mass basis observed.
at 100 C (entry 3), and fructose loss increases substantially at 110 and
◦
1
20 C (entries 4 and 5). On comparing the loss of fructose in the ex-
◦
periments performed at 120 C on different resins (entries 5, 8–14), no
clear trend is seen. On the one hand, it seems that humins are formed in
higher amounts on the more active resins (Dowex 50Wx2 and Dowex
5
0Wx4). As in the less active resins (Amberlyst 16 and Amberlyst 15),
the amount of HMF is still high at 6 h (Fig. 5B, Table 3), this fact could
highlight the contribution of HMF to the formation of humins. However,
the highest values observed on Amberlyst 31 and, particularly, CT124
that have the same DVB% and a similar morphology in the swollen state
as Dowex 50Wx4 (Fig. 7, Table 5), suggest that humins formation is a
very complex process deserving a specific study. On the other hand,
experiments with a different mass of Dowex 50Wx2 (entries 5–7) show
that the formation of humins decreases slightly on increasing catalyst
mass. This fact could indicate that homogeneous reactions play a role in
the degradation of fructose and HMF to humins, and the formation of
HMF, BMF, LA, and BL accelerates by using larger catalyst mass limiting
in this way the extent of catalytic degradation reactions of fructose and
HMF.
If data of the reused resin are plotted against contact time on an acid
site basis, activity per acid site as determined by titration (Table 8) of
fresh and reused resin is very similar, in particular in the second reuse
cycle (Fig. 10). This fact suggests that humins block predominantly acid
sites located in the most impervious resin zones. Such acid sites hardly
3
.8. Catalyst reusability
Experiments with reused Dowex 50Wx2 were performed at 120 C.
◦
After each run, resin samples were washed and dried by the procedure
Table 7
Fructose moles and fructose carbon atoms lost forming humins at 6h reaction time (RBuOH/F =79; RBuOH/W =1.2).
Entry
Catalyst
Experimental conditions
C atoms lost (%)
Fructose mole lost (%)
◦
◦
1
2
3
4
5
6
7
8
9
Dowex 50Wx2
Dowex 50Wx2
Dowex 50Wx2
Dowex 50Wx2
Dowex 50Wx2
Dowex 50Wx2
Dowex 50Wx2
Dowex 50Wx4
CT 124
80 C, 1 g, fresh resin
0
0
90 C, 1 g, fresh resin
0
0
◦
◦
◦
◦
◦
◦
◦
◦
◦
◦
◦
◦
100 C, 1 g, fresh resin
1,7
3.6
110 C, 1 g, fresh resin
13.1
16.7
16.6
12.3
18.0
33.0
22.5
18.9
16.1
14.9
14.7
19.6
18.7
16.7
20.1
20.3
17.0
21.4
36.9
25.6
22.6
19.3
18.8
16.7
23.1
22.4
120 C, 1 g, fresh resin
120 C, 0.5 g, fresh resin
120 C, 2.0 g, fresh resin
120 C, 1 g, fresh resin
120 C, 1 g, fresh resin
10
11
12
13
14
15
16
Amberlyst 31
Dowex 50Wx8
Amberlyst 39
Amberlyst 16
Amberlyst 15
Dowex 50Wx2
Dowex 50Wx2
120 C, 1 g, fresh resin
120 C, 1 g, fresh resin
120 C, 1 g, fresh resin
120 C, 1 g, fresh resin
120 C, 1 g, fresh resin
◦
st
120 C, 1 g, 1 reuse cycle
◦
nd
120 C, 0.7 g, 2 reuse cycle
1
2

E. Ramírez et al.
Applied Catalysis A, General 612 (2021) 117988
Fig. 9. Fructose conversion (A) and HMF (B), LA (C), BMF (D), and BL (E) yield over fresh and reused Dowex 50Wx2 against contact time on a catalyst mass basis. T
◦
=
120 C; RBuOH/F = 79; RBuOH/W = 1.19; fresh and first reuse cycle: 1 g dried resin (catalyst loading 1.7 wt. %); second cycle: 0.7 g dried resin (catalyst loading 1.2
wt. %).
3
.9. Comparison with literature data
Table 8
SEM/EDS analysis and acid capacity of fresh and reused Dowex 50Wx2.
Data of the BL synthesis from fructose found in the open literature on
EDX analysis
solid catalysts are compared in Table 9 with the best ones obtained in the
present work. The comparison is not straightforward since different
reaction conditions and devices are used in the open literature. For easy
comparison, the reaction conditions are shown through parameters as
initial molar ratio 1-butanol/fructose, etc. As seen, in all these works 1-
butanol free of water is used in large excess compared to fructose. In this
way, the system shifts to BL production, and, at the same time, humins
formation is reduced [57,69]. It is to be noted that the information
supplied is generally limited, and important details of the course of re-
actions are not given i.e. humins formation, the FA (plus BF) excess over
LA (plus BL), or the yield of products other than BL. Only some of these
papers report the yield of BMF produced.
a
Catalyst
[H+] (meq/g)
%
C atoms
% O atoms
% S atoms
Fresh resin
4.98
4.13
3.35
2.65
74.54
74.69
72.74
73.24
15.78
17.45
20.02
20.34
8.68
7.56
7.25
6.37
Fresh resin, after use
st
After 1 reuse cycle
nd
After 2 reuse cycle
a
Titration against standard base by the Fischer and Kunin method [45].
take part in the reaction since such zones are almost inaccessible to
reactants, even in a swollen state. Sulfonic groups blocked by humins
cannot be measured by titration because humins prevent the reaction
with the base. It confirms that BL production occurs in the swollen gel-
phase of the resin, wide and flexible enough for the reactions of the
series-parallel scheme (Fig. 2) to take place.
As Table 8 shows, the best BL yield obtained with Dowex 50Wx2
(
entry 1) is lower than the yields found in the literature. This is probably
due to the presence of water initially since water highly inhibits the
1
3

E. Ramírez et al.
Applied Catalysis A, General 612 (2021) 117988
Fig. 10. Fructose conversion (A) and HMF (B), LA (C), BMD (D), and BL (E) yield over fresh and reused Dowex 50Wx2 against contact time on an acid site basis. T =
◦
1
20 C; RBuOH/F = 79; RBuOH/W = 1.24; fresh and first reused resin: 1 g dry resin (catalyst loading 1.7 wt. %); second reuse: 0.7 g dried resin (catalyst loading 1.2
wt. %).
catalytic activity of ion exchange resins in alcohol media [44,70–72]. An
two works, very large catalyst loading, and contact time were used and
such excessive catalyst loading might likely cause saturation problems
by solids inside the reactor.
additional experiment on Dowex 50Wx2 was performed with water-free
1
-butanol (entry 2). As seen, BL yield highly increases, and it is of the
same order as those found in the open literature. It is to be noted that
some BMF is in the reaction medium at 8 h and therefore BL yield could
still increase with time. Regarding the experiment carried out with a
As for inorganic catalysts (entries 8 and 9), runs were performed at a
higher temperature and a faster reaction is expected. However, BL yield
on TiO
2
(entry 8) is lower than that reported with Dowex 50Wx2 in entry
(not reported) may be important
1
-butanol/water mixture (entry 1), it is also noticeable the decrease in
2. The formation of humins on TiO
2
humins production that is consistent with the reduction of the stoi-
chiometric excess of FA (plus BF) observed (1.29 vs. 1.45).
enough to decrease BL formation since inorganic catalysts show deac-
tivation after a few cycles.
Experiments performed over Amberlyst 15 (entry 6) and CNT-PSSA
In summary, the use of ion-exchange resins, particularly those highly
swollen in aqueous-alcoholic media, is interesting because good BL
yields can be obtained at a moderate temperature. Energy savings could
be significant in industrial practice so that the production of BL from
fructose is a greener process
◦
at 120 C (entry 7) show BL yields higher than entry 2 although a
higher excess of 1-butanol is used. Brown et al. [73] reported on
◦
Amberlyst 15 at 100 C (entry 3) a similar BL yield as that found by us
with water-free 1-butanol (entry 2). However, Balakrishnan et al. re-
◦
ported at 110 C very low BL yields on Amberlyst 15 (entry 4) and
Dowex 50Wx8 (entry 5) [40]. In line with our experiments, a large
amount of BMF remains in the reactor yet what suggests that a higher BL
amount could be obtained at longer contact times. However, in the last
4. Conclusions
Ion exchange resins are effective catalysts to produce butyl
1
4

Table 9
Comparison with literature data on transformation of fructose to butyl levulinate.
◦
0
BL
BMF
Entry
1
Catalyst
T( C)
Reaction
Reaction
t(h)
8
Wt/n
F
X
F
Y
F
Y
F
(%)
Observation
Ref
a
conditions
Parameters
gh/mol
(%)
(%)
Dowex 50Wx2
120
Catalyst: 2 g
Fructose: 1.5 g
BuOH: 60 mL
Water: 10 mL
Catalyst: 1 g
R
BuOH/F = 79
1920
>99
43.4
25.1
Humins: 17.0%
FA excess : 1.45
This work
R
BuOH/W = 1.19
m
F
/W = 0,75
BL
Cat. Load. = 3.3%
F
Reu0073e: Y decreases by 22% after 3 cycles
2
3
4
5
6
7
8
Dowex 50Wx2
Amberlyts 15
Amberlyst 15
Dowex 50Wx8
Amberlyst 15
CNT-PSSA
120
100
110
110
120
120
150
R
BuOH/F = 79
8
800
3600
1950
1950
2160
2160
135
100
na
73.4
77
10.4
0
Humins: 9.8%
FA excess: 1.29
This work
[70]
Fructose: 1.8 g
BuOH: 70 mL
Water: 0 mL
m
F
/W = 1.8
Cat. Load. = 1.7%
Catalyst: 20 g
Fructose: 20 g
BuOH: 150 mL
Water: 0 mL
R
BuOH/F = 14.7
/W = 1
Cat. Load. : 12.4%
20
30
30
24
24
3
m
F
Catalyst 0.65 g
Fructose: 0.18 g
BuOH: 2g
R
BuOH/F = 270
97
16
71
56
[40]
m
F
/W = 0.28
Cat. Load. = 23%
Water: 0 mL
Catalyst 0.65 g
Fructose: 0.18 g
BuOH: 2g
R
BuOH/F = 270
97
14
m
F
/W = 0.28
[
40]
Cat. Load. = 23%
Water: 0 mL
Catalyst 20 mg
Fructose: 50 mg
BuOH: 4 mL
R
BuOH/F = 156
>99
>99
100
89
[41]
[41]
[39]
m
F
/W = 2.5
Cat. Load. = 0.6%
Water: 0 mL
BL
F
Catalyst 20 mg
Fructose: 50 mg
BuOH: 4 mL
R
BuOH/F = 156
87
Reuse: Y
decreases by 18% after 6 cycles
m
F
/W = 2.5
Cat. Load. = 0.6%
Water: 0 mL
TiO
nanoparticles
2
Catalyst 0.2 g
Fructose: 0.8 g
BuOH: 40 mL
R
BuOH/F = 98
62.8
m
F
/W = 4
Cat. Load. = 4.8%
Water: 0 mL
Catalyst 0.1 g
Fructose: 0.18 g
BuOH: 20 mL
Water: 0 mL
BL
F
9
Fe
2
(SO
4
)
3
190
R
BuOH/F = 217
3
300
>99
75
Reuse: Y
decreases by 27% after 5 cycles
[37]
m
F
/W = 1.8
Cat. Load. = 0.6%
a
m
F
/W = quotient of fructose to catalyst mass; Cat. Load. = Catalyst loading.
b
FA excess = quotient between the mole of FA +BF produced and those of BL + LA.

E. Ramírez et al.
Applied Catalysis A, General 612 (2021) 117988
[5] F.X. Covers, Solvent refining oil, U.S. Patent 2087473 (1937).
levulinate from fructose. BL formation has been studied in the temper-
ature range 80ꢀ 120 ◦C at 2.0 MPa in the liquid phase using a BuOH/
[6]
P.D. Bloom, Levulinic acid ester derivatives as reactive plasticizers and coalescent
solvents U.S. Patent 2010/0216915 A1 (2010).
water mixture (RBUOH/W = 1.19). The process is highly temperature-
[
[
7] J.R. Rieth, C.M. Leibig, J.D. Pratt, M. Jackson, U.S. Latex coating compositions
including carboxy ester ketal coalescents, methods of manufacture and uses
thereof. U.S. Patent Application 2012/0041110 A1 (2012).
8] D.J. Yontz, Fragrant formulations, methods of manufacture thereof and articles
comprising the same, U.S. Patent 2011/0274643 A1 (2011).
◦
sensitive, and BL production is detected only above 100 C. Butyl lev-
ulinate formation follows two pathways. Firstly, fructose dehydrates to
HMF, which in turn reacts with water or 1-butanol. HMF hydrolysis
gives rise to LA and FA, and the etherification of HMF yields BMF.
Finally, esterification of LA with butanol and alcoholysis of BMF yield
butyl levulinate. BL synthesis is highly influenced by the morphology of
the acidic ion-exchange resins in the liquid reaction medium. Acidic
resins greatly swell in butanol-water mixtures, and the quality of spaces
formed in the swollen polymer network is decisive. Resins showing the
best catalytic behavior are those that highly swell in the reaction me-
dium and have less acid site density in the swollen state. In this way, gel-
type Dowex 50Wx2 (2% DVB) is the most effective resin in terms of BL
yield. The formation of butyl levulinate is accompanied by the formation
of humins due to the degradation of fructose, HMF, and fructose-HMF
cross-reaction. The mole balance of fructose and the carbon atom bal-
ance of fructose show that between 17 and 37 % of monosaccharide is
lost forming humins. Finally, experiments with reused resin show that
the activity of fresh and twice-reused catalysts is similar on an acid site
basis. The loss of activity on a resin weight basis can be explained by
desulfonation and fouling because of humins deposition over the most
impervious polymer zone. The acid centers of such impervious polymer
zone hardly take part in the reaction, so the activity per accessible site
does not change.
[9] L. Lomba, B. Giner, I. Bandr ´e s, C. Lafuente, M.R. Pino, Green Chem. 13 (2011)
2
062–2070, https://doi.org/10.1039/c0gc00873b.
[
[
10] H.G. P o¨ ttering, P. Necas, Off. J. Eur. Union. 140 (2009) 88–112.
11] J. Yanowitz, E. Christensen, R.L. McCormick, Utilization of Renewable Oxygenates
As Gasoline Blending Components, Technical report NREL/TP-5400-50791, 2011.
12] E. Christensen, J. Yanowitz, M. Ratcliff, R.L. McCormick, Energy Fuels 25 (2011)
[
[
[
[
4
723–4733, https://doi.org/10.1021/ef2010089.
13] H. Joshi, B.R. Moser, J. Toler, W.F. Smith, T. Walker, Biomass Bioenergy 35 (2011)
3262–3266, https://doi.org/10.1016/j.biombioe.2011.04.020.
14] R.W. Jenkins, M. Munro, S. Nash, C.J. Chuck, Fuel 103 (2013) 593–599, https://
doi.org/10.1016/j.fuel.2012.08.019.
15] A. Demolis, N. Essayem, F. Rataboul, ACS Sustain. Chem. Eng. 2 (2014)
1338–1352, https://doi.org/10.1021/sc500082n.
[16] P.P.T. Sah, S.Y. Ma, J. Am. Chem. Soc. 52 (1930) 4880–4883, https://doi.org/
1
0.1021/ja01375a033.
[
17] H.A. Schuette, M.A. Cowley, J. Am. Chem. Soc. 53 (1931) 3485–3489, https://doi.
org/10.1021/ja01360a039.
[18] G.J. Cox, M.L. Dodds, J. Am. Chem. Soc. 55 (1933) 3391–3394, https://doi.org/
0.1021/ja01335a59.
19] H.J. Bart, J. Reidetschmeger, K. Schatka, A. Lehmann, Ind. Eng. Chem. Res. 33
1
[
(
1994) 21–25, https://doi.org/10.1021/ie00025a004.
[20] K.C. Maheria, J. Kozinsji, A. Dalai, Catal. Lett. 143 (2013) 1220–1225, https://doi.
org/10.1007/s10562-013-1041-3.
[
21] K.Y. Nandiwale, V.V. Bokade, Chem. Eng. Technol. 38 (2015) 246–252, https://
doi.org/10.1002/ceat.201400326.
[22] F.G. Cirujano, A. Corma, F.X. Llabr ´e s, Chem. Eng. Sci. 124 (2015) 52–60, https://
doi.org/10.1016/j.ces.2014.09.047.
[
[
[
23] S. Dharme, V.V. Bokade, J. Nat. Gas Chem. 20 (2011) 18–24, https://doi.org/
0.1016/S1003-9953(10)60147-8.
24] K. Manikandan, K.K. Cheralathan, Appl. Catal. A Gen. 547 (2017) 237–247,
https://doi.org/10.1016/j.apcata.2017.09.007.
25] M. Popova, A. Szegedi, H. Lazarova, A. Ristic, Y. Kalvachev, G. Atanasova,
N. Wilde, N. Novak Tusar, R. Glaser, Microporous Mesoporous Mater. 235 (2016)
Funding
1
Grant CTQ2014-56618-R from the Spanish Ministry of Economy and
Competitiveness is kindly acknowledged.
5
0–58, https://doi.org/10.1016/j.micromeso.2016.07.047.
[
[
[
[
[
[
[
26] F. Su, Q. Wu, D. Song, X. Zhang, M. Wang, Y. Guo, J. Mat. Chem. A 1 (2013)
CRediT authorship contribution statement
1
3209–13221, https://doi.org/10.1039/c3ta12412f.
27] D. Song, S. An, Y. Sun, Y. Guo, J. Catal. 333 (2016) 184–199, https://doi.org/
10.1016/j.jcat.2015.10.018.
28] W. Ciptonugroho, M.G. Al-Shaal, J.B. Mensah, R. Palkovits, J. Catal. 340 (2016)
Eliana Ramírez: Investigation, Methodology. Roger Bringu e´ :
Visualization, Writing - review & editing. Carles Fit e´ : Validation,
Formal analysis. Montserrat Iborra: Investigation, Resources, Meth-
odology. Javier Tejero: Supervision, Writing - original draft, Concep-
tualization, Project administration. Fidel Cunill: Conceptualization,
Writing - review & editing.
1
7–29, https://doi.org/10.1016/j.jcat.2016.05.001.
29] X. Zhou, Z. Xain Li, C. Zhang, X. Ping Gao, Y. Zhi Dai, G. Ying Wang, J. Mol. Catal.
A: Chem. 417 (2016) 71–75, https://doi.org/10.1016/j.nolcata.2016.03.006.
30] A.N. Chermahini, M. Nazeri, Fuel. Proc. Technol. 167 (2017) 442–450, https://doi.
org/10.1016/j.fuproc.2017.07.034.
31] M.A. Tejero, E. Ramírez, C. Fit ´e , J. Tejero, F. Cunill, Appl. Catal. A: Gen. 517
(
2016) 56–66, https://doi.org/10.1016/j.apcata.2016.02.032.
32] V. Trombettoni, L. Bianchi, A. Zupanic, A. Porciello, M. Cuomo, O. Piermatti,
A. Marrochi, L. Vaccaro, Catalysts 7 (2017) 235, https://doi.org/10.3390/
catal7080235.
Declaration of Competing Interest
The authors report no declarations of interest.
Acknowledgments
[33] G.D. Yadav, I.V. Borkar, Ind. Eng. Chem. Res. 47 (2008) 3358–3363, https://doi.
org/10.1021/ie800193f.
[
[
[
[
[
[
34] Y. Hishikawa, M. Yamaguchi, S. Kubo, Y. Yamada, J. Wood Sci. 59 (2013)
179–182, https://doi.org/10.1007/s10086-013-1324-8.
35] A. Demolis, M. Eternot, N. Essayem, F. Rataboul, New J. Chem 40 (2016)
3
747–3754, https://doi.org/10.1039/c5nj02493e.
We thank Rohm and Haas France for supplying Amberlyst resins, and
Purolite for supplying CT resin.
36] F. Yu, R. Zhong, H. Chong, M. Smet, W. Dehaen, B.F. Sels, Green Chem. 19 (2017)
153–163, https://doi.org/10.1039/c6gc02130a.
37] R. An, G. Xu, Ch. Chang, J. Bai, S. Fang, J. Energy Chem. 26 (2017) 556–563,
https://doi.org/10.1016/j.jechem.2016.11.015.
Appendix A. Supplementary data
38] L. Deng, Ch. Chang, R. An, X. Qi, G. Xu, Cellulose 24 (2017) 5403–5415, https://
doi.org/10.1007/s10570-017-1530-4.
39] C.H. Kuo, A.S. Poyraz, L. Jin, Y. Meng, L. Pahalagedara, S.Y. Chen, D.A. Kriz,
C. Guild, A. Gudz, S.L. Suib, Green Chem. 16 (2014) 785–791, https://doi.org/
Supplementary material related to this article can be found, in the
online version, at doi:https://doi.org/10.1016/j.apcata.2021.117988.
1
0.1039/c3gc40909k.
[
[
[
[
[
[
[
40] M. Balakrishnan, E.R. Sacia, A.T. Bell, Green Chem. 14 (2012) 1626–1634, https://
doi.org/10.1039/c2gc35102a.
41] R. Liu, J. Chen, X. Huang, L. Chen, L. Ma, X. Li, Green Chem. 15 (2013)
References
2
895–2903, https://doi.org/10.1039/c3gc41139g.
[
1] T. Werpy, G. Petersen, Top Value Added Chemicals From Biomass. Volume I –
Results of Screening for Potential Candidates From Sugars and Synthesis Gas, U.S.
Department of Energy, 2004, https://doi.org/10.2172/15008859. DOE/GO-
42] R. Bringu ´e , E. Ramírez, M. Iborra, J. Tejero, F. Cunill, J. Catal. 304 (2013) 7–21,
https://doi.org/10.1016/j.jcat.2013.03.006.
43] C. Casas, R. Bringu ´e , E. Ramírez, M. Iborra, J. Tejero, Appl. Catal. A: Gen. 396
1
02004-1992.
2] J.J. Bozell, G.R. Petersen, Green Chem. 12 (2010) 539–554, https://doi.org/
0.1039/b922014c.
3] J.J. Bozell, L. Moens, D.C. Elliott, Y. Wang, G.G. Neuenschwander, S.W. Fitzpatrick,
R.J. Bilski, J.L. Jarnefeld, Resour. Conserv. Recycl. 28 (2000) 227–239, https://doi.
org/10.1016/S0921-3449(99)00047-6.
(
2011) 129–139, https://doi.org/10.1016/j.apcat.2011.02.006.
[
[
44] M.A. P ´e rez, R. Bringu ´e , M. Iborra, J. Tejero, F. Cunill, Appl. Catal. A: Gen. 482
1
(
2014) 38–48, https://doi.org/10.1016/j.apcata.2014.05.017.
45] S. Fischer, R. Kunin, Anal. Chem. 27 (1955) 1191–1194, https://doi.org/10.1021/
ac60103a052.
46] R. Soto, M. Sv ¨a rd, V. Verma, L. Pradela, K. Ryan, A.C. Rasmuson, Int. J. Pharma.
[
4] A. Demirbas, Appl. Energy 88 (2011) 17–28, https://doi.org/10.1016/j.
apenergy.2010.07.016.
5
88 (2020) 119686, https://doi.org/10.1016/j.ijpharm.2020.119686.
1
6

E. Ramírez et al.
Applied Catalysis A, General 612 (2021) 117988
[
[
[
[
[
[
[
[
[
[
[
[
[
47] R. Stephenson, J. Stuart, J. Chem. Eng. Data 31 (1986) 56–70, https://doi.org/
0.1021/je00043a19.
48] L.A. Reber, W.M. McNabb, W.W.- Lucasse, J. Phys. Chem. 46 (1942) 500–511,
https://doi.org/10.1021/j150418a010.
49] N. Ebrahimi, R. Sadeghi, B.A.H. Ameen, J. Mol. Liquids 278 (2019) 164–174,
https://doi.org/10.1016/j.molliq.2019.01.011.
50] I. Thapa, B. Mullen, A. Saleem, C. Leibig, R.T. Baker, J.B. Giorgi, Appl. Catal. A
Gen. 539 (2017) 70–79, https://doi.org/10.1016/j.apcata.2017.03.016.
51] E. Ahmad, M.I. Alam, K.K. Pant, M.A. Haider, Green Chem. 18 (2016) 4804–4823,
https://doi.org/10.1039/c6gc01523a.
[60] Y. Sano, N. Yamaguchi, T. Adachi, J. Chem. Eng. Jpn. 7 (4) (1974) 255–261.
[61] B.E. Poling, J.M. Prausnitz, J.P. O’Connell, The Properties of Gases and Liquids,
fifth edition, McGraw-Hill, New York, 2001. Pages 11.41–11.42.
[62] K. Jerabek, ACS Symp. Ser. 635 (1996) 211–224.
1
[63] K. Jerabek, Anal. Chem. 57 (1985) 1595–1597, https://doi.org/10.1021/
ac00285a022.
[64] K. Jerabek, Anal. Chem. 57 (1985) 1598–1602, https://doi.org/10.1021/
ac00285a023.
[65] G. Ogston, Trans. Faraday Soc. 54 (1958) 1754–1757, https://doi.org/10.1039/
TF958540174.
[66] K. Jerabek, L. Holub, in: 13th International IUPAC Conference on Polymers and
Organic Chemistry, Montreal, 2009.
[67] D.R. Fernandes, A.S. Rocha, E.F. Mai, C.J.A. Mota, V. Teixeira da Silva, Appl. Catal.
A Gen. 425-426 (2012) 199–204, https://doi.org/j.apcata.2012.03.020.
[68] G. Garces, L. Faba, E. Diaz, S. Ord o´ n˜ ez, ChemSusChem 12 (2019) 1–12, https://
doi.org/10.1002/cssc.201802315.
52] T. Flannelly, M. Lopes, L. Kupianen, S. Dooley, J.J. Leahy, RSC Adv. 6 (2016)
5
797–5804, https://doi.org/10.1039/c5ra25172a.
53] T.D. Swift, C. Bagia, V. Choudhary, G. Peklaris, V. Nikolakis, D.V. Vlachos, ACS
Catal. 4 (2014) 259–267, https://doi.org/10.1021/cs4009495.
54] L. Deng, J. Li, D.M. Lai, Y. Fu, Q.X. Guo, Angew. Chem. Int. Ed. 48 (2009)
6
529–6532, https://doi.org/10.1002/anie.200902281.
55] W. Zeng, D.G. Cheng, H. Zhang, F. Chen, Zhan Xiaoli, Reac. Kinet. Mech. Cat. 100
[69] X. Hu, S. Kadarwati, S. Wang, Y. Song, M.D. Mahmudul Hasan, Ch. Li, Fuel. Proc.
Technol. 137 (2015) 212–219, https://doi.org/10.1016/j.fuproc.2015.04.024.
[70] R. Bringu ´e , E. Ramírez, M. Iborra, J. Tejero, F. Cunill, Chem. Eng. J. 246 (2014)
71–78, https://doi.org/10.1016/j.cej.2014.02.042.
(
2010) 377–384, https://doi.org/10.1007/s11144-010-0187-x.
56] G. Tsilomelekis, M.J. Orella, Z. Lin, Z. Cheng, W. Zheng, V. Nikolakis, D.G. Vlachos,
Green Chem. 18 (2016) 1983–1993, https://doi.org/10.1039/c5gc01938a.
57] X. Hu, C. Lievens, A. Larcher, C.Z. Li, Bioresour. Technol. 102 (2011)
[71] C. Casas, E. Ramírez, C. Fit ´e , M. Iborra, J. Tejero, AIChE J. 63 (2017) 3966–3978,
https://doi.org/10.1002/aic.15741.
1
0104–10113, https://doi.org/10.1016/j.biortecg.2011.08.040.
58] J.M. Smith, Chemical Engineering Kinetics, 3rd edition, McGraw-Hill Kogakusha,
Aukland, 1981. Pages 389-412.
[72] R. Bringu ´e , J. Tejero, M. Iborra, J.F. Izquierdo, C. Fit ´e , F. Cunill, Top. Catal. 45
(2007) 181–186, https://doi.org/10.1007/s11244-007-0262-3.
[73] D.W. Brown, A.J. Floyd, R.G. Kinsman, Y. Roshan-Ali, J. Chem. Tech. Biotechnol.
32 (1982) 920–924, https://doi.org/10.1002/jctb.5030320730.
59] G.B. Froment, K.B. Bischoff, Chemical Reactor Analysis and Design, second edition,
Wiley, New York, 1990. Page 168.
1
7