Direct versus acetalization routes in the reaction network of catalytic HMF etherification

Article abstract of DOI:10.1039/c7cy02339a

The etherification of HMF (5-hydroxymethylfurfural) to EMF (5-(ethoxymethyl)furan-2-carbaldehyde) is studied over a series of MFI-type zeolite catalysts containing different heteroatoms (B, Fe, Al), aiming to understand the effect of different isomorph substitutions in the MFI framework on the reaction pathways of HMF conversion. The rate constants in the reaction network are determined for these different catalysts and analyzed with respect to the amount of Br?nsted and Lewis acid sites determined by FT-IR pyridine adsorption. Two different pathways of EMF formation, i.e. direct etherification and via acetalization, were evidenced. The Lewis acid sites generated from the presence of aluminum are primarily active in catalyzing direct HMF etherification to EMF, which has a rate constant about one order of magnitude lower than the etherification of the corresponding acetals. This behaviour is due to the competitive chemisorption between hydroxyl and aldehyde groups (both present in HMF) on the Lewis acid sites catalyzing the etherification. A cooperation phenomenon between Br?nsted and Lewis acid sites is observed for the HMF acetal etherification to EMF acetal. In the reactions of direct HMF acetalization and deacetalization of the EMF acetal, the turnover frequencies for Silicalite-1 and B-MFI samples are about twice those for Fe-MFI and Al-MFI samples. This is attributed to the different reactivity of strong silanol groups associated with surface defects on the external surface in Silicalite-1 and B-MFI. These sites are also responsible for the EMF-to-EOP (ethyl 4-oxopentanoate) reaction step. In the deacetalization reaction of the EMF acetal, the behavior is determined from the presence of water (product of reaction) favouring the back reaction (aldehyde formation).

Full text of DOI:10.1039/c7cy02339a

Catalysis
Science &
Technology
View Article Online
View Journal
PAPER
Direct versus acetalization routes in the reaction
network of catalytic HMF etherification†
Cite this: DOI: 10.1039/c7cy02339a
a
a
a
a
M. Migliori, *b
P. Lanzafame, * G. Papanikolaou, S. Perathoner, G. Centi,
b
b
b
E. Catizzone, A. Aloise and G. Giordano
The etherification of HMF (5-hydroxymethylfurfural) to EMF (5-(ethoxymethyl)furan-2-carbaldehyde) is
studied over a series of MFI-type zeolite catalysts containing different heteroatoms (B, Fe, Al), aiming to un-
derstand the effect of different isomorph substitutions in the MFI framework on the reaction pathways of
HMF conversion. The rate constants in the reaction network are determined for these different catalysts
and analyzed with respect to the amount of Brønsted and Lewis acid sites determined by FT-IR pyridine
adsorption. Two different pathways of EMF formation, i.e. direct etherification and via acetalization, were
evidenced. The Lewis acid sites generated from the presence of aluminum are primarily active in catalyzing
direct HMF etherification to EMF, which has a rate constant about one order of magnitude lower than the
etherification of the corresponding acetals. This behaviour is due to the competitive chemisorption be-
tween hydroxyl and aldehyde groups (both present in HMF) on the Lewis acid sites catalyzing the ether-
ification. A cooperation phenomenon between Brønsted and Lewis acid sites is observed for the HMF ace-
tal etherification to EMF acetal. In the reactions of direct HMF acetalization and deacetalization of the EMF
acetal, the turnover frequencies for Silicalite-1 and B-MFI samples are about twice those for Fe-MFI and Al-
MFI samples. This is attributed to the different reactivity of strong silanol groups associated with surface de-
fects on the external surface in Silicalite-1 and B-MFI. These sites are also responsible for the EMF-to-EOP
Received 16th November 2017,
Accepted 29th January 2018
(ethyl 4-oxopentanoate) reaction step. In the deacetalization reaction of the EMF acetal, the behavior is de-
DOI: 10.1039/c7cy02339a
termined from the presence of water (product of reaction) favouring the back reaction (aldehyde
formation).
rsc.li/catalysis
1
5
from methanol and isobutene. In general, it is known that
both Brønsted and Lewis acid sites could be active in these re-
actions and that it is necessary to tune the acidity because of
the presence of other reactions, such as acetalization, compet-
ing with etherification in HMF conversion, although it may
also lead to interesting products (biodegradable surfac-
Introduction
Etherification is a relevant reaction in the liquid phase trans-
formation of biomass platform molecules, for example to
produce i) diesel additives by etherification of furanyl alcohol
or HMF (5-hydroxymethylfurfural), ii) mono-alkylated prod-
1
,2
3
4
–6
1
6
ucts used to obtain high quality alkyl naphthenic kerosene by
tants). In fact, different types of functional groups (alco-
holic, aldehydic) are present in HMF as well as in other plat-
form molecules, and the control of catalyst reactivity is thus a
7
HMF
self-etherification
to
5,5′-oxyIJbismethylene)-2-
furaldehyde (OBMF), which is an interesting prepolymer and
8
17
antiviral precursor, and 5-alkoxymethylfuroate for a new class
relevant aspect from the application perspective.
4
a,6
18
3
of biodegradable anionic surfactants by HMF selective ether-
Although oxide-,
carbon- and ion-exchange resin-type
9
ification. Other relevant etherification reactions include glyc-
catalysts have been used for HMF etherification, zeolite-type
1
0,11
4b,19,20
erol etherification with isobutene,
methanol etherification
catalysts
offer the advantage of fine tuning the acidity
1
2–14
to DME,
and methyl tert-butyl ether (MTBE) synthesis
features (amount, nature and strength of the Brønsted and
2
0c
Lewis sites),
allowing better understanding of the role of
these active sites in the reaction network of HMF ether-
ification, including the possible presence of cooperative ef-
fects. In fact, recently, the presence of cooperative effects be-
tween Lewis and Brønsted acid sites has been reported in
a
Departments of ChiBioFarAm and MIFT- Section of Industrial Chemistry, Univer-
sity of Messina, ERIC aisbl and CASPE-INSTM, V.le F. Stagno d'Alcontres 31,
9
b
8166 Messina, Italy. E-mail: planzafame@unime.it
Department of Environmental and Chemical Engineering, University of Calabria,
2
1
glucose transformation over zeolites, H-transfer in furfural
via Bucci, 87036 Rende, Italy. E-mail: massimo.migliori@unical.it
Electronic supplementary information (ESI) available. See DOI: 10.1039/
2
2
23
conversion, multistep biomass conversion and glycerol
†
2
4
c7cy02339a
dehydration.
This journal is © The Royal Society of Chemistry 2018
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
The aim of this paper is to analyze how the rate constants
in the reaction network of HMF etherification with ethanol
depend on the nature and amount of the Lewis and Brønsted
acid sites, which are changed in a homogenous series of
MFI-type zeolites containing different heteroatoms, T-
Silicalite-1 (T = B, Fe and Al). The analysis of the results pro-
vides some novel insight into the role of these acid sites in
the different steps of the reaction network of HMF ether-
ification and the presence of cooperative effects, leading to
different reaction pathways over different catalysts.
Results and discussion
Characterization of the catalysts
The XRD patterns of the synthesized samples (Fig. 1a) show
characteristic peaks of the MFI-type topology without any evi-
2
5
dence of the presence of other phases. The comparison of
the XRD patterns in the 2θ range from 20 to 28°, reported in
Fig. 1b, reveals the presence of a doublet reflection around
2
θ = 25° for Silicalite-1 and Al-MFI, which is characteristic of
a monoclinic symmetry. For B-MFI, a single reflection at 2θ =
4.4° ascribed to orthorhombic symmetry is observed, while both
Fig. 2 SEM images of the samples: Silicalite-1 (a), B-MFI (b), Fe-MFI (c)
and Al-MFI (d).
2
26
framework symmetries are present for the Fe-MFI sample.
The results of scanning electron microscopy (SEM) for all
prepared samples are shown in Fig. 2. For Silicalite-1, a
round-like shape morphology for crystals with a relatively
4
b
uniform size of 2 μm is observed. The intergrowth of
smaller crystals with a typical coffin-shape was also observed
for this sample. Crystals with different sizes and morphol-
ogies were observed for the other samples. The image in
Fig. 2b indicates that B-MFI consists of heavily twinned near-
spherical crystals with a size of about 10 μm. The incorpora-
tion of iron in the MFI-type structure led to a more jagged
morphology of elliptical crystallites with a size of about 5 μm
(
Fig. 2c). Furthermore, the SEM images confirmed that the
preparation of the Al-MFI sample produced more elongated
crystals, which present non-homogeneous dimension
Fig. 2d).
The textural characteristics of the investigated catalysts
were determined by nitrogen adsorption measurement
a
(
(
4
Table 1). All samples exhibit high BET surface areas (366–
2
−1
3
−1
08 m g ) and micropore volumes (0.079–0.110 cm g ) in
2
7–29
the typical range of an MFI structure,
in accord with
their good XRD crystallinity (Fig. 1a). Furthermore, a marked
reduction in the external surface area was observed for T-
Silicalite-1 containing heteroatoms.
3
The acidity of the catalysts was studied by NH -TPD and
pyridine FT-IR measurements. The profiles of NH -TPD (see
3
Fig. S1 in the ESI†) can be deconvoluted into two peaks re-
lated to weak and medium–strong acid sites, in the tempera-
ture range of 100–300 °C and 300–450 °C, respectively. The
total acidity decreases following the order Al-MFI > Fe-MFI >
B-MFI > Silicalite-1. After the introduction of the hetero-
atoms, a different distribution in the strength of acid sites
was also observed. From the ratio of the two deconvolution
curve areas, an increase of the weak acidity percentage from
15% of Silicalite-1 to 22%, 25% and 31% for B-MFI, Fe-MFI
Fig. 1 X-ray diffractograms of MFI-type catalysts in the 2θ ranges
5–45° (a) and 20–28° (b).
and Al-MFI was respectively found.
Catal. Sci. Technol.
This journal is © The Royal Society of Chemistry 2018

View Article Online
Catalysis Science & Technology
Paper
Table 1 Textural properties of MFI-type samples
a
b
2
−1
c
2
−1
c
3
−1
Sample
Si/Tbulk
S.A.BET (m g
)
Areaexter (m g
)
V
micropor. (cm g
)
dpore (Å)
Silicalite-1
B-MFI
Fe-MFI
Al-MFI
∞
408
366
406
405
345
195
204
233
0.110
0.087
0.094
0.079
33
23
29
32
95
100
100
a
b
c
Measured by A.A. Calculated using the BET method. Calculated using the t-plot method.
The nature of acid sites present in Silicalite-1, B-MFI, Fe-
MFI and Al-MFI was investigated by FT-IR adsorption of pyri-
dine, having a steric encumbrance similar to the molecules
vent, at 140 °C and autogenous pressure. Fig. 4 reports the
time course of HMF conversion for all synthesized catalysts.
All catalysts were active in the etherification reaction of HMF
showing a similar trend with an increase of HMF conversion
during the reaction time until reaching a plateau after 5
hours. The catalytic activity of the samples follows the total
acidity trend, obtaining a higher HMF conversion by using
the Al-MFI catalyst, which reaches 91% at the end of the
reaction.
Table 3 reports the pseudo-first order rate constants for
HMF depletion, which can be estimated from the data
reported in Fig. 4 and the turnover frequencies determined
with respect to the overall amount of Lewis and Brønsted
sites, as determined by FT-IR measurements (Table 2).
The pseudo-first-order rate constant for the HMF overall
conversion increases in going from Silicalite-1 to Al-MFI, in
agreement with the conversion data (Fig. 4). If the turnover
frequency is analyzed with respect to the amount of acid sites
determined by pyridine adsorption, it could be observed that
the turnover frequency is essentially independent on the
heteroatom when considering the sum of strong Lewis and
Brønsted acid sites. An expected maximum in the turnover
frequency for the B-MFI sample is observed due to the minor
content of strong Lewis acid sites (see Table 2).
3
0,31
involved in the etherification reaction.
Fig. 3 shows the
−
1
infrared spectra in the range of 1700–1400 cm , commonly
used to identify Lewis and Brønsted acid sites, of a represen-
tative sample (Al-MFI), where the red line indicates the spec-
trum obtained after adsorption of pyridine, and black and
green lines refer to the profiles after outgassing at room tem-
perature and 150 °C, respectively.
Table 2 reports the concentration of the Lewis and
Brønsted acid sites of the catalysts, obtained after
−
1
−1
deconvolution of the peaks at 1445 cm and 1545 cm , re-
spectively.
3
1,32
Silicalite-1 presents the lowest acidity with a
major contribution of Lewis acid sites and a very low content
−1
of strong Brønsted acid sites (9 μmol gcat ). As expected and
according to TPD results, the modification of Silicalite-1
structure with B, Fe and Al increases the acidity of Silicalite-1
and introduces both Lewis and Brønsted acid sites. The Al-
MFI catalyst presents the highest acidity, with a major contri-
−1
bution of weak Lewis sites (307.7 μmol gcat ) and a concen-
−1
tration of strong Brønsted acid sites of 56.6 μmol gcat
.
Catalytic activity and reaction network
In this series of catalysts, both strong Lewis and Brønsted
acid sites are active in HMF conversion. The behaviour is de-
termined from the amount of strong acid sites, which is
influenced by the introduction of heteroatoms. These find-
ings are also confirmed by the direct correlation observed be-
tween the acid site density and the rate constant of HMF con-
version (Fig. 5). Also in this case, a deviation was observed
for B-MFI which presents the lowest acid site density.
The catalytic performance of the samples was studied in the
etherification of HMF, using ethanol as the reactant and sol-
The reaction products, 5-(ethoxymethyl)furan-2-carb-
aldehyde
(EMF),
ethyl
4-oxopentanoate
(EOP),
Table 2 Brønsted and Lewis acid sites distribution of the catalysts
obtained from pyridine FT-IR measurements
Lewis acid sites
Brønsted acid sites
−1
−1
(μmol gcat
)
(μmol gcat
)
a
b
c
Sample
Weak
Strong
Strong
Silicalite-1
B-MFI
Fe-MFI
Al-MFI
30.4
124.4
122.3
307.7
20.7
10.8
23.3
31.1
9.0
13.4
35.4
56.6
−1
a
−1
b
Fig. 3 FT-IR spectra of the Al-MFI sample in the 1700–1400 cm re-
gion after adsorption of pyridine (red line) and evacuation at room
temperature (black line) and at 150 °C (green line).
Calculated using the band at 1445 cm , r.t. Calculated using the
1 c
−
−1
band at 1445 cm , 150 °C. Calculated using the band at 1545 cm
150 °C.
,
This journal is © The Royal Society of Chemistry 2018
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
Fig. 5 Trend of the rate constant of HMF depletion with the acid sites
density.
Fig. 4 Time course of HMF conversion for all synthesized catalysts.
5
5
-(hydroxymethyl)furfural diethyl acetal (HMFDEA) and
-(ethoxymethyl)furfural diethyl acetal (EMFDEA), were identi-
The increase of the catalyst acidity due to the introduction
of different heteroatoms in the Silicalite-1 structure favours
the formation of EMF. For the Al-MFI catalyst, which presents
the major quantity of Lewis acid sites, an EMF yield of 70%
is observed after five hours of reaction. Moreover, it is possi-
ble to prove that the increase of the acidity also affects the
pathway to HMFDEA, which shows the highest yields during
the first hour of reaction and then decreases faster than in
the case of Silicalite-1 catalyst. The distribution of the reac-
tion products during the course of the etherification reaction
allows us to hypothesize two different routes of HMF ether-
ification, i.e. direct etherification and a parallel pathway via
acetalization.
In order to confirm this hypothesis, specific tests
starting from an ethanolic solution of HMFDEA were car-
ried out under the same experimental conditions. Fig. 7
shows that almost 90% of HMFDEA is consumed during
the first hour of reaction, leading to the formation of
EMFDEA, EMF and EOP. Between 1 and 5 h, a complete
conversion of HMFDEA and EMFDEA is observed, yielding
EMF as the main product.
fied and quantified by GC-FID analyses in the final solution
Scheme 1). The direct etherification of the hydroxyl group of
(
HMF leads to the formation of EMF, whilst EOP is obtained
by the opening of the furanic ring.
4
b,33
In previous studies,
we revealed two main conver-
sion pathways for HMF, leading to the formation of EMF
and EOP, activated by the different nature of acidic sites.
A secondary route for the formation of EOP starting from
EMF was furthermore proposed. In the present work, it
was possible to identify the formation of acetals, HMFDEA
obtained by acetalization of HMF and EMFDEA formed
through etherification of the free hydroxyl functional group
of HMFDEA.
Fig. 6 shows the product distribution for each catalyst as a
function of reaction time. For the Silicalite-1 catalyst, the
yield of EMF, the main reaction product, increases with time
and reaches its maximum value after 5 hours of reaction.
A second aspect to be highlighted is the high yield of
HMFDEA at the lowest conversion values in the first hour of
reaction.
Moreover, the highest value of EMFDEA yield is observed
at the third hour of reaction with a subsequent drop when in-
creasing the reaction time. The yield of EOP always remains
below 2%, in accord with the low Brønsted acidity of this
catalyst.
4
a
At the end of all the reactions, the mass balance is always
about 90%; the missing 10% is due to the formation of
humins and other condensed products (see Table S1 in the
Table 3 Pseudo-first order rate constants and turnover frequencies for
HMF depletions
a
Rate constant,
Turnover frequency
−
1
−1
−1
Sample
k
HMF (h
)
(h μmol
)
Silicalite-1
B-MFI
Fe-MFI
Al-MFI
0.360
0.623
0.685
1.074
0.121
0.257
0.117
0.122
a
−1
Rate constant (h ) divided by the amount of strong Lewis and
Brønsted sites (μmol) determined by FT-IR pyridine measurements.
Scheme 1 Reaction products of HMF etherification.
Catal. Sci. Technol.
This journal is © The Royal Society of Chemistry 2018

View Article Online
Catalysis Science & Technology
Paper
Fig. 6 Evolution of product yields for the catalysts Silicalite-1 (a), B-MFI (b), Fe-MFI (c) and Al-MFI (d).
ESI†). The results discussed above are in accordance with the
reaction scheme presented in Scheme 2, in which a pathway
of EMF formation via acetals is evidenced, with two different
equilibria between HMF ↔ HMFDEA and EMF ↔ EMFDEA.
The proposed reaction pathway is also in agreement with the
the reaction volume fixed at 3.5 ml and t is the reaction time
(h).
The absence of mass transfer limitations verified as well
that the first-order kinetics correctly describes the behaviour.
Fig. 8 shows the fitting of the experimental data, whilst
Fig. 9 compares the model predictions and experimental data
to evidence the overall satisfactory correlation.
Table 4 summarizes the estimated rate constants as a
function of catalyst composition for all the reaction steps of
the reaction network indicated in Scheme 2.
34
results of Lewis et al.
The reaction system was modelled as an isotherm–iso-
choric uniformly mixed batch reactor, assuming a first-order
kinetic equation for all reaction steps, after verification that
this simplified model provides a reasonably good description
of the catalytic behavior. Due to the large excess of ethanol,
the concentration of the alcohol was assumed constant and
not considered in the kinetic equation.
Role of acid sites in various steps of the reaction network
The following material balances can be written:
The analysis of the dependence of the rate constants in the
reaction network of HMF etherification on the catalyst com-
position (Table 4), in relation to the quantification of their
(
(
(
1)
2)
3)
(
(
4)
5)
where n and C are the mole number (mmol) and molar con-
centration (mmol l ) of i-th species, respectively, whilst V is
i
i
Fig. 7 Evolution of EMFDEA, EMF and EOP products during the
etherification of HMFDEA.
−
1
This journal is © The Royal Society of Chemistry 2018
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
From the analysis of the k parameter in Table 4, it clearly
1
appears that the presence of aluminum is necessary to pro-
mote the direct etherification of HMF to EMF, since only Al-
MFI shows a non-zero k value. Although there is no detailed
1
analysis of the reaction network in this paper, previous works
showed that this reaction step is catalysed preferably by
4
b
Lewis acid sites. Despite the fact that the B and Fe samples
also presented some Lewis acidity, no activity towards direct
HMF-to-EMF conversion was observed, suggesting that only
Lewis acid sites generated from aluminum species could be
effective for this reaction step.
On the contrary, if the EMF-to-EOP step is considered,
Table 4 shows an inverse correlation, i.e. the lower the overall
Scheme 2 Proposed reaction network in the etherification of HMF
with ethanol.
acidity, the higher the rate constant k . In a previous work
5
4
b
carried out over Silicalite-1 samples, it was shown that the
EMF ring opening to EOP was promoted by strong Brønsted
acid sites, whilst the results presented here show an opposite
trend. However, this issue could be clarified by considering
the different features of Brønsted acid sites. In fact, it is well
known that Brønsted acid sites in zeolites commonly refer to
both bridging OH groups (SiOH-T, T = B, Fe, Al) and silanol
groups. In Silicalite-1, the strong Brønsted sites detected via
acidity by FT-IR pyridine adsorption (Table 2), provides some
interesting insight into the relationship between the type and
amount of acid sites and different reactivities over the pro-
posed reaction scheme.
As mentioned before analysing the turnover frequency of
this series of catalysts (Table 3), the HMF conversion is a
function of the total amount of Lewis and Brønsted acid
sites, rather than the nature of heteroatoms.
4
b
FT-IR analysis can be associated with silanol groups. There-
fore, since Silicalite-1 showed the maximum k value, Si–OH
5
This may suggest that the introduction of the heteroatoms
in the synthesis changes the amount of acid sites rather than
their specific activity. However, this general observation may
be not valid if the specific rates in the reaction network are
analysed. It is noteworthy that the proposed discussion is
based on the acidity measured via FT-IR pyridine adsorption
data, providing a good quantification of the amount of the
groups could be assumed as the most active towards the
EMF-to-EOP reaction step. It should be also noted that the
higher activity of Silicalite-1 is related to its higher external
surface area (Table 1) where these silanol groups are prefer-
entially located. FT-IR characterization of these samples, not
reported here for conciseness, is in good agreement with this
interpretation.
different acid sites. The acid site estimation via NH
3
-TPD
If the reversible reaction step HMF ↔ HMFDEA is consid-
measurement are in line with the FT-IR results, although
with some differences since i) it is not possible to discrimi-
nate between Brønsted and Lewis acid sites in the NH -TPD
3
test and ii) ammonia might reach some sites not accessible
by pyridine, due to its reduced molecular size. On the other
hand, accessibility by pyridine is more representative of the
accessibility by HMF.
2
ered, it was found that the k /k2r ratio linearly depends on
the Brønsted acid site concentration (Fig. 10), confirming
that the acetalization reaction is favoured by the increase of
3
4
Brønsted acidity. Therefore, also for this step, Al-MFI shows
the highest activity, followed by Fe-MFI, B-MFI and Silicalite-
1. The conversion of EMFDEA to EMF (deacetalization) is also
favoured by the increase of the Brønsted acid site concentra-
tion (Fig. 10). Table 5 reports the turnover frequencies esti-
mated for direct HMF acetalization (HMF ↔ HMFDEA) and
EMFDEA deacetalization (EMFDEA ↔ EMF) reactions, by
considering the k /k or k /k ratio for HMF ↔ HMFDEA or
2
2r
4r
4
EMFDEA ↔ EMF, respectively, as a measure of the net reac-
tion divided by the amount of strong Brønsted acid sites (de-
termined by FT-IR pyridine adsorption).
We may note that i) the turnover frequencies for Silicalite-
1
and B-MFI samples are about twice those for Fe-MFI and
Al-MFI samples for both direct acetalization and
deacetalization reactions, and ii) the turnover frequency of
deacetalization (EMFDEA ↔ EMF step) is at least 4 times
higher than that of the direct acetalization step (HMF ↔
HMFDEA). The first aspect indicates that the intrinsic reactiv-
ity of the strong Brønsted acid sites in Silicalite-1 and B-MFI
is about two times higher than that of Fe-MFI and Al-MFI in
the acetalization reaction, contrary to what may be expected.
Fig. 8 Comparison between the model (line) and experimental data
(
symbol).
Catal. Sci. Technol.
This journal is © The Royal Society of Chemistry 2018

View Article Online
Catalysis Science & Technology
Paper
Fig. 9 Comparison between experimental species moles and model predictions.
This is likely related to the preferential presence of strong
acid silanol groups associated with surface defects on the ex-
ternal surface in Silicalite-1 and B-MFI rather than in Fe-
MFI and Al-MFI samples, in agreement with the SEM data
tion. With the progress of the reaction, the water formed
tends to favour the deacetalization reaction, and this explains
why with the progress of the reaction the deacetalization (re-
action step 4) is favoured over the direct acetalization (reac-
tion step 2).
4
b
(Fig. 2) and FT-IR characterization of these samples, which is
not reported here for conciseness.
The results discussed up to now indicate that either Lewis
or Brønsted acid sites catalyze the reaction steps in the HMF
etherification reaction network. However, we could note that
reaction step 3 (HMFDEA to EMFDEA) depends instead on
the ratio between strong Brønsted and Lewis acid sites,
rather than on their specific amounts.
The second aspect may appear rather contradictory. In
fact, the same type of strong Brønsted acid sites apparently
catalyzes the direct acetalization (HMF ↔ HMFDEA) and the
EMFDEA deacetalization (EMFDEA ↔ EMF). While there is a
difference in the electron donating character of the ethoxy
group in EMF with respect to the hydroxyl group in HMF, it
is unlikely that this is the main reason for this effect. How-
ever, we should consider that even if not indicated in
Schemes 1 and 2, the acetalization reaction leads to the for-
mation of water (Scheme 3) as does the etherification reac-
Table 4 Pseudo-first order rate constants for the reaction steps in the
HMF etherification network
−
1
Rate constant (h
)
Al-MFI
Fe-MFI
B-MFI
Silicalite-1
k
k
k
k
k
k
k
1
0.172
0.653
0.487
1.190
0.235
1.210
0.030
0.992
0
0
0
2
1.200
1.740
0.901
1.070
3.780
0.032
0.980
1.020
1.590
0.814
2.140
6.010
0.051
0.961
1.630
3.980
0.357
0.200
0.375
0.106
0.969
2r
3
4
4r
Fig. 10 The effect of Brønsted acid sites concentration on k
and k4r/k rate coefficient ratios.
2
/k2r
5
2
R
4
This journal is © The Royal Society of Chemistry 2018
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
Table 5 Turnover frequency for reversible acetalization reaction steps
a
Turnover frequency
Sample
HMF ↔ HMFDEA
EMFDEA ↔ EMF
Silicalite-1
B-MFI
Fe-MFI
Al-MFI
0.46
0.48
0.19
0.24
2.08
2.10
1.00
0.91
a
2 4
k /k2r (HMF ↔ HMFDEA) or k4r/k (EMFDEA ↔ EMF) divided by
the amount of strong Brønsted acid sites (μmol) determined by FT-IR
pyridine measurements.
This is evidenced in Fig. 11 which shows that the rate con-
stants of this step k linearly depend on the B/L ratio. A rele-
3
Fig. 11 The effect of strong Brønsted acid site-to-strong Lewis acid
vant result emerging from the analysis of the rate constants
of the various steps (Table 4) is that the HMFDEA ether-
ification (reaction step 3) showed a rate constant about one
order of magnitude higher than the direct HMF etherification
3
site ratio on the k rate coefficient.
both acetalization and etherification reactions.
A high
(
reaction step 1).
This could be explained with the different types of coordi-
Brønsted acid sites concentration is required in order to
achieve a fast transformation of acetals into EMF. According
to these observations, Al-MFI can be considered as the prefer-
able catalyst among the investigated zeolites, showing the
highest HMF conversion, the highest EMF productivity, and
the lowest amount of intermediates after 5 h of reaction. The
superiority of Al-MFI can be related to the presence of highly
active Lewis sites that promote the HMF to EMF reaction and
the large amount of Brønsted acid sites that favour the trans-
formation of intermediates into EMF.
The analysis of the turnover frequency provides some fur-
ther insights. The values of turnover frequency calculated for
HMF depletion indicate a simple correlation with the total
amount of strong Brønsted and Lewis acid sites without a
specific effect of the heteroatoms in changing the strength of
the acid sites. Moreover, the analysis of the single reaction
steps provides some different observations, in addition to the
observation that only the strong Lewis acid sites present in
Al-MFI are able to catalyze the direct EMF synthesis from
HMF (reaction step 1).
In the direct HMF acetalization (HMF ↔ HMFDEA, reac-
tion step 2) and EMFDEA deacetalization (EMFDEA ↔ EMF,
reaction step 4) reactions, the turnover frequencies, calcu-
lated considering the strong Brønsted acid sites, for Silicalite-
1 and B-MFI samples are about twice those for Fe-MFI and
Al-MFI samples. This is likely related to the different reactiv-
ity of strong silanol groups generated by external surface de-
fects in Silicalite-1 and B-MFI, as confirmed by SEM and FT-
IR results. These sites are also responsible for the EMF-to-
EOP reaction.
nation. Lewis acid sites catalyse the etherification by non-
dissociative adsorption of the hydroxyl group, which as a re-
3
5
sult becomes a better leaving group. The CO group in
aldehyde, however, can compete with chemisorption, while
this competition is inhibited in the acetal. This result thus
indicates that, even limited from the reversible reactions of
direct HMF acetalization (reaction step 2) and deacetalization
of the EMFDEA, the formation of EMF via acetal is preferable
over the direct route (reaction step 1). However, a fine control
of the amount of water in the reaction medium should be
necessary to favour first the acetalization (reaction step 2)
and then the deacetalization (reaction step 4).
Conclusions
The analysis of the dependence of the rate constants for the vari-
ous steps in the reaction network of HMF etherification evi-
dences the different roles played by Lewis and Brønsted acid
sites and some cooperation effects. The results thus provide new
insights into the effect of catalyst properties on HMF conversion.
The strong Lewis acid sites generated from the presence
of aluminum are primarily active to catalyze the direct HMF
etherification to EMF. If the amount and the strength of
Lewis acid sites are not sufficient to promote the direct ether-
ification, the formation of EMF occurs via acetalization on
Brønsted acid sites. Brønsted acid sites promote acetalization
reactions, when H O is not present, whilst the deacetalization
reaction is favoured in the presence of water, formed from
2
Scheme 3 Formation of acetal.
Catal. Sci. Technol.
This journal is © The Royal Society of Chemistry 2018

View Article Online
Catalysis Science & Technology
Paper
Finally, the acetal etherification (HMFDEA to EMFDEA, re-
action step 3) depends on the ratio between the Brønsted
and Lewis acid sites, suggesting that their cooperation in this
reaction is different from reaction step 1 (direct ether-
ification), whose rate constants are about one order of magni-
tude lower with respect to those of the acetal etherification.
The etherification proceeds via non-dissociative adsorption of
the hydroxyl group on Lewis acid sites, making the hydroxyl
functionality a better leaving group. However, the aldehyde
group in HMF can compete with chemisorption, while this
competition is inhibited in the acetal, and as a consequence
the HMFDEA to EMFDEA reaction proceeds faster.
The Si/T ratio (T = B, Fe or Al) in the solid was measured
using an atomic absorption spectrometer GBC 932 AA.
The morphology of the crystalline phase of the samples
has been examined using a PhenomProX desktop scanning
electron microscope (SEM).
The specific surface area and the micropore volume were
estimated by BET and t-plot models, respectively, by measur-
ing nitrogen adsorption isotherms at −196 °C using an ASAP
2020 Micromeritics instrument.
NH -TPD measurements were conducted using
a
3
TPDRO1100 apparatus (ThermoFisher). A dried sample (100
mg, pellet mesh 90–150 μm) was loaded in a tubular quartz
micro-reactor and pretreated at 300 °C in helium flow for 1
hour to remove adsorbed water. The sample was then cooled
Experimental section
Materials
3
down to 150 °C and saturated with a 10% v/v NH /He mixture
−
1
with a flow rate of 20 STP mL min for 2 hours. Physically
adsorbed ammonia was removed by purging in helium at 150
The investigated samples were synthesized by using tetra-
propyl ammonium bromide as the structure directing agent
°
C for 1.5 hours until TCD baseline stabilization. Desorption
(SDA) and by adopting an alkaline synthesis gel with compo-
measurements were carried out in the temperature range of
sition reported in Table 6. Precipitated silica (Merck) was
used as the silica source for the synthesis of Silicalite-1,
B-MFI and Al-MFI whilst fumed silica was used to prepare Fe-
−1
1
2
00–700 °C (10 °C min ) using a helium carrier (flow rate of
0 STP mL min ).
−1 13
The nature of the acid sites was analyzed by the FT-IR pyr-
3 3 3 2 3 3
MFI. AlIJOH) , FeIJNO ) ·9H O and H BO were used as triva-
idine adsorption technique. FT-IR spectra were collected
lent atom sources for the synthesis of the respective T-MFI-
type samples (T = B, Fe, Al). In order to obtain T-MFI-type
materials with a similar Si/T ratio in the final solid (about
−1
using a Nicolet iS50 FT-IR spectrometer (resolution 4 cm )
by means of OMNIC software. All the samples were finely
ground in a mortar and pressed in self-supporting wafers. Be-
1
00), the gel composition was optimized on purpose. The Si/
fore IR analysis, all samples were activated under vacuum
T ratio in the gel was then different, due to the different ca-
pacity of boron, iron and aluminum to be incorporated into
the MFI framework.
−4
(
10 Torr) for 1 hour at 400 °C in order to desorb any possi-
ble physisorbed species in an IR cell allowing in situ thermal
treatments and pyridine dosage. The desorption procedure of
pyridine was monitored in a stepwise manner by evacuating
the sample at room temperature (r.t.) and 150 °C and cooling
down after each step to record the corresponding spectrum.
The quantification of Lewis and Brønsted strong acid sites
was performed by integrating the area underneath the bands
Crystallization was carried out under hydrothermal condi-
tions using a Teflon-lined stainless-steel autoclave kept in a
static oven at 170 °C for 24 hours for Silicalite-1 and Fe-MFI.
A crystallization period of 72 hours was necessary to synthe-
size B-MFI and Al-MFI. After crystallization, the solid was re-
covered by filtration, washed several times with distilled wa-
ter and dried at 110 °C overnight. A calcination treatment at
−1
−1
at 1445 and 1545 cm , using 2.2 cm μmol and 1.67 cm
−1
μmol as the integrated molar extinction coefficient (IMEC),
−
1
5
50 °C in an air flow for 8 hours (heating rate of 5 °C min )
respectively.
was carried out to remove the organic SDA molecules. The
protonic form of the catalysts was obtained by double ion-
Catalytic tests
4
exchange with NH Cl (1 M) at 80 °C, followed by calcination
under static air conditions at 550 °C for 8 hours.
HMF etherification with ethanol has been carried out in a
Parr autoclave reactor (Teflon-lined) provided with a Parr
1
3
4
848 controller, using 2.5 mmol of HMF, 100 mg of catalyst
Catalyst characterization
and 3.5 ml of ethanol at 140 °C for 1, 3 and 5 hours under
autogenous pressure. The identification of the products was
performed using a Finnigan Trace-GC-FID equipped with an
Rxi-5 ms capillary column (length 30 m, diameter 0.25 mm
4
X-Ray powder diffraction spectroscopy (APD 2000 Pro) was
used to evaluate the crystallinity and purity of the synthesized
phases.
Table 6 Synthesis conditions for MFI-type zeolites
Synthesis
Sample
Molar gel composition
T (°C)
t (h)
Silicalite-1
B-MFI
Fe-MFI
Al-MFI
0.08Na
0.14Na
0.14Na
0.12Na
2
2
2
2
O–0.08TPABr–1SiO
O–0.08TPABr–1SiO
O–0.08TPABr–1SiO
O–0.08TPABr–0.0042Al
2
–20H
–0.033H
–0.005Fe
–1SiO
2
O
170
170
170
170
24
72
24
72
2
3
BO
–0.015H
2
–20H O
3 2
–25H O
2
2
O
3
3 4 2
PO –30H O
2
O
3
2
This journal is © The Royal Society of Chemistry 2018
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
and film thickness 0.25 μm) using pure compounds for the
calibration curves. 5-Hydroxymethyl furfural (HMF),
15 J. G. Goodwin Jr, S. Natesakhawat, A. A. Nikolopoulos and
S. Y. Kim, Catal. Rev.: Sci. Eng., 2002, 44, 287.
5
4
-(ethoxymethyl)furan-2-carbaldehyde
-oxopentanoate (EOP) were purchased from Sigma Aldrich,
(EMF)
and
ethyl
16 K. S. Arias, S. I. Al-Resayes, M. J. Climent, A. Corma and S.
Iborra, ChemSusChem, 2013, 6, 123.
while acetal compounds were synthesized and purified
according to methods described by Balakrishnan et al. and
used as standards for the quantification.
The absence of mass and heat transfer limitations during
experiments have been verified with conventional tests and
estimations.
17 (a) R. J. van Putten, J. C. van der Waal, E. de Jong, C. B.
Rasrendra, H. J. Heeres and J. G. de Vries, Chem. Rev.,
2013, 113, 1499; (b) J. N. Chheda, G. W. Huber and J. A.
Dumesic, Angew. Chem., Int. Ed., 2007, 46, 7164.
18 H. Wang, Y. Wang, T. Deng, C. Chen, Y. Zhu and X. Hou,
Catal. Commun., 2015, 59, 127.
3
b
The carbon balance during catalytic tests was typically
good (over 90%), the difference to 100% being due to the for-
mation of some humins and products of condensation,
which become relevant only at longer reaction times.
Model parameters were estimated by using the fourth or-
der Runge–Kutta method (MATLAB R2012a) to solve the
above differential equation set, coupled with the Levenberg–
Marquardt algorithm for non-linear regression analysis. A
19 M. M. Antunes, S. Lima, A. Fernandes, A. L. Magalhaes, S.
Neves, C. M. Silva, M. F. Ribeiro, D. Chadwick, K.
Hellgardt, M. Pillinger and A. A. Valente, ChemCatChem,
2017, 9, 2747.
20 (a) K. Barbera, P. Lanzafame, S. Perathoner, G. Centi, M.
Migliori, A. Aloise and G. Giordano, New J. Chem., 2016, 40,
4300; (b) P. Lanzafame, K. Barbera, G. Papanikolaou, K.
Barbera, S. Perathoner, G. Centi, M. Migliori, A. Aloise and
G. Giordano, Catal. Today, 2018, 304, 97–102; (c) P. Frontera,
A. Macario, A. Aloise, P. L. Antonucci, G. Giordano and J. B.
Nagy, Catal. Today, 2013, 218–219, 18–29.
9
5% confidence interval was adopted during the analysis.
Conflicts of interest
2
1 (a) G. Li, E. A. Pidko and E. J. M. Hensen, Catal. Sci.
Technol., 2014, 4, 2241; (b) G. Li, E. A. Pidko and E. J. M.
Hensen, ACS Catal., 2016, 6, 4162.
There are no conflicts to declare.
2
2 H. Li, J. He, A. Riisager, S. Saravanamurugan, B. Song and S.
Yang, ACS Catal., 2016, 6, 7722.
Notes and references
1
2
3
H. Li, S. Yang, A. Riisager, A. Pandey, R. S. Sangwan, S.
Saravanamurugan and R. Luque, Green Chem., 2016, 18, 5701.
M. J. Climent, A. Corma and S. Iborra, Green Chem.,
23 J. Dijkmans, M. Dusselier, D. Gabriels, K. Houthoofd,
P. C. M. M. Magusin, S. Huang, Y. Pontikes, M. Trekels, A.
Vantomme, L. Giebeler, S. Oswald and B. F. Sels, ACS Catal.,
2015, 5, 928.
24 Z. Wang, L. Wang, Y. Jiang, M. Hunger and J. Huang, ACS
Catal., 2014, 4, 1144.
2
014, 16, 516.
(a) E. R. Sacia, M. Balakrishnan and A. T. Bell, J. Catal.,
014, 313, 70; (b) M. Balakrishnan, E. R. Sacia and A. T. Bell,
2
Green Chem., 2012, 14, 1626.
(a) K. Barbera, P. Lanzafame, A. Pistone, S. Millesi, G.
Malandrino, A. Gulino, S. Perathoner and G. Centi, J. Catal.,
25 (a) T. A. J. Hardenberg, L. Mertens, P. Mesman, H. C. Muller
and C. P. Nicolaides, Zeolites, 1992, 12, 685; (b) F. Feng and
K. J. Balkus Jr, Microporous Mesoporous Mater., 2004, 69, 85.
26 A. Zecchina, S. Bordiga, G. Spoto and L. Marchese, J. Phys.
Chem., 1992, 96, 4985.
27 S. Mintova and V. Valtchev, Microporous Mesoporous Mater.,
2002, 55, 171.
28 (a) M. Migliori, A. Aloise and G. Giordano, Catal. Today,
2014, 227, 138; (b) R. Srivastava, N. Iwasa, S. Fujita and M.
Arai, Chem. – Eur. J., 2008, 14, 9507.
29 G. Centi, S. Perathoner, F. Pino, R. Arrigo, G. Giordano, A.
Katovic and V. Pedulà, Catal. Today, 2005, 110, 211.
30 M. I. Zaki, M. A. Hasan, F. A. Al-Sagheer and L. Pasupulety,
Colloids Surf., A, 2001, 190, 261.
31 R. Buzzoni, S. Bordiga, G. Ricchiardi, C. Lamberti, A.
Zecchina and G. Bellussi, Langmuir, 1996, 12, 930.
32 C. A. Emeis, J. Catal., 1993, 141, 347.
33 P. Lanzafame, D. M. Temi, S. Perathoner, G. Centi, A. Macario,
A. Aloise and G. Giordano, Catal. Today, 2011, 175, 435.
34 J. D. Lewis, S. Van de Vyver, A. J. Crisci, W. R. Gunther, V. K.
Michaelis, R. G. Griffin and Y. Román-Leshkov,
ChemSusChem, 2014, 7, 2255.
4
2
015, 323, 19; (b) P. Lanzafame, K. Barbera, S. Perathoner,
G. Centi, A. Aloise, M. Migliori, A. Macario, J. B. Nagy and G.
Giordano, J. Catal., 2015, 330, 558.
5
6
7
E. Salminen, N. Kumar, P. Virtanen, M. Tenho, P. Maeki-
Arvela and J.-P. Mikkola, Top. Catal., 2013, 56, 765.
J. Luo, J. Yu, R. J. Gorte, E. Mahmoud, D. G. Vlachos and
M. A. Smith, Catal. Sci. Technol., 2014, 4, 3074.
K. S. Arias, M. J. Climent, A. Corma and S. Iborra, Energy
Environ. Sci., 2015, 8, 317.
O. Casanova, S. Iborra and A. Corma, J. Catal., 2010, 275, 236.
K. S. Arias, M. J. Climent, A. Corma and S. Iborra,
ChemSusChem, 2014, 7, 210.
8
9
1
1
0 C. J. A. Mota and B. P. Pinto, Curr. Phys. Chem., 2012, 2, 211.
1 O. D. Bozkurt, F. M. Tunc, N. Baglar, S. Celebi, I. D. Gunbas
and A. Uzun, Fuel Process. Technol., 2015, 138, 780.
1
1
1
2 G. Bozga, I. T. Apan and R. E. Bozga, Recent Pat. Catal.,
2
013, 2, 68.
3 E. Catizzone, A. Aloise, M. Migliori and G. Giordano,
Microporous Mesoporous Mater., 2017, 243, 102.
4 E. Catizzone, A. Aloise, M. Migliori and G. Giordano, Appl.
Catal., A, 2015, 502, 215.
35 M. Calatayud, A. M. Ruppert and B. M. Weckhuysen, Chem.
– Eur. J., 2009, 15, 10864.
Catal. Sci. Technol.
This journal is © The Royal Society of Chemistry 2018