Catalytic hydrogenation of furfural to tetrahydrofurfuryl alcohol using competitive nickel catalysts supported on mesoporous clays

Article abstract of DOI:10.1016/j.apcata.2020.117903

Nickel catalysts supported on mesoporous clays with different acid properties, such as montmorillonite MK-10, Al-pillared montmorillonite, mesoporous Na-saponite and mesoporous H-saponite, were prepared, characterized and tested for the hydrogenation of furfural to tetrahydrofurfuryl alcohol (THFA). Clays were also modified introducing basicity through magnesium oxide in different amounts. Catalysts with higher acidity or low amounts of metallic centres favoured deactivation and/or selectivity to the non-desired products. Interestingly, the addition of MgO both neutralized the acidity of the montmorillonite supports and improved the hydrogenation of the furanic ring, resulting in higher selectivity to THFA. The best catalyst was the one prepared with montmorillonite MK-10 covered by 30 wt% of magnesium oxide and with 8.8 % of the Ni metal phase achieving total conversion and total selectivity to THFA. The activity of this catalyst was maintained after several reuses.

Full text of DOI:10.1016/j.apcata.2020.117903

Journal Pre-proof
Catalytic hydrogenation of furfural to tetrahydrofurfuryl alcohol using
competitive nickel catalysts supported on mesoporous clays
C. Sunyol (Investigation) (Visualization) (Writing - original draft), R.
English Owen (Investigation) (Visualization) (Writing - original draft),
M.D. Gonz a´ lez (Conceptualization) (Validation) (Writing - original
draft), P. Salagre (Conceptualization) (Resources) (Supervision), Y.
Cesteros (Conceptualization) (Resources) (Supervision) (Writing -
original draft) (Writing - review and editing)
PII:
S0926-860X(20)30496-8
DOI:
https://doi.org/10.1016/j.apcata.2020.117903
APCATA 117903
Reference:
To appear in:
Applied Catalysis A, General
Received Date:
Revised Date:
Accepted Date:
29 July 2020
8 October 2020
25 October 2020
Please cite this article as: Sunyol C, English Owen R, Gonz a´ lez MD, Salagre P, Cesteros Y,
Catalytic hydrogenation of furfural to tetrahydrofurfuryl alcohol using competitive nickel
catalysts supported on mesoporous clays, Applied Catalysis A, General (2020),
doi: https://doi.org/10.1016/j.apcata.2020.117903

This is a PDF file of an article that has undergone enhancements after acceptance, such as
the addition of a cover page and metadata, and formatting for readability, but it is not yet the
definitive version of record. This version will undergo additional copyediting, typesetting and
review before it is published in its final form, but we are providing this version to give early
visibility of the article. Please note that, during the production process, errors may be
discovered which could affect the content, and all legal disclaimers that apply to the journal
pertain.
©
2020 Published by Elsevier.

Catalytic hydrogenation of furfural to tetrahydrofurfuryl
alcohol using competitive nickel catalysts supported on
mesoporous clays
a
a
a
a
a*
C. Sunyol ; R. English Owen , M. D. González , P. Salagre , Y. Cesteros
aUniversitat Rovira i Virgili, Departament de Química Física i Inorgànica, C/ Marcel·lí
Domingo 1, 43007 Tarragona, Spain
Prof. Yolanda Cesteros
Dpt. Química Física i Inorgànica
Universitat Rovira i Virgili
C/Marcel·lí Domingo 1
4
3007 TARRAGONA, Spain
Tel: +34 977558785
Fax: +34 977559563
e-mail: yolanda.cesteros@urv.cat
*
To whom correspondence should be addressed.
Graphical abstract
1

Highlights
-
-
-
Furfural hydrogenation depended on Ni amount and the acidity provided by the support.
MgO in the support neutralized the acidity of the clay favouring THFA selectivity.
Total conversion and total selectivity to THFA was achieved with Ni(40)/MgO(30)-M.
Abstract
Nickel catalysts supported on mesoporous clays with different acid properties, such as
montmorillonite MK-10, Al-pillared montmorillonite, mesoporous Na-saponite and
mesoporous H-saponite, were prepared, characterized and tested for the hydrogenation
of furfural to tetrahydrofurfuryl alcohol (THFA). Clays were also modified introducing
basicity through magnesium oxide in different amounts. Catalysts with higher acidity or
low amounts of metallic centres favoured deactivation and/or selectivity to the non-
desired products. Interestingly, the addition of MgO both neutralized the acidity of the
montmorillonite supports and improved the hydrogenation of the furanic ring, resulting
2

in higher selectivity to THFA. The best catalyst was the one prepared with
montmorillonite MK-10 covered by 30 wt% of magnesium oxide and with 8.8 % of the
Ni metal phase achieving total conversion and total selectivity to THFA. The activity of
this catalyst was maintained after several reuses.
Keywords: mesoporous clays; nickel catalysts; THFA; hydrogenation of furfural;
acidity; MgO; mesoporous clays.
3

1
. Introduction
Production of chemical compounds and fuels from biomass sources is acquiring
importance as a more environmentally friendly alternative to non-renewable crude oil.
Lignocellulose, one of the most abundant and low-price biomass feedstocks, mainly
consists of three types of polymers: cellulose, hemicellulose, and lignin. Catalytic
dehydration of xylose, extracted from hemicellulose, led to the formation of furfural, an
interesting platform molecule, which can be later converted into high-added value
products [1]. Thus, catalytic hydrogenation of furfural results in the formation of
different products of interest, such as furfuryl alcohol (FOL), cyclopentanone (CPO) or
tetrahydrofurfuryl alcohol (THFA), among others [2]. Of all of them, THFA is the most
important hydrogenated product because of its use as environmental friendly solvent
[
2].
Classical synthesis of tetrahydrofurfuryl alcohol involves two steps, with the formation
of furfuryl alcohol as intermediate. The reaction can be carried out in liquid or gas phase
[
3]. However, side reactions can also occur, including rearrangements, resulting mainly
in the formation of cyclopentanone and cyclopentanol (CPL) in slightly acid aqueous
conditions [4] (Scheme 1).
Good results were achieved using noble metals for the direct hydrogenation of furfural
to THFA, especially with Pd, Rh and Ru [5-7], but their high cost make researchers
look for cheaper alternatives, mainly catalysts based on Ni [7] or Cu [8]. Nickel
supported on carbon nanotubes catalyst led to 84.3 % yield to THFA at 130 ºC and 40
bars of H
2
pressure after 10 h of reaction [9], a Ni/C catalyst derived from MOFs
resulted in total conversion and total selectivity to THFA at 120 ºC and 10 bars of H
2
pressure after 2 h of reaction [10], and a Ni/γ-Al
2
3
O catalyst gave conversion up to 99.8
4

%
and selectivity to THA of 99.5 % at 80 ºC and 40 bars of H
reaction [11]. More recently, bimetallic MOF derived Ni-Co alloy catalyst presented
almost total conversion and 99.1 % THFA selectivity at 80 ºC, 30 bars of H pressure
2
pressure after 2 h of
2
and 8 h of reaction [12], and Ni-Cu catalyst supported on mesoporous hectorite
prepared with Ni:Cu ratio of 1:1 showed total conversion and selectivity to THFA of 95
%
at 140 ºC and 40 bar H
Different catalytic systems were studied modifying the reaction conditions by applying
higher H pressure or much lower reaction temperatures. Nickel-cobalt catalysts
supported on SBA-15 achieved 100 % of conversion and 90.4% of selectivity to THFA
at 210 ºC, 70 bars of H pressure and 6 h of reaction [14]. Ruan et al. reported high
selectivity to furfuryl alcohol (FOL) (90.0% at 5 ºC and 92.4% at 10 ºC) at 50 bars of
pressure and 1 h of reaction using a Pd/Ni/Ni(OH) /C catalyst [15]. Nakagawa et al.
obtained 94 % of selectivity to THFA using a Pd–Ir/SiO bimetallic catalyst at very low
temperature (2 ºC) after 6 h of reaction but 80 bar of hydrogen pressure was required
16].
2
pressure after 4 h of reaction [13].
2
2
H
2
2
2
[
The presence of Brønsted and Lewis acidity can result in the production of humines, an
undesired by-product, which can affect the activity of the metallic phase. Y. Yang et al.
modified the support with basic oxides of alkaline metals (Mg, Ca, Sr, or Ba) [3]. Thus,
changing the basic conditions with a Ni/Ba-Al
2
3
O catalyst they achieved >99% of
furfural conversion and 99% of THFA selectivity. Therefore, finding a support with
appropriate acid–base properties and porosity (to improve the metal phase dispersion
and reducibility) becomes a challenge since it has not been practically studied.
Clay minerals are widespread materials on Earth. Saponite, Mx-y/n[(Mg6-yAl
y
)Td[(Si8-
)Oh.O20(OH)
].nH O, is a trioctahedral microporous clay of the smectite group
x
Al
x
4
2
5

containing an octahedral (Oh) sheet of MgO between tetrahedral (Td) sheets of SiO .
2
Isomorphous substitution of Si4+ by Al3+ generates a negative layer charge, which
n+
should be compensated by the presence of interlayer cations (M ). Mesoporous
saponites can be synthesized by adding quaternary ammonium salts or polymers during
the hydrothermal treatment, and later removing of the template by calcination resulting
in high-surface area materials [17-18].
In this work, several nickel supported on mesoporous clays catalysts, with different acid
properties, were tested for the hydrogenation of furfural to obtain tetrahydrofurfuryl
alcohol in liquid phase. Commercial acid montmorillonite clays (MK-10 and Al-
Pillared) and synthesized Na-, H-saponites, directly and modified with different
amounts of MgO, were used as supports since they have acidity that can be easily
modified, high surface areas, mesoporosity and are of low cost. Special attention will be
paid to the study the effect of adding MgO to the supports on the acidity of the final
catalysts, and therefore, on the catalytic reaction. The percentage of metal phase was
also modified changing the percentage of metal precursors and applying different
reduction temperatures. The amount of metallic centres, responsible for the obtention of
THFA, could also contribute to partially cover the acid sites responsible for undesired
products. All catalysts were widely characterized.
2
. Experimental
2
.1.Reagents and commercial materials
Furfural was distilled under reduced pressure before use. Tetrahydrofurfuryl alcohol,
montmorillonite MK-10 (M), aluminium pillared montmorillonite (PM), nickel (II)
nitrate hexahydrate, magnesium nitrate hexahydrate, aluminium (III) nitrate
6

nonahydrate, copper (II) chloride dihydrate, ammonium nitrate, sodium silicate solution,
ethylenediamine, cyclohexylamine and 1-butanol were of analytic grade and purchased
from Sigma-Aldrich. Sodium hydroxide and sodium bicarbonate were purchased from
EBRA reagents. Dodecyltrimethylammonium chloride, furfuryl alcohol were supplied
by Acros Organics. Cyclopentanol and cyclopentanone were purchased from Alfa
Aesar.
2
.2.Preparation of mesoporous saponites
Saponite with general formula Na1.2Si6.8Al1.2Mg
6
O20(OH)
4
·nH O was prepared. The
2
ratios of the elements were determined from the moles of each reagent used in the initial
slurry. The general method used to prepare these materials was that proposed by
Gebretsadik et al. [18] with several modifications. A buffer solution of CO 2
-/HCO -
3
3
was prepared at pH 13 by dissolving 3.63 g of NaOH and 6.6 g of NaHCO in 50 mL
3
distilled water. 2.7 g of the surfactant, dodecyltrimethylammonium chloride was added
and stirred mechanically until no lumps resided. After which 5.6 mL of sodium silicate
solution (SiO
a further 10 minutes. A second solution of 2.31 g of Al(NO
Mg(NO ·6H O in 5 mL distilled water was prepared. This was added dropwise to the
2
·NaOH, SiO
2
27 wt%, density 1.39 g/mL) was then added and stirred for
3
)
3
·9H O with 7.91 g of
2
3
)
2
2
stirring mixture over a period of half an hour. The final slurry was then stirred for a
further 30 minutes ensuring homogenised stirring. The gel was then transferred into a
conventional reactor and aged at 180 ºC for 72 h. The resulting solid was then filtered
and washed with doubly deionised water to neutrality and then dried overnight in an
oven at 80 ˚C. The surfactant was removed by calcination in a quartz tubular reactor in a
two-step process, in the first at 550 ºC under 2 mL/s flow of air for 12 h, and in the
7

second at 600 ºC under 2 mL/s flow of oxygen for 12 h (sample S). The delaminated
saponite should have Na cation in the interlamellar space. A saponite with Brønsted
acidity will be also prepared by exchange with a 1 M NH
4
(NO ) solution by refluxing
3
for 1 h, resulting in the support HS.
2
.3. Basic-modification of synthesized and commercial mesoporous clays
Basic modification of the supports was made through impregnation. First, an ethanolic
solution 100 mL Mg(NO 1 M was prepared to impregnate 1 g of the desired clay (M,
PM and S) with 23 mL and 4.6 mL of the Mg(NO solution to prepare catalysts with
0 wt% and 10 wt%, respectively. Then, the solvent was removed using a rotary
)
3 2
)
3 2
3
evaporator, dried at 80 ºC and calcined at 350 ºC for 5 h in order to obtain supported
magnesium oxide. The modified supports were named as MgO(10)-M, MgO(30)-M,
MgO(30)-PM and MgO(30)-S.
2
.4. Preparation of catalysts
Catalysts were prepared by impregnation. Firstly, an ethanolic solution Ni(NO
3
2
) 0.5 M
was prepared. Secondly, 6 and 25 mL of nickel (II) nitrate solution were added to 1 g of
the support in order to achieve 15 wt% or 40 wt% of nickel, respectively. Then, the
solvent was rotary-evaporated and the residing solid heated at 80 ºC overnight. The
samples were calcined at 450 ºC for 5 h in order to obtain the catalytic precursors
named as P-Ni(15)/M, P-Ni(40)/M, P-Ni(40)/PM, P-Ni(40)/S, P-Ni(40)/HS, P-
Ni(40)/MgO(10)-M,
Ni(40)/MgO(30)-S. The reduction of the catalytic precursors was carried out in a
fluidized bed tubular reactor at 1.25 mL/s of H flow at 450 ºC for 6 h resulting in
P-Ni(40)/MgO(30)-M,
P-Ni(40)/MgO(30)-PM
and
P-
2
8

catalysts Ni(15)/M, Ni(40)/M, Ni(40)/PM, Ni(40)/S, Ni(40)/HS, Ni(40)/MgO(10)-M,
Ni(40)/MgO(30)-M, Ni(40)/MgO(30)-PM and Ni(40)/MgO(30)-S. One more catalyst,
Ni(40)/M (LT), was obtained by reducing P-Ni(40)/M at 1.25 mL/s of H
for 6 h.
2
flow at 350 ºC
2
.3.Characterization techniques
X-Ray diffraction was used to determine the crystalline phases, estimate the NiO and Ni
crystallite size using the Scherrer equation [19] and quantify crystalline phases. The
experiments were carried out with a Siemens D-500 instrument by using CuKα radition
with a Ni filter and detecting values 2θ between 0 and 70º. Crystallite size was
determined by Rietveld refinement [20] and it was performed with the TOPAS v6
software [21-22]. The background was modelized with a 2nd order Chebyschev
polynomial. The instrumental contribution to the diffraction profile was calculated with
the Fundamental Parameters Approach [23]. The relative quantitative phase analysis
was obtained by refining the Rietveld scale factor for each phase and applying the
corresponding well-known equations [24]. The peak width of each phase was modelled
with the Double-Voigt Approach [25] by considering only the Lorentzian contribution
of the crystallite size effect and discarding any contribution of the microstrain to the
peak width. The averaged integral breadth was obtained from the resulting fitted Voigt
function to the whole diffractogram.
The cation-exchange capacity was determined following the Bergaya and Vayer method
2
+
[
26]. The samples were exchanged with a 1 M solution of [Cu(en) ] complex, and the
2
exchanged amount was calculated by spectroscopic UV-Visible absorption at λ=546 nm
9

using a calibration curve at different concentrations. All measurements were done with
a Shimadzu spectrometer following the Beer-Lambert law.
The surface area of the samples was calculated by nitrogen physisorption in a
Quadrasorb Surface Analyzer instrument. Water and other physically adsorbed species
were removed by pre-degassing the samples at 85 ºC for 12 h. The BET surface area
was calculated using the Brunauer-Emmett-Teller equation. The pore size distribution
was obtained using the Barrett-Joyner-Halenda (BJH) model.
Thermogravimetric analysis (TGA) of adsorbed cyclohexylamine (CHA) was carried
out using a Setaram Labsys TG to determine acidity of the materials following the
method previously applied for clays, zeolites and mesoporous materials [27-29],. This
method involves thermogravimetric analysis following adsorption of the base on the
catalysts and determines the number of acid sites, which are accesible and sufficiently
strong to interact with the base after heat treatment at 250º C. Firstly, the samples were
left in a solution of cyclohexylamine at room temperature overnight. Then, the samples
were heated at 80 ºC for 4 h in order to eliminate excess of cyclohexylamine and for a
further 2 h at 250 ºC to remove the physisorbed CHA. Finally, the samples were
analysed using the TGA from 50 ºC to 800 ºC at 10 ºC/min under N . Acidity values
2
were calculated from the weight loss observed between 250 ºC and 450 ºC.
Temperature Programmed Reduction (TPR) studies of the catalytic precursors were
performed from room temperature until 900 ºC at 5 ºC/min under Ar/H
same Setaram Labsys TG.
2
(95:5) in the
The morphologies and particle sizes of the samples were determined by transmission
electron spectroscopy (TEM) on a JEOL JSM1011 electron microscopy operating at 80
kV.
1
0

2
.4 Catalytic tests
Catalytic hydrogenation of furfural was performed in a 50 mL stainless steel autoclave
equipped with a stirrer, a pressure valve and automatic temperature control apparatus.
The reactor was connected to hydrogen flow. 30 g of aqueous solution was loaded into
the reactor (1.5 g furfural) with 600 mg of catalyst. A small aliquot of this sample was
kept aside for analysis by gas chromatography (GC). Then, the reactor was sealed and
purged three times with H
2
for air removal. Afterwards, H was fully charged into the
2
reactor until 40 bar pressure, temperature programmed to 140 ºC and 600 rpm. The
reaction lasted 4 h. The resulting solution was then left to cool and centrifuged in order
to separate the reaction products from the catalyst. Afterwards, the filtrate was analysed
by GC with an internal standard.
In order to study the reusability, the catalyst Ni(40)/MgO(30)-M was washed several
times with ethanol and centrifuged to recover the catalyst and reused thrice at the same
reaction conditions.
2
.5. Analysis of the reaction products
GC measurements were performed on Shimadzu GC-2010A series equipped with an
AOC-20i Series autoinjector and FID. The column was a Suprawax-280 (60 m x 0.25
µm x 0.50 µm). 1-Butanol was the internal standard. The quantification of products was
determined based on GC data using internal standard method, which allowed for the
determination of conversion and selectivity of the main products calculated as below.
퐶표푛푐푒푛푡푟푎푡푖표푛 표푓 푐표푛푣푒푟푡푒푑 푓푢푟푓푢푟푎푙
퐶표푛푣푒푟푠푖표푛 (%) =
푥100
퐶표푛푐푒푛푡푟푎푡푖표푛 표푓 푖푛푖푡푖푎푙 푓푢푟푓푢푟푎푙
1
1

ꢀ
ꢁꢂꢃꢄꢂꢅꢆꢇꢅꢈꢁꢂ ꢁꢉ ꢉꢊꢆꢉꢊꢆꢇꢋ ꢃꢁꢂꢌꢄꢆꢅꢄꢍ ꢅꢁ ꢅℎꢄ 푝ꢆꢁꢍꢊꢃꢅ
푆푒푙푒푐푡푖푣푖푡푦 (%) =
x 100
ꢀꢁꢂꢃꢄꢂꢅꢆꢇꢅꢈꢁꢂ ꢁꢉ ꢃꢁꢂꢌꢄꢆꢅꢄꢍ ꢉꢊꢆꢉꢊꢆꢇꢋ
3
. Results and Discussion
3
.1. Characterization of supports
The four supports used in this study were materials related to smectite-type clays: two
commercial mesoporous montmorillonites (M, PM) and two synthesized mesoporous
saponites (S and HS). XRD pattern of the commercial montmorillonite MK-10 (M)
allowed us to identify the presence of crystalline quartz and muscovite together with
kaolinite and montmorillonite clay with disordered stacking of the layers. This mixture
of components is because of the mineral nature of this sample. For commercial pillared
clay (PM), in addition to the peaks corresponding to montmorillonite, two reflections
0
01, related to non-homogeneous pillars distribution, were also observed at low angles.
Finally, synthesized mesoporous saponite samples (S and HS) showed the characteristic
reflections corresponding to the saponite clay. The low definition of the 001 reflection
in both samples can be related to disorder in the lamellae stacking as a result of
delamination [18]. The common characteristics of these materials were their lamellar
structure and mesoporosity, which was confirmed by TEM (not shown here). The
textural and acid properties of the fourth supports are shown in Table 1.
Physisorption analysis of commercial montmorillonite MK10 (M) showed the capillary
effect in the adsorption-desorption curve at high relative pressures (Isotherm type IV),
2
typical of mesoporous materials, with a surface area of 220 m /g and an unimodal
1
2

porous size distribution with a radius around 18 Å. The isotherm of the aluminium
2
pillared montmorillonite (PM) and its BET area (250 m /g) were similar to those
corresponding to M (Table 1) but the pore size distribution was in this case bimodal,
with an approximate pore radius of 20 and 30 Å, respectively. The area values obtained
for both samples can be explained because the disorder in the lamellar structure for M,
due to acid pre-treatment of the original mineral, and the presence of pillars for PM. In
the case of the synthesized mesoporous saponites, an isotherm type IV was also
observed for both samples, with an average pore size radius of around 20 Å and higher
2
BET area (447-448 m /g), as expected, due to the delamination of small lamella [17-18].
Acidity information was obtained by thermogravimetric analysis of adsorbed
cyclohexylamine (CHA-TGA). Due to its strong basicity, cyclohexylamine interacts
with acid sites of varying strength including those that are weak. In order to exclude the
weakest acid sites and therefore determine only those sites responsible for catalysis the
samples were heat-treated at 250ºC [27]. Figure 1 shows three representative
thermograms of the supports M, PM and HS. The mass loss percentage (blue line) was
used to determine the mEq CHA/g while the temperature of the minimum of the first
derivative (orange line) can be related to the acidity strength. The general tendency
observed in these thermograms was the following: first, an initial mass loss,
corresponding to the loss of water; and secondly, a loss in the range of 250-450 ºC
associated to the desorption of CHA, related to the amount of acid sites [27-29]. The
results showed an acidity for the M support of 0.42 mEq CHA/g and a temperature of
the minimum of the first derivative at 347 ºC while for PM the acidity was 0.6 mEq
CHA/g with a temperature of the minimum of the first derivative at 349 ºC (Table 1).
For mesoporous saponites, the acidity was 0.7 and 1.2 mEq CHA/g for S and HS,
1
3

respectively, with a temperature of the minimum of the first derivative at 323 ºC and
57 ºC, respectively (Table 1). Moreover, outside of the range corresponding to the
3
CHA desorption, a wide and less intense peak in the thermogram of the three samples
was observed. This could be mainly associated with structural dehydroxylation [30]
Thus, we could evaluate the strength of acidity: the further to the left that the minimum
of the first derivative, due to the acid site-CHA interactions, exists, the easier is to lose
CHA and, in consequence, the weaker the acid sites. Therefore, HS presented the
highest acidity and strength. In the case of S, mainly with Lewis acid sites, the acidity
was higher than for M and PM, but CHA desorbed at a lower temperature (323 ºC).
Besides, cation exchange capacity showed values of 0.56 for M, 0.17 for PM and 0.78
for S (Table 1). These results confirmed the higher capacity of S to exchange cations
compared to M and PM. In fact, PM should have, probably, more protons that are more
difficult to be exchanged, according to the measured value of acidity.
Regarding the supports modified with magnesium oxide, the isotherm was of type IV
for MgO(10)-M but a significant decrease of the BET surface area, resulting in a value
2
of 67 m /g, was observed when compared to M due to the partial coverage of the
surface by MgO. This also reduced the initial pore radius of M. However, for MgO(30)-
M, MgO(30)-PM and MgO(30)-HS, the isotherms were of type III, showing a N -
2
surface interaction lower than that of adsorbate-adsorbate. Thus, a determination of the
area was not possible for these samples. The acidity of the MgO-modified samples
decreased as the amount of MgO on the clay was increased, as expected, caused by the
neutralization of acid sites. Thus, the acidity value of M (0.42 mEq CHA/g) reduced to
0
.21 mEq CHA/g for MgO(10)-M, and to 0 mEq CHA/g for MgO(30)-M.
1
4

3
.2.Characterization of the catalytic precursors
X-ray diffraction patterns of the non-basic modified precursors confirmed the presence
of nickel oxide for all of them. The NiO crystallite size values are shown in Table 2. An
increase of the Ni content, from 15 to 40 wt%, increased the crystallinity of the NiO
phase, as observed in Figure 2 when comparing P-Ni(15)/M (a) and P-Ni(40)/M (b).
Thus, the characteristics peaks at 2of 31.3º and 37.9º due to the NiO crystalline
phase, were less intense and wider for P-Ni(15)/M, as confirmed by its lower NiO
crystallite size (Table 2). By changing the support for the same Ni content (40 wt%),
some differences in the crystallinity of NiO were observed. The NiO supported on
mesoporous saponites, P-Ni(40)/S and P-Ni(40)/HS, presented smaller crystallite sizes
(Table 2) than those supported on M or MP, probably due to the higher surface area of S
and HS resulting in a greater dispersion. The smallest NiO crystallite size was observed
for P-Ni(40)/HS (6.2 nm). This could be explained, in addition to its area, by the higher
acidity of HS (Table 1), which can favor the dispersion of a basic oxide such as NiO.
Interestingly, for the NiO supported on MgO modified supports, the XRD patterns did
not show separate peaks corresponding to MgO and NiO crystalline phases. In fact, the
signal appeared between the two expected peaks. Considering that NiO has a space
̅
group Fm3m, the same as MgO, and dimensions of a=b=c=4.1771 Å and 4.1980 Å,
respectively, a solid solution may have formed. This hypothesis was based on the
Vegard’s law, an empirical rule that establishes a linear correlation, at constant
temperature, between the cell parameters of the crystal of an alloy and the concentration
of the constituent elements. Therefore, a solid solution type Ni Mg1-xO was formed.
x
This occurred in all the samples containing MgO. The greater the amount of MgO, the
further the signal was shifted to lower angles (see Figure 2c). The crystallite size values
1
5

of the NiO-MgO solid solution phase were smaller than that of the NiO crystallites for
the non-basic modified catalytic precursors (Table 2). The most significant decrease was
observed when the support was PM, in which the size of the crystallite changed from
2
1.9 nm in P-Ni(40)/PM to 6.2 nm in P-Ni(40)/MgO(30)-PM. Therefore, in all cases the
solid solution favoured dispersion.
Transmission electron microscopy showed, in detail, the supported particles of NiO, as
darker zones, complementing the X-ray diffraction results. At 150k magnifications
some NiO particles in P-Ni(15)/M were crystallized in the form of octahedrons with
sizes around 10 nm in a good dispersion (Figure 3a). When the amount of NiO was
increased (P-Ni(40)/M) the dispersion was not as good and the range of particle size
was greater (10-80 nm) (Figure 3b). For the catalytic precursors containing magnesium
oxide it was difficult to differentiate the particles because of the formation of the solid
solution, showing a less intense colour of the NiO-MgO particles.
Nitrogen physisorption analysis of the catalytic precursors without basic modification
showed a decrease of the BET area as the amount of NiO was increased (Table 2). The
pore size, compared to the supports themselves, did not change significantly. This
suggests that there was no priority occupation of pores. Precursors containing MgO,
however, showed different values depending on the amount. In the samples containing
1
0 wt% of MgO, the decrease in area was associated with the occupation of the support
surface with particles of solid solution MgO-NiO. Pore size distribution and average
pore radius values exhibited some differences between samples. P-Ni(40)/M had a value
of 17 Å while in P-Ni(40)/MgO(10)/M the pore distribution was bimodal, with average
pore radius values of 19.5 and 40 Å, probably related to inter-particle space of particles
with different size. On the other hand, when the sample contained 30 wt% of MgO (P-
1
6

Ni(40)/MgO(30)/M) the surface area increased and the bimodality disappeared leaving
only a single pore size distribution of 20 Å. The conclusion was that the greater the
amount of MgO, the greater the amount of solid solution and the better the dispersion of
NiO. This resulted in the formation of smaller particles, in turn causing an increase in
the surface area whilst the pore size distribution remained unimodal.
Acidity studies of the catalytic precursors were performed by TGA of adsorbed
cychlohexylamine (CHA). The results are also shown in Table 2. The increase in the
amount of NiO and MgO resulted in a decrease of acidity, as expected, to the extent
that, for the catalytic precursors, which had a support modified with the 30 wt% of MgO
and loaded with the highest amount of nickel, the acidity was completely neutralised
(Table 2).
By comparing the TGA profiles of the samples P-Ni(15)/M, P-Ni(40)/M and P-
Ni(40)/MgO(10)-M (Figure 4), a first derivative peak was observed around 347 ºC, at
similar temperature to that of the support M itself (347 ºC, Figure 1). This means that
neutralization was not preferential on stronger acid sites. At higher temperature (> 450
ºC), outside of the range corresponding to the CHA desorption, there was a narrow and
intense peak in the thermogram of the three samples that could be mainly associated
with dehydroxylation of hydrated NiO and/or the formation of CHA complexes with the
nickel of NiO together with some structural dehydroxylation [30]. Finally, one more
derivative peak was detected at 677 ºC in the catalytic precursors P-Ni(40)/M and P-
Ni(40)/MgO(10)-M that was larger for the latter. This peak could be related to the
formation of CHA complexes with interlamellar cations incorporated during
impregnation with Ni or Mg. When the supports were PM, S and HS, acidity reduced as
a result of the presence of NiO and MgO in the samples (Table 2), although the
1
7

derivative peak corresponding to CHA desorption showed similar temperatures to those
observed for the corresponding support (Figure 5). The better dispersion of the NiO
particles in these supports decreased the acidity amount of the saponites, meaning that
the acid sites of the support were covered more efficiently. Additionally, the peak at 677
ºC, which was not observed for P-Ni(40)/HS, appeared for P-Ni(40)/S. This confirms
that this peak should be related to the interaction between interlamellar Ni cations
incorporated during impregnation and CHA since protonated saponite is more difficult
to exchange than saponite.
In order to compare the NiO reducibility for the different catalytic precursors,
temperature-programmed reduction studies were performed. The results are indicated in
Table 2. The initial reduction temperature was similar for the precursors P-Ni(15)/M, P-
Ni(40)/M and P-Ni(40)/PM with values around 338-345 ºC, but for the sample P-
Ni(15)/M the rate of reduction was slower since the minimum corresponding to the first
derivative appeared at higher temperature (480 ºC). In the group of the catalytic
precursors supported on saponites, when the cation was Na (P-Ni(40)/S) the
thermogram showed two mass losses in the range of determination (250-650 ºC) (Figure
6
). Thus, it was apparent that there was a quantity of nickel that was more easily
reduced (at 352 ºC) and a second quantity with lower reducibility (at 650 ºC). These
peaks could be related to the different degrees of interaction between NiO and the
saponite supports. A likely explanation is that the loss corresponding to the higher
temperature could be due to Ni2+ introduced by exchange in the interlaminar space
during impregnation of the support, resulting in a more difficult reduction. A similar
phenomenon to this was observed in the case of saponites that were exchanged with a
solution of nickel cations [31]. The catalytic precursors, whose supports were modified
1
8

with MgO, showed that the reduction started later and also finished later in all the cases.
This can be explained by the presence of solid solution.
3
.3.Characterization of the catalysts
Table 3 shows the XRD results obtained for all catalysts. The catalysts supported on M
and PM, which were reduced at 450 ºC, had total reduction of nickel oxide, both for 15
wt% and 40 wt% of nickel content. The Ni crystallite size values were higher for the
catalysts with the 40 wt% metal, as expected. When P-Ni(40)/M was reduced at 350 ºC
(Ni(40)/M(LT)), the reduction was not complete (83%) and the Ni crystallite size was
smaller. For the catalysts supported on saponites, which were reduced at 450 ºC,
catalyst Ni(40)/HS showed a complete reduction while for catalyst Ni(40)/S the % of
reduction was 55 %, in agreement with the lower reducibility of its precursor, as
observed by TPR (Table 2). The Ni crystallite sizes of saponite supported nickel
catalysts were smaller than those of the catalysts supported on commercial
montmorillonite (Table 3).
Figure 7 shows the XRD patterns for the catalysts Ni(40)/MgO(10)-M,
Ni(40)/MgO(30)-M and Figure 8 for the Ni(40)/MgO(30)-S. The presence of the NiO-
MgO solid solution in all these catalysts, explains the lower reducibility of their NiO, as
expected taking into account TPR results (Table 2). In the catalyst Ni(40)/MgO(30)-S
(
Figure 8) the % of Ni phase was very low (12%). In this case, both Ni2+ in the
interlamellar space of saponite and the formation of the NiO-MgO solid solution made
more difficult the NiO reduction.
4
. Catalytic activity
1
9

With respect to the catalysts studied for this reaction, the effect of the two types of
active sites present should be analysed: metal and acid sites. The metal sites are
necessary for the reduction of the aldehyde functional group in furfural to furfuryl
alcohol (the reaction intermediate) - a process that is easy to facilitate [5]. However, the
second hydrogenation of the C=C linkage in the ring is more challenging [3] and the
metal sites are in competition with the acid sites that catalyse the formation of
secondary products, such as CPO, CPL and resinification products [32]. Therefore, the
catalytic results depend on the relative amounts of the metal and acid sites and the
relative activity of these sites. The results are shown based on the conversion of furfural
and the selectivity to THFA, FOL, CPO and CPL that were calculated from the data
obtained from gas chromatography. The rest of the products until 100 % should be
resinification products, which cannot be detected by GC.
Figure 9 shows the catalytic activity results for the hydrogenation of furfural for all
catalysts. Regarding the non-basic modified catalysts prepared with commercial
supports and reduced at 450 ºC, high conversion (90-100 %) but low selectivity to
THFA (13-25 %) values were obtained. Conversion values were slightly higher for
those synthesized with montmorillonite MK-10 while the selectivity to THFA was
slightly higher for that supported on PM. Higher amounts of CPO and CPL were formed
with Ni(15)/M and Ni(40)/M whereas FOL was the main product for Ni(40)/PM. These
small differences should be related to the relative activity of the acid and metal sites
present in each catalyst. By comparing catalysts Ni(40)/M and Ni(40)/M (LT), the
second one reduced at 350 ºC, conversion was total for both but the selectivity to THFA
was lower for the catalyst reduced at lower temperature. This can be related to less
amount of reduced Ni present in catalyst Ni(40)/M (LT) (Table 3). Furthermore, Ni
2
0

could not compete enough with the acid sites, so acid catalysis had higher contribution
to catalysis than metal catalysis.
For catalysts using saponites as supports the conversion values were moderate (40-47
%) with similar low selectivity to THFA (around 15 %). Additionally, the selectivity to
FOL was much higher than the selectivity to THFA for Ni(40)/S. An increase in the
acidity of the support (Ni(40)/HS) resulted in a slight decrease of the conversion and the
formation of higher amounts of CPO (Figure 9). These results confirm again the
importance of the role of acid sites for this reaction, since this catalyst had 100 % of Ni
phase (Table 3). The acidity probably promotes catalyst deactivation by the formation
of resinification products that block the metal sites.
On the other hand, the introduction of basic sites in the M and PM supported catalysts,
notably favoured the selectivity to tetrahydrofurfuryl alcohol. This could be explained
by the lower/null amount of acid sites together with the lower amount of metallic sites
present in these catalysts compared to their homologues Ni(40)/M and Ni(40)/PM
(Tables 2 and 3). Thus, total selectivity and almost total selectivity to THFA was
achieved for the catalysts with higher magnesium content, Ni(40)/MgO(30)-M and
Ni(40)/MgO(30)-PM, respectively. Nevertheless, the presence of some acidity for the
catalyst with the lowest MgO content, Ni(40)/MgO(10)-M, reduced the selectivity to
THFA at expenses of increasing the formation of by-products like FOL and CPO
although this catalyst had higher Ni content than Ni(40)/MgO(30)-M (Table 3). This
confirms the importance of balancing acid and metal sites for this reaction. This
supports again the idea that neutralisation of acid sites is essential for a successful
selective reaction to THFA.
2
1

From our results and taking into account the results previously reported by Yang et al.
3], we believe that, in our catalysts, magnesium could coordinate with the oxygen from
[
FOL, bringing it closer into the metallic surface and improving the hydrogenation of the
furanic ring, to continue with the catalytic hydrogenation.
Conversely, the basic modified support in catalyst Ni(40)/MgO(30)-S yielded an
increase of conversion (61%) but a decrease in the selectivity to THFA until 7 % at
expenses of the formation mainly of high amounts of FOL. The low amount of metallic
sites for this catalyst could explains these results.
The best catalyst, Ni(40)/MgO(30)-M, was chosen to study its reusability. Catalyst was
reused thrice at the conditions commented in the experimental section. The results
showed good stability of the catalyst (Figure 10). The loss of catalytic activity,
including conversion and selectivity of THFA, was, on average, 5%. The test showed
some increase in the production of FOL. This suggested the contamination by residual
products on the surface of the catalyst that deactivated the catalyst especially after the
first use. In order to evaluate the nature of the products remaining in the catalytic pores
after reaction, the resulting solution obtained after washing and centrifugation of the
reused catalyst was analyzed by gas chromatography. The chromatogram exhibited the
presence of small peaks corresponding to furfural and some reaction products.
However, XRD pattern of the reused catalyst did not show significant variations with
respect to that taken for the fresh catalyst. An efficient wash of the catalyst was a crucial
factor in obtaining good values of conversion and selectivity in later tests.
2
2

4
. Conclusions
The hydrogenation of furfural is highly dependent on the amount of metallic nickel but
also on the acidity of the catalysts. Higher values of acidity result in catalyst
deactivation due to resinification products or higher selectivity to acid-catalysed
products.
Catalysts containing MgO showed lower percentage of metallic nickel than their non-
basic modified counterparts due to the formation of a NiO-MgO solid solution, which
makes difficult the NiO reduction. For the catalyst supported on basic-modified
saponite, in addition to the solid solution, the Ni2+ incorporated by exchange during
impregnation in the interlamellar space was also very difficult to reduce. This resulted
in the least amount of reduced nickel (12 %).
Catalytic results of non-basic modified catalysts showed high or moderate conversion
but low selectivity values to the product of interest, THFA. The presence of acidity in
these catalysts favoured the formation of byproducts, such as CPO, CPL or FOL.
Interestingly, the addition of MgO neutralized the acidity of the clay support. This was
essential for a successful reaction. As a result of that, total conversion and total
selectivity to THFA was obtained with the catalyst Ni(40)/MgO(30)-M, which had the
highest content of MgO and appropriate amounts of reduced nickel. This confirms the
importance of balancing acid and metal sites for this reaction. After reusing thrice, this
catalyst showed good stability with a loss of catalytic activity on an average of 5 %.
Credit author statement
Carles Sunyol: Investigation. Visualization. Writing-Original Draft; Roma English Owen:
Investigation. Visualization. Writing-Original Draft María Dolores González: Conceptualization.
Validation. Writing-Original Draft. Pilar Salagre: Conceptualization. Resources. Supervision.
Yolanda Cesteros: Conceptualization. Resources. Supervision. Writing-Original Draft. Writing-
Review & Editing.
Declaration of interests
2
3

☒
The authors declare that they have no known competing financial interests or personal
relationships that could have appeared to influence the work reported in this paper.
Acknowledgments
Authors acknowledge the finantial support of the Ministerio de Economía y
Competitividad of Spain and Feder Funds (CTQ2015-70982-C3-3-R), and the
recognition from the Generalitat de Catalunya (2017 SGR 798).
2
4

5
. References
[
(
[
1] C. M. Cai, T. Zhang, R. Kumar, C. E. Wyman, J. Chem. Technol. Biotechnol. 89
2014) 2-10. https://doi.org/10.1002/jctb.4168
2] Y. Nakagawa, H. Nakazawa, H. Watanabe, K. Tomishige, ChemCatChem 4 (2012)
1
791-1797. https://doi.org/10.1002/cctc.201200218
3] Y. Yang, J. Ma, X. Jia, Z. Du, Y. Duan, J. Xu, RSC Adv. 6 (2016) 51221-51228.
https://doi.org/10.1039/C6RA05680F
4] Y. Yang, Z. Du, Y. Huang, F. Lu, F. Wang, J. Gao, J. Xu, Green Chem. 15 (2013)
932-1940. https://doi.org/10.1039/C3GC37133F
5] Y. Nakagawa, K. Tomishige, Catal. Commun. 12 (2010) 154-156.
https://doi.org/10.1016/j.catcom.2010.09.003
6] B. M. Matsagar, C.-Y. Hsu, S. S. Chen, T. Ahamad, S. M. Alshehri, D. C. W. Tsang,
K. C.-W. Wu, Sustainable Energy Fuels. (2020) 293-301.
https://doi.org/10.1039/C9SE00681H
7] Y. Nakagawa, M. Tamura, K. Tomishige, ACS Catal. 3 (2013) 2655-2668.
https://doi.org/10.1021/cs400616p
8] J. Wu, G. Gao, J. Li, P. Sun, X. Long, F. Li, Appl. Catal. B: Environ. 203 (2017)
27-236. https://doi.org/10.1016/j.apcatb.2016.10.038
9] L. Liu, H. Lou, M. Chen, Int. J. Hydrogen Energy 41 (2016) 14721-14731.
https://doi.org/10.1016/j.ijhydene.2016.05.188
10] Y. Su, C. Chen, X. Zhu, Y. Zhang, W. Gong, H. Zhang, H. Zhao, G. Wang, Dalton
Trans. 46 (2017) 6358–6365. https://doi.org/10.1039/C7DT00628D
11] S. Sang, Y. Wang, W. Zhu, G. Xiao, Res. Chem. Intermed.
[
[
1
[
[
4
[
[
2
[
[
[
2
5

4
3 (2017) 1179-1195. https://doi.org/10.1007/s11164-016-2691-8
12] H. Wang, X. Li, X. Lan, T. Wang, ACS Catal. 8 (2018) 2121-2128.
https://doi.org/10.1021/acscatal.7b03795
[
[
(
[
13] V. Sánchez, P. Salagre, M. D. González, J. Llorca, Y. Cesteros, Mol. Catal. 490
2020) 110956. https://doi.org/10.1016/j.mcat.2020.110956.
14] J. Parikh, S. Srivastava, G. C. Jadeja, Ind. Eng. Chem. Res. 58 (2019) 16138–
1
6152. https://doi.org/10.1021/acs.iecr.9b01443
15] L. Ruan, H. Zhang, M. Zhou, L. Zhu, A. Pei, J. Wang, K. Yang, C. Zhang, S. Xiao,
B. H. Chen, Mol. Catal. 480 (2020), 110639.
https://doi.org/10.1016/j.mcat.2019.110639
16] Y. Nakagawa, K. Takada, M. Tamura, K. Tomishige, ACS
Catal. 4 (2014) 2718–2726. https://doi.org/10.1021/cs500620b.
17] F.B. Gebretsadik, P. Salagre, Y. Cesteros, Appl. Clay Sci. 87 (2014) 170–178.
https://doi.org/10.1016/j.clay.2013.10.027.
18] F.B. Gebretsadik, D. Mance, M. Baldus, P. Salagre, Y. Cesteros, Appl. Clay Sci.
14 (2015) 20–30. https://doi.org/10.1016/j.clay.2015.05.004.
19] A. R. Stokes, A. J. C. Wilson, Math. Proc. Cambridge Philos. Soc. 38 (1942) 313–
22. https://doi.org/10.1017/S0305004100021988.
20] H. M. Rietveld, J. Appl.
https://doi.org/10.1107/S0021889869006558.
[
[
[
[
1
[
3
[
Cryst.
2
(1969)
65–71.
[
[
21] Bruker AXS GmbH, Software TOPAS 6, (2017).
22] A. A. Coelho, J. Appl. Cryst.
51
(2018)
210-218.
https://doi.org/10.1107/S1600576718000183.
23] R. W. Cheary, A. A. Coelho, J. P. Cline, J. Res. Natl. Inst. Stand. Technol. 109
[
2
6

(2004) 1-25. https://doi.org/10.6028/jres.109.002.
[
24] R. J. Hill, C. J. Howard, J. Appl. Cryst. 20 (1987) 467-474.
https://doi.org/10.1107/S0021889887086199.
25] D. Balzar, in: R.L. Snyder, J. Fiala, H.J. Bunge (Eds.), Defect and Microstructure
[
Analysis by Diffraction (International Union of Crystallography Monographs on
Crystal), Oxford University Press, Oxford, 1999, pp. 94-124.
[
26] F. Bergaya, M. Vayer, Appl. Clay Sci. 12 (1997) 275-280.
https://doi.org/10.1016/S0169-1317(97)00012-4.
27] C. Breen, Clay Minerals
https://doi.org/10.1180/claymin.1991.026.4.04
28] R. Mokaya, W. Jones, S. Moreno, G. Poncelet, Catal. Lett. 49 (1997) 87-94.
https://doi.org/10.1023/A:1019084617120.
29] M. D. González, Y. Cesteros, P. Salagre, Appl. Catal. A: Gen. 450 (2013) 178-
88. https://doi.org/10.1016/j.apcata.2012.10.028.
30] M. Holtzer, A. Bobrowski, S. Zymankowska-Kumon, J. Mol. Struct. 1004 (2011)
02-108. https://doi.org/10.1016/j.molstruc.2011.07.040.
31] I. Vicente, P. Salagre, Y. Cesteros, Appl. Clay Sci. 53 (2011) 212–219.
https://doi.org/10.1016/j.clay.2010.12.025.
[
26
(1991)
487.
[
[
1
[
1
[
[
32] S. Koso, N. Ueda, Y. Shinmi, K. Okumura, T. Kizuka, K. Tomishige, J. Catal. 267
2009) 89-92. https://doi.org/10.1016/j.jcat.2009.07.010.
(
2
7

Table 1. Characterization of the supports.
Acidity
mEq
CHA/g)
TGA)*
Surface
Average
Minimum
1st derivative
of TGA (ºC)
(
C.E.C
(mEq/g)
Support
BET area pore radius
2
(
m /g)
(BJH) (Å)
(
M
PM
S
220
250
448
447
18
20 and 30
20
20
0.42
0.60
0.70
1.20
347
349
323
357
0.56
0.17
0.78
-
HS
*
calculated from cyclohexylamine desorbed between 250 and 450 ºC by TGA; C.E.C.:
cation exchange capacity.
2
8