Synthesis of isopropyl levulinate from furfural: Insights on a cascade production perspective

Article abstract of DOI:10.1016/j.apcata.2019.02.018

The present work explores the production of isopropyl levulinate from furfural by a two-step microwave assisted cascade process. Furfural is a versatile biomass-derived industrial feedstock with high annual production volume. Alkyl levulinates are promising bio-based molecules with several applications in many sectors, in particular, as biofuels, blended with transportation fuels including biodiesel, these compounds can significantly reduce the formation of soot in engines. Thus, in the first step, the catalytic transfer hydrogenation of furfural to furfuryl alcohol was studied employing a simply “ad hoc” synthesized magnetically recoverable Cu catalyst and 2-propanol as H-donor. Subsequently, the alcoholysis of the previously obtained liquors rich in furfuryl alcohol or of neat furfuryl alcohol solutions to isopropyl levulinate was investigated using commercial solid acid catalysts such as niobium phosphate and Amberlyst sulfonic resins (A15 and A70). The cascade process resulted feasible leading to good furfuryl alcohol yields in the transfer hydrogenation process with Cu-Fe₃O₄ magnetic catalyst using much lower Cu to furfural molar ratios than commonly reported. The subsequent alcoholysis step of furfuryl alcohol-rich liquors was highly efficient with A70 resin even in presence of unreacted furfural from the first step.

Full text of DOI:10.1016/j.apcata.2019.02.018

Accepted Manuscript
Title: Synthesis of isopropyl levulinate from furfural: Insights
on a cascade production perspective
´
Authors: Hilda Gomez Bernal, Patricia Benito, Enrique
´
´
Rodrıguez-Castellon, Anna Maria Raspolli-Galletti, Tiziana
Funaioli
PII:
S0926-860X(19)30074-2
DOI:
Reference:
https://doi.org/10.1016/j.apcata.2019.02.018
APCATA 16988
To appear in:
Applied Catalysis A: General
Received date:
Revised date:
Accepted date:
15 December 2018
5 February 2019
10 February 2019
´
´
´
Please cite this article as: Gomez Bernal H, Benito P, Rodrıguez-Castellon E,
Raspolli-Galletti AM, Funaioli T, Synthesis of isopropyl levulinate from furfural:
Insights on a cascade production perspective, Applied Catalysis A, General (2019),
https://doi.org/10.1016/j.apcata.2019.02.018
This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.

Synthesis of isopropyl levulinate from furfural: insights on a cascade production
perspective
Hilda Gómez Bernala, Patricia Benitob, Enrique Rodríguez-Castellónc, Anna Maria Raspolli-
Gallettia*, Tiziana Funaiolia
aDipartimento di Chimica e Chimica Industriale, Università di Pisa, Via Giuseppe Moruzzi, 13,
56124 Pisa, Italy
bDipartimento di Chimica Industriale “Toso Montanari”, Università di Bologna, Viale Risorgimento
4, 40136, Bologna, Italy
cDepartamento de Química Inorgánica, Facultad de Ciencias, Universidad de Málaga, 29071
Málaga, Spain
*Corresponding author, e-mail: anna.maria.raspolli.galletti@unipi.it
Graphical abstract
Highlights
 A promising two step cascade catalytic process for furfural upgrading to isopropyl levulinate is
studied.
 Simply “ad hoc” synthesized Cu-Fe3O4 magnetic material selectively catalyzes furfural to
furfuryl alcohol transfer hydrogenation.
 Raw furfuryl alcohol liquors are readily transformed to isopropyl levulinate with commercial
sulfonic resin Amberlyst 70 with very high yields.

 Additionally, commercial Niobium phosphate was tested in the alcoholysis of furfuryl alcohol
resulting in high yields of isopropyl furfuryl ether.
Abstract
The present work explores the production of isopropyl levulinate from furfural by a two-step
microwave assisted cascade process. Furfural is a versatile biomass-derived industrial feedstock
with high annual production volume. Alkyl levulinates are promising bio-based molecules with
several applications in many sectors, in particular, as biofuels, blended with transportation fuels
including biodiesel, these compounds can significantly reduce the formation of soot in engines.
Thus, in the first step, the catalytic transfer hydrogenation of furfural to furfuryl alcohol was studied
employing a simply “ad hoc” synthesized magnetically recoverable Cu catalyst and 2-propanol as
H-donor. Subsequently, the alcoholysis of the previously obtained liquors rich in furfuryl alcohol or
of neat furfuryl alcohol solutions to isopropyl levulinate was investigated using commercial solid
acid catalysts such as niobium phosphate and Amberlyst sulfonic resins (A15 and A70).
The cascade process resulted feasible leading to good furfuryl alcohol yields in the transfer
hydrogenation process with Cu-Fe3O4 magnetic catalyst using much lower Cu to furfural molar
ratios than commonly reported. The subsequent alcoholysis step of furfuryl alcohol-rich liquors was
highly efficient with A70 resin even in presence of unreacted furfural from the first step.
Keywords: Biomass, furfural, furfuryl alcohol, isopropyl levulinate, catalytic transfer
hydrogenation, alcoholysis.
1. Introduction
Nowadays the search for greener processes and sustainability has become a must. In line with this
premise, biomass derived platform molecules, particularly if recovered from waste, represent a
viable route for the transition from a fossil-based era into a green circular economy. Furan
compounds, for instance, can be subsequently upgraded to a great variety of value-added chemicals.
Furfural (F), with an annual production of more than 200 000 tons per year [1], is the feedstock
from which commercial furan derivatives stem from, and along with furfuryl alcohol (FA)
constitute remarkable specialty and reactive solvents which confer interesting properties to resins
such as corrosion resistance, stability at high temperature, excellent physical strength, etc. [2].
FA production utilizes 65% of the overall F produced [1]. Besides its extensive use in the
production of resins, it is an important building block for the synthesis of several value added

compounds as tetrahydrofurfuryl alcohol (THFA), levulinic acid (LA), fragrances and
pharmaceuticals as ranitidine and vitamin C. Recently, FA has been used as precursor for the
production of alkyl levulinates which are promising bio based molecules with several applications
as intermediates to produce coatings, resins, anti-cancer agents, pesticides, insecticides, and in the
fine chemical industry as flavouring and fragrances [3-6]. These compounds are regarded as green
solvents due to their low vapor pressures, which allows their use in metallic surface degreasing
processes instead of chlorinated solvents. Of particular interest are their properties as fuel additives,
since alkyl levulinates can raise the oxygen content in gasoline to burn more completely and reduce
the release of harmful gases and the formation of soot in engines.
First reports on alkyl levulinate production involved LA esterification with an alcohol in presence
of acid catalysts such as HCl [7]. Revisited literature favors heterogeneous catalysts not only due to
their simple separation from the reaction mixture but also to the lack of corrosion and disposal
issues of solid acids, which are inherent problems to the use of mineral acids. Recently, FA has
revealed itself as a better precursor for alkyl levulinates due to its easy accessibility from F, lower
cost, high alcoholysis reactivity and higher carbon economy [8,9]. The conversion of neat FA to
alkyl levulinates in presence of C1-C4 alcohols has been tested with several catalysts such as
sulfuric acid, metal salts, sulfonated carbons and resins, heteropolyacids, zeolites and metal oxides
[3-6, 8-11]. Although this reaction leads to good yields, it is industrially more amenable to put off
purifying procedures to the final product. Thus, several one pot processes for alkyl levulinate
production are being studied starting from F or even from poly- and monosaccharides in alcoholic
medium. Such an approach necessitates of bifunctional catalytic systems in order to hydrogenate F
to FA and perform the acid catalyzed FA alcoholysis to the alkyl levulinate. Chen et al. [12]
proposed a one pot process for the production of ethyl levulinate (EL) from F catalyzed by 2% wt.
Pt nanoparticles supported on a ZrNb binary phosphate. The reaction was performed at 130°C with
5 MPa of H2 pressure and reached complete F conversion and up to 76% EL after 6 h of reaction. In
spite of the good results obtained with this system, the fact of integrating conventional reductive
technologies in presence of molecular hydrogen, although technically viable, can be considered
challenging in regard to process safety and economics. Instead, catalytic transfer hydrogenation
(CTH) using alcohols as hydrogen donors reduces production costs and hazards regarding H2
handling, transportation and storage [13] and also renders the process more sustainable, since
alcohols used could be derived from biomass avoiding petroleum derived H2. Moreover, the lack of
high H2 pressure potentially enhances selectivity in hydrogenation and/or hydrogenolysis reactions.
The unconverted alcohol could be recycled to the feed while the ketone/aldehyde could be also
separated and integrated into other biorefinery processes involving carbon chain growth by

condensation to give valuable jet fuels, as already proposed in furfural based integrated
biorefineries [1, 14, 15].
Moreover, also Chen et al. [16] studied Nb2O5-ZrO2 mixed oxide microspheres for producing
isopropyl levulinate (iPrL) from furfural with 2-propanol as hydrogen donor and solvent. ZrO2 has
higher activity in transfer hydrogenation and Nb2O5 provides relatively strong surface acidity. This
process setup leads to 93% F conversion with 72% selectivity to iPrL at 180°C, but presents some
problems concerning recyclability of the catalyst.
Also, Zr/Al-containing silicates have been tested for F conversion to butyl levulinate displaying
only up to 18% yield to the ester product after 24 h of reaction at 120°C, due to simultaneous
promotion of side reactions. It is to say that not even changing the substrate to FA brought about
any improvement in levulinate yield [4].
In the field of transition metals-catalyzed processes, magnetically recoverable (Pd, Ir and Fe)-
exchanged beta zeolites have been prepared by Jorge et al. [17] following a quite complex
procedure and tested in the one pot F conversion to isopropyl levulinate. Pd-exchanged catalysts,
which afford simple separation from the reaction medium, showed almost full conversion to the
levulinate product at 130°C and 24 h of reaction with conventional heating. The microwave assisted
reaction displayed equal results after one h of reaction time, demonstrating the efficiency of
microwave heating. However, also this catalyst presented issues on recyclability, showing a net
decline in F conversion after the first recycle and a clear decrease in levulinate selectivity when
performing the sole FA alcoholysis.
Hence, great advances have been performed on F and FA conversion to alkyl levulinates but still
several drawbacks have to be overcome such as the use of highly diluted feedstocks, high catalyst
loadings, recyclability, etc. Different process approaches can also be evaluated that could better
meet industrial requirements, such as a cascade process for F valorization to alkyl levulinates has
not been evaluated before. In this study, F transfer hydrogenation is attempted using a magnetically
recoverable Cu-Fe3O4 catalyst and 2-propanol as solvent and H-donor. Subsequently, FA
alcoholysis is studied with commercial solid acids such as niobium phosphate and Amberlyst
sulfonic resins to produce isopropyl levulinate (iPrL). The best catalyst was further employed to
convert the obtained CTH liquors to iPrL. The whole cascade process approach is depicted in
Scheme 1.
Scheme 1 near here
Both steps were performed under microwave (MW) assisted heating, which can afford energy
efficiency and time savings [18]. This last aspect is particularly relevant for laboratory-scale

studies, allowing to perform a wide catalytic screening in a sustainable time. Although mostly
adopted on laboratory-scale, very recently the potential for scale-up of MW reactors has been
considered for the catalytic valorization of biobased chemicals [19].
2. Experimental
2.1. Materials
Furfural (F, Sigma-Aldrich, 99%), furfuryl alcohol (FA, Aldrich, 98%), dodecane (Sigma-Aldrich,
99%), 2-propanol (Merck, 99.8% puriss p.a.), ethanol absolute anhydrous (Carlo Erba), iron(II)
acetate (Fe(C2H3O2)2, Aldrich, 95%) and Amberlyst resins (A15 and A70, Rohm and Haas) were
used as received. Niobium phosphate (NbP) was kindly provided by CBMM (Companhia Brasileira
de Metalurgia e Mineracão) and treated at 255°C for 6 h under high vacuum (5 Pa) before reaction.
The synthesis of the copper carbamate precursor (Cu2(O2CNEt2)4·2NHEt2) and the magnetic
materials (Cu-Fe3O4 and Fe3O4) are reported elsewhere [20, 21]. In particular, Cu-Fe3O4
nanocatalyst was prepared by a green hydrothermal method consisting in the contemporaneous
decomposition of iron(II) acetate and copper carbamate in 96% aqueous ethanol through MW-
irradiation at 120°C for 15 minutes. At the end of this treatment, the particles were magnetically
collected, washed and dried [21]. Prior to catalytic tests the Cu-Fe3O4 catalyst was reduced ex-situ
in 2-propanol (13mg catalyst/ml) under 50 atm H2 at 180°C for 5 h.
2.2. Catalyst characterization
X-ray diffraction (XRD) analysis was carried out using a PANalytical X’Pert diffractometer
equipped with a copper anode (λmean = 0.15418 nm) and a fast X’Celerator detector. Wide-angle
diffractograms were collected over 2θ range from 5 to 70° with a step size of 0.05° and counting
time 15 s.
Micro-Raman measurements were performed in a Renishaw Raman Invia configured with a Leica
DMLM microscope (obj. 5×, 20×, 50×). The available sources are an Ar+ laser (514.5 nm, Pmax =30
mW) and a diode-laser (780.0 nm, Pmax =300 mW). The system was equipped with edge filters to
cut Rayleigh scattering, monochromators (1800 lines/mm for Ar+ laser, and 1200 lines/mm for
diode laser) and a Charge-Coupled Device (CCD) thermoelectrically cooled (203 K) detector.
Measurements were performed with the Ar+ Laser (514.5 nm) at power level Pout= 3 mW (10%
power). Each spectrum was recorded by four accumulations (30 s for each).

X-ray photoelectron spectroscopy (XPS) studies were carried on a Physical Electronics PHI 5700
spectrometer with non-monochromatic MgK radiation (300 W, 15 kV, and 1253.6 eV) with a
multi-channel detector. Spectra were recorded in the constant pass energy mode at 29.35 eV, using
a 720 m diameter analysis area. Charge referencing was measured against adventitious carbon (C
1s at 284.8 eV). The spectrometer energy scale was calibrated using Cu 2p3/2, Ag 3d5/2, and Au 4f7/2
photoelectron lines at 932.7, 368.2 and 84.0 eV, respectively. A PHI ACCESS ESCAV6.0 F
software package was used for acquisition and data analysis. A Shirley-type background was
subtracted from the signals. Recorded spectra were always fitted using Gaussian–Lorentzian curves
in order to determine the binding energy of the different element core levels more accurately.
High resolution transmission electron microscopy (HRTEM) characterization was carried out by a
TEM/STEM FEI TECNAI F20 microscope, equipped with an EDS analyzer. Powder catalysts were
suspended in ethanol under ultrasounds for 20 min.The suspension was subsequently deposited on
o
multi foil carbon film Au grid and dried at 100 C before doing the measurement. Particle size
distribution was processed considering around 150 particles in three different zones for each
sample.
Carbon and hydrogen analyses were performed on a Vario MICRO cube instrument (Elementar).
The metal content was determined by inductively coupled plasma-optical emission spectroscopy
(ICP-OES) with a Spectro-Genesis instrument.
2.3. Catalytic tests
The catalytic tests performed in this study were of two different types, namely CTH of F to yield
FA and the subsequent alcoholysis of FA to yield iPrL. Both reaction setups were carried out under
MW assisted heating, using a commercially available mono-mode MW unit (CEM Discover SP
system) comprised of a calibrated external infrared sensor.
In a typical liquid phase CTH procedure, a 10 ml glass reactor equipped with a magnetic Teflon stir
bar was loaded with F (0.25 g), 5ml of a 0.047M solution of dodecane, as internal standard, in 2-
propanol and the desired amount of the copper catalyst. This procedure was carried out under an Ar
atmosphere using standard Schlenk techniques and deaerated reagents and solvents. The reactor was
capped inside the Schlenk and transferred to the MW unit to be heated to the desired temperature
(160-190°C) under maximum stirring at varying reaction times from 1 to four h, at due time the
reactor was rapidly cooled with an air jet.
Also, a comparative CTH test was performed in a conventionally heated and mechanically stirred
Parr 4560 autoclave (300 ml) equipped with a P.I.D. controller 4843 (comprising a thermocouple

inside the autoclave). This reaction was performed loading F (2 g) to 40 mL of a 0.047M solution
of dodecane, in 2-propanol and the desired amount of the copper catalyst. The reactor was closed,
purged three times with nitrogen and pressurized to 30 bar N2. After reaction time was finished, the
reactor was rapidly cooled to room temperature.
To perform CTH recyclability tests, the spent catalyst was separated from the reaction solution with
the aid of a magnet. The recovered Cu-Fe3O4 catalyst was washed with deaerated 2-propanol under
stirring and vacuum dried before adding the fresh reaction solution.
After the catalytic recycles, the catalyst was washed thoroughly with deareated acetone to remove
organic matter deposited and vacuum dried before adding the fresh reaction solution.
FA alcoholysis in presence of 2-propanol was carried out in a 10 ml glass reactor using either neat
FA or FA-rich CTH liquors. In this case, 1.5 ml of a 0.047M solution of dodecane (internal
standard) in 2-propanol containing neat FA (50 mg) or the FA-rich CTH liquor and the desired
amount of the acid catalyst were fed into the reactor comprising a magnetic Teflon stir bar. The
reaction mixture was MW heated to the desired temperature (120-150°C) under maximum stirring
at varying reaction times from 15 to 120 min and then cooled rapidly with an air jet.
A comparative FA alcoholysis test was carried out in a 10 ml sealed glass vial comprising a
magnetic stirring bar, heated in a conventional oil bath at 120°C for five hours of reaction time,
using the same amounts of reagents stated for the MW assisted reaction.
Alcoholysis recyclability tests were performed after washing twice the used resin with 2-propanol
and then a 1:1 solution of 2-propanol/acetone under mixing for 30 minutes.
The reaction products were identified on an Agilent 7890B/5977A GC-MS with a HP-5MS column
(Agilent Technologies, Palo Alto, CA, USA). Reagents and commercial products concentrations
were analyzed using a DANI GC1000 gas chromatograph equipped with a flame ionization detector
and Zebron capillary column (Methylpolysiloxane, 30 m × 0.25 mm × 0.25 µm film thickness).
Calibration curves of F and FA were performed using dodecane as internal standard for quantitative
analysis. Non commercial products (iPrL and isopropyl furfuryl ether) were quantified obtaining the
response factors from combined GC and NMR analyses. NMR spectra of the solutions were
recorded at 298 K on a Bruker Avance II DRX400 instrument equipped with a BBFO broadband
probe.
Heavy degradation products were not quantified. 2-furaldehyde diisopropyl acetal was identified by
GC-MS but not quantified by GC due to its low amount.
Reagents conversions, yields and selectivity to products were defined as follows:

Conversions: XReagent (mol%) =MMoolleessooffrienaitgieanl treraegaectnetd × 100
Yields: YProduct (mol%)=MMoolelessooffpirnoidtiualctrefoargmenetd × 100
Selectivity: SProduct (mol%)=Moles of reagent reacted
Moles of product formed × 100
3. Results and discussion
3.1. Liquid phase Catalytic Transfer Hydrogenation of furfural in presence of Cu-Fe3O4 catalyst
MW assisted F CTH to yield FA was carried out using 2-propanol as solvent and H-donor and the
copper-based Cu-Fe3O4 (denoted as CuOx[ca]/Fe3O4 in [21]) was tested as magnetically
recoverable catalyst, at temperatures up to 190°C. This catalyst was prepared by the MW assisted
hydrothermal co-decomposition of Cu(II) carbamate and Fe(II) acetate precursors in alcoholic
solvent. As previously reported, the presence of nanomagnetite was confirmed by XRD, Raman and
IR data. Typical reflections of CuO were not detected by XRD presumably due to good Cu
dispersion and low loading, however the presence of 3.9 wt.% Cu was confirmed by ICP. XPS
spectra indicated that the copper oxide on the catalyst surface was mostly Cu(II) along with a small
amount of Cu(I).
Preliminary tests (not shown) with the as obtained catalysts at temperatures below 160°C resulted in
minor catalyst activity, for instance at 150°C less than 10 mol% F conversion and practically no FA
yield was obtained after 1 h of reaction. Table 1 displays the results obtained from 160°C on. At
160°C and 1 h of reaction, the as obtained catalyst reaches 10% F conversion and a yield of 4% FA
is obtained. Further increasing temperature to 185°C and reaction time to 2 h, F conversion only
reaches 14% and 10% FA yield is obtained.
Table 1 near here
Cu based catalysts have often been proposed in a reduced state for the F transfer hydrogenation in
alcohol. Villaverde et al. [22] states that metal copper is necessary for this reaction to occur at
110°C with Cu-Mg-Al catalysts and Scholtz et al. [23] reaches up to 28% FA yields with a Cu-
Fe2O3 catalyst at 180°C and 7.5 h of reaction time, stating that this catalyst is partially reduced in
situ. Thus, the catalyst was reduced in autoclave with 2-propanol under 50 atm of H2 at 180°C for 5

h. After the pretreatment, the characterization of the catalyst confirms the reduction of Cu. In the
diffraction pattern (Fig. 1a), the small reflection at ca. 50.4° (2θ) is related to the (002) plane of Cu0
(note that the main Cu0(111) reflection at 43.3° (2θ) is overlapped with the (400) reflection of the
magnetite spinel).While in the XPS spectrum, the low intensity Cu2p3/2 signal at 932.0 eV could be
related to Cu0 (Fig. 1c) [24, 25]. The Auger Cu LMM signal was also recorded to determine the
presence of Cu1+, but the very low intensity of the observed signal avoided to make any conclusion.
The well dispersed copper species in the fresh catalyst were transformed into large particles with a
broad size distribution (89 ± 37 nm) after the reduction, which were deposited on the surface of the
magnetite nanoparticles of 11 ± 4 nm. (Fig 2a, 2b, 2c). The cubic magnetite spinel structure is
mostly intact according to XRD and Raman (Fig. 1a and 1b), though some large Fe particles are
observed in the HRTEM images. These particles could be related to metallic iron since in the
diffraction pattern the reflections at ca. 44.7 and 65.1° (2θ) are due to the (011) and (002) planes of
Fe0. The XPS Fe 2p core level spectra only shows two peaks at binding energies 710.0 and 723.6
eV due to Fe 2p3/2 and Fe 2p1/2 in magnetite (Fig. 1d) [26, 27] and the peak at ca. 707.7 eV due to
Fe0 was not observed. The O 1s signal was deconvoluted into two contributions at 529.8 and 531.7
eV (Fig. 1e). The former at low binding energy is close to that observed for lattice oxygen (530.2
eV) of iron oxide. The latter at 531.7 eV, is assigned to hydroxyl species.
Figure 1 near here
Figure 2 near here
The reduced catalyst is noticeably more active already at 160 °C reaching 31% F conversion and
17% FA yield after 1 h of reaction. Further prolonging reaction to 2 h, F conversion increases to
38% and a clear enhancement in FA selectivity is observed leading to 27% FA yield.
Increasing reaction temperatures results in higher catalyst activity up to 185°C, which turned out to
yield 61% FA with a F conversion of 64%. The selectivity enhancement obtained by increasing
reaction temperature is related to the reversible formation of 2-furaldehyde diisopropyl acetal, such
formation is promoted at lower temperatures [28], while transfer hydrogenation is boosted at higher
temperatures. Peculiarly, a slightly higher temperature of 190°C presented lower catalyst activity
and selectivity probably due to F polymerization reactions leading to humins formation, known to
be promoted at higher temperatures. Humins deposition on the catalyst surface partially deactivate
the catalyst causing its lower activity. Then, in order to assess the contribution of the “support”,
bespoke magnetite synthesized as the copper-modified one [21] was tested at the best ascertained
reaction conditions. Bare Fe3O4 presented higher activity than the non-reduced catalyst but lower
FA selectivity, yielding only 14% FA (run 9).

A comparative CTH test was performed with conventional heating and the same reaction conditions
(run 10) showing a clear effect of microwaves mainly reducing reaction times.
Thus, the presence of reduced copper species in Cu-Fe3O4 magnetic catalyst highly increases
activity and selectivity (compare run 2 with 7) probably by activating 2-propanol through the metal
hydride route [13]. Magnetite “support”, as observed in run 9, also contributes to the catalyst
activity perhaps by coordinating F through its carbonyl group on Fe2+/Fe3+ and also by facilitating
the activation of the O-H bond in 2-propanol due to the oxophilic nature of Fe [23]. It is remarkable
that the activity of the as obtained Cu-Fe3O4 is significantly lower respect to that of the reduced Cu-
Fe3O4 system (compare run 2 and 7) so the presence of Cu0 has a marked effect on the enhancement
of the selectivity to FA, confirming, as yet observed also for the runs 3 to 8, that higher conversions
determine also an increase of the selectivity, that reached 96% at the maximum conversion up here
obtained of 64%. It is strongly believed that intermediate formation of 2-furaldehyde diisopropyl
acetal occurs at shorter reaction times which is successively hydrogenated to FA. Such acetal
formation was confirmed by GC-MS.
As it was mentioned above, temperature has a clear effect increasing the reaction rate, reaching a
maximum at 185°C. The effect of reaction time was also analyzed and results are displayed in
figure 3.
Figure 3 near here
A slightly lower reaction temperature (180°C) leads to noticeable lower FA yields after 2 h of
reaction and, in this case, longer reaction times result in higher FA yields. Instead, at the optimum
temperature (185°C) longer reaction times do not show any clear improvement on reaction yield,
leading only to a slight increase in F conversion.
The effect of catalyst loading on the transfer hydrogenation of F was also studied. Three different
catalyst loadings were tested, which account for F to Cu molar ratios from 171 to 43, the obtained
results are displayed in figure 4, including a blank reaction for completeness. Reaction tests
presented up to now were performed with a F/Cu mol ratio of 86. The blank reaction resulted
mainly in F polymerization byproducts observed as a dark precipitate. When the amount of catalyst
was halved, the catalytic activity was proportionally reduced, obtaining 38% of F conversion and
30% FA yield. When the catalyst was doubled (F/Cu mol ratio of 43), up to 71% FA yield was
obtained. In fact, at lower F/Cu ratios the catalytic reaction consumes F more rapidly withdrawing it
from degradation processes thus enhancing the selectivity to FA. It is to highlight that when
quantitative FA yields have been reported in literature for the transfer hydrogenation of F with Cu
based catalytic systems, these have always been obtained using quite higher metal loadings namely,
F/Cu mol ratio in the range of 10-1 and more diluted F solutions (<2 wt.% F) [22, 29, 30].

Figure 4 near here
To test catalyst stability in this reaction, four reaction cycles were performed (Fig. 5). After each
test, catalyst and reaction liquors were separated with the aid of a magnet, under inert atmosphere.
The catalyst was washed with 2-propanol and vacuum dried before adding fresh substrate in 2-
propanol.
The catalyst showed superior stability during the first three reaction cycles and only at the fourth
one suffered from a 20% decrease in FA yield. Such decrease in catalyst activity can be most likely
ascribed to organic matter deposition on the catalyst surface. In fact, the catalyst is stable in terms
of structure and morphology. The diffraction pattern (Fig. 1a), Raman spectrum (Fig. 1b) and
HRTEM images (Fig. S1) did not evidence significant modifications in Fe3O4 particles (11 ± 3 nm),
while the average Cu size is 86 ± 34 nm after 4 cycles vs 89 ± 37 nm in the reduced catalyst (Fig.
2d and c). The only difference observed in the catalyst after 4 reaction cycles is a Raman band at
1580 cm-1 due to G mode of carbon (Fig. 1b), in addition the carbon content of the spent catalyst
was found to increase up to 4% (see supporting information for elemental analysis Table S1). The
Cu-Fe3O4 catalyst was then washed thoroughly with deareated acetone and vacuum dried. This led
to good restoring of the catalytic sites as observed in fig. 5.
Figure 5 near here
3.2. Alcoholysis of furfuryl alcohol to yield isopropyl levulinate (iPrL)
In the second stage of the MW assisted cascade process, the obtained CTH liquors rich in FA were
converted into iPrL in presence of solid acid catalysts.
Initially, in order to get a clearer picture of the alcoholysis reaction, neat FA in 2-propanol was used
as substrate for the sake of simplicity. The first catalyst tested was niobium phosphate (NbP), a
solid acid active also in the acid hydrolysis of raw biomass. This catalyst presents both terminal
POH and NbOH groups which behave as Brønsted acidic sites, whereas Nb5+ coordinatively
unsaturated exposed sites confer Lewis acidity of medium strength; the acid capacity of this
material is 0.33 meq H+/g [31, 32]. Table 2 presents the screening results of this catalyst with
different loadings and different reaction temperatures.
Table 2 near here
At 120°C and the lowest catalyst loading (run IL1) only 40% FA was converted reaching barely 4%
yield of iPrL and 36% yield of isopropyl furfuryl ether (iPrFE), the latter has been reported as an
intermediate for alkyl levulinate formation [33, 34, 35]. Increasing the amount of catalyst (run IL2)
and reaction time to 60 min leads to a high increase in conversion up to 94% as well as higher

yields of the ether product and 15% iPrL. The highest amount of catalyst (mmol FA/meq H+=10)
leads to full conversion of FA already at 120°C but lower amounts of products (iPrFE+iPrL) are
detected at long reaction times, meaning that significant degradation occurs. Still, the major product
obtained is iPrFE. Increasing reaction temperature to 150°C with such high catalyst loading leads to
a slight increase in the iPrL product, mostly at the expense of iPrFE (compare IL 3 with 5). Both
reaction tests at this high temperature presented the highest amounts of degradation products, these
being more pronounced at long reaction times. In any case, ether remains the major product
obtained with this catalyst. It is worth noting that also alkyl furfuryl ethers are promising bio-based
molecules with applications as flavors and fuel blends (up to 30% v/v) in diesel engines [4, 36, 37].
To the best of our knowledge, NbP appears to be one of the best catalysts to obtain iPrFE with high
yield and selectivity from FA.
It has been reported that the conversion of the intermediate alkyl-furfuryl ether to alkyl levulinate
presents quite low reaction rates [34] and that higher Bronsted acidity is needed for its conversion to
the levulinate product [8]. Commercial ion exchange resins with sulfonated polymer frameworks
present high Bronsted acid sites density; this type of resins have recently proven to be active in the
esterification of levulinic acid with butanol [38].
Hence, two different sulfonic resins were tested, namely Amberlyst 15 (A15) and, Amberlyst 70
(A70). Their catalytic activity was analyzed during 2 h of reaction and the obtained kinetic profiles
are shown in figure 6. As observed in figure 6A, at 120° with 6.4 mmolFA/meq H+, A15 reaches
almost full FA conversion already after 15 minutes of reaction. At this point, iPrFE presents a
maximum in concentration which progressively decreases as reaction continues, displaying the
typical behavior of an intermediate product, although without the initial induction time. After 2 h of
reaction, iPrL yield reaches 78% and complete conversion of both FA and intermediate iPrFE is
achieved.
Figure 6 near here
Figure 6B shows that A70 reaches total FA conversion also after 15min of reaction, achieving 68%
iPrL yield and less than 20% iPrFE. Longer reaction times prompt to increasing iPrL yields,
achieving up to 96% iPrL and practically no iPrFE.
A comparative reaction performed with conventional heating (not shown) leads to similar kinetic
profiles but reaction times must be extended to five hours.
A70 compared to A15, is a macroreticular styrene-divinilbenzene resin with lower cross-linking
degree as well as lower acid capacity, that is 2.65 vs. 4.7 meq H+/g, respectively. A15 instead,
contains 20% divinylbenzene, which renders it less prone to swelling in the reaction medium [39].
Such a stiff matrix with high acid density might promote internal transport resistances and increase

side reactions such as solvent etherification and FA polymerization/resinification, the latter
observed by the darkening of the solution. It is to note that the FA concentration employed in this
work is slightly higher than the ones frequently used in FA alcoholysis studies [3, 8-10, 33, 34].
Higher FA concentrations promote undesirable polymerization reactions, however, using more
concentrated feedstocks is industrially desirable.
A70 contains chlorine atoms in its structure which confer high thermal stability to this resin [39], so
a higher reaction temperature was tested. The reaction performed at 150°C results in increased iPrL
yield after 15 minutes of reaction as observed in figure 6C. However, after 2h of reaction, up to
87% iPrL yield is obtained but low amounts of iPrFE remain unreacted. The higher temperature
probably enhances degradation reactions. For instance, 2-propanol dehydration to diisopropyl ether,
although not quantified in this work, is clearly observed to increase with temperature in accordance
with previous reports [9]. These side reactions probably compete with FA alcoholysis for acid sites,
leading to lower iPrL yields at long reaction times.
The mechanism of FA alcoholysis catalyzed by Amberlyst sulfonic resins according to the literature
[34, 35] is displayed in Scheme 2. As shown, different pathways are possible for the alkyl levulinate
formation but all of them are initiated from the Bronsted acid addition to FA which causes the
protonation of the hydroxyl moiety. The protonated FA can evolve completely detaching water and
adding one alcohol molecule to form A (alkyl-furfuryl ether) or adding three alcohol molecules to
form B which were both observed and identified as intermediates in the alkyl levulinate synthesis.
Scheme 2 near here
Recyclability of A70 was evaluated after washing the resin with 2-propanol and acetone. Two
recycles were performed at 120°C, showing good stability of the resin also at the third use, as
observed in Fig. 7.
Figure 7 near here
Therefore, A70 showed excellent catalytic activity and stability at 120°C, reaching up to 96% iPrL
yield after 2 h of reaction employing a substrate to catalyst ratio of 6.4mmol FA/meq H+ and a FA
concentration of 0.34M. These reaction conditions were adopted for the conversion of FA-rich CTH
liquors, which contain mainly FA and unreacted F in concentrations of 0.37M and 0.13M,
respectively. Selectivity to FA in the CTH reactions was usually over 90%, so only little amounts of
byproducts are present, among them, 2-furaldehyde diisopropyl acetal was qualitatively detected by
GC-MS.
To better understand the role of unconverted F in the FA alcoholysis of raw liquors to iPrL, a model
solution of FA + F was employed as substrate, keeping the same FA/H+ molar ratio. The results are

displayed in figure 8A. It was observed that F itself slightly lowers the yields of isopropyl
levulinate. In fact a F conversion of nearly 40% was observed in this test, indicating a clear
interaction of F and the active acid sites of the catalyst. At these reaction conditions, F conversion
only leads to degradation products and small amounts of acetal remain. It is also interesting to note
that although the total mmol of substrates (F+FA) to meq H+ was increased to 9, the catalyst activity
remains high reaching total FA conversion, yet iPrL yields decreased to 81% after 1.5 h.
Figure 8 near here
On the base of the previous results, A70 was tested in the alcoholysis of FA-rich CTH liquors. It
can be observed in figure 8B that up to 84% yield of iPrL can be obtained, a very interesting
performance considering the use of raw substrate. As above observed on the FA + F model solution,
CTH liquors lead to somewhat lower iPrL yield respect to the use of neat FA, mainly due to the
interaction of F with the acid catalyst, as confirmed by the high F conversion (up to 53%).
Moreover, small amounts of furfurylideneacetone were found among the unquantified byproducts
due to condensation of F and acetone in CTH liquors in presence of the acid catalyst.
The obtained result is quite promising since it demonstrates the possibility of reaching high
isopropyl levulinate yields from FA rich streams in the presence of simple commercial sulfonic
resins, without the intermediate expensive FA isolation. Furfural, mainly derived from the C5
sugars in the hemicellulose fraction, is a large scale biomass-platform and its upgrading to alkyl
levulinates is highly desirable, specially from an integrated biorefinery viewpoint. The pentose
fraction exploitation would increase the overall yield of an alkyl levulinate production plant from
lignocellulose [40], taking into account the high research efforts that have recently been devoted to
the one pot production of alkyl levulinates directly from hexose sugars, which requires quite well
tuned acid catalysts to perform dehydration, etherification and ring opening reactions [41, 42].
4. Conclusions
A promising two step MW-assisted cascade process for upgrading furfural to isopropyl levulinate
was studied. In the first step, consisting in the transfer hydrogenation of furfural with simply self-
synthesized Cu-Fe3O4 magnetic catalyst, up to 71% molar yield of furfuryl alcohol was obtained
using 2-propanol as solvent and hydrogen donor, working with a high furfural/Cu molar ratio and
adopting an appreciable furfural starting concentration. The need of reduced Cu species was
ascertained by catalytic reactivity tests. The magnetic catalyst showed good stability in four reaction
cycles and, although organic matter deposition occurred, simple washing with acetone restored the
catalytic sites.

The obtained FA-rich CTH liquors were successfully converted to isopropyl levulinate with
Amberlyst 70 sulfonic resins displaying molar yields up to 80%, showing the robustness of this
commercial resin although in presence of unreacted and highly labile furfural.
The most remarkable feature of the proposed cascade process is that adequate tuning of each
catalytic step avoids irreversible passivation of both catalysts used which is a common issue of the
involved reactions.
The alcoholysis of neat FA was also performed with niobium phosphate, A15 and A70. Niobium
phosphate displayed high activity in FA conversion leading to quite high and selective yields of
isopropyl furfuryl ether intermediate (up to 85% yield and 86% selectivity) while the stronger
Bronsted acidity of Amberlyst sulfonic resins resulted in high yields of the levulinate product. It
was also observed that the resin with higher acid site density and higher crosslinking degree (A15)
led to lower isopropyl levulinate yields perhaps due to transport hindrances given the more stiff
matrix.
Acknowledgements
The authors thank Prof. Guido Pampaloni for providing copper carbamate.
References
1. R. Mariscal, P. Maireles-Torres, M. Ojeda, I. Sádaba and M. López Granados, Energy
Environ. Sci., 9 (2016) 1144-1189.
2. H.E. Hoydonckx, W.M. Van Rhijn, W. Van Rhijn, D.E. De Vos and P.A. Jacobs, Furfural
and Derivatives, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, 2012.
3. S.S. Enumula, K.S. Koppadi, V.R. Babu Gurram, D.R. Burri, S.R. Rao Kamaraju,
Sustainable Energy Fuels, 1 (2017) 644–651.
4. M.M. Antunes, S. Lima, P. Neves, A.L. Magalhae, E. Fazio, F. Neri, M.T. Pereira, A.F.
Silva, C.M. Silva, S.M. Rocha, M. Pillinger, A. Urakawa, A.A. Valente, Appl. Catal. B:
Environ., 182 (2016) 485-503.
5. A. Demolis, N. Essayem, F. Rataboul, ACS Sustain. Chem. Eng., 2 (2014) 1338-1352.
6. E. Ahmad, M.I. Alam, K.K. Pant, M.A. Haider, Green Chem., 18 (2016) 4804-4823.
7. P.P.T. Sah, S.Y. Ma, J. Am. Chem. Soc., 52 (1930) 4880-4883.
8. Y.B. Huang, T. Yang, M.C. Zhou, H. Pan, Y. Fu, Green Chem., 18 (2016) 1516-1523.
9. M. Paniagua, J.A. Melero, J. Iglesias, G. Morales, B. Hernández, C. López-Aguado, Appl.
Catal. A: Gen., 537 (2017) 74-82.

10. P. Neves, S. Lima, M. Pillinger, S. M. Rocha, J. Rocha, A. A. Valente, Catal. Today, 218-
219 (2013) 76-84.
11. P. Neves, P.A. Russo, A. Fernandes, M.M. Antunes, J. Farinha, M. Pillinger, M.F.Ribeiro,
J.E. Castanheiro, A.A. Valente, Appl. Catal. A: Gen., 487 (2014) 148-157.
12. B. Chen, F. Li, Z. Huang, T. Lu, Y. Yuan, G. Yuan, ChemSusChem, 7 (2014) 202-209.
13. M.J. Gilkey, B. Xu, ACS Catal., 6 (2016) 1420-1436.
14. H. Olcay, A.V. Subrahmanyam, R. Xing, J. Lajoie, J.A. Dumesic, G.W. Huber, Energy
Environ. Sci., 6 (2013) 205-216.
15. J. Yang, N. Li, S. Li, W. Wang, L. Li, A. Wang, X. Wang, Y. Cong, T. Zhang, Green
Chem., 16 (2014) 4879-4884.
16. B. Chen, F. Li, Z. Huang, G. Yuan, J. Energy Chem., 25 (2016) 888-894.
17. E.Y.C. Jorge, T.M. Lima, C.G.S. Lima, L. Marchini, W.N. Castelblanco, D.G. Rivera,
E.A. Urquieta-González, R. S. Varma, M.W. Paixão, Green Chem., 19 (2017) 3856-3868.
18. C. Antonetti, M. Melloni, D. Licursi, S. Fulignati, E. Ribechini, S. Rivas, J.C. Parajó, F.
Cavani, A.M. Raspolli Galletti, Appl. Catal. B: Environ., 206 (2017) 364-377.
19. P. Priecel, J.A. Lopez-Sanchez, ACS Sustainable Chem. Eng., 7 (2019) 3-21.
20. E. Agostinelli, D. Belli Dell’Amico, F. Calderazzo, D. Fiorani, G. Pelizzi, Gazz. Chim. Ital.,
118 (1988) 729.
21. N. Sgrolli, N. Imlyhen, J. Volkman, A.M. Raspolli-Galletti, P. Serp, Molecular Catalysis,
438 (2017) 143-151.
22. M.M. Villaverde, T. F. Garetto, A. J. Marchi, Catal. Commun., 58 (2015) 6-10.
23. D. Scholz, C. Aellig, I. Hermans, ChemSusChem, 7 (2014) 268-275.
24. J.F. Moulder, W.F. Stickle, P.E. Sobol, K.D. Bomben, Standard Spectra for Identification
and Interpretation of XPS data, Perkin Elmer, Eden Prairie, MN,1992.
25. E. Moretti, M. Lenarda, P. Riello, L. Storaro, A. Talon, R. Frattina, A. Reyes-Carmona, A.
Jiménez-López, E. Rodríguez-Castellón, Appl. Catal. B: Environ., 129 (2013) 556–565.
26. A.P. Grosvenor, B.A. Kobe, M.C. Biesinger, N.S. McIntyre, Surf. Interface Anal., 36 (2004)
1564−1574.
27. T. Yamashita, P. Hayes, Appl. Surf. Sci., 254 (2008) 2441−2449.
28. J.M. Rubio-Caballero, S. Saravanamurugan, Р. Maireles-Torres, A. Riisager, Catal. Today,
234 (2014) 233-236.
29. W. Gong, C. Chen, Y. Zhang, H. Zhou, H. Wang, H. Zhang, Y. Zhang, G. Wang, H. Zhao,
ACS Sustainable Chem. Eng., 5 (2017) 2172-2180.
30. J. Zhang, J. Chen, ACS Sustainable Chem. Eng., 5 (2017) 5982-5993.

31. H. Gómez Bernal, L. Bernazzani, A.M. Raspolli Galletti, Green Chem., 16 (2014) 3734-
3740.
32. H. Gómez Bernal, A.M. Raspolli Galletti, G. Garbarino, G. Busca, E. Finocchio, Appl.
Catal. A: Gen., 502 (2015) 388-398.
33. Z.H. Zhang, K. Dong, Z. B. Zhao, ChemSusChem, 4 (2011) 112-118.
34. G.M. González Maldonado, R.S. Assary, J. Dumesic, L.A. Curtiss, Energy Environ. Sci., 5
(2012) 8990-8997.
35. G. Wang, Z. Zhang, L. Song, Green Chem., 16 (2014) 1436-1443.
36. E. de Jong, T. Vijlbrief, R. Hijkoop, G. J. M. Gruter and J. C. van der Waal, Biomass
Bioenergy, 36 (2012) 151–159.
37. A.J.J.E. Eerhart, W.J.J. Huijgen, R.J.H. Grisel, J.C. Van der Waal, E. de Jong, A.D.S. Dias,
A.P.C. Faaij, M.K. Patel, RSC Adv., 4 (2014) 3536-3549.
38. M.A. Tejero, E. Ramírez, C. Fité, J. Tejero, F. Cunill, Appl. Catal. A: Gen., 517 (2016) 56–
66.
39. R. Bringué, E. Ramírez, M. Iborra, J. Tejero, F. Cunill, J. Catal., 304 (2013) 7-21.
40. D. Zhao, P. Prinsen, Y. Wang, W. Ouyang, F. Delbecq, C. Len, R. Luque, ACS Sustainable
Chem. Eng., 6 (2018) 6901-6909.
41. L. Jiang, L. Zhou, J. Chao, H. Zhao, T. Lu, Y. Su, X. Yang, J. Xu, Appl. Catal. B: Environ.,
220 (2018) 589-596.
42. Y.B. Huang, T. Yang, Y.T. Lin, Y.Z. Zhu, L.C. Li, H. Pan, Green Chem., 20 (2018) 1323-
1334.

Captions for Schemes and Figures
Figure 1. Characterization of reduced and used Cu-Fe3O4 catalysts: a) XRD patterns; b) Raman
spectra and c, d, e) XPS spectra for Cu 2p, Fe 2p, and O 1s, respectively.
Figure 2. Characterization of reduced Cu-Fe3O4 catalyst: a) HR-TEM image; b) STEM-HAADF
image (element composition of each respective location is shown in the Table); c) Cu particle
distribution; d) Cu particle distribution of used Cu-Fe3O4 catalyst.
Figure 3. Liquid phase transfer hydrogenation of F to FA with 3.9 wt% Cu-Fe3O4 magnetic catalyst.
Conversion and yield as a function of reaction time. Reaction conditions: F (2.6 mmol) in 2-
propanol (5 ml) under inert atmosphere, F/Cu molar ratio of 86.
Figure 4. Influence of the catalyst loading on the CTH of F in presence of 3.9 wt% Cu-Fe3O4
magnetic catalyst. Reaction conditions: F (2.6 mmol) in 2-propanol (5 ml) under inert atmosphere
for 2h of reaction at 185°C using different F to Cu molar ratios.
Figure 5. Recyclability test of the 3.9 wt% Cu- Fe3O4 magnetic catalyst. Reaction conditions: F (2.6
mmol) in 2-propanol (5 ml) with F/Cu mol ratio=43, under inert atmosphere for 2h of reaction at
185°C.
Figure 6. Kinetic profiles of FA alcoholysis at different temperatures in presence of A15 and A70:
A) 120°C, A15; B) 120°C, A70; C) 150°C, A70. Reaction conditions: FA (0.51 mmol) in 2-
propanol (1.5 ml) with a substrate to catalyst ratio of 6.4 mmolFA/meq H+.
Figure 7. Recyclability of A70 at 120°C after 1.5 h of reaction. Reaction conditions: FA (0.51
mmol) in 2-propanol (1.5 ml) with a substrate to catalyst ratio of 6.4 mmol FA/meq H+.
Figure 8. Alcoholysis of FA-rich liquors in presence of A70 at 120°C. A) Effect of F addition on
the kinetic profiles of FA alcoholysis. Reaction conditions: FA (0.51 mmol), and F (0.21 mmol) in
2-propanol (1.5 ml) with a substrate to catalyst ratio of 6.4 mmol FA/meq H+; B) Alcoholysis of
FA-rich CTH liquors. Reaction conditions: FA (0.56 mmol) and F (0.2 mmol) in 2-propanol (1.5
ml) with a substrate to catalyst ratio of 6.9 mmol FA/meq H+.

Scheme 1. Cascade process for furfural valorization to isopropyl levulinate.
Scheme 2. Proposed reaction mechanism for FA alcoholysis.

Table 1. Furfural CTH with Cu-Fe3O4 at different conditions a
Run F CTH Pre- treatment T (°C) t (h) X (mol%) YFA (mol%) SFA (mol%)
1
2
-
160
185
160
160
170
180
185
190
185
185
1
2
1
2
2
2
2
2
2
2
10
14
31
38
48
61
64
55
25
35
4
43
75
53
70
90
92
96
86
54
90
-
10
17
27
43
56
61
47
14
31
3
reduced
reduced
reduced
reduced
reduced
reduced
Fe3O4
reduced
4
5
6
7
8
9b
10c
a Reaction conditions: F (2.6 mmol) in 2-propanol (5 ml) under inert atmosphere and MW heating,
F/Cu molar ratio of 86 which corresponds to an F/catalyst weight ratio of 5. b F/catalyst weight ratio
of 5. c Conventional heating under equal reaction conditions.

Table 2. FA alcoholysis to iPrL in presence of NbP a
run mmolFA /meq H+ T (°C) t (min) X (%) YiPrFE (%) YiPrL (%)
IL 1
IL 2
IL 3
IL 4
IL 5
IL 6
50
30
10
10
10
10
120
120
120
120
150
150
15
60
15
60
15
60
40
94
36
77
85
65
63
44
4
15
13
20
20
21
99
100
100
100
a Reaction conditions: Neat FA (0.51 mmol) in 2-propanol (1.5 ml) under MW heating.