A novel Ru-polyethersulfone (PES) catalytic membrane for highly efficient and selective hydrogenation of furfural to furfuryl alcohol

Article abstract of DOI:10.1039/c7ta10575d

A novel catalytic membrane has been synthesised, characterised and evaluated for the selective hydrogenation of furfural to furfuryl alcohol. Unlike conventional methods, involving high pressure and high H₂:feed ratios, this work proposes an innovative ruthenium based Catalytic Membrane Reactor (CMR) to overcome mass transfer limitations, resulting in low H₂ requirements, high catalytic activity and high selectivity towards furfuryl alcohol. A UV-curable hydrophilic anionic monomer acrylic acid was used as a coating material on a commercial PES membrane and subsequently Ru nanoparticles were added. The hydrogenation of furfural was carried out in a customised catalytic membrane reactor under mild conditions: 70 °C and 7 bar, exhibiting high catalytic activity towards furfuryl alcohol (selectivity >99%) with turnover frequency (TOF) as high as 48000 h^-1, 2 orders of magnitude higher than those obtained so far.

Full text of DOI:10.1039/c7ta10575d

Journal of
Materials Chemistry A
View Article Online
View Journal
COMMUNICATION
A novel Ru–polyethersulfone (PES) catalytic
membrane for highly efficient and selective
hydrogenation of furfural to furfuryl alcohol†
Cite this: DOI: 10.1039/c7ta10575d
a
a
A. Figoli,b C. Ursino,b F. Galianob and A. Sanna
Received 1st December 2017
Accepted 22nd February 2018
*
G. Bagnato,
DOI: 10.1039/c7ta10575d
rsc.li/materials-a
A novel catalytic membrane has been synthesised, characterised and agents and lubricants.7 Furfuryl alcohol can be manufactured
evaluated for the selective hydrogenation of furfural to furfuryl by the chemoselective hydrogenation of furfural, which is an
alcohol. Unlike conventional methods, involving high pressure and aldehyde with a heterocyclic structure and a common product
high H2 : feed ratios, this work proposes an innovative ruthenium of lignocellulose pyrolysis and/or cellulose hydrolysis.8–10 The
based Catalytic Membrane Reactor (CMR) to overcome mass transfer production of furfuryl alcohol from renewable sources has
limitations, resulting in low H2 requirements, high catalytic activity and recently attracted great interest.
high selectivity towards furfuryl alcohol. A UV-curable hydrophilic
Furfural is mainly derived from a variety of agricultural prod-
anionic monomer acrylic acid was used as a coating material on ucts, including corn, oats, wheat, bran and sawdust. Industrially,
a commercial PES membrane and subsequently Ru nanoparticles were the hydrogenation of furfural is used to obtain furfuryl alcohol, an
added. The hydrogenation of furfural was carried out in a customised intermediate product for the industrial production of resin.11
catalytic membrane reactor under mild conditions: 70 ꢀC and 7 bar, Fig. 1 shows the reaction pathway for the hydrogenation of
exhibiting high catalytic activity towards furfuryl alcohol (selectivity furfural. Furfuryl alcohol formation is in competition with tetra-
>99%) with turnover frequency (TOF) as high as 48 000 hꢁ1, 2 orders of hydrofurfural formation via the saturation of different functional
magnitude higher than those obtained so far.
groups, aldehyde and the heterocyclic group, respectively.
Other undesirable products derived from the hydrogenation/
dehydration of furfural are 2 methylfuran, 2 methyltetrahy-
drofuran on further hydrogenation and pentanol on consecu-
tive hydrogenation. Furthermore, decarbonylation reactions are
favoured under the same conditions of the hydrogenation
reaction, leading to the formation of furan and subsequently,
butanol.12 All these competing reactions render it difficult to
control the selectivity of the desired product.
The production of bio-fuels or chemicals from biomass deriv-
atives, such as by pyrolysis/liquefaction of bio-oils or cellulose
hydrolysis, is a promising pathway to reduce the dependence on
fossil fuels and reduce the emission of greenhouse gases. Bio-
oils are obtained by fast pyrolysis of lignocellulosic biomass,
marine biomass and bio-waste.1,2 Bio-oil is a dark brown liquid
with high viscosity and high oxygen and water contents, which
limit its use as a fuel in common engines. To decrease the
oxygen content and convert highly reactive functionalities into
more stable ones, bio-oils can be subjected to hydrogenation
reactions3–6 resulting in compounds with a higher economic
value that can be used for the production of polymers,
cosmetics, food additives or drop-in fuels. Furfuryl alcohol is an
important chemical intermediate for the production of chem-
ical products, such as vitamin C, lysine, plasticizers, dispersing
The hydrogenation of furfural has been studied in both gas
and liquid phases,13–15 in the presence of metal catalysts, in
particular the elements of IX (Co, Rh, and Ir) and X (Ni, Pt, and
Pd) groups and Cu, on supports such as SiO2, C or Al2O3.13,16–19
O'Driscol et al.20 synthesized a Pt–Sn/SiO2 catalyst to study
furfural hydrogenation at different reaction temperatures
ꢀ
between 25 and 150 C at 20 bar and with different solvents in
a batch reactor. The authors obtained a complete conversion
aer 48 h at 100 ꢀC in the presence of toluene. Chen et al.14
achieved a complete conversion aer 5 h at 100 ꢀC and 10 bar in
the presence of Pt supported on g-C3N4. Rh and Pt nanoparticles
stabilized by phosphine-functionalized silica were studied by
Llop et al. for the hydrogenation of different bio-oil
compounds.21 At 80 ꢀC and 40 bar, furfural was completely
converted into furfuryl alcohol in the presence of Pt, while the
conversion was about 16% using Rh.
aInstitute of Mechanical, Process and Energy Engineering, School of Engineering &
Physical Sciences, Heriot-Watt University, Edinburgh EH14 4AS, UK. E-mail: a.
sanna@hw.ac.uk
bInstitute on Membrane Technology (ITM-CNR), Via P. Bucci, 17/c, 1-87030 Rende, CS,
Italy
† Electronic supplementary information (ESI) available. See DOI:
10.1039/c7ta10575d
This journal is © The Royal Society of Chemistry 2018
J. Mater. Chem. A

View Article Online
Journal of Materials Chemistry A
Communication
Fig. 1 Reaction pathways.
The above studies indicate that a high pressure (>10 bar) and/ hydrogenation of nitrobenzene in the presence of the Pd/g-Al2O3
or long residence time (hours) are required for the successful catalyst. The authors noted that the membrane with the cata-
selective conversion of furfural into furfuryl alcohol, using lytically active layer showed the best performance in terms of
traditional reactors (continuous and batch). The main disad- conversion and catalytic stability, ꢂ85% for 10 hours. Instead,
vantage of hydrogenation reactions is represented by the mass the MR with the separate catalytic packed bed achieved
transport limitation, due to the reactions taking place in a conversion of about 20% and 60%, respectively, aer 4 and 6
gaseous, liquid and solid phases. The system must operate under hours. A limited number of studies have been focused on the
high pressure, which improves the gas solubility in the liquid specic catalytic hydrogenation of furfural and all of them were
system, and at high temperature, which benets the kinetics. done using batch reactors.14,16,17 The outcomes of these studies
However, the hydrogen solubility decreases under those condi- are discussed and compared to the ndings of this work in the
tions. The choice of reaction temperature and pressure is last part of this work. The aim of this work was to develop and
dictated by the conversion of the limiting reagent and the test a novel catalytic membrane reactor in order to selectively
selectivity of the desirable product. With the purpose of over- hydrogenate bio-oil derived compounds by reducing the amount
coming the mass transfer limitation, the use of a membrane of hydrogen used. To the best of our knowledge, Ru-
reactor (MR) represents a valid choice.22–24 According to the functionalised PES membranes have never been synthesized
IUPAC denition, a MR is a device for simultaneously carrying and tested for hydrogenation reactions. Recently, Mengistie
out a reaction and a membrane-based separation, in the same et al.29 used a Pd–PES membrane for the hydrogenation of the
physical enclosure.25 The membrane can have extractor, –NO2 group in nitrophenol in a ow-through catalytic membrane
distributor or active layer functionalities. For example, Bagnato reactor, denoting the stability of PES as a support for Pd.
et al.26 used a dense Pd–Ag membrane to extract H2 from the
In this work, a novel catalytic membrane with an active layer
reaction zone during the steam reforming of a model bio-ethanol was synthesised by modifying a commercial PES membrane by
in the presence of the Ni/CeO2 catalyst. They compared the photochemical gra30,31 polymerization of acrylic acid and
performance of a MR and a packed bed reactor (PBR) at 400 ꢀC, doping its surface with Ru nanoparticles. The modied
at a gas hourly space velocity (GHSV) of 800 hꢁ1, obtaining the membrane was characterized in terms of morphology, porosity,
best conversion and hydrogen yield by using the MR. Catalytic pore size diameter, contact angle and Ru content. Aer that, the
membrane reactors were recently proposed for the conversion of catalytic membrane was tested for the hydrogenation of furfural
biomass substrates into bio-chemicals. The hydrogenation of under mild process conditions for evaluating its turnover
levulinic acid was carried out using a porous expanded poly- frequency (TOF). The commercial PES membranes (supplied
tetrauoroethylene (ePTFE) membrane with Ru catalyst particles from Hangzhou Cobetter Filtration Equipment Co., China),
by Stanford et al.27 Different membranes, with and without with different pore sizes (from 50 to 800 nm), were dipped in
a dense Matrimid layer, were tested to evaluate the hydrogen ux a graing solution composed of 25 wt% acrylic acid, 2.83 wt%
through them. The hydrogenation reaction was studied in the N,N0-methylene-bisacrylamide and 0.0753 wt% 4-hydrox-
reaction temperature and pressure range of 40–90 ꢀC and 0.7–5.6 ybenzophenone in water. The PES membrane exposed to UV
bar, respectively. Furthermore, the authors compared the results light (two lamps from General Electric, UV output: 2.2 watt, gap
obtained with a PBR considering the kinetic rate of the gamma- from the samples: 6 cm), and in the presence of photo-
valerolactone product over grams of Ru. The MR without the initiators, led to the dissociation of free radicals with consecu-
Matrimid layer resulted in the best performance (rate was 4 tive polymerization of acrylic acid on the membrane surface.
times higher than that of the PBR), but with a low conversion of
The irradiation time was between 5 and 30 min, equivalent
0.0065%, while the MR with the control layer (Matrimid) showed to 4.84–14.53 J cmꢁ2 energy ux. Subsequently, the graed
a kinetic rate 2 times lower than that of the PBR. Two different membranes were washed with distilled water to remove the
MRs were studied by Liu et al.,28 who compared (i) a MR plus excess graing solution and dried. The modied PES
a catalytic packed bed, where the membrane acted as a H2 membrane was dipped in hexaammine-ruthenium(II) chloride
distributor in the reaction zone (where the catalyst was allocated) solution (0.01 M) for 18 hours at room temperature.29 The
and (ii) a catalytic MR, where the membrane was modied by solution reacted with the carboxyl group of acrylic acid on the
adding a catalytic layer. The two systems were tested for the PES membrane (Fig. 2).
J. Mater. Chem. A
This journal is © The Royal Society of Chemistry 2018

View Article Online
Communication
Journal of Materials Chemistry A
Fig. 2 Precursor loading.
A Na2BH4 0.1 M solution was added for reducing the Ru from
ions to metallic form for 3 hours (Fig. 3).
Aer that, the membrane was washed with distilled water
and dried at 80 ꢀC. Three different commercial PES membranes
(pore sizes of 50, 220 and 800 nm) were modied by different UV
light exposure, from 5 to 30 min. The modied membranes
were then characterized by SEM analysis and the Ru was
quantied. The catalytic membrane with an initial pore size of
220 nm was selected to perform furfural hydrogenation tests. In
order to evaluate the effective presence of the acrylic layer and
the Ru content on the PES membrane, different character-
isation techniques were employed (ESI†). The FTIR analysis was
used for measuring the effectiveness of UV graing, with the
acrylic acid polymerization promoted by UV in the presence of
compounds such as N,N-methylenebisacrylamide and hydrox-
ybenzophenone. Fig. 4 shows the mid IR spectrum of a modied
and unmodied PES membrane with an initial pore size of
50 nm. The presence of acrylic acid on the membrane is high-
lighted by the peaks at 1720, 1620 and about 1100 cmꢁ1. These
peaks represent the stretching of C]O, C]C and ]C–O–
bonds. The different time periods of exposure to UV light
resulted in different polymerization grades of acrylic acid on the
membrane. The presence of the acrylic layer led to the loading
of different amounts of Ru on the membrane and different
transport phenomena. The morphological changes in the
membrane and the formation of the active layer were evaluated
by SEM (see Fig. 5).
Fig. 4 FTIR spectrum of modified PES.
and a water drop. The measurement conrmed the presence of
a coating layer on the PES membrane which resulted in changes
in the contact angle, as shown in Fig. 6. For the 50 nm PES
membrane and for the bottom layer of the 200 nm PES
membrane, there was an increase of the contact angle (decrease
of hydrophilic properties). This phenomenon is linked to the
presence of Ru nanoparticles on the membrane surface and to
their relative orientation.32,33 The bubble point, pore size
distribution and overall porosity were determined, as reported
in the ESI,† before and aer the UV light exposure.
The formation of the acrylic layer resulted in a slight
decrease of the porosity from 83 to about 77% and from 85 to
about 81%, respectively, for the PES membranes with 50 and
A longer UV light exposure involved a larger absorption of
energy by the PES membrane, and the presence of photo-
initiators led to the dissociation of free radicals with consecu-
tive acrylic monomer addition. Fig. 5 shows the modied PES
membranes using different UV light exposure. A thicker acrylic
layer was obtained using an UV energy ux of 4.84 J cmꢁ2
(20 mm) rather than 2.42 J cmꢁ2 (15 mm), which is due to a major
concentration of free radicals.
The hydrophilic properties of the modied membrane were
quantied by the contact angle between the membrane surface
Fig. 5 SEM surface and cross-section analyses of the PES-220 nm
Fig. 3 Redox reaction.
modified membrane: (a) UV ¼ 2.42 J cmꢁ2 and (b) UV ¼ 4.84 J cmꢁ2
.
This journal is © The Royal Society of Chemistry 2018
J. Mater. Chem. A

View Article Online
Journal of Materials Chemistry A
Communication
aer the coating of the membrane PES-220 nm, while it was
almost unchanged for the PES-50 nm and diminished aer the
coating of the 800 nm PES membrane.
This different behaviour can be attributed to the different
morphological structures of the pristine PES membranes,
symmetric for the 50 and 220 nm ones and asymmetric for the
800 nm one.
The PES-220 nm also presented an increase of the elongation
at break compared to the other two membranes that showed
a decrease of the elongation at break aer treatment (Fig. 9).
The Young's modulus and the elongation break results have
been used to assess how the mechanical membrane structure
changed aer the coating. The high values of Young's modulus
for 50 and 220 nm membranes indicate a better resistance to
deformation from stress, therefore resulting in a more rigid
structure. The elongation at break instead indicates how the
membrane absorbs deformation without breaking. For the 50
and 800 nm membranes there was a clear ductility reduction
aer the coating.
Fig. 6 Contact angle for PES before and after the coating (UV light
7.26 J cmꢁ2).
The inductively coupled plasma optical emission spectrom-
etry (ICP-EOS) analysis was used to determine the Ru content
loaded on the coated PES membranes with different pore sizes.
Fig. 10 shows the concentration of Ru for the tested
membranes: 50, 200 and 800 nm. A constant Ru content of
about 9 mg cmꢁ2 was detected for the PES membranes with
a pore size of 800 nm at different UV energies supplied
(depending on the UV exposure time). An increase of Ru
concentration on the PES membrane surface was noted for
the PES membrane with a pore size of 220 nm, with about
15 mg cmꢁ2 for the maximum exposure time of UV light (cor-
responding to the maximum energy used).
Fig.
7 Porosity of PES before and after the coating (UV light
7.26 J cmꢁ2).
An unusual trend was instead observed for the PES-50 nm,
with an initial decrease in the amount of Ru deposited on the
membrane up to an energy ux of about 10 J cmꢁ2 and an
increase in the amount of the metal deposited at longer UV
exposure times up to 15 mg cmꢁ2. From the SEM analysis (see
Fig. S2†), it is possible to note that for energy ux lower than
9.69 J cmꢁ2 (intermediate UV exposure time), the pore sizes on
the membrane surface are relatively larger compared to those
obtained with longer UV exposure times. For the longest UV
800 nm pore size, as can be seen in Fig. 7. Instead, the porosity
did not change for the PES-220 nm. The distribution of the pore
size is reported in Table 1, indicating that the presence of the
coating layer has no signicant effect on the pore size for the
PES-50 and 220 nm, while it has the contrary effect on the PES-
800 nm. In order to study the effect of acrylic acid and Ru
doping on the mechanical properties of the PES membrane
structure, the tensile stress was analysed by calculating the
Young's modulus and the elongation at break, which resulted to
be in the range of z150–250 N mmꢁ2 and 17–32%, respectively.
Fig. 8 shows a considerable increase of the Young's modulus
exposure time, corresponding to an energy ux of 14.53 J cmꢁ2
,
the Ru content on the modied membrane was almost equal to
that obtained for an exposure of 2.42 J cmꢁ2, but the pore size
distribution was different, ve times bigger for the lower energy
ux. This difference was caused by the different exposure of UV
Table 1 Bubble point, pore size and maximum pore size distribution for PES before and after the coating (UV light 7.26 J cmꢁ2
)
Bubble point
[bar]
Mean ow pore
diameter [nm]
Diameter at maximum pore
size distribution [nm]
Membrane
PES-50 nm not coated
PES-50 nm coated
PES-220 nm not coated
PES-220 nm coated
PES-800 nm not coated
PES-800 nm coated
1.55
0.78
1.14
0.62
0.51
0.54
120
110
280
240
850
770
70
72
240
220
740
700
J. Mater. Chem. A
This journal is © The Royal Society of Chemistry 2018

View Article Online
Communication
Journal of Materials Chemistry A
oxidation during the sample preparation for the XPS analysis.35
TEM images show the internal structure of the uncoated and
coated active layers of Ru–PES membranes. As shown in Fig. 12,
in all cases the presence of Ru nanoparticles is clearly visible,
and the coating is homogeneous. In respect to the unmodied
membranes, the coated membranes present a thin darker
(electron dense) Ru-rich part of the skin and penetration of the
Ru nanoparticles into the pores of the support, according to
pore size and porosity measurements. Energy-dispersive X-ray
analysis was also performed, using the coated and uncoated
membranes. The spectra demonstrate that the coated
membranes are rich in ruthenium. The additional peaks belong
to copper used for TEM grids.
Fig. 8 Young's modulus before and after the coating (UV light 7.26 J
In order to evaluate its activity, once characterised, the PES
cmꢁ2).
220 nm modied membrane (UV light graing at 7.26 J cmꢁ2
)
light. The increased number of free radicals produced at high
energy ux inuenced the amount of acrylic deposited and thus
the nal membrane structure. The pore size reects the capacity
of the membrane to work as a contactor among the phases.
Small pore sizes increase the mass transfer resistance of the
liquid phase, with consequently reduced contact of the organics
in the liquid with the active layer. On the contrary, large pore
sizes lead to short contact in the reaction system (liquid/
catalyst/gas), and thus a low conversion. The PES-220 nm
modied membrane showed an intermediate pore size
(maximum pore size distribution of 240 nm). More specic
analyses were conducted for the Ru NPs on the PES membrane
by Transmission Electron Microscopy (TEM) and X-ray photo-
electron spectroscopy (XPS).
The surface composition of the Ru–PES membrane was
studied using an XPS technique, as shown in Fig. 11a. The
survey-scan XPS spectrum shows all the framework elements of
the Ru–PES membrane, which is in agreement with the SEM-
EDS data. The high-resolution XPS spectrum shown in
Fig. 11b shows two signicant bands at 284.2 and 280.6 eV,
which can be readily assigned to Ru(0) and Ru oxide, respec-
tively.34 The presence of Ru oxide can be attributed to Ru
Fig. 9 Elongation at break before and after the coating (UV light Fig. 10 ICP analysis of modified (a) PES-50 nm, (b) PES-220 nm and (c)
7.26 J cmꢁ2).
PES-800 nm.
This journal is © The Royal Society of Chemistry 2018
J. Mater. Chem. A

View Article Online
Journal of Materials Chemistry A
Communication
a catalyst under the studied conditions was excluded a priori
based on previous assessments. Chen et al.14 did not observe
any conversion (less than 1%) at 10 bar and 100 ꢀC without
a catalyst.
Moreover, Sanna et al.,4 who investigated the homogeneous
(without the presence of a catalyst) continuous ow hydroge-
nation of furfural at 125 ꢀC and 51 bar H2 for 36 h, dꢀid not
observe any change in the furfural concentration at 125 C.
In order to compare the hydrogenation reaction of furfural
with previous studies, the operating conditions, conversion of
furfural, product selectivity and TOF present in the literature,
are summarized in Table 2.
Since all the previous studies were carried out using batch
reactors, similar process conditions can be used to generally
compare a PBR reactor vs. a CMR. However, a precise compar-
ison of the catalyst properties is difficult due to the different
reactor set-ups and hence the use of TOF should be preferred. In
eqn (3), the TOF represents the reactant converted per catalyst
mole per unit time (typically the second), with the number
being a function of the operating conditions. Analysing the
TOFs obtained in the available studies on furfural hydrogena-
tion, the highest value of 48 000 hꢁ1 was obtained in this work,
mainly due to different factors: (i) the high hydrogenation
activity of Ru at low temperature; (ii) a good stability of Ru in the
aqueous phase and (iii) the high hydrogenation selectivity
towards the carbonyl group.16,18,19 Ru catalysts are widely used
for hydrotreating biomass-derived substrates at low tempera-
tures.34 Tan et al.35 studied the Ru activity on different supports
for the hydrogenation of levulinic acid to g-valerolactone from
ꢁ10 to 100 ꢀC, with a complete selectivity in g-valerolactone and
a conversion of levulinic acid of about 70 and 100%, respec-
tively. (iv) The enhanced contact between the H2 and the Ru
active sites due to the reduced mass transfer limitation36 in the
presence of the modied PES membrane. Under the studied
conditions, the presence of three phases is due to a series of
mass transfer limitations, where the main limitation step is
represented by the hydrogen absorption to the catalyst surface.
A method to improve the driving force of the solubility process
is increasing the pressure, very common for hydrogenation
reactions in the liquid phase, or the enhancement of the contact
surface area between the liquid and gas phase. For a PBR, the
contact surface area is about 100 m2 per m3 of the reactor, while
the contact surface area is in the order of 1500–7000 m2 per m3
of module for hollow ber membranes.37,38 PES was mainly used
as an enhanced contactor because of its chemical resistivity in
the presence of furfural under the used conditions. Since H2 in
the gas phase has low solubility in the other phase, a higher
surface area contact between these phases decreases the need
for higher pressure that could have been applied to less soluble
Fig. 11 (a) XPS spectrum of Ru–PES-220 nm and (b) the Ru 3d XPS
spectrum of Ru–PES-220 nm.
was tested for the hydrogenation of furfural. The test was
carried out in duplicate at 70 ꢀC with a pressure of 7 bar in the
gas and liquid zones, a liquid ow rate of 0.497 mL minꢁ1
,
a water feed with 5 wt% furfural and a hydrogen/furfural molar
ratio of 1.
For a complete evaluation and comparison with previous
studies, the conversion of furfural, the selectivity for the prod-
ucts (i-compound) and the turn over frequency (TOF) have been
dened as:
ꢀ
ꢀ
ꢀ
ꢀ
Furfuralin ꢁ Furfuralout
x ¼
Si ¼
ꢃ 100½¼ꢄ%
ꢃ 100½¼ꢄ%
(1)
(2)
(3)
Furfuralin
molar
ꢀ
ꢀ
ꢀ
ꢀ
i-compoundout
Furfuralin ꢁ Furfuralout
molar
mole of furfural reacted
TOF ¼
½¼ꢄhꢁ1
mole of catalyst ꢃ reaction time
Among the potential reaction products reported in Fig. 1,
only furfuryl alcohol was detected under the tested conditions
(selectivity >99%), with a furfural conversion of 26% aer 30
minutes. The conversion and selectivity remained constant for 4
hours. Fig. 13 shows the conversion of furfural for the whole
reaction test, where the conversion achieved the steady state
condition aer 30 min. Aer 4 h, the conversion of furfural
decreased to 17%. A potential contribution to the furfural
conversion from homogeneous reactions in the absence of
components. Moreover, supplying H2 through
a porous
membrane allowed the hydrogen to be adsorbed directly onto
the Ru surface, and partly solubilised in the liquid phase. The
catalytic MR favoured the reaction in terms of TOF; 48 000
moles of furfural have been converted in 1 hour by using 1 mol
of Ru catalyst. The conversion of furfural was obtained to be
about 26%, a value that is lower than some data reported in the
literature, but reasonable if the low ratio of hydrogen/furfural
J. Mater. Chem. A
This journal is © The Royal Society of Chemistry 2018

View Article Online
Communication
Journal of Materials Chemistry A
Fig. 12 TEM and EDX analysis of the PES-220 nm membrane (a) not coated and (b) coated (magnification ꢃ2000 and ꢃ20k, scale bar: left, 5mm;
right, 500 nm).
(1 : 1) and the fact that only about 780 mg (2–3 orders of same conditions, 4% Ir/SiO2 was the most selective to FOL
magnitude less than the literature data) of Ru was loaded onto (96%) with a conversion of 14% and a TOF of 67. Fang et al.18
the membrane are considered. Moreover, it has to be stressed loaded different amounts of Ru nanoparticles, from 1 to 5 wt%,
that the residence time of the reactants in the membrane layer on an acidic MOF material (MIL-101) obtaining a total conver-
was very low, at about 1 s, compared to typical residence times > sion of furfural at 160 ꢀC and 40 bar and in the presence of
1 hour in PBRs. For comparison, Nakagawa et al.16 reported 3 wt% Ru/MIL-101, but with a furfuryl alcohol selectivity of only
a similar conversion (14%) using 2% Pd and 2% Ru on silica 1%. A complete conversion and selectivity to furfuryl alcohol
with 49% selectivity on furfuryl alcohol aer 1 h, but a much was obtained by Yang et al.19 at 20 ꢀC and 5 bar in the presence
higher pressure (80 bar) and H2 : furfural molar ratio (125) were of Ru on an aluminium-based MOF (Al-MIL-53-BDC). However,
used, resulting in a TOF of 337. In the same study and under the the TOF in this case was 21 hꢁ1, indicating that a large amount
of catalyst was used for converting furfural (to convert
21 mol hꢁ1 of furfural, 101.7 gr of Ru was used, which corre-
sponds to 3.5 kg of 2.9 wt% Ru/Al-MIL-53-BDC). Chen et al.14
also achieved a 99% selectivity to FOL at a comparable
temperature (80 ꢀC) and pressure (10 bar) with 32% conversion
and a TOF of 120 using 5% Pt on g-C3N4 nanosheets, but with
a H2 : furfural molar ratio of 2. In order to improve the furfural
conversion, other reaction tests were carried out at the same
temperature and reaction pressure, but varying the H2/furfural
feed molar ratio to 4 : 1. A maximum furfural conversion of 21%
was obtained aer 1 hour with a virtually complete selectivity to
tetrahydro-furfuryl alcohol. This can be explained by the
complete hydrogenation of the heterocyclic ring due to the H2
excess. Moreover, the absence of tetrahydrofurfural among the
products indicates that an increase of H2 favoured further
Fig. 13 Furfural conversion vs. time and Ru content before and after
the reaction test.
hydrogenation of furfuryl-alcohol to tetrahydrofurfuryl alcohol,
This journal is © The Royal Society of Chemistry 2018
J. Mater. Chem. A

View Article Online
Journal of Materials Chemistry A
Communication
J. Mater. Chem. A
This journal is © The Royal Society of Chemistry 2018

View Article Online
Communication
Journal of Materials Chemistry A
rather than the formation of tetrahydrofurfural and then its
hydrogenation to tetrahydrofurfuryl alcohol, as shown in Fig. 1.
Aer 150 min from the beginning of the 4 : 1 test, the
furfural conversion was reduced to 12%, suggesting that the
modied Ru–PES membrane was not stable for the furfural
hydrogenation. As can be seen in Fig. 11, the furfural conversion
was relatively stable for the rst 3 h and then declined to 12%
aer 6.5 h. To understand the reason for the decreased
conversion, the presence of Ru in both (i) the membrane (before
and aer the experiments) and (ii) the product-solutions was
analysed.
The Ru concentration was detected in the product stream in
the whole experimental campaign by ICP-EOS and GC-MS
analyses, as shown in Fig. 14.
The highest Ru concentration of 40 mg Lꢁ1 was found in the
product solution aer 30 min, while only half of that concen-
tration (18 mg Lꢁ1) was in the product aer 180 min. At the same
time, the GC-MS data did not show the presence of the Ru–
Acrylic acid (AA) residue in the product solution. Instead, the
Ru–AA residue started appearing from the product sampled
aer 240 min as shown in Fig. 14.
This suggests that the large presence of Ru in the product
aer 30 min (about 50% of Ru lost in solution) is most likely
related to the removal of “Ru metal clusters” not bonded to the
AA that were not removed by the water pre-washing step under
ambient conditions. This explains the stability of the catalytic
membrane in the rst three hours, while the appearance of the
Ru–AA residue from the forth hour suggests that the active layer
(acrylic monomer and Ru) was also partially removed in the
liquid phase, so that a more resistant metal support rather than
acrylic acid would need to be used for the furfural hydrogena-
tion process. Moreover, from the TEM image aer the hydro-
genation test (Fig. 15), it is possible to see the presence of Ru on
the membrane, which appears to have preserved the homoge-
neity and dispersion of Ru in the coated layer. However, the EDX
conrms that overall, the amount of Ru is lower than that before
the hydrogenation test. To further elucidate this, at the end of
the hydrogenation tests, the metal on the membrane was
Fig. 14 Ru concentration in product solutions.
This journal is © The Royal Society of Chemistry 2018
J. Mater. Chem. A

View Article Online
Journal of Materials Chemistry A
Communication
3 H. Bergem, R. Xu, R. C. Brown and G. W. Huber, Green
Chem., 2017, 19, 3252–3262.
4 A. Sanna, T. P. Vispute and G. W. Huber, Appl. Catal., B, 2015,
165, 446–456.
5 G. Chieffi, C. Giordano, M. Antonietti and D. Esposito,
J. Mater. Chem. A, 2014, 2, 11591–11596.
6 B. Sarkar, C. Pendem, L. N. S. Konathala, T. Sasaki and
R. Bal, J. Mater. Chem. A, 2014, 2, 18398–18404.
7 A. Corma, S. Iborra and A. Velty, Chem. Rev., 2007, 107, 2411–
2502.
8 H. Li, X. Chen, J. Ren, H. Deng, F. Peng and R. Sun,
Biotechnol. Biofuels, 2015, 8, 127.
Fig. 15 TEM image after the hydrogenation test (magnification ꢃ2000
and ꢃ20k, scale bar: left, 5mm;right, 500 nm).
9 D. Santos, U. F. Silva, F. A. Duarte, C. A. Bizzi, E. M. M. Flores
and P. A. Mello, Ultrason. Sonochem., 2018, 40, 81–88.
10 K. Kuroda, K. Miyamura, H. Satria, K. Takada, K. Ninomiya
and K. Takahashi, ACS Sustainable Chem. Eng., 2016, 4,
3352–3356.
11 H. Deka, M. Misra and A. Mohanty, Ind. Crops Prod., 2013,
41, 94–101.
12 V. V. Pushkarev, N. Musselwhite, K. An, S. Alayoglu and
G. A. Somorjai, Nano Lett., 2012, 12, 5196–5201.
quantied before and aer the reaction, conrming a decrease
from 10.9 mg mꢁ2 to 5.76 mg mꢁ2, half of which can be linked to
the active layer degradation. In conclusion, Ru nanoparticles
were added onto the surface of the modied PES membrane
using a UV source.
The catalytic membrane was characterized with different
techniques in order to evaluate the effective presence of the
active layer and the Ru nanoparticles. The membrane with an
initial pore size of 220 nm and 10.9 mg cmꢁ2 of Ru catalyst was
tested for the hydrogenation of furfural under mild conditions
(7 bar, 70 ꢀC) resulting in >99% selectivity towards FOL with
a TOF of 48 000 hꢁ1, when a H2/furfural molar ratio of 1 : 1 was
used. The resulting TOF was considerably higher than those
reported in the literature suggesting that the catalytic
membrane reactor enhances the catalytic activity of Ru in the
selected reaction, while, a >99% selectivity to THF was obtained
by increasing the H2/furfural molar ratio to 4 : 1 under the same
process conditions. The use of the CMR also resulted in a resi-
dence time of the reactants in the membrane layer of less than 2
seconds, compared to a residence time higher than 1 h for
conventional PBRs. Therefore, this work indicates that selective
hydrogenation of furfural to FOL can be successfully performed
using CMRs under mild conditions, but alternative polymeric
support materials need to be developed/tested and the recy-
clability of the catalysts need to be addressed for rendering this
pathway commercially viable.
´
´
´
´
13 C. P. Jimenez-Gomez, J. A. Cecilia, I. Marquez-Rodrıguez,
´
´
´
R. Moreno-Tost, J. Santamarıa-Gonzalez, J. Merida-Robles
and P. Maireles-Torres, Catal. Today, 2017, 279, 327–338.
14 X. Chen, L. Zhang, B. Zhang, X. Guo and X. Mu, Sci. Rep.,
2016, 6, 28558.
15 M. W. Nolte, A. Saraeian and B. H. Shanks, Green Chem.,
2017, 19, 3654–3664.
16 Y. Nakagawa, K. Takada, M. Tamura and K. Tomishige, ACS
Catal., 2014, 4, 2718–2726.
17 S. Liu, Y. Amada, M. Tamura, Y. Nakagawa and
K. Tomishige, Catal. Sci. Technol., 2014, 4, 2535–2549.
18 R. Fang, H. Liu, R. Luque and Y. Li, Green Chem., 2015, 17,
4183–4188.
19 J. Yang, J. Ma, Q. Yuan, P. Zhang and Y. Guan, RSC Adv.,
2016, 6, 92299–92304.
´
´
20 A. O'Driscoll, T. Curtin, W. Y. Hernandez, P. Van Der Voort
and J. J. Leahy, Org. Process Res. Dev., 2016, 20, 1917–1929.
21 J. Llop Castelbou, K. C. Szeto, W. Barakat, N. Merle,
C. Godard, M. Taouk and C. Claver, Chem. Commun.,
2017, 53, 3261–3264.
Conflicts of interest
22 I. Romanenko, M. Lechner, F. Wendler, C. Horenz, C. Streb
and F. H. Schacher, J. Mater. Chem. A, 2017, 5, 15789–15796.
23 L. Meng, M. Kanezashi, X. Yu and T. Tsuru, J. Mater. Chem. A,
2016, 4, 15316–15319.
There are no conicts to declare.
Acknowledgements
24 H. Li, A. Caravella and H. Y. Xu, J. Mater. Chem. A, 2016, 4,
14069–14094.
25 W. J. Koros, Y. H. Ma and T. Shimidzu, Pure Appl. Chem.,
1996, 68(7), 1479–1489.
The authors thank Dr Ida Daniela Perrotta, Department of
Biology, Ecology and Earth Sciences (Di.B.E.S.T.), University of
Calabria-Arcavacata di Rende, (CS), Italy, for the TEM analysis.
`
26 G. Bagnato, A. Iulianelli, A. Vita, C. Italiano, M. Lagana,
C. Fabiano, C. Rossi and A. Basile, Int. J. Membr. Sci.
Technol., 2015, 2, 48–56.
27 J. P. Stanford, M. C. Soto, P. H. Pfromm and M. E. Rezac,
Catal. Today, 2016, 268, 19–28.
28 M. Liu, X. Zhu, R. Chen, Q. Liao, H. Feng and L. Li, Chem.
Eng. J., 2016, 301, 35–41.
References
1 T. P. Vispute, H. Zhang, A. Sanna, R. Xiao and G. W. Huber,
Science, 2010, 330, 1222–1227.
2 J. Shi, M. Zhao, Y. Wang, J. Fu, X. Lu and Z. Hou, J. Mater.
Chem. A, 2016, 4, 5842–5848.
J. Mater. Chem. A
This journal is © The Royal Society of Chemistry 2018

View Article Online
Communication
Journal of Materials Chemistry A
29 E. C. Mengistie, J.-F. Lahitte, #xe7 and ois, Int. J. Chem. Eng., 35 J. Tan, J. Cui, X. Cui, T. Deng, X. Li, Y. Zhu and Y. Li, ACS
2017, 2017, 1–8. Catal., 2015, 5, 7379–7384.
30 W.-S. Lee, J.-C. Lee, H.-T. Oh, S.-W. Baek, M. Oh and 36 M. D. Wales, L. B. Joos, W. A. Traylor, P. Pfromm and
C.-H. Lee, Energy, 2017, 134, 731–742. M. Rezac, Catal. Today, 2016, 268, 12–18.
31 R. Bernstein, E. Anton and M. Ulbricht, ACS Appl. Mater. 37 H. A. Kramers and K. R. Westerterp, Elements of chemical
´
Interfaces, 2012, 4, 3438–3446.
32 D. X. Ye, T. M. Lu and T. Karabacak, Phys. Rev. Lett., 2008,
100, 256102.
reactor design and operation, Netherlands Univ. Press, 1st
British edn, 1963, p. 245, Chapman
B0000CLV7S.
& Hall, ASIN:
33 L. Y. Ng, A. Ahmad and A. W. Mohammad, Arabian J. Chem., 38 J. C. Crittenden, R. R. Trussell, D. W. Hand, K. J. Howe and
2017, 10, S1821–S1834.
G. Tchobanoglous, Water Treatment: Principles and Design,
34 H. Wang, S.-J. Lee, M. V. Olarte and A. H. Zacher, ACS
Sustainable Chem. Eng., 2016, 4, 5533–5545.
2012.
This journal is © The Royal Society of Chemistry 2018
J. Mater. Chem. A