Investigating hydrogenation and decarbonylation in vapor-phase furfural hydrotreating over Ni/SiO2 catalysts: Propylene production

Article abstract of DOI:10.1016/j.apcata.2021.118020

Furfural can be mass-produced from lignocellulose biomass and can be a platform chemical for producing valuable chemicals. In this study, we examine Ni/SiO₂ catalysts for the conversion of furfural under a hydrogen atmosphere. The reactivity an

Full text of DOI:10.1016/j.apcata.2021.118020

Applied Catalysis A, General 613 (2021) 118020
Contents lists available at ScienceDirect
Applied Catalysis A, General
journal homepage: www.elsevier.com/locate/apcata
Investigating hydrogenation and decarbonylation in vapor-phase furfural
hydrotreating over Ni/SiO2 catalysts: Propylene production
,
,
Szu-Hua Chena, Ya-Chun Tsenga, Sheng-Chiang Yanga b,*, Shawn D. Lina **
a Department of Chemical Engineering, National Taiwan University of Science and Technology, 43 Keelung Rd. Sec. 4, Taipei, 10617, Taiwan, ROC
b Department of Chemical and Materials Engineering, Lunghwa University of Science and Technology, No. 300, Section 1, Wanshou Road, Taoyuan, 33306, Taiwan,
ROC
A R T I C L E I N F O
A B S T R A C T
Keywords:
Furfural can be mass-produced from lignocellulose biomass and can be a platform chemical for producing
valuable chemicals. In this study, we examine Ni/SiO2 catalysts for the conversion of furfural under a hydrogen
atmosphere. The reactivity and the product selectivity are governed by the reaction temperature and the Ni
particle size. A catalyst pretreatment by including calcination prior to hydrogen reduction leads to Ni/SiO2–CR
with large Ni particles (~ 15 nm) and a high selectivity to furfuryl alcohol (FA) at below 200 ◦C. The Ni/SiO2-R is
hydrogen-pretreated without a prior calcination and it contains small Ni particles (~ 5 nm) and exhibits rela-
Bio-refinery
Furfural hydrotreating
Structure sensitive
Hydrogenation
Decarbonylation
tively high selectivity to furan. The turnover frequency (TOF) of furfural conversion is 239 and 408 hꢀ 1
,
respectively, on Ni/SiO2-CR and Ni/SiO2-R at 175 ◦C, when the former shows 100% selectivity to FA and the
latter exhibits a selectivity of around 38% and 62%, to FA and furan, respectively. Moreover, the furan can be
reacted to produce propylene and CO by the Ni catalysts at above 200 ◦C and Ni/SiO2-R exhibits a higher activity
than Ni/SiO2-CR. The results suggest that the furfural hydrotreating reaction over Ni catalysts is structure sen-
sitive and a proper design of catalyst and operating temperature can provide a full furfural utilization, either to
FA (a biofuel), or to furan and propylene (as chemical feedstock).
1. Introduction
furfural and HMF to valuable chemicals have attracted great interests [6,
7].
Biomass can be processed through the so-called bio-refinery pro-
cesses to sustainably produce valuable chemicals and fuels. Bio-refinery
involves the breakdown of high-molecular-weight to platform chemicals
and the uses of which to produce chemicals and fuels. In addition to
biogas, bio-oil, and bio-char harvest by cracking via pyrolysis or hy-
drothermal methods, procedures like depolymerization and hydro-
deoxygenation are indispensable in biomass utilization [1–3]. Both
sugar platform and syngas platform are considered as viable approaches
for producing bio fuels and chemicals [4,5]. The sugar platform uses
furfural and hydroxymethyl furfural (HMF) from 5-carbon and 6-carbon
sugar units as the platform chemicals, while the syngas platform refers to
conversions of syngas from biomass gasification. The sugar platform is
theoretically more efficient than the syngas platform because it does not
Furfural was first mass-produced from oat hulls by Quaker Oats
Company in 1922 [8], and furfural has become a precursor [9] for fur-
furyl alcohol (FA), solvents, pharmaceuticals, chemical intermediates,
flavors & fragrance, herbicides, and pesticides, etc. One type of furfural
processing is its reaction with hydrogen wherein hydrogenation,
decarbonylation, ring opening (hydrogenolysis) and combinations of
these can occur [9,10]. These results are somewhat similar to the
hydrotreating process in oil refinery for reducing aromatics (via hy-
drogenation and hydrogenolysis) and heteroatoms (e.g., oxygen, sulfur,
nitrogen). Product selectivity from such furfural hydrotreating is
important, and catalyst plays a determining role.
Many catalysts, mainly Group IB and VIII metal catalysts, have been
tested for furfural hydrotreating. The early works of Wilson et al. [11,
12] show that product selectivity depends on both the type of catalysts
–
require breaking all of the C C bonds; therefore, the conversions of
* Corresponding author at: Department of Chemical Engineering, National Taiwan University of Science and Technology, 43 Keelung Rd. Sec. 4, Taipei, 10617,
Taiwan ROC.
** Corresponding author at: Department of Chemical Engineering, National Taiwan University of Science and Technology, 43 Keelung Rd. Sec. 4, Taipei, 10617,
Taiwan, ROC.
E-mail addresses: scy@mail.ntust.edu.tw, scy@gm.lhu.edu.tw (S.-C. Yang), sdlin@mail.ntust.edu.tw (S.D. Lin).
https://doi.org/10.1016/j.apcata.2021.118020
Received 6 July 2020; Received in revised form 25 December 2020; Accepted 23 January 2021
Available online 1 February 2021
0926-860X/© 2021 Elsevier B.V. All rights reserved.

S.-H. Chen et al.
Applied Catalysis A, General 613 (2021) 118020
and the reaction conditions (temperature, concentration, solvent, and
extent of reaction). In general, Group IB catalysts (such as Cu) catalyze
hydrogenation of the aldehyde group of furfural to produce furfuryl
alcohol (FA) selectively and then to methylfuran (MF) [13]. The group
VIII catalysts can produce FA or THFA selectively under mild conditions,
and many studies have been carried out in liquid phase. Gong et al. [14]
used Ni supported on N-doped active carbon achieving 100% selectivity
to THFA and concluded a synergistic effect of Ni sites and suitable N
doping content to the catalytic performance. Tayler et al. [15] used Pt
catalysts on different supports and concluded that polar solvents and
support acidity could influentially lead to high FA selectivity.
2.2. Catalyst characterization
Temperature-programmed reduction (TPR) analysis was performed
at 5 ◦C/min using 10% H2 in nitrogen and a TCD detector; the H2O
formed during TPR was trapped by a molecular sieve column installed
prior to the detector. The catalyst was typically purged by He at 50 ◦C till
TCD became stable and then switched to H2/N2 flow and the tempera-
ture was ramped. The Ni content was analyzed by inductively coupled
plasma-atomic emission spectrometer (ICP-AES, JY 2000). The Ni size
(calculated with Scherrer equation) and the structure were examined by
X-ray diffraction (XRD) analysis using a synchrotron radiation source
(Beamline #1C, National Synchrotron Radiation Research Center,
Hsinchu, Taiwan). The electron storage ring was operated at 1.5 GeV
with a beam current of 100ꢀ 200 mA. The XRD patterns were recorded
and processed to be consistent with using a wavelength of 1.5418 Å (the
energy of Cu Kα1) and at a scan rate of 10◦ minꢀ 1 with steps of 0.05◦ from
10 to 90◦. The metal (Ni) dispersion was calculated by assuming a Had/
MS ratio of 1, based on H2 chemisorption from dual-isotherm analysis at
room temperature using the BELSORP-max instrument. Thermo-
gravimetric analysis (TGA, Perkin-Elmer, Diamond TG/DTA) was per-
formed to study catalyst coking after furfural hydrotreating, in dry air
At elevated temperatures, decarbonylation (e.g., of furfural to furan),
hydrodeoxygenation (e.g., of FA to MF), ring opening (e.g., of furfural to
butanal), and hydrogenation (e.g., of furan ring and aldehyde group)
can occur simultaneously, especially when using Group VIII catalysts.
Lee et al. [16] used high throughput method to examine the initial ac-
tivity of aqueous-phase hydrogenation over different Group VIII cata-
lysts and found that Ni and Pt were the two having ring-opening
catalytic activity. Sitthisa and Resasco [17] reported that vapor-phase
furfural hydrotreating over Ni catalysts leads to FA, furan, and
ring-open products (C4s, including butanol and butane). Nakagawa
et al. [18] examined gas-phase furfural hydrotreating at 130 ◦C over
Ni/SiO2 and found FA and tetrahydrofurfuryl alcohol (THFA) as the
major products. They mentioned that furfural hydrogenation to FA was
not specifically influenced by changes in Ni particle size but FA hydro-
genation to THFA was influenced. No other study mentioned about such
particle size dependent reaction characteristics during furfural hydro-
treating over Ni catalysts. In contrast, the ring-opening product was
mainly propylene over Pt catalysts [19], instead of C4s frequently noted
over Ni catalysts. Furthermore, Pushkarev et al. [20] clearly demon-
strated that the reaction products from furfural hydrotreating over Pt
catalysts depended on Pt particle size; smaller Pt particles preferred
decarbonylation to form furan while large Pt particles led to more FA
formation from aldehyde hydrogenation. Thus, it would be interesting to
examine the furfural hydrotreating of Ni catalysts with different metal
particle size.
◦
◦
flow (25 mL/min) from 30 to 800 C at a heating rate of 10 C/min.
Transmission Electron Microscopy (TEM) analysis was performed using
a Philips Tecnai F30 FEI-TEM at an acceleration voltage of 200 kV. The
catalyst powders were sprayed on resin gel and a 100-nm thin slice was
cut after the resin is set using FIB (Focused Ion Beam) for TEM
measurement.
2.3. Catalytic reaction tests
Furfural (Acros, 99%) was distilled at 60 ◦C under vacuum to become
clear liquid before feeding to the reactor. Reaction tests were performed
in a quartz-tube packed-bed reactor (7 mm, I.D.) at atmospheric pressure
using a home-made reaction system. The catalyst was reduced in line
with 80 vol% H2/N2 (50 mL/min) at 350 ◦C for 1 h. Furfural was fed by a
syringe pump and carried by H2/He flows into the reactor at PFurfural = 8
kPa, PH2 = 80 kPa and a WHSV (weight hourly space velocity, furfural
mass flow rate divided by cat. weight) of 5~ 25 hꢀ 1. Both He and H2
were of 99.995% purity and were flown through moisture and oxygen
trap columns before entering the system. The reaction was carried out in
a stepwise temperature-ascending-descending mode from 175 to 300 ◦C,
wherein the steady reactions under isothermal conditions were re-
ported. The reactor effluent was analyzed using gas chromatography
with TCD detector and two columns (5% Bentone34 + 5% Dinonyl
Phthalate on Chromosorb W-HP and Porapak-T) via a 10-port valve. The
mass balance of the effluent was analyzed and typically within ± 10% of
the feed except when the temperature approached 300 ◦C. The steady-
state performance was observed typically after from 30 to 90 min on-
stream and the average performance during that period, typically with
a ± 5% error, is used for data analysis. Turnover frequency was calcu-
lated based on the observed kinetic rate and the metal (Ni) surface
measured by H2 chemisorption assuming Had/Nis = 1.
Nickel catalysts are relatively inexpensive and have been used in
different industrial catalytic processes. In this study, we examined Ni/
SiO2 catalysts for vapor-phase furfural hydrotreating attempting to un-
derstand how the reaction products can be controlled. In particular,
whether the reaction is governed by metal particle size, i.e., a structure
sensitive reaction where the specific activity and the product selectivity
can be altered by varying surface sites (edge, step or terrace), is studied.
The results indicate that furfural hydrotreating over Ni/SiO2 is indeed
dependent on metal particle size. In addition, the decarbonylation
products include propylene, which has not been reported over Ni cata-
lysts by flow reactor system although propylene was found to desorb
from furfural over Ni(111) surface by temperature-programmed
desorption (TPD) and density functional theory (DFT) calculation
[21]. The sequential decarbonylation of furfural to furan and to pro-
pylene is a molecular efficient reaction route as that CO would be the
only byproduct.
2. Experimental
3. Results and discussion
2.1. Catalyst preparation
3.1. Catalyst characterization
Supported 5 wt% Ni/SiO2 catalysts were prepared by an incipient-
wetness impregnation of Ni nitrate (Aldrich, 98%) aqueous solution
on a SiO2 support (Davison 952, 300 m2/g, 80~120 mesh, used after
calcination at 550 ◦C). After impregnation, catalysts were put at room
temperature, 12 h for the equilibrium of capillarity, and then dried to
120 ◦C, 12 h, is denoted Ni/SiO2. The catalyst calcined in air at 400 ◦C, 4
h, is denoted as Ni/SiO2-C. The catalyst with/without calcination was
reduced under 10% H2 in N2 at 350 ◦C, 1 h, and is denoted by a suffix R
as Ni/SiO2-R and Ni/SiO2-CR.
The as-prepared catalyst was pretreated in two ways; one was
calcined prior to hydrogen reduction (Ni/SiO2-CR) and the other was
directly reduced with hydrogen (Ni/SiO2-R). The calcined catalyst after
reduction can result in larger Ni particles than that without calcination
[22]. Both the Ni/SiO2 and the Ni/SiO2-C catalysts were examined using
TPR analysis as shown in Fig. 1. The former showed the characteristics
of Ni nitrate reduction (including nitrate decomposition signal) while
the latter indicated the reduction of NiO. NiO phase is typically reduced
by H2 at 300–400 ◦C in TPR and its dispersion (particle size) can
2

S.-H. Chen et al.
Applied Catalysis A, General 613 (2021) 118020
Table 1
Texture properties of the prepared catalysts.
Catalyst
BET SA
Nia
DbNi
dcNi
ddNi
(m2/g)
(wt%)
(%)
(nm)
(nm)
SiO2
297
222
267
0
–
–
–
Ni/SiO2-R
Ni/SiO2-CR
4.9
4.9
21.4
8.4
4.7
11.9
5.3
14.9
a. Ni content from ICP-AES results.
b. The dispersion (%) is calculated with (H2 chemisorption)/(theoretical H2
consumption*reduction degree)*100%.
c. The Ni particle size calculated from chemisorption data using dNi = 1/DNi.
d. The particle size calculated by Scherrer equation based on XRD Ni(111) peak
width.
groups to simplify the discussion, based on their possible routes of for-
mation shown in Scheme 1. Group A products indicate the hydrogena-
tion of aldehyde side chain to the formation of FA and methyl furan
(MF). Group B contains furan and tetrahydrofuran (THF), resulting from
decarbonylation of furfural with/without ring hydrogenation. Group C
refers to the decarbonylation of Group B products with propylene as the
main observed species. Group D refers to the ring opening products that
include butanal, butanol, and butane (summed up as C4 products). The
observed products of this study with selectivity are listed in Tables 2 and
3. The consistent amount of CO byproduct and decarbonylation products
were observed corresponding to Scheme 1, Tables 2, and 3 in our study.
Fig. 3 shows the results of furfural hydrotreating over the Ni/SiO2
catalysts. The unreduced catalysts and the SiO2 support showed negli-
gible activity. Ni/SiO2-R and Ni/SiO2-CR exhibited different activity and
product selectivity. Ni/SiO2-R resulted in a higher furfural conversion
than Ni/SiO2-CR at each reaction temperature. At 175 ◦C, Ni/SiO2-R had
85% furfural conversion with 41% selectivity of Group A (mainly FA)
and 59% selectivity of Group B (mainly furan); Ni/SiO2-CR had 29%
furfural conversion with 100% selectivity of Group A (FA). The Ni/SiO2-
R was more active than Ni/SiO2-CR; the different product selectivity will
be examined and discussed later.
Fig. 1. TPR traces of as-prepared Ni/SiO2 and that after calcination, Ni/SiO2-C.
Thin solid lines represent TPR after the samples were reduced at 350 ◦C by H2
for.1 h.
influence the reduction temperature [23]. The remnant low intensity
TPR signal after 400 ◦C suggests a possible presence of a small amount of
difficultly reduced Ni species, for which Ni-silicate is one possible
explanation [22,23].
XRD results in Fig. 2 show that Ni/SiO2-C contained NiO crystal
phase which was not found in the as-prepared Ni/SiO2. After hydrogen
reduction (at 350 ◦C), both samples contained metallic Ni crystal
structure. The Ni particle size calculated with Scherrer equation is 5.3
and 14.9 nm, respectively, for Ni/SiO2-R and Ni/SiO2-CR, in consistent
with the previous report [22] that a calcination before reduction lead to
an increased Ni particle size. The Ni/SiO2-CR catalyst showed a small
shoulder peak near 43◦, possibly due to the presence of remnant unre-
duced NiO phase or reoxidation of nanoparticles by air during ex situ
XRD measurement. The texture properties of the prepared catalysts are
listed in Table 1. The Ni dispersion (Table 1) calculated based on H2
chemisorption analysis is 8.4% and 21.4%, respectively, for Ni/SiO2-CR
and Ni/SiO2-R. With a generally accepted correlation of dispersion ~
1/dNi (nm), this corresponds to 11.9 nm and 4.7 nm, respectively, for the
two catalysts which are consistent with that analyzed from XRD.
In the ascending-temperature sequence, both Ni/SiO2-R and Ni/
SiO2-CR showed a similar temperature-driven product shift toward less
products of Group A. Up to 225 ◦C, Ni/SiO2-R had an increasing furfural
conversion to 100%, a decreasing selectivity of Group A to 10%, an
increasing selectivity of Group B to 79%, and an appearing selectivity of
Group C (propylene) to 9%. The Ni/SiO2-CR had an increasing furfural
conversion to 70%, a decreasing selectivity of Group A to 50%, and an
increasing selectivity of Group
B to 48%. This indicates that
3.2. Furfural hydrotreating
The observed products from furfural hydrotreating are divided into 4
Fig. 2. XRD pattern of Ni/SiO2 and Ni/SiO2-R for before and after H2 reduction
Scheme 1. Possible reaction routes of furfural hydrotreating over Ni/SiO2
at 350 ◦C using synchrotron light source.
catalysts leading to the observed products and their grouping.
3

S.-H. Chen et al.
Applied Catalysis A, General 613 (2021) 118020
Table 2
Furfural hydrotreating of Ni/SiO2-R.
Selectivity (%)
Temperature (ºC)
Conversion (%)
A
B
C
D
FA
MF
FR
THF
PP
BAL
BOL
175
200
225
250
275
300
250
200
84.6
39.0
6.8
6.5
0.0
0.0
0.0
0.0
0.0
1.8
4.4
3.5
0.3
0.0
0.0
0.0
0.0
59.2
72.9
78.4
73.2
70.5
70.6
85.1
100.0
0.0
2.2
0.7
0.5
0.5
0.7
0.0
0.0
0.0
0.0
1.1
0.7
2.4
3.3
2.2
0.0
0.0
0.0
3.4
1.2
0.0
0.3
0.0
0.0
0.0
97.9
9.2
100.0
100.0
100.0
74.5
9.0
23.6
25.4
26.5
14.9
0.0
25.9
18.4
Table 3
Furfural hydrotreating of Ni/SiO2-CR.
Selectivity (%)
Temperature (ºC)
Conversion (%)
A
B
C
D
FA
MF
FR
THF
PP
BAL
BOL
175
200
225
250
275
300
250
200
28.8
49.7
70.2
90.9
90.0
13.9
7.6
100.0
87.0
50.0
7.8
0.0
5.7
0.0
3.5
2.3
0.0
0.0
–
0.0
0.0
0.0
1.6
0.7
2.8
2.3
0.0
0.0
–
0.0
4.1
0.0
14.1
18.2
0.0
0.0
–
0.0
1.6
1.6
5.3
5.6
0.0
0.0
–
0.0
0.0
0.0
0.0
0.0
0.0
0.0
–
47.6
66.5
69.2
100.0
100.0
–
2.4
0.0
0.0
0.0
–
FA: furfuryl alcohol.
MF: methylfuran.
FR: furan (typically observed with a concurrent CO formation at 1:1 M ratio).
THF: tetrahydrofuran.
PP: propylene (typically observed with a concurrent CO formation at 1:1 M ratio).
BAL: butanal.
BOL: butanol.
decarbonylation (Group B and Group C products) was preferentially
promoted over hydrogenation (Group A) by the ascending temperature.
At 275 ◦C, 100% furfural conversion was achieved over Ni/SiO2-R;
the selectivity of Group B dropped while the selectivity of Group C
gained with a simultaneous small gain in the selectivity of Group D. The
Group A was no more present at this temperature. Similar shifts in
product selectivity were also found with Ni/SiO2-CR. The results showed
that the decarbonylation (to Group B and C) became the dominant re-
SiO2-CR showed a similar temperature-driven product shift; THF (from
ring hydrogenation) was the major product at low temperatures and
propylene (from decarbonylation) with correspondingly produced CO
were the only products at high temperatures. There was no ring opening
product (Group D product mentioned earlier). Therefore, we proposed
in Scheme 1 that Group D products were from the furfural’s ring
opening, not from furan, over the Ni catalysts. Ni/SiO2-R showed higher
furan conversions than Ni/SiO2-CR but the trends of product selectivity
shift with temperature were nearly the same. That the product contained
only THF at low temperature was consistent with that observed during
furfural hydrotreating, i.e., the Ni catalysts preferentially catalyzed
hydrogenation at low temperatures. Similarly, the propylene product at
high temperature indicates that Ni catalysts preferentially catalyze
decarbonylation at high temperatures. During the descending sequence
after 300 ◦C, both catalysts showed deactivation; Ni/SiO2-CR exhibited
more severe deactivation than Ni/SiO2-R and its hydrogenation activity
was more subject to deactivation than decarbonylation, similar with
furfural hydrotreating.
◦
action at 275 C. The C-balance data (in Fig. 8) indicates that carbo-
naceous deposit (coking) became obvious at above 275 ◦C which caused
deactivation and changes in product selectivity. At 300 ◦C and the
subsequent descending sequence, both catalysts showed a deactivation
in furfural conversion; the decay was more dramatic over Ni/SiO2-CR
than Ni/SiO2-R. Moreover, the catalysts produced only decarbonylation
products (Groups B and C), no hydrogenation product, during the sub-
sequent descending sequence. It suggests that the sites for hydrogena-
tion are more subject to deactivation than those for decarbonylation.
3.3. Furan hydrotreating
3.4. Reaction route
Furan and propylene, the decarbonylation products of furfural, were
found over our Ni/SiO2 catalysts. The formation of propylene from
furfural hydrotreating was reported over Pt catalysts [20,24] but it was
not previously reported over Ni catalysts. One report [21] indicates that
propylene desorbed from Ni(111) surface at around 30 ◦C during
furfural TPD. To confirm the sequential decarbonylation and hydroge-
nation pathway over Ni catalysts, we performed furan hydrotreating
over the Ni catalysts and the results are shown in Fig. 4. The observed
products with selectivity are listed in Tables 4 and 5.
Fig. 5 shows the influence of space time on furfural hydrotreating
over Ni/SiO2-R and Ni/SiO2-CR at 175 ◦C. The linear relation at short
contact time (WHSVꢀ 1 ≤ 0.1 h) in Fig. 5a was used to calculate the
turnover frequency (TOF) based on the slope, corresponding to kinetic
rate, and the surface metal fraction from hydrogen chemisorption. The
calculated furfural TOFs are 408 and 236 hꢀ 1, respectively, for Ni/SiO2-
R and Ni/SiO2-CR at 175 ◦C. Fig. 5b shows that the furfural hydro-
treating product selectivity to FA and furan remained constant at around
38% and 62%, respectively, over Ni/SiO2-R at 175 ◦C when space time
In the temperature-ascending sequence, both Ni/SiO2-R and Ni/
4

S.-H. Chen et al.
Applied Catalysis A, General 613 (2021) 118020
Fig. 4. (a) Furan hydrotreating conversion of Ni/SiO2-R and Ni/SiO2-CR ; (b) is
product selectivity of Ni/SiO2-R and Ni/SiO2-CR. Reaction conditions: WHSV of
5 hꢀ 1; PFuran = 8 kPa, PH2 = 80 kPa, balance with Helium at 1 bar; temperature
sequence of ascending from 175 to 300 ◦C and then descending from 300 to 200
◦C. The presented data were the average of 1.5 h onstream time at isothermal
conditions with a typical error of ± 5%.
Table 4
Furan hydrotreating of Ni/SiO2-R.
Selectivity (%)
THF
Conversion
(%)
Temperature (ºC)
PP
175
200
225
250
275
300
250
200
31.5
46.6
54.1
77.6
95.0
99.3
45.9
40.6
100.0
98.7
64.9
20.4
0.0
0.0
1.3
35.1
79.6
100.0
100.0
65.4
0
Fig. 3. (a) Furfural conversion of Ni/SiO2-R, Ni/SiO2-CR, and SiO2 catalysts
and the product selectivity over (b) Ni/SiO2-R and (c) Ni/SiO2-CR during
furfural hydrotreating tests. Reaction conditions: WHSV of 5 hꢀ 1; PFurfural = 8
kPa, PH2 = 80 kPa, balance with Helium at 1 bar; temperature ascending
sequence from 175 to 300 ◦C and then descending from 300 to 200 ◦C. The
presented data were the average of 1.5 h onstream time at isothermal condi-
tions with a typical error of ± 5%. The products of Group A to D are depicted in
Scheme 1.
0.0
34.6
100
was varied. This suggests that the two products were formed from par-
allel reaction pathways, not from a sequential pathway. The results
confirm that furan formation is from the decarbonylation of furfural but
not from the hydrogenolysis (side chain cleavage) of FA. The selectivity
to FA over Ni/SiO2-CR at 175 ◦C was 100% in the range of furfural
conversion observed, which was higher than that of Ni/SiO2-R at the
5

S.-H. Chen et al.
Applied Catalysis A, General 613 (2021) 118020
Fig. S1. However, the THF was the only product observed in the tests at
175 ◦C. The results indicate that at this temperature, the hydrogenation
is the dominant reaction pathway and Ni/SiO2-CR had a higher hydro-
genation activity than Ni/SiO2-R under the conditions of this study.
Table 5
Furan hydrotreating of Ni/SiO2-CR.
Selectivity (%)
THF
Conversion
Temperature (ºC)
(%)
PP
175
200
225
250
275
300
250
200
50.2
47.2
51.8
46.9
53.4
58.4
8.5
100.0
100.0
51.6
18.8
0.0
0.0
3.5. Coking analysis
0.0
48.4
81.2
100.0
100.0
100.0
–
The catalysts after hydrotreating tests were analyzed by TGA and the
weight loss attributable to carbonaceous residue are shown in Fig. 6 and
Table 6. Although not active, SiO2 contained carbonaceous deposit after
reaction test. This indicates that adsorbed furfural formed residue over
SiO2 under the reaction conditions. Ni/SiO2-CR showed a lower amount
of carbonaceous residue (weight basis) than Ni/SiO2-R and the furan
hydrotreating reaction resulted in less residue than furfural hydro-
treating. Derivative thermogravimetry (DTG) was also examined to
identify different types of carbonaceous residue (Fig. S2). Either the type
or the amount of carbonaceous residue can be correlated to the extent of
deactivation noticed in Fig. 3a. ICP analysis indicated that the Ni content
in the spent catalysts was almost the same as that before reaction. XRD of
spent catalysts indicates the presence of Ni3C carbide phase (PDF#06-
0697), whose formation can be caused by carbon deposition on Ni [25,
26]. The spent Ni/SiO2-R had a higher TGA weight loss (Fig. 6) and a
more obvious Ni3C diffraction peak (Fig. 7) than the spent Ni/SiO2-CR.
Carbonaceous deposit monitored by carbon balance at reactor effluent
(Fig. 8) indicates that carbonaceous deposit formed starting from 275 ◦C
and became significant at 300 ◦C. The results suggest that the furfural
hydrotreating over Ni/SiO2 should not go above 250 ◦C to avoid
carbonaceous deposit and deactivation.
0.0
0.0
0
–
THF: tetrahydrofuran.
PP: propylene (typically observed with a concurrent CO formation at 1:1 M
ratio).
3.6. Morphology
The TEM images of reduced and spent catalysts are shown in Fig. 9.
The results indicate that Ni/SiO2-R contains Ni particles mostly in the
range from 3 to 8 nm while Ni/SiO2-CR contains Ni particles mostly
larger than 10 nm. This is consistent with the particle size analysis from
XRD. The spent catalysts show similar particles as the corresponding
reduced catalysts, with no observable fibrous coke. The high-resolution
images of selected Ni particles in Ni/SiO2-R and Ni/SiO2-CR are shown
in Fig. 9(e) and (f), wherein the revealedD-spacing of 0.206 and 0.204
nm, respectively, corresponds to that of Ni (111). The difference in the
exposed facets between smaller particle in Ni/SiO2-R and larger particle
in Ni/SiO2-CR are not significant. According to TEM results, the different
product selectivity between the two catalysts may be related to the
exposed terrace dimension.
Fig. 5. Influence of space time on furfural hydrotreating over Ni/SiO2 catalysts
at 175 ◦C: (a) the furfural conversion vs. WHSVꢀ 1 and (b) the product selec-
tivity vs. furfural conversion.
same levels of furfural conversion. The TOF toward furan formation over
Ni/SiO2-R at 175 ◦C is estimated as 253 hꢀ 1 (408 hꢀ 1 × 62%, equivalent
to 7 × 10–2 s–1), similar to that over Pt nanoparticles of 1.5 nm size (8.8
× 10–2 s–1 at 473 K, 9.3 kPa furfural, 93 kPa H2) [20]. The TOF toward
furfuryl alcohol formation over Ni/SiO2-R at 175 ◦C is 155 hꢀ 1, less than
the 236 hꢀ 1 observed over Ni/SiO2-CR. This indicates that the selectivity
to side chain aldehyde hydrogenation is favorable over the larger Ni
particles of Ni/SiO2-CR and the furfural hydrotreating reaction over Ni
catalysts is a structure-sensitive reaction. We also examined the effect of
space velocity on furan hydrotreating, and the results are shown in
Fig. 6. Thermogravimetric analysis of Ni/SiO2 before and after furfural/furan
hydrotreating reaction.
6

S.-H. Chen et al.
Applied Catalysis A, General 613 (2021) 118020
According to the model, the metal dispersion, D (number of surface
atoms / total number of atoms), can be approximated by D = 1/d, where
d represents the particle size in nm. This leads to a Ni dispersion of 0.2
and 0.067, respectively, for 5 nm and 15 nm size Ni particle. Further-
more, the theoretical fraction of terrace, edge, and corner sites can be
calculated according to truncated octahedral model [27]. Accordingly,
the fraction of terrace (facet) sites is around 0.8 and 0.92, respectively,
over 5 nm and 15 nm Ni particle, while the fraction of edge/corner sites
is 0.2 and 0.08, respectively. This results in a calculated ratio of the
number edge/corner sites over the two sizes of Ni particle as 0.2*0.2 /
0.067*0.08 = 7.5. This strongly indicates that the Ni/SiO2-R catalyst
contains more than seven times of edge/corner sites comparing to the
Ni/SiO2-CR.
Table 6
Comparison Ni wt% before and after reaction and the carbon lost wt%.
Catalyst
Ni fresh
(%)
Ni used
(%)
wt. Loss
(%)
Carbon
Species
Ni/SiO2-R-Furfural
Ni/SiO2-CR-
4.9
4.9
4.8
4.8
8
330, 360 ◦
330, 360 ◦
C
C
4.3
Furfural
SiO2-Furfural
Ni/SiO2-R-Furan
Ni/SiO2-CR-Furan
–
4.9
4.9
–
4.9
4.9
2.4
1.6
0.8
400 ◦
260 ◦
310 ◦
C
C
C
The reaction of furfural + H2 ↔ FA is exothermic, with ΔH◦ = ꢀ 61
kJ/mol based on the enthalpy of formation of gas at standard conditions
listed in National Institute of Standards and Technology (NIST) website.
Sitthisa et al. [28] reported an equilibrium constant of 4.56, 2.98, and
2.01, respectively, at 230, 270, and 290 ◦C. From that and an assumption
of ideal gas, we calculated the equilibrium conversion of furfural to FA
◦
as near 100% at 230ꢀ 290 C under the feed conditions of this study.
This indicates that the reaction is not thermodynamically limited under
the conditions of this study. We showed that this reaction is favored over
Ni/SiO2-CR at lower temperature but it is kinetically suppressed with a
lower selectivity over Ni/SiO2-R (having smaller Ni particles). To the
best of our knowledge, this has not been reported for Ni catalysts,
although similar observation was reported over Pt [20] and Pd [29,30]
catalysts.
Pushkarev et al. [20] reported that both Pt particle size and shape
caused significant influences on the product selectivity of furfural
hydrotreating; for Pt/SiO2 in the range of 443–513 K, the dominant
product was furan over polyhedral Pt of < 2 nm while furfuryl alcohol
was preferred over larger Pt nanoparticles. Cai et al. [31] performed DFT
calculations and showed consistent conclusions in that the furfural
decarbonylation can be promoted by decreasing particle size and by
increasing temperature. They showed that terrace sites would favor
hydrogenation and corner sites would favor decarbonylation. Our re-
sults showed a similar trend in having more furan production over
smaller Ni particles. Based on thermodynamic minimum surface energy
model of fcc nanoparticles, more than 7 times of edge/corner sites can be
presented over the Ni/SiO2-R compared to the Ni/SiO2-CR. This suggests
that the corner/step sites of Ni favor the furfural decarbonylation while
terrace sites favor side chain hydrogenation.
Fig. 7. XRD pattern of Ni/SiO2-R and Ni/SiO2-CR after furfural hydrotreating
using synchrotron light source.
Sitthisa and Resasco [17] reported that Pd/SiO2 showed approxi-
mately 1:1 yield of furan and furfuryl alcohol at 210 ◦C but furan became
the dominant product at higher temperature. Recent results showed that
selective modification of Pd surface sites can result in a significant shift
in product selectivity, indicating a structure sensitivity furfural hydro-
treating reaction. Pang et al. [29] reported that Pd/Al2O3 catalysts had
high furan selectivity at 190 ◦C and a modification with octadecanethiol
self-assembled monolayers significantly suppressed the furan formation
rate resulting in a shift in product selectivity toward furfuryl alcohol.
Zhang et al. [30] reported that Pd/Al2O3 catalysts had high furfuryl
Fig. 8. The carbonaceous deposits were monitored by carbon balance at the
reactor effluent (typically with + 5% error) during furfural hydrotreating. The
reaction could maintain nearly steady state at an on-stream time range of 0.5 to
2.5 h.
◦
alcohol selectivity at 190 C and showed that a modification by ALD
(atomic layer deposition) led to increased furan selectivity. Both studies
proposed that step sites of Pd/Al2O3 are active for furfural side chain
hydrogenation and terrace sites are active for furfural decarbonylation.
This is in contradiction to the results of Pt [20] and Ni catalysts (this
study) wherein small metal particles favor decarbonylation. The exact
reason of such discrepancy is not known but it is suspected that the
adsorption configuration of furfural may be influenced by the presence
of long-chain alkane thiols and/or acidic sites of support [29]. Never-
theless, these results also demonstrated the structure sensitivity of
furfural hydrotreating.
3.7. Discussion
That propylene can be produced through sequential decarbonylation
of furfural has not been reported over Ni catalysts to the best of our
knowledge. We confirm the sequential decarbonylation to propylene via
the tests of furan hydrotreating. The two Ni/SiO2 catalysts examined in
this study indicate that both the reaction rate and the product selectivity
of vapor-phase furfural hydrotreating change with Ni particle size,
indicating a structure-sensitive reaction. From thermodynamic mini-
mum surface energy consideration (Wulff construction), truncated
octahedral is the most stable model of nanoparticles of fcc metals.
We did not observe THFA (tetrahydro furfuryl alcohol) or THMF
(tetrahydro methyl furan) in our reaction products. This may be caused
by the influence of H2/furfural feed ratio, reaction temperature and
7

S.-H. Chen et al.
Applied Catalysis A, General 613 (2021) 118020
Fig. 9. TEM image of (a)Ni/SiO2-R and (b) Ni/SiO2-CR (hydrogen reduction); (d) Ni/SiO2-R-rxn and (d)Ni/SiO2-CR-rxn (after furfural hydrotreating); HR-TEM
images indicate the Ni (111) lattice of (e) Ni/SiO2-R and (f) Ni/SiO2-CR.
maybe space velocity. Nakagawa et al. [18] studied Ni/SiO2 catalyst
under a high H2/furfural ratio and a low space time (H2/furfural = 36/1,
0.884 gcath/molfurfural) and THFA was the major product with a selec-
tivity of 94% and 77%, respectively, at 140 ◦C and 170 ◦C. Sitthisa and
Resasco [17] examined Ni/SiO2 under H2/furfural = 25/1 and a space
time (19.4 gcat h / molfurfural) at temperatures from 175 to 300 ◦C. This
may be the reason why we did not observed THFA and THMF. This also
suggests that the reaction product from furfural hydrotreating can be
tuned by adjusting operating conditions.
◦
time of 9.6 gcath/molfurfural at 210 C and recorded the product selec-
4. Conclusions
tivity of furan (32%), FA (31%) and THFA (5%). These results indicate
that the selectivity to FA and THFA decrease when reaction temperature
is increased and when H2/furfural ratio is decreased. In our study, the
reaction conditions were under low H2/furfural (10/1) and high space
The product distribution of furfural hydrotreating is governed by
both the Ni particle size of Ni/SiO2 and the reaction temperature. Over
Ni/SiO2, a hydrogenation route (to furfuryl alcohol) is preferred at <
8

S.-H. Chen et al.
Applied Catalysis A, General 613 (2021) 118020
200 ◦C and a decarbonylation route (to furan and propylene) becomes
dominant at > 200 ◦C under the conditions of this study. The Ni/SiO2-CR
with ca. 15 nm particle size exhibits a high hydrogenation activity at low
temperature. The Ni/SiO2-R with ca. 5 nm particle size exhibits a high
selectivity to furan at low temperature, and the decarbonylation of furan
to propylene becomes obvious at around 250 ◦C. These results illustrate
the structure sensitivity of the furfural hydrotreating over Ni catalysts,
suggesting that the terrace sites preferentially catalyze side chain hy-
drogenation and the edge or step sites catalyze decarbonylation. The Ni/
SiO2-CR can selectively produce FA at 175 ◦C and it is applicable for this
purpose. With Ni/SiO2-R, furan and propylene can be produced at above
200 ◦C via sequential decarbonylation with CO as the byproduct,
demonstrating a high molecular efficiency furfural utilization to pro-
duce CO, propylene, and furan. This can be an attractive furfural
hydrotreating route for bio-refinery, as a viable alternative to replace
propylene production from petroleum.
References
[1] M. Grilc, B. Likozar, J. Levec, ChemCatChem 8 (2016) 180–191.
[2] M. Grilc, B. Likozar, J. Levec, Appl. Catal. B 150-151 (2014) 275–287.
[3] M. Grilc, B. Likozar, J. Levec, Catal. Today 256 (2015) 302–314.
[4] X. Li, P. Jia, T. Wang, ACS Catal. 6 (2016) 7621–7640.
[5] B. Kamm, Angew. Chemie Int. Ed. 46 (2007) 5056–5058.
¨
[6] S. Larsson, E. Palmqvist, B. Hahn-Hagerdal, C. Tengborg, K. Stenberg, G. Zacchi,
N.-O. Nilvebrant, Enzyme Microb. Technol. 24 (1999) 151–159.
[7] J.N. Chheda, Y. Roman-Leshkov, J.A. Dumesic, Green Chem. 9 (2007) 342–350.
[8] H.J. Brownlee, C.S. Miner, Ind. Eng. Chem. 40 (1948) 201–204.
[9] J.-P. Lange, E. van der Heide, J. van Buijtenen, R. Price, ChemSusChem 5 (2012)
150–166.
[10] G.W. Huber, S. Iborra, A. Corma, Chem. Rev. 106 (2006) 4044–4098.
[11] C.L. Wilson, J. Chem. Soc. (Resumed) (1945) 58–61.
[12] C.L. Wilson, J. Chem. Soc. (Resumed) (1945) 48–51.
[13] J.G.M. Bremner, R.K.F. Keeys, J. Chem. Soc. (Resumed) (1947) 1068–1080.
[14] W. Gong, C. Chen, H. Zhang, Y. Zhang, Y. Zhang, G. Wang, H. Zhao, Mol. Catal. 429
(2017) 51–59.
[15] M.J. Taylor, L.J. Durndell, M.A. Isaacs, C.M.A. Parlett, K. Wilson, A.F. Lee,
G. Kyriakou, Appl. Catal. B 180 (2016) 580–585.
[16] J. Lee, Y. Xu, G.W. Huber, Appl. Catal. B 140-141 (2013) 98–107.
[17] S. Sitthisa, D.E. Resasco, Catal. Lett. 141 (2011) 784–791.
[18] Y. Nakagawa, H. Nakazawa, H. Watanabe, K. Tomishige, ChemCatChem 4 (2012)
1791–1797.
CRediT authorship contribution statement
Szu-Hua Chen: Formal analysis, Investigation, Data curation. Ya-
Chun Tseng: Formal analysis, Investigation, Data curation. Sheng-
Chiang Yang: Conceptualization, Methodology, Writing - review &
editing. Shawn D. Lin: Conceptualization, Methodology, Writing - re-
view & editing, Funding acquisition.
[19] R.C. Runnebaum, T. Nimmanwudipong, J. Doan, D.E. Block, B.C. Gates, Catal. Lett.
142 (2012) 664–666.
[20] V.V. Pushkarev, N. Musselwhite, K. An, S. Alayoglu, G.A. Somorjai, Nano Lett. 12
(2012) 5196–5201.
[21] W. Yu, K. Xiong, N. Ji, M.D. Porosoff, J.G. Chen, J. Catal. 317 (2014) 253–262.
[22] C. Louis, Z.X. Cheng, M. Che, J. Phys. Chem. 97 (1993) 5703–5712.
[23] F. Yang, D. Liu, Y. Zhao, H. Wang, J. Han, Q. Ge, X. Zhu, ACS Catal. 8 (2018)
1672–1682.
Declaration of Competing Interest
The authors report no declarations of interest.
Acknowledgments
[24] C.J. Kliewer, C. Aliaga, M. Bieri, W. Huang, C.-K. Tsung, J.B. Wood,
K. Komvopoulos, G.A. Somorjai, J. Am. Chem. Soc. 132 (2010) 13088–13095.
[25] Y. Goto, K. Taniguchi, T. Omata, S. Otsuka-Yao-Matsuo, N. Ohashi, S. Ueda,
H. Yoshikawa, Y. Yamashita, H. Oohashi, K. Kobayashi, Chem. Mater. 20 (2008)
4156–4160.
[26] Z.L. Schaefer, K.M. Weeber, R. Misra, P. Schiffer, R.E. Schaak, Chem. Mater. 23
(2011) 2475–2480.
[27] P.M. Mortensen, J.-D. Grunwaldt, P.A. Jensen, A.D. Jensen, Catal. Today 259
(2016) 277–284.
This work was supported by the Ministry of Science and Technology,
Taiwan (107-2221-E-011-064, 107-2811-E-011-514, and 107-2914-I-
011-008-A1), the National Taiwan University of Science and
Technology.
[28] S. Sitthisa, T. Sooknoi, Y. Ma, P.B. Balbuena, D.E. Resasco, J. Catal. 277 (2011)
1–13.
[29] S.H. Pang, C.A. Schoenbaum, D.K. Schwartz, J.W. Medlin, Nat. Commun. 4 (2013)
2448.
[30] H. Zhang, X.-K. Gu, C. Canlas, A.J. Kropf, P. Aich, J.P. Greeley, J.W. Elam, R.
J. Meyers, J.A. Dumesic, P.C. Stair, C.L. Marshall, Angew. Chemie Int. Ed. 53
(2014) 12132–12136.
Appendix A. Supplementary data
[31] Q.-X. Cai, J.-G. Wang, Y.-G. Wang, D. Mei, Aiche J. 61 (2015) 3812–3824.
Supplementary material related to this article can be found, in the
online version, at doi:https://doi.org/10.1016/j.apcata.2021.118020.
9