Formic acid enabled selectivity boosting in transfer hydrogenation of 5-hydroxymethylfurfural to 2,5-furandimethanol on highly dispersed Co-Nxsites

Article abstract of DOI:10.1039/d0cy01969k

Catalytic transfer hydrogenation (CTH) reaction is considered as a potential route for upgrading bio-based carbonyl compounds to their corresponding alcohols. Herein, ordered mesoporous N-doped carbon confined Co-N_x(Co-NC) was adopted as a catalyst for converting cellulose-derived 5-hydroxymethylfurfural (HMF) to 2,5-furandimethanol (FDM) using formic acid (FA) as a hydrogen donor. Different catalysts and preparation methods were screened, by varying cobalt phases and template removal procedures. It is found that highly dispersed N-confined Co species (Co-N_x) other than naked Co NPs acted as catalytic species for the CTH of HMF with FA, which gave 86% yield of FDM at 100% HMF conversion. Kinetic experiments revealed that, compared with molecular hydrogen, Co-NC could effectively accelerate HMF hydrogenation and suppress as-formed FDM hydrogenolysis in the presence of FA, which is ascribed to its superior activity toward hydrogen transfer from FA and fast desorption toward FDM. Mechanism studies indicated that C-H dissociation of FA could be the rate-determining step in the CTH reaction, and the hydrogenation of HMF could proceed through an intermolecular hydride transfer route. This work shows that the bifunctional nature of the catalyst is critical in the efficient CTH of biomass-derived carbonyl compounds and provides insights toward the rational design of such catalysts.

Full text of DOI:10.1039/d0cy01969k

Catalysis
Science &
Technology
View Article Online
View Journal | View Issue
PAPER
Formic acid enabled selectivity boosting in
transfer hydrogenation of 5-hydroxymethylfurfural
to 2,5-furandimethanol on highly dispersed Co–Nx
sites†
Cite this: Catal. Sci. Technol., 2021,
11, 1451
b
a
a
a
a
Ling Xu,a Renfeng Nie,
*
*
Xujie Chen, Yanchen Li, Yuxi Jiang and Xiuyang Lu
Catalytic transfer hydrogenation (CTH) reaction is considered as a potential route for upgrading bio-based
carbonyl compounds to their corresponding alcohols. Herein, ordered mesoporous N-doped carbon
confined Co–Nx (Co–NC) was adopted as
a
catalyst for converting cellulose-derived
5-hydroxymethylfurfural (HMF) to 2,5-furandimethanol (FDM) using formic acid (FA) as a hydrogen donor.
Different catalysts and preparation methods were screened, by varying cobalt phases and template removal
procedures. It is found that highly dispersed N-confined Co species (Co–Nx) other than naked Co NPs
acted as catalytic species for the CTH of HMF with FA, which gave 86% yield of FDM at 100% HMF
conversion. Kinetic experiments revealed that, compared with molecular hydrogen, Co–NC could
effectively accelerate HMF hydrogenation and suppress as-formed FDM hydrogenolysis in the presence of
FA, which is ascribed to its superior activity toward hydrogen transfer from FA and fast desorption toward
FDM. Mechanism studies indicated that C–H dissociation of FA could be the rate-determining step in the
CTH reaction, and the hydrogenation of HMF could proceed through an intermolecular hydride transfer
route. This work shows that the bifunctional nature of the catalyst is critical in the efficient CTH of
biomass-derived carbonyl compounds and provides insights toward the rational design of such catalysts.
Received 9th October 2020,
Accepted 2nd December 2020
DOI: 10.1039/d0cy01969k
rsc.li/catalysis
furanmethanol (MFM),8,9 and selective hydrogenation of
carbonyl groups produces 2,5-furandimethanol (FDM). In
Introduction
Transformation of renewable biomass into high value-added
chemicals and fuels has been recognized as an important
approach to alleviate the current dependence on limited fossil
fuels and cut the level of carbon dioxide emission. In this
context, 5-hydroxymethylfurfural (HMF) dehydrated from
lignocellulosic compounds (e.g., glucose and fructose) has
been hailed as a ‘sleeping giant’ due to its rich chemistry and
potential availability.1,2 Association with –CO, furan ring
and –OH groups make it more feasible for HMF to be used in
downstream transformations.3–5 Both the oxidation and
reduction of HMF can produce a variety of valuable building
blocks. For example, oxidation of HMF could generate
2,5-furandicarboxylic acid (FDCA),6,7 hydrogenolysis of HMF
results in 2,5-dimethylfuran (DMF) and 5-methyl-2-
particular, FDM is a promising diol that can be used either
directly for the synthesis of shape self-healing and memory
polymers or for the production of 1,6-hexanediol, drugs and
crown ethers.10
Selective hydrogenation of HMF to FDM is still
challenging, because HMF is intrinsically unstable and can
be transformed to many other undesirable byproducts when
using traditional routes.11–14 For instance, FDM can be
obtained via Cannizzaro reaction, but an equimolar amount
of 5-hydroxymethylfuranoic acid (HMFA) by-products is also
formed,15,16 leading to low atom economy and high cost for
product separation. Moreover, catalytic hydrogenation of
HMF to FDM catalyzed by Pt,17 Ru (ref. 18) and Au (ref. 19)
catalysts is widely explored, however, high selectivity to FDM
is hard to achieve since the as-formed FDM is easily
hydrogenolyzed especially in the presence of high-pressure-
H2 and noble metal catalysts.
a Key Laboratory of Biomass Chemical Engineering of Ministry of Education,
College of Chemical and Biological Engineering, Zhejiang University, Hangzhou
310027, P.R. China. E-mail: luxiuyang@zju.edu.cn
b College of Chemical Engineering, Zhengzhou University, Zhengzhou, 450001, P.R.
China. E-mail: refinenie@163.com
Catalytic transfer hydrogenation (CTH) has emerged as a
promising alternative for the upgrading of biomass-derived
compounds without using high-pressure-H2.20–23 Formic acid
(FA), a co-product from biomass degradation processes,24 has
attracted much recent interest as a promising hydrogen
† Electronic supplementary information (ESI) available: Additional experimental
results. See DOI: 10.1039/d0cy01969k
This journal is © The Royal Society of Chemistry 2021
Catal. Sci. Technol., 2021, 11, 1451–1457 | 1451

View Article Online
Paper
Catalysis Science & Technology
donor for hydrogenation reactions.25–28 As for CTH of HMF,
rare examples using FA as a hydrogen donor and noble Pd or
Ir catalysts were reported,29,30 although >90% yield of FDM
was achieved successively. Considering their earth-abundant
and low cost features, non-noble metals are more desirable
for CTH of HMF with FA.31 However, selectively yielding FDM
with non-noble metals and FA via a CTH route has not been
reported until now.
N2 adsorption/desorption isotherms (Fig. 1a) show that both
Co–NC-A and Co–NC-B display a typical type IV curve with an
H4-type hysteresis loop (p/p0 = 0.5–0.9), along with which a clear
steep at relative pressures of <0.05 could be observed, revealing
the existence of both mesopores and micropores for the two
catalysts. Pore size distributions (Fig. 1a and S2†) further
confirm that Co–NC-A exhibits more distinct mesopores in the
range of 2–10 nm. As shown in Table S1,† the BET surface area
and pore volume of Co–NC-A are calculated to be 883.5 m2 g−1
and 0.3 cm3 g−1, respectively, much higher than those of Co–
NC-B (610.6 m2 g−1 and 0.2 cm3 g−1). These results indicate that
the meso-structure of Co–NC-A is replicated via the base-
removal of SBA-15 and strengthened further via the acid-
removal of unwrapped Co species.
Wide-angle X-ray powder diffraction patterns (XRD) of Co–
NCs are shown in Fig. 1b. Apart from C(002) peak at 25.8°
(PDF 01-0646), Co phases exist differently in Co–NC-A and
Co–NC-B. The diffraction peaks of Co–NC-B that are located
at 19.1°, 31.4°, 36.9°, 59.5° and 65.3° could be attributed to
(111), (220), (311), (511), and (440) reflections of Co3O4 (PDF
43-1003), respectively.40 Meanwhile, CoO and metallic Co
(Co0) are also observed in Co–NC-B. When etching with acid,
neither Co3O4 nor CoO is observed. Although the Co0 phase
is preserved in Co–NC-A, the intensity relative to C(002) is
weakened remarkably. ICP-OES (Table S1†) shows that the Co
content of Co–NC-A is 2.8 wt%, which is much lower than
that of Co–NC-B (21.5 wt%). These results indicate that those
exposed Co species mainly include Co3O4, CoO and Co0, and
after etching with acid, almost all Co3O4 and CoO are
removed, along with which only a small fraction of Co
species (Co–Nx) is preserved within Co–NC-A.
Recently, N-doped carbon has been proved to be a very
attractive catalytic material for Co-catalyzed CTH
processes,32–38 but the nature of its active sites as well as
reaction mechanism toward FA and reactants is still unclear.
In this contribution, we fabricated ordered mesoporous
N-doped carbon confined Co catalysts (Co–NC-A) by using a
Co(phen)2(OAc)2 complex as a precursor and SBA-15 as a hard
template. The resulting Co–NC-A catalyst exhibits a high
surface area (883.5 m2 g−1) and highly dispersed Co–Nx
species in carbon skeleton, which contributed to 88.2% HMF
conversion along with 92.3% FDM selectivity at 160 °C for 5
h. By screening catalysts with different structured Co, the role
of Co species as well as N species was explored. The influence
of FA and H2 on the CTH performance of Co–NC-A was
studied. Wide controlled experiments, including deuterium
labelling experiment, NMR analysis and in situ DRIFT, were
employed for tracking the hydrogen transfer as well as
substrate adsorption over the catalyst. Finally, with the
above-mentioned studies, a possible reaction mechanism was
proposed. This work would be helpful in providing the
feasibility of CTH of HMF to DFM with biomass-derived FA
in the presence of non-noble Co catalysts, which will advance
the green preparation of DFM from biomass-based materials.
Surface composition of Co–NCs is detected by X-ray
photoelectron spectroscopy (XPS). As shown in Fig. 1c, the Co
2p peak intensity of Co–NC-A is much weaker than that of
Co–NC-B owing to the acid-removal of most Co species. The
Co 2p spectra could be fitted into three characteristic peaks
Results and discussion
Characterization of mesoporous Co–NCs
Co–NCs were synthesized by employing SBA-15 as a mesopore
template through a carbonization procedure. In this process,
Co(phen)2(OAc)2 was used as Co, C and N precursors and
inserted into the ordered channels of SBA-15. After
carbonization under a flow of nitrogen, HF was adopted to
remove SBA-15 and exposed cobalt species, affording the
catalyst denoted as Co–NC-A. For comparison, replacing HF
with NaOH only got rid of SBA-15 but retained all Co species
including both confined and exposed cobalt, which was
denoted as Co–NC-B.
Firstly, the pore structure of Co–NCs was studied by small-
angle X-ray diffraction (SAXRD) and is shown in Fig. S1.† Bare
SBA-15 exhibits three well-resolved peaks below 2.5°, which
are ascribed to the (100), (110) and (200) diffraction peaks
derived from ordered mesopores.39 The (110) diffraction is
preserved but shifted to a higher 2θ value as shown in the
pattern of Co–NC-A, suggesting the successful replication of
mesopores with the aid of SBA-15. Comparatively, nearly no
diffraction peaks are observed for Co–NC-B, revealing its low
ordered meso-structure.
Fig. 1 (a) N2 adsorption/desorption isotherms, (b) XRD patterns, and
(c) Co2p and (d) N1s spectra of Co–NC-A and Co–NC-B.
1452 | Catal. Sci. Technol., 2021, 11, 1451–1457
This journal is © The Royal Society of Chemistry 2021

View Article Online
Catalysis Science & Technology
Paper
could be confined by the N-dopants (Fig. S4d†). Apart from a
few Co NPs observed in Co–NC-A, the signals of Co element
are well overlapped with those of N element and spread all
over the N-doped carbon (Fig. 2d and h). This observation
indicates that, although a fraction of the Co species exist as
NPs which are surrounded by N-doped carbon (Fig. S4d†),
the rest of the Co species (Co–Nx) could be highly dispersed
within the N-doped carbon and interact intimately with the N
dopants
(especially
for
pyridinic
N),
since
the
Co(phen)2(OAc)2 complex was widely adopted as a precursor
for fabricating single atom Co catalysts.41
Fig. 2 (a and e) Bright field TEM and (b and f) HAADF-STEM images of
Co–NC-A; (c, d, g and h) elemental mapping of Co–NC-A.
of Co0, Co3+ and Co2+ with BEs at 778.7, 780.3 and 782.8 eV,
respectively.34 The fitting parameters (Table S2†) show that
Co–NC-A exhibits higher Co0 (7.6%) but lower Co3+ (42.6%)
percentage than Co–NC-B, revealing the stabilization role of
N-dopants for Co species. The N1s spectra were deconvoluted
into peaks corresponding to pyridinic N (398.0 eV), pyrrolic
N/pyridinic N–Co (399.6 eV), graphitic N (401.1 eV) and
pyridine oxide (403.3 eV) for the two catalysts (Fig. 1d). The
fitting parameters (Table S2†) show that Co–NC-B exhibits
lower pyridinic N (31.7%) but higher pyrrolic N/pyridinic N–
Co (52.3%) percentage than Co–NC-A, indicating that those
unwrapped Co species in Co–NC-B also interact with N
species. After etching with acid, the pyridinic N-Co species
were partially recovered into pyridinic N. This speculation is
also proved as shown in the C1s spectra (Fig. S3†), in which a
relatively lower percentage of C–C/CC species is observed
for Co–NC-B.
CTH of HMF with FA
Initially, catalyst screening was carried out for CTH of HMF
with FA at 160 °C for 5 h (Table 1). In the blank experiment,
only 6.8% HMF conversion along with negligible FDM yield
was obtained in the absence of a catalyst (entry 1). Among Fe-,
Ni-, Cu- and Co-based catalysts (entries 2–5), Co–NC-A
accounted for the highest HMF conversion (88.2%) and 92.3%
FDM selectivity, exhibiting better performance than Fe–NC-A,
Ni–NC-A and Cu–NC-A, in which 24.2, 31.7 and 15.8% HMF
conversions were obtained, respectively. Compared with those
reported
catalysts
in
literature
studies
(Table
S3†),18,19,29,30,43–46 Co–NC-A exhibited comparable catalytic
performance for transfer hydrogenation of HMF. Other noble
Ru/C, Pd/C and Pt/C catalysts only resulted in 41.7, 46.1 and
37.8% HMF conversions, respectively (entries 6–8), along with
relatively low FDM selectivities (70.0–88.9%). It should be
noted that high HMF conversion was commonly accompanied
with high conversion of FA, indicating that efficient FA
dehydrogenation should be a prerequisite for CTH of HMF.
In order to explore the active sites of Co–NC-A, several
controlled catalysts were evaluated for CTH of HMF with FA
(Fig. 3a). Co, Co3O4 and CoO accounted for similar HMF
conversions (16.4–18.7%), but achieved 71.2, 39.7 and 36.7%
FDM selectivities, respectively. As a precursor of Co–NC-A,
Co(phen)2 only resulted in 19.3% conversion which was
higher than with phen (10.6%). These catalysts are much less
The morphology of Co–NCs is disclosed using
a
transmission electron microscope (TEM) (Fig. 2 and S4†).
TEM images show that the morphology of Co–NC-B and Co–
NC-A is analogous to that of parent SBA-15. Base-etching
resulted in more Co particles as well as wider particle size
distribution in comparison to Co–NC-A. For instance, the
mean particle sizes of Co NPs in Co–NC-B and Co–NC-A are
22.9 and 10.5 nm, respectively. After etching with acid, most
exposed Co species were washed away, and the retained ones
Table 1 Transfer hydrogenation of HMF with FA over different catalystsa
Selec./%
Molar ratio of
Conv. of
Entry
Catalysts
HMF to Co
Conv./%
FDM
MFM
LA
FA/%
1
2
3
4
5
6
7
8
No catalyst
Co–NC-A
Fe–NC-A
Ni–NC-A
Cu–NC-A
5 wt% Ru/C
5 wt% Pd/C
5 wt% Pt/C
—
6.8
88.2
24.2
31.7
15.8
41.7
46.1
37.8
0
0
14.7
3.4
5.0
3.5
1.9
5.8
5.6
5.6
1.6
89.5
16.7
33.5
23.1
34.0
40.7
40.0
25.0
22.8
26.3
35.3
24.1
25.3
46.4
92.3
90.1
88.6
84.8
70.0
88.9
83.3
2.4
2.1
2.2
0
3.4
2.4
4.2
a
Reaction conditions: HMF (30 mg), catalyst (20 mg), nFA/HMF = 7.5, 1,4-dioxane (3.5 mL), 0.5 MPa N2, 160 °C, 5 h, 500 rpm.
This journal is © The Royal Society of Chemistry 2021
Catal. Sci. Technol., 2021, 11, 1451–1457 | 1453

View Article Online
Paper
Catalysis Science & Technology
the FDM selectivity to 87.8%, ascribing to over-hydrogenation
of FDM in the presence of excess FA.
As shown in Fig. S6,† reaction temperature had
a
remarkable effect on the catalyst performance in CTH of
HMF. Increasing the temperature from 120 to 180 °C resulted
in an increase of the reaction rate but a decrease of FDM
selectivity at a high HMF conversion. It is found that ln(c/c0)
followed a linear relationship with the initial reaction time.
On the basis of the Arrhenius plot, the apparent activation
energies (Ea) for the transformation of HMF was calculated to
be 27.5 kJ mol−1. Meanwhile, a gram-scale experiment was
performed using one gram of HMF as a reactant at 180 °C
(Fig. S7†). After 10 h, 92.1% HMF was converted,
accompanied with 76.3% FDM yield.
Catalyst recycling shows that Co–NC-A is highly stable and
could be recycled five times without a significant loss in the
efficiency (Fig. 3d). ICP-OES analysis for the reaction liquid
shows that the Co concentration was below the detection
limits, but 28.8% Co leached into solution for Co–NC-B.
Fig. 1b shows that Co3O4, CoO and metallic Co are observed
in the XRD pattern of fresh Co–NC-B. However, after being
used, neither Co3O4 nor CoO is observed (Fig. S8b†). By
contrast, XRD, TEM and XPS characterization techniques
(Fig. S8–S10†) show that the morphology or crystalline phase
of the recycled Co–NC-A did not change too much,
confirming the stable physicochemical properties of Co–NC-A
in acidic media.
Fig. 3 (a) CTH of HMF over the controlled catalysts. (b) Effect of
solvents and (c) FA usage on the CTH of HMF over Co–NC-A. (d)
Recycling test of Co–NC-A for the CTH of HMF. Reaction conditions:
HMF (30 mg), catalyst (20 mg), nFA/HMF =7.5, solvent (3.5 mL), 0.5 MPa
N2, 160 °C, 5 h, 500 rpm. Co–NC-A(H) catalyst was prepared via
treatment of Co–NC-A with HNO3 solution (3 M, 20 mL) under reflux
at 100 °C for 5 h. Equivalent Co in the controlled catalysts was used in
Fig. 3(a).
active than Co–NC-A, indicating the important role of
N-doped carbon for promoting the CTH performance of Co-
based catalysts, in which intimate interaction between cobalt
and N-dopants (e.g. Co–Nx) could be existed.47,48
With the optimal Co–NC-A in hand, the influence of
solvents on the CTH of HMF was studied and is shown in
Fig. 3b. Four polar protic solvents, e.g. methanol, ethanol,
water and isopropanol, resulted in moderate to good HMF
conversions (40.2–92.5%) but low FDM selectivities (6.5–
30.4%). It is reported that strong proton solvent would favor
the generation of H+ ions from FA, thus aggravating the
polymerization or condensation of HMF.49 Other four aprotic
solvents, e.g. ethyl acetate, dichloromethane, DMF and
cyclohexane, exhibited 31.6–92.3% HMF conversions but
better FDM selectivities (55–64%). Comparatively, 1,4-dioxane
is the best solvent for achieving both high conversion and
FDM selectivity. Moreover, catalyst loading has a profound
influence on the CTH performance (Fig. S5†). Increasing the
catalyst loading from 5 to 20 mg resulted in a remarkable
increase of HMF conversion from 47.3 to 88.2%, while the
FDM selectivity remained almost constant at above 80%.
However, excess catalyst loading led to a drop of FDM
selectivity, probably due to its over-hydrogenation or
degradation of HMF.
Considering that FA acts as a hydrogen source, the mole
ratio of FA/HMF could affect the CTH performance of Co–NC-
A (Fig. 3c). It can be found that no FDM was formed without
adding of FA. Increasing the mole ratio of FA/HMF from 1 to
9 led to a significant increase of HMF conversion from 11.7
to 92.5% but a slight drop of FA conversion from 100 to
85.8%. FDM selectivity was increased dramatically from 31.4
to 92.9% with the increase of the mole ratio of FA/HMF from
1 to 7.5. However, further increasing this value to 9 decreased
Mechanism of CTH of HMF with FA
NH3-TPD analyses for Co–NC-A and Co–NC-B were performed
and are shown in Fig. S11.† It is found that nearly no
desorption peaks could be observed for the two samples,
revealing that negligible acid sites exist in the two catalysts.
This result indicates that Lewis acid might not be an active
center for the CTH reactions with FA. Detailed studies of the
CTH performance of Co–NC-A and base-treated counterpart
(Co-NC-B) were carried out and are shown in Table S4.† It is
found that Co–NC-B exhibited 58.1 and 95.2% HMF
conversions at 2 h and 5 h, respectively, which are slightly
higher than those obtained on Co–NC-A (54.9 and 88.2%).
Co–NC-B contains nearly an equivalent amount of acid-
insoluble cobalt species (existing as Co–Nx) as Co–NC-A, but
a
much higher amount of acid-soluble cobalt species
(existing as Co or CoOx) than Co–NC-A, indicating that the
Co–Nx species could be the active sites that are responsible
for CTH of HMF with FA. In order to verify this assumption,
Co–NC-A was treated with 3 M HNO3 (denoted as Co-NC-
A(H)) so as to destroy the Co–Nx species (Fig. 3a). It is found
that HNO3 treatment led to ∼13% Co removal as well as a
9.4% drop of HMF conversion, although FDM selectivity
remained nearly constant. This result indicates that those
Co–Nx species are relatively stable in acid and could be acted
as main active centers.
Fig. 4 shows the hydrogenations of HMF/FDM with FA or
H2 over Co–NC-A. Using FA as a hydrogen donor resulted in
1454 | Catal. Sci. Technol., 2021, 11, 1451–1457
This journal is © The Royal Society of Chemistry 2021

View Article Online
Catalysis Science & Technology
Paper
in the presence of FA and H2, selective hydrogenation of
FDM was investigated as shown in Fig. 4c and d. It is found
that FDM was not stable and easy to polymerize, and the
main detected product was MFM when using FA as a
hydrogen donor. Meanwhile, the as-formed MFM could be
quickly hydrogenolyzed into DMF in the presence of H2,
further revealing the strong hydrogenolysis capacity of Co–
NC-A with the aid of H2.
In order to figure out the reason, H2-TPD studies were
conducted and are shown in Fig. 5b. It is found that Co–NC-
B exhibited three large peaks at 147 °C, 265 °C and 317 °C,
attributing to chemically adsorbed hydrogen on the exposed
metallic cobalt.50,51 Meanwhile, Co–NC-A showed a weak
desorption peak at 288 °C and exhibited 13.3 times lower H2
adsorption capacity (0.14 mmol g−1) than Co–NC-B (1.84
mmol g−1), revealing its weak capacity for activating H2.
Meanwhile, the adsorption capacity of HMF over Co-NC-A is
0.86 μg g−1 (Fig. 5c and S12†), much higher than that of FDM
(0.23 μg g−1), indicating that Co–NC-A could facilitate HMF
conversion but restrain over-hydrogenation of FDM due to its
easy desorption.
In order to study the hydrogenation mechanism of HMF
with FA over Co–NC-A, furfural was used as a model reactant
for the in situ DRIFT studies (Fig. 6). As for FA adsorption at
room temperature (Fig. 6a), the absorption bands at 2946,
1795, 1735 and 1214 cm−1 can be ascribed to C–H, CO
(gaseous FA), CO (absorbed FA) and C–O vibrations of FA,
respectively. Increasing temperature to 135 oC led to
significant weakening of CO (gaseous FA) and C–O
vibrations, indicating enhanced interaction between FA and
Co–NC-A, in which the protons of FA could be scrambled by
N-dopants while the as-formed HCOO* was stabilized by
adjacent cobalt. With prolonged reaction time, CO
(absorbed FA) and C–H vibrations weakened gradually, which
followed with the appearance of CO2 stretching at 2324 cm−1.
During the whole reaction time, no signals of CO and water
were observed, indicating that FA decomposition was
dominated by dehydrogenation other than dehydration.52–54
Fig. 6b shows the FTIR spectra with introducing both furfural
and FA. The weakening of characteristic vibrations of FA was
accelerated compared with FA alone, indicating that the
introduced furfural facilitated its interaction with FA, which
decelerated the FA dehydrogenation. Meanwhile, new peaks
at ∼3200 and 1061 cm−1 were strengthened with prolonged
reaction time, which could be ascribed to the –OH and C–O
Fig. 4 Using (a and c) FA or (b and d) H2 as hydrogen donors for the
CTH of HMF and FDM. Reaction conditions: HMF or FDM (30 mg), Co-
NC-A (20 mg), nFA/HMF = 7.5, 1,4-dioxane (3.5 mL), 0.5 MPa N2 or 0.5
MPa H2, 160 °C, 500 rpm.
FDM as the main product and no formation of DMF (Fig. 4a).
Comparatively, H2 led to much a slower reaction rate and
lower FDM selectivity (Fig. 4b). For instance, prolonging the
reaction time from 0.5 to 8 h, HMF conversion increased
slowly from 14.8 to 48.1%, while FDM selectivity decreased
from 80.3 to 39.5%, which was accompanied with an increase
of DMF selectivity from 18.3 to 55.3%. When replacing 0.5
MPa N2 with H2, allowing both FA and H2 as hydrogen
donors (Fig. 5a), although the reaction rate was comparable
with FA alone, significant DMF formation could still be
observed. This result indicates that, compared with FA,
although molecular H2 is hard to be activated, it facilitates
hydrogenolysis of FDM over Co–NC-A. To further
demonstrate the different catalyst performances of Co–NC-A
Fig. 5 (a) The effect of hydrogen donor to the CTH of HMF: reaction
conditions: HMF (30 mg), Co–NC-A (20 mg), nFA/HMF = 7.5, 1,4-dioxane
(3.5 mL), 0.5 MPa H2/N2, 160 °C, 500 rpm, 3 h; (b) H2-TPD spectra of
Co–NC-A and Co–NC-B; (c) reactant absorption over Co–NC-A; (d)
deuterium-labelling studies of FAL using FA.
Fig. 6 In situ DRIFT of (a) FA and (b) FA + FAL adsorption over Co–NC-A.
Catal. Sci. Technol., 2021, 11, 1451–1457 | 1455
This journal is © The Royal Society of Chemistry 2021

View Article Online
Paper
Catalysis Science & Technology
exhibiting better performance than other frequently-used
cobalt-based catalysts and commercial noble Pd-, Pt- and Ru-
based catalysts. Control experiments suggest that FA
facilitated HMF hydrogenation but restrained FDM
hydrogenolysis in comparison to molecular hydrogen.
Mechanism studies reveal that FA dehydrogenation could be
rate-limiting for CTH of HMF, in which hydrogen was
transferred directionally from formate to the α-carbon of
carbonyl group via an intermolecular hydride transfer
process. Additionally, Co–NC-A was acid-resistant and could
catalyze scalable CTH of HMF without an obvious loss of its
activity. The atom-level understanding gained regarding the
reaction pathway lays the foundation for the development of
efficient catalytic process for biomass upgrading, showing
potential applications in sustainable chemical industry.
Fig. 7 Possible reaction mechanism of CTH of HMF over Co–NC-A.
vibrations of the newly generated furfuryl alcohol,
respectively.55
The deuterium-labelling studies (Fig. 5d and S13†) reveal
that, replacing HCOOH with DCOOH resulted in a much
lower reaction rate for both furfural and FA. As for furfural
conversion, the reaction rate constant ratio (kH/kD) with
HCOOH and HCOOD was calculated to be 2.03, while this
value was 2.53 for FA dehydrogenation. The relatively low
reaction rate along with high kH/kD of FA confirm that the
dissociation of the C–H bond in FA could be the rate
determining step in the CTH reaction.42 Moreover, using
DCOOH as a hydrogen donor resulted in 61% conversion as
Conflicts of interest
There are no conflicts to declare.
Acknowledgements
This work was supported by the National Natural Science
Foundation of China (No. 21676243, 22078290), and the
Fundamental Research Funds for the Central Universities (No.
2020BCE002).
1
determined by H-NMR (Fig. S14†), in which 88% deuterium
was incorporated into the α-carbon of furfural.
Notes and references
In light of the above observations, a possible mechanism for
CTH of HMF with FA was illustrated in Fig. 7. Firstly, the
electron-rich N species in Co–NC-A capture H* from FA to form
N–H*, and cobalt–formate (Co–HCOO*) species are generated
due to the coordination between electronegative formate
anions and the empty d orbitals of adjacent Co species.42,56
Then, carbonyl oxygen in HMF is adsorbed on Co owing to its
oxophilic nature57,58 and coordinates with the Co–HCOO*
species. The hydrogen is transferred from formate to the
carbonyl group via an intermolecular hydride transfer
mechanism with a six-membered-ring intermediate, similar to
the Meerwein–Ponndorf–Verley (MPV) mechanism. Finally,
FDM is formed and desorbed from the catalyst, accompanied
with the release of CO2. The proposed mechanism reveals that
the transfer of hydrogen in formate species should be the rate-
determining step in the CTH of HMF with FA.
1 M. J. Climent, A. Corma and S. Iborra, Green Chem.,
2014, 16, 516–547.
2 L. Hu, L. Lin and S. Liu, Ind. Eng. Chem. Res., 2014, 53,
9969–9978.
3 N.-T. Le, P. Lakshmanan, K. Cho, Y. Han and H. Kim, Appl.
Catal., A, 2013, 464–465, 305–312.
4 T. S. Hansen, I. Sádaba, E. J. García-Suárez and A. Riisager,
Appl. Catal., A, 2013, 456, 44–50.
5 A. Takagaki, M. Takahashi, S. Nishimura and K. Ebitani, ACS
Catal., 2011, 1, 1562–1565.
6 O. Casanova, S. Iborra and A. Corma, ChemSusChem, 2009, 2,
1138–1144.
7 O. Casanova, S. Iborra and A. Corma, J. Catal., 2009, 265,
109–116.
8 Z.-Z. Yang, J. Deng, T. Pan, Q.-X. Guo and Y. Fu, Green
Chem., 2012, 14, 2986–2989.
9 J. Ma, Z. Du, J. Xu, Q. Chu and Y. Pang, ChemSusChem,
2011, 4, 51–54.
Conclusion
Ordered mesoporous N-doped carbon confined Co–Nx
catalysts (Co–NCs) were synthesized by facile carbonization
using SBA-15 as a hard template with Co(phen)2(OAc)2 as Co,
C and N sources. As revealed by catalyst screening, different
catalyst preparations and cobalt phases remarkably affected
the catalyst performance in CTH of HMF with FA as a
hydrogen donor. It is found that Co–Nx species other than
naked Co species are responsible for FA dehydrogenation
and carbonyl group hydrogenation. With Co–NC-A as the
optimal catalyst, 86.0% FDM yield could be obtained,
10 J. M. Timko and D. J. Cram, J. Am. Chem. Soc., 1974, 96,
7159–7160.
11 H. Cai, C. Li, A. Wang and T. Zhang, Catal. Today, 2014, 234,
59–65.
12 A. S. Amarasekara and B. Wiredu, Ind. Eng. Chem. Res.,
2015, 54, 824–831.
13 R. Alamillo, M. Tucker, M. Chia, Y. Pagán-Torres and J.
Dumesic, Green Chem., 2012, 14, 1413–1419.
14 M. Chatterjee, T. Ishizaka and H. Kawanami, Green Chem.,
2014, 16, 1543–1551.
1456 | Catal. Sci. Technol., 2021, 11, 1451–1457
This journal is © The Royal Society of Chemistry 2021

View Article Online
Catalysis Science & Technology
Paper
15 E.-S. Kang, D. W. Chae, B. Kim and Y. G. Kim, J. Ind. Eng.
Chem., 2012, 18, 174–177.
37 H. Guo, R. Gao, M. Sun, H. Guo, B. Wang and L. Chen,
ChemSusChem, 2019, 12, 487–494.
16 S. Subbiah, S. P. Simeonov, J. M. S. S. Esperança, L. P. N.
Rebelo and C. A. M. Afonso, Green Chem., 2013, 15,
2849–2853.
17 M. Chatterjee, T. Ishizaka and H. Kawanami, Green Chem.,
2014, 16, 4734–4739.
18 J. Han, Y.-H. Kim, H.-S. Jang, S.-Y. Hwang, J. Jegal, J. W. Kim
and Y.-S. Lee, RSC Adv., 2016, 6, 93394–93397.
19 J. Ohyama, A. Esaki, Y. Yamamoto, S. Arai and A. Satsuma,
RSC Adv., 2013, 3, 1033–1036.
20 J. Jae, W. Zheng, R. F. Lobo and D. G. Vlachos,
ChemSusChem, 2013, 6, 1158–1162.
38 H. Guo, B. Wang, P. Qiu, R. Gao, M. Sun and L. Chen, ACS
Sustainable Chem. Eng., 2019, 7, 8876–8884.
39 M. Kruk, M. Jaroniec, C. H. Ko and R. Ryoo, Chem. Mater.,
2000, 12, 1961–1968.
40 L. Xue, C. Zhang, H. He and Y. Teraoka, Appl. Catal., B,
2007, 75, 167–174.
41 W. Liu, L. Zhang, W. Yan, X. Liu, X. Yang, S. Miao, W. Wang,
A. Wang and T. Zhang, Chem. Sci., 2016, 7, 5758–5764.
42 P. Zhou, Z. Zhang, L. Jiang, C. Yu, K. Lv, J. Sun and S. Wang,
Appl. Catal., B, 2017, 210, 522–532.
43 J. Jae, E. Mahmoud, R. F. Lobo and D. G. Vlachos,
ChemCatChem, 2014, 6, 508–513.
44 J. Wei, X. Cao, T. Wang, H. Liu, X. Tang, X. Zeng, Y. Sun, T.
Lei, S. Liu and L. Lin, Catal. Sci. Technol., 2018, 8, 4474–4484.
45 T. Wang, J. Zhang, W. Xie, Y. Tang, D. Guo and Y. Ni,
Catalysts, 2017, 7, 92.
46 W. Hao, W. Li, X. Tang, X. Zeng, Y. Sun, S. Liu and L. Lin,
Green Chem., 2016, 18, 1080–1088.
47 R. Gao, H. Guo, B. Wang, P. Qiu, M. Sun and L. Chen, Appl.
Catal., A, 2019, 579, 99–105.
48 S. Chen, L.-L. Ling, S.-F. Jiang and H. Jiang, Green Chem.,
2020, 22, 5730–5741.
21 D. Scholz, C. Aellig and I. Hermans, ChemSusChem, 2014, 7,
268–275.
22 J. Du, J. Zhang, Y. Sun, W. Jia, Z. Si, H. Gao, X. Tang, X.
Zeng, T. Lei and S. Liu, J. Catal., 2018, 368, 69–78.
23 Z. Zhang, S. Yao, C. Wang, M. Liu, F. Zhang, X. Hu, H. Chen,
X. Gou, K. Chen and Y. Zhu, J. Catal., 2019, 373, 314–321.
24 T. Lu, Y. Hou, W. Wu, M. Niu and Y. Wang, Fuel Process.
Technol., 2018, 171, 133–139.
25 J. Tuteja, H. Choudhary, S. Nishimura and K. Ebitani,
ChemSusChem, 2014, 7, 96–100.
26 E. J. García-Suárez, A. M. Balu, M. Tristany, A. B. García,
K. Philippot and R. Luque, Green Chem., 2012, 14,
1434–1439.
49 Z. Fu, Z. Wang, W. Lin, W. Song and S. Li, Appl. Catal., A,
2017, 547, 248–255.
27 D. Zhang, F. Ye, T. Xue, Y. Guan and Y. M. Wang, Catal.
Today, 2014, 234, 133–138.
50 D. Xu, W. Li, H. Duan, Q. Ge and H. Xu, Catal. Lett.,
2005, 102, 229–235.
28 R. Javaid, S.-i. Kawasaki, A. Suzuki and T. M. Suzuki,
Beilstein J. Org. Chem., 2013, 9, 1156–1163.
51 X. Liu, W. Jia, G. Xu, Y. Zhang and Y. Fu, ACS Sustainable
Chem. Eng., 2017, 5, 8594–8601.
29 T. Thananatthanachon and T. B. Rauchfuss, Angew. Chem.,
Int. Ed., 2010, 49, 6616–6618.
30 T. Thananatthanachon and T. B. Rauchfuss, ChemSusChem,
2010, 3, 1139–1141.
31 Y. Sun, C. Xiong, Q. Liu, J. Zhang, X. Tang, X. Zeng, S. Liu
and L. Lin, Ind. Eng. Chem. Res., 2019, 58, 5414–5422.
32 L. Xu, R. Nie, H. Xu, X. Chen, Y. Li and X. Lu, Ind. Eng.
Chem. Res., 2020, 59, 2754–2760.
33 L. Xu, R. Nie, X. Lyu, J. Wang and X. Lu, Fuel Process.
Technol., 2020, 197, 106205.
52 K. Tedsree, T. Li, S. Jones, C. W. A. Chan, K. M. K. Yu, P. A. J.
Bagot, E. A. Marquis, G. D. W. Smith and S. C. E. Tsang, Nat.
Nanotechnol., 2011, 6, 302–307.
53 T. C. Johnson, D. J. Morris and M. Wills, Chem. Soc. Rev.,
2010, 39, 81–88.
54 B. Loges, A. Boddien, F. Gärtner, H. Junge and M. Beller,
Top. Catal., 2010, 53, 902–914.
55 Y. Zhu, W. Zhao, J. Zhang, Z. An, X. Ma, Z. Zhang, Y. Jiang,
L. Zheng, X. Shu, H. Song, X. Xiang and J. He, ACS Catal.,
2020, 8032–8041.
34 H. Yang, R. Nie, W. Xia, X. Yu, D. Jin, X. Lu, D. Zhou and Q.
Xia, Green Chem., 2017, 19, 5714–5722.
56 J. Tan, J. Cui, X. Cui, T. Deng, X. Li, Y. Zhu and Y. Li, ACS
Catal., 2015, 5, 7379–7384.
35 Q.-Y. Bi, J.-D. Lin, Y.-M. Liu, H.-Y. He, F.-Q. Huang and Y.
Cao, Angew. Chem., Int. Ed., 2016, 55, 11849–11853.
36 L. Tao, Q. Zhang, S.-S. Li, X. Liu, Y.-M. Liu and Y. Cao, Adv.
Synth. Catal., 2015, 357, 753–760.
57 M. Mavrikakis and M. A. Barteau, J. Mol. Catal. A: Chem.,
1998, 131, 135–147.
58 Z. Gao, L. Yang, G. Fan and F. Li, ChemCatChem, 2016, 8,
3769–3779.
This journal is © The Royal Society of Chemistry 2021
Catal. Sci. Technol., 2021, 11, 1451–1457 | 1457