Insights into the active sites and catalytic mechanism of oxidative esterification of 5-hydroxymethylfurfural by metal-organic frameworks-derived N-doped carbon

Article abstract of DOI:10.1016/j.jcat.2019.11.029

Directly oxidative esterification of Biomass-derived 5-hydroxymethylfurfural (HMF) into dimethyl furan dicarboxylate (DMFDCA) is a promising route for the replacement of petroleum-derived commodity chemical terephthalic acid (TPA) extensively employed in polyester synthesis. Co-based N-doped carbon materials are one of the most promising applied catalysts for oxidative esterification reaction, however, the active sites and reaction pathway of these catalysts have not been clearly clarified, which is crucial to the practical application. Herein, we report that ZIF-67 (a zeolitic imidazolate framework (ZIF)-type cobalt-containing MOF) derived Co@C-N material is a highly effective catalyst for the selective conversion of HMF into DMFDCA in 95% yield. The high activity of the ZIF-67 derived nanocarbon composites Co@C-N can be attributed to the electron transfer between nitrogen-doped carbon shells and Co nanoparticles. The appropriate graphitic N and pyridinic N doping increases the electronic mobility and active sites. Density functional theory (DFT) simulations indicated that oxygen, HMF and methanol molecules are adsorbed and activated on C-N materials. Furthermore, no 2, 5-diformylfuran (DFF) was captured as an intermediate because the oxidative esterification of aldehyde preferentially occurred than the oxidation of hydroxyl group in HMF. We anticipate that these results can drive progress in the bio-based polymers sector and oxidative esterification reaction.

Full text of DOI:10.1016/j.jcat.2019.11.029

Journal of Catalysis 381 (2020) 570–578
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Insights into the active sites and catalytic mechanism of oxidative
esterification of 5-hydroxymethylfurfural by metal-organic
frameworks-derived N-doped carbon
d
e
Yunchao Feng a, Wenlong Jia a, Guihua Yan a, Xianhai Zeng a,b,c, , Jonathan Sperry , Binbin Xu ,
⇑
Yong Sun a,b,c, Xing Tang a,b,c, Tingzhou Lei f, Lu Lin a,b,c
a College of Energy, Xiamen University, Xiamen 361102, China
b Fujian Engineering and Research Center of Clean and High-valued Technologies for Biomass, Xiamen 361102, China
c Xiamen Key Laboratory of Clean and High-valued Utilization for Biomass, Xiamen 361102, China
d Centre for Green Chemical Science, University of Auckland, Auckland 1142, New Zealand
e College of Chemistry and Chemical Engineering, Xiamen University, Xiamen 361102, China
f Henan Key Lab of Biomass Energy, Huayuan Road 29, Zhengzhou, Henan 450008, China
a r t i c l e i n f o
a b s t r a c t
Article history:
Directly oxidative esterification of Biomass-derived 5-hydroxymethylfurfural (HMF) into dimethyl furan
dicarboxylate (DMFDCA) is a promising route for the replacement of petroleum-derived commodity
chemical terephthalic acid (TPA) extensively employed in polyester synthesis. Co-based N-doped carbon
materials are one of the most promising applied catalysts for oxidative esterification reaction, however,
the active sites and reaction pathway of these catalysts have not been clearly clarified, which is crucial to
the practical application. Herein, we report that ZIF-67 (a zeolitic imidazolate framework (ZIF)-type
cobalt-containing MOF) derived Co@C-N material is a highly effective catalyst for the selective conversion
of HMF into DMFDCA in 95% yield. The high activity of the ZIF-67 derived nanocarbon composites Co@C-N
can be attributed to the electron transfer between nitrogen-doped carbon shells and Co nanoparticles. The
appropriate graphitic N and pyridinic N doping increases the electronic mobility and active sites. Density
functional theory (DFT) simulations indicated that oxygen, HMF and methanol molecules are adsorbed
and activated on C-N materials. Furthermore, no 2, 5-diformylfuran (DFF) was captured as an intermediate
because the oxidative esterification of aldehyde preferentially occurred than the oxidation of hydroxyl group
in HMF. We anticipate that these results can drive progress in the bio-based polymers sector and oxidative
esterification reaction.
Received 26 September 2019
Revised 12 November 2019
Accepted 19 November 2019
Keywords:
5-Hydroxymethylfurfural
Dimethyl furan dicarboxylate
Mechanism
Oxidation
Catalysis
Ó 2019 Elsevier Inc. All rights reserved.
1. Introduction
~50% reduction in the greenhouse gas emissions associated with
current PET production process [3]. Despite great advancements
Substituting petroleum-derived terephthalic acid (TPA) with
biomass-derived furan-2,5-dicarboxylic acid (FDCA) in polyester
production is a longstanding goal in the field of renewable polymer
chemistry [1]. The novel polyethylene furandicarboxylate (PEF,
prepared from FDCA or its esters) is reported to have superior
material properties than polyethylene terephthalate (PET, pre-
pared from TPA), including better biodegradability, higher isolation
rate of oxygen, CO2 and water, higher glass transition temperature
and lower melting temperature [2]. Moreover, the production of
PEF from lignocellulosic biomass has been predicted to lead to a
for FDCA production, however, another issue to consider is the
poor solubility of FDCA in most industrial solvents. Purification
by traditional distillation and recrystallization is unfeasible, hin-
dering the large-scale applications of FDCA [4]. A promising alter-
native is the direct oxidative esterification of HMF into dimethyl
furan dicarboxylate (DMFDCA), which is readily soluble in most
common solvents and can be easily purified by sublimation at
low-temperature [5]. Furthermore, the production of PEF from
DMFDCA is easier to proceed than FDCA.
As the conversion of HMF to DMFDCA can be regarded as the
tandem oxidation esterification of a heteroaromatic aldehyde and
alcohol, a series of intermediates including 2,5-diformylfuran
(DFF), 5-hydroxymethyl-2-methyl-furoate (HMMF) and 5-formyl-
2-methyl-furoate (FMF), could be formed (Scheme 1). Indeed,
⇑
Corresponding author at: College of Energy, Xiamen University, Xiamen
361102, China.
E-mail address: xianhai.zeng@xmu.edu.cn (X. Zeng).
https://doi.org/10.1016/j.jcat.2019.11.029
0021-9517/Ó 2019 Elsevier Inc. All rights reserved.

Y. Feng et al. / Journal of Catalysis 381 (2020) 570–578
571
O
O
O
H
H
O
O
O
O
O
O
O
O
OH
H
H
MeO
OMe
MeO
O
O
OH
MeO
Scheme 1. Reaction pathway for the oxidation esterification of HMF to DMFDCA.
significant research efforts have been reported for the oxidation-
esterification of HMF using O2 and methanol. In 2008, Christensen
et al. [4] firstly reported the oxidation esterification of HMF using
Au/TiO2 catalyst, achieving a high DMFDCA yield of 99%. Subse-
quently, a series of Au or other noble metals based catalysts were
employed, providing satisfactory selectivity and yield [6–13]. In
2013, Beller et al. [14] reported that Co3O4-N@C was a highly effec-
tive catalyst for direct oxidative esterification of alcohols. As a
result, unremitting efforts have been devoted to design and syn-
thesis of Co-based N-doped carbon catalysts for the oxidative
esterification of alcohols or HMF [15–21]. However, despite these
achievements, the active sites and the catalytic pathway of these
Co-based catalysts for the oxidative esterification of HMF or alco-
hols have not been clearly clarified.
for 2 h, and then pyrolyzed at desired temperature (600, 700,
800 and 900 °C) for 6 h with a heating rate of 5 °C/min under argon
atmosphere to obtain Co@C-N(x), where ‘‘x” represents the pyrol-
ysis temperature.
Synthesis of Co@C (800): The preparation procedure was as fol-
low: Co(NO3)2ꢀ6H2O (0.58 g) and trimesic acid (0.15 g) were dis-
solved in 35 mL of DMF solvent and stirred for 2 h. The obtained
light-red solution was then transferred to a 50 mL Teflon autoclave
and heated to 120 °C for 24 h. The as-obtained pink precipitates
were centrifuged and washed with MeOH for three times, and
finally dried under vacuum at 80 °C for 24 h, resulting in the Co-
MOF (Co-BTC). Then, after complete pyrolysis at 800 °C (5 °C minꢁ1
)
for 6 h in argon, Co-BTC derived Co@C was obtained.
Recently, metal-organic frameworks (MOFs), a new class of por-
ous crystalline materials, have emerged as a new platform to syn-
thesize nanocarbon composites for applications as catalysts in
electrocatalysis and traditional heterogeneous catalysis [22–25],
because of their ordered structures and relatively low thermal sta-
bility [26–29]. Based on our recent report on ZIF-8 derived cata-
lysts for hydrogenation of HMF [30], here we report a ZIF-67
derived nanoscale Co-based catalysts (Co@C-N) for the oxidation
of HMF to DMFDCA (>99% conversion and 95% yield). ZIF-67 was
selected as a catalyst precursor also because of its abundant carbon
and nitrogen species with zero oxygen content. The detailed exper-
imental analysis, catalysts characterization and DFT simulations
shed light on the active sites and reaction mechanism, which we
anticipate will drive progress in the bio-based polymers sector
and oxidative esterification reaction.
2.3. Catalyst characterization
Powder X-ray diffraction patterns of the samples were recorded
on a Rigaku D/max X-ray diffractometer using Cu K radiation
a
(40 kV, 30 mA, k = 0.1543 nm). Raman spectra were measured
on a Renishaw inVia Raman microscope with 633 nm laser excita-
tion. The laser power was 10 mW, and the spot size was 2 m. BET
surface areas measurements were performed with N2 adsorption/
desorption isotherms using a TriStar 3000 with the BET and HK
methods. Before measurements, the samples were degassed at
100 °C for 12 h. Element contents of the samples were determined
using an Elementar Vario EL III instrument and Inductively Coupled
Plasma Optic Emission Spectrometer (ICP-OES, Varian (720)). The
size and morphology of materials were studied by scanning
electron microscopy (SEM, SUPRA 55) and high-resolution trans-
mission electron microscopy (HR-TEM, JEM-2100). Thermo-
gravimetry (TG) analysis of ZIF-67 were recorded on an SDT
Q600TGA thermal gravimetric analyzer under N2. The magnetic
properties of Co@C-N were measured by a Lake Shore 7404 vibrat-
ing sample magnetometer (VSM).
2. Materials and methods
2.1. Materials
The X-ray photoelectron spectroscopy (XPS) measurements
were performed on a Thermo ESCALAB 250XI electron spectrome-
All reagents were supplied by Aladdin Chemical Technology Co.
Ltd. (Shanghai, China) and used without further purification.
ter with a monochromatic Al K
spot size of all spectra were 650
qualitatively by a Thermo-Fisher Trace 1300 & ISQ LT GC–MS
instrument equipped with a TR-5MS column (15.0 m ꢂ 250 m ꢂ 0.
25 m) and quantified by the Agilent 7890A GC equipped with a
m) and a flame
a
radiation (hv = 1486.8 eV) and the
lm. The DMFDCA was identified
2.2. Synthesis of catalysts
l
Synthesis of ZIF-67: In a typical synthesis, 0.65 g of Co(NO3)2-
l
ꢀ6H2O were dissolved in 15 mL of MeOH/EtOH mixture (1:1) to
DB-WAXetr column (30.0 m ꢂ 0.25 mm ꢂ 0.25
l
form
a clear solution, which was subsequently poured into
ionization detector (FID) that was operated at 270 °C.
15 mL of MeOH/EtOH mixture (1:1) containing 0.74 g of 2-
methylimidazole. After the mixture was stirred for 20 h at room
temperature, the as-obtained purple precipitates were centrifuged
and washed with EtOH for three times, and finally dried under vac-
uum at 80 °C for 24 h, resulting in the Co-based zeolitic imidazo-
late frameworks (ZIF-67).
2.4. Computational method
The binding energies of O2 were performed by using the Vienna
Ab-initio Simulation Package [31,32] (VASP), employing the den-
sity functional theory (DFT) and the Projected Augmented Wave
[33] (PAW) method. The Perdew-Burke-Ernzerhof (PBE) functional
Synthesis of Co@C-N: The preparation procedure was listed as
follow: ZIF-67 was placed in a tube furnace and heated to 300 °C

572
Y. Feng et al. / Journal of Catalysis 381 (2020) 570–578
was used to describe the exchange and correlation effect [34]. In all
the cases, the cut-off energy was set to be 450 eV. The Monkhorst-
Pack grids [35] were set to be 3 ꢂ 3 ꢂ 1 for the all the surface cal-
culations. At least 16 Å vacuum layer was applied in z-direction of
the slab models, preventing the vertical interactions between slabs.
Spin-polarization was contained in all the cases. The Co@NC model
was constructed as one mono-layer N doped graphene staying on
the top of the (111) surface of Co.
carbon, respectively (Fig. 2b) [37]. The IG/ID band intensity ratios
of Co@C-N(600), Co@C-N(700), Co@C-N(800) and Co@C-N(900)
materials are 0.86, 0.97, 1.29 and 1.97, respectively, revealing that
the crystallization degree of graphitic carbon becomes better with
temperature, which was consistent with the XRD observations.
SEM images indicated that the pyrolyzed nanoparticles roughly
retained the polyhedral shape of parent ZIF-67, and shrunk slightly
(Fig. 3b). As shown in Fig. 3(d) and Figure S1, the presence of Co
nanoparticles encapsulated by a few layered carbon shells and
the particle size of Co nanoparticles increased with the pyrolysis
temperature. Furthermore, the highly dispersed Co nanoparticles
of materials were also observed, which was mainly due to the
ordered organic ligands of ZIF-67. Fig. 3(e, f) revealed that the car-
bon matrices and Co nanoparticles of Co@C-N(800) were crys-
talline and the lattice fringes with an inter-planar distance of
0.334 nm and 0.204 nm correspond to the C(002) plane and Co
(111) [38]. The diffraction rings in the SAED image (Fig. 3c) can
be attributed to Co, in good agreement with XRD results.
The adsorption energy of O2 was defined as
D
EB ¼ Eads ꢁ Eslab ꢁ EO2
where Eads is the electronic energy of the slab with an adsorbed O2,
Eslab is the electronic energy of the clean slab, and EO2 is the elec-
tronic energy of gaseous oxygen molecule. Under this definition, a
more negative value indicates a stronger binding system.
2.5. Catalytic oxidation esterification of HMF
In a typical run, HMF (5.0 mmol), catalyst (100 mg), Na2CO3
(30% mol relative to HMF), and methanol (5 mL) were added in a
stainless steel reactor equipped with a magnetic stirrer. The reac-
tion mixture was stirred at 100 °C and at 2 MPa O2. After comple-
tion of the reaction, the catalyst was separated and a sample of the
liquid mixture was subjected to GC analysis. The conversion of 5-
HMF and the yield of products were calculated according to the fol-
lowing equations:
3.2. Catalyst activity and stability
The selective oxidative esterification of HMF to DMFDCA was
performed in the presence of sodium carbonate using methanol
as the solvent at 100 °C under 2 MPa oxygen. The results of
exploratory catalytic experiments with the different catalytic
materials were summarized in Table 1. The parent ZIF-67 showed
no activity (entry 1). Pleasingly, the ZIF-67 pyrolysed materials
were effective and the yields of DMFDCA were disproportionately
affected by the pyrolysis temperature of the ZIF-67 precursors
(entries 2–5). Among the four Co@C-N(T) samples, Co@C-N(800)
was the most active for the HMF conversion, affording DMFDCA
in 99% conversion and 91% yield (entry 4).
ꢀ
ꢁ
Moles of HMF
Moles of HMF loaded
HMF conversion ¼ 1 ꢁ
ꢂ 100%
ꢂ 100%
ꢀ
ꢁ
Moles of product
Product yield ¼ 1 ꢁ
Moles of HMF converted
With the optimum Co@C-N(800) catalyst in hand, the reaction
conditions for the conversion of HMF into DMFDCA were subse-
quently investigated. It was found that the reaction temperature
of 100 °C and oxygen pressure of 2 MPa was optimum in terms
of both HMF conversion and DMFDCA yield (Table 1. entries 6–11).
Next, the effect of different strength bases on the Co@C-N(800)-
catalyzed oxidation of HMF to DMFDCA in methanol was investi-
gated (Fig. 4a). The introduction of weak bases (i.e., NaHCO3,
KHCO3) resulted in a slight increase in DMFDCA yield, while a sig-
nificant increase in yield from 83% to 91% was obtained when using
Na2CO3 and K2CO3. However, stronger bases (i.e., NaOH, KOH,
NaOMe and NaOEt) led to a high HMF conversion but a decrease
in DMFDCA yield, indicating that some undesired byproducts were
formed. The Co@C-N(800) material can efficiently catalyze the
oxidative esterification of HMF, even in the absence of base, to
afford DMFDCA in 83% yield (Figure 4 and S2), which can be attrib-
uted to the strong basic sites of Co@C-N(800) as evidenced with
CO2-TPD (Figure S3). The DMFDCA yield firstly increased and then
decreased with the increasing base concentration, and the
optimized content of Na2CO3 was 30 mol% with respect to HMF
(Figure S4). As shown in Fig. 4b, the yield of DMFDCA reached
83% in only one hour, and such high efficiency have not been
reported to date. A 95% yield of DMFDCA was obtained after a pro-
longed reaction time (12 h). These high conversion and selectivity
achieved by Co@C-N can compare well with those reported noble
metal based catalysts [4,6–13], even in a lower O2 pressure
(Table 1, entry 7). Furthermore, the non-noble metal catalysts are
cost-effective and feasible, especially for large-scale production.
The recyclability of the catalyst was also investigated. After the
reaction, the catalyst was dispersed in ethanol solution and was
easily separated with an external magnet given the magnetic prop-
erties of Co-based material (Fig. 4c). ICP-OES analysis of the reac-
tion solution conformed that the content of Co was below the
detection limit, suggesting no obvious Co leaching during the
3. Results and discussion
3.1. Catalyst preparation and characterization
The two-step synthesis of Co@C-N is shown in Fig. 1a. ZIF-67
was firstly synthesized by the modified room-temperature precip-
itation method [36]. The powder X-ray diffraction (XRD) patterns
of synthetic ZIF-67 (Fig. 1b) matched well with a simulated ana-
logue, confirming the formation of pure ZIF-67 crystals. As shown
in the SEM image (Fig. 3a), synthetic ZIF-67 was composed of
microcrystals with a typical rhombic dodecahedral shape, uniform
morphology and high crystallinity.
The Co@C-N materials were formed by a simple thermal decom-
position of ZIF-67 under an argon atmosphere and metal ions were
reduced in situ. Thermogravimetric (TG) analysis indicated that the
organic ligands in ZIF-67 began to decompose at ca. 500 °C
(Fig. 1c). Therefore, the final set of pyrolyzed temperatures chosen
were 600, 700, 800 and 900 °C, resulting in Co@C-N(T), where T
represents the thermolysis temperature. The corresponding mass
loss is about 28.3–46.6 wt%, which is significantly lower than the
theoretical values in the transformation from ZIF-67 to metallic
Co (73.3%), implying that carbon-containing materials are gener-
ated during the pyrolysis process.
XRD patterns of the Co@C-N materials were shown in Fig. 2a,
the peak at about 26.3° corresponds to a typical (0 0 2) interlayer
of graphite-type carbon sheets, and other peaks at about 44.2°,
51.5°, and 75.8° are attributed to metallic Co (PDF#. 15-0806).
The enhanced peak intensities of the Co diffraction peaks for the
Co@C-N(T) at higher calcination temperatures suggests the forma-
tion of a Co phase with a higher crystallization degree. The Raman
spectra of the samples also revealed characteristic carbon G and D
bands, corresponding to the graphitic sp2-carbon and disoriented

Y. Feng et al. / Journal of Catalysis 381 (2020) 570–578
573
Fig. 1. (a) Synthetic scheme for the preparation of Co@C-N, (b) XRD pattern of ZIF-67, (c) TG curve of ZIF-67 in N2.
Fig. 2. (a) XRD patterns of (1) Co@C-N(600), (2) Co@C-N(700), (3) Co@C-N(800), and (4) Co@C-N(900); (b) Raman spectra of Co@C-N.
reaction. The isolated catalyst was washed several times with etha-
nol and subsequently reused in HMF conversion to DMFDCA under
identical reaction conditions. These results showed that the activ-
ity of the reused catalyst was significantly decreased. Fortunately,
after being reduced with H2 at 400 °C for 1 h, the catalyst could be
reused for up to five runs without any apparent loss in efficiency
(Fig. 4d). The XRD patterns indicated that the Co diffraction inten-
sity of the reused catalyst (without reduction in H2) became much
weaker (Fig S5), suggesting a partial oxidation of metallic Co
occured during the oxidation reaction. After reduction in H2 at
400 °C for 1 h, the XRD profile was similar to the freshly prepared
catalyst (Figure S5), which led to high reactivity of the regenerated
Co@C-N(800) material.
3.3. Mechanism study on catalyst activity
To verify the nature of the active sites in Co@C-N(T), a series of
Co or carbon-based samples were prepared for the aerobic oxida-
tive esterification of HMF under identical conditions (Table 2). In
the absence of any catalyst, HMF conversion was 17% and DMFDCA
yield was less than 1% (entry 1). A similar result was obtained
when using activated carbon as catalyst, demonstrating that there
was no obvious catalytic activity with activated carbon or without
the catalyst (entry 2). Note that the C-N composite, synthesized by
dipping Co@C-N(800) into aqua regia for 48 h to remove Co
(Figure S5), exhibited some activity in the conversion of HMF to
DMFDCA (entry 3). Subsequently, the Co@C-N(800) was treated

574
Y. Feng et al. / Journal of Catalysis 381 (2020) 570–578
Fig. 3. (a) SEM image of ZIF-67, (b) SEM images of Co@C-N(800), (c) SAED pattern of Co@C-N(800), and (d-f) HRTEM images of Co@C-N(800). Arrows in d and e indicate the
direction of the graphitic layers.
Table 1
Conversion of HMF to DMFDCA by Co@C-N.a
Entry
Catalyst
Temperature(°C)
O2 (MPa)
Con.(%)
Sel.b
(%)
Yield (%)
DMFDCA
HMMF
FMF
1
2
3
4
ZIF-67
100
100
100
100
100
100
100
100
100
80
2
2
2
2
Co@C-N(600)
Co@C-N(700)
Co@C-N(800)
Co@C-N(900)
Co@C-N(800)
Co@C-N(800)
Co@C-N(800)
Co@C-N(800)
Co@C-N(800)
Co@C-N(800)
92
96
99
88
54
91
97
99
88
75
93
86
98
90
<4
98
98
97
94
89
70
82
91
52
<1
77
85
88
60
45
13
7
5
23
<1
11
8
6
19
13
3
1
1
4
2
1
2
2
4
9
5
2
6c
7d
8
0 (air)
0.1
1
3
2
9
10
11
120
2
a
b
c
Condition: HMF (0.5 mmol), Catalyst (100 mg), Na2CO3 (30 mol% relative to HMF), methanol (5 mL), 5 h.
Sel. = Yield(DMFDCA + HMMF + FMF)/Con.
42% of DFF was detected as by-product.
12h.
d
at 400 °C in air for 1 h and then reduced under H2 for another hour
to obtain metallic Co (Figure S6), resulting in a negligible DMFDCA
yield (entry 4). The physical mixture of metallic Co and activated
To elucidate how the Co nanoparticles work in coordination
with N-doping Carbon for the selective oxidation of HMF, the
chemical composition and structural features were further charac-
terized. The nitrogen mass contents in Co@C-N(600), Co@C-N(700),
Co@C-N(800) and Co@C-N(900) are 9.7, 5.8, 2.1 and 0.6 wt%,
respectively (Table S1). Two high-resolution N1s peaks of
Co@C-N(T) suggest the presence of two types of nitrogen species,
pyridinic N at ~399 eV and graphitic N at ~401 eV (Fig. 5a and e)
[19]. The results showed that the relative contents of graphitic N
increased with increasing thermolysis temperature, while that of
pyridinic N decreased (Fig. 5b), implying a higher graphitization
degree at higher temperatures, which are beneficial for high
carbon also showed poor activity (entry 5). Interestingly,
a
significantly enhanced yield was obtained when mixed with C-N
composite (entry 6), suggesting an absolutely essential role of
N-doping. For comparison, Co-MOF (Co-BTC) without N was pyro-
lyzed at 800 °C to generate Co@C(800). However, a relatively low
DMFDCA yield of 18% was obtained (entry 7). These control exper-
iments demonstrated the important synergic interactions between
Co and C-N composite in the activity of Co@C-N(800) in the oxida-
tive esterification reaction.

Y. Feng et al. / Journal of Catalysis 381 (2020) 570–578
575
Fig. 4. (a) Conversion of HMF to DMFDCA upon addition of various strength bases. Conditions HMF, (0.5 mmol), catalyst (100 mg), base (30 mol%, relative to HMF), 5 mL
methanol, 2 MPa O2, 100 °C, 5 h; (b) Conversion of HMF to DMFDCA by Co@C-N(800). Conditions: HMF (0.5 mmol), catalyst (100 mg), Na2CO3 (30 mol%, relative to HMF), 5 mL
methanol, 2 MPa O2, 100 °C; (c) Magnetic separation of the catalyst after reaction; (d) Reuses of the Co@C-N(800) catalyst for HMF oxidation.
Table 2
to a higher value than that of Co nanoparticles (778.1) as well as
the formation of Co–C bond further prove the electron transfer
The oxidative esterification of HMF in the presence of various catalysts.a
from metal nanoparticles to C-N composites (Fig. 5f).
Entry
Catalyst
Conversion %
DMFDCA yield %
Based on the aforementioned analysis, the most active Co@C-N
(800) material showed an appropriate nitrogen content and a rel-
atively high ratio of graphitic-N/C, contributing to its balance
between the contents and species of nitrogen (Fig. 5d). This bal-
ance can result in higher electronic mobility and more active sites
[42], which is beneficial for activation and reduction of O2. The
strong electronic interaction between Co atoms and N-doping car-
bon is prone to generate strong electron acceptors (Co nanoparti-
cles) and electron donors (C-N composites), leading to the high
catalytic activity of Co@C-N(800) in the oxidative esterification
reaction. As previously reported, both of the electron-poor metal
nanoparticles and the electron-rich carbon shells should attract
O2 [43–46] which results in unclear O2 adsorption sites. Thus, a
series of DFT simulations were performed to reveal the O2 adsorp-
tion sites on Co@C-N(800). The Co@NC model was constructed as
one mono-layer N doped graphene staying on the top of the
(111) surface of Co. As illustrated in Fig. 6a, O2 adsorption on Co
sites resulted in the O–O bond cleavage. This indicated that O2 can-
not be adsorbed on Co sites, which also resulted in high adsorption
energy of O2 (3.74 eV). However, the O–O linkage was well main-
tained since O2 was adsorbed on C–N sites (Fig. 6b) and the low
binding energy (ꢁ0.05 eV) further demonstrated that the oxygen
molecule was adsorbed and activated on the C–N materials.
It can also be assumed from the simulation results that the
–CHO and –OH group of HMF are activated at C–N sites because
1
2
3
4
5
6
7
–
C
C-N
Co
Co + C
Co + C-N
Co@C(800)
21
25
85
38
28
86
60
<1
<1
12
<1
2
33
18
a
Condition: Catalyst (100 mg), HMF (0.5 mmol), Na2CO3 (30 mol% relative to
HMF), methanol (5 mL), 2 MPa O2, 100 °C, 5 h.
electronic mobility. The result is in good agreement with the XRD
and Raman observations. In addition, the pyridinic N doping pro-
duces structural defects in C-N composites to form the O2
adsorption sites and basic sites. Moreover, the Co nanoparticles
encapsulated by carbon shells are believed to affect the properties
of outer C-N composites. It has been previous reported that the
strong interaction between metal nanoparticles and the graphitic
walls can enhance activity of nanocarbon catalysts owing to the
electron transfer from metal atoms to carbon shells until their
Fermi level reaches equilibrium [39,40]. It is beyond doubt that
N-doping facilitates the electronic interaction with nearby car-
bon/metal atoms to accelerate the formation of the oxygen radical.
The Co 2p3/2 spectrum exhibits two prominent bands at ~778.7 eV
and ~781.0 eV, readily assigned to Co–Co and Co–C bond respec-
tively (Fig. 5c) [41]. The phenomena that Co–Co bond peak shifts

576
Y. Feng et al. / Journal of Catalysis 381 (2020) 570–578
Fig. 5. (a) High-resolution N 1s XPS spectra of Co@C-N (600–900). (b) The relative contents of N species of Co@C-N (600–900). (c) High-resolution Co 2p3/2 XPS spectra of
Co@C-N (600–900). (d) N content and the corresponding DMFDCA yield for Co@C-N (600–900), (e) Possible bonding configurations of N-doped carbon sheels of Co@C-N(800).
(f) Schematic illustration of electron transfer between Co nanoparticles and N-doped carbon materials.
Fig. 6. Schematic model of (a) O2 adsorption on Co sites, (b) O2 adsorption on N-doped carbon sites, (c)–OH (HMF) adsorption on C-N sites and (d) –CHO (HMF) adsorption on
C-N siteThe brown, gray, blue, and red balls represent C, N, Co and O atoms, respectively.
the size of HMF molecular is larger than O2 molecualr. As shown
in Fig. 6(c, d), the low adsorption energies supported the
hypothesis that HMF is activated at C–N sites. On the basis of
the conversion-time profiles (Fig. 4b, S2), almost no DFF was
detected (Figure S7a). The factor of selective and preferential
adsorption for alcohol or aldehyde group in HMF should be
excluded as the closely calculated adsorption energy between
the –CHO (–1.9 eV) and –OH (–1.97 eV) group. Note that
hemiacetal reactions of the aldehyde group can be easily pro-
ceeded in alcoholic solvents. Thus, this result was attributed to

Y. Feng et al. / Journal of Catalysis 381 (2020) 570–578
577
Fig. 7. Plausible mechanism for the conversion HMF to DMFDCA by Co@C-N. (a) Adsorption of O2 and formation of superoxide anion radical; (b) Superoxide anion radical
reacts with alcohol; (c) b-hydride elimination; (d) Oxidation of metal–hydride.
the fact that the oxidative esterification of aldehyde was prefer-
entially occurred than the oxidation of hydroxyl groups of HMF
(Figure S7b).
reactivity, which can be attributed to the balance of the contents
and species of nitrogen. DFT simulations concluded that oxygen
molecular and HMF were adsorbed and activated on C-N sheels.
No DFF was observed indicating that the oxidative esterification
of aldehyde was preferentially proceed than the oxidation of
hydroxyl group of HMF. Most importantly, the detail studied
Co@C-N catalyzed oxidative esterification mechanism could pro-
mote a devolpement of oxidative esterification reaction, especially
for bio-based polymers sector.
When di-tert-butyl-4-methylphenol (BHT, a free-radical scav-
enger) was introduced, the yield of DMFDCA decreased
(Figure S8a), suggesting that oxygen radical(s) were formed after
O2 adsorption onto the C-N sites. The oxygen radical(s) was(were)
further checked by Electron Paramagnetic Resonance (EPR) as
shown in Figure S8b. These peaks were typical EPR signals of
DMPO-ꢃO2H adducts, demonstrating the of superoxide anion radi-
cal (ꢃO2ꢁ) was formed in our catalytic system. It’s worth noted that
superoxide anion radical was also produced when using CoxOy@C-N
Declaration of Competing Interest
as catalyst alone [12], but mixing with Ru/C, K-OMS-2 or
a-MnO2
The authors declare that they have no known competing finan-
cial interests or personal relationships that could have appeared
to influence the work reported in this paper.
were required to achieve high conversion [12,17,18], indicating an
superior role Co@C-N in the oxidative esterification reaction.
On the basis of the reaction process and above analysis, a pos-
sible reaction mechanism for the oxidation of HMF to DMFDCA is
presented in Fig. 7. First, oxygen molecules are adsorbed and acti-
vated at the electron-rich carbon sites adjacent to the pyridinic N
(where more structural defects exist), accepting electrons to pro-
duce superoxide anion radical. At the same time, aldehyde/alcohol
(HMF) can be attracted on C-N sites. Then, the superoxide anion
radical reacts with alcohol (MeOH/HMF), producing the corre-
sponding ester/aldehyde by a b-hydride elimination (oxidation of
aldehyde to ester via a hemiacetal coordinated intermediate). It
is well established that the b-hydride elimination is commonly
associated with the formation of H-metal species on the catalyst
surface [47–49], so the introduction of an appropriate base can
facilitate deprotonation of the H-metal species to regenerate the
active catalyst surface.
Acknowledgements
This work was supported by the National Natural Science Foun-
dation of China (No. 21978248), the Natural Science Foundation of
Fujian Province of China (No. 2019J06005), and the Energy Devel-
opment Foundation of the College of Energy, Xiamen University
(No. 2017NYFZ02). We also thank the New Zealand Ministry of
Business, Innovation and Employment (MBIE) for support through
the Catalyst Fund (16-UOA-049-CSG).
Appendix A. Supplementary material
Supplementary data to this article can be found online at
https://doi.org/10.1016/j.jcat.2019.11.029.
References
4. Conclusions
[1] T.A. Werpy, J.E. Holladay, J.F. White, Synthetic Fuels, 2004.
[2] J. Zhu, J. Cai, W. Xie, P.H. Chen, M. Gazzano, M. Scandola, R.A. Gross,
Macromolecules 46 (2013) 796–804.
[3] A.J.J.E. Eerhart, A.P.C. Faaij, M.K. Patel, Energy Environ. Sci. 5 (2012) 6407–
6422.
[4] E. Taarning, I.S. Nielsen, K. Egeblad, R. Madsen, C.H. Christensen,
ChemSusChem 1 (2008) 75–78.
[5] G.Z. Papageorgiou, D.G. Papageorgiou, Z. Terzopoulou, D.N. Bikiaris, Eur. Polym.
J. 83 (2016) 202–229.
In summary, starting with Co-MOF (ZIF-67) particles as the sin-
gle template, we prepared Co@C-N materials that are cost effective,
effective, stable and reusable catalysts for the oxidative esterifica-
tion of HMF to DMFDCA. The synthetic Co@C-N composites exhib-
ited excellent activity and selectivity for the DMFDCA production, a
result of the nitrogen doping and the electron transfer between Co
and C-N composite. Among the four Co@C-N samples synthesized
at different temperatures, Co@C-N(800) showed the highest
[6] F. Menegazzo, M. Signoretto, D. Marchese, F. Pinna, M. Manzoli, J. Catal. 326
(2015) 1–8.

578
Y. Feng et al. / Journal of Catalysis 381 (2020) 570–578
[7] F. Li, X.-L. Li, C. Li, J. Shi, Y. Fu, Green Chem. 20 (2018) 3050–3058.
[8] J. Du, H. Fang, H. Qu, J. Zhang, X. Duan, Y. Yuan, Appl. Catal. A 567 (2018) 80–
89.
[29] B. Liu, H. Shioyama, T. Akita, Q. Xu, J. Am. Chem. Soc. 130 (2008) 5390–5391.
[30] Y. Feng, G. Yan, T. Wang, W. Jia, X. Zeng, J. Sperry, Y. Sun, X. Tang, T. Lei, L. Lin,
Green Chem. 21 (2019) 4319–4323.
[9] O. Casanova, S. Iborra, A. Corma, J. Catal. 265 (2009) 109–116.
[10] A. Cho, S. Byun, J.H. Cho, B.M. Kim, ChemSusShem 12 (2019) 2310–2317.
[11] A. Buonerba, S. Impemba, A.D. Litta, C. Capacchione, S. Milione, A. Grassi,
ChemSusChem 11 (2018) 3139–3149.
[12] A. Salazar, P. Hünemörder, J. Rabeah, A. Quade, R.V. Jagadeesh, E. Mejia, ACS.
Sustain. Chem. Eng. 7 (2019) 12061–12068.
[13] D.S. Mannel, J. King, Y. Preger, M.S. Ahmed, T.W. Root, S.S. Stahl, ACS Catal. 8
(2018) 1038–1047.
[14] R.V. Jagadeesh, H. Junge, M.-M. Pohl, J.r. Radnik, A. Brückner, M. Beller, J. Am.
Chem. Soc. 2013, 135, 10776–10782.
[31] G. Kresse, J. Furthmüller, Phys. Rev. B 54 (1996) 11169–11186.
[32] G. Kresse, J. Hafner, Phys. Rev. B 49 (1994) 14251–14269.
[33] P.E. Blöchl, Phys. Rev. B 50 (1994) 17953–17979.
[34] J.P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 77 (1996) 3865–3868.
[35] H.J. Monkhorst, J.D. Pack, Phys. Rev. B 13 (1976) 5188.
[36] Y. Lu, Y. Wang, H. Li, Y. Lin, Z. Jiang, Z. Xie, Q. Kuang, L. Zheng, A.C.S. Appl,
Mater. Interfaces 7 (2015) 13604–13611.
[37] M. Zhang, Q. Dai, H. Zheng, M. Chen, L. Dai, Adv. Mater. (2018) 30.
[38] W. Feng, Y. Wang, J. Chen, B. Li, L. Guo, J. Ouyang, D. Jia, Y. Zhou, J. Mater. Chem.
C 6 (2018) 10–18.
[15] C. Bai, A. Li, X. Yao, H. Liu, Y. Li, Green Chem. 18 (2016) 1061–1069.
[16] K.-K. Sun, S.-J. Chen, Z.-L. Li, G.-P. Lu, C. Cai, Green Chem. 21 (2019) 1602–1608.
[17] J. Deng, H.J. Song, M.S. Cui, Y.P. Du, Y. Fu, ChemSusChem 7 (2014) 3334–3340.
[18] Y. Sun, H. Ma, X. Jia, J. Ma, Y. Luo, J. Gao, J. Xu, ChemCatChem 8 (2016) 2907–
2911.
[19] W. Zhong, H. Liu, C. Bai, S. Liao, Y. Li, ACS Catal. 5 (2015) 1850–1856.
[20] Q. Zhu, F. Wang, F. Zhang, Z. Dong, Nanoscale 11 (2019) 17736–17745.
[21] H. Zhou, S. Hong, H. Zhang, Y. Chen, H. Xu, X. Wang, Z. Jiang, S. Chen, Y. Liu,
Appl. Catal., B (2019) 117767.
[22] K. Shen, X. Chen, J. Chen, Y. Li, ACS Catal. 6 (2016) 5887–5903.
[23] K. Murugesan, T. Senthamarai, M. Sohail, A.S. Alshammari, M.M. Pohl, M.
Beller, R.V. Jagadeesh, Chem. Sci. 9 (2018) 8553–8560.
[24] R.V. Jagadeesh, K. Murugesan, A.S. Alshammari, H. Neumann, M.-M. Pohl, J.
Radnik, M. Beller, Science 358 (2017) 326–332.
[39] D. Deng, L. Yu, X. Chen, G. Wang, L. Jin, X. Pan, J. Deng, G. Sun, X. Bao, Angew.
Chem. Int. Ed. 52 (2013) 371–375.
[40] G. Wu, K.L. More, C.M. Johnston, P. Zelenay, Science 332 (2011) 443–447.
[41] J. Meng, C. Niu, L. Xu, J. Li, X. Liu, X. Wang, Y. Wu, X. Xu, W. Chen, Q. Li, Z. Zhu,
D. Zhao, L. Mai, J. Am. Chem. Soc. 139 (2017) 8212–8221.
[42] C. Hu, L. Dai, Angew. Chem. Int. Ed. 55 (2016) 11736–11758.
[43] L. Zhang, A. Wang, W. Wang, Y. Huang, X. Liu, S. Miao, J. Liu, T. Zhang, ACS
Catal. 5 (2015) 6563–6572.
[44] H. Watanabe, S. Asano, S.-I. Fujita, H. Yoshida, M. Arai, ACS Catal. 5 (2015)
2886–2894.
[45] H. Chen, K. Shen, Q. Mao, J. Chen, Y. Li, ACS Catal. 8 (2018) 1417–1426.
[46] H. Su, K.X. Zhang, B. Zhang, H.H. Wang, Q.Y. Yu, X.H. Li, M. Antonietti, J.S. Chen,
J. Am. Chem. Soc. 139 (2017) 811–818.
[47] M. Huang, Y. Li, Y. Li, J. Liu, S. Shu, Y. Liu, Z. Ke, Chem. Commun. 55 (2019)
6213–6216.
[48] S. Sarina, S. Bai, Y. Huang, C. Chen, J. Jia, E. Jaatinen, G.A. Ayoko, Z. Bao, H. Zhu,
Green Chem. 16 (2014) 331–341.
[25] X. Ma, Y.X. Zhou, H. Liu, Y. Li, H.L. Jiang, Chem. Commun. 52 (2016) 7719–7722.
[26] A.J. Howarth, Y. Liu, P. Li, Z. Li, T.C. Wang, J.T. Hupp, O.K. Farha, Nat. Rev. Mater.
1 (2016) 15018.
[27] J.K. Sun, Q. Xu, Energy Environ. Sci. 7 (2014) 2071–2100.
[28] O.M. Yaghi, M. O’Keeffe, N.W. Ockwig, H.K. Chae, M. Eddaoudi, J. Kim, Nature
423 (2003) 705.
[49] S.E. Davis, M.S. Ide, R.J. Davis, Green Chem. 15 (2013) 17–45.