Identification of the dehydration active sites in glycerol hydrogenolysis to 1,2-propanediol over Cu/SiO2 catalysts

Article abstract of DOI:10.1016/j.jcat.2019.12.032

Metal–support interaction is a hot topic in catalysis, but attention has seldom been drawn to the promotional effect of metal–irreducible SiO₂ interfaces. Here, Cu/SBA-15 catalysts with Cu[sbnd]O[sbnd]Si[sbnd]O[sbnd] interface structures were prepared by simple grinding, and showed high activity and selectivity to 1,2-propanediol above 97% in liquid-phase glycerol hydrogenolysis. Glycerol conversion of ground Cu/SBA-15 with 10% Cu loading was about seven times that of its impregnation counterpart. The linear relationship between turnover frequency and (Cu[sbnd]O[sbnd]Si[sbnd]O[sbnd] induced Lewis acid sites)/Cu⁰ ratio indicated that Cu[sbnd]O[sbnd]Si[sbnd]O[sbnd] structures were crucial for achieving excellent hydrogenolysis performance. More importantly, in situ Fourier transform infrared spectroscopy of glycerol adsorption and density functional theory calculation results confirmed that Cu[sbnd]O[sbnd]Si[sbnd]O[sbnd] species were the dominant active dehydration sites of glycerol hydrogenolysis, while the adjacent Cu sites were involved in subsequent hydrogenation of the generated hydroxyacetone.

Full text of DOI:10.1016/j.jcat.2019.12.032

Journal of Catalysis 383 (2020) 13–23
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Identification of the dehydration active sites in glycerol hydrogenolysis
to 1,2-propanediol over Cu/SiO catalysts
2
Jianfeng Shan a,b,c,1, Huan Liu a,c,1, Kuan Lu a,c,d, Shanhui Zhu a, , Junfen Li , Jianguo Wang a,c, Weibin Fan a,
⇑
a
⇑
a
State Key Laboratory of Coal Conversion, Institute of Coal Chemistry, Chinese Academy of Sciences, 27 South Taoyuan Road, Taiyuan 030001, People’s Republic of China
College of Chemistry and Material Science, Shandong Agricultural University, 61 Daizong Road, Tai’an 271018, People’s Republic of China
b
c
University of Chinese Academy of Sciences, Beijing 100049, People’s Republic of China
d
National Energy Center for Coal to Clean Fuels, Synfuels China Co., Huairou District, Beijing 101400, People’s Republic of China
a r t i c l e i n f o
a b s t r a c t
Article history:
Metal–support interaction is a hot topic in catalysis, but attention has seldom been drawn to the promo-
tional effect of metal–irreducible SiO interfaces. Here, Cu/SBA-15 catalysts with CuAOASiAOA interface
Received 30 July 2019
Revised 18 December 2019
Accepted 18 December 2019
2
structures were prepared by simple grinding, and showed high activity and selectivity to 1,2-propanediol
above 97% in liquid-phase glycerol hydrogenolysis. Glycerol conversion of ground Cu/SBA-15 with 10% Cu
loading was about seven times that of its impregnation counterpart. The linear relationship between
turnover frequency and (CuAOASiAOA induced Lewis acid sites)/Cu0 ratio indicated that
CuAOASiAOA structures were crucial for achieving excellent hydrogenolysis performance. More impor-
tantly, in situ Fourier transform infrared spectroscopy of glycerol adsorption and density functional the-
ory calculation results confirmed that CuAOASiAOA species were the dominant active dehydration sites
of glycerol hydrogenolysis, while the adjacent Cu sites were involved in subsequent hydrogenation of the
generated hydroxyacetone.
Keywords:
Glycerol
Hydrogenolysis
1
,2-Propanediol
Cu/SBA-15
Interface
Ó 2019 Elsevier Inc. All rights reserved.
1
. Introduction
Utilization of biomass-derived feedstock provides a way to
and Ni) has been studied extensively [2,5,8–21]. Among them,
copper-based catalysts have been demonstrated to be the best can-
didates for their poor ability to cleave CAC bonds, while giving
reduce consumption of diminishing fuel and benefit the global cli-
mate by reducing greenhouse gas emissions [1,2]. Recently, biodie-
sel has emerged as a promising substitute for conventional fossil
fuels. It is well known that glycerol accounts for approximately
excellent 1,2-PDO selectivity [3,6,22]. Various metallic/nonmetallic
2 3 2
oxides, such as chromite, Al O , MgO, MgAlO, AlOOH, ZrO , ZnO,
and SiO , have been employed as supports for Cu-based catalysts
2
for hydrogenolysis of glycerol to 1,2-PDO [23–32]. A key question
that remains is the exact dehydration active site for the first step
of glycerol hydrogenolysis (Scheme 1). Although glycerol dehydra-
tion is generally accepted to be catalyzed by acid or base supports,
some disparate opinions still exist. For instance, Sun et al. [2] spec-
ulated that the carriers might not be able to catalyze glycerol dehy-
dration to hydroxyacetone and behaved just as inert supports for
dispersion of Cu after extensive literature investigation. They
deduced that metallic Cu not only worked as an active hydrogena-
tion center, but also involved glycerol dehydration. Especially,
Raney Cu has been established to catalyze glycerol dehydration
to hydroxyacetone [33]. Moreover, it is hard to produce hydroxy-
acetone selectively from glycerol dehydration over supports such
1
0 wt% of the total product in the biodiesel process. Although the
production of glycerol is growing rapidly, its market condition is
poor [3–5]. Therefore, transformation of glycerol into value-
added products is highly desirable for increasing profits in the
biodiesel industry. 1,2-Propanediol (1,2-PDO) is widely used in
the synthesis of unsaturated polyester resins, pharmaceuticals,
cosmetics, and so on [6,7]. Currently, 1,2-PDO is produced
from propylene oxide that comes from selective oxidation of
petroleum-derived propylene. Thus, the hydrogenolysis of glycerol
to 1,2-PDO is a promising route for saving fossil fuel and upgrading
glycerol utilization.
During the past decade, the hydrogenolysis of glycerol to 1,2-
PDO over different metal catalysts (e.g., Pt, Ru, Ag, Rh, Ir, Cu, Co
as Al
glycerol conversion of 61% with 93.3% 1,2-PDO selectivity, while
pure Al gives a conversion of 7% without forming hydroxyace-
2 3 2 2 2 3
O , ZrO , and TiO [33,34]. For example, Cu/Al O exhibits
2 3
O
⇑
Corresponding author.
E-mail addresses: zhushanhui@sxicc.ac.cn (S. Zhu), fanwb@sxicc.ac.cn (W. Fan).
These authors contributed equally to this work.
tone under identical reaction conditions [23]. Nevertheless, up to
now, no direct experimental evidence has been obtained to
1
https://doi.org/10.1016/j.jcat.2019.12.032
021-9517/Ó 2019 Elsevier Inc. All rights reserved.
0

1
4
J. Shan et al. / Journal of Catalysis 383 (2020) 13–23
Scheme 1. Reaction route of glycerol hydrogenolysis into 1,2-PDO [2].
illustrate the exact roles of Cu and supports in glycerol dehydra-
tion. In this context, further research is necessary.
In situ diffuse reflectance Fourier transform infrared spectra
(DRIFTS) were collected on a Nicolet iS10 FTIR spectrophotometer
equipped with a SMART collector and an MCT detector cooled with
In the present work, excellent Cu/SBA-15 catalysts were pre-
pared by a simple grinding method, and these catalysts were
employed to identify the exact dehydration active site with some
other catalysts as comparisons. It was found that the interfacial
CuAOASiAOA structure played a decisive role in the dehydration
process, which was confirmed by both experiments and density
functional theory (DFT) calculation results. Copper species evolu-
tion in catalyst preparation processes was thoroughly investigated
for the first time on Cu/SBA-15 by a grinding method.
2
liquid N . Typically, a certain amount of KBr was first placed into
the sample cell as background. Then about 5 mg of dried catalyst
was added. The sample was heated to 693 K at a ramp rate of
ꢁ1
ꢁ1
2 K min in an air flow (15 mL min ), and the DRIFTS spectra
were simultaneously recorded.
The reduction behavior of catalysts was investigated by H
temperature-programmed reduction (H -TPR) conducted in a
2
2
Micromeritics AutoChem II 2920 chemisorption analyzer. Typi-
cally, a 50-mg sample was loaded into a U-type quartz tube and
heated at 593 K for 30 min in flowing Ar. After the sample was
cooled to room temperature, the pretreated catalyst was heated
2
. Experimental
ꢁ
1
again to 973 K at a rate of 5 K min in a 10 vol% H
(
2
/Ar flow
30 mL min ) and the consumed hydrogen concentration was
monitored by a TCD at the same time.
O chemisorption was performed to determine copper specific
surface area using the same apparatus as for H TPR. First, 30 mg of
catalyst was prereduced at 593 K for 2 h in a 10 vol% H /Ar flow
15 mL min ). Then it was immediately cooled to room tempera-
ture. Subsequently, the surface copper of the catalyst was oxidized
2.1. Catalyst preparation
ꢁ1
The template-free SBA-15 was purchased from XF Nano Materi-
als Tech. Nanjing Co. Ltd. xCu/SBA-15(G) catalysts (x represents
nominal Cu mass percentage) were obtained by grinding SBA-15
N
2
2
2
with a certain amount of Cu(NO
3
)
2
ꢀ3H
2
O at room temperature for
ꢁ
1
(
1
h in a planetary ball mill and subsequently calcined at 773 K
for 5 h in an air flow. For comparison, 10%Cu/SBA-15(IM) was pre-
pared by an incipient wetness impregnation method. The impreg-
nated sample was dried at 373 K and calcined under the same
conditions as those used for calcining xCu/SBA-15(G).
ꢁ1
to Cu
the catalyst was reduced by a 10 vol% H
with the temperature increasing from room temperature to
93 K, and the consumed hydrogen amount (HCu2O) was moni-
tored. To determine the total hydrogen consumption, the sample
was first oxidized to CuO at 573 K for 1 h in 1 vol% O /Ar
15 mL/min). After that, it was heated from room temperature to
2 2
O in 5 vol% N O/Ar (15 mL min ) at 323 K for 20 min. Then
ꢁ
1
2
/Ar flow (15 mL min )
4
2
Cu@SiO was fabricated using the modified method reported by
Zheng [35]. First, Cu powder (Alfa Aesar) was calcined at 573 K in
air. After that, 0.3 g of the obtained powder and 2 g of cetyltri-
methyl ammonium bromide (CTAB) were added to 300 mL deion-
ized water. Then, 0.6 g tetraethyl orthosilicate (TEOS) and 5 mL
ethanol were added to the mixture and it was stirred for 1 h. The
solid products were then separated, washed, and dried, followed
by calcination at 773 K for 5 h.
2
(
ꢁ1
1
173 K at a rate of 10 K/min in a 10 vol% H
and the corresponding hydrogen consumption amount (H
2
/Ar flow (15 mL min ),
) was
T
detected. The specific surface area and dispersion of copper species
can be calculated by the equations [36,37]
ꢀ
ꢁ
1
9
S
Cu ¼ 2 ꢂ HCu2O ꢂ NAv= H
T
ꢂ MCu ꢂ 1:4 ꢂ 10
2.2. Characterization
À
Á
2
¼
1353 ꢂ HCu2O=H
T
m =g
;
Cu
The metal loadings in the catalysts were determined by induc-
tively coupled plasma atomic emission spectroscopy (ICP-AES,
Thermo iCAP6300).
D
Cu ¼ ð2 ꢂ HCu2O=H
T
Þ ꢂ 100%;
1
9
N
2
physisorption measurements were performed at 77 K on a
where NAv = Avogadro’s constant, MCu = 63.46 g/mol, and 1.4 ꢂ 10
19
TriStar3020 II gas adsorption analyzer. Prior to the measurement,
all the samples were treated at 593 K for 10 h under high vacuum
to remove water and other physically adsorbed impurities.
The powder X-ray diffraction (XRD) patterns of xCu/SBA-15(G)
and 10%Cu/SBA-15(IM) were recorded on a Rigaku MiniFlex II X-
refers to the 1.4 ꢂ 10 copper atoms per square meter, because the
2
average surface area of a copper atom is 0.0711 nm .
The Cu nanoparticle images were measured by transmission
electron microscopy (TEM, JEM-2001F, JEOL, Japan) at 200 kV.
The X-ray photoelectron spectra (XPS) and X-ray induced Auger
electron spectra (XAES) were measured on a Kratos AXIS ULTRA
ray diffractometer with CuK
scanning speed of 4°min . The samples reduced at 593 K for 2 h
a
radiation (30 kV and 15 mA) at a
ꢁ1
DLD spectrometer equipped with monochromated AlK
and a multichannel detector. Notably, reduced samples were first
pretreated in a 10% H /Ar flow at different temperatures for 2 h
a radiation
2
under 10 vol% H /Ar\ were protected with liquid paraffin for XRD
measurement.
2
Temperature-programmed desorption of ammonia (NH
was carried out in a Micromeritics Autochem II 2920 chemisorp-
tion analyzer equipped with a thermal conductivity detector
3
-TPD)
in an auxiliary pretreatment chamber before XPS measurements.
Then they were transferred to the XPS measurement chamber
under high vacuum conditions without air exposure. The collected
binding energy values were referenced to the C1s peak at 284.6 eV
with an experimental error of ±0.1 eV.
(
TCD). Before adsorption, an approximately 0.05-g sample was pre-
ꢁ1
2
reduced at 593 K for 2 h in 10 vol% H /Ar (30 mL min ) and then
cooled to 393 K. Subsequently, the sample was purged with an
NH /He mixture (1:9 vol ratio) for 30 min. After that, the physi-
sorbed NH was removed by flushing the sample with pure helium
Fourier transform infrared spectra of pyridine adsorbing (Py-IR)
samples were collected on a Bruker Tensor 27 FT-IR spectrometer.
First, the sample was pressed into a self-supported wafer and
3
3
ꢁ
1
(
30 mL min ) at 393 K for 2 h. Finally, the sample was heated from
placed in a vacuum cell. Prior to the measurement, the sample
ꢁ1
ꢁ1
3
93 to 773 K at a rate of 10 K min
.
wafer was reduced by 10 vol% H
2
/Ar (50 mL min ) at 593 K for

J. Shan et al. / Journal of Catalysis 383 (2020) 13–23
15
2
h, followed by evacuation at 10ꢁ2 Pa for 2 h. Finally, pyridine
moles of unreacted glycerol
moles of glycerol in the feed
Glycerol conversion ¼ ð1 ꢁ
Þ
vapor was introduced in the sample cell at 308 K for 1 h and the
spectra were recorded after evacuation at 423 K for 15 min.
ꢂ 100%
2
.3. In situ Fourier transform infrared experiments of glycerol
moles of 1; 2 ꢁ PDO
1; 2 ꢁ PDO selectivity ¼
ꢂ 100%
adsorption
moles of all liquid products
In situ FT-IR experiments of glycerol adsorption were conducted
TOF ðhꢁ
1
Þ
on a Bruker Tensor 70 FT-IR spectrometer equipped with a LN-MCT
Mid VP detector. A self-supported sample wafer was reduced with
conversion ꢂ moles of glycerol in the feed ðmolÞ
ꢁ1
¼
:
1
0 vol% H
2
/Ar flow (50 mL min ) at 593 K for 2 h. The spectra were
total Cu amount ðmolÞ ꢂ Cu dispersion ꢂ reaction time ðhÞ
collected at selected temperatures in the course of cooling down
and used as background. After cooling to room temperature,
2
0 mL 5 wt% glycerol tetrahydrofuran solution was dripped onto
3
. Results and discussion
the wafer homogeneously. The wafer was then heated, and the IR
spectra were recorded under vacuum at 323, 373, 423, and
4
3.1. Formation of CuAOASiAOA species and highly dispersed Cu
73 K. Tetrahydrofuran was used as a solvent because it can easily
be pumped out of the vacuum system at room temperature for its
low boiling point. The spectra of adsorbed species can be acquired
by subtracting the background spectra. The weight of the sample
wafer is about 10 mg for pure SBA-15, 10%Cu/SBA-15(IM), and
XPS analysis was used to determine the chemical states of cop-
per. The XPS spectra of calcined and reduced 10%Cu/SBA-15(G)
and 10%Cu/SBA-15(IM) are shown in Fig. 1A. For calcined 10%
Cu/SBA-15(G), two Cu 2p3/2 peaks (933.1 eV and 935.4 eV) can
be distinguished, corresponding to CuO and CuAOASiAOA spe-
cies, respectively [32,43–45]. As a comparison, the 935.4eV peak
of calcined 10%Cu/SBA-15(IM) is much smaller and enveloped
by the CuO peak at 933.1 eV, meaning fewer CuAOASiAOA spe-
cies in calcined 10%Cu/SBA-15(IM). These results are supported
by the green color of 10%Cu/SBA-15(G) and the gray color of
1
0%Cu/SBA-15(G). Cu+SBA-15 is a physical mixture of 50 wt%
metallic Cu (Alfa-Aesa) with particle size 0.5–1.5
and 50 wt% SBA-15.
lm diameter
2.4. Density functional theory calculations
First principles DFT calculations were carried out with the
1
0%Cu/SBA-15(IM) shown in Fig. S1 in the Supplementary Mate-
rial, because the colors of CuAOASiAOA and bulk CuO are green
and black, respectively [31].
Vienna ab initio simulation package (VASP) [38,39] with
projector-augmented wave (PAW) [40,41] pseudo-potentials. The
generalized gradient approximation (GGA)-type functional, param-
eterized by Perdew, Burke, and Ernzerhof (PBE), was implemented
to describe electron exchange and correlation energy [42]. The
plane-wave cutoff energy for the wave function was set to
From XPS spectra of reduced samples, it can be found that no
satellite peak appears at 940–947 eV, suggesting that almost all
2+
of the Cu species were reduced [46]. Additionally, the Cu2p2/3
binding energy of reduced 10%Cu/SBA-15(G) is higher than that
of the sample prepared by impregnation. The shift of the reduced
0%Cu/SBA-15(G) to higher binding energy indicates that a strong
4
00 eV. All structures were considered relaxed until the total
-5
energy was converged to 10 eV and all atomic forces were smaller
1
ꢁ1
than 0.05 eV Å . Spin-polarized calculations were performed. We
adopted a four-layer SiO (1 1 0) slab with a Cu adsorbing struc-
ture (Cu /SiO ) to investigate the CuAOASiAOA interface interac-
interaction exists between Cu and SiO resulting from formation of
2
2
4
CuAOASi bonds [47]. In contrast, only a few CuAOASi bonds were
formed in 10%Cu/SBA-15(IM), as indicated by the asymmetry of the
4
2
tion. More details concerning DFT methods and the calculated
model are described in the Supplementary Material.
Cu2p2/3 peak. The electron shift in CuAOASi bonds should result in
d+
the formation of Cu
sites. To confirm the existence of
CuAOASiAOA species and determine the surface content of Cu
species with different valence states, Cu LMM XAES spectra were
recorded for reduced catalysts (Fig. 1B). All samples presented
two peaks at kinetic energies of about 918.1 and 914.2 eV in the
Cu LMM XAES spectra, which could be ascribed to Cu0 and Cu
2
.5. Catalytic test
Glycerol hydrogenolysis to 1,2-PDO was carried out in a Teflon-
d+
lined stainless steel autoclave (NS-10-316 L, Anhui Kemi Machin-
ery Technology Co., Ltd.) equipped with an electromagnetic stirrer
and a temperature controller. The specific reaction conditions were
as follows: a certain amount of catalyst, 6.67 g n-butanol solution
of glycerol (12 wt%), 483 K, 4.0 MPa hydrogen pressure, and
sites
(CuAOASiAOA
species),
respectively
[3,37].
0
(CuAOASiAOA)/Cu was calculated by the area ratio of peaks cen-
tered at 914.2 and 918.1 eV. As clearly illustrated in Fig. 1B, the
0
(CuAOASiAOA)/Cu ratio of 10%Cu/SBA-15(G) (1.098) was much
5
00 rpm stirring speed. All the catalysts were prereduced in a flow
higher than that of 10%Cu/SBA-15(IM) (0.706), coincident with
XPS results.
ꢁ1
2
of 10 vol% H /Ar (50 mL min ) at 593 K for 2 h before reaction.
After designed reaction time, the reactor was cooled with an ice
water bath. Finally, the used catalyst was separated from the liquid
phase by centrifugation.
The TEM images of reduced 10%Cu/SBA-15(IM) and xCu/SBA-15
(G) are shown in Fig. S2. Initially, no obvious Cu nanoparticles
could be found in the general microstructure of reduced ground
samples (Figs. S2A, S2C, and S2E); the Cu nanoparticles are proba-
bly located inside the mesopores of SBA-15 [48]. After being
exposed to electron beam irradiation at high resolution for about
5 min, the ordered mesopores of SBA-15 were damaged and Cu
nanoparticles showed up (Figs. S2B, S2D, and S2F). The Cu
nanoparticles in xCu/SBA-15(G) are highly dispersed and the parti-
cle sizes are small. However, the large Cu particles could easily be
identified for 10%Cu/SBA-15(IM) (Fig. S2H), suggesting that small
Cu nanoparticles could be obtained more easily by grinding than
impregnation.
The collected liquid samples were analyzed on a Shimadzu GC-
2
010 Plus gas chromatograph equipped with a flame ionization
detector (FID), an HP-PONA capillary column, and an autosampler.
The gas phase was analyzed on an Agilent 7890A gas chro-
matograph equipped with two FIDs, a TCD, two capillary columns
(
2 3
DB-1 and GS-OxyPlot), and a Al O /KCl column. The gaseous prod-
ucts contain little CO . The contents of products and unreacted
2
glycerol were determined using an external standard method.
The conversion of glycerol, selectivity to identified products, and
turnover frequency (TOF) were calculated as follows:

1
6
J. Shan et al. / Journal of Catalysis 383 (2020) 13–23
Fig. 1. (A) XPS spectra of calcined and reduced samples with 10 wt% Cu loading synthesized by grinding and impregnation. (B) Deconvolution of Cu XAES spectra of reduced
catalysts.
To illustrate the effects of the two preparation methods, XRD
patterns of 10%Cu/SBA-15(G) and 10%Cu/SBA-15(IM) at different
stages were recorded to analyze the structural evolution during
synthesis. As displayed in Fig. 2A, the diffraction peaks at 12.6°,
these two peaks did not show much change over 10%Cu/SBA-15
(IM) (Fig. 3B). The absorption bands around 800 and 1120 cm
ꢁ1
are assigned to different vibration modes of SiAO bonds of silica
[3,50,53]. Because the existence of a CuAOASiAOA structure has
been confirmed by the color of the catalyst (Fig. S1), the XPS peak
at 935.4 eV (Fig. 1A), and XAES peaks at 914.2 eV (Fig. 1B), the
2
5.6°, and 33.4° are ascribed to copper hydroxynitrate (Cu
NO ) of dried 10%Cu/SBA-15(IM) [22]. The sharp diffraction peaks
indicate aggregation of copper salt in the drying process. After cal-
cination at 773 K for 5 h, the Cu (OH) NO diffraction peaks were
2 3
(OH) -
3
ꢁ1
increasing intensity of SiAO vibration (1120 cm ) was probably
also relevant to the formation of large numbers of SiAO bonds
related to CuAOASiAOA structure in 10%Cu/SBA-15(G). As
2
3
3
totally replaced by CuO peaks at 35.4°, 38.6°, and 48.7°. After
reduction at 593 K, these peaks for CuO disappeared and peaks
of Cu showed up. However, no obvious diffraction peaks related
to copper species can be observed in XRD spectra of the dried, cal-
cined, and reduced 10%Cu/SBA-15(G) (Fig. 2B). These results sug-
2
+
reported by Gong and his co-workers [43], Cu can react with
II
the silanol groups of a silica surface to produce ꢃSiOCu monomers
and form CuAOASiAOA via consecutive polymerization. After
grinding, the well-dispersed copper precursor can react with sila-
gest that good dispersion of Cu
obtained at different stages for the ground catalyst. Xie et al. [49]
pointed that nitrates can be highly dispersed on the surfaces of
2
(OH)
3
NO
3
, CuO, and Cu can be
nol groups at the interface of Cu
sonable to deduce that better-dispersed Cu
2
(OH)
3
NO
3
and SBA-15. It is rea-
(OH) NO obtained
2
3
3
by grinding can interact more easily with silanol groups on the
SBA-15 surface, resulting in more CuAOASiAOA species forma-
tion. The CuAOASiAOA structure between the interface of copper
species and SBA-15 can stabilize the CuO and Cu nanoparticles in
the calcination and reduction steps. For 10%Cu/SBA-15(IM), bulk
Cu (OH) NO results in a small interface between Cu (OH) NO
2 3 3 2 3 3
and SBA-15 and much fewer CuAOASiAOA species can be formed,
leading to formation of bulk CuO and Cu.
supports (
than their melting temperatures. During grinding, strong friction
leads to an increase in the surface temperature of Cu (OH) NO
3
2 3 2
c-Al O , TiO , zeolites, etc.) at temperatures much lower
2
3
particles, which facilitates the dispersion of the Cu precursor. How-
ever, with the impregnation method, the copper precursor tends to
aggregate at the drying stage, as confirmed by XRD. By comparing
the XRD patterns of 10%Cu/SBA-15(G) and 10%Cu/SBA-15(IM), it
can be inferred that better dispersion of Cu
2
(OH)
3
NO
3
on SBA-15
From these discussions, it can be found that the formation of
fine dispersed copper precursor by grinding plays a key role in
the synthesis of highly dispersed Cu/SBA-15 catalyst. These find-
ings suggest pursuing high dispersion of metallic precursors to
construct well supported catalysts.
leads to smaller CuO and Cu particles after calcination and reduc-
tion, respectively, and the mechanical mixing step is a decisive fac-
tor for well-dispersed Cu on xCu/SBA-15(G).
IR spectroscopy is a superior tool to distinguish Cu(OH)
2
,
Cu
2
(OH) NO , and CuAOASiAOA species [43]. To confirm the
3
3
decisive role of grinding, in situ DRIFTS was employed to study
the evolution of copper species in calcination processes. Fig. 3
shows the DRIFTS spectra of dried 10%Cu/SBA-15(G) and 10%Cu/
SBA-15(IM) collected in situ at different calcination temperatures.
At 313 K, two absorption bands were observed at approximately
3.2. Structure–performance relationship
The reduced catalysts were evaluated at 483 K and 4.0 MPa
hydrogen pressure with n-butanol as solvent for different reaction
times. As displayed in Fig. 4A, the conversion of glycerol gradually
increased with increasing reaction time. For 10%Cu/SBA-15(G),
glycerol conversion was 59.1% at 1.5 h and increased to 88.9% in
another 4.5 h. Comparatively, the impregnated catalyst exhibited
much lower glycerol conversion for all the tests, despite that both
catalysts showed excellent selectivity to 1,2-PDO > 97% due to the
higher intrinsic ability of Cu-based catalysts to selectively cleave
CAO bonds over CAC bonds [2].
Fig. 4B illustrates the effects of Cu loading on the catalytic per-
formance of glycerol hydrogenolysis over Cu/SBA-15(G). Glycerol
conversion improved from 40.1% to 70.6% with Cu content increas-
ing from 5% to 20% at reaction time of 1.5 h, while the selectivity to
1,2-PDO declined only slightly. Reactions performed under other
conditions (different temperatures and atmospheres) were also
ꢁ1
1
430 and 1349 cm , which are related to NO
and imply the presence of Cu (OH) NO
With the increase in calcination temperature, Cu
decomposed and its characteristic peaks disappeared. The dOH band
3
groups [50,51]
2
3
3
.
2 3 3
(OH) NO
ꢁ1
could be found at 670 cm for both calcined catalysts, and is gen-
erally assigned to OH deformation vibration of copper phyllosili-
cate in previous Refs. [3,43,50,52]. In our work, the surface
CuAOASiAOA species have formation processes and chemical
structures similar to those of copper phyllosilicate, and thus we
deduce that the dOH may be related to surface hydroxyl groups of
ꢁ1
CuAOASiAOA species. Notably, the peak around 800 cm in the
ꢁ
1
spectra of 10%Cu/SBA-15(G) became weaker and the 1120 cm
band became stronger (Fig. 3A) during calcination. Nevertheless,

J. Shan et al. / Journal of Catalysis 383 (2020) 13–23
17
Fig. 2. XRD patterns of dried, calcined, and reduced (A) 10%Cu/SBA-15(IM) and (B) 10%Cu/SBA-15(G). (C) Comparison of XRD results for calcined 10%Cu/SBA-15(IM) and xCu/
SBA-15(G).
Fig. 3. In situ DRIFTS spectra of 10%Cu/SBA-15(G) (A) and 10%Cu/SBA-15(IM) (B).
investigated for 10%Cu/SBA-15(IM) and 10%Cu/SBA-15(G) (Table 1).
An increase of reaction temperature greatly enhanced the conver-
sion of glycerol.
its structure. For all the calcined catalysts, the surface area and
pore volume are proportionally decreased with increased copper
content.
The structural and surface properties of employed catalysts
were analyzed to disclose their different catalytic performance.
The exposed copper surface area, dispersion, and Cu crystallite
size were measured by the N O dissociative chemisorption
2
Fig. S3 shows the N
2
adsorption–desorption isotherms and pore
method. As shown in Table 2, 10%Cu/SBA-15(IM) shows the lowest
dispersion (8.8%), while the dispersion of 10%Cu/SBA-15(G) is
about six times higher. These results clearly suggest that the grind-
ing method facilitates the formation of small, highly dispersed Cu
nanoparticles. Among the ground samples, 5%Cu/SBA-15(G) shows
size distribution curves of calcined catalysts, SBA-15, and ground
SBA-15 (SBA-15(G)). The surface area and pore volume are listed
in Table 2. After ball-milling, a decrease in volume and surface area
was observed for SBA-15, indicating that the milling step damaged

1
8
J. Shan et al. / Journal of Catalysis 383 (2020) 13–23
Fig. 4. Glycerol conversion and 1,2-PDO selectivity of 10%Cu/SBA-15(G) and 10%Cu/SBA-15(IM) at different reaction times (A), and of xCu/SBA-15(G) at a TOS of 1.5 h (B).
Reaction conditions: 4.0 MPa, 483 K, weight ratio of catalyst/glycerol = 1:10.
Table 1
Catalytic results for hydrogenolysis of glycerol on 10%Cu/SBA-15(IM) and 10%Cu/SBA-15(G) under different conditions.a
Sample
T (K)
Atmosphere
Conversion (%)
Selectivity (%)
,2-PDO
Carbon balance (%)
1
Hydroxyacetoneb
Others
1
0%Cu/SBA-15(IM)
0%Cu/SBA-15(G)
443
H
2
H
2
H
2
H
2
N
2
H
2
H
2
H
2
H
2
N
2
4 MPa
4 MPa
4 MPa
4 MPa
4 MPa
4 MPa
4 MPa
4 MPa
4 MPa
4 MPa
1.5
3.3
8.1
13.6
1.6
99.3
98.9
99.2
97.5
0
99.5
99.2
98.7
97.3
0
—
—
—
—
50.5
—
—
—
—
~0.7c
95.4
95.9
96.1
95.1
—
94.1
95.2
93.8
94.6
—
c
4
4
5
4
63
83
03
83
~1.1
c
~0.8
~2.5
c
d
49.5
c
1
443
8.7
~0.5
~0.8
c
4
4
5
4
63
83
03
83
24.9
59.1
90.3
7.5
c
~1.3
c
~2.7
56.9
43.1d
a
b
c
Other reaction conditions: 1.5 h, weight ratio of catalyst/glycerol = 1:10.
Hydroxyacetone can be detected by GC, but its quantity is too small to quantify.
Others include ethylene glycerol and propanols.
d
Compounds that cannot be defined.
Table 2
The physiochemical properties of different samples.
Cu content (wt.%)a
S
BET (m2
g
ꢁ1
)
V
p
(cm3
g
ꢁ1
)
Calculated by N
D (%)b
O chemisorption results
Acid site amounte
(lmol/g)
Sample
2
ꢁ1
d
Cu (nm)c
S
Cu (m 2C u
g
SBA-15
—
9.6
—
4.9
9.7
19.3
608
490
471
406
366
267
0.99
0.85
0.65
0.56
0.51
0.43
—
8.8
—
70.0
54.9
38.6
8.3
7.1
—
11.3
—
1.4
1.8
2.6
—
5.9
—
23.2
37.1
50.4
5.6
4.8
0.004
0.105
0.004
0.560
0.851
1.047
1
0%Cu/SBA-15(IM)
SBA-15(G)
5
1
2
%Cu/SBA-15(G)
0%Cu/SBA-15(G)
0%Cu/SBA-15(G)
f
Cu powder
Cu@SiO
2
0.072
a
b
c
d
e
f
Determined by ICP.
Dispersion degree of metallic copper.
Average size of Cu particles.
Cu exposed surface area calculated from N O chemisorption (Fig. S4).
2
3
Determined by NH TPD.
Metallic Cu powder purchased from Alfa-Aesar.
0
the highest Cu dispersion (70.0%). Although the dispersion degree
decreases with the increase in the copper amount, it still remains
at a high value of 38.6% when the sample supports 20% Cu.
the low-temperature peak, and only one peak at 460–470 K can
be found when Cu content is decreased to 10% or 5%. This gives
another piece of evidence for the conclusion that most copper spe-
cies in 10%Cu/SBA-15(IM) are bulk CuO, while the samples pre-
pared by the grinding method contain mainly highly dispersed
CuO nanoparticles. The slight shift of the reduction peak at about
460 K to higher temperature with increasing Cu loading in the
ground catalysts indicates an increase of CuO nanoparticle size.
The H
2
-TPR profiles of both 10%Cu/SBA-15(IM) and 20%Cu/SBA-
5(G) show two reduction peaks (Fig. 5). The low-temperature
1
peak at 460–470 K is related to the reduction of well-dispersed
CuO nanoparticles, while the high-temperature one (550 K) can
be ascribed to bulk CuO [22,23]. It is noteworthy that the high-
temperature reduction peak area is much larger than the low-
temperature peak area in the profile of 10%Cu/SBA-15(IM),
0
The above results confirm that much higher Cu dispersion could
be achieved by grinding than by the conventional impregnation
2
whereas the H -TPR profile of 20%Cu/SBA-15(G) is dominated by
2
method, in line with XRD, N O dissociative chemisorption, and

J. Shan et al. / Journal of Catalysis 383 (2020) 13–23
19
Table 3
0
TOF and LAS/Cu of xCu/SBA-15(G).
Conversion (%)a
LAS/Cu0b
TOF (h )c
ꢁ1
Samples
5
1
2
1
%Cu/SBA-15(G)
0%Cu/SBA-15(G)
0%Cu/SBA-15(G)
0%Cu/SBA-15(IM)
7.3
1.017
0.985
0.862
0.758
57.2
52.8
48.3
44.2
10.5
13.5
1.4
a
Reaction conditions: 4.0 MPa, 483 K, 1.0 h, weight ratio of catalysts/
glycerol = 0.025.
b
0
The Cu site was determined from the metallic Cu dispersion (DCu) measured by
0
N
2
O chemisorption using the equation Cu site amount per gram of catalyst =
1
gꢂCu mass percentage
ꢂ DCu (mol/g), where MCu = 63.46 g/mol.
MCu
c
TOF was calculated on the basis of metallic Cu dispersion at conversions lower
than 15% in this table.
Cu0 ratio and the linear relationship between TOF and
0
(
CuAOASiAOA)/Cu . As shown in Fig. 1B, the (CuAOASiAOA)/
0
Cu ratios of the reduced 10%Cu/SBA-15(IM), 5%Cu/SBA-15(G),
1
1
0%Cu/SBA-15(G), and 20%Cu/SBA-15(G) were 0.706, 0.834,
2
Fig. 5. H TPR profiles of xCu/SBA-15(G) and 10%Cu/SBA-15(IM).
.098, and 1.292, respectively. From Fig. 6B, TOF increased linearly
0
with the rising ratio of (CuAOASiAOA)/Cu , indicating the crucial
role of the CuAOASiAOA structure in glycerol hydrogenolysis.
It can be speculated that Lewis acid sites originated from
CuAOASiAOA might be the active center for catalyzing glycerol
dehydration to hydroxyacetone, and the generated intermediate
would adsorb onto adjacent Cu sites for subsequent hydrogena-
tion. Then, the high TOF of 5%Cu/SBA-15(G) can be ascribed to large
numbers of CuAOASiAOA structures around Cu sites.
Former research [55–57] have found that the rate-determining
step of glycerol hydrogenolysis is the cleavage of CAOH bonds in
the dehydration step rather than the hydrogenation of the dehy-
dration product (hydroxyacetone). To prove this point, first, hydro-
genation of hydroxyacetone was investigated on different
catalysts. As shown in Table 4, the conversion of hydroxyacetone
for 10 min at 443 K was 70.3% and 100% on 10%Cu/SBA-15(IM)
and 10%Cu/SBA-15(G), respectively. However, glycerol conversion
was only 1.5% and 8.7% on 10%Cu/SBA-15(IM) and 10%Cu/SBA-15
(G), respectively, even for a much longer time (1 h) at the same
temperature (Table 1). Thus, the hydrogenation of hydroxyacetone
was proved not to be the rate-determining step, and could be effec-
tively catalyzed by metallic Cu. These results evidenced that the
promotion effect of CuAOASiAOA structures was mainly present
in the dehydration step in this study.
TEM characterization results. It is generally accepted that metallic
Cu is the active site for hydroxyacetone hydrogenation [2], which is
the second step of glycerol hydrogenolysis (Scheme 1). Thus, more
,2-PDO could be obtained over a larger exposed Cu surface area if
more hydroxyacetone could be provided from the first dehydration
step.
1
0
0
It was mentioned in [29] that metallic Cu species might also
catalyze glycerol dehydration to hydroxyacetone. To identify the
0
exact role of Cu , glycerol hydrogenolysis reactions were per-
formed on SBA-15, Cu powder, and a mixture of Cu powder and
SBA-15 (Cu+SBA-15). The catalytic results listed in Table S1 show
that Cu powder exhibited a much lower reaction rate than 10%
Cu/SBA-15(IM) under the same conditions, although their exposed
0
Cu surfaces were very close. Because 10%Cu/SBA-15(IM) possesses
a
certain amount of CuAOASiAOA structure, absence of
CuAOASiAOA structure seems very likely to be the reason for poor
catalytic activity of Cu powder and Cu+SBA-15.
Py-IR spectra of reduced samples in Fig. S5 illustrate that 10%
Cu/SBA-15(IM) and xCu/SBA-15(G) possess only Lewis acid sites
LAS), while the acid site amount is negligible for SBA-15 and Cu
SBA-15. According to XPS and XAES results of reduced catalysts
Fig. 1), the CuAOASiAOA structure should be the only source of
Lewis acid [42,51,52,54]. The acid site amounts determined by
NH -TPD of the reduced 10%Cu/SBA-15(IM), 5%Cu/SBA-15(G), 10%
(
+
(
For ground Cu/SBA-15 catalysts, the (CuAOASiAOA induced
0
3
Lewis acid sites)/Cu ratio increased with the decrease of Cu load-
Cu/SBA-15(G), and 20%Cu/SBA ꢁ15(G) are 0.105, 0.560, 0.851,
and 1.047 mmol/g, respectively (Table 2). The much lower Lewis
acid amount of 10%Cu/SBA-15(IM) than of xCu/SBA-15(G) further
confirms that many more CuAOASiAOA structures were formed
in xCu/SBA-15(G).
ing (Table 3). Thus, two other ground catalysts with 1% and 2% Cu
loading were synthesized to get a high (CuAOASiAOA)/Cu ratio
0
for finding out the boundary of the linear relationship between
0
TOF and (CuAOASiAOA)/Cu . However, a good linear relationship
0
between TOF and LAS/Cu can still be observed over 1% and 2% Cu/
TOF can be employed to reflect the intrinsic activity of a cata-
lyst, which has been estimated in terms of the surface-active Cu
sites using initial glycerol conversion (<15%) (Table 3). Of all the
examined catalysts, 5%Cu/SBA-15(G) achieved optimal perfor-
SBA-15(G) in Fig. S6. Although the accurate boundary cannot be
deduced with this result, the promotion effect of CuAOASiAOA
structures was further confirmed.
ꢁ1
mance, 57.2 h , which is superior to that of most reported Cu-
based catalysts [2,5,24–29]. More interestingly, a strong linear
3.3. In situ Fourier transform infrared experiments on glycerol
adsorption
0
relationship between TOF and LAS/Cu was found, as displayed in
0
0
Fig. 6A. LAS/Cu indicates the LAS amount per Cu site. It can be
demonstrated that Lewis acid sites derived from CuAOASiAOA
structures benefit the catalytic activity of Cu/BA-15. Increasing
numbers of CuAOASiAOA structures facilitate the continuous
increase of TOF. The linear relationship between TOF and LAS/Cu0
implies that CuAOASiAOA structures are crucial for achieving
excellent hydrogenolysis performance. To make the conclusion
more credible, Cu LMM X-ray excited XAES spectra were analyzed
to reconfirm the (Lewis acid sites originated from CuAOASiAOA)/
To probe the intrinsic active sites of glycerol dehydration, in situ
FT-IR spectra were recorded at specified temperatures after
adsorption of glycerol onto pure SBA-15, Cu+SBA-15, and reduced
Cu/SBA-15 catalysts (Fig. 7). As shown in Fig. 7A, the initial glycerol
loadings were almost the same on four samples because nearly
identical infrared bands of adsorbed glycerol were observed at
ꢁ1
ꢁ1
ꢁ1
1
[
462 cm
(dCH2 of glycerol) and 1413 cm
(dOH of glycerol)
was ascribed to
58]. The negative peak at about 1626 cm
2
adsorbed H O [34]. When the temperature was increased to

2
0
J. Shan et al. / Journal of Catalysis 383 (2020) 13–23
Fig. 6. (A) Relationship between LAS/Cu0 and TOF of Cu . (B) Relationship between (CuAOASiAOA)/Cu and TOF of Cu0.
0
0
Table 4
Catalytic performance of hydroxyacetone hydrogenation on different catalysts.
than onto the other samples, which suggested stronger interaction
between glycerol and 10%Cu/SBA-15(G). With increasing tempera-
ꢁ1
Sample
Conversion (%)
1,2-PDO Selectivity (%)
ture, the dOH bands of glycerol at about 1413 cm [58] disap-
peared in the spectra of 10%Cu/SBA-15(G) and 10%Cu/SBA-15(IM)
taken at 423 K and 473 K (Fig. 7C and D) because of dehydration
of glycerol. For SBA-15 and Cu+SBA-15, the dOH bands of glycerol
became weaker and underwent a red shift at high temperature.
Cu powder
55.9
78.1
100
>99
>99
>99
1
1
0%Cu/SBA-15(IM)
0%Cu/SBA-15(G)
2
Note: Reaction conditions: 4.0 MPa H ; 443 K; 10 min; molar quantity of hydrox-
ꢁ
1
yacetone equal to that of glycerol in hydrogenolysis reactions.
It should be noted that a broad peak in the 1600–1500 cm
region started to appear at 373 K and became much more obvious
at higher temperatures for 10%Cu/SBA-15(G) and 10%Cu/SBA-15
(IM) (Fig. 7). In contrast, this band was much smaller in the spec-
trum of Cu+SBA-15 and could not be found for pure SBA-15. The
3
73 K (Fig. 7B), the amount of adsorbed glycerol decreased due to
desorption of physically adsorbed glycerol at higher temperature.
Nevertheless, more glycerol adsorbed onto the 10%Cu/SBA-15(G)
ꢁ1
band in the 1600–1500 cm region was attributed to carboxylate
Fig. 7. In situ FT-IR spectra recorded at different temperatures after adsorbing glycerol.

J. Shan et al. / Journal of Catalysis 383 (2020) 13–23
21
species; it comes from the redox reaction between the copper spe-
cies and hydroxyacetone under hydrogen-deficient conditions [59–
ꢁ
1
6
1]. The much stronger band at 1600–1500 cm implies that more
hydroxyacetone was produced on 10%Cu/SBA-15(G). Besides, the
formation of hydroxyacetone can also be verified by the peak of
ꢁ
1
m
C=O at 1730 cm
[59,60]. The assignment of the 1600–
ꢁ1
1
500 cm
band was further demonstrated by hydroxyacetone
adsorption experiments. Fig. S7 shows that this band appeared
immediately after introduction of hydroxyacetone at 423 K. Thus,
low conversion of glycerol under N can be explained (Table 1).
2
Under hydrogen-deficient conditions, the generated carboxylate
species covered the Cu surface and inhibited the dehydration
reaction.
The above results indicate that Cu and SBA-15 are not effective
catalysts for glycerol dehydration. Comparatively, 10%Cu/SBA-15
(
G) and 10%Cu/SBA-15(IM) are much more active, as confirmed
by the high disappearance rate of glycerol hydroxyls with increas-
ing temperature (Fig. 7). Thus, it can be concluded that metallic Cu
and SBA-15 are not the primary dehydration active site, whereas
Lewis acid sites derived from CuAOASiAOA structures can greatly
promote the first step, dehydration of glycerol to hydroxyacetone.
It should be noted that the exposed surface area of Cu powder is
very close to that of 10%Cu/SBA-15(IM) (Table 2). This further con-
firms that the CuAOASiAOA interface structures are responsible
for the discrepancy in catalytic activity between Cu powder and
Fig. 8. Energy profile for cleavage of the terminal hydroxyl group of glycerol on Cu
(
4 2
1 1 1) and Cu /SiO . The local configurations of the adsorbates on catalysts at initial
states (IS), transition states (TS), and final states (FS) along the minimum-energy
pathway are shown in the inserts.
Yun et al. [55]. The adsorption energy of glycerol on Cu (1 1 1) is
ꢁ1.11 eV, and the terminal CAOH bond is 1.44 Å. For Cu
4 2
/SiO ,
1
0%Cu/SBA-15(IM). When it comes to 10%Cu/SBA-15(IM) and 10%
the adsorption energy of glycerol is ꢁ1.66 eV, and the terminal
CAOH bond is 1.47 Å, which is longer than CAOH bonds in glycerol
adsorbed onto Cu (1 1 1). The lower Eads and longer CAOH bond
indicate that the adsorption and activation of glycerol on
CuAOASiAOA structure is more favorable than Cu surface. We fur-
ther investigated the energy barrier of terminal CAOH cleavage of
ꢁ1
Cu/SBA-15(G), the peak is much larger in the 1600–1300 cm
region for 10%Cu/SBA-15(G). First, more CuAOASiAOA structures
on 10%Cu/SBA-15(G) give more hydroxyacetone. Second, the
exposed copper surface area of 10%Cu/SBA-15(G) is far larger than
that of 10%Cu/SBA-15(IM) and the generation of hydroxyacetone
can interact with Cu nanoparticles of 10%Cu/SBA-15(G), which
facilitates the subsequent hydrogenation step.
glycerol on Cu (1 1 1) and Cu
the removal of terminal hydroxyl group on Cu
4
/SiO
2
(Fig. 8). The energy barrier for
/SiO is calculated
4
2
To support the above conclusions, Cu powder coated with silica
to be 1.50 eV, lower than that on Cu (1 1 1) (1.85 eV). Additionally,
the energy barrier for the elimination of center hydroxyl groups
was also calculated on Cu /SiO (2.44 eV) and Cu (1 1 1)
4 2
(
Cu@SiO
acetone hydrogenation reactions (Table S1). The existence of
CuAOASiAOA structures in Cu@SiO has been reported by Zheng’s
2
) was evaluated in glycerol hydrogenolysis and hydroxy-
(2.00 eV), as illustrated in Fig. S8. In both models, the removal of
center hydroxyl groups is much more difficult than that of terminal
hydroxyl groups, coincidence with the high selectivity to 1,2-PDO
(Scheme 1). The energy barrier of hydroxyacetone hydrogenation
is 1.20 eV (Fig. S9), which confirms the presumption that CAOH
cleavage is the rate-determining step in glycerol hydrogenolysis.
From the above results, we can conclude that the CuAOASiAOA
interface structures enhance the adsorption of glycerol molecules
and lower the energy barrier of terminal hydroxyl group cleavage,
which boost the glycerol dehydration step. The DFT calculation
results are well consistent with our experimental results that
CuAOASiAOA interface structures are the primary active sites
for glycerol dehydration and can significantly improve the catalytic
2
group [35]. Besides, it is reported in Section 3.2 that the
CuAOASiAOA structure is the only source of Lewis acid, so the
appearance of Lewis acid sites in Cu@SiO
strates the formation of the CuAOASiAOA structure. As for cat-
alytic performance, glycerol conversion on Cu@SiO (6.5%,
2
(Fig. S4) clearly demon-
2
Table S1) was about six times that on Cu powder (1.2%, Table S1)
in the hydrogenolysis reaction. Glycerol hydrogenolysis and
hydroxyacetone hydrogenation reaction results of Cu@SiO
Table S1) again confirmed the crucial role of the CuAOASiAOA
structure in dehydration. These experimental results indicate that
the interfacial CuAOASiAOA structures between Cu and SiO (not
just SBA-15) could dramatically promote the catalytic activity of
Cu/SiO catalysts, irrespective of the type of SiO
2
(
2
2
.
2
performance of Cu/SiO catalysts in glycerol hydrogenolysis.
2
3.4. Density functional theory calculation results
3.5. Catalytic stability
As confirmed in Section 3.2, CAOH cleavage in the dehydration
step was the rate-determining step of glycerol hydrogenolysis
55,57]. Therefore, investigation of CAOH cleavage can provide a
Reusability was tested for 10%Cu/SBA-15(G) and 10%Cu/SBA-15
(IM) in a third reaction run. After every reaction, the catalyst was
separated and dried at 353 K, followed by calcination in air flow
[
rational interpretation of catalytic behavior for glycerol
hydrogenolysis and give information about the active sites for glyc-
erol dehydration.
Because the Cu (1 1 1) surface was the most exposed surface in
our experiments, the catalyst model of Cu (1 1 1) was constructed
2
at 773 K and reduction under a 10 vol% H /Ar flow at 593 K. Table 5
lists glycerol conversion and 1,2-PDO selectivity in different reac-
tion runs. In the case of 10%Cu/SBA-15(G), glycerol conversion
slightly decreased from 59.1% to 52.3% during three catalytic
cycles, while 10%Cu/SBA-15(IM) exhibited rather poor stability.
to simulate metallic Cu, while an SiO
adsorbed Cu cluster structure (Cu /SiO
the CuAOASiAOA interface. First, the adsorption energy of the
most stable structure of glycerol on Cu (1 1 1) and Cu /SiO was
investigated (Fig. 8), which is similar to the work reported by
2
(1 1 0) surface with an
After three times of reaction, declines in Cu content (from 9.6%
to 7.9%) and Cu exposed surface area (from 5.9 to 3.4 m 2C u
g
cat
ꢁ1
)
4
4
2
) was built to represent
could be observed for 10%Cu/SBA-15(IM), but the physicochemical
properties of 10%Cu/SBA-15(G) showed no obvious changes. It can
be deduced that abundant CuAOASiAOA interface structures in
4
2

2
2
J. Shan et al. / Journal of Catalysis 383 (2020) 13–23
Table 5
Reusability results of glycerol hydrogenolysis over 10%Cu/SBA-15(G) and 10%Cu/SBA-15(IM).
Sample
Glycerol conversion, 1,2-PDO selectivity (%)
S
Cu (m 2C u
g
cat)a
Cu content (wt.%)b
First run
Second run
Third run
1
1
0%Cu/SBA-15(IM)
0%Cu/SBA-15(G)
8.1, 99.2
59.1, 98.7
4.5, 99.0
49.3, 98.3
2.3, 99.2
52.3, 98.8
3.4
34.2
7.9
9.3
a
Cu exposed surface area of used catalyst after the third reaction.
Cu content of used catalyst after the third reaction.
b
10%Cu/SBA-15(G) could not only improve catalytic performance but
References
also stabilize Cu nanoparticles, resulting in superior catalytic activity
and stability in glycerol hydrogenolysis.
[
1] A. Corma, S. Iborra, A. Velty, Chemical routes for the transformation of biomass
into chemicals, Chem. Rev. 107 (2007) 2411–2502.
[
2] D. Sun, Y. Yamada, S. Sato, W. Ueda, Glycerol hydrogenolysis into useful C3
chemicals, Appl. Catal. B Environ. 193 (2016) 75–92.
4
. Conclusions
[3] S. Zhu, X. Gao, Y. Zhu, Y. Zhu, H. Zheng, Y. Li, Promoting effect of boron oxide on
Cu/SiO2 catalyst for glycerol hydrogenolysis to 1,2-propanediol, J. Catal. 303
(
2013) 70–79.
In this work, highly dispersed copper nanoparticles on SBA-15
[
4] C.H. Zhou, J.N. Beltramini, Y.X. Fan, G.Q. Lu, Chemoselective catalytic
conversion of glycerol as a biorenewable source to valuable commodity
chemicals, Chem. Soc. Rev. 37 (2008) 527–549.
5] Y. Nakagawa, K. Tomishige, Heterogeneous catalysis of the glycerol
hydrogenolysis, Catal. Sci. Technol. 1 (2011) 179–190.
6] S. Zhu, X. Gao, Y. Zhu, Y. Li, Tailored mesoporous copper/ceria catalysts for the
selective hydrogenolysis of biomass-derived glycerol and sugar alcohols,
Green Chem. 18 (2016) 782–791.
7] C. Wang, H. Jiang, C. Chen, R. Chen, W. Xing, Solvent effect on hydrogenolysis of
glycerol to 1,2-propanediol over Cu–ZnO catalyst, Chem. Eng. J. 264 (2015)
344–350.
were fabricated by a simple grinding method. XRD, XPS, and
in situ DRIFTS results demonstrated that good dispersion of CuO
nanoparticles on calcined 10%Cu/SBA-15(G) resulted from the
finely dispersed copper precursor formed by grinding. In the grind-
ing and calcination processes, CuAOASiAOA structures were pro-
duced in the interface of the nanoscale copper species and SBA-15
surface, and finally worked as anchor sites to stabilize Cu nanopar-
ticles. For the impregnation sample, low dispersion of copper pre-
cursor and fewer CuAOASiAOA structures resulted in the
formation of bulk Cu.
[
[
[
[
8] Y. Kang, X. Bu, G. Wang, X. Wang, Q. Li, Y. Feng, A highly active Cu–Pt/SiO2
bimetal for the hydrogenolysis of glycerol to 1,2-propanediol, Catal. Lett. 146
(
2016) 1408–1414.
The ground Cu/SBA-15 catalysts with more exposed copper sur-
face area and interfacial CuAOASiAOA structures presented excel-
lent catalytic activity and stability in glycerol hydrogenolysis. The
conversion of glycerol on 10%Cu/SBA-15(G) was much higher than
that of 10%Cu/SBA-15(IM), and both catalysts showed high selec-
[9] E.S. Vasiliadou, E. Heracleous, I.A. Vasalos, A.A. Lemonidou, Ru-based catalysts
for glycerol hydrogenolysis—effect of support and metal precursor, Appl. Catal.
B Environ. 92 (2009) 90–99.
10] I. Gandarias, J. Requies, P.L. Arias, U. Armbruster, A. Martin, Liquid-phase
glycerol hydrogenolysis by formic acid over Ni–Cu/Al2O3 catalysts, J. Catal.
290 (2012) 79–89.
[
0
[11] S. Xia, R. Nie, X. Lu, L. Wang, P. Chen, Z. Hou, Hydrogenolysis of glycerol over
8.6 catalysts: The role of basicity and hydrogen spillover,
tivity to 1,2-PDO above 97%. The TOF of Cu increased linearly with
Cu0.4/Zn5.6ꢁxMgxAl
2
O
0
the amount of LAS/Cu , indicating that Lewis acid sites derived
J. Catal. 296 (2012) 1–11.
from CuAOASiAOA structures are a crucial prerequisite for excel-
lent hydrogenolysis performance. More importantly, in situ FT-IR
experiments of glycerol adsorption and DFT calculation results evi-
denced that CuAOASiAOA structures were the primary active sites
for glycerol dehydration to hydroxyacetone. The hydrogen species
adsorbed onto adjacent Cu sites would perform hydroxyacetone
hydrogenation smoothly to generate 1,2-PDO.
[12] J. ten Dam, U. Hanefeld, Renewable chemicals: dehydroxylation of glycerol and
polyols, ChemSusChem 4 (2011) 1017–1034.
[
13] T. Miyazawa, Y. Kusunoki, K. Kunimori, K. Tomishige, Glycerol conversion in
the aqueous solution under hydrogen over Ru/C + an ion-exchange resin and
its reaction mechanism, J. Catal. 240 (2006) 213–221.
[
14] M. Lee, Y.K. Hwang, J.-S. Chang, H.-J. Chae, D.W. Hwang, Vapor-phase
hydrogenolysis of glycerol to 1,2-propanediol using a chromium-free Ni-Cu-
2
SiO nanocomposite catalyst, Catal. Commun. 84 (2016) 5–10.
[15] I. Gandarias, P.L. Arias, J. Requies, M.B. Güemez, J.L.G. Fierro, Hydrogenolysis of
glycerol to propanediols over a Pt/ASA catalyst: The role of acid and metal sites
on product selectivity and the reaction mechanism, Appl. Catal. B Environ. 97
(
2010) 248–256.
Declaration of Competing Interest
[
[
[
16] I. Furikado, T. Miyazawa, S. Koso, A. Shimao, K. Kunimori, K. Tomishige,
Catalytic performance of Rh/SiO
Chem. 9 (2007) 582–588.
17] F. Auneau, S. Noël, G. Aubert, M. Besson, L. Djakovitch, C. Pinel, On the role of
the atmosphere in the catalytic glycerol transformation over iridium-based
catalysts, Catal. Commun. 16 (2011) 144–149.
2
in glycerol reaction under hydrogen, Green
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.
2 3
18] J. Zhou, J. Zhang, X. Guo, J. Mao, S. Zhang, Ag/Al O for glycerol hydrogenolysis
to 1,2-propanediol: activity, selectivity and deactivation, Green Chem. 14
(2012) 156–163.
19] A. Perosa, P. Tundo, Selective hydrogenolysis of glycerol with Raney nickel, Ind.
Eng. Chem. Res. 44 (2005) 8535–8537.
Acknowledgments
[
The authors gratefully acknowledge financial support from the
National Natural Science Foundation of China (21878321), the Nat-
ural Science Foundation of Shanxi Province (2016021033), the
Science Foundation for Youth Scholars of the State Key Laboratory
of Coal Conversion (2016BWZ002), and the Youth Innovation Pro-
motion Association CAS (2015140). The authors declare that they
have no competing interests.
[
20] X. Guo, Y. Li, W. Song, W. Shen, Glycerol hydrogenolysis over Co catalysts
derived from a layered double hydroxide precursor, Catal. Lett. 141 (2011)
1458–1463.
[
21] X. Guo, Y. Li, R. Shi, Q. Liu, E. Zhan, W. Shen, Co/MgO catalysts for
hydrogenolysis of glycerol to 1, 2-propanediol, Appl. Catal. A Gen. 371
(2009) 108–113.
[22] S. Zhu, X. Gao, Y. Zhu, W. Fan, J. Wang, Y. Li, A highly efficient and robust Cu/
SiO2 catalyst prepared by the ammonia evaporation hydrothermal method for
glycerol hydrogenolysis to 1,2-propanediol, Catal. Sci. Technol. 5 (2015) 1169–
1
180.
23] P. Hirunsit, C. Luadthong, K. Faungnawakij, Effect of alumina hydroxylation on
glycerol hydrogenolysis to 1,2-propanediol over Cu/Al combined
[
Appendix A. Supplementary material
2 3
O :
experiment and DFT investigation, RSC Adv. 5 (2015) 11188–11197.
24] Z. Yuan, J. Wang, L. Wang, W. Xie, P. Chen, Z. Hou, X. Zheng, Biodiesel derived
glycerol hydrogenolysis to 1,2-propanediol on Cu/MgO catalysts, Bioresour.
Technol. 101 (2010) 7088–7092.
[
Supplementary data to this article can be found online at
https://doi.org/10.1016/j.jcat.2019.12.032.

J. Shan et al. / Journal of Catalysis 383 (2020) 13–23
23
[
[
[
[
[
[
25] Z. Yuan, L. Wang, J. Wang, S. Xia, P. Chen, Z. Hou, X. Zheng, Hydrogenolysis of
glycerol over homogenously dispersed copper on solid base catalysts, Appl.
Catal. B Environ. 101 (2011) 431–440.
structure, and catalytic performance in hydrogenation of dimethyl oxalate to
ethylene glycol, J. Catal. 257 (2008) 172–180.
[45] C.F. Zhou, Y.M. Wang, Y. Cao, T.T. Zhuang, W. Huang, Y. Chun, J.H. Zhu, Solvent-
free surface functionalized SBA-15 as a versatile trap of nitrosamines, J. Mater.
Chem. 16 (2006) 1520–1528.
26] D. Durán-Martín, M. Ojeda, M.L. Granados, J.L.G. Fierro, R. Mariscal, Stability
2
and regeneration of Cu–ZrO catalysts used in glycerol hydrogenolysis to 1,2-
propanediol, Catal. Today 210 (2013) 98–105.
27] M. Balaraju, V. Rekha, P.S.S. Prasad, R.B.N. Prasad, N. Lingaiah, Selective
hydrogenolysis of glycerol to 1, 2 Propanediol over Cu–ZnO catalysts, Catal.
Lett. 126 (2008) 119–124.
2
[46] Z. Huang, F. Cui, J. Xue, J. Zuo, J. Chen, C. Xia, Cu/SiO catalysts prepared by
hom- and heterogeneous deposition–precipitation methods: texture,
structure, and catalytic performance in the hydrogenolysis of glycerol to 1,2-
propanediol, Catal. Today 183 (2012) 42–51.
28] E.S. Vasiliadou, T.M. Eggenhuisen, P. Munnik, P.E. de Jongh, K.P. de Jong, A.A.
[47] A. Yin, X. Guo, W.L. Dai, K. Fan, The nature of active copper species in Cu-HMS
catalyst for hydrogenation of dimethyl oxalate to ethylene glycol: new insights
2
Lemonidou, Synthesis and performance of highly dispersed Cu/SiO catalysts
0
+
for the hydrogenolysis of glycerol, Appl. Catal. B Environ. 145 (2014) 108–119.
29] H. Liu, Z. Huang, Z. Han, K. Ding, H. Liu, C. Xia, J. Chen, Efficient production of
methanol and diols via the hydrogenation of cyclic carbonates using copper–
silica nanocomposite catalysts, Green Chem. 17 (2015) 4281–4290.
30] J. Zheng, W. Zhu, C. Ma, Y. Hou, W. Zhang, Z. Wang, Hydrogenolysis of glycerol
to 1,2-propanediol on the high dispersed SBA-15 supported copper catalyst
prepared by the ion-exchange method, React. Kinet. Mech. Catal. 99 (2010)
on the synergetic effect between Cu and Cu , J. Phys. Chem. C 113 (2009)
11003–11013.
[48] Y. Wang, J. Liao, J. Zhang, S. Wang, Y. Zhao, X. Ma, Hydrogenation of methyl
acetate to ethanol by Cu/ZnO catalyst encapsulated in SBA-15, AIChE J. 63
(2017) 2839–2849.
[49] Y.C. Xie, Y.Q. Tang, Spontaneous monolayer dispersion of oxides and salts onto
surface of supports: application to heterogeneous catalysis, Adv. Catal. 37
(1990) 1–43.
4
55–462.
[
31] Y.M. Wang, Z. Ying Wu, J.H. Zhu, Surface functionalization of SBA-15 by the
solvent-free method, J. Solid State Chem. 177 (2004) 3815–3823.
32] Y.M. Wang, Z.Y. Wu, L.Y. Shi, J.H. Zhu, Rapid functionalization of mesoporous
materials: directly dispersing metal oxides into as-prepared SBA-15 occluded
with template, Adv. Mater. 17 (2005) 323–327.
[50] T. Toupance, M. Kermarec, J.F. Lambert, C. Louis, Conditions of formation of
copper phyllosilicates in silica-supported copper catalysts prepared by
selective adsorption, J. Phys. Chem. B 106 (2002) 2277–2286.
[51] Z. Huang, H. Liu, F. Cui, J. Zuo, J. Chen, C. Xia, Effects of the precipitation agents
and rare earth additives on the structure and catalytic performance in glycerol
[
[
[
33] S. Sato, M. Akiyama, R. Takahashi, T. Hara, K. Inui, M. Yokota, Vapor-phase
reaction of polyols over copper catalysts, Appl. Catal. A Gen. 347 (2008) 186–
2
hydrogenolysis of Cu/SiO catalysts prepared by precipitation-gel method,
Catal. Today 234 (2014) 223–232.
[52] Y. Zhu, X. Kong, J. Yin, R. You, B. Zhang, H. Zheng, X. Wen, Y. Zhu, Y. Li,
Covalent-bonding to irreducible SiO leads to high-loading and atomically
2
dispersed metal catalysts, J. Catal. 353 (2017) 315–324.
[53] G. Díaz, R. Pérez-Hernández, A. Gómez-Cortés, M. Benaissa, R. Mariscal, J.L.G.
1
91.
34] D. Stosic, S. Bennici, S. Sirotin, P. Stelmachowski, J. Couturier, J. Dubois, A.
Travert, A. Auroux, Examination of acid-base properties of solid catalysts for
gas phase dehydration of glycerol: FTIR and adsorption microcalorimetry
studies, Catal. Today 226 (2014) 167–175.
2
Fierro, CuO–SiO Sol-Gel catalysts: characterization and catalytic properties
[
35] C. Xu, G. Chen, Y. Zhao, P. Liu, X. Duan, L. Gu, G. Fu, Y. Yuan, N. Zheng,
Interfacing with silica boosts the catalysis of copper, Nat. Commun. 9 (2018)
for NO reduction, J. Catal. 187 (1999) 1–14.
[54] J. Ding, T. Popa, J. Tang, K.A.M. Gasem, M. Fan, Q. Zhong, Highly selective and
3
367.
2
stable Cu/SiO catalysts prepared with a green method for hydrogenation of
[
36] H. Xi, X. Hou, Y. Liu, S. Qing, Z. Gao, Cu-Al spinel oxide as an efficient catalyst
for methanol steam reforming, Angew. Chem. Int. Ed. 53 (2014) 11886–11889.
37] H. Yang, Y. Chen, X. Cui, G. Wang, Y. Cen, T. Deng, W. Yan, J. Gao, S. Zhu, U.
Olsbye, J. Wang, W. Fan, A highly stable copper-based catalyst for clarifying the
catalytic roles of Cu0 and Cu+ species in methanol dehydrogenation, Angew.
Chem. Int. Ed. 57 (2018) 1836–1840.
diethyl oxalate into ethylene glycol, Appl. Catal. B Environ. 209 (2017) 530–
542.
[55] Y.S. Yun, T.Y. Kim, D. Yun, K.R. Lee, J.W. Han, J. Yi, Understanding the reaction
[
mechanism of glycerol hydrogenolysis over a CuCr
10 (2016) 442–454.
2 4
O catalyst, ChemSusChem
[56] E.S. Vasiliadou, A.A. Lemonidou, Kinetic study of liquid-phase glycerol
hydrogenolysis over Cu/SiO catalyst, Chem. Eng. J. 231 (2013) 103–112.
[57] Z. Zhou, X. Li, T. Zeng, W. Hong, Z. Cheng, W. Yuan, Kinetics of hydrogenolysis
[
38] G. Kresse, J. Furthmuller, Efficiency of ab-initio total energy calculations for
metals and semiconductors using a plane-wave basis set, Comput. Mater. Sci. 6
2
(
1996) 15–50.
of glycerol to propylene glycol over Cu-ZnO-Al
Eng. 18 (2010) 384–390.
[58] J.R. Copeland, G.S. Foo, L.A. Harrison, C. Sievers, In situ ATR-IR study on
2 3
O catalysts, J. Chin. J. Chem.
[
[
[
[
[
39] G. Kresse, J. Furthmuller, Efficient iterative schemes for ab initio total-energy
calculations using a plane-wave basis set, Phys. Rev. B 54 (1996) 11169–11186.
40] P.E. Blöchl, Projector augmented-wave method, Phys. Rev. B 50 (1994) 17953–
aqueous phase reforming reactions of glycerol over
catalyst, Catal. Today 205 (2013) 49–59.
2 3
a Pt/gamma-Al O
1
7979.
41] G. Kresse, D. Joubert, From ultrasoft pseudopotentials to the projector
augmented-wave method, Phys. Rev. B 59 (1999) 1758–1775.
42] J.P. Perdew, K. Burke, M. Ernzerhof, Generalized gradient approximation made
simple, Phys. Rev. Lett. 77 (1996) 3865–3868.
[59] F. Vila, M. López Granados, R. Marisca, Significance of isomeric reaction
intermediates in the hydrogenolysis of glycerol to 1,2-propanediol with Cu-
based catalysts, Catal. Sci. Technol. 7 (2017) 3119–3127.
[60] V.A. Yaylayan, S. Harty-Majors, A.A. Ismail, Monitoring carbonyl-amine
reaction and enolization of 1-hydroxy-2-propanone (acetol) by FTIR
spectroscopy, J. Agr. Food Chem. 47 (1999) 2335–2340.
43] J. Gong, H. Yue, Y. Zhao, S. Zhao, L. Zhao, J. Lv, S. Wang, X. Ma, Synthesis of
0
+
2
ethanol via syngas on Cu/SiO catalysts with balanced Cu –Cu sites, J. Am.
Chem. Soc. 134 (2012) 13922–13925.
44] L.F. Chen, P.J. Guo, M.H. Qiao, S.R. Yan, H.X. Li, W. Shen, H.L. Xu, K.N. Fan, Cu/
[61] M.T. Hernández, M. González, Synthesis of resins as alpha-alumina precursors
by the Pechini method using microwave and infrared heating, J. Eur. Ceram.
Soc. 22 (2002) 2861–2868.
[
SiO
2
catalysts prepared by the ammonia-evaporation method: Texture,