One-pot direct conversion of levulinic acid into high-yield valeric acid over a highly stable bimetallic Nb-Cu/Zr-doped porous silica catalyst

Article abstract of DOI:10.1039/c9gc03516h

The direct conversion of levulinic acid (LA) to valeric biofuel is highly promising for the development of biorefineries. Herein, LA is converted into valeric acid (VA) via one-pot direct cascade conversion over non-noble metal-based Nb-doped Cu on Zr-doped porous silica (Nb-Cu/ZPS). Under mild reaction conditions (150 °C and 3.0 MPa H₂ for 4 h), LA was completely converted into VA in high yield (99.8%) in aqueous medium with a high turnover frequency of 0.038 h^-1. The Lewis acid sites of ZPS enhanced the adsorption of LA on the catalyst surface, and both the Lewis and Br?nsted acidity associated with Nb₂O₅ and the metallic Cu⁰ sites promoted catalysis of the cascade hydrogenation, ring cyclization, ring-opening, and hydrogenation reactions to produce VA from LA. The bimetallic Nb-Cu/ZPS catalyst was also effective for the conversion of VA into various valeric esters in C1-C5 alcohol media. The presence of Nb₂O₅ effectively suppressed metal leaching and coke formation, which are serious issues in the liquid-phase conversion of highly acidic LA during the reaction. The catalyst could be used for up to five consecutive cycles with marginal loss of activity, even without catalyst re-activation.

Full text of DOI:10.1039/c9gc03516h

View Article Online
View Journal
Green
Chemistry
Cutting-edge research for a greener sustainable future
Accepted Manuscript
This article can be cited before page numbers have been issued, to do this please use: N. Karanwal, D.
Verma, P. Butolia, S. M. Kim and J. Kim, Green Chem., 2020, DOI: 10.1039/C9GC03516H.
Volume 20
Number
May 2018
Pages 1919-2160
This is an Accepted Manuscript, which has been through the
Royal Society of Chemistry peer review process and has been
accepted for publication.
9
7
Green
Chemistry
Cutting-edge research for a greener sustainable future
rsc.li/greenchem
Accepted Manuscripts are published online shortly after acceptance,
before technical editing, formatting and proof reading. Using this free
service, authors can make their results available to the community, in
citable form, before we publish the edited article. We will replace this
Accepted Manuscript with the edited and formatted Advance Article as
soon as it is available.
You can find more information about Accepted Manuscripts in the
Information for Authors.
Please note that technical editing may introduce minor changes to the
text and/or graphics, which may alter content. The journal’s standard
Terms & Conditions and the Ethical guidelines still apply. In no event
shall the Royal Society of Chemistry be held responsible for any errors
or omissions in this Accepted Manuscript or any consequences arising
from the use of any information it contains.
ISSN 1463-9262
PAPER
Paul T. Anastas et al.
The Green ChemisTREE: 20 years after taking root with the 12
principles
rsc.li/greenchem

Page 1 of 54
Green Chemistry
View Article Online
DOI: 10.1039/C9GC03516H
One-pot direct conversion of levulinic acid into high-yield valeric acid
over a highly stable bimetallic Nb-Cu/Zr-doped porous silica catalyst
a,§
a,b,c,§
c
d
* a,b,c
Neha Karanwal, Deepak Verma,
Paresh Butolia, Seung Min Kim, Jaehoon Kim
aSKKU Advanced Institute of Nanotechnology (SAINT), Sungkyunkwan University,
2
066 Seobu-Ro, Jangan-Gu, Suwon, Gyeong Gi-Do 16419, Republic of Korea.a
bSchool of Mechanical Engineering, Sungkyunkwan University,
2
066 Seobu-Ro, Jangan-Gu, Suwon, Gyeong Gi-Do 16419, Republic of Korea.
cSchool of Chemical Engineering, Sungkyunkwan University,
2
066 Seobu-Ro, Jangan-Gu, Suwon, Gyeong Gi-Do 16419, Republic of Korea
dInstitute of Advanced Composite Materials, Korea Institute of Science and Technology,
Chudong-ro 92, Bongdong-eup, Wanju-gun, Jeonranbuk-do, 55324, Republic of Korea.
df
§Equally contributed as first authors
*
Corresponding author, e-mail: jaehoonkim@skku.edu
1

Green Chemistry
Page 2 of 54
View Article Online
DOI: 10.1039/C9GC03516H
ABSTRACT
The direct conversion of levulinic acid (LA) to valeric biofuel is highly promising for the development
of biorefineries. Herein, LA is converted into valeric acid (VA) via one-pot direct cascade conversion
over non-noble metal-based Nb-doped Cu on Zr-doped porous silica (Nb-Cu/ZPS). Under mild
reaction conditions (150 °C and 3.0 MPa H for 4 h), LA was completely converted into VA in high
2
–
1
yield (99.8%) in aqueous medium with a high turnover frequency of ~300 h . The Lewis acid sites of
ZPS enhanced the adsorption of LA on the catalyst surface, and both the Lewis and Brønsted acidity
0
associated with Nb O and the metallic Cu sites promoted catalysis of the cascade hydrogenation, ring
2
5
cyclization, ring-opening, and hydrogenation reactions to produce VA from LA. The bimetallic Nb-
Cu/ZPS catalyst was also effective for the conversion of VA into various valeric esters in C1–C5
alcohol media. The presence of Nb O effectively suppressed metal leaching and coke formation,
2
5
which are serious issues in the liquid-phase conversion of highly acidic LA during the reaction. The
catalyst could be used for up to five consecutive cycles with marginal loss of activity, even without
catalyst re-activation.
Keywords: Biofuel, Levulinic acid, Valeric acid, Valerate esters, Copper, Niobium.
2

Page 3 of 54
Green Chemistry
View Article Online
DOI: 10.1039/C9GC03516H
1
. Introduction
The chemoselective conversion of cellulose or cellulose-derived sugars into transportation fuels
and value-added chemicals is one of the most promising strategies for meeting the current urgent
demand for renewable and sustainable energy and chemicals.1-4 During the last two decades, a number
of different pathways for converting cellulose into value-added chemicals have been explored.5, 6 One
of the most optimistic pathways is the conversion of C6 sugars to levulinic acid (LA) through 5-
hydroxymethyl furfural (5-HMF) via acid-catalyzed hydrolysis, followed by the transformation of LA
to -valerolactone (GVL) by hydrogenation, and of GVL to valeric acid (VA) by ring-opening and
subsequent hydrogenation (Scheme 1).7-10 VA can then be esterified to produce valeric acid alkyl esters
(VE, e.g., methyl valerate and ethyl valerate), which are recognized as a new type of cellulose-derived
biofuel.8, 11 Other potential applications of VA include in perfumes, cosmetics, food additives,
lubricants, and plasticizers, as shown in Fig. 1. Several studies have focused on the conversion of LA to
GVL over noble metals on neutral supports (e.g., Ru/C and Ru/TiO ) under relatively mild conditions
2
(
25–150 °C, 2–4.5 MPa H ).12-14 However, the ring-opening of GVL is a highly thermodynamically
2
–
1
unfavorable reaction that requires a large increase in the free energy (G = +25 kJ mol in water and
–
1
15, 16

G = +1.5 kJ mol in methanol).
Although numerous catalysts and reaction conditions have been explored for the conversion of LA
to GVL14, 17-21 and the conversion of GVL to VA and VE,22-25 the one-pot direct cascade conversion of
LA to VA and VE is a more economically viable and simpler approach, because the separation and
purification (which are typically high-cost, energy-intensive, and time-consuming steps) of reaction
intermediates (in this case, GVL) from the unreacted feed, byproducts, and catalysts can be avoided.
The direct conversion of cellulose or cellulose-derived C6/C5 sugars to LA via acid hydrolysis is could
be a feasible pathway,7, 26 but because of the environmental pollution and low product concentration
issues, hydrolysis of furfuryl alcohol is an industrially established pathway to produce LA.
3

Green Chemistry
Page 4 of 54
View Article Online
DOI: 10.1039/C9GC03516H
Consequently, the one-pot, direct conversion of LA to VA and VE would enable the development of a
practical-scale cellulosic biofuel process. A variety of multifunctional catalysts consisting of acidic
sites and metal sites and various reaction conditions (e.g., organic solvents, alcoholic solvents, and
water) have been explored for this purpose (Table S1).21, 25, 27-35 Most previous approaches are based on
the use of noble metal (Pt, Pd, or Ru) catalysts. The first example was reported by Lou et al., who
explored the conversion of LA to VA using 1 wt% Ru loaded on strongly acidic HZSM-5.21
A
moderate yield of VA + VE mixtures (45.8%) was obtained at 4.0 MPa H and 200 °C after 10 h of
2
reaction in dioxane. However, the Ru/HZSM-5 catalyst was quickly deactivated due to leaching of the
Ru and Al species during the conversion of highly corrosive and polar LA, thereby necessitating further
+
work. Later, the same group examined the effects of Ru precursors, extra-framework cations (H and
+
NH ), and the Si/Al ratio in the ZSM-5 support for the conversion of LA to VA at 200 °C and 4.0 MPa
4
H in dioxane over 10 h. Increasing the number of acid sites of the zeolite support enhanced the yield of
2
the VA + VE mixture to 91.3%; however, serious coke formation then decreased the yield of the VA +
VE mixture to 66.4%, and a reactivation step was required to recover the initial activity of the
2
7
catalyst. Pan et al. developed a Brønsted acid-supported Ru-based catalyst (5 wt% Ru/SBA–SO H)
3
3
5
for the conversion of LA into VA + VE; the high VA + VE yield (94%) at 240 °C and 4 MPa H after
2
6
h of reaction was attributed to the efficient conversion of GVL over the Brønsted acid sites. However,
a significant loss of sulfur (45.7%) from the SBA–SO H support during LA conversion resulted in a
3
loss of the catalyst acidity and a decrease in the yield of the VA + VE mixture. The use of non-noble
metals (Ni and Cu) deposited on a SBA–SO H support generated the intermediate GVL as the main
3
product. Kon et al. further studied the selective conversion of LA to VA over Pt/HMFI zeolite at
3
4
2
00 °C using 0.8 MPa H for 6 h under solvent-free conditions, achieving a high yield of 99%. When
2
LA conversion was carried out in methanol at 200 °C and 0.8 MPa H for 3 h, 87% methyl valerate was
2
produced. Recently, Zhou et al. reported that the use of metal triflates as organic Lewis acids and Pd/C
4

Page 5 of 54
Green Chemistry
View Article Online
DOI: 10.1039/C9GC03516H
2
8
resulted in a 92% VA yield at 150 °C and 5 MPa H in the reaction for 6 h in n-octane. However, it
2
was difficult to recover the soluble metal triflates after completing the reaction. In addition, Pd leaching
and the formation of coke on the catalyst decreased the yield of VA when the catalyst was reused.
Wang et al. encapsulated Pd nanoparticles in the mesopores of AlMCM-41 to prepare a robust catalyst;
the reaction at 270 °C and 6 MPa H for 10 h in n-octane furnished a high LA conversion (>99%), but
2
2
9
the VA yield was low (45%). The Pd-encapsulated catalyst showed marginal loss of activity during
four cycles of the catalytic reaction, indicating that Pd leaching and coke formation were suppressed.
Zhu et al. reported the hydrogenolysis of GVL to VA over the metal triflates + Pd/C catalyst at 135 °C
3
6
and 1 atm H for 12 h under a solvent-free condition ; the strong Lewis acid site of metal triflates and
2
oxophilic metal promoted the ring opening of GVL, and further hydrogenation proceeded by the H·
radical generated from Pd/C to produce VA.
In contrast with the numerous reports of noble metal-based bifunctional catalysts, only a few
3
0
studies have examined the use of non-noble metal-based catalysts. Sun et al. investigated the selective
hydrogenation of LA to VA and ethyl valerate over an encapsulated Co metal-based bifunctional
catalyst (Co@HZSM-5) using both batch and fixed-bed reactors at 240 °C under 3 MPa H in H O or
2
2
ethanol. In ethanol, 100% LA conversion and a VA + VE mixture yield of 97% were achieved. The Co
NPs encapsulated in the framework of the HZSM-5 zeolite showed high stability, as they suppressed
Co leaching and sintering under the liquid-phase reaction conditions. Later, the same group
3
3
investigated coke formation over an acidity-regulated K-modified Ni/HZSM-5 catalyst. After 80 h of
continuous LA conversion in a fixed bed reactor, a gradual reduction in the VA + VE mixture yield
was noted due to coke formation. Kumar et al. reported that the conversion of LA over Brønsted acid-
enhanced W-modified Ni/TiO232 and W-modified Ni/HZSM-5 at 200 °C and 0.1 MPa H in a fixed bed
2
reactor produced 36% VA yield and 26% VA yield, respectively.31
5

Green Chemistry
Page 6 of 54
View Article Online
DOI: 10.1039/C9GC03516H
As discussed above, the direct conversion of LA to VA or VE over heterogeneous catalysts
without isolating GVL remains a great challenge. The main issues that must be carefully addressed
include: (1) the production of mixture of VA and VE, as encountered in previous studies. This lack of
product selectivity necessitates additional highly energy-intensive and expensive separation procedures
to recover and purify the products. Therefore, the selective production of either VA or VE directly from
LA is highly desirable. (2) In most previous studies (Table S1), the conversion of LA was conducted in
organic solvents.21, 27-30, 35 Because LA is typically produced by the hydrolysis of monosaccharides,
7
polysaccharides, or even directly from lignocellulosic biomass in aqueous medium, it is logical to
perform LA conversion in the same aqueous medium to avoid the highly energy-intensive and
expensive recovery of LA from the aqueous reaction medium. In addition, the use of toxic organic
solvents in the practical-scale conversion of LA would produce a large amount of organic waste. (3)
The extreme scarcity and high cost of noble metals (Pt, 50.86 USD/g; Pd, 31.99 USD/g, Ru, 9.27
3
7
USD/g) are major drawbacks that prevent practical-scale biofuel production. Thus, the development
of a highly efficient and inexpensive non-noble metal-based catalyst is desirable. (4) The highly
corrosive nature of LA and the liquid-phase reaction conditions make it difficult to suppress metal
leaching during the reaction. Encapsulation of the metal sites in the support framework is a promising
solution for suppressing metal leaching,29, 30 but the inherently low active material loading and sluggish
diffusion of the reactants into the encapsulated active phase lower the catalytic activity. (5) To activate
the ring-opening of GVL, acidic sites such as those present on zeolitic supports, metal triflates,
2
7, 30, 33, 38
AlMCM-41, and SBA–SO H are promising,
but the instability of zeolitic supports under
3
hydrothermal conditions and the accelerated coke formation on the acid sites are major drawbacks.
Therefore, the acidity of the catalyst must be tuned to optimize the tradeoff between GVL activation
and coke formation. (6) Lastly, in most previous cases (Table S1), relatively harsh reaction conditions
were used in the conversion of LA (high hydrogen pressure, high temperature, extended reaction time,
6

Page 7 of 54
Green Chemistry
View Article Online
DOI: 10.1039/C9GC03516H
and high catalyst-to-feed ratios); such conditions are highly unfavorable for scale-up. Therefore, the
development of an economically viable, efficient, and robust non-noble metal-based catalyst for the
selective conversion of LA to VA or VE under mild reaction conditions in an environmentally benign
solvent is highly desired.
Herein, we demonstrate that a highly-efficient, robust, and non-noble metal-based bimetallic
catalyst, Nb-doped Cu on Zr-doped porous silica (Nb-Cu/ZPS), can directly convert LA almost
exclusively into VA (>99.8%) under mild reaction conditions (150 °C, 3 MPa H , 4 h, catalyst-to-feed
2
ratio of 0.125) in aqueous medium. The VA yield achieved in this study is unprecedentedly high
compared to those in previous studies (Table S1). The bimetallic Nb-Cu/ZPS catalyst is also highly
active for the conversion of VA to alkyl valerates in high yield under mild reaction conditions (150 °C
for 2 h) using various alcohols as the medium. The rationale behind the design of the bimetallic Nb-
Cu/ZPS catalyst in this study is as follows: (1) the incorporation of Zr into the porous silica framework
can create Lewis acid sites on the surface of the support; these sites can enhance the adsorption strength
4
+
39
of carbonyl (>C=O) group-containing substrates through coordination with the unsaturated Zr ions,
and (2) the doping of Nb as a promoter on Cu nanoparticles (NPs) not only creates Brønsted acid sites
when contact with water molecules,40-43 but also helps to reduce Cu leaching and aggregation of the Cu
NPs during the reaction.
2
. Experimental section
2
.1 Materials
–
1
Pluronic P-123 (molar mass: 5800 g mol ) and tetraethyl orthosilicate (TEOS, 98% purity) were
purchased from Sigma-Aldrich (USA). Zirconium oxychloride (ZrOCl , 98% purity), copper(II) nitrate
2
hemipentahydrate (Cu(NO ) ∙2.5H O, 98% purity), and niobium(V) pentachloride (NbCl , 99% purity)
3
2
2
5
were procured from Alfa-Aesar (USA). Hydrochloric acid (HCl, 37% aqueous solution), methanol
(99.9% purity), ethanol (99.9% purity), isopropanol (99.9% purity), tetrahydrofuran (THF, 99.5%
7

Green Chemistry
Page 8 of 54
View Article Online
DOI: 10.1039/C9GC03516H
purity), n-hexane (99.9% purity), N,N-dimethylformamide (DMF, 99.9% purity), and acetonitrile
ACN, 99.9% purity) were purchased from Daejung Chemicals (South Korea). LA (98.9% purity),
GVL (99.7% purity) and 2,5-hexanedione (2,5-HED, 98.9% purity) were procured from Sigma Aldrich
USA) and used as standards for quantification. The chemicals were used without further purification.
(
(
Distilled and deionized (DDI) water was prepared using an Aqua Max Basic 363 water purification
system equipped with a 0.22 μm filter (Young Lin Instrument Co., Ltd., South Korea). 5% H /Ar
2
(
99.999% purity), air (99.9% purity), He (99.999% purity), oxygen (99.999% purity), 10% NH /Ar
3
(
99.999% purity), and 10% N O/He (99.999% purity) were purchased from JC Company (South
2
Korea).
2
2
.2 Catalyst preparation
.2.1 Synthesis of Zr-doped porous silica (ZPS)
For the synthesis of ZPS, 0.5 g of Pluronic P-123, 4 mL of DDI water, and 13.3 mL of HCl were
mixed and stirred at room temperature until a clear solution was obtained. TEOS (1.15 mL) was then
added dropwise to the aqueous solution, and then ZrOCl ·8H O (0.162 g) was added into the mixture.
2
2
The whole mixture was then stirred at 35 °C for one day. The weight ratio of
P123:H O:HCl:TEOS:ZrOCl in the mixture was 2.9:19.4:69.9:6.7:0.94. The solution was
2
2
subsequently heated at 90 °C for one day in a beaker on a hot plate. The solid product was recovered by
filtration, then washed with DDI water, dried overnight at 100 °C, and calcined at 550 °C for 5 h under
–
1
air at a flow rate of 30 mL min to produce ZPS. The procedure for synthesis of porous silica (PS)
without Zr was the same as that for the production of ZPS, except that ZrOCl was not added. To
2
synthesize PS, the weight ratio of P123:H O:HCl:TEOS in the mixture was set at 3.33:22.4:66.7:7.67.
2
8

Page 9 of 54
Green Chemistry
View Article Online
DOI: 10.1039/C9GC03516H
2
.2.2 Synthesis of bimetallic Nb-doped Cu/ZPS catalysts
ZPS was loaded with Nb and Cu via a wet impregnation method using an aqueous solution of
Cu(NO ).2.5H O and NbCl . The amount of Nb and Cu loaded on the ZPS support was controlled by
3
2
5
adjusting the concentrations of NbCl (in the range 1–6 wt%) and Cu(NO ) ∙2.5H O (in the range 10–
5
3 2
2
5
0 wt%). An experimentally desired amount of Cu(NO ).2.5H O was dissolved in 10 mL DDI water
3 2
prepare Cu(NO ).2.5H O aqueous solution. Because of its high hydrolysis activity, NbCl was first
3
2
5
dissolved in 5 mL ethanol and then additional 5 mL DDI water was added to the NbCl ethanol solution.
5
Then the Cu(NO ).2.5H O and NbCl solutions were mixed together, and then the mixture was
3
2
5
dropwise added to pre-weighed ZPS. The reaction mixture was stirred at room temperature for 2 h, and
the liquid phase was then evaporated at 80 °C. The resulting solid product was calcined at 500 °C for 5
–
1
–1
h at a ramp rate of 2 °C min under 5% H /Ar gas flowing at 30 mL min . The synthesized catalysts
2
are designated as xNb-yCu/ZPS, where x and y represent the experimentally desired weight
percentages of Nb and Cu impregnated in the ZPS support, respectively.
2
.3 Catalyst characterization
The phase structure and crystallite size were investigated by powder X-ray diffraction (XRD) at a
−
1
scanning rate of 3–90° (scanning speed: 0.02° s ) using Cu-Kα (λ = 1.5418 Å) Ni-filtered radiation at
0 kV and 50 mA with a D/Max-2500V/PC Rigaku X-ray diffractometer (Japan). The specific surface
4
area, external surface area, and micropore volume of the catalyst were measured using a Belsorp-mini
II apparatus (BEL Inc., Japan). The composition of the samples and surface elements were analyzed by
X-ray photoelectron spectroscopy (XPS). XPS spectra were recorded using an ESCALAB250 (Thermo
Scientific, UK) spectrometer. The binding energy of the C–C bond at 284.8 eV was used as a reference
for the peak assignment. The acidity of the catalysts was examined by temperature-programmed
desorption of ammonia (NH –TPD) using a BELCAT instrument (BEL Inc.). Prior to ammonia
3
–
1
adsorption, the catalyst samples were pretreated at 250 °C for 3 h under a flow of He at 30 mL min .
9

Green Chemistry
Page 10 of 54
View Article Online
DOI: 10.1039/C9GC03516H
The samples were exposed to 10% NH in Ar for 1 h to facilitate the adsorption of NH at 100 °C.
3
3
Desorption of NH from the catalyst samples was carried out from 100 to 1000 °C at a ramp rate of
3
−
1
−1
1
0 °C min under He flowing at a rate of 30 mL min . To investigate the reduction properties of the
catalysts, temperature programmed reduction (TPR) was carried out by using a BELCAT instrument.
Before the H –TPR analysis, each catalyst sample (0.05 g) was subjected to heat treatment under Ar at
2
2
50 °C for 1 h. After cooling to 50 °C, the sample was oxidized at 50 °C for 30 min under oxygen
(99.9%) atmosphere. The catalysts were reduced by increasing the temperature to 900 °C at a heating
−
1
−1
rate of 10 °C min under a 5% H /Ar flow of 30 mL min . The hydrogen consumption was monitored
2
using a thermal conductivity detector (TCD). Pyridine Fourier-transform infrared spectroscopy (Py–
FTIR) was used to investigate the nature of the acid sites in the range of 1400–1700 cm–1 at a
–
1
resolution of 4 cm . All samples were pressed into self-standing wafers (15 mg sample + 30 mg KBr),
−
3
which were then pretreated at 400 °C under a vacuum of 10 mbar for 1 h prior to pyridine adsorption.
After injecting 5 µL of pyridine at 150 °C, the FT-IR cell was saturated for 10 min, followed by
evacuation at 150 °C for 30 min to remove excess pyridine. Each sample was scanned 32 times and the
Py–FTIR spectra were recorded at room temperature in the transmittance mode. The morphology of the
catalysts was determined using field-emission scanning electron microscopy (FE-SEM, Hitachi S-4100,
2
Japan) and high-resolution transmission electron microscopy (HR-TEM, Tecnai G FEI Co. Ltd., USA).
High-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) images
were collected using a Tecnai G2 instrument operated at 200 kV. The samples were suspended in
ethanol and deposited directly on a gold grid prior to analysis. To evaluate corrosion of the catalyst,
cyclic voltammetry (CV) was performed in a three-electrode cell configuration using a
potentiostat/galvanostat (Model VSP, BioLogic Science Instruments, France). The catalyst-containing
electrodes were prepared by suspending ~70 mg of the catalyst and 20 mg of carbon black in 10 mg of
polyvinylidene fluoride in N-methyl-2-pyrrolidinone. The slurry was then coated uniformly on a 1 cm2
1
0

Page 11 of 54
Green Chemistry
View Article Online
DOI: 10.1039/C9GC03516H
stainless steel coupon, which was dried overnight in a vacuum oven at 80 °C. An aqueous solution
–
1
(
85.0 mmol L ), similar to that used for LA conversion, was used as the electrolyte solution. The
reference and counter electrodes were Ag/AgCl and Pt foil, respectively. Prior to the CV measurements,
the electrolyte solution was degassed with high-purity N . The electrodes were cycled in the voltage
2
–
1
window of +0.1 to –0.5 V at a scan rate of 20 mV s . A Q50 thermogravimetric analyzer (TGA, TA
Instruments, USA) was used to investigate coke deposition on the spent and fresh catalysts under an air
−
1
−1
flow of 60ꢀmLꢀmin , at temperatures ranging from 30 to 800ꢀ°C, at a heating rate of 10ꢀ°Cꢀmin . The
loading of Cu and Nb on the ZPS support was measured by using an Optima 7300 V inductively
coupled plasma-optical emission spectrometer (ICP–OES).
The surface area of the Cu NPs loaded on the ZPS support was measured via N O chemisorption
2
at 75 °C. Typically, a flow of 5% H /Ar was used to reduce the catalyst at 350 °C for 30 min, and
2
–
1
subsequently, the reactor was cooled from 350 to 75 °C under a He gas flow at 30 mL min . Thereafter,
0
.5 mL of N O was injected via pulse mode through a 988 µL sample loop, and the decomposition of
2
N O to N on the surface of the metallic Cu NPs was examined using a gas chromatograph (GC)
2
2
equipped with a TCD and a column (activated carbon, 250 mm, BELCAT). The surface density of Cu
–
3
was set to 8.96 g cm for measurement of the surface area of the metallic Cu NPs. The N O pulse
2
titrations were replicated to avoid possible sub-surface oxidation of the catalyst, which is typically
caused by excess pulses. A maximum of three or four injections of N O pulses was carried out for each
2
0
experiment. The surface area of Cu was calculated using the N O titration method based on the
2
equation: N O(g) + 2Cu(s) → Cu O(s) + N (g).
2
2
2
CO-diffuse reflectance infrared Fourier transform spectroscopy (CO-DRIFTS) analysis was used
to investigate the nature of the metal catalytic sites, the oxidation state of the metals, and coordination
mode of the as-synthesized catalysts. A PerkinElmer Frontier spectrometer equipped with a mercury
cadmium telluride (MCT) detector and a Harrick Praying Mantis cell (USA) was used for this purpose.
1
1

Green Chemistry
Page 12 of 54
View Article Online
DOI: 10.1039/C9GC03516H
Prior to analysis, the catalyst samples (mixed with KBr) were pretreated at 150 °C under N at a flow
2
–
1
rate of 40 mL min for 1 h. The catalyst was then reduced at 450 °C under H flowing at a rate of 40
2
–
1
mL min for 4 h and subsequently flushed with N for 30 min at the same temperature. The samples
2
were then cooled to ambient temperature, and background spectra were collected. The sample was then
–
1
exposed to a controlled stream of 5 vol% CO in He at a rate of 50 mL min for 1 h. The samples were
flushed with N to remove non-adsorbed CO molecules and the IR spectra were collected at ambient
2
temperature. All spectra were converted into Kubelka-Munk units.
To gain insight into the possible interactions between the adsorbate and adsorbent, an adsorption
study was conducted at room temperature. The role of Zr in the ZPS support as an active center for
interaction with different types of oxygenated functional groups in the reactant was evaluated by using
LA, VA, and 2,5-HED as the selected adsorbents. For the adsorption study, 20 mL of a 10000 ppm
aqueous solution of each adsorbent was separately transferred to two 25 mL vials, and 100 mg of the
PS or ZPS support was added to each vial. The mixtures were sonicated for 10 min and were left
undisturbed to allow the catalyst to settle. The catalyst was separated, and the liquid product was
analyzed using high-performance liquid chromatography (HPLC).
2
.4 Catalyst activity and product analysis
The catalytic reactions for the conversion of LA to VA were performed in a 140 mL stainless steel
batch reactor equipped with a mechanical stirrer. Prior to the LA conversion experiments, the as-
synthesized catalysts were reduced at 500 °C under H /Ar atmosphere for 5 h with a temperature ramp
2
–
1
of 2 °C min . In a typical experiment, 3.4 mmol of LA, 50 mg of the activated catalyst, and 30 mL of
DDI water were added to the reactor. The reduced catalyst was immediately charged into the reactor,
and the reactor was sealed and purged three times with H and then pressurized to the experimentally
2
desired H pressure at room temperature to suppress air oxidation of Cu NPs. The reactor was heated to
2
various temperatures in the range of 120–200 °C for varying reaction times of 1–4 h with a fixed
1
2

Page 13 of 54
Green Chemistry
View Article Online
DOI: 10.1039/C9GC03516H
stirring rate of 600 rpm. After the reaction, the reactor was rapidly quenched with cold water to cool the
reactor to room temperature and then depressurized to atmospheric pressure. 1,6-Hexanediol as an
internal standard was added to the reaction mixture. The entire reaction mixture was then collected in a
beaker by washing the reactor with DDI water (20 mL × 2). The used catalyst was separated from the
liquid product via filtration with CHMLAB qualitative filter paper (C1.F093.055, Spain). Prior to the
HPLC analysis of the liquid product, the product was filtered through a 0.45 µm syringe filter. An
HPLC (Waters e2695) equipped with a Bio-Rad Aminex HPX-87H column (300 × 7.8 mm) and
Waters 2487 UV detector (220 nm) was used to quantify the reaction products. The HPLC column
temperature was set at 40 °C. The injection volume was 10 µL, and the mobile phase was 0.0013 N
–
1
H SO at a flow rate of 0.5 mL min . The conversion of LA and the yield and selectivity of VA were
2
4
calculated from the HPLC data using Eqs. 1–3:
푚표푙푒푠 표푓 푟푒푎푐푡푒푑 퐿퐴
퐶표푛푣푒푟푠푖표푛 (%) =
(
1 ― 푚표푙푒푠 표푓 푖푛푖푡푖푎푙 퐿퐴
)
× 100
(1)
(2)
푚표푙푒푠 표푓 푉퐴
푌푖푒푙푑 (%) =
× 100
푚표푙푒푠 표푓 푖푛푖푡푖푎푙 퐿퐴
푚표푙푒푠 표푓 푉퐴
푆푒푙푒푐푡푖푣푖푡푦 (%) =
× 100
(3)
푚표푙푒푠 표푓 퐿퐴 푐표푛푣푒푟푡푒푑
For comparative purposes, the LA conversion in organic solvents was also evaluated. The reaction
products collected in organic solvents were analyzed using an Agilent 7890N GC instrument with a time-of
flight mass spectrometer (TOF-MS) detector. The GC/TOF-MS was equipped with an auto-injector system
(Agilent 7860N) and a Rxi-5Sil-MS column (30 m × 0.25 mm × 0.25 μm). Typically, 1 μL of the reaction
products recovered after the reaction was injected into the column in split mode (50:ꢁ1). The injector
temperature was set at 250 °C and the transfer line temperature was fixed at 260 °C. The initial temperature
of the column was set at 40 °C with a holding time of 2 min. The temperature was increased to 300 °C at a
−1
ramping rate of 10 °C min . The detectable mass range for the MS analysis was set at 35–650 m/z. An
Agilent 6890N instrument equipped with a flame ionization detector (FID) and a Rxi-5Sil-MS capillary
column (30 m × 0.25 μm × 0.25 mm) was used for quantitative analysis of the products collected from the
1
3

Green Chemistry
Page 14 of 54
View Article Online
DOI: 10.1039/C9GC03516H
reactions in organic solvent. A 1.0 μL aliquot of the product sample was injected at the inlet temperature of
50 °C with the detector temperature set to 250 °C, using a split ratio of 1:10. The initial column temperature
was maintained at 40 °C for 2 min, and thereafter ramped to the final temperature of 240 °C at a rate of
2
−1
1
0 °C min . The conversion of LA and the yield and selectivity of VA were calculated from the GC-
FID results using Eqs. 1–3. After the reaction, the number of moles of unreacted reactant and products
were quantified from the calibration curves constructed using standard reference compounds by using
HPLC analysis for the reactions in aqueous medium and by GC-FID for those in organic solvents (see
Fig. S1).
3
. Results and discussion
3
.1 Catalyst characterization
Fig. 2a shows the XRD patterns of the porous silica (PS), Zr-doped porous silica (ZPS), and the
bimetallic Nb-Cu/ZPS catalysts with varying Nb and Cu loadings. The XRD pattern of the PS support
exhibited a broad peak centered at 23.2°, which is characteristic of amorphous SiO .44 The
2
incorporation of Zr into the porous silica did not change the crystallinity of PS, and no peaks associated
with ZrO were observed in the profile of ZPS, indicating that the Zr species were well-incorporated
2
into the silica framework or that the particle size of ZrO was below the detection limit of XRD (2
2
4
5
–1
nm). As shown in Fig. 2b, the appearance of the stretching vibration of the Si–O–Zr bond at 965 cm
in the FT-IR spectrum of ZPS, which was not observed in the spectrum of PS, confirmed the
4
6
incorporation of Zr into the SiO framework. The XRD patterns of the Nb-Cu/ZPS catalysts showed
2
0
peaks at 43.3°, 50.2°, and 75.2° attributed to the (111), (200), and (220) planes of Cu (JCPDS file No.
4
-0836), respectively, and the peaks at 22.5°, 28.4°, and 36.4° were attributed to the (001), (100), and
4
7
(
101) planes of Nb O , respectively (JCPDS file No. 28-317). The average crystal sizes of the
2 5
0
metallic Cu NPs in the Nb-Cu/ZPS catalysts were evaluated by using the XRD peak corresponding to
the (111) plane at 43.6° and the Scherrer equation, and the results are listed in Table 1. When the Nb
1
4

Page 15 of 54
Green Chemistry
View Article Online
DOI: 10.1039/C9GC03516H
0
doping level was increased to 6 wt%, the crystallite size of Cu decreased from 35.0 nm (Cu/ZPS) to
4.3 nm (6Nb-40Cu/ZPS), indicating that Nb doping decreased the crystallite size of the Cu NPs. This
may be due to suppression of the growth of the Cu NPs in the presence of Nb O , as discussed in detail
2
2
5
in a later section. As shown in Fig. 2a, no shift of the Cu (111) peaks was observed in the XRD profile
of the bimetallic Nb-Cu/ZPS catalysts, indicating that alloy formation between the Cu and Nb phases
was negligible.
The textural properties of the bimetallic Nb-Cu/ZPS catalysts were evaluated from the
corresponding N adsorption–desorption isotherms. As shown in Fig. 2c, the isotherms of the ZPS
2
support exhibited type IV behavior with a H1 hysteresis loop in the p/p range of 0.5–1.0, indicative of
0
2
–1
a mesoporous structure. The BET surface area of the ZPS support was 658 m g . The external surface
2
–1
area of the ZPS support determined by the t-plot method was 554 m g , which corresponds to 84% of
the total BET surface area. This indicates that the ZPS support had a high surface area and large
number of exposed active sites, which would facilitate the transport of the bulky LA molecules to the
surface of the support. After loading with 40 wt% Cu, the BET surface area decreased slightly to 511
2
–1
m g . When the Nb doping level of the bimetallic Nb-Cu/ZPS catalysts was increased from 1 to 6
2
–1
wt%, the BET surface area decreased from 496 to 399 m g . In addition, as shown in Fig. 2d,
increasing the Nb doping to 6 wt% resulted in a significant reduction of the pore size to 2–10 nm, as
estimated using the BJH model. This indicates that the incorporation of Nb O suppressed the growth
2
5
0
of the metallic Cu NPs; consequently, some of the pores of the ZPS support were blocked by the ultra-
small Cu NPs, which nucleated near the surface of the pores.
N O pulse chemisorption analysis was used to further investigate the effect of Nb doping on the
2
size of the Cu NPs in the bimetallic Nb-Cu/ZPS catalysts (Table 2). When the Nb doping level was
increased from 0 to 6 wt%, the Cu dispersion increased from 2.8% to 3.5%, the surface area of the Cu
2
–1
NPs increased from 18.1 to 22.4 m g , and the particle size of Cu decreased from 37.1 to 30.0 nm.
1
5

Green Chemistry
Page 16 of 54
View Article Online
DOI: 10.1039/C9GC03516H
This indicates that the use of Nb O as a promoter led to an increase in the Cu surface area and Cu
2
5
4
8
dispersion. These results are consistent with the XRD and N adsorption–desorption results.
2
The FE-SEM images and elemental mapping data for the catalysts are shown in Fig. S2. The ZPS
support comprised spherical particles with an average size of 500–700 nm. EDS mapping of the ZPS
support illustrated uniform distribution of Zr throughout the ZPS framework. The overall morphology
of 4Nb-40Cu/ZPS was quite similar to that of the ZPS support, indicating that impregnation of the
support with the active metals did not change the morphology of the support significantly. The Nb and
Cu species were uniformly distributed over the surface of the support. To further investigate the
morphology of the catalyst, HR-TEM and HAADF-STEM images of the ZPS support, monometallic
4
Nb/ZPS and 40Cu/ZPS, and bimetallic Nb-Cu/ZPS catalysts were collected, as shown in Fig. 3. The
unique porous structure of the ZPS support was apparent from the HR-TEM and HAADF-STEM
images. No fringes associated with ZrO were observed in the FFT images, indicating that Zr was
2
uniformly distributed throughout the framework of the porous silica (Fig. 3a). The morphology of
4
Nb/ZPS was quite similar to that of the support (Fig. 3b), indicating that the low Nb O loading did
2 5
not change the morphology of the ZPS support. The 40Cu/ZPS catalyst comprised spherical Cu NPs
35–40 nm in diameter) uniformly distributed on the surface of ZPS (Fig. 3c). However, large areas of
(
the surface of ZPS were not covered with Cu NPs. In sharp contrast, for 4Nb-40Cu/ZPS, the surface of
ZPS was more widely covered by the Cu NPs (Fig. 3d). Compared to that of the Cu NPs in 40Cu/ZPS,
the surface of the Cu NPs in 4Nb-40Cu/ZPS was highly crumpled, and the size of the Cu NPs
decreased to 30–35 nm in the latter. The interplanar spacings of 0.208 nm and 0.180 nm (Figs. 3d and
0
S3) were attributed to the (111) and (200) planes of Cu , indicating the presence of Cu with a fcc
structure (JCPDS file number 04-0836). The interplanar spacing of 0.39 nm (Fig. S3a) indicates the
presence of monoclinic Nb O (JCPDS file number 04-0836). The HAADF-STEM images of 4Nb/ZPS,
2
5
4
0Cu/ZPS, and 4Nb-40Cu/ZPS (Fig. 3) and their EDS images (Fig. S3b) clearly revealed the uniform
1
6

Page 17 of 54
Green Chemistry
View Article Online
DOI: 10.1039/C9GC03516H
and homogenous coverage of the surface of ZPS by the Cu and Nb species in the bimetallic 4Nb-
0Cu/ZPS catalyst. The drastic change in the morphology of bimetallic 4Nb-40Cu/ZPS compared to
4
that of monometallic Cu/ZPS indicates that Nb doping significantly altered the nucleation and growth
of the Cu NPs. The mechanism of formation of bimetallic 4Nb-40Cu/ZPS is schematically represented
Fig. 3e. As discussed previously, some dopants (e.g., Zn, Zr, Nb, and Ga) can act as promoters and
influence the size and shape of the host nanocrystals by affecting the rate of crystal growth.49-51 During
crystal growth, the dopant is adsorbed on the facets of the host nanocrystals, thus changing the facet
energy, thereby altering the nucleation and subsequent growth rate. Thus, Nb incorporation
significantly altered the size and shape of the host Cu NPs.
The chemical composition of the bimetallic Nb-Cu/ZPS catalysts with different Nb doping levels
was examined using XPS; the high-resolution Zr 3d, Cu 2p, and Nb 3d spectra are presented in Fig. 4.
Prior to the XPS analysis, all the catalysts were reduced by treatment in an atmosphere of 5 vol% H /Ar
2
at 500 °C for 5 h. The Zr 3d spectra were fitted to two peaks at 182.8 eV and 185.2 eV, which
correspond to Zr 3d and Zr 3d , respectively These binding energies are higher than those of bulk
5
/2
3/2
.
5
2
ZrO at 182.2 eV (Zr 3d ) and 184.5 eV (Zr 3d ), as referenced to the C–C peak at 284.8 eV, and
2
5/2
3/2
5
3, 54
are close to those of ZrSiO (183.2 eV and 185.3 eV, respectively).
The shift of the Zr 3d peak to
4
5
3
higher binding energy may be due to the formation of Si–O–Zr bonds in ZPS; the shift in the binding
energies of Zr may be attributed to the higher electronegativity of Si compared to that of Zr. Therefore,
the shift in the Zr 3d peak confirmed that the Zr ions were very well incorporated into the silica
framework rather than being localized outside the framework as bulk ZrO . A doublet Cu 2p peak
2
3/2
was observed at 931.9 and 934.0 eV and a doublet Cu 2p peak was observed at 952.4 and 954.5 eV
1
/2
(Fig. 4b) in the Cu 2p spectra, with a binding energy difference between the Cu 2p3/2 and Cu 2p1/2
2
+
doublets of approximately 20.5 eV. The appearance of the Cu satellite peak at 942.9 eV in the spectra
0
of the reduced catalysts indicates that Cu was oxidized when exposed to air during sample collection
1
7

Green Chemistry
Page 18 of 54
View Article Online
DOI: 10.1039/C9GC03516H
2
+
and mounting in the XPS sample holder. The area percentage of the Cu peak at 934.0 eV decreased
significantly from 28.3% to 17.2% when the Nb content was increased from 1 to 6 wt%, indicating that
0
Nb acted as a promoter to suppress the re-oxidation of Cu . Because the binding energy difference
0
+
+
0
between Cu and Cu was very small (0.7 eV), it was difficult to differentiate between the Cu and Cu
peaks in the Cu 2p spectra. The Nb 3d spectra exhibited two peaks at 207.1 eV and 209.8 eV, which
3
/2
5
+
55
were assigned to the Nb species. This indicates that most of the Nb in the catalyst was present as
Nb O .
2
5
NH –TPD was used to characterize the active sites in the as-synthesized catalysts (Fig. 5a, Table
3
3
). The desorption peaks in the temperature ranges of 100–275 °C, 275–500 °C, and 500–600 °C can
be attributed to weak, medium, and strong acidic sites, respectively. The NH –TPD profile of the ZPS
3
support contained high-temperature desorption peaks in the 700–800 °C region, with a total acidity of
–
1
46,
0
.88 mmol g . The strong acidity of ZPS is due to the presence of Zr in the neutral SiO framework.
2
5
6
After loading Cu onto the surface of ZPS, low-temperature desorption peaks were observed, and the
high-temperature desorption peaks shifted towards the low-temperature direction. The wide peaks in
the TPD profiles of the bimetallic Nb-Cu/ZPS catalysts in the weak, medium, and strong acidity
regions were deconvoluted into five or six peaks. Overall, the total acidity of the bimetallic Nb-Cu/ZPS
–
1
catalysts increased from 0.83 to 1.08 mmol g when the Nb doping level was increased from 0 to 6
wt%.
Py–FTIR analysis was used to further investigate the types of acid sites in the catalyst (Fig. 5b).
–
1
The bands at 1450, 1550, and 1488 cm can be assigned to the Lewis acid (L), Brønsted acid (B), and
–
1
55, 57
L+B acid sites, respectively. The peaks at 1588 and 1608 cm were assigned to the L sites.
L sites
were predominant in the as-synthesized monometallic Cu/ZPS and bimetallic Nb-Cu/ZPS catalysts
–
1
(
coordinated pyridine: 1450 and 1608 cm ) with fewer B sites (protonated pyridine: 1550 and 1645
–
1
+
cm ). The presence of Lewis acid sites in the monometallic Cu/ZPS resulted from Zr in the ZPS
1
8

Page 19 of 54
Green Chemistry
View Article Online
DOI: 10.1039/C9GC03516H
5
+
support, and the Nb sites in the bimetallic Nb-Cu/ZPS catalysts contributed additional Lewis acid
sites. When the Nb doping level in the bimetallic Nb-Cu/ZPS catalysts was increased, a low intensity
–
1
band at 1550 cm began to emerge, indicating the presence of exposed B sites on the surface of the
catalyst. The B sites in the bimetallic Nb-Cu/ZPS catalyst resulted from the creation of highly polarized
Nb–O bonds in the distorted NbO octahedral and NbO tetrahedral arrangements in Nb O after co-
6
4
2
5
4
0, 43
–1
–1
ordination with H O molecules.
The total integral area of the bands at 1550 cm and 1450 cm
2
was used to quantify the L and B acid sites; the L/B ratios of the catalysts are presented in Table 3.
When the Nb doping level was increased from 1 to 6 wt%, the L/B ratio increased significantly from
3
.91 to 5.78. The enrichment of L sites with increasing Nb content is attributed to the presence of
5
+
40, 43, 58
unsaturated Nb cations in the exposed Nb O .
2
5
The reducibility of the active sites in the catalysts was evaluated by H –TPR analyses (Fig. 5c). A
2
small and broad peak centered at 690 °C was observed in the profile of the ZPS support; the low H2
uptake (0.06 mmol gcat–1) of the support indicates the absence of reducible metal on the surface of ZPS
4
+
59
as Zr is an irreducible cation below 800 °C. On the other hand, low-temperature H uptake peaks
2
were observed in the profiles of the monometallic Cu/ZPS and bimetallic Nb-Cu/ZPS catalysts,
indicating the reduction of easily reducible CuO species. As listed in Table 3, the H uptake increased
2
from 0.568 to 0.601 mmol gcat–1 when the Nb doping level of the bimetallic Nb-Cu/ZPS catalysts was
0
increased from 1 to 6 wt%. The reduction of highly-dispersed CuO to Cu is known to occur via a two-
2
+
1+
1+
step reduction process: Cu is reduced to Cu at temperatures below 250 °C and Cu is reduced to
0
60
2+
1+
Cu at temperatures above 250 °C. The strong interaction of Cu /Cu species with the ZPS support
6
0
may account for the high-temperature reduction peaks centered around 270–510 °C. The H –TPR
2
profile of 40Cu/ZPS showed four distinguishable reduction regions at temperatures of 270–510 °C,
indicating the presence of Cu ions in different chemical states in the catalyst; the two peaks at 274 °C
2
+
and 330 °C are attributed to the reduction of isolated Cu species and the reduction of highly-dispersed
1
9

Green Chemistry
Page 20 of 54
View Article Online
DOI: 10.1039/C9GC03516H
CuO crystallites located on the external surface and in the pores of the ZPS support, respectively. The
2
+
+
high-temperature peaks at 429 °C and 510 °C are assigned to the reduction of Cu –O–Zr or Cu –O–Zr
1
8
structures and the reduction of CuO in the bulk phase, respectively. Nb doping significantly changed
the reduction behavior of CuO. In the Nb doping range of 1−4 wt%, the number of reduction peaks was
reduced from four to three, with lowering of the reduction temperature. In contrast with monometallic
Cu/ZPS (510 °C), no high-temperature reduction peak was observed in the profile of bimetallic Nb-
Cu/ZPS, suggesting that the reduction temperature of bulk CuO decreased. This difference indicates
that the Nb dopant acted as a promoter to decrease the reduction temperature by suppressing the growth
of the Cu NPs, which is consistent with the XRD and the Cu dispersion results.
To further understand the reductive behavior of bimetallic Nb-Cu/ZPS, CO–DRIFT was employed
to investigate possible changes in the electronic states of Cu due to Nb doping. CO–DRIFT has
previously been used to explore the effects of the electronic structure on the performance of catalysts
6
1-63
comprising a metal (Mo, Ru, Pt, Rh) on acidic (zeolites, Nb O , TiO ) and neutral (SiO ) supports.
2
5
2
2
Fig. 5d shows the DRIFT spectra of CO adsorbed on 40Cu/ZPS, 4Nb-40Cu/ZPS, and Nb O . The CO–
2
5
DRIFT profile of monometallic 40Cu/ZPS showed two broad peaks of adsorbed CO at 2050–2220 cm–
1,
which were further deconvoluted into six assignable peaks. The region below 2130 cm (2127 and
–1
–
1
+
2
110 cm for the 40Cu/ZPS sample) is attributed to CO weakly adsorbed on metallic Cu in a linear
geometry through π-bond interaction.64, 65, 66, 67 The band at 2083 cm is attributed to CO adsorbed on
–1
0
68
highly dispersed nano-structured metallic Cu in the form of Cu –CO. On the other hand, the
–
1
adsorption peaks appearing at 2174 and 2153 cm are attributed to the symmetric and asymmetric
+
69
–1
stretching vibrations of Cu –(CO) species, respectively. The peak at 2200 cm is assigned to the
2
2
+
65
characteristic peak of Cu –CO species. After doping the 40Cu/ZPS catalyst with 4 wt% Nb, all the
–
1
deconvoluted peaks of CO adsorbed on Cu, except for the peak at 2200 cm , shifted to higher
wavenumber. This blue shift is caused by the Lewis-acidic Nb O , which lowers the electron density of
2
5
2
0

Page 21 of 54
Green Chemistry
View Article Online
DOI: 10.1039/C9GC03516H
the Cu NPs in 4Nb-40Cu/ZPS. In the case of Nb O , no CO adsorption peak was observed, which
2
5
confirms that the blue shift is caused by the Lewis acid characteristics of Nb O . The CO-DRIFT
2
5
results clearly indicate that the major species in the Cu NPs in the bimetallic 4Nb-40Cu/ZPS are
+
2+
+
Cu/Cu rather than Cu , which agrees well with the XPS data. The presence of highly populated Cu
species on the 4Nb-40Cu/ZPS catalyst can be explained by approximately 0.5 V above the valance
7
0, 71
0
+
band edge of Nb O .
Notably, the presence of Cu /Cu species can facilitate
2
5
hydrogenation/hydrodeoxygenation.72, 73 The decrease in electron density of Cu species in the 4Nb-
+
4
0Cu/ZPS catalyst can increase their reduction temperatures and keep the reducing metal in their high
oxidation state.74, 75 For example, the electron withdrawing characteristic of Mn O created low
2
4
+
0
75
electron density over Cu NPs, resulting in increasing the Cu fractions rather than Cu in Cu@Mn O .
2
4
+
Similarly, the Lewis acid characteristics of Nb O decreased the electron density of Cu species, which
2
5
would increase the reduction temperature. On the other hand, the H –TPR profiles revealed that the
2
bimetallic Nb-Cu/ZPS catalysts exhibited slightly lower reduction temperatures (Fig. 5c). This
discrepancy could be because of the decrease in Cu NPs size by the Nb O doping, which facilitate the
2
5
reduction reaction.
3
.2 Catalyst evaluation
The conversion of LA over the ZPS support, monometallic Cu/ZPS catalysts, and bimetallic Nb-
Cu/ZPS catalysts was examined at 150 °C using an initial H pressure of 3 MPa, a catalyst-to-feed ratio
2
of 0.125, and a reaction time of 2 h in aqueous medium; the results are listed in Table 4. The LA
conversion and GVL yield achieved with the ZPS support were 25.9% and 19.8%, respectively,
accompanied by the production of a small amount of -angelica lactone (-AL, see Fig. S4 for the
HPLC results). These observations suggest that the Lewis acid sites in the ZPS support were somewhat
active for the direct cyclization of LA. It was previously reported that Zr-based catalysts such as ZrO2
and UiO-66 are active in the conversion of LA to GVL in the presence of both external and internal
2
1

Green Chemistry
Page 22 of 54
View Article Online
DOI: 10.1039/C9GC03516H
hydrogen sources;18, 19 this is possible because Zr-containing catalysts with acidic Zr4+ sites can
activate the carbonyl group of LA.19, 39 The detailed reaction mechanism is discussed in Section 3.6.
After verifying the presence of active acid sites in the ZPS support for the conversion of LA, the effect
of impregnation of ZPS with Cu metal on the hydrogenation reaction was evaluated. After Cu loading,
both the LA conversion and GVL yield increased; when the Cu loading in the monometallic Cu/ZPS
catalysts was increased from 10 to 40 wt%, the LA conversion increased from 50.2% to 80.3% and the
GVL yield increased from 45.0% to 75.8% (entries 3–6, Table 4). Further increasing the Cu loading to
5
0 wt% resulted in a slight decrease in the LA conversion and GVL yield, plausibly due to the
decreased availability of the Zr-based Lewis acid sites required for the adsorption and conversion of
LA due to the extensive coverage of Cu on the ZPS surface. As indicated in Table 1, the BET surface
2
–1
area decreased significantly from 511 to 429 m g when the Cu loading was increased from 40 to 50
wt%. Even though GVL was produced in high yield over the monometallic Cu/ZPS, the targeted
product VA was not observed in the reaction mixture. The liquid mixture had a light brick red
coloration due to leaching of Cu metal from the monometallic Cu/ZPS catalysts (Fig. S5a).
When the Nb doping level in the bimetallic Nb-Cu/ZPS catalysts was increased from 1 to 4 wt%,
the LA conversion increased from 81.5% to 95.6% and the VA yield increased significantly from
2
1.0% to 92.5%, while the GVL yield decreased significantly from 58.0% to 2.2% (entries 8–11). As
indicated in Table S1, the VA yield over 4Nb-40Cu/ZPS was unprecedentedly high compared to those
of previously reported noble metal and non-noble metal-based catalysts, even under much milder
reaction conditions. In addition, the aqueous mixture collected after the reaction was completely
transparent and colorless (Fig. S5b), which demonstrates that Nb doping effectively suppressed
0
oxidation of the Cu NPs during the reaction. Leaching of the Cu-based catalyst is explained in detail in
Section 3.4. Further increasing the Nb doping level to 6 wt% (6Nb-40Cu/ZPS) did not change the LA
conversion or VA yield significantly because the increase in the Nb content from 4 to 6 wt% did not
2
2

Page 23 of 54
Green Chemistry
View Article Online
DOI: 10.1039/C9GC03516H
significantly affect the Cu dispersion (Table 2) or acidity (Table 3) of the bimetallic catalysts. To
investigate the effect of Nb on LA conversion, the reaction was conducted over the Cu-free catalyst, i.e.,
4
Nb/ZPS (entry 13). Under the optimized reaction conditions (150 °C, 3.0 MPa), 26.7% LA conversion
with GVL as the major product was achieved with 80.9% selectivity, indicating that Nb O effectively
2
5
facilitated the conversion of LA to GVL.
Having confirmed that Nb doping of Cu/ZPS could trigger the conversion of GVL to VA, the
activation energy for LA conversion and VA production were calculated using the time-course reaction
data and the Arrhenius equation, as shown in Figs. 6a and b, respectively. The activation energy for LA
–
1
conversion decreased significantly from 55.2 to 27.7 kJ mol when 4 wt% Nb dopant was added to the
0Cu/ZPS catalyst. When the Nb doping level in the 40Cu/ZPS catalysts was increased from 2 to 4
4
–
1
wt%, the activation energy for the production of VA decreased from 69.2 to 51.0 kJ mol . This
indicates that the bimetallic Nb-Cu/ZPS catalyst enhanced both the LA conversion and VA selectivity.
–
1
The turnover frequency (TOF, h ) of the bimetallic Nb-Cu/ZPS catalysts with varying Nb doping
levels was calculated using Eq. 4, and the results are shown in Fig. 6c.
푉퐴 푝푟표푑푢푐푒푑 (푚표푙푒)
푇푂퐹
(
ℎ ―1)
=
(4)
퐶푢 푙표푎푑푖푛푔 (푚표푙푒푠) × 퐶푢 푑푖푠푝푒푟푠푖표푛 × 푟푒푎푐푡푖표푛 푡푖푚푒 (ℎ)
Increasing the Nb doping level from 1 to 4 wt% induced an approximately four-fold increase in the
–
1
TOF from 0.011 to 0.038 h . Further increasing the Nb doping level to 6 wt% resulted in a slight
–
1
decrease in the TOF to 0.037 h . Based on the amount of total Cu metal in the bimetallic 4Nb-
4
0Cu/ZPS catalyst, TOF was 81.5 mmol VA g–1Cu h–1, a value which is much larger than TOF of noble
metal-based Pt/HMFI catalyst (33 mmol VA g–1Pt h–1)76. The LA conversion, VA selectivity, activation
energy, and TOF data clearly demonstrate the significant role of Nb doping in the one-pot conversion
of LA to VA.
To further investigate the roles of Zr and Nb in the bimetallic Nb-Cu/ZPS catalyst, the adsorption
of various types of molecules on the surface of PS, ZPS, and 4Nb-40Cu/ZPS was investigated. The
2
3

Green Chemistry
Page 24 of 54
View Article Online
DOI: 10.1039/C9GC03516H
adsorbents evaluated herein include LA, which has one carboxylic end group and one ketone group,
VA, which has one carboxylic end group, and 2,5-hexanedione (2,5-HED), which has two ketone
groups. VA and 2,5-HED were selected for evaluating the adsorption behavior of unifunctional
catalysts. As shown in Fig. 7a, the amount of molecules adsorbed increased significantly when Zr was
incorporated into the silica framework. This indicates that the Lewis acid sites associated with Zr+
enhanced the adsorption of oxygen-containing molecules. For the ZPS support, the adsorption followed
the order: VA ≈ LA >> 2,5-HED, indicating that the ketone group interacted less strongly with the
+
Zr -based-Lewis acid sites compared to the carboxylic acid group. The amounts of LA, VA, and 2,5-
HED adsorbed on the surface of the 4Nb-40Cu/ZPS catalyst decreased slightly compared to those on
+
the ZPS support because some of the Zr sites were covered with Cu NPs. Fig. 7b presents a plausible
mechanism for the adsorption and conversion of LA to generate 4-hydroxypentanoic acid (4-HPA).
+
The carboxylic acid group of LA is adsorbed on a Lewis acid site of Zr , and the LA ketone group is
hydrogenated to an alcohol group over the Cu site to produce 4-HPA. Based on the discussion thus far,
the adsorption order of the oxygen-containing molecules can be arranged according to the polarity of
the group; carboxyl group > hydroxyl group > ester group > ketone carbonyl group. To verify this, the
adsorption study was extended to GVL and D–α hydroxy isovaleric acid (HVA). We selected HVA
(which has both the carboxyl group and hydroxyl group as in the case of 4-HPA) because 4-HPA is not
available. As shown in Fig. S6, the amount of HVA adsorbed was much higher than GVL. This
indicates that 4-HPA does not desorb well from the catalyst surface, and thus is further dehydrated to
form GVL. In order to confirm the proposed reaction mechanism, LA conversion was conducted at a
low temperature of 100 °C under 1 MPa H in aqueous medium over the 40Cu/ZPS catalyst, and the
2
reaction mixture was analyzed using GC-TOF/MS. As shown in Fig. S7, 4-HPA was detected in the
reaction mixture, which clearly indicates that 4-HPA was a reaction intermediate formed by
hydrogenation of the ketone group of LA.
2
4

Page 25 of 54
Green Chemistry
View Article Online
DOI: 10.1039/C9GC03516H
3
.3 Optimization of the reaction conditions in aqueous medium
In the previous section, we discussed the effectiveness of 4 wt% Nb doping of Cu/ZPS for the one-
pot direct cascade conversion of LA to VA under mild reaction conditions in aqueous medium. Here,
the reaction conditions were varied to optimize the LA conversion and VA selectivity using the 4Nb-
4
0Cu/ZPS catalyst. Fig. 8a shows the effect of temperature on the LA conversion and product yields at
a catalyst-to-feed ratio of 0.125 and an initial H pressure of 3 MPa for 2 h of reaction. At the low
2
reaction temperature of 120 °C, 70.5% LA conversion, 45.7% VA yield, and 20.5% GVL yield were
obtained, indicating that this low temperature could activate LA conversion or the ring-opening of
GVL in the presence of the catalytic system. Further, as shown in Fig. S8, 3-pentenoic acid (3-PEA)
was detected in the reaction mixture, indicating that 3-PEA is a reaction intermediate formed by the
ring-opening of GVL. When the temperature was increased to 150 °C, the LA conversion increased to
9
5.6% and the VA yield increased to 92.5%, while the GVL yield was very low (2.2%). This indicates
that a relatively mild reaction temperature of 150 °C was sufficient to activate the conversion of LA
and ring-opening of GVL over the 4Nb-40Cu/ZPS catalyst. When the temperature was further
increased to 200 °C, complete conversion of LA and GVL was observed. On the other hand, the VA
yield leveled off at approximately 93% at temperatures of 150–200 °C because of the formation of
other products. These may include side products such as 1,4-pentanediol (1,4-PDO, which could be
produced by the breakage of the C–O bond of O=C–O in GVL and subsequent hydrogenation), methyl
tetrahydrofuran (MTHF, which can be produced by the dehydration of 1,4-PDO), pentanol (which can
be produced by the hydrogenation of VA ), and/or 5-nonanone (which can be produced by ketonization
between PA molecules). As shown in Fig. S9a, 1,4-PDO, MTHF, and 5-nonanone were observed in the
reaction mixture when THF was used.
The time-course of the reaction was monitored under conditions employing a temperature of
1
50 °C, initial H pressure of 3 MPa, and catalyst-to-feed ratio of 0.125 (Fig. 8b). As the reaction time
2
2
5

Green Chemistry
Page 26 of 54
View Article Online
DOI: 10.1039/C9GC03516H
approached 1 h, the yield of the intermediate GVL increased to 29.4%, while the VA yield was low
18.7%). Further increasing the reaction time to 2 h resulted in a significant decrease in the GVL yield
(
to 2.2% and an increase in the VA yield to 92.5%. Further increasing the reaction time to 4 h did not
decrease the VA yield, indicating that VA was stable in the reaction mixture under these mild
conditions.
The effect of pressure on the conversion of LA at 150 °C with a catalyst-to-feed ratio of 0.125 in
the reaction for 2 h is shown in Fig. 8c. Low H pressures of 1–2 MPa were not sufficient for the one-
2
pot conversion of LA to VA. At 1 MPa H , the major product was GVL, while at 3 MPa H , the LA
2
2
conversion increased to 95.6% and the VA yield increased to 92.5%. Further increasing the H pressure
2
to 4 MPa resulted in a slight decrease in the VA yield to 89.7% due to the formation of side products
such as 1,4-PDO.
When the catalyst-to-feed ratio was decreased from 0.125 to 0.0625 at 150 °C using 3 MPa H for
2
2
6
h, the LA conversion decreased from 94.6% to 73.0%, and the VA yield decreased from 92.8% to
0.3% (Fig. 8d). On the other hand, when the catalyst-to-feed ratio was increased two-fold from 0.125
to 0.250, almost complete conversion of LA to VA (99.1% yield) was achieved. Thus, it can be
concluded that at the higher catalyst loading, complete conversion of LA to VA could be achieved,
while the generation of other products was suppressed. When the LA concentration was increased from
1
.0 to 1.6 wt% at 150 °C using 3 MPa H for 2 h, the LA conversion increased from 88.9% to 96.1%,
2
and the VA yield increased from 62.2% to 90.1% (Fig. 8e)
3
.4 Solvent screening and Cu metal leaching
As discussed in the Introduction, the use of water as the reaction medium in the conversion of LA
is expedient because of its environmental friendliness. To investigate the other beneficial effects of
water as a reaction medium, LA conversion was conducted in different solvents over 40Cu/ZPS and
4
Nb-40Cu/ZPS at 150 °C under 3 MPa H using a catalyst-to-feed ratio of 0.125, over the course of 2 h
2
2
6

Page 27 of 54
Green Chemistry
View Article Online
DOI: 10.1039/C9GC03516H
(
Fig. 9). The LA conversion in the various solvents followed the order: N,N-dimethylformamide
DMF) ≈ water (>95%) > methanol (80.5%) > tetrahydrofuran (THF, 75.5%) > n-hexane (60.1%).
(
Although high LA conversion was achieved in DMF, the product mixtures contained a large amount of
residual GVL (58.3%), and thus the targeted VA yield was much lower (15.8%) than that achieved in
water (92.5%). When n-hexane was used, the low solubility of LA in the hydrophobic solvent resulted
in low LA conversion. The selectivity for VA achieved with bimetallic Nb-Cu/ZPS was much higher in
aqueous medium than in organic solvent because the enhanced Brønsted acidity of Nb O in aqueous
2
5
medium facilitated the conversion of the intermediate GVL by activating the ring-opening reaction.40, 43
In THF, the side products 1,4-PDO, 5-nonanone, and MTHF were also detected (Fig. S9); methyl
valerate was observed in methanol; dihydro-3,5-methyl-2(3H)-furanone was observed in DMF; and 4-
hydroxy-2-pentenoic acid was observed in hexane.
In light of the previous observation of leaching of Cu metal from the Cu/ZPS catalyst in aqueous
medium, Cu leaching in the various solvents with different polarities (including isopropanol (IPA),
methanol, ethanol, THF, acetonitrile (ACN), and DMF) was evaluated (Fig. 9b). After the reaction, the
IPA, methanol, ethanol, and THF solutions were blue due to dissolution of the metallic Cu NPs from
2
+
2+
the catalyst to generate solvated Cu ions. The dissolved Cu ions could be stabilized by forming
chelates with the carboxylic acid groups of LA in the reaction media. In ACN and DMF, the reaction
mixture was brownish because N-containing solvents such ACN and DMF can act as reducing agents
or protecting agents to stabilize the Cu NPs.77, 78 The color change in the organic solvents indicates that
the Cu NPs of the monometallic Cu/ZPS were highly susceptible to leaching during the conversion of
LA. On the other hand, as shown in Fig. 9a, the reaction over 4Nb-40Cu/ZPS generated transparent and
colorless solutions in all the solvents tested in this study. This indicates that Cu leaching from the
bimetallic Nb-Cu/ZPS catalyst was effectively suppressed, which was further confirmed by ICP–OES
(Fig. S10). To further evaluate the Cu leaching, the redox properties of the monometallic Cu/ZPS and
2
7

Green Chemistry
Page 28 of 54
View Article Online
DOI: 10.1039/C9GC03516H
bimetallic Nb-Cu/ZPS catalysts were evaluated via CV measurements in aqueous LA solution. The CV
profiles of the Cu NPs, 40Cu/ZPS, 8.0 wt% Nb loaded on Cu NPs (8Nb92Cu), and 4Nb-40Cu/ZPS are
shown in Fig. 9c. The CV profile of the Cu NPs exhibited two broad redox peak pairs centered at –
2
+
0
.14/–0.24 V during the cathodic and anodic sweeps, which are attributed to the reduction of Cu to
0
0
2+
Cu and oxidation of Cu to Cu , respectively. 40Cu/ZPS exhibited a similar broad redox peak pair
centered at –0.21/–0.31 V; the shift of these peaks to more negative potential is attributed to the change
in the electronic conductivity of the Cu NPs when loaded on ZPS. On the other hand, no redox peak
pair was observed for 8Nb92Cu and 4Nb-40Cu/ZPS, indicating that the redox reaction caused by Cu
+
leaching was effectively suppressed. This suggests that the Cu species in the Cu NPs in the 4Nb-
+
2+
4
0Cu/ZPS catalysts are highly stable in the presence of Nb O , and thus the oxidation of Cu to Cu is
2 5
7
0
effectively suppressed. As discussed previously, active metal leaching is one of the most deleterious
issues in the conversion of LA in various media.18, 21, 25, 27-29, 79-81 Leaching of the base metal can be
effectively suppressed by re-oxidation of the metal during the reaction, which can be achieved either by
facilitating H transfer to the catalyst surface or by doping the host material with metals or metal oxides
2
to tune the electronic states, which can promote dispersion and reduction of the metals and enhance the
interaction between the active sites and the support.17, 50 For example, Zhang et al. reported the
conversion of LA into GVL over a bimetallic Cu-Ag (1:1 ratio) catalyst; doping with an equimolar
amount of Ag suppressed Cu leaching by maintaining Cu in the metallic state during the reaction in
THF.
17
3
.6 Reaction pathway
Based on the discussion thus far, a plausible reaction pathway for the direct conversion of LA to
VA over the bimetallic Nb-Cu/ZPS catalyst was proposed (Scheme 2). As discussed in the previous
sections, the reaction intermediates formed in the reactions involving incomplete conversion of LA
include 4-HPA, -AL, GVL, and 3-PEA. The direct conversion of LA to VA involves several steps
2
8

Page 29 of 54
Green Chemistry
View Article Online
DOI: 10.1039/C9GC03516H
that depend on various factors such as the concentration and species of the acidic sites, the metallic
sites available for hydrogenation, the solvent polarity, and the reaction parameters (temperature,
pressure, and catalyst-to-feed ratio). In the case of the bimetallic Nb-Cu/ZPS catalyst, the carbonyl

group of LA is first adsorbed on the Lewis acid sites in the ZPS support (Zr ). LA then undergoes
8
2
83
hydrogenation over the Cu NPs to form 4-HPA. In our recent previous study and a study by others,
Cu-based catalysts were found to be active towards the selective hydrogenation/hydrodeoxygenation of
aryl ketones. After the hydrogenation step, 4-HPA undergoes intramolecular cyclization to form GVL
8
4
via dehydration over the acid sites. During the cyclization of 4-HPA, one water molecule is removed
to form GVL, which is a very strong and hydrothermally stable intermediate. An alternative pathway
for producing GVL is intramolecular cyclization of LA and subsequent removal of water to produce -
AL. To verify -AL as an intermediate for producing GVL and VA, -AL was converted over the
bimetallic Nb-Cu/ZPS, and the results are shown in Fig. S11 and Table 4; high-yield VA was achieved
and residual GVL was observed, while other chemical species were not detected. The Brønsted acid
sites of Nb O that are generated in aqueous media can donate an activated hydrogen to GVL,
2
5
8
5
converting GVL into a stable carbenium ion and producing 3-PEA by proton elimination. The role of
Nb O as the Brønsted acid site was confirmed by the reaction over the monometallic Cu/ZPS catalyst
2
5
without Nb O , in which GVL was the major product (Table 4). In contrast, the bimetallic Nb-Cu/ZPS
2
5
catalyst yielded VA as the main product and GVL as a minor product. Brønsted acid sites are
4
0, 43
reportedly created in Nb O upon contact with water;
these sites can accelerate the conversion of
2
5
GVL to 3-PEA. It is noted that Lewis acid (e.g., metal triflates) is also effective in the ring opening of
GVL to produce 3-PEA.36 3-PEA is then converted into the final targeted product, VA, by
hydrogenation of the C=C bond over the Cu NPs. As reported previously, the conversion of GVL is the
rate-limiting step in the direct conversion of LA to VA because of the high stability of GVL. As shown
in Table 4, the conversion of GVL to VA was promoted when the Nb O doping level was increased to
2
5
2
9