Catalytic Response and Stability of Nickel/Alumina for the Hydrogenation of 5-Hydroxymethylfurfural in Water

Article abstract of DOI:10.1002/cssc.201501225

The catalytic response of Ni on Al₂O₃ obtained from Ni-Al layered double hydroxides was studied for the liquid-phase hydrogenation of hydroxymethyl furfural to tetrahydrofuran-2,5-diyldimethanol (THFDM) in water. The successive calcination and reduction of the precursors caused the removal of interlayer hydroxyl and carbonate groups and the reduction of Ni²⁺ to Ni⁰. Four reduced mixed oxide catalysts were obtained, consisting of different amount of Ni metal contents (47-68 wt %) on an Al-rich amorphous component. The catalytic activity was linked to Ni content whereas selectivity was mainly affected by reaction temperature. THFDM was formed in a stepwise manner at low temperature (353 K) whereas 3-hydroxymethyl cyclopentanone was generated at higher temperature. Coke formation caused deactivation; however, the catalytic activity can be regenerated using heat treatment. The results establish Ni on Al₂O₃ as a promising catalyst for the production of THFDM in water.

Full text of DOI:10.1002/cssc.201501225

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
DOI: 10.1002/cssc.201501225
Full Papers
Catalytic Response and Stability of Nickel/Alumina for the
Hydrogenation of 5-Hydroxymethylfurfural in Water
NoØmie Perret, Alexios Grigoropoulos, Marco Zanella, Troy D. Manning, John B. Claridge,
and Matthew J. Rosseinsky*[a]
The catalytic response of Ni on Al2O3 obtained from Ni–Al lay-
ered double hydroxides was studied for the liquid-phase hy-
drogenation of hydroxymethyl furfural to tetrahydrofuran-2,5-
diyldimethanol (THFDM) in water. The successive calcination
and reduction of the precursors caused the removal of interlay-
er hydroxyl and carbonate groups and the reduction of Ni2+ to
Ni0. Four reduced mixed oxide catalysts were obtained, consist-
ing of different amount of Ni metal contents (47–68 wt%) on
an Al-rich amorphous component. The catalytic activity was
linked to Ni content whereas selectivity was mainly affected by
reaction temperature. THFDM was formed in a stepwise
manner at low temperature (353 K) whereas 3-hydroxymethyl
cyclopentanone was generated at higher temperature. Coke
formation caused deactivation; however, the catalytic activity
can be regenerated using heat treatment. The results establish
Ni on Al2O3 as a promising catalyst for the production of
THFDM in water.
Introduction
As the production of biofuels and bioenergy from biomass re-
mains economically challenging, there is a growing demand
for the co-production of value-added chemicals to render the
process cost effective. Cellulose and hemicelluloses, the two
major components of lignocellulosic biomass, can be broken
down and converted to monosaccharides.[1] The subsequent
dehydration of six-carbon sugars can generate 5-hydroxyme-
thylfurfural (HMF), which is regarded as a primary renewable
building block.[2–5] The hydrogenolysis of biomass and the sub-
sequent dehydration are mainly carried out in water, but the
extraction of furanic products remains challenging.[3,6,7] There-
fore the use of water as solvent for the subsequent transfor-
mation of HMF is of great importance, and these reactions re-
quire heterogeneous catalysts that exhibit high activity, selec-
tivity and stability in aqueous solutions.
Scheme 1. Reaction pathway for the transformation of HMF to some of the
targeted compounds in the literature.
The conversion of HMF into value-added chemicals has re-
ceived increasing interest over the last decade[2,3,8,9] during
which a wide variety of heterogeneous metal (i.e., Ni, Cu, Pd,
Pt, Ru)-supported catalysts were studied.[10,11] Although most
research focused on CÀO hydrogenolysis (e.g., dimethyl furan)
and ring-opening products (e.g., hexanetriol, C5 and C6 poly-
ols), there are fewer reports on the selective hydrogenation of
HMF towards tetrahydrofuran-2,5-diyldimethanol (THFDM;
Scheme 1). This chemical finds application as a solvent (e.g.,
for the dehydration of fructose), a building block for the syn-
thesis of polymers (e.g., synthesis of polyesters via caprolac-
tone)[12] and high-value chemicals (e.g., 8-oxa-3-aza-bicy-
clo(3.2.1)octane hydrochloride).[13]
A number of factors have been proposed to influence the
catalytic response, including the nature of the catalysts and re-
action conditions, that is, the solvent,[14] pH value (acidic or
neutral),[15] hydrogen partial pressure,[16] temperature,[16] metal,
isoelectric point (IEP), acid and basic properties of the sup-
port.[7,15] Indeed, the nature of the metal affects the selectivity
for which Cu- and Ru-based catalysts can be selective for CÀO
hydrogenolysis with formation of dimethyl furan[17,18] whereas
polymerisation of the intermediate (furan-2,5-diyldimethanol
(FDM); Scheme 1) occurs over Pd and Pt catalysts.[15]
[a] N. Perret, A. Grigoropoulos, M. Zanella, T. D. Manning, J. B. Claridge,
Prof. M. J. Rosseinsky
Department of Chemistry
University of Liverpool
Crown Street, Liverpool, L69 7ZD (UK)
E-mail: m.j.rosseinsky@liv.ac.uk
On the other hand, incorporation of Ni on Pd/SiO2 and Co–
Al mixed-oxide catalysts results in concomitant hydrogenation
of the ring and formation of the saturated THFDM derivative,
Supporting Information for this article is available on the WWW under
http://dx.doi.org/10.1002/cssc.201501225. Underlying data can be ac-
cessed at http://dx.doi.org/10.17638/datacat.liverpool.ac.uk/61.
[11]
with a 96% yield over Ni–Pd/SiO2 and 89% over Ni–Co–Al
ChemSusChem 2016, 9, 521 – 531
521
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Full Papers
mixed oxides.[16] It is worth noting that Raney Ni[12,13,19] exhibits
100% selectivity towards THFDM, but it is less active than
other supported Ni catalysts (e.g., Ni–Pd/SiO2) for this reac-
tion.[11,16] It was also shown that supports with a high IEP (e.g.,
Al2O3) favour hydrogenation of the ring whereas supports ex-
hibiting Brønsted acidity, such as SiO2, generate polyols and
polymers via ring opening.[15,20] Despite these re-
ports,[11,13,15,16,20] which suggest that Ni metal and Al2O3 favour
the formation of THFDM, the hydrogenation of HMF over Ni
on Al2O3 has not been studied in a systematic manner.
effect of catalyst composition and reaction conditions is also
systematically investigated; the chemical kinetics of this reac-
tion and the catalytic stability of the catalysts are also consid-
ered.
Results and Discussion
Characterisation measurements
Layered double hydroxides precursors
The environment can cause degradation of heterogeneous
catalysts, and stability in water remains challenging as changes
in physical structure easily occur.[21] The studies over Raney Ni
were conducted in organic solvents such as alcohols (e.g., eth-
anol)[19] and ethyl acetate.[13] Nakagawa and Tomishige[11] tested
Ni–Pd/SiO2 as a catalyst for the same reaction in water and
they observed that 16% of Ni leached into the liquid phase
after 2 h.
Layered double hydroxides are composed of hydroxide bru-
cite-like sheets where two metals (M2+, M3+) occupy octahe-
dral sites; the presence of M3+ cations generates a net positive
charge, which is balanced by interlayers composed of anions
and water.[28] In this work, M3+ =Al3+, M2+ =Ni2+ have been
2À
introduced as cations whereas CO3 is the charge-balancing
II
1Àx
anion; the general formula can be written as [Ni AlIII (OH)2]
x
2À
[CO3
]x/2·mH2O. The precursors were synthesised by precipita-
Raney Ni can exhibit 100% selectivity towards THFDM. How-
ever, its preparation requires dissolving Ni in molten Al,
quenching and treatment of the alloy with NaOH (ca. 5m) to
leach Al out. As a result, a large amount of concentrated NaOH
is used during synthesis, the material is pyrophoric and its
chemical composition cannot be easily controlled. Therefore,
catalysts that are easier to synthesise and handle are required,
if the hydrogenation of HMF is to be carried out in a more en-
vironmentally benign process using water as the solvent. The
use of water as a solvent in sustainable chemical processes is
preferred as it is non-toxic, non-flammable, low cost, renew-
able and widely available.[22,23]
tion from an aqueous solution of NiCl2·6H2O and AlCl3·6H2O
using urea, based on the method of Costantino et al.[29] The
change in temperature (from 295 to 368 K at 1 KminÀ1) results
in the decomposition of urea according to Reaction (1).[30]
À
NH2ÀCOÀNH2 þ 3 H2O ! 2 NH4þ þ OHÀ þ HCO3
ð1Þ
The pH value progressively increased to circa 8.5, which is
suitable to precipitate the metal hydroxides. The urea hydroly-
sis takes place slowly and uniformly, so the precipitation is car-
ried out under a low degree of supersaturation.[31] The pre-
dominant species in the carbonate equilibria at final pH (ꢀ8.5)
is hydrogen carbonate,[32] providing the required interlayer
anion. Four layered double hydroxides (NiAl-P) were prepared
with different Ni-Al ratios (x=0.24, 0.28, 0.36, 0.47); the in-
creasing Z number (NiAl-ZP) relates to the increasing Al con-
tent (Table 1). The crystalline structure and textural properties
were investigated by inductively coupled plasma-optical emis-
sion spectroscopy (ICP-OES), X-ray diffraction (XRD), CHN analy-
sis, thermogravimetric analysis (TGA), scanning electron micros-
copy–energy-dispersive X-ray spectroscopy (SEM-EDX), Fourier
transform infrared (FTIR) and nitrogen physisorption experi-
ments.
There has been a growing interest, in recent years, in the
synthesis of mixed oxides by thermal pre-treatment of layered
double hydroxides (LDH) and their use for hydrogenation reac-
tions.[24] The mixed oxides obtained can exhibit distinct proper-
ties compared to traditional impregnated metal catalysts in-
cluding small crystallite size, large surface area, good stability,
high metal loading, basic properties and distinct catalytic per-
formance.[25–27] After reduction, the catalysts display well-dis-
persed metallic particles on the surface and enhanced metal–
oxide interaction.[24] Hence, the supported Ni catalysts used in
this study were derived from Ni–Al layered double hydroxides.
In this work, the catalytic response of Ni on Al2O3 catalysts
for the hydrogenation of HMF in water is presented. The cata-
lysts are prepared from layered double hydroxides of readily
available non-noble metals under mild conditions in water. The
The XRD pattern of the Ni–Al precursors (taking the sample
with the lowest Al content, NiAl-1P, as representative, Fig-
ure 1a) shows seven peaks (2q=13–788) that are characteristic
of Ni–Al layered double hydroxides.[33] The degree of crystallini-
Table 1. Composition, surface area and crystal parameters associated with the precursors.
[e]
Catalyst
Composition[a]
x[b]
SA[c]
[m2 gÀ1
a[d]
[Š]
c[d]
[Š]
dLDH
]
[nm]
NiAl-1P
NiAl-2P
NiAl-3P
NiAl-4P
Ni0.76Al0.24(OH)2(CO3)0.12·0.64H2O
Ni0.72Al0.28(OH)2(CO3)0.14·0.58H2O
Ni0.64Al0.36(OH)2(CO3)0.18·0.45H2O
Ni0.53Al0.47(OH)2(CO3)0.23·0.30H2O
0.24/0.26
0.28/0.28
0.36/0.34
0.47/0.51
65
67
84
3.061(1)
3.057(1)
3.048(1)
3.037(1)
23.46(1)
23.52(1)
23.37(1)
23.06(1)
10(1)
10(1)
10(1)
10(1)
107
[a] based on ICP, CHN and TGA analysis. [b] Al ratio determined from ICP analysis after digestion in HCl/from SEM-EDX measurements. [c] BET surface area
ꢀ
calculated from N2 adsorption–desorption isotherms. [d] Lattice parameters based on R3m space group, errors equate to 3s. [e] Crystallite size based on
Double-Voigt approach, errors equate to 1s.
ChemSusChem 2016, 9, 521 – 531
www.chemsuschem.org
522
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Full Papers
^
Figure 1. (a) XRD patterns of NiAl-1 precursor (black), calcined (red) and reduced (blue) showing also the main planes associated with Ni ( ), NiO (*) and hy-
drotalcite (~); (b) lattice parameter a of the precursors as a function of Al bulk molar ratio (x) where the linear fit obeys Vegard’s law; (c) TGA profiles for the
precursors (P) and (d) H2-TPR profiles (with temperature ramp in dotted line) generated for the calcined (C) samples of NiAl-1, NiAl-2, NiAl-3 and NiAl-4.
ty and textural properties depend on various reaction parame-
ters such as pH value, temperature, concentration and aging
of the precipitate.[26,34] The samples present good crystallinity
in agreement with the work of Costantino et al.,[29] who report-
ed improved crystallinity with longer digestion time (>36 h) at
temperatures of 363–373 K.
related to the interlayer and depends on factors such as the
nature and concentration of the anion, the state of hydration,
the strength of hydrogen bond between the anions and the
hydroxyl groups.[28] a is associated with the cation–cation dis-
tance in the hydroxide layer; as the radius of Al3+ (0.50 Š) is
smaller than Ni2+ (0.69 Š), a decreases with increasing Al con-
tent. Applicability of Vegard’s law can be tested (Figure 1b)
using Equation (2),
The FTIR spectra (taking NiAl-1P as representative, Figure S1
in the Supporting Information) show
a broad band at
3530 cmÀ1, which is mainly attributed to Al-coordinated OH
groups with small contributions of H-bonded interlayer water,
CO32À–H2O bridging and hydroxyl coordinated by both Ni and
Al.[35] The region between 1200–1800 cmÀ1 is characterised by
the bending mode of interlayer water around 1655 cmÀ1
(1640–1700 cmÀ1)[35] and interlayer carbonate at 1366 cmÀ1
(1350–1400 cmÀ1).[35] These results confirm the presence of H2O
a ¼ aNi þ x  ðaAl À aNiÞ
ð2Þ
where x and aNi and aAl correspond to the molar fraction of Al
and the lattice parameters of the pure constituents, respective-
ly. The linear fitting confirms the validity of Vegard’s law; more-
over, the extrapolation of aNi (x=0, aNi =3.08 Š) is close to that
of brucite-like Ni(OH)2 (3.11 Š).[37]
2À
and CO3 in the interlayer and are in agreement with hydrox-
ide sheets.
A representative SEM image using NiAl-4P as example is pre-
sented in Figure 2a where the lamellar/platelet morphology
observed for the four samples is characteristic of a well-devel-
oped layered structure. EDX measurements confirm the homo-
geneity of the samples, and the Al ratios (Table 1) are close
(Æ8%) to those obtained by ICP. The Brunauer–Emmett–Teller
(BET) surface areas (SA=65–107 m2 gÀ1, Table 1) are within the
range of values reported in the literature for Ni–Al layer
double hydroxides (15–98 m2 gÀ1)[38–40] and increase with the Al
content whereas the crystallite size (dLDH =10 nm) remains con-
stant.
The analysis by ICP of Al and Ni contents of the supernatant
(<0.1% metal in solution) and the precursors after digestion
(xAl, Table 1) verified that complete precipitation of the metal
hydroxides occurred. The CHN results were in agreement
(Æ5%) with the theoretical amount of CO32À and H2O in the in-
terlayer (according to the Miyata formula);[36] the compositions
of the four layered double hydroxides are presented in Table 1.
XRD patterns over a larger 2q range (108–1208) and with an
internal standard were measured to determine the lattice pa-
rameters (a, c; Table 1); they were obtained by fitting the pat-
ꢀ
terns using a R3m rhombohedral symmetry unit cell. c corre-
sponds to three times the thickness of a unit layer; hence, it is
ChemSusChem 2016, 9, 521 – 531
www.chemsuschem.org
523
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Full Papers
Catalyst pre-treatments
surface area of the calcined samples as it increases (Table 2)
with a decreasing Ni content.
The TGA curves of the four systems are shown in Figure 1c.
The first mass loss can be attributed to dehydration and partial
dehydroxylation whereas the second step is due to the loss of
interlayer hydroxyl and carbonate groups.[41] NiAl-4P has the
lowest amount of water and highest quantity of carbonate in
the interlayer (see composition, Table 1); hence, it has the
lowest mass loss during the first step and the highest during
the second step. The four layered double hydroxides have sim-
ilar temperatures for the two steps (ca. 373 and 550 K). The
loss of the interlayer groups is associated with an increase of
the surface area by 50–80%. Based on the TGA results, all sam-
ples were calcined at 773 K (denoted as NiAl-C). The XRD pat-
terns of NiAl-C after calcination (taking NiAl-1C as representa-
tive, Figure 1a) show three peaks (2q=43–758) that are charac-
teristic of cubic NiO. There is no detectable signal correspond-
ing to Al2O3 or spinel (e.g., NiAl2O4); this is not surprising as
higher calcination temperatures (>1100 K) are usually required
for their formation.[42] Thus, the samples are composed of
a mixture of an Al-rich amorphous component and a NiO-like
oxide. The decrease of the NiO lattice parameters (a, Table 2)
with increasing Al content and the lower values compared to
pure NiO (4.177 Š) suggest that Al3+ is incorporated into the
lattice, which is in agreement with literature.[26] The substitu-
tion of a divalent host (Ni2+) by a trivalent cation (Al3+) is com-
pensated by the formation of cation vacancies that balance
the charge.[43] The introduction of Al3+ is associated with
a slight decrease in NiO mean crystallite size (from 6 to 4 nm).
The Al-rich amorphous phase contributes significantly to the
The hydrogen temperature-programmed reduction (H2-TPR)
of the four calcined samples generated the profiles shown in
Figure 1d where positive peaks, corresponding to H2 consump-
tion, are evident. The amounts of H2 consumed (per gram of
material) are included in Table 2 and correspond (Æ4%) to the
amount of H2 required for the complete reduction of Ni2+ to
Ni0. The temperatures of the peaks at maximum signals (Tmax
)
are also included in Table 2 and range from 819 to 859 K. The
introduction of Al3+ into the NiO lattice enhances the stability
of Ni2+ and hinders its reduction; this is a well-established
phenomenon[26,42] and explains the shift of Tmax to higher tem-
perature with increasing Al content. Finally, the calcined sam-
ples were reduced under H2 (at 773 K) and passivated; after
these pre-treatments, the samples are denoted as NiAl-R.
Reduced mixed-oxide catalysts
The composition of the four catalysts (Table 3) was determined
by ICP, CHN and TGA analysis. As expected, the Al bulk ratios
of the reduced catalysts are similar (Æ5%) to the ones of the
precursors (NiAl-P, Table 1). The XRD patterns of NiAl-R (taking
NiAl-1R as representative, Figure 1a) show two peaks (2q=52–
618) that can be attributed to cubic nickel metal (ICDD 04-
0850). The samples must consist of Ni particles dispersed on
the Al-rich amorphous component as no Al-containing phases
were observed in the diffraction patterns. Ni0 mean crystallite
sizes (dNi, Table 3) are in the range of 10–14 nm, with an in-
crease in Ni loading corresponding to an increase in crystallite
size. Ni0 mean crystallite sizes of the reduced samples NiAl-R
are twice the size of NiO crystallites (NiAl-C, dNiO, Table 2); how-
ever, the presence of Al3+ in NiO lattices after calcination may
broaden the peaks due to strain and a direct comparison be-
tween crystallite sizes might not be valid.
Table 2. Ni content and particle properties after calcination as well as H2
consumed and maximum temperature during H2-TPR of the calcined
samples.
The morphology of the catalysts was investigated by per-
forming SEM and TEM imaging. Despite the modifications to
the composition during pre-treatment, the reduced catalysts
(Figure 2b) present a lamellar/platelet structure similar to that
of the precursor layered double hydroxides (Figure 2a); the re-
tention of the morphology after calcination and reduction has
been already reported elsewhere.[28,44,45] TEM-EDX analysis was
conducted on NiAl-4R as representative sample; the results ob-
tained suggest that the samples have a good homogeneity
(xAl =0.43Æ0.03) and the value obtained is in agreement with
[d]
Catalyst Ni content[a] SA[b]
[wt%]
[m2 gÀ1
a[c]
[Š]
dNiO
H2 consumed Tmax
]
[nm] [mmolgÀ1
]
[K]
NiAl-1C 68
117
121
142
162
4.159(1) 6.2(6) 11.5(2)
4.155(1) 5.1(4) 10.9(2)
4.149(1) 4.3(5)
4.145(1) 3.7(6)
819
828
850
859
NiAl-2C 61
NiAl-3C 58
NiAl-4C 47
9.5(2)
8.1(2)
[a] Based on ICP analysis after digestion in HCl. [b] Calculated from N2 ad-
sorption–desorption isotherms. [c] Lattice parameter based on Fm3m
space group, errors equate to 3s. [d] Based on Double-Voigt approach,
errors equate to 1s.
ꢀ
Table 3. Composition, particle properties and IEP associated with the reduced catalysts.
[b]
[d]
Catalyst
Composition[a]
Ni content[a]
[wt%]
SA[b]
[m2 gÀ1
PV[b]
[cm3 gÀ1
dpore
a[c]
[Š]
dNi
IEP
]
]
[nm]
[nm]
NiAl-1R
NiAl-2R
NiAl-3R
NiAl-4R
Ni0.76Al0.24O0.79H0.28
Ni0.73Al0.27O0.97H0.24
Ni0.64Al0.36O0.96H0.33
Ni0.53Al0.47O1.10H0.39
68
60
56
47
89
101
113
129
0.25
0.25
0.27
0.28
10
10
9
3.5227(3)
3.5226(4)
3.5241(4)
3.5217(8)
14(1)[e]
10.0(2)
10.2(2)
10.3(2)
10.1(2)
13(1)[e]
12(1)[e]
10(1)[e]/9(3)[f]
9
[a] Based on ICP, CHN and TGA analysis. [b] BET surface area, total pore volume and mean pore diameter calculated from N2 adsorption–desorption iso-
ꢀ
therms. [c] Based on Fm3m space group, errors equal to 3s. [d] Mean Ni crystallite size. [e] Based on Double-Voigt approach, errors equate to 1s. [f] From
TEM, error equates to 1s.
ChemSusChem 2016, 9, 521 – 531
www.chemsuschem.org
524
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim