Characterization and catalytic properties of Sn-modified rapidly quenched skeletal Ni catalysts in aqueous-phase reforming of ethylene glycol

Article abstract of DOI:10.1016/j.jcat.2006.05.001

Skeletal Ni (RQ Ni) catalyst was prepared by alkali leaching of rapidly quenched Ni₅₀Al₅₀ alloy. By impregnating RQ Ni with SnCl₄ followed by thermal treatment in inert atmosphere, homogeneous Sn-modified skeletal Ni (RQ N

Full text of DOI:10.1016/j.jcat.2006.05.001

Journal of Catalysis 241 (2006) 211–220
www.elsevier.com/locate/jcat
Characterization and catalytic properties of Sn-modified rapidly quenched
skeletal Ni catalysts in aqueous-phase reforming of ethylene glycol
Fuzhong Xie a, Xianwen Chu a, Huarong Hu a, Minghua Qiao a,∗, Shirun Yan a, Yuanlong Zhu a,
Heyong He a, Kangnian Fan a,∗, Hexing Li b, Baoning Zong c, Xiaoxin Zhang c
a
Department of Chemistry and Shanghai Key Laboratory of Molecular Catalysis and Innovative Materials, Fudan University, Shanghai 200433, P.R. China
b
Department of Chemistry, Shanghai Normal University, Shanghai 200234, P.R. China
c
Research Institute of Petroleum Processing, Beijing 100083, P.R. China
Received 25 January 2006; revised 28 April 2006; accepted 1 May 2006
Abstract
Skeletal Ni (RQ Ni) catalyst was prepared by alkali leaching of rapidly quenched Ni Al alloy. By impregnating RQ Ni with SnCl followed
50 50
4
by thermal treatment in inert atmosphere, homogeneous Sn-modified skeletal Ni (RQ Ni–Sn) catalysts were obtained. It was found that thermal
treatment induces alloying of the deposited metallic Sn with Ni, forming Ni Sn alloy segregated on the catalyst surface. In aqueous-phase
3
reforming of ethylene glycol, the RQ Ni catalyst was less selective in producing H but more selective in producing alkanes than the Raney Ni
2
catalyst described in the literature. This is attributed to the expansion of the lattice of RQ Ni favoring the dissociation of CO and, consequently,
the methanation reaction consuming H . Moreover, the structural difference influences the reaction pathway, with the undesired C–O cleavage
2
pathway obstructed on RQ Ni. Modification of RQ Ni with Sn drastically improves the H selectivity. On the RQ Ni Sn catalyst, alkane
2
80 20
production was virtually retarded, whereas H selectivity as high as 98 mol% was achieved at high conversion. Based on the characterizations
2
and previous findings, it is suggested that Sn may block the active sites for CO adsorption and/or dissociation, thus suppressing the undesired
methanation reaction. On the other hand, bifunctional Ni–Sn ensembles may form, in which Sn facilitates H O dissociation while neighboring Ni
2
adsorbs CO, thus promoting the desired water–gas shift reaction, leading to more H .
2
© 2006 Elsevier Inc. All rights reserved.
Keywords: Skeletal Ni; Rapid quenching; Sn modification; Aqueous-phase reforming; Ethylene glycol; H production
2
1. Introduction
a C:O stoichiometry of 1:1 and a H2 content relative to car-
bon >1. Their results indicate that catalytic APR of glucose,
sorbitol, glycerol, and ethylene glycol is feasible in generating
H2-rich fuel gas. In addition, the APR process is energy saving,
because vaporization of water and the hydrocarbons is not re-
quired. An even more attractive property is that the water–gas
shift (WGS) reaction is thermodynamically favored at the low
temperatures of APR, which can lead to lower CO content in
H2 than that achieved in steam reforming processes.
Several types of catalysts have been designed for the APR
process, including the supported Pt catalysts [5–8,11], SiO2-
supported metal catalysts [8], and Sn-modified Raney Ni cat-
alysts [9–12]. Among these, the Pt/Al2O3 catalyst and the Sn-
modified Raney Ni catalyst are the most promising. The high
H2 selectivity from the reforming of ethylene glycol can be
achieved over the Pt/Al2O3 catalyst [5,6], whereas the addition
H2 is considered a potential source of clean energy because
it does not emit greenhouse gases after direct combustion or
use in a H2 fuel cell. The current industrialized methods for
H2 production are based mainly on nonrenewable fossil fuels
[1–4]; thus H2 from renewable sources is highly desirable.
Recently, Dumesic and coworkers [5–13] demonstrated that
H2 can be produced from sugars and sugar alcohols (derived
from biomass) at temperatures near 500 K in a single-reactor
aqueous-phase reforming (APR) process, and that the selec-
tivity for H2 increases significantly when hydrocarbons have
*
Corresponding authors. Fax: +86 21 65642978.
E-mail addresses: mhqiao@fudan.edu.cn (M.H. Qiao),
knfan@fudan.edu.cn (K.N. Fan).
0021-9517/$ – see front matter © 2006 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2006.05.001

212
F.Z. Xie et al. / Journal of Catalysis 241 (2006) 211–220
of Sn is necessary to avoid alkane formation by the methanation
reaction over Raney Ni [9–12]. Although the Pt/Al2O3 cata-
lyst is more active in terms of specific activity, the Sn-modified
Raney Ni catalyst is less expensive, making it economically
more competitive than the former in the APR of ethylene gly-
col, a sustainable feedstock derived from biomass, sugars, and
sugar-alcohols [8]. Moreover, because Raney Ni is metallic,
it exhibits higher thermal conductivity compared with oxide-
supported reforming catalysts, which is crucial to the startup
and transient response of the reformer [14].
Ni95Sn5, RQ Ni90Sn10, and RQ Ni80Sn20, respectively. The re-
sulting RQ Ni–Sn catalysts were washed with distilled water
until free of chlorine ions and kept in water before activity test-
ing.
For characterizations, the RQ Ni and RQ Ni–Sn catalysts
were also washed with ethanol to replace water and finally
stored in ethanol. All of the measurements were made within
24 h of catalyst preparation. Because the catalysts are py-
rophoric, care must be taken to avoid air oxidation during sam-
ple handling.
In this paper, we extend the scope of the APR catalysts to
the rapidly quenched skeletal Ni catalyst (designated RQ Ni)
prepared by alkali leaching of the rapidly quenched Ni50Al50
alloy (Ni/Al, w/w, designated RQ Ni50Al50). It is noted that
the Raney Ni catalyst is leached from the naturally solidified
Ni50Al50 alloy. We reported previously that the RQ Ni catalyst
behaves differently in the hydrogenation of unsaturated organic
compounds compared with the Raney Ni catalyst irrespective of
their similar compositions, and attributed the underlying cause
to their discrepancies in textural and structural properties [15].
Thus, it will be interesting to explore the performance of the
RQ Ni catalyst in the APR of ethylene glycol. In contrast, the
Sn-modified Ni catalysts, either supported or unsupported, are
usually prepared by surface reaction of organotin compounds
(e.g., Sn(C4H9)3OAc [9–12], Sn(CH3)4 [16–18], Sn(n-C4H9)4
[19–21]) with Ni preadsorbed with hydrogen or in hydrogen
atmosphere. Here we used the less hazardous and inexpensive
SnCl4 and a simple impregnation procedure to deposit Sn on
the RQ Ni catalyst (designated RQ Ni–Sn). The effects of Sn on
the texture and structure of the RQ Ni catalyst were investigated
in detail. The catalytic behaviors of the RQ Ni–Sn catalysts in
APR of ethylene glycol were correlated and discussed based on
the characterizations.
2.2. Catalyst characterization
Because the melting point of metallic Sn is as low as 505 K
and the APR of ethylene glycol is usually conducted below
543 K, before characterization, the RQ Ni and RQ Ni–Sn cata-
lysts were subjected to thermal treatment at 543 K under inert
atmosphere for at least 1 h. The bulk compositions of the cat-
alysts were determined by inductively coupled plasma-atomic
emission spectroscopy (ICP-AES), using a Hitachi P-4010 in-
strument, after total dissolution in aqua regia. Their morpholo-
gies were observed by scanning electron microscopy (SEM),
using a Philips XL30 microscope. Sn distribution was obtained
on an energy-dispersive X-ray emission analyzer (EDX) at-
tached to the SEM apparatus.
The nitrogen isotherms and subsequent BET surface areas
(SBET) and pore distributions were measured by N2 adsorption
at 77 K on a Micromeritics TriStar3000 apparatus. The catalyst
was transferred to the adsorption glass tube with the storage liq-
uid and pretreated at 543 K under N2 (99.9995%) flow for 2 h.
It was weighed by difference in the adsorption tube on comple-
tion of the measurement.
Powder X-ray diffraction (XRD) was executed on a Bruker
AXS D8 Advance X-ray diffractometer using Cu-Kα radiation
(0.15418 nm), at a tube voltage of 40 kV and a current of
40 mA. Catalyst with solvent was heated in the hot stage to
543 K for 1 h, cooled to room temperature, and detected. All of
the operations were shielded by Ar (99.9995%) flow.
2. Experimental
2.1. Catalyst preparation
The RQ Ni50Al50 alloy was prepared by a single-roller melt-
spinning technique. A mixture containing equal weights of
metallic Ni and Al of >99.9% purity was melted at 1573 K in
an induction furnace for sufficient time to ensure homogeneity
of the melt. The ribbon (ca. 5 mm wide and 20 µm thick) was
fabricated by spraying the melt on a water-cooled high-speed
X-ray photoelectron spectroscopy (XPS) was performed on
a Perkin-Elmer PHI5000C instrument to determine the surface
electronic states and surface composition of the catalysts. After
being treated at 543 K under N2 (99.9995%) flow for 1 h, the
catalyst was immersed in ethanol, pressed into a self-supported
disc, and mounted on the sample plate. It was degassed in the
pretreatment chamber for 2 h at 383 K in vacuo before being
transferred to the analyzing chamber, in which the background
pressure was <2 × 10−9 Torr. All of the binding energy (BE)
values were obtained after the surface oxides were removed by
Ar ion sputtering and were referenced to the C 1s peak of con-
taminant carbon at 284.6 eV with an uncertainty of 0.2 eV.
H2 temperature-programmed desorption (H2-TPD) profiles
were obtained in the following manner. After the catalyst was
treated at 543 K for 1 h under Ar flow (99.9995%, deoxy-
genated by an Alltech Oxy-trap filter), it was cooled to room
temperature before saturation chemisorption of H2 by pulsed
injection, as confirmed by the constant eluted peak area mon-
itored with a thermal conductivity detector (TCD). The maxi-
rotating copper wheel with a cooling rate of ca. 107 Ks−1
.
The alloy was subsequently ground, and the fraction of 100–
200 mesh was used throughout the experiments.
The RQ Ni–Sn catalysts were prepared as described below.
Under gentle stirring, 1 g of the RQ Ni50Al50 alloy was added
to the NaOH aqueous solution (10 mL, 6.0 M) at 363 K. Af-
ter addition, the mixture was stirred at 363 K for 1 h, to achieve
further alkali leaching. The black powder was washed free of
alkali with distilled water. The as-prepared RQ Ni catalyst was
then soaked in 11 mL of ethanol solution containing a desired
amount of SnCl4 and stirred gently for 1 h at 303 K. The nom-
inal Sn/Ni molar ratios were 1/99, 2/98, 5/95, 10/90, and
20/80 for the catalysts labeled RQ Ni99Sn1, RQ Ni98Sn2, RQ

F.Z. Xie et al. / Journal of Catalysis 241 (2006) 211–220
213
mum desorption temperature, 850 K, was achieved at a ramp
Table 1
rate of 20 K min−1. The catalyst was weighed after being
cooled to room temperature. The active surface area (SH) was
calculated from the total volume of H2 desorbed by assuming an
equal distribution of atoms among the (100), (110), and (111)
planes of Ni and a H/Ni stoichiometry of unity.
Physicochemical properties of the RQ Ni and RQ Ni–Sn catalysts
Catalyst
RQ Ni
Bulk composition
(atomic ratio)
S
V
d
S
H
pore
3
pore
BET
2
−1
2
−1
−1
(m g
)
(nm) (m g
)
)
(cm g
Ni
Ni
Ni
Ni
Ni
Ni
Al
77
77
78
79
73
71
0.092
0.095
0.107
0.112
0.119
0.137
4.7
4.9
5.0
5.7
6.4
7.9
14.3
12.0
11.9
11.4
10.6
10.0
79.8 20.2
RQ Ni Sn
Sn Al
99
1
2
5
79.2 0.8 20.0
RQ Ni Sn
Sn Al
98
78.8 1.4 19.8
RQ Ni Sn
Sn Al
76.8 3.4 19.8
95
2.3. Activity test and product analysis
RQ Ni Sn
Sn Al
90 10
74.8 6.1 19.1
RQ Ni Sn
Sn Al
71.2 11.2 17.6
80 20
The reactor system for the APR of ethylene glycol was estab-
lished according to the setup described by Shabaker et al. [7].
The as-prepared RQ Ni or RQ Ni–Sn catalyst with water was
loaded in the stainless steel tubular reactor (6 mm i.d.). No ad-
ditional activation step of the catalysts was executed except for
the thermal treatment at 543 K for 1 h under Ar (99.9995%)
flow. An aqueous solution containing 5 wt% ethylene glycol
was fed to the reactor in an up-flow configuration. Ar was used
to regulate the system pressure. The reforming was typically
conducted with a catalyst mass of 0.5 g, a liquid hourly space
velocity (LHSV; volumetric flow rate of feed solution/catalyst
bed volume) of 3.60 h−1, a temperature of 498 K, and a system
pressure of 2.58 MPa, unless otherwise specified. The reaction
course was monitored by sampling the gas and liquid products
at intervals, followed by gas chromatography (GC) analysis.
For the gas effluent, hydrogen, carbon monoxide, methane, and
carbon dioxide were separated by a 5-Å molecular sieve packed
column and evaluated by a TCD. Methane, ethane, and propane
were separated by a Porapak R packed column and examined
by a flame ionization detector (FID). The liquid products, in-
cluding methanol, ethanol, acetone, 2-propanol, and ethylene
glycol, were separated by an HP-5 capillary column and exam-
ined by the FID. The liquid products were also qualified by gas
chromatography–mass spectroscopy (Finnigan Voyager) with
an HP-5 capillary column. According to Shabaker et al. [10],
H2 selectivity is defined as
Sn incorporated in the RQ Ni–Sn catalysts accordingly. How-
ever, the relative ratio of Sn between the amount in the result-
ing catalyst and the nominal amount kept decreasing from 98
to 41% when the nominal Sn/Ni molar ratio was increased
from 1/99 to 20/80. On the other hand, the BET surface areas
of the catalysts first increased marginally at a nominal Sn/Ni
ratio < 5/95, then decreased slightly at higher Sn/Ni ratios,
consistent with the finding that the BET surface areas of the
Raney Ni–Sn catalysts did not change significantly when trib-
utyl tin acetate was used as the Sn source [10]. Table 1 also
shows that the mean pore diameter and pore volume increased
monotonously with respect to Sn/Ni ratio.
Fig. 1 shows the SEM images and EDX results for the RQ
Ni99Sn1 catalyst. Note that the morphologies of the RQ Ni
and RQ Ni–Sn catalysts are similar. Fig. 1a shows that the
RQ Ni99Sn1 catalyst has a relatively flat surface and a rough
cross-section, likely inherited from the ribbon-like morphology
of the pristine RQ Ni50Al50 alloy. Fig. 1b shows that unlike
the smooth surface of the RQ Ni50Al50 ribbon (not shown),
the surface of the leached catalyst is highly fragmentized. The
cross-sectional images (Figs. 1a and c) clearly demonstrate that
the RQ Ni99Sn1 particle is composed of even smaller secondary
particles (diameter ca. 2 µm). These secondary particles have
round contours, in sharp contrast to the angular morphologies
of Raney Ni [15]. The addition of Sn decreased the dimensions
of the secondary particles; for the RQ Ni80Sn20 catalyst, the
size of the secondary particles was ca. 1 µm. It is possible that
the acidic SnCl4 solution dissolved additional residual metal-
lic Al and Ni from RQ Ni, leading to further fracturing of the
secondary particles.
H2 selectivity (%)
= (moles of H2 produced/moles of C in gas phase)
× (5/2) × 100,
(1)
which takes into account the WGS reaction, and alkane selec-
tivity is defined as
Alkane selectivity (%)
EDX experiments were performed to investigate the distrib-
ution of Sn in the RQ Ni–Sn catalysts. A typical result is shown
in Fig. 1d, demonstrating that the evolution of the Sn concen-
tration closely follows that of the Ni concentration, indicating
the homogeneity of the distribution of Sn associated with Ni.
Sn was also uniformly dispersed on the RQ Ni80Sn20 catalyst
with the highest Sn content studied here. This verifies that the
present SnCl4-impregnation method is a promising alternative
for the preparation of homogeneous Sn-modified Ni catalyst.
= (mol of C in gaseous alkanes
/total mol of C in gas products) × 100.
(2)
It should be noted that according to the definitions, the sum-
mation of H2 selectivity and alkane selectivity does not lead to
unity, because they are calculated based on independent hydro-
gen and carbon balances, respectively.
3. Results
3.2. Phase composition
3.1. Bulk composition and texture
Before being treated at 543 K in Ar atmosphere, all of the
RQ Ni and RQ Ni–Sn catalysts had XRD patterns similar to
that of fcc Ni, but with a slightly broadened peak width at higher
Sn content. No Sn-related phase emerged, implying the homo-
The bulk composition, BET surface area, pore volume, and
mean pore diameter of the RQ Ni and RQ Ni–Sn catalysts are
summarized in Table 1. Chemical analysis shows that the addi-
tion of more SnCl4 during preparation increased the amount of

214
F.Z. Xie et al. / Journal of Catalysis 241 (2006) 211–220
Fig. 1. (a) SEM image of the RQ Ni Sn catalyst, and the corresponding surface (b) and cross-sectional images (c). (d) X-Ray line profiles of Ni and Sn in the RQ
99
1
Ni Sn catalyst. The line drawn on (d) denotes the range being scanned.
99
1
the skeletal Ni catalyst has been linked to the more confined
relaxation of the Ni atoms in the Ni50Al50 alloy at a faster cool-
ing rate [15]. An alternative explanation may be that the lattice
expansion results from the defects introduced by rapid quench-
ing. Recent atomically resolved STM images of surface defects
on Ru(0001) have shown regions with lattice expansion of up
to 10% in the immediate vicinity of the defects [25]. Because
these two situations have the same origin, distinguishing be-
tween them can be difficult.
Besides the features arising from fcc Ni, the intensity of the
shoulder peak at 2θ of 44.8◦ was developed at higher Sn con-
tent, attributable to the (201) diffraction of hexagonal Ni3Sn
[22] due to alloying between metallic Ni and Sn during ther-
mal treatment. Thus, the asymmetric tail at the lower side of
the Ni(111) diffraction for the RQ Ni–Sn catalysts, which also
gained in intensity at higher Sn content, can be assigned to the
(002) diffraction of hexagonal Ni3Sn. The formation of Ni–Sn
alloys at 543 K was not unexpected; Onda et al. obtained Ni–Sn
alloys using chemical vapor deposition of Sn(CH3)4 on Ni/SiO2
at 448 K [17].
Fig. 2. XRD patterns of the RQ Ni and RQ Ni–Sn catalysts with different Sn
contents.
geneous dispersion and extremely small crystallite size of the
deposited metallic Sn. Thermal treatment noticeably changed
the XRD patterns. As shown in Fig. 2, all catalysts exhibited
diffractions at 2θ of 44.2, 51.4, and 75.7◦, corresponding to
the (111), (200), and (220) planes of fcc Ni, respectively [22].
Broadening of the Ni(111) peak was also observed with the
increment of Sn. It is noteworthy that the unit-cell parameter
of fcc Ni derived from these Bragg angles was ∼0.3551 nm,
larger than the values for Ni crystal (0.35238 nm) [23] and
Raney Ni (0.3528–0.3535 nm) [24]. The lattice expansion of
3.3. Surface composition
Fig. 3 illustrates the XPS spectra of the Ni 2p and Sn 3d lev-
els for the RQ Ni and RQ Ni–Sn catalysts. Fig. 3a shows only
one Ni 2p3/2 peak at 852.7 eV assignable to metallic Ni [26].
There is essentially no oxidized nickel species on these cata-
lysts, consistent with the observations of Fouilloux [27] and
Delannay et al. [28] on Raney Ni. In contrast, Fig. 3b reveals
two chemical states of Sn after curve fitting: metallic Sn, with

F.Z. Xie et al. / Journal of Catalysis 241 (2006) 211–220
215
Fig. 3. XPS spectra of (a) Ni 2p and (b) Sn 3d levels of the RQ Ni and RQ Ni–Sn catalysts.
Table 2
Surface compositions of the RQ Ni–Sn catalysts determined by XPS
0
0
δ+ a
)
Catalyst
Sn/Ni (mol%)
Sn /(Sn +Sn
(mol%)
Bulk
Surface
RQ Ni Sn
1.0
1.8
4.4
8.2
15.7
4.3
6.8
7.4
8.5
10.3
73
74
75
78
81
99
1
2
5
RQ Ni Sn
98
RQ Ni Sn
95
RQ Ni Sn
90 10
RQ Ni Sn
80 20
a
δ+
4+
2+
Sn represents Sn or Sn which cannot be differentiated by XPS.
a Sn 3d5/2 BE of 484.8 eV, and oxidized tin species (Sn4+ or
Sn2+), with a Sn 3d5/2 BE of 486.8 eV [26], with >70% of Sn
in the metallic state (Table 2).
Table 2 compares the bulk and surface Sn/Ni ratios in the
RQ Ni–Sn catalysts and shows that the surface Sn/Ni ratio
increased with the bulk one, with the former exceeding the
latter for the catalysts with Sn/Ni nominal ratios <10/90. At
higher Sn/Ni nominal ratios, the increase in surface Sn/Ni ra-
tio became less evident. For RQ Ni80Sn20, the surface Sn/Ni
ratio was even lower than the bulk one. Because, based on
the argument that metallic Sn has much lower surface energy
(710 mJ m−2) than Ni (2450 mJ m−2) [29], the surface en-
richment of Sn itself was not applicable for the lower surface
Sn/Ni ratio than the bulk one for RQ Ni80Sn20, the Ni3Sn
alloy was most likely segregated on the catalyst surface in
the form of island-like crystallites. It is anticipated that the
higher the Sn content, the larger the Ni3Sn crystallites, and
thus the portion of Sn beyond the detection depth of the XPS
technique is increased. This interpretation can satisfactorily
account for the evolution of the surface Sn/Ni ratios in Ta-
ble 2.
Fig. 4. H -TPD features of the RQ Ni and RQ Ni–Sn catalysts.
2
3.4. Surface sites
To obtain the dependence of surface active sites on Sn con-
tent, H2-TPD profiles were acquired; these are plotted in Fig. 4.
Fig. 4 shows that RQ Ni was characterized by two hydrogen
desorption maxima at ca. 401 and 637 K, indicating the pres-
ence of at least two kinds of active sites. Because the low-
temperature peak occurred in a temperature range similar to
that on Ni single crystals with low Miller indices [30], follow-
ing Arai et al. [31], we ascribe this to hydrogen chemisorbed
on low-Miller index planes of RQ Ni. On the other hand, the
high-temperature peak, which was dominant on RQ Ni treated
at lower temperature [15], can be assigned to defects or lattice
expansion. This assignment is in agreement with the theoretical

216
F.Z. Xie et al. / Journal of Catalysis 241 (2006) 211–220
calculation of Greeley et al., who identified that surface expan-
sion can noticeably improve the stability of surface hydrogen
on the Ni(111) surface [32]. It is conceivable that thermal treat-
ment at high temperature removes some defects, as evidenced
by the sharpened diffractions compared with the freshly pre-
pared RQ Ni, leading to the appearance of the low-temperature
peak.
Table 3
Comparison of the catalytic behaviors of RQ Ni and Raney Ni in APR of 5 wt%
ethylene glycol solution
a
a
Raney Ni [10] RQ Ni Raney Ni [10] RQ Ni
498 498 538 538
2.58 2.58 5.13 5.13
4.13 3.60 8.26 3.60
92 104 102
Temperature (K)
Pressure (MPa)
LHSV (h
Conversion of C to gas (%) 97
−1
)
Fig. 4 shows that after Sn modification, the low-temperature
peak was attenuated gradually and became indiscernible when
the nominal Sn/Ni ratio was 5/95. The reduced number of des-
orption peaks suggests that the active sites over the RQ Ni–Sn
catalysts became more uniform, which may have decreased the
adsorption modes of the reactant and suppressed side reactions.
At the first sight, considering that Sn itself does not adsorb hy-
drogen and that the amount of hydrogen adsorbed on the Ni3Sn
alloy is at least one order of magnitude lower than that ad-
sorbed on Ni [18], it appears that Sn or Ni3Sn preferentially
blocked the low-Miller index planes on RQ Ni, leaving the ad-
sorption sites for high-temperature hydrogen intact. However,
it was found that SH decreased only slightly after Sn modifica-
tion (Table 1); thus, we can also assume that more defects were
generated during Sn modification, as can be inferred from the
broadened Ni(111) diffraction in Fig. 2 and the smaller particle
size found on SEM. On the other hand, the loss of hydrogen due
to Sn occupation can be partly compensated for by the newly
generated defects and also may be impeded by the clustering of
Ni3Sn on the catalyst surface, as suggested by XPS.
H
selectivity (%)
35
44
35
47
28
47
21
59
2
Alkane selectivity (%)
Gas products (mol%)
H
47.6
30.4
0.02
20.7
0.97
47.4
30.8
n.d.
22.8
1.12
41.4
31.9
0.03
25.6
0.92
34.9
27.2
n.d.
36.4
1.43
2
CO
CO
2
b
b
Methane
Ethane
Propane
Butane
0.28
0.02
0.23
0.00
0.18
0.01
0.15
0.00
Liquid products (mol%, excluding unreacted EG)
Methanol
Ethanol
10.2
48.3
30.2
7.3
4.0
0.0
85.1
12.9
0.0
0.0
0.0
34.3
95.1
2.5
63.2
0.0
0.0
0.0
3.5
0.0
0.0
0.0
0.2
1.2
Acetic acid
Acetaldehyde
Glycolaldehyde
2-Propanol
Acetone
0.2
1.8
0.0
0.0
a
0.5 g of catalyst.
Below the detection limit.
b
hyde (the main liquid byproducts on Raney Ni) was identified in
the liquid products when RQ Ni was used as the catalyst. Acetic
acid is known to be troublesome due to its high stability in
aqueous solution [10] and corrosivity [12]. Moreover, reform-
ing of these liquid byproducts cannot be carried out with high
H2 selectivity, because methane is generated concurrently [10].
Instead, a small amount of 2-propanol and acetone were ob-
tained over RQ Ni, with the latter being the dehydrogenation
product of the former. Shorthouse et al. [34] identified selective
dehydrogenation of 2-propanol to acetone as the major reaction
pathway on the Ni(111) surface, and also reported that acetone
desorption competes very effectively with unselective decom-
position to smaller species.
3.5. Aqueous-phase reforming of ethylene glycol
3.5.1. Comparison between RQ Ni and Raney Ni
Before addressing the effect of Sn on the RQ Ni catalyst in
the APR of ethylene glycol to H2, we compared RQ Ni with
Raney Ni to elucidate the effect of rapid quenching on catalytic
performance. We found that the carbon balance was within 10%
for the catalytic runs.
Table 3 compiles the H2 and alkane selectivities, and gas
and liquid product distributions, at 498 and 538 K over the RQ
Ni and the previously reported Raney Ni catalysts [10]. The
table shows that RQ Ni exhibited comparable or lower H2 se-
lectivity and higher alkane selectivity than Raney Ni. However,
it is noteworthy that the RQ Ni catalyst produced less CO in
the outlet gas, below the detection limit of 50 ppm. This result
is significant especially in the field of H2 fuel cells, consid-
ering that the anode materials are usually vulnerable to CO
poisoning [33]. Based on the increased alkane concentrations
in the gas products, we propose that CO depletion on the RQ
Ni catalyst occurred mainly due to the enhanced activity in the
methanation or Fischer–Tropsch (F-T) reaction rather than in
the WGS reaction.
As for the liquid products, Table 3 shows that the RQ Ni cat-
alyst favored formation of lower-alcohol methanol rather than
ethanol, especially at high temperature. Shabaker et al. [10] re-
ported that methanol is a desirable product because it contains a
C–O linkage and may be further reformed to H2 at high yields.
For comparison, ethanol is preferred on Raney Ni particularly
at low temperature. No acetic acid, acetaldehyde, or glycolalde-
3.5.2. The effect of Sn
Fig. 5 shows the evolutions of H2 and alkane selectivities
against the amount of Sn incorporated in the bulk of the RQ
Ni catalyst. Similar to the promoting effect of Sn on Raney Ni
[9,10], the selectivity to H2 increased when the Sn content in-
creased in the RQ Ni–Sn catalysts. As shown, a bulk Sn/Ni
ratio as low as 1.0 mol% was sufficient to improve the H2
selectivity from 35% on RQ Ni to 54% on RQ Ni99Sn1. H2 se-
lectivity exceeded 90% when the bulk Sn/Ni ratio was >4.4%.
Meanwhile, alkane selectivity followed an opposite trend to that
of H2; on RQ Ni80Sn20, alkane selectivity was <1%, whereas
H2 selectivity reached as high as 98%.
Although Sn modification led to similar changes in H2 and
alkane selectivities on RQ Ni and Raney Ni, the kinetic data
compiled in Table 4 suggest that the modification mechanisms
were not the same for these two catalysts. The presence of
transport limitations under the reaction conditions specified in

F.Z. Xie et al. / Journal of Catalysis 241 (2006) 211–220
217
Fig. 5. Effect of bulk Sn/Ni molar ratio of the RQ Ni and RQ Ni–Sn catalysts on
Fig. 6. H production rate per gram of catalyst as a function of the weight of
2
the H and alkane selectivities in APR of ethylene glycol. Reaction conditions:
2
RQ Ni used in APR of ethylene glycol. Reaction conditions: 5 wt% of ethylene
−1
0.5 g of catalyst, 5 wt% of ethylene glycol aqueous solution, LHSV of 3.60 h
temperature of 498 K, and pressure of 2.58 MPa.
,
−1
glycol aqueous solution, LHSV of 22.6 h , temperature of 498 K, and pressure
of 2.58 MPa.
Table 4 was tested following the method proposed by Koros
and Nowak [35]. It has been claimed that the reaction rate per
gram of catalyst is proportional to S, the number of active sites
per unit volume of the reactor, in the absence of any transport
limitations. Thus, for the skeletal Ni catalysts, the reaction rate
per gram of catalyst should remain constant at various catalyst
weights in the kinetically controlled regime. As shown in Fig. 6,
the H2 production rates per gram of catalyst are almost identi-
cal irrespective of the weight of RQ Ni used, strongly indicating
that the data in Table 4 are intrinsic. According to Shabaker et
al. [10], for the Raney Ni–Sn catalysts, the turnover frequen-
cies (TOFs) of the alkanes decreased due to the addition of
Sn, whereas the TOFs of H2 were nearly unaffected. For the
present RQ Ni–Sn catalysts, Table 4 reveals that aside from the
decrement of the TOFs of the alkanes, the TOFs of H2 and CO2
increased steadily with the addition of Sn. It is also noted that
the TOFs as well as the H2 production rates over the RQ Ni
and RQ Ni–Sn catalysts are lower than those over the Raney Ni
and Raney Ni–Sn catalysts [10]. However, the deviations are
generally within one order of magnitude and may be due to the
experimental differences among laboratories.
Fig. 7. Effect of system pressure on the H and alkane selectivities for APR
2
of ethylene glycol over the RQ Ni and RQ Ni Sn catalysts. Other reaction
80 20
conditions: 0.5 g of catalyst, 5 wt% of ethylene glycol aqueous solution, LHSV
−1
of 3.60 h , and temperature of 498 K. Conversion is above 90% for all points.
from 35 to 11% and alkane selectivity rose from 47 to 57%.
Although the H2 and alkane selectivities on the RQ Ni80Sn20
catalyst evolved similarly to those on RQ Ni, the extents of
the changes were less remarkable; the H2 selectivities were al-
ways >90%, and the alkane selectivities were well below 10%,
demonstrating that RQ Ni80Sn20 is a H2-specific catalyst even
Fig. 7 compares the effects of system pressure on H2 and
alkane selectivities over the RQ Ni and RQ Ni80Sn20 catalysts.
The reaction temperature and the LHSV of ethylene glycol were
maintained at 498 K and 3.6 h−1. For RQ Ni, when the pressure
was increased from 2.58 to 3.62 MPa, H2 selectivity dropped
Table 4
a
The H and alkane selectivities and the reaction rates over the RQ Ni and RQ Ni–Sn catalysts in APR of ethylene glycol
2
−1
Catalyst
Conversion of C to gas
(%)
TOF (min
)
H
production
2
−3
−1
)
(µmol cm reactor min
H
CO
CH
C H
2
C H
3 8
2
2
4
6
RQ Ni
RQ Ni Sn
RQ Ni Sn
RQ Ni Sn
RQ Ni Sn
RQ Ni Sn
11.8
9.6
7.8
7.6
7.5
7.1
0.22
0.33
0.41
0.64
0.72
0.77
0.16
0.17
0.19
0.24
0.26
0.27
0.159
0.130
0.061
0.013
0.005
0
0.0069
0.0053
0.0021
0
0
0
0.0021
0.0002
0
0
0
0
21
28
33
51
54
57
99
1
2
5
98
95
90 10
80 20
a
−1
Reaction conditions: 0.5 g of catalyst, 5 wt% of ethylene glycol aqueous solution, LHSV of 22.6 h , temperature of 498 K, and pressure of 2.58 MPa.

218
F.Z. Xie et al. / Journal of Catalysis 241 (2006) 211–220
Based on the XRD results, we tentatively attribute the dif-
ferences in the catalytic behaviors of RQ Ni and Raney Ni in
APR of ethylene glycol to their structure differences due to lat-
tice expansion. The expanded lattice of the RQ Ni catalyst can
be expected to affect the reaction pathways of ethylene glycol,
thus altering the distributions of the gas and liquid products.
The higher alkane selectivity on RQ Ni can also be explained
in light of lattice expansion. For simplicity, we confine our dis-
cussion to the formation of methane, which is the main alkanic
product (see Table 3). Experimental and theoretical work has
demonstrated that CO dissociation on Ni is the rate-determining
step for the methanation reaction, CO + 3H2 → CH4 + H2O
[37–41]. Based on self-consistent density functional calcula-
tions, Mavrikakis et al. identified that when the lattice constant
parallel to the surface of Ru(0001) is increased or the coordina-
tion number is smaller, the d states have to increase in energy,
leading to stronger interaction with CO and a lowered dissocia-
tion barrier of CO [42]. The correlation between the interaction
strength and the d-band center can be generalized to other ad-
sorbates and metals. Because the lattice of RQ Ni is expanded
relative to that of Raney Ni, we expect the dissociation probabil-
ity of CO on RQ Ni to be enhanced, favoring the methanation
reaction. Although it may be argued that the enhanced inter-
action of CO with Ni will also promote the WGS reaction,
CO+H2O → CO2 +H2, which competes for CO and improves
H2 production, Vannice [43] found that the optimum metal–
CO interaction required for maximum WGS activity is weaker
than the methanation reaction. Thus, H2 selectivity is lower and
alkane selectivity higher on RQ Ni than on Raney Ni.
It is interesting to note that on going from left to right and
from 3d to 5d in the periodic table of transition-metal elements,
dissociative adsorption of CO is generally suppressed [44]. Dis-
sociative adsorption of CO is observed at high surface tem-
peratures on Ni surfaces, but not on Pt surfaces. This trend is
consistent with the finding that Pt is more favorable than Ni in
the APR of ethylene glycol to H2 [8,13], further proving the va-
lidity of our understanding of the effect of lattice expansion on
the reforming behavior of RQ Ni.
The effect of Sn on H2 selectivity over the Sn-modified
Raney Ni catalysts has been revealed and discussed in the pio-
neering work of Dumesic and coworkers [9–12]. These authors
have suggested that Sn may geometrically modify the catalytic
behavior of Ni by preferentially occupying the active sites for
CO adsorption and/or dissociation, or by destroying the sur-
face ensembles composed of multiple Ni atoms necessary for
CO dissociation [10], thus resulting in higher H2 selectivity and
lower alkane selectivity on the Raney Ni–Sn catalysts.
Fig. 8. Effect of LHSV on H and alkane selectivities over the RQ Ni and RQ
2
Ni Sn catalysts in APR of ethylene glycol. Other reaction conditions: 0.5 g
80 20
of catalyst, 5 wt% of ethylene glycol aqueous solution, temperature of 538 K,
and pressure of 5.13 MPa. Conversion is above 87% for all points.
under high pressure. In contrast, the activities of the catalysts
decreased with rising pressure, in line with the behavior of the
Raney Ni–Sn catalysts; Shabaker et al. [7] linked this effect to
inhibition by high product partial pressures and reactor dynam-
ics.
Fig. 8 shows the influence of the space velocity of ethylene
glycol on the H2 and alkane selectivities over the RQ Ni and RQ
Ni80Sn20 catalysts. The reaction temperature and the pressure
were fixed at 538 K and 5.13 MPa. It was found that the H2 and
alkane selectivities were affected less by the space velocity of
the feed than by the pressure. Moreover, the decreased space ve-
locity or longer residence time did not lead to increased alkane
selectivity, as was observed on Raney Ni–Sn [10]. Analysis of
the liquid products reveals that the decrease in the H2 selectiv-
ity on RQ Ni80Sn20 is due to the formation of more alcohols.
Figs. 7 and 8 clearly demonstrate that pressure and space veloc-
ity have a lower affect on the H2 selectivity for the RQ Ni80Sn20
catalyst than for the R-Ni14Sn catalyst [10], suggesting that the
formation of methane from CO2 and H2 is less favorable on RQ
Ni80Sn20.
4. Discussion
Referring to the reaction pathways outlined by Cortright et
al. [5] for the reaction of ethylene glycol with water, C–O cleav-
age is responsible for the formation of ethanol, whereas the
dehydrogenation/rearrangement pathway leads to the formation
of organic acids and aldehydes. According to Table 3, the dis-
tribution of the liquid products readily rules out the presence of
the undesired dehydrogenation/rearrangement pathway on RQ
Ni. Moreover, the high methanol/ethanol ratio strongly suggests
that C–O cleavage in ethylene glycol was virtually absent on
RQ Ni. It is not surprising that RQ Ni exhibited lower activity
toward C–O cleavage. We recently found that the temperature
required to break the C–S bond in thiophene was about 100 K
higher on RQ Ni than on Raney Ni [36].
Because over the RQ Ni and RQ Ni–Sn catalysts the con-
versions of ethylene glycol to gas products are always >92%,
and the liquid products only contains small portion of unre-
acted feed and <5% of byproducts in carbon under the reaction
conditions specified in Fig. 5, the reaction pathways of APR
of ethylene glycol can be reasonably simplified as Scheme 1,
which involves C–C cleavage, the methanation reaction, and the
WGS reaction. Competition between the methanation and the
WGS reactions determines the H2 selectivity. According to the
evolution of the TOFs of H2 and alkanes in Table 4, the above

F.Z. Xie et al. / Journal of Catalysis 241 (2006) 211–220
219
Acknowledgments
This work was supported by Fok Ying Tong Education Foun-
dation (grant 104022), the National Science Foundation of
China (grant 20421303), and Shanghai Science and Technol-
ogy Committee (grant 03QB14004).
Scheme 1. Simplified reaction pathway for APR of ethylene glycol on the RQ
Ni and RQ Ni–Sn catalysts.
References
[1] S.H. Clarke, A.L. Dicks, K. Pointon, T.A. Smith, A. Swann, Catal. To-
day 38 (1997) 411.
[2] L.K. Rath, J.R. Longanbach, Energy Sources 13 (1991) 443.
[3] J.R. Rostrup-Nielsen, Steam Reforming Catalysts, Danish Technical
Press, Copenhagen, 1975.
[4] J.R. Rostrup-Nielsen, Phys. Chem. Chem. Phys. 3 (2001) 283.
[5] R.D. Cortright, R.R. Davda, J.A. Dumesic, Nature 418 (2002) 964.
[6] J.W. Shabaker, G.W. Huber, R.R. Davda, R.D. Cortright, J.A. Dumesic,
Catal. Lett. 88 (2003) 1.
[7] J.W. Shabaker, R.R. Davda, G.W. Huber, R.D. Cortright, J.A. Dumesic,
J. Catal. 215 (2003) 344.
interpretations on Sn by Dumesic et al. can be directly adopted
to account for the positive role of Sn in the RQ Ni–Sn catalysts.
On the other hand, Onda et al. [16] found that on increasing the
Sn content in the Ni–Sn intermetallic compounds, the density
of state at Fermi level decreased, the width of the Ni 3d peak
narrowed, and the peak top of Ni 3d shifted toward higher BE.
We expect that such an electronic effect may be operative to
a greater or lesser extent on the RQ Ni–Sn catalysts, weaken-
ing the bonding of CO [42] and consequently suppressing the
methanation reaction. The shift of the high-temperature H2 des-
orption peak to lower temperature after Sn addition shown in
Fig. 4 also can be rationalized within this framework. In addi-
tion, it has been reported that on a PtRu or Pt3Sn alloy surface,
the alloying Ru or Sn is much better than Pt at dissociating H2O
and providing OH groups to react with the CO at Pt sites, which
thus promotes the electrochemical oxidation of CO on Pt [45,
46]. We suggest that Sn in the RQ Ni–Sn catalysts may play a
role similar to that of Sn in the Pt3Sn alloy, functioning as an
reactive center for H2O activation. Once the surface OH group
is formed on Sn, it readily oxidizes the CO adsorbed on neigh-
boring Ni sites, thus facilitating the WGS reaction and leading
to higher H2 selectivity. This interpretation is compatible with
the observation that the TOFs of H2 and CO2 increased con-
comitantly with the addition of Sn (see Table 4).
[8] R.R. Davda, J.W. Shabaker, G.W. Huber, R.D. Cortright, J.A. Dumesic,
Appl. Catal. B 43 (2003) 13.
[9] G.W. Huber, J.W. Shabaker, J.A. Dumesic, Science 300 (2003) 2075.
[10] J.W. Shabaker, G.W. Huber, J.A. Dumesic, J. Catal. 222 (2004) 180.
[11] J.W. Shabaker, J.A. Dumesic, Ind. Eng. Chem. Res. 43 (2004) 3105.
[12] J.W. Shabaker, D.A. Simonetti, R.D. Cortright, J.A. Dumesic, J. Catal. 231
(2005) 67.
[13] R.R. Davda, J.W. Shabaker, G.W. Huber, R.D. Cortright, J.A. Dumesic,
Appl. Catal. B 56 (2005) 171.
[14] D.A. Morgenstern, J.P. Fornango, Energy Fuels 19 (2005) 1708.
[15] H.R. Hu, M.H. Qiao, Y. Pei, K.N. Fan, H.X. Li, B.N. Zong, X.X. Zhang,
J. Catal. 221 (2004) 612.
[16] A. Onda, T. Komatsu, T. Yashima, Phys. Chem. Chem. Phys. 2 (2000)
2999.
[17] A. Onda, T. Komatsu, T. Yashima, J. Catal. 201 (2001) 13.
[18] A. Onda, T. Komatsu, T. Yashima, J. Catal. 221 (2003) 378.
[19] P. Gallezot, P.J. Cerino, B. Blanc, G. Flèche, P. Fuertes, J. Catal. 146
(1994) 93.
[20] P. Lesage, O. Clause, P. Moral, B. Didillon, J.P. Candy, J.M. Basset,
J. Catal. 155 (1995) 238.
[21] N.N. Nichio, M.L. Casella, G.F. Santori, E.N. Ponzi, O.A. Ferretti, Catal.
Today 62 (2000) 231.
[22] PDFMaint Version 3.0, Powder Diffraction Database, Bruker Analytical
X-Ray Systems GmbH, 1997.
5. Conclusion
Skeletal Ni catalyst prepared from a rapidly quenched
Ni50Al50 alloy is intrinsically less selective in producing H2
but more selective in producing alkanes than Raney Ni in the
aqueous-phase reforming of ethylene glycol. This effect is at-
tributed to the expanded lattice of RQ Ni in favor of the dis-
sociation of CO and consequently the methanation reaction,
which consumes H2. Inspection of the gas and liquid products
on RQ Ni suggests that the change in the lattice constant influ-
ences the reaction pathways of ethylene glycol with H2O, with
complete suppression of the undesired C–O cleavage pathway.
Modification of RQ Ni with Sn by simply impregnating RQ Ni
in SnCl4 ethanolic solution drastically improves H2 selectiv-
ity. Moreover, H2 selectivity over the RQ Ni80Sn20 catalyst is
not sensitive to system pressure and space velocity. Based on
our characterizations and previous findings, we suggest that Sn
may block the active sites for CO adsorption and/or dissocia-
tion, thus suppressing the undesired methanation reaction. On
the other hand, the water–gas shift reaction is promoted, most
likely due to the activation of H2O by Sn.
[23] H.E. Swanson, E. Fuyat, Natl. Bur. Stand. Circ. 1 (1953) 13.
[24] S.D. Robertson, R.B. Anderson, J. Catal. 23 (1971) 286.
[25] J. Wintterlin, T. Zambelli, J. Trost, J. Greeley, M. Mavrikakis, Angew.
Chem. Int. Ed. 42 (2003) 2850.
[26] J.F. Moulder, W.F. Stickle, P.E. Sobol, K.D. Bomben, in: J. Chastain (Ed.),
Handbook of X-Ray Photoelectron Spectroscopy, Perkin–Elmer, Eden
Prairie, MN, 1992.
[27] P. Fouilloux, Appl. Catal. 8 (1983) 1.
[28] F. Delannay, J.P. Damon, J. Masson, B. Delmon, Appl. Catal. 4 (1982)
169.
[29] C. Padeste, D.L. Trimm, Catal. Lett. 17 (1993) 333.
[30] K. Christmann, O. Schober, G. Ertl, M. Neumann, J. Chem. Phys. 60
(1974) 4528.
[31] M. Arai, K. Suzuki, Y. Nishiyama, Bull. Chem. Soc. Jpn. 66 (1993) 40.
[32] J. Greeley, W.P. Krekelberg, M. Mavrikakis, Angew. Chem. Int. Ed. 43
(2004) 4296.
[33] R.A. Lemons, J. Power Sources 29 (1990) 251.
[34] L.J. Shorthouse, A.J. Roberts, R. Raval, Surf. Sci. 480 (2001) 37.
[35] R.M. Koros, E.J. Nowak, Chem. Eng. Sci. 22 (1967) 470.
[36] H.R. Hu, M.H. Qiao, F.Z. Xie, K.N. Fan, H. Lei, D.L. Tan, X.H. Bao, H.L.
Lin, B.N. Zong, X.X. Zhang, J. Phys. Chem. B 109 (2005) 5186.
[37] D.W. Goodmann, R.D. Kelley, T.E. Madey, J.M. White, J. Catal. 64 (1980)
479.

220
F.Z. Xie et al. / Journal of Catalysis 241 (2006) 211–220
[38] A.T. Bell, Catal. Rev. Sci. Eng. 23 (1981) 203.
[42] M. Mavrikakis, B. Hammer, J.K. Nørskov, Phys. Rev. Lett. 81 (1998)
2819.
[39] J.W.E. Coenen, P.F.M.T. van Nisselroy, M.H.J.M. de Croon, P.F.H.A. van
Dooren, R.Z.C. van Meerten, Appl. Catal. 25 (1986) 1.
[43] M.A. Vannice, J. Catal. 50 (1977) 228.
[44] R.I. Masel, Principles of Adsorption and Reaction on Solid Surfaces, In-
terscience, New York, 1996.
[45] P. Liu, A. Logadottir, J.K. Nørskov, Electrochim. Acta 48 (2003) 3731.
[46] M.T.M. Koper, Surf. Sci. 548 (2004) 1.
[40] H.S. Bengaard, J.K. Nørskov, J. Sehested, B.S. Clausen, L.P. Nielsen,
A.M. Molenbrock, J.R. Rostrup-Nielsen, J. Catal. 209 (2002) 365.
[41] J. Sehested, S. Dahl, J. Jacobsen, J.R. Rostrup-Nielsen, J. Phys. Chem.
B 109 (2005) 2432.