Highly efficient Pd-ZnO catalyst doubly promoted by CNTs and Sc 2O3 for methanol steam reforming

Article abstract of DOI:10.1016/j.apcata.2013.01.039

A type of Pd-ZnO catalyst doubly promoted by CNTs and Sc₂O ₃ for methanol steam reforming (MSR) was developed, and displayed excellent activity and operation stability for the selective formation of H ₂ and CO₂. Over a Pd_0.15Zn₁Sc _0.067-10%CNTs catalyst under the reaction conditions of 0.5 MPa and 548 K, the STY(H₂) can maintain stable at the level of 1.56 mol h^-1 g^-1 at 75 h after the reaction started, which was 1.7 times that of the corresponding (CNTs and Sc)-free counterpart Pd _0.15Zn₁. Characterization of the catalyst revealed that the highly conductive CNTs could promote hydrogen spillover from the PdZn/ZnO-sites to the CNTs adsorption-sites, and then combine to form H ₂(a), followed by desorbing to H₂(g), which would help increase the rate of a series of surface dehydrogenation reactions in the MSR process. The pronounced modification action of Sc³⁺ may be due to the high solubility of Sc₂O₃ in ZnO lattice. Solution of a small amount of Sc₂O₃ in ZnO lattice resulted in the formation of Schottky defects in the form of cationic vacancies at the surface of ZnO, where the (PdZn)⁰-Pd²⁺ clusters can be better stabilized through the Pd²⁺ accommodated at the surface vacant cation-sites. This would be conducive to inhibiting the sintering of the catalytically active (PdZn)⁰ nanoparticles, and thus, markedly prolonging the life of the catalyst.

Full text of DOI:10.1016/j.apcata.2013.01.039

Applied Catalysis A: General 455 (2013) 137–144
Contents lists available at SciVerse ScienceDirect
Applied Catalysis A: General
journal homepage: www.elsevier.com/locate/apcata
Highly efficient Pd–ZnO catalyst doubly promoted by CNTs and Sc2O3
for methanol steam reforming
Lu Yang, Guo-Dong Lin, Hong-Bin Zhang∗,1
Department of Chemistry, College of Chemistry and Chemical Engineering, State Key Laboratory of Physical Chemistry for Solid Surfaces and National
Engineering Laboratory for Green Chemical Productions of Alcohols-Ethers-Esters, Xiamen University, Xiamen 361005, China
a r t i c l e i n f o
a b s t r a c t
Article history:
A type of Pd–ZnO catalyst doubly promoted by CNTs and Sc2O3 for methanol steam reforming (MSR)
was developed, and displayed excellent activity and operation stability for the selective formation of H2
and CO2. Over a Pd0.15Zn1Sc0.067–10%CNTs catalyst under the reaction conditions of 0.5 MPa and 548 K,
the STY(H2) can maintain stable at the level of 1.56 mol h−1 g−1 at 75 h after the reaction started, which
was 1.7 times that of the corresponding (CNTs and Sc)-free counterpart Pd0.15Zn1. Characterization of the
catalyst revealed that the highly conductive CNTs could promote hydrogen spillover from the PdZn/ZnO-
sites to the CNTs adsorption-sites, and then combine to form H2(a), followed by desorbing to H2(g), which
would help increase the rate of a series of surface dehydrogenation reactions in the MSR process. The
pronounced modification action of Sc3+ may be due to the high solubility of Sc2O3 in ZnO lattice. Solution
of a small amount of Sc2O3 in ZnO lattice resulted in the formation of Schottky defects in the form of
cationic vacancies at the surface of ZnO, where the (PdZn)0–Pd2+ clusters can be better stabilized through
the Pd2+ accommodated at the surface vacant cation-sites. This would be conducive to inhibiting the
sintering of the catalytically active (PdZn)0 nanoparticles, and thus, markedly prolonging the life of the
catalyst.
Received 15 December 2012
Received in revised form 28 January 2013
Accepted 31 January 2013
Available online xxx
Keywords:
MWCNTs
Pd–ZnO catalyst
Sc2O3
Methanol steam reforming
© 2013 Elsevier B.V. All rights reserved.
1. Introduction
of the Pd/ZnO catalyst was also reported by Chin et al. [9]. Nev-
ertheless, from the practical viewpoint, the activity, especially
There has been a lot of attention recently on the develop-
ment of suitable catalysts for steam reforming of methanol (MSR;
CH3OH + H2O → CO2 + 3H2) [1–10]. Most early studies focused on
Cu-based catalysts, which exhibit high activity and selectivity to
the formation of CO2 and H2 [1–3]. However, they suffer from poor
heat and oxidation resistances and fast deactivation at reaction
temperatures above 543 K.
On the other hand, novel Pd/ZnO systems [5–9] (as well as
Pd/Ga2O3 [10]) have attracted increasing attention because of their
high selectivity to the formation of CO2 and H2 and excellent ther-
mal stability. Iwasa et al. [5] were the first to report that, when
supported on ZnO and reduced at temperatures higher than 573 K,
Pd displayed high activity and selectivity to form CO2 and H2.
Extremely low CO selectivity was attributed to the formation of
bimetallic PdZn alloy dispersed on the ZnO matrix, which modi-
fied the catalytic function of Pd [6–8]. The high MSR performance
operation stability, of the Pd/ZnO or Pd/Ga2O3 catalysts can still
be improved.
Multi-walled carbon nanotubes (simplified as CNTs in later text),
as a novel nano-carbon material, have been drawing a lot of atten-
tion for catalysis applications in recent years [11–13]. This kind
of nanotube-C possesses several unique features, such as highly
graphitized tube-wall, nano-sized channel and sp2-C-constructed
surface [14,15]. They also display high thermal/electrical con-
ductivity, medium to high specific surface areas, and excellent
performance for adsorbing hydrogen and promoting hydrogen
spillover [15,16], all of which render them full of promise to be
a novel catalyst carrier and/or promoter. The catalytic studies con-
ducted so far on CNT-based systems have shown promising results
in terms of activity and selectivity [11–13].
In the present work, a type of highly efficient Pd–ZnO catalyst
doubly promoted by CNTs and Sc2O3 was developed. The catalyst
displayed higher activity and operation stability for MSR, compared
to the corresponding (CNTs and Sc)-free counterpart. The catalyst
was characterized by means of TEM/EDX, XRD, XPS, H2-TPR and H2-
TPD, and the nature of promoter action by the CNTs and Sc2O3 was
discussed. The results shed light on the design of practical catalysts
for the MSR.
∗
Corresponding author. Tel.: +86 592 2184591; fax: +86 592 2086116.
E-mail addresses: zenobiapost@yahoo.com.cn (L. Yang), gdlin@xmu.edu.cn
(G.-D. Lin), hbzhang@xmu.edu.cn (H.-B. Zhang).
1
Tel.: +86 592 2181888; fax: +86 592 2081888.
0926-860X/$ – see front matter © 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.apcata.2013.01.039

138
L. Yang et al. / Applied Catalysis A: General 455 (2013) 137–144
2. Experimental
2.1. Catalyst preparation
STY(H2)) could be calculated through the yields of CO2 and CO. All
data were taken 24 h after the reaction started (unless otherwise
specified).
The CNTs used in the present work were prepared following
the method reported previously [14]. The freshly prepared CNTs
were purified with treatment of nitric acid (8 mol/L at 363 K) for
8 h, followed by rinsing with de-ionized water twice, and then
drying at 383 K under N2-atmosphere. Open-ended CNTs with
somewhat hydrophilic surface were then obtained. For the puri-
fied CNTs, contents of the total carbon and the graphitized carbon
were ≥99.5% and ≥90% (mass%), respectively.
2.3. Catalyst characterization
Transmission electron microscopy (TEM) and energy-dispersive
X-ray spectroscopy (EDX) measurements were performed on
JEM-1400 and Hitachi S-4800 electron microscopes, respectively.
XRD measurements were carried out on an X’Pert PRO X-
ray diffractometer (PANalytical) with Cu K␣ (ꢀ␣1 = 0.15406 nm,
ꢀ␣2 = 0.15443 nm) radiation. A continuous scan mode was used
to collect 2ꢁ data from 10◦ to 90◦. The voltage and current were
40 kV and 30 mA, respectively. X-ray photoelectron spectroscopy
(XPS) measurements were done on a Quantum 2000 Scanning
ESCA Microprobe instrument with Al K˛ radiation (15 kV, 25 W,
hꢂ = 1486.6 eV) under ultrahigh vacuum (5 × 10−7 Pa), calibrated
internally by the carbon deposit C(1s) (Eb = 284.7 eV).
Specific surface area (SSA) was determined by N2 adsorption
using a Micromeritics ASAP 2020 system. Measurement of CO
chemisorption on the catalysts was performed by a Micromeritics
ASAP-2010 Micropore Analyzer. 0.1–0.2 g of catalyst sample was
used for each test. The sample was put into a quartz tube, followed
by evacuating for 10 min at 393 K, then switching to a purified H2
stream (30 mL min−1) as reducing gas to conduct an in situ H2-TPR
treatment of the catalyst sample, subsequently evacuating for 1 h
at the reduction temperature and another 1 h after cooling down
to room temperature, and then switching to gaseous CO (of 99.99%
purity) to conduct the CO chemisorption measurement. From the
determined amount of chemisorbed CO, the dispersion and surface
area of metallic palladium were calculated [17].
A series of Pd–ZnO or Pd–ZnO–Sc2O3 catalysts doped with CNTs,
denoted as PdiZnj(or PdiZnjSck)–x%(mass percentage)CNTs, were
prepared by a co-precipitation method. An aqueous solution con-
taining calculated amounts of PdCl2 and Zn(NO3)2·6H2O (or as well
as Sc(NO3)3·6H2O) was added dropwise under vigorous stirring
into a Pyrex flask containing a certain amount of aqueous Na2CO3
solution at 333 K. The addition was adjusted to maintain the pH of
the suspension at 9–10. The suspension was continuously stirred
for 30 min at 333 K, followed by cooling down to room tempera-
ture before filtering. The filtered cake was repeatedly washed with
deionized water until the filtrate became neutral in pH. The washed
solid was added into a suspension prepared in advance contain-
ing calculated amounts of CNTs, followed by stirring vigorously for
4 h, and then centrifuging-filtering. The obtained solid was dried at
383 K for 12 h and calcined at 633 K for 2 h, yielding the precursor of
PdiZnj–x%CNTs or PdiZnjSck–x%CNTs catalysts (in oxidation state).
The CNT-free counterparts were prepared in the similar man-
ner, and used as reference. All samples of catalyst precursor were
pressed, crushed, and sieved to a size of 20–40 mesh for the activity
evaluation.
Tests of H2-temperature-programmed reduction (H2-TPR) of
oxidation precursor of the catalysts were conducted on a fixed-bed
continuous-flow micro-reactor. A NaOH-column and a 3A-zeolite
column were installed in sequence at the reactor-exit to remove
water vapor formed by the reduction of metal oxides of the cat-
alyst sample. Fifty mg of catalyst sample was used for each test.
The sample was first flushed by an Ar (of 99.999% purity) stream
(60 mL min−1) at 393 K for 60 min to clean its surface, and then
cooled down to room temperature, followed by switching to a
N2-carried 5 vol% H2 gaseous mixture as reducing gas to start the
2.2. Catalyst evaluation
Activity tests of the catalyst for MSR were carried out in a
fixed-bed continuous-flow reactor-GC combination system. Cat-
alyst (0.200 g) was mixed with 4.0 g quartz sand (inert diluents,
20–40 mesh) to order to maintain isothermal conditions, and
placed in the reactor. Prior to the reaction, the catalyst was pre-
reduced in situ under purified H2 stream (0.1 MPa and 1800 mL h−1).
The reduction temperature was programmed to rise from room
temperature to 523 K and maintain at that temperature for 12 h,
before being brought to the desired temperature for the cata-
lyst test. The MSR reaction was conducted at a stationary state
under the reaction conditions of 0.5 MPa and 473–673 K. A pre-
mixed feed-liquid of methanol and water (molar ratio of 1:1) was
introduced into the reactor by using a syringe pump (Series II
Pump, 10 mL Heads). Prior to entering the reactor, the feed-liquid
was fully vaporized through a vaporizer, operating at 473 K. N2 (of
99.99% purity) was used as the dilution gas to maintain the mol%
of the pre-mixed CH3OH–H2O (molar ratio of 1:1) in the feed-gas
((CH3OH–H2O) + N2) at the desired level (controlled by Model D08-
1F flow control valve). The corresponding gas hourly space velocity
(GHSV) of the feed-gas ranged in 54,000–216,000 mL h−1 g−1. A
glass condenser at 278 K was used to separate liquid products from
gaseous products. The gaseous products were analyzed by an on-
line GC (Model GC-2014C, Shimadzu) equipped with a TCD detector
and a column filled with carbon molecular sieve (TDX-01, 2.0-m
length), which was used for the analysis of N2 (as internal standard),
CO and CO2. The experimental results showed that CO2 and CO
were the only two carbon-containing products of the MSR reaction,
without other possible carbon-containing products detected. Thus,
selectivity to CO2 and CO (symbolized as S(CO2) and S(CO)) could
be determined by an internal normalization method, and methanol
conversion (noted as X(CH3OH)) and H2 space-time-yield (noted as
TPR observation. The rate of temperature increase was 5 K min−1
.
Change of hydrogen-signal was monitored by an on-line GC (Shi-
madzu GC-8A) with a TC detector.
Tests of H2-temperature-programmed desorption (H2-TPD) of
the catalysts were conducted on an adsorption/desorption system.
Two hundred milligram of the catalyst precursor was used in each
test. Prior to the H2-TPD test, the sample of catalyst-precursor was
in situ pre-reduced in a H2 (of 99.999% purity) stream (900 mL h−1
)
at 523 K for 2 h and then flushed by an Ar (of 99.999% purity) stream
(1800 mL h−1) at 433 K for 30 min to clean its surface, followed
by switching to the H2 (of 99.999% purity) stream for hydrogen
adsorption for 30 min and subsequently at room temperature for
4 h. Afterwards, the sample was flushed by the Ar stream at room
temperature until the stable baseline in GC appeared. TPD measure-
ments were then conducted from 298 K to 1073 K. The rate of tem-
perature increase was 5 K min−1. Change of hydrogen-signal was
monitored by an on-line GC (Shimadzu GC-8A) with a TC detector.
3. Results and discussion
3.1. Optimization of the catalyst composition
The reactivity of MSR over a series of Pd0.15Zn1–x%CNTs
catalysts
with
varied
amounts
of
CNTs
was
first

L. Yang et al. / Applied Catalysis A: General 455 (2013) 137–144
139
Fig. 2. Reactivity of MSR over the Pd0.15Zn1–10%CNTs catalyst with vary-
ing temperatures; the reaction conditions: 0.5 MPa, the feed-gas composition
CH3OH/H2O/N2 = 7.5/7.5/85 (molar ratio), GHSV = 108,000 mL h−1 g−1
.
Fig. 1. XRD patterns of the Pd0.15Zn1–10%CNTs catalyst reduced at different tem-
peratures, as indicated in the figure.
indicating the formation of PdZn alloy-phase as the predominant
Pd-containing crystallite-phase. Using the well-known Scherrer’s
equation, the average particle size of PdZn-alloy crystallites of the
four samples in Fig. 1 was estimated to be: (b) 7.5 nm, (c) 13.7 nm,
(d) 16.3 nm, and (e) 19.7 nm, respectively. The upward trend of
the particle size of PdZn-alloy crystallites with rising reduction-
temperature implied that the sequence of the exposed area of
PdZn-alloy in those samples was: (b) > (c) > (d) > (e), in line with
the aforementioned observed sequence of catalytic activity of those
catalysts for MSR.
As observed by Datye et al. [8] on the Pd–ZnO catalyst reduced at
temperatures of <523 K, PdZn alloy and monometallic Pdx0 crystal-
lites coexisted in the Pd0.15Zn1–10%CNTs catalyst reduced at 473 K.
The average particle size of the former was estimated to be 4.5 nm,
while the latter was unable to display the characteristic peak in
the corresponding XRD pattern (Fig. 1(a)) probably due to their
particle sizes of ≤3 nm. Lower-temperature reduction led to incom-
plete transformation of Pd into the PdZn alloy, therefore resulting
in lower activity of the corresponding catalyst for MSR.
investigated. The results show that CO2 and CO were the only
two carbon-containing products observed. Among the five tested
catalysts of Pd0.15Zn1–x%CNTs (x% = 5%, 7.5%, 10%, 12.5%, 15%),
Pd0.15Zn1–10%CNTs showed the highest activity for MSR. Over
this catalyst under the reaction conditions of 0.5 MPa, 523 K,
feed-gas composition of CH3OH/H2O/N2 = 7.5/7.5/85 (molar ratio)
and GHSV = 108,000 mL h−1 g−1, X(CH3OH) reached 67.0%, with the
corresponding STY(H2) being 0.72 mol h−1 g−1. X(CH3OH) of the
other four catalysts with x% = 5%, 7.5%, 12.5%, 15% was 61.6%, 65.2%,
64.3%, 59%, with the corresponding STY(H2) being 0.66, 0.70, 0.69,
0.63 mol h−1 g−1, successively.
Next, the Pd/Zn molar ratio (i/j) of the PdiZnj-10%CNTs cat-
alysts was optimized. The results showed that the catalyst
with i/j = 0.15 displayed the highest activity for MSR. Over the
Pd0.15Zn1–10%CNTs catalyst under the aforementioned reaction
conditions, X(CH3OH) reached 67.0%, with the corresponding
STY(H2) being 0.72 mol h−1 g−1. X(CH3OH) of the other four cat-
alysts with i/j = 0.05, 0.10, 0.2, 0.25 was 43.9%, 57.1%, 66.3%,
58.4%, with the corresponding STY(H2) being 0.47, 0.61, 0.71,
0.63 mol h−1 g−1, successively.
The above XRD-analysis results, in connection with above-
mentioned those of catalyst assay, demonstrated that the
catalytically active phase associated more closely with the MSR was
PdZn alloy, rather than metallic Pd, especially under reaction tem-
peratures of ≥523 K, in line with the results reported in the previous
literatures [6,8].
3.2. Optimization of the reaction operation conditions
3.2.1. Reduction temperature
The reduction temperature for the precursor of the
Pd0.15Zn1–10%CNTs catalyst has a marked effect on the perfor-
mance of catalyzing MSR. The test results showed that the catalyst
reduced by H2 at 523 K displayed the highest activity for MSR, with
X(CH3OH) and STY(H2) reaching 67% and 0.72 mol h−1 g−1 under
the aforementioned reaction conditions; while the X(CH3OH)
of the other four catalysts reduced by H2 at 473 K, 573 K, 673 K
and 773 K, respectively, was 61%, 66%, 58% and 38%, with the
3.2.2. Reaction temperature
Reaction temperature has a pronounced effect on the conver-
sion of MSR and the selectivity of CO2 formation. Fig. 2 displayed the
reactivity of MSR over the catalyst of Pd0.15Zn1–10%CNTs at varying
temperatures. With the reaction temperature rising progressively
from 473 K, X(CH3OH) ascended quickly before reaching a maxi-
mum at 598 K, and then slowly descended; while SCO descended
monotonously. In order to achieve both high S(CO2) and2 X(CH3OH),
548 K was taken as the optimal operating temperature.
corresponding STY(H2) being 0.65, 0.71, 0.62 and 0.41 mol h−1 g−1
,
successively, under the same reaction conditions. Therefore, 523 K
is taken as the optimal reduction temperature for this catalyst in
order to achieve both high X(CH3OH) and STY(H2).
The results of XRD measurement (Fig. 1) provided direct evi-
dence of the formation of PdZn alloy upon the reduction of the
CNT-doped Pd–ZnO catalyst. In the XRD patterns taken on the
four samples of Pd0.15Zn1–10%CNTs reduced at 523, 573, 673 and
773 K, respectively, the peak at 2ꢁ = 40.1◦ (originating from the
(1 1 1) diffraction of metallic Pdx0) completely vanished, while the
two peaks at 2ꢁ = 41.2◦ and 44.1◦ (due to the (1 1 1) and (2 0 0)
diffractions of PdZn alloy, respectively [18]) markedly enhanced,
3.2.3. Feed-gas GHSV
In order to find out the working conditions with great extent of
reaction, the reactivity of MSR over the Pd0.15Zn1–10%CNTs catalyst
was tested under a series of reaction conditions with varying GHSV
of the feed-gas. The results in Fig. 3 showed that, with GHSV of the
feed-gas rising progressively from 54,000 mL h−1 g−1, X(CH3OH)
decreased monotonously and S(CO2) maintained continuously at
a high level of ∼98%. Meanwhile, STY(H2) increased in the lower

140
L. Yang et al. / Applied Catalysis A: General 455 (2013) 137–144
Fig. 3. Reactivity of MSR over the Pd0.15Zn1–10%CNTs catalyst with varying
GHSV of the feed-gas; the reaction conditions: 0.5 MPa, 548 K, the feed-gas
composition CH3OH/H2O/N2 = 7.5/7.5/85 (molar ratio) and GHSV ranging in
Fig.
5. Operating
stability
of
MSR
over
the
catalysts
of:
(a)
Pd0.15Zn1Sc0.067–10%CNTs; (b) Pd0.15Zn1–10%CNTs; (c) Pd0.15Zn1; at reaction
conditions of 0.5 MPa, 548 K, CH3OH/H2O/N2 = 30/30/40 (molar ratio) and
56,000–216,000 mL h−1 g−1
.
GHSV = 180,000 mL h−1 g−1
.
GHSV region, and approached a plateau when the GHSV reached
180,000 mL h−1 g−1 and above. In order to simultaneously achieve
relatively high X(CH3OH) and S(CO2), as well as STY(H2), GHSV of
3.3. Performance of CNT-promoted Pd0.15Zn1-based catalysts and
reference system for MSR
the feed-gas was set at 180,000 mL h−1 g−1
.
In order to evaluate the performance of the CNT-promoted
Pd–ZnO based catalysts under the working conditions with great
extent of reaction, the MSR reaction was conducted under the
aforementioned optimized reaction conditions: 0.5 MPa, 548 K,
feed-gas composition CH3OH/H2O/N2 = 30/30/40 (molar ratio) and
GHSV = 180,000 mL h−1 g−1. Table 1 lists the results taken at 75 h
after the reaction started. Over the Pd0.15Zn1–10%CNTs catalyst,
X(CH3OH) reached 19.3%, with the corresponding STY(H2) being
1.50 mol h−1 g−1. This STY(H2) value was 1.67 times that of the cor-
responding CNT-free counterpart, Pd0.15Zn1.
Fig. 5(b) and (c) shows the operation stability of the
Pd0.15Zn1–10%CNTs catalyst and the corresponding CNT-free coun-
terpart for the MSR lasting 150 h. Over the two catalysts, throughout
the reaction, S(CO2) maintained stable at the level of ∼95%,
while X(CH3OH) and STY(H2) descended gradually, and came
down to 10.8% and 0.96 mol h−1 g−1 for the former, and 8.1% and
0.56 mol h−1 g−1 for the latter, after 150 h of reaction.
3.2.4. Feed-gas composition
The composition of feed-gas has a marked effect on the extent of
MSR reaction. The MSR reactivity over the Pd0.15Zn1–10%CNTs cata-
lyst with varying feed-gas compositions was evaluated. The results
in Fig. 4 showed that, with mol% of the pre-mixed CH3OH–H2O
(molar ratio of 1:1) in the feed-gas ((CH3OH–H2O) + N2) rising pro-
gressively from 15% to 75%, X(CH3OH) decreased monotonously,
and S(CO2) maintained continuously at a high level of 98–96%.
Meanwhile, STY(H2) slowly descended after reaching a maximum at
the mol% = 60%. In order to achieve both high STY(H2) and low con-
sumption of raw-material methanol, (CH3OH–H2O):(N2) = 60:40
(mol/mol) was taken as the optimal composition of feed-
gas.
We observe that doping an appropriate amount of Sc2O3 into the
Pd–ZnO based catalysts can significantly improve their operation
stability for the MSR. The activity of three catalysts of Pd0.15Zn1Sck-
10%CNTs with k = 0.05, 0.067, 0.1 for the MSR was evaluated. The
results showed that the catalyst with k = 0.067 displayed the opti-
mal activity and operating stability. Over this catalyst under the
reaction condition of 0.5 MPa, 548 K, CH3OH/H2O/N2 = 30/30/40
(molar ratio) and GHSV = 180,000 mL h−1 g−1, X(CH3OH) reached
21.4% at 75 h after the reaction started, with the corresponding
STY(H2) being 1.56 mol h−1 g−1 (see Table 1). Meanwhile over the
other two catalysts with k = 0.05, 0.1, X(CH3OH) was 18.9% and
12.6%, respectively, with the corresponding STY(H2) being 1.31 and
0.91 mol h−1 g−1, under the same reaction conditions.
Fig. 5(a) showed the operating stability of the
Pd0.15Zn1Sc0.067–10%CNTs catalyst for MSR lasting 150 h. After
around 24 h of “running in” stage of the reaction, the catalyst
reached a stable operating state, and after 150 h of reaction,
STY(H2) maintained stable at the level of 1.29 mol h−1 g−1, with no
sign of obvious deactivation observed.
The apparent activation energy (Ea) of the MSR reaction
over the three catalysts tested was measured under the
reaction conditions with mass transfer limitation ruled out.
Fig. 4. Reactivity of MSR over the Pd0.15Zn1–10%CNTs catalyst with varying mol% of
the premixed CH3OH–H2O (molar ratio of 1:1) in the feed-gas ((CH3OH–H2O) + N2);
at reaction conditions of 0.5 MPa, 548 K, GHSV = 180,000 mL h−1 g−1 and the mol% of
the pre-mixed CH3OH–H2O in the feed-gas ranging in 15–75%.

L. Yang et al. / Applied Catalysis A: General 455 (2013) 137–144
141
Table 1
Reactivity of MSR over the CNTs and Sc2O3 doubly promoted Pd–ZnO catalyst and the reference systems.a
Catalyst
Specific surface
area (m2/g)
Mean pore
diameter (nm)
PdZn-alloy SA
(m2/g)
X(CH3OH) (%)
Selectivity of C-containing
products (%)
STY(H2)
(mol h−1 g−1
)
S(CO2)
S(CO)
Pd0.15Zn1
Pd0.15Zn1–10%CNTs
Pd0.15Zn1Sc0.067–10%CNTs
82.3
75.0
100.6
11.2
11.5
10.9
27.0
26.8
31.5
12.7
19.3
21.4
94.4
95.6
90.0
5.6
4.4
10.0
0.90
1.50
1.56
a
The reaction conditions: 0.5 MPa, 548 K, the feeding-gas composition CH3OH/H2O/N2 = 30/30/40 (molar ratio), GHSV = 180,000 mL h−1 g−1; the data were taken 75 h after
the reaction started.
The results showed that the Ea of MSR reaction observed
type of CNTs was a “Herringbone-type” of CNTs, with the outer
diameters of 10–50 nm, inner diameters of 3–8 nm, and N2-BET
surface area of ∼125 m2/g. Fig. 6(b)–(d) showed the TEM images
of the three used catalysts of Pd0.15Zn1, Pd0.15Zn1–10%CNTs and
Pd0.15Zn1Sc0.067–10%CNTs, respectively, from which the particle
size of the PdZn-alloy crystallites was estimated to be 12–16 nm,
10–16 nm and 4–7 nm, respectively. The results of the correspond-
ing EDX measurement showed that the element compositions of
the surface of the three used catalysts were: Pd/Zn/O = 6.4/38.7/54.9
(mol%) for the Pd0.15Zn1, Pd/Zn/O/C = 4.0/28.9/39.2/27.9 (mol%) for
the Pd0.15Zn1–10%CNTs, and Pd/Zn/O/C/Sc = 3.7/24.6/44.4/26.7/0.6
(mol%) for the Pd0.15Zn1Sc0.067–10%CNTs. This result indicated that
the doping a minor amount of Sc2O3 into the Pd0.15Zn1–10%CNTs
catalyst led to an increase by an appreciable amount in
mol% of oxygen at the quasi-functioning surface of the used
Pd0.15Zn1Sc0.067–10%CNTs catalyst, implying an obvious decrease
of mol% of the Pd reduced to Pd0 in the total surface Pd amounts.
on the three catalysts, Pd0.15Zn1, Pd0.15Zn1–10%CNTs and
Pd0.15Zn1Sc0.067–10%CNTs, was 87.8, 86.4 and 86.2 kJ mol−1
,
respectively, under the reaction conditions of 0.5 MPa, 473–573 K,
feed-gas composition CH3OH/H2O/N2 = 30/30/40 (molar ratio) and
GHSV = 432,000 mL h−1 g−1. These Ea values were fairly close to
each other, indicating that appropriate incorporation of a minor
amount of either simple CNTs or CNTs and Sc2O3 into the Pd0.15Zn1
host catalyst did not cause a marked change in the Ea for the MSR
reaction, implying that such doping did not alter the major reaction
pathway of steam reforming of CH3OH to yield H2 and CO2.
3.4. Characterization of the catalysts
3.4.1. TEM and EDX characterization
Fig. 6(a) showed the TEM image of the CNTs used in the
present work. Previous studies [14,15] have demonstrated that this
Fig. 6. TEM images of: (a) CNTs; (b) the used Pd0.15Zn1; (c) the used Pd0.15Zn1–10%CNTs; (d) the used Pd0.15Zn1Sc0.067–10%CNTs.

142
L. Yang et al. / Applied Catalysis A: General 455 (2013) 137–144
Fig. 7. TPR profiles of the catalyst precursors in oxidation state by H2: (a) Pd0.15Zn1
(×90%); (b) Pd0.15Zn1–10%CNTs; (c) Pd0.15Zn1Sc0.067–10%CNTs.
Fig. 8. XRD patterns of the used catalysts: (a) Pd0.15Zn1; (b) Pd0.15Zn1–10%CNTs; (c)
Pd0.15Zn1Sc0.067–10%CNTs.
3.4.2. H2-TPR test of catalysts in oxidation state
H2-TPR of the catalyst provided useful information about the
reducibility of catalyst. Fig. 7 showed the H2-TPR profiles taken
on the three catalysts in oxidation-state. The H2-reduction of
the Pd0.15Zn1 catalyst (Fig. 7(a)) started at ∼323 K and the main
TPR-peaks appeared at 509 K and 628 K. The lower-temperature
reduction peak (spanning from 323 K to 509 K) may be ascribed
to the continuous multi-step single-electron reduction of a large
portion of Pdn+ species, while the higher-temperature peak (628 K)
likely included the contribution from the reduction of the Zn2+
XRD-detection limit. Using the Scherrer’s equation, the PdZn-alloy
particle size for the three used catalysts was estimated. The cal-
culated average particle size of the PdZn-alloy crystallites was
14.1 nm, 12.5 nm and 4.5 nm, for Pd0.15Zn1, Pd0.15Zn1–10%CNTs and
Pd0.15Zn1Sc0.067–10%CNTs, respectively, very close to those esti-
mated from the corresponding TEM images (Fig. 6(b)–(d)). This
result showed an upward trend of PdZn-alloy dispersity with the
addition of small amounts of Sc3+
.
XPS-analysis of the used catalysts revealed that certain differ-
ence existed among the three catalysts in the valence-states of
the surface Pd species and their relative concentration. Fig. 9 dis-
played the Pd(3d)-XPS spectra taken on the three used catalysts.
The observed Pd(3d5/2) and (3d3/2) XPS peaks appeared at ∼335.6
and ∼340.9 eV. These values are characteristic of Pd-species with
mixed valences. According to Ref. [19] and through computer-
fitting, it could be found that each of those Pd(3d)-XPS spectra
involved the contribution from two or three types of surface Pd-
species: Pd0(PdZn-alloy) and Pd2+(PdO), as well as Pd4+(PdO2), with
0
species near the Pdx to form PdZn alloy, in addition to that of the
subsequent reduction of the remaining Pdn+ species, which were
difficult to be reduced at the lower temperatures. Interestingly, the
doping of CNTs resulted in the appearance of a new peak at ∼373 K,
in addition to the 509-K peak downshifting to 504 K or 488 K (see
Fig. 7(b) and (c) vs. (a)).
The specific H2-consumed amounts (i.e., the amount of hydro-
gen consumed due to reduction of unit mass of Pd) of the three
catalysts in the region of 323–673 K were estimated, and the
obtained relative ratio was: S(a)/S(b)/S(c) = 90.6/93.8/100. The higher
specific H2-consumed amount indicated the higher percentage
of the Pdn+ species reducible to lower valence in the total Pd
amount. Based upon the difference in the specific H2-consumed
amounts and the position (i.e. reduction temperature) of the cor-
responding main H2-reduction peaks, it could be inferred that
the sequence of reducibility of the three catalysts was as fol-
lows: Pd0.15Zn1 < Pd0.15Zn1–10%CNTs < Pd0.15Zn1Sc0.067–10%CNTs.
Conceivably, the lower reduction-temperature would be benefi-
cial to inhibiting the aggregation of PdZn-alloy crystallites formed
by H2-reduction, thus favorable to increasing the Pd0(PdZn-alloy)
exposed area (see Table 1), and subsequently, improving the cat-
alytic activity for MSR.
the B.E. value of Pd0(3d5/2, 3/2), Pd2+(3d5/2, 3/2) and Pd4+(3d5/2, 3/2
)
3.4.3. Post-reaction analysis of the used catalysts by XRD and XPS
The XRD post-reaction analysis (Fig. 8) showed that there were
few differences among the three used catalysts in their XRD fea-
tures related with the Pd and Zn components, except the feature
at 2ꢁ = 26.1◦ for the CNT-containing systems, which was due to the
diffraction of (0 0 2) plane of graphite-like tube-wall of the CNTs
[15]. In these used catalysts, the Pd and Zn components existed
mainly in the forms of ZnO and PdZn alloy (with the observed XRD
features at 2ꢁ = 31.8◦/34.5◦/36.3◦/47.6◦ and 41.2◦/44.1◦, respec-
tively) [18]. The XRD feature at 2ꢁ = 40.1◦ (due to the diffraction
0
of (1 1 1) plane of metallic Pdx crystallites [18]) was not detected,
Fig. 9. XPS spectra of Pd(3d) of the used catalysts: (a) Pd0.15Zn1; (b)
Pd0.15Zn1–10%CNTs; (c) Pd0.15Zn1Sc0.067–10%CNTs.
suggesting that the content of metallic Pdx0-phase was below the

L. Yang et al. / Applied Catalysis A: General 455 (2013) 137–144
143
being 335.5/340.8 eV, 336.3/341.6 eV and 337.9/343.2 eV, respec-
tively (see Fig. 9). At the quasi-functioning surface of the used
Pd0.15Zn1–10%CNTs catalyst (Fig. 9(b)), the mol% of Pd0-species in
the total surface Pd-amount reached 79.9%. This value was higher
than that (71.8 mol%) of the CNTs-free host catalyst, Pd0.15Zn1
(Fig. 9(a)), suggesting that the addition of an appropriate amount
of CNTs to the host catalyst improve the reducibility of the cata-
lyst, resulting in the increase of the mol% of Pd0-species in the total
surface Pd-amount.
Iwasa et al. [5] were the first to report that the catalytic perfor-
mance of Pd/ZnO in MSR is greatly improved by previously reducing
the catalyst at higher temperatures. Afterwards, Datye et al. [8]
found that PdZn alloy particles could be formed after reaction at
523 K, due to the facile reduction of ZnO in the presence of Pd and
H2. Currently, it is well acknowledged that the original catalytic
functions of metallic Pd are modified by the formation of PdZn
alloys upon the reduction of the Pd/ZnO [6–8]. The present XRD
and XPS analysis results, combined with the results of the afore-
mentioned catalyst tests (Table 1), once again confirmed that there
existed close correlation between the surface Pd0-species in form
of PdZn alloys and the excellent selectivity for MSR. The high con-
centration of the surface Pd0-species in the form of PdZn alloys was
conducive to the selective formation of H2 and CO2 in MSR.
Unlike the case of adding CNTs, the doping of small amounts
of Sc3+ into the catalyst system significantly enhanced the mol%
Fig. 10. TPD profiles of H2 adsorption on the pre-reduced catalysts: (a) Pd0.15Zn1;
(b) Pd0.15Zn1–10%CNTs; (c) Pd0.15Zn1Sc0.067–10%CNTs.
H-adspecies at the surface of functioning catalysts was those cor-
responding to the 473–823 K region (region-II), suggesting that it
was those H-adsorption-sites that were closely associated with the
reactivity for MSR.
of Pd2+(as well as Pd4+)-species (with the mol% of Pd2+- and Pd4+
-
species ascending to 42.7% and 14.4%, respectively, correspondingly
with the mol% of Pd0-species descending to 42.9%) at the catalyst
surface (see Fig. 9(c)). The pronounced modification action of Sc3+
may be due to high solubility of Sc2O3 in ZnO lattice, as the ionic
radius of Sc3+ (0.0745 nm) is close to that of Zn2+ (0.074 nm). Solu-
tion of a small amount of Sc2O3 in ZnO lattice will result in the
formation of cationic vacancies unless it is compensated by solution
of an equivalent amount of Pd+. Yet Pd+ is not stable and apparently
does not dissolve appreciably in ZnO lattice because of the much
larger ionic radius (known r(Pd2+) = 0.085 nm). Thus the Schottky
defects in the form of cationic vacancies due to dissolved Sc2O3
in ZnO lattice will diffuse to the ZnO surface, where (PdZn)0–Pdn+
(n = 1 or 2) clusters can be better stabilized through the Pdn+ accom-
modated at the surface vacant cation-sites to achieve valence and
charge compensation for subsurface layers of Sc3+-doped ZnO. This
would be conducive to inhibiting removal and sintering of the
catalytically active (PdZn)0 nanoparticles, and thus, markedly pro-
longing the life of the catalyst (see Fig. 5(a)).
In the MSR process catalyzed by CNT-promoted PdZn/ZnO, the
H(a) generated from dehydrogenation of CH3OH (as well as the
intermediate products, HCHO and HCOOH) may in situ transfer
to the H-adsorption-sites at the CNTs surface, since the highly
conductive CNTs might promote hydrogen spillover from the
PdZn/ZnO-sites to the adsorption-sites of CNT-surface, combine
to form H2(a) and then desorb to H2(g), a scenario similar to the
case of Cu10Cr1/CNTs-catalyzed deep dehydrogenation of CH3OH
to form H2 and CO [12,20]. This would be favorable to increas-
ing the rate of a series of surface dehydrogenation reactions in
the MSR process. The ratio of relative area-intensities of H2-TPD
profiles in the region-II for those catalysts was estimated to be:
S(a)/S(b)/S(c) = 47/56/100. This implied that the sequence of increas-
ing H-adsorption sites at the functioning surface of those catalysts
was: Pd0.15Zn1 < Pd0.15Zn1–10%CNTs < Pd0.15Zn1Sc0.067–10%CNTs,
in line with the observed sequence of reactivity for MSR on those
catalysts.
3.4.4. H2-TPD test of hydrogen adsorption on pre-reduced
catalysts
4. Concluding remarks
The previous TPD investigation [16] of H2/CNTs adsorption sys-
tem showed that hydrogen adsorption on the CNTs can occur at
ambient temperatures, and that H2 adsorption on the CNTs may
occur in two forms: associative (molecular state) and dissocia-
tive (atomic state), which had been previously confirmed by laser
Raman spectroscopy characterization of the H2/CNTs adsorption
system [15].
Fig. 10 showed the H2-TPD profiles of hydrogen adsorbed on
the three pre-reduced catalysts. Overall, each profile contained
a lower-temperature peak centered at ∼353 K and a higher-
temperature peak at ∼748 K (spanning from 473 to 823 K). The
lower-temperature peaks resulted from the desorption of weakly
adsorbed H-species, most probably the molecularly adsorbed
hydrogen H2(a), while the higher-temperature peaks may be
attributed to the desorption of strongly adsorbed H-species, per-
haps the dissociatively adsorbed hydrogen H(a). It is conceivable
that in the reaction-temperature region for MSR (≥473 K for the
present work), the surface concentration of H-adspecies associated
with the 298–473 K region (region-I) was very low, and most of
(1) A type of highly efficient Pd–ZnO catalysts doubly promoted by
CNTs and Sc2O3 for MSR was developed.
(2) The CNTs could act as a novel promoter of the Pd–ZnO based
catalyst for MSR. The highly conductive CNTs might pro-
mote hydrogen spillover from the PdZn/ZnO-sites to the CNTs
adsorption-sites. The H(a) generated from dehydrogenation
of CH3OH (as well as the intermediate products, HCHO and
HCOOH) may in situ transfer to the surface adsorption-sites of
CNTs, and combine to form H2(a) and then desorb to H2(g),
which would be favorable to increasing the rate of a series of
surface dehydrogenation reactions in the MSR process.
(3) Sc2O3 could act as a novel modifier of the Pd–ZnO based cata-
lyst for MSR. Solution of a small amount of Sc2O3 in ZnO lattice
resulted in the formation of Schottky defects in form of cationic
vacancies at the surface of ZnO, where the (PdZn)0–Pdn+ clus-
ters can be better stabilized through the Pdn+ accommodated
at the surface vacant cation-sites. This would be conducive
to inhibiting removal and sintering of the catalytically active
(PdZn)0 nanoparticles.

144
L. Yang et al. / Applied Catalysis A: General 455 (2013) 137–144
Acknowledgements
[6] N. Iwasa, S. Masuda, N. Ogawa, N. Takezawa, Appl. Catal. A: Gen. 125 (1995)
145–157.
[7] Y. Suwa, S. Ito, S. Kameoka, K. Tomishige, K. Kunimori, Appl. Catal. A: Gen. 267
(2004) 9–16.
[8] A. Karim, T. Conant, A. Datye, J. Catal. 243 (2006) 420–427.
[9] Y.H. Chin, R. Dagle, J. Hu, A.C. Dohnalkova, Y. Wang, Catal. Today 77 (2002)
79–88.
The work is supported by “973” project (2011CBA00508), NSFC
project (20923004), PCSIRT (No. IRT1036) and Fujian Provincial Key
Sci. projects (2009HZ0002-1) of China.
[10] H. Lorenz, S. Penner, W. Jochum, C. Rameshan, B. Klötzer, Appl. Catal. A: Gen.
358 (2009) 203–210.
Appendix A. Supplementary data
[11] P. Serp, M. Corrias, P. Kalck, Appl. Catal. A: Gen. 253 (2003) 337–358.
[12] H.B. Zhang, G.D. Lin, Y.Z. Yuan, Curr. Topics Catal. 4 (2005) 1–21.
[13] H.B. Zhang, X.L. Liang, X. Dong, H.Y. Li, G.D. Lin, Catal. Surv. Asia 13 (2009)
41–58.
[14] P. Chen, H.B. Zhang, G.D. Lin, Q. Hong, K.R. Tsai, Carbon 35 (1997) 1495–1501.
[15] H.B. Zhang, G.D. Lin, Z.H. Zhou, X. Dong, T. Chen, Carbon 40 (2002) 2429–2436.
[16] Z.H. Zhou, X.M. Wu, Y. Wang, G.D. Lin, H.B. Zhang, Acta Phys. Chim. Sin. 18
(2002) 692–698.
[17] P.A. Webb, C. Orr, Analytical Methods in Fine Particle Technology, Micromeritics
Instrument Co., Norcross, GA, USA, 1997.
[18] XRD Data Bank Attached to X’Pert PRO X-ray Diffractometer (PANalytical, The
Netherlands), 2003.
Supplementary data associated with this article can be
found, in the online version, at http://dx.doi.org/10.1016/j.apcata.
2013.01.039.
References
[1] A.P. Tsai, M. Yoshimura, Appl. Catal. A: Gen. 214 (2001) 237–241.
[2] Y. Liu, T. Hayakawa, K. Suzuki, S. Hamakawa, T. Tsunoda, T. Ishii, M. Kumagai,
Appl. Catal. A: Gen. 223 (2002) 137–145.
[19] J.F. Moulder, W.F. Stickle, P.E. Sobol, K.D. Bomben, Handbook of X-ray
Photoelectron Spectroscopy – A Reference Book of Standard Spectra for Identi-
fication and Interpretation of XPS Data, Physical Electronics Inc., Eden Prairie,
1995.
[20] H.B. Zhang, X. Dong, Y.Z. Yuan, G.D. Lin, K.M. Dong, K.R. Tsai, pp. 2–156, 267–268
13th ICC (Paris, 2004) Book of Abstract 1, 2004.
[3] X.R. Zhang, P. Shi, J. Zhao, M. Zhao, C. Liu, Fuel Process. Technol. 83 (2003)
183–192.
[4] S.R. Segal, K.B. Anderson, K.A. Carrado, C.L. Marshall, Appl. Catal. A: Gen. 231
(2002) 215–226.
[5] N. Iwasa, S. Kudo, H. Takahashi, S. Masuda, N. Takezawa, Catal. Lett. 19 (1993)
211–216.