Ethanol steam reforming over Co-based catalysts: Role of oxygen mobility

Article abstract of DOI:10.1016/j.jcat.2008.11.006

The effect of oxygen mobility on the bio-ethanol steam reforming of ZrO₂- and CeO₂-supported cobalt catalysts was investigated. The supported catalysts were prepared by incipient wetness impregnation (IWI) and characterized through N

Full text of DOI:10.1016/j.jcat.2008.11.006

Journal of Catalysis 261 (2009) 66–74
Contents lists available at ScienceDirect
Journal of Catalysis
www.elsevier.com/locate/jcat
Ethanol steam reforming over Co-based catalysts: Role of oxygen mobility
∗
Hua Song, Umit S. Ozkan
Department of Chemical & Biomolecular Engineering, The Ohio State University Columbus, OH 43210, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 10 July 2008
Revised 4 November 2008
Accepted 4 November 2008
Available online 2 December 2008
The effect of oxygen mobility on the bio-ethanol steam reforming of ZrO - and CeO -supported cobalt
2
2
catalysts was investigated. The supported catalysts were prepared by incipient wetness impregnation
IWI) and characterized through N2 physisorption, X-ray photoelectron spectroscopy, temperature
(
programmed oxidation, laser Raman spectroscopy, diffuse reflectance infrared Fourier transform spectro-
scopy, O2 pulse chemisorption, isotopic labeling, and transmission electron microscopy techniques at
various life stages of the catalyst. The results indicated that the catalyst deactivation was due mostly to
deposition of various types of carbon on the surface although cobalt sintering could also be contributing
to the deactivation. The addition of ceria was found to improve the catalytic stability as well as activity,
primarily due to the higher oxygen mobility of ceria.
Keywords:
Cobalt catalyst
Zirconia
Ceria
Bio-ethanol steam reforming
Deactivation
© 2008 Elsevier Inc. All rights reserved.
Carbon deposition
Carbon nanofiber
Oxygen mobility
◦
1. Introduction
C–C bond cleavage at temperatures as low as 350–400 C [10–12].
At these temperatures, researchers have reported good selectivity
to CO2 and H2 with CH4 being the only by-product.
As the wide-spread application of fuel cells becomes closer to
reality, increased attention has been focused on hydrogen produc-
tion technologies. Bio-ethanol steam reforming (BESR) offers an
environmentally friendly route for hydrogen production from re-
newable sources, with potentially little contribution to green house
effect, since CO2 can be recycled through photosynthesis during
the plant growth. Bio-ethanol can be obtained from fermentation
of biomass (e.g. sugar cane, cellulose, corn). Because of its low toxi-
city and ease of deliverability, ethanol steam reforming lends itself
very well to a distributed-production strategy. Development of a
non-precious catalytic system with high activity and stability will
be an important step in making this technique economically com-
petitive.
Catalyst deactivation during BESR has been reported over vari-
ous catalyst systems [13–16], especially over supported Ni catalysts
[17–24]. Furthermore, in most of these studies, the metal sinter-
ing and the deposition of carbonaceous species such as amorphous
carbon, carbon filament, and carbonate species have been recog-
nized as the main causes of catalyst deactivation. The number
of studies examining the activity of supported cobalt catalysts is
fewer [25–27], most detailed studies being reported by Llorca and
co-workers [11,12,28–32] and the most commonly used support
being ZnO.
In recent years, increased attention has been given to the role
of oxygen mobility in improving the catalytic stability during BESR
through preventing metal particles from sintering [33,34] and sup-
pressing the formation of carbonaceous species by gasification.
While most of these studies are focused on the utilization of
CeO2 or ZrO2–CeO2 mixed oxide supported noble-metal catalysts
With increased interest in hydrogen as an energy carrier for
fuel cells, studies in the literature on ethanol steam reforming cat-
alysts have also increased significantly in recent years. Catalysts
utilized are mainly Ni, Cu, Co and supported noble metals, such
as Rh, Ru, Re, Pd, Ir and Pt [1–4], which have been reviewed by
Haryanto et al. [5] and Vaidya et al. [6]. Although supported no-
ble metal catalysts have been shown to have significant activity
[35–37], the importance of a similar phenomenon for non-precious
metal catalysts such Co has received much less attention.
In our earlier publications [38,39], we have reported on the
catalytic activity of Co/ZrO2 catalysts in ethanol steam reform-
ing and the effect of synthesis parameters on catalytic perfor-
mance. In this study, we focus on the deactivation behavior of
these Co-based catalysts and the effect of the oxygen mobility
of the support on catalytic activity and stability. Ceria, which is
known for its high oxygen storage capacity, has been reported
to enhance the catalytic activity and stability in several reactions
◦
in 500–600 C range and at high space velocities [7–9], high cost
of these metals limits their application. As a less expensive alter-
native, cobalt-based catalysts have been reported to have superior
ethanol steam reforming performance due to their high activity for
Corresponding author.
E-mail address: ozkan.1@osu.edu (U.S. Ozkan).
*
0
021-9517/$ – see front matter © 2008 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2008.11.006

H. Song, U.S. Ozkan / Journal of Catalysis 261 (2009) 66–74
67
such as CO oxidation, water–gas shift, and steam reforming of
methane [40–42]. In this paper, the role of CeO2 addition in im-
proving the stability of Co-based catalysts, as examined through
characterization studies that included N2 physisorption, X-ray pho-
toelectron spectroscopy (XPS), temperature-programmed oxidation
were sandwiched between two layers of quartz wool and subjected
to an oxidative cleaning step in air for 30 min, at the calcination
temperature of the sample, followed by a reduction step at 400 C
◦
for 2 h. After the system was flushed with He to remove any mois-
◦
ture, the O2 pulses were introduced at 300 C. The m/z = 32 signal
(TPO), laser Raman spectroscopy (LRS), diffuse reflectance infrared
was continuously detected by the MS until there was no variation
between two consecutive peaks.
Fourier transform spectroscopy (DRIFTS), O2 pulse chemisorption,
16
18
O2/ O2 exchange, and transmission electron microscopy (TEM)
techniques, is reported.
16
18
O2/ O2 exchange experiments were performed using
a
Thermo Finnegan Trace Ultra DQC GC/MS. ∼50 mg samples were
placed in a U-tube quartz reactor. Following a cleaning step un-
2. Experimental
◦
der He at 400 C for 30 min to remove impurities adsorbed during
◦
storage, the reactor was cooled down to 300 C, at which tempera-
2.1. Catalysts preparation
ture the 16 O2/ O2 exchange took place. The m/z = 32, 34, and 36
18
(
mass-to-charge ratio) signals were monitored by the mass spec-
Supported cobalt catalysts with 10 wt% metal loading were
trometer during the exchange process. 10% Ar was included in the
prepared in air by incipient wetness impregnation from cobalt(II)
nitrate hexahydrate (Aldrich 99.999%) aqueous solutions. The pure
16
2%
O2/He stream to account for the gas-phase hold-up time, as
2
described previously [43–45]. In addition, blank experiments were
also performed and showed no exchange in the gas phase when
no catalyst was present.
The TEM experiments were performed by using Philips Tecnai
TF-20 TEM instrument operated at 200 kV. An X-ray analyzer for
EDS is incorporated into the instrument for elemental analysis un-
der STEM mode for improving image contrast between C and Co
phases. The sample was first dispersed in ethanol and supported
on Lacey-formvar carbon on a 200 mesh Cu grid before the TEM
images were recorded.
supports used were ZrO2 (Saint Gobain, surface area: 55 m /g,
3
pore volume: 0.21 cm /g) and CeO2 (powder <5 μm, 99.9%
2
3
Aldrich, surface area: 71 m /g, pore volume: 0.34 cm /g). The
0 wt%CeO2–ZrO2 support was prepared by impregnating the ZrO2
1
support using an aqueous solution of cerium(III) nitrate (Aldrich
◦
9
9.999%). All the supports were calcined for 3 h in air at 500 C
prior to metal impregnation. After repeating the impregnation of
◦
cobalt precursor and drying steps (at 95 C) as many times as de-
termined by the pore volume of the corresponding supports, the
◦
resulting samples were calcined at 400 C for 3 h in the air and
stored for use. The term “fresh sample” is used to represent the
sample after calcination; the “reduced sample” denotes the sam-
DRIFTS was performed with a Thermo NICOLET 6700 FTIR spec-
trometer equipped with a liquid-nitrogen-cooled MCT detector and
a KBr beam splitter. The in situ experiments were performed using
a Smart collector DRIFT environmental chamber with ZnSe win-
◦
ple reduced at 400 C for 2 h; and the “spent sample” refers to the
sample after reduction treatment and exposure to reaction atmo-
sphere for various time periods.
◦
dows. Following the pretreatment under He at 400 C for 30 min
◦
and reduction under 5%H2/He at 400 C for 2 h, the environmen-
tal chamber was heated to 450 C for 1 h under He for removing
2
.2. Catalysts characterization and reaction performance measurement
◦
moisture generated from the reduction step. The reactant vapors
generated from a two-bubbler system were then flowed over the
sample for 1 h at room temperature using He as a carrier gas. The
sample was then flushed with He for 10 min. Spectra were taken at
The surface areas of the fresh and spent catalysts were mea-
sured using nitrogen adsorption at 77 K (Micromeritics ASAP 2010).
◦
Before each measurement, the sample was degassed under 130 C
overnight to remove any impurities adsorbed from the atmosphere
during storage.
pre-set intervals while the sample temperature was ramped from
◦
2
5 to 500 C.
XPS analysis was performed using an AXIS His, 165 Spectrom-
eter manufactured by Kratos Analytical with a monochromatized
AlKα X-ray source. 2.3 V voltage was chosen to make the charge
balance. A stainless steel sample holder was used. Survey scans
were performed to identify all the elements within the sample,
followed by more detailed regional scans for Co 2p, C 1s, O 1s, Zr
The catalytic performance measurement and analysis methods
used were reported elsewhere [38]. Briefly, all catalysts were first
◦
pretreated at 400 C for 30 min. under He and then reduced in
◦
situ at 400 C for 2 h under 5%H2/He. The reactant liquid consist-
ing of ethanol and water at 1:10 molar ratio was delivered into an
evaporator. The generated reactant vapor was carried by He and in-
troduced into the reactor. The dilution ratio varied between 40 and
75 (inert-to-ethanol molar ratio). For neat experiments, gas phase
reactants were directly fed to the reactor without dilution with an
inert gas. Subsequently the catalytic performances were tested in
3
d orbitals in order to achieve the high resolution for these ele-
ments of interest. A controlled-atmosphere transfer chamber was
used for transferring the sample to the XPS instrument without
exposure to atmosphere.
Temperature-programmed oxidation (TPO) experiments were
performed using Autochem-2920 (Micromeritics) with an online
mass spectrometer (MS) (MKS Instruments, 1–300 amu). The sam-
◦
the temperature range of 300 to 550, in 50 C increments. The cat-
alyst was held at each temperature for at least 2 h. At the end of
◦
ples were first pretreated at 300 C with He for 30 min in order to
the catalytic test, the flow of EtOH +H2O was stopped and the cat-
remove adsorbed contaminants during storage. After cooling down
to room temperature under helium, 10%O2/He (30 ml/min) was in-
troduced into the reactor and TPO experiments were subsequently
alyst was cooled under He stream. The hydrogen yield is defined
moles of H2 produced
×(moles of ethanol fed)
as H2 yield % = 6
× 100. The time-on-stream
◦
◦
(TOS) tests were performed at 450 C for different time periods
depending on the deactivation rate of various samples. The spent
samples after TOS experiments were cooled down to room temper-
ature under helium before characterization.
performed with a heating rate of 10 C/min after the MS signal
was stable.
Raman spectra were taken with a LabRAM HR-800 spectrometer
equipped with an OLYMPUS BX41 microscope (50× magnification)
and a CCD detector. An argon ion green laser (514.5 nm, operated
at 3 mW) was used as the excitation source during spectra collec-
tion.
O2 pulse chemisorption experiments were conducted using Au-
toChem II 2920 (Micrometrics) connected with a Cirrus Mass Spec-
trometer (MKS Instruments, 1–300 amu). Catalysts of ∼200 mg
The turnover frequency (TOF) reported in the paper is calcu-
lated based on the ethanol conversion rate divided by the total
available metallic cobalt active sites contained over the samples
charged in the reactor. The cobalt dispersion is estimated using
a H2 chemisorption technique, which is described in detail previ-
ously [38].

6
8
H. Song, U.S. Ozkan / Journal of Catalysis 261 (2009) 66–74
Fig. 2. O2 (m/z = 32) and CO2 (m/z = 44) traces during TPO over spent 10%Co/ZrO2.
Fig. 1. Variation of ethanol conversion and product yields with temperature dur-
ing steady-state steam reforming over 10%Co/ZrO2 (EtOH:H2O = 1:10 (molar ratio),
GHSV = 5000 h− , and CEtOH = 1.2%).
1
3
. Results and discussion
3
.1. The deactivation behavior of Co/ZrO2 catalysts in BESR
As we have reported previously [38], Co/ZrO2 catalysts with
a 10%Co loading gave high H2 yields, at temperatures as low as
◦
4
5
50 C. The H2 yield at this temperature is 77% and reaches 92% at
◦
50 C, which is equivalent to 5.5 mol of H2 produced per mol of
ethanol fed. What is also worth noting in Fig. 1 is that at or above
◦
4
4
50 C, there are no liquid products remaining in the system. At
◦
50 C and above, the only by products are CO (less than 2%) and
◦
CH4 (less than 7%). At 550 C, the only H-containing product is
H2, giving it a 100% selectivity. The only other product observed
at this temperature besides H2 and CO2 is CO. At lower tempera-
tures, in addition to the main products which are expected to be
formed during BESR reaction, (i.e., H2 and CO2), small quantities
of acetone, CO, CH4, C2H4, and acetaldehyde were also observed
(a)
◦
at different levels. As seen in the figure, above 350 C, the ethanol
conversion is complete, hence the yield of H2 is determined by
the competing reactions, including ethanol decomposition, dehy-
dration, dehydrogenation, methanation, WGS and reverse WGS re-
actions.
Although the activity and H2 yield over these Co/ZrO2 cata-
lysts were quite promising, the TOS experiments showed that after
◦
about 25 h on stream at 450 C, the activity declined rapidly. There
was also significant pressure build-up in the reactor.
Post-reaction characterization of the spent 10%Co/ZrO2 cata-
lysts provided information as to the nature of catalyst deactivation.
Fig. 2 shows the results of TPO experiment performed over a deac-
tivated catalyst. As the temperature is raised under an oxygen flow,
significant levels of CO2 formation were observed over the catalyst,
indicating that carbon was deposited over the surface during reac-
(
b)
Fig. 3. (a) Co 2p and (b)
C 1s region of the X-ray photoelectron spectra of
10%Co/ZrO2: (1) fresh, (2) reduced, and (3) spent catalyst.
◦
tion. The maximum in the CO2 signal, which takes place at 520 C
is accompanied by a minimum in the O2 signal, showing the oxy-
gen depletion during the oxidation of the carbon deposited on the
surface.
signal in the Co 2p region, suggesting that most of the Co sites
were covered by (or encased in) carbon. Fig. 3b shows C 1s region
of the XPS spectra of 10%Co/ZrO2 catalyst taken at different life
stages, i.e., fresh, reduced and spent. Compared to the fresh and
reduced samples, there was a very strong signal observed over the
spent catalyst, indicating carbon deposition on the surface. Zr 2p
and O 1s regions of the spectra (not shown) also showed reduced
intensities compared to the fresh and reduced samples. However,
this was especially severe for the Co 2p signal.
The XPS spectra of 10%Co/ZrO2 were taken at different life
stages, i.e., fresh, reduced and spent, which are shown in Fig. 3.
The Co 2p3/2 peak centered at 780.3 eV over fresh sample could
be assigned to Co3O4. The peak shift to 778.4 eV was observed
when the reduced sample was analyzed, indicating the existence of
metallic Co [46–49]. Moreover, the absence of the shake-up lines
2
+
nearby the two main Co 2p peaks characteristic of Co
[50–52]
implied the full reduction of cobalt oxide under these reduction
conditions. The spectrum of the spent sample showed a very weak
Raman spectroscopy was used to characterize the carbon de-
posited on the post-reaction catalysts. Fig. 4 shows the Raman

H. Song, U.S. Ozkan / Journal of Catalysis 261 (2009) 66–74
69
3
.2. The effect of support modification with CeO2 on catalytic
performance
While the Co/ZrO2 catalysts deactivated rapidly due to car-
bon deposition on the surface, addition of ceria to the sup-
port showed significant improvement in catalyst stability. Fig. 6
shows a comparison of the TOS performance of 10%Co/ZrO2 and
◦
1
0%Co/10%CeO2–ZrO2 catalysts at 450 C. While the initial activity
and hydrogen yields observed over these two catalysts are compa-
◦
rable at 450 C (∼76%), Co/ZrO2 catalyst shows rapid deactivation
after about 30 h of TOS. There was not much change in selectivity
during the rapid deactivation. The catalyst that contains 10% CeO2
in the support maintains its activity even after 110 h TOS. Ceria ad-
dition also imparts the catalysts with a higher reforming activity,
especially at lower temperatures, as seen in the inset.
The surface areas and pore volume of these two catalysts
are compared in fresh and spent form and the results obtained
through N2 physisorption are presented in Table 1. The Co/ZrO2
catalysts show increased surface area and pore volume after TOS,
while ceria-containing sample does not show much change. The
increase in surface area and pore volume may be due to carbon
deposition on the surface of the Co/ZrO2 catalyst.
Fig. 4. Raman spectra of 10%Co/ZrO2 (a) fresh, (b) spent.
spectra taken over 10%Co/ZrO2 sample, before reaction (reduced)
◦
(
Fig. 4a) and after 45 h on stream at 450 C (Fig. 4b). The two
When Co/CeO2–ZrO2 catalysts were characterized with LRS af-
ter being kept on stream for 110 h, there was no Raman signal that
corresponded to carbon (data not shown), indicating that there
was either no carbon deposition or the carbon on the surface was
below the detection limits of the Raman spectrometer.
Post-reaction TEM characterization of the Co/CeO2–ZrO2 cata-
lysts was also performed (Fig. 7). Surface was found to be mostly
free of any carbon deposition with very few carbon fibers detected.
Co particles appeared to remain intact on the surface. EDS analy-
sis also verified that there was not much carbon deposited on the
surface.
Since incorporation of CeO2 into the zirconia support demon-
strated significantly better catalytic performance, Co catalysts sup-
ported on CeO2-only were tested under more demanding reac-
tion conditions (i.e., higher ethanol concentrations and higher
GHSV). Fig. 8 shows a comparison of the H2 yields achieved over
broad bands observed over the spent catalyst that are centered
−1
at 1340 and 1590 cm
are characteristic of disordered carbona-
ceous and ordered graphitic species [28] and are referred as D
band and G band, respectively. The G band is known to be very
sensitive to the extent of two-dimensional graphitic ordering. The
degree of the disordering of the carbonaceous materials can be
roughly estimated quantitatively based on the integrated intensity
of D and G bands. The larger the microcrystalline planar size La
defined as 44(IG /ID ) is [53], the higher the degree of ordering is
in the carbonaceous materials. In our case, La is estimated to be
2.5 nm after peak fitting by using two Gaussian type bands, which
is higher than the reported value obtained from carbon deposits
over Co-based catalysts [12], indicating the higher degree of order-
ing.
Fig. 5 shows TEM and STEM images of the Co/ZrO2 sam-
ples at three different life stages—calcined (a), reduced (b) and
spent (c–f). The Co3O4 crystallites are represented as darker par-
ticles (a), while the ZrO2 supports are seen as lighter particles.
The lattice fringes are visible for both phases (i.e., Co3O4 and
ZrO2). The average particle size decreased from 24 nm for the
Co3O4 to 12 nm for Co particles, following a reduction treat-
ment. The decrease in size was confirmed by in situ-XRD results
as well [40]. For spent samples, in addition to amorphous carbon,
carbon nanofibers are also observed over 10%Co/ZrO2 with vary-
ing diameters. The cobalt particles appear to be encased at the tip
of the fibers. Figs. 5d and 5f are the STEM images of the same
sample area represented in Figs. 5c and 5e, respectively. Super-
imposed on the STEM images are the EDS analysis results from
the highlighted areas. The area highlighted in Fig. 5d is at the
tip of a fiber and the analysis shows a very strong Co signal. The
area highlighted in Fig. 5f is on a fiber and EDS analysis shows
a very strong C signal. The Cu signal observed in both cases is
due to the copper grid used to support the samples. The carbon
fibers are seen to vary in diameter substantially, depending on the
size of the Co particle that catalyzes its growth. While most of
the fibers are shorter with diameters around 20 nm, there are
some fibers with much larger diameters (∼150 nm). Compared
to the average size of the Co particles of freshly reduced sam-
ples, some of the Co particles encased in carbon fibers appear
to be much larger, suggesting that there may have been sinter-
ing of Co particles during reaction, resulting in larger particle
sizes.
10%Co/10CeO2–ZrO2 and 10%Co/CeO2 catalysts. Under these reac-
tion conditions, Co/ZrO2 catalysts showed very rapid deactivation
and pressure build-up, not allowing data collection at steady-state.
As shown in Fig. 8, Co catalysts supported on ceria gave substan-
tially higher yields for H2 than the ones supported on CeO2–ZrO2
supports in the entire temperature range studied.
The three catalysts (Co/ZrO2, Co/CeO2–ZrO2 and Co/CeO2) were
also compared using TOFs calculated from ethanol conversion data
◦
at 450 C. For these measurements, neat reaction conditions were
used. The TOF data, which are presented in Table 3, show that the
intrinsic activities of the three catalysts increase in the order of
Co/ZrO2 < Co/CeO2–ZrO2 < Co/CeO2.
The long term stability of the three catalysts were tested un-
der neat reaction conditions (without the dilution of an inert
−1
gas) with EtOH:H2O = 1:10 (molar ratio), GHSV = 5000 h
and
◦
CEtOH = ∼7.5%. The Co/CeO2 catalyst was kept on-line at 450 C
for 45 h and showed no change in activity or product distribu-
tion. The H2 yield under these conditions was slightly higher than
60%. The Co/ZrO2 deactivated within the 1st hour and showed se-
vere pressure build-up. Co/CeO –ZrO2 catalyst showed deactivation
2
starting after ∼15 h. These experiments showed that the stability
also increased in the order of Co/ZrO < Co/CeO –ZrO < Co/CeO .
2
2
2
2
Since Co catalysts supported on ceria or ceria-modified zirco-
nia gave much higher yields and showed much better stability and
resistance to coking, the differences between the ceria-containing
and ceria-free catalytic systems were examined using different
techniques.

70
H. Song, U.S. Ozkan / Journal of Catalysis 261 (2009) 66–74
Fig. 5. Electron microscopy of 10%Co/ZrO2: (a) TEM image after calcination, (b) TEM image after reduction, (c and d) TEM images of spent catalysts, (e and f) STEM images of
the same sample areas shown in (c) and (d), respectively. EDS analyses of the highlighted areas are superimposed.
Co crystal sizes were determined using X-ray diffraction. Table 2
summarizes the evolution of cobalt crystal sizes at different life
stages for all three catalysts calculated using Scherrer equation. The
diffraction lines used for these calculations are (311) and (111) for
the Co3O4 and Co phases, respectively. All three catalysts show a
decrease in crystal size upon reduction. The percent increase in
crystal size upon exposure to reaction medium is the largest for
Co/ZrO2 catalyst. Co/CeO2 catalyst shows no increase in crystal size
during reaction.
situ DRIFT spectra taken during ethanol TPD over 10%Co/ZrO2 and
10%Co/CeO2, respectively. At room temperature, in the region be-
−1
tween 3600–3800 cm , the negative features originate from the
molecularly adsorbed ethanol through the formation of hydrogen
bridge bonding with the OH groups of the support. Furthermore,
the molecularly adsorbed ethanol produces the vibrational bands
−1
located at 1323 (δ(CH3)) and 1280 cm
(δ(OH)) [1]. Ethanol ad-
sorption also leads to the formation of monodentate and biden-
tate ethoxy species through disassociation as identified by the
−
1
DRIFTS technique was used to monitor the surface species dur-
ing BESR reaction over 10%Co/ZrO2. Fig. 9 a and b show the in
CH3 bending (1443, 1381 cm ) and CCO stretching (1161, 1110,
−1
1066 cm ) vibrations [29]. The C–H stretching located within

H. Song, U.S. Ozkan / Journal of Catalysis 261 (2009) 66–74
71
Fig. 8. Comparison of H2 yields obtained over 10%Co/10%CeO2–ZrO2 and 10%Co/CeO2
(EtOH:H2O = 1:10 (molar ratio), GHSV = 20,000 h− and CEtOH = 2%).
1
Fig. 6. (a) Product yields obtained over 10%Co/ZrO2 and 10%Co/10%CeO2–ZrO2 dur-
◦
ing ethanol steam reforming collected at 450 C (EtOH:H2O = 1:10 (molar ratio),
GHSV = 5000 h− and CEtOH = 1.2%).
1
Table 2
Cobalt/cobalt oxide crystal size estimation from XRD (nm).
Table 1
Catalyst
Fresha
Reducedb
Spent
Surface area, pore volume and pore size of fresh and spent 10%Co/ZrO2 and
1
0%Co/10%CeO2–ZrO2 catalysts measured using N2 physisorption.
Co/ZrO2
Co/CeO2–ZrO2
Co/CeO2
27
29
31
15
20
25
24
25
25
Catalyst
Surface area
Pore volume
(cm /g)
Pore size
(Å)
2
3
(
m /g)
a
Calculations are based on (311) diffraction line for Co3O4.
Calculations are based on (111) diffraction line for Co.
Co/ZrO2 Fresh
Co/ZrO2 Spent
Co/CeO2–ZrO2 Fresh
Co/CeO2–ZrO2 Spent
49.5
56.4
38.2
38.6
0.202
0.251
0.160
0.163
158.0
170.7
160.3
164.8
b
Table 3
−
1
The TOFs (s ) based on ethanol conversion collected under neat reaction condi-
tions at 450 C: EtOH:H2O = 1:10 (molar ratio), GHSV = 5000 h
◦
−1
and CEtOH = 7.5%.
Co/ZrO2
Co/CeO2–ZrO2
Co/CeO2
0.054
0.214
0.476
catalyst, the appearance and disappearance of each surface species
mentioned above occur at much lower temperatures (shown in
Fig. 9b). For instance, even at room temperature, the vibration
peaks characteristic for acetate species are present, implying the
abundance of oxygen on the surface available for oxidation of ad-
sorbed ethanol species. The fact that almost no species remain
◦
on the surface of ceria-supported sample at 400 C while very
◦
strong acetate bands are still visible even at 500 C over the ZrO2-
supported catalyst is an indication that the oxidation of surface
species is easily facilitated due to the oxygen mobility of the ce-
ria support. This observation is consistent with the resistance to
coking seen over the ceria-containing catalysts.
The in situ DRIFT spectra obtained during ethanol/water TPD
over 10%Co/ZrO2 and 10%Co/10%CeO2–ZrO2 are shown in Fig. 10.
Compared to Fig. 9a, the spectra collected during ethanol/water
TPD spectra over Co–ZrO2 at room temperature shows much
Fig. 7. TEM image taken over spent 10%Co/10%CeO2–ZrO2.
−
000–2700 cm 1 (2970, 2928, 2867, 2710 cm ) comes from the
−1
3
−1
stronger adsorption bands around 3600–3800 cm
due to ad-
CH3– and CH3CH2–groups. Initial temperature increase favors the
ethoxide species adsorbed in bidentate form (stronger 1161 cm
−1
−1
sorbed water. The peak located around 1653 cm
[56] is also due
to the O–H scissoring resulting from adsorbed water. As a result
of water dilution, the vibration peaks corresponding to the molec-
ular or dissociative adsorption of ethanol are much weaker than
the ethanol TPD spectra taken at the same temperature. Many of
the same features are observed as those seen in Fig. 9a, except for
slight shifts. The effect of water addition to facilitate ethanol con-
version can be seen from the disappearance of the surface acetate
species peaks at lower temperatures. It is worth noting that similar
phenomena are observed over CeO2 containing catalyst, however,
the increased accessibility of oxygen and hence the ease of oxida-
peak). The bands characteristic of ethoxy species disappeared with
further temperature increase due to oxidation with the lattice
oxygen from the support. Surface acetate species were observed
−1
−1
to form subsequently at 1552 cm
(νassym(COO)), 1441 cm
−1
(νsym(COO)), and 1346 cm
(δ(CH3)) [54]. These species could
first evolve to monodentate carbonate [1,55] as an intermediate,
then dissociate into CO2, which is seen through the bands at 2361
−
1
and 2338 cm . It was noted that in the region from 2200 to
2
−1
000 cm where adsorbed CO bands would be [56] there were no
peaks during the entire run. Interestingly, over the CeO2-supported

72
H. Song, U.S. Ozkan / Journal of Catalysis 261 (2009) 66–74
(
a)
(a)
(b)
(b)
Fig. 9. In situ DRIFTS over (a) 10%Co/ZrO2 and (b) 10%Co/CeO2 during ethanol TPD.
Fig. 10. In situ DRIFTS over (a) 10%Co/ZrO2 and (b) 10%Co/10%CeO2–ZrO2 during
ethanol/water TPD.
tion is apparent through the disappearance of acetate species at
lower temperatures (Fig. 10b).
The oxygen storage characteristics of Co catalysts supported on
ceria or zirconia and the bare supports were examined through O2
pulse chemisorption. The samples were pre-reduced in situ prior
◦
to pulsing oxygen and monitoring oxygen uptake at 300 C. The
results are presented in Fig. 11. ZrO2 support has no reducibil-
ity and shows no oxygen uptake. Ceria without any Co loading
shows, although small, a non-negligible oxygen uptake, indicating
modest reducibility. Partial reducibility of ceria near the surface
at lower temperatures has been previously observed by other re-
searchers [57,58]. Compared to blank supports, cobalt-loaded sam-
ples display much higher oxygen uptakes, however ceria-supported
samples have a higher oxygen uptake compared to ZrO2-supported
ones, suggesting a greater accessibility of oxygen from the ceria
lattice compared to zirconia. The difference between the oxygen
uptake values is more pronounced when the Co-loaded samples
is compared, suggesting a role for Co in facilitating oxygen mobil-
ity within the support. A similar phenomenon has been previously
reported over noble-metal samples [59–62]. In fact, the quantifica-
tion of the results shows that part of the oxygen uptake for the
ceria-supported samples is attributable to the support.
Fig. 11. Oxygen uptake during O2 pulse chemisorption over ZrO2, CeO2, 10%Co/ZrO2,
and 10%Co/CeO2.
Isotopic oxygen exchange technique has been widely accepted
as a useful tool for investigating oxygen mobility in oxides [43–45].
The inset of Fig. 12a shows the typical profiles collected during

H. Song, U.S. Ozkan / Journal of Catalysis 261 (2009) 66–74
73
oxygen plays a key role in determining both the H2 yield and the
stability of the Co-based catalysts. Ethanol steam reforming can be
considered as a redox reaction, where ethanol is oxidized by the
oxygen species originating from water. If there is sufficient oxygen
available/accessible, ethanol can be fully oxidized to CO2. However,
if replenishment of oxygen on the surface is slower compared to
its depletion, other byproducts with intermediate oxidation states
will be formed, including CH4, carbon and CO. Hydrogen yield will
be maximized when carbon in ethanol is oxidized all the way to
CO2. Therefore, effective delivery of oxygen to the oxidation sites
plays a key role in determining the hydrogen yield and maintaining
catalyst stability. Ethanol adsorbs preferentially on the Co sites and
water molecules tend to adsorb onto support surface. The ethanol
oxidation seems to take place at the interface of cobalt particles
and the support. The higher oxygen mobility benefits the oxygen
transfer across entire sample surface, resulting in complete oxida-
tion of ethanol and in turn maximization of hydrogen production.
(
a)
4
. Conclusions
Steady-state reaction experiments coupled with post-reaction
characterization experiments showed significant deactivation of
Co/ZrO2 catalysts through deposition of carbon on the surface,
mostly in the form of carbon fibers, the growth of which is cat-
alyzed by the Co particles. The addition of ceria appears to improve
the catalyst stability due to its high OSC and high oxygen mobil-
ity, allowing gasification/oxidation of deposited carbon as soon as
it forms. Although Co sintering is also observed, especially over the
ZrO2-supported catalysts, it does not appear to be the main mode
of deactivation. The high oxygen mobility of the catalyst not only
suppresses carbon deposition and helps maintain the active sur-
face area, but it also allows delivery of oxygen to close proximity
of ethoxy species, promoting complete oxidation of carbon to CO2,
resulting in higher hydrogen yields. Overall, oxygen accessibility of
the catalyst plays a significant role on catalytic performance during
BESR.
(
b)
Fig. 12. (a) Oxygen exchange over ZrO2, CeO2, 10%Co/ZrO2, and 10%Co/CeO2 mea-
sured using 16 O2/ O2 switch and (b) Effect of temperature on oxygen exchange
over 10%Co/CeO2.
18
16 O2-to-18
O2 and 18 O2-to-16 O2 switches. All the signals are nor-
Acknowledgments
16
16
malized by the total counts. During the first switch,
O O sig-
nal (m/z = 32) decreases, accompanied by a rise of 18 O18 O signal
We gratefully acknowledge the funding from the US Depart-
ment of Energy through the Grant DE-FG36-05GO15033.
1
6
18
(
m/z = 36). The
O
O signal (m/z = 34) is shown to go through
18
1
6
a maximum. The
O
O formation is derived from the dissocia-
1
8
18
References
tive adsorption of O2 and subsequent recombination of the
O
atom created and 16 O atom originally attached to the sample. This
mechanism has been referred as going through a “three-atom com-
plex” by Winter [63]. The doubly labeled oxygen results from a
[
1] A. Erd o˝ helyi, J. Raskó, T. Kecskés, M. Tóth, M. Dömök, K. Baán, Catal. Today 116
2006) 367–376.
2] D.K. Liguras, D.I. Kondarides, X.E. Verykios, Appl. Catal. B 43 (2003) 345–354.
(
[
18
16
“four-atom” complex on the surface. Following the O2-to- O2
[3] J.R. Salge, G.A. Deluga, L.D. Schmidt, J. Catal. 235 (2005) 69–78.
[4] V. Fierro, O. Akdim, C. Mirodatos, Green Chem. 5 (2003) 20–24.
switch, relaxation and rise curves are essentially identical as well
as the signal for the cross-labeled oxygen. After integrating the
peak areas of corresponding oxygen species and subtracting the
gas-phase holdup contribution determined from the Ar signal as
described previously [56–58], the amount of exchanged oxygen
atoms was calculated using a technique described previously in the
literature [64] for the two bare supports and ceria- and zirconia-
supported catalysts. As seen in Fig. 12a, not only the CeO2 support,
but also the CeO2-supported Co sample presents notably higher
oxygen mobility than their zirconia counterparts. The effect of the
exchange temperature was further examined over Co–CeO2 sample
and shown in Fig. 12b. As expected, higher temperatures facilitate
the lattice diffusion of oxygen atom, resulting in higher oxygen
[
5] A. Haryanto, S. Fernando, N. Murali, S. Adhikari, Energy Fuels 19 (2005) 2098–
106.
6] P.D. Vaidya, A.E. Rodrigues, Chem. Eng. J. 117 (2006) 39–49.
2
[
[7] S. Cavallaro, V. Chiodo, S. Freni, N. Mondello, F. Frusteri, Appl. Catal. A 249
(2003) 119–128.
[
[
8] C. Diagne, H. Idriss, A. Kiennemann, Catal. Commun. 3 (2002) 565–571.
9] S. Cavallaro, Energy Fuels 14 (2000) 1195–1199.
[
[
10] J.R. Mielenz, Curr. Opin. Microbiol. 4 (2001) 324–329.
11] J. Llorca, P.R. de la Piscina, J.-A. Dalmon, J. Sales, N. Homs, Appl. Catal. B 43
(2003) 355–369.
[
[
[
12] J. Llorca, N. Homs, J. Sales, P.R. de la Piscina, J. Catal. 209 (2002) 306–317.
13] E.C. Wanat, K. Venkataraman, L.D. Schmidt, Appl. Catal. A 276 (2004) 155–162.
14] T. Montini, L.D. Rogatis, V. Gombac, P. Fornasiero, M. Graziani, Appl. Catal. B 71
(2007) 125–134.
[15] J.R. Hansen, C.H. Christensen, J. Sehested, S. Helveg, J.R.R. Nielsen, S. Dahl,
Green Chem. 9 (2007) 1016–1021.
16] B.C. Zhang, X.L. Tang, Y. Li, W.J. Cai, Y.D. Xu, W.J. Shen, Catal. Commun. 7 (2006)
67–372.
17] A.L. Alberton, M.M. Souza, M. Schmal, Catal. Today 123 (2007) 257–263.
◦
mobility. When the temperature reaches 550 C, almost all of the
oxygen atoms stored in the sample become accessible to partici-
pate in the exchange process.
When the results from different characterization experiments
are combined, the emerging picture suggests that the availability of
[
3
[
[
18] M.C. Sánchez, R.M. Navarro, J.L.G. Fierro, Int. J. Hydrogen Energy 32 (2007)
1462–1471.

74
H. Song, U.S. Ozkan / Journal of Catalysis 261 (2009) 66–74
[19] M.N. Barroso, M.F. Gomez, L.A. Arrúa, M.C. Abello, Appl. Catal. A 304 (2006)
[41] L.V. Mattos, F.B. Noronha, J. Catal. 233 (2005) 453–463.
1
16–123.
[42] N. Laosiripojana, S. Assabumrungrat, Appl. Catal. B 66 (2006) 29–39.
[43] S.A. Driscoll, U.S. Ozkan, J. Catal. 97 (1993) 11524–11529.
[44] U.S. Ozkan, Y. Cai, J. Catal. 149 (1994) 375–389.
[
20] F. Frusteri, S. Freni, V. Chiodo, L. Spadaro, G. Bonura, S. Cavallaro, J. Power
Sources 132 (2004) 139–144.
[
21] H.V. Fajardo, F.L.D. Probst, Appl. Catal. A 306 (2006) 134–141.
[45] U.S. Ozkan, Y. Cai, M.W. Kumthekar, J. Catal. 149 (1994) 390–403.
[46] D. Pietrogiacomi, S. Tuti, M.C. Campa, V. Indovina, Appl. Catal. B 28 (2000) 43–
54.
[22] A.N. Fatsikostas, D.I. Kondarides, X.E. Verykios, Catal. Today 75 (2002) 145–155.
23] S. Freni, S. Cavallaro, N. Mondello, L. Spadaro, F. Frusteri, J. Power Sources 108
[
(
2002) 53–57.
[47] K. Takanabe, K. Nagaoka, K. Nariai, K.-I. Aika, J. Catal. 230 (2005) 75–85.
[48] R. Riva, H. Miessner, R. Vitali, G.D. Piero, Appl. Catal. A 196 (2000) 111–123.
[49] A.A. Khassin, T.M. Yurieva, V.V. Kaichev, V.I. Bukhtiyarov, A.A. Budneva, E.A.
Paukshtis, V.N. Parmon, J. Mol. Catal. A 175 (2001) 189–204.
[50] L.F. Liotta, G.D. Carlo, G. Pantaleo, A.M. Venezia, G. Deganello, Appl. Catal. B 66
(2006) 217–227.
[
[
[
24] A.J. Akande, R.O. Idem, A.K. Dalai, Appl. Catal. A. 287 (2005) 159–175.
25] H. Wang, J.L. Ye, Y. Liu, Y.D. Li, Y.N. Qin, Catal. Today 129 (2007) 305–312.
26] S. Tuti, F. Pepe, Catal. Lett. 122 (2008) 196–203.
[
27] S.S.Y. Lin, D.H. Kim, S.Y. Ha, Catal. Lett. (2008) 295–301.
28] J. Llorca, N. Homs, J. Sales, J.L.G. Fierro, P.R. de la Piscina, J. Catal. 222 (2004)
70–480.
29] J.M. Guil, N. Homs, J. Llorca, P.R. de la Piscina, J. Phys. Chem. B 109 (2005)
0813–10819.
30] N. Homs, J. Llorca, P.R. de la Piscina, Catal. Today 116 (2006) 361–366.
31] J. Llorca, J.A. Dalmon, P.R. de la Piscina, N. Homs, Appl. Catal. A 243 (2003)
61–269.
32] J. Llorca, A. Casanovas, T. Trifonov, A. Rodríguez, R. Alcubilla, J. Catal. 255 (2008)
28–233.
33] J.C. Vargas, S. Libs, A.C. Roger, A. Kiennemann, Catal. Today 107–108 (2005)
17–425.
[
[
[
4
[51] F.B. Noronha, C.A. Perez, M. Schmal, R. Frety, Phys. Chem. Chem. Phys. 1 (1999)
2861–2867.
1
[52] M. Haneda, Y. Kintaichi, N. Bion, H. Hamada, Appl. Catal. B 46 (2003) 473–482.
[53] F. Tuinstra, J.L. Koenig, J. Chem. Phys. 53 (1970) 1126.
[54] J. Raskó, M. Dömök, K. Baán, A. Erdöhelyi, Appl. Catal. A 299 (2006) 202–211.
[55] L.V. Mattos, F.B. Noronha, J. Catal. 233 (2005) 453–463.
[56] S.Yu. Venyaminov, F.G. Prendergast, Anal. Biochem. 248 (1997) 234–245.
[57] A. Trovarelli, Catal. Rev. Sci. Eng. 38 (1996) 439–445.
[58] W. Cai, B. Zhang, Y. Li, Y. Xu, W. Shen, Catal. Commun. 8 (2007) 1588–1594.
[59] A. Trovarelli, Catal. Rev. Sci. Eng. 38 (1996) 439.
[
2
[
[
2
4
[
[
34] J.C. Vargas, S. Libs, A.C. Roger, A. Kiennemann, Chem. Eng. Trans. 4 (2004) 247.
35] M. Virginie, M. Araque, A.C. Roger, J.C. Vargas, A. Kiennemann, Catal. Today 138
[60] J. Mikulová, J. Barbier Jr, S. Rossignol, D. Mesnard, D. Duprez, C. Kappenstein,
J. Catal. 251 (2007) 172–181.
(
2008) 21–27.
[61] E.M. Sadovskaya, Y.A. Ivanova, L.G. Pinaeva, G. Grasso, T.G. Kuznetsova, A. van-
Veen, V.A. Sadykov, C. Mirodatos, J. Phys. Chem. A 111 (2007) 4498–4505.
[62] W. Cai, F. Wang, E. Zhan, A.C. Van Veen, C. Mirodatos, W. Shen, J. Catal. 257
(2008) 96–107.
[
36] S.M. Lima, I.O. Cruz, G. Jacobs, B.H. Davis, L.V. Mattos, F.B. Noronha, J. Catal. 257
(2008) 356–368.
[
37] H.S. Roh, Y. Wang, D.L. King, Top. Catal. 49 (2008) 32–37.
[38] H. Song, L.Z. Zhang, U.S. Ozkan, Catal. Today (2007) 346–354.
[39] H. Song, L.Z. Zhang, U.S. Ozkan, Green Chem. 9 (2007) 686–694.
[40] A. Yee, S.J. Morrison, H. Idriss, J. Catal. 186 (1999) 279–295.
[63] E.R.S. Winter, J. Chem. Soc. A 2889 (1968).
[64] J.S.J. Hargreaves, S.D. Jackson, G. Webb, Imprerial College Press, 2006, pp. 115–
130.