Study of catalyst deactivation and reaction mechanism of steam reforming, partial oxidation, and oxidative steam reforming of ethanol over Co/CeO 2 catalyst

Article abstract of DOI:10.1016/j.jcat.2009.09.025

The mechanisms of Co/ceria catalyst deactivation during steam reforming, oxidative steam reforming, and partial oxidation of ethanol were explored by comparing the results from different characterization techniques with those obtained from catalytic testi

Full text of DOI:10.1016/j.jcat.2009.09.025

Journal of Catalysis 268 (2009) 268–281
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Study of catalyst deactivation and reaction mechanism of steam reforming,
partial oxidation, and oxidative steam reforming of ethanol over Co/CeO catalyst
2
a,1
a
a
b
b
Sania M. de Lima , Adriana M. da Silva , Lídia O.O. da Costa , Uschi M. Graham , Gary Jacobs ,
b
a
a,
*
Burtron H. Davis , Lisiane V. Mattos , Fábio B. Noronha
a
Instituto Nacional de Tecnologia – INT, Av. Venezuela 82, CEP 20081-312, Rio de Janeiro, Brazil
Center for Applied Energy Research, The University of Kentucky, 2540 Research Park Drive, Lexington, KY 40511, USA
b
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 29 June 2009
Revised 25 September 2009
Accepted 26 September 2009
Available online 4 November 2009
The mechanisms of Co/ceria catalyst deactivation during steam reforming, oxidative steam reforming,
and partial oxidation of ethanol were explored by comparing the results from different characterization
techniques with those obtained from catalytic testing in a fixed-bed reactor. The nature of carbon depo-
sition and the reaction conditions played critical roles in determining the extent of a catalyst deactiva-
tion. To shed light on the modes of carbon deposition under different reaction conditions, the
mechanisms by which the adsorbed surface species turned over on the catalyst surface were evaluated
using diffuse reflectance infrared spectroscopy under reaction conditions and temperature-programed
desorption of adsorbed ethanol. In steam reforming, ethoxy species were converted to acetate and steam
promoted forward acetate demethanation. The resulting methane decomposed on Co metal particles. In
this case, carbon diffused through the Co particle, nucleating growth sites for filamentous carbon behind
Keywords:
Hydrogen production
Ethanol steam reforming
Ethanol partial oxidation
Ethanol oxidative steam reforming
Co/CeO
2
catalyst
it, with the resulting filaments lifting Co from the support. High H
2
O/ethanol ratios and oxygen promoted
Deactivation mechanism
cleaning of the cobalt surface.
Ó 2009 Elsevier Inc. All rights reserved.
1
. Introduction
reaction and as such, high operation temperatures are necessary
1]. In addition, large amounts of water are required, implying high-
[
The environmental goal of decreasing atmospheric CO
2
by
er energy consumption requirements for vaporization. An alterna-
tive reaction for hydrogen production is POX, an exothermic
reaction exhibiting quick start-up and response times. This choice
also offers the potential of a more compact reactor design, a desir-
able feature for mobile fuel cell applications [10]. The main disad-
vantages are the likelihood of hot-spot formation, making control
of the reaction difficult, and low hydrogen yield. OSR, also known
as autothermal reforming, combines the SR and POX reactions in
one reactor, and thus, if carefully balanced, precludes the need for
an external heat supply [14].
switching in part to renewable resources, combined with the rela-
tively low cost of sugar cane and other plant-based resources in
certain countries, has been the driving forces behind hydrogen pro-
duction research aimed at utilizing bio-ethanol as an environmen-
tally benign resource.
Different strategies have been proposed to produce hydrogen
from ethanol: steam reforming (SR) (reaction 1); partial oxidation
POX) (reaction 2); and oxidative steam reforming (OSR) (reaction
) [1].
(
3
Depending on the fuel cell application, one technology may of-
fer potential advantages over another. However, all these technol-
ogies face one common drawback: several reaction pathways may
occur depending on the reaction conditions and catalyst used. Con-
sequently, some of these reactions lead to the formation of coke,
which can in turn induce catalyst deactivation.
The nature of the support and metal directly influences the
product distribution and catalyst stability during ethanol conver-
sion reactions [2,3]. In spite of their lower activity relative to the
supported metals, metal oxide catalysts can be used to produce
hydrogen that is virtually free of CO and with little carbon deposi-
tion on the catalyst surface, provided that the appropriate reaction
conditions are selected [18–21]. However, a wide range of undesir-
able by-products (e.g., ethylene, acetaldehyde, and acetone) is
C
C
C
2
2
2
H
H
H
5
5
5
OH þ 3H
OH þ 1:5O
OH þ 2H O þ 0:5O
2
O ! 2CO
2
þ 6H
þ 3H
! 2CO
2
ð1Þ
ð2Þ
ð3Þ
2
! 2CO
2
2
2
2
2
þ 5H
2
While SR has been extensively studied [2–9], there are few studies
in the open literature that focus on the POX [10–13] and OSR [14–
1
7] reactions. The advantage of SR for hydrogen production applica-
tions is high hydrogen yield. However, it is a highly endothermic
*
Corresponding author. Fax: +55 21 2123 1166.
E-mail address: fabio.bellot@int.gov.br (F.B. Noronha).
Present address: Universidade Estadual do Oeste do Paraná – Unioeste, Campus
1
de Toledo, Rua da Faculdade, 645, Jd. La Salle, CEP 85903-000, Toledo, Brazil.
0
021-9517/$ - see front matter Ó 2009 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2009.09.025

S.M. de Lima et al. / Journal of Catalysis 268 (2009) 268–281
269
formed during SR of ethanol over metal oxides in comparison with
the supported metal catalysts, depending on the metal oxide
properties.
crease in hydrogen selectivity and an increase in acetaldehyde
selectivity. According to the authors, the surfaces of Co particles
were oxidized by oxygen from the feed and the cobalt oxide
formed favored the dehydrogenation of ethanol to acetaldehyde.
The addition of a noble metal (e.g., Rh, Ru) was suggested to pre-
vent or at least reduce the oxidation of cobalt particles. It is impor-
tant to stress that the oxidation state of cobalt was not investigated
by surface techniques such as X-ray photoelectron spectroscopy
(XPS) or XANES. Raman spectroscopy results demonstrated the
presence of two types of carbon after reaction, which were re-
moved by oxygen treatment after OSR.
2 3
Al O is generally used as a support but its acid sites promote
the dehydration of ethanol to ethylene [7]. MgO contains strongly
basic sites, which are proposed to be highly active for ethanol
dehydrogenation to acetaldehyde [21]. Ceria and ceria-containing
mixed oxides have also been proposed to be catalytically active
components of supported metal catalysts for ethanol conversion
reactions due to their high oxygen storage capacities, suggested
to improve catalyst stability [16,22–24], and because of their abil-
ity to promote the dissociation of molecules of the type ROH (e.g.,
Therefore, the mechanism of Co-based catalyst deactivation is
still unclear. The aim of this work is to perform a systematic study
of the deactivation mechanism of Co/ceria catalysts during the dif-
ferent ethanol conversion reactions. To achieve this goal, the reac-
2
H O and ethanol) [25]. In addition, the strong metal-support inter-
action has been postulated to prevent metal particle sintering,
which can contribute to catalyst deactivation [26].
Different base metals (Ni, Co, and Cu) and noble metals (Pd, Pt,
Rh, and Ru) have been tested for ethanol conversion reactions.
Since cobalt is a much less costly alternative to noble metals, Co-
based catalysts have been extensively studied for SR
2
tions mechanisms of SR, POX, and OSR over Co/CeO catalyst were
first investigated using a combination of reaction testing, temper-
ature-programed desorption (TPD), and diffuse reflectance infrared
Fourier transform spectroscopy (DRIFTS) measurements recorded
under in situ reaction conditions. In addition, high-resolution
transmission electron microscopy (HR-TEM), scanning transmis-
sion electron microscopy (STEM) and thermogravimetric analysis
(TG) were employed to characterize the carbonaceous deposits
formed. Finally, a deactivation mechanism is proposed.
[
4,17,20,21,26–30], though less so for POX [11] and OSR [17]. In
spite of reportedly high activities and selectivities, Co-based cata-
lysts are not immune to the problem of deactivation. In the litera-
ture, there are only a few publications dedicated to investigating
the root causes of the deactivation of Co-based catalysts [26,31–
3
4] and virtually all of them focus on SR. There are no studies in
the literature comparing the causes of catalyst deactivation for
the different ethanol conversion reactions (SR, POX, and OSR).
The carbon deposits following SR of ethanol have been studied
by Raman spectroscopy, temperature-programed oxidation (TPO),
and transmission electron microscopy (TEM), and include both fil-
amentous and amorphous carbon covering the metallic particle
and the support [31–34]. TEM images of a deactivated Co/ZnO cat-
alyst after SR revealed Co particles either covered by carbonaceous
deposits or located inside carbon filaments [32]. Wang et al. [34]
2
. Experimental
2.1. Catalyst preparation
Cerium oxide was prepared by calcination of cerium (IV) ammo-
nium nitrate at 1073 K (CeO
port by incipient wetness impregnation using an aqueous solution
of Co(NO ).6H O. After impregnation of 10 wt.% cobalt, the samples
were dried at 373 K and calcined under flowing air (50 mL/min) at
73 K, for 2 h.
2 2
). Cobalt was added to the CeO sup-
3
2
2
studied the deposition of carbon during SR over Co/CeO as a func-
6
tion of reaction temperature; between 623 and 723 K, where se-
vere deactivation was observed, the cobalt particles were found
to be completely encapsulated by coke. They proposed that the ac-
tive sites responsible for ethanol dehydrogenation were covered by
coke and as a result, ethanol is preferentially dehydrated to ethyl-
ene. However, the product distribution observed may not corre-
spond to this explanation, as acetaldehyde concentration was
found to increase with time on stream (TOS) at 723 K, contrary
to the view that the dehydrogenation sites were blocked. When
the reaction was carried out at 773 or 823 K, carbon filaments were
detected but the catalyst remained quite stable for 8 h TOS. Above
2
.2. BET surface area
The BET surface areas of the samples were measured using a
Micromeritics ASAP 2020 analyzer by nitrogen adsorption at the
boiling temperature of liquid nitrogen.
2.3. X-ray diffraction (XRD)
XRD measurements were recorded using a RIGAKU diffractom-
eter employing CuKa radiation (k = 1.5406 Å). Data were collected
over the 2h range of 25–75° using a scan rate of 0.04°/step and a
8
73 K, carbon deposits were no longer detected by TEM analysis.
Galeti et al. [33] also studied the effect of reaction temperature
on the performance of CuCoZnAl catalyst during SR. The nature
of the carbon deposits formed was evaluated by Raman spectros-
copy and temperature-programed oxidation experiments. CuCoZ-
nAl rapidly deactivated during SR at 673 and 773 K, whereas the
activity remained unchanged at 873 K. Regardless of the reaction
temperature used, Raman spectra revealed the presence of two
types of carbon: ordered (e.g., graphitic, ordered filamentous)
and disordered (e.g., amorphous, defect-laden filamentous) carbon.
Scanning electron microscopy (SEM) confirmed the formation of
carbon filaments after reaction at 773 K while amorphous carbon
was present at 873 K. They proposed that coke was removed from
the surface of the catalyst by the reverse Boudouard reaction,
improving catalyst stability at this temperature.
scan time of 1 s/step. Scherrer equation was used to estimate the
crystallite mean diameter based on the reflection (311) of Co
3 4
O
particles.
2.4. Temperature-programed reduction (TPR)
The catalyst was pretreated at 473 K for 1 h under a flow of ar-
gon prior to the TPR experiment in order to remove traces of water.
The reducing mixture (1.5% hydrogen in argon) was passed
through the sample (300 mg) at a flow rate of 30 mL/min and the
temperature was increased to 1273 K at a heating rate of 10 K/min.
2.5. Diffuse reflectance spectroscopy (DRS)
Recently, Pereira et al. [17] proposed that the loss of activity of
Co-based catalysts is directly related to the oxidation of metallic Co
2
Diffuse reflectance spectra of a Co/CeO catalyst and a reference
particles during reaction. Co/SiO
2
, Co–Rh/SiO
2
, and Co–Ru/SiO
2
cat-
compound were recorded between 200 and 800 nm on a UV–Vis
NIR spectrometer (Cary 5 - Varian) equipped with a Harrick inte-
grating sphere using the support as a reference.
alysts deactivated during OSR carried out between 623 and 673 K.
The decrease in ethanol conversion was accompanied by a de-

2
70
S.M. de Lima et al. / Journal of Catalysis 268 (2009) 268–281
2.6. TPD of ethanol
tan imaging filter. TEM images were recorded with a Gatan 794
slow-scan charge coupled device (CCD) camera. Prior to the inves-
tigations, minute quantities of powdered sample material were
dispersed onto copper grids equipped with lacey carbon support
structures.
TPD experiments of adsorbed ethanol were carried out in a mi-
cro-reactor coupled to a quadrupole mass spectrometer (Omnistar,
Balzers). Prior to TPD analyses, the samples were reduced under
2
flowing H (30 mL/min) by ramping to 1023 K (10 K/min) and
holding at this temperature for 1 h. After reduction, the system
was purged with helium at 1023 K for 30 min and cooled to room
temperature. The adsorption of ethanol was carried out at room
temperature using an ethanol/He mixture, which was obtained
by flowing He through a saturator containing ethanol at 298 K.
After adsorption, the catalyst was heated at 20 K/min to 773 K un-
der flowing helium (60 mL/min). The products were monitored
using a quadrupole mass spectrometer.
2
.10. Reaction conditions
Direct ethanol decomposition (ED), SR, POX, and OSR were per-
formed in a fixed-bed reactor at atmospheric pressure. Prior to
reaction, catalysts were reduced under pure hydrogen (30 mL/
min) at 1023 K for 1 h and then purged under N
2
at the same tem-
perature for 30 min. All reactions were carried out at 773 K except
for SR, which was also performed at 1073 K. POX was conducted
using an O
tio was based on the work of Cavallaro et al. [35], who showed that
the ethanol conversion is around 100% and the H production
reaches a maximum using an O /ethanol ratio between 0.4 and
2 2
/ethanol molar ratio of 0.5. The choice of O /ethanol ra-
2
.7. DRIFTS
2
DRIFTS spectra were recorded using a Nicolet Nexus 870 spec-
2
trometer equipped with a DTGS-TEC detector. A Thermo Spectra-
Tech cell capable of high pressure/high temperature operation
and fitted with ZnSe windows served as the reaction chamber for
in situ adsorption and reaction measurements. Scans were taken
at a resolution of 4 to give a data spacing of 1.928 cm . The num-
ber of scans taken was 512. The amount of catalyst was ꢁ40 mg.
Samples were first pre-reduced in an external fixed-bed reactor
0
2
.5. For SR, H O/ethanol molar ratios of 3.0 and 10.0 were utilized.
OSR was performed employing a H O/ethanol molar ratio of 3.0
2
2
and an O /ethanol molar ratio of 0.5. The reactant mixtures were
obtained using two saturators containing water and ethanol, which
were maintained at the temperature required to obtain the desired
H O/ethanol and O /ethanol molar ratios. For ED and POX, N
ꢀ
1
2
2
2
(
30 mL/min) and 5.6% O
2
/N
2
mixture (30 mL/min), respectively,
using 200 mL/min H
0 K/min to 1023 K, holding at this temperature for 1 h, and cool-
ing to room temperature. The reactor was purged with helium
prior to switching to 1%O /He (25 ml/min) for 4 h to passivate
the sample. Samples were re-reduced in situ by flowing 200 ml/
min H :He (1:1) and increasing the temperature at 10 K/min to
73 K and holding at this condition for 1 h. The catalyst was purged
2
:He (1:1) by increasing the temperature at
were passed through the saturator with ethanol, and then the reac-
tant mixtures obtained were diluted with N (each N stream flo-
wed at 30 mL/min). In the case of SR, the reactant mixture was
obtained by flowing two N streams (30 mL/min) through each sat-
urator containing ethanol and water separately. For OSR, a flow of
.6% O /N (30 mL/min) and a flow of N (30 mL/min) were passed
1
2
2
2
2
2
5
2
2
2
7
through the saturators containing ethanol and water, respectively.
The partial pressure of ethanol was maintained constant for all
experiments. The variation of partial pressure of water was com-
in flowing He at 773 K, prior to cooling in flowing He to 313 K.
For ethanol adsorption and SR reaction tests, He was bubbled at
ꢁ
15 mL/min through a saturator filled with ethanol and held at
73 K. For SR tests, a second helium stream (ꢁ15 mL/min) was
2
pensated by a decrease in the partial pressure of N .
2
In order to observe the catalyst deactivation within a short
timeframe, a small amount of catalyst was used (20 mg). The sam-
ples were diluted with inert SiC (SiC mass/catalyst mass = 3.0). The
reaction products were analyzed by gas chromatography (Micro
GC Agilent 3000 A) containing two channels for dual thermal con-
ductivity detectors (TCDs) and two columns: a molecular sieve and
a Poraplot U column. The ethanol conversion and selectivity to
products were determined from:
bubbled through a saturator filled with water and held at 298 K.
The two streams were joined at a tee-junction, prior to which 1
psig check valves were placed on the lines to prevent a back-flow
condition. The saturator gas flows and temperatures were set to
2 3 2
provide a H O:CH CH OH stoichiometric ratio of 2:1. Adsorption/
reaction measurements were started at 323 K, and then the tem-
perature was increased at 10 K/min; measurements were recorded
at 373, 473, 573, 673, and 773 K.
ðnethanol
Þ
ꢀ ðnethanolÞexit
For POX, 30 mL/min of 0.75% O
urator filled with ethanol held at 273 K. For OSR, 15 mL/min of
0.75%O /He was bubbled through the H O saturator held at
98 K and another 15 mL/min of 0.75% O /He was bubbled through
2
/He was bubbled through a sat-
fed
X
ethanol
¼
ꢂ 100
ð4Þ
ð5Þ
ðnethanolÞfed
ꢁ
2
2
ðn Þ
x
produced
S
x
¼
ꢂ 100
2
2
ðn Þproduced
total
an ethanol saturator held at 273 K. The two streams were com-
bined at a tee-junction, prior to which 1 psig check valves were
added to prevent the possibility of back-flow. The same tempera-
tures and ramp rates were used as in SR.
where (n
methane, acetaldehyde, or ethylene) and (ntotal
+ moles of CO + moles of CO + moles of methane + moles of
x
)
produced = moles of x produced (x = hydrogen, CO, CO2,
)produced = moles of
H
2
2
acetaldehyde + moles of ethylene (i.e., the moles of water produced
are not included).
2.8. TG analysis
TPO experiments were performed using a TA Instruments TGA
analyzer (SDT Q 600) in order to determine the amount of carbon
formed over the catalyst. Approximately 10 mg of spent catalyst
was heated under air flow from room temperature to 1273 K at a
heating rate of 20 K/min and the weight change was measured.
3. Results and discussion
3.1. Catalyst characterization
2
The BET surface area of the CeO
2
support was very low (9.9 m /
2
2.9. Transmission electron microscopy
g) and not measurably affected by the addition of cobalt (9.8 m /g).
The X-ray diffraction pattern of Co/CeO displayed the main lines
characteristic of CeO (2h = 28.5°, 47.5°, and 56.4°) and Co
(2h = 36.8° and 65.4°) (Fig. 1a). The peak broadening procedure
was used to estimate the mean Co crystallite size by Scherrer
2
HR-TEM and STEM investigations were conducted with a 200-
keV JEOL-2010f field-emission electron microscope equipped with
an Oxford energy-dispersive X-ray detector, a STEM unit and a Ga-
2
3 4
O
3 4
O

S.M. de Lima et al. / Journal of Catalysis 268 (2009) 268–281
271
(
A) Co O4
(
a )
( b )
3
CeO2
(B) 10% Co/CeO2
Co O4
3
468
738
(B)
(A)
20
30
40
50
60
70
80
90
200 300 400 500 600 700 800 900 1000
2θ (°)
λ (nm)
688
(
c )
601
Co O
3
4
603
724
859
1204
1200
Co/CeO2
400
600
800
1000
1400
Temperature (K)
Fig. 1. (a) XRD pattern of Co/CeO
2
catalyst; (b) DRS spectra of the Co/CeO
2
catalyst and Co
3
O
4
; (c) TPR profile of Co/CeO
2
catalyst.
equation. The calcined sample exhibited a Co
9 nm. The DRS spectra obtained from the Schuster–Kubelka–
Munk (SMK) equation for Co and Co/CeO catalysts are pre-
sented in Fig. 1b. The spectrum of Co presented bands at 470
and 734 nm. The DRS spectrum of Co/CeO catalyst exhibited two
bands at 468 and 738 nm. A comparison between the DRS spectra
of the Co/CeO catalyst and reference compound suggests that co-
balt in the calcined Co/CeO catalyst must be present, for the most
part, as Co . The bands around 468 and 738 nm could be attrib-
uted to octahedral Co [36,37]. The TPR profiles of Co
CeO are presented in Fig. 1c and are in agreement with the DRX
and DRS results. The TPR profile of Co presents two hydrogen
uptakes at 601 and 688 K, indicating that reduction occurs in two
steps: Co ? CoO ? Co° (Fig. 1c) [37]. The reduction profile of
Co/CeO catalyst exhibits two peaks at 603 and 724 K that corre-
spond to the two-step reduction of Co particles. This result indi-
cates that the Co/CeO catalyst mainly comprises supported Co
particles. However, we cannot rule out the presence of a small frac-
3
O
4
crystallite size of
3.2. Reaction mechanism
3.2.1. TPD of ethanol
1
3
O
4
2
3
O
4
Fig. 2 presents the TPD profiles of adsorbed ethanol for the Co/
CeO catalyst. At low temperatures, Co/CeO catalyst exhibited the
simultaneous formation of H and CH (397 K). Moreover, ethanol
2
2
2
2
4
2
desorption and CO production were not detected within this tem-
perature range. Some authors [12,39,40] observed the formation of
2
3
O
4
2 4
H , CH , and CO in the low-temperature region during TPD of eth-
3
+
3
O
4
and Co/
anol over Al
2
O
3
- and CeO
2
-supported metals. They suggested that
2
3 4
O
580
541
3
O
4
2
3
O
4
397
H (x 0.3)
2
3
O
4
2
2
+
tion of Co probably linked to the support. The broad reduction
peak at around 700–800 K could also comprise the reduction of
396
5
04
2
+
Co species. In addition to the peaks corresponding to the reduc-
tion of Co , a shoulder was detected at around 859 K as well as
a small peak at 1204 K. The H uptake at 859 K is attributed to
the surface reduction of CeO , while the H consumption at around
204 K is related to the formation of bulk Ce . Reduction of Co
CH (x 4)
4
3 4
O
565
2
C H4
2
2
2
1
2 3
O
oxides to metal occurs along with the reduction of the surface shell
of ceria, which is promoted by the presence of the cobalt metal,
and is located in the broad peak between 700 and 800 K [38].
One view is that, during activation, the metal can facilitate reduc-
560
CH CHO (x 6)
3
tion of the surface shell of ceria from Ce4 to Ce by H
+
3+
dissociation
C H OH
2
2
5
and spillover, leading to the production of bridging OH groups on
the surface associated with reduced Ce3 centers. The bridging
OH groups can be interpreted to be dissociated H O located at an
2
O vacancy. Reduction at 1023 K will likely completely reduce the
ceria surface shell, as well as a fraction of the subsurface layers.
+
3
00
400
500
600
700
800
Temperature (K)
2
Fig. 2. TPD profiles of ethanol desorption obtained for Co/CeO catalyst.

2
72
S.M. de Lima et al. / Journal of Catalysis 268 (2009) 268–281
ethanol adsorbs on the catalyst surface as ethoxy species that can
be decomposed to H , CH , and CO over metal particles. The ab-
sponds to the m(CO) stretching mode of bridge-bonded CO ad-
sorbed on Co particles [44].
2
4
sence of CO desorption in our work suggests that CO still remained
adsorbed in this range of temperature.
When the catalyst was heated to 373 and 473 K, the intensity of
the bands corresponding to ethoxy species decreased and the band
related to molecularly adsorbed ethanol was no longer detected.
Furthermore, the bands assigned to acetaldehyde (1650 and
At high temperatures, significant peaks for H
2
production (541
and 580 K) and a much smaller peak for CH formation (504 K)
4
ꢀ
1
were detected. In addition, peaks corresponding to acetaldehyde
and ethylene formation were observed at 560 and 565 K, respec-
tively. Erdohelyi et al. [8] also observed the production of ethylene
and acetaldehyde at around 500 K over CeO - and Al O -supported
2 2 3
noble metal catalysts. In that case, ethylene and acetaldehyde were
suggested to be produced by dehydration and dehydrogenation of
1713 cm ) were observed [12]. This result confirms that a fraction
of ethanol decomposes to H , CH , and CO and another fraction
2
4
dehydrogenates to acetaldehyde, as suggested by TPD analysis.
At 573 K, weak bands were observed at 1342, 1431, and
ꢀ
1
1533 cm
s 3 s a
corresponding to the d (CH ), m (OCO), and m (OCO)
vibrational modes of acetate species, respectively [12,41,42].
Moreover, the bands corresponding to acetaldehyde were no long-
er detected. The transformation from ethoxy species to acetate
species involves the dehydrogenation of ethoxy species to acetal-
dehyde, which may be further dehydrogenated to acetyl species.
Erdohelyi et al. [8] suggested that acetate species are formed via
two reaction pathways: (i) the reaction between the acetyl species
and oxygen from the support; and/or (ii) the reaction of acetalde-
ethanol, respectively. No significant peaks for CO, CO
hyde, acetone, or benzene were observed by TPD analysis. Accord-
ing to the literature [12,39], the formation of CO and CH at high
temperatures (above 500 K) during TPD of ethanol over CeO and
Al supports is due to the decomposition of carbon species
e.g., acetaldehyde and acetate species) previously formed. Recall-
2
, crotonalde-
4
2
2 3
O
(
ing that acetate species are formed by oxidation of the dehydroge-
nated species, a fraction of the acetate species can be oxidized to
2
hyde with surface OH groups. Since the Co/CeO catalyst was pre-
carbonate species that are, in turn, decomposed to CO
this work, the small peak for CH and the absence of CO and CO
at high temperatures suggest that the formation of acetate species
was not favored over Co/CeO catalyst during ethanol TPD.
Finally, the large production of H detected at high tempera-
tures in the TPD profiles of the Co/CeO catalyst can be assigned
to the desorption of H previously formed during the different
steps of ethanol dehydrogenation.
2
. Then, in
viously reduced at 1023 K, the acetate species were likely produced
from the reaction between adsorbed acetaldehyde and surface OH
groups. Increasing the temperature to 673 K led to a significant de-
crease in the bands related to acetate species. At 773 K, the bands
of acetate species were no longer detected. These results indicate
that the formation of acetate species was not favored over Co/
4
2
2
2
2
2
2
CeO catalyst for the ED reaction, consistent with the decomposi-
tion products arising during TPD.
Recently, Song and Ozkan [45] carried out DRIFTS experiments
after ethanol adsorption over a Co/CeO catalyst. The spectra re-
2
3
.2.2. DRIFTS analysis of ethanol desorption
Fig. 3 shows the IR spectra of adsorbed ethanol on Co/CeO cat-
2
vealed the presence of acetate species even at room temperature,
attributed to the large availability of oxygen from the ceria sup-
port. The difference between their work and ours likely stems from
the difference in reduction temperature used in each work. We
previously reduced our catalyst under pure hydrogen at 1023 K
in order to assure the complete reduction of cobalt oxides, the sur-
face shell reduction of ceria, and the reduction of a fraction of sub-
surface ceria, as previously discussed in the section on TPR. On the
alyst at different temperatures. At room temperature, the bands at
1
spond to different vibrational modes of ethoxy species, which were
formed by dissociative adsorption of ethanol over Ce cations
ꢀ
1
049, 1087, 1356, 1406, 1444, 2862, 2925, and 2969 cm corre-
[
12,41,42]. In addition to the bands of ethoxy species, there are also
ꢀ1
bands at 1265, 1624, and 1892 cm . The bands at 1265 and
ꢀ1
1
624 cm
and to the
tively. The acetyl species are produced by the dehydrogenation of
are assigned to molecularly adsorbed ethanol [42]
2
other hand, Song and Ozkan [45] reduced the catalyst with a H /He
m(CO) vibrational mode of acetyl species [43], respec-
mixture at 673 K, which likely only facilitates surface shell reduc-
tion. Thus, the amount of oxygen on the support may be higher in
reference [45], owing to the diffusion of subsurface O to the sur-
ꢀ1
ethoxy species via acetaldehyde. The band at 1892 cm corre-
1049
1087
1
431
1
406
1
533
1
342
573 K
2925
1265
2969
1444
2862
1892
1
624
313 K
373 K
1
356
1049
1095
1
444
673 K
1265
1624
1356
1433
1038
1
713
1097
473 K
773 K
1
650 1356
3
000 2500 2000 1500 1000 500
3
000 2500 2000 1500 1000 500
-1
-1
Wavenumber (cm )
Wavenumber (cm )
Fig. 3. DRIFTS spectra of adsorbed ethanol obtained on Co/CeO
2
catalyst at different temperatures.

S.M. de Lima et al. / Journal of Catalysis 268 (2009) 268–281
273
face, and thereby favoring the oxidation of ethoxy species to
acetate.
resulting from the dissociative adsorption of ethanol. Even at
low temperature, some acetate species are clearly present
(m (OCO) = 1425 cm ; mas(OCO) = 1556 cm ), indicating that in
s
ꢀ1
ꢀ1
3
.2.3. DRIFTS analysis under reaction conditions
Figs. 4–6 display DRIFTS results of the impact of temperature on
the presence of water acetate formation is favored. When the
temperature was steadily increased to 373, 473, and 573 K, the
intensities of ethoxy bands clearly decreased, whereas those of
SR, POX, and OSR, respectively. It is noteworthy that there are no
DRIFTS experiments reported in the literature carried out under
reaction conditions over supported cobalt catalysts. We have re-
cently demonstrated that IR experiments under reaction condi-
tions are fundamental to obtain reliable mechanistic information
that more closely reflects the reaction conditions of the true cata-
lytic test [16].
ꢀ
1
ꢀ1
acetate species (1552–1558 cm
increased.
and 1425–1433 cm
)
At 573 K, the different branches of the gas-phase CO
2
band
ꢀ1
(ꢁ2350 cm ) are detected, and the intensity is higher at 673
and 773 K. The intensity of acetate bands significantly decreased
in moving from 573 to 673 K. At 673 K, new bands in addition to
those related to acetate species, likely corresponding to carbonate
3.2.3.1. DRIFTS analysis under ethanol + water mixture. In Fig. 4,
complexes, are detected in the m(OCO) stretching range; the cover-
which covers SR, at low temperature (e.g., 313 K), the coverage
of partially reduced ceria is dominated by ethoxy species (main
age of all species is quite low at 773 K.
Comparing these results with the spectra obtained during etha-
nol desorption (Fig. 3) significant differences in the relative band
ꢀ1
ꢀ1
bands:
m(CO) = 1057, 1101 cm
and m(CH) = 2881, 2974 cm )
1
433
1
552
1
101
1
556
1
057
1383
2974
8
83
3
1334
2881
13 K
1
026
1
313
573 K
1
558
425
1427
1558
1
053
1
1
101
883
1
330
373 K
2362 2320
673 K
1
307
1
425
1
558
1
101
1
047
473 K
773 K
1330
3
000 2500 2000 1500 1000 500
3000 2500 2000 1500 1000 500
-1
-1
Wavenumber (cm )
Wavenumber (cm )
Fig. 4. DRIFTS spectra obtained on Co/CeO
2
catalyst at different temperatures and under the reaction mixture containing ethanol and water (H
2
O/ethanol ratio = 2.0).
1
097
1
547
1
056
1434
1
1406
1558
2981
1309
883
2906
049
313 K
2
364 2320
573 K
1
547 1444
1
547
1
406
240
1
1328
373 K
673 K
473 K
773 K
3
000 2500 2000 1500 1000 500
3000 2500 2000 1500 1000 500
-1
-1
Wavenumber (cm )
Wavenumber (cm )
2
Fig. 5. DRIFTS spectra obtained on Co/CeO catalyst at different temperatures and under the reaction mixture containing ethanol and oxygen (oxygen/ethanol ratio = 0.5).

2
74
S.M. de Lima et al. / Journal of Catalysis 268 (2009) 268–281
1054
1432
1
558
1097
1
407
2976
2898
1558
8
1323
83
313 K
1
049
2360
2326
573 K
1
510
1433
1547
1339
373 K
673 K
1423
1533
473 K
773 K
3
000 2500 2000 1500 1000 500
3000 2500 2000 1500 1000 500
-1
-1
Wavenumber (cm )
Wavenumber (cm )
2 2
Fig. 6. DRIFTS spectra obtained on Co/CeO catalyst at different temperatures and under the reaction mixture containing ethanol, water, and oxygen (H O/ethanol ratio = 2.0;
oxygen/ethanol ratio = 0.5).
intensities are observed. Adding water to the feed promoted the
formation of acetate species as well as the acetate decomposition
reaction, in agreement with our previous study [16].
perature (473 K) than previously observed with SR (573 K), but
essentially at the same temperature observed with POX. At
673 K, bands corresponding to both acetate and carbonate are ob-
served, and the coverages of these species are apparently higher
than those observed with SR but lower than the ones found with
POX at the same temperature. There appears to be a higher surface
concentration of acetate species relative to carbonate during OSR
than observed under POX conditions, since weak bands at around
3.2.3.2. DRIFTS analysis under an ethanol + oxygen mixture. A similar
pattern emerges with POX (Fig. 5). At low temperature (313 K),
ꢀ1
there are ethoxy species (main bands:
m
(CO) = 1056, 1097 cm
while acetate is now present
mas(OCO) = 1558 cm ). Increasing temperature to the 373–573 K
ꢀ1
and
m(CH) = 2906, 2981 cm )
ꢀ1
ꢀ1
2
850 cm corresponding to the main
modes are still observed in the DRIFTS spectrum during OSR. At
73 K, it seems that mainly carbonate is present, as (OCO) bands
m(CH) vibrational stretching
(
range led to a decrease in the coverage of ethoxy species and a sig-
nificant increase in the coverage of acetate species; moreover, the
acetate bands appear to be of higher intensity relative to the case
7
m
ꢀ1
are detected in the broad range of 1000–1700 cm . To our knowl-
edge, these are the first DRIFTS spectra under OSR reaction condi-
tions reported in the literature for supported cobalt catalysts.
The DRIFTS results are better explained in the context of our
of SR (i.e., compare Fig. 5 with Fig. 4). CO
temperature (473 K) than previously observed with SR (573 K). At
73 K, both acetate and carbonate bands are observed, and their
coverages are higher than observed during SR at the same temper-
ature. The results suggest that H O plays an important role in the
2
can be detected at lower
6
earlier proposed mechanism for SR of ethanol over Pt/CeO
2
and
Pt/CeO –ZrO catalysts [16,25].
2
2
2
In the proposed mechanism (Scheme 1), which includes a junc-
tion effect with the support, ethoxy species are generated on the
surface of partially reduced ceria by dissociative adsorption. Oxida-
tive dehydrogenation via support –OH groups generates the ace-
tate intermediate, liberating H in the process at the support-
2
metal interface [16]. In POX and OSR, the dehydrogenated species
may react with oxygen from the support to produce the acetate
decomposition of acetate species, and one can speculate based on
the reactivity data that both the decomposition rate and selectivity
to acetate are influenced by the presence/absence of H
2
O. At 773 K,
it appears that some carbonate is present, as (OCO) bands are de-
m
.
ꢀ
1
tected in the broad range of 1000–1700 cm
We have recently reported DRIFTS results under POX conditions
over Pt/CeZrO catalyst. The addition of oxygen was suggested to
2
species. In the presence of coadsorbed H
sition reaction is favored, producing CO
2
O, the acetate decompo-
and CH species.
promote the oxidation of ethoxy species to acetate species via reac-
tion with support oxygen adatoms, as vacancies of the support
could be continuously replenished by oxygen from the feed. In
addition, the bands corresponding to carbonate species were more
intense when the ethanol–oxygen mixture was used. The current
results are in accordance with our previous findings [16].
x
x
3
3
.3. Reactions
.3.1. ED, SR, POX, and OSR at 773 K over Co/CeO
The performance of the Co/CeO catalyst for ED, SR, POX, and
2
2
3
.2.3.3. DRIFTS analysis under ethanol + water + oxygen mix-
OSR reactions at 773 K was evaluated and the results are displayed
in Fig. 7a–d. Initial ethanol conversion for ED was approximately
65% but significantly increased to 100% when water (SR), oxygen
(POX), or both (OSR) were added to the feed. Kugai et al. [14] also
ture. Similar trends emerge with OSR (Fig. 6). At low temperature
ꢀ1
(
313 K), ethoxy species (main bands:
m
(CO) = 1054, 1097 cm and
ꢀ
1
m(CH) = 2898, 2976 cm ) dominate the coverage while some ace-
ꢀ1
tate is already present [
m
as(OCO) = 1558 cm ]. Increasing temper-
2
compared the performance of Ni–Rh/CeO catalyst on a series of
ature to the 373–573 K range led to decreases in the intensities of
bands corresponding to ethoxy species and increases in the band
intensities of surface acetate. In addition, the spectra reflect a com-
posite between the DRIFTS spectra recorded under SR conditions
hydrogen production reactions: ED, SR, POX, and OSR. The addition
of water, oxygen, or both to the ethanol feed also increased ethanol
conversion, and this finding was attributed to a higher propensity
for C–C bond cleavage in the presence of water and/or oxygen. Cai
et al. [22] carried out a comparative study of SR, POX, and OSR
2
and those obtained during POX. CO can be detected at lower tem-

S.M. de Lima et al. / Journal of Catalysis 268 (2009) 268–281
275
Scheme 1. Reaction mechanism proposed for the ethanol conversion reactions.
reactions over Ir/CeO
2
. Ethanol conversion rates for POX and OSR
2
the H O/ethanol molar ratio in the feed increased during SR of eth-
were higher than that for SR, indicating that oxygen promoted
the surface reaction of ethanol. They proposed that the lattice oxy-
gen atoms of ceria should assist in activating the ethanol molecule.
We previously reported that initial ethanol conversion rates for SR,
anol over alumina-supported Pt catalysts. Ethene is produced by
ethanol dehydration over the support as described by reaction
(6) and, as such, addition of water inhibits dehydration of ethanol.
C
2
H
5
OH ! C
2
H
4
þ H
2
O
ð6Þ
2
POX, and OSR reactions over Pt/CeZrO were higher than those de-
tected for the ED reaction [16].
The effect of the H
2
O/ethanol molar ratio during SR at 773 K was
Ethanol conversion significantly decreased at the beginning of
the SR reaction tests while oxygen addition to the feed significantly
improved catalyst stability. Although a different catalyst system
was used, Fierro et al. [15] reported that the introduction of a high-
also evaluated. Ethanol conversion and product distributions as a
function of TOS obtained for Co/CeO catalyst during SR with a
O/ethanol molar ratio of 10.0 are shown in Fig. 8. Increasing
the H O/ethanol molar ratio from 3.0 (Fig. 7b) to 10.0 (Fig. 8) signif-
icantly promoted long-term catalyst stability. The final ethanol con-
version for SR under a H O/ethanol molar ratio of 10.0 remained
around 100% during 50 h TOS and it was higher than when the
O/ethanol molar ratio was 3.0 (52%). Increasing the H O/ethanol
2
H
2
2
er amount of oxygen increased the stability of Ni–Cu/SiO
2
catalysts
during OSR.
2
The selectivity to hydrogen decreased at the beginning of the ED
and POX reaction tests, while it remained constant for the SR and
H
2
2
OSR runs. After the first 2 h of TOS, H
2
production remained
molar ratio from 3.0 to 10.0 resulted in a decrease in the selectivity
to methane and CO while acetaldehyde formation was no longer
observed.
Recently, we demonstrated that the support plays an important
role in the SR reaction mechanism [10]. In order to decouple the
role of the support from the metal, the SR reaction was carried
out over the ceria support alone. Fig. 9 shows the ethanol conver-
approximately constant for all the different reactions tested. In
addition, hydrogen selectivity increased in the following order:
POX < OSR < SR, indicating that oxygen addition likely leads to
the production of water. The selectivities to CO and CH
atively constant during SR, POX, and OSR tests. However, the pro-
duction of CO and CH was negligible after 1 h TOS for ED,
indicating that the ED reaction to CO and CH was no longer ob-
served. The selectivity to CO increased as oxygen was introduced
4
were rel-
4
2
sion and product distributions obtained for the CeO support dur-
4
ing SR at 773 K, using a H O/ethanol molar ratio of 3.0. Ethanol
2
2
in the feed and was higher in the case of POX. This result indicates
that the ethanol combustion reaction occurs when oxygen is added
to the feed.
conversion slightly decreased during 6 h of TOS (from 55% to
50%). This is precisely the same final ethanol conversion achieved
by the Co/CeO
The similarity between the final ethanol conversion of Co/CeO
and unpromoted CeO
2
catalyst after the deactivation period (Fig. 7b).
The selectivity to acetaldehyde slightly increased during the
first 3 h of TOS and then leveled off at approximately 5% in the
cases of SR and OSR. On the other hand, a higher amount of acetal-
dehyde was formed during ED and POX. The latter result regarding
POX agrees well with the findings of Kugai et al. [14], who pro-
posed that oxygen takes part in ethanol dehydrogenation to acet-
aldehyde rather than C–C bond rupture. We also previously
reported that the introduction of oxygen in the feed favors acetal-
dehyde production during ethanol reforming reactions over Pt/
2
2
suggests that the deactivation is likely due
to the loss of activity of the metallic phase.
2
Regarding the product distributions, hydrogen, CO , and ethene
2
were the main products obtained over unpromoted CeO . Small
quantities of acetaldehyde (ꢃ5%) and trace amounts of methane
were also observed. In addition, the production of CO during test-
ing of the unpromoted CeO
2
catalyst was very low (<200 ppm).
CeZrO
2
[16]. Significant amounts of ethene were only detected dur-
3.3.2. Effect of the reaction temperature on SR over Co/CeO
2
ing ED and its selectivity continuously increased during 6 h of TOS.
Dömök et al. [46] reported a decrease in ethene production when
SR was also carried out at high temperature (1073 K) and the re-
sults are presented in Fig. 10. The catalyst exhibited the same ini-

2
76
S.M. de Lima et al. / Journal of Catalysis 268 (2009) 268–281
1
00
X ethanol
H2
CO2
1
00
(b)
(a)
X ethanol
H2
acetaldehyde
CO
CO2
ethene
CH4
8
6
0
0
ethene
CH4
acetaldehyde
8
6
4
2
0
0
0
0
0
1
1
5
0
5
0
0
10
20
30
40
50
0
1
2
3
4
5
6
Time (h)
Time (h)
1
00
100
80
(
c)
X ethanol
H2
(d)
CO2
X ethanol
H2
8
6
4
2
0
0
0
0
0
ethene
CH4
acetaldehyde
CO
CO2
ethene
CH4
acetaldehyde
CO
60
3
2
2
1
1
0
5
0
5
0
5
0
0
5
10
15
20
25
30
0
5
10 15 20 25 30 35 40 45 50
Time (h)
Time (h)
Fig. 7. Ethanol conversion (Xethanol) and product distributions versus TOS over Co/CeO
2
catalyst obtained during (a) ED; (b) SR under H
2
O/ethanol molar ratio = 3.0; (c) POX
under O
2
/ethanol = 0.5; and (d) OSR under H
2
O/ethanol molar ratio = 3.0 and O /ethanol molar ratio = 0.5 (mass of catalyst = 20 mg; Treaction = 773 K and residence
2
time = 0.02 g s/mL).
100
100
X ethanol
CO2
H2
Acetaldehyde
Ethene
90
80
70
60
CH4
8
0
0
X ethanol
H2
6
CO2
CO
CH4
50
25
20
15
10
5
0
40
30
20
10
0
0
5
10 15 20 25 30 35 40 45 50
Time (h)
0
1
2
3
4
5
6
Time (h)
Fig. 8. Ethanol conversion (Xethanol) and product distributions versus TOS over
Co/CeO catalyst obtained during SR under H O/ethanol molar ratio = 10.0 (mass of
catalyst = 20 mg; Treaction = 773 K and residence time = 0.02 g s/mL).
Fig. 9. Ethanol conversion (Xethanol) and product distributions versus TOS over
unpromoted CeO obtained during SR under H O/ethanol molar ratio = 3.0 (mass of
catalyst = 20 mg; Treaction = 773 K and residence time = 0.02 g s/mL).
2
2
2
2

S.M. de Lima et al. / Journal of Catalysis 268 (2009) 268–281
277
Low reaction temperatures favor the formation of carbon
through reactions (7) (Boudouard reaction) and (8) (reverse of
carbon gasification). However, carbon formation via reaction (9)
(methane decomposition) is the main route at high temperature.
The mechanism of coke formation over supported Ni catalysts
during steam reforming of methane is well described in the litera-
ture [1,47–49]. Methane dissociates on nickel surface, producing
highly reactive carbon species [47]. This carbon may undergo a
number of processes, including (1) reaction with water; (2) encap-
sulation of the Ni particle surface; or (3) dissolution in the Ni crys-
tallite followed by the nucleation and growth of carbon filaments
1
00
X ethanol
H2
CO
CO2
8
0
0
CH4
ethene
6
1
5
0
5
0
(
e.g., whiskers).
The studies concerning carbon formation over Ni and Co-based
1
catalysts during SR of ethanol are more scarce and the mechanism
is not well understood. Typically, two types of carbonaceous
deposits have been reported in the literature: amorphous carbon
and filamentous carbon [31–34]. The nature of carbon formed de-
pends on the reaction conditions.
We carried out TG and TEM analyses in order to study the effect
of reaction conditions on the nature of the carbon deposits formed.
This may provide useful guidance in determining the optimal oper-
ating conditions that minimize coke formation during ethanol con-
version reactions for hydrogen production.
0
5
10
15
Time (h)
20
25
30
Fig. 10. Ethanol conversion (Xethanol) and product distributions versus TOS over 10%
Co/CeO catalyst obtained during SR under H O/ethanol molar ratio = 3.0 (mass of
catalyst = 20 mg; Treaction = 1073 K and residence time = 0.02 g s/mL).
2
2
2
Fig. 11 shows the TPO profile of the Co/CeO catalyst after dif-
ferent reaction conditions. The TPO profile exhibited one peak in
the range of 721–730 K and a shoulder positioned at 753–760 K
tial ethanol conversion level at both temperatures (Fig. 7b: 773 K
and Fig. 10: 1073 K). However, increasing the reaction temperature
from 773 to 1073 K greatly improved the stability of the catalyst,
such that it remained quite stable at the higher temperature. Con-
cerning product distributions, H , CO, CO , and CH were the only
2 2 4
products detected at high temperature, whereas acetaldehyde
was also observed during the run at 773 K. When reaction temper-
after subjecting the catalyst to SR reaction under different H
2
O/
ethanol molar ratios. In addition, increasing the H O/ethanol molar
2
ratio from 3.0 to 10.0 decreased the intensity of the peak at low
temperature. After POX and OSR, one narrow peak and one broad
peak were observed at approximately 719 and 765 K, respectively.
The TPO profile did not exhibit weight loss following SR at 1073 K.
The amounts of carbon deposited on the Co/CeO catalyst after
2
exposure to different reaction conditions are listed in Table 1. The
amount of carbon deposited on the catalyst after SR at 773 K and
ature was increased, the selectivities to H
2 2
and CO decreased
while the selectivities to CO and CH increased. Galetti et al. [33]
4
carried out SR at different temperatures (673, 773, and 873 K) over
a CuCoZnAl catalyst. They also observed an improvement in cata-
lyst stability and increases in selectivities to CO and CH as the
4
2
H O/ethanol molar ratio of 3.0 was higher than the one observed
after POX and OSR at 773 K. Increasing the H O/ethanol molar ratio
2
reaction temperature was risen. Moreover, TG analysis revealed a
decrease in the amount of carbon formed upon increasing reaction
temperature, in good agreement with the activity test. They pro-
posed that the occurrence of the reverse of Boudouard reaction
from 3.0 to 10.0 resulted in a decrease in the amount of carbon
deposited on the catalyst surface following SR at 773 K. Carbon
deposition was no longer observed during SR at high temperature.
These two different oxidation regions can be due to the pres-
ence of two different types of carbon on the surface of the catalyst.
(
Eq. 7) at 873 K improved the resistance of the catalyst to carbon
deposition. In another study, the deposition of carbon over Co/
CeO was not detected by either TG or TEM analysis when SR
2
was carried out above 873 K [34]. According to Wang et al. [34],
the absence of carbon deposition was attributed to the occurrence
2
of steam and CO reforming reactions.
7
65
OSR
Catalyst stability is probably the most important issue during
ethanol conversion reactions to produce hydrogen. Most authors
in the open literature report that carbon formation is the main cause
of catalyst deactivation, and that the deleterious impact of carbon
deposits on catalyst stability depends on both the reaction condi-
tions and the catalyst used. In the next section, the main drivers of
catalyst deactivation observed in this work will be discussed in
greater detail for all the ethanol conversion reactions. To our knowl-
edge, studies comparing the modes of deactivation during SR, POX,
and OSR over Co-based catalysts have not yet been reported.
730
760
SR-10:1
721
7
53
SR-3:1
POX
3.4. Catalyst deactivation
7
19
There are three main reactions that may contribute to carbon
formation [46–48].
CO ! C þ CO
CO þ H ! H O þ C
CH þ C
! 2H
4
00
600
800
1000
2
2
ð7Þ
ð8Þ
ð9Þ
Temperature (K)
2
2
2
Fig. 11. TPO profile of 10% Co/CeO catalyst obtained after subjecting the catalyst to
different reaction conditions.
4
2

2
78
S.M. de Lima et al. / Journal of Catalysis 268 (2009) 268–281
Table 1
Amount of carbon deposited on the catalyst surface as determined by TGA after
Using the TPO peak assignments from the literature as a basis,
our TPO profile indicates the presence of amorphous carbon and
exposing the catalyst to different reaction conditions. During TGA, 5%O
and a heating rate of 10 K/min were used.
2
:He mixture
2
filamentous carbon over the Co/CeO catalyst after running SR un-
der H O/ethanol ratios of 3.0 and 10.0 and following exposure to
2
Reaction condition
mg carbon/g
catalyst/h
OSR reaction. In addition, the fraction of amorphous carbon ap-
pears to be higher in the cases of SR as compared to OSR.
Electron microscopy techniques have been used to characterize
the nature of the carbonaceous species formed over Co-based cat-
SR at 773 K under H
SR at 773 K under H
2
O/ethanol molar ratio = 3.0
O/ethanol molar ratio = 10.0
4.3
2.3
1.8
1.5
2
POX at 773 K under O
OSR at 773 K under H
and O
SR at 1073 K under H
2
/ethanol = 0.5
2
O/ethanol molar ratio = 3.0
2
/ethanol molar ratio = 0.5
O/ethanol molar ratio = 3.0
2
0.0
The group of Prof. Resasco regularly uses TPO as an important tool
to characterize the types of carbonaceous species present on Co-
based catalysts used in the production of carbon nanotubes [50].
The oxidation peak located below 673 K is ascribed to amorphous
carbon, whereas the peak above 773 K corresponds to carbon
nanotubes. The oxidation of graphite occurs at higher tempera-
tures (ꢁ973 K).
There are few studies available in the open literature that made
use of TPO after SR of ethanol to study the nature of carbonaceous
deposits and to quantify the amount of carbon formed over sup-
ported Co catalysts [17,26,31,34]. Wang et al. [34] performed TG
analyses after SR of ethanol at different reaction temperatures over
2
Co/CeO . The TPO profile after SR at 773 K displayed two peaks at
around 698 and 848 K, while no peaks were detected during TGA
following SR at 873–973 K. However, the authors did not correlate
these peaks to the types of carbonaceous species formed. Two
peaks at 573 and 773 K were also present in the TPO profile of
Co/CeZrO
attributed to carbonaceous deposits over the surfaces of the parti-
cles and to carbon filaments. The TPO profile of CoRh/SiO catalyst
2
catalyst after SR of ethanol at 813 K [31]. They were
2
following OSR at 673 K exhibited two peaks below 673 K but they
were not assigned to any particular carbon species [17]. No TGA or
TPO results following exposure of a Co-based catalyst to the POX
reaction were found in the literature.
Fig. 13. HR-TEM image of the Co/CeO
under an O /He mixture. The inset corresponds to selected area electron diffraction
patterns recorded on cobalt particle.
2
catalyst reduced at 1023 K and passivated
2
2 2
Fig. 12. STEM and HR-TEM images of the Co/CeO catalyst reduced at 1023 K and passivated under an O /He mixture.

S.M. de Lima et al. / Journal of Catalysis 268 (2009) 268–281
279
alysts during SR. HR-TEM images of Co/CeO
2
catalyst following SR
vere deactivation. On the other hand, at 773 and 823 K, carbon fil-
aments were detected but the catalyst did not deactivate. At high
reaction temperature (973 K), carbon deposits were not observed
and the catalyst remained stable.
of ethanol at 723 K displayed Co particles covered by amorphous
carbon with a minor amount extending over the support [28].
TEM analyses revealed the presence of two types of carbon depos-
its on Co/CeZrO
2
catalyst after SR at 813 K: a carbonaceous layer
In our work, TEM images were carried out after different reac-
tion conditions to shed light on the mechanism of carbon forma-
residing on the surface of the metallic particle, and carbon fila-
ments. Galetti et al. [33] observed that the morphologies of the
spent catalysts varied significantly as a function of reaction tem-
perature. Carbon filaments were detected by SEM on the surface
of CuCoZnAl catalyst after SR of ethanol at 773 K, but were absent
in images taken after reaction at 873 K. Wang et al. [34] also con-
tion over Co/CeO
Co/CeO catalyst after exposure to different reaction conditions.
Fig. 12 displays the STEM image of the Co/CeO catalyst reduced
in hydrogen and passivated under an O /He mixture. After reduc-
2
. Figs. 12–16 display TEM micrographs of the
2
2
2
tion, cobalt particles are formed in a size range from 3 nm up to
20 nm with the majority of particles ranging between 5 and
10 nm in size. The largest Co particles can be identified in HR-
TEM to actually be a composite of several smaller Co crystals.
The Co catalyst particles have a hexagonal or pseudo-hexagonal
morphology as shown by the HR-TEM inset in Fig. 12. A represen-
cluded that the nature of carbon deposits during SR over Co/CeO
2
catalyst is dependent on the reaction temperature. TEM images re-
vealed the presence of carbon encapsulating the cobalt particle
when the reaction was performed at 623 and 723 K, resulting in se-
Fig. 14. HR-TEM images of the Co/CeO
2
following SR under H
2
O/ethanol molar
2 2
Fig. 15. HR-TEM images of the Co/CeO following OSR under H O/ethanol molar
ratio = 3.0 at 773 K.
ratio = 3.0 and O /ethanol molar ratio = 0.5 at 773 K.
2

2
80
S.M. de Lima et al. / Journal of Catalysis 268 (2009) 268–281
2
The images of Co/CeO catalyst following exposure to SR under
H O/ethanol ratios of 3.0 and OSR at 773 K are shown in Figs. 14
2
and 15 and illustrate the presence of carbon deposits, which are al-
ways in proximity to the Co particles.
Cobalt catalyst nanoparticles in the spent catalyst are well-crys-
tallized (hexagonal and pseudo-hexagonal crystals), ranging from
ꢁ
3 to 30 nm in size and are located on the ceria agglomerates.
The larger Co particles appear to have formed from close spatial
arrangements of several <5 nm particles. Most importantly, Co
nanoparticles occur in either of two spatial arrangements: (1) dis-
persed on ceria surfaces and in this case, lesser amounts of carbon
deposits formed nanocoatings on the Co and CeO
alternately, a high fraction of Co nanoparticles occur separated
from the CeO host surface (Co must have been lifted off the sup-
2
exterior; or (2)
2
port) and in this case, those Co nanoparticles initiated the forma-
tion of carbon tubular nanostructures shown in Fig. 14 (Co
particles are located at the ends of carbon tubules). The carbon
tubular morphology entails multiple outer walls which are not par-
allel to the inner channel, but rather in a configuration best de-
scribed as a herringbone structure with numerous capped areas
that disrupt or cap the inner channel. Thus the carbon nanostruc-
tures are not carbon nanotubes, but resemble more likely carbon
nanofibers or filaments.
The relative sizes (i.e., diameters) of the carbon nanostructures
are clearly controlled by the Co nanoparticle size. In some in-
stances, two or more Co nanoparticles form a larger cluster that
initiates the growth of rather large carbon nanotubular structures
with diameters of up to 35 nm. Several carbon nanotubular struc-
tures also engulfed ultra small (typically <3 nm) Co nano-catalyst
particles as shown in Fig. 14b. Therefore, some of the Co is confined
to the internal structure of the carbon nanotubular host. In addi-
tion, several carbon filaments are characterized by an open end
where a Co nanoparticle became detached as illustrated in
Fig. 15a and b. The morphology of the carbon tubular nanostruc-
tures or filaments was observed to be similar in both samples,
but the frequency and location of the carbon deposits vary signifi-
cantly. Carbon nanostructures were not detected over the Co/CeO
2
catalyst following SR at 1073 K (Fig. 16). In this case, some regions
revealed only the presence of amorphous carbon in proximity to
the ceria agglomerates. Fig. 16a shows a large ceria agglomerate
and under high magnification (Fig. 16b) one can observe well-crys-
tallized cobalt catalyst particles on the ceria surface. The cobalt
catalyst particles clustered to form an approximately 20 nm grain
which is overgrown by a very thin (ꢁ2–3 nm) layer of amorphous
carbon nanocoating.
According to the proposed reaction mechanism based on
DRIFTS experiments over Co/CeO catalyst under different reaction
2
conditions, the decomposition of dehydrogenated species (e.g.,
acetaldehyde, acetyl) and acetate species produce hydrogen, CO,
and CH species, which may in turn result in carbon formation
x
2 2
Fig. 16. HR-TEM images of the Co/CeO following SR under H O/ethanol molar
ratio = 3.0 at 1073 K.
depending on the rate of hydrogen recombination. We have re-
cently demonstrated that the presence of the metal promotes the
tative diffraction pattern for the cobalt nanoparticles was obtained
during the TEM investigation, displaying characteristic rings for a
Co phase (Fig. 13). The arrangement of concentric rings (diffrac-
decomposition of acetate species over Pt/CeZrO
Therefore, the catalyst should deactivate when the rate of this
reaction pathway is higher than the rate of desorption of CH spe-
cies as CH . The CH species formed may block the Pt-support
interface resulting in catalyst deactivation. The CH species may
2
catalyst [16].
0
tion pattern was obtained from several randomly oriented cobalt
x
particles) corresponds to the presence of the Co (111), (200),
4
x
0
(
220), and (310) planes assigned to Co confirming our conclusion
x
that reduction indeed leads to the formation of cobalt metal nano-
particles. In addition, the EELS spectra of the Co nanoparticle
be further dehydrogenated to H and C. In the case of Co-based cat-
alysts, the formed carbon will either (a) diffuse behind the metal to
showed the L
and L characteristic of Co (L
). Furthermore, the oxygen K-edge is missing in the spectra at
30 eV.
Ceria nanoparticles are typically agglomerated in spent cata-
lysts. The individual ceria grain sizes are on the order of 4–10 nm
while ceria agglomerates can be observed in excess of 0.5 m.
2
/L
3
edges with a typical intensity difference for L
2
assist in carbon filament growth or (b) react with O
2 2
(or H O) to
0
3
3
is significantly taller compared with
produce CO species (Scheme 1). The addition of an excess of water
X
L
5
2
or oxygen to the feed helps to clean the surface of the metal, keep-
ing it active for a longer period of time. Therefore, the metal can re-
main active despite having a considerable amount of carbon
deposited behind the particles, since the top surface will remain
exposed to reactants and gas-phase intermediates. According to
l

S.M. de Lima et al. / Journal of Catalysis 268 (2009) 268–281
281
Trimm [47], this is a special case where the coke formation does
References
not result in catalyst deactivation.
When SR is carried out at high temperature (1073 K), carbon
does not accumulate, and this may be due to: (1) the steam and
[
[
[
1] P.R. de la Piscina, N. Homs, Chem. Soc. Rev. 37 (2008) 2459.
2] P.D. Vaidya, A.E. Rodrigues, Ind. Eng. Chem. Res. 45 (2006) 6614.
3] A. Haryanto, S. Fernando, N. Murali, S. Adhikari, Energy Fuels 19 (2005) 2098.
CO
2
reforming of methane reactions; (2) the reverse of the dispro-
[4] H. Song, L. Zhang, R.B. Watson, D. Braden, U.S. Ozkan, Catal. Today 129 (2007)
46.
[
3
portion reaction (Eq. 7); and/or (3) the carbon gasification reaction,
which is favored at high temperatures.
5] R.M. Navarro, M.C. Alvarez-Galvan, M.C. Sanchez-Sanchez, F. Rosa, J.L.G. Fierro,
Appl. Catal. B 55 (2004) 223.
[
[
[
6] M. Veronica, B. Graciela, A. Norma, L. Miguel, Chem. Eng. J. 138 (2008) 602.
7] A.N. Fatsikostas, X.E. Verykios, J. Catal. 225 (2004) 439.
8] A. Erdohelyi, J. Raskó, T. Kecskés, M. Tóth, M. Dömök, K. Baán, Catal. Today 116
4
. Conclusions
(
2006) 367.
[
9] V. Fierro, V. Klouz, O. Akdim, C. Mirodatos, Catal. Today 75 (2002) 141.
[
[
[
10] L.V. Mattos, F.B. Noronha, J. Power Sources 145 (2005) 10.
11] L.V. Mattos, F.B. Noronha, J. Power Sources 152 (2005) 50.
12] L.V. Mattos, F.B. Noronha, J. Catal. 233 (2005) 453.
DRIFTS, reaction testing, and TPD measurements provided the
following conclusions regarding the reaction network. With ED,
ethanol adsorbs dissociatively at low temperature to form ad-
sorbed ethoxy species and a bridging –OH group. At moderate tem-
peratures, ethoxy species diminish and acetaldehyde increases. A
[13] J.L. Bi, S.N. Hsu, C.T. Yeh, C.B. Wang, Catal. Today 129 (2007) 330.
[
[
[
14] J. Kugai, S. Velu, C. Song, Catal. Lett. 101 (2005) 255.
15] V. Fierro, O. Akidim, C. Mirodatos, Green Chem. 5 (2003) 20.
16] S.M. de Lima, I.O. da Cruz, G. Jacobs, B.H. Davis, L.V. Mattos, F.B. Noronha, J.
Catal. 257 (2008) 356.
2 4
fraction of the ethanol decomposes to H , CH , and CO, and another
fraction dehydrogenates to acetaldehyde. At higher temperatures,
acetaldehyde is converted to acetate via support bound –OH
groups. In the absence of steam, the dehydration product of ethyl-
ene is favored.
With SR, OSR, and POX, the increased coverage of –OH and lat-
tice O adatoms promotes the formation of acetate at lower temper-
atures. At higher temperatures, steam promotes the forward
[17] E.B. Pereira, N. Homs, S. Marti, J.L.G. Fierro, P.R. de la Piscina, J. Catal. 257
2008) 206.
(
[
[
18] N. Laosiripojana, S. Assabumrungrat, Appl. Catal. B 66 (2006) 29.
19] T. Nishiguchi, T. Matsumoto, H. Kanai, K. Utani, Y. Matsumurab, W-J. Shenc, S.
Imamura, Appl. Catal. A 279 (2005) 273.
[
[
[
20] J. Llorca, N. Homs, P.R. de la Piscina, J. Catal. 227 (2004) 556.
21] J. Llorca, P.R. de la Piscina, J. Sales, N. Homs, Chem. Commun. (2001) 641.
22] W. Cai, F. Wanga, E. Zhan, A.C. Van Veen, C. Mirodatos, W. Shen, J. Catal. 257
(2008) 96.
[
[
23] J. Kugai, V. Subramani, C. Song, M.H. Engelhard, Y.H. Chin, J. Catal. 238 (2006)
decomposition of acetate to carbonate and CH
decomposes over Co metal particles to evolve H
Replacing steam with oxygen resulted in losses in H
4
, the latter of which
and carbon.
selectivity
430.
2
24] H. Roh, A. Platon, Y. Wang, D.L. King, Catal. Lett. 110 (2006) 1.
2
[25] G. Jacobs, R.A. Keogh, B.H. Davis, J. Catal. 245 (2007) 326.
[
26] F. Romero-Sarria, J.C. Vargas, A. Roger, A. Kiennemann, Catal. Today 133 (2008)
49.
27] L.P.R. Profeti, E.A. Ticianelli, E.M. Assaf, J. Power Sources 175 (2008) 482.
as the forward decomposition was hindered; thus, higher acetalde-
hyde production was observed. In OSR, although acetaldehyde pro-
1
[
duction was inhibited by the presence of H
2
O in the feed, the H
2
[28] J. Llorca, N. Homs, J. Sales, P.R. de la Piscina, J. Catal. 209 (2002) 306.
[
[
[
29] B. Zhang, X. Tang, Y. Li, W. Cai, Y. Xu, W. Shen, Catal. Commun. 7 (2006) 367.
30] H. Wang, J.L. Ye, Y. Liu, Y.D. Lin, Y.N. Qin, Catal. Today 129 (2007) 305.
31] J.C. Vargas, S. Libs, A.C. Roger, A. Kiennemann, Catal. Today 107 (2005) 417.
selectivity was lower relative to the case of SR alone.
Regarding catalyst stability, the reaction type, feed ratios, and
temperature strongly influence the nature of the carbon deposits
that form and their impact on catalyst stability. The results suggest
that the carbon diffuses behind the particle, nucleating the growth
of carbon filaments. The filamentous carbon lifts the Co particles
from the support, but this does not directly lead to catalyst deacti-
[32] J.M. Guil, N. Homs, J. Llorca, P.R. de la Piscina, J. Phys. Chem. B 109 (2005)
10813.
[
33] A.E. Galetti, M.F. Gomez, L.A. Arrua, A.J. Marchi, M.C. Abello, Catal. Commun. 9
2008) 1201.
34] H. Wang, Y. Liu, L. Wang, Y.N. Qin, Chem. Eng. J. 145 (2008) 25.
(
[
[35] S. Cavallaro, V. Chiodo, A. Vita, S. Freni, J. Power Sources 123 (2003) 10.
[
[
36] P. Gajardo, P. Grange, B. Dalmon, J. Phys. Chem. 83 (1979) 1771.
37] F.B. Noronha, C.A. Perez, R. Frety, M. Schmal, Phys. Chem. Chem. Phys. 1 (1999)
2
vation. In cases where high H O/ethanol feed ratios are employed,
or if oxygen is included in the feed, enough of the active surfaces of
both the support and metal remains exposed and accessible to
reactants and intermediates, such that the catalyst deactivation
2861.
[38] G. Jacobs, E. Chenu, P.M. Patterson, L. Calico-Williams, D.E. Sparks, G. Thomas,
B.H. Davis, Appl. Catal. A 258 (2004) 203.
[
[
39] E.M. Cordi, J.L. Falconer, J. Catal. 162 (1996) 104.
40] L.F. de Mello, F.B. Noronha, M. Schmal, J. Catal. 220 (2003) 358.
2
rate is alleviated. On the other hand, when less of H O or no oxygen
was present, amorphous carbon tended to cover and block active
sites, leading to more significant losses in activity.
[41] A. Yee, S.J. Morrison, H. Idriss, J. Catal. 191 (2000) 30.
[
[
[
[
[
[
[
42] J. Raskó, A. Hancz, A. Erdohelyi, Appl. Catal. A 269 (2004) 13.
43] M. Mavrikakis, M.A. Barteau, J. Mol. Catal. A: Chem. 131 (1998) 135.
44] P.V. Menacherry, G.L. Haller, J. Catal. 177 (1998) 175.
45] H. Song, U.S. Ozkan, J. Catal. 261 (2009) 66.
46] M. Dömök, M. Tóth, J. Raskó, A. Erdohelyi, Appl. Catal. B 69 (2007) 262.
47] D.L. Trimm, Catal. Today 37 (1997) 233.
Acknowledgments
48] D.L. Trimm, Catal. Today 49 (1999) 49.
This work received financial support of CTENERG/FINEP-
[49] R.M. Navarro, M.A. Pena, J.L.G. Fierro, Chem. Rev. 107 (2007) 3952.
[50] B. Kitiyanan, W.E. Alvarez, J.H. Harwell, D.E. Resasco, Chem. Phys. Lett. 317
0
1.04.0525.00. CAER acknowledges the Commonwealth of Ken-
(
2000) 497.
tucky for financial support.