Catalytic partial oxidation of ethanol over noble metal catalysts

Article abstract of DOI:10.1016/j.jcat.2005.07.021

The catalytic partial oxidation of ethanol and ethanol-water was investigated over noble metal and metal plus ceria-coated alumina foams at catalyst contact times <10 ms. The effects of catalyst, flow rate, and water addition on selectivity and conversion were examined. Rh-Ce catalysts were the most active and stable. Without water addition, ethanol was converted directly to H₂ with >80% selectivity and >95% conversion with Rh-Ce catalysts. Rh, Pt, Pd, and Rh-Ru produced less H₂, with Pt and Pd producing <50% H₂. Pt, Pd, and Rh also produced more CH ₄ and C₂H₄ than Rh-Ce. There was a smaller dependence on flow rate for Rh-Ce catalysts than other catalysts. Variation of a factor of 2 produced small changes in H₂, and lower flow rates produced less CH₄ and C₂H₄. Autothermal operation was achieved at as low as 10 mol% ethanol in water. Adding water to the Rh-Ce catalyzed reactor increased H₂ selectivity and reduced selectivity to CO to <50% due to increased water-gas shift and steam reforming activity. With added water, the selectivity to H₂ exceeds 100%, because both ethanol and water contribute H₂. Also, the total selectivity of all unwanted products, mostly CH₄, is <3% at the H₂ production maximum with water addition.

Full text of DOI:10.1016/j.jcat.2005.07.021

Journal of Catalysis 235 (2005) 69–78
www.elsevier.com/locate/jcat
Catalytic partial oxidation of ethanol over noble metal catalysts
∗
J.R. Salge, G.A. Deluga, L.D. Schmidt
Department of Chemical Engineering and Materials Science, University of Minnesota, 421 Washington Ave SE, Minneapolis, MN 55455, USA
Received 12 May 2005; revised 15 July 2005; accepted 20 July 2005
Available online 19 August 2005
Abstract
The catalytic partial oxidation of ethanol and ethanol-water was investigated over noble metal and metal plus ceria-coated alumina foams
at catalyst contact times <10 ms. The effects of catalyst, flow rate, and water addition on selectivity and conversion were examined. Rh–Ce
catalysts were the most active and stable. Without water addition, ethanol was converted directly to H with >80% selectivity and >95%
2
conversion with Rh–Ce catalysts. Rh, Pt, Pd, and Rh–Ru produced less H , with Pt and Pd producing <50% H . Pt, Pd, and Rh also produced
2
2
more CH and C H than Rh–Ce. There was a smaller dependence on flow rate for Rh–Ce catalysts than other catalysts. Variation of a factor
4
2 4
of 2 produced small changes in H , and lower flow rates produced less CH and C H . Autothermal operation was achieved at as low as
2
4
2 4
1
0 mol% ethanol in water. Adding water to the Rh–Ce catalyzed reactor increased H selectivity and reduced selectivity to CO to <50% due
2
to increased water–gas shift and steam reforming activity. With added water, the selectivity to H exceeds 100%, because both ethanol and
2
water contribute H . Also, the total selectivity of all unwanted products, mostly CH , is <3% at the H production maximum with water
2
4
2
addition.

2005 Elsevier Inc. All rights reserved.
1
. Introduction
Therefore, heat must be added to heat the reactor to the
◦
8
00 C needed to achieve high conversions, and this requires
Research in our laboratory over the past decade has fo-
residence times of ∼1 s. Because external heat is required,
reactor designs are typically limited by heat transfer rather
than by reaction kinetics. There are ways to increase heat
transfer while steam reforming fuels, such as in microchan-
nel reactor systems [7] or catalytic wall reactors, where both
endothermic and exothermic reactions take place on high-
surface area catalysts [2]. However, the use of multiple fuels
and expensive materials makes steam reforming unsuitable
for transportation applications.
cused on understanding partial oxidation reactions over no-
ble metal catalysts at millisecond contact times. These par-
tial oxidation systems can be run autothermally, thus elim-
inating the need for external heat. They can also be easily
scaled up or down for different applications, enabling the
use of a compact reactor suitable for stationary and portable
applications. These autothermal reactor systems are also ad-
vantageous because they can be rapidly started up and can
respond to process fluctuations at response times <5 s [1].
There have been numerous studies on hydrogen produc-
tion by steam reforming of ethanol [2–6]. However, a major
drawback to this approach is the fact that steam reforming is
strongly endothermic,
There has been discussion of the desirability of autother-
mal reforming of ethanol [8], and the catalytic partial oxida-
tion and autothermal reforming of ethanol have been active
areas of research [9–12]. Verykios et al. carried out their re-
◦
0
actions by preheating ethanol to ∼500 C over lanthanates,
C2H5OH + H2O → 2CO + 4H2, ꢀH = +256 kJ/mol.
Ru, and Ni [10,11]. They accomplished their reactions at res-
idence times ∼10 times longer than needed with millisecond
contact time reactors. Cavallaro et al. carried out their re-
(1)
*
Corresponding author. Fax: +1-612-626-7246.
E-mail addresses: salge@cems.umn.edu (J.R. Salge),
◦
actions by preheating ethanol to ∼650 C over Rh/Al2O3
schmi001@umn.edu (L.D. Schmidt).
powder [12]. Metal sintering was an issue in these experi-
0021-9517/$ – see front matter  2005 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2005.07.021

70
J.R. Salge et al. / Journal of Catalysis 235 (2005) 69–78
ments, which were carried out in a furnace and focused on
diluted ethanol solutions.
2. Experimental
We recently reported that ethanol can be converted di-
rectly to H2 with >80% selectivity and >95% conversion in
an autothermal reformer using Rh–Ce catalysts [9]. Here we
extend this research to examine the effects of catalyst and
flow rate on the autothermal reforming of ethanol at catalyst
contact times <10 ms. Noble metals (Rh, Pt, and Pd), noble
metal alloys (Rh–Ru), and the addition of ceria (Rh–Ce) are
examined. The effect of flow rate is also investigated.
Unlike hydrocarbons, the partial oxidation of ethanol is
slightly endothermic, so it alone will not produce the heat
necessary for autothermal operation:
2.1. Reactor
Ethanol–air mixtures are flammable over a wide compo-
sition range [15]. To avoid the formation of flames and to
help limit other homogeneous reactions before the catalyst,
we used an automotive fuel injector [16], which can deliver
and vaporize fuel rapidly by creating small droplets. Because
vaporization and mixing of the fuel with air occur almost
simultaneously, upstream regions containing a combustible
mixture are reduced or avoided.
The reactor was identical to that described previously and
consists of a quartz tube, 50-cm long with an 18-mm i.d. [9].
The fuel (ethanol or ethanol–water mixture), which is liquid
at room temperature, was introduced at the top of the reactor
using a fuel injector. Pressurized fuel at 20 psig fed the injec-
tor, controlled using LabVIEW at frequencies of 10–20 Hz
and at duty cycles (i.e., the percentage of time that the injec-
tor remains open) of 3–15%. The liquid flow rate delivered
by the injector was accurately controlled by the pressure in
the fuel supply tank and by the duty cycle. The fuel delivery
rate was calibrated at different frequencies and duty cycles
and was accurate to within ± 2%. In all experiments, the re-
actor ran at atmospheric pressure.
0
C2H5OH +(1/2)O2 → 2CO + 3H2, ꢀH = +14 kJ/mol.
(2)
Thus some total oxidation is needed to generate the heat for
◦
operation at the 700–1200 C necessary for sustained fast
reaction:
C2H5OH + 3O2 → 2CO2 + 3H2O,
0
ꢀ
H = −1277 kJ/mol.
(3)
Although the combustion reaction is highly exothermic, it
can also produce flames. Flames are intolerable because they
lead to unsteady operation and form coke and soot, which
deactivate the catalyst. Homogeneous reactions also form
unwanted products, such as acetaldehyde and ethylene [13].
To generate more hydrogen, the steam reforming and par-
tial oxidation reactions can be combined with the water–gas
shift (WGS) reaction:
The fuel injector accurately dispersed the fuel into a
2
1
5-cm-long preheated section of the reactor. Preheating of
40 ± 10 C was maintained above the catalyst using heat-
◦
ing tape wrapped around this preheat section. Oxygen and
nitrogen were mixed at air stoichiometry (N2/O2 = 3.76)
and then introduced at the beginning of the preheat section.
The flow rates of high purity oxygen and nitrogen entering
the system from high-pressure cylinders were adjusted using
mass flow controllers via LabVIEW; these controllers are ac-
curate to within ± 5% of their setpoint. Mixing of the air and
fuel was improved by adding a blank ceramic foam at the
end of the preheat section, about 4 cm above the catalyst.
The catalyst-coated foam had uncoated foams placed up-
stream and downstream to prevent axial heat loss by radi-
ation. The upstream heat shield also promoted additional
radial mixing of the reactants. The foams were sealed in the
quartz tube using an alumina-silicate cloth gasket that pre-
vents bypassing of gases. A hole was bored along the axis of
the downstream heat shield so that a chromel–alumel k-type
thermocouple could touch the back face of the catalyst in the
same location for each experiment. Alumina–silica insula-
tion was placed around the outside of the reactor to minimize
radial heat losses. A sample of the products was obtained for
analysis at the reactor outlet using a gas-tight syringe.
0
CO + H2O → CO2 + H2, ꢀH = −41 kJ/mol.
(4)
Along with generating more H2, the WGS reaction is im-
portant because it reduces the CO concentration, and CO
is a poison to PEM fuel cells. The exothermic WGS reac-
tion generally goes to equilibrium at the high temperatures
required for the reforming reactions, but it requires low tem-
peratures for favorable H2 equilibrium, at which point it
becomes kinetically limited [14].
By combining oxidation with steam reforming and WGS,
one can take advantage of the heat generated by total oxi-
dation and the extra H2 produced by steam reforming and
WGS. Stoichiometrically, these three reactions can be writ-
ten as
C2H5OH + 2H2O + (1/2)O2 → 2CO2 + 5H2,
0
ꢀ
H = −68 kJ/mol.
(5)
Thus adding water with air should maximize hydrogen
production and minimize CO in an exothermic process.
Ethanol–water mixtures are also beneficial because the need
to remove all the water is a significant cost in producing fuel
grade ethanol. Fermentation produces 10–20 moles of water
per moles of ethanol. Distillation and water separation from
the azeotrope using zeolite adsorption are then necessary to
produce fuel-grade ethanol.
2.2. Catalyst preparation
Ceramic foams (92% Al O , 8% SiO ) were used as cata-
lyst supports. They were supplied at 80 pores per linear inch
(ppi) in cylindrical segments 10 mm long and 17 mm in di-
ameter. The foams have a nominal surface area ∼1.0 m /g,
2
3
2
2

J.R. Salge et al. / Journal of Catalysis 235 (2005) 69–78
71
an average channel diameter of ∼200 µm, and a void fraction
of ∼80%. These foams were coated by the wet impregnation
method as described previously [17].
For single metal catalysts (Rh, Pt, and Pd), the met-
als were coated on the supports by soaking them in an
aqueous solution of the corresponding metal precursor salt
into the system, then H2 selectivity can exceed 100%. This is
possible if water is consumed in the process, because the H
atoms from the H2O can also be converted into H2. Accord-
ing to the reaction of interest (Eq. (5)), complete conversion
of the ethanol and water could generate 5 H2 per C2H5OH,
which gives a maximum H2 selectivity of 5/3, or 167%. The
consumption of added water results in negative selectivity to
H2O, but total H atom selectivity still sums to unity.
(
Rh(NO3)3, H2PtCl6, or PdCl3). The foams were dried and
◦
calcined at 600 C for 6 h. A calculated amount of metal salt
was used to ensure ∼5 wt% metal loading based on the mass
of the foam.
A γ -Al2O3 washcoat was added to select Rh catalysts to
decrease channel size and increase surface area [17]. The
supports were washcoated using a 3 wt% γ -Al2O3 slurry in
3. Results
In these experiments the C/O ratio was varied at a con-
stant total flow rate. The C/O ratio for ethanol and ethanol–
water mixtures is defined as the amount of C atoms entering
the reactor with ethanol divided by the number of O atoms
entering with oxygen and ethanol. Therefore, a C/O ratio of
2.0 corresponds to pure ethanol, a ratio of 1.0 corresponds
to syngas (H2 and CO) stoichiometry according to Eq. (2),
and a ratio of 0.29 corresponds to combustion according to
Eq. (3). The lowest C/O ratios shown represent the point
at which experiments were halted because of pulsing in the
catalyst, where compositions were within the flammability
regime of ethanol–air mixtures. The maximum C/O ratios
shown are the highest ratios that could be maintained in au-
tothermal operation without the catalyst extinguishing due
to a fuel-rich mixture.
◦
water, followed by drying and heating at 600 C for 6 h in
a closed furnace. Typical washcoats were ∼5% by weight
of the foam, for which an average alumina film thickness
of 10 µm was measured by scanning electron microscope
analysis. After the washcoat was applied, the foam was
coated with Rh as described previously.
The Rh–Ru catalysts were prepared by dipping the
foams in an aqueous mixture of Rh(NO3)3 and 0.1 mol
K2RuCl5(H2O) in 1.0 mol H2SO4. A calculated amount of
solution was used to ensure a metal loading of ∼2.5 wt%
each of Rh and Ru. The catalysts were dried and calcined at
◦
6
00 C for 6 h.
For the catalysts combining Rh with ceria, a calculated
amount of Rh(NO3)3 and Ce(NO3)3·6H2O that when de-
posited on the support would lead to ∼2.5 wt% each of Rh
and ceria were mixed together in an aqueous solution. The
3.1. Effect of catalyst
foam was soaked in this solution, dried, and then calcined at
◦
6
00 C for 6 h.
We investigated several noble metal and metal plus ceria-
coated alumina foams. The single metal catalysts (Rh, Pt,
and Pd) were ∼5 wt% of the metal based on the total mass of
the foam. A γ -Al2O3 washcoat was added to select Rh cat-
alysts (denoted as Rh wc in Fig. 1) to decrease the channel
size and increase the surface area of the catalyst, which has
been found to increase syngas selectivity and reduce olefins
All experiments were run for up to 30 h on a given cat-
alyst, and most experiments were repeated on several nom-
inally identical catalyst samples with no systematic differ-
ences or deactivation noted.
2
.3. Product analysis
[
17]. The addition of ceria to noble metal catalysts has been
Once the back-face temperature of the catalyst had sta-
shown to increase WGS activity [18] and is also commonly
used as an oxygen donor for automotive catalytic convert-
ers [19]. Therefore, foams were loaded with ∼2.5 wt% each
of Rh and Ce. The catalytic partial oxidation of ethanol
over structured Ru catalysts has been shown to produce high
selectivity to syngas, although these experiments were con-
ducted in a furnace and at longer contact times [10]. Here
Rh and Ru were combined with ∼2.5 wt% each based on
the mass of the foam.
The effect of the catalyst on reactor temperature, conver-
sion, and selectivity for the reforming of ethanol is shown
in Fig. 1. The total flow rate for these experiments was
6 slpm, resulting in a calculated catalyst contact time of
bilized (ꢀ10 min), a sample of the product gases was ob-
tained through a septum at the exit of the reactor. A gas-tight
syringe was used to withdraw product samples and inject
them into a dual-column gas chromatograph equipped with
thermal conductivity and flame ionization detectors. Column
retention times and response factors were determined by in-
jecting known species. Because nitrogen is an inert species,
it was used as the calibration gas for mass balances. Mass
balances on carbon and hydrogen typically closed to within
±
5%.
Product selectivities were calculated on an atomic basis.
Both C and H atom selectivities were calculated as the ra-
tio of the moles of a specific product to the total moles of all
products, accounting for stoichiometry. The three H2 mole-
cules from C2H5OH represent 100% H2 selectivity. Thus, if
pure ethanol is fed to the system, then maximum H2 selectiv-
ity is 100%; however, if an ethanol–water mixture is injected
◦
∼7 ms at 700 C. The corresponding gas hourly space ve-
5
−1
locity (GHSV) was ∼1.5×10 h . Ethanol conversion was
>85% and oxygen conversion (not shown) was >99% for all
ratios and catalysts considered. Pt and Pd had a higher back-
face temperature, while the other catalysts operated within a

72
J.R. Salge et al. / Journal of Catalysis 235 (2005) 69–78
Fig. 1. Comparison of the reforming of ethanol over Rh, Rh–Ce, Rh–Ru, Pt, and Pd coated foams at a total flow rate of 6 SLPM. Ethanol conversion (X),
catalyst back-face temperature (T ), and product selectivities are shown as a function of C/O ratio.

J.R. Salge et al. / Journal of Catalysis 235 (2005) 69–78
73
◦
similar temperature range, about 200 C cooler. As the feed
produced at higher flow rates. Selectivity to CH4 and C2H4
both increased sharply with higher flow rates, and their se-
lectivity peaked at C/O ∼ 1.1 for all flow rates. Meanwhile,
selectivity to C2H6 and CH3CHO (not shown) were indepen-
dent of flow rate, but increased with increasing C/O ratio.
The Rh–Ce catalysts had a less pronounced dependence
on flow rate. Fig. 3 shows the dependence of temperature,
conversion and product selectivities on flow rate for Rh–Ce
catalysts. Ethanol conversion was >95% for all flow rates.
The catalyst back-face temperature increased with increas-
ing flow rate because at higher flow rates the rate of heat
generation increases, causing the reactor to operate closer to
adiabatic.
Selectivities to syngas (H2 and CO) and combustion prod-
ucts (CO2 and H2O) depended only slightly on flow rate.
Over this range of flow, selectivity to H2 varied by ∼10%
at a given C/O ratio, whereas selectivity to CO and com-
bustion products exhibited a smaller variation. As was the
case with Rh catalysts, more CH4 and C2H4 were produced
at higher flow rates, whereas CH3CHO and C2H6 produc-
tion (not shown) remained relatively constant. Production of
all the minor products increased as the feed became more
fuel-rich for all flow rates.
becomes more fuel-rich, the back-face temperature of all of
the catalysts decreases, as expected. As shown in Fig. 1, the
order of effectiveness in syngas production is Rh–Ce >Rh–
wc >Rh–Ru >Rh >Pd >Pt. Rh–Ce is more stable and gives
greater WGS activity than noble metals alone. The selectiv-
ity to H2 peaks at ∼80% at a C/O ∼0.7 for Rh–Ce. Rh, Pt,
Pd, and Rh–Ru produced less H2, with Pt and Pd producing
<
50% H2. The range of operation of Pt and Pd was also lim-
ited. Pt was difficult to ignite and was unstable at low C/O
ratios, and Pd showed immediate coke formation, eventually
causing extinction in the reactor. Rh–Ru showed ∼65% se-
lectivity to H2; however, it was difficult to maintain steady
state at lower C/O ratios.
The minor products observed consisted of CH4, C2H4,
CH3CHO, and C2H6. At low C/O ratios, the production of
minor products was <3% for Rh–Ce. Meanwhile, Pt and
Pd produced ∼15% each of CH4 and C2H4. Only small
amounts of CH3CHO and C2H6 (not shown) were produced
on any of the catalysts at low C/O ratios. CH3CHO was
produced by ethanol dehydrogenation and was completely
reformed at high temperatures (low C/O ratios). As the fuel
in the feed increased, more CH3CHO was produced.
For the experimental conditions investigated, the major
products predicted by thermodynamic equilibrium are H2,
H2O, CO, CO2, and CH4. The selectivity to CH4 is high
at low temperatures (>50%), but decreases rapidly above
3.3. Water addition
Because Rh–Ce is more stable and gives greater WGS
activity, it was used as the catalyst for water addition ex-
periments. Typical results for the reforming of ethanol over
Rh–Ce with added water are shown in Fig. 4. These exper-
iments were done at a constant total flow rate of 6 slpm,
resulting in a calculated catalyst contact time of ∼7 ms
◦
◦
5
00 C and becomes negligible above 800 C. This decrease
in CH4 is accompanied by a corresponding increase in H2
selectivity. At low C/O ratios, product selectivities from
reforming ethanol over Rh–Ce coated foams were within
±
3% of those predicted by thermodynamic equilibrium. At
◦
a C/O ∼ 0.7, the Rh–Ce catalyst back-face temperature was
at 700 C. Curves are shown for pure ethanol (100%) and
◦
∼
810 C. At this temperature, equilibrium predicts a H2 se-
ethanol–water mixtures of 75, 50, 25, 20, and 10% ethanol
by mole. On a weight basis, 10 mol% ethanol corresponds
to 22 wt% ethanol, or 53 “proof.” This is close to the upper
limit of ethanol from fermentation.
lectivity of 82.9% and a H2O selectivity of 16.9%. Selec-
tivities to CO, CO2, and CH4 are 82.0, 17.9, and <0.2%,
respectively. Only negligible amounts of CH3CHO, C2H4,
and C2H6 are predicted.
For pure ethanol and ethanol–water mixtures, ethanol
conversion remained >95% for all C/O compositions.
Adding water increased the selectivity to H2. For pure
ethanol, the selectivity to H2 peaked at ∼80% at C/O ∼ 0.7,
whereas for 10% ethanol, the selectivity to H2 exceeded
100%, because both ethanol and water contribute H2. Mean-
while, the selectivity to CO decreased with added water, due
to increased WGS and steam reforming activity. For pure
ethanol, selectivity to CO peaked at ∼80% at C/O ∼ 0.7,
whereas for 10% ethanol, the selectivity to CO was <50%.
Thus, the H2/CO ratio rose to 6.3/1 and the CO/CO2 ratio
fell to 1/2.3 for 10% ethanol.
As expected, the minor products were CH4, C2H4, C2H6,
and CH3CHO, and the total of these products was <3% at
the H2 maximum of C/O ∼ 0.7, rising as C/O increased.
CH4 is the major byproduct of pure ethanol and ethanol–
water mixtures. Selectivity to CH4 rose quickly with in-
creasing C/O ratio. Adding up to 50% water changed the
selectivity to CH4 only slightly. However, at ethanol com-
3
.2. Effect of flow rate
The effect of flow rate on the reforming of ethanol was
studied over Rh and Rh–Ce catalysts. For these experiments,
the total flow rate was varied from 8, 6, and 4 slpm (GHSV
5
−1
∼
2, 1.5, and 1 × 10 h ). These correspond to catalyst
◦
contact times of 5–10 ms at 700 C. Oxygen conversion was
>
99% for all ratios and flow rates considered.
Fig. 2 shows the effect of flow rate on the reformation
of ethanol over Rh catalysts. Ethanol conversion was >95%
for all flow rates, whereas the back-face temperature of the
catalyst increased slightly with increasing flow rate. This
increase in temperature was more apparent at lower C/O
ratios. Reducing the total flow rate increased the syngas se-
lectivity and decreased higher products. Selectivity to H2
reached a maximum of ∼75% at a flow rate of 4 slpm, but
was reduced to <50% at 8 slpm. More minor products were

74
J.R. Salge et al. / Journal of Catalysis 235 (2005) 69–78
Fig. 2. The effect of flow rate on the reforming of ethanol over Rh at 8, 6, and 4 SLPM, which corresponds to catalyst contact times of 5 to 10 ms. Ethanol
conversion (X), catalyst back-face temperature (T ), and product selectivities are shown as a function of C/O ratio.
◦
positions <25% ethanol, more CH4 is produced due to de-
creased catalyst temperature. Methane is undesirable in this
process because it competes with H2 for hydrogen atoms,
but is inert in PEM fuel cells.
Unlike CH4, selectivity to C2H4 and C2H6 (results not
shown) decreased when water was added. For pure ethanol
and ethanol–water mixtures, C2H4 formation peaked at a
C/O ∼ 1.1 and then quickly declined. A negligible amount
of C2H6 was formed for ethanol–water mixtures with <50%
ethanol. The selectivity to CH3CHO (results not shown) had
no clear dependence on water addition.
was heated between 250 and 900 C. Ethanol and air were
introduced at a total flow rate of 6 slpm, which corresponds
to residence times of the gases in the furnace of 200–400 ms.
Ethanol conversion remained >90% and oxygen conversion
was >99%, which are feasible because ethanol can decom-
pose completely at high temperatures even in the absence of
a catalyst [3]. However, the selectivity to H2 fell to ∼25%,
and carbon formation on the blank foam was observed. Se-
lectivity to CH4 and C2H4 rose with temperature to ∼25 and
◦
∼20% at 900 C, respectively. Production of CH3CHO fell
◦
as the temperature increased, but was still ∼5% at 900 C.
3
.4. Blank experiment
3.5. Homogeneous modeling
To further study the effectiveness of producing syngas on
An elementary reaction mechanism to describe high-
temperature ethanol oxidation has been proposed by Mari-
nov [13]. This comprehensive model incorporates 57 species
and more than 370 elementary reactions. Note that this
model has not been validated within our operating range.
these catalysts, experiments were carried out in which an
uncoated alumina foam was inserted into a quartz tube and
placed in a furnace. The fuel was vaporized and mixed with
air at a C/O ∼ 1.0, then introduced to the furnace, which

J.R. Salge et al. / Journal of Catalysis 235 (2005) 69–78
75
Fig. 3. The effect of flow rate on reforming of ethanol over Rh–Ce at 8, 6, and 4 SLPM, which corresponds to catalyst contact times of 5 to 10 ms. Ethanol
conversion (X), catalyst back-face temperature (T ), and product selectivities are shown as a function of C/O ratio.
Using ChemKin, the Marinov model was solved assuming
a plug flow reactor with an isothermal temperature profile.
A similar approach was used by Hudgins et al., who com-
pared the Marinov model to empty tube experiments in a
version of ethanol and oxygen (data not shown) at these
conditions was predicted. The presence of the Rh–Ce cata-
lyst dramatically increased syngas selectivity. Production of
H2 increased by a factor of 4 at C/O ∼ 0.7, whereas selectiv-
ity to CO increased by ∼20%. The model predicts that CH4
selectivity will be fairly constant at ∼20%, whereas over
Rh–Ce, <3% was observed at a C/O ∼ 0.7, but it increased
with C/O to ∼20%. Selectivity of C2H4 over Rh–Ce was
<1% at C/O ∼ 0.7 and increased to only ∼10% at higher
C/O. However, the model predicts ∼10% at the low C/O
and >35% at the higher C/O. Analysis of this mechanism
identifies the primary oxidation pathway proceeding through
CH3CHO. However, the model predicts that CH3CHO will
be completely reformed to CO and CH4 at the conditions
chosen here.
◦
temperature range of 600–800 C, a residence time of 4 s,
and a C/O of 1.0 [20]. These authors found that the Mari-
nov model predicts trends, but does not reproduce the ex-
perimental data accurately. They attributed this difference to
experimental error in recording temperatures. Our results in
the present work show better agreement between the model
and our blank tube experiments.
The Marinov model was used to predict conversions and
product selectivities at our experimental conditions. Results
obtained over Rh–Ce are compared with those predicted
from homogeneous chemistry using the Marinov model in
Fig. 5. A total flow rate of 6 slpm was used for both. The
The predicted selectivities from the homogeneous model
at C/O ∼ 1.0 are comparable to those from the blank exper-
◦
model assumes a constant temperature of 900 C and a res-
◦
idence time of 200 ms, which is greater than the catalyst
iment at 900 C, shown to the right in Fig. 5. The residence
contact time of <10 ms in these experiments. Complete con-
time of the gases for both was ∼200 ms. The model pre-

76
J.R. Salge et al. / Journal of Catalysis 235 (2005) 69–78
Fig. 4. The effect of water on the reforming of ethanol over Rh–Ce at 6 SLPM. Curves are shown for pure ethanol (100%) and ethanol–water mixtures of 75,
0, 25, 20, and 10% ethanol on a mole basis. Ethanol conversion (X), catalyst back-face temperature (T ), and product selectivities are shown as a function of
C/O ratio.
5

J.R. Salge et al. / Journal of Catalysis 235 (2005) 69–78
77
allacycle, which decomposes to adsorbed carbon, hydrogen,
and oxygen species. These species then recombine to form
the syngas products observed in the present study. Surface
reactions forming products more complex than C1 carbon
species do not exist [18]. Most other species are probably
formed by homogeneous reactions.
Ethanol and acetaldehyde do not decarbonylate via a
common pathway on Rh. Once acetaldehyde is formed, it ad-
2
sorbs as η -acetaldehyde, with the α-carbon and the oxygen
bonded to the metal. This is followed by C–C bond scission
to form CO and methyl, then methane [21]. This could ex-
plain the high methane selectivity observed at high C/O.
4
.2. Gas phase reactions
Under the conditions used here, all reactions appear to
occur on the surface, except for the formation of minor
products such as acetaldehyde and ethylene. Analysis of
the Marinov mechanism identifies the primary ethanol ox-
idation pathway proceeding through acetaldehyde [13]. The
α-CH bonds are the weakest in ethanol, and the model pre-
dicts that abstraction of one of these produces acetaldehyde
and H atoms. This explains why acetaldehyde was observed
with ethanol, when previous work on the partial oxidation
of alkanes over noble metal coated foams produced no oxy-
genates [24].
Selectivities were predicted using the homogeneous gas
phase mechanism. As shown in Fig. 5, much more CH4
and C2H4 would have been formed had gas phase chemistry
dominated. It has been observed in this laboratory that with
alkanes, syngas is seen at low C/O and olefins are seen at
higher C/O. However, with ethanol, olefin production never
increased above ∼10%, even at the highest C/O before au-
tothermal operation extinguished.
Fig. 5. Product selectivities observed experimentally over Rh–Ce and those
predicted using a homogeneous model are shown on the left as a function
of C/O ratio. On the right are product selectivities observed in the blank
4
.3. Effect of catalyst
◦
experiment at C/O ∼ 1.0 and 900 C. The total flow rate was 6 SLPM and
◦
The choice of catalyst is a crucial factor in determining
the model assumes a constant temperature of 900 C and a residence time
of 200 ms.
syngas yield. Rh–Ce was the most stable and gave greater
syngas selectivity than noble metals alone. This is perhaps
because Ce was able to store oxygen and make it available
for reaction via a redox reaction [18]. Ru cannot undergo this
redox reaction, so the addition of Ru to Rh did not increase
syngas yield as did the addition of Ce. Pt and Pd produce
less syngas than the Rh-containing catalysts. They run at
a higher back-face temperature, which can in turn gener-
ate more products of homogeneous chemistry. The increase
in higher products on these catalysts led to coking, which
caused the Pd catalyst to extinguish.
dicts H2, H2O, CO, CO2, and CH4 selectivities to within 2%.
However, the model overpredicts C2H4 selectivity and un-
derpredicts CH3CHO selectivity by ∼5%.
4
. Discussion
4
.1. Surface reactions
Ethanol adsorption and decomposition have been care-
fully examined on well-defined single crystal surfaces.
Ethanol adsorbs dissociately via scission of the O–H bond as
an ethoxide species [21–23]. Over Pt and Pd, ethoxides are
subsequently dehydrogenated to acetaldehyde [23]. How-
ever, over Rh, ethoxides do not yield acetaldehyde [21], but
rather dehydrogenate via β-CH scission to form an oxamet-
4.4. Effect of washcoat
Application of a γ -Al2O3 washcoat to Rh catalysts has
been found to decrease channel size and increase surface
area of the catalyst [17]. Because Rh promotes syngas pro-
duction, the increased surface area of Rh due to washcoating

78
J.R. Salge et al. / Journal of Catalysis 235 (2005) 69–78
was found to improve syngas yields for the partial oxidation
of alkanes. This work shows that adding a washcoat to Rh
catalyst increases syngas selectivity ∼10%. However, Rh–
Ce is still a superior producer of syngas.
ethoxide species and then decomposed to carbon, oxygen,
and hydrogen species. These then reacted rapidly on the sur-
face to form H2 and CO; no reaction pathways on the surface
to form greater than C1 products were found to exist. If ho-
mogeneous reactions had dominated, then less syngas and
much more CH4 and C2H4 would have been observed.
4
.5. Rh stability
After 4–6 h of operation, the Rh catalyst was no longer
Acknowledgments
stable, and it began to deteriorate and crumble. There was
no noticeable change in the operation of the reactor or in
product yields, but when the reactor was dismantled, it was
found that the Rh-coated foam had turned to powder. The
addition of a washcoat to the Rh catalyst produced the same
result. However, this behavior was not observed for Rh–Ce,
Rh–Ru, Pt, or Pd.
This research was partially supported by grants from the
Minnesota Corn Growers Association and the Initiative for
Renewable Energy and the Environment at the University of
Minnesota.
Deterioration of a Rh/Al2O3 catalyst is not unique to this
system. Cavallaro et al. noticed that when oxygen is added
during the steam reforming of ethanol, Rh crystallites sin-
ter [12]. They hypothesized that this was due to strong local
temperature increases (hot spots) produced by the total oxi-
dation of ethanol. Their experiments were done using pow-
der catalyst, so they did not observe the destruction of the
foam. The reason for the deterioration of Rh/Al2O3 catalysts
during the autothermal reforming of ethanol is currently un-
der investigation.
References
[
[
1] C.A. Leclerc, J.M. Redenius, L.D. Schmidt, Catal. Lett. 79 (2002) 39.
2] E.C. Wanat, K. Venkataraman, L.D. Schmidt, Appl. Catal. A 276
(2004) 155.
[3] A.N. Fatsikostas, D.I. Kondarides, X.E. Verykios, Chem. Com-
mun. 2001 (2001) 851.
[
4] A.N. Fatsikostas, D.I. Kondarides, X.E. Verykios, Catal. Today 75
2003) 145.
5] D.K. Liguras, D.I. Kondarides, X.E. Verykios, Appl. Catal. B 43
2003) 345.
(
[
(
[6] S. Cavallaro, Energy Fuels 14 (2000) 1195.
[7] A.Y. Tonkovich, S. Perry, Y. Wang, D. Qui, T. LaPlante, W.A. Rogers,
Chem. Eng. Sci. 59 (2004) 4819.
5
. Conclusions
[
[
8] L.F. Brown, Int. J. Hydrogen Energy 26 (2001) 381.
9] G.A. Deluga, J.R. Salge, L.D. Schmidt, Science 303 (2004) 993.
High selectivities to H2 were achieved from the catalytic
[10] D.K. Liguaras, K. Goundani, X.E. Verykios, Int. J. Hydrogen En-
ergy 29 (2004) 419.
partial oxidation of ethanol and ethanol–water. Short-contact
time reactors can be tuned by changing catalyst, flow, and
feed conditions. Rh–Ce catalysts gave the highest selectivity
to syngas and were the most stable, perhaps due to the redox
capabilities of Ce. Less H2 and more minor products were
produced over Rh, Rh–Ru, Pt, and Pd catalysts. The addition
of washcoat to Rh increased syngas selectivity, but had no
effect on stability. Flow variation by a factor of 2 produced
only small changes in H2 production for Rh–Ce compared
with other catalysts.
Autothermal operation was achieved down to 10 mol%
ethanol in water. The addition of water to the Rh–Ce ca-
talyzed reactor increased H2 selectivity to >100%, because
both ethanol and water contribute H2. Due to increased WGS
and steam reforming activity, CO selectivity decreased to
[
11] D.K. Liguaras, K. Goundani, X.E. Verykios, J. Power Sources 130
2004) 30.
12] S. Cavallaro, V. Chiodo, A. Vita, S. Freni, J. Power Sources 123 (2003)
0.
[13] N.M. Marinov, Int. J. Chem. Kinet. 31 (1999) 183.
(
[
1
[
[
14] T. Bunluesin, R.J. Gorte, G.W. Graham, Appl. Catal. B 15 (1998) 107.
15] B. Lewis, G. von Elbe, Combustion, Flames and Explosions of Gases,
Academic Press, New York, 1987, p. 706.
[16] J.J. Krummenacher, K.N. West, L.D. Schmidt, J. Catal. 215 (2003)
3
32.
[17] A.S. Bodke, S.S. Bharadwaj, L.D. Schmidt, J. Catal. 179 (1998) 138.
[18] C. Wheeler, A. Jhalani, E.J. Klein, S. Tummala, L.D. Schmidt, J. Ca-
tal. 223 (2004) 191.
[
19] K.C. Taylor, in: J.R. Anderson, M. Boudart (Eds.), Automobile Cat-
alytic Converters, Catalysis: Science and Technology, vol. 5, Springer,
Berlin, 1984.
[20] D.O. Christensen, P.L. Silveston, E. Croiset, R.R. Hudgins, Ind. Eng.
Chem. Res. 43 (2004) 2636.
<
50% with added water. At the H2 production maximum,
total selectivity to unwanted products was <3% with water
addition.
[21] C.J. Houtman, M.A. Barteau, J. Catal. 130 (1991) 528.
[22] M. Mavrikakis, M.A. Barteau, J. Mol. Catal. A: Chem. 131 (1998)
135.
Under the conditions used in these experiments, surface
reactions dominated. Ethanol adsorbed dissociately as an
[23] A. Yee, S.J. Morrison, H. Idriss, Catal. Today 63 (2000) 327.
[24] M. Huff, P.M. Torniainen, L.D. Schmidt, Catal. Today 21 (1994) 113.