5.pdf

Hydrogen production from methanol decomposition over Pt/Al2O3
and ceria promoted Pt/Al2O3 catalysts
J. Christopher Brown, Erdogan Gulari *
Chemical Engineering Department, University of Michigan, Ann Arbor, MI 48109-2136, USA
Received 26 August 2003; received in revised form 7 May 2004; accepted 15 May 2004
Available online 20 June 2004
Abstract
We present the activities and selectivities of Pt/Al2O3-based catalysts in the production of hydrogen from methanol through
catalytic decomposition. For the decomposition reaction, the Pt and Pd catalysts promoted by ceria exhibit similar activity reaching
100% conversion at 315 °C. Promoting with ceria had a positive effect on activity and selectivity. Among all catalysts, the 9% Pt–
10% CeO2/Al2O3 catalyst demonstrated the best activity. This catalyst compares favorably to a commercial copper-based catalyst.
The hydrogen selectivity is exceptionally high at 99+%. In all experiments, we found that methane formation was not a major
factor.
Ó 2004 Elsevier B.V. All rights reserved.
Keywords: Methanol decomposition; Pt/alumina; Ceria; Hydrogen; PEM fuel cell
1. Introduction
There is a growing necessity to find alternative ways
to produce energy with lower emissions of pollutants
and higher efficiencies compared to internal combustion.
One such option is the use of proton exchange mem-
brane (PEM) fuel cell system. PEM fuel cells convert
hydrogen gas into useful electric power with an effi-
ciency that is not limited by thermodynamics and the
only by product is water [1,2]. Due to current infra-
structure, storage technology and safety concerns, hy-
drogen gas cannot be stored on-board in adequate
amounts for mobile applications. On demand reforming
of liquid hydrocarbons to produce hydrogen is seen as
an attractive means of providing the necessary hydrogen
to the fuel cell. With the exception of formic acid,
methanol is the easiest hydrocarbon to reform. On
board methanol reforming has a couple of advantages
over other potential fuels: unlike gasoline or diesel fuel,
liquid methanol can be readily produced from biomass,
it is easily adaptable to the current infrastructure, it
is easily transported and stored, and finally it has a
high hydrogen density [3–5]. The most basic way to
generate H2 from methanol is the decomposition of pure
methanol.
CH3OH ! 2H2 þ CO DEC DH0
298 ¼ þ90:7 kJ=mol
ð1Þ
The decomposition reaction is endothermic with a 90.7
kJ/mol heat of reaction. Therefore, heat must be pro-
vided to maintain the reaction. Apart from being
harmful to health, carbon monoxide produced is an
atmospheric pollutant and a poison for the fuel cell [6].
So, some downstream processing must be done to
eliminate the CO from the system before releasing the
product stream to the fuel cell. The decomposition can
also produce undesired by products such as di-methyl-
ether and methane.
In this communication, we present the effect of cat-
alyst formulation and also the reaction conditions on
the production of hydrogen from methanol through
decomposition over Pt and Pd catalysts.
*
Corresponding author. Tel.: +1-734-763-5941; fax: +1-734-763-
0459.
E-mail address: gulari@umich.edu (E. Gulari).
1566-7367/$ - see front matter Ó 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.catcom.2004.05.008
Catalysis Communications 5 (2004) 431–436
www.elsevier.com/locate/catcom

2. Catalyst preparation
2.1. Sol–gel method
To prepare Pt/Al2O3 and Pd/Al2O3 in a single step
with the sol–gel method, a known amount of aluminum
isopropoxide (AIP), 98% (from Alfa Aesar Inc.) was
hydrolyzed in water at 85 °C and then a small amount
of nitric acid was added to obtain a clear sol [7]. After
that, the necessary amount of either H2PtCl6 or
Pd(NO3)2 precursor was added to obtain the desired
metal loading (from 1 to 10 wt%) in the final catalyst
formulation. Ceria doped catalysts were also prepared
in the same way as the un-doped catalysts but cerium
(III) acetate (99% from Alfa Aesar Inc.) was added to
the clear sol solution. Pt or Pd precursor solution was
added last. All sol solutions were aged overnight at
room temperature and then turned into gels by slowly
evaporating the excess solvent. Once gels were ob-
tained, they were dried at 110 °C for 24 h to remove
water and solvent. The dried gels were calcined in air,
in a furnace by heating from room temperature to 500
°C with an 8 °C/min temperature ramp. Once 500 °C
was reached, they were kept at this temperature for 24
h. Finally, all catalysts were ground and sieved to 80–
120 mesh size. The ZrO2 promoted catalysts were made
in a similar manner.
2.2. Activity tests
A vertical Pyrex glass tube (i.d. 3 mm) was used as a
reactor. 0.060 g of a catalyst was held in place by a
glass wool plug. The temperature of the top of the
catalyst in the reactor was monitored with a K-type
thermocouple. Before the activity studies, all catalysts
were reduced at 300 °C under a gas mixture containing
20% hydrogen in helium at a flow rate of 100 ml/min.
After the end of H2 reduction step, all catalysts were
purged with a flow of He at 300 °C and then activity
tests were conducted from 160 to 400 °C. The product
stream of the reactor was analyzed with a HP 5710A
GC equipped with a TCD detector after reaching a
steady state in 1 h. Products were separated in a 10 ft
Carbosphere 80/100 column operated under a temper-
ature program (from 25 to 120 °C with a 10 °C/min
temperature ramp).
2.3. Catalyst characterization
The crystalline phases in the catalysts were identified
with a Rigaku powder diffractometer. The average
crystallite size was computed from the Debye–Scherrer
equation. The BET (Brunauer, Emmet, Teller) surface
area was measured with a MicroMeritics ASAP2010
instrument.
3. Results and discussion
3.1. Effect of Pt loading
In this set of experiments, we explored the effect of
changing the platinum loading on decomposition ac-
tivity of methanol and the composition of the product
stream. One of the goals was to find the optimal amount
of platinum to be incorporated into the pure alumina
support. Three loadings of platinum were investigated;
2%, 5% and 9%. Pure methanol diluted with helium gas
was decomposed by these catalysts. The results of the
activity tests for different Pt loading catalysts on sol–gel
alumina are displayed in Fig. 1. Decomposition activity
was increased when the Pt loading was increased from
2% to 9%. The complete conversion temperature drop-
ped from 405 to 390 °C. Further increasing the loading
had no effect on the activity. The 5% and 9% Pt catalysts
exhibited essentially 100% hydrogen selectivity for all
conversions (hydrogen selectivity is defined as the frac-
tion of hydrogen in methanol that ends up as molecular
hydrogen). The increases observed in the decomposition
rate are much less than the increase in the catalyst
loading. This can be due to increased crystallite size or
to some other limitation such as mass transfer. Another
possibility is the encapsulation of Pt inside the support
(past measurements indicated that the amount of en-
capsulated Pt could be as high as 30%).
The 2% Pt catalyst started forming methane at 360 °C
and the selectivity for hydrogen dropped to 92% at the
maximum conversion temperature of 405 °C. Since
production of maximum amount of hydrogen is the
goal, the methanation reaction is undesirable
CO þ 3H2 $ CH4 þ H2O MET ð2Þ
Fig. 1. Methanol conversion for various loading of Pt catalyst.
432 J.C. Brown, E. Gulari / Catalysis Communications 5 (2004) 431–436

The low methane selectivity of Pt compares favorably
with the selectivity for Ni-based catalysts, which gener-
ally exhibit higher amounts of methane. Matsumura
et al. [8] reported a selectivity of CH4 higher than 15%
over a nickel supported on silica sol–gel catalyst. From
the methanation (2) reaction, H2O is formed. Water can
participate in the water gas shift reaction (WGS) (3).
CO þ H2O $ H2 þ CO2 WGS DH0
298 ¼ 39:4 kJ=mol
ð3Þ
Under our reaction conditions, no CO2 was observed
in the exit stream so it is likely that no WGS is occur-
ring. This is unlike what Imamura et al. [9] found for a
Pt/CeO2 catalyst, which produced a considerable
amount of methane and CO2 under similar conditions.
The production rates of hydrogen are 289 mmol/s kg for
the 9% Pt catalyst, for the 2% Pt it is 273 mmol/s kg
catalyst because of the lower selectivity. To assure that
the activity of the catalysts were due to the platinum, an
activity testing of the sol–gel alumina was conducted
and no activity was observed between 175 and 405 °C.
The 390 °C complete conversion temperature for the
9% Pt catalyst we observed is rather high compared to
reported complete conversion temperatures for other
catalysts [9,10]. However, the space velocity of our sys-
tem is generally an order of a magnitude higher than
those reported. Imamura used a similar 3% Pt/Al2O3
catalyst and obtained complete conversion temperature
at 290 °C, but at a space velocity of 4200 h1
and 3.8%
volume of methanol in the feed [9]. The complete con-
version temperature is 100 °C lower but the amount of
hydrogen produced will be an order of magnitude less.
The hydrogen selectivities for our catalysts are ex-
ceptionally high. The hydrogen selectivity appears to
increase as the platinum loading is increased. This is
most likely due to complete conversion occurring at
lower temperatures with the higher loadings. Methana-
tion competes for the hydrogen atoms and it is primarily
seen at the higher temperatures, thus hydrogen selec-
tivity suffers at these temperatures. Our selectivities of
100% compares favorably with the 93.9% selectivity
reported for a 3% Pt/Al2O3 by Imamura et al. [9] and to
the 82% selectivity of a rhodium/alumina catalyst tested
by Imamura [11]. Our selectivities are also better than
the selectivities reported for some Cu [12] and Ni-based
catalysts [13].
3.2. Effect of ceria loading
Ceria was incorporated during the sol formation into
the catalysts at 5%, 10%, and 15% loadings. The plati-
num was then impregnated into the support with a
loading of 5% or 9%.
The methanol decomposition activity results for
varying ceria amounts with the 5% and 9% Pt loadings
are shown in Table 1. The addition of ceria has a sig-
nificant positive effect, decreasing the complete conver-
sion temperature when compared to the pure alumina
supported catalysts. The catalysts with 10% ceria show
the best results, with complete conversion temperatures
of 345 and 315 °C, respectively, for the 5% and 9% Pt
catalysts. A further increase in ceria amount to 15% had
a negative effect on the activity, increasing the complete
temperature to 360 °C. Again, the hydrogen selectivity
for all the catalysts are very high with the selectivity
always remaining above 97% in the tested temperature
ranges. The catalyst without ceria exhibited the lowest
selectivity, 97%, at its complete conversion temperature.
Overall, the differences in selectivity between various
ceria loadings are within experimental error and addi-
tion of ceria increases the selectivity at the highest
conversions. The data also seems to indicate that the
temperature of complete conversion determines whether
methane will be observed or not. The increased activity
results are similar to those observed by Mansilp [7], for
CO oxidation.
Another interesting finding from this study is the ef-
fect the incorporation of ceria on the optimum platinum
loading. With pure alumina as a support 5% Pt/Al2O3
and 9% Pt/Al2O3 exhibited very similar activities with
Table 1
Complete conversion temperature and selectivity for methanol decomposition at 100% conversion for catalysts containing ceria
Catalyst Complete conversion temperature Selectivity (%)
H2 CO CH4
5% Pt/Al2O3 390 97.0 97.1 2.9
5% Pt–5% CeO2/Al2O3 360 98.2 98.5 1.5
5% Pt–10% CeO2/Al2O3 345 99.1 99.1 0.9
5% Pt–15% CeO2/Al2O3 360 97.6 97.6 2.4
9% Pt/Al2O3 390 97.9 98.1 1.9
9% Pt–5% CeO2/Al2O3 360 97.4 97.4 2.5
9% Pt–10% CeO2/Al2O3 315 99.1 99.5 0
9% Pt–15% CeO2/Al2O3 345 98.2 98.3 1.7
Hydrogen selectivity is defined as % of the original hydrogen in methanol and carbon selectivity is defined as the % of the methanol carbon ending
up as either CO or CH4. Since the only by product is methane having the same number of atoms as methanol the carbon and hydrogen selectivities
for methane are identical as a result only 1% is given.
J.C. Brown, E. Gulari / Catalysis Communications 5 (2004) 431–436 433

both reaching complete conversion at 390 °C. However,
after addition of 10% ceria, the 5% Pt catalyst had a
complete conversion temperature of 345 °C and the 9%
Pt catalyst had a complete conversion temperature of
315 °C. This can most likely be explained by the ability
of ceria to stabilize small Pt crystallites even at fairly
high metal loadings. This finding suggests that the op-
timum platinum loading changes with the incorporation
of ceria, again similar to the observations made in the
selective CO oxidation study [7].
Stenger and Choi [14] have demonstrated that the
commercial Cu-based catalyst exhibits complete con-
version at 330 °C, at a space velocity of 4400
ml1
h1
g cat1
. Their system also had significant
amount of by-products, with methyl formate amounts
as high as 6%. These results can be compared to 99.5%
hydrogen selectivity and 315 °C complete conversion
temperature obtained at an order of magnitude higher
space velocity in this study.
Since most commercial catalysts containing ceria also
have zirconia added to improve dispersion and activity
we also tested adding some zirconia along with ceria.
Zirconia was incorporated into the catalyst at loadings
of 2%, 3.5%, and 5%. The ceria amount was adjusted so
that the total zirconia and ceria would add up 10%. The
support was made by the sol–gel process and contained
alumina, ceria and zirconia. The zirconia precursor,
zirconium-n-butoxide, was added to the sol after the
ceria precursor. When zirconia containing catalysts were
tested, we found no improvement due to the presence of
zirconia, indeed both the activity and the selectivity of
the catalysts were worse than the ceria only catalysts.
These results differ from results obtained from other Zr/
CeO2 systems. Kapoor et al. [15] reported an optimal
Zr/Ce ratio of 3–7 with an increase of conversion by
more than 50% compared to the ceria only catalyst. Liu
et al. [16] used a Zr/Ce ratio of 1:4 in their support with
palladium to obtain complete conversion at a tempera-
ture of 250 °C at a space velocity of one tenth ours.
3.3. Space velocity effect
Space velocity is one of the key operating conditions
that affect production rates. Generally, an increase in
space velocity will result in lower conversions. The cat-
alyst used in these experiments was the 5% Pt/10%
CeO2–Al2O3 catalyst. The concentration of methanol in
the feed remained constant at 14% while the total flow
was varied. The space velocities tested were 42,150,
84,300, and 126,450 ml1
h1
g cat1
. The results from
this experiment are displayed in Fig. 2. As expected the
42,150 condition performed the best with a complete
conversion temperature near 330 °C. The middle space
velocity of 84,300 reached complete conversion at 345
°C. While a further increase to 126,450 ml1
h1
g cat1
,
increased the complete conversion to 360 °C. The hy-
drogen production rates for the 42,150, 84,300, and
126,450 ml1
h1
g cat1
experiments at complete con-
version temperatures were 148, 294, 424 mmol/s kg cat-
alysts, respectively. Even though our studies were
carried out in integral conversion mode, a check of
the conversions as a function of flow rate at a moderate
conversion temperature of 260 °C shows that the
conversion is almost inverse of the flow rate ratio, evi-
dence that there are no serious mass transfer limitations
in the system.
3.4. Direct comparison to Cu/Ce–Al2O3 and Cu/Zn/Al2O3
catalysts
Copper-based catalysts have been well studied for
methanol reforming and are often referenced for com-
parison. Due to the relatively high space velocity used in
this study and the lack of any copper-based catalyst
studies at these levels, a commercial Cu/Zn/Al2O3 cat-
alyst was obtained from Sud Chemie [17], and we pre-
pared a pure Cu/CeO2–Al2O3 catalyst. The methanol
conversions as a function of temperature for these cat-
alysts are compared to the 9% Pt/10% CeO2–Al2O3
catalyst in Fig. 3. Although the complete conversion
temperature is a few degrees °C lower for the commer-
cial catalyst, at lower conversions and temperatures the
platinum catalyst has slightly better activity. The cop-
per–ceria catalyst exhibits the worst activity out of the
three catalysts, with a complete conversion temperature
of 340 °C. The products from the methanol conversion
of these catalysts are different and thus the hydrogen
selectivity is also slightly different. The hydrogen selec-
tivity for the catalysts is shown as a function of con-
version in Fig. 4. The platinum catalyst hydrogen
selectivity stays at essentially 100% for all conversions.
The copper–ceria catalyst selectivity ranged from 86% to
91%. The commercial copper catalyst starts with a se-
Fig. 2. Methanol conversion by 5% Pt/10% CeO2–Al2O3 formulations
at different space velocities.

lectivity of 92% at 10% conversion. The selectivity de-
creases reaching a minimum of 83% at a conversion of
40% (temperature of 270 °C). It then increases back to a
level slightly above 92% at the complete conversion
temperature of 310 °C.
While a trace of methane is the only by-product ob-
served in the platinum system, methane, methyl formate,
dimethyl ether and carbon dioxide were all observed at
some point in the copper systems. The methyl formate
levels reached as high as 4% of the products. At higher
temperatures and conversions the amount of methyl
formate decreased to a minimum of 1%. Dimethyl ether
levels in the product stream remained relatively constant
throughout the tested temperature range ranging from
1% to 1.6%. Small amounts of carbon dioxide and
methane (0.5%) were also observed at the higher
temperatures.
3.5. Direct comparison to Pd/CeO2–Al2O3 catalyst
In methanol decomposition, more work has been
done with palladium catalysts. To determine if there
were differences between the Pt and we made Pd cata-
lysts and tested them under identical conditions.
Instead of platinum, palladium was incorporated into
the catalyst supports at loadings of 2%, 5%, and 10%
palladium. The decomposition curves for the palladium
catalysts are shown in Fig. 5. The performance of the
10% Pd catalyst is basically identical with the 9% Pt
catalyst having a complete conversion temperature of
315 °C. The activity decreases with decreasing Pd
loading. The 5% Pd catalyst has a complete conversion
temperature of approximately 330 °C. The 2% Pd cat-
alyst has a complete conversion temperature close to 390
°C. These results indicate that the higher the palladium
loading in the catalyst, the better the activity will be, at
least up to a loading of 10% palladium. The palladium
catalysts in these studies did not exhibit as much low
temperature activity as some other palladium catalysts
mentioned in literature, mainly due to our high space
velocity and lower Pd loading. At a palladium loading
of 17% on pure ceria support, Liu et al. [16] reported
conversions near 100% at 250 °C, but at a space velocity
of one-eighth of that used in this study.
3.6. Catalyst characterization
The BET surface areas of the catalysts ranged from a
low of 244–270 m2
/g . The pure alumina support had a
slightly higher surface area of 280 m2
/g. This suggests
that the metals incorporated in the ceria–alumina sup-
port have little effect, if any, on the surface area. The
similarity was also carried over to the pore sizes. The
average pore diameters for all the catalysts was around
50
A.
Fig. 3. Methanol conversion for various catalysts supported on alu-
mina.
Fig. 4. Hydrogen selectivity based on temperature for various catalysts
supported on alumina.
Fig. 5. Methanol conversion for different X% Pd/10% CeO2–Al2O3
formulations X ¼ varying Pd amount and a 9% Pt/10% CeO2–Al2O3
catalyst.
J.C. Brown, E. Gulari / Catalysis Communications 5 (2004) 431–436 435

The crystallinity of the catalysts was measured by X-
ray powder diffraction. For the best catalyst (9% Pt–10%
CeO2–Al2O3), the support appears to contain a combi-
nation of c-alumina, g-alumina, and ceria crystalline
phases. The spectrum also shows that the Pt is in a
metallic state. By using the Debye–Scherrer equation,
we found a rough estimate of the average crystallite sizes
of the components. The alumina crystallites were 39
A,
the ceria crystallites were 53
A, the Pt crystallites were
160
A. Similar Pt crystallite sizes were observed for the
lower Pt loading catalysts.
CO chemisorption was used to determine the metal
dispersion of the best performing catalyst, 9% Pt/10%
CeO2–Al2O3. It was found to have a platinum metal
dispersion of 25%. This is a low dispersion but typical of
high platinum loadings. The platinum surface area was
found to be 6.32 m2
/g catalyst and 70.23 m2
/g platinum.
The surface area obtained from the chemisorption
measurement is higher than the 15% dispersion cal-
culated from the average crystallite size obtained from
the Debye–Scherrer equation. This is not surprising gi-
ven the strong bias of XRD towards the larger crystallite
sizes. X-ray results gave the same Pt crystallite size es-
sentially independent of loading. This we speculate is
due to Pt sol formation after adding the Pt precursor to
the transparent alumina (or alumina ceria sol).
4. Conclusions
Our results show that Pt/ceria–alumina catalysts are
very selective and active methanol decomposition cata-
lysts. The activity increases with Pt loading. Ceria pro-
moted catalysts are significantly more active than
unpromoted catalysts. We find that the optimum ceria
loading is in the neighborhood of 10%. Promotion with
zirconia results in lower activity. The performance of Pd
catalysts was found to be comparable to Pt catalysts of
the same wt% metal loading. The 9% Pt–10% CeO2/
Al2O3 catalyst showed the highest activity and yield
among the catalysts that we studied under our reaction
conditions. Its performance is comparable or better than
the commercial copper catalyst tested. Over all the Pt-
based catalysts are more robust than the Cu–Zn/alu-
mina catalysts because they are not pyrophoric and can
tolerate an exposure to the atmosphere without suffering
catastrophic decay in activity.
Acknowledgements
We gratefully acknowledge financial support of this
research by the National Science Foundation and the
Department of Education and Department of Energy.
References
[1] L. Alejo, R. Lago, M.A. Pe~
na, J.L.G. Fierro, Appl. Catal. A:
General 162 (1–2) (1997) 281.
[2] C. Bracchini, V. Indovina, S. De Rossi, L. Giorgi, Catal. Today 55
(1–2) (2000) 45–49.
[3] J. Agrell, K. Hasselbo, K. Jansson, S.G. J€
ar
as, M. Boutonnet,
Appl. Catal. A: General 211 (2) (2001) 239.
[4] P. Mizsey, E. Newson, T. Truong, P. Hottinger, Appl. Catal. A:
General 213 (2) (2001) 233.
[5] S. Velu, K. Suzuki, M. Okazaki, M.P. Kapoor, T. Osaki, F.
Ohashi, J. Catal. 194 (2) (2000) 373.
[6] S. Gottesfeld, J. Pafford, J. Electrochem. Soc. 135 (10) (1988)
2651.
[7] A. Manasilp, E. Gulari, Appl. Catal. B: Environ. 37 (1) (2002)
17–25.
[8] Y. Matsumura, K. Tanaka, N. Tode, T. Yazawa, M. Haruta,
J. Mol. Catal. A: Chemi. 152 (3) (2000) 157.
[9] S. Imamura, T. Higashihara, Y. Saito, H. Aritani, H. Kanai,
Y. Matsumura, N. Tsuda, Catal. Today 50 (2) (1999) 369.
[10] T. Tsoncheva, S. Vankova, D. Mehandjiev, Fuel 82 (7) (2003) 755.
[11] S. Imamura, T. Yamashita, R. Hamada, Y. Saito, Y. Nakao, N.
Tsuda, C. Kaito, J. Mol. Catal. A: Chemi. 129 (2–3) (1998) 249.
[12] W.H. Cheng, C.Y. Shiau, T.H. Liu, H.L. Tung, H.H. Chen, J.F.
Lu, C.C. Hsu, Appl. Catal. B: Environ. 18 (1–2) (1998) 63.
[13] Y. Matsumura, N. Tode, T. Yazawa, M. Haruta, J. Mol. Catal.
A: Chemical 99 (3) (1995) 183.
[14] Y. Choi, H.G. Stenger, Appl. Catal. B: Environ. 38 (4) (2002) 259.
[15] M.P. Kapoor, Anuj Raj, Yasuyuki Matsumura, Microporous
Mesoporous Mater. 44 (565) (2001).
[16] Y. Liu, T. Hayakawa, T. Ishii, M. Kumagai, H. Yasuda, K.
Suzuki, S. Hamakawa, K. Murata, Appl. Catal. A: General 210
(1–2) (2001) 301.
[17] This was part of a sample given by Sud Chemie to Professor Levi
Thompson’s group at the University of Michigan.

5.pdf

More Related Content

Similar to 5.pdf (20)

More from ZeenathulFaridaAbdul1 (11)

Recently uploaded (20)

5.pdf