First use of macroporous silicon loaded with catalyst film for a chemical reaction: A microreformer for producing hydrogen from ethanol steam reforming

Article abstract of DOI:10.1016/j.jcat.2008.02.006

Macroporous silicon with millions of parallel microchannels per square centimeter can be used as a micromonolithic support for high-speed catalysis in miniaturized systems, opening a new and exciting field of research and application. We used macroporous silicon membranes with >10⁶ parallel microchannels with a diameter of 3.3 μm and depth/width ratio >60 as microreactors for the generation of hydrogen through the steam reforming of ethanol. The microchannels walls were coated with a thin layer of Co₃O₄-ZnO catalyst by a complexation-decomposition method, resulting in a specific inner surface of 10⁵-10⁶ m² m^-3. A constant flow of 20 μL s^-1 H₂ (0.054 mmol min^-1 H₂) was obtained from a liquid water-ethanol mixture at 773 K with a steam-to-carbon ratio of S/C = 3 and LHSV = 2 × 10⁴ h^-1 (0.1 mL min^-1). The turnover frequency under these conditions was ca. 1.1 s^-1, based on surface Co atoms.

Full text of DOI:10.1016/j.jcat.2008.02.006

Journal of Catalysis 255 (2008) 228–233
www.elsevier.com/locate/jcat
First use of macroporous silicon loaded with catalyst film
for a chemical reaction: A microreformer for producing hydrogen
from ethanol steam reforming
a,∗
a
b
b
b
Jordi Llorca , Albert Casanovas , Trifon Trifonov , Angel Rodríguez , Ramon Alcubilla
a
Institut de Tècniques Energètiques, Universitat Politècnica de Catalunya, Diagonal 647, ed. ETSEIB, 08028 Barcelona, Spain
b
Departament d’Enginyeria Electrònica, Universitat Politècnica de Catalunya, Campus Nord, Mòdul C4, c. Jordi Girona 1-3, 08034 Barcelona, Spain
Received 21 December 2007; revised 11 February 2008; accepted 11 February 2008
Available online 14 March 2008
Abstract
Macroporous silicon with millions of parallel microchannels per square centimeter can be used as a micromonolithic support for high-speed
catalysis in miniaturized systems, opening a new and exciting field of research and application. We used macroporous silicon membranes with
6
>
10 parallel microchannels with a diameter of 3.3 µm and depth/width ratio >60 as microreactors for the generation of hydrogen through the
steam reforming of ethanol. The microchannels walls were coated with a thin layer of Co O –ZnO catalyst by a complexation–decomposition
3
4
5
6
2
−3
−1
−1
method, resulting in a specific inner surface of 10 –10 m m . A constant flow of 20 µL s H (0.054 mmol min H ) was obtained from a
liquid water–ethanol mixture at 773 K with a steam-to-carbon ratio of S/C = 3 and LHSV = 2 × 10 h (0.1 mL min ). The turnover frequency
2
2
4
−1
−1
−
1
under these conditions was ca. 1.1 s , based on surface Co atoms.
2008 Elsevier Inc. All rights reserved.
©
Keywords: Macroporous silicon; Micromonolith; Hydrogen; Ethanol; Steam reforming
1
. Introduction
conditions with higher product yields [9]. The improved heat
transfer properties in microreactors are particularly relevant for
endothermic reactions, such as the steam reforming of alcohols.
Although several microreactors for onsite production of hydro-
gen from methanol steam reforming at 533–723 K have been
reported [10–15], the higher temperatures required for ethanol
steam reforming has prevented extensive work in this field [16].
Nonetheless, interest in ethanol steam reforming is growing
rapidly, because ethanol is a renewable fuel readily produced
from biomass (bioethanol) and yields more hydrogen on a mo-
lar basis,
A major impediment to the success of fuel cells is the dif-
ficulties and hazards involved in the storage and handling of
hydrogen [1]. Onsite steam reforming is an alternative to pro-
duce hydrogen from organic sources with the aid of a catalyst,
with alcohols (i.e., methanol and ethanol) among the best can-
didates for portable applications [2]. Using alcohols for steam
reforming is attractive due to their high volumetric energy den-
sity, low cost, safety, and easy transportation. Although con-
siderable work has been done on catalytic steam reforming
of methanol and ethanol for hydrogen production using con-
ventional macroscale reactors [3–8], the use of microreactors
is relatively new. Besides their compactness and light weight,
the microreactor systems provide rapid mass and heat trans-
port due to large surface area-to-volume ratios and short dif-
fusion length, low pressure drop, and precise control of process
C H OH + 3H O → 6H + 2CO .
(1)
2
5
2
2
2
Herein we report for the first time the use of macroporous sil-
icon membranes containing straight, parallel channels of a few
micrometers in diameter for the generation of hydrogen for mi-
crodevices, and validate their capability to serve as catalyst
support structures by performing steam reforming of ethanol
at 698–773 K. (Silicon is stable to oxidation in the presence
of steam up to 1073 K.) A new method was developed for de-
*
Corresponding author.
E-mail address: jordi.llorca@upc.edu (J. Llorca).
0021-9517/$ – see front matter © 2008 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2008.02.006

J. Llorca et al. / Journal of Catalysis 255 (2008) 228–233
229
positing a ca. 100-nm-thick layer of Co3O4–ZnO catalyst on the
microchannel wall that is catalytically active. ZnO-supported
cobalt has shown good activity and selectivity toward ethanol
steam reforming at moderate temperatures in fixed-bed reactors
[
17,18]. Very recently, Si micromonoliths have been used as
supports for zeolite films [19], but no application in a chemical
reaction has been yet attempted.
2
. Experimental
Macroporous silicon membranes were prepared by photo-
assisted electrochemical etching [20] and subsequent opening
of the pores from the back side. The substrate material was
ꢀ
100ꢁ n-type float-zone silicon, with a resistivity of 2–6 ꢀ cm.
Before the electrochemical etching, the wafer surface was pre-
structured by lithography and tetramethylammonium hydroxide
(
TMAH) etching was used to create a square array of inverted
pyramids. These pyramids pointed toward the bulk of the wafer
and defined the positions of pore growth. The electrochemical
etching was then carried out at 288 K in 5 wt% HF solution, us-
ing TritonX-100 surfactant (0.1 mmol) as a wetting agent. The
structured (front) side of the wafer was in contact with the elec-
trolyte, while the backside was illuminated through an array of
LEDs with an 880-nm peak emission wavelength. A constant
anodic potential of 2 V was applied. The etching current was
computer-controlled and was reduced as a nonlinear function
of the etching time, to compensate for the effects of limited HF
diffusion in the narrow pores [20,21]. Under these conditions,
about 330 min of etching time was required to achieve a pore
depth of 220 µm. After the electrochemical etching, the macro-
porous samples were oxidized in dry O2 at 1373 K for 30 min.
The oxide layer on the backside was removed, and the remain-
ing backside silicon was etched off in 25 wt% TMAH solution
at 358 K until the pore tips were reached. Finally, the samples
were dipped in 5 wt% HF to remove any oxide remaining on
the pore walls and on the wafer’s front side. The resulting struc-
tures were silicon membranes with pores (channels) opened at
both sides and arranged in a square lattice with a periodicity of
Fig. 1. SEM image of the macroporous silicon membrane as prepared showing
the regular array of parallel microchannels.
(
Carbolite, accuracy ± 0.1 K). Two sets of experiments were
carried out with different reactions mixtures. A reaction mix-
−1
ture consisting of gaseous ethanol (0.24 mL min , or 1.07 ×
−2
−1
1
0
mmol min ) and water (fed separately) diluted in He was
passed at atmospheric pressure through the silicon membrane at
−1
4
6–82 STP mL min , with C2H5OH:H2O molar ratios of 1:3
−1
and 1:6 (steam-to-carbon ratios of 1.5 and 3 mol mol , respec-
tively). In addition, an undiluted mixture of C2H5OH:H2O =
1
:6 (molar ratio) was provided directly by a syringe pump (Ge-
−1
nie Plus, Kent Scientific) at a constant rate of 0.1 mL min
.
Reaction products were analyzed continuously by online mass
spectrometry (Cirrus MKS) every 15 s through a capillary in-
let held at 353 K. The composition values were obtained after
proper calibration using a standard mixture of known com-
position and by parallel injection into a gas chromatograph
equipped with MS 5 Å, PlotU, and Stabilwax columns (Agilent
3
000A). Scanning electron microscopy (SEM) was performed
with a JEOL JSM-6400 electron microscope. Microstructural
characterization by high-resolution transmission electron mi-
croscopy (HRTEM) was carried out at 200 kV with a JEOL
JEM-2010F electron microscope equipped with a field emis-
sion gun.
4
µm.
The walls of the macroporous silicon channels were coated
with a catalyst layer containing zinc and cobalt oxides. During
the first step in the preparation procedure, the channels were
completely filled with a dimethylketone solution containing
equal amounts of Zn2 and Co (0.35 M, nitrate precursors),
and urea (1.4 M). The precursor solution was forced to pass
through the macroporous silicon channels by applying a pres-
sure gradient with a vacuum pump and a pressure regulator set
at 75 kPa. After ensuring complete air removal and complete
filling of channels, the silicon membrane was heated first at
+
2+
3
. Results and discussion
Macroporous silicon channels were imaged by SEM and
analyzed by EDX before and after deposition of the catalyst
layer. Figs. 1 and 2a show typical SEM images of the macro-
porous Si membrane as prepared, which has a regular distrib-
ution of channels with smooth walls. The channels are 3.3 µm
in diameter and 210 µm in length, producing an aspect ratio
(depth/width) of about 64. Deposition of a catalyst coating on
such small channels with high aspect ratio is challenging, and
classical preparation methods such as washcoating, deposition,
and precipitation have proven inadequate, resulting in strong
agglomeration of catalyst particles and poor dispersion. In con-
trast, the novel complexation–decomposition route reported in
3
48 K for 3 h, then at 393 K for 15 h, and finally at 673 K for 2 h
−1
at 2.5 K min . Before catalytic testing, the catalytic layer was
activated by thermal treatment at 723 K for 1 h (5 K min ) un-
−
1
−
1
der a helium-hydrogen mixture (10% H2, 50 mL min total).
For catalytic testing, macroporous silicon membranes were
glued with epoxy into stainless steel washers (external diam-
eter 19 mm, internal diameter 8 mm), which were then in-
corporated within a stainless steel housing inside a furnace

230
J. Llorca et al. / Journal of Catalysis 255 (2008) 228–233
Fig. 3. Bright field and high resolution transmission electron microscopy
(
HRTEM) images of the catalyst film after activation.
Fig. 2b corresponds to macroporous Si after deposition of
the catalyst layer by this method. The catalyst coating was about
00 nm thick and was well dispersed and adhered over the
Fig. 2. SEM images of a single microchannel before (a) and after (b) catalyst
deposition.
1
channel interior. SEM and EDX analysis revealed that the cat-
alytic coating consisted of a homogeneous, fine-grained layer
of intermixed ZnO and Co O particles. Along with this well-
this study produced adequate results. Urea served both as a
complexating agent and as a basic medium during the decom-
position of complexes at increasing calcination temperature. In
a separate experiment (without macroporous silicon), the pres-
ence of intermediate complexes was inferred from infrared (IR)
spectroscopy. IR spectra after drying at 393 K (before decom-
position) showed absorption bands at 3000–3500, 2200–2400,
3
4
dispersed layer, occasional ZnO crystallites about 1 µm in size
also were present. Considering a channel diameter of 3.3 µm
and a mean catalyst layer thickness of 0.1 µm, we can deduce
a microchannel surface area after catalyst coating of ca. 1.3 ×
6
2
−3
10 m m . After activation under H , Co O was partially re-
2
3
4
−1
1
763, 1680, 1420, 1145, 1050, 1025, 825, and 775 cm . Bands
duced to metallic cobalt, as determined by XRD and HRTEM.
Fig. 3 shows representative low- and high-magnification TEM
images of the catalyst film after activation. Lattice fringes cor-
responding to ZnO, metallic Co, and Co O can be identified.
−
1
in wavenumber range 3000–3500 cm were assigned to the
stretching vibrations of hydroxyl ions. The absorption bands at
−1
2
200–2400 cm were attributed to the existence of interca-
3
4
lated NH3 species [22] and/or cyanate ion [23]. The broad band
Cobalt particles of ca. 10–20 nm diameter are well mixed with
ZnO and partially covered by a layer of cobalt oxide. The redox
pair Co /Co is an active species for ethanol steam reforming
[24,25].
The effective area of the macroporous Si membranes for cat-
alytic testing after assembly with stainless steel washers was
−1
2−
at 1420 cm is characteristic of the CO3 ion, whereas bands
at lower wavenumbers are unspecific and may be assigned to
various species, including coordinated carbonate, nitrate, and
hydroxyl groups. The presence of carbonate and cyanate ions
should not be surprising, because these have been known to
result from the heating of urea solutions [22,23]. The band at
δ+
0
2
6
38 mm , which means about 1.5 × 10 microchannels. Taking
into account the pitch distance between channels of 4.0 µm and
the membrane thickness of 210 µm, this huge number of mi-
−1
1
763 cm was attributed to the stretching mode of carbonyl in
coordinated urea. These bands shifted (probably due to the for-
mation of dimer and polymeric species) [23] and vanished on
heating. Thermal decomposition was followed by TGA, with
the main weight loss occurring at 453–533 K. After calcination
at 673 K, only ZnO and Co3O4 were identified by FTIR as well
as by XRD.
2
crochannels yields a total inner surface of 30 cm with only ca.
10 g of total catalyst loading, and a specific inner surface,
per m , of 3.5 × 10 m . This value is as high as those re-
ported in high-surface area porous materials prepared by flame
pyrolysis, sol–gel techniques, templates, or supramolecular as-
−
3
3
5
2

J. Llorca et al. / Journal of Catalysis 255 (2008) 228–233
231
In contrast, when the microchannels of the Si membrane were
loaded with catalyst, the production rate of hydrogen increased
significantly, particularly when the reaction was carried out
−
1
with S/C = 3. In this case, up to 5.72 µL of H2 s (0.015 mmol
−
1
H2 min ) was obtained at 773 K. The rate of hydrogen pro-
duction obtained with S/C = 1.5 (stoichiometric value for the
reforming reaction [Eq. (1)]) was lower than that obtained with
S/C = 3 at the same gaseous ethanol inlet rate and similar con-
version. This can be explained by the distribution of products
obtained in each case (Table 1). When the reaction was carried
out with the stoichiometric mixture, ethanol transformed into
H , CO , CO, and minor amounts of CH . In addition, the ratio
2
2
4
CO/CO2 increased with increasing temperature. This is in ac-
cordance with a reaction scheme in which first ethanol reforms
into a mixture of H2 and CO [Eq. (3)], and then CO transforms
into CO2 through the water–gas shift (WGS) reaction [Eq. (4)]:
C2H5OH + H2O → 4H2 + 2CO
and
(3)
Fig. 4. Rate of hydrogen production from ethanol steam reforming at increas-
ing temperature obtained over macroporous Si before (1) and after coating the
−
1
microchannels with catalyst film (!, P). 0.24 mL min C H OH, GHSV =
CO + H2O → CO2 + H2.
(4)
2
5
6
−1
1
.5–2 × 10 h , S/C = 3 (1, !), S/C = 1.5 (P).
In this case, an increase in the reaction temperature resulted
in an increase in ethanol conversion and thus also of hydro-
gen production (Fig. 4), along with in an increase in the relative
amount of CO due to the reverse WGS reaction (Table 1), which
is favored at increasing temperature [27]. In contrast, when the
reaction was performed with S/C = 3, the distribution of prod-
ucts indicated that the steam reforming of ethanol was more
selective toward the final reforming products, and a hydrogen-
rich stream composed of 71–73 vol% H2, 23–24 vol% CO2,
and 3–5 vol% CH4 was obtained on a dry basis (Table 1). The
absence of CO among the reaction products is likely due to a
higher partial pressure of water, which has a strong influence
on the WGS reaction [28], and methane likely was produced at
low temperature from the decomposition of ethanol or acetalde-
hyde. The activation energy of the reaction calculated from the
rates reported in Table 1 and additional experiments conducted
semblies [26]; however, the regular arrangement of parallel mi-
crochannels with exactly the same dimensions and high aspect
ratio makes the structure of macroporous silicon membranes
unique for microreaction technology. Our catalytic tests were
performed at a residence time of about 1.9–3.2 ms per mi-
crochannel and a total equivalent gas hourly space velocity
6
−1
(
GHSV) of about 1.5–2 × 10 h
.
Fig. 4 shows the production rate of hydrogen obtained in the
catalytic tests using C2H5OH–H2O mixtures diluted in inert gas
at increasing temperature under different S/C ratios and GHSV
values, as well as the results obtained in a blank experiment (Si
membrane assembled with stainless steel washers with no cat-
alyst coating). It is evident that the hydrogen production rate in
the blank experiment was much lower than that of the Si mem-
brane with microchannels coated with catalyst. The products
obtained in the blank run were mainly H2, CO, and CH4 in sim-
ilar amounts, indicating that the main process occurring in the
absence of catalyst was the thermal decomposition of ethanol
through
−1
at lower conversions (see below) is 59 ± 2 kJ mol . Unfor-
tunately, performing a direct comparison with reaction rates
obtained in conventional fixed-bed reactors loaded with pow-
dered ZnO-supported cobalt catalysts [17,18] is not possible,
because the ethanol conversion reported is total in the tempera-
ture range used in this work.
C2H5OH → CH4 + CO + H2.
(2)
Table 1
−
1
Catalytic results obtained at atmospheric pressure with a gaseous stream containing C H OH (0.24 mL min ) and H O diluted in He. Standard gas-flow velocity
2
5
2
−
1
6
−1
)
0.06–0.1 m s (GHSV = 1.5–2 × 10 h
a
C H OH:H O
T
Activity
C H OH
2 5
Selectivity (%)
2
5
2
(K)
(mol C H OH/mol Co h)
conversion (%)
2
5
H
2
CO
CO
CH
2
4
1:6
698
16.8
26.4
37.7
45.9
33.2
52.2
74.6
90.7
69.3
70.1
71.1
73.4
24.3
24.9
24.2
23.2
–
–
–
–
6.4
5.0
4.7
3.4
723
748
773
1
:3
a
698
16.4
22.9
33.6
41.8
32.5
45.3
66.5
82.6
70.5
69.0
68.1
67.3
23.4
23.8
17.5
15.3
6.1
7.2
11.6
13.7
–
–
2.8
3.7
723
748
773
Molar percentage of products (water excluded).

232
J. Llorca et al. / Journal of Catalysis 255 (2008) 228–233
surface cobalt per s) was estimated using the Co particle size
−1
determined by TEM. At 773 K, the TOF was ca. 1.1 s
. Conclusion
Macroporous silicon is envisioned as a new and excellent
.
4
micromonolithic support for catalysts in chemical microreac-
tors. In this work, we have demonstrated the high efficiency of
macroporous silicon membranes for producing hydrogen from
ethanol steam reforming at moderate temperature after coat-
ing the microchannels with a thin layer of an appropriate cat-
alyst (Co3O4–ZnO) by a novel complexation–decomposition
route. High and stable selectivity toward the reforming prod-
ucts, H2 and CO2, with >73 vol% H2 was attained at 773 K
and S/C = 3 at contact times of <5 ms per channel. Micro-fuel
cells, which are expected to be involved in the energy supply
for portable electronic devices in the future, may benefit from
the microtechnology derived from macroporous silicon-based
microreformers that generate hydrogen.
(a)
Acknowledgments
This work was supported by MEC projects ENE2006-
0
6925, TEC2004-06854-C03-03/MIC, and TEC2005-02716/
MIC. T.T. thanks the Spanish Government for a Juan de la
Cierva scholarship.
References
[1] D.R. Palo, J.D. Holladay, R.T. Rozmiarek, C.E. Guzman-Leong, Y. Wang,
J. Hu, Y.-H. Chin, R.A. Dagle, E.G. Baker, J. Power Sources 108 (2002)
28.
[2] L.F. Brown, Int. J. Hydrogen Energy 26 (2001) 381.
[3] M.A. Peña, J.P. Gómez, J.L.G. Fierro, Appl. Catal. A 144 (1996) 7.
[4] V. Twigg, S. Spencer, Top. Catal. 22 (2003) 191.
(b)
[5] P.D. Vaidya, A.E. Rodrigues, Chem. Eng. J. 117 (2006) 39.
[6] A. Haryanto, S. Fernando, N. Murali, S. Adhikari, Energy Fuels 19 (2005)
Fig. 5. (a) Variation of hydrogen partial pressure (m/z = 2) in a temperature
2098.
pulse experiment. (b) Product distribution (molar fraction) obtained at 773 K.
[7] J. Llorca, N. Homs, J. Sales, P. Ramírez de la Piscina, J. Catal. 209 (2002)
306.
8] J.A. Torres, J. Llorca, A. Casanovas, M. Domínguez, J. Salvadó, D. Mon-
tané, J. Power Sources 169 (2007) 158.
−
1
6
−1
.
S/C = 1.5, 0.24 mL min C H OH, GHSV = 1.5 × 10 h
2
5
[
We carried out several experiments to evaluate the response
and stability of the microreformer. We applied temperature
pulses under steady-state conditions, and found that the gen-
eration of hydrogen was completely reversible at each temper-
ature (Fig. 5a). Further, we detected no significant variations in
product distribution during stability tests conducted at different
temperatures (Fig. 5b). A micro-fuel cell capable of running a
portable device of several watts requires a hydrogen flow of sev-
[
9] W. Ehrfeld, V. Hessel, H. Löwe, Microreactors—New Technology for
Modern Chemistry, Wiley–VCH, Weinheim, 2000.
[10] T. Kim, J.S. Hwang, S. Kwon, Lab Chip 7 (2007) 835.
[11] Y. Kawamura, N. Ogura, T. Yamamoto, A. Igarashi, Chem. Eng. Sci. 61
(2006) 1092.
[
[
12] A.V. Pattekar, M.V. Kothare, J. Microelectromech. Syst. 13 (2004) 7.
13] J.D. Holladay, E.O. Jones, M. Phelps, J. Hu, J. Power Sources 108 (2002)
21.
[14] G.G. Park, D.J. Seo, S.H. Park, Y.G. Yoon, C.S. Kim, W.L. Toon, Chem.
Eng. J. 101 (2004) 87.
15] V. Cominos, S. Hardt, V. Hessel, G. Kolb, H. Löwe, M. Wichert, R. Zapf,
Chem. Eng. Commun. 192 (2005) 685.
16] Y. Men, G. Kolb, R. Zapf, V. Hessel, H. Löwe, Process Saf. Environ.
Prot. 85 (2007) 413.
[17] J. Llorca, P. Ramírez de la Piscina, J.A. Dalmon, J. Sales, N. Homs, Appl.
Catal. B 43 (2003) 355.
18] J. Llorca, N. Homs, J. Sales, J.L.G. Fierro, P. Ramírez de la Piscina,
J. Catal. 222 (2004) 470.
19] M.A. Urbiztondo, E. Valera, T. Trifonov, R. Alcubilla, S. Irusta, M.P. Pina,
A. Rodríguez, J. Santamaría, J. Catal. 250 (2007) 190.
−1
eral mL min . This can be achieved with our microreformer
by increasing the amount of ethanol in the reactant mixture.
When a C2H5OH:H2O = 1:6 molar mixture was supplied di-
[
[
−
1
rectly from a syringe pump (total liquid flow, 0.1 mL min
;
4
−1
−1
LHSV = 2 × 10 h ), up to 1.2 mL min H2 (0.054 mmol
−1
H2 min ) was obtained at 773 K. In this case, the overall
ethanol conversion was low (ca. 1.5%), but higher values could
be readily obtained by stacking several Si membranes. (The
macroporous Si membrane is only 0.21 mm thick.) A turnover
frequency (TOF; number of hydrogen molecules produced per
[
[
[20] V. Lehmann, J. Electrochem. Soc. 140 (1993) 2836.

J. Llorca et al. / Journal of Catalysis 255 (2008) 228–233
233
[
[
21] G. Barillaro, F. Pieri, J. Appl. Phys. 97 (2005) 116105.
22] P. Jeevanandam, Y. Koltypin, A. Gedanken, Y. Mastai, J. Mater. Chem. 10
2000) 511.
Mater. 16 (2004) 3573.
[26] I.-K. Sung, C.M. Mitchell, D.-P. Kim, P.J.A. Kenis, Adv. Funct. Mater. 15
(
(2005) 1336.
[
[
23] T. Ishikawa, E. Matijevic, Colloid Polym. Sci. 269 (1991) 179.
24] J. Llorca, J.A. Dalmon, P. Ramírez de la Piscina, N. Homs, Appl. Catal.
A 243 (2003) 261.
[27] I. Fishtik, A. Alexander, R. Datta, D. Geana, Int. J. Hydrogen Energy 25
(2000) 31.
[28] Y. Tanaka, T. Utaka, R. Kikuchi, K. Sasaki, K. Eguchi, Appl. Catal. A 238
(2003) 11.
[25] J. Llorca, P. Ramírez de la Piscina, J.A. Dalmon, N. Homs, Chem.