Sustainable H2 gas production by photocatalysis

Article abstract of DOI:10.1016/j.jphotochem.2010.06.022

The photocatalytic, ambient temperature, reforming of oxy-hydrocarbons, used to generate hydrogen gas, has been studied in some detail over Pd-titania systems and a general mechanism is developed. In the case of water-methanol mixtures, methanol adsorption on the metal surface results in hydrogen gas evolution and the formation of adsorbed CO. The latter is subsequently oxidised by an oxygen species (O^-) generated in the semiconductor by UV light. Desorption of the resulting CO₂ enables another methanol molecule to adsorb, restarting the cycle. Water is reduced at the titania hole replacing the lost anion and generating further hydrogen. We will discuss the evidence for this mechanism and extend it to consider the photocatalysis of higher alcohols including ethanol, butanol and glycerol.

Full text of DOI:10.1016/j.jphotochem.2010.06.022

Journal of Photochemistry and Photobiology A: Chemistry 216 (2010) 115–118
Contents lists available at ScienceDirect
Journal of Photochemistry and Photobiology A:
Chemistry
journal homepage: www.elsevier.com/locate/jphotochem
Sustainable H2 gas production by photocatalysis
Hasliza Bahruji∗, Michael Bowker, Philip R. Davies, Layla Saeed Al-Mazroai, Amanda Dickinson,
Jane Greaves, David James, Lucy Millard, Fabien Pedrono
Wolfson Nanoscience Laboratory and Cardiff Catalysis Institute, School of Chemistry, Cardiff University, Main Building, Cardiff, CF10 3AT, UK
a r t i c l e i n f o
a b s t r a c t
Article history:
Available online 23 June 2010
The photocatalytic, ambient temperature, reforming of oxy-hydrocarbons, used to generate hydrogen
gas, has been studied in some detail over Pd-titania systems and a general mechanism is developed. In
the case of water–methanol mixtures, methanol adsorption on the metal surface results in hydrogen gas
evolution and the formation of adsorbed CO. The latter is subsequently oxidised by an oxygen species (O−)
generated in the semiconductor by UV light. Desorption of the resulting CO2 enables another methanol
molecule to adsorb, restarting the cycle. Water is reduced at the titania hole replacing the lost anion and
generating further hydrogen. We will discuss the evidence for this mechanism and extend it to consider
the photocatalysis of higher alcohols including ethanol, butanol and glycerol.
Keywords:
Photocatalysis
Hydrogen production
Water splitting
Methanol
Alcohol reforming
Titania-Pd catalysts
© 2010 Elsevier B.V. All rights reserved.
1. Introduction
the surface. This is further dehydrogenated to CO and gas phase
hydrogen. The aim of this paper is to explore this proposed mech-
The use of fossil fuels as the main energy source in modern
society not only brings benefits of convenience and an estab-
lished technology, but also has the well-known disadvantage of
CO2 release. This gas which is produced by most combustion pro-
cesses is believed to be the main culprit involved in global warming
and shows a significant percentage increase since global industri-
alisation [1]. An attractive alternative to fossil fuels as an energy
source is H2 which is widely considered to be the most promis-
ing clean energy carrier for the future and its use is being explored
in many areas of technology [2]. The fundamental problem that
underlies this approach however is finding an environmentally
friendly source of hydrogen.
Water is an obvious source for both hydrogen and oxygen and
there is therefore a great deal of interest in methods by which the
individual components might be economically obtained. The use
of photocatalysis for water splitting to generate H2 has risen in
interest over the past few decades. Previously we [3], and oth-
ers [4], have reported on hydrogen production using ambient
temperature reforming of methanol and glycerol on a variety of
titania-supported catalysts. A mechanism for methanol reforming
over Pd modified TiO2 catalysts was reported earlier [5] in which
the precious metal provides the active site for alcohol dehydro-
genation whilst the titania acts as a photon absorber creating an
active oxidising species. The dehydrogenation of alcohols on Pd
surfaces involves O–H bond cleavage leaving an alkoxy group on
anism in more detail by comparing the photocatalytic reactivity of
a number of primary, secondary and tertiary alcohols: ethanol, 1-
propanol, 1-butanol, cyclohexanol and glycerol and glucose. From
the observed activity of these molecules we deduce further details
of the mechanism.
2. Experimental
The reaction vessel consists of a Pyrex reaction flask with a
purge line and a septum for sampling. The catalyst was prepared
using the incipient wetness impregnation method in which the
0.01 g of PdCl2 (99.9%, Sigma–Aldrich) dissolved in 1.5 ml of dis-
tilled water acidified with one drop of HNO3 was added drop wise
into the 2 g of titania (P25, Degussa) and thoroughly grind. This
was heated at 200 ◦C for 2 h followed by calcination at 500 ◦C in
air oven for 3 h. The catalysts used here had a loading of 0.5 wt%
Pd with respect to the support, this loading being shown previ-
ously to give optimum photocatalytic yields [5]. The as-prepared
catalyst was ground in a pestle and mortar and sieved to an aggre-
gate size of less than 53 ␮m. The reaction mixture, consisting of
0.2 g catalysts and 100 ml deionised water, were placed in the flask
and purged with Ar for 30 min to evacuate dissolved gases. 100 ␮l
of alcohols – methanol (Sigma–Aldrich 99.8%), ethanol, 1-propanol,
1-butanol, 2-butanol, 3-butanol and glycerol (Fisher Scientific, 99%)
2-propanol (Acros Organic 99%), and 0.5 g of d-glucose (Analar) and
0.5 g sucrose (FSA Laboratory Supplies) were added respectively,
and the mixture was stirred with a magnetic stirrer and illumi-
nated with a 400 W Xe arc lamp (Oriel model 66084) from the side
of the flask. Samples of 0.2 ml gas were extracted by syringe and
∗
Corresponding author. Tel.: +44 (0) 29 2087 4072.
E-mail address: bahrujih@cardiff.ac.uk (H. Bahruji).
1010-6030/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.jphotochem.2010.06.022

116
H. Bahruji et al. / Journal of Photochemistry and Photobiology A: Chemistry 216 (2010) 115–118
Fig. 2. A plot of H2 production from the photocatalytic reforming of related
molecules with alcohol functional group: ring alcohol (cyclohexanol), aromatic alco-
hol (benzyl alcohol), and sugar (d-glucose, sucrose and glycerol).
Fig. 1. A plot of H2 production from the photocatalytic reforming of alcohols: pri-
mary alcohol (methanol, ethanol, 1-propanol and 1-butanol), secondary alcohol
(2-propanol and 2-butanol) and tertiary alcohol (3-butanol).
an excellent hydrogenation catalyst and can carry out such reac-
tions at low temperature [7]. We suggest here that the dehydration
reaction is possibly driven by photo-stimulated reaction followed
by hydrogenation of the alkene, due to the availability of hydrogen
in the reactor from the photo-driven reforming reaction.
analysed using a Varian 3300 gas chromatograph with a thermal
conductivity detector and MS 13X column, 2 m long. Calibration of
the hydrogen gas volume measurement within the photoreactor
showed an error of approximately 0.8 ml. For identification of the
by-products, 0.2 ml of gas samples after 3 h of reaction respectively
were analysed by injection into a sampling system attached to a
Hiden Analytical quadrupole Hal 201 mass spectrometer.
H2 production from molecules with a hydroxyl group on the
structure and somewhat more complex alcohols (cyclohexanol,
benzyl alcohol, glycerol, d-glucose and sucrose) are shown in
Fig. 2. 3-Butanol and benzyl alcohol produce very little hydrogen,
whilst cyclohexanol gives 2 ml of hydrogen after 3 h of reaction.
Sugar-type alcohols d-glucose and sucrose show good reactivity in
producing hydrogen with 8 ml and 4 ml in 3 h respectively and glyc-
erol gives 12 ml in 3 h. From all of these data there is evidence to
suggest that an important linking factor is the presence of hydro-
gen atoms in the molecule at the alpha carbon (␣) position but
the absence of any significant hydrogen from benzyl alcohol indi-
cates that whilst the presence of an ␣-hydrogen may be a necessary
condition for reaction it is not sufficient. We return to this point
below.
That the rate of hydrogen production does not increase when
long chain alcohols were employed implies that the hydrogen gen-
erated does not come from the liberation of all the hydrogen atoms
in the alcohol. Instead, the amount of H2 produced can be esti-
mated by the number of ␣-hydrogen atoms present. In methanol
and glycerol reforming, for example, where all the carbon atoms
possess ␣-hydrogens, complete dehydrogenation occurs on the Pd,
leaving absorbed CO on the surface, and converting all the hydrogen
atoms into gas phase molecular hydrogen; the rate of hydrogen pro-
duction from glycerol reforming is therefore approximately twice
that for methanol [8]. d-Glucose follows the prediction with 8 ml of
hydrogen produce in 3 h. However, the rate does not depend solely
on the number of ␣-hydrogens since sucrose with approximately
twice the number of ␣-hydrogens as d-glucose produces only 5 ml
of hydrogen in 3 h. Ethanol has ␤-hydrogens, but these are not
lost in the photocatalysis, instead C–C bond-breaking occurs as evi-
denced by the production of methane during the reaction, together
with the CO2 and H2. Based on the number of ␣-hydrogens in
ethanol and the presence of methane in the products, we estimate
that 2 moles of H2 should be produced from the photoreforming of
ethanol–water mixtures as written in Eq. (1). However, Fig. 1 shows
that the rate is not significantly different from that of methanol
(3 mol of H2, Eq. (2))
3. Results and discussion
Fig. 1 shows H2 production from a series of primary alcohols
(C1–C4), secondary alcohols (2-propanol and 2-butanol) and a ter-
tiary alcohol (3-butanol) over a 0.5% Pd/TiO2 photocatalyst. The
hydrogen production rate from the primary and secondary alco-
hols is fairly similar over 3 h of reaction. Note that in the absence
of a photocatalyst or UV light irradiation, no hydrogen was pro-
duced. Further analysis of the gas phase was carried out after 3 h
of reaction using mass spectrometry. This showed that ethanol
produces methane as the only major by-product, 1-propanol pro-
duces ethane and 2-propanol produces methane. 1-Butanol and
2-butanol produce propane and ethane respectively. However for
the C3 and C4 alcohols, the alkane derivative of its parent alco-
hol was also observed with propane being obtained from both
1-propanol and 2-propanol and butane from both 1-butanol and
2-butanol. A summary of the products obtained from each reaction
is given in Table 1. The dehydration of alcohols, and in particular
propanol, is frequently used as a probe of the acid–base properties
of metal oxide surfaces: acid sites catalysing dehydration to pro-
duce propene whereas basic sites lead to acetone formation. With
titania, propene is the dominant product of propanol reaction at
400 K [6]. In the photocatalytic reforming reaction only trace quan-
tities of propene were observed indicating that the production of
propane is due to further hydrogenation at the metal sites. Pd is
Table 1
Products of the photocatalytic reactions of different alcohols over 0.5 wt% Pd TiO2
catalysts determined by mass spectrometry.
Alcohol
Products
Methanol
Ethanol
H2, CO2
H2, CO2, CH4
1-Propanol
2-Propanol
1-Butanol
2-Butanol
H2, CO2, C2H6, C3H8
H2, CO2, CH4, C3H8
H2, CO2, C3H8, C4H10
H2, CO2, C2H6, C4H10
C2H5OH + H2O → CO2 + 2H2 + CH4
CH3OH + H2O → CO2 + 3H2
(1)
(2)

H. Bahruji et al. / Journal of Photochemistry and Photobiology A: Chemistry 216 (2010) 115–118
117
Scheme 2. Proposed mechanistic steps for secondary alcohol photoreforming via
beta-hydride elimination.
CO2 and H2. Studies carried out by Davis and Barteau on primary
alcohol decomposition on Pd surface showed that the decomposi-
tion occurs by the elimination of CO to form a hydrocarbon one
carbon atom shorter than the parent alcohol chain [11]. These
primary alcohols undergo stoichiometric decarbonylation on the
Pd surface which involves ␣ C–C bond scission [12,13]. The TiO2
semiconductor facilitates this process by providing energy and an
oxidiser species for restarting the cycle. There is some evidence,
in the case of the longer primary alcohols that the Pd surface is
poisoned by C species [11]; this might explain the decreasing H2
production rate in the present results as the number of carbons
in the chain increases (ethanol – 6.08 × 10−4 ml s−1, 1-butanol –
3.49 × 10−4 ml s−1).
Scheme 1. Proposed mechanism for the photoreforming of primary alcohols.
Secondary alcohols can produce H2 by two possible mech-
anistic pathways. The absorbed alkoxide could decompose via
non-selective decomposition yielding (in the case of 2-propanol)
two molecules of methane, H2 and CO2 (step (13-a), Scheme 2).
Alternatively, the molecule could decompose by a beta-hydride
elimination mechanism involving a ketone intermediate to get sim-
ilar products [14]. Shekhar and coworkers showed that 2-propanol
decomposes on palladium to produce CO, methane, hydrogen and
surface carbon. However, methane desorption from 2-propanol
was observed to start at 345 K, nearly 50 K higher than for ethanol
decomposition, suggesting that C–C scission in the secondary alco-
hol exhibits a higher activation barrier than that for a primary
alcohol [12]. The importance of ␣ C–C bond scission in longer
chain alcohol reactions is further emphasised by the reaction of
benzyl alcohol which failed to give significant quantities of hydro-
gen. Although the benzyl alcohol possesses two ␣ hydrogen atoms,
the ␣ C–C bond is attached to an aromatic ring and is expected to
have much higher dissociation barrier (∼427 kJ mol−1) than that of
aliphatic C–C bonds (350–370 kJ/mol) [15].
The second mechanistic pathway by which 2-propanol decom-
position might occur is beta-hydride elimination (Scheme 2); after
dehydration of the alcohol the resulting 2-propoxide eliminates a
second hydrogen atom (step (13-b(i))) to form adsorbed acetone.
No acetone was observed as a product of 2-propanol in the present
experiments. However, in a study to be published separately we
have observed the photocatalysed decomposition of acetone into
CO2 and methane on palladium (steps (13-b(ii)) and (14)).
It was stated above that the critical factor for generating hydro-
gen from water in the presence of alcohol is the availability of
a strongly bonded CO molecule on the Pd surface which facili-
tates reduction on the titania. The CO on the Pd surface acts as a
hole scavenger and prevents recombination of the photogenerated
electron–hole pairs. Once the electrophilic oxygen species is con-
sumed in oxidising the CO, the resulting oxygen vacancy on the
titania acts as a reducing agent, decomposing water into hydro-
gen. Therefore, the rate of alcohol dissociation on metal surfaces
at ambient temperature can be used to determine the reactivity in
producing hydrogen gas as well as the reaction rate.
According to Gates et al., the rate determining step in alkoxide
decomposition on palladium surfaces is the elimination of the
␤-hydrogen in the alkoxide [9] (where “␤” refers to the posi-
tion relative to the palladium and is therefore equivalent to the
␣-hydrogens discussed above). This may account for the similar
rate in methanol and ethanol decomposition on Pd surface and
hence similar H2 production rate. Further identification of the by-
products from the primary alcohol reactions indicates the presence
of alkane groups with the loss of one carbon from the parent alcohol
(for example, ethane from 1-propanol, propane from 1-butanol).
The tertiary alcohol has no ␣-hydrogen and therefore shows little
hydrogen production. However, although benzyl alcohol possesses
two ␣-hydrogens, it shows no reactivity in 3 h; this apparent con-
tradiction is discussed further below.
Based on the detailed mechanism previously reported for
methanol and glycerol on Pd/TiO2 catalysts [3,8], the present find-
ings are used to generalise the mechanism for longer chain primary
alcohols as summarised in Scheme 1.
The first step in ethanol photocatalytic reforming involves the
dissociation of the hydroxyl on the palladium leaving an ethoxy
group and an adsorbed H. Studies previously carried out on ethanol
oxidation on Pd surfaces at room temperature, have shown that the
major pathway for ethanol decomposition is via surface ethoxy fol-
lowed by C–C bond-breakage to produce a methyl group, carbon
monoxide and hydrogen adatoms [10]. The methyl groups pro-
duce methane by recombination with hydrogen atoms absorbed
on the Pd surface. This however leaves a surface poisoned by a
strongly adsorbed CO which is only removed with light irradiation.
Electrophilic O− species generated on titania migrate to oxidise
the CO to CO2. The vacant site on the metal is then replaced by
another alcohol molecule liberating another mole of hydrogen gas.
The oxygen vacancy on the titania is filled by an oxygen supplied
by the decomposition of a water molecule since the experiment
was carried out under anaerobic conditions. Longer chain primary
alcohols, 1-propanol and 1-butanol have been studied and produce
mainly ethane and propane respectively as by-products alongside

118
H. Bahruji et al. / Journal of Photochemistry and Photobiology A: Chemistry 216 (2010) 115–118
4. Conclusion
[3] L.S. Al-Mazroai, M. Bowker, P. Davies, A. Dickinson, J. Greaves, D. James, L. Mil-
lard, The photocatalytic reforming of methanol, Catalysis Today 122 (2007)
46–50.
Photocatalytic reforming of alcohols can offer sustainable
hydrogen gas production at ambient temperatures when bio-fuels
are used as sacrificial reductants. The main characteristic of an alco-
hol that determines its reactivity is the availability of ␣-hydrogens
and the ease of dissociation of the ␣ C–C bond. The decomposition
mechanism for alcohols on the Pd surface is similar for primary
and secondary alcohol involving O–H bond scission and ␣ C–C
bond cleavage. Branched alcohol decomposition probably involves
a ketone intermediate before it reforms into an alkane, H2 and
absorbed CO on the Pd surface. The adsorbed CO effectively under-
goes a photocatalytic water-gas-shift reaction, splitting water and
producing more hydrogen.
[4] Y. Sasaki, A. Iwase, H. Kato, A. Kudo, The effect of co-catalyst for Z-scheme
photocatalysis systems with an Fe3+/Fe2+ electron mediator on overall water
splitting under visible light irradiation, Journal of Catalysis 259 (2008) 133–137.
[5] A. Dickinson, D. James, N. Perkins, T. Cassidy, M. Bowker, The photocatalytic
reforming of methanol, Journal of Molecular Catalysis A: Chemical 146 (1999)
211–221.
[6] D. Haffad, A. Chambellan, J.C. Lavalley, Propan-2-ol transformation on simple
metal oxides TiO2, ZrO2 and CeO2, Journal of Molecular Catalysis A: Chemical
168 (2001) 153–164.
[7] A.M. Doyle, S.K. Shaikhutdinov, H.J. Freund, Alkene chemistry on the palla-
dium surface: nanoparticles vs single crystals, Journal of Catalysis 223 (2004)
444–453.
[8] M. Bowker, P.R. Davies, L.S. Al-Mazroai, Photocatalytic reforming of glycerol
over gold and palladium as an alternative fuel source, Catalysis Letters 128
(2009) 253–255.
[9] S.M. Gates, J.N. Russell Jr., J.T. Yates Jr., Scanning kinetic spectroscopy (SKS):
a new method for investigation of surface reaction processes, Surface Science
159 (1985) 233–255.
[10] M. Bowker, R.P. Holroyd, R.G. Sharpe, J.S. Corneille, S.M. Francis, D.W. Goodman,
Molecular beam studies of ethanol oxidation on Pd(1 1 0), Surface Science 370
(1997) 113–124.
Acknowledgements
The authors acknowledge Ministry of Higher Education
Malaysia and University Pendidikan Sultan Idris (UPSI) for financial
sponsorship to H. Bahruji.
[11] J.L. Davis, M.A. Barteau, Decarbonylation and decomposition pathways of alco-
hol’s on Pd(1 1 1), Surface Science 187 (1987) 387–406.
[12] R. Shekhar, M.A. Barteau, Structure sensitivity of alcohol reactions on (1 1 0)
and (1 1 1) palladium surfaces, Catalysis Letters 31 (1995) 221–237.
[13] J.L. Davis, M.A. Barteau, Spectroscopic identification of alkoxide, aldehyde, and
acyl intermediates in alcohol decomposition on Pd(1 1 1), Surface Science 235
(1990) 235–248.
References
[1] M.A.L. Caetano, D.F.M. Gherardi, T. Yoneyama, Optimal resource management
control for CO2 emission and reduction of the greenhouse effect, Ecological
Modelling 213 (2008) 119–126.
[2] D.A.J. Rand, R.M. Dell, Hydrogen Energy, Challenges and Prospects, RSC Pub-
lishing, 2008.
[14] S. Hoang, M. Pan, C.B. Mullins, Surface chemistry of 2-propanol on clean
and oxygen precovered Ir(1 1 1), Journal of Physical Chemistry C 113 (2009)
21745–21754.
[15] T.L. Cottrell, The Strengths of Chemical Bonds, Butterworths Scientific Publica-
tions, London, 1954.