Hydrogen production by photocatalytic steam reforming of methanol on noble metal-modified TiO2

Article abstract of DOI:10.1016/j.jcat.2010.05.012

The photocatalytic production of hydrogen by methanol steam reforming was studied over a series of pristine or noble metal (Ag, Au, Au-Ag alloy and Pt) - modified TiO₂ photocatalysts, synthesised by flame spray pyrolysis or by the deposition of preformed noble metal nanoparticles on TiO₂. A closed recirculation apparatus was employed, with the photocatalyst bed continuously fed with methanol/water vapours. Methanol underwent oxidation up to CO₂ through the formation of formaldehyde and formic acid. Carbon monoxide, methane, methyl formate, acetaldehyde and dimethyl ether were identified as side products. Hydrogen evolved at constant rate, which significantly increased upon noble metal addition, Pt being the most effective co-catalyst, followed by gold and silver, according to their work function values. A systematic investigation into the effects of the inlet gas composition gave valuable information on the prevailing reaction paths and on the conditions of process optimisation, also in terms of distribution of CH ₃OH oxidation products.

Full text of DOI:10.1016/j.jcat.2010.05.012

Journal of Catalysis 273 (2010) 182–190
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Hydrogen production by photocatalytic steam reforming of methanol on noble
metal-modified TiO2
a
b
a,*
Gian Luca Chiarello , Myriam H. Aguirre , Elena Selli
a
Dipartimento di Chimica Fisica ed Elettrochimica, CIMAINA and ISTM-CNR, Università degli Studi di Milano, Via Golgi 19, I-20133 Milano, Italy
Empa – Swiss Federal Laboratories for Materials Testing and Research Solid State Chemistry and Catalysis, Ueberlandstrasse 129, CH-8600 Duebendorf, Switzerland
b
a r t i c l e i n f o
a b s t r a c t
Article history:
The photocatalytic production of hydrogen by methanol steam reforming was studied over a series of
Received 9 March 2010
Revised 22 April 2010
Accepted 21 May 2010
Available online 29 June 2010
pristine or noble metal (Ag, Au, Au–Ag alloy and Pt) – modified TiO
spray pyrolysis or by the deposition of preformed noble metal nanoparticles on TiO
tion apparatus was employed, with the photocatalyst bed continuously fed with methanol/water
vapours. Methanol underwent oxidation up to CO through the formation of formaldehyde and formic
2
photocatalysts, synthesised by flame
2
. A closed recircula-
2
acid. Carbon monoxide, methane, methyl formate, acetaldehyde and dimethyl ether were identified as
side products. Hydrogen evolved at constant rate, which significantly increased upon noble metal addi-
tion, Pt being the most effective co-catalyst, followed by gold and silver, according to their work function
values. A systematic investigation into the effects of the inlet gas composition gave valuable information
on the prevailing reaction paths and on the conditions of process optimisation, also in terms of distribu-
Keywords:
Photocatalytic H production
2
Photo-steam reforming
Methanol
Flame spray pyrolysis
Noble metals on TiO
2
3
tion of CH OH oxidation products.
Ó 2010 Elsevier Inc. All rights reserved.
1
. Introduction
Hydrogen is widely considered the clean energy vector of the
Because of the endothermic nature of the steam reforming reac-
tion, heat must be supplied for the reaction to proceed, and this is
usually provided by the combustion of part of the feed stock, with a
consequent decrease in the net yield of the process.
The recent growing concern about global climate change,
mainly tied to the exploitation of fossil fuels, led to great efforts
in the development of new alternative, environmentally friendly
energy sources. The photocatalytic production of hydrogen from
photo-reforming of biomass on metal oxide semiconductors can
respond to this urgent need. Indeed, since biomasses are renew-
future. However, nowadays, about two-thirds of the world’s hydro-
gen production is employed in ammonia synthesis to produce fer-
tilizers; large amounts are used in the synthesis of methanol, by
reaction with carbon monoxide, and in the catalytic hydrogenation
of organic compounds [1].
Although the technologies of energy production from hydrogen,
e.g. fuel cells and internal hydrogen combustion engines, are
already mature, hydrogen production remains a major problem.
Indeed, nearly all hydrogen production is still based on fossil raw
materials, and only 4% is produced via water electrolysis. The most
important industrial route of hydrogen production consists in the
catalytic steam–reforming of hydrocarbons, implying gaseous or
vaporised hydrocarbons treatment with steam at high pressure
able and consume atmospheric CO
a small net CO impact compared to fossil fuels.
Photocatalytic reactions on semiconductors are initiated by the
absorption of photons with energy h equal to, or greater than, the
2
during their growth, they have
2
m
semiconductor band gap [2]. This promotes an electron from the
valence band (VB) to the conduction band (CB), with the conse-
ꢁ
þ
(
15–40 bar) and high temperature (650–950 °C) over nickel-based
quent formation of an electron (e )–hole (h ) pair. The so pro-
CB VB
catalysts. In the case of methane, the reaction is as follows:
duced charge carriers can induce the reduction of electron
acceptor species, having a reduction potential lower in energy than
the CB, and the oxidation of electron donor species, having a reduc-
tion potential higher in energy than the VB, respectively. In the
case of the photo-reforming reaction, H ions are the electron
acceptors, and the organic substrates, which are oxidised up to
ꢀ
H ¼ 205 kJ molꢁ1Þ
CH
4
þ H
2
O ! CO þ 3H
2
ð
D
Additional hydrogen can be recovered by a subsequent, lower
temperature ‘‘water gas shift” step, in which steam further oxidises
CO yielding CO
+
2
:
ꢀ
H ¼ ꢁ42 kJ molꢁ1Þ
CO
and can be regarded as a sort of artificial photosynthesis, able to
convert radiation energy into chemical energy, i.e. into H . Con-
2
, are the electron donors. The overall reaction is endoergonic
CO þ H O ! CO
2
2
þ H
2
ðD
2
trarily to traditional steam reforming, the photo-assisted process
occurs at room temperature and atmospheric pressure and no feed
*
Corresponding author. Fax: +39 02 503 14300.
E-mail address: elena.selli@unimi.it (E. Selli).
0
021-9517/$ - see front matter Ó 2010 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2010.05.012

G.L. Chiarello et al. / Journal of Catalysis 273 (2010) 182–190
183
stock needs to be burnt, because the required energy is totally sup-
2. Experimental
plied by photons. With photocatalysts able to absorb solar light
efficiently, the process represents a fascinating way of harvesting
and converting solar energy.
2.1. Photocatalysts preparation
Although different mixed metal oxide semiconductors, also
with rather complex structure, have been proposed as photocat-
alysts in recent years [3–7], titanium dioxide modified by noble
metal nanoparticles still remains the best photocatalyst for
hydrogen production [8,9]. Indeed, the Fermi level of noble met-
als is usually lower in energy than the conduction band energy
The investigated photocatalysts are listed in Table 1. They were
all home-prepared, apart from P25 TiO2 from Degussa, frequently
employed as benchmark in photocatalytic studies. FP-TiO2 and
FP-0.5 wt.%Pt/TiO2 were synthesised in a continuous and single
step by the already described FP method [16], starting from a solu-
tion containing 10 mL of titanium(IV)-isopropoxide dissolved in a
xylene (35 mL)/acetonitrile (5 mL) mixture, also containing the re-
quired amount of platinum precursor (Pt acetylacetonate), when
necessary. The so obtained solution was fed at constant rate
of the TiO
migrate and be captured by the noble metal, whereas photopro-
duced holes remain in the TiO valence band. The rate of photo-
2
semiconductor. Thus, photopromoted electrons can
2
ꢁ1
catalytic hydrogen production is further increased in the
presence of organic compounds able to act as hole scavengers,
undergoing relatively rapid and irreversible oxidation on the
(4 mL min ) to the flame reactor, together with oxygen
ꢁ1
(5 L min ). The solution was so sprayed and ignited by a crown
of twelve methane/oxygen supporting flamelets surrounding the
central nozzle of the reactor, forming a vertical main flame in
which the photocatalyst nanoparticles were produced. The ob-
tained powder was then collected on the electrodes of an electro-
static precipitator.
2
TiO surface. Of course, also, organics deriving from renewable
sources, i.e. biomasses, may be profitably employed as hole scav-
engers. For instance, methanol is widely used as a sacrificial
agent in the liquid-phase photocatalytic production of hydrogen
[
10–13].
However, the photo-steam reforming reaction did not receive
The series of 1.0 (or 0.5) wt.% NM/TiO2 (NM = Ag, Au and Pt)
photocatalysts, as well as the photocatalyst containing an Au/Ag
much attention so far. Steam addition was found to increase meth-
anol conversion in the gas-phase anaerobic photocatalytic oxida-
alloy (i.e. sample 1%Au–1%Ag/TiO ), were synthesised by the depo-
sition of surfactant-stabilised preformed NM nanoparticles on
2
tion of methanol over Fe/TiO
an increase in hydrogen production rate when an Au/TiO
2
[14]. Greaves et al. [13] evidenced
photo-
commercial TiO2 Degussa P25 or on FP-TiO2 (sample 0.5%Pt/FP-
2
TiO ), adopting a reverse micelle method similar to that described
2
catalyst film was illuminated in contact with methanol–water va-
pours, with respect to the same photocatalyst in liquid suspension.
The photocatalytic steam reforming of methane has been recently
by Liu et al. [19]. Chloroauric acid, hexachloroplatinic acid and sil-
ver nitrate were used as NM precursors. Proper amounts of NM
precursor were dissolved in 95 mL of an n-dodecyltrimethylammo-
nium chloride aqueous solution, to obtain a 40:1 surfactant to NM
molar ratio. NM colloidal suspensions were then obtained by add-
ing 5 mL of a NaBH aqueous solution (4:1 NaBH to NM molar ra-
investigated by Yoshida et al. [15] employing Pt/TiO
reactor.
2
in a flow
In the present work, a systematic investigation has been per-
formed on hydrogen production by photocatalytic steam reforming
of methanol, employed as a volatile and simple organic electron
donor acting as sacrificial reagent able to combine with photopro-
duced holes, in a model reaction for photocatalytic hydrogen for-
mation from renewable sources. A series of noble metal (NM)
4
4
tio) under vigorous stirring. The colour turned from yellowish to
deep dark, as a consequence of NM reduction. The proper amount
of TiO2 powder, dispersed in water (50 mL) in an ultrasonic bath,
was then added to the NM colloidal suspension under vigorous
stirring. When stirring was stopped, a coloured powder was pre-
cipitated, whereas the supernatant liquid was perfectly clear, indi-
2
containing TiO -based photocatalysts has been employed, either
synthesised in a single and continuous step by flame spray pyroly-
sis (FP) [16], according to a procedure leading to very photoactive
high surface area titania, also modified with noble metals nanopar-
ticles [17,18], or obtained by the deposition of preformed noble
cating the total deposition of the NM colloids on the TiO surface.
Finally, the powder was recovered after at least four cycles of pre-
cipitation, supernatant separation and washing with water. The fi-
nal wet precipitated powder was centrifuged to minimise its water
content and dried overnight in oven at 70 °C. All chemicals were
purchased from Aldrich and used as received.
2
2
metals (Ag, Au, Au–Ag alloy and Pt) nanoparticles on TiO . Particu-
lar attention has been made on the identification of all photoreac-
tion intermediates and by-products and on their production rates,
also as a function of the inlet gas composition, aiming at identify-
ing the main reaction paths and at ascertaining which photocata-
lyst features are required to optimise the process, not only in
2.2. Photocatalysts characterisation
2
The BET-specific surface area was measured by N adsorption/
desorption at 77 K on a Micromeritics ASAP 2010 apparatus, after
out-gassing in vacuo at 300 °C for at least 6 h. X-ray diffraction pat-
terms of hydrogen yield, but also in terms of distribution of CH
oxidation products.
3
OH
Table 1
Photocatalytic performance of the investigated photocatalysts in methanol photo-steam reforming, in terms of rates of products formation, r, and per cent selectivity in relation to
ꢁ
1
hydrogen production, S. Reaction conditions: 0.014 g of photocatalyst fed in recirculation mode with 40 mL min of a 2% CH
3
OH/3% H
2
O/N
2
(balance) gas mixture.
ꢁ
1
ꢁ1
Photocatalyst
Production rate (mmol h
CO
g
)
Selectivity in relation to H
2
production (%)
cat
H
2
2
CO
H
2
CO
2
HCO H
CH
4
CO
2
CO
H
2
CO
HCO
2
H
Sum
4
4
3
3
3
3
3
3
3
TiO
FP-TiO
2
0.36
0.72
1.17
13.30
12.82
18.60
7.75
0.013
0.022
0.032
1.61
1.24
2.88
0.45
0.70
0.032
0.044
0.056
0.479
0.603
0.468
0.322
0.190
0.078
0.25
0.48
0.81
5.17
5.23
7.16
4.40
4.65
7.85
tr.
tr.
9.1 ꢂ 10ꢁ
10.7
9.1
8.2
36.3
29.0
46.4
17.5
25.1
21.6
17.7
12.2
9.6
7.2
9.4
5.0
8.3
4.6
1.1
69.4
67.0
68.8
38.9
40.8
38.5
56.7
55.6
55.1
–
–
98
88
98
95
96
103
98
97
ꢁ
2
7.6 ꢂ 10
ꢁ
1%Ag/TiO
1%Au/TiO
1%Au–1%Ag/TiO
1%Pt/TiO
0.5%Pt/TiO
0.5%Pt/FP-TiO
2
0.06
0.84
1.11
1.23
0.58
0.49
1.44
1.1 ꢂ 10
11.0
12.7
17.3
13.2
14.9
11.8
20.3
ꢁ
2
1.2 ꢂ 10
ꢁ
2
1.6 ꢂ 10
ꢁ
2
8.7 ꢂ 10
ꢁ
2
4.4 ꢂ 10
ꢁ
2
2
8.36
14.23
2.8 ꢂ 10
FP-0.5%Pt/TiO
1.02
2.3 ꢂ 10ꢁ
98

1
84
G.L. Chiarello et al. / Journal of Catalysis 273 (2010) 182–190
terns were recorded on a Philips PW3020 powder diffractometer,
by using the Cu K radiation (k = 1.54056 Å) and compared with
Three consecutive irradiation cycles were always performed for
each photocatalyst bed, in order to check the reproducibility of the
a
literature data [20] for phase recognition. Quantitative phase anal-
2
photoactivity tests, with 30-min N purging in the dark between
ysis was made by the Rietveld refinement method [21], using the
each irradiation cycle. No remarkable variation in the rate of prod-
ucts formation in the gas phase was ever observed, which was thus
evaluated as the average value in the three runs. In some cases, the
same photocatalyst bed was tested also on the next day in three
more subsequent irradiation cycles, with no significant variation
in photoactivity.
‘
‘Quanto” software [22]. UV–vis diffuse reflectance was measured
by a Perkin-Elmer Lambda 35 apparatus equipped with an integra-
tion sphere (Labsphere RSA-PE-20).
High Resolution Transmission Electron Microscopy (HRTEM)
and High Angular Annular Dark Field image in Scanning mode
(
2
STEM–HAADF) were taken on a JEOL FS2200-FEG operating at
00 kV. The catalysts were dispersed in ethanol, dropped on car-
bon-film coated copper grids and dried in air. In STEM and EDX
Energy Dispersive X-ray) analysis the spot size was 0.7 nm. The
.7.0 analysis station with the JED-2200 software from Jeol was
used to process EDX spectra.
3
. Results and discussion
(
3
3.1. Photocatalysts characterisation
Both FP-made materials, with or without Pt, consisted of micro-
aggregates of single crystal nanospheres, 10–25 nm in diameter, as
revealed by HRTEM analysis (Fig. 1A), and displayed almost identi-
cal surface areas (70 m g from BET analysis) and crystal phase
2
.3. Photocatalytic tests
2
ꢁ1
The photocatalytic activity in hydrogen production by methanol
photo-steam reforming was tested using an apparatus similar to
that already described [18]. The photocatalyst powder (14 mg)
was deposited on 3 g of 20–40 mesh (0.85–0.42 mm) quartz beads
by mixing them with 1.2 mL of distilled water, followed by drying
in oven at 70 °C for 6 h. The so obtained photocatalyst bed was in-
serted in the photoreactor, consisting in a flat cylindrical Plexiglas
cell, having a central 2 mm thick and 50 mm in diameter round
hollow, frontally closed with a Pyrex glass optical window (irradi-
A
2
ation surface ca. 20 cm ). The photoreactor was connected to a
closed stainless steel system, where the gas phase was recirculated
at constant rate by means of a metal bellows pump. The system
was preliminarily purged with nitrogen in order to remove any
oxygen trace. During the photocatalytic tests, the inert gas was sat-
urated with methanol/water vapours by continuously bubbling it
into an aqueous solution containing different methanol amounts,
kept at 30 °C, and then fed to the photoreactor. In particular, the
photocatalytic activity tests of different photocatalysts were car-
ried out employing 20 vol.% methanol in water solutions (standard
conditions), whereas different solutions containing from 1 vol.% up
to pure methanol were employed when investigating the effect of
methanol content. The photocatalyst bed was typically irradiated
for 2 h, while the recirculating gas was analysed on-line by sam-
pling it every 20 min by means of a pneumatic sampling valve
placed at the exit of the photoreactor. Gas samples were automat-
ically injected into an Agilent 6890 N gas-chromatograph (GC),
equipped with two columns (HP-PlotU and Molesive 5A), two
detectors (thermo conductivity and flame ionisation) and a Ni-cat-
5
nm
B
alyst kit for CO and CO
The GC response was first calibrated by injecting known volumes
of H , CO, CO and CH into the recirculation system through the
2 2
methanation. N was used as carrier gas.
2
2
4
loop of a six ways sampling valve. The amount of formic acid accu-
mulated in the liquid solution during the photocatalytic tests was
determined by ion chromatography (IC), employing a Metrohm
7
61 Compact IC instrument equipped with an anionic Metrosep
A column.
The irradiation source was an iron halogenide mercury arc lamp
Jelosil, 250 W), placed at 20 cm from the reactor, emitting in the
(
3
1
50–450 nm wavelength range with a full irradiation intensity of
.67 ꢂ 10 Einstein s cm on the reactor, as determined by
ꢁ
7
ꢁ1
ꢁ2
ferrioxalate actinometry [23]. Its constancy was checked daily by
ꢁ2
an UVA-meter, giving a constant reading of 37 mW cm
0 cm from the lamp.
The reactor temperature during irradiation was 55 ± 5 °C, as
at
2
5
nm
monitored by a thermocouple placed inside the cell. The absolute
pressure was 1.2 bar at the beginning of the runs and slightly in-
creased during irradiation, as a consequence of the accumulation
of products in the gas phase.
2 2
Fig. 1. HRTEM images of: (A) FP-0.5%Pt/TiO and (B) 1%Ag/TiO . The arrows point to
the noble metal nanoparticles.

G.L. Chiarello et al. / Journal of Catalysis 273 (2010) 182–190
185
composition (ca. 53% anatase and 47% rutile, from XRD analysis).
The high rutile content is mainly due to the high temperature of
the flame (ca. 2000 °C) [24], when xylene is used as solvent. The
smaller and better dispersed NM nanoparticles, on a relatively
higher surface area. No NM particle aggregation was observed in
2 2
the case of 1%Au/TiO (Fig. 2D), 1%Ag/TiO (Fig. 2E) and 1%Au–
typical structure features of commercial P25 TiO
2
, consisting of
1%Ag/TiO (Fig. 2F). However, the gold-containing sample exhibits
2
widely condensed, irregularly shaped, ca. 20 nm in size crystalline
aggregates, was maintained after NM nanoparticles’ deposition
bigger NM nanoparticles (3–8 nm on average, together with some
rare bigger particles up to 20 nm) than the silver-containing one
(1.5–3 nm). The Z-contrast of silver is poorer than those of gold
and platinum, silver being a lighter element. Finally, much bigger
NM particles, ranging from 5 to 20 nm, were detected in 1%Au–
(
4
see for example Fig. 1B). Also, the specific surface area,
8 m g according to our BET analysis, and mixed crystal phase
2
ꢁ1
composition of ca. 80% anatase and 20% rutile, according to XRD
measurements, were not altered by the here adopted low temper-
ature noble metals’ deposition procedure.
2
1%Ag/TiO , and two types of surface nanoparticles could be distin-
guished in this case, i.e. bigger and brighter ones made of Au–Ag
alloy, together with some smaller and less bright ones consisting
of pure silver, as confirmed by the EDX analysis reported in Fig. 3.
The UV–vis absorption spectra of the Au, Ag and Au–Ag alloy
colloidal suspensions before deposition on P25 are reported in
Fig. 4A. They exhibit the plasmonic bands centred at 400 and at
530 nm, typical of colloidal silver and gold nanoparticles, respec-
The STEM–HAAD images of the FP-made Pt/TiO
played in Fig. 2A evidence the presence of well-dispersed, ca.
.5–3-nm-sized Pt nanoparticles, appearing as bright dots depos-
ited on TiO , due to the different Z-contrast of the noble metal with
respect to the titania support. By contrast, the 0.5%Pt/FP-TiO (not
shown), 0.5%Pt/TiO (Fig. 2B) and 1%Pt/TiO (Fig. 2C) samples pre-
pared by the deposition of preformed Pt nanoparticles on FP-TiO
or P25 TiO exhibit the same Pt particles size distribution (5–
nm), with some extent of aggregation. Thus, the FP method al-
lows the direct synthesis of noble metal-deposited materials, with
2
sample dis-
1
2
2
2
2
2
2
6
2
1
5
0 nm
1
00
Au
Ag
Ti
1
8
6
4
2
0
0
0
0
0
O
0
2
4
6
8
10
12
Distance / nm
100
Au
2
80
60
40
20
0
Ag
Ti
O
0
2
4
6
8
10 12 14 16
Distance / nm
2
Fig. 3. STEM image of 1%Au–1%Ag/TiO and EDX analysis (lower panels) of areas 1
Fig. 2. STEM–HAADF images of: (A) FP-0.5%Pt/TiO
D) 1%Au/TiO , (E) 1%Ag/TiO and (F) 1%Ag-1%Au/TiO
appear as bright dots on the titania support because of their higher Z-contrast.
2
, (B) 0.5%Pt/TiO
2
, (C) 1%Pt/TiO
2
,
and 2 outlined in the STEM image, showing the per cent atom composition moving
along the marked points on: (1) an Au–Ag alloy nanoparticle and (2) a pure silver
nanoparticle.
(
2
2
2
. Noble metal nanoparticles

1
86
G.L. Chiarello et al. / Journal of Catalysis 273 (2010) 182–190
0
0
0
0
0
0
0
.6
.5
.4
.3
.2
.1
.0
4
0.20
Ag
Au-Ag alloy
A
A
0.16
0.12
0.08
3
Au
2
1
H O
2
CH OH
0.04
0.00
3
H CO
2
0
0.003
3
00
400
500
λ / nm
600
700
800
16
B
0.002
0.001
0.000
1
2
8
4
0
2
1
1
0
0
0
.0
.6
.2
.8
.4
.0
B
0.0
0.5
1.0
1.5
2.0
Irradiation time / h
H2
CO2
CH4
CO
Ag-Au / TiO2
Au / TiO2
FP-Pt / TiO2
Pt / TiO2
0.015
0.010
Ag / TiO2
0
.20
C
FP-TiO2
CH OCH
TiO2
3
3
0.15
HCO
CH
3
2
3
00
400
500
600
700
800
0
0
0
.10
.05
.00
CH CHO
3
λ / nm
0
.005
Fig. 4. UV–vis absorption spectra of: (A) Au, Ag and Au–Ag alloy colloidal aqueous
suspensions; (B) TiO , FP-TiO , FP-0.5%Pt/TiO , (labelled as FP-Pt/TiO ) and 1%M/
TiO (M = Au, Ag, Au–Ag and Pt) powders, recorded in reflectance mode.
2
2
2
2
2
0.000
0
.0
0.5
1.0
1.5
2.0
tively [25], arising from the collective oscillation of free conduction
band electrons, induced by interaction with an incident electro-
magnetic radiation, whose wavelength far exceeds the particles
size. The Ag–Au alloy colloidal suspension shows a single plasmon-
ic band centred at 430 nm, i.e. intermediate between those of pure
Au and Ag colloids, confirming the formation of an alloy.
Irradiation time / h
Fig. 5. Typical gas-phase composition profiles (concentration expressed in mole per
cent) during irradiation, in the methanol photo-steam reforming reaction. Inset:
magnification of the methane concentration profile. Specific reaction conditions:
.014 g of FP-0.5%Pt/TiO photocatalyst fed in recirculation mode with 40 mL min
of 2% CH OH/3% H O/N (balance) gas mixture.
ꢁ
1
0
2
3
2
2
2
All of the investigated TiO -based photocatalysts exhibit a band
gap adsorption threshold below 400 nm (Fig. 4B); metal-contain-
ing samples also exhibit plasmonic bands in the visible region, cen-
tred at 555 nm for Au, 445 nm for Ag and 525 nm for the Au–Ag
alloy, i.e. all of them red-shifted compared to that of the corre-
sponding colloidal suspension, indicating electronic interaction be-
2 2 4
of H , CO , CO and CH (Fig. 5B) increased linearly with time be-
cause they accumulated in the recirculating gas phase, according
to a pseudo-zero order rate law, due to the high excess of reactants.
By contrast, the concentration of formaldehyde, methyl formate,
acetaldehyde and dimethyl ether (Fig. 5A and C) rapidly increased
at the beginning of irradiation and then remained almost constant
in the gas phase, because such species accumulated in the metha-
nol–water liquid solution when the gas phase bubbled into it, as
confirmed by GC analysis of the liquid at the end of photocatalytic
tests. Formic acid, another possible intermediate species, was
never detected by GC analysis of the gas phase, very likely because
its amount always remained below the detection limit. However,
its formation and accumulation in the liquid solution was con-
firmed by IC analysis at the end of the runs.
2 2
tween NM nanoparticles and TiO support. Pt/TiO samples,
appearing as a grey powders, exhibit a broad absorption in the vis-
ible region, without any specific plasmonic band.
3
3
.2. Photocatalytic tests
.2.1. Identification of products and intermediate species
Besides the two main photoreaction products, i.e. hydrogen and
carbon dioxide, the presence of formaldehyde, carbon monoxide,
dimethyl ether, methyl formate, acetaldehyde and traces of meth-
ane and ethane as by-products, was detected during all photocat-
alytic tests. Fig. 5 displays the typical composition vs. time
profiles of the recirculating gas during irradiation, obtained when
On the basis of the main final oxidation products, by-products
and intermediate species detected in both gas and liquid phase,
the following reaction scheme should be considered:
hm;TiO
2
2
testing FP-made 0.5%Pt/TiO . The concentration of water and
CH
3
OH ƒƒƒƒ! H
2
CO þ H
2
ð1Þ
ð2Þ
methanol (Fig. 5A) did not significantly vary under steady-state
conditions, due to their low per cent conversion. The concentration
hm;TiO
2
H
2
CO þ H
2
O ƒƒƒƒ! HCO
2
H þ H
2

G.L. Chiarello et al. / Journal of Catalysis 273 (2010) 182–190
187
hm;TiO
2
1
00
20
16
12
8
HCO
2
H ƒƒƒƒ! CO
2
þ H
2
ð3Þ
ð4Þ
ð5Þ
ð6Þ
ð7Þ
rH
2
hm;TiO
2
H
2
CO ƒƒƒƒ! CO þ H
2
8
6
4
2
0
0
0
0
0
SCO
2
hm;TiO
2
2CH
2CH
2CH
3
3
3
OH ƒƒƒƒ! HCO
2
CH
3
þ 2H
2
SCO
hm;TiO
2
OH
!
CH
3
CHO þ H
2
þ H
2
O
hm;TiO
2
OH ƒƒƒƒ! CH
3
OCH
3
2
þ H O
According to reaction (7), the formation of dimethyl ether does
4
r
not contribute to hydrogen production.
The results of photocatalytic tests performed with the investi-
gated photocatalysts series are reported in Table 1. The rates of
0
2 2 4
H , CO , CO and CH production were taken as the slope of the
straight lines of the produced amount (normalised per unit catalyst
weight) vs. irradiation time plots (see for example Fig. 5B). The rate
of formaldehyde production (rH2CO) was estimated by the average
GC peak area recorded in gas samples collected during irradiation
at the exit of the photoreactor, taking into account the flow rate of
the recirculating gas, under the very reasonable assumption that
almost all formaldehyde present in the gas phase at the exit of
the photoreactor was trapped by the methanol/water solution
and not fed back into the photoreactor. The average rate of formic
acid production was estimated as the overall amount accumulated
into the liquid solution, determined by IC at the end of the runs, di-
vided by the overall irradiation time and the weight of catalyst,
thus assuming a constant production rate during irradiation also
for this intermediate species.
Fig. 6. Overview of the rate of hydrogen production, rH2 , and per cent selectivity to
CO2 and to CO obtained with the investigated photocatalysts. Reaction conditions
are reported in Table 1.
plane,
[
U
= 4.74 eV for Ag,
26], whereas values of 4.6–4.7 eV are reported in literature for
[27]. Consequently, Pt is a more efficient electron trapper than
. By
, suggests
U = 5.31 eV for Au and U = 5.93 eV for Pt
U
TiO
2
gold, in line with the higher photoactivity of Pt-modified TiO
contrast, the value of Ag, very close to that of TiO
2
U
2
scarce electron transfer, resulting in less efficient charge separation
and consequent little improvement in the photocatalytic perfor-
The selectivity in hydrogen production was also calculated from
mance upon Ag addition to TiO
The same trend is reflected by rCO2 and SCO2 in the series of phot-
ocatalysts containing 1% metal on P25 TiO (see Fig. 6), 1%Pt/TiO
being the most selective photocatalyst (46%), followed by 1%Au/
TiO (36%) and 1%Ag/TiO (8%). Moreover, when considering the
2
.
the rates of H
tween the rate of H
the four products, by taking into account the stoichiometric coeffi-
cients of reactions (1)–(4), and the overall rate of H photoproduc-
production (SCO2 ) was
2
2 2
CO, HCO H, CO or CO formation, as the ratio be-
2
production from methanol to give one of
2
2
2
2
2
tion. For example, the selectivity of CO
calculated as follows:
2 2
in H
use of hydrogen as a feedstock for fuel cells, CO certainly is the
most undesired by-product, being a well-known poison for the
Pt-based catalysts in fuel cells. From the point of view of CO pro-
CO2 ¼ 3
ꢃ rCO2
S
ꢃ 100
duction, bare TiO
because of the low rH2 value, but also for its high SCO value, even
higher than SCO2 (e.g. 18% vs. 11% for TiO P25, see Table 1). Noble
metal deposition on TiO significantly increases SCO2 and decreases
CO. In particular, 1%Pt/TiO showed the lowest SCO, followed by
Au/TiO and Ag/TiO . This clearly indicates that the more efficient
2
is the less performing photocatalyst, not only
r
H
2
The sum of the so obtained selectivity values, also reported in
Table 1, is always close to 100%, indicating a very good mass bal-
ance agreement. Thus, under standard experimental conditions,
the contribution to H production of reactions (5) and (6) is negli-
2
gible. However, this does not apply for higher methanol to water
molar ratios, as will be discussed in Section 3.2.3.
2
2
S
2
2
2
is the separation between photoproduced charge carriers in the
ꢁ
photocatalyst, the more efficient are not only the e – involving
CB
reduction paths, mainly leading to H
2
production, but also the
þ
3
.2.2. Effects of noble metals deposition on TiO
The rate of hydrogen production r 2 on flame-made TiO
higher than on unmodified TiO P25, as already observed in our
2
h
3
– initiated complete CH OH oxidation to CO
2
.
VB
H
2
was
Although a systematic investigation into the effects of metal
loading on titania is beyond the scope of the present work, we no-
2
previous studies [18], and it dramatically increased upon noble
metals’ deposition on the semiconductor oxide, as clearly evi-
denced by the overview, reported in Fig. 6, of the hydrogen produc-
tion rate and selectivity values obtained with the investigated
photocatalysts. Silver resulted the less effective co-catalyst, with
tice that a more than doubled rH2 value was attained, when the
amount of Pt deposited on TiO
2
P25 was doubled from 0.5% and
1% (Fig. 6), and even more remarkable was the effect observed in
r
CO2 and SCO2 values (Table 1), increasing from 0.45 (SCO2 = 17.5%)
ꢁ1
ꢁ1
to 2:88 mmolCO2
h
g
(SCO2 = 46.4%).
cat
only a fourfold rH2 increase with respect to rH2 on bare P25 TiO
whereas rH2 increased by 37 and 52 times when the same amount
of gold or platinum, respectively, was deposited on TiO . The differ-
ence in the photocatalytic performance of the three metal co-cata-
lysts can be related to their work function values ( ), i.e. the
energy required to promote an electron from the Fermi energy le-
vel into vacuum (the higher is , the lower in energy is the Fermi
level). In fact, the greater is the difference between the metal work
function and that of the TiO support, the higher is the Schottky
2
,
A comparison between the performance of the three 0.5 wt.%Pt-
containing photocatalysts, prepared either by deposition of pre-
2
formed metal nanoparticles on P25 TiO
one-step FP (FP-0.5%Pt/TiO , see Fig. 6), evidences that, although
bare FP-TiO was more active than P25, much more closer rH2 val-
ues were attained when the same amount of Pt was deposited on
the two TiO supports, with a similar metal nanoparticles size dis-
tribution, whereas rCO2 and rCO varied more significantly (Table 1).
Indeed, as in the case of the bare photocatalysts, 0.5%Pt/FP-TiO en-
sured higher rCO2 and SCO2 values and lower rCO and SCO values than
0.5%Pt/TiO . These observations could be explained by considering
that H is expected to be mainly produced through the involve-
ment of photopromoted electrons on the NM nanoparticles surface,
2 2
and on FP-TiO , or by
2
U
2
U
2
2
2
barrier [8,9], the electronic potential barrier generated by the band
alignment at the metal–semiconductor heterojunction, with con-
sequent increased efficiency of photogenerated electron transfer
2
2
H
and trapping by the metal, leading to higher r 2 . For the 111 crystal

1
88
G.L. Chiarello et al. / Journal of Catalysis 273 (2010) 182–190
which are preformed before deposition on TiO
2
. Therefore, they are
photocatalysts,
fractions x (0.0045 6 x 6 1). At this temperature, the vapour pres-
sure of methanol and water are 163.97 and 31.82 mm Hg, respec-
tively; their partial pressure ratios in the gas phase, pH2O=pCH3OH,
were calculated by assuming an ideal behaviour, according to Rao-
similar for the two 0.5%Pt/TiO and 0.5%Pt/FP-TiO
2
2
and this leads to similar rH2 values. On the other hand, methanol
photo-oxidation, involving valence band holes, occurs on the tita-
nia surface, and this explains why the noble metal-modified phot-
ult’s law. The complete series of results concerning H
2
, CO
2
, CO,
pro-
ocatalysts retain memory of the higher photoactivity of bare TiO
2
.
H
2
CO, HCO H and CH formation rates and selectivities in H
2
4
2
Finally, the one-step prepared FP-0.5%Pt/TiO exhibited an excep-
2
duction is reported in Table 2. The effects of the fed mixture com-
position on the rate of formation of each product can better be
appreciated in Fig. 7, where rate values are plotted as a function
of x, the molar fraction of methanol in the liquid phase.
Fig. 7 clearly shows that the rates of hydrogen, formaldehyde
and formic acid production display bell-shaped curves as a func-
tion of x, whereas carbon dioxide exhibits a hyperbolic decay curve
tionally higher performance with respect to those of the two mate-
rials prepared by deposition of preformed Pt nanoparticles, despite
of the same noble metal content, with doubled rH2 and rCO2 values,
together with an exceptionally low rCO value and only 1% selectiv-
ity to CO. The higher photoactivity of FP-0.5%Pt/TiO
ably a consequence of the lower Pt particle size and higher Pt
dispersion on the TiO surface (see Fig. 2A), typical of flame-made
catalysts, ensuring a more extended Pt surface area, in line with the
H2 increase and rCO decrease with decreasing the Au particles’ size
on TiO in the liquid-phase methanol photo-reforming over Au/
TiO photocatalysts [10].
Silver was the least performing co-catalyst on TiO
2
is most prob-
2
with increasing x. Thus, the rate of CO
2
production and the selectiv-
ity to CO were high at relatively high water partial pressure and
2
r
rapidly decreased as the partial pressure of water decreased and
that of methanol increased, clearly showing that water is needed
2
2
to achieve complete CH
of H
teau (ca. 17 mmolH2
3
OH oxidation. On the other hand, the rate
2
(Fig. 6), and
2
production increased with increasing x, reaching a sort of pla-
ꢁ
1
ꢁ1
its addition to gold to form Au–Ag alloy nanoparticles induced a
decrease in photoactivity with respect to that of pure gold nano-
h
g
) for 0.2 < x < 0.6, and decreased down
cat
ꢁ1
ꢁ1
to 2:8 mmolH2
h
g
, when the photocatalyst bed was fed with
cat
particles on TiO
tained with the 1%Au–1%Ag/TiO
the corresponding values attained with 1%Au/TiO
2
. Indeed, the rH2 and selectivity to CO
photocatalyst were lower than
, with a slightly
2
values at-
pure methanol. It is worth underlining, however, that whereas
hydrogen and formaldehyde production did not drop to zero under
such conditions, no HCO H and CO were produced in the absence
2 2
2
2
higher selectivity to CO. It is reasonable to assume that silver addi-
tion induced a work function decrease with respect to that of pure
gold, leading to less efficient separation of the photoproduced
charge pairs.
Finally, the rate of formation of the other reduction product, i.e.
methane, was always very low with respect to that of hydrogen
production rH2 , and it only slightly increased upon noble metal
deposition (Table 1). This might be an indication that photopro-
moted electrons are not directly involved in methane formation.
On the other hand, formaldehyde was always the main by-product
formed on NM containing photocatalysts, with a selectivity always
of water, in agreement with reaction (2), followed by reaction (3).
By contrast, the rate of CO production was almost constant up to
x = 0.8, and it dropped to ca. zero when pure methanol was fed
to the photocatalyst, with the selectivity to CO never exceeding
1.5%. Furthermore, CO production rate as a function of x has a trend
more similar to that of formic acid production, than to that of form-
aldehyde production, which did not drop to zero in the absence of
water. This indicates that CO does not form directly from metha-
nol, but it only forms from methanol oxidation intermediates,
and in particular mainly through formic acid dehydration, reaction
(8), rather than through formaldehyde oxidation, reaction (4).
exceeding 40%, except in the case of 1%Pt/TiO
2
, mainly yielding
hm;TiO
2
CO
2
.
HCOOH ƒƒƒƒ! CO þ H O
ð8Þ
2
The two CH
3
CHO and HCO
2
CH
3
by-products exhibit a null pro-
3.2.3. Effect of methanol and water partial pressures
duction rate at low x and a production rate increase for x > 0.1, with
a maximum rate shifted towards very high x values (ca. 0.9). This
trend is consistent with reactions (5) and (6), involving two adja-
cent surface-adsorbed methanol molecules, because with increas-
ing x and methanol partial pressure, the surface coverage by
methanol increased, together with the probability that two vicinal
adsorbed methanol molecules interact with each other, rather than
In order to ascertain how the rates of product formation are
influenced by the methanol/water molar ratio in the recirculating
gas phase under irradiation, and in particular to discern the role
of water in hydrogen production and methanol oxidation, a sys-
tematic investigation was performed by varying the methanol
and water partial pressures. The rate of formation of the main reac-
tion products was thus measured by feeding the FP-0.5%Pt/TiO
2
3 2 3
with water. However, CH CHO and HCO CH production from pure
photocatalyst with vapours in equilibrium with water/methanol li-
quid solutions, kept at 30 °C, containing different methanol molar
methanol occurred at lower rate, suggesting that water plays a role
also in their formation.
Table 2
Effect of methanol molar fraction in the aqueous solution, xCH3 OH, and of water-to-methanol partial pressure ratio in the recirculating gas phase in the photocatalytic performance
of FP-0.5%Pt/TiO in terms of rates of products formation, r, and per cent selectivity in relation to hydrogen production, S.
2
p
H
O
ꢁ1 ꢁ1
x
CH3OH
2
Production rate (mmol h
g
)
Selectivity in H
2
production (%)
CO
cat
p
3
CH OH
H
2
CO
2
CO
2
H CO
HCO
2
H
CH
4
CO
2
CO
H
2
2
HCO H
Sum
1.000
0.956
0.894
0.800
0.640
0.400
0.229
0.100
0.047
0.023
0.0090
0.0045
0
2.81
6.18
8.71
0
0.003
0.026
0.061
0.078
0.071
0.076
0.064
0.078
0.086
0.065
0.067
0.067
1.75
3.42
4.66
5.49
8.03
9.10
9.49
7.85
5.40
4.25
2.96
1.41
0
0
0
0
0
1.7
3.0
3.7
5.8
9.3
14.4
21.6
34.1
47.8
59.9
76.7
0.22
0.83
1.40
1.23
0.85
0.88
0.77
1.10
1.35
1.14
1.34
1.45
62.3
55.4
53.5
43.1
47.9
52.3
57.1
55.1
42.3
37.2
29.7
15.3
0
5.8
8.1
6.3
13.6
18.6
23.8
16.6
16.4
13.8
6.8
63
64
66
54
68
81
96
94
94
100
98
99
0.009
0.023
0.05
0.11
0.29
0.66
1.75
3.93
8.30
21.4
43.2
0.03
0.10
0.16
0.32
0.54
0.80
1.02
1.45
1.82
1.99
2.35
0.18
0.35
0.40
1.14
1.62
1.98
1.18
1.05
0.79
0.34
0.26
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
4
3
3
3
3
3
3
3
3
12.74
16.76
17.38
16.63
14.23
12.77
11.43
9.98
9.6 ꢂ 10
1.6 ꢂ 10
1.8 ꢂ 10
2.0 ꢂ 10
2.3 ꢂ 10
2.2 ꢂ 10
2.4 ꢂ 10
3.2 ꢂ 10
9.17
4.5 ꢂ 10ꢁ
5.7

G.L. Chiarello et al. / Journal of Catalysis 273 (2010) 182–190
189
2
2.
2.
1.
1.
0.
0.
5
100
H2
HCO H
2
1
0
8
6
4
2
0
0
0
1
5
00
8
4
0
0
00
5
00
0
0
.0
0
.2
0
.4
0
.6
0
.8
1
.0
0
.0
0
.2
0
.4
0
0
.
.
6
6
0
0
.
.
8
8
1.0
1.0
C
3
H OH molar fraction
3
CH OH molar fractionn
1
1
2
0
8
6
4
2
0
100
2.5
2.0
1.5
1.0
0.5
0.0
100
80
60
40
20
0
H CO
2
CO2
8
6
4
2
0
0
0
0
0
0
.0
0
2
0.4
0
6
0
.8
1
.0
0.0
0.2
0.4
0.6
0.8
1.0
C
3
H O
H
m
o
o
n
3
CH OH molar fraction
0
0
0
0
0
0
.10
.08
.06
.04
.02
.00
5
250
200
150
100
50
CO
Serie1
HCO CH
2
3
4
3
2
1
0
C
H
Serie 32 CHO
0
0
.0
0.2
0.4
0.6
0.8
1.0
0.0
0.2
0.4
0.6
0.8
1.0
3
CH OH molar fraction
3
CH OH molar fraction
Fig. 7. H
2
, H
2
CO, HCO
2
H, CO
2
2 3 3 2
, CO, HCO CH and CH CO production rates (N) and per cent selectivity in relation to hydrogen production (h) obtained with the FP-0.5%Pt/TiO
photocatalyst, as a function of the methanol molar fraction in the aqueous solution.
Finally, the sum of the selectivity SCO2 , SCO, SH2CO and SHCO2
H
val-
(xHCO2
H
¼ 1:0); (ii) 80 vol.% HCO
20 vol.% HCO H/H
¼ 0:1).
The results, shown in Fig. 8, first of all demonstrate that both H
and CO production rates significantly increased with increasing
2
H/H
2
O (xHCO2
H
¼ 0:7); and (iii)
ues (Table 2, last column on the right) gives a good mass balance
for the overall reactions, i.e. this sum approaches 100%, only for
low x values, indicating that reactions (1)–(4) relative to overall
photocatalytic steam reforming, satisfactorily accounts for hydro-
gen production and simultaneous oxidation products, whereas this
sum decreases for x > 0.4, i.e. when the contribution of bimolecular
side-reactions (5) and (6) to hydrogen production becomes
significant.
2
2
O (xHCO
2
H
2
2
the water partial pressure in the reaction mixture. Furthermore,
the rate of hydrogen production was always higher than that of
CO production. Both facts clearly point to an involvement of water
2
molecules in the photocatalytic oxidation of formic acid in the ab-
sence of dioxygen. In fact, if only formic acid was involved in a
simultaneous oxidation and reduction process according to reac-
3
.2.4. Effect of water on the gas-phase anaerobic oxidation of formic
acid
In order to have a better insight into the role played by water in
2 2
tion (3), equal amounts of H and CO would have been produced
under irradiation. So, a much more complex mechanism is at work,
part of the photoproduced hydrogen deriving from water (protons)
reduction by photopromoted electrons, most probably occurring
on the NM nanoparticles, whereas formic acid may undergo oxida-
tion either through a direct mechanism, by interaction with photo-
the final steps of the investigated reaction, and in particular on the
origin of CO production, we performed some tests on the effect of
water in the anaerobic photo-oxidation of formic acid, one of the
intermediate species of methanol oxidation. The rates of hydrogen,
carbon dioxide and carbon monoxide production were thus mea-
ꢀ
produced valence band holes, or through the attack of OH radical
produced from water oxidation on the irradiated photocatalyst
sured with the FP-0.5%Pt/TiO
2
photocatalyst fed with vapours in
surface [2]. The difference between the rates of H
2 2
and CO produc-
equilibrium with liquid phases, kept at 30 °C: (i) pure formic acid
tion observed for xHCO2
H
¼ 1 might be accounted for by the pres-

1
90
G.L. Chiarello et al. / Journal of Catalysis 273 (2010) 182–190
6
5
4
3
2
1
0
2
0.7. Complete methanol oxidation to CO very rapidly declines
5
.4
with increasing methanol amount, starting from very low metha-
nol content. Carbon monoxide is mainly produced from formic acid
dehydration.
H2
CO2
4.1
CO
Acknowledgments
2
.9
Financial support from the University of Milan PUR funds, with-
in the Project Photoinduced Redox Processes on Semiconductor Sur-
faces and from the CARIPLO Foundation through the Project
Development of nanostructured photocatalytic films for energy con-
version on microplatforms is gratefully acknowledged.
1
.9
1.7
0.7
1
.2
0
.04
0.08
0.08
References
1
.0
0.1
HCOOH molar fraction
[1] K. Weissermel, H.J. Arpe, Industrial Organic Chemistry, fourth ed., Wiley-VCH,
Weinheim, 2003.
[
[
[
[
2] A. Fujishima, X. Zhang, D.A. Tryk, Surf. Sci. Rep. 63 (2008) 515.
3] Z. Zou, J. Ye, K. Sayama, H. Arakawa, Nature 414 (2001) 625.
4] Z. Zou, H. Arakawa, J. Photochem. Photobiol. A 158 (2003) 145.
5] R. Abe, M. Higashi, K. Sayama, Y. Abe, H. Sugihara, J. Phys. Chem. B 109 (2005)
Fig. 8. H
of formic acid on the FP-0.5%Pt/TiO
acid or formic acid/water solutions.
2
, CO
2
and CO production rates in the gas-phase anaerobic photo-oxidation
2
photocatalyst, fed with vapours of pure formic
16052.
[
6] W. Yao, J. Ye, Chem. Phys. Lett. 435 (2007) 96.
ence of the small amount of water (ca. 2%) in the employed formic
acid.
[7] H. Kato, A. Kudo, Catal. Today 78 (2003) 561.
[
[
8] P. Kamat, J. Phys. Chem. B 106 (2002) 7729.
9] A.L. Linsebigler, G. Lu, J.T. Yates, Chem. Rev. 95 (1995) 735.
Finally, the rate of CO production from pure HCO
2
H was much
[
10] G. Wu, T. Chen, W. Su, G. Zhou, X. Zong, Z. Lei, C. Li, Int. J. Hydrogen Energy 33
(2008) 1243.
[11] O. Rosseler, M.V. Shankar, M. Karkmaz-Le Du, L. Schmidlin, N. Keller, V. Keller,
J. Catal. 269 (2010) 179.
[12] N. Strataki, V. Bekiari, D.I. Kondarides, P. Lianos, Appl. Catal. B: Environ. 77
(2007) 184.
[13] J. Greaves, L. Al-Mazroai, A. Nuhu, P. Davies, M. Bowker, Gold Bull. 39 (2006)
ꢁ
1
higher than that from pure CH OH (0.04 vs. 0.003 mmolCO
3
h
ꢁ
1
g
), and it also increased in the presence of water (Fig. 8), reaching
cat
ꢁ
1
ꢁ1
a value of ca. 0:08 mmolH2
h
g
. This is in agreement with the
cat
above proposed mechanism of CO production in the process under
study, also in line with Wu’s results [10], i.e. CO is formed prefer-
entially from formic acid through reaction (8). Water has a benefi-
216.
[
[
14] M. Kang, J. Mol. Catal. A: Chem. 197 (2003) 173.
15] H. Yoshida, K. Hirao, J. Nishimoto, K. Shimura, S. Kato, H. Itoh, T. Hattori, J. Phys.
Chem. C 112 (2008) 5542.
16] G.L. Chiarello, I. Rossetti, L. Forni, J. Catal. 236 (2005) 251.
17] G.L. Chiarello, E. Selli, L. Forni, Appl. Catal. B: Environ. 84 (2008) 332.
ꢀ
cial effect also on this reaction, most probably by providing OH
radical-mediated alternative paths.
[
[
4
. Conclusions
[18] G.L. Chiarello, L. Forni, E. Selli, Catal. Today 144 (2009) 69.
[
[
19] J.H. Liu, A.Q. Wang, Y.S. Chi, H.P. Lin, C.Y. Mou, J. Phys. Chem. B 109 (2005) 40.
20] M.D. Advanced Selected Powder Diffraction Data, J.C.P.D.S., Swarthmore, PA,
FP-made photocatalysts containing NM nanoparticles are best
1974.
performing in the photo-steam reforming of methanol, in terms
of both hydrogen production efficiency and low CO production, be-
cause of their smaller NM particle size and higher dispersion on an
oxide higher surface area. Pt confirmed to be the most effective co-
catalyst also for this gas phase reaction, followed by gold and sil-
ver, in agreement with their work function values.
Although the presence of methanol, as an electron donor, is
beneficial in the photocatalytic hydrogen production from water,
its amount in the fed gas mixture should be kept low, the highest
hydrogen production rate being attained for water-to-methanol
molar ratios in the feeding gas phase comprised between 0.1 and
[21] H.M. Rietveld, J. Appl. Cryst. 2 (1969) 65.
[
[
22] A. Altomare, M.C. Burla, C. Giacovazzo, A. Guagliardi, A.G.G. Moliterni, G.
Polidori, R. Rizzi, J. Appl. Cryst. 34 (2001) 392.
23] C.G. Hatchard, C.A. Parker, Proc. Roy. Soc. (London) A235 (1956) 518.
[24] G.L. Chiarello, I. Rossetti, P. Lopinto, G. Migliavacca, F. Forni, Catal. Today 117
(2006) 549.
[
[
[
25] M.M. Alvarez, J.T. Khoury, T.G. Schaaff, M.N. Shafigullin, I. Vezmar, R.L.
Whetten, J. Phys. Chem. B 101 (1997) 3706.
26] CRC Handbook of Chemistry and Physics, 87th ed., Taylor & Francis, 2006, pp.
12–114.
27] C. Young, T.M. Lim, K. Chiang, J. Scott, R. Amal, Appl. Catal. B: Environ. 78
(
2008) 1.