Sie sind auf Seite 1von 15

Journal of Catalysis 329 (2015) 499–513

Contents lists available at ScienceDirect

Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat

Effect of TiO2 polymorph and alcohol sacrificial agent on the activity of


Au/TiO2 photocatalysts for H2 production in alcohol–water mixtures
Wan-Ting Chen a, Andrew Chan a, Zakiya H.N. Al-Azri a,b, Aubrey G. Dosado a,c, Muhammad A. Nadeem d,
Dongxiao Sun-Waterhouse a, Hicham Idriss d,e, Geoffrey I.N. Waterhouse a,b,c,⇑
a
School of Chemical Sciences, The University of Auckland, Auckland, New Zealand
b
The MacDiarmid Institute for Advanced Materials and Nanotechnology, New Zealand
c
The Dodd-Walls Centre for Photonic and Quantum Technologies, New Zealand
d
SABIC, Corporate Research Institute (CRI), KAUST, Saudi Arabia
e
Department of Chemistry, University College London, London, UK

a r t i c l e i n f o a b s t r a c t

Article history: This study compares the performance of 5 different Au/TiO2 photocatalysts for H2 production in alcohol–
Received 5 May 2015 water mixtures under UV excitation, placing particular emphasis on the role of the TiO2 support and alco-
Revised 16 June 2015 hol sacrificial reagent. 1.5 wt.% Au/TiO2 photocatalysts were prepared by the deposition–precipitation
Accepted 17 June 2015
with urea method, using P25 TiO2 (85% anatase and 15% rutile, 49 m2 g1), sol–gel anatase (39 m2 g1),
hydrothermally synthesised anatase (115 m2 g1), commercial brookite (33 m2 g1) and hydrothermally
synthesised rutile (29 m2 g1) as supports. UV–Vis, TEM, XRD, XPS and Au L3-edge EXAFS characterisation
Keywords:
studies confirmed the presence of gold nanoparticles of mean size 5–10 nm on all the Au/TiO2 photocat-
Hydrogen
Photocatalysis
alysts, which generated intense localised surface plasmon resonances (LSPRs) centred between 560 and
TiO2 polymorph 590 nm. Gold nanoparticle deposition suppressed electron–hole pair recombination in each TiO2 support
Gold following UV excitation, and created cathodic sites on TiO2 for H2 evolution. Photocatalytic H2 production
Alcohols tests were conducted in water, methanol–water, ethanol–water, ethylene glycol–water and glycerol–wa-
Sacrificial agent ter mixtures (alcohol concentration 10 vol.%) at a UV flux of 6.5 mW cm2. H2 production rates decreased
in the order Au/P25 > Au/anatase SG  Au/brookite > Au/anatase HT  Au/rutile HT when the rates
were normalised against photocatalyst surface area. For each Au/TiO2 photocatalyst, the rates decreased
in the order glycerol > ethylene glycol > methanol > ethanol. Good correlations were established between
H2 production rates and the number of hydroxyl groups, polarity and relative hole scavenging ability of
the alcohols. Results improve knowledge about the role of sacrificial reagents in photocatalytic hydrogen
production.
Ó 2015 Elsevier Inc. All rights reserved.

1. Introduction and have a significant carbon footprint [5–7]. Various alternative


H2 production technologies are presently being pursued [8–12],
There is growing consensus that hydrogen (H2) will replace fos- with water splitting using sunlight and semiconductor photocata-
sil fuels as the principal energy carrier in the near future, though a lysts considered one of the best long term prospects for the
seamless transition to the envisaged ‘‘hydrogen economy’’ requires delivery of a carbon neutral H2 supply [13–15].
major technical advancements in hydrogen production and stor- M/TiO2 photocatalysts, where M represents certain high work
age, of which the former is arguably the most challenging [1–4]. function metals including Ni, Pd, Pt or Au, demonstrate good activ-
Current hydrogen production technologies based around steam ity and stability for water splitting in the presence of sacrificial
methane reforming (SMR) and the water-gas shift reaction are effi- reagents under UV or full solar irradiation [16–21]. The band edges
cient and mature, but are not sustainable long term as they are of TiO2 are appropriate for water splitting, since the top of the
energy intensive, based on a non-renewable natural gas feedstock valence band (+2.7 V vs. NHE for anatase) is more positive than
the O2/H2O redox couple (+1.23 V vs. NHE), and the bottom of
⇑ Corresponding author at: School of Chemical Sciences, The University of the conduction band (0.5 V vs. NHE for anatase) is more negative
Auckland, Auckland, New Zealand. Fax: +64 9 373 7422. than the H2O/H2 redox couple (0.0 V vs. NHE). The band gap energy
E-mail address: g.waterhouse@auckland.ac.nz (G.I.N. Waterhouse). of TiO2 (Eg = 3.0, 3.2 and 3.3 eV for rutile, anatase and brookite,

http://dx.doi.org/10.1016/j.jcat.2015.06.014
0021-9517/Ó 2015 Elsevier Inc. All rights reserved.
500 W.-T. Chen et al. / Journal of Catalysis 329 (2015) 499–513

respectively) is also sufficiently small to allow solar excitation, that of anatase across interfacial heterojunctions (the conduction
which is a critical requirement for any practical water-splitting band of anatase is 0.2 eV more positive versus NHE than that of
photocatalyst. Whilst bulk absorption of electromagnetic radiation rutile), and hole migration from the valence band of anatase to that
with E > Eg generates electron–hole pairs (e–h+) in TiO2 that could of rutile, facilitates charge separation and increases the number of
potentially drive water splitting, bare TiO2 surfaces demonstrate charge carriers available for photoreactions [33]. Photoreaction
negligible activity for H2 production due to (1) rapid electron–hole rates generally scale in proportion to charge carrier concentrations.
pair recombination after photoexcitation; and (2) the high overpo- Similar positive synergies have also been reported in brookite–ru-
tential for H2 evolution on TiO2 surfaces. Deposition of Ni, Pd, Pt or tile and anatase–brookite systems (possible since the conduction
Au on TiO2 creates rectifying Schottky junctions [16–21], the for- bands of anatase, brookite and rutile are offset at 0.5, 0.6 and
mation of which is highly beneficial for hydrogen evolution as 0.7 V versus SHE, respectively). The effect of the TiO2 support
the Fermi level (EF) for each metal sits between the bottom of composition on the activity of M/TiO2 photocatalysts for H2 pro-
the TiO2 conduction band and the H2O/H2 redox couple (for 5 nm duction in alcohol–water systems has been the subject of a number
Au nanoparticles on TiO2, EF = 0.27 V) [21]. Electrons of investigations [27,33–35], but very few studies actually encom-
photo-excited into the conduction band of TiO2 and thus migrate pass all three TiO2 polymorphs and biphasic supports such as P25.
onto the supported metal nanoparticles, thereby suppressing elec- Gauging the relative merits of P25, anatase, brookite and rutile as
tron hole pair recombination in TiO2, whilst the metal nanoparti- supports from existing literature is not straightforward, as addi-
cles themselves function as cathodic sites for H2 evolution. The tional factors including TiO2 crystallite size, crystallite shape, sur-
addition of sacrificial agents (typically electron donors such as face area and photocatalytic test conditions also strongly impact
methanol, ethanol, ethylene glycol or glycerol) with oxidation H2 production rates in M/TiO2 systems.
potentials lower than water further suppresses electron hole pair Arguably the least researched aspect of photocatalytic H2 pro-
recombination in TiO2. In such systems, both water splitting duction in alcohol–water systems is the effect of the alcohol sacri-
(H2O ? 1/2O2 + H2) and alcohol photoreforming occur (e.g. ficial agent. Very few literature studies compare H2 rates for
CH3OH + H2O ? CO2 + 3H2 and CH3CH2OH + 3H2O ? 2CO2 + 6H2), M/TiO2 photocatalysts in different alcohol–water mixtures, or even
which under carefully optimised reaction conditions can yield H2 at a single alcohol at different alcohol–water concentrations [16–
production rates as high as 30–40 mmol g1 h1 at UV fluxes com- 40]. Bowker et al. reported hydrogen production rates for the pho-
parable to those in sunlight. For a given M/TiO2 photocatalyst, H2 toreforming of a variety of bio-derivable feedstocks over Pd/P25
production rates are known to depend closely on the co-catalyst under UV excitation, and observed that the rate decreased in the
loading, the TiO2 support composition (anatase, brookite, rutile order glycerol > aq. D-glucose > methanol > ethanol  aq. sucro-
or biphasic combinations thereof) and the type of sacrificial agent se > 1-butanol [17]. Bahruji et al. presented a more comprehensive
used. A gap in the current literature is the lack of comprehensive study of the Pd/P25 system using 20 different sacrificial reagents,
studies simultaneously examining the effects of both the TiO2 sup- and established the reactivity series triols > diols > 2° alcohols > 1°
port and the alcohol hole scavenger on H2 production rates in alcohols > 3° alcohols [18]. From this pattern of reactivity, it was
M/TiO2 systems. It is unclear whether the relative activities of concluded that the sacrificial hole scavenger must have an a-H
M/TiO2 photocatalysts prepared using different TiO2 supports are adjacent to the OH group(s), with the other main by-products of
altered by the type of alcohol hole scavenger used in the photocat- alcohol photoreforming being largely predictable based on the
alytic tests, motivating further investigations. alcohol structure. Bowker et al. later investigated the Au/P25 sys-
TiO2 exists in three main polymorphic forms, all of which have tem, and found that H2 production rates decreased in the order
electronic band gaps in the near UV region [22–24]. The three poly- methanol > 1-propanol > ethanol > 1-butanol > 2-propanol  ter-
morphs are constructed from TiO6 octahedra, differing primarily in tiary butanol [19]. Yang et al. examined UV irradiated suspensions
the connectivity of the octahedra. Rutile has a tetragonal unit cell of Pt/anatase in different alcohol–water mixtures, and observed
(P42/mnm, a = 4.5937 Å, c = 2.9587 Å; Z = 2), with a structure that that H2 evolution rates decreased in the order metha-
can be described as chains of edge sharing TiO6 octahedra running nol  ethanol > 1-propanol  2-propanol > 1-butanol [36]. A corre-
parallel to the c-axis [25]. Rutile is the most stable TiO2 polymorph, lation was established between the H2 production rate and alcohol
though generally has poor photocatalytic activity owing to the fact polarity. Li et al. investigated Cu2O/P25 photocatalysts for H2 evo-
it is a direct band gap semiconductor which allows fast electron– lution from aqueous solution with different hole scavengers, and
hole pair recombination following photoexcitation [26,27]. established the activity series methanol > ethylene glycol  glyc-
Anatase also crystallises with a tetragonal unit cell (I41/amd, erol > ethanol under UV–Vis excitation [37]. Wang et al. examined
a = 3.7845 Å, c = 9.5143 Å; Z = 4), with each TiO6 octahedra edge photocatalytic H2 production from aliphatic alcohol–water mix-
sharing with four adjacent octahedra [25]. Anatase is an indirect tures over Pt/TiO2 nanotubes [38]. The rate of H2 production fol-
band gap semiconductor, typically displaying photocatalytic activ- lowed the trend methanol > ethanol > i-propanol > n-propanol >
ities 1–2 orders of magnitude higher than rutile due to longer sec-butanol > tert-butanol. Recently, Su et al. demonstrated that
charge carrier lifetimes. In brookite, each octahedron shares three Pd(shell)–Au(core) nanoparticles immobilised on TiO2 exhibit high
edges with adjoining octahedra, to form an orthorhombic structure quantum efficiencies for H2 production in a wide range of
(Pbca, a = 5.4558 Å, b = 9.1819 Å, c = 5.1429 Å, Z = 8) [25]. To date, aqueous alcohols, observing that photocatalyst activity followed
only a limited amount of data is available about the photocatalytic the sequence methanol > glycerol > ethanol > 1,2-propanediol >
activity of brookite, owing to historical difficulties in synthesising ethylene glycol > aq. glucose > aq. fructose [39]. These studies
this polymorph in pure form [28,29]. Though it has yet to be provide a reasonable guide as to the general reactivity pattern fol-
unequivocally confirmed, brookite is considered to be an indirect lowed by M/TiO2 photocatalysts in alcohol–water mixtures. What
band gap semiconductor, which is consistent with its photocat- is not clear is the influence of the TiO2 support on these reactivity
alytic activity being comparable to anatase in many photoreac- series (all of the above studies used a single TiO2 support).
tions. Mixed phase TiO2 catalysts have also been widely studied, This work aimed to fill several key gaps in the scientific
owing to synergistic electron transfer phenomena that can be rea- literature concerning photocatalytic H2 production over Au/TiO2
lised in such systems. Degussa P25 TiO2 (a 6:1 mixture of anatase photocatalysts in alcohol–water systems. 1.5 wt.% Au/TiO2
and rutile by weight) typically demonstrates superior activity to photocatalysts were prepared using P25, anatase, brookite and
pure anatase for most photocatalytic reactions [30–32]. Transfer rutile supports, and their performance was evaluated for H2
of photo-excited electrons from the conduction band of rutile to production in 10 vol.% methanol–water, ethanol–water, ethylene
W.-T. Chen et al. / Journal of Catalysis 329 (2015) 499–513 501

glycol–water and glycerol–water mixtures under UV excitation to a 500 mL glass Schott bottle, then heated with constant stirring
(6.5 mW cm2). The objectives of this study were 3-fold: (i) to rank at 80 °C for 8 h. The obtained pale yellow powders (likely
the activity of the different Au/TiO2 photocatalysts for H2 produc- Au(OH)3/TiO2 or Au(OH)3xClx/TiO2) were collected by vacuum fil-
tion in each alcohol-water system; (ii) to rank the alcohol sacrifi- tration, washed repeatedly with milli-Q and then air dried at 60 °C
cial agents in terms of their ability to promote hydrogen overnight. The powders were then calcined at 300 °C for 2 h to
production; and (iii) to explore relationships between photocat- reduce surface Au(III) to Au0.
alytic H2 production rates and specific properties of the alcohols
(e.g. number of a-H, number of hydroxyl groups, alcohol polarity
2.5. TiO2 and Au/TiO2 photocatalyst characterisation
or the standard oxidation potential, Eoox).
UV–Vis absorbance spectra for the TiO2 powders and Au/TiO2
2. Experimental section powders were collected over the wavelength range 220–1400 nm
on a Shimadzu UV-2600 spectrophotometer fitted with an inte-
2.1. Materials grating sphere attachment. BaSO4 powder was used as a reference.
TEM images were collected using a TECNAI 12 transmission
P25 TiO2 (P99.5%), brookite (99.99%), titanium(IV) isopropox- electron microscope operated at an accelerating voltage of
ide (97%), 2-propanol (99.5%), HAuCl43H2O (99%), urea (P99.5%), 120 kV. Powder samples were dispersed in absolute ethanol and
HCl (34–37 wt.%) and glycerol (P99%), were all obtained from then 1 lL of the resulting dispersion was placed on carbon coated
Sigma–Aldrich and used without further purification. Methanol copper TEM grids for analysis.
(99.8%), absolute ethanol (P99.5%) and ethylene glycol (>95%) XRF data were taken on a Siemens SRS3000 spectrometer. Data
were obtained from ECP Ltd and used without further purification. reduction used Siemens SPECTRA 3000 software. Samples were
All solutions were prepared using milli-Q water (18.2 MX cm). analysed directly as powders supported on Mylar films.
Powder XRD patterns were taken on a PANalytical Empyrean
2.2. Synthesis of sol–gel anatase diffractometer equipped with a Cu anode X-ray tube and a curved
graphite filter monochromator. XRD data were collected from
An amorphous sol–gel TiO2 powder was synthesised by the 2h = 10–90° (step 0.02°, scan rate 2° min1) using Cu Ka X-rays
hydrolysis of titanium(IV) isopropoxide in 2-propanol [41]. (k = 1.5418 Å, 40 mA, 40 kV). Anatase, brookite and rutile crystal-
Briefly, titanium (IV) isopropoxide (28.4 g) and 2-propanol lite sizes (L) were determined from the powder XRD data using
(125 mL) were poured into a 500 mL glass Schott bottle. With vig- the Scherrer equation and line-widths of the anatase (1 0 1)
orous stirring, water (108.1 g) was slowly added dropwise to the reflection at 2h = 25.3°, the brookite (1 2 1) reflection at 2h = 30.8°
titanium isopropoxide solution over a period of 1 h at 20 °C. The and rutile (1 1 0) reflection at 2h = 27.4°, respectively [28,40]. The
solution was then stirred for 24 h, and the amorphous TiO2 powder rutile:anatase ratio in P25 TiO2 was determined according to the
obtained was collected by vacuum filtration and washed repeat- method described by Ding et al. [44].
edly with milli-Q water. The product was then dried at 70 °C over-
1
night to remove any residual water or 2-propanol. The dried %Rutile ¼  100 ð1Þ
amorphous TiO2 was then ground to a fine powder using a mortar ½1 þ 0:8ðIA =IR Þ
and pestle, before being calcined at 475 °C for 4 h. This calcination
where IA is the peak intensity for the anatase (1 0 1) reflection, and IR
treatment yielded an anatase powder with a crystallite size similar
is the peak intensity for the rutile (1 1 0) reflection.
to the anatase fraction in P25 TiO2 [42]. This anatase powder is
XPS data were collected using a Kratos Axis UltraDLD equipped
denoted as anatase SG in the text below.
with a hemispherical electron energy analyser and an analysis
chamber of base pressure 1  109 Torr. Spectra were excited
2.3. Hydrothermal anatase and rutile synthesis using monochromatic Al Ka X-rays (1486.69 eV), with the X-ray
source operating at 150 W. Samples were gently pressed into thin
The amorphous sol–gel TiO2 powder synthesised in Section 2.2 pellets of 0.1 mm thickness for the analyses. A charge neutralisa-
was also used to make nanocrystalline anatase and rutile powders tion system was used to alleviate sample charge build-up during
via hydrothermal (HT) routes. An anatase HT powder was prepared analysis. Survey scans were collected at a pass energy of 80 eV over
by adding the amorphous sol–gel TiO2 (1 g) to 50 mL of an etha- the binding energy range 1200–0 eV, whilst core level scans were
nol–water mixture (2:1 by volume), and then transferring the collected with a pass energy of 20 eV. The spectra were calibrated
resulting dispersion into a teflon-lined stainless steel autoclave against the C 1s signal at 285.0 eV from adventitious hydrocarbons.
(the total internal volume of the teflon cup was 100 mL). The auto- Ti L-edge and O K-edge NEXAFS data were collected on the soft
clave was then heated to 160 °C for 16 h. The obtained anatase X-ray Beamline at the Australian Synchrotron. NEXAFS data were
powder was collected by centrifugation, washed repeatedly with taken in the drain current mode at an analysis chamber pressure
milli-Q water, and dried in air at 100 °C. A rutile HT powder was of 1  1010 Torr. The drain current data were normalised against
prepared in a similar manner, using 4 mol L1 HCl as the dispersion a current measured simultaneously on a gold mesh in the beamline
medium and a higher reaction temperature of 175 °C. to eliminate potential spectral artefacts caused by fluctuations in
the beam intensity whilst scanning. The measurements were car-
2.4. Au/TiO2 photocatalyst synthesis ried out in high resolution (HR) mode by increasing the photon
energies in steps of 0.05 or 0.1 eV.
Au nanoparticles were deposited on the 5 different TiO2 sup- EXAFS analyses were carried out on beamline BL01B1 at the
ports (P25, brookite, anatase SG, anatase HT and rutile HT) at a Japan Synchrotron Radiation Research Institute (SPring-8). The
nominal Au loading of 1.5 wt.% using the deposition-precipitation storage ring was operated at 8 GeV with a ring current of 44–
with urea method described by Zanella et al. [43]. Briefly, a 65 mA for the duration of the experiments. Transmission mode
4.2 mM Au3+ stock solution was prepared by dissolving measurements were taken over the Au LIII-edge (11.7–13.3 keV)
HAuCl43H2O (1.65 g) in 1 L of milli Q water. TiO2 powder (2 g), region using a double crystal Si(1 1 1) monochromator. Samples
Au3+ stock solution (37.5 mL for target nominal Au loading of were pressed into 1.5 mm thick pellets of 7 mm diameter for the
1.5 wt.%), urea (5.04 g) and milli-Q water (162.5 mL) were added analysis.
502 W.-T. Chen et al. / Journal of Catalysis 329 (2015) 499–513

N2 physisorption isotherms were determined at liquid nitrogen pyrex reactor (105 mL volume). Photocatalyst (6.5 mg) was placed
temperature (77 K) using a Micromeritics Tristar 3000 instrument. in the reactor and flushed under a N2 flow for 30 min to remove
Specific surface areas were calculated from the N2 adsorption data oxygen. Then, 20 mL of a 10 vol.% alcohol–90 vol.% water mixture
according to the Brunauer–Emmett–Teller (BET) method using P/Po (alcohols tested were methanol, ethanol, ethylene glycol and glyc-
values in the range 0.05–0.2 [45]. Cumulative pore volumes and erol) was injected into the reactor through a rubber septum and
pore diameters were calculated from the adsorption isotherms by the resulting photocatalyst dispersion briefly sonicated then stir-
the Barrett–Joyner–Halenda (BJH) method [46]. Samples were red continuously for 1 h in the dark (no UV excitation). A 1 mL
degassed at 100 °C under vacuum for 1 h prior to the N2 physisorp- sample of the headspace gas was taken to confirm that no hydro-
tion measurements. gen formed by dark reactions, after which the reactor was exposed
Photoluminescence measurements were performed in air at to UV light supplied from a Spectroline model SB-100P/F lamp
room temperature using a Perkin–Elmer LS-55 Luminescence (100 W, 365 nm) at a distance of 10 cm from the reactor. The pho-
Spectrometer. Samples were excited at 310 nm and photolumines- ton flux at the sample was approximately 6.5 mW cm2. Hydrogen
cence spectra were recorded over a range of 330–600 nm using a evolution was monitored by taking gas headspace samples (1 mL)
standard photomultiplier. at 20 min intervals and injecting these into a Shimadzu GC 2014
equipped with a TCD detector and Carboxen-1010 plot capillary
2.6. Photocatalytic hydrogen production tests column (L  I.D. 30 m  0.53 mm, average thickness 30 lm). H2
evolved was quantified against an external calibration curve of
Photocatalytic hydrogen production tests on the Au/TiO2 photo- peak area versus moles of H2. Photocatalytic tests for each sample
catalysts and the bare TiO2 supports were carried out in a tubular were carried out in triplicate.

Table 1
Summarised structural, optical and chemical composition data for the different TiO2 supports and the 1.5 wt.% Au/TiO2 photocatalysts.

Photocatalyst Phase and crystallite Band gap BET area Au by XRF Mean Au size Au LSPR Au LSPR At.% by XPS Au:Ti
size (nm) (eV) (m2 g1) (wt.%) (nm) maximum (nm) Abs.
Ti O Ca Au
P25 TiO2 29 (A), 55 (R) 3.16 49.1 – – – – 22.3 52.5 25.2 – –
Anatase SG 23 (A) 3.18 41.4 – – – – 22.3 55.5 22.2 – –
Anatase HT 10 (A) 3.23 115.0 – – – – 24.5 53.8 21.7 – –
Brookite 26 (B) 3.26 32.9 – – – – 23.0 51.8 25.2 – –
Rutile HT 57 (R) 3.00 29.1 – – – – 23.4 56.1 20.5 – –
Au/P25 28 (A), 57 (R) 3.21 45.8 1.4 6±1 560 0.85 21.3 51.2 27.1 0.4 0.019
Au/anatase SG 25 (A) 3.24 39.4 1.4 7±2 590 0.80 20.6 49.9 29.1 0.4 0.019
Au/anatase HT 12 (A) 3.27 114.3 1.5 5±1 563 0.66 20.5 53.0 26.1 0.4 0.020
Au/brookite 26 (B) 3.33 33.7 1.4 10 ± 4 573 1.00 20.0 43.4 36.2 0.4 0.020
Au/rutile HT 60 (R) 3.03 29.2 1.6 7±3 556 0.60 24.4 56.7 18.4 0.5 0.021

A = anatase, B = brookite, R = rutile.


a
Adventitious hydrocarbons.

L3 L2 t2g
(a) (b) eg
t2g eg t2g eg
O (2p) - Ti (4sp)
dz 2
dx 2-y 2

O (2p) P25 TiO2


Intensity (arbitrary units)
Intensity (arbitrary units)

P25 TiO2
Anatase HT

anatase HT

Brookite

brookite

Rutile HT

rutile HT

455 460 465 470 475 530 540 550 560


Photon energy (eV) Photon energy (eV)

Fig. 1. (a) Ti L-edge NEXAFS data for selected TiO2 photocatalysts; and (b) O K-edge NEXAFS data for selected TiO2 photocatalysts. Anatase, brookite and rutile can readily be
distinguished on the basis of the Ti L-edge spectra.
W.-T. Chen et al. / Journal of Catalysis 329 (2015) 499–513 503

3. Results and discussion with the manufacturers’ specifications (6:1 anatase:rutile by


weight). Crystallite sizes were estimated from the linewidths of
3.1. TiO2 support characterisation the anatase (1 0 1), rutile (1 1 0) or brookite (1 2 1) reflections, using
the Scherrer equation. Results are summarised in Table 1, along
Fig. S1 shows powder X-ray diffraction patterns for the five dif- with other important optical and physico-chemical data for the
ferent TiO2 supports used in this study. The data confirm that the TiO2 supports.
anatase SG, anatase HT, brookite and rutile HT samples were all Ti L-edge and O K-edge NEXAFS measurements were used as a
pure, single phase titanias. The P25 TiO2 contained both anatase further tool for discriminating the anatase, brookite and rutile sup-
and rutile, with the anatase phase being dominant in accordance ports. As outlined in Section 1, the three TiO2 polymorphs are

(a)

1.8 1.8
3.0 anatase SG 3.0
(b) P25 TiO2 P25 TiO2 (c) Au/anatase SG anatase SG
2.5 1.6
(αhυ)1/2 (eV1/2 cm -1/2 )

1.6 Au/P25 Au/P25 2.5 Au/anatase SG


Au/anatase HT

(αhυ)1/2 (eV1/2 cm -1/2 )


Au/anatase HT
2.0 2.0
1.4 1.4
1.5 1.5

1.0
1.2 1.2 1.0

0.5
Absorbance

0.5
Absorbance

1.0 0.0
1.0
0.0
2.0 2.5 3.0 3.5 4.0 2.0 2.5 3.0 3.5 4.0
Energy (eV) Energy (eV)
0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0.0 0.0
400 600 800 1000 400 600 800 1000
Wavelength (nm) Wavelength (nm)

1.8 1.8
3.0 20
brookite (e) rutile HT
(d) 2.5
brookite
Au/brookite Au/rutile HT
rutile HT
Au/brookite Au/rutile HT
(αhυ)1/2 (eV1/2 cm -1/2 )

1.6 1.6
2

15
(αhυ)2 (eV cm -1)

2.0

1.4 1.5
1.4 10

1.0
1.2 1.2 5
0.5
Absorbance
Absorbance

1.0 0.0 1.0 0


2.0 2.5 3.0 3.5 4.0 2.0 2.5 3.0 3.5 4.0
Energy (eV) Energy (eV)
0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0.0 0.0
400 600 800 1000 400 600 800 1000
Wavelength (nm) Wavelength (nm)

Fig. 2. (a) Digital photograph of the 1.5 wt.% Au/TiO2 photocatalysts prepared using the different TiO2 supports. Corresponding UV–Vis absorbance spectra and Tauc plots
(insets) for (b) P25 TiO2 and Au/P25; (c) anatase SG, Au/anatase SG and Au/anatase HT; (d) brookite and Au/brookite; and (e) rutile HT and Au/rutile HT. The absorbance
spectrum for anatase HT is not shown, but was identical to that shown in (c) for anatase SG.
504 W.-T. Chen et al. / Journal of Catalysis 329 (2015) 499–513

(a) (b)

rutile

Au
anatase

(c) (d)

Au

Au

(e) (f)

Au
Au

Fig. 3. TEM images for (a) P25 TiO2; (b) Au/P25; (c) Au/anatase SG; (d) Au/anatase HT; (e) Au/brookite; and (f) Au/rutile HT. Scale bar = 50 nm.

constructed from TiO6 octahedra, but differ in the connectivity of by 5.8 eV which is equivalent to the spin orbit splitting between
the octahedra. The octahedra are the subject of Jahn-Teller distor- the occupied Ti 2p3/2 and 2p1/2 states in TiO2. The L3-edge (Ti
tions due to bonding and crystal packing effects in the solid state, 2p3/2 ? Ti 3d) and L2-edge (Ti 2p1/2 ? Ti 3d) features each show
which changes the crystal field about the Ti4+ centres and energy two sublevels (t2g and eg) which arise from crystal-field splitting
levels of the unoccupied Ti 3d states. Ti-L edge NEXAFS spectra caused by the octahedral ligand fields about the Ti4+ cations [47].
for P25 TiO2, anatase HT, brookite and rutile HT are shown in The pre-edge peaks are assigned to the particle–hole coupling
Fig. 1(a). Data for anatase SG support were identical to those between a 2p63d0 ground and 2p53d1 final-state configuration
shown for anatase HT support and thus excluded. The Ti L-edge [47,48]. The main difference between the three TiO2 polymorphs
spectra show two sets of peaks, denoted L3 and L2, and separated was seen in the eg feature at the L3 edge, which is split into two
W.-T. Chen et al. / Journal of Catalysis 329 (2015) 499–513 505

(a) (b)
Anatase (101)
d = 0.352 nm

Au Au

(c) (d)
Au

Au

(e) (f)

Au

Au

Fig. 4. HRTEM images for (a) and (b) Au/P25; (c) Au/anatase SG; (d) Au/anatase HT; (e) Au/brookite; and (f) Au/rutile HT. Scale bar = 10 nm.

unresolved components (i.e. dz2 and dx2y2 states). The eg levels accord with Ti L-edge data reported for mineral and synthetic tita-
have orbitals which are directed towards the oxygens. For anatase, nias [47,49–51]. The Ti L-edge spectrum of P25 (85% anatase, 15%
the dz2 peak is more intense than the dx2y2 peak, whereas for rutile) agreed with expectations, being effectively the weighted
rutile the relative peak intensities are reversed. This change in sum of the anatase and rutile spectra. The O K-edge NEXAFS spec-
the relative intensity of the dz2 and dx2y2 peaks on going from tra for all the TiO2 polymorphs showed two intense peaks between
anatase to rutile is explained by distortion of the Ti4+ site from 530 and 537 eV, separated by 2.6–2.7 eV (Fig. 1(b)]. These features
D2h (in anatase) to D2d (in rutile). For brookite, the dz2 and dx2y2 can readily be assigned to O 1s ? Ti 3d type transitions involving
peaks have similar intensities. These observations are in good unoccupied Ti 3d–O 2p hybrid orbitals [52,53], with the two peaks
506 W.-T. Chen et al. / Journal of Catalysis 329 (2015) 499–513

arising from the t2g and eg states of the Ti4+ octahedral field. stronger than that of Au with the other polymorphs. The rutile
Anatase, brookite and rutile differ in their spectral features at pho- HT support exposed a high proportion of (1 1 0) planes. A previous
ton energies between 538 and 550 nm, which involve transitions study by Bae et al. [63] demonstrated that Pt nanoparticles have a
from O 1s to O 2p-Ti 4sp hybrid states (O 1s ? 3a1g and 4t1u orbi- high affinity for rutile (1 1 0) faces. Hence, by analogy, we conclude
tals of TiO2) [53–55]. This region can be used to ‘‘fingerprint’’ the that Au nanoparticles also adsorb strongly on rutile (1 1 0) surfaces,
individual polymorphs. The O K-edge spectrum of P25 was the resulting in non-spherical Au particle shapes. A further important
weighted sum of the anatase and rutile O K-edge spectra. observation made for the all Au/TiO2 samples, and especially
Au/P25, was the preferential location of the Au nanoparticles on
3.2. Au/TiO2 photocatalyst characterisation interparticle heterojunctions. Such heterojunctions are known to
be catalytic ‘‘hot spots’’, and the preference of Au nanoparticles
Fig. 2(a) shows a digital photograph of the 5 different 1.5 wt.% for such heterojunctions has important implications for the photo-
Au/TiO2 photocatalysts prepared using P25, anatase SG, anatase catalytic activity of Au/TiO2, as we explore below. Collectively the
HT, brookite and rutile HT supports. All Au/TiO2 photocatalysts dis- data in Figs. 2–4 and Table 1 enable some meaningful conclusions
played a distinctive purple colour characteristic of nanocrystalline to be drawn regarding the optical properties of the Au/TiO2 photo-
Au0 on TiO2, though the intensity and shade of the purple colour catalysts. The Au/P25, Au/anatase HT and Au/brookite photocata-
varied considerably depending on the support. The purple colour lysts all exhibit quite spherical Au nanoparticle shapes, and
is due to the localised surface plasmon resonance (LSPR) of the accordingly gave symmetrical Au LSPR features (Fig. 2) at similar
supported Au nanoparticles, the wavelength of which is almost wavelengths (560, 563 and 573 nm, respectively). For these three
independent of Au particle size in the range 3–12 nm, but is very photocatalysts, the intensity of the LSPR feature decreased in the
sensitive to the dielectric constant of the surrounding medium order Au/brookite > Au/P25 > Au/anatase, reflecting a correspond-
[56–60]. Thus, it is not surprising that the Au/TiO2 photocatalysts ing decrease in the mean Au nanoparticle size (10 nm, 6 nm and
exhibited different optical responses. Au particle shape and 5 nm, respectively). For Au colloids in solution, the dependence
strength of the metal–support interaction are also important, and of the LSPR intensity on the Au particle size is well established
explored in detail below. [64], and appears to translate to supported Au nanoparticle sys-
Fig. 2(b)–(e) shows UV–Vis absorbance spectra and Tauc plots tems also (the Au loading by XRF was the same for all Au/TiO2 sam-
for the Au/TiO2 photocatalysts and their bare TiO2 supports. All ples and so the intensity differences are not simply the result of
supports showed intense absorption below 410 nm, due to different Au loadings). The LSPR feature for the Au/anatase SG pho-
band gap excitation in TiO2. Band gap energies determined from tocatalyst was much broader and red-shifted compared to the
the Tauc plots were P25 (Eg = 3.16 eV), anatase SG and anatase Au/anatase HT photocatalyst (Fig. 2(c)), and consistent with the
HT (Eg = 3.18–3.23 eV), brookite (Eg = 3.26 eV) and rutile atypical grey-purple colour of that particular sample (Fig. 2(a)).
(Eg = 3.00 eV). These values were all in excellent accord with previ-
ous literature reports for the P25 and the individual TiO2 poly-
morphs [23,24,60]. For the Au/TiO2 photocatalysts, an intense
LSPR absorption (Au 5d ? 6sp transitions) was seen that extended
from UV–NIR wavelengths, with a maximum centred between 560
and 590 nm. Au LSPR wavelengths and absorbances for the differ- (a)
ent samples are listed in Table 1. For unsupported spherical Au
nanoparticles of a particular size in solution, the LSPR redshifts lin-
early with increasing solvent refractive index [61]. Reported refrac-
tive indices for mineral anatase, brookite and rutile are 2.49, 2.58
and 2.61, respectively [60]. Thus for a constant Au nanoparticle size
on these supports, a redshift in the LSPR could be expected on (b)
going from Au/anatase to Au/brookite to Au/rutile. Indeed, the
Intensity (arbitrary units)

Au/rutile LSPR is predicted to be 40 nm higher than that of


Au/anatase at the same Au nanoparticle size [20,62]. However,
the fact that no such dependence of the Au LSPR maxima on the
dielectric properties of the support was observed here suggests (c)
that other factors, such as Au particle shape and bonding interac-
tion with the support were more influential. We examine these
factors below.
TEM images for P25 TiO2 and the 1.5 wt.% Au/TiO2 photocata- (d)
lysts are shown in Fig. 3. All images were taken at the same mag-
nification to allow direct comparison of the samples regarding the
relative size and size distribution of the TiO2 crystallites and sup-
ported Au nanoparticles. Fig. 4 shows higher resolution TEM
images for the Au/TiO2 photocatalysts, and provides more precise (e)
information concerning the bonding interaction of the individual
Au nanoparticles with each TiO2 support. Mean Au particle sizes
determined from the TEM images are listed in Table 1, and ranged
between 5 and 10 nm (the mean Au sizes are the average of mea- (f)
surements performed on >50 individual Au nanoparticles). For
Au/P25, Au/anatase and Au/brookite, the Au nanoparticles were
20 30 40 50 60 70 80
mainly spherical with contact angles greater than 100° on the
2θ (degrees)
TiO2 surface. For Au/rutile HT, the contact angles were typically
less than 100° and the particle shape is more lozenge-like suggest- Fig. 5. Powder XRD patterns for (a) Au/P25; (b) Au/anatase SG; (c) Au/anatase HT;
ing the metal-support interaction (MSI) for Au with rutile is (d) Au/brookite; (e) Au/rutile HT; and (f) Au foil.
W.-T. Chen et al. / Journal of Catalysis 329 (2015) 499–513 507

Since Au/anatase SG and Au/anatase HT have similar mean Au par- with no surface cleaning before XPS analyses. The Au:Ti atom ratio
ticle sizes (7 and 5 nm, respectively), the differences in their for the Au/TiO2 photocatalysts was similar (0.019–0.021), corre-
absorption spectra must be due to factors other than Au particle sponding to ‘‘XPS-derived’’ Au loadings of 3.2–4.5 wt.%. This is
size. One plausible explanation is that in the Au/anatase SG sample, clearly an overestimation of the true Au loading (1.4–1.6 wt.% by
the individual Au nanoparticles differ considerably in the extent of XRF for all samples, Table 1), which arises from the fact that XPS
their contact with the anatase support. If the Au nanoparticles only probes the top few nanometres of the Au/TiO2 powders where
interact with multiple TiO2 crystallites, and to different extent Au is preferentially located. The Ti 2p spectra for the Au/TiO2 pho-
(the anatase SG support had a far broader anatase crystallite size tocatalysts are shown in Fig. 6(a), and show peaks at 458.9 eV
distribution than anatase HT), then this could broaden/redshift 464.6 eV in a 2:1 area ratio that are readily assigned to photoemis-
the Au LSPR signal appreciably due to the high refractive index of sion from the Ti 2p3/2 and Ti 2p1/2 states of Ti4+ in TiO2 [65]. No dif-
TiO2 and the sensitivity of the LSPR to the surrounding dielectric ferences could be discerned between Ti 2p spectra of the different
medium. For the Au/rutile sample, the LSPR feature was distinctly Au/TiO2 photocatalysts. Au 4f XPS measurements confirmed that
asymmetric on the long wavelength side of the LSPR peak maxi- metallic Au was the only surface gold species (Fig. 6(b)). The Au
mum, which can be rationalised in terms of the lozenge-like Au 4f spectra for all photocatalysts showed peaks at 83.8 and
nanoparticle shape and the non-equivalence of transverse and lon- 87.5 eV in a 4:3 area ratio, which are assigned to Au 4f7/2 and Au
gitudinal LSPRs [64]. 4f5/2 peaks, respectively, of metallic Au0 species on TiO2 [65]. The
XRD patterns for the Au/TiO2 photocatalysts (Fig. 5) were very Au 4f peaks were observed at a slightly lower binding energy than
similar to those of the bare TiO2 supports (Fig. S1), confirming that the same peaks for a metallic Au foil (Au 4f7/2 = 84.0 eV, not
Au deposition had no effect on the TiO2 supports in terms of phase shown). The small shift could be due to the transfer of electron
composition or mean crystallite size. Additional very weak and density from the oxide support to the Au nanoparticles, as has been
broad peaks could be discerned in the XRD patterns of the reported elsewhere [66], or alternatively to the presence of hydro-
Au/TiO2 photocatalysts, which were consistent with the presence xyl groups which modify the electronic properties of TiO2. Au
of supported Au nanoparticles possessing a fcc structure similar L3-edge EXAFS examines electronic transitions from Au 2p3/2 core
to the Au reference powder. The Au(1 1 1) and Au(2 0 0) reflections levels into unoccupied d-like states near the Fermi level of the
at 2h = 38.2° and 44.4°, respectively, were the most conspicuous of absorbing Au atoms, and is a sensitive probe of coordination envi-
the Au peaks. ronment and next nearest neighbour effects [67,68]. Au L3-edge
XPS and Au L3-edge EXAFS were used to probe the speciation of EXAFS data for 1.5 wt.% Au/P25 TiO2 and its uncalcined precursor
Au on the surface of the Au/TiO2 photocatalysts, in particular to are shown in Fig. 7(a). Note that the EXAFS data presented here
confirm the complete reduction of Au(III) species, such as initially for the P25-based photocatalysts are representative of data
deposited Au(OH)3xClx or Au(OH)4xClx, to Au0. Near surface obtained for all the Au/TiO2 photocatalysts and their uncalcined
region chemical compositions, including Au:Ti atom ratios, deter- precursors. An intense peak at 11919 eV was seen in the Au L3
mined for the Au/TiO2 photocatalysts from XPS survey spectra absorbance spectrum of the Au/P25 precursor, which is character-
are shown in Table 1. The quantification procedure used peak areas istic for Au3+ species, and arises from Au 2p3/2 ? 5d electronic
for the Ti 2p, O 1s, C 1s and Au 4f regions and appropriate transitions [69]. This feature disappeared after sample calcination
instrument-modified Scofield cross sections. The C 1s signal origi- at 300 °C, with the resulting absorption spectrum being character-
nates from adventitious hydrocarbons. The O:Ti ratio determined istic for metallic Au (5d106s1) and almost indistinguishable from
for the bare TiO2 supports and Au/TiO2 photocatalysts ranged the Au foil reference spectrum. The r-space plots for the Au/P25
between 2.2 and 2.6, as expected for TiO2-based photocatalysts precursor showed peaks at 1.1 and 1.9 Å (Fig. 7(b)), which can

(a) Ti 2p ) (b) Au 4f )

Au/P25
Intensity (arbitrary units)
Intensity (arbitrary units)

Au/P25

Au/anatase HT

Au/anatase HT

Au/brookite

Au/brookite

Au/rutile HT
Au/rutile HT

470 465 460 455 92 90 88 86 84 82 80


Binding energy (eV) Binding energy (eV)

Fig. 6. (a) Ti 2p XPS spectra for selected photocatalysts; and (b) Au 4f XPS spectra for selected photocatalysts. The vertical lines added at 458.8 eV and 83.8 eV in the Ti 2p and
Au 4f spectra, respectively, show that the electronic properties of all the Au/TiO2 photocatalysts are similar.
508 W.-T. Chen et al. / Journal of Catalysis 329 (2015) 499–513

brookite < anatase < P25 < rutile. This order reflects the relative
(a) band gap energies of the supports (Eg = 3.26, 3.18, 3.16 and
Au reference foil 3.00 eV, respectively, Table 1). The photoluminescence signal
arises from electron de-excitation across the band gap to annihilate
Normalized absorbance (arbitrary units)

holes in the valence band. Deposition of Au nanoparticles on TiO2


was expected to create a rectifying Schottky junction and suppress
electron–hole pair recombination, since the Fermi level of sup-
ported 5 nm Au nanoparticles is 0.27 V versus NHE, more positive
Au/P25 than the bottom of the conduction bands of anatase (0.5 V versus
NHE), brookite (0.6 V versus NHE) and rutile (0.7 V versus NHE).
Photo-excited electrons reaching the surface of TiO2 will thus
migrate onto the Au nanoparticles, thereby attenuating the PL sig-
nal [19,20,32,56]. The experimental data of Fig. 8 supported this
argument, with the PL signals for Au/P25, Au/anatase SG,
Au/P25 precursor Au/brookite and Au/rutile being 71%, 45%, 63% and 40% weaker,
respectively, than the PL intensities of the corresponding bare
TiO2 supports. The PL spectrum for Au/anatase HT is not shown,
but was attenuated by 48% after Au deposition.

3.3. Photocatalytic hydrogen production tests


11850 11900 11950 12000 12050 12100
Photon energy (eV) The activity of the bare TiO2 supports and the Au/TiO2 photocat-
alysts for H2 production was evaluated in a range of different alco-
hol–water mixtures under UV excitation, at a fixed alcohol
(b) concentration of 10 vol.%. Alcohols used were methanol, ethanol,
ethylene glycol and glycerol, and were selected so the effects of
alcohol structure (1°, diol or triol) and alcohol physical properties
(especially polarity and oxidation potential) on the H2 production
Au reference foil rate could be examined. The bare TiO2 supports exhibited very
Amplitude (arbitrary units)

low activities for H2 production in all the alcohol–water mixtures,


with the activity of P25 being fractionally superior to the other
TiO2 supports. H2 production rates determined for P25 are sum-
marised in Table 2, and ranged from 1.3 to 1.9 mmol g1 h1. The
low rates observed for the bare TiO2 supports can be attributed
Au/P25 to rapid electron–hole recombination following photoexcitation,
and the lack of H:H recombination or H+ reduction sites. The
Au/TiO2 photocatalysts demonstrated significantly improved H2
production activities (Figs. 9 and 10, Table 2), with the activity by
catalyst mass (in mmol g1 h1) decreasing in the order Au/P25
TiO2 > Au/anatase SG  Au/anatase HT  Au/brookite > Au/rutile
Au/P25 precursor HT in all the alcohol–water mixtures tested. The activity did not
change much if normalised to unit area of the semiconductor
(Fig. 10(b)), with the activities following the order Au/P25
0 1 2 3 4 5 6 TiO2 > Au/anatase SG  Au/brookite > Au/anatase HT  Au/rutile
Radial distance (Å) HT. The fact that the Au/P25 TiO2 photocatalyst afforded the high-
est H2 production rates is easily understood in terms of increased
Fig. 7. (a) Au L3-edge EXAFS spectra for different Au-containing samples including
charge carrier availability, resulting from synergistic electron and
Au/P25; and (b) corresponding r-space plots obtained from Fourier transforms on
k3-weighted Au L3-edge EXAFS spectra. hole transfer across anatase–rutile heterojunctions that was
described in Section 1. For photocatalytic H2 production, the pres-
ence of Au nanoparticles on anatase–rutile heterojunctions is
be assigned to Au–Cl and Au–O bond distances, respectively [69]. highly beneficial [32,33], since electrons photo-excited into the
The data support the hypothesis that the species formed on TiO2 conduction bands of rutile or anatase can readily migrate onto
using the deposition-precipitation with urea method was the Au nanoparticles which in turn act as cathodic centres for H2
Au(OH)3xClx or Au(OH)4xClx, not surprising given that evolution. The lower H2 production activities observed for the
HAuCl43H2O was the Au salt used and the final pH in the synthesis Au/anatase SG and Au/brookite photocatalysts, compared to
was 9.5 [69]. The r-space plot for the Au/P25 photocatalyst was Au/P25, can be rationalised in terms of lower charge carrier con-
dominated by a peak at 2.86 Å, which is the characteristic Au– centrations since both anatase and brookite are indirect band gap
Au bond distance in fcc gold [68,69]. The EXAFS transmittance semiconductors. The activity of the Au/anatase HT sample was
measurements thus provided further confirmation that metallic low per unit area compared to Au/anatase SG, which can be attrib-
Au0 was the only Au species on the surface of the Au/TiO2 uted to the small crystallite size (10–12 nm) and low crystallinity
photocatalysts. in the anatase HT support as seen by XRD (Table 1). For some
Photoluminescence (PL) spectra collected in air for the different photocatalysts, small crystallite sizes can be beneficial for
TiO2 supports and the Au/TiO2 photocatalysts are presented in photoreactions by reducing diffusion lengths, though small
Fig. 8. Each of the bare TiO2 supports gave an intense crystallites usually have high defect concentrations which can
photoluminescence signal following UV excitation, with the serve as electron–hole pair recombination sites. The anatase SG
emission maximum wavelength red-shifting in the order supports appear to have a near ideal anatase crystallite size for
W.-T. Chen et al. / Journal of Catalysis 329 (2015) 499–513 509

800 800
P25 anatase HT
(a)) 1.5 wt.% Au/P25 (b)) 1.5 wt.% Au/anatase
1.5 wt.% Au/anatase HT
Photoluminescence intensity

Photoluminescence intensity
600 600

400 400

200 200

0 0
3.6 3.4 3.2 3.0 2.8 2.6 2.4 2.2 3.6 3.4 3.2 3.0 2.8 2.6 2.4 2.2
Energy (eV) Energy (eV)

800 800
brookite
rutile HT
(c)) 1.5 wt.% Au/brookite (d)) 1.5 wt.% Au/rutile HT
Photoluminescence intensity

Photoluminescence intensity
600 600

400 400

200 200

0 0
3.6 3.4 3.2 3.0 2.8 2.6 2.4 2.2 3.6 3.4 3.2 3.0 2.8 2.6 2.4 2.2
Energy (eV) Energy (eV)

Fig. 8. Photoluminescence spectra for (a) P25 TiO2 and Au/P25; (b) anatase HT and Au/anatase HT; (c) brookite and Au/brookite; and (d) rutile HT and Au/rutile HT. PL data for
anatase SG (not shown) and Au/anatase SG were similar to those shown in (b). The sharp features at low energies in the PL spectra are due to the instrument.

Table 2
Summarised H2 production data for the 1.5 wt.% Au/TiO2 photocatalysts in different alcohol–water mixtures. All H2 production tests were carried out at an alcohol concentration
of 10 vol.% and a UV flux of 6.5 mW cm2.

Photocatalyst H2 production rate at an alcohol concentration of 10 vol.%


Methanol Ethanol Ethylene glycol Glycerol
(mmol g1 h1) (mmol m2 h1) (mmol g1 h1) (mmol m2 h1) (mmol g1 h1) (mmol m2 h1) (mmol g1 h1) (mmol m2 h1)
P25 TiO2 1.4 0.028 1.3 0.026 1.4 0.029 1.9 0.039
Au/P25 13.5 0.294 9.8 0.213 20.9 0.456 27.9 0.609
Au/anatase SG 7.0 0.179 6.6 0.168 12.0 0.304 15.5 0.393
Au/anatase HT 8.5 0.074 7.3 0.064 12.0 0.105 15.0 0.131
Au/brookite 6.7 0.199 4.9 0.156 11.0 0.325 13.8 0.410
Au/rutile HT 0.9 0.031 0.4 0.015 1.8 0.062 3.2 0.109

photocatalytic H2 production. The Au/rutile photocatalyst exhib- that all the Au/TiO2 photocatalysts were stable during the photo-
ited the worst performance, reflecting the fact that rutile is a direct catalytic H2 production tests, as evidenced by the strong linearity
band gap semiconductor. Fast electron–hole pair recombination of H2 production with time in all alcohol–water mixtures.
following photo-excitation results in low charge carrier concentra- Previous studies of photocatalytic H2 production in the Au/P25,
tions for photoreactions. Other considerations, such as the faces Au/anatase and Au/rutile systems have demonstrated that H2 pro-
exposed on the HT rutile sample, comprising mainly rutile (1 1 0) duction rates are independent of Au nanoparticle size in the range
and (1 1 1), may also contribute to the low activity. Fig. 9 shows 3–12 nm, so differences in the activity seen in Figs. 9 and 10 are not
510 W.-T. Chen et al. / Journal of Catalysis 329 (2015) 499–513

90 90
methanol methanol
80 ethanol (a) Au/P25 ) 80 ethanol
(b) Au/anatase SG )
ethylene glycol ethylene glycol
glycerol glycerol
70 70
H 2 produced (mmol g )

H 2 produced (mmol g )
-1

-1
60 60

50 50

40 40

30 30

20 20

10 10

0 0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0
Time (h) Time (h)

90 20
methanol (c) Au/brookite methanol (d) Au/rutile HT
80 ethanol ethanol
ethylene glycol ethylene glycol
70 glycerol glycerol
15
H 2 produced (mmol g )

H 2 produced (mmol g )
-1

-1

60

50
10
40

30
5
20

10

0 0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0
Time (h) Time (h)

Fig. 9. Plots of H2 production versus time for selected Au/TiO2 photocatalysts in 10 vol.% aqueous solutions of methanol, ethanol, ethylene glycol or glycerol under UV
irradiation. (a) Au/P25; (b) Au/anatase SG; (c) Au/brookite; and (d) Au/rutile HT. Stable H2 production rates were observed for each Au/TiO2 photocatalyst in all alcohol–water
systems.

due to Au particle size effects (Table 1). Table S1 summarises some It is of interest to examine the relative H2 production rates in
photocatalytic H2 production data reported previously by different terms of the structure and physical properties of the different alco-
research groups for Au/TiO2 photocatalysts in alcohol or alcohol– hols. Table 3 summarises key physical properties for the four alco-
water mixtures under UV or UV–Vis excitation. The data in hols. Fig. 11 shows plots of H2 production for each Au/TiO2
Table S1 provide a useful benchmark to that reported in the cur- photocatalyst versus selected physical properties of the alcohols.
rent study. In all cases, reasonable linear relationships can be established,
Comparing the data in Fig. 10 and Table 2, it is evident H2 pro- the basis for which we discuss below. The photoreforming of the
duction rates were strongly dependent on the alcohol used as the alcohols can be represented by the following general equation
sacrificial hole scavenger, with the rate decreasing in the general [21]:
order glycerol > ethylene glycol > methanol > ethanol for all the
Cx Hy Oz þ ð2x  zÞH2 O ! xCO2 þ nHþ þ ne ð2Þ
Au/TiO2 photocatalysts. The Au/P25 photocatalyst afforded
the highest rates, 27.9 mmol g1 h1 in 10 vol.% glycerol, where x, y and z are the number of carbon, hydrogen and oxygen
20.9 mmol g1 h1 in 10 vol.% ethylene glycol, 13.5 mmol g1 h1 atoms, respectively, in the alcohol and n is the number of protons
in 10 vol.% methanol and 9.8 mmol g1 h1 in 10 vol.% ethanol. or electrons liberated. In aqueous media, the photoreaction will
We determined the H2 production rate for the Au/P25 photocata- proceed via the formation of alpha hydroxyl radicals, which have
lyst in pure water to be 0.25 mmol g1 h1; thus, the addition of been identified previously on TiO2 by EPR [18,70].
the alcohols to water increased the H2 evolution rates by 40–112 þ
times depending on the alcohol. The relative rates in the different RCH2  OHðaÞ þ h ðVBÞ ! RCH:  OHðaÞ þ Hþ ð3Þ
alcohol–water mixtures were remarkably similar for each Au/TiO2 þ þ
R2 CH  OHðaÞ þ h ðVBÞ ! R2 C:  OHðaÞ þ H ð4Þ
photocatalyst, and can be considered independent of the TiO2 poly-
morph (at least at the alcohol concentration and Au loading used in By this mechanism, it is easy to envisage why the presence of a-H
this study). This is a key finding of the current study. atoms on the alcohol is critical to achieving high rates of H2
W.-T. Chen et al. / Journal of Catalysis 329 (2015) 499–513 511

30 Considering the physical adsorption of the alcohols and alpha


(a) 10 vol.% methanol hydroxyl radicals on the TiO2 surface, the interaction of lone pairs
10 vol.% ethanol on the OH groups with unoccupied Ti 3d states is important. This
25 10 vol.% ethylene glycol largely explains the dependence of the hydrogen production rates
10 vol.% glycerol on the number of OH groups on the alcohols [Fig. 11(b)], and further
H 2 production rate (mmol g h )

pointed to a probable relationship between the H2 production rates


-1

and alcohol polarity. Fig. 11(c) confirmed the existence of a relation-


-1

20
ship for all Au/TiO2 photocatalysts, in accord with the findings of
Yang et al. for the Pt/TiO2 system [36]. A similar relationship could
potentially be drawn between the H2 production rates and alcohol
15 polarisability (Table 3). In Fig. 11(a)–(c), methanol showed slightly
different behaviour to the other alcohols, affording a H2 production
rate lower than expected. C1 chemistry often differs from that of C2,
10 C3, and longer chain organics, and that appears to be the case
here. For methanol, photocatalytic oxidation to formaldehyde
then CO may have partially poisoned the photocatalyst
surface, with the kinetics of the photocatalytic water-gas shift
5
reaction (CO + H2O ? CO2 + H2) impacting the overall rate of H2
production [71].
Using standard Gibbs free energies of formation (DfGo) of each
0 alcohol, we calculated the standard oxidation potentials (Eoox) of
5

HT
kite

methanol, ethanol, ethylene glycol and glycerol (see Table S2).


HT
P2

SG
/P2

tile
oo
se

The calculations gave values of 0.016, 0.084, 0.009 and 0.004 V ver-
Au

se

/br
ata

/ru
ata

sus NHE, respectively, all significantly lower than the oxidation


Au

Au
/an
/an

potential of water (1.23 V versus NHE). Hence, during photocatal-


Au
Au

ysis in alcohol–water systems, the alcohols will be preferentially


0.7 oxidised, though water will also be oxidised through direct water
(b) 10 vol.% methanol splitting and alcohol photo-reforming. Glycerol and ethylene glycol
Area normalized H2 production rate (mmol m h )

afforded the highest H2 production rates for all Au/TiO2 photocat-


-1

10 vol.% ethanol
0.6 alysts (Table 2), which is not surprising given they have the lowest
-2

10 vol.% ethylene glycol


10 vol.% glycerol oxidation potentials, making them the best sacrificial hole scav-
engers. According to current theories of electron transfer reactions
0.5
between an electron donor and valence band holes in a semicon-
ductor [72], the experimental rate constant (kexp) for such reac-
0.4 tions follows the relation
ðEo Eoox Þ=RT
kexp / exp VBðTiO2 Þ
ð5Þ
0.3
where EoVBðTiO2 Þ and Eoox are the valence band potential of TiO2 and the
oxidation potential of the donor (water of alcohol), respectively. If
0.2 electron transfer from water or alcohols to TiO2 was the overall rate
limiting step in photocatalytic H2 production over M/TiO2 surfaces,
then H2 production rates could be expected to correlate with
0.1  
 Eo Eoox
exp VBðTiO2 Þ . In Fig. 11(d) we have plotted H2 production rates
against the exponential of energy change for the alcohol oxidation
0.0
reaction; [exp(DEo)], where DEo ¼ EoVBðTiO2 Þ  Eoox . The analysis con-
5

HT
kite
T
P2

SG
/P2

firms a linear relationship between the H2 production rates and


H

tile
oo
se
Au

se

exp(DEo) for all the Au/TiO2 photo-catalysts (data for ethanol


/br
ata

/ru
ata

Au

Au

were excluded since it did not have a O:C ratio = 1 like the other
/an
/an

Au
Au

alcohols). We can conclude that the larger the potential gap


between the TiO2 valence band (2.7 V versus NHE for anatase)
Fig. 10. (a) H2 production rates for the Au/TiO2 photocatalysts in 10 vol.% aqueous
and Eoox, the higher the H2 production rate. Further, the data suggest
solutions of methanol, ethanol, ethylene glycol or glycerol. (b) Surface area
normalised H2 production rates for Au/TiO2 photocatalysts in 10 vol.% aqueous that electron transfer reactions between the alcohols (or water) and
solutions of methanol, ethanol, ethylene glycol or glycerol. valence band holes of TiO2 may be rate limiting in H2 production on
Au/TiO2 surfaces.
Currently we are attempting to quantify other products formed
production (a study we performed using tertiary butanol as the sac- during the H2 production tests, including O2, CO, CO2, CH4, C2H4,
rificial hole scavenger, which has no a-H, gave a low H2 production C2H6 and CH3CHO in the case of the ethanol–water system, to bet-
rate of 1.2 mmol g1 h1 for Au/P25). By plotting H2 production ter understand the photoreaction mechanism(s) and also to deter-
rates versus the number of a-H atoms or number of OH groups mine the relative contributions of water splitting and alcohol
on the alcohol, reasonable linear relationships were obtained for photoreforming to the high H2 yields reported here. We are also
all the Au/TiO2 photocatalysts (Fig. 11(a) and (b), respectively.) extending the current study by conducting photocatalytic tests
Bowker and co-workers previously identified the importance of under visible and full solar excitation, to determine the extent to
a-H adjacent to the OH groups as being important for achieving which the plasmonic effect can enhance H2 production rates
high rates of H2 production in the Pd/TiO2 system [19]. [70,73].
512 W.-T. Chen et al. / Journal of Catalysis 329 (2015) 499–513

Table 3
Summarised physical properties of the alcohol hole scavengers used in the H2 production tests. H2 production rates for Au/P25 taken from Table 2 are also included.

Alcohol No. of No. of O/C Permittivity Polaritya Refractive Polarisabilityb Alcohol oxidation H2 production rate
a-H OH ratio index potential (V)c (mmol g1 h1)
Methanol 3 1 1 32.7 0.914 1.328 0.203 0.016 13.5
Ethanol 2 1 0.5 24.6 0.887 1.361 0.221 0.084 9.8
Ethylene glycol 4 2 1 37.7 0.924 1.429 0.258 0.009 20.9
Glycerol 5 3 1 47.0 0.939 1.475 0.282 0.004 27.9
a
Polarity, Y = (es  1)/(es + 2).
b
Polarisability, p = (n2  1)/(n2 + 2).
c
Alcohol oxidation potentials were calculated from Gibbs free energies of reaction (see Table S2).

35 35
1.5 wt.% Au/P25
(a) 1.5 wt.% Au/P25
1.5 wt.% Au/anatase 475
(b) 1.5 wt.% Au/anatase 475
30 1.5 wt.% Au/anatase HT 30 1.5 wt.% Au/anatase HT
H 2 production rate (mmol g h )

1.5 wt.% Au/brookite glycerol


-1

1.5 wt.% Au/brookite glycerol

H 2 production rate (mmol g h )


1.5 wt.% Au/rutile HT

-1
1.5 wt.% Au/rutile HT
-1

25 25

-1
ethylene glycol
ethylene glycol

20 20

15 15 methanol
methanol
ethanol
10 10 ethanol

5 5

0 0
2 3 4 5 1.0 1.5 2.0 2.5 3.0
Number of alpha-hydrogens Number of OH groups

35 35
1.5 wt.% Au/P25
(c) 1.5 wt.% Au/anatase 475
(d) 1.5 wt.% Au/P25
1.5 wt.% Au/anatase 475
30 1.5 wt.% Au/anatase HT glycerol 30 glycerol 1.5 wt.% Au/anatase HT
H 2 production rate (mmol g h )

H 2 production rate (mmol g h )


-1

1.5 wt.% Au/brookite 1.5 wt.% Au/brookite


-1

1.5 wt.% Au/rutile HT 1.5 wt.% Au/rutile HT


-1

-1

25 25 ethylene glycol
ethylene glycol

20 20

methanol
15 methanol 15

ethanol
10 10

5 5

0 0
0.88 0.89 0.90 0.91 0.92 0.93 0.94 0.95 0.0674 0.0676 0.0678 0.0680 0.0682 0.0684
o o
Alcohol polarity − (E VB ( TiO 2 ) − Eox )
exp

Fig. 11. H2 production rates for the Au/TiO2 photocatalysts versus (a) number of a-H on the alcohol; (b) number of OH groups on the alcohol; (c) alcohol polarity; and (d)
 
 Eo Eoox
VBðTiO2 Þ
exp . Data for ethanol was excluded from (d).

4. Conclusion heterojunctions, and reduce electron–hole pair recombination


rates in TiO2 by accepting photo-excited electrons from the con-
The activity of Au/TiO2 photocatalysts for H2 production in alco- duction band and creating cathodic sites for H2 evolution, thereby
hol–water mixtures under UV excitation is strongly dependent on explaining their positive role in promoting H2 generation on each
the polymorphic composition of the TiO2 support. Au/TiO2 photo- TiO2 support. H2 production rates were highest for the Au/P25
catalyst activity decreased in the order P25 (anatase + rutile) > ana- photocatalyst (27.9 mmol g1 h1 in 10 vol.% glycerol), and
tase  brookite  rutile for each alcohol–water system tested, decreased with the alcohol in the order glycerol > ethylene gly-
which can be rationalised in terms of the relative charge carrier col > methanol > ethanol. For all the Au/TiO2 photocatalysts, corre-
concentrations in each TiO2 support under UV irradiation. lations could be established between the H2 production rate and
Au nanoparticles demonstrate a preference for interparticle key physical properties of the alcohols, including the number of
W.-T. Chen et al. / Journal of Catalysis 329 (2015) 499–513 513

a-H or OH groups on the alcohol, alcohol polarity and more [29] L. Liu, H. Zhao, J.M. Andino, Y. Li, ACS Catal. 2 (2012) 1817–1828.
  [30] K. Woan, G. Pyrgiotakis, W. Sigmund, Adv. Mater. 21 (2009) 2233–2239.
 Eo Eoox [31] R.I. Bickley, T. Gonzalez-Carreno, J.S. Lees, L. Palmisano, R.J. Tilley, J. Solid State
VBðTiO2 Þ
importantly on exp as suggested from electron Chem. 92 (1991) 178–190.
transfer theory. [32] V. Jovic, W.-T. Chen, D. Sun-Waterhouse, M.G. Blackford, H. Idriss, G.I.N.
Waterhouse, J. Catal. 305 (2013) 307–317.
[33] D.O. Scanlon, C.W. Dunnill, J. Buckeridge, S.A. Shevlin, A.J. Logsdail, S.M.
Acknowledgments Woodley, C.R.A. Catlow, M.J. Powell, R.G. Palgrave, I.P. Parkin, Nat. Mater. 12
(2013) 798–801.
EXAFS experiments were performed on the BL01B1 (XAFS) [34] Q. Xu, Y. Ma, J. Zhang, X. Wang, Z. Feng, C. Li, J. Catal. 278 (2011) 329–335.
[35] Q. Tay, X. Liu, Y. Tang, Z. Jiang, T.C. Sum, Z. Chen, J. Phys. Chem. C 117 (2013)
beamline at SPring-8 with the approval of the Japan Synchrotron
14973–14982.
Radiation Research Institute (JASRI) (Proposal No. 2014B1065). [36] Y.Z. Yang, C.H. Chang, H. Idriss, Appl. Catal. B: Environ. 67 (2006) 217–222.
The authors would like to acknowledge funding support from [37] Y. Li, B. Wang, S. Liu, X. Duan, Z. Hu, Appl. Surf. Sci. 324 (2015) 736–744.
[38] J. Wang, P. Yang, B. Cao, J. Zhao, Z. Zhu, Appl. Surf. Sci. 325 (2015) 86–90.
Saudi Basic Industries Corporation (SABIC), the University of
[39] R. Su, R. Tiruvalam, A.J. Logsdail, Q. He, C.A. Downing, M.T. Jensen, N.
Auckland FRDF fund, the Energy Education Trust of New Zealand, Dimitratos, L. Kesavan, P.P. Wells, R. Bechstein, H.H. Jensen, S. Wendt, C.R.A.
the MacDiarmid Institute for Advanced Materials and Catlow, C.J. Kiely, G.J. Hutchings, F. Besenbacher, ACS Nano 8 (2014) 3490–
Nanotechnology, the Dodd-Walls Centre for Photonic and 3497.
[40] U. Pal, S. Ghosh, D. Chatterjee, Transition Met. Chem. 37 (2012) 93–96.
Quantum Technologies and the Ministry of Higher Education at [41] A. Zachariah, K. Baiju, S. Shukla, K. Deepa, J. James, K. Warrier, J. Phys. Chem. C
the Sultanate of Oman. 112 (2008) 11345–11356.
[42] W.-T. Chen, A. Chan, V. Jovic, D. Sun-Waterhouse, K.-I. Murai, H. Idriss, G.I.N.
Waterhouse, Top. Catal. 58 (2015) 85–102.
Appendix A. Supplementary data [43] R. Zanella, S. Giorgio, C.R. Henry, C. Louis, J. Phys. Chem. B 106 (2002) 7634–
7642.
[44] X.-Z. Ding, X.-H. Liu, Y.-Z. He, J. Mater. Sci. Lett. 15 (1996) 1789–1791.
Supplementary data associated with this article can be found, in
[45] S. Brunauer, P.H. Emmett, E. Teller, JACS 60 (1938) 309–319.
the online version, at http://dx.doi.org/10.1016/j.jcat.2015.06.014. [46] E.P. Barrett, L.G. Joyner, P.P. Halenda, JACS 73 (1951) 373–380.
[47] P. Guttmann, C. Bittencourt, S. Rehbein, P. Umek, X. Ke, G. Van Tendeloo, C.P.
References Ewels, G. Schneider, Nat. Photon 6 (2012) 25–29.
[48] P. Krüger, Phys. Rev. B 81 (2010) 125121.
[49] G.S. Henderson, X. Liu, M.E. Fleet, Phys. Chem. Miner. 29 (2002) 32–42.
[1] J.D. Holladay, J. Hu, D.L. King, Y. Wang, Catal. Today 139 (2009) 244–260. [50] J.P. Crocombette, F. Jollet, Phys. Rev. B: Condens. Matter 6 (1994) 10811.
[2] G.W. Crabtree, M.S. Dresselhaus, M.V. Buchanan, Phys. Today 57 (2004) 39–44. [51] F.M.F. de Groot, M.O. Figueiredo, M.J. Basto, M. Abbate, H. Petersen, J.C. Fuggle,
[3] M. Balat, Int. J. Hydrogen Energy 33 (2008) 4013–4029. Phys. Chem. Miner. 19 (1992) 140–147.
[4] X. Chen, C. Li, M. Grätzel, R. Kostecki, S.S. Mao, Chem. Soc. Rev. 41 (2012) 7909– [52] S. Lazar, G.A. Botton, M.Y. Wu, F.D. Tichelaar, H.W. Zandbergen,
7937. Ultramicroscopy 96 (2003) 535–546.
[5] N. Armaroli, V. Balzani, ChemSusChem 4 (2011) 21–36. [53] R. Brydson, H. Sauer, W. Engel, J.M. Thomass, E. Zeitler, N. Kosugi, H. Kuroda,
[6] L. Barelli, G. Bidini, F. Gallorini, S. Servili, Energy 33 (2008) 554–570. Phys. Rev. B: Condens. Matter 1 (1989) 797.
[7] J.P. Van Hook, Cat. Rev. 21 (1980) 1–51. [54] J.G. Chen, Surf. Sci. Rep. 30 (1997) 1–152.
[8] J.R. Bolton, Sol. Energy 57 (1996) 37–50. [55] T. Mizoguchi, I. Tanaka, S. Yoshioka, M. Kunisu, T. Yamamoto, W.Y. Ching, Phys.
[9] L.J. Minggu, W.R. Wan Daud, M.B. Kassim, Int. J. Hydrogen Energy 35 (2010) Rev. B 70 (2004) 045103.
5233–5244. [56] M. Murdoch, G.I.N. Waterhouse, M.A. Nadeem, J.B. Metson, M.A. Keane, R.F.
[10] M.G. Walter, E.L. Warren, J.R. McKone, S.W. Boettcher, Q. Mi, E.A. Santori, N.S. Howe, J. Llorca, H. Idriss, Nat. Chem. 3 (2011) 489–492.
Lewis, Chem. Rev. 110 (2010) 6446–6473. [57] K. Kimura, S.-I. Naya, Y. Jin-nouchi, H. Tada, J. Phys. Chem. C 116 (2012) 7111–
[11] A. Steinfeld, Sol. Energy 78 (2005) 603–615. 7117.
[12] D. Das, T.N. Veziroğlu, Int. J. Hydrogen Energy 26 (2001) 13–28. [58] R. Zanella, S. Giorgio, C.-H. Shin, C.R. Henry, C. Louis, J. Catal. 222 (2004) 357–
[13] A. Kudo, Y. Miseki, Chem. Soc. Rev. 38 (2009) 253–278. 367.
[14] K. Maeda, K. Domen, J. Phys. Chem. Lett. 1 (2010) 2655–2661. [59] E. Kowalska, O.O.P. Mahaney, R. Abe, B. Ohtani, PCCP 12 (2010) 2344–2355.
[15] R. Navarro, M. Pena, J. Fierro, Chem. Rev. 107 (2007) 3952–3991. [60] D.P. Macwan, C. Balasubramanian, P.N. Dave, S. Chaturvedi, J. Saudi Chem. Soc.
[16] W.-T. Chen, A. Chan, D. Sun-Waterhouse, T. Moriga, H. Idriss, G.I.N. 18 (2014) 234–244.
Waterhouse, J. Catal. 326 (2015) 43–53. [61] J.J. Mock, D.R. Smith, S. Schultz, Nano Lett. 3 (2003) 485–491.
[17] M. Bowker, Catal. Lett. 142 (2012) 923–929. [62] K. Kimura, S.-I. Naya, Y. Jin-nouchi, H. Tada, Phys. Chem. C 116 (2012) 7111–
[18] H. Bahruji, M. Bowker, P.R. Davies, F. Pedrono, Appl. Catal. B: Environ. 107 7117.
(2011) 205–209. [63] E. Bae, N. Murakami, T. Ohno, J. Mol. Catal. A: Chem. 300 (2009) 72–79.
[19] M. Bowker, C. Morton, J. Kennedy, H. Bahruji, J. Greves, W. Jones, P.R. Davies, C. [64] H. Chen, X. Kou, Z. Yang, W. Ni, J. Wang, Langmuir 24 (2008) 5233–5237.
Brookes, P.P. Wells, N. Dimitratos, J. Catal. 310 (2014) 10–15. [65] J.F. Moulder, W.F. Stickle, W.F. Sobol, K.D. Bomben, Handbook of X-Ray
[20] V. Jovic, Z.H. Al-Azri, W.-T. Chen, D. Sun-Waterhouse, H. Idriss, G.I.N. Photoelectron Spectroscopy, 1992.
Waterhouse, Top. Catal. 56 (2013) 1139–1151. [66] P. Rodriguez, D. Plana, D.J. Fermin, M.T.M. Koper, J. Catal. 311 (2014) 182–189.
[21] K. Shimura, H. Yoshida, Energy, Environ. Sci. 4 (2011) 2467–2481. [67] T. Li, S. Chattopadhyay, T. Shibata, R.E. Cook, J.T. Miller, N. Suthiwangcharoen,
[22] A. Di Paola, M. Bellardita, L. Palmisano, Catalysts 3 (2013) 36–73. S. Lee, R.E. Winans, B. Lee, J. Mater. Chem. 22 (2012) 14458–14464.
[23] D. Reyes-Coronado, G. Rodriguez-Gattorno, M.E. Espinosa-Pesqueira, C. Cab, [68] K. Yu, T. Yao, Z. Pan, S. Wei, Y. Xie, Dalton Trans. (2009) 10353–10358.
R.d. Coss, G. Oskam, Nanotechnology 19 (2008) 145605. [69] J. Ma, Y. Zou, Z. Jiang, W. Huang, J. Li, G. Wu, Y. Huang, H. Xu, PCCP 15 (2013)
[24] M. Landmann, E. Rauls, W.G. Schmidt, J. Phys.: Condens. Matter 24 (2012) (1908) 11904–11908.
195503. [70] J.B. Priebe, J. Radnik, A.J.J. Lennox, M.-M. Pohl, M. Karnahl, D. Hollmann, K.
[25] S. Sood, P. Gouma, Nanomater. Energy 2 (2013) 82–96. Grabow, U. Bentrup, H. Junge, M. Beller, A. Brückner, ACS Catal. 5 (2015) 2137–
[26] D. Yang, H. Liu, Z. Zheng, Y. Yuan, J.-C. Zhao, E.R. Waclawik, X. Ke, H. Zhu, J. Am. 2148.
Chem. Soc. 131 (2009) 17885–17893. [71] L. Millard, M. Bowker, J. Photochem. Photobiol. A: Chem. 148 (2002) 91–95.
[27] A. Luis, M. Neves, M. Mendonça, O. Monteiro, Mater. Chem. Phys. 125 (2011) [72] V. Balzani, F. Scandola, in: Michael Gratzel (Ed.), Energy Resources through
20–25. Photochemistry and Catalysis, Academic Press Inc., New York, 1983, pp. 2–48.
[28] G.L. Chiarello, A. Di Paola, L. Palmisano, E. Selli, Photochem. Photobiol. Sci. 10 [73] X. Zhang, Y.L. Chen, R.-S. Liu, D.P. Tsai, Rep. Prog. Phys. (2013) 046401. 41.
(2011) 355–360.

Das könnte Ihnen auch gefallen