1. Introduction
Generation of hydrogen by means of photocatalytic,
solar–thermal and photoelectrochemical water splitting
has gained considerable importance in recent years
due to the demand for renewable hydrogen [1–4].
Plants convert solar energy to chemical energy using
2016 The Author(s) Published by the Royal Society. All rights reserved.
Downloaded from http://rsta.royalsocietypublishing.org/ on December 22, 2016
Figure 1. Comparison of (a) one-step and (b) two-step processes of water splitting. Reproduced with permission from [2].
Copyright c 2014, The Royal Society Chemistry. (Online version in colour.)
only 0.03% of the solar energy with a quantum yield of near unity. In natural photosynthesis,
two photosensitizers simultaneously absorb light and the electron transfer chains separate the
charges efficiently [5]. Several efforts have been made to mimic natural photosynthesis in the
laboratory artificially, to produce hydrogen and oxygen by the splitting of water involving two
redox reactions [6,7]:
and
Photochemical water splitting comprises three steps [1]. They are absorption of light, generation
and separation of charges and redox reactions at the surface as given by the above equations.
The semiconductors used for the purpose should possess a negative conduction band (CB)
minimum relative to water reduction and a positive valance band (VB) maximum relative to the
water-oxidation potential (figure 1a). Furthermore, one should have reasonable surface catalytic
reaction rates. Reduction of water to hydrogen involves two electrons while the oxidation of
water to oxygen involves four electrons. Water oxidation is a crucial step in water splitting
and determines the overall efficiency. Sacrificial electron acceptors (AgNO3 , Na2 S2 O8 ) or donors
(Na2 S–Na2 SO3 ) are used to remove either electrons or holes, and the counter charge retained in
the photocatalyst reacts with water to yield either oxygen or hydrogen. Artificial photosynthesis
can be a one-step or two-step process depending on the involved mechanism as shown in figure 1
[2]. In the one-step process, the photocatalyst absorbs the light and generates charges, and the
electrons reduce water while the holes oxidize water. In the two-step process, two sensitizers
absorb the light generating electron–hole pairs and the process is akin to the Z-scheme in natural
photosynthesis [5].
Another strategy to convert solar energy to chemical energy is by thermochemical splitting of
H2 O into H2 , a strategy that can also be used to reduce CO2 to CO. Thermochemical processes
performed using solar concentrators can achieve high solar-to-fuel conversion efficiencies as they
allow the utilization of the entire solar spectrum besides the thermal energy. Direct thermolysis
of H2 O is not favoured due to the large energy penalty (T > 2500 K) and the unavoidable mixing
of explosive H2 and O2 gases at high temperatures. With this purpose, catalytic cycles which split
H2 O at moderate temperatures in two or multiple steps are being explored [8–11]. In this context,
redox-active metal oxides (MOn ) which split H2 O (or CO2 ) in two discrete steps are of interest.
The basic reactions are
δ
Endothermal step: MOn → MOn−δ + O2 (1.3)
2
Downloaded from http://rsta.royalsocietypublishing.org/ on December 22, 2016
and
3
Exothermal step: MOn−δ + δH2 O → MOn + δH2
(1.4)
(a) (b)
4
Chl* Ru(II)*
photons CO2 (CHO)n photons S2O82– SO42– + SO4–
Figure 2. Mechanism of oxidation of water in (a) natural photosynthesis and (b) artificial photosynthesis. Typical catalysts in
(b) are LaCoO3 and Mn2 O3 . Adapted with permission from [6]. Copyright
c 2014, Indian Academy of Sciences, India. (Online
version in colour.)
(a) (b)
Li2Co2O4 100
(mmol.mole–1 min–1)
(i)
transition metal . m2)
45
500
(mmol/mole of
Li2Co2O4 (ii)
800
Li1.1Co2O4
50 30 (i)
400 (iii)
15
O2 evolved
600 (iv)
0
300 0 5 10 15 0 2 4 6 8 10 12 14 16
(ii)
increasing Li
400
200
(iii)
200
100
(iv)
0 0
0 5 10 15 0 5 10 15
time (min) time (min)
Figure 3. (a) Amount of O2 evolved per mole of Co by Lix Co2 O4 with varying Li content. (Circles) Li2 Co2 O4 and (squares) Li1.1 Co2 O4 .
(b) Amount of O2 evolved per mole of transition metal by (i) LaCoO3 , (ii) Li2 Co2 O4 , (iii) Mn2 O3 and (iv) LaMnO3 . Reproduced with
permission from [17].
seems to be essential in determining the catalytic activity for the oxidation of water. In view of the
importance of the e1g configuration, deposition of such catalysts on semiconducting materials such
as BiVO4 , α-Fe3 O4 , WO3 , etc., may provide a useful combination to yield high oxygen evolution
rates. It is found that the presence of Pt enhances the catalytic activity of LaCoO3 and other
catalysts [21].
dxy, xz, yz H2 0
0 2 4 6 8 10
EY 1T-MoS2
light time (h)
Figure 4. (a) The crystal-field-splitting induced electronic configuration of 2H-MoS2 and 1T-MoS2 and proposed mechanism for
catalytic activity of 1T-MoS2 . Orange, S; blue-grey, Mo. (b) Time course of H2 evolved by freshly prepared 1T-MoS2 . Reproduced
with permission from [25]. Copyright c 2013, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim. (Online version in colour.)
the electronic structure and decreases the absorption edge. However, it leads to charge imbalance
due to replacing O2− with N3− and defective. Co-substitution of N and F in place of O is
a promising method to alter the electronic structure while retaining the charge balance [23].
Recently, co-substitution of N and F in TiO2 and ZnO has been carried out in our laboratory
which leads to substantial decrease in band gaps of TiO2 and ZnO with a change in colour from
white to yellow and orange, respectively [23,24]. TiO2−x (N,F)x exhibited a hydrogen evolution
rate of 60 µmol h−1 g−1 without the presence of any noble-metal cocatalyst under visible light
irradiation [24]. Hence co-substitution of N and F in several wide band gap oxides such as SrTiO3 ,
NaTaO3 , BaTaO3 , etc., is a potential method to make and use them as visible light-sensitive
photocatalysts.
(a) Dyes
Dyes have been successfully used as visible light-responsive components in photocatalysis.
Combinations of dyes with wide variety of materials, ranging from large band gap to small band
gap and metallic systems have been studied (table 1) [25,30,31]. For example, in dye-assisted
TiO2 , the dye absorbs light and injects the photo-excited electron to the CB of the TiO2 and these
electrons participate in the reduction of H+ to H2 [1]. Use of small band gap materials such
as MoS2 with Eosin Y dye has been studied recently [25,30]. Hydrogen evolution activities of
0.05 and 2–3 mmol h−1 g−1 were obtained with 2H-MoS2 and graphene–MoS2 , respectively [25].
Graphene acts as an electron channel in the latter case. Use of N-doped graphene with MoS2
results in the formation of p–n junction which results in the induced electric field at the interface
of n-type N-doped graphene and p-type MoS2 . As a result, enhanced hydrogen evolution activity
of 10.8 mmol h−1 g−1 is obtained (table 1) [25]. 1T-MoS2 exhibits an activity of 30 mmol h−1 g−1 ,
superior to those obtained with 2H-MoS2 . In figure 4a is presented the mechanism of dye-
assisted photocatalytic hydrogen evolution. Upon visible light irradiation EY forms EY1∗ and
then transforms to triplet EY3∗ . It receives electrons from sacrificial electron donor and forms the
EY− anion. The electron from the anion is transferred to MoS2 (1T, metallic form) either directly
or through graphene channel. Recently, 1T and 2H forms of MoSe2 have also been used along
with Eosin Y for visible light-induced hydrogen evolution [26]. 1T-MoSe2 exhibits an activity
of 60 mmol h−1 g−1 , superior to those obtained with 2H-MoSe2 (0.05 mmol h−1 g−1 ), 1T-MoS2
(30 mmol h−1 g−1 ) and 2H-MoS2 (0.05 mmol h−1 g−1 ) with a turnover factor of 19.2 (figure 5 and
table 1). In table 1, these results are compared with those obtained with similar other systems
[27–29].
Downloaded from http://rsta.royalsocietypublishing.org/ on December 22, 2016
600
6
500
300
h –1
g –1
200
ol
mm
100
75
1T-MoSe2
0
0 4 8 12
time (h)
Figure 5. Time course of hydrogen evolution by 1T-MoSe2 and using Eosin Y as a sensitizer. Reproduced with permission from
[26]. Copyright
c 2014, AIP Publishing. (Online version in colour.)
Table 1. Visible light-induced H2 generation by dye sensitization. Values of turnover factor are given in parentheses. RGO,
reduced graphene oxide; NRGO, nitrogen-doped reduced graphene oxide.
(b) Semiconductors
Very few of the narrow band gap semiconductors possess desirable CB and VB positions. Among
them CdS, CdSe (quantum dots), TaON, Y2 Ta2 O5 N2 , etc., have been well explored [1]. Recently,
g-C3 N4 has gained attention owing to its desirable VB and CB positions in addition to its visible
light absorption and surface catalytic sites [32]. Modification of the electronic structure of these
materials, for example P and Cl co-substituted CdS, has been shown to enhance the activities
approximately three times probably due to the capability for using longer wavelengths [33]. These
materials themselves exhibit very weak H2 evolution activities because of the recombination
of either charge carriers or the reaction intermediates. Often redox reactions of water splitting
occur on the time scales of microseconds or longer, whereas the charge carrier recombination
and relaxation occur within a few nanoseconds [34]. In order to achieve maximum yields, there
is a need to prolong the lifetime of charge carriers and suppress the recombination of reaction
intermediates.
Kato et al. [35] have demonstrated efficient water splitting to hydrogen and oxygen using NiO–
NaTaO3 : La by physical separation of water reduction and oxidation sites. This has motivated
several researchers to take steps to develop such kinds of structures and evaluate their properties
[36–38]. A very important contribution has been made by Alivisatos and co-workers who have
developed multicomponent nanoheterostructures with tunable properties [36]. These structures
Downloaded from http://rsta.royalsocietypublishing.org/ on December 22, 2016
(a) (b)
QE = 20% 7
H2 D = 2.3 nm
Pt
CdSe
CdS h+
60 nm 20 nm 27 nm 40 nm 70 nm 60 nm
samples of Pt-tipped seeded and unseeded rods
Figure 6. (a) Schematic of process and energy level diagram of hydrogen generation by CdSe–CdS–Pt heterostructrures. (b)
Photocatalytic hydrogen evolution as a function of length of the nanorod and diameter of the seed. Reproduced with permission
from [36]. Copyright
c 2010, American Chemical Society. (Online version in colour.)
ZnO–Pt–CdS Pt 32 36 [37]
..........................................................................................................................................................................................................
CdS/(Pt–TiO2 ) Pt 9 5 [39]
..........................................................................................................................................................................................................
b
CdS/Pt/TiO2 Pt 15 14 [40]
..........................................................................................................................................................................................................
a Visible light irradiation.
b Light sources used for activity and AQY calculations are different; AQY, apparent quantum yield.
contain Pt-tipped CdS nanorod with embedded CdSe seed. Upon photoexcitation of CdS, the
photo-generated electrons are injected into Pt and holes are confined to CdSe seeds (figure 6).
These authors could tune the properties by tuning the distance between the reactions sites and
extent of charge separation through band alignment of CdS and CdSe. From this study, it was
concluded that increasing nanorod length (i.e. the distance between the reaction sites, Pt and
CdSe) results in increased hydrogen evolution activities. Similarly, a smaller seed (2.3 versus
3.1 nm) results in higher yields (figure 6). CdSe/CdS/Pt heterostructures with seed diameter of
2.3 nm and nanorod length of 60 nm exhibited a hydrogen evolution of 40 mmol h−1 g−1 with an
apparent quantum yield (AQY) of 20% at 420 nm in the presence 2-propanol (table 2). And their
study also revealed that hydrogen bound to Pt is stable and the reaction is of first order with
respect to the photon flux.
Similar strategies have been used recently, by forming heterostructures of several other
semiconductors. Heterostructures of ZnO/CdS and TiO2 /CdS have been well explored [39–42].
Band positions in these structures form a type II alignment which is suitable for charge separation.
Therefore, superior hydrogen evolution activities are obtained with ZnO/CdS and TiO2 /CdS
relative to their individual components. Deposition of co-catalysts such as RuO2 and Pt provides
active sites for surface redox reactions and improves the activity. Wang et al. [41] have observed
that the charge carriers in ZnO/CdS exhibit lifetimes up to 220 ns, whereas in ZnO and CdS have
only 60 ns decay lifetime.
Downloaded from http://rsta.royalsocietypublishing.org/ on December 22, 2016
0 8
–3
e– e– e– –1
H2
.........................................................
–4 e– e– e– H+/H2
Pt 0
E versus NHE
–5 CdS ZnO
H+ 1
–6 h+ h+ O2/H2O
2
–7 CH2OH CHO
3
–8
Figure 7. Schematic illustration of process and energy level diagram of photocatalytic hydrogen production in the presence of
benzyl alcohol. Adapted with permission from [37]. Copyright c 2013, The Royal Society of Chemistry. (Online version in colour.)
(a) (b)
visible ZnO/Pt/Cd1–xZnxS 40 ZnO/Pt/Cd0.8Zn0.2S
35 UV–visible activity (mmol h–1 g–1) 35 ZnO/Pt/CdS
activity (mmol h–1 g–1)
30 30 l > 395 nm
25 25
20 20
15 15
10 10
5 5
0 0
0 0.2 Na2S,Na2SO3 PhCH2OH
x
Figure 8. (a) Photocatalytic H2 evolution as a function of Zn substitution in ZnO/Pt/Cd1−x Znx S and (b) comparison of effect
of sacrificial agent on H2 evolution activity. Reproduced with permission from [37]. Copyright c 2013, The Royal Society of
Chemistry. (Online version in colour.)
(a) (b) 9
20
CdS
0
0 60 120 180 240 300
5 nm
time (min)
Figure 9. (a) High-resolution transmission electron microscopy image of a-TiO2 /Pt/CdS and (b) time course of H2 evolution of
a-TiO2 /Pt/CdS as a function of time in the presence of benzyl alcohol–acetic acid. (Online version in colour.)
36.5 mmol h−1 g−1 , respectively, under visible light irradiation (figure 8b). In table 2, we compare
the yields obtained by these systems with other similar systems.
We have carried out a detailed investigation of the visible light-induced hydrogen evolution
by using TiO2 /Pt/Cd1−x Znx S (x = 0.0, 0.2) heterostructures, wherein poorly crystalline or nearly
amorphous TiO2 (anatase) as well as crystalline anatase (denoted as a-TiO2 and c-TiO2 ,
respectively) have been used. a-TiO2 was prepared by the hydrolysis of aqueous TiCl4 with
dilute KOH solution and c-TiO2 was prepared by thermal treatment of a-TiO2 at 300◦ C. With
the increasing proportion of CdS, the absorption in the visible region increases as expected.
We first studied visible light-induced hydrogen evolution in the presence of Na2 S–Na2 SO3 as
sacrificial agents and found maximum activity at a loading of 20 mol% of CdS relative to TiO2
in a-TiO2 /CdS. The fraction of CdS that effectively forms the interface with the TiO2 decreases
beyond 20 mol%. This reduces the transfer of excited electrons to the TiO2 and also prevents
the reactant molecules from reaching the TiO2 surfaces. We have, therefore, maintained 20 mol%
of CdS relative to TiO2 as the optimum loading in subsequent studies with the a-TiO2 /Pt/CdS
heterostructures. We must note that the role of the Pt co-catalyst in hydrogen evolution has been
well explored. A transmission electron microscopy image of a-TiO2 /Pt/CdS is shown in figure 9a.
This image depicts uniform distribution of Pt and CdS nanoparticles on the a-TiO2 . The lattice
fringes d spacing of 0.22 nm is assigned to the (111) plane of Pt. Similarly, the d spacing of 0.35 nm
is assigned to the (111) plane of the cubic CdS. Upon photoirradiation, the charges generated in
CdS are transferred to the interface of TiO2 and participate in the water reduction. The activity
in this kind of heterostructure, therefore, depends on the surface/interface properties rather
than the bulk properties of the TiO2 . As hydrogen evolution activity is considerably enhanced
in the presence of benzyl alcohol–acetic acid as a sacrificial agent [37,43], we have used this
mixture as the hole scavenger and carried out hydrogen evolution reaction (HER) studies. We
indeed found excellent improvement in hydrogen evolution, yielding 31 mmol h−1 g−1 with AQY
of 29% under visible light irradiation (figure 9b and table 2). As modification of the electronic
structure of CdS by partial substitution of Zn as in Cd1−x Znx S is reported to improve the visible
light-induced HER activity in these heterostructures [37], we have prepared heterostructures by
partially substituting the Zn in CdS. These heterostructures (a-TiO2/Pt/Cd0.8 Zn0.2 S) exhibited
a maximum H2 evolution activity of 40 mmol h−1 g−1 with an AQY of 48% under visible light
irradiation. Hence we assume the use of amorphous TiO2 in such kinds of heterostructures has a
potential to reduce the processing cost and technologies.
The photocatalytic properties of ZnO/Pt/CdS heterostructures depend on the electronic,
optical and physical properties of ZnO. Conducting ZnO (F-doped ZnO) nanoparticles were
therefore obtained by the treatment of ZnO with NH4 F (1 : 2 mole ratio) at 600◦ C in N2
atmosphere [23]. We have used F-doped ZnO for the purpose of forming ZnO/Pt/Cd1−x Znx S
type heterostructures, and used them in H2 evolution studies. Our preliminary results suggest
Downloaded from http://rsta.royalsocietypublishing.org/ on December 22, 2016
(a) (b)
10
Na2S and Na2SO3
1.0 ZnO 35
Figure 10. (a) Comparison of UV–visible absorption spectrum of ZnO1−x (N,F)x with that of undoped ZnO (inset shows the
change in colour from white to orange). (b) Comparison of effect of co-doping of N and F in ZnO on hydrogen evolution of
ZnO/Pt/CdS (25%) type hybrid nanostructures using Na2 S–Na2 SO3 or benzyl alcohol–acetic acid mixture as sacrificial agent
under visible light irradiation. Reproduced with permission from [43]. Copyright
c 2014, The Royal Society of Chemistry. (Online
version in colour.)
that the H2 evolution activity indeed increases with F doping in ZnO under visible light
irradiation. The effect of conductivity of ZnO is most remarkable in the absence of Pt. Detailed
investigations will be published elsewhere.
Narrow band gap of CdS (2.4 eV) relative to ZnO (3.2 eV) causes only CdS to get excited
upon visible light irradiation. As such ZnO does not participate in the light absorption. Recently,
we have developed a new approach to modify the electronic structure of wide band gap
oxides by co-substitution of N and F in place of O. For example, N, F co-substitution in ZnO
decreases substantially its band gap to 2.4 eV (figure 10a) [43]. Therefore, the structures containing
ZnO0.8 (NF)0.2 and CdS provide two photosensitive centres. These structures exhibited superior
hydrogen evolution activities in the presence of both Na2 S–Na2 SO3 and benzyl alcohol–acetic
acid (figure 10b). Partial substitution of Zn in CdS in these heterostructures also enhances the
activity, as in the case of ZnO/Pt/Cd1−x Znx S [37].
We have investigated the effect of the morphology and surface area of oxide nanostructures
in ZnO/Pt/CdS type heterostructures for visible light-induced hydrogen generation [44].
TiO2 /Pt/Cd0.8 Zn0.2 S heterostructures containing TiO2 powder, TiO2 nanoparticles (NPs),
H2 Ti3 O7 nanotubes (NTs) and TiO2 NTs with surface areas of 40, 123, 248 and 249 m2 g−1 ,
respectively, exhibited hydrogen evolution activity of 0.55, 1.01, 1.76 and 1.20 mmol h−1 g−1 ,
respectively (figure 11a). Structures containing H2 Ti3 O7 NTs having highest surface area show
highest H2 evolution activity, whereas those containing TiO2 powder having least surface area
show least H2 evolution activity. We have also observed that ZnO/Pt/CdS heterostructures
containing (i) ZnO NPs, (ii) ZnO nanorod 1 (NR1), (iii) ZnO nanorod 2 (NR2) and (iv) ZnO
nanorod 3 (NR3) with surface areas of 45, 41, 18 and 9 m2 g−1 , respectively, exhibited hydrogen
evolution activities of 5.36, 6.88, 5.29 and 2.55 mmol h−1 g−1 , respectively (figure 11b). Hence H2
evolution from these heterostructures follows the trend in the BET surface area of the oxide
nanostructures where higher surface area shows higher activity. A slight variation from the trend
is observed in the case of ZnO NR1 which showed higher photocatalytic activity than ZnO NP. It
is probably due to better separation of photo-generated charge carriers in the case of ZnO NR1.
Replacing expensive Pt with NiO in these heterostructures is a promising approach in order
to obtain cost-effective renewable energy production. NiO has several advantages over Pt and
other co-catalysts as it does not allow back reaction or recombination of H2 and O2 [45]. In
our further studies, we have replaced Pt with NiO in ZnO/Pt/Cd1−x Znx S heterostructures
(figure 12) [46]. Although ZnO/NiO/CdS shows higher activities compared with ZnO/CdS,
the most remarkable improved activity is observed with partial substitution of Zn in CdS as
in ZnO/NiO/Cd0.8 Zn0.2 S. The activities obtained are comparable with those obtained with
Downloaded from http://rsta.royalsocietypublishing.org/ on December 22, 2016
(a) (b)
2.0 1. Degussa P25 4 10 1. ZnO NR3 11
0 0
0 60 120 180 240 0 10 20 40 50
surface area (m2 g–1) surface area (m2 g–1)
Figure 11. Phtocatalytic H2 evolution of (a) TiO2 /Pt/CdS and (b) ZnO/Pt/CdS heterostructures with oxides having different
surface areas. Reproduced with permission from [44]. (Online version in colour.)
8
10
NiO
6 8 Pt
6
4
Au
4
2
2
0 n 0.2S 0
/CdS /Cd 0.8Z
/CdS /NiO /NiO
ZnO ZnO ZnO
Figure 12. Comparison of photocatalytic hydrogen evolution with (a) different photocatalysts and (b) ZnO/X/Cd0.8 Zn0.2 S
(X = Pt, NiO and Au). Reproduced with permission from [46]. Copyright 2014, Elsevier BV. (Online version in colour.)
ZnO/Pt/Cd0.8 Zn0.2 S and superior to those obtained with ZnO/Au/Cd0.8 Zn0.2 S (figure 12).
Similarly, we have prepared heterostructures of TiO2 /NiO/Cd0.8 Zn0.2 S by the deposition of
Cd0.8 Zn0.2 S on TiO2 /NiO. TiO2 /NiO was prepared by hydrolysis of aqueous solution of nickel
acetate in a solution containing TiO2 dispersion and subsequent thermal treatment at 400◦ C for
2 h in air. We have used these structures for photocatalytic hydrogen evolution studies. Our
preliminary results show that use of NiO in TiO2 /NiO/Cd0.8 Zn0.2 S results in H2 yields superior
to those obtained with TiO2 /Pt/Cd0.8 Zn0.2 S with a wide range of molar ratios of TiO2 : NiO.
Further studies on determining the role of NiO in these heterostructures in separating charges and
catalysing proton reduction are in progress. The detailed results will be published soon elsewhere.
There is considerable scope for developing heterostructures which are efficient, stable
and inexpensive for the reduction of water. More importantly, the structures should exhibit
H2 evolution activities at longer wavelengths (λ > 600 nm). Detailed discussion on use of
heterojunctions or heterostructures for producing hydrogen is available in the literature [9,11,33].
at CeO2 results in δ ∼ 0.066 (TRED = 1773 K, pO2 = 10−5 atm), producing nearly stoichiometric
12
quantity of H2 [49]. CeO2 can be modified by several metal ions and metal oxides as in the case
of ferrites. Oxidation rates improve due to the addition of +3 dopants, especially La3+ adds
(a) Perovskites
The large degree of reduction capability at elevated temperatures and reversible reoxidation are
among the basic criteria needed for a material to be used for the solar–thermal reduction of
H2 O and CO2 . Large numbers of perovskite oxides of ABO3−δ class are known to create oxygen
non-stoichiometry. Nalbandian et al. [57] have investigated La1−x Srx MO3 (M = Mn, Fe; x = 0–1)
perovskites for thermochemical syngas production using CH4 and H2 O as reactants at 1273 K.
Notably, the addition of 5% NiO on La1−x Srx FeO3 is reported to produce maximum H2 [58].
Beside the fact of lowering the reaction temperature, use of CH4 and H2 O as reactants suffers from
significant coke formation. The La1−x Srx MnO3 (LSM) family of perovskites has been successfully
applied for reduction of CO2 and H2 O by using the two-step process.
In LSM, replacement of trivalent La by divalent Sr creates holes located at the Mn sites,
creating Mn3+ /Mn4+ pairs. A detailed thermodynamic investigation by Scheffe et al. [56] reveals
the greater oxygen non-stoichiometry of LSM perovskites compared with ceria. Experimental
investigations show LSM35 (La0.65 Sr0.35 MnO3 ) reduces to a greater extent than CeO2 (TRED =
1773 K, figure 13b). Despite having less favourable oxidation thermodynamics, LSM35 produces
more CO than CeO2 (figure 13b). Thermochemical H2 O splitting was studied using LSM (x = 0–
0.5) by Yang et al. [59] showing higher fuel production capability than CeO2 . For higher Sr content
stoichiometric reoxidation needs high concentration of H2 O as dictated by thermodynamic and
kinetic assessments. A similar observation has been reported by Demont et al. [60] for CO2
splitting. La1−x Cax MnO3 (LCM; x = 0.35, 0.5, 0.65) class of perovskites has been recently explored
for the thermochemical CO2 and H2 O splitting from our laboratory and shows even more
promising results [61]. The extent of reduction increases with the Ca content with a gradual
decrease in the reduction temperature (figure 14a). The amount of O2 evolved from LCM35,
LCM50 and LCM65 is 109, 315 and 653 µmol g−1 , respectively, at TRED of 1673 K (figure 14a).
Under similar experimental conditions, LCM50 produces 1.6 times and 5 times more O2 than
La0.5 Sr0.5 MnO3 (201 µmol g−1 ) and CeO2 (63 µmol g−1 ), respectively (figure 14b). LCM reoxidizes
stoichiometrically upto x = 0.5 by 40% CO2 compositions at TGS = 1373 K (figure 14a). Importantly,
the CO productivity of LCM50 (525 µmol g−1 ) is also about 1.6 times and 5 times larger than
La0.5 Sr0.5 MnO3 (325 µmol g−1 ) and CeO2 (112 µmol g−1 ), respectively (figure 14b). Constant fuel
production activity of LCM50 during multiple cycles makes it even more promising. LCM50
shows superior performance during the thermochemical H2 O splitting keeping the reaction
kinetics similar to that of La0.5 Sr0.5 MnO3 . Structural investigations reveal that the smaller size
of Ca causes LCM50 to be orthorhombic whereas LSM50 is rhombohedral. The superior activity
of LCM50 over LSM50 is related to their structural differences.
Electronic and magnetic properties of the rare earth manganites are known to be strongly
dependent on the size of rare earth ions [62]. In this regard, two series of perovskites,
Ln0.5 Sr0.5 MnO3 and Ln0.5 Ca0.5 MnO3 (Ln = La, Nd, Sm, Gd, Dy, Y), were recently used to
thermally split CO2 (H2 O) by our group [63]. Oxygen non-stoichiometry of rare earth manganites
increases with the decreasing size of the rare earth ion from La to Y (figure 15a, b). The yttrium
derivatives evolve highest amount of O2 . Decrease in the radius of rare earth ions decreases
the tolerance factor (τ ). Decrease in τ indicates the increase in lattice distortion and reduction
Downloaded from http://rsta.royalsocietypublishing.org/ on December 22, 2016
(a) (b)
100.2 13
2.00
temperature (K)
1773
1473
1173
mass %
CeO2 99.8
2–d
1.90
Zr05
Zr10 99.6
Zr15
1.85 Zr20
99.4
1.80 LSM35
–25 –20 –15 –10 –5 0
0 100 200 300 400 500
log10(pO2/atm)
time (min)
Figure 13. (a) Oxygen non-stoichiometry of Ce1−x Zrx O2−δ (x = 0–0.2) at 800◦ C. Symbols represent thermogravimetric
analysis results. Solid lines are model fits as presented in [55]. Vertical lines present the pO2 during two-step cycles. Reproduced
with permission from [55]. Copyright c 2014, American Chemical Society. (b) Thermochemical CO2 splitting profile of
La0.65 Sr0.35 MnO3 (LSM35) in comparison with ceria as indicated by dashed and black lines, respectively. Reduction (pO2 =
1.5 × 10−3 atm) and oxidation (pCO2 = 0.40 atm) temperatures are 1773 K and 1073–1273 K, respectively. Total gas flow rate is
300 sccm. Reproduced with permission from [56]. Copyright c 2013, American Chemical Society. (Online version in colour.)
(a) (b)
100.5 1400 100.2
1400
temperature (°C)
1. La0.65Ca0.35MnO3
1200 1. CeO2
temperature (°C)
2. La0.50Ca0.50MnO3 100.0 1200
100.0 2. La0.5Sr0.5MnO3
3. La0.35Ca0.65MnO3 1000
3. La0.5Ca0.5MnO3 1000
800 99.8
99.5 1 1 800
weight (%)
weight (%)
600
99.6 600
99.0
increasing Ca
2 99.4 2
98.5
99.2
3 3
98.0 99.0
Figure 14. Thermogravimetric analysis plots of thermochemical CO2 splitting of (a) La1−x Cax MnO3 composites (x =
0.35, 0.5, 0.65) and (b) La0.5 Ca0.5 MnO3 in comparison with La0.5 Sr0.5 MnO3 and CeO2 . Reproduced with permission from [61].
Copyright c 2015, Royal Society. (Online version in colour.)
of spatial overlap of Mn eg and O 2pσ orbitals, which in turn favours the O2 removal (figure 15c)
[64]. The amount of O2 evolved by Y0.5 Sr0.5 MnO3 (τ = 0.965) is higher (481 µmol g−1 ) compared
with La0.5 Sr0.5 MnO3 (τ = 0.996). Y0.5 Ca0.5 MnO3 (τ = 0.948) exhibits further enhancement in
the O2 evolution (575 µmol m g−1 ). Notably, the O2 release temperature gradually decreases
on decreasing the rare earth ion size in the Ln0.5 Sr0.5 MnO3 series (figure 15d). Disorder due
to the size mismatch between the Ln and A cations (designated as size variance factor, σ 2 )
affects the reduction temperature of Ln0.5 A0.5 MnO3 perovskites [63]. Y0.5 Sr0.5 MnO3 with the
highest σ 2 (15.6 × 10−3 ) starts to lose O2 around 860◦ C. The yttrium derivatives evolve the
highest amount of CO (figure 15a,b). Y0.5 Sr0.5 MnO3 is one of the best perovskite oxides with a
production of 624 µmol g−1 of CO for the TRED /TGS pair of 1300◦ C/900◦ C (table 3). Production
of 418 µmol g−1 CO even in the TRED /TGS pair of 1200◦ C/900◦ C indicates Y0.5 Sr0.5 MnO3 as one
of the beneficial material for energy application (table 3). Y0.5 Sr0.5 MnO3 and Y0.5 Ca0.5 MnO3 are
further used to produce H2 by splitting H2 O at TRED /TGS pair of 1400◦ C/1100◦ C.
Downloaded from http://rsta.royalsocietypublishing.org/ on December 22, 2016
temperature (°C)
temperature (°C)
1400 1400
1200 1200
99.9 100.0
1000
weight (%)
600
1
99.3 1
2 99.0 2
3
99.0 3
4 4
98.5
98.7 5 5
98.0 6
98.4 6 50 75 100 125 150 175 200
50 75 100 125 150 175 200 time (min)
time (min)
(c) 500
(d )
2.92 100.0
2.90
O2 produced (mmol g–1)
400 1
2.88 99.5
weight (%)
2
3–d
2.86 3
4
300
2.84 99.0 5
2.82 6
200
2.80 98.5
0.966 0.973 0.980 0.987 0.994 700 800 900 1000 1100 1200 1300
tolerance factor (t)
temperature (°C)
Figure 15. Thermogravimetric analysis curves to represent the thermochemical splitting of CO2 by (a) Ln0.5 Sr0.5 MnO3 and (b)
Ln0.5 Ca0.5 MnO3 . (c) Variation of oxygen non-stoichiometry and total O2 produced with tolerance factor and (d) weight loss profile
with temperature for Ln0.5 Sr0.5 MnO3 ; Ln = La (1), Nd (2), Sm (3), Gd (4), Dy (5) and Y (6). Reduction and oxidation temperatures
are 1400◦ C and 1100◦ C, respectively. Reproduced with permission from [63]. Copyright c 2015, Wiley-VCH Verlag GmbH&Co.
KGaA, Weinheim.
Table 3. O2 production and CO (H2 ) evolution results of perovskite oxides using thermochemical two-step process. Adapted
with permission from [63].
material TRED (◦ C) O2 evolved ( µmol g−1 ) Tox (◦ C) CO(H2 ) evolved ( µmol g−1 ) references
Y0.5 Sr0.5 MnO3 1400 483 900 757a [63]
..........................................................................................................................................................................................................
400
O2 produced by 15
350 La0.5Sr0.5Mn1–xAlxO3
300
150
100
50
Figure 16. Histograms showing the O2 production profile of Ln0.5 Sr0.5 Mn1−x Alx O3 perovskites (x = 0, 0.25, 0.5). Reduction is
performed at temperature of 1673 K (heating rate 20 K min−1 ) in Ar. (Online version in colour.)
In Ln1−x Ax MnO3 type manganites, B-site substitution is known to induce disorder in the
MnO6 octahedral network. In several half-doped manganites, even a small amount of Mn
substitution results in drastic changes of magnetic as well as of transport properties. LaAlO3−δ
perovskites substituted with Mn on B sites and Sr on A sites (termed SLMA) have been
investigated for two-step process by McDaniel et al. [65]. SLMA perovskites produce multiple
times more fuel than CeO2 keeping the reaction kinetics unaltered and are reported to minimize
the carbonate formation substantially [66]. Our preliminary results indicate gradual doping of
Al enhances the O2 evolution of La0.5 Sr0.5 MnO3 significantly (figure 16). The doping of Al is
performed by the Pecini route. It gives high surface area material in contrast with solid-state
synthesis route. Further studies on determining the role of Al3+ and related cations in LSM are
in progress.
Authors’ contributions. C.N.R.R. suggested the problems and gave guidance, S.R.L., S.D. and A.R. carried out
16
experiments and the manuscript was written jointly.
Competing interests. We have no competing interests.
References
1. Kudo A, Miseki Y. 2009 Heterogeneous photocatalyst materials for water splitting. Chem. Soc.
Rev. 38, 253–278. (doi:10.1039/b800489g)
2. Hisatomi T, Kubota J, Domen K. 2014 Recent advances in semiconductors for photocatalytic
and photoelectrochemical water splitting. Chem. Soc. Rev. 43, 7520–7535. (doi:10.1039/
c3cs60378d)
3. Kodama T, Gokon N. 2007 Thermochemical cycles for high-temperature solar hydrogen
production. Chem. Rev. 107, 4048–4077. (doi:10.1021/cr050188a)
4. Licht S. 2005 Thermochemical solar hydrogen generation. Chem. Commun. 4635–4646.
(doi:10.1039/b508466k)
5. Tachibana Y, Vayssieres L, Durrant JR. 2012 Artificial photosynthesis for solar water-splitting.
Nat. Photon. 6, 511–518. (doi:10.1038/nphoton.2012.175)
6. Maitra U, Lingampalli SR, Rao CNR. 2014 Artificial photosynthesis and the splitting of water
to generate hydrogen. Curr. Sci. 106, 518–527.
7. Rao CNR, Lingampalli SR. In press. Generation of hydrogen by visible-light induced water-
splitting with the use of semiconductors and dyes. Small. (doi:10.1002/smll.201500420)
8. Roeb M, Neises M, Monnerie N, Call F, Simon H, Sattler C, Schmucker M, Pitz-Paal R. 2012
Materials-related aspects of thermochemical water and carbon dioxide splitting: a review.
Materials 5, 2015–2054. (doi:10.3390/ma5112015)
9. Abanades S, Charvin P, Flamant G, Neveu P. 2006 Screening of water-splitting
thermochemical cycles potentially attractive for hydrogen production by concentrated solar
energy. Energy 31, 2805–2822. (doi:10.1016/j.energy.2005.11.002)
10. Scheffe JR, Steinfeld A. 2014 Oxygen exchange materials for solar thermochemical splitting of
H2 O and CO2 : a review. Mater. Today 17, 341–348. (doi:10.1016/j.mattod.2014.04.025)
11. Miller JE, McDaniel AH, Allendorf MD. 2014 Considerations in the design of materials for
solar-driven fuel production using metal-oxide thermochemical cycles. Adv. Energy Mater.
4, 1300469. (doi:10.1002/aenm.201300469)
12. Chueh WC, Falter C, Abbott M, Scipio D, Furler P, Haile SM, Steinfeld A. 2010 High-flux solar-
driven thermochemical dissociation of CO2 and H2 O using nonstoichiometric ceria. Science
330, 1797–1801. (doi:10.1126/science.1197834)
13. Moniz SJA, Shevlin SA, Martin DJ, Guo Z-X, Tang J. 2015 Visible-light driven heterojunction
photocatalysts for water splitting—a critical review. Energy Environ. Sci. 8, 731–759.
(doi:10.1039/c4ee03271c).
14. Wang D, Li R, Zhu J, Shi J, Han J, Zong X, Li C. 2012 Photocatalytic water oxidation on BiVO4
with the electrocatalyst as an oxidation cocatalyst: essential relations between electrocatalyst
and photocatalyst. J. Phys. Chem. C 116, 5082–5089. (doi:10.1021/jp210584b).
15. Gardner GP, Go YB, Robinson DM, Smith PF, Hadermann J, Abakumov A, Greenblatt M,
Dismukes GC. 2012 Structural requirements in lithium cobalt oxides for the catalytic oxidation
of water. Angew. Chem. Int. Ed. 51, 1616–1619. (doi:10.1002/anie.201107625)
16. Robinson DM, Go YB, Greenblatt M, Dismukes GC. 2010 Water oxidation by λ-MnO2 :
catalysis by the cubical Mn4 O4 subcluster obtained by delithiation of spinel LiMn2 O4 . J. Am.
Chem. Soc. 132, 11 467–11 469. (doi:10.1021/ja1055615)
17. Maitra U, Naidu BS, Govindaraj A, Rao CNR. 2013 Importance of trivalency and the e1g
configuration in the photocatalytic oxidation of water by Mn and Co oxides. Proc. Natl Acad.
Sci. USA 110, 11 704–11 707. (doi:10.1073/pnas.1310703110)
18. Zhu J, Yin Z, Yang D, Sun T, Yu H, Hoster HE, Hng HH, Zhang H, Yan Q. 2013
Hierarchical hollow spheres composed of ultrathin Fe2 O3 nanosheets for lithium storage and
photocatalytic water oxidation. Energy Environ. Sci. 6, 987–993. (doi:10.1039/c2ee24148j)
19. Townsend TK, Sabio EM, Browning ND, Osterloh FE. 2011 Photocatalytic water oxidation
with suspended alpha-Fe2 O3 particles—effects of nanoscaling. Energy Environ. Sci. 4, 4270–
4275. (doi:10.1039/c1ee02110a)
Downloaded from http://rsta.royalsocietypublishing.org/ on December 22, 2016
20. Naidu BS, Gupta U, Maitra U, Rao CNR. 2014 Visible light induced oxidation of water by rare
17
earth manganites, cobaltites and related oxides. Chem. Phys. Lett. 591, 277–281. (doi:10.1016/
j.cplett.2013.10.089)
39. Park H, Choi W, Hoffmann MR. 2008 Effects of the preparation method of the ternary
18
CdS/TiO2 /Pt hybrid photocatalysts on visible light-induced hydrogen production. J. Mater.
Chem. 18, 2379–2385. (doi:10.1039/B718759A).
58. Nalbandian L, Evdou A, Zaspalis V. 2011 La1−x Srx My Fe1−y O3−δ perovskites as oxygen-
19
carrier materials for chemical-looping reforming. Int. J. Hydrogen Energy 36, 6657–6670.
(doi:10.1016/j.ijhydene.2011.02.146)