You are on page 1of 8

Catalysis Today 185 (2012) 270–277

Contents lists available at ScienceDirect

Catalysis Today
journal homepage: www.elsevier.com/locate/cattod

Heterojunction semiconductors: A strategy to develop efficient photocatalytic


materials for visible light water splitting
Jum Suk Jang a , Hyun Gyu Kim b , Jae Sung Lee a,∗
a
Department of Chemical Engineering, Division of Advanced Nuclear Engineering, Pohang University of Science and Technology (POSTECH), San 31 Hyoja-dong, Pohang 790-784,
Republic of Korea
b
Busan Center, Korea Basic Science Institute (KBSI), Busan 609-735, Republic of Korea

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

Article history: Heterojunction semiconductors are discussed as a strategy to develop efficient visible light photocatalysts
Received 29 May 2011 for water splitting. The concept has been demonstrated in photovoltaic cells and optoelectronic devices,
Received in revised form 8 July 2011 for which junction-type semiconductors show greatly enhanced efficiency compared to the devices con-
Accepted 11 July 2011
sisting of a single semiconductor. We applied this proven concept to fabricate photocatalysts of inorganic
Available online 1 September 2011
semiconductors. Thus heterojunction structures of Shottky junctions, p–n junctions (or p–n diode), p–n
junctions with Ohmic layer, and bulk heterojunctions were fabricated and their photoactivity was tested
Keywords:
for reduction or oxidation of water under visible light. The formation of heterojunctions results in the
Heterojunction
Photocatalysts
efficient separation of electron–hole pairs to minimize the energy-wasteful electron–hole recombination,
Water splitting which leads to the high photocatalytic activity. As the complexity and sophistication of the photocatalyst
Recombination fabrication increased, the photoactivity also increased. Modern nanomaterial synthetic techniques were
employed to bring into reality the highly engineered material configurations.
© 2011 Elsevier B.V. All rights reserved.

1. Introduction alysts are needed to meet the requirements of future environmental


and energy technologies driven by solar energy.
Sunlight and water is a clean, renewable and most abundant Like other photocatalytic reactions, PWS is initiated when
energy source and natural resource on the Earth, respectively. Their a semiconductor photocatalyst absorbs photons with energies
conversion to hydrogen has been considered as an ideal solution to greater than its band gap energy (Eg). This absorption creates
counter the depletion and environmental problems of fossil fuels. excited photoelectrons (e− ) in the conduction band and holes (h+ )
Photocatalytic water splitting (PWS) is a promising and sustainable in the valence band of the semiconductor as schematically shown
technology for the purpose, since H2 could be produced directly in Fig. 1 [11]. The photoelectrons and holes reduce and oxidize
from water and solar light through the process. Photocatalysts water, respectively, to produce the 2:1 mixture of H2 and O2 by
that respond to visible light are needed to utilize the main part of the following reactions:
the solar spectrum for production of hydrogen energy by splitting
water. Traditional visible-light photocatalysts are either unstable Oxidation : H2 O + 2h+ → 2H+ + 1/2O2 (1)
in water upon illumination (e.g., CdS, CdSe) [1] or have low activity
Reduction : 2H+ + 2e− → H2 (2)
(e.g., WO3 , Fe2 O3 ) [2]. Recently, some UV-active oxides are mod-
ified to function as visible-light photocatalysts by substitutional Overallreaction : H2 O → H2 + 1/2O2 (3)
doping of metals as in Nix In1−x TaO4 [3] and (V-, Fe-, or Mn-)TiO2 [4],
reduction of TiO2 (as in TiOx (x < 2)) [5], and anion doping or making The overall reaction involves four-electron transfer (per O2
compounds with N, C, and S (as in TiO2−x Nx [6],TiO2−x Cx [7], TaON molecule) and is usually promoted by metal or metal oxide co-
[8], and Sm2 Ti2 O2 S2 [9]). However, these doped materials, in gen- catalysts (denoted as cat 1 and cat 2 in Fig. 1) deposited on the
eral, show only a little absorption in the visible-light region, leading semiconductor surfaces. These co-catalysts are known to collect
to low activities [10]. New and more efficient visible-light photocat- charge carriers and provide catalytic reaction sites for these reac-
tions [12]. The reaction involves the standard Gibbs free energy
change G◦ of 237 kJ/mol or 1.23 eV (per electron). In prac-
tice, some overpotentials are needed to drive the reactions fast
enough, and thus the semiconductor should possess Eg greater than
∗ Corresponding author. Tel.: +82 54 279 2266; fax: +82 54 279 5528. 1.6–1.8 eV to be used for water splitting. In order to absorb a large
E-mail address: jlee@postech.ac.kr (J.S. Lee). part of the visible light irradiation, however, its value should be

0920-5861/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.cattod.2011.07.008
J.S. Jang et al. / Catalysis Today 185 (2012) 270–277 271

Fig. 1. A model that illustrates the principle of photocatalytic water splitting.

less than ca. 2.2 eV. In addition, for the facile electron and hole sol–gel route was adopted to synthesize the CaFe2 O4 nanoparti-
transfer to reduce and oxidize water, the conduction band should cles. A stoichiometric ratio of Ca(NO3 )2 ·4H2 O (Aldrich, 99.999%)
be located at a potential more negative than the reduction poten- and Fe(NO3 )3 ·9H2 O (Aldrich, 99.999%) was mixed in 30% aqueous
tial of water (0.0 V vs. NHE), while the valence band more positive NH3 solution, and the mixture was stirred at room temperature
than oxidation potential of water (1.23 V vs. NHE) as indicated in for 24 h. The composite of CaFe2 O4 /PbBi2 Nb2−x Wx O9 was made
Fig. 1. Therefore, the photocatalytic materials for visible light water by hydrothermal treatment of preformed PbBi2 Nb1.9 W0.1 O9 and
splitting should satisfy these two requirements of proper band gap CaFe2 O4 sol at 150 ◦ C for 7 day to obtain 2 wt% CaFe2 O4 loading on
energy (1.6–1.8 eV < Eg < 2.2 eV) and band positions. Other require- PbBi2 Nb1.9 W0.1 O9 .
ments for photocatalytic materials include durability in aqueous The W on surfaces of PbBi2 Nb2−x Tix O9 powders was deposited
electrolyte solutions, competitive cost, high crystallinity with little by using a cold-wall CVD chamber with a shower-head. The base
defects, and good conductivity. pressure of the reactor was below 10−3 Torr, and the operating
In contrast to single photocatalysts, heterojunction semicon- pressure was in a range of 0.1–1 Torr during the W deposition. The
ductors are an efficient system for electron–hole separation to white colored tungsten salt, viz., W(CO)6 with a high vapor pressure
minimize the energy wasteful electron–hole recombination. It has was used as the source for metal oxide chemical vapor deposition
been demonstrated in photovoltaic cells and photoelectrochemical (MOCVD). The solid tungsten hexacarbonyl source was contained
(PEC) cells that junction semiconductors show greatly enhanced in a stainless steel bubbler. During the W deposition, the bubbler
activities compared to the devices consisting of a single semi- temperature was maintained at 80 ◦ C. Delivery line was also heated
conductor. For example, the fundamental unit of current silicon to 110 ◦ C to prevent the source condensation. Argon was used as
solar cells is made of p–n junction of p-type and n-type silicon. a carrier gas with the flow rate of 1.5 mol/s. Deposition tempera-
Nozik constructed an n-TiO2 /p-GaP diode electrode for a PEC cell, ture was varied from 450 to 500 ◦ C. Thus produced W metal-loaded
which was found to be much more active than either n-TiO2 or PbBi2 Nb2−x Tix O9 catalysts were then oxidized in air (25 ␮mol/s) at
p-GaP single-component electrode for the decomposition of water 473 K for 1 h.
[13,14]. Multi-junction solar cells of remarkable energy conversion The synthesis of CaFe2 O4 was carried out by the polymer com-
efficiencies have been reported [15]. plex method. Calcium acetate monohydrate (Ca(CH3 CO2 )2 , 99.0%,
There are several junction types: (i) Schottky junction, (2) Aldrich), magnesium carbonate (MgCO3 , 99.9%, Aldrich), ethylene
p–n diode, (3) p–n junction with Ohmic layer, and (4) bulk glycol (C2 H6 O2 , Kanto Chemicals), citric acid (C6 H8 O7 , Wako) and
heterojunction. In this critical review, we highlight particulate iron nitrate nonahydrate (Fe(NO3 )3 9H2 O, 99.99%, Aldrich) were
versions of the heterojunction semiconductors that we have used as starting materials. The citric acid (CA) was added into water
developed for photocatalytic applications based on these four het- under constant agitation, at 60–70 ◦ C. Next, the salts of calcium car-
erojunction types. There have been many reports on composite bonate and iron nitrate hydrate were dissolved in citric acid–water
photocatalysts, but not many works deliberately fabricated the het- solution to obtain metal–citrate complex. Finally, ethylene glycol
erojunctions to improve the performance of the photocatalysts, (EG) was added to the mixture to yield a mass proportion of 60% CA
or attributed the observed improvements to the junction effects. to 40% EG. The mixture was kept on hot plate (80 ◦ C) till it became
New concepts and methods are explored to fabricate more effi- a transparent colorless solution. The colorless solution was then
cient photocatalysts by utilization of modern material processing heated at 130 ◦ C for several hours to obtain a polymeric gel. The
techniques. These heterojunction photocatalysts are highly active viscous polymeric product was pyrolyzed at about 300–500 ◦ C to
under visible light and stable under typical photocatalytic reaction form the precursor powders. Thus obtained powder was pressed
conditions. in the form of pellets, which were calcined at 500–1200 ◦ C for 2 h
in an electric furnace to obtain nanocrystalline CaFe2 O4 /MgFe2 O4 .
RuO2 was first loaded on CaFe2 O4 /MgFe2 O4 photocatalyst by an
2. Experimental impregnation method, and then, 1 wt% of Pt metals were deposited
on RuO2 /CaFe2 O4 /MgFe2 O4 photocatalysts by using a photodepo-
2.1. Photocatalyst preparation sition method using metal H2 PtCl6 under a 450-W Xe-arc lamp
(Oriel) with UV cut-off filter ( ≥ 420 nm) and then was dried in
The basic catalyst PbBi2 Nb2 O9 was synthesized by the conven- an oven at 373 K in air for 2 h.
tional solid-state method. Thus, a stoichiometric mixture of PbO
(Aldrich, 99.999%), Bi2 O3 (Aldrich, 99.99%), and Nb2 O5 (Aldrich,
99.999%), was ground in a mortar. The pelletized powders were 2.2. Characterization and photocatalytic reactions
calcined at 1123 K for 24 h in static air and sintered at 1323 K for
24 h. The oxide form of doping component (WO3 , TiO2 ) was added The crystal structure photocatalysts thus obtained were deter-
in the precursor mixture to synthesize doped photocatalysts. The mined by X-ray diffractometer (Mac Science Co., M18XHF), and
272 J.S. Jang et al. / Catalysis Today 185 (2012) 270–277

the morphology was determined by scanning electron microscopy 3.2. Heterojunction between two p–n semiconductors
(SEM, Hitachi, S-2460N) and high-resolution transmission elec-
tron microscopy (HRTEM, Phillips model CM 200). Light absorption A p–n junction is formed by joining p-type and n-type semicon-
property was determined by UV/Vis diffuse reflectance spec- ductors in close contact. It forms an elementary building block of
troscopy (Shimadzu, UV 2401). almost all semiconductor electronic devices such as diodes, tran-
The photocatalytic reactions in aqueous solution were car- sistors, solar cells, and light emitting diodes (LED). As p–n type
ried out at room temperature in a closed system using a 450-W semiconductors are joined, electrons near the p–n interface tend
Xe-arc lamp (Oriel) with UV cut-off filter ( > 420 nm) placed to diffuse into the p-region. As electrons diffuse, they leave posi-
in an inner irradiation-type 200-mL Pyrex reaction cell. The tively charged ions in the n-region. Similarly, holes near the p–n
H2 evolution was examined in an aqueous solution (distilled interface begin to diffuse into the n-type region leaving ions with
H2 O(170 mL) + CH3 OH(30 mL)), and the O2 evolution in an aque- negative charge. The regions nearby the p–n interfaces lose their
ous AgNO3 solution (0.05 mol L−1 , 200 mL), both containing 0.3 g neutrality and become charged, forming the space charge region
of the catalyst. The quantum yield (QY) was calculated as twice the or depletion layer. Then the electric field is created such that elec-
number of generated H2 molecules or four times the number of gen- trons and holes diffuse in opposite direction to reduce the chance
erated O2 molecules divided by the number of photons absorbed of recombination.
by the photocatalyst. The number of absorbed photons was deter- We have tried to apply this concept of p–n heterojunction
mined with a light flux meter (1815-C, Newport) with the light to the synthesis of particulate photocatalysts [24]. This type of
sensor attached to the photocatalytic reactor. junction structures reported in the literature includes CaFe2 O4 -
PbBi2 Nb1.9 W0.1 O9 [24], CdS–AgGaS2 [25], CoAl2 O4 –Fe2 O3
[26], CuBi2 O4 –WO3 [27], Cu–Ti–O nanotube arrays [28],
and so on [29–31]. Here a heterojunction photocatalyst of
CaFe2 O4 –PbBi2 Nb1.9 W0.1 O9 is discussed. As the suitable n-type
3. Results and discussion semiconductor, we employed PbBi2 Nb2 O9 , the layered perovskite
semiconductor discussed above [23]. The material was doped with
3.1. Schottky junction photocatalysts hexavalent tungsten WVI to obtain n-type PbBi2 Nb1.9 W0.1 O9 . In
addition to the valency, the similarity of its size (0.74 nm) to that
The Schottky junction is a commonly adopted strategy to induce of NbV (0.78 nm) was also considered in the choice of tungsten
an effective charge separation in optoelectronic devices, photocata- dopant. CaFe2 O4 is a p-type semiconductor with the band gap
lysts or photoelectrochemical cells [16,17]. In general, the Schottky of 1.9 eV as reported earlier [24]. In order to combine these two
junction consists of n- or p-type semiconductor (SC) and a metal semiconductor particles to form a p–n junction, we devised the
contact spread over SC [13,14]. These single Schottky junction semi- configuration of the diode particles: (i) nano-crystals of p-CaFe2 O4
conductors are the most common type of photocatalysts, which on the surface of (ii) n-PbBi2 Nb1.9 W0.1 O9 as submicron-sized
can be fabricated with nanoscale metal islands on semiconductor base material. Thus, CaFe2 O4 nanoparticles and PbBi2 Nb1.9 W0.1 O9
surface. The Schottky junction forms an electric field that facili- submicron particles were synthesized individually by sol–gel
tates the separation of photogenerated electrons and holes [18–20]. and solid-state reaction methods, respectively. Then, produced
Most of particulate semiconductor photocatalysts employ a metal individual components were combined in by a hydrothermal
or a metal oxide forming the Schottky junction, called co-catalyst. reactor (150 ◦ C for 7 days) to obtain CaFe2 O4 nano-islands formed
Without it, the semiconductor alone shows little or no activity. The over larger particles of PbBi2 Nb1.9 W0.1 O9 .
critical role of the co-catalyst has been demonstrated in a recent Fig. 3a shows a HRTEM image of a typical
example by Domen et al. who developed Rh/Cr2 O3 (core/shell) CaFe2 O4 –PbBi2 Nb1.9 W0.1 O9 heterojunction photocatalyst par-
nanoparticles photodeposited on (Ga1−x Znx )(N1−x Ox ) [21,22]. The ticle. It clearly exhibits the existence of nano-islands dispersed
carefully designed co-catalyst suppressed water formation from over the perovskite base material. The magnified view of a nano-
the back-reaction of evolved H2 and O2 during the water splitting island reveals the nanocrystallinity of 5–10 nm CaFe2 O4 (Fig. 3b)
reaction. and it also shows a highly crystalline, layered base material cor-
Our example of Schottky junction photocatalyst is responding to PbBi2 Nb1.9 W0.1 O9 (layer thickness 0.4 nm, Fig. 3c).
Pt/PbBi2 Nb2 O9 [23]. The semiconductor PbBi2 Nb2 O9 , is an The phase identification was also confirmed by elemental analysis
Aurivillius-phase layered perovskite, and 1 wt% Pt was loaded with in situ energy-dispersive X-ray microanalysis (EDX) carried
on its surface by the photodecomposition method. As shown in out during the microscopic study. Each island forms a nanoscale
Fig. 2a, Pt forms islands of 5–8 nm on the surface of PbBi2 Nb2 O9 p–n junction at the interface, which plays a crucial role in the
forming a Schottky junction. Thus, PbBi2 Nb2 O9 represents an photocatalytic process.
undoped, single phase oxide photocatalyst with Eg of 2.88 eV. This The working principle of p–n heterojunction is shown with an
small band gap was attributed to the formation of a hybrid valence energy band model for one p–n junction (Fig. 3c). Here, band posi-
band between O2p and Pb6s/Bi6s orbitals as revealed by the band tions were determined by the flat-band potential measurements
structure calculation [23]. [32]. At thermal equilibrium, the Fermi levels of two semiconduc-
The operation of a Schottky junction photocatalyst is depicted tors align. When the device is immersed in an electrolyte, the band
schematically in Fig. 2b. Across the n-type semiconductor/metal edges of the conduction and valence bands bend as indicated. The
junction, the electric field is established as depicted. Visible light is operation of the p–n heterojunction is initiated by the absorption of
absorbed by the semiconductor creating electron–hole pairs. Pho- the band-gap photons in both p–n semiconductors. Photogenerated
toelectrons slide down the bent conduction band to the metal and holes and electrons separate under the influence of the electric field,
holes move up towards the electrolyte inducing electron–hole sep- corresponding to the energy diagram in Fig. 4c. Thus, holes move to
aration. Here, the conduction band position of SC and work function the p-CaFe2 O4 side, and electrons to the n-PbBi2 Nb1.9 W0.1 O9 side.
of the metal should be matched as in Fig. 2b to allow the pho- To establish this charge flow, the band positions of two semicon-
toelectrons to flow from SC to the metal. The other role of metal ductors should stagger as shown with that of p-type semiconductor
is to provide catalytic sites for the reduction of water to produce located at a more negative potential. The net effect of the p–n
hydrogen. These factors are expected to contribute to an improved diode formation is efficient separation of electron–hole pairs by
efficiency. physically separating reduction and oxidation sites, which limits
J.S. Jang et al. / Catalysis Today 185 (2012) 270–277 273

Fig. 2. TEM images of 1 wt% Pt–PbBi2 Nb2 O9 and a schematic energy band model of Schottky junction.

Fig. 3. Proposed photocatalytic nanodiode consisting of p-CaFe2 O4 and n-PbBi2 Nb1.9 W0.1 O9 : (a) HRTEM image showing nanocrystalline p-CaFe2 O4 islands decorating a large
n-PbBi2 Nb1.9 W0.1 O9 particle, (b) schematic of the p–n heterojunction working for water oxidation, and (c) an energy band model of a p–n heterojunction (after Ref. [24]).

Fig. 4. (a) Schematic illustration of fabrication process of PNHO. (b) Typical HRTEM of the fabricated PNHO showing WO3 nanodimensional entities stacking over the surfaces
of perovskite base material. (c) Cross-sectional schematic of a typical single WO3 -encapsulated W layer that stacks over base material. (d) Schematic energy band model
diagram of the system (after Ref. [34]).
274 J.S. Jang et al. / Catalysis Today 185 (2012) 270–277

Fig. 5. (a) A TEM image of a CFO/MFO particle. (b) A HRTEM image of interfaces between semiconductor nanoparticles. (c) A schematic model of interpenetrating domains
of CFO (white) and MFO (gray). (d) An energy band diagram showing band gap energies and band positions of p-CFO and n-MFO: CB, conduction band; VB, valence band. (e)
Operating principle of a BHJ promoted by co-catalysts (after Ref. [46]).

the energy-wasteful electron–hole recombination. By introducing experimental observations, we propose a schematic model for
semiconductors of different band gap energies, the configuration PNHO as illustrated in Fig. 4c. The WO3 -encapsulated W layer on
can also extend the range of absorbed light in solar spectrum. These PbBi2 Nb1.9 Ti0.1 O9 appears to play a critical role of Ohmic junction in
effects will lead to the higher photocatalytic activity. Formation of the photocatalytic process. Its presence was confirmed by observ-
heterojunction between semiconductors of the same kinds or with ing both W0 and W6+ in X-ray photoelectron spectroscopy (XPS)
a neutral semiconductor can also contribute to the charge separa- analysis where the deconvolution of peaks indicated a WOx /W kind
tion if their band positions are correct for charge transfer in the of configuration in the Ar+ etched samples which otherwise domi-
desired directions [33]. But without band bending cross the deple- nantly showed peaks only due to WO3 .
tion region, the beneficial effect on charge separation should be less The working principle of PNHO is schematically shown in Fig. 4d.
pronounced than p–n junction semiconductors. Again, the band positions for PbBi2 Nb1.9 Ti0.1 O9 were determined
from the flat band potential measurements. Upon absorption of
3.3. Heterojunction with Ohmic layer the band gap photons both in the p–n semiconductors, the holes
and electrons are separated in space-charge regions of respective
It has been noted that the interface between two semicon-
ductors forming p–n heterojunction would contain an ample
amount of defects that could hamper charge transfer pro-
cesses by trapping charge carriers. In an attempt to alleviate
the problem, we have adopted the concept of a p–n hetero-
junction structure with the Ohmic junction (PNHO) [34]. Thus,
PNHO photocatalyst powders were fabricated with the configu-
ration of p-semiconductor/metal/n-semiconductor by combining
solid-state reaction (SSR) with chemical vapor deposition (CVD).
Thus, WO3 /W/PbBi2 Nb1.9 Ti0.1 O9 photocatalyst was fabricated by
depositing the tungsten metal clusters over the p-type perovskite
base material by CVD with W(CO)6 , and later partly oxidizing the
surfaces of these W clusters to obtain n-WO3 overlayers on the top
of metallic W layers. Fig. 4a illustrates the schematic fabrication
process of WO3 /W/PbBi2 Nb1.9 Ti0.1 O9 . The WO3 is a well-known
n-type semiconductor, and PbBi2 Nb1.9 Ti0.1 O9 is derived from
PbBi2 Nb2 O9 doped with Ti4+ in order to obtain p-type semicon-
ductivity.
Fig. 4b shows a HRTEM image of typical PNHO particles
that clearly exhibit the existence of 30–40 nm WO3 clusters
stacking over the surface of a large PbBi2 Nb1.9 Ti0.1 O9 parti-
cle of 150–200 nm. The phase identification was also confirmed
by elemental analysis with in situ EDX analysis. Based on the Fig. 6. Optical transitions probed by UV–vis spectroscopy for various photocatalysts.
J.S. Jang et al. / Catalysis Today 185 (2012) 270–277 275

Table 1
Band gap energies and photocatalytic activities for H2 and O2 evolution.

Catalysts Band gap energy Hydrogen evolution Oxygen evolution

Eg (eV) ab (nm) ␮mol/h Q.Y. (%) ␮mol/h Q.Y. (%)

RuO2 /MgFe2 O4 /CaFe2 O4 /Pt 2.0 620 82.7 10.1 – –


WO3 /W/PbBi2 Nb1.9 Ti0.1 O9 2.79 443 49.3 6.1 741 41
CaFe2 O4 /PbBi2 Nb1.9 W0.1 O9 2.75 450 34.8 4.2 675 38
PbBi2 Nb2 O9 2.88 431 7.6 0.95 520 29
CaFe2 O4 1.99 623 Trace 0 Trace 0
TiO2−x Nx 2.73 451 Trace 0 221 14

Catalysts loaded with 1 wt% Pt, 0.3 g; light source, 450 W W-Arc lamp (Oriel) with UV cut-off filter ( ≥ 420 nm). Reaction was performed in aqueous methanol solution for
H2 generation (methanol 30 ml + distilled water 170 ml) or in an aqueous AgNO3 solution for O2 generation (0.05 mol/l, 200 ml).

semiconductors and become available for redox reactions. Accord- that electrons of n-SC and holes of p-SC are drawn to the Ohmic
ingly, the generated minority carriers, viz., holes in n-type and layer.
electrons in p-type semiconductors, move to the front of the sur-
face and are injected to the electrolyte solution to initiate oxidation 3.4. Bulk heterojunction
and reduction reactions, respectively. In addition, the photogen-
erated majority carriers in each semiconductor are then bound The concept of bulk heterojunction (BHJ) originated from poly-
to migrate towards the Ohmic W layer interfaced between two mer photovoltaic cells, which showed dramatically improved
semiconductors, where the carriers undergo nonradiative recom- efficiency [35,36]. In the BHJ, electron donor and acceptor
bination. Thus, one-photon is lost by this recombination. Yet the molecules are blended together. If the domain size in either mate-
remaining electron–hole pair can promote the redox reactions at rial is similar to the exciton (electron–hole pair) diffusion length,
the semiconductor-fluid interface, thereby facilitating the reduc- then wherever an exciton is photogenerated, it is likely to diffuse to
tion reaction on p-PbBi2 Nb1.9 Ti0.1 O9 and the oxidation reaction the interface and break up. The situation is more suitable for high
on n-WO3 . In addition to this physical separation of the reduc- efficiency than a planar heterojunction, or bi-layer device, where
tion and oxidation sites, PNHO provides higher net photon energies much longer diffusion of the exciton is required to reach the inter-
needed for the redox reactions by partial addition of two band face. The device proposed here is an all-inorganic version of BHJ,
gaps (Eg-composite), while individual band gaps (Eg-p and Eg- where p-type and n-type semiconductors form interpenetrating
n) still absorb visible light. The large energy difference between networks on nano-scale, and high efficiency is expected by the
photoelectron and hole (Eg-composite) also reduces the chance same principle. Although there is no precedent examples claim-
of recombination as for the large band-gap UV active photocata- ing BHJ formation, there are many examples in the literature of
lysts. These effects lead to the higher photocatalytic activity. For two semiconductors forming interpenetrating networks on nano-
the operation of PNHO, the work function of the Ohmic layer metal scale: CdS–TiO2 [37], MoS2 –CdS [38], Pb[Zrx Ti1−x ]O3 (PZT)–TiO2
should be matched with band positions of two semiconductors so [39], LaVO4 /TiO2 [40], and so on [41–45].

Fig. 7. A schematic summary of (a) Schottky junction, (b) p–n heterojunction between two semiconductors, (c) p–n heterojunction with Ohmic Layer, and (d) bulk
heterojunction.
276 J.S. Jang et al. / Catalysis Today 185 (2012) 270–277

Here this concept is exploited with a configuration of RuO2 /n- showed a slight red shift of the absorption edge from the base mate-
MgFe2 O4 (MFO)/p-CaFe2 O4 (CFO)/Pt [46], These spinel ferrites are rial PbBi2 Nb2 O9 giving Eg of 2.75 eV, and WO3 /W/PbBi2 Nb1.9 Ti0.1 O9
the most important magnetic materials, yet they also show inter- PNHO showed Eg of 2.79 eV.
esting photocatalytic properties for hydrogen or oxygen evolution The photocatalytic activity of the heterojunction photocatalysts
from water [47–51]. Desirable properties of these materials as pho- discussed above was evaluated for two half-reactions of water split-
tocatalysts are resistance to photocorrosion in aqueous solutions ting; hydrogen evolution from an aqueous solution of methanol as
and narrow band gaps near 2 eV. But their activity as single phase a hole scavenger and oxygen evolution from an aqueous solution of
photocatalysts was very low. As a means to enhance the photocat- silver nitrate as an electron scavenger. The results are summarized
alytic activity of the ferrites, a bulk heterojunction structure was in Table 1. The quantum yield (QY) is defined as 2 × number of H2
fabricated, which was promoted by two co-catalysts, Pt and RuO2 . molecules (or 4 × number of O2 molecule) generated per absorbed
Fig. 5(a) shows a TEM image of a typical BHJ CFO/MFO particle photon. The exact comparison among different heterojunction pho-
of 150–200 nm decorated with 2–5 nm clusters on its surface. The tocatalysts is not possible because compositions are not the same.
small nanoparticles on the surface were found to be Pt and RuO2 by Yet, the data are still interesting because all photocatalysts with
in situ EDX analysis. The elemental mapping with EDX also clearly similar band gap energies were tested under the same reaction
showed that the elements of Ca, Fe, Mg, and O were uniformly conditions.
distributed over the BHJ particle, confirming the composition of The single phase PbBi2 Nb2 O9 photocatalyst forming a Schottky
CFO/MFO. According to HRTEM image in Fig. 5(b), the 150–200 nm junction with Pt showed hydrogen evolution corresponding to QY
BHJ particle was made of many randomly mixed and interfacing of 0.95%. The QY for oxygen evolution was very high at 29%. The ref-
20–30 nm particles. When n-MFO [45] and p-CFO [47] make the erence photocatalyst, Pt-loaded TiO2−x Nx , showed trace amount of
interface, a p–n diode heterojunction is formed in nano-scale. There H2 evolution, but QY for O2 evolution was ca. 12%. The high QY for O2
are many such junctions randomly distributed in the large par- evolution indicates that PbBi2 Nb2 O9 has high oxidation potential.
ticle forming a bulk heterojunction. The conceptual structure of Indeed, the activity of PbBi2 Nb2 O9 for oxidative decomposition of
the BHJ is schematically shown in Fig. 5(c). Fig. 5(d) shows a band isopropyl alcohol was about twice as high as that of TiO2−x Nx [23].
diagram for a single p-CFO/n-MFO heterojunction. To form BJH, it The p–n heterojunction catalyst of CaFe2 O4 –PbBi2 Nb1.9 W0.1 O9
is essential to have separate phases of p- and n-semiconductors exhibited better performance than single phase Pt/PbBi2 Nb2 O9
instead of solid solution or mixed phase. Our synthesis technique even without any co-catalyst. The QY for hydrogen evolution was
(polymer complex method) provided one-pot production of the greatly improved to ca. 4.2%. The QY of O2 evolution over p–n het-
two separate semiconducting, stable metal oxide phases of nano- erojunction catalyst was estimated to be ca. 38%. When Ohmic
dimensions. The formation of the separate phases was confirmed by layer was introduced to form PNHO (WO3 /W/PbBi2 Nb1.9 Ti0.1 O9 ),
X-ray diffraction analysis and it may be due to different structures the QY for hydrogen evolution was further increased to 6.1% and
of the two ferrites (orthorhombic vs. cubic) and the considerable that for O2 evolution was estimated to be ca. 41%. To the best of
size difference between Ca2+ (0.99 Å) and Mg2+ (0.65 Å). our knowledge, this is the highest value ever reported for the reac-
This bulk heterojunction was further loaded with Pt and RuO2 tion over semiconductor photocatalysts under visible light. Finally,
co-catalysts to facilitate the faster electron–hole transfer to the the BHJ photocatalyst (RuO2 /n-MgFe2 O4 /p-CaFe2 O4 /Pt) showed a
electrolyte. The working principle of such a multi-junction device very high QY for H2 evolution of 10.1%, which also represents the
can be explained with reference to Fig. 5(e). Like p–n junction dis- highest QY for H2 evolution ever reported for oxide photocatalysts
cussed above, a nano-scale p–n junction is formed at the interface. [46]. The results in Table 1 thus demonstrate that formation of het-
When photons are absorbed by the device, electron–hole pairs or erojunctions is a successful strategy to fabricate highly efficient
excitons are generated. The excitons diffuse in each semiconductor photocatalysts.
and break up when they arrive at the p–n junction. According to the
potential established across the junction, holes move to the p-CFO
side and electrons to the n-MFO side. For a homojunction of a single 4. Conclusion
semiconductor or a planar heterojunction, the diffusion distance
for the exciton would be long and the electron–hole recombina- This critical review discussed heterojunction semiconductors
tion would take place before it reaches the interface. In the bulk as an effective strategy to develop a highly efficient photocata-
heterojunction where p- and n-type semiconductors are dispersed lyst working under visible light for water splitting. The concepts
interfacing each other, the probability that the exciton reaches the of heterojunction are summarized in Fig. 7. These proven con-
interface and dissociates is higher, especially when the domain size cepts in solar cells and other optoelectronic devices have been
of the semiconductor particles is similar to the exciton diffusion successfully applied here to fabricate photocatalysts of inor-
length. The co-catalyst Pt provides sites for electron collection and ganic semiconductors. The formation of heterojunctions results
reduction reactions, whereas RuO2 becomes the site for hole collec- in the efficient separation of electron–hole pairs to minimize the
tion and oxidation reactions, although in reality these co-catalysts energy-wasteful electron–hole recombination, which leads to the
should be dispersed randomly on both types of semiconductors. All high photocatalytic activity. As the complexity and sophistica-
these factors would lead to enhanced photocatalytic activity. tion of the photocatalyst fabrication increased, the photoactivity
also increased. Modern nanomaterial synthetic techniques were
3.5. Photocatalytic activity for reduction or oxidation of water employed to bring into reality the highly engineered material con-
figurations. At the present time, our results as well as those of
The UV–visible diffuse reflectance spectra of various photocata- other groups show still low efficiencies of solar light-to-hydrogen
lysts are compared in Fig. 6 and the band gap energies determined conversion. Also, the photoactivity reported here was obtained in
from them are summarized in Table 1. As a reference photocatalyst, sacrificial agent systems (CH3 OH and AgNO3 ), which have a lim-
nitrogen-doped TiO2 (TiO2−x Nx ) was employed for comparison. Our ited practical value. There still remains the grand challenge to find
PbBi2 Nb2 O9 sample, being a single phase material, showed a steep materials to absorb more photons from solar light and convert them
onset of absorption spectrum around green wavelengths sweeping to chemical energy more efficiently with minimal loss by charge
major visible region. This sharp absorption edge indicates single recombination. These materials could be utilized to their maxi-
band gap absorption and is expected to be beneficial for high photo- mal potential by applying concepts of heterojunction formation
activity [10]. The spectrum of CaFe2 O4 –PbBi2 Nb1.9 W0.1 O9 powders introduced in this review.
J.S. Jang et al. / Catalysis Today 185 (2012) 270–277 277

Acknowledgment [24] (a) H.G. Kim, P.H. Borse, W. Choi, J.S. Lee, Angew. Chem., Int. Ed. 117 (2005)
4661–4665;
(b) Y. Matsumoto, J. Solid State Chem. 126 (1996) 227–234.
This work was supported by Hydrogen Energy Center and Korea [25] (a) J.S. Jang, D.W. Hwang, J.S. Lee, Catal. Today 120 (2007) 174–181;
Center for Artificial Photosynthesis (KCAP) funded by the Ministry (b) J.S. Jang, S.H. Choi, N. Shin, C. Yu, J.S. Lee, J. Solid State Chem. 180 (2007)
of Science and Technology of Korea. 1110–1118.
[26] K.S. Ahn, Y. Yan, M.S. Kang, J.Y. Kim, S. Shet, H. Wang, J. Turner, M. Al-Jassim,
Appl. Phys. Lett. 95 (2009), 022116/1–3.
References [27] T. Arai, M. Yanagida, Y. Konishi, Y. Iwasaki, H. Sugihara, K. Sayama, J. Phys. Chem.
C 111 (2007) 7574–7577.
[1] G.C. De, A.M. Roy, S.S. Bhattacharya, Int. J. Hydrogen Energy 21 (1996) 19–23. [28] G.K. Mor, O.K. Varghese, R.H.T. Wilke, S. Sharma, K. Shankar, T.J. Latempa, K.S.
[2] (a) D.W. Hwang, J. Kim, T.J. Park, J.S. Lee, Catal. Lett. 80 (2002) 53–57; Choi, C.A. Grimes, Nano Lett. 8 (2008) 1906–1911.
(b) P.H. Borse, H. Jun, S.H. Choi, S.J. Hong, J.S. Lee, Appl. Phys. Lett. 93 (2008), [29] M. Long, W. Cai, H. Kisch, J. Phys. Chem. C 112 (2008) 548–554.
173103/1–3; [30] M.F. Weber, M.J. Dignam, Int. J. Hydrogen Energy 11 (1986) 225–232.
(c) J.S. Jang, J. Lee, H. Ye, F.R.P. Fan, A.J. Bard, J. Phys. Chem. C 113 (2009) [31] M. Grätzel, Nature 414 (2001) 338–344.
6719–6724. [32] (a) J.R. White, A.J. Bard, J. Phys. Chem. 89 (1985) 1947–1954;
[3] Z. Zou, J. Ye, K. Sayama, H. Arakawa, Nature 414 (2001) 625–627. (b) A.M. Roy, G.C. De, N. Sasmal, S.S. Bhattacharyya, Int. J. Hydrogen Energy 20
[4] H. Yamashita, H. Harada, J. Misaka, M. Takeuchi, K. Ikeue, M. Anpo, J. Photochem. (1995) 627–630.
Photobiol. A 148 (2002) 257–261. [33] S.J. Hong, S. Lee, J.S. Jang, J.S. Lee, Energy Environ. Sci. 4 (2011) 1781–1787.
[5] R.G. Breckenridge, W.R. Hosler, Phys. Rev. 91 (1953) 793–802. [34] H.G. Kim, E.D. Jeong, P.H. Borse, S. Jeon, K. Yong, J.S. Lee, W. Li, S.H. Oh, Appl.
[6] R. Asahi, T. Morikawa, T. Ohwaki, K. Aoki, Y. Tao, Science 293 (2001) 269–271. Phys. Lett. 89 (2006), 064103/1–3.
[7] (a) S.U.M. Khan, M. Al-Shahry, W.B. Ingler, Science 297 (2002) 2243–2245; [35] G. Yu, A.J. Heeger, J. Appl. Phys. 78 (1995) 4510–4515.
(b) S. Sakthivel, H. Kisch, Angew. Chem., Int. Ed. 42 (2003) 4908–4911. [36] J.J.M. Halls, C.A. Walsh, N.C. Greenham, E.A. Marseglia, R.H. Friend, S.C. Moratti,
[8] G. Hitoki, T. Takata, J. Kondo, M. Hara, H. Kobayashi, K. Domen, Chem. Commun. A.B. Holmes, Nature 376 (1995) 498–500.
16 (2002) 1698–1699. [37] (a) J.S. Jang, W. Li, S.H. Oh, J.S. Lee, Chem. Phys. Lett. 425 (2006) 278–282;
[9] A. Ishikawa, T. Takata, J.N. Kondo, M. Hara, H. Kobayashi, K. Domen, J. Am. Chem. (b) J.S. Jang, H.G. Kim, P.H. Borse, J.S. Lee, Int. J. Hydrogen Energy 32 (2007)
Soc. 124 (2002) 13547–13553. 4786–4791;
[10] D.W. Hwang, H.G. Kim, J.S. Lee, J. Kim, W. Li, S.H. Oh, J. Phys. Chem. B 109 (2005) (c) J.S. Jang, S. Ji, S.W. Bae, H.C. Son, J.S. Lee, J. Photochem. Photobiol. A: Chem.
2012–2093. 188 (2007) 112–119.
[11] J.S. Lee, Catal. Surv. Asia 9 (2005) 217–227. [38] X. Zong, H. Yan, G. Wu, G. Ma, F. Wen, L. Wang, C. Li, J. Am. Chem. Soc. 130
[12] T. Sakata, in: N. Serpone, E. Pelizzetti (Eds.), Photocatalysis: Fundamentals and (2008) 7176–7177.
Applications, Wiley, New York, 1989. [39] H. Huang, D. Li, Q. Lin, Y. Shao, W. Chen, Y. Hu, Y. Chen, X. Fu, J. Phys. Chem. C
[13] A.J. Nozik, Appl. Phys. Lett. 29 (1976) 150–153. 113 (2009) 14264–14269.
[14] A.J. Nozik, Appl. Phys. Lett. 30 (1977) 567–569. [40] H. Huang, D. Li, Q. Lin, W. Zhang, Y. Shao, Y. Chen, M. Sun, X. Fu, Environ. Sci.
[15] O. Khaselev, J. Turner, Science 280 (1998) 425–427. Technol. 43 (2009) 4164–4168.
[16] S.U.M. Khan, S.A. Majumder, Int. J. Hydrogen Energy 14 (1989) 653–660. [41] M. Lim, Y. Zhou, Y. Guo, C. Sun, B. Woodd, L. Wanga, D. Hulicova-Jurcakova, J.
[17] J. Akikusa, S.U.M. Khan, Int. J. Hydrogen Energy 27 (2002) 863–870. Zou, V. Rudolph, G.Q.M. Lu, Electrochem. Commun. 11 (2009) 509–514.
[18] (a) K. Gurunathan, Int. J. Hydrogen Energy 29 (2004) 933–940; [42] X. Lin, J. Xing, W. Wang, Z. Shan, F. Xu, F. Huang, J. Phys. Chem. C 111 (2007)
(b) J.S. Jang, S.H. Choi, H.G. Kim, J.S. Lee, J. Phys. Chem. C 112 (2008) 18288–18293.
17200–17205. [43] X. Zhang, L. Zhang, T. Xie, D. Wang, J. Phys. Chem. C 113 (2009) 7371–7378.
[19] (a) B. Kraeutler, A.J. Bard, J. Am. Chem. Soc. 100 (1978) 4317–4318; [44] L. Zheng, Y. Zheng, C. Chen, Y. Zhan, X. Lin, Q. Zheng, K. Wei, J. Zhu, Inorg. Chem.
(b) J.S. Jang, K.Y. Yoon, X. Xiao, F.F. Fan, A.J. Bard, Chem. Mater. 21 (2009) 48 (2009) 1819–1825.
4803–4810. [45] J.S. Jang, S.J. Hong, J.Y. Kim, J.S. Lee, Chem. Phys. Lett. 475 (2009) 78–81.
[20] A. Fujishima, K. Honda, Nature 238 (1972) 37–38. [46] H.G. Kim, P.H. Borse, J.S. Jang, E.D. Jeong, O.S. Jung, Y.J. Suh, J.S. Lee, Chem.
[21] K. Maeda, K. Teramura, D. Lu, T. Takata, N. Saito, Y. Inoue, K. Domen, Nature 440 Commun. 39 (2009) 5889–5891.
(2006) 295. [47] L.G.J. De Haart, G. Blasse, Solid State Ionics 16 (1985) 137–140.
[22] K. Maeda, K. Teramura, D. Lu, N. Saito, Y. Inoue, K. Domen, J. Phys. Chem. C 111 [48] Y. Matsumoto, M. Omae, K. Sugiyama, E. Sato, J. Phys. Chem. 91 (1987) 577–581.
(2007) 7554–7560. [49] U. Steinike, K. Tkacova, J. Mater. Synth. Process. 8 (2000) 197–203.
[23] (a) H.G. Kim, D.W. Hwang, J.S. Lee, J. Am. Chem. Soc. 126 (2004) 8912–8913; [50] S. Zhuiykov, T. Ono, N. Yamazoe, N. Miura, Solid State Ionics 152–153 (2002)
(b) H.G. Kim, O.S. Becker, J.S. Jang, P.H. Borse, J.S. Lee, J. Solid State Chem. 179 801–807.
(2006) 1214–1218. [51] B.F. Decker, J.S. Kasper, Acta Crystallogr. 10 (1957) 332–337.

You might also like