Titanium dioxide crystals with tailored facets. - PDF Download Free (2024)

Review pubs.acs.org/CR

Titanium Dioxide Crystals with Tailored Facets Gang Liu,† Hua Gui Yang,‡,§ Jian Pan,†,∥ Yong Qiang Yang,†,⊥ Gao Qing (Max) Lu,*,∥ and Hui-Ming Cheng*,†,# †

Shenyang National Laboratory for Materials Science, Institute of Metal Research, Chinese Academy of Sciences, 72 Wenhua Road, Shenyang 110016, China ‡ Key Laboratory for Ultrafine Materials of Ministry of Education, School of Materials Science and Engineering, East China University of Science and Technology, 130 Meilong Road, Shanghai 200237, China § Centre for Clean Environment and Energy, Gold Coast Campus, Griffith University, Queensland 4222, Australia ∥ ARC Centre of Excellence for Functional Nanomaterials, Australian Institute for Bioengineering and Nanotechnology, The University of Queensland, Queensland 4072, Australia ⊥ Department of Materials Science & Technology, School of Chemistry and Materials Science, University of Science and Technology of China, 96 Jinzhai Road, HeFei 230026, China # Chemistry Department, Faculty of Science, King Abdulaziz University, Jeddah 21589, Saudi Arabia S Supporting Information *

3.4. Anisotropic Insertion of Lithium and Hydrogen 3.5. Synergistic Effects of Different Facets 3.6. Anisotropic Etching on Crystal Facets 3.7. Thermal Stability of Faceted TiO2 Crystal 4. Modification of the Electronic Structure and Interfacial Properties of Faceted TiO2 4.1. Doping 4.2. Creating Intrinsic Defects 4.3. Heterostructuring 4.3.1. Metal−TiO2 Heterostructures 4.3.2. Carbon−TiO2 Heterostructures 4.3.3. Semiconductor−TiO2 Heterostructures 5. Applications in Environment and Energy 5.1. Photocatalysis 5.2. Solar Cells 5.3. Lithium Ion Batteries 5.4. Other Applications 6. Concluding Remarks Associated Content Supporting Information Author Information Corresponding Authors Notes Biographies Acknowledgments References

CONTENTS 1. Introduction 2. Synthesis of TiO2 Crystals with Different Facets 2.1. Basic Strategies for Synthesis of Faceted TiO2 2.2. Faceted Anatase TiO2 2.2.1. Crystals with {101} Facets 2.2.2. Crystals with {001} Facets 2.2.3. Crystals with {010} Facets 2.2.4. Crystals with {110} Facets 2.2.5. Crystals with High-Index Facets 2.3. Faceted Rutile TiO2 2.4. Faceted Brookite TiO2 2.5. Faceted TiO2 Mesocrystals 2.6. Hollow Faceted TiO2 Crystals 2.7. Mesoporous Single Crystal TiO2 2.8. Films of Faceted TiO2 3. Unusual Properties of TiO2 Crystals with Different Predominant Facets 3.1. Surface Reconstruction 3.2. Anisotropic Surface Electronic Structures 3.3. Anisotropic Molecule/Cluster Adsorption

© 2014 American Chemical Society

9559 9561 9561 9562 9563 9564 9567 9569 9570 9571 9573 9576 9578 9579 9580

9587 9587 9589 9590 9591 9592 9594 9596 9596 9596 9597 9598 9598 9600 9602 9602 9603 9604 9604 9604 9604 9604 9604 9605 9605

1. INTRODUCTION Titanium dioxide (TiO2) has been the most intensively investigated binary transition metal oxide in the past four decades as indicated by Figure S1. Furthermore, the annual

9582 9582 9584 9585

Special Issue: 2014 Titanium Dioxide Nanomaterials Received: October 31, 2013 Published: May 22, 2014 9559

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

of Pt on the {110} facets and PbO2 particles on the {011} facets of rutile TiO2 because of the different surface energy levels of two facets.73 In addition, the interfaces in the heterostructures can also be modified by involving different facets of TiO2. Developing TiO2 crystals with specific facets is therefore highly desirable in many applications of TiO2. Although TiO2 crystals, particularly nanostructures with various shapes (i.e., nanowire, nanorod, nanobelt, nanotube, nanosphere, and nanosheet) have been realized in the past decades,58 the synthetic TiO2 crystals enclosed by clear facets, particularly high-energy facets were very rare before 2008.74 The breakthrough in synthesizing micrometer-sized high quality anatase crystals with a high percentage of reactive high-energy {001} facets was not made until 2008.75 The crystals synthesized by Yang et al. were enclosed by 47% {001} and 53% {101} facets (the % values referring to the facet area and not the number of facets in this paper). On the other hand, Vittadini et al. predicted the unusual dissociative adsorption of water on anatase (001) surface in 1998.69 In the following decade intense theoretical and experimental studies focused on the surface structure and surface chemistry of the anatase (001) surface.59,60 However, most relevant experimental studies were conducted with macroscopic anatase (001) single crystals (usually obtained by the epitaxial growth on the SrTiO3 or LaTiO3 (001) surfaces76−81) due to the lack of synthetic mesoscopic anatase crystals with a high percentage of {001} facets. The emergence of synthetic anatase crystals with a high percentage of {001} facets immediately generated enormous excitement in the scientific community especially among those working in TiO2 related fields and aroused strong interest in TiO2 with specific facets. In the past five years, different synthesis strategies have been used to produce TiO 2 polymorphs with different facets and the research topic has experienced an explosive growth. So far, the production of lowindex facets ({101}, 8 2 − 9 9 {001}, 7 1 , 7 2 , 8 8 , 9 5 , 9 6 , 1 0 0 − 2 1 9 {010},98,136,174,183,220−229 and {110}115,230) and high-index facets97,222,231−233 ({103}, {105}, {107}, {201}, {401}, {301}, and {106}) has been achieved in anatase crystals. For rutile crystals the {110},234−241 {011},242−244 {001},242,245 and {111}244,246−250 facets have been realized. Brookite, which is the most difficult phase to be synthesized among three natural phases (anatase, rutile, and brookite), has also been produced with {210}, {201}, {101}, {111}, {100}, and {001} facets.251−257 Among them, anatase crystals, especially those with {001} facets have been most intensively studied. Schematic 1 summarizes the main shapes and applications of crystals of three TiO2 polymorphs with their surfaces consisting of different facets. Many unusual properties associated with the faceted crystals have been revealed. Extensive studies on modification of electronic structure and interfacial properties of the faceted crystals have been reported. These faceted crystals have been actively used in nearly all research fields (particularly photocatalysis, DSSCs, lithium ion batteries), where crystals without specific facets were earlier used. Encouraging performance improvements have been achieved. Furthermore, the concept of tailoring crystal facets has been extended to the controllable synthesis of other metal oxides for various applications (Scheme 1). Clearly, TiO2 crystals with tailored facets have been one of the hottest research topics on various metal oxides in the past five years. A comprehensive summary of this important topic is lacking and highly desirable in order to rationally promote the

number of papers published on TiO2 has seen a continuous increase, particularly since the beginning of this century (Figure S2). This is understandable when one considers the wide range of applications of TiO2 from the conventional areas (i.e., pigment, cosmetic, toothpaste, and paint) to the later developed functional areas such as photoelectrochemical cell,1−3 dye-sensitized solar cells (DSSCs),4−11 photocatalysis,12−24 catalysis,25−31 photovoltaic cell,32−34 lithium ion batteries,35−41 sensors,42−46 electron field emission,47−51 microwave absorbing material, biomimetic growth, and biomedical treatments.52−57 Nearly all these functional applications of TiO2 fall in the scope of energy, environment, and health, which are definitely the three most important and challenging themes facing the Human race that need to be addressed in this century. Besides the apparent merits including nontoxicity, elemental abundance, good chemical stability, and easy synthesis, TiO2 has attracted strong research interest worldwide due to its physicochemical properties for realizing various functions.15,58,59 Especially, very encouraging progresses in photocatalysis and DSSCs with the involvement of TiO2 have greatly stimulated the rapid development of TiO2 crystals with controllable phase, size, shape, defect, and heteroatom.58,60−68 The interaction between molecules/ions and surface of TiO2 crystals is essential nearly in all the functional applications listed above. Photocatalysis requires the effective adsorption of reactant molecules/ions (i.e., water, O2, CO2, and organic compounds) on the surface of TiO2 photocatalysts before surface electron transfer. In DSSCs, dye molecules as sensitizers are anchored on the surface of TiO2 nanocrystals by covalent bonding. The discharge/charge processes in lithium ion batteries with TiO2 based anodes are accompanied by the transport of lithium ions across the surface of TiO2. For specific molecules/ions, their adsorption states are intrinsically determined by surface atomic structures (atomic coordination and arrangement) of TiO2. The substantial role of different surfaces in affecting the interaction can be demonstrated by comparing water adsorption, dye anchoring, and lithium insersion on anatase (001) and (101) surfaces. (1) The anatase (001) surface allows the dissociative adsorption of water, while anatase (101) surface only allows the molecular adsorption of water.69 (2) The adsorption of C101 dye molecules on anatase (001) and (101) surfaces form two anchoring geometries differing in the binding of the dye carboxylic groups to the TiO2 surfaces (bridged bidentate vs monodentate),70 and this is responsible for the remarkably different dye loading and back electron transfer dynamics observed in the corresponding DSSCs.71,72 (3) An electrochemical study on anatase single crystals with the (101) and (001) facets exposed showed that Li+ insertion is favorable on the (001) surface but is unfavorable on (101) as evidenced by a higher standard rate constant for charge transfer (10−8 vs 2 × 10−9 cm/s).41 Controlling the essential interaction between molecules or ions and TiO2 surfaces required in various applications is therefore one inherent driving force to develop TiO2 with specific facets. The development of TiO2 crystals with specific facets is also driven by the inherent necessity of constructing heterostructures of TiO2 in a controllable manner. Constructing the heterostructures is attractive in not only improving the performance of the individual materials but also introducing some new and unusual properties the individual materials do not possess. Different facets with different surface atomic structures could exhibit distinct abilities in hosting guest materials. One typical example is the selective photodeposition 9560

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

surface energies (i.e., anatase (001) surface, rutile (001) surface, and brookite (100) surface) always grow rapidly so that they usually take up a very small fraction of the surface of the final crystals or even vanish.262−264 Surface energies of different surfaces can be effectively decreased by the selective adsorption of appropriate organic molecules, inorganic ions or their mixtures as capping agents so that the growth rates along different orientations can be controlled.74,75,265−272 This consequently leads to the formation of crystals with tunable percentages of different facets as indicated by Figure 1. The

Scheme 1. Summary of Main Shapes and Applications (i.e., Lithium Ion Batteries, Photocatalytic Hydrogen Evolution, Photodegradation, and Solar Cells) of Anatase, Rutile, and Brookite TiO2 Crystals with Their Surfaces Consisting of Different Facetsa

a

Figure 1. Schematic of the effect of solvent and additive/impurity molecules or ions on the morphological control of crystal facets.260 Reprinted with permission from ref 260. Copyright 2011 Royal Society of Chemistry.

The unit cell of each phase is also given.

further development of TiO2 crystals with tailored facets and also other faceted transition metal oxides. Several previous feature articles, perspective and review paper258−261 have given preliminary summaries of the progresses made in faceted TiO2 crystals mainly with anatase phase before 2011. However, new results of faceted crystals of three main TiO2 polymorphs (anatase, rutile and brookite) in the last three years have been far greater than in earlier years as indicated by the increased number of publications (Figure S3). This review is devoted to fully reviewing the progress in faceted TiO2 crystals mainly achieved after the breakthrough in the synthesis of anatase crystals with 47% {001} facets in 2008. The review focuses on four aspects of TiO2 crystals with tailored facets, namely, synthesis, unusual properties, modification, and their applications.

selectivity and capability of the adsorption of capping agents on different surfaces is basically controlled by the density of undercoordinated Ti atoms and/or the distance between two adjacent undercoordinated Ti atoms on the surfaces. Therefore, the choice of capping agents is central in controlling the facets grown in the crystals. So far, organic capping agent takes a dominant role in controlling the shape of TiO2 crystals largely due to the diversity of organic molecules. For example, oleic acid (OA) and oleylamine (OM) with different binding strengths can act as the capping agents of anatase {001} and {101} facets, respectively.272−275 However, increasing success in controlling the shape of TiO2, particularly the formation of specific crystal facets using inorganic capping agents has been reported in recent years.75,271 One impressive example is using F− as the capping agent to obtain a large percentage of anatase {001} facets.75 In the nonhydrolytic methods, which have revolutionized the production of highly monodisperse and shape-controlled nanocrystals of metal oxides and metals in the past two decades, the use of appropriate capping agents is also an important factor.88,229 Accompanying the rapidly increased ability to control the facets of crystals using inorganic capping agents, the choice of capping agents has also experienced a change from being empirical to being theoretically predicted. The basis of the theoretical prediction is surface energy variation of a facet before and after the adsorption of capping agents. One of the early theoretical studies demonstrated that the removal of OH groups from the anatase (001) surface leads to an increase of surface energy and thus a decrease of the area of {001} facets in the truncated bipyramid with major {101} and minor {001} facets at thermodynamic equilibrium.270 Barnard et al. predicted that hydrogen-poor adsorbates and oxygenated

2. SYNTHESIS OF TIO2 CRYSTALS WITH DIFFERENT FACETS 2.1. Basic Strategies for Synthesis of Faceted TiO2

Clearly faceted TiO2 crystals can be synthesized by many synthesis methods including wet-chemistry route (hydrothermal, solvothermal and nonhydrolytic), gas oxidation route, topotactic transformation, crystallization transformation from amorphous TiO2, and epitaxial growth. Among them the hydrothermal and solvothermal methods are most widely used to tailor the exposure of crystal facets due to the versatile ability in manipulating the nucleation and growth behaviors (particularly growth rates along different orientations) of crystals. It is well established that the shape evolution of crystals during growth in a given environment is largely driven by the inherent necessity of minimizing the total surface energy. Under natural or equilibrium conditions, surfaces with high 9561

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

subsequent growth are considered to be responsible for the formation of the high-energy facets in this approach.232 Appropriate lattice match between the desired facets and the surfaces of the substrates is essential in the epitaxial growth of TiO2 crystals with specific facets. This means that the epitaxial growth process can only be used to prepare limited kinds of faceted TiO2 crystals, such as anatase (001) crystals on the SrTiO3 or LaTiO3 (001) single crystal surfaces.78,278 The topotactic transformation from faceted mother precursors with similar crystal structures to that of TiO2 polymorphs provides the possibility for the oriented transformation to form corresponding TiO2 crystals with the same facets exposed. This approach is only valid in some compound precursors (i.e., NH4TiOF3).279 In addition, an interesting generic way of increasing the percentage of anatase {001} by using external strains was predicted by Jia et al.280

surfaces (basic conditions) are favorable for the presence of {010} facets in anatase crystals,74 which has been subsequently examined by many groups.225,227,276 Yang et al. theoretically investigated the effects of 12 nonmetallic atoms (X) as adsorbates (X = H, B, C, N, O, F, Si, P, S, Cl, Br, or I) on changing the surface energies of anatase (001) and (101) surfaces (Figure 2a−d).75 As shown in Figure 2e, the

2.2. Faceted Anatase TiO2

Anatase with an indirect bandgap of around 3.2 eV is a thermodynamically metastable phase. It is the most intensely studied TiO2 polymorph among the three natural phases in catalysis, photocatalysis, and DSSCs. It is easy to synthesize anatase crystals using wet-chemistry approaches largely due to two favorable points: (a) in air anatase is the most stable phase of TiO2 below 11 nm,281 and (b) anatase is more stable compared with brookite and rutile in very basic conditions.282 Most applications of anatase are sensitive to the surface properties and numerous studies have therefore focused on controlling the shape of the anatase crystals. The equilibrium shape of an anatase crystal according to the Wulff construction and surface energies calculated in vacuum is a slightly truncated bipyramid enclosed by more than 94% {101} and fewer 6% {001} facets (The left panel in Figure 3).263 The surface Figure 2. Slab models and calculated surface energies of anatase TiO2 (001) and (101) surfaces. (a and b) Unrelaxed, clean (101) and (001) surfaces. Ti and O atoms are represented by gray and red spheres, with 6-fold Ti, 5-fold Ti, 3-fold O, and 2-fold O labeled as 6c-Ti, 5c-Ti, 3cO, and 2c-O, respectively. (c and d) Unrelaxed (101) and (001) surfaces surrounded by adsorbate X atoms. (e) Calculated energies of the (001) and (101) surfaces surrounded by X atoms. (f) Plots of the optimized value of B/A and percentage of {001} facets for anatase single crystals with various adsorbate atoms X.75 Reprinted with permission from ref 75. Copyright 2008 Nature Publishing Group.

adsorption of F reverses the relative order of surface energy and thus order of stability of (101) and (001) surfaces, compared to the clean ones. This means that a F-terminated (001) is preferable for the development of a high percentage of {001} facets (Figure 2f). This prediction has been validated by Yang et al. in their first synthesis of anatase crystals with 47% {001} facets with the assistance of F75 and subsequently by numerous studies from other groups.102,103,108,114,116,119,123,130,135,136,277 Rationally screening based on the calculated surface energy changes provides an efficient approach for choosing an appropriate capping agent for the targeted crystal facets of TiO2 and may promote the design of new capping agents for the production of desired crystal facets. High temperature gas oxidation is an attractive approach for the large-scale production of the faceted TiO2 crystals with high crystallinity. So far, this approach has produced anatase crystals with {001} and {101} facets100 and anatase crystals with highindex {105} facets.232 The synergistic effects of thermodynamic and kinetic factors that control crystal nucleation and

Figure 3. Equilibrium crystal shape of anatase TiO2 through the Wulff construction263 and the evolved other shapes.

energies of low-index facets are in the order (110) (1.09 J m−2) > (001) (0.90 J m−2) > (010) (0.53 J m−2) > (101) (0.44 J m−2).263,283 The surface energy is closely related to the density of undercoordinated Ti atoms. The (110) surface with the highest surface energy is the only low-index surface with 4-fold coordinated Ti (Ti4c) atoms. Although the surface energy of (010) is much lower than that of (001), {010} facets do not appear in the predicted shape in vacuum, and this difference is explained by considering practical surface chemistry in that the 9562

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

Cl− is a rutile mineralizer.94 A relatively low concentration of H2O2 acts as the crystallographic directing agent to oxidize Ti(III) to generate a titanium peroxo complex, [Ti(O2)3]2−, for the formation of anatase crystals. There is a wide window of 0.7−3.3 M H2O2 at a fixed concentration of 1.6 M HCl for the formation of pyramidal anatase crystals with nearly 100% {101} facets. The stronger binding of HCl on (101) than (001) results in a far lower surface energy of (101) than (001),224 which retards growth along the [101] direction and allows rapid growth along the [001] direction. However, a higher concentration of H2O2 acting as the capping agent of {001} facets causes the increased percentage of {001} facets produced. The crystals obtained in the two cases are micrometer-sized.93,285 Anatase crystals with the shape of a quasi-octahedron with {101} facets can also be prepared using water-insoluble titanium precursors (i.e., titanate, TiB2, and electrospun nanofibers of amorphous TiO2).82,94,101 One of the remarkable features using this kind of precursor is the controllable and gradual release of the titanium complex in the reaction system, which results in a low nucleation rate and is thus favorable for controllable crystal growth. The hydrothermal acidic hydrolysis of TiB2 crystals, a conventional but important ceramic, in a SO42− mediated solution leads to the formation of micrometersized anatase crystals with a slightly truncated bipyramidal shape.94 The gradual release of the titanium complex is evidenced by the presence of TiB2 crystals in the first 5 h of reaction. Furthermore, the freshly released Ti species could be trivalent judging from the pale purple reaction solution, which together with the gradual release of the Ti precursor is favorable for the growth of faceted crystals. By hydrothermally treating the nanofibers consisting of amorphous TiO2 and poly(vinylpyrrolidone) (PVP) at low pH values, octahedral anatase nanocrystals with {101} facets were obtained.101 Layered titanate nanostructures, prepared by treating TiO2 crystals in concentrated alkaline hydroxide solutions, were also used to synthesize anatase crystals with {101} facets. Amano et al. synthesized octahedral anatase crystals with a c-axis length of 16 oC min−1) and long enough nanotubes (at least 30 μm for the full transformation) to guarantee some NH4F in the nanotubes, besides a sufficiently high temperature (>500 °C). The solid state transformation of TiOF2 cubes or plates to anatase crystals whose surface area is mainly {001} facets by heating in different atmospheres were reported.163,293 A gas phase oxidation process using Ar gas bubbled TiCl4 vapor as the Ti precursor and excessive O2 flow was developed to prepare 50−250 nm decahedral anatase single crystals with {001} and {101} facets at 1300 °C (Figure 9).100 It is considered that uniform and rapid heating at a high temperature enables the hom*ogeneous nucleation and subsequent growth of well faceted crystals with few defects, and the low concentration of TiCl4 and the narrow heating zone prevent the formation of large particles and polycrystalline aggregates with grain boundaries. This result represents the

shapes. One of the key issues determining the selective adsorption of the capping agent molecules on the targeted (001) surface is the match of the distance between two active functional groups of the capping agent molecule(s) with that between two consecutive unsaturated O2c atoms in the (001) surface. For example, it was theoretically revealed that the distance between ammine-H and hydroxyl-H in DEA (0.46 nm) is close to the distance between two consecutive 2-fold coordinated O (O2c) atoms of the anatase (001) surface (0.53 nm), but much larger than the distance between two consecutive O2c of the anatase (101) surface (0.28 nm).211 DEA can therefore stabilize anatase (001) facets through Hbonding. Roy et al. synthesized anatase nanocrystals with 35− 42 area% {001} facets using DEA as the capping agent.211 In addition, Chen et al. synthesized hierarchical anatase microspheres assembled from the high-surface-area nanosheets with nearly 100% {001} facets (Figure 8) by the hydrothermal

Figure 8. (A) FESEM image and (B) TEM image of as-prepared anatase TiO2 nanosheet hierarchical spheres.110 Reprinted with permission from ref 110. Copyright 2010 American Chemical Society.

reaction of titanium isopropoxide in isopropyl alcohol containing diethylenetriamine as the capping agent and subsequent calcination to remove the capping agent.110 The formation of nanosheets whose surface is almost all {001} facets is very sensitive to the chain length of the amine used. Chen et al. further synthesized large 2D nanosheets from nanomosaic building blocks of anatase nanosheets with (001) facets by hydrothermally treating the mixture of titanium isopropoxide, HF and PVP.129 The PVP molecules adsorbed on the (001) facets were considered to serve as the linker that brings building blocks together on the ab-plane, at the same time prevents them from stacking along the c-axis. Using appropriate flat 2D templates to regulate the crystal growth is a unique method of controlling the shape of the crystals. Wang et al. showed the feasibility of using DVMT as the stabilizer of {001} facets as a result of the selective interaction of the DVMT layers with {001} facets.183 The resultant anatase nanorods and nanocuboids with a tunable percentage of {001} facets were prepared depending on the number of DVMT layers in the reaction system. Vermiculite is a natural clay mineral, composed of an MgO2(OH)4 octahedral sheet symmetrically coupled to an octahedral sheet of silica, and has a net negative charge that originates from isomorphic substitution of Al3+ in the Si4+ sites of the tetrahedral sheets which are balanced by some interlayer cations. The external surfaces of the DVMT layers were terminated by Si−OH groups (where the O is an apical oxygen atom Oa of the SiO4 tetrahedral sheet). The origin of the selective interaction of the DVMT layers with {001} facets is due to the equality of the distance between the bridging Oa atoms within the hexagonal arrangement of DVMT to that of the bridging O2c atoms in the 9566

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

al. reported two modes of the epitaxial growth of square anatase islands with a top (001) facet on a SrTiO3 (001) surface prepared by extended annealing in UHV. The modes depended on the annealing temperature.78 At an annealing temperature of 1000−1030 °C, above a critical edge length of 1.3 μm the islands evolve into rectangular shapes with increased length and diminished width. This evolution follows the Tersoff-Tromp model, where the continuous attachment and detachment of titanium and oxygen atomic precursors from the islands is feasible due to the temperature being high enough. At an annealing temperature of 930−1000 °C, above a critical edge of 0.53 μm the islands relieve strain by the formation of trenches in the middle of each side of the square, finally evolving into crosses. In both modes, the driving force for the shape changes is to relieve the increasing strain energy in the square islands. Oropeza et al. realized the growth of epitaxial anatase nano islands by high temperature annealing of ultrathin anatase TiO2 (001) thin films deposited on SrTiO3 (001) substrates by a dip coating method.278 Relatively uniform islands with lateral size of over 150 nm obtained at 1000 °C (Figure 10a,b) have surface terminations with top (001) and side {103} facets (Figure 10c), which are different from the dominant side {101} facets of unsupported nanocrystals.

Figure 9. SEM image of TiO2 particles prepared by gas-phase reaction of TiCl4 with oxygen.100 Reprinted with permission from ref 100. Copyright 2009 American Chemical Society.

capping agent-free synthesis of nanosized anatase crystals with high energy facets. This method was subsequently modified by Jiang et al. to prepare anatase crystals with high-index facets.232 Various Ti precursors have been used in solution methods to synthesize anatase crystals with dominant {001} facets. These precursors can be simply divided into three classes: (a) easily hydrolyzed precursors when exposed to water or moisture such as titanium isopropoxide,292 TiCl4,212 and tetrabutyl titanate,102 (b) water-soluble precursors such as TiF 4 , 75 TiCl 3 , 93 TiOSO 4,136 and Ti(SO4 ) 2,113 and (c) water insoluble precursors such as Ti,137,148 TiB2,104,219 TiN,105 TiS2,112 TiSi2,231 TiC,158 TiCN,201 titanate,98 and even TiO2.176 One of the merits using the water-soluble or insoluble precursors is the insensitivity of the reactivity to the surrounding environment thus making it easy to manipulate the synthesis process. It is noted that crystals derived from soluble or insoluble precursors usually have much large particle sizes than those from easily hydrolyzed precursors. This could be rationalized by the low density of nuclei and their progressive growth during the reaction in the former. In addition, with TiB2, TiN, TiS2, TiSi2 and TiCN as the precursors, anion doping can be realized simultaneously with the facet control as a result of some residual anion elements in the resultant crystals. This will be introduced in section 4.1. Epitaxial growth of a crystalline phase on the surfaces of appropriate crystalline substrates is an important method of preparing high quality crystals with specific facets. Compared to rutile TiO2, anatase crystals are suitable for epitaxial growth on SiTiO3 (001) and LaTiO3 (001) substrates because of the small lattice mismatch between the anatase (001) surface and the substrate surfaces. The lattice mismatch between tetragonal anatase (001) (3.785 Å × 3.785 Å) and cubic perovskite SiTiO3 (001) (3.905 Å × 3.905 Å) is 3.1%, which is larger than the 0.2% between cubic perovskite LaTiO3 (001) (3.793 Å × 3.793 Å) and anatase (001). This means that LaTiO3 is a better substrate for the epitaxial growth of anatase crystals with (001) facets, as indicated by the higher crystallinity of anatase crystals on LaTiO3.79 In addition, the fact of the larger bandgap (∼5 eV) of LaTiO3 substrates than the supported anatase films makes it easy to study the optical properties of anatase films without disturbance from the substrates.80 Wen et al. gave a good summary of the epitaxial growth of anatase (001) films or islands on both these substrates,258 mainly conducted before 2009. Here we focus on some progresses after 2009. Marshall et

Figure 10. (a) 5 μm × 5 μm AFM image of TiO2 sample obtained after two dip-hydrolysis cycles calcined at the temperature of 1000 °C. (b) Expanded 2 μm × 2 μm AFM image of two-dip sample annealed at 1000 °C. (c) Cross-sectional TEM image of an anatase island viewed down the [010] crystal direction of the substrate (calcined at 1000 °C).278 Reprinted with permission from ref 278. Copyright 2013 American Chemical Society.

2.2.3. Crystals with {010} Facets. {010} facets with a medium surface energy of 0.53 J m−2 do not exist in the equilibrium shape of the anatase crystal according to the Wulff construction,263 where surfaces are in vacuum. However, it is theoretically predicted that {010} facets dominate the crystal shape when the effect of basic solution chemistry on the shape is considered.74 Following this prediction, anatase crystals with {010} facets have been obtained by treating titanate nanowires/ nanotubes/nanosheets or particles as the Ti precursor in an appropriate basic aqueous solution under hydrothermal conditions.97,98,225,227,276,299 The earliest report by Li et al. 9567

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

subsequent release of Na+ from titanate further increases the pH value. The question then arises as to whether there is coadsorption of OH− and alkali metal ions on the surfaces. Based on the fact that in the suspension of sodium titanate, the additional Li+ ions lead to the formation of bipyramids with {101} facet but nanorods with dominant {010} facets are formed without Li+, one can reasonably infer the presence of such synergy. Submicrometer particulate titanates (Cs0.68Ti1.83O4 or H0.68Ti1.83O4) were also suitable precursors for the synthesis of anatase nanorods with a predominance of {010} facets.227 Hydrothermally treating Cs0.68Ti1.83O4 in water or H0.68Ti1.83O4 in a solution of Cs2CO3 as the in situ source of OH− leads to the formation of micrometer or nanosize rods with {010} facets. The much larger rod size in the case of Cs0.68Ti1.83O4 is related to the role of Cs+ ions in the interlayer spaces in controlling the dissolution rates of layers and thus the release of Ti precursor, which is favorable for a low nucleation density and continuous growth of the crystals. The merit of using particulate titanates as the precursor, prepared by solid state reaction, is that the process is easy to scale-up. It is noted that products using titanate as the precursor always appear as rods. The possible reason is that the collapse of a layered titanate structure under hydrothermal conditions directly produces fragments of Ti(OH)4 instead of Ti complexes. The rearrangement of these fragments and subsequent nucleation−growth of TiO2 crystals with the preferential adsorption of adsorbates on specific facets ({010}) favors anisotropic growth along the c-axis and retards growth along the a/b-axes of the anatase crystals. F− ions are proven to be very effective in stabilizing high energy anatase {001} facets by saturating surface unsaturated coordinated Ti5c atoms. Similar to the {001} facet, all titanium atoms on the {010} surface are Ti5c. In principle, it is feasible to use fluorine species as the capping agent to stabilize {010} facets. By carefully controlling the synthesis parameters (concentration of Ti precursor and HF, reaction time) in the reaction system of TiOSO4 and HF, Pan et al. synthesized micrometer-sized anatase crystals with 53% {010}, 14% {001}, and 33% {101} facets (Figure 11b).136 In this method, the short duration (2 h) is one key to obtaining a large percentage area of {010} facets. Longer times usually lead to a decrease and even a loss of {010} facets, accompanying by an increase of {101} facets, which is also confirmed by the result of Wu et al.230 This suggests that the fluorine-stabilized {010} facets are still less stable than fluorine-stabilized {101} facets with fluorine as the sole capping agent. Using dual capping agents provides an effective way to stabilize anatase crystals with dominant {010} facets. Several studies showed that, regardless of their sources, the coexistence of F− and Cl− in the reaction system is favorable for the synthesis of anatase crystals with {010} facets.130,174,224,301 One remarkable example is that the replacement of the TiF4 precursor with TiCl3 in a reaction solution of HF changes the resultant product from anatase crystals with {001}/{101} facets to anatase crystals with additional {010} facets.130 Wu et al. theoretically revealed the origin of the synergistic effects of the mixture of HF and HCl as the capping agents.224 Their key findings are that (a) the adsorption of HF on each surface is always stronger than that of HCl; (b) HCl binds to the (101) surface more strongly than to (001), while HF adsorption gives the reverse result; (c) (010) is the plane to which HF or HCl binds the most weakly. All these results indicate that HF tends

used Na-titanate nanotubes in pure water to produce anatase nanorods with dominant {100} facets in the center part and minor {101} facets at two ends (Figure 11a).225 The initial pH

Figure 11. (a) High SEM images of tetragonal faceted nanorods.225 Reprinted with permission from ref 225. Copyright 2010 Royal Society of Chemistry. (b) SEM image of anatase TiO2 crystals.136 Reprinted with permission from ref 136. Copyright 2011 Wiley-VCH. (c) Anatase TiO2 nanocuboids obtained from the TTIP/H2O/[bmim][BF4]/HAc system with molar ratios at 1:1.66:2:210.292 Reprinted with permission from ref 292. Copyright 2011 Wiley-VCH.

value in the suspension was 10.1 and gradually increased to 11.8 in the final stage with the release of Na+ ions from the Natitanate and the formation of anatase nanorods. The gradually increased concentration of hydroxyl groups instead of a fixed concentration is believed to be crucial for the formation of faceted nanorods. Hydroxyl groups would preferentially adsorb on {010} surfaces to lower the surface energy of {010} and thus retard growth along the a- and b-axes. Several other groups further conducted the synthesis of anatase nanorods with {010} facets using sodium titanate or hydrogen titanate nanostructures as the precursor.97,276,300 Although the products are quite similar, there is a discrepancy in terms of the stabilizer of {010} facets. The selective adsorption of Cl− ions on the {010} facets was proposed to be responsible for stabilizing {010} in the system of hydrogen titanate nanosheets and NaCl by comparing the morphology of products at different concentration of NaCl.300 However, this mechanism was challenged by the results of Yang et al. with sodium titanate as the precursor in the presence of different alkaline metal ion salts.97 They proposed that the adsorption of alkali metal ions (Li+, Na+, and K+) with different charge densities instead of Cl− anions on the {001} and {100} planes affects the shape of anatase crystals (production of {101}, {103}, and {010} facets). This is supported by the fact that the replacement of alkali chlorides with corresponding alkali bromides or iodides does not change the crystal shape. The initial pH value in the reaction system is around 12 and the 9568

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

binding of OA and OM molecules on {001} and {101} facets, respectively, restricts the growth rates along the corresponding directions, which is the basis of controlling the shape of nanocrystals by changing the ratio of OA:OM. It is remarkable that the amount of TB precursor and reaction temperature play an equally important role in forming a high percentage area of {010} facets in the nanocrystals by elongating the crystals along the [001] direction at fixed ratios of OA:OM. The elongation of nanocrystals with increasing amount of TB is considered to arise from the anisotropic crystal growth at a high monomer concentration. Very similar to the fluorine species as the capping agent, a low temperature is favorable for the formation of nanobars with lateral {010} facets. Anatase crystals with {010} facets were also obtained by some unusual synthesis methods. Wang et al. synthesized anatase nanocuboids or nanorods dominated with {010} facets using tetramethylammonium hydroxide as the stabilizer of {010} facets.183 Porous anatase microspheres composed of {010}-faceted nanobelts were synthesized by simple thermal treatment of a titanium glycerolate precursor.228 Roy et al. showed a dependence on the time of the reaction in the production of different facets in anatase crystals when keeping all other parameters (reaction temperature and initial molar concentrations of precursor, diethanolamine, and tetramethylammonium hydroxide) fixed.211 Longer reaction times led to a decreased percentage area of {010} facets but an increase in the area of {101} facets, which is consistent with the trend observed using F− ions as the capping agent.230 2.2.4. Crystals with {110} Facets. Among all the lowindex facets, anatase {110} facets are most rarely found in synthetic crystals largely due to their extremely high surface energy of 1.09 J m−2, which is even higher than that (0.90 Jm2−) of {001} facets. Liu et al. reported the first example of anatase crystals with major {101} and {001} facets and minor {110} facets by treating Ti powder in a mixture of HF and H2O2 under hydrothermal conditions (Figure 13a).115 The

to adsorb on (001) and (010) surfaces, while HCl tends to adsorb on the (101) surface. Further Wulff construction plots confirm that HF-covered {010}/{001} and HCl-covered {101} does generate a cubic shape. In contrast, crystals covered with only HF or HCl give bipyramidal shapes with just different ratios of {001} to {101}. The selective adsorption of HF and HCl on different surfaces is crucial for the stabilization of both {010} and {001} facets. In addition, the most weak binding of HF on the (010) surface among the three surfaces might explain the instability of any initially formed {010} facets for the prolonged hydrothermal reaction time, and the absence of {010} facets at the higher hydrothermal treating temperature.230,301 The synthesis temperature dependent stability of {010} facets can be strongly indicated by the results of Lai et al.174 In the reaction system of TiCl4, NaBF4/NaF and HCl under solvothermal conditions, temperatures below 140 °C always lead to the creation of {010} facets in various shapes of anatase crystals without changing other parameters, while temperatures above 140 °C result in the absence of {010} facets. However, the underlying mechanism proposed in this case is that the surface lattice F (in Ti−F−Ti bonds) generated in the reaction is considered as the stabilizer of the high surface energy {100} and {001} facets at low temperature. Dual capping agents can also be generated by one compound. Uniform anatase nanocuboids with {100} and {001} facets with controllable aspect ratios (Figure 11c) were synthesized through the hydrolysis of titanium tetraisopropoxide using acetic acid as the solvent and the ionic liquid [bmim][BF4] as the capping agent.292 It is demonstrated that the F− and [bmim]+ ions released from [bmim][BF4] can preferentially stabilize the {001} and {100} facets, respectively. Fluorine-free capping agents were also used to stabilize the {010} facets of anatase crystals. A typical example by Dinh et al. is using OA and OM with different binding strengths as the capping agents of {001} and {101} facets, respectively, to control the growth of anatase nanocrystals in the solvothermal method with TB as the Ti precursor and water vapor as the hydrothermal agent.287 Various shapes (rhombic, truncated rhombic, spherical, dog-bone, truncated and elongated rhombic, and bar) of the anatase nanocrystals can be obtained by controlling the ratio of TB:OA:OM together with the reaction temperature, as summarized in Figure 12. The selective

Figure 13. (a) Small rhombus {110} facets (indicated by red circle) of the anatase single crystals. Scale bar: 1 μm.115 Reprinted with permission from ref 115. Copyright 2010 Royal Society of Chemistry. (b) SEM images of TiO2 single crystals prepared in the presence of different concentrations (1.5 mL) of H2O2 at 150 °C for 10 h.230 Reprinted with permission from ref 230. Copyright 2012 Elsevier.

presence of H2O2 in the reaction system is considered to be crucial for the formation of {110} facets by further retarding the hydrolysis of Ti precursor through the peroxotitanium complex formed. By controlling the amount of TiOSO4 precursor in HF solution, minor {110} facets can be formed in anatase crystals that have major {001}, {010}, and {101} facets.221 Wu et al. systematically studied the influence of various synthesis parameters (amounts of HF and H2O2, reaction temperature and time) on the formation of {110} facets in a reaction system

Figure 12. Schematic illustration of the overall formation and shape evolution of TiO2 nanocrystals.287 Reprinted with permission from ref 287. Copyright 2009 American Chemical Society. 9569

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

part. These findings could be instructive to understand the different stages of crystal growth from thermodynamically unstable high-index surfaces to stable low-index surfaces and pave the way for the fabrication of some other unusual facets as well. In addition, the surface energy of (105) is estimated to be ca. 0.84 J m−2, which is lower than that of anatase (001) (ca. 1.0 J m−2) but is much higher than that of anatase (101) (ca. 0.5 Jm2−). The use of titanate as the precursor in hydrothermal methods is favorable for the synthesis of anatase crystals with high-index facets. Polyhedral anatase crystals with 16 {101} and {103} facets can be obtained in the reaction system of potassium titanate nanowires and hexamethylenetetramine.168 By introducing the additional capping agent PVP in the reaction system, bipyramidal anatase crystals with {102} facets can be obtained. The photodegradation efficiency of different facets follows the order {001} > {102} ≈ {103} > {101}, as estimated by the degeneration of methylene blue. Yang et al. synthesized anatase crystals with tunable shapes and facets by using sodium titanate as the precursor in the presence of different alkali metal ion salts. {101} faceted anatase bipyramids were synthesized in a LiCl solution.97 Spindle-like anatase crystals with main lateral high-index {301} facets were obtained in a NaCl solution. Needle-like anatase crystals, whose two ends are terminated with {301} facets while the middle part has {100} facets, were obtained in a KCl solution. The alkali metal ion-dependent facet production in anatase crystals is explained by the different restraints of alkali metal ion adsorption on the anisotropic stacking of the {100} and {001} planes. The charge density of alkali metal ions follows the order of Li+ > Na+ > K+. It is considered that alkali metals with a large charge density were attracted to both {100} and {001} planes, restraining the growth along the related directions. This leads to the growth of small particles with a low aspect ratio, whereas cations with a low charge density tended to be absorbed only on the {100} planes, producing particles with a high aspect ratio. However, crystals dominated by {101} facets show a superior photocatalytic activity in decomposing methylene blue to those with {301} facets. The rarely observed asymmetric bipyramidal anatase TiO2 nanocrystals (Figure 15), whose short and long ends have

with Ti powder as the precursor.230 All investigated parameters exhibited an obvious influence on the percentage area of {110} facets. Each parameter provides a narrow window for growing {110} facets. {110} seems to act as an intermediate plane during crystal growth. Although surface termination with fluorine can substantially reduce the surface energy of {110} facets, fluorine terminated {001}/{101}/{010} facets seem to be more stable so that the fraction of {110} in the anatase crystals is very low. The highest claimed percentage area of {110} facets is around 11% (Figure 13b). Developing other capping agents is highly necessary in order to further increase the {110} facet area. 2.2.5. Crystals with High-Index Facets. High-index facets with high densities of atomic steps, ledges, kinks and dangling bonds usually possess high chemical reactivity. They are, however, always absent from the crystals because of their high surface energies. For example, anatase (103) has a surface energy of 0.93 and 0.83 J m−2 for the faceted and smooth configurations, respectively, which is comparable to that (0.90 J m−2) of anatase (001).283 Crystals are typically terminated with low-index facets. The production of high-index facets has been actively pursued in noble metal nanocrystals in order to improve their catalytic or electrocatalytic reactivity. For example, tetrahexahedral Pt nanocrystals enclosed by 24 highindex facets such as {730}, {210}, and {520} exhibit an increased (up to 400%) catalytic activity for equivalent Pt surface areas for the electro-oxidation of small molecule organic fuels.302 So far only limited reports have focused on the synthesis of anatase TiO2 crystals with high-index facets. Impressive examples include anatase crystals with a predominance of high-index {105},232 {201},231 {103},168 {301},97 and {401}/{201}233 facets. Anatase crystals with {105} facets (Figure 14) were prepared by a modified high-temperature (1000 °C) gas-phase oxidation

Figure 14. SEM images of the as-obtained anatase TiO2 crystals dominated by high-index {105} facets.232 Reprinted with permission from ref 232. Copyright 2011 Wiley-VCH.

method with oxygen bubbled TiCl4 as the Ti source. The key modification is to use an additional thin spiral tube as the reactant feeder. Without highlighting the spiral tube in the synthesis method of Amano et al.,100 highly truncated bipyramid crystal with {001} and {101} facets were obtained. By monitoring the morphologies of the early products, a twostage growth mechanism was proposed for the growth of anatase crystals with high-index facets. In the first stage, a highly truncated morphology was formed. In the second stage, {001} facets acted as substrates to induce the gas-phase epitaxial growth of high index facets as a result of the good lattice matching. Furthermore, the crystallographic structure is well retained between the mother crystals and the newly formed

Figure 15. (A) TEM image of a bundle consisting of several TiO2 needles. (B) TEM image of an individual TiO2 needle (inset: corresponding SEAD pattern).233 Reprinted with permission from ref 233. Copyright 2011 Royal Society of Chemistry.

nearly 100% {201} and {401} facets, respectively, were synthesized by a simple solvothermal approach.233 The reaction system is titanium butoxide as the Ti precursor in a mixed solvent of acetic acid and N,N-dimethylformamide. The proposed growth process of the asymmetric crystals has three stages. The formation of symmetric rhombic crystals bound 9570

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

entirely by {401} facets occurs in the first stage. In the second stage, these crystals tend to assemble on the high energy {401} facets into dandelion-like TiO2 to minimize the free energy. The outer unstable {401} facets exposed to the reaction solution continue to grow into lower-index crystal facets until forming {201} facets stabilized probably by acetic ions. Besides the high-index facet dominated anatase crystals described above, some other high-index facets (i.e., {116} and {106}) appear in crystals with a high percentage area of lowindex facets.222,303,304

{110} facets with the lowest surface energy dominate the shape of crystal. The remaining area of the crystal is almost entirely composed of {011} and {100} facets. {001} facets are present but almost invisible in the equilibrium shape due to their having highest surface energy and near instability with respect to the formation of {011} facets. Some possible typical shapes of rutile crystals evolved from the equilibrium shape are given in the right panel of Figure 16. Most of them have been realized in synthetic rutile crystals. Figure S5 compares the atomic structures of some low-index facets. It is noted that similar to anatase (001) and (101), rutile (001) has 100% unsaturated Ti5c at the surface and rutile (110) has 50% unsaturated Ti5c and 50% saturated Ti6c at the surface. Rutile crystals can be prepared by many synthesis methods such as wet-chemistry, chemical vapor deposition, and phase transformation from metastable TiO2 polymorphs (anatase, brookite, or other artificial phases). Because of the dependence of the phase stability of TiO2 polymorphs on pH value (at small sizes, rutile is more stable than anatase in very acidic solutions and anatase is more stable than rutile/brookite in very basic solutions),282 an acidic environment seems to be the favorable condition for the nucleation and growth of rutile crystals. The dependence of phase stability on crystal size (anatase is most stable below 11 nm and rutile above 35 nm in air)281 makes the nucleation and growth of rutile crystals more challenging in acidic solutions if no suitable Ti complex favorable for rutile nucleation is formed. It is well documented that the presence of Cl− ions as a mineralizer in the acidic synthesis system is favorable for rutile formation, particularly at a high concentration of Cl− ions.309 In contrast, SO42− ions usually act as a mineralizer for the anatase phase. The basis of using mineralizer-modulating phase is that mineralizers with different coordination ability and spatial steric effects can affect the linkage of 6-fold coordinated monomers in different bonding modes. Many previous studies on the synthesis of rutile TiO2 crystals overwhelmingly used chlorine-containing Ti precursors (TiCl4, TiCl3, and TiOCl2) in wet-chemistry methods.239,250,286,310−313 Otherwise, concentrated HCl or chloride solutions are usually necessary if chlorine-free Ti precursors were used.94,245,250,314 For synthetic faceted rutile crystals, nanorods or nanowires with top {111}/{001} facets and lateral {110} facets are frequently produced in most production methods without the use of specific capping agents. This can be understood by the fact that the growth along the top {111} or {001} facets with a high surface energy is much more rapidly than the growth along {110} facets with a low surface energy. It is also suggested that Cl− ions tends to adsorb on the {110} facets of rutile, which further retards growth along the [110] direction.315 These nanorods always tend to aggregate into spheres or arrays by sharing the lateral {110} facets in order to further lower its total free energy. One of the early studies by Kakiuchi et al. showed the dependence of the degree of perfection of facets on rutile nanorods grown on substrates on hydrothermal temperature in the system TiCl3 and NaCl.286 At a low temperature of 80 °C, no clear facet was formed in needle-like nanorods. The sharp nanorods with clear lateral {110} and top {111} facets were grown at 200 °C. High grow temperature leads to a low density of defects in the crystals with high crystallinity, which is favorable for the formation of well-developed crystal facets. Also with the system TiCl3+NaCl under hydrothermal conditions, Sun et al. synthesized rutile microspheres constructed of well-crystallized faceted nanorods with 100%

2.3. Faceted Rutile TiO2

Rutile TiO2 with an indirect bandgap of around 3.0 eV is the thermodynamically most stable phase among the three natural polymorphs of TiO2. It has a wide range of applications due to its unusual properties (i.e., high optical stability, high chemical stability, high refractive index, high dielectric constant, excellent scattering efficiency). In comparison to the small dielectric constant of 2.48 for anatase, rutile has a very large dielectric constant (180 along the c-axis and 90 along the a-axis).305−307 Although it is typically considered to be less active in photocatalytic reactions (particularly degradation of organic compounds) than anatase, increasing results show the unique behavior of rutile in realizing overall water splitting. For example, it has been demonstrated that direct water splitting into hydrogen and oxygen can occur on Pt-loaded rutile TiO2 but fails on Pt loaded anatase TiO 2 under bandgap irradiation.308 The fact that the intrinsic surface structure of rutile is less sensitive to the photoreduction of O2 is believed to play a key role in promoting overall water splitting on rutile. The potential promising performance of rutile crystals in different applications is closely related to the surface or crystallographic orientation, which requires intense studies in controlling the shape of crystals. According to the Wulff construction with computed surface energies of the facets, Ramamoorthy et al. predicted that the shape of rutile under equilibrium conditions is a macroscopic crystal constructed with {110}, {100}, {011}, and {001} facets.264 (See the left panel of Figure 16). Their surface energies are 15.6, 19.6, 24.4, and 28.9 meV au−2, respectively.

Figure 16. Left panel: Equilibrium crystal shape of rutile TiO2 through the Wulff construction264 and (right panel) the evolved other shapes of rutile TiO2. 9571

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

{111} facets on the sphere’s surface.250 A similar hydrothermal temperature dependent growth of rutile nanowires with lateral {110} facets and top {001} facets, which were assembled into microspheres with {001} facets on the surface of the microsphere, was observed in the reaction system TiCl4+H2O.311 Completely different from the general hydrothermal methods, Bai et al. introduced Resorcinol−Formaldehyde (RF) method to prepare rutile spheres consisting of nanorods with lateral {110} and top {001} facets.239 The preparation method involves the formation of Ti-RF spheres by mixing an aqueous TiCl3 solution, resorcinol and formaldehyde solution, and subsequent calcination of the dried Ti-RF spheres under atmospheric conditions. Carbon dopants were introduced in the resultant rutile nanorods due to a small carbon residue after the calcination. Some water-insoluble Ti precursors such as TiN and Ti powder have also been used to prepare rutile nanorods or nanowires.247,249,314 Liu et al. synthesized a free-standing, large area, oriented single-crystal rutile nanowire arrays with a controlled length in the range 10−80 μm by treating Ti powder in a mixed solution of HCl+H2O2 under hydrothermal conditions.314 The nanowires as the building blocks preferentially grow along the [001] direction. By hydrothermally treating TiN powder in the aqueous solution of HCl and H2O2, rutile TiO2 films consisting of nanorod arrays with welldeveloped top {111} facets and lateral {110} facets on fluorinedoped tin oxide substrates were prepared.247,249 The surface is 100% pyramid-shaped {111} faces. In this case, a film-based photoanode shows excellent visible light activity as a result of the formed bulk Ti3+. In both cases, the growth mechanism follows typical dissolution and recrystallization processes. Compared to the conventional titanium precursors, which are easy to hydrolyze when exposed to water, TiN and Ti as precursors have the remarkable feature of the slow release rate of dissolved Ti complexes with low oxidation states, which provides a flexible platform for both controlling the growth process and changing the composition of the crystals. Although it is easy to synthesize rutile nanorods with a preferred growth orientation along [001] in the form of the assembled films or microspheres using wet-chemistry methods, the synthesis of the clearly faceted rutile crystals without assembly remains challenging. Murakami et al. developed a two-step synthesis method to prepare well-dispersed rutile nanorods with lateral {110} and top {111}/{001} facets.246 In the first step, the hydrolysis of TiCl3 in an aqueous solution containing NaCl/NaClO4 at room temperature for 1 week led to the formation of aggregations of needle-like rutile TiO2. The use of ClO4− ions induced a radial aggregation of narrower and longer needle-like TiO2, whereas the sample prepared using Cl− ions exhibited a random aggregation of wider and shorter needle-like TiO2. In the second step, a suspension of needlelike rutile from the first step with/without additional NaClO4 was treated hydrothermally to prepare rutile nanorods with tunable aspect ratios. The nanorod length increases as the length of the starting needle-like TiO2 increases. A large percentage of high energy facets {001}/{111} are usually formed in the rutile microspheres or films assembled from the nanorods with top {001}/{111} facets (Figure 17a,b).239,250 The {111}/{001} facets always take a low surface fraction in most individual faceted rutile crystals due to the high aspect ratio. Lai et al. reported the controllable synthesis of rutile crystals with tunable percentages of {111} facets with NaF as a capping agent (Figure 18).316 Surface terminated Ti−

Figure 17. (a) High resolution FESEM image of the TiO 2 nanorods.239 Reprinted with permission from ref 239. Copyright 2012 Royal Society of Chemistry. (b) Magnified spherical surface structure (of rutile TiO2).250 Reprinted with permission from ref 250. Copyright 2011 Royal Society of Chemistry. (c) TEM image of an individual crystal scratched from the substrate projected along the [001] zone axis with the related SAED pattern taken from the whole particle (inset).234 Reprinted with permission from ref 234. Copyright 2009 American Chemical Society. (d) Scanning electron micrographs reveal a broad and flat nanostructure with taper and tip angles that are coplanar and perpendicular to the large facet, respectively.242 Reprinted with permission from ref 242. Copyright 2010 IOP Publishing Ltd.

Figure 18. FESEM images of TiO2 synthesized at 220 °C for 12 h in 3.5 M HCl solution with different concentrations of NaF: (a) 0.015 M, (b) 0.030 M, (c) 0.0375 M and (d) 0.060 M. The insets of panels a−d are the corresponding high resolution FESEM images.316 Reprinted with permission from ref 316. Copyright 2013 Royal Society of Chemistry.

F bonds are considered to effectively lower the surface energy of {111} facets. With the decrease of molar ratio of titanium precursor to NaF from 1:0.5 to 1:3 in the reaction system, the percentage of {111} facets in the resultant rutile crystals 9572

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

increases from 25% to 98.5%. This is the first evidence that surface Ti−F bonds can also stabilize high-energy facets in the rutile phase. Many studies have suggested that the presence of F species is favorable for both forming the anatase phase and stabilizing high energy facets of anatase such as {001} and {010} due to the strong bonding between Ti and F. However, the Ti−F bonds may inhibit the nucleation and growth of rutile TiO2. The key in this synthesis method is the high concentration (3.5 M) of HCl in the reaction system so that the nucleation of rutile is promoted by Cl−. This result indicates that F species might be used to stabilize high energy facets of other TiO2 polymorphs. Several strategies have been developed to prepare rutile crystals with a high percentage of {001} facets.234,242,245 Different from the conventional preferred growth orientation of the [001] direction, a unique preferred growth orientation along the [110] direction occurred in the cross-medal-like Ta doped rutile TiO2 arrays grown in situ on a Ta substrate. These are obtained by treating tetrabutyl titanate in a concentrated HCl solution (6 M) in the presence of a Ta foil under hydrothermal conditions.234 This unique growth behavior is controlled by two factors. One is the crystalline lattice microstrain from the distortion of the Ti(Ta)O6 octahedra along the c-axis produced by the replacement of Ti with Ta, which retards the growth along this direction. The other is the closeness of the lattice spacing of {200} and {110} in rutile to those of {110} and {100} in cubic tantalum, which is favorable for the [110] oriented and vertical epitaxial growth of rutile from the {100} plane of tantalum foil. The surface of rutile crystals in this case is formed by major high energy {001} and minor low energy {110} facets (Figure 17c). Another strategy for synthesizing rutile crystals with a large percentage of {001} facets is to use surface amorphous MoO3 to stabilize the {001} surface. The hydrothermal reaction system used includes TiF4 and (NH4)6Mo7O24.4H2O in an acidic solution containing HCl and HNO3.245 The resultant single crystalline nanosheets with dominant {001} facets aggregate randomly to form “desert rose”-like hierarchical microspheres. The uniform dispersion of amorphous MoO3 on the high-energy {001} facets of rutile TiO2 provides sufficient stability to preserve these surfaces. The nature of the interaction between MoO3 and TiO2 remains unclear. In addition, the morphology of rutile in this case is sensitive to the amounts of HNO3 and TiF4. The merit of this strategy is in situ construction of hybrid structures with potential unusual properties. The challenge is, however, how to remove surface MoO3 to get clean {001} rutile nanosheets. One-dimensional rutile nanostructures with various facets have also been prepared by chemical vapor decomposition (CVD).242,317,318 Sosnowchik et al. prepared two types of sword-shaped rutile TiO2 nanostructures.242 The first grows along the [100] direction, enclosed by two {110} facets, resulting in a broad, more reactive {001} surface, while the second grows along the [101], enclosed by four {110} facets, resulting in the exposure of {101} facets (Figure 17d). It is claimed that over 90% of nanostructures have high energy facets. Furthermore, the large-scale synthesis can be completed in minutes. Besides the above typical synthesis methods, a unique method with an amorphous TiO2 layer on a α-Ti substrate as the starting material was developed to prepare a film of rutile pillars with well developed facets.319 The details will be presented in section 2.8.

Compared to the synthesis of anatase crystals with clear facets or brookite crystals (to be discussed later), the synthesis of rutile crystals with clear facets using wet-chemistry methods rarely uses organic capping agents. Nearly all methods use a reaction system containing Cl− ions due to the strong role of Cl− ions as a mineralizer in facilitating the nucleation and growth of the rutile phase. Developing the Cl−-free methods together with exploring organic capping agents could provide a way to synthesize rutile crystals with other new facets, particularly, high-index ones. 2.4. Faceted Brookite TiO2

Brookite is the least studied of the three natural phases of TiO2 largely because of the extreme difficulty in synthesizing singlephase brookite. Brookite always appears as a minor byproduct of anatase or rutile in most synthesis methods, although the calorimetric data for the transformation enthalpies of anatase to rutile, and of brookite to rutile suggests the thermodynamic phase stability should be in the order rutile > brookite > anatase.320 In comparison with anatase (tetragonal, I41/amd) and rutile (tetragonal, P42/mnm), brookite (orthorhombic, Pbca) has a low structural symmetry. Figure 19 compares the

Figure 19. Representations of the TiO2 anatase, rutile, and brookite forms.321 Reprinted with permission from ref 321. Copyright 2010 American Chemical Society.

structures of the unit cells of the three phases. Among them, only anatase is built by sharing the edges of TiO6 octahedra along all three directions. The octahedra in rutile are connected by sharing the corners along the a and b directions but the edges along the c direction. Although the connection of the octahedra along the c direction in brookite is similar to that in rutile, the connection configurations along the a and b directions are very complex, sharing both the corners and edges.321 The complexity and the appearance of anatase and rutile characteristics in the brookite structure might partially explain the great difficulty of growing single-phase brookite and the coexistence of brookite with anatase or rutile. Gong et al. determined the equilibrium crystal shape of brookite TiO2 through the Wulff construction with surface formation energies computed for different facets, and analyzed the features of different facets (Figure 20).262 Its surface consists of seven facets: (100) with a computed surface formation energy of 0.88 J m−2, (010) with 0.77 J m−2, (001) with 0.62 J m−2, (011) with 0.74 J m−2, (101) with 0.87 J m−2, (111) with 0.72 J m−2 and (210) with 0.70 J m−2, whose relative areas are 2%, 7%, 18%, 4%, 2%, 34% and 33%, respectively. Most of the surface area is taken by the (010), (111), (210), and reconstructed (001) facets with relatively low surface energies. Similar to other metal oxides, the concentration of exposed Ti atoms with low coordination numbers is the decisive factor affecting the relative stabilities of the different facets of brookite. The higher the concentration of 9573

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

molecules of a brookite nanostructured TiO2 thin film (spherical particles and nanorods) at an optimum working temperature.324 All these promising applications have greatly stimulated the increasing research interest in developing various methods of synthesizing pure brookite phase, particularly with controllable shapes. So far, many brookite crystals with different morphologies such as nanotube, nanorod, nanoflower, nanocube and nanosheet have been obtained. Here we mainly introduce the development of brookite crystals with well recognized facets. Several impressive methods have been developed to prepare brookite TiO2 nanocrystals with the preferential growth orientation along the c-axis.255,256,332,334,335 Buonsanti et al. developed a surfactant-assisted nonaqueous strategy for synthesizing high-quality anisotropically shaped brookite nanorods with a tunable aspect ratio.256 The strategy relies on controlling the time of the aminolysis reaction of titanium oleate complexes, which takes place during treatment at 290 °C of mixtures of titanium tetrachloride (TiCl4), oleic acid (OLAC), and oleyl amine (OLAM) under both water- and oxygen-free conditions. The formation process of brookite nanorods can be divided into two stages. In a slow-heating stage, spherical or rodlike titania nanocrystals that act as seeds were formed. Subsequent feeding of the seed at a controllably slow rate in an equimolar TiCl4/OLAC mixture contributes to the continuous anisotropic growth of the nanocrystals. A selfregulated phase-switching seed-catalyzed mechanism was rationalized to explain the growth of the nanorods, where the initially generated anatase seeds trigger heterogeneous nucleation of brookite and promote subsequent growth of the nanorods in the second stage. It was considered that brookite originates from the direct solid-state phase transformation of the initially formed anatase seeds elongated along the c-axis, by highly correlated atomic displacements. This hypothesis was supported by the high degree of similarity between the two TiO2 crystal structures, the observation of hybrid nanocrystals with epitaxially interconnected domains of the two polymorphs during the crystal-phase transition stage, and a characteristic lattice elongation in the c-axis direction for both the anatase seeds and the final brookite nanostructures. The nanorods prepared by this method have a very uniform particle size as shown in Figure 21a. The nanorod surface consists of longitudinal {210}/{100} facets and basal {001} facets. In comparison to most methods of synthesizing brookite using a precursor of TiCl4 or titanium alkoxides that are very sensitive to moisture, methods using water-soluble precursor complexes have the major advantages of being able to easily manipulate the precursors in air and being low cost. Such studies with the commercially available precursor TiOSO4 or titanium bis(ammonium lactate) dihydroxide (TALH) or homemade water-soluble complexes (e.g., a titanium-glycolate complex) have been made.255,332,334 By hydrothermally treating aqueous solutions of TALH in the presence of urea as an in situ OH− source, anatase/brookite with tunable ratios can be obtained.332 The high quality brookite nanorods with preferential growth along the [001] direction (Figure 21b) can be obtained in the presence of a high concentration of urea (≥6M). It was revealed that the formation of the brookite is governed by the high concentration of the in situ produced OH− instead of the high final pH. Furthermore, different from the nonaqueous synthesis of brookite nanorods by Buonsanti et al.,256 where anatase nanocrystals were first formed as the seeds of the brookite nanorods, the brookite seeds here were directly

Figure 20. Equilibrium crystal shape of brookite TiO2, as obtained through the Wulff construction. Different facets have different colors, and the orientations are given.262 Reprinted with permission from ref 262. Copyright 2007 American Physical Society.

such atoms, the higher the surface energy. The atomic structure models of each facet are given in Figure S6. Of seven facets in the equilibrium shape, the reconstructed (001) facet with four Ti5c and two Ti4c atoms exposed per unit cell has the lowest surface energy of 0.62 J m−2. (111) with the third lowest surface energy of 0.72 J m−2 but accounting for the largest surface area of the crystal (34%) has one Ti4c and three Ti5c atoms exposed per unit cell. Due to the similarities in the bulk structures of the three polymorphs, it is reasonable to observe some structural similarities between brookite surfaces and those of the anatase and rutile phases. It is interesting that the structure of brookite (210) with both Ti5c and Ti6c exposed is very close to that of anatase (101), the most stable facet in anatase. Brookite (210) can be considered as a distorted anatase (101) based on the structural similarities. However, they possess quite different electronic and chemical properties.262 Although the theoretically predicted shape (Figure 20) is not strictly consistent with the observed shape of natural or synthetic brookite crystals, the most common facets in real samples usually have low surface energies. Such a difference is attributed to the fact that the calculation refers to surfaces in vacuum while the samples are influenced by the environmental growth conditions. In recent years, theoretical results and increasing success in synthesizing the pure phase have revealed the unusual properties and relevant applications of brookite TiO2 in electronic,255 electrochemical,322 catalytic,323 sensor,324 photocatalytic,325−327 and photovoltaic328 fields. Brookite is theoretically predicted to be a promising dielectric with a static dielectric constant much higher than both anatase and rutile.329 In contrast to the conventionally considered photocatalytic inactivity of brookite, many studies have clearly shown the superior photocatalytic activities of brookite or brookite-rich particles to anatase or rutile.325,327,330−333 For example, Kandiel et al. reported that the photocatalytic hydrogen evolution activity of anatase/brookite mixtures and of pure brookite is higher than that of pure anatase nanoparticles despite the lower surface area of the former.332 The superior activity was explained to be the result of the higher conduction band edge by 0.14 eV of brookite than anatase so that the driving force for proton reduction to release hydrogen is higher in brookite. Dambournet et al. showed that brookite had a higher volumetric energy density and lower irreversible capacity than comparable anatase or rutile nanomaterials.321 In addition, it was demonstrated that UV radiation can significantly enhance the conductivity and sensitivity toward NO2 oxidizing gas 9574

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

hydrothermal reaction stage. The coexistence of Na+ and NH4+ ions in the interlayer spaces of titanate is considered to play a key role in controlling the formation of the pure brookite phase. The easy release of NH4+ from the galleries during hydrothermal treatment facilitates the collapse of the layered structure, and some of the layered structure stabilized by Na+ delays and changes the structural transformation pathway to brookite. The other is to use the reaction system of TiOSO4 and NaOH, which is similar to that for nanospindle-like brookite.255 Layered titanate containing Na+ is also recognized as the intermediate to grow brookite TiO2 with a nanoflowerlike shape. By replacing NaOH with LiOH or KOH, no brookite but only rutile or anatase is formed even when the pH value in the system remains.253 In both cases, Na+ plays a crucial role in controlling the brookite formation, where layered titanate containing Na+ as the intermediate is formed. Pseudocubic brookite nanocrystals with a particle size of approximately 40 nm were prepared with titanium-glycolate complex as the precursor by an oleate-modified hydrothermal growth method.257 The resultant nanocrystals are surrounded by four {210} and two {001} facets, accompanied by small {111} facets, as shown in Figure 22a.257 The water-soluble

Figure 21. (a) Low- and high-magnification (left and right panels, respectively) TEM overview of TiO2 nanocrystals synthesized at a high OLAM/OLAC ratio of 52/0.5 (20 mmol of an equimolar TiCl4/ OLAM mixture was subsequently added).256 Reprinted with permission from ref 256. Copyright 2008 American Chemical Society. (b) FE-SEM micrograph of as-synthesized nanocrystalline TiO2 powder obtained by thermal hydrolysis of aqueous solution of the TALH precursor at 160 °C for 24 h in the presence of urea (6.0 M).332 Reprinted with permission from ref 332. Copyright 2010 American Chemical Society. (c) TEM image of phase-pure brookite nanospindles prepared after reaction at 220 °C for 48 h.255 Reprinted with permission from ref 255. Copyright 2013 Royal Society of Chemistry.

formed from the reaction solution. Zhao et al. reported that a Ti(OH)4(OH2)2 intermediate, derived from the hydrolysis of TiOSO4 in a NaOH solution (2.5 M), can be directly transformed into brookite nanospindles by hydrothermal treatment in a NaOH solution (1.0 M).255 The complete removal of SO42− ions from Ti(OH)4(OH2)2 by sufficient washing with deionized water before hydrothermal treatment is crucial to the formation of the brookite phase because SO42− ions favor the formation of the anatase phase. The surface of brookite nanospindles is enclosed with four lateral {210} facets and eight top {111} facets (Figure 21c). It is considered that in a strong alkaline solution, the abundant OH− ions tend to absorb on the {210} facets, which lowers the surface energy of {210} and thus somewhat inhibits crystal growth along the [210] orientation and preferential growth along the [001] direction. In addition, brookite nanorods enclosed with lateral {210} and claimed top {212} facets were also obtained.334 Besides the well dispersed nanorods described above, brookite nanoflowers consisting of nanorods with the preferential growth orientation along the [001] direction were prepared by a hydrothermal process in two different reaction systems.255,335 One uses the system tetrabutyl titanate, NaCl and ammonia.335 Depending on the concentration of NaCl, different phases can be obtained. The NaCl-free method only leads to the formation of the anatase phase. Pure brookite nanoflowers with nanorods as building blocks can be obtained when the concentration of NaCl is 0.25 M. A concentration of NaCl higher than 1.5 M results in a pure phase of layered H2Ti2O5.H2O. The formation of brookite is derived from the Na2‑x(NH4)xTi2O5.H2O intermediate formed in the initial

Figure 22. (a) HRTEM image of a brookite particle dispersed on on a carbon grid viewed along [120].257 Reprinted with permission from ref 257. Copyright 2011 American Chemical Society. (b) HRTEM image and (inset) corresponding FFT result for horizontally oriented (brookite) nanosheet. (c) Stereoshape model proposed for brookite nanosheets in panel b.254 Reprinted with permission from ref 254. Copyright 2012 American Chemical Society.

titanium-glycolate complex was obtained by dissolving Ti powder in a mixture of NH3 and H2O2 solution to first form a peroxo-titanium complex solution and the subsequent addition of glycolic acid. The central role of the oleate-based addition to the reaction system is the selective absorption of oleic acid molecules on specific crystal faces of the metal oxide nanoparticles to lower surface energy. In the case of brookite, the addition of oleic acid in the synthesis is thought to cause preferential absorption on the {210} and {001} facets, which induces the particles to exhibit a pseudocube morphology. In contrast, without the addition of sodium oleate, only brookite 9575

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

nanorods were formed.336 The time-dependent crystal growth shows that the growth of brookite crystals may be dependent on a transformation of the classical Ostwald ripening principle involving dissolution of smaller nanoparticles and reprecipitation of larger particles by increasing the hydrothermal temperature and time. Furthermore, the smaller nanoparticles formed in the initial stage were anatase. This might be understood by the fact that anatase is most stable when the particle size of TiO2 is below 11 nm, and brookite is most stable when the size is between 11 and 35 nm.281 Liu et al. prepared impressive brookite single crystal nanosheets surrounded with four {210}, two {101}, and two {201} facets (Figure 22b and c) from a system with TiCl4 as a titanium source, urea as an in situ OH− source, and sodium lactate as the complexant and surfactant.254 The in situ generated precursor (chiral space group P2 1 3), [Ti(C3H4O3)3]2−, is more asymmetric than the commercial TALH complex used for brookite synthesis by Kandiel et al.332 Considering the low symmetry of the brookite structure, the [Ti(C3H4O3)3]2− complex is believed to be more efficient in forming brookite under mild conditions. Similar to the growth of the pseudocube shaped brookite nanocrystals by Ohno et al.,257 the precursor [Ti(C3H4O3)3]2− can only be transformed to anatase when the concentration of OH− released by the decomposition of urea is low at the early stage of the reaction. The moderately alkaline solution generated by the complete decomposition of urea in a prolonged reaction leads to the condensation of the precursor to form brookite through oxolation during crystallization process. It was proposed that the initial anatase structure progressively turns into brookite as a result of assimilation triggered by the primary brookite lattice in the presence of Na+. Equally important, the selective absorption of extra lactate ions as the surfactant onto the specific planes of brookite to reduce the surface energy thus causing the kinetically selective control of the growth rates of various faces of a grain seed is responsible for the exposure of specific facets of brookite nanosheets. As presented above, brookite crystals with specific facets can be prepared by many synthesis routes. Compared to the synthesis of anatase or rutile crystals with specific facets, in which attention is mainly paid to the shape control, the synthesis of the brookite crystals with specific facets is much more challenging because of the requirement to control not only the shape of the crystals but also the nucleation/growth of brookite. The formation of the brookite phase is guaranteed by a combination of modulating solution chemistry and using an appropriate Ti precursor or in situ generated intermediates. Due to the fact that very acidic solutions favor the formation of nanorutile and extremely basic solutions fully transform nanobrookite to nanoanatase without the formation of rutile,282 most existing synthesis methods for brookite were conducted in moderately basic solutions. Besides the pH dependent phase stability, the phase stability of titania polymorphs is also dependent on the crystal size.281 Anatase is most stable when the particle size is below 11 nm so that anatase nanocrystals are always formed at the initial reaction stage. The fate of these anatase nanocrystals in the prolonged reaction process highly depends on the degree of crystal structure distortion of the nanocrystals themselves and/or solution chemistry variation. As demonstrated by Buonsanti et al.,256 anatase nanocrystals elongated along the c-axis formed initially can act as seeds for the subsequent growth of brookite after the addition of the precursor TiCl4/OLAC. The initially formed anatase nano-

crystals can also disappear following the classic Ostwald ripening principle involving dissolution of anatase nanocrystals and reprecipitation into larger brookite particles.257 In contrast, anatase nanocrystals formed at the initial stage are retained in the final product of an anatase/brookite mixture when the solution environment (i.e., 1.0 M urea in the case of Kandiel et al.)332 is not appropriate for phase transformation or dissolution. On the other hand, layered titanate is recognized as the intermediate in the early stage of brookite formation.253,335 The presence of Na+ plays an important role in stabilizing the monomers from layered titanate for the formation of brookite. The good structural similarity of the Ti precursor to the brookite structure336 and asymmetry254 of the Ti precursors are considered to be favorable for the formation of brookite. Without additional surfactants in the systems, brookite nanorods with preferential growth orientation along the [001] are the typical structures formed. The presence of oleic acid molecules or lactate ions as surfactant leads to pseudocubes257 or nanosheets254 of brookite crystals due to the selective adsorption of a surfactant on specific surfaces to lower surface energies. 2.5. Faceted TiO2 Mesocrystals

Mesocrystals (MCs) are oriented superstructures of nanocrystals that have common crystal facets, and are a new class of colloidal crystals formed from nonspherical, crystalline building units, as summarized by Cölfen et al.337,338 MCs are considered as intermediates between classical single crystals and polycrystals whose units do not have the same orientation.337 Different from a classical crystallization process, where crystal formation is controlled by monomer-by-monomer assembly, the MC formation follows a nonclassical crystallization process involving the spontaneous self-organization of adjacent nanocrystals as building blocks. The relevant formation mechanisms include nanoparticle alignment by an organic matrix, interparticle forces, ordering by physical ordering, mineral bridges, space constraints and self-similar growth and topotactic reactions.338 MCs formed in both inorganic and organic crystals are usually single-crystal-like due to the oriented assembly of nanocrystal building blocks with the same crystallographic orientation and have high porosity and surface area stemming from nonuniform nanocrystals or nonclose packing of nanocrystals or the removal of additives. All these properties make MCs an attractive class of crystalline materials for various applications. TiO2 MCs with defined facets have experienced a rapid development in recent years.279,339 One impressive example is based on the topotactic transformation mechanism. Zhou et al. prepared anatase MCs by the thermal decomposition, or aqueous hydrolysis with H3BO3, of the faceted NH4TiOF3 MCs.279 The similarities in the crystal structures of NH4TiOF3 and anatase TiO2 provide the possibility for the oriented transformation shown in Figure 23. The positions of titanium atoms in the [001] planes are similar in both structures, but in NH4TiOF3, these are separated by ammonium ions in a layered structure so that contraction along the [001] direction occurs during the topotactic transformation from NH4TiOF3 to anatase because of the greater atom density in TiO2. The overall crystal orientation, however, remains unchanged as a result of the crystallographically matched template of NH4TiOF3 MCs for the subsequent growth of the TiO2 MCs. The resultant anatase MC platelets are single-crystal9576

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

Figure 24. (a) SEM image of Meso-TiO2-500 showing a plate-like morphology. (b) Magnified SEM image of a crystal surface revealing porous structures.340 Reprinted with permission from ref 340. Copyright 2012 American Chemical Society. (c) Representative image of TiO2 MCs (MC-4).95 Reprinted with permission from ref 95. Copyright 2012 Wiley-VCH. (d) TEM image of rutile TiO2 mesocrystal obtained in the presence of SDBS (the molar ratio of titanate/SDBS is 0.09).345 Reprinted with permission from ref 345. Copyright 2012 Elsevier.

Figure 23. Illustration of the oriented transformation of NH4TiOF3 mesocrystal to TiO2 (anatase) mesocrystal.279 Reprinted with permission from ref 279. Copyright 2008 American Chemical Society.

the major thermodynamically stable {101} facets and minor high energy {001} facets.95,339 One remarkable example is the additive-free synthesis of nanoporous anatase MCs with a single-crystal-like structure using tetrabutyl titanate (TBT) as the titanium source and acetic acid (CH3COOH) as the solvent under the solvothermal conditions.339 The solvent acetic acid was considered to play multiple roles in the mesoscale assembly of the MCs by lowering the reactivity of TBT, enabling the slow release of the soluble titanium containing species, temporarily stabilizing the tiny anatase nanoparticles against immediate single crystal formation or uncontrolled aggregation, and producing the template butyl acetate. The formed spindleshaped MCs elongated along the [001] direction are built from tiny nanoparticles with diameters 10−20 nm that have mostly {101} facets as a result of the relatively weak binding of acetic acid on {001} facets. The size of the MCs can be readily tuned by changing the TBT concentration in the TBT-acetic acid system. The higher the concentration, the smaller the size. The reason for this change trend is that the TBT concentration positively determines the concentration of preformed nascent nanoparticles for the subsequent oriented attachment. In addition, the MCs can be stable up to 900 °C without phase transformation from anatase to rutile because interfacial sites of contacting anatase grains for the nucleation of rutile are largely eliminated. However, calcination at 900 °C led to the formation of solid single crystals. The above features may apply to other MCs. Chen et al. used a similar combination of formic acid (HCOOH) and titanium isopropoxide (TTIP) as reactants to synthesize anatase MCs with tunable ratios of {101} to {001} facets.95 The ratio can be tuned from 60:40 to 98:2 with an increase of the solvothermal duration from 4 to 10 h as a result of the preferable adsorption of HCOOH on {101} facets and thus oriented attachment along the [001] direction. Different from the obtuse shape of the MCs from the TBT-acetic acid

like and their surface consists of the major top {001} facets and minor lateral {010} facets. By using different surfactants, the thickness of NH4TiOF3 and thus anatase MCs can be controlled. This could be one advantage of the topotactic reaction in preparing MCs. However, it was not stated whether F species released from the NH4TiOF3 mesocrystals plays an important role in stabilizing a large percentage of the highenergy {001} facets of anatase MCs. Several other studies used a similar topotactic transformation strategy to prepare anatase MCs.111,218,340,341 In particular, porous plate-like anatase MC sheets with dominant {001} facets (Figure 24a,b) were obtained by heating a thin layer of an aqueous solution containing TiF 4, NH 4F and NH 4NO3 covering a silicon wafer.340 In the annealing process, the materials underwent changes in two stages. An intermediate plate-like NH 4TiOF 3 was first formed by a series of combination reactions of Ti4+, F−, NH4+, and H2O at low annealing temperatures (250−300 °C) and this was then transformed into TiO2 at a higher temperature (400−800 °C). The unique microstructures of the mesocrystals leads to the improved charge separation efficiency, large surface area, and highly efficient transportation of reactants/products, whose synergistic effects are responsible for the improved photocatalytic activities. This synthesis method provides a simple way of directly immobilizing MCs on various substrates to fabricate electrodes. In addition, hexagonal microrods of anatase mesocrystals with {001} and {101} facets were prepared by directly heating a TiO2 anodic layer at 500 °C in air. The formation of the microrods is by self-directed self-assembly within intermediate scaffolds.342 TiO2 MCs can also be prepared by the oriented attachment growth of TiO2 nanocrystal building units derived from the hydrolysis of titanium precursors in appropriate reaction media. Most anatase MCs based on oriented attachment growth have 9577

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

2.6. Hollow Faceted TiO2 Crystals

system, the MCs were composed of faceted crystals about 30− 40 nm in diamter, as shown in Figure 24c. The above two examples demonstrate the substantial role of the solvent (CH3COOH and HCOOH) in forming anatase MCs by the preferential adsorption of solvent onto {101} facets. Several studies showed the formation of anatase MCs in methods involving SO42− anions that were considered to preferentially adsorb on anatase {001} to stabilize {001} facets due to the higher density of 5-fold Ti on the (001) surfaces.90,125,240,343 Bian et al. synthesized single-crystal-like anatase mesocages by the attachment along their {101} facets of 8 nm crystal units with {001} surfaces protected by SO42− anions. The resultant mesocages with mostly {001} facets have a disordered mesoporous network, which originates from the voids between the building units. Confining the attachment growth of the building crystals in the channels of mesoporous silica templates also led to the single-crystal-like mesocages but with ordered mesopores of 4.6 nm in diameter. These mesocages have a spheroidal shape instead of a similar shape to the building crystals. This can be explained as the balancing effects of minimizing surface energy by adapting a spherical shape and oriented growth of the building crystals into selfsimilar crystals. However, it was shown that H+ ions play a more important role than SO42− ions in the assembly of anatase nanoparticles in the presence of OA and OM as capping agents (they are tightly bonded to the (001) and (101) crystal faces of anatase TiO2, respectively) because of their interaction with OA and OM under hydrothermal conditions.343 In comparison to the widely studied anatase MCs, the formation of rutile MCs is the subject of fewer reports. This might be explained by the instability of rutile nanocrystals with a particle size below ca. 35 nm281 so that the nanocrystals tend to aggregate randomly to minimize surface energy before the occurrence of oriented attachment growth. An intriguing synthesis strategy for preparing rutile MCs developed by Hong et al. is to use titanate nanowires as the primary building blocks for the formation of MCs.344 By heating the titanate nanowires in a HNO3 solution under stirring at low temperature of 50 °C for 7 days and subsequently calcining the resultant product at 400 °C, nanorod-like single-crystal-like MCs constructed from ultrathin rutile TiO2 nanowires were obtained. The formation of such MCs follows the mechanism of the face-to-face oriented attachment of the ultrathin nanowires by hom*oepitaxial aggregation, accompanied and promoted by a simultaneous phase transformation from protonated titanate to rutile. The crystal phases of MCs derived from titanate nanowires are controlled by the acids used in the reaction media.240 With a H2SO4 solution only anatase MCs are obtained. By introducing sodium dodecyl benzenesulfonate (SDBS) into the above reaction system,344 nanoporous rutile MCs with tunable shapes can be achieved by controlling the molar ratios of titanate to SDBS.345 Ratios lower than 0.11 lead to the formation of MCs with the Wulff shape (Figure 24d), and higher ratios lead to rod-like MCs. The formation process of the MCs with the Wulff shape involves the aggregation and assembly of titanate nanowires along the [101] direction, accompanied and promoted by the simultaneous phase conversion from titanate to rutile with the increased reaction time. These rutile MCs show a large reversible charge− discharge capacity, excellent cycling stability and high rate performance when used in the anodes of LIBs due to their unique microstructures.

A hollow structure is intriguing for many applications of TiO2 crystals because it provides an inner surface and constrained inner space for trapping light by scattering, confined reactions, ion transport, and hosting guest materials. Many TiO2 hollow structures have been prepared by various synthesis strategies in the past decades,346,347 but they are usually polycrystalline and lack faceted surfaces. Promoted by the development of solid faceted crystals, some hollow TiO2 crystals with specific facets have been produced. One preparation method is to selectively etch faceted crystals along some specific crystallographic orientations based on different resistances to the corrosion in the reaction medium to which they are exposed. By hydrothermally treating rutile nanorods with major lateral {110} and minor top {001} facets in a HCl solution (Figure 25a), hollow rutile nanotubes with {110} facets on the four

Figure 25. (a) FESEM image of (rutile) TiO2 nanorods grown on a glass side (top view). (b) FESEM image of (rutile) TiO2 nanotubes (derived from rutile nanorods in panel a by the anisotropic corrosion).235 Reprinted with permission from ref 235. Copyright 2010 Royal Society of Chemistry.

tube walls (Figure 25b) were obtained as a result of the selective etching of the (001) surface along the [001] direction.235 This method has also been used to prepare hollow anatase crystals with facets, particularly, {101} facets. By treating amorphous TiO2 nanotubes in a H2O/HF solution under hydrothermal conditions, the microsphere assembled from single-crystal anatase dous with one {001} and eight {101} facets (Figure 26a) was prepared.103 Similar open and hollow structures were also reported by several other groups by prolonging the hydrothermal reaction time of Ti foil in HF solution or increasing the concentration of fluorine species in the reaction solution.91,141,148 The elongated truncated tetragonal bipyramid anatase crystals with {001}, {101}, and {001} facets (Figure 26b) were partially etched at the top {001} facets by increasing the hydrothermal reaction of TiOSO4 precursor in HF solution from 4 to 12 h.221 A unique double-hollow core−shell morphology of anatase crystals (Figure 26c) was obtained by increasing the hydrothermal reaction time to 31 h in a mixture solution containing TiF4, H2O, iPrOH, and HF.156 The formation of all these anatase hollow (or partially etched) crystals can be explained as the etching of {001} facets with the highest surface energy along [001], and will be discussed in detail in section 3.4. Jiao et al. reported a hydrothermal synthesis system containing Ti(SO4)2, Na3PO4 and HF to prepare hollow anatase single crystals (Figure 26d) and mesocrystals with dominant {101} facets.90 In contrast to the usually observed trend that a high HF concentration favors the formation of 9578

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

nanoplates that form the boxes.151,164,293 In addition, Wang et al. showed a topotactic transformation from TiOF2 nanocubes to hollow ordered three-dimensional anatase nanoboxes at a temperature of 80 °C in a NaOH solution.294 The six walls of the box consist of oriented aligned nanorod arrays along the [001] direction. The high concentration of OH− in the reaction solution favors the preferential growth of anatase nanorods along the [001] direction due to its key role in stabilizing {100} facets. The very small lattice mismatch between the two phases enables the epitaxial growth of anatase TiO2 nanorods on the surface of the underlying TiOF2, which acts as both the Ti precursor and a hard template to restrict the alignment of TiO2 nanorods. In all cases, the consumption of internal TiOF2 causes the hollow structure of TiO2 boxes. Hollow spherical and ellipsoidal nanostructures with their shells assembled from {001} anatase nanosheets were constructed using hard templates.128,131,348−350 With polystyrene hollow spheres or silica nanospheres as the templates, hollow TiO2 spheres with high surface areas were prepared by the assembly of {001} anatase nanosheets on the template followed by the removal the template.348 SnO2@TiO2 hollow spheres with an outer shell of assembled anatase nanosheets were also prepared by this approach. By using silica-coated αFe2O3 nanospindles as the starting templates, hollow ellipsoidal nanostructures of α-Fe2O3@SnO2@TiO2 and α-Fe2O3@TiO2 with a shell of assembled {001} anatase nanosheets were obtained. All these hollow crystals with the described facets show superior performance to reference solid crystals in various applications including photocatalysis, lithium batteries and DSSCs. The superiority can be attributed to their merits of unique hollow structure.

Figure 26. SEM or TEM images of (a) the microsphere synthesized with 40 mg of TiO2 nanotubes under the H2O2 (30%) and 0.05 mL of HF (40%) at 180 °C for 5 h.103 Reprinted with permission from ref 103. Copyright 2009 American Chemical Society. (b) Calcined etched elongated truncated tetragonal bipyramid (ETTB), mTi = 48 mg, th = 12 h.221 Reprinted with permission from ref 221. Copyright 2012 American Chemical Society. (c) The product synthesized through the solvothermal reaction with a mixture containing TiF4 solution (0.04 M, 2 mL), H2O (3 g), iPrOH (20 g), and HF (10 wt %, 0.4 mL) at 180 °C for 31 h.156 Reprinted with permission from ref 156. Copyright 2011 Wiley-VCH. (d) Hollow anatase TiO2 single crystal (with dominant {101} exposed).90 Reprinted with permission from ref 90. Copyright 2012 American Chemical Society. (e) Hollow structured TiO2 nanoparticles synthesized by introducing H2SO4 into the OA and OM system at 180 °C.343 Reprinted with permission from ref 343. Copyright 2012 Royal Society of Chemistry. (f) The hollow TiO2 boxes.151 Reprinted with permission from ref 151. Copyright 2011 Royal Society of Chemistry.

hollow anatase structure as a result of the strong etching of HF, a high HF concentration (500 mM) here leads to the formation of solid octahedral single crystals with {101} facets, and a low concentration (400 mM) leads to hollow octahedral-like single crystals with dominant {101} facets. Furthermore, unlike the etching process occurring on the (001) surface of anatase crystals with both {101} and {001} facets in the presence of HF, the initial etching site for these octahedral crystals is a shaper corner where facets meet because the corner has a higher energy than facets. This results in the hollow structure. Luo et al. reported hollow spindle-like anatase nanoparticles prepared using ahydrothermal system containing Ti(SO4)2, oleic acid (OA), oleylamine (OM), and H2SO4 (Figure 26e). A channel exists in the [001] direction.343 The formation of hollow nanoparticles is attributed to the effects of the adsorption and protection of OA and OM sufactants on the surface of TiO2 nano crystals, and of damage in functional groups or even in carbon chains of OA and OM and the corrosion in TiO2 nanoparticles caused by H2SO4 and also the coeffect of both oriented attachment and Ostwald ripening. Anatase boxes enclosed by six equivalent single crystal square {001} nanoplates (Figure 26f) were prepared by heat-treating solid precursor TiOF2 cubes with {100} facets at a temperature ranging from 450 to 600 °C in an air or argon atmosphere.151 The formation of the boxes starts from the surface of TiOF2 cubes and the internal TiOF2 acts as a hard template to restrict the newly formed anatase nanoplates. The very small lattice mismatch (0.34%) between anatase {001} and TiOF2 {100} surfaces, and the F species released by the decomposition of TiOF2 that acts as a capping agent are considered to be responsible for the large percentage of {001} facets on the

2.7. Mesoporous Single Crystal TiO2

A large accessible surface with the favorable structural properties of crystalline TiO2 is highly desirable in various applications such as lithium batteries, solar cells, (photo)catalysis and (bio)sensing, where the interface between the TiO2 surface and the electrolyte is a key factor affecting the performance. Besides decreasing the particle size, the formation of a mesoporous structure is an actively used strategy to increase the specific surface area of the crystals. Past efforts to develop mesoporous TiO2 have been very successful in obtaining large specific surface areas as summarized by many reviews.346,351−354 The synthesized mesoporous TiO2 crystals with disordered or ordered pores produced by various routes (soft/hard template and template-free) are basically polycrystalline so that many interparticle boundaries exist in them. However, the characteristics of long-range electronic connectivity and structural coherence are equally important in order to enable both the high conductivity and electron mobility required by many applications. The integration of being mesoporous and single crystalline in TiO2 crystals is the solution to achieving both demands but long remains a challenge.9 In the early attempts, a top-down method was used to fabricate porous single crystal TiO2. For instance, Sanz et al. fabricated well-ordered nanopore arrays in rutile TiO2 single crystals using nanolithography involving heavy ion bombardment through a porous anodic alumina mask.355 The formed nanopore arrays, with a pore diameter of approximately 70 nm, have a high-aspect-ratio of 16. The lithography can be performed on both rutile (111) and (110) substrates to 9579

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

from 250 to 20 nm. It was shown that both isolated mesoporous single crystals and ensembles incorporated into films have substantially higher conductivities and electron mobilities than does nanocrystalline TiO2. All-solid-state sensitized solar cells with the film of MSCs as the anode had reached the highest energy efficiency of 7.3%. Following the above synthesis strategy, Jiao et al. synthesized mesoporous rutile single crystals with a rod shape,356 whose surface is enclosed by lateral {110} facets and top {111} facets (Figure 28). Compared to the solid rutile single crystals with

produce porous single crystals, and nanopatterned areas, up to 4 mm2, can be obtained in a single radiation spot on singlecrystal TiO2. Recently, Crossland et al. developed a bottom-up method to prepare faceted, high-surface-area mesoporous single crystals of anatase TiO2 (Figure 27), whose external dimensions grow to

Figure 28. (a and b) High-magnification SEM and TEM images of mesoporous ruitle single crystal particles. 356 Reprinted with permission from ref 356. Copyright 2013 Royal Society of Chemistry.

Figure 27. (a) Schematic of mesoporous single-crystal nucleation and growth within a mesoporous template. (b) Pristine silica template made up of quasi-close-packed silica beads (fast Fourier transform (FFT) inset showing 6-fold symmetry with 49 ± 3 nm spacing). (c) Nonporous truncated bipyramidal TiO2 crystal hom*ogeneously nucleated in a 20-mM TiF4 solution at 210 °C. (d) Templatenucleated variant of the crystal type shown in c, grown from a seed at or near the external template surface such that a nonporous volume coexists with a mesoporous region within a single faceted microcrystal. (e) Replication of the mesoscale pore structure within the templated region (FFT inset, 47 ± 3 nm 6-fold symmetry) with crystal lattice vectors implied from the particle symmetry overlaid (reaction conditions 170 °C, 40 mM TiF4). (f and g) Fully mesoporous TiO2 crystals grown by seeded nucleation in the bulk of the silica template.9 Reprinted with permission from ref 9. Copyright 2013 Nature Publishing Group.

{110} and {111} facets, the mesoporous crystals showed improved photocatalytic activities in hydrogen or oxygen generation. The photoelectrode fabricated with the mesoporous crystals also gave a much higher photoelectrochemical water splitting than did the solid crystals. Based on the similar synthesis strategy, Zheng et al. synthesized rutile and anatase TiO2 mesoporous single crystals with diverse morphologies, which show better activities in photocatalytic hydrogen evolution and degradation of methyl orange.357 2.8. Films of Faceted TiO2

TiO2 is widely used in the form of electrodes in various electronic devices. Developing the preparation methods of the TiO2 films with tailored facets on the substrates is therefore an important issue for the use of TiO2. Two key points need to be considered for the film preparation. One is the intimate contact between the TiO2 and the substrates in order to enable a smooth interface charge carrier transfer. The other is the orientation control of the faceted particles in the film in order to maximize the advantage of targeted facets. Several ex situ methods such as doctor-blading, electrophoretic deposition, spin-coating and spray painting, have been widely used to prepare TiO2 film electrodes. Post-thermal treatment is usually used to increase the interface contact in these methods. The doctor-blading method has been also used to fabricate films of TiO2 crystals with a dominance of different facets on fluorine or indium doped tin oxide (FTO or ITO) glass in DSSCs.144,227,358,359 In this method the faceted particles always tend to randomly distribute in the films, which may counter the underlying merit of using faceted TiO2 crystals to improve the performance of the electrodes. Li et al. compared the influence of the orientation distribution of {001} anatase nanosheets in the electrode film on its lithium storage.360 It is demonstrated that an electrode consisting of the well patterned nanosheets with the [001] direction parallel to the substrate has three times the lithium storage capability at a high rate of 100 C compared to a film consisting of the spheres of randomly distributed {001} nanosheets. This strongly indicates the significance of

orders of magnitude larger than their internal mesopores.9 The central point of this method is to use a quasi-close packed array of silica beads (Figure 27b) seeded by pretreatment in a diluted TiCl4 solution as the template with microscopic nucleation sites for the subsequent hom*ogeneous nucleation and growth of crystals in the template. The removal of the silica template by selective etching in a NaOH aqueous solution recovers the mesoporous TiO2 crystals (Figure 27d−g). A combination of several techiniques conclusively reveals the single crystal nature of the mesoporous TiO2 crystals, whose external symmetry matches that of hom*ogeneously nucleated bulk crystals with {001} and {101} facets grown without a template (Figure 27c). The morphology (shape and particle size) of such mesoporous crystals varies much depending on the hydrothermal conditions (temperature, concentration of the TiF4 precursor), the density of seeds on the template, and the diameter of the silicon beads.9 For example, a mesoporous crystal synthesized at 130 °C, 120 mM TiF4 is elongated along [010] direction, and its surface consists of major {010} and {001}, and minor {101} facets, which is different from the truncated bipyramids (Figure 27f,g). The crystal size decreases with an increase in the density of seeds on the template. In addition, the pore sizes in mesoporous crystals can be easily tuned by adjusting the diameter of the silica beads in the range 9580

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

controlling the orientation distribution of faceted crystals in the films to achieve excellent performance. To control the orientation distribution of crystals, Van et al.361 used a manual assembly method to prepare oriented anatase TiO2 films with exposed {100} facets from two kinds of anatase crystals with dominant {100} facets. Polyethylenimine (PEI), which is positively charged, was first spin-coated on the substrates. The anatase particles were then placed on the PEIcoated substrates and rubbed smooth by a finger for a while. The obtained film was calcined to remove the PEI linker and also increase the adhesion between the TiO2 crystals and the substrates. The procedure surprisingly resulted in monolayer film with the [100] axis of the crystals perpendicular to the surface. The electrostatic attraction between PEI molecules and negatively charged TiO2 crystals could play an important role in forming the monolayer film on the substrate. XRD pattern analysis suggests that this method is sensitive to the shape of crystals. The monolayer formed from rod-type crystals give a better [100] orientation perpendicular to the surface than that from cubic-type crystals (Figure 29a,b), as indicated by the

Figure 30. (a) Typical top-view image of the resultant film on the Ti foil.137 Reprinted with permission from ref 137. Copyright 2011 Royal Society of Chemistry. (b) FESEM image (top-view) of anatase TiO2 film with oriented {001} facets deposited on FTO substrate after 4 h of hydrothermal growth.135 Reprinted with permission from ref 135. Copyright 2011 Royal Society of Chemistry. (c) High magnification SEM images of the fluorine-doped tin oxide surface hydrothermally reacted at 150 °C for 12 h in the hydrochloric acid/water (v/v by 1:1) solution with addition of 0.5 g ammonium hexafluorotianate.134 (d) XRD patterns of the FTO before and after hydrothermal reaction.134 Reprinted with permission from ref 134. Copyright 2011 Wiley-VCH.

morphology-controlling agent to reduce the {001} surface energy. A similar synthesis procedure was also used to synthesize TiO2 films consisting of anatase microspheres with exposed {001} facets on the Ti substrate.141,148 Liu et al. prepared anatase TiO2 films (Figure 30b), made up of partially agglomerated crystals with major {001} facets, on FTO glass substrates by keeping the substrates in the aqueous solution of TiF4 during the hydrothermal process.135 The HF or F− ions released through the hydrolysis of TiF4 can act as the morphology controlling agent to stabilize the {001} facets of TiO2. This procedure is, however, not suitable for ITO, silicon and glass substrates due to the instability of these substrates in a solution containing HF. Ichimura et al. reported anatase films exhibiting ∼100% (001) reactive facets grown hydrothermally on a gold substrate from a hom*ogeneous solution of TiF4 and NaF.171 The properties of these films are insensitive to the fluorine source used, as shown by investigating different fluorine salts. It is noted that all these anatase films on various substrates give a remarkably high (004) diffraction intensity in XRD patterns as a result of a high degree of preferred orientation along the c-axis of the anatase crystals with major {001} facets normal to the substrate surfaces. In contrast, the modified procedures by Feng et al. led to the formation of a film of mostly {001} anatase sheets (Figure 30c),134 where the sheets adhere to the FTO substrate with their c-axis basically parallel to the substrate surface. As a result of this high orientation, compared to the powder diffraction pattern, the (101) diffraction peak is significantly increased (Figure 30d), and some diffraction peaks including (110) and (001) are absent. Besides the above hydrothermal routes, CVD and electrochemical anodization processes have also been used to grow the films of anatase crystals with a majority of {001}.172,175 High-

Figure 29. SEM images of the axis oriented monolayer films of (a) cubic-type and (b) rod-type TiO2 single crystals, (c) XRD pattern of the monolayers on glass substrate, and (d) the photoelectrochemical property of the TiO2 particle-monolayered films.361 Reprinted with permission from ref 361. Copyright 2013 Wiley-VCH.

much greater intensity of the (200) diffraction peak compared with its standard intensity as well as the intensity from the film composed of cubic-type crystals (Figure 29c). In situ preparation routes have the apparent merit of producing strong interface adhesion by the direct nucleation and growth of TiO2 on the substrates, and controlling the orientation distribution of faceted crystals. The attempt by Wang et al. prepared a TiO2 film with oriented anatase {001} facets directly grown on a titanium foil substrate by hydrothermally treating the foil in a HF aqueous solution (Figure 30a).137 The film consists of a top anatase layer, a middle rutile layer and a bottom TiOx layer. The anatase layer is made up of truncated pyramids with a major top square {001} facet and minor lateral {101} facets. Most {001} facets are parallel to the substrate. The use of HF plays a dual role. One is to partially dissolve the Ti foil to produce soluble titanium complex for the nucleation and growth of TiO2 crystals on the Ti substrate. The other is to act as a 9581

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

was conducted by Kakiuchi et al., where TiCl3 was used as precursor together with a NaCl additive.286 The degree of perfection of the top surface of the rods depends on the hydrothermal synthesis temperature. Only needle-like rods with a rough top-surface were formed at a low temperature (80 °C), while the faceted rods with lateral {110} and top {111} were obtained at 200 °C for 24 h. Rutile films with 100% triangular (111) surfaces that were reactive to visible light were prepared on FTO substrates were also prepared by hydrothermally treating TiN in a solution of HCl and H2O2.247 The bulk Ti3+ is considered to be responsible for the visible light absorption. The presence of H2O2 in the reaction solution plays a key role in the formation of such films. The formation of peroxotitanium complexes by Ti4+ reacting with H2O2 during the hydrothermal process is believed to slow down the hydrolysis of the titanium precursor and favor the formation of {111} facets. A substrate-free, free-standing and two-side oriented single crystal rutile TiO2 nanorod array film with a thickness of over ten micrometers was prepared by a hydrothermal reaction of titanium powder in a mixture solution of H2O2 and HCl.243 The film surface consists of well recognized (101) and (011) facets. The formation of such a film is explained by a three-step growth process (formation, agglomeration, and self-assembly of rutile crystal nuclei). In addition, several other hydrothermal routes have been developed to prepare rutile nanorods with major lateral {110} facets.310,362,363 In all the above cases, Cl− ions play an essential role in controlling the formation of the rutile phase by affecting the linkage of 6-fold coordinated monomers, and obtaining rods with major lateral {110} facets by preferentially adsorbing and retarding the growth rate of (110) surfaces. An alternative route based on the lattice matching effect between substrate and supported crystals was developed to prepare a film of faceted rutile crystals on a Ti substrate.319 An amorphous TiO2 compact layer on an α-Ti substrate as the starting material is derived from the anodized Ti foil with the top nanotube arrays and middle compact layer of TiO2 by removing the top nanotube arrays. By carefully controlling the crystallization process in an oxygen atmosphere, a rutile film (Figure 32A), consisting of isolated domains with a size distribution from several to tens of micrometers, was formed. The similar size distribution of the domains to that of the Ti grains in the α-Ti foil (Figure 32B) suggests that the crystallization of the amorphous film on the polycrystalline Ti foil is affected by the underlying Ti grains. This phenomenon is attributed to the lattice matching effect between (001) of α-Ti and (110) of rutile. The pillars in each domain with high crystallinity are single crystals, and have a clearly faceted shape close to the equilibrium shape of a rutile TiO2 crystal (Figure 32D,E). Such process is basically controlled by the synergistic effects of lattice matching between rutile and α-Ti, crystallization temperature and also the pressure of the oxygen atmosphere.

density anatase TiO2 nanosheets with {001} facets were synthesized on silicon and silicon-coated substrates by CVD in two conventional tube furnaces connected to each other in series.175 Depending on gas flow conditions, either randomly oriented nanosheets or aligned nanosheets were prepared. Similar to the films prepared hydrothermally by Feng et al.,134 the angle between the (001) surface of most nanosheets and substrate is close to 90° due to the one axis orientation growth of nanosheets. In this CVD process, TiCl4 was used as the precursor in the presence of an Ar/O2/H2 mixture and no fluorine, the typical stabilizer of {001}, was used. Furthermore, the sheet growth temperature (800/450 °C) is much lower than that (1300 °C) used in the preparation of the powder of truncated bipyramids from TiCl4 by CVD. The proposed key point for the sheet growth at medium temperature is the suppressed growth of TiO2 crystals along [001] by the adsorption of silicon atoms, generated by the evaporation of silicon during H2 autoignition. The ultrahigh temperature (2000−2400 °C) of the H2 flame is close to the melting point of silicon (2335 °C) so that evaporation of silicon is possible. In this case some silicon dopants may exist in the nanosheets. Films consisting of an array of anatase TiO2 nanotubes on a Ti foil substrate, prepared by anodization of the Ti foil in an electrolyte containing fluorine ions and subsequent thermal crystallization, have been widely investigated because of the potential merit of their tube structure in promoting charge transfer and/or providing a large surface area for loading dyes or other semiconductors. These polycrystalline nanotubes have major {101} facets as do common anatase particles. Surprisingly, Jung et al. prepared single-crystal-like anatase nanotubes with mainly {001} facets (Figure 31a) by

Figure 31. (a) SEM images of a TiO2 nanotube array formed on a Ti substrate using an anodic etching procedure and (b) XRD patterns of TiO2 nanotubes annealed at 550 °C.172 Reprinted with permission from ref 172. Copyright 2012 Royal Society of Chemistry.

introducing PVP and acetic acid in the electrolyte during the anodization synthesis of TiO2 nanotubes.172 An oriented attachment mechanism was proposed for the formation of such nanotubes. PVP in the electrolyte solution acting as a surfactant and controller of crystal growth was preferentially adsorbed onto the (101) surfaces so that the (001) facets grew more quickly, leading to single-crystal-like anatase nanotubes with major (001) facets. The preferential growth of nanotubes along the [001] direction is strongly indicated by the remarkably increased relative intensity of (004) from 20% to 100% in the XRD pattern (Figure 31b). The percentage area of {001} facets in the nanotubes is estimated to be approximately 77%. Films of rutile TiO2 with tailored facets can be prepared by several methods. An early study of the preparation of films of arrays of rutile rods with high energy {111} on glass substrates

3. UNUSUAL PROPERTIES OF TIO2 CRYSTALS WITH DIFFERENT PREDOMINANT FACETS 3.1. Surface Reconstruction

Surface reconstruction as 2D defect always exists in most crystals. Numerous studies in Surf. Sci. have been conducted to recognize the geometric structures of surface reconstructions on various facets of TiO2 crystals, and their influence on surface adsorption and reactivity. Several review papers have provided 9582

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

Figure 33. Side view of (a) bulk-terminated and (b) 1 × 4 reconstructed anatase (001) slabs. The bottom trilayer (shaded area) is fixed at bulk positions in the DFT optimizations. Red and gray spheres represent oxygen and Ti atoms, respectively.366 Reprinted with permission from ref 366. Copyright 2013 American Chemical Society.

used to remove surface F in the experiments. Compared to the unconstructed surface, the reconstructed surface is weakly reactive in terms of its dissociative water adsorption. Giorgi et al. provided another possible explanation of the higher photocatalytic activity of the anatase (001) surface with the unreconstructed (1 × 1) structure than that with the (1 × 4) reconstruction by comparing the spatial behavior and optical signatures of excitions on the two different surfaces.368 As a result of the distinct valence band frontier orbitals of the (1 × 4) and (1 × 1) surfaces (Figure 34), the remarkably different

Figure 32. (A) Low magnification SEM image of the rutile TiO2 film on an α-Ti foil substrate prepared at 550 °C under an O2 pressure of 103 Pa; (B) optical image of the a-Ti foil used showing grain boundaries; (C) SEM image of the rutile TiO2 pillars constrained within a grain; (D and E) top view and cross-sectional view of the rutile TiO2 pillars. The inset in panel E is the cross-sectional view and top-view equilibrium shapes of a rutile TiO2 crystal according to the Wulff construction.319 Reprinted with permission from ref 319. Copyright 2012 Royal Society of Chemistry.

comprehensive summaries of the results of surface reconstructions for TiO2, which are usually achieved on macroscopic single crystals supported on substrates (SrTiO3-(001) or LaTiO3-(001)).59,364 Here, we will not consider the progress on surface reconstructions of TiO2 crystals themselves in detail, but present some representative results showing the effects of surface reconstructions on the electronic structure, reactivity and doping. Considering the overwhelming dominance of {001} facets on anatase crystals with the size of tens of nanometers to several micrometers in such studies, we chose the (1 × 4) reconstruction on anatase (001) as an example. This reconstruction always exists on a sputtered and heated anatase (001) surface in ultrahigh vacuum.59 One geometric structure of the (1 × 4) reconstruction that is revealed theoretically is based on the “ad-molecule” (ADM) model.365 This model can be considered as the periodic replacement of rows of surface bridging oxygen of the (1 × 1) surface by the rows of TiO3 species, as shown in Figure 33.366 With respect to the (1 × 1) surface with 100% Ti5c and 100% O2c atoms exposed, the (1 × 4) reconstruction causes the reduction of Ti5c atoms by 25% and the formation of Ti4c atoms.367 The reconstructed (001) surface has a decreased surface energy of 0.51 J m−2 compared to the original value of 0.90 J m−2 and therefore has an improved stability.365 One open question already mentioned by Selloni et al.367 is whether the (1 × 4) reconstruction also occurs in micrometersize and even nanosize anatase crystals with the dominant {001} facets synthesized by the wet-chemistry methods with HF as the stabilizer. This question has been preliminarily addressed in crystals of μm size by Selçuk et al.366 It is theoretically shown that the anatase (001) surface exhibits a bulk-terminated (1 × 1) structure when in contact with concentrated HF solutions but the (1 × 4) reconstruction is always stable at temperatures of 400−600 °C, the temperatures

Figure 34. (a) Large yellow (small, blue) spheres represent Ti (O) atoms for the (1 × 1)-(001) surface. Magenta isosurface: |Ψ|2 of the VBM, at Γ (at 1% of its maximum value). (b) As in (a) but for the (1 × 4)-(001) surface. (c) DOS of bulk anatase (black full line), (1 × 1)(001) (dotted-dashed yellow line), of (1 × 4)-(001) (red dashed line).368 Reprinted with permission from ref 368. Copyright 2011 American Physical Society.

optical spectra exist in the two cases. The complete exciton delocalization in the subsurface and bulk part of the slab (a small probability that both electrons and holes are immediately available at surface sites and involved in subsequent chemical reactions) is related to the weak reactivity of the (1 × 4) surface. On the other hand, the two basic characteristics of photocatalytic processes of (a) spatial separation of photoexcited electrons and holes and (b) their immediate copresence at the surface are well satisfied in the (1 × 1) surface. A recent study showed completely different results in that the reconstructed surfaces (oxidized or reduced) are totally inert to the spontaneous dissociation of H2O molecules at room temperature, and the dissociation of H2O molecules only occurs on the reduced reconstructed surface with Ti3+ at 80 K.369 Furthermore, the authors proposed a different surface 9583

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

than that of rutile phase. Further study based on two anatase single crystal electrodes with exposed (001) and (101) surfaces showed a more negative flat-band potential of anatase (001) by 0.06 ± 0.02 V on average.41 The different flat-band potentials are attributed to the different chemisorptions of water on two surfaces.69 It was theoretically predicted that a clean anatase (101) surface with 50% Ti5c and 50% Ti6c atoms exposed adsorbs water nondissociatively, while a clean anatase (001) surface with 100% Ti5c atoms exposed adsorbs water dissociatively to form isolated hydroxyl groups anchored on Ti atoms.69 Consequently, the more acidic anatase (001) surface, which attracts less protons, is responsible for its more negative flat-band potential. However, no data show the possible difference in bandgap of the two anatase single crystal surfaces. The rapid development of the controllable synthesis of micrometer-size or nanosize TiO2 crystals with different facets makes it possible to also study the dependence of the surface electronic structure (bandgap and band edge) on the crystallographic orientation at the microscopic scale. One remarkable example is that nanosize anatase crystals with 82% {101} and 18% {001} facets have a blue-shift absorption edge by around 9 nm with respect to micrometer-sized anatase crystals with 28% {101} and 72% {001} facet areas (Figure 35).113 This is the first evidence for the larger bandgap of {101}

structural model from the widely accepted ADM model to satisfactorily explain the reactivity inertness of the reconstructed surface. In the newly proposed model, the oxidized (1 × 4) reconstructed surface grown under an O2 atmosphere is terminated with Ti6c atoms instead of Ti4c atoms in the ADM model. The reduced reconstructed surface contains a Ti4c atom by removing a TiO2 species and then intercalating two Ti atoms in the vacancies left from the oxidized reconstructed surface. All these theoretical results support the presence of the (1 × 4) reconstruction on the {001} facets of synthesized anatase crystals, which much impairs the reactivity of the facets or even makes the surface inert. Therefore, it is highly necessary to develop effective strategies to further modulate the structure of {001} facets on synthetic anatase crystals in order to generate favorable properties for targeted applications. Introducing heteroatoms or creating intrinsic defects such as oxygen vacancies and interstitial Ti atoms might be useful toward this goal. On the other hand, the (1 × 4) reconstruction with the ADM model is demonstrated to play a pivotal role in mediating the incorporation of nonmetal dopants (N and C) in the anatase (001) surface by Lee et al.370 They identified various subsurface interstitial binding sites for dopants and corresponding surface to subsurface penetration pathways on the reconstructed surface. The reconstructed surface is also favorable for the presence of subsurface oxygen vacancies, to which the adsorbed species can migrate to form substitutional dopants. Furthermore, the nonexposed oxygen sites just below the reconstructed surface play a key role in the incorporation of nitrogen and carbon in TiO2 (001). The reconstruction mediated incorporation of N and C not only modifies the optical properties of TiO2 but also changes the geometric structure of the (001) surface, to some extent. The change might counteract the negative effect of the reconstruction on the reactivity of the (001) surface. Besides reconstruction on the anatase (001) surface, reconstructions also occur on other surfaces of TiO2 crystals such as the (1 × 2) reconstruction on the rutile (110) surface and the (1 × 3) reconstruction on the rutile (100) surface under reduced conditions.59 The reconstruction itself definitely affects the properties of specific surfaces and thus causes a substantial difference in the activity between the real and ideal surfaces. Therefore, the role of surface reconstruction needs to be fully considered in tailoring the reactivity of faceted crystals.

Figure 35. UV−visible absorption spectra of nanosized anatase crystals with 18% {001} facets and micrometer-sized anatase crystals with 72% {001} facets.113 Reprinted with permission from ref 113. Copyright 2010 Royal Society of Chemistry.

dominant anatase crystals over that of {001} dominant crystals (3.3 vs 3.2 eV), which is also supported by the calculated electronic structure results. The origin of the bandgap difference in the two crystals is attributed to the different atomic configurations on {001} and {101}. Further study also draws the conclusion that {101} facets have a larger bandgap than {001} facets but have a bandgap very close to {010} facets by comparing micrometer-sized anatase crystals with a predominance of {001}, {101} and {010} facets, respectively.136 The larger bandgap of {101} than {001} is also supported by the theoretical calculations by Zheng et al.162 However, the theoretical calculations by De Angelis et al. suggested the same bandgap, within 0.01 eV, of anatase (001) and (101) surfaces.70 A difference in the bandgap of a crystal, depending on the exposed surface, is also observed in other semiconductors such as WO3,372 ZnO,373 and Ag3PO4.374 Clearly, bandgap anisotropy is likely general in semiconductors with different exposed facets.

3.2. Anisotropic Surface Electronic Structures

Anisotropic surface electronic structures are logically expected because of the different atomic arrangements and configurations in each surface. Crystallographic orientation dependent surface electronic structure anisotropy was first observed in early studies based on macroscopic single crystals of TiO2 with different orientations.41,371 The first study of the (101) plane of an anatase single crystal containing 0.22% Al and traces of V, Zr, Nb, and La as an electrode revealed that the flat-band potential of anatase (101) is negatively shifted by 0.2 V vs the flat-band potential of rutile (001), and the main difference between anatase (101) and rutile (001) is the higher conduction band edge of anatase (101).371 The bandgap of rutile is larger than that of anatase by 0.2 eV as observed in polycrystalline material. It seems that the more negative flatband potential of the anatase (101) surface by 0.2 V is relevant to the fact that the bandgap of the anatase phase is 0.2 eV larger 9584

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

nanosheet photoanodes with respect to the photoanodes of {101} faceted anatase nanocrystals as a result of the more negative flat-band potential of {001} facets than {101} facets.71 This is because the open-circuit voltage is controlled by the difference between the Fermi level of TiO2 and redox potential of the redox couple in DSSCs.6 Martsinovich et al. theoretically investigated the influence of the positions of the conduction band edge of different anatase surfaces on the efficiency of electron injection from the model carboxylated dye to anatase surfaces.379 As shown in Figure 36, the injection times are in agreement with the position of the slabs’ conduction band relative to the LUMO of dye when the injection energy is at the LUMO of dye.

So far, a bandgap change of around 10 nm has been achieved by changing the percentage areas of {001} and {101} facets in anatase crystals. It is still not clear how large a bandgap change can be produced by changing the facets involved. The facet itself may be considered as an ultrathin nanosheet of the minimum layers of TiO6 octahedral units along the specific orientation. By doing this, we may get some indications from two important examples. One is that the titania (Ti0.91O2) nanosheets with a thickness of 0.7 nm, delaminated from the layered lepidocrocite-type titanate, have a bandgap of around 3.8 eV.375 A nanosheet with negligible bulk, consisting of only two layers of TiO6 units, can be considered as one quasi-facet. This means that the bandgap might increase by around 0.6 eV by tailoring facets with respect to that of the bulk of anatase (3.2 eV). The other is the two-dimensional Ti3+/ impurity free TiO2 surface phase on rutile (011) produced by the oxidation of bulk titanium interstitials.376 The bandgap of the true surface phase with a different hexagonal arrangement from the rutile (011) substrate is as small as 2.1 eV, which means that the bandgap might decrease by more than 1 eV by tailoring the crystal facets of TiO2. Therefore, tailoring the facets could be an alternative strategy for bandgap modification. Besides the bandgap, the band edge is also a key issue to be considered in the applications of TiO2 polymorphs, particularly in photocatalysis and DSSCs. Based on the XPS valence band spectra of nanosize anatase crystals with dominant {101} facets and micrometer-size anatase crystals with dominant {001} facets, Liu et al.113 and Pan et al.136 proposed that anatase (101) has a higher conduction band edge than anatase (001). However, this result is inconsistent with the result of the more negative flat-band potential of anatase (001) than anatase (101) shown by electrochemical measurements41,71,377 and theoretical results.70,162 In addition, Zhao et al. theoretically gave the following order of the work function values of (101) > (010) > (001).378 The discrepancy observed might be explained by two factors. One is that the claimed higher conduction band edge of anatase (101) than (001) by Liu et al.113 might be within experimental error of the determination of the XPS valence band spectra. The other is the influence of surface adsorbed H2O on the properties of the crystal facets. All the electrochemical measurements used to determine the flatband potentials of the different facets were conducted in aqueous solution. Dissociative adsorption of water on (001) could exert a great influence on the acidic degree and thus the flat-band potential as suggested by Hengerer et al.41 In contrast, the measurements of XPS valence band spectra were conducted in vacuum so that the influence of water adsorption may be neglected. The anisotropy of surface electronic structure in terms of bandgap and band edges, demonstrated in the case of anatase crystals with different facets, could be general for other TiO2 polymorphs with different facets, though the studies on other polymorphs are rare. Such anisotropy plays an important role in determining the behavior of faceted TiO2 crystals in different applications. The fact is that the spatial separation of the photoexcited charge carriers between different facets exists in anatase crystals with {001} and {101} facets, rutile crystals with {110} and {011} facets,73 and brookite crystals with {210}, {101}, and {201} facets,254 as a result of different flat-band potentials (or conduction band edges) of the different facets. This effect will be discussed in detail in section 3.5. In the DSSCs, Laskova et al. reported a voltage improvement by 45 ± 2 mV at 100% sun illumination of {001} dominant anatase

Figure 36. Injection times (at injection energy = dye’s LUMO energy) for the most stable adsorption configurations, compared with the energies of the conduction band minimum for the clean anatase slabs. The arrow shows the calculated position of the dye’s LUMO.379 Reprinted with permission from ref 379. Copyright 2012 Royal Society of Chemistry.

3.3. Anisotropic Molecule/Cluster Adsorption

The effective adsorption of molecules on a TiO2 surface is the basic process in many applications of TiO2 such as DSSCs and photocatalysis. The states of the adsorbates are closely related to the exposed atomic structures of the surfaces. The water molecule is one of the most important molecules in photocatalytic reactions including water splitting, photodecomposition of organic pollutants and CO2 reduction for solar fuels. The effective adsorption of water molecules on the surfaces of TiO2 is the prerequisite for the subsequent photocatalytic reactions. Numerous experimental and theoretical studies59,380,381 have focused on this important topic. A review by Sun et al. gave a comprehensive summary of the major theoretical outcomes of titania-water interactions.382 It was pointed out that in most cases the central issue is whether or not water will dissociate on TiO2 surfaces, which is intrinsically determined by the coordination numbers of surface atoms and their separation. Brief summaries of the adsorption of water on the low-index surfaces of anatase and rutile TiO2 are given here. Most theoretical and experimental results support the molecular adsorption of water on anatase (101) surface with 50% Ti6c and 50% Ti5c atoms,383 and the dissociative adsorption of water on anatase (001) surface with 100% Ti5c atoms at low coverages.69 The dissociative adsorption of water is also favorable on an anatase (100) surface with 100% Ti5c atoms. Zhao et al. reported the chemical 9585

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

activity for the water decomposition reaction decreases in the order (001) > (103)f > (100) > (110) > (103)s > (101).384 For rutile, a defect free (110) surface adsorbs water molecules at low coverages in most studies.59,385,386 The adsorption states of water on the rutile (100) surface are controversial. Current option tends to believe that water can dissociate on rutile (100),59 to some extent. The dissociative adsorption of water is favorable on both rutile (011) and (001) surfaces.387,388 Surface reconstructions and defects sensitively affect the water adsorption. For example, the presence of bridging oxygen vacancies makes anatase (101) or rutile (110) active in the dissociative adsorption of water.389−391 A mixture of molecular/ dissociative adsorption on (1 × 2) reconstructed rutile (011) was identified as the most stable configuration at monolayer coverage.392,393 Studies of water adsorption on brookite surfaces are rare largely because brookite TiO2 is usually considered to be photocatalytically inactive. A dye together with a semiconductor oxide and a redox mediator are the three ingredients in DSSCs. Except for the properties of the dyes themselves (optical absorption range, extinction coefficient), the effective anchoring of dyes as sensitizers on surfaces of an oxide semiconductor plays an important role in affecting the conversion efficiency of light to electricity. For a specific dye molecule, the surface atomic structures control the dye anchoring geometries and thus affect the dye loading, electron injection from photoexcited dyes to TiO2 and the back electron transfer reaction. Several studies have experimentally shown the importance of modulating the exposure of different facets of anatase TiO2 crystals in the photoanodes of DSSCs. Wu et al. reported an increased conversion efficiency with the increase of percentage area of exposed {001} facets of anatase nanocrystals in DSSCs with the N719 dye sensitizer.144 The increased percentage area of {001} facets is favorable for both dye adsorption and retarding charge recombination. By comparing the adsorption behavior of the N719 dye on anatase nanoparticles and nanosheets with dominant {101} and {001} facets, respectively, Fan et al. found a larger specific adsorption capability of N719 on anatase nanosheets because of the higher dissociative adsorption of reactant molecules on the (001) surface.132 Laskova et al.71 showed that using a C101 dye sensitizer, compared to a photoanode of anatase (101) nanoparticles, a photoanode of anatase (001) nanosheets exhibited an increased voltage but decreased short-circuit current in the fabricated DSSCs. Furthermore, the back electron transfer for the anatase (001) nanosheets is 6 times slower than the same process on the anatase nanoparticles with dominant {101} facets, which was also observed in other studies.72,169 The difference in the back electron transfer kinetics in the two photoanodes is rationalized by the different C101 dye-anchoring geometries on the two surfaces.71 Two different C101 dye-anchoring geometries were theoretically found for the (001) and (101) surfaces,70 as shown in Figure 37. These two anchoring geometries differ in the binding of the dye carboxylic groups to the TiO2 surfaces (bridged bidentate vs monodentate), leading to a different tilting of the anchored bipyridine plane with respect to the TiO2 surface. The different geometries result in a smaller dye loading and slower charge recombination kinetics on the anatase (001) surface. Several theoretical studies compared the adsorption energy of different molecules on anatase surfaces and/or their influence on the electronic structure of different surfaces. Kusama et al. showed that N-containing heterocycles, which

Figure 37. Left: Isodensity plot of the hom*o for C101 in adsorption mode A on the (001)-TiO2 surface. Right: Isodensity plot of the hom*o for C101 in adsorption mode B on the (101)-TiO2 surface.70 Reprinted with permission from ref 70. Copyright 2012 American Chemical Society.

can be used as coadsorbates with other dyes in DSSCs, adsorb strongly on the surfaces of anatase in the order (100) < (101) < (001).394 Furthermore, the adsorption of N-containing heterocycles negatively shifts the Fermi-level of TiO2 due to the adsorbate dipole moment component normal to the TiO2 surface. Martsinovich et al. showed that the strength of formic acid adsorption of the most stable configurations on the surfaces of anatase follows the order (103)-smooth > (001) > (100) > (103)-faceted > (110) > (101).379 The strength of adsorption on anatase surfaces basically follows the surface energy order.379 Surface reconstruction has a potential influence on the adsorption of molecules on a specific surface. Ç akır et al. investigated the interaction of perylenediimid-based dye compounds (BrPDI, BrGly, and BrAsp) with both unreconstructed and reconstructed anatase (001) surfaces in DSSC applications.395 It was found that all the dyes at the most stable adsorption structures form strong bonds with the two surfaces. Furthermore, the adsorption of BrPDI on the reconstructed surface leads to a bandgap narrowing of TiO2, which suggests an alternative way of obtaining visible light responsive TiO2 photocatalysts. The presence of heteroatoms on the surface of TiO2 can modify the surface chemistry and thus the adsorption of molecules. Fluorine, the most frequently used capping agent, exists in the as-prepared faceted TiO2 crystals. Most studies suggest that the removal of surface fluorine results in the improved photocatalytic activities in photocatalytic hydrogen evolution from the aqueous solution containing methanol, and OH radical generation.107,136,292 Pan et al. demonstrated that fluorine terminated anatase crystals with different percentages of {001}/{101}/{010} facets have similar photocatalytic activities in generating hydrogen or OH radicals but have lower activities than their fluorine-free counterparts.136 This could be rationalized by the re-exposure of Ti5c atoms, which are passivated by surface F. The Ti5c atoms in some surfaces as mentioned above are the active sites for the dissociative adsorption of water. Sun et al. predicted that water molecules remain intact with F-terminated anatase (001) surface due to repulsive interactions between surface fluorine and oxygen.382 For the photocatalytic decomposition of organic molecules the situation is quite complex. Some studies also suggest an improved activity by removing surface fluorine from TiO2. However, Liu et al. reported the tunable photocatalytic selectivity of anatase microspheres with exposed {001} facets in decomposing methylene blue and methylene orange dyes.116 The origin of the photocatalytic selectivity is the tunable 9586

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

barriers for both surfaces are much higher than that for bulk diffusion (0.35−0.65 eV), implying that surface insertion is the rate-determining step. This finding is significant, particularly for lithium anodes composed of nanocrystals with a large surface area. Considering the apparently different atomic structures of some surfaces of other TiO2 polymorphs, orientation-dependent lithium storage anisotropy is also expected. Heating TiO2 crystals in a H2 atmosphere is a frequently used method to modify the properties (i.e., electronic conductivity, optical absorption) of TiO2, where the insertion of hydrogen in TiO2 crystals is a basic process. Motivated by the formation of black TiO2, prepared by heating anatasespherical-like nanocrystals in a H2 atmosphere at high pressure, for use as an efficient visible light photocatalysts,62 there is an increasing interest in preparing H2-treated TiO2 for photocatalysis and lithium ion batteries. Sun et al. showed that orientation dependent hydrogen insertion and storage in anatase nanocrystals.400 It was found that the insertion of hydrogen through the (101) surface is more favorable than through the (001) surface due to the much stronger adsorption energy (2.21 vs 0.69 eV) of hydrogen on O2c than on O3c atoms. Consequently, anatase nanocrystals with a predominance of {101} and {001} facets have hydrogen storage capabilities of 1.4 and 1.0 wt %, respectively (hydrogen stays in the interstitial sites between TiO6 octahededra). Hydrogenated {001}-anatase and {101}-anatase nanocrystals are blue and black, respectively. Subsequent study also supports the blue color of hydrogenated anatase nanosheets with dominant {001} facets.186

adsorption selectivity produced by surface chemistry and surface structure modifications. In contrast, Xiang et al. reported that the fluorine terminated {001} dominant anatase nanosheets show a much higher activity in photocatalytic degradation of acetone in air than do fluorine-free nanosheets as a result of the synergistic effect of surface fluorine and {001} facets.120 Similarly, F-terminated anatase nanosheets with dominant {001} facets showed a superior heterogeneous catalytic ozonation activity in decomposing oxalic acid compared to fluorine-free anatase nanosheets.170 The superiority of the former is attributed to the high concentration of oxygen vacancies, at which OH groups are formed by dissociating water. The OH groups facilitate the adsorption of oxalic acid on the TiO2 surface. It is not stated whether oxygen vacancies themselves will increase the chemisorption of oxalic acid. The adsorption of clusters or quantum dots of noble metals or semiconductors on TiO2 surfaces has important applications in catalysis, photocatalysis and solar cells. The strong interaction between clusters/quantum dots and substrates is crucial for communication between the two phases. Sun et al. theoretically compared the adsorptions of gold clusters (Aun, n = 1−10) on anatase (001) and (101) surfaces.396 It was found the adsorption of gold on the (001) surface is much stronger than that on the (101) surface as a result of the strong interfacial bonding (Au−O and Au−Ti) in the former. In CdS quantum dot sensitized solar cells, anatase nanosheets with dominant {001} or {100} facets have the larger number of active sites for the adsorption of CdS quantum dots.397 It was shown that the {001} and {100} nanosheet based photoanodes accommodate a number of cadmium ions per unit area as high as 11.7 × 10−9 and 8.3 × 10−9 mol cm−2, respectively, which was much larger than that on corresponding photoanodes made from Degussa P25 (1.9 × 10−9 mol cm−2). Clearly, the strong interfacial bonding between Au clusters and anatase (001), and the high loading of CdS quantum dots on anatase (001) and (100) surfaces are related to the highly unsaturated nature of high-energy (001) and (100) surfaces.

3.5. Synergistic Effects of Different Facets

Encouraged by the attractive dissociative adsorption of water on the anatase (001) surface predicted in 1998,69 it is logical to anticipate that the higher the percentage area of exposed {001} facets on anatase crystals, the higher activity the crystals will have. Anatase crystals with higher percentage areas of {001} facets have been pursued since the breakthrough in the synthesis of micrometer-sized anatase crystals with 47% {001} facets of high quality.75 In the past five years, the percentage area of {001} facets in anatase crystals has been increased to nearly 100%.102,110,140 Most studies on photocatalytic decomposition of organic pollutants draw the consistent conclusion that the photocatalytic activities of anatase crystals increase with an increase of the percentage area of {001} facets.167,290,401 For example, Han et al. showed that the photodegradation percentage of methylene blue increases from 46% to 59%, 73%, and 98% with the increased percentage area of anatase {001} facets from nearly 0% to 5%, 20%, and 60% (the remaining surface is composed of {101} facets).167 This means that the anatase (001) surface is indeed more reactive in decomposing organic molecules than anatase (101). In addition, the photoexcited charge transfer from fluorophores to TiO2 nanoparticles can be increased by a factor of more than 10 in the quenching rate constant at maximum with an increase of {001} percentage area from 30% to 69%.402 Some photocatalytic hydrogen evolution studies also support the trend of the higher activities of anatase crystals with higher percentage areas of {001} facets.154 The increasing number of studies on crystal facet dependent photocatalytic activities, particularly the photocatalytic hydrogen or OH radical generation, show quite complex trends. One remarkable example by Pan et al. reported that the micrometersize anatase crystals with 24% {001} and 76% {101} facet areas

3.4. Anisotropic Insertion of Lithium and Hydrogen

Lithium storage in TiO2 nanocrystals is significant due to its use in lithium ion batteries. An early electrochemical study on two macroscopic anatase single crystals with exposed (001) and (101) surfaces revealed anisotropic lithium insertion along the [001] and [101] directions.41 Li+ insertion is favorable on the (001) surface as shown by a higher standard rate constant for charge transfer (10−8 vs 2·10−9 cm/s) and a higher chemical diffusion coefficient for Li+ insertion (4 × 10−13 vs 2 × 10−13 cm2/s) for propagation along the c axis. Favorable lithium insertion along the c axis is attributed to the less dense atomic structure of the (001) surface. The success in the synthesis of anatase nanocrystals with tunable ratios of {001} to {101} facet areas provides the mesoscopic materials that can be used to examine the generality of anisotropic lithium insertion along different directions. By investigating the electrochemical behavior of lithium storage in anatase nanocrystals with {101} and {001} facets exposed Bousa et al. confirmed the validity of the higher chemical diffusion coefficient for Li+ insertion and faster interfacial charge transfer along the c axis.398 Sun et al. proposed the much faster transport of Li+ ions across the (001) surface than the (101) surface because of the smaller energy barriers for Li+ insertion into anatase (001) and (101) surfaces (1.33 eV vs 2.73 eV).399 Furthermore, the 9587

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

exhibit higher activities in both hydrogen and OH radical generation than that with 40% {001} and 60% {101} facet areas, and crystals with 14% {001}, 33% {101} and 53% {010} facet areas have the highest activities.136 The observed activity difference between crystals with different facets was explained by the effect of surface band structures on the redox abilities of photoexcited charge carriers in each surface associated with the conventionally considered density of undercoordinated atoms exposed on each surface. This explanation is based on the different bandgaps of three anatase low-index facets as discussed in section 3.2, which follows the order of {010} ≈ {101} > {001}. Subsequent studies from other groups also showed a similar trend that anatase crystals with a higher ratio of {001} to {101} areas do not always have a higher activity.88,292 For example, Gordon et al. demonstrated an ∼8 times higher photocatalytic hydrogen generation reactivity of 1 wt % Pt-loaded anatase nanocrystals with {101} dominant facets than with {001} dominant facets.88 Zhao et al. reported that anatase cuboids with dominant {100} facets exhibit a 3 times higher photocatalytic activity in generating OH radicals than do anatase sheets with dominant {001} facets.292 In this series, the experimental results have suggested that anatase crystals with major {001} and minor {101} or {010} facet areas are not really reactive in photocatalytic hydrogen evolution or OH radical generation compared to crystals with minor {001} and major {101} or {010}. The tentative explanation for the reversal of reactivity order of {001} and {101} by Pan et al. introduced the effect of an orientation-dependent surface band structure anisotropy.136 However, the different facets coexisting in one particle are considered as isolated surfaces in this explanation, and possible synergistic effects between the coexisting different facets with different surface band structures (i.e., charge transfer between different facets) are not considered. The early studies by Ohno et al. in 2002 observed the selective photodecomposition of Pt and PbO2 particles on different facets of rutile and anatase crystals,73 which were occasionally found in commercial TiO2 powder from Toho Titanium Co. Based on the photoreduction reaction of Pt4+ (Pt4+ + e− → Pt) and photooxidation reaction of Pb2+ (Pb2+ + H2O + h+ → PbO2), it is clearly indicated that rutile {110} and {011} facets provide reduction and oxidation sites, respectively (Figure 38a,c). The obvious separation of reduction and oxidation sites on faceted rutile crystals is attributed to photoexcited electron and hole transfer between {011} and {110} facets, which is driven by the higher electronic energy levels of {011}. Although the selective distribution of photodeposited Pt and PbO2 particles on anatase {101} and {001} facets (Figure 38b,d) is not as obvious as that on different rutile facets, anatase {101} and {001} facets can still be considered to be more reductive and more oxidative, respectively. This claim is supported by the more negative flatband potential of anatase (001) than anatase (101) by around 0.06 V.41 Ohno’s group subsequently conducted a series of studies on the spatial separation of redox sites on the TiO2 polymorphs with different shapes. By monitoring the distribution of photodeposited PbO2 and Pt particles, the lateral {110} and top {111}/{001} facets in rutile nanorods provide reduction and oxidation sites, respectively.246,403,404 For faceted anatase nanocrystals, the reduction and oxidation sites are located on {101} and {001} facets, respectively.106 The lateral {210} and top {212} facets in brookite nanorods provide reduction and

Figure 38. SEM images of (a) a rutile paritlce and (b) an anatase particle on which Pt fine particles were deposited by UV-irradiation in a solution of 1.0 mM H2PtCl6; SEM images of (c) a rutile particle and (d) an anatase particle showing PbO2 deposits, which were loaded on the particles by UVirradiation of the Pt-deposited TiO2 powder in a solution of 0.1 M Pb(NO3)2.73 Reprinted with permission from ref 73. Copyright 2002 Royal Society of Chemistry.

oxidation sites, respectively.334 All these results reveal the generality of the spatial separation of reduction and oxidation sites in TiO2 crystals with appropriate different facets. Consequently, the corresponding spatial separation of photoexcited charge carriers on different facets improves the photocatalytic activities of photocatalysts, as illustrated in Figure 39.

Figure 39. Schematic of the spatial separation of redox sites on anatase crystals with {101} and {001} facets106 and rutile TiO2 particle with {110} and {011} facets. Reprinted with permission from ref 106. Copyright 2009 American Chemical Society.

Some studies from other groups confirm the presence of a spatial separation of redox sites in faceted TiO2 crystals and its great influence on optimizing the activity.162,254,405 It was revealed that in brookite nanosheets {201} facets are oxidation sites, and {210} and {101} facets act as reduction sites,254 which makes the brookite photocatalytically active. Zheng et al. demonstrated the gradually decreased activities in photocatalytic H2 generation and OH radical generation with the 9588

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

reactions. One may argue that the initial nucleation of Pt or PbO2 on the TiO2 surface, or the exposure of the TiO2 surface to an aqueous reaction solution might change the properties of the pristine facets. This argument could be partially addressed by using single-molecule fluorescence spectroscopy to probe the interfacial electron transfer on the surface of an individual TiO2 particle,407−410 as demonstrated by Tachikawa et al. By counting the numbers of fluorescence bursts of HN-BODIPY molecules from nonfluorescent DN-BODIPY molecules by the photoexcited electrons on anatase {001} and {101} facets, the fluorescence bursts were found to be preferentially located on {101} facets.407 Further studies controlling the location of the incident light spot on only {001} or {101} facets clearly suggest unidirectional electron flow from {001} to {101} facets as indicated by Figure 41.410 These results are consistent with the

increase of {001} facet area in hierarchical anatase microspheres,162 whose surface is enclosed by {001} and {101} facets. The improvement is attributed to the efficient separation of charge carriers among the coexisting facets. Liu et al. provided a very straightforward proof of the facet-induced spontaneous separation of photoexcited electrons and holes by comparing the photocatalytic H2 productions of a group of bare anatase nanocrystals with tunable percentages of {001} facet area from 0 to 51.2% (the remaining surface is {101}),405 as shown in Figure 40. The highest activity is achieved for the

Figure 40. Hydrogen production after 6 h irradiation time using TiO2 nanocrystals (NCs) with different exposed {001} facets as photocatalysts.405 Reprinted with permission from ref 405. Copyright 2013 Wiley-VCH. Figure 41. (a) Illustration of the remote photocatalytic reaction on the {101} facets with DN-BODIPY during photoirradiation onto the {001} facets. The irradiated area was limited by a pinhole (the spot diameter is 2 μm on the crystal surface); (b and c), Location of fluorescence bursts on the {001} (blue) and {101} (red) facets. The UV irradiation areas are inside the black circles (diameter 2 μm).410 Reprinted with permission from ref 410. Copyright 2011 American Chemical Society.

sample with 14.9% {001} area as a result of the spatial separation of charge carriers on {001} and {101} facets. In contrast, the crystals with 100% {101} facets show negligible activity due to the appearance of photoexcited electrons and holes on the same surface. The spatial separation of photoexcited charge carriers on different facets does not occur between any two different facets. In principle, staggered band alignments of two facets are necessary to induce charge carrier transfer. Ye et al. demonstrated effective charge carrier transfer between {001} and {101} facets but not between {001} and {010} facets, which leads to the much higher activity of anatase nanosheets with exposed {001}-{101} facets than {001}-{010} facets.301 Roy et al. also showed the importance of the presence of both oxidative {001} and reductive {101} facets in obtaining high photocatalytic activities of anatase nanocrystals with coexisting {010} facets.211 The effect of the spatial separation of redox sites (or charge carriers) on different facets of TiO2 crystals can be used to construct highly efficient heterostructured photocatalysts. Liu et al. constructed photocatalysts by the selective deposition of a Pt particle cocatalyst on the reductive {101} facets of anatase crystals with {101} and {001} facets,405 which shows much higher activities than do anatase crystals where the Pt is nonselectively deposited. Very recently, Guo et al. showed that the selective deposition of ZnO2 on {201} facets as oxidation sites leads to much superior activity of faceted brookite nanorods than does nonselective deposition.406 The spatial separation of reduction and oxidation sites on different facets of TiO2 shown above is revealed by tracking the distribution of Pt and PbO2 deposits on the TiO2 surface, which were formed by photoreduction and photooxidation

conclusion drawn based on the distribution of Pt and PbO2 deposits on the surface. In addition, Tachikawa et al. ranked the reduction reactivity order as {101} > {001} ≥ {010} and oxidation reactivity order as {101} ≈ {001} ≈ {010} based on the numbers of fluorescence bursts appearing on each facet in {101}-{001} anatase crystals and {101}-{010}-{001} anatase crystals.408 The reactivity orders can explain the observed effective charge carrier transfer between {001} and {101} facets but not between {001} and {010} facets.301 By monitoring photogenerated defects in shape-controlled anatase nanocrystals by electron spin resonance spectroscopy, D’Arienzo et al. revealed that in vacuum conditions the {001} facets play a major role in the photocatalytic process by providing oxidation sites, while the {101} facets as the reductive sites are only indirectly involved; in an O2 atmosphere the {101} facets indirectly contribute to the photoxidative processes.411 All these results again suggest crystal facet dependent photocatalytic reduction and oxidation reactions. 3.6. Anisotropic Etching on Crystal Facets

As presented in section 2.7, many hollow TiO2 crystals with clear facets can be obtained by carefully modifying the hydrothermal synthesis parameters. The underlying mechanism of forming such hollow structures is related to the anisotropic etching of crystal facets with different surface energies when 9589

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

HF in reaction media.139 The higher the concentration of TiF4 (HF will be released with the hydrolysis of TiF4) was used, the greater the etching of the {001}. It was shown theoretically that three key steps are involved for a high HF concentration, namely the dissociative adsorption of HF to form a full HF covered surface (Figure 43b), the formation of −TiOF2 groups

exposed to specific reaction media. Facets with a high surface energy are experimentally seen to be more easily etched than those with a low surface energy. For example, rutile {001} facets with surface energy of 28.9 meV a.u.−2 tend to dissolve in a HCl solution under hydrothermal conditions while rutile {110} facets with a surface energy of 15.6 meV a.u.−2 remain unchanged in the same system (Figure 25).235 Only {001} facets of anatase crystals with both {001} and {101} facets, whose surface energies are 0.90 and 0.44 J m−2, respectively, are prone to dissolve in a solution containing fluorine species by hydrothermal treatment. The determination of the central factors controlling the anisotropic etching process and understanding the mechanism at the atomic level are scientifically and technically significant. Yang et al. studied the influence of solvents containing HF on the etching of anatase crystals with around 35% {001} and 65% {101} facet areas (Figure 42a) under hydrothermal

Figure 43. DFT calculated reaction energies and structures for different stages of HF interaction with single crystal anatase TiO2(101) (left) and (001) (right) surfaces. (a) Clean surfaces; (b) full HFcovered surfaces; (c) complete fluorinated surfaces; (d) etched surfaces. All structures are optimized structures.139 Reprinted with permission from ref 139. Copyright 2011 Royal Society of Chemistry.

Figure 42. (a) SEM image of anatase TiO2 single crystals with 35% of {001} facets synthesized with an aqueous TiF4 solution (5.33 × 10−3 M) at 180 °C f3or 14 h. (b and c) Typical SEM images of the samples prepared through hydrothermal treatment of the product with H2O (25 g) containing HF (10 wt %, 0.4 mL) at 180 °C for 6 and 9 h. (d) Corresponding TEM image of an individual particle in panel c.156 Reprinted with permission from ref 156. Copyright 2011 Wiley-VCH.

on the completely fluorinated surface by the replacement of surface exposed −OH groups with F (Figure 43c), and the dissolution of the surface −TiOF2 by further interaction with HF to create surface vacancies (etched surface) (Figure 43d). The first two steps are thermodynamically favorable on both (001) and (101) surfaces. However, the third step is highly dependent on the crystallographic orientation. The calculated dissolution reaction energies for (101) and (001) surfaces are −0.41 and +0.04 eV, respectively, which implies that etching is energetically permitted solely for the {001} surface. This is considered as the intrinsic reason for the anisotropic etching between {101} and {001} facets. Further theoretical calculations suggest that a partially HF covered (001) surface, usually formed at low HF concentrations, is unfavorable for the process (reaction energy: −0.56 eV) of replacing a surface −OH group by F to form a −TiOF2 group compared to the process (reaction energy: −0.78 eV) of further HF adsorption to form a completely covered (001) surface. The formation of −TiOF2 groups is a precondition of etching. This can explain why the etching of a (001) surface occurs only if the HF concentration is higher than a critical value.

conditions.156 It was found that etching of the (001) surface is highly dependent on the solvent. In an aqueous solution containing HF, 6 h hydrothermal treatment led to significant etching traces on (001) and the formation of inverted pyramidal pits at the center of the {001} facets (Figure 42b). A 9 h treatment caused the disappearance of the initial {001} facets as a result of continuous etching on them in the [001] direction (Figure 42c). In contrast, the presence of additional isopropanol in the aqueous solution containing HF obviously retards the etching activity by working as a synergistic stabilizer of {001} facets. Furthermore, uniform etching activity on the whole {001} crystal surface in a mixture of water/isopropanol with HF is dramatically different from the etching that originates from the center of the {001} facets in an aqueous solution with HF. The proposed etching mechanism is the replacement of −OH groups on the crystal surfaces of anatase TiO2 with F ions and the subsequent dissolution of Ti atoms in the form of TiIV complexes, such as TiF62−. Wang et al. experimentally observed the direct dependence of the etching of anatase {001} facets on the concentration of

3.7. Thermal Stability of Faceted TiO2 Crystal

High energy surfaces such as anatase {001} of TiO2 crystals obtained by hydrothermal processes are usually stabilized by surface adsorbed species. The removal of the surface adsorbed 9590

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

nanosheets with 80% {001} facet area as a model starting material by Yang et al.155 Similar to the crystal growth behavior observed when heating the nanosheets in air, the nanosheets under hydrothermal conditions (200 °C in deionized water) can also grow into larger single crystals with a truncated bipyramidal shape through an oriented attachment process along the [001] direction. It was proposed that under hydrothermal coarsening conditions, hydrolysis of Ti−F groups on the surfaces of TiO2 nanosheets will form Ti−OH groups, whose subsequent condensation leads to the combination of adjacent nanosheets in deionized water by the Ti−O−Ti linkages. A further surface fusion process is involved in generating the well faceted crystals. The above crystal growth mechanism is summarized in Figure 45. In addition, the

species is always necessary for fundamental studies and practical applications of faceted TiO2 crystals, though both experimental and theoretical results show that clean high energy surfaces are unstable and tend to reconstruct. Heating the product in air and treating a suspension of TiO2 crystals by ultrasonic or hydrothermal processes are two basic strategies for removing surface adsorbed species, where phase and morphology stabilities are the critical issues. Lv et al. systematically investigated the morphology and phase evolution of surface fluorine-terminated anatase nanosheets with around 88% {001} facet area with the increasing temperature up to 1250 °C.412 It was surprising to find that the nanosheets showed no phase change from anatase to rutile before 1100 °C. The typical phase change temperature of nanosized anatase crystals is around 700 °C. The ultrahigh temperature phase stability is attributed to the positive role of surface adsorbed F− in suppressing the nucleation of the rutile phase at the interface between different anatase nanosheets. The temperature dependent crystallite size variations of anatase TiO2 suggest the crystallite size along the [001] direction increases much faster than that in the [100] direction (Figure 44). Meanwhile, the percentage area of {001}

Figure 45. Proposed growth mechanism of stacked anatase TiO2 nanosheets dominated by {001} facets at their interfacial regions: (a) Stacking of TiO2 nanosheets with F-terminated surfaces; (b) hydrolysis of Ti−F groups on the surfaces and accordingly the formation of Ti−OH; (c) formation of Ti−O−Ti linkages via condensation between Ti−OH on the (001) surfaces of the adjacent TiO2 nanosheets and the resulting elimination of (001) surfaces.155 Reprinted with permission from ref 155. Copyright 2011 American Chemical Society.

Figure 44. Dependence of average crystalline sizes of the photocatalysts along [001] and [100] direction on calcination temperatures, together with the percentages of {001} facets calculated according to geometric shape of Wulff construction, indicating that the photocatalyst has the preferential growth along c-axis ([001] direction) (inset is the corresponding shape simulation).412 Reprinted with permission from ref 412. Copyright 2011 Elsevier.

reaction medium plays a crucial role in the morphology evolution of the nanosheets by controlling the hydrolysis of surface Ti−F groups. Some reaction media including a 1.5 M HCl solution, ethanol, propanol, and butanol, can effectively suppress the hydrolysis of surface Ti−F, and the morphology of anatase nanosheets remains unchanged. In addition, Zhu et al. investigated the influence of the type and density of benzyl groups that decorate the basal plane of {001} dominant anatase nanosheets on their self-assembly into layered structures under hydrothermal conditions.413 The πstacking between benzyl groups in neighboring nanosheets was determined to play a key role in forming layered structures with strong interlayer interactions. The stacked layered structures can form oriented films supported on Ti foil substrates and act as photonic crystals, showing two visible light absorption bands centered at 400 and 640 nm.

rapidly decreases with the temperature increase. All the results indicate the preferential growth of anatase nanosheets along the [001] direction during heating. The fusion of different anatase sheets through adjacent fluorine-free {001} facets at the elevated temperature is the reason for the preferential growth of crystals along the [001]. Different from the case of Lv et al.,412 the nanosheets can grow into anatase TiO2 microsheets with a large percentage of {001} surface.215 The proposed growth mechanism follows a similar fusion process of {001} dominated nanosheets along the [001] direction. However, the presence of gaseous HF as the capping agent, released from NH4F during heating, can effectively stabilize {001}. Due to the partial overlap of some nanosheets by sharing a {001} surface, the crystal can simultaneously grow along the [100] direction so that the micrometer-size sheets are finally obtained. The hydrothermal stability of {001} dominated anatase TiO2 was systematically investigated by using single-crystal anatase

4. MODIFICATION OF THE ELECTRONIC STRUCTURE AND INTERFACIAL PROPERTIES OF FACETED TIO2 For photocatalytic solar energy conversion, faceted TiO2 crystals suffer from little visible light absorption as do common TiO2 crystals because of their large bandgap (anatase, 3.2 eV; rutile, 3.0 eV; brookite, 3.2 eV), and rapid recombination of photoexcited charge carriers. The modification of the electronic 9591

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

structure of faceted TiO2 to fully absorb visible light and promote the separation and surface transfer of photoexcited charge carriers in faceted TiO2 is greatly needed. The widely used strategy for modifying the electronic structures of widebandgap semiconductors is to incorporate heteroatoms, namely doping, which has shown attractive potential in the modification of the electronic structure of common TiO2 crystals without specific facets as summarized in several reviews. Moreover, the modified electronic structure also affects other applications of TiO2 such as lithium ion storage by improving electronic conductivity. The challenge of obtaining the doped TiO2 crystal with tailored facets is how to introduce dopants in the crystal without destroying the growth environment needed to control facet growth for in situ preparation methods. Creating intrinsic defects such as oxygen vacancies or related Ti3+ ions can also significantly modify the electronic structure by introducing localized states in the band gap so that many electronic properties such as the light absorption range, density of charge carriers and conductivity can be significantly changed. Furthermore, defects in the TiO2 surface usually act as active sites for reactant molecules/ions. Introducing defects in TiO2 crystals with tailored facets certainly creates new opportunities for modulating the functionalities of TiO2. Heterostructuring has attracted increasing interest due to its effectiveness in improving the performance of TiO2 in various applications. For example, coupling TiO2 crystals in different shapes with carbon nanostructures (nanotubes, graphene, and quantum dots) can substantially improve both lithium storage capability and photocatalytic activity compared to bare TiO2 crystals. In addition to the added properties of the second material, the properties of interface between it and the TiO2 in the constructed heterostructure play an important role in determining the performance. The rapid development of TiO2 crystals with tailored facets provides a great possibility to tune the interfacial properties of TiO2 based heterostructures by controlling the exposure of different facets of TiO2.

Figure 46. UV−visible absorption spectrum of nitrogen doped anatase TiO2 sheets; the insets in the upper right and lower left corners are SEM image and optical photo of nitrogen doped anatase TiO2 sheets, respectively.105 Reprinted with permission from ref 105. Copyright 2009 American Chemical Society.

doped TiO2 sheets is yellow due to the additional visible light absorption band ranging from 400 to 570 nm (Figure 46). The origin of the visible light absorption is associated with localized states in the band gap formed by the nitrogen dopant. N-doped anatase sheets with a loaded Pt cocatalyst can photocatalytically generate hydrogen from a mixture of water and methanol under visible light. To decrease the particle size of such N doped TiO2 sheets, Xiang et al. developed a modified solvothermal production method by treating TiN in a HNO3−HF ethanol solution, which leads to the formation of N doped TiO2 nanosheets with ca. 67% {001} facet area.150 The nanosheets with a side length of ca. 80 nm and a ca. 20 nm thickness have a similar visible light absorption band to that of microsize N doped sheets. As a result of the remarkably decreased particle size, the N doped nanosheets show a much higher photocatalytic H2 production activity in visible light than do N doped TiO2 microcrystals with {001} facets (ca. 60% area) by a factor of 4.1. Nanoscaled N doped anatase TiO2 with around 70% {001} and 30% {101} facet areas can also be prepared by the solvothermal treatment of TiN in acidic NaBH4.199 TiN was also used to prepare codoped anatase TiO2 sheets with dominant {001} facets. By the hydrothermal treatment of a thermal sprayed TiN/Ti coating with a HF aqueous solution containing chromium powder, Wu et al. reported N−F−Crdoped anatase TiO2 microspheres with a nearly all-(001) surface.143 Furthermore, a band-to-band like visible light absorption band with its absorption edge at around 1.75 eV is formed as a result of a more hom*ogeneous incorporation of Cr ions in the resulting TiO2 together with the N/F dopants. Mo and N codoped anatase sheets with dominant {001} facets were prepared by hydrotheramlly treating TiN and MoO3 in the mixture solution of HF and HNO3.194 The generality of the above synthesis strategy has been validated by the successful preparation of S-doped,112 Bdoped,219 C-doped,158 Si-doped,231 even C/N-doped anatase TiO2 crystals with well-developed facets with TiS2, TiB2, TiC, TiSi2, and TiCN as the respective precursors. The realization of nonmetal doping without affecting the formation of well developed facets in this strategy can be rationalized as follows: (1) Some Ti-X bonds, which can be left in the TiO2 crystals formed during the hydrolysis of the TiX precursor, are

4.1. Doping

Transition metal doping (i.e., Cr3+, Fe3+, and Cu2+) has been extensively conducted in TiO2 since the 1980s.20,414,415 Although doping is effective in increasing the visible light absorption of TiO2 photocatalysts, doped materials always suffer from a thermal instability, and an increased number of recombination centers of photoexcited electrons and holes. Nonmetal doping (N, C, B, S, P, and I)416−421 has emerged as an alternative approach to modify the electronic structure of TiO2 and also other metal oxides since the first report of Ndoped TiO2 by Asahi et al. in 2001.61 Various preparation methods have been developed to produce nonmetal doped TiO2. The in situ methods usually use compounds containing dopant(s) as precursors, for example, hexamine (C6H12N4) or NH3 for N doping. The presence of such compounds in the reaction media inevitably destroys the nucleation and growth of TiO2 with tailored facets, to some extent, so that it is a great challenge to prepare nonmetal doped TiO2 with well-tailored facets. To solve this problem, Liu et al. developed a unique hydrothermal route where a class of compounds TiX (X = nonmetal dopant) containing both Ti and dopant were used as the only precursor.105 Following this idea, micrometer-size N doped TiO2 sheets with major {001} and minor {101} facets (inset in Figure 46) were prepared by hydrothermally treating TiN crystals in a HF aqueous solution.105 The product of N 9592

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

media.152,239 Although heteroatoms can be effectively doped into TiO2 crystals, the shape of the doped crystals obtained in most cases is not as well-defined as the well faceted crystals prepared with TiX as the precursor. This is basically because the presence of an amount of such compounds seriously destroys the environment required for the growth of wellfaceted TiO2 crystals. An alternative synthesis method is to use the insoluble precursors doped with the targeted heteroatoms for the synthesis of doped TiO2 crystals. Zhang et al. reported N/Ni-doped anatase crystals with {001} facets by hydrothermally treating N/Ni doped H2Ti3O7 nanotubes in a HF solution. N/Ni doping leads to a bandgap narrowing from the original 3.1 to 2.9 eV by lowering the conduction band minimum with the involved Ni 3d states, and an additional visible light absorption band related to N 2p states is formed in the band gap.195 Lin et al. reported that Te doped anatase nanorods can be obtained by treating Te doped titanate nanotubes under hydrothermal conductions (deionized water, 200 °C).276 Te doped anatase nanorods are clearly faceted with the lateral major {100} facets (Figure 47A). EDX mapping

responsible for the corresponding nonmetal doping. (2) The total amount of byproducts from the hydrolysis of TiX could be very low and exert a weak influence on the nucleation and growth of crystals. In some cases, the byproducts are even gaseous and have no influence on the crystal growth. All these doped anatase crystals with specific facets except the B-doped one219 have an increased visible light absorption band as a result of the dopants introduced in the crystals. In most cases, however, only a shoulder-like visible light absorption band is formed, probably due to the inhom*ogeneous distribution of dopants within the particles. A redshift of the intrinsic absorption edge is realized in C doped anatase sheets with ca. 58% {001} and 42% {101} facet areas, prepared by hydrothermally treating TiC powder in the mixed solution of HNO3 and HF (HNO3 acting as the solvent of TiC, and HF acting as the stabilizer of anatase {001}).158 The bandgap is narrowed from the original 3.18 to 2.80 eV with the incorporation of the substitutional C for lattice O. In visible light C-doped anatase sheets, as stable photocatalysts, exhibit strong photocatalytic activity in generating OH radicals and decomposing methylene blue and phenol. Different from other doped TiO2 crystals with the dopant located in the O sites, Wu et al. reported anatase Ti0.89Si0.11O2 crystals with high-index {201} facets and high-energy {001} facets,231 where Si dopants take the position of Ti lattice atoms. The crystals have a strong absorption in the whole visible light range. Theoretical calculations suggest that the (201) surface can induce spontaneous dissociation of water molecules as does the (001) surface but has a higher stability. Furthermore, Si doping does not affect such water dissociation. Boron-doped anatase microspheres were prepared by the hydrolysis of TiB2 in a HCl solution containing Na2SO4 under hydrothermal conditions.219 Their surface consists of nanosize truncated pyramids with top major {001} and lateral minor {101} facets. The substitution of B for lattice O occurs in the core of as-prepared microspheres. After thermal treatment, a boron gradient with a maximum content at the outer surface is formed in the shell as a result of thermal diffusion of boron from the core to the shell along the [001] direction (The shell thickness is around 50 nm.). A conversion from substitutional boron (Bδ−) to interstitial boron (Bσ+, σ ≤ 3) occurs upon thermal treatment because the former is much less stable than the latter. Neither bandgap narrowing nor an obvious visible light absorption band is realized with the incorporation of B in the microspheres regardless of the chemical state of the boron. This is consistent with the theoretical results that both interstitial and substitutional boron lead to no change in the intrinsic bandgap of anatase TiO2, and only a very low density of localized states of B is formed in the band gap.422 The remarkable electronic structure change produced by the incorporation of interstitial B in the shell is a downward shift of valence and conduction band edges (by 0.26 eV at the outer surface) with respect to the boron-free shell. The origin of the downward shift is attributed to the band bending effect caused by the extra electrons from Ti3+. The formation of Ti3+ is generated by contributing valence electrons of the interstitial boron to neighboring Ti4+. Such an electronic structure modification provides a platform based on {001} facets to investigate the effect of heteroatom on modulating photocatalytic reduction and oxidation reaction preferences. Other studies have attempted to prepare nonmetal doped TiO2 crystals with tailored facets by directly introducing compounds containing the targeted dopant in the reaction

Figure 47. (A) SEM image of Te-TiO2 nanorods. (B) HAADFSTEM-EDX mapping of one representative Te-TiO2 nanorod for Ti, O, and Te elements, separately.276 Reprinted with permission from ref 276. Copyright 2013 Wiley-VCH.

confirms the good dispersion of Te dopants throughout the nanorod (Figure 47B). The doping of Te into the TiO2 lattice induces a shift of the maximum absorption wavelength from 300 to 350 nm. The absorption edge covers the whole visible light range. Doping of Se in Te doped TiO2 nanorods further improves absorption in the visible region. As a consequence, both Te doped and Se/Te-doped TiO2 nanorods exhibit higher antibacterial activities against E. coli and S. aureus than P25 TiO2 when activated by visible light. Using a doped titanate as the precursor to obtain doped TiO2 crystals with well-developed facets in the above two cases is essentially consistent with that of using TiX as precursor. The key to perfectly integrating the doping with crystal facet engineering is to minimize the influence of dopant-related precursors on the reaction environment for the nucleation and growth of faceted TiO2 crystals. Two studies used TiOF2 cubes as the precursor to prepare doped crystals with tailored facets.163,293 By heating TiOF2 cubes in a gaseous NH3 atmosphere, N/F codoped anatase sheets with dominant {001} facets were prepared by Zong et al.163 The resulting sample has an extended absorption edge up to 570−580 nm and high absorbance. Furthermore, such sheets show an excellent oxygen evolution rate from the photocatalytic water oxidation reaction under visible light (500 μmol h−1 g−1 for the sample heated at 773 K). However, only negligible 9593

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

oxygen was generated with N-doped TiO2. This might be associated with the unique electronic structure modification produced by N/F codoping. in addtion, N-doped anatase TiO2 (001) films can be grown on LaAlO3 (001) by plasma-assisted molecular beam epitaxy. The molecular formula of the film is determined to be TiO1.998N0.002.423 The substitution of N for lattice O results in a new low energy light absorption band from 2.5 to 3.4 eV, which corresponds to the excitation from Nderived states above the valence band into the conduction band. Ex situ doping process was also used to modify the electronic structure of faceted TiO2 crystals. Compared to in situ doping, the remarkable advantage of ex situ doping should be its versatility for various TiO2 crystals by simply heating the crystals in appropriate gaseous atmospheres or solid media. However, it is challenging to conduct effective doping for faceted crystals. This is because the well-developed facets, particularly on micrometer-size crystals, usually have a high crystallinity and low concentration of defects so that the incorporation of dopants is difficult. Although nanosize faceted crystals with a large surface area and high concentration of defects may make the incorporation of dopants easy, heating itself during doping may cause some unexpected changes to the shapes of crystals, in particular, those with high energy facets. So far, several studies have been conducted to prepare nonmetal doped anatase sheets with dominant {001} facets.149,424 For example, Xiang et al. prepared N/S-codoped anatase crystals with {001} facets by heating a mixture of {001} dominant anatase nanosheets and thiourea at 500 °C in air.149 The introduced interstitial N and major substitutional S for lattice Ti causes an additional visible light absorption band up to 750 nm by forming isolated N 2p states and hybridization states of S 3s and O 2p in the band gap. However, compared to the pristine ca. 6 nm thick nanosheets, the thickness of the N/ S-doped sheets increase to 20−25 nm probably as a result of the fusion and restructuring of several original TiO2 nanosheets in close-contact during heating at 500 °C, which may impair their photocatalytic activity. An unusual red anatase TiO2 was obtained by heating the earlier-described boron-containing anatase microspheres in a gaseous NH3 atmosphere at 600 °C.425 They are composed of crystals with mainly {001} facets and have an interstitial boron gradient within a ∼50 nm thick shell. The basic shape of the anatase microspheres is retained but the original sharp edges and corners of the truncated pyramids (Figure 7), the unit of the microsphere surface, become obtuse after the treatment (Figure 48a). This is largely due to the high energy of the sharp edges and corners themselves, which become blunted in order to decrease the system energy during heating. UV−visible absorption spectra in Figure 48c show a remarkably high absorbance in the whole visible light region, which is consistent with the red color of the sample (Figure 48b). At the atomic level the red color is attributed to the substantially improved amount (N/Ti+O = 4.1 at%) of substitutional N dopant for lattice O with the assistance of interstitial boron in TiO2, which can weaken the surrounding Ti−O bonds for easy substitution of N for lattice O. The origin at the electronic structure level is the upward shift of the valence band maximum by forming additional electronic states of main N 2p above the original valence band. The bandgap of B/N codoped TiO2 is theoretically predicted to linearly decrease with increasing nitrogen dopant. Experimentally, a nitrogen gradient with the highest level on the surface exists in the ca. 50 nm thick shell of

Figure 48. (a) SEM image of a red TiO2 microsphere; (b) optical photograph of the prepared red TiO2 sample; (c) UV−visible absorption spectra of the white TiO2 and red TiO2; (d) schematic of band structures of the red TiO2 with a bandgap gradient.425 Reprinted with permission from ref 425. Copyright 2012 Wiley-VCH.

the red TiO2 microspheres. A bandgap gradient varying from 1.94 eV on the surface to 3.22 eV in the core is therefore inferred (Figure 48d). A photoanode fabricated with the red TiO2 microspheres for photoelectrochemical water splitting gives a photocurrent under the irradiation up to around a wavelength of 700 nm. 4.2. Creating Intrinsic Defects

Oxygen vacancies and related Ti3+ ions are common intrinsic defects in n-type TiO2 crystals. Many previous studies have focused on the important topic of changing the electronic structure by the introduction of oxygen vacancies and applications of TiO2 crystals without specific facets.426−431 The dominant mechanism involved in modifying the electronic structure by this method has been shown, both experimentally and theoretically, to be the introduction of some localized states at 0.75−1.18 eV below the conduction band.432 With increasing number of oxygen vacancies the localized states are widened into a sub-band, which can overlap the conduction band minimum if the concentration of oxygen vacancies is high enough.431 With the emergence of faceted TiO2 crystals synthesized by wet chemical methods, studies began to use oxygen vacancies to further modulate their properties. Liu et al. synthesized anatase sheets with a large percentage area of {001} facets and a dark blue color (the insets in Figure 49) by the acidic hydrolysis of TiB2 under hydrothermal conditions (180 °C, HF solution).104 The blue color is associated with oxygen vacancies and/or related Ti3+ in the TiO2, which are generated by reducing Ti4+ with the released H2 from the hydrolysis of metallic TiB2. Compared to anatase TiO2 sheets with no oxygen deficiency, an additional strong absorption band covering the visible light and near-infrared range (vis−NIR) appears in oxygen-deficient anatase sheets (Figure 49). Such an absorption band is attributed to the low-energy photon and/or thermal excitations of trapped electrons in the localized states of the defects just below the conduction band minimum to the conduction band, or the excitation of free electrons in the conduction band. The oxygen deficiency defects plus surface fluorine species exert a great influence on the geometric 9594

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

tetragonal prisms by heating the sample in air results in the loss of visible light photocatalytic activity, suggesting the essential role of Ti3+ in contributing to the observed visible light activity. It is noted that Ti3+ can also be introduced in situ in faceted TiO2 crystals in the synthesis methods without the involvement of reductive gases. The dependence of oxygen vacancy and Ti3+ formation on the synthesis temperature of the hydrothermal conditions (Ti(OC4H9)4, HF) was reported.185 Anatase TiO2 nanosheets with dominant {001} facets synthesized at 180 °C, a widely used temperature for the growth of faceted TiO2 crystals, contain few oxygen vacancies and Ti3+. An increased temperature of higher than 200 °C, particularly 240 °C leads to the generation of Ti3+ in the anatase nanosheets so that a few VIS-NIR absorption band is formed in the absorption spectra. Nanosheets with Ti3+ show improved photocatalytic activity in decomposing Rhodamine B and generating OH radicals under visible light. No clear mechanism is proposed for the formation of oxygen vacancies and Ti3+ in this case. Wang et al. reported Ti3+ doped TiO2 crystals in mesoporous nanosheets with dominant {001} facets prepared by supercritical treatment (240−280 °C) of a precursor obtained from the sol−gel hydrolysis of mixed Ti(n-OCH4H9)4 and TiF4.214 The generation of Ti3+ species is attributed to the reduction of Ti4+ by organic substances and/or residues under supercritical conditions. The obtained Ti3+ doped TiO2 with visible light absorption shows the ability of simultaneous pollutant degradation (methylene blue, methylene orange, Rhodamine B, and 4-chlorophenol) and H2 production under visible light. Gordon et al. reported the nonaqueous synthesis of anatase nanocrystals with the tunable ratios of {001}/{101} that contained oxygen vacancies.88 This was performed using standard Schlenk line techniques under a nitrogen atmosphere at 290 °C. The presence of TiF4 in this case is found to play an essential role in generating oxygen vacancies in the resulting blue TiO2 nanocrystals that have a strong VIS-NIR absorption. It was proposed that the very strong bond between Ti and F (the only bond to titanium stronger than Ti−O) facilitates oxygen vacancy formation. The blue coloration is stable for months without noticeable change but disappears after heating at 300 °C in air. The measurements of photocatalytic hydrogen generation from a mixed methanol/water solution under simulated solar light show an increased activity of Pt-loaded nanocrystals with a decreased percentage of {001} area. Besides the clearly faceted single crystals with oxygen vacancies and Ti3+ discussed above, mesocrystals containing Ti3+ with tailored facets were also developed. Chen et al. synthesized anatase mesocrystals with both tunable facets and Ti3+ concentration by the solvothermal treatment of a mixture of formic acid and titanium isopropoxide at 160 °C.95 The resulting submicrometer-size mesocrystals with a blue color consist of nanocrystal subunits with an average crystal size of 38 nm, and have the shape of a truncated bipyramid with lateral {101} and top {001} facets. An increased reaction time leads to the gradual increase of the {101} percent area of the mesocrystals as a result of the preferential assembly of the subunits along the [001] direction. Importantly, the concentration of Ti3+ in the mesocrystals simultaneously increases with the increased area of {101} facets. The formation mechanism of Ti3+ is the in situ reduction of Ti4+ by formic acid, which acts as a reducing agent in an anaerobic system. The mesocrystals containing Ti3+ with a higher percentage area of {101} facets possess higher activities in both the photo-oxidation of terephthalic acid and the photoreduction of nitrosobenzene

Figure 49. (a) UV−vis absorption spectra of (a) the oxygen-deficient anatase TiO2 sheets and (b) anatase TiO2 sheets free of oxygen deficiency by calcining (a) at 600 °C in air.104 Reprinted with permission from ref 104. Copyright 2009 American Chemical Society.

structure of the surface layer as indicated by the appearance of two new Raman modes at 155 and 171 cm−1. Compared to 1 wt % Pt-loaded oxygen vacancy-free anatase sheets, 1 wt % Ptloaded oxygen-deficient anatase sheets show an improved photocatalytic H2 evolution from a mixture of water/methanol by a factor of 1.75. The improvement is attributed to both the surface reconstruction and the strengthened interaction between Pt and TiO2. However, no visible light photocatalytic activity is claimed in this work. Some other in situ methods probably involving the release of reductive gas H2, were developed to prepare TiO2 containing defects with tailored facets. Under hydrothermal conditions (TiCl3, NaF, 200 °C), pseudoisotropic mosaic spheres of anatase almost completely covered with (001) facets with a steelblue color were prepared.145 The product in this case has a similar VIS-NIR absorption band to that of the product from TiB2. It is proposed that the color may originate from surface defects produced by F adsorption. However, there are other possible reasors for the formation of blue color, namely, no oxidation of a small amount of Ti3+ to Ti4+ or the reduction of Ti4+ to Ti3+ by reducting H2 gas released from the hydrolysis of TiCl3. In addition, the relatively high synthesis temperature could be also favorable for the formation of oxygen vancancies.185 Zuo et al. synthesized Ti3+ doped rutile TiO2 tetragonal prisms by hydrothermally treating Ti powder in a HCl solution.433 The tetragonal prisms are enclosed by lateral {110} and top {111} facets, which are respectively prone to collect photoexcited electrons and holes. The EPR spectrum with a strong signal at g = 1.97 suggests the presence of Ti3+ in the bulk instead of at the surface of the TiO2 (surface Ti3+ would be reduced to O2− by adsorbing atmospheric O2 and will show an EPR signal at g = 2.02). The generation of Ti3+ is believed to result from the reduction of Ti4+ by the H2 released in the reaction between the Ti powder and HCl. Ti3+ with a concentration of ca. 4.95 μmolg−1 causes a visible light absorption band. It is remarkable that such crystals after loading with 1 wt % Pt by photodeposition have the ability of stable photocatalytic hydrogen generation (18.1 μmol h−1/0.1 g) from a mixture of H2O/methanol under visible light (λ > 400 nm). The effective spatial separation induced by the different facets of the prism is confirmed by the fact that Pt nanoparticles accumulated selectively on reductive {110}, and no Pt was found on oxidative {111}. This suggests that Ti3+ does not affect the spatial separation of photoexcited charge carriers on different facets. The removal of Ti3+ from the 9595

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

under UV−visible light. The removal of Ti3+ lowers the activities but does not change the reactivity order of the mesocrystals, which is consistent with the results derived from single crystals.88,433 The superior activities of mesocrystals containing Ti3+ are attributed to both the downward shift of the valence band maximum as a result of band bending effects and the defect states introduced in the band gap, which facilitate the generation of strongly reductive electrons. Like the ex situ preparation of nonmetal-doped TiO2 crystals with tailored facets, ex situ preparation methods of heating the samples in an atmosphere of H2 were also explored. Heating fluorine-terminated anatase nanosheets with dominant {001} facets in a high pressure H2 atmosphere (10 bar) also results in a color change of the product from white to blue due to a large number of Ti3+ and oxygen vacancies introduced.186 The hydrogenated product gives a much improved activity in decomposing methylene blue under visible light. Wang et al. reported that hydrogenated N/F codoped anatase sheets show a superior photocatalytic activity in decomposing RhB to that obtained using hydrogenated F doped anatase sheets under visible light.184 This superiority is related to the formation of a wide defect energy state band instead of some isolated defect energy level. Such a band not only narrows the band gap and significantly enhances light absorption but also facilitates the photo carrier’ migration. In addition, the formation of oxygen vacancies on the surface of the anatase nanosheets with dominant {001} facets in the selective oxidation of benzyl alcohol has been reported.208

Figure 50. TEM images of the as-prepared core−shell Au@TiO2 nanoparticles (a) and individual particle image (b).435 Reprinted with permission from ref 435. Copyright 2009 American Chemical Society.

direction into a wedge-shaped morphology. This synthesis strategy can be extended to grow core−shell heterostructures of Pt-TiO2, Ag-TiO2, and metal ion (Fe3+ or Cr3+) doped AuTiO2.435,436 The resulting Au-TiO2 heterostructure exhibits superior photocatalytic removal of acetaldehyde to P25 TiO2 under both UV and visible light as a result of the unique structure. Heterostructures of TiO2 nanorod supported Ag nanoparticles were also prepared by a photocatalytic synthesis strategy or in situ reduction of silver acetate and further growth of Ag on TiO2.437,451 The Ag-anatase TiO2 heterostructures as electron transport layers in the fabrication of organic solar cells showed notable improvement in power conversion efficiency than pure TiO2 nanorods due to improved charge separation and transfer by the presence of Ag nanoparticles.437 4.3.2. Carbon−TiO2 Heterostructures. Studies on graphene, a one-atom thick two-dimensional crystal, have seen an explosive increase in recent years due to its unusual properties (high carrier mobility, high electronic conductivity, high light transparency, etc.) and expected attractive applications.452−454 The fabrication of heterostructures of TiO2-graphene has been actively pursued in both lithium storage455,456 and photocatalytic solar energy conversion.457,458 Graphene plays a key role in improving the electronic conductivity of the TiO2 electrodes of lithium batteries and promoting the separation of charge carriers of TiO2 photocatalysts. For example, nanostructured TiO2-graphene hybrid materials show more than twice the lithium insertion/extraction specific capacity at high discharge/charge rates than does the pure TiO2 phase.459 This is a result of the increased electrode conductivity in the presence of a percolating graphene network embedded in the metal oxide electrodes. To use the advantages of both rapid lithium diffusion along the [001] direction of anatase TiO2 and high electronic conductivity of graphene, Ding et al. designed and synthesized a unique hybrid structure by directly growing ultrathin anatase TiO2 nanosheets with exposed {001} faces on a graphene support for fast lithium storage.460 Both ex situ and in situ fabrications of TiO2-graphene hybrids have been conducted. Among them, several groups reported the in situ synthesis of anatase crystals with tailored facets, particularly {001} dominant anatase nanosheets, on graphene by the hydrolysis of Ti precursors in the solutions containing fluorine ions in the presence of graphene oxide.182,187,442−444,461 As an example, Jiang et al. synthesized graphene supported TiO2 nanoparticles with {001} facets (Figure 51), 461 which gives an improved activity in decomposing methylene blue under UV light compared to pure TiO2. The retarded recombination and prolonged life of photoexcited charge carriers are responsible for the improve-

4.3. Heterostructuring

Faceted TiO2 crystals have been hybridized with various substrates including noble metals,434−437 carbon nanostructures,187,438−446 and semiconductors137,440,447−450 in order to improve the functionalities of TiO2. In this section, some typical examples of faceted TiO2 involved heterostructures will be presented. 4.3.1. Metal−TiO2 Heterostructures. Metal−semiconductor heterostructures are an important class of materials attracting extensive interest in exploring their unusual physicochemical properties. Noble metal-TiO2 heterostructure photocatalysts are popular due to the effective charge carrier transfer from TiO2 to metal through the interface, which retards the recombination of photoexcited electrons and holes in TiO2 and thus improves photocatalytic activity. The conventional way to do this is to load noble nanoparticles on various TiO2 substrates. As already discussed in section 3.5, the noble metal nanoparticles can be selectively deposited on the reductive faces of faceted TiO2 crystals. It is, however, considered that the exposure of these nanoparticles to the surrounding media might suffer from corrosion or dissolution of metal nanoparticles in long-term practical applications.435 The heterostructure of a metal-TiO2 core−shell is demonstrated to be a candidate to solve this problem. By the hydrolysis of TiF4 in the presence of preformed Au nanoparticles under hydrothermal conditions, Wu et al. synthesized core−shell Au-TiO2 nanoparticles with a truncated wedge-shaped TiO2 morphology, whose lateral and top facets are recognized as {101} and {004} (Figure 50),435 respectively. A segmenting process within radial wedge-shaped shells during TiO2 shell formation is revealed, where adjacent to the local region of Au nanocrystals, the (101) planes of TiO2 crystals first grow epitaxially to stabilize the Au-TiO2 heterojunction, then gradually change to preferential growth along the [001] 9596

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

spheres with major {001} and {101} facets have a higher photocatalytic degradation efficiency for gaseous styrene than does simple TiO2.446 4.3.3. Semiconductor−TiO2 Heterostructures. Many heterostructures of faceted TiO2−semiconductors (CdS, Cu2O, and Bi2O3) have also been used.137,440,448−450 The CdS−TiO2 heterostructure is a widely investigated system due to the narrow bandgap (around 2.3 eV) for wide visible light absorption and high flat band potential for the injection of photoexcited electrons from CdS to TiO2. Strong anchoring of CdS nanoparticles on TiO2 substrates is essential for the effective electron injection. Considering the large percentage of unsaturated atoms on high energy facets such as anatase {001}, the interaction between the anchored nanoparticles and the high energy facets could be strong. Qi et al. prepared CdSsensitized Pt/anatase-TiO2 nanosheets with (001) facets,448 which give a much improved stable photocatalytic hydrogen production from a lactic acid aqueous solution under both UV and visible light in comparison with Pt/TiO2 nanosheets. The advantage of {001} facets in constructing efficient CdS-TiO2 (anatase) heterostructures was further validated by the demonstrated superior photocatalytic activity and stability of CdS-TiO2 with dominant {001} to that with dominant {101}.450 Using one-pot synthesis methods, ultrasmall particles of targeted semiconductors can be formed in situ on TiO2 crystals with tailored facets. Liu et al. reported the in situ loading of ultrasmall Cu2O particles on anatase nanosheets by the simultaneous hydrolysis of Cu(Ac)2 and Ti(OBu)4 in the presence of HF.449 Cu2O particles with narrow size distribution and tunable sizes from 1.5 to 3 nm were mainly located on {001} facets. As a result of the induced visible light absorption by Cu2O (bandgap, 2.0 eV; p-type) and the promoted charge carrier transfer from Cu2O to TiO2 by the p−n junction formed at the interface between TiO2 and Cu2O, the Cu2O-TiO2 heterostructure with an optimized particle size (2 nm) of Cu2O exhibits almost 3 times the visible light photocatalytic activity in decomposing phenol than do N doped anatase nanosheets with major {001} facets. Anatase nanosheets with dominant {001} facets decorated with Bi2O3 quantum dots (around 2 nm in diameter), were synthesized by a similar one-pot method.440 They show a significantly improved activity in the decomposition of RhB under visible light in comparison with undecorated anatase because of the decreased recombination of charge carriers by interface charge carrier transfer and increased visible light absorption. The loading of Bi2O3decorated anatase on graphene can further improve the activity. Different from the faceted TiO2−CdS/Cu2O/Bi2O3 heterostructures presented above, where a sensitization mechanism dominates the visible light activity of TiO2, Iwaszuk et al. theoretically studied the effects of tin oxide nanoclusters on changing the electronic structure of the anatase (001) surface itself.462 The change in the electronic structure was found to depend on the tin oxidation state. Sn(II)O modification of the anatase (001) surface narrows the band gap of TiO2 by an upward shift of the valence band. This arises from the presence of the Sn 5s−O 2p lone pair in the nanocluster, which is also beneficial for charge separation compared to the localized O 2p states. In contrast, a Sn(IV)O2 nanocluster with no Sn-derived lone pair is inactive in modifying the bandgap of anatase. The significance of these findings is the indication of the feasibility of narrowing the bandgap and promoting charge separation by

Figure 51. (a) TEM images of the TGCS-1, showing the thin graphene sheet and uniform distribution of high reactive TiO2. (b) Part of enlarged TEM images of TGCS-1 is corresponding to the black circle part in panel a.461 Reprinted with permission from ref 461. Copyright 2011 American Chemical Society.

ment. The fact that the conduction band edge of TiO2 is higher than the Fermi level of graphene facilities the transfer of photoexcited electrons from TiO2 to graphene. In addition, the strong interfacial contact between TiO2 and graphene increases the visible light absorption of TiO2 by forming Ti−O−C or Ti−C bonds, and this could be particularly effective for the case of high energy facets with a large percentage of unsaturated atoms. This is indicated by the reported visible light activity in decomposing methylene blue with a {001} dominant anatase nanosheet-graphene hybrid. Some other carbon nanostructures with supported TiO2 with tailored facets are also intriguing. It is theoretically predicted that anatase (001)-graphdiyne hybrids possess superior charge separation and oxidation properties, having the longest lifetimes of photoexcited carriers among all the two-dimensional hybrids containing TiO2 with different facets.441 Experimental results further confirm that for the photocatalytic degradation of methylene blue, the rate constant of TiO2 (001)-graphdiyne hybrid is 1.27 ± 0.12 times that of the TiO2 (001)-graphene. Guo et al. reported the direct growth of the arrays of anatase nanosheets with 40−90% {001} area on carbon fibers (Figure 52A) for the efficient photocatalytic decomposition of

Figure 52. (A) SEM top-view image of uniform and dense arrays of TiO2 nanosheets (NSs) grown on carbon fibers at 453 K for 4 h. (B) A single TiO2 NS grown on a carbon fiber. The (001) facet is labeled.445 Reprinted with permission from ref 445. Copyright 2012 Wiley-VCH.

methylene orange.445 The nanosheet with two lateral {001} facets is perpendicular to the fiber (Figure 52B). In comparison with TiO2 nanosheets grown on a FTO glass substrate, the arrays on carbon fibers show an improved activity by a factor of more than three as a result of a large surface area, exposed high energy surfaces, and electron transfer from TiO2 to carbon. In view of practical applications, the integration of the nanosheets on a micrometer-size, flexible, conductive and stable carbon fiber for easy recycling is definitely significant. Multiwall carbon nanotubes wrapped with submicrometer size anatase TiO2 9597

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

of the hierarchical photoanode is considered due to the integrated effects of more efficient light energy utilization, a larger concentration of photocarriers, and better separation/ mobility of electrons and holes for (001)A/R-TiO2 as indicated in Figure 53c. Besides the heterostructures of rutile-anatase TiO2, the anatase-TiO2(B) heterostructure with a type-II band alignment (TiO2(B) has a similar bandgap to anatase, while its conduction band edge is above that of anatase.) has been also demonstrated to be effective in improving photocatalytic activity of TiO2.473−475 Yang et al. prepared the heterostructure of TiO2(B) nanofiber with a shell of anatase nanocrystals via two consecutive partial phase transformation processes with titanate nanofiber as the starting material.473 Anatase (001) planes and (100) planes of TiO2(B) are coherently connected together to form a stable interface of the heterostructures, which is crucial for the effective charge transfer between two phases. Magnetic-semiconductor heterostructures are a class of intriguing materials with multifunctional capabilities. Buonsanti et al. reported a group of asymmetric binary nanocrystals, comprising one c-axis elongated TiO2 (anatase or brookite) section and one FexOy spherical domain attached together, which were synthesized by heterogeneous nucleation of iron oxide onto the longitudinal facets of TiO2 nanorods as seeds.476−478 It is revealed that the interfacial planes of the heterodimers of anatase TiO2 nanorods-γ-Fe2O3, identified as the {040} of maghemite and the {040} of anatase, are deformed to match with each other so that a true inorganic junction between two lattices is formed.476 In the case of the heterodimers of one c-axis enlongated brookite TiO2 nanorod with richly faceted termination (longitudinal {210}, {220} and {100} facets; basal high-index facets, {001} facets) and one iron oxide spherical domain, a single spherical domain of iron oxide can be epitaxially grown at either one apex or any location along their longitudinal sidewalls, depending on the concentration of iron oxide precursor used.477

modifying the surface with appropriate nanocluster instead of introducing foreign atoms in the bulk of photocatalysts. An anatase−rutile mixed TiO2 photocatalyst has long been actively pursued due to its superior activity to photocatalysts consisting of only one phase. It is widely accepted that the superiority of the mixed phases stems from photoexcited electron transfer between them. However, controversy on the transfer direction still exists.463−465 For example, Kawahara et al. showed that the photoexcited electrons transfer from anatase to rutile in patterned anatase/rutile bilayer-type junction photocatalysts.463 On the contrary, Hurum et al. explained the excellent activity of Degussa P25 with around 80% anatase and 20% rutile as a result of the allowed rapid electron transfer from rutile to anatase.464 Very recently, using a combination of materials simulation techniques and X-ray photoemission experiments, Scanlon et al. revised the conventional understanding of the band alignment between anatase and rutile, and showed a higher work function of anatase than rutile by around 0.4 eV,465 which supports photoexcited conduction electron transfer from rutile to anatase. In spite of the controversy, much effort has focused on the construction of anatase−rutile mixed photocatalysts.21,466−471 Much progress has been achieved with TiO2 crystals with no specific dominant facet.21,467−471 There is no doubt that anatase−rutile heterostructures constructed from faceted TiO2 crystals can be more efficient photocatalysts as a result of rapid charge carrier transfer across the interface, and exposed reactive facets. Tian et al. constructed a hierarchical anatase/rutile heterojunction photoanode, where ultrathin anatase nanosheets with a high {001} percentage area were anchored perpendicular on the four lateral {110} facets of rutile nanorods (Figure 53a).472 Compared to control photoanodes

5. APPLICATIONS IN ENVIRONMENT AND ENERGY So far, the applications of TiO2 have mainly been in the fields of environment and energy, particularly solar energy conversion and lithium storage. In this section, we will focus on the remarkably improved and unusual performance produced by tailoring the crystal facets of TiO2. Some interesting progress in bioapplications related to faceted TiO2 will also be briefly introduced. 5.1. Photocatalysis

Photocatalysis has four important application aspects, namely, degradation of pollutants, water splitting, reduction of CO2 and organic synthesis. TiO2 based photocatalysts are active in all four aspects. A recent study on the formation of nitrate from atmospheric nitrogen and oxygen on nanosize TiO2 under UV light extended the application aspect of TiO2 photocatalysts.479 For photocatalytic degradation of pollutants and organic synthesis numerous TiO2 photocatalysts, particularly with small particle size and large surface area, have been reported to be efficient due to the long-time generation of strong oxidative species (for example, •OH, O2−, and h+)12,13 under light irradiation. However, these oxidative species are usually less selective to reactants so that photocatalysts have a poor selectivity. The selectivity of photocatalysts can be affected by many factors such as properties (size, polarity, structure, etc.) of

Figure 53. (a) FE-SEM image of (001)A/R-TiO2; (b) j-t curves, under simulated solar light illumination [(A) A-TiO2; (B) A/R-TiO2; (C) (001)A/R-TiO2]; (c) separation and mobility of photogenerated electrons and holes (e−/h+) for (001)A/R-TiO2.472 Reprinted with permission from ref 472. Copyright 2002 Wiley-VCH.

(anatase-TiO2: A-TiO2; anatase/rutile-TiO2: A/R-TiO2), the constructed (001) anatase/rutile photoanode, namely (001)A/ R-TiO2, gives substantially increased photoelectrochemical water oxidation (Figure 53b) and photoelectrocatalytic decomposition of bispenol A. Specifically, the photon-tocurrent conversion efficiency for (001)A/R-TiO2 is 8.2% at around 0.4 V bias, which is respectively 82 and 27 times higher than that of A-TiO2 and A/R-TiO2. The excellent performance 9598

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

the reactant, surface charging, solvent, surface atomic structure, and redox potential of the active species. Various strategies have been developed to control the selectivity of photocatalysts. These include controlling the surface charge of photocatalysts by adjusting the pH, anchoring specific molecules to the surface for the selective adsorption of reactants, coating them with a thin layer of molecularly imprinted polymer that can recoganize the template molecules, encapsulating them with a porous silica shell with tunable pores for the controlled diffusion of reactant molecules, and loading them on substrates, as summarized in the previous paper.260,480 Besides these strategies driven by the extrinsic influences, crystal facet tailoring might provide a platform for producing the preferential reactions and thus selectivity. The basis for this consideration is that the surface atomic structure affects the adsorption/desorption of reactants and intermediate products, and surface electronic structure controls the redox potentials of photogenerated electrons and holes. Liu et al. showed a tunable photocatalytic selectivity of hollow microspheres composed of anatase polyhedra with ca. 20% {001} facet area toward azo dyes.116 A F terminated surface produces the preferential decomposition of MO in comparison to MB. Removing most surface fluorine by NaOH washing or calcination reverses the preferential decomposition selectivity to MB. This reverse is related to the much improved adsorption to MB on the rehydroxylated surface after removing F. Furthermore, the degree of photocatalytic selectivity can be further tuned with an increased concentration of surface F, indicating the key role of surface atomic structure in affecting photocatalytic selectivity. On the other hand, the effect of adsorbed F−, Cl−, and OH− ions on the formaldehyde adsorption performance and mechanism of anatase TiO2 nanosheets with dominant {001} facets was also investigated by the experimental analysis and theoretical simulations.481 The photocatalytic decomposition of azo dyes is basically dominated by an oxidation half reaction. In many situations such as overall water splitting and reduction of CO2 with H2O, both the oxidation and the reduction half reactions play a key role in determining the efficiency. The ability to change both half reactions is desirable for designing efficient photocatalysts. Liu et al. investigated the heteroatom modulated switching of the photocatalytic hydrogen (reduction reaction) and oxygen (oxidation reaction) evolution preferences of anatase TiO2 microspheres with dominant {001} facets (see Figure 7).219 The photocatalytic hydrogen producing reaction is favorable when the shell of the microspheres is boron-free (Figure 54a). By moving the boron from the core to shell of the microspheres, the photocatalytic oxygen producing reaction is favorable (Figure 54b). This switching stems from the downward shift of electronic band edges of the shell by a band bending effect that originates from the extra electrons coming from the interstitial boron. Specifically, photocatalysis requires the simultaneous occurrence of reduction and oxidation induced by the photoexcited conduction band electrons and valence band holes, respectively. In the presence of an electron (hole) donor that easily consumes the photoexcited holes (electrons), the splitting water activity to produce hydrogen (oxygen) by the photoexcited electrons (holes) as the rate-limiting step is determined by the redox power of the electrons (holes). The electrons from the boronfree shell have a stronger reducing power than those from the shell with boron, while the holes from the latter have a stronger oxidative power than those from the former. This can well

Figure 54. Boron distribution-dependent photocatalytic activities. (a) Photo catalytic hydrogen evolution from 1 wt % Pt loaded TiO2 microsphere photocatalysts in the presence of methanol as an electron donor. (b) Photocatalytic oxygen evolution from the TiO2 microsphere photocatalysts in the presence of AgNO3 as a hole donor: (i) TiO2 microspheres with a boron-free shell and (ii) TiO2 microspheres with a boron-containing shell.219 Reprinted with permission from ref 219. Copyright 2012 Wiley-VCH.

explain the hydrogen and oxygen evolution preference of the microspheres resulting from the introduction of boron in the shell. These results suggest the substantial role of surface electronic structure in controlling photocatalytic reaction preferences. As presented in section 3.5, the spatial separation of photoexcited charge carriers by different facets exists in faceted TiO2 crystals. The {101} and {001} facets of anatase and the {110} and {111}/{101} facets of rutile respectively act as photocatalytic reduction and oxidation reaction locations.73,246 This phenomenon plays an important role in rationally designing highly efficient photocatalysts and photoelectrodes. Liu et al. reported that anatase crystals with {101} and {001} facets, with selective deposition of Pt particles by a photoreduction process only on the reductive {101} facets show a much higher photocatalytic hydrogen evolution (reduction reaction) from a mixture of H2O and methanol than do the same crystals with Pt particles on both {101} and {001} facets (Figure 55).405 The superior activity of the former is also

Figure 55. (a) Hydrogen production amounts of Pt deposited TiO2 NCs by means of photochemical-reduction (■) and chemicalreduction routes (▲). SEM images of TiO2 NCs with 0.5% Ptloading amount prepared by (b) chemical reduction and (c) photochemical reduction, respectively.405 Reprinted with permission from ref 405. Copyright 2013 Wiley-VCH.

shown by monitoring the formation of •OH radicals from the capture of the photogenerated holes by the surface-bound OH (oxidation reaction). The above results demonstrate the great potential of the spatial separation effect of the photoexcited charge carriers by the coorperative effect of different facets in one single crystal. There have been many reports of the great influence of deposition process of cocatalyst on photocatalytic 9599

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

charge carriers. Furthermore, surface atomic structure linked to electronic structure variations might change open circuit voltage of DSSCs. Driven by the above potential possibilities, faceted anatase TiO2 crystals have been introduced into the electrodes of DSSCs as summarized in Table 1. Many studies used anatase crystals with {001} or {010} dominant facets to fabricate photoanodes of DSSCs [see refs 72, 123, 138, 144, 153, 165, 227, 358, 359, and 482]. These crystals can be used as an active layer, a scattering layer or both. Most of these results show the obvious advantages of the photoanodes consisting of crystals with dominant {001} facets in improving the energy conversion efficiency compared to P25 reference photoanodes. To the best of our knowledge, the highest overall conversion efficiency so far obtained with such an anatase-based photoanode is 8.49%.144 By investigating a set of anatase crystals with different percentage areas of {001} (10%, 38% and 80%) and {101} facets, the conversion efficiency was found to be positively dependent on the percentage area of {001} facets. The performance superiority of anatase crystals with dominant {001} facets in the photoanodes is usually explained as the synergistic effects of retarding electron recombination, high dye adsorption capability, and the superior light scattering effect related to anatase {001} facets. Based on the comparisons of open-circuit voltage decays and transient absorbance decays in the photoanodes made from crystals with either {001} or {101} dominant facets,71,144 it was revealed that the former has the slower charge recombination kinetics, whose origin could be the way the dye is adsorbed on (001), making it favorable for electron transfer. However, such an adsorption mode leads to a low dye (heteroleptic ruthenium dye C101) coverage on (001) and thus lowers the amount of adsorbed dyes.70 This is also observed experimentally.71 The substantial role of the crystals, particularly micrometer-size crystals, in improving the light harvesting efficiency of the photoanodes is demonstrated in many studies.123,138,153,165 For example, Zhang et al. showed a stronger light scattering effect of anatase microspheres with {001} as the scattering layer than both microspheres without {001} facets and a P25 TiO2 film (Figure 57).123 In addition, because of the higher conduction band edge of {001} than {101} facets, a photoanode made from anatase with {001} facets has a larger open-circuit voltage.71 Anatase crystals with dominant {101} facets are attractive for constructing highly efficient photoanodes for DSSCs.87,99,223 One potential merit associated with the {101} facets is the high density of dye adsorption resulting from the unique adsorption mode on the (101) surface. Yan et al. developed a novel double layer photoanode for DSSCs made of an underlayer of highly crystalline anatase octahedral nanocrystals with smooth {101} facets and an overlayer of agglutinated mesoporous anatase microspheres.87 The integration of the underlayer for primary light absorption and charge collection, with the overlayer for the additional functions of light scattering and electrolyte permeation is responsible for the high energy conversion efficiency of 8.72%. Shiu et al. developed another double layer photoanode made of an active layer of around 30 nm-size octahedron-like TiO2 single crystals with major {101} facets and a scattering layer of around 300 nm-size octahedron-like single crystals with major {101} facets.99 The best DSSC with an optimum photoanode gives a conversion efficiency as high as 10.2%. The outstanding performance of the device is attributed to the greater electron transport rate and the more negative potential for the octahedron-like crystal system than

activities of photocatalysts. The general explanation for the activity difference is the different particle size and chemical states of cocatalysts prepared by different processes. It is now clear that the difference in spatial distribution of cocatalysts on photocatalysts is an important but unfortunately neglected factor. The spatial separation of photoexcited electrons and holes on different facets was also seen in brookite nanosheets surrounded with four {210} and two {101} facts as reduction sites, and two {201} facets as oxidation sites. Consequently, such spatial separation produces an excellent photocatalytic activity of brookite, which is conventionally inactive. Compared to photocatalysts in powder form where electron and hole induced reduction and oxidation reactions occur simultaneously on the surface of single particle, the photoelectrodes in photoelectrochemical applications provide an exposed surface only for either hole induced oxidation reactions (anode) or electron induced reduction reactions (cathode). In this situation, it is necessary to maximize the percentage of oxidative (reductive) facets exposed in the photoanode (photocathode). Zhen et al. showed that a photoanode consisting of rutile pillars with oxidative {111} and {011} facets, has remarkably improved photoelectrochemical water oxidation in comparison to a photoanode consisting of anatase nanotubes with major reductive {101} facets (Figure 56),319

Figure 56. Applied potential bias-dependent photocurrent density of a rutile TiO2 pillar photoelectrode and anatase TiO2 nanotube photoelectrode under irradiation.319 Reprinted with permission from ref 319. Copyright 2012 Royal Society of Chemistry.

although anatase is usually considered to be more active in photocatalysis than rutile. The crystal facet-dependent PEC water splitting performance can also be indicated by the studies of Van et al.361 who showed that a photoanode composed of anatase particles with a better [100] axis orientation has a higher PEC water splitting performance (Figure 29d). All these results shed some light on how to improve the photoelectrode performance by carefully tailoring the exposed surface to have appropriate facets for targeted reactions. 5.2. Solar Cells

Nanosize TiO2 crystals are the most widely used semiconductor as a support for dye molecules to harvest sunlight in the electrodes of DSSCs. Besides the large surface area and high crystallinity usually required for high dye adsorption capability and rapid electron transport, the surface atomic structure is also an important factor affecting light-to-electric energy conversion efficiency. This is because the surface atomic structure characteristics determine the adsorption modes of dye molecules on the crystal surface, which thus affect the amount of dye adsorbed, electron injection from photoexcited dyes to TiO2, and even the recombination probability of photoexcited 9600

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

Table 1. Summary of faceted TiO2 crystal based photoanodes of DSSCs and their corresponding performancea photoanode active layer: P25 TiO2 scattering layer: anatase TiO2 microspheres with major {001} anatase nanosheets with around 75% (001) facet anatase nanosheet-based spheres with over 90% {001} facets anatase sheets with 80% {001} anatase TiO2 nanosheets-based microspheres with dominated {001} TiCl4 treated anatase TiO2 microspheres with (001) exposed UV irradiated anatase TiO2 nanosheets with dominant (001) yolk@shell anatase spheres with 90% {001} single-crystal-like anatase nanotubes exposed (001) anatase nanorods with major (010) and minor (101) anatase nanorods with major (010) and minor (101) anatase nanooctahedra with enclosed {101} facets and agglutinated mesoporous anatase microspheres active layer: octahedron-like 30 nm-sized anatase crystals with major (101)/minor (001) scattering layer: octahedronlike 300 nm-sized anatase crystals with major (101) and minor (001) active layer: cube-like anatase nanosized crystals with (101)/(011) scattering layer: 200 nm-sized TiO2 particles a

performance

refs

η = 7.91%, Jsc = 15.46 mA/cm2, Voc = 729 mV, FF = 0.70 η = 4.56%, Jsc = 12.5 mA/cm2, Voc = 583 mV, FF = 0.627 η = 7.51%, Jsc = 17.9 mA/cm2, Voc = 650 mV η = 8.49%, Jsc = 17.3 mA/cm2, Voc = 805 mV, FF = 0.631 η = 6.64%, Jsc = 15.2 mA/cm2, Voc = 650 mV η = 7.57%, Jsc = 18.2 mA/cm2, Voc = 720 mV, FF = 0.58 η = 6.14%, Jsc = 11.71 mA/cm2, Voc = 694 mV, FF = 0.757 η = 6.01%, Jsc = 13.1 mA/cm2, Voc = 760 mV, FF = 0.60 η = 3.28%, Jsc = 7.02 mA/cm2 η = 7.73%, Jsc = 16.5 mA/cm2 η = 7.73%, Jsc = 16.5 mA/cm2, Voc = 712 mV, FF = 0.66 η = 8.72%, Jsc = 18.18 mA/cm2, Voc = 740 mV, FF = 0.648 η = 10.2%, Jsc = 15.87 mA/cm2, Voc = 825 mV, FF = 0.776 η = 7.06%, Jsc = 15.90 mA/cm2, Voc = 740 mV, FF = 0.60

123 359 153 144 138 72 193 165 172 227 482 87 99 223

η: energy conversion efficiency; Jsc: short circuit current; Voc: open circuit voltage; FF: fill factor.

electron transport due to the use of reactive TiO2 nanosheets. Feng et al. synthesized a double-layer film, consisting of an underlayer of rutile nanorods and an overlayer of anatase nanosheets with dominant {001} facets to load CdSe QDs as photoanodes in QDSSCs.483 The CdSe QD loaded doublelayer photoanode gives an almost five times higher photocurrent density than the photoanode of CdS QD sensitized rutile nanorods because of the increased amount of CdS deposited and its light scattering ability. Different from DSSCs and QDSSCs as a class of photoelectrochemical solar cells, Etgar et al. fabricated QD heterojunction solar cells working on the separation of photogenerated electrons and holes driven by the local electronic field formed in a depletion layer extending from the TiO2 film to the PbS QD film.32 It is shown that a solar cell fabricated with a 30 nm-anatase (001) nanosheet and PbS QDs with Eg of 1.38 eV gives an energy conversion efficiency of 4.73%, which is higher than that (4.04%) from 18 nm-anatase nanoparticles/PbS QDs. The better photovoltaic performance of the nanosheets compared to the nanoparticles is attributed to the higher ionic charge of the (001) compared to the (101) facets which strengthens the attachment of the QDs to the TiO2 surface. Besides the use of well-dispersed anatase crystals with different dominant facets in solar cells, a series of hyperbranched anatase TiO2 nanocrystals/nanorods with their surface consisting of major {101} facets were also used to fabricate the photoelectrodes of DSSCs and heterojunction solar cells.275,484−486 As a result of the substantial role of the hyperbranched nanostructures in optimizing light harvesting and charge collection efficiency of the electrodes of DSSCs, a high power conversion efficiency of 10.26% was obtained.486 Similarly, rutile branched nanostructures with branches grown along backbones of rutile nanowires with lateral major {110} and minor {100} facets gave a higher conversion efficiency than

Figure 57. Diffuse reflectance spectra of the TiO2 microsphere with exposed mirror-like plane {001} crystalline facets, TiO2 microsphere without exposed mirror-like plane {001} crystalline facets, and P25 films.123 Reprinted with permission from ref 123. Copyright 2010 Royal Society of Chemistry.

for the conventional nanoparticle system. Besides the use of anatase crystals with a predominance of {001} and {101} facets in DSSCs, anatase crystals with dominant{010} facets have been also used in the active layer of photoanodes of DSSCs. With similar anatase nanorods with major {010} facets in DSSCs, a 7.73% energy conversion efficiency was achieved by Pan et al.227 and Yang et al.482 Faceted TiO2 crystals were also used in quantum dotsensitized solar cells (QDSSCs).397,483 You et al. demonstrated CdS QDSSCs based on anatase nanosheets with exposed reactive {001} and {100} facets, which gave a conversion efficiency of 2.29% and 2.18%, respectively.397 These values are larger than that (1.46%) of a QDSSC based on P25 TiO2 due to the improved adsorption abilities for CdS QDs, a high surface area, reduced electron recombination, and increased 9601

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

rutile nanowires when used as photoanode material in DSSCs (4.3% vs 2.6%).487 The superiority is attributed to both increased specific surface area and roughness factor of the branched nanostructures. On the other hand, the DSSC electrode fabricated using a mixture of anatase (81%) and rutile (19%) nanorods showed a conversion efficiency of 3.83%.488 In contrast, the electrode of the nanorods of pure anatase or rutile gave an efficiency of 3.54% and 2.12%, respectively. 5.3. Lithium Ion Batteries

Figure 58. Illustrations of (a) anisotropic TiO2/graphene sandwich papers (A-TO/GSP) and (b) isotropic TiO2/graphene sandwich paper (I-TO/GSP) for Li+ ion transport.360 Reprinted with permission from ref 360. Copyright 2013 Royal Society of Chemistry.

TiO2 crystals have been actively studied as a typically safe anode material in lithium-ion batteries with an operating voltage higher than 1 V, which is above the potential where most types of electrolytes or solvents are reduced. Numerous studies in the past decade have been conducted to improve lithium extraction/insertion properties at high charge− discharge rates by controlling particle size, crystallinity and morphology, doping, and hybridizing with other substances, in particular highly conductive substances. The dependence of lithium ion insertion/extraction behavior on crystallographic orientation in both macroscopic anatase single crystals and mesoscopic anatase crystals have been shown.41,398 Encouraged by the demonstrated advantage of the anatase (001) surface in Li+ storage and the development of methods for the synthesis of faceted TiO2 crystals, many studies have focused on high rate lithium storage in anodes made of anatase crystals with {001} facets.110,131,191,360,399,460,489,490 Sun et al. theoretically predicted a smaller energy barrier (1.33 eV vs 2.73 eV) for Li+ insertion into anatase (001) than into the (101) surface, and considered Li+ surface insertion as the rate-limiting step because of the much larger barriers of surface insertion than bulk diffusion (0.35−0.65 eV).399 Experimental results further demonstrated that an anode made of anatase nanosheets with around 80% {001} facet area gives improved lithium storage at high rates with respect to that of anatase octahedral nanocrystals with around 98% {101} facet area. By studying the lithium insertion electrochemistry of polycrystalline anatase crystals with a dominance of (001) and (101) facets, Bousa et al. concluded that in comparison with the (101) face the greater activity of the (001) face for Li+ insertion stems from synergistic contributions of a faster interfacial charge transfer at this surface and the easier Li+ transport within the more open structure of the anatase lattice parallel to the caxis.398 Subsequently, various anatase nanostructures with large percentage areas of {001} facets have been further developed to fabricate Li+ battery anodes with a high rate performance. In particular, the integration of {001} dominated anatase nanosheets with graphene shows great potential in obtaining a large lithium storage capability at high rates by providing additional electronic conductivity paths. Li et al. further demonstrated the substantial role of the orientation relationship between the anatase {001} surface and the surface of graphene in affecting the lithium storage of TiO2-graphene hybrid materials.360 An anisotropic electrode with the anatase {001} surface parallel to the graphene surface shows a specific capacity of 112 mAhg−1 at 100 C. This is 3 times higher than that of an isotropic electrode with anatase {001} nanosheets randomly distributed on graphene (Figure 58). The outstanding performance of the anisotropic electrode could be attributed to the synergetic effects of easy Li+ insertion along the c axis, the short Li+ transport path in the nanosheets, and full interfacial contact between graphene and the nanosheets.

A spherical structure composed of radially oriented anatase bipyramidal nanocrystals was also used as an anode material for lithium-ion batteries.233 Each crystal was elongated with a short pyramid bound by (201) facets on the outside and a long pyramid bound by (401) facets on the inside of the sphere. The material shows a very low initial irreversible capacity loss (15% vs the usual 30−50%) and excellent cyclic capacity retention (0.3% capacity loss per cycle for 100 cycles) probably due to the exposed high-index facets and hierarchical structures. Compared to the widely investigated anatase crystals with tailored facets as an active material in the anodes of lithium batteries, faceted rutile crystals have been rarely studied. Rutile also shows a strong crystallographic orientation dependent lithium storage capability. It has been suggested that the diffusion coefficient (10−6 cm2s−1) of Li+ ions along the cdirection is 9 orders of magnitude higher than that in the abplace.491 Therefore, rutile crystals for lithium storage should have a high percentage of {001} facets. However, the equilibrium shape of a rutile crystal according to the Wulff construction mainly consists of major {110} and {011} facets with low surface energy, but has little {001} facets with the highest surface energy (28.9 meV a.u.−2).264 So far, the preparation of rutile with a large percentage area of {001} facets has had limited success. Chen et al. studied the lithium storage capability of the “desert rose”-like microspheres reassembled from unusual rutile nanosheets with dominant {001} facets that had been stabilized by amorphous MoO3.245 The microspheres show a greatly superior lithium storage capability at a current rate of 600 mAg−1 compared to rutile nanoparticles, nanosize rod-like rutile, or even graphenesupported rutile nanorods. All the above results clearly demonstrate the significance of tailoring the facets of TiO2 crystals in improving the lithium storage capability and cycling stability of the anodes using them as an active material. 5.4. Other Applications

So far, faceted TiO2 crystals have been mainly used in the fields of photocatalysis, solar cells and lithium batteries. Tailoring the facets of TiO2 crystals could also play a pivotal role in other applications such as sensor, biomimetic growth, and biomedical treatments by modulating the interaction between molecules or ions in the environment and surfaces of TiO2 crystals. However, the application of faceted crystals in these other fields is still in its infancy. Some early studies have demonstrated that the abundant basic Ti−OH groups on TiO2 are essential for the nucleation and crystallization of apatite in simulated body fluid (SBF).492 The apatite, however, is not biomimetically formed on a single crystal of 9602

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

rhombohedral anatase.493 Tailoring the surface of TiO2 crystals to specific reactive facets, which can generate abundant Ti−OH groups, might be desirable for promoting the formation of apatite. One impressive related result is that reactive {001} facets of anatase crystals are found to be active in promoting the biomimetic growth of CaP.56 As shown in Figure 59, a large

generated charge carriers between two different facets of all three polymorphs. So far, many facet combinations such as {101}/{001}, {010}/{001}, {101}/{010}, and {001}/{101}/ {010} have been produced in anatase crystals. For rutile and brookite both the controlled production of different facets and selective combinations of them is still in its infancy. Continuous effort is urgently needed to boost the production of rutile and brookite crystals with controlled facets. The rational choice of suitable capping agents for the targeted facets based on computer-assisted surface energy variations could greatly promote the synthesis of the faceted crystals. Also, a combination of different appropriate capping agents could be a powerful way of the production of selective combinations of different facets in the same crystal. Unconventional strategies for changing the surface energies of different facets also need to be explored. For example, it has been predicted that, compared to the case where no strain is applied, the area percentage of the anatase (001) surface can be increased 5 times when a 5% compressive strain is applied biaxially along [100] and [010] directions.280 Furthermore, the moderate strain in the strategy does not introduce extrinsic defects into the material. The availability of high quality faceted crystals makes it possible to more convincingly clarify the structure−propertyperformance relationships of TiO2 crystals due to the welldefined surface structures involved in various applications. The facet (orientation) dependent properties that have been demonstrated (i.e., surface band structure, adsorption of ions and molecules, interaction between clusters/quantum dots and supported surfaces, insertion of lithium and hydrogen, and synergy between different facets) provide important implications for not only understanding the performance differences of TiO2 crystals with various shapes but also for guiding the design of high performance materials based on crystal facet engineering. For example, the anisotropic surface electronic structure effect and the accompanying synergy between different facets have been applied to construct highly efficient photocatalyst systems by controlling the spatial distribution of cocatalysts for reduction and oxidation reactions on reductive and oxidative facets. Three key issues remain to be carefully concerned in future studies. The first is to further reveal new facet dependent unusual properties. The second is to examine the generality of the revealed facet-dependent properties for a wide range of facets. The third is to estimate the influence of surface reconstruction on the properties of micrometer- or even nanosize facets. Faceting itself cannot solve all the shortcomings of crystalline TiO2 (i.e., little visible light absorption for photocatalysis and poor electronic conductivity for Li+ batteries). Further modifications are indispensible in order to maximize the performance of faceted TiO2 crystals. So far, such modifications focus on the conventionally used approaches including doping and heterostructuring. For doping the central point is how to introduce dopants into the material without destroying the growth environment of faceted crystals using wet-chemistry methods. The strategy of using stable TiX (X = nonmental dopant) compounds as the precursors of both TiO2 and the dopant is effective to prepare in situ doped faceted crystals. The doped crystals show much increased visible light absorption compared to undoped faceted crystals. Success in doping by ex situ routes is mainly limited to anatase crystals with dominant {001} and {010} facets. The synthesis of doped TiO2 crystals with other facets still remains challenging but deserves more effort considering the importance of both visible light

Figure 59. FE-SEM microphotographs of anatase TiO2 samples after soaking in SBF for 3 days.56 Reprinted with permission from ref 56. Copyright 2013 American Chemical Society.

surface area of a CaP layer can be formed on microspheres consisting of anatase cuboids with {001} facets. As indicated above, the most attractive advantage of using the crystals with tailored facets in these applications is the possible existence of some specific facets in boosting the activities of TiO2 crystals or even opening up some unique functionality.

6. CONCLUDING REMARKS Undoubtedly, the development of micrometer- and nanosize TiO2 crystals with tailored facets has been one of the most significant events in TiO2 related research in the past five years. The unprecedented progress in the controllable synthesis of the faceted TiO2 crystals provides new opportunities for constructing high performance material systems by finely tuning surface and interfacial properties. The existing gap between the understanding from macroscopic single crystal samples and the results from the practical systems using micrometer- and nanosize crystals without specific facets has been bridged, to a large extent, by the faceted crystals. Furthermore, some conventional beliefs have been revised and unknown physicochemical properties have been revealed as a result of studies on the faceted crystals, which have greatly accelerated the development of TiO2 related applications. Regarding the synthesis of faceted TiO2 crystals, the central task is to control the percentage areas of specific facets and selective combinations of two or more facets in crystals with different sizes. Compared to rutile and brookite, faceted anatase crystals have been most intensely investigated. The controllable synthesis of anatase with {001} and coexisting {101}/{010} facets has been essentially realized. The percentage area of {001} facets in the crystals can be tuned from close to zero to nearly 100% (98.7%). Correspondingly, the percentage areas of {101}/{010} facets can also be well controlled. The thickness of these particles along the c-axis can vary from several micrometers to a minimum of 1.6 nm. However, the controllable formation of other anatase facets including {110} and high-index facets is far from satisfactory. Selective combinations of different facets in one crystal are significant because of the potential synergy between different facets, as shown experimentally by the intriguing transfer of photo9603

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

absorption and surface atomic structure in photocatalysis. Doping specific facets of TiO2 is also meaningful to the application of TiO2 in lithium ion batteries by increasing electronic conductivity.494 One issue to be considered for doping is the possible influence of dopant incorporation on surface atomic structure changes, as indicated by the occurrence of surface reconstruction of nitrogen-doped TiO2 single crystals in the Surf. Sci. field. For heterostructuring, the current results show an improvement in the performance but do not fully reflect the potential ability of the facet dependent interfacial properties to substantially improve the performance of the heterostructures. In addition, the coupling of organic substances with faceted TiO2 crystals has been rarely explored. Synthetic TiO2 crystals with different facets are used in nearly all applications of common TiO2 crystals. The encouraging improvements in performance and new performance features (e. g., facet-dependent reaction selectivity in photocatalysis) are stimulating increased interest in studying TiO2 crystals. One important but neglected issue is to develop specific facets that can be matched with other components in various applications to obtain an optimum performance. For example, designing special dyes and electrolytes for specific facets of TiO2 crystals might further improve the efficiency of DSSCs. Overall, the development of TiO2 crystals with tailored facets provides a valuable platform for the rational design and fabrication of efficient material systems, revealing facet dependent properties, and improving the performance in various applications. Furthermore, the knowledge accumulated by studying faceted TiO2 has prompted the development of other metal oxides with different facets.372,495−507

at Institute of Metal Research (IMR), Chinese Academy of Sciences (CAS) in 2009. During his Ph.D. study, he worked at Prof. G. Q. Max Lu’s laboratory for 1.5 years in Australia. He was the recipient of the T. S. Kê RESEARCH FELLOWSHIP founded by Shenyang National Laboratory for Materials Science (SYNL), IMR CAS. Now he is a professor of SYNL, IMR CAS. His current main research interest is to develop photocatalytic materials to utilize solar energy for clean environment and renewable energy.

Hua Gui Yang received his Bachelor degree from Qingdao University in 1996 and completed his Ph.D. in 2005 at the National University of Singapore. He then joined the General Electric (GE) Company as a Research Scientist and moved to the University of Queensland (UQ) in 2007 as a Postdoctoral Research Fellow. Currently he is a professor

ASSOCIATED CONTENT

at East China University of Science and Technology. Currently he has

S Supporting Information *

interests in design and synthesis of metallic and semiconducting

Data on publications and citations (Figures S1−S3). Slab models of different surfaces (Figures S4−S6). This material is available free of charge via the Internet at http://pubs.acs.org.

functional materials for renewable clean energy and environmental protection applications.

AUTHOR INFORMATION Corresponding Authors

*E-mail: [emailprotected]. *E-mail: [emailprotected]. Notes

The authors declare no competing financial interest. Biographies

Jian Pan received his Bachelor degree in Material Chemistry in China University of Petroleum in 2007. Now he is a Ph.D. candidate in a joint training program of the Graduate University of the Chinese Academy of Sciences and the University of Queensland (UQ). He studied at the lab of Professor Lianzhou Wang in UQ as a joint Ph.D. student for 2 years. His current research is focusing on the morphology control of titanium-based semiconductor for solar energy conversion under the supervision of Professor Hui-Ming Cheng,

Gang Liu received his Bachelor degree in Materials Physics in Jilin University in 2003. He obtained his Ph.D. degree in Materials Science

Professor Gang Liu, and Professor Lianzhou Wang. 9604

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

Hui-Ming Cheng is Professor and Director of Advanced Carbon Research Division of Shenyang National Laboratory for Materials Science, Institute of Metal Research, Chinese Academy of Sciences. His research activities focus on the synthesis, properties and applications of carbon nanotubes, graphene, energy storage materials, and photocatalytic materials. He has published over 400 papers with >20 000 citations. He has received several international and national awards, including the 2nd class National Natural Science Prize, the Charles E. Pettinos Award, and the Prize for Scientific and Technological Progress of Ho Leung Ho Lee Foundation. He is the Editor of Carbon since 2000 and Editor-in-Chief of New Carbon Materials since 1998 and has given more than 70 plenary/keynote/ invited talks in international conferences and symposia.

Yongqiang Yang received his Bachelor Degree in Material Chemistry in University of Science and Technology of China in 2011. Now he is a Ph.D. candidate in a joint Ph.D. training program of Institute of Metal Research, Chinese Academy of Sciences and University of Science and Technology of China. His current research is focusing on photocatalysts with engineered surfaces for photocatalytic water splitting under the supervision of Professor Gang Liu and Professor Hui-Ming Cheng.

ACKNOWLEDGMENTS The authors thank the Major Basic Research Program (2014CB239401), Ministry of Science and Technology of China, and NSFC (No. 51172243, 51002160, 21090343, and 51221264). REFERENCES (1) Fujishima, A.; Honda, K. Nature 1972, 238, 37. (2) Gratzel, M. Nature 2001, 414, 338. (3) Cho, I. S.; Lee, C. H.; Feng, Y. Z.; Logar, M.; Rao, P. M.; Cai, L. L.; Kim, D. R.; Sinclair, R.; Zheng, X. L. Nat. Commun. 2013, 4, doi:10.1038/ncomms2729. (4) Oregan, B.; Gratzel, M. Nature 1991, 353, 737. (5) Bach, U.; Lupo, D.; Comte, P.; Moser, J. E.; Weissortel, F.; Salbeck, J.; Spreitzer, H.; Gratzel, M. Nature 1998, 395, 583. (6) Gratzel, M. J. Photochem. Photobiol. C Photochem. Rev. 2003, 4, 145. (7) Hagfeldt, A.; Boschloo, G.; Sun, L. C.; Kloo, L.; Pettersson, H. Chem. Rev. 2010, 110, 6595. (8) Yella, A.; Lee, H. W.; Tsao, H. N.; Yi, C. Y.; Chandiran, A. K.; Nazeeruddin, M. K.; Diau, E. W. G.; Yeh, C. Y.; Zakeeruddin, S. M.; Gratzel, M. Science 2011, 334, 629. (9) Crossland, E. J. W.; Noel, N.; Sivaram, V.; Leijtens, T.; Alexander-Webber, J. A.; Snaith, H. J. Nature 2013, 495, 215. (10) Chung, I.; Lee, B.; He, J. Q.; Chang, R. P. H.; Kanatzidis, M. G. Nature 2012, 485, 486. (11) Wang, Z. S.; Kawauchi, H.; Kashima, T.; Arakawa, H. Coord. Chem. Rev. 2004, 248, 1381. (12) Linsebigler, A. L.; Lu, G. Q.; Yates, J. T. Chem. Rev. 1995, 95, 735. (13) Hoffmann, M. R.; Martin, S. T.; Choi, W. Y.; Bahnemann, D. W. Chem. Rev. 1995, 95, 69. (14) Fox, M. A.; Dulay, M. T. Chem. Rev. 1993, 93, 341. (15) Fujishima, A.; Zhang, X. T.; Tryk, D. A. Surf. Sci. Rep. 2008, 63, 515. (16) Chen, H. H.; Nanayakkara, C. E.; Grassian, V. H. Chem. Rev. 2012, 112, 5919.

G. Q. Max Lu received his Ph.D. in Chemical Engineering from University of Queensland, Australia. He has been a faculty member at the University of Queensland since 1994, and last held the position of Chair in Nanotechnology before moving into senior executive position at the University where he is currently Provost and Senior Vice President. Dr Lu has received numerous prestigious awards including a Federation Fellowship. He was the foundation Director for the Australian Research Council Centre of Excellencefor Functional Nanomaterials. Dr Lu is coauthor of over 500 papers (with 23 000 citations and h-index of 77). His research interests include carbon, silicate and oxide nanoparticles and nanoporous materials for energy and environmental applications. 9605

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

(17) Chen, X.; Shen, S.; Guo, L.; Mao, S. S. Chem. Rev. 2010, 110, 6503. (18) Kubacka, A.; Fernandez-Garcia, M.; Colon, G. Chem. Rev. 2012, 112, 1555. (19) Tada, H.; Fujishima, M.; Kobayashi, H. Chem. Soc. Rev. 2011, 40, 4232. (20) Anpo, M.; Takeuchi, M. J. Catal. 2003, 216, 505. (21) Zhang, J.; Xu, Q.; Feng, Z.; Li, M.; Li, C. Angew. Chem., Int. Ed. 2008, 47, 1766. (22) Chen, C. C.; Ma, W. H.; Zhao, J. C. Chem. Soc. Rev. 2010, 39, 4206. (23) Dhakshinamoorthy, A.; Navalon, S.; Corma, A.; Garcia, H. Energy Environ. Sci. 2012, 5, 9217. (24) Henderson, M. A. Surf. Sci. Rep. 2011, 66, 185. (25) Liu, L. M.; McAllister, B.; Ye, H. Q.; Hu, P. J. Am. Chem. Soc. 2006, 128, 4017. (26) Maeda, Y.; Iizuka, Y.; Kohyama, M. J. Am. Chem. Soc. 2013, 135, 906. (27) Froschl, T.; Hormann, U.; Kubiak, P.; Kucerova, G.; Pfanzelt, M.; Weiss, C. K.; Behm, R. J.; Husing, N.; Kaiser, U.; Landfester, K.; Wohlfahrt-Mehrens, M. Chem. Soc. Rev. 2012, 41, 5313. (28) Chen, M. S.; Goodman, D. W. Science 2004, 306, 252. (29) Kaden, W. E.; Wu, T. P.; Kunkel, W. A.; Anderson, S. L. Science 2009, 326, 826. (30) Green, I. X.; Tang, W. J.; Neurock, M.; Yates, J. T. Science 2011, 333, 736. (31) Enache, D. I.; Edwards, J. K.; Landon, P.; Solsona-Espriu, B.; Carley, A. F.; Herzing, A. A.; Watanabe, M.; Kiely, C. J.; Knight, D. W.; Hutchings, G. J. Science 2006, 311, 362. (32) Etgar, L.; Zhang, W.; Gabriel, S.; Hickey, S. G.; Nazeeruddin, M. K.; Eychmueller, A.; Liu, B.; Graetzel, M. Adv. Mater. 2012, 24, 2202. (33) Kim, Y. G.; Walker, J.; Samuelson, L. A.; Kumar, J. Nano Lett. 2003, 3, 523. (34) Slooff, L. H.; Kroon, J. M.; Loos, J.; Koetse, M. M.; Sweelssen, J. Adv. Funct. Mater. 2005, 15, 689. (35) Natarajan, C.; Setoguchi, K.; Nogami, G. Electrochim. Acta 1998, 43, 3371. (36) Wagemaker, M.; Kentgens, A. P. M.; Mulder, F. M. Nature 2002, 418, 397. (37) Kavan, L.; Kalbac, M.; Zukalova, M.; Exnar, I.; Lorenzen, V.; Nesper, R.; Graetzel, M. Chem. Mater. 2004, 16, 477. (38) Hu, Y. S.; Kienle, L.; Guo, Y. G.; Maier, J. Adv. Mater. 2006, 18, 1421. (39) Guo, Y. G.; Hu, Y. S.; Sigle, W.; Maier, J. Adv. Mater. 2007, 19, 2087. (40) Bruce, P. G.; Scrosati, B.; Tarascon, J. M. Angew. Chem., Int. Ed. 2008, 47, 2930. (41) Hengerer, R.; Kavan, L.; Krtil, P.; Gratzel, M. J. Electrochem. Soc. 2000, 147, 1467. (42) Zhou, W. J.; Du, G. J.; Hu, P. G.; Li, G. H.; Wang, D. Z.; Liu, H.; Wang, J. Y.; Boughton, R. I.; Liu, D.; Jiang, H. D. J. Mater. Chem. 2011, 21, 7937. (43) Lu, H. F.; Li, F.; Liu, G.; Chen, Z. G.; Wang, D. W.; Fang, H. T.; Lu, G. Q.; Jiang, Z. H.; Cheng, H. M. Nanotechnology 2008, 19, 405504. (44) Varghese, O. K.; Gong, D. W.; Paulose, M.; Ong, K. G.; Dickey, E. C.; Grimes, C. A. Adv. Mater. 2003, 15, 624. (45) Mor, G. K.; Varghese, O. K.; Paulose, M.; Grimes, C. A. Sens. Lett. 2003, 1, 42. (46) Zheng, Q.; Zhou, B. X.; Bai, J.; Li, L. H.; Jin, Z. J.; Zhang, J. L.; Li, J. H.; Liu, Y. B.; Cai, W. M.; Zhu, X. Y. Adv. Mater. 2008, 20, 1044. (47) Liu, G.; Li, F.; Wang, D. W.; Tang, D. M.; Liu, C.; Ma, X. L.; Lu, G. Q.; Cheng, H. M. Nanotechnology 2008, 19, 025606. (48) Zhu, W. D.; Wang, C. W.; Chen, J. B.; Li, D. S.; Zhou, F.; Zhang, H. L. Nanotechnology 2012, 23, 455204. (49) Chen, J. B.; Wang, L. Q.; Wang, C. W.; Li, D. S.; Li, Y.; Wang, J.; Zhou, F. Vacuum 2013, 96, 18. (50) Alivov, Y.; Molloi, S. J. Appl. Phys. 2010, 108, 024303.

(51) Alivov, Y.; Klopfer, M.; Molloi, S. Nanotechnology 2010, 21, 505706. (52) Schwartz, J. J.; Stavrakis, S.; Quake, S. R. Nat. Nanotechnol 2010, 5, 127. (53) Wang, N.; Li, H. Y.; Lu, W. L.; Li, J. H.; Wang, J. S.; Zhang, Z. T.; Liu, Y. R. Biomaterials 2011, 32, 6900. (54) Sugita, Y.; Ishizaki, K.; Iwasa, F.; Ueno, T.; Minamikawa, H.; Yamada, M.; Suzuki, T.; Ogawa, T. Biomaterials 2011, 32, 8374. (55) Liu, X. Y.; Zhao, X. B.; Fu, R. K. Y.; Ho, J. P. Y.; Ding, C. X.; Chu, P. K. Biomaterials 2005, 26, 6143. (56) Ruso, J. M.; Verdinelli, V.; Hassan, N.; Pieroni, O.; Messina, P. V. Langmuir 2013, 29, 2350. (57) Yin, Z. F.; Wu, L.; Yang, H. G.; Su, Y. H. Phys. Chem. Chem. Phys. 2013, 15, 4844. (58) Chen, X.; Mao, S. S. Chem. Rev. 2007, 107, 2891. (59) Diebold, U. Surf. Sci. Rep. 2003, 48, 53. (60) Thompson, T. L.; Yates, J. T. Chem. Rev. 2006, 106, 4428. (61) Asahi, R.; Morikawa, T.; Ohwaki, T.; Aoki, K.; Taga, Y. Science 2001, 293, 269. (62) Chen, X. B.; Liu, L.; Yu, P. Y.; Mao, S. S. Science 2011, 331, 746. (63) Bak, T.; Nowotny, J.; Sucher, N. J.; Wachsman, E. J. Phys. Chem. C 2011, 115, 15711. (64) Liu, G.; Wang, L.; Yang, H. G.; Cheng, H.-M.; Lu, G. Q. J. Mater. Chem. 2010, 20, 831. (65) Chen, D.; Caruso, R. A. Adv. Funct. Mater. 2013, 23, 1356. (66) Mor, G. K.; Varghese, O. K.; Paulose, M.; Shankar, K.; Grimes, C. A. Sol. Energy Mater. Sol. Cells 2006, 90, 2011. (67) Anpo, M.; Dohshi, S.; Kitano, M.; Hu, Y.; Takeuchi, M.; Matsuoka, M. Annu. Rev. Mater. Res. 2005, 35, 1. (68) Yang, P. D.; Zhao, D. Y.; Margolese, D. I.; Chmelka, B. F.; Stucky, G. D. Nature 1998, 396, 152. (69) Vittadini, A.; Selloni, A.; Rotzinger, F. P.; Gratzel, M. Phys. Rev. Lett. 1998, 81, 2954. (70) De Angelis, F.; Vitillaro, G.; Kavan, L.; Nazeeruddin, M. K.; Graetzel, M. J. Phys. Chem. C 2012, 116, 18124. (71) Laskova, B.; Zukalova, M.; Kavan, L.; Chou, A.; Liska, P.; Wei, Z.; Bin, L.; Kubat, P.; Ghadiri, E.; Moser, J. E.; Graetzel, M. J. Solid State Electrochem. 2012, 16, 2993. (72) Wang, H.; Liu, M.; Yan, C.; Bell, J. Beilstein J. Nanotechnol. 2012, 3, 378. (73) Ohno, T.; Sarukawa, K.; Matsumura, M. New J. Chem. 2002, 26, 1167. (74) Barnard, A. S.; Curtiss, L. A. Nano Lett. 2005, 5, 1261. (75) Yang, H. G.; Sun, C. H.; Qiao, S. Z.; Zou, J.; Liu, G.; Smith, S. C.; Cheng, H. M.; Lu, G. Q. Nature 2008, 453, 638. (76) Herman, G. S.; Gao, Y.; Tran, T. T.; Osterwalder, J. Surf. Sci. 2000, 447, 201. (77) Herman, G. S.; Gao, Y. Thin Solid Films 2001, 397, 157. (78) Marshall, M. S. J.; Castell, M. R. Phys. Rev. Lett. 2009, 102, 146102. (79) Kennedy, R. J.; Stampe, P. A. J. Cryst. Growth 2003, 252, 333. (80) Murakami, M.; Matsumoto, Y.; Nakajima, K.; Makino, T.; Segawa, Y.; Chikyow, T.; Ahmet, P.; Kawasaki, M.; Koinuma, H. Appl. Phys. Lett. 2001, 78, 2664. (81) Jeong, B. S.; Norton, D. P.; Budai, J. D.; Jellison, G. E. Thin Solid Films 2004, 446, 18. (82) Amano, F.; Yasumoto, T.; Prieto-Mahaney, O.-O.; Uchida, S.; Shibayama, T.; Ohtani, B. Chem. Commun. 2009, 2311. (83) Wang, H.; Shao, W.; Gu, F.; Zhang, L.; Lu, M.; Li, C. Inorg. Chem. 2009, 48, 9732. (84) Wu, N.; Wang, J.; Tafen, D. N.; Wang, H.; Zheng, J.-G.; Lewis, J. P.; Liu, X.; Leonard, S. S.; Manivannan, A. J. Am. Chem. Soc. 2010, 132, 6679. (85) Chen, C.; Hu, R.; Mai, K.; Ren, Z.; Wang, H.; Qian, G.; Wang, Z. Cryst. Growth Des. 2011, 11, 5221. (86) Pan, J. H.; Han, G.; Zhou, R.; Zhao, X. S. Chem. Commun. 2011, 47, 6942. (87) Yan, K.; Qiu, Y.; Chen, W.; Zhang, M.; Yang, S. Energy Environ. Sci. 2011, 4, 2168. 9606

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

(88) Gordon, T. R.; Cargnello, M.; Paik, T.; Mangolini, F.; Weber, R. T.; Fornasiero, P.; Murray, C. B. J. Am. Chem. Soc. 2012, 134, 6751. (89) Horvat, B.; Recnik, A.; Drazic, G. J. Cryst. Growth 2012, 347, 19. (90) Jiao, W.; Wang, L.; Liu, G.; Lu, G. Q.; Cheng, H.-M. Acs Catal. 2012, 2, 1854. (91) Pan, H.; Qian, J.; Cui, Y.; Xie, H.; Zhou, X. J. Mater. Chem. 2012, 22, 6002. (92) Chen, Q.; Liu, H.; Xin, Y.; Cheng, X.; Li, J. Appl. Surf. Sci. 2013, 264, 476. (93) Li, T.; Tian, B.; Zhang, J.; Dong, R.; Wang, T.; Yang, F. Ind. Eng. Chem. Res. 2013, 52, 6704. (94) Liu, G.; Yang, H. G.; Sun, C.; Cheng, L.; Wang, L.; Lu, G. Q.; Cheng, H.-M. CrystEngComm 2009, 11, 2677. (95) Chen, Q.; Ma, W.; Chen, C.; Ji, H.; Zhao, J. Chem.Eur. J. 2012, 18, 12584. (96) Wu, D.; Gao, Z.; Xu, F.; Chang, J.; Gao, S.; Jiang, K. CrystEngComm 2013, 15, 516. (97) Yang, M.-H.; Chen, P.-C.; Tsai, M.-C.; Chen, T.-T.; Chang, I. C.; Chiu, H.-T.; Lee, C.-Y. CrystEngComm 2013, 15, 2966. (98) Li, J.; Yu, Y.; Chen, Q.; Li, J.; Xu, D. Cryst. Growth Des. 2010, 10, 2111. (99) Shiu, J.-W.; Lan, C.-M.; Chang, Y.-C.; Wu, H.-P.; Huang, W.-K.; Diau, E. W.-G. ACS Nano 2012, 6, 10862. (100) Amano, F.; Prieto-Mahaney, O.-O.; Terada, Y.; Yasumoto, T.; Shibayama, T.; Ohtani, B. Chem. Mater. 2009, 21, 2601. (101) Dai, Y.; Cobley, C. M.; Zeng, J.; Sun, Y.; Xia, Y. Nano Lett. 2009, 9, 2455. (102) Han, X.; Kuang, Q.; Jin, M.; Xie, Z.; Zheng, L. J. Am. Chem. Soc. 2009, 131, 3152. (103) Hu, X.; Zhang, T.; Jin, Z.; Huang, S.; Fang, M.; Wu, Y.; Zhang, L. Cryst. Growth Des. 2009, 9, 2324. (104) Liu, G.; Yang, H. G.; Wang, X.; Cheng, L.; Lu, H.; Wang, L.; Lu, G. Q.; Cheng, H.-M. J. Phys. Chem. C 2009, 113, 21784. (105) Liu, G.; Yang, H. G.; Wang, X.; Cheng, L.; Pan, J.; Lu, G. Q.; Cheng, H.-M. J. Am. Chem. Soc. 2009, 131, 12868. (106) Murakami, N.; Kurihara, Y.; Tsubota, T.; Ohno, T. J. Phys. Chem. C 2009, 113, 3062. (107) Yang, H. G.; Liu, G.; Qiao, S. Z.; Sun, C. H.; Jin, Y. G.; Smith, S. C.; Zou, J.; Cheng, H. M.; Lu, G. Q. J. Am. Chem. Soc. 2009, 131, 4078. (108) Zhang, D.; Li, G.; Yang, X.; Yu, J. C. Chem. Commun. 2009, 4381. (109) Zheng, Z.; Huang, B.; Qin, X.; Zhang, X.; Dai, Y.; Jiang, M.; Wang, P.; Whangbo, M.-H. Chem.Eur. J. 2009, 15, 12576. (110) Chen, J. S.; Tan, Y. L.; Li, C. M.; Cheah, Y. L.; Luan, D.; Madhavi, S.; Boey, F. Y. C.; Archer, L. A.; Lou, X. W. J. Am. Chem. Soc. 2010, 132, 6124. (111) Feng, J.; Yin, M.; Wang, Z.; Yan, S.; Wan, L.; Li, Z.; Zou, Z. CrystEngComm 2010, 12, 3425. (112) Liu, G.; Sun, C.; Smith, S. C.; Wang, L.; Lu, G. Q.; Cheng, H.M. J. Colloid Interface Sci. 2010, 349, 477. (113) Liu, G.; Sun, C.; Yang, H. G.; Smith, S. C.; Wang, L.; Lu, G. Q.; Cheng, H.-M. Chem. Commun. 2010, 46, 755. (114) Liu, M.; Piao, L.; Lu, W.; Ju, S.; Zhao, L.; Zhou, C.; Li, H.; Wang, W. Nanoscale 2010, 2, 1115. (115) Liu, M.; Piao, L.; Zhao, L.; Ju, S.; Yan, Z.; He, T.; Zhou, C.; Wang, W. Chem. Commun. 2010, 46, 1664. (116) Liu, S.; Yu, J.; Jaroniec, M. J. Am. Chem. Soc. 2010, 132, 11914. (117) Ma, X. Y.; Chen, Z. G.; Hartono, S. B.; Jiang, H. B.; Zou, J.; Qiao, S. Z.; Yang, H. G. Chem. Commun. 2010, 46, 6608. (118) Shan, G.-B.; Demopoulos, G. P. Nanotechnology 2010, 21, 025604. (119) Wang, X.; Huang, B.; Wang, Z.; Qin, X.; Zhang, X.; Dai, Y.; Whangbo, M.-H. Chem.Eur. J. 2010, 16, 7106. (120) Xiang, Q.; Lv, K.; Yu, J. Appl. Catal., B 2010, 96, 557. (121) Yu, J.; Qi, L.; Jaroniec, M. J. Phys. Chem. C 2010, 114, 13118. (122) Yu, J.; Xiang, Q.; Ran, J.; Mann, S. CrystEngComm 2010, 12, 872.

(123) Zhang, H.; Han, Y.; Liu, X.; Liu, P.; Yu, H.; Zhang, S.; Yao, X.; Zhao, H. Chem. Commun. 2010, 46, 8395. (124) Zhu, J.; Wang, S.; Bian, Z.; Xie, S.; Cai, C.; Wang, J.; Yang, H.; Li, H. CrystEngComm 2010, 12, 2219. (125) Bian, Z.; Zhu, J.; Wen, J.; Cao, F.; Huo, Y.; Qian, X.; Cao, Y.; Shen, M.; Li, H.; Lu, Y. Angew. Chem., Int. Ed. 2011, 50, 1105. (126) Cai, C.; Wang, J.; Cao, F.; Li, H.; Zhu, J. Chin. J. Catal. 2011, 32, 862. (127) Cao, F.-L.; Wang, J.-G.; Lv, F.-J.; Zhang, D.-Q.; Huo, Y.-N.; Li, G.-S.; Li, H.-X.; Zhu, J. Catal. Commun. 2011, 12, 946. (128) Chen, J. S.; Chen, C.; Liu, J.; Xu, R.; Qiao, S. Z.; Lou, X. W. Chem. Commun. 2011, 47, 2631. (129) Chen, J. S.; Liu, J.; Qiao, S. Z.; Xu, R.; Lou, X. W. Chem. Commun. 2011, 47, 10443. (130) Cuong, Ky. N.; Cha, H. G.; Kang, Y. S. Cryst. Growth Des. 2011, 11, 3947. (131) Ding, S.; Chen, J. S.; Wang, Z.; Cheah, Y. L.; Madhavi, S.; Hu, X.; Lou, X. W. J. Mater. Chem. 2011, 21, 1677. (132) Fan, J.; Cai, W.; Yu, J. Chem.Asian J. 2011, 6, 2481. (133) Fang, W. Q.; Zhou, J. Z.; Liu, J.; Chen, Z. G.; Yang, C.; Sun, C. H.; Qian, G. R.; Zou, J.; Qiao, S. Z.; Yang, H. G. Chem.Eur. J. 2011, 17, 1423. (134) Feng, S.; Yang, J.; Zhu, H.; Liu, M.; Zhang, J.; Wu, J.; Wan, J. J. Am. Ceram. Soc. 2011, 94, 310. (135) Liu, B.; Aydil, E. S. Chem. Commun. 2011, 47, 9507. (136) Pan, J.; Liu, G.; Lu, G. Q.; Cheng, H.-M. Angew. Chem., Int. Ed. 2011, 50, 2133. (137) Wang, X.; Liu, G.; Wang, L.; Pan, J.; Lu, G. Q.; Cheng, H.-M. J. Mater. Chem. 2011, 21, 869. (138) Wang, Y.; Yang, W.; Shi, W. Ind. Eng. Chem. Res. 2011, 50, 11982. (139) Wang, Y.; Zhang, H.; Han, Y.; Liu, P.; Yao, X.; Zhao, H. Chem. Commun. 2011, 47, 2829. (140) Wen, C. Z.; Zhou, J. Z.; Jiang, H. B.; Hu, Q. H.; Qiao, S. Z.; Yang, H. G. Chem. Commun. 2011, 47, 4400. (141) Wu, J.-M.; Song, X.-M.; Ma, L.-Y.; Wei, X.-D. J. Cryst. Growth 2011, 319, 57. (142) Wu, J.-M.; Tang, M.-L. J. Hazard. Mater. 2011, 190, 566. (143) Wu, J.-M.; Tang, M.-L. Nanoscale 2011, 3, 3915. (144) Wu, X.; Chen, Z.; Lu, G. Q.; Wang, L. Adv. Funct. Mater. 2011, 21, 4167. (145) Xiang, G.; Li, T.; Wang, X. Inorg. Chem. 2011, 50, 6237. (146) Xiang, G.; Wu, D.; He, J.; Wang, X. Chem. Commun. 2011, 47, 11456. (147) Xiang, Q.; Yu, J. Chin. J. Catal. 2011, 32, 525. (148) Xiang, Q.; Yu, J.; Jaroniec, M. Chem. Commun. 2011, 47, 4532. (149) Xiang, Q.; Yu, J.; Jaroniec, M. Phys. Chem. Chem. Phys. 2011, 13, 4853. (150) Xiang, Q.; Yu, J.; Wang, W.; Jaroniec, M. Chem. Commun. 2011, 47, 6906. (151) Xie, S.; Han, X.; Kuang, Q.; Fu, J.; Zhang, L.; Xie, Z.; Zheng, L. Chem. Commun. 2011, 47, 6722. (152) Xing, M.-Y.; Qi, D.-Y.; Zhang, J.-L.; Chen, F. Chem.Eur. J. 2011, 17, 11432. (153) Yang, W.; Li, J.; Wang, Y.; Zhu, F.; Shi, W.; Wan, F.; Xu, D. Chem. Commun. 2011, 47, 1809. (154) Yang, X. H.; Li, Z.; Liu, G.; Xing, J.; Sun, C.; Yang, H. G.; Li, C. CrystEngComm 2011, 13, 1378. (155) Yang, X. H.; Li, Z.; Sun, C.; Yang, H. G.; Li, C. Chem. Mater. 2011, 23, 3486. (156) Yang, X. H.; Yang, H. G.; Li, C. Chem.Eur. J. 2011, 17, 6615. (157) Yu, H.; Tian, B.; Zhang, J. Chem.Eur. J. 2011, 17, 5499. (158) Yu, J.; Dai, G.; Xiang, Q.; Jaroniec, M. J. Mater. Chem. 2011, 21, 1049. (159) Zhang, D.; Yang, X.; Zhu, J.; Zhang, Y.; Zhang, P.; Li, G. J. SolGel Sci. Technol. 2011, 58, 594. (160) Zhang, H.; Liu, P.; Li, F.; Liu, H.; Wang, Y.; Zhang, S.; Guo, M.; Cheng, H.; Zhao, H. Chem.Eur. J. 2011, 17, 5949. 9607

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

(199) Zhou, X.; Peng, F.; Wang, H.; Yu, H.; Fang, Y. Chem. Commun. 2012, 48, 600. (200) Bai, Y.; Luo, P. Y.; Wang, P. Q.; Liu, J. Y. Catal. Commun. 2013, 37, 45. (201) Dai, G.; Liu, S.; Liang, Y.; Liu, H.; Zhong, Z. J. Mol. Catal. A: Chem. 2013, 368, 38. (202) Ding, X.; Hong, Z.; Wang, Y.; Lai, R.; Wei, M. J. Alloys Compd. 2013, 550, 475. (203) He, Z. Q.; Cai, Q. L.; Wu, M.; Shi, Y. Q.; Fang, H. Y.; Li, L. D.; Chen, J. C.; Chen, J. M.; Song, S. Ind. Eng. Chem. Res. 2013, 52, 9556. (204) Liu, M.; Li, H.; Zeng, Y.; Huang, T. Appl. Surf. Sci. 2013, 274, 117. (205) Liu, N.; Zhao, Y.; Wang, X.; Peng, H.; Li, G. Mater. Lett. 2013, 102, 53. (206) Luan, Y. B.; Jing, L. Q.; Xie, Y.; Sun, X. J.; Feng, Y. J.; Fu, H. G. ACS Catal. 2013, 3, 1378. (207) Miao, J.; Liu, B. RSC Adv. 2013, 3, 1222. (208) Pan, X.; Zhang, N.; Fu, X.; Xu, Y.-J. Appl. Catal., A 2013, 453, 181. (209) Pomoni, K.; Georgakopoulos, T.; Sofianou, M. V.; Trapalis, C. J. Alloys Compd. 2013, 558, 1. (210) Pomoni, K.; Sofianou, M. V.; Georgakopoulos, T.; Boukos, N.; Trapalis, C. J. Alloys Compd. 2013, 548, 194. (211) Roy, N.; Sohn, Y.; Pradhan, D. ACS Nano 2013, 7, 2532. (212) Wang, B.; Guo, L.; He, M.; He, T. Phys. Chem. Chem. Phys. 2013, 15, 9891. (213) Wang, J.; Bian, Z.; Zhu, J.; Li, H. J. Mater. Chem. A 2013, 1, 1296. (214) Wang, J.; Zhang, P.; Li, X.; Zhu, J.; Li, H. Appl. Catal., B 2013, 134, 198. (215) Wang, W.; Lu, C.; Ni, Y.; Xu, Z. CrystEngComm 2013, 15, 2537. (216) Yu, J.; Zhang, L.; Huang, B.; Liu, H. Int. J. Electrochem. Sci. 2013, 8, 5810. (217) Zhang, H.; Liu, X.; Li, Y.; Li, Y.; Zhao, H. Sci. China: Chem. 2013, 56, 402. (218) Zhou, L.; Chen, J.; Ji, C.; Zhou, L.; O’Brien, P. CrystEngComm 2013, 15, 5012. (219) Liu, G.; Pan, J.; Yin, L.; Irvine, J. T. S.; Li, F.; Tan, J.; Wormald, P.; Cheng, H.-M. Adv. Funct. Mater. 2012, 22, 3233. (220) Li, J.; Cao, K.; Li, Q.; Xu, D. CrystEngComm 2012, 14, 83. (221) Pan, L.; Zou, J.-J.; Wang, S.; Liu, X.-Y.; Zhang, X.; Wang, L. ACS Appl. Mater. Interfaces 2012, 4, 1650. (222) Ding, X.; Ruan, H.; Zheng, C.; Yang, J.; Wei, M. CrystEngComm 2013, 15, 3040. (223) Shuang, Y.; Hou, Y.; Zhang, B.; Yang, H. G. Ind. Eng. Chem. Res. 2013, 52, 4098. (224) Wu, L.; Yang, B. X.; Yang, X. H.; Chen, Z. G.; Li, Z.; Zhao, H. J.; Gong, X. Q.; Yang, H. G. CrystEngComm 2013, 15, 3252. (225) Li, J.; Xu, D. Chem. Commun. 2010, 46, 2301. (226) Miao, Y.; Gao, J. Micro Nano Lett. 2011, 6, 848. (227) Pan, J.; Wu, X.; Wang, L.; Liu, G.; Lub, G. Q.; Cheng, H.-M. Chem. Commun. 2011, 47, 8361. (228) Zhao, J.; Zou, X.-X.; Su, J.; Wang, P.-P.; Zhou, L.-J.; Li, G.-D. Dalton Trans. 2013, 42, 4365. (229) Wu, B.; Guo, C.; Zheng, N.; Xie, Z.; Stucky, G. D. J. Am. Chem. Soc. 2008, 130, 17563. (230) Wu, Q.; Wu, Z.; Li, Y.; Gao, H.; Piao, L.; Zhang, T.; Du, L. Chin. J. Catal. 2012, 33, 1743. (231) Wu, L.; Jiang, H. B.; Tian, F.; Chen, Z.; Sun, C.; Yang, H. G. Chem. Commun. 2013, 49, 2016. (232) Jiang, H. B.; Cuan, Q.; Wen, C. Z.; Xing, J.; Wu, D.; Gong, X.Q.; Li, C.; Yang, H. G. Angew. Chem., Int. Ed. 2011, 50, 3764. (233) Wu, H. B.; Chen, J. S.; Lou, X. W.; Hng, H. H. Nanoscale 2011, 3, 4082. (234) Yang, X.; Chen, J.; Gong, L.; Wu, M.; Yu, J. C. J. Am. Chem. Soc. 2009, 131, 12048. (235) Liu, L.; Qian, J.; Li, B.; Cui, Y.; Zhou, X.; Guo, X.; Ding, W. Chem. Commun. 2010, 46, 2402.

(161) Zhang, H.; Wang, Y.; Liu, P.; Han, Y.; Yao, X.; Zou, J.; Cheng, H.; Zhao, H. ACS Appl. Mater. Interfaces 2011, 3, 2472. (162) Zheng, Z.; Huang, B.; Lu, J.; Qin, X.; Zhang, X.; Dai, Y. Chem.Eur. J. 2011, 17, 15032. (163) Zong, X.; Xing, Z.; Yu, H.; Chen, Z.; Tang, F.; Zou, J.; Lu, G. Q.; Wang, L. Chem. Commun. 2011, 47, 11742. (164) Chen, L.; Shen, L.; Nie, P.; Zhang, X.; Li, H. Electrochim. Acta 2012, 62, 408. (165) Fang, W. Q.; Yang, X. H.; Zhu, H.; Li, Z.; Zhao, H.; Yao, X.; Yang, H. G. J. Mater. Chem. 2012, 22, 22082. (166) Gu, L.; Wang, J.; Cheng, H.; Du, Y.; Han, X. Chem. Commun. 2012, 48, 6978. (167) Han, X.; Wang, X.; Xie, S.; Kuang, Q.; Ouyang, J.; Xie, Z.; Zheng, L. Rsc Adv. 2012, 2, 3251. (168) Han, X.; Zheng, B.; Ouyang, J.; Wang, X.; Kuang, Q.; Jiang, Y.; Xie, Z.; Zheng, L. Chem.Asian J. 2012, 7, 2538. (169) Hao, F.; Wang, X.; Zhou, C.; Jiao, X.; Li, X.; Li, J.; Lin, H. J. Phys. Chem. C 2012, 116, 19164. (170) He, Z.; Cai, Q.; Hong, F.; Jiang, Z.; Chen, J.; Song, S. Ind. Eng. Chem. Res. 2012, 51, 5662. (171) Ichimura, A. S.; Mack, B. M.; Usmani, S. M.; Mars, D. G. Chem. Mater. 2012, 24, 2324. (172) Jung, M.-H.; Chu, M.-J.; Kang, M. G. Chem. Commun. 2012, 48, 5016. (173) Kao, C.-H.; Tsai, J.-H.; Yeh, S.-W.; Huang, H.-L.; Gan, D.; Shen, P. Jpn. J. Appl. Phys. 2012, 51. (174) Lai, Z.; Peng, F.; Wang, Y.; Wang, H.; Yu, H.; Liu, P.; Zhao, H. J. Mater. Chem. 2012, 22, 23906. (175) Lee, W.-J.; Sung, Y.-M. Cryst. Growth Des. 2012, 12, 5792. (176) Li, H.; Zeng, Y.; Huang, T.; Piao, L.; Liu, M. ChemPlusChem. 2012, 77, 1017. (177) Li, H.; Zeng, Y.; Huang, T.; Piao, L.; Yan, Z.; Liu, M. Chem. Eur. J. 2012, 18, 7525. (178) Li, X.; Zhu, J.; Li, H. Catal. Commun. 2012, 24, 20. (179) Liu, B.; Huang, Y.; Wen, Y.; Du, L.; Zeng, W.; Shi, Y.; Zhang, F.; Zhu, G.; Xu, X.; Wang, Y. J. Mater. Chem. 2012, 22, 7484. (180) Liu, P.; Wang, Y.; Zhang, H.; An, T.; Yang, H.; Tang, Z.; Cai, W.; Zhao, H. Small 2012, 8, 3664. (181) Shi, W.; Yang, W.; Li, Q.; Gao, S.; Shang, P.; Shang, J. K. Nanoscale Res. Lett. 2012, 7, 1. (182) Sun, L.; Zhao, Z.; Zhou, Y.; Liu, L. Nanoscale 2012, 4, 613. (183) Wang, L.; Zang, L.; Zhao, J.; Wang, C. Chem. Commun. 2012, 48, 11736. (184) Wang, W.; Lu, C.; Ni, Y.; Su, M.; Xu, Z. Appl. Catal., B 2012, 127, 28. (185) Wang, W.; Lu, C.-H.; Ni, Y.-R.; Song, J.-B.; Su, M.-X.; Xu, Z.-Z. Catal. Commun. 2012, 22, 19. (186) Wang, W.; Ni, Y.; Lu, C.; Xu, Z. Rsc Adv. 2012, 2, 8286. (187) Wang, W.-S.; Wang, D.-H.; Qu, W.-G.; Lu, L.-Q.; Xu, A.-W. J. Phys. Chem. C 2012, 116, 19893. (188) Wang, X.; He, H.; Chen, Y.; Zhao, J.; Zhang, X. Appl. Surf. Sci. 2012, 258, 5863. (189) Wu, Q.; Liu, M.; Wu, Z.; Li, Y.; Piao, L. J. Phys. Chem. C 2012, 116, 26800. (190) Yu, Y.; Cao, C.; Li, W.; Li, P.; Qu, J.; Song, W. Nano Res. 2012, 5, 434. (191) Yu, Y.; Wang, X.; Sun, H.; Ahmad, M. Rsc Adv. 2012, 2, 7901. (192) Zhang, J.; Chen, W.; Xi, J.; Ji, Z. Mater. Lett. 2012, 79, 259. (193) Zhang, J.; Wang, J.; Zhao, Z.; Yu, T.; Feng, J.; Yuan, Y.; Tang, Z.; Liu, Y.; Li, Z.; Zou, Z. Phys. Chem. Chem. Phys. 2012, 14, 4763. (194) Zhang, J.; Xi, J.; Ji, Z. J. Mater. Chem. 2012, 22, 17700. (195) Zhang, Y.; Li, C.; Pan, C. J. Am. Ceram. Soc. 2012, 95, 2951. (196) Zhao, Y.; Zhao, Q.; Li, X.; Hou, Y.; Zou, X.; Wang, J.; Jiang, T.; Xie, T. Mater. Lett. 2012, 66, 308. (197) Zhao, Z.; Sun, Z.; Zhao, H.; Zheng, M.; Du, P.; Zhao, J.; Fan, H. J. Mater. Chem. 2012, 22, 21965. (198) Zheng, Y.; Lv, K.; Wang, Z.; Deng, K.; Li, M. J. Mol. Catal. A: Chem. 2012, 356, 137. 9608

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

(236) Cao, T.; Li, Y.; Wang, C.; Shao, C.; Liu, Y. J. Nanomater. 2011, 267415. (237) Hao, F.; Lin, H.; Zhou, C.; Liu, Y.; Li, J. Phys. Chem. Chem. Phys. 2011, 13, 15918. (238) Hayashi, K.; Nakamura, M.; Makita, Y.; Fujiwara, R.; Kori, T.; Ishimura, K. Mater. Lett. 2011, 65, 3037. (239) Bai, H.; Liu, Z.; Sun, D. D. J. Mater. Chem. 2012, 22, 18801. (240) Hong, Z.; Xu, Y.; Liu, Y.; Wei, M. Chem.Eur. J. 2012, 18, 10753. (241) Cha, S. I.; Hwang, K. H.; Kim, Y. H.; Yun, M. J.; Seo, S. H.; Shin, Y. J.; Moon, J. H.; Lee, D. Y. Nanoscale 2013, 5, 753. (242) Sosnowchik, B. D.; Chiamori, H. C.; Ding, Y.; Ha, J.-Y.; Wang, Z. L.; Lin, L. Nanotechnology 2010, 21, 485601. (243) Liu, Y.; Wang, H.; Wang, Y.; Xu, H.; Li, M.; Shen, H. Chem. Commun. 2011, 47, 3790. (244) Aoyama, Y.; Oaki, Y.; Ise, R.; Imai, H. CrystEngComm 2012, 14, 1405. (245) Chen, J. S.; Lou, X. W. Chem. Sci. 2011, 2, 2219. (246) Murakami, N.; Katayama, S.; Nakamura, M.; Tsubota, T.; Ohno, T. J. Phys. Chem. C 2011, 115, 419. (247) Liu, X.; Zhang, H.; Yao, X.; An, T.; Liu, P.; Wang, Y.; Peng, F.; Carroll, A. R.; Zhao, H. Nano Res. 2012, 5, 762. (248) Pang, C. L.; Lindsay, R.; Thornton, G. Chem. Rev. 2013, 113, 3887. (249) Zhang, H.; Liu, X.; Wang, Y.; Liu, P.; Cai, W.; Zhu, G.; Yang, H.; Zhao, H. J. Mater. Chem. A 2013, 1, 2646. (250) Sun, L.; Qin, Y.; Cao, Q. Q.; Hu, B. Q.; Huang, Z. W.; Ye, L.; Tang, X. F. Chem. Commun. 2011, 47, 12628. (251) Kobayashi, M.; Petrykin, V.; Tomita, K.; Kakihana, M. J. Cryst. Growth 2011, 337, 30. (252) Katsumata, K.-i.; Ohno, Y.; Tomita, K.; Taniguchi, T.; Matsush*ta, N.; Okada, K. ACS Appl. Mater. Interfaces 2012, 4, 4846. (253) Hu, W. B.; Li, L. P.; Li, G. S.; Tang, C. L.; Sun, L. Cryst. Growth Des. 2009, 9, 3676. (254) Lin, H. F.; Li, L. P.; Zhao, M. L.; Huang, X. S.; Chen, X. M.; Li, G. S.; Yu, R. C. J. Am. Chem. Soc. 2012, 134, 8328. (255) Zhao, M. L.; Li, L. P.; Lin, H. F.; Yang, L. S.; Li, G. S. Chem. Commun. 2013, 49, 7046. (256) Buonsanti, R.; Grillo, V.; Carlino, E.; Giannini, C.; Kipp, T.; Cingolani, R.; Cozzoli, P. D. J. Am. Chem. Soc. 2008, 130, 11223. (257) Ohno, Y.; Tomita, K.; Komatsubara, Y.; Taniguchi, T.; Katsumata, K.; Matsush*ta, N.; Kogure, T.; Okada, K. Cryst. Growth Des. 2011, 11, 4831. (258) Wen, C. Z.; Jiang, H. B.; Qiao, S. Z.; Yang, H. G.; Lu, G. Q. J. Mater. Chem. 2011, 21, 7052. (259) Fang, W. Q.; Gong, X.-Q.; Yang, H. G. J. Phys. Chem. Lett. 2011, 2, 725. (260) Liu, G.; Yu, J. C.; Lu, G. Q.; Cheng, H.-M. Chem. Commun. 2011, 47, 6763. (261) Liu, S.; Yu, J.; Jaroniec, M. Chem. Mater. 2011, 23, 4085. (262) Gong, X. Q.; Selloni, A. Phys. Rev. B 2007, 76, 235307. (263) Lazzeri, M.; Vittadini, A.; Selloni, A. Phys. Rev. B 2001, 63, 155409. (264) Ramamoorthy, M.; Vanderbilt, D.; Kingsmith, R. D. Phys. Rev. B 1994, 49, 16721. (265) Lovette, M. A.; Browning, A. R.; Griffin, D. W.; Sizemore, J. P.; Snyder, R. C.; Doherty, M. F. Ind. Eng. Chem. Res. 2008, 47, 9812. (266) Bourne, J. R.; Davey, R. J. J. Cryst. Growth 1976, 36, 278. (267) Bourne, J. R.; Davey, R. J. J. Cryst. Growth 1976, 36, 287. (268) Lahav, M.; Leiserowitz, L. Chem. Eng. Sci. 2001, 56, 2245. (269) Oliver, P. M.; Watson, G. W.; Kelsey, E. T.; Parker, S. C. J. Mater. Chem. 1997, 7, 563. (270) Arrouvel, C.; Digne, M.; Breysse, M.; Toulhoat, H.; Raybaud, P. J. Catal. 2004, 222, 152. (271) Joo, J.; Chow, B. Y.; Prakash, M.; Boyden, E. S.; Jacobson, J. M. Nat. Mater. 2011, 10, 596. (272) Dinh, C. T.; Nguyen, T. D.; Kleitz, F.; Do, T. O. ACS Nano 2009, 3, 3737.

(273) Jun, Y. W.; Casula, M. F.; Sim, J. H.; Kim, S. Y.; Cheon, J.; Alivisatos, A. P. J. Am. Chem. Soc. 2003, 125, 15981. (274) Joo, J.; Kwon, S. G.; Yu, T.; Cho, M.; Lee, J.; Yoon, J.; Hyeon, T. J. Phys. Chem. B 2005, 109, 15297. (275) Buonsanti, R.; Carlino, E.; Giannini, C.; Altamura, D.; De Marco, L.; Giannuzzi, R.; Manca, M.; Gigli, G.; Cozzoli, P. D. J. Am. Chem. Soc. 2011, 133, 19216. (276) Lin, Z.-H.; Roy, P.; Shih, Z.-Y.; Ou, C.-M.; Chang, H.-T. ChemPlusChem. 2013, 78, 302. (277) Alivov, Y.; Fan, Z. Y. J. Phys. Chem. C 2009, 113, 12954. (278) Oropeza, F. E.; Zhang, K. H. L.; Regoutz, A.; Lazarov, V. K.; Wermeille, D.; Poll, C. G.; Egdell, R. G. Cryst. Growth Des. 2013, 13, 1438. (279) Zhou, L.; Smyth-Boyle, D.; O’Brien, P. J. Am. Chem. Soc. 2008, 130, 1309. (280) Jia, L.; Shu, D.-J.; Wang, M. Phys. Rev. Lett. 2012, 109, 156104. (281) Zhang, H. Z.; Banfield, J. F. J. Phys. Chem. B 2000, 104, 3481. (282) Finnegan, M. P.; Zhang, H. Z.; Banfield, J. F. J. Phys. Chem. C 2007, 111, 1962. (283) Lazzeri, M.; Vittadini, A.; Selloni, A. Phys. Rev. B 2002, 65, 119901. (284) Penn, R. L.; Banfield, J. F. Geochim. Cosmochim. Acta 1999, 63, 1549. (285) Hosono, E.; Fujihara, S.; Lmai, H.; Honma, I.; Masaki, I.; Zhou, H. S. ACS Nano 2007, 1, 273. (286) Kakiuchi, K.; Hosono, E.; Imai, H.; Kimura, T.; Fujihara, S. J. Cryst. Growth 2006, 293, 541. (287) Dinh, C.-T.; Nguyen, T.-D.; Kleitz, F.; Do, T.-O. ACS Nano 2009, 3, 3737. (288) Shang, S.; Jiao, X.; Chen, D. ACS Appl. Mater. Interfaces 2012, 4, 860. (289) Yu, Y. X.; Xu, D. S. Appl. Catal., B 2007, 73, 166. (290) Menzel, R.; Duerrbeck, A.; Liberti, E.; Yau, H. C.; McComb, D.; Shaffer, M. S. P. Chem. Mater. 2013, 25, 2137. (291) Sofianou, M.-V.; Trapalis, C.; Psycharis, V.; Boukos, N.; Vaimakis, T.; Yu, J.; Wang, W. Environ. Sci. Pollut. Res. 2012, 19, 3719. (292) Zhao, X.; Jin, W.; Cai, J.; Ye, J.; Li, Z.; Ma, Y.; Xie, J.; Qi, L. Adv. Funct. Mater. 2011, 21, 3554. (293) Wen, C. Z.; Hu, Q. H.; Guo, Y. N.; Gong, X. Q.; Qiao, S. Z.; Yang, H. G. Chem. Commun. 2011, 47, 6138. (294) Wang, Z.; Huang, B.; Dai, Y.; Zhang, X.; Qin, X.; Li, Z.; Zheng, Z.; Cheng, H.; Guo, L. CrystEngComm 2012, 14, 4578. (295) Liu, S.; Yu, J.; Cheng, B.; Jaroniec, M. Adv. Colloid Interface Sci. 2012, 173, 35. (296) Lv, K.; Cheng, B.; Yu, J.; Liu, G. Phys. Chem. Chem. Phys. 2012, 14, 5349. (297) Bai, Y.; Luo, P.-Y.; Wang, P.-Q.; Liu, J.-Y. Catal. Commun. 2013, 37, 45. (298) Alivov, Y.; Fan, Z. Y. Nanotechnology 2009, 20, 405610. (299) Xu, D.; Li, J.; Yu, Y.; Li, J. Sci. China: Chem. 2012, 55, 2334. (300) Wang, C.; Zhang, X.; Zhang, Y.; Jia, Y.; Yuan, B.; Yang, J.; Sun, P.; Liu, Y. Nanoscale 2012, 4, 5023. (301) Ye, L.; Liu, J.; Tian, L.; Peng, T.; Zan, L. Appl. Catal., B 2013, 134, 60. (302) Tian, N.; Zhou, Z. Y.; Sun, S. G.; Ding, Y.; Wang, Z. L. Science 2007, 316, 732. (303) Jiao, Y.; Peng, C.; Guo, F.; Bao, Z.; Yang, J.; Schmidt-Mende, L.; Dunbar, R.; Qin, Y.; Deng, Z. J. Phys. Chem. C 2011, 115, 6405. (304) Li, F.; Xu, J.; Chen, L.; Ni, B.; Li, X.; Fu, Z.; Lu, Y. J. Mater. Chem. A 2013, 1, 225. (305) Lee, C.; Ghosez, P.; Gonze, X. Phys. Rev. B 1994, 50, 13379. (306) Parker, R. A.; Wasilik, J. H. Phys. Rev. 1960, 120, 1631. (307) Lee, Y. H.; Chan, K. K.; Brady, M. J. J. Vac. Sci. Technol., A 1995, 13, 596. (308) Maeda, K. Chem. Commun. 2013, 49, 8404. (309) Yin, H. B.; Wada, Y.; Kitamura, T.; Kambe, S.; Murasawa, S.; Mori, H.; Sakata, T.; Yanagida, S. J. Mater. Chem. 2001, 11, 1694. (310) Liu, B.; Aydil, E. S. J. Am. Chem. Soc. 2009, 131, 3985. 9609

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

(311) Zhang, J.; Li, L.; Li, G. Phys. Chem. Chem. Phys. 2012, 14, 11167. (312) Inada, M.; Mizue, K.; Enomoto, N.; Hojo, J. Sci. Adv. Mater. 2010, 2, 102. (313) Inada, M.; Mizue, K.; Enomoto, N.; Hojo, J. J. Ceram. Soc. Jpn. 2009, 117, 819. (314) Liu, Y.; Wang, H.; Li, H.; Zhao, W.; Liang, C.; Huang, H.; Deng, Y.; Shen, H. J. Colloid Interface Sci. 2011, 363, 504. (315) Guo, W. X.; Xu, C.; Wang, X.; Wang, S. H.; Pan, C. F.; Lin, C. J.; Wang, Z. L. J. Am. Chem. Soc. 2012, 134, 4437. (316) Lai, Z.; Peng, F.; Wang, H.; Yu, H.; Zhang, S.; Zhao, H. J. Mater. Chem. A 2013, 1, 4182. (317) Shi, J.; Wang, X. Cryst. Growth Des. 2011, 11, 949. (318) Wang, X. D.; Shi, J. J. Mater. Res. 2013, 28, 270. (319) Zhen, C.; Liu, G.; Cheng, H. M. Nanoscale 2012, 4, 3871. (320) Ranade, M. R.; Navrotsky, A.; Zhang, H. Z.; Banfield, J. F.; Elder, S. H.; Zaban, A.; Borse, P. H.; Kulkarni, S. K.; Doran, G. S.; Whitfield, H. J. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 6476. (321) Dambournet, D.; Belharouak, I.; Amine, K. Chem. Mater. 2010, 22, 1173. (322) Koelsch, M.; Cassaignon, S.; Guillemoles, J. F.; Jolivet, J. R. Thin Solid Films 2002, 403, 312. (323) Yan, W. F.; Chen, B.; Mahurin, S. M.; Dai, S.; Overbury, S. H. Chem. Commun. 2004, 1918. (324) Manera, M. G.; Taurino, A.; Catalano, M.; Rella, R.; Caricato, A. P.; Buonsanti, R.; Cozzoli, P. D.; Martino, M. Sens. Actuators, B 2012, 161, 869. (325) Iskandar, F.; Nandiyanto, A. B. D.; Yun, K. M.; Hogan, C. J.; Okuyama, K.; Biswas, P. Adv. Mater. 2007, 19, 1408. (326) Addamo, M.; Augugliaro, V.; Bellardita, M.; Di Paola, A.; Loddo, V.; Palmisano, G.; Palmisano, L.; Yurdakal, S. Catal. Lett. 2008, 126, 58. (327) Li, J. G.; Tang, C. C.; Li, D.; Haneda, H.; Ishigaki, T. J. Am. Ceram. Soc. 2004, 87, 1358. (328) Magne, C.; Dufour, F.; Labat, F.; Lancel, G.; Durupthy, O.; Cassaignon, S.; Pauporte, T. J. Photochem. Photobiol., A 2012, 232, 22. (329) Mo, S. D.; Ching, W. Y. Phys. Rev. B 1995, 51, 13023. (330) Kominami, H.; Ishii, Y.; Kohno, M.; Konishi, S.; Kera, Y.; Ohtani, B. Catal. Lett. 2003, 91, 41. (331) Ohtani, B.; Handa, J.; Nishimoto, S.; Kagiya, T. Chem. Phys. Lett. 1985, 120, 292. (332) Kandiel, T. A.; Feldhoff, A.; Robben, L.; Dillert, R.; Bahnemann, D. W. Chem. Mater. 2010, 22, 2050. (333) Ismail, A. A.; Kandiel, T. A.; Bahnemann, D. W. J. Photochem. Photobiol., A 2010, 216, 183. (334) Zhang, L. J.; Menendez-Flores, V. M.; Murakami, N.; Ohno, T. Appl. Surf. Sci. 2012, 258, 5803. (335) Zhao, B.; Chen, F.; Huang, Q. W.; Zhang, J. L. Chem. Commun. 2009, 5115. (336) Tomita, K.; Petrykin, V.; Kobayashi, M.; Shiro, M.; Yoshimura, M.; Kakihana, M. Angew. Chem., Int. Ed. 2006, 45, 2378. (337) Colfen, H.; Antonietti, M. Angew. Chem., Int. Ed. 2005, 44, 5576. (338) Song, R. Q.; Colfen, H. Adv. Mater. 2010, 22, 1301. (339) Ye, J. F.; Liu, W.; Cai, J. G.; Chen, S. A.; Zhao, X. W.; Zhou, H. H.; Qi, L. M. J. Am. Chem. Soc. 2011, 133, 933. (340) Bian, Z.; Tachikawa, T.; Majima, T. J. Phys. Chem. Lett. 2012, 3, 1422. (341) Liu, Y.; Zhang, Y.; Li, H.; Wang, J. Cryst. Growth Des. 2012, 12, 2625. (342) Zhang, A. Y.; Long, L. L.; Li, W. W.; Wang, W. K.; Yu, H. Q. Chem. Commun. 2013, 49, 6075. (343) Luo, L.; Hui, J.; Yu, Q.; Zhang, Z.; Jing, D.; Wang, P.; Yang, Y.; Wang, X. CrystEngComm 2012, 14, 7648. (344) Hong, Z. S.; Wei, M. D.; Lan, T. B.; Jiang, L. L.; Cao, G. Z. Energy Environ. Sci. 2012, 5, 5408. (345) Hong, Z. S.; Wei, M. D.; Lan, T. B.; Cao, G. Z. Nano Energy 2012, 1, 466.

(346) Joo, J. B.; Dahl, M.; Li, N.; Zaera, F.; Yin, Y. D. Energy Environ. Sci. 2013, 6, 2082. (347) Bao, Y.; Yang, Y. Q.; Ma, J. Z. J. Inorg. Mater. 2013, 28, 459. (348) Chen, J. S.; Luan, D.; Li, C. M.; Boey, F. Y. C.; Qiao, S.; Lou, X. W. Chem. Commun. 2010, 46, 8252. (349) Chen, J. S.; Archer, L. A.; Lou, X. W. J. Mater. Chem. 2011, 21, 9912. (350) Chen, J. S.; Lou, X. W. Mater. Today 2012, 15, 246. (351) Zhang, R. Y.; Elzatahry, A. A.; Al-Deyab, S. S.; Zhao, D. Y. Nano Today 2012, 7, 344. (352) Zhou, W.; Fu, H. G. ChemCatChem. 2013, 5, 885. (353) Korzhak, A. V.; Ermokhina, N. I.; Stroyuk, A. L.; Bukhtiyarov, V. K.; Raevskaya, A. E.; Litvin, V. I.; Kuchmiy, S. Y.; Ilyin, V. G.; Manorik, P. A. J. Photochem. Photobiol., A 2008, 198, 126. (354) Li, W.; Wu, Z. X.; Wang, J. X.; Elzatahry, A. A.; Zhao, D. Y. Chem. Mater. 2014, 26, 287. (355) Sanz, R.; Johansson, A.; Skupinski, M.; Jensen, J.; Possnert, G.; Boman, M.; Vazquez, M.; Hjort, K. Nano Lett. 2006, 6, 1065. (356) Jiao, W.; Xie, Y. P.; Chen, R. Z.; Zhen, C.; Liu, G.; Ma, X. L.; Cheng, H. M. Chem. Commun. 2013, 49, 11770. (357) Zheng, X. L.; Kuang, Q.; Yan, K. Y.; Qiu, Y. C.; Qiu, J. H.; Yang, S. H. ACS Appl. Mater. Interfaces 2013, 5, 11249. (358) Wu, X.; Lu, G.; Wang, L. J. Colloid Interface Sci. 2013, 391, 70. (359) Yu, J.; Fan, J.; Lv, K. Nanoscale 2010, 2, 2144. (360) Li, N.; Zhou, G. M.; Fang, R. P.; Li, F.; Cheng, H. M. Nanoscale 2013, 5, 7780. (361) Van, T. K.; Nguyen, C. K.; Kang, Y. S. Chem.Eur. J. 2013, 19, 9376. (362) Feng, X. J.; Shankar, K.; Varghese, O. K.; Paulose, M.; Latempa, T. J.; Grimes, C. A. Nano Lett. 2008, 8, 3781. (363) Yang, X. F.; Zhuang, J. L.; Li, X. Y.; Chen, D. H.; Ouyang, G. F.; Mao, Z. Q.; Han, Y. X.; He, Z. H.; Liang, C. L.; Wu, M. M.; Yu, J. C. ACS Nano 2009, 3, 1212. (364) Diebold, U.; Ruzycki, N.; Herman, G. S.; Selloni, A. Catal. Today 2003, 85, 93. (365) Lazzeri, M.; Selloni, A. Phys. Rev. Lett. 2001, 87, 266105. (366) Selcuk, S.; Selloni, A. J. Phys. Chem. C 2013, 117, 6358. (367) Selloni, A. Nat. Mater. 2008, 7, 613. (368) Giorgi, G.; Palummo, M.; Chiodo, L.; Yamash*ta, K. Phys. Rev. B 2011, 84, 073404. (369) Wang, Y.; Sun, H. J.; Tan, S. J.; Feng, H.; Cheng, Z. W.; Zhao, J.; Zhao, A. D.; Wang, B.; Luo, Y.; Yang, J. L.; Hou, J. G. Nat. Commun. 2013, 4, DOI: 10.1038/ncomms3214. (370) Lee, J. H.; Fernandez Hevia, D.; Selloni, A. Phys. Rev. Lett. 2013, 110, 016101. (371) Kavan, L.; Gratzel, M.; Gilbert, S. E.; Klemenz, C.; Scheel, H. J. J. Am. Chem. Soc. 1996, 118, 6716. (372) Xie, Y. P.; Liu, G.; Yin, L.; Cheng, H.-M. J. Mater. Chem. 2012, 22, 6746. (373) Wang, X. W.; Yin, L. C.; Liu, G.; Wang, L. Z.; Saito, R.; Lu, G. Q.; Cheng, H. M. Energy Environ. Sci. 2011, 4, 3976. (374) Bi, Y.; Ouyang, S.; Umezawa, N.; Cao, J.; Ye, J. J. Am. Chem. Soc. 2011, 133, 6490. (375) Sakai, N.; Ebina, Y.; Takada, K.; Sasaki, T. J. Am. Chem. Soc. 2004, 126, 5851. (376) Tao, J. G.; Luttrell, T.; Batzill, M. Nat. Chem. 2011, 3, 296. (377) Kawakita, M.; Kawakita, J.; Sakka, Y.; Shinohara, T. J. Electrochem. Soc. 2010, 157, H65. (378) Zhao, Z.; Li, Z.; Zou, Z. J. Phys.: Condens. Matter 2010, 22, 175008. (379) Martsinovich, N.; Troisi, A. Phys. Chem. Chem. Phys. 2012, 14, 13392. (380) Onda, K.; Li, B.; Zhao, J.; Jordan, K. D.; Yang, J. L.; Petek, H. Science 2005, 308, 1154. (381) He, Y. B.; Tilocca, A.; Dulub, O.; Selloni, A.; Diebold, U. Nat. Mater. 2009, 8, 585. (382) Sun, C.; Selloni, A.; Du, A.; Smith, S. C. J. Phys. Chem. C 2011, 115, 17092. (383) Selloni, A.; Vittadini, A.; Gratzel, M. Surf. Sci. 1998, 402, 219. 9610

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

(384) Zhao, Z.; Li, Z.; Zou, Z. J. Phys. Chem. C 2012, 116, 7430. (385) Brookes, I. M.; Muryn, C. A.; Thornton, G. Phys. Rev. Lett. 2001, 87, 266103. (386) Henderson, M. A. Surf. Sci. 1996, 355, 151. (387) Fahmi, A.; Minot, C. Surf. Sci. 1994, 304, 343. (388) Barnard, A. S.; Zapol, P.; Curtiss, L. A. J. Chem. Theory Comput. 2005, 1, 107. (389) Wendt, S.; Matthiesen, J.; Schaub, R.; Vestergaard, E. K.; Laegsgaard, E.; Besenbacher, F.; Hammer, B. Phys. Rev. Lett. 2006, 96, 066107. (390) Bikondoa, O.; Pang, C. L.; Ithnin, R.; Muryn, C. A.; Onishi, H.; Thornton, G. Nat. Mater. 2006, 5, 189. (391) Tilocca, A.; Selloni, A. J. Phys. Chem. B 2004, 108, 4743. (392) Beck, T. J.; Klust, A.; Batzill, M.; Diebold, U.; Di Valentin, C.; Tilocca, A.; Selloni, A. Surf. Sci. 2005, 591, L267. (393) Di Valentin, C.; Tilocca, A.; Selloni, A.; Beck, T. J.; Klust, A.; Batzill, M.; Losovyj, Y.; Diebold, U. J. Am. Chem. Soc. 2005, 127, 9895. (394) Kusama, H.; Orita, H.; Sugihara, H. Langmuir 2008, 24, 4411. (395) Cakir, D.; Gulseren, O.; Mete, E.; Ellialtioglu, S. Phys. Rev. B 2009, 80, 035431. (396) Sun, C.; Smith, S. C. J. Phys. Chem. C 2012, 116, 3524. (397) You, T.; Jiang, L.; Han, K.-L.; Deng, W.-Q. Nanotechnology 2013, 24, 245401. (398) Bousa, M.; Laskova, B.; Zukalova, M.; Prochazka, J.; Chou, A.; Kavan, L. J. Electrochem. Soc. 2010, 157, A1108. (399) Sun, C. H.; Yang, X. H.; Chen, J. S.; Li, Z.; Lou, X. W.; Li, C.; Smith, S. C.; Lu, G. Q.; Yang, H. G. Chem. Commun. 2010, 46, 6129. (400) Sun, C.; Jia, Y.; Yang, X.-H.; Yang, H.-G.; Yao, X.; Lu, G. Q.; Selloni, A.; Smith, S. C. J. Phys. Chem. C 2011, 115, 25590. (401) Zhang, D.; Li, G.; Wang, H.; Chan, K. M.; Yu, J. C. Cryst. Growth Des. 2010, 10, 1130. (402) Maitani, M. M.; Tanaka, K.; Mochizuki, D.; Wada, Y. J. Phys. Chem. Lett. 2011, 2, 2655. (403) Bae, E.; Murakami, N.; Ohno, T. J. Mol. Catal. A: Chem. 2009, 300, 72. (404) Bae, E.; Ohno, T. Appl. Catal., B 2009, 91, 634. (405) Liu, C.; Han, X.; Xie, S.; Kuang, Q.; Wang, X.; Jin, M.; Xie, Z.; Zheng, L. Chem.Asian J. 2013, 8, 282. (406) Guo, M. C.; Li, L. P.; Lin, H. F.; Zuo, Y.; Huang, X. S.; Li, G. S. Chem. Commun. 2013, 49, 11752. (407) Tachikawa, T.; Wang, N.; Yamash*ta, S.; Cui, S. C.; Majima, T. Angew. Chem., Int. Ed. 2010, 49, 8593. (408) Tachikawa, T.; Majima, T. Chem. Commun. 2012, 48, 3300. (409) Tachikawa, T.; Majima, T. Chem. Soc. Rev. 2010, 39, 4802. (410) Tachikawa, T.; Yamash*ta, S.; Majima, T. J. Am. Chem. Soc. 2011, 133, 7197. (411) D’Arienzo, M.; Carbajo, J.; Bahamonde, A.; Crippa, M.; Polizzi, S.; Scotti, R.; Wahba, L.; Morazzoni, F. J. Am. Chem. Soc. 2011, 133, 17652. (412) Lv, K.; Xiang, Q.; Yu, J. Appl. Catal., B 2011, 104, 275. (413) Zhu, J.; Wang, J.; Lv, F.; Xiao, S.; Nuckolls, C.; Li, H. J. Am. Chem. Soc. 2013, 135, 4719. (414) Choi, W. Y.; Termin, A.; Hoffmann, M. R. J. Phys. Chem. 1994, 98, 13669. (415) Choi, J.; Park, H.; Hoffmann, M. R. J. Phys. Chem. C 2010, 114, 783. (416) Khan, S. U. M.; Al-Shahry, M.; Ingler, W. B. Science 2002, 297, 2243. (417) Zhao, W.; Ma, W. H.; Chen, C. C.; Zhao, J. C.; Shuai, Z. G. J. Am. Chem. Soc. 2004, 126, 4782. (418) Umebayashi, T.; Yamaki, T.; Itoh, H.; Asai, K. Appl. Phys. Lett. 2002, 81, 454. (419) Hong, X. T.; Wang, Z. P.; Cai, W. M.; Lu, F.; Zhang, J.; Yang, Y. Z.; Ma, N.; Liu, Y. J. Chem. Mater. 2005, 17, 1548. (420) Lin, L.; Lin, W.; Zhu, Y. X.; Zhao, B. Y.; Xie, Y. C. Chem. Lett. 2005, 34, 284. (421) Liu, G.; Wang, L. Z.; Yang, H. G.; Cheng, H. M.; Lu, G. Q. J. Mater. Chem. 2010, 20, 831.

(422) Finazzi, E.; Di Valentin, C.; Pacchioni, G. J. Phys. Chem. C 2009, 113, 220. (423) Cheung, S. H.; Nachimuthu, P.; Engelhard, M. H.; Wang, C. M.; Chambers, S. A. Surf. Sci. 2008, 602, 133. (424) Wang, W.; Lu, C.; Su, M.; Ni, Y.; Xu, Z. Chin. J. Catal. 2012, 33, 629. (425) Liu, G.; Yin, L. C.; Wang, J. Q.; Niu, P.; Zhen, C.; Xie, Y. P.; Cheng, H. M. Energy Environ. Sci. 2012, 5, 9603. (426) Nakamura, I.; Negishi, N.; Kutsuna, S.; Ihara, T.; Sugihara, S.; Takeuchi, E. J. Mol. Catal. A: Chem. 2000, 161, 205. (427) Jing, L. Q.; Xin, B. F.; Yuan, F. L.; Xue, L. P.; Wang, B. Q.; Fu, H. G. J. Phys. Chem. B 2006, 110, 17860. (428) Schaub, R.; Thostrup, P.; Lopez, N.; Laegsgaard, E.; Stensgaard, I.; Norskov, J. K.; Besenbacher, F. Phys. Rev. Lett. 2001, 87, 266104. (429) Wendt, S.; Schaub, R.; Matthiesen, J.; Vestergaard, E. K.; Wahlstrom, E.; Rasmussen, M. D.; Thostrup, P.; Molina, L. M.; Laegsgaard, E.; Stensgaard, I.; Hammer, B.; Besenbacher, F. Surf. Sci. 2005, 598, 226. (430) Setvin, M.; Aschauer, U.; Scheiber, P.; Li, Y. F.; Hou, W. Y.; Schmid, M.; Selloni, A.; Diebold, U. Science 2013, 341, 988. (431) Justicia, I.; Ordejon, P.; Canto, G.; Mozos, J. L.; Fraxedas, J.; Battiston, G. A.; Gerbasi, R.; Figueras, A. Adv. Mater. 2002, 14, 1399. (432) Cronemeyer, D. C. Phys. Rev. 1959, 113, 1222. (433) Zuo, F.; Bozhilov, K.; Dillon, R. J.; Wang, L.; Smith, P.; Zhao, X.; Bardeen, C.; Feng, P. Angew. Chem., Int. Ed. 2012, 51, 6223. (434) Zhu, S.; Liang, S.; Gu, Q.; Xie, L.; Wang, J.; Ding, Z.; Liu, P. Appl. Catal., B 2012, 119, 146. (435) Wu, X.-F.; Song, H.-Y.; Yoon, J.-M.; Yu, Y.-T.; Chen, Y.-F. Langmuir 2009, 25, 6438. (436) Wu, X.-F.; Chen, Y.-F.; Yoon, J.-M.; Yu, Y.-T. Mater. Lett. 2010, 64, 2208. (437) Lu, Q. P.; Lu, Z. D.; Lu, Y. Z.; Lv, L. F.; Ning, Y.; Yu, H. X.; Hou, Y. B.; Yin, Y. D. Nano Lett. 2013, 13, 5698. (438) Gao, H. T.; Li, X. H.; Lv, J.; Liu, G. J. J. Phys. Chem. C 2013, 117, 16022. (439) Gu, L.; Wang, J.; Cheng, H.; Zhao, Y.; Liu, L.; Han, X. ACS Appl. Mater. Interfaces 2013, 5, 3085. (440) Hou, J.; Yang, C.; Wang, Z.; Jiao, S.; Zhu, H. Appl. Catal., B 2013, 129, 333. (441) Yang, N.; Liu, Y.; Wen, H.; Tang, Z.; Zhao, H.; Li, Y.; Wang, D. ACS Nano 2013, 7, 1504. (442) Gu, L. A.; Wang, J. Y.; Cheng, H.; Zhao, Y. Z.; Liu, L. F.; Han, X. J. ACS Appl. Mater. Interfaces 2013, 5, 3085. (443) Wang, Z.; Huang, B.; Dai, Y.; Liu, Y.; Zhang, X.; Qin, X.; Wang, J.; Zheng, Z.; Cheng, H. CrystEngComm 2012, 14, 1687. (444) Liu, B. T.; Huang, Y. J.; Wen, Y.; Du, L. J.; Zeng, W.; Shi, Y. R.; Zhang, F.; Zhu, G.; Xu, X. H.; Wang, Y. H. J. Mater. Chem. 2012, 22, 7484. (445) Guo, W.; Zhang, F.; Lin, C.; Wang, Z. L. Adv. Mater. 2012, 24, 4761. (446) An, T.; Chen, J.; Nie, X.; Li, G.; Zhang, H.; Liu, X.; Zhao, H. ACS Appl. Mater. Interfaces 2012, 4, 5988. (447) Hou, J. G.; Yang, C.; Wang, Z.; Jiao, S. Q.; Zhu, H. M. Appl. Catal., B 2013, 129, 333. (448) Qi, L.; Yu, J.; Jaroniec, M. Phys. Chem. Chem. Phys. 2011, 13, 8915. (449) Liu, L.; Gu, X.; Sun, C.; Li, H.; Deng, Y.; Gao, F.; Dong, L. Nanoscale 2012, 4, 6351. (450) He, D.; Chen, M.; Teng, F.; Li, G.; Shi, H.; Wang, J.; Xu, M.; Lu, T.; Ji, X.; Lv, Y.; Zhu, Y. Superlattices Microstruct. 2012, 51, 799. (451) Yang, D. P.; Yang, N. T.; Ge, J. P. CrystEngComm 2013, 15, 7230. (452) Geim, A. K.; Novoselov, K. S. Nat. Mater. 2007, 6, 183. (453) Geim, A. K. Science 2009, 324, 1530. (454) Geim, A. K.; Grigorieva, I. V. Nature 2013, 499, 419. (455) Pumera, M. Chem. Soc. Rev. 2010, 39, 4146. (456) Chen, D.; Tang, L. H.; Li, J. H. Chem. Soc. Rev. 2010, 39, 3157. 9611

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Chemical Reviews

Review

(457) Xiang, Q. J.; Yu, J. G.; Jaroniec, M. Chem. Soc. Rev. 2012, 41, 782. (458) Zhang, N.; Zhang, Y. H.; Xu, Y. J. Nanoscale 2012, 4, 5792. (459) Wang, D. H.; Choi, D. W.; Li, J.; Yang, Z. G.; Nie, Z. M.; Kou, R.; Hu, D. H.; Wang, C. M.; Saraf, L. V.; Zhang, J. G.; Aksay, I. A.; Liu, J. ACS Nano 2009, 3, 907. (460) Ding, S. J.; Chen, J. S.; Luan, D. Y.; Boey, F. Y. C.; Madhavi, S.; Lou, X. W. Chem. Commun. 2011, 47, 5780. (461) Jiang, B.; Tian, C.; Pan, Q.; Jiang, Z.; Wang, J.-Q.; Yan, W.; Fu, H. J. Phys. Chem. C 2011, 115, 23718. (462) Iwaszuk, A.; Nolan, M. J. Mater. Chem. A 2013, 1, 6670. (463) Kawahara, T.; Konishi, Y.; Tada, H.; Tohge, N.; Nishii, J.; Ito, S. Angew. Chem., Int. Ed. 2002, 41, 2811. (464) Hurum, D. C.; Agrios, A. G.; Gray, K. A.; Rajh, T.; Thurnauer, M. C. J. Phys. Chem. B 2003, 107, 4545. (465) Scanlon, D. O.; Dunnill, C. W.; Buckeridge, J.; Shevlin, S. A.; Logsdail, A. J.; Woodley, S. M.; Catlow, C. R. A.; Powell, M. J.; Palgrave, R. G.; Parkin, I. P.; Watson, G. W.; Keal, T. W.; Sherwood, P.; Walsh, A.; Sokol, A. A. Nat. Mater. 2013, 12, 798. (466) Ohno, T.; Tokieda, K.; Higashida, S.; Matsumura, M. Appl. Catal., A 2003, 244, 383. (467) Kho, Y. K.; Iwase, A.; Teoh, W. Y.; Madler, L.; Kudo, A.; Amal, R. J. Phys. Chem. C 2010, 114, 2821. (468) Wu, C. Y.; Yue, Y. H.; Deng, X. Y.; Hua, W. M.; Gao, Z. Catal. Today 2004, 93−5, 863. (469) Zachariah, A.; Baiju, K. V.; Shukla, S.; Deepa, K. S.; James, J.; Warrier, K. G. K. J. Phys. Chem. C 2008, 112, 11345. (470) Liu, G.; Yan, X. X.; Chen, Z. G.; Wang, X. W.; Wang, L. Z.; Lu, G. Q.; Cheng, H. M. J. Mater. Chem. 2009, 19, 6590. (471) Su, R.; Bechstein, R.; So, L.; Vang, R. T.; Sillassen, M.; Esbjornsson, B.; Palmqvist, A.; Besenbacher, F. J. Phys. Chem. C 2011, 115, 24287. (472) Tian, H.; Zhao, G.; Zhang, Y.-n.; Wang, Y.; Cao, T. Electrochim. Acta 2013, 96, 199. (473) Yang, D. J.; Liu, H. W.; Zheng, Z. F.; Yuan, Y.; Zhao, J. C.; Waclawik, E. R.; Ke, X. B.; Zhu, H. Y. J. Am. Chem. Soc. 2009, 131, 17885. (474) Liu, H. W.; Zheng, Z.; Yang, D. J.; Ke, X. B.; Jaatinen, E.; Zhao, J. C.; Zhu, H. Y. ACS Nano 2010, 4, 6219. (475) Liu, B.; Khare, A.; Aydil, E. S. ACS Appl. Mater. Interfaces 2011, 3, 4444. (476) Buonsanti, R.; Grillo, V.; Carlino, E.; Giannini, C.; Curri, M. L.; Innocenti, C.; Sangregorio, C.; Achterhold, K.; Parak, F. G.; Agostiano, A.; Cozzoli, P. D. J. Am. Chem. Soc. 2006, 128, 16953. (477) Buonsanti, R.; Grillo, V.; Carlino, E.; Giannini, C.; Gozzo, F.; Garcia-Hernandez, M.; Garcia, M. A.; Cingolani, R.; Cozzoli, P. D. J. Am. Chem. Soc. 2010, 132, 2437. (478) Buonsanti, R.; Snoeck, E.; Giannini, C.; Gozzo, F.; GarciaHernandez, M.; Garcia, M. A.; Cingolani, R.; Cozzoli, P. D. Phys. Chem. Chem. Phys. 2009, 11, 3680. (479) Yuan, S. J.; Chen, J. J.; Lin, Z. Q.; Li, W. W.; Sheng, G. P.; Yu, H. Q. Nat. Commun. 2013, 4, doi:10.1038/ncomms3249. (480) Robert, D.; Piscopo, A.; Weber, J. V. Sol. Energy 2004, 77, 553. (481) Zhou, P.; Zhu, X.; Yu, J.; Xiao, W. ACS Appl. Mater. Interfaces 2013, 5, 8165. (482) Yang, W.; Wang, Y.; Shi, W. CrystEngComm 2012, 14, 230. (483) Feng, S.; Yang, J.; Liu, M.; Zhu, H.; Zhang, J.; Li, G.; Peng, J.; Liu, Q. Thin Solid Films 2012, 520, 2745. (484) De Marco, L.; Manca, M.; Buonsanti, R.; Giannuzzi, R.; Malara, F.; Pareo, P.; Martiradonna, L.; Giancaspro, N. M.; Cozzoli, P. D.; Gigli, G. J. Mater. Chem. 2011, 21, 13371. (485) Agosta, R.; Giannuzzi, R.; De Marco, L.; Manca, M.; Belviso, M. R.; Cozzoli, P. D.; Gigli, G. J. Phys. Chem. C 2013, 117, 2574. (486) De Marco, L.; Manca, M.; Giannuzzi, R.; Belviso, M. R.; Cozzoli, P. D.; Gigli, G. Energy Environ. Sci. 2013, 6, 1791. (487) Oh, J. K.; Lee, J. K.; Kim, H. S.; Han, S. B.; Park, K. W. Chem. Mater. 2010, 22, 1114. (488) Koo, B.; Park, J.; Kim, Y.; Choi, S. H.; Sung, Y. E.; Hyeon, T. J. Phys. Chem. B 2006, 110, 24318.

(489) Liu, J.; Liu, X.-W. Adv. Mater. 2012, 24, 4097. (490) Chen, J. S.; Lou, X. W. Electrochem. Commun. 2009, 11, 2332. (491) Deng, D.; Kim, M. G.; Lee, J. Y.; Cho, J. Energy Environ. Sci. 2009, 2, 818. (492) Ohtsuki, C.; Iida, H.; Hayakawa, S.; Osaka, A. J. Biomed. Mater. Res. 1997, 35, 39. (493) Li, P. J.; Ohtsuki, C.; Kokubo, T.; Nakanishi, K.; Soga, N.; Degroot, K. J. Biomed. Mater. Res. 1994, 28, 7. (494) Jiao, W.; Li, N.; Wang, L.; Wen, L.; Li, F.; Liu, G.; Cheng, H. M. Chem. Commun. 2013, 49, 3461. (495) Xi, G. C.; Ye, J. H. Chem. Commun. 2010, 46, 1893. (496) Li, R. G.; Zhang, F. X.; Wang, D. G.; Yang, J. X.; Li, M. R.; Zhu, J.; Zhou, X.; Han, H. X.; Li, C. Nat. Commun. 2013, 4, doi:10.1038/ncomms2401. (497) Wang, D.; Jiang, H.; Zong, X.; Xu, Q.; Ma, Y.; Li, G.; Li, C. Chem.Eur. J. 2011, 17, 1275. (498) Zhou, X. M.; Lan, J. Y.; Liu, G.; Deng, K.; Yang, Y. L.; Nie, G. J.; Yu, J. G.; Zhi, L. J. Angew. Chem., Int. Ed. 2012, 51, 178. (499) Jiang, J.; Zhao, K.; Xiao, X. Y.; Zhang, L. Z. J. Am. Chem. Soc. 2012, 134, 4473. (500) Kuo, C. H.; Yang, Y. C.; Gwo, S.; Huang, M. H. J. Am. Chem. Soc. 2011, 133, 1052. (501) Liang, Y. H.; Shang, L.; Bian, T.; Zhou, C.; Zhang, D. H.; Yu, H. J.; Xu, H. T.; Shi, Z.; Zhang, T. R.; Wu, L. Z.; Tung, C. H. CrystEngComm 2012, 14, 4431. (502) Zhang, Y.; Deng, B.; Zhang, T. R.; Gao, D. M.; Xu, A. W. J. Phys. Chem. C 2010, 114, 5073. (503) Zheng, Z. K.; Huang, B. B.; Wang, Z. Y.; Guo, M.; Qin, X. Y.; Zhang, X. Y.; Wang, P.; Dai, Y. J. Phys. Chem. C 2009, 113, 14448. (504) Zhao, H.; Yin, W.; Zhao, M.; Song, Y.; Yang, H. Appl. Catal., B 2013, 130, 178. (505) Wang, J.; Teng, F.; Chen, M.; Xu, J.; Song, Y.; Zhou, X. CrystEngComm 2013, 15, 39. (506) Wang, H.; Lang, X.; Gao, J.; Liu, W.; Wu, D.; Wu, Y.; Guo, L.; Li, J. Chem.Eur. J. 2012, 18, 4620. (507) Bi, Y.; Hu, H.; Ouyang, S.; Jiao, Z.; Lu, G.; Ye, J. J. Mater. Chem. 2012, 22, 14847.

9612

dx.doi.org/10.1021/cr400621z | Chem. Rev. 2014, 114, 9559−9612

Titanium dioxide crystals with tailored facets. - PDF Download Free (2024)

References

Top Articles
Latest Posts
Article information

Author: Mr. See Jast

Last Updated:

Views: 5964

Rating: 4.4 / 5 (55 voted)

Reviews: 86% of readers found this page helpful

Author information

Name: Mr. See Jast

Birthday: 1999-07-30

Address: 8409 Megan Mountain, New Mathew, MT 44997-8193

Phone: +5023589614038

Job: Chief Executive

Hobby: Leather crafting, Flag Football, Candle making, Flying, Poi, Gunsmithing, Swimming

Introduction: My name is Mr. See Jast, I am a open, jolly, gorgeous, courageous, inexpensive, friendly, homely person who loves writing and wants to share my knowledge and understanding with you.