This content has been downloaded from IOPscience. Please scroll down to see the full text.

Download details:

IP Address: 141.24.47.169

This content was downloaded on 13/07/2014 at 12:01

Please note that terms and conditions apply.

Observation of defect state in highly ordered titanium dioxide nanotube arrays

View the table of contents for this issue, or go to the journal homepage for more

2014 Nanotechnology 25 275603

(http://iopscience.iop.org/0957-4484/25/27/275603)

Home Search Collections Journals About Contact us My IOPscience

Observation of defect state in highly orderedtitanium dioxide nanotube arrays

Hongchao Zhang1,3, Min Zhou2,3, Qun Fu1, Bo Lei1, Wei Lin1, Heshuai Guo1,Minghong Wu1 and Yong Lei1,2

1 School of Environmental and Chemical Engineering, Shanghai University, Shanghai 200444, People’sRepublic of China2 Institute for Physics and IMN MacroNano®, Technical University of Ilmenau, Ilmenau 98693, Germany

E-mail: yong.lei@tu-ilmenau.de

Received 9 March 2014, revised 17 April 2014Accepted for publication 12 May 2014Published 24 June 2014

AbstractFor the first time, a conductive-substrate induced electro-deposition approach is employed toachieve highly ordered TiO2 nanotube arrays based on an anodic aluminum oxide template.Different from other methods, the morphology and parameters of arrays can be adjusted easilythrough changing the exposure area of a conductive layer. All these arrays are used as matrixesto explore the defect state emission by photoluminescence (PL) spectra. Interestingly, we findthat the emission from blue edge to red edge (∼450 nm–600 nm) are apparently quenched in theordered nanotube arrays, especially when compared to the PL spectra of nanowire arrays, singlenanotube and nanoparticles. This distinct result originates from passivation of oxygen vacanciesresiding along the tube walls when the tubes are interconnected, which is further evidenced bythe observation of PL spectra with crystalline phase and sintering. The passivation of defectssuggests valuable charge transport perpendicular to the long axis of the tubes in the orderedarrays. This point is particularly significant to the design of highly efficient devices and theapplications in various energy-related fields.

S Online supplementary data available from stacks.iop.org/NANO/25/275603/mmedia

Keywords: nanotube array, defect state, photoluminescence

(Some figures may appear in colour only in the online journal)

1. Introduction

Titanium dioxide (TiO2), also known as titanium (IV) oxideor titania, is the naturally existing oxide of titanium. With itscommercial production appearing in the early 20th century,this material has drawn extensive attention due to its wideapplications ranging from energy to environmental fields[1–4]. Recently, enormous efforts have been devoted to thestudies of TiO2 nanomaterials, which has led to many pro-mising applications in areas of photovoltaics, photo-/photo-electro-catalysis, photo-/electro-chromics, sensors, electronicdata storage medium, battery, etc. Aiming at pursuing highconversion efficiency in all these areas, a deep understandingof defect states in both bulk and surface is primarily important

[5–8], because the characteristic electronic and geometricalstructure associated with defect states endow TiO2 withunique macroscopic properties [9, 10]. For example, oxygenvacancies are able to induce room temperature ferromagnet-ism of TiO2 [11]. Intra-band-gap states originating fromdefects lead to charge recombination and the transport prop-erties related to phonons in TiO2-based dye-sensitized solarcells and photo-/photoelectrochemical-catalytic cells [12–14].Excess electrons located on the oxygen vacancy states wouldaffect surface absorption and reactivity of key adsorbates,such as O2 or H2O on TiO2 [15, 16]. Therefore, there is agreat significance in deeply understanding defect states andexploiting the dependence of the properties of TiO2 on thedefect states.

Among the unique properties of TiO2, the movement ofcharge carries is primarily important for many applications

Nanotechnology

Nanotechnology 25 (2014) 275603 (10pp) doi:10.1088/0957-4484/25/27/275603

3 These authors contributed equally to this work.

0957-4484/14/275603+10$33.00 © 2014 IOP Publishing Ltd Printed in the UK1

[1, 17]. Theoretical and experimental results indicate a strongrelationship between the defect states and charge transport ofTiO2 [13, 18, 19]. With regard to conventional TiO2 nano-particles, inefficient interparticle electron transfer has beenknown as an obstacle for charge collection [6, 20]. It isexpected that one-dimensional (1D) ordered structure, espe-cially nanotube arrays with high surface-to-volume ratio,would result in fast charge transport [21]. However, previousstudies show that charge transport in nanotube-array-baseddevices is not obviously faster than common nanoparticle-based devices [22]. In general, different distributions of trapstate can accelerate the charge transport in TiO2 nanotubesthan in TiO2 nanoparticles, while excitonic states deceleratethe charge transport in TiO2 nanotubes [21–23]. It has beenimplied that charge transport in nano-TiO2 occurs by diffu-sion mediated by shallow traps, but meanwhile, deep trapslead to unfavorable recombination [24–26]. Trapping anddetrapping states may be attributed to lower mobility in nano-TiO2 than the bulk material, and thus it is of great interest tounderstand how defect states of conventional TiO2 nanotubesmay differ from those of TiO2 nanoparticles and bulkmaterials.

In previous work, our group has employed photo-luminescence (PL) spectroscopy to illuminate the defect statesof TiO2 nanowires and anodic aluminum oxide (AAO) tem-plates [27, 28]. We presented a growth model of TiO2

nanowires in an AAO template by sol-gel approach to clearlyexplain why the nanowires grew from center to the walls ofpores. This approach was not suitable for uniform nanotubearrays in a large scale. Inspired by these studies, a novelconductive-substrate induced electro-deposition approach is

rationally designed to obtain controllable TiO2 nanotubearrays. Considering that materials can only be deposited onconductive substrate within an electric-field-assisted hydro-lysis approach, which is quite different from the electrolyteelectrolysis in conventional electro-deposition, there are twogrowth directions within the AAO template as shown inscheme 1. One is along the conductive substrate, the other oneis vertical to the conductive substrate. If the spurring time ofsubstrate metal (e.g. Cu) can be adjusted randomly, theexposed area of conductive substrate would be changed cor-respondingly, along with the horizontal growth rate. Suitablethickness of Cu contributes to proper horizontal growth,leading to nanotube array. Exposure of too little conductivesubstrate results in few growth trends, so we cannot get any1D nanostructures. Meanwhile, long spurring time of Curesults in continuous conductive layer, and thus nanowirearrays can be finally obtained rather than nanotube arrays.This approach is obviously different from other widely usedmethods for 1D nanostructures, such as sol-gel, hydrothermal,and atomic layer deposition, where the materials fully coverand depend more on the template. It is expected that a facilepreparation of pore-through nanotube arrays can be realizedwithout any complex processes and expensive equipment,serving as a matrix with clean surface for the full studies ofbasic physical/chemical theory and practical application.

Herein, a conductive-substrate induced electro-depositionapproach is first employed to achieve highly ordered nano-tube arrays based on AAO template. Different from othermethods, we can easily achieve 1D nanotube arrays throughadjusting the exposure area of the conductive layer. All thesearrays serve as matrixes to explore the dependence of defect

Scheme 1. The schematic diagram of growth mechanism demonstration of TiO2 nanotube arrays. (a) deposition of Cu on the back side of thetemplate through physical vapor deposition; (b) electro-deposition; (c) removing the template.

2

Nanotechnology 25 (2014) 275603 H Zhang et al

state emission on crystalline phase and sintering by PLspectra, which can provide significant information forintrinsic carrier transport and further expanding the applica-tions of TiO2 nanotube arrays in energy and environmentalfields.

2. Experimental details

2.1. Fabrication of AAO template

The template is prepared by a two-step electro-chemicalanodization process of aluminum foil [29–31]. Before oxidi-zation, the aluminum foils are rinsed in acetone and water toremove the adsorbed contamination and electro-polished inperchloric acid (in ethanol with volume ratio of 1:9) under18 V. Then the alumina foil is oxidized twice under 55 V in0.3 mol L−1 oxalic acid solution for 12 h and 6 h, respectively.The alumina originating from the first oxidation is removed toform patterned reliefs to induce the second oxidation. Afteroxidization, a template with 90–100 nm pore diameter and135 nm interpore distance can be obtained. The pore-widen-ing process is conducted in 5 wt% phosphoric acid at 32 °C toremove the back barrier layer and enlarge the diameters ofnanopores.

2.2. Deposition of metal electrode

Before the electrodeposition, we sputter 40 nm Cu film on theback of the AAO template as an electrode by vacuum thermalevaporation using physical vapor deposition. During thethermal evaporation, the vacuum degree in the vacuumchamber is 8 × 10−4 Pa, and the evaporation rate of0.3∼ 0.5 nm s−1. The thickness of Cu can be changed by thedeposition time.

2.3. Electrochemical deposition

The electro-deposition is conducted in the constant voltagemode with a three-electrode system of potentiostatic system(PGSTAT302N), containing an AAO working electrode, Ag/AgCl reference electrode and a platinum foil counter elec-trode. The electrolyte consists 0.05 g L−1 TiF4 and 0.05 g L−1

NiCl2·6H2O, where the pH value is adjusted to 1.8 withNaHCO3.

2.4. Removing the template

The template is removed by NaOH (1 mol L−1) after electro-deposition for further characterization.

2.5. Characterization

The scanning electron microscope (SEM) images are taken byJSM-6700F, JEOL (Japan Electron Optics Laboratory Co.Ltd). The transmission electron microscope (TEM) imagesand selected area electron diffraction (SAED) patterns areachieved on JEM-2010F (JEOL). The x-ray diffraction(XRD) measurements are recorded on a 3KW D/MAX2200VPC (RIGAKU Corporation) x-ray diffractometer (Cu-Kαradiation) equipment. The photoluminescence (PL) measure-ments are carried out on a Hitachi, F-70 000. The excitationwavelength is 330 nm.

3. Results and discussion

The free-standing ordered TiO2 nanotube arrays were fabri-cated by the conductive-substrate induced electro-depositionapproach according to the following steps (scheme 2), and thedetails can be found in the experimental sections.

Inspired by the consideration that the horizontal growthrate of materials is directly related to the conductive substrate,we found the sputtering time of Cu, which is related to thethickness of the Cu layer, indeed plays an essential role in thefinal morphology of arrays during the electro-depositionprocesses. Under the same deposition time (TiO2 for 10 min),the dependence of the final morphology on the thickness ofthe Cu layer can be observed clearly, as shown in figure 1(a).When the thickness of Cu is less than 30 nm, it is hard to getany 1D nanostructure through electrochemical deposition(figure 1(b)) no matter how long the deposition time is. This isbecause the exposure area of conductive Cu is too small tosupport electric-field-assisted hydrolysis processes. If thethickness increases to 30–50 nm, the growth of TiO2 alongthe template can be seen in figure 1(c) due to having enoughexposure of Cu as conductive sites for electro-deposition,which is evidenced by the shortened diameter of pores fromthe back view of the template after Cu deposition as shown inthe inset image of figure 1(c), and thus the formation ofnanotube arrays is achievable within expectation. Moreover,nanowires appear when the thickness of Cu is more than

Scheme 2. Schematic diagram of cross section of TiO2 nanotubes arrays.

3

Nanotechnology 25 (2014) 275603 H Zhang et al

50 nm (figure 1(d)) owing to a too-rapid hydrolysis rateinduced by the continuous conductive layer. Totally differentfrom other electro-deposition processes, the types of nanos-tructure are determined by the thickness of the conductivelayer rather than only the deposition time, which has pre-viously been long neglected.

Following these steps, the nanotube arrays can beachievable. Each nanotube is uniform and nearly parallel, butconnects with one another. The hollow structure is clearlyobserved from the top view in figure 1(c). The out-diameter ofeach nanotube is about 90–100 nm, which is in good agree-ment with the pore diameter of the AAO template. Thethickness of each nanotube is ∼36 nm and can be adjustedeasily by changing the deposition time of TiO2 (figure S1),indicating the high controllability of this conductive-substrateinduced electro-deposition approach.

With the help of this fabrication approach, various TiO2

nanotube arrays can be obtained so as to fully understand thedefect state emission. The as-obtained TiO2 nanotube arrayswith 10 min deposition time were post-treated for 5 h at hightemperatures of 350 °C, 450 °C, 550 °C, 650 °C and 750 °C,respectively. The x-ray powder diffraction (XRD) in

figure 2(a) shows the pure rutile phase TiO2 when the tem-perature is higher than 550 °C. If the annealing temperature isbelow 550 °C, the nanotube arrays consist of both the rutilephase and the anatase phase. Shown in figures 2(b)–(f) arecorresponding SEM images. The overall nanotube morphol-ogy is maintained, implying the obtained TiO2 arrays have ahighly stable structure. However, when the annealing tem-perature is higher than 550 °C, a close observation revealsthat the slight collapse of the nanotube array can be largelyattributed to deintercalation of H2O and phase transition. Forconvenience, hereafter, we designate various TiO2 nanotubearrays as ‘NT-Temperature’, where NT refers to nanotube andthe temperature is the annealing temperature.

In order to confirm the nature of defect states, photo-luminescence (PL) spectroscopy was performed on nanotubearrays with different annealing temperatures (figure 3(a)). Theobserved line shapes of the five samples are clearly similar,except for the intensity. From the projection on PL intensity-temperature section, we find the intensity of peak increasedrapidly from 350 °C to 450 °C until reaching the maximumvalue at 550 °C, and then decreased at higher temperatures.Since TiO2 is an indirect bandgap semiconductor, the free

Figure 1. (a) The relationship between the thickness of the spurring Cu layer at the bottom of AAO templates and types of nanostructuresobtained by electrochemical deposition. (b) The typical SEM image of the samples after electro-depositing for 10 min with differentthicknesses of Cu: (b) 10 nm and (c) 40 nm.The right inset image is a back view of the template, while the left one is a side view of thetemplate. (d) 70 nm. The templates in both (c) and (d) have already been removed.

4

Nanotechnology 25 (2014) 275603 H Zhang et al

exciton recombination can be excluded [11]. The observedbroad visible luminescence is primarily determined by thedefect states. Detailed analysis of the PL spectrum providesmore insight into the nature of defect states in differentsamples, and thus the deconvolution of the peak is necessary.Taking NT-350 for example, the broad emission analyzedwith multi-peak Gaussian fitting method in figure 3(b) can befitted with three Gaussian sub-bands centered at 415 nm (peak1), 471 nm (peak 2) and 531 nm (peak 3).

In general, PL spectra of anatase TiO2 are attributed tothree kinds of physical origins: (I) self-trapped excitonslocated at TiO6 octahedra (STE); (II) oxygen vanancies; and(III) surface state [27, 28]. As for the 415 nm sub-band, it hasbeen reported from PL spectra of TiO2 anatase single crystals,nanowire arrays and nanoparticles that this peak should beassigned to self-trapped excitons localized on TiO6 octahedra[11, 32]. Meanwhile, the PL bands at 471 nm and 531 nmshould originate from the oxygen vanacies associated with

Figure 2. (a) XRD patterns and (b)–(f) SEM images of TiO2 nanotube arrays annealing for 5 h at different temperatures: (b) 350 °C; (c)450 °C; (d) 550 °C; (e) 650 °C; (f) 750 °C.

5

Nanotechnology 25 (2014) 275603 H Zhang et al

Ti3+/F and F+, respectively [33]. Simply, due to the loss of anO atom in TiO2 lattice, the electron pair trapped in thevacancy cavity (VO) will generate an F center [34]. One of theelectrons in F centre may occupy the adjacent Ti4+ ion,yielding a Ti3+ center and a F+ (a single electron associatedwith oxygen vacancies) center, which will form shallow anddeep trap states, respectively, as shown in the followingequations.

+ →+ → +

+ →+ →

−

+ + +

− +

+ − +

V e F

F Ti F TiV e F

Ti e Ti

2O

O

4 3

4 3

Hence, Ti3+ center and F+ center are formed due to the loss ofan oxygen atom. Shallow and deep-trap states are establishedat 0.51 [35] and 0.8 eV [36] below the conduction band. Thesub-band centered at 471 nm, which is coincident with shal-low trap, can be assigned to Ti3+ states, while PL emissioncentered at 531 nm results from the deep-trap state associatedwith the F+ center [27]. The literature has suggested that the415 nm sub-band probably originates from the intrinsic states

in bulk rather than from surface states [27], while the 471 nmand 531 nm sub-bands of anatase TiO2 powders can beassigned to surface defect states [37]. Some studies also reportan intraband surface state of a nanocrystalline anatase TiO2

film and identify the trap with Ti atoms adjacent to oxygenvanancies at the surface. Hence, it is reasonable to assignthese two bands to the oxygen vanancies on the surface areaof the nanotube arrays. Overall, all the locations of these trapcenters responsible for the strong visible PL from TiO2

nanotube arrays can be understood from the schematic banddiagram shown in figure 3(d). The photogenerated electronsare initially excited to the conduction band of TiO2 on irra-diating UV-light (330 nm) and then relaxed to the defectstates.

Quite similar with NT-350, observation of defect statescan also be seen from the broad peak in the PL spectrum ofNT-550 in figure 3(c), which is fitted with only two Gaussianbands centered at 422 nm (peak 1) and 468 nm (peak 2). It isof interest to pay attention to the following two parts: (I) Theoverall intensity of the broad peak and the separated intensityof the two Gauss fit peaks are stronger than those of NT-350,

Figure 3. (a) Room temperature PL spectra of TiO2 nanotube arrays annealing at different temperatures (350, 450, 550, 650, 750 °C); (b), (c)Comparison of PL intensity for NT-350 and NT-550, the spectra of which are fitted with Gaussian peaks. The deconvoluted spectra areshown with solid lines and experimental data are shown with symbols. (d) Schematic diagram illustrating the physical origin of visible PLemissions.

6

Nanotechnology 25 (2014) 275603 H Zhang et al

implying the existing of more intrinsic STE and oxygenvacancies at the surface. Considering the constant morphol-ogy of nanotube arrays, this increase results from the gradu-ally increased crystallization of TiO2, which contributes moreto the PL signals; (II) The area ratio of 422 nm sub-band to468 nm sub-band is obviously increased. Note that the peakcentered at 422 nm is more related to intrinsic STE, while thepeak around 468 nm results from oxygen vacancies at thesurface. It is concluded that the concentration of surfaceoxygen vanacies increases more rapidly than STE with risingtemperature. (III) The sub-peak centered at ∼531 nm is hardto be seen in the detailed spectrum. Even more remarkable,looking through all the PL spectra of TiO2 nanotube arrays,we find an extremely distinct point that the emissions fromblue edge to red edge (∼450 nm–600 nm) are apparentlyquenched in the ordered nanotube arrays, especially whencomparing them to the PL spectra of TiO2 nanowire arrays[27], nanoribbons [11], a single nanotube [6] and nano-particles [32]. The PL bands induced by other kinds ofdefects, such as hydroxyl (OH) species, are also missing inthe spectra. It is regarded that photoluminescence is mostly asurface phenomenon, and thus a change in the surfaceenvironment would lead to significant influences on the PLprocess. Although the nanotube arrays possess a large surfacearea, the defect-related PL emission is found to be weaker.

Many factors have been regarded to influence the PLintensity, such as morphology [6], incident electric-fieldvector [6, 38], coordination number [39] and so on. MaHaleet al has checked the possible effects of PL spectra undervertically and horizontally polarized incidental light, respec-tively [6]. There is no difference in the PL spectra for thesetwo experiments; thus, the difference in the spectra of nano-tube arrays and other morphologies is not a consequence ofthe different alignments of the tubes relative to the incidentalelectric field vector. Additionally, PL quenching by oxygenmay require a sufficient number of O2 binding sites, wherehole traps associated with oxygen vacancies are located on(101) facets, while electron traps associated with under-coordinated Ti atoms prevail on the (001) surfaces [39].Considering the polycrystalline nature of nanotubes, which isevidenced by polycrystalline rings in SAED patterns of figureS2, the influences of different distribution of emissive trapsdue to coordination number on independent facets can also beexcluded when compared to conventional TiO2 morphology.Alternatively, the interparticle connections in the nanotubearrays could passivate a subset of more blue-to-red emittingtraps, or permit interparticle transport that competes with theradiative recombination from these traps. The diminished red-edge PL in the ordered arrays may result from passivation oftraps residing along the tube walls when the tubes areinterconnected.

To evidence this hypothesis, we should pay more atten-tion on at least two results. One is the PL spectrum of samplesunder annealing temperatures higher than 550 °C. When theannealing temperature keeps rising, the resulting intensity ofblue-to-red edge PL is weaker and weaker (figure 3(a)). Thesechanges are attributed to more exposure of samples in air athigh temperatures, which allow increased Ti4+ transformed

from Ti3+ and corresponding phase transition from anatase torutile as shown in the XRD patterns in figure 2(a). Moreimportantly, this transformation leads to the change of themorphology of the nanotube array. Except for the effects ofphase transition, more and more interconnection betweennanotubes further decreases the surface area. As indicated bythe comparison between NT-650 and NT-750, the emissionsat the blue-to-red edge decrease because all these detectableemissions result from defect states at the surface. The secondevidence comes from the dependence of PL spectra on ther-mal treatment time. The nanotube arrays were annealed at550 °C for different periods of time as shown in figure 4,figure 5 shows the similar trend of the change on the mor-phology and phase. The blue-shift and weaker peaks evidencethe decrease of blue-to-red edge PL and as-mentionedconclusion.

Well-known research has indicated that deep traps lead tounfavorable recombination [24–26]. The dominating chargecarrier recombination is the Shockley–Read–Hall recombi-nation, which occurs when a charge is trapped by the defect[40, 41]. From this viewpoint, decreasing deep-trap defectstates are essential to reduce the recombination. The highlyordered TiO2 nanotube arrays obtained in our approach arebeneficial to suppress this PL radiative recombination andhence to increase the lifetime of the charge carriers fortransport, especially for the one between interconnectednanotubes within the arrays. As a result, our data showed thatthere is significant charge transport perpendicular to the longaxis of the tubes in highly ordered arrays [6], which is sig-nificant to the design of highly efficient devices.

Although a deep understanding of the role of defect statesis in its infancy and there are some contradictory statementson the role of oxygen vacancies related to the performance ofsemiconductors, it has been demonstrated that the defectstates of TiO2 play an important role on controllable adjust-ment of macroscopic properties, such as visible lightresponse, charge separation, charge transport and absorptionselectivity. We are looking forward to further efforts in thisarea to rationally engineer the defects in TiO2, aiming to tuneits properties in a desired manner for advancing environ-mental and energy-related applications.

4. Conclusions

In conclusion, we first employed a conductive-substrateinduced electro-deposition approach to achieve highlyordered TiO2 nanotube arrays based on an anodic aluminumoxide template. This approach is totally different from othermethods, such as sol-gel, hydrothermal and atomic layerdeposition, where the materials fully cover and depend moreon the template. It is expected that a facile preparation ofpore-through nanotube arrays can be realized without anycomplex processes and expensive equipment, while theparameters of nanotube arrays can be adjusted easily bychanging the exposure area of the conductive layer. All thesearrays can serve as a matrix with clean surfaces for the fullstudies of basic physical and chemical theory and practical

7

Nanotechnology 25 (2014) 275603 H Zhang et al

application. The photoluminescence (PL) spectroscopy is anefficient tool to explore the dependence of defect-stateemissions on the various TiO2 nanotube arrays. The PLemission of TiO2 nanotube arrays are attributed to two kindsof physical origins: (I) self-trapped excitons located at TiO6

octahedra (STE) and (II) oxygen vacancies, including Ti3+, Fand F+. Interestingly, we find the emission from blue edge tored edge (∼450 nm–600 nm) is apparently quenched in theordered nanotube arrays, especially when compared to the PL

spectra of nanowire array, single nanotubes and nanoparticles.This distinct result originates from passivation of oxygenvacancies residing along the tube walls when the tubes areinterconnected, which is further evidenced by the observationof PL spectra with crystalline phase and sintering. The pas-sivation of defects suggest significant charge transport per-pendicular to the long axis of the tubes in the ordered arrays,which is significant to the design of highly efficient devicesand the applications in various energy-related fields.

Figure 4. (a) XRD patterns and (b)–(f) SEM images of TiO2 nanotube arrays annealing at 550 °C for different times: (b) 3 h; (c) 8 h; (d) 10 h;(e) 12 h; (f) 16 h.

8

Nanotechnology 25 (2014) 275603 H Zhang et al

Acknowledgments

This work is supported by Shanghai Thousand Talent Planand Program, Innovative Research Team in University (No.IRT13078), European Research Council (ThreeDsurface:240144), BMBF (ZIK-3DNanoDevice: 03Z1MN11), BMBF(Meta-ZIK-BioLithoMorphie: 03Z1M511), and Volkswagen-Stiftung (Herstellung funktionaler Oberflächen: I/83 984).

References

[1] Chen X and Mao S S 2007 Titanium dioxide nanomaterials:synthesis, properties, modifications, and applications Chem.Rev. 107 2891

[2] Fujishima A and Honda K 1972 Electrochemical photolysis ofwater at a semiconductor electrode Nature 238 37

[3] Tachikawa T, Wang N, Yamashita S, Cui S-C and Majima T2010 Design of a highly sensitive fluorescent probe forinterfacial electron transfer on a TiO2 surface Angew. Chem.Int. Ed. 49 8593

[4] Xu Y, Memarzadeh Lotfabad E, Wang H, Farbod B, Xu Z,Kohandehghan A and Mitlin D 2013 Nanocrystallineanatase TiO2: a new anode material for rechargeable sodiumion batteries Chem. Comm. 49 8973

[5] Pan X, Yang M-Q, Fu X, Zhang N and Xu Y-J 2013 DefectiveTiO2 with oxygen vacancies: synthesis, properties andphotocatalytic applications Nanoscale 5 3601

[6] Mercado C C, Knorr F J and McHale J L 2012 Observation ofcharge transport in single titanium dioxide nanotubes bymicro-photoluminescence imaging and spectroscopy ACSNano 6 7270

[7] Zhou M, Bao J, Bi W, Zeng Y, Zhu R, Tao M and Xie Y 2012Efficient water splitting via a heteroepitaxial BiVO4

photoelectrode decorated with Co-Pi catalystsChemSusChem 5 1420

[8] Chen X, Liu L, Yu P Y and Mao S S 2011 Increasing solarabsorption for photocatalysis with black hydrogenatedtitanium dioxide nanocrystals Science 331 746

[9] Pacchioni G 2003 Oxygen vacancy: the invisible agent onoxide surfaces ChemPhysChem 4 1041

[10] Polarz S, Strunk J, Ischenko V, van den Berg M W E,Hinrichsen O, Muhler M and Driess M 2006 On the role of

oxygen defects in the catalytic performance of zinc oxideAngew. Chem. Int. Ed. 45 2965

[11] Santara B, Giri P K, Imakita K and Fujii M 2013 Evidence ofoxygen vacancy induced room temperature ferromagnetismin solvothermally synthesized undoped TiO2 nanoribbonsNanoscale 5 5476

[12] Jing L, Xin B, Yuan F, Xue L, Wang B and Fu H 2006 Effectsof surface oxygen vacancies on photophysical andphotochemical processes of zn-doped TiO2 nanoparticlesand their relationships J. Phys. Chem. B 110 17860

[13] Wang J, Liu P, Fu X, Li Z, Han W and Wang X 2008Relationship between oxygen defects and the photocatalyticproperty of ZnO nanocrystals in nafion membranesLangmuir 25 1218

[14] Nowotny J 2008 Titanium dioxide-based semiconductors forsolar-driven environmentally friendly applications: impactof point defects on performance Energy Environ. Sci.1 565

[15] Zhang Z, Bondarchuk O, White J M, Kay B D and Dohnálek Z2006 Imaging adsorbate O−H bond cleavage: methanol onTiO2(110) J. Am. Chem. Soc. 128 4198

[16] Schaub R, Thostrup P, Lopez N, Lægsgaard E, Stensgaard I,Nørskov J K and Besenbacher F 2001 Oxygen vacancies asactive sites for water dissociation on rutile Phys. Rev. Lett.87 266104

[17] Chen X, Shen S, Guo L and Mao S S 2010 Semiconductor-based photocatalytic hydrogen generation Chem. Rev.110 6503

[18] Thompson T and Yates J Jr 2005 TiO2-based photocatalysis:surface defects, oxygen and charge transfer Top. Cata.35 197

[19] Nakamura I, Negishi N, Kutsuna S, Ihara T, Sugihara S andTakeuchi K 2000 Role of oxygen vacancy in the plasma-treated TiO2 photocatalyst with visible light activity for noremoval J. Mol. Cata. A: Chem. 161 205

[20] Benkstein K D, Kopidakis N, van de Lagemaat J and Frank A J2003 Influence of the percolation network geometry onelectron transport in dye-sensitized titanium dioxide solarcells J. Phys. Chem. B 107 7759

[21] Grimes C A 2007 Synthesis and application of highly orderedarrays of TiO2 nanotubes J. Mater. Chem. 17 1451

[22] Mohammadpour R, Iraji zad A, Hagfeldt A and Boschloo G2010 Comparison of trap-state distribution and carriertransport in nanotubular and nanoparticulate TiO2 electrodesfor dye-sensitized solar cells ChemPhysChem 11 2140

[23] Richter C and Schmuttenmaer C A 2010 Exciton-like trapstates limit electron mobility in TiO2 nanotubes Nat. Nano.5 769

[24] Jennings J R, Ghicov A, Peter L M, Schmuki P andWalker A B 2008 Dye-sensitized solar cells based onoriented TiO2 nanotube arrays: transport, trapping, andtransfer of electrons J. Am. Chem. Soc. 130 13364

[25] Cahen D, Hodes G, Grätzel M, Guillemoles J F and Riess I2000 Nature of photovoltaic action in dye-sensitized solarcells J. Phys. Chem. B 104 2053

[26] Nelson J 1999 Continuous-time random-walk model ofelectron transport in nanocrystalline electrodes Phys. Rev. B59 15374

[27] Lei Y, Zhang L D, Meng G W, Li G H, Zhang X Y, Liang C H,Chen W and Wang S X 2001 Preparation andphotoluminescence of highly ordered TiO2 nanowire arraysAppl. Phys. Lett. 78 1125

[28] Lei Y and Zhang L D 2001 Fabrication, characterization, andphotoluminescence properties of highly ordered TiO2

nanowire arrays J. Mater. Res. 16 1138[29] Lei Y, Yang S, Wu M and Wilde G 2011 Surface patterning

using templates: concept, properties and device applicationsChem. Soc. Rev. 40 1247

Figure 5. PL spectra of different samples annealing at 550 °C fordifferent times.

9

Nanotechnology 25 (2014) 275603 H Zhang et al

[30] Lei Y, Jiao Z, Wu M and Wilde G 2007 Ordered arrays ofnanostructures and applications in high-efficient nano-generators Adv. Eng. Mater. 9 343

[31] Wang C, Wen L, Kups T, Mi Y, Vellacheri R, Fang Y,Schaaf P, Zhao H and Lei Y 2013 Growth control ofAgTCNQ nanowire arrays by using a template-assistedelectro-deposition method J. Mater. Chem. C 1 8003

[32] Abazović N D, Čomor M I, Dramićanin M D, Jovanović D J,Ahrenkiel S P and Nedeljković J M 2006Photoluminescence of anatase and rutile TiO2 particlesJ. Phys. Chem. B 110 25366

[33] Shi J, Chen J, Feng Z, Chen T, Lian Y, Wang X and Li C 2006Photoluminescence characteristics of TiO2 and theirrelationship to the photoassisted reaction of water/methanolmixture J. Phys. Chem. C 111 693

[34] Serpone N 2006 Is the band gap of pristine TiO2 narrowed byanion- and cation-doping of titanium dioxide in second-generation photocatalysts? J. Phys. Chem. B 110 24287

[35] Serpone N, Lawless D and Khairutdinov R 1995 Size effectson the photophysical properties of colloidal anatase TiO2

particles: size quantization versus direct transitions in thisindirect semiconductor? J. Phys. Chem. 99 16646

[36] Kshirsager S T, Ogale S B, Sainkar S R, Patil S I and Saraf L V1998 Synthesis of nanophase TiO2 by ion beam sputteringand cold condensation technique Int. J. Modern Phys. B12 2635

[37] Forss L and Schubnell M 1993 Temperature dependence of theluminescence of TiO2 powder Appl. Phys. B 56 363

[38] Hendry E, Koeberg M, O’Regan B and Bonn M 2006 Localfield effects on electron transport in nanostructuredTiO2 revealed by terahertz spectroscopy Nano Lett. 6755

[39] Mercado C C, Knorr F J, McHale J L, Usmani S M,Ichimura A S and Saraf L V 2012 Location of hole andelectron traps on nanocrystalline anatase TiO2 J. Phys.Chem. C 116 10796

[40] Foley J M, Price M J, Feldblyum J I and Maldonado S2012 Analysis of the operation of thin nanowirephotoelectrodes for solar energy conversion EnergyEnviron. Sci. 5 5203

[41] Zhou M, Wu H B, Bao J, Liang L, Lou X W and Xie Y 2013Ordered macroporous BiVO4 architectures with controllabledual porosity for efficient solar water splitting Angew. Chem.Int. Ed. 52 8579

10

Nanotechnology 25 (2014) 275603 H Zhang et al