Epitaxial growth of TiO thin films on SrTiO , LaAlO and yttria ...STO,LaAlO 3 (LAO)andyttria-stabilized zirconia(YSZ) single crystal substrates [21]. They found that no epitaxial TiO

007) 3439–3447www.elsevier.com/locate/tsf

Thin Solid Films 515 (2

Epitaxial growth of TiO2 thin films on SrTiO3, LaAlO3 and yttria-stabilizedzirconia substrates by electron beam evaporation

A. Lotnyk ⁎, S. Senz, D. Hesse

Max Planck Institute of Microstructure Physics, Weinberg 2, 06120 Halle, Germany

Received 11 April 2006; received in revised form 8 September 2006; accepted 9 October 2006Available online 5 December 2006

Abstract

Epitaxial anatase, rutile and mixed (brookite and anatase) TiO2 thin films have been obtained by reactive electron beam evaporation. AnataseTiO2 thin films have been prepared on LaAlO3 (LAO) and SrTiO3 (STO) substrates of (100) and (110) orientation. The epitaxial relationshipbetween anatase films (A) and (110) LAO/STO substrates is (012) A ∣∣ (110) LAO/STO, [100] A ∣∣ [001] LAO/STO. The anatase films weretransformed into a cubic structure similar to TiO when in situ exposed to an electron beam. On the other hand, TiO2 thin films with a mixture ofepitaxial anatase and brookite grains have been grown on (100)-oriented yttria-stabilized zirconia (YSZ) at a substrate temperature of 600 °C whilethe pure rutile phase has been obtained at a substrate temperature of 900 °C on (100) YSZ. The crystalline structure and surface morphology of thethin films were investigated by X-ray diffraction, transmission electron microscopy and atomic force microscopy. The crystallographic reason forthe epitaxy between anatase films and perovskite substrates is discussed.© 2006 Elsevier B.V. All rights reserved.

Keywords: Thin film; Anatase; Rutile; Brookite

1. Introduction

Epitaxial TiO2 thin films are of interest due to their scientificand technological importances. The applications of TiO2

include photocatalytic, catalytic and optical devices as well asgas sensors and waveguides [1–3]. Hsieh et al. propose futureapplications of TiO2 thin films on (110) SrTiO3 (STO) in thefield of microwave devices, because (001) oriented YBa2Cu3O7

(YBCO) films can be produced on TiO2-templated STO, thusyielding (103) YBCO/(001) YBCO biepitaxial junctions on a(110) STO substrate [4].

There are at least five TiO2 structures [5]. The naturallyoccurring TiO2 polymorphs are rutile (tetragonal), anatase(tetragonal), brookite (orthorhombic) and TiO2 (B) (mono-clinic) that are composed of octahedrally coordinated Ti4+

cations. Rutile is the most stable form of TiO2 whereasanatase and brookite are metastable and transform to the rutilephase on heating [6,7]. There is no well-defined temperature

⁎ Corresponding author. Tel.: +49 345 5582 692; fax: +49 345 5511223.E-mail address: [email protected] (A. Lotnyk).

0040-6090/$ - see front matter © 2006 Elsevier B.V. All rights reserved.doi:10.1016/j.tsf.2006.10.106

for these transformations. However, the anatase-rutile transi-tion was observed at temperatures between 400 °C and1000 °C, depending on many factors such as presence ofdopants and size of corresponding precipitates, deviations ofstoichiometry, surface area, particle size and surroundingatmosphere [6–12].

Rutile single crystals are commercially available withdifferent surface orientations. Rutile surfaces have been studiedextensively using surface science techniques [13]. In addition,rutile single crystals were used as substrates in modelexperiments to study the formation of BaTiO3 thin films bysolid state reactions with BaCO3 and BaO [14,15]. Anatasesingle crystals are more difficult to obtain with appropriate size[16–18]. Thus, relatively few surface studies have beenconducted on anatase surfaces and no experiments concerningsolid state reactions with anatase surfaces were performed. Thisproblem has been addressed by growing epitaxial anatase thinfilms [19] or by using mineral samples [20].

Epitaxial TiO2 films can be prepared by several physical andchemical deposition techniques. Various substrates withdifferent lattice misfits were used. S. Yamamoto et al. haveused pulsed laser deposition (PLD) to grow TiO2 thin films on

3440 A. Lotnyk et al. / Thin Solid Films 515 (2007) 3439–3447

STO, LaAlO3 (LAO) and yttria-stabilized zirconia (YSZ) singlecrystal substrates [21]. They found that no epitaxial TiO2 thinfilms can be obtained on (110) STO and (110) LAO. (001)anatase films can be grown on (100) STO even at a substratetemperature of 1000 °C [4]. Single crystal epitaxial (001) TiO2

anatase films were also obtained on (100) LAO by PLD [22].Thus, only the (001) crystallographic surface of anatase can begrown by PLD and studied by surface techniques. No otherepitaxial orientations of anatase thin films were reported up tonow. Rutile thin films have been prepared on α-Al2O3

substrates [21,22] or on (100) STO substrates by oxidizingTiN films [4]. Epitaxial brookite films have obviously not beenobtained so far. Polycrystalline brookite was grown togetherwith anatase and rutile by PLD [23].

The deposition of TiO2 thin films on fused silica andtitanium-coated fused silica by reactive electron beam evapo-ration (REBE) has been reported previously [24]. The structureand properties of thin films under different conditions such asoxygen partial pressure and substrate temperature wereinvestigated. The structure of the obtained film depends onthe oxygen partial pressure and on the substrate used. In thepresent work, we report the epitaxial growth of anatase, rutileand mixed phase TiO2 thin films on STO, LAO and YSZsubstrates by REBE. The influence of substrate orientation andsubstrate temperature on formation and phase growth of TiO2

films is investigated.

2. Experimental details

The growth of epitaxial TiO2 thin films was carried out byREBE of titanium (IV) oxide (TiO2) powder tablets (Merck,Germany). The base pressure of the system was less than2×10−5 mbar. Pure oxygen was introduced during depositionto establish a pressure of (1–4)×10−4 mbar. The depositionrate (0.01 nm/s) and the thickness of the thin film (100 nm)were monitored in situ by a quartz microbalance. Thesubstrates were heated in a tube furnace at temperaturesbetween 400 °C and 1200 °C directly by thermal irradiationduring deposition. The temperatures were measured by a Pt/PtRh10 thermocouple. After deposition the samples were keptin the vacuum chamber and allowed to cool to roomtemperature.

One-side polished (100), (110) and (111) surfaces ofcommercial STO as well as (100) and (110) surfaces of LAOand (100) surfaces of YSZ single crystals were used as substratematerials. Before the experiments the (100) STO substrateswere chemically cleaned in buffered HF solution and thermallytreated in air at 950 °C. (110) and (111) STO substrates wereheated in air at 1100 °C for 60 min while (100) and (110) LAOand (100) YSZ were thermally treated in air at 900 °C and1200 °C for 10 min, respectively.

The phases present in the thin films after deposition and theirorientation relationships were investigated by X-ray diffraction(XRD, Philips X'Pert MRD) with CuKα radiation. θ–2θ and 2θmeasurements were performed to identify the oriented andpolycrystalline products. Rocking curve scans in θ were madeto determine the orientation quality of the thin films. Pole figure

measurements (center: ψ=0°, rim: ψ=90°) were performed tofind out the crystallographic orientation of the thin films relativeto the substrates. The surfaces of the deposited films werestudied by an atomic force microscope (AFM, DigitalInstruments 5000 microscope) working in tapping mode.Microstructure analysis of the interface between thin filmsand substrates was performed by cross sectional transmissionelectron microscopy (TEM). The samples for TEM were madeby standard methods. They were glued together face to facewith epoxy resin, then mechanically polished to a thickness ofabout 100 μm, dimpled from one side to get a thickness of about20 μm at the center followed by ion milling using a Gatanprecision ion polishing system. TEM investigations werecarried out in a Philips CM 20 T (at 200 keV) and a Jeol4010 (at 400 keV).

3. Results

In order to facilitate better understanding of the resultspresented in this paper, the crystal lattice parameters of TiO2

(anatase), TiO2 (rutile), TiO2 (brookite), SrTiO3, LaAlO3 andYSZ are listed:

TiO2 (anatase): tetragonal, space group I41/amd (No. 141),lattice parameters: a= b=0.378 nm andc=0.951 nm.

TiO2 (rutile): tetragonal, space group P42/mnm (No. 136),lattice parameters: a = b = 0.459 nm andc=0.296 nm.

TiO2 (brookite): orthorhombic, space group Pbca (No. 61),lattice parameters: a=0.92 nm, b=0.547 andc=0.517 nm.

SrTiO3: cubic, space group Pm3̄m (No. 221), lattice parameter:a=0.39 nm.

LaAlO3: rhombohedral at room temperature and usuallydefined as hexagonal, space group R3̄c (No. 167),lattice parameters: a=0.536 nm, c=1.311 nm. It canalso be represented as pseudocubic witha=0.3793 nm and derived from SrTiO3 by replacingthe Sr ions by La and the Ti ions by Al. We will usethe cubic indexing for LAO in this work.

Yttria-stabilized zirconia (YSZ): cubic, space group Fm3̄m

3.1. TiO2 film growth on (100) STO and (100) LAO

Pure anatase TiO2 (A) films were obtained on (100) STO and(100) LAO at substrate temperatures ranging from 500 °C to1000 °C. θ–2θ scans and pole figure measurements (notshown) reveal the epitaxial nature of the TiO2 films. Fig. 1shows a TEM micrograph of the film made on (100) STO at asubstrate temperature of 600 °C. Columnar grains are clearlyseen in the film. High-resolution TEM (HRTEM) images of theanatase/STO interface are shown in Fig. 2a and b. The interfacebetween substrate and film is sharp. The strain in the film isreleased partly by interfacial misfit dislocations (as the onemarked by white lines in Fig. 2b). The epitaxial orientation

Fig. 1. Cross-section transmission electron micrograph of an anatase thin filmgrown on (100) STO at a substrate temperature of 600 °C.

3441A. Lotnyk et al. / Thin Solid Films 515 (2007) 3439–3447

relationship obtained from the HRTEM image, from electrondiffraction and from pole figure analyses is:

ð001Þ A jj ð100Þ STO; ½100� A jj ½001� STO:

This orientation relationship is also observed on (100) LAO.The crystal quality of the prepared TiO2 thin films was

analyzed by rocking curves. The corresponding full-width at halfmaximum (FWHM) values of the (004) anatase peaks for the

Fig. 2. High-resolution TEM images of the sample made at 600 °C: a) Overviewof the (001) anatase/(100) STO interface; and b) Interfacial defect (misfitdislocation marked by white lines) between substrate and the anatase film.

films prepared on (100) STO at substrate temperatures up to900 °C are ranging from0.54° to 0.69°. The FWHMof theKβ lineof the (100) STO substrate is 0.13°. At a substrate temperature of1000 °C, a relatively broad peak, split into three peaks, wasobserved in the θ scan which means that the anatase film consistsof three kinds of tilted domains. For the thin films deposited on(100) LAO, the FWHM values of the rocking curves are about0.2° showing the better crystallinity of the anatase films grown on(100) LAO compared to those on (100) STO. It should be notedthat LAO substrates usually consist of several twin domains. Therocking curve of the (200) LAO peak measured by high-resolution XRD with a Ge (220) monochromator shows a broadpeak and the FWHM was estimated to be about 0.1°.

AFM images of thin films deposited on (100) STO and (100)LAO at a substrate temperature of 800 °C are shown in Fig. 3aand b, respectively. The surface morphology shows an averageisland diameter of about 180 nm in Fig. 3a, and of about 450 nmin Fig. 3b. The islands in Fig. 3a and b are square-shaped. Thesurface of each island consists of spiral terraces due to a spiralgrowth mechanism. With increasing substrate temperature the

Fig. 3. AFM images of (001) anatase thin films grown on a) (100) STO and b)(100) LAO at a substrate temperature of 800 °C.

Fig. 4. X-ray diffraction patterns of epitaxial anatase thin films grown on a)(110) STO and b) (110) LAO substrates. The films were deposited at 700 °C.

3442 A. Lotnyk et al. / Thin Solid Films 515 (2007) 3439–3447

average island diameter of the films grown on (100) STO isincreasing up to 600 nm at a temperature of 1000 °C.

At a substrate temperature of 1100 °C the films grown on(100) STO consist of a mix of (110) rutile and (001) anatasegrains. A splitting of the {101} anatase peaks in pole figures

Fig. 5. X-ray pole figure (2θ=25.3°) of the TiO2 film deposited on (110) LAO ata substrate temperature of 800 °C. The peaks situated at ψ=16°, 61° and 75° arefrom the anatase {101} family. The positions of marks “a” and “b” correspond tothe φ values of the ð1̄10Þ and ð11̄0Þ substrate planes, respectively.

into two peaks occurs (not shown), which means that theanatase grains consist of two kinds of tilted domains.

3.2. TiO2 film growth on (110) STO and (110) LAO

In the case of (110) STO and (110) LAO, pure epitaxialanatase films were obtained at substrate temperatures between500 °C and 750 °C, and 500 °C and 900 °C, respectively. Fig. 4a,b show the θ–2θ scans of epitaxial anatase films grown at700 °C on (110) STO and (110) LAO, respectively. The peak at62.6° in Fig. 4a and 4b corresponds to (024) anatase. Fig. 5shows a pole figure recorded at fixed 2θ=25.3° of the sampleafter deposition at a substrate temperature of 800 °C on (110)LAO. The epitaxial orientation relationship between the thinfilm and (110) LAO is:

ð012Þ A jj ð110Þ LAO; ½100� A jj ½001� LAO:

This orientation relationship was also observed on (110)STO. In distinction from thin films grown by PLD [21],epitaxial (012) anatase films can be prepared by REBE. Fig. 6shows an HRTEM image of the film deposited on (110) STO ata substrate temperature of 700 °C. The above orientationrelationship is in good agreement with the one deduced fromHRTEM and from electron diffraction (not shown). Theinterface in Fig. 6 is not ideal. Small grains were found at theinterface. These grains have an orientation deviating from thatof the film.

The anatase film was converted into a cubic phase similar toTiO when extensively exposed to the 400 keVelectron beam fora few minutes. Fig. 7a shows an HRTEM micrograph of a placeafter exposure of the anatase film under the electron beam for a

Fig. 6. High-resolution TEM images of the TiO2 film grown on (110) STO at asubstrate temperature of 700 °C: a) Overview of (012) anatase/(110) STOinterface; and b) magnified image of the area marked in (a).

Fig. 7. TEM cross-section micrographs of anatase film and substrate afterextended exposure to electron beam: a) Interface between the damaged film and(110) STO; b) and c) Lattice plane images of TiO and the STO substrate,respectively.

Fig. 8. AFM image of a (012)-oriented anatase thin film grown on (110) LAO ata substrate temperature of 800 °C.

Fig. 9. Cross-section transmission electron micrograph of a TiO2 thin film grownon (100) YSZ at a substrate temperature of 600 °C.

3443A. Lotnyk et al. / Thin Solid Films 515 (2007) 3439–3447

few minutes. The film shows strong black and white contrastindicating the instability of the film under the electron beam.Damaged parts (white areas) are clearly seen in Fig. 7a. At thesame time, the contrast of the substrate is uniform showing thatno distortions occur. Fig. 7b shows an enlarged image of thedamaged part of Fig. 7a. The angle between the lattice planes ofthe converted phase is 90°, with a spacing of 0.42 nm. Thisphase can thus be attributed to a cubic phase similar to titaniummonoxide (TiO).

In rocking curves of the TiO2 films grown on (110) STO theFWHM values of the (024) anatase peak are from 0.56° to0.72°. The crystal quality of the films obtained on the (110)LAO is better than on (110) STO. The FWHM values of the(024) anatase peak are 0.4° for the films prepared at substratetemperatures up to 700 °C and 0.2° for the films prepared atsubstrate temperatures higher than 800 °C, showing the goodcrystallinity of the thin films. An AFM image of the thin film

deposited on (110) LAO at a substrate temperature of 800 °C isshown in Fig. 8. The islands are elongated in the ½11̄0� directionof LAO (the orientation of the substrate was confirmed byXRD), with an average grain size of about 150 nm×100 nm.The same surface morphology of the anatase islands wasobserved on (110) STO. As shown by AFM, most of the surfaceof large (012)-oriented anatase grains is plane and parallel to thesubstrate surface, so that it should represent a crystallographic(012) plane.

A small amount of the rutile phase was detected afterdeposition on (110) STO at temperatures of 800 °C and 900 °C,as shown by one very weak peak in the pole figures recorded at2θ=27.4°. The amount of rutile grains is about the same at atemperature of 1000 °C but its orientation quality is improved.The maximum intensity of the rutile phase at this temperaturetaken from a pole figure with 2θ=27.4° is 38 cps. This is a verylow value compared to the intensity of anatase taken from a polefigure with 2θ=25.3°, which is 9216 cps. The orientationquality and the amount of the anatase phase are decreased at asubstrate temperature of 1100 °C. However, the total intensityobtained from anatase is 3516 cps while from rutile it is 85 cps.At a substrate temperature of 1200 °C the anatase phase is not

Fig. 10. High-resolution TEM images of the TiO2 film grown on (100) YSZ at asubstrate temperature of 600 °C: a) Thin film/substrate interface; b) Lattice planeimages of brookite and anatase grains.

3444 A. Lotnyk et al. / Thin Solid Films 515 (2007) 3439–3447

present any more in the films. Now the film consists mainly oftilted rutile grains.

3.3. TiO2 film growth on (111) STO

For the (111) STO substrates, no pure anatase film wasobtained at a substrate temperature of 600 °C. This filmconsists of a mixture of (001) and (112) anatase and (100)rutile epitaxial grains. The epitaxial orientation relationships ofthe anatase and rutile (R) grains were found by pole figures(not shown) as follows:

ð001Þ A jj ð111Þ STO; ½100� A jj ½3P41� STO;

ð112Þ A jj ð111Þ STO; ½1P10� A jj ½1P1 0� STO;

ð100Þ R jj ð111Þ STO; ½010� R jj ½1P21� STO:

3.4. TiO2 film growth on (100) YSZ

The phase compositions of the films grown on (100) YSZsubstrates are different from the films obtained on the cubicperovskites (100) LAO and (100) STO. A pure anatase phasewas not obtained on (100) YSZ. This is in contrast with the dataobtained in Ref. [21]. XRD investigations reveal that epitaxialthin films of mixed anatase and brookite (B) structures wereformed after deposition at substrate temperatures up to 600 °C.The epitaxial orientation relationships of the brookite andanatase grains with respect to the (100) YSZ substrate wereidentified by pole figures (not shown) as follows:

ð100Þ B jj ð100Þ YSZ; ½001� B jj ½001� YSZ;

ð001Þ A jj ð100Þ YSZ; ½100� A jj ½011� YSZ:

The orientation relationships show that the two phases growin two different directions relative to the substrate. For TEMinvestigations part of the cross-section samples were preparedparallel to the substrate edge (0°, YSZ (001) cut) and the otherpart with a rotation by 45° (45°, YSZ (011) cut). Fig. 9 shows across section TEM image of a film grown at a substratetemperature of 600 °C and prepared in the 0° view. The filmconsists of grains with plane boundaries. Fig. 10 shows anHRTEM image of the interface between the substrate and thethin film. HRTEM investigations show that the anatase grainsare separated by brookite grains. Apparently there is no well-defined, regular shape of the brookite and anatase grains. Theorientation relationships of the brookite and anatase grainsdeduced from HRTEM are in a good agreement with the onesobtained by pole figure measurements. The brookite phase wasstable under the electron beam while the anatase grains wereagain transformed into a cubic phase under the electron beam.

Rutile grains were found in the films starting from a substratetemperature of 700 °C. With increasing substrate temperature

the amount of the rutile phase increased. Some remaininganatase grains were detected at a temperature of 850 °C while apure rutile thin film was obtained at a substrate temperature of900 °C. Fig. 11 shows a pole figure (2θ=27.4°, rutile {110}family) of a thin film grown at this temperature. The epitaxialorientation relationship between (100) YSZ and the rutile thinfilm is:

ð100Þ R jj ð100Þ YSZ; ½010� R jj ½023�YSZ:

4. Discussion

Aswasmentioned above, rutile is the most stable form of TiO2

whereas anatase and brookite are metastable and transform to therutile phase on heating. Experimental data [10] and theoreticalanalysis [10,11] show that anatase becomes more stable withrespect to rutile at very small crystallite sizes (≈10–14 nm).These findings are not directly applicable to our results. The pureanatase films deposited on STO and on LAO have an averagegrain size from 100 nm to 600 nm depending on the substratetemperature used. In addition, the stability of anatase depends

Fig. 11. X-ray pole figure (2θ=25.3°) of a rutile film deposited on (100) YSZ ata substrate temperature of 900 °C. Peaks are situated at ψ=45°, which are fromthe TiO2 {110} family. The positions of marks “a” and “b” are corresponding tothe φ values of the (023) and ð02̄ 3̄Þ substrate planes, respectively.

Fig. 13. Schematic representation of the TiO6 octahedra of a) the TiO2 anatasecell and b) the TiO cell, respectively.

3445A. Lotnyk et al. / Thin Solid Films 515 (2007) 3439–3447

strongly on the substrate temperature, the substrate orientation,and the kind of substrate used. Previous discussions on thestability of the anatase phase and on the epitaxial growth werefocused on the similarities between the local atomic arrangementsin the grown layer and the substratematerials [4,22]. Here, wewill

Fig. 12. Schematic representation of the TiO6 octahedra of a) the (001) A/(100)STO interface and b) the (012) A/(110) STO interface, respectively.

apply a fundamental building block approach in order to betterunderstand the epitaxial relations between anatase and thesubstrates — an approach that was formerly adopted in Ref.[25] to understand epitaxial relations among the TiO2 minerals.TiO2 and perovskite structures can be constructed from AO6

octahedra (in our caseA=Ti, Al). Fig. 12a and b show a schematicrepresentation of building blocks (TiO6 octahedra) for (001)A/(100)STO and (012)A/(110)STO interfaces, respectively. Thesimilarities between the orientations of the TiO6 octahedra of STOand anatase are clearly visible. The (100) surface of STOpresentedin Fig. 12a is SrO-terminated. According to the work of Ohnishi etal. [26], the experimental treatment of the (100) STO substratesperformed in our work (heating in air and then in vacuum of10−5 mbar), however, results in mixed SrO- and TiO2-terminatedsurfaces of the STO. The substrate termination is important for theinterface because the TiO2-terminated (100) SrTiO3 surfacecontains truncated TiO6 octahedra. On the other hand, the (100)surface of LaAlO3 is terminated by a La–O layer at hightemperatures (above 523 K) [27,28] and thus contains completeAlO6 octahedra. TheAO6 octahedra (A=Ti, Al) are assumed to beimportant in determining the final phase formation and theepitaxial growth of the thin films. Moreover, the STO surfacesshow a good lattice match with the anatase surface. The in-planelatticemismatch between (100)A and (100) STO is−3.04%whilethe lattice mismatch between ð013̄Þ A (d=0.243 nm) and ð11̄0ÞSTO (d=0.275 nm) is −11.6%. It should be noted that the anglebetween the ð013̄Þ and (012) planes of anatase is not exactly 90°.A calculation results in the value of 88.5°. This means that theð013̄Þ plane of a perfectly (012)-oriented anatase grain would

3446 A. Lotnyk et al. / Thin Solid Films 515 (2007) 3439–3447

deviate from the ð11̄0Þ STO plane by 1.5°, resulting in a kink by1.5° at the anatase/STO interface. In the same scenario theepitaxial formation of anatase films on (100) LAO and (110) LAOsurfaces can be explained. The only difference is the better latticematch of (001)/(012) anatase with (100)/(110) LAO surfaces. Thelattice mismatch is −0.01% between (100) A and (001) LAO and−9.4% between ð013̄Þ A and ð11̄0Þ LAO (d=0.268 nm).

It was found that the irradiation of anatase cross-sectionTEM specimens by the electron beam results in a transformationinto a cubic phase. A similar transformation was observed inRef. [29] by secondary-electron imaging and low energyelectron diffraction. The authors found that sputtering with500 eV Ne+ ions of (001) anatase surfaces leads to atransformation from tetragonal anatase to the face-centeredcubic titanium monoxide TiO. Our TEM investigations alsoshow this phenomenon. The new transformed structure isepitaxial with anatase. A comparison of anatase and TiO isshown in Fig. 13a and b. The building blocks of the anatasestructure are distorted TiO6 octahedra in a zig-zag alignment(Fig. 13a) whereas the TiO6 octahedra in TiO are highlysymmetric and without distortions (Fig. 13b). The cubic NaCltype lattice of TiO is known to be stable with very strongdeviations from an ideal 1:1 stoichiometric ratio [30].Comparing one anatase unit cell with two TiO unit cells, theauthors of Ref. [29] found that filling intersitial sites in theanatase structure with titanium ions would lead to the TiOstructure (see Fig. 6 in Ref. [29]). Thus, anatase can beconsidered as an ordered phase similar to Ti0.5O1.0 with missingoctahedra. The electron beam induced disorder in anataseresults in a cubic oxide with an intermediate composition in theneighbourhood of Ti0.5O1.0. A transition from TiO2 anatase toideal TiO would require a strong shrinkage since the densitiesare very different (ρanatase=3.9 g/cm3, ρTiO=5.84 g/cm3).

The phase formation on (100) YSZ substrates dependsstrongly on the substrate temperature used. The YSZ structuredoes not contain AO6 octahedra. Thus, the nucleation and growthof TiO2 phases is defined by the lattice match between thesubstrate and the thin film. Considering the orientation relation-ships [001] B ∣∣ [001] YSZ and [100] A ∣∣ [011] YSZ thedifferences between (001) B and (001) YSZ and (100) A and(011) YSZ (d=0.363 nm) are 0.6% and 4.1%, respectively. Thelattice mismatch between (010) R and (023) YSZ (3d=0.429 nm)is 6.7% and between (001) R and ð03̄2Þ YSZ (2d=0.286 nm) is3.4% considering the orientation relationships [010] R ∣∣ [023]YSZ and [001] R ∣∣ ½03̄2� YSZ, respectively. Obviously, such arelatively small misfit may play an important role in the nu-cleation of both the brookite and anatase phases at low substratetemperatures (up to 600 °C) while the formation of a pure rutilephase is predominant at high temperature (900 °C).

5. Conclusions

Epitaxial anatase, rutile and mixed TiO2 films were preparedby reactive electron beam evaporation of TiO2 targets. Epitaxial(001)-oriented anatase films were grown on (100) STO and(100) LAO substrates, while (012)-oriented anatase films wereobtained on (110) STO and (110) LAO. Epitaxial mixed anatase

and brookite thin films were prepared on (100) YSZ surfaces atsubstrate temperatures up to 600 °C while phase-pure rutilefilms were grown on the same substrate at a temperature of900 °C. The similarities in the orientations of the AO6 octahedraof STO/LAO and anatase structures determine the final phaseformation and the epitaxial growth of the thin films. On theother hand, the structure of (100) YSZ cannot be represented byAO6 octahedra. The choice in the nucleation and growth of TiO2

phases on (100) YSZ is defined by the lattice match betweensubstrate and the thin film. The anatase films were transformedinto a cubic phase similar to TiO when in situ exposed to anelectron beam.

We have demonstrated the possibility to produce epitaxialanatase thin films of good crystalline quality with differentsurface orientations by REBE. The mean orientation of thesurface of (012)-oriented anatase grains is close to (012). Thequestion, whether the (012) surface is stable or whether micro-faceting occurs, remains for further work. This work opensalso the way to investigate the solid state reactions of anatasesurfaces with BaCO3 and BaO. The latter is presently inprogress.

Acknowledgments

Work supported by Deutsche Forschungsgemeinschaft(DFG) via SFB 418 at Martin-Luther-Universität Halle-Wittenberg. The authors are thankful to Ms. Sina Swatek forTEM sample preparations.

References

[1] B. Weinberger, R. Garber, Appl. Phys. Lett. 66 (1995) 2409.[2] J.P. Chen, R.T. Yang, J. Catal. 139 (1993) 277.[3] G.E. Jellison, F.A. Modine, L.A. Boatner, Opt. Lett. 22 (1997) 1808.[4] C.C. Hsieh, K.H.Wu, J.Y. Jung, T.M.Uen, J.-Y. Lin, Y.S. Gou, J. Appl. Phys.

92 (2002) 2518.[5] J.F. Banfield, B.L. Bischoff, M.A. Anderson, Chem. Geol. 110 (1993) 211.[6] A.W. Czanderna, C.N.R. Rao, J.M. Honig, Trans. Faraday Soc. 54 (1958)

1069.[7] C.N.R. Rao, S.R. Yoganarasimhan, P.A. Faeth, Trans. Faraday Soc. 57

(1961) 504.[8] R. Rodriguez-Talavera, S. Vargas, R. Arroyo-Muirillo, R. Montiel-Campos,

E. Haro-Poniatoiwski, J. Mater. Res. 12 (1997) 439.[9] F.C. Gennari, D.M. Pasquevich, J. Mater. Sci. 33 (1998) 1571.[10] A.A. Gribb, J.F. Banfield, Am. Mineral. 82 (1997) 717.[11] H. Zhang, J. Banfield, J. Mater. Chem. 8 (1998) 2073.[12] R.D. Shannon, J. Appl. Phys. 35 (1964) 3414.[13] U. Diebold, Surf. Sci. Rep. 48 (2003) 53.[14] A. Graff, S. Senz, D. Völtzke, H.-P. Abicht, D. Hesse, J. Eur. Ceram. Soc.

25 (2005) 2201.[15] A. Lotnyk, S. Senz, D. Hesse, Solid State Ionics 177 (2006) 429.[16] E. Wäsch, Krist. Tech. 7 (1972) 187.[17] F. Izumi, H. Kodama, A. Ono, J. Cryst. Growth 47 (1979) 139.[18] H. Berger, H. Tang, F. Levy, J. Cryst. Growth 130 (1993) 108.[19] G.S. Herman, Y. Gao, T.T. Tran, J. Osterwalder, Surf. Sci. 447 (2000) 201.[20] N. Ruzycki, G.S. Herman, L.A. Boatner, U. Diebold, Surf. Sci. Lett. 529

(2003) L239.[21] S. Yamamoto, T. Sumita, T. Yamaki, A. Miyashita, H. Naramoto, J. Cryst.

Growth 237–239 (2002) 569.[22] J.Y. Huang, B.H. Park, D. Jan, X.Q. Pan, Y.T. Zhu, Q.X. Jia, Philos. Mag.,

A 82 (2002) 735.[23] M.P. Moret, R. Zallen, D.P. Vijay, S.B. Desu, Thin Solid Films 366 (2000) 8.

3447A. Lotnyk et al. / Thin Solid Films 515 (2007) 3439–3447

[24] R. Krol, A. Goossens, J. Vac. Sci. Technol., A 21 (2003) 76.[25] J.F. Banfield, D.R. Veblen, Am. Mineral. 77 (1992) 545.[26] T. Ohnishi, K. Shibuya, M. Lippmaa, D. Kobayashi, H. Kumigashira, M.

Oshima, H. Koinuma, Appl. Phys. Lett. 85 (2004) 272.[27] J. Yao, P.B. Merril, S.S. Perry, D. Marton, J.W. Rabalais, J. Chem. Phys.

108 (1998) 1645.

[28] R.J. Francis, S.C. Moss, A.J. Jacobson, Phys. Rev., B. 64 (2001) 235425.[29] R. Hengerer, B. Bolliger, M. Erbudak, M. Grätzel, Surf. Sci. 460 (2000) 162.[30] R.W.G. Wyckoff, Crystal Structures, vol. 1, Interscience Publishers, New

York, 1963, p. 94.