Applied Surface Science 632 (2023) 157586 Contents lists available at ScienceDirect Applied Surface Science journal homepage: www.elsevier.com/locate/apsusc Full Length Article Heteroepitaxial growth of anatase (0 0 1) films on SrTiO3 (0 0 1) by PLD and MBE A. Crespo a, J. Gallenberger b, c, M. De Santis b, V. Langlais d, F. Carla e, J.M. Caicedo f, J. Rius a, X. Torrelles a, * a Institut de Ciència de Materials de Barcelona (ICMAB-CSIC), Campus de la UAB, 08193 Catalonia, Spain b Université Grenoble Alpes, CNRS, Grenoble INP, Institut Néel, 38042 Grenoble, France c Technische Universität Darmstadt, 64287 Darmstadt, Germany d Centre d’Elaboration de Matériaux et Etudes Structurales (CEMES-CNRS), 31055 Toulouse, France e Diamond Light Source, Harwell Science and Innovation Campus, Didcot, Oxfordshire OX11 0DE (UK) f Catalan Institute of Nanoscience and Nanotechnology (ICN2), Campus UAB, Building ICN2, Bellaterra, Catalonia E-0193, Spain A B S T R A C T The epitaxial growth of anatase (0 0 1) films deposited by pulsed laser deposition (PLD) and molecular beam epitaxy (MBE) on SrTiO3 (0 0 1) (STO) single crystals has been studied using X-ray diffraction and surface sensitivity UHV techniques. The evolution of the strain represented by the microstrain and the change of the in-plane and out-of-plane lattice parameters with film growth temperature, the effect of the annealing temperature and the influence of the oxygen content of the film have been investigated. The out-of-plane lattice strain shows a compressive (− 0.2 %) or expansive (+0.3 %) behavior, in the range 600–900 ◦ C, for temperatures below or above 700 ◦ C, respectively. The in-plane lattice parameters, as well as the cell volume of the film, remain under compression over the entire temperature range explored. PLD films grow into square islands that align with the surface lattice directions of the STO substrate. The maximum size of these islands is reached at growth temperatures close to 875–925 ◦ C. Film annealing at temperatures of 800 ◦ C or higher melts the islands into flat terraces. Larger terraces are reached at high annealing temperatures of 925 ◦ C for extended periods of 12 h. This procedure allows flat surface terrace sizes of up to 650 nm to be achieved. The crystalline quality achieved in anatase films prepared by PLD or MBE growth methods is similar. The two-step anatase growth process used during the synthesis of the films with both methods: film growth and post-annealing treatment in oxygen or air at ambient pressure, using temperature and time as key pa­ rameters, allows to control the surface terrace size and stoichiometry of the films, as well as the anatase/rutile intermixing rates at sufficiently high temperatures. This growth process could allow the substitution of their equivalent single crystals. The range of applicability of these films would include their use as structural and electronic model systems, or in harsh experimental conditions due to their low production cost. 1. Introduction UV portion of the spectrum. However, most energy in the solar spectrum is carried by lower energy photons. Consequently, attempts at modifying Titanium dioxide (TiO2) is a wide band-gap semiconductor, nowa­ the absorption edge of titania to increase its efficiency against visible days still considered a very promising material because of its physical light have been performed either by a partial substitution of oxygen with and chemical properties and its great variety of applications in different another electronegative atom or by the absorption of molecular species scientific and technological areas of relevance such as photovoltaic that can take in visible light and donate electrons or holes to the titania conversion of light to electricity [1], photocatalysis [2] or solar cells as [9–13]. electrodes [3]. Among the different polymorphs of TiO2, rutile is the The anatase phase has attracted particular attention due to its longer most studied one in terms of both experimental and theoretical funda­ carrier lifetime, [14] longer diffusion length, [15] and higher electron mental properties, although it is anatase that shows both the best cata­ mobility [16] with respect to the rutile phase. Different facets of anatase- lytic properties [4] and a number of interesting behaviors [1,5–8]. TiO2 exhibit different activities for different reactants. To understand One aspect that limits the photocatalytic applications of titania is its the origin of such morphology-dependent activity, numerous experi­ band gap. Since the absorption edges of rutile and anatase are at ≈3.0 eV mental studies have been carried out. Since the pioneering work of [17] and ≈3.2 eV, respectively, light absorption will only be significant in the on the preparation of a well-defined anatase TiO2 single crystal * Corresponding author. E-mail address: torrelles@icmab.es (X. Torrelles). https://doi.org/10.1016/j.apsusc.2023.157586 Received 13 March 2023; Received in revised form 18 May 2023; Accepted 20 May 2023 Available online 22 May 2023 0169-4332/© 2023 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC license (http://creativecommons.org/licenses/by- nc/4.0/). A. Crespo et al. Applied Surface Science 632 (2023) 157586 containing 47 % (0 0 1) facet, researchers devoted intensive efforts to conditions, temperature (T) and oxygen pressure P(O2) of anatase (0 0 1) enhance its efficiency by varying the (0 0 1)-to-(1 0 1) facet ratio [18]. films on SrTiO3 (0 0 1) substrates were determined. Derived from this Theoretical studies studying the mechanisms of H2 dissociation in the work, a correlation is established between both synthesis procedures. three anatase faces (0 0 1) (1 0 0) and (1 0 1) indicate that this process is Furthermore, the evolution of different structural and morphological more favorable in anatase (0 0 1) [19]. Most of the works provided in­ properties related to the films have been investigated with film growth sights into the higher photoactivity of the dominant (0 0 1) facet for H2 temperature, such us: 1) grain size and strain (represented by micro­ production. In contrast, other studies invoked the low-energy (1 0 1) strain and the change of in- and out-of-plane lattice parameters), 2) the facet for higher photocatalytic activity [20,21]. In the absence of effect of annealing temperature, and 3) the influence of the oxygen comprehensive fundamental studies pointing out the role of each crystal content in the film. facet in the photocatalytic redox reactions, the debate still remains un­ clear and needs to be well clarified. 2. Experiment details A second limiting aspect is the scarce knowledge and restricted availability of the anatase-TiO2 (0 0 1) face. Oriented rutile substrates, The crystalline structure and crystallographic orientation of the among which the (1 1 0) face is the most stable one, can be synthesized in titania thin films were characterized by X-ray diffraction (XRD) (Mal­ the laboratory and have been exhaustively used as model surfaces in the vern PANalytical X’pert Pro MRD) using Cu Kα radiation (λ = 1.540 Å). investigation of fundamental surface phenomena [22]. In contrast, in­ The surface morphology was studied by atomic force microscopy (AFM) formation about (0 0 1)-oriented anatase is sparse because: 1) it is (Keysight 5100). difficult to synthesize in the laboratory [23]; 2) it is the most scarce in nature and shows the highest surface energy [24]. Therefore, access to 2.1. SrTiO3 (0 0 1) substrate preparation anatase (0 0 1) surfaces with large crystal terrace sizes at low cost is of paramount importance to advance in the fundamental knowledge of the SrTiO3 (0 0 1) substrates from Crystec GmbH (miscut < 0.3◦ , di­ structural and electronic properties of this surface and for its applica­ mensions: 5 × 5 × 0.5 mm) were used for TiO2 film growth. The pro­ tions. In the last two decades, the improvements on sample preparation cedure to obtain atomically flat Ti-STO terminated surfaces involves and characterization techniques permitted some experimental studies annealing at 1000 ◦ C for 1 h in air and further sonication in deionized using well defined anatase surfaces [25–28]. However, the synthesis water etching for 15 min [33]. After this treatment, AFM measurements techniques still need some improvements in order to grow suitable show Ti-terminated flat terraces with step heights between them of one anatase crystals in the laboratory. unit cell, i.e. 3.9 Å. These substrates were placed in their respective PLD/ The possibility of growing TiO2 over substrates oriented within a MBE chambers and before starting the growth process they were cleaned specific plane such as SrTiO3 (0 0 1) allows the formation of highly or­ to remove residual carbon from their surface due to exposure to air by dered films of anatase (0 0 1). In. addition, the use of extended synthesis annealing them at 850 ◦ C at 10− 6 mbar oxygen pressure during 20 min. techniques such as PLD (Pulsed Laser Deposition) and MBE (Molecular Beam Epitaxy) could help improve access to this type of material for use 2.2. PLD growth in studying its properties, i.e., structural, electronic and photocatalytic. SrTiO3 (0 0 1) (STO) offers one of the best initial conditions for the A commercially available TiO2 target (diameter: 5 cm, purity: 99.99 heteroepitaxial growth of anatase (0 0 1). The mismatch between its in- %, Mateck) was used to prepare the titania thin films by PLD. TiO2 films plane parameters is of approx. 3 %. On the contrary, substrates like of about 50 nm thick were grown using a 248 nm KrF excimer laser with DyScO3 (1 1 0), with low surface roughness and with a 4 % mismatch a repetition rate of 2 Hz and an energy density of ~1.8 J/cm2. The between them, produce a mixture of granulated anatase crystals of number of laser pulses necessary to grow a 50 nm anatase film is of considerable size with crystals of other orientations and/or composi­ 1500. The growing process takes 12.5 min at a speed rate of 0.033 nm/ tions. The percentage of these anatase (0 0 1) facets decreases with pulse. annealing T while the others increase as indicated by X-rays and AFM Substrates were introduced in a non UHV-PLD chamber until static measurements [29]. The LaAlO3 (0 0 1) (LAO) system has in-plane lattice vacuum pressures close to 1.33⋅10− 8 mbar were achieved. Then STO parameters close to those of anatase (0 0 1). The lower mismatch with substrates were annealed at a P(O2) of 0.11 mbar up to the TiO2 depo­ LAO reduces the crystal quality when the film thickness increases, while sition temperature to guarantee carbon removal. The films were grown in STO it improves due to the relief of the lattice strain [30]. in a range of temperatures between 600 and 925 ◦ C at identical P(O2) = The possibility of employing other techniques such as sol–gel or 0.11 mbar. After deposition, films were cooled down at 7 ◦ C/min at P atomic layer deposition over ordered substrates would be possible. (O2) = 13.33 mbar. The evolution of the film growth was monitored by There are two well-known sol–gel procedures to prepare anatase films in-situ reflectivity high energy electron diffraction (RHEED). Ex-situ by dip- or spin-coating. One involves the preparation of an amorphous XRD (θ-2θ) and AFM measurements were performed to study the film that is converted to anatase after a thermal process [31]. The second structure and morphology of the films. is based on the early preparation of a sol of anatase nanoparticles, avoiding a thermal process for the crystallization from amorphous 2.3. MBE growth titania [32]. In both cases, surfactants can be utilized for the control of the porosity of the resulting film. However, these methods still need TiO2 films in the nanometric range were grown at the Néel Institute more efforts to develop a process for the growth of highly oriented by Molecular Beam Epitaxy (MBE) using a UHV set-up composed by a anatase surfaces. preparation chamber coupled to an analysis chamber. The latter is The experimental work has involved the optimization of the syn­ equipped with a commercial scanning tunneling microscope (Omicron thesis conditions of anatase (0 0 1) films grown on SrTiO3 (0 0 1) sub­ VT STM/AFM), a low-energy electron diffractometer (LEED) and an strates with techniques that allow their growth with a preferential Auger electron spectrometer (AES). The substrate temperature was orientation such as PLD and MBE. During this process, some similarities monitored using an infrared pyrometer (IMPAC Infrared GmbH IPE140). were discovered between both methods that, despite being different, Cleanliness was checked by AES. Metallic titanium was evaporated share those steps related to the preparation of the surface substrate and (Titanium pellets 3 × 3 mm 99.995 %; Mateck) by an electron beam subsequent annealing treatment of the grown films. evaporator (EFM 4 Omicron), with the chamber backfilled with mo­ Consequently, the main objective presented in this work is to reach lecular oxygen at a partial pressure of 10− 6 mbar at 600 ◦ C. The titanium flat anatase (0 0 1) surfaces with large terrace sizes (>600 nm) synthe­ flux was previously calibrated by using a quartz crystal microbalance. sized by PLD and MBE. To achieve such a goal, the best growth The deposition rate was about 0.012 Å/s. The deposition of 8 nm of 2 A. Crespo et al. Applied Surface Science 632 (2023) 157586 titanium is equivalent to 16 nm of TiO2 anatase. The same relation is initial stage of growth, Ti reaches the terrace steps and condenses at the maintained for the other thicknesses. edges during a step-flow growth mode. The intermediate stage is The STO substrates were treated as described for the PLD samples, compatible with a Volmer-Weber growth mode in which the interaction but were additionally cleaned by two successive cycles of Ar+ ion between the incoming surface atoms is small, giving rise to the forma­ sputtering (0.6 keV, P(Ar) = 5 * 10− 6 mbar, t = 5 min and t = 10 min) tion of islands. During the initial stage of formation of islands, TiO2 is and annealing at 800 ◦ C under a P(O2) of 10− 6 mbar for 20 and 40 min, proposed to grow at the interface between the islands and the upper STO respectively. This step could be replaced by annealing at 850 ◦ C at an layers by forming misfit dislocations to reduce the mismatch between oxygen pressure of 10− 6 mbar for 20 min. The cleanliness was checked the in-plane cell parameters of anatase with those of the STO, as by AES. TiO2 films of 2, 4, 8 and 16 nm thickness were prepared by observed in anatase (1 0 0) films grown on STO(0 0 1) [35]. The right depositing Ti at 600 ◦ C as mentioned above and characterized in situ by image in Fig. 1 shows the process in a projected 2D X-Z plane involving Auger and low energy electron diffraction (LEED) techniques. Subse­ 10 STO cells (X direction) and 5 anatase cells along the Z direction, quently, all samples were annealed ex situ in a cylindrical ceramic however, the distortions are noticeably reduced after 2 or 3 cells are furnace at 800 ◦ C in 1 bar oxygen for 6 h and reintroduced into the UHV grown. In this initial slab of 10 STO cells, the reduction of the mismatch chamber. Their surfaces were refreshed by performing two successive between film and STO is 1/3 of the total 3 % mismatch between them. cycles of Ar+ ion sputtering (0.6 keV, P(Ar) = 5 * 10− 6 mbar, t = 5 min) The 3rd stage involves growth through the flow of crystal steps where and annealing at 700 ◦ C under a P(O2) of 10− 6 mbar for 30 min. The the gaps between adjacent islands are filled [34]. resulting films were re-analyzed with the same UHV techniques, but only the one with the best LEED diffractogram was inspected by STM. 3.2. XRD characterization of PLD films 3. Results The growth temperatures of the different PLD anatase films are listed in Table 1. All films have similar thicknesses (about 50 nm) due to the The synthesis of anatase films by PLD has been already described in identical number of pulses used during the PLD growth. The structure Section 2.2. Willfully, the process developed for the preparation of PLD and surface morphology of sample H was studied after each of the 4 anatase films (involving substrate preparation and film annealing steps) consecutive annealing treatments (6 h in air each) at 600 ◦ C, 700 ◦ C, was also applied to the growth of TiO2 films by MBE. This allows to 800 ◦ C and 900 ◦ C carried out in a cylindrical ceramic furnace. check the similarities/capacities between the two synthesis procedures The analysis of the inner structure of the anatase (0 0 1) films was and, eventually, to unify the growth conditions of the anatase films or carried out with standard XRD laboratory techniques, i.e. (θ–2θ) and their respective preliminary and subsequent temperature treatments. reflectivity measurements. One of the diffractograms, measured for the The last part of each section focuses on the effect of annealing on the samples listed in Table 1, shows the indexed peaks relative to sample A grown films. (Fig. 2a) corresponding to the STO (S00L) and anatase (A00L) re­ flections (2θ1,2 = 37.6◦ and 80.5◦ ). The diffractogram shows the spec­ 3.1. Anatase growth mode ular reflections of the cubic STO cell and those of the tetragonal anatase cell. Their lattice vectors along the (X,Y,Z) directions are parallel to each The growth morphology of the anatase film on STO(0 0 1) by PLD has been well described by Radovic et al. [34] and would involve 3 stages Table 1 during the deposition process: pseudomorphic STO growth, intermedi­ List of the TiO2 thin films on STO substrates, grown by PLD, with corresponding growth temperatures. ate stage where TiO2 initially grows forming islands and anatase for­ mation (Supplementary information, section II). TiO2 film A B C D E F G H Fig. 1 shows a schematic view of the proposed structural growth T( C) ◦ 600 650 700 750 800 875 925 600 morphology of the anatase (0 0 1) film on SrTiO3 (0 0 1). The left side of Film H was annealed 4 times at 4 different temperatures. Film G was not the Fig. 1 simulate several processes that take place during growth: Sr- annealed. diffusion to the surface, islands formation and step flow growth. In the Fig. 1. Schematic diagram of TiO2 anatase (0 0 1) growth. Green, blue spheres correspond to Sr and Ti atoms, respectively. The largest/smallest blue spheres correspond to atoms placed in the cell with Y = 0 or Y = 1/2 coordinates, respectively. The red empty circles correspond to Sr diffusion from bulk to the surface. The images on the left side show an orientation (X, Z) identical to that on the right. 3 A. Crespo et al. Applied Surface Science 632 (2023) 157586 Fig. 2. a) X-ray diffractogram of sample A grown at 600 ◦ C and zoom into the anatase peaks before (black) and after (red) annealing. b) Sample A (600 ◦ C and 900 ◦ C). c) Sample F (875 ◦ C and 900 ◦ C). other, i.e., STO (aSTO, bSTO, cSTO) where aSTO = bSTO = cSTO = 3.905 Å equivalently the increase of the c-lattice parameter, depends on the final T and for anatase film (afilm, bfilm, cfilm) with afilm = bfilm and cfilm values applied to the film: GT and annealing treatment. A more detailed analysis listed in table TS1 at the corresponding preparation temperature (T). on this point is given in the discussion section. The same Fig. 3 shows the Fig. 2b and c show the (0 0 4) and (0 0 8) anatase peaks before (black) evolution of the particle size obtained from the AFM image analysis with and after (red) annealing at 900 ◦ C during 6 h in air using a ceramic the GT (red curve) so, this size defines the lateral dimensions of the oven. The anatase peaks move towards smaller 2θ-values which is islands that form the film. The particle size is not related to the crystal equivalent to an increase of the c-lattice parameter of the film. These 2θ- size, as particles can be formed by several crystals. Particle size, grain shifts are also observed in the same anatase films grown on substrates and island could be considered equivalent, so these terms will be used with lower lattice mismatch as a function of annealing temperature interchangeably throughout the document. [36]. As expected, the peak shift is much smaller for sample F than for The afilm values of the samples typed in Table 1 were obtained by sample A due to the proximity of the growth and the annealing tem­ performing X-rays-RSM (Reciprocal Space Mappings) measurements perature, confirming the T dependence of the peak shift. around the (1 0 3) reflection. These values are listed in Table TS1 and The c-lattice parameters (cfilm) of the freshly grown anatase (0 0 1) indicate constant biaxial compression over the explored T-range. The films {from 9.5014 to 9.5464 ± 0.0005 Å in the range 600–925 ◦ C} were evolution of the anatase cell volume with T is a relevant magnitude to measured at room temperature (RT) and normalized to the c-bulk understand the properties of these films related with their inner struc­ parameter (cbulk) of the tetragonal anatase crystal (cnorm = cfilm/cbulk), ture and stoichiometry. The evolution of the normalized film cell volume (abulk = 3.7845 Å, cbulk = 9.5143 Å [37]). Fig. 3 shows the values of the (Vnorm = Vfilm/Vbulk) is shown in Fig. 3 as a dashed black line crossing the normalized c-lattice parameters (cnorm) as function of the growth tem­ experimental black dots. The volume values of the film were calculated perature (GT) (upper black dots curve). The magnitude of a shift, or using the average in-plane lattice parameter < afilm> = 3.7733 Å ob­ tained from the values listed in Table TS1 and those of the cfilm given in the same figure, i.e. Vfilm = 2 * cfilm. Note that the Vnorm values are less than one indicating that the anatase film cell is under compression in this T-range. The compressive behaviour of the afilm lattice parameter has also been observed in similar thin films [30]. Furthermore, the measurement of similar afilm values corresponding to two samples with identical GT (875 ◦ C) using synchrotron diffraction techniques also shown smaller values than those of the bulk, i.e. 3.7784 +/− 0.0005 Å for PLD and 3.7811+/− 0.0005 Å for MBE (8 nm film). The alignment between the substrate and the film lattice vectors was also confirmed from this experiment, i.e. the (H,K) values of the anatase (10) reflection expressed in STO basis are (1.0335, − 0.0008). The thickness of the films listed in Table 2 were obtained from reflectivity measurements. Fig. 4 shows the X-ray diffractogram measured for sample A after its preparation by PLD. The data in Fig. 4 is expressed in units of L calculated using the cSTO = 3.905 Å lattice parameter. In this reciprocal space notation, the film thickness was calculated using the expression: ∑ m t= Δ cSTO n [TiO2 ]STO (1) n=1 Fig. 3. Evolution of the particle size of the anatase (0 0 1) films vs. growth T where n is the number of TiO2 cells forming the film, [TiO2] includes (red curve); normalized cfilm dependence on T (black dots); fit of the experi­ the atomic coordinates of the anatase (0 0 1) unit cell expressed in units mental points to a Gumbel distribution (blue line) [38]. 4 A. Crespo et al. Applied Surface Science 632 (2023) 157586 Table 2 mean square deviation of a sampling including 5 horizontal line profiles Content: thickness values calculated using the Snell’s law, the asterisk indicates (5 µm long) from AFM images. the value obtained from expression (1), and crystallite sizes (D) obtained using The θ–2θ diffraction patterns of anatase films are characterized by the Debye-Sherrer expression or Williamson-Hall method for the as-prepared the presence of peak broadening and by shifts of these peaks from the samples listed. ideal (or bulk) positions. The experimental profile of the diffraction TiO2 Thickness D D εmic (WH) Grain Terrace peaks results from the convolution of the intrinsic profile with the film (±1 nm) (Sherrer) (WH) size size instrumental contribution. Therefore, once the instrumental contribu­ (±6 nm) (±6 AFM (nm) tion is removed, the remaining contribution to peak broadening is nm) (nm) caused solely by the intrinsic properties of the material, e.g. the presence A 52/54* 32 39 0.0012710 357 12025 of uni- or multimodal distributions of small crystallites (coherent B 44 25 41 0.0017210 5510 30050* C 49 34 40 0.0010510 5710 diffraction domains) or structural imperfections (crystalline distortions, D 50 29 50 0.0014610 6010 40070* dislocations, stacking defects). In the case of heteroepitaxial films such E 44 26 39 0.0013510 6410 as the anatase of this work, the peak widths only reflect the projection of F 51 32 49 0.0016110 12025 the magnitudes of these intrinsic properties along the surface normal G 54 28 (<30 45 0.0014510 14027 650** 100 direction. In this work, all causes contributing to the additional peak >) (<40 >) broadening (with respect to the proper width of the film caused by its crystalline size) will be interpreted in terms of microstrain (εmic) [42] Their averages are indicated in the last row. The errors were obtained from the which can be understood as the presence in the film of a lattice constant discrepancies between the used methods: Snell/diffraction and Sherrer/Wil­ distribution due to the crystalline defects. X-ray broadening can be used liamson applied to their respective distributions. The microstrain (εmic) in fresh films does not follow a clear trend with T. The last column shows the evolution of to investigate the microstrain or the defect distribution in the film the terrace sizes measured by AFM after 6 or 12* hours annealing at 900 ◦ C. [43–45]. Conversely, when strain is uniform throughout the lattice, ** Sample G was annealed at 925 ◦ C during 12 h. lattice deformation manifests itself as a shift of the diffraction peaks (εLat) [46,47]. The main difference between the respective θ–2θ diffraction patterns of epitaxial films and powder samples concerns the structural informa­ tion they contain. In epitaxial films, the θ–2θ patterns provide access to structural information along one direction as, for example in the present case, where only reflections (00L) are available. In contrast, indexed powder diffraction patterns allow the study of crystal size and strain in all directions. Some of these defects commented above are reduced during annealing, as indicated by the decrease in microstrain with annealing temperature performed for a single sample, as shown in Table 3. A detailed XRD analysis of the evolution of the widths (β) and in­ tensities subjected to various annealing treatments at different temper­ atures (6 h each) was performed for sample H. Line broadening analysis of anatase film peaks performed using integral breadth (IB) methods [48,49] in combination with the Williamson-Hall method [41] offers a valuable tool for a preliminary assessment of the IB-related effects of crystallite size and lattice strain. This formalism has been applied to the anatase 2θ1,2 peaks as shown in Table 3. The estimated errors associated to these angles were obtained from their sensitivity to induce changes in strain. Similar procedure was followed for the estimation of the strain Fig. 4. Experimental reflectivity curve (black) for the anatase (0 0 1) film errors. The normalized cfilm lattice parameters are obtained from the 2θ1- (sample A) in units of the Miller index L corresponding to the STO cell (r.L.u. = positions of the table. reciprocal lattice units). The red curve shows the best fit. The maximum GT explored, as indicated in Table 1, was 925 ◦ C. of the STO, Δ is an offset of the anatase diffractogram, or equivalently, a normalized parameter that modifies the cSTO lattice value used to define Table 3 the TiO2 cell to correct for the diffractogram shift along L. In this base, Integral breadths measured on sample H versus temperature. the anatase TiO2 cbulk value is 2.4364 times greater than that of the cSTO. T (◦ ) a,b β1/2θ1 (◦ ) β2/2θ2(◦ ) εmic D (Å) cfilm/ The resulting film thickness is approximately 54.1 nm corresponding to cbulk n = 57 anatase cells with cfilm = cbulk Δ = 9.5143 * 0.9985 Å or 57 TiO2 600 a 0.3520/ 0.5823/ 0.0012710 39210 0.99851 cells expressed in STO units: 57 * (2.4364 * 3.905) * 0.9985 Å = 54.1 37.8511 80.8521 600b 0.3513/ 0.5817/ 0.0012710 39410 0.99991 nm. 37.8311 80.7901 The film thicknesses and crystallite (or crystal) sizes of the samples 700b 0.3223/ 0.5177/ 0.0009610 40310 1.00011 listed in Table 2 were calculated from the θ–2θ XRD diffractograms using 37.7881 80.6811 Snell’s law, Debye-Sherrer and Williamson-Hall (WH) formalisms, 800 b 0.3147/ 0.4809/ 0.0006510 37610 1.00241 respectively [39–41]. The table also includes the microstrain (εmic) 37.6991 80.4801 900b 0.3093/ 0.4723/ 0.0006210 38210 1.00311 values for each of the film growth temperatures, the particle sizes 37.6711 80.3851 measured by AFM and some terrace sizes after annealing at 900 ◦ C (G: 925 ◦ C) for 6 or 12 h. Note that the crystal size and strain values a and b super-indexes indicate the temperature of the ‘as-prepared’ or the calculated at each temperature via the WH formalism only use the two annealed sample. The β-widths have been calculated using the expression β = Area /Imax, from the area and maximum intensity of the peak (in units of detector anatase reflections contained in the measured θ-2θ X-ray region. The counts). The values of the microstrain (εmic) and the crystallite size (D) were experimental errors reported in the last two columns of Table 2, refer­ obtained from the Williamson-Hall method. The microstrain reduces with T as ring to island and terrace sizes, were obtained by calculating the root already reported [50,51] while cfilm increases (last right column). 5 A. Crespo et al. Applied Surface Science 632 (2023) 157586 However, the effect of high temperature annealing (T > 950 ◦ C) was also AFM images in Fig. 6. In the temperature range between 600 and 800 ◦ C investigated with the intention of establishing a maximum annealing the increase in grain size is small (only 25 nm), however, above 800 ◦ C temperature value before the anatase to rutile phase transition appears. the size increases markedly. This transformation starts to appear at 950 ◦ C showing weak contribu­ tions from the rutile and brookite phases which add to the anatase dif­ 3.3.1. Effect of heat treatment on PLD films fractogram, Figure S4. A relevant data obtained from the cfilm evolution with T (Fig. 3) is the The X-ray diffractogram of sample E after annealing it at 1000 ◦ C, indication of the temperature range (800–900 ◦ C) in which the Fig. 5, shows some reflections coming from the 3 polymorphs of TiO2. annealing process efficiently transforms the degree of crystallinity of According to its indexation, the rutile phase shows the (1 1 0) reflection both the film and its surface. Post-growth annealing alters the surface with respect to the STO substrate while brookite shows reflections of grain morphology of the films into larger flat terraces. grains with orientations (0 0 1) and (1 0 0) although this phase looks The overall effect of T-treatments after deposition has been described residual when comparing its intensity with those of the anatase and using sample H. This film initially presents a surface distribution of small rutile. The figure is expressed in units of the L-Miller index, as it is grains with an average size of 35 nm which transform into larger flat intended to reflect that the displacements detected in the anatase peaks terraces of about 120 nm after several annealing treatments. This evo­ are not detected in the STO peaks, as their positions coincide with lution can be observed in Fig. 7 and reveals the possibility to obtain large integer L-values. flat ordered terraces (around 120 nm) quite independent of the initial The first anatase peak in Fig. 5 is at LE,1 = 1.642 while the same peak grain size obtained at their corresponding GT. Successive long annealing in sample F is found at LF,1 = 1.637. The calculation of these L-values on steps (6 h each) at progressively increasing temperatures in ambient air the anatase base using the expression cr = cbulk/cSTO = 9.5143/3.905, conditions have the effect of smoothing the film surface by replacing the gives the values L’E,1= LE,1 * cr = 4.0006 and L’F,1 = LF,1 * cr = 3.9885. surface grain morphology by a flatter plate-like one. Fig. 7a shows the The influence of P(O2) during the growth process was also verified. surface grain morphology of anatase from the initial grown film. The Samples prepared at a lower P(O2): 0.013 mbar, show similar x-ray circular-shaped grain morphology changes to a square-shaped diffractograms and average grain sizes as those prepared at higher morphology after annealing the film at 700 ◦ C (Fig. 7b). At 800 ◦ C, pressures. The main difference between the samples grown at low and the grains coalesce into plates (Fig. 7c). Maximum coalescence is ach­ high P(O2) is observed in their respective topographies, which is more ieved at 900 ◦ C giving the largest possible terrace sizes (Fig. 7d) in this homogeneous when pressure is fixed at 0.11 mbar. A further increase in series of measurements. P(O2) to 0.153 mbar does not improve the surface morphology and the The terrace sizes shown in Fig. 7d are similar to those obtained by PLD pumping system becomes unstable. MBE by Du et al. [28]. In both cases the T and annealing time (900 ◦ C, 6 The P(O2) can also modify the cfilm magnitude of the prepared films. h) used in PLD were equivalent to the T and deposition time (750 ◦ C, 6 h) Figure S5 shows the effects of oxygen pressure during the growth process used in MBE. Both procedures show similar surfaces where the islands in cfilm for two different cases. The oxygen deficit, that is, P(O2): 0.013 coalesce to form flat terraces of approximately 100 nm. However, the mbar, increases the normalized cfilm value up to 1.01, much higher than internal structure, although similar, should not be stoichiometric Ti/O2 those shown in Fig. 3. Oxygen pressures >0.11 mbar do not cause in the MBE case, as demonstrated in Figure S5. The oxygen pressure used additional changes in the cfilm (Figure S5). during the growth process was 10− 8 mbar, below the critical value of 0.1 mbar shown in figure S5 for PLD, and below the 10− 6 mbar (section 2.3) used to grow already oxygen-deficient thin films by MBE. In any 3.3. AFM characterization of PLD films case, it is shown that similar film quality can be obtained using PLD and MBE in a two-step process: first growing the film to a given T (in oxygen- The surface morphology of the PLD films, i.e., the evolution of the saturated atmosphere if possible at least for PLD) followed by annealing lateral size of the TiO2 particles with deposition and T-annealing was at ambient pressure at temperatures above that of coalescence investigated by AFM as detailed in Fig. 3 (red curve). The size of the (>800 ◦ C). The MBE growth process as Ref. [28] represents the con­ islands observed on the surface of the anatase films measured by AFM is ventional option where operating conditions, O2 pressures and T, must highly dependent on GT, as shown in Fig. 3 (red curve). Film A prepared be compatible with UHV. at 600 ◦ C shows the smallest grain size (~35 nm) while film G grown at The effect of larger annealing times can be seen on Figure S6 (sample the highest T (925 ◦ C) exhibits the largest island size (~150 nm), see D). Twelve hours of annealing at 900 ◦ C generates terraces with sizes of approximately 400 nm. A similar 12-hour annealing of sample G at temperatures of 925 ◦ C increases the flat terraces to 650 nm, Fig. 8. This figure shows the AFM image of sample G grown at 925 ◦ C (left image) and subsequently annealed in air for 12 h (middle image: 3 h temperature raise from room temperature to 925 ◦ C, 6 h constant T at 925 ◦ C and 3 h drop to RT). The film surface shows the evolution from the typical distribution of square geometry islands formed during high temperature growth (Fig. 6G) to long flat terraces after annealing in air. Comparing this annealing with those performed in sample D (Figure S6) the size of flat terraces in­ creases from 400 nm to 650 nm. 3.4. LEED/STM characterization of MBE films TiO2 films of different thicknesses (2, 4, 8, and 16 nm) were grown by MBE under identical T and oxygen pressure conditions. Ti was evapo­ rated by keeping the STO substrate at 650 ◦ C and the chamber at 10− 6 mbar O2 pressure. At this stage, all prepared MBE films shown a diffuse LEED diffractogram with broad spots evidencing the formation of a Fig. 5. XRD diffractogram of sample E annealed at 1000 ◦ C. Main TiO2 phases poorly ordered surface structure. Fig. 9a shows the LEED diffractogram are rutile (R) and anatase (A). Weaker peaks belong to the brookite phase (B). of an 8 nm thick film. The samples obtained by MBE were subsequently 6 A. Crespo et al. Applied Surface Science 632 (2023) 157586 Fig. 6. AFM images of TiO2 PLD-fresh prepared films at different growth temperatures for the indicated samples. The sizes of images [(A–F), and (G)] are [(1 × 1 μm; zoom: 250 × 250 nm) and (5 × 5 μm; zoom: 1 × 1 μm)], respectively. Fig. 7. AFM images of sample H: (a) ‘as-prepared’ (600 ◦ C), and annealed (b) 700 ◦ C, (c) 800 ◦ C, and (d) 900 ◦ C. annealed ex-situ at 850 ◦ C in oxygen (ambient pressure) for 6 h. The pressure). After this process, the LEED image of the 8 nm thick sample films were returned to the UHV chamber and were sputtered/annealed shows a sharp pattern with two domains of the well-known (4 × 1) (Ar+ 10− 6 mbar, energy 500 eV, 2 min)/(850 ◦ C at 10− 6 mbar of O2 reconstruction routinely observed in anatase (0 0 1) single crystals [52] 7 A. Crespo et al. Applied Surface Science 632 (2023) 157586 Fig. 8. AFM images of sample G after PLD preparation (left) and annealed at 925 ◦ C during 12 h (middle). Line profile along the surface of the aneealed sample (right). The average surface terrace size is about 600 nm. Fig. 9. LEED diffractogram of an 8 nm thick anatase film prepared by MBE (top): (a) sample as deposited and (b) sample after annealing ex-situ at 850 ◦ C with. sputtering/annealing treatment. LEED diffractogram of a 55 nm thick anatase film prepared by PLD (bottom): (c) sample as deposited and (d) sample after sputtering and annealing at 850 ◦ C in O2. (Fig. 9b). Only in the thinnest deposit (2 nm) no LEED pattern was STM technique. Fig. 10a shows an AFM image of the surface of an 8 nm observed. To compare the surface structures of MBE and PLD films, a thick anatase film. The macroscopic AFM image of the PLD-grown PLD film grown at 875 ◦ C was introduced into the UHV system and samples (Fig. 7) shows flat terraces of about 100 nm, probably formed studied by LEED. This shows a diffuse pattern and faint spots (1 × 1) by two smaller grains than those observed in the STM images (Fig. 9c); however, after a soft sputtering/annealing cycle under con­ (Fig. 10b–c). These images show well-defined regions of the anatase ditions similar to those used to restore the surface of the MBE samples, a surface showing parallel rows along the main crystallographic directions comparable LEED pattern (4 × 1) was also obtained (Fig. 9d). This of the substrate surface with average gaps between them of approxi­ procedure shows that PLD and MBE techniques can be used to synthesize mately 1.6 nm, about four times the unit cell distance (1 × 1) ~15.12 Å. anatase (0 0 1) films with very similar surface structure. The images also show smaller crystal sizes of the anatase surface than To better understand the reconstruction (4 × 1), the atomic structure those of the PLD samples, probably due to the difference in thickness and surface morphology of anatase films were investigated using the between them. Indeed, a certain amount of material (thickness) is 8 A. Crespo et al. Applied Surface Science 632 (2023) 157586 Fig. 10. (a) 5 × 5 μm2 AFM image of a 8 nm thick film (b) STM images of a 8 nm thick anatase (0 0 1) film measured at several resolutions. The upper image corresponds to a zoom of the marked TiO2 grain. At the nanometer scale the typical rows of the (4 × 1)/(1 × 4) reconstruction are observed (c) line profile of the yellow line. needed to fill the spaces between the grains formed during growth. already been observed in anatase films [50]. The continuous blue curve These “square” shaped grains, easily identifiable on the AFM image of in Fig. 3 represents the calculated cnorm values derived by fitting the PLD films, accumulate holes or spaces between them that decrease in experimental ones to a cumulative distribution function (CDF) of the number after high-temperature annealing, where the spaces between Gumbel distribution [38] defined by expression (2) grains are substantially filled during their coalescence (Figures S6 and (2) e− (x− μ)/βo 10). The topography of the samples also indicates a correlation between cnorm (calc) = p + q(e− ) the amount of material deposited with the maximum achievable domain The exponential part, CDF(x : μ, βeta) = e , is a probability dis­ − X(x) size [28]. tribution function defined between 0 and 1. Here, p denotes a back­ ground baseline that shifts the function to values of cnorm(x) close to 1; q 4. Discussion is a scaling factor able to reduce the Gumbel distribution in the range 0.990 < cnorm(x) < 1.004; x is the temperature variable. The quantities 4.1. X-ray diffraction analysis defining the distribution are the standard deviation = βo π/ √6 with βo = 72.22479 and the mean = μ + γβo with γ ~ 0.5572 (Euler constant) and The X-ray diffractograms in Fig. 2 reveal that the anatase peaks: shift μ = 711.08209. βo and μ result from the adjustment to the experimental towards lower Bragg angles evidencing an increase in cfilm with T data (correlation value = 0.9989). (Fig. 2b); the increment between peak shifts is proportional to the dif­ The microstrain values shown in Table 2 do not follow a clear ference in T; they reduce their widths (as remarked in Table 3); and dependence on the GT of the film. The crystallite sizes (Table 2) derived increase their intensity. In addition, Table 3 shows the strongest from the WH-method are close to 40 nm, in good agreement with the decrease in peak widths in the T-region ≥ 700 ◦ C, which is related to an thickness values obtained from the XRR measurement for the same increase in film crystallinity along the [0 0 1] direction. Furthermore, samples. The crystallite sizes obtained from the Sherrer equation are upon annealing at high temperatures (Fig. 2c) the anatase peaks systematically smaller than those of the WH method, probably because continue to increase in intensity. However, the major structural rear­ the calculation does not take into account the microstrain of the film. rangement occurs when temperatures above 800 ◦ C trigger the coales­ Thus, the WH method gives out-of-plane crystallite sizes, as the re­ cence process, which fuses the crystalline grains that make up the film to flections do not include the H and K components, and microstrain values form large terraces on the surface, visible by AFM. This lateral structural that are of the same order of magnitude as those of the c-lattice strain rearrangement cannot be observed in the θ–2θ diffractograms. shown in Fig. 3. Table 2 also shows the evolution of the terrace size after The evolution of the peak shifts of anatase (0 0 1) films with T high temperature annealing at 900 ◦ C for 6 h (sample H) or 12 h (sample (growth or annealing), as mentioned in section 3.2, is related to a G). The effect of prolonged anneals at 900–925 ◦ C is to increase the continuous expansion of cfilm. The evolution of the normalized (cnorm) initial surface size of the islands obtained at the corresponding growth cfilm-parameter is defined by the upper black dotted curve in Fig. 3. temperature (600 ◦ C < T < 875 ◦ C) to much larger terrace sizes as a Below 700 ◦ C, cnorm is smaller than one, indicating a compression of the consequence of inter-island coalescence. cfilm parameter of 0.20 % at 600 ◦ C (compared to the bulk value at RT). The effect of annealing-T on strain (εmic) in anatase films from peak Conversely, at higher growth temperatures cfilm expands by up to 0.33 %, line broadening analysis using IB-methods is shown in Table 3. These i.e. at around 700 ◦ C the out-of-plane lattice parameter changes from a values are different, but similar in magnitude, to those of (εLat) obtained compressive to a tensile strain regime. This type of regime change has 9 A. Crespo et al. Applied Surface Science 632 (2023) 157586 from the conversion of the evolution of cfilm with T, from Fig. 3 (Table 2) strained lattice is considered to be possible [56]. In fact, the same system using the expression: such as SrRuO3 can grow with both types of strain depending on whether cfilm − cbulk it grows in SrTiO3 (0 0 1) or DyScO3 (1 1 0). εLat = (3) Anatase (0 0 1) films grown on STO in the T-range of 600 to 925 ◦ C cbulk show in-plane (afilm) biaxial compressive strain, while out-of-plane (cfilm) where cfilm and cbulk are the c-lattice parameters of the anatase film compressive strain occurs at temperatures below 700 ◦ C; at higher and bulk, respectively. temperatures cfilm changes to tensile strain. Biaxial compressive strain The maximum cfilm increase over the temperature range is 0.474 %. could then be the driving force favouring the growth of (0 0 1) anatase on The cfilm strain varies from − 0.0015 (600 ◦ C) < εLat < 0.0034 (925 ◦ C), STO (0 0 1) [57]. This surface possesses the largest surface stress among values which are of the same order as the microstrain values obtained by all rutile/anatase surfaces and occurs along the [1 0 0] direction [58]. the WH methods in Tables 2 and 3. Consequently, the relative stability of (0 0 1) surface can improve under Note the differences between the WH microstrain values given in the effect of in-plane compressive strain. Experiments applying external Tables 2 and 3. This presumed discrepancy observed in the different compressive biaxial strain show that the increase in the proportion of the microstrain evolution with temperature calculated using this method is anatase surface (0 0 1) is favoured [57]. The origin of this compressive not such since the values in Table 2 are from fresh films grown at stress could be attributed to the effect of the surface stress inducing different temperatures, while those of Table 3 come from a single sample compression in the islands before they become firmly attached to the that has been annealed several times at different temperatures. As surface. The compressive strain in the islands could be maintained and mentioned in Section 3.2, film reflections contain two different types of propagated throughout the growth process [59]. The inter-island stress deformation contributions, εmic and εLat, which could be related to each at their borders could help to its stabilization along the film. Further­ other. more, it is also reported that c-lattice strain depends on the thickness of The progressive decrease of the microstrain versus T in Table 3 is due the film, showing its compressive character for thin films; this behavior to the annealing effect that reduces the number of defects and micro­ was even detected in the plane for LAO substrates that have smaller distortions present in the films, manifesting as a decrease of the peak mismatch than STO [30]. Moreover, the same system synthesized using widths, while the evolution of the WH microstrain values in Table 2 can sol–gel technique on STO shows a range of temperatures from 700 ◦ C to be considered constant with T because the peak widths oscillate around 1100 ◦ C where cfilm shows a progressive and continuous increase in a mean value, i.e. ( = 0.34◦ ± 0.01◦ ). In this case, tensile strain [50]. similar levels of defects are generated in the freshly prepared films that are independent of temperature due to the rapid growth process asso­ 4.3. Relationship between lattice strain and oxygen vacancies in anatase ciated with the PLD technique [53]. On the other side, the progressive films decreasing of the 2θ peak positions in Table 3 is compatible with a progressive expansion of the cfilm parameter whose normalized values The values of the cell lattice parameters could delimit or suppress the are very similar to those shown in Fig. 3 for identical temperatures. presence of oxygen vacancies in the film; conversely, their presence The overall effect of annealing is the reduction of temperature- could also modify the magnitude of the cell volume. According to this dependent structural/morphological stress components accumulated premise, the evolution of the lattice strain associated with the anatase throughout the film during its PLD growth. This behavior is quantified in growth process, as a function of T shown in Fig. 3, could be assigned to Table 3 as a WH-strain (εmic) that decreases with annealing temperature. different concentrations of oxygen vacancies during film preparation. In The values of the widths and positions of the reflections collected in this order to know if the prepared films presented in this work are affected by table can be divided into two temperature regions: less (R1) or greater a dependence of the oxygen vacancies with T, or if the lattice values act (R2) than 700 ◦ C. The boundary between them marks the change of the as a conditioning factor to avoid their presence, the literature on the strain regime of the c-lattice parameter from compressive to tensile. In matter will be revisited and some experimental measurements, already the R1 interval, the morphology of the islands changes slightly from a presented in the results and supplementary material sections, will be spherical cap (Fig. 7a) to square shape (Fig. 7b) while the β-widths of the discussed to clarify this point. peaks become narrower (Table 3). The square shape of the islands is In agreement with the experiments, the diffusion coefficient of ox­ clearly visible in AFM images measured on films grown at high tem­ ygen ions is found to substantially increase under tensile biaxial stress peratures (Fig. 6: samples F and G). In the R2 region, the square islands and to decrease under compressive biaxial stress [60]. This effect has merge, restructuring the surface and forming larger terraces that been observed in various oxide film systems as CeO2 [61] and SmNiO3 generate a further reduction in structural strain [54,55]. This release of on STO and LAO [62]. More specifically, [63] investigated that oxygen microstructural deformation seems to be related to an increase of the defects increase lattice stress and expand the vacuum-annealed ceria cfilm strain, as the c-lattice parameter increases (last column of Table 3). lattice. However, with heat treatment these oxygen defects are removed However, this would not be entirely true in this system, since annealed and grain growth occurs. or grown anatase films at the same temperature show similar levels of In this work, the grown films show compressive biaxial strain over cLat strain, as shown in Table 3 and Fig. 3. Due to the lattice mismatch the range of temperatures explored, as shown in Table TS1, while the between the substrate and the film, an intrinsic lattice strain appears in out-of-plane lattice parameters show a change from compressive to the film, which would normally increase with thickness and relax above tensile strain with increasing temperature. Fig. 3 shows the evolution of a certain value [38]. In our 50 nm films, the out-of-plane lattice strain normalized cfilm (upper curve with solid black cicles) and Vfilm (lower component evolves from compressive to expansive with T, keeping the dashed black curve) against the growth temperature. The film cell vol­ film volume in the compressive regime (lower dashed black curve in ume was calculated considering the average value < afilm> = 3.7733 Å Fig. 3) over the whole T range. The decrease of the microstrain with obtained from Table TS1, and the values of cfilm given in Fig. 3. annealing T is not able to reduce the lattice strain generated during The film cell volume varies from 135.26 Å3 at 600 ◦ C to 135.89 Å3 at growth by the lattice mismatch. 925 ◦ C, while the bulk cell volume is 136.28 Å3 at RT. The cell volume of the film remains under compression throughout the entire temperature 4.2. Influence of biaxial lattice strain range (Fig. 3). This behavior would not agree with a film with a distri­ bution of oxygen vacancies homogeneously distributed throughout the The type of strain in the film, compressive/tensile, may be largely film, since such a distribution would produce an increase in the cell due to the difference in lattice parameters between the film and the volume of the anatase. substrate. If the misfit between both can be kept in the range ±7 % fully Figure S5 shows the reflection (0 0 4) (red curve) of two films grown 10 A. Crespo et al. Applied Surface Science 632 (2023) 157586 at 600 ◦ C at two different P(O2) of 0.013 mbar (red curve) and 0.11 mbar PLD: (H,K,L = 0,1,1)film = (0,1.0335, 0.4091)STO: (afilm, cnorm)PLD = (black curve); this last pressure is the one used to grow the films pre­ (3.77845, 1.00338). sented in this work. The oxygen deficient film increases the normalized MBE: (H,K,L = 0,2,4)film = (0, 2 * 1.0328, 1.6397)STO: (afilm, cfilm from 0.9982 (0.11 mbar) to 1.010 (0.013 mbar), a much higher cnorm)MBE = (3.78115, 1.00138). value than any of those shown in Fig. 3. The cnorm(PO2 = 0.11 mbar) = 1.010 generates a Vfilm (=136.80 Å3) > Vbulk which could be related to which are very close but not identical probably due to their different the presence of oxygen vacancies in the film. The blue and green dashed film thicknesses. curves, in Figure S5, correspond to two films grown at 875 ◦ C and P(O2) The preparation methods by PLD and MBE shown in Figs. 8 and 10 of 0.11 mbar and 0.153 mbar, respectively. In this case, both reflections generate patterns of coalescence of islands like those observed by other have identical values of 2θ. This experiment proves that the oxygen authors [28,34] with similar sizes, however, there are some differences pressure selected to grow the anatase films is adequate to avoid oxygen between them. The 10 nm film synthesized by PLD at 800 ◦ C [34] in­ vacancies in the films. dicates that island coalescence is reached at the temperature of growth, Furthermore, the comparison between the cnorm values of freshly however, the work does not offer any information on whether or not any grown and annealed films at different temperatures, in Fig. 3 and post-heat treatment was applied to the film. In our 50 nm anatase thin Table 3, are very similar. This means that a possible loss of oxygen films (Table 2), coalescence is always achieved after 6 h of annealing at during the annealing treatments would give higher normalized lattice 750–800 ◦ C or higher (Fig. 7). The samples prepared at any growth parameters than freshly grown samples, and this assumption has not temperature by PLD, without post annealing, never coalesce as evi­ been experimentally detected. denced in Fig. 6. Consequently, annealing of the film at high tempera­ These results show that the biaxial compressive strain detected in the tures (and in oxygen to avoid its deficiency) is mandatory to induce films, as well as the behaviour of the compressive strain along the out-of- coalescence of the grains or islands. The large domain sizes achieved by plane direction at temperatures below 700 ◦ C, are not an artefact MBE at a deposition temperature of 750 ◦ C [28] are recovered in the generated by the presence of oxygen vacancies in the films. present films after a long air annealing at high temperatures, although it is true that we have not carried out long anneals at 750 ◦ C because in our 4.4. PLD versus MBE growth techniques films the higher the annealing temperature is the larger the terraces are. The largest terrace size achieved in this work has been reached at an Film characterization using diffraction techniques (synchrotron and annealing temperature of 925 ◦ C (Fig. 8). In accordance with this trend, conventional) show that the [1 0 0], [0 1 0] and [0 0 1] directions of the the effect of annealing closer to the anatase/rutile transformation tem­ anatase film are aligned with those of the STO (0 0 1) substrate, as perature, detected at 950 ◦ C, would further increase the size of the mentioned in Section 3.2, and the orientation of the terraces follow this terraces. alignment. Coalescence results in the melting of the molten islands to form The SrTiO3 (0 0 1) substrates used in this work are TiO2-terminated larger terraces on the film surface. This process enhances atomic diffu­ with a double-height layer between neighbouring steps, with inter-step sion and consequently reduces the concentration of holes or gaps terrace widths of 250 nm (miscut ~0.1◦ measured from AFM images). observed between neighbouring islands (Fig. 6). However, these holes The two-step cleaning process, consisting of an annealing followed by a do not disappear completely. This atomic rearrangement induced during sonic bath in ultrapure water applied to the STO substrates, produces the melting process of the islands makes them larger, while also pro­ atomically flat Ti-terminated surfaces with a roughness level of <0.01 ducing a rearrangement of the distribution of the holes, reducing their nm. The main advantages of using this process are related to the safety of number but making them larger. These larger holes would presumably the chemicals required, as well as the absence of vacuum conditions, be formed by flat terraces of anatase towards deeper levels [28], which facilitates handling. In addition, samples can be stored in low- although the holes could also reach the substrate surface (Supplemen­ pressure chambers for a long time, whose surfaces can be regenerated tary information, section I). The maximum hole depth detected is close in the PLD chamber by pre-annealing at moderate temperatures in ox­ to 15 nm, as shown in Fig. 8, after long anneals at high T. The larger ygen, i.e. up to 850 ◦ C for one to two hours. However, the use of alter­ number of “hole-defects” in the GT films would result in a higher con­ native etching procedures [64] should be explored for the establishment centration of Ti3+ than the annealed samples, which would presumably of comparisons between the resulting TiO2 films obtained by PLD and make them more interesting for catalysis [29]. MBE from different substrates prepared with STO. The anatase to rutile transformation begins to appear at 950 ◦ C Fig. 6 demonstrates that the PLD growth time, even at high tem­ (Figure S4) but the anatase phase is still the majority phase present in peratures, cannot be considered an annealing since the morphology of the film at 1000 ◦ C (Fig. 5). This window of temperatures allows the the islands is visible over the entire range of T explored, i.e. up to 925 ◦ C. possibility of controlling the percentages of each phase in the film for In addition, Figure S6 also shows that island coalescence already takes specific purposes, for example, to study effects of the phase ratio on the place at annealing temperatures of 800 ◦ C, so any PLD film growing in properties of these catalysts…, since the anatase to rutile transformation the range of 600–925 ◦ C could change its surface morphology over time is progressive with T in these films. Even at higher temperatures, such as and annealing temperatures above 800 ◦ C. For such reasons the anatase 1100 ◦ C, complete conversion of anatase to rutile is still not achieved film synthesis involves two steps: growth at a given temperature plus a ([65] & ref. therein). long annealing in air or oxygen at high temperatures, e.g. 900 ◦ C. These The quality of the anatase (0 0 1) MBE and PLD films was tested by steps allow the anatase islands to coalesce into flat terraces (Fig. 7). This measuring the (4 × 1) surface reconstruction using STM techniques, as result also offers a host temperature range to grow anatase films by MBE mentioned in Section 3.4. The average sizes of the crystalline domains at temperatures equal or higher than 800 ◦ C since the island coalescence for both types of films were on average 120 nm. This size is still insuf­ process is already activated in the vicinity of this temperature [28]. ficient to use these samples for surface structure determination. How­ The two-step growth method (film growth and post-annealing ever, the average terrace sizes of 600 nm obtained after annealing at treatment in oxygen or air at ambient pressure) used during the syn­ high T for extended times (Fig. 8) are promising candidates for surface thesis of anatase films by the PLD and MBE techniques, produces films of X-ray diffraction experiments. So, the proposed annealing process could similar crystalline quality (Section 3.4) and show between them a very help to overcome this problem. similar lattice strain response to STO(0 0 1) mismatch. The lattice con­ Another aspect to be investigated is the quality of MBE growth of stants obtained from synchrotron diffraction experiments [29] for these these films by introducing monoatomic oxygen into the UHV chamber films after averaging several reflections to compensate miscut mis­ instead of gaseous molecular oxygen. In this case, the higher reactivity alignments along the surface normal direction are: of monoatomic oxygen would accelerate the growth process and reduce 11 A. Crespo et al. Applied Surface Science 632 (2023) 157586 the evaporation time. Supervision, Funding acquisition. V. Langlais: Investigation, Visuali­ In summary, as mentioned throughout the article, PLD and MBE zation, Formal analysis, Methodology. F. Carla: Supervision, Validation. techniques can share common steps that facilitate the film growth pro­ J.M. Caicedo: Methodology. J. Rius: Software, Supervision, Writing – cess: 1) surface preparation of the substrates by ultrasonic baths in review & editing. X. Torrelles: Conceptualization, Methodology, deionized water and annealing at thigh temperature, 2) introduction Investigation, Funding acquisition, Resources, Writing – original draft, into their corresponding chambers and cleaning of the carbon substrate Supervision, Project administration. surface by annealing at moderate temperature under oxygen pressure and 3) post annealing heat treatments after the film growth process at Declaration of Competing Interest high temperatures under ambient or oxygen pressure. The authors declare that they have no known competing financial 5. Conclusions interests or personal relationships that could have appeared to influence the work reported in this paper. Systematic growth of epitaxial anatase (0 0 1) films by PLD and MBE techniques shows similar film qualities after annealing for several hours Data availability in oxygen-rich atmospheres. The effects of temperature and time on film growth and the role of these two parameters during the film annealing Data will be made available on request. process have been studied independently. The post-growth annealing process enlarges the size of the ordered surface terraces in 50 nm thick Acknowledgements films to 400 nm as a consequence of the coalescence of the grains into flat terraces after 6 h of annealing at 900 ◦ C. The following funding is acknowledged: grant PID2021-123276OB- In addition, it is possible to control the average lateral size of the I00 funded by MCIN/AEI/10.13039/501100011033 and by “ERDF A terraces or domains of the anatase (0 0 1) surface, almost independently way of making Europe”; Severo Ochoa Programme for Centre of Excel­ of the growth temperature, by performing a long post-growth anneal at lence in R&D, FUNFUTURE (CEX2019-000917-S) funded by MCIN/AEI high temperatures (900–925 ◦ C for 6–12 h) in air or oxygen. A larger /10.13039/501100011033; grant LINKA20394 funded by CSIC, pro­ increase in terrace size would be expected as a result of island coales­ gram i-LINK 2021; synchrotron experiment was awarded by SOLEIL and cence from larger grain sizes, such as the films synthesized at 925 ◦ C performed on the CRG IF beamline at ESRF, France (proposal number (Fig. 6G). In fact, a 12-hour long anneal at 925 ◦ C on such a sample 20200213). J.M.C. also thanks Severo Ochoa Programme (Grant increases the surface terrace dimensions up to 650 nm (Fig. 8), high­ CEX2021-001214-S funded by MCIN/AEI/10.13039/501100011033). lighting the relevance of the combined effects of temperature and time on each other in these films. Experiments performed near the anatase-to- Appendix A. Supplementary material rutile temperature transformation (950 ◦ C–1000 ◦ C) offer a possible temperature window between 925 ◦ C and 950 ◦ C to optimize the Supplementary data to this article can be found online at https://doi. recrystallisation of anatase films before the anatase-to-rutile trans­ org/10.1016/j.apsusc.2023.157586. formation occurs. The evolution of the in-plane and out-of-plane lattice parameters References with temperature has been described for freshly grown anatase films. The out-of-plane lattice parameter evolves from a compressive to a [1] B. O’regan, M. Grätzel, A low cost, high-efficciency solar cell based on dye- tensile regime with increasing temperature, while the in-plane lattice sensitized colloidad TiO2 films, Nature 353 (1991) 737–740, https://doi.org/ 10.1038/353737a0. parameters remain under a compressive regime in the same temperature [2] A.L. Linsebigler, G. Lu, J.T. Yates, Photocatalysis on TiO2 surfaces: principles, range explored. The biaxial compressive strain behaviour of the in-plane mechanisms, and selected results, Chem. Rev. 95 (1995) 735–758, https://doi.org/ film lattice parameters induces a reduction of the cell volume over the 10.1021/cr00035a013. [3] M. Grätzel, Solar energy conversion by dye-sensitized photovoltaic cells, Inorg. whole temperature range explored. Chem. 44 (2005) 6841–6851, https://doi.org/10.1021/ic0508371. The evolution of the positions, widths and intensities of the anatase [4] S. Selcuk, A. Selloni, Facet-dependent trapping and dynamics of excess electrons at XRD peaks of the grown and annealed films with temperature indicates anatase TiO2 surfaces and aqueous interfaces, Nat. Mat. 15 (2016) 1107–1112, that two different deformation strain contributions have to be consid­ https://doi.org/10.1038/nmat4672. [5] S.Y. Huang, L. Kavan, I. Exnar, M. Grätzel, Rocking chair lithium battery based on ered: one generates the strain that causes the continuous expansion of nanocrystalline TiO2 (anatase), J. Electrochem. Soc. 142 (1995) L142, https://doi. the c-lattice parameter (εLat), the other (εmic) is temperature dependent org/10.1149/1.2048726. and reduces the defects and microdistortions present in the films during [6] L. Forro, O. Chauvet, D. Emin, L. Zuppiroli, H. Berger, F. Levy, High mobility n-type charge carriers in large single crystals of anatase (TiO2), J. Appl. Phys. 75 (1994) the annealing process. The reduction of microstructural strain comes at 633, https://doi.org/10.1063/1.355801. the cost of increased lattice strain in case the anneals are performed at [7] H. Tang, F. Levy, H. Berger, P.E. Schmid, Urbach tail of anatase TiO2, Phys. Rev. B temperatures above GT. The continuous increase of the c-lattice 52 (1995) 7771, https://doi.org/10.1103/PhysRevB.52.7771. [8] M. Kamei, T. Mitsuhashi, Hydrophobic drawings on hydrophilic surfaces of single parameter with T, annealing or GT, suggests that the deformation crystalline titanium dioxide: surface wettability control by mechanochemical associated with it is independent of the level of defects in the films and treatment, Surf. Sci. 463 (2000) L609–L612, https://doi.org/10.1016/S0039-6028 its dependence is mainly established through T, while the microstruc­ (00)00635-X. [9] K. Arifin, R.M. Yunus, L.J. Minggu, M.B. Kassim, Improvement of TiO2 nanotubes ture seems to be related mainly to the lattice mismatch as it does not for photoelectrochemical water splitting: review international, J. Hydrogen Energy evolve with GT. 46 (2021) 4998–5024, https://doi.org/10.1016/j.ijhydene.2020.11.063. In both types of films, PLD and MBE, the 4 × 1/1 × 4 anatase surface [10] Y.C. Nah, I. Paramasivam, P. Schmuki, Doped TiO2 and TiO2 nanotubes: synthesis and applications, ChemPhysChem 11 (2010) 2698–2713, https://doi.org/ reconstruction is observed in UHV conditions after two successive cycles 10.1002/cphc.201000276. of Ar+ ion sputtering (0.6 keV, PAr = 5 * 10− 6 mbar, t = 5 min) and [11] Y. Wang, S. Cao, Y. Huan, T. Nie, Z. Ji, Z. Bai, X. Cheng, J. Xi, X. Yan, The effect of annealing at 850 ◦ C under an oxygen pressure of 10− 6 mbar. composite catalyst on Cu2O/TiO2 heterojunction photocathodes for efficient water splitting, Appl. Surf. Sci. 526 (2020), 146700, https://doi.org/10.1016/j. apsusc.2020.146700. CRediT authorship contribution statement [12] B.Y. Cheng, J.S. Yang, H.W. Cho, J.J. Wu, Fabrication of an effcient BiVO4-TiO2 heterojunction photoanode for photoelectrochemical water oxidation, ACS Appl. A. Crespo: Conceptualization, Investigation, Visualization, Super­ Mater. Interfaces 8 (2016) 20032–20039, https://doi.org/10.1021/ acsami.6b05489. vision, Formal analysis. J. Gallenberger: Conceptualization, Investi­ [13] C.W. Kim, S.P. Suh, M.J. Choi, Y.S. Kang, Fabrication of SrTiO3–TiO2 gation. M. De Santis: Conceptualization, Visualization, Investigation, heterojunction photoanode with enlarged pore diameter for dye-sensitized solar 12 A. Crespo et al. Applied Surface Science 632 (2023) 157586 cells, J. Mater. Chem. 1 (2013) 11820–11827, https://doi.org/10.1039/ [41] G.K. Williamson, W.H. Hall, X-ray line broadening from filed aluminium and C3TA12604H. wolfram, Acta Metall. 1 (1953) 22–31, https://doi.org/10.1016/0001-6160(53) [14] M. Xu, Y. Gao, E.M. Moreno, M. Kunst, M. Muhler, Y. Wang, H. Idriss, C. Wöll, 90006-6. Photocatalytic activity of bulk TiO2 anatase and rutile single crystals using infrared [42] M.A. Krivoglaz, X-Ray and neutron diffraction in nonideal crystals, Springer, absorption spectroscopy, Phys. Rev. Lett. 106 (2011), 138302, https://doi.org/ Berlin, 1996. 10.1103/PhysRevLett.106.138302. [43] I. Groma, T. Ungár, M. Wilkens, Assymmetric X-ray line broadening of plastically [15] T. Luttrell, S. Halpegamage, J. Tao, A. Kramer, E. Sutter, M. Batzill, Why is anatase deformed crystals, I. Theory, J. Appl. Cryst. 21 (1988) 47–54, https://doi.org/ a better photocatalyst than rutile? - Model studies on epitaxial TiO2 films, Sci. Rep. 10.1107/S0021889887009178. 4 (2014) 4043, https://doi.org/10.1038/srep04043. [44] M. Wilkens, The determination of density and distribution of dislocations in [16] H. Tang, K. Prasad, R. Sanjines, P.E. Schmid, F. Levy, Electrical and optical deformed single crystals from broadened X-ray diffraction profiles, Phys. Status properties of TiO2 anatase thin films, J. Appl. Phys. 75 (1994) 2042–2047, https:// Solidi A 2 (1970) 359–370, https://doi.org/10.1002/pssa.19700020224. doi.org/10.1063/1.356306. [45] V.M. Kaganer, K.K. Sabelfeld, Strain distributions and diffraction peak profiles [17] H.G. Yang, C.H. Sun, S.Z. Qiao, J. Zou, G. Liu, S.C. Smith, H.M. Cheng, G.Q. Lu, from crystals with dislocations, Acta Crystallogr. Sect. A 70 (2014) 457–471, Anatase TiO2 single crystals with a large percentage of reactive facets, Nature 453 https://doi.org/10.1107/S2053273314011139. (2008) 638–641, https://doi.org/10.1038/nature06964. [46] S. Dolabella, A. Borzì, A. Dommann, A. Neels, Lattice strain and defects analysis in [18] X. Han, Q. Kuang, M. Jin, Z. Xie, L. Zheng, Synthesis of titania nanosheets with a Nanostructured semiconductor materials and devices by high-resolution X-ray high percentage of exposed (001) facets and related photocacatalytic properties, diffraction: theoretical and practical aspects, Small Methods 6 (31) (2022), J. Am. Chem. Soc. 131 (2009) 3152–3153, https://doi.org/10.1021/ja8092373. 2100932, https://doi.org/10.1002/smtd.202100932. [19] B. Wei, M. Calatayud, Hydrogen activation on anatase TiO2: effect of surface [47] W. Qin, T. Nagae, Y. Umakoshi, J.A. Spuznar, Relationship between microstrain termination, Catal. Today 397–399 (2022) 113–120, https://doi.org/10.1016/j. and lattice parameter change in nanocrystalline materials, Phylosoph. Magazine cattod.2021.11.020. Lett. 88 (2008) 169–179, https://doi.org/10.1080/09500830701840155. [20] N. Murakami, Y. Kurihara, T. Tsubota, T. Ohno, Shape-controlled anatase titanium [48] H.P. Klug, L.E. Alexander, X-ray diffraction procedures for polycrystalline and (IV) oxide particles prepared by hydrothermal treatment of peroxo titanic acid in amorphous materials, 2nd ed, Wiley, New York, 1974. the presence of polyvinyl alcohol, J. Phys. Chem. C 113 (2009) 3062–3069, [49] P. Scardi, M. Leoni, R. Delhez, Line broadening analysis using integral breadth https://doi.org/10.1021/jp809104t. methods: a critical review, J. Appl. Crystal. 37 (2004) 381–390, https://doi.org/ [21] G. Liu, C. Sun, H.G. Yang, S.C. Smith, L. Wang, G.Q.M. Lu, H.-M. Cheng, Nanosized 10.1107/S0021889804004583. anatase TiO2 single crystals for enhanced photocatalytic activity, Chem. Commun. [50] F.E. Oropeza, K.H.L. Zhang, A. Regoutz, V.K. Lazarov, D. Wermeille, C.G. Poll, R. 46 (2010) 755–757, https://doi.org/10.1039/B919895D. G. Egdell, Growth of epitaxial anatase nano islands on SrTiO3(001) by dip coating, [22] U. Diebold, The surface science of titanium dioxide, Surf. Sci. Rep. 48 (2003) Crys. Growth Des. 13 (2013) 1438–1444, https://doi.org/10.1021/cg301455q. 53–229, https://doi.org/10.1016/S0167-5729(02)00100-0. [51] H. Albetran, H. Hroosh, Y. Dong, V.M. Prida, B.H. O’Connoor, Low Phase [23] G.S. Herman, Y. Gao, Growth of epitaxial anatase (001) and (101) films, Thin Solid transformatios and crystallization kinetics in electrospun TiO2 nanofibers in air and Films 397 (2001) 157–161, https://doi.org/10.1016/S0040-6090(01)01476-6. argon atmospheres, Appl. Phy. A 116 (2014) 161–169, https://doi.org/10.1007/ [24] M. Lazzeri, A. Vittadini, A. Selloni, Structure and energetics of stoichiometric TiO2 s00339-014-8246-1. anatase surfaces, Phys Rev. B 63 (2001), 155409, https://doi.org/10.1103/ [52] M. Lazzeri, A. Selloni, Stress-driven reconstruction of an oxide surface: the Anatase PhysRevB.63.155409. TiO2 (001)-(1x4) surface, Phys. Rev. Lett. 87 (26) (2001), 266105, https://doi.org/ [25] G.S. Herman, Y. Gao, T.T. Tran, J. Osterwalder, X-ray photoelectron diffraction 10.1103/PhysRevLett.87.266105. study of an anatase thin film: TiO2(001), Surf. Sci. 447 (2000) 201–211, https:// [53] V.D. Mote, Y. Purushotam, B.N. Dole, Williamson-Hall analysis in estimation of doi.org/10.1016/S0039-6028(99)01186-3. lattice strain in nanometer-sized ZnO particles, J. Theor. Appl. Phys. 6 (2012) 6, [26] R. Hengerer, P. Bolliger, M. Erbudak, M. Grätzel, Structure and stability of the https://doi.org/10.1186/2251-7235-6-6. anatase TiO2 (101) and (001) surfaces, Surf. Sci. 460 (2000) 162–169, https://doi. [54] Y. Mukai, X-ray diffraction study on inelastic deformation behavior of Ni-base org/10.1016/S0039-6028(00)00527-6. single crystal superalloy, Proc. Struct. Integrity 2 (2016) 895–902, https://doi.org/ [27] Y. Liang, S. Gan, S.A. Chambers, E.I. Altman, Surface structure of anatase 10.1016/j.prostr.2016.06.115. TiO2(001): Reconstruction, atomic steps, and domains, Phys. Rev. B 63 (2001), [55] T. Horazdovsky, J. Drahokoupil, J. Jindra, P. Vlcak, Crystallite size and microstrain 235402, https://doi.org/10.1103/PhysRevB.63.235402. evolution in low-temperature annealed titanium, NANOCON (2019), in: [28] Y. Du, D.J. Kim, T.C. Kaspar, S.E. Chamberlin, I. Lyubinetsky, S.A. Chambers, In- Proceedings 11th intern. conf. on nanomaterials – research & application (2019), situ imaging of the nucleation and growth of epitaxial anatase TiO2(001) films on NANOCON (2019). https://www.confer.cz/nanocon/2019/download/32-cryst SrTiO3(001), Surf. Sci. 606 (2012) 1443–1449, https://doi.org/10.1016/j. allite-size-and-microstrain-evolution-in-low-temperature-annealed-titanium.pdf. susc.2012.05.010. [56] A. Chen, Q. Su, H. Han, E. Enriquez, Q. Jia, Metal oxide nanocomposites: A [29] Unpublished data. Home laboratory experiments, Surface X-ray diffraction and perspective from strain, defect, and interface, Adv. Mater. 31 (2019) 1803241, photoemission measurements in synchrotron radiation sources: ESRF and SOLEIL. https://doi.org/10.1002/adma.201803241. [30] R.J. Kennedy, P.A. Stampe, The influence of lattice mismatch and film thickness on [57] L. Jia, Da-Jun Shu, Mu Wang, Tuning the area percentage of reactive surface of the growth of TiO2 on LaAlO3 and SrTiO3 substrates, J. Cryst. Growth 252 (2003) TiO2 by strain engineering, Phys. Rev. Lett. 109 (2012), 156104, https://doi.org/ 333–342, https://doi.org/10.1016/S0022-0248(02)02514-9. 10.1103/PhysRevLett.109.156104. [31] G.Q. Liu, Z.G. Jin, X.X. Liu, T. Wang, Z.F. Liu, Anatase TiO2 porous thin films [58] J. Zheng, D. Shu, Regulation of surface properties of photocatalysis material TiO2 prepared by sol-gel method using CTAB surfactant, J. Sol-Gel Sci. Technol. 41 by strain engineering, J. Semicond. 41 (2020), 091703, https://doi.org/10.1088/ (2007) 49–55, https://doi.org/10.1007/s10971-006-0122-9. 1674-4926/41/9/091703. [32] P. Alphonse, A. Varghese, C. Tendero, Stable hydrosols for TiO2 coatings, J. Sol-Gel [59] G. Abadias, E. Chason, J. Keckes, M. Sebastiani, G.B. Thompson, E. Barthel, G. Sci. Technol. 56 (2010) 250, https://doi.org/10.1007/s10971-010-2301-y. L. Doll, C.E. Murray, C.H. Stoessel, L. Martinu, Stress in thin films and coatings: [33] J.G. Connell, B.J. Isaac, G.B. Ekanayake, D.R. Strachan, S.S.A. Seo, Preparation of Current status, challenges, and prospects, J. Vac. Sci. Technol. A 36 (2018), atomically flat SrTiO3 surfaces using a deionized-water leaching and thermal 020801, https://doi.org/10.1116/1.5011790. annealing procedure, Appl. Phys. Lett. 101 (2012), 251607, https://doi.org/ [60] Z. Zhou, D. Chu, Cl. Cazorla, Ab initio description of oxygen vacancies in 10.1063/1.4773052. epitaxially strained SrTiO3 at finite temperatures, Sci. Rep. 11 (2021) 11499, [34] M. Radovic, M. Salluzzo, Z. Ristic, R. Di Capua, N. Lampis, R. Vaglio, F. https://doi.org/10.1038/s41598-021-91018-4. M. Granozio, In situ investigation of the early stage of TiO2 epitaxy on (001) [61] D. Ma, Z.h. Lu, Y. Tang, T. Li, Z.h. Tang, Z. Yang, Effect of lattice strain on the SrTiO3, J. Chem. Phys. 135 (2011), 034705, https://doi.org/10.1063/1.3613637. oxygen vacancy formation and hydrogen adsorption at CeO2(111) surface, Phys. [35] A. Lotnyk, S. Senz, D. Hesse, Epitaxial growth of TiO2 thin films on SrTiO3, LaAlO3 Lett. A 378 (2014) 2570–2575, https://doi.org/10.1016/j.physleta.2014.07.006. and yttria-stabilized zirconia substrates by electron beam evaporation, Thin Solid [62] F. Conchon, A. Boulle, R. Guinebretière, C. Girardot, S. Pignard, J. Kreisel, F. Weiss, Films 515 (2007) 3439–3447, https://doi.org/10.1016/j.tsf.2006.10.106. E. Dooryhee, J.L. Hodeau, Effect of tensile and compressive strains on the transport [36] H.S. Jung, J.-K. Lee, J. Lee, B.S. Kang, Q. Jia, M. Nastasi, Strain relaxation in sol-gel properties of SmNiO3 layers epitaxially grown on (001) SrTiO3 and LaAlO3 epitaxial anatase thin films, J. Phys. Chem C 112 (2008) 4205–4208, https://doi. substrates, Appl. Phys. Lett. 91 (2007), 192110, https://doi.org/10.1063/ org/10.1021/jp076194n. 1.2800306. [37] C.J. Howard, T.M. Sabine, F. Dickson, Structural and thermal parameters for rutile [63] B. Choudhury, P. Chetri, A. Choudhury, Annealing temperature and oxygen- and anatase, Acta Cryst. B 47 (1991) 462–468, https://doi.org/10.1107/ vacancy dependent variation of lattice strain, band gap and luminescence S010876819100335X. properties of CeO2 nanoparticles, J. Exp. Nanosci. 10 (2015) 103–114, https://doi. [38] I.A.M. Huijben, W. Kool, M.B. Paulus, R.u.J.G. van Sloun, A Review of the Gumbel- org/10.1080/17458080.2013.801566. max trick and its extensions for discrete stochasticity in machine learning, IEEE [64] M. Kawasaki, K. Takahashi, T. Maeda, R. Tsuchiya, M. Shinohara, O. Ishiyama, Trans. Pattern Anal. Mach. Intell. (2022), https://doi.org/10.1109/ T. Yonezawa, M. Yoshimoto, H. Koinuma, Atomic control of the SrTiO3 crystal TPAMI.2022.3157042. surface, Science 266 (1994) 1540–1542, https://doi.org/10.1126/ [39] M.L. Vera, A.E. Ares, M.R. Rosenberg, D.G. Lamas, C.E. Schvezov, Determinación science.266.5190.1540. por reflectometría de rayos x del espesor de recubrimientos de TiO2 obtenidos por [65] H. Rath, S. Anand, M. Mohapatra, P. Dash, T. Som, U.P. Singh, N.C. Mishra, oxidación anódica, J. Anales AFA 21 (2009) 174–178. https://ri.conicet.gov.ar/ D. Kanjilal, D.K. Avasthi, Phase transformation of TiO2 from anatase to rutile by bitstream/handle/11336/59658/CONICET_Digital_Nro.a900e321-07e2-4ea9- thermal annealing and swift heavy irradiation, AIP Conf. Proc. 1063 (2008) 8f14-bdab21d354d1_A.pdf?sequence=2&isAllowed=y. 250–255, https://doi.org/10.1063/1.3027167. [40] A. Patterson, The Scherrer formula for X-ray particle size determination, Phys. Rev. B 56 (1939) 978–982, https://doi.org/10.1103/PhysRev.56.978. 13