Epitaxial Stabilization of Perovskite ATeO₃ Thin Films

Abstract

Tellurium oxides of the ATeO₃ form typically do not crystallize in perovskite structures. Here, we show that perovskite-like ATeO₃ (A = Ca, Sr, Ba) thin films can be grown on perovskite single-crystal substrates via epitaxial stabilization. These films are stable with high optical bandgaps, low dielectric losses, and a high electric breakdown strength. Hysteretic dielectric behavior found in SrTeO₃ and BaTeO₃ strongly suggests the presence of antiferroelectricity and ferroelectricity, respectively. These properties make perovskite tellurium oxides possibly appealing candidates for thin film coating or insulator materials in advanced microelectronics. Tellurium oxides constitute a largely unexplored class of materials that might show new and interesting functionalities in epitaxial thin-films. Our work encourages new work within this field.

1. Introduction

Ternary oxides have great potential to show properties that make them favorable for diverse applications in the science of materials. As such, perovskites with ABO₃ unit blocks have been largely explored, where A and B are typically metal cations. The stability range of these perovskites can be described surprisingly well using the Goldschmidt factor t = (r_A + r_B)/$\surd 2$ (r_B + r_O), where r_A, r_B, and r_O are the ionic radii of the anions and cations, respectively [1]. Goldschmidt originally proposed that materials with 0.8 < t < 1.0 are typically of a perovskite structure, where a deviation from the ideal value of 1.0 leads to distortions from the ideal cubic perovskite structure. Materials with t < 0.7 generally stabilize in other energetically more favorable structures. High-throughput ab-initio calculations and machine learning approaches have recently helped to refine the stability range of perovskites in order to facilitate structural predictions. Figure 1a shows a sketch of the perovskite stability range derived by Filip et al. [2] as a function of the tolerance factor t and the octahedral factor μ = r_B/r_O.

Based on this stability range, ternary oxides with alkaline earth metal ions A = Ca, Sr, Ba, and B-site tellurium cations fall on the edge of this perovskite stability range (see Figure 1a). According to Shannon et al. [3], the ionic radius of Te⁴⁺ (0.97 pm) is comparable to that of Ce⁴⁺ (0.94 pm). It is therefore reasonable to assume a similar structural behavior between ATeO₃ and ACeO₃ compounds. SrCeO₃ and BaCeO₃ both stabilize in perovskite structures with large oxygen octahedral rotations, whereas perovskite CaCeO₃ is unstable and decomposes to CaO and CeO₂ [4]. However, ATeO₃ does not form perovskite structures in bulk form. Due to the high covalent character of Te ions, tellurium oxides typically form highly complex phases with multiple polymorphs in close energetic proximity to the ground state phase. A particularly interesting example is SrTeO₃, with a room temperature non-polar phase that transforms into a ferroelectric phase above 310 °C [5]. The dielectric constant of SrTeO₃ reaches a value of about 10⁴ near the Curie temperature of 485 °C, comparable to the highest dielectric constant in the prototype ferroelectric BaTiO₃ [6].

This example suggests that new phases and functionalities might be found in thin films. However, to the best of our knowledge, epitaxial AteO₃ thin films have not been studied so far. In this paper, we report the successful growth of epitaxial ternary tellurium oxide thin films via pulsed laser deposition. We show experimental and theoretical evidence that ATeO₃ can be epitaxially stabilized as perovskites in thin films grown on single-crystal substrates.

2. Materials and Methods

2.1. Thin Film Preparation

Stoichiometric targets of CaTeO₃, SrTeO₃ and BaTeO₃ were prepared via standard solid-state synthesis. The targets were found to be of their bulk phases (see Figure 1b–d), with structural and dielectric properties closely resembling those reported in the literature [7,8,9]. Then, roughly 100 nm thick films were grown via pulsed laser deposition at a temperature of 550 °C and an energy density of 1.5 J/cm². The oxygen pressure was p_O2 = 0.1 mbar for CaTeO₃ and SrTeO₃, and 10⁻⁴ mbar for BaTeO₃. After growth, the samples were annealed for 5 min and cooled in 600 mbar of oxygen to ensure full oxygenation of the films. The laser deposition was carried out onto single-crystalline SrTiO₃ (STO) and LaAlO₃ (LAO) substrates with (001) orientation.

2.2. Characterization

X-ray diffraction (XRD) was performed with a Bruker Advance D8 (Billerica, MA, USA) diffractometer and a copper source. Density functional theory (DFT) calculations were performed with the Quantum Espresso software package version 6.4. [10,11] and are based on Perdew–Burke–Ernzerhof (PBE) exchange functionals and a 6 × 6 × 6 Monkhorst k-point grid. Projector augmented-wave pseudopotentials with an energy cut-off of 700 eV were employed. Raman measurements were only conducted for the films on LAO (001) substrates, since the scattering of SrTiO₃ substrates is too large. A Nanomics Imaging Multiview 1000^TM setup was used with a 532 nm laser.

Optical properties of the films have been determined via variable angle spectroscopic ellipsometry using a J.A. Woolam M-2000. The recorded ellipsometry data are fitted to a simple two-layer model consisting of the substrate and a film with a certain film surface roughness. The film layers are first approximated using a Cauchy model in order to obtain a good starting point for refinement using a Kramers–Kronig consistent B-spline fit with 19 data points over the full energy range. The B-spline fit was necessary accurately capture the absorption above the band gaps [12]. The substrates’ optical properties were fixed to dielectric functions, independently determined by fitting the data of the bare substrates.

In order to perform an electrical characterization, ATeO₃ films were grown on Nb–SrTiO₃ substrates and capped with 50 × 50 µm² large Au-top electrodes deposited by magnetron sputtering. Capacitance-voltage hysteresis loops were then recorded with an HP Agilent 4278A LCR meter. A low signal voltage and frequency of 0.2 V and 100 kHz, respectively, was used. The dielectric constants were calculated based on the top electrode area and the thin film thickness. Polarization loops were measured with an aixACCT TF Analyzer 1000.

3. Results and Discussion

3.1. X-ray Diffraction

Figure 2a,b show wide-angle XRD 2θ-θ scans of the ATeO₃ films grown on (001)-oriented STO and LAO substrates, respectively. All films are phase pure and grown epitaxially; i.e., the XRD pattern can fully be explained by a single structure with a homogeneous crystal orientation. At first sight, the XRD patterns resembled those of typical patterns of perovskite thin films grown with cube-on-cube epitaxy. In order to corroborate this observation, the existence of a variety of peak reflections was checked. It was found that for all three materials, all peak reflections were consistent with a perovskite structure.

Pseudocubic lattice parameters of the perovskite films were determined from reciprocal space maps measured around the (103)_pc reflections. Exemplary maps for the CaTeO₃/STO (001) film are shown in Figure 2c,d, respectively. It is apparent that the films are not coherently strained to the substrate, but show a large degree of strain relaxation. The calculated pseudocubic lattice parameters are listed in

Table 1. List of pseudocubic lattice parameters determined from the positions of XRD film peaks.

	CaTeO3 /STO	CaTeO3 /LAO	SrTeO3 /STO	SrTeO3 /LAO	BaTeO3 /STO	BaTeO3 /LAO
a [Å]	4.010	4.012	4.160	4.151	4.356	4.336
c [Å]	4.014	4.014	4.107	4.117	4.342	4.324

. In general, the lattice parameters increase within the Ca–Sr–Ba series, as can be expected according to the increasing ion size. For CaTeO₃, the in-plane and out-of-plane lattice parameters (a and c) are similar, which indicates full strain relaxation. In contrast, SrTeO₃ and BaTeO₃ show some degree of tetragonality. Apart from incomplete strain relaxation, this fact could be attributed to a variety of other reasons. For example, the sustained tetragonality could be due to a thermal expansion mismatch between the substrate and the film material. A second cause could be the creation of oriented defects during the deposition of non-stoichiometric films, as this has been previously observed in BaTiO₃ films [13]. Lastly, the films might show polar behavior, and the tetragonality could be a reflection of ferroelectric displacement. Since the out-of-plane lattice parameter is smaller than the in-plane parameter, the ferroelectric polarization would be oriented in the film plane.

We have observed that the films only grow in perovskite form when grown on perovskite substrates, while no epitaxial growth is observed on other substrates with similar lattice parameters (for example, MgO). We thus deduct that the structural similarity is important. For the ATeO₃ films, it is energetically more favorable to stay in a perovskite structure than introduce large amount of defects and heavily reconstruct at the interface. In summary, our XRD results suggest that perovskite ATeO₃ films can be stabilized with epitaxial growth on perovskite substrates. Our measurements also show that the phases remain stabled over the months and no noticeable destabilization was observed in a humid atmosphere. We do note, however, that the BaTeO₃ films needed to be grown in a low-oxygen pressure environment to produce a decent crystalline quality. We speculate that during PLD growth, this tellurate is in competition with another oxide with a different tellurium valence state, most likely BaTeO₄.

3.2. Density Functional Theory Calculations

DFT calculations were performed on the bulk ATeO₃ phases and their perovskite counterparts. Since perovskites with a tolerance factor of t < 1 typically lower their energies through oxygen octahedral rotations, it is very likely that the ATeO₃ films do not stabilize in the ideal perovskite structure. A complete exploration of the full phase space of various octahedral rotation patterns using DFT would be beyond the scope of this work. Therefore, we restrained our calculations to the ideal cubic perovskite phase (space group Pm3m) and the perovskite phase with an a⁻a⁻c⁺ octahedral rotation pattern (space group Pnma). The latter is often found in perovskites with large B-site ions (i.e., low tolerance factors), such as SrCeO₃ or SrZrO₃.

The calculated energy differences between all three structures are depicted in Figure 3. The DFT results show three results that we highlight as significant:

• The distorted perovskite phases of all three ATeO₃ compounds are very close in energy to their bulk ground states, with energy differences <50 meV/atom. These small energy differences directly suggest that the perovskite structures can be energetically favorable over the bulk phases, which are geometrically less compatible with perovskite single-crystal substrates and would require large lattice alterations at substrate–film interfaces.
• The energy difference between the ideal Pm3m perovskite phases and the bulk phases reduces with increases in the A-site ion radius, from CaTeO₃, over SrTeO₃, to BaTeO₃. This observation is in agreement with the fact that the ATeO₃ moves farther into the perovskite stability range as the tolerance factor is increased.
• The energy difference between the cubic Pm3m and the orthorhombic Pnma phases decreases with increases in the A-site ion radius. This is another indication that the energy gained through oxygen octahedral rotations and tilts is reduced as the tolerance factor approaches one. This same trend can, for example, be observed in the AZrO₃ series, where CaZrO₃ has strong a⁻a⁻c⁺ octahedral rotation, while BaZrO₃ has an ideal cubic perovskite structure.

It should be noted that further theoretical and experimental work is needed to determine the exact lattice structure of the perovskite ATeO₃ phases. Since Te⁴⁺ is stereoactive, cation displacements may be present that would lead to polar or anti-polar phases.

3.3. Raman Studies

Raman spectra of both the actual samples and the bare substrates were recorded in order to subtract the substrates’ background and obtain spectra of the film material. These results are presented in Figure 4. For all three films, we find a strong Raman peak near 720–750 cm⁻¹. A weaker peak with varying prominence and width is observed about 100 cm⁻¹ below the main peak. Especially for SrTeO₃, a broader shoulder is evident that is reminiscent of second-order scattering in SrTiO₃ [14].

Raman modes in the 500–800 cm⁻¹ wavenumber range can usually be attributed to Te–O stretching motions [15]. The typical Te–O_x geometries in tellurium oxides are TeO₃ pyramids, TeO₄ disphenoids, and TeO₆ octahedra. The symmetric stretching Ag mode of the TeO₃ pyramid, as is present in, for example, bulk BaTeO₃, has previously been assigned to a Raman peak at 750 cm⁻¹, i.e., one significantly higher than in our film [16]. Also, an anti-stretching mode of similar intensity is observed at lower wavenumbers, which is not present in our spectra. We therefore conclude that the presence of TeO₃ pyramids is unlikely in the AteO₃ films.

TeO₄ disphenoids are present in both the α- and γ-phase of TeO₂. For both of these two phases, the most intense Raman line is located below 680 cm⁻¹ and multiple strong lines are found in the range of 400–600 cm⁻¹ [17]. This observation is in disagreement with our Raman data and we deduce that TeO₄ disphenoids are unlikely to be present in all ATeO₃ films.

The Raman spectra of tellurium oxides with TeO₆ octahedra, such as Ln₂TeO₆ (Ln = Y, La, Sm) (Ref. [18]), do show a strong Raman peak at 720–750 cm⁻¹ and a second weaker peak about 100 cm⁻¹ lower. Thus, our data suggest the presence of TeO₆ octahedra in our ATeO₃ films.

We have also used DFT to calculate Raman wavenumbers and intensities for Pnma-SrTeO₃ (see Figure 4b). The strong A_g Raman mode at 750 cm⁻¹ agrees very well with the experimentally observed peak. The calculation also provides a pair of modes at 710 cm⁻¹, which could be assigned to the weaker peak found in the measurement. Computationally, much more demanding calculations would be necessary to increase the reliability of these Raman calculations and include second-order effects. However, we take the DFT results presented here as another indication that the tellurium in the ATeO₃ films is coordinated with O₆ octahedra, as one would expect in a perovskite.

3.4. Optical Properties

Optical properties of the films have been determined using variable angle spectroscopic ellipsometry. Figure 5 shows a Tauc plot of the absorption coefficient α for the ATeO₃ films grown on STO (001) substrates. All three perovskite films show little absorption at lower energies and a sharp increase above their band gaps. This behavior is typical for perovskite large-bandgap insulators. A linear extrapolation of the absorption coefficient to zero provides a rough estimate of the optical band gap. The perovskite BaTeO₃ and SrTeO₃ films have a gap of around 3.9 and 4.2 eV, respectively, while that of CaTeO₃ is even higher. This indicates that these films could be used as transparent, and possibly highly insulating, buffer layers in perovskite heterostructures. The increase in the band gap with decreasing ion size is consistent with general trends in metal oxides, such as, for example, the ATiO₃ and ACeO₃ families [19,20].

The refractive index of the ATeO₃ films, as determined from the ellipsometry, is plotted in inset of Figure 5. A trend of an increasing refraction index with increasing photon energy and ion size is observed.

3.5. Dielectric Characterization

The relative dielectric permittivity ε_r as a function of bias voltage is shown in Figure 6a for all three materials. We find that CaTeO₃ shows predominately dielectric or paraelectric behavior with a weak non-linearity. SrTeO₃, on the other hand, exhibits a clear hysteretic behavior, with a loop shape that is typical for antiferroelectrics. In order to corroborate this finding, a polarization loop was measured. The loop is depicted in Figure 6b. When the data are corrected for parasitic capacitance and leakage current, clear characteristics of an antiferroelectric hysteresis loop become evident.

A notable hysteretic behavior is also seen in the dielectric constant of BaTeO₃. In this case, the butterfly-like hysteresis loop is reminiscent of ferroelectric behavior. Unfortunately, we were not able to obtain clean polarization loops to confirm this observation. We also note that piezoresponse force microscopy was conducted for all three films, but it revealed no clear evidence of ferroelectric behavior.

All three materials exhibit quite a high dielectric permittivity (ε_r = 18–37). Due to its large bandgap, CaTeO₃ in particular shows low dielectric losses and a high electric breakthrough strength of several MV/cm. This observation makes it a possible candidate to consider as a gate oxide or coating material for thin-film devices. The increasing ε_r with increasing ion size is a further indication of polar tendencies in the ATeO₃ perovskites and is typical for metal oxides. Good examples of this are titanates, where CaTiO₃ and SrTiO₃ are incipient ferroelectrics and a ferroelectric state can be induced by small lattice distortions. BaTiO₃ is a classic ferroelectric.

Ferroelectric and antiferroelectric behavior are of interest for potential applications and deserve further investigation. Further optimization of thin film growth conditions and capacitor geometry is needed to gain a conclusive result. Nevertheless, our initial data indicate that the stereoactive Te⁴⁺ ions in perovskite ATeO₃ films promotes polar behavior.

4. Conclusions

Epitaxial films of ATeO₃ (A = Ca, Sr, Ba) were grown on perovskite single-crystal substrates. X-ray diffraction and Raman spectroscopy were used to show that the epitaxial films are stabilized in perovskite structures, phases that are not observed in bulk ceramics for these materials. An array of DFT calculations show small total energy differences (<50 meV/atom) between the perovskite phases and their bulk structures. Our measurements show that the films have large optical bandgaps, high dielectric permittivities, and low dielectric losses. They can be considered as potential materials to use as insulators in thin film applications. Hysteretic behavior was found in the SrTeO₃ and BaTeO₃ films, strongly suggesting the presence of antiferroelectricity and ferroelectricity, respectively. The formation of ternary tellurium oxides with perovskite structures can motivate future exploration of new functionalities based on the properties of their stereoactive chalcogen cations.