Tuning piezoelectric properties through epitaxy of La2Ti2O7 and related thin films

Current piezoelectric sensors and actuators are limited to operating temperatures less than ~200 °C due to the low Curie temperature of the piezoelectric material. Strengthening the piezoelectric coupling of high-temperature piezoelectric materials, such as La2Ti2O7 (LTO), would allow sensors to operate across a broad temperature range. The crystalline orientation and piezoelectric coupling direction of LTO thin films can be controlled by epitaxial matching to SrTiO3(001), SrTiO3(110), and rutile TiO2(110) substrates via pulsed laser deposition. The structure and phase purity of the films are investigated by x-ray diffraction and scanning transmission electron microscopy. Piezoresponse force microscopy is used to measure the in-plane and out-of-plane piezoelectric coupling in the films. The strength of the out-of-plane piezoelectric coupling can be increased when the piezoelectric direction is rotated partially out-of-plane via epitaxy. The strongest out-of-plane coupling is observed for LTO/STO(001). Deposition on TiO2(110) results in epitaxial La2/3TiO3, an orthorhombic perovskite of interest as a microwave dielectric material and an ion conductor. La2/3TiO3 can be difficult to stabilize in bulk form, and epitaxial stabilization on TiO2(110) is a promising route to realize La2/3TiO3 for both fundamental studies and device applications. Overall, these results confirm that control of the crystalline orientation of epitaxial LTO-based materials can govern the resulting functional properties.

One way to increase the macroscopic piezoelectric response of LTO for use in sensors and devices is to align the piezoelectric direction across the material by synthesizing oriented 7 or single 6 crystals. One method to accomplish this is by deposition of epitaxial thin films of LTO on lattice-matched substrates. A similar approach was recently shown to enhance the piezoelectric response of epitaxial films of ferroelectric (Bi 0.5 Na 0.5 )TiO 3 -(Bi 0.5 K 0.5 ) TiO 3 with the (100) orientation 12 . Previously, epitaxial thin films of LTO have been deposited by molecular beam epitaxy (MBE) 13,14 , pulsed laser deposition (PLD) [15][16][17] , and sputtering 18 . SrTiO 3 (STO) is the typical perovskite substrate choice (cubic perovskite, a = 3.905 Å). The best lattice match is found to be LTO(100) || STO(110), which places the additional oxygen planes parallel to the growth surface 13,14 . Epitaxial LTO (210) films have also been obtained on STO(001) substrates; in this case, the monoclinic distortion of the LTO crystal structure results in a tilt of ~4.5° between LTO [210] and STO[001] 16 . In this orientation, the additional oxygen planes, and thus the ferroelectric direction, are oriented approximately 45° out of the film plane.
We have identified rutile TiO 2 (110) as a potential substrate upon which to deposit epitaxial LTO(010) films, which has the potential to orient the piezoelectric direction out of the film plane. Rutile TiO 2 (110) possesses a reasonable lattice match to LTO in one direction, and a coincident lattice match in the perpendicular direction. In this paper, we present the crystallographic and piezoelectric properties of LTO epitaxial thin films deposited by PLD on STO(110), STO(001), and rutile TiO 2 (110) substrates.

Results and Discussion
LTO on STO(110). As shown in Fig. 1(a), deposition of LTO on lattice-matched STO(110) did not result in a well-crystallized PLS phase for any deposition conditions (750-1000 °C, 0.5-120 mTorr O 2 ). Many films produced an XRD pattern similar to curve (i) in Fig. 1(a), with a broad peak at ~25-30° 2θ. At best, the well-crystallized peaks observed in curve (iii) were obtained, although the broad peak was consistently present as well. Surprisingly, the sharp peaks in curve (iii) could not be indexed to any known La n Ti n O 3n+2 phase. The peak spacing is consistent with a cubic lattice of repeat distance 6.75 Å. When films with this unidentified phase are annealed at 1100 °C for 4 h in air, the XRD pattern completely transforms to well-crystallized PLS with little or no evidence of the previous diffraction features, as shown in curve (iv). Annealing the poorly-crystallized film (curve (i)) under similar conditions produces a pattern (curve (ii)) which indicates partial crystallization in the PLS phase, but the broad peak widths and low intensities indicate that this film is poorly crystallized compared to the annealed film in curve (iv).
The XRD results for LTO/STO(110) are corroborated by STEM-HAADF imaging of the films before and after annealing. Figure 2(a) presents a cross-sectional STEM-HAADF image of LTO/STO(110) analogous to the XRD pattern of curve (iii) in Fig. 1(a). Although the film has a sharp interface with the substrate, significant disorder and phase separation are observed in the bulk of the film. A thick (10-15 nm) amorphous layer is present at the film surface. The high-resolution image in Fig. 2(b) reveals the complex microstructure of this film. On the left side of the image, a square lattice pattern is observed which appears similar to that of the STO(110) substrate; we tentatively assign this to perovskite LaTiO 3 . A disordered region separates this LaTiO 3 phase from the well-crystallized phase that dominates the film. The repeat distance of this crystalline phase is ~7 Å, matching reasonably well (within the error of the STEM measurement) to the 6.75 Å repeat distance observed in the XRD pattern for the unidentified phase. At the LTO/STO interface, a unit cell of La 5 Ti 5 O 17 (n = 5) has nucleated ( Fig. 2(b)); La 5 Ti 5 O 17 is similar to La 2 Ti 2 O 7 , but with five perovskite layers between additional oxygen planes 19 . It should be noted that the diffraction peaks observed in Fig. 1(a) do not correspond to the diffraction pattern of La 5 Ti 5 O 17 20 . After annealing, the unidentified phase has transformed to a well-crystallized and stoichiometric PLS phase, as shown in Fig. 2(c) and (d). As expected from the XRD patterns, the additional oxygen planes lie parallel to the STO substrate, with LTO[001] // STO [110]. In addition to the PLS phase, regions of perovskite LaTiO 3 still remain.
The behavior of La-Ti-O films deposited on STO(110) as a function of oxygen pressure during deposition is counterintuitive. Deposition at low oxygen pressure (0.5 mTorr) results in an unidentified phase; annealing in air recovers the PLS LTO phase. It might be expected that deposition at low oxygen pressure results in an oxygen-deficient phase, and increasing the oxygen pressure would result in the direct formation of PLS LTO. Instead, deposition with increased oxygen pressure results in poorly-ordered or nearly amorphous films. This points to the substantial energy required to form the PLS phase: in our PLD system, the relatively large distance between target and substrate (7.5 cm) means that an increase in background oxygen pressure significantly decreases the kinetic energy of species in the ablation plume, and this reduced kinetic energy is insufficient to form the PLS phase at the growth temperature employed (≥900 °C).
The unidentified phase observed in Fig. 1(b), curve (iii) and Fig. 2(b) closely resembles both the La 2 Ti 2 O 7 and La 5 Ti 5 O 17 crystal structures, and in fact appears to match the A 2 B 2 O 8 layered structure 10 . Although "La 2 Ti 2 O 8 " cannot exist as a charge-neutral compound (only BaMF 4 with M = Mn, Fe, Co, Ni, Zn, Mg compounds are known 10 ), we hypothesize that the unidentified phase maintains the La 2 Ti 2 O 7 stoichiometry, but the deposition conditions have not provided enough kinetic energy to sufficiently order the oxygen planes as well-defined layers between every four perovskoite slabs. Instead, partially complete oxygen planes have formed between every two perovskite slabs, and this appears as the A 2 B 2 O 8 structure in TEM (which is insensitive to precise oxygen stoichiometry). The crystal structure of the La-Ti-O system does not necessarily directly correlate to the oxygen stoichiometry 10 , and thus we speculate that the La 2 Ti 2 O 7 stoichiometry can exist in both the A 2 B 2 O 8 and the ABO 3 crystal structures with disordered oxygen vacancies or excess oxygen dopants to maintain charge neutrality.  (001), a single high-resolution θ−2θ XRD scan will not capture both film and substrate peaks 16 . Fig. 1(b) shows the out-of-plane XRD patterns when the diffractometer is aligned to the substrate (curve (i)), and when it is aligned to the diffraction feature at ~41° 2θ (curve (ii)). In curve (ii), the diffraction peaks corresponding to the STO(001) substrate do not appear, but three diffraction features remain. Two of these peaks can be indexed to LTO(021) (2θ = 21.135°) and LTO(042) (2θ = 43.034°), which is the expected orientation for epitaxial LTO on STO(001) 16 . The diffraction feature at ~47.7-48.1° 2θ cannot be indexed to an LTO plane in the 〈021〉 family 11 . As revealed by the reciprocal space map (RSM) plotted in Fig. 1(c), this diffraction feature exhibits the same alignment with the STO substrate as does the LTO(420) reflection, and thus corresponds to an epitaxial phase. Three lobes from each feature are present in the RSM, corresponding to planes aligned parallel to the STO(001) surface (intensity at 0° from the surface normal), and planes tilted ±~4° away from the surface normal. An additional diffraction feature between the LTO(021) and STO(001) peaks (~22° 2θ) is present when aligned to the substrate, but does not appear when aligned to the LTO film. This peak arises from a secondary phase of perovskite LaTiO 3 (001), which is well aligned to the STO(001) substrate. After annealing at 1100 °C for 4 h in air, little change is observed in the diffraction features (curves (iii) and (iv)).
The primary epitaxial orientation, LTO(021)//STO(001), is confirmed by the STEM-HAADF images in Fig. 3. In these images, the LTO(001) direction lies approximately 45° to the substrate, and thus the additional oxygen planes also lie along this direction. On the STO(001) surface, which possesses a cubic surface net of atoms, there are four equivalent possible orientations (separated by 90° around the STO(001) surface normal) for the LTO(021) film to nucleate and grow. In Fig. 3, twinned left and right orientations of LTO[001], separated by 180°, are shown. The other two equivalent orientations (LTO[001] pointed out of the image and into the image) are likely also present, but are difficult to distinguish; in these orientations, the atomic columns of the PLS structure are not highly aligned, and thus these orientations would appear weakly cubic but fairly disordered in the STEM images. Disordered regions are observed in Fig. 3(a) which might correspond to these orientations. However, the cubic atomic structure observed in portions of these regions may also arise from perovskite LaTiO 3 secondary phases, since LaTiO 3 (001) was also observed in the XRD patterns ( Fig. 1(b)).

LTO on TiO2(110).
To orient the film such that the direction of ferroelectric coupling (the LTO[010] direction) is pointed out of the plane of the film, LTO depositions were attempted on rutile TiO 2 (110). There exists a reasonable lattice match of 1.3% between the Ti spacing along the TiO 2 [110] direction (d = 6.496 Å) and the average La spacing along the LTO[100] direction (d average = 6.58 Å). The lattice match in the perpendicular in-plane direction is not as favorable: the Ti spacing in the TiO 2 [001] direction is 2.958 Å, while the average spacing between La cations in the LTO[001] direction is 3.92 Å, resulting in a lattice misfit of 32.5%. However, there exists a coincident lattice match between LTO and TiO 2 in this direction, with four TiO 2 units (11.832 Å) nearly equal to three LTO units (11.76 Å). Combined with the favorable lattice match in the LTO[100] direction, epitaxy may be possible. As shown in curve (i) of Fig. 4(a), deposition on TiO 2 (110) does not result in an LTO film with the (010) orientation. Instead, surprisingly, the same A 2 B 2 O 8 -like phase as observed in depositions on STO(110) results, with the same [001] out-of-plane orientation. As with LTO/STO(110), annealing the LTO/TiO 2 (110) film at 1000 °C for 8 h in air transforms this phase (Fig. 4(a), curve (ii)). However, in contrast to the deposition on STO(110), the dominant orientation on TiO 2 (110) appears to be LTO(010); this is the expected orientation from the lattice matching argument above. A weaker set of diffraction peaks corresponding to a secondary epitaxial orientation of LTO(001) is also observed. Interestingly, deposition of a thicker (1000 Å) LTO film on Nb-doped TiO 2 (110), annealed at 1000 °C in air for 8 h, results in an XRD pattern (curve (iii)) which is free of secondary orientations, and appears to consist only of LTO 〈0l0〉 peaks. However, glancing-angle μXRD shown in Fig. 4(b) reveals that the dominant crystal structure matches best to La 2/3 TiO 3 , an orthorhombic perovskite phase with ordered A-site vacancies 21,22 , not La 2 Ti 2 O 7 . The orthorhombic (102) plane (pseudocubic (101)) of La 2/3 TiO 3 is  Fig. 4(c), the orthorhombic phase exhibits in-plane as well as out-of-plane epitaxial relationships to the Nb:TiO 2 substrate. The correspondence of the La 2/3 TiO 3 (102) reflection position to the "LTO(020)" reflection in Fig. 2(a) curve (ii) confirms that the oriented phase in this film is also La 2/3 TiO 3 (note that the assignment of the PLS LTO(001) orientation doesn't change).
The STEM-HAADF images presented in Fig. 5 provide more insight into the structure of La 2/3 TiO 3 and LTO on TiO 2 (110). As shown in Fig. 5(a), the as-deposited LTO film (curve (ii) in Fig. 2(a)) is crystalline at the TiO 2 (110) interface and extending approximately halfway through the film thickness, but the top portion of the film (~30 nm) is amorphous. This amorphous phase may arise from excess La that has been expelled from the La 2/3 TiO 3 crystal structure. The same unidentified cubic phase that is present in LTO/STO(110) depositions ( Fig. 2(b)) can be clearly observed on the right side of Fig. 5(b). On the left side of the image are regions which we interpret as in-plane rotational domains of this same structure. In the center of the image is a triangular defect of another phase. Similar triangular defects were observed in other regions of the film. There appears to be a specific orientation relationship between the two phases at their interface, but identification of either phase is non-trivial and beyond the scope of this paper.
After annealing at 1000 °C for 8 h, the film thickness becomes highly non-uniform (not shown). Despite the annealing treatment, which was sufficient to promote this bulk transfer of material, the top portion of the film remains amorphous. The thickness of the amorphous region has decreased (~16 nm), and this thickness remains fairly uniform regardless of the variation in the overall film thickness. In the high-resolution lattice image presented in Fig. 5(c), the LTO lattice has transformed from the unidentified phases shown in Fig. 5(b); however, the lattice image does not exhibit the same pattern as the PLS lattice images in Figs 2(d) and 3(c). This provides another indication that the predominant phase is La 2/3 TiO 3 , not La 2 Ti 2 O 7 . Schematic diagrams of both the La 2/3 TiO 3 atomic positions in a multi-unit-cell slab of the cubic perovskite structure (for simplicity, a cubic unit cell, not the orthorhombic doubled cell 21 , is shown) and the LTO atomic positions in a six-unit-cell slab of the PLS structure are sketched in Fig. 5(d). The correlation between the schematic atomic positions and the lattice positions in the image confirms that the epitaxial phase is La 2/3 TiO 3 with the orthorhombic [102] direction perpendicular to the interface. In this crystalline orientation, the superstructure contrast arising from the ordered A-site vacancies is not visible, in contrast to STEM images of the [100] projection 23 Fig. 6(a) and (b), respectively. The piezoelectric coupling direction in the PLS structure of LTO is parallel to the additional oxygen planes, along the LTO[010] direction. From the XRD and STEM data above, LTO/STO(001) possesses this direction at an approximately 45° angle to the substrate surface. PFM measurements in Fig. 6(a) confirm that clear piezoelectric coupling is observed for LTO/Nb:STO(001). The piezoresponse exhibits similar magnitude and spatial structure in-plane and out-of-plane, consistent with a 45° piezoelectric coupling direction with components of similar magnitude in-plane and out-of-plane. A comparison of the top two images in Fig. 6(a) indicates that each film grain in the amplitude image (left) consists of two, opposing polarization directions separated by a domain wall (seen as a color change, indicating a phase reversal, in the right-hand image). In the plane of the film (bottom images), the grains each consist of a single polarization domain.
In contrast, no structure indicative of a polarization reversal is observed in the out-of-plane images for LTO/ Nb:STO(110), as seen in Fig. 6(b) (top images). This is expected for LTO deposited on STO(110) with the piezoelectric LTO[001] direction lying entirely in the film plane, with no out-of-plane component. Structure is observed in the lateral amplitude and in-plane phase images (Fig. 6(b), bottom), confirming an in-plane piezoelectric A strong out-of-plane piezoelectric response is expected for LTO/Nb:TiO 2 (110) if the piezoelectric LTO[010] direction is oriented out-of-plane. Instead, the dominant phase is epitaxial La 2/3 TiO 3 , an orthorhombic perovskite structure which is not piezoelectric. In Fig. 7, as expected, no strong piezoelectric signals are observed either out-of-plane or in-plane. Quantitative values of the piezoelectric coefficient (d 33 ) cannot be obtained by our PFM measurements. However, the strength of piezoelectric coupling can be qualitatively evaluated by comparing the maximum vertical PFM amplitude, normalized to the applied voltage, for each film. These values are 22, 2.2, and 1.7 pm/V for LTO films on STO(001), STO(110), and TiO 2 (110), respectively. Clearly, the LTO film on STO(001) shows an enhanced piezoelectric response compared to the films with other crystalline orientations. Epitaxial deposition of LTO[021] on STO(001) would be preferred for thin film piezoelectric devices. In addition, further optimization of the LTO deposition parameters to stabilize epitaxial LTO[010] on TiO 2 (110) (or another suitable substrate) is expected to promote even stronger out-of-plane piezoelectric coupling.
While La 2/3 TiO 3 may not be of interest as a piezoelectric material, it has been explored as a microwave dielectric with promising properties [23][24][25] . La 2/3 TiO 3 is also the parent compound of ionic 22 and lithium ion 26 conductors. The orthorhombic structure of La 2/3 TiO 3 , with ordered A-site cation vacancies, is difficult to stabilize in bulk form without the addition of dopant cations 23,25 . Epitaxial stabilization of undoped La 2/3 TiO 3 on TiO 2 (110) is a promising route to realize pure La 2/3 TiO 3 for both fundamental studies and device applications.
In summary, epitaxial thin films of the high-temperature piezoelectric material La 2 Ti 2 O 7 were deposited on STO(110), STO(001), and rutile TiO 2 (110) substrates by pulsed laser deposition. Reasonable control of the film orientation can be achieved when depositing LTO on substrates with various orientations, as confirmed by XRD patterns and STEM measurements. PFM measurements confirmed that LTO/STO(001) possesses strong piezoelectric coupling in a direction approximately 45° to the film plane; both in-plane and out-of-plane piezoelectric signals were observed. In the out-of-plane direction, each film grain consists of two domains with opposite polarization direction; in-plane, only one polarization direction is observed for each grain. For LTO films deposited on STO(110), only in-plane polarization is observed, as expected. The case of LTO on TiO 2 (110) is more complex. The film structure consists of several phases and orientations, and the dominant orientation is epitaxial La 2/3 TiO 3 (102). This phase is not piezoelectric, but is of interest as a microwave dielectric material and an ion conductor. The results presented here confirm that the strength of piezoelectric coupling can be enhanced in epitaxial LTO thin films, which may be of interest for high-temperature sensors and piezoelectric devices.
Crystalline properties. High-resolution XRD patterns were collected on a Philips X'Pert Materials Research Diffractometer (MRD) using Cu K α1 radiation monochromated with a hybrid mirror/4 crystal monochromator and fixed-slit detector optics. A Rigaku D/MAX RAPID II microdiffractometer with a curved imaging plate and a rotating Cr anode (Cr Kα = 2.2897 Å) operating at 35 kV and 25 mA was used to collect microbeam XRD (μXRD) and pole figure patterns. Cross-sectional STEM samples were fabricated using a focused ion beam (FIB) lift out technique with a FEI Helios microscope operating at 0.5-30 keV ion beam energy. High-angle annular dark field (STEM-HAADF) images were acquired using a JEOL ARM-200CF microscope operating at 200 keV with a 27.5 mrad convergence and 82.6 mrad inner collection semi-angles, respectively. Additional STEM-HAADF images were acquired using an FEI Titan 80-300 microscope operating at 300 keV.