Structure and piezo-ferroelectricity relationship study of (K0.5Na0.5)0.985La0.005NbO3 epitaxial films deposited on SrTiO3 by sputtering

This work demonstrates that the rf-sputtering technique, combined with appropriate heat treatments, is potentially effective to develop new materials and devices based on oxide-interface and strain engineering. We report a study of the structural-physical properties relationship of high crystalline quality, highly oriented and epitaxial thin films of the lead-free (K0.5Na0.5)0.985La0.005NbO3 (KNNLa) compound which were successfully deposited on Nb-doped SrTiO3 substrates, with orientations [100] (NSTO100) and [110] (NSTO110). The crystalline growth and the local ferroelectric and piezoelectric properties were evaluated by piezoresponse force microscopy combined with transmission electron microscopy and texture analysis by X-ray diffraction. Conditioned by the STO surface parameters, in the KNNLa films on NSTO100 coexist a commensurate [001]-tetragonal phase and two incommensurate [010]-monoclinic phases; while on NSTO110 the KNNLa films grew only in an incommensurate [101]-monoclinic phase. Both samples show excellent out-of-plane polarization switching patterns consistent with 180° domains walls; while for KNNLa/NSTO100 ferroelectric domains grow with the polarization pointing down, for KNNLa/NSTO110 they prefer to grow with the polarization pointing up. Comparing with previous reports on epitaxial KNN films, we find our samples to be of very high quality regarding their crystalline growth with highly ordered ferroelectric domains arrangements and, consequently, great potential for domain engineering.

they cannot explain satisfactorily the physical properties (polarizability, spontaneous polarization, domain structure, piezoelectric response, among others). The reason is that they do not correspond with the temperature dependent phase transitions, characteristic of the ferroelectric perovskite-type compounds, that occur from the monoclinic Pm, rhombohedral R3m or orthorhombic Pnma (or Pbnm) distortions at lower temperatures to the tetragonal P4mm distortion at intermediate temperatures and to the cubic Pm ¯m 3 structure at higher temperatures 11,14,16,17 . As we discuss below, all this takes special significance when KNN thin films are grown on monocrystalline surfaces which induce or constrain the film´s lattice parameters and preferential growth orientations leading to epitaxiallity 18,19 .
Another challenge is to bring the ferroelectric-ferroelectric transition temperature (T F-F ) of (K 0.5 Na 0.5 )NbO 3 closer to room temperature (RT) from the reported value of 200 °C 2,20 , to take advantage of the better properties of the tetragonal phase, and the coexistence of different phases. In this respect, many works have been reported about the effect of the dopants on the structural and physical properties of KNN; among them, Li doping has been extensively investigated as a representative modifier for this system 5,11,21 . In 2004, Saito et al. 2 reported textured ceramics of Li-, Ta-, and Sb-modified KNN with dielectric constant of ε ε = / 1570 T 33 0 , normalized strain of S max / E max = 750 (pm/V), and amazing values of the piezoelectric constant up to d 33 ~ 416 pC/N, comparable to that of soft PZT ceramics. Moreover, Hao et al. 20 showed that the crystalline structure of (Na 0.5 K 0.5 ) 1−3x La x Nb 0.95 Ta 0.05 O 3 ceramics changed from orthorhombic to pseudo-cubic perovskite phase with the increase of La-doping at room temperature.
In the meantime, driven by the miniaturization and integration tendencies 22,23 , great efforts have been made to fabricate high-quality KNN or KNN-based thin films using different growth techniques such as chemical solution deposition (CSD) via sol-gel procedure 3,13,18,19,[24][25][26] , pulsed laser deposition (PLD) 27,28 , and rf-magnetron sputtering [29][30][31] . Among them, rf-magnetron sputtering is a particularly attractive technique for applications because it allows large area deposition and is commonly used in actual microelectronic manufactures.
The aim of this work, is to study the structure-piezo-ferroelectric properties relationship of La doped KNN (KNNLa) thin films grown by the rf-magnetron sputtering technique using Nb doped SrTiO 3 (STO) single-crystal substrates with [100] and [110] crystallographic orientations. The ferroelectric and piezoelectric response of the highly oriented and epitaxial films thus obtained was investigated using piezo-force microscopy, and the strong correlation with the structural and morphological characterization is discussed.

Experimental
A (K 0.5 Na 0.5 ) 0.985 La 0.005 NbO 3 (KNNLa) ceramic target was synthetized by conventional solid-state reaction using K 2 CO 3 (99%), Na 2 CO 3 (99.5%), Nb 2 O 5 (99.9%), and La 2 O 3 (≥99.98%) powders as raw materials. The mixed and milled powders were calcined at 700 °C for 2 h, and milled again, pressed at 450 MPa (using PVA as binder), and sintered in air at 950 °C for 2 h. The ∼60 nm thick KNNLa thin films, were deposited by reactive rf-magnetron sputtering on Nb doped SrTiO 3 single-crystal substrates with [100] (NSTO100) and [110] (NSTO110) crystallographic orientations. The parameters for the deposit were: a base pressure of 2.0 × 10 −5 Torr, 5 cm target-substrate distance, 585 °C substrate temperature, and 175 Watts rf-power. The deposition time was 1 h and a 4:1 argon/oxygen partial pressure totaling 20 mTorr was used. To enhance the crystalline growth, NSTO substrates and KNNLa/NSTO samples were heat treated under an O 2 atmosphere at the annealing temperature of 585 °C, before and after the deposition processes. The crystallographic structure and orientation was examined by X-ray diffraction (XRD) using a Panalytical X-Pert Pro MRD diffractometer with monochromatic Cu-K α1 radiation (1.540598 Å). The cross-section and the crystal structure of the epitaxial film were analyzed using transmission electron microscopy (TEM) with a JEOL JEMF-2010 microscope. The cross-sectional specimens preparation were realized using a JEOL JIB-4500 scanning electron microscope equipped with focused ion beam technique (SEM + FIB) at room temperature. Structural simulation was carried out using the VESTA software (version 3.4) 32 . Surface topography, out-of-plane and in-plane ferroelectric domain structure and domain switching analysis were studied by piezoresponse force microscopy (PFM) using a Park Systems Launch XE7 atomic force microscope (AFM) with an SR865 lock-amplifier by Stanford Research Systems. Platinum top electrodes were deposited by rf-sputtering. The electromechanical resonance response was measured with an Agilent Precision LCR Meter E4980A source coupled to a two arms CPX-VF cryogenic probe station by LakeShore Cryotronics Inc.

Results and Discussion
In part A of this section, a combination of XRD, TEM, and AFM techniques and structural simulation are used to investigate the crystalline quality of the epitaxial KNNLa thin films grown on both NSTO substrates (NSTO100 and NSTO110) in regards to the crystalline phases, preferential growth orientations, and epitaxial matching; whereas the in-plane lattice parameters are determined by crystallography texture analysis. In part B, the ferroelectric and piezoelectric behavior of the highly ordered ferroelectric domains arrangements are studied by PFM analysis focusing on the structural characteristics.
A. Structural and morphological characterization. KNNLa on NSTO100. Figure 1a shows the XRD pattern of a KNNLa thin film grown on a NSTO100 substrate (labeled K/S1). The strongest peaks correspond to the (100) C , (200) C , and (300) C planes of the substrate's cubic perosvkite structure. The peaks at 2θ = 21.87°, 45.01° and 70.20°, belong to the pure KNNLa thin film and correspond to a strong out-of-plane preferential growth orientation induced by the STO substrate and its in-plane lattice parameters at the surface. It is well known from the literature that KNN has a perovskite-like monoclinic structure, stable at room temperature (RT) 3,11,14,33 ; however, the asymmetric broadening of the peak at 2θ = 70.20° suggests the coexistence of single symmetry phase (monoclinic, orthorhombic, or tetragonal) with different lattice parameters and/or different orientations [such as (h00), (0l0), and (00l)] or more than one phase at RT. The inset of Fig. 1a shows the fitting results, where three peaks centered at 2θ = 70.20°, 70.56°, and 71.59° may be deconvoluted and only the third peak at 71.59° is affected Scientific RepORTS | (2017) 7:17721 | DOI:10.1038/s41598-017-17767-3 by the overlapping with the corresponding peak of the substrate. Consequently, as expected for the (300) C peak (centered at 2θ = 72.55°), it is found that the substrate lattice parameter is a STO = 3.9060 Å.
On the other hand, Fig. 2a shows the AFM topography image of the as-grown K/S1 sample. Parallel and perpendicular rod-shape grains with different sizes, forming a maze-like mesh where the rods are oriented at ~45° with respect to the [001] and [010] directions of the STO substrate, are observed. Such arrangement, different to the herringbone domain patterns reported for KNN ceramics 9,12 , may be associated to separated orthogonal, tetragonal or orthorhombic phases, growing locally epitaxial as will be discused below. To date, similar crystalline growth has not been reported for KNN thin films even for epitaxial growth claimed in different previous works where KNN films were obtained on STO substrates using chemical solution deposition via sol-gel technique 3,19 . In all previous reports, as-grown KNN films show granular morphology with domains distributed randomly in-plane, without developing highly ordered crystalline arrangements as observed in Fig. 2.
Moreover, in Fig. 1b, a representative low-resolution TEM cross section micrograph of the K/S1 sample (part A) shows a relaxed KNNLa thin film with an average thickness of 60 nm and a flat and continuous KNNLa/STO interface connection. The high-resolution TEM image (part B) of the KNNLa/STO interface illustrates an epitaxial growth of the KNNLa film following the STO orientation; meanwhile, the Fourier transform of the same region (part C) shows an excellent coupling between the in-plane lattice parameters of the KNNLa film and STO substrate with minimal mismatch.
With all those arguments and evidence, and according to the atomic arrangement of the STO's (100) surface ( Fig. 3), the three peaks describing the peak profile at 2θ = 70.20° (Fig. 1a) can be associated to the possible coexistence of monoclinic phases found in the vicinity of the MPB or polymorphic regions for x = 0.5 5 or to the coexistence of such monoclinic phases and a tetragonal one 9,[11][12][13]15 . We consider that the coexistence of tetragonal and monoclinic phases is the best picture to explain the observed results.
Starting with the tetragonal phase, it is assumed to exist at room temperature as result of a strong lattice constriction by compressive stress generated by the STO surface, promoting a strained-epitaxial-like growth; this, in combination with the fact that La doping decreases the transition temperature T F-F to around 200 °C 2,20 . However, since for the doped KNN compound the lattice parameters of the reported tetragonal phases are larger than those of the monoclinic phases 4,11,15 , we have assumed here that the peak maxima at 2θ = 21.87°, 45.01° and the first deconvoluted peak at 70.20° correspond mainly to the tetragonal contribution. From the XRD pattern, using the 2θ values at the maxima of such peaks, where the overlapping is minimum, we can obtain the crystalline parameters with higher precision. Thus, the interplanar spacing for 2θ = 70.20°, corresponding to crystalline planes of the KNNLa film parallel to the surface is d 1 = 1.3396 Å. Taking this into account, Fig. 3a shows the model of an epitaxial commensurate growth of a KNNLa film on NSTO100 with tetragonal structure and preferential orientation along [001] assuming that the cell parameters a T = b T have values close to those of the substrate (3.906 Å) and c T = 3d 1 = 4.0188 Å obtained from XRD pattern. Therefore, the peaks at 21.87°, 45.01° and 70.20° in the pattern of Fig. 1a can be indexed for the tetragonal phase as (001) T , (002) T , and (003) T , respectively.
As a second possibility, the square array of the STO surface permits to consider an incommensurate growth of the KNN monoclinic primary cell (see Fig. 3b) as has been reported previously 3,19 . In the literature, it has been reported that the monoclinic cell is a special case where two of the parameters are nearly equal (a M ≈ c M ) forming an angle β (>90°) and the third one, b M , is shorter and perpendicular to a M and c M 4,11,14,15 . Therefore, the peaks at 2θ = 70.56° and 71.59° (Fig. 1a), corresponding to interplanar spacing of d 2 = 1.3337 Å and d 3 = 1.3170 Å, can  19 .
To obtain further information of the in-plane structure (lattice parameters, orientations, and epitaxial quality) of the KNNLa films, crystallography texture analysis was carried out. The out-of-plane parameters obtained from XRD patterns and the in-plane parameters taken from those reported in the literature for tetragonal and monoclinic phases in correspondence with the STO parameters, were used as starting values. Thus, measurements of pole figures (PFs) for selected (hkl) crystalline planes and ψ-polar and Φ-azimuthal angle scans were realized to obtain the optimal values of 2θ hkl and ψ hkl . Table 1 summarizes the obtained results for those PFs used to calculate the lattice parameters of each crystalline phase. The analysis for the M2 phase was not considered because of the higher overlapping with the substrate.
The results demonstrate the coexistence of the tetragonal and monoclinic phases in the KNNLa film of the K/ S1 sample. The PFs in Fig. 1c show the expected 4-fold symmetry for the STO and the tetragonal phase present in the KNNLa film. Moreover, due to overlapping between the diffraction peaks of the tetragonal and monoclinic phases, the PF of the tetragonal phase exhibits the presence of the expected 2-fold symmetry of the monoclinic phase, and vice versa, as can be seen from the maxima in the first and third quadrants. Furthermore, the Φ-scans in Fig. 1d confirm the strong in-plane preferential orientation induced by the STO substrate and the high quality of the epitaxial growth of the KNNLa thin film.  35 .
With all this into consideration, the coexistence of tetragonal and monoclinic phases can explain the maze-like topography (Fig. 2a) as is illustrated in the simulation of Fig. 3c. It can also be seen in Fig. 3c, due to the square array of the STO surface, the orthorhombic (monoclinic) growth is possible in the four diagonals forcing the tetragonal growth to follow two directions 90° apart.
KNNLa on NSTO110. Figure 1e shows the X-ray diffraction pattern of the KNNLa thin film grown on an NSTO110 substrate (labeled K/S11). The high intensity peaks at 2θ = 32.49° y 2θ = 67.81° correspond to the (110) C and (220) C planes of STO, respectively; while, the peaks at 2θ = 32.01° y 2θ = 66.77° correspond to the KNNLa film indicating high crystallinity and a strong preferential orientation conditioned by the [110] orientation of the substrate and its in-plane lattice parameters at the surface. From the (220) C peak position a value of a STO = 3.9058 Å was obtained for the STO in good correspondence with those of NSTO100; thus, the NSTO110 surface is characterized by a rectangular array 3.9058 Å × 5.5236 Å as shown in the simulation in Fig. 4a. Meanwhile, the interplanar spacing obtained from the KNNLa peak at 2θ = 66.77° is d 4 = 1.3999 Å. Additionally, Fig. 2b shows the AFM topography image of the as-grown K/S11 sample, where rod-shape grains grow in-plane on the substrate surface, mutually parallel along the [001] direction and perpendicular to the [1ī0] directions of the STO substrate.
In this case, we could also think of a (101)-oriented monoclinic phase coexisting with a (110)-oriented tetragonal phase, both of them following the (110) orientation of the substrate as is illustrated in Fig. 4b. Using the parameter value 2d 4 = 2.7998 Å corresponding to planes (101) M or (110) T (consistent with the peak at 2θ = 32.01°) the value of 3.9595 Å is obtained for the lattice parameters a M ≈ c M and a T = b T for the monoclinic and tetragonal phases, respectively. In the light of these results, we can consider that the KNNLa film grows incommensurate for both phases; however, attending to the strong lattice constriction that the NSTO110 surface imposes on the third parameter (b M or c T ) to be coupled to its own lattice parameter (a STO = 3.9058 Å) by compressive stress, the presence of the tetragonal phase is discarded since the values reported for c T are higher than 4.001 Å 4 as is expected in perovskite like compounds.
Thus, using the experimental value of d 101 = 2.7998 Å and the assumption of a M ≈ c M = 3.9595 Å as starting model for the monoclinic phase, measurements of PFs for selected (hkl) crystalline planes (corresponding to high values of 2θ to minimize the overlapping between the diffraction from the KNNLa film and the STO substrate), and ψand Φ-scans were realized to obtain the optimal values of 2θ hkl and ψ hkl . Table 1 summarizes the obtained results. The PFs in Fig. 1f show the expected 2-fold and mirror (2 m) symmetry for the STO and the mirror-symmetry expected for the monoclinic KNNLa film. Meanwhile, the Φ-scans in Fig. 1g  symmetries and the strong in-plane preferential orientation induced by the STO substrate and the high quality of the epitaxial growth of the KNNLa thin film. As can be seen in Table 1, the calculated lattice parameter b M = 3.9114 Å is very close to the calculated a STO = 3.9056 Å and the calculated d 101 = 2.8145 Å is consistent with those obtained from the XRD pattern; meanwhile, the rectangular array 3.9114 Å × 5.6386 Å for KNNLa film establishes a mismatch along the [ī01] direction of the STO substrate (Fig. 4a), justifying the assumption made above. Additionally, all computed lattice parameters (a M = 3.948 Å, b M = 3.9114 Å, c M = 4.0197 Å, and β = 90.09°) are in good correspondence with those reported for the monoclinic (orthorhombic) phase of KNN 3,8,36,37 . With this in mind, the peaks in the pattern of B. Ferroeletric and piezoelectric response. PFM in the resonance mode was used to study the ferroelectric and piezoelectric properties. To investigate the domain structure, the out-of-plane polarization switching patterns were explored in a DC regime working at optimal applied voltages on K/S1 (K/S11) sample as follows. In a total area of 12 × 12 µm 2 of the as-grown sample, an initial poling was carried out by applying a −30 V (+5 V) bias voltage in a concentric 8 × 8 µm 2 inner area, followed by applying +15 V (−5 V) in a concentric 4 × 4 µm 2 area enclosed within the previous one. After poling, topography, amplitude, and phase PFM images were simultaneously taken of the full 12 × 12 µm 2 area. In this work, all measurements were realized with an AC voltage signal of 1 V pp , at a frequency value near below the maximum resonance in both samples, just where the phase began to change in the electromechanical spectrum. The results are shown in Fig. 5 for K/S1 and Fig. 6 for K/S11. In the topography images of K/S1 (Fig. 5a) and K/S11 (Fig. 6a) no damage was observed as result of the applied electric fields, indicating that the collected PFM signals were coming only from the KNN layers piezoresponse. Moreover, to distinguish if the switching processes, hysteresis loops, and electromechanical properties are originated from the spontaneous polarization, the first/second harmonic criterion was used 38,39 using AC excitation. In Figs 5b and 6b, the amplitude of the first harmonic is notably higher than that of the second harmonic for both K/S1 and K/S11 samples, respectively, as is expected for good FE materials 39 .
As can be seen in Figs 5c and 6c, the concentric square patterns in the amplitude piezoresponse for both samples show well defined perimeter borders, separating regions with similar expansion and contraction amplitude values, associated to FE domain walls (DW); while, the corresponding concentric patterns in the phase piezoresponse (Figs 5d and 6d) show well defined bright and dark contrast corresponding to up and down FE domains, respectively; i.e., domains with opposite polarization switched along the out-of-plane orientation, characteristic of 180° DWs 9,40 . However, there are some differences between the samples' behaviors.
Higher domain switching voltages (−30 V and +15 V) were required for the K/S1 sample (Fig. 5) probably due to the coexistence of tetragonal and monoclinic phases where the spontaneous polarization lies along the directions [001] parallel and [101] 90° away from the out-of-plane direction, respectively (Fig. 3) 9,12,13,15 . The influence of the coexistence of different crystalline phases on the coercive electric field and the piezoelectric response have been reported for KNN and PZT systems 8,11,16,17,41 . Besides, only half of the voltage (+15 V) was required to reverse the polarization in the innermost area where the phase values are similar to those of the external area, as indication that the FE domains grow with the polarization pointing down in good agreement with the preferential growth orientations discussed above.
Meanwhile, lower and symmetric domain switching voltages (+5 V and −5 V) were required for the K/S11 sample (Fig. 6) since only (101)-oriented monoclinic crystallites are aligned parallel to the same in-plane substrate direction; where the spontaneous polarization lies along the [101] direction of the monoclinic cell, parallel to the out-of-plane direction (Fig. 4b). Moreover, switching is relatively easy for the FE domains grown with the polarization pointing up in the external area as can be observed in Fig. 6. It is worth noting that in-plane PFM measurements were realized in both samples on regions as grown, but no contrast was obtained. Figure 6e shows the in-plane (left) and out-of-plane (right) phase PFM images obtained after poling with switching voltages of +6 V and −6 V applied on concentric square areas of the K/S11 sample. Only the out-of-plane phase image shows clearly two areas with opposite polarization, similar to Fig. 6d. In the in-plane image no contrast was observed. The piezoresponse signal is very low, compared to those of Fig. 6d,   Figure 6. Out-of-plane PFM images, for the K/S11 sample, after poling processes of (a) topography, (c) amplitude, and (d) phase (the bias voltages are indicated). (b) First and second harmonic spectra. (e) In-plane (left) and out-of-plane (right) phase PFM images, after poling, using a test frequency far away below the first harmonic resonant frequency.
Additionally, local polarization hysteresis loops obtained through the phase vs DC voltage curves and local strain loops through the displacement amplitude vs DC voltage curves (butterfly loops) were recorded to obtain quantitative information of the FE domain structure and its electromechanical properties. Such measurements were done with a pulsed triangular DC voltage signal in the ON and OFF field modes; however, as in the ON field mode the loops contain electromechanical and electrostatic effects induced by the capacitive force between the cantilever and the sample surface, only the OFF field mode response was used to minimize such effects 42 . In both samples, a positive bias voltage was applied first. Figure 7a and b show the phase and butterfly loops for K/S1 and K/S11, respectively. The phase loops are square, typical of 180° DWs, and in both cases the loops are shifted from the origin but in opposite directions, corresponding with the preferential orientation of the polarization, i.e., down for K/S1 and up for K/S11 as was observed before. Additionally, both samples show low average coercive fields of ~3.7 V for K/S1 and ~2.8 V for K/ S11. Meanwhile, according with the phase loops, the samples exhibit asymmetric butterfly loops and the values of the piezoelectric constant d 33 = 29 pm/V for K/S1 and d 33 = 19 pm/V for K/S11 were calculated.
Finally, Fig. 7c shows the global piezoelectric response of the K/S1 sample, obtained through the small-signal impedance modulus spectrum as function of frequency near the electromechanical resonance response. As can Figure 7. Phase-voltage and amplitude-voltage loops of (a) K/S1 and (b) K/S11 samples obtained in OFF field mode. (c) Impedance modulus spectrum of K/S1 sample.
Scientific RepORTS | (2017) 7:17721 | DOI:10.1038/s41598-017-17767-3 be seen, the values of the resonance frequency f r , 396 kHz, and anti-resonance frequency f a , 397 kHz, are in good agreement with the values of the first harmonic in Fig. 5d used in the PFM characterization.

Conclusions
High quality and highly oriented thin films of the lead-free ferroelectric La-doped (K 0.5 Na 0.5 )NbO 3 (KNNLa) compound were successfully deposited on SrTiO 3 :Nb[100] and SrTiO 3 :Nb[110] substrates by rf-sputtering. The KNNLa films on [100]-substrate is characterized by the coexistence of a commensurate [001]-oriented tetragonal phase and two incommensurate (010)-oriented monoclinic phases, while on [110]-substrate they grow only in an incommensurate (101)-oriented monoclinic phase, in both cases conditioned by the lattice parameters of the substrates.
The KNNLa films exhibit out-of-plane polarization switching patterns of up and down ferroelectric domains corresponding to 180° domain walls. The as-grown KNNLa films on [100]-substrates show pointing down polarization and higher coercivity values depending on the spontaneous polarization direction in each of the coexisting phases. For as-grown KNNLa films on [110]-substrates the domains grow with the polarization pointing up with lower coercivity. For both kinds of samples, the obtained piezoelectric constant values are comparable with those in other reports on KNN films. With such ferroelectric and piezoelectric properties, KNNLa films grown by rf-sputtering are potential candidates for applications on ferroelectric lead-free films based devices.