Extraordinarily Large Contribution Ratio of Ferroelastic Domain Switching to Piezoresponse in Monoclinic (K, Na)NbO3 Films

Ferroelectric monoclinic phases have attracted exceptional attention as the origin of giant piezoelectricity, whilst the detailed contributions of ferroelastic domain switching and electric‐field induced lattice strain to the piezoelectric response remain still challenging to clarify. In this work, these contributions to the piezoelectric response are deconvoluted in a (K0.4Na0.6)NbO3 (KNN) film epitaxially grown in monoclinic phase on Nb‐doped SrTiO3, where the as‐deposited film feature (111) and ( 1¯11$\bar{1}11$ ) non‐180° domains. By time‐resolved synchrotron X‐ray diffraction, the ferroelastic domain switching and electric‐field induced lattice strain subjected to an ultrafast electric‐field pulse are quantitatively probed. The switching of ≈4% volume fraction of (111) domains into ( 1¯11$\bar{1}11$ ) ones by a moderate electric field and its response within 30 µs as well are unambiguously unveiled. Interestingly, the contribution of domain switching to the strain is larger than the total strain of the film, which is enabled because of the negative electric‐field‐induced lattice strain. The present study connects macroscopic piezoelectric response in KNN films with the underlying microscopic origins unveiled by separating two contributions, which may provide a knowledge platform that allows for significant achievement of practical lead‐free piezoelectric microelectromechanical and nanoelectromechanical systems in the future.


Introduction
[3][4][5][6][7][8][9] The piezoelectric response of ferroelectrics is stemmed from both intrinsic and extrinsic mechanisms.[17] It is widely known for PZT thin films that the latter contributes far more to piezoelectric response than the former. [18,19]Therefore, non-180°ferroelastic domain switching is critical for boosting piezoelectric response and developing environment-friendly alternatives to the prototype PZT.
Since the breakthrough by Saito et al., (K, Na)NbO 3 (KNN)based solid solutions have been one of the most intriguing lead-free piezoelectric candidates, in view of whose closest piezoelectric characteristics comparable to those of PZT. [20]xtensive efforts are devoted to elucidating intricate domain patterns such as checkerboard, stripe, and herringbone-like patterns, [21][22][23] that arise from the competition between electrostatic and elastic energy in pure KNN and KNN-based films.[26] In addition, various advanced techniques such as piezo-response force microscopy (PFM), transmission electron microscopy, and optical means were employed to reveal the domain switching, yielding intriguing and novel results.However, the majority of them are ex situ or semi-in situ methods, accompanied by time resolution limitations, [27][28][29] which impedes the insight into the domain switching dynamics.Here, it is worthwhile to mention that the low symmetry monoclinic phase in several lead-based systems manifesting large piezoelectric response was verified by first-principles calculations and Xray diffraction (XRD). [30]On the other hand, to date, experimental studies on the domain switching in monoclinic KNN thin films have not been fully explored yet and remain to be clarified to achieve a high-performance piezoelectric response.Thus, epitaxial monoclinic KNN films could provide an unambiguous insight into the domain switching dynamics for high-performance piezoelectric response.33][34] In this work, the detailed contributions of ferroelastic domain switching and electric-field induced lattice strain to the piezoelectric response were simultaneously revealed in pure (K 0.4 Na 0.6 )NbO 3 film epitaxially grown on Nb-doped SrTiO 3 (Nb:STO).The (K 0.4 Na 0.6 )NbO 3 film in monoclinic phase where featuring (111) and ( 111) non-180°domains were proven by XRD reciprocal space mapping (RSM) in conjunction with PFM.By time-resolved synchrotron XRD, the ferroelastic domain switching and electric-field induced lattice strain subjected to fast electric-field pulses were quantitatively probed.It was found that the moderate bias enhanced the extrinsic contribution over the total strain of the film due to the appearance of negative intrinsic contribution, whereas the intrinsic response becomes positive at a larger bias.Both time for inducing the domain switching by bias applications and relaxing back to the original state after removal of the bias was only ≈30 μs.

Structural Property and Phase Transition
Crystallinity and epitaxial growth of the KNN film were determined by XRD. Figure 1a displays the XRD 2/ scan of the 850 nm-thick KNN film on Nb:STO (111), where only (hhh) pc (h = 1, 2) diffraction peaks indexed to the KNN film were observed, indicating the perfect (111) pc orientation of the KNN film with the absence of secondary phases.The XRD ϕ scans for KNN (110) pc and Nb:STO (110) were performed, as shown in Figure 1b.The same in-plane orientation of KNN film with that of the Nb:STO substrate was confirmed by 3-fold symmetry with an equal 120°i nterval, demonstrating that the KNN film was epitaxially grown and has the cube-on-cube relationship with the substrate; namely, [ 110] pc KNN (111) pc || [ 110] Nb:STO (111).
To inspect the detailed crystal and domain structures of the as-deposited KNN films, temperature-dependent XRD RSMs for both the symmetric and asymmetric planes were carried out.Figure 1c-e,f-h show the evolution of RSMs around Nb:STO 111 and 112 with incremental temperature, respectively.At room temperature, the film has two diffraction peaks around Nb:STO 111 that correspond to the ( 111) and (111) domains as shown in Figure 1c, and similar but much broader two peaks are observed around Nb:STO 112 (Figure 1f).The latter broader two peaks can be due to the two sets of similar lattice spacings for 121 and 121, and for 211, 112, 211, and 112 in the monoclinic phase.Here note that the monoclinic phase has comparable energy with the orthorhombic phase, albeit the more energy-stable monoclinic structure (M c ) at room temperature was confirmed. [35]At 300 °C, the 111 and 111 peaks merged together (Figure 1d), and the two peaks around Nb:STO 112 become clearer and can be assigned to 211 and 112 (Figure 1g), indicating the tetragonal phase.At 500 °C, the 211 and 112 peaks merged almost together (Figure 1h), corresponding to the cubic phase.The tetragonal and cubic phases in the KNN film at high temperatures agree with those in KNN bulks of the same chemical composition.
To accurately clarify the phase at room temperature, the structural characterization of 2 μm-thick KNN film deposited on Nb:STO (001) substrate under the same deposition condition as that on Nb:STO (111) substrate was also performed.The typical XRD 2/ scans, and ϕ scans for KNN (110) pc and Nb:STO (110) shown in Figure S1a  corresponding to monoclinic M c symmetry (Figure S2, Supporting Information).It is worth mentioning that similar monoclinic (001) pc KNN films were also reported. [24,26]The stabilization into the monoclinic phase is also supported by the fact that its chemical composition of (K 0.4 Na 0.6 )NbO 3 exhibits the monoclinic phase in bulk KNN at room temperature. [36]Furthermore, as the temperature increased, the crystal symmetry transformed from monoclinic to tetragonal and cubic, in accordance with that of bulk KNN (Figure S1c-e,f-h,Supporting Information).Here, we recall that the broader two peaks in RSM around Nb:STO 112 for the (111) pc KNN film at room temperature (Figure 1f) can be also reasonably explained by the monoclinic phase as discussed above; thus, it can be concluded that both of the (001) pc and (111) pc KNN films are in the monoclinic phase.
The set of structural analyses provided above lets us know the polarization directions of the (111) pc and (001) pc -oriented KNN films as depicted in Figure 1i and Figure S1i (Supporting Information), respectively.In the case of (001) pc -oriented KNN film, there are four variants with different in-plane polarization directions (Figure S1i, (Supporting Information).However, all the variants have the same out-of-plane polarization component; they are equivalent in terms of piezoelectric response d 33 when the electric field is applied along with the film thickness, and there will be no domain switching induced by the electric field.On the other hand, the (111) pc -oriented KNN film has nonequivalent domains; the ( 111) and (111) domains have different polarization components along not only the in-plane direction but also the out-of-plane directions (Figure 1i).Therefore, these are nonequivalent in terms of the piezoelectric response d 33 , and domain switching by the electric field is possible.

Ferroelectric Property and Domain Structure
The polarization-electric (P-E) hysteresis loops for the KNN films were recorded at a maximum electric field of 300 kV cm −1 and a frequency of 10 kHz, as shown in Figure 2a and Figure S3 (Supporting Information).These films, as can be seen, possess well-developed P-E hysteresis loops with negligible leakage current.For the (001) pc KNN film (Figure S3, Supporting Information), the remnant polarization, P r , and saturation polarization, P sat , were 13.5 and 15.7 μC cm −2 , respectively.P r and P sat are comparable to those reported for single crystals with analogous compositions listed in Table S1 (Supporting Information), proving that the XRD observed complete c domain of M c phase.The (111) pc KNN film demonstrated less square hysteresis (Figure 2a), where P r and P sat were 2.6 and 6.3 μC cm −2 , respectively.Based on the polarization axis in the M c phase, P r and P sat for the (001) pc KNN film, and the volume fraction of the ( 111) domain in the (111) pc KNN film, the P r and P sat for the present (111) pc KNN film can be roughly predicted to be 8 and 9 μC cm −2 .The observed P r was somewhat smaller than the expectation, but P sat was not far from it.
Prior to the piezoelectric property, it is comparatively significant to investigate the domain structure of the (111) pc KNN film having a possibility of domain switching.As described above, it is worthy to note that there are (111) and ( 111) domains and the polarization components of nonequivalent ( 111) and ( 111) domains are distinct along not only the out-of-plane but also the in-plane directions, as schematically shown in Figure 1i.Namely, there are 6 in-plane variants for (111) domains and 3 in-plane variants for ( 111) domains.Figure 2b,c shows the vertical and lateral PFM images of the (111) pc KNN film after poling treatment of 25 V and the corresponding component histogram of these images.Here, Acos(′) is so-called mixed response combining the PFM amplitude A and phase ′; in the present study, the latter was adjusted so that the phase at the resonance frequency is 90°to correct the slight offset of the measured frequency dependence of the phase.As can be found in Figure 2b, Acos(′) values in the vertical PFM image are all positive, indicating the poled state of the film.On the other hand, there are two different regions labeled as V+1 and V+2, indicating the coexistence of ( 111) and ( 111) domains with different out-of-polarization components.The fraction estimated from the histogram was 44%:56%, which is approximately  111) and ( 111) domains, though the difference in the in-plane polarization components was not clearly visualized within the accuracy of our PFM measurements.Although further investigation is needed to clarify the detailed domain patterns, the present PFM study showed the domain structure having two different out-of-plane polarization components, which agrees with our XRD study.

Domain Switching Dynamics
Figure 3a shows the schematic configuration of the in situ timeresolved synchrotron XRD measurement.The pulsed voltages were applied to clarify the effect of the electric field on the domain structure.The films were poled before the measurements, and the unipolar pulse voltage of the time width of 50 μs with the same direction as the poling bias was repeatedly applied with the time interval of 1000 μs at each diffraction angle.The time width of 50 μs for pulse voltage was selected from the viewpoint of the saturation time of the domain switching as discussed below.By integrating the photon intensity during the measurement, the synchrotron XRD spectra with and without bias were constructed. [37,38]The synchrotron XRD spectra of the (111) pc KNN film are shown in Figure 3b.As aforementioned, ( 222) and (222) peaks correspond to ( 111) and (111) domains, The of these two peaks barely alter with increasing the bias the bias is ≤ 36 kV cm −1 , but their intensity ratio varies prominently: ( 222) increases, (222) decreases.Subsequently, the peak positions slightly shift toward lower angles  111) pc KNN film with that of state-of-the-art lead-based and lead-free materials.PPT and MPB represent the polymorphic phase transition and morphotropic phase boundary, respectively.[45][46][47][48][49][50][51][52][53] upon the bias is ≥ 48 kV cm −1 .The variation in the volume fractions of the domains is determined by the alteration in the peak intensity ratio.Figure 3c shows the volume fraction of the ( 111) domains, V ( 111) , which can be expressed as follows: where I ( 222) and I (222) refer to the intensities of ( 222) and (222) peaks.As displayed in the Figure 3c, V ( 111) was increased nonlinearly and saturated at a 4% larger value till to the electric field ≥ 48 kV cm −1 , where the shift of the peak position evolved.The increment of V ( 111) denotes that the part of the (111) domain was reconfigured to the ( 111) domain possessing the larger out-ofplane polarization component; namely, the electric-field-driven domain switching.The small volume fraction change of 4% in the present epitaxial (111) pc KNN film would be due to the substrate clamping, by which the large tensile stress is induced in the film when the domain switching takes place.At the same time, the substrate clamping may provide the driving force to get back to the original domain state when the electric field is removed.
Taking the volume fractions and field-induced strain of ( and (111) domains into account, total strain, S, can be quantitatively determined by:  2) can be simplified as Equation (3). Figure 3d plots the field-induced strain in view of the intrinsic and extrinsic contributions, and the total strain.As observed, when the bias is ≤ 36 kV cm −1 , the strain for both ( 111) and ( 111) domains from intrinsic contribution were nearly zero, even slightly negative.On the contrary, the strain arising from the extrinsic contribution, i.e., domain switching, was large and increased with increasing the bias.Extrinsic contribution-driven strain was saturated above 36 kV cm −1 because of the substrate clamping effect, but intrinsic contribution increased afterward.Consequently, the intrinsic and extrinsic piezoelectric responses are highly dependent on the applied electric field; the moderate bias enhances the extrinsic contribution via domain switching, whereas the intrinsic contribution emerges only under the large bias after the saturation of domain switching.When the extrinsic contribution is dominant, the domain switches from (111) to ( 111) involving the large polarization rotation induces the tensile stress in the entire film as the in-plane lattice constant of ( 111) domain is smaller than that of (111) domain.As a consequence, the strain from the intrinsic contribution is suppressed to be nearly zero or below.However, the domain switching reaches the limitation at a certain bias due to the large elastic energy accumulated by the domain switching; thus, the intrinsic positive contribution becomes noticeable under the large bias.Although the strains induced by the intrinsic and extrinsic piezoelectric responses are nonlinear, the overall field-induced strain increases monotonously with the electric field.On average, the effective piezoelectric parameter d 33,f of the KNN (111) pc was 50.2 pm V −1 , which is comparable to that macroscopic piezoelectric constant (51.8 pm V −1 ) measured by DBLI (Figure S4, Supporting Information).It should be emphasized here that just a small fraction of domain switching significantly enhances d 33,f in the current film.Because of the slightly negative intrinsic response under the moderate bias, the extrinsic contribution ratio to overall d 33,f surpasses 100% as shown in Figure 3e.
[41][42][43] Figure 4 shows the comparison of the extrinsic contribution in the present (111) pc KNN film with that of state-of-the-art lead-based and lead-free ferroelectric materials.It was found that the extrinsic contribution in the (111) pc KNN film is far superior to other lead-based and lead-free single crystals, ceramics as well as thin films, except for the most recent lead-based PZT thin films.This is an interesting finding since such a large extrinsic contribution induced by a small domain switching has never been reported for lead-free materials before.Especially, the large extrinsic contribution is usually found in KNN-based systems with the composition in the vicinity of polymorphic phase boundary providing an easy polarization rotation path. [46,54]The results of the present work may give a different strategy apart from the phase boundary engineering to further enhance the piezoelectric performance.

Domain Switching Speed
Figure 5a shows the time evolutions of the ( 222) and (222) diffraction peak profiles, and corresponding time dependencies of the volume fraction of the ( 111) domain under the electric fields of 36 kV cm −1 .In accordance with Figure 3b, the intensities of ( 222) and (222) peaks respectively increase and decrease under the field.Moreover, the peak intensities recover back to initial values after the removal of the field, demonstrating that the domain switching is fully reversible.Confirmed from the time-dependent ΔV ( 111) in Figure 5b, this domain switching saturated during the interval of the voltage pulse of 50 μs.Both the times required to induce the domain switching by the field and relax back to the original state after withdrawing the field were ≈30 μs, indicating that the extrinsic piezoelectric response can follow at the frequency of 30 kHz (Figure 5b).They are substantially longer than those (25 and 95 ns) of (111) pc -epitaxial Pb(Zr 0.65 Ti 0.35 )O 3 film with rhombohedral phase from our previous report. [15]The physical explanations to pinpoint elaborately why the major domains are unswitchable and the slow domain switching speed are not clear yet.Nonetheless, there are two probable explanations proposed here: one is the difference of polarization axis.Un- like the rhombohedral (111) pc PZT film, [15] the domain switching from (111) to ( 111) in the (111) pc KNN film involves the rotation of the polarization from the axes nearly lying in the in-plane as shown in Figure 5c, which would require a much larger bias or a longer time when the bias is minimal. [55]Indeed, the applied electric field for the rhombohedral (111) pc PZT films reported by Ehara et al. was 1000 kV cm −1 , [15] which is almost 30 times for the present KNN films.Another one is that the reported (111) pc PZT film is 200 nm thick, but the present (111) pc KNN film is 850 nm thick.Thinner films would respond faster than thicker films because domain switching takes place with the domain wall motion. [8,56,57]As a consequence, a comprehensive investigation is further needed to understand the more detailed domain switching dynamics of KNN films in the future.

Conclusion
In summary, this work explicitly deconvolutes the contributions of ferroelastic domain switching and electric-field induced lattice strain to the piezo-response by quantitatively probing the piezoelectric dynamics in (111)-epitaxial (K 0.4 Na 0.6 )NbO 3 film under electric-field pulses.The (K 0.4 Na 0.6 )NbO 3 film featuring (111) and ( 111) non-180°domains in the monoclinic phase, were proven by XRD RSMs in conjunction with PFM.Time-resolved synchrotron XRD reveals that the extrinsic contribution to strain by 4 vol% domain switching under the moderate bias exceeded the total strain due to the negative intrinsic contribution, whereas the intrinsic response became positive under large bias.Meanwhile, both times for inducing the domain switching by applied bias and relaxing back to the original state after withdrawing the bias were only ≈30 μs.Intriguingly, the extrinsic contribution ratio in the (111) pc KNN film is far superior to the state-of-the-art lead-free ferroelectric materials.Our findings provide a deep understanding of the underlying mechanism and would allow for fine-tuning and significant improvement of the piezoelectric performance, eventually hastening to replacement of the lead-based thin films in piezoelectric MEMS and NEMS devices.

Experimental Section
KNN Films Pulsed laser deposition with a KrF excimer laser ( = 248 nm) was used to grow KNN films on 0.5 wt.% Nb-doped SrTiO 3 (001) and (111) substrates.The films were deposited at 673 °C under 1 Torr oxygen pressure.The laser energy and repetition rate were 95 mJ and 7 Hz, respectively.The KNN targets with K:Na = 1:1 fabricated by a solid-state reaction from K 2 CO 3, Na 2 CO 3, and Nb 2 O 5 were used.The thickness of the deposited films on the (001) and (111) substrates was 2 μm and 850 nm, and their chemical composition was determined to be around (K 0.4 Na 0.6 )NbO 3 .Pt top electrodes with diameters of 100 μm and 200 μm were deposited using electron beam evaporation on the surface of KNN films via a shadow mask.
Structural Characterizations: XRD using Cu K  1 X-rays (Bruker, D8 DIS-COVER) was used to determine the crystallographic structure and orientation of the films.A field emission scanning electron microscope (FESEM) equipped with energy-dispersive X-ray spectroscopy (EDX) was used to examine the cross-section and chemical composition of the films (Hitachi, S-4800).
Electrical and Electromechanical Characterizations: A ferroelectric tester (Toyo, FCE-1) and a double-beam laser interferometer (aixACCT Systems, aixDBLI) were used to record room-temperature ferroelectric loops and macroscopic piezoelectric response, respectively.Local piezoelectric response was acquired by PFM using Asylum Research MFP-3D equipped with a conductive Ti/Ir-coated tip (spring constant k of 1.4-5.8Nm −1 , resonant frequency f of 58-97 kHz), for which the dual AC resonance-tracking (DART) mode with AC voltage of 0.5 V was employed to avoid crosstalk with topographic information.
In Situ Time-Resolved Synchrotron XRD: The electric-field driven lattice strain and domain switching were evaluated using in situ time-resolved synchrotron XRD at the SPring-8 in Japan using the BL13XU and BL15XU beamlines.The diffraction patterns were acquired using an a-Si avalanche photodiode detector, and the collimated X-ray beam was focused on the electrode that was utilized for applying pulsed voltages.The setup for the in situ time-resolved synchrotron XRD measurements had been detailed elsewhere. [13,15,58,59] ,b (Supporting Information) clearly indicate the (001) pc out-of-plane orientation with 4-fold symmetry with an equal 90°interval, confirming the cube-on-cube relationship with the substrate, i.e., [100] pc KNN (001) pc || [100] Nb:STO (001).In Figure S1c (Supporting Information), a single diffraction peak from the film was observed in the symmetric RSM around Nb:STO 002.The asymmetric RSM around Nb:STO 203, however, consists of three splitting peaks (Figure S1, Supporting Information).Two of them are shifted upward and downward along Q z with the same Q x , and the third one has a larger Q x and is located at the midpoint of the other two along Q z .These results suggest that the film is in either the monoclinic or orthorhombic phase.However, based on the lattice parameters estimated from RSMs (a = 3.990 Å, b = 3.940 Å, c = 4.001 Å, and  = 90.28°), it can be clearly identified that the fabricated (001) pc KNN film is in the monoclinic phase and has the out-ofplane along c-axis and the in-plane composed of a and b axes,

Figure 1 .
Figure 1.a) XRD 2/ scan of the KNN film grown on Nb:STO (111) substrate.b) ϕ scan profiles for KNN (110) pc and Nb:STO (110).c-e) XRD RSMs around Nb:STO 111 with the temperature rising.f-h) XRD RSMs around Nb:STO 112 with the temperature rising.i) schematic illustration of possible polarization directions in the KNN film, for which the downward directions are excluded.Yellow and orange arrows are the polarization directions respectively for the ( 111) and (111) domains.

Figure 2 .
Figure 2. a) Room temperature P-E hysteresis loop measured at 10 kHz.b) Vertical PFM Acos(′) image and the corresponding component histogram.c) Lateral PFM Acos(′) image and the corresponding component histogram.

Figure 3 .
Figure 3. a) Schematic of in situ time-resolved synchrotron XRD measurement setup.In situ time-resolved synchrotron XRD with incremental electric fields for the (111) pc KNN film: b) 2/ scans, c) field dependence of volume fraction of ( 111) domain, d) field-induced strain, and e) extrinsic piezoelectric-response contribution.

Figure 5 .
Figure 5.Time evolution for the (111) pc KNN film with the applied electric field of 36 kV cm −1 .a) Color map for the ( 222) and (222) XRD peak profiles as a function of time (horizontal axis) and diffraction angle (vertical axis).b) Time-dependent volume fraction of ( 111) domain estimated from (a).The dashed lines mark the start and finish times of the application of the electric field.c) The schematic illustration of domain switching from (111) to ( 111).The orange and yellow arrows denote the respective polarization vectors, respectively.