Scholarly article on topic 'Magnetoelectrics for magnetic sensor applications: status, challenges and perspectives'

Magnetoelectrics for magnetic sensor applications: status, challenges and perspectives Academic research paper on "Materials engineering"

CC BY-NC-ND
0
0
Share paper
Academic journal
Materials Today
OECD Field of science
Keywords
{}

Abstract of research paper on Materials engineering, author of scientific article — Yaojin Wang, Jiefang Li, D. Viehland

The magnetoelectric (ME) effect, with cross-correlation coupling between magnetic and electric degrees of freedom, is associated with two promising application scenarios: magnetic field sensors and electric-write magnetic-read memory devices. In this review, we highlight recent progress in ME laminates for sensor applications, in particular with regards to the most difficult technical obstacle to their practical use (i.e. reduction of equivalent magnetic noise), while presenting an evolution of ME materials. The challenges and perspectives for the technical obstacles that would enable ME composites for sensor applications are emphasized.

Academic research paper on topic "Magnetoelectrics for magnetic sensor applications: status, challenges and perspectives"

Materials Today • Volume 00, Number 00• June 2014

RESEARCH

.^twxsume. ELSEVIER

Magnetoelectrics for magnetic sensor applications: status, challenges and perspectives

Yaojin Wang*, Jiefang Li and D. Viehland

Virginia Polytechnic Institute and State University, Materials Science & Engineering, Blacksburg, VA 24061, United States

The magnetoelectric (ME) effect, with cross-correlation coupling between magnetic and electric degrees of freedom, is associated with two promising application scenarios: magnetic field sensors and electric-write magnetic-read memory devices. In this review, we highlight recent progress in ME laminates for sensor applications, in particular with regards to the most difficult technical obstacle to their practical use (i.e. reduction of equivalent magnetic noise), while presenting an evolution of ME materials. The challenges and perspectives for the technical obstacles that would enable ME composites for sensor applications are emphasized.

Introduction

Magnetic field sensors have assisted mankind in achieving and operating thousands of functions for many decades [1], which are becoming popular methods of implementing the non-contacting location of moving objects, and electronic guiding. There are many approaches to sense magnetic fields, most of them based on the intimate correlation between magnetic and electric phenomena: including search coil, fluxgate, optical pumps, SQUID, Hall-effect, magneto-resistance, and giant magnetoimpedance, etc. [1]. The subjects and comparisons can be found in previous review articles [1-5]. Since the turn of the millennium, ME composites, incorporating both ferroelectric and ferromagnetic components, have drawn much attention. They typically yield giant ME coupling above room temperature, which has stimulated technological development based on their significant promise in next generation magnetic sensor applications. However, the practical usefulness of a magnetic sensor is determined not only by the output signal of the sensor in response to an incident magnetic field, but also by the equivalent magnetic noise generated in the absence of an incident field [6], together with other important specifications: such as bandwidth, operation range and linearity, power consumption and spatial resolutions, temperature effects, etc. [4]. In particular, the reduction of equivalent magnetic noise is perhaps the most difficult technical obstacle to the practical use of magnetoelectrics, or any other, magnetic sensors. To tackle this

*Corresponding author. Wang, Y. (yaojin@vt.edu)

1369-7021/© 2014 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.mattod.2014.05.004

challenge, two corresponding strategies have been employed: (i) enhancement of the ME coefficients or voltage gain and (ii) rejection of sources of noise. We describe the evolution and recent progress on these two strategies using Metglas/piezofiber laminates as a model.

Evolution of the ME effect and materials

The magnetoelectric (ME) effect is the appearance of an electric polarization P in a material upon applying a magnetic field H, and/or conversely, the appearance of magnetization M in a material when an electric field E is applied [7,8]. The intrinsic ME effect was theoretically predicted by Pierre Curie in 1894 [9], and was experimentally first observed in antiferromagnetic Cr2O3 single crystals in 1961 [10]. In the past century, from a perspective of material constituents and dimensions, ME materials have evolved from single phase compounds to particulate composites, to laminated composites, and finally to micro-/ nano-thin films (see Fig. 1). Until now, over ten different single-phase compound families have been widely studied as ME materials. These single-phase ME material studies have been summarized in recent review papers [9,11-14], such as the well-known BiFeO3 (BFO) and rare-earth manganates [12,15,16]. Investigations were motivated by potential applications in information storage and spintronics [9,13,17-23]; however, a high inherent ME coupling, especially above room temperature, has not yet been found in single-phase materials as most of them have low Curie temperatures [17,24].

Please cite this article in press as: Y. Wang, et al., Mater. Today (2014), http://dx.doi.Org/10.1016/j.mattod.2014.05.004

RESEARCH

FIGURE 1

Evolution on the development of ME materials: from single-phase compounds to multi-phase ferromagnetic/ferroelectric composites and from bulk laminates to micro-/nano-thin films. The main applications, advantages and/or disadvantages, research emphasis or challenges of each generation are also summarized in this figure.

The difficulties associated with uniting electric and magnetic orderings in a single phase material have been circumvented by forming multi-phase ME composites of ferromagnetic and ferroelectric components that can be electromagnetically coupled by stress mediation [24,25]. The ME effect in composites is a product tensor property first proposed by van Suchtelen in 1972 [24,26]. Upon application of a magnetic field to ME composites, the ferromagnetic component produces a deformation due to magnetostriction that is transferred to the ferroelectric component via interfacial bonding, in turn this induces an electric charge across the piezoelectric phase due to piezoelectricity. Soon after van Suchtelen proposed the possibility of enhanced ME effects in composites, scientists experimentally and theoretically developed particulate ferromagnetic/ferroelectric ME composites in BaTiO3 (BTO)-CoFe2O3 (CFO) and ferrite-Pb(Zr, Ti)O3 (PZT) systems with various phase connectivity schemes (i.e. 0-3,1-3 and 3-3) [27-30]. A historical perspective of research progress on particulate composites has been summarized in recent review articles [9,24,31]. In particular, relatively low ME coefficients and high dielectric losses make this type of ME composite technologically challenging for sensor and other applications [9]. Accordingly, this review will not discuss the particulate composites in any further detail.

The limitations that ME materials have faced were finally overcome in 2001 by using laminate, instead of particulate composites [32,33]. In laminated PZT/Terfenol-D composite samples, Ryu et al. acquired an ME voltage coefficient of up to 4.7 V/cm x Oe, Oe, which far exceeded the highest value reported from any particulate composite [9,33]. To date, several kinds of ME laminate

composite have been experimentally and theoretically investigated [9,24,34,35]: (a) magnetic ferrite (i.e. CFO, NFO) and piezoelectric ceramics (i.e. PZT) [36-40]; (b) magnetic alloys (i.e. Terfenol-D, Ni, Metglas) and piezoelectric polymer/ceramics/crystals [25,4154] (i.e. Pb(Mgi/3Nb2/3)O3-PbTiO3 (PMN-PT), Pb^n^Nb^^-PbTiO3 (PZN-PT));and (c) magnetic alloys, with interdigitated (ID)-electrodes and piezofibers (i.e. PZT, PMN-PT, PZN-PT) [5564]. The value of the ME coupling in laminated composites is determined by three issues: (i) the basic material parameters of the constituent phases (dielectric constant, magnetic permeability, elastic stiffness, and the piezoelectric and piezomagnetic coefficients) [61,64-66]; (ii) the volume/thickness ratio of the constituent layers [34,42,59,67]; and (iii) the operation mode (i.e. orientation of the constituent phases and an applied magnetic field) [24,25,42,53,57,67-70]. To date, the multi-push-pull configuration of ME laminates, consisting of magnetostrictive Metglas alloys and piezofibers with ID-electrodes [25,56], as shown in Fig. 2a,b, possess the largest ME coefficients and the highest sensitivity to magnetic field variations [25]. Their discovery can be regarded as a milestone in the development of ME materials for magnetic sensor applications. The progress of this particular type of ME laminates and its related magnetic sensor applications will be addressed in more detail in the next section.

In view of device miniaturization and component connectivity at the atomic-level, ME micro-/nano-thin films have emerged and recently flourished due to advances in thin-film growth techniques [17,71]. Most of the research activity in ME thin films was motivated by promising applications in electric-write/magnetic-read memories

MATTOD-329; No of Pages 7 ARTICLE IN PRESS

Materials Today • Volume 00, Number 00• June 2014 RESEARCH

FIGURE 2

(a) Schematic diagram of a multi-push-pull configuration ME composite and (b) exploded view photo of constituent components: six-layers of magnetostrictive Metglas and a piezoelectric core composite consisting of piezofibers interrogated by a pair of Kapton® interdigitated (ID) electrodes. (c) A summary on the development of ME coefficient and noise floor at 1 Hz for the multi-push-pull mode ME sensors. (d) Optical micrograph of a longitudinally poled push-pull element in the core composite. Photographs of the complete sensor detection unit consisting of (e) a multi-push-pull mode Metglas/ piezofiber ME composite and (f) a low noise charge amplifier (arranged from Ref. [56,79]). (g) Basic detection circuit and noise model for a ME sensor and an example of noise contributions of a sensor unit comprised of a Metglas/PZT-fiber laminate (C = 420 pF, tan 8 = 0.02, R = 60 GV) and 2SK369 based charge amplifier.

and spintronics [19-23,63,64], and the status and future perspectives have been comprehensively reviewed by authors in the field [16,17,71-73]. It is worth noting that micro thin films, fabricated by a MEMS technique, have exhibited an extremely high ME coefficient of 737V/cm x Oe at mechanical resonance and of 3.1V/ cm x Oe off resonance [74], providing a low equivalent magnetic noise of several pT Hz~1/2 at resonance [75,76].

Recent progress in ME sensor development

In the last five years, our research group has developed Metglas/ID-electrodes/piezofiber multi-push-pull configuration ME laminates and low-noise charge amplifiers for sensing low-frequency minute magnetic field variations, as summarized in Fig. 2c. We have mainly focused on: (i) optimization of the laminate constitutive characteristics (i.e. category [56,64,77,78], geometry [59,60,7981]); (ii) improvement of fabrication techniques [58,61,82]; (iii) packaging of sensor units;(iv) optimization of signal processing

conditions [83,84]; and (v) gradiometric configuration to reduce environmental noise sources [85].

Analysis of noise contributions

The sensing capabilities of an ME sensor to magnetic field variations are determined by both the output ME signal in response to an incident magnetic field and the intrinsic/extrinsic noise generated in the absence of an incident field. Moreover, the equivalent magnetic noise floor is influenced by the properties of the ME laminates and the parameters of the detection circuits. Accordingly, the noise sources and their corresponding contributions to the total equivalent magnetic noise must be considered. Based on these results, the detection circuitry and laminate design was optimized to reduce the equivalent magnetic noise floor. Generally, either voltage or charge detection schemes can be employed to collect and amplify small signals detected by ME laminates [75]. However, efforts have focused on a charge mode circuit, built

RESEARCH

around a low-noise operational amplifier, as it was more suitable for multi-push-pull mode Metglas/piezofiber laminates. Figure 2g presents a schematic for a charge mode detection circuit, including various noise sources. Details can be found in recent publications [83,84]. Figure 2h shows typical noise contributions from various noise sources for a typical Metglas/piezofiber sensor and accompanying low-noise operational amplifier (JFET 2SK369) based detection circuit (see Fig. 2f). From these results, one can clearly see that the dielectric loss noise and DC resistance noise of the laminate, and the current noise of the charge amplifier are the main contributing sources to the total noise floor. At a frequency of f = 1 Hz, the total noise was dominated by the dielectric loss noise source, which was two times larger than that of the DC resistance and current noises. Clearly, measurement circuit optimization and laminate design (i.e. realization of high ME coefficients, low dielectric loss and high DC resistance) are crucial for ME sensor applications.

Noise reduction in ME sensors Fabrication technique

Since the ME coefficients of laminates are significantly influenced by interfacial bonding characteristics [9,65], studies were initiated to optimize the mechanical and dielectric properties of the interfacial bonding layer beginning in 2009 (i.e. the interface between ID-electrodes and the piezofibers, as shown in Fig. 2d). It was found that the dielectric loss was significantly influenced by the type and thickness of the interfacial bonding epoxy, and that the ME coefficient was slightly enhanced and the equivalent magnetic noise floor notably reduced, via the reduction of the dielectric loss by a mechanized spin-coating method and better-control of the epoxy amount [58,61]. Although these results indicated that ID-electrodes/piezofiber core composites had smaller capacitances (C) and ME charge coefficients (aQ when a stycast epoxy (1264, USA) was used relative to other commercially available epoxies, much lower dielectric loss factors were found with laminate epoxied with stycast resulting in notably lower equivalent magnetic noise floors [58].

Furthermore, with regards to the piezoelectric layers, the properties were quite sensitive to the poling conditions, in particular the dielectric loss. In 2010, a systematic study of the influence of the poling process on the ME coefficient and dielectric loss factor was performed. The results revealed a notable reduction in the equivalent magnetic noise floor when using a lab-made automatic poling system [82].

Detection circuit optimization

In an effort to reduce the contributions from electronic noise sources from the detection circuits, several different detection schemes have been designed based on various operational amplifiers (i.e. AD795, LTC6240, LMC6040, JFET 2SK369) [84]. After optimization was completed in 2010, the noise floor for a Metglas/ piezofiber ME sensors connected to a JFET 2SK369 based circuit was reduced by a factor of ~2x relative to that with a LMC6040 based one. In particular, the results revealed a significant reduction in the spectral noise density for f > 1 Hz. This improvement is primarily due to the use of lower-noise, higher-voltage op-amps in the first stage of the detection circuit scheme and to the use of larger feedback resistors.

High-performance piezoelectric single crystal sensors

Piezoelectric single crystals, such as PMN-PT or PZN-PT, exhibit ultrahigh longitudinal piezoelectric coefficients of d33 « 2000 pC/ N and low dielectric losses of tan 8 « 0.005. This provides opportunities for the realization of higher magnetic field sensitivity through a combination of giant ME effects and ultralow equivalent magnetic noises. Recent investigations have shown about a 2x enhancement in the magnetic field sensitivity for ME laminates constructed with piezoelectric single crystals. Accordingly, the equivalent magnetic noise floors are about 3-4 times lower for laminates with PMN-PT or PZN-PT fibers than those with PZT ones [56,64]. After optimization of the fabrication techniques and detection circuit conditions, high-performance PMN-PT crystal based ME sensors were reported in 2011 [54], as shown in Fig. 2e. These sensors exhibited low dielectric losses (tan 8 ~ 0.007), high DC resistances (R ~ 80 GV), large ME charge coefficients (aQ ~ 2700pC/0e) and extremely low equivalent magnetic noise floors of around 5 pT Hz~1/2 at f = 1 Hz. In particular, the equivalent magnetic noise of the ME sensor unit was as low as about 1 pT Hz~1/2 at a frequency of only several Hz [58]. Moreover, the repeatability of the equivalent magnetic noise floor was well controlled (error ~ 5%) using the fabrication technique described in Refs. [56,58]. The superior properties and fabrication repeatability make these Metglas/PMN-PT fiber laminates particularly promising for use in ultralow magnetic field sensing.

Following these reports, it has proven increasingly difficult to further reduce the equivalent magnetic noise in a single ME sensor unit. However, the equivalent magnetic noise floor was reduced, and the magnetic field sensitivity was enhanced, by the use of sensor arrays connected either in serial or parallel modes [62,86]. In 2012, a giant ME charge coefficient of ~6500 pC/Oe at f = 1 Hz, and in turn an extremely low equivalent magnetic noise of ~3.6 pT Hz~1/2, was obtained by stacking four Metglas/PMN-PT sensors in a parallel connection. Please bear in mind that stacking of the sensor units has an obvious disadvantage of requiring a larger volume.

For comparison with the ME composite-based magnetic field sensors, we briefly review the performance of other commercial-supplied and widely used instruments. Representative noise floors are plotted in Fig. 3. The noise floors of a variety of fluxgate magnetometers, being vector instruments similar to ME sensors, have been reviewed in previous reports [1,6,87]. The most sensitive instruments have a noise level as low as ~10~12 T Hz~1/2 at 1 Hz, with a spectral density scaling as 1/f at lower frequencies. Induction coils measure the magnetic flux density, whose sensitivity is significantly dependent on the size and number of turns. The output of the coil is connected to an amplifier that at low frequencies inevitable exhibits 1/fnoise. Because the voltage response of the coil falls off as f, the spectral density of the magnetic field noise scales as 1/f3 at low frequencies. The noise floor above 1 Hz can be remarkably low, however, a typical value is 3 x 10~n T Hz~1/2 [6]. SQUID fabricated from low transition temperature superconductors and operated at extremely low temperatures are routinely used as ultrasensitive magnetometers. In particular, an extremely low noise floor of ~10~14 T Hz~1/2 at 1 Hz can be obtained . However, the physical requirement makes SQUID not as popular as fluxgate instruments in industrial applications in consideration of its disadvantages of high cost, large size, etc.

Materials Today • Volume 00, Number 00• June 2014

RESEARCH

FIGURE 3

Performance of various magnetic field sensors. The measured and estimated noise floor for ME magnetic field sensor was represented by a Metglas/PMN-PT sensor with low-noise charge amplifier [ref.56]. The noise floors for other types of magnetic field sensors are representative [following ref.6], and the white noise of 4.2 K SQUID, fluxgate and coil has been arbitrarily chosen to be 10~14, 10~12, and 3 x 10~" THz~1/2, respectively.

Challenges and future perspectives

Current and future challenges

In the past few years, although considerable progress has been made, the exploitation of high magnetic field sensitivity in two-phase ferromagnetic/ferroelectric laminate composites requires further development and the identification of end users. The quest for multi-phase ME materials exhibiting high magnetic field sensitivity, especially in a real-world environment, remains a challenge.

Interfacial bonding

It has been found that Mn substituents in PMN-PT are effective in achieving an extremely low dielectric loss factor of tan 8 < 0.001 due to the selective pining of 180° domain wall motions. Similar values of tan 8 can be obtained in a longitudinal-transverse mode sensor [49]. However, the values of tan 8 for a multi-push-pull mode sensor are generally notably higher of ~0.005, due to limitation of the interfacial bonding between the ID-electrodes and piezofibers. This aspect still needs improvement. The question remains open as to: 'how does one control the formation of interfacial bonding between ID-electrodes and piezofibers in a manner such that the dielectric loss of the laminate is equal to that of the piezofibers?'

As can be seen in Fig. 3, the ME magnetic sensor, a new alternative generation of magnetometers, has a comparable noise floor to the fluxgate and optical pumping ones at 1 Hz, and even lower values above 10 Hz. One should bear in mind, of course, that the sensitivity of the ME sensors is significantly influenced by external vibration sources. The development of ultra-highly sensitive ME sensors for end users will require considerable progress in reduction and elimination of external vibration noises. Recently, in our group, we have emphasized effort in modulation techniques, which are expected to significantly reduce the external vibration noise.

External vibration noise

With regards to applications of ME sensors in real-world environments, difficulties in enhancing the magnetic field sensitivity are aggravated by contamination of the ME signals by external vibration noise sources, which act to increase the equivalent magnetic noise floor. For instance, with regards to the moving vehicles [1]: the objects produce a magnetic field variation of low frequency (fo), whereas the field is inevitably contaminated by acoustic noise sources, which may have a close frequency to that of the magnetic field, as shown in Fig. 4a. Due to the piezoelectric component in the ME sensor (see Fig. 4b), the output signal from the ME sensor is

FIGURE 4

Schematic illustrations of the challenges and future perspectives. (a) Fluctuations in low frequency noise (i.e. 1/fn, with n as an integer) and electromagnetic interference make decrease in the noise floor increasingly challenging. (b) The working principle of ME sensors in passive detection mode, illustrating that the acoustic and electromagnetic interference induced external noise is inevitable. (c) The working principle and prototype ME sensor in active detection mode, in which a modulation field with carrier frequency (fC » >f0) is applied to the sensor. (d) The output signals from passive and active sensors. The low frequency H field can be sensed via the modulation technique at the higher frequencies (fC ± f0), where the noise floor is lower than f0 (i.e. white noise range), and the output signals from low frequency acoustic noise sources can be significantly reduced.

RESEARCH

comprised of an H-field induced voltage via the ME effect and vibration induced ones via the piezoelectric effect. Thus, the reduction or elimination of the external noise is an important challenge in the future for ME sensors, as it is for all sensors.

Furthermore, thermal noise can also be induced via the pyro-electric effect in an environment with a temperature gradient [88]. Sensor packaging and adiabatic shielding is a remaining obstacle that can restrict ME sensor applications.

Theoretical model for multi-push-pull mode laminates

A simple model for ME coefficients and equivalent magnetic noise floors in multi-push-pull mode laminates was proposed based on the assumption that the polarization in the push-pull configuration was uniform. The experimental and theoretical properties were in agreement with each other, over a moderate operational range. It has been experimentally found that the equivalent magnetic noise floor decreases with increasing ID-electrode spacing over some range, although the theory predicts that the equivalent magnetic noise should be independent of the ID-electrode geometry. A more precise model is of importance to guide the design of ME laminates.

Future perspectives

In order to overcome the current and future challenges, future directions should include the following.

Self-biasing effect for sensor arrays

As mentioned above, sensor stacking is an effective method to reduce the equivalent magnetic noise. However, the additional permanent-magnets that provide the required optimum external bias are opposite in trends to the need of device miniaturization, and also introduce mutual inductance effects [62,63]. Recently, self-biased ME sensors have been proposed with a relatively high ME coefficient due to exchange bias and gradient magnetostrictive effects [78,89,90]. In the future, self-biased ME sensors with higher ME coefficients should be a research issue. The equivalent magnetic noise could then be further reduced via sensor stacking using high-performance self-biased sensors.

Gradiometer

Gradiometry has carried over as an effective method to reduce environmental noises. It employs two (or more) magnetic sensors in a differential mode configuration. Such configurations are capable of rejecting common noise sources that are coherently shared between two sensors spatially separated across a baseline or by sensor arrays [85].

The noise rejection efficiency of the gradiometer system is limited by internal incoherent noise sensors and differences in properties between the ME sensors in a pair, especially with regards to their phase responses. Thus, advancements in reducing the equivalent magnetic noise floor and phase shifts of individual ME sensors would be beneficial to ME gradiometer developments in the future.

Modulation technique

Based on a non-linear ME effect, it has been established that a modulation-demodulation technique can be employed to reduce the equivalent magnetic noise at arbitrary frequencies [91,92]. The

technological principle is essentially from the frequency dependence of noise, as illustrated in Fig. 2h. The low frequency magnetic signal that needs to be detected can be modulated to the high frequency region due to the non-linear ME effect, where the noise floor is much smaller, providing the possibility to reduce the equivalent magnetic noise, as shown in Fig. 4c,d.

Acknowledgments

This review is based upon work supported as part of DARPA and ONR. Authors would like to greatly acknowledge the help and discussion from Dr. Keith McLaughlin in SAIC and Prof. Christophe Dolabdjian in GREYC. Authors would also like to acknowledge Prof. Haosu Luo in Shanghai Institute of Ceramics for providing high-performance piezoelectric single crystals.

References

[1] J. Lenz, A.S. Edelstein, IEEE Sens. J. 6 (3) (2006) 631.

[2] J.E. Lenz, Proc. IEEE 78 (6) (1990) 973.

[3] D. Robbes, et al. IEEE Trans. Appl. Supercond. 11 (1) (2001) 629.

[4] D. Robbes, Sens. Actuators A: Phys. 129 (1-2) (2006) 86.

[5] P. Ripka, J. Magn. Magn. Mater. 320 (20) (2008) 2466.

[6] John Clarke, R.H. Koch, Science 242 (1988) 217.

[7] L.D. Landau, E. Lifshitz, Electrodynamics of Continuous Media, Perfamon, Oxford, 1960.

[8] N.A. Spaldin, M. Fiebig, Science 309 (5733) (2005) 391.

[9] M. Fiebig, J. Phys. D: Appl. Phys. 38 (8) (2005) R123.

[10] V.J. Folen, et al. Phys. Rev. Lett. 6 (11) (1961) 607.

[11] W. Prellier, et al. J. Phys. Condens. Mater. 17 (30) (2005) R803.

[12] A.M. Kadomtseva, et al. Low Temp. Phys. 32 (8-9) (2006) 709.

[13] S.W. Cheong, M. Mostovoy, Nat. Mater. 6 (1) (2007) 13.

[14] Y. Tokura, S. Seki, Adv. Mater. 22 (14) (2010) 1554.

[15] S. Valencia, et al. Nat. Mater. 10 (10) (2011) 753.

[16] C.A.F. Vaz, et al. Adv. Mater. 22 (26-27) (2010) 2900.

[17] J. Ma, et al. Adv. Mater. 23 (9) (2011) 1062.

[18] R. Ramesh, Nat. Nanotechnol. 5 (2010) 761.

[19] S.A. Wolf, et al. Proc. IEEE 98 (12) (2010) 2155.

[20] J. Atulasimha, S. Bandyopadhyay, Appl. Phys. Lett. 97 (17) (2010) 173105.

[21] N.A. Pertsev, H. Kohlstedt, Nanotechology 21 (47) (2010) 475202.

[22] N. Tiercelin, et al. Appl. Phys. Lett. 99 (19) (2011) 192507.

[23] K. Roy, et al. Appl. Phys. Lett. 99 (6) (2011) 063108.

[24] C.W. Nan, et al. J. Appl. Phys. 103 (3) (2008) 031101.

[25] J. Jin, et al. Adv. Mater. 23 (33) (2011) 3853.

[26] J.V. Suchtelen, Philips Res. Rep. 27 (1972) 28.

[27] A. Vanrun, et al. J. Mater. Sci. 9 (10) (1974) 1710.

[28] Y. Benveniste, Phys. Rev. B 51 (22) (1995) 16424.

[29] C.W. Nan, Phys. Rev. B 50 (9) (1994) 6082.

[30] G. Srinivasan, et al. Phys. Rev. B 64 (21) (2001) 214408.

[31] S. Priya, et al. J. Electroceram. 19 (1) (2007) 147.

[32] J. Ryu, et al. Jpn. J. Appl. Phys. 40 (8) (2001) 4948.

[33] J. Ryu, et al. J. Electroceram. 8 (2) (2002) 107.

[34] M.I. Bichurin, et al. J. Appl. Phys. 107 (5) (2010) 053904.

[35] M.I. Bichurin, et al. J. Appl. Phys. 107 (5) (2010) 053905.

[36] M. Bichurin, et al. Phys. Rev. B 68 (5) (2003) 054402.

[37] M.I. Bichurin, et al. Phys. Rev. B 68 (13) (2003) 132408.

[38] M.I. Bichurin, et al. Phys. Rev. B 72 (6) (2005) 060408.

[39] G. Srinivasan, et al. Phys. Rev. B 67 (1) (2003) 014418.

[40] Y. Yan, et al. Appl. Phys. Lett. 102 (5) (2013) 052907.

[41] V.M. Laletin, et al. Appl. Phys. Lett. 87 (22) (2005) 222507.

[42] S.X. Dong, et al. IEEE Trans. Ultrason. Ferroelectr. Freq. Control 51(7) (2004) 794.

[43] S. Dong, et al. Appl. Phys. Lett. 87 (6) (2005) 062502.

[44] S.X. Dong, et al. Appl. Phys. Lett. 87 (22) (2005) 222504.

[45] S.X. Dong, et al. Appl. Phys. Lett. 91 (2) (2007) 022915.

[46] J.Y. Zhai, et al. Appl. Phys. Lett. 89 (8) (2006) 083507.

[47] Z.P. Xing, et al. Appl. Phys. Lett. 89 (11) (2006) 112911.

[48] Y.J. Wang, et al. J. Appl. Phys. 103 (12) (2008) 124511.

[49] Y. Wang, et al. Appl. Phys. Lett. 101 (2) (2012) 022903.

[50] Y. Wang, et al. Phys. Status Solidi (RRL) - Rapid Res. Lett. 5 (7) (2011) 232.

[51] J. Jiao, et al. Appl. Phys. Lett. 101 (23) (2012) 232906.

Materials Today • Volume 00, Number 00• June 2014

RESEARCH

[52] J.G. Wan, et al. Appl. Phys. Lett. 86 (20) (2005) 202504.

[53] Y. Wang, et al. Appl. Phys. Lett. 100 (20) (2012) 202903.

[54] D.V. Chashin, et al. Appl. Phys. Lett. 92 (10) (2008) 102511.

[55] S.X. Dong, et al. Appl. Phys. Lett. 89 (25) (2006) 252904.

[56] Y.J. Wang, et al. Adv. Mater. 23 (2011) 4111.

[57] D. Hasanyan, et al. J. Appl. Phys. 112 (1) (2012) 013908.

[58] Y. Wang, et al. J. Alloys Compd. 513 (2012) 242.

[59] Y. Wang, et al. IEEE Trans. Ultrason. Ferroelectr. Freq. Control 60 (6) (2013) 1227.

[60] J. Das, et al. Appl. Phys. Lett. 95 (9) (2009) 092501.

[61] M. Li, et al. J. Am. Ceram. Soc. 94 (11) (2011) 3738.

[62] M. Li, et al. J. Appl. Phys. 111 (10) (2012) 104504.

[63] M. Li, et al. J. Appl. Phys. 111 (3) (2012) 033923.

[64] J. Gao, et al. J. Appl. Phys. 108 (8) (2010) 084509.

[65] G. Liu, et al. J. Appl. Phys. 95 (5) (2004) 2660.

[66] M.I. Bichurin, et al. Phys. Rev. B 64 (9) (2001) 094409.

[67] S.X. Dong, et al. IEEE Trans. Ultrason. Ferroelectr. Freq. Control 50 (10) (2003) 1253.

[68] S.X. Dong, et al. J. Appl. Phys. 96 (6) (2004) 3382.

[69] J. Gao, et al. IEEE Trans. Ultrason. Ferroelectr. Freq. Control 58 (8) (2011) 1545.

[70] V.M. Petrov, et al. J. Appl. Phys. 105 (6) (2009) 063911.

[71] R. Ramesh, N.A. Spaldin, Nat. Mater. 6 (1) (2007) 21.

[72] L. Yan, et al. J. Mater. Sci. 44 (19) (2009) 5080.

[73] N.A. Spaldin, R. Ramesh, MRS Bull. 33 (11) (2008) 1047.

[74] H. Greve, et al. Appl. Phys. Lett. 96 (18) (2010) 182501.

[75] R. Jahns, et al. IEEE Trans. Instrum. Meas. 60 (8) (2011) 2995.

[76] S. Marauska, et al. J. Micromech. Microeng. 22 (6) (2012) 065024.

[77] Y. Yang, et al. Mater. Res. Bull. 46 (2) (2011) 266.

[78] M. Li, et al. Appl. Phys. Lett. 102 (8) (2013) 082404.

[79] Y. Wang, et al. IEEE Trans. Ultrason. Ferroelectr. Freq. Control 59 (5) (2012) 859.

[80] J. Gao, et al. Appl. Phys. Lett. 99 (15) (2011) 153502.

[81] Y. Wang, et al. Appl. Phys. Lett. 101 (9) (2012) 092905.

[82] Y. Wang, et al. J. Alloys Compd. 519 (2012) 1.

[83] J. Gao, et al. Mater. Lett. 85 (2012) 84.

[84] Y. Wang, et al. Philos. Trans. R. Soc. A 372 (2009) (2014) 20120455.

[85] Y. Shen, et al. Sens. Actuators A: Phys. 171 (2) (2011) 63.

[86] Z.P. Xing, et al. IEEE Electron. Device Lett. 30 (5) (2009) 445.

[87] P. Ripka, Sens. Actuators A: Phys. 33 (3) (1992) 129.

[88] J.Y. Zhai, et al. Appl. Phys. Lett. 93 (7) (2008) 072906.

[89] E. Lage, et al. Nat. Mater. 11 (6) (2012) 523.

[90] U. Laletin, et al. Phys. Rev. B 85 (10) (2012) 104404.

[91] R. Jahns, et al. Sens. Actuators A: Phys. 183 (2012) 16.

[92] X. Zhuang, et al. IEEE Sens. J. 11 (10) (2011) 2266.