- This work explores the electric field induced strain mechanisms in environmentally benign Bi0.5Na0.5TiO3 (BNT) based piezoelectric thin films. Although BNT-based materials show promise as replacements for toxic Pb-based piezoelectrics, the displacement mechanisms are not consistent between thin films and bulk materials, and the differences have yet to be well understood. Here, in-situ structural characterization (2-dimensional X-Ray Diffraction (XRD2)) under applied field has been utilized and the effects of substrate clamping have been shown to be a large contributor to the differences between bulk and thin film embodiments.
Piezoelectric materials convert mechanical strain into dielectric displacement, and the converse, making these materials suitable for use as sensors, actuators, and transducers. Lead-based materials have received the most attention due to their high strain response and tunability, but the toxicity of lead is leading to legislative directives limiting its use. This is driving a search for suitable replacement materials, and BNT-based ergodic relaxor materials in bulk embodiments display properties that make them ideal for actuator applications. In thin film embodiments of the same materials, however, the response appears to be quite different.
Initial work was done to optimize the deposition of the bottom platinum electrode for the metal-insulator-metal devices commonly used to study piezoelectric films. A dense, smooth, and crystallographically textured Pt will yield better active, piezoelectric, layers by acting as crystal template and a diffusion barrier (true for both lead-based and non-lead-based piezoelectric thin films). Pt is sputter deposited on TiOx, which is oxidized from Ti that was sputter deposited on SiO2/Si. The thickness of the initial Ti was found to affect the achievable Pt surface roughness and crystal texture. Pt was smoothest and most textured with 30 nm of Ti.
With quality Pt bottom electrodes understood, chemical solution deposition of the active layer was then optimized to achieve optimal properties. The morphotropic phase boundary (MPB) composition, where structural instabilities result in enhanced dielectric, ferroelectric, and piezoelectric properties, 80Bi0.5Na0.5TiO3-20Bi0.5K0.5TiO3 (80-20 BNKT) and ergodic relaxor compositions [75-x/2](Bi0.5Na0.5)TiO3 - [25-x/2](Bi0.5K0.5)TiO3 - [x]Bi(Mg0.5Ti0.5)O3 (BNKT-xBMgT) where x = 2.5, 5, 10 were explored. Main process parameters; cation excess, pyrolysis temperature and time, and crystallization ramp rate, hold time, hold temperature, and atmosphere were studied.
X-ray diffraction was used to confirm phase purity and atomic force microscopy was used to understand morphology changes with processing parameters. The dielectric, ferroelectric, and piezoelectric properties were characterized and values of the d33,f piezoelectric coefficient were extracted from double beam laser interferometry measurements. For 675 μm diameter top electrodes dielectric constant, loss, and d33,f ranged from 450-900, 2-10%, and 16-96 pm/V respectively. Best case process parameters were:
- 6-12-12 % excess for Bi-Na-K in 80-20 BNKT and 4-8-8% excess for Bi-Na-K for BNKT-xBMgT
- 400 °C for 4 mins pyrolysis
- 100 °/sec ramp rate to 700 °C hold for 5 min all under 2 SLPM O2 crystallization
While 80-20 BNKT is the most researched BNT-based system, there remain differences in displacement response between films and bulk that have not been explained. In the BNT-BKT-xBMgT system, the differences between films and bulk are even more pronounced, with bulk systems showing a displacement response almost 8 times that of thin films. The in-situ XRD2 experiments have revealed that the fully clamped film embodiments undergo an irreversible strain with application of field but not a phase change as seen in bulk. Also, strain is almost entirely due to intrinsic effects and little extrinsic (non-180° domain rotation) effects. In partially released films, however, signs of phase change under applied field are present, and extrinsic effects are more prominent. These results show that the clamping stresses due to the substrate may be causing some of the differences between bulk and thin film embodiments. Further studies on fully released films should be completed to confirm these results.