Piezoelectric Micromachined Ultrasound Transducers Using Lead Zirconate Titanate Films
Open Access
- Author:
- Cheng, Christopher
- Graduate Program:
- Materials Science and Engineering
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- October 01, 2021
- Committee Members:
- Susan Trolier-Mckinstry, Chair & Dissertation Advisor
Tom Jackson, Outside Unit & Field Member
Sri-Rajasekhar Kothapalli, Outside Field Member
Clive Randall, Major Field Member
John Mauro, Program Head/Chair - Keywords:
- PMUT
Piezoelectrics
Acoustics
Ultrasound
Transducers
MEMS
Micromachined Ultrasound Transducers - Abstract:
- The goal of this research was to fabricate Piezoelectric Micromachined Ultrasound Transducers (PMUTs) and a) evaluate their suitability for 1D deterministic particle manipulation, b) establish their potential for ultrasound and photoacoustic imaging, and c) develop ways to improve their acoustic properties. In addition, the crack propagation was evaluated as a function of the Zr/Ti ratio in the lead zirconate titanate (PZT) films, as well as the failure behavior of clamped versus released multilayer-actuator devices. PMUT arrays were fabricated with high quality {001} oriented PZT that produced 9.5 kPa at 7.5 mm distance in water and 40 nm/V deflection in air at 6-8 MHz resonant frequency with a bandwidth of 62.5%. SiO2 beads 4 μm in diameter were successfully manipulated in 1D without the need of microfluidic control. In addition, multiple levitation planes were observed for driving frequencies greater than 30 MHz, well outside the operational bandwidth of 62.5%. The homogeneity between elements was high, with the permittivity variation between elements < 2%. The PMUT arrays were also used for ultrasound and photoacoustic imaging. For ultrasound imaging, an 18 mm by 18 mm clay target was imaged. A compact handheld photoacoustic imaging device was fabricated by wirebonding the array to a printed circuit board integrated with a fiber optic light guide. Targets such as pencil leads, a single follicle of hair, and tubes filled with indocyanine green (ICG) or blood were successfully imaged with photoacoustics. Pencil lead targets and the tube filled with ICG tubes were imaged even under 5 mm thick chicken tissue. However, the tubes filled with blood under chicken tissue could not be imaged well. In addition, for ultrasound imaging, scanning metal needle targets of 0.5 mm – 0.7 mm in diameter was difficult, even though 120 μm resolution was expected. This may have been due to scattering from a cylindrical target using a single element for transmit and receive. Three methods were investigated to increase the transmit and receive sensitivities of PZT based PMUTs. For transmit sensitivity, by increasing the effective area by 1.4 times and changing to a single rectangular diaphragm as opposed to a series of circular diaphragms, the pressure output increased by ~ 2 times. For receive sensitivity, two methods were investigated. Firstly, a DC bias was used to increase the receive sensitivity (or voltage coefficient) by approximately two times when 15 V (75 kV/cm) was applied to the PMUT with a PZT composition of 52/48. This increased the piezoelectric coefficient e31,f and decreased the relative permittivity. 15 V DC bias applied to the 52/48 PMUT generated stronger photoacoustic signals and enabled images of targets at greater depths. Secondly, PZT films of different compositions (with Zr/Ti ratios of 52/48, 40/60, 30/70, and 20/80) were tested for suitability in receive. When 0 DC bias was applied, PZT 20/80 had the highest voltage coefficient compared to PZT 52/48, 40/60, and 30/70. When 15 V (75 kV/cm) DC bias was applied, PZT 52/48 matched the receive sensitivity of PZT 20/80. The crack initiation stress and strength of the entire stack of 2 μm-thick PZT 52/48, 40/60, and 30/70 on Pt/TiO2/SiO2/Si substrates were tested. For PZT 52/48 films, it was found that the PZT films cracked prior to failure of the stack. As a result, the PZT film thickness served as the critical flaw size that drove failure. This produced a Weibull modulus of 17 (11 – 24 at 95% confidence interval) and a stack strength of 1137 MPa (1091 – 1183 at 95% confidence interval). As the PZT compositions moved towards PbTiO3, the characteristic strength increased while the Weibull modulus decreased to ~ 1580 MPa (1355 – 1807 MPa at 95% confidence interval) and 3.0 (2.5 – 3.6 at 95% confidence interval) respectively; these values are comparable to those of the substrate itself, suggesting a much broader distribution of critical flaw sizes. Two potential contributions to the change in cracking behavior were investigated: ferroelastic toughening and residual thermal stresses associated with the thermal expansion coefficient between the film and substrate upon cooling from the crystallization temperature to the Curie Temperature. The residual thermal stresses did not vary significantly with Zr/Ti ratio. It was found, however, that the volume fraction of c-domains increased as compositions became more Ti-rich. This could increase ferroelastic toughening, as more c-domains can switch to a-domains when the sample is subjected to biaxial tensile stress. It was found that the PZT films had a mixture of a-domains, c-domains, and rhombohedral phases for all average compositions. A line scan via TEM indicated that Zr/Ti concentration gradients were present for all compositions, which may account for the observation of a rhombohedral phase for average compositions in the tetragonal phase field. Finally, the breakdown behavior of multilayer actuator PZT based devices was investigated. Four main observations were made. First, released films showed two families of behaviors in the current versus time graphs: one in which the current exceeded two orders of magnitude higher than the base current along with various current spikes, and the other in which the diaphragm breaks before the current increases two orders of magnitude. Secondly, the lateral size of the thermal breakdown events for released films were much smaller (typically less than 5 μm) compared to clamped films. Thirdly, breakdown events were not connected via cracks in released films, in contrast to observations on clamped films. Fourthly, the released films were more likely to experience immediate membrane breakage upon voltage application. Fifthly, the activation energy for failure was found to be 1.02 ± 0.17 eV and 0.46 ± 0.20 eV for clamped and released films, respectively. The t50 lifetime between released and clamped devices were similar at temperatures ~ 150 °C – 170 °C but diverged at ~ 200 °C, where the released films exhibited higher lifetimes. Thermal imaging of the electrode indicated no significant changes in the temperature profile over time under voltage, so the difference in apparent activation energy is not thermal in origin. The thermal breakdown behavior for clamped films was also examined. It was found that these breakdown events were not randomly distributed in space, but were often within 50 μm from each other. The characteristic time to reach 50% of the cumulative number of breakdown events decreased with increasing applied electric field. Moreover, at lower fields, breakdown events were spaced out over a longer period of time.