UC Berkeley

A variety of physical mechanisms have been exploited to capture electronic images of a human fingerprint, including optical, capacitive, pressure, and acoustic mechanisms. Compared to other technologies, ultrasonic fingerprint sensors have two major advantages (1) they are insensitive to contamination and moisture on the finger (2) they have the ability to measure images at multiple depths hundreds of microns from the sensor surface. With the maturity of the thin film piezoelectric materials technology and MEMS-CMOS eutectic wafer-bonding process, piezoelectric micromachined ultrasonic transducers (PMUTs) arrays with electrical addressing to individual pixel have been proposed and developed for ultrasonic fingerprint sensors.

This research focuses on the modeling and characterization of the PMUT-based ultrasonic fingerprint sensors. First of all, a 51.7% fill-factor, 110 x 56 rectangular PMUTs array is demonstrated with improved pressure output, receive sensitivity, and image resolution. With a customized CMOS design, time-gated images collected at two imaging depths was demonstrated to construct overlaid patterns separated axially by 127 um. Using another 65 x 42 circular PMUTs array ultrasonic fingerprint sensor design, transmit beamforming is implemented. The measured TX pressure output with beamforming is 25 kPa and the 3 dB beam-width is 50 um, a 1.6 x increase in pressure and 6.4 x decrease in beam-width relative to non-beamformed measurements. Beamforming increases the receive voltage by a factor of 1.4 and the SNR by 7 dB. On the other hand, the mode shapes of the dense rectangular PMUTs array with shared mechanical anchors are more complicated than a single-MUT dynamic model. Finite element model was used to model the mode shapes exhibited by the entire array, while experimental measurements of the mode-shape via laser Doppler vibrometry (LDV) was used to compute the volume velocity, which correlates to the measured far field pressure. In order to have an accurate prediction of the 2D near field pressure output from dense transducer arrays, time domain acoustic simulation using k-wave MatLab toolbox and vibration displacement measurements using LDV were utilized. Effects of mechanical crosstalk, acoustic surface waves, and anchor motions were also characterized. To further improve the fingerprint sensors performance, equivalent circuit model of circular PMUTs operating at the same resonant frequency in air is formulated. PMUT displacement is proved to be inversely proportional to the device thickness. Sub-micro thick piezoelectric layer devices were designed and fabricated to demonstrate the thickness scaling benefits.