High Aspect-ratio Biomimetic Hair-like Microstructure Arrays for MEMS Multi-Transducer Platform

Abstract

Many emerging applications of sensing microsystems in health care, environment, security and transportation systems require improved sensitivity and selectivity, redundancy, robustness, increased dynamic range, as well as small size, low power and low cost. Providing all of these features in a system consisting of one sensor is not practical or possible. Micro electro mechanical microsystems (MEMS) that combine a large sensor array with signal processing circuits could provide these features. To build such multi-transducer microsystems we get inspiration from “hair”, a structure frequently used in nature. Hair is a simple yet elegant structure that offers many attractive features such as large length to cross-sectional area ratio, large exposed surface area, ability to include different sensing materials, and ability to interact with surrounding media in sophisticated ways. In this thesis, we have developed a microfabrication technology to build 3D biomimetic hair structures for MEMS multi-transducer platform. Direct integration with CMOS will enable signal processing of dense arrays of 100s or 1000s of MEMS transducers within a small chip area. We have developed a new device structure that mimics biological hair. It includes a vertical spring, a proof-mass atop the spring, and high aspect-ratio narrow electrostatic gaps to adjacent electrodes for sensing and actuation. Based on this structure, we have developed three generations of 3D high aspect-ratio, small-footprint, low-noise accelerometers. Arrays of both high-sensitivity capacitive and threshold accelerometers are designed and tested, and they demonstrate extended full-scale detection range and frequency bandwidth. The first-generation capacitive hair accelerometer arrays are based on Silicon-on-Glass (SOG) process utilizing 500 µm thick silicon, achieving a highest sensor density of ~100 sensors/mm2 connected in parallel. Minimum capacitive gap is 5 μm with device height of 400 μm and spring length of 300 μm. A custom-designed Bosch deep-reactive-etching (DRIE) process is developed to etch ultra-deep (> 500 µm) ultra-high aspect-ratio (UHAR) features (AR > 40) with straight sidewalls and reduced undercut across a wide range of feature sizes. A two-gap dry-release process is developed for the second-generation capacitive hair accelerometers. Due to the large device height at full wafer thickness of 1 mm and UHAR capacitive transduction gaps at 2 µm that extend > 200 µm, the accelerometer achieves sub-µg resolution (< 1µg/√Hz) and high sensitivity (1pF/g/mm2), having an area smaller than any previous precision accelerometers with similar performance. Each sensor chip consists of devices with various design parameter to cover a wide range. Bonding with metal interlayers at < 400 °C allows direct integration of these devices on top of CMOS circuits. The third-generation digital threshold hair accelerometer takes advantage of large aspect-ratio of the hair structure and UHAR DRIE structures to provide low noise (< 600 ng/√Hz per mm2 footprint proof-mass due to small contact area) and low power threshold acceleration detection. 16-element (4-bit) and 32-element (5-bit) arrays of threshold devices (total chip area being < 1 cm2) with evenly-spaced threshold gap dimensions from 1 µm to 4 µm as well as with hair spring cross-sectional area from 102 µm to 302 µm are designed to suit specific g-ranges from < 100 mg to 50 g. This hair sensor and sensor array technology is suited for forming MEMS transducer arrays with circuits, including high performance IMUs as well as miniaturized detectors and actuators that require high temporal and spatial resolution, analogous to high-density CMOS imagers.