Development of MEMS neuristors via capacitive micromachined transducer technology

Date

2025-5-12

Author

Vize, Berre

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

259
views

0
downloads

Cite This

The rapidly growing demand for technology-driven applications requires substantial computational resources, yet traditional transistor-based analog and digital electronics face significant energy efficiency and scalability limitations. Although brain-inspired neuromorphic hardware offers a promising alternative, existing memristor-based neuristors suffer from reliability challenges, limiting their practical application. This study introduces a novel MEMS-based neuristor utilizing capacitive micromachined ultrasonic transducers (CMUTs) to provide a CMOS-integrable, scalable, and high-performance alternative for next-generation neuromorphic electronics. This design achieves threshold-based spiking without relying on memristive elements, significantly improving reliability. The MEMS neuristor utilizes collapse-snapback hysteresis in CMUTs and Fowler-Nordheim tunneling, which determines the electrical contact onset voltage (ECOV), enabled by doped polysilicon dimples. The microfabricated CMUTs operate in two distinct modes, resistive and capacitive, depending on whether the ECOV (~40 V) is exceeded. Resistive mode CMUTs, with a collapse voltage of ~77 V, exhibit a shark fin-type resistance behavior with an average contact resistance of 4.5 kΩ. Due to the spring softening effect, their resonance frequency is tunable between 260-190 kHz via DC bias (70-75.5 V). Capacitive mode CMUTs, with a collapse voltage of ~33 V, operate between 140-90 kHz with a DC bias of 25-30 V and demonstrate short- and long-term plasticity (STP, LTP) depending on the DC bias relative to the snapback voltage. Resistive mode CMUTs inherently exhibit STP. The design is validated through experimental measurements and circuit-level simulations, confirming sub- and super-threshold responses essential for neuromorphic computing. The MEMS neuristor holds significant potential for energy-efficient and high-performance neuromorphic computing systems.

Subject Keywords

Neuristor, CMUT, MEMS, Collapse-Snapback Hysteresis, Quantum Tunneling

URI

https://hdl.handle.net/11511/114946

Collections

Graduate School of Natural and Applied Sciences, Thesis