University of Southampton
Displacement detection in silicon nano-optoelectromechanical systems (NOEMS) for the development of artificial intelligence hardware
Abstract
dc:description.abstractArtificial intelligence (AI) is increasingly prolific in the technology sectors and across society, but it is becoming constrained by the physical limits of the hardware platforms which are used in its deployment, which are primarily conventional complementary metal-oxide-semiconductor (CMOS) transistor-transistor logic (TTL). This motivates the development of more specialised hardware platforms. In this thesis, the foundational technologies of nano-electromechanical systems (NEMS) and nano-optoelectromechanical systems (NOEMS) are introduced and the details of transduction schemes for the detection of nanomechanical motion are reviewed, combined with a consideration of the role for NEMS and NOEMS devices within future, non-conventional AI hardware approaches. Their use necessitates accurate detection of their nanoscale and picoscale displacements. The detection of mechanical motion by electrical methods in NEMS devices, encompassing both motion which is actuated by frequency modulated input signals and which originates from thermomechanical noise, was investigated. Signal processing methods were applied for the detection of picometre-scale displacements of the latter. Free space optical detection methods were investigated through simulation and experimentation, suggesting strongly sensitive detection by Fabry-Pérot interferometry. New NEMS and NOEMS devices were designed as part of this investigation, the latter using photonic crystal cavities to enhance optical detection. The electrical characteristics and the mechanical and optical degrees of freedom of the devices were optimised for detection as part of the device fabrication. Integrated optical nanomechanical devices, including devices with engineered optomechanical coupling between adjacent nanobeams, were designed and built using a silicon photonics prototyping platform, and optimised for a telecommunications wavelength. The optical transmission of these devices was probed and their design was iteratively improved. Finally, the further work necessary for the development of NOEMS-based AI hardware platforms was considered and outlined.
Degree
thesis:*- Name dc:type.qualificationname
- Ph.D.
- Level dc:type.qualificationlevel
- doctoral
- Grantor dc:publisher.institution
- University of Southampton
- Year dc:date.issued
- 2024
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Fernando, James
- Advisors dc:contributor.advisor
-
- Tsuchiya, Yoshishige
- Ou, Bruce (Jun-Yu)
- ORCID person:identifier.orcid
Rights
- Language dc:language
- English, eng