Richard D. Smith, EE PhD Defense

Wafer-scale fabrication of transducers directly on CMOS dies can minimize device size, increase the throughput and improve uniformity. Doing so with materials sensitive to the conditions encountered during typical microfabrication processes, such as enzymes, remains a challenge.

This thesis investigates aspects of the performance and fabrication processes of electrochemical enzyme-based glucose-sensing transducers, intended for monolithic, implantable wireless sensors. The described work builds on past efforts in the Scherer group and focuses on the transducer fabrication compatible with CMOS wafers.

In the first part, the pre-existing, enzyme-film lift-off patterning process is analyzed. The topography of the film is related to the geometry of the patterns and to the performance of these transducers. Modifications are then made to reduce variation and improve yield. A process to optically profile such structures was also developed to better interpret the non-uniformity from thin film interference.

The second part of this thesis describes the development and processing for plasma etch patterning the enzyme films. This aims to separate the uniformity of the film deposition from the definition of the boundaries, as occurs in many microfabrication processes with less sensitive materials. Strategies were developed that limit the optical, thermal and chemical degradation of the enzyme activity. The resulting process demonstrated the feasibility of dry etch patterning functional enzyme films without loss of activity. Further, it clarified that improved structural uniformity can yield improved performance uniformity.

A final, tangential section investigates a positive tone electron beam lithography process that can be entirely performed in vacuum. Myo-inositol, an electron beam sensitive material, was unexpectedly found and refined. Dry processed negative tone resists avoid pattern collapse during wet development, but analogous positive-tone processes remain elusive. Myo-inositol films, deposited with thermal evaporation, were exposed with electron beams and then developed by subsequent heating. With dry etching and plasma stripping, the full dry process could be implemented in a vacuum cluster tool. While early in development and with challenges remaining, sub 100 nm features were transferred into an underlying thin metal film with this process.