MEMS Design: The Geometry of Silicon Micromachining

Ted J. Hubbard

Ph.D. Thesis, California Institute of Technology
(April 1994).


The design of MEMS (Micro Electro Mechanical Systems) on the millimeter to micron length scales will be examined in this thesis.

A very broad base of knowledge has been developed concerning the etching processes commonly used in MEMS fabrication. The fundamental problem we have sent out to study is how to model the shape transformations that occur in MEMS fabrication. The ultimate goal is to determine the required input mask geometry for a desired output etched shape.

The body of work begins with the crystal structure of silicon and ends with etched shapes. The underlying crystal structure causes different rates for different directions; this behavior has been modeled to obtain rate models. The information in these rate models has then been used in a number of shape modelers. High level models like the Eshape model provide not only simulation but a framework for true design. Other models such as the Cellular Automata model take a different approach and provide flexible and robust simulators. The tools were used to develop real world MEMS applications such as compensation structures.

As important as the individual models, is the ability to integrate them together into a coherent design tool and allow information to flow between different parts. This synthesis allows a fuller understanding of the etching process from start to finish.

It is important to note that while this thesis deals with etching, the methods developed are very general and are applicable to many shape transformation processes.