(Semi-)automated synthesis of at least some classes of MEMS devices appears to be a realistic near-term result of research into structured design methods for MEMS. This work will draw heavily on the prior developments in digital VLSI, but will also clearly depart in several significant ways. However, the need for a widely-used representation format (language); the clear value of libraries of reusable elements, and the need of efficient, high-quality simulations are all elements in common with the prior developments in structured design methods for digital VLSI.
These elements may take the form of a language to facilitate structured design methods (perhaps an HDL-like language), or may consist of libraries of successful (sub-)elements or devices (although the language/representation that will be used to store these prior designs in the library remains a research issue), or may be analytical methodologies that can rapidly transform a desired functional description into a the description of a physical device (or perhaps into a description of the instructions (e.g., mask-layout) and other processing instructions to create the device), or may be methods to efficiently and rapidly explore the highly complex design space. Languages, libraries and simulation will form the basis for creating a "clean separation" between design and fabrication of MEMS. Research into all of these approaches should be pursued.
The goal of all of these approaches is to free the MEMS designer from the necessity of intimate knowledge of the details of each fabrication process. This approach to creating a "clean separation" between design and fabrication will not only greatly enlarge the community of MEMS designers, and decrease the time and number of prototypes required for each new MEMS device, and increase the quality of MEMS designs (by increasing their robustness to uncontrolled variations), but will also free the fabrication process developers to (more) independently improve their processes.