Motivated by exciting early work in the area KEP:82, in 1988 the National Science Foundation sponsored a series of three workshops on Microelectromechanical Systems Research. These workshops resulted in a report entitled: Small Machines, Large Opportunities: A Report on the Emerging Field of Microdynamics NSF:88. This report initiated NSF funded research in the MEMS field. The motivation described in the report remains viable today.
"The miniaturization of electronics has produced a far-reaching technological revolution. Now mechanics is poised on the brink of a similar miniaturization, and its own revolution. Researchers are working toward creating microdynamical systems, the microscale derivatives of conventional large-scale electromechanical systems." [page 1] NSF:88
"The technology of microdynamics is based on that of microelectronics but calls for important advances over it. The goal is to make fully assembled devices and systems that can do what large-scale electromechanical systems cannot do as cheaply, or at all." [page 1] NSF:88
"In recent years these techniques have provided the basis for a viable and growing sensor industry. This industry's greatest commercial success is pressure sensors for automobiles." [page 4] NSF:88
"Current mask design and creation programs were written in response to the fabrication requirements of silicon-based electronic devices and are now highly optimized for the technology of microelectronics. Many of the program features are, at best, useless and, at worst, contrary to the needs of silicon mechanical device fabrication." "Current programs typically [create] rectangular features arranged in Manhattan grids." "... pattern features other than rectangular, for example, curvilinear and freeform, will be necessary for making microdynamical items such as springs, gears, and bearings." [page 15] NSF:88
"Microfabrication technologies, based on batch fabrication, lithography, and selective etching, impose new constraints on the design process. The conventional iterative fabrication [and] sequential refinement ... [is] inappropriate [for MEMS]. The entire design must be completed before fabrication is begun." [page 16] NSF:88
"Thus, simulating designs before they are fabricated, as is done in electronics and in large-scale mechanics, is highly beneficial. A set of computer-aided design tools can reduce the overall cost and/or time between conception and prototype and improve designs for better performance. However, for several reasons, these simulation tools are not now readily used in microdynamical systems design." [page 16] NSF:88
Though there have been many remarkable and revolutionary advances made in the MEMS area since 1988, the need for structured design methods for MEMS remains KEP:94. For example, there is not yet an equivalent to CIF or the other descriptive languages which are commonly used in VLSI design. The MEMS fabrication processes are maturing rapidly, but they are many and varied. The time is now ripe to develop structured design methods, and to take advantage of the still formative nature of the field. The experiences of the VLSI research community of 25 years ago in developing design methodologies and fabrication processes should provide useful guidance.
Several elements appear to contribute to the successes in developing structured design methods for VLSI:
Many have long argued that macroscopic mechanical design has none of these virtues, which largely explains the lack of structured design methods for mechanical design NSF:94. The question remains, however, is MEMS enough like VLSI so that structured design methods can be developed? If so, which part or parts of the developments in VLSI will translate to MEMS?
One approach to developing these new design formalisms and methods is shown in Figure 1 (MEMS Design and Analysis Processes.). The boxes represent physical artifacts, the arrows represent processes. The central straight line indicates: a mask is processed (by fabrication) to create a shape, which when put into operation exhibits a function. The lower arrows represent engineering analysis. For example, a simulation of the fabrication processes can be used to process a mask-layout into a geometric model of the expected shape, which can be analyzed with a finite element method to predict the function of the device. The "backwards" pointing arrows in the upper portion of the figure represent synthetic (rather than analytic) processes, corresponding to creating a shape that will exhibit a desired function, and generating a mask that will create a desired shape.
An alternative view of this general approach is shown in Figure 2 (MEMS Design Flowcharts.). The left-hand flowchart in Figure 2 (MEMS Design Flowcharts.) shows a typical MEMS design process of today. The right-hand flowchart combines a number of new structured approaches, including reusable component libraries, advanced simulation and analysis, and (similar to Figure 1 (MEMS Design and Analysis Processes.) a direct synthesis of layout from schematic.
While the challenges to developing structured design methods for MEMS are significant, the benefits of such methods will greatly enhance MEMS developments. A set of initial research steps appear to be reasonably clear, and are discussed in the following sections of this report.