In order to efficiently design complex microelectromechanical systems (MEMS) having large numbers of multi-domain components, a hierarchically structured design approach that is compatible with standard IC design is needed. A graphical-based schematic, or structural, view is presented as a geometrically intuitive way to represent MEMS as a set of interconnected lumpedparameter elements. An initial library focuses on suspended-MEMS technology from which inertial sensors and other mechanical mechanisms can be designed. The schematic representation has a simulation interface enabling the designer to simulate the design at the component level. Synthesis of MEMS cells for common topologies provides the system designer with rapid, optimized component layout and associated macro-models. A synthesis module is developed for the popular folded-flexure micromechanical resonator topology. The algorithm minimizes a combination of total layout area and voltage applied to the electromechanical actuators. Synthesis results clearly show the design limits of behavioral parameters such as resonant frequency for a fixed process technology.

This paper presents a comprehensive CAD environment for MEMS/MST design. The design flow includes MEMS mechanical schematic entry, parameterized cell layout generation, LVS style error checking, lumped parameter macro-model generation, and system level modeling of the micro-system. The design environment also provides error checking between different design views. The implemented macromodel extraction technique can be used for micro-systems that can be represented as multi-component massspring-dashpot structures. Examples include accelerometers, gyros, and other structures that have rigid masses and compliant springs. A mass-produced capacitive MEMS accelerometer (Analog Devices, Inc.'s ADXL76), is employed as an illustrative example of this procedure. The accuracy of the developed method is also verified by comparing macro-model results for several other MEMS structures to the results of full 3-D physics simulations. Good accuracy is demonstrated in both spatialdomain and frequency-domain dynamic behavior of the models.