A Computational Infrastructure for Microelectromechanical Systems (MEMS)
N. R. Aluru
Award year: 1999-2000
The breakthrough discovery of silicon as a mechanical material and the advent of micromachining has enabled the design and fabrication of several innovative microelectromechanical devices and systems (MEMS). There is now a critical need for robust, efficient and radically simpler-to-use computer-aided design (CAD) tools to support the revolutionary growth of MEMS. Our research group in the Beckman Institute at UIUC is currently developing advanced computer-aided design tools to support the design and invention of MEMS technology for the next century.
The design of microelectromechanical devices typically involves several interacting energy domains e.g. electrical, mechanical (structural and fluidic), thermal, optical and other energy domains. The goal of device-level modeling is to characterize the behavior of the device by performing a self-consistent analysis of all the interacting energy domains. The present approach is to generate a mesh for each energy domain and then to perform a self-consistent analysis of the mixed-energy domains. This approach is very complicated and cumbersome, especially because of the presence of several interacting energy domains. We are currently developing new device-level modeling capabilities for MEMS by employing meshless methods. In meshless methods only points need to be sprinkled (and no connectivity between points is assumed) instead of the usual brick or tetrahedral elements in mesh-based methods. Meshless methods will enable seamless integration of several interacting energy domains and this radically simplifies the CAD for MEMS. The critical dimension in some microdevices is less than a micron and classicalmathematical model based (also referred to as continuum modeling) approaches are not accurate for device modeling. Instead, atomic scale techniques (such as Monte Carlo and molecular dynamics approaches) will be employed in the critical regions of the device and the entire device will be simulated by a mixed continuum/atomic-scale modeling. Such approaches are critical for accurate characterization of microdevices. Meshless methods are particularly appealing for mixed continuum/atomic-scale modeling as the particles in the atomic scale modeling can be treated as points for continuum modeling. This completely eliminates the need for existing interpolation approaches between meshes for continuum modeling and particles for atomic scale modeling. In this work, we hope to develop meshless methods for continuum modeling, Monte Carlo approaches for atomic scale modeling and multi-scale approaches by combining Meshless and Monte Carlo approaches.