Internet Based Prototyping of Micro-Electro-Mechanical Systems

This paper details preliminary work investigating Internet-based simulation of microsystems. It is in no means exhaustive, and focuses on using mainstream, current web-based technology while highlighting anticipated future directions in relevant internet technology. It proposes a paradigm of automation that limits the information that needs to pass back and forth between a browser-based client and the back-end server. In the present implementation, geometric models are generated from layouts and a process specification, then meshed automatically, and simulated using hp-adaptive finite element techniques. The weband browserbased environment provides file transfer, simulation control, and visualization of simulated results. Introduction Without question, the Internet has revolutionized the world we live in. It has had a profound effect on many aspects of computing, yet has had less impact on computer simulation. Three key technical factors have delayed the success of internet-based simulation for products of engineering interest: 1. Software. Software has been written to run on a specific machine and platform (i.e. operating system) using application programming interfaces (APIs) that require local access to the machine. 2. Bandwidth. Engineering analysis typically requires the use and manipulation of large amounts of data. 3. Hardware. Simulation and visualizing of an engineering analysis often requires powerful computers with memory and graphics capabilities beyond that of the typical office desktop machine. Currently available commercial simulation tools are designed for a “traditional” computing environment. A preliminary system is detailed in this paper that demonstrates an Internet-based prototyping environment for microsystems. The field of MEMS provides an ideal target for Internet-based simulation software for the following four important reasons: 1. Emerging market. Unlike the automotive and aerospace industries which have been around and utilizing simulation tools for decades, the MEMS market is relatively new and use of simulation in the design process has not yet become deeply entrenched using existing simulation paradigms. 2. Cost. The automotive and aerospace industry are driven by a few large companies that can devote significant manpower and financial resources to computational prototyping. In sharp contrast, MEMS is a field driven by smaller firms and universities with more financial constraints. 3. Size of models. A typical model for an aircraft involves millions of unknowns (i.e. degrees of freedom). However, many micromechanical devices of interest can be reasonably simulation with tens-of-thousands of degrees of freedom. 4. Standardized processes. Due to inherent difficulties in IC fabrication, several popular commercial processes (e.g. MUMPS) currently exist. Since fabrication runs take several months, significant time is lost between runs if one is iteratively designing a prototype. This provides opportunities for “virtual fabrication runs” which can be done using internet-based simulation tools in between fabrication of devices. This paper details a MEMS computational prototyping system designed from the ground up to take full advantage of state-of-the-art technology in computer hardware, software, and the Internet. Key design decisions in the software architecture and engineering tradeoffs of computational efficiency for automation are made to enable internet-based prototyping. Client-Server Overview The system takes advantage of a client-server paradigm as shown schematically in Fig. 1. The light gray boxes (masks/layout, processing flow, boundary conditions, and material properties) indicate the user provided information. The left side of the diagram also indicates the steps that are carried out on the designer’s machine (referred to as the “client”). The right side of the figure corresponds to the processes run on the “server.” The server can be the same machine as the client, a different machine on the same local network, or a remote machine accessed via the Internet, even through a proxy. There are three main simulation steps carried out on the server: 1. Geometry. This step takes the user-defined process flow and masks for the layout and creates a solid model representing the three-dimensional device. 2. Discretization. This step automatically breaks the solid model into smaller non-overlapping discrete pieces (called a mesh) needed for simulation. 3. Simulation. This step simulates the device using multi-physics Finite Element Analysis techniques to obtain electrical and mechanical characteristics of the structure.