Abstract

This research aims at analyzing the effective mechanical properties of thin film materials that are used in MEMS. Using the effective mechanical properties, reliable simulations of new or slightly altered designs can be performed successfully. The main reason for investigating effective material properties of MEMS devices is that the existing techniques can not provide consistent prediction of the mechanical properties without time-consuming and costly physical prototyping if the device or the fabrication recipe is slightly altered. To achieve this goal, two approaches were investigated: soft computing and analytical. In the soft computing approach, the effective material properties are empirically modeled and estimated based on experimental data and the relationships between the parameters affecting the mechanical properties of devices are discovered. In this approach, 2D-search, Micro Genetic Algorithms, Neural networks, and Radial Basis Functions Networks were explored for the search of the effective material properties of the thin films with the help of a Finite Element Analysis (FEA) and modeling the mechanical behavior such that the effective material properties can be estimated for a new device. In the analytical approach, the physical behavior of the thin films is modeled analytically using standard elastic theories such as Stoney’s formulae. As a case study, bilayer cantilevers of various dimensions were fabricated for extracting the effective Young’s modulus of thin film materials: Aluminum, TetraEthylOrthoSilicate (TEOS)-based SiO2, and Polyimide. In addition, a Matlab® graphical user interface (GUI), STEAM, is developed which interfaces with Ansys®. In STEAM, a fuzzy confidence factor is also developed to validate the reliability of the estimates based on factors such as facility and recipe-dependent variables. The results obtained from both approaches generated comparable effective material properties which are in accord with the experimental measurements. The results show that effective material properties of thin films can be estimated so that reliable MEMS devices can be designed without timely and costly physical prototyping.

Library of Congress Subject Headings

Thin films--Mechanical properties; Thin films--Mechanical properties--Computer simulation; Microelectromechanical systems--Materials

Publication Date

2006

Document Type

Dissertation

Student Type

Graduate

Department, Program, or Center

Microsystems Engineering (KGCOE)

Advisor

Sahin, Ferat

Advisor/Committee Member

Palmer, Harvey

Advisor/Committee Member

Pasupuleti, Ajay

Comments

Note: imported from RIT’s Digital Media Library running on DSpace to RIT Scholar Works. Physical copy available through RIT's The Wallace Library at: TA418.9.T45 P37 2007

Campus

RIT – Main Campus

Share

COinS