Page 50 - MEMS Mechanical Sensors

CHAPTER 3
MEMS Simulation and Design Tools

3.1 Introduction

Simulation of micromachined systems and sensors is becoming increasingly impor-
tant. The motivation here is similar to that of the simulation of purely electronic
VLSI circuits: before fabricating a prototype, one wishes to virtually build the device
and predict its behavior. This allows for the optimization of the various design
parameters according to the specifications. As it is a virtual device, parameters can
be changed much more quickly than actually fabricating a prototype, then redesign-
ing and fabricating it again. This considerably reduces the time to market and also
the cost to develop a commercial device. Simulation software tools for electronic
circuits are very mature nowadays, and the level of realism is striking. Often the first
fabricated prototype of a novel circuit works in a very similar way as predicted by
the simulation. In MEMS, however, this degree of realism cannot be achieved in
many cases for two reasons. First, the simulation tools have not reached a similar
maturity as their electronic equivalents; and second, and more importantly, simula-
tion of MEMS devices is much more complex. A MEMS device typically comprises
many physical domains such as mechanical, electrical, thermal, and optical. All
these domains interact and influence each other, making the problem orders of mag-
nitude more difficult.
Any MEMS simulation software uses either of two approaches:

• System level (or behavioral or reduced order or lumped parameter) modeling:
This approach captures the main characteristics of a MEMS device. It pro-
vides a quick and easy method to predict the main behavior of a MEMS
device. The requirement is that the device can be described by sets of ordinary
differential equations and nonlinear functions at a block diagram level. This
approach originated from control system engineering. The multidomain prob-
lem is avoided since, typically, the simulation tools are physically dimension-
less—only the user interprets the input and output of the various blocks in a
physically meaningful way.
• Finite element modeling (FEM): This approach originated from mechanical
engineering where it was used to predict mechanical responses to a load, such
as forces and moments, applied to a part. The part to be simulated is broken
down into small, discrete elements—a process called meshing. Each element
has a number of nodes and its corners at which it interacts with neighboring
elements. The analysis can be extended to nonmechanical loads, for example,
temperature. Additionally, finite element simulation techniques have been

39