John R. Gilbert and Stephen F. Bart, Microcosm Technologies, Inc.

A new technology is replacing traditional sensors with microfabricated devices. Microelectromechanical systems (MEMS)[1] sensors—sometimes called microsystems technology (MST)[2] sensors—are often less expensive and perform better than traditional devices, and they can more easily be integrated with control electronics. Manufacturers now offer or are developing MEMS devices that detect several chemical types or measure pressure, acceleration, angular rotation, strain, temperature, force, tilt, proximity, electrical and magnetic fields, gas and liquid flow, and viscosity and density of liquids. MEMS-based systems are also being developed to perform DNA, protein, and biochemical analysis.

Using IC microfabrication techniques, manufacturers can create hundreds or thousands of sensor chips on one silicon wafer. The manufacturing technology used to build MEMS sensors also allows additional components (e.g., onchip signal conditioning) to be incorporated with the sensing devices. However, because MEMS are manufactured with techniques used to produce ICs but operate with mechanical devices, there is a real demand for design tools that can connect and integrate the tools of both domains.

Supporting the design of MEMS sensors means providing tools to help the MEMS engineer in three broad areas:

• MEMS sensor design
• System integration
• Application-specific package design

Sensors can usually be thought of as transducers between one physical domain and another; to understand such devices, the engineer must be able to solve coupled physics problems. MEMCAD addresses this need by providing the means to deal with coupled thermomechanics, electromechanics, electrothermal, piezoresistance-thermomechanics, and thermal-electromechanics. Figure 1(A) shows a 3D simulation of a pressure sensor in which MEMCAD has solved a coupled piezoresistance-mechanics problem. Figure 2 shows a microchannel simulation used to design DNA microanalysis systems, which requires a detailed simulation of electrophoretic transport. These tools and others in MEMCAD help the MEMS sensor designer create new, manufacturable MEMS sensors.[4]

Figure 1. (A) In this piezoresistive pressure sensor simulation, the colors indicate pressure-induced stress. Blue represents the smallest stress, and red indicates the largest stress. The cross is the piezoresistive element. This simulation allows the designer to place the piezoresistor at the maximum stress point to obtain maximum output. (B) The curves show the results of a simulation of the piezoresistive pressure sensor shown in (A). The curves show the output voltage of the resistor vs. the location of the resistor relative to the edge of the membrane.

MEMCAD introduces the concept of simulation for devices (SFD). Simply put, SFD means that the simulation tools do not simply solve a physics problem

Figure 2. In the 3D simulation of electrophoretic chemical transport in a microchannel, the solid color indicates the density of a diluted species in the flow. The extruded color visualization (i.e., the broken curves) uses column integration techniques to match the observed experimental results.

under a given set of conditions but are wrapped in a layer of software that focuses on answers to the device designer's questions.

The piezoresistor shown in Figure 1(A) is a good example of SFD software. The underlying physics solved here is 3D coupled piezoresistive-thermomechanics. But a real designer question is where should the piezoresistor be placed on the pressure sensor to get the maximum signal? In MEMCAD, this is answered by generating a graph of the signal in the piezoresistor vs. the location of the resistor on the pressure sensor membrane. Figure 1(B) shows the output voltage response vs. the location of the piezoresistor and gives the designer insight into the placement problem, and it comes automatically out of the SFD layer in MEMCAD.

In most cases, users place sensors in control or monitoring systems to enable or improve the systems. Two examples of systems enabled by sensors are industrial process control systems and automotive air bag subsystems. Designers of control systems often use system-level modeling tools—such as Simulink, SABER, or SPICE—to describe the systems and understand their performance.

High-level system models allow end users (and the sensor designers working for them) to quantitatively understand the sensor specifications required by an application, as well as the tradeoffs between sensor specifications and end-user system design. To make a system model for the sensor user, the manufacturer must be able to make an accurate model of the sensor component. That component—which is a complex system in its own right—may contain micromechanical elements and circuit elements.

The sensor manufacturer must have an accurate model of the micromechanical component to use in its modeling of the sensor, as well as to produce a model of the sensor to be given to its customers. MEMCAD automates the process of extracting models of the micromechanical components of a sensor directly from the manufacturing instructions for the MEMS device.[5] Figure 3(A) shows a sketch of the hierarchy of systems and subsystems, from the MEMS component to the whole sensor and up to the user system into which the sensor must be inserted.

Figure 3. (A) This diagram represents the hierarchy of a MEMS sensor system: the lowest level is the MEMS sensing element (C), which is included in the whole accelerometer chip (B) and (D). The accelerometer system is then included in a user's air bag triggering module. (B) In the mask layout for an Analog Devices ADXL76 accelerometer, the MEMS sensing element is in the center, surrounded by signal conditioning circuitry. (C) The 3D solid model of the MEMS sensing element was automatically created in MEMCAD from the mask layouts. (D) The block diagram of the ADXL76 shows the MEMS component model as well as the surrounding onchip circuitry. This diagram represents the actual circuit schmatic that would be used for system-level circuit simulations. The MEMS component model for these simulations is generated by MEMCAD from 3D simulations of the solid model. (Figures (B) and (D) are courtesy of Analog Devices, Inc.)

The methodology for using MEMCAD couples all layers of design. This is summarized in Figure 3, in which MEMCAD aids the user in going from layout [Figure 3(B)], to full 3D model of the device [Figure 3(C)] and then extracts a model of the MEMS component to fit into the sensor system model [Figure 3(D)]. This leads to accurate modeling of the specifications and behavior of the sensor so that the users of the sensor can have reliable models to place in their user system simulations.

One key difference between the IC industry and the MEMS sensor industry

Figure 4. In this 3D visualization of a thermomechanical package stress simulation, done in the MEMCAD 4.0 coupled package-device simulator, the stress values (indicated by the color) on the top surface of the die (in the center of the package) are extracted from the package simulation and used as boundary conditions for the simulation of the MEMS device (B). (B) This visualization of a thermomechanical stress simulation of an accelerometer device uses the package stresses in (A) as boundary conditions. The color indicates the stress magnitude (blue is the lowest, and red is the highest). The simulation allows the designer to understand the temperature-dependent effects of the package on the device.

is that sensors require far more application-specific packaging. Often critical differences in whole system cost, reliability, and usability of the sensor stem directly from the design of its packaging. By their nature, sensors are more sensitive to stress and temperature variations induced by their packages than are ICs.

MEMCAD introduces the capability to perform coupled analysis of packages and MEMS devices. Figure 4(A) shows a thermomechanical package simulation. Figure 4(B) shows a 3D simulation of a MEMS accelerometer, which displays the temperature-dependent effects of the package on the device. These tools allow designers to quickly and easily build libraries of package and device models and optimize their coupled design.

References
1. K.E. Peterson. December 1985. "Silicon Sensor Technologies," Technical Digest, IEEE Electron Devices Meeting, Washington, DC: 2-7.

2. J.H.J. Fluitman. 1994. "Micro Systems Technology," IEEE Custom Integrated Circuits Conference: 471-478.

3. S.D. Senturia. June 26-29, 1995. "CAD for Microelectromechanical Systems," International Conference on Solid State Sensors and Actuators, Transducers, Stockholm.

4. S.F. Bart. April 15-17, 1996. "CAD for Integrated Surface-Micromachined Sensors: Present and Future," Proc Fifth ACM/SIGDA Physical Design Workshop, Reston VA: 72-75.

5. N.R. Swart, et al. January 25-29, 1998. "AutoMM: Automatic Generation of Dynamic Macromodels for MEMS Devices," Proc Eleventh Annual IEEE MEMS Workshop, Heidelberg, Germany: 178-183.

Home | Contact Us | Advertise © 2009 Questex Media Group, Inc.. All rights reserved. Reproduction in whole or in part is prohibited. Please send any technical comments or questions to our webmaster.