MEMS Design Optimization with FEA

One of the most promising new areas of engineering is Micro Electro Mechanical Systems (MEMS). At the size of a grain of salt or the eye of a needle, these micromachines integrate mechanical elements, sensors, actuators and electronics on a common silicon substrate. Already in use in optical switches within telecommunication and networking systems, accelerometers in automotive airbags, inkjets in desktop printers and sensors in medical testing equipment, engineers are continuing to find new applications for MEMS technology in an ever-growing range of industries.

MEMS technology has the potential to make the next generation of electronic products smarter and cheaper. Using the same technologies developed for computer chips, MEMS can be very economical to produce, especially in large quantities. Due to their small size, they require very little material in manufacturing and tiny amounts of energy for power. Since MEMS often integrates computer chips and sensors, they can provide “smart” control over other mechanisms. One example of a MEMS device that offers smart control is the automotive gas sensor that makes engines burn more efficiently by sampling and adjusting combustion gasses.

The task of designing and optimizing MEMS devices also brings with it unique challenges, including analyzing the many interdependent physical phenomena to which MEMS devices are sensitive and working with small-scale geometry. “MEMS simulation demands multidisciplinary, multiphysics software and good CAD support,” said Walied Moussa, Ph.D., Assistant Professor at the University of Alberta, in Edmonton, Alberta, Canada. Dr. Moussa is currently researching and teaching on the subject of MEMS and FEA. “ALGOR has these capabilities and is therefore a very important player in the MEMS simulation industry. In addition, the software has been easy for my students to learn. The software is well-documented and Internet-based instructional Webcasts aid with the students’ learning curve.”

Advancing Automotive MEMS Research

In one recent experiment, Dr. Moussa collaborated with ALGOR to model an experimental automotive gas sensor. “ALGOR software was selected for this project because it had the multiphysics and CAD support capabilities that were necessary,” said Dr. Moussa. Each sensor functions at a unique temperature, which is controlled by a micro-heater. Simulating the stresses that result from thermal expansion for the MEMS device is critical to ensuring its performance. 

This photograph shows the top surface of a MEMS micro-heater for an automotive gas sensor. Automotive gas sensor arrays make engines burn more efficiently by sampling and adjusting combustion gasses.

First, the micro-heater was modeled in Pro/ENGINEER and the geometry was captured using ALGOR’s InCAD technology. “Whereas many other MEMS simulation solutions are tied to a particular design tool, ALGOR offers the flexibility of built-in modeling capabilities as well as CAD support for a number of popular CAD solid modelers,” commented Dr. Moussa. 

This exploded view of the micro-heater in Pro/ENGINEER shows the many layers of the device.

After automatically generating a finite element mesh, FEA and material properties were defined. The micro-heater consists of layers of Silicon Dioxide, Silicon Nitride and Platinum/Titanium alloy on a Silicon substrate. For all of these materials, published material properties were available and input into a user-defined material library for use by the software. Although not applicable in the case of the micro-heater, MEMS often employ piezoelectric, composite and nonlinear materials in addition to common linear materials. Many MEMS devices use materials whose properties are not available in the public domain. “MEMS simulation software often requires a wide choice of material capabilities, wizards and data management tools that enable engineers to easily define and store experimental material data,” said Dr. Moussa.

The convection, heat generation and radiation loads that the micro-heater experiences were then applied to the model. The heat generation load represented the effects from Joule heating from the power source. The convection and radiation loads simulated the loss of heat to 25°C air. A steady-state heat transfer analysis was performed on the micro-heater and the resulting maximum temperature agreed with experimental results published in Sensors and Actuators by Yaowu Mo, et. al. 

Temperature results from a steady-state heat transfer analysis of the MEMS micro-heater (left) agreed with a published experimental temperature profile from a micro-area pyrometer (right).

A linear static stress analysis was then performed using the heat transfer analysis results in order to calculate the thermal expansion and resulting stresses. Constraints were added to fix the bottom surface of the micro-heater in the vertical direction. The only loads applicable were the thermal effects, so no additional structural loads were necessary. 

The use of heat transfer and static stress analyses to determine thermal stress for the micro-heater is an example of multiphysics analysis. Multiphysics capabilities are important for engineers designing MEMS because the effect of multiphysics is often more significant at the micro-scale than at the macro-scale. For example, a typical MEMS device called a comb drive is entirely driven by electrostatic forces, a phenomena that engineers rarely have to consider in large-scale parts. In addition to electrostatic effects, stress, motion, heat, fluid flow and linear dynamics can play an important part in designing and optimizing MEMS devices. Just as important as using each of these analysis capabilities separately is the capability to consider the interaction of different phenomena, or the effects of one phenomena on another, as was the case with the micro-heater model. 

“At the macro-scale, engineers can neglect possible environmental forces such as static in the air,” explained Dr. Moussa. “This is because the driving forces for macro-scale devices are usually many orders of magnitude higher than environmental forces, which are therefore negligible. That is not the case with MEMS. Small environmental forces can be a help or hindrance, but they must always be considered.”

The stress analysis revealed high stresses on the top surface near the center where the micro-heater attaches to the sensor. “The agreement between the heat transfer analysis results and the experimental profile gave me confidence in the model and the software,” said Dr. Moussa. “The stress results were useful because such results would be difficult to obtain experimentally. Furthermore, the stress analysis results indicate the need to optimize the sensor geometry to reduce the thermal stresses. This is an important benefit of FEA because it means that engineers can optimize designs like this one without creating expensive prototypes.”

ALGOR’s linear static stress analysis results based on thermal loading reveals high stresses on the top surface near the center where the micro-heater attaches to the sensor. This indicates the need to optimize the sensor geometry to reduce the thermal stresses.

The optimization of the micro-heater design is currently underway to reduce the thermal stress. In addition, packaging will need to be designed to contain the micro-heater and protect it from environmental forces. MEMS packaging often includes heat sinks, fans, insulation from static and covers that keep out dust. How well the packaging protects the MEMS device from its environment is another area to study through multiphysics analyses. 

The automotive gas sensor micro-heater is just one of many MEMS devices under development today. As illustrated in this example, the use of multiphysics analysis software with CAD support is an effective means for facilitating the design optimization of MEMS devices.