Shahrzad (Sherry) Towfighian is Associate Professor of Mechanical Engineering at Binghamton University's Thomas J. Watson School of Engineering and Applied Sciences. She and her team have developed a capacitive MEMS actuator with a linear transfer characteristic that will enable major improvements in the manufacture of microphones and other MEMS devices such as gyroscopes and accelerometers.
Tech Briefs: What got you started on this particular project?
Professor Shahrzad Towfighian: I’ve been studying microelectromechanical systems (MEMS) sensors and actuators since 2007. At Binghamton University I’ve been working on a new mechanism, called levitation, which is different from other capacitive devices. Conventional devices use two electrodes for actuation and this one uses four, so there are more parameters for tuning the characteristics of the system. We have known that nonlinearity has always been a limitation in MEMS devices that operate based on electrostatics and I have been looking for a solution to that for years.
I have been working on two different mechanisms based on electrostatic devices — parallel plate and levitation. In our current work, we combined the two and saw that we could solve the nonlinearity problem that way. Making the system behave linearly, has advantages for many devices, for example, optical filters, accelerometers, gyroscopes, and microphones.
Tech Briefs: How did you get the idea that combining parallel plate and levitation mechanisms would produce linear behavior?
Professor Towfighian: I had been working on the two separately, and then I would, say it was a matter of luck. I wanted to see what would happen if I combined them and we observed that the response was linear. Since we had the theoretical background, we were able to explain it. I had been experimenting with levitation to see if adding that to the parallel plate arrangement — the conventional method of electrostatic actuation —would give us an additional parameter for tuning the characteristics of the system. It was not intended to achieve linearity, we just noticed it was linear when we combined the two mechanisms.
Tech Briefs: I think that luck plays a part in many discoveries.
Professor Towfighian: Yes, it does. We just wanted to know that if we added the levitation, for example to tune for the frequency, how fast these devices would respond. Then we noticed the linear relationship.
Tech Briefs: In the parallel plate design, is the device essentially a variable capacitor?
Professor Towfighian: Exactly, it has two plates, which have two different voltages, one from ground and the other to establish a voltage difference between the two plates.
Tech Briefs: And then you vary the voltage to vary the spacing between the two plates?
Professor Towfighian: Exactly, we change the voltage to change the distance between the plates, which makes an actuator.
Tech Briefs: Could you explain levitation?
Professor Towfighian: You place the parallel plate electrodes in the middle between two other electrodes on the sides — one on the left, one on the right — of this capacitor. If you put high voltages on those two side electrodes, you are adding another variable with which you can tune the properties of the system. Since the force between the parallel plates is always attractive, they collapse onto each other above a certain voltage. Therefore, the relationship between the voltage and the force is nonlinear. This is the problem of using these conventional capacitors for sensing — for sensors, you want to have a linear response. Once you add the two side electrodes, they create a force in the outward direction, away from the substrate so we call that a repulsive force. It is in the opposite direction of the force created by the parallel plate capacitor. We were measuring the amount of force generated by the side electrodes and found that If we played with the voltage on those, we were able to create a linear variation of distance versus voltage. We also observed linear change of the frequency, which means you can adjust the speed of the device. If we have a linear system, that simplifies the design of the controller, which simplifies the whole system, which also reduces the cost. And with a simpler mechanism, because it’s electrostatic, you have low power consumption.
Tech Briefs: Do you use ac or dc voltage?
Professor Towfighian: We use dc with ac superimposed on it. The ac is the frequency you apply to the system. The applied dc voltage changes the resonant frequency of the device — each device has its own resonant frequency. When we use this mechanism, we are able to tune the device linearly. We often set the frequency of the applied ac voltage to the resonant frequency so the device can have the largest possible response. Changing the frequency linearly simplifies the actuator control algorithm.
Tech Briefs: How does this work in a microphone?
Professor Towfighian: A microphone receives an external vibration from sound, so that is the reverse of an actuator. With the parallel plate mechanism, one of the electrodes moves. If we measure the capacitance between the plates, we can determine how much it has moved. Since the motion is related to the sound pressure, this is how the microphone measures the intensity of the sound it’s receiving. It’s like an acoustic sensor — it senses the amount of pressure it receives, and we measure the pressure by seeing how much the plate has moved. With our new mechanism, the relationship between sound pressure and capacitance is linear. This is an advantage for MEMS microphones because for that application, you want to have a sensor such that if you give it 1 volt it moves, for example, 1 micron. Rather than going to high voltage and seeing the motion drop off, or like having a ruler with uneven spacing, you want to have a sensor that changes linearly with displacement.
Tech Briefs: You said that with your microphone, you can work with higher voltage signals to eliminate the problem of background noise.
Professor Towfighian: Yes, in the microphone, we use capacitive sensing — the pressure moves the sensor and the capacitance changes. So, we need an electronic circuit to measure that change. For that circuit, we want to be able to supply a dc voltage that is higher than the background noise. That can’t be done with a traditional parallel plate mechanism, because for example, a parallel plate of the same size will collapse above 10 volts, which poses a limitation on the sensing circuit. The device will collapse because the force between the two plates is always attractive. But with the side electrodes providing a force that pushes up instead of down, we have an additional variable — we can increase the voltage on the side electrodes. That enables us to increase the signal above the noise level, which gives us a higher signal to noise ratio, and gives us high resolution.
Tech Briefs: What is the order of magnitude of the size of these devices?
Professor Towfighian: The cantilever beam itself is about 500 microns long, 10 microns wide, and 2 microns thick — the size of the whole chip is about half a centimeter square. That’s what we made using a standard fabrication method, but it can be much smaller. The switch itself, in the middle, occupies about less than 1 mm square.
Tech Briefs: That’s amazing to me. I’m old fashioned enough that’s it’s very hard to wrap my head around these dimensions.
Professor Towfighian: That’s what I was feeling when I first started this work, so I understand.
Tech Briefs: What do you see for the future?
Professor Towfighian: We demonstrated this concept at a basic level, but it has wide applications, not only for microphones, but also, for example, gyroscopes, accelerometers, and pressure sensors — it can improve the functioning of many MEMS-based devices, so the impact could be huge.
An edited version of this interview appeared in the November Issue of Tech Briefs.