To get an idea of the current state of the MEMS industry and trends for the future, I interviewed Tim Brosnihan, Executive Director of SEMI’s MEMS & Sensors Industry Group (MSIG).

Tech Briefs: Could we start with a brief seminar of sorts on what MEMS are, how they work, how they’re made, how they’re used. It boggles my mind when I think of the sizes of these things.

Tim Brosnihan: That’s what fascinated me about this technology and drew me to it. I’ve been doing this for about 25 years. When I started, it was very new and kind of miraculous that you could make these tiny little mechanical devices. I worked first with Analog Devices who was a bit of a pioneer in the field — they produced some of the first inertial sensors for airbag deployment. Nowadays, almost every new car has MEMS accelerometers that deploy the airbag when a crash occurs. I’ve worked at other companies, like Qualcomm, on MEMS-based displays, and then at Cirrus Logic on MEMS microphones. There’s a whole variety of these little mechanical structures. It’s amazing that you can have a mechanical device so small you can’t even see it.

Tech Briefs: How can that be made? I can see 3D manufacturing of microscopically small things, but I can’t imagine how you can make it moveable.

Brosnihan: MEMS are made in the same way as every microprocessor or memory chip that’s built on a silicon wafer — the MEMS industry pretty much copied and leveraged those techniques. They’re called subtractive techniques. For example, to make a resistor you put down a layer of resistive material, say some metal, that covers the whole wafer. Light is then used with a photolithography mask that has micron or sub-micron level dimensions. The light exposes the mask so what is left behind protects the film and the pattern you want. You use a dry or wet etch chemistry to remove the material you don’t want, and the IC is then built layer by layer. The MEMS device is built up in the same way. The big difference is there’s a final step where one of the underlying layers is removed. We call that the “release process.” That’s the key magic — you take away an underlying layer, which leaves a suspended structure that can move. It can move in plane, out of plane, or twist. It all depends upon what type of mechanical motion you need. There’s a whole variety of sensors and actuators that utilize all kinds of mechanical motion.

Tech Briefs: Could you tell me how the MEMS devices connect to the electronics.

Figure 2. VM1001 piezoelectric MEMS microphone from Vesper Technologies, Inc. (Boston, MA) provides IP57 rating — no external mesh is required for water and dust protection. (Credit: Vesper Technologies)

Brosnihan: MEMS and electronics can coexist on the same wafer because the mechanical element is fabricated using the same techniques as an IC. You can run a metal trace just like you would to connect one transistor to another using a metal line in an IC. We use those same metal lines to connect the on-chip electronics to the mechanical elements. That’s called an integrated process.

What’s actually more common though is two chips: a stand-alone MEMS chip and a stand-alone circuit chip that are then assembled together, either side-by-side, wire-bonded together, or you can flip chips one onto the other and make vertical connections through vias.

Tech Briefs: I can see the MEMS conveying information to the electronics. Do the electronics ever activate the MEMS?

Brosnihan: Absolutely, the way I like to think about it is that there are two categories of MEMS: sensors and actuators — they’re two sides of the same coin.

Figure 3. Using a real-time vibration analysis sensor to monitor automotive production machine. (Credit: Shutterstock/Zapp2Photo)

A MEMS microphone is an example of a sensor. It has a moveable diaphragm that moves in response to sound pressures created by your voice. Since that moving membrane functions as one plate of a capacitor, it creates an electrical signal.

A speaker is an example of an actuator. There, instead of a pressure input, you have an electrical input — a signal representing a sound wave moves the speaker diaphragm, creating sound pressure.

Figure 4. Wrist-based heartbeat monitor. (Credit: Shutterstock/Andrey_Popov)

Tech Briefs: How do you implement an electrical input for an actuator?

Brosnihan: It can be done with a magnetic coil. Or you can have an electrostatic actuator, with two parallel plates separated by an air gap. When you apply a voltage to one side, it attracts the other side. That’s probably one of the most common techniques.

Then there are thermal actuators, where a resistive element gets heated up and expands. It can bend or move depending on the heat gradient through the film. That’s the one used in inkjet printing.

There’s also piezoelectric film. If you have a voltage you apply to a piezoelectric film, it will expand or bend and that can work as a actuator.

Tech Briefs: What do you see as some emerging growth areas for MEMS applications?

Brosnihan: Although it’s been growing for some time, there’s no sign of it slowing down: the big market for MEMS sensors is consumer electronics. The first cell phone had one microphone — my iPhone 11 has 16 sensors. For example, when you put your phone up to your head to speak, a proximity sensor shuts the screen off to save power. Some of the more advanced phones have up to four microphones. There are gyroscopes to measure the rotation and jitter of your phone because people are taking pictures and need dynamic stabilization. Every year they’re putting more sensors into the phones.

The other consumer electronics are following suit, especially with microphones — now everybody wants to talk to their devices — they want to tell the TV what they want to watch. In cars, people have to do hands-free driving now, so voice activation is a growing market there.

Another growth area is in machine learning, which needs large amounts of data in order to function. Large amounts of data require a large number of sensors, for example, to measure the condition of factory machinery.

Tech Briefs: How about powering all those MEMS sensors and actuators?

Brosnihan: That is certainly one of the key features. Being small, and having electronics integrated with them, they use lower power than with traditional sensors — a coin cell battery can power some sensors for years. A temperature sensor in a smart building, can typically last for three years. We try and keep the power as low as possible. Even for things you want to recharge, like your phone. With 16 sensors in a phone, nobody wants it to die partway through the day. So, there’s continuous pressure to reduce power. And it’s not just so you don’t have to charge your phone many times a day, but so all of those sensors can remain on all of the time. It used to be your microphone was only on when you were making a call; now it’s always on so when you say “Hey Siri” the phone wakes up (as mine just did). Things are listening all the time and that takes power, so the demand to have a low-power microphone, for example, is tremendous. If you have to listen all the time, you can’t burn a lot of power to do it.

Before MEMS came along, microphones were analog electret microphones. They had to house a separate analog to digital converter (ADC) to create a digital stream that you could process with your audio processing software. Now, a large percent of the microphones on the market have already moved to digital, where it’s still an analog transducer, but there is an ADC in the sensor package. So, when we put the microphone device into your phone, it outputs the digital bitstream that you want your codec to process. And they’re working to embed even more smarts right into the microphone — like being able to detect a trigger word to wake it up.

Tech Briefs: How about some other areas?

Brosnihan: We’re seeing a lot of health care monitoring. An elderly person can wear a bracelet or belt with sensors that will alert someone if they have a fall, or maybe their heartbeat or breathing becomes irregular.

There are also some that are just fun — shoes are being instrumented with sensors, and they can tell you about your running style or about your basketball play and feed back information on your performance so you can get the most out of your workout. For example, Fitbit was the first with that massive option: wear this thing on your wrist and it’ll tell you how much exercise you’re getting. Things that used to be daily activity are now exercise because Fitbit deems them so. That caught on and people want to know: everything I’m doing, is it healthy, am I making progress, am I doing better than I did yesterday?

A third area, although it hasn’t been growing as fast as we thought it would, is electrified and self-driving vehicles. There are a huge number of sensors in electric and autonomous automobiles.

Autonomous cars need to sense the environment to make sure that a car can drive down the street without hitting an obstacle or a pedestrian. Less critical, but also important, is to park. We think that market is absolutely happening, but it’s slower to be adopted, partly because it’s government regulated. But we know that’ll be an explosion in sensor demand.

This article was written by Ed Brown, Editor of Sensor Technology. For more information, visit here  or here .