What’s New On The MEMS Horizon?

While the more well-known MEMS such as accelerometers, gyroscopes, microphones and pressure sensors continue to grow in volume as they find new applications in things like wireless earbuds, drones and virtual reality headsets, for example, there’s an emerging class of MEMS devices that are destined to, dare I say it, “improve our lives.” These include fingerprint sensors (FPS), MEMS-based speakers, retinal scanners, MEMS-based LIDAR (Light Detection and Ranging), gesture recognition, energy harvesters, force feedback actuators/sensors (essentially piezo-MEMS for haptic feedback on input devices such as smartphones and tablets), gas/particle sensors, and the list goes on and on.

Some of these new devices are in the early proof-of-concept stages, but many others are well into development and are already being demoed with perspective system-level OEMs. Let’s take a look at a couple of the latter stage technologies and see how these new classes of MEMS devices might change our lives (yet again).

Fingerprint sensors
These are used today in a number of security-centric applications. According to Yole Developpement, the market value for these devices was $2.9 billion in 2016 with roughly 800M units shipped. Yole is forecasting the number of FPS units to increase to over 2 billion units with a market value approaching $3.7 billion by 2022. There are currently four different fundamental operating technologies for FPS: optical, capacitive, thermal and ultrasonic. The most prolific device technologies used today are capacitive based. You’re probably most familiar with the capacitive FPS device in your smartphone or laptop computer.

While each technology has differing pixel count/densities and overall levels of effectiveness, it’s the sensing area that ultimately determines their utility and volume adoption in the consumer market. In this regard, capacitive and ultrasonic technologies beat out the competition with somewhat conservative sensing areas of ~28mm² and ~36mm², respectively. Combined with pixel counts north of 10k, pixel densities >500ppi and extremely low power consumption in stand-by mode, it’s easy to see why these two technologies are prime candidates for use in mobile products. And, while capacitive devices currently enjoy broad adoption, the piezo-based ultrasonic devices are just now entering the market. With increased sensitivity capable of imaging both epidermal and dermal layers of the finger, piezo-based devices offer greater security and, as reported in a previous post, are impervious to the effects of dust and moisture on the scanning surface.

Today, piezo-based FPS chips can be fabricated using either aluminum nitride (AlN), scandium-doped aluminum nitride (ScAlN) with scandium concentrations in the range of 20-30%, or lead zirconium titanate (PZT) piezo-materials, each with an increased electromechanical coupling coefficient (kt²). While the CMOS compatibility of the AlN family of films is preferred, there are device architectures that might enable a PZT-based device by taking advantage of CMOS/sensor integration based on wafer bonding. And, although various deposition technologies can be utilized depending on the film parameters of interest, FPS chips will likely rely on physical vapor deposition (PVD) as a high-productivity manufacturing approach. So, that only leaves one big question: is this a 200mm or 300mm play? Well, for now it seems 200mm (and small gen panel) is the way forward, however this pundit will not rule out a 300mm play in the future for this high-volume, increasingly ubiquitous device.

MEMS-based LIDAR
LIDAR is a technique whereby laser light is used to illuminate a specific target and measure its distance from the source. Encased in a spinning “can” mounted on the roof of a car, it’s how many autonomous vehicles “see” their surroundings today. While effective, it’s hard to imagine driving around with one atop a sleek new autonomous vehicle. Nor are they cheap; LIDAR prices range between $10k – $30k per unit, but can climb up to $75k for the most advanced models.

Addressing this cost issue, several companies are touting discrete, MEMS-based LIDAR solutions at price points targeting sub-$100 (and possibly a few tens of dollars). With die sizes on the order of 25mm², these devices are small, 2D scanning mirror architectures operated by a series of electromechanical comb drive structures. There are several R&D efforts underway to develop new and interesting technologies in this space. The Berkeley Sensor & Actuator Center (BSAC) has its own approach for MEMS-based LIDAR.

Using an Optical Phased Array (OPA), a die-based mirror composed of many smaller reflective micromechanical elements is able to carry out sophisticated beam forming, steering and tracking of multiple objects. Each mirror element in the OPA is ~2µm wide and 35µm long. The full 1D mirror device is capable of scan angles >22° at scanning speeds in excess of 500kHz with operating voltages <10V. While similar to the MEMS-based LIDAR devices I mentioned previously that have mirror elements manipulated by electromechanical comb structures, these comb structures can be as thin as 300nm with equivalent 300nm spacing. So, while metal deposition will be important in forming high-quality mirror surfaces for these devices, it seems that deep reactive ion etch (DRIE) will be the key enabling process – especially if device designs call on reliable sub-micron process control.

While the future of LIDAR looks bright (ok, pun intended), there is a lot of competition between component OEMs looking to enable this exciting segment of the automotive industry.

I’ll provide more exciting updates from the MEMS space in 2017 as Applied continues to tackle the tough problems and support its customers in this growing market.