The first MEMS Business Forum, sponsored by MEMS Journal and MEPTEC (MicroElectronics Packaging and Test Engineering Council) was held May 24 at the Santa Clara Biltmore Hotel. Ten speakers presented on topics ranging from near- and mid-term business opportunities to roles of MEMS in broad visions for the future. Registered attendance was just over 90.
Sandhi Bhide, Intel’s Senior Strategist and Futurist, opened the meeting with a discussion of emerging uses and applications for sensors in 2016 and beyond. He opened with a theme from yesterday’s MEPTEC meeting: machines need to become aware of the user’s context and intent, rather than humans conforming to the machine requirements. The fusion of sensors and interpretive software can enable devices to be aware of their connectivity surroundings, environmental changes, and user activity context to modify the presentation and responsiveness of the device. Domains include personal health and safety, social interactions, and relevant business and news events. The notion of proximity detection to help enforce restraining orders against ex-spouses seemed to resonate with this audience. I’m just saying… Extension of this concept to machine-to-machine awareness and interaction melds nicely into the CeNSE (central nervous system for the earth) being promoted by HP. Demands on data processing and storage, bandwidth and energy consumption will be huge, and system reliability will be critical. He anticipates it will take another 8-10 years for the market to reach the trillion sensor level, including both MEMS and other sensor technologies.
Jérémie Bouchaud of IHS iSuppli speculated whether the market for MEMS in consumer electronics will keep growing forever. MEMS microphone revenue jumped 63% in 2011, with background noise suppression driving redundancy and media tablets opening new demand. RF MEMS tuners are starting to show up in mobile handsets in place of analog, facilitating a smaller antenna volume. Based on the current consumer product mix, MEMS annual growth is projected to drop below 10% in 2015 following a +29% peak in 2011. However, there are plenty of ‘next big thing’ candidates that collectively seem to hold the promise of sustaining a healthy growth rate. MEMS for drug delivery and energy scavenging are two leading prospects, with environmental and biosensors in handsets and tablets also very promising. Breath analyzers for a variety of target compounds range from alcohol (excessive consumption) to acetone (hunger and fat burning) to many other biomarkers still in early development. I’ve been hearing more and more that future growth is dependent, at least in part, on “whatever Apple is working on.”
Shahin Farshchi of Lux Capital Management described the role of MEMS in enabling distributed healthcare. VC activity today is weighted toward mobile/social/internet opportunities and healthcare, with lower emphasis on the energy and semiconductors & related equipment segments. The rising prevalence of diabetes and obesity in the US suggests some promising targets. Over 22% of healthcare expenditures are spent on the sickest 1% of the population; 50% of the expenditures care for the 10% most chronically ill. The smart phone platform is gaining broader acceptance as a diagnostic and health data communications tool. Perspiration, respiration, excretion, blood, heart signals and brain waves are relatively accessible data sources with technology that is available or being readied for market today.
Jeffrey Hilbert of wiSpry presented the emerging opportunities for RF MEMS in mobile applications. They manufacture the first MEMS tuner to be implemented in high volume manufacturing in a smart phone (Samsung). Demands for multi-band and multi-mode support, high data rates, power efficiency and decreasing form factor make this an exercise in compromised RF performance. The heart of wiSpry’s approach is an array of MEMS parallel plate capacitor tunable over a wide range using conventional CMOS interconnect materials built atop a 4LM CMOS circuit. The infamous ‘Antennagate’ episode brought to light by the iPhone 4 was resolved by implementing the wiSpry tuner. (The core issue was that the RF antenna tuning and available power changes the instant you touch your phone or put it near your ear. Touchless levitating iPhones were not considered a viable solution.) Tuners alone are a $150M market opportunity. Other applications include tunable antennae, notch filters, power amps and duplex filters. Field demonstrations of complete 1st generation multiband tunable radio front ends can be anticipated in ~6 months.
Prof. Gabriel Rebeiz at UCSD and Director of the DARPA/Industry Center on RF MEMS elucidated the commercialization trends and business opportunities presented by RF MEMS switches and tuners. One shortcoming of RF MEMS is that they require voltage up conversion to 25V or higher for reliable operation; they do not operate at 5V. From 2002 to 2012, front end RF integration has allowed the industry to reduce the RF front end area 13.5x from 5400mm2 to 400mm2 while increasing the number of radios in that space 8x from 2 to 16. For antenna tuning, RF MEMS competitors Cavendish Kinetics and wiSpry compete very well on performance, but SOS/SOI alternatives like Peregrine do much better on manufacturing and device cost. Antenna tuning will be the big market driver through 2018, followed by power amp tuning with filter tuning trailing; the CAGR over the next 7 years will be an impressive +99%. He expects that all smart phones will have tuners by 2014.
Frank Melzer, CEO of Bosch Sensortec shared his perspectives on the interplay between technologies, software and sensor fusion as they conspire to deliver MEMS sensor performance. Barometric pressure devices for geopositional sensing require much greater accuracy than pressure sensors for automotive applications. Ten DoF (degree of freedom) systems have become the benchmark challenge for smart phone integration. The gyroscope component is the limiting factor in terms of size, power management and complexity in 10 DoF systems.
Prof. Al Pisano of UC Berkeley updated the group on the requirements for harsh environment MEMS wireless sensors for energy and power applications. Unlike the miniature, low cost, low power devices needed for smart phones and tablets, these devices may operate at 600°C and cost $100k each. A 1% efficiency gain in a large gas or steam generator can allow it to produce an additional 17 GWH/year of energy, worth almost $2M in additional revenue. In one design, MEMS sensors are bonded to the surface of gas turbine blades, without degrading the necessary aerodynamics, to fine tune the gas flow direction for a more uniform burn and squeeze out an additional percent or two or efficiency. Sensors based on SiC and AlN have been developed for geothermal energy systems that operate at 400°C to 600°C in order to allow in situ subterranean imaging for optimum well identification. This is 4x hotter than oil and gas wells. JFET and bipolar transistors have been fabricated in SiC to operate in this environment. A 20nm layer of graphene is the secret sauce to preserve ohmic contacts and prevent formation of PtSi. The subterranean systems are powered by vibrational energy harvesters based on rapid local pressure fluctuations rather than mechanical vibration.
Harmeet Bhugra of IDT’s MEMS Division explained why people want to buy MEMS alternatives to quartz crystal oscillators. MEMS resonators claim greater reliability and tunability than quartz, with the ability to offer reliable just-in-time delivery akin to related semiconductor technologies. Piezoelectric MEMS oscillators require no DC bias or narrow gaps like capacitive MEMS oscillators, so piezoelectric is the focus for future commercialization. Current aging data shows a frequency stability of ±0.5ppm over 21 months so far.
Prof. Marc Madou of UC Irvine described a sensor technology based on suspended carbon nanowires as a proxy for insight into several new and creative directions into which MEMS fabrication may develop. Structures are formed lithographically using polymer precursors that are pyrolized to glassy carbon. These materials have excellent electrochemical electrode performance, and can be intercalated with materials such as Li (think Li+ battery applications). The suspended nanowires are fabricated with an electrospinning technique that string the fiber between individual posts. In a sense, this addresses the controlled assembly deficiencies associated with carbon nanotubes. Suspension provides 360° access to the sensing element without contamination or undesirable interactions with the surface. In addition to this family of applications, the technique can be applied to fabricate structural colors, which are surfaces that derive their color from microstructural light scattering effects.
Kurt Petersen of KP-MEMS wrapped up the meeting with a retrospective view on MEMS: how did we get here? He published a seminal paper “Silicon as a structural material” 30 years ago, and had been working in MEMS 7 years already before that. The MEMS market today is 3.5% the size of the semiconductor industry, up from 1% in 1986. Market ‘desperation’ can be identified for each incremental jump in the MEMS market dating back to 1985 and HP’s response to dot matrix printer shortcomings with inkjets. The proliferation of MEMS microphones in cell phones was largely driven by the fact that conventional microphones were incompatible with wave soldering, thus requiring that they be hand soldered at assembly. As to the elusive goal of $1T or 1T units, we are presently short by a factor of 100x. As a benchmark, it took the semiconductor industry 35 years to grow 100x. Kurt foresees a 25 year runway for MEMS to achieve this 100x milestone, gated largely by ‘the transformation of the infrastructure of the planet’ associated with CeNSE and the internet of things.