The panel ended up more or less back at the one fact that has dominated MEMS and has both enabled it and hindered it. The first law of MEMS is: "One product, one process, one package." But the panel also highlighted the potential risk of MEMS oversupply either from over-exuberant established manufacturers or possibly from Chinese entrants into the field.
The panel pointed out that the 3Ps rule of MEMS is a double-edged sword; it favors customization and fragmentation rather than standardization and scaling but it has created a rich garden of exotic sensors and actuators based on a wide variety of transduction principles.
However, the panel discussion did raise the possibility of an incipient equivalent to Moore's Law based around packaging and how the awareness amongst the members of industry bodies such as MEMS & Sensors Industry Group, organizer of the conference, could help drive standardization.
The session was moderated by Peter Merz, MEMS business unit manager at X-Fab Silicon Foundries AG and the panel comprised Dave Monk, general manager of the motion sensors business unit at NXP Semiconductor, Markus Sonnemann, vice president of engineering at Robert Bosch GmbH and Yannick Pilloux, business development manager for MEMS at manufacturing equipment supplier PlasmaTherm LLC.
Their task was to enable another decade of MEMS industry growth, if they could.
The Internet of Things (IoT) would certainly appear to be boon to MEMS vendors with the enormous volumes of simple machines that are expected to be connected to the Internet in 2025, 50 billion devices according to Cisco.
Projections showed by Merz put the demand at 12 MEMS per person across the globe in 2025 or 105 MEMS per "industrialized" person with the majority shared between the automobile, the public environment, in the home and in wearable/mobile equipment.
But to achieve such volumes at average selling prices (ASPs) that are affordable to the public AND motivating to industry players appears difficult. Not least, the panel agreed, because the MEMS sector does not scale the same way that CMOS and digital semiconductors have done.
Next: A Moore's Law of MEMS?