The MEMS market is exploding as smart phones, tablets, games and other mobile devices swallow billions of components. Motion processing and location sensing technologies are central to the functionality of today’s handheld products that most of us find indispensable. Indeed, the merging of sensing with computing power and communications is transforming the MEMS market. But, if it is to meet consumer demand for the fastest growing products, the MEMS industry will need to see the design ecosystem supporting it evolve.
For starters, pure-play foundries will need to offer standardized MEMS processes and collaborate with fabless MEMS makers to meet fast time-to-market and high-volume demands. Next, the design flow will need to combine MEMS and traditional IC technology, bringing together MEMS and analog/mixed-signal design in a single toolset and design flow. To complete a new MEMS design ecosystem, designers will also need process design kits, MEMS IP libraries, and reference flows from foundries. In short, the changing MEMS market calls for nothing short of a new silicon and software ecosystem – much of it based on the fabless model familiar to IC design companies – to enable designers to turn out products for the cost-driven, high-volume consumer market.
The MEMS market will continue to see double-digit growth for the next six years with 20% compound average annual growth rate in units and 13% growth in revenues to become a $21 billion market by 2017, reports Laurent Robin, Activity Leader, Inertial MEMS Devices and Technologies at Yole Development (Lyon, France). MEMS market analyst Yole expects motion sensing and microfluidics to make up almost half the overall market by 2017, with accelerometers, gyroscopes, magnetometers and combos making up about 25% market share and microfluidics 23%. Yole predicts the growth for inertial combo solutions to be huge.
The promise of the MEMS market goes far beyond today’s growth as impressive as that is. The Hewlett-Packard project Central Nervous System of the Earth, appropriately dubbed CeNSE, will eventually include a trillion tiny sensors installed globally to collect data about the world around us. Ten years from now, reports Peter Hartwell, nanotechnology researcher at HP, the company will have a single sensor the size of a pushpin to measure sound, vibration and temperature. Built into this device will be a processor, storage, radio, battery and energy scavenging technology. To sense the vibration of a bridge structure, HP already has a new MEMS accelerometer that’s 1,000 times more sensitive than devices used in airbags and game controllers. HP will first deploy CeNSE to help Shell drill for oil.
Up until now, IDMs have dominated the MEMS industry. Their huge investments in proprietary processes plus long R&D and production cycles have led to the “one process, one product” adage descriptive of the MEMS industry. To move beyond these dedicated MEMS fabrication lines toward standardized MEMS processes, the industry can leverage excess available capacity in many 8 inch wafer fabs. They can also put to use equipment still needing to be amortized. Pure-play foundries, MEMS makers and design tool suppliers can work together to attain today’s market goals. Such collaboration has paid off for fabless InvenSense which has experienced a 67% growth rate to $144 million after working in a familiar fabless model to bring to market its gyroscope and motion sensing MEMS.
Design and integration innovation will, of course, be central to fulfill the promise of the MEMS market. As MEMS makers increasingly integrate multiple sensors like the combo accelerometers/magnetometers and accelerometers/gyros that shipped in volume last year, they’ll require new design software and methodologies. These software tools will simplify the integration of MEMS into larger systems and modules. Combining the two very different disciplines of multi-physics, MEMS-specific design with anaog/mixed-signal IC development will go a long way to accelerate design cycles. Although MEMS are almost always on the same substrate or in the same package as the electronics, the two components have been developed separately. The cost and lagging time-to-market of this ad-hoc approach is no longer viable.
Models that encompass the complex physical behavior of MEMS-based inertial devices, resonators, microphones, optical MEMS devices, etc. will replace the Spice and/or FEA-based models that MEMS specialists have hand-crafted in a build-and-test design approach. Non-MEMS specialists will be able to use these models without a steep learning curve, considerably speeding the design cycle. These models will simulate MEMS plus electronics, thus avoiding errors likely to occur with manual handoff of a design from MEMS to IC designers.
Integrating the design flow for MEMS designers and their analog/mixed-signal IC colleagues all the way through to tape-out is a major strategy of Coventor. At each stage of its design flow—through simulation, functional verification, layout, physical verification and parasitic extraction—MEMS and electronic designers can work with tools from Coventor and its partners in the electronic design worlds – such as Cadence and The MathWorks. At every level of design, simulation compresses the design cycle. What needs to fall into place now is for pure-play foundries to offer process design kits, MEMS IP libraries and reference flows for developers to use throughout the design flow.
In the meantime, we can look forward to HP’s CeNSE to “get at the heartbeat of the earth.” CeNSE looks at the climate, shows how things are moving, where people are going and what they’re doing. Its brainpower can be used to do something as simple as turn off the lights in a room where nobody is, reveals HP’s Hartwell. We can also look forward to emerging MEMS-based personal medical devices that have already begun to be integrated or plugged into mobile devices. Wearable health monitors are also on the horizon. An ecosystem that unites the design of MEMS and ICs will accelerate time to market for these and other innovative MEMS products.