The third major transformation is dawning. With the successful proliferation of nanotechnology and architectural engineering, MEMS are the component set of tomorrow. While MEMS commoditization is underway due to a confluence of variables, MEMS+IC designs are proliferating and challenging some tenets and traditional designs and packaging in the semiconductor industry.
Micro- and nanotechnology are not new to the semiconductor industry, nor are the dramatic changes to electronics that these smaller and smaller architectures have offered. Particularly noteworthy among these technologies are Microelectromechanical Systems (MEMS). At a recent session at the Globalpress Electronics Summit in April 2009 (http://esummit09.globalpresspr.com/), Vijay Ullal, group president at Maxim Integrated Products Inc., was quoted by EE Times Asia:
The interesting thing about MEMS is that it is part of an evolution that started 200 years ago with the Industrial Revolution where energy was converted in to [sic] motion. The next major step was computation, the invention of the transistor. The third major revolution with the machine is all about sensors. […] MEMS is not just a new gadget but the third massive transformation of our society.
Figure 1. A spider mite on a polysilicon MEMS gear-train. (Source: Photo courtesy of Sandia National Laboratories, SUMMiTTM Technologies, www.mems.sandia.gov)
What’s in a MEMS?
The domain of MEMS is highly sophisticated, micro- and nanotechnology focusing on microsensors and microactuators. MEMS are rooted in the ‘electromechanical’ domain and their greatest contributions to electronics are the addition of sensing capabilities in the form of accelerometers, gyroscopes, air flow and pressure monitors, among other sensory functions. Having reduced the size of such an array of sensors and actuators to the micro- and nano-levels has opened many markets for MEMS and increased adoption by triple digits over the past few years.
It is this sudden, widescale adoption of MEMS coupled with new manufacturing solutions and barrier reductions that are dramatically changing the role and position of MEMS in the semiconductor industry, the related supply chains, and even the architecture of System on Chip (SoC).
Why MEMS is revolutionary
To explore the validity of Mr. Ullal’s proposition, that MEMS is the next transformative revolution, we need to consider the market reach and the depth of industry impact by MEMS, particularly any technological or architectural impacts.
MEMS originally saw their greatest market share in the automotive sector where the number of ‘insertion points’ has increased steadily to 30-40 for most cars in the U.S. and Europe. Pressure sensors are the biggest revenue generators and the most prolific type of MEMS, with accelerometers and gyroscopes rounding out the top three categories. The downturn in the automotive market has significantly impacted MEMS, with some sectors only set to return to 2007 levels over the next few years. New legislative requirements for enhanced stabilization, tire pressure sensors, and other safety features go into effect in the U.S in 2012 and Europe in 2014. The MEMS automotive sector is set to near US $2 billion by 2014, with some subclasses performing better than others.
Products like the iPhone and Nintendo’s Wii opened the consumer electronics (CE) floodgates for MEMS. While consumers had been using MEMS in their autos and their GPS/navigation devices, it was the popularity of these two CE products that changed what consumers would expect from their electronics and how they would demand to interact with them. Since 2006, games and mobile phones are increasingly offering MEMS based features as differentiators; from motion sensors and autofocus camera lenses to improved microphones and headsets with noise delimiters, and an array of other MEMS devices. The mid- to high-end CE market now expects interactive features in their electronic devices.
As MEMS has proliferated, more companies have entered the supply chain, and the average selling price (ASP) has declined. With price barriers lowered, more industries have been attracted to design new devices based on MEMS. The medical sector in particular is experiencing an increased adoption rate for MEMS. From pain sensors and pacemakers to spinal stability monitors, there is an increase in medical research targeting the application of MEMS. With increased funding through the American Recovery and Reinvestment Act (ARRA) for Healthcare IT, MEMS research and design (R&D) is certain to benefit as a solution path for monitoring patients and for giving feedback and updates directly transmitted from the medical device to the patient’s records.
Industrial, aerospace and defense, among other sectors, are also increasing both R&D and orders for MEMS based devices. This adoption increase comes in the wake of improvements in equipment operation, shock and vibration sensors, enhanced digital imagery, radar and sonar, improved pre-impact response times, among other advanced capabilities leveraging MEMS technology (e.g., night-time ‘vision’, image projection, training simulators, pre-collision/impact safety, new types of sonar, HDD data back up, and gesture or other motion detection).
Figure 2. Market overview and forecast for MEMS by industry subsector, compiled by iSuppli (Source: http://www.memsindustrygroup.org/images/newsletter/June2009/isuppli_figure1.jpg from http://www.memsindustrygroup.org/i4a/pages/index.cfm?pageid=3739)
Figures 2 and 3, compiled by iSuppli for the MEMS Industry Group (MIG), present breakdowns of the market proliferation and growth for MEMS by industry sector and by device type, respectively. These data present clear pictures of the proliferation rates and strong forecasts for MEMS throughout the industry along various trajectories.
Figure 3. Market overview and forecast for MEMS by device type, compiled by iSuppli (Source: http://www.memsindustrygroup.org/images/newsletter/June2009/isuppli_figure2.jpg from http://www.memsindustrygroup.org/i4a/pages/index.cfm?pageid=3739)
Meeting the MEMS demand
During 2008, Taiwan Semiconductor Manufacturing Company (TSMC), United Microelectronics Corporation (UMC), and Freescale Semiconductor all engaged significant ramps in MEMS fabrication both in Taiwan and in the U.S., respectively. As utilization rates for integrated circuits (ICs) were drastically reduced, the move to MEMS production provided a welcomed and important diversification opportunity during extreme economic conditions. Although MEMS also suffered during the recession, demand outside of automotive did not wane as greatly as other component sectors and the rebound has already begun.
Once major foundries entered the MEMS arena, original design manufacturers (ODMs), and industry specialized players began to flesh out the emerging supply chain. Alongside the MEMS sector growth came the breakthrough of the US $1 barrier by accelerometers in 2008 (http://www.isuppli.com/NewsDetail.aspx?ID=20268). With this critical price barrier breached, came an ensuing proliferation of MEMS adoption that continues and is forecasted to see significant, double digit growth through at least 2014. According to iSuppli, the CE and mobile MEMS markets alone are expected to see global revenues of US $2.6 billion by 2012, and the PC MEMS market will rise to US $185.9 million in 2012 (in DigiTimes 2/2/2009, “Popular consumer and mobile products fuel MEMS boom, says iSuppli”).
Figure 4. Global MEMS Market Revenue 2007-2013 (Source: iSuppli 5/28/2009, http://www.isuppli.com/NewsDetail.aspx?ID=20268)
As new players enter the MEMS supply chain, some previous mainstays of the MEMS automotive sector are facing a double-blow. The auto sector slow down coupled with aggressive pricing by new competitors with similarly graded products is proving too great of an obstacle to overcome, and a number are leaving the market. As iSuppli recently reported, Systron Donner Automotive (SDA) was closed down by Schneider; Infineon sold off Sensonor, though remains in the Tire Pressure Monitoring Systems (TPMS) business; Freescale will shut its 6” MEMS fab in Sendai, Japan during 2011, transferring all of MEMS to their 8” Oak Hill, Texas MEMS fab opened in 2008. Also, according to iSuppli sources, various MEMS CE suppliers are presently being vetted by Tier 1 auto suppliers. This is an important indicator of increased competition, hence volatility, in the MEMS supply chain. What was once the exclusive territory of automotive MEMS companies is now a more open market with new, fierce competition and thinner margins.
The commoditization of MEMS reshapes ICs
While the rapid adoption and supply chain changes for MEMS are important in and of themselves, there is a deeper impact that MEMS is having on the semiconductor industry, and it started at the complementary-metal-oxide semiconductor (CMOS) fabs. IC foundries presented the transition to MEMS production as viable based on similarities to CMOS manufacturing that offered opportunities for depreciated lines to be easily converted. These CMOS fabs then began producing MEMS, but with this transition came an interesting side-development: the integration of MEMS and IC chips into single packages. Given that CMOS is also a cross-roads for analog-digital chips, perhaps it is not as surprising that it would be the birthplace of MEMS and IC packaging into single chip solutions, arriving at new SoC configurations.
As MEMS and ICs are increasingly being packaged together, the question of the continued co-existence of the largest IC class, CMOS, alongside of MEMS at fabs has come to the forefront during the past year. This question arose due, in part, to the nearly double-digit growth levels of MEMS over CMOS, particularly noteworthy during the most drastic utilization cuts in fab history. MEMS production has reached as much as 10% of total wafer turnover for a foundry, as cited by SEMI (www.semi.org/en/ctr_026808). Because of the portability of the MEMS manufacturing to existing, depreciating CMOS and through silicon-via (TSV) fabs, “you can have a 40 to 1 volume ratio on your processes over specialty MEMS technologies. That means tremendous changes in the cost structure,” according to Janusz Bryzek, CTO and chairman of LV Sensors, as quoted by SEMI (www.semi.org/en/ctr_026808).
Another barrier to continued CMOS development, hence investment, has arisen for today’s nano-levels. There are new problems in transistor theory that impact progress for CMOS ICs. Namely, the viability of the elastic tunneling model:
Researchers at the national Institute of Standards and Technology in the U.S. have warned of a fundamental flaw in the understanding of transistor noise […]. The theory – or the elastic tunneling model – predicts that as transistors shrink the noise frequency should get higher. […] But [researchers] have now shown that even in nanometer sized transistors, the noise frequency remains the same.(http://www.analog-europe.com/217600929;jsessionid=ZFIHDFWI53QDIQSNDLRSKHSCJUNN2JVN)
While nanotechnology for transistors must now battle this ‘bottleneck’ to move forward with low-power applications, R&D continues to move steadily ahead for MEMS.
Figure 5. MEMS foundry growth projection from 2008 in SEMI.org, original (Source: http://www.semi.org/en/ctr_026808, SEMI’s source: Yole Developpement)
Another driver in the proliferation of MEMS into IC chipsets are cost realizations. As demand for MEMS rises, the de-specialization of the chipsets combined with decreasing ASPs and increasing volume to meet the demand further favors moving MEMS into SoCs. Additional advantages of these new MEMS+IC chipsets are increased power efficiency, greater flexibility, wider application options, and quicker production time by stacking MEMS onto the package. In short, the commoditization of MEMS is underway. A result of this commoditization is the emergence of a new type of multi-functional SoC that may only require a final layer of programmability, or minimal additional components, to make it application specific.
Bearing witness to the wider industry impact of MEMS, as recently as May 2009, Cadence Design Systems Inc. and Coventor Inc., announced a new 3-dimentional (3D) computer-aided design (CAD) system, MEMS+IC, that unites 3D MEMS and CMOS ICs into a single design effort “enabling full co-simulation and co-verification.” (http://www.analog-europe.com/217500743;jsessionid=30IJNA1IASWXGQSNDLPSKH0CJUNN2JVN?pgno=1) One example of a new CMOS + MEMS (CMEMS) technology are MEMS timing devices that are “fabricated atop a customer’s existing CMOS chip designs, eliminating the need for separate timing chips, according to Silicon Clocks.” (http://www.eetimes.com/showArticle.jhtml?articleID=217800320)
Underscoring the importance of the MEMS+IC integration trend, Randolph Fish, director of product marketing for Virtuoso from Cadence, drew a parallel with MEMS+IC to the analog and digital convergence. Mr. Fish further offered, “At first the MEMS element was fairly decoupled from its electronics, but that’s no longer true; the functionality of a MEMS component itself is now dependent upon the [electronic] component.” (ibid, http://www.analog-europe.com/217500743;jsessionid=30IJNA1IASWXGQSNDLPSKH0CJUNN2JVN?pgno=1)
An interesting and strong future
Alongside of the research arena, the major foundries, design houses and OEMs have embraced MEMS as the consumer is demanding more devices and applications that leverage this new interactivity between the person and the electronic equipment. As a result, the market forecasts are strong, even in today’s tough economic climate, showing average forecasts squarely in the double digit range. As presented by DigiTimes 5/25/2009:
According to market research firm Yole Developpement, sales of MEMS equipment will match the 17% average annual growth of the MEMS device market from 2007 through 2012, while the materials market maintains a steady 12% growth rate. By 2012, MEMS makers will be shipping 8.1 billion units worth some US$15.5 billion a year, with nearly half that market to be consumer devices.
There are significant improvements and engineering advances driving new understandings of what System-on-Chip (SoC) will mean as we move forward. The ASICs period may be taking a step closer to exiting as a financially viable component. It is likely that stacking improvements and engineering advances will further the packaging of MEMS and ICs into single SoCs with a final layer of programmability that will allow for customization at a post-production phase of a de-specialized, or commoditized, chipset. Consumers and the semiconductor industry have embraced MEMS. These new MEMS+IC designs are rapidly becoming integral to both the economics of chip fabrication as well as the end-product design that users across all industries have come to expect and now demand as top ranked features.