by Jérémie Bouchaud
Director and Principal Analyst, MEMS & Sensors, iSuppli
Like many new technologies, RF MEMS and varactors followed a well documented trajectory. The problem as the technology emerged early in 2003 was that the devices could not live up to their initial promise, which led to disillusionment among the companies eagerly awaiting these new devices. Another problem was that the devices were offered by startups, which often burn brightly for only a short time and then disappear. The list of startups that fall into this category is quite long, and includes Magfusion, Siverta, TeraVicta, Simpler Networks, Sagamore, etc., but there are also larger groups that failed to bring the cost down like Panasonic Electric Works, which sampled from the end of 2005 to 2008.
As a result, and seven years later, only three suppliers are in (small) serial production with a total of well under a million per year, and only one of these with RF switches: Asulab with magnetic MEMS switch (although note that these are not RF switches) for hearing aids and high-end watches, MEMSCAP with magnetic MEMS switches for so-called “smart pills”, and Advantest with its captive RF MEMS relays used in Advantest Automated Test Equipment (ATE). But all this is about to change as the technology for RF MEMS switches and varactors has now overcome technology and commercialization hurdles and is now firmly lodged in the enlightened stage. And, in fact, it will soon reach the plateau of productivity (see figure below) via strong offers from credible companies like Omron, Analog Devices, and XCOM with its partner Teledyne.
Also, already a total of six companies are sampling. Targeting high-end applications, (e.g. test, instrumentation and defense), are predominantly U.S. companies Analog Devices, Radant and Teledyne / XCOM as well as Japan’s Omron. And aiming for high volume (mobile phone) applications are WiSpry (U.S.) and TDK-Epcos (Japan).
A market for RF MEMS emerges
Although 2009 saw nearly a million units shipped, mostly for non-RF applications of MEMS switches, iSuppli now expects the market to take off starting in 2011 as the first cell phones start to implement MEMS based antenna tuning and matching networks from the company WiSpry.
At the same time, the market will expand with test and instrumentation applications as more price competitive MEMS relays appear from companies like ADI, Omron, and XCOM (Teledyne will do packaging and marketing for XCOM's products). The target for a switch here is $10 and instrumentation companies like Advantest and Rohde & Schwarz, which patiently awaited RF MEMS, are now much more confident in the offer. The growth continues to climb rapidly (overall GACR 2009 to 2014 = 131%) to reach $223 million by 2014 from these very modest beginnings.
The market segments are as follows in order of opportunity size: (1) tuning and matching circuits used in the front-end modules of mobile handsets will hold the "lion’s share" of the market, (2) test and Instrumentation, including Automated Test Equipment (ATE) for semiconductors, will the be second largest opportunity, (3) a small opportunity for wireless infrastructures, e.g. for base stations and femto cells, which is expected to take off in 2013, and (4) medical applications will take off in 2010 (non-RF), while aerospace and defense applications will develop after 2014.
2011 will be a breakthrough year
Interest among the mobile handset OEMs for tuning and matching of the antenna or the power amplifier has grown significantly since 2008, tied to the growth in the number of frequency bands and advent of new standards -- like LTE for 4G. iSuppli believes that by 2014, more than 50% of mobile handsets will feature some level of front-end module tuning. In support of this, the industry is actively exploring different architectures and their benefits, and a working group “tuning components and architectures” was launched at the International Wireless Industry Consortium in 2009.
The first mobile handset with tunable antenna or antenna matching is expected in H2 2010 and will use either MEMS, BST (ferromagnetic) or silicon-on-sapphire switches. WiSpry has already several design wins under its belt with MEMS variable capacitors and should head into serial production by Q4 of this year. The volume remains modest in 2010 but should rapidly rise in 2011 as the first customers adopt MEMS in more platforms, and TDK-EPC should join the fray.
For test and instrumentation, the high price of MEMS relays -- typically from $30 to $50 -- has limited its implementation until now, but it now seems the target of $10 for high performance RF MEMS relays at this price point should be available in 2011, e.g. from companies like XCOM and Analog Devices.
Other, albeit smaller, untapped opportunities for MEMS switches and varactors include wireless infrastructure and cellular base stations, where today’s switches would be, outperformed on price and performance. The femto cell industry could adopt MEMS rapidly in the 2012 to 2014 time period. Another potential market is defense and aerospace applications, including radio systems, especially after 2014 for phased array antennas, which offer the best volume opportunity with millions of units.
Next technology developments
Some additional hurdles have been cleared in the last 5 years, for example the requirement for a charge pump to increase the voltage of the electrostatic switches to usable levels up to 70V was long considered an issue for cell phones, which operate at 3V. WiSpry has since proved it can integrate this function in the CMOS circuitry of the MEMS switch at a competitive price, while concerns regarding power handling of RF MEMS switches and variable capacitors and the number of switching cycles have evaporated since 2005.
Some major technology breakthroughs are still expected though. First, thin-film packaging has so-far been an unattainable goal of the RF MEMS supply industry for handsets, and no company appears to have made a competitive hermetically sealed thin film cap. WiSpry uses thin-film encapsulation but must add a metal lead on top of the chipset. In addition, probably another 2 to 3 intermediate solutions are needed for thin-film encapsulation to be used solely as the encapsulation. TDK-EPC excluded thin-film packaging for hermetic reasons and went with a more expensive but hermetic wafer-to-wafer bonding with a metal sealing solution. Second, other approaches under investigation include the “in CMOS MEMS” method from the likes of U.S. company Cavendish Kinetics and Baolab in Spain, both of which adopt a revolutionary approach that builds the MEMS between the metal layers of the CMOS structure. This allegedly alleviates additional packaging and presents great advantages in term of cost, although to date little is known about the performance of these switches.
A final word
It is worth noting in this field that MEMS is one technology among many. For mobile phones, MEMS will compete with BST (ferroelectric) varactors that will enter serial production in 2010 for antenna tuning. While MEMS is better in terms of linearity, losses and tuning range, a strong BST manufacturing chain is already in place with the cooperation of STMicroelectronics and Paratek. Meanwhile, silicon-on-sapphire switches exhibit high linearity and although MEMS outperforms on losses, silicon-on-sapphire also benefits from a strong supply chain, e.g. the Peregrine-IBM cooperation. SOI switches are also a credible candidate.
Currently in the test and instrumentation field, Reed relays and EMR dominate the DC-6 GHz frequency band. Silicon-on-sapphire switches, again from Peregrine, are emerging at 7 GHz to 13 GHz, while GaAs PHEMT devices are becoming available for DC-20 GHz applications from companies like Hittite.
In the defense and aerospace realm, SiGe phase shifters will be less expensive than MEMS and GaAs is already used today. In addition, SiGe cost and size are comparable to MEMS. Therefore, iSuppli believes that MEMS will be used on a case-by-case basis for applications requiring high performance and low power.
Jérémie Bouchaud is unique to the MEMS industry, his breadth of MEMS device and application knowledge is unmatched, particularly in terms of automotive, consumer markets and RF MEMS. He was a founder and head of MEMS research for WTC - Wicht Technologie Consulting, acquired by iSuppli in April 2008. At iSuppli, Jérémie is responsible for the MEMS service area. The success of WTC - Wicht Technologie Consulting is due to a large degree to the MEMS research which Jérémie developed. In the course of his career, Jérémie has led over 100 MEMS-related market research endeavors. Prior to WTC, he oversaw technology transfer for sensors and MEMS at the German office of CEA-LETI. Jérémie is a graduate of the Munich University of Applied sciences and of Ecole Supérieur de Commerce of Grenoble. He speaks German, English and French.
Copyright 2010 MEMS Investor Journal, Inc.