Can RF MEMS Master The Mass-Market Challenge?

April 12, 2007
Reliable, high-quality RF devices can be produced at reasonable prices. But questions linger on when costs will drop enough for them to launch into the consumer world.

After many fits and starts, RF MEMS technology may finally create some waves within the electronics market this year, thanks to recent advances in manufacturing and improved device reliability. Whether or not this year's new RF MEMS products can pass the litmus test for modest entry into mass-market applications, though, remains to be seen. Nevertheless, expect RF MEMS to gradually penetrate such applications. As for getting below the magical $1 and $2 price barriers needed to gain widespread adoption in mobile phones? Well, that will have to wait a few more years (see "Markets And Expectations").

Opinions are tempered by the history of RF MEMS IC developments over the last three decades, particularly from 2000 to 2003, that failed to meet market expectations. The key will be in honing the necessary manufacturing process steps to produce very reliable devices in very large numbers and bring down the cost to low enough levels that suit large consumer applications. RF MEMS holds some key advantages over other approaches. Dan Hyman, president and CTO of XCom Wireless, cites its low loss, high isolation, and near-perfect linearity. Also, conventional mechanical and semiconductor technologies simply can't compete with its unbelievably large instantaneous bandwidth.

Its advantages are well-known: cost, size, speed, ruggedness, reliability, repeatability, lifetime. Components such as tunable capacitors can displace traditional reed and electromechanical relays, as well as solid-state varactor-based devices, for greatly improved agility and performance.

Now, many companies can successfully make RF MEMS components that lay the foundation for RF circuit application. These include switches from Matsushita Electric, Radant MEMS Inc., and TeraVicta Technologies Inc.; inductors from Enpirion and Walsin Technology Corp.; filters from Avago Technologies, Epcos, Fujitsu, Infineon, and NXP; and tunable capacitors from NXP and WiSpry.

Even manufacturers of MEMS resonators for timing circuits (e.g., SiTime, Discera, and Silicon Clocks) can easily use such components in RF MEMS for mobile phones—if the market pull is there. For example, TeraVicta and Radant MEMS recently succeeded in using such circuits (switches) for commercial and military applications.

IMEC is developing a generic technology platform to implement most of its RF MEMS devices (Fig. 1). The platform employs thin-film surface micromachining techniques at the wafer level. The company believes that the future of RF MEMS technology lies in the ability to integrate RF MEMS devices with other passive and active components, preferably in a hybrid manner. This would result in a system-in-a-package (SiP) approach. It also believes such a platform is one means of achieving successful RF MEMS products.

The most common applications for RF MEMS ICs are mobile-phone switches. Today's RF MEMS switches are typically used by automatic-test-equipment (ATE) manufacturers in the commercial sector (supplied mostly by TeraVicta) and in radar and other military applications (supplied mostly by Radant MEMS).

The tough technical challenges presented by large-scale production of low-cost RF MEMS ICs have been greatly underestimated. There's a big difference between developing a few samples and prototypes and consistently mass-producing products with high levels of reliability and quality. Consequently, many market expectations have been dashed, leading to skepticism about the technology's usefulness (see "Demystifying RF MEMS,").

According to industry estimates by RF MEMS device manufacturers like WiSpry Inc., over half of the world's shipments of mobile phones will contain at least one duplexer circuit, and most of them will be multiband, multimode devices. Combining the RF switches, filters, and duplexers needed to deliver this level of flexibility— and still meeting stringent cost and performance targets—represents one of the biggest obstacles facing RF engineers.

Issues include reliability, temperature drift, lifetimes, dielectric breakdown and leakage, and stiction. Packaging is another major hurdle, since RF MEMS ICs need to stand free in a cavity. They also must be well protected from harmful contaminants during the manufacturing process. "The two biggest issues are stiction and contact contamination," says Ray Burgess, CEO of TeraVicta. "We've now successfully solved them. Cost is no longer an issue."

Burgess also believes that RF MEMS switch technology is ready to compete with electromechanical relays and reed relays in terms of cost, reliability, performance, and size. Ter-aVicta offers three lines of RF MEMS switches that consist of baseline, higher-cost high-bandwidth, and low-cost devices that will be available in volume.

The baseline dc to 7-GHz TT712-CSP line of single-pole double-throw (SPDT) switches, which have been shipping in volume since early last year, will be joined by derivative double-pole double-throw (DPDT), single-pole four-throw (SP4T), and custom switches (XPXT) during the first half of this year. Aimed at instrumentation, ATE, and radio communications applications, the switches go for $10 to $20 each, depending on volumes and configurations.

Also emerging in the first half of 2007 will be dc to 26.5GHz switches for military, aerospace, radar, and microwave communications. The low-cost line for industrial automation, digital-channel testing, and cellular communications applications includes switches that operate from dc to 2 GHz and cost less than $8 each (again depending on volumes and configurations). Expect these to arrive in the second half of this year.

Key to TeraVicta's switch design is a patented high-force disk actuator technology and in-line chip-scale hermetic packaging. The combination provides switching lifetimes of hundreds of billions of cycles and high levels of reliability (Fig. 2). The switch is built directly on an alumina (ceramic) conductive layer with metal vias (Fig. 3).

A proprietary gold alloy is used for the switch contact, helping minimize stiction and contact resistance problems. A low-voltage (3 to 5 V) potential, via a separate charge-pump IC, circumvents the need for a high-voltage (65 V) actuation force for high-force disk actuation. The end result is a small (3.25 by 4.5 mm), hermetically sealed package with a Kovar lid. Radant MEMS has already demonstrated high-performance RF MEMS switches for the ATE and military markets (Fig. 4). Lifetimes exceeding 700 billion switching cycles were achieved, and prototypes featured 500 billion cycles. These switches are part of electronically steerable antennas used by the U.S. Air Force for fire-control radar mounted on balloons (aerostats).

Developed under funding by the U.S. Defense Advanced Research Projects Agency (DARPA), they're manufactured using wafer-level processing and assembled in a rigid, lightweight structure fabricated from graphite composite rigid foam and a flexible RF substrate. Each switch occupies just 1 mm3. The technology was originally developed in the late 1990s by researchers from Northeastern University and Analog Devices and subsequently licensed to Radant MEMS.

Other RF MEMS approaches include bulk acoustic wave (BAW) and film bulk acoustic-resonator (FBAR) technologies. Agilent Technologies, Avago Technologies, Infineon, and others have enjoyed considerable success with FBAR for RF applications.

A BAW device, which is a metal-insulator-metal (MIM) capacitor, uses two metal layers that sandwich a piezoelectric dielectric. An FBAR consists of a piezoelectric material, like aluminum nitride, sandwiched between two electrodes and acoustically decoupled from the surrounding medium.

Avago recently announced two FBAR duplexers for handsets, PC data cards, and other wireless products operating in the U.S. Personal Communications Service (PCS) and Universal Mobile Telecommunications System (UMTS) frequency bands. Both duplexers are housed in ultra-thin packages featuring a height of 1.3 mm, with a 3.8- by 3.8-mm footprint. These dimensions also enable miniature RF modules with increased functionality to be embedded into other portable consumer appliances.

Motorola is pushing the integration of passive components with MEMS moving structures on the printed-circuit board (PCB). Coining the term "mesoMEMS," the company developed and implemented a mesoMEMS structure for RF switching applications in mobile phones. According to Motorola, mesoMEMS structures cost less to develop than monolithic RF MEMS structures, since they can be fabricated using PCB processing techniques.

Many experts feel confident that manufacturing costs for RF MEMS ICs will decrease, making their application more ubiquitous. "You only need to look at MEMS accelerometers now in widespread use in mobile phones and other consumer electronics products to appreciate how rapid technology advances can influence applications," says David L. Yuknis, VP of marketing and product manager at Akustica.

"It wasn't that long ago that these MEMS ICs were considered too expensive," he continues. "This has now been proven otherwise. The same thing is bound to happen with RF MEMS ICs."

Most components used in mobile phones are passive elements, such as inductors, variable capacitors, and filters. Using RF MEMS devices to replace these components holds great potential. Passive components require specific filters for specific bandwidths, making a phone's front end more complex and costly. RF MEMS devices not only reduce component count and save space, they also lower noise and reduce power dissipation.

John McKillop, CTO for TeraVicta Technologies, is very optimistic about the future of RF MEMS technology. He believes three key trends will drive new applications for RF MEMS switches over the next three to five years: the proliferation of a wide variety of new product configurations, a substantial improvement in reliability, and significant reductions in switch size and costs.

Also, he foresees the development of lower-cost materials and a drastic reduction in packaged switch chip size. "We can expect a 90% size reduction in switch form factors, down to less than 1.5 mm2, over the next couple of years," he says.


To join the conversation, and become an exclusive member of Electronic Design, create an account today!