Thinking that they were facing a “normal” set of technical challenges, MEMS switch developers consistently underestimated the difference between being able to sample a few prototype devices from a wafer and consistently manufacturing products that deliver high levels of quality and reliability. This created market expectations that are only now beginning to be met, and generated a considerable degree of skepticism concerning both MEMS companies and their products that still exists today. A number of technical issues contributed to this miscalculation, and to the difficulties in developing a reliable MEMS switch. Fortunately, hindsight allows us to highlight the most important of these issues and to identify the key solutions that were required to develop the first reliable RF MEMS switch products.
Unlike many other emerging technologies, MEMS switches must compete against established products based on very mature and stable technologies. Market entry products must have an initial defect rate of less than one percent and an operating life of at least 100 million cycles—10 times higher than a traditional electromechanical relay with otherwise comparable performance. This is a phenomenal expectation for an emerging technology product and a significant challenge in itself.

In parallel with this hurdle were two other serious technical issues that impacted nearly every MEMS switch development effort. The first was a subtle but important conflict between the market requirements for MEMS switches and stiction, the typical early failure mechanism of these devices (see Figure A1).

RF MEMS switch users need low loss devices that are capable of reliably handling a few watts of RF power in a frequency band from DC to 5 GHz. This demand can only be met by ohmic contact switches with very low contact resistance (insertion loss). The requirement for low contact resistance, in turn, favors the use of relatively soft contact metals and/or high contact force, which makes the contacts stick together and fail (a phenomenon known in the MEMS community as stiction). Reliable low loss operation therefore requires stiff switch elements that provide enough return force to overcome stiction. High return force in turn requires a high enough operating voltage to generate enough closing force to overcome this stiffness and achieve an efficient contact. The correlation between design parameters and key failure mechanisms necessitates the use of high force actuator designs.

Low voltage actuator operation requires a compliant switch element (spring) that will necessarily have a limited return force. Thus, there is a fundamental conflict between the requirement for low switching voltage and the requirement for reliable, low loss operation. Although a variety of low voltage designs were investigated, all commercial solutions to date use “high force” actuator designs. This in turn requires high closing force, which can only be generated by high switch voltages.

Early MEMS switch designs were also plagued by the lack of hermetic packaging expertise within the MEMS industry. Contact degradation is one of two primary failure mechanisms for MEMS switches with ohmic contacts. This is primarily caused by contamination of the switch contacts and leads to steadily increasing values of switch resistance and insertion loss until the switch fails.

Long life MEMS switches thus require both high force actuation and manufacturing and packaging techniques that fabricate contaminant free switch contacts and keep them clean. This in turn requires cavity style packages that can protect the moving parts of the MEMS device from damage and contamination. A viable solution to this problem only emerged in the last several years as MEMS switch manufacturers adapted wafer bonding techniques to the fabrication of hermetic cavity packages. This advancement in packaging technology has shown the incredible sensitivity of MEMS switches to contamination. The crucial importance of hermetic packaging is highlighted by the fact that early MEMS switches attempted to get by with non-hermetic packaging technology. All failed to meet the 100 M cycle lifetime required for an entry-level product to compete with incumbent technologies. The MEMS switches manufactured by TeraVicta and Radant MEMS use in-line chip scale hermetic packages that are sealed in a clean room. This approach helps eliminate contamination of the MEMS switch contact and is a key factor in achieving operating lifetimes in excess of 100 M cycles.

Looking back with perfect hindsight, it is now possible to see that two key achievements were required to develop a commercial MEMS switch. The first was the realization that high force actuation was essential. The second was that in-line chip scale hermetic packaging was equally important in delivering highly reliable MEMS switches. TeraVicta and Radant MEMS have both solved the puzzle.