Degradation of RF MEMS capacitive switches

In this study, degradation mechanisms in Radio Frequency Micro-Electromechanical systems (RF MEMS) capacitive switches have been investigated. We presented a methodology to characterize device degradation in a way that minimizes the in- °uence of the measurement on the device itself, even if relatively slow equipment is used. An automated fast RF-CV measurement setup was developed, which enabled us to measure a whole C-V curve with minimal influence on the device. With the setup a wide array of systematic, fast and accurate degradation experiments were done. With these experiments we gained more insight in the time dependence of laterally homogeneous dielectric charging. Also, it enabled us to develop a framework to distinguish mechanical degradation from laterally inhomogeneous dielectric charging, two totally di®erent degradation mechanisms which, despite their di®erences, have similar e®ects on the device. The framework also allowed us to choose stress patterns that speci¯cally target mechanical degradation. This was used to investigate the in°uence of various operating conditions on mechanical degradation. The phenomenon of laterally inhomogeneous dielectric charging was also studied more directly with Scanning Kelvin Probe Microscopy. The SKPM measurements also showed that trapped charge di®use in the lateral direction. This contributes to the recovery e®ects we observed with electrical measurements. Finally, due to the level of automation of the developed measurement setup, and the availability of enough measurement material, it was possible to do a large number of measurements. This allowed us to study how spread in device degradation impacts the way characterization of those degrading devices should be done. We saw that measuring whole distributions of degrading devices is important when studying the various mechanisms that contribute to degradation of RF MEMS capacitive switches, as di®erences in device degradation due to for instance a change in stress intensity can easily be masked by the spread in degradation rates.