Failure of Au RF-MEMS switches subjected to dynamic loading

The dynamic failure of Au RF-MEMS was investigated over a wide range of loading rates by three different experimental setups: a drop weight tower, which induced a maximum peak acceleration of 600g (g: acceleration of gravity), a Hopkinson pressure bar with a maximum peak acceleration of 300,000g, and a pulsed laser loading technique with a maximum peak acceleration of 1.8 × 10⁸g. In the drop weight tower the total load pulse duration was in the milliseconds range - much longer than the 28 μs resonant period of the devices - and no failure of any kind occurred in the RF-MEMS devices or their substrate. At 90,000g (generated in the Hopkinson bar) no damage in either the substrate or the devices was observed. However, at 200,000g, which corresponds to a loading duration of a few microseconds, i.e., comparable to the device resonant period, 10% of the switches failed although postmortem imaging showed no damage to the substrate. Damage increased after this acceleration and at 300,000g 20% of the switches failed, but, in addition, significant failure in the quartz substrate was recorded. Lastly, the pulsed laser loading technique, which has a loading pulse duration of a few tens of nanoseconds, was applied to accelerate the Au switches to 1.8 × 10⁸g, and the probability of failure at this loading ranged from 50% to 80%. At even larger accelerations, 10⁹g, the probability of failure was 100%. The results of this study establish the severity of dynamic failure in MEMS, despite their small mass, and its dependence on the level of acceleration which spanned about 7 orders of magnitude.