Experimental Investigation of an Embedded Heating Mechanism to Improve RF-MEMS Switches Reliability

Up to date, the remarkable performances and characteristics of MEMS switches and lumped components for Radio Frequency applications (i.e. RF-MEMS) have been demonstrated by several Authors [1, 2]. On the other hand, the reliability of such a technology still has to be fully addressed in order to enable a successful penetration of RF-MEMS technology into the market [3]. Reliability of MEMS/RF-MEMS involves several physical phenomena that can jeopardize their normal operation as well as the stability of their characteristic vs. time [4]. Among such different effects, the authors believe that one of the most important source of malfunctioning is the stiction (i.e. the switch remains stuck in the actuated position when the controlling bias is removed) due to the charge entrapment into the insulator layer and/or the formation of micro-welding [5]. In order to counteract stiction, the authors already presented an innovative RF-MEMS switch design employing an active restoring mechanism, based on an high-resistivity serpentine heater (see Figure 1) to bring it back to its normal operability when stiction occurs [6]. In this work we report on the experimental testing recently performed on such test structures employing the active mechanism, fabricated in the FBK RF-MEMS technology. Firstly we used the Laser Doppler Vibrometer (LDV) integrated into a Polytec MSA-500 optical profilometer to verify the effectiveness of the heating mechanism to induce a movement of the suspended bridge. Figure 2 shows the vibration velocities as a function of the frequency of the bridge central part when a periodic chirp signal (iheater = 2 mA) is applied to the heater. We have also verified the transient behavior over time, measuring the bridge displacement induced by the heating (not by the actuator), when a square pulsed signal is applied to the heater pads, see Figure 3. Once we checked the good response of the bridge to the heating mechanism, we tested the effectiveness of the heating on the possibility to release the switch whether stiction occurred. Being very difficult to predict if a switch is stuck, we applied the following procedure, controlled by a LabView interface: (1) actuate the switch at VBIAS = 80 V; (2) decrease VBIAS at a value slightly higher than the release voltage; (3) switch-on the heating mechanism for a user selectable time; (4) switch-off the heater. Monitoring continuously the Sparameters over time it was possible to study the bridge behavior. Figure 4 shows an example of such procedure, as well as the release of the bridge induced by the embedded heater. In order to better understand the behavior, and the limits, of such restoring mechanism, we acquired thermal images of two pieces of wafer with different area (“small” ~ 1 cm2, “big” ~ 2.4 cm2) with a FLIR A20 infrared camera. Figure 5 shows the comparison of the two dies temperature images, taken at fixed intervals, when IHEATER = 4 mA (DC). It is clear the impact on heating of the bigger die, limiting the increase of the substrate temperature because of the higher thermal dissipation. This is confirmed also by Figures 6 and 7, that show the substrate temperature increase over time at different heater currents on small and big die respectively (slower increase on bigger die).