This Small Business Innovation Research Phase II SBIR project will develop manufacturing capabilities for MEMS electrical switches with a novel dual substrate design approach. The approach consists of dividing the switch components between two substrates, with the moving portion on an upper substrate, and the stationary contacts on a lower substrate. The moving portion will be formed from a stress-free layer of single crystal silicon, and so has no tendency to warp or distort. Using two substrates allows the contacts to be fully exposed throughout processing, and cleaned just before the substrates are bonded together to form the switch, thereby minimizing the contact resistance of the switch. Because the contacts are exposed, they can be effectively cleaned just prior to sealing in the hermetic seal between the two wafers, thereby reducing the contact resistance of the junctions. This Phase II effort will take the improved design into volume manufacturing to produce higher power, higher frequency, lower contact resistance and/or smaller footprint switches than competing ones while being produced at lower costs.
If successful, the approach described here will be used to produce MEMS cantilevered switches for a broad range of applications, from DC power handling applications to RF and radar applications. Because of their high current-carrying, high frequency characteristics with small size and low cost, the MEMS switches may serve as viable replacements for FET switches or micro relays in a wide range of devices. The approach may also be applicable to other sorts of MEMS devices, such as sensors and actuators, which may have a movable component suspended over a substrate which interacts with a fixed component on the substrate. This approach may therefore fundamentally alter how these devices are manufactured, and open up a wide range of applications not presently served by MEMS devices.
This Small Business Innovation Research project tested and confirmed a novel approach to the manufacture of small, cantilevered MEMS electrical switches. The approach promises to allow the manufacture of small, electrical switches which operate at higher power, higher frequency, and with lower contact resistance, yet still have a footprint of about 200 microns by 300 microns. The approach obviates or ameliorates a multitude of problems which exist using the present surface machining approach to the fabrication of these switches. The approach divides the switch components between two substrates, with the moving cantilevered portion on an upper substrate, and the stationary contacts on the lower substrate. The moving portion may be formed from a stress-free layer of single crystal silicon, and therefore has no tendency to warp or distort. Using two substrates allows the contacts to be fully exposed throughout processing, until the substrates are bonded together to form the switch. Because the contacts are exposed, they can be effectively cleaned just prior to sealing in the hermetic seal between the two wafers, thereby reducing the contact resistance of the junctions. In addition, the dual substrate approach affords a number of design options not available using the surface machining approach. An illustration of the developed switch is shown in Fig. 1, and the performance attributes are summarized in Table 1. The techniques developed in this project here may be used to produce MEMS cantilevered switches for a broad range of applications, from DC power handling applications to RF and radar applications. Because of their high current-carrying, high frequency characteristics with small size and low cost, the MEMS switches may serve as viable replacements for FET switches or micro relays in a wide range of devices. The approach may also be applicable to other sorts of MEMS devices, such as sensors and actuators, which may have a movable component suspended over a substrate which interacts with a fixed component on the substrate. This approach may therefore fundamentally alter how these devices are manufactured, and open up a wide range of applications not presently served by MEMS devices. In the course of this effort, the Dual Substrate MEMS switch design proved to be highly manufacturable, achieving die-level yields in excess of 90%. In fact, on the last four wafers manufactured, we achieved functional yields in excess of 70%, with the highest yielding designs having functional yields better than 90%. As we continued to improve the contact resistance distributions and the leakage performance of the switches, and to reduce the size of the design set fabricated on each wafer, we expect that we will eventually be able to obtain 95% or better yields in full production. Given the small size of the device and the number that can be accommodated on each wafer, this represent an enormous value generation on every wafer fabricated. Accordingly, with this work, we successfully demonstrated one of the major goals of this project, which was to produce a highly manufacturable, high yielding device. However, the design was dogged throughout development by a single performance issue: failure of the switch to open after completing many open/close cycles, or after being held closed for an extended period of time. Evidence suggests that to proceed further, additional research will be needed to formulate a contact metallurgy with sufficient softness to achieve the sub-ohm contact resistance, but with sufficient hardness to avoid a fatal amount of stiction at the junction. It would appear that these issues are fundamental in nature and would affect virtually any MEMS contact switch design, as the restoring forces in these very small devices and proportionately small, and may thus be overwhelmed by even a small amount of stiction.
