In the area of MEMS switches two different approaches are being investigated: (a) a compliant switch for low activation voltage and high power handling capability (b) a switch pair for high isolation. The two switches are based on different designs and being are developed in parallel.
In the past, several different MEMS structures for building micromachined switches were developed successfully, rivaling the performance of conventional solid-state switching devices. However, none of them could be operated at high frequencies and with low actuation voltages, as necessary for applications in wireless and space- or airborne systems.
The switch consists of a 220 µm x 220 µm electrode that is placed 4.2 µm above a finite ground coplanar waveguide (FGCPW) line and is held in place by serpentine springs at its sides (Figure 7). The name "compliant switch" derives from the low spring constant of these springs, resulting in low forces needed to actuate the structure in order to decrease the pull-in voltage of the switch.
Figure 7: Scanning Electron Microscope (SEM) photo of a compliant switch7
A second electrode placed above the switch keeps the switch from moving
unintentionally due to high pressure or accelerations of the system. A constant
voltage is applied, keeping the switch clamped to the top electrode when
not in use. For switching, this voltage is removed, and the necessary pull-in
voltage (~ 15 V) between the ground plane of the transmission line and the
switch is applied, clamping it down and thus providing a high capacitance
that serves as virtual short at high frequencies. The holes in the structure
are meant to release the air between the electrodes at the time of switching
very quickly, improving the switching speed.
Figure 7: Switching principle of a micromachined compliant switch. (a)
Switch is not activated. (b) Switch is activated and pulled down to the
Table 1 will be opened in a new window and includes all the steps needed for the microfabrication of a compliant switch. The process is shown diagrammatically here. XX
The final structure shows the transmission lines at the surface of the wafer (red), the top electrode for keeping the switch up and the switch itself in the middle between top electrode and transmission line.
The supercritical CO2 release in the very end of the production process (steps  - ) is necessary to circumvent the so-called "stiction" problem that otherwise would occur when finally drying the rinsed structure: Due to the surface tension of the deionized water used for rinsing, the structures are pulled down to the surface of the wafer when dried by conventional means (N2 drying). There they are held by forces that have not been investigated completely. Trying to free the switch results in damaging it.
To prevent this undesired effect from happening, one approach is to immerse the sample in isopropyl alcohol (IPA) and two baths of ethanol after rinsing it with water. Then the sample is placed in a cooling chamber where the alcohol is replaced by liquid CO2 at a very low temperature. The CO2 is brought to a supercritical state by raising the chamber temperature and pressure, where no surface tension occurs. After releasing the CO2 as gas the final structure can be removed from the chamber.
The high isolation switches consist of four air-bridges above a transmission line, which are either lined up behind each other or in a cross-like layout like shown in Figure 8 below. For switching, the air-bridges are pulled down by a DC voltage, resulting in a capacitive short at high frequencies, just like the compliant switch described above. The difference is the higher pull-down voltage for the bridges of about 20 to 25 V DC (opposed to 15 V for the compliant switch).
Figure 8: Top view of a cross-bridge high isolation switch
Figure 9: Cross section of one of the air-bridges in the switch shown
The process steps shown in Table 2 are similar to those of the compliant switch described in Table 1. Since the structure of the air-bridges is more rigid, the supercritical CO2 release in the end of the process is not necessary, rinsing and drying the final structure is sufficient. No descum is done after developing or stripping photoresist, and gold is used for electroplating the bridge (in contrast to nickel applied to the compliant switch).
In general, the process steps and the exact layers applied in order to produce the desired structure are still under development and in a constant state of change to improve the characteristics of the switch. As stated above, the process documented in the table is the one currently under development at the University of Michigan.