Characterization and optimization of dry releasing for the fabrication of RF MEMS capacitive switches

Abstract

This paper discusses fabrication aspects of photoresist sacrificial layers for fabricating metal bridges of capacitive radio frequency (RF) microelectromechanical systems (MEMS) switches. First, reflow of the photoresist layer after lithography is investigated for reducing mechanical fracture of the metal layer by smoothing the edges of the sacrificial layer. Second, the dry-etch releasing process of the structures in an O 2 plasma has been investigated by identifying suitable etching parameters. The findings in this paper reveal that the mechanical performance of the released bridges strongly depends on the etch parameters. It is shown that especially the etching power affects the mean stress and the stress gradient in the bridge, which results in buckling and deformed bridge shape for an etching power above 500 W, drastically increasing the actuation voltage and reducing the down-state capacitance. Finally, the paper presents a suitable parameter set for the release etching of capacitive MEMS metal bridges.

RF MEMS Switches and Integrated Switching Circuits

“…This down-state capacitance degradation problem is usually a result of nonplanarization of metal bridge, surface roughness of capacitance area and etching hole in the metal bridge. [10,11]. Fig.…”

“…The surface roughness of the capacitance area can be improved by using a refractory metal layer underneath the metal bridge [9]. Flat metal bridge can be obtained by planarizing the sacrificial layer underneath the metal bridge [10,11].…”

Characterization and optimization to improve uneven surface on MEMS bridge fabrication

“…The roughness average value before and after the hard baking for 5 lm 2 , measured by an AFM, are 0.98 nm and 0.36 nm, respectively. This is because the resin material of the photoresist becomes soften and reflows at the temperature above the glass transition temperature, resulting in atomically smooth surface and also a planar surface [12]. However, the thermal planarization of the resist could stop at very high temperature due to the thermal cross linking of the resin.…”

“…Saha et al [11] used two experiments to release a membrane. In the first experiment, the sacrificial layer under the bridge is etched and the membrane is released at 400 W. However, using high power O 2 plasma for releasing the bridge increases the compressive stress in the membrane, resulting in the buckling of the membrane [12,13]. In the second experiment, etching holes are incorporated in the membrane to avoid the problem of buckling.…”

“…Although introducing the etching holes on the beam results in a faster and convenient dry release process, it reduces the resonant frequency of the beam, as well as the isolation of the switch while the switch is in contact [14][15][16]. The parameters of the releasing process by O 2 plasma, power, pressure, and photoresist thickness, were comprehensively investigated by Rahman et al and Yu et al [17,12]. The releasing became more critical when the photoresist was hard baked as it made a thermal cross link.…”