Published January 1, 2023 | Version v1
Journal article Open

Nonlinear restructuring of patterned thin films by residual stress engineering into out-of-plane wavy-shaped electrostatic microactuators for high-performance radio-frequency switches

  • 1. Sabanci Univ, Fac Engn & Nat Sci, TR-34956 Istanbul, Turkiye

Description

Electrostatic microelectromechanical (MEMS) switches are the basic building blocks for various radio-frequency (RF) transceivers. However, conventional cantilever-based designs of MEMS switches require a large actuation voltage, exhibit limited RF performance, and suffer from many performance tradeoffs due to their flat geometries restricted in two dimensions (2D). Here, by leveraging the residual stress in thin films, we report a novel development of three-dimensional (3D) wavy microstructures, which offer the potential to serve as high-performance RF switches. Relying on standard IC-compatible metallic materials, we devise a simple fabrication process to repeatedly manufacture out-of-plane wavy beams with controllable bending profiles and yields reaching 100%. We then demonstrate the utility of such metallic wavy beams as RF switches achieving both extremely low actuation voltage and improved RF performance owing to their unique geometry, which is tunable in three dimensions and exceeds the capabilities of current state-of-the-art flat-cantilever switches with 2D-restricted topology. As such, the wavy cantilever switch presented in this work actuates at voltages as low as 24 V while simultaneously exhibiting RF isolation and insertion loss of 20 dB and 0.75 dB, respectively, for frequencies up to 40 GHz. Wavy switch designs with 3D geometries break through the design limits set by traditional flat cantilevers and provide an additional degree of freedom or control knob in the switch design process, which could enable further optimization of switching networks used in current 5G and upcoming 6G communication scenarios.

Files

bib-2409822f-c004-4981-b25e-44164a0251c4.txt

Files (302 Bytes)

Name Size Download all
md5:b2accdafde8fed1bb6e2ac4a07a73636
302 Bytes Preview Download