Alignment teaching assisted fully automated mechanical dicing of MEMS and NEMS devices

Abstract

In the dynamic landscape of semiconductor manufacturing, the demand for innovative and efficient techniques is ever-growing. Dicing is a singulation process where a machine known as a dicing saw or dicer uses a diamond blade or laser to separate dies from a wafer through a manual, semi-automated, or fully automated process. In diamond blade or mechanical dicing, the dicing saw utilizes a thin blade to cut through a wafer. This paper presents the design and implementation of an alignment teaching-assisted fully automated dicing process for the singulation of microelectromechanical systems (MEMS) devices. A pseudo-MEMS device with potential alignment targets was designed and manufactured by conventional microfabrication techniques. Alignment teaching operation was optimized for the dicing saw by finding the most appropriate alignment targets, as alignment teaching is as a pre-requisite for realizing both auto-alignment and automated dicing processes. A systematic trial-and-error approach was employed to discover the most suitable alignment targets from a pool of twenty-three potential target patterns. A circle was identified as an excellent macro target, while the addition symbol, hash symbol, and rectangle-pair were determined to be the most appropriate micro targets. The developed versatile singulation process is capable of executing an alignment teaching assisted fully automated (i.e. a total of one-click to initiate and finalize) dicing for singulating MEMS device chips, irrespective of alignment target color, die size, or wafer material. Furthermore, we developed, and experimentally validated, a mathematical model to estimate the total process time for the automated dicing.