Characterization and modeling of oxide chemical-mechanical polishing using planarization length and pattern density concepts

TLDR

Chemical‑mechanical polishing is the leading dielectric planarization technique, yet pattern‑dependent thickness variations still compromise chip performance and yield. The study develops a semiphysical, pattern‑dependent CMP model with integrated parameter extraction to predict post‑CMP oxide thickness accurately. The model is calibrated by polishing test wafers and extracting parameters from standard layouts that vary feature density, pitch, and include step‑density structures to enhance extraction. The planarization length is identified as the key parameter characterizing the CMP process.

Abstract

Chemical-mechanical polishing (CMP) has emerged as the dominant dielectric planarization method due to its ability to reduce topography over longer lateral distances than earlier techniques. However, CMP still suffers from pattern dependencies that result in large variation in polished oxide thickness across typical chips, which can impact circuit performance and yield. A comprehensive semiphysical pattern dependent model of the CMP process, integrated with a parameter extraction and process characterization methodology, has been developed to enable accurate and efficient prediction of post-CMP oxide thickness across patterned chips. In the characterization phase, test wafers are polished to obtain model parameters for the desired CMP process. Standard test layouts have been defined which consist of regions with different feature density and pitch; a new contribution is the inclusion of "step density" structures which provide large abrupt post-CMP thickness variations to improve parameter extraction. The key extracted parameter which characterizes the particular CMP process is the planarization length.