Design Technology Co-optimization for Early Intercept on Maturing Process Nodes

Raj Varada, Pascal Meinerzhagen, Srikanth Venkataraman

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

In this paper, we describe a novel design-technology co-optimization methodology to support yield-sensitive design optimizations in parallel with process defect learning. This work enables the simultaneous early start of multiple designs in parallel with process development and provides significant immunity to design convergence from the changes in yield maturity. It must be noted that yield-sensitive design optimizations can be deployed when the systematic defects have been primarily addressed and the defect density is in the single digits. Previously, [1] described a method to bias design area by a measure of cell yield. We believe our work is the first to attempt to holistically improve design yield through the entire design creation flow. The results presented in this work are from applying this co-optimization methodology on early cell-based networking digital designs in a 4nm class technology [2]; however, the methods are general enough to be applied to any process node.

Original languageEnglish
Title of host publicationDTCO and Computational Patterning IV
EditorsNeal V. Lafferty, Harsha Grunes
PublisherSPIE
ISBN (Electronic)9781510686366
DOIs
StatePublished - 2025
Externally publishedYes
EventDTCO and Computational Patterning IV 2025 - San Jose, United States
Duration: 25 Feb 202528 Feb 2025

Publication series

NameProceedings of SPIE - The International Society for Optical Engineering
Volume13425
ISSN (Print)0277-786X
ISSN (Electronic)1996-756X

Conference

ConferenceDTCO and Computational Patterning IV 2025
Country/TerritoryUnited States
CitySan Jose
Period25/02/2528/02/25

Bibliographical note

Publisher Copyright:
© 2025 SPIE.

Fingerprint

Dive into the research topics of 'Design Technology Co-optimization for Early Intercept on Maturing Process Nodes'. Together they form a unique fingerprint.

Cite this