Agile prototyping of AI hardware LSIs using multi-layer aluminum interconnects in a minimal fab environment

View at publisher Summarise with VIRAT
Authors: Hirofumi Sumi, Hideharu Amano, Naonobu Shimamoto, Atsutake Kosuge, Yukinori Ochiai, Tohru Mogami, Yoshio Mita, Makoto Ikeda
Cited by: 0
References: 12

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Abstract We report a multi-layer aluminum interconnect technology developed within a minimal fab environment to enable rapid prototyping of Edge AI accelerators. To address the gap between FPGA flexibility and ASIC performance, we established a hybrid manufacturing flow integrating standard CMOS wafers. Recognizing that transport cleanliness alone guarantees insufficient yield for mega-scale integration, we incorporated a unified yield-reliability model. The analysis revealed that a 1 ppm defect rate restricts yield to approximately 37% for 10 6 vias, necessitating a strict

Fields of Study & Subjects

VLSI and FPGA Design Techniques 3D IC and TSV technologies Copper Interconnects and Reliability

Publication details

Journal / source
Japanese Journal of Applied Physics
Volume
65
Issue
8
Pages
088004–088004
Publication date
25 March 2026
Language
EN