A Multifault Testing Method for TSVs Based on GAF-DRSN and Mirror Constant Current Source Structure

Three-dimensional stacked integration based on through-silicon via (TSV) meets the high-speed development requirements of integrated circuits (ICs). However, TSV is a sensitive unit prone to manufacturing defects, with common faults being void faults and leakage faults. When TSV exhibits multifault,...

Full description

Saved in:
Bibliographic Details
Published in:IEEE transactions on components, packaging, and manufacturing technology (2011) packaging, and manufacturing technology (2011), 2024-10, Vol.14 (10), p.1744-1752
Main Authors: Shang, Yuling, Geng, Longlu, Li, Chunquan, Song, Zhuofan, Duan, Gefei, Zhang, Jintao, Li, Junji
Format: Article
Language:English
Subjects:
Automation
Circuit faults
Circuit reliability
Current leakage
Current sources
Deep residual shrinking network (DRSN)
Discharge
Electrical faults
Fault detection
Faults
Image acquisition
Integrated circuits
Interconnections
Machine learning
Manufacturing defects
mirror constant current source
Mirrors
multifault testing
Packaging
Switching circuits
through-silicon via (TSV)
Through-silicon vias
Citations: Items that this one cites
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Three-dimensional stacked integration based on through-silicon via (TSV) meets the high-speed development requirements of integrated circuits (ICs). However, TSV is a sensitive unit prone to manufacturing defects, with common faults being void faults and leakage faults. When TSV exhibits multifault, such as simultaneous resistive void fault and current leakage fault, the reliability of 3-D ICs is significantly reduced compared to single faults. To address this, this article proposes a TSV multifault testing method based on a Gramian angular field (GAF), deep residual shrinking network (DRSN), and a mirror constant current source structure. The mirror constant current source circuit is initially designed in the method, with the load structure being TSV and TSV charge/discharge rates are measured as testing parameters. Then, the GAF is employed to transform the acquired charge/discharge signals into 2-D images. Ultimately, the DRSN model should be applied to precisely detect faults in the transformed TSV images representing different fault types. The results demonstrate the effectiveness of the proposed method in classifying TSV fault types, with an average accuracy exceeding 98%. The method exhibits notable advantages, including high accuracy and robust generalization capabilities.
ISSN:2156-3950
2156-3985
DOI:10.1109/TCPMT.2024.3456228