Scaling Effect Of Through-Silicon Via (Tsv) On Stress And Reliability For 3D Interconnects

Keywords

Extrusion; Microstructure; Reliability; Scaling; Stress; TSV

Abstract

Through-silicon vias (TSVs) enable full three-dimensional integration by providing high-density vertical interconnections for improved device bandwidth and power consumption. However, TSVs pose unique reliability risks due to via extrusion, which is caused by thermal stress induced by the mismatch in the coefficients of thermal expansion (CTE) for the silicon and the copper. These effects can degrade device performance and it has been proposed that optimal post-plating annealing and shrinking via dimensions can be effective in mitigating negative stress effects for TSVs. In this paper, the scaling effect on TSV stress and reliability is investigated by examining the evolution of the copper microstructure during annealing and its effect on the plasticity and extrusion statistics for 10, 5, and 2μm-diameter TSVs. TSV stress and plasticity are correlated to the cross-sectional microstructure using synchrotron x-ray microdiffraction and electron backscatter diffraction. Annealing was found to increase scatter in the grain size distribution and in the via extrusion statistics, as well as significantly increase the extrusion heights for each TSV set. These results can be traced to the elastic anisotropy of copper, as the abnormal grain growth, which increases the statistical spreads in the via extrusion distributions and is controlled by the strain energy originating from the CTE mismatch. Interestingly, the results from the 2μm-diameter TSVs offer no advantages over larger TSVs, as additional annealing to 400°C was found to increase the statistical variability in grain structure and via extrusion. Since such annealing processes are required for via-middle fabrication, it seems that via reliability will continue to be a challenge as TSV scaling continues.

Publication Date

3-1-2017

Publication Title

Advancing Microelectronics

Volume

44

Issue

2

Number of Pages

12-15

Document Type

Article

Personal Identifier

scopus

Socpus ID

85019888035 (Scopus)

Source API URL

https://api.elsevier.com/content/abstract/scopus_id/85019888035

This document is currently not available here.

Share

COinS