Answers are grounded in this paper and cite their sources.
Achieving wafer-scale reliability in Cu–Cu hybrid bonding remains challenging due to process-induced non-uniformities and the limited understanding of buried interfacial contamination. Here, we employ a correlative multiscale microscopy framework combining scanning electron microscopy, transmission electron microscopy, and atom probe tomography (APT) to investigate bond degradation from the wafer center to the edge. We observe progressive bond deterioration characterized by interfacial delamination, reduced Cu grain size, and the formation of a nanometer-scale amorphous interlayer. While such interlayers have been widely observed in hybrid bonding, their chemical nature has remained largely unresolved. APT directly reveals that the interlayer consists of Si-, C-, and N-bearing species consistent with contributions from the adjacent SiCN dielectric, together with Cu oxide and organic-related contaminants. By correlating interlayer chemistry and thickness with microstructural evolution, we establish a direct link between interfacial contamination, suppressed cross-interface grain growth, and bond degradation. Furthermore, we observe a progressive thickness-dependent degradation, where interfacial thickening from ∼1 to 2 nm at the center to ∼2 to 3 nm in the donut region is accompanied by deterioration in mechanical bond integrity and localized delamination. These findings provide atomistic insight into the chemistry governing hybrid bond reliability and establish quantitative guidelines for process optimization toward wafer-scale integration.
No takes yet. Share an insight, caveat, or question.
Shukla et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: