PiFM for Defect Review in Advanced Packaging Processes

The importance of defect review in advanced packaging manufacturing 

Advanced packaging is a relatively recent innovation in semiconductor fabrication in which multiple devices (or dies) are bonded together prior to encapsulation. It has become an essential approach for sustaining performance scaling, enabling higher interconnection density and shorter signal paths, which together can increase processing speed [1, 2].  

In production, an important step is wafer binding. In this process, single or multi-layer thin-film via wafers (MTVs) with copper through-silicon vias (TSVs) are temporarily attached to carrier wafers. The wafers are then bound together by aligning corresponding Cu-TSVs and bringing them into contact, which, at the nanoscale, chemically bonds them. The carrier wafers are then debonded, the bound wafers are cleaned, and the process is repeated [Figure 1]. 

While improvements observed in chip performance from advanced-packaged systems are substantial, manufacturing is a very delicate process that is highly sensitive to surface condition. Particularly in the binding step, each wafer surface must be exceptionally clean for successful TSV-to-TSV binding to reliably occur, and cleaning steps must successfully and consistently remove all defects. Even a small number of residual defects, whether introduced during handling or bypassing cleaning steps, can greatly reduce chip performance further downstream. Importantly, defects must be easily detectable and identifiable throughout the manufacturing process so that corrective actions can be implemented to eliminate them quickly. Accordingly, pushing the limits of the optical resolution of inspection tools has become central to driving future innovation in the semiconductor industry [3].

Where PiFM steps in: a case study 

We were recently approached by a group looking to evaluate their adhesive cleaning process after a carrier wafer debonding step. Upon analyzing the topography image centered on a Cu-TSV, we observed two sets of small, spherical features: one faint set with diameters of ~20 nm, and another more prominent set with ~50 nm-sized diameters. However, if only topography was used for analysis, the identities of those sets were ambiguous, and we could not confidently classify either set as defects. 

However, we were able to prove that their cleaning process failed after PiFM analysis. On the cleaned wafer, we could identify the ~20 nm-sized group as adhesive residue, and the ~50 nm-sized group as merely aggregations of copper (I) oxide resulting from the sample’s exposure to air. Additionally, we were able to clearly observe that the adhesive residue was uniformly spread on both the Cu-TSV and the dielectric surface. All this was discovered through imaging at different wavenumbers present in the spectra we took [Figure 2]. 

In summary, we present PiFM as a promising new analytical technique to detect defects in a semiconductor manufacturing setting, specifically in advanced packaging, where defect review is particularly critical and difficult. By providing chemical information at the nanoscale to supplement topographical data, the use of PiFM can help distinguish true contaminants from benign morphological features and support rapid process optimization. 

References 

1. M. LaPedus, “Advanced Packaging’s Next Wave,” Semiconductor Engineering, (2021).
2. “Advanced Packaging Fundamentals for Semiconductor Engineers,” (2025).
3. G. Haley, “How Advanced Packaging Is Reshaping Inspection,” Semiconductor Engineering, (2025).