AI-Driven Metrology for the Yield-Critical Layer in Advanced Packaging
Heterogeneous integration is redefining semiconductor scaling, shifting system performance from front-end transistor advances to package-level integration of chiplets, logic, and High Bandwidth Memory (HBM).
As packaging evolves to finer pitches, higher bump density, and multi-die stacking, process margins tighten significantly. Small variations in wafer topography, morphology and warpage, as well as bump height, shape, or coplanarity, can directly impact interconnect reliability and yield, driving the need for high-accuracy, inline dimensional metrology.
Camtek plays a central role in this transition, leveraging advanced inspection and metrology capabilities to deliver actionable data for tighter control of bonding uniformity, interconnect geometry, and warpage-driven variations.
In this talk, I will introduce Camtek’s latest metrology platforms designed to address the most challenging measurement problems in advanced packaging. Beyond sensor innovation and engineering breakthroughs, the presentation will demonstrate how AI-enhanced algorithms can significantly extend the capabilities of classical metrology models.
Two real-world case studies will be presented. The first shows how AI-enhanced reflectometry improves multi-layer film thickness measurements by reducing median error from several nanometers to sub-nanometer levels while simultaneously accelerating computation time. The second demonstrates how AI-driven modeling overcomes traditional White Light Interferometry (WLI) limitations in TSV reveal metrology, eliminating measurement artefacts and improving agreement with reference measurements. Quantitative results from production-relevant datasets will be shared, highlighting the practical impact of AI on metrology accuracy, robustness, and time-to-solution.
Camtek welcomes discussions with customers interested in advancing next-generation metrology solutions and is actively seeking partners willing to contribute relevant problem statements, representative samples, and ground-truth reference data to help train, validate, and accelerate the development of emerging AI capabilities.
Key Technologies Covered
- AI-empowered algorithms
- White Light Interferometry
- Small spot reflectometry
