Senior Engineer ZEISS USA
Juan received his PhD in Physics from the University of Illinois at Urbana-Champaign in 2014. In 2015, he joined Intel in Oregon as Process Development Engineer to work in CMP/Wafer thinning for TSV as well as M0-M5 Back-End Dry Etch for N10 Cannon Lake. In 2017, Juan joined ZEISS in the PCS group working on X-ray Microscopy for Advanced Packaging and then moved to DUV and EUV semiconductor mask metrology, owning the applications support for the ZEISS SMS division for the USA customer base. Juan is now part of the ZEISS Expert Ladder as a Senior Engineer and drives solution development projects across metrology platforms in ZEISS USA.
Abstract
New Developments in Metrologies to Characterize Particle Size Distribution in CMP Slurries
Recent advances in logic and memory devices for AI-focused data centers are creating new challenges for characterizing consumables and chemicals used in semiconductor manufacturing, including CMP slurries. Shrinking critical dimensions, increasingly complex device architectures, progress in monolithic integration and packaging, a growing number of CMP steps, the introduction of new polishing-layer materials, and tighter requirements for wafer defectivity and planarity all demand greater accuracy, reliability, and repeatability in CMP slurry characterization.
Currently, there is a lack of documented industry consensus and comparative analysis of metrologies that are required to characterize such critical properties of CMP slurries as Particle Size Distribution (PSD). In addition, there is a lack of new pioneering approaches to reliably measure PSD, especially to measure fine particles with diameters of 2 nm or smaller. As such, accurate characterization of PSD of CMP slurries is becoming increasingly important to predict fab performance and avoid wafer defects.
In the presentation, the advantages and shortcomings of legacy and more recent metrologies for characterization of PSD of CMP layers will be analyzed, including DLS, laser diffraction, nano-tracking, BET adsorption, spinning disk centrifuge, SEM, TEM, SAXS, NMR, CHDF, nebulization, and particle counting.
In addition, capabilities of the new approach based on analyzing SEM images using Machine Learning (ML) algorithms will be reviewed. The suggested SEM/ML approach shows the potential to identify peaks of PSD for fine particles in the range from 5 nm to below 2 nm, reliably separate the peaks of bimodal or multimodal PSD, and recognize and exclude particle aggregates from PSD analysis.