Application of Scatterometry-Based Machine Learning to Control Multiple Electron Beam Lithography

Authored by: Nivea Figueiro, Francisco Sanchez, Roy Koret, Michael Shifrin, Yoav Etzioni, Shay Wolfling, Matthew Sendelbach, Yoann Blancquaert, Thibault Labbaye, Guido Rademaker, Jonathan Pradelles, Lucie Mourier, Stephane Rey, Laurent Pain | ASMC 2018, May 1, 2018


The evaluation of scatterometry and machine learning for the monitoring of intended critical dimension (CD) variations within scatterometry targets is presented. Such variations mimic non-uniformities potentially caused by massively parallel e-beam Maskless Lithography (ML2). Although previous results [1] demonstrate that traditional model-based scatter-ometry can properly quantify these within-target variations, the current work shows that the application of scatterometry-based machine learning complements the model-based scatterometry results. While model-based scatterometry can provide information about structure profile, which can be used to detect parameter shifts even in the absence of a reference, machine learning provides superb correlation to a defined reference.
Keywords—machine learning, scatterometry, alternative lithography, e-beam lithography, multibeam, multiple e-beam, dose variation, TMU, TMU analysis