Large-range two-dimensional sub-nano-misalignment sensing for lithography with a piecewise frequency regression network

Abstract

Circular gratings have been traditionally used as coarse alignment markers rather than fine ones for carrying out two-dimensional (2D) large-range misalignment measurements. This is primarily due to its complex phase distribution, which renders the extraction of information from high-precision alignment challenging using conventional frequency filtering methods. Along these lines, in this work, a novel, to the best of our knowledge, convolutional regression filter capable of achieving a 2D misalignment measurement with an impressive accuracy of 0.82 nm across a 3 mm range was introduced. Importantly, the proposed approach exhibited robustness against system errors and noise. It is anticipated that this strategy will provide an effective solution for similar misalignment sensing applications and hold promise for addressing future challenges in these fields.