Uncertainty Estimation and Design Optimization of 2D Diffraction-Based Overlay Metrology Targets

Abstract

Scatterometry is an optical metrology technique in which light scattered from a specifically designed grating stack (overlay target) is measured in the far-field. Using 1D periodic overlay target designs, the technique has been shown to have nanometer-scale sensitivity to spatial misalignments of subsequent patterned layers, which are also known as overlay errors. However, while scatterometry is highly sensitive to overlay errors, multiple sources of systematic errors hinder its absolute accuracy. Here, we investigate how an extended version of scatterometry called Fourier scatterometry, in combination with more complex overlay target designs, can help addressing those challenges. To this end, we developed a statistical method that can determine the influence of 2D overlay targets on the overlay measurement uncertainty. We study periodic and deterministic aperiodic designs as well as designs that emerged from simulated annealing optimizations. Our results suggest that current overlay target designs could be augmented by more complex 2D designs to fulfill specific purposes, such as fabrication robustness and high sensitivity over a large overlay range.

Process window improvement for fin cut layer in self-aligned double-patterning process based on backscattered electron imaging

“…As fin damage is obviously caused by the insufficient coverage of the SOC to the closest covered fin, the OV, defined as the pattern positional accuracy between the bottom and current layers, 19 – 21 and PW, which determines the CD uniformity of the spaces between neighboring fins, 22 should also be considered. To characterizing the PW performance, the etch process is suspended at the step of SOC etch [Fig.…”

Neural network driven sensitivity analysis of diffraction-based overlay metrology performance to target defect features

“…Although the present study provides a basis for subsequent research on DBO target, more research directions still need to be explored, including but not limited to: (1) the impacts of more random combinations of various defects in different types of DBO targets, (2) the relationship between broadband metrology wavelength, the OVL target defects and their corresponding metrology performance; (3) more adaptive and robust metamodel techniques for more efficiently handling varying DBO targets and configurations; (4) the performance of deep neural network based forward and inverse models for more complex targets and defect combinations; (5) the influence of measurement wavelength on the model fitting and defect feature sensitivity analysis along with how to optimize the wavelength selection; (6) Besides, our proposed method can be applied into various DBO targets (including the cDBO targets and potential metamaterials or metasurface based targets [47][48][49]) and extended into other metrology techniques (e.g. Muller matrix ellipsometer) by some model modifications.…”

Calibration-based overlay sensing with minimal-footprint targets

“…Our minimal-footprint targets seem to encode overlay error in radiation patterns through far-field interference, rather than through more strongly position-dependent multiple scattering mechanisms. With this in mind, it is interesting to ask how more precise overlay sensors may be designed for a given measurement 19,20 . One path could be to exploit phenomena with a stronger position dependence, such as plasmonic coupling between individual nanoparticles 21 or molecular ruler-like systems 22 .…”