OPC and modeling solution to support 0.55 NA EUV stitching

Abstract

Background The new high numerical aperture (0.55 NA) extreme ultraviolet lithography (EUVL) machine has been developed, which uses an anamorphic projection system with the demagnification of 4× in x-direction and 8× in y-direction. Due to the unchanged 6 in. mask, 0.55 NA EUVL reduces the exposure field size to half-field (26×16.5 mm2). Therefore, the in-die stitching between two exposures might be needed for applications requiring a larger than half-field size. To achieve in-die stitching in practical applications at advanced nodes, performing model-based optical proximity correction (OPC) is an essential step.

Aim To build an accurate OPC model, the interaction effects between two stitching fields require some special considerations, including aerial image interaction, optical proximity effect among the stitching patterns, mask absorber reflection, black border proximity effect, and the stray light from the neighboring fields effect. All these effects must be captured by specific models and corrected during OPC. In this paper, we will study the model accuracy and design decomposition rules at the stitching region and provide a solution from an EDA perspective.

Approach In this paper, the in-die stitching effects and solutions are investigated using a Ta-based dark-field mask. To study the model accuracy at the stitching region, various stitching test patterns have been designed and placed on imec 0.55 NA test masks, and the wafer data are collected from the 0.55 NA EUV scanner at the joint ASML-imec High NA EUV Lithography Lab. To enable effective in-die stitching, the impact of design decompositions on the stitching performance is investigated by performing stitching OPC with the built double exposure OPC model.

Results The model accuracy has been evaluated using the obtained wafer data, both for single and double exposures in the stitching region. It is important to use a smart cut approach to decompose the design before running OPC. Comprehensive results provide a detailed comparison between double exposure in the stitching region and single exposure through simulation, with particular emphasis on their respective impacts on overall stitching performance.

Conclusions In this paper, the in-die stitching effects and OPC solution have been investigated under idealized lithographic conditions, where mask errors, overlay errors between two exposures, and the resist delay effect are not considered. Our initial investigations indicate that double exposure shows promising model accuracy with the existing calibrated resist model (based on single exposure data), and new resist model calibrations might not be necessary for the stitching region. Notably, the wafer critical dimension (CD) in the double exposure region is sensitive to flare values, resulting in wafer CD shifts and requiring accurate flare characterization. The smart cut approach for design decomposition effectively mitigates edge placement error issues across the stitching line. By employing this smart cut approach, the stitching OPC can deliver patterning performance similar to single exposure OPC, with only slightly degraded exposure latitude.