https://scholarworks.bwise.kr/erica/handle/2021.sw.erica/323 2026-03-11T20:03:02Z https://scholarworks.bwise.kr/erica/handle/2021.sw.erica/126597 Title: Development and validation of Generative AI Competence Scale (GenAIComp) among university students Authors: Lee, Seul Chan; Baby, Tiju; Vongvit, Rattawut; Lee, Jieun; Kim, Young Woo; Cha, Min Chul; Yoon, Sol Hee Abstract: The rapid development of Generative Artificial Intelligence (Generative AI) across several sectors underscores the need for a systematic tool to evaluate AI competence. Current digital literacy frameworks lack AI-specific competencies, resulting in inconsistencies in the assessment of AI competence. This study aims to establish a standardized assessment framework for Generative AI competence by identifying key skill factors and empirically validating a structured evaluation tool called the Generative AI Competence Scale (GenAIComp). The proposed GenAIComp has five essential factors: Information and Data Literacy, Communication and Collaboration, Digital Content Creation, Safety and Ethics, and Problem-Solving. A quantitative approach was employed, incorporating expert validation, pilot testing, and extensive empirical evaluation involving 1000 participants, principally university students. The factor analysis confirmed a robust 5-factor structure with strong psychometric properties. The final model demonstrated excellent fit indices, confirming its reliability and validity in assessing Generative AI competence across the five key factors. Research demonstrates that educational background considerably impacts AI competence, with individuals from technical disciplines showing a greater aptitude for problem-solving and content generation. Gender-based disparities were noted, with males achieving marginally higher scores in several factors, but with minimal effect sizes. Correlation analysis indicated that perceived AI expertise and frequency of AI utilization significantly influenced competence, especially in data literacy and problem-solving, and exhibited less correlation with ethical awareness. GenAIComp provides a reliable tool for assessing AI competence, helping educators, industry experts, and policymakers to design AI training programs and integrate AI literacy into curricula and thereby AI technology advancement in society. Future research should explore its applicability across cultures and include performance-based assessments to enhance AI competence. 2026-03-01T00:00:00Z https://scholarworks.bwise.kr/erica/handle/2021.sw.erica/127336 Title: Overlay deviation caused by mask heating during EUV exposure Authors: Kang, Ji-won; Ko, Heechang; Kim, Minwoo; Chae, Yu-jin; Oh, Hyekeun; Son, Seung-woo Abstract: As the resolution of EUV lithography continues to improve, the allowable overlay tolerance has decreased to the sub-nanometer level, increasing the significance of mask thermal behavior on process stability. However, the measured overlay used for correction in actual processes includes the combined influence of mask heating, optical system aberrations, stage motion, and alignment errors, making it difficult to isolate the direct contribution of mask heating. To overcome this limitation, the thermo-mechanical response of the mask during exposure was modeled independently through simulation and translated into the corresponding wafer overlay. The temporal and spatial evolution of mask temperature and deformation during repeated exposures was analyzed, and the overlay magnitude and distribution were compared under various exposure conditions, including pattern type, dose, and numerical aperture (NA). Under the 0.55 NA condition, pronounced layer-to-layer errors were observed at in-die stitching regions where adjacent half-fields meet. These results indicate that mask heating is one of the dominant sources of overlay variation in high-resolution EUV lithography processes. © 2025 SPIE. All rights reserved. 2025-11-01T00:00:00Z https://scholarworks.bwise.kr/erica/handle/2021.sw.erica/127335 Title: Source optimization for less aggressive optical proximity correction in 0.55 NA logic single patterning Authors: Kim, Minwoo; Chae, Yu-jin; Ko, Heechang; Kang, Ji-won; Yeung, Michael S.; Oh, Hyekeun; Son, Seung-woo Abstract: Logic metal layers, consisting of randomly arranged metal lines with various pitches, are intended to be printed by EUV single exposure. However, when such layers are exposed using a simple source, differences in pattern density and shape cause significant critical dimension (CD) variation, preventing the printed features from matching the target design. Moreover, the complexity of optical proximity correction (OPC) increases with pattern diversity and reduced feature size, resulting in higher computational cost and more challenging mask fabrication. In this study, we investigate the potential of pixelated source optimization to achieve high-fidelity patterning of logic metal layers without any use of OPC or bias. The optimized source is evaluated for different absorbers under 0.55 NA conditions. By comparing optimal source shapes for each absorber, we demonstrate that pixelated sources can reproduce the intended patterns while achieving high normalized image log slope (NILS) and a wide process window. This approach offers a pathway toward less aggressive OPC and improved manufacturability for future logic single patterning. © 2025 SPIE. All rights reserved. 2025-11-01T00:00:00Z https://scholarworks.bwise.kr/erica/handle/2021.sw.erica/127358 Title: Pixelated source polarization optimization for hyper-NA EUV lithography Authors: Chae, Yu-jin; Kim, Minwoo; Kang, Ji-won; Ko, Heechang; Son, Seung-woo; Yeung, Michael S.; Oh, Hyekeun Abstract: As extreme ultraviolet (EUV) lithography advances into the hyper-numerical aperture (hyper-NA, NA > 0.75) regime, achieving sub-8 nm resolution while maintaining a sufficient process window (PW) becomes increasingly critical for enabling next-generation semiconductor nodes. According to the latest scaling roadmaps, metal pitches are expected to shrink below 16 nm, driving the need for advanced lithographic techniques that overcome the limitations of conventional polarized illumination methods. Although transverse electric (TE) and transverse magnetic (TM) polarization approaches have shown effectiveness in high-NA systems down to 15 nm pitch, their impact saturates as vectorial light–matter interactions become more dominant. To address these challenges, we present a pixelated source polarization optimization (SPO) technique tailored for hyper-NA EUV lithography. This method enables spatially resolved polarization control at each pixel within the source pupil, expanding the degrees of freedom beyond conventional uniform-polarization schemes. Rigorous electromagnetic simulations using the Fastlitho platform demonstrate that pixelated SPO extends resolution capabilities beyond previous limits, achieving 11 nm pitch for line-space (L/S) and 15 nm pitch for contact-hole (C/H) patterns—while maintaining peak normalized image log-slope (NILS) > 2.0 and robust normalized depth of focus (nDOF) across X and Y directions. Beyond resolution improvement, pixelated SPO simultaneously enhances NILS and DOF, thereby expanding process margins and ensuring patterning feasibility for complex DRAM and logic layouts. Previous results also confirm consistent CD control and lithographic success for mixed-pattern DRAM configurations incorporating L/S, diagonal, and C/H arrays. This study establishes pixelated SPO not only as an enhancement technique, but as a foundational enabler for future hyper-NA nodes, potentially supporting beyond-A2 scaling as projected in the ASML/imec roadmap. By pushing resolution boundaries while preserving process stability and pattern fidelity, pixelated SPO opens a new design space for advanced memory and logic device generations. 2025-11-01T00:00:00Z