Invited Speakers

Abstract
Confocal microscopy is a cornerstone of modern metrology and bio-imaging, widely utilized in diverse fields such as the cross-sectional imaging of living cells, the topographic evaluation of precision mechanical components, and high-accuracy stage positioning measurements. The fundamental principle relies on a focused light beam, known as a confocal probe, which acts as a non-contact optical sensor. By employing a pinhole aperture to reject out-of-focus light, this technique achieves exceptional spatial resolution and high contrast in optical-axis displacement measurements, ensuring that only signals from the exact focal plane are detected.
To further push the boundaries of measurement sensitivity and axial range, researchers have increasingly turned to dual-detection confocal probes. These systems utilize two independent detectors, often placed at specific offsets relative to the focal point, to implement advanced signal processing methods. Such configurations are designed to mitigate common issues found in single-detector systems, such as sensitivity to light source fluctuations and limited linear measurement ranges.
However, the performance of a confocal probe is not universal; it varies significantly depending on the spatial placement of the detectors and the specific mathematical signal processing algorithms applied (such as differential or ratio-based methods). In this study, we systematically classify various detector configurations and signal processing strategies. By conducting a comparative analysis of their axial response curves and sensitivity profiles, we aim to clarify how specific setups influence measurement accuracy. This paper provides critical insights into the optimization of dual-detection confocal probes, offering a detailed categorization that serves as a guide for developing next-generation optical metrology tools.

Abstract
To advance manufacturing technologies, it is essential to develop a deeper and more quantitative understanding of machining phenomena. Many machining processes—such as chemical mechanical polishing (CMP), electrical discharge machining (EDM), cutting, and abrasive machining—are performed under wet conditions, where the presence of cutting fluids or slurry complicates direct observation of the tool–workpiece interactions. Although visualization plays a crucial role in elucidating material removal mechanisms and process dynamics, conventional microscopy and optical measurement techniques cannot be easily applied due to issues such as light scattering, fluid flow, and difficulty maintaining optical access in wet environments.
To overcome these limitations, a fluorescence-based measurement technique that enables robust visualization and surface shape assessment even in wet, in-situ machining conditions is proposed. When machining fluids, such as grinding or cutting oils, are excited using ultraviolet illumination, they emit strong fluorescence. By gathering this fluorescence and utilizing it as a measurement signal, the proposed method realizes surface geometry measurement without requiring the fluid to be removed. This approach supports high-speed areal imaging of the entire observation region, enabling rapid acquisition of dynamic surface information that would be difficult to obtain using conventional point-based sensors. Furthermore, the technique incorporates an in-situ calibration procedure, allowing accurate reconstruction of surface profiles despite the harsh conditions such as the presence of fluid layers.
Because of these characteristics, fluorescence-based sensing provides substantial advantages for on-machine metrology, where non-contact, fast, and robust measurements are required. In this presentation, it is introduced that the application of the proposed method to the measurement of mechanical machining tools, demonstrating its measurement performance, stability, and sensitivity to fine surface features. It is also discussed the broader potential of fluorescence-enhanced metrology for real-time monitoring and analysis of machining processes, highlighting its value as a versatile tool for understanding complex material removal behaviors in wet machining environments.

Abstract
Diamond is an indispensable material for advanced optical, electronic, and mechanical applications, yet its extreme hardness and chemical stability make conventional machining inefficient and prone to severe tool wear. Ultrashort pulse laser processing offers a non-contact alternative capable of high-precision diamond machining; however, achieving both high material removal efficiency and acceptable surface quality remains a critical challenge. In this lecture, I will present our systematic studies on improving the efficiency of ultrashort pulse laser processing of single-crystalline diamond through controlled laser irradiation strategies. First, I will discuss the effects of beam shaping on material removal behavior. Using sub-picosecond laser pulses combined with top-hat beam modulation, we demonstrate enhanced and stable ablation, achieving material removal rates on the order of 1.5 × 10⁻⁴ mm³/s with surface roughness around 250 nm Sa. These results clarify how spatially uniform energy deposition contributes to efficient material removal and surface smoothing. Next, I will focus on the role of laser-induced surface modification layers, which strongly influence ablation efficiency in ultrashort pulse laser processing. We propose and validate a two-step processing concept in which the modified layer formed during irradiation is intentionally removed before subsequent pulses. This strategy significantly increases the material removal volume per pulse and provides new insight into the dynamic interaction between transient surface states and laser ablation efficiency. Finally, I will briefly outline how these findings can be extended toward more precise and high-yield diamond processing, including feedback-assisted approaches for surface control. By integrating efficient material removal with controlled surface modification, ultrashort pulse laser processing can become a scalable and versatile technique for next-generation diamond manufacturing.

Abstract
The accurate extraction of phase information from these fringe patterns is vital to ensure high-precision measurements. Over the years, various methods have been developed to achieve this, ranging from conventional phase-shifting techniques to the more recent approaches leveraging machine learning. Conventional phase-shifting interferometry has long been the standard in optical metrology as shown in Fig. 1. These techniques involve capturing a series of interference fringe images at different phase shifts and then using mathematical algorithms to extract the phase distribution. While phase-shifting methods are widely used and offer high precision, they do have limitations. For instance, the necessity for multiple images can lead to longer measurement times, making them less suitable for real-time or dynamic applications. Furthermore, these techniques often require precise control over the phase shifts and can be sensitive to noise, especially when the fringe contrast is low or the system experiences mechanical vibrations. The limitations of conventional phase-shifting methods have prompted researchers to explore alternative techniques, including the integration of machine learning (ML) to improve phase extraction. Machine learning algorithms, particularly deep learning networks, have shown promise in overcoming some of the inherent challenges in traditional phase extraction. In ML-based phase extraction, deep neural networks or convolutional neural networks are trained on large datasets of interference fringe patterns to learn the underlying phase directly. These methods can handle noisy and incomplete data more robustly than conventional techniques. For instance, they can extract phase information from a single fringe image, eliminating the need for multiple phase-shifted images, which reduces measurement time and system complexity.

Abstract
Optics, semiconductors, and quantum technologies require atomic-scale manufacturing capabilities. However, the challenge lies in discovering or applying phenomena distinct from conventional macroscopic processes to realize atomic-scale and damage-free material removal. By employing biological enzymatic hydrolysis, an atomic-scale and abrasive-free machining technique, enzymatic hydrolysis machining (EHM), is proposed. Enzymatic hydrolysis regenerates abundant hydroxyl groups on the polymer surface, enabling the atomic-scale removal of Si through C-O-Si bridging bond formation and shearing. Static incubation experiments confirm the enzymatic hydrolysis effect, with hole/crack formation and carbonyl intensity reduction directly verifying ester bond scission. The removal depth of Si using polybutylene succinate (PBS) in pure water was 16.4 nm after 12 h, whereas it increased to 213.5 nm when Humicola insolens cutinase (HiC) was used. Pad-type and wheel-type tools obtain W-shaped and Gaussian-like removal functions for polishing and figuring, respectively. The volume removal rate was 3.8 × 10-5 mm3/min. EHM effectively achieved an atomically smooth Si surface (Sq = 0.05 nm, 5 × 5 μm2 viewing area). The generality of EHM is demonstrated by extending it to other materials (SiO2, 4H-SiC, and GaN). Furthermore, numerically controlled EHM was applied to a 100 × 50 mm² Si mirror, achieving a uniform atomically smooth surface across the entire area. This work shows that EHM is a novel and effective technique for atomic-scale manufacturing.

Abstract
As the Korea Basic Science Institute (KBSI) leads the construction of the 4th Generation Storage Ring (4GSR), securing domestic expertise in ultra-precision X-ray optics is a strategic priority. This presentation focuses on our recent advancements in overcoming technical bottlenecks in sub-nanometer scale fabrication, specifically targeting the efficiency of the deterministic finishing process.
As a significant milestone, we present a joint research project with Electronics and Telecommunications Research Institute (ETRI) and Pohang Accelerator Laboratory (PAL) focused on the development of 300 mm freeform mirrors for 2D digital X-ray sources. A major challenge in this process was the exponential increase in processing time during final finishing. To address this, we implemented a surface-refinement model based on spatial frequency-response characteristics within the Magnetorheological Finishing (MRF) process. By optimizing material removal according to specific spectral requirements, we significantly enhanced fabrication efficiency and reduced overall correction time. The resulting mirror achieved a figure error of 5 nm RMS and a reflectivity of 45%, as validated through X-ray beamline evaluations at the PAL.
Building on these foundational results, we are now expanding our capabilities toward a total in-house fabrication chain for 500 mm mirrors. This initiative aims to integrate the entire workflow—from bulk specimen preparation via DTM to robotic polishing and MRF correction. Our current research also focuses on developing specialized robotic tool heads and ensuring the metrological reliability of stitching interferometry. These efforts represent a critical step toward establishing a robust, independent manufacturing ecosystem for next-generation synchrotron X-ray optics.

Abstract
As the silicon-based chip industry approaches its physical limits, carbon-based semiconductors are considered one of the disruptive technologies of the “post Moore era”. Carbon nanotubes, with their excellent electrical properties, are expected to become interconnect materials for the next generation of large-scale integrated circuits. Due to challenges such as excessive channel size, irregular arrangement of carbon nanotubes, and high interface resistance between carbon nanotubes, there has been no qualitative breakthrough in the performance of carbon nanotube integrated circuits. The report will introduce the progress of laser micro nano manufacturing methods and mechanisms from the perspectives of nanochannel processing, regular self-assembly of carbon nanotubes, and controllable connection of carbon nanotubes.

Abstract
Lithography is one of the most complex processes in semiconductor manufacturing. To monitor and ensure lithography quality, the layer-to-layer lateral shift in process structures, known as overlay (OVL), serves as a critical performance indicator. As the technical node of lithography approaches the nanoscale limit, the allowable OVL budget becomes increasingly stringent. Consequently, precise and advanced OVL metrology techniques are essential for lithography process control. Currently, there are two mainstream optical methods for overlay metrology: diffraction-based overlay (DBO) and image-based overlay (IBO). In this presentation, we will report several research efforts aimed at advancing DBO and IBO techniques. First, nonuniform-exposure fusion methods are introduced for angle-resolved scatterometry-based overlay metrology. Next, a novel overlay indicator based on the polarization eigenstate of the retardance Mueller matrix will be presented for ellipsometry-based overlay metrology. We will then discuss the potential of computational intelligence methods to enhance overlay metrology performance, including neural network-driven IBO image enhancement with adaptive aberration correction, as well as intelligent design optimization for both Moiré fringe-based IBO targets and microstructure-array-based DBO targets. Finally, we will share insights on the opportunities and challenges in future research to further drive innovation in advanced OVL metrology techniques.