MLMyungjun Lee

Myungjun Lee

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Myungjun Lee has a total of 220 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Special Section Guest Editorial: High-Density Sampling and Massive Metrology

    by Myungjun Lee, Daniel Schmidt

    Guest Editors Myungjun Lee and Daniel Schmidt introduce the Special Section on High-Density Sampling and Massive Metrology.


  • AI-powered super-resolution and denoising of scanning electron microscopy images for high-throughput metrology and inspection

    by Leslie Ching Ow Tiong, Jhongkwon Kim, Daewon Choi, Chaehoon Park, Junhui Lee, Songhyun Yu, Yeny Yim, Wonhee Lee, Jaehoon Lee, Suil Son, et al.

    Scanning electron microscopy (SEM) imaging plays a crucial role in metrology and inspection (MI) systems across various fields, including semiconductor manufacturing and nanotechnology. This method benefits for providing detailed surface and compositional analysis with a high-quality (HQ) image. Yet, producing HQ images with SEM imaging not only requires a long acquisition time but also introduces new challenges for maintaining reasonable throughput, which potentially disrupts the MI workflows. To leverage the semiconductor industry with a better approach, we propose an effective artificial-intelligence-powered image enhancement solution, namely, MI-image enhancement solution (MI-IES), that integrates deep learning techniques. MI-IES is designed to enhance the efficiency of massive metrology applications within MI workflows by processing low-quality SEM images as an input target and subsequently processing them using image enhancement algorithms. Specifically, MI-IES contains two main modules: super-resolution and denoising, to overcome the stated challenges in our study. Extensive experiments across various actual devices were conducted using SEM imaging. The results highlight MI-IES enhanced capability in massive metrology applications, as evidenced by the superior R2 analysis that yielded a score ≥0.90 in terms of SEM image quality. In addition, MI-IES achieved remarkable throughput processing time by a decrease of up to 75% (3.9 times faster) in comparison to HQ, underscoring its efficiency. Our solution not only delivers superior performance but also benefits massive metrology with sufficient throughput and high structural integrity of the SEM imaging.

  • Massive metrology at wafer and die level for hybrid bonding in VNAND

    by Samantha Mazzamuto, Seokhan Kim, KyungSouk Yi, Su-Young Lee, Seok Park, SungHoon Park, Kyong Wook Noh, Min Hong, Kwang Soo Kim, Myungjun Lee, et al.

    Hybrid bonding is a critical part of advanced packaging development, enabling wafer-to-wafer bonding and die-to-wafer bonding for 3D stacking of chips. This process, which is carried out post-individual wafer fabrication, can suffer from void formation either between copper PADs or between the dielectric, leading to failed interconnections in the former case, affecting performance at the die level and rupture of the dielectric layers in the latter case, leading to significant yield loss at the wafer level. Control of critical parameters such as surface roughness, degree of cleanliness, flatness, recess level of the copper PADs, and dielectric surface is necessary at the wafer level, essentially calling for massive metrology. We present such a massive metrology measurement performed on production wafers in the fab using a high-throughput in-line Atomic Force Microscopy-based metrology tool. Measurements were performed on three wafers, which were designed to be the “top” wafers during the bonding step, where a split condition was applied on these wafers, whereas an additional three wafers were measured, which were to be used as the “bottom” wafers during bonding and were all in POR condition. The analysis of the results obtained is also presented for understanding the expected impact on the final performance of the bonding step.

  • Full-wafer 3D metrology solution based on imaging ellipsometry for process uniformity control

    by Hyukjoon Cho, Yong-Ju Jeon, Jhongkwon Kim, Jinwoo Ahn, Jaewon Lee, Taeyong Jo, Juntaek Oh, Changhyeong Yoon, Jaehyeon Son, Eunsoo Hwang, et al.

    As semiconductor manufacturing technology becomes more complex, process control becomes more important, and in particular, technologies for discovering and controlling structural defects become more important. However, considering the productivity of manufacturing, it is impossible to achieve full coverage measurement with existing measurement technologies. Here, we propose imaging ellipsometry-based full wafer coverage metrology (IFM) that acquires spectra of all cell blocks of DRAM devices using large-area imaging ellipsometry measurements with a 22 mm x 22 mm field of view (FOV) and 6.5 μm spatial resolution. A prediction model learned from tens to hundreds of reference data points and the acquired spectra helps generate a massive amount of measurement data exceeding 12 million points per wafer. The vast amount of data generated allows us to extend the spatial coverage of existing thickness measurements, which typically cover only 0.001%, to up to 47%. The proposed method can play a critical role in detecting suspected equipment failures, especially when repetitive and periodic patterns indicating problems appear in the fault map. This novel approach identified and mitigated equipment-induced distributions that negatively impacted product yield that were difficult to observe using conventional metrology. The effectiveness of this approach was verified through measurements performed over 100 steps on the DRAM device.


  • Nondestructive in-cell conductivity metrology for memory semiconductors using near-field THz spectroscopy

    by Sunhong Jun, Inkeun Baek, Hyunsoo Kwak, Hyeong Seok Yun, Yoonkyung Jang, Seonmyeong Kim, Hanjung Ryu, Wontae Kim, Martin Priwisch, Sooyeong Lee, et al.

    We have developed a non-destructive, in-line wafer metrology technique using terahertz (THz) waves that overcomes conventional limitations in the handling of electrical properties in the thin film processes. By combining a near-field microprobe with THz time-domain spectroscopy (TDS) and establishing an accurate alignment system, we have achieved sub-10 μm resolution. It has been integrated as an in-line metrology system in a memory production line and demonstrated on device wafers. Our signal correction method effectively suppresses signal variations in the lower part of the multi-stack enabling the non-invasive measurement of metal film resistivity through THz TDS, which is used as the gate metal or various contacts. This was utilized to non-destructively measure resistance in the cell, where the actual transistors are formed during the memory semiconductor integration process. Currently, the sheet resistance (Rs) is measured at the observation site (OS) with a 4-point probe (4PP), but as the distance from the OS increases, the resistance value changes and the representativeness of the actual cell decreases. We found that after the tungsten contact deposition process of vertical NAND flash, the agreement with the Rs of the OS drops to R2 0.42 in the cell area away from the OS. On the other hand, using THz transmittance, we were able to measure Rs in the cell without contact, and found that the agreement with Rs of 4PP at the same location was R2 0.97. In the case of the TiN metal gate of a DRAM cell transistor, it is not possible to measure the electrical properties with 4PP after the etching process and thermal process where the actual structure is formed, but it is possible with THz TDS. The RsTHz obtained from the THz transmittance shows that the resistivity changes significantly after the etching and thermal process from the initial metal deposition state.

  • Semiconductor image analysis using data-efficient machine learning

    by Songhyun Yu, Sangho Yoon, Sungjin Lim, Hoonseok Park, Subong Shon, Suil Son, Chulmoo Kang, Yoonsung Bae, Hansaem Park, Sungsik Jo, et al.

    Recent advancements in AI have intensified interest in automating Transmission/Scanning Electron Microscopy (TEM/SEM) image analysis for semiconductor metrology and inspection (MI). However, the scarcity of annotated datasets in the industry poses a critical challenge, as model performance heavily relies on the quantity of precisely labeled training data. To address this limitation, we propose a data-efficient machine learning framework optimized for structural and categorical semiconductor image analysis. Our approach integrates three key innovations: (1) Transfer learning of a foundation model for semiconductor image segmentation, which outperforms generic counterparts by 0.1920 in mIoU; (2) Semi-supervised learning with TEM-specialized augmentation strategy with limited labeled data, achieving competitive mIoU of 0.877 even with a few labeled examples; and (3) Self-supervised contrastive pre-training for SEM defect image classification, leveraging magnification-invariant consistency to enhance feature robustness. Evaluated under limited-data regimes, our framework demonstrates superior performance over conventional methods while significantly reducing annotation dependency. This solution directly tackles the semiconductor industry’s persistent challenge of high labeling costs and rapidly expanding unlabeled data pools, offering a practical pathway for AI adoption in semiconductor developments.


  • Open Access

    Ultra-Wide-Field Imaging Mueller Matrix Spectroscopic Ellipsometry for Semiconductor Metrology

    by Taeyong Jo, Juntaek Oh, Jaehyeon Son, Chnaghyeong Yoon, Eunsoo Hwang, Jinwoo Ahn, Jaewon Lee, Jinsoo Lee, Jiyong Shin, Donggun Lee, et al.

    We propose an ultra-wide-field imaging Mueller matrix spectroscopic ellipsometry (IMMSE) system for semiconductor metrology. The IMMSE system achieves large-area measurements with a 20 mm × 20 mm field of view (FOV)—the largest FOV reported to date—and a spatial resolution of 6.5 µm. It enables the acquisition of over 10 million Mueller matrix (MM) spectra within the FOV, while a unique signal correction algorithm ensures spectrum consistency across the FOV. Leveraging this numerous MM spectra and machine learning, spatially dense metrology across the entire wafer area is achieved. This approach provides over 12,800 times more metrology data and 580 times higher throughput compared to conventional point-based methods, such as scanning electron microscopy. We experimentally demonstrate the potential of the IMMSE for yield enhancement in semiconductor manufacturing by identifying spatial variations of dynamic random access memory (DRAM) structures within individual chips as well as across the wafer.

  • Blind Image Deblurring with Noise-Robust Kernel Estimation

    by Chanseok Lee, Jeongsol Kim, Seungmin Lee, Jaehwang Jung, Yunje Cho, Taejoong Kim, Taeyong Jo, Myungjun Lee, Mooseok Jang


  • Massive on-cell overlay metrology solution based on Mueller-matrix analysis for advanced process control

    by Changhyeong Yoon, Hyukjoon Cho, Ji Yong Shin, Kihun Kang, Juntaek Oh, Donggun Lee, Jaehyun Son, Seunga Lim, Eunsoo Hwang, Jinsoo Lee, et al.

    We have developed a massive on-cell overlay metrology system based on Mueller matrix measurements. By integrating microscopic techniques into ellipsometry, we achieved high-throughput and extensive sampling coverage, with 1-shot/field per 1-field of view (FOV) measurement capability within a 34 x 34 mm2 FOV. Analyzing the off-diagonal components of the Mueller matrix allowed for on-cell overlay measurement across the wafer. This system provides measurement sensitivity comparable to e-beam-based technologies while offering high coverage, enabling precise reticle correction or high-order overlay correction in photolithography processes. This advancement represents a significant improvement in overlay metrology, offering both sensitivity and resolution for enhanced semiconductor manufacturing processes.

  • Open Access

    Detection of defective chips from nanostructures with a high-aspect ratio using hyperspectral imaging and deep learning

    by Sunhong Jun, Wonjun Choi, Yong-Ju Jeon, Jeongsu Ha, Kyuhwan Kim, Sungyoon Ryu, Myungjun Lee, Yongdeok Jeong, Younghoon Sohn

    We have developed an imaging spectroscopic reflectometry (ISR) method based on hyperspectral imaging and deep learning to detect defects in the bottom region of high-aspect-ratio nanostructures. ISR enables fast and non-destructive imaging of the bottom critical dimension (BCD) of channel holes (CHH) on a chip die of vertical NAND (V-NAND). A supervised learning model is built to predict the BCD by associating a pre-measured hyperspectral cube with scanning electron microscopy images after decapsulation of the top of the sample. The BCD predicted by ISR shows a high correlation of R2=0.72 with the actual BCD, and the distribution of CHH not open (NOP) defects on the chip die identified by bright field inspection after decapsulation is consistent with the BCD image obtained by ISR. In addition, ISR can detect defects that occur at arbitrary positions relative to the optical critical dimension (OCD) of the die. On fully integrated V-NAND chips, the ISR result showed a high correlation (R2=0.82) with the failure rate caused by CHH NOP, while the conventional spot OCD showed only R2=0.41. Thus, ISR can be used to optimize the etch process for weak wafer edge regions and to detect defective etch equipment.