Generative Adversarial Network-Based Fault Detection in Semiconductor Equipment with Class-Imbalanced Data
Jeong Eun Choi, Da Hoon Seol, Chan Young Kim, Sang Jeen Hong
IF 3.5
Sensors
This research proposes an application of generative adversarial networks (GANs) to solve the class imbalance problem in the fault detection and classification study of a plasma etching process. Small changes in the equipment part condition of the plasma equipment may cause an equipment fault, resulting in a process anomaly. Thus, fault detection in the semiconductor process is essential for success in advanced process control. Two datasets that assume faults of the mass flow controller (MFC) in equipment components were acquired using optical emission spectroscopy (OES) in the plasma etching process of a silicon trench: The abnormal process changed by the MFC is assumed to be faults, and the minority class of Case 1 is the normal class, and that of Case 2 is the abnormal class. In each case, additional minority class data were generated using GANs to compensate for the degradation of model training due to class-imbalanced data. Comparisons of five existing fault detection algorithms with the augmented datasets showed improved modeling performances. Generating a dataset for the minority group using GANs is beneficial for class imbalance problems of OES datasets in fault detection for the semiconductor plasma equipment.
Wafer Type Ion Energy Monitoring Sensor for Plasma Diagnosis
Chansu Han, Yoonsung Koo, Jaehwan Kim, Kwang‐Wook Choi, Sang Jeen Hong
IF 3.5
Sensors
We propose a wafer-type ion energy monitoring sensor (IEMS) that can measure the spatially resolved distribution of ion energy over the 150 mm plasma chamber for the in situ monitoring of the semiconductor fabrication process. The IEMS can directly be applied to the semiconductor chip production equipment without further modification of the automated wafer handling system. Thus, it can be adopted as an in situ data acquisition platform for plasma characterization inside the process chamber. To achieve ion energy measurement on the wafer-type sensor, the injected ion flux energy from the plasma sheath was converted into the induced currents on each electrode over the wafer-type sensor, and the generated currents from the ion injection were compared along the position of electrodes. The IEMS operates without problems in the plasma environment and has the same trends as the result predicted through the equation.
Generative Adversarial Network-Based Fault Detection in Semiconductor Equipment with Class-Imbalanced Data
Jeong Eun Choi, Da Hoon Seol, Chan Young Kim, Sang Jeen Hong
IF 3.5
Sensors
This research proposes an application of generative adversarial networks (GANs) to solve the class imbalance problem in the fault detection and classification study of a plasma etching process. Small changes in the equipment part condition of the plasma equipment may cause an equipment fault, resulting in a process anomaly. Thus, fault detection in the semiconductor process is essential for success in advanced process control. Two datasets that assume faults of the mass flow controller (MFC) in equipment components were acquired using optical emission spectroscopy (OES) in the plasma etching process of a silicon trench: The abnormal process changed by the MFC is assumed to be faults, and the minority class of Case 1 is the normal class, and that of Case 2 is the abnormal class. In each case, additional minority class data were generated using GANs to compensate for the degradation of model training due to class-imbalanced data. Comparisons of five existing fault detection algorithms with the augmented datasets showed improved modeling performances. Generating a dataset for the minority group using GANs is beneficial for class imbalance problems of OES datasets in fault detection for the semiconductor plasma equipment.
Wafer Type Ion Energy Monitoring Sensor for Plasma Diagnosis
Chansu Han, Yoonsung Koo, Jaehwan Kim, Kwang‐Wook Choi, Sang Jeen Hong
IF 3.5
Sensors
We propose a wafer-type ion energy monitoring sensor (IEMS) that can measure the spatially resolved distribution of ion energy over the 150 mm plasma chamber for the in situ monitoring of the semiconductor fabrication process. The IEMS can directly be applied to the semiconductor chip production equipment without further modification of the automated wafer handling system. Thus, it can be adopted as an in situ data acquisition platform for plasma characterization inside the process chamber. To achieve ion energy measurement on the wafer-type sensor, the injected ion flux energy from the plasma sheath was converted into the induced currents on each electrode over the wafer-type sensor, and the generated currents from the ion injection were compared along the position of electrodes. The IEMS operates without problems in the plasma environment and has the same trends as the result predicted through the equation.
Effects of the Applied Power of Remote Plasma System With Green Alternative Chamber Cleaning Gas of Carbonyl Fluoride
Se Yun Jo, A. ‐H. A. Park, Sang Jeen Hong
IF 2.3
IEEE Transactions on Semiconductor Manufacturing
An effort to find an alternative dry-cleaning process gas with low global warming potential (GWP) has been conducted to decrease greenhouse gas emissions. Carbonyl fluoride (COF2) is one of the candidates as an alternative gas for plasma-enhanced chemical vapor deposition (PECVD) chamber cleaning because of its lower GWP compared to the currently employed NF3 gas. The dry-cleaning process conditions containing the power amount of the plasma source is related to the dissociation rate of the cleaning gas and dry-cleaning performance. We investigated the effects of the amount of remote plasma power to the chamber cleaning rate with COF2, and its effects with diluted gases of O2 and Ar. By the comparison of both numerical analysis and experiment, we found that the change of the amount of power induced different production rates of species in the gas mixture. In the case of O2 dilution, oxygen radicals prevail in the plasma, and it produces stable by-product of CO2 with the reaction of oxygen radicals to yield more fluorine atoms and radicals. We conclude that oxygen radicals have a significant role in the dissociation of the COF2, production of fluorine radicals, and it helps to reduce the amount of cleaning inhibitors such as C-C and C-F compounds. Additional dilution gases for cleaning gas affect production mechanisms and rates of species.
In-Situ Plasma Monitoring Using Multiple Plasma Information in SiO₂ Etch Process
Min Ho Kim, Jeong Eun Jeon, Sang Jeen Hong
IF 2.3
IEEE Transactions on Semiconductor Manufacturing
Optical emission spectroscopy (OES) data analysis with inert gas, called rare gas tracing method, has become a widely accepted method for the monitoring of plasma process. However, it is becoming less desirable due to the need for a higher hardmask selectivity in etch. Conventional OES analysis focuses on bulk plasma properties, such as electron temperature and density, but fail to capture the full complexity of etch rate changes influenced by both ohmic heating and ion acceleration. To address these limitations, we propose an alternative approach that incorporates multiple plasma information (PI), offering a more comprehensive view of plasma mechanisms. This new framework was applied to develop an OES-based monitoring technique without inert gases. By modulating source and bias powers to vary both ohmic heating and ion acceleration, the multiple PI model demonstrated a higher R2 score ( 0.97) compared to the traditional Ar-based PI model ( 0.8). In addition, explainable artificial intelligence (XAI) indicated that multiple PI had greater importance, demonstrating its effectiveness in monitoring etch rates in non-inert gas processes. It not only detects changes in the etch process, but also identifies whether the variations stem from chemical or physical reactions to be useful for advanced process control.
Impedance Monitoring of Capacitively Coupled Plasma Based on the Vacuum Variable Capacitor Positions of Impedance Matching Unit
H. Kim, Jiseok Lee, Sang Jeen Hong
IF 2.6
Electronics
Plasma impedance monitoring in semiconductor manufacturing processes is performed using external sensors, such as voltage-current (VI) probes or directional couplers. Plasma chamber impedance measurements, conducted in non-50 Ω matched transmission lines, suffer from a lack of clean signals due to phase variations and the nonlinearity of plasma, thus, sensor calibration is required for each installment. In this study, we monitored plasma impedance in situ based on the position of the vacuum variable capacitor within the matching network, without employing an external VI probe. We observed changes in the matching position according to parameter variations and subsequently confirmed that the calculated plasma impedance also varied accordingly. This study demonstrates the feasibility of real-time plasma impedance monitoring under 50 Ω-matched conditions without the use of external sensors, thereby simplifying plasma diagnostics.
Understanding the Chamber Wall-Deposited Thin Film of Plasma Deposition Equipment for the Efficiency of In Situ Dry-Cleaning
J. O. Lee, Jiwon Jang, Sang Jeen Hong
IF 2.8
Coatings
In plasma-enhanced chemical vapor deposition (PECVD) processes, thin films can accumulate on the inner chamber walls, resulting in particle contamination and process drift. In this study, we investigate the physical and chemical properties of these wall-deposited films to understand their spatial variation and impact on chamber maintenance. A 6-inch capacitively coupled plasma (CCP)-type PECVD system was used to deposit SiO2 films, whilst long silicon coupons were attached vertically to the chamber side walls to collect contamination samples. The collected contamination samples were comparatively analyzed in terms of their chemical properties and surface morphology. The results reveal significant differences in hydrogen content and Si–O bonding configurations compared to reference films deposited on wafers. The top chamber wall, located near the plasma region, exhibited higher hydrogen incorporation and larger Si–O–Si bonding angles, while the bottom wall exhibited rougher surfaces with larger particulate agglomerates. These variations were closely linked to differences in gas flow dynamics, precursor distribution, and the energy state of the plasma species at different chamber heights. The findings indicate that top-wall contaminants are more readily cleaned due to their high hydrogen content, while bottom-wall residues may be more persistent and pose higher risks for particle generation. This study provides insights into wall contamination behavior in PECVD systems and suggests strategies for spatially optimized chamber cleaning and conditioning in high-throughput semiconductor processes.
Virtual Metrology of Multiple Dielectric Layer Thickness for 3D-NAND Deposition Process
Hye Eun Sim, Min Uk Lee, Sang Jeen Hong
IF 2.3
IEEE Transactions on Semiconductor Manufacturing
With the growing emphasis on three-dimensional (3D) vertical structures for enhancing the storage capacity and performance of 3D-NAND flash memory devices, precise control and prediction of process results to minimize process variability are important. Herein we predicted the layer thickness of an oxide/nitride (ON) dielectric stack for a 3D-NAND deposition process using artificial intelligence (AI). We investigated the key variables influencing the thicknesses of multiple dielectrics to propose strategies for mitigating thickness variation. We constructed a virtual metrology (VM) model based on status-variable identification (SVID) and optical emission spectroscopy (OES) data collected from plasma deposition equipment, and employed explainable AI (XAI) algorithms to interpret the significance of variables affecting the process results. XAI also supports the reliability of AI-predictive models for determining the thicknesses of the deposited multi-layered ON stack. Using variables derived from the SVID and OES data, the models for predicting oxide layer thickness, nitride layer thickness, and the thicknesses of both oxide and nitride layers achieved accuracies of 99%, 88% and 99%, respectively. This study highlights the importance of developing high-performance VM models and interpreting predictive outcomes for precise process control in semiconductor plasma processes.
An Alternative PECVD Chamber Cleaning Gas of COF<sub>2</sub> for Low-GWP Consideration
Ah Hyun Park, Y. Lee, Seyun Jo, Sang Jeen Hong
IF 2.3
IEEE Transactions on Semiconductor Manufacturing
Continuous deposition processes in PECVD environments are critical for ensuring the uniformity and reproducibility of thin films across various applications. Silicon dioxide (SiO2), widely used in these processes for its excellent properties, can leave residual materials in PECVD chambers, leading to material buildup that compromises process consistency and reproducibility. A representative example of compromised process consistency and reproducibility is found in the manufacturing of 3D-NAND flash memory, which involves oxide-nitride (ON) stacking processes. Effective chamber cleaning is essential to ensure consistent and reproducible performance in continuous deposition processes. Nitrogen trifluoride (NF3), a commonly used as chamber cleaning gas, is expected to be newly belong to the greenhouse gas regulations due to its high global warming potential (GWP), which may pose both environmental and industrial risks. In this study, we explored the potential of carbonyl fluoride (COF2) as an alternative chamber cleaning gas with low GWP, albeit with an inferior cleaning rate compared to NF3. This study investigates gas dissociation in the plasma environment and analyzes plasma species and changes in the deposited film surface affecting the cleaning rate. Based on the results, proposed improvements are made to the cleaning process design for COF2, considering factors influencing plasma enhanced chemical vapor deposition (PECVD) chamber cleaning efficiency.
SSUKF-FA-RBF: A Kalman-Enhanced High-Precision Positioning Framework for BeiDou Navigation Using Firefly-Optimized Neural Estimation
Liang Li, Sang Jeen Hong, Xueqin Liu
Informatica
This study addresses the high-precision positioning requirements of the BeiDou Navigation System (BDS) by focusing on the commonly adopted BDS/Inertial Navigation System integrated navigation mode. A novel Spherical Simplex Unscented Kalman Filter (SSUKF) algorithm is proposed, featuring an improved sigma-point sampling strategy that enhances filtering accuracy while reducing computational overhead. In parallel, the Time Difference of Arrival (TDOA) method is combined with the Firefly Algorithm (FA) to optimize a Radial Basis Function (RBF) neural network, further enhancing positioning precision. Evaluation is conducted using an Ultra-Wideband TDOA dataset. Results show that the SSUKF algorithm significantly reduces positioning error. Specifically, the root means square error (RMSE) achieved by SSUKF is 0.1614 m-a reduction of 62.2% compared to the Extended Kalman Filter and 52.1% compared to the Unscented Kalman Filter. When integrated with the FA-optimized RBF neural network, the hybrid SSUKF-FA-RBF model achieves an RMSE of 0.127 m under high-noise conditions, demonstrating strong robustness and accuracy. In addition to its accuracy, the SSUKF algorithm offers improved computational efficiency, making it suitable for real-time, high-precision applications. Error analysis confirms the robustness and stability of the SSUKF-FA-RBF model across various environments. Under zero standard deviation noise, the model achieves 96.4% accuracy, 95.6% precision, and a 96.1% recall ratesubstantially outperforming comparative models. This study contributes an enhanced Kalman filtering method and an optimized positioning framework, advancing both accuracy and computational efficiency for the BDS. The proposed approach offers effective technical support for a wide range of high-precision positioning applications.
