In the past, display devices have undergone many changes, such as plasma TVs and LCDs, and have continued to develop. Recently, new display technologies, such as Organic Light Emitting Diode displays and Inorganic Light Emitting Diode displays, have been developed. Among them, Micro LED displays have the potential to improve performance more than LCDs and OLEDs, but a lot of effort and time are needed until the mass production technology (transfer and bonding) of Micro LED displays is developed. We have developed a new Micro LED display light source that can be produced using existing transfer and bonding process technologies to enable faster commercialization of Micro LED in the industry. This light source is TFT deposition on LED. TFT deposition on LED has the advantage of being able to produce displays using existing process technology, making early commercialization of display application products possible. In this study, we applied the Active Driving method to verify the performance of TFT deposition on LED as a display to determine its commercialization potential. Additionally, to facilitate faster application of Micro LED in the industry, we applied TFT deposition on LED to Optical Wireless Communication systems, which are widely used in application service areas such as safety/security and sensors, to verify its communication performance. The experimental results confirmed that TFT deposition on LED is not only capable of AM driving but can also be applied to OWC systems.

High-potential data can be predicted and stored in the cache to prevent cache misses, thus reducing the processor's request and wait times. As a result, the processor can work non-stop, hiding memory latency. By utilizing the temporal/spatial locality of memory access, the prefetcher introduced to improve the performance of these computers predicts the following memory address will be accessed. We propose a prefetcher that applies the GRU model, which is advantageous for handling time series data. Display the currently accessed address in binary and use it as training data to train the Gated Recurrent Unit model based on the difference (delta) between consecutive memory accesses. Finally, using a GRU model with learned memory access patterns, the proposed data prefetcher predicts the memory address to be accessed next. We have compared the model with the multi-layer perceptron, but our prefetcher showed better results than the Multi-Layer Perceptron.

EtherCAT is an Ethernet-based fieldbus system standardized in IEC 61158 and SEMI, and widely used in the fields of factory automation, semiconductor equipment and robotics. In this paper, without operating system, we have implemented an EtherCAT master with an open source EtherCAT master stack SOEM on STM32 Nucleo-144 board with an STM32F767 microcontroller. And its jitter performance has been evaluated at the output of the network port to include all the effects of the entire system in the results. The results show that the implemented EtherCAT master has precise control performance for control frequencies from 1KHz to 8KHz and relatively superior jitter performance compared to the EtherCAT masters with real-time patched Linux operating system.

In semiconductor manufacturing processes, a cleaning process is important that can remove sub-micron particles. Conventional wet cleaning methods using chemical have limits in removing nano-particles. Thus, physical forces of a mechanical vibration up to 1 MHz frequency, was tried to aid in detaching them from the substrates. In this article, we developed a 1 MHz quartz megasonic for a bare wafer cleaning using finite element analysis. At first, a 1 MHz megasonic prototype was manufactured. Using the results, a main product which can improve a particle removal performance, was analyzed and designed. The maximum impedance frequency was 992 kHz, which agreed well with the experimental value of 986 kHz (0.6% error). Acoustic pressure distributions were measured, and the result showed that maximum / average was 400.0~432.4%, and standard deviation / average was 46.4~47.3%. Finally, submicron particles were deposited and cleaned for the assessment of the system performance. As a result, the particle removal efficiency (PRE) was proved to be 92% with 11 W power. Reflecting these results, the developed product might be used in the semiconductor cleaning process.

In this paper, we propose a real-time bus arrival notification system using deep learning to guarantee movement rights for the visually impaired. In modern society, by using location information of public transportation, users can quickly obtain information about public transportation and use public transportation easily. However, since the existing public transportation information system is a visual system, the visually impaired cannot use it. In Korea, various laws have been amended since the 'Act on the Promotion of Transportation for the Vulnerable' was enacted in June 2012 as the Act on the Movement Rights of the Blind, but the visually impaired are experiencing inconvenience in using public transportation. In particular, from the standpoint of the visually impaired, it is impossible to determine whether the bus is coming soon, is coming now, or has already arrived with the current system. In this paper, we use deep learning technology to learn bus numbers and identify upcoming bus numbers. Finally, we propose a method to notify the visually impaired by voice that the bus is coming by using TTS technology.

Since dishing in the CMP process is a major factor that hinders the uniformity of the semiconductor thin film, many studies have focused this issue to improve the non-uniformity of the film due to dishing. In the metal layer, the dishing mainly occurs in the central part of the metal due to a difference in a selection ratio between the metal and the dielectric, thereby generating a step on the surface of the metal layer. Factors that cause dishing include the shape of the thin film, the chemical reaction of the slurry, thermal deformation, and the rotational speed of the pad and head, and dishing occurs due to complex interactions between them. This study analyzed the stress generated on the metal layer surface in the CMP process using ANSYS software, a commercial structure analysis program. The stress caused by the vertical load applied from the pad was analyzed by changing the area density and line width of the dummy metal. As a result of the analysis, the stress in the active region decreased as the pattern density and line width of the dummy metal increased, and it was verified that it was valid compared with the previous study that studied the dishing according to the dummy pattern density and line width of the metal layer. In conclusion, it was confirmed that there is a relationship between dishing and normal stress.

Silicon oxide thin films have been deposited by plasma-enhanced chemical vapor deposition in SiH4 and N2O plasma along the variation of the gas flow ratio. Optical emission spectroscopy was employed to monitor the plasma and ellipsometry was employed to obtain refractive index of the deposited thin film. The atomic ratio of Si, O, and N in the film was obtained using XPS depth profiling. Fourier Transform Infrared Spectroscopy was used to analyze structures of the films. RI decreased with the increase in N2O/SiH4 gas flow ratio. We noticed the increase in the Si-O-Si bond angles as the N2O/SiH4 gas flow ratio increased, according to the analysis of the Si-O-Si stretching peak between 950 and 1,150 cm-1 in the wavenumber. We observed a correlation between the optical emission intensity ratio of (ISi+ISiH)/IO. The OES intensity ratio is also related with the measured refractive index and chemical composition ratio of the deposited thin film. Therefore, we report the added value of OES data analysis from the plasma related to the thin film characteristics in the PECVD process.

The defect detection based on deep learning requires minimal loss and high accuracy to pinpoint product defects. In this paper, we confirm the loss rate of deep learning training based on disc-shaped artificial scaffold images. It is intended to compare the performance of Cross-Entropy functions used in object detection algorithms. The model was constructed using normal, defective artificial scaffold images and category cross entropy and sparse category cross entropy. The data was repeatedly learned five times using each loss function. The average loss rate, average accuracy, final loss rate, and final accuracy according to the loss function were confirmed.

When blockchain technology, which shows various applicability, is utilized as a component of a database system, the characteristics of open verification and integrity/transparency of blockchain technology can bring new functionalities or enhanced results to the existing database system. However, when applying this blockchain technology to a database system, the cost versus expected effect in various performance perspectives must be evaluated. These costs include execution time and required storage space, and the performance of the converged system may vary in analysis method depending on the configuration method of the characteristics of the blockchain. This paper aims to propose an analysis model for the entire architecture by considering aspects that are not considered in the performance analysis models of database systems and the unique characteristics of blockchains. In doing so, we are trying to build a theoretical framework as an important conceptual technique that should be considered in the evaluation process of the performance results that can be obtained through the utilization of blockchain components in database systems. What we hope is that this work is expected to provide a useful foundation for researchers interested in the convergence of database systems and blockchain technology in order to construct a system with new future functionalities.

Due to the unavailability of global positioning system (GPS) indoors, various indoor pedestrian positioning methods have been designed to estimate the position of the user using received signal strength (RSS) measurements from radio beacons, such as wireless fidelity (WiFi) access points and Bluetooth low energy (BLE) beacons. In indoor environments, radio-frequency (RF) signals are unpredictable and change over space and time because of multipath associated with reflection and refraction, shadow fading caused by obstacles, and interference among different devices using the same frequencies. Therefore, the outliers in the positional information obtained from the indoor positioning method based on RSS measurements occur often. For this reason, the performance of the positioning method can be degraded by the characteristics of the RF signal. To resolve this issue, a map-matching (MM) algorithm based on maximum probability (MP) estimation is applied to the indoor positioning method in this study. The MM algorithm locates the aberrant position of the user estimated by the positioning method within the limits of the adjacent pedestrian passages. Empirical experiments show that the positioning method can achieve higher positioning accuracy by leveraging the MM algorithm.

This paper presents a method using finite element analysis and topology optimization to address the issue of overdesign in pressure-explosion proof enclosure doors for semiconductor manufacturing processes. The design conducted in this paper focuses on the pattern design of the enclosure door and its fixation components. The process consists of a solid-filled model, a topology optimization model, and a post-processing model. By applying environmental conditions to each model and comparing the maximum displacement, maximum equivalent stress, and weight values, it was confirmed that a reduction of about 13% in weight is achievable.

If an impedance mismatch occurs between the source and load in a Radio Frequency transmission system, reflected power is generated. This results in incomplete power transmission and the generation of Reflected Power, which returns to the Radio Frequency generator. To minimize this Reflected Power, Impedance matching is performed. Fast and efficient Impedance matching, along with converging reflected power towards zero, is advantageous for achieving desired plasma characteristics in semiconductor processes. This paper explores Impedance matching by adjusting the Vacuum Variable Capacitor of an L-type Matching Module based on the trends observed in the voltage of the Phase Sensor and Electromotive Force voltage. After assessing the impedance matching characteristics, the findings are described.

The mobile card system used for mobile access control is connected to the door lock mounted on the door, enabling non-contact control. The RSSI (Received Signal Strength Indicator) of the BLE (Bluetooth Low Energy) communication packet used here can help to know the direction and distance of the mobile device. In this study, the desirable access control distance was calculated and implemented by setting the RSSI of the transmitter of the BLE packet used in mobile access control and processing the RSSI of the receiver.

Since fires in uncontrolled environments pose serious risks to society and individuals, many researchers have been investigating technologies for early detection of fires that occur in everyday life. Recently, with the development of deep learning vision technology, research on fire detection models using neural network backbones such as Transformer and Convolution Natural Network has been actively conducted. Vision-based fire detection systems can solve many problems with physical sensor-based fire detection systems. This paper proposes a fire detection method using the latest YOLOv8, which improves the existing fire detection method. The proposed method develops a system that detects sparks and smoke from input images by training the Yolov8 model using a universal fire detection dataset. We also demonstrate the superiority of the proposed method through experiments by comparing it with existing methods.

Recently, there has been an increased demand for light-emitting diode (LED) due to the growing emphasis on environmental protection. However, the use of GaN-based sapphire in LED manufacturing leads to the generation of defects, such as dislocations caused by lattice mismatch, which ultimately reduces the luminous efficiency of LEDs. Moreover, most inspections for LED semiconductors focus on evaluating the luminous efficiency after packaging. To address these challenges, this paper aims to detect defects at the wafer stage, which could potentially improve the manufacturing process and reduce costs. To achieve this, image processing and deep learning-based defect detection techniques for Sapphire Epi-Wafer used in Green LED manufacturing were developed and compared. Through performance evaluation of each algorithm, it was found that the deep learning approach outperformed the image processing approach in terms of detection accuracy and efficiency.

TiO₂ thin films were grown using the Atomic Layer Deposition (ALD) and their structural and electrical properties were investigated. The crystal structure, dielectric constant, and surface roughness of the TiO₂ thin films grown by the ALD deposition method were studied. The grown TiO₂ thin films showed an anatase crystal structure, and their properties varied with temperature. In particular, the properties of the TiO₂ thin films were confirmed by changing the process temperature. The electrical properties of Metal-Insulator-Silicon (MIS) capacitor structures were analyzed using a probe station. The performance improvement of capacitors using TiO₂ as a dielectric was confirmed by measuring capacitance through Capacitance-Voltage (C-V) curves.

This paper presents a robust position tracking controller for a motor used in semiconductor equipment, utilizing the motor angle measurement. Precise position control is challenging due to the presence of uncertainties in various motor applications. The proposed controller consists of a PD (Proportional-Derivative) controller and a PIO (Proportional-Integral Observer) to estimate the system's state and equivalent disturbance compensating for the uncertainties. Since the stability alternates as the observer gain increases, we have investigated it through the closedloop root locus under the system parameters change. The analysis has showed that the inertia of the motor is the main parameter that affects it, and by adjusting the control gain appropriately, the system can be rendered to be stable even when the inertia of the motor changes. The effectiveness of the proposed control algorithm is validated through computer simulations, followed by a comparison of its performance with the results of a previous study.

The semiconductor industry, which relies on global supply chains, has recently been facing longer lead time for material procurement due to supply chain uncertainties. Moreover, since increasing customer satisfaction and reducing inventory costs are in a trade-off relationship, it is challenging to determine the appropriate safety stock level under demand and lead time uncertainties. In this paper, we propose a framework for determining safety stock levels by utilizing the optimization method to determine the optimal safety stock level. Additionally, we employ a linear regression method to analyze customer satisfaction scores and inventory costs based on variations in lead time and demand. To verify the effectiveness of the proposed framework, we compared safety stock levels obtained by the regression equations with those of the conventional method. The numerical experiments demonstrated that the proposed method successfully reduces inventory costs while maintaining the same level of customer satisfaction when lead time increases.

In urban environments, signals of Global Positioning System (GPS) can be blocked and reflected by tall buildings, large vehicles, and complex components of road network. Therefore, the performance of the positioning system using the GPS module in urban areas can be degraded due to the loss of GPS signals necessary for the position estimation. To deal with this issue, various localization schemes using inertial measurement unit (IMU) sensors, such as gyroscope and accelerometer, and Bayesian filters, such as Kalman filter (KF) and particle filter (PF), have been designed to enhance the performance of the GPS-based positioning system. Among Bayesian filters, the PF has been widely used for the target tracking and vehicle navigation, since it can provide superior performance in estimating the state of a dynamic system under nonlinear/non-Gaussian circumstance. This paper presents a positioning system that uses the double-stacked particle filter (DSPF) as well as the accelerometer, gyroscope, and GPS receiver on the smartphone to provide higher pedestrian positioning accuracy in urban environments. The DSPF employs a nonparametric technique (Parzen-window) to create the multimodal target distribution that approximates the posterior distribution. Experimental results show that the DSPF-based positioning system can provide the significant improvement of the pedestrian position estimation in urban environments.

Recently, metaverse has attracted great attention worldwide with the development of related technologies and the demand for on-tact systems. The ideal metaverse should constitute a virtual world more similar to the realistic society with physical interactions. However, there are many constraints when we construct a virtual world using the existing commercialized metaverse platform. In this paper, we design and implement a university campus-based metaverse which minimizes the constraints, and create a virtual environment maps very similar to the real-world using Unity engine, Photon cloud and Playfab SDK for multi-user accessible to virtual world as a personal avatar. Also, this work includes attractive interaction based on exchange between users. Lastly, we illustrate our implemented metaverse works on Android phone, and it works without difficulty in terms of performance.

Recently, with the development of deep learning technology, research on pedestrian attribute recognition technology using deep neural networks has been actively conducted. Existing pedestrian attribute recognition techniques can be obtained in such a way as global-based, regional-area-based, visual attention-based, sequential prediction-based, and newly designed loss function-based, depending on how pedestrian attributes are detected. It is known that the performance of these pedestrian attribute recognition technologies varies greatly depending on the type of backbone network that constitutes the deep neural networks model. Therefore, in this paper, several backbone networks are applied to the baseline pedestrian attribute recognition model and the performance changes of the model are analyzed. In this paper, the analysis is conducted using Resnet34, Resnet50, Resnet101, Swin-tiny, and Swinv2-tiny, which are representative backbone networks used in the fields of image classification, object detection, etc. Furthermore, this paper analyzes the change in time complexity when inferencing each backbone network using a CPU and a GPU.

This study the problem of deposition uniformity observed during Mist-CVD deposition experiments. The TRIZ's ADRIGE algorithm, a problem-solving technique, is utilized to systematically analyze the issue and propose solutions. Through problem and resource analysis, technical contradictions are identified regarding the precursor's volume and its path when it encounters the substrate. To resolve these contradictions, the concept of applying the principle of dimensional change to transform the precursor's three-dimensional path into a one-dimensional path is suggested. The chosen solution involves the design of an enhanced Mist-CVD system, which is evaluated for feasibility and analyzed using computational fluid dynamics. The analysis confirms that the deposition uniformity consistently follows a pattern and demonstrates an improvement in uniformity. The improved Mist-CVD equipment is validated through analysis, providing evidence of its feasibility and yielding satisfactory results.