Machine Engineering

• Advanced Equipment & Process Quality through Pre-securing Diagnostic Capabilities for Newly Applied Technologies.
• Need for equipment diagnostic tech. will continue ➔ Increasing complexity and difficulty of core tech. ➔ Need for Revolutionary diagnostic & analysis tech. ➔ Development of a proprietary automated/agentic diagnostic & analysis system ➔ Advancement of Equipment/Process Quality ➔ Ultimately, intended for application in a Smart/Autonomous Fab

Necessity of Technical Diagnosis for Equipment Components

• Increased complexity & difficulty in upgrading specifications and performance for equip. and component technologies.
• Technology gap: Speed of new technology adoption >> Speed of diagnostic technology development.
• "Black-box" nature of new technologies: Difficulty in securing and analyzing hardware specifications, performance/quality verification data, and process-related s-parameters

Integrated Diagnostic Framework for Equipment & Process Quality Enhancement

• Quality Enhancement ➔ Maximizing equip./component quality(HW TTTM, specification optimization, durability assurance) and process quality(etch/deposition rates, uniformity, reproducibility, yield) through core diagnostic technologies.
• We aim to continuously enhance our diagnostic capabilities by establishing correlations through mutual feedback between equipment/parts quality diagnostic data and process results, ultimately internalizing our proprietary diagnostic technology.

Revolutionary Path of future Equipment-Process Diagnosis

• Evolutionary Path ➔ Build diagnostic infrastructure aligned with equipment hardware specifications (RF Power, Gas, Chemistry, Temp, etc.), Invasive-Type, Compatible commercial sensors, Utilization of equipment FDC
• Revolutionary Path (Paradigm Shift) ➔ ① Non-Invasive, ② Miniaturization, ③ Target-specific specialized sensors, ④ Direct diagnostics using wireless wafer sensors, ⑤ Internalization through in-house development of high-performance sensors with high sensitivity & precision & resolution, ⑥ optimization of customized algorithms for each process tool, and ⑦ Advancement of diagnostic agents (AI Assistant ➔ AI Agent ➔ Agentic AI)
• "Black-box" nature of new technologies: Difficulty in securing and analyzing hardware specifications, performance/quality verification data, and process-related s-parameters
• As process sensitivity of equipment increases, non-invasive diagnostic methods are required to support mass-production tools. And, there is a need to develop diagnostic technologies capable of quantifying real-time plasma spatial distribution (uniformity), real-time arc monitoring, and electromagnetic field values.

Advanced Solution for Equipment-Process Interlinked Analysis

• Existing manpower-dependent equipment diagnosis methods have shown limitations such as data loss, repetitive tasks, and non-standardized analysis processes. Through the establishment of the MADA system, we intend to transition to an automated solution that integrates sensor, equipment, and process yield data. Based on the automation of data collection and preprocessing, this system is equipped with advanced AI diagnosis models for anomaly detection, root cause analysis, and yield prediction. It is being developed to enable real-time anomaly judgment and prediction via a web-based dashboard without the need for expert engineers to be on-site, and we are working toward providing recommendations for optimal corrective actions in the future. In particular, we believe that the decision-making system incorporating natural language processing (NLP) and the automatic analysis report generation feature will drastically reduce analysis time and enable the rapid identification of key parameters affecting processes and yield. Beyond simple diagnosis automation, this is an AI-based standard platform that realizes the specialization and intelligence of analysis; it will serve as an innovative foundation for future advancement into an automated equipment evaluation and real-time predictive maintenance system, where MADA collaborates with multiple AI Agents to optimize yield.