半導體產業為現今台灣最重要且資金稠密度最高的產業之一。隨著科技的發展與製程技術的改善,晶圓尺寸日益增大;而其製作出之元件/晶片的尺寸則日趨輕薄短小。針對此高精密度的製程,如何即時有效瞭解機台的健康狀況及其對應之製程表現,進而提升製程良率、增進產出並降低成本,實為一極具挑戰性的重要議題。 此外,由於半導體製程之步驟繁複且歷時甚久,產品的品質特徵值在各製程加工中難以即時取得。如何仰賴生產線之機台製程資料,定義機台各子系統的健康指標並且有效偵測出製程之異常表現,為現今半導體業製程監控工作之重大挑戰。本研究提出一系統性的半導體製程品質管制監控系統,依據生產線之機台製程資料,可有效並快速的偵測異常製程並且提供機台的即時健康情況之分析。
Semiconductor industry is one of the most important and capital-intensive industries in Taiwan. As the trend of extended wafer size and narrowed feature size of chips and dies, the powerful and suitable control scheme for such highly accurate manufacturing becomes the essential issue to improve the yield, enhance the throughput and decrease the cost. Due to the rare output observation known as metrology compared with the huge sensor data set collected on-line over the whole process, it is the crucial challenge to detect the abnormal performance of process and analyze the health conditions of equipment effectively within the shortest possible time. In this research, the novel fault detection and classification, FDC, system is provided to perform the fast and precise diagnosis for process/equipment healthy performance determination.