半導體晶圓廠設備管理績效之研究 –以A公司十二吋廠先進製程控制系統為例

A case study of wafer Fab equipment management performance –base on advanced process control system for A company 12” fab.

黃 錳 樺(Meng-Hwa Huang)

指導教授 ： 何建德 博士

元智大學/管理學院/管理研究所/碩士(2007年)

摘要

根據經濟部工業局半導體產業推動辦公室的統計，台灣半導體廠商積體電路(IC)製造能力，全球排名第四，台灣半導體產業因具備完整的半導體產業鏈支援，群聚效果顯著；故不論是專業晶圓代工(foundry)廠商或者是動態隨機存取記憶體(Dynamic Random Access Memory，DRAM)廠商在大量生產的製造能力上實力堅強，進而帶動了上下游產業的蓬勃發展，這些完整健全的矽產業結構，使我國擁有了全球唯一的半導體專業分工體系。因此有足夠的機會將我國帶入下一代的半導體產業。
雖然我國半導體產業站在如此有利的制高點上，但依據國際晶片製造聯盟(International SEMATECH Manufacturing Initiative，ISMI)的研究，從2003年起以一座月產能四萬片的十二吋晶圓廠投資高達三十億美元，加上DRAM產品製造週期(cycle time)長達50〜60天且製造步驟(step)多達400〜600道，使得在生產過程中所面臨的設備機台運籌管理相關的問題日趨複雜，這已成為台灣晶圓製造業者必須注意的課題。再說設備機台佔半導體晶圓製造廠成本比例最高，因此設備機台在生產的管理上有很大的改進空間，因此本研究認為建構一套晶圓廠先進製程控制(Advanced Process Control，APC)系統，一方面除了可以減少晶圓廠因設備機台異常或人為疏失時所造成的重大損失;另一方面更可以提供設備管理者不同於傳統的設備維護模式，使其運用此系統之各項功能來達到提昇設備管理績效與改善設備功能之參考。
以晶圓廠而言，因設備機台及其零件故障頻率不一，加上設備機台之結構複雜，設備管理工作不僅僅是定期預防保養以及故障排除與零件的修護工作；更應積極地運用資訊管理系統，以有效率、有效果的方法將價值昂貴、數量繁多的生產機台妥善的運用，以產生綜效(synergy)降低營運風險。本研究將藉由個案研究法( Case Study )，透過訪談、參與觀察及檔案紀錄等多重證據來源作為佐證，論述個案公司應用先進製程控制系統，對應於晶圓廠提升設備管理績效上所扮演之意涵。
本研究之機台參數與機台警訊管理提供了一概念性之架構，使晶圓廠管理者可參考此一架構，發展一套適用於自己公司體質的設備管理模式。另一方面本研究也發現善用APC系統之歷史資料、機台即時監控等功能，可以建立設備之故障預測模式，來縮短設備機台的各種停機時間。這將是晶圓廠未來之重要工作目標。

關鍵字

先進製程控制 ； 設備管理 ； 製程參數 ； 警訊管理

並列摘要

According to the statistics provided by Semiconductor Industrial Promotion office of Industrial Development Bureau (IDB), Ministry of Economic Affairs (MOEA), Taiwan semiconductor manufacturing capacity ranks number four in the world. Because the supply chain of Taiwan semiconductor industrial is very complete, its cluster effect is very remarkable. The mass production capability of foundry and DRAM businesses are very strong, and as a result, it leads to the industry vigorous development. The solid semiconductor industrial structure enables Taiwan to become an important player in the global semiconductor supply system.

Although Taiwan semiconductor industry has many advantages, a 300 mm fab with the capacity of 40,000 wafers per month may cost up to 3 billion US dollars. In addition, the DRAM manufacturing cycle time is about 50 to 60 days and the whole process involves 400 to 600 manufacturing steps. These entire factors make the management of equipment during the manufacturing process complicated, and have become the issues that Taiwan semiconductor manufacturer must pay attention to. Because the equipment cost accounts for the largest portion among the semiconductor manufacturing expenditure, and the management of equipment in the production still have lots of space for improvement, this research seeks to construct a semiconductor wafer fab advanced process control (APC) system, which can reduces the loss that caused by exceptional equipment or human errors. The system can provide the equipment superintendent with an equipment maintenance model which is different from the traditional one. It improves the efficiency of equipment management and is used as a reference of the equipment function transformation.

Due to complex equipment systems in the wafer fab, equipment management covers regular preventive maintenance, trouble shooting and components maintenance. Wafer fab should positively utilize information management system for better productivity. This research uses case study research method to explore the adoption process of the APC system in a DRAM manufacturing company. Qualitative and quantitative data were collected interviewing key persons, participating observation and file records. Multiple data resources ensure evidence validity and weave rich information regarding the application of the APC system and productivity improvement in the case company.

The equipment parameter and alarm message management resulted from this research provide a conceptual structure. Wafer fab managers should be able to use the structure to develop an equipment management model that is suitable for their company. This research also discovered that using historical data of the APC system and instant monitoring function of the equipment can establish an equipment failure predication model to reduce equipment down time.

並列關鍵字

advanced process control ； equipment management ； process parameter ； alarm message management

參考文獻

1.Andrews, K. R (1951)， “Executive raining by the Case Method”, Harvard Business Review.

2.Fredrik Dahlstrand (2002)，”Consequence analysis theory for alarm analysis”，Knowledge-Based Systems 15，27-36。

3.Fumihiko Yamanaka and Takushi Nishiya (1997)，”Application of intelligent alarm system for the plant operation”，Computer chemical engineering，Vol.21，pp. S625-S630

6.L.Pfitzner&P. Kücher, (2003)，”A roadmap towards cost efficient 300mm equipment”，Materials Science in semiconductor processing，Vol. 5

9.Yoshitaka Yuki，(2002)，”Alarm system optimization for increasing operations productivity” ，ISA Transaction 41，383-387。