標題: 半導體晶圓離子植入製程派工輔助系統建構之研究
A DISPATCH DECISION SUPPORT SYSTEM FOR SEMICONDUCTOR IMP PROCESS TOOL
作者: 唐亦賢
林君信
管理學院管理科學學程
關鍵字: 晶圓製造;離子植入;決策支援系統;派工;Chip Manufacturing;Ion Implantation;Decision Support System;Dispatching
公開日期: 2007
摘要: 半導體ICT產業的產值,可以看出電子工業興盛與否,而IC晶圓製造目前也是屬於台灣重點發展產業。晶圓製造如今從0.15微米製程更進步到深次微米(奈米)。其中晶粒的密度得以不斷增加,原因之ㄧ源自於半導體製造在離子植入及高溫回火的製程技術上不斷突破,因此業界亦將半導體離子植入製程視為關鍵製程。 在晶圓製造的過程當中,製造程序由多達上百道的製程階段(Stage)所組成,每一階段又包含多道製程步驟(Step)需要被執行,才能完成一整段的關鍵製程(如離子植入)。而其中順序上的重複、乃至於運送上的交叉與逆回等複雜性都難以避免。高效率以及合理化的派工與指派,以確保關鍵製程機台能夠完全的發揮產能,生產作業能夠維持高效能與彈性化、低成本化,是所有晶圓製造管理者的挑戰。 在多樣變化以及複雜資訊判斷下,傳統派工僅利用EDD(最早到期時間,由交貨時間反推)設定優先順序(Priority),再輔以個人經驗判斷,是目前以本人所任職企業而言所採用之方式。本論文研究考量系統化管理方式建立模式,目的為 1.研究最適化指派 2.知識管理,將指派方式與經驗組合成知識庫 3.系統支援決策,藉由技術知識庫發展決策支援人工智慧系統,輔助管理者派工的工作以及知識分享傳遞 4. 自動性與相容性,系統自動執行基本判斷並相容於現有優先順序系統 5. 使用者需求導向,系統適用於管理者與現場工作人員。 系統使用OR(Operation research)指派模式計算最適解,並採用Delphi程式語言開發管理系統。最後經由驗證相關KPI(關鍵績效指標)-機台使用效率、最大產出皆優於原派工方式,製造管理者與執行製造工作人員皆可利用此系統增進派工管理效能。
The prosperity of electronic industry can mainly be attributed and highly correlated to the GDP of semiconductor manufacturing industry (Information & Communications Technology; ICT) which is one of the most crucial developing industries in Taiwan. Nowadays, Chip Manufacturing is even more advanced and sophisticated from 150mm to 90mm (even 65mm) in technological arena. One of the quintessential root causes to bring about the density enhancement of chips is the incessant breakthrough and consistent metamorphosis in process technique in ion implantation and furnace anneal technology so as to be regarded as critical process within semiconductor manufacturing industry. Among series of procedure towards chip manufacturing sphere, one single procedure is composed of hundreds of stages including multi-layer steps (such as ion implantation). Therefore, it’s unavoidable in complexity on process sequence, transportation arrangement, etc. Moreover, outstanding effectiveness and high degree of rationality in production dispatching and bottleneck-equipment assignment is a major challenge to all managing levels in chip manufacturing so as to corroborate the performance exertion towards key machines among all critical bottleneck procedures. In common practices towards current chip manufacturing under versatile metamorphosis and complexity in informational judgment, EDD technique (Earliest Due Date Method, which is calculated from and based on ship date to end customers) for facilitating fundamental setting to production priority and personal judgment in manual mode are the most frequent utilized methodologies. The main focus in this paper is to establish proactive and interactive production model in systematic approaches which is projected to acquire the following advantages: (1) Dispatching Optimization Research (DOR); (2) Knowledge Management (KM) and Data Mining (DM) to feasibly interlink between dispatching methodologies and individual judgment so as to optimally create well-organized knowledge base; (3) Decision Support System (DSS) to develop artificial intelligence interface in decision support approaches via technological knowledge pool so as to efficiently lead to highly-than-normal performance in assisting mangers for making the most optimal and feasible solutions and facilitate transmission in knowledge sharing; (4) Degree towards Automation and Compatibility to automatically execute baseline operation and induce production interface to be compatible with current priority sequence system; (5) user-demand orientation so as to enable the algorithm to trigger beneficial synergy to on-line employees and managers. The approach in this paper is mainly based on assignment model in Operation Research (OR) category to calculate the most feasible solution and develops management system through Delphi programming. By verifying pertinent KPIs (Key Production Indices, such as Equipment Efficiency, WPH, etc), the performance is much higher than traditional methodologies to activate managers and on-line workforce to excel in dispatching effectiveness.
URI: http://140.113.39.130/cdrfb3/record/nctu/#GT009462535
http://hdl.handle.net/11536/82362
Appears in Collections:Thesis