薄膜電晶體陣列廠考量光罩資源限制之主生產排程機制設計

Abstract

在TFT-LCD陣列段黃光製程中，曝光作業需搭配特定的光罩才可加工，且機台更換光罩需要耗費整備時間，是故排程作業是非常困難的。由於現今陣列段製程為存貨式生產的環境，同時也為玻璃基板在TFT-LCD三階段中所歷經的第一道步驟，故若是無法達到預定產出計畫則會影響到後面製程進度，進而使顧客訂單延誤。 過去研究陣列廠主生產排程之論文中，大多未考慮光罩資源。有鑑於此，針對各產品各期之預定產出目標，吾人建構數學模式規劃考量光罩限制的陣列廠主生產排程，並最小化每一產品延誤數量為目標進行求解。本文首先粗估機台產能是否能負荷規劃幅度內之產品預定產出量，其次在使產品層級延誤需求量最少的目標下，考量光罩數量限制與再回流特性，以線性規劃模式進行求解，若產品層級在規劃幅度內延誤數量為0，則改以產出最大化方式生產以避免浪費產能。此外，本文考量管理人員依其意願修改主生產排程的限制下，設計一決策流程以善用光罩與機台產能，並計算更改前後缺貨成本與機台整備次數等相關資訊，提供給管理人員作為參考依據，以確認是否需修改原先主生產排程結果。 範例驗證結果顯示，本文所估算之光罩與機台配置結果，可使每一產品層級之延誤數量最少，且能充分利用機台與光罩產能；而透過與顏氏[36]方法比較，本文方法在使產品延誤數量最小化上，確實具有不錯成效。整體而言，本文所發展之模式，兼顧求解品質以及實用性，可有效地規劃考量光罩限制的陣列廠主生產排程。

In the photolithography process of TFT-Array, the exposure process will be tackled only when the specific reticle is ready on the machine and the exchanges of reticles also take time to setup. Owning to the characters mentioning above, it is quite difficult to schedule. TFT-Array, the fisrt stage of the process of TFT-LCD, is the circumstance making to order. If the factory can not reach the planned amount of throughputs, it will contribute to a mess after the stage and the order will be postponed. Since the resources of reticles are less considered in the recent essays concerning MPS of TFT-Array, the work brings up a MPS of TFT-Array by MIP model considering the constraint of reticles and solving by minimizing the quantity of delay of each product . For one thing, the study roughly estimate the capacity to see whether we can satisfy the amount of throughputs that is scheduled among the planning horizon or not. For the purpose of minimizing the postponed number of products’ layers, this paper also solves the problem that take the constrain of quantity of reticles and the character of reentry into account by LP model. If the amount of delay is at zero in the planning horizon, we shift the object to maximize the throughputs for avoiding the waste of capacity. Furthermore, this work think about that the manager is prone to modify MPS and design a decision process employing reticles and capacity efficiently and count the information of cost of shortage supply as well as times of setup. In doing so, the process can provide the supervisor with the information to confirm whether it is necessary to adjust previous scheduling or not. The experimental results reveals that the reticle and machine match not only minimize the quantity of delay of each product, but also utilize the capacity of reticle and machine adequately; the study certainty have correct effect of minimize the quantity of delay of each product through compare with Ru-Min Yan. Consequently, the mechanism propose in this research is effective and efficient, and can be used to plan MPS for the TFT-Array Production with reticle resource constraint.