晶圓針測廠考量針測卡資源限制下主生產排程系統之設計

Abstract

晶圓針測製程(Wafer Probing)接續於晶圓製造之後，為半導體業之前段製程中最後一道加工步驟，其訂單達交績效最被重視。以往相關之排程文獻僅探討系統中主資源之運用而忽略了附屬資源之重要性，因此本文探討之晶圓針測主生產排程規劃問題，除了考量訂單交期、順序相關之設置時間、機台產能與製程規格能力等限制外，還加入了附屬資源－針測卡之數量限制，以符合實務所需。 有鑑於此，本文首先初估機台與針測卡產能是否足以負荷總體訂單之需求；其次為兼顧訂單達交績效並有效利用瓶頸資源與附屬資源之產能，本文以一混合整數規劃模式搭配一產出規劃流程，求解各訂單各產品最佳的生產順序與針測卡之使用配置。此外文中另發展一針測卡新增機制，協助公司與其協力廠間決定出最有效的針測卡調度模式，以善用未被利用之機台產能。而若所規劃之主生產排程中，尚有未被安排排程之產能缺口，本文則將之作為臨時到臨訂單需求之允諾依據，以迅速確認此筆訂單需求是否能夠及時滿足。 實證結果顯示，本文中各筆訂單在交期限制下，各機台皆以最小之設置時間進行排程，以充分利用機台與針測卡產能。另外當機台產能充裕卻出現未能滿足之需求量時，能有效的決定出須借用之針測卡別以善用原本未能利用之機台產能。整體而言，本文發展之規劃模式，兼顧求解效率與實務性，可作為回覆上層需求管理單位之交貨依據，並能給予下層生產控制單位明確之生產排程，以利其進行生產活動控制。

The wafer probing process is the last operation in the front-end of wafer fabrication, and its due-date oriented performance is extremely concerned. Formerly researches related to wafer probing scheduling problem merely investigate the utilization of main resource in the system, while the auxiliary resources were out of consideration. In order to fit in with reality, the wafer probing scheduling system in this thesis will consider constraints of order due date, sequence dependent setup time, machine capacity, and process capability. It futhermore considers the constraint of probe card resource. In this study, we first estimate the capacity of machines and probe cards to see if their supply can bear the aggregate demand of orders. Then, we construct a mixed integer programming model that matched with a production planning procedure so as to simultaneously considering the delivery performance of orders, the highly capacity utilization of the bottleneck and auxiliary resource. The planning results contain the production sequence of each product of orders and the schedule of each probe card. Furthermore, to fully utilize machine’s capcacity, we develop a “probe card adding procedure” to assist the company borrowing the right types of probe cards from his supplier. Finally, if there exists spare capacity based on the result of MPS, this system can rapidly determine whether a new coming order can be accepted or not. The experimental results showed that based on due date constraints, all orders will be scheduled with minimum setup time to properly utilize the capacity of machines and probe cards. Moreover, when there is some surplus capacity on machines but the demand is not fully satisfied, this system will determine which kind of probe cards would be added to make good use of the capacity of the machine. Hence, the mechanism proposed in this thesis can provide effective and efficient planning results, which are valuable for replying committable amount to demand management department and for detail scheduling in shop floor production control.