考慮製程規格限制之晶圓廠產能需求規劃

Capacity Requirements Planning with Capability Constraint for Semiconductor Manufacturing Fabs

陳佳雯(Chia-Wen Chen)

指導教授 ： 陳建良

中原大學/工學院/工業工程研究所/博士(2008年)

https://doi.org/10.6840/cycu200900022

摘要

晶圓製造對製程規格之要求嚴謹，機台須具備該加工步驟所需之製程規格方可進行加工，此為製程規格限制。此外，為了確保產品良率，同一批晶圓需要相同製程規格之不同步驟須使用相同機台進行加工，此為綁機限制。製程規格及綁機為晶圓製造微影製程之特殊限制，增加晶圓廠生產規劃之困難度。本研究考慮上述兩項限制發展晶圓廠產能需求規劃系統。

本研究所發展之晶圓廠產能需求規劃系統包括兩大部份：多廠無限產能需求規劃及單廠有限產能需求規劃。多廠無限產能需求規劃藉由訂單分配及負荷配置兩部份來決定各廠適當之產品組合。訂單分配考慮各批量加工途程之機台負荷決定各廠產品組合，負荷配置進一步針對具製程規格及綁機限制之步驟，選擇適當之機台進行加工，以平衡設備間負荷。無限產能需求規劃之結果可作為安排設備製程規格、設備間備用及外包決策之依據。

單廠有限產能需求規劃接著用來發展適當之生產計劃，有限產能需求規劃在製程規格及綁機限制下，考量各設備所具備之最小製程規格，搭配後推並考慮各批量加工途程之機台負荷，決定各批量適當之投料時點及加工設備，以最小化訂單之平均絕對延遲及具相同製程規格設備使用率之標準差。有限產能需求規劃之結果可作為與顧客協商交期之依據。

本研究收集晶圓製造廠之資料，用來評估所發展晶圓廠產能需求規劃系統之績效。無限產能需求規劃部份，訂單分配考慮修正投料日及加工途程機台負荷可有效平衡各廠、各日、各設備之負荷；負荷配置考慮未分配訂單所需負荷可有效平衡具相同製程規格之設備間負荷。有限產能需求規劃部份，考慮批量加工途程之機台負荷、設備所具備之最小製程規格及訂單交期，可得到較佳之訂單平均絕對延遲及平衡具相同製程規格設備之使用率。以上結果顯示，本研究所發展之晶圓廠產能需求規劃系統藉由適當地分派訂單至各廠及各設備，可獲得較佳之生產計劃，以有效利用設備產能、縮短生產週期時間、提高達交率及降低在製品。

關鍵字

綁機限制 ； 製程規格限制 ； 晶圓製造 ； 產能需求規劃 ； 有限負荷 ； 無限負荷

並列摘要

Process specifications in wafer fabrication require precise definition. Whether or not an operation can be worked at a specific machine depends on the machine’s ability, and this is the process capability constraint. Moreover, processing all the layers of a lot on the same stepper assures better yield rate, and this is the machine dedication constraint. The capability and dedication are specific constraints at photolithography of wafer fabrication that makes it difficult to obtain the optimal production plan in the real production environments. This research develops a Capacity Requirements Planning system considering equipment’s Capacity, Capability, and Dedication of (CRP-CCD) for semiconductor manufacturing fabs.

CRP-CCD consists of two major parts: Infinite Capacity Requirements Planning System (ICRPS) for multiple fabs and Finite Capacity Requirements Planning System (FCRPS) for single fabs. Based on an assumption of infinite loading, ICRPS determines the product mix for each fab by Order Assignment (OA) and Capacity Allocation (CA). Using Path Load (PL) concept, OA selects appropriate product mix for each fab and determines the loading occurrence times of each of the required capability of each order to better balance the equipment workload among various fabs, on various days, and across various equipment at various levels of demands by taking into account equipment’s capacity and capability at multiple fabs. CA uses Loading Ratio (LR) concept to select the machine used to process each order’s capabilities considering equipment’s capacity, capability, and dedication to smooth the loading among machines with the same capability. ICRPS can help fab engineers to arrange equipment capability, equipment backup and subcontract.

After the application of ICRPS, FCRPS is then used to develop an appropriate production plan for single fab. Based on an assumption of finite loading and by taking into account equipment’s capacity, capability, and dedication, FCRPS combines Backward Finite Loading (BFL), Maximum Minimum Capability (MMC), and Path Load (PL) to determine the release time and equipment capability for each lot to minimize Mean Absolute Lateness (MAL) of customer orders and the Standard Deviation of the Utilization of equipment with the same Capability (SDUC). FCRPS can help fab salespersons to negotiate with customers regarding order due dates.

Data from a real foundry fab are collected and used by simulation based on experimental design to evaluate the performance of CRP-CCD. The following conclusions are drawn from the analysis of simulation results. In ICRPS, OA based on Adjusted Release Time (ART) and PL can efficiently balance the equipment workload among various fabs, on various days, and across various equipments at various levels of demands. CA using LR can efficiently balance the equipment workload among different machines with the same capability. FCRPS combining BFL, MMC, and PL can efficiently balance the loading on various machines with the same capability and minimize MAL of customer orders. These findings demonstrate CRP-CCD can obtain the right production plan for semiconductor manufacturing fabs by assigning the right orders to the right fab and assigning the right orders to the right equipment with the right capability. Thus, CRP-CCD can help semiconductor foundry companies to obtain better equipment utilization, more balanced equipment loading among different fabs and also among different machines with the same capability, shorter manufacturing cycle time, better on-time delivery, and lower work-in-process.

並列關鍵字

capability ； dedication ； infinite loading ； capacity requirements planning ； wafer fabrication ； finite loading

參考文獻

I. Ahmed and W. W. Fisher, 1992, Due Date Assignment, Job Order Release, and Sequencing Interaction in Job Shop Scheduling, Decision Sciences, Vol. 23, No. 3, pp. 633-647.

E. Akçalt, K. Nemoto, and R. Uzsoy, 2001, Cycle-Time Improvements for Photolithography Process in Semiconductor Manufacturing, IEEE Transactions on Semiconductor Manufacturing, Vol.14, No. 1, pp. 48-56.

E. Akçalt and R. Uzsoy, 2000, A Sequential Solution Methodology for Capacity Allocation and Lot Scheduling Problems for Photolithography, In Proceedings of 2000 IEEE/CPMT international Electronics Manufacturing Technology Symposium, pp. 374-381.

D. Bergamaschi, R. Cigolini, M. Perona, and A. Portioli, 1997, Order Review and Release Strategies in a Job Shop Environment: A Review and a Classification, International Journal of Production Research, Vol. 35, No. 2, pp. 399-420.

J. W. Breithaupt, M. Land, and P. Nyhuis, 2002, The Workload Control Concept: Theory and Practical Extensions of Load Oriented Order Release, Production Planning and Control, Vol.13, No. 7, pp. 625-638.