臺灣博碩士論文加值系統

在先進的製程技術下，因為光罩影響的範圍已經大於佈局圖最小間距，所以非相鄰的佈局圖也可能會降低良率，而所謂的製程熱點(process-hotspot)就是在佈局圖中會影響良率的部分佈局圖。本論文所要解決的問題就是利用含有可忽略區域的範圍圖形偵測製程熱點。當製程熱點包含有可忽略區域時，很難將其所有可能性描述出來，因此在以往的文獻中，無法找出包含有可忽略區域的製程熱點。我們提出兩個方法，在第一個方法中是建立二元搜尋樹，找出佈局圖中符合製程熱點的所有可能性，雖然可以找到全部製程熱點，但是需要的執行時間過長。在第二個方法是利用多元搜尋樹，搜尋出分配佈局圖的所有的可能性，透過適當的分配找出最大製程熱點。實驗結果顯示，利用含有可忽略區域的範圍圖形，在第二個方法中可以成功找出最大製程熱點，此外在製程熱點中的可忽略區域我們允許多個矩形存在。

A process-hotspot is a layout pattern detrimental to chip's fabrication yield. With continuing use of 193nm lithography for advanced technology nodes, the number of layout patterns that represent process hotspots increases dramatically. Hence, layout pattern description for process hotspots using range rectangles has been employed to cope with the explosion problem. This kind of pattern is called range pattern. Several effective methods using range patterns for process-hotspot detection have been proposed. However, no effective method has been presented to deal with a range pattern that contains don’t-care regions. Layout patterns within a don’t-care region can be ignored during hotspot detection. In this thesis, we propose two approaches to detecting process-hotspot using range pattern with don't-care regions. The first approach uses a binary search tree that explicitly explores all possible process hotspots. Although this approach can find out all process-hotspots, it requires a very long runtime. The second approach uses a multi-way search tree that implicitly considers all matched possibilities. Experimental results show that we find out all the process-hotspots in the given layout using range pattern with don't-cares. We can find out process hotspots not found by a commercial tool.

目 錄
書 名 頁 II
致 謝 III
中 文 摘 要 IV
英 文 摘 要 V
目 錄 VII
表 目 錄 IX
圖 目 錄 X
CHAPTER 1. 緒論 1
1.1、 研究背景 1
1.2、 研究動機 1
1.3、 論文貢獻 2
1.4、 論文架構 3
CHAPTER 2. 相關研究 4
2.1、 利用字串編碼偵測製程熱點 4
2.2、 利用關鍵製程規則偵測製程熱點 8
2.3、 利用機器學習偵測製程熱點 12
CHAPTER 3. 研究方法 17
3.1 研究架構 17
3.2 演算法 19
3.3 證明 32
CHAPTER 4. 實驗結果 75
CHAPTER 5. 結論 80
參考文獻 81

[1] H. Yao et al. “Efficient process-hotspot detection using range pattern matching.” In Proc. ICCAD, pp.625–632, 2006.
[2] R. Zhu and T. Takaoka. “A technique for two-dimensional pattern matching.” In Comm. ACM, 32(9), pages 1110–1120, 1989.
[3] Yen-Ting Yu; Ya-Chung Chan; Sinha, S.; Jiang, I.H.-R.; Chiang, C., "Accurate process-hotspot detection using critical design rule extraction," Design Automation Conference (DAC), 2012 49th ACM/EDAC/IEEE , vol., no., pp.1163,1168, 3-7 June 2012
[4] D. Ding et al. High performance lithographic hotspot detection using hierarchically refined machine learning. In Proc. ASP-DAC, pp. 775–780, 2011.
[5] J.-Y. Wuu et al. Rapid layout pattern classification. In Proc. ASPDAC,pp. 781–786, 2011.