| DC 欄位 |
值 |
語言 |
| DC.contributor | 電機工程學系在職專班 | zh_TW |
| DC.creator | 林正田 | zh_TW |
| DC.creator | Cheng-Tien Lin | en_US |
| dc.date.accessioned | 2012-1-17T07:39:07Z | |
| dc.date.available | 2012-1-17T07:39:07Z | |
| dc.date.issued | 2012 | |
| dc.identifier.uri | http://ir.lib.ncu.edu.tw:88/thesis/view_etd.asp?URN=975301018 | |
| dc.contributor.department | 電機工程學系在職專班 | zh_TW |
| DC.description | 國立中央大學 | zh_TW |
| DC.description | National Central University | en_US |
| dc.description.abstract | 產品量產之良率( Yield )高低與成本有著密不可分的關係。隨著進入奈米時代，顯現的晶圓瑕疵圖樣更加多變。由瑕疵數的分解比例，透過等比例的累積運算，我們可觀察到晶圓圖受到不同瑕疵密度的影響。並且可從晶圓圖驗證泊松分佈的可加性。
另外由於製程的變動增加以及複雜度的提高，不再只是單純由隨機微粒瑕疵所造成的故障晶粒，使得晶圓圖出現群聚瑕疵現象。而群聚瑕疵現象會導致泊松良率模型( Poisson Yield Model )預測的缺失。本文提出一種新穎的龍捲風良率模型，它是以晶圓圖故障晶粒數( NBD )和故障晶粒群聚方向數( NCD )兩項特徵為觀察座標--龍捲風圖。以最符合實際晶圓圖所顯示的情況，觀察兩項特徵值座落在龍捲風圖上的位置。
無論是製程中的隨機瑕疵或是設計參數的製程偏移，甚至於是兩者因素的結合，皆符合龍捲風圖的檢測。並應用統計分析的可加性特質，瑕疵密度對於良率的可乘性質相互呼應，可管控量產的品質並即時監測製程的穩定度。
| zh_TW |
| dc.description.abstract | Product yield and cost of production has a close relationship. With the arrival of nanometer era, the wafer defect patterns appear to be more various. By decomposition of the ratio of the number of defects, such as through the cumulative proportion of operations, we observe wafer maps affected by different density of defects. And it can be verified that the parameters of Poisson distribution on wafer maps are additive.
In addition, as the increase on process variations and complexity, it is no longer simply a random particle defects causing the distributed failure dice. The defect wafer maps appear the phenomenon of clustering, and cluster defects will lead to the yield prediction of Poisson model failed. This these proposes a novel tornado yield model, which is based on the number of bad dice (NBD) and Clustering Degrees (NCD) in wafer map for the observation of two characteristic coordinates -- Tornado Chart. Wafer maps of real cases shows the situation observed two characteristics located in the Tornado Chart.
Whether random defect or parameter variation, or even then the combination of both factors, it is shown that the detection of failure by a tornado chart is valid. By applying the statistical analysis, the defect density for yield multiplicative nature echo to the parameter additive nature. It can be used to control the quality of production and real-time monitoring of the stability of the process.
| en_US |
| DC.subject | 製程偏移 | zh_TW |
| DC.subject | 瑕疵 | zh_TW |
| DC.subject | 可加性 | zh_TW |
| DC.subject | 泊松分佈 | zh_TW |
| DC.subject | 良率 | zh_TW |
| DC.subject | 龍捲風圖 | zh_TW |
| DC.subject | Defect | en_US |
| DC.subject | Tornado Chart | en_US |
| DC.subject | Additivity | en_US |
| DC.subject | Yield | en_US |
| DC.subject | Poisson Distribution | en_US |
| DC.subject | Process variation | en_US |
| DC.title | 就隨機瑕疵源角度之晶圓圖分析 | zh_TW |
| dc.language.iso | zh-TW | zh-TW |
| DC.title | Wafer Map Analysis from a Random-Defect-Source Perspective | en_US |
| DC.type | 博碩士論文 | zh_TW |
| DC.type | thesis | en_US |
| DC.publisher | National Central University | en_US |