分割晶圓圖分析以增強系統性錯誤的解析;Wafer Map Partition Analysis to Enhance Systematic Error Resolution

NCUIR > College of Electrical Engineering & Computer Science > Graduate Institute of Electrical Engineering > Electronic Thesis & Dissertation > Item 987654321/77539

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/77539

Title:	分割晶圓圖分析以增強系統性錯誤的解析;Wafer Map Partition Analysis to Enhance Systematic Error Resolution
Authors:	吳雅軒;HSUAN, WU YA
Contributors:	電機工程學系
Keywords:	晶圓圖;隨機性錯誤;系統性錯誤;良率;泊松良率;Wafer map;Random errors;Systematic errors;Yield;Poisson yield
Date:	2018-04-12
Issue Date:	2018-08-31 14:47:44 (UTC+8)
Publisher:	國立中央大學
Abstract:	一般晶圓圖是由隨機性和系統性的兩種錯誤組成，本篇論文針對九種系統性錯誤中的其中六種有症狀的錯誤做分割分析，觀察六種錯誤的特徵後，我們將晶圓圖分割成該特徵圖形的晶圓圖，因此可強化該特徵的系統性錯誤，且分開晶圓由隨機性錯誤組成的部分，再觀察分割後提高系統性錯誤的判別效果。然而分割成各種形狀後，我們研究的種種特徵參數是否會因為形狀不同而不適用是個議題，所以本研究也針對種種分割圖形做參數驗證。另外，利用柏松良率能判別晶圓圖的隨機性，過去我們利用晶圓圖故障晶粒數( NBD )和故障晶粒連續線( NCL )可以描述部分的柏松良率，本文提出校正先前實驗參數缺陷的改善，讓未來研究方法可以描述整個柏松良率以方便判別隨機性。除了利用分割延續先前迴力棒透過群聚情形的判別方式外，再提出兩種判別系統性錯誤的分割方法，分別是扇形及甜甜圈形損壞晶粒分布檢測，和迴力棒判別方式不同，提出的兩種方法是預測損壞晶粒分布情況做判別，最後把三種方式綜合在一起互補判別，以提高即時判別系統性錯誤的能力，進而達到提高良率、測試效率以及降低成本的目的。 ;The general wafer map is composed of random and systematic errors. In this paper, we analyze the six kinds of symptomatic errors among the nine kinds of systematic errors. After observing the characteristics of the six kinds of errors, the wafer map is partitioned into wafer maps of the feature graphics so that systematic errors of the feature can be enhanced and the part of random errors can be separated. Then we observe the effect of improving system errors resolution after partition. However, after the partition into various shapes, it is an issue whether all the characteristic parameters of our research will not be applicable due to the different shapes. Therefore, this study also verifies the parameters of all the partition patterns. In addition, the use of the Poisson Yield can determine the randomness of the wafer map, in the past we use the number of bad dice (NBD) and the number of contiguous line(NCL) can describe a part of Poisson Yield. In this paper, we propose to correct the defects of the previous experimental parameters so that future research methods can describe the entire Poisson Yield to facilitate the determination of randomness. In addition to partitioning the continuation of the previous use of Boomerang to determine the clustering situation, and then proposed two methods to determine the systematic error partition, namely sector-shaped and donut-shaped bad dice distribution detection. The two methods are to predict the distribution of bad dice to discriminate. Finally, the three methods are combined to complement each other, so as to improve the ability of discrimination systematic errors promptly, so as to improve the yield, test efficiency and reduce costs.
Appears in Collections:	[Graduate Institute of Electrical Engineering] Electronic Thesis & Dissertation

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	128	View/Open

社群 sharing

Loading...