|
Title:
|
Semiconductor defect identification, analysis, and reduction through machine vision
|
|
Author:
|
Patton, James C.
|
|
Abstract:
|
Defect monitoring is a necessity in the semiconductor manufacturing industry. It is a process characterized by the regular inspection of its product during the significant portions of assembly. Defect monitoring is accomplished using automated inspection equipment capable of machine-vision. Machine-vision allows for the regular collection of defect data used to quantify and objectionably judge the quality of the processes and equipment responsible for the semiconductor manufacturing. NEC Electronics in Roseville, California employs such a system.
Their defect monitoring process uses machine vision to collect defect data. This data affords their Defect Detection Engineering department the opportunity to understand how to control and reduce the typical and atypical defects observed.
This project centers on a critical, atypical defect discovered during the metallization portion of process assembly. The defect was named, “Gouged SiO2.” This project will reveal how machine vision was able to detect Gouged SiO2 using template-matching processes. It will detail how engineering was able to investigate for its root cause using a machine-vision function called, “Defect-Source Analysis.” Finally, it will reveal how the defect was resolved and its countermeasure validated using feature-extraction analysis; another machine-vision capability.
All of these techniques used to characterize and solve Gouged SiO2 are functions of the machine-vision defect detection device called the “KLA2367.” The KLA2367 was the keystone equipment to understanding and solving the Gouged SiO2 defect, resulting in an improved quality process validated through higher wafer test-yield results. |
|
Description:
|
Project (M.S., Mechanical Engineering) -- California State University, Sacramento, 2010. |
|
URI:
|
http://hdl.handle.net/10211.9/305
|
|
Date:
|
2010-07-23 |
