Image analysis – Applications – Manufacturing or product inspection
Reexamination Certificate
2000-03-10
2002-01-15
Boudreau, Leo (Department: 2721)
Image analysis
Applications
Manufacturing or product inspection
C382S145000, C382S147000, C382S149000, C382S152000, C348S086000, C348S092000, C348S125000, C348S126000
Reexamination Certificate
active
06339653
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention concerns visual inspection for a product or a part being manufactured and more particularly to an inspection data analyzing system which is capable of inspecting defects or particles on a surface of the product or part and analyzing the inspection data.
In the manufacture of a semiconductor device or the like, product defects often result from particles or other defects existing on a surface of a work piece (noting that particles are one type of defect that can occur). It is, therefore, necessary to quantitatively inspect particles or other defects for normally monitoring if a problem occurs in the manufacturing machine or the environment around it. And, it is necessary to grasp how the particles or other defects have an adverse effect on yield and take effective measures for the particles or other defects for improving the yield. Hereinafter, the terms “particles or other defect” will sometimes be generally referred to as “defects”, but when specific reference is made only to particles, it will be identified as such.
As an example, the use of an automatic visual inspection machine for data analysis in the manufacture of semiconductors has been disclosed in an article entitled “How does the automatic wafer inspection improve a yield?”, Solid State Technology (Japanese Version), July 1988, pages 44 to 48. The visual inspection is carried out for wafers in more than one manufacturing process. Hence, the inspection data includes data for managing the inspection data itself. The managing data contains a product name of a inspected wafer, a lot number, a wafer number, and an inspected process, data, and time, for example. It is necessary to analyze not only the inspection data but also the managing data. The conventional visual inspection machine includes a function of measuring sizes of defects and where the defects are located on a wafer coordinate, a function of measuring the number of defects existing on a wafer, and a means for allowing an operator to determine a category of defects, and the like. The machine inspects the change of the number of defects on each wafer, the distribution of defect frequency on a wafer-size basis, and the like. Further, the machine serves to analyze the correlation between the number of defects on each wafer (defects density) and the yield of the wafer as well.
And, each wafer has to be identified in more than one visual inspection process in the data analysis. Conventionally, the operator has visually recognized a wafer number. To reduce the burden of this operation, an automatic particle inspection machine having a means for automatic recognition of a wafer number has been disclosed in JP-A-63-213352.
Known automatic visual inspection machines have been categorized into two groups. One is referred to as an automatic particle inspection machine which is an inspection machine employing a light-scattering system. This machine serves to inspect particles existing on a wafer. It is thus unable to always inspect defects other than particles. The other group is an inspection machine employing a pattern recognition system. It is referred to as an automatic visual inspection machine or an automatic defect inspection machine, which has a function of accurately recognizing other defects in addition to particles. The automatic visual inspection machine needs an inspection time which is about 1000 times as long as the time required by the automatic particle inspection machine. The former machine can thus inspect a far smaller number of wafers than the latter. For monitoring how defects are caused in a mass production line, two methods are provided. The first method is to restrict the processes to be visually inspected to a specific process (Solid State Technology (Japanese Version), July 1988, pages 44 to 48). The second method is to take the steps of matching the particle inspection data to the visual inspection data over all the processes and machines, checking the correlation between particles and defects, and presuming how defects are caused on the particle inspection data (Semiconductor World, May 1989, pages 118 to 125). Further, in analyzing data, these methods require an operator who serves to analyze data, because there exist a lot of data and various kinds of data analysis methods in analyzing data.
The conventional method is uncapable of grasping how defects are caused on each chip. Hence, they can merely perform correlation analysis between the number of defects per wafer and a yield. That is, these methods have a disadvantage that they cannot grasp the relation between defects per wafer and a product character. In addition, one semiconductor for one wafer is provided at this time, while two or more semiconductors for one wafer will be provided in future. It is necessary to enhance the data processing unit from a wafer unit to a chip-unit basis. A new data analysis technique is expected accordingly.
And, for inspecting how many defects are caused in a mass production line, the foregoing first method is designed to determine the process to be visually inspected on the basis of the knowledge of an operator and the result of a probing test. The foregoing second method requires large labor for matching the particle inspection data to the visual inspection data over all the processes and machines.
Moreover, an operator who is mainly in charge of maintaining and managing the manufacturing machine does not have spare time to analyze the inspection data of a wafer given by his or her machine. Hence, the operator requests the data analysis of another operator who is mainly in charge of it. However, novel data analysis method and means are expected which anyone can operate easily and quickly and which serve to output the analyzed data.
SUMMARY OF THE INVENTION
It is therefore an object of the present invention to provide an inspection data analysis system which is capable of analyzing data per chip for the purpose of grasping the relation between the occurrence condition of defects per chip and the product character of each chip.
And, it is a further object of the present invention to provide an inspection data analyzing system which is capable of easily determining a manufacturing process which causes problems and the contents or the problems.
It is another object of the present invention to provide an inspection data analyzing system which is capable of monitoring the overall production line and efficiently inspecting the quantity of caused defects in a mass production line.
To achieve the foregoing objects, the present invention offers a probing tester, an automatic particle inspection machine, and an automatic visual inspection machine respectively having data analysis stations. Each data analysis station has chip arrangement information for each product and serves to describe the locations of defects on the coordinate system on which the chip disposition is described. And, the station provides a function for determining on which chip each defect is caused. These data analysis stations are linked with a communication line.
Further, for inspecting the quantity of caused defects in the mass production line, the particle inspection machine operated at a higher inspection speed employs the step of monitoring the overall manufacturing line, inspecting the portions around caused defects, and monitoring the quantity of caused defects.
And, in order for anyone to use the machine, the data analysis station is designed to offer a routine data retrieval method, a routine operation method, and a routine analysis result output format.
As mentioned above, each data analysis station provides chip disposition information, a function of describing the locations of caused defects on the coordinate system on which the chip disposition is described, and a function of determining on which chip each defect is caused. It is thus possible to grasp how particles are attached and defects are caused on each chip. By linking these data analysis stations with a communication line, therefore, the data analysis station for pro
Ebara Yutaka
Hanashima Shuichi
Hashimoto Taizo
Ishikawa Seiji
Matsuoka Kazuhiko
Boudreau Leo
Mariam Daniel G.
LandOfFree
Inspection data analyzing system does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Inspection data analyzing system, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Inspection data analyzing system will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2857203