Method for analyzing defect data and inspection apparatus and review system

• US 20020181756A1
• Filed: 04/10/2002
• Published: 12/05/2002
• Est. Priority Date: 04/10/2001
• Status: Active Grant

• Current Assignee: Hitachi High-Tech Company Limited (Hitachi, Ltd.)
• Original Assignee: Hitachi High-Technologies Corporation (Hitachi, Ltd.)
• Inventors: Takagi, Yuji, Shibuya, Hisae
• Granted Patent: US 6,876,445 B2
• Time in Patent Office: Days
• US Class Current: 382/145
• CPC Class Codes: G01N 21/9501 Semiconductor wafers manufacturing processes per se of semiconductor devices implementing a measuring step H01L22/10
• CPC Class Codes: G01N 21/956 Inspecting patterns on the surface of objects contactless testing of electronic circuits G01R31/308; testing currency G07D; manufacturing processes per se of semiconductor devices implementing a measuring step H01L22/10
• CPC Class Codes: G06T 2207/30148 Semiconductor; IC; Wafer
• CPC Class Codes: G06T 7/0006 using a design-rule based approach
• CPC Class Codes: G06T 7/001 using an image reference approach

Abstract

The distribution states of defects are analyzed on the basis of the coordinates of defects detected by an inspection apparatus to classify them into a distribution feature category, or any one of repetitive defect, congestion defect, linear distribution defect, ring/lump distribution defect and random defect. In the manufacturing process for semiconductor substrates, defect distribution states are analyzed on the basis of defect data detected by an inspection apparatus, thereby specifying the cause of defect in apparatus or process.

Claims

1. A method of analyzing defect data, comprising the step of classifying defects detected as a result of inspecting a processed substrate that results from processing said substrate to form a circuit pattern thereon, wherein said classification is made to specify at least any one of repetitive defect, congestion defect, linear distribution defect, ring/lump distribution defect and random defect of distribution feature categories on the basis of the position coordinates of said defects.

2. A method of analyzing defect data, comprising the steps of:

generating a chip overlap map showing the positions of all defects by coordinate systems having the origins of chips used as reference points, said map being produced by use of defect data including at least the position coordinates of said defects obtained by inspecting a processed substrate that results from processing said substrate to form a circuit pattern;

calculating the distance between the adjacent ones of said defects by using a closest point Voronoi diagram corresponding to said chip overlap map;

discriminating a repetitive defect on the basis of said distance between the adjacent defects;

generating a wafer map with the positions of all defects expressed by a wafer-origin coordinate;

calculating the distance between adjacent defects and a Voronoi region area by using said closest point Voronoi diagram corresponding to said wafer map;

discriminating a congestion defect on the basis of said distance between the adjacent defects;

discriminating a linear distribution defect on the basis of said distance between the adjacent defects and said Voronoi region area;

discriminating a ring/lump distribution defect on the basis of said distance between the adjacent defects and said Voronoi region area; and

extracting a defect other than said repetitive defect, said congestion defect, said linear distribution defect and said ring/lump distribution defect, i.e., a random defect.

3. A method of analyzing defect data according to claim 2, wherein, in said step of discriminating said repetitive defect, the adjacent defects on said chip overlap map are removed not to be processed when they are located on the same chip and when the distance between the adjacent defects is equal to or smaller than a predetermined threshold, the remaining defects are connected to be grouped when the distance between the adjacent defects is equal to or smaller than a certain threshold, and said group of defects of which the number is equal to or larger than a predetermined value is detected as said repetitive defect.

4. A method of analyzing defect data according to claim 2, wherein, in said step of discriminating said congestion defect, said adjacent defects are connected to be grouped when said distance between the adjacent defects on said wafer map is equal to or smaller than a predetermined threshold, and said group of defects of which the number is equal to or larger than a certain value is detected as said congestion defect.

5. A method of analyzing defect data according to claim 2, wherein, in said step of discriminating said linear distribution defect, said wafer map is subjected to Hough transform after being weighted for each defect in association with said distance between the adjacent defects or said Voronoi region area, and then, coordinates (θ

, ρ

) corresponding to the maximum value on said Hough transform plane are detected, and said linear distribution defect is discriminated on the basis of a dark and light image that represents the defect density around the position on said wafer map, corresponding to the coordinates (θ

, ρ

), when the width of a portion determined for the maximum value on said Hough transform plane is equal to or smaller than a predetermined value.

6. A method of analyzing defect data according to claim 2, wherein, in said step of discriminating said ring/lump distribution defect, an optimum image size is determined on the basis of the frequency distribution of said Voronoi region area, a high-density region image of said optimum image size that represents a relatively high defect-density portion is generated on the basis of said wafer map, geometrical dictionary patterns of said optimum image size are automatically generated, the degree of coincidence between said high-density region image and each of said geometrical dictionary patterns is calculated, and said geometrical dictionary pattern having the highest degree of coincidence is selected.

7. A method of analyzing defect data according to claim 2, wherein, in said step of discriminating said ring/lump distribution defect, an optimum image size is determined on the basis of the frequency distribution of said Voronoi region area, a density image of which the pixel values represent defect density values at their positions is generated to have said optimum image size after said wafer map is weighted for each defect in association with said distance between the adjacent defects or said Voronoi region area, a high-density region image that represents a relatively high defect-density portion is generated by binarizing said density image, geometrical dictionary patterns of said optimum image size are automatically generated, the degree of coincidence between said high-density region image and each of said geometrical dictionary patterns is calculated, and said geometrical dictionary pattern having the highest degree of coincidence is selected.

8. A method of analyzing defect data according to claim 2, wherein, in said step of discriminating said ring/lump distribution defect, said discrimination is executed when the highest frequency of said Voronoi region area is equal to or smaller than a first threshold value, an optimum image size is determined on the basis of said highest frequency, a density image of which the pixel values represent the defect density values at their positions is generated to have said optimum image size after said wafer map is weighted for each defect in association with said distance between the adjacent defects or said Voronoi region area, a high-density region image representing a relatively high defect-density portion is generated by binarizing said density image, geometric dictionary patterns of said optimum image size are automatically generated, the degree of coincidence between said high-density region image and each of said geometric dictionary patterns is calculated, said geometric dictionary pattern having the highest degree of coincidence is selected, said density image and said selected geometric dictionary pattern are superimposed, the ratio between the discrimination analysis values within and out of a pattern is calculated, and the presence or absence of pattern is decided by comparing said ratio between said discrimination analysis values with a second threshold.

9. A method of analyzing defect data according to claim 8, wherein said first and second thresholds are specified by the user.

10. A method of analyzing defect data according to claim 7, wherein an inspection region image of said optimum image size is generated on the basis of information associated with an inspection condition, and the degree of coincidence between said high-density region image and each of said geometric dictionary patterns is calculated by referring to said inspection region image so that only the inspection region can be contributed to the calculation of the degree of coincidence.

11. A method of analyzing defect data according to claim 8, wherein an inspection region image of said optimum image size is generated on the basis of information associated with an inspection condition, and the degree of coincidence between said high-density region image and each of said geometric dictionary patterns is calculated by referring to said inspection region image so that only the inspection region can be contributed to the calculation of the degree of coincidence.

12. A method of analyzing defect data according to claim 9, wherein an inspection region image of said optimum image size is generated on the basis of information associated with an inspection condition, and the degree of coincidence between said high-density region image and each of said geometric dictionary patterns is calculated by referring to said inspection region image so that only the inspection region can be contributed to the calculation of the degree of coincidence.

15. An inspection apparatus according to claim 13, further comprising:

sampling condition generating means for generating sampling conditions associated with said dictionary patterns and other sampling conditions not associated with said patterns; and

sampling means for sampling defects to be reviewed on the basis of said sampling conditions.

16. An inspection apparatus according to claim 14, further comprising:

sampling condition generating means for generating sampling conditions associated with said dictionary patterns and other sampling conditions not associated with said patterns; and

sampling means for sampling defects to be reviewed on the basis of said sampling conditions.

13. An inspection apparatus for a semiconductor substrate, comprising:

inspection means for detecting a foreign substance or pattern defect by irradiating light or an electron beam on the surface of said substrate to scan said surface, thereby producing an output of defect data including at least position coordinates of a defect;

defect data analyzing means for generating a high-density region image representing a high-density portion of said defect on the basis of said position coordinates of said defect included in said defect data, automatically generating a plurality of geometric dictionary patterns, calculating the degree of coincidence between said high-density region image and each of said automatically generated dictionary patterns, and selecting the dictionary pattern having the highest degree of coincidence, thereby executing the discrimination of defect distribution patterns;

pattern information adding means for adding information associated with said patterns to said dictionary patterns;

storage means for storing said defect data, said pattern information and the result of said defect distribution pattern discrimination; and

display means for displaying information associated with said substrate to be inspected or analyzed, said selected dictionary patterns, and said pattern information associated with said selected dictionary patterns.

14. An inspection apparatus for a semiconductor substrate, comprising:

inspection means for detecting a foreign substance or pattern defect by irradiating light or an electron beam on the surface of said substrate to scan said surface, thereby producing an output of defect data including at least position coordinates of a defect;

defect data analyzing means for generating a high-density region image representing a high-density portion of said defect on the basis of said position coordinates of said defect included in said defect data, automatically generating a plurality of geometric dictionary patterns, calculating the degree of coincidence between said high-density region image, each of said automatically generated dictionary patterns and a previously registered user'"'"'s registered dictionary pattern, and selecting the dictionary pattern having the highest degree of coincidence, thereby executing the discrimination of defect distribution patterns;

pattern information adding means for adding information associated with said patterns to said dictionary patterns;

storage means for storing said defect data, said pattern information and the result of said defect distribution pattern discrimination;

display means for displaying information associated with said substrate to be inspected or analyzed, said selected dictionary patterns, and said pattern information associated with said selected dictionary patterns; and

user'"'"'s pattern registration means for registering said user'"'"'s registered dictionary pattern.

17. A review system for a semiconductor substrate, comprising:

at least a single inspection apparatus for detecting a foreign substance or pattern defect by irradiating light or an electron beam on the surface of a substrate to scan said surface, and producing an output of defect data including at least the position coordinates of said defect;

a defect data analyzing apparatus for generating a high-density region image representing a high-density portion of said defect on the basis of the position coordinates of said defect included in said defect data, automatically generating a plurality of geometric dictionary patterns, calculating the degree of coincidence between said high-density region image, each of said automatically generated geometric dictionary patterns, and a previously registered user'"'"'s registered dictionary pattern, and selecting the dictionary pattern having the highest degree of coincidence, thereby executing the discrimination of defect distribution patterns, sampling a point to be reviewed on the basis of said discriminated pattern and a predetermined sampling condition, and generating a report on the basis of said discriminated defect distribution pattern, the review result from said review system and said discriminated defect distribution pattern, and information analyzed according to said review result; and

at least a single light-type or electron-beam-type review apparatus for acquiring a defect image and automatically or manually classifying types of defect.