System and method for identifying systematic defects in wafer inspection using hierarchical grouping and filtering

• US 9,142,014 B2
• Filed: 05/30/2013
• Issued: 09/22/2015
• Est. Priority Date: 05/30/2013
• Status: Active Grant

• Current Assignee: Applied Materials Incorporated
• Original Assignee: DMO Systems Limited (Applied Materials Incorporated)
• Inventors: Lin, Jason Zse-Cherng, Juang, Shauh-Teh
• Primary Examiner(s): Shah, Utpal
• Assistant Examiner(s): TRAN, MAI H
• Application Number:
  US13/906,303
• Publication Number:
  US 20140355867A1
• Time in Patent Office: 845 Days
• Field of Search: None
• US Class Current: 1/1
• CPC Class Codes: G06T 2207/10061 from scanning electron microscope
• CPC Class Codes: G06T 2207/20016 Hierarchical, coarse-to-fine, multiscale or multiresolution image processing; Pyramid transform
• CPC Class Codes: G06T 2207/30148 Semiconductor; IC; Wafer
• CPC Class Codes: G06T 7/001 using an image reference approach

Abstract

A number of wafers of a same semiconductor device are inspected to generate a plurality of candidate defect lists for identifying systematic defects. Each candidate defect list comprises a plurality of candidate defects obtained from inspecting one of the wafers. Each candidate defect is represented by a plurality of defect attributes including a defect location. The candidate defects in every one or more candidate defect lists are processed as a set by stage one grouping and filtering to generate a stage one defect list for each set. The candidate defects in all the stage one defect lists are then processed together by stage two grouping and filtering to generate a final defect lists for systematic defects. The defect attributes of each defect and a design pattern clip extracted from a design database based on the defect location are used in the hierarchical grouping and filtering.

Claims

1. A method of identifying systematic defects in wafer inspection using hierarchical grouping and filtering, comprising the steps of:

collecting a plurality of candidate defect lists obtained from inspecting a plurality of wafers of a same semiconductor device, each of said candidate defect lists consisting of a plurality of candidate defects obtained from inspecting one of said plurality of wafers, each of said candidate defects being represented by a plurality of defect attributes including a defect location;

forming a plurality of data sets by dividing said plurality of candidate defect lists into a plurality of groups with each data set consisting of all the candidate defects of one corresponding group;

performing first stage grouping and filtering for said plurality of data sets to generate a plurality of first stage defect lists with each data set being processed independently to generate a corresponding first stage defect list consisting of a portion of the candidate defects in the corresponding data set by grouping candidate defects of similar or identical features into a same defect group and filtering out candidate defects of nuisance type based on a first pre-set criteria;

performing second stage grouping and filtering for all the candidate defects in said plurality of first stage defect lists collectively together to generate a final defect list for said systematic defects by grouping candidate defects having similar or identical features into a same systematic defect group and filtering out candidate defects of non-systematic type based on a second pre-set criteria;

wherein the defect attributes of each candidate defect are used in said first stage grouping and filtering and said second stage grouping and filtering;

forming a group of previous first stage defect lists from all the first stage defect lists used in generating a previous final defect list by excluding one of the first stage defect lists;

collecting a new data set consisting of all the candidate defects of at least one new candidate defect list obtained from inspecting at least one new wafer of the same semiconductor device;

performing first stage grouping and filtering for said new data set to generate a new first stage defect list consisting of a portion of the candidate defects in the new data set; and

performing second stage grouping and filtering for all the candidate defects in said new first stage defect list and the group of previous first stage defect lists collectively together to generate a new final defect list for said systematic defects.

2. The method as claimed in claim 1, wherein each of said plurality of data sets consists of the candidate defects of only one candidate defect list.

3. The method as claimed in claim 1, wherein each of said plurality of data sets consists of the candidate defects of at least two candidate defect lists.

4. The method as claimed in claim 1, wherein said second stage grouping and filtering uses a recipe different from the recipe used in said first stage grouping and filtering.

5. The method as claimed in claim 1, wherein said final defect list comprises a plurality of defect groups and each defect group is considered as a systematic defect if said second pre-set criteria is satisfied.

6. The method as claimed in claim 5, wherein said second pre-set criteria is the number of candidate defects in the defect group exceeding a pre-set threshold.

7. The method as claimed in claim 5, wherein the plurality of defect groups in said final defect list are ranked based on the number of candidate defects in each defect group for determining criticality of the systematic defects.

8. The method as claimed in claim 5, wherein a critical area is defined for an identified systematic defect according to the defect locations of the candidate defects in the defect group corresponding to the identified systematic defect and the critical area is used as a micro care area in future wafer inspection of the same semiconductor device.

9. The method as claimed in claim 5, wherein a design pattern clip extracted from a design database based on the defect location of each candidate defect is used in said first stage grouping and filtering and said second stage grouping and filtering.

10. The method as claimed in claim 9, wherein in said first stage grouping and filtering and said second stage grouping and filtering, candidate defects having an identical design pattern are grouped in a same defect group.

11. The method as claimed in claim 10, wherein the design pattern of the defect group corresponding to an identified systematic defect is used to search the design database for identifying a plurality of critical areas having the same design pattern, and the plurality of critical areas are used as micro care areas in future wafer inspection of the same semiconductor device.

12. The method as claimed in claim 5, wherein a high-resolution electron-beam image pattern is acquired based on the defect location of each candidate defect and used in said first stage grouping and filtering and said second stage grouping and filtering.

13. The method as claimed in claim 1, wherein the method is used in a wafer production line to sequentially generate final defect lists for monitoring trends of systematic defects by inspecting every group of a predetermined number of wafers to obtain said plurality of candidate defect lists for the hierarchical grouping and filtering, and generating an updated final defect list for each group of wafers.

14. The method as claimed in claim 13, wherein two adjacent groups of the predetermined number of wafers have at least one common wafer.

15. The method as claimed in claim 1, wherein in the step of performing first stage grouping and filtering for said plurality of data sets, each data set is processed independently with said plurality of data sets being processed concurrently.

16. The method as claimed in claim 1, wherein a distributed computing system comprising a plurality of computing devices and a data storage system is used for performing the method, said plurality of computing devices being configured as a plurality of computing nodes for performing the hierarchical grouping and filtering.

17. The method as claimed in claim 16, wherein the step of performing first stage grouping and filtering for said plurality of data sets to generate a plurality of first stage defect lists is accomplished by processing each data set independently in one of said plurality of computing nodes so that said plurality of data sets are processed concurrently.

18. The method as claimed in claim 1, wherein the method is used in a wafer production line to sequentially generate new final defect lists for monitoring trends of systematic defects by sequentially obtaining said at least one new candidate defect list from inspecting at least one new wafer to form said new data set.