Creating defect classifiers and nuisance filters

Patent metadata

Field	Value
Publication number	US-9613411-B2
Application number	US-201414505446-A
Country	US
Kind code	B2
Filing date	Oct 2, 2014
Priority date	Mar 17, 2014
Publication date	Apr 4, 2017
Grant date	Apr 4, 2017

Abstract

Methods and systems for setting up a classifier for defects detected on a wafer are provided. One method includes generating a template for a defect classifier for defects detected on a wafer and applying the template to a training data set. The training data set includes information for defects detected on the wafer or another wafer. The method also includes determining one or more parameters for the defect classifier based on results of the applying step.

First claim

What is claimed is:

1. A method for setting up a classifier for defects detected on a wafer, comprising:
generating a template for a defect classifier for defects detected on a wafer;
applying the template to a training data set, wherein the training data set comprises information for defects detected on the wafer or another water, wherein applying the template to the training data set automatically creates an initial version of the defect classifier, and wherein the training data set does not comprise classifications for defects in the training data set; and
determining one or more parameters for the defect classifier based on results of said applying, wherein determining the one or more parameters comprises tuning one or more parameters of the initial version of the defect classifier to determine the one or more parameters for the defect classifier, and wherein the generating, applying, and determining steps are performed with a computer system.

2. The method of claim 1 , wherein the template is a level-based template, and wherein the defect classifier is a level-based defect classifier.

3. The method of claim 1 , wherein applying the template comprises separating the defects in the training data set into stable populations.

4. The method of claim 1 , wherein tuning the one or more parameters of the initial version of the defect classifier comprises sampling one or more defects from different nodes in the initial version of the defect classifier resulting from the applying step and determining a classification of the one or more sampled defects.

5. The method of claim 1 , wherein the template comprises information for one or more nodes of the defect classifier.

6. The method of claim 1 , wherein generating the template comprises receiving parameters for the template from a user via a user interface provided by the computer system.

7. The method of claim 1 , wherein the template comprises information about user input to be received during one or more steps of the method.

8. The method of claim 1 , wherein the applying and determining steps are performed by executing one or more algorithms from nodes of the initial version of the defect classifier.

9. The method of claim 1 , further comprising displaying results of said applying in a user interface and receiving a user selection of one or more nodes in the results for analysis via the user interface, wherein the determining step further comprises executing one or more algorithms from the one or more selected nodes to analyze the one or more selected nodes.

10. The method of claim 9 , wherein said determining further comprises displaying results of executing the one or more algorithms to the user via the user interface and allowing the user to select at least some of the one or more parameters for the defect classifier.

11. The method of claim 1 , wherein said determining further comprises interrupting the determining to request input from a user for a node of an intermediate version of the defect classifier and upon receipt of the input from the user, resuming the determining step for other nodes in the defect classifier.

12. The method of claim 11 , wherein the input is requested from the user by displaying a sorted list of suggestions for the node from which the user can select the one or more parameters for the node.

13. The method of claim 1 , wherein the applying and determining steps further comprise searching all values of all attributes of the defects in the training set data for one or more best segmentation candidates for one or more nodes of the defect classifier.

14. The method of claim 1 , further comprising determining different defect classifiers for different wafer inspection recipes using the template.

15. The method of claim 1 , wherein the applying and determining steps produce detect classification results for the training data set, and wherein the method further comprises determining information for a wafer inspection recipe for the wafer, the other wafer, or an additional wafer based on the defect classification results.

16. The method of claim 15 , wherein the information for the wafer inspection recipe comprises a determination of whether the wafer inspection recipe is valid.

17. The method of claim 15 , wherein the information for the wafer inspection recipe comprises information for one or more differences between the wafer inspection recipe and another wafer inspection recipe.

18. The method of claim 15 , wherein the information for the wafer inspection recipe comprises information for whether one or more nuisance filters in the wafer inspection recipe adhere to the template.

19. The method of claim 1 , wherein the generating, applying, and determining steps are performed automatically by the computer system.

20. The method of claim 1 , wherein at least one of the generating, applying, and determining steps are performed manually by a user of the computer system.

21. A non-transitory computer-readable medium, storing program instructions executable on a computer system for performing a computer-implemented method for setting up a classifier for defects detected on a wafer, wherein the computer-implemented method comprises:
generating a template for a defect classifier for defects detected on a wafer;
applying the template to a training data set, wherein the training data set comprises information for defects detected on the wafer or another wafer, wherein applying the template to the training data set automatically creates an initial version of the defect classifier, and wherein the training data set does not comprise classifications for defects in the training data set; and
determining one or more parameters for the defect classifier based on results of said applying, wherein determining the one or more parameters comprises tuning one or more parameters of the initial version of the defect classifier to determine the one or more parameters for the defect classifier.

22. A system configured to set up a classifier for defects detected on a wafer, comprising a computer system configured for:
generating a template for a defect classifier for defects detected on a wafer;
applying the template to a training data set, wherein the training data set comprises information for defects detected on the wafer or another wafer, wherein applying the template to the training data set automatically creates an initial version of the defect classifier, and wherein the training data set does not comprise classifications for defects in the training data set; and
determining one or more parameters for the defect classifier based on results of said applying, wherein determining the one or more parameters comprises tuning one or more parameters of the initial version of the defect classifier to determine the one or more parameters for the defect classifier.

23. The system of claim 22 , wherein the template is a level-based template, and wherein the detect classifier is a level-based defect classifier.

24. The system of claim 22 , wherein applying the template comprises separating the defects in the training data set into stable populations.

25. The system of claim 22 , wherein tuning the one or more parameters of the initial version of the defect classifier comprises sampling one or more defects from different nodes in the initial version of the defect classifier resulting from the applying step and determining a classification of the one or more sampled defects.

Assignees

• Kla Tencor Corp
• Kla Tencor Corp

Inventors

• Konuru Raghavan
• Bhatti Naema
• Lennek Michael
• Plihal Martin

Classifications

• G06V10/7788 Primary

  the supervisor being a human, e.g. interactive learning with a human teacher · CPC title

• G06T7/0004 Primary

  Industrial image inspection · CPC title

• G06F18/41

  Interactive pattern learning with a human teacher · CPC title

• G06F18/24323

  Tree-organised classifiers · CPC title

• G06K9/6254

  Physics · mapped topic