Single image detection

Patent metadata

Field	Value
Publication number	US-10186026-B2
Application number	US-201615353210-A
Country	US
Kind code	B2
Filing date	Nov 16, 2016
Priority date	Nov 17, 2015
Publication date	Jan 22, 2019
Grant date	Jan 22, 2019

Abstract

Methods and systems for detecting defects on a specimen are provided. One system includes a generative model. The generative model includes a non-linear network configured for mapping blocks of pixels of an input feature map volume into labels. The labels are indicative of one or more defect-related characteristics of the blocks. The system inputs a single test image into the generative model, which determines features of blocks of pixels in the single test image and determines labels for the blocks based on the mapping. The system detects defects on the specimen based on the determined labels.

First claim

What is claimed is:

1. A system configured to detect defects on a specimen, comprising:
an imaging subsystem configured for generating images of a specimen, wherein the imaging subsystem comprises at least an energy source configured to direct energy to the specimen and at least a detector configured to detect energy from the specimen;
one or more computer subsystems coupled to the imaging subsystem, wherein the one or more computer subsystems are configured for acquiring a single test image for a portion of the specimen generated by the imaging subsystem; and
one or more components executed by the one or more computer subsystems, wherein the one or more components comprise:
a generative model, wherein the generative model comprises a non-linear network configured for mapping blocks of pixels of an input feature map volume into labels, and wherein the labels are indicative of one or more defect-related characteristics of the blocks;
wherein the one or more computer subsystems are further configured for inputting the single test image into the generative model;
wherein the generative model is configured for:
separating the single test image into multiple blocks of pixels;

for at least one of the multiple blocks of pixels, determining a feature of the at least one of the multiple blocks based on only the pixels in the at least one of the multiple blocks; and
selecting one of the labels for the at least one of the multiple blocks based on the determined feature and the mapping of the blocks of the pixels of the input feature map volume into the labels; and

wherein the one or more computer subsystems are further configured for detecting defects in the portion of the specimen based on the selected label for the at least one of the multiple blocks.

2. The system of claim 1 , wherein the generative model is a deep generative model.

3. The system of claim 1 , wherein the generative model is a machine learning model.

4. The system of claim 1 , wherein the generative model is a convolution neural network.

5. The system of claim 1 , wherein detecting the defects does not comprise aligning the single test image to any other image.

6. The system of claim 1 , wherein detecting the defects does not comprise comparing the single test image to any other image.

7. The system of claim 1 , wherein detecting the defects does not comprise statistical based defect detection.

8. The system of claim 1 , wherein the labels indicate whether input features in the input feature map volume are associated with defects or are not associated with defects.

9. The system of claim 1 , wherein the labels indicate a type of defect to which input features in the input feature map volume are associated.

10. The system of claim 1 , wherein the one or more computer subsystems are further configured for determining a type of the defects in the portion of the specimen based on the selected label for the at least one of the multiple blocks.

11. The system of claim 1 , wherein selecting one of the labels for the at least one of the multiple blocks comprises selecting only one of the labels for a combination of the multiple blocks.

12. The system of claim 1 , wherein the one or more computer subsystems are further configured for generating a training dataset used for training the generative model, and wherein the training dataset comprises a set of pairs of a portion of design information for the specimen and an image generated for the portion of the design information.

13. The system of claim 12 , wherein at least one of the pairs in the set is generated for a defect detected on the specimen or another specimen.

14. The system of claim 12 , wherein at least one of the pairs in the set is generated for a synthetic defect.

15. The system of claim 12 , wherein at least one of the pairs in the set is generated for a simulated defect.

16. The system of claim 12 , wherein at least one of the pairs in the set is generated for a defect detected by process window qualification.

17. The system of claim 12 , wherein the design information comprises design data.

18. The system of claim 12 , wherein the design information comprises simulated images generated from design data.

19. The system of claim 12 , wherein at least one of the images in the pairs comprises an actual image generated of the specimen or another specimen by the imaging subsystem.

20. The system of claim 12 , wherein at least one of the images in the pairs comprises a synthetic image generated based on 1) the design information for the at least one image and 2) other images generated by the imaging subsystem for other specimens.

21. The system of claim 12 , wherein at least one of the images in the pairs comprises a simulated image generated based on 1) the design information for the at least one image and 2) one or more characteristics of the imaging subsystem.

22. The system of claim 12 , wherein the one or more computer subsystems are further configured for detecting other defects in the images in the pairs in the set and associating labels with pixels in the images corresponding to results of said detecting the other defects.

23. The system of claim 22 , wherein the one or more computer subsystems are further configured for performing supervised training of the generative model by progressively modifying parameters of the generative model until features determined by the generative model for the images in the pairs are mapped to the labels associated with the pixels in the images.

24. The system of claim 23 , wherein the one or more computer subsystems are further configured for modifying the training dataset by adding a new pair to the set, detecting one or more additional defects in an image in the new pair, associating one or more additional labels with pixels in the image corresponding to results of said detecting the one or more additional defects, and re-performing the supervised training by progressively modifying the parameters until the features determined for the images in the pairs and the new pair by the generative model are mapped to the labels associated with the pixels in the images in the pair and the new pair.

25. The system of claim 1 , wherein the one or more computer subsystems are further configured for generating multiple perspectives of the single test image, and wherein the single test image input by the one or more computer subsystems into the generative model comprises the multiple perspectives of the single test image.

26. The system of claim 1 , wherein the one or more computer subsystems are further configured for detecting the defects based on the selected label in combination with design information for the specimen.

27. The system of claim 1 , wherein the one or more computer subsystems are further configured for detecting the defects based on the selected label and without design information for the specimen.

28. The system of claim I, wherein the imaging subsystem is an electron beam based imaging subsystem.

29. The system of claim 1 , wherein the imaging subsystem is an optical based imaging subsystem.

30. The system of claim 1 , wherein the imaging subsystem is an inspection subsystem.

31. The system of claim 1 , wherein the imaging subsystem is a defect review subsystem.

32. The system of claim 1 , wherein the imaging subsyst

Assignees

• Kla Tencor Corp
• Kla Tencor Corp

Inventors

• Karsenti Laurent
• Bhaskar Kris
• Jordan Iii John Raymond
• Venkataraman Sankar
• Carmon Yair

Classifications

• G01N2021/8893

  providing a video image and a processed signal for helping visual decision · CPC title

• G01N2021/8858

  Flaw counting · CPC title

• G01N21/956

  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 H10P74/20)} · CPC title

• G01N21/8851 Primary

  Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges (G01N21/8806 and G01N21/93 - G01N21/95692 take precedence; optical measurement of dimensions G01B11/00; optical scanning G02B26/10; image transformation G06T3/00; computerised image enhancement G06T5/00; image processing per se for flaw detection G06T7/0002) · CPC title

• G06T2207/20084

  Artificial neural networks [ANN] · CPC title