Generating simulated output for a specimen

• US 10,043,261 B2
• Filed: 01/09/2017
• Issued: 08/07/2018
• Est. Priority Date: 01/11/2016
• Status: Active Grant

• Current Assignee: KLA-Tencor Corporation
• Original Assignee: KLA-Tencor Corporation
• Inventors: Bhaskar, Kris, Zhang, Jing, Chen, Grace Hsiu-Ling, Kulkarni, Ashok, Karsenti, Laurent
• Primary Examiner(s): Kassa, Yosef
• Application Number:
  US15/402,094
• Publication Number:
  US 20170200265A1
• Time in Patent Office: 575 Days
• Field of Search: 382141, 382145, 382147, 382149, 382152, 348 86, 348125, 348126
• CPC Class Codes: G03F 1/84 Inspecting
• CPC Class Codes: G03F 1/86 by charged particle beam [CPB]
• CPC Class Codes: G03F 7/7065 Defects, e.g. optical inspection of patterned layer for defects
• CPC Class Codes: G06T 2207/10061 from scanning electron microscope
• CPC Class Codes: G06T 2207/20081 Training; Learning
• CPC Class Codes: G06T 2207/30148 Semiconductor; IC; Wafer
• CPC Class Codes: G06T 7/0004 Industrial image inspection
• CPC Class Codes: G06T 7/0006 using a design-rule based approach
• CPC Class Codes: G06T 7/001 using an image reference approach

Abstract

Methods and systems for generating simulated output for a specimen are provided. One method includes acquiring information for a specimen with one or more computer systems. The information includes at least one of an actual optical image of the specimen, an actual electron beam image of the specimen, and design data for the specimen. The method also includes inputting the information for the specimen into a learning based model. The learning based model is included in one or more components executed by the one or more computer systems. The learning based model is configured for mapping a triangular relationship between optical images, electron beam images, and design data, and the learning based model applies the triangular relationship to the input to thereby generate simulated images for the specimen.

Claims

1. A system configured to generate simulated output for a specimen, comprising:

one or more computer subsystems configured for acquiring information for a specimen, wherein the information comprises at least one of an actual optical image of the specimen, an actual electron beam image of the specimen, and design data for the specimen; and

one or more components executed by the one or more computer subsystems, wherein the one or more components comprise a learning based model, wherein the learning based model is configured for mapping a triangular relationship between optical images, electron beam images, and design data, wherein the one or more computer subsystems are configured to input the information for the specimen into the learning based model, and wherein the learning based model applies the triangular relationship to the input to thereby generate simulated output for the specimen.

2. The system of claim 1, wherein the information that is input to the learning based model comprises the actual optical image of the specimen, and wherein the simulated output comprises a simulated electron beam image representing an actual electron beam image generated for the specimen by an electron beam tool.

3. The system of claim 1, wherein the information that is input to the learning based model comprises the actual optical image of the specimen, and wherein the simulated output comprises simulated design data for the specimen.

4. The system of claim 1, wherein the information that is input to the learning based model comprises the actual electron beam image of the specimen, and wherein the simulated output comprises simulated design data for the specimen.

5. The system of claim 1, wherein the information for the specimen further comprises actual optical images of the specimen generated at different values of a parameter of an optical tool, wherein the information that is input to the learning based model comprises the actual optical images of the specimen, and wherein the simulated output comprises a simulated electron beam image of the specimen representing an actual electron beam image generated for the specimen by an electron beam tool.

6. The system of claim 1, wherein the information for the specimen further comprises actual optical images of the specimen generated at different values of a parameter of an optical tool, wherein the information that is input to the learning based model comprises the actual optical images of the specimen, wherein the input to the learning based model further comprises another different value of the parameter of the optical tool, and wherein the simulated output comprises a simulated optical image of the specimen representing an optical image generated at the other different value of the parameter of the optical tool.

7. The system of claim 1, wherein the information that is input to the learning based model comprises the actual optical image of the specimen and the design data for the specimen, and wherein the simulated output comprises a simulated electron beam image for the specimen representing an actual electron beam image generated for the specimen by an electron beam tool.

8. The system of claim 1, wherein the information for the specimen further comprises actual optical images of the specimen corresponding to different values of a parameter of a process performed on the specimen, wherein the information that is input to the learning based model comprises the actual optical images of the specimen, wherein the input to the learning based model further comprises another different value of the parameter of the process, and wherein the simulated output comprises a simulated optical image of the specimen corresponding to the other different value of the parameter of the process.

9. The system of claim 1, wherein the information for the specimen further comprises actual optical images of the specimen generated at different values of a parameter of an optical tool, wherein the information that is input to the learning based model comprises the actual optical images of the specimen and the design data for the specimen, and wherein the simulated output comprises defect classifications for defects detected on the specimen.

10. The system of claim 1, wherein the information that is input to the learning based model comprises run time input, and wherein the information for the run time input comprises information for a pattern not included in input to the learning based model used for training the learning based model.

11. The system of claim 1, wherein the learning based model is not configured for generating the simulated output for the specimen by performing pixel-value interpolation.

12. The system of claim 1, wherein the learning based model is not configured for generating the simulated output for the specimen by performing pixel-value extrapolation.

13. The system of claim 1, wherein the learning based model is further configured for mapping a relationship between patterned features on the specimen and one or more of the optical images, the electron beam images, and the design data.

14. The system of claim 1, wherein the information for the specimen further comprises two or more actual optical images of the specimen, and Wherein the two or more actual optical images comprise two or more actual optical images corresponding to different values of a parameter of an optical tool, two or more actual optical images of a layer on the specimen generated before and after a process is performed on the specimen, two or more actual optical images of different physical layers on the specimen, two or more actual optical images of the specimen generated by different optical tools, two or more actual optical images of the specimen corresponding to different values of a parameter of a process performed on the specimen, or a combination thereof.

15. The system of claim 1, wherein the information for the specimen further comprises two or more actual electron beam images of the specimen, and wherein the two or more actual electron beam images comprise two or more actual electron beam images corresponding to different values of a parameter of an electron beam tool, two or more actual electron beam images of a layer on the specimen generated before and after a process is performed on the specimen, two or more actual electron beam images of different physical layers on the specimen, two or more actual electron beam images of the specimen generated by different electron beam tools, two or more actual electron beam images of the specimen corresponding to different values of a parameter of a process performed on the specimen, or a combination thereof.

16. The system of claim 1, wherein the design data for the specimen comprises the design data stacked by:

different selected patterns in the same layer, different layers, different materials, or a combination thereof.

17. The system of claim 1, wherein the information for the specimen further comprises information for patterned features formed on the specimen, and wherein the information for the patterned features is generated experimentally or theoretically.

18. The system of claim 1, wherein the learning based model comprises a discriminative model.

19. The system of claim 18, wherein the discriminative model comprises a support vector machine, a support vector regression, a convolutional neural network, or a recurrent neural network.

20. The system of claim 1, wherein the learning based model comprises a parametric or non-parametric Bayesian approach.

21. The system of claim 20, wherein likelihood probability is modeled for the parametric or non-parametric Bayesian approach by a Gaussian Mixture Model, a Dirichlet Mixture Model, a Gaussian process, or a deep Gaussian process.

22. The system of claim 20, wherein prior and posterior are modeled for the parametric or non-parametric Bayesian approach by a variational inference or a Markov Chain Monte Carlo.

23. The system of claim 1, wherein the learning based model comprises a generative model.

24. The system of claim 23, wherein the generative model comprises an autoencoder variant, a generative adversarial network, a conditional generative adversarial network, or a deep generative model.

25. The system of claim 1, wherein the one or more computer subsystems are configured for training the learning based model using a conditional generative adversarial network.

26. The system of claim 25, wherein the training comprises inputting a noise prior to a generative portion of the conditional generative adversarial network.

27. A non-transitory computer-readable medium, storing program instructions executable on one or more computer systems for performing a computer-implemented method for generating simulated output for a specimen, wherein the computer-implemented method comprises:

acquiring information for a specimen with the one or more computer systems, wherein the information comprises at least one of an actual optical image of the specimen, an actual electron beam image of the specimen, and design data for the specimen; and

inputting the information for the specimen into a learning based model, wherein the learning based model is included in one or more components executed by the one or more computer systems, wherein the learning based model is configured for mapping a triangular relationship between optical images, electron beam images, and design data, and wherein the learning based model applies the triangular relationship to the input to thereby generate simulated output for the specimen.

28. A computer-implemented method for generating simulated output for a specimen, comprising:

acquiring information for a specimen with one or more computer systems, wherein the information comprises at least one of an actual optical image of the specimen, an actual electron beam image of the specimen, and design data for the specimen; and

inputting the information for the specimen into a learning based model, wherein the learning based model is included in one or more components executed by the one or more computer systems, wherein the learning based model is configured for mapping a triangular relationship between optical images, electron beam images, and design data, and wherein the learning based model applies the triangular relationship to the input to thereby generate simulated output for the specimen.