Accelerating semiconductor-related computations using learning based models

Patent metadata

Field	Value
Publication number	US-2017200260-A1
Application number	US-201715402169-A
Country	US
Kind code	A1
Filing date	Jan 9, 2017
Priority date	Jan 11, 2016
Publication date	Jul 13, 2017
Grant date	—

Abstract

Methods and systems for performing one or more functions for a specimen using output simulated for the specimen are provided. One system includes one or more computer subsystems configured for acquiring output generated for a specimen by one or more detectors included in a tool configured to perform a process on the specimen. The system also includes one or more components executed by the one or more computer subsystems. The one or more components include a learning based model configured for performing one or more first functions using the acquired output as input to thereby generate simulated output for the specimen. The one or more computer subsystems are also configured for performing one or more second functions for the specimen using the simulated output.

First claim

What is claimed is:

1 . A system configured to perform one or more functions for a specimen using output simulated for the specimen, comprising:
one or more computer subsystems configured for acquiring output generated for a specimen by one or more detectors included in a tool configured to perform a process on 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 configured for performing one or more first functions using the acquired output as input to thereby generate simulated output for the specimen, and wherein the one or more computer subsystems are further configured for performing one or more second functions for the specimen using the simulated output.

2 . The system of claim 1 , wherein the learning based model comprises a machine learning model.

3 . The system of claim 1 , wherein the learning based model comprises a deep learning model.

4 . The system of claim 1 , wherein the learning based model comprises a neural network.

5 . The system of claim 1 , wherein the learning based model comprises a convolution neural network.

6 . The system of claim 1 , wherein the learning based model comprises a deep neural network.

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

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

9 . The system of claim 1 , wherein the learning based model is further configured for convolution with upsampled filters.

10 . The system of claim 1 , wherein the one or more computer subsystems are further configured for training the learning based model with results generated from one or more non-learning based algorithms, and wherein the training is performed before the one or more first functions are performed by the learning based model.

11 . The system of claim 1 , wherein the one or more computer subsystems are further configured for pre-training the learning based model using a mother network and training the pre-trained learning based model, and wherein the pre-training and the training are performed before the one or more first functions are performed by the learning based model.

12 . The system of claim 1 , wherein the one or more computer subsystems are further configured for supervised training of the learning based model with user provided information, and wherein the training is performed before the one or more first functions are performed by the learning based model.

13 . The system of claim 1 , wherein the one or more computer subsystems are further configured for generating synthetic data for the specimen and training the learning based model using the synthetic data.

14 . The system of claim 1 , wherein the learning based model is further configured for performing the one or more first functions using the acquired output as a first input and information for the specimen as a second input.

15 . The system of claim 14 , wherein the information for the specimen comprises design data for the specimen.

16 . The system of claim 15 , wherein the learning based model is further configured for altering the one or more first functions based on the design data.

17 . The system of claim 1 , wherein the one or more first functions comprise labeling pixels in the output as associated with defects or as not associated with defects.

18 . The system of claim 1 , wherein the one or more first functions comprise identifying a classification of defects detected on the specimen.

19 . The system of claim 1 , wherein the one or more first functions comprise generating a reference for the specimen, and wherein the one or more second functions comprise comparing the reference to the output and detecting defects on the specimen based on results of the comparing.

20 . The system of claim 1 , wherein the one or more first functions comprise aligning at least a portion of the output, to a reference.

21 . The system of claim 1 , wherein the one or more second functions comprise detecting defects on the specimen based on the simulated output.

22 . The system of claim 1 , wherein the output generated for the specimen by the one or more detectors comprises output generated in multiple modes.

23 . The system of claim 1 , wherein the one or more computer subsystems are further configured for acquiring additional output generated for an additional specimen by one or more additional detectors included in an additional tool configured to perform a process on the additional specimen, wherein the learning based model is further configured for performing the one or more first functions using the acquired additional output as the input to thereby generate the simulated output for the additional specimen, and wherein the one or more computer subsystems are further configured tor performing the one or more second functions for the additional specimen using the simulated output.

24 . The system of claim 1 , wherein the one or more computer subsystems are further configured for training the learning based model with results generated from one or more non-learning based algorithms, wherein the training is performed before the one or more first functions are performed by the learning based model, and wherein the results generated from the one or more non-learning based algorithms comprise defect classification results.

25 . The system of claim 24 , wherein the defect classification results are generated by running a hot scan on another specimen to thereby detect defects on the other specimen, reviewing at least a portion of the detected defects, and classifying at least the portion of the detected defects using at least results of the reviewing.

26 . The system of claim 25 , wherein classifying at least the portion of the detected defects is performed by a user.

27 . The system of claim 24 , wherein the one or more first functions comprise predicting which defects detected with the output are defects of interest.

28 . The system of claim 27 , wherein said predicting is performed based on defect attributes determined from the output.

29 . The system of claim 27 , wherein said predicting is performed based on the output.

30 . The system of claim 27 , wherein said predicting is performed based on images generated from the output.

31 . The system of claim 1 , wherein the learning based model is further configured for generating a classifier for nuisance filtering for an inspection process for the specimen based on results of the inspection process generated for multiple specimens in multiple lots, and wherein the results of the inspection process comprise image characteristics, defect attributes, or a combination thereof.

32 . The system of claim 1 , wherein the learning based model is further configured for generating a difference filter for detecting defects for an inspection process for the specimen based on results of the inspection process generated for multiple specimens in multiple lots, and wherein the results of the inspection process comprise image characteristics, defect attributes, or a combination thereof.

33 . The system of claim 1 , wherein the process comprises a metrology process.

34 . The s

Assignees

• Kla Tencor Corp

Inventors

• Bhaskar Kris
• Young Scott
• Roulo Mark
• Zhang Jing
• Karsenti Laurent
• Mahadevan Mohan
• Brauer Bjorn

Classifications

• G06T7/0004 Primary

  Industrial image inspection · CPC title

• G06V30/19173

  Classification techniques · CPC title

• G06V10/82

  using neural networks · CPC title

• G06F18/214

  Generating training patterns; Bootstrap methods, e.g. bagging or boosting · CPC title

• G06F18/24133

  Distances to prototypes · CPC title