Method of deep learning-based examination of a semiconductor specimen and system thereof

Patent metadata

Field	Value
Publication number	US-11010665-B2
Application number	US-201715668623-A
Country	US
Kind code	B2
Filing date	Aug 3, 2017
Priority date	Dec 22, 2015
Publication date	May 18, 2021
Grant date	May 18, 2021

Abstract

There are provided system and method of segmentation a fabrication process (FP) image obtained in a fabrication of a semiconductor specimen. The method comprises: upon obtaining a Deep Neural Network (DNN) trained to provide segmentation-related data, processing a fabrication process (FP) sample using the obtained trained DNN and, resulting from the processing, obtaining by the computer segments-related data characterizing the FP image to be segmented, the obtained segments-related data usable for automated examination of the semiconductor specimen. The DNN is trained using a segmentation training set comprising a plurality of first training samples and ground truth data associated therewith, each first training sample comprises a training image; FP sample comprises the FP image to be segmented.

First claim

The invention claimed is:

1. A method of segmentation of a fabrication process (FP) image obtained in a fabrication of a semiconductor specimen, the method comprising:
upon obtaining, by a computer, a Deep Neural Network (DNN) trained to provide segmentation-related data, using the obtained trained DNN to process a fabrication process (FP) sample, wherein the DNN is trained using a segmentation training set comprising a plurality of first training samples and ground truth data associated therewith, each first training sample comprises a training image, and the segmentation-related data is selected from a group comprising a grayscale image in which different color values for each pixel correspond to different segments in the training image, a representation of edges of each segment, and a representation of vertices of each segment, and wherein the FP sample comprises the FP image, wherein the segmentation training set further comprises a plurality of augmented training samples, wherein at least a part of the augmented training samples is obtained by tone mapping; and
resulting from the processing, obtaining, by the computer, segments-related data characterizing the FP image, wherein the obtained segments-related data are usable for automated examination of the semiconductor specimen.

2. The method of claim 1 , wherein the obtained segments-related data are informative of at least one of per-pixel values, per-pixel labels, CAD polygons, CAD models, regions of interest in the FP image, background of the FP image and foreground of the FP image.

3. The method of claim 1 , wherein the FP image is selected from the group consisting of images resulting from different examination modalities and a design data-based image of the semiconductor specimen.

4. The method of claim 1 , wherein the obtained segments-related data are usable for automatic defect classification (ADC) to define a location of a defect with regard to a background.

5. The method of claim 1 , wherein the obtained segments-related data are usable for automatic defect review (ADR) to apply segment-specific detection thresholds.

6. The method of claim 1 , wherein the obtained segments-related data are usable for detecting locations of potential defects.

7. The method of claim 1 , wherein the obtained segments-related data are usable by another examination-related application using another DNN trained using a training set comprising ground truth data specific for the another examination-related application.

8. The method of claim 1 , wherein each of the first training samples comprises at least one training image obtained by an examination modality that is not available when obtaining the FP sample.

9. The method of claim 1 , wherein at least a part of the segmented training samples is obtained by augmenting at least part of the first training samples and augmented ground truth data associated with the augmented training samples.

10. The method of claim 9 , wherein at least a part of the augmented training samples is obtained by one or more augmentation techniques from the group consisting of adding noise, blurring, and generating synthetic images.

11. A non-transitory computer readable medium comprising instructions that, when executed by a computer, cause the computer to perform a method of segmentation of a fabrication process (FP) image in a fabrication of a semiconductor specimen, the method comprising:
upon obtaining, by a computer, a Deep Neural Network (DNN) trained to provide segmentation-related data, using the obtained trained DNN to process a fabrication process (FP) sample, wherein the DNN is trained using a segmentation training set comprising a plurality of first training samples and ground truth data associated therewith, each first training sample comprises a training image, and the segmentation-related data is selected from a group comprising a grayscale image in which different color values for each pixel represent different segments in the training image, a representation of edges of each segment, and a representation of vertices of each segment, and wherein the FP sample comprises the FP image, wherein the segmentation training set further comprises a plurality of augmented training samples, wherein at least a part of the augmented training samples is obtained by tone mapping; and
resulting from the processing, obtaining, by the computer, segments-related data characterizing the FP image, wherein the obtained segments-related data are usable for automated examination of the semiconductor specimen.

12. The non-transitory computer readable medium of claim 11 , wherein the obtained segments-related data are informative of at least one of per-pixel segmentation, per-pixel labels, CAD polygons, CAD models, regions of interest in the FP image, background and foreground of the FP image.

13. The non-transitory computer readable medium of claim 11 , wherein the FP image is selected from the group consisting of images resulting from different examination modalities and a design data-based image of the semiconductor specimen.

14. The non-transitory computer readable medium of claim 11 , wherein at least a part of the segmentation training set is obtained by augmenting at least part of the first training samples and augmented ground truth data associated with the augmented training samples and informative of segments-related data associated with the augmented training samples.

15. The non-transitory computer readable medium of claim 14 , wherein at least a part of the augmented training samples is obtained by one or more augmentation techniques from the group consisting of adding noise, blurring, and generating synthetic images.

16. A system usable for examination of a semiconductor specimen, the system comprising a processing and memory block (PMB) operatively connected to an input interface and an output interface, wherein:
the input interface is configured to receive one or more fabrication process (FP) samples each comprising a FP image to be segmented; the PMB is configured:
to obtain a Deep Neural Network (DNN) trained to provide segmentation-related data and to process a fabrication process (FP) sample using the obtained trained DNN, wherein the DNN is trained using a segmentation training set comprising a plurality of first training samples and ground truth data associated therewith, each first training sample comprises a training image, and the segmentation data is selected from a group comprising a grayscale image in which different color values for each pixel represent different segments in the training image, a representation of edges of each segment and a representation of vertices of each segment and wherein the FP sample comprises the FP image to be segmented, wherein the segmentation training set further comprises a plurality of augmented training samples, wherein at least a part of the augmented training samples is obtained by tone mapping; and
resulting from the processing, to obtain segments-related data characterizing the FP image, wherein the obtained segments-related data are usable for automated examination of the semiconductor specimen, and the output interface is configured to output the obtained segments-related data.

17. The system of claim 16 , wherein at least a part of the segmentation training set is obtained by augmenting at least part of the first training samples and augmented ground truth data associated with the augmented training samples and informative of segments-related data associated with the augmented training samples.

18. The system of claim 17 , wherein at least a

Assignees

• Applied Materials Israel Ltd

Inventors

• Karlinsky Leonid
• Cohen Boaz
• Kaizerman Idan
• Rosenman Efrat
• Batikoff Amit
• Ravid Daniel
• Rosenweig Moshe

Classifications

• G06V10/82

  using neural networks · CPC title

• G06V10/764

  using classification, e.g. of video objects · CPC title

• G06F18/24133

  Distances to prototypes · CPC title

• G06N3/045

  Combinations of networks · CPC title

• G06V10/993 Primary

  Evaluation of the quality of the acquired pattern · CPC title