Systems and methods for predicting defects and critical dimension using deep learning in the semiconductor manufacturing process

Patent metadata

Field	Value
Publication number	US-11275361-B2
Application number	US-201715814561-A
Country	US
Kind code	B2
Filing date	Nov 16, 2017
Priority date	Jun 30, 2017
Publication date	Mar 15, 2022
Grant date	Mar 15, 2022

Abstract

An initial inspection or critical dimension measurement can be made at various sites on a wafer. The location, design clips, process tool parameters, or other parameters can be used to train a deep learning model. The deep learning model can be validated and these results can be used to retrain the deep learning model. This process can be repeated until the predictions meet a detection accuracy threshold. The deep learning model can be used to predict new probable defect location or critical dimension failure sites.

First claim

What is claimed is:

1. A method comprising:
scanning a wafer with an inspection tool;
confirming presence of at least one defect with a defect review tool, wherein the confirming includes sampling defects;
inputting parameters into a deep learning model, wherein the parameters include a location of the defect with respect to one or more of a design, a care area, or a design clip and wherein the parameters optionally further include one or more of:
focus;
exposure;
a type of the defect;
neighboring design sites; and
a layer type;

predicting defect sites, using a controller, after the inputting based on the deep learning model;
collecting images of the defect sites predicted by the deep learning model;
validating the defect sites, wherein the validating includes confirming a presence of at least one defect at the defect sites in the images; and
retraining the deep learning model with wafer data from the validating, wherein the predicting, the validating, and the retraining are repeated until the deep learning model meets a detection accuracy threshold.

2. The method of claim 1 , wherein the scan is a hot scan.

3. The method of claim 1 , wherein the defect review tool is a scanning electron microscope.

4. The method of claim 1 , wherein the parameters further include optical proximity correction.

5. The method of claim 1 , further comprising generating a heat map of the defect sites.

6. The method of claim 1 , further comprising using the deep learning model to predict defects on a new wafer.

7. The method of claim 1 , further comprising using the deep learning model to predict defects for a new semiconductor manufacturing process on the wafer.

8. A non-transitory computer readable medium storing a program configured to instruct a processor to execute the method of claim 1 .

9. A method comprising:
scanning a wafer with a scanning electron microscope;
calculating, using a controller, critical dimension variation across the wafer at sample sites;
inputting parameters into a deep learning model, wherein the parameters include a location where critical dimension is measured with respect to one of a design, a care area, or a design clip and wherein the parameters optionally further include one or more of:
focus;
exposure;
neighboring design sites; and
a layer type;

predicting the critical dimension at sites across the wafer after the inputting, using the controller, based on the deep learning model;
collecting images of the sites from the deep learning model;
validating the critical dimension at the sites, wherein the validating includes confirming the critical dimension at the sites in the images; and
retraining the deep learning model with wafer data from the validating, wherein the predicting, the validating, and the retraining are repeated until the deep learning model meets a detection accuracy threshold.

10. The method of claim 9 , wherein the parameters further include optical proximity correction.

11. The method of claim 9 , further comprising generating a heat map of critical dimension variation.

12. The method of claim 9 , further comprising using the deep learning model to predict critical dimension on a new wafer.

13. The method of claim 9 , further comprising using the deep learning model to predict critical dimension for a new semiconductor manufacturing process on the wafer.

14. A non-transitory computer readable medium storing a program configured to instruct a processor to execute the method of claim 9 .

15. A system comprising:
a controller in electronic communication with a scanning electron microscope or with the scanning electron microscope and an optical inspection tool, wherein the controller includes a processor and an electronic data storage unit in electronic communication with the processor, and wherein the controller is configured to:
receive results of a review of a wafer, wherein the results are one of defect locations or critical dimension;
input parameters in a deep learning model, wherein the parameters include a location where the results are measured with respect to one of a design, a care area, or a design clip and wherein the parameters optionally further include one or more of:
focus;
exposure;
neighboring design sites; and
a layer type;

predict additional results at sites across the wafer after the inputting, wherein the additional results are the same as the one of the defect locations or the critical dimension;
receive collected images of the sites from the deep learning model;
receive validation of the results at the sites across the wafer, wherein the validation confirms a presence of the additional results at the defect sites in the images; and
retrain the deep learning model with wafer data from the validating, wherein the predict step, the validate step, and the retrain step are repeated until the deep learning model meets a detection accuracy threshold.

16. The system of claim 15 , further comprising the scanning electron microscope in electronic communication with the controller.

17. The system of claim 15 , further comprising the optical inspection tool and the scanning electron microscope in electronic communication with the controller.

18. The system of claim 15 , wherein the controller is configured to generate a heat map of results variation.

19. The system of claim 15 , wherein the controller is configured to use the deep learning model to predict results on a new wafer.

20. The system of claim 15 , wherein the controller is configured to use the deep learning model to predict results for a new semiconductor manufacturing process on the wafer.

Assignees

• Kla Tencor Corp

Inventors

• Yati Arpit

Classifications

• H10P74/203 Primary

  Structural properties, e.g. testing or measuring thicknesses, line widths, warpage, bond strengths or physical defects · CPC title

• H10P74/23

  characterised by multiple measurements, corrections, marking or sorting processes · CPC title

• G05B19/41875 Primary

  characterised by quality surveillance of production · CPC title

• G05B13/027 Primary

  using neural networks only · CPC title

• G06N3/0499

  Feedforward networks · CPC title