Optical Die to Database Inspection

Patent metadata

Field	Value
Publication number	US-2017191948-A1
Application number	US-201615391753-A
Country	US
Kind code	A1
Filing date	Dec 27, 2016
Priority date	Jan 4, 2016
Publication date	Jul 6, 2017
Grant date	—

Abstract

Methods and systems for detecting defects on a wafer are provided. One system includes one or more computer subsystems configured for generating a rendered image based on information for a design printed on the wafer. The rendered image is a simulation of an image generated by an optical inspection subsystem for the design printed on the wafer. Generating the rendered image includes one or more steps, and the computer subsystem(s) are configured for performing at least one of the one or more steps by executing a generative model. The computer subsystem(s)) are also configured for comparing the rendered image to an optical image of the wafer generated by the optical inspection subsystem. The design is printed on the wafer using a reticle. In addition, the computer subsystem(s) are configured for detecting defects on the wafer based on results of the comparing.

First claim

What is claimed is:

1 . A system configured to detect defects on a wafer, comprising:
an optical inspection subsystem comprising at least a light source and a detector, wherein the light source is configured to generate light that is directed to a wafer, and wherein the detector is configured to detect light from the wafer and to generate images responsive to the detected light; and one or more computer subsystems configured for:
generating a rendered image based on information for a design printed on the wafer, wherein the rendered image is a simulation of an image generated by the optical inspection subsystem for the design printed on the wafer, wherein generating the rendered image comprises one or more steps, and wherein the one or more computer subsystems are further configured for performing at least one of the one or more steps by executing a generative model;
comparing the rendered image to an optical image of the wafer generated by the optical inspection subsystem, wherein the design is printed on the wafer using a reticle; and
detecting defects on the wafer based on results of the comparing.

2 . The system of claim 1 , wherein the at least one step is further performed by executing the generative model and an additional generative model.

3 . The system of claim 2 , wherein the one or more steps comprise a first step and a second step, and Wherein the one or more computer subsystems are further configured for performing the first step by executing the generative model and performing the second step by executing the additional generative model.

4 . The system of claim 2 , wherein the one or more computer subsystems are further configured for separately training the generative model and the additional generative model.

5 . The system of claim 1 , wherein the generative model is configured as a convolutional neural network.

6 . The system of claim 1 , wherein the generative model is configured as an auto-encoder.

7 . The system of claim 1 , wherein the one or more steps comprise converting polygons in the information for the design to a gray scale image.

8 . The system of claim 7 , wherein said converting the polygons is performed with sub-pixel accuracy.

9 . The system of claim 7 , wherein the one or more steps further comprise generating a modified gray scale image by applying bias correction and corner rounding to the gray scale image.

10 . The system of claim 9 , wherein the one or more steps further comprise estimating a near field of the wafer based on the modified gray scale image and the information for the design printed on the wafer.

11 . The system of claim 10 , wherein the one or more steps further comprise generating an initial rendered image that is another simulation of the image generated by the optical inspection subsystem for the design printed on the wafer based on the near field and an optical model of the optical inspection subsystem.

12 . The system of claim 11 , wherein the one or more steps further comprise generating the rendered image from the initial rendered image by modifying the initial rendered image to minimize differences between the initial rendered image and the optical image generated by the optical inspection subsystem.

13 . The system of claim 1 , wherein the one or more steps comprise calculating a near field of the wafer based on the information for the design printed on the wafer, and wherein the information for the design printed on the wafer comprises geometry and material characteristics.

14 . The system of claim 13 , wherein the one or more steps further comprise generating an initial rendered image that is another simulation of the image generated by the optical inspection subsystem for the design printed on the wafer based on the near field and an optical model of the optical inspection subsystem.

15 . The system of claim 14 , wherein the one or more steps further comprise generating the rendered image from the initial rendered image by modifying the initial rendered image to minimize differences between the initial rendered image and the optical image generated by the optical inspection subsystem.

16 . The system of claim 1 , wherein the one or more computer subsystems are further configured for training the generative model based on: one or more additional rendered images for one or more selected sites on one or more other wafers generated by performing said generating for the one or more selected sites; and one or more optical images generated by the optical inspection subsystem for the one or more selected sites on the one or more other wafers.

17 . The system of claim 16 , wherein generating the rendered image further comprises modifying an initial rendered image to minimize differences between the initial rendered image and the optical image generated by the optical inspection subsystem, wherein the initial rendered image is generated offline, and wherein generating the rendered image is performed online.

18 . The system of claim 17 , wherein the initial rendered image is generated for a whole die in the design printed on the wafer, and wherein generating the rendered image is further performed online for an entirety of the wafer.

19 . The system of claim 1 , wherein the one or more computer subsystems are further configured for training one or more additional models used for said generating based on: one or more additional rendered images for one or more selected sites on one or more other wafers generated by performing said generating for the one or more selected sites; and one or more optical images generated by the optical inspection subsystem for the one or more selected sites on the one or more other wafers.

20 . The system of claim 19 , wherein generating the rendered image further comprises modifying an initial rendered image to minimize differences between the initial rendered image and the optical image generated by the optical inspection subsystem, wherein the initial rendered image is generated offline, and wherein generating the rendered image is performed online.

21 . The system of claim 20 , wherein the initial rendered image is generated for a whole die in the design printed on the wafer, and wherein generating the rendered image is further performed online for an entirety' of the wafer.

22 . The system of claim 1 , wherein said generating is performed for only one or more areas in the design printed on the wafer such that said generating is not performed for an entirety of the design.

23 . The system of claim 1 , wherein the one or more computer subsystems are further configured for training the generative model based on: two or more additional rendered images for two or more areas on one or more other wafers generated by performing said generating for the two or more areas; and two or more optical images generated by the optical inspection subsystem for the two or more areas on the one or more other wafers, and wherein said training performed for a first of the two or more areas is performed differently than said training performed for a second of the two or more areas.

24 . The system of claim 23 , wherein generating the rendered image further comprises modifying an initial rendered image to minimize differences between the initial rendered image and the optical image generated by the optical inspection subsystem, wherein the initial rendered image is generated offline, and wherein generating the rendered image is performed o

Assignees

• Kla Tencor Corp

Inventors

• Gao Lisheng
• Luo Tao
• Wells Keith
• Li Xiaochun

Classifications

• G01N2201/12

  Circuits of general importance; Signal processing · CPC title

• G01N21/95607 Primary

  using a comparative method · CPC title

• G06T2207/30148

  Semiconductor; IC; Wafer · CPC title

• G06T7/001

  using an image reference approach · CPC title

• G06T2207/20084

  Artificial neural networks [ANN] · CPC title