Method, electronic device and operating method of electronic device and manufacture of semiconductor device
Disclosed is an operating method of an electronic device for manufacture of a semiconductor device. The method includes receiving, at the electronic device, a computer-aided design (CAD) image for a lithography process of the semiconductor device, and generating, at the electronic device, a first scanning electron microscope (SEM) image and a first segment (SEG) image from the CAD image by using a machine learning-based module, and the first SEG image includes information about a location of a defect.
1. An operating method of an electronic device for manufacture of a semiconductor device, the method comprising:
receiving, at the electronic device, a computer-aided design (CAD) image for a lithography process of the semiconductor device;
generating, at the electronic device, a first scanning electron microscope (SEM) image and a first segment (SEG) image, the first SEM image and the first SEG image generated from the CAD image by using a machine learning-based circuit,
wherein the first SEG image includes information about a location of a defect;
calculating, at the electronic device, a SEM loss based on the first SEM image and a SEG loss based on the first SEG image;
calculating, at the electronic device, a cost based on the SEM loss and the SEG loss; and
updating, at the electronic device, the machine learning-based circuit based on the cost.
2. The method of claim 1 , further comprising:
receiving, at the electronic device, a second SEM image and a second SEG image that correspond to the CAD image.
3. The method of claim 2 , wherein the second SEM image and the second SEG image are obtained from a semiconductor device at least partially manufactured by using the CAD image.
4. The method of claim 2 , further comprising:
identifying, at the electronic device, one of a set of the CAD image, the first SEM image, and the first SEG image and a set of the CAD image, the second SEM image, and the second SEG image with a “true” label.
5. The method of claim 4 , further comprising:
performing, at the electronic device, machine learning based on a result of the identifying.
6. The method of claim 2 , further comprising:
calculating, at the electronic device, a generative adversarial network (GAN) loss;
wherein calculating, at the electronic device, the SEM loss is further based on the second SEM image; and
wherein calculating, at the electronic device, the SEG loss is further based on the second SEG image.
7. The method of claim 6 , further comprising:
calculating, at the electronic device, a summed loss based on a sum of the GAN loss, the SEM loss, and the SEG loss;
wherein calculating, at the electronic device, the cost is based on the summed loss.
8. The method of claim 2 , further comprising:
blending the CAD image, the first SEM image, and the first SEG image; and
blending the CAD image, the second SEM image, and the second SEG image.
9. The method of claim 2 , further comprising:
generating a first set of three channels with the CAD image, the first SEM image, and the first SEG image; and
generating a second set of three channels with the CAD image, the second SEM image, and the second SEG image.
10. The method of claim 2 , further comprising:
generating a first cascade image with the CAD image, the first SEM image, and the first SEG image; and
generating a second cascade image with the CAD image, the second SEM image, and the second SEG image.
11. The method of claim 1 , wherein the machine learning-based circuit is based on one or both of a conditional GAN or a pixe2pixel.
12. The method of claim 1 , wherein the first SEG image indicates the location of the defect in one or more of the shape of Gaussian blurring, pixels corresponding to the defect, a circle, or a quadrangle.
13. An electronic device comprising:
a memory; and
a processor configured to drive a machine learning-based circuit by using the memory,
wherein, in response to executing the machine learning-based circuit, the processor is configured to,
receive a computer-aided design (CAD) image for a lithography process of a semiconductor device,
generate a first scanning electron microscope (SEM) image and a first segment (SEG) image, the first SEM image and the first SEG image generated from the CAD image by using the machine learning-based circuit, wherein the first SEG image includes information about a location of a defect;
calculate a SEM loss based on the first SEM image and a SEG loss based on the first SEG image;
calculate a cost based on the SEM loss and the SEG loss; and
update the machine learning-based circuit based on the cost.
14. The electronic device of claim 13 , wherein the processor is configured to further receive a second SEM image and a second SEG image that correspond to the CAD image, and
the second SEM image and the second SEG image are obtained from a semiconductor device at least partially manufactured by using the CAD image.
15. The electronic device of claim 14 , wherein the processor is configured to identify one of a set of the CAD image, the first SEM image, and the first SEG image and a set of the CAD image, the second SEM image, and the second SEG image as a “true” label.
16. The electronic device of claim 14 , wherein the processor is configured to calculate a GAN loss, wherein calculating the SEM loss is further based on the second SEM image, and wherein calculating SEG loss is further based on the second SEG image.
17. The electronic device of claim 16 , wherein the processor is configured to update the machine learning-based circuit based on the GAN loss, the SEM loss, and the SEG loss.
18. A method of manufacturing a semiconductor device by using at least one electronic device, the method comprising:
receiving, at the at least one electronic device, a computer-aided design (CAD) image for a lithography process of the semiconductor device;
generating, at the at least one electronic device, a scanning-electron microscope (SEM) image and a segment SEG image from the CAD image, the SEM image and the SEG image generated by using a machine learning-based circuit;
performing, at the at least one electronic device, a follow-up operation based on the SEM image and the SEG image, wherein the SEG image includes information about a location of a defect;
calculating, at the at least one electronic device, a SEM loss based on the SEM image and a SEG loss based on the SEG image;
calculating, at the at least one electronic device, a cost based on the SEM loss and the SEG loss; and
updating, at the at least one electronic device, the machine learning-based circuit based on the cost.
19. The method of claim 18 , wherein the performing of the follow-up operation includes:
revising the CAD image based on the SEM image and the SEG image.
20. The method of claim 18 , wherein the performing of the follow-up operation includes:
at least partially manufacturing the semiconductor device based on the CAD image; and
inspecting a defect of the semiconductor device based on the SEM image and the SEG image.