Using deep learning based defect detection and classification schemes for pixel level image quantification
A heat map of probable defects in an image can be represented as a matrix of defect probability index corresponding to each pixel. The image may be generated from data received from a detector of a scanning electron microscope or other inspection tools. A number of pixels in the image that exceed a corresponding threshold in the matrix can be quantified.
1. A system comprising:
an electron beam source that generates an electron beam;
a stage configured to hold a wafer in a path of the electron beam;
a detector configured to receive the electron beam returned from the wafer;
a processor in electronic communication with the detector; and
a deep learning module operated by the processor, wherein the deep learning module is configured to:
receive an image generated from data received from the detector: and
perform defect detection and defect classification on the image thereby determining probable defects:
wherein in the processor is configured to:
represent a heat map of the probable defects from the deep learning module in the image as a matrix of defect probability index corresponding to each pixel of the image, wherein the heat map includes at least some of the wafer and is based on measurements of features in the image; and
quantify a number of pixels in the image that exceed a corresponding threshold in the matrix.
2. The system of claim 1 , wherein the processor is further configured to determine the heat map.
3. The system of claim 1 , wherein the corresponding threshold for one of the pixels is at a same location on the image as the one of the pixels.
4. The system of claim 1 , wherein the quantifying is used in pixel level image quantification.
5. The system of claim 1 , wherein the probable defects are EUV stochastics.
6. The system of claim 1 , wherein the probable defects are critical dimension defects.
7. The system of claim 1 , wherein the image is a scanning electron microscope image.
8. A method comprising:
receiving an image at a processor:
performing defect detection and defect classification on the image using a deep learning module of the processor thereby determining probable defects:
representing a heat map of the probable defects from the deep learning module in the image as a matrix of defect probability index corresponding to each pixel using a processor, wherein the image is generated from data received from a detector, and wherein the heat map includes at least some of the wafer and is based on measurements of features in the image; and
quantifying, using the processor, a number of pixels in the image that exceed a corresponding threshold in the matrix.
9. The method of claim 8 , further comprising determining the heat map using the processor.
10. The method of claim 8 , wherein the corresponding threshold for one of the pixels is at a same location on the image as the one of the pixels.
11. The method of claim 8 , wherein the quantifying is used in pixel level image quantification.
12. The method of claim 8 , wherein the probable defects are EUV stochastics.
13. The method of claim 8 , wherein the probable defects are critical dimension defects.
14. The method of claim 8 , wherein the image is a scanning electron microscope image.
15. The method of claim 8 , further comprising:
directing an electron beam at the wafer;
collecting electrons returned from the wafer with a detector; and
generating, using the processor, the image of the wafer.
16. A non-transitory computer readable medium storing a program configured to instruct a processor to execute the method of claim 8 .
17. The system of claim 1 , wherein the heat map displays mean diameters.
18. The method of claim 8 , wherein the heat map displays mean diameters.