System and method for conducting adaptive fourier filtering to detect defects in dense logic areas of an inspection surface

• US 20070081154A1
• Filed: 12/20/2005
• Published: 04/12/2007
• Est. Priority Date: 09/27/2002
• Status: Active Grant

• Current Assignee: Kla-Tencor Technologies Corporation (KLA Corporation)
• Original Assignee: Kla-Tencor Technologies Corporation (KLA Corporation)
• Inventors: Shortt, David, Chen, Grace, Mapoles, Evan, Bevis, Christopher
• Granted Patent: US 7,505,619 B2
• Time in Patent Office: Days
• US Class Current: 356/237.500
• CPC Class Codes: G01N 2021/8822 Dark field detection
• CPC Class Codes: G01N 21/47 Scattering, i.e. diffuse reflection G01N21/25, G01N21/41 take precedence G01N21/55 takes precedence
• CPC Class Codes: G01N 21/95607 using a comparative method
• CPC Class Codes: G01N 21/95623 using a spatial filtering method per se G02B

Abstract

A dark field surface inspection tool and system are disclosed herein. The tool includes an illumination source capable of scanning a light beam onto an inspection surface. Light scattered by each inspection point is captured as image data by a photo detector array arranged at a fourier plane. The images captured are adaptively filtered to remove a portion of the bright pixels from the images to generate filtered images. The filtered images are then analyzed to detect defects in the inspection surface. Methods of the invention include using die-to-die comparison to identify bright portions of scattering patterns and generate unique image filters associated with those patterns. The associated images are then filtered to generate filtered images which are then used to detect defects. Also, data models of light scattering behavior can be used to generate filters.

Claims

1. A dark field surface inspection tool comprising:

an illumination system configured to scan a light beam across an inspection surface, thereby generating light scattering patterns associated with selected portions of the inspection surface;

a photodetector array arranged at a fourier plane to receive the light scattering patterns from the work piece, the array comprising a two-dimensional array of light sensitive pixels configured to capture two-dimensional images of the light scattering patterns; and

circuitry for receiving the images and conducting adaptive fourier filtering of image data to selectively filter image pixels to enhance signal to noise ratio in the remaining image pixels.

2. The dark field surface inspection tool of claim 1 wherein the circuitry enables the identification of defects at the selected portions of the inspection surface.

3. The dark field surface inspection tool of claim 1 wherein the inspection surface has a plurality of substantially similar semiconductor dies formed thereon;

wherein the light beam illuminates the same selected portion of each of the plurality semiconductor dies to generate light scattering pattern associated with the selected portion of the inspection surface;

wherein the photodetector array captures two-dimensional images of the light scattering patterns of the same selected portion of the plurality semiconductor dies with each image corresponding to an equivalent portion on a different die;

wherein the circuitry includes,

circuitry configured to process the images to generate a filter template capable of selectively filtering the image pixels of each image to enhance signal to noise ratio of each filtered image;

circuitry configured to process each image with an associated template to obtain the filtered image for the associated selected portion of the inspection surface; and

circuitry configured to process the filtered image to detect defects for the associated selected portion of the inspection surface.

4. The dark field surface inspection tool of claim 1 wherein the photodetector array a two-dimensional array of light sensitive pixels comprising 256 photosensitive elements.

5. The dark field surface inspection tool of claim 4 wherein the photosensitive elements of the photodetector array are positioned such that the array subtends a solid angle of greater than about 40°

×

40°

.

6. The dark field surface inspection tool of claim 3 wherein the light scattering pattern generated by the light beam forms a pattern of bright regions and dark regions associated with the selected portion of a die;

wherein the photodetector array captures two-dimensional images of the pattern of bright regions and dark regions associated with the selected portion of the die;

wherein the circuitry configured to process the images to generate a filter template are configured to define a set of bright pixels for each selected portion of the inspection surface and generate a filter template associated with the set of bright pixels; and

wherein the circuitry configured to process the image with an associated template to obtain the filtered image is configured to selectively remove the pixels associated with the set of bright pixels from the image to generate the filtered image.

7. A method of adaptively filtering of image data, the method comprising:

illuminating selected portions of an inspection surface with a light beam, thereby generating a light scattering pattern associated with the selected portions of the inspection surface;

capturing image data at a fourier plane wherein the image data is associated with the light scattering pattern for each of the selected portions of said surface, wherein the image data comprises a two-dimensional image comprised of an array of image pixels; and

adaptively filtering the image data to produce an adaptively filtered image having selected image pixels removed from the adaptively filtered image.

8. The method of claim 7, wherein the inspection surface comprises a surface having a plurality of substantially similar semiconductor die patterns formed thereon;

and

adaptively filtering the image pixels includes,

conducting a die to die comparison of a same portion of each die pattern to identify a set of pixels suitable for adaptive filtering; and

generating a unique adaptive filter template for each inspected portion of the die pattern wherein the filter template enables the selected image pixels to be filtered out from image data.

9. The method of claim 8, wherein the die to die comparison comprises:

obtaining a set of images for the portion of each die pattern;

comparing the pixels of each set of images to identify a pattern of bright and dark pixels wherein the bright pixels are identified as the set of pixels suitable for filtering; and

wherein generating the unique adaptive filter template for each portion of the die pattern comprises generating a filter configured to filter out data from said bright pixels.

10. The method of claim 9, wherein the bright pixels are defined as pixels in a set of images associated with a portion of a die having a brightness above a predetermined brightness level.

11. The method of claim 9, wherein the bright pixels are defined as the brightest percentage of pixels in an image associated with a portion of a die wherein the percentage comprises a predetermined brightness threshold percentage.

12. The method of claim 7, wherein illuminating selected portions of an inspection surface comprises illuminating a die pattern wherein the selected portions comprise random logic features.

13. The method of claim 7, wherein adaptively filtering the image pixels includes,

analyzing a data model associated with the inspection surface to identify, for each inspected portion of the surface, a set of pixels suitable for adaptive filtering;

and

incorporating each set of pixels into a unique adaptive filter template, wherein each filter template is suitable for filtering data associated with its associated inspected portion of the surface.

14. The method of claim 7, wherein the inspection surface comprises a surface having a plurality of substantially similar semiconductor die patterns formed thereon;

and

adaptively filtering the image pixels includes, using a data model of the die pattern to identify a set of pixels suitable for adaptive filtering; and

generating a unique adaptive filter template for each inspected portion of the die pattern wherein the filter template enables the selected image pixels to be filtered out from image data.

15. The method of claim 14, wherein using the data model of the die pattern to identify a set of pixels suitable for adaptive filtering includes analyzing the light scattering pattern predicted by the model to identify a pattern of bright and dark pixels wherein the bright pixels are identified as the set of pixels suitable for filtering;

and

wherein generating the unique adaptive filter template comprises generating a filter configured to filter out data from said bright pixels.

16. The method of claim 15, wherein the bright pixels are defined as pixels in a set of images associated with a portion of a die having a brightness above a predetermined brightness level.

17. The method of claim 9, wherein the bright pixels are defined as the brightest percentage of pixels in an image associated with a portion of a die wherein the percentage comprises a predetermined brightness threshold percentage.

18. The method of claim 15, wherein the bright pixels are defined as a group of pixels having a cumulative energy level above a predetermined energy threshold percentage.

19. The method of claim 9, wherein the bright pixels are defined as a group of pixels having a cumulative energy level above a predetermined energy threshold percentage.

20. The method of claim 7, wherein adaptively filtering the image data includes,

generating an image pattern of bright and dark pixels associated with each inspection point on the inspection surface;

selecting a set of bright pixels for each image pattern to identify a set of filtering pixels suitable for adaptive filtering; and

using the set of filtering pixels to generate a unique adaptive filter template for each image pattern, the filtering pixels enabling selected image data to be filtered out from image data.

21. The method of claim 20, wherein the bright pixels are defined as pixels having a brightness above a predetermined brightness level.

22. The method of claim 20, wherein the bright pixels are defined as the brightest percentage of pixels in each image associated with a particular inspection point.

23. A method of adaptively filtering of image data, the method comprising:

illuminating selected portions of an inspection surface with a light beam, thereby generating a light scattering pattern associated with the selected portions of the inspection surface;

capturing image data at a fourier plane wherein the image data is associated with the light scattering pattern for each of the selected portions of said surface, wherein the image data comprises a two-dimensional image comprised of an array of image pixels;

analyzing the image data to produce an adaptive filter template for each selected portion of said surface enabling the selective removal of particular image pixels from an image associated with the selected portion of said surface;

applying the template to the image data captured at the fourier plane thereby selectively excluding said particular image pixels to obtain a filtered image; and

analyzing the filtered image to identify defects in the inspection surface.

24. The method of claim 23 wherein

the two-dimensional array of image pixels captured at a fourier plane comprises a pattern of bright and dark pixels;

wherein analyzing the image data to produce an adaptive filter template for each selected portion of said surface comprises,

defining which of the image pixels are bright and identifying the position of the bright pixels in the array of image pixels,

defining an adaptive filter template for each selected portion of the inspection surface, the template configured to selectively remove the bright pixels from the image while still enabling data associated with the dark pixels to be evaluated; and

wherein applying the template to the image data captured at the fourier plane comprises selectively excluding the bright pixels to obtain a filtered image.

25. The method of claim 23, wherein the inspection surface comprises a surface having a plurality of substantially similar semiconductor die patterns formed thereon;

and

wherein analyzing the image data to produce an adaptive filter template for each selected portion of said surface enabling the selective removal of particular image pixels from an image associated with the selected portion of said surface comprises,

conducting a die to die comparison of a same portion of each die pattern to identify a set of pixels suitable for adaptive filtering; and

generating a unique adaptive filter template for each inspected portion of the die pattern wherein the filter template enables the selected image pixels to be filtered out from image data.

26. The method of claim 23, wherein illuminating selected portions of an inspection surface comprises illuminating a die pattern wherein the selected portions comprise random logic features.

27. A computer program product embodied on a computer readable media including computer program code for accomplishing adaptively filtering of image data, program product including:

computer program code instructions for illuminating selected portions of an inspection surface with a light beam, thereby generating a light scattering pattern associated with the selected portions of the inspection surface;

computer program code instructions for capturing image data at a fourier plane wherein the image data is associated with the light scattering pattern for each of the selected portions of said surface, wherein the image data comprises a two-dimensional image comprised of an array of image pixels;

computer program code instructions for analyzing the image data to produce an adaptive filter template for each selected portion of said surface enabling the selective removal of particular image pixels from an image associated with the selected portion of said surface;

computer program code instructions for applying the template to the image data captured at the fourier plane thereby selectively excluding said particular image pixels to obtain a filtered image; and

computer program code instructions for analyzing the filtered image to identify defects in the inspection surface.

28. The computer program product of claim 27 wherein

the two-dimensional array of image pixels captured at a fourier plane comprises a pattern of bright and dark pixels;

wherein the computer program code instructions for analyzing the image data to produce an adaptive filter template for each selected portion of said surface comprises,

computer program code instructions for defining which of the image pixels are bright and identifying the position of the bright pixels in the array of image pixels,

computer program code instructions for defining an adaptive filter template for each selected portion of the inspection surface, the template configured to selectively remove the bright pixels from the image while still enabling data associated with the dark pixels to be evaluated; and

wherein the computer program code instructions for applying the template to the image data captured at the fourier plane comprises computer program code instructions for selectively excluding the bright pixels to obtain a filtered image.

29. The computer program product of claim 27 wherein the inspection surface comprises a surface having a plurality of substantially similar semiconductor die patterns formed thereon;

and

wherein the computer program code instructions for analyzing the image data to produce an adaptive filter template for each selected portion of said surface enabling the selective removal of particular image pixels from an image associated with the selected portion of said surface comprises,

computer program code instructions for conducting a die to die comparison of a same portion of each die pattern to identify a set of pixels suitable for adaptive filtering; and

computer program code instructions for generating a unique adaptive filter template for each inspected portion of the die pattern wherein the filter template enables the selected image pixels to be filtered out from image data.

30. The computer program product of claim 27 wherein the computer program code instructions for illuminating selected portions of an inspection surface includes illuminating a die pattern on the inspection surface wherein the selected portions comprise random logic features.