Method for Inspection and Detection of Defects on Surfaces of Disc-Shaped Objects and Computer System with a Software Product for Carrying out the Method

• US 20120163698A1
• Filed: 11/10/2011
• Published: 06/28/2012
• Est. Priority Date: 12/22/2010
• Status: Active Grant

• Current Assignee: KLA-Tencor MIE GmbH (KLA Corporation)
• Original Assignee: KLA-Tencor MIE GmbH (KLA Corporation)
• Inventors: Michelsson, Detlef
• Granted Patent: US 8,705,837 B2
• Time in Patent Office: Days
• US Class Current: 382/141
• CPC Class Codes: G01N 2021/9513 Liquid crystal panels
• CPC Class Codes: G01N 21/9501 Semiconductor wafers manufacturing processes per se of semiconductor devices implementing a measuring step H01L22/10
• CPC Class Codes: G01N 21/9506 Optical discs
• CPC Class Codes: G01N 21/956 Inspecting patterns on the surface of objects contactless testing of electronic circuits G01R31/308; testing currency G07D; manufacturing processes per se of semiconductor devices implementing a measuring step H01L22/10
• CPC Class Codes: G06F 17/153 Multidimensional correlation or convolution
• CPC Class Codes: G06T 2207/10024 Color image
• CPC Class Codes: G06T 2207/30121 CRT, LCD or plasma display
• CPC Class Codes: G06T 2207/30148 Semiconductor; IC; Wafer
• CPC Class Codes: G06T 7/001 using an image reference approach

Abstract

An image (30) of a disc-shaped object (100) is recorded, wherein the entire surface (100_O) is captured with a plurality of fields (60). A difference image (31) is formed, by subtracting a reference from each field (60) of the surface (100_O) of the disc-shaped object (100), and subject to a color transformation, wherein by a suitable choice of transformation signals in one channel are maximized, while at the same time undesired variations, caused by production, of the fields (60) are moved to a different channel. That combination of transformation and detection channel is chosen for which the largest number of pixels of a field (60) with the defect to be found are located outside the spread of the pixels of the respective field (60) on the surface (100_O) of the respective disc-shaped object (100) from production and provide the largest signals for the defect.

Claims

1. A method for inspecting and detecting defects on surfaces of disc-shaped objects, comprising the following steps:

capturing an image of the respective disc-shaped object, wherein the entire surface is captured with a plurality of fields;

deriving a defect-free reference of a pre-defined size from plural fields equal in size on the surface of the disc-shaped object;

generating a difference image of the surface of the disc-shaped object, wherein the derived reference is subtracted from each field on the surface of the disc-shaped object;

applying a color transformation to the difference image, wherein, by a suitable choice of transformation, signals in one channel are maximized, while undesired variations of the fields, caused by the production, are moved to a different channel;

evaluating the signals, wherein a mean value, a variance, a maximum value and a minimum value are determined for each field;

determining a distribution of the signals for each pixel of a field and a distribution of the signals for each defect to be examined, wherein the distribution is determined for each combination of transformation and detection channel used; and

selecting that combination of transformation and detection channel, for which the largest number of pixels of a field with the defect to be found lie outside the spread of the pixels of the respective field on the surface of the respective disc-shaped object from production, and which results in the largest signals for the defect.

2. The method of claim 1, wherein the fields of the surface of the disc-shaped object are chosen such that the fields of a structured surface of the disc-shaped object contain at least a rational fraction of a DIE.

3. The method of claim 1, wherein the fields of the surface of the disc-shaped object are chosen such that for an unstructured surface of the disc-shaped object all fields to be examined are distributed such that they cover the surface of the disc-shaped object and lie within a circumference of the disc-shaped object.

4. The method of claim 1, wherein, via a user interface, plural pixels are combined into a detection area, in order to minimize pixel noise, and wherein a choice between spatial resolution and noise suppression is made.

5. The method of claim 1, wherein for n×

m pixels differences of the pixels of the respective field to be examined and the corresponding pixels of the reference are formed, and wherein the differences are summed up separately for each color channel to form a difference image.

6. The method of claim 5, wherein the difference image formed is subject to a normalization to an illumination intensity, in order to achieve an independence from the illumination intensity.

7. The method of claim 6, wherein a constant value is added to each pixel of the difference image formed, in order to be able to represent also negative values of the difference image.

8. The method of claim 5, wherein the difference image is separated into individual channels, whence result three visual representations as greylevel images.

9. The method of claim 8, wherein a filter with a radial polynomial is superimposed on the difference image.

10. The method of claim 5, wherein a detection of a defect is achieved by applying a threshold to the difference image and classifying all positions in the difference image the values of which are larger than the threshold as defects.

12. The method of claim 1, wherein in case of a structured surface of a disc-shaped object each field comprises plural different regions, exhibiting different specificities of the defect.

13. The method of claim 12, wherein for each region within the field it is determined how strong the signals for the fields from production vary, and wherein a mean value and the variance for each field and in addition a minimum and maximum value are determined

14. The method of claim 13, wherein for the fields containing the defect the values of the regions are captured separately, and wherein the distribution of the signals for the fields and the regions is determined for each combination of transformation and detection channel.

15. The method of claim 12, wherein a detection of a defect is achieved by dividing the difference image of the respective field by the variance of the region in the respective field.

16. The method of claim 1, wherein in a learning phase based on a disc-shaped object with defined defects an evaluation of the signals of these defects and a determination of the optimal combination of transformation and channel are carried out, and wherein that transformation is chosen, for which the largest number of fields containing the defect are outside the spread of the field from production and provide the largest defect signals.

17. The method of claim 1, wherein in the learning phase an operator trains the detection of the system by trial or by experience, and determines the fields in which defects are to be expected.

18. The method of claim 1, wherein a morphological gradient is formed on the transformed difference image, so that changes of the signal are evaluated.

11. The method of claim 11, wherein the threshold is a function of the distance from a center of mass of the disc-shaped object.

19. A computer system with a software product, wherein the software product together with the computer system carries out an inspection and detection of defects on surfaces of disc-shaped objects, wherein the software product comprises the following steps:

capturing an image of the respective disc-shaped object with a camera of a system, wherein the entire surface of the disc-shaped object is captured with a plurality of fields, and wherein an image capturing window of the camera essentially corresponds to the size of the field;

forming a defect-free reference of a pre-defined size from plural fields, equal in size, of the surface of the disc-shaped object;

generating a difference image of the surface of the disc-shaped object, wherein the formed reference is subtracted from every field of the surface of the disc-shaped object;

applying a color transformation to the difference image, wherein by a suitable choice of the transformation signals in one channel are maximized, while at the same time undesired variations, caused by production, of the fields are moved to a different channel;

evaluating the signals, wherein a mean value, a variance, a maximum value and a minimum value are determined for each field;

determining a distribution of the signals for each pixel of a field and a distribution of the signals for each defect to be examined, wherein the distribution is determined for each combination of transformation used and detection channel used; and

selecting that combination of transformation and detection channel, for which the largest number of pixels of a field with the defect to be found are located outside the spread of the pixels of the respective field on the surface of the respective disc-shaped object from production, and which yields the largest signals for the defect.

20. The computer system of claim 19, wherein the fields of the surface of the disc-shaped object are chosen such that the fields of a structured surface of the disc-shaped object contain at least a rational fraction of a DIE.

21. The computer system according to claim 19, wherein, via a user interface of the system, plural pixels are combined into a detection area in order to minimize the pixel noise, and a choice between a spatial resolution and a noise suppression can be made.

22. The computer system according to claim 19, wherein for n×

m pixels the differences of the pixels of the respective field to be examined and the corresponding pixels of the reference are formed, and wherein the differences are summed up separately for each color channel.

23. The computer system according to claim 19, wherein in a learning phase, based on a disc-shaped object with defined defects, an evaluation of the signals of these defects and the determination of the optimal combination of transformation and channel are carried out, and wherein that transformation is chosen, for which the largest number of fields containing the defect are located outside the spread of the field from production, and which yields the largest defect signals.