A method for detecting backside defects

Abstract

The present invention discloses a method for detecting crystal back defects, comprising obtaining a standard crystal back image and a crystal back image to be detected after each layer process is completed, and dividing the two into blocks to form a standard block and a block to be detected; on the basis of judging the block to be detected based on a conventional detection method, the normal block judged by the conventional detection method is compared with the corresponding standard block again, and finally a judgment result of whether the block to be detected is normal is formed, thereby improving the accuracy of detection, and the defect schematic diagram formed is more consistent with the crystal back image to be detected, which is more beneficial to the subsequent statistical work of crystal back defects.

Description

Technical Field

[0001] The invention relates to semiconductor manufacturing technology, and in particular to a method for detecting crystal back defects. Background Art

[0002] In the semiconductor chip manufacturing process, with the downward trend of technology nodes, the molding requirements of devices are becoming more and more stringent, and the defects on the back of the wafer (hereinafter referred to as back defects) have an increasingly serious impact on the front graphics of the wafer. Therefore, the detection requirements for back defects are becoming more and more stringent, and the number of back defects is becoming more and more important. Back defects refer to defects formed during the contact between the back of the wafer and the process machine during wafer processing. At present, the conventional detection method for wafer defects is based on the two-dimensional grayscale image obtained by the inspection machine scanning the back of the wafer to be inspected, and the two-dimensional grayscale image is converted into a defect schematic diagram with the same shape as the back of the wafer to be inspected, and in the converted defect schematic diagram, two different colors are used to represent the normal area and the defect area respectively. When the conventional detection method is used for detection, the normal area and the defect area are formed based on dividing the wafer image to be inspected into multiple blocks of equal size and judging whether the divided blocks are normal or not. However, the conventional detection method will lead to misjudgment of larger defects. The reasons for the misjudgment are as follows: Figure 1 As shown, Figure 1 In the figure, the white area is the normal area determined by the conventional detection method, the gray area is the abnormal area determined by the conventional detection method, the area marked as A is the actual normal area, and the area marked as B is the actual abnormal area, that is, when determining a defect with a large area including multiple blocks, the conventional detection method makes a misjudgment and determines the middle area of ​​the defect with a large area as a normal block; Figure 2a shows a two-dimensional grayscale backside image of the wafer to be inspected. Figure 2b The defect diagram is generated by processing the two-dimensional grayscale back image of the back of the wafer to be inspected using a conventional inspection method. Figure 2a In the figure, A is the actual normal area and B is the actual abnormal area, but in Figure 2b In the above example, the actual abnormal area B is judged as a normal area, that is, a misjudgment occurs; this misjudgment is very unfavorable for the statistical work of crystal back defects, for example, the number of crystal back defects cannot be accurately counted. Summary of the invention

[0003] The technical problem to be solved by the present invention is to provide a detection method for crystal back defects, which can accurately form a crystal back pattern according to a two-dimensional grayscale crystal back image, improve the accuracy of crystal back defect statistics, and facilitate subsequent statistics of crystal back defects.

[0004] In order to solve the above technical problems, the present invention provides a method for detecting wafer back defects, comprising the following steps:

[0005] Step S01, obtaining a standard wafer back image after each layer process is completed, wherein the standard wafer back image is a two-dimensional grayscale image;

[0006] Step S02, obtaining a wafer back image to be tested after each process is completed, wherein the wafer back image to be tested is a two-dimensional grayscale image, and the wafer back image to be tested of each process layer corresponds to the standard wafer back image;

[0007] Step S03, performing a block processing on the wafer back image to be detected of each process layer obtained in step S02, dividing the wafer back image to be detected into a plurality of equal-sized blocks to be detected, wherein the block to be detected includes at least one pixel area;

[0008] Step S04, using the same partitioning and blocking processing method as in step S03, the standard wafer back image of each process layer obtained in step S01 is partitioned and blocked to obtain standard blocks, wherein the standard blocks correspond to the blocks to be detected;

[0009] Step S05, using a conventional detection method to detect each of the blocks to be detected obtained in step S03. When a block to be detected is detected as abnormal, the block to be detected is directly determined to be an abnormal block, and the detection result is output; when a block to be detected is detected as normal, a second determination is performed: the grayscale value of the block to be detected is compared with the grayscale value of the corresponding standard block. When the grayscale difference between the two is less than the set grayscale threshold, the block to be detected is finally determined to be a normal block, and the detection result is output; when the grayscale difference between the two is greater than or equal to the set grayscale threshold, the block to be detected is finally determined to be an abnormal block, and the detection result is output.

[0010] Preferably, the detection method further comprises the following steps:

[0011] Step S06, counting the number of defects in the wafer back image to be detected according to the detection result output in step S05.

[0012] Preferably, the number of defects in the wafer back image to be detected is counted based on the block to be detected that is determined to be an abnormal block:

[0013] When all other blocks to be inspected adjacent to the block determined to be abnormal are determined to be normal blocks, the abnormal block is counted as a defect;

[0014] When an abnormal block appears in other blocks to be detected adjacent to the block determined to be abnormal, all adjacent abnormal blocks are combined and counted as one defect.

[0015] Preferably, the test result outputted in step S05 eventually forms a defect schematic diagram having the same shape as the back side of the wafer to be tested, and the normal detection result and the abnormal detection result have different colors in the defect schematic diagram.

[0016] Preferably, the color output of the normal inspection result is white, and the color output of the abnormal detection result is red.

[0017] Preferably, the size of the block to be detected in step S03 is a pixel range; the size of the standard block in step S04 is a pixel range.

[0018] Preferably, the standard wafer back image is obtained through historical scanning product records of a wafer back inspection machine.

[0019] Preferably, the wafer back image to be inspected is obtained by scanning the wafer to be inspected by a wafer back inspection machine.

[0020] The present invention improves the accuracy of detection by performing a secondary comparative detection on the normal block determined by the conventional detection method and the standard block on the basis of the conventional detection method. The defect schematic diagram formed is more consistent with the crystal back image to be detected, which is more beneficial to the subsequent statistical work of crystal back defects. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solution of the present invention, the following briefly introduces the drawings required for use in the present invention. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0022] Figure 1 It is a case of misjudgment caused by conventional detection methods;

[0023] Figure 2a It is a two-dimensional grayscale image of the backside of the wafer to be inspected;

[0024] Figure 2b The conventional detection method is based on Figure 2a A schematic diagram of defects formed by a two-dimensional back image of the wafer to be inspected;

[0025] In the figure, A is the actual normal area, and B is the actual abnormal area. DETAILED DESCRIPTION

[0026] The following will be combined with the accompanying drawings to clearly and completely describe the technical solutions in the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0027] Figure 1 A flow chart of a method for detecting backside defects according to an embodiment of the present invention is shown, comprising the following steps:

[0028] Step S01, obtaining a standard wafer back image after each layer process is completed, wherein the standard wafer back image is a two-dimensional grayscale image;

[0029] Step S02, obtaining a wafer back image to be tested after each process is completed, wherein the wafer back image to be tested is a two-dimensional grayscale image, and the wafer back image to be tested of each process layer corresponds to the standard wafer back image;

[0030] Step S03, performing a block processing on the wafer back image to be detected of each process layer obtained in step S02, dividing the wafer back image to be detected into a plurality of equal-sized blocks to be detected, wherein the block to be detected includes at least one pixel area;

[0031] Step S04, using the same partitioning and blocking processing method as in step S03, the standard wafer back image of each process layer obtained in step S01 is partitioned and blocked to obtain standard blocks, wherein the standard blocks correspond to the blocks to be detected;

[0032] Step S05, using a conventional detection method to detect each of the blocks to be detected obtained in step S03, when a block to be detected is detected as abnormal, directly determine the block to be detected as an abnormal block, and output the detection result; when a block to be detected is detected as normal, perform a second determination: compare the gray value of the block to be detected with the gray value of the corresponding standard block, when the gray difference between the two is less than the set gray threshold, finally determine the block to be detected as a normal block, and output the detection result; when the gray difference between the two is greater than or equal to the set gray threshold, finally determine the block to be detected as an abnormal block, and output the detection result. Preferably, the test result output in step S05 eventually forms a defect schematic diagram with the same shape as the back of the wafer to be tested, and the normal detection result and the abnormal detection result have different colors in the defect schematic diagram. More preferably, the color output of the normal inspection result is white, and the abnormal detection result is red.

[0033] In an embodiment of the present invention, a standard crystal back image is first obtained: in the process of detecting the crystal back, it is found that after the preparation of each process layer of the wafer is completed, the grayscale of the two-dimensional grayscale images of the crystal back of different process layers obtained when detecting the crystal back is different, but for different wafers, the grayscale values ​​of the two-dimensional grayscale images of the same process layer are basically the same. Therefore, based on the two-dimensional grayscale image records of the historical crystal backs of each layer that have been detected, a standard crystal back image is produced, and when editing the detection program, the produced standard crystal back image is edited into the detection program for standby use.

[0034] In the embodiment of the present invention, a re-determination of the normal block determined by the conventional detection method is added, and the basis of the determination is the gray value difference between the block to be detected and the standard block. When the back image to be detected and the standard back image are processed in blocks, the same processing method is adopted. After processing, each block to be detected in the back image to be detected corresponds to each standard block in the standard back image, that is, the size and relative position are the same to ensure their comparability. The re-determination of the normal block determined by the conventional detection method is added. For the normal detection block determined by the conventional detection method, after comparing it with the standard block again, if the gray value difference is greater than the set gray threshold, it will eventually be determined as abnormal. Therefore, the misjudgment generated when the conventional detection method is used for determination is effectively avoided. The shape, size and number of defects in the defect schematic diagram finally output in the embodiment of the present invention are consistent with the back image to be detected, which is more convenient for subsequent related statistics of back defects, such as quantity statistics. Regarding the setting of grayscale threshold, the following explanation is also required: the grayscale value of each process layer is different, therefore, the grayscale threshold of each layer is different, but in the production process, some standardized algorithms are usually used to determine the standard threshold of each layer. For example, 3σ of the grayscale value of the standard block is used as the standard threshold. If the difference is less than 3σ, it is judged that the block to be tested is normal, and if the difference is greater than 3σ, it is judged that the block to be tested is abnormal. Based on the detection method and the actual production situation, other standards can also be used to define the grayscale threshold, which are not listed here one by one.

[0035] In a specific embodiment, the detection method further comprises the following steps:

[0036] Step S06, according to the detection result output in step S05, the number of defects in the backside image to be detected is counted. The number of defects in the backside image to be detected is counted based on the blocks to be detected that are determined to be abnormal blocks:

[0037] When all other blocks to be inspected adjacent to the block determined to be abnormal are determined to be normal blocks, the abnormal block is counted as a defect;

[0038] When an abnormal block appears in other blocks to be detected adjacent to the block determined to be abnormal, all adjacent abnormal blocks are combined and counted as one defect.

[0039] This specific implementation is an application of the back-of-crystal defect detection method of the present invention, specifically the statistics of the number of defects. When the defect diagram produced by the conventional detection method is counted, due to the existence of misjudgment, the statistics of the number of defects are also inaccurate; but when the back-of-crystal defect detection method provided by the present invention is adopted, since the defect diagram formed is consistent with the back-of-crystal image to be detected, there is no misjudgment, so the defect number statistics are also very accurate. It should be noted that when judging a single defect, the adjacent principle is generally adopted, that is, all adjacent abnormal blocks judged as abnormal blocks are identified as one defect; but it is also possible to set abnormal blocks separated by several blocks to be identified as one defect.

[0040] In a specific implementation, the size of the block to be detected in step S03 is a pixel range; the size of the standard block in step S04 is a pixel range. It can be understood that using a single pixel as a block division can improve the accuracy of the detection result.

[0041] In a specific embodiment, the standard back image is obtained by scanning the product history of the back inspection machine. Understandably, since the grayscale values ​​of the two-dimensional grayscale images of the same process layer of different crystals are basically the same, although there are back defects in the two-dimensional grayscale image of a single back image, the standard back images of each layer without defects can be obtained based on the two-dimensional grayscale images of many historical back images.

[0042] In a specific implementation manner, the wafer back image to be inspected is obtained by scanning the wafer to be inspected by a wafer back inspection machine.

[0043] In summary, the present invention improves the accuracy of detection by performing a secondary comparative detection on the normal blocks determined by the conventional detection method and the standard blocks on the basis of the conventional detection method. The defect schematic diagram formed is more consistent with the crystal back image to be detected, which is more beneficial to the subsequent statistical work of crystal back defects.

[0044] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for detecting crystal back defects, characterized in that: The steps include: Step S01, obtaining a standard wafer back image after each layer process is completed, wherein the standard wafer back image is a two-dimensional grayscale image; Step S02, obtaining a back image of the wafer to be detected after each process is completed, wherein the back image of the wafer to be detected is a two-dimensional grayscale image, and the back image of the wafer to be detected of each process layer corresponds to the standard back image of the wafer; Step S03, performing a block processing on the wafer back image to be detected of each process layer obtained in step S02, dividing the wafer back image to be detected into a plurality of equal-sized blocks to be detected, wherein the block to be detected includes at least one pixel area; Step S04, using the same partitioning and blocking processing method as in step S03, the standard wafer back image of each process layer obtained in step S01 is partitioned and blocked to obtain standard blocks, wherein the standard blocks correspond to the blocks to be detected; Step S05, using a conventional detection method to detect each of the blocks to be detected obtained in step S03. When a block to be detected is detected as abnormal, the block to be detected is directly determined to be an abnormal block, and the detection result is output; when a block to be detected is detected as normal, a second determination is performed: the grayscale value of the block to be detected is compared with the grayscale value of the corresponding standard block. When the grayscale difference between the two is less than the set grayscale threshold, the block to be detected is finally determined to be a normal block, and the detection result is output; when the grayscale difference between the two is greater than or equal to the set grayscale threshold, the block to be detected is finally determined to be an abnormal block, and the detection result is output.

2. The detection method according to claim 1, characterized in that The detection method further comprises the following steps: Step S06, counting the number of defects in the wafer back image to be detected according to the detection result output in step S05.

3. The detection method according to claim 2, characterized in that The number of defects in the wafer back image to be detected is counted based on the blocks to be detected that are determined to be abnormal blocks: When all other blocks to be inspected adjacent to the block determined to be abnormal are determined to be normal blocks, the abnormal block is counted as a defect; When an abnormal block appears in other blocks to be detected adjacent to the block determined to be abnormal, all adjacent abnormal blocks are combined and counted as one defect.

4. The detection method according to claim 1, characterized in that The test result outputted in step S05 eventually forms a defect schematic diagram having the same shape as the back side of the wafer to be tested, and the normal detection result and the abnormal detection result have different colors in the defect schematic diagram.

5. The detection method according to claim 4, characterized in that: The color output of the normal detection result is white, and the color output of the abnormal detection result is red.

6. The detection method according to claim 1, characterized in that The size of the block to be detected in step S03 is a pixel range; the size of the standard block in step S04 is a pixel range.

7. The detection method according to claim 1, characterized in that The standard wafer back image is obtained through the historical scanning product history of the wafer back inspection machine.

8. The detection method according to claim 1, characterized in that The wafer back image to be inspected is obtained by scanning the wafer to be inspected by a wafer back inspection machine.

Images