Debris determination method

An image processing method for detecting debris around hard laser marks on wafers accurately identifies and excludes dot portions, addressing the failure of conventional techniques to reliably detect debris, thereby preventing device manufacturing defects.

JP7584471B2Active Publication Date: 2024-11-15SHIN ETSU HANDOTAI CO LTD +1
View PDF 7 Cites 0 Cited by

Patent Information

Application Number
JP2022097355
Authority / Receiving Office
JP · JP
Patent Type
Patents
Current Assignee / Owner
Priority Date
2021-07-13
Filing Date
2022-06-16
Publication Date
2024-11-15
Estimated Expiration
2042-06-16

AI Technical Summary

Technical Problem

Conventional methods fail to reliably detect debris around hard laser marks on wafer back surfaces, which can interfere with device manufacturing processes, and existing image processing techniques are inadequate for identifying gentle protrusions.

Method used

A method involving image processing to calculate reference brightness, extract the hard laser mark printing area, exclude dot portions, and determine debris areas based on pixel brightness thresholds, allowing for accurate detection of debris that conventional shape measuring machines miss.

Benefits of technology

The method reliably detects debris that conventional methods cannot, reducing inspection variations due to equipment differences and preventing device manufacturing defects.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 0007584471000001
    Figure 0007584471000001
  • Figure 0007584471000002
    Figure 0007584471000002
  • Figure 0007584471000003
    Figure 0007584471000003
Patent Text Reader

Abstract

To provide a debris determination method capable of reliably detecting debris that cannot be detected by a shape measuring instrument and determines whether there is debris.SOLUTION: A debris determination method is a method for determining debris generated around a hard laser mark (HLM) on the backside of a wafer using an image obtained by an appearance inspection device, and includes the steps of: calculating reference luminance from a grayscale image obtained by the appearance inspection device; extracting a printing area including the HLM from the grayscale image; excluding a dot portion of the HLM from the printing area; extracting a debris area from the printing area excluding the dot portion of the HLM on the basis of the reference luminance; and determining whether there is debris in the printing area based on the debris area.SELECTED DRAWING: Figure 1
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] The present invention relates to a method for determining debris generated around a hard laser mark on the back surface of a wafer using an image obtained by a visual inspection device. [Background technology]

[0002] In order to identify individual silicon wafers, there is a process in which an individual number is printed on the flat part at the edge of the back surface of the wafer using a solid-state laser (hard laser marking process). Hard laser marking involves melting the wafer itself with a high-power solid-state laser while intermittently forming dots to imprint the characters. This causes the silicon around the dots to become amorphous, and in the subsequent polishing process, the amorphous parts cannot be polished at the same polishing speed as other single crystal parts. It is believed that this causes localized protrusions with a relatively gentle slope to form in the amorphous parts around the dots. This is called debris, and it has been pointed out that if debris interferes with the stage in the device process, it can cause problems in device manufacturing. For this reason, it is necessary to detect debris that occurs in the laser mark area.

[0003] Conventionally, a method was used in which debris in hard laser marks was identified as a shape abnormality due to thickness changes using a shape measuring machine, but there have been cases where debris that cannot be detected by a shape measuring machine has caused problems in the device manufacturing process. Therefore, it is necessary to reliably detect such debris that cannot be detected by a shape measuring machine.

[0004] As a conventional technique, a method for detecting surface irregularities and surface defects by image processing has been disclosed. For example, Patent Document 1 shows an example of image processing of a spherical concave portion and a spherical convex portion, and discloses that in REV mode (reverse position defocus), a convex shape is imaged brightly (a concave shape is imaged darkly), and that in FOW mode (forward position defocus), a concave shape is imaged brightly (a convex shape is imaged darkly).

[0005] This technology is intended to detect depression-like defects caused by processing or crystallization, and is not intended to detect gentle protrusions (debris). When attempting to detect surface irregularities in an area printed with hard laser marks using this method, only the irregularities in the printed areas of the hard laser marks are detected, and it is not possible to detect gentle protrusions (debris).

[0006] Furthermore, Patent Document 2 discloses a defect inspection method in which the surface of an object to be inspected is photographed with a differential interference microscope and the number of defects observed on the surface is counted by image processing, characterized in that defects are detected based on points where brightness changes in the photographed image. However, this technology is a method for measuring the number of defects on the surface, and cannot detect gentle protrusions (debris) in the hard laser mark printing area on the back surface. [Prior art documents] [Patent documents]

[0007] [Patent Document 1] JP 2017-53764 A [Patent Document 2] JP 2002-365236 A Summary of the Invention [Problem to be solved by the invention]

[0008] As described above, conventionally, a shape measuring machine is used to identify debris in hard laser marks as shape abnormalities due to thickness changes, but there have been cases where debris that cannot be detected by a shape measuring machine has caused problems in the device manufacturing process.

[0009] Therefore, an object of the present invention is to provide a debris determination method that can reliably detect debris that cannot be detected by such shape measuring instruments and determine the presence or absence of debris. [Means for solving the problem]

[0010] In order to achieve the above object, the present invention provides a method for determining debris generated around a hard laser mark on a back surface of a wafer after forming the hard laser mark on the back surface, or after polishing the back surface after forming the hard laser mark, by using an image obtained by a visual inspection device, the method comprising: A step A of calculating a reference luminance from the grayscale image obtained by the appearance inspection device; A process B of extracting a hard laser mark printed area including the hard laser mark from the grayscale image; A process C of excluding the dot portion of the hard laser mark from the extracted hard laser mark printing area; A process D of extracting the debris area from the hard laser mark printing area excluding the dot portion of the hard laser mark based on the reference brightness; A step E of determining the presence or absence of debris in the hard laser mark printing area based on the extracted debris area; The present invention provides a debris determination method comprising the steps of:

[0011] The debris determination method of the present invention makes it possible to reliably extract only debris regions that cannot be detected by shape measuring instruments, and can determine the presence or absence of debris more reliably than conventional methods. Furthermore, since the debris region is extracted based on the reference brightness, it is possible to suppress detection variations due to individual variations (machine differences) between appearance inspection devices.

[0012] At this time, in the step A, The average brightness of the portions of the grayscale image where the hard laser marks are not engraved can be calculated as the reference brightness.

[0013] In this way, a reference brightness can be established that is not influenced by the brightness of the hard laser mark printing area, and the pixel brightness threshold for extracting the debris area based on this reference brightness can be set with high accuracy.

[0014] In addition, in the step B, A pixel brightness threshold value (specific frame) capable of detecting a white specific frame displayed so as to surround the hard laser mark printing area in the grayscale image is set in advance; A region equal to or larger than the threshold (specific frame) is determined to be the white specific frame and detected; The area surrounded by the detected specific white frame can be extracted as the hard laser mark printing area.

[0015] Since the white specific frame (automatic function of the visual inspection device) surrounding the hard laser mark printing area has a higher pixel brightness than other areas, it is possible to set the pixel brightness value as a threshold (specific frame) for extracting only the white specific frame based on the difference in pixel brightness values ​​between the white specific frame and other areas. By extracting only the specific white frame in this way, the hard laser mark printed area therein can be extracted, and it is possible to simply target only the hard laser mark printed area where debris is generated for image processing.

[0016] In the step C, a pixel brightness threshold value (dot area) capable of extracting only the dot area is set in advance based on a relationship between the dot area constituting the hard laser mark and the pixel brightness of the dot area; A region having a pixel brightness equal to or less than the threshold value (dot region) is determined to be the dot region and extracted; An area expanded by one pixel or more around the periphery of the extracted dot area can be excluded as the dot portion.

[0017] The dot area (the area that is one pixel or more larger than the dot area) will not become a debris area, so it must be excluded from the hard laser mark printing area (i.e., the image processing area). First, the pixel brightness of the dot area that constitutes the hard laser mark is particularly low, so the dot area can be easily identified from the pixel brightness value. In this case, by expanding the area to be removed as the dot area to an area that is at least one pixel larger than the dot area, it is possible to prevent the periphery of the dot from being erroneously detected as debris due to changes in pixel brightness caused by the influence of the dot area.

[0018] In addition, in the step D, a relationship between the pixel brightness of the high-brightness debris and the pixel brightness of the high-brightness debris, and a relationship between the pixel brightness of the low-brightness debris and the pixel brightness of the low-brightness debris and a pixel brightness threshold value (high-brightness debris) that can extract the high-brightness debris based on the reference brightness, and a pixel brightness threshold value (low-brightness debris) that can extract the low-brightness debris; Areas with pixel brightness equal to or greater than the threshold (high brightness debris) and areas with pixel brightness equal to or less than the threshold (low brightness debris) can be determined as debris areas and extracted.

[0019] By extracting the debris area based on the reference brightness in this manner, it is possible to more reliably prevent detection variations due to individual variations in the appearance inspection device.

[0020] In addition, in the step E, A relationship between an area ratio of the debris area in the hard laser mark printing area and quality defects caused by debris in a device manufacturing process is obtained in advance, and a threshold value (with debris) of the area ratio of the debris at which the quality defects caused by the debris occur is set; calculating an area ratio of the debris area in the hard laser mark printing area from the number of pixels of the extracted debris area; When the area ratio of the debris region is equal to or greater than the threshold value (debris present), it can be determined that debris is present.

[0021] In this way, by determining the area ratio of debris, a quantitative evaluation becomes possible. Furthermore, if it is known that defects in quality characteristics occur in the device manufacturing process due to debris, then making a judgment as described above can provide an extremely accurate and effective method for determining debris.

[0022] The present invention also provides a wafer having a hard laser mark on a back surface, comprising: There is provided a wafer having the hard laser marking area which has been determined to be free of debris by the debris determination method of the present invention.

[0023] Such a wafer of the present invention is an acceptable product that has been determined to be free of debris that cannot be detected by a shape measuring instrument, and is a non-defective product that can prevent problems caused by debris even when it is subsequently subjected to device manufacturing processes. Effect of the Invention

[0024] The debris determination method of the present invention can reliably detect debris that cannot be detected by a shape measuring instrument and determine whether or not it is present. Moreover, this method can suppress the inspection variation between appearance inspection devices. [Brief description of the drawings]

[0025] [Figure 1] FIG. 2 is a flow diagram showing an example of a debris determination method of the present invention. [Diagram 2] FIG. 11 is a measurement diagram showing an example of a grayscale image obtained by a visual inspection device. [Diagram 3] FIG. 2 is an explanatory diagram showing changes in luminance from 0 to 255 in the gray scale. [Figure 4] 1 is a graph showing the relationship between the marking depth and brightness of a hard laser mark. [Diagram 5] 11 is a graph showing an example of an image portion in the vicinity of a dot region and a change in luminance of the image portion. [Figure 6]11 is a graph showing a part of an area printed with hard laser marks and a change in luminance when the part is scanned. [Figure 7] FIG. 13 is an explanatory diagram showing a state when debris is irradiated with light by a visual inspection device. [Figure 8] 1 is a graph showing an example of the relationship between the area ratio of the debris region and the occurrence of defocusing in a device manufacturing process. [Figure 9] 13 is an image of a print region in the first embodiment where the area ratio of the debris region is 0% (debris-free determination). [Figure 10] 13 is an image of a print region in Example 1 where the area ratio of the debris region is 1.5% (debris presence determination). [Figure 11] 1 is a graph showing an example of the relationship between ESFQR and the occurrence of defocus in a device manufacturing process. [Figure 12] 13 is a graph showing an instrumental difference in the area ratio of the debris region in Example 2. [Figure 13] 13 is a graph showing an instrumental difference in the area ratio of the debris region in Comparative Example 2. [Figure 14] FIG. 13 is an explanatory diagram showing a cell at a notch position including a hard laser mark. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0026] Hereinafter, the present invention will be described with reference to the drawings. However, the present invention is not limited to the embodiments. For simplicity, hereinafter, the hard laser mark may be referred to as HLM, and the hard laser mark printed area may be referred to as printed area. The present invention is a method for determining debris generated around the HLM on the back surface of a wafer on which an HLM is formed as an individual number, or on the back surface that has been further polished thereafter, by using an image (grayscale image) obtained by a visual inspection device. As the visual inspection device, for example, a conventional commercially available one can be used. One example is SIFTer300. In particular, it is preferable to use a device that has a function of automatically surrounding an area (printing area) including the HLM with a specific white frame and leaving the print area as an image.

[0027] An example of the debris determination method of the present invention is shown in Figure 1. It is roughly divided into process A: calculation of reference brightness, process B: extraction of print area, process C: removal of dot parts, process D: extraction of debris area, and process E: determination. An explanatory diagram for each process is also shown. Each step is described in detail below.

[0028] <Step A: Calculating the reference brightness> This is the process of calculating the reference luminance from the grayscale image obtained by the appearance inspection device. The grayscale image taken by the visual inspection device is shown in the upper left of Figure 1. The HLM and a white identification frame that identifies the position of the HLM are displayed. An example of an actual grayscale image is also shown in Figure 2. This is a hard laser mark printing position inspection image B_T7 (BMP format) taken by a visual inspection device, and is 8-bit grayscale (256 gradations).

[0029] Of such grayscale images (256 levels), the average luminance of the areas where HLM is not imprinted is calculated as the reference luminance. Differences in brightness occur even when the same wafer is used due to differences in appearance inspection equipment. Therefore, the accuracy of debris detection is improved by calculating the average brightness value of the areas where the HLM is not marked for each wafer being measured. In this case, the area for calculating the average brightness value is not particularly limited, but it can be, for example, a portion where the HLM is not engraved, on the same circumference as the white specific frame that identifies the position of the HLM, and 10 pixels away from the right end of the specific frame. More specifically, the average brightness value of each wafer can be calculated in a square area whose one side is the length of the short side of the specific frame (or the printing area).

[0030] <Process B: Print area extraction> This is the process of extracting the print area including the HLM from the grayscale image. First, in order to extract only the print area, a pixel brightness threshold (specific frame) is set that can detect a white specific frame from the grayscale image (256 gradations). For example, the specific frame is detected by identifying pixels with a pixel brightness of ≧240. Since this specific frame is displayed in white (in the example shown in Figure 1, it is displayed in white in the grayscale image, while it is highlighted in the drawing of the area of ​​process A for easier understanding), the threshold value does not change depending on the wafer or equipment. Figure 3 shows the change in brightness from 0 to 255 in the grayscale. The brightness of the white specific frame is 255 out of the grayscale image (256 gradations), and in order to reliably detect the specific frame, as described above, for example, 240 is set as the threshold (specific frame), and the grayscale image is processed from left to right to detect the white square specific frame, and the enclosed area is extracted as the hard laser mark printing area. In addition, the upper limit of pixel brightness in the "area equal to or greater than the threshold (specific frame)" that is determined to be a white specific frame can be set to 255 because it is a grayscale. In this way, the printing area where debris occurs can be easily extracted.

[0031] Of course, the present invention is not limited to this method. Even if the appearance inspection device is not equipped with an automatic function of drawing a specific white frame, it is possible for an operator to manually trim the area including the HLM in the grayscale image to extract the printing area.

[0032] <Process C: Excluding dotted parts> This is the process of excluding the HLM dots from the extracted print area. First, let us briefly explain the terminology used here. The part that constitutes the HLM print itself is called the dot area. In addition, the area that extends beyond the dot area by one pixel or more (for example, 1 to 5 pixels) is called the dot part.

[0033] Now, a pixel brightness threshold (dot area) that can extract only dot areas is set in advance based on the relationship between the dot area and its pixel brightness. This is because areas with pixel brightness below this threshold (dot area) are determined to be dot areas and extracted. For example, Figure 4 is a graph showing the relationship between HLM printing depth and brightness when HLMs of different depths are printed on a 300 mm diameter wafer. The target printing depth for a typical HLM is 45μm±5μm, but here, taking into consideration the variation in required specifications, the brightness for a printing depth of 35μm to 110μm is shown. In this case, if the pixel brightness threshold (dot area) is set to 65 or less, the HLM dot area can be reliably excluded from debris detection targets. The lower limit of pixel brightness in the "area below the threshold (dot area)" that is determined to be a dot area can be set to 0 because of the grayscale.

[0034] At this time, the pixel brightness decreases in the outer periphery of the dot area (exclusion area around the dot area) as shown in FIG. 5. FIG. 5 is a graph showing an image portion near the dot area and the change in brightness of the image portion. This change in brightness may be erroneously counted as debris, so to prevent such erroneous detection, the dot area is enlarged by at least one pixel and excluded. As described above, depending on the size of the image to be processed, the number of pixels to be enlarged can be varied, for example, between 1 and 5 pixels. Naturally, it is possible to enlarge by more than 5 pixels if necessary, and there is no upper limit as it depends on the size of the image, etc.

[0035] When actually processing the image (i.e., the print area extracted in step B), the area with pixel brightness below a threshold (dot area) is extracted as the dot area, and then the dot portion obtained by enlarging it is excluded. This makes it possible to prevent erroneous detection of debris.

[0036] <Step D: Debris area extraction> This is the process of extracting the debris area from the printing area excluding the HLM dots, using a reference brightness as a reference. Two relationships are determined in advance: the relationship between debris with a higher brightness than the reference brightness and its pixel brightness, and the relationship between debris with a lower brightness than the reference brightness and its pixel brightness. From these relationships, two thresholds are set: a pixel brightness threshold (high brightness debris) that allows high brightness debris to be extracted based on the reference brightness, and a pixel brightness threshold (low brightness debris) that allows low brightness debris to be extracted based on the reference brightness. Then, from the actual image (i.e., the printing area excluding the dots in step C), areas above this threshold (high brightness debris) and areas below the threshold (low brightness debris) are extracted as debris areas. In the example shown in Figure 1, the number of pixels counted as debris areas is 40 and 50, totaling 90.

[0037] For example, the threshold (low brightness debris) can be "pixel brightness≦[reference brightness−15]", and the threshold (high brightness debris) can be "pixel brightness≧[reference brightness+20]". Then, for the image of the printed area from which the dots have been removed as described above, the printed area is scanned according to the above criteria, and the number of pixels in the area where the debris occurs (debris area) is counted. Figure 6 shows an example of a part of the printed area and the change in brightness when this part is scanned. There are areas where the threshold is exceeded on both the high brightness side and the low brightness side of the reference brightness. The upper limit of pixel brightness relative to the reference brightness in the "area above the threshold (high brightness debris)" extracted as high brightness debris can be set to +255 because of the grayscale. The lower limit of pixel brightness relative to the reference brightness in the "area below the threshold (low brightness)" extracted as low brightness debris can be set to -255 because of the grayscale.

[0038] We will now explain high brightness debris and low brightness debris. Figure 7 is an explanatory diagram showing what happens when debris is irradiated with light by a visual inspection device. Debris forms protrusions with a relatively gentle slope. The visual inspection device uses a lens + coaxial lighting to shine light onto the wafer. If a gentle protrusion occurs and the light is scattered so that it does not return to the camera, the brightness will be low, and if the light returns strongly, the brightness will be bright, and high and low brightness will be detected relative to the reference brightness. In the present invention, for example, a wafer that has become a problem in a device manufacturing process is used as a sample, and the thresholds for the high brightness side and the low brightness side described above can be determined. In particular, it is preferable to appropriately determine the thresholds (high brightness debris) and (low brightness debris) taking into consideration the correlation with the device process and the differences between equipment. Since the debris area is extracted based on the reference brightness, this can be useful in preventing detection variations due to differences in appearance inspection devices.

[0039] <Process E: Judgment> This is a process for determining the presence or absence of debris in the printing area based on the extracted debris area. First, the relationship between the area ratio of the debris in the printing area and quality defects caused by debris in the device manufacturing process is obtained in advance. Furthermore, from that relationship, a threshold value of the area ratio of the debris (with debris) at which quality defects caused by debris occur is set. Then, the area ratio of the debris area in the printed area is calculated from the number of pixels of the debris area extracted in process D on the actual wafer being evaluated, and if this is equal to or greater than the above threshold (debris present), it is determined that debris is present (debris that may cause quality defects).

[0040] Figure 8 shows an example of the relationship between the area ratio of the debris region and the occurrence of defocusing in the device manufacturing process. The sample levels (slots) on the horizontal axis are sorted from high to low by the area ratio of the debris region for easy understanding. In this case, defocusing problems begin to occur when the area ratio of the debris area is around 0.22% of the processing area (printing area), so 0.2% can be set as the threshold (debris present), and it can be determined that debris is present when it is 0.2% or more.

[0041] Therefore, for the debris area actually extracted in process D, the area ratio of the debris area is calculated from the ratio of [number of pixels in the debris area] / [number of pixels in the HLM printing area], and the presence or absence of debris can be judged with high accuracy based on whether or not the area ratio meets the pass / fail criterion of 0.2% of the processing target area (printing area). In particular, it is possible to reliably select wafers that will not suffer from defocusing defects in the device manufacturing process. In the example shown in Figure 1, the number of pixels in the print area is 40,000 (400 vertical x 100 horizontal), but the debris area is 90, giving an area ratio of 2.25%. Since this is greater than the above threshold value of 0.2% (debris present), it is determined that debris is present, and a defocusing failure is expected to occur. Incidentally, the upper limit of the area ratio of the debris region cannot be determined because it depends on the number of pixels in the debris region, but due to the calculation formula for the area ratio of the debris region mentioned above, the maximum is 100%.

[0042] The present invention as described above can reliably detect debris that cannot be detected by conventional inspection methods using shape measuring instruments, and can accurately determine the presence or absence of debris. Moreover, it is possible to detect debris while suppressing detection variations due to differences in the appearance inspection device.

[0043] Furthermore, an acceptable wafer having a printing area that has been determined to be debris-free using the determination method of the present invention has passed the highly accurate determination described above, and therefore can prevent problems such as poor defocus caused by debris from occurring during the device manufacturing process. EXAMPLES

[0044] The present invention will be described in more detail below with reference to examples and comparative examples of the present invention, but the present invention is not limited to these. Example 1 The wafers were prepared with a diameter of 300 mm, a crystal plane orientation of (110), and a number of (25+2). The hard laser mark was engraved at an angle of 5±1° from the notch on the back surface of the wafer, and both sides were polished. Of the total 27 images, 25 were used to investigate the relationship between the area ratio of debris regions and the occurrence of defocusing during the device manufacturing process, and the remaining two images were used to determine whether or not there was debris that could affect the occurrence of defocusing based on that relationship.

[0045] For these wafers, 8-bit grayscale (256 gradations) images in BMP format obtained by a visual inspection device (SIFTer300) were processed to detect debris.

[0046] <Step A: Calculating the reference brightness> For the areas where the hard laser mark was not engraved, the average brightness value of each wafer was calculated in a square area with one side equal to the length of the short side of the specific frame, located on the same circumference as the printed area where the hard laser mark was printed (or the area surrounded by a white specific frame), 10 pixels away from the right end of the specific frame. As a result, the minimum value was 118 and the maximum value was 119 for the 25 samples.

[0047] <Process B: Print area extraction> For the extraction process of the hard laser mark printing area, the position of a white specific frame was identified using a threshold (specific frame) of "pixel brightness ≧ 240", and the hard laser mark printing area was extracted from an angle of 5° from the notch with a pixel size of approximately 105 × 430.

[0048] <Process C: Excluding dotted parts> Within the hard laser mark printing area, the dot area of ​​the hard laser mark was identified by setting a threshold (dot area) of "pixel brightness ≦ 65", and the area to be removed was defined as an area (dot portion) expanded by one pixel from the outer periphery of the dot area.

[0049] Here, a specific image processing method for removing this portion will be described. The upper left pixel within the hard laser mark printing area is used as the starting point and scanned using a raster scan. If a pixel with "pixel brightness ≦ 65" is detected, it is recorded as the starting coordinate for the dot area detection process. The surrounding eight pixels are examined counterclockwise around the starting coordinate, and the coordinates of the first pixel that satisfies "pixel brightness ≦ 65" are recorded as the dot perimeter and become the new starting coordinates. The above process is repeated until the initially set starting coordinates are returned to. To expand the dot perimeter by 1 pixel, the coordinates of the eight pixels surrounding each recorded pixel are also similarly recorded as dot perimeters. The area inside the dot perimeter is recorded as the dot area. The dot periphery and dot area obtained as described above are excluded from the debris detection process.

[0050] <Step D: Debris area extraction> The specific method for counting the number of pixels where debris is generated is as follows. The upper left pixel in the hard laser mark printing area is set as the starting point and the area is scanned by raster scanning. If the pixel brightness is less than or equal to the threshold value of [reference brightness - 15] (low brightness debris), or if the pixel brightness is greater than or equal to the threshold value of [reference brightness + 20] (high brightness debris), the pixel is counted as a debris-generating pixel.

[0051] <Process E: Judgment> The area ratio of the debris region was calculated from the number of debris-generating pixels obtained, and the relationship with the occurrence of defocusing in the device manufacturing process was examined, resulting in a graph with a relationship similar to that shown in Figure 8. Regarding the area ratio of the debris region where a specific problem (defocusing) occurs, 0.22% was recognized as the boundary for occurrence of defocusing, and 0.2% or more was considered to be debris occurrence. These were the results obtained from the 25 wafers investigated.

[0052] The two test sheets were then judged according to the pass / fail criteria of 0.2% obtained from the graph in Figure 8. One had debris, but only a small amount, with the area ratio of the area being about 0%, and was judged to have no debris and passed. The other had debris at about 1.5%, and was judged to have no debris, and failed. Each image shows the location of debris in the hard laser mark printing area (representative areas are indicated by arrows). Figure 9 shows an image with an area ratio of 0%, and Figure 10 shows an image with an area ratio of 1.5%. Moreover, when the device was actually subjected to the manufacturing process, no defocusing defect occurred in the 0% judgment, whereas a huge defocusing defect occurred in the 1.5% judgment.

[0053] Comparative Example 1 The shape of the hard laser mark printed area was measured (ESFQR) using a commercially available shape measuring device (WaferSight; manufactured by KLA Tencor), and the results were analyzed to see whether separation was possible based on the defocus judgment results in the device process. The wafers prepared were the same as the 25 wafers used in the investigation in Example 1. ESFQR is calculated by the range of positive and negative deviations from the least squares method within a rectangular area (cell). The roughly rectangular area to be measured is surrounded by an arc equivalent to 10 mm in the diameter direction from the outer edge and 18° in the circumferential direction, and the cell at the wafer central angle of 270° (the cell at the notch position) is the area that contains the hard laser mark. Figure 14 shows the cell at the notch position that contains the hard laser mark.

[0054] 11 shows an example of the relationship between ESFQR and the occurrence of defocusing in a device manufacturing process. The arrangement of sample levels on the horizontal axis is the same as that in the first embodiment. The 270° rectangular region where the shape was measured corresponds to the position where the hard laser mark was formed, but it is clear that the presence or absence of a defocus defect caused by debris in the device process cannot be determined based on the ESFQR value. In other words, when we tried to classify problematic wafers based on the ESFQR (Estimated Shape Change in Thickness) as a conventional debris anomaly screening method, we got the following results. Some wafers that had defocusing problems in the device process showed high values, but some did not show high values ​​even with huge defocusing. Conversely, some wafers that had no problems showed high values, so in the end, ESFQR cannot be used to screen for the presence or absence of debris.

[0055] (Example 2, Comparative Example 2) One wafer from Slot-15 was selected from the same wafers as in Example 1, and debris was determined using six visual inspection devices. At this time, the inter-device (machine difference) differences in debris determination were compared between a case in which the reference brightness for process A was calculated for each wafer using six visual inspection devices (Example 2) and a case in which the reference brightness for process A was not calculated for each wafer using six visual inspection devices (note that in this case, low-brightness debris was extracted when "pixel brightness ≦ 105" and high-brightness debris was extracted when "pixel brightness ≧ 140") (Comparative Example 2). The respective results are shown in FIG. 12 and FIG.

[0056] In Example 2, as shown in FIG. 12, the minimum value of the area ratio of the debris region among six units was 0.07%, the maximum value was 0.13%, and the standard deviation was 0.024. By using the reference brightness, it is possible to absorb differences between instruments and reduce the variation in the debris area ratio depending on the instrument, allowing any instrument to perform accurate debris determination.

[0057] On the other hand, in Comparative Example 2, as shown in FIG. 13, the minimum value was 0%, the maximum value was 0.17%, and the standard deviation was 0.063. Due to differences between instruments, the calculated debris area ratios vary greatly, making it difficult to make accurate debris determinations.

[0058] The present specification includes the following aspects. [1]: A method for determining debris generated around a hard laser mark on a back surface of a wafer after forming the hard laser mark on the back surface, or after polishing the back surface after forming the hard laser mark, using an image obtained by an appearance inspection device, comprising: A step A of calculating a reference luminance from the grayscale image obtained by the appearance inspection device; A process B of extracting a hard laser mark printed area including the hard laser mark from the grayscale image; A process C of excluding the dot portion of the hard laser mark from the extracted hard laser mark printing area; A process D of extracting the debris area from the hard laser mark printing area excluding the dot portion of the hard laser mark based on the reference brightness; A step E of determining the presence or absence of debris in the hard laser mark printing area based on the extracted debris area; A debris determination method comprising: [2]: In the step A, The debris determination method according to the above [1], wherein an average brightness of a portion of the grayscale image in which the hard laser mark is not engraved is calculated as the reference brightness. [3]: In the step B, A pixel brightness threshold value (specific frame) capable of detecting a white specific frame displayed so as to surround the hard laser mark printing area in the grayscale image is set in advance; A region equal to or larger than the threshold (specific frame) is determined to be the white specific frame and detected; The debris determination method according to the above [1] or [2], wherein the area surrounded by the detected specific white frame is extracted as the hard laser mark printing area. [4]: In the step C, a pixel brightness threshold value (dot area) capable of extracting only the dot area is set in advance based on a relationship between the dot area constituting the hard laser mark and the pixel brightness of the dot area; A region having a pixel brightness equal to or less than the threshold value (dot region) is determined to be the dot region and extracted; The debris determination method according to any one of [1] to [3] above, wherein an area expanded by at least one pixel around the periphery of the extracted dot area is excluded as the dot portion. [5]: In the step D, a relationship between the pixel brightness of the high-brightness debris and the pixel brightness of the high-brightness debris, and a relationship between the pixel brightness of the low-brightness debris and the pixel brightness of the low-brightness debris and a pixel brightness threshold value (high-brightness debris) that can extract the high-brightness debris based on the reference brightness, and a pixel brightness threshold value (low-brightness debris) that can extract the low-brightness debris; A debris determination method according to any one of [1] to [4] above, which determines and extracts areas having pixel brightness equal to or greater than the threshold (high brightness debris) and areas having pixel brightness equal to or less than the threshold (low brightness debris) as debris areas. [6]: In the step E, A relationship between an area ratio of the debris area in the hard laser mark printing area and quality defects caused by debris in a device manufacturing process is obtained in advance, and a threshold value (with debris) of the area ratio of the debris at which the quality defects caused by the debris occur is set; calculating an area ratio of the debris area in the hard laser mark printing area from the number of pixels of the extracted debris area; A debris determination method according to any one of [1] to [5] above, which determines that debris is present when the area ratio of the debris region is equal to or greater than the threshold (debris present). [7]: A wafer having a hard laser mark on the backside, A wafer having the hard laser marking area that is determined to be free of debris by any of the debris determination methods [1] to [6] above.

[0059] The present invention is not limited to the above-described embodiment. The above-described embodiment is merely an example, and anything that has substantially the same configuration as the technical idea described in the claims of the present invention and exhibits similar effects is included in the technical scope of the present invention.

Claims

1. A method for determining debris generated around a hard laser mark on a back surface of a wafer after forming the hard laser mark on the back surface, or after polishing the back surface after forming the hard laser mark, using an image obtained by an appearance inspection device, comprising: A step A of calculating a reference luminance from the grayscale image obtained by the appearance inspection device; A process B of extracting a hard laser mark printed area including the hard laser mark from the grayscale image; A process C of excluding the dot portion of the hard laser mark from the extracted hard laser mark printing area; A process D of extracting the debris area from the hard laser mark printing area excluding the dot portion of the hard laser mark based on the reference brightness; A step E of determining whether or not debris is present in the hard laser mark printing area based on the extracted debris area; A debris determination method comprising:

2. In the step A, The debris determination method according to claim 1 , further comprising the step of calculating an average brightness value of a portion of the grayscale image in which the hard laser mark is not engraved as the reference brightness.

3. In the step B, a pixel brightness threshold value (specific frame) capable of detecting a white specific frame displayed so as to surround the hard laser mark printing area in the grayscale image is set in advance; A region equal to or larger than the threshold (specific frame) is determined to be the white specific frame and detected; 2. The debris determination method according to claim 1, wherein an area surrounded by the detected specific white frame is extracted as the hard laser mark printing area.

4. In the step B, a pixel brightness threshold value (specific frame) capable of detecting a white specific frame displayed so as to surround the hard laser mark printing area in the grayscale image is set in advance; A region equal to or larger than the threshold (specific frame) is determined to be the white specific frame and detected; 3. The debris determination method according to claim 2, wherein an area surrounded by the detected specific white frame is extracted as the hard laser mark printing area.

5. In the step C, a pixel brightness threshold value (dot area) capable of extracting only the dot area is set in advance based on a relationship between a dot area constituting the hard laser mark and the pixel brightness of the dot area; A region having a pixel brightness equal to or less than the threshold value (dot region) is determined to be the dot region and extracted; 2. The debris determination method according to claim 1, wherein an area expanded by at least one pixel around the periphery of the extracted dot area is excluded as the dot area.

6. In the step C, a pixel brightness threshold value (dot area) capable of extracting only the dot area is set in advance based on a relationship between a dot area constituting the hard laser mark and the pixel brightness of the dot area; A region having a pixel brightness equal to or less than the threshold value (dot region) is determined to be the dot region and extracted; 3. The debris determination method according to claim 2, wherein an area expanded by at least one pixel around the periphery of the extracted dot area is excluded as the dot portion.

7. In the step C, a pixel brightness threshold value (dot area) capable of extracting only the dot area is set in advance based on a relationship between a dot area constituting the hard laser mark and the pixel brightness of the dot area; A region having a pixel brightness equal to or less than the threshold value (dot region) is determined to be the dot region and extracted; 4. The debris determination method according to claim 3, wherein an area expanded by at least one pixel around the periphery of the extracted dot area is excluded as the dot portion.

8. In the step C, a pixel brightness threshold value (dot area) capable of extracting only the dot area is set in advance based on a relationship between a dot area constituting the hard laser mark and the pixel brightness of the dot area; A region having a pixel brightness equal to or less than the threshold value (dot region) is determined to be the dot region and extracted; 5. The debris determination method according to claim 4, wherein an area expanded by at least one pixel around the periphery of the extracted dot area is excluded as the dot area.

9. In the step D, a relationship between the pixel brightness of the high-luminance debris and the pixel brightness of the high-luminance debris, and a relationship between the pixel brightness of the low-luminance debris and the pixel brightness of the low-luminance debris and a pixel brightness threshold value (high-luminance debris) that can extract the high-luminance debris based on the reference brightness, and a pixel brightness threshold value (low-luminance debris) that can extract the low-luminance debris; 9. The debris determination method according to claim 1, further comprising determining and extracting areas having pixel brightness equal to or greater than the threshold (high brightness debris) and areas having pixel brightness equal to or less than the threshold (low brightness debris) as debris areas.

10. In the step E, A relationship between an area ratio of the debris area in the hard laser mark printing area and quality defects caused by debris in a device manufacturing process is obtained in advance, and a threshold value (with debris) of the area ratio of the debris at which the quality defects caused by the debris occur is set; calculating an area ratio of the debris area in the hard laser mark printing area from the number of pixels of the extracted debris area; 9. The debris determination method according to claim 1, wherein if the area ratio of the debris region is equal to or greater than the threshold value (debris present), it is determined that debris is present.

11. In the step E, A relationship between an area ratio of the debris area in the hard laser mark printing area and quality defects caused by debris in a device manufacturing process is obtained in advance, and a threshold value (with debris) of the area ratio of the debris at which the quality defects caused by the debris occur is set; calculating an area ratio of the debris area in the hard laser mark printing area from the number of pixels of the extracted debris area; 10. The debris determination method according to claim 9, further comprising determining that debris is present when the area ratio of the debris region is equal to or greater than the threshold value (debris present).

Citation Information

Patent Citations

  • Dot mark reading apparatus and reading method

    EP1073097A2

  • Hard laser marked wafer and manufacture thereof

    JP2000286173A

  • Reader and reading method for dot mark

    JP2001101337A

  • Defect inspection method and apparatus

    JP2002365236A

  • Mark forming method of semiconductor device

    JP2007150174A