Method of calculating inter-layer overlay offset and electronic device

Abstract

A method for calculating an overlay shift between layers and an electronic device. First, a target image is obtained. Next, a first region conforming to a first template is found in the target image. A second region conforming to a second template is found in the first region. A third region conforming to a third template corresponding to a first side and at least one fourth region conforming to at least one fourth template corresponding to a second side are found in the second region. Then, a center point of the fourth region with the highest similarity and a center point of the third region are used to calculate the overlay shift.

Description

Technical Field

[0001] This invention relates to an automation mechanism, and more particularly to a method and electronic device for calculating interlayer stack-up offsets in semiconductor processes. Background Technology

[0002] In semiconductor manufacturing, due to its multi-layered structure, incorrect alignment of layers during exposure can lead to defects such as short circuits and open circuits, as the patterns between layers cannot closely match the circuit design. This misalignment is generally called overlay error. Currently, after exposure and development, images are taken using an inline automatic optical inspection (AOI) system, and the overlay is manually checked and the offset is calculated. This offset is then input into the exposure machine to adjust parameters for re-exposure.

[0003] Pairing errors have a significant impact on the yield of the entire semiconductor process. Therefore, the detection of pairings and the measurement of pairing offsets are among the issues that need to be addressed. Summary of the Invention

[0004] This invention provides a method and electronic device for calculating interlayer stacking offset, which can automatically calculate the offset.

[0005] The present invention provides a method for calculating interlayer stacking offset in semiconductor processes, suitable for execution by a processor. The method includes: obtaining a target image; identifying a first region in the target image that conforms to a first template; identifying a second region in the first region that conforms to a second template; identifying a third region in the second region that conforms to a third template corresponding to a first side; identifying at least one fourth region that conforms to at least one fourth template corresponding to the second side; and calculating the offset by identifying the center point of the most similar region in the at least one fourth region and the center point of the third region.

[0006] In one embodiment of the present invention, the method further includes: taking multiple images at each of a plurality of shooting positions of the target object via an imaging device to obtain a plurality of shooting images corresponding to each shooting position; and analyzing the sharpness of the plurality of shooting images obtained at each shooting position to select the one with the best sharpness among these shooting images as the target image for each shooting position.

[0007] In one embodiment of the present invention, each shooting position has a corresponding template combination, the template combination including a first template, a second template, a third template, and at least a fourth template. The method further includes: retrieving the corresponding template combination based on a shooting position corresponding to the target image.

[0008] In one embodiment of the present invention, after calculating the offset, the method further includes: marking the offset at the corresponding position in the target image, and visually indicating whether the offset needs to be adjusted.

[0009] In one embodiment of the present invention, before finding the first region that conforms to the first template in the target image, the method further includes: performing color block cutting on the target image using a color mask to retain the target color block, so as to find the first region that conforms to the first template in the target image after color block cutting.

[0010] The electronic device of the present invention includes: a memory comprising: a template database storing template combinations, the template combinations including a first template, a second template, a third template, and at least a fourth template; and a processor coupled to the memory and configured to perform the method for calculating interlayer stack-up offsets in a semiconductor process.

[0011] Based on the above, this disclosure designs an automatic measurement mode that automatically calculates the interlayer stacking offset using captured images, and then transmits the calculation results back to the exposure machine. This reduces the time spent on manual calculations and avoids losses in production yield. Attached Figure Description

[0012] Figure 1 This is a block diagram of an electronic device according to an embodiment of the present invention.

[0013] Figure 2 This is a schematic diagram of an interlayer overlay according to an embodiment of the present invention.

[0014] Figure 3 This is a flowchart of a method for calculating interlayer stacking offset according to an embodiment of the present invention.

[0015] Figure 4 This is a schematic diagram of the shooting position configuration according to an embodiment of the present invention.

[0016] Figure 5A and Figure 5B This is a schematic diagram of the region matching convolution according to an embodiment of the present invention.

[0017] Figure 6 This is a schematic diagram of obtaining a first region in a target image according to an embodiment of the present invention.

[0018] Figure 7This is a schematic diagram of obtaining a second region in a first region according to an embodiment of the present invention.

[0019] Figure 8 This is a schematic diagram of obtaining the third and fourth regions in the second region according to an embodiment of the present invention.

[0020] Figure 9 This is a schematic diagram of obtaining a third region and a plurality of fourth regions in a second region according to an embodiment of the present invention.

[0021] Figure 10 This is a schematic diagram of the output result according to an embodiment of the present invention.

[0022] Figure 11 This is a schematic diagram of a user interface according to an embodiment of the present invention.

[0023] Explanation of reference numerals in the attached figures:

[0024] 100: Electronic devices

[0025] 110: Processor

[0026] 120: Memory

[0027] 130: Template Database

[0028] 210-R, 210-G, 210-B: Drawings

[0029] 220: Black light-blocking layer

[0030] 410: Substrate

[0031] 500: Target Image

[0032] 501, 502, 503, 511, 521: Areas

[0033] 610: Area 1

[0034] 710: Second Zone

[0035] 810: Third Zone

[0036] 820, 830: Fourth Zone

[0037] 1000: Output result

[0038] 1100: User Interface

[0039] A: Overlapping

[0040] d1, d2: Length

[0041] P82: Overlapping Regions

[0042] T4-1, T4-2: Fourth Template

[0043] X1, Y1: Dimensions

[0044] S305~S325: Steps for calculating interlayer stack offset Detailed Implementation

[0045] Figure 1 This is a block diagram of an electronic device according to an embodiment of the present invention. Please refer to... Figure 1 The electronic device 100 can be implemented using any device with computing capabilities. For example, the electronic device 100 can be implemented using a smartphone, tablet, laptop, personal computer, server, etc. The electronic device 100 includes at least the following components: a processor 110 and a memory 120. The processor 110 is coupled to the memory 120.

[0046] The processor 110 may be, for example, a central processing unit (CPU), a physical processing unit (PPU), a programmable microprocessor, an embedded control chip, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), or other similar devices.

[0047] The memory 120 is, for example, any form of fixed or removable random access memory (RAM), read-only memory (ROM), flash memory, hard disk, or other similar device or combination thereof. The memory 120 includes a template database 130 and one or more program code segments, which, after being installed, will be executed by the processor 110.

[0048] Figure 2 This is a schematic diagram of an interlayer overlay according to an embodiment of the present invention. Please refer to... Figure 2 This embodiment describes a color filter using a black light-shielding layer and a filter layer, but it is not limited to this and can be applied to any two-layer structure in semiconductor processes. The structure of the color filter involves forming an anti-reflective black light-shielding layer 220, also called a BM (black matrix) layer, on a glass substrate. Then, a transparent red, green, and blue (RGB) three-primary-color pattern (such as...) is formed on the black light-shielding layer 220. Figure 2The filter layers (patterns 210-R, 210-G, and 210-B) shown are finally sputtered with a transparent conductive film. The shape, size, color, and arrangement of the filter layers (patterns 210-R, 210-G, and 210-B) are determined based on the application of different liquid crystal displays.

[0049] like Figure 2 As shown, patterns 210-R, 210-G, and 210-B were not aligned with the black light-blocking layer 220 during the formation process, resulting in misalignment gaps between patterns 210-R, 210-G, and 210-B and the black light-blocking layer 220, i.e., overlap A.

[0050] The following example illustrates the method for calculating interlayer stack offset. Figure 3 This is a flowchart of a method for calculating interlayer stacking offset according to an embodiment of the present invention. Please refer to... Figure 1 and Figure 3 In step S305, the target image is obtained.

[0051] In one embodiment, an image-capturing device can capture images multiple times at each of multiple shooting positions of the target object, obtaining multiple images corresponding to each shooting position. The image with the best clarity is then selected from these images to represent the target image corresponding to each shooting position. The image-capturing device may be, for example, a zoom lens, capturing images at multiple different focal lengths at each shooting position. In one embodiment, the image-capturing device may be located within an automatic optical recognition (AOI) system.

[0052] For example, Figure 4 This is a schematic diagram illustrating the configuration of the shooting positions according to an embodiment of the present invention. Please refer to... Figure 4 In this embodiment, the substrate 410 (target object) is divided into four regions (exposure times 4 times), and the four corners of each region are set as shooting positions (as indicated by the dots), for a total of 16 shooting positions. In other embodiments, the substrate 410 may also be divided into six regions (exposure times 6 times), with 24 shooting positions set; the configuration of the shooting positions is not limited here. The imaging device performs multiple image captures (e.g., 3 to 6 times) at different focal lengths at these 16 shooting positions to obtain multiple images corresponding to each shooting position.

[0053] After acquiring multiple sets of images corresponding to the target object (substrate 410) (one set of images per shooting position), the processor 110 first selects the image with the best clarity from the set of images corresponding to each shooting position as the target image for subsequent processing. In one embodiment, algorithms such as the Laplacian gradient function can be used to calculate the clarity of each image, and the image with the highest clarity is selected as the target image. For 16 shooting positions, 16 corresponding target images will be acquired.

[0054] Before obtaining the target image (highest resolution), the processor 110 can further perform image preprocessing on each group of captured images. The image preprocessing includes image processing techniques based on OpenCV, such as image breaking, noise removal, and brightness correction. For example, the original input captured image is subjected to erosion processing to initially remove noise, followed by dilation processing to restore it to its original range. Dilation processing can also be used to further connect broken objects (image breaking). Afterwards, the R, G, B distribution of the image after image breaking and noise removal is adjusted to align with the R, G, B distribution of a preset template (brightness correction) to obtain the corrected image (i.e., the target image). In one embodiment, a color scale map can be used for color alignment.

[0055] Furthermore, during the formation of the filter layer with R, G, and B primary color patterns, one color pattern is formed on the black light-blocking layer, followed by another color pattern. For example, a red pattern is formed first, then a green pattern, and finally a blue pattern. After forming one color pattern, the AOI's image-capturing device is driven to capture images multiple times at different focal lengths at multiple shooting positions, thereby obtaining multiple captured images corresponding to each shooting position. After the processor 110 obtains the captured images, it can further utilize a color mask to perform color block segmentation on the captured images to retain the target color blocks. Then, image preprocessing operations such as erosion, dilation, and brightness correction are performed on the image with only the target color blocks retained.

[0056] In one embodiment, after the red pattern processing is completed, the target image contains only the red pattern. A red mask is used for color block segmentation to retain the red pattern (target color block) in the target image. After the green pattern processing is completed, the target image contains only the red and green patterns. A green mask is used for color block segmentation to retain the green pattern (target color block) in the target image. After the blue pattern processing is completed, the target image contains patterns of red, green, and blue. A blue mask is used for color block segmentation to retain the blue pattern (target color block) in the target image.

[0057] In one embodiment, the red, green, and blue masks are masks based on the HSV (Hue, Saturation, Value) color space. In the HSV color space, H values ​​range from 0 to 180; S values ​​range from 0 to 255; and V values ​​range from 255. The red, green, and blue masks are obtained by adjusting the H values ​​(hue). For example, the H value of the red mask falls between 10 and 18, the H value of the green mask falls between 50 and 58, and the H value of the blue mask falls between 95 and 103.

[0058] After obtaining the target image, at least four templates are used to find the region with the highest similarity. The first template is used to cut out the region to be compared later, the second template is used to further narrow down the search range, and then dual templates (the third and fourth templates) are used for localization. For RGB filter layers, corresponding template combinations can be pre-set for different color patterns. For example, assuming there are 16 shooting locations, three sets of template combinations are set for each shooting location based on the three RGB colors.

[0059] bottom pairing Figure 6 , Figure 7 , Figure 8 Let's explain steps S310 to S325. Figure 6 This is a schematic diagram of obtaining a first region in a target image according to an embodiment of the present invention. Figure 7 This is a schematic diagram of obtaining a second region in a first region according to an embodiment of the present invention. This is also a schematic diagram of obtaining a third and a fourth region in a second region according to an embodiment of the present invention.

[0060] In step S310, the processor 110 identifies a first region 610 in the target image 500 that matches the first template. Here, the processor 110 uses convolutional matching to identify the first region 610 in the target image 500. In one embodiment, the processor 110 may start from the upper left corner of the target image 500 and extract regions with the same dimensions as the first template at intervals of length d1 in the vertical direction for similarity comparison with the first template, and also extract regions with the same dimensions as the first template at intervals of length d2 in the horizontal direction for similarity comparison with the first template.

[0061] Figure 5A and Figure 5B This is a schematic diagram of the convolution matching region according to an embodiment of the present invention. In this embodiment, it is assumed that the size of the first template is X1×Y1. Figure 5AAs shown, starting from the top left corner of the target image 500, region 501 is obtained by taking a length of 0 to X1 vertically downwards and a length of Y1 to the right. Next, region 511 is obtained by taking a length of d1 to d1+X1 vertically and a length of Y1 horizontally. Then, region 521 is obtained by taking a length of 2×d1 to 2×d1+X1 vertically and a length of Y1 horizontally. This process is repeated to obtain multiple regions in the first row vertically.

[0062] After that, as Figure 5B As shown, next, lengths d2 to d2+Y1 are taken in the horizontal direction, and length X1 is taken in the vertical direction to obtain region 502, and then... Figure 5A Similar to the example shown, multiple regions corresponding to the second row at positions d2 to d2+Y1 in the horizontal direction are sequentially obtained in the vertical direction. Then, regions 503 are obtained by taking lengths of 2×d2 to 2×d2+Y1 in the horizontal direction and length X1 in the vertical direction, and then... Figure 5A Similar to the example shown, multiple regions corresponding to the third row at positions 2×d2 to 2×d2+Y1 in the horizontal direction are sequentially obtained in the vertical direction. This process is repeated to obtain multiple regions for each row in the vertical direction.

[0063] Alternatively, in another embodiment, multiple regions can be obtained first in a horizontal column, and then multiple regions can be obtained in another horizontal column after moving down in the vertical direction.

[0064] After obtaining regions of the same size as the first template (501–503, 511, 521, etc.), the extracted regions are compared with the first template for similarity. For example, the mean squared error (MSE) algorithm can be used to calculate the distance (MSE) between pixels in each region and the first template in both images. The region with the smallest MSE is selected as the first region 610. If there are multiple regions with the smallest MSE, the structural similarity (SSIM) algorithm is used to calculate the similarity between these regions and the first template, and the region with the highest similarity is selected as the first region 610. The SSIM algorithm can compare the brightness, contrast, and structural similarity of two images.

[0065] Next, in step S315, a second region 710 that matches the second template is found in the first region 610. In this step, the processor 110 may use MSE and SSIM to find the second region 710.

[0066] Next, in step S320, a third region 810 matching the third template is found in the second region 710, and a fourth region 820 matching the fourth template is found. Here, to accurately determine the offset of the overlapping region P82, corresponding templates (third template, fourth template) need to be set for both sides of the overlapping region P82. In one embodiment, the first side (outer side) of the overlapping region P82 can be pre-set. Figure 7 , Figure 8 Set a third template on the right side of the overlapping area P82 (inner side). Figure 7 , Figure 8 The fourth template is set on the left side. Then, the processor 110 uses convolution matching, MSE and SSIM to find the third region 810 and the fourth region 820 located on both sides of the overlapping region P82 in the second region 710.

[0067] Then, in step S325, the offset is calculated using the center point of the fourth region 820 and the center point of the third region 810. Specifically, the processor 110 calculates the number of pixels between the center point of the fourth region 820 and the center point of the third region 810, and then converts the number of pixels into the actual length. For example, the conversion relationship between pixels and actual length can be obtained based on the imaging parameters of the target image.

[0068] In addition, to improve accuracy, multiple fourth templates can be set for the second (inner) side of the overlap.

[0069] For example, Figure 9 This is a schematic diagram illustrating the acquisition of a third region and multiple fourth regions in the second region according to an embodiment of the present invention. Please refer to... Figure 9 In this embodiment, for the second side (inner side) of the stack, Figure 7 , Figure 9 Two fourth templates, T4-1 and T4-2, are set on the left side of the second region 710, but this is not a limitation. The third region 810, which conforms to the third template, and the fourth regions 820 and 830, which conform to the fourth templates T4-1 and T4-2, are extracted from the second region 710. Then, the similarity between the fourth region 820 and the fourth template T4-1 is calculated, and the similarity between the fourth region 830 and the fourth template T4-2 is calculated. The region with the highest similarity is used for subsequent judgment. Assuming that the similarity between the fourth region 830 and the fourth template T4-2 is higher than that between the fourth region 820 and the fourth template T4-1, the offset is calculated using the center point of the fourth region 830 and the center point of the third region 810. For example, the distance (offset) can be calculated using the X or Y coordinate of the center points of the third and fourth regions, depending on whether the overlapping region P82 is vertical or horizontal. Alternatively, the shortest distance between the two center points can be calculated.

[0070] by Figure 8 For example, the absolute value obtained by subtracting the X-coordinates of the center point of the third region 810 and the center point of the fourth region 820 is used as the offset. Figure 9 For example, the absolute value obtained by subtracting the X coordinates of the center point of the third region 810 and the center point of the fourth region 830 is used as the offset.

[0071] exist Figure 8 and Figure 9 In the text, different patterns are used to represent different color gradients. Figure 8 , Figure 9 This is merely an example and is not intended to be limiting.

[0072] After obtaining the offset, the offset can be further marked at the corresponding position in the target image, and the need to adjust the offset can be indicated visually.

[0073] Figure 10 This is a schematic diagram of the output result according to an embodiment of the present invention. Please refer to... Figure 10 The output result 1000 indicates the calculated offset. Furthermore, a threshold value (e.g., 3μm) can be set. Offsets smaller than this threshold are marked with a first color, and offsets greater than or equal to this threshold are marked with a second color. These different colors allow users to easily identify which offsets need adjustment.

[0074] In another embodiment, a user interface may be provided for the user to further select captured images. Figure 11 This is a schematic diagram of a user interface according to an embodiment of the present invention. Please refer to... Figure 11 The user interface 1100 provides options such as the date of the photograph, the type of image to be displayed, and the identification code of the glass substrate (selecting the target object) for the user to choose from. After the user makes a selection, the processor 110 outputs the data according to the user's choice. For example, if the photograph date is selected as "2021 / 7 / 23", the glass substrate identification code is selected as "R20_AFF0X670FVL", and the image type is selected as "raw image", then the following output can be obtained: Figure 10 The output result shown is 1000 (displayed on the monitor).

[0075] Table 1 shows the average error between the actual measured offset and the offset obtained using three calculation methods in the target images obtained from shooting positions CD1 to CD4 at the four corners of the same exposure area. The first calculation method uses only general template matching. For example, a template image is slid across the target image, and the similarity between the image area under the template image and the template image is compared to find the most matching area. The second calculation method uses general template matching and image enhancement processing. The third calculation method uses the methods described in steps S305 to S325 of the above embodiments.

[0076] Table 1

[0077]

[0078] As shown in Table 1, the error value using the first calculation method is greater than 0.5 μm. Compared to the first calculation method, the error value using the second calculation method decreases to below 0.5 μm. Furthermore, the error value at shooting positions CD2 and CD3 also decreases significantly when using the third calculation method (the embodiment of this disclosure).

[0079] In summary, this disclosure presents an automatic measurement mode that automatically calculates the interlayer stacking offset using captured images, and then transmits the calculation results back to the exposure machine. This reduces the time spent on manual calculations and avoids losses in production yield.

Claims

1. A method for calculating interlayer stack-up offsets in a semiconductor process, suitable for execution by a processor, the method comprising: Obtain an image of a target; Find a first region in the target image that matches a first template; Find a second region in the first region that matches a second template; In the second region, find a third region that corresponds to a third template to a first side, and find at least a fourth region that corresponds to at least a fourth template to a second side; as well as An offset is calculated by finding the center point of the most similar element in at least one fourth region and the center point of the third region. The region with the highest similarity is the fourth region in the at least one fourth region that has the highest similarity to its corresponding fourth template.

2. The method of claim 1, further comprising: Multiple images are captured at each of multiple shooting positions of a target object using an imaging device to obtain multiple images corresponding to each of those shooting positions; as well as The sharpness of the images obtained from each of the shooting positions is analyzed, and the image with the best sharpness is selected as the target image for each of the shooting positions.

3. The method of claim 2, wherein each of the shooting locations has a corresponding template combination, the template combination including the first template, the second template, the third template, and the at least one fourth template, and the method further includes: The corresponding template combination is extracted based on one of the shooting positions corresponding to the target image.

4. The method of claim 1, wherein after calculating the offset, it further comprises: The offset is marked at the corresponding position in the target image, and whether the offset needs to be adjusted is indicated visually.

5. The method of claim 1, wherein before finding the first region in the target image that matches the first template, the method further comprises: A color mask is used to perform a color block cutting on the target image to retain a target color block, so as to find the first region that matches the first template in the target image after the color block cutting.

6. An electronic device comprising: A memory includes: a template database storing a template combination, the template combination including a first template, a second template, a third template, and at least a fourth template; and A processor, coupled to the memory, is configured to: Obtain an image of a target; Find a first region in the target image that matches a first template; Find a second region in the first region that matches a second template; In the second region, identify a third region that corresponds to a third template on a first side, and identify at least one fourth region that corresponds to at least one fourth template on a second side; and An offset is calculated by finding the center point of the most similar element in at least one fourth region and the center point of the third region. The region with the highest similarity is the fourth region in the at least one fourth region that has the highest similarity to its corresponding fourth template.

7. The electronic device of claim 6, wherein the processor is configured to: Multiple images are captured at each of multiple shooting positions of a target object using an imaging device to obtain multiple images corresponding to each of those shooting positions; The sharpness of the images obtained from each of the shooting positions is analyzed, and the image with the best sharpness is selected as the target image for each of the shooting positions.

8. The electronic device of claim 7, wherein the template database stores the template combination corresponding to each of the shooting positions. The processor is configured to retrieve the corresponding template combination based on one of the shooting positions corresponding to the target image.

9. The electronic device of claim 6, wherein the processor is configured to: The offset is marked at the corresponding position in the target image, and whether the offset needs to be adjusted is indicated visually.

10. The electronic device of claim 6, wherein the processor is configured to: A color mask is used to perform a color block cutting on the target image to retain a target color block, so as to find the first region that matches the first template in the target image after the color block cutting.

Images