Graph correction method, storage medium and storage terminal
By determining the pattern error and density calculation window in the mask and combining machine learning methods to correct the initial pattern to be corrected, the problem of pattern density imbalance when combining multiple exposure technology and self-aligned imaging technology is solved, and the uniformity and yield of the pattern after etching are improved.
Patent Information
- Application Number
- CN202410660520.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-05-24
- Publication Date
- 2025-11-25
AI Technical Summary
When existing multiple exposure techniques are combined with self-aligned dual or quadruple imaging techniques, there is a problem that pattern density imbalance leads to large differences in pattern size after etching, which can cause device failure.
By determining the pattern error of the mask pattern in the mask, the target pattern density calculation window with the greatest correlation with the pattern error is obtained. Machine learning methods are used to determine the target influence error with the greatest impact of the pattern error after etching, and the initial pattern to be corrected is initially corrected to obtain the pattern to be corrected.
It improves the uniformity and yield of the etched pattern, reduces the dimensional deviation of the etched pattern, and improves the accuracy of pattern correction.
Smart Images
Figure CN121008441A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of semiconductor technology, and in particular to a pattern correction method, a storage medium, and a storage terminal. Background Technology
[0002] Different sizes of immersion lithography machines can provide half-pitch resolution for the corresponding size, while smaller sizes require double or even multiple exposure techniques.
[0003] Double or multiple exposure techniques require multiple independent exposures, each producing a portion of the pattern. These patterns are then transferred together onto the substrate via etching. The advantage is that there are no special requirements for the regularity of the design layout, but the disadvantage is the need for two or more independent exposures. In contrast, self-aligned double or quadruple imaging techniques (SADP / SAQP) require only one exposure, using non-photolithography processes to spatially multiply the photolithographic pattern. However, this requires the design layout to conform to certain rules. In actual production, multiple exposure techniques are often combined with self-aligned double or quadruple imaging techniques to meet practical production requirements.
[0004] However, there are still some problems with combining existing multiple exposure techniques with self-aligned dual or quadruple imaging techniques. Summary of the Invention
[0005] The technical problem solved by this invention is to provide a pattern correction method, storage medium, and storage terminal to address the problems existing in the process of combining multiple exposure technology with self-aligned dual or quadruple imaging technology.
[0006] To address the aforementioned technical problems, the present invention provides a graphic correction method, comprising: providing a plurality of initial layouts to be corrected, the initial layouts to be corrected including a plurality of initial graphics to be corrected; obtaining a test layout based on the plurality of initial layouts to be corrected, the test layout including a plurality of test graphics, the positions of the plurality of test graphics corresponding to the positions of the plurality of initial graphics to be corrected; obtaining a mask based on the test layout, the mask including a plurality of mask graphics, the plurality of mask graphics corresponding to the test graphics; determining the graphic error of the mask graphics corresponding to the mask graphics in the mask; obtaining a plurality of graphic density calculation windows corresponding to the test graphics; obtaining a target graphic density calculation window with the highest correlation to the graphic error based on the graphic density of the test graphics in each graphic density calculation window; and performing initial correction on the initial graphics to be corrected in the initial layouts to be corrected using the target graphic density calculation window and the graphic error to obtain the layout to be corrected.
[0007] Optionally, determining the pattern error corresponding to the mask pattern in the mask includes: obtaining several mask pattern errors between the size of several mask patterns and the target value; obtaining several developed pattern errors between the size of several developed patterns formed based on the mask and the target value; obtaining several etched pattern errors between the size of several etched patterns formed based on the mask and the target value; and obtaining several etch errors based on the developed pattern errors and the etched pattern errors.
[0008] Optionally, determining the pattern error corresponding to the mask pattern in the mask also includes: using a machine learning method to determine the target influence error that has the greatest impact on the etched pattern error from the mask pattern error, the developed pattern error, and the etch error; the initial correction of the initial pattern to be corrected in the initial pattern to be corrected in the initial pattern to be corrected using the target pattern density calculation window and the pattern error includes: initial correction of the initial pattern to be corrected in the initial pattern to be corrected using the target pattern density calculation window and the target influence error.
[0009] Optionally, a machine learning method is used to determine the target influence error that has the greatest impact on the etched pattern error from the mask pattern error, the developed pattern error, and the etching error. This includes using a random forest model, taking the etched pattern error as the target value, and the mask pattern error, the developed pattern error, and the etching error as feature vectors, and obtaining the influence error with the largest weight through machine learning as the target influence error, wherein the target influence error has the greatest impact on the etched pattern error.
[0010] Optionally, obtaining several post-development pattern errors between the dimensions of several developed patterns formed based on the photomask and the target value includes: providing a substrate and a photoresist layer located on the substrate; exposing and developing the photoresist layer according to the photomask to obtain a photoresist pattern, wherein the photoresist pattern is the developed pattern; and obtaining several post-development pattern errors between the dimensions of the photoresist pattern and the target value.
[0011] Optionally, obtaining several etched pattern errors between the dimensions of several etched patterns formed based on the photomask and the target value includes: etching the substrate using the photoresist pattern as a mask to form etched patterns on the substrate; and obtaining several etched pattern errors between the dimensions of the etched patterns and the target value.
[0012] Optionally, several etching errors are obtained based on several post-development pattern errors and several post-etching pattern errors, including: obtaining the difference between the post-development pattern error and the post-etching pattern error, wherein the difference is the etching error.
[0013] Optionally, based on the graphic density of the test graphic within each graphic density calculation window, a target graphic density calculation window with the highest correlation to the graphic error is obtained, including: using several graphic densities as feature vectors and the graphic error as the target value, machine learning is used to obtain correlation coefficients between several graphic densities and the graphic error; obtaining the graphic density with the highest correlation coefficient; and obtaining the corresponding target graphic density calculation window based on the graphic density with the highest correlation coefficient.
[0014] Optionally, the initial correction of the initial graphic to be corrected in the initial layout to be corrected using the target graphic density calculation window and graphic error includes: dividing the initial layout to be corrected by the area of the target window to obtain several regions to be corrected; obtaining the graphic density of the initial graphic to be corrected in each sampling point in each region to be corrected, wherein the graphic density of the initial graphic to be corrected is the ratio between the area of the initial graphic to be corrected in the sampling point and the area of the region to be corrected; dividing the graphic densities of the several initial graphics to be corrected into several segments in ascending order; and using the corresponding graphic error to perform initial correction on the initial graphic to be corrected in each segment to obtain the layout to be corrected.
[0015] Optionally, the initial correction of the initial graphic to be corrected in each segment is performed using the corresponding graphic error to obtain the layout to be corrected, including: segmenting and compensating the key dimensions of the initial graphic to be corrected in the initial layout to be corrected using the target graphic density calculation window and the graphic error.
[0016] Optionally, the graphic density is the ratio of the area of the test graphic within the graphic density calculation window to the area of the graphic density calculation window.
[0017] Optionally, it also includes: performing optical proximity effect correction on the pattern to be corrected.
[0018] Accordingly, the present invention also provides a storage medium storing computer instructions, which execute the steps of the above method when the computer instructions are run.
[0019] Accordingly, the present invention also provides a storage terminal, including a memory and a processor, wherein the memory stores computer instructions that can be executed on the processor, and the processor executes the steps of the above method when executing the computer instructions.
[0020] Compared with the prior art, the technical solution of the present invention has the following beneficial effects:
[0021] The pattern correction method of the present invention determines the pattern error corresponding to the mask pattern in the mask, obtains a target pattern density calculation window with the highest correlation to the pattern error, and uses the target pattern density calculation window and the pattern error to perform initial correction on the initial pattern to be corrected in the initial pattern to be corrected, thereby obtaining the pattern to be corrected. This results in a smaller deviation between the size of the etched pattern and the target size after optical proximity effect correction, thus improving the uniformity and yield of the etched pattern. Attached Figure Description
[0022] Figures 1 to 5 This is a flowchart illustrating the graphic correction method in an embodiment of the present invention;
[0023] Figures 6 to 8 This is a schematic diagram of the graphic correction process in an embodiment of the present invention;
[0024] Figure 9 This is a schematic diagram illustrating the segmented compensation of key dimensions of the initial graphic to be corrected in the initial layout to be corrected, as described in an embodiment of the present invention. Detailed Implementation
[0025] As described in the background section, there are still some problems with the process of combining existing multiple exposure techniques with self-aligned dual or quadruple imaging techniques.
[0026] Specifically, when using a combination of double exposure and self-aligned double or quadruple imaging techniques to fabricate design layouts with a half-pitch resolution of less than 36–40 nm, the mask needs to be split into three layers. Two masks (mask1 and mask2) are used for double exposure, and one mask (mask3) is used to form the target pattern, such as a metal line pattern, using self-aligned double or quadruple imaging. In this case, the pattern area on mask3 used for self-aligned double or quadruple imaging has a densely packed, regular pattern, while the non-self-aligned double or quadruple imaging area on mask3 is completely blank. This severe imbalance in pattern density on mask3 causes significant differences in the same critical dimension (CD) at different locations within the pattern area for self-aligned double or quadruple imaging on the actual wafer. This leads to the pattern on the wafer deviating from the target size, resulting in device failure. For example, metal lines deviating from the target size are prone to defects and breakage.
[0027] To address the aforementioned problems, the present invention provides a pattern correction method, a storage medium, and a storage terminal. By determining the pattern error corresponding to the mask pattern in the mask, a target pattern density calculation window with the highest correlation to the pattern error is obtained. The initial pattern to be corrected in the initial pattern to be corrected is then corrected using the target pattern density calculation window and the pattern error, resulting in a pattern to be corrected. This ensures that after optical proximity effect correction, the size deviation between the etched pattern and the target size is small, improving the uniformity and yield of the etched pattern.
[0028] To make the above-mentioned objectives, features and beneficial effects of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.
[0029] Figures 1 to 5 This is a flowchart illustrating the graphic correction method in an embodiment of the present invention.
[0030] Please refer to Figure 1 The graphic correction method includes:
[0031] Step S10: Provide several initial layouts to be corrected, wherein the initial layouts to be corrected include several initial graphics to be corrected;
[0032] Step S20: Obtain a test layout based on several initial layouts to be corrected, wherein the test layout includes several test patterns, and the positions of the test patterns correspond to the positions of the initial layouts to be corrected.
[0033] Step S30: Obtain a mask based on the test pattern, wherein the mask includes a plurality of mask patterns, and the plurality of mask patterns correspond to the test pattern;
[0034] Step S40: Determine the pattern error corresponding to the mask pattern in the mask;
[0035] Step S50: Obtain several graphic density calculation windows corresponding to the test graphic;
[0036] Step S60: Based on the graphic density of the test graphic within each graphic density calculation window, obtain the target graphic density calculation window that has the highest correlation with the graphic error;
[0037] Step S70: Perform initial correction on the initial graphic to be corrected in the initial layout to be corrected using the target graphic density calculation window and the graphic error, and obtain the layout to be corrected.
[0038] The pattern correction method determines the pattern error corresponding to the mask pattern in the mask, obtains the target pattern density calculation window with the highest correlation to the pattern error, and uses the target pattern density calculation window and the pattern error to perform initial correction on the initial pattern to be corrected in the initial pattern to be corrected, thereby obtaining the pattern to be corrected. This ensures that after optical proximity effect correction, the size deviation between the etched pattern and the target size is small, improving the uniformity and yield of the etched pattern.
[0039] Next, each step will be analyzed and explained.
[0040] Please continue to refer to this. Figure 1 Step S10: Provide several initial layouts to be corrected, the initial layouts to be corrected including several initial graphics to be corrected.
[0041] The initial pattern to be corrected is used to transfer the image onto the mask after recalibration and optical proximity correction. The image in the initial pattern to be corrected is the image used in subsequent self-aligned dual or quadruple imaging techniques.
[0042] Please refer to Figure 6 , Figure 6 This diagram illustrates an initial layout 100 to be corrected, which includes a graphics area ( Figure 6 The blue area and the blank area are defined as a pattern area containing several initial patterns to be corrected. The blank area is a pattern area formed on the wafer by other photomasks.
[0043] Please continue to refer to this. Figure 1 Step S20: Obtain a test layout based on several initial layouts to be corrected. The test layout includes several test patterns, and the positions of the test patterns correspond to the positions of the initial layouts to be corrected.
[0044] Obtaining a test layout based on several initial layouts to be corrected includes: obtaining a test layout based on the distribution pattern of several initial graphics to be corrected on several initial layouts to be corrected, wherein the several test graphics on the test layout include the initial graphics to be corrected on several initial layouts to be corrected.
[0045] In this embodiment, the shape of the test pattern includes rectangles, circles, ellipses, or other irregular shapes.
[0046] Please refer to Figure 7 , Figure 7 The test layout 200 shown includes a graphic area 201 (yellow area) and a blank area, with the test graphic located in the graphic area 201; please refer to... Figure 8 , Figure 8 for Figure 7 An enlarged schematic diagram of the graphic area 201, which includes several test graphics 202.
[0047] Please continue to refer to this. Figure 1 Step S30: Obtain a mask based on the test pattern 200. The mask includes several mask patterns, and the several mask patterns correspond to the test pattern 202.
[0048] Obtaining a mask based on the test layout 200 includes: transferring the test pattern 202 on the test layout 200 onto the mask, wherein the mask includes a plurality of mask patterns.
[0049] Please continue to refer to this. Figure 1 Step S40: Determine the graphic error corresponding to the mask pattern in the mask.
[0050] Please refer to Figure 2 In this embodiment, determining the pattern error corresponding to the mask pattern in the mask includes:
[0051] Step S401: Obtain several mask pattern errors between the dimensions of several mask patterns and the target values;
[0052] Step S402: Obtain several developed pattern errors between the dimensions of several developed patterns formed based on the mask and the target value;
[0053] Step S403: Obtain several etching pattern errors between the dimensions of several etched patterns formed based on the mask and the target value;
[0054] Step S404: Based on the pattern error after development and the pattern error after etching, obtain several etching errors;
[0055] Step S405: Using machine learning methods, determine the target influence error that has the greatest impact on the post-etching pattern error from the mask pattern error, the post-development pattern error, and the etching error.
[0056] Please continue to refer to this. Figure 2 Step S401: Obtain several mask pattern errors between the dimensions of several mask patterns and the target value.
[0057] The size of the pattern on the mask has a target value, and the difference between the size of the mask pattern and the target value is the mask pattern error. The target value is a preset size of the pattern on the mask, set according to actual design requirements.
[0058] Please continue to refer to this. Figure 2Step S402: Obtain several developed pattern errors between the size of several developed patterns formed based on the mask and the target value.
[0059] Please refer to Figure 3 In this embodiment, obtaining several developed pattern errors between the dimensions of several developed patterns formed based on the mask and the target value includes:
[0060] Step S4021: Provide a substrate and a photoresist layer on the substrate;
[0061] Step S4022: Expose and develop the photoresist layer according to the photomask to obtain a photoresist pattern, wherein the photoresist pattern is the developed pattern;
[0062] Step S4023: Obtain several post-development pattern errors between the size of the photoresist pattern and the target value.
[0063] The developed pattern after developing and exposing the photoresist layer has a target value, and the difference between the size of the photoresist pattern and the target value is the developed pattern error. The target value is a preset pattern size on the photoresist layer, set according to actual design requirements.
[0064] Please continue to refer to this. Figure 2 Step S403: Obtain several etching pattern errors between the dimensions of several etched patterns formed based on the mask and the target value.
[0065] Obtaining several etched pattern errors between the dimensions of several etched patterns formed based on the photomask and the target value includes: etching the substrate using the photoresist pattern as a mask to form etched patterns on the substrate; and obtaining several etched pattern errors between the dimensions of the etched patterns and the target value.
[0066] The etched pattern formed on the substrate has a target size, and the difference between the etched pattern size and the target size is the etched pattern error. The target size is a preset etched pattern size, set according to actual design requirements.
[0067] Please continue to refer to this. Figure 2 Step S404: Obtain several etching errors based on several post-development pattern errors and several post-etching pattern errors.
[0068] A plurality of etching errors are obtained based on a plurality of post-development pattern errors and a plurality of post-etching pattern errors, including: obtaining the difference between the post-development pattern errors and the post-etching pattern errors, wherein the difference is the etching error.
[0069] The pattern errors include: mask pattern error, post-development pattern error, post-etching pattern error, and etching error.
[0070] Please continue to refer to this. Figure 2 Step S405: Using machine learning methods, determine the target influence error that has the greatest impact on the etched pattern error from the mask pattern error, the developed pattern error, and the etching error.
[0071] Using machine learning methods, the target influence error that has the greatest impact on the etched pattern error is determined from the mask pattern error, the developed pattern error, and the etching error. This includes: using a random forest model, taking the etched pattern error as the target value, and using the mask pattern error, the developed pattern error, and the etching error as feature vectors, the influence error with the largest weight is obtained through machine learning as the target influence error, and the target influence error has the greatest impact on the etched pattern error.
[0072] The target influence error is one of the following: mask pattern error, post-development pattern error, and etching error.
[0073] Please continue to refer to this. Figure 1 Step S50: Obtain several graphic density calculation windows corresponding to the test graphic.
[0074] The density window is a region divided when performing OPC correction on the test layout. A test layout includes several density windows, and the test graphics in multiple density windows are corrected for OPC respectively.
[0075] In this embodiment, the areas of the various graphics density calculation windows are different.
[0076] Please continue to refer to this. Figure 1 Step S60: Based on the graphic density of the test graphic within each graphic density calculation window, obtain the target graphic density calculation window that has the highest correlation with the graphic error.
[0077] The graphic density is the ratio of the area of the test graphic within the graphic density calculation window to the area of the graphic density calculation window.
[0078] Please refer to Figure 4 In this embodiment, based on the graphic density of the test graphic within each graphic density calculation window, a target graphic density calculation window with the highest correlation to the graphic error is obtained, including:
[0079] Step S 601: Using several of the aforementioned graphic densities as feature vectors and the graphic errors as target values, machine learning is employed to obtain the correlation coefficients between the aforementioned graphic densities and the graphic errors;
[0080] Step S 602: Obtain the density of the graph with the highest correlation coefficient;
[0081] Step S603: Obtain the corresponding target graphic density calculation window based on the graphic density with the highest correlation coefficient.
[0082] The correlation is the Pearson correlation coefficient. The target graphic density calculation window has the highest correlation with the graphic error, meaning that the target graphic density calculation window has the greatest impact on the graphic error. Graphic correction performed under this target graphic density calculation window has a better effect.
[0083] The pattern errors include: mask pattern error, post-development pattern error, post-etching pattern error, and etching error.
[0084] In this embodiment, machine learning is used to obtain correlation coefficients between several of the pattern densities and the pattern errors, including: using machine learning to obtain correlation coefficients between several of the pattern densities and mask pattern errors, post-development pattern errors, post-etching pattern errors, and etching errors.
[0085] In other embodiments, machine learning is employed to obtain correlation coefficients between several of the aforementioned pattern densities and one or more of the mask pattern error, post-development pattern error, post-etching pattern error, and etching error.
[0086] Please continue to refer to this. Figure 1 Step S70: Perform initial correction on the initial graphic to be corrected in the initial layout to be corrected using the target graphic density calculation window and the graphic error, and obtain the layout to be corrected.
[0087] In this embodiment, the initial correction of the initial graphic to be corrected in the initial layout to be corrected using the target graphic density calculation window and the graphic error includes: initial correction of the initial graphic to be corrected in the initial layout to be corrected using the target graphic density calculation window and the target influence error.
[0088] The target influence error is one of the following: mask pattern error, post-development pattern error, and etching error.
[0089] Please refer to Figure 5 In this embodiment, the initial correction of the initial graphic to be corrected in the initial layout to be corrected is performed using the target graphic density calculation window and graphic error, including:
[0090] Step S701: Divide the initial layout to be corrected into several regions based on the area of the target window;
[0091] Step S702: Obtain the initial graphic density of the graphic to be corrected in each sampling point of each region to be corrected, wherein the graphic density of the initial graphic to be corrected is the ratio between the area of the initial graphic to be corrected in the sampling point and the area of the region to be corrected.
[0092] Step S703: Divide the graphic density of the initial graphics to be corrected into several segments in ascending order;
[0093] Step S704: Use the corresponding graphic error to perform initial correction on the initial graphic to be corrected in each segment, and obtain the layout to be corrected.
[0094] In this embodiment, the initial correction of the initial graphic to be corrected in each segment is performed using the corresponding graphic error to obtain the layout to be corrected, including: segmenting compensation of the key dimensions of the initial graphic to be corrected in the initial layout to be corrected using the target graphic density calculation window and the graphic error.
[0095] In this embodiment, the layout to be corrected includes a plurality of graphics to be corrected, and the plurality of graphics to be corrected correspond to the initial graphics to be corrected of the plurality of initial graphics to be corrected.
[0096] Please combine Figure 9 Continue to refer to Figure 5 , Figure 9 The density of the graphic is divided into several intervals in ascending order, with nodes including 0, X1, X2, X3, ..., Xn. A represents the initial size of the graphic to be corrected. The target influence error values a1, a2, a3, ..., an are assigned to the initial graphic to be corrected corresponding to the density of the graphic in each interval. n The size of the compensated graphic to be corrected is A+a n .
[0097] Source of compensation value: Compensation value bias = R * AEI_error, where AEI_error is the pattern error after etching. The responsivity (R) refers to the change in the pattern (AEI) after etching when the size of the pattern (ADI) changes by 1 nm, i.e., R = d(ADI) / d(AEI).
[0098] The described pattern correction method uses a machine learning approach to identify the target influence error that has the greatest impact on the etched pattern error from among several influencing errors. It then calculates a target pattern density window based on the pattern density, finding the window with the highest correlation to the mask pattern error, the post-development pattern error, the post-etching pattern error, and the etching error. Finally, it performs segmented compensation on the key dimensions of the initial pattern to be corrected based on the target pattern density calculation window and the target influence error. This allows the influencing factors on the etched pattern to be transformed into quantitatively describable data. Furthermore, by segmenting the key dimensions of the initial pattern to be corrected within the initial pattern to be corrected, the size deviation between the etched pattern and the target size is smaller after optical proximity effect correction, thus improving the uniformity and yield of the etched pattern.
[0099] In this embodiment, the graphic correction method further includes: performing optical proximity effect correction on the graphic to be corrected.
[0100] The uniformity and yield of the etched pattern are improved after optical proximity effect correction.
[0101] Accordingly, embodiments of the present invention also provide a storage medium storing computer instructions thereon, which are executed when the computer instructions are run. Figures 1 to 5 The steps of the method are described.
[0102] Accordingly, embodiments of the present invention also provide a storage terminal, including a memory and a processor, wherein the memory stores computer instructions that can be executed on the processor, and the processor executes the computer instructions. Figures 1 to 5 The steps of the method are described.
[0103] While the present invention has been disclosed above, it is not limited thereto. Any person skilled in the art can make various modifications and alterations without departing from the spirit and scope of the invention; therefore, the scope of protection of the present invention should be determined by the scope defined in the claims.
Claims
1. A method for correcting graphics, characterized in that, include: Several initial layouts to be corrected are provided, wherein the initial layouts to be corrected include several initial graphics to be corrected; A test layout is obtained based on several initial layouts to be corrected, wherein the test layout includes several test patterns, and the positions of the test patterns correspond to the positions of the initial layouts to be corrected. A mask is obtained based on the test pattern, the mask comprising a plurality of mask patterns, the plurality of mask patterns corresponding to the test pattern; Determine the pattern error corresponding to the mask pattern in the aforementioned mask; Obtain several graphic density calculation windows corresponding to the test graphic; Based on the graphic density of the test graphic within each graphic density calculation window, obtain the target graphic density calculation window that has the highest correlation with the graphic error; The initial pattern to be corrected is initially corrected using the target pattern density calculation window and the pattern error to obtain the pattern to be corrected.
2. The graphic correction method as described in claim 1, characterized in that, The determination of the graphic error corresponding to the mask pattern in the mask plate includes: Obtain several mask pattern errors between the dimensions of several mask patterns and the target value; Obtain several developed pattern errors between the dimensions of several developed patterns formed based on the mask and the target value; Obtain several etching pattern errors between the dimensions of several etched patterns formed based on the mask and the target value; Based on the pattern error after development and the pattern error after etching, several etching errors are obtained.
3. The graphic correction method as described in claim 2, characterized in that, The determination of the pattern error corresponding to the mask pattern in the mask also includes: using a machine learning method to determine the target influence error that has the greatest impact on the etched pattern error from the mask pattern error, the developed pattern error and the etching error; The initial correction of the initial graphic in the initial layout to be corrected using the target graphic density calculation window and the graphic error includes: initial correction of the initial graphic in the initial layout to be corrected using the target graphic density calculation window and the target influence error.
4. The graphic correction method as described in claim 3, characterized in that, Using machine learning methods, the target influence error that has the greatest impact on the etched pattern error is determined from the mask pattern error, the developed pattern error, and the etching error. This includes: using a random forest model, taking the etched pattern error as the target value, and using the mask pattern error, the developed pattern error, and the etching error as feature vectors, the influence error with the largest weight is obtained through machine learning as the target influence error, and the target influence error has the greatest impact on the etched pattern error.
5. The graphic correction method as described in claim 2, characterized in that, Obtain several developed pattern errors between the dimensions of several developed patterns formed based on the mask and the target value, including: A substrate and a photoresist layer on the substrate are provided; the photoresist layer is exposed and developed according to the photomask to obtain a photoresist pattern, the photoresist pattern being the developed pattern; and several developed pattern errors are obtained between the size of the photoresist pattern and a target value.
6. The graphic correction method as described in claim 5, characterized in that, Obtain several etching pattern errors between the dimensions of several etched patterns formed based on the mask and the target value, including: The substrate is etched using the photoresist pattern as a mask to form an etched pattern on the substrate; several etched pattern errors between the size of the etched pattern and the target value are obtained.
7. The graphic correction method as described in claim 2, characterized in that, A plurality of etching errors are obtained based on a plurality of post-development pattern errors and a plurality of post-etching pattern errors, including: obtaining the difference between the post-development pattern errors and the post-etching pattern errors, wherein the difference is the etching error.
8. The graphic correction method as described in claim 1, characterized in that, Based on the graphic density of the test graphic within each graphic density calculation window, obtain the target graphic density calculation window that has the highest correlation with the graphic error, including: Using several graphic densities as feature vectors and the graphic errors as target values, machine learning is employed to obtain correlation coefficients between the graphic densities and the graphic errors; the graphic density with the largest correlation coefficient is obtained; and a corresponding target graphic density calculation window is obtained based on the graphic density with the largest correlation coefficient.
9. The graphic correction method as described in claim 1, characterized in that, The initial correction of the initial graphic to be corrected in the initial layout to be corrected, based on the target graphic density calculation window and graphic error, includes: dividing the initial layout to be corrected into several regions to be corrected by the area of the target window; obtaining the graphic density of the initial graphic to be corrected in each sampling point of each region to be corrected, wherein the graphic density of the initial graphic to be corrected is the ratio between the area of the initial graphic to be corrected in the sampling point and the area of the region to be corrected; dividing the graphic densities of the initial graphic to be corrected into several segments in ascending order; and using the corresponding graphic error to perform initial correction on the initial graphic to be corrected in each segment to obtain the layout to be corrected.
10. The graphic correction method as described in claim 9, characterized in that, The initial correction of the initial graphic to be corrected in each segment is performed using the corresponding graphic error, to obtain the layout to be corrected, including: The key dimensions of the initial graphic to be corrected in the initial layout are segmented and compensated using the target graphic density calculation window and the graphic error.
11. The graphic correction method as described in claim 1, characterized in that, The graphic density is the ratio of the area of the test graphic within the graphic density calculation window to the area of the graphic density calculation window.
12. The graphic correction method as described in claim 1, characterized in that, Also includes: Optical proximity effect correction is performed on the pattern to be corrected.
13. A storage medium storing computer instructions thereon, characterized in that, When the computer instructions are executed, they perform the steps of the method according to any one of claims 1 to 12.
14. A storage terminal, comprising a memory and a processor, wherein the memory stores computer instructions capable of running on the processor, characterized in that, When the processor executes the computer instructions, it performs the steps of the method according to any one of claims 1 to 12.