A method and apparatus for optical proximity correction

Abstract

The present application relates to the field of integrated circuit manufacturing, in particular to an optical proximity correction method and device, by receiving a target wafer pattern; determining a target wafer grid image according to the target wafer pattern according to a preset grid size; taking the target wafer pattern as a current mask pattern; inputting the current mask pattern into a pre-trained three-dimensional optical proximity correction model, and assisting with the grid size to obtain a simulation wafer grid image; calculating the edge height error corresponding to each edge segment; judging whether the edge height error exceeds a preset adjustment threshold; when the edge height error exceeds the preset adjustment threshold, adjusting the position of each edge segment in the corresponding normal direction according to the edge height error to obtain an updated current mask pattern; repeating steps four to seven until the edge height error does not exceed the adjustment threshold. The present application reduces defects such as photoresist height loss and improves product yield.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of integrated circuit manufacturing, and in particular to an optical proximity correction method and device. BACKGROUND

[0002] In semiconductor manufacturing, OPC (optical proximity correction) correction verification, i.e. comparison of the difference between the simulated profile and the target pattern, is an indispensable step. However, the optical proximity correction technology in the prior art is based on a two-dimensional target wafer pattern, i.e. the mask pattern is adjusted so that the wafer profile after the simulation of the lithography (or etching) process on a preset z plane is consistent with the target wafer pattern. However, the actual wafer is a three-dimensional structure, and the correction based on the two-dimensional target wafer pattern cannot guarantee the matching of the three-dimensional wafer topography, and the mismatch often leads to process defects such as top loss of photoresist, as shown in Figs. 1(a) and 1(b). For the target wafer pattern shown at the dashed interface in Fig. 1(a), the critical dimension of the wafer height z0 on the wafer after lithography is basically consistent with the target pattern, as shown in Fig. 1(b). However, for the heights other than the wafer height z0, there is a significant photoresist height loss at positions A and B in Fig. 1(a), i.e. there is a significant difference between the three-dimensional topography of the wafer after lithography and the target three-dimensional topography.

[0003] Therefore, how to reduce the pattern lithography and etching defects caused by photoresist thickness loss and fill the gap that the current optical proximity correction cannot guarantee the correction accuracy in the thickness direction of the photoresist is a problem to be solved by those skilled in the art. SUMMARY

[0004] The purpose of the present application is to provide an optical proximity correction method and device to solve the problem that the optical proximity correction in the prior art cannot be corrected in the thickness direction of the photoresist, resulting in a large number of defects caused by photoresist thickness loss.

[0005] To solve the above technical problems, the present application provides an optical proximity correction method, comprising:

[0006] Step one, receiving a target wafer pattern; the target wafer pattern is surrounded by a plurality of edge segments;

[0007] Step two, determining a target wafer grid image according to the target wafer pattern according to a preset grid size; each grid in the target wafer grid image records a corresponding height value;

[0008] Step three, taking the target wafer pattern as a current mask pattern;

[0009] Step four, inputting the current mask pattern into a pre-trained three-dimensional optical proximity correction model and assisting with the grid size to obtain a simulation wafer grid image; each grid in the simulation wafer grid image records a corresponding height value;

[0010] Step five, calculating an edge height error corresponding to each edge segment by comparing the simulation wafer grid image with the target wafer grid image;

[0011] Step six, judging whether the edge height error exceeds a preset adjustment threshold;

[0012] Step seven, when the edge height error exceeds the preset adjustment threshold, adjusting the position of each edge segment in a corresponding normal direction according to the edge height error to obtain an updated current mask pattern;

[0013] Step eight, repeating steps four to seven until the edge height error does not exceed the adjustment threshold, and outputting the current mask pattern as a target mask pattern.

[0014] Optionally, in the optical proximity correction method, the calculating of the edge height error corresponding to each edge segment by comparing the simulation wafer grid image with the target wafer grid image comprises:

[0015] The edge segment comprises a control point, and the control point is located at the center of the edge segment; the control point corresponds to a control point coordinate grid, and two grids adjacent to the control point coordinate grid in a normal direction corresponding to the edge segment are taken as calculation domain grids;

[0016] The height difference between the height value of the control point coordinate grid in the simulation wafer grid image and the height value of the control point coordinate grid in the target wafer grid image is calculated; the height difference between the height value of each of the two calculation domain grids in the simulation wafer grid image and the height value of the calculation domain grid in the target wafer grid image is calculated;

[0017] The three height differences are added to obtain the edge height error corresponding to the edge segment.

[0018] Optionally, in the optical proximity correction method, the adjusting of the position of each edge segment in a corresponding normal direction according to the edge height error comprises:

[0019] When the edge height error indicates that the height value of a grid in the simulation wafer grid image is higher than the height value of a grid in the target wafer grid image, the edge segment is moved to the inside of the pattern along the corresponding normal direction;

[0020] When the edge height error indicates that the height value of the grid in the simulation wafer grid image is lower than the height value of the grid in the target wafer grid image, the edge segment is moved outward along the corresponding normal direction.

[0021] Optionally, in the optical proximity correction method, the moving distance of the edge segment during the adjustment in the corresponding normal direction is determined by the following formula:

[0022] L = -f x (EHE / MF) ;

[0023] wherein, L is the moving distance of the edge segment, f is a feedback coefficient, MF is a mask error magnification coefficient corresponding to the edge segment, and EHE is the edge height error of the edge segment.

[0024] Optionally, in the optical proximity correction method, after judging whether the edge height error exceeds a preset adjustment threshold, the method further comprises:

[0025] judging whether the update times of the current mask pattern exceed a preset tolerance threshold;

[0026] when the update times of the current mask pattern exceed the tolerance threshold, outputting the current mask pattern as a target mask pattern.

[0027] Optionally, in the optical proximity correction method, the height value of the grid where the pattern edge is located in the wafer grid image is calculated by the following formula:

[0028] O = r x O1 + (1-r) x O2;

[0029] wherein, O is the height value of the grid where the pattern edge is located, O2 is the height value of the pattern covered area, O1 is the height value of the pattern uncovered area, and r is the area ratio of the pattern uncovered area to the grid.

[0030] Optionally, in the optical proximity correction method, the training method of the three-dimensional optical proximity correction model comprises:

[0031] establishing an uncalibrated three-dimensional optical proximity correction model; the three-dimensional optical proximity correction model can first determine a light intensity distribution image according to an inputted to-be-processed pattern through a preset cross-transmission coefficient, and then calculate a photoresist height image corresponding to the light intensity distribution image through a preset photoresist height model term and a corresponding linear coefficient;

[0032] according to a preset set of modeling patterns, manufacturing a test mask and obtaining an acquired photoresist height image corresponding to the test mask;

[0033] The cross-transfer coefficient, the photoresist height model term and the linear coefficient are calibrated using the modeling pattern set and the corresponding test three-dimensional topography image, so that a difference between a simulated photoresist height image obtained by inputting the modeling pattern set as a to-be-processed pattern into the three-dimensional optical proximity correction model and the acquired photoresist height image is less than a preset calibration threshold.

[0034] Optionally, in the optical proximity correction method, the three-dimensional optical proximity correction model obtains the light intensity distribution image by the following formula:

[0035] ;

[0036] wherein I(x, y) is the light intensity distribution image, TCC is the cross-transfer coefficient, is a convolution operation, and M(x, y) is a rasterized to-be-processed pattern;

[0037] The three-dimensional optical proximity correction model obtains the photoresist height image by the following formula:

[0038] ;

[0039] wherein H(x, y) is the photoresist height image, F j is the jth photoresist height model term, c j is the linear coefficient corresponding to the jth photoresist height model term, and I(x, y) is the light intensity distribution image.

[0040] Optionally, in the optical proximity correction method, when a loss function value of the following formula is less than a preset calibration threshold, it is considered that a difference between a simulated photoresist height image obtained by inputting the modeling pattern set as a to-be-processed pattern into the three-dimensional optical proximity correction model and the acquired photoresist height image is less than a preset calibration threshold:

[0041] ;

[0042] wherein Cost is the loss function value; H ik is a height value of the kth grid in the ith modeling pattern in the modeling pattern set in the simulated photoresist height image, h ik is a height value of the kth grid in the ith modeling pattern in the acquired photoresist height image, w i is a weight coefficient of the ith modeling pattern in the modeling pattern set, w k is a weight coefficient of the kth grid of the ith modeling pattern in the modeling pattern set.

[0043] An optical proximity correction device comprises:

[0044] receiving a target wafer pattern, wherein the target wafer pattern is surrounded by a plurality of edge segments;

[0045] determining a target wafer grid image according to the target wafer pattern and a preset grid size, wherein each grid in the target wafer grid image records a corresponding height value;

[0046] using the target wafer pattern as a current mask pattern;

[0047] inputting the current mask pattern into a pre-trained three-dimensional optical proximity correction model and obtaining a simulated wafer grid image according to the grid size, wherein each grid in the simulated wafer grid image records a corresponding height value;

[0048] calculating edge height errors of each of the edge segments in the simulated wafer grid image compared with the target wafer grid image;

[0049] judging whether the edge height errors exceed a preset adjustment threshold;

[0050] adjusting positions of each of the edge segments in a corresponding normal direction according to the edge height errors when the edge height errors exceed the preset adjustment threshold, and obtaining an updated current mask pattern;

[0051] repeating steps four to seven until the edge height errors do not exceed the adjustment threshold, and outputting the current mask pattern as a target mask pattern.

[0052] The optical proximity correction method provided by the present application comprises the following steps: step one, receiving a target wafer pattern; the target wafer pattern is surrounded by a plurality of edge segments; step two, determining a target wafer grid image according to the target wafer pattern and a preset grid size; each grid in the target wafer grid image records a corresponding height value; step three, taking the target wafer pattern as a current mask pattern; step four, inputting the current mask pattern into a pre-trained three-dimensional optical proximity correction model and assisting with the grid size to obtain a simulation wafer grid image; each grid in the simulation wafer grid image records a corresponding height value; step five, calculating edge height errors corresponding to each edge segment in the simulation wafer grid image compared with the target wafer grid image; step six, judging whether the edge height errors exceed a preset adjustment threshold; step seven, when the edge height errors exceed the preset adjustment threshold, adjusting the positions of each edge segment in the corresponding normal direction according to the edge height errors to obtain an updated current mask pattern; and step eight, repeating steps four to seven until the edge height errors do not exceed the adjustment threshold, and outputting the current mask pattern as a target mask pattern.

[0053] In the present application, the difference between the simulated simulation wafer grid image and the target wafer grid image is not measured from a specific height plane and corrected, but the images are divided into grids, so that the height values of the grids at the same position in the simulation image and the target image are as consistent as possible, thereby greatly improving the accuracy of the optical proximity correction in the thickness direction of the photoresist, reducing potential defects such as photoresist height loss, and improving the yield of finished products. The present application also provides an optical proximity correction device with the above beneficial effects. BRIEF DESCRIPTION OF DRAWINGS

[0054] In order to more clearly illustrate the technical solutions of the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiment or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and those skilled in the art can also obtain other drawings according to these drawings without creative labor.

[0055] Fig. 1(a) is a schematic diagram of a specific surface of a photoresist pattern after optical proximity correction in the prior art;

[0056] Fig. 1(b) is a schematic diagram of a cross section of the photoresist pattern in the thickness direction after optical proximity correction in the prior art;

[0057] Figure 2 Fig. 1(b) is a schematic diagram of a cross section of the photoresist pattern in the thickness direction after optical proximity correction in the prior art;

[0058] Figure 3 A schematic diagram of a grid for calculating edge height error corresponding to a single edge segment of one embodiment of the optical proximity correction method provided by the present application;

[0059] Figure 4 A flowchart of the training method of the three-dimensional optical proximity correction model provided by the present application;

[0060] Figure 5 A structural schematic diagram of one embodiment of the optical proximity correction device provided by the present application.

[0061] Reference signs:

[0062] 100 - receiving module; 200 - gridding module; 300 - mask determining module; 400 - simulation module; 500 - height difference module; 600 - judging module; 700 - adjusting module; 800 - circulating module. DETAILED DESCRIPTION

[0063] In order to make the personnel in the technical field better understand the present application scheme, the present application will be further described in detail below in combination with the drawings and specific embodiments. Obviously, the described embodiments are only some of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by the person skilled in the art without creative labor belong to the scope of protection of the present application.

[0064] The core of the present application is to provide an optical proximity correction method, a flowchart of one embodiment of which is shown as Figure 2 The present application is called embodiment one, which includes:

[0065] S101: Step one, receiving a target wafer pattern; the target wafer pattern is surrounded by a plurality of edge segments.

[0066] The target wafer pattern is the pattern that needs to be generated on the wafer; the edge segment can be moved along the corresponding normal direction in the subsequent steps to adjust the edge position of the pattern.

[0067] S102: Step two, determining a target wafer grid image according to the target wafer pattern according to a preset grid size; each grid in the target wafer grid image records a corresponding height value.

[0068] S103: Step three, taking the target wafer pattern as a current mask pattern.

[0069] S104: Step four, input the current mask pattern into the pre-trained three-dimensional optical proximity correction model, and assist with the grid size to obtain a simulated wafer grid image; each grid in the simulated wafer grid image records a corresponding height value.

[0070] Of course, the three-dimensional optical proximity correction model can output a simulated wafer pattern, and then the simulated wafer pattern is divided by the grid size to obtain the simulated wafer grid image, or the three-dimensional optical proximity correction model can directly output the simulated wafer grid image by inputting the previous mask pattern and the grid size together.

[0071] S105: Step five, calculate the edge height error corresponding to each edge segment by comparing the simulated wafer grid image with the target wafer grid image.

[0072] As a preferred embodiment, this step includes:

[0073] A1: The edge segment includes a control point, and the control point is located at the center of the edge segment; the grid corresponding to the control point is taken as a control point coordinate grid, and two grids adjacent to the control point coordinate grid in the normal direction corresponding to the edge segment are taken as calculation domain grids.

[0074] Reference can be made to Figure 3 , Figure 3 A schematic diagram of a control point coordinate grid and two calculation domain grids corresponding to a single edge segment in a specific embodiment is given by way of illustration, in which the normal line passing through the control point is indicated by a dashed line.

[0075] A2: Calculate the height difference between the height value of the control point coordinate grid in the simulated wafer grid image and the height value in the target wafer grid image; and calculate the height difference between the height value of each of the two calculation domain grids in the simulated wafer grid image and the height value in the target wafer grid image.

[0076] This step finally obtains three height differences, respectively corresponding to the height value differences of the above three grids in the two wafer grid images.

[0077] A3: Add the three height differences to obtain the edge height error corresponding to the edge segment.

[0078] That is, the edge height error corresponding to the edge segment is the sum of the three height differences, and the corresponding edge height error can be obtained by comprehensively referring to the height differences of the above three grids. Specifically, the edge height error can be obtained by the following formula (1):

[0079] ; (1)

[0080] wherein, EHE is the edge height error, i=1, 2, 3 respectively correspond to the case of the grid being a control point coordinate grid and two calculation domain grids, H i is the height value of the corresponding grid in the simulation wafer grid image, H' i is the height value of the corresponding grid in the target wafer grid image.

[0081] Directly taking the sum of the height differences of the three grids as the edge height error can greatly reduce the required computing power and improve the calculation efficiency, and provide an evaluation standard for the change of the pattern similarity before and after the edge segment moves. Of course, the average value of the height differences of the three grids can also be used, and the present application is not limited thereto.

[0082] Further, different weights can be assigned to the three grids, thereby further improving the degree of freedom and accuracy of adjustment. The corresponding formula is shown in the following formula (2):

[0083] ; (2)

[0084] Referring to formula (1), wherein, q i is the weight coefficient of the corresponding grid (i.e. one control point coordinate grid and two calculation domain grids).

[0085] S106: Step six, judging whether the edge height error exceeds the preset adjustment threshold.

[0086] S107: Step seven, when the edge height error exceeds the preset adjustment threshold, adjusting the position of each edge segment in the corresponding normal direction according to the edge height error to obtain an updated current mask pattern.

[0087] As a specific embodiment, adjusting the position of each edge segment in the corresponding normal direction according to the edge height error includes:

[0088] B1: When the edge height error shows that the height value of the grid in the simulation wafer grid image is higher than the height value of the grid in the target wafer grid image, moving the edge segment to the inside of the pattern along the corresponding normal direction.

[0089] For example, in the formula (1), the height difference is the height value of the grid in the simulation wafer grid image minus the height value of the grid in the target wafer grid image. When the edge height error calculated by the formula (1) is positive, it is considered that the height value of the grid in the simulation wafer grid image is higher than the height value of the grid in the target wafer grid image. If the calculation process of the edge height error is the height value of the grid in the target wafer grid image minus the height value of the grid in the simulation wafer grid image, when the edge height error is negative, it is considered that the height value of the grid in the simulation wafer grid image is higher than the height value of the grid in the target wafer grid image.

[0090] Moving the edge segment to the inside of the pattern will reduce the overall height value of the grid region composed of the control point coordinate grid in the simulation wafer grid image and the calculation domain grid, and reduce the edge height error.

[0091] B2: When the edge height error shows that the height value of the grid in the simulation wafer grid image is lower than the height value of the grid in the target wafer grid image, move the edge segment to the outside of the pattern along the corresponding normal direction.

[0092] The description of this step can refer to the description of step B1, but the situation is opposite, which will not be expanded here. In the preferred embodiment, the coverage area of the pattern, that is, the coverage area of the photoresist, is changed by moving the edge segment along the normal direction. Obviously, the larger the coverage area of the photoresist, the greater the overall thickness on the wafer. Obviously, the smaller the coverage area of the photoresist, the smaller the overall thickness on the wafer. In this way, the height values of each edge segment in the simulation wafer grid image and the height values of each edge segment in the target wafer grid image are more and more close, which can effectively improve the precision and accuracy of optical proximity correction.

[0093] Further, the moving distance of the edge segment in the adjustment process in the corresponding normal direction is determined by the following formula (3):

[0094] L=-f×(EHE / MF) (3)

[0095] Wherein, L is the moving distance of the edge segment, f is the feedback coefficient, MF is the mask error amplification coefficient corresponding to the edge segment, and EHE is the edge height error of the edge segment.

[0096] The mask error amplification coefficient is the change amount of the overall thickness of the wafer (the sum of the thickness of the wafer itself and the thickness of the photoresist) caused by the movement of the edge segment by a unit distance. The formula (3) gives the calculation method of the moving amount of the edge segment, which occupies low computing power and has high adjustment precision of the edge height error.

[0097] S108: Step eight, loop step four to step seven until the edge height error does not exceed the adjustment threshold, output the current mask pattern as the target mask pattern.

[0098] As a preferred embodiment, after judging whether the edge height error exceeds the preset adjustment threshold, it further comprises:

[0099] C1: judging whether the update times of the current mask pattern exceed the preset tolerance threshold.

[0100] C1: when the update times of the current mask pattern exceed the tolerance threshold, output the current mask pattern as the target mask pattern.

[0101] In this preferred embodiment, the current mask pattern will not be updated endlessly due to the substandard edge height error, but the update of the current mask pattern will be stopped after the update times reach the tolerance threshold, and the current mask pattern with the update times is directly output as the target mask pattern, which avoids spending too much time on the calculation of a single target mask pattern and reduces the correction efficiency.

[0102] As another preferred embodiment, the height value of the grid where the pattern edge is located in the wafer grid image is calculated by the following formula (4):

[0103] O = r x O1 + (1-r) x O2; (4)

[0104] Wherein, O is the height value of the grid where the pattern edge is located, O2 is the height value of the pattern coverage area, O1 is the height value of the pattern non-coverage area, and r is the area ratio of the pattern non-coverage area to the grid.

[0105] Whether it is the determination of the target wafer grid image in step two or the determination of the simulation wafer grid image in step four, the height value corresponding to each grid in the wafer grid image needs to be determined, and the height of the grid completely covered by the pattern should be the sum of the thickness of the wafer itself and the thickness of the photoresist (corresponding to O2 above), and conversely, the height of the grid not covered by the pattern should be equal to the thickness of the wafer itself (corresponding to O1 above). However, if a grid is only partially covered by the pattern and the other part is not covered by the pattern (i.e. the grid where the pattern edge is located), its height value cannot be simply equal to either of the above two, and the present preferred embodiment provides a method for calculating the height value of the grid where the pattern edge is located, which is simple in method, low in computing power occupation, high in calculation efficiency, representative, and can improve the accuracy of mask correction.

[0106] The optical proximity correction method provided by the application comprises the following steps: step one, receiving a target wafer pattern; the target wafer pattern is surrounded by a plurality of edge sections; step two, determining a target wafer grid image according to the target wafer pattern according to a preset grid size; each grid in the target wafer grid image records a corresponding height value; step three, taking the target wafer pattern as a current mask pattern; step four, inputting the current mask pattern into a pre-trained three-dimensional optical proximity correction model and assisting with the grid size to obtain a simulation wafer grid image; each grid in the simulation wafer grid image records a corresponding height value; step five, calculating edge height errors corresponding to each edge section in the simulation wafer grid image compared with the target wafer grid image; step six, judging whether the edge height errors exceed a preset adjustment threshold; step seven, when the edge height errors exceed the preset adjustment threshold, adjusting the positions of each edge section in the corresponding normal direction according to the edge height errors to obtain an updated current mask pattern; step eight, repeating steps four to seven until the edge height errors do not exceed the adjustment threshold, and outputting the current mask pattern as a target mask pattern. In the application, the difference between the simulated simulation wafer grid image and the target wafer grid image is not measured from a specific height plane and corrected, but the images are divided into grids, so that the height values of the grids at the same position in the simulation image and the target image are as consistent as possible, thereby greatly improving the accuracy of the optical proximity correction in the photoresist thickness direction, reducing potential defects such as photoresist height loss, and improving the yield of finished products.

[0107] The application further provides a training method of a three-dimensional optical proximity correction model, a flowchart of one specific embodiment of which is shown in Figure 4 The specific embodiment two comprises the following steps:

[0108] S201: establishing an uncalibrated three-dimensional optical proximity correction model; the three-dimensional optical proximity correction model can first determine a light intensity distribution image according to an inputted to-be-processed pattern through a preset cross-transmission coefficient, and then calculate a photoresist height image corresponding to the light intensity distribution image through a preset photoresist height model term and a corresponding linear coefficient.

[0109] As a specific embodiment, the three-dimensional optical proximity correction model obtains the light intensity distribution image through the following formula (5):

[0110] ; (5)

[0111] Wherein, I(x, y) is the light intensity distribution image, TCC is the cross-transmission coefficient, is a convolution operation, and M(x, y) is a rasterized to-be-processed pattern.

[0112] The rasterized to-be-processed pattern is an image obtained by rasterizing the to-be-processed pattern according to a preset grid size.

[0113] The three-dimensional optical proximity correction model obtains the photoresist height image by the following formula (6):

[0114] ; (6)

[0115] Wherein, H(x, y) is the photoresist height image, F j is the jth photoresist height model term, c j is the linear coefficient corresponding to the jth photoresist height model term, and I(x, y) is the light intensity distribution image.

[0116] The above two formulas provide a method for processing the to-be-processed pattern, and the corresponding photoresist height image is an image composed of height values of each coordinate position after lithography.

[0117] S202: According to a preset set of modeling patterns, a test mask is made, and a collected photoresist height image corresponding to the test mask is obtained.

[0118] The set of modeling patterns includes one-dimensional and two-dimensional patterns of different geometric shapes and sizes.

[0119] The photoresist height image, that is, the three-dimensional image of the wafer after the photoresist is set, can be a three-dimensional topographic image of the modeling pattern on the wafer after lithography obtained by using an atomic force microscope (Atomic Force Microscopy), that is, the height distribution of each pixel point on the pattern is obtained.

[0120] S203: Using the set of modeling patterns and the corresponding test three-dimensional topographic image, the cross-transmission coefficient, the photoresist height model term and the linear coefficient are calibrated, so that the difference between the simulated photoresist height image obtained by inputting the set of modeling patterns as a to-be-processed pattern into the three-dimensional optical proximity correction model and the collected photoresist height image is less than a preset calibration threshold.

[0121] Further, when the loss function value of the following formula (7) is less than the preset calibration threshold, it is considered that the difference between the simulated photoresist height image obtained by inputting the set of modeling patterns as a to-be-processed pattern into the three-dimensional optical proximity correction model and the collected photoresist height image is less than the preset calibration threshold:

[0122] ; (7)

[0123] Wherein, Cost is the loss function value; H ikh is a height value of the kth grid in the ith modeling pattern in the modeling pattern set in the simulated photoresist height image ik w is a height value of the kth grid in the ith modeling pattern in the modeling pattern set in the collected photoresist height image i w is a weight coefficient of the ith modeling pattern in the modeling pattern set k w is a weight coefficient of the kth grid of the ith modeling pattern in the modeling pattern set.

[0124] In the specific embodiment, the loss function for calibrating the above parameters is further defined, specifically, the difference between the height values of the grids at the same position in the simulated photoresist height image obtained by simulation and the collected photoresist height images obtained on both sides is calculated, and the weight coefficient of each modeling pattern and the weight coefficient of each grid are supplemented, and the sum of the weighted height value differences is taken as the function loss value. It is intuitive and simple, that is, the smaller the total difference between the simulation result and the actual result, the more accurate the correction of the cross transfer coefficient, the photoresist height model term and the linear coefficient, and it is also closer to the actual situation. Of course, other ways can also be used to correct the above parameters, which are not limited in the present application.

[0125] The training method of the three-dimensional optical proximity correction model provided by the present application comprises the following steps: establishing an uncalibrated three-dimensional optical proximity correction model; the three-dimensional optical proximity correction model can first determine a light intensity distribution image according to an input to-be-processed pattern through a preset cross transfer coefficient, and then calculate a photoresist height image corresponding to the light intensity distribution image through a preset photoresist height model term and a corresponding linear coefficient; a test mask is made according to a preset modeling pattern set, and a collected photoresist height image corresponding to the test mask is obtained; the cross transfer coefficient, the photoresist height model term and the linear coefficient are calibrated by using the modeling pattern set and the corresponding test three-dimensional topography image, so that the difference between the simulated photoresist height image obtained by inputting the modeling pattern set as the to-be-processed pattern into the three-dimensional optical proximity correction model and the collected photoresist height image is less than a preset calibration threshold. In the present application, the difference between the simulated simulation wafer grid image and the target wafer grid image is not measured from a specific height plane and corrected, but the images are divided into grids, so that the height values of the grids at the same position in the simulation image and the target image are as consistent as possible, thereby greatly improving the accuracy of the optical proximity correction in the thickness direction of the photoresist, reducing potential defects such as photoresist height loss, and improving the yield of finished products.

[0126] The optical proximity correction device provided by the embodiment of the present application is described below. The optical proximity correction device described below can be correspondingly referred to the optical proximity correction method described above.

[0127] Figure 5 The structure block diagram of the optical proximity correction device provided by the embodiment of the present application is shown in FIG. 1. Figure 5 The optical proximity correction device can comprise:

[0128] A receiving module 100 is configured to receive a target wafer pattern in step one; the target wafer pattern is surrounded by a plurality of edge segments;

[0129] A gridding module 200 is configured to determine a target wafer grid image according to the target wafer pattern based on a preset grid size in step two; each grid in the target wafer grid image records a corresponding height value;

[0130] A mask determining module 300 is configured to take the target wafer pattern as a current mask pattern in step three;

[0131] A simulation module 400 is configured to input the current mask pattern into a pre-trained three-dimensional optical proximity correction model and obtain a simulation wafer grid image by assisting the grid size in step four; each grid in the simulation wafer grid image records a corresponding height value;

[0132] A height difference module 500 is configured to calculate edge height errors corresponding to each edge segment in the simulation wafer grid image compared with the target wafer grid image in step five;

[0133] A judging module 600 is configured to judge whether the edge height errors exceed a preset adjustment threshold in step six;

[0134] An adjustment module 700 is configured to adjust the positions of each edge segment in a corresponding normal direction according to the edge height errors when the edge height errors exceed the preset adjustment threshold, and obtain an updated current mask pattern in step seven;

[0135] A loop module 800 is configured to loop steps four to seven until the edge height errors do not exceed the adjustment threshold, and output the current mask pattern as a target mask pattern in step eight.

[0136] As a preferred embodiment, the height difference module 500 comprises:

[0137] A grid determining unit is configured to take a control point located at the center of the edge segment as a control point coordinate grid, and take two grids adjacent to the control point coordinate grid in a normal direction corresponding to the edge segment as a calculation domain grid;

[0138] a height difference calculation unit configured to calculate a height difference between a height value of the control point coordinate grid in the simulation wafer grid image and a height value of the control point coordinate grid in the target wafer grid image, and calculate a height difference between a height value of each of the two calculation domain grids in the simulation wafer grid image and a height value of each of the two calculation domain grids in the target wafer grid image, respectively;

[0139] an edge height error adding unit configured to add the three height differences to obtain an edge height error corresponding to the edge segment.

[0140] As a preferred embodiment, the adjustment module 700 comprises:

[0141] an inward moving unit configured to move the edge segment inward along a corresponding normal direction when the edge height error indicates that a height value of a grid in the simulation wafer grid image is higher than a height value of a grid in the target wafer grid image;

[0142] an outward moving unit configured to move the edge segment outward along a corresponding normal direction when the edge height error indicates that a height value of a grid in the simulation wafer grid image is lower than a height value of a grid in the target wafer grid image.

[0143] As a preferred embodiment, the adjustment module 700 comprises:

[0144] a moving distance calculation unit configured to determine a moving distance of the edge segment in the adjustment process along the corresponding normal direction by using the following formula (3):

[0145] L = -f x (EHE / MF); (3)

[0146] wherein L is the moving distance of the edge segment, f is a feedback coefficient, MF is a mask error amplification coefficient corresponding to the edge segment, and EHE is the edge height error of the edge segment.

[0147] As a preferred embodiment, the judgment module 600 further comprises:

[0148] an update times judgment unit configured to judge whether an update times of the current mask pattern exceeds a preset tolerance threshold;

[0149] a stop updating unit configured to output the current mask pattern as a target mask pattern when the update times of the current mask pattern exceeds the tolerance threshold.

[0150] As a preferred embodiment, the rasterization module 200 and the simulation module 400 comprise:

[0151] The height value calculation unit is configured to calculate a height value of a grid where the pattern edge is located in the wafer grid image according to the following formula (4):

[0152] O = r x O1 + (1-r) x O2; (4)

[0153] wherein O is the height value of the grid where the pattern edge is located, O2 is a height value of a pattern coverage area, O1 is a height value of a pattern non-coverage area, and r is an area ratio of the pattern non-coverage area to the grid.

[0154] As a preferred embodiment, the training device of the three-dimensional optical proximity correction model comprises:

[0155] The model establishing module is configured to establish an uncalibrated three-dimensional optical proximity correction model, and the three-dimensional optical proximity correction model is configured to determine an optical intensity distribution image according to an inputted to-be-processed pattern through a preset cross transfer coefficient, and calculate a photoresist height image corresponding to the optical intensity distribution image through a preset photoresist height model term and a corresponding linear coefficient.

[0156] The test sample manufacturing module is configured to manufacture a test mask according to a preset modeling pattern set, and acquire an acquired photoresist height image corresponding to the test mask.

[0157] The calibration module is configured to calibrate the cross transfer coefficient, the photoresist height model term and the linear coefficient by using the modeling pattern set and the corresponding test three-dimensional topography image, so that a difference between a simulated photoresist height image obtained by inputting the modeling pattern set as the to-be-processed pattern into the three-dimensional optical proximity correction model and the acquired photoresist height image is less than a preset calibration threshold.

[0158] As a preferred embodiment, the model establishing module comprises:

[0159] The optical intensity distribution image unit is configured to make the three-dimensional optical proximity correction model obtain the optical intensity distribution image according to the following formula (5):

[0160] ; (5)

[0161] wherein I(x, y) is the optical intensity distribution image, TCC is the cross transfer coefficient, is a convolution operation, and M(x, y) is a rasterized to-be-processed pattern.

[0162] The photoresist height image unit is configured to make the three-dimensional optical proximity correction model obtain the photoresist height image according to the following formula (6):

[0163] ; (6)

[0164] wherein H(x, y) is the photoresist height image, F j is the jth photoresist height model term, c j is the linear coefficient corresponding to the jth photoresist height model term, and I(x, y) is the light intensity distribution image.

[0165] As a preferred embodiment, the calibration module comprises:

[0166] a threshold formula unit configured to consider that the difference between the simulated photoresist height image obtained by inputting the set of modeling patterns into the three-dimensional optical proximity correction model and the collected photoresist height image is less than the preset calibration threshold value when the loss function value of the following formula (7) is less than the preset calibration threshold value:

[0167] ; (7)

[0168] wherein Cost is the loss function value; H ik is the height value of the kth grid in the ith modeling pattern in the set of modeling patterns in the simulated photoresist height image, h ik is the height value of the kth grid in the ith modeling pattern in the set of modeling patterns in the collected photoresist height image, w i is the weight coefficient of the ith modeling pattern in the set of modeling patterns, w k is the weight coefficient of the kth grid of the ith modeling pattern in the set of modeling patterns.

[0169] The optical proximity correction device provided by the application comprises a receiving module 100 for receiving a target wafer pattern in step one; the target wafer pattern is surrounded by a plurality of edge sections; a gridding module 200 for determining a target wafer grid image according to the target wafer pattern according to a preset grid size in step two; each grid in the target wafer grid image records a corresponding height value; a mask determining module 300 for taking the target wafer pattern as a current mask pattern in step three; a simulation module 400 for inputting the current mask pattern into a pre-trained three-dimensional optical proximity correction model and obtaining a simulation wafer grid image with the aid of the grid size in step four; each grid in the simulation wafer grid image records a corresponding height value; a height difference module 500 for calculating edge height errors corresponding to each edge section in the simulation wafer grid image compared with the target wafer grid image in step five; a judging module 600 for judging whether the edge height errors exceed a preset adjustment threshold in step six; an adjustment module 700 for adjusting the positions of each edge section in the corresponding normal direction according to the edge height errors when the edge height errors exceed the preset adjustment threshold to obtain an updated current mask pattern in step seven; and a loop module 800 for looping steps four to seven until the edge height errors do not exceed the adjustment threshold and outputting the current mask pattern as a target mask pattern in step eight. In the application, the difference between the simulated simulation wafer grid image and the target wafer grid image is not measured from a specific height plane and corrected, but the images are divided into grids, so that the height values of the grids at the same position in the simulation image and the target image are as consistent as possible, thereby greatly improving the accuracy of the optical proximity correction in the photoresist thickness direction, reducing potential defects such as photoresist height loss, and improving the yield of finished products.

[0170] The optical proximity correction device of the embodiment is used to implement the optical proximity correction method described above, so the specific embodiments in the optical proximity correction device can be seen from the embodiment part of the optical proximity correction method in the foregoing, for example, the receiving module 100, the gridding module 200, the mask determining module 300, the simulation module 400, the height difference module 500, the judging module 600, the adjustment module 700, and the loop module 800 are respectively used to implement steps S101, S102, S103, S104, S105, S106, S107, and S108 in the optical proximity correction method, so the specific embodiments can be referred to the description of the corresponding embodiment part, and will not be described here.

[0171] The application also provides an optical proximity correction device, comprising:

[0172] a memory for storing a computer program;

[0173] The processor is used for executing the computer program to realize the steps of the optical proximity correction method. The optical proximity correction method provided by the application comprises the following steps: step one, receiving a target wafer pattern; the target wafer pattern is surrounded by a plurality of edge segments; step two, determining a target wafer grid image according to the target wafer pattern according to a preset grid size; each grid in the target wafer grid image records a corresponding height value; step three, taking the target wafer pattern as a current mask pattern; step four, inputting the current mask pattern into a pre-trained three-dimensional optical proximity correction model and assisting with the grid size to obtain a simulation wafer grid image; each grid in the simulation wafer grid image records a corresponding height value; step five, calculating edge height errors of each edge segment corresponding to the simulation wafer grid image compared with the target wafer grid image; step six, judging whether the edge height errors exceed a preset adjustment threshold; step seven, when the edge height errors exceed the preset adjustment threshold, adjusting positions of each edge segment in a corresponding normal direction according to the edge height errors to obtain an updated current mask pattern; step eight, repeating steps four to seven until the edge height errors do not exceed the adjustment threshold, and outputting the current mask pattern as a target mask pattern. In the application, the difference between the simulated simulation wafer grid image and the target wafer grid image is not measured from a specific height plane and corrected, but the images are divided into grids, so that the height values of the grids at the same position in the simulation image and the target image are as consistent as possible, thereby greatly improving the accuracy of the optical proximity correction in the photoresist thickness direction, reducing potential defects such as photoresist height loss, and improving the yield of finished products.

[0174] The application also provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the steps of the optical proximity correction method. The optical proximity correction method provided by the application comprises the following steps: step one, receiving a target wafer pattern; the target wafer pattern is surrounded by a plurality of edge segments; step two, determining a target wafer grid image according to the target wafer pattern according to a preset grid size; each grid in the target wafer grid image records a corresponding height value; step three, taking the target wafer pattern as a current mask pattern; step four, inputting the current mask pattern into a pre-trained three-dimensional optical proximity correction model and assisting with the grid size to obtain a simulation wafer grid image; each grid in the simulation wafer grid image records a corresponding height value; step five, calculating the edge height error of each edge segment corresponding to the simulation wafer grid image compared with the target wafer grid image; step six, judging whether the edge height error exceeds a preset adjustment threshold; step seven, when the edge height error exceeds the preset adjustment threshold, adjusting the position of each edge segment in the corresponding normal direction according to the edge height error to obtain an updated current mask pattern; step eight, repeating steps four to seven until the edge height error does not exceed the adjustment threshold, and outputting the current mask pattern as a target mask pattern. In the application, the difference between the simulated simulation wafer grid image and the target wafer grid image is not measured from a specific height plane and corrected, but the images are divided into grids, so that the height values of the grids at the same position in the simulation image and the target image are as consistent as possible, thereby greatly improving the accuracy of the optical proximity correction in the photoresist thickness direction, reducing potential defects such as photoresist height loss, and improving the yield of finished products.

[0175] The embodiments in the specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts of each embodiment can be referred to each other.

[0176] It should be noted that, in the specification, the terms such as first and second, etc. are merely intended to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "contain" or any other variants thereof are intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or device. Without more limitations, the element defined by the statement "including a" does not exclude the presence of additional identical elements in the process, method, article or device including the element.

[0177] The skilled person can further realize that the units and algorithm steps of the examples described in connection with the embodiments disclosed herein can be realized in electronic hardware, computer software, or a combination of both. In order to clearly illustrate the interchangeability of hardware and software, the components and steps of the examples have been described in general terms in the above description. Whether the functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. The skilled person can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0178] The steps of the method or algorithm described in connection with the embodiments disclosed herein can be directly implemented by hardware, a software module executed by a processor, or a combination of both. The software module can be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0179] The optical proximity correction method and device provided by the present application are described in detail above. The principles and implementation modes of the present application are described by applying specific examples in this paper, and the above description of the examples is only used to help understand the method of the present application and its core idea. It should be pointed out that, for ordinary skilled persons in the technical field, some improvements and modifications can be made to the present application without departing from the principles of the present application, and these improvements and modifications also fall within the protection scope of the present application.

Claims

1. An optical proximity correction method, characterized in that, include: Step 1: Receive the target wafer pattern; the target wafer pattern is composed of multiple edge segments; Step 2: Determine the target wafer grid image based on the target wafer pattern according to the preset grid size; each grid in the target wafer grid image records its corresponding height value; Step 3: Use the target wafer pattern as the current mask pattern; Step 4: Input the current mask pattern into the pre-trained 3D optical proximity correction model and supplement it with the grid size to obtain a simulated wafer grid image; each grid in the simulated wafer grid image records the corresponding height value; Step 5: Calculate the edge height error of each edge segment compared to the simulated wafer mesh image and the target wafer mesh image; Step 6: Determine whether the edge height error exceeds a preset adjustment threshold; Step 7: When the edge height error exceeds the preset adjustment threshold, adjust the position of each edge segment in the corresponding normal direction according to the edge height error to obtain the updated current mask pattern. Step 8: Repeat steps 4 to 7 until the edge height error does not exceed the adjustment threshold, and output the current mask pattern as the target mask pattern.

2. The optical proximity correction method as described in claim 1, characterized in that, Calculating the edge height error of each edge segment compared to the simulated wafer mesh image and the target wafer mesh image includes: The edge segment includes a control point, which is located at the center of the edge segment; the grid corresponding to the control point is used as the control point coordinate grid, and the two grids adjacent to the control point coordinate grid in the normal direction corresponding to the edge segment are used as the computational domain grid; Calculate the height difference between the height value of the control point coordinate grid in the simulated wafer grid image and the height value in the target wafer grid image; calculate the height difference between the height values ​​of the two computational domain grids in the simulated wafer grid image and the height values ​​in the target wafer grid image respectively; The three height differences are added together to obtain the edge height error corresponding to the edge segment.

3. The optical proximity correction method as described in claim 2, characterized in that, Based on the edge height error, the position of each edge segment is adjusted in the corresponding normal direction, including: When the edge height error indicates that the height value of the grid in the simulated wafer grid image is higher than the height value of the grid in the target wafer grid image, the edge segment is moved into the interior of the graphic along the corresponding normal direction; When the edge height error indicates that the height value of the grid in the simulated wafer grid image is lower than the height value of the grid in the target wafer grid image, the edge segment is moved outward along the corresponding normal direction.

4. The optical proximity correction method as described in claim 3, characterized in that, The distance the edge segment moves during the adjustment process in the corresponding normal direction is determined by the following formula: L = -f × (EHE / MF); Where L is the moving distance of the edge segment, f is the feedback coefficient, MF is the mask error amplification coefficient corresponding to the edge segment, and EHE is the edge height error of the edge segment.

5. The optical proximity correction method as described in claim 1, characterized in that, After determining whether the edge height error exceeds a preset adjustment threshold, the method further includes: Determine whether the number of times the current mask pattern has been updated exceeds a preset tolerance threshold; When the number of updates to the current mask pattern exceeds the tolerance threshold, the current mask pattern is output as the target mask pattern.

6. The optical proximity correction method according to any one of claims 1 to 5, characterized in that, In a wafer mesh image, the height of the mesh containing the pattern edge is calculated using the following formula: O = r × O1 + (1 - r) × O2; Where O is the height value of the grid where the edge of the pattern is located, O2 is the height value of the area covered by the pattern, O1 is the height value of the area not covered by the pattern, and r is the area ratio of the area not covered by the pattern to the area of ​​the grid.

7. The optical proximity correction method as described in claim 1, characterized in that, The training method for the three-dimensional optical proximity correction model includes: An uncalibrated three-dimensional optical proximity correction model is established. The three-dimensional optical proximity correction model can first determine the light intensity distribution image based on the input pattern to be processed through a preset cross-transfer coefficient, and then calculate the photoresist height image corresponding to the light intensity distribution image through a preset photoresist height model term and the corresponding linear coefficient. A test mask is fabricated based on a preset modeling pattern set, and the corresponding image of the photoresist height is acquired. Using the modeling pattern set and the corresponding test 3D topographic image, the cross-transfer coefficient, the photoresist height model term, and the linear coefficient are calibrated so that the difference between the simulated photoresist height image obtained by inputting the modeling pattern set as the pattern to be processed into the 3D optical proximity correction model and the acquired photoresist height image is less than a preset calibration threshold.

8. The optical proximity correction method as described in claim 7, characterized in that, The three-dimensional optical proximity correction model obtains the light intensity distribution image using the following formula: ； Wherein, I(x,y) is the light intensity distribution image, and TCC is the cross-transfer coefficient. For convolution operation, M(x,y) is the rasterized pattern to be processed; The three-dimensional optical proximity correction model obtains the photoresist height image using the following formula: ； Where H(x,y) is the height image of the photoresist, and F j For the j-th photoresist height model term, c j is the linear coefficient corresponding to the j-th photoresist height model term, and I(x,y) is the light intensity distribution image.

9. The optical proximity correction method as described in claim 7, characterized in that, When the loss function value of the current formula is less than a preset calibration threshold, it is considered that the difference between the simulated photoresist height image obtained by inputting the modeled pattern set as the pattern to be processed into the three-dimensional optical proximity correction model and the acquired photoresist height image is less than the preset calibration threshold. ； Where Cost is the value of the loss function; H ik h is the height value of the k-th mesh in the i-th modeling pattern set in the simulated photoresist height image. ik w is the height value of the k-th grid in the i-th modeling pattern set in the acquired photoresist height image. i w is the weight coefficient of the i-th modeling pattern in the modeling pattern set. k The weight coefficient is the k-th grid of the i-th modeling pattern in the modeling pattern set.

10. An optical proximity correction device, characterized in that, include: The receiving module is used in step one to receive the target wafer pattern; the target wafer pattern is formed by multiple edge segments; The rasterization module is used in step two to determine the target wafer grid image based on the target wafer pattern according to the preset grid size; each grid in the target wafer grid image records the corresponding height value; The mask determination module is used in step three to use the target wafer pattern as the current mask pattern. The simulation module is used in step four to input the current mask pattern into a pre-trained three-dimensional optical proximity correction model and supplement it with the grid size to obtain a simulated wafer grid image; each grid in the simulated wafer grid image records the corresponding height value; The height difference module is used in step five to calculate the edge height error corresponding to each edge segment compared with the simulated wafer mesh image and the target wafer mesh image; The judgment module is used in step six to determine whether the edge height error exceeds a preset adjustment threshold. The adjustment module is used in step seven to adjust the position of each edge segment in the corresponding normal direction according to the edge height error when the edge height error exceeds the preset adjustment threshold, so as to obtain the updated current mask pattern. The loop module is used in step eight to loop steps four through seven until the edge height error does not exceed the adjustment threshold, and then outputs the current mask pattern as the target mask pattern.

Images