Inverse photoetching inversion method, device and equipment and storage medium

Abstract

The invention relates to the field of integrated circuit manufacturing, in particular to an inverse photoetching inversion method, device and equipment and a storage medium. Determining angular points of the to-be-processed chip pattern; according to a preset fillet parameter, performing fillet processing on the corner points, and generating a fillet transformation pattern in the to-be-processed chip pattern; rasterizing a to-be-processed chip pattern including the fillet transformation pattern, and assigning a value to each grid to obtain mask target image distribution; determining mask pixel distribution according to the mask target image distribution; and inputting the mask pixel distribution into a preset optical model, and obtaining the optimal pattern distribution of the mask through multiple iterations. According to the method, the spatial intensity in optical simulation is converted into the sigmod image by means of the sigmod function, so that the convergence efficiency of an iteration process is improved, the calculation power consumption and the calculation time consumption are reduced, and the exposure quality of an integrated circuit chip is improved.

Description

Technical Field

[0001] This invention relates to the field of integrated circuit manufacturing, and in particular to a reverse lithography method, apparatus, equipment, and storage medium. Background Technology

[0002] As the critical dimensions of integrated circuits continue to shrink, traditional optical proximity correction (OPC) is gradually becoming insufficient to meet the requirements of mask pattern calibration, and inverse lithography (ILT) is playing an increasingly important role. ILT is a method that uses the desired pattern on the silicon wafer as the target and inversely calculates the required pattern on the mask. It sets the exposed image of the wafer surface after photolithography as the ideal imaging result, and inversely calculates the mask image based on the transformation model of the spatial image of the imaging system.

[0003] Inductively coupled plasma (ILT) starts from wafer exposure images and reverse-engineers the optimal mask pattern, greatly improving the flexibility and accuracy of optimization to meet the stringent requirements of advanced processes for pattern precision. However, current ILT methods achieve spectral calibration of the mask pattern by decomposing it into pixel values ​​with multiple degrees of freedom, resulting in high computational complexity and a huge computational burden, posing a significant challenge to optimization calculations. In particular, ILT is a typical large-scale optimization problem, and the optimization process is multi-solution and ill-conditioned. Current methods often employ iterative solutions based on conjugate gradient methods and minimum residual margin methods to obtain pixel-level photomask pattern distributions. However, the optimization results of these iterative methods are strictly dependent on the initial values ​​of the iterations; different initial values ​​usually yield local minima, which cannot meet the requirements for optimal mask pattern design.

[0004] Therefore, how to simultaneously improve the computational efficiency of reverse lithography inversion calculation of mask patterns, reduce computing power consumption, and improve convergence efficiency during the iteration process, thereby improving the exposure quality and yield of integrated circuit chips, is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0005] The purpose of this invention is to provide a reverse lithography inversion method, apparatus, device and storage medium to solve the problems of low computational efficiency, high computing power consumption and poor convergence efficiency in the reverse lithography inversion calculation of mask patterns in the prior art, which leads to a decrease in the exposure quality and yield of integrated circuit chips.

[0006] To solve the above-mentioned technical problems, the present invention provides a reverse lithography inversion method, comprising: Receive the chip pattern to be processed; Determine the corner points of the chip pattern to be processed; According to the preset corner rounding parameters, the corner points are rounded to generate a corner rounding transformation pattern inside the chip pattern to be processed. The chip pattern to be processed, including the rounded corner transformation pattern, is rasterized, and a value is assigned to each grid to obtain the mask target image distribution; wherein, the value corresponding to the grid that completely overlaps with the rounded corner transformation pattern and the value corresponding to the grid that does not completely overlap with the rounded corner transformation pattern are different, and the value corresponding to the grid that partially overlaps with the rounded corner transformation pattern is the area ratio of the rounded corner transformation pattern in the grid. The mask pixel distribution is determined based on the distribution of the target image. The mask pixel distribution is input into a preset optical model, and the optimal mask pattern distribution is obtained through multiple iterations. In each iteration, the optical model simulates the mask pixel distribution using a preset optical kernel function to obtain the corresponding spatial intensity. Then, the spatial intensity is converted into a sigmoid image using a sigmoid function, and the sigmoid image is compared with the mask target image distribution until the difference between the sigmoid image and the mask target image distribution is minimized. When the difference between the sigmoid image and the mask target image distribution is minimized, the mask pixel distribution corresponding to the sigmoid image is taken as the optimal mask pattern distribution.

[0007] Optionally, in the reverse lithography inversion method, the difference between the distribution of the sigmoid image and the mask target image is minimized when the function value of the loss function in the following equation (1) is minimized: ; (1) Where N is the total number of grid cells, T(x, y) is the distribution of the mask target image, and t i IS(x, y) is the value corresponding to the i-th grid in the mask target image distribution, and IS(x, y) is the sigmoid image. i This is the value corresponding to the i-th raster in the sigmoid image.

[0008] Optionally, in the reverse lithography inversion method, the mask pixel distribution is input into a preset optical model, and the optimal mask pattern distribution is obtained through multiple iterations, including: The mask pixel distribution is input into a preset optical model, and the optimal mask pattern distribution is obtained through multiple iterations using the Adam algorithm and the gradient of the mask pixel distribution with respect to the loss function.

[0009] Optionally, in the reverse lithography inversion method, determining the mask pixel distribution based on the mask target image distribution includes: In the mask target image distribution, the values ​​in the grids that do not overlap with the rounded corner transformation pattern are replaced with random values ​​to obtain the mask pixel distribution.

[0010] Optionally, in the reverse lithography inversion method, the chip pattern to be processed, including the rounded corner transformation pattern, is rasterized, and a value is assigned to each raster to obtain the mask target image distribution, including: The chip pattern to be processed, including the rounded corner transformation pattern, is rasterized, and each raster is assigned a value to obtain the mask target image distribution; wherein, the value corresponding to the raster that does not overlap with the rounded corner transformation pattern is 0, and the value corresponding to the raster that overlaps with the rounded corner transformation pattern is 1. Accordingly, the step of randomly replacing the values ​​in the grid cells that do not overlap with the rounded corner transformation pattern in the mask target image distribution to obtain the mask pixel distribution includes: In the mask target image distribution, the values ​​in the grids that do not overlap with the rounded corner transformation pattern are replaced with random values ​​between 0 and 1 to obtain the mask pixel distribution.

[0011] Optionally, in the reverse lithography inversion method, the mask pixel distribution is input into a preset optical model, and the optimal mask pattern distribution is obtained through multiple iterations. The method further includes: The mask pixel distribution is input into a preset optical model, and the optimal mask pattern distribution is obtained after a preset first number of iterations; the mask pixel distribution corresponding to the sigmoid image of the mask pixel distribution of the first number of iterations is taken as the optimal mask pattern distribution.

[0012] A reverse lithography inversion apparatus, comprising: The receiving module is used to receive the chip pattern to be processed; A corner point module is used to determine the corner points of the chip pattern to be processed; The corner rounding module is used to round the corners according to preset corner rounding parameters and generate a corner rounding transformation pattern inside the chip pattern to be processed. A rasterization module is used to rasterize the chip pattern to be processed, including the rounded corner transformation pattern, and assign a value to each grid to obtain the mask target image distribution; wherein, the value corresponding to the grid that completely overlaps with the rounded corner transformation pattern and the value corresponding to the grid that does not completely overlap with the rounded corner transformation pattern are different, and the value corresponding to the grid that partially overlaps with the rounded corner transformation pattern is the area ratio of the rounded corner transformation pattern in the grid. An initial distribution module is used to determine the mask pixel distribution based on the distribution of the mask target image; The model iteration module is used to input the mask pixel distribution into a preset optical model and obtain the optimal mask pattern distribution through multiple iterations. In each iteration, the optical model simulates the mask pixel distribution using a preset optical kernel function to obtain the corresponding spatial intensity. Then, the spatial intensity is converted into a sigmoid image using a sigmoid function, and the sigmoid image is compared with the mask target image distribution until the difference between the sigmoid image and the mask target image distribution is minimized. When the difference between the sigmoid image and the mask target image distribution is minimized, the mask pixel distribution corresponding to the sigmoid image is taken as the optimal mask pattern distribution.

[0013] Optionally, in the aforementioned reverse lithography inversion apparatus, the model iteration module includes: The loss function unit is used to determine the minimum difference between the distribution of the sigmoid image and the mask target image when the function value in the following equation (1) is minimized: ; (1) Where N is the total number of grid cells, T(x, y) is the distribution of the mask target image, and t i IS(x, y) is the value corresponding to the i-th grid in the mask target image distribution, and IS(x, y) is the sigmoid image. i This is the value corresponding to the i-th raster in the sigmoid image.

[0014] A reverse lithography inversion device includes: Memory, used to store computer programs; A processor, configured to implement the steps of any of the above-described reverse lithography inversion methods when executing the computer program.

[0015] A computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of any of the above-described reverse lithography inversion methods.

[0016] The reverse lithography inversion method provided by this invention involves: receiving a chip pattern to be processed; determining the corner points of the chip pattern; rounding the corner points according to preset rounding parameters to generate a rounded corner transformation pattern inside the chip pattern; rasterizing the chip pattern including the rounded corner transformation pattern and assigning a value to each raster to obtain a mask target image distribution; wherein the values ​​corresponding to the raster that completely overlaps with the rounded corner transformation pattern and the values ​​corresponding to the raster that does not completely overlap with the rounded corner transformation pattern are different, and the value corresponding to the raster that partially overlaps with the rounded corner transformation pattern is the area ratio of the rounded corner transformation pattern within the raster; and determining the mask target image distribution based on the mask target image distribution. The mask pixel distribution is input into a preset optical model, and the optimal mask pattern distribution is obtained through multiple iterations. In each iteration, the optical model simulates the mask pixel distribution using a preset optical kernel function to obtain the corresponding spatial intensity. Then, the spatial intensity is converted into a Sigma-Aldrich image using a Sigma-Aldrich function, and the Sigma-Aldrich image is compared with the mask target image distribution until the difference between the Sigma-Aldrich image and the mask target image distribution is minimized. The mask pixel distribution corresponding to the Sigma-Aldrich image with the minimum difference is taken as the optimal mask pattern distribution. In this invention, the spatial intensity in the optical simulation is converted into a Sigma-Aldrich image using a Sigma-Aldrich function. By comparing the difference between the distribution of grid values ​​in the Sigma-Aldrich image and the distribution of the mask target image presented in Sigma-Aldrich image form, the similarity between the simulation result and the target is determined, which greatly improves the convergence efficiency in the iteration process and also greatly reduces the computational power consumption and computation time, thereby improving the exposure quality and yield of integrated circuit chips. The present invention also provides a reverse lithography inversion apparatus, device and storage medium having the above-mentioned beneficial effects. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 A flowchart illustrating a specific implementation of the reverse lithography inversion method provided by the present invention; Figure 2 A chip pattern to be processed according to a specific embodiment of the reverse lithography inversion method provided by the present invention; Figure 3A chip pattern to be processed, including a rounded corner transformation pattern, is provided as a specific embodiment of the reverse lithography inversion method provided by the present invention. Figure 4 This is a schematic diagram of a specific embodiment of the reverse lithography inversion device provided by the present invention.

[0019] Figure label: 100 - Receiver module, 200 - Corner module, 300 - Rounded corner module, 400 - Rasterization module, 500 - Initial distribution module, 600 - Model iteration module. Detailed Implementation

[0020] To enable those skilled in the art to better understand the present invention, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are merely some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] The core of this invention is to provide a reverse lithography inversion method, the flowchart of one specific implementation of which is shown below. Figure 1 As shown, this is referred to as Specific Implementation Method One, which includes: S101: Receive the pattern of the chip to be processed.

[0022] The chip pattern to be processed is the final pattern that needs to be printed on the chip. (Refer to...) Figure 2 , Figure 2 The polygon in the diagram represents the chip pattern to be processed in a specific implementation.

[0023] S102: Determine the corner points of the chip pattern to be processed.

[0024] Following the previous text, the patterns that need to be drawn on the chip are usually composed of Manhattan patterns, which are graphics composed of horizontal and vertical lines. The corner points are the top corner positions of the Manhattan patterns.

[0025] S103: According to the preset corner rounding parameters, the corner points are rounded to generate a corner rounding transformation pattern inside the chip pattern to be processed.

[0026] The rounding parameters can be adjusted according to specific circumstances, and this invention does not impose any limitations on them. After the rounding process, the resulting rounded corner transformation pattern no longer has the apex corner formed by the intersection of two straight lines, but instead presents a curved edge at the original corner point. Figure 2 The rounded corner transformation pattern formed after the chip pattern to be processed in the image can be referenced. Figure 3 , Figure 3 The image still shows the chip pattern to be processed. This is because the rounded corner transformation pattern is just an edge variation of the chip pattern to be processed. The two patterns can actually be considered to be in the same position. The rounded corner transformation pattern after rounding is marked with a diagonal line area.

[0027] S104: The chip pattern to be processed, including the rounded corner transformation pattern, is rasterized, and a value is assigned to each grid to obtain the mask target image distribution; wherein, the value corresponding to the grid that completely overlaps with the rounded corner transformation pattern and the value corresponding to the grid that does not completely overlap with the rounded corner transformation pattern are different, and the value corresponding to the grid that partially overlaps with the rounded corner transformation pattern is the area ratio of the rounded corner transformation pattern in the grid.

[0028] In this step, it can be considered that the direct approach is... Figure 3 The image is rasterized, that is, the chip pattern to be processed, including the rounded corner transformation pattern, is divided into numerous grids, and values ​​are assigned according to whether the rounded corner transformation pattern exists in the grid and how many rounded corner transformation patterns there are. After the assignment, the resulting mask target image is essentially a grid array labeled with the corresponding values.

[0029] In this step, the value assigned to the grid that partially overlaps with the rounded corner transformation pattern is determined by the area ratio of the rounded corner transformation pattern in the grid. For example, if the rounded corner transformation pattern occupies half of the grid area, then the value corresponding to that grid is 1 / 2.

[0030] It should be noted that the rounding parameters mentioned above are based on the actual photolithography mask parameters and experimental results. Therefore, the rounding transformation pattern can be directly determined as the target pattern that the mask needs to achieve in order to realize the chip pattern to be processed. Correspondingly, the corresponding grid distribution becomes the target image distribution of the mask.

[0031] S105: Determine the mask pixel distribution based on the mask target image distribution.

[0032] After determining the target of the iteration (i.e., the mask target image distribution), it is also necessary to set the initial distribution for the iteration. In this invention, the values ​​in the grid are modified directly based on the mask target image distribution, and the resulting grid distribution can be used as the initial image distribution for input optical model, that is, the mask pixel distribution is the initial distribution.

[0033] S106: Input the mask pixel distribution into a preset optical model, and obtain the optimal mask pattern distribution through multiple iterations; in each iteration, the optical model simulates the mask pixel distribution using a preset optical kernel function to obtain the corresponding spatial intensity, and then uses the sigmoid function to convert the spatial intensity into a sigmoid image, and compares the sigmoid image with the mask target image distribution until the difference between the sigmoid image and the mask target image distribution is minimized. When the difference between the sigmoid image and the mask target image distribution is minimized, the mask pixel distribution corresponding to the sigmoid image is taken as the optimal mask pattern distribution.

[0034] In this step, the mask pixel distribution obtained in step S105 is input into the optical model to perform optical simulation, and the mask pixel distribution is continuously iterated, with the expectation that the mask pixel distribution after the iteration is the same as the mask target image distribution.

[0035] This step requires multiple iterations of the mask pixel distribution. Specifically, in each iteration, the mask pixel distribution needs to be converted into spatial intensity using the following formula (2): ; (2) Where I(x, y) is the spatial intensity, h k (x, y) represents the k-th optical kernel function, and K is the number of effective truncation optical kernel functions. For the convolution operator, u k is the weighting coefficient corresponding to the k-th optical kernel function, and P(x,y) is the mask pixel distribution.

[0036] After obtaining the spatial intensity, it is necessary to further utilize the sigmoid function to obtain the sigmoid image using the following equation (3): ; (3) Where IS(x, y) is the sigmoid image.

[0037] Specifically, the sigmod function can be represented by the following equation (4): ; (4) Where a and t r These are preset control parameters; the two parameters are used to determine the smoothness of the sigmoid function.

[0038] As a preferred embodiment, the difference between the distribution of the sigmoid image and the mask target image is minimized when the function value of the loss function in equation (1) is minimized: ; (1) Where N is the total number of grid cells, T(x, y) is the distribution of the mask target image, and t i IS(x, y) is the value corresponding to the i-th grid in the mask target image distribution, and IS(x, y) is the sigmoid image. i This is the value corresponding to the i-th raster in the sigmoid image.

[0039] In this embodiment, the difference between the distribution of the mask target image and the value on the corresponding grid in the sigmoid image is calculated, and the difference of all grid values ​​is used as the function value of the loss function. This method has a fast convergence speed, which can further improve the computational efficiency. Moreover, the calculation process is clear and direct, and the results obtained are more accurate.

[0040] Furthermore, the mask pixel distribution is input into a preset optical model, and the optimal mask pattern distribution is obtained through multiple iterations, including: The mask pixel distribution is input into a preset optical model, and the optimal mask pattern distribution is obtained through multiple iterations using the Adam algorithm and the gradient of the mask pixel distribution with respect to the loss function.

[0041] In this preferred embodiment, the Adam algorithm (Adaptive Moment Method) is used to obtain the gradient of the loss function with respect to the membrane pixel distribution. Based on the magnitude of the gradient change, the minimum value of the loss function can be quickly determined, thus achieving rapid convergence of the iterative process.

[0042] Based on the preferred embodiments described above, the gradient of the loss function with respect to the mask pixel distribution can be expressed as the following equation (5): ; (5) Where T represents matrix transpose, and e represents the element-wise gradient. Substituting Adam into the equations, we can update the grid and iterate, specifically including the following equations (6), (7), (8), (9), and (10): ; (6) ; (7) ; (8) ; (9) ; (10) Where, the subscript t represents the number of iterations, b1 and b2 are the corresponding hyperparameters, a is the gradient update step size, e is the error threshold (used to prevent the denominator from being 0), and m, v, ... , All of these are intermediate variables in the recursion and have no physical meaning. Substituting equations (1), (2), and (3) into equations (6) to (10) allows for iterative updates until the difference between the distribution of the sigmoid image and the mask target image is minimized.

[0043] Further, determining the mask pixel distribution based on the mask target image distribution includes: In the mask target image distribution, the values ​​in the grids that do not overlap with the rounded corner transformation pattern are replaced with random values ​​to obtain the mask pixel distribution.

[0044] In this preferred embodiment, a method of randomly replacing the values ​​in the grid that does not overlap with the rounded corner transformation pattern is used to obtain the initial mask pixel distribution for inputting the optical model. This can greatly reduce the computational power consumption and shorten the time required to obtain the initial mask pixel distribution. At the same time, since only the values ​​in the grid that does not overlap with the rounded corner transformation pattern are changed, the number of iterations can also be reduced, the iteration time can be shortened, and the final simulation accuracy can be improved.

[0045] Furthermore, the chip pattern to be processed, including the rounded corner transformation pattern, is rasterized, and a value is assigned to each raster to obtain the mask target image distribution, including: A1: The chip pattern to be processed, including the rounded corner transformation pattern, is rasterized, and a value is assigned to each raster to obtain the mask target image distribution; wherein, the value corresponding to the raster that does not overlap with the rounded corner transformation pattern is 0, and the value corresponding to the raster that completely overlaps with the rounded corner transformation pattern is 1.

[0046] Accordingly, the step of randomly replacing the values ​​in the grid cells that do not overlap with the rounded corner transformation pattern in the mask target image distribution to obtain the mask pixel distribution includes: A2: In the mask target image distribution, the values ​​in the grids that do not overlap with the rounded corner transformation pattern are replaced with random values ​​between 0 and 1 to obtain the mask pixel distribution.

[0047] In this preferred embodiment, the value corresponding to the grid that does not overlap with the rounded corner transformation pattern is defined as 0, and the value corresponding to the grid that completely overlaps with the rounded corner transformation pattern is defined as 1. At the same time, in the process of determining the mask pixel distribution, the values ​​in the grid that does not overlap with the rounded corner transformation pattern are replaced with random values ​​between 0 and 1, so that the values ​​of all grids in the obtained mask pixel distribution and the values ​​of all grids in the mask target image are between 0 and 1. This is beneficial for the Adam algorithm to iterate, can greatly improve the efficiency of Adam iteration, and further shorten the iteration time.

[0048] In another specific embodiment, the mask pixel distribution is input into a preset optical model, and the optimal mask pattern distribution is obtained through multiple iterations, further comprising: The mask pixel distribution is input into a preset optical model, and the optimal mask pattern distribution is obtained after a preset first number of iterations; the mask pixel distribution corresponding to the sigmoid image of the mask pixel distribution of the first number of iterations is taken as the optimal mask pattern distribution.

[0049] In this specific embodiment, an upper limit is set for the number of iterations. In other words, once the number of iterations exceeds the first number, even if the difference between the distribution of the sigmoid image and the mask target image has not reached the minimum, the iteration will not continue. Instead, the mask pixel distribution of the last iteration will be directly output as the optimal mask pattern distribution. This avoids the simulation results of some patterns failing to converge and dragging down the reverse lithography inversion of other chip patterns, thus improving the working stability of the reverse lithography inversion method.

[0050] The reverse lithography inversion method provided by this invention involves: receiving a chip pattern to be processed; determining the corner points of the chip pattern; rounding the corner points according to preset rounding parameters to generate a rounded corner transformation pattern inside the chip pattern; rasterizing the chip pattern including the rounded corner transformation pattern and assigning a value to each raster to obtain a mask target image distribution; wherein the values ​​corresponding to the raster that completely overlaps with the rounded corner transformation pattern and the values ​​corresponding to the raster that does not completely overlap with the rounded corner transformation pattern are different, and the value corresponding to the raster that partially overlaps with the rounded corner transformation pattern is the area ratio of the rounded corner transformation pattern within the raster; and determining the mask target image distribution based on the mask target image distribution. The mask pixel distribution is input into a preset optical model, and the optimal mask pattern distribution is obtained through multiple iterations. In each iteration, the optical model simulates the mask pixel distribution using a preset optical kernel function to obtain the corresponding spatial intensity. Then, the spatial intensity is converted into a Sigma-Aldrich image using a Sigma-Aldrich function, and the Sigma-Aldrich image is compared with the mask target image distribution until the difference between the Sigma-Aldrich image and the mask target image distribution is minimized. The mask pixel distribution corresponding to the Sigma-Aldrich image with the minimum difference is taken as the optimal mask pattern distribution. In this invention, the spatial intensity in the optical simulation is converted into a Sigma-Aldrich image using a Sigma-Aldrich function. By comparing the difference between the distribution of grid values ​​in the Sigma-Aldrich image and the distribution of the mask target image presented in Sigma-Aldrich image form, the similarity between the simulation result and the target is determined, which greatly improves the convergence efficiency in the iteration process and also greatly reduces the computational power consumption and computation time, thereby improving the exposure quality and yield of integrated circuit chips.

[0051] The reverse lithography inversion apparatus provided in the embodiments of the present invention will be described below. The reverse lithography inversion apparatus described below can be referred to in correspondence with the reverse lithography inversion method described above.

[0052] Figure 4 This is a structural block diagram of the reverse lithography inversion apparatus provided in an embodiment of the present invention, with reference to... Figure 4 The reverse lithography inversion apparatus may include: Receiver module 100 is used to receive the chip pattern to be processed; Corner point module 200, used to determine the corner points of the chip pattern to be processed; The corner rounding module 300 is used to round the corners according to preset corner rounding parameters and generate a corner rounding transformation pattern inside the chip pattern to be processed. The rasterization module 400 is used to rasterize the chip pattern to be processed, including the rounded corner transformation pattern, and assign a value to each grid to obtain the mask target image distribution; wherein, the value corresponding to the grid that completely overlaps with the rounded corner transformation pattern and the value corresponding to the grid that does not overlap with the rounded corner transformation pattern are different, and the value corresponding to the grid that partially overlaps with the rounded corner transformation pattern is the area ratio of the rounded corner transformation pattern in the grid. The initial distribution module 500 is used to determine the mask pixel distribution based on the distribution of the mask target image; The model iteration module 600 is used to input the mask pixel distribution into a preset optical model and obtain the optimal mask pattern distribution through multiple iterations. In each iteration, the optical model simulates the mask pixel distribution using a preset optical kernel function to obtain the corresponding spatial intensity. Then, the spatial intensity is converted into a sigmoid image using a sigmoid function, and the sigmoid image is compared with the mask target image distribution until the difference between the sigmoid image and the mask target image distribution is minimized. When the difference between the sigmoid image and the mask target image distribution is minimized, the mask pixel distribution corresponding to the sigmoid image is taken as the optimal mask pattern distribution.

[0053] In a preferred embodiment, the model iteration module 600 includes: The loss function unit is used to determine the minimum difference between the distribution of the sigmoid image and the mask target image when the function value in the following equation (1) is minimized: ; (1) Where N is the total number of grid cells, T(x, y) is the distribution of the mask target image, and t i IS(x, y) is the value corresponding to the i-th grid in the mask target image distribution, and IS(x, y) is the sigmoid image. i This is the value corresponding to the i-th raster in the sigmoid image.

[0054] In a preferred embodiment, the model iteration module 600 includes: The Adam iteration unit is used to input the mask pixel distribution into a preset optical model, and through the Adam algorithm, using the gradient of the loss function on the mask pixel distribution, obtain the optimal mask pattern distribution through multiple iterations.

[0055] In a preferred embodiment, the initial distribution module 500 includes: The non-overlapping grid random substitution unit is used to randomly replace the values ​​in the grids that do not overlap with the rounded corner transformation pattern in the mask target image distribution to obtain the mask pixel distribution.

[0056] In a preferred embodiment, the rasterization module 400 includes: The zero-one grid fill unit is used to rasterize the chip pattern to be processed, including the rounded corner transformation pattern, and assign a value to each grid to obtain the mask target image distribution; wherein, the value corresponding to the grid that does not overlap with the rounded corner transformation pattern is 0, and the value corresponding to the grid that completely overlaps with the rounded corner transformation pattern is 1. Accordingly, the initial distribution module 500 includes: The zero-one random substitution unit is used to replace the values ​​in the grid cells that do not overlap with the rounded corner transformation pattern in the mask target image distribution with random values ​​between 0 and 1 to obtain the mask pixel distribution.

[0057] In a preferred embodiment, the model iteration module 600 further includes: The iteration number threshold unit is used to input the mask pixel distribution into a preset optical model, and obtain the optimal mask pattern distribution after a preset first number of iterations; the mask pixel distribution corresponding to the sigmoid image of the mask pixel distribution of the first number of iterations is taken as the optimal mask pattern distribution.

[0058] The reverse lithography inversion method provided by this invention includes a receiving module 100 for receiving a chip pattern to be processed; a corner point module 200 for determining the corner points of the chip pattern to be processed; a corner rounding module 300 for rounding the corner points according to preset corner rounding parameters, generating a corner rounding transformation pattern inside the chip pattern to be processed; a rasterization module 400 for rasterizing the chip pattern to be processed including the corner rounding transformation pattern and assigning a value to each raster to obtain a mask target image distribution; wherein, the values ​​corresponding to the raster that completely overlaps with the corner rounding transformation pattern and the values ​​corresponding to the raster that does not completely overlap with the corner rounding transformation pattern are different, and the value corresponding to the raster that partially overlaps with the corner rounding transformation pattern is the area ratio of the corner rounding transformation pattern within the raster; and an initial distribution module 5. 00, used to determine the mask pixel distribution based on the mask target image distribution; Model iteration module 600, used to input the mask pixel distribution into a preset optical model, and obtain the optimal mask pattern distribution through multiple iterations; In each iteration, the optical model simulates the mask pixel distribution through a preset optical kernel function to obtain the corresponding spatial intensity, and then uses the sigmoid function to convert the spatial intensity into a sigmoid image, and compares the sigmoid image with the mask target image distribution until the difference between the sigmoid image and the mask target image distribution is minimized. When the difference between the sigmoid image and the mask target image distribution is minimized, the mask pixel distribution corresponding to the sigmoid image is taken as the optimal mask pattern distribution. In this invention, the spatial intensity in the optical simulation is converted into a sigmoid image using the sigmoid function. By comparing the difference between the distribution of grid values ​​in the sigmoid image and the distribution of the mask target image presented in the form of a sigmoid image, the similarity between the simulation result and the target is determined, which greatly improves the convergence efficiency in the iteration process and also greatly reduces the computing power consumption and computation time, thereby improving the exposure quality and yield of integrated circuit chips.

[0059] The reverse lithography inversion apparatus of this embodiment is used to implement the aforementioned reverse lithography inversion method. Therefore, the specific implementation of the reverse lithography inversion apparatus can be found in the embodiment section of the reverse lithography inversion method above. For example, the receiving module 100, corner module 200, rounding module 300, rasterization module 400, initial distribution module 500, and model iteration module 600 are respectively used to implement steps S101, S102, S103, S104, S105, and S106 in the above-mentioned reverse lithography inversion method. Therefore, its specific implementation can be referred to the description of the corresponding embodiments, which will not be repeated here.

[0060] The present invention also provides a reverse lithography inversion apparatus, comprising: Memory, used to store computer programs; A processor is configured to execute the computer program to implement the steps of any of the above-described reverse lithography inversion methods. The reverse lithography inversion method provided by this invention involves: receiving a chip pattern to be processed; determining the corner points of the chip pattern to be processed; performing corner rounding processing on the corner points according to preset corner rounding parameters to generate a corner rounding transformation pattern inside the chip pattern to be processed; rasterizing the chip pattern to be processed including the corner rounding transformation pattern and assigning a value to each raster to obtain a mask target image distribution; wherein, the values ​​corresponding to the raster that completely overlaps with the corner rounding transformation pattern and the values ​​corresponding to the raster that does not completely overlap with the corner rounding transformation pattern are different, and the value corresponding to the raster that partially overlaps with the corner rounding transformation pattern is the area ratio of the corner rounding transformation pattern within the raster; determining the mask target image distribution based on the mask target image distribution. The mask pixel distribution is input into a preset optical model, and the optimal mask pattern distribution is obtained through multiple iterations. In each iteration, the optical model simulates the mask pixel distribution using a preset optical kernel function to obtain the corresponding spatial intensity. Then, the spatial intensity is converted into a Sigma-Aldrich image using a Sigma-Aldrich function, and the Sigma-Aldrich image is compared with the mask target image distribution until the difference between the Sigma-Aldrich image and the mask target image distribution is minimized. The mask pixel distribution corresponding to the Sigma-Aldrich image with the minimum difference is taken as the optimal mask pattern distribution. In this invention, the spatial intensity in the optical simulation is converted into a Sigma-Aldrich image using a Sigma-Aldrich function. By comparing the difference between the distribution of grid values ​​in the Sigma-Aldrich image and the distribution of the mask target image presented in Sigma-Aldrich image form, the similarity between the simulation result and the target is determined, which greatly improves the convergence efficiency in the iteration process and also greatly reduces the computational power consumption and computation time, thereby improving the exposure quality and yield of integrated circuit chips.

[0061] This invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of any of the above-described reverse lithography inversion methods. The reverse lithography inversion method provided by this invention involves: receiving a chip pattern to be processed; determining the corner points of the chip pattern to be processed; performing corner rounding processing on the corner points according to preset corner rounding parameters to generate a corner rounding transformation pattern inside the chip pattern to be processed; rasterizing the chip pattern to be processed including the corner rounding transformation pattern and assigning a value to each raster to obtain a mask target image distribution; wherein the values ​​corresponding to the raster that completely overlaps with the corner rounding transformation pattern and the values ​​corresponding to the raster that does not completely overlap with the corner rounding transformation pattern are different, and the value corresponding to the raster that partially overlaps with the corner rounding transformation pattern is the area ratio of the corner rounding transformation pattern within the raster; determining the mask target image distribution based on the mask target image distribution. The mask pixel distribution is input into a preset optical model, and the optimal mask pattern distribution is obtained through multiple iterations. In each iteration, the optical model simulates the mask pixel distribution using a preset optical kernel function to obtain the corresponding spatial intensity. Then, the spatial intensity is converted into a Sigma-Aldrich image using a Sigma-Aldrich function, and the Sigma-Aldrich image is compared with the mask target image distribution until the difference between the Sigma-Aldrich image and the mask target image distribution is minimized. The mask pixel distribution corresponding to the Sigma-Aldrich image with the minimum difference is taken as the optimal mask pattern distribution. In this invention, the spatial intensity in the optical simulation is converted into a Sigma-Aldrich image using a Sigma-Aldrich function. By comparing the difference between the distribution of grid values ​​in the Sigma-Aldrich image and the distribution of the mask target image presented in Sigma-Aldrich image form, the similarity between the simulation result and the target is determined, which greatly improves the convergence efficiency in the iteration process and also greatly reduces the computational power consumption and computation time, thereby improving the exposure quality and yield of integrated circuit chips.

[0062] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to in the method section.

[0063] It should be noted that, in this specification, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0064] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0065] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0066] The foregoing provides a detailed description of the reverse lithography inversion method, apparatus, device, and storage medium provided by this invention. Specific examples have been used to illustrate the principles and implementation methods of this invention. The descriptions of the embodiments above are merely for the purpose of helping to understand the method and core ideas of this invention. It should be noted that those skilled in the art can make various improvements and modifications to this invention without departing from its principles, and these improvements and modifications also fall within the protection scope of this invention.

Claims

1. A reverse photolithography inversion method, characterized in that, include: Receive the chip pattern to be processed; Determine the corner points of the chip pattern to be processed; According to the preset corner rounding parameters, the corner points are rounded to generate a corner rounding transformation pattern inside the chip pattern to be processed. The chip pattern to be processed, including the rounded corner transformation pattern, is rasterized, and a value is assigned to each grid to obtain the mask target image distribution; wherein, the value corresponding to the grid that completely overlaps with the rounded corner transformation pattern and the value corresponding to the grid that does not completely overlap with the rounded corner transformation pattern are different, and the value corresponding to the grid that partially overlaps with the rounded corner transformation pattern is the area ratio of the rounded corner transformation pattern in the grid. The mask pixel distribution is determined based on the distribution of the target image. The mask pixel distribution is input into a preset optical model, and the optimal mask pattern distribution is obtained through multiple iterations. In each iteration, the optical model simulates the mask pixel distribution using a preset optical kernel function to obtain the corresponding spatial intensity. Then, the spatial intensity is converted into a sigmoid image using a sigmoid function, and the sigmoid image is compared with the mask target image distribution until the difference between the sigmoid image and the mask target image distribution is minimized. When the difference between the sigmoid image and the mask target image distribution is minimized, the mask pixel distribution corresponding to the sigmoid image is taken as the optimal mask pattern distribution.

2. The reverse lithography inversion method as described in claim 1, characterized in that, The difference between the distributions of the sigmoid image and the mask target image is minimized when the loss function of the following formula is minimized: ； Where N is the total number of grid cells, T(x, y) is the distribution of the mask target image, and t i IS(x, y) is the value corresponding to the i-th grid in the mask target image distribution, and IS(x, y) is the sigmoid image. i This is the value corresponding to the i-th raster in the sigmoid image.

3. The reverse lithography inversion method as described in claim 2, characterized in that, The mask pixel distribution is input into a preset optical model, and the optimal mask pattern distribution is obtained through multiple iterations, including: The mask pixel distribution is input into a preset optical model, and the optimal mask pattern distribution is obtained through multiple iterations using the Adam algorithm and the gradient of the mask pixel distribution with respect to the loss function.

4. The reverse lithography inversion method as described in claim 3, characterized in that, Based on the distribution of the target image in the mask, the distribution of mask pixels is determined, including: In the mask target image distribution, the values ​​in the grids that do not overlap with the rounded corner transformation pattern are replaced with random values ​​to obtain the mask pixel distribution.

5. The reverse lithography inversion method as described in claim 4, characterized in that, The chip pattern to be processed, including the rounded corner transformation pattern, is rasterized, and a value is assigned to each raster to obtain the mask target image distribution, including: The chip pattern to be processed, including the rounded corner transformation pattern, is rasterized, and each raster is assigned a value to obtain the mask target image distribution; wherein, the value corresponding to the raster that does not overlap with the rounded corner transformation pattern is 0, and the value corresponding to the raster that completely overlaps with the rounded corner transformation pattern is 1. Accordingly, the step of randomly replacing the values ​​in the grid cells that do not overlap with the rounded corner transformation pattern in the mask target image distribution to obtain the mask pixel distribution includes: In the mask target image distribution, the values ​​in the grids that do not overlap with the rounded corner transformation pattern are replaced with random values ​​between 0 and 1 to obtain the mask pixel distribution.

6. The reverse lithography inversion method as described in claim 1, characterized in that, The process of inputting the mask pixel distribution into a preset optical model and obtaining the optimal mask pattern distribution through multiple iterations also includes: The mask pixel distribution is input into a preset optical model, and the optimal mask pattern distribution is obtained after a preset first number of iterations; the mask pixel distribution corresponding to the sigmoid image of the mask pixel distribution of the first number of iterations is taken as the optimal mask pattern distribution.

7. A reverse lithography inversion apparatus, characterized in that, include: The receiving module is used to receive the chip pattern to be processed; A corner point module is used to determine the corner points of the chip pattern to be processed; The corner rounding module is used to round the corners according to preset corner rounding parameters and generate a corner rounding transformation pattern inside the chip pattern to be processed. A rasterization module is used to rasterize the chip pattern to be processed, including the rounded corner transformation pattern, and assign a value to each grid to obtain the mask target image distribution; wherein, the value corresponding to the grid that completely overlaps with the rounded corner transformation pattern and the value corresponding to the grid that does not completely overlap with the rounded corner transformation pattern are different, and the value corresponding to the grid that partially overlaps with the rounded corner transformation pattern is the area ratio of the rounded corner transformation pattern in the grid. An initial distribution module is used to determine the mask pixel distribution based on the distribution of the mask target image; The model iteration module is used to input the mask pixel distribution into a preset optical model and obtain the optimal mask pattern distribution through multiple iterations. In each iteration, the optical model simulates the mask pixel distribution using a preset optical kernel function to obtain the corresponding spatial intensity. Then, the spatial intensity is converted into a sigmoid image using a sigmoid function, and the sigmoid image is compared with the mask target image distribution until the difference between the sigmoid image and the mask target image distribution is minimized. When the difference between the sigmoid image and the mask target image distribution is minimized, the mask pixel distribution corresponding to the sigmoid image is taken as the optimal mask pattern distribution.

8. The reverse lithography inversion apparatus as described in claim 7, characterized in that, The model iteration module includes: The loss function unit is used to determine the minimum difference between the distributions of the sigmoid image and the mask target image when the function value in the following formula is minimized: ； Where N is the total number of grid cells, T(x, y) is the distribution of the mask target image, and t i IS(x, y) is the value corresponding to the i-th grid in the mask target image distribution, and IS(x, y) is the sigmoid image. i This is the value corresponding to the i-th raster in the sigmoid image.

9. A reverse lithography inversion device, characterized in that, include: Memory, used to store computer programs; A processor for executing the computer program to implement the steps of the reverse lithography inversion method as described in any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the reverse lithography inversion method as described in any one of claims 1 to 6.

Images