Light source-mask optimization method for improving full-view-field photoetching imaging quality

Abstract

The invention provides a light source-mask optimization method for improving full-field photoetching imaging quality, which comprises the following steps of: under the condition of comprehensively considering point wave aberration of each field of view of a photoetching objective lens, firstly, constructing an optimization objective function into a linear weighted sum of sub-objective functions containing point wave aberration information of each field of view; secondly, optimizing a light source and a mask based on the optimization objective function, adaptively adjusting a weight factor of each view field point sub-objective function in the optimization process, and correcting the gradient between the light source and the mask by different view field point sub-objective functions in a gradient projection mode; gradient conflicts among the sub-objective functions of different view field points are reduced; and finally, updating the light source pattern and the mask pattern by adopting a gradient descent method based on the corrected gradient information, thereby improving the fidelity and uniformity of full-field photoetching imaging and improving the yield of an integrated circuit.

Description

Technical Field

[0001] The present invention belongs to the technical field of lithography resolution enhancement, and particularly relates to a source-mask optimization method for improving the lithography imaging quality of the entire field of view. Background Art

[0002] Optical lithography is a key process in the manufacture of ultra-large scale integrated circuits. The lithography system is a diffraction-limited system with a diffraction limit. Therefore, resolution enhancement techniques must be adopted to improve the lithography imaging resolution and pattern fidelity. The source and mask optimization (SMO) technique is one of the key resolution enhancement techniques for integrated circuit manufacturing at the 28 nm and below technology nodes. By co-optimizing the source pattern and the mask pattern and regulating the diffraction orders and amplitude distributions of the light waves participating in lithography imaging, the SMO technique can effectively compensate for lithography imaging errors and improve the pattern fidelity of lithography imaging.

[0003] Differences in the distribution of aberrations of the projection objective lens caused by factors such as optical design, processing, and alignment exist across the entire exposure field, resulting in uneven lithography imaging at each field point and seriously affecting the yield of integrated circuits.

[0004] Chinese Patent No. CN108693715A proposes a multi-objective source-mask optimization method for improving the uniformity of lithography imaging across the entire field of view for the polarization aberration of the projection lithography objective lens. In this method, the optimization objective function is constructed as the average value of the pattern error (PAE) containing the polarization aberration information of each field point to balance lithography imaging across the entire field of view.

[0005] The known source-mask optimization methods for the entire field of view have the following problems: On the one hand, the existing technologies adopt inefficient weight strategies and cannot effectively balance the lithography imaging quality across the entire field of view. On the other hand, the gradient conflict between the sub-objective functions of each field point during the optimization process will prevent the optimization objective function from converging to the optimal solution, reducing the fidelity and uniformity of lithography imaging across the entire field of view. Summary of the Invention

[0006] To solve the above problems, the present invention provides a source-mask optimization method for improving the lithography imaging quality of the entire field of view, constructs the objective function as a linear weighted sum of sub-objective functions containing the wave aberration information of each field point, and reduces the gradient conflict between the sub-objective functions of different field points through gradient projection, thereby improving the fidelity and uniformity of lithography imaging across the entire field of view and improving the yield of integrated circuits.

[0007] A source-mask optimization method for improving the lithography imaging quality of the entire field of view includes the following steps:

[0008] Step 1: Initialize the light source to a size of N S ×N S of the initial light source pattern J, and initialize the mask to a size of N M ×N M of the initial mask pattern M, where N S and N M are integers;

[0009] Step 2: Construct the objective function D as follows:

[0010]

[0011] where i = 1, 2,..., T, T is the number of field points, ω i is the weight corresponding to the i-th field point, F i is the sub-objective function corresponding to the i-th field point, and there is:

[0012]

[0013] where, is the target pattern matrix of size N M ×N M ; Z(W i ) represents the photoresist image pattern matrix of the i-th field point calculated using the lithography imaging model considering the wave aberration W i corresponding to the i-th field point;

[0014] Step 3: Optimize the light source and the mask based on the objective function D.

[0015] Furthermore, the pixel value of the light-emitting area on the initial light source pattern J is set to 1, and the pixel value of the non-light-emitting area is set to 0, and the initial light source pattern J satisfies where, Ω J is the pre-set light source variable matrix of size N S ×N S ;

[0016] The pixel value of the effective area of the initial mask pattern M is set to 1, and the background area is set to 0, and the initial mask pattern M satisfies where, Ω M is the pre-set mask variable matrix of size N M ×N M ;

[0017] Furthermore, in Step 3, the specific process of optimizing the light source and the mask based on the optimization objective function D is as follows:

[0018] Step 301: Obtain the light source gradient matrix and the mask gradient matrix of each field point according to the following steps:

[0019] Step 301a: Calculate the sub-objective function F corresponding to the i-th field of view point i of the scale factor where the moving average function EMA t [F i = βEMA t-1 [F i + (1 - β)F i t-1 , β represents the moving average weight factor, t represents the iteration round number, and F j is the sub-objective function corresponding to the j-th field of view point;

[0020] Step 301b: Calculate the weight factor of the sub-objective function F corresponding to the i-th field of view point i of where C is the temperature parameter used to regulate the influence of the scale factor H i on the weight factor ω i ;

[0021] Step 301c: Calculate the gradient of the sub-objective function F corresponding to the i-th field of view point i for the current light source variable matrix Ω J , denoted as the light source gradient matrix Calculate the gradient of the sub-objective function F corresponding to the i-th field of view point i for the current mask variable matrix Ω M , denoted as the mask gradient matrix

[0022] Step 301d: Arrange the rows of the light source gradient matrix corresponding to the i-th field of view point in sequence to convert them into a row vector, denoted as the light source gradient vector Arrange the rows of the mask gradient matrix corresponding to the i-th field of view point in sequence to convert them into a row vector, denoted as the mask gradient vector

[0023] Step 301e: Update the light source gradient vector of the i-th field of view point according to the inner product between the light source gradient vector of the i-th field of view point and the light source gradient vectors of the remaining field of view points and denote the finally updated light source gradient vector as Update the mask gradient vector of the i-th field of view point according to the inner product between the mask gradient vector of the i-th field of view point and the mask gradient vectors of the remaining field of view points and denote the finally obtained mask gradient vector as

[0024] Step 301f: Restore the updated light source gradient vector of the i-th field of view point to a matrix of size N S ×N S , denoted as the light source gradient matrix Restore the updated mask gradient vector of the i-th field of view point to a matrix of size N M ×N M , denoted as the mask gradient matrix

[0025] Step 302: Calculate the gradient matrix of the objective function D with respect to the current light source variable matrix Ω J and the gradient matrix of the objective function D with respect to the current mask variable matrix Ω based on the light source gradient matrix and the mask gradient matrix of each field of view point M and the gradient matrix of the objective function D with respect to the current mask variable matrix Ω

[0026] Step 303: Use the steepest descent method to update the light source variable matrix Ω J to Obtain the light source pattern J* corresponding to the current , where S J is the preset light source optimization step size; use the steepest descent method to update the mask variable matrix Ω M to Obtain the light source pattern M* corresponding to the current , where S M is the preset mask optimization step size;

[0027] Step 304: Calculate the value of the objective function D corresponding to the current light source pattern J* and the current mask pattern M*; determine whether the value of the objective function D is less than the predetermined threshold or whether the update times of the light source variable matrix Ω J and the mask variable matrix Ω M reach the predetermined upper limit value. If one of the judgment results is yes, terminate the optimization and determine the current light source pattern J* and the current mask pattern M* as the optimized light source pattern and mask pattern. If both judgment results are no, increment the iteration round number t by 1 and re-execute steps 301 to 304.

[0028] Furthermore, in step 301e, for any field of view point i, the method for obtaining the finally updated light source gradient vector is as follows:

[0029] Sort the remaining field of view points j other than the i-th field of view point according to their respective serial numbers, where j = 1,..., i - 1, i + 1,..., T;

[0030] Calculate the original light source gradient vector of the i-th field of view point The original light source gradient vector with the first remaining field of view point in the sequence Inner product

[0031] Judge whether the currently obtained inner product Is less than zero. If so, use the projection conflict gradient method to update the light source gradient vector of the i-th field of view point to Where Is the light source gradient vector obtained by this iterative update; if not, the light source gradient vector of the i-th field of view point remains unchanged, that is

[0032] Calculate the light source gradient vector obtained by the previous update of the i-th field of view point The inner product with the original light source gradient vector of the next remaining field of view point in the sequence, and then judge whether the currently obtained inner product is less than zero. If so, use the projection conflict gradient method to update the light source gradient vector obtained by the previous update of the i-th field of view point again. If not, the light source gradient vector obtained by the previous update of the i-th field of view point remains unchanged; and so on, until all the remaining field of view points in the sequence are traversed, and the light source gradient vector obtained by the last update of the i-th field of view point with the last remaining field of view point in the sequence is used as the final light source gradient vector

[0033] Furthermore, in step 301e, for any field of view point i, the method for obtaining its finally updated mask gradient vector Is as follows:

[0034] Sort the remaining field of view points j other than the i-th field of view point according to their respective serial numbers, where j = 1,..., i - 1, i + 1,..., T;

[0035] Calculate the original mask gradient vector of the i-th field of view point The original mask gradient vector with the first remaining field of view point in the sequence Inner product

[0036] Judge whether the currently obtained inner product Is less than zero. If so, use the projection conflict gradient method to update the mask gradient vector of the i-th field of view point to Where Is the mask gradient vector obtained by this iterative update; if not, the mask gradient vector of the i-th field of view point remains unchanged, that is

[0037] Calculate the mask gradient vector obtained by the previous update of the i-th field of view point The inner product with the original mask gradient vector of the next remaining field point in the sequence, and then determine whether the currently obtained inner product is less than zero. If so, the projection conflict gradient method is used to update the mask gradient vector obtained by the previous update of the i-th field point again. If not, the mask gradient vector obtained by the previous update of the i-th field point remains unchanged; and so on, until all the remaining field points in the sequence are traversed, and the mask gradient vector obtained by the last update of the i-th field point with the last remaining field point in the sequence is used as the final mask gradient vector

[0038] Beneficial effects:

[0039] The present invention provides a light source-mask optimization method for improving the full-field lithography imaging quality. Considering the wavefront aberration of each field point of the lithography objective lens comprehensively, first, the optimization objective function is constructed as a linear weighted sum of sub-objective functions containing the wavefront aberration information of each field point; then, based on the optimization objective function, the light source and the mask are optimized. During the optimization process, the weight factors of the sub-objective functions of each field point are adaptively adjusted, and the gradients between the sub-objective functions of different field points for the light source and the mask are corrected by the gradient projection method to reduce the gradient conflict between the sub-objective functions of different field points; finally, based on the corrected gradient information, the gradient descent method is used to update the light source pattern and the mask pattern, thereby improving the fidelity and uniformity of the full-field lithography imaging and improving the yield rate of integrated circuits Description of the drawings

[0040] Figure 1 It is a flowchart of a light source-mask optimization method for improving the full-field lithography imaging quality provided by the present invention

[0041] Figure 2 It is a schematic diagram of the position of the field point of the lithography objective lens provided by the present invention

[0042] Figure 3 It is a schematic diagram of the initial light source and the initial mask provided by the present invention

[0043] Figure 4 It is a schematic diagram of the optimized light source pattern and mask pattern obtained by using the prior art CN108693715A provided by the present invention

[0044] Figure 5 It is a schematic diagram of the optimized light source pattern and mask pattern obtained by using the present invention provided by the present invention Detailed implementation manners

[0045] In order to enable those skilled in the art of the present technology to better understand the solution of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application

[0046] Principle of the present invention: The present invention provides a light source-mask optimization method for improving the lithographic imaging quality of the full field of view. This method constructs the optimization objective function as a linear weighted sum of sub-objective functions containing wavefront aberration information of each field point. Based on the optimization objective function, the light source and the mask are optimized. During the optimization process, the weight factors of the sub-objective functions of each field point are adaptively adjusted, and the gradients of the sub-objective functions of each field point with respect to the light source and the mask are corrected by gradient projection to reduce the gradient conflict between the sub-objective functions of different field points. Based on the corrected gradient information, the light source pattern and the mask pattern are updated using the gradient descent method.

[0047] As Figure 1 shown, the present invention proposes a light source-mask optimization method for improving the lithographic imaging quality of the full field of view, and the main steps are as follows:

[0048] Step 1: Initialize the light source to an initial light source pattern J of size N S ×N S and initialize the mask to an initial mask pattern M of size N M ×N M , where N S and N M are integers;

[0049] Specifically, set the pixel value of the light-emitting area on the initial light source pattern J to 1 and the pixel value of the non-light-emitting area to 0, and the initial light source pattern J satisfies where Ω J is a pre-set light source variable matrix of size N S ×N S ; set the pixel value of the effective area of the initial mask pattern M to 1 and the background area to 0, and the initial mask pattern M satisfies where Ω M is a pre-set mask variable matrix of size N M ×N M .

[0050] Step 2: Construct the objective function D as follows:

[0051]

[0052] where i = 1, 2,..., T, T is the number of field points, ω i is the weight corresponding to the i-th field point, F i is the sub-objective function corresponding to the i-th field point, and there is:

[0053]

[0054] where is the target pattern matrix of size N M ×N M ; Z(W i ) represents the photoresist image pattern matrix of the i-th field point calculated using the lithographic imaging model considering the wave aberration W i corresponding to the i-th field point; thus, it can be seen that the objective function D is the linear weighted sum of the sub-objective functions F i of each field point of the lithographic objective.

[0055] It should be noted that based on the Abbe imaging principle, considering the flare effect and the photoresist diffusion effect, the spatial image corresponding to the i-th field point of the current light source pattern and the mask pattern is calculated as:

[0056]

[0057] where J(x s , y s ) is the intensity of the light source point (x s , y s ), represents the illumination intensity normalization factor; the symbol represents convolution, · represents the multiplication of the corresponding elements of two matrices, and || represents taking the modulus of each element in the matrix; is the mask diffraction matrix corresponding to the point light source J(x s , y s ), representing the translation of the mask diffraction spectrum caused by the oblique incidence effect of the point light source; H r is the convolution kernel representing the flare effect and the photoresist diffusion effect.

[0058] is the equivalent point spread function of the lithographic system corresponding to the light source point (x s , y s ), which includes the influence of wave aberration on the electric field vector and can be calculated as:

[0059]

[0060] where W i represents the wave aberration of the i-th field point, U is a low-pass filter, and R is the reduction factor of the projection optical device.

[0061] Adopting the photoresist hard threshold model approximated by the Sigmoid function, the photoresist image matrix corresponding to the i-th field point is calculated according to the spatial image intensity I(W i ) as:

[0062]

[0063] where a represents the steepness of the Sigmoid function, and t r characterizes the exposure threshold of the photoresist.

[0064] According to the above calculation process, comprehensively considering the wave aberration W corresponding to each field point i , the specific value of the sub-objective function F of each field point and the optimization objective function D can be calculated. i

[0065] It should be noted that when constructing the sub-objective function in the present invention, the pattern error (PAE) is used as the lithography imaging evaluation index. Other evaluation indexes such as edge placement error and normalized logarithmic slope can also be used to construct the sub-objective function, which will not be elaborated in the present invention.

[0066] Step 3: Optimize the light source and mask based on the objective function D. The specific steps are as follows:

[0067] Step 301: Obtain the light source gradient matrix and mask gradient matrix of each field point according to the following steps:

[0068] Step 301a: Calculate the scale factor of the sub-objective function F corresponding to the i-th field point i where the moving average function EMA t [F i = βEMA t-1 [F i + (1 - β)F i t-1 , β represents the moving average weight factor, t represents the iteration round number, and F j is the sub-objective function corresponding to the j-th field point;

[0069] Step 301b: Calculate the weight factor of the sub-objective function F corresponding to the i-th field point i where C is the temperature parameter used to control the influence of the scale factor H i on the weight factor ω i ;

[0070] Step 301c: Calculate the gradient of the sub-objective function F corresponding to the i-th field point i with respect to the current light source variable matrix Ω J , denoted as the light source gradient matrix Calculate the gradient of the sub-objective function F corresponding to the i-th field point i with respect to the current mask variable matrix Ω M , denoted as the mask gradient matrix

[0071] ​​​Step 301d: Arrange each row of the light source gradient matrix corresponding to the i-th field of view point in sequence to convert it into a row vector, denoted as the light source gradient vector Arrange each row of the mask gradient matrix corresponding to the i-th field of view point in sequence to convert it into a row vector, denoted as the mask gradient vector

[0072] Step 301e: Update the light source gradient vector of the i-th field of view point according to the inner product between the light source gradient vector of the i-th field of view point and the light source gradient vectors of the remaining field of view points And denote the finally updated light source gradient vector as Update the mask gradient vector of the i-th field of view point according to the inner product between the mask gradient vector of the i-th field of view point and the mask gradient vectors of the remaining field of view points And denote the finally obtained mask gradient vector as

[0073] Furthermore, in Step 301e, for any field of view point i, the method for obtaining its finally updated light source gradient vector is as follows:

[0074] Sort the remaining field of view points j other than the i-th field of view point according to their respective serial numbers, where j = 1, …, i - 1, i + 1, …, T;

[0075] Calculate the inner product between the original light source gradient vector of the i-th field of view point and the original light source gradient vector of the first remaining field of view point in the sequence

[0076] Judge whether the currently obtained inner product is less than zero. If so, adopt the projection conflict gradient method to update the light source gradient vector of the i-th field of view point to where is the light source gradient vector obtained by this iterative update; if not, the light source gradient vector of the i-th field of view point remains unchanged, that is

[0077] Calculate the light source gradient vector The inner product with the original light source gradient vector of the next remaining field of view point in the sequence, and then determine whether the currently obtained inner product is less than zero. If so, the projection conflict gradient method is used to update the light source gradient vector obtained from the previous update of the $i$-th field of view point again. If not, the light source gradient vector obtained from the previous update of the $i$-th field of view point remains unchanged; and so on, until all the remaining field of view points in the sequence are traversed. The light source gradient vector obtained from the last update of the $i$-th field of view point and the last remaining field of view point in the sequence is used as the final light source gradient vector

[0078] For example, assume there are 10 fields of view, the final light source gradient vector of the first field of view is updated as follows:

[0079] First, calculate the inner product between the original light source gradient vector of the first field of view and the original light source gradient vector of the second field of view, and select different update methods to update the light source gradient vector of the first field of view according to whether the inner product is less than zero; then calculate the inner product between the light source gradient vector obtained from the first update of the first field of view and the original light source gradient vector of the third field of view, and select different update methods to update the light source gradient vector obtained from the first update of the first field of view according to whether the inner product is less than zero; then calculate the inner product between the light source gradient vector obtained from the second update of the first field of view and the original light source gradient vector of the fourth field of view, and so on, until the 10th field of view is traversed; the result of the ninth update of the light source gradient vector obtained from the eighth update of the first field of view and the original light source gradient vector of the 10th field of view is used as the final light source gradient vector of the first field of view

[0080] The final light source gradient vector of the second field of view is obtained in a similar way to the final light source gradient vector of the first field of view The acquisition method is also to first calculate the inner product between the original light source gradient vector of the second field of view point and the original light source gradient vector of the first field of view point, and select different update methods to update the light source gradient vector of the second field of view point according to whether the inner product is less than zero; then calculate the inner product between the light source gradient vector obtained by the first update of the second field of view point and the original light source gradient vector of the third field of view point, and select different update methods to update the light source gradient vector obtained by the first update of the second field of view point according to whether the inner product is less than zero; then calculate the inner product between the light source gradient vector obtained by the second update of the second field of view point and the original light source gradient vector of the fourth field of view point, and so on until the tenth field of view point is traversed; the ninth update result of the light source gradient vector obtained by the eighth update of the second field of view point and the original light source gradient vector of the tenth field of view point is used as the final light source gradient vector of the second field of view point

[0081] Similarly, the final light source gradient vectors of the remaining field of view points can be obtained, and the present invention will not elaborate on this

[0082] Further, in step 301e, for any field of view point i, the finally updated mask gradient vector The acquisition method is as follows:

[0083] Sort the remaining field of view points j other than the i-th field of view point according to their respective serial numbers, where j = 1,..., i - 1, i + 1,..., T;

[0084] Calculate the original mask gradient vector of the i-th field of view point And the original mask gradient vector of the first remaining field of view point in the sequence Inner product

[0085] Judge whether the currently obtained inner product Is less than zero. If so, use the projection conflict gradient method to update the mask gradient vector of the i-th field of view point to Among them, Is the mask gradient vector obtained by this iterative update; if not, the mask gradient vector of the i-th field of view point remains unchanged, that is

[0086] Calculate the mask gradient vector obtained by the previous update of the i-th field of view point The inner product with the original mask gradient vector of the next remaining field point in the sequence, and then determine whether the currently obtained inner product is less than zero. If so, the projection conflict gradient method is used to update the mask gradient vector obtained from the previous update of the i-th field point again. If not, the mask gradient vector obtained from the previous update of the i-th field point remains unchanged; and so on, until all the remaining field points in the sequence are traversed, and the mask gradient vector obtained from the last update of the i-th field point and the last remaining field point in the sequence is used as the final mask gradient vector

[0087] It should be noted that the method for obtaining the final mask gradient vector of each field point is similar to the method for obtaining the final light source gradient vector of each field point, and the present invention will not elaborate on this

[0088] Step 301f, the updated light source gradient vector of the i-th field point is restored to a matrix of size N S ×N S and denoted as the light source gradient matrix The updated mask gradient vector of the i-th field point is restored to a matrix of size N M ×N M and denoted as the mask gradient matrix

[0089] Step 302, calculate the gradient matrix of the objective function D with respect to the current light source variable matrix Ω J and the gradient matrix of the objective function D with respect to the current mask variable matrix Ω and the gradient matrix of the objective function D with respect to the current mask variable matrix Ω M and denoted as the gradient matrix

[0090] Step 303, use the steepest descent method to update the light source variable matrix Ω J to obtain the light source pattern J* corresponding to the current , where S J is a preset light source optimization step size; use the steepest descent method to update the mask variable matrix Ω M to obtain the light source pattern M* corresponding to the current , where S M is a preset mask optimization step size;

[0091] Step 304, calculate the value of the objective function D corresponding to the current light source pattern J* and the current mask pattern M*; determine whether the value of the objective function D is less than a predetermined threshold or the light source variable matrix Ω J and the mask variable matrix ΩM Whether the update count reaches a predetermined upper limit value. If one of the judgment results is yes, the optimization is terminated, and the current light source pattern J* and the current mask pattern M* are determined as the optimized light source pattern and mask pattern. If both judgment results are no, the iteration round number t is incremented by 1 and steps 301 to 304 are executed again.

[0092] Furthermore, as Figure 2 shown in the schematic diagram of the field point position of the lithographic objective lens, the wave aberration data of each field point can be directly exported by the optical design software CODE V.

[0093] As Figure 3 shown in the schematic diagram of the initial light source and the initial mask. In Figure 3 , 301 is the initial light source pattern, where white represents the light-emitting part and black represents the non-light-emitting part. 302 is the initial mask pattern, which is also the target pattern. White represents the effective area and black represents the background area, and its feature size is 27 nm.

[0094] As Figure 4 shown for the target pattern in Figure 3 , the schematic diagrams of the light source pattern and the mask pattern optimized by the related patent (CN108693715A) (hereinafter simply referred to as method A). In Figure 4 , 401 is the light source pattern optimized by method A; 402 is the mask pattern optimized by method A.

[0095] As Figure 5 shown for the target pattern in Figure 3 , the schematic diagrams of the light source pattern and the mask pattern optimized by the light source-mask optimization method proposed in the present invention (hereinafter simply referred to as method B). In Figure 5 , 501 is the light source pattern optimized by method B; 502 is the mask pattern optimized by method B.

[0096] Here, the PAE data is used to describe the lithographic imaging quality. Table 1 gives the statistical data of the PAE distribution of method A and method B at different field points:

[0097] Table 1 Statistical data of PAE distribution corresponding to different methods at each field point

[0098]

[0099] The data in Table 1 shows that the average value, standard deviation, and range of variation of PAE corresponding to method B are all smaller than those of method A. The average value has decreased by 2.5%, the standard deviation has decreased by 34.9%, and the range of variation has decreased by 39.0%. The comparison shows that the proposed method B is superior to method A in terms of lithographic imaging fidelity and uniformity, demonstrating the superiority of the present invention.

[0100] In summary, the present invention discloses a method for optimizing a light source - mask to improve the lithographic imaging quality of the entire field of view. This method constructs the optimization objective function as a weighted sum of sub - objective functions containing wavefront aberration information of each field - of - view point. During the optimization process, the weight factors of the sub - objective functions of each field - of - view point are adaptively adjusted, and the gradients of the sub - objective functions of each field - of - view point with respect to the light source and the mask are corrected by means of gradient projection. Using the corrected gradient information, the light source pattern and the mask pattern are updated, improving the fidelity and uniformity of the lithographic imaging of the entire field of view.

[0101] Of course, the present invention may also have many other embodiments. Without departing from the spirit and essence of the present invention, those skilled in the art can certainly make various corresponding changes and deformations according to the present invention. However, these corresponding changes and deformations should all fall within the protection scope of the appended claims of the present invention.

Claims

1. A light source-mask optimization method for improving full-field lithography imaging quality, characterized in that: The following steps are involved: Step 1: Initialize the light source size to N S ×N S The initial light source pattern J is initialized to a size of N M ×N M The initial mask pattern M, where N S and N M is an integer; Step 2: Construct the objective function D as follows: Where i = 1, 2, ..., T, T is the number of viewing points, ω i is the weight corresponding to the i-th field of view point, F i is the sub-objective function corresponding to the i-th field of view point, and: in, For size N M ×N M The target graphics matrix; Z(W i ) represents the wave aberration W corresponding to the i-th field of view point i In the case of , the photoresist image matrix of the i-th field of view point is calculated using the photolithography imaging model; Step three: Based on the objective function D, optimize the light source and the mask.

2. The light source-mask optimization method for improving full-field lithography imaging quality according to claim 1, characterized in that: The pixel value of the luminous area on the initial light source graph J is set to 1, and the pixel value of the non-luminous area is set to 0, and the initial light source graph J satisfies Among them, Ω J The preset size is N S ×N S The light source variable matrix; The pixel value of the effective area of ​​the initial mask pattern M is set to 1, and the pixel value of the background area is set to 0, and the initial mask pattern M satisfies Among them, Ω M The preset size is N M ×N M The mask variable matrix.

3. The light source-mask optimization method for improving full-field lithography imaging quality according to claim 2, characterized in that: In step 3, based on the optimization objective function D, the specific process of optimizing the light source and the mask is as follows: Step 301: Obtain the light source gradient matrix and mask gradient matrix of each field of view point according to the following steps: Step 301a: Calculate the sub-objective function F corresponding to the i-th field of view point i The scaling factor Among them, the moving average function EMA t [F i ]=βEMA t-1 [F i ]+(1-β)F i t-1 , β represents the moving average weight factor, t represents the iteration round number, F j is the sub-objective function corresponding to the j-th field of view point; Step 301b: Calculate the sub-objective function F corresponding to the i-th field of view point i The weight factor Among them, C is used to control the proportional factor H i For the weight factor ω i The influence of temperature parameters; Step 301c: Calculate the sub-objective function F corresponding to the i-th field of view point i For the current light source variable matrix Ω J The gradient of Calculate the sub-objective function F corresponding to the i-th field of view point i For the current mask variable matrix Ω M The gradient of is recorded as the mask gradient matrix Step 301d: The light source gradient matrix corresponding to the i-th field of view point The rows are sequentially converted into a The row vector is recorded as the light source gradient vector The mask gradient matrix corresponding to the i-th field of view point The rows are sequentially converted into a The row vector of is denoted as the mask gradient vector Step 301e: Update the light source gradient vector of the ith field of view point according to the inner product between the light source gradient vector of the ith field of view point and the light source gradient vectors of the remaining field of view points. And the final updated light source gradient vector is recorded as Update the mask gradient vector of the i-th field of view point according to the inner product between the mask gradient vector of the i-th field of view point and the mask gradient vectors of the remaining field of view points The final mask gradient vector is recorded as Step 301f: Update the light source gradient vector of the i-th field of view point Restore to a size of N S ×N S The matrix is ​​recorded as the light source gradient matrix The updated mask gradient vector of the i-th field of view point Restore to a size of N M ×N M The matrix is ​​recorded as the mask gradient matrix Step 302: Calculate the objective function D for the current light source variable matrix Ω according to the light source gradient matrix and mask gradient matrix of each field of view point. J The gradient matrix And the objective function D for the current mask variable matrix Ω M The gradient matrix Step 303: Use the steepest descent method to convert the light source variable matrix Ω J Updated to Get the corresponding current The light source graph J*, where S J Optimize the step size for the preset light source; use the steepest descent method to convert the mask variable matrix Ω M Updated to Get the corresponding current The light source pattern M*, where S M Optimize the step size for a pre-set mask; Step 304: Calculate the value of the objective function D corresponding to the current light source pattern J* and the current mask pattern M*; determine whether the value of the objective function D is less than a predetermined threshold or the light source variable matrix Ω J With the mask variable matrix Ω M Whether the number of updates reaches the predetermined upper limit value, if one of the judgment results is yes, the optimization is terminated, and the current light source pattern J* and the current mask pattern M* are determined as the optimized light source pattern and mask pattern; if the judgment results are all no, the iteration round number t is incremented by 1 and steps 301 to 304 are re-executed.

4. The light source-mask optimization method for improving full-field lithography imaging quality according to claim 3, characterized in that: In step 301e, for any field of view point i, the light source gradient vector finally updated is The method to obtain is: Sort the remaining view points j except the i-th view point according to their respective serial numbers, where j = 1, ..., i-1, i+1, ..., T; Calculate the original light source gradient vector of the i-th field of view point The original light source gradient vector of the first remaining field point in the sequence The inner product of Determine the inner product currently obtained Is it less than zero? If so, the projection conflict gradient method is used to update the light source gradient vector of the i-th field of view point to in, is the light source gradient vector obtained by this iteration update; if not, the light source gradient vector of the i-th field of view point remains unchanged, that is, Calculate the light source gradient vector obtained from the last update of the i-th field of view point The inner product of the original light source gradient vector of the next remaining field of view point in the sequence is then determined to be less than zero. If so, the projection conflict gradient method is used to update the light source gradient vector of the last update of the ith field of view point again. If not, the light source gradient vector of the last update of the ith field of view point remains unchanged; and so on, until all the remaining field of view points in the sequence are traversed, and the light source gradient vector obtained by the last update of the ith field of view point and the last remaining field of view point in the sequence is used as the final light source gradient vector 5. The light source-mask optimization method for improving full-field lithography imaging quality according to claim 3, characterized in that: In step 301e, for any field of view point i, the mask gradient vector finally updated is The method to obtain is: Sort the remaining view points j except the i-th view point according to their respective serial numbers, where j = 1, ..., i-1, i+1, ..., T; Calculate the original mask gradient vector of the i-th field of view point The original mask gradient vector of the first remaining field point in the sequence The inner product of Determine the inner product currently obtained Is it less than zero? If so, the projection conflict gradient method is used to update the mask gradient vector of the i-th field of view point to in, is the mask gradient vector obtained by this iteration update; if not, the mask gradient vector of the i-th field of view point remains unchanged, that is, Calculate the mask gradient vector obtained from the last update of the i-th field of view point The inner product of the original mask gradient vector of the next remaining field of view point in the sequence is calculated, and then it is determined whether the inner product currently obtained is less than zero. If so, the projection conflict gradient method is used to update the mask gradient vector obtained by the last update of the i-th field of view point again. If not, the mask gradient vector obtained by the last update of the i-th field of view point remains unchanged; and so on, until all the remaining field of view points in the sequence are traversed, and the mask gradient vector obtained by the last update of the i-th field of view point and the last remaining field of view point in the sequence is used as the final mask gradient vector

Images