A method for obtaining diffraction near-field distribution
By segmenting the target mask pattern and calculating its corresponding core matrix, the problems of low accuracy of diffraction near-field distribution calculation and long calculation time in the prior art are solved, and efficient and accurate near-field distribution recovery is achieved, suitable for DUV light High-precision calculations for etching.
Patent Information
- Application Number
- CN202111335555.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-11-11
- Publication Date
- 2024-12-03
- Estimated Expiration
- 2041-11-11
AI Technical Summary
The existing diffraction near-field distribution calculation method has low accuracy and is too long in the manufacturing of small-size chips. In particular, the Kirchoff approximation method has low accuracy and high computational capacity for electromagnetic field calculation, which cannot effectively simulate large-scale graphics. and the convolution method cannot restore the 'corrier' field strength distribution.
By determining the segmentation pattern of the target mask pattern and the corresponding kernel matrix, the diffraction near-field distribution of the target area is calculated and spliced to obtain the near-field distribution of the entire mask pattern. Matrix operation is used instead of the convolution kernel, and the minimum two are used. Multiplication calculates the kernel matrix.
It improves the accuracy of diffraction near-field distribution calculation, reduces the calculation time, and can effectively restore the 'corrigated' field strength distribution, significantly improving the calculation accuracy and efficiency of DUV lithography.
Smart Images

Figure CN114092490B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of photolithography technology, and in particular to a method for obtaining diffraction near-field distribution. Background Art
[0002] like Figure 1 As shown, photolithography technology is an extremely critical step in the manufacturing process of very large-scale integrated circuits. Its significance lies in using optical projection methods to ultimately transfer the pattern of circuit devices on the mask to the photoresist coated on the silicon wafer.
[0003] With the continuous development of integrated circuit technology, the size of circuit patterns on chips has gradually shrunk, which has improved the performance of chips while increasing the density of transistors. However, it has also brought great challenges to chip manufacturing. At present, on advanced chips, the minimum distance between patterns is only tens of nanometers, and the mainstream lithography technology in the industry uses a light source wavelength of 193 nanometers (ArF deep ultraviolet lithography as an example). At this size, the diffraction effect of light will be very serious. Therefore, resolution enhancement technology can be used to improve the quality of lithography patterns, such as Optical Proximity Correction (OPC) and Source Mask Optimization (SMO). Among these technologies, calculating the diffraction near-field distribution of the target mask pattern is an extremely important step.
[0004] The existing methods for calculating the diffraction near-field distribution can be divided into Kirchhoff's approximation method, electromagnetic field rigorous calculation method, and fast calculation method. Kirchhoff's approximation method uses a scalar model and assumes that the mask satisfies Kirchhoff's boundary conditions: in the light-transmitting area of the mask, the field distribution and its partial derivatives are exactly the same as when the mask does not exist; in the light-shielding area of the mask, the field distribution and its partial derivatives are completely zero. In this way, the near-field distribution of the mask can be expressed as a two-dimensional amplitude transmittance function of the mask pattern. It can be seen that the Kirchhoff approximation method is simple and convenient, but it does not consider the three-dimensional effect of the mask and the influence of its boundary on the light field. Therefore, the near-field distribution results obtained using the Kirchhoff approximation method are of low accuracy. Correspondingly, the electromagnetic field rigorous calculation method calculates the diffraction near field of the thick mask by solving the Maxwell equations. The near-field results obtained using this method are theoretically completely accurate, but at the same time, the calculation time of this method is too long, making it impossible to apply it to the calculation of large-scale diffraction near-field distribution.
[0005] The existing fast calculation method is based on convolution calculation. First, the strict calculation method of electromagnetic field is used to calculate the accurate near-field distribution corresponding to a part of the graphics and use it as a training set. Then, using the training set as the standard, the convolution kernel that can reflect the relationship between the graphics and its corresponding near-field distribution is calculated. Finally, the convolution kernel is used to convolve the entire graphics, and the mask graphics are calculated and restored point by point to finally obtain the near-field distribution corresponding to the target graphics. The results of the fast calculation method have smaller errors than the accurate results of the strict calculation of the electromagnetic field, and the calculation speed is significantly improved.
[0006] In summary, the Kirchhoff approximation method can be used to quickly obtain the near-field distribution corresponding to the mask pattern. However, as the chip manufacturing process decreases year by year, the pattern spacing of the lithography is smaller than the wavelength of the light source, which causes the Kirchhoff approximation to no longer hold, and the near-field distribution obtained using the Kirchhoff approximation method has low accuracy. The accurate near-field distribution can be calculated using the strict electromagnetic field calculation method, but the amount of calculation is too large, so that its simulation of large-scale near-field distribution will consume unbearable computing time. The existing fast calculation method is based on convolution calculation. First, the electromagnetic field strict calculation method is used to calculate the accurate near-field distribution corresponding to a part of the graphics, and it is used as a training set. Then, using the training set as the standard, the convolution kernel that can reflect the relationship between the graphics and its corresponding near-field distribution is calculated. Finally, the convolution kernel is used to convolve the entire graphics to obtain the near-field distribution corresponding to the target graphics. The results of the fast calculation method have a smaller error than the accurate results of the strict electromagnetic field calculation, and the calculation speed is greatly improved, making it possible to simulate the near-field distribution of large-scale graphics with higher accuracy. However, on the one hand, the accuracy and speed of the calculation using the fast calculation method still have room for further improvement; on the other hand, for DUV lithography, since the spacing of the prepared patterns is much smaller than the wavelength of the light source, the diffraction effect of its near-field distribution is obvious, and a "wavy" field intensity distribution will appear inside the pattern. As shown in Figure 2(a) and Figure 2(b), when using the convolution method for calculation, if the image pixel corresponding to the convolution kernel is completely within a certain pattern, the calculation result will be independent of its position, so that a uniform intensity is obtained inside the pattern, and the "wavy" field intensity distribution caused by the diffraction effect cannot be restored. It can be seen that the fluctuations in the field intensity distribution inside the pattern are not well restored. Summary of the invention
[0007] In view of the above analysis, an embodiment of the present invention aims to provide a method for obtaining diffraction near-field distribution, which can not only improve the accuracy of the calculation result of the diffraction near-field distribution, but also reduce the calculation time.
[0008] The present invention discloses a method for obtaining diffraction near-field distribution, comprising:
[0009] Determine a target mask pattern for which diffraction near-field distribution is to be obtained, and segment the target mask pattern according to the pattern type to obtain target segmented patterns constituting the target mask pattern;
[0010] Determine a target kernel matrix corresponding to the target segmentation pattern, and obtain a target area diffraction near-field distribution corresponding to the target segmentation pattern according to the target segmentation pattern and the target kernel matrix; each column element in the target kernel matrix corresponds to a pixel position of the target segmentation pattern after rearrangement, and represents a weight of a near-field intensity corresponding to the pixel position;
[0011] The diffraction near-field distributions of all target areas are spliced to obtain the diffraction near-field distribution of the target mask pattern.
[0012] Further, the graphic type is a Manhattan graphic, and includes at least one of a convex corner portion, a concave corner portion, a vertical edge portion, and a horizontal edge portion; accordingly, segmenting the target mask graphic according to the graphic type to obtain a target segmented graphic constituting the target mask graphic includes:
[0013] The target mask pattern is segmented according to at least one of the Manhattan patterns, and the obtained target segmented pattern includes at least one of the convex corner portion, the concave corner portion, the vertical edge portion, and the horizontal edge portion.
[0014] Furthermore, obtaining the target area diffraction near-field distribution corresponding to the target segmentation pattern according to the target segmentation pattern and the target kernel matrix includes:
[0015] Matrixing the target segmentation pattern to obtain a target segmentation pattern matrix; the number of rows of the target segmentation pattern matrix is 1, the number of columns is the total number of pixels corresponding to the target segmentation pattern, and the elements of the target segmentation pattern matrix represent whether the pixel position corresponding to the target segmentation pattern is light-transmitting or light-shielding;
[0016] The diffraction near-field distribution of the target area corresponding to the target segmentation pattern is calculated according to the following formula:
[0017] X1·A1=Y1
[0018] Among them, X1 is the target segmentation pattern matrix, A1 is the target kernel matrix, · represents multiplication calculation, and Y1 is the diffraction near-field distribution of the target area.
[0019] Furthermore, before the step of determining a target mask pattern for obtaining a diffraction near-field distribution, the method for obtaining a diffraction near-field distribution further includes:
[0020] Determine whether there is a kernel matrix; each column element in the kernel matrix corresponds to a pixel position after rearrangement of the training set graph, and represents a weight of the near field intensity corresponding to the pixel position;
[0021] If the core matrix exists, the step of determining the target mask pattern of the diffraction near-field distribution to be acquired and subsequent steps are performed.
[0022] Furthermore, the method for obtaining the diffraction near-field distribution also includes:
[0023] If the kernel matrix does not exist, obtaining a training set graph, and calculating the overall reference diffraction near-field distribution of the training set graph;
[0024] Segmenting the training set pattern and the overall reference diffraction near-field distribution corresponding thereto to obtain a segmented mask pattern and a reference diffraction near-field distribution of an area corresponding to the position of the mask pattern;
[0025] The kernel matrix is calculated according to the mask pattern and the reference diffraction near-field distribution.
[0026] Further, the calculating the kernel matrix according to the mask pattern and the reference diffraction near-field distribution includes:
[0027] Matrixing the mask pattern to obtain a mask pattern matrix; the number of rows in the mask pattern matrix is 1, the number of columns is the total number of pixels corresponding to the mask pattern, and the elements of the mask pattern matrix represent whether the pixel position corresponding to the mask pattern is light-transmitting or light-shielding;
[0028] Matrixing the reference diffraction near-field distribution to obtain a reference diffraction near-field distribution matrix; the number of rows in the reference diffraction near-field distribution matrix is 1, the number of columns is the total number of pixels in the area corresponding to the mask pattern position, and the elements of the reference diffraction near-field distribution matrix represent the near-field intensity at the pixel position;
[0029] The kernel matrix is calculated according to the following formula:
[0030] A=(X T X) -1 X T Y
[0031] Where A is the kernel matrix, X is the mask pattern matrix, X T is the transposed matrix of X, (X T X) -1 Represents X T The inverse of X, Y, is the reference diffraction near-field distribution matrix.
[0032] Further, obtaining the overall reference diffraction near-field distribution includes:
[0033] The overall reference diffraction near-field distribution is obtained by calculation according to a rigorous electromagnetic field calculation method.
[0034] Furthermore, the kernel matrix is calculated according to the following formula:
[0035] A=(X T X) -1 X T Y
[0036] Where A is the kernel matrix, X is the mask pattern matrix, X T is the transposed matrix of X, (X T X) -1 Represents X T The inverse of X, Y is the reference diffraction near-field distribution matrix, including:
[0037] The kernel matrix is obtained by using the least square method and calculating according to the above formula.
[0038] Further, the determining of the target kernel matrix corresponding to the target segmentation graph includes:
[0039] A target kernel matrix corresponding to the target segmentation graph is selected from the kernel matrices.
[0040] Furthermore, the method for obtaining the diffraction near-field distribution also includes:
[0041] If the mask pattern is a convex corner portion or a concave corner portion, performing a circumscribed minimum square process on the convex corner portion or the concave corner portion;
[0042] The image except the convex corner part or the concave corner part in the minimum square is marked, and the marked image is no longer calculated.
[0043] Compared with the prior art, the present invention can achieve at least one of the following beneficial effects:
[0044] The method for obtaining the diffraction near-field distribution provided in the embodiment of the present invention determines the target mask pattern for which the diffraction near-field distribution is to be obtained, and segments the target mask pattern according to the pattern type to obtain the target segmented pattern constituting the target mask pattern; determines the target core matrix corresponding to the target segmented pattern, and obtains the target area diffraction near-field distribution corresponding to the target segmented pattern according to the target segmented pattern and the target core matrix; and splices all the target area diffraction near-field distributions to obtain the diffraction near-field distribution of the target mask pattern. The method can not only improve the accuracy of the calculation result of the diffraction near-field distribution, but also reduce the calculation time.
[0045] In the present invention, the above-mentioned technical solutions can also be combined with each other to achieve more preferred combination solutions. Other features and advantages of the present invention will be described in the subsequent description, and some advantages can become obvious from the description, or can be understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained through the contents particularly pointed out in the description and the drawings. BRIEF DESCRIPTION OF THE DRAWINGS
[0046] The drawings are only for the purpose of illustrating particular embodiments and are not to be considered limiting of the present invention. Like reference symbols denote like components throughout the drawings.
[0047] Figure 1 It is a schematic diagram of a prior art lithography model;
[0048] FIG2( a ) is a schematic diagram of the “ripple effect” of the near-field distribution due to diffraction;
[0049] FIG2( b ) is a schematic diagram of the result calculated using the existing convolution method;
[0050] Figure 3 It is a schematic flow chart of a method for obtaining diffraction near-field distribution in an embodiment of the present invention;
[0051] Figure 4 Schematic diagram for explaining the convex corner part and the concave corner part in the embodiment of the present invention;
[0052] Figure 5 It is a schematic flow chart of a method for obtaining diffraction near-field distribution in another embodiment of the present invention;
[0053] FIG6( a ) is a schematic diagram of accurate results obtained using the electromagnetic field rigorous calculation method in an embodiment of the present invention;
[0054] FIG6( b ) is a schematic diagram of simulation results obtained using the existing convolution method in an embodiment of the present invention;
[0055] FIG6( c ) is a schematic diagram of a simulation result obtained using the method of the embodiment of the present invention in the embodiment of the present invention. DETAILED DESCRIPTION
[0056] The preferred embodiments of the present invention are described in detail below in conjunction with the accompanying drawings, wherein the accompanying drawings constitute a part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not used to limit the scope of the present invention.
[0057] A specific embodiment of the present invention discloses a method for obtaining diffraction near-field distribution, the flow chart of which is as follows: Figure 1 As shown, the following steps are included:
[0058] Step S1: determining a target mask pattern for which diffraction near-field distribution is to be obtained, and segmenting the target mask pattern according to pattern types to obtain target segmented patterns constituting the target mask pattern.
[0059] Step S2: Determine a target kernel matrix corresponding to the target segmentation pattern, and obtain the target area diffraction near-field distribution corresponding to the target segmentation pattern based on the target segmentation pattern and the target kernel matrix; each column element in the target kernel matrix corresponds to the pixel position of the target segmentation pattern after rearrangement, and represents the weight of the near-field intensity corresponding to the pixel position.
[0060] Step S3: splicing the diffraction near-field distributions of all target areas to obtain the diffraction near-field distribution of the target mask pattern.
[0061] In step S1, the graph type may be a Manhattan graph, and may include at least one of a convex corner portion, a concave corner portion, a vertical edge portion, and a horizontal edge portion. Figure 4 As shown, the convex corner part is Figure 4 The dotted box 2 marks the concave corners. Figure 4 The dashed box 1 is marked.
[0062] Further, segmenting the target mask pattern according to the pattern type to obtain the target segmented pattern constituting the target mask pattern includes:
[0063] The target mask pattern is segmented according to at least one of the Manhattan patterns, and the target segmented pattern obtained includes at least one of the convex corner portion, the concave corner portion, the vertical edge portion, and the horizontal edge portion. Referring to the above description, the target mask pattern is segmented using the convex corner portion, the concave corner portion, the vertical edge portion, and the horizontal edge portion.
[0064] The purpose of segmenting the target mask pattern according to the pattern type is to make the target mask patterns independent of each other and be one of the above-mentioned convex corner portion, concave corner portion, vertical edge portion, and horizontal edge portion.
[0065] In step S2, the segmented convex corner part, concave corner part, vertical edge part, and horizontal edge part have target kernel matrices corresponding to their respective parts. The target kernel matrix can be expressed as (a1 2 ,b1 2 ), indicating that there is a1 2 rows, indicating the total number of pixels corresponding to the target segmentation graph, and b1 2 Columns represent the total number of pixels in the area corresponding to the mask pattern position, where the mask pattern is obtained by segmenting the training set pattern, and the area corresponding to the mask pattern position has a corresponding reference diffraction near-field distribution.
[0066] The target segmentation graph is rearranged into the target segmentation graph matrix (1, a1 2 ), that is, row 1 a1 2 columns, the i-th column element in the target kernel matrix is equal to a1 2 The pixel position of the ith column in the columns corresponds to the pixel position of the ith column, which represents the weight of the near-field intensity corresponding to the pixel position of the ith column. The weight of the near-field intensity is obtained by training the kernel matrix based on the benchmark diffraction near-field distribution. The benchmark diffraction near-field distribution can be calculated by the most accurate method for calculating the diffraction near-field distribution.
[0067] The target area diffraction near-field distribution can be obtained by multiplying the target segmentation graph and the target kernel matrix, that is, the target area diffraction near-field distribution can be expressed by the target area diffraction near-field distribution matrix, which can be expressed as (1, b1 2 ), where the element represents the near-field intensity value of the corresponding pixel position of the diffraction near-field of the target area.
[0068] Further, according to the target segmentation pattern and the target core matrix, a target area diffraction near-field distribution corresponding to the target segmentation pattern is obtained, including:
[0069] The target segmentation pattern is matrixed to obtain a target segmentation pattern matrix; the number of rows of the target segmentation pattern matrix is 1, the number of columns is the total number of pixels corresponding to the target segmentation pattern, and the elements of the target segmentation pattern matrix represent whether the pixel position corresponding to the target segmentation pattern is light-transmissive or light-shielding; taking the target segmentation pattern as a square as an example, the total number of pixels corresponding to the target segmentation pattern can be represented by a1*a1.
[0070] The target segmentation matrix can be expressed as (1, a1 2 ), where "1" means 1 row, "a1 2 " indicates that the number of columns is a1 2 , the elements in the matrix can be 1 or 0, which respectively indicate that the corresponding pixel position of the target segmentation pattern is light-transmitting or light-shielding (i.e., the light is blocked).
[0071] The diffraction near-field distribution of the target area corresponding to the target segmentation pattern is calculated according to the following formula:
[0072] X1·A1=Y1
[0073] Where X1 is the target segmentation pattern matrix, A1 is the target kernel matrix, · represents multiplication calculation, and Y1 is the target area diffraction near field distribution. Y1 can be represented by the target area diffraction near field distribution matrix, in which the elements represent the near field intensity value of the corresponding pixel position of the target area diffraction near field. The target kernel matrix is a matrix with a size of (a1 2 ,b1 2 ) can be described with reference to the above embodiment.
[0074] Furthermore, if Figure 5 As shown, before the step of determining the target mask pattern of the diffraction near-field distribution to be obtained, the method for obtaining the diffraction near-field distribution further includes:
[0075] Determine whether there is a kernel matrix; each column element in the kernel matrix corresponds to the pixel position after rearrangement of the training set graphics, and represents the weight of the near-field intensity corresponding to the pixel position; whether there is a kernel matrix can be understood as, if the weight of the near-field intensity corresponding to the pixel position is obtained, it means that the kernel matrix training is completed, and the kernel matrix exists at this time, otherwise the kernel matrix does not exist. The kernel matrix of the embodiment of the present invention refers to the kernel matrix with trained weights.
[0076] The training set graph can be understood as a set of standard graphs used for training, which are Manhattan graphs, including vertical rectangles and horizontal rectangles. The convex corners and concave corners of the graph can be obtained. The convex corners and concave corners can be formed at the intersection of the horizontal rectangle and the vertical rectangle, and points are selected at a certain distance on the horizontal and vertical edges. Based on these points, the entire graph is decomposed and summarized into four categories: convex corner part, concave corner part, vertical edge part, and horizontal edge part.
[0077] The exact near-field distribution corresponding to these graphs can be calculated using rigorous electromagnetic field calculation methods.
[0078] The description of the kernel matrix can refer to the description of the target kernel matrix above, which will not be repeated here.
[0079] If the core matrix exists, the step of determining the target mask pattern of the diffraction near-field distribution to be acquired and subsequent steps are performed.
[0080] Further, the determining of the target kernel matrix corresponding to the target segmentation graph includes:
[0081] A target kernel matrix corresponding to the target segmentation figure is selected from the kernel matrix. For example, if the target segmentation figure is a concave corner part, a kernel matrix corresponding to the concave corner part is selected and used as the target kernel matrix corresponding to the concave corner part.
[0082] Furthermore, the method for obtaining the diffraction near-field distribution also includes:
[0083] If the kernel matrix does not exist, the training set graphics are obtained, and the overall benchmark diffraction near-field distribution of the training set graphics is calculated; since the training set graphics are a set of standard graphics, the overall benchmark diffraction near-field distribution corresponding to this set of standard graphics can be calculated at one time.
[0084] The training set graphics and the corresponding overall reference diffraction near-field distribution are segmented to obtain the segmented mask graphics and the reference diffraction near-field distribution of the area corresponding to the position of the mask graphics; the instructions for segmenting the training set graphics can refer to the instructions for segmenting the target mask graphics above, and will not be repeated here.
[0085] The kernel matrix is calculated according to the mask pattern and the reference diffraction near-field distribution.
[0086] Further, the calculating the kernel matrix according to the mask pattern and the reference diffraction near-field distribution includes:
[0087] The mask pattern is matrixed to obtain a mask pattern matrix; the number of rows in the mask pattern matrix is 1, the number of columns is the total number of pixels corresponding to the mask pattern, and the elements of the mask pattern matrix represent whether the pixel position corresponding to the mask pattern is transparent or shading; the mask pattern can be represented by X, and it can be transformed into a row to represent (1, a 2 ), whose elements are 1 and 0, indicating that the corresponding position of the mask pattern is light-transmitting or light-shielding.
[0088] It should be noted that if the mask pattern is a vertical edge portion or a horizontal edge portion, the total number of pixels corresponding to the vertical edge portion or the horizontal edge portion can be expressed by the area of the minimum circumscribed rectangle or the minimum circumscribed square corresponding to the vertical edge portion or the horizontal edge portion, such as the length multiplied by the width.
[0089] Furthermore, the method for obtaining the diffraction near-field distribution also includes:
[0090] like Figure 4 As shown, if the mask pattern is a convex corner portion or a concave corner portion, a circumscribed minimum square process is performed on the convex corner portion or the concave corner portion;
[0091] The images other than the convex corners or concave corners in the minimum square are marked, and the marked images are no longer calculated, so that the marked images are no longer calculated, saving computing resources.
[0092] The reference diffraction near-field distribution is matrixed to obtain a reference diffraction near-field distribution matrix; the number of rows of the reference diffraction near-field distribution matrix is 1, the number of columns is the total number of pixels in the area corresponding to the mask pattern position, and the elements of the reference diffraction near-field distribution matrix represent the near-field intensity of the pixel position; the area corresponding to the mask pattern position can be understood as the same as the mask pattern position, and the reference diffraction near-field distribution has been calculated and transformed into a row representation, the size of the area is b*b, then after transformation, Y is (1, b 2 )-sized matrix.
[0093] The above a and b may be the same or different. Since the purpose of the present invention is to calculate the near field distribution through a given mask pattern, the area of the mask pattern selected logically should cover the area of the calculated near field distribution (i.e., preferably a>b). A is the kernel matrix corresponding to the pattern type (convex angle, concave angle, vertical edge, horizontal edge) in the mask pattern, and the specific description is not repeated here.
[0094] The kernel matrix is calculated according to the following formula:
[0095] A=(X T X) -1 X T Y
[0096] Where A is the kernel matrix, X is the mask pattern matrix, X T is the transposed matrix of X, (X T X) -1 Represents X T The inverse of X, Y, is the reference diffraction near-field distribution matrix.
[0097] Furthermore, the kernel matrix can be obtained by using the least squares method and calculating according to the above formula. The least squares method is a conventional algorithm and will not be described in detail.
[0098] Further, obtaining the overall reference diffraction near-field distribution includes:
[0099] The reference diffraction near-field distribution is obtained by calculating according to the electromagnetic field rigorous calculation method. The electromagnetic field rigorous calculation method is a relatively accurate method for calculating the diffraction near-field distribution at present, and will not be described in detail.
[0100] In step S3, all target area diffraction near-field distributions corresponding to each target segmentation pattern after segmentation may be spliced according to the positions of each target segmentation pattern before segmenting the target mask pattern, thereby obtaining the diffraction near-field distribution of the target mask pattern.
[0101] In an embodiment of the present invention, the target segmentation pattern is multiplied by the target kernel matrix, which can effectively solve the problem of being unable to restore the "wavy" field intensity distribution caused by the diffraction effect, thereby improving the accuracy of restoration. Moreover, the calculation object is the kernel matrix, rather than a large number of convolution kernels, which greatly reduces the amount of calculation and thus can reduce the calculation time.
[0102] The fast algorithm for DUV lithography diffraction near-field distribution proposed in the present invention can greatly improve the calculation accuracy of DUV lithography near-field distribution while ensuring high calculation efficiency. Compared with the existing DDM algorithm, the operation time of this method is reduced to a certain extent, and the absolute error can be reduced by dozens of times. This method can be applied in the OPC process to improve the calculation efficiency and accuracy of the diffraction near-field distribution, which will ultimately improve the resolution of the lithography pattern and save calculation time, as shown in Table 1:
[0103] Table 1
[0104]
[0105]
[0106] As shown in FIG. 6( a ), FIG. 6( b ) and FIG. 6( c ), it can be determined that the effect of FIG. 6( c ) is closest to the effect of FIG. 6( a ) and is much better than the effect of FIG. 6( b ).
[0107] The present invention uses a kernel matrix method to replace the existing single convolution kernel method. After using the least squares method to calculate the kernel matrix corresponding to the near-field distribution, a matrix multiplication is used to directly obtain the near-field distribution of an area, rather than using the convolution method to calculate the near-field intensity of each point in an area one by one. The kernel matrix method can well restore the "corrugated" field intensity distribution caused by the diffraction effect, improve the accuracy of the calculation results, and reduce the calculation time.
[0108] Compared with the prior art, the method for obtaining the diffraction near-field distribution provided in the embodiment of the present invention determines the target mask pattern for which the diffraction near-field distribution is to be obtained, and segments the target mask pattern according to the pattern type to obtain the target segmented pattern constituting the target mask pattern; determines the target core matrix corresponding to the target segmented pattern, and obtains the target area diffraction near-field distribution corresponding to the target segmented pattern according to the target segmented pattern and the target core matrix; splices all the target area diffraction near-field distributions to obtain the diffraction near-field distribution of the target mask pattern. This method can not only improve the accuracy of the calculation result of the diffraction near-field distribution, but also reduce the calculation time.
[0109] Those skilled in the art will appreciate that all or part of the processes of the above-mentioned embodiments can be implemented by instructing related hardware through a computer program, and the program can be stored in a computer-readable storage medium, wherein the computer-readable storage medium is a disk, an optical disk, a read-only storage memory, or a random access memory, etc.
[0110] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by any technician familiar with the technical field within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.
Claims
1. A method for obtaining diffraction near-field distribution, It is characterized in that include: Determine if a nuclear matrix exists; Each column element in the kernel matrix corresponds to a pixel position after rearrangement of the training set graph, and represents the weight of the near field intensity corresponding to the pixel position; If the kernel matrix does not exist, a training set pattern is obtained, and the overall reference diffraction near-field distribution of the training set pattern is calculated; the training set pattern and the overall reference diffraction near-field distribution corresponding thereto are segmented to obtain the segmented mask pattern and the reference diffraction near-field distribution of the area corresponding to the position of the mask pattern; Calculating the kernel matrix according to the mask pattern and the reference diffraction near-field distribution; Determine a target mask pattern for which diffraction near-field distribution is to be obtained, and segment the target mask pattern according to the pattern type to obtain target segmented patterns constituting the target mask pattern; A target kernel matrix corresponding to the target segmentation pattern is selected from the kernel matrix, and a target area diffraction near-field distribution corresponding to the target segmentation pattern is obtained according to the target segmentation pattern and the target kernel matrix; each column element in the target kernel matrix corresponds to a pixel position of the target segmentation pattern after rearrangement, and represents a weight of a near-field intensity corresponding to the pixel position; The diffraction near-field distributions of all target areas are spliced to obtain the diffraction near-field distribution of the target mask pattern.
2. The method for obtaining diffraction near-field distribution according to claim 1, It is characterized in that The graphic type is a Manhattan graphic, and includes at least one of a convex corner portion, a concave corner portion, a vertical edge portion, and a horizontal edge portion; accordingly, segmenting the target mask graphic according to the graphic type to obtain a target segmented graphic constituting the target mask graphic includes: The target mask pattern is segmented according to at least one of the Manhattan patterns, and the obtained target segmented pattern includes at least one of the convex corner portion, the concave corner portion, the vertical edge portion, and the horizontal edge portion.
3. The method for obtaining diffraction near-field distribution according to claim 1, It is characterized in that The step of obtaining the target area diffraction near-field distribution corresponding to the target segmentation pattern according to the target segmentation pattern and the target core matrix includes: Matrixing the target segmentation pattern to obtain a target segmentation pattern matrix; the number of rows of the target segmentation pattern matrix is 1, the number of columns is the total number of pixels corresponding to the target segmentation pattern, and the elements of the target segmentation pattern matrix represent whether the pixel position corresponding to the target segmentation pattern is light-transmitting or light-shielding; The diffraction near-field distribution of the target area corresponding to the target segmentation pattern is calculated according to the following formula: X1·A1=Y1 Among them, X1 is the target segmentation pattern matrix, A1 is the target core matrix, · represents multiplication calculation, and Y1 is the diffraction near-field distribution of the target area.
4. The method for obtaining diffraction near-field distribution according to any one of claims 1 to 3, It is characterized in that If the core matrix exists, the step of determining the target mask pattern of the diffraction near-field distribution to be acquired and subsequent steps are performed.
5. The method for obtaining diffraction near-field distribution according to claim 1, It is characterized in that The step of calculating the kernel matrix according to the mask pattern and the reference diffraction near-field distribution comprises: Matrixing the mask pattern to obtain a mask pattern matrix; the number of rows in the mask pattern matrix is 1, the number of columns is the total number of pixels corresponding to the mask pattern, and the elements of the mask pattern matrix represent whether the pixel position corresponding to the mask pattern is light-transmitting or light-shielding; Matrixing the reference diffraction near-field distribution to obtain a reference diffraction near-field distribution matrix; the number of rows in the reference diffraction near-field distribution matrix is 1, the number of columns is the total number of pixels in the area corresponding to the mask pattern position, and the elements of the reference diffraction near-field distribution matrix represent the near-field intensity at the pixel position; The kernel matrix is calculated according to the following formula: A=(X T X) -1 X T Y Where A is the kernel matrix, X is the mask pattern matrix, X T is the transposed matrix of X, (X T X) -1 Represents X T The inverse of X, Y, is the reference diffraction near-field distribution matrix.
6. The method for obtaining diffraction near-field distribution according to claim 1, It is characterized in that Obtaining the overall reference diffraction near-field distribution includes: The overall reference diffraction near-field distribution is obtained by calculation according to a rigorous electromagnetic field calculation method.
7. The method for obtaining diffraction near-field distribution according to claim 5, It is characterized in that The kernel matrix is calculated according to the following formula: A=(X T X) -1 X T Y Where A is the kernel matrix, X is the mask pattern matrix, X T is the transposed matrix of X, (X T X) -1 Represents X T The inverse of X, Y is the reference diffraction near-field distribution matrix, including: The kernel matrix is obtained by using the least square method and calculating according to the above formula.
8. The method for obtaining diffraction near-field distribution according to claim 4, It is characterized in that The step of determining a target kernel matrix corresponding to the target segmentation graph includes: A target kernel matrix corresponding to the target segmentation graph is selected from the kernel matrices.
9. The method for obtaining diffraction near-field distribution according to claim 5, It is characterized in that The method for obtaining the diffraction near-field distribution also includes: If the mask pattern is a convex corner portion or a concave corner portion, performing a circumscribed minimum square process on the convex corner portion or the concave corner portion; The image except the convex corner part or the concave corner part in the minimum square is marked, and the marked image is no longer calculated.
Citation Information
Patent Citations
Method for determining illumination intensity distribution of light source and mask pattern of photo-etching process
CN102169295A
Quick three-dimensional mask diffraction near-field calculation method based on sample library and data fitting
CN105045033A
Method for detecting phase type defect of extreme ultraviolet mask based on polarization changes of scattered light
CN106290372A
Diffraction optical element manufacturing method, diffraction optical element and reticle mask used in the manufacturing method
JP2007155927A
