Modeling method of OPC correction model and OPC correction method

Abstract

The invention discloses a modeling method of an OPC correction model and an OPC correction method for performing OPC correction on a to-be-corrected graph based on the OPC correction model established by the modeling method. The method comprises the following steps: establishing OPC test patterns with different pattern densities, performing a wafer photoetching process to obtain corresponding actual pattern CD values under different pattern densities, calculating a difference value between an actual value and a preset value, drawing a relation curve, obtaining a deviation value correction table under different pattern densities, and establishing an OPC correction model and the deviation value correction table; according to the OPC correction method, a to-be-corrected graph is pre-scanned on the basis of an established OPC correction model to obtain the graph density of the to-be-corrected graph, the correction amount of the to-be-corrected graph is matched in the OPC correction model according to the graph density, then OPC correction is carried out again, and process implementation is carried out on the basis of the corrected OPC graph. According to the OPC modeling and OPC correction, the correction parameter based on the graph density is introduced, but the graph density parameter does not directly serve as a third dimension during modeling and correction to participate in calculation, and the calculation amount is reduced.

Description

Technical Field

[0001] The present invention relates to the field of semiconductor device manufacturing processes, and in particular to a method for establishing a model in the OPC process and a method for performing OPC correction using the established model. Background Art

[0002] In the semiconductor manufacturing process, as the process node becomes smaller and smaller, the layout pattern density will become higher and higher. In order to better transfer the integrated circuit pattern on the mask to the silicon wafer through lithography exposure, it is often necessary to use the method of optical proximity correction (OPC: Optical Proximity Correction) to correct the layout pattern on the mask. Optical proximity correction is a technology to increase the resolution of the lithography process. In the process of lithography, ideally, the image pattern on the silicon wafer should be exactly the same as the layout design on the photomask. Unfortunately, in the deep sub-micron semiconductor manufacturing process, when the critical dimension of the photomask pattern is less than the exposure wavelength, due to the fact that the size of the pattern and the size of the lithography wavelength are close or even smaller, due to reasons such as light interference, diffraction, and development, the pattern exposed on the photoresist will be inconsistent with the pattern on the mask layout, resulting in distortion of the optical proximity effect (OPE). The imaging on the silicon wafer will be distorted, thus not matching well with the layout pattern of the photomask, resulting in pattern distortion. Therefore, in order to form the required pattern on the photoresist, it is necessary to perform OPC correction on the pattern on the mask layout.

[0003] The process of modeling is to expose multiple sets of sample data, analyze each set of sample data, and use multiple sets of sample data to correct the OPC model (auxiliary patterns are also included in the process of collecting sample data). Finally, the mapping relationship between the actual critical dimension of the main pattern Pi on the silicon wafer and the pattern density change of the OPC test pattern is obtained, which is to establish the OPC model.

[0004] Most of the existing OPC correction models do not consider the influence of pattern density, and neither the corrected OPC model (Model) nor the menu (Recipe) can reflect the actual influence of pattern density on pattern accuracy.

[0005] Since neither the OPC model nor the menu considers the influence of pattern density, there is no requirement for the pattern density parameter when collecting data during current OPC modeling. However, with the continuous reduction of feature sizes due to technological progress, when the lithography process approaches the resolution limit, the influence of pattern density, especially local pattern density, cannot be ignored.

[0006] An OPC correction method considering pattern density proposed by the applicant in CN115598922 put forward a modeling method that uses OPC test patterns with different pattern densities to establish an OPC correction model capable of performing different corrections according to different pattern densities.

[0007] The OPC correction algorithm transforms the Mask (reticle) pattern M(x, y) into the corresponding photoresist pattern M'(x, y) through the TCC transfer matrix (using the mathematical expression of the TCC transfer matrix to characterize the deformation of the original pattern M by optical diffraction, chemical diffusion, and development processes).

[0008] M′(x,y) = TCC * M(x,y)

[0009] The OPC modeling method is to measure the actual topography M'(x, y) of the corresponding photoresist through a large number of test patterns M(x, y), and then use their difference δ(x, y) and optical system parameters (a series of dozens of process-related parameters such as D - dose, NA - numerical aperture, σ - coherence coefficient, L - diffusion length, Eth - reaction threshold energy, etc.) to inversely calculate and determine the TCC, thereby establishing a mathematical model based on TCC to predict M'(x, y) from M(x, y). This calculation usually cannot obtain an analytical solution, but determines all components of the TCC by solving a numerical solution with a computer to minimize the root mean square difference and max(δ(x, y)) of all δ(x, y).

[0010] If the pattern density η is directly introduced into the TCC transfer matrix, and the pattern density η variable is added to M(x, y) and M'(x, y) to become M(x, y, η) and M'(x, y, η), as follows:

[0011] M′(x,y,η) = TCC * M(x,y,η)

[0012] After considering the pattern density, calculate the difference δ:

[0013] δ i (x,y) = M i ′(x,y) - M i (x,y)

[0014] δ i (x,y,η) = M i ′(x,y,η) - M i (x,y,η)

[0015] The relevant variables in the TCC transfer matrix include at least the following components:

[0016] Optical imaging (4F imaging system Fourier transform / Hopkins transform);

[0017] Chemical diffusion (Gaussian equivalent diffusion);

[0018] Chemical threshold reaction (LPM Model……);

[0019]

[0020] TCC transfer matrix

[0021] Therefore, if the pattern density η is directly introduced into the TCC transfer matrix and the pattern density variables are added to M(x, y) and M'(x, y) to become M(x, y, η) and M'(x, y, η), the TCC transfer matrix will change from two - dimensional to three - dimensional, and the computational complexity during OPC modeling will increase exponentially, consuming a lot of time and effort. Summary of the Invention

[0022] The technical problem to be solved by the present invention is to provide a modeling method for an OPC correction model.

[0023] The present invention also provides a correction method for OPC correction according to the established OPC correction model.

[0024] A modeling method for an OPC correction model according to the present invention includes the following steps:

[0025] a). Establish multiple groups of OPC test patterns with different pattern densities, and use the OPC test patterns for process processing to obtain wafer experimental pieces for modeling;

[0026] b). Measure the CD values of different pattern densities and the differences between the measured values and the design / theoretical values based on the wafer experimental pieces;

[0027] c). Determine a stable pattern density interval, determine the standard CD and standard difference, and establish an OPC test model OPC model A and a standard deviation value table SSA Table S based on these;

[0028] d). Establish a target value table SSA table corresponding to different pattern densities according to steps b and c.

[0029] Further, the modeling method specifically includes:

[0030] Step 1, design N groups of OPC test patterns with different pattern densities η, where N is a natural number greater than or equal to 2; each group of OPC test patterns with different pattern densities η includes at least one main pattern Pi and at least one Dummy pattern Ai arranged around the main pattern Pi and having a different design distance from it, so that the OPC test patterns P1 - PN have different pattern densities η1 - ηN; a preset pattern area PX includes several test patterns M(x, y) related to the position coordinates (x, y);

[0031] Step 2: Use the OPC test pattern obtained in Step 1 to perform exposure and development of the lithography process, and measure the corresponding post-lithography photoresist pattern M'(x, y) of M(x, y) in the OPC test patterns P1 to PN in the preset pattern area PX at different pattern densities η.

[0032] Step 3: Calculate the position deviation values ηδ i (x, y), where i ∈ (1 to N);

[0033] Step 4: Plot the relationship curve between the position deviation value ηδ i (x, y) and the pattern density η, and find the pattern density η within the range where ηδ i (x, y) meets the OPC modeling accuracy requirement δ;

[0034] Step 5: Based on ηδ i (x, y) obtained in Step 4, find the deviation values of ηδ i (x, y) at different pattern densities η;

[0035] Step 6: Based on Step 4, with ηδ i (x, y) in Step 4 as the reference, establish the OPC model OPC model A and the SSA table;

[0036] And based on Step 4, take the set of CD values in the relatively stable pattern density η interval as the standard value, and establish OPC modeling for different line widths and space periods, including the standard target value table SSA table S;

[0037] Step 7: Use ηδ i (x, y) and the SSA table S to establish the deviation value correction table SSAtable(ηi) at different pattern densities η.

[0038] Furthermore, in Step 1, the N groups of the OPC test patterns are N groups of OPC test patterns with different pattern densities, where N ≥ 2; and each group of the OPC test patterns with different pattern densities contains at least one same main pattern Pi and at least one dummy pattern Ai set around the main pattern Pi and having a different design distance or different total area from it; using the OPC test pattern, establish an optical proximity correction model in which the critical dimension of the chip layout pattern changes with the pattern density.

[0039] Further, the graphic density η refers to the local graphic density and / or the global graphic density; the local graphic density refers to the ratio of the sum of the design areas of all the auxiliary graphics Ai within an arbitrarily defined preset area in an OPC graphic to the area of the preset area: the arbitrary graphic density η i refers to the arbitrary graphic density not included in the actual layout specifically determined by the N groups of OPC test graphic densities.

[0040] Further, the dummy graphics Ai have the same or different sizes, areas, or morphologies; the dummy graphics Ai in different groups of the OPC test graphics P1 to PN have different distances from the OPC test graphics P1 to PN, and / or, the dummy graphics Ai in different groups of the OPC test graphics P1 to PN have different total areas, so as to adjust and form different graphic densities η1 to ηN.

[0041] Further, the modeling accuracy δ is determined according to the product design redundancy and is designed to be 1% to 5% of the minimum CD value of this graphic layer.

[0042] Further, the SSA table or SSA table (ηi) is a matrix, where the independent variables in the horizontal and vertical directions are the line widths and spatial periods or line spacings in the layout graphics, and the matrix elements at the intersection positions are the target values or target moving distances for this combination.

[0043] Further, in step 6, take the CD values of the group with the relatively stable graphic density η interval as the standard values. The standard values of the CD values are one group, or the average of several groups within the stable graphic density interval;

[0044] The definition of the stable graphic density η interval includes the following situations:

[0045] a) Within this graphic density η interval, the CD values and deviation values of different graphic densities meet the accuracy requirements of OPC modeling;

[0046] b) Within this graphic density η interval, any graphic density can be selected as a representative for modeling;

[0047] c) Within this graphic density η interval, the CD values corresponding to any number of graphic density intervals can be averaged, and the deviation values are calculated for the averaged CD values, and modeling is performed with the averaged CD values and the deviation values corresponding to the averaged CD values.

[0048] Further, the SSA table is composed of one or no less than two SSA tables; each SSA table can set its applicable graphic environment, including corresponding different graphic densities.

[0049] Further, the SSA table does not contain specific CD values, but positive and negative correction amounts under different graphic densities.

[0050] An OPC correction method includes:

[0051] Using the OPC correction model established by the foregoing OPC modeling method, the OPC correction model is a target value table SSA table and a standard target value table SSA table S including a correction function relationship between the actual graphic CD value and its corresponding graphic density η variable;

[0052] Using the target value table SSA table and the standard target value table SSA table S to perform OPC correction on the OPC graphics of the pre-fabrication layout based on the correction information of the graphic density ηi to be corrected to obtain a target layout, and then performing a normal mass production lithography process on the target layout to obtain a corrected ideal wafer graphic.

[0053] Further, before performing OPC correction on the graphic to be corrected using the OPC correction model, a scanning calculation step is first performed on the graphic to be corrected to determine the value of the graphic density ηi to be corrected, including:

[0054] Scanning the graphic M(x, y) to be corrected within the coordinate ranges X and Y, and measuring the graphic density ηi of the X*Y area range of the graphic to be corrected.

[0055] Further, according to the measured graphic density ηi of the graphic to be corrected, calling the OPC correction model, including OPC model A and the target value table SSA table or the standard target value table SSA table S, to correct the target graphic M(x, y) to obtain the graphic M”(x, y);

[0056] Calling the deviation value correction table SSA table(ηi) corresponding to the graphic density ηi to correct the graphic M”(x, y), and adjusting it to the target value M’(x, y), that is, the OPC correction is completed.

[0057] Further, the graphic density ηi is not used as the third dimension during modeling / correction, and the traditional two-dimensional plane x, y dimensions are still retained. The graphic density is a correction amount after the correction of the existing OPC technology.

[0058] Furthermore, the target value table SSA table is a matrix, where the independent variables in the horizontal and vertical directions are line width and spatial period or line pitch, and the matrix element at the intersection position is the target value or target movement distance of this combination.

[0059] Furthermore, the target value table SSA table consists of one or no less than two parts; when there are no less than two, each SSA table can be applicable to different graphic environments.

[0060] An OPC correction system includes an OPC model establishment module that establishes a functional relationship between the graphic density η of a to-be-corrected graphic and the CD value of the actually formed device based on the graphic density η, and establishes an OPC correction model based on the connection between the functional relationship and the variables of the graphic density η, including OPC model A, target value table SSA table, and standard target value table SSA table S.

[0061] The OPC correction module uses the OPC correction model to perform graphic density scanning calculation on the to-be-corrected graphic to obtain the graphic density ηi of the to-be-corrected graphic, compares and searches the graphic density ηi with the OPC correction model and the deviation value correction table to obtain the correction amount of the to-be-corrected graphic, and then performs OPC correction on the to-be-corrected graphic.

[0062] The modeling method of the OPC correction model of the present invention, by establishing an OPC correction model including graphic density parameters, and the graphic density parameters do not directly participate in the operation of the TCC transfer matrix, and the graphic density parameters do not serve as the third dimension of the modeling calculation, so that the calculation amount in the process of OPC modeling and OPC correction remains a two-dimensional operation, with a small OPC modeling calculation amount, and at the same time having high modeling accuracy and good correction effect. Brief Description of the Drawings

[0063] Figure 1 is a schematic diagram of the OPC correction method for traditional sub-resolution assist features.

[0064] Figure 2 is a schematic structural diagram of an OPC test pattern provided in an embodiment of the present invention, where an assist feature is set near the main graphic to consider the influence of the one-dimensional graphic density near the test pattern on its critical dimension.

[0065] Figure 3 is a schematic structural diagram of an OPC test pattern provided in an embodiment of the present invention, where a set area is framed around the main graphic to consider the influence of the two-dimensional graphic density in the surrounding area of the test pattern on its critical dimension.

[0066] Figure 4It shows the CD values and DOF (depth of focus) of the same pattern under different pattern densities and different substrate conditions, and is a schematic diagram for obtaining the corresponding CD under a relatively stable pattern density range as the standard value according to the modeling accuracy requirements.

[0067] Figure 5 It is the target value table SSA table established by the present invention and the flowchart for performing OPC correction in the actual process based on the established OPC correction model. Detailed implementation manners

[0068] In order to make the content of the present invention clearer and easier to understand, the content of the present invention will be described in detail below in combination with specific embodiments and drawings. However, the technical content involved in the present invention is not limited to the specific embodiments given.

[0069] The present invention will be further described in detail below in combination with the drawings and specific embodiments. The advantages and features of the present invention will be clearer according to the following description and the claims. It should be noted that the drawings are all in a very simplified form and use non-precise ratios, only for conveniently and clearly assisting in explaining the purpose of the embodiments of the present invention.

[0070] As described in the technical background part of the present invention, directly changing the pattern density into three dimensions results in a huge amount of calculation and is not advisable. Therefore, it is changed to a bias table method by sampling and extraction. When modeling, the following test patterns are used: set N groups of OPC test patterns with different pattern densities, where N is greater than or equal to 2; each group of OPC test patterns with different pattern densities includes at least one main pattern Pi and at least one dummy auxiliary pattern Ai arranged around the main pattern Pi and having a different design distance from it. Among them, the main patterns of each group of OPC test patterns are the same, and different sizes of dummy auxiliary patterns Ai are paired to form different pattern densities. For details, reference can be made to CN115598922A. Using multiple groups of OPC test patterns with different pattern densities, an optical proximity correction model is established in which the critical dimension of the test pattern changes with the pattern density. The wafer data of each group of OPC test patterns with different pattern densities are collected respectively. The wafer data includes the actual critical dimension of the main pattern Pi fabricated on the wafer, and the wafer data for measuring the deviation value between the CD values of the OPC test pattern and the actual process output under different pattern densities. A mapping relationship between the actual critical dimension of the main pattern Pi on the wafer and the pattern density of the OPC test pattern is established, and a deviation value query table SSA table of δ(x, y) under different pattern densities is established, that is, the main process of the modeling method for establishing the OPC correction model. The deviation value is the difference between the actual CD value of the wafer after the actual processing technology and the design value or theoretical value corresponding to the pattern.

[0071] The main steps of the described modeling method are as follows:

[0072] Step 1. Set N groups of OPC test patterns P1 to PN with different pattern densities η, such that P1 to PN have different pattern densities η1 to ηN, where N is a natural number greater than or equal to 2. The preset area PX contains several test patterns M(x, y) related to the position coordinates x and y.

[0073] x and y are the horizontal and vertical coordinate ranges of the preset area, and the values of x and y can each be any value within the range of 30 to 1000 nm.

[0074] Step 2. Use the test patterns from Step 1 for exposure and development, and measure the post-lithography positions / CDs M'(x, y) of M(x, y) in P1 to PN corresponding to different η.

[0075] Step 3. Respectively calculate the differences ηδ(x, y) between M(x, y) and M'(x, y) corresponding to different Pi. i (x, y);

[0076] Step 4. Plot the relationship curve between ηδ(x, y) and η, and find the η within the range that meets the OPC modeling accuracy requirements for ηδ(x, y). i (x, y) and η; i (x, y) that meets the OPC modeling accuracy requirements.

[0077] Step 5. Based on Step 4, taking ηδ(x, y) from Step 4 as a reference, calculate the differences in ηδ(x, y) for any pattern density η. Since the pattern densities η (η1 to ηN) of the OPC test patterns are artificially set, and the pattern density η of the actual layout pattern may be any value within a certain range, it is necessary to obtain the correction differences ηδ(x, y) corresponding to any pattern density η through the relationship curve plotted with the pattern densities η of the limited OPC test patterns. i (x, y) as a reference, calculate the difference in ηδ(x, y) for any pattern density η. Since the pattern density η (η1 to ηN) of the OPC test pattern is artificially set, and the pattern density η of the actual layout pattern i may be any value within a certain range, so it is necessary to obtain the correction difference ηδ(x, y) corresponding to any pattern density η i through the relationship curve plotted with the pattern density η of the limited OPC test pattern; i may be any value within a certain range, so it is necessary to obtain the corresponding correction difference ηδ(x, y) for any pattern density η i through the relationship curve plotted with the pattern density η of the limited OPC test pattern; i (x, y);

[0078] Step 6. Based on Step 4, taking ηδ(x, y) from Step 4 as a reference, establish the OPC correction model OPC model A and the target value table SSA table. i (x, y) as a reference, establish the OPC correction model OPC model A and the target value table SSA table;

[0079] Step 7. Based on Step 4, select the group of CD values with relatively stable pattern density intervals and establish the standard target value table SSA table S for OPC for different line widths and space periods.

[0080] Among them, the graphic density interval is relatively stable. The standard value of the CD value is the average of one group or several groups in the stable graphic density interval, not limited to only taking one group. Its definition includes the following several types, which can be selected and determined according to the actual situation:

[0081] a) Within this graphic density interval, the CD values and deviation values of different graphic densities meet the accuracy requirements of OPC modeling;

[0082] b) Within this graphic density interval, any graphic density can be selected as a representative for modeling;

[0083] c) Within this graphic density interval, the CD values corresponding to any number of graphic density intervals can be averaged, and the deviation value of the averaged CD value can be calculated. Then, the averaged CD value and the deviation value corresponding to the averaged CD value are used for modeling;

[0084] Step 8. Use ηδ i (x,y) and the SSA table S to establish the SSA table (ηi) at any graphic density ηi.

[0085] Figure 5 The figure shows a schematic diagram of a simple OPC correction method. In the figure, the SSA table only shows the schematic correction of the deviation value and does not involve specific correction values. Each intersection point of the actual SSA table is a specific correction amount including positive and negative signs.

[0086] Of course, as described in the background art section, the actually established OPC correction model also includes a large number of other OPC correction parameters, including optical system parameters, etc. This is not the focus of the technical solution of the present invention and will not be elaborated here.

[0087] Specifically, the modeling method of the OPC correction model includes:

[0088] Step 1. Design N groups of OPC test patterns. Each group correspondingly includes the same main pattern Pi (P1 to PN). Then, dummy patterns Ai (i.e., auxiliary patterns Ai1 to AiN) are placed around each main pattern Pi from P1 to PN, so that the N groups of OPC test patterns have different graphic densities η1 to ηN, where N is a natural number greater than or equal to 2.

[0089] When modeling, the N groups of OPC test patterns with different graphic densities η, such as Figure 2 and Figure 3As shown, the OPC test pattern on a preset area includes at least one main pattern Pi and at least one auxiliary pattern Ai. Different groups of main patterns Pi and their corresponding at least one auxiliary pattern Ai have different design distances, and / or the ratios of the total areas of different groups of at least one auxiliary pattern Ai to the total areas of their corresponding regions are different, so as to characterize different pattern densities η. For a chip layout, the pattern density η can refer to the overall pattern density or the local pattern density. The local pattern density is the percentage of the sum of the areas of multiple auxiliary patterns Ai included in a preset area in the total area of the preset area.

[0090] In any preset area PX, there are several test patterns M(x, y) related to the position coordinates (x, y), and the position coordinates (x, y) are all within the design area PX.

[0091] The pattern density η of the OPC test pattern needs to be determined by scanning and calculating before establishing the OPC model. For a preset area PX with a coordinate range of (X, Y) (i.e., the size range can be X*Y, the value range of X can be 30 - 1000 nm, and the value range of Y can be 30 - 1000 nm), the pattern density η is the percentage of the total area of at least one auxiliary pattern Ai included in the preset area PX in the total area of the preset area PX.

[0092] Step 2: Using the OPC test pattern, test the actual CD values of the main pattern Pi on the actual semiconductor silicon wafer at different pattern densities η.

[0093] That is, use the lithography process of exposure and development with the N groups of OPC test patterns in Step 1 to measure the actual pattern M'(x, y) of the position / CD after lithography of the test pattern M(x, y) corresponding to different pattern densities η in P1 - PN;

[0094] Step 3: Respectively calculate the topographic difference ηδ i (x, y) between the test pattern M(x, y) and the actual pattern M'(x, y) corresponding to the main pattern Pi at different positions

[0095] Step 4: Plot the relationship curve between ηδ i (x, y) and the pattern density η, and find the pattern density η within the range that meets the OPC modeling accuracy δ requirement in ηδ i (x, y) iThe modeling accuracy δ is generally between 1% and 5% of the minimum CD value of the graphic layer, and is determined according to the product design redundancy. Drawing the relationship curve requires multiple repeated tests on different graphic densities η to obtain CD value data in a relatively stable and reliable graphic density range, such as Figure 4 η4 shown in. The OPC modeling accuracy is determined according to requirements. For example, the gate CD of the 130nm node is 130nm, and the accuracy requirement is δ < 5nm; for the 55nm node, the gate CD is 65nm, and the accuracy requirement is δ < 3nm.

[0096] Then, taking ηδ i (x,y) in step 4 as the reference, determine ηδ i (x,y) under different graphic densities η, that is, the deviation value ηδ i (x,y) between the OPC theoretical value or design value containing graphic density information and the CD value formed on the actual wafer;

[0097] Similarly, taking ηδ i (x,y) in step 4 as the reference, that is, establishing the OPC correction model OPC model A and the target value table SSA table;

[0098] Based on step 4 again, for the set of CD values under relatively stable graphic densities, take the set of CDs in the relatively stable graphic density range η (as mentioned above, it can also be the average of several groups) as the standard value, such as Figure 4 η4 in.

[0099] Establish the standard target value table SSA table S for different line widths CD and space periods of OPC;

[0100] Figure 4 simply shows four graphic density ranges η1 - η4. Among them, the curves of η1 - η3 are CD values with unstable deviation values under different graphic densities, and different SSA tables can be formed; while η4 has relatively stable CD values, the curve is relatively straight, and for different substrates (the dotted line and the solid line correspond to substrates with different reflection coefficients), the curve coincidence degree is relatively high, and the standard target value table SSA table S can be established.

[0101] Using ηδ i (x,y) and the target value table SSA table S to establish the deviation value correction table SSA table(ηi) containing any graphic density ηi.

[0102] There is a mapping relationship between the critical dimension CD of the pattern in the OPC correction model and the pattern density-related variables, including: the functional relationship between the actual critical dimension of the main pattern Pi on the silicon wafer and the designed area of the auxiliary pattern Ai and the designed distance between the main pattern Pi and the auxiliary pattern Ai; and / or, the functional relationship between the actual critical dimension of the main pattern Pi on the silicon wafer and the ratio of the designed area of the auxiliary pattern Ai to the area of the preset region within a preset region.

[0103] The OPC correction method proposed by the present invention is to use the above-mentioned OPC model A and SSA tableS to perform correction to obtain the pattern M”(x,y), and then call the deviation value correction table SSAtable(ηi) corresponding to the pattern density ηi of the pattern to be corrected to adjust the pattern M”(x,y) to the target value M’(x,y). In the correction method of the present invention, the pattern density is not directly used as the third dimension during modeling / correction, and the two-dimensional plane x,y dimensions in the traditional correction method are still retained. The pattern density parameter forms a deviation value correction table SSA table(ηi) and then corrects the correction amount after the existing OPC model is corrected, which can greatly reduce the computational amount of model establishment and ensure relatively accurate correction at the same time.

[0104] Before using the model OPC model A and the target value table SSA table S for correction, it is necessary to calculate the pattern density ηi of the pattern to be corrected in advance. The calculation method is as described above. Scan a certain range X, Y of the target pattern area M(x,y) to be corrected, and measure the pattern density ηi of the X*Y area.

[0105] Use the previously established model OPC model A and SSA table S to perform correction to obtain the pattern M”(x,y);

[0106] Call the SSA table(ηi) corresponding to ηi to perform deviation value adjustment on the pattern M”(x,y) again to the target value M’(x,y);

[0107] ηδ i (x,y) = δ i ′(x,y) - δ i (x,y)

[0108] Then, use the lithography layout of the corrected OPC pattern for the lithography process to obtain the actual wafer pattern that is more consistent with the expected parameter target.

[0109] The SSA table described in the present invention is a matrix that contains the CD deviation value correction amounts at different pattern densities instead of the actual CD sizes. The independent variables in the horizontal and vertical directions are the line widths and the space periods (or line spacings), and the matrix element at the intersection position is the target value or the target movement distance of this combination. In different application scenarios, there may not be only one SSA table, but rather multiple SSA tables can be composed. Each SSA table can be set to be applicable to a certain pattern density or environment. Figure 5 A schematic diagram of the correction using the SSA table is shown in Figure 5 . However, according to the different pattern densities calculated by scanning, different SSA tables including SSA table S need to be called for correction, not limited to only one group among them. The present invention can call different SSA tables according to different pattern densities for different corrections, which can significantly improve the accuracy of the correction.

[0110] The above are only the preferred embodiments of the present invention and are not used to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A modeling method for an OPC correction model, characterized in that: It includes the following steps: a). Establish multiple groups of OPC test patterns with different graphic densities, and use the OPC test patterns to perform process processing to obtain modeled wafer test chips; b). Measure the CD values of different graphic densities and the differences from the design values / theoretical values based on the wafer test chips; c). Determine the stable graphic density range, determine the standard CD and standard difference, and establish the OPC test model OPCmodel A and the standard deviation value table SSA Table S based on these; d). Establish the target value table SSA table corresponding to different graphic densities according to steps b and c.

2. The modeling method of the OPC correction model according to claim 1, characterized in that: The specific modeling method includes: Step 1, design N groups of OPC test patterns with different graphic densities η, where N is a natural number greater than or equal to 2; each group of OPC test patterns with different graphic densities η includes at least one main pattern Pi and at least one Dummy pattern Ai arranged around the main pattern Pi and having a different design distance from it, so that the OPC test patterns P1 to PN have different graphic densities η1 to ηN; a preset graphic area PX contains several test patterns M(x, y) related to the position coordinates (x, y); Step 2, use the OPC test patterns in Step 1 for exposure and development of the lithography process, and measure the post-lithography photoresist patterns M’(x, y) corresponding to M(x, y) in the OPC test patterns P1 to PN in the preset graphic area PX at different graphic densities η; Step 3: Calculate the position deviation values ηδ of M(x, y) and M’(x, y) corresponding to the OPC test patterns P1 to PN at different pattern densities η, where i ∈ (1 to N); i (x, y), i ∈ (1 to N); Step 4, draw the position deviation value ηδ i The relationship curve between (x, y) and the pattern density η, and find ηδ i The pattern density η where (x, y) meets the OPC modeling accuracy requirement range of δ; Step 5, taking ηδ obtained in the said Step 4 i (x, y) as a reference, find the deviation value of ηδ of different graphic densities η i (x, y); Step 6, based on Step 4, using ηδ of Step 4 i (x, y) as a reference, establish an OPC model OPC model A and an SSA table; And based on Step 4, take the group of CD values in the relatively stable graphic density η range as the standard value, and establish an OPC model for different line widths and space periods, including the standard target value table SSA table S; Step 7, using ηδ i (x, y) and the SSA table S to establish a deviation value correction table SSA table(ηi) at different graphic densities η.

3. The modeling method of the OPC correction model according to claim 2, characterized in that: In Step 1, the N groups of OPC test patterns are sets of OPC test patterns with different graphic densities, where N≥2; and each group of OPC test patterns with different graphic densities includes at least one same main pattern Pi and at least one dummy pattern Ai arranged around the main pattern Pi and having a different design distance or different total area from it; use the OPC test patterns to establish an optical proximity correction model in which the critical dimensions of the chip layout patterns change with the graphic density.

4. The modeling method of the OPC correction model according to claim 2, characterized in that: The described pattern density η refers to the local pattern density and / or the global pattern density; the local pattern density refers to the ratio of the sum of the design areas of all the auxiliary patterns Ai within a randomly defined preset area in an OPC pattern to the area of the preset area: the arbitrary pattern density η i refers to the arbitrary pattern density not included in the actual layout specifically determined by the N sets of OPC test pattern densities.

5. The modeling method of the OPC correction model according to claim 2, characterized in that: The dummy pattern Ai has the same or different dimensions, areas or morphologies; the dummy pattern Ai in different groups of the OPC test patterns P1 to PN has a different distance from the OPC test patterns P1 to PN, and / or, the dummy pattern Ai in different groups of the OPC test patterns P1 to PN has a different total area, so as to adjust to form different graphic densities η1 to ηN.

6. The modeling method of the OPC correction model according to claim 2, characterized in that: The modeling accuracy δ is determined according to the product design redundancy and is designed to be 1% - 5% of the minimum CD value of this graphic layer.

7. The modeling method of the OPC correction model according to claim 2, characterized in that: The SSA table or SSAtable(ηi) mentioned above is a matrix. The independent variables in the horizontal and vertical directions are the line widths and spatial periods or line spacings in the layout pattern, and the matrix element at the intersection position is the target value or target movement distance of this combination.

8. The modeling method of the OPC correction model according to claim 2, characterized in that: Take the set of CD values in the relatively stable graphic density η interval in step 6 as the standard value. The standard value of the CD value is a set, or the average of several sets, in the stable graphic density interval. The definition of the stable graphic density η interval includes the following situations: a) Within this graphic density η interval, the CD values and deviation values of different graphic densities meet the accuracy requirements of OPC modeling. b) Any graphic density can be selected as a representative for modeling within this graphic density η interval. c) Within this graphic density η interval, the CD values corresponding to any number of graphic density intervals can be averaged, and the deviation value of the averaged CD value can be calculated. Then, the averaged CD value and the deviation value corresponding to the averaged CD value are used for modeling.

9. The modeling method of the OPC correction model according to any one of claims 1 to 8, characterized in that: The SSAtable mentioned above is composed of one or no less than two SSA tables; each SSA table can set its applicable graphic environment, including corresponding to different graphic densities.

10. The modeling method of the OPC correction model according to any one of claims 1 to 9, characterized in that: The SSAtable does not contain specific CD values, but positive and negative correction amounts under different graphic densities.

11. An OPC correction method, characterized in that: The correction method mentioned above includes: Using the OPC correction model established by the OPC modeling method according to any one of claims 1 - 8. The OPC correction model is a target value table SSA table and a standard target value table SSA table S that contain the correction function relationship between the actual graphic CD value and its corresponding graphic density η variable. Using the target value table SSA table and the standard target value table SSA table S to perform OPC correction on the OPC graphics of the pre - fabricated layout based on the correction information of the graphic density ηi to be corrected to obtain the target layout, and then performing the normal mass - production lithography process on the target layout to obtain the corrected ideal wafer pattern.

12. The OPC correction method according to claim 11, characterized in that: Before using the OPC correction model to perform OPC correction on the graphic to be corrected, first perform a scanning calculation step on the graphic to be corrected to determine the value of the graphic density ηi to be corrected, including: Scanning the graphic M(x, y) to be corrected within the coordinate ranges X and Y, and measuring the graphic density ηi of the X*Y area range of the graphic to be corrected.

13. The OPC correction method according to claim 11, characterized in that: According to the measured graphic density ηi of the graphic to be corrected, call the OPC correction model, including OPCmodel A and the target value table SSA table or the standard target value table SSA table S, to correct the target graphic M(x, y) to obtain the graphic M”(x, y). Call the deviation value correction table SSA table(ηi) corresponding to the pattern density ηi to correct the pattern M”(x,y), and adjust it to the target value M’(x,y), that is, the OPC correction is completed.

14. The OPC correction method according to claim 11, wherein: The pattern density ηi described above is not used as the third dimension during modeling / correction. The traditional two-dimensional plane x,y dimensions are still retained. The pattern density is a correction amount after the correction of the existing OPC technology is completed.

15. The OPC correction method according to claim 11, wherein: The SSA table described above is a matrix. The independent variables in the horizontal and vertical directions are the line width and the space period or the line pitch. The matrix element at the intersection position is the target value or the target movement distance of this combination.

16. The OPC correction method according to claim 11, wherein: The SSA table described above consists of one or not less than two; when there are not less than two, each SSA table can be applicable to different pattern environments.

17. An OPC correction system, characterized in that: The correction system described above includes an OPC model establishment module, which establishes a functional relationship between the pattern density η and the device CD value actually formed based on the pattern density η of the pattern to be corrected, and establishes an OPC correction model based on the relationship between the functional relationship and the variables of the pattern density η, including OPC model A, the target value table SSA table, and the standard target value table SSA table S; The OPC correction module uses the OPC correction model to perform a pattern density scan calculation on the pattern to be corrected to obtain the pattern density ηi of the pattern to be corrected, compares the pattern density ηi with the OPC correction model and the deviation value correction table to find the correction amount of the pattern to be corrected, and then performs OPC correction on the pattern to be corrected.

Images