Modeling method and system of optical proximity correction model and computer equipment

Abstract

The invention discloses a modeling method and system of an optical proximity correction model and computer equipment, and belongs to the field of integrated circuit manufacturing, and the modeling method of the optical proximity correction model comprises the following steps: setting a cost function comprising a standard deviation of an asymmetric graph; under each parameter, inputting the target graph data into the optical proximity correction model, obtaining simulation graph data of the target graph, and obtaining a cost function under each parameter according to the simulation graph data and the wafer graph data; outputting an optical proximity correction model when the cost function is minimum in the parameter range; and when the minimum cost function is smaller than a preset cost, taking the optical proximity correction model corresponding to the minimum cost function as a target optical proximity correction model. According to the modeling method of the optical proximity correction model disclosed by the invention, the accuracy of optical proximity correction of the asymmetric photomask graph can be improved.

Description

Technical Field

[0001] The present invention belongs to the field of integrated circuit manufacturing, and particularly relates to a method, system and computer device for modeling an optical proximity correction model. Background Art

[0002] In semiconductor manufacturing, optical proximity correction (OPC) is a key step in the lithography process. Optical proximity correction compensates for image distortion caused by optical effects during the lithography process by fine-tuning the mask pattern.

[0003] Optical proximity correction models typically focus on symmetrically designed mask patterns. Since symmetric mask patterns allow for the use of simplified mathematical models and algorithms to estimate and correct optical distortion, the optical proximity correction model works well when dealing with symmetric mask patterns. However, many important mask patterns are asymmetric mask patterns, such as the mask patterns of static random access memory (SRAM). Due to their unique layout and optical properties, these asymmetric mask patterns make it impossible for traditional optical proximity correction models to effectively predict and correct their optical distortion. Only by manually adjusting the process steps and parameters by engineers can the asymmetric mask patterns be adjusted, and the whole process is time-consuming and requires a large amount of manpower. Summary of the Invention

[0004] The purpose of the present invention is to provide a method, system and computer device for modeling an optical proximity correction model, which can solve the problem that asymmetric mask patterns cannot be optically proximity corrected.

[0005] To solve the above technical problems, the present invention is realized through the following technical solutions: The present invention provides a method for modeling an optical proximity correction model, including the following steps: Obtain a plurality of target graphic data and the corresponding wafer graphic data for each of the target graphic data; Classify the target graphics into symmetric graphics and asymmetric graphics; Set the initial kernel and parameter range of the optical proximity correction model, and set a cost function representing the fitting degree, where the cost function includes the standard deviation of all target graphics and the standard deviation of the asymmetric graphics; Under each parameter within the parameter range, input the target graphic data into the optical proximity correction model, obtain the simulated graphic data of the target graphics, and obtain the cost function under each parameter based on the simulated graphic data and the wafer graphic data; Output the optical proximity correction model when the cost function within the parameter range is minimized; and Determine whether the minimum cost function is less than a preset cost; if the minimum cost function is less than the preset cost, use the output optical proximity correction model as the target optical proximity correction model; if the minimum cost function is greater than or equal to the preset cost, adjust the kernel and parameter range of the optical proximity correction model until the minimum cost function is less than the preset cost.

[0006] In an embodiment of the present invention, obtaining the cost function for each parameter includes the following steps: Select a parameter within the parameter range; Input all target graphic data into the optical proximity correction model to obtain the simulated data graphics corresponding to each target graphic data; Obtain the cost function for this parameter based on the simulated graphic data corresponding to each target graphic data and the wafer graphic data; and Determine whether the parameters within the parameter range have been traversed. If the parameters within the parameter range have been traversed, end the process. If the parameters within the parameter range have not been traversed, select another parameter within the parameter range and repeat the process of obtaining the cost function.

[0007] In an embodiment of the present invention, the simulated graphic data is the position data of the two intersection points of the optical signal curve of the wafer graphic formed by simulating the target graphic on the wafer using the optical proximity correction model and the exposure threshold.

[0008] In an embodiment of the present invention, the cost function is obtained through the following formula: Cost function = A × (RMS - all / RMS - all - spec) + B × (RMS - part / RMS - part - spec); Where, Cost function is the cost function, RMS - all is the standard deviation of all input target graphics, RMS - all - spec is the specified value of the standard deviation of all input target graphics, A is the weight coefficient of all input target graphics, RMS - part is the standard deviation of all asymmetric graphics, RMS - part - spec is the specified value of the standard deviation of all asymmetric graphics, and B is the weight coefficient of all asymmetric graphics.

[0009] In an embodiment of the present invention, RMS - all is the root mean square of the fitting error of all target graphics to characterize the error between the simulated graphics and the wafer graphics of all target graphics, and RMS - part is the root mean square of the fitting error of all asymmetric graphics to characterize the error between the simulated graphics and the wafer graphics of all asymmetric graphics.

[0010] In an embodiment of the present invention, the root mean square (RMS) of all input target patterns, RMS-all, is obtained by the following formula: RMS-all = SQRT(SUMSQ E(1, N) / COUNTA E(1, N)); The root mean square (RMS) of all asymmetric patterns, RMS-part, is obtained by the following formula: RMS-part = SQRT(SUMSQ E(b + 1, N) / COUNTA E(b + 1, N)); Wherein, SQRT represents the square root function, SUMSQ represents the sum of squares function, COUNTA represents the function of counting non-empty cells within a given range, E(1, N) is the set of errors between the simulated pattern data corresponding to all the target patterns and the wafer pattern data, and E(b + 1, N) is the set of errors between the simulated pattern data corresponding to the asymmetric patterns and the wafer pattern data.

[0011] In an embodiment of the present invention, the value of the weight coefficient A for all input target patterns is 0.8, and the value of the weight coefficient B for all the asymmetric patterns is 0.2.

[0012] In an embodiment of the present invention, classifying the target patterns into symmetric patterns and asymmetric patterns includes the following steps: Inputting the target pattern data and the wafer pattern data into the optical proximity correction model; Using the optical proximity correction model to simulate the optical signal curve of the wafer pattern formed by the target pattern on the wafer; Determining the exposure threshold based on the fact that the distance between the two intersection points of the optical signal curve and the exposure threshold is equal to the wafer pattern data; Obtaining the distances from the two intersection points of the optical signal curve and the exposure threshold to the target pattern data to determine the two intersection points of the optical signal curve and the exposure threshold; and Judging whether the distances from the two intersection points of the optical signal curve and the exposure threshold to the target pattern data are equal. If the distances from the two intersection points of the optical signal curve and the exposure threshold to the target pattern data are equal, then the target pattern is determined to be the symmetric pattern. If the distances from the two intersection points of the optical signal curve and the exposure threshold to the target pattern data are not equal, then the target pattern is determined to be the asymmetric pattern.

[0013] The present invention also provides a modeling system for an optical proximity correction model, including: A data acquisition module, which acquires a plurality of target pattern data and the wafer pattern data corresponding to each of the target pattern data; A graphic discrimination module that discriminates target graphics into symmetric graphics and asymmetric graphics; A kernel and parameter range adjustment module that sets the initial kernel and parameter range of the optical proximity correction model; A cost function setting module that sets a cost function representing the fitting degree, where the cost function includes the standard deviation of all target graphics and the standard deviation of the asymmetric graphics; An optical proximity correction and cost function output module that inputs the target graphic data into the optical proximity correction model at each parameter within the parameter range, obtains the simulated graphic data of the target graphics, and obtains the cost function at each parameter based on the simulated graphic data and the wafer graphic data; An optical proximity correction model output module that outputs the optical proximity correction model when the cost function within the parameter range is minimized; and A judgment module that judges whether the minimum cost function is less than a preset cost; and A target optical proximity correction model output module that, when the minimum cost function is less than the preset cost, uses the output optical proximity correction model as the target optical proximity correction model; Wherein, if the minimum cost function is greater than or equal to the preset cost, the kernel and parameter range adjustment module adjusts the kernel and parameter range of the optical proximity correction model until the minimum cost function is less than the preset cost.

[0014] The present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. The feature is that when the processor executes the computer program, it implements the steps of the modeling method of the optical proximity correction model as described in any one of the above.

[0015] In summary, the present invention provides a modeling method, system, and computer device for an optical proximity correction model. The unexpected effect is that through the modeling of the optical proximity correction model, the processing ability of the optical proximity correction model for asymmetric design graphics is significantly improved, especially the correction accuracy when processing asymmetric graphics of key components such as static random access memories. The modeled optical proximity correction model can more accurately predict and correct the optical distortion of asymmetric graphics, thereby directly improving the result of optical proximity correction and the yield of products. By enabling the modeled optical proximity correction model to adapt to asymmetric graphics at the initial stage, the cumbersome adjustment and compensation work that engineers need to perform in the subsequent process is greatly reduced, thereby accelerating the R & D progress and improving the production efficiency. In addition, the application of this technical solution reduces the dependence on subsequent manual adjustment process steps and parameter compensation, reduces the risk of compensation failure, and further ensures the accuracy of electrical performance and the reliability of products.

[0016] Of course, it is not necessary for any product implementing the present invention to achieve all of the above advantages simultaneously. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention, and those of ordinary skill in the art can obtain other drawings based on these drawings without creative efforts.

[0018] Figure 1 It is a flowchart of a method for modeling an optical proximity correction model in an embodiment.

[0019] Figure 2 It is a flowchart of distinguishing a target pattern into a symmetric pattern and an asymmetric pattern in an embodiment.

[0020] Figure 3 It is a schematic diagram of the principle of distinguishing a target pattern into a symmetric pattern and an asymmetric pattern in an embodiment.

[0021] Figure 4 It is a flowchart of obtaining a cost function under each parameter in an embodiment.

[0022] Figure 5 It is a schematic diagram of symmetric pattern correction principle in an embodiment.

[0023] Figure 6 It is a schematic diagram of asymmetric pattern correction principle in an embodiment.

[0024] Figure 7 It is a structural block diagram of a modeling system for an optical proximity correction model in an embodiment.

[0025] Figure 8 It is a structural block diagram of a computer device in an embodiment.

[0026] Reference Signs Description: 201, data acquisition module; 202, pattern distinction module; 203, kernel and parameter range adjustment module; 204, cost function setting module; 205, optical proximity correction and cost function output module; 206, optical proximity correction model output module; 207, judgment module; 208, target optical proximity correction model output module; 210, optical signal curve; 211, exposure threshold. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0027] The following describes the implementation modes of the present invention through specific examples. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific implementation modes. Various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention.

[0028] It should be noted that the diagrams provided in this embodiment only illustrate the basic concept of the present invention in a schematic manner. Therefore, only the components related to the present invention are shown in the diagrams, rather than being drawn according to the number, shape, and size of the components in actual implementation. The type, quantity, and proportion of each component in actual implementation can be arbitrarily changed, and the layout type of its components may also be more complex.

[0029] In the present invention, it should be noted that when terms such as "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc. appear, the orientation or positional relationship indicated is based on the orientation or positional relationship shown in the drawings. This is only for the convenience of describing this application and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it should not be understood as a limitation to this application. In addition, when terms such as "first" and "second" appear, they are only used for descriptive and differentiating purposes and cannot be understood as indicating or implying relative importance.

[0030] In the process of semiconductor integrated circuit manufacturing, through lithography, the mask pattern on the mask is transferred to the wafer to form a wafer pattern. Before lithography, an optical proximity correction model is used to correct the mask pattern so that the wafer pattern formed on the wafer by the mask pattern highly fits the designed wafer pattern. Specifically, the optical proximity correction model first needs to identify the edges of the mask pattern, allow each edge to move freely, and then correct the mask pattern to form a target pattern. After that, the optical model and the resist chemical reaction model simulate the wafer pattern formed on the wafer by the target pattern, that is, the simulated pattern. Then, the simulated pattern is compared with the designed wafer pattern. When the error between the simulated pattern and the designed wafer pattern is small, it means that the simulated pattern is close to the designed wafer pattern. The correction software moves the edge position of the mask pattern during operation, obtains the error between the simulated pattern and the designed wafer pattern in real time, and repeats this process continuously until the error reaches an acceptable value.

[0031] An initial optical proximity correction model is established in the lithography machine. The kernel and parameter range in the initial optical proximity correction model can be set by empirical values. The process of modeling the optical proximity correction model is the process of determining the kernel and parameter range in the optical proximity correction model. After inputting the target graphic data and the designed wafer graphic data into the optical proximity correction model, the optical proximity correction model iteratively calculates repeatedly to obtain the optical signal curve of the simulated graphic and the exposure threshold of the lithography machine. When the distance between the two intersection points of the exposure threshold and the optical signal curve is exactly equal to the wafer graphic data, it indicates that the simulated graphic output by the optical proximity correction model at this time completely coincides with the designed wafer graphic, that is, the target optical proximity correction model is obtained, and the modeling of the optical proximity correction model is completed.

[0032] Please refer to Figure 1 As shown, a method, system, and computer device for modeling an optical proximity correction model provided by the present application can improve the fitting degree between the simulated graphic of the asymmetric graphic and the wafer graphic, and improve the accuracy of the proximity correction model for correcting the asymmetric graphic. Specifically, a method for obtaining an optical proximity correction model provided by the present invention includes steps S10 to S80.

[0033] Step S10: Obtain a plurality of target graphic data and the corresponding wafer graphic data for each target graphic data.

[0034] Step S20: Classify the target graphics into symmetric graphics and asymmetric graphics.

[0035] Step S30: Set the initial kernel and parameter range of the optical proximity correction model.

[0036] Step S40: Set a cost function representing the fitting degree.

[0037] Step S50: Under each parameter within the parameter range, input the target graphic data into the optical proximity correction model to obtain the corresponding simulated graphic data of the target graphic data, and obtain the cost function under each parameter based on the simulated graphic data and the wafer graphic data.

[0038] Step S60: Output the optical proximity correction model when the cost function within the parameter range is the smallest.

[0039] Step S70: Determine whether the minimum cost function is less than a preset cost. If the minimum cost function is less than the preset cost, then execute Step S80: Use the optical proximity correction model corresponding to the minimum cost function as the target optical proximity correction model; if the minimum cost function is greater than or equal to the preset cost, then return to Step S30 and reset the kernel and parameter range in Step S30 until the minimum cost function is less than the preset cost.

[0040] Please refer to Figure 1As shown, in an embodiment of the present invention, in step S10, the target pattern is the corrected reticle pattern, and the target pattern data is the critical dimension (mask CD) of the corrected reticle pattern. The wafer pattern data corresponding to the target pattern data is the designed wafer pattern data, that is, the critical dimension (wafer CD) of the designed wafer pattern.

[0041] Please refer to Figures 1 to 3 As shown, in an embodiment of the present invention, in step S20, the target pattern is classified into a symmetric pattern and an asymmetric pattern, and it can be determined whether the target pattern is a symmetric pattern or an asymmetric pattern based on the target pattern data and the corresponding wafer pattern data. Specifically, it includes steps S201 to S207.

[0042] Step S201: Input the target pattern data and the wafer pattern data into the optical proximity correction model.

[0043] Specifically, input the target pattern data and the wafer pattern data into the optical model and the photoresist chemical reaction model of the optical proximity correction model. Among them, the target pattern data is, for example, Figure 3 the Mask CD shown in part (a) of Figure 3 and the wafer pattern data is, for example, the wafer CD shown in part (a) of

[0044] Step S202: Use the optical proximity correction model to simulate the optical signal curve of the wafer pattern formed by the target pattern on the wafer.

[0045] Specifically, use the optical proximity correction model to simulate the optical signal curve 210 of the wafer pattern formed by the target pattern on the wafer, for example Figure 3 as shown in part (b) of

[0046] Step S203: Determine the exposure threshold based on the fact that the distance between the two intersection points of the optical signal curve and the exposure threshold is equal to the wafer pattern data.

[0047] Specifically, as Figure 3 shown in part (c) of, adjust the size of the exposure threshold 211 so that the distance between the two intersection points of the optical signal curve 210 and the exposure threshold 211 is equal to the size of the wafer pattern data, that is, obtain the exposure threshold 211. It should be noted that at this time, the obtained optical signal curve 210 and the exposure threshold 211 do not necessarily meet the requirements of the subsequent cost function, and here it is only used as a judgment basis for whether the target pattern is a symmetric pattern.

[0048] Step S204: Obtain the distances from the two intersection points of the optical signal curve and the exposure threshold to the target pattern data, and determine the two intersection points where the optical signal curve and the exposure threshold intersect.

[0049] Specifically, asFigure 3 part (d) in Figure 3 As shown in part (e) in

[0050] Step S205: Determine whether the distances from the two intersection points where the optical signal curve intersects the exposure threshold to the target graphic data are equal. If the distances from the two intersection points where the optical signal curve intersects the exposure threshold to the target graphic data are equal, then execute Step S206: Determine that the target graphic is a symmetric graphic; if the distances from the two intersection points where the optical signal curve intersects the exposure threshold to the target graphic data are not equal, then execute Step S207: Determine that the target graphic is an asymmetric graphic.

[0051] In this embodiment, as Figure 3 part (d) in Figure 3 As shown in part (e) in

[0052] Please refer to Figures 1 to 2 As shown, in an embodiment of the present invention, after all target graphics are classified into symmetric graphics and asymmetric graphics, the symmetric graphics and asymmetric graphics are placed into different sets, namely the symmetric graphic set and the asymmetric graphic set. In this embodiment, all target graphics are represented by the set (1, N), the symmetric graphic set is represented by the set (1, b), and the asymmetric graphic set is represented by the set (b + 1, N). Among them, N is the number of all target graphics, b is the number of symmetric graphics, and N - b is the number of asymmetric graphics.

[0053] Please refer to Figure 1As shown, in an embodiment of the present invention, in step S30, the kernel refers to the basic function or algorithm for describing the pattern transformation process of transferring the mask pattern to the wafer surface, and these transformation processes include effects such as light diffraction, reflection, and interference. The parameter range refers to the parameter setting range of these kernel functions when simulating the actual lithography process, and these parameters will affect the accuracy of optical proximity correction and the final correction effect. The selection of the kernel and its parameters is directly related to the effectiveness and efficiency of the optical proximity correction model modeling process. In this application, when setting the initial kernel and parameter range of the optical proximity correction model, any kernel and the corresponding parameter range can be selected from the modeling program library according to experience.

[0054] Please refer to Figure 1 As shown, in an embodiment of the present invention, the cost function in step S40 is obtained through the following formula: Cost function = A × (RMS-all / RMS-all-spec) + B × (RMS-part / RMS-part-spec); Wherein, Cost function is the cost function, RMS-all is the standard deviation of all input target patterns, RMS-all-spec is the specified value of the standard deviation of all input target patterns, A is the weight coefficient of all input target patterns, RMS-part is the standard deviation of all asymmetric patterns, RMS-part-spec is the specified value of the standard deviation of all asymmetric patterns, and B is the weight coefficient of all asymmetric patterns.

[0055] Please refer to Figure 1 As shown, in an embodiment of the present invention, RMS-all is the root mean square of the fitting error of all target patterns, which is used to characterize the error between the simulated pattern of all target patterns and the wafer pattern, and RMS-part is the root mean square of the fitting error of all asymmetric patterns, which is used to characterize the error between the simulated pattern of all asymmetric patterns and the wafer pattern. The fitting error is the data difference between the simulated pattern output by the optical proximity correction model and the designed wafer pattern.

[0056] Specifically, the standard deviation RMS-all of all input target patterns is obtained through the following formula: RMS-all = SQRT(SUMSQ E(1, N) / COUNTA E(1, N)); The standard deviation RMS-part of all asymmetric patterns is obtained through the following formula: RMS-part = SQRT(SUMSQ E(b + 1, N) / COUNTA E(b + 1, N)); Among them, SQRT represents the square root function, SUMSQ represents the sum of squares function, COUNTA represents the function of counting non-empty cells within a given range, E(1, N) is the set of errors between the simulated pattern data and the wafer pattern data corresponding to all target patterns, and E(b + 1, N) is the set of errors between the simulated pattern data and the wafer pattern data corresponding to asymmetric patterns.

[0057] Please refer to Figure 1 As shown, in an embodiment of the present invention, RMS-all-spec and RMS-part-spec are specific specification values of the standard deviation set to meet the lithography quality standard. The specific values of RMS-all-spec and RMS-part-spec are set based on process requirements and the characteristics of the process node, which can ensure the accuracy and consistency of the simulated pattern and the target pattern. Among them, the value range of RMS-all-spec is, for example, 1.6 to 2.0, and specifically, it can be 1.6, 1.8, or 2.0. The value range of RMS-part-spec is, for example, between 1.4 and 1.6, and specifically, it can be 1.4, 1.5, or 1.6.

[0058] Please refer to Figure 1 As shown, in an embodiment of the present invention, the value of the weight coefficient A of all input target patterns is, for example, 0.8, and the weight coefficient B of all asymmetric patterns is, for example, 0.2. Therefore, the cost function can be expressed as: Cost function = 0.8×(RMS-all / RMS-all-spec) + 0.2×(RMS-part / RMS-part-spec).

[0059] In other embodiments, the specific values of the weight coefficient A of all input target patterns and the weight coefficient B of all asymmetric patterns can also be readjusted.

[0060] Please refer to Figure 1 and Figure 4 As shown, in an embodiment of the present invention, step S50 specifically includes steps S501 to S504.

[0061] Step S501: Select a parameter within the parameter range.

[0062] Step S502: Input all target pattern data into the optical proximity correction model to obtain the simulated pattern data corresponding to each target pattern data.

[0063] Specifically, input all target pattern data into the optical model and the resist chemical reaction model in the optical proximity correction model, and use the optical proximity correction model to simulate the wafer pattern data formed by all target pattern data on the wafer, that is, the simulated pattern data. Specifically, please combineFigure 3 As shown, when the optical proximity correction model simulates the wafer pattern data formed by the target pattern data on the wafer, it adjusts the optical signal curve 210 and the exposure threshold 211 according to the kernel and parameters. Among them, the simulated pattern data is the position data of two intersection points of the optical signal curve 210 and the exposure threshold 211 of the wafer pattern formed by simulating the target pattern using the optical proximity correction model on the wafer. In this embodiment, the simulated pattern data is the position data of the left intersection point L and the right intersection point R where the optical signal curve 210 intersects the exposure threshold 211. Therefore, in the process of repeatedly iterating and executing steps S501 to S504, in step S502, for each target pattern under each parameter within the parameter range, a position data of the left intersection point L and the right intersection point R where the optical signal curve 210 intersects the exposure threshold 211 will be obtained, that is, the simulated pattern data of each target pattern.

[0064] Step 503: Obtain the cost function under this parameter according to the simulated pattern data and the wafer pattern data corresponding to each target pattern data.

[0065] Specifically, under the parameters set in step S501, obtain the simulated pattern data and the wafer pattern data corresponding to all target pattern data, including the simulated pattern data and the wafer pattern data corresponding to symmetric patterns, and the simulated pattern data and the wafer pattern data corresponding to asymmetric patterns. At this time, the cost function under this parameter can be calculated according to the formula of the cost function set in step S40. Specifically, obtain the root mean square of the fitting errors of all target patterns, that is, RMS-all, according to the simulated pattern data and the wafer pattern data corresponding to all target pattern data. Obtain the root mean square of the fitting errors of all asymmetric patterns, that is, RMS-part, according to the simulated pattern data and the wafer pattern data corresponding to all asymmetric patterns. Therefore, in the process of repeatedly iterating and executing steps S501 to S504, a cost function corresponding to the parameter can be obtained in step S503.

[0066] Step S504: Determine whether the parameters within the parameter range have been traversed. If the parameters within the parameter range have been traversed, end the process; if the parameters within the parameter range have not been traversed, return to step S501.

[0067] It should be noted that in each loop, the parameters selected in step S501 within the parameter range are not repeated. Under each parameter within the set parameter range, repeatedly iterate and execute the content of steps S501 to S504, and a cost function can be obtained under each parameter within the parameter range. In this embodiment, if the number of target patterns is N, then under each parameter, the loop of repeatedly iterating and executing the content of steps S501 to S504 is also N times.

[0068] Please refer to Figure 1As shown, in an embodiment of the present invention, in step S60, multiple cost functions obtained in step S50 can be compared to obtain the minimum cost function, and the optical proximity correction model when the cost function is minimized is output.

[0069] Please refer to Figure 1 As shown, in an embodiment of the present invention, in step S70, it is determined whether the minimum cost function is less than a preset cost. If the cost function is less than the preset cost, it means that the optical proximity correction model corresponding to the current minimum cost function meets the requirements, and the optical proximity correction model corresponding to the minimum cost function is used as the target optical proximity correction model to complete the modeling process.

[0070] Please refer to Figure 1 As shown, in an embodiment of the present invention, in step S70, if the cost function is greater than or equal to the preset cost, it means that the optical proximity correction model corresponding to the current minimum cost function still does not meet the requirements, and it is necessary to add a new kernel and adjust the parameter range, and then re-determine the optical proximity correction model.

[0071] Please refer to Figure 1 As shown, in an embodiment of the present invention, in step S70, through the setting of the cost function, the kernel and parameter range in the optical proximity correction model can be adjusted in a specific target direction. In this application, data of asymmetric graphics is introduced into the cost function, so that the kernel and parameter range in the optical proximity correction model are adjusted towards the range beneficial to asymmetric graphics, improving the accuracy of asymmetric graphics correction.

[0072] Please refer to Figure 1 、 Figure 5 and Figure 6As shown, during the modeling process of the optical proximity correction model, the position data (simulated pattern data) of the left intersection point L and the right intersection point R where the optical signal curve 210 intersects with the exposure threshold 211 are mainly adjusted to coincide with the designed wafer pattern data. When the target pattern is a symmetric pattern, since the symmetric pattern is centrosymmetric, it is easy to confirm the position data of the left intersection point L and the right intersection point R. For example, by subtracting the designed wafer pattern data from the target pattern data and then dividing by 2, the first distance Y1 from the left intersection point L to the target pattern data and the second distance Y2 from the right intersection point R to the target pattern data can be obtained. Then, subtracting the first distance Y1 from the target pattern data can obtain the data of the left intersection point L, and adding the second distance Y2 to the target pattern data can obtain the data of the right intersection point R. However, for an asymmetric pattern, if the left intersection point L and the right intersection point R are obtained in this way, there will be a large error between the simulated pattern data and the designed wafer pattern data. Since it is difficult to determine the first distance Y1 from the left intersection point L to the target pattern data and the second distance Y2 from the right intersection point R to the target pattern data, only by adjusting the kernel and parameter range can the position data of the left intersection point L and the right intersection point R where the optical signal curve 210 intersects with the exposure threshold 211 be continuously approximated to the designed wafer pattern data. In this application, by adding the relevant data of the asymmetric pattern to the cost function and adjusting the weights corresponding to the relevant data of the asymmetric pattern, the obtained target optical proximity correction model can improve the fitting result of the asymmetric pattern while ensuring the fitting degree of the symmetric pattern. Specifically, the modeling method of the optical proximity correction model in this application can reduce the root mean square of the fitting error of the asymmetric pattern in the active layer from 2.5 to 1.5, with the accuracy rate increased by 40%, and reduce the root mean square of the fitting error of the asymmetric pattern in the metal layer from 2.8 to 1.4, with the accuracy rate increased by 50%.

[0073] Please refer to Figure 7As shown in the figure, the present invention further provides a modeling system for an optical proximity correction model, and the modeling system includes a data acquisition module 201, a graphic discrimination module 202, a kernel and parameter range adjustment module 203, a cost function setting module 204, an optical proximity correction and cost function output module 205, an optical proximity correction model output module 206, a judgment module 207, and a target optical proximity correction model output module 208. Among them, the data acquisition module 201 executes step S10, acquires a plurality of target graphic data, and the wafer graphic data corresponding to each target graphic data. The graphic discrimination module 202 executes step S20, and discriminates the target graphics into symmetric graphics and asymmetric graphics. The kernel and parameter range adjustment module 203 executes step S30, and sets the initial kernel and parameter range of the optical proximity correction model. The cost function setting module 204 executes step S40, and sets a cost function representing the fitting degree. The optical proximity correction and cost function output module 205 executes step S50. Under each parameter within the parameter range, the target graphic data is input into the optical proximity correction model to obtain the simulated graphic data corresponding to the target graphic data, and based on the simulated graphic data and the wafer graphic data, the cost function under each parameter is obtained. The optical proximity correction model output module 206 executes step S60, and outputs the optical proximity correction model when the cost function within the parameter range is the smallest. The judgment module 207 executes step S70, and judges whether the minimum cost function is less than a preset cost. If the minimum cost function is less than the preset cost, then step S80 is executed; if the minimum cost function is greater than or equal to the preset cost, then return to step S30. The target optical proximity correction model output module 208 executes step S80, and uses the optical proximity correction model corresponding to the minimum cost function as the target optical proximity correction model.

[0074] Please refer to Figure 8 As shown in the figure, the present invention further provides a computer device, and this computer device can be a server. The computer device includes a processor, a memory, a network interface, and a database connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile and / or volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external client through a network connection. When the computer program is executed by the processor, it realizes the functions or steps of a modeling method for an optical proximity correction model.

[0075] In summary, the present invention provides a method, a system and a computer device for modeling an optical proximity correction model. The method for modeling the optical proximity correction model includes: obtaining a plurality of target graphic data and the corresponding wafer graphic data for each of the target graphic data; classifying the target graphics into symmetric graphics and asymmetric graphics; setting an initial kernel and parameter range of the optical proximity correction model, and setting a cost function representing the fitting degree, where the cost function includes the standard deviation of all target graphics and the standard deviation of the asymmetric graphics; inputting the target graphic data into the optical proximity correction model under each parameter within the parameter range, obtaining the simulated graphic data of the target graphics, and obtaining the cost function under each parameter based on the simulated graphic data and the wafer graphic data; outputting the optical proximity correction model when the cost function within the parameter range is the smallest; and determining whether the minimum cost function is less than a preset cost; if the minimum cost function is less than the preset cost, using the output optical proximity correction model as the target optical proximity correction model; if the minimum cost function is greater than or equal to the preset cost, adjusting the kernel and parameter range of the optical proximity correction model until the minimum cost function is less than the preset cost.

[0076] The method, the system and the computer device for modeling an optical proximity correction model provided by the present invention have unexpected effects: By modeling the optical proximity correction model, the processing ability of the optical proximity correction model for asymmetric design graphics is significantly improved, especially the correction accuracy when processing asymmetric graphics of key components such as static random access memories. The modeled optical proximity correction model can more accurately predict and correct the optical distortion of asymmetric graphics, thereby directly improving the result of optical proximity correction and the yield of products. By enabling the modeled optical proximity correction model to adapt to asymmetric graphics at the initial stage, the cumbersome adjustment and compensation work that engineers need to perform in the subsequent process is greatly reduced, thereby accelerating the R & D progress and improving the production efficiency. In addition, the application of this technical solution reduces the dependence on subsequent manual adjustment process steps and parameter compensation, reduces the risk of compensation failure, and further ensures the accuracy of electrical performance and the reliability of products.

[0077] The embodiments of the present invention disclosed above are only used to help illustrate the present invention. The embodiments do not describe all the details in detail, nor do they limit the invention to the specific embodiments described. Obviously, many modifications and variations can be made according to the content of this specification. These embodiments are selected and specifically described in this specification to better explain the principles and practical applications of the present invention, so that those skilled in the relevant technical fields can understand and utilize the present invention well. The present invention is only limited by the claims and their full scope and equivalents.

Claims

1. A method for modeling an optical proximity correction model, characterized in that Including the following steps: Obtain a plurality of target graphic data, and the wafer graphic data corresponding to each of the target graphic data; Classify the target graphics into symmetric graphics and asymmetric graphics; Set the initial kernel and parameter range of the optical proximity correction model, and set a cost function representing the fitting degree, where the cost function includes the standard deviation of all target graphics and the standard deviation of the asymmetric graphics; Under each parameter within the parameter range, input the target graphic data into the optical proximity correction model, obtain the simulated graphic data of the target graphics, and based on the simulated graphic data and the wafer graphic data, obtain the cost function under each parameter; Output the optical proximity correction model when the cost function within the parameter range is minimized; and Judge whether the minimum cost function is less than a preset cost; if the minimum cost function is less than the preset cost, use the output optical proximity correction model as the target optical proximity correction model; if the minimum cost function is greater than or equal to the preset cost, adjust the kernel and parameter range of the optical proximity correction model until the minimum cost function is less than the preset cost.

2. The modeling method of the optical proximity correction model according to claim 1, wherein Obtaining the cost function under each parameter includes the following steps: Select a parameter within the parameter range; Input all the target graphic data into the optical proximity correction model to obtain the simulated data graphics corresponding to each of the target graphic data; Based on the simulated graphic data corresponding to each of the target graphic data and the wafer graphic data, obtain the cost function under this parameter; and Judge whether the parameters within the parameter range have been traversed. If the parameters within the parameter range have been traversed, end the process. If the parameters within the parameter range have not been traversed, select another parameter within the parameter range and repeat the process of obtaining the cost function.

3. The modeling method of the optical proximity correction model according to claim 1, wherein The simulated graphic data is the position data of the two intersection points between the optical signal curve of the wafer graphic formed by simulating the target graphic on the wafer using the optical proximity correction model and the exposure threshold.

4. The modeling method of the optical proximity correction model according to claim 1, characterized in that, The cost function is obtained through the following formula: Cost function = A × (RMS - all / RMS - all - spec) + B × (RMS - part / RMS - part - spec); Where, Cost function is the cost function, RMS - all is the standard deviation of all input target graphics, RMS - all - spec is the specified value of the standard deviation of all input target graphics, A is the weight coefficient of all input target graphics, RMS - part is the standard deviation of all asymmetric graphics, RMS - part - spec is the specified value of the standard deviation of all asymmetric graphics, and B is the weight coefficient of all asymmetric graphics.

5. The modeling method of the optical proximity correction model according to claim 4, characterized in that, RMS-all is the root mean square of the fitting errors of all the target patterns, which characterizes the error between the simulated patterns of all the target patterns and the wafer patterns. RMS-part is the root mean square of the fitting errors of all the asymmetric patterns, which characterizes the error between the simulated patterns of all the asymmetric patterns and the wafer patterns.

6. The modeling method of the optical proximity correction model according to claim 4, characterized in that The standard deviation RMS-all of all the input target patterns is obtained by the following formula: RMS-all = SQRT(SUMSQ E(1, N) / COUNTA E(1, N)); The standard deviation RMS-part of all the asymmetric patterns is obtained by the following formula: RMS-part = SQRT(SUMSQ E(b + 1, N) / COUNTA E(b + 1, N)); Wherein, SQRT represents the square root function, SUMSQ represents the sum of squares function, COUNTA represents the function of counting non-empty cells within a given range, E(1, N) is the set of errors between the simulated pattern data and the wafer pattern data corresponding to all the target patterns, and E(b + 1, N) is the set of errors between the simulated pattern data and the wafer pattern data corresponding to the asymmetric patterns.

7. The modeling method of the optical proximity correction model according to claim 4, characterized in that The value of the weight coefficient A of all the input target patterns is 0.8, and the value of the weight coefficient B of all the asymmetric patterns is 0.

2.

8. The modeling method of the optical proximity correction model according to claim 1, characterized in that Classifying the target patterns into symmetric patterns and asymmetric patterns includes the following steps: Inputting the target pattern data and the wafer pattern data into the optical proximity correction model; Using the optical proximity correction model to simulate the optical signal curve of the wafer pattern formed by the target pattern on the wafer; Determining the exposure threshold according to the fact that the distance between the two intersection points of the optical signal curve and the exposure threshold is equal to the wafer pattern data; Obtaining the distances from the two intersection points of the optical signal curve and the exposure threshold to the target pattern data to determine the two intersection points of the optical signal curve and the exposure threshold; and Judging whether the distances from the two intersection points of the optical signal curve and the exposure threshold to the target pattern data are equal. If the distances from the two intersection points of the optical signal curve and the exposure threshold to the target pattern data are equal, then the target pattern is determined to be the symmetric pattern. If the distances from the two intersection points of the optical signal curve and the exposure threshold to the target pattern data are not equal, then the target pattern is determined to be the asymmetric pattern.

9. A modeling system for an optical proximity correction model, characterized in that, Including: A data acquisition module, which acquires a plurality of target pattern data and the wafer pattern data corresponding to each target pattern data; A pattern classification module, which classifies the target patterns into symmetric patterns and asymmetric patterns; A kernel and parameter range adjustment module, which sets the initial kernel and parameter range of the optical proximity correction model; A cost function setting module, which sets a cost function representing the fitting degree, and the cost function includes the standard deviation of all the target patterns and the standard deviation of the asymmetric patterns; An optical proximity correction and cost function output module, at each parameter within the parameter range, inputs the target pattern data into the optical proximity correction model, obtains the simulated pattern data of the target pattern, and obtains the cost function at each parameter based on the simulated pattern data and the wafer pattern data; An optical proximity correction model output module, outputs the optical proximity correction model when the cost function within the parameter range is minimized; and A judgment module, judges whether the minimum cost function is less than a preset cost; and A target optical proximity correction model output module, when the minimum cost function is less than the preset cost, takes the output optical proximity correction model as the target optical proximity correction model; wherein, if the minimum cost function is greater than or equal to the preset cost, the kernel and parameter range adjustment module adjusts the kernel and parameter range of the optical proximity correction model until the minimum cost function is less than the preset cost.

10. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the modeling method of the optical proximity correction model according to any one of claims 1 to 8.

Images