OPC model data collection method, data collection system and OPC model optimization method

Abstract

Disclosed are an OPC model data collection method, a data collection system, and an OPC model optimization method. The OPC model data collection method includes: simulating a Manhattan layout of a test pattern using an initial OPC model to obtain a simulation result; inputting the simulation result into a measurement tool to establish a feature size measurement standard; and collecting the feature size of an actual wafer pattern of the test pattern using the feature size measurement standard to obtain collected data. The OPC model data collection method and data collection system simulate the Manhattan layout of the test pattern using an initial OPC model, and then use the simulation result to establish a feature size measurement standard. The simulated layout has a small difference from the actual wafer size. The feature size measurement standard established in this way can better match the actual wafer pattern, ensuring good data collection results, making data collection accurate, and thus obtaining more accurate model calibration results.

Description

Technical Field

[0001] The present invention relates to the field of semiconductor manufacturing technology, and in particular to an OPC model data collection method, a data collection system and an OPC model optimization method. Background Art

[0002] Photolithography is a crucial industrial step in semiconductor device production, transferring the pattern structure printed on a photomask to the surface of a wafer. With the advancement of integrated circuits, semiconductor manufacturing technology has continuously progressed toward smaller dimensions, with the feature size of semiconductor devices even smaller than the wavelength of the light source used in the photolithography process. In this situation, due to the diffraction of light, the pattern on the mask can be distorted during transfer, a phenomenon known as the optical proximity effect. This optical proximity effect can cause significant discrepancies between the actual pattern projected onto the wafer and the designed target pattern, compromising the lithography quality of adjacent pattern areas on the mask, ultimately impacting circuit performance and production yield.

[0003] To eliminate the effects of the optical proximity effect, Optical Proximity Correction (OPC) is commonly used. This method uses computer software to correct the original pattern to be exposed on the semiconductor substrate of the silicon wafer, obtaining a target pattern that is different from the original. A photomask is then created based on this target pattern. During photolithography, the resulting pattern projected onto the semiconductor substrate using this photomask is virtually identical to the original pattern, thus compensating for the effects caused by the optical proximity effect.

[0004] Currently, OPC model corrections utilize CD-SEM (Critical Dimension-Scanning Electron Microscope) testing to collect critical dimensions of chips on masks or wafers. This collected data is then filtered and fed back to the OPC model for comprehensive calculations based on the modeling information. However, due to the optical proximity effect of the wafer itself during data collection, the dimensions of the patterns measured using the CD-SEM standard deviate significantly from the actual pattern dimensions, making it impossible to accurately capture the wafer's feature dimensions. This makes data collection and screening complex and time-consuming, which invisibly increases the accuracy of the feedback data, prolongs the correction process, and increases production costs. Summary of the Invention

[0005] In view of the above problems, the object of the present invention is to provide an OPC model data collection method, a data collection system and an OPC model optimization method to solve the problems in the prior art.

[0006] According to a first aspect of the present invention, there is provided an OPC model data collection method, comprising:

[0007] The initial OPC model is used to simulate the Manhattan layout of the test pattern to obtain the simulation results;

[0008] Inputting the simulation results into a measurement tool to establish a feature size measurement standard; and

[0009] The characteristic size measurement standard is used to collect the characteristic size of the actual wafer pattern of the test pattern to obtain collected data.

[0010] Optionally, after the step of collecting the characteristic dimensions of the actual wafer pattern of the test pattern using the characteristic dimension measurement standard to obtain collected data, the method further includes:

[0011] Inputting the collected data into the initial OPC model to perform model optimization to obtain an iterative OPC model;

[0012] Data collection was repeated using the iterative OPC model.

[0013] Optionally, before the step of simulating the Manhattan layout of the test pattern using the initial OPC model to obtain a simulation result, the method further includes:

[0014] An initial OPC model is established based on the basic data of the test pattern and the historical collected data.

[0015] Optionally, after the step of re-collecting data using the iterative OPC model, the method further includes:

[0016] The collected data is input into the iterative OPC model to calibrate the iterative OPC model.

[0017] Optionally, the step of simulating the Manhattan layout of the test pattern using the initial OPC model to obtain a simulation result includes:

[0018] The initial OPC model is used to simulate the Manhattan layout of the test pattern of the one-dimensional structure to obtain the first simulation result.

[0019] The characteristic dimension measurement standard established by using the first simulation result is the first characteristic dimension measurement standard, and the collected data collected by using the first characteristic dimension measurement standard is the first collected data.

[0020] Optionally, re-collecting data using the iterative OPC model includes:

[0021] Using the iterative OPC model to simulate the Manhattan layout of the test pattern of the two-dimensional structure to obtain a second simulation result;

[0022] Inputting the second simulation result into a measurement tool to establish a second feature size measurement standard;

[0023] The second characteristic size measurement standard is used to collect characteristic sizes of actual wafer patterns of the test pattern of the two-dimensional structure to obtain second collected data.

[0024] Optionally, after the step of simulating the Manhattan layout of the test pattern of the one-dimensional structure using the initial OPC model to obtain a first simulation result, the method further includes:

[0025] Data with a simulation data size smaller than a threshold value in the first simulation result is filtered out.

[0026] According to a second aspect of the present invention, there is provided an OPC model optimization method, comprising:

[0027] The above-mentioned OPC model data collection method is used for data collection;

[0028] The OPC model is optimized according to the difference between the acquired feature size of the actual wafer pattern and the feature size of the test pattern.

[0029] According to a third aspect of the present invention, there is provided an OPC model data collection system, comprising:

[0030] A first simulation module, using an initial OPC model to simulate a Manhattan layout of a test pattern of a one-dimensional structure to obtain a first simulation result;

[0031] A first standard establishing module, inputting the first simulation result into a measurement tool to establish a first feature size measurement standard; and

[0032] The first collection module collects the characteristic dimensions of the actual wafer pattern of the test pattern using the first characteristic dimension measurement standard to obtain first collected data.

[0033] Optionally, the OPC model data collection system further includes:

[0034] Initial model building module, which builds the initial OPC model based on the basic data of the test pattern and the historical collected data;

[0035] A data screening module, screening out data in the first simulation result whose simulation data size is smaller than a threshold;

[0036] An iteration module, inputting the first collected data into the initial OPC model to perform model optimization to obtain an iterative OPC model;

[0037] a second simulation module, simulating a Manhattan layout of a test pattern of a two-dimensional structure using the iterative OPC model to obtain a second simulation result;

[0038] a second standard establishing module, inputting the second simulation result into a measurement tool to establish a second characteristic dimension measurement standard;

[0039] a second collecting module, which collects characteristic dimensions of an actual wafer pattern of the test pattern of the two-dimensional structure using the second characteristic dimension measurement standard to obtain second collected data;

[0040] The calibration module inputs the collected data into the iterative OPC model to calibrate the iterative OPC model.

[0041] The OPC model data collection method, data collection system, and OPC model optimization method provided by the present invention first use an initial OPC model to simulate the Manhattan layout of a test pattern, then use the obtained simulation results to establish a feature size measurement standard, and use the established feature size measurement standard to measure the actual size of the wafer to achieve data collection. Because the pattern obtained after the Manhattan layout is simulated by the OPC model is relatively consistent with the shape of the etched pattern of the actual wafer, the size of the data collected according to the feature size measurement standard is relatively close to the size of the actual wafer, thereby ensuring good data collection results, making data collection accurate, saving data collection and data screening time, reducing production costs, and further obtaining more accurate model calibration results.

[0042] Furthermore, the data collection method of the OPC model of the present invention first uses the Manhattan layout of the test pattern of the one-dimensional structure to simulate, establishes the feature size measurement standard according to the simulation results, collects data based on this, and then returns the collected results to the OPC model for optimization and iteration to obtain an iterative OPC model. The iterative OPC model is then used to simulate the Manhattan layout of the test pattern of the two-dimensional structure, establishes the feature size measurement standard according to the simulation results, and collects data based on this, thereby realizing the gradual collection of one-dimensional data to two-dimensional data. By collecting and screening the one-dimensional data, the model can be made more accurate when collecting two-dimensional data, reducing the difficulty of directly collecting two-dimensional data, and improving the collection accuracy of two-dimensional data, so that the data collection results are more accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The above and other objects, features and advantages of the present invention will become more apparent through the following description of the embodiments of the present invention with reference to the accompanying drawings, in which:

[0044] Figure 1a-Figure 1e The electron microscope images obtained by collecting multiple data using the traditional data collection method are shown respectively;

[0045] Figure 2 A flow chart of an OPC model data collection method according to a first embodiment of the present invention is shown;

[0046] Figure 3a-3c Comparison diagrams before and after simulation of multiple Manhattan layouts according to the OPC model data collection method according to an embodiment of the present invention are respectively shown;

[0047] Figure 4a-4b Two exemplary schematic diagrams of data collection results of the OPC model data collection method according to an embodiment of the present invention are respectively shown;

[0048] Figure 5 A flow chart of an OPC model data collection method according to a second embodiment of the present invention is shown;

[0049] Figure 6 A simplified flowchart of an OPC model optimization method according to an embodiment of the present invention is shown;

[0050] Figure 7 A schematic block diagram of an OPC model data collection system according to an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0051] Various embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. In each of the accompanying drawings, identical elements are represented by identical or similar reference numerals. For the sake of clarity, the various parts in the accompanying drawings are not drawn to scale. In addition, some well-known parts may not be shown.

[0052] The specific implementation of the present invention is further described in detail below with reference to the accompanying drawings and examples.

[0053] Figure 1a-Figure 1e Electron microscope images obtained by collecting multiple data using traditional data collection methods are shown respectively.

[0054] During the OPC model building process, data collection is required. The model must be adjusted and optimized based on the collected data before the mask can be repaired. The most time-consuming part of the modeling process is data collection and organization. Currently, data collection primarily uses Hitachi's measuring instruments and Applied Materials' measurement tools. By inputting the Manhattan layout (GDS) (Graphic Data Stream) of the test pattern, measurement positions and boundaries are identified. CD-SEM measurement standards are then established offline, and the measurement standards are then used to measure the actual wafer size online to obtain collected data. Data processing involves processing collected data containing invalid, failed, or erroneous measurements. After removing these problematic data, manual offline measurement corrections or remeasurements are performed to recollect the data.

[0055] When using the above measurement tools to input Manhattan layout to establish CD-SEM measurement standards, problems such as invalid data measurement, measurement failure and measurement error may occur. Figure 1a-Figure 1e .

[0056] like Figure 1a As shown, the feature size of the measured pattern is too small, which easily causes the patterns to be joined together, or it is too small to be captured. The white dotted line in the figure is the positioning point, and the required data cannot be measured, resulting in invalid measurement.

[0057] like Figure 1b and Figure 1c As shown in the figure, the white dashed lines represent measurement points, and the pattern surrounded by the solid white lines represents an image of the actual wafer under a scanning electron microscope. It is clearly observed that some of the measurement points along the white dashed lines extend beyond the actual wafer locations that should be measured, resulting in measurement failure. This failure is primarily due to the complexity of the wafer pattern, which prevents the offline CD-MES measurement frame from accurately capturing the pattern boundaries.

[0058] like Figure 1d and Figure 1e As shown in the figure, the white dotted line is the boundary of the Manhattan layout, and the pattern surrounded by the white solid line is the image of the actual wafer under a scanning electron microscope. The two overlap. When the Manhattan layout is used to identify the measurement position and boundary of the actual wafer, it is obvious that there is a large deviation between the Manhattan layout pattern and the actual wafer pattern, for example Figure 1d Square boundary graphics and elliptical boundary graphics, for example Figure 1e Square and circular graphics. Therefore, the existing measurement standards established using Manhattan graphics cannot intuitively reflect the actual wafer graphics shape, resulting in incorrect selection of measurement points and thus measurement errors.

[0059] In summary, during the data collection phase, the CD-SEM measurement standard established offline through the Manhattan layout of the traditional OPC model cannot accurately present the actual shape of the wafer due to the large difference between the actual wafer pattern and the actual shape of the wafer, resulting in a series of measurement problems, making data collection and screening difficult, time-consuming and costly. In response to this, this application proposes an improved OPC model data collection method that can improve the accuracy of data collection. Figure 2-Figure 7 Detailed introduction.

[0060] Figure 2 A flow chart of an OPC model data collection method according to a first embodiment of the present invention is shown.

[0061] like Figure 2 As shown, an OPC model data collection method is provided, and the specific steps include:

[0062] In step S101 , an initial OPC model is established based on basic data of a test pattern and historically collected data.

[0063] In this step, an initial OPC model is established based on the basic data of the test pattern collected and evaluated in the previous process window and the historical collection data. The OPC model is based on the historical collection data of the test pattern. The initial OPC model is subsequently used to optimize data collection. Data collection is also called data acquisition.

[0064] In step S102 , the initial OPC model is used to simulate the Manhattan layout of the test pattern to obtain a simulation result.

[0065] In this step, the initial OPC model established in the previous step is used to simulate the Manhattan layout of the test pattern to obtain a simulated graphic of the Manhattan layout, and the simulated graphic has a high degree of conformity with the shape of the actual pattern of the wafer. Generally, generating an OPC test mask requires generating multiple Manhattan layout GDS (Graphic DataStream, Graphic Data Stream) files of the test pattern first, each GDS file includes a type of OPC test pattern, and then several GDS files are arranged in the effective exposure area of the mask as required, and JDV (Job Deck View, Photomask Data Detection) file is generated and output. The shape of the Manhattan layout is relatively consistent with the ideal test pattern, and the corners are relatively square. Due to the influence of light diffraction, the corners of the pattern formed on the silicon substrate of the wafer have a certain curvature. It is not easy to capture the measurement points of the actual pattern of the wafer by directly establishing the CD-MES measurement standard through the Manhattan layout, resulting in measurement errors, failures and other problems. The simulated pattern obtained by simulating the Manhattan layout using the OPC model is actually a pattern with curved corners, which is very consistent with the actual etched pattern of the wafer. In this way, the established CD-MES measurement standard can accurately capture the shape and size of the actual pattern of the wafer.

[0066] In step S103 , the simulation results are input into a measurement tool to establish a feature size measurement standard.

[0067] In this step, the simulation results of the Manhattan layout obtained in the previous step are input into the Hitachi Designgauge (Hitachi test tool) or AMAT OPCC (Applied Materials test tool) to establish the CD-SEM feature size measurement standard. The feature size measurement standard established based on the simulated pattern can accurately capture the feature size of the actual wafer pattern.

[0068] In step S104 , the feature size of the actual wafer pattern of the test pattern is collected using a feature size measurement standard to obtain collected data.

[0069] In this step, the feature size measurement standard established in the previous step is used to collect the feature size of the actual wafer pattern of the test pattern as collected data.

[0070] In step S105 , the collected data is input into the initial OPC model for model optimization to obtain an iterative OPC model.

[0071] In this step, the collected data obtained by adopting the above-mentioned feature size measurement standard is highly consistent with the feature size data of the actual pattern of the wafer. This collected data is input into the initial OPC model for model optimization to obtain an iterative OPC model. The new model is then used to collect data again, so that the results obtained are more accurate.

[0072] In step S106 , data collection is performed again using the iterative OPC model.

[0073] In this step, the obtained iterative OPC model is used to re-collect data to obtain more accurate collection results, so as to make subsequent adjustments and optimizations to the OPC model. The difference between the mask template and the actual wafer size can be accurately obtained, so as to better correct the model and make the wafer etching more accurate.

[0074] The OPC model data collection method provided by the present invention first uses an initial OPC model to simulate the Manhattan layout of a test pattern, then uses the obtained simulation results to establish a feature size measurement standard, and uses the established feature size measurement standard to measure the actual size of the wafer to achieve data collection. Because the pattern obtained after the Manhattan layout is simulated by the OPC model is highly consistent with the shape of the etched pattern on the actual wafer, the size of the data collected according to the feature size measurement standard is relatively close to the size of the actual wafer, thereby ensuring good data collection results, making data collection accurate, saving data collection and data screening time, reducing production costs, and further obtaining more accurate model calibration results.

[0075] Figure 3a-3c Comparison diagrams before and after simulation of multiple Manhattan layouts by the OPC model data collection method according to an embodiment of the present invention are respectively shown.

[0076] Figure 3a-3c reflects Figure 2 The comparison diagram before and after the simulation of step S102 shows that by simulating the Manhattan layout using the OPC model, a layout that completely approximates the actual wafer pattern morphology can be obtained. By inputting this simulated layout into the test tool, a correct and reliable CD-SEM measurement standard can be established to achieve correct measurement of subsequent graphics.

[0077] Figure 3a-3cIn the figure, the left side is the Manhattan layout of the test pattern, and the right side is the simulated pattern after simulation. By comparison, we can see that the simulated pattern has a larger change in shape than the Manhattan layout, which can more realistically reflect the actual wafer morphology. The CD-MES standard established by the simulated pattern can basically solve the problem of Figure 1a-Figure 1e The system solves the problems of invalid measurement, measurement failure and measurement error during the measurement process. At the same time, it improves the measurement accuracy, saves a lot of time and cost, and provides a reliable foundation for the automatic offline establishment of CD-SEM feature size measurement standards.

[0078] Figure 4a-4b Two exemplary schematic diagrams of data collection results of the OPC model data collection method according to an embodiment of the present invention are shown respectively.

[0079] Figure 4a-4b The display shows the display results under an electron microscope of the actual wafer pattern captured using the data collection method of the first embodiment of the present invention. The black solid line in the figure represents the measurement point for data capture using the characteristic size measurement standard, and the black pattern represents the display pattern of the actual wafer under the electron microscope. It can be seen that the capture boundary of the measurement point is highly consistent with the pattern of the actual wafer, and the wafer size can be accurately collected. This further verifies that the characteristic size of the actual wafer can be accurately measured in one go using the OPC model data collection method of the first embodiment of the present invention, reducing the time spent in data collection and data collation, reducing the difficulty of data collation, and effectively solving the problems in the prior art.

[0080] Figure 5 A flow chart of an OPC model data collection method according to a second embodiment of the present invention is shown.

[0081] The OPC model data collection method of the second embodiment of the present invention includes steps S201-S210, which are described step by step below. This embodiment is substantially similar to the first embodiment, but differs from the first embodiment in that, in this embodiment, simulation is first performed using data from a one-dimensional test pattern. The collected data is then returned to the OPC model for optimization iteration. Simulation is then performed again using a two-dimensional test pattern to obtain collected data again, thereby ensuring the accuracy of the collected data, improving the precision of the model, and enhancing the accuracy of the final data collection results.

[0082] In step S201 , an initial OPC model is established based on basic data of a test pattern and historically collected data.

[0083] This step and Figure 2 Step S101 of the embodiment is the same and will not be described again here.

[0084] In step S202 , the initial OPC model is used to simulate the Manhattan layout of the test pattern of the one-dimensional structure to obtain a first simulation result.

[0085] In this step, using the initial OPC model, a one-dimensional test pattern is first selected as the first set of measurement points. The initial OPC model is then used to simulate the Manhattan layout of this one-dimensional test pattern to obtain the first simulation result. Generally, test patterns can be divided into one-dimensional and two-dimensional patterns. One-dimensional patterns are simple patterns with periodicity in only one direction, such as lines. Two-dimensional patterns are complex patterns with periodicity in both directions, such as contact holes, L-shaped patterns, and patchwork patterns. Because of the simplicity of the lines, testing one-dimensional patterns is simpler than testing two-dimensional patterns, making it easier to obtain simulation results.

[0086] In step S203 , data in the first simulation result whose simulation data size is smaller than a threshold is filtered out.

[0087] In this step, based on the simulation results of the test pattern of the one-dimensional structure, graphics with too small data size are screened out. That is, when the data size of the graphics obtained after simulation is less than a certain threshold, the graphic data is deleted. Such data will cause subsequent measurements to be invalid. After screening out these data and then establishing the feature size measurement standard, the problem of invalid measurement of some points can be solved, avoiding repeated screening in the data sorting stage.

[0088] In step S204 , the first simulation result is input into a measurement tool to establish a first feature size measurement standard.

[0089] This step is similar to step S103 , where the first simulation result of the one-dimensional structure test pattern is input to the test tool to establish a first feature size measurement standard (CD-SEM measurement standard).

[0090] In step S205 , a first feature size measurement standard is used to collect feature sizes of actual wafer patterns of the test pattern to obtain first collected data.

[0091] In this step, the first feature size measurement standard established in the previous step is used to collect the feature size of the actual wafer pattern of the test pattern, and the obtained data is used as the first collected data.

[0092] In step S206 , the first collected data is input into the initial OPC model for model optimization to obtain an iterative OPC model.

[0093] This step is the same as step S105. The first collected data is re-input into the initial OPC model to perform model optimization iterations to simulate the test pattern of the two-dimensional structure and establish the measurement standard. This can reduce the difficulty and pressure of directly establishing the measurement standard of the two-dimensional structure test pattern and improve the accuracy of the measurement standard.

[0094] Figure 2 Step S106 may include steps S207 to S209 of this embodiment, and the specific steps are as follows.

[0095] In step S207 , the iterative OPC model is used to simulate the Manhattan layout of the test pattern of the two-dimensional structure to obtain a second simulation result.

[0096] In this step, the Manhattan layout of the test pattern of the two-dimensional structure is simulated again using the iterative OPC model to obtain a second simulation result of the two-dimensional pattern.

[0097] In step S208 , the second simulation result is input into a measurement tool to establish a second feature size measurement standard.

[0098] In this step, the second simulation result of the two-dimensional structure test pattern is input to the test tool to establish the second feature size measurement standard (CD-SEM measurement standard). This step mainly solves the problem of Figure 1b-Figure 1e Given the problems of measurement failure and measurement error.

[0099] In step S209 , a second feature size measurement standard is used to collect feature sizes of actual wafer patterns of the two-dimensional test pattern to obtain second collected data.

[0100] In this step, the characteristic dimensions of the actual wafer pattern of the test pattern are collected using the second characteristic dimension measurement standard established in the previous step, and the obtained data is used as the second collected data. In this way, the data collection of the one-dimensional pattern and the two-dimensional pattern is completed.

[0101] In step S210 , the collected data is input into the iterative OPC model to calibrate the iterative OPC model.

[0102] After steps S201-S209, data collection is essentially complete, and data organization takes virtually no time. This effectively addresses the data measurement inaccuracies inherent in existing technologies. Following these steps, the final OPC model calibration and calibration can be performed directly. Therefore, in this step, the collected data from both iterations is input into the iterative OPC model for calibration.

[0103] Of course, the calibration of the iterative OPC model may also be performed directly in the last step of the first embodiment.

[0104] The data collection method of the OPC model of this embodiment first uses the Manhattan layout of the test pattern of the one-dimensional structure to simulate, establishes the feature size measurement standard based on the simulation results, collects data based on this, and then returns the collected results to the OPC model for optimization and iteration to obtain an iterative OPC model. The iterative OPC model is then used to simulate the Manhattan layout of the test pattern of the two-dimensional structure, establishes the feature size measurement standard based on the simulation results, and collects data based on this, thereby realizing the gradual collection of one-dimensional data to two-dimensional data. By collecting and filtering the one-dimensional data, the model can be made more accurate when collecting two-dimensional data, reducing the difficulty of directly collecting two-dimensional data, and improving the collection accuracy of two-dimensional data, so that the data collection results are more accurate.

[0105] Figure 6 The following is a simplified flowchart of an OPC model optimization method according to an embodiment of the present invention. The OPC model optimization method of this embodiment actually uses the above-mentioned OPC model data collection method, and then uses the collected data to optimize the model. The specific steps are as follows:

[0106] In step S301 , an OPC model is established using basic data, historical data, and finite element analysis data.

[0107] and Figure 2 and Figure 5 The two embodiments are similar in that a preliminary OPC model is established using basic data, historical data, finite element analysis data, etc.

[0108] In step S302 , the test pattern data is simulated using the OPC model.

[0109] In this step, the OPC model established in the previous step is used to simulate the Manhattan layout of the test pattern. Here, the one-dimensional pattern and the two-dimensional pattern can also be simulated separately.

[0110] In step S303 , the simulation results are input into the CD-SEM, a measurement standard is established, and data is collected based on the standard to complete data simulation and testing.

[0111] In this step, the simulation results are used to establish CD-SEM measurement standards for data collection, completing the simulation and testing of the data. After step S303, the process returns to step S301 to perform model optimization iterations, and then re-execute steps S302 and S303 to complete the data collection after the iteration.

[0112] In step S304, data collection is completed.

[0113] After the above steps, the data collection process is completed, and then the model optimization is achieved through the collected data.

[0114] In step S305, the collected data is used to perform difference analysis to achieve model optimization.

[0115] In this step, the difference between the collected characteristic dimensions of the actual wafer and the characteristic dimensions of the test pattern is analyzed to achieve model correction and optimization. Specifically, the characteristic dimensions of the original pattern and the actual wafer pattern photoetched on the semiconductor substrate are sampled using the above-mentioned data collection method. This data is input into the OPC model, which is trained and the model parameters are adjusted to minimize the error between the characteristic dimensions of the simulated pattern and the characteristic dimensions of the actual pattern. The model with these parameters is then used to correct the mask pattern of the actual pattern. The resulting OPC model, after the mask is corrected, has a very small difference between the actual wafer pattern and the ideal pattern, and the wafer etching results are accurate.

[0116] Figure 7 A schematic block diagram of an OPC model data collection system according to an embodiment of the present invention is shown.

[0117] like Figure 7 The OPC model data collection system 700 includes: an initial model establishment module 701, a first simulation module 702, a data screening module 703, a first standard establishment module 704, a first collection module 705, an iteration module 706, a second simulation module 707, a second standard establishment module 708, a second collection module 709 and a calibration module 710.

[0118] The initial model establishment module 701 establishes an initial OPC model based on the basic data of the test pattern and the historical collected data; the first simulation module 702 uses the initial OPC model to simulate the Manhattan layout of the test pattern of the one-dimensional structure to obtain a first simulation result; the data screening module 703 screens out the data whose simulation data size is less than the threshold in the first simulation result; the first standard establishment module 704 inputs the first simulation result into the measurement tool to establish a first feature size measurement standard; the first collection module 705 uses the first feature size measurement standard to collect the feature size of the actual wafer pattern of the test pattern to obtain the first collected data; the iteration module 706 06 Input the first collected data into the initial OPC model for model optimization to obtain an iterative OPC model; the second simulation module 707 uses the iterative OPC model to simulate the Manhattan layout of the test pattern of the two-dimensional structure to obtain a second simulation result; the second standard establishment module 708 inputs the second simulation result into the measurement tool to establish a second feature size measurement standard; the second collection module 709 uses the second feature size measurement standard to collect the feature size of the actual wafer pattern of the test pattern of the two-dimensional structure to obtain second collected data; the calibration module 710 inputs the collected data into the iterative OPC model to calibrate the iterative OPC model.

[0119] In summary, the OPC model data collection method, data collection system, and OPC model optimization method provided by the present invention first use the initial OPC model to simulate the Manhattan layout of the test pattern, and then use the obtained simulation results to establish a feature size measurement standard, and use the established feature size measurement standard to measure the actual size of the wafer to achieve data collection. Since the shape of the Manhattan layout obtained after the OPC model simulation is relatively consistent with the shape of the etched pattern of the actual wafer, the size of the data collected according to the feature size measurement standard is relatively close to the size of the actual wafer, thereby ensuring good data collection results, making data collection accurate, saving data collection and data screening time, reducing production costs, and thus obtaining more accurate model calibration results. The OPC model data collection method is used to collect data and then optimize the OPC model so that the collected results are closer to the size of the actual wafer, and the true difference between the test pattern and the actual wafer can be effectively calculated, and the optimization of the OPC model can be better completed.

[0120] Furthermore, a Manhattan layout of a one-dimensional test pattern is first simulated, and a feature size measurement standard is established based on the simulation results. Data is then collected based on this. The collected results are then returned to the OPC model for optimization and iteration to obtain an iterative OPC model. The iterative OPC model is then used to simulate the Manhattan layout of a two-dimensional test pattern. A feature size measurement standard is established based on the simulation results, and data is collected based on this, achieving a gradual acquisition from one-dimensional data to two-dimensional data. By collecting and filtering one-dimensional data, the model for two-dimensional data acquisition can be made more accurate, reducing the difficulty of directly collecting two-dimensional data, improving the collection accuracy of two-dimensional data, and making the data collection results more accurate. At the same time, during the one-dimensional data collection process, graphic data with simulation results that are too small are screened out to avoid the impact of this data on the collection results, reducing the difficulty of subsequent data clearing, and ensuring the accuracy of two-dimensional data collection.

[0121] While embodiments of the present invention have been described above, these embodiments do not exhaustively describe all details and do not limit the invention to the specific embodiments described. Obviously, many modifications and variations are possible based on the above description. These embodiments are selected and described in detail in this specification in order to better explain the principles and practical applications of the present invention, thereby enabling those skilled in the art to better utilize the present invention and its modifications. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. An OPC model data collection method, comprising: The initial OPC model is used to simulate the Manhattan layout of the test pattern to obtain the simulation results; Inputting the simulation results into a measurement tool to establish a feature size measurement standard; as well as The characteristic size measurement standard is used to collect the characteristic size of the actual wafer pattern of the test pattern to obtain collected data.

2. The OPC model data collection method according to claim 1, wherein: After the step of collecting the characteristic dimensions of the actual wafer pattern of the test pattern using the characteristic dimension measurement standard to obtain collected data, the method further includes: Inputting the collected data into the initial OPC model to perform model optimization to obtain an iterative OPC model; Data collection was repeated using the iterative OPC model.

3. The OPC model data collection method according to claim 1, wherein: Before the step of simulating the Manhattan layout of the test pattern using the initial OPC model to obtain a simulation result, the method further includes: An initial OPC model is established based on the basic data of the test pattern and the historical collected data.

4. The OPC model data collection method according to claim 2, wherein: After the step of re-collecting data using the iterative OPC model, the method further includes: The collected data is input into the iterative OPC model to calibrate the iterative OPC model.

5. The OPC model data collection method according to claim 2, wherein: The steps of simulating the Manhattan layout of the test pattern using the initial OPC model to obtain simulation results include: The initial OPC model is used to simulate the Manhattan layout of the test pattern of the one-dimensional structure to obtain the first simulation result. The characteristic dimension measurement standard established by using the first simulation result is the first characteristic dimension measurement standard, and the collected data collected by using the first characteristic dimension measurement standard is the first collected data.

6. The OPC model data collection method according to claim 5, wherein: Re-collecting data using the iterative OPC model includes: Using the iterative OPC model to simulate the Manhattan layout of the test pattern of the two-dimensional structure to obtain a second simulation result; Inputting the second simulation result into a measurement tool to establish a second feature size measurement standard; The second characteristic size measurement standard is used to collect characteristic sizes of actual wafer patterns of the test pattern of the two-dimensional structure to obtain second collected data.

7. The OPC model data collection method according to claim 5, wherein: After the step of simulating the Manhattan layout of the test pattern of the one-dimensional structure using the initial OPC model to obtain a first simulation result, the method further includes: Data with a simulation data size smaller than a threshold value in the first simulation result is filtered out.

8. An OPC model optimization method, comprising: The OPC model data collection method according to any one of claims 1 to 7 is used for data collection; The OPC model is optimized according to the difference between the acquired feature size of the actual wafer pattern and the feature size of the test pattern.

9. An OPC model data collection system comprising: A first simulation module, using an initial OPC model to simulate a Manhattan layout of a test pattern of a one-dimensional structure to obtain a first simulation result; a first standard establishing module, inputting the first simulation result into a measurement tool to establish a first characteristic dimension measurement standard; as well as The first collection module collects the characteristic dimensions of the actual wafer pattern of the test pattern using the first characteristic dimension measurement standard to obtain first collected data.

10. The OPC model data collection system according to claim 9, further comprising: Initial model building module, which builds the initial OPC model based on the basic data of the test pattern and the historical collected data; A data screening module, screening out data in the first simulation result whose simulation data size is smaller than a threshold; An iteration module, inputting the first collected data into the initial OPC model to perform model optimization to obtain an iterative OPC model; a second simulation module, simulating a Manhattan layout of a test pattern of a two-dimensional structure using the iterative OPC model to obtain a second simulation result; a second standard establishing module, inputting the second simulation result into a measurement tool to establish a second characteristic dimension measurement standard; a second collecting module, which collects characteristic dimensions of an actual wafer pattern of the test pattern of the two-dimensional structure using the second characteristic dimension measurement standard to obtain second collected data; The calibration module inputs the first collected data and the second collected data into the iterative OPC model to calibrate the iterative OPC model.

Images