Method and apparatus for determining a lithography process window, terminal, computer program product

Abstract

The present disclosure provides a method and device for determining a lithography process window, a terminal and a computer program product. The method comprises: determining design layout data, mask parameters and a preset process parameter range; determining a plurality of virtual exposure conditions according to the preset process parameter range, and performing a virtual focus exposure matrix simulation to generate a simulation exposure result; obtaining a virtual process window formed by the virtual exposure conditions according to the simulation exposure result and using a first screening condition; determining a center of the virtual process window as an optimal virtual exposure condition; defining a physical experiment region in the virtual process window according to a second screening condition; selecting a part of the physical experiment region as an experiment condition set based on a third screening condition; using the virtual exposure condition specified by the experiment condition set to perform a physical focus exposure experiment, and determining an actual exposure dose and an actual focus distance according to an experiment result. The above technical solution can improve the efficiency of determining the lithography process window.

Description

Technical Field

[0001] This disclosure relates to the field of semiconductor manufacturing technology, and in particular to a method and apparatus, terminal, and computer program product for determining a photolithography process window. Background Technology

[0002] In semiconductor manufacturing processes, determining the lithography process window is a crucial step in ensuring chip yield and performance stability. Traditionally, the industry widely adopts the focus-exposure matrix (FEM) method, which involves conducting numerous physical exposure experiments under different dose and focal length combinations to plot the range of process parameters corresponding to a qualified pattern.

[0003] However, a large number of physical exposure experiments require multiple wafers and several weeks of equipment time, resulting in high resource and time costs.

[0004] Therefore, how to provide technical solutions to reduce the cost of determining the lithography process window and improve the efficiency of determining the lithography process window has become an urgent technical problem to be solved. Summary of the Invention

[0005] In view of this, the present disclosure provides a method, apparatus, terminal, and computer program product for determining a photolithography process window, which can improve the efficiency of determining the photolithography process window.

[0006] This disclosure provides a method for determining a photolithography process window, comprising: determining design layout data, mask parameters, and a preset process parameter range; wherein the design layout data includes at least one of a design graphic, a critical dimension of the design graphic, and an edge placement error of the design graphic; the mask parameters include at least one of a graphic to be exposed, a critical dimension of the graphic to be exposed, and an edge placement error of the graphic to be exposed; and the preset process parameter range includes a preset exposure dose range and a preset focal length range; determining multiple virtual exposure conditions based on the preset process parameter range, and performing virtual focus exposure matrix simulation to generate simulated exposure results under multiple virtual exposure conditions; and performing a pass / fail determination based on the simulated exposure results using a first screening condition to obtain a virtual process window formed by the screened virtual exposure conditions, wherein the first screening condition includes: critical dimension tolerance, edge placement error tolerance, and contour completion tolerance. At least one of the following: determining the virtual exposure conditions at the center of the virtual process window as the optimal virtual exposure conditions, including: optimal exposure dose and optimal focal length; defining the physical experimental area of ​​the virtual process window according to a second screening condition, wherein the second screening condition includes: the fluctuation range of the exposure dose is the optimal exposure dose ± dose offset, and the fluctuation range of the focal length is the optimal focal length ± focal length offset; selecting a portion of the virtual exposure conditions in the physical experimental area as an experimental condition set based on a third screening condition, wherein the third screening condition is that the number of virtual exposure conditions selected in the central area of ​​the physical experimental area is less than the number of virtual exposure conditions selected in the edge area; wherein the virtual exposure conditions specified in the experimental condition set are used to conduct physical focus exposure experiments, and the actual exposure dose and actual focal length of the final lithography process window are determined based on the experimental results.

[0007] Optionally, the preset exposure dose range and preset focal length range satisfy one or more of the following: the dose step of the preset exposure dose range is selected from 0.1mJ to 0.5mJ; the preset focal length range is selected from -120nm to +120nm; the focal length step corresponding to the preset focal length range is selected from 3nm to 10nm.

[0008] Optionally, the first screening criteria include one or more of the following: the acceptable range for the critical dimension is an error of less than or equal to 1 nm; the acceptable range for the edge placement error is less than or equal to 1 nm; the contour integrity includes: no broken lines, no bridging, no wavy defects, and end indentation of less than or equal to 2 nm.

[0009] Optionally, the design graphic includes an ISO-type isolated graphic structure; wherein, in the first screening condition, the contour integrity further includes the detection of necking defects and footing defects in the ISO-type isolated graphic structure.

[0010] Optionally, based on the third screening condition, a portion of the virtual exposure conditions in the physical experimental area is selected as the experimental condition set, including: the number of virtual exposure conditions in the experimental condition set is less than or equal to 1 / 8 of the number of virtual exposure conditions generated by performing virtual focus exposure matrix simulation.

[0011] Optionally, before using the first screening condition to determine the pass / fail status based on the simulated exposure results, the method further includes: using a machine learning model to analyze the design layout data, mask parameters, preset process parameter range, and the simulated exposure results, and outputting a predicted virtual exposure condition range, wherein the first screening condition is used to determine the pass / fail status of the simulated exposure results obtained from the predicted virtual exposure condition range; wherein the training process of the machine learning model includes: acquiring training data, the training data including: design layout sample data, mask sample parameters, sample process parameter range, and sample physical exposure pattern; using the training data as input features and the corresponding sample exposure dose and sample focal length as output features to construct a training sample set; and using the training sample set to train the machine learning model; wherein the machine learning model is selected from one of random forest, support vector regression, neural network, or Gaussian process regression.

[0012] Optionally, the third screening condition includes: the number of virtual exposure conditions in the experimental condition set is less than or equal to 10; and / or, the ratio of the number of virtual exposure conditions selected in the edge region to the number of virtual exposure conditions selected in the center region of the physical experimental area is not less than 3:1.

[0013] This disclosure also provides an apparatus for determining a photolithography process window, comprising: a data input module for determining design layout data, mask parameters, and a preset process parameter range, wherein the design layout data includes at least one of a design graphic, a critical dimension of the design graphic, and an edge placement error of the design graphic; the mask parameters include at least one of a graphic to be exposed, a critical dimension of the graphic to be exposed, and an edge placement error of the graphic to be exposed; and the preset process parameter range includes a preset exposure dose range and a preset focal length range; a virtual simulation module for determining multiple virtual exposure conditions based on the preset process parameter range and performing virtual focus exposure matrix simulation to generate simulated exposure results under multiple virtual exposure conditions; and a first screening module for performing a pass / fail determination based on the simulated exposure results using a first screening condition to obtain a virtual process window formed by the screened virtual exposure conditions, wherein the first screening condition includes: critical dimension tolerance, edge placement error tolerance, and contour integrity. At least one of the following: determining the virtual exposure conditions at the center of the virtual process window as the optimal virtual exposure conditions, including: optimal exposure dose and optimal focal length; a second screening module, used to define the physical experimental area of ​​the virtual process window according to the second screening conditions, wherein the second screening conditions include: the fluctuation range of the exposure dose is the optimal exposure dose ± dose offset, and the fluctuation range of the focal length is the optimal focal length ± focal length offset; a third screening module, used to select a portion of the virtual exposure conditions in the physical experimental area as an experimental condition set based on the third screening conditions, wherein the third screening condition is that the number of virtual exposure conditions selected in the central area of ​​the physical experimental area is less than the number of virtual exposure conditions selected in the edge area; a physical experimental interface module, used to output the virtual exposure conditions specified by the experimental condition set for physical focus exposure experiments, and determine the actual exposure dose and actual focal length of the final lithography process window according to the experimental results.

[0014] This disclosure also provides a terminal, including a memory and a processor, wherein the memory stores a computer program that can run on the processor, and the processor executes the steps of any of the above-described methods for determining a photolithography process window when running the computer program.

[0015] This disclosure also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of any of the above-described methods for determining a photolithography process window; and / or, a computer program product comprising a computer program / instructions that, when executed by a processor, implement the steps of any of the above-described methods for determining a photolithography process window.

[0016] Compared with the prior art, the technical solution of the present disclosure has the following advantages: The method, apparatus, terminal, and computer program product for determining a photolithography process window provided in this disclosure include: determining design layout data, mask parameters, and a preset process parameter range; determining multiple virtual exposure conditions based on the preset process parameter range, and performing virtual focus exposure matrix simulation to generate simulated exposure results; obtaining a virtual process window formed by the virtual exposure conditions based on the simulated exposure results and using a first screening condition; determining the center of the virtual process window as the optimal virtual exposure condition; defining a physical experimental area within the virtual process window based on a second screening condition; selecting a portion of the physical experimental area as a set of experimental conditions based on a third screening condition; using the virtual exposure conditions specified by the set of experimental conditions to conduct a physical focus exposure experiment, and determining the actual exposure dose and actual focal length based on the experimental results. By first replacing numerous physical experiments with computer simulations, the reliance on physical silicon wafers and exposure machine time is reduced. Then, an initial screening criterion eliminates unqualified virtual exposure conditions, narrowing the range of virtual exposure conditions. Next, a second screening criterion delineates the physical experiment area near the boundary, further reducing the number of virtual exposure conditions requiring physical verification. Finally, a third screening criterion selects more virtual exposure conditions in the edge area and fewer in the central area. Using fewer virtual exposure conditions for physical experiments yields more accurate exposure dose and focal length inflection points, thus determining the size of the process window. Therefore, this method reduces the cost and improves the efficiency of determining the lithography process window. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments disclosed in this specification, the drawings used in the description of the embodiments disclosed in this specification or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this specification. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a schematic diagram of the result of a physical focus exposure matrix; Figure 2 This is a flowchart illustrating a method for determining a photolithography process window according to an embodiment of this disclosure; Figure 3 This is a flowchart illustrating the training process of a machine learning model according to an embodiment of this disclosure. Figure 4 This is a schematic diagram of the structure of a device for determining a photolithography process window according to an embodiment of the present disclosure; Figure 5 This is a schematic diagram of the hardware structure of a device for determining a photolithography process window according to an embodiment of this disclosure. Detailed Implementation

[0019] The technical solutions described herein will be described in detail below with reference to specific embodiments and accompanying drawings. The embodiments described herein are specific implementations of this disclosure and are used to illustrate the concept of this disclosure. These descriptions are illustrative and exemplary and should not be construed as limiting the implementation methods or the scope of protection of this disclosure. In addition to the embodiments described herein, those skilled in the art can employ other obvious technical solutions based on the content disclosed in the claims and specification of this application. These technical solutions include those that make any obvious substitutions and modifications to the embodiments described herein.

[0020] It should be noted that the accompanying drawings in this embodiment are schematic diagrams used to illustrate the concept of this disclosure, and to schematically show the shape and interrelationship of each part. It should be understood that, in order to clearly show the structure of each component of this disclosure, the drawings are not drawn to the same scale, and the same reference numerals are used to indicate the same parts in the drawings.

[0021] As described in the background section, determining the lithography process window is a crucial step in ensuring chip yield and performance stability during semiconductor manufacturing. Traditionally, the industry widely employs the focus-exposure matrix (FEM) method, which involves conducting numerous physical exposure experiments under different dose and focal length combinations to plot the range of process parameters corresponding to a qualified pattern.

[0022] However, a large number of physical exposure experiments require multiple wafers and several weeks of equipment time, resulting in high resource and time costs.

[0023] See Figure 1 , Figure 1 This is a schematic diagram of the result of a physical focus exposure matrix.

[0024] In traditional physical focus exposure matrix experiments, a large number of experimental points need to be uniformly selected within a wide range of exposure dose and focal length. A significant portion of these experimental conditions (such as areas far from the center of the window) are simply ineffective or redundant, failing to produce acceptable patterns. These redundant experiments not only waste valuable equipment time and wafer resources but also require multiple wafers, multiple exposure cycles, development, and measurements, making the entire process cumbersome and inefficient. Despite the large number of experimental points, the actual density of valuable data is very low, making the determination of the process window both time-consuming and labor-intensive.

[0025] Specifically, traditional methods suffer from the following significant drawbacks: First, they consume enormous resources, heavily relying on physical silicon wafers and expensive scanning exposure machines. Multiple wafers, exposures, and measurements are typically required to cover the complete exposure dose and focal length combination, resulting in high costs. Second, the experimental cycle is lengthy. From experimental design, condition selection, exposure and development to data measurement and analysis, the entire process is cumbersome and complex, severely slowing down the pace of process development. Third, experimental design is often unpredictable. The selection range and step size of exposure dose and focal length rely heavily on operator experience, resulting in many experimental points falling outside the process window and becoming invalid data. Even points within the window are often redundant and repetitive, leading to low data value density. Fourth, they are slow to respond to changes. Once inputs such as mask critical dimensions, photoresist thickness, or baking conditions change, a complete physical focus exposure matrix experiment must be repeated, lacking rapid response capability. These drawbacks make traditional methods not only inefficient but also ill-suited to the demands of rapid process iteration.

[0026] Therefore, how to provide technical solutions to reduce the cost of determining the lithography process window and improve the efficiency of determining the lithography process window has become an urgent technical problem to be solved.

[0027] The method, apparatus, terminal, and computer program product for determining a photolithography process window provided in this disclosure include: determining design layout data, mask parameters, and a preset process parameter range; determining multiple virtual exposure conditions based on the preset process parameter range, and performing virtual focus exposure matrix simulation to generate simulated exposure results; obtaining a virtual process window formed by the virtual exposure conditions based on the simulated exposure results and using a first screening condition; determining the center of the virtual process window as the optimal virtual exposure condition; defining a physical experimental area within the virtual process window based on a second screening condition; selecting a portion of the physical experimental area as a set of experimental conditions based on a third screening condition; using the virtual exposure conditions specified by the set of experimental conditions to conduct a physical focus exposure experiment, and determining the actual exposure dose and actual focal length based on the experimental results. By first replacing numerous physical experiments with computer simulations, the reliance on physical silicon wafers and exposure machine time is reduced. Then, an initial screening criterion eliminates unqualified virtual exposure conditions, narrowing the range of virtual exposure conditions. Next, a second screening criterion delineates the physical experiment area near the boundary, further reducing the number of virtual exposure conditions requiring physical verification. Finally, a third screening criterion selects more virtual exposure conditions in the edge area and fewer in the central area. Using fewer virtual exposure conditions for physical experiments yields more accurate exposure dose and focal length inflection points, thus determining the size of the process window. Therefore, this method reduces the cost and improves the efficiency of determining the lithography process window.

[0028] To enable those skilled in the art to better understand and implement the embodiments of this disclosure, the concepts, schemes, principles, and advantages of the embodiments of this disclosure are described in detail below with reference to the accompanying drawings and through specific application examples.

[0029] See Figure 2 , Figure 2 This is a schematic flowchart of a method for determining a photolithography process window according to an embodiment of this disclosure. The method may perform the following steps S11 to S16.

[0030] Step S11: Determine the design layout data, mask parameters, and preset process parameter range. The design layout data includes at least one of the design graphic, the key dimensions of the design graphic, and the edge placement error of the design graphic. The mask parameters include at least one of the graphic to be exposed, the key dimensions of the graphic to be exposed, and the edge placement error of the graphic to be exposed. The preset process parameter range includes a preset exposure dose range and a preset focal length range.

[0031] Step S12: Determine multiple virtual exposure conditions according to the preset process parameter range, and perform virtual focus exposure matrix simulation to generate simulation exposure results under multiple virtual exposure conditions.

[0032] Step S13: Based on the simulated exposure results, the first screening condition is used to determine the passability, and a virtual process window is obtained formed by the screened virtual exposure conditions. The first screening condition includes at least one of the following: critical size tolerance, edge placement error tolerance, and contour integrity.

[0033] Step S14: Determine the virtual exposure conditions at the center of the virtual process window as the optimal virtual exposure conditions, including: optimal exposure dose and optimal focal length.

[0034] Step S15: Within the virtual process window, define the physical experimental area of ​​the virtual process window according to the second screening conditions, wherein the second screening conditions include: the fluctuation range of the exposure dose is the optimal exposure dose ± dose offset, and the fluctuation range of the focal length is the optimal focal length ± focal length offset.

[0035] Step S16: Based on the third screening condition, select a portion of the virtual exposure conditions in the physical experimental area as the experimental condition set, wherein the third screening condition is that the number of virtual exposure conditions selected in the central area of ​​the physical experimental area is less than the number of virtual exposure conditions selected in the edge area. The virtual exposure conditions specified in the experimental condition set are used to conduct physical focus exposure experiments, and the actual exposure dose and actual focal length of the final photolithography process window are determined based on the experimental results.

[0036] Understandably, the method for determining the photolithography process window may include more or fewer steps, and the order of the steps may be the same or different.

[0037] In step S11, the design layout data, mask parameters, and preset process parameter range are determined.

[0038] The design layout data includes at least one of the following: design graphic, key dimensions of the design graphic, and edge placement error of the design graphic.

[0039] Specifically, the design layout data may include design graphics, such as dense line arrays, isolated feature (ISO) structures, and sensitive graphics such as corners / ends. The critical dimensions (CD) of the design graphics are extracted from these graphics.

[0040] In this embodiment, taking the polysilicon gate layer photolithography process of a certain 55nm node logic device as an example, the key dimension (such as gate width) can be 55nm.

[0041] In some embodiments, the design layout data further includes requirements for the edge placement error (EPE) of the design graphic. The edge placement error does not refer to the error of the design graphic, but rather to the deviation between the edge of the actual graphic or the edge of the simulated graphic after manufacturing and the edge of the design graphic.

[0042] In this embodiment, the edge placement error tolerance of the design pattern is ±1nm.

[0043] The mask parameters include at least one of the following: the image to be exposed, the key dimensions of the image to be exposed, and the edge placement error of the image to be exposed.

[0044] Specifically, the mask parameters include the pattern to be exposed, which basically corresponds to the design pattern of the design layout. However, due to mask manufacturing deviations or optical proximity effects, the actual pattern to be exposed is slightly offset. The critical dimensions of the pattern to be exposed on the mask basically correspond to the critical dimensions of the design pattern. However, due to mask manufacturing deviations or optical proximity effects, the actual pattern to be exposed is slightly offset (e.g., offset by 1nm).

[0045] In some embodiments, the mask parameters include an edge placement error requirement of ±1nm for the pattern to be exposed.

[0046] The preset process parameter range includes a preset exposure dose range and a preset focal length range.

[0047] Specifically, the preset process parameter range includes: a preset exposure dose range and a preset focal length range. The preset exposure dose range can be input by the operator based on process experience. The preset exposure dose range can also be input externally, such as from an external file or system. The preset exposure dose range can be a range of ±7mJ, using the optimal exposure dose predicted by machine learning as the median.

[0048] The preset focal length range is -120nm to +120nm.

[0049] The preset focal length range can be input by the operator based on their process experience.

[0050] The preset focal length range can be input externally, such as from an external file or system.

[0051] In some embodiments, the preset focal length range can be a range of ±30nm with the optimal focal length predicted by machine learning as the median.

[0052] The preset exposure dose range and preset focal length range will be used to generate multiple virtual exposure conditions in subsequent virtual simulations.

[0053] In step S12, multiple virtual exposure conditions are determined according to the preset process parameter range, and a virtual focus exposure matrix simulation is performed to generate simulated exposure results under multiple virtual exposure conditions.

[0054] In a traditional physical focus exposure matrix (PSME) experiment, taking a typical example, the exposure focal length matrix contains 156 conditions. Each wafer can only handle a limited number of conditions, typically requiring three wafers to complete the entire experiment. Each exposure requires scanning the exposure machine time, followed by development, CD measurement, and other steps, totaling approximately 2.5 weeks. If critical dimensions or other conditions of the mask change, the entire process must be repeated from scratch, lacking rapid response capability.

[0055] In this embodiment, the dose step of the preset exposure dose range is selected from 0.1mJ to 0.5mJ; the focal length step corresponding to the preset focal length range is selected from 3nm to 10nm.

[0056] In one specific embodiment, taking a preset exposure dose range of 18mJ to 32mJ and a preset focal length range of -120nm to +120nm as an example, the dose step size of the preset exposure dose range can be 0.2mJ; the focal length step size corresponding to the preset focal length range can be 5nm. Therefore, the number of dose points obtained in the preset exposure dose range = (32-18) / 0.2+1 = 14 / 0.2+1 = 70+1 = 71; the number of focal length points obtained in the preset focal length range = (120-(-120)) / 5+1 = 240 / 5+1 = 48+1 = 49; thus, the number of virtual exposure conditions can be determined to be 3479 (71×49=3479).

[0057] Perform a virtual focus exposure matrix simulation.

[0058] Specifically, computational lithography software (such as Dr. LiTHO or Sentaurus Lithography) is used to input each (dose, focal length) combination one by one, run the lithography simulation model, and obtain the simulated exposure results under various conditions. The simulation results include the three-dimensional photoresist profile on the wafer, critical dimension (CD), edge placement error (EPE), and profile integrity indicators (broken lines, bridging, wavy lines, etc.).

[0059] The 3D photoresist outline visually displays the three-dimensional morphology of the photoresist after exposure and development, allowing observation of the steepness of the sidewalls, bottom residue, and top rounded corners. The critical dimension (CD) refers to the actual size of the line width or spacing of the pattern. By comparing it with the critical dimension of the designed pattern, it can be determined whether the specifications are met. Edge placement error (EPE) comprehensively reflects the offset of the edge of the pattern relative to the design position, including the total deviation caused by CD changes and overlay deviations. The outline integrity index is used to detect whether there are defects such as broken lines (line breaks in the middle), bridging (abnormal connection between adjacent lines), and wavy lines (edges showing periodic undulations). These defects directly affect device performance and yield.

[0060] The simulated exposure results can be saved in structured data format, which facilitates subsequent batch screening and statistical analysis.

[0061] In this embodiment, 3479 virtual exposure conditions were computed in parallel using a 20-core server, taking a total of approximately 8 hours.

[0062] In comparison, traditional physical focus exposure matrix experiments only test 156 focus exposure conditions, while this approach generates 3479 virtual exposure conditions, which is more than 22 times the amount of data of the traditional method. Moreover, the step size of exposure dose and focal length is smaller, so the final process window is more accurate. Traditional physical focus exposure matrix experiments require 3 wafers and a time cycle of 2.5 weeks, while this approach only requires one wafer for subsequent verification, thus achieving the goal of significantly reducing costs and shortening the cycle.

[0063] The method further includes: using a machine learning model to analyze the design layout data, mask parameters, preset process parameter ranges, and the simulated exposure results, and outputting a predicted range of virtual exposure conditions. Through this machine learning model, after performing a complete virtual focus exposure matrix simulation, the selection range and density of virtual exposure conditions can be further optimized, thereby improving overall screening efficiency.

[0064] See Figure 3 , Figure 3 This is a flowchart illustrating the training process of a machine learning model according to an embodiment of this disclosure. The training process of the machine learning model can be performed by executing the following steps S21 to S23.

[0065] In step S21, training data is acquired, which includes: design layout sample data, mask sample parameters, sample process parameter range, and sample physical exposure pattern.

[0066] The design layout sample data is taken from typical circuit layouts at known process nodes (e.g., 55nm, 28nm, or 14nm nodes), including pattern structures (such as gate layers, metal interconnect layers, or contact via layers) with different pattern densities (dense regions, isolated regions, semi-dense regions) and critical dimensions (CD). The data format is usually GDS or OASIS, from which characteristic parameters such as pattern perimeter, area, curvature, and proximity effect intensity can be extracted.

[0067] The mask sample parameters include the critical dimension (CD) of the pattern to be exposed, the edge placement error (EPE) requirement, and the pattern outline after optical proximity correction (OPC) corresponding to the design pattern sample data.

[0068] The sample process parameters cover the actual exposure dose range (e.g., 18mJ~32mJ, step size 0.2mJ~0.5mJ) and focal length range (e.g., -120nm~+120nm, step size 3nm~10nm) used in historical physical focus exposure matrix (FEM) experiments, as well as process conditions such as light source wavelength, numerical aperture, photoresist type and thickness, baking temperature, etc.

[0069] The physical exposure pattern of the sample refers to the photoresist profile measurement results obtained by scanning electron microscope (SEM) or optical measurement equipment in physical FEM experiments. It includes the critical dimension (CD), edge placement error (EPE), and the type of defect in profile integrity (such as broken lines, bridging, wavy lines, necking, footing, etc.) under various conditions, which truly reflects the pattern quality under the input conditions.

[0070] It should be noted that the design layout sample data, mask sample parameters, sample process parameter ranges, and sample physical exposure patterns are interconnected. A one-to-one correspondence can be established using experimental batch numbers, wafer numbers, or condition numbers to ensure that the input features and output labels of each training sample accurately match, providing a sufficient and reliable training foundation for subsequent machine learning models.

[0071] In step S22, the training data is used as input features, and the corresponding sample exposure dose and sample focal length are used as output features to construct a training sample set.

[0072] Specifically, the input features include: one or more graphic features extracted from the design layout sample data and mask sample parameters, such as line width density, area ratio of isolated graphics to dense graphics, number of graphic corners, and perimeter-to-area ratio of graphics; and one or more process features extracted from the sample process parameter range, such as exposure dose range, dose step size, focal length range, and focal length step size. The output features refer to the optimal exposure dose and optimal focal length corresponding to the sample, which are defined as the optimal exposure dose and optimal focal length combination that simultaneously meets the requirements for critical dimension (CD) and edge placement error (EPE) in the sample physical exposure pattern.

[0073] For each set of training samples, the corresponding input feature is paired with the output feature to form a training sample. Multiple training samples together constitute the training sample set, which is used for the subsequent learning and training of the machine learning model.

[0074] In step S23, the machine learning model is trained using the training sample set.

[0075] During training, the machine learning model learns the mapping relationship between input features (design layout, mask parameters, and process parameter range) and output features (optimal exposure dose and optimal focal length). After sufficient training, when new design layout data, mask parameters, and preset process parameter ranges are input into the model, the model can quickly predict the range of predicted virtual exposure conditions under the new conditions. For example, the range centered on the optimal exposure dose and offset by a certain dose (e.g., ±7mJ) above and below it, and the range centered on the optimal focal length and offset by a certain focal length (e.g., ±30nm) above and below it.

[0076] The machine learning model is selected from one of random forest, support vector regression, neural network, or Gaussian process regression. Random forest has good anti-overfitting ability and the ability to handle high-dimensional features; support vector regression is suitable for small sample nonlinear regression; neural network can fit highly complex mapping relationships; and Gaussian process regression can provide the uncertainty range of prediction. Those skilled in the art can flexibly choose according to the amount of training data, computing resources, and accuracy requirements.

[0077] In some embodiments, the predicted virtual exposure condition range output by the machine learning model is used as the optimization input for the preset process parameter range in step S12. That is, the predicted range can be used to replace or narrow the initial input range, thereby reducing the total number of conditions in the virtual focus exposure matrix simulation. For example, if the model predicts an optimal dose of 50mJ and an optimal focal length of 0nm, the preset dose range is set to 43mJ~57mJ (±7mJ), and the preset focal length range is set to -30nm~+30nm (±30nm), so that the subsequent virtual FEM simulation focuses more on the high-probability window region, further improving the simulation efficiency.

[0078] In this embodiment of the disclosure, the machine learning model can automatically optimize the selection range of virtual exposure conditions using historical experimental data, reduce invalid or redundant simulation calculations, and further reduce the time cost and computing resource consumption for determining the lithography process window.

[0079] See also Figure 2 In step S13, based on the simulated exposure results, the first screening condition is used to determine the passability, and a virtual process window is obtained formed by the screened virtual exposure conditions.

[0080] The first screening criteria include at least one of the following: critical size tolerance, edge placement error tolerance, and contour integrity.

[0081] In this embodiment, it is preferable to use at least one of the following three criteria for comprehensive judgment: critical size tolerance, edge placement error tolerance, and contour integrity.

[0082] The acceptable range for the critical dimensions is an error of less than or equal to ±1nm, meaning that the deviation of the line width or spacing of the simulated graphic from the critical dimensions of the design graphic does not exceed 1nm.

[0083] The acceptable range for edge placement error is less than or equal to ±1nm, meaning the positional deviation of the simulated graphic edge relative to the designed graphic edge does not exceed 1nm.

[0084] The outline integrity includes: no broken lines (i.e., continuous and unbroken graphic lines), no bridging (i.e., no abnormal connections between adjacent lines), no wavy defects (i.e., no periodic undulations at the edges of the graphic), and end indentation of less than or equal to 2nm (i.e., the amount of retraction of the line end relative to the design position does not exceed 2nm).

[0085] In some embodiments, the first screening condition is used to determine the simulated exposure results under each virtual exposure condition within the range of predicted virtual exposure conditions, and retain the virtual exposure conditions that meet the first screening condition, thereby obtaining a virtual process window composed of the screened virtual exposure conditions.

[0086] In some embodiments, the design layout data includes ISO-type isolated pattern structures (e.g., isolated lines or isolated spaces). Since ISO-type isolated patterns are most sensitive to exposure dose and focal length fluctuations, their pattern quality often determines the process window of the weakest point on the entire chip. Therefore, in the first screening criterion, the contour integrity also includes the detection of necking defects and footing defects in the ISO-type isolated pattern structures.

[0087] The necking defect refers to a localized narrowing in the middle of an isolated line, forming a "bottleneck" shape; the footing defect refers to an abnormal expansion of the bottom of an isolated line to both sides, forming a "foot"-like protrusion. Both of these defects significantly affect the channel length control and leakage characteristics of transistors, and are therefore explicitly included as non-compliance criteria.

[0088] In some embodiments, for each virtual exposure condition, if the simulated exposure result simultaneously meets all the above-mentioned CD, EPE, and contour integrity requirements (including necking and pedimentation detection of ISO graphics), it can be determined as qualified. The set of all qualified virtual exposure conditions constitutes the virtual process window.

[0089] In some embodiments, the first screening condition is used not only to determine all the simulated exposure results generated in step S12, but also to determine the pass / fail status of the simulated exposure results within the range of predicted virtual exposure conditions output by the machine learning model.

[0090] Specifically, before step S12 or step S13, the machine learning model can be used to analyze the design layout data, mask parameters, and preset process parameter ranges to output a predicted virtual exposure condition range (e.g., a narrowed dose range centered on the predicted optimal dose). Then, virtual focus exposure matrix simulation is performed only on the virtual exposure conditions within this predicted range, and the simulation exposure results are input into step S13 for the qualification determination of the first screening condition. Since the predicted range has already excluded a large number of obviously unqualified areas, the simulation computation load can be further reduced without affecting the integrity of the virtual process window.

[0091] In one specific embodiment, a set of specific values ​​is used as an example. The preset process parameter range is: exposure dose 18mJ~32mJ (step 0.2mJ), focal length -120nm~+120nm (step 5nm), generating a total of 3479 virtual exposure conditions. After performing virtual FEM simulation, the simulation exposure results for each condition are subject to the first screening criteria: CD error ≤ ±1nm, EPE ≤ ±1nm, no broken lines, no bridging, no wavy defects in the contour, and end indentation ≤ 2nm; for ISO-type isolated graphics in the design layout, there are additional requirements for no necking and no footing defects. After evaluation, 156 virtual exposure conditions meet all the above requirements, and these conditions together constitute the virtual process window. The center position of this window (e.g., dose 25mJ, focal length 0nm) is determined as the optimal virtual exposure condition, including the optimal exposure dose and the optimal focal length.

[0092] In this embodiment, the criteria in the first screening condition can be dynamically adjusted according to specific process nodes and device requirements; conditions can be added, deleted, or modified. For example, for the more advanced 28nm node, the critical size tolerance may be tightened to ±0.5nm, and the edge placement error tolerance may be tightened to ±0.5nm; while for the mature 90nm node, the above tolerances can be relaxed to ±2nm. Similarly, the allowable size of defects such as end indentation, necking, and leading in contour integrity can also be set with different thresholds according to device performance requirements. Those skilled in the art will understand that the specific values ​​of the above criteria should not constitute a limitation on the scope of this disclosure.

[0093] By filtering according to the first screening criteria, especially for specific defect detection of ISO-type isolated patterns, the simulated exposure results corresponding to the virtual exposure conditions within the virtual process window are ensured to have high reliability in terms of image quality. The virtual process window, as the basic input for subsequent steps (second screening, third screening, and physical experiments), can significantly reduce the number of invalid physical experiments while ensuring the accuracy of the finally determined lithography process window.

[0094] In step S14, the virtual exposure conditions at the center of the virtual process window are determined as the optimal virtual exposure conditions, including: optimal exposure dose and optimal focal length.

[0095] The virtual process window is a set of qualified virtual exposure conditions obtained after screening in step S13. This set is presented as a continuous or nearly continuous region on the two-dimensional plane of exposure dose and focal length.

[0096] Specifically, by analyzing the simulation exposure results corresponding to each condition within the virtual process window (such as the values ​​of critical dimension CD, edge placement error EPE, and contour integrity index), the point with the best graphic quality is selected as the center of the window.

[0097] In some embodiments, the absolute value of the deviation between the CD value and the critical dimension of the design graphic under all qualified conditions within the virtual process window is calculated, and the point with the smallest deviation is taken as the center; if there are multiple virtual exposure conditions with the same deviation, the EPE values ​​are further compared, and the one with the smallest EPE is taken as the center. The virtual exposure condition corresponding to the center is the optimal virtual exposure condition, the corresponding exposure dose is the optimal exposure dose, and the corresponding focal length is the optimal focal length.

[0098] In some embodiments, when the virtual process window is irregularly shaped, the center of the window can be determined by the geometric center method.

[0099] Specifically, in the exposure dose-focal length plane coordinate system, the coordinates of the boundary points of the virtual process window are extracted, and the geometric centroid of the region enclosed by the boundary is calculated. The exposure dose and focal length values ​​corresponding to this centroid are taken as the optimal exposure dose and optimal focal length. If the centroid is not on an actual existing virtual exposure condition grid point, the qualified virtual exposure condition closest to the centroid is selected as the optimal virtual exposure condition. This method solves the problem of complex virtual process window shapes.

[0100] In some embodiments, the determination of the optimal virtual exposure conditions also incorporates the simulated exposure results of ISO-type isolated pattern structures. Since ISO-type isolated patterns are most sensitive to process variations, their pattern quality often deteriorates before other pattern structures. Therefore, within the virtual process window, the optimal virtual exposure conditions are selected that allow the critical dimensions of the ISO-type isolated pattern to most closely approximate the critical dimensions of the designed pattern, without necking or footing defects. This approach ensures that the optimal conditions can accommodate the process requirements of the most sensitive pattern structures, thereby improving the robustness of the final process window.

[0101] In one specific embodiment, a set of specific values ​​is used as an example for illustration. In step S13, 156 qualified conditions are selected from 3479 virtual exposure conditions, forming a virtual process window. All virtual exposure conditions and their corresponding simulated exposure results within the virtual process window are analyzed: on the dose coordinate, the dose range of the qualified conditions is approximately 23mJ to 27mJ; on the focal length coordinate, the focal length range of the qualified conditions is approximately -30nm to +25nm. The deviation of the CD value under each qualified condition from the critical dimension 55nm of the design pattern is calculated. It is found that when the dose is 25mJ and the focal length is 0nm, the CD deviation is -0.2nm, the EPE is 0.3nm, and the contour integrity is optimal (no defects). Therefore, the optimal exposure dose is 25mJ, and the optimal focal length is 0nm.

[0102] It should be noted that the virtual exposure conditions for determining the center of the virtual process window can also be achieved using methods such as curve fitting or taking median values. Curve fitting involves performing surface fitting on the CD or EPE data for each condition within the window, and then taking the dose and focal length corresponding to the extreme points. Taking median values ​​directly calculates the average of the maximum and minimum dose values ​​and the average of the maximum and minimum focal length values ​​for the acceptable conditions. This disclosure does not impose any restrictions on the specific determination method.

[0103] In step S15, the physical experimental area of ​​the virtual process window is defined according to the second screening criteria.

[0104] The second screening criteria include: the fluctuation range of the exposure dose is the optimal exposure dose ± dose offset, and the fluctuation range of the focal length is the optimal focal length ± focal length offset.

[0105] The dose offset and focal length offset can be preset according to specific process nodes and device requirements.

[0106] In this embodiment, the dose offset is 1 mJ; the focal length offset is 30 nm.

[0107] In one specific embodiment, taking the optimal exposure dose of 25mJ and the optimal focal length of 0nm determined in step S14 as an example, the dose offset is set to ±1mJ and the focal length offset to ±30nm. Then, the physical experimental area defined by the second screening condition is a rectangular area with a dose between 24mJ and 26mJ and a focal length between -30nm and +30nm. The virtual exposure conditions within the physical experimental area are located inside the virtual process window, representing the most reliable area that can still stably produce qualified patterns under normal process fluctuations. In other words, the virtual exposure conditions within the physical experimental area are virtual exposure conditions within the virtual process window with a dose between 24mJ and 26mJ and a focal length between -30nm and +30nm.

[0108] In some embodiments, the dose offset and focal length offset are not fixed values, but are adaptively determined based on the shape of the virtual process window. For example, the total width of the virtual process window in the dose direction (maximum dose minus minimum dose) is calculated, and 20% of this width is taken as the dose offset; similarly, 30% of the total width in the focal length direction is taken as the focal length offset.

[0109] In step S16, based on the third screening condition, a portion of the virtual exposure conditions in the physical experimental area is selected as the experimental condition set.

[0110] The third screening criterion is that the number of virtual exposure conditions selected in the central region of the physical experimental area is less than the number selected in the edge region. In other words, denser sampling is performed near the boundary of the physical experimental area, while sparser sampling is performed in the central region near the optimal exposure dose and optimal focal length.

[0111] The number of virtual exposure conditions in the experimental condition set is less than or equal to 1 / 8 of the number of virtual exposure conditions (initial virtual simulation conditions) generated when performing virtual focus exposure matrix simulation. In other words, after three layers of screening, the number of virtual exposure conditions finally used for physical verification does not exceed one-eighth of the total number of virtual exposure conditions determined in step S12.

[0112] In some embodiments, the third screening condition includes: the number of virtual exposure conditions in the experimental condition set is less than or equal to 10; and / or, the ratio of the number of virtual exposure conditions selected in the edge region to the number of virtual exposure conditions selected in the center region of the physical experimental area is not less than 3:1.

[0113] Taking a specific set of values ​​as an example. In step S12, the virtual focus exposure matrix simulation generates a total of 3479 virtual exposure conditions. After screening by the first screening condition in step S13, a virtual process window consisting of 156 qualified conditions is obtained; after defining by the second screening condition in step S15, a physical experimental area is obtained, which contains several virtual exposure conditions (e.g., 88). In step S16, according to the third screening condition, an experimental condition set is selected from these 88 conditions. The total number of experimental conditions is set to 8 (less than or equal to 10), of which 6 points are selected in the edge area and 2 points are selected in the center area, with a ratio of 3:1 between the edge and center areas. These 8 points constitute the experimental condition set. This number 8 is less than or equal to 1 / 8 of 3479 (i.e., 434.875), satisfying the aforementioned ratio requirement.

[0114] In some embodiments, the central and edge regions of the physical experimental area are divided as follows: a central rectangular area is defined with the optimal exposure dose and optimal focal length as the center point, and 50% of the dose offset and 50% of the focal length offset as the semi-axis; the portion of the physical experimental area excluding the central area is defined as the edge region. For example, if the optimal dose is 25 mJ, the optimal focal length is 0 nm, the dose offset is ±1 mJ, and the focal length offset is ±30 nm, then the central region is a rectangle with a dose of 24.5 mJ to 25.5 mJ and a focal length of -15 nm to +15 nm; the edge region is the portion of the physical experimental area (dose 24 mJ to 26 mJ, focal length -30 nm to +30 nm) minus the central region.

[0115] In some embodiments, the specific sampling points within the edge region are not randomly selected, but are determined based on the boundary curvature changes of the virtual process window.

[0116] Specifically, the curvature of each point on the boundary of the physical experimental area is calculated, and virtual exposure conditions are preferentially selected at locations with greater curvature (i.e., drastic changes in window shape); the number of sampling points is appropriately reduced in areas with relatively flat boundaries. Within the central region, only the point corresponding to the optimal exposure dose and optimal focal length, along with 1-2 neighboring points, are selected to confirm the stability of the central region and maximize boundary recognition accuracy.

[0117] It should be noted that having 10 or fewer virtual exposure conditions in the experimental condition set is a preferred range, but not an absolute limitation. For certain process nodes with larger physical experimental areas or higher precision requirements, the number of experimental conditions can be appropriately increased to 15 or 20, but should still be less than or equal to 1 / 8 of the initial total number of virtual simulation conditions. For example, when the total number of virtual simulation conditions is 5000, the maximum allowed number of experimental conditions is 625, but in practice it is still recommended to control it within 10-20, thereby significantly reducing wafer consumption, equipment time, and data analysis costs of physical experiments, achieving efficient determination of the lithography process window, and maintaining the economic efficiency of physical experiments.

[0118] The virtual exposure conditions specified by the set of experimental conditions selected in step S16 are used to conduct physical focus exposure experiments.

[0119] Specifically, each virtual exposure condition in the set of experimental conditions is used as the input parameter for the physical experiment, and standard photolithography processes such as coating, exposure, and development are sequentially performed on the same wafer. The exposure equipment can be a scanning exposure machine (e.g., an ArF immersion lithography machine). After development, a scanning electron microscope (SEM) or an optical linewidth meter is used to measure the pattern on the wafer to obtain the actual critical dimension (CD), edge placement error (EPE), and pattern outline integrity information (including defects such as broken lines, bridging, wavy lines, necking, and footing) under each experimental condition.

[0120] Based on the results of physical experiments, the actual exposure dose and actual focal length of the final photolithography process window are determined.

[0121] Specifically, within the dose-focal length range covered by the set of physical experimental conditions, physical experimental points that simultaneously meet the aforementioned qualification requirements (e.g., CD error ≤ 1nm, EPE ≤ 1nm, no significant defects) are marked as "qualified". If multiple consecutive points are qualified, the area enclosed by these points is the final process window. The exposure dose and focal length corresponding to the point with the best pattern quality within this window (e.g., minimum CD deviation, minimum EPE, and no defects) are taken as the actual optimal exposure dose and actual optimal focal length.

[0122] In this embodiment, the physical focus exposure experiment can use only one wafer. Since the number of conditions in the experimental condition set is usually no more than 10 (e.g., 8), different doses and focal lengths can be set on different exposure areas (shots) of the same wafer to complete all exposures at once.

[0123] When determining the final process window, if some points in the edge region of the physical experiment condition set (i.e., the boundary points selected according to the third screening condition) show as unqualified in physical measurements, but are qualified in the virtual simulation, it indicates a deviation in the simulation model. In this case, the boundary of the virtual process window can be corrected using interpolation methods: for example, the midpoint between the unqualified point measured in the physical experiment and the adjacent qualified point can be used as the actual boundary position to obtain a more accurate actual process window. Conversely, if all points in the physical experiment are consistent with the virtual simulation results (qualified / unqualified matching degree ≥ 95%), the area of ​​the virtual process window verified by the physical experiment points can be directly used as the final process window, and the center point measured in the physical experiment can be used as the actual optimal exposure dose and the actual optimal focal length.

[0124] Therefore, the method can reduce the cost of determining the lithography process window and improve the efficiency of determining the lithography process window.

[0125] Accordingly, this disclosure also provides an apparatus for determining the photolithography process window.

[0126] See Figure 4 , Figure 4 This is a schematic diagram of a device for determining a photolithography process window according to an embodiment of the present disclosure. The device for determining the photolithography process window may include: The data input module 31 is used to determine the design layout data, mask parameters and preset process parameter range. The design layout data includes at least one of the design graphic, the key dimensions of the design graphic and the edge placement error of the design graphic. The mask parameters include at least one of the graphic to be exposed, the key dimensions of the graphic to be exposed and the edge placement error of the graphic to be exposed. The preset process parameter range includes a preset exposure dose range and a preset focal length range. The virtual simulation module 32 is used to determine multiple virtual exposure conditions according to the preset process parameter range, and to perform virtual focus exposure matrix simulation to generate simulation exposure results under multiple virtual exposure conditions. The first screening module 33 is used to determine the passability of the virtual process window based on the simulated exposure results and the first screening conditions, thereby obtaining a virtual process window formed by the screened virtual exposure conditions. The first screening conditions include at least one of the following: critical size tolerance, edge placement error tolerance, and contour integrity. The virtual exposure conditions at the center of the virtual process window are determined as the optimal virtual exposure conditions, including the optimal exposure dose and the optimal focal length. The second screening module 34 is used to define the physical experimental area of ​​the virtual process window according to the second screening conditions within the virtual process window. The second screening conditions include: the fluctuation range of the exposure dose is the optimal exposure dose ± dose offset, and the fluctuation range of the focal length is the optimal focal length ± focal length offset. The third filtering module 35 is used to select a portion of the virtual exposure conditions in the physical experimental area as a set of experimental conditions based on the third filtering condition, wherein the third filtering condition is that the number of virtual exposure conditions selected in the central area of ​​the physical experimental area is less than the number of virtual exposure conditions selected in the edge area. The physical experiment interface module 36 is used to output the virtual exposure conditions specified by the set of experimental conditions for conducting physical focus exposure experiments, and to determine the actual exposure dose and actual focal length of the final photolithography process window based on the experimental results.

[0127] This application also provides a computer-readable storage medium storing a computer program thereon. When the computer program is run by a computer, the above-described method is executed. The storage medium may include read-only memory (ROM), random access memory (RAM), a magnetic disk, or an optical disk, etc. The storage medium may also include non-volatile memory or non-transitory memory, etc.

[0128] This application embodiment also provides a terminal, including a memory and a processor, wherein the memory stores a computer program that can run on the processor, and the processor executes the steps of the method for determining the photolithography process window described above when running the computer program.

[0129] See Figure 5 , Figure 5 This is a schematic diagram of the hardware structure of a device for determining a photolithography process window according to an embodiment of this disclosure.

[0130] Figure 5 The terminal shown includes a memory 41, a processor 42, and a transceiver 43. The processor 42 is coupled to the memory 41 and the transceiver 43. The memory 41 can be located inside or outside the terminal. The memory 41, processor 42, and transceiver 43 can be connected via a communication bus. The transceiver 43 is used to communicate with other devices or communication networks.

[0131] This application also provides a computer program product, including a computer program / instructions, which, when executed by a processor, implement the steps of the method for determining the photolithography process window described above.

[0132] It should be understood that in the embodiments of this application, the processor can be a central processing unit (CPU), or it can be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor.

[0133] It should also be understood that the memory in the embodiments of this application can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory can be ROM, programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous linked dynamic random access memory (SLDRAM), and direct rambus RAM (DR RAM).

[0134] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer program are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer program can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another. For example, the computer program can be transferred from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless means.

[0135] While the embodiments disclosed above are described in this application, this application is not limited thereto. Any person skilled in the art can make various modifications and alterations without departing from the spirit and scope of this application; therefore, the scope of protection of this application should be determined by the scope defined in the claims.

[0136] The foregoing describes several embodiments of the method for determining the photolithography process window. The various optional methods described in each embodiment can be combined and cross-referenced without conflict, thereby extending to a variety of possible embodiments. These can all be considered as embodiments disclosed in this disclosure.

Claims

1. A method of determining a lithography process window, characterized by, include: The design layout data, mask parameters, and preset process parameter range are determined. The design layout data includes at least one of the design graphic, the key dimensions of the design graphic, and the edge placement error of the design graphic. The mask parameters include at least one of the graphic to be exposed, the key dimensions of the graphic to be exposed, and the edge placement error of the graphic to be exposed. The preset process parameter range includes a preset exposure dose range and a preset focal length range. Multiple virtual exposure conditions are determined based on the preset process parameter range, and a virtual focus exposure matrix simulation is performed to generate simulated exposure results under multiple virtual exposure conditions. Based on the simulated exposure results, a first screening condition is used to determine the passability, resulting in a virtual process window formed by the screened virtual exposure conditions. The first screening condition includes at least one of the following: critical size tolerance, edge placement error tolerance, and contour integrity. The virtual exposure conditions at the center of the virtual process window are determined as the optimal virtual exposure conditions, including: optimal exposure dose and optimal focal length. Within the virtual process window, the physical experimental area of ​​the virtual process window is defined according to the second screening conditions, wherein the second screening conditions include: the fluctuation range of the exposure dose is the optimal exposure dose ± dose offset, and the fluctuation range of the focal length is the optimal focal length ± focal length offset. Based on the third screening condition, a portion of the virtual exposure conditions in the physical experimental area are selected as the experimental condition set, wherein the third screening condition is that the number of virtual exposure conditions selected in the central area of ​​the physical experimental area is less than the number of virtual exposure conditions selected in the edge area. The virtual exposure conditions specified in the experimental condition set are used to conduct physical focus exposure experiments, and the actual exposure dose and actual focal length of the final photolithography process window are determined based on the experimental results.

2. The method of claim 1, wherein, The preset exposure dose range and preset focal length range satisfy one or more of the following: The dose step size of the preset exposure dose range is selected from 0.1 mJ to 0.5 mJ; The preset focal length range is selected from -120nm to +120nm; The focal length step corresponding to the preset focal length range is selected from 3nm to 10nm.

3. The method of claim 1, wherein, The first screening criteria include one or more of the following: The acceptable range for the critical dimension is an error of less than or equal to 1 nm; The acceptable range for the edge placement error is less than or equal to 1 nm; The outline integrity includes: no broken lines, no bridging, no wavy defects, and end indentation of less than or equal to 2nm.

4. The method of claim 3, wherein, The design graphics contain an ISO-type isolated graphic structure; Among the first screening criteria, the contour integrity also includes the detection of necking defects and footing defects in the ISO-type isolated graphic structure.

5. The method of claim 1, wherein, Based on the third screening criterion, a subset of the virtual exposure conditions in the physical experimental area is selected as the experimental condition set, including: The number of virtual exposure conditions in the experimental condition set is less than or equal to 1 / 8 of the number of virtual exposure conditions generated by performing virtual focus exposure matrix simulation.

6. The method of claim 1, wherein, Before determining the pass / fail status using the first screening criteria based on the simulated exposure results, the process also includes: A machine learning model is used to analyze the design layout data, mask parameters, preset process parameter range and the simulation exposure results, and outputs a predicted virtual exposure condition range. The first screening condition is used to determine the passability of the simulation exposure results obtained from the predicted virtual exposure condition range. The training process of the machine learning model includes: Acquire training data, which includes: design layout sample data, mask sample parameters, sample process parameter range, and sample physical exposure pattern; The training data is used as input features, and the corresponding sample exposure dose and sample focal length are used as output features to construct a training sample set; The machine learning model is trained using the training sample set; The machine learning model is selected from one of random forest, support vector regression, neural network, or Gaussian process regression.

7. The method of claim 1, wherein, The third screening criteria include: The number of virtual exposure conditions in the experimental condition set is less than or equal to 10; And / or, the ratio of the number of virtual exposure conditions selected in the edge region to the number of virtual exposure conditions selected in the center region of the physical experimental area is not less than 3:

1.

8. An apparatus for determining a lithography process window, characterized in that, include: The data input module is used to determine the design layout data, mask parameters, and preset process parameter range. The design layout data includes at least one of the design graphic, the key dimensions of the design graphic, and the edge placement error of the design graphic. The mask parameters include at least one of the graphic to be exposed, the key dimensions of the graphic to be exposed, and the edge placement error of the graphic to be exposed. The preset process parameter range includes a preset exposure dose range and a preset focal length range. The virtual simulation module is used to determine multiple virtual exposure conditions based on the preset process parameter range, and to perform virtual focus exposure matrix simulation to generate simulation exposure results under multiple virtual exposure conditions. The first screening module is used to determine the passability of the virtual exposure based on the simulated exposure results using the first screening conditions, and to obtain a virtual process window formed by the screened virtual exposure conditions. The first screening conditions include at least one of the following: critical size tolerance, edge placement error tolerance, and contour integrity. The virtual exposure conditions at the center of the virtual process window are determined as the optimal virtual exposure conditions, including: optimal exposure dose and optimal focal length. The second screening module is used to define the physical experimental area of ​​the virtual process window according to the second screening conditions within the virtual process window. The second screening conditions include: the fluctuation range of the exposure dose is the optimal exposure dose ± dose offset, and the fluctuation range of the focal length is the optimal focal length ± focal length offset. The third filtering module is used to select a portion of the virtual exposure conditions in the physical experimental area as a set of experimental conditions based on the third filtering condition. The third filtering condition is that the number of virtual exposure conditions selected in the central area of ​​the physical experimental area is less than the number of virtual exposure conditions selected in the edge area. The physical experiment interface module is used to output the virtual exposure conditions specified by the set of experimental conditions for conducting physical focus exposure experiments, and to determine the actual exposure dose and actual focal length of the final photolithography process window based on the experimental results.

9. A terminal comprising a memory and a processor, said memory having stored thereon a computer program that is operable on said processor, characterized in that, When the processor runs the computer program, it performs the steps of the method for determining the photolithography process window according to any one of claims 1 to 7.

10. A computer-readable storage medium having stored thereon a computer program, characterized in that, When the program is executed by the processor, it implements the steps of the method for determining the photolithography process window according to any one of claims 1 to 7; And / or, a computer program product comprising a computer program / instructions, characterized in that, when executed by a processor, the computer program / instructions implement the steps of the method for determining a photolithography process window as claimed in any one of claims 1 to 7.

Images