Self-adaptive spraying control method and system for semiconductor wafer

Abstract

The invention provides a self-adaptive spraying control method and system for a semiconductor wafer, and relates to the technical field of spraying control, and the method comprises the steps: collecting all feature information of a to-be-sprayed wafer, and determining the expected coating information of a t-th node according to a spraying process chain; and constructing a registration space by registering spraying control features, and activating the optimization evaluation channel to perform optimization analysis to generate a spraying control scheme meeting the rule. And finally, in combination with the expected coating information and the spraying control scheme, a complete wafer spraying control scheme is formulated, and spraying control is carried out. According to the method and the device, the problem that the performance and the yield of the semiconductor device are further influenced due to the fact that the spraying precision and the coating uniformity are difficult to guarantee due to lack of self-adaptive adjustment capability in the prior art can be solved, and accurate spraying control is performed on the wafer to be sprayed according to the wafer spraying control scheme, so that the spraying quality and the spraying efficiency are improved.

Description

Technical Field

[0001] This application relates to the field of spray coating control technology, and in particular to an adaptive spray coating control method and system for semiconductor wafers. Background Technology

[0002] In semiconductor manufacturing processes, photoresist is a key material used to create patterns on integrated circuits. Photolithography is a core step in semiconductor manufacturing, involving coating the wafer surface with photoresist and projecting patterns onto the photoresist using ultraviolet light or laser irradiation to form the desired pattern. Traditional photoresist spraying equipment often uses a hose-like integrated nozzle structure. This structure makes it difficult to ensure that the photoresist is accurately sprayed onto the exact center of the wafer. Furthermore, due to the high viscosity of photoresist and the large volume of sprays required per pass, the nozzle structure is prone to contamination and is inconvenient to clean or replace, severely impacting the efficiency of photoresist spraying on wafers. With the development of semiconductor technology, the requirements for spraying efficiency and coating uniformity are becoming increasingly stringent. Especially in the central region of the wafer, where centrifugal force is weaker, traditional spin coating methods easily lead to uneven coating in the central area, thus affecting wafer yield and the performance of semiconductor devices.

[0003] Currently, existing semiconductor wafer coating technologies largely rely on traditional coating methods and equipment. While these methods are mature, they lack detailed consideration for different wafer characteristics and application scenarios. Coating parameters are often set based on experience rather than being dynamically adjusted according to the actual conditions of the wafer, thus proving inadequate when facing diverse and complex coating requirements.

[0004] In summary, the lack of adaptive adjustment capability in existing technologies makes it difficult to guarantee the accuracy and uniformity of spraying, which further affects the performance and yield of semiconductor devices. Summary of the Invention

[0005] The purpose of this application is to provide an adaptive spraying control method and system for semiconductor wafers, in order to solve the problem that the existing technology lacks adaptive adjustment capabilities, which makes it difficult to guarantee the accuracy and uniformity of spraying, and further affects the performance and yield of semiconductor devices.

[0006] In view of the above problems, this application provides an adaptive spraying control method and system for semiconductor wafers.

[0007] In a first aspect, this application provides an adaptive spraying control method for semiconductor wafers. The method is implemented through an adaptive spraying control system for semiconductor wafers. The method includes: collecting structural feature information, material feature information, and application scenario feature information of the wafer to be sprayed, generating a wafer feature data sequence; extracting the t-th node of the wafer spraying process chain according to the wafer spraying process chain, wherein the spraying process chain includes T wafer spraying nodes of the wafer to be sprayed, where t and T are both positive integers, and t ∈ T; and performing operations based on the t-th node of the wafer spraying process chain according to the wafer feature data sequence. The coating expectation is analyzed to determine the expected coating information for node t. Based on the expected coating information for node t and the wafer feature data sequence, spray control feature registration is performed to construct the spray control registration space for node t. The spray control optimization evaluation channel is activated, and optimization analysis is performed in conjunction with the spray control registration space for node t to generate a spray control scheme for node t that satisfies the spray control optimization rules. Based on the expected coating information for node t and the spray control scheme for node t, a wafer spray control scheme is generated. Based on the wafer spraying end, spray control is performed on the wafer to be sprayed according to the wafer spray control scheme.

[0008] Secondly, this application also provides an adaptive spraying control system for semiconductor wafers, used to execute the adaptive spraying control method for semiconductor wafers as described in the first aspect, wherein the system includes: a feature data sequence generation module, used to collect structural feature information, material feature information, and application scenario feature information of the wafer to be sprayed, and generate a wafer feature data sequence; a spraying node extraction module, used to extract the t-th node of wafer spraying according to the spraying process chain of the wafer to be sprayed, wherein the spraying process chain includes T wafer spraying nodes of the wafer to be sprayed, t and T are both positive integers, and t belongs to T; and an expectation parsing module, used to perform analysis based on the t-th node of wafer spraying and the wafer feature data sequence. The system comprises the following modules: a coating expectation analysis module to determine the expected coating information for node t; a feature registration module to perform spraying control feature registration based on the expected coating information for node t and the wafer feature data sequence to construct the spraying control registration space for node t; an optimization analysis module to activate the spraying control optimization evaluation channel and perform optimization analysis in conjunction with the spraying control registration space for node t to generate a spraying control scheme for node t that satisfies the spraying control optimization rules; a spraying scheme generation module to generate a wafer spraying control scheme based on the expected coating information for node t and the spraying control scheme for node t; and a spraying control module to perform spraying control on the wafer to be sprayed based on the wafer spraying end and the wafer spraying control scheme.

[0009] One or more technical solutions provided in this application have at least the following technical effects or advantages: By collecting structural, material, and application scenario information of the wafer to be coated, a wafer feature data sequence is generated. Based on the coating process chain of the wafer to be coated, the t-th node of the wafer coating is extracted, where the coating process chain includes T wafer coating nodes, t and T are positive integers, and t ∈ T. Based on the t-th node of the wafer coating, coating expectation analysis is performed according to the wafer feature data sequence to determine the expected coating information for the t-th node. Coating control feature registration is performed based on the expected coating information for the t-th node and the wafer feature data sequence to construct the coating control registration space for the t-th node. Finally, the coating control optimization evaluation is activated. The channel, combined with the coating control registration space of the t-th node, performs optimization analysis to generate a coating control scheme for the t-th node that satisfies the coating control optimization rules. Based on the desired coating information of the t-th node and the coating control scheme of the t-th node, a wafer coating control scheme is generated. Based on the wafer coating end, the wafer to be coated is coated according to the wafer coating control scheme. This effectively solves the problem that the lack of adaptive adjustment capability in the existing technology makes it difficult to guarantee the accuracy and uniformity of coating, which further affects the performance and yield of semiconductor devices. The wafer coating control scheme enables precise coating control of the wafer to be coated, thereby improving coating quality and efficiency.

[0010] The above description is merely an overview of the technical solution of this application. To better understand the technical means of this application and to facilitate its implementation according to the description, and to make the above and other objects, features, and advantages of this application more apparent, specific embodiments of this application are described below. It should be understood that the content described in this section is not intended to identify key or important features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent through the following description. Attached Figure Description

[0011] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0012] Figure 1 This is a schematic flowchart of the adaptive spraying control method for semiconductor wafers used in this application; Figure 2 This is a schematic diagram of the adaptive spraying control system for semiconductor wafers used in this application.

[0013] Explanation of reference numerals in the attached figures: Feature data sequence generation module 11, spraying node extraction module 12, expectation parsing module 13, feature registration module 14, optimization analysis module 15, spraying scheme generation module 16, spraying control module 17. Detailed Implementation

[0014] This application provides an adaptive spraying control method and system for semiconductor wafers, which solves the problem that the lack of adaptive adjustment capability in the prior art makes it difficult to guarantee the accuracy and uniformity of the spraying, which further affects the performance and yield of semiconductor devices. The method provides precise spraying control of the wafer to be sprayed according to the wafer spraying control scheme, thereby improving the spraying quality and efficiency.

[0015] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. It should be understood that this application is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application. It should also be noted that, for ease of description, only the parts related to this application are shown in the accompanying drawings, not all of them. Example 1

[0016] Please see the appendix Figure 1 This application provides an adaptive spraying control method for semiconductor wafers, wherein the method is applied to an adaptive spraying control system for semiconductor wafers, and the method specifically includes the following steps: S1: Collect structural feature information, material feature information, and application scenario feature information of the wafer to be coated, and generate a wafer feature data sequence.

[0017] Specifically, structural characteristic information includes the wafer's size, shape, and surface morphology. High-precision measuring instruments, such as optical microscopes, scanning electron microscopes, or atomic force microscopes, are used to precisely measure the wafer surface and obtain parameters such as dimensions, flatness, and roughness. Laser scanners or white light interferometers are used to perform three-dimensional morphology scanning of the wafer surface to obtain more comprehensive surface structure information. Material characteristic information involves the wafer's material, chemical composition, and physical properties, which directly affect the selection of coating materials and the formulation of coating processes. Energy dispersive spectroscopy (EDS), X-ray diffraction (XRD), or Raman spectroscopy are used to determine the wafer's chemical composition and crystal structure. The wafer's hardness, density, thermal conductivity, and other physical properties are measured to better select matching coating materials and process parameters. Application scenario characteristic information includes the environmental conditions in which the wafer is located and its intended use. The collected structural characteristic information, material characteristic information, and application scenario characteristic information are organized and standardized. Data analysis techniques are used to extract key characteristic parameters from the integrated data, forming a characteristic data sequence that comprehensively describes the wafer's properties.

[0018] S2: Based on the coating process chain of the wafer to be coated, extract the t-th node of the wafer coating, wherein the coating process chain includes T wafer coating nodes of the wafer to be coated, and t and T are both positive integers, and t belongs to T.

[0019] Specifically, the composition of the entire coating process chain needs to be clearly defined. This chain includes multiple stages such as preparation, pretreatment, coating, and post-treatment, each of which can be considered a coating node. For example, wafer cleaning before coating can be considered a node, primer coating another node, and topcoat coating yet another node. The coating process chain consists of T wafer coating nodes, where T is a positive integer representing the total number of nodes in the entire chain. After clarifying the entire coating process chain and the number of nodes, the t-th coating node is extracted based on specific needs or problems. Here, t is a positive integer and belongs to the set T, meaning it is a specific stage in the process chain. After extracting the t-th node, a detailed analysis of this node is required. This includes identifying the specific operational steps, materials used, and required process parameters for that node.

[0020] S3: Based on the t-th node of the wafer coating, perform coating expectation analysis according to the wafer feature data sequence to determine the expected coating information of the t-th node.

[0021] Specifically, the specific coating requirements and objectives for node t are identified, such as coating thickness, uniformity, and adhesion. Based on the application characteristics of the wafer, such as requirements for corrosion resistance, conductivity, and optical performance, the performance expectations for the node t coating are initially set. Key performance indicators of the coating are determined. Data analysis techniques are used to explore the potential relationship between wafer characteristic data and coating performance. Through statistical analysis and machine learning models, the impact of different wafer characteristics on coating performance is predicted. Based on the results of correlation analysis, the specific expectations for the node t coating are clarified, such as the expected thickness range, uniformity standard, and surface roughness.

[0022] S4: Based on the desired coating information of node t and the wafer feature data sequence, perform spraying control feature registration to construct the spraying control registration space of node t.

[0023] Specifically, the desired coating information at node t is integrated with the wafer feature data sequence. The impact of wafer features, such as size, shape, and material, on the coating effect is analyzed. Based on the desired coating information, such as coating thickness, uniformity, and roughness, key coating control parameters affecting these coating characteristics are determined, such as spraying speed, spraying pressure, spray gun distance, and spraying angle. Data analysis techniques, such as regression analysis and correlation analysis, are used to identify the relationship between wafer features and coating control parameters. Which wafer features have a significant impact on the coating effect are identified, and the corresponding coating control parameters are found. A multi-dimensional coating control registration space is constructed. This space associates wafer features with coating control parameters, forming a mapping relationship. In this space, each combination of wafer features corresponds to a specific set of coating control parameters to achieve the desired coating effect.

[0024] S5: Activate the spray control optimization evaluation channel, perform optimization analysis in conjunction with the spray control registration space of the t-th node, and generate a spray control scheme for the t-th node that satisfies the spray control optimization rules.

[0025] Specifically, the optimization and evaluation function of the spraying control system is activated to ensure automatic searching and evaluation of the merits of various spraying control schemes. Based on the desired coating information at node t, a clear optimization objective is set, such as minimizing coating thickness variation and maximizing coating uniformity. Spraying control optimization rules are defined, including the adjustment range of spraying parameters, optimization step size, and stopping conditions. The wafer features are matched with the spraying control parameters using the spraying control registration space at node t. Within the registration space, changes in spraying control parameters, such as spraying speed, pressure, and spray gun distance, are observed and evaluated to assess coating quality changes. Optimization algorithms, such as genetic algorithms and particle swarm optimization algorithms, are used to search for the optimal combination of spraying control parameters within the spraying control registration space. Based on the set optimization objectives and rules, the coating quality under different parameter combinations is continuously iterated and evaluated. After optimization analysis, the optimal combination of spraying control parameters that satisfies the spraying control optimization rules is selected. Based on these optimal parameters, a spraying control scheme for node t is generated, including specific key control parameters such as spraying speed, pressure, and spray gun distance.

[0026] S6: Generate a wafer coating control scheme based on the desired coating information of node t and the coating control scheme of node t.

[0027] Specifically, based on the spraying parameters determined in the node t spraying control scheme, such as spraying speed, spraying pressure, spray gun distance, and spraying angle, the position and role of node t in the entire spraying process chain, as well as its relationship with other nodes, are analyzed. Based on the node t spraying control scheme and considering the requirements of the entire spraying process chain, a comprehensive wafer spraying control scheme is developed. The scheme clearly defines the spraying parameters and key operational points for each node, ensuring that each step meets the requirements of the desired coating information.

[0028] S7: Based on the wafer coating end, control the coating of the wafer to be coated according to the wafer coating control scheme.

[0029] Specifically, ensure that the coating equipment, such as spray guns and coating robots, is in good working order. Prepare the necessary coating materials, such as paints and thinners. Perform necessary pretreatment on the wafers to be coated, such as cleaning and drying, to ensure that the wafer surface is clean and free of contamination. Load the generated wafer coating control scheme into the control system of the coating equipment. Ensure that the control system can accurately identify and execute the parameter settings in the scheme. Adjust the parameters of the coating equipment, such as spraying speed, spraying pressure, and spray gun distance, according to the coating control scheme. Conduct a test spray to check whether the coating effect meets expectations, and make fine adjustments if necessary. Place the wafers to be coated on the coating equipment. Start the coating equipment and spray the wafers to be coated according to the coating control scheme. Monitor the coating process to ensure that the coating is uniform and without omissions. After coating, perform post-processing on the wafers, such as drying and curing. Inspect the coated wafers to ensure that the coating quality meets the requirements and that there are no defects, bubbles, or other problems.

[0030] Furthermore, step S3 of this application also includes: Based on the wafer coating node t, a coating feature record retrieval is performed to obtain a wafer feature data sequence record set and a node t coating feature record set. A predetermined coating expectation learning model is then trained under supervision based on the wafer feature data sequence record set and the node t coating feature record set. At each predetermined number of training iterations, a mean squared loss coefficient for coating expectation prediction is obtained. When the mean squared loss coefficient for coating expectation prediction is less than a mean squared loss threshold for coating expectation prediction, a node t coating expectation parsing channel is generated. The wafer feature data sequence is then input into the node t coating expectation parsing channel to obtain the node t expected coating information.

[0031] Specifically, the process involves retrieving coating feature records based on the t-th node of the wafer coating process. A set of wafer feature data sequences and a set of coating feature records for the t-th node are extracted from the database. The wafer feature data sequence set contains feature data of the wafer at each coating node, such as size, shape, and material composition. The t-th node coating feature record set refers to the coating features of the wafer after coating at the t-th node, such as coating thickness, uniformity, color, and hardness. The retrieved wafer feature data sequence set and the t-th node coating feature record set are used to supervise the training of a predetermined coating expectation learning model. This model, such as a regression model or neural network in machine learning, is used to learn the relationship between wafer features and coating features. The model is trained using known data, i.e., the retrieved record set, to predict or resolve the coating expectation under given wafer features. During training, the mean squared error between the model's predicted coating features and the actual coating features is calculated; this error value reflects the accuracy of the model's prediction. When the mean squared loss coefficient of the coating expectation prediction is less than a preset threshold, it indicates that the model's predictive ability has reached a certain level of accuracy. At this point, we can generate the coating expectation parsing channel for node t. The coating expectation parsing channel is a channel built based on the trained model. It receives the wafer feature data sequence as input and outputs the corresponding expected coating information for node t. Finally, the wafer feature data sequence is input into the coating expectation parsing channel for node t. Through model calculation and analysis, the expected coating information for node t is obtained. This information includes key performance indicators such as coating thickness, color, hardness, and adhesion.

[0032] Furthermore, step S4 of this application also includes: Based on the wafer coating end, multiple wafer coating control record groups corresponding to the t-th node of the wafer coating are retrieved; the multiple wafer coating control record groups are registered and selected according to the desired coating information of the t-th node and the wafer feature data sequence to obtain the wafer coating control registration source of the t-th node; the coating control trigger feature analysis is performed based on the wafer coating control registration source of the t-th node to obtain the wafer coating control trigger feature space of the t-th node; based on the wafer coating end, the wafer coating control constraint information corresponding to the wafer coating equipment corresponding to the t-th node of the wafer coating is collected to establish the wafer coating control constraint feature space of the t-th node; the intersection of the wafer coating control constraint feature space of the t-th node and the wafer coating control trigger feature space of the t-th node is obtained to generate the coating control registration space of the t-th node.

[0033] Specifically, multiple wafer coating control record sets corresponding to node t are retrieved from the database at the wafer coating end. These record sets contain various coating control parameters and corresponding coating results used for similar wafer feature data sequences during historical coating processes. Using the desired coating information of node t and the current wafer feature data sequence, the retrieved multiple wafer coating control record sets are registered and selected. The coating control record that best matches the current wafer features and desired coating information is found and used as the benchmark for subsequent analysis, i.e., the wafer coating control registration source for node t. Based on the wafer coating control registration source for node t, the coating control triggering features are analyzed. The study of coating control parameters identifies which parameter combinations can trigger the desired coating effect. Through this analysis, a wafer coating control triggering feature space for node t can be constructed, which describes the coating control features that need to be triggered to achieve the desired coating effect. Simultaneously, coating control constraint information of the wafer coating equipment corresponding to node t is collected from the wafer coating end. This information includes the equipment's physical limitations, operating procedures, safety standards, etc. Based on this information, a wafer coating control constraint feature space at node t can be established, defining the constraints that must be followed during the actual coating process. Finally, by finding the intersection of the wafer coating control constraint feature space and the wafer coating control trigger feature space at node t, the coating control registration space at node t can be generated. This registration space considers both the trigger features required to achieve the desired coating effect and the constraints in the actual coating process.

[0034] Furthermore, this application also includes: Based on the plurality of wafer coating control record groups, a first wafer coating control record group is extracted, wherein the first wafer coating control record group includes first historical node expected coating information, first historical wafer feature data sequence, and first historical wafer coating control scheme; based on the t-th node expected coating information and the wafer feature data sequence, the first historical node expected coating information and the first historical wafer feature data sequence are registered and identified to determine a first coating control registration index; it is determined whether the first coating control registration index is greater than or equal to a coating control registration threshold; if the first coating control registration index is greater than or equal to the coating control registration threshold, the first historical wafer coating control scheme is recorded as the first wafer coating control registration scheme, and the first wafer coating control registration scheme is added to the t-th node wafer coating control registration source; based on the coating control registration threshold, the plurality of wafer coating control record groups are further registered and selected according to the t-th node expected coating information and the wafer feature data sequence to generate the t-th node wafer coating control registration source.

[0035] Specifically, from multiple wafer coating control record groups, the first wafer coating control record group is selected and extracted for analysis. This record group contains three main parts: the first historical node expected coating information, which specifies the expected coating effect for a certain node during the historical coating process; the first historical wafer feature data sequence, which contains the feature data of historical wafers corresponding to the expected coating information, such as size, shape, and material; and the first historical wafer coating control scheme, which is the coating control scheme adopted during the historical coating process for the aforementioned wafer features and expected coating information. Using the current t-th node expected coating information and wafer feature data sequence, the first historical node expected coating information and the first historical wafer feature data sequence are registered and identified. The degree of matching between the historical data and the current data is identified by comparing their similarity and differences. The first coating control registration index is a quantitative indicator used to represent the degree of matching between historical records and current requirements. The higher the index value, the better the matching degree. The first coating control registration index is compared with a preset coating control registration threshold. This threshold is a standard used to determine whether the historical coating control scheme is suitable for the current coating requirements. If the first coating control registration index is greater than or equal to the coating control registration threshold, it indicates that the historical coating control scheme has a high degree of matching with the current requirements and can be considered for adoption. If the first coating control registration index meets the standard, the first historical wafer coating control scheme is marked as the first wafer coating control registration scheme and added to the wafer coating control registration source at node t. Registration selection continues for the remaining multiple wafer coating control record groups. By repeatedly performing the above steps—extracting record groups, registration identification, determining the registration index, judging whether the index meets the standard, recording and adding registration schemes—a complete wafer coating control registration source at node t is finally generated. This registration source will contain multiple historical coating control schemes with a high degree of matching with the current coating requirements for subsequent selection and use.

[0036] Furthermore, step S5 of this application also includes: Spraying control parameters are set according to the spraying control registration space of the t-th node to obtain a first spraying control decision; based on the first spraying control decision, a first spraying control optimization index is calculated according to the spraying control optimization evaluation channel; a second spraying control decision is generated according to the spraying control registration space of the t-th node; based on the second spraying control decision, a second spraying control optimization index is calculated according to the spraying control optimization evaluation channel; based on the first and second spraying control optimization indices, current optimization is performed in combination with the first and second spraying control decisions to determine the current winning spraying control optimization index and the current winning spraying control decision; based on the current winning spraying control optimization index, the current winning spraying control decision is iteratively optimized according to the spraying control registration space of the t-th node to generate the t-th node spraying control scheme that satisfies the spraying control optimization rule; wherein, the spraying control optimization rule includes a predetermined number of spraying control optimizations.

[0037] Specifically, spray control parameters are set according to the spray control registration space at node t, firstly obtaining the first spray control decision. This decision is based on the parameters and optimization objectives within the registration space. Using the spray control optimization evaluation channel, a first spray control optimization index is calculated for the first spray control decision. This index is a quantitative indicator measuring the optimization degree of the decision. A second spray control decision is generated based on the spray control registration space at node t, and a second spray control optimization index is calculated using the same channel. Combining the first and second spray control optimization indices with their corresponding spray control decisions, a comparison is made for the current round of optimization. Through comparison, the winning spray control optimization index and the corresponding current winning spray control decision are determined. This winning decision is the best performing decision among the currently considered options. Based on the current winning spray control optimization index, the process returns to the spray control registration space at node t, iteratively optimizing the current winning spray control decision. Parameters are continuously adjusted within the registration space to find a better decision. During the iterative optimization process, the spray control optimization rules are followed, including a predetermined number of optimization attempts. This means the optimization process will stop after reaching the predetermined number of attempts to ensure efficiency and prevent infinite loops. Through iterative optimization, a spray control scheme for node t that satisfies the spray control optimization rules is finally generated. This scheme represents the best-performing combination of spray control parameters across multiple iterations, meeting the expected optimization objective.

[0038] Furthermore, this application also includes: The coating control optimization evaluation channel includes a coating control simulation channel and a coating control evaluation channel. Based on the coating control simulation channel, the wafer to be coated is subjected to multiple simulated coatings according to the first coating control decision to obtain multiple wafer simulated coatings. Based on the expected coating information at node t, deviation analysis is performed on the multiple wafer simulated coatings to generate a first coating control expected deviation vector. The first coating control expected deviation vector is input into the coating control evaluation channel to obtain the first coating control optimization index.

[0039] Specifically, the spray control optimization evaluation channel comprises two main parts: a spray control simulation channel and a spray control evaluation channel. Based on the spray control simulation channel, multiple simulated spraying processes are first performed on the wafer to be sprayed according to the first spray control decision. Through these multiple simulations, multiple simulated coatings are obtained on the wafer. These simulated coatings reflect the possible coating effects under different spraying parameters. Next, deviation analysis is performed on the multiple simulated coatings on the wafer using the desired coating information at node t. The purpose of this analysis is to compare the differences between the simulated coatings and the desired coatings, thereby evaluating the effectiveness of the spray control decision. The results of the deviation analysis are used to generate the first spray control desired deviation vector. This vector quantifies the degree of deviation between the simulated coatings and the desired coating. The first spray control desired deviation vector is input into the spray control evaluation channel. This channel evaluates the optimization degree of the spray control decision based on the value of the deviation vector. Through processing by the spray control evaluation channel, the first spray control optimization index is obtained. This index is a quantitative representation of the optimization effect of the spray control decision; a higher index indicates a better decision.

[0040] Furthermore, this application also includes: Based on the desired coating information at node t, a desired wafer coating is constructed; based on the desired wafer coating, coating size feature deviation analysis is performed on the multiple simulated wafer coatings to obtain multiple coating size expected deviation datasets; based on the desired wafer coating, coating uniformity deviation analysis is performed on the multiple simulated wafer coatings to obtain multiple coating uniformity expected deviation datasets; based on the desired wafer coating, coating adhesion deviation analysis is performed on the multiple simulated wafer coatings to obtain multiple coating adhesion expected deviation datasets; based on the desired wafer coating, coating surface feature deviation analysis is performed on the multiple simulated wafer coatings to obtain multiple coating surface feature expected deviation datasets; lumped value calculation is performed on the multiple coating size expected deviation datasets, the multiple coating uniformity expected deviation datasets, the multiple coating adhesion expected deviation datasets, and the multiple coating surface feature expected deviation datasets to obtain the first spraying control expected deviation vector.

[0041] Specifically, based on the desired coating information at node t, a desired wafer coating is constructed. This desired coating represents the ideal coating effect the wafer should achieve after spraying, including requirements for coating size, uniformity, adhesion, and surface features. Based on the desired wafer coating, coating size characteristic deviation analysis is performed on multiple simulated wafer coatings. The difference between the size of the simulated coating and the desired coating size is compared. By calculating the dimensional deviation of each simulated coating from the desired coating, multiple datasets of desired coating size deviations can be obtained. Similarly, based on the desired wafer coating, coating uniformity deviation analysis is performed on multiple simulated wafer coatings. The uniformity of the simulated coating distribution on the wafer surface is evaluated and compared with the uniformity of the desired coating. By calculating the uniformity deviation of each simulated coating from the desired coating, multiple datasets of desired coating uniformity deviations can be obtained. Continuing with the desired wafer coating, coating adhesion deviation analysis is performed on multiple simulated wafer coatings. Whether the adhesion of the simulated coating meets the requirements of the desired coating is evaluated. By calculating the adhesion deviation of each simulated coating from the desired coating, multiple datasets of desired coating adhesion deviations can be obtained. Finally, based on the desired wafer coating, surface feature deviation analysis was performed on multiple simulated wafer coatings. The differences between the surface features of the simulated coatings and the desired coating were compared. By calculating the deviations in surface features between each simulated coating and the desired coating, multiple datasets of expected deviations in coating surface features were obtained. Central tendency values, such as the mean and median, were calculated for multiple datasets of expected deviations in coating size, coating uniformity, coating adhesion, and coating surface features. These central tendency values ​​reflect the central trends of each type of deviation dataset.

[0042] In summary, the adaptive spraying control method for semiconductor wafers provided in this application has the following technical advantages: By collecting structural, material, and application scenario information of the wafer to be coated, a wafer feature data sequence is generated. Based on the coating process chain of the wafer to be coated, the t-th node of the wafer coating is extracted, where the coating process chain includes T wafer coating nodes, t and T are positive integers, and t ∈ T. Based on the t-th node of the wafer coating, coating expectation analysis is performed according to the wafer feature data sequence to determine the expected coating information for the t-th node. Coating control feature registration is performed based on the expected coating information for the t-th node and the wafer feature data sequence to construct the coating control registration space for the t-th node. Finally, the coating control optimization evaluation is activated. The channel, combined with the coating control registration space of the t-th node, performs optimization analysis to generate a coating control scheme for the t-th node that satisfies the coating control optimization rules. Based on the desired coating information of the t-th node and the coating control scheme of the t-th node, a wafer coating control scheme is generated. Based on the wafer coating end, the wafer to be coated is coated according to the wafer coating control scheme. This effectively solves the problem that the lack of adaptive adjustment capability in the existing technology makes it difficult to guarantee the accuracy and uniformity of coating, which further affects the performance and yield of semiconductor devices. The wafer coating control scheme enables precise coating control of the wafer to be coated, thereby improving coating quality and efficiency. Example 2

[0043] Based on the adaptive spraying control method for semiconductor wafers described in the foregoing embodiments, and using the same inventive concept, this application also provides an adaptive spraying control system for semiconductor wafers. Please refer to the appendix. Figure 2 The system includes: The feature data sequence generation module 11 is used to collect structural feature information, material feature information and application scenario feature information of the wafer to be coated, and generate wafer feature data sequence.

[0044] The spraying node extraction module 12 is used to extract the t-th node of the wafer spraying according to the spraying process chain of the wafer to be sprayed, wherein the spraying process chain includes T wafer spraying nodes of the wafer to be sprayed, and t and T are both positive integers, and t belongs to T.

[0045] The expectation parsing module 13 is used to perform coating expectation parsing based on the wafer feature data sequence at the t-th node of the wafer coating to determine the expected coating information at the t-th node.

[0046] The feature registration module 14 is used to perform spraying control feature registration based on the desired coating information of the t-th node and the wafer feature data sequence, and to construct the spraying control registration space of the t-th node.

[0047] The optimization analysis module 15 is used to activate the spray control optimization evaluation channel, perform optimization analysis in conjunction with the spray control registration space of the t-th node, and generate a spray control scheme for the t-th node that satisfies the spray control optimization rules.

[0048] The coating scheme generation module 16 is used to generate a wafer coating control scheme based on the desired coating information of the t-th node and the coating control scheme of the t-th node.

[0049] The spraying control module 17 is used to control the spraying of the wafer to be sprayed based on the wafer spraying end and according to the wafer spraying control scheme.

[0050] Furthermore, the expected parsing module 13 in the system is also used for: Based on the wafer coating node t, a coating feature record retrieval is performed to obtain a wafer feature data sequence record set and a node t coating feature record set. A predetermined coating expectation learning model is then trained under supervision based on the wafer feature data sequence record set and the node t coating feature record set. At each predetermined number of training iterations, a mean squared loss coefficient for coating expectation prediction is obtained. When the mean squared loss coefficient for coating expectation prediction is less than a mean squared loss threshold for coating expectation prediction, a node t coating expectation parsing channel is generated. The wafer feature data sequence is then input into the node t coating expectation parsing channel to obtain the node t expected coating information.

[0051] Furthermore, the feature registration module 14 in the system is also used for: Based on the wafer coating end, multiple wafer coating control record groups corresponding to the t-th node of the wafer coating are retrieved; the multiple wafer coating control record groups are registered and selected according to the desired coating information of the t-th node and the wafer feature data sequence to obtain the wafer coating control registration source of the t-th node; the coating control trigger feature analysis is performed based on the wafer coating control registration source of the t-th node to obtain the wafer coating control trigger feature space of the t-th node; based on the wafer coating end, the wafer coating control constraint information corresponding to the wafer coating equipment corresponding to the t-th node of the wafer coating is collected to establish the wafer coating control constraint feature space of the t-th node; the intersection of the wafer coating control constraint feature space of the t-th node and the wafer coating control trigger feature space of the t-th node is obtained to generate the coating control registration space of the t-th node.

[0052] Furthermore, the system also includes a control registration source generation module, used for: Based on the plurality of wafer coating control record groups, a first wafer coating control record group is extracted, wherein the first wafer coating control record group includes first historical node expected coating information, first historical wafer feature data sequence, and first historical wafer coating control scheme; based on the t-th node expected coating information and the wafer feature data sequence, the first historical node expected coating information and the first historical wafer feature data sequence are registered and identified to determine a first coating control registration index; it is determined whether the first coating control registration index is greater than or equal to a coating control registration threshold; if the first coating control registration index is greater than or equal to the coating control registration threshold, the first historical wafer coating control scheme is recorded as the first wafer coating control registration scheme, and the first wafer coating control registration scheme is added to the t-th node wafer coating control registration source; based on the coating control registration threshold, the plurality of wafer coating control record groups are further registered and selected according to the t-th node expected coating information and the wafer feature data sequence to generate the t-th node wafer coating control registration source.

[0053] Furthermore, the optimization analysis module 15 in the system is also used for: Spraying control parameters are set according to the spraying control registration space of the t-th node to obtain a first spraying control decision; based on the first spraying control decision, a first spraying control optimization index is calculated according to the spraying control optimization evaluation channel; a second spraying control decision is generated according to the spraying control registration space of the t-th node; based on the second spraying control decision, a second spraying control optimization index is calculated according to the spraying control optimization evaluation channel; based on the first and second spraying control optimization indices, current optimization is performed in combination with the first and second spraying control decisions to determine the current winning spraying control optimization index and the current winning spraying control decision; based on the current winning spraying control optimization index, the current winning spraying control decision is iteratively optimized according to the spraying control registration space of the t-th node to generate the t-th node spraying control scheme that satisfies the spraying control optimization rule; wherein, the spraying control optimization rule includes a predetermined number of spraying control optimizations.

[0054] Furthermore, the system also includes an offset vector generation module, used for: The coating control optimization evaluation channel includes a coating control simulation channel and a coating control evaluation channel. Based on the coating control simulation channel, the wafer to be coated is subjected to multiple simulated coatings according to the first coating control decision to obtain multiple wafer simulated coatings. Based on the expected coating information at node t, deviation analysis is performed on the multiple wafer simulated coatings to generate a first coating control expected deviation vector. The first coating control expected deviation vector is input into the coating control evaluation channel to obtain the first coating control optimization index.

[0055] Furthermore, the system also includes a lumped value calculation module, used for: Based on the desired coating information at node t, a desired wafer coating is constructed; based on the desired wafer coating, coating size feature deviation analysis is performed on the multiple simulated wafer coatings to obtain multiple coating size expected deviation datasets; based on the desired wafer coating, coating uniformity deviation analysis is performed on the multiple simulated wafer coatings to obtain multiple coating uniformity expected deviation datasets; based on the desired wafer coating, coating adhesion deviation analysis is performed on the multiple simulated wafer coatings to obtain multiple coating adhesion expected deviation datasets; based on the desired wafer coating, coating surface feature deviation analysis is performed on the multiple simulated wafer coatings to obtain multiple coating surface feature expected deviation datasets; lumped value calculation is performed on the multiple coating size expected deviation datasets, the multiple coating uniformity expected deviation datasets, the multiple coating adhesion expected deviation datasets, and the multiple coating surface feature expected deviation datasets to obtain the first spraying control expected deviation vector.

[0056] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Figure 1 The adaptive coating control method and specific examples for semiconductor wafers in Embodiment 1 are also applicable to the adaptive coating control system for semiconductor wafers in this embodiment. Through the foregoing detailed description of the adaptive coating control method for semiconductor wafers, those skilled in the art can clearly understand the adaptive coating control system for semiconductor wafers in this embodiment. Therefore, for the sake of brevity, it will not be described in detail here. As the system disclosed in the embodiments corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant details can be found in the method section.

[0057] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0058] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application also intends to include such modifications and variations.

Claims

1. An adaptive spraying control method for semiconductor wafers, characterized in that, The method includes: Collect structural, material, and application scenario information of the wafer to be coated, and generate a wafer feature data sequence. Based on the coating process chain of the wafer to be coated, extract the t-th node of wafer coating, wherein the coating process chain includes T wafer coating nodes of the wafer to be coated, and t and T are both positive integers, and t belongs to T; Based on the t-th node of the wafer spraying, the expected coating information of the t-th node is determined by analyzing the wafer feature data sequence. Based on the desired coating information of node t and the wafer feature data sequence, spray control feature registration is performed to construct the spray control registration space of node t. Activate the spray control optimization evaluation channel, and perform optimization analysis in conjunction with the spray control registration space of the t-th node to generate a spray control scheme for the t-th node that satisfies the spray control optimization rules. Based on the desired coating information of node t and the spraying control scheme of node t, a wafer spraying control scheme is generated; Based on the wafer coating end, the wafer to be coated is controlled by the coating control scheme.

2. The adaptive spraying control method for semiconductor wafers as described in claim 1, characterized in that, Based on the t-th node of the wafer coating, the desired coating information for the t-th node is determined by analyzing the wafer feature data sequence, including: Based on the t-th node of the wafer coating, a coating feature record retrieval is performed to obtain a wafer feature data sequence record set and a coating feature record set for the t-th node. The predetermined coating expectation learning model is trained under supervision based on the wafer feature data sequence record set and the coating feature record set at node t. The mean square loss coefficient of coating expectation prediction is obtained after each predetermined number of training iterations. When the expected mean square loss coefficient of the coating is less than the expected mean square loss threshold of the coating, the expected coating parsing channel of node t is generated. The wafer feature data sequence is input into the desired coating parsing channel at node t to obtain the desired coating information at node t.

3. The adaptive spraying control method for semiconductor wafers as described in claim 1, characterized in that, Based on the desired coating information of node t and the wafer feature data sequence, spraying control feature registration is performed to construct the spraying control registration space of node t, including: Based on the wafer coating end, retrieve multiple wafer coating control record groups corresponding to the t-th node of the wafer coating; Based on the desired coating information of node t and the wafer feature data sequence, the multiple wafer spraying control record groups are registered and selected to obtain the wafer spraying control registration source of node t. Based on the wafer spraying control registration source at node t, spraying control triggering feature analysis is performed to obtain the wafer spraying control triggering feature space at node t. Based on the wafer coating end, collect the wafer coating control constraint information corresponding to the wafer coating equipment corresponding to the t-th node of the wafer coating, and establish the wafer coating control constraint feature space of the t-th node. Find the intersection of the wafer coating control constraint feature space at node t and the wafer coating control trigger feature space at node t, and generate the coating control registration space at node t.

4. The adaptive spraying control method for semiconductor wafers as described in claim 3, characterized in that, Based on the desired coating information of node t and the wafer feature data sequence, the multiple wafer coating control record groups are registered and selected to obtain the wafer coating control registration source for node t, including: Based on the plurality of wafer coating control record groups, the first wafer coating control record group is extracted, wherein the first wafer coating control record group includes the expected coating information of the first historical node, the first historical wafer feature data sequence, and the first historical wafer coating control scheme; Based on the desired coating information of node t and the wafer feature data sequence, the desired coating information of the first historical node and the first historical wafer feature data sequence are registered and identified to determine the first spraying control registration index. Determine whether the first spraying control registration index is greater than or equal to the spraying control registration threshold; If the first coating control registration index is greater than or equal to the coating control registration threshold, the first historical wafer coating control scheme is recorded as the first wafer coating control registration scheme, and the first wafer coating control registration scheme is added to the t-th node wafer coating control registration source. Based on the spraying control registration threshold, the multiple wafer spraying control record groups are further registered and selected according to the desired coating information of the t-th node and the wafer feature data sequence to generate the wafer spraying control registration source of the t-th node.

5. The adaptive spraying control method for semiconductor wafers as described in claim 1, characterized in that, Activate the spray control optimization evaluation channel, perform optimization analysis in conjunction with the spray control registration space at node t, and generate a spray control scheme for node t that satisfies the spray control optimization rules, including: The spraying control parameters are set according to the spraying control registration space of the t-th node to obtain the first spraying control decision; Based on the first spraying control decision, the first spraying control optimization index is calculated according to the spraying control optimization evaluation channel. Based on the spraying control registration space of the t-th node, a second spraying control decision is generated; Based on the second spraying control decision, the second spraying control optimization index is calculated according to the spraying control optimization evaluation channel. Based on the first spraying control optimization index and the second spraying control optimization index, and combined with the first spraying control decision and the second spraying control decision, the current optimization is performed to determine the current winning spraying control optimization index and the current winning spraying control decision. Based on the current winning spray control optimization index, the current winning spray control decision is iteratively optimized according to the spray control registration space of the t-th node to generate the t-th node spray control scheme that satisfies the spray control optimization rule. The spraying control optimization rule includes a predetermined number of spraying control optimization attempts.

6. The adaptive spraying control method for semiconductor wafers as described in claim 5, characterized in that, Based on the first spraying control decision, and according to the spraying control optimization evaluation channel, the first spraying control optimization index is calculated, including: The spraying control optimization evaluation channel includes a spraying control simulation channel and a spraying control evaluation channel; Based on the spraying control simulation channel, the wafer to be sprayed is subjected to multiple simulated sprayings according to the first spraying control decision to obtain multiple wafer simulated coatings. Based on the desired coating information of the t-th node, deviation analysis is performed on the multiple simulated coatings on the wafers to generate a first spraying control desired deviation vector. The first expected deviation vector of the spraying control is input into the spraying control evaluation channel to obtain the first spraying control optimization index.

7. The adaptive spraying control method for semiconductor wafers as described in claim 6, characterized in that, Based on the desired coating information at node t, deviation analysis is performed on the multiple simulated coatings on the wafers to generate a first desired coating control deviation vector, including: Based on the desired coating information of the t-th node, construct the desired coating for the wafer; Based on the desired wafer coating, a coating size characteristic deviation analysis is performed on the multiple simulated wafer coatings to obtain multiple coating size expected deviation datasets. Based on the desired wafer coating, a coating uniformity deviation analysis is performed on the multiple simulated wafer coatings to obtain multiple coating uniformity expected deviation datasets. Based on the desired wafer coating, the coating adhesion deviation analysis is performed on the multiple simulated wafer coatings to obtain multiple datasets of desired coating adhesion deviations. Based on the desired wafer coating, surface feature deviation analysis is performed on the multiple simulated wafer coatings to obtain multiple datasets of desired coating surface feature deviations. The first spray control expected deviation vector is obtained by performing lumped value calculations on the multiple coating size expected deviation datasets, the multiple coating uniformity expected deviation datasets, the multiple coating adhesion expected deviation datasets, and the multiple coating surface feature expected deviation datasets, respectively.

8. An adaptive spraying control system for semiconductor wafers, characterized in that, The system is used for implementing the adaptive spraying control method for semiconductor wafers according to any one of claims 1 to 7, wherein the system comprises: The feature data sequence generation module is used to collect structural feature information, material feature information and application scenario feature information of the wafer to be coated, and generate wafer feature data sequence. The spraying node extraction module is used to extract the t-th node of the wafer spraying process according to the spraying process chain of the wafer to be sprayed, wherein the spraying process chain includes T wafer spraying nodes of the wafer to be sprayed, and t and T are both positive integers, and t belongs to T; The expectation parsing module is used to perform coating expectation parsing based on the wafer feature data sequence at the t-th node of the wafer coating to determine the expected coating information at the t-th node. The feature registration module is used to perform spraying control feature registration based on the desired coating information of the t-th node and the wafer feature data sequence, and to construct the spraying control registration space of the t-th node. The optimization analysis module is used to activate the spray control optimization evaluation channel, and perform optimization analysis in combination with the spray control registration space of the t-th node to generate the spray control scheme of the t-th node that satisfies the spray control optimization rules. The coating scheme generation module is used to generate a wafer coating control scheme based on the desired coating information of the t-th node and the coating control scheme of the t-th node. The coating control module is used to control the coating of the wafer to be coated based on the wafer coating end and the wafer coating control scheme.

Images