A wafer pre-treatment method and device before coating

Abstract

The application discloses a wafer pre-treatment method and device before film coating, and relates to the technical field of semiconductor manufacturing. The application discloses a wafer pre-treatment method and device before film coating, and relates to the technical field of semiconductor manufacturing. The wafer pre-treatment method and device before film coating can significantly improve the process consistency of the wafer pre-treatment stage and the stability of the subsequent film coating quality.

Description

Technical Field

[0001] This invention relates to the field of semiconductor manufacturing technology, and more specifically to a pretreatment method and apparatus for wafer coating. Background Technology

[0002] In semiconductor manufacturing, wafers typically undergo pretreatment steps such as cleaning, activation, or etching before entering the coating process to ensure the wafer surface achieves good cleanliness and activity, thereby improving the uniformity and stability of thin film deposition. The types of chemical solutions used in pretreatment are diverse, and their concentration distribution, batch variations, and changes over time directly affect the actual treatment effect on the wafer surface. Therefore, the consistency of the pretreatment process has become one of the important factors affecting coating quality.

[0003] While existing manufacturing systems are generally equipped with automated chemical solution supply modules, significant shortcomings remain. On one hand, the chemical solutions at different processing stations may originate from different batches, or their concentration may drift due to variations in storage and usage cycles. On the other hand, even if the same batch of chemical solution is distributed to multiple processing devices, differences in distribution paths, usage frequency, and environmental conditions can still lead to deviations in the actual concentration of the chemical solution encountered by each device. These factors can all result in inconsistent pretreatment effects on wafers in different devices, leading to variations in surface characteristics and affecting the uniformity and adhesion of subsequent coating layers. Furthermore, existing technologies largely rely on data collected from single stations for concentration monitoring, failing to provide comprehensive analysis of concentration trend changes between different batches. When abnormal concentration distributions occur, there is a lack of ability to identify key deviation points, and it is difficult to generate targeted correction parameters to regulate the chemical solution state of each device. Simultaneously, the lack of an effective data synchronization mechanism between devices makes it difficult to maintain consistent chemical solution preparation parameters across multiple stations, resulting in a lack of stable coordination in the overall pretreatment process. Furthermore, during wafer processing, the equipment typically only records basic operating parameters and lacks the ability to dynamically compare deviation information in the processing log. It cannot adjust the equipment adaptation parameters in a timely manner according to the deviation, which may result in the wafer coming into contact with chemical solutions of a concentration that deviates from the target concentration when processed in different batches or at different times, affecting the repeatability of the pretreatment steps. Summary of the Invention

[0004] The purpose of this invention is to provide a pretreatment method and apparatus for wafers before coating, thereby solving the problems existing in the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a pretreatment method for wafers before coating, comprising transporting the wafer to be processed to a pretreatment station, selecting a corresponding chemical solution for wafer surface cleaning, activation, or etching according to process requirements, collecting batch data and concentration distribution information of the chemical solution from each processing station through a sensor network, and performing multi-dimensional verification on the real-time acquisition frequency and data sampling accuracy to obtain the initial change trend within the concentration distribution range and determine the concentration distribution characteristics between batches; based on the concentration distribution characteristics between batches, if the concentration range exceeds a preset threshold interval, an anomaly detection rule is triggered, and combined with historical trends... The degree of anomaly in concentration distribution is determined by comparison and analysis of environmental interference factors. Based on this anomaly, dynamic concentration data around relevant marker points is extracted from the real-time monitoring system. Time-series analysis is performed based on equipment response delays to obtain the distribution pattern of potential deviations and identify key points requiring adjustment. According to the deviation distribution pattern of these key points, a deviation calculation formula and correction value generation logic are used, combined with historical data weights and compensation range limitations, to calculate targeted correction values. These correction values ​​are then input to the chemical liquid distribution module of each device, enabling real-time adjustment of chemical liquid concentration through the allocation actuator. A data synchronization protocol is used to synchronize the correction values ​​to all stations.

[0006] Preferably, the process involves transporting the wafer to be processed to a pre-processing station, selecting a corresponding chemical solution for wafer surface cleaning, activation, or etching according to process requirements, collecting batch data and concentration distribution information of the chemical solution from each processing station via a sensor network, and performing multi-dimensional verification on the real-time acquisition frequency and data sampling accuracy to obtain the initial change trend within the concentration distribution range. This is used to determine the concentration distribution characteristics between batches. The process includes: collecting batch data and concentration distribution information of the chemical solution from each processing station via a sensor network, performing multi-dimensional verification on the real-time acquisition frequency and data sampling accuracy to obtain the initial change trend within the concentration distribution range; fusing temperature monitoring data obtained from the processing station based on the initial change trend to obtain the fluctuation parameters of the concentration distribution between batches, and determining the correspondence between the fluctuation parameters and the liquid volume adjustment; adjusting the chemical solution supply based on the fluctuation parameters, obtaining a stable concentration distribution from the adjusted supply, determining whether the stable concentration distribution meets a preset threshold, and if so, recording it as optimized batch data; and obtaining a comparison between the optimized batch data and the initial change trend to determine the concentration distribution characteristics between batches.

[0007] Preferably, based on the concentration distribution characteristics between batches, if the concentration range exceeds a preset threshold range, an anomaly detection rule is triggered. The determination of the anomaly degree of concentration distribution by combining historical trend comparison and environmental interference factor analysis includes obtaining the range from the concentration distribution characteristics between batches, performing preliminary filtering of the range using environmental interference factors to eliminate noise influence and obtain fluctuation parameters of the range; for the fluctuation parameters, if the range exceeds a preset threshold range, an anomaly detection rule is triggered, integrating data synchronization between sites to consolidate real-time information and determine the detection basis; based on the detection basis, historical trend comparison is used, calculating the difference between current data and historical data to obtain trend comparison differences, and combining interference analysis to determine the distribution anomaly parameters; the anomaly degree is obtained through the distribution anomaly parameters, and the supply is dynamically calibrated to balance the distribution and obtain stable characteristics of the concentration distribution.

[0008] Preferably, the process for determining the degree of concentration distribution anomaly involves extracting dynamic concentration data around relevant marker points from the real-time monitoring system, performing time-series analysis based on equipment response delay to obtain potential deviation distribution patterns, and identifying key points requiring adjustment. This includes extracting concentration anomaly data around marker points from the real-time monitoring system, adjusting data timing based on equipment delay parameters through response delay correction, and integrating environmental interference factors to obtain preliminary time-series deviation analysis results. Historical trend comparisons are then used to evaluate the time-series deviation analysis results. If the deviation exceeds a preset threshold, distribution pattern recognition is triggered. Potential distribution patterns are obtained by comparing the current deviation with historical patterns. Key points are selected based on the potential distribution patterns. Supply adjustment and optimization are used to integrate site data synchronization and calculate quantification values ​​to determine anomaly degree quantification parameters. Stable distribution equilibrium characteristics are obtained through anomaly degree quantification parameters to determine key points requiring adjustment.

[0009] Preferably, the step of calculating a targeted correction value based on the deviation distribution pattern of key points, using a deviation calculation formula and correction value generation logic, combined with historical data weights and compensation range limitations, includes: obtaining the deviation distribution pattern from the key point deviations; fusing the deviation distribution pattern with historical data weights to obtain a preliminary weighted deviation value, wherein the fusing is achieved by multiplying the historical data weights by the sum of the values ​​of each element of the deviation distribution pattern using a weighted average method; determining the intermediate correction deviation based on the preliminary weighted deviation value using a deviation calculation formula combined with the correction value generation logic, wherein the deviation calculation formula is D=(PH)×W, where D represents the deviation value, P represents the current point value, H represents the historical average value, and W represents the weight; after calculation using this formula, the correction value generation logic adjusts to the intermediate correction deviation; obtaining the compensation range limitation and fusing the intermediate correction deviation, if it exceeds a preset threshold, adjusting the point weights to obtain a limited deviation pattern, wherein the fusing is achieved by summing the components of the intermediate correction deviation within the compensation range limitation; calculating the targeted correction value using the limited deviation pattern, and determining the control parameters synchronously to obtain the equipment control basis, wherein the calculation is achieved by calculating the average value of the limited deviation pattern to obtain the targeted correction value.

[0010] Preferably, the step of inputting the correction value to the chemical solution distribution module of each device, and adjusting the chemical solution concentration in real time through the dispensing actuator, and synchronizing the correction value to all stations using a data synchronization protocol to ensure a unified preparation status includes: obtaining correction value data through the chemical solution distribution module; performing format verification on the input correction value to obtain a standardized correction dataset; based on the standardized correction dataset, calling the actuator interface to transmit the correction value to the chemical solution concentration adjustment unit to determine the concentration adjustment execution command; encrypting the execution command using the data synchronization protocol to obtain the encrypted command data packet and determining the integrity of the command data packet; if the command data packet is complete, distributing the encrypted command data packet to each station through the network channel to obtain reception confirmation information from each station; parsing the status field in the reception confirmation information of each station to determine whether each station has synchronized the correction value data; if a station has not synchronized the correction value data, resending the encrypted command data packet through a backup channel to obtain secondary confirmation information from the station; updating the preparation status record of each station based on the secondary confirmation information to determine whether all stations have reached a consistent status.

[0011] Preferably, the method further includes comparing deviation records in the wafer surface treatment log with a unified concentration status. If the deviation record is within a preset threshold range, the device adaptation parameters are automatically reloaded to ensure that the wafer is always exposed to a consistent chemical solution concentration in subsequent processing stages, resulting in optimized concentration consistency data. Specifically, this includes obtaining specific data of deviation records from the wafer surface treatment log, comparing the deviation records with preset thresholds to determine whether the deviation is within an acceptable range; if the deviation record is within the preset threshold range, obtaining relevant device adaptation parameters through the system interface to determine the device adaptation update requirements; and adjusting the device adaptation parameters according to the update requirements. The parameter loading module transmits the adaptation parameters to the device control unit and obtains status feedback after parameter loading. For this status feedback, the execution fields are parsed to determine if the device has completed the adaptation parameter update process. If the device has completed the adaptation parameter update, real-time concentration data is obtained through the concentration status monitoring module to determine if the concentration consistency meets the expected standard. Based on the concentration consistency data, a record file of the optimization results is generated, and the optimization results are saved through the data storage unit, obtaining the saved index identifier. Based on the saved index identifier, the relevant fields in the wafer surface treatment log are updated to determine if the log data has been synchronized to the latest status.

[0012] Preferably, the method further includes generating targeted feedback reports based on optimized concentration consistency data, and iteratively updating system parameters by integrating environmental interference factors and historical trends. This drives the system to continuously optimize the preparation path in multi-site and multi-batch processing. Specifically, this involves obtaining the current chemical solution preparation status information through concentration consistency data, analyzing key fields in the status information to determine the initial deviation of the preparation path; obtaining environmental interference factor data based on the initial deviation, extracting corresponding influence weights from a pre-established interference factor database, and judging the intensity of the interference factor's effect on the preparation path; if the intensity of the interference factor's effect on the preparation path exceeds a preset threshold, obtaining the change patterns in multi-batch processing through historical trend data to determine the adjustment direction of the preparation path in different time periods; and, based on the adjustment direction, obtaining the differences in operating status between multiple sites, extracting the system parameter update requirements from the status differences, and obtaining the priority ranking of parameter iterative updates.

[0013] Preferably, the step of generating a targeted feedback report based on the optimized concentration consistency data, and iteratively updating the system parameters by integrating environmental interference factors and historical trends to drive the system to continuously optimize the preparation path in multi-site and multi-batch processing also includes adjusting the system parameters step by step using an iterative update module according to priority, obtaining the adjusted parameter configuration status, and determining whether the configuration meets the stability standards for multi-site operation; if the configuration meets the stability standards for multi-site operation, the updated parameters are loaded through the chemical solution preparation module, real-time operating data of the pretreatment stage is obtained, and the monitoring results of process stability are determined; based on the monitoring results of process stability, an optimized path record for multi-batch processing is generated, the path record is saved through the data storage unit, and the saved identification information is obtained.

[0014] A pre-processing apparatus for wafer coating is provided to implement the steps of the pre-processing method for wafer coating. The apparatus includes a pre-processing and data acquisition module, which transports the wafer to be processed to a pre-processing station and selects a corresponding chemical solution for wafer surface cleaning, activation, or etching according to process requirements. A sensor network is used to collect batch data and concentration distribution information of the chemical solution from each processing station, and multi-dimensional verification is performed on the real-time acquisition frequency and data sampling accuracy to obtain the initial change trend within the concentration distribution range and determine the concentration distribution characteristics between batches. An anomaly detection module, based on the concentration distribution characteristics between batches, triggers anomaly detection rules if the concentration range exceeds a preset threshold range, and determines the degree of anomaly in the concentration distribution by combining historical trend comparison and environmental interference factor analysis. A deviation localization module, based on the degree of anomaly in the concentration distribution, extracts dynamic concentration data around relevant marker points from the real-time monitoring system, performs time-series analysis based on equipment response delay, obtains the distribution pattern of potential deviations, and determines the required deviations. The key points to be adjusted include: a correction value generation module, which calculates targeted correction values ​​based on the deviation distribution pattern of the key points using deviation calculation formulas and correction value generation logic, combined with historical data weights and compensation range limitations; a concentration allocation module, which inputs the correction values ​​to the chemical solution allocation modules of each device, and realizes real-time adjustment of chemical solution concentration through allocation actuators, and synchronizes the correction values ​​to all sites using a data synchronization protocol; a processing log comparison and device adaptation module, which compares the deviation records in the wafer surface processing logs according to the unified concentration status, and if the deviation record is within the preset threshold range, it automatically triggers the reloading of device adaptation parameters, so that the wafer always comes into contact with a consistent chemical solution concentration in subsequent processing stages, and obtains optimized concentration consistency data; and a feedback update module, which generates targeted feedback reports based on the optimized concentration consistency data, and integrates environmental interference factors and historical trends to iteratively update system parameters, driving the system to continuously optimize the preparation path in multi-site and multi-batch processing.

[0015] As can be seen from the above technical solution, the present invention has the following beneficial effects: This pretreatment method and apparatus for wafer coating achieves dynamic consistency control of pretreatment chemical solution concentration by introducing chemical solution concentration distribution analysis, key deviation point identification, correction value generation, multi-site synchronous allocation, and process closed-loop optimization mechanisms during the pretreatment process. Compared to existing technologies that rely on single-site detection, cannot identify batch differences, and lack equipment coordination, this invention can acquire batch data and concentration distribution information of chemical solutions at each processing station through a sensor network, and judge concentration anomalies by combining historical trends and environmental interference factors, thereby accurately identifying key deviation points affecting pretreatment stability. Through deviation calculation and correction value generation logic, the chemical solution allocation module of each device can be adjusted in real time, and the unified preparation status between multiple stations is ensured through a data synchronization protocol, significantly improving the consistency of chemical solution supply. On the other hand, this invention achieves automatic updating of equipment adaptation parameters by comparing deviation information in the wafer processing log, ensuring stable pretreatment effects for wafers in different batches and on different devices. Simultaneously, by conducting feedback analysis on the optimized concentration consistency data and iteratively updating the system parameters in conjunction with environmental factors and historical models, a closed-loop process control system for multiple batches and multiple sites can be formed. In summary, this invention effectively solves the problems of chemical solution concentration drift, insufficient multi-equipment coordination, and inconsistent pretreatment effects in existing technologies, thereby significantly improving the process consistency in the wafer pretreatment stage and the stability of subsequent coating quality. Attached Figure Description

[0016] Figure 1 This is a flowchart of the pretreatment method for wafer coating according to the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] like Figure 1As shown, the present invention provides a technical solution: a pretreatment method for wafers before coating, characterized by comprising: S1, transporting the wafer to be processed to a pretreatment station, and selecting a corresponding chemical solution for wafer surface cleaning, activation, or etching according to process requirements; collecting batch data and concentration distribution information of the chemical solution from each processing station through a sensor network, and performing multi-dimensional verification on the real-time acquisition frequency and data sampling accuracy to obtain the initial change trend within the concentration distribution range, which is used to determine the concentration distribution characteristics between batches; S2, based on the concentration distribution characteristics between batches, if the concentration range exceeds a preset threshold range, triggering anomaly detection rules, and judging the degree of anomaly in the concentration distribution by combining historical trend comparison and environmental interference factor analysis; S3, based on the degree of anomaly in the concentration distribution, extracting dynamic concentration data around relevant marker points from the real-time monitoring system, performing time-series analysis based on equipment response delay, obtaining the distribution pattern of potential deviations, and determining the key points that need adjustment; S4, according to the relevant... The deviation distribution pattern of key points is calculated using a deviation calculation formula and correction value generation logic, combined with historical data weights and compensation range limitations, to generate targeted correction values. S5: The correction values ​​are input to the chemical solution distribution module of each device. Real-time adjustment of the chemical solution concentration is achieved through the allocation actuator. A data synchronization protocol is used to synchronize the correction values ​​to all sites, ensuring a consistent preparation status. S6: Based on the consistent concentration status, deviation records in the wafer surface treatment log are compared. If the deviation record is within a preset threshold range, the device adaptation parameters are automatically reloaded, ensuring that the wafer always encounters a consistent chemical solution concentration in subsequent processing stages, resulting in optimized concentration consistency data. S7: Based on the optimized concentration consistency data, a targeted feedback report is generated. Environmental interference factors and historical trends are integrated to iteratively update system parameters, driving the system to continuously optimize the preparation path in multi-site and multi-batch processing, ultimately ensuring chemical solution consistency and process stability in the wafer pretreatment stage.

[0019] In the above implementation, each pretreatment station continuously collects batch information of chemical solutions, local concentration distribution, and sampling frequency and accuracy parameters through a deployed sensor network. A multi-dimensional verification mechanism is used to confirm the validity of real-time data to ensure the reliability of concentration change trends. The system first constructs batch-to-batch concentration distribution characteristics based on the verified data. When a concentration fluctuation exceeds a threshold range, the anomaly detection module automatically activates, combining historical concentration change trends and environmental interference factors (such as temperature, humidity, and equipment load status) to identify the anomaly type and its severity. Subsequently, the system extracts dynamic concentration data near key marker points from the monitoring system based on the anomaly severity and performs sequence analysis using equipment response delay parameters to infer the path of deviation and its change pattern over time, thereby accurately locating key points requiring adjustment. Based on the deviation model of these key points, the system introduces a deviation calculation formula and a correction value generation algorithm, generating the final correction value by superimposing historical data weights and compensation range limitations. After the correction value is input into the chemical solution distribution module of each device, the dispensing actuator adjusts the chemical solution component ratio and flow parameters in real time, and uses a data synchronization protocol to ensure all stations maintain a consistent preparation status. After achieving uniform concentration, the system performs deviation detection on the wafer processing logs. If the deviation falls within a set range, the system automatically reloads the equipment adaptation parameters to ensure that subsequent wafers are always exposed to a consistent chemical concentration environment. Finally, by analyzing the optimized concentration consistency data, the system generates a feedback report, which is used for continuous iteration of the algorithm and parameters. This continuously optimizes the multi-site, multi-batch processing path and maintains the stability of the preprocessing process.

[0020] S1 includes collecting batch data and concentration distribution information of chemical solutions from each processing station through a sensor network, performing multi-dimensional verification on the real-time acquisition frequency and data sampling accuracy to obtain the initial trend of change within the concentration distribution range; integrating temperature monitoring data obtained from the processing station based on the initial trend of change to obtain the fluctuation parameters of concentration distribution between batches, and determining the correspondence between the fluctuation parameters and the liquid volume adjustment; adjusting the chemical solution supply based on the fluctuation parameters, obtaining a stable concentration distribution from the adjusted supply, determining whether the stable concentration distribution meets the preset threshold, and if it does, recording it as optimized batch data; obtaining the comparison difference between the optimized batch data and the initial trend of change to determine the concentration distribution characteristics between batches.

[0021] In this embodiment, the sensor network continuously collects data on each batch of chemical solution entering each processing station. Specifically, when each batch of wafers enters the preprocessing station, the concentration distribution information of the chemical solution is continuously recorded at a pre-set real-time acquisition frequency, starting from the moment the batch of wafers comes into contact with the chemical solution. Each record includes the processing station number, acquisition time, location of the chemical solution, concentration value at that location, and temperature monitoring data at that moment. To ensure data reliability, at each acquisition time point, the sensor repeatedly measures multiple concentration values ​​at the same location within a very short time interval, for example, three consecutive measurements. Then, the three values ​​are verified in multiple dimensions: The first dimension is acquisition frequency verification, specifically comparing the interval between two adjacent acquisition times with a pre-set target acquisition interval. If the deviation of the interval does not exceed 10% of the target interval, the acquisition frequency is considered valid; otherwise, the acquisition is marked as invalid and discarded. The second dimension is sampling accuracy verification, specifically calculating the maximum value among the three repeated measurements at the same location. The difference between the concentration value and the minimum concentration value is considered acceptable if the difference does not exceed the allowable error range given by the calibrated measuring instrument; otherwise, the data set is discarded. The third dimension is concentration rationality verification, which involves comparing the current concentration value with the average concentration of the same batch at adjacent time points or adjacent spatial locations. If the difference is within the allowable fluctuation range, it is considered reasonable; otherwise, it is recorded as an anomaly and does not participate in subsequent trend calculations. After the above multi-dimensional verification, the retained concentration distribution information is sorted in chronological order. In each fixed time interval, the concentration values ​​of all sampling points in that time interval are summed and divided by the number of valid sampling points in that time interval to obtain the average concentration value of that time interval. Then, the average concentration values ​​of all time intervals in a batch are connected sequentially in chronological order to form the initial trend of the batch in the entire concentration distribution range. The initial trend is described in words as whether the concentration of the chemical solution in the batch gradually increases, gradually decreases, or first increases and then decreases in a certain period of time.After obtaining the initial trend, the temperature monitoring data recorded within the same time interval are paired with the corresponding average concentration values ​​one by one. All paired data for each batch are statistically analyzed. First, the maximum and minimum concentration values ​​for that batch during the entire observation period are calculated, and the difference between them is taken as the concentration fluctuation amplitude for that batch. Then, the average concentration values ​​of all time intervals are summed and divided by the number of time intervals to obtain the average concentration for that batch. The average concentration values ​​of adjacent time intervals are subtracted, and the absolute value is taken. All differences are then summed and divided by the number of differences to obtain the average change for that batch. Simultaneously, the maximum, minimum, and average temperature values ​​for all time intervals within the same batch, as well as the average change between adjacent time intervals, are calculated. By comparing the concentration changes with the temperature changes one by one, it is determined whether the concentration increases or decreases when the temperature rises, and the direction and magnitude of the concentration change when the temperature decreases. This yields batch-to-batch concentration distribution fluctuation parameters reflecting the relationship between temperature and concentration. In this embodiment, these fluctuation parameters include the concentration fluctuation amplitude, average concentration, average concentration change, and the average change in concentration caused by temperature changes. Before the system is put into formal production, several representative trial production batches are selected. For each trial production... Different chemical solution supply volumes are set for each batch. The corresponding fluctuation parameters are calculated using the method described above. Each set of fluctuation parameter values ​​is mapped one-to-one with the current chemical solution supply volume. Batches are sorted from smallest to largest fluctuation amplitude, and those that meet the subsequent coating quality requirements are selected as reference batches. The fluctuation parameter values ​​of these reference batches and their corresponding chemical solution supply volumes are recorded, forming a correspondence table between fluctuation parameters and solution volume adjustments. In actual production, when the concentration data of a new batch is verified through multi-dimensional analysis to obtain the fluctuation parameters for that batch, the current fluctuation amplitude of that batch is compared with the fluctuation amplitude of the reference batches in the correspondence table. When the current fluctuation range is greater than the fluctuation range of the reference batch, the corresponding relationship table is used to find the supply volume fraction that needs to be reduced. For example, when the difference falls in the first interval, the current chemical liquid supply is reduced by 5%, and when the difference falls in the second interval, the current chemical liquid supply is reduced by 10%. Conversely, when the current fluctuation range is less than the fluctuation range of the reference batch, the corresponding relationship table is used to find the supply volume fraction that needs to be increased. For example, when the difference falls in the first interval, it is increased by 5%, and when the difference falls in the second interval, it is increased by 10%. When the current fluctuation range falls within the allowable deviation range of the fluctuation range of the reference batch, the supply volume remains unchanged.After each adjustment, under the new supply conditions, concentration distribution information is collected again at the same sampling frequency and accuracy requirements. The maximum concentration value, minimum concentration value, and the difference between them are recalculated over at least three consecutive observation periods, and a new average change is calculated. When the change in the difference between the maximum and minimum concentration values ​​is lower than a preset change threshold and the difference itself falls within a preset stable concentration range over these three or more consecutive periods, a stable concentration distribution is considered to have been obtained under the current supply. This stable concentration distribution is then compared with a preset threshold. When the difference between the concentration value at each sampling point and the allowable deviation range of the target concentration does not exceed the upper limit of the preset threshold and is not lower than the lower limit of the preset threshold, the sample is considered stable. The stable concentration distribution of each batch, along with corresponding fluctuation parameters and supply volume, is recorded as optimized batch data. Finally, the optimized batch data is compared point-by-point with the initial trend before supply volume adjustment. The difference between the stable concentration value and the initial average concentration value at the same time interval or spatial location is calculated, and this difference is statistically analyzed across multiple batches to identify recurring patterns in the difference distribution across most batches. For example, the concentration difference between the inlet and outlet positions often concentrates within a fixed range, or the inflection point where the concentration rises and stabilizes within a certain time period concentrates within a specific interval. These recurring regularity differences across multiple batches are used as the concentration distribution characteristics between batches.

[0022] S2 includes obtaining the amplitude range from the concentration distribution characteristics between batches, initially filtering the amplitude range through environmental interference factors to remove noise influence and obtain the fluctuation parameters of the amplitude range; for the fluctuation parameters, if the amplitude range exceeds the preset threshold range, an anomaly detection rule is triggered, and data synchronization between sites is integrated to consolidate real-time information to determine the detection basis; based on the detection basis, historical trend comparison is adopted, and the difference between the current data and historical data is calculated to obtain the trend comparison difference, and combined with interference analysis to determine the distribution anomaly parameters; the degree of anomaly is obtained through the distribution anomaly parameters, and the supply is dynamically calibrated to balance the distribution and obtain the stable characteristics of the concentration distribution.

[0023] In this embodiment, after completing step S1 and obtaining the batch-to-batch concentration distribution characteristics for each batch of wafers, the system first organizes the concentration distribution characteristics of each batch. For all effective concentration measurements of the same batch at the same pre-processing station, they are arranged in chronological order. The maximum and minimum concentration values ​​of the batch at that station are then identified. The minimum concentration value is subtracted from the maximum concentration value to obtain the range of the batch at that station. Simultaneously, the ranges of the same batch at different stations are summarized to form the set of ranges for that batch. In this embodiment, the batch-to-batch concentration distribution characteristics are defined as: the average concentration value, maximum concentration value, and minimum concentration value of each batch of wafers at each pre-processing station and each time interval. The statistical results of low concentration values ​​and their changing trajectories during processing are obtained in step S1 by comparing concentration data from multiple batches of wafers with optimized batch data, and are used as input data for step S2. After obtaining the amplitude range of each batch, the system simultaneously reads the environmental interference factor data recorded throughout the entire processing period of that batch. In this embodiment, environmental interference factors include temperature, humidity, and equipment operating status. The temperature is the temperature value synchronously recorded by the temperature sensor located near the pre-processing station at each concentration acquisition. The humidity is the humidity value synchronously recorded by the humidity sensor located in the same area. The equipment operating status is whether the equipment is in a positive state, as recorded by the equipment control system at each moment. The system displays status indicators for normal operation, start / stop switching, or maintenance modes. To initially filter the amplitude range based on environmental interference factors and eliminate noise impact, the system pre-selects 10 reference batches operating under stable environmental conditions during the process verification phase. Temperature, humidity, and equipment operating status, along with corresponding amplitude range changes, are recorded at all time points in these 10 batches. By comparing these 10 batches of data point by point, the system determines the natural fluctuation range of the amplitude range when temperature and humidity are within stable ranges and the equipment is continuously operating normally. Simultaneously, it determines the magnitude of instantaneous fluctuations added to the amplitude range when temperature, humidity, or equipment status changes abruptly. Based on this, the system defines the normal range of temperature, normal range of humidity, and normal operating range of the equipment. The range of the state is written into the system parameter library as the normal range of environmental interference factors. In actual production, after the system calculates the amplitude range of a certain batch at each site, it will compare the environmental interference factor data corresponding to each time period with the above normal range point by point. When the temperature and humidity are both within the normal range and the equipment is operating normally within a certain time period, the amplitude change corresponding to that time period is considered to be an effective amplitude change that truly reflects the process state. When the temperature or humidity exceeds the normal range or the equipment operation state changes from start to stop within a certain time period, that time period is considered to be a period of significant environmental interference. The amplitude change generated within that time period is considered to be noise, and the system will remove this part of the data from subsequent calculations.After removing all time periods marked with significant environmental interference, the system recalculates the maximum and minimum concentration values ​​and their differences for each remaining time period in each batch. It also calculates the average amplitude range of all time periods and the average amplitude range variation between adjacent time periods. The recalculated maximum, minimum, average amplitude range, and average amplitude variation are used as the amplitude range fluctuation parameters for that batch. In this embodiment, the fluctuation parameters are defined as a set of specific values, including the maximum and minimum amplitudes after noise removal, the arithmetic mean of the amplitudes of all time periods, and the arithmetic mean of the differences in amplitudes between adjacent time periods. These values ​​are directly calculated by the system based on the filtered amplitude data and do not include any undefined statistics. For the obtained fluctuation parameters, the system compares the average and maximum amplitudes of each batch with a preset threshold range. In this embodiment, the preset threshold range is determined during the process verification stage as follows: First, 20 batches of normal batches that have been tested and confirmed to meet quality requirements for coating thickness uniformity and defect density in actual production are selected. The maximum and average amplitudes of each of these 20 batches after filtering are calculated. The system selects the largest value from the maximum amplitudes of 20 batches as the upper limit candidate value for the allowable amplitude, and the smallest value from the maximum amplitudes of 20 batches as the lower limit candidate value for the allowable amplitude. Then, it sorts the average amplitudes of these 20 batches and selects the average amplitude closest to the middle as the target stable amplitude. Based on process experience, a fixed deviation range is set for this target stable amplitude, resulting in a central amplitude interval. This central amplitude interval is then compared with the aforementioned upper and lower limit candidate values. If the candidate upper limit is higher than the upper boundary of the central interval, the upper boundary of the central interval is selected as the final upper limit; if the candidate lower limit is lower than the lower boundary of the central interval, the lower boundary of the central interval is selected as the final lower limit. This forms a final preset threshold interval, which is stored in the system parameters with a defined upper limit amplitude value and a defined lower limit amplitude value, and remains unchanged thereafter. In actual operation, when the average amplitude and maximum amplitude of a batch both fall within the preset threshold interval, the system determines that the batch's fluctuation is normal and does not trigger the anomaly detection rule. When the maximum amplitude of any batch exceeds the preset upper limit or the average amplitude significantly deviates from the target stable amplitude, the system determines that the batch has abnormal fluctuations and immediately triggers the anomaly detection rule.When an anomaly detection rule is triggered, the system uses the aforementioned data synchronization protocol to perform data synchronization between sites. Specifically, within a fixed detection time window, the system uploads the concentration measurements, amplitude ranges, fluctuation parameters, environmental interference factors, and current supply adjustment records of all pretreatment sites to the same data recording unit, aligning them according to a unified timestamp and site number. This synchronized and organized data set is defined as the detection basis in this implementation. The detection basis explicitly records the concentration level, amplitude level, and corresponding supply and environmental conditions of each site for each time period. The data source is unique and synchronized with the actual situation on-site. After the detection basis is constructed, the system adopts the detection basis... By comparing historical trends, the system calculates the difference between current and historical data. In this implementation, historical data is defined as the filtered amplitude range and average concentration values ​​recorded at the same pretreatment station and time period from the 30 most recent batches of data that were deemed normal during the process validation phase and early stages of formal production. For each station and time period, the system first takes the arithmetic mean of the amplitude ranges of these 30 batches of normal data to obtain the historical amplitude reference value for that station and time period. Then, it takes the arithmetic mean of the average concentration values ​​of these 30 batches of normal data to obtain the historical concentration reference value for that station and time period. Finally, the system subtracts the filtered amplitude of the current batch at that station and time period from the historical amplitude reference value. The absolute value of the difference is taken. The average concentration of the current batch at this site and time period is subtracted from the historical concentration reference value, and the absolute value of the difference is taken. These two types of differences constitute the trend comparison difference for this site and time period. In this embodiment, the trend comparison difference is a set of two specific values: one is the absolute value of the amplitude difference, and the other is the absolute value of the concentration difference. After obtaining the trend comparison difference, the system combines it with interference analysis to determine the distribution anomaly parameters. In this embodiment, interference analysis refers to the analysis process of the relationship between the trend comparison difference and environmental interference factors. For each time period and each site, the system checks whether all environmental interference factors for that time period are within the aforementioned normal range. When all environmental interference factors are within the normal range and there is a significant trend difference, the system determines that the trend difference is entirely caused by the process itself. The system records the corresponding absolute value of the amplitude difference, the absolute value of the concentration difference, the time location of the time period, and the site number as a distribution anomaly parameter. When any environmental interference factor exceeds the normal range and the trend difference is also significant, the system marks this difference as an environmentally related difference and does not include it in the distribution anomaly parameter. After the above screening, the distribution anomaly parameter is defined in this embodiment as a set of multiple records. Each record contains a time location, a site number, an amplitude difference value, and a concentration difference value.After obtaining the distribution anomaly parameters, the system calculates the anomaly level by statistically analyzing these parameters. Specifically, it calculates the maximum value of the amplitude difference among all distribution anomaly parameter records within a batch to obtain the maximum anomaly amplitude for that batch. Simultaneously, it counts the number of time periods with distribution anomaly parameter records and the number of sites involved. Then, it compares the maximum anomaly amplitude, the number of abnormal time periods, and the number of abnormal sites with a pre-set grading standard established during the process validation phase. This grading standard is determined during the process validation phase through statistical analysis of multiple batches of test data with intentionally altered supply volumes. Specifically: when the maximum anomaly amplitude is below a certain fixed value, the number of abnormal time periods does not exceed 10% of the total number of time periods, and only one site is involved, the anomaly level is classified as Level 1. When the maximum anomaly amplitude is between the first and second fixed values, or the percentage of abnormal time periods is between 10% and 30%, or the number of sites involved is between 2 and 3, the anomaly level is classified as Level 2. When the maximum anomaly amplitude is above the second fixed value, or the percentage of abnormal time periods exceeds 30%, or the number of sites involved is greater than 3, the anomaly level is classified as Level 3. The second fixed value, along with the time period percentage limit and the number of stations, are determined during the process validation phase and written into the system parameters as fixed values, and are not changed arbitrarily. Based on the determined degree of anomaly, the system adjusts the supply volume to perform dynamic calibration to balance the distribution. Specifically, when the degree of anomaly for a batch is Level 1, the system only makes a small adjustment to the chemical liquid supply volume at the stations involved in the distribution anomaly parameter record. When the amplitude anomaly is manifested as an excessively large amplitude range at the station, the system reduces the current supply volume at that station by 5%. When the amplitude anomaly is manifested as an excessively small amplitude range, the system increases the current supply volume at that station by 5%. After adjustment, the new supply volume is maintained for a fixed period of time. During this period, concentration data is re-collected and the above amplitude calculation, filtering, fluctuation parameter, and trend comparison difference calculation process are repeated to verify whether the amplitude range falls back into the preset threshold range. When the degree of anomaly for a batch is Level 2, the system synchronously adjusts the supply volume at each station involved in the distribution anomaly parameter record, increasing or decreasing the supply volume at these stations by 10%, and re-collecting and calculating over a longer observation period until the amplitude range and trend comparison difference return to the allowable range.When the anomaly level of a batch is Level 3, the system performs significant adjustments at all sites involved in the anomaly, increasing or decreasing the supply by 15%. Concentration data is continuously collected over multiple consecutive observation periods, and the system repeatedly executes a cycle of "calculating the amplitude range, filtering environmental noise, calculating fluctuation parameters, comparing with the threshold range, calculating trend comparison differences, and updating distribution anomaly parameters and anomaly level" until the filtered amplitude range of all sites falls within the preset threshold range, and the corresponding distribution anomaly parameter records are compressed to a level where the maximum anomaly amplitude is below the first fixed value, the proportion of anomaly time periods is below 10%, and only one site is involved. At this point, the system saves the filtered amplitude range, average concentration value, and statistical results of amplitude and concentration differences between sites for that batch at each site and time period as stable characteristics of the concentration distribution. In this embodiment, stable characteristics of the concentration distribution are defined as: maintaining a stable amplitude range, average concentration within the preset threshold range, and a stable, small-deviation concentration distribution pattern between sites and time periods. This characteristic is used for historical trend reference and process stability evaluation of subsequent batches.

[0024] S3 includes: extracting concentration anomaly data around the marked points from the real-time monitoring system; adjusting the data timing based on equipment latency parameters through response delay correction and integrating environmental interference factors to obtain preliminary timing deviation analysis results; comparing the timing deviation analysis results with historical trends; if the deviation exceeds a preset threshold, triggering distribution pattern identification; obtaining potential distribution patterns by comparing the current deviation with historical patterns; selecting key locations based on potential distribution patterns; optimizing and integrating site data synchronization through supply adjustment and calculating quantitative values ​​to determine the anomaly degree quantification parameters; obtaining stable distribution balance characteristics through anomaly degree quantification parameters to determine key locations that need adjustment.

[0025] In this embodiment, after the degree of concentration distribution anomaly between batches has been obtained in step S2 and one or more batches have been identified as having anomalies, the system first reads the marker point information related to these anomalies one by one from the real-time monitoring system. In this embodiment, the marker point is a sampling location where the concentration deviates from the aforementioned concentration distribution characteristics between batches by a preset difference in spatial or temporal location. Each marker point corresponds to a fixed time window centered on the time of that marker point in the real-time monitoring system. In this embodiment, this window is set to extend forward and backward by a number of sampling times, for example, 10 sampling times forward and 10 sampling times backward. The real-time monitoring system stores the concentration measurement value at each sampling time, the spatial identifier of the measurement location, and the corresponding environmental interference factor data within this time window. In this embodiment, the environmental interference factors include temperature, humidity, and equipment operating status. These data have been recorded in the aforementioned steps in a manner synchronized with the concentration. When reading, the system arranges the concentration values ​​of all sampling times within the time window in chronological order. This ordered set of concentration data is defined as concentration anomaly extraction data. Since there is a fixed flow time difference between the chemical liquid flowing from the supply adjustment position to the corresponding measurement position, and this time difference varies between different pretreatment stations, during the equipment commissioning phase, tracer liquid is injected into the supply pipeline for each pretreatment station and each measurement position. The start time is recorded when injection begins at the supply end, and the arrival time is recorded when a clear increase in concentration is detected at the measurement position. The arrival time is subtracted from the start time to obtain the equipment delay parameter for that position. The equipment delay parameter is recorded as a definite value in seconds and stored in the real-time monitoring system. In actual operation, when the system acquires concentration anomaly extraction data for a certain marker point, the equipment delay parameter corresponding to the measurement position to which the marker point belongs is read out, and the time stamps of all sampling moments within that time window are uniformly corrected. That is, the original recorded time of each sampling moment is subtracted from the corresponding equipment delay value, and the corrected time is used as the new analysis time. This allows the chemical liquid supply adjustment moment and the concentration change moment to be aligned on the time axis, completing the response delay correction.After time correction, the system maps the temperature, humidity, and equipment operating status data for each sampling moment to the corrected time. When all environmental interference factors at a given sampling moment fall within the normal range determined statistically during the process verification phase, that sampling moment is marked as an environmentally stable moment. When any environmental interference factor exceeds the normal range, that sampling moment is marked as an environmentally abnormal moment. The normal range is determined during the process verification phase by continuously recording multiple batches of data under no-load or stable load conditions and statistically analyzing the maximum and minimum values ​​of temperature and humidity, as well as their corresponding natural concentration fluctuation ranges. Ultimately, the range where natural concentration fluctuations are still within the normal range is selected. The maximum values ​​of temperature and humidity within acceptable ranges are used as the normal upper limit, while the minimum values ​​of temperature and humidity within acceptable ranges for natural concentration fluctuations are selected as the normal lower limit. The continuous operation of the equipment without start-up or shutdown is defined as the normal state. These values ​​and states are stored in the system parameters as the normal range for environmental interference factors. Subsequently, within this time window, the system selects several consecutive sampling points before the marked point time, all of which are environmentally stable moments. For example, it selects five environmentally stable moments before the marked point, calculates the arithmetic mean of the concentration values ​​at these five moments, and defines this average value as the baseline concentration near the marked point. The time window is then defined as... The concentration values ​​at all stable environmental moments are sequentially subtracted from the baseline concentration to obtain the concentration deviation value at each stable moment. These deviation values ​​are then arranged in chronological order to form a deviation sequence, excluding deviations corresponding to abnormal environmental moments from subsequent statistics. This results in a deviation sequence that only reflects the variation of process factors over time. The system calculates multiple indicators for this deviation sequence. First, it finds the maximum absolute value of the deviation in the sequence and uses this maximum value as the peak deviation value at that marker point. Then, it finds the minimum non-zero absolute value of the deviation and records it as the minimum effective deviation. Simultaneously, it counts the continuous time length for the absolute value of the deviation to be greater than a certain fixed minimum value. During the process verification phase, the threshold value is determined by using the value of the deviation in several batches of normal data that is small enough not to affect the coating quality. The length of time that the absolute value of the deviation is continuously greater than this small value is the deviation duration of the marker point. Then, the start time and end time of the deviation from near zero to near the peak value and the time of the deviation falling back to near zero are recorded. Finally, the set of indicators such as the deviation peak value, deviation duration, deviation start time, deviation end time and deviation fall time are defined as the preliminary time-series deviation analysis result of the marker point. The result is stored in the real-time monitoring system in the form of a defined numerical set and corresponds one-to-one with the marker point.After obtaining the preliminary time-series deviation analysis results, the system performs historical trend comparison for each marker point. To this end, during the process verification phase and early production, the system has extracted deviation sequences from a large number of batches judged as normal or optimized through the aforementioned steps, finding those with the same spatial and temporal positions as the current marker point. These deviation sequences are aligned according to the time scale. For each time scale, the system calculates the average deviation and standard deviation of all normal batches at that scale, and defines the entire curve composed of the average deviation as the historical reference deviation curve for that position. Simultaneously, the system records the maximum peak deviation value and the longest deviation duration among the normal batches. The maximum value and the average deviation are used as the upper limit of historical normal fluctuations. During actual comparison, the system aligns the deviation sequence of the current marker point with the same time scale as the historical reference deviation curve. For each time scale, the absolute value of the difference between the current deviation and the reference deviation is calculated to obtain a difference sequence. Then, the maximum value, the average value, and the duration for which the absolute value of the difference is continuously greater than a certain fixed minimum value are calculated for the difference sequence. These three values ​​are used as the time-series deviation index for the current marker point. The preset threshold is determined in this embodiment as follows: During the process verification stage, the system also performs the above steps on several normal batches. By comparing historical trends, the maximum, average, and duration of differences in these normal batches were obtained. Statistical analysis was conducted to select the maximum, average, and duration of these differences from these normal batches. Combined with the coating quality evaluation results, a critical point was identified where a significant quality decline occurred after the difference index exceeded a certain combination of values. The maximum, average, and duration of the difference corresponding to this critical point were used as preset thresholds. These three values ​​were recorded as the upper limit of the threshold range in a fixed numerical format. Typical values ​​in normal batches where the difference was close to zero but still exhibited measurable deviation were used as the threshold range. The lower limit; in actual operation, when any time-series deviation of the current marker point deviates from the index by more than the corresponding preset threshold upper limit, the system determines that the deviation of the marker point is abnormal and immediately triggers the distribution pattern recognition process; the distribution pattern recognition process is carried out in this embodiment according to the following steps: the system has classified a large number of different types of deviation sequences during the process verification stage and summarized several typical distribution patterns. Each distribution pattern is composed of a description of the shape of the deviation changing over time, such as rapid rise in the early stage followed by stabilization, peak shape in the middle stage, slow rise throughout the process followed by slow fall, etc., and the standard deviation curve corresponding to each pattern is stored in the historical pattern library;When a marker triggers distribution pattern recognition, the system aligns the deviation sequence of that marker with the standard deviation curve of each historical pattern along the time dimension. For each time scale, it calculates the absolute value of the difference between the current deviation and the reference deviation of that pattern. It then sums all the absolute values ​​of these differences within a time window to obtain the total difference between the marker and the pattern. Alternatively, it averages all the absolute values ​​of these differences to obtain the average difference. The system then sorts all patterns by their total or average differences, selecting the pattern with the smallest difference as the first potential distribution pattern, the pattern with the second smallest difference as the second potential distribution pattern, and records the specific values ​​of the current deviation's difference among these patterns. The system also names the pattern with the smallest difference. Information such as the time range of the deviation peak and the duration of the deviation are combined and recorded as the potential distribution pattern of the marker point. This allows for explicit numerical calculations to categorize the current deviation into one of a limited number of known patterns. After obtaining the potential distribution pattern for each marker point, the system selects key points from all marker points in the same batch. In this implementation, key points are defined as marker points with large deviation peaks, long deviation durations, and high overlap in deviation occurrence times with other marker points. The system first groups the marker points according to their potential distribution patterns. Within each group, the system counts the deviation peak value, deviation duration value, and the number of times the deviation time overlaps with other marker points for each marker point. Within each group, the deviation peak value is multiplied by the deviation duration and... Multiplying by the number of overlapping points yields an impact value. This impact value is then sorted among all markers within the group. The markers with the highest impact values ​​are selected as candidates for key points in that group. All key point candidates from all groups are merged to form the initial set of key points for that batch. Subsequently, without changing the overall process parameters, the system performs supply adjustment trials on the supply path of each key point to obtain a quantitative value for quantifying the degree of anomaly. Specifically, the system sequentially selects each point in the initial set of key points, reads the chemical liquid supply volume used by the pretreatment station at that point in the current time period, and increases or decreases this supply volume by a fixed percentage. In this embodiment, this percentage is set to 5%. The system maintains this new supply for a fixed observation period, which in this embodiment is set to several sampling cycles, such as 20 sampling cycles. Before each supply adjustment, the system first performs a site data synchronization operation at all preprocessing sites, that is, unifies the time base of each site and records the concentration data and environmental interference factor data of each site at the same time, ensuring that the data collected during the observation period has the same time scale across all sites. Before the supply adjustment, the system first calculates the deviation sequence of the key point and other markers in the same potential distribution pattern within a complete time window, and calculates the sum of the absolute values ​​of all deviations as the total deviation before adjustment, and uses the total deviation as the initial deviation base of the key point.After the supply adjustment is completed, the system again calculates the deviation sequence of the key point and related markers within the same time window using the aforementioned method, and obtains the sum of the new absolute values ​​of the deviations as the total deviation after adjustment. The total deviation before adjustment is subtracted from the total deviation after adjustment to obtain the deviation reduction. The ratio of this deviation reduction to the total deviation before adjustment is defined as the deviation improvement ratio corresponding to the key point, and this deviation improvement ratio is used as the first type of quantification value. Simultaneously, the changes in the total deviation before and after adjustment are recorded for other markers in the potential distribution pattern to which the key point belongs. The ratio of the sum of the deviation reductions for these related markers to the original total deviation is used as the second type of quantification value. This constitutes a set of quantified response values ​​for each key point under a single supply adjustment trial operation. The system sequentially performs the aforementioned supply increase and supply decrease trial operations on all points in the initial set of key points, recording the quantified value sets for each direction. After obtaining the quantified values ​​for all key points, the system determines the anomaly degree quantification parameters based on these values. In this embodiment, the anomaly degree quantification parameters include three defined values: the first value is the ratio of the initial deviation peak value of the key point to the aforementioned upper limit of the preset threshold for time-series deviation; the second value is the absolute value of the deviation improvement ratio; and the third value is the absolute value of the reduction ratio of the total deviation of the relevant indicator points. These three values ​​are determined by... The aforementioned calculation method is directly derived from measured data and does not include undefined parameters. During the process verification phase, the system uses multiple batches of test batches with intentionally introduced anomalies of varying severity to statistically analyze the coating quality when the three values ​​fall into different combination ranges. Value combinations that can be restored to the target quality level are categorized as stable distribution equilibrium characteristics, and the corresponding ranges are fixed in the form of numerical intervals. For example, a combination where the first value does not exceed a certain upper limit, the second value is greater than a certain lower limit, and the third value is greater than a certain lower limit is defined as reaching an equilibrium state. These three ranges are written into the system parameters as stable distribution equilibrium characteristics. In actual operation, when a batch completes supply adjustments to key locations… After the initial trial, the system compares the three anomaly quantification parameters of each key point with the corresponding numerical ranges in the stable distribution equilibrium characteristics. When all three values ​​of a key point fall within the corresponding range, the system determines that the key point has reached a stable distribution equilibrium state under the current supply conditions and is no longer included in the list of key points that must be adjusted. When any value of a key point exceeds the corresponding range, the system marks the key point as one that still needs adjustment and will prioritize calculating and applying correction values ​​for these key points in the subsequent S4 step. This allows the system to obtain the stable distribution equilibrium characteristics through the anomaly quantification parameters and determine the key points that need adjustment accordingly.

[0026] S4 includes obtaining a deviation distribution pattern from key point deviations, fusing the deviation distribution pattern with historical data weights to obtain a preliminary weighted deviation value, wherein the fusing is performed by multiplying the historical data weights by the sum of the element values ​​of the deviation distribution pattern using a weighted average method; for the preliminary weighted deviation value, a deviation calculation formula is used, combined with the correction value generation logic to determine the intermediate correction deviation, wherein the deviation calculation formula is D=(PH)×W, where D represents the deviation value, P represents the current point value, H represents the historical average value, and W represents the weight. After calculation using this formula, the correction value generation logic is adjusted to the intermediate correction deviation; obtaining the compensation range limit for fusing the intermediate correction deviation, if it exceeds a preset threshold, the point weights are adjusted to obtain a post-limited deviation pattern, wherein the fusing is performed by summing the components of the intermediate correction deviation within the compensation range limit; calculating a targeted correction value using the post-limited deviation pattern, and determining the control parameters to obtain the basis for equipment control, wherein the calculation is performed by calculating the average value of the post-limited deviation pattern to obtain the targeted correction value.

[0027] In this embodiment, after completing the aforementioned steps, the system has obtained the key points and their deviation values ​​corresponding to each batch of wafers based on time-series deviation analysis and potential distribution pattern recognition. In this embodiment, the key point deviation is defined as the difference between the current actual concentration value at the key point and the target concentration value in the stable distribution equilibrium feature. This difference is divided into several discrete time periods, each time period corresponds to a signed deviation value, and the deviation values ​​of all time periods are arranged in chronological order to form the deviation sequence of the key point. The system first obtains the deviation distribution pattern from all key point deviations. In this embodiment, the deviation distribution pattern is defined as the set of deviation sequences of each key point in each time period along the entire process path. Specifically, the system sorts all key points in the batch according to the batch number. For each key point, the deviation values ​​of each time period in the deviation sequence of the point within the preset analysis time window are extracted in sequence to form the deviation sub-pattern of the point. Then, the deviation sub-patterns of each key point are sequentially spliced ​​together according to the process sequence to obtain a one-dimensional numerical array composed of multiple deviation values. Each element in the array corresponds to a specific key point and a specific time period. This array is the deviation distribution pattern corresponding to the wafers in this batch. In this implementation, historical data weights are defined as a set of values ​​used to characterize the degree of influence of historical data on the current deviation distribution. During the process verification stage, the system selects 30 historical batches that were judged to be normal in quality inspection. For each element position in these batches that corresponds to the current deviation distribution pattern, the average value of the actual absolute deviation of that position in the 30 batches is calculated. Positions with larger absolute deviation values ​​are assigned larger weight values, and positions with smaller absolute deviation values ​​are assigned smaller weight values. Then, the weight values ​​of all positions are summed, and the weight value of each position is divided by the sum of the weights. The result is used as the final historical data weight of that position, thereby ensuring that the sum of the weights of all positions is equal to 1. After obtaining the deviation distribution pattern from the key point deviations, the system performs a fusion calculation on the deviation distribution pattern using historical data weights to obtain a preliminary weighted deviation value. The specific calculation process is as follows: for each element in the deviation distribution pattern, the historical data weight corresponding to that element's position is taken. The deviation value of that element is multiplied by the corresponding historical data weight to obtain the weighted deviation value for that position. Then, the weighted deviation values ​​for all positions in the deviation distribution pattern are summed, and this summation is divided by the sum of all historical data weights (which equals 1 after the aforementioned normalization process) to obtain a definite value. This value is the preliminary weighted deviation value for this batch under the current process conditions. In this embodiment, the preliminary weighted deviation value is defined as a single representative deviation value obtained by fusing the deviation distribution pattern and historical data weights according to a weighted average rule, used to summarize the overall deviation degree of the current batch relative to the historical stable state.For the initial weight deviation value, the system uses a deviation calculation formula and combines it with the correction value generation logic to determine the intermediate correction deviation. In this embodiment, the deviation calculation formula is defined as follows: the deviation value is equal to the product of the difference between the current point value and the historical average value and the weight coefficient. The current point value refers to the actual average concentration value at a certain key point after response delay correction and environmental interference filtering during the processing of this batch of wafers. The historical average value refers to the arithmetic mean of the concentration values ​​recorded at the same key point and the same time period in the aforementioned 30 batches of historical normal batches. In this embodiment, the weight coefficient is taken as the product of the historical data weight corresponding to the key point and the importance coefficient of the point in the stable distribution balance feature. The importance coefficient in the stable distribution balance feature is determined in the process verification stage through statistical analysis of the impact of different point deviations on the coating quality. Points with a greater than 1 are assigned a coefficient greater than 1, and points with a smaller impact on the quality are assigned a coefficient less than 1 but greater than 0. For each key point, the system first subtracts its historical average value from the current value to obtain the difference. Then, it multiplies this difference by the corresponding weighting coefficient to obtain the deviation value. All deviation values ​​for key points are arranged in process order to form a deviation value sequence. In this embodiment, the correction value generation logic is defined as a rule that converts deviation values ​​into intermediate correction deviations used for actual adjustment of the direction and amplitude of chemical liquid supply. For each key point, the system determines the sign of its deviation value. When the deviation value is greater than 0, it indicates that the current value is higher than the historical average value. The system multiplies the deviation value by -1 and uses the result as the intermediate correction deviation for that point, indicating the need for compensation towards lower concentration. When the deviation value is less than 0, it indicates that the current value is lower than the historical average value. The system uses the absolute value of the deviation value as the intermediate correction deviation for that point, indicating the need for compensation towards higher concentration. When the deviation value is equal to 0, the system sets the intermediate correction deviation to 0. After this logical processing, the system obtains an intermediate correction deviation sequence corresponding one-to-one with the key points. Each element is a specific value. This sequence is used for subsequent compensation range limitation and weight adjustment. In this embodiment, the compensation range limitation is defined as the maximum allowable correction range that can be applied in a single instance without causing process oscillations or reverse quality deterioration. It is determined as follows: During the process verification stage, 10 test batches are selected. By controlling the chemical solution supply to change the concentration of the key points at different levels, the adjustment amount corresponding to the intermediate correction deviation is gradually increased or decreased in each test batch. The coating thickness uniformity and defect density are detected. When it is found that the coating quality deteriorates significantly after the correction amplitude exceeds a certain value, this value is taken as the compensation upper limit for that point. The value slightly lower than this value that still ensures quality is taken as the compensation lower limit. Finally, a pair of upper and lower limit values ​​are formed for each key point, which serve as the compensation range limitation for that point.In this embodiment, the preset threshold is defined as a threshold value used to determine whether the point weight needs to be adjusted. The system calculates the absolute value of the intermediate correction deviation for each key point in the same batch of test data. When the absolute value of the intermediate correction deviation approaches the compensation limit for multiple batches, the corresponding point exhibits significant instability. The average absolute value of the intermediate correction deviation under this sustained high deviation state is used as the preset threshold, and this value is stored in the system parameters for subsequent judgment. After obtaining the compensation range limit, the system performs a compensation range fusion operation on the intermediate correction deviation. Specifically, the absolute value of the intermediate correction deviation for each key point is taken and compared with the compensation limit for that point. When the absolute value of the intermediate correction deviation is less than or equal to the compensation limit, the intermediate correction deviation is directly retained as the initial component of that point in the limited deviation mode. When the absolute value of the intermediate correction deviation is greater than the compensation limit, the intermediate correction deviation is replaced with a value of the same sign and an absolute value equal to the compensation limit as the initial component of that point in the limited deviation mode. Simultaneously, the weight value of that point is reduced by a fixed proportion. In this embodiment, the original weight value of the point is multiplied by 0.5 and then updated as the new point weight. In this implementation, the point weight is defined as the numerical value representing the degree of influence of each key point on the overall correction value when calculating the targeted correction value later. Initially, this value is the product of the aforementioned weight coefficient and the point's importance coefficient. When a limit is exceeded, the weight is reduced according to the aforementioned rules. When the absolute value of the intermediate correction deviation of a key point exceeds a preset threshold, the system not only performs the aforementioned truncation process but also multiplies the point weight by 0.5 again to further reduce the point's influence on the overall correction value, thereby preventing extreme anomalies from dominating the overall adjustment direction. Subsequently, within the compensation range limit, the system performs a summation operation on each component of the intermediate correction deviation. Specifically, the initial component of all key points in the post-limited deviation mode is multiplied by the corresponding point weight to obtain the weighted correction component of each point. All weighted correction components are then summed to obtain a total weighted correction sum. Simultaneously, all point weight values ​​are summed to obtain the total point weight. Finally, the weighted correction sum is divided by the total point weight to obtain a representative value of the post-limited deviation mode. This value, along with the weighted correction components of each point, constitutes the post-limited deviation mode. In this embodiment, the restricted deviation pattern is defined as the set of intermediate corrected deviations after compensation range restriction and point weight adjustment, which is the result of pruning and reweighting the original intermediate corrected deviation sequence.The system calculates a targeted correction value by limiting the deviation mode. Specifically, the weighted correction component of each key point in the limiting deviation mode is taken as a component, all components are added together and divided by the number of key points to obtain a certain value, which is the targeted correction value. In this embodiment, the targeted correction value is defined as a global correction amount that directly affects the chemical liquid supply parameters. Its sign indicates whether the overall adjustment needs to be made in the direction of increasing or decreasing the concentration, and its absolute value indicates the overall adjustment range. In this embodiment, the control parameters are defined as a set of parameters used to control the chemical solution preparation and supply behavior of the pretreatment equipment. These parameters include at least the target concentration setpoint, flow rate setpoint, and ratio setpoint for each treatment station. The equipment control basis in this embodiment is defined as a set of control parameter target values ​​determined by targeted correction values ​​and the location weights of each key point. After obtaining the targeted correction values, the system distributes these values ​​among the stations according to their location weights. For each treatment station, it calculates the required concentration and flow rate corrections. These corrections are then used to add to or subtract from the original control parameters to obtain new control parameter target values. The aforementioned data synchronization protocol is then invoked to synchronously write these new control parameter target values ​​into the control units of all relevant pretreatment equipment, achieving control parameter synchronization and thus obtaining a unified equipment control basis. This ensures that each device is adjusted according to the same targeted correction value during subsequent operation.

[0028] S5 includes obtaining correction value data through the chemical solution dispensing module, performing format verification on the input correction values ​​to obtain a standardized correction dataset; based on the standardized correction dataset, calling the actuator interface to transmit the correction values ​​to the chemical solution concentration adjustment unit to determine the concentration adjustment execution command; encrypting the execution command using a data synchronization protocol to obtain the encrypted command data packet and determining the integrity of the command data packet; if the command data packet is complete, distributing the encrypted command data packet to each station through the network channel to obtain reception confirmation information from each station; parsing the status field in the reception confirmation information from each station to determine whether each station has synchronized the correction value data; if a station has not synchronized the correction value data, resending the encrypted command data packet through a backup channel to obtain secondary confirmation information from the station; updating the preparation status record of each station based on the secondary confirmation information to determine whether all stations have reached a consistent status.

[0029] In this embodiment, after completing step 5, the correction value data calculated for each key point has been obtained. In this embodiment, the correction value data is defined as a data set consisting of several records. Each record contains at least a site number field, a parameter type field, and a correction value field. The site number field is an integer number representing the chemical liquid treatment site. The parameter type field is an enumerated value representing the object to which the correction value is applied, including three types: concentration setting value, flow rate setting value, and ratio setting value. The correction value field is the value that the corresponding parameter needs to be increased or decreased. In this embodiment, the unit is uniformly uniformly set as volume fraction percentage or flow rate unit. The chemical solution dispensing module periodically reads the entire set of correction value data from the output buffer of step 5, and first performs format validation on each record. Format validation includes four parts: field integrity validation, data type validation, numerical range validation, and precision unification processing. Field integrity validation checks whether each record simultaneously contains the site number field, parameter type field, and correction value field. If any field is missing, the record is marked as invalid and removed from the correction value data set. Data type validation checks whether the site number field is an integer, whether the parameter type field is one of three predefined values, and whether the correction value field is numeric and does not contain illegal characters. If any of these conditions are not met, the record is removed. Numerical range validation checks the correction value field... The absolute value is compared with the predetermined maximum allowable correction range for a single instance. In this embodiment, during the process verification stage, a set of maximum correction ranges that will not cause process oscillations is obtained by changing the concentration and flow rate of multiple test batches. This maximum correction range is used as the upper limit, and its opposite is used as the lower limit. If the correction value exceeds the upper or lower limit, it is truncated according to the upper or lower limit. The precision unification process is to round the correction value according to the predetermined decimal place requirements. In this embodiment, the concentration setting value is retained to 3 decimal places, the flow rate setting value is retained to 2 decimal places, and the ratio setting value is retained to 3 decimal places. All the legal records obtained after the above processing form a standardized correction dataset, which is sorted by station number and parameter type for subsequent sequential retrieval.Based on the standardized correction dataset, the system calls the actuator interface to convert each record into a specific execution instruction and transmit it to the chemical solution concentration adjustment unit. The actuator interface calculates the correction value according to fixed rules. Specifically, it first reads the target concentration, target flow rate, or target ratio currently in effect at the corresponding site, adds or subtracts the target value from the correction value to obtain a new target setting value, and then determines the execution time of the adjustment process based on the magnitude of the correction. In this embodiment, when the absolute value of the correction is less than or equal to the preset first amplitude threshold, the execution time is set to several sampling periods. When the absolute value of the correction is greater than the first amplitude threshold but less than or equal to the second amplitude threshold, the execution time is set to twice the previous level. When the absolute value of the correction is greater than the second amplitude threshold, the execution time is set to three times the previous level. The first amplitude threshold and the second amplitude threshold are determined during the process verification stage by observing the smoothness of concentration changes and wafer quality fluctuations under different correction gradients, and are stored in the parameter table in a defined numerical form. After obtaining the new target setting value and the corresponding execution time, the actuator interface generates an execution instruction for each correction record. The execution instruction includes fields such as station number, parameter type, new target setting value, execution time, and instruction number. The instruction number is a monotonically increasing integer used to distinguish instructions from different batches.

[0030] Subsequently, a data synchronization protocol is used to encrypt the aforementioned execution instructions and encapsulate them into instruction data packets. In this embodiment, the data synchronization protocol is defined as a set of fixed communication rules to ensure that all stations receive and execute the same instructions within the same time window. Specifically, the following steps are taken: First, the fields in the execution instructions are arranged and concatenated into an original instruction sequence according to a predetermined order. Then, header information is added before the sequence, including the instruction number, instruction generation time, and the total number of target stations. A reserved checksum and digest area are added to the end of the sequence, forming a complete original instruction data structure. Then, the original instruction data structure is segmented and encrypted using a symmetric encryption key uniformly distributed during the equipment debugging phase. The encryption process involves dividing the data structure into segments of a fixed length. The system divides the data packet into several segments. For each segment, several rounds of numerical permutations and mixing operations are performed to obtain ciphertext segments. All ciphertext segments are then concatenated in their original order to form an encrypted instruction data packet. Next, the system calculates a checksum and a digest value based on the encrypted instruction data packet. The checksum is calculated by summing the values ​​of all bytes in the data packet, then taking the remainder of this sum with a fixed integer (set to 65536 during the system design phase). This remainder is written as the checksum in the checksum area. The digest value is calculated by dividing the data packet into several segments of fixed length, performing multiple rounds of addition, subtraction, and shift operations on each segment, and then combining the results of each segment in order to obtain a fixed-length digest value, which is written into the digest area. After the system completes the above operations, it performs an integrity check on the instruction data packet. Specifically, it recalculates the checksum of the current data packet and compares the new checksum with the checksum stored in the data packet. If they are equal, it is considered that no error occurred during the generation of the data packet. At the same time, it compares the digest value recorded in the digest area with the new digest value obtained by the digest algorithm of the current data packet. If they are equal, it is considered that the data content has not been tampered with. Only when both the checksum and the digest value match is the data packet considered complete. Otherwise, the data packet is discarded and the error reason is recorded in the log. In this embodiment, incomplete data packets are not sent out.

[0031] When a command data packet is deemed complete, the system distributes the encrypted command data packet to each station via the network channel. In this embodiment, the network channel is defined as the main communication link connecting the chemical liquid distribution module and each processing station. This link has undergone bandwidth and latency testing during the system deployment phase, and a transmission priority has been assigned to each data packet. The system sends command data packets via the network channel either through broadcast or in order of station number. Simultaneously, a receive timer is started, and an acknowledgment timeout is set. In this embodiment, the acknowledgment timeout is determined to be 3 seconds through multiple network performance tests, meaning that under most normal circumstances, the time from sending the data packet to receiving acknowledgment information from the station will not exceed 3 seconds. Upon receiving the encrypted instruction data packet, each station performs the decryption process using the same key and algorithm. After successful decryption, it verifies the data structure using checksums and digest values. If verification is successful, the instruction is written to the local execution buffer, and a reception confirmation message is generated. This message includes fields such as station number, received instruction number, decryption and verification result status value, and local configuration version number. The status value is a specific integer: 0 indicates successful reception and readiness for execution; 1 indicates decryption failure; 2 indicates integrity verification failure; and 3 indicates the instruction number is earlier than the current version and has been ignored. The configuration version number is an integer representing the batch of control parameters that have been applied at the current station. All stations return the reception confirmation message to the chemical liquid distribution module via the network channel. The module collects all reception confirmation messages returned by all stations within the confirmation timeout period. For stations that do not return confirmation messages within the timeout period, the system automatically generates a timeout confirmation record with a status value of 4.

[0032] Subsequently, the system parses the received confirmation information from each station. During parsing, it reads the station number, instruction number, and status value fields from each confirmation message. When the status value is 0 and the instruction number matches the currently issued instruction number, the system determines that the station has synchronized the correction value data, meaning it has received the data and is ready to perform concentration adjustments according to the current correction value. When the status value is any one of 1, 2, 3, or 4, the system determines that the station has not synchronized the correction value data and needs to enter the backup channel retransmission process. In this embodiment, the backup channel is defined as a second communication path relative to the network channel. This path is physically or logically isolated from the main channel during system deployment. Its bandwidth requirement can be slightly lower than that of the main channel, but its reliability requirement must not be lower than that of the main channel. For each station determined to be out of sync, the system retransmits an encrypted instruction data packet with the same instruction number using the backup channel and restarts the receive timer. In this embodiment, the confirmation timeout for the backup channel is set to twice the timeout for the main channel, i.e., 6 seconds, to ensure sufficient time to complete the transmission even when the backup link conditions are poor. After receiving the data packet through the backup channel, the station repeats the decryption, verification, and digest verification process, generating a secondary confirmation message. The structure of the secondary confirmation message is the same as the first confirmation message, but a value of 5 is added to the status field to indicate the reception result through the backup channel, signifying successful reception via the backup channel after the primary channel failed. The chemical liquid distribution module collects the secondary confirmation messages from all stations within the timeout period. For stations that have not yet received the secondary confirmation message, the system generates a secondary timeout record with a status value of 6.

[0033] Based on the secondary confirmation information, the system updates the preparation status records of each site and determines whether all sites have reached a consistent status. In this implementation, the preparation status record is defined as a data table used to track the current chemical solution preparation status of each site. Each record contains four fields: site number, most recently successfully issued instruction number, current configuration version number, and synchronization status. The most recently successfully issued instruction number is the instruction number recorded when the site's status value is 0 or 5. The current configuration version number is the version identifier converted from the instruction number. The synchronization status is an enumerated value, with 0 indicating no synchronization and 1 indicating synchronization. When processing the secondary confirmation information, for sites with a status value of 0 or 5, the system updates the most recently successfully issued instruction number in its corresponding preparation status record to the current instruction number, updates the current configuration version number to the new version number, and sets the synchronization status to 1. For sites with status values ​​of 1, 2, 3, 4, or 6, the synchronization status is maintained or updated to 0, and the cause and time of the anomaly are recorded in the system log. The system then scans all records in the configuration status log table. When it finds that the synchronization status of all stations is 1 and the current configuration version number is equal to the version number corresponding to this instruction number, it is determined that all stations have reached a consistent status and the data synchronization of this correction value is completed. If any station has a synchronization status of 0 or the current configuration version number of that station is behind the latest version number, it is determined that the consistency status has not yet been reached. In this embodiment, the system will output the list of these unsynchronized stations to the upper-level monitoring interface, and the manual investigation and handling will be carried out based on the fault records.

[0034] S6 includes obtaining specific data of deviation records through wafer surface treatment logs, comparing the deviation records with preset thresholds to determine whether the deviation is within an acceptable range; if the deviation records are within the preset threshold range, obtaining relevant device adaptation parameters through the system interface to determine the device adaptation update requirements; based on the device adaptation update requirements, calling the parameter loading module to transmit the adaptation parameters to the device control unit and obtaining status feedback after parameter loading; parsing the execution fields in the status feedback after parameter loading to determine whether the device has completed the adaptation parameter update process; if the device has completed the adaptation parameter update, obtaining real-time concentration data through the concentration status monitoring module to determine whether the concentration consistency meets the expected standard; generating a record file of optimization results based on the concentration consistency data, saving the optimization results through the data storage unit, and obtaining the saved index identifier; updating the relevant fields in the wafer surface treatment log based on the saved index identifier to determine whether the log data has been synchronized to the latest status.

[0035] In this embodiment, after completing the aforementioned steps, the concentration data and target concentration data of each wafer at each pre-processing station and processing time period are continuously recorded in the wafer surface treatment log. A deviation record is generated in the log for each processing step. In this embodiment, the deviation record is explicitly defined as the signed value obtained by subtracting the target concentration from the actual concentration, along with the corresponding wafer number, station number, timestamp, and process recipe number. These records are stored in chronological order in the wafer surface treatment log. During execution S6, the system first uses the wafer surface treatment log's retrieval function to read the specific data of all deviation records for the current batch, using the current batch number as the condition. The deviation value field in each deviation record is organized into a deviation data set according to station and time order, while retaining the corresponding wafer number and station number information to ensure that subsequent comparisons have a clear target. In this embodiment, the preset threshold is defined as the upper and lower limits of the allowable deviation. This range is determined in the process verification stage as follows: Several batches of qualified batches whose coating quality has passed the test are selected. For each pretreatment station and each type of process formula, the maximum, minimum and average values ​​of the deviation values ​​of these batches under the corresponding conditions are statistically analyzed. When the absolute value of the deviation is greater than a certain specific value, the uniformity of coating thickness or defect density deteriorates significantly. This specific value is taken as the upper limit of the allowable absolute value of the deviation. At the same time, based on the typical value of the deviation in normal batches that is close to zero but still has measurement error, the lower limit of the allowable absolute value of the deviation is set to a certain value that is greater than zero. This forms the preset threshold range of the deviation. The upper and lower limits of the range are recorded in the parameter table for different stations and different process formulas. After acquiring the deviation data set, the system performs a comparison operation on each deviation record in the set, comparing the absolute value of the deviation with the upper limit of the preset threshold range under the corresponding site and process formula. When the absolute value of the deviation is less than or equal to the upper limit and greater than or equal to the lower limit, the record is marked as "within acceptable range"; otherwise, it is marked as "out of acceptable range". Then, the system performs a statistical analysis on the marking results of all deviation records in the current batch. When all records are marked as "within acceptable range", the system determines that the deviation records of the batch are within the preset threshold range and enters the equipment adaptation and update process; otherwise, the batch is recorded as having unacceptable deviation, and subsequent adaptation parameter updates are not performed. Only the reason for the deviation exceeding the limit is recorded in the log. For batches with deviations recorded within a preset threshold range, the system reads the equipment adaptation-related parameters corresponding to the current process formula from the equipment through the system interface. In this embodiment, the equipment adaptation-related parameters include a set of parameters used to align the equipment status with the process target, such as the local target concentration fine-tuning amount, local flow fine-tuning amount, allowable response time, and internal control gain configured for each pretreatment station. These parameters are written into the equipment control unit during the equipment factory calibration and process introduction stages and are provided to the outside world in a structured form through the system interface for reading and writing functions.In this step, the system interface sequentially reads the currently effective adaptation parameter values ​​for each site according to the site number, and compares these values ​​one by one with the reference adaptation parameter range obtained from the previous optimization batches. The reference adaptation parameter range was determined through multiple batches of stable production data during the process verification phase, and a target value and allowable deviation range are set for each parameter. For each adaptation parameter, the system calculates the difference between the current parameter value and the target value. When the absolute value of the difference is greater than the upper limit of the reference deviation range, the parameter is marked as "needs adjustment," and the suggested adjustment direction and suggested adjustment magnitude are recorded. For example, when the local target concentration adjustment amount at a certain site is higher than the target value, it is recommended to adjust it to a certain percentage in the downward direction; when it is lower than the target value, it is recommended to adjust it to a certain percentage in the upward direction. When the absolute value of the difference is less than or equal to the upper limit of the reference deviation range, the parameter is marked as "remain unchanged." All parameters that need adjustment and their suggested adjustment directions and magnitudes together constitute the equipment adaptation update requirements. The update requirements are represented by a set of structured records, each containing the site number, parameter name, current value, target value, and suggested adjustment value. Based on the equipment adaptation and update requirements, the system calls the parameter loading module to load the adaptation parameters that need to be adjusted in batches. In this embodiment, the parameter loading module is responsible for transmitting the updated adaptation parameters to the equipment control unit and obtaining the parameter loading results. The specific process is as follows: For each parameter marked as "needs adjustment", the loading module calculates a new target value for the parameter based on the suggested adjustment value. This target value is equal to the original parameter value plus or minus the suggested adjustment value. When necessary, the calculation result is subjected to a limiting process, that is, the new target value is constrained within the parameter range allowed by the process. This range is determined through experiments on the stability of equipment operation during the process verification phase. Subsequently, the loading module packages each new target value with the corresponding site number and parameter name into a parameter update instruction, and sends it to the corresponding equipment control unit through the communication interface provided by the equipment control unit. When sending, a unique serial number and timestamp are attached to each instruction. After receiving the parameter update instruction, the equipment control unit performs a validity check and write operation, and returns a status feedback message after the write is completed. In this implementation, the status feedback message includes fields such as instruction sequence number, site number, execution result code, and execution completion time. The execution result code is a fixed value: 0 indicates successful writing and effectiveness; 1 indicates the parameter value exceeds the device's supported range, causing writing failure; 2 indicates the device is currently in a write-unwritable state; and 3 indicates other internal errors. After receiving the status feedback messages from each device, the parameter loading module aggregates these messages into a status feedback set after parameter loading.The system parses the status feedback set, reading the execution result code and related fields line by line. When the execution result codes of all parameter update instructions for a certain site are 0 and the execution completion time is no earlier than the start time of the current batch, the system determines that the site has completed the adaptation parameter update process and marks it as "adaptation complete" internally. When the execution result code of any parameter update instruction is not 0, the system determines that the adaptation of the site has failed, records the specific reason for the failure, and does not proceed to the subsequent concentration consistency confirmation process. For all sites marked as "adaptation complete", the system obtains real-time concentration data after parameter updates through the concentration status monitoring module. In this embodiment, the concentration status monitoring module continuously collects concentration values ​​from the concentration sensors of each site after the aforementioned calibration and filtering, and summarizes them by wafer batch within a fixed time window. After adaptation, the system performs statistical analysis on the concentration data of each batch of wafers by site and time within several sampling periods, calculating the concentration consistency index for each batch. In this implementation, concentration consistency is defined as the difference between the maximum and minimum concentration values ​​across all sites and key time periods, the average absolute concentration deviation, and the maximum difference in average concentration between sites. The system compares these indicators one by one with the expected standards used to characterize good concentration consistency during the process verification phase. These expected standards are determined by calculating the above consistency indicators from multiple batches of process-stable and quality-compliant reference batches and selecting the maximum allowable value. For example, the expected standards might be: the difference between the maximum and minimum concentration values ​​not exceeding a certain specific value; the average absolute concentration deviation not exceeding another specific value; and the maximum difference in average concentration between sites not exceeding a third specific value. If all concentration consistency indicators for the current batch are less than or equal to the corresponding expected standards, then the concentration consistency is considered to have met expectations. The system then generates a record file of the optimization results based on this concentration consistency data. The record file includes fields such as batch number, key parameter values ​​for each site before and after adaptation, concentration consistency indicators before and after adaptation, deviation record statistics, and whether the expected standards have been met. This record file is stored by the data storage unit. When saving record files, the data storage unit assigns a unique index identifier to each file. In this embodiment, the index identifier is a non-repeating integer number or a string obtained by concatenating a timestamp and a sequence number, used to quickly locate the corresponding record file in subsequent searches. After successful writing, the data storage unit returns the index identifier. The system uses the index identifier to update relevant fields in the wafer surface treatment log. The log reserves fields such as "recent optimization result index" and "log version number" for each batch. The "recent optimization result index" points to the corresponding optimization result record file in the data storage unit, and the "log version number" indicates the number of times the log for that batch has been updated.For the current batch, the system updates the "Recent Optimization Result Index" field to the index identifier returned by the data storage unit, increments the "Log Version Number" by 1, and updates the deviation status field to "Processed according to the latest adapted parameters." Then, it performs a consistency check on all relevant log records for the batch. When the check results show that all records have the same batch number, the "Recent Optimization Result Index" field is consistent, and the "Log Version Number" has been updated to the current version number, it is determined that the wafer surface treatment log data has been synchronized to the latest state. If some records are found to still retain the old index identifier or the old version number, these records are listed and the update operation is re-executed until all records correspond to the latest index identifier.

[0036] S7 includes obtaining the current chemical solution preparation status information through concentration consistency data, parsing key fields in the status information to determine the initial deviation of the preparation path; based on the initial deviation, obtaining environmental interference factor data, extracting corresponding influence weights from a pre-established interference factor database, and judging the intensity of the interference factor's effect on the preparation path; if the intensity of the interference factor's effect on the preparation path exceeds a preset threshold, obtaining the change pattern in multiple batches of processing through historical trend data to determine the adjustment direction of the preparation path in different time periods; for the adjustment direction, obtaining the operating status differences between multiple sites, extracting the system parameter update requirements from the status differences, and obtaining the priority ranking of parameter iteration updates; according to the priority ranking, using the iteration update module to adjust the system parameters step by step, obtaining the adjusted parameter configuration status, and judging whether the configuration meets the stability standards of multi-site operation; if the configuration meets the stability standards of multi-site operation, loading the updated parameters through the chemical solution preparation module, obtaining real-time operating data of the pretreatment stage, and determining the monitoring results of process stability; based on the monitoring results of process stability, generating optimized path records for multiple batches of processing, saving the path records through the data storage unit, and obtaining the saved identification information.

[0037] In this embodiment, based on the aforementioned steps, concentration consistency data for each batch of wafers at each pretreatment station and time period has been obtained. This concentration consistency data is defined in this embodiment as a set of three numerical indicators: the difference between the maximum and minimum actual concentration values ​​for the batch across all stations and key time periods; the average absolute deviation of the concentration at all sampling points relative to the target concentration; and the maximum difference between the average concentrations at each station. The system first reads these three indicators from the results generated in the previous step using the current batch number as an index, and then reorganizes them according to the preparation path. In this embodiment, the preparation path is defined as the path from the starting point of the chemical solution preparation to each pretreatment station. The specific delivery channels are defined, with each preparation path having a unique number in the system, and associated with one or more downstream processing stations. The system aggregates the concentration consistency data of stations belonging to the same path according to the path number, and calculates the difference between the maximum and minimum concentration values ​​of all stations within the path, the average absolute deviation of the concentration of all stations within the path, and the maximum difference in average concentration of stations within the path for each path, thus obtaining the path-level concentration consistency data for each preparation path. In this embodiment, the current status information of the chemical solution preparation is the aforementioned path-level concentration consistency data organized by path and the corresponding target concentration consistency standard. Key fields include path number, maximum and minimum concentration difference within the path, average absolute concentration deviation within the path, maximum average concentration difference at stations within the path, and target upper limit value under the corresponding process requirements. The system parses each of these key fields, comparing each path-level indicator with its corresponding target upper limit value. When an indicator is less than or equal to the target upper limit value, it is marked as qualified; when it is greater than the target upper limit value, it is marked as out of tolerance. Then, for each path, the number of items marked as out of tolerance and the absolute value of the out-of-tolerance are counted. These results are combined to define the initial deviation status of the preparation path for that path. The initial deviation status of the preparation path in this implementation method... The formula uses a defined deviation level and three deviation quantification values. The deviation level is divided into several levels according to the number of out-of-tolerance items and the maximum out-of-tolerance amount. It is determined by analyzing a large number of batch data during the process verification stage. For example, when all three indicators are within the target upper limit, the deviation level is 0. When only one indicator slightly exceeds the upper limit and the out-of-tolerance amount does not exceed the preset slight deviation value, the deviation level is 1. When two or more indicators exceed the upper limit or the out-of-tolerance amount of any indicator exceeds the preset larger deviation value, the deviation level is 2. The preset slight deviation value and the preset larger deviation value are obtained by statistical analysis of the relationship between coating quality and concentration consistency during the process verification stage and stored in the parameter table in a defined numerical form.After obtaining the initial deviation status of each preparation path, the system extracts environmental interference factor data recorded within the same time window as that batch from the operating database based on the deviation level and deviation quantification value. In this embodiment, the environmental interference factor data includes time-indexed numerical sequences such as ambient temperature and humidity in the preparation area, fluctuations in the liquid level of the preparation tank, fluctuations in the main supply pump outlet pressure, and changes in exhaust volume. These sequences are continuously collected by corresponding sensors and stored by timestamp during system operation. The interference factor database is a dataset pre-established during process verification and historical operation phases. For each environmental interference factor, the database records three items: first, the statistical range of the interference factor under normal operating conditions, including the upper and lower limits; second, the quantification value of the degree of concentration consistency deterioration caused by the interference factor deviating from the normal range; and third, the influence weight of the interference factor on different preparation paths based on the statistical results. In this embodiment, the influence weight is between 0 and... The system determines a value between 1 and 1, and normalizes the influence weights of all interference factors along the same path so that the sum equals 1. Based on the time window of the current batch, the system extracts the environmental interference factor data on the time axis corresponding to the preparation and pretreatment process of this batch. For each interference factor, the system calculates the absolute value of the difference between the average value within the segment and the center value of the normal range, and divides this difference by the allowable deviation of the normal range to obtain the interference deviation ratio. When the interference deviation ratio is less than 1, the factor is considered to have a small degree of deviation; when it is greater than 1, the factor is considered to have a large degree of deviation. Then, the system reads the influence weight of each interference factor on the current preparation path from the interference factor database, multiplies the deviation ratio of the factor by the influence weight to obtain the contribution intensity of the factor to the path, and then sums the contribution intensity of all interference factors along the same path to obtain the total interference intensity of the path. In this embodiment, the total interference intensity is the influence intensity of the interference factors on the preparation path. In this embodiment, the preset threshold is a boundary value used to determine whether the intensity of interference requires intervention to adjust the formulation path. During the process verification stage, a large amount of historical batch data is selected. For each formulation path, the total value of the interference intensity without environmental compensation and the concentration consistency result of the corresponding batch are calculated. When it is found that the total value of the interference intensity exceeds a certain value, the concentration consistency index of that path continues to deteriorate and affects the subsequent coating quality. This value is used as the upper limit threshold of the interference intensity of that path. The lower limit threshold is selected as a typical value where the interference intensity is significantly greater than zero but the impact on concentration consistency is still acceptable. Both the upper and lower limits are written into the interference factor database in a defined numerical form. In actual operation, the system compares the total value of the interference intensity calculated for each path with the upper limit of the preset threshold for that path. When the total value is less than or equal to the upper limit, the path is marked as "interference negligible" and continues to be executed according to the existing formulation path. When the total value is greater than the upper limit, the path is marked as "interference significant" and the historical trend analysis process is triggered.In this implementation, historical trend data comprises the initial deviations of the preparation paths, the intensity of interference, and subsequent path adjustments recorded for several recent batches, along with the corresponding concentration consistency results. To obtain the variation patterns across multiple batches, the system selects batches from the historical trend data where the interference intensity has exceeded the upper limit threshold, organized chronologically into historical records. Each record includes the total interference intensity at that time, the direction and magnitude of parameter modifications at each node in the preparation path, and the changes in the concentration consistency index after adjustment. The system performs statistical analysis on these historical records. For each preparation path, all historical records are divided into several intervals based on the intensity of interference. For example, an interval is defined as the interference intensity between 1 and 2 times the upper limit threshold, and another interval is defined as the interference intensity between 2 and 3 times the upper limit threshold. In another interval, the average value and frequency of concentration consistency improvement when different adjustment directions are adopted within that interval are statistically analyzed. In this embodiment, the adjustment direction is specifically the direction of increasing or decreasing the chemical solution concentration setting value, increasing or decreasing the liquid supply flow rate, or adjusting the preparation time sequence at different nodes of the preparation path. The system determines the direction with the largest average improvement and the most occurrences as the priority adjustment direction under that interference intensity interval, and uses it as the historical change law of the path within the corresponding interference intensity range. The system further analyzes the order of concentration consistency changes after adjusting the direction at different preparation stages in chronological order, thereby obtaining the adjustment direction of the preparation path in different time periods. In this embodiment, the adjustment direction is constituted by the path node number, the time period number, and a definite action description such as "increase" or "decrease". The system then obtains the operational status differences between multiple sites based on the adjustment direction determined for each preparation path with significant interference. In this embodiment, the operational status differences between multiple sites are defined as the differences between the current pretreatment sites and the average level of similar sites in key operational indicators such as capacity utilization, number of downtimes, load rate, and local concentration fluctuations. These differences are directly calculated from the aforementioned operational monitoring data. The system reads the operational status data of all sites based on the same time window, calculates the average value of all sites for each operational indicator, and then subtracts the average value from the current value of the indicator for each site to obtain the status deviation of the site for that indicator. The status deviations of all indicators are combined into the operational status vector of the site. The operational status differences between multiple sites are the set of differences between the operational status vectors of different sites.In this implementation, system parameters are defined as a set of basic settings used to control the joint operation behavior of the preparation path and pretreatment stations. These include the target concentration setting, target flow setting, start and end times of preparation for each path, receiving cycle time setting for each station, and compensation coefficients associated with environmental interference factors. When analyzing the differences in the operating status of multiple stations, the system uses rules to map indicators with large state deviations and directly affected by the preparation path to the corresponding system parameter update requirements. For example, when the load rate of a downstream station of a path is systematically high and the concentration deviation of that path is positive, the system explicitly reduces the target concentration setting of that path by a certain percentage as one of the update requirements. When the number of downtimes of a downstream station of a path is significantly higher than the average level and the flow setting of that path is low, the system explicitly increases the target flow setting of that path by a certain percentage as one of the update requirements. All such changes obtained by mapping the relationship between state deviation and preparation path constitute the system parameter update requirement set. In this implementation, the priority ranking of parameter iteration updates is generated based on three criteria: the initial deviation level of the preparation path corresponding to the parameter, the historical statistical impact of the parameter change on the improvement effect of concentration consistency, and the magnitude of the difference in the operating status of the relevant stations. The system assigns a score to each parameter under these three criteria. For example, a parameter with a deviation level of 2 is assigned 3 points, a parameter with a deviation level of 1 is assigned 2 points, and a parameter with a deviation level of 0 is assigned 1 point. The historical improvement effect is divided into three levels according to the average improvement amount: a large improvement amount is assigned 3 points, a medium improvement amount is assigned 2 points, and a small improvement amount is assigned 1 point. The difference in operating status is divided into three levels according to the absolute value of the deviation, and is assigned 3 points, 2 points, and 1 point respectively. Then, the three scores are added together to obtain the total priority score of the parameter. All parameters are then sorted from high to low according to the total score, thus obtaining the priority ranking list of parameter iteration updates.The system then invokes the iterative update module to adjust the system parameters step by step according to priority. In this implementation, the iterative update module executes according to a fixed iterative strategy: starting with the parameter with the highest priority, it calculates a single-step adjustment amount based on the difference between the current value and the target value of this parameter. This single-step adjustment amount is equal to the difference multiplied by a fixed proportional coefficient less than 1. During the process verification phase, this proportional coefficient was set to 0.5 through multiple experiments to ensure that each step adjustment does not exceed half of the entire difference. The system applies this single-step adjustment amount to the current value of the parameter to obtain a new parameter value, and calculates the concentration of the preparation path based on the new parameter value in the internal prediction model. The predicted changes in the consistency index are made using the parameter change and concentration change relationship table established from the aforementioned historical trend data. If the prediction results show that all indicators are changing in an improving direction and will not exceed the set safety limit, the adjustment is confirmed to be effective, and the new parameter values ​​are written into the current parameter configuration state. In this embodiment, the parameter configuration state is a snapshot set of all current system parameter values. Then the system moves to the next parameter in the priority ranking and repeats the above steps. When the prediction result of a certain parameter shows that it may cause a certain indicator to exceed the safety limit, the parameter is not adjusted in this round, the total score is reduced by a certain value, and the evaluation is postponed to the next iteration. After completing a round of step-by-step adjustments to all parameters, the system calculates the stability standard compliance of multi-site operation based on the prediction results and actual feedback. In this implementation, the stability standard of multi-site operation is defined by two constraints: first, the three indicators of path-level concentration consistency for each preparation path do not exceed the target upper limit value corresponding to that path; second, the absolute value of all indicators in the differences in the operating status of each site does not exceed the allowable deviation upper limit determined by the system during the process verification stage based on equipment capacity and production capacity configuration. When the above two conditions are met simultaneously, it is determined that the current parameter configuration status meets the stability standard of multi-site operation. If it does not meet the standard, the next round of iteration is continued within the iterative update module until the stability standard is reached or the pre-set maximum number of iterations is reached.Once the current parameter configuration status is determined to meet the multi-site operational stability standard, the system loads the updated parameters through the chemical solution preparation module. In this embodiment, the chemical solution preparation module is responsible for writing the iteratively updated system parameters into the actual preparation execution flow. During loading, the chemical solution preparation module sequentially reads the latest target concentration setpoint, flow rate setpoint, and preparation time setpoint according to the preparation path, and sends them to the preparation tank control unit and the supply pump control unit, thus initiating a new pretreatment operation cycle. During this operation cycle, the system continuously collects real-time operating data from the pretreatment stage. The real-time operating data includes the actual concentration, flow rate, preparation start and end times, and processing cycle time of each path at each time period, etc., and the system once again calculates the concentration consistency data. According to the definition, new concentration consistency indicators are statistically analyzed, and the changing trends of these indicators are continuously monitored in one or more batches. The monitoring results of process stability in this embodiment are the clear conclusions and corresponding values ​​obtained by comparing the above-mentioned new concentration consistency indicators and operating status difference indicators with the expected standards. The clear conclusions include three states: "process stable", "process tending to be stable" and "process unstable". The judgment rules are clarified through the analysis of a large amount of historical data during the process verification stage. When the concentration consistency and operating status difference indicators of several consecutive batches are stable within the expected standards, it is marked as "process stable". When the indicators gradually approach the standards but a small number of batches still exceed them, it is marked as "process tending to be stable". When the indicators continue to exceed the standards, it is marked as "process unstable". Based on the monitoring results of process stability, the system generates optimized path records for multiple batches of processing. In this embodiment, the optimized path record is a structured record file containing fields such as batch range, specific paths and values ​​of parameter adjustments in each iteration, concentration consistency index of the corresponding batch, intensity of interference factors, differences in operating status, and final process stability conclusion. This record is stored by the data storage unit. When writing, the data storage unit assigns a unique identifier to each optimized path record. In this embodiment, the identifier is a string or integer encoding obtained by combining a timestamp and an auto-incrementing sequence number to ensure that it is not repeated throughout the entire system lifecycle. After the record is successfully saved, the data storage unit returns the identifier to the upper-level system. The upper-level system associates the identifier with the corresponding batch range and parameter configuration status and archives it for subsequent querying and traceability.

[0038] A pre-processing device for wafer coating is also provided, used to implement the steps of the pre-processing method for wafer coating. The device includes a pre-processing and data acquisition module, which transports the wafer to be processed to the pre-processing station and selects the corresponding chemical solution for wafer surface cleaning, activation, or etching according to process requirements. It collects batch data and concentration distribution information of the chemical solution from each processing station through a sensor network, and performs multi-dimensional verification on the real-time acquisition frequency and data sampling accuracy to obtain the initial change trend within the concentration distribution range, used to determine the concentration distribution characteristics between batches; an anomaly detection module, based on the concentration distribution characteristics between batches, triggers anomaly detection rules if the concentration range exceeds a preset threshold range, and judges the degree of anomaly in the concentration distribution by combining historical trend comparison and environmental interference factor analysis; a deviation location module, based on the degree of anomaly in the concentration distribution, extracts dynamic concentration data around relevant marker points from the real-time monitoring system, performs time-series analysis based on equipment response delay, obtains the distribution pattern of potential deviations, and determines the key points that need adjustment; and a correction value generation module. The system employs a multi-stage process: a first stage, a second stage, and a third stage. The first stage calculates targeted correction values ​​based on the deviation distribution patterns at key points, using deviation calculation formulas and correction value generation logic, combined with historical data weighting and compensation range limitations. The second stage, a third stage, inputs these correction values ​​into the chemical solution distribution modules of each device, enabling real-time adjustment of chemical solution concentration via the distribution actuator. A data synchronization protocol is used to synchronize the correction values ​​to all sites, ensuring a consistent preparation status. The third stage, a second stage, compares deviation records in the wafer surface treatment logs with the consistent concentration status. If the deviation record falls within a preset threshold range, the device adaptation parameters are automatically reloaded, ensuring the wafer consistently encounters a uniform chemical solution concentration in subsequent processing stages, resulting in optimized concentration consistency data. The fourth stage, a fifth stage, generates targeted feedback reports based on the optimized concentration consistency data and iteratively updates system parameters by incorporating environmental interference factors and historical trends. This drives the system to continuously optimize the preparation path across multiple sites and batches, ultimately ensuring chemical solution consistency and process stability in the wafer pretreatment stage.

[0039] In the aforementioned apparatus, the preprocessing and data acquisition module first transports the wafer to the preprocessing station and selects a chemical solution for cleaning, activation, or etching based on process requirements. This module collects batch data and concentration distribution information of the chemical solution at fixed sampling intervals through a sensor network, and performs multi-dimensional verification on sampling frequency, sampling accuracy, and data stability to obtain a reliable initial concentration change trend for judging concentration distribution differences between batches. The anomaly detection module receives concentration distribution characteristics and determines whether the deviation from the preset threshold is out of tolerance by calculating the deviation of the concentration amplitude range. When out of tolerance is confirmed, the degree of anomaly is analyzed by combining historical trends and the offset of environmental interference factors. The deviation location module extracts concentration data around the marked point from the real-time monitoring system based on the degree of anomaly, performs time-series calibration according to equipment response delay, and obtains the deviation distribution pattern by comparing time-series differences to determine key points. The correction value generation module reads the deviation distribution pattern of key points, weights the deviation amount using historical data weights, determines the deviation direction and deviation amplitude according to deviation calculation rules, and then limits excessive deviations based on compensation range limitations, finally generating targeted correction values ​​that can be directly used for adjustment. The concentration adjustment module converts the correction value into an execution command and sends it to the chemical solution distribution module. The actuator adjusts the chemical solution concentration or flow rate in real time, while a data synchronization protocol ensures that all stations apply the same correction information within the same control cycle. The processing log comparison and equipment adaptation module, after unifying the concentration status, reads deviation records from the wafer surface treatment log and compares them with preset thresholds. When the deviation meets the required range, it automatically triggers the loading of equipment adaptation parameters and confirms the adaptation completion using the execution status returned by the equipment. The feedback update module evaluates process stability based on the optimized concentration consistency data and iteratively updates system parameters based on environmental interference factors and historical trends. This allows the preparation path to gradually converge and remain stable over multiple batches, thereby continuously improving the chemical solution consistency and process stability in the pretreatment stage.

[0040] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A pretreatment method for wafers before coating, characterized in that, include: The wafers to be processed are transported to the pre-processing station, and the corresponding chemical solutions are selected according to the process requirements for wafer surface cleaning, activation or etching. The batch data and concentration distribution information of the chemical solutions are collected from each processing station through a sensor network, and the real-time acquisition frequency and data sampling accuracy are verified in multiple dimensions to obtain the initial change trend within the concentration distribution range and determine the concentration distribution characteristics between batches. Based on the concentration distribution characteristics between batches, if the concentration range exceeds the preset threshold range, the anomaly detection rule is triggered, and the degree of anomaly in the concentration distribution is judged by combining historical trend comparison and environmental interference factor analysis. To determine the degree of anomaly in concentration distribution, dynamic concentration data around relevant marker points are extracted from the real-time monitoring system. Time-series analysis is performed based on equipment response delay to obtain the distribution pattern of potential deviations and identify key points that need adjustment. Based on the deviation distribution pattern of key points, the deviation calculation formula and correction value generation logic are used, combined with historical data weights and compensation range limitations, to calculate targeted correction values. The correction value is input to the chemical solution distribution module of each device, and the concentration of the chemical solution is adjusted in real time through the adjustment actuator. The correction value is synchronized to all stations using a data synchronization protocol.

2. The pretreatment method for wafer coating according to claim 1, characterized in that: The process involves transporting the wafer to be processed to the pre-processing station, and selecting the appropriate chemical solution for wafer surface cleaning, activation, or etching according to process requirements. A sensor network is used to collect batch data and concentration distribution information of the chemical solution from each processing station, and multi-dimensional verification is performed on the real-time acquisition frequency and data sampling accuracy to obtain the initial trend of concentration distribution within the range, which is used to determine the concentration distribution characteristics between batches. The sensor network collects batch data and concentration distribution information of chemical solutions from each processing station, and performs multi-dimensional verification on the real-time acquisition frequency and data sampling accuracy to obtain the initial trend of change within the concentration distribution range. By integrating the initial trend of change with the temperature monitoring data obtained from the processing station, the fluctuation parameters of the concentration distribution between batches are obtained, and the correspondence between the fluctuation parameters and the liquid volume adjustment is determined. Adjust the chemical solution supply for fluctuating parameters, obtain a stable concentration distribution from the adjusted supply, and determine whether the stable concentration distribution meets the preset threshold. If it does, record it as optimized batch data. By comparing the optimized batch data with the initial trend, the concentration distribution characteristics between batches can be determined.

3. The pretreatment method for wafer coating according to claim 1, characterized in that: Based on the concentration distribution characteristics between batches, if the concentration range exceeds a preset threshold interval, an anomaly detection rule is triggered. The degree of anomaly in the concentration distribution is determined by combining historical trend comparisons and environmental interference factor analysis, including: The amplitude range is obtained from the concentration distribution characteristics between batches, and the amplitude range is initially filtered by environmental interference factors to remove the influence of noise and obtain the fluctuation parameters of the amplitude range. For fluctuating parameters, if the amplitude range exceeds the preset threshold range, the anomaly detection rule is triggered, and data synchronization between sites is integrated to combine real-time information to determine the detection basis. Based on the detection criteria, historical trend comparison is adopted. The difference between current data and historical data is calculated to obtain the trend comparison difference, and combined with interference analysis to determine the distribution abnormal parameters. The degree of anomaly is obtained by distributing anomaly parameters, and the supply is dynamically calibrated to balance the distribution and obtain stable characteristics of the concentration distribution.

4. The pretreatment method for wafer coating according to claim 1, characterized in that: Regarding the degree of anomaly in concentration distribution, dynamic concentration data around relevant marker points is extracted from the real-time monitoring system. Time-series analysis is performed based on equipment response delays to obtain the distribution pattern of potential deviations and determine key points requiring adjustment, including: Data on concentration anomalies around marker points are extracted from the real-time monitoring system. The data timing is adjusted based on the equipment delay parameters by response delay correction and environmental interference factors are fused to obtain preliminary timing deviation analysis results. The results of time series deviation analysis are compared with historical trends. If the deviation exceeds a preset threshold, distribution pattern recognition is triggered. Potential distribution patterns are obtained by comparing the current deviation with historical patterns. Key locations are selected based on potential distribution patterns, and the degree of anomaly is determined by quantifying and integrating site data through supply adjustments and optimizations, and calculating quantitative values. By quantifying the anomaly degree parameters, we can obtain the stable distribution equilibrium characteristics and determine the key points that need adjustment.

5. The pretreatment method for wafer coating according to claim 1, characterized in that: Based on the deviation distribution pattern of key points, the deviation calculation formula and correction value generation logic are used, combined with historical data weights and compensation range limitations, to calculate targeted correction values, including: The deviation distribution pattern is obtained from the deviation of key points. The deviation distribution pattern is then fused with historical data weights to obtain a preliminary weighted deviation value. The fusion is achieved by multiplying the historical data weights by the values ​​of each element of the deviation distribution pattern and summing them up using a weighted average method. The initial weight deviation value is calculated using a deviation calculation formula. Combined with the correction value generation logic, the intermediate correction deviation is determined. The deviation calculation formula is D=(PH)×W, where D represents the deviation value, P represents the current point value, H represents the historical average value, and W represents the weight. After calculation using this formula, the correction value generation logic is adjusted to the intermediate correction deviation. The compensation range limit is obtained by fusing intermediate correction deviation. If it exceeds the preset threshold, the point weight is adjusted to obtain the deviation mode after the limit is obtained. The fusion is achieved by adding the components of intermediate correction deviation within the compensation range limit. By calculating the targeted correction value through the post-deviation mode, the control parameters are synchronized to obtain the basis for equipment control. The targeted correction value is obtained by calculating the average value of the post-deviation mode.

6. The pretreatment method for wafer coating according to claim 1, characterized in that: The process of inputting correction values ​​into the chemical solution distribution modules of each device, adjusting the chemical solution concentration in real time through the dispensing actuator, and synchronizing the correction values ​​to all stations using a data synchronization protocol to ensure a uniform preparation status includes: The correction value data is obtained through the chemical liquid dispensing module. The format of the input correction value is validated to obtain a standardized correction dataset. Based on the standardized correction dataset, the actuator interface is invoked to transmit the correction value to the chemical solution concentration adjustment unit, and the execution instruction for concentration adjustment is determined. A data synchronization protocol is used to encrypt the executed instructions, obtain the encrypted instruction data packet, and determine the integrity of the instruction data packet. If the instruction data packet is complete, the encrypted instruction data packet is distributed to each station through the network channel, and the receiving confirmation information is obtained from each station. For each site's confirmation message, the status field in the confirmation message is parsed to determine whether each site has synchronized the corrected value data; If a site fails to synchronize the corrected data, the encrypted instruction data packet will be resent through the backup channel to obtain the site's secondary confirmation information. Based on the secondary confirmation information, update the configuration status records of each site to determine whether all sites have reached a consistent status.

7. The pretreatment method for wafer coating according to claim 1, characterized in that, It also includes comparing deviation records in the wafer surface treatment log based on a unified concentration status. If the deviation record is within a preset threshold range, the device adaptation parameters are automatically reloaded, ensuring that the wafer always comes into contact with a consistent chemical solution concentration in subsequent processing stages, resulting in optimized concentration consistency data. Specifically, this includes: By using the wafer surface treatment log, specific data of deviation records are obtained, and the deviation records are compared with preset thresholds to determine whether the deviation is within an acceptable range. If the deviation is recorded within the preset threshold range, the relevant parameters of device adaptation are obtained through the system interface to determine the update requirements of device adaptation. Based on the device adaptation update requirements, the parameter loading module is invoked to transmit the adaptation parameters to the device control unit and obtain the status feedback after the parameters are loaded. Based on the status feedback after parameter loading, the execution field in the feedback is parsed to determine whether the device has completed the parameter update process. If the device has completed the adaptation parameter update, the concentration status monitoring module will obtain real-time concentration data to determine whether the concentration consistency meets the expected standard. Based on the concentration consistency data, a record file of the optimization results is generated, the optimization results are saved through the data storage unit, and the saved index identifier is obtained; For the saved index identifier, update the relevant fields in the wafer surface treatment log to determine whether the log data has been synchronized to the latest state.

8. The pretreatment method for wafer coating according to claim 1, characterized in that, It also includes generating targeted feedback reports based on optimized concentration consistency data, and iteratively updating system parameters by integrating environmental interference factors and historical trends, driving the system to continuously optimize the preparation path in multi-site and multi-batch processing. Specifically, this includes: By using concentration consistency data, the current status information of the chemical solution preparation is obtained. Key fields in the status information are analyzed to determine the initial deviation of the preparation path. Based on the initial deviation, environmental interference factor data is obtained, and the corresponding influence weights are extracted from the pre-established interference factor database to determine the intensity of the interference factor's effect on the preparation path. If the influence of the interference factor on the preparation path exceeds the preset threshold, the change pattern in multiple batches of processing is obtained through historical trend data to determine the adjustment direction of the preparation path in different time periods. To determine the direction of adjustment, the differences in operating status among multiple sites are obtained, and the update requirements for system parameters are extracted from these differences to obtain the priority ranking of parameter iteration updates.

9. The pretreatment method for wafer coating according to claim 8, characterized in that: The process of generating targeted feedback reports based on optimized concentration consistency data, and iteratively updating system parameters by integrating environmental interference factors and historical trends to drive continuous optimization of the preparation path across multiple sites and batches, also includes: Based on priority, the system parameters are adjusted step by step using an iterative update module to obtain the adjusted parameter configuration status and determine whether the configuration meets the stability standards for multi-site operation. If the configuration meets the stability standards for multi-site operation, the updated parameters are loaded through the chemical solution preparation module to obtain real-time operating data of the pretreatment stage and determine the monitoring results of process stability. Based on the monitoring results of process stability, optimized path records for multi-batch processing are generated, and the path records are saved through the data storage unit to obtain the saved identification information.

10. A pretreatment apparatus for wafer deposition, used to implement the steps of the pretreatment method for wafer deposition according to any one of claims 1-9, characterized in that, The device includes: The preprocessing and data acquisition module transports the wafer to be processed to the preprocessing station and selects the corresponding chemical solution for wafer surface cleaning, activation or etching according to process requirements. It collects batch data and concentration distribution information of chemical solution from each processing station through a sensor network, and performs multi-dimensional verification on the real-time acquisition frequency and data sampling accuracy to obtain the initial change trend within the concentration distribution range and determine the concentration distribution characteristics between batches. The anomaly detection module, based on the concentration distribution characteristics between batches, triggers anomaly detection rules if the concentration range exceeds a preset threshold range. It judges the degree of anomaly in the concentration distribution by combining historical trend comparison and environmental interference factor analysis. The deviation location module extracts dynamic concentration data around relevant marker points from the real-time monitoring system based on the degree of abnormality in concentration distribution, performs time-series analysis based on equipment response delay, obtains the distribution pattern of potential deviations, and determines the key points that need to be adjusted. The correction value generation module calculates targeted correction values ​​based on the deviation distribution pattern of key points, using deviation calculation formulas and correction value generation logic, combined with historical data weights and compensation range limitations. The concentration adjustment module inputs the correction value to the chemical solution distribution module of each device, realizes the real-time adjustment of the chemical solution concentration through the adjustment actuator, and uses a data synchronization protocol to synchronize the correction value to all stations; The processing log comparison and device adaptation module compares the deviation records in the wafer surface treatment log according to the unified concentration status. If the deviation record is within the preset threshold range, the device adaptation parameters are automatically reloaded so that the wafer always comes into contact with a consistent chemical solution concentration in subsequent processing steps, and the optimized concentration consistency data is obtained. The feedback update module generates targeted feedback reports based on the optimized concentration consistency data, and iteratively updates system parameters by integrating environmental interference factors and historical trends, driving the system to continuously optimize the preparation path in multi-site and multi-batch processing.

Images