A method, apparatus, and medium for pressure regulation of a wafer bake cavity

Abstract

The application discloses a pressure adjusting method, device and medium of a wafer drying cavity, and relates to the technical field of semiconductor manufacturing. The method comprises the following steps: collecting process data of the wafer drying cavity in real time; based on the process data, calculating the internal pressure value of the wafer drying cavity and the corresponding confidence interval through a pre-constructed cavity pressure model; generating a feedback adjustment instruction set through a feedback control engine according to the pressure difference value between the internal pressure value and a preset pressure standard value; collecting multiple preset process event signals of the wafer drying cavity in parallel, and judging whether a disturbance event is triggered; when the disturbance event is triggered, generating an active adjustment instruction set through a feedforward control engine based on the disturbance type; and generating a final adjustment driving instruction and executing the same. Through the construction of the cavity pressure model and the combination of the feedforward feedback double-engine control mechanism, accurate adjustment can be carried out based on the pressure deviation, and foreseeable disturbance can be compensated in advance.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of semiconductor manufacturing, in particular to a wafer drying cavity pressure adjusting method, device and medium. BACKGROUND

[0002] Wafer drying is a key link in the semiconductor manufacturing process. By removing moisture and residual solvents on the surface of the wafer, while avoiding the introduction of impurities, scratches or structural damage during the drying process, a clean and dry surface is provided for subsequent key processes such as photolithography, thin film deposition, etching, etc. on the semiconductor wafer. Any surface contamination or process defects that may occur during the drying process will result in a decrease in product quality and affect the performance of semiconductor devices. To isolate external pollution and control the diffusion of internal pollutants, the drying process is usually carried out in a sealed cavity. High-purity inert gas is introduced into the cavity before drying to replace air, so that the cavity is in a drying negative pressure environment, and a stable micro-negative pressure inert gas environment needs to be maintained throughout the drying process.

[0003] However, the existing wafer drying equipment usually adopts a closed-loop feedback control system composed of a pressure sensor, a programmable logic controller (PLC) and an electric regulating valve. However, when the wafer drying cavity is working, this traditional proportional-integral-derivative (PID) feedback control mode is essentially a passive response during the wafer drying process. Only when the pressure value deviates from the set value, the correction operation is performed, it is difficult to predict the possible and foreseeable strong disturbances, such as opening or closing of the cavity door cover, loading or unloading of the wafer, sudden change of the heating base power, step change of the process gas flow, etc., resulting in inherent lag in control response.

[0004] In addition, the existing drying equipment usually collects the negative pressure in the cavity through a diaphragm type sealing pressure sensor, which is connected to the cavity through a process interface sealed by a metal diaphragm. One side of the diaphragm contacts the process cavity, and the other side is inert gas. Although this method physically isolates the sensor from the process environment, the diaphragm needs to deform repeatedly to respond to pressure changes. Long-term high-frequency deformation can cause the metal diaphragm to enter the fatigue stage, and the elastic recovery ability decreases, resulting in a lag in the pressure value collected by the sensor, thereby causing the controller to control the regulating valve inaccurately, causing frequent fluctuations in negative pressure, breaking the micro-negative pressure balance, increasing the risk of external unclean air infiltration, and causing production accidents. SUMMARY

[0005] To solve the above problems, the present application provides a wafer drying cavity pressure adjusting method, applied to a wafer drying cavity with an air inlet system and an air outlet system, comprising:

[0006] collecting process data of the wafer baking cavity in real time; the process data includes inlet gas flow data and inlet pipe pressure data in the inlet pipe, outlet gas flow data and outlet pipe pressure data in the outlet pipe, and current execution parameters of the inlet flow controller and current opening degree parameters of the outlet pressure control valve;

[0007] based on the process data, calculating the internal pressure value of the wafer baking cavity and the corresponding confidence interval through a pre-constructed cavity pressure model;

[0008] generating a feedback adjustment instruction set through a feedback control engine according to the pressure difference between the internal pressure value and a preset pressure standard value;

[0009] At the same time, a plurality of preset process event signals of the wafer baking cavity are collected in parallel to determine whether a disturbance event is triggered; the process event signals include wafer transmission instruction signals, heating base power setting signals, and process gas flow setting signals;

[0010] When the disturbance event is triggered, an active adjustment instruction set is generated through a feedforward control engine based on the triggered disturbance type;

[0011] Based on the feedback adjustment instruction set and the active adjustment instruction set, a final adjustment driving instruction is generated and executed, and the execution parameters of the inlet flow controller and the opening degree parameters of the outlet pressure control valve are adjusted to maintain the pressure stability of the wafer baking cavity.

[0012] In another aspect, the present application also provides a wafer baking cavity pressure adjustment device, comprising:

[0013] at least one processor; and,

[0014] a memory in communication connection with the at least one processor; wherein,

[0015] the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform a wafer baking cavity pressure adjustment method as described in the above examples.

[0016] In another aspect, the present application also provides a non-volatile computer storage medium storing computer executable instructions, which are configured to perform a wafer baking cavity pressure adjustment method as described in the above examples.

[0017] The wafer baking cavity pressure adjustment method provided by the present application can bring the following beneficial effects:

[0018] By constructing a cavity pressure model to realize non-contact pressure sensing, and combining a feedforward-feedback dual-engine control mechanism, not only can the pressure deviation be accurately adjusted, but also foreseeable disturbances such as wafer transmission and process parameter adjustment can be compensated in advance, which significantly improves the dynamic response speed of the control, so that the pressure can be quickly stabilized when a process disturbance occurs, avoiding the continuous fluctuation of the pressure caused by the response delay of the traditional pure feedback control.

[0019] By calculating the internal pressure value based on the external parameters, the potential pollution risk caused by installing physical sensors in the cavity and the fatigue drift problem of the sensors after long-term use are completely avoided. By outputting the confidence interval and establishing a linkage mechanism between the confidence interval and the control parameters, the control strategy can be automatically adjusted according to the predicted reliability, thereby greatly improving the long-term stability and reliability of the entire pressure control while ensuring the control accuracy. BRIEF DESCRIPTION OF DRAWINGS

[0020] The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this application, illustrate embodiments of the application and together with the description serve to explain the application. In the drawings:

[0021] Figure 1 A flowchart of a pressure regulation method for a wafer drying cavity in an embodiment of the application;

[0022] Figure 2 A communication diagram of a computer control system and a control component of a wafer drying cavity in an embodiment of the application;

[0023] Figure 3 A structural diagram of a wafer drying cavity in an embodiment of the application;

[0024] Figure 4 A schematic diagram of a pressure regulation device for a wafer drying cavity in an embodiment of the application. DETAILED DESCRIPTION

[0025] To make the objectives, technical solutions and advantages of the application clearer, the technical solutions of the application will be described below in conjunction with specific embodiments of the application and corresponding drawings. Obviously, the described embodiments are only some of the embodiments of the application, not all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those of ordinary skill in the art without creative labor fall within the scope of the application.

[0026] The technical solutions provided by the embodiments of the application will be described in detail below with reference to the accompanying drawings.

[0027] As Figure 1As shown, the embodiment of the present application provides a pressure adjustment method of a wafer drying cavity, comprising:

[0028] S101, real-time acquisition of process data of the wafer drying cavity; the process data includes gas flow data and pressure data of the inlet pipeline, exhaust gas flow data and pressure data of the exhaust pipeline, and current execution parameters of the inlet flow controller and current opening degree parameters of the exhaust pressure control valve.

[0029] The computer control system directly reads the real-time gas flow value stored in the internal register of the inlet flow controller (MFC) through the device control network to obtain the inlet gas flow data. The inlet flow controller is integrated with a flow sensor, and the inlet pipeline pressure data is collected through the first pressure sensor installed between the outlet end of the inlet flow controller and the cavity inlet.

[0030] The exhaust pipeline pressure data is collected through the vacuum gauge installed at the front end of the exhaust pressure control valve, and the exhaust gas flow data is collected through the flow sensor (such as a thermal mass flow meter) installed at the rear end of the exhaust pressure control valve, or the exhaust gas flow data is indirectly calculated through the measured pipeline pressure and valve opening degree.

[0031] At the same time, the current set value and actual opening percentage of the inlet flow controller are obtained by querying the state register of the inlet flow controller as the current execution parameters of the inlet flow controller; the current opening degree parameters of the exhaust pressure control valve are obtained by reading the feedback value of the servo motor encoder of the driven exhaust pressure control valve or the analog signal of the valve position sensor.

[0032] It should be noted that, as Figure 2 As shown, the computer control system is a software module integrated in the upper control system of the wafer drying device, running on the existing processor, using a switch device such as an industrial Ethernet switch, connected with each sensor and actuator through a device control network such as an industrial Ethernet, a fieldbus, etc.

[0033] In the embodiment of the present application, as Figure 3As shown, the main body of the wafer drying cavity is a closed container made of high-purity, corrosion-resistant material. A wafer transfer window is provided on one side wall of the cavity, which is the only passage for the wafer to enter or exit the cavity. The wafer is transferred by a transfer device controlled by a wafer transfer robot control system, so that the wafer is accurately transferred into or out of the cavity. Inside the cavity, a heating base is provided, and a wafer supporting device is provided on the upper part of the heating base for supporting the wafer to be processed. Inside the heating base, a heating assembly such as an embedded resistance heating base is integrated, which is electrically connected with an external heating base controller to receive control signals from the heating base controller, so as to heat the wafer to a specific temperature required by the process.

[0034] In addition, an air inlet is provided at the bottom or side wall of the cavity, which is connected to the outlet end of the mass flow controller through a pipeline and an air inlet joint, and the inlet end of the mass flow controller is connected to a high-purity nitrogen gas inlet device or a high-purity argon gas inlet device. At the same time, a first pressure sensor is installed on the air inlet pipeline for collecting air inlet pipeline pressure data.

[0035] Similarly, an air outlet is provided at the bottom or side wall of the cavity, which is usually opposite or staggered with the air inlet to ensure uniform airflow. The air outlet is connected to the inlet of the exhaust pressure control valve through a pipeline and an exhaust joint. The outlet end of the exhaust pressure control valve is connected to the vacuum pump to provide stable exhaust power. A vacuum gauge is installed on the exhaust pipeline for measuring the exhaust pipeline pressure, and a flow meter or pressure conversion can be optionally installed to obtain the exhaust gas flow data.

[0036] It should be noted that the mass flow controller, the exhaust pressure control valve, the first pressure sensor, the vacuum gauge, the flow sensor, the heating base controller, and the wafer transfer robot control system are all connected in communication with the pressure adjustment unit of the computer control system through the equipment control network.

[0037] S102, based on the process data, the internal pressure value of the wafer drying cavity and the corresponding confidence interval are calculated through a pre-constructed cavity pressure model.

[0038] Specifically, the process data is pre-processed, the inlet gas flow data, the inlet pipeline pressure data, the exhaust gas flow data and the exhaust pipeline pressure data are denoised by a moving average filtering algorithm, the abnormal value detection and elimination are performed by a 3 sigma rule, and the identified abnormal data points are replaced by linear interpolation of adjacent normal data points. The pre-processed process data is subjected to feature extraction, the cavity pressure features of each process data are extracted, the cavity pressure feature set of the current time process data is constructed, and the feature set includes the instantaneous value of the inlet gas flow and the first-order differential change rate of the instantaneous value, the pressure gradient of the exhaust pipeline, and the change trend index of the inlet flow controller and the exhaust pressure control valve opening, etc.

[0039] Among them, the cavity pressure feature is a feature vector of a sample data in the process data, and the cavity pressure feature set is a feature vector set of all sample data in the process data.

[0040] Further, the cavity pressure feature set is input into the pre-trained cavity pressure model, and in the model, each cavity pressure feature in the cavity pressure feature set is sequentially matched and calculated with each real sample feature in the preset calibration data set to obtain a feature similarity corresponding to each real sample feature. After the matching and calculation of each cavity pressure feature in the cavity pressure feature set are completed, the feature similarity corresponding to each cavity pressure feature is obtained to form a feature similarity set of the cavity pressure feature set.

[0041] It should be noted that the distance between each cavity pressure feature in the cavity pressure feature set and each real sample feature in the multi-dimensional feature space is determined by the feature similarity, and the matching calculation process can be performed by using a cosine similarity or Euclidean distance algorithm.

[0042] Further, according to the feature similarity set, the real cavity pressure value corresponding to the real sample feature is weighted and fused to obtain the internal pressure value and the corresponding confidence interval.

[0043] Specifically, for each cavity pressure feature in the cavity pressure feature set, the feature similarity is set as the initial weight of the corresponding real sample feature, and the initial weights are summed to obtain an initial weight sum. In a specific implementation, the feature similarity is mapped to the initial weight of the real sample feature corresponding to the matching calculation of the cavity pressure feature by a preset exponential decay function. Wherein, the preset exponential decay function is configured to assign a significantly higher weight to the sample with smaller distance.

[0044] Based on the initial weight sum, the initial weights are normalized to obtain the normalized weights corresponding to each real sample feature. In a specific implementation, each initial weight is divided by the initial weight sum to obtain the normalized weight corresponding to each real sample feature, so as to ensure that the sum of all initial weights is equal to 1.

[0045] Based on the normalized weight, the real sample feature corresponding real cavity pressure value in the preset calibration data set is weighted and fused to generate the internal pressure value component of the cavity pressure feature. In a specific implementation, the real sample feature corresponding real cavity pressure value in the preset calibration data set is multiplied by the normalized weight corresponding to the real sample feature to obtain a weighted pressure value. The sum of all weighted pressure values is calculated to generate the internal pressure value component corresponding to the cavity pressure feature.

[0046] Based on the internal pressure value components corresponding to all cavity pressure features in the cavity pressure feature set, the internal pressure value is generated. In a specific implementation, after calculating the internal pressure value components corresponding to all cavity pressure features in the cavity pressure feature set, the internal pressure value components are weighted and averaged based on the quality indicators of the feature vectors to obtain the internal pressure value.

[0047] The total feature similarity of the cavity pressure feature set is calculated, and the confidence interval corresponding to the internal pressure value is calculated based on the total feature similarity. In a specific implementation, each feature similarity in the feature similarity set is preprocessed to eliminate outliers in the feature similarity set and exclude invalid feature similarities caused by noise or transient interference to obtain a preprocessed effective feature similarity set. The effective feature similarity set is weighted and averaged to obtain the total feature similarity. The distribution feature parameters of the effective feature similarity set are calculated, including the standard deviation and skewness of the similarity values of the feature similarity. The total feature similarity is taken as the basic similarity value, and the distribution feature parameters are modified to finally generate the confidence interval width.

[0048] It should be noted that the confidence interval width is determined by two key factors. One is the standard deviation of the real cavity pressure value assigned with an effective weight. The larger the standard deviation, the wider the interval. The other is the total feature similarity of the current cavity pressure feature set and all sample feature sets. The lower the total feature similarity, the wider the interval. The standard deviation and the total feature similarity are input into a preset interval calculation function to finally output the confidence interval corresponding to the internal pressure value. The total feature similarity has a value range of [0, 1] and is used to quantify the overall matching degree of the cavity pressure feature set and the sample feature set in the preset calibration data set. The higher the value, the better the overall matching degree.

[0049] The total feature similarity is input into a preset mapping function, and the total feature similarity is converted into a basic uncertainty coefficient through the mapping function. The mapping function is configured to output a smaller uncertainty coefficient when the total feature similarity value is larger, and output a larger uncertainty coefficient when the total feature similarity value is lower. The dispersion of the real cavity pressure value of the effective feature similarity is added, and the standard deviation of the real cavity pressure value is introduced as a dispersion compensation factor into the confidence interval calculation. The basic uncertainty coefficient and the compensation factor corresponding to the dispersion degree are weighted and combined to obtain a dispersion compensation adjustment amount.

[0050] The confidence interval width and the dispersion compensation adjustment amount are input into a preset interval calculation function, and the confidence interval width value is obtained by adding the confidence interval calculation function. The confidence interval is output by applying the calculated confidence interval width value to the internal pressure value as the center.

[0051] In addition, a pre-training process of the cavity pressure model is also included. In the embodiment of the present application, a reference pressure sensor calibrated by metrology is temporarily installed in the wafer baking cavity. The reference pressure sensor is connected through a reserved process interface on the side wall of the cavity, and is used to collect the real cavity pressure value inside the cavity as a training label. At the same time, all normally operating air inlet flow controllers, first pressure sensors, vacuum gauges, flow meters, and exhaust pressure control valves are kept in normal working condition, and the readings of each sensor are collected in a synchronous time stamp manner to ensure the time consistency of the data in subsequent modeling.

[0052] Under the empty state of the cavity, the computer control system applies a composite excitation signal to the air inlet flow controller and the exhaust pressure control valve. The pseudo-random binary sequence excitation is applied to the air inlet flow controller to make the air inlet flow controller randomly switch different flow set values within the safe flow range. The sweep signal excitation is applied to the exhaust pressure control valve to make the exhaust pressure control valve opening change at different frequencies within the effective range. At the same time, the application time stamp of the excitation signal and the sweep signal is recorded to ensure accurate association with the response data. The excitation signal is composed of a series of randomly distributed opening and closing pulses, and the frequency range covers the main frequency band of the system working, aiming to stimulate the cavity pressure to produce rich dynamic response in the whole working interval.

[0053] During the application of the excitation signal, a plurality of sets of training sample data are synchronously collected, each set of training sample data including a real cavity pressure value measured by a reference pressure sensor, intake gas flow data and intake pipeline pressure data in an intake pipeline, exhaust gas flow data and exhaust pipeline pressure data in an exhaust pipeline, a current set value and actual opening degree of an intake flow controller, a current opening degree parameter of an exhaust pressure control valve, a power set value of a heating base, and actual temperature data. The real cavity pressure value measured by the reference pressure sensor is a target variable; the input features are the intake gas flow data and the intake pipeline pressure data in the intake pipeline, the exhaust gas flow data and the exhaust pipeline pressure data in the exhaust pipeline, the current set value and the actual opening degree of the intake flow controller, the current opening degree parameter of the exhaust pressure control valve, the power set value of the heating base, and the actual temperature data.

[0054] The collected raw training sample data are preprocessed, including removing abnormal data points due to signal transmission errors or sensor transient response; and using a moving average filter to smooth the data to reduce the influence of random noise. Based on the preprocessed data, a set of cavity pressure time sequence features is extracted, including flow dynamic features such as change rates and fluctuation amplitudes of intake and exhaust flow, pressure gradient features such as change trends of intake and exhaust pressure, valve state features such as change rates and maintenance durations of valve opening degree, and process coupling features such as interaction indicators of intake flow and exhaust opening degree. The extracted set of cavity pressure time sequence features and the corresponding real cavity pressure value are combined to construct a calibration data set, which is divided into a training subset and a validation subset.

[0055] A cavity pressure model is constructed through a Gaussian process regression framework, the training subset is input into the cavity pressure model, and the Gaussian process regression framework of the cavity pressure model is trained. Radial basis functions are used as the main components of the kernel function to capture the nonlinear relationship between the input features and the output pressure. Meanwhile, monotonicity constraints are introduced into the kernel function by imposing non-negativity restrictions on the derivative of the kernel function, to ensure that the model prediction output presents a monotonically decreasing physical law with the increase of the valve opening degree.

[0056] The hyperparameters of the model are optimized by maximizing the marginal likelihood function. Specifically, a probability distribution function about the hyperparameters is constructed, and the conjugate gradient method is used to iteratively adjust the values of the hyperparameters. In each iteration, the likelihood value of the training sample data under the current hyperparameter setting is calculated, and the optimal solution of the hyperparameters that can maximize the likelihood value is determined to determine the final form and parameters of the kernel function.

[0057] The generalization performance of the evaluation model is verified on the subset. The mean absolute error and the root mean square error between the predicted value and the true value of the model on the validation set are calculated. When the error is less than the preset tolerance threshold, it is determined that the model training is successful, and the temporarily installed reference pressure sensor is removed. The trained cavity pressure model is integrated into the computer control system.

[0058] In the embodiments of the present application, the confidence interval width of the internal pressure value output by the Gaussian process regression model is continuously monitored, which represents the uncertainty degree of the model prediction result. The monitoring process includes: calculating the difference between the upper and lower boundaries of the confidence interval of the internal pressure value in each sampling period in real time as the confidence interval width value at the current time, constructing a time series data of the confidence interval width value, analyzing the change trend of the time series data by using a sliding window statistical method, setting a plurality of time scale monitoring windows including a short-term fluctuation monitoring window and a long-term trend monitoring window, which are respectively used to capture sudden abnormalities and gradual performance degradation.

[0059] When it is detected that the confidence interval width presents a continuously increasing trend, a multi-level warning mechanism is started.

[0060] The process data collected by the intake is extracted from the historical database to form a new training data set. The data quality of the new training data set is evaluated and the abnormal values are filtered to ensure the effectiveness and representativeness of the training data. On the basis of maintaining the original model architecture, the Gaussian process regression model is retrained by using an incremental learning method, which not only retains the original knowledge but also adapts to the new working condition characteristics. The same hyperparameter optimization method and physical constraint condition as the initial model are used in the training process to ensure the consistency and physical rationality of the updated model.

[0061] After the model is updated, the control parameters of the feedback control engine are adjusted based on the characteristics of the updated model. The change of the confidence interval width output by the updated model is monitored to evaluate the reliability improvement effect of the model. According to the convergence of the confidence interval width, the parameters of the proportional-integral-derivative control algorithm are dynamically adjusted. When the confidence interval width is significantly reduced, the proportional gain and integral gain of the control system are appropriately increased to improve the response speed and accuracy of the control system. When the confidence interval width is not significantly improved, the gain parameters of the control system are maintained or appropriately reduced to ensure the stability and robustness of the control system. The performance comparison data before and after the model is updated, including the change of the confidence interval width and the improvement of the control effect, are recorded for subsequent model life cycle management.

[0062] S103、According to the pressure difference between the internal pressure value and the preset pressure standard value, a feedback adjustment instruction set is generated by a feedback control engine.

[0063] The preset pressure standard value corresponding to the wafer drying process is obtained, the internal pressure value is compared with the preset pressure standard value set by the process, the pressure difference value is calculated, and the pressure difference value is input into the feedback control engine. The feedback control engine is optimized based on a PID control algorithm. Through the proportional, integral, and differential operations of the engine on the difference value, a feedback adjustment instruction set is generated. The feedback adjustment instruction set includes feedback parameter adjustment instructions of the air inlet flow controller and feedback opening adjustment instructions of the exhaust pressure control valve.

[0064] In specific implementation, the pressure difference value between the preset pressure standard value and the internal pressure value is calculated, and at the same time, the confidence interval width corresponding to the internal pressure value output by the cavity pressure model is obtained. The width value corresponding to the confidence interval is compared with the preset width threshold in real time.

[0065] When the width value exceeds the preset width threshold, the gain coefficient of the feedback control engine is adjusted based on a preset attenuation factor. The gain coefficient includes a proportional gain coefficient, an integral gain coefficient, and a differential gain coefficient. In specific implementation, when the width value corresponding to the confidence interval exceeds the preset width threshold, indicating that the model prediction uncertainty is high, the proportional gain coefficient, the integral gain coefficient, and the differential gain coefficient of the feedback control engine are adjusted according to a first preset rule. The first preset rule is that the proportional gain coefficient, the integral gain coefficient, and the differential gain coefficient are multiplied by an attenuation factor less than 1, so as to reduce the response speed of the control system and avoid excessive adjustment when the prediction is inaccurate.

[0066] Further, the pressure difference value of the preset pressure standard value and the internal pressure value of the wafer drying cavity is calculated, and the adjustment components corresponding to the pressure difference value are calculated according to the adjusted gain coefficient. The adjustment components include a proportional adjustment component, an integral adjustment component, and a differential adjustment component. In specific implementation, in a preset control period, for the proportional adjustment component, the pressure difference value is multiplied by the adjusted proportional gain coefficient; for the integral adjustment component, the integral of the pressure difference value of the current control period and the historical pressure difference value of the last control period is multiplied by the adjusted integral gain coefficient; and for the differential adjustment component, the difference between the pressure difference value of the current control period and the historical pressure difference value of the last control period is multiplied by the adjusted differential gain coefficient.

[0067] In each control period, the current control period length and the current pressure difference value are read, the integral increment in the current control period is calculated, the integral increment calculated this time is added to the integral increment summation of all previous control periods, and the integral of the pressure difference value with respect to time is obtained. In each control period, the current pressure difference value and the pressure difference value of the last control period are read, and the current control period length is read. The rate of change of the pressure difference value with respect to time is calculated. The rate of change is used to represent the difference of the pressure difference value with respect to time.

[0068] Further, based on the adjustment component, a base feedback control value is calculated to generate a feedback adjustment instruction set, the feedback adjustment instruction set including a feedback parameter adjustment instruction of the intake flow controller and a feedback opening adjustment instruction of the exhaust pressure control valve. In a specific implementation, the proportional adjustment component, the integral adjustment component and the differential adjustment component are superimposed to generate the base feedback control value, and the base feedback control value is decomposed into specific instructions for different actuators according to a preset distribution strategy. A preset proportional part of the total control value is distributed to the intake flow controller to generate the feedback parameter adjustment instruction containing a specific flow set value, and the remaining control value is distributed to the exhaust pressure control valve to generate the feedback opening adjustment instruction containing a target opening value.

[0069] S104, simultaneously, a plurality of preset process event signals of the wafer drying cavity are collected in parallel to determine whether a disturbance event is triggered; the process event signals include a wafer transfer instruction signal, a heating base power setting signal and a process gas flow setting signal.

[0070] Specifically, through the equipment control network, a plurality of preset process event signals of the wafer drying cavity are collected in real time and in parallel, the process event signals are analyzed, and when any process event signal meets the corresponding trigger condition, it is determined that a disturbance event of the corresponding disturbance type is triggered.

[0071] Through the equipment control network, a state instruction message from the wafer transfer robot control system is subscribed and received to obtain a wafer transfer instruction signal. In a specific implementation, for the collection of the wafer transfer instruction signal, a host communication connection is established with the wafer transfer robot control system through the SECS / GEM communication protocol, and a state instruction message of the wafer transfer robot control system is subscribed and received in real time to obtain the wafer transfer instruction signal.

[0072] The content of the wafer transfer instruction signal is analyzed, when there is a wafer loading start instruction in the content, it is determined that a wafer loading event is triggered, and when there is a wafer unloading start instruction in the content, it is determined that a wafer unloading event is triggered. In a specific implementation, when the content is parsed as a predefined "WaferTransferStart" command and the parameter identifier is "LOAD", that is, the content has a wafer transfer start command and the parameter indicates loading, it is determined that the content has a wafer loading start instruction, and it is determined that a wafer loading event is triggered, and the event is immediately captured; when the parsed message instruction is also "WaferTransferStart" instruction but the parameter identifier is "UNLOAD", that is, the content has a wafer transfer start command but the parameter indicates unloading, it is determined that the content has a wafer unloading start instruction, and it is immediately determined that a wafer unloading event is triggered.

[0073] The power setting value of the heating base controller is read based on a preset period through the equipment control network to collect the heating base power setting signal. In a specific implementation, for the collection of the heating base power setting signal, a master-slave communication connection is established with the heating base controller through the Profibus-DP field bus, and the setting value register of the heating base controller is periodically requested to be read based on a preset period. After receiving the request, the heating base controller returns the heating base power setting signal containing the current latest power setting value.

[0074] The change amount and the change rate of the heating base power setting signal in a unit time are calculated, and when the change amount is greater than a first preset threshold and the change rate is greater than a second preset threshold, it is determined that the heating power step event is triggered. In a specific implementation, the absolute difference between the power setting value at the current sampling time and the power setting value at the last sampling time is calculated as the change amount of the power in a unit time, and the change amount is divided by the sampling period to obtain the change rate of the power in a unit time. The calculated change amount is compared with the first preset threshold, and the change rate is compared with the second preset threshold. When and only when the change amount is greater than the first preset threshold and the change rate is greater than the second preset threshold, it is determined that a power mutation with a significant amplitude and speed occurs, and the heating power step event is triggered.

[0075] The flow setting value of the inlet flow controller is read based on a preset period through the equipment control network to collect the process gas flow setting signal. In a specific implementation, the Modbus TCP protocol is used to communicate with the inlet flow controller, and a request to access the holding register is periodically sent to the inlet flow controller based on a preset period. After receiving the request, the inlet flow controller returns the process gas flow setting signal containing the current real-time flow setting value.

[0076] The absolute difference between the current time flow setting value and the previous time flow setting value of the process gas flow setting signal is calculated, and when the absolute difference is greater than a third preset threshold, it is determined that the gas introduction change event is triggered. In a specific implementation, the absolute difference between the current time flow setting value and the previous time flow setting value is calculated, and the absolute difference is compared with the third preset threshold. When the absolute difference is greater than the third preset threshold, it is determined that the introduction state of the process gas has changed with process significance, and the gas introduction change event is triggered.

[0077] It should be noted that the feedforward control process and the feedback control process are executed in parallel.

[0078] S105, when the disturbance event is triggered, the feedforward control engine generates an active adjustment instruction set based on the triggered disturbance type.

[0079] When a disturbance event is determined to be triggered, a type identifier of the process event signal triggering the disturbance is determined. In specific embodiments, the feedforward control engine identifies and classifies the event type according to the disturbance type corresponding to the process event signal triggering the disturbance, and parses the received process event signal into a standardized type identifier, including specific types such as wafer loading, wafer unloading, heating power step, gas flow change, etc.

[0080] Further, based on the type identifier, a query is performed in the feedforward strategy library to determine an ideal compensation control curve corresponding to the current disturbance event. In specific embodiments, the feedforward control engine accesses the feedforward strategy library stored in the non-volatile memory, uses the type identifier as a query key, performs a search in the hash table of the feedforward strategy library, and obtains the corresponding compensation strategy parameter set. The compensation strategy parameter set includes recommended control parameter values for the gas inlet flow controller and the exhaust pressure control valve, and based on the compensation strategy parameter set, the ideal compensation control curve is constructed.

[0081] It should be noted that the feedforward strategy library is implemented using a hash table data structure, with the process event type identifier as the key value, and the corresponding compensation strategy parameter set is stored. The ideal compensation control curve refers to the recommended execution parameter value of the gas inlet flow controller and the recommended opening parameter value of the exhaust pressure control valve over time from the time when the disturbance event is triggered to the subsequent time period, which is composed of the first ideal compensation curve for the gas inlet flow controller and the second ideal compensation curve for the exhaust pressure control valve.

[0082] The trigger time of the current disturbance event is set as the time starting point, and based on the preset control period, the recommended control parameter values at the starting time of each period in the ideal compensation control curve are read in sequence, including the recommended execution parameter value of the gas inlet flow controller and the recommended opening parameter value of the exhaust pressure control valve. In specific embodiments, at the starting time of each control period, the internal timer is queried to obtain the current time offset of the current time relative to the starting point, and the first ideal compensation curve and the second ideal compensation curve are simultaneously queried, with the current time offset as an input parameter, and interpolation calculation is performed on the two curves respectively to obtain the recommended execution parameter value of the gas inlet flow controller and the recommended opening parameter value of the exhaust pressure control valve corresponding to the starting time of the current period.

[0083] The recommended execution parameter value of the gas inlet flow controller includes a basic compensation flow value, a compensation duration, a flow change curve type, and a time constant corresponding to the flow change curve type; the recommended opening parameter value of the exhaust pressure control valve includes an opening compensation reference value, a compensation action time length, and valve action curve configuration parameters; the cooperative control parameter is used to define the timing relationship of the actions of the two actuators, including a starting time difference and a mutual coupling coefficient.

[0084] For the intake flow controller, the feedforward control engine generates feedforward parameter adjustment instructions according to the recommended execution parameter values; for the exhaust pressure control valve, corresponding feedforward opening adjustment instructions are generated according to the recommended opening parameter values. In particular, for such disturbance events with clear timing characteristics as wafer transfer, a timestamp-based prediction compensation mechanism is adopted to generate compensation instructions in advance a few milliseconds before the actual start of the mechanical arm action, ensuring precise synchronization of the compensation opportunity with the occurrence of the disturbance.

[0085] It should be noted that the generated active adjustment instruction set is encapsulated in a unified data structure, including fields such as instruction type identifier, target device address, instruction value, effective timestamp, and instruction validity period. The encapsulated instruction set is sent simultaneously to the drivers of the intake flow controller and the exhaust pressure control valve through the real-time Ethernet network, ensuring that the two actuators can perform coordinated actions based on a unified time reference.

[0086] Among them, the feedforward instruction provides fast dynamic compensation, and the feedback instruction provides accurate steady-state adjustment. This parallel architecture ensures that the system can quickly suppress sudden disturbances while maintaining long-term control accuracy, achieving an optimal balance between response speed and control accuracy.

[0087] In the embodiments of the present application, the feedforward strategy library is constructed based on the mapping relationship between the process event signals and the optimal compensation instructions.

[0088] For each process event signal, a corresponding special preset test action is performed, and the current execution parameter of the intake flow controller and the current opening parameter of the exhaust pressure control valve are kept unchanged. In specific implementation, for each type of process event for which a feedforward strategy needs to be established, a special single-trigger test is performed. For example, the special test action for the wafer transfer event can be that the wafer transfer mechanical arm control system performs a wafer loading and unloading operation cycle, and sends wafer loading start instructions and wafer unloading start instructions respectively; the special test action for the heating power step event can be that the heating pedestal controller changes the heating pedestal power from the reference value to the preset target value to achieve rapid power switching; the special test action for the process gas flow change event can be that the intake flow controller changes the process gas flow from the reference value to the preset target flow to perform rapid flow switching.

[0089] During each test process, while keeping the current execution parameter of the intake flow controller and the current opening parameter of the exhaust pressure control valve unchanged, the internal pressure values are continuously collected through the cavity pressure model to form the disturbance pressure curve caused by the specific event, and the event start time, duration, and basic working condition parameters are recorded.

[0090] The characteristic parameters of the disturbance pressure curve are extracted, and first ideal compensation curves of the intake flow controller and second ideal compensation curves of the exhaust pressure control valve are respectively generated according to the characteristic parameters and preset response parameters. In specific implementation, the collected disturbance pressure curve is subjected to characteristic analysis, and the characteristic parameters are extracted, including disturbance amplitude, response delay, disturbance duration, change characteristics, etc. Among them, the disturbance amplitude is the maximum deviation value of pressure change, quantifying the strength of the disturbance; the response delay measures the interval time from event triggering to the start of pressure change; the change characteristics are dynamic parameters such as the rising rate and stable time of pressure change; the disturbance duration is the total length of time from the start of pressure change to the recovery to stability. The preset response parameters refer to the static performance indicators and operating performance limits of the intake flow controller and the exhaust pressure control valve, and are a set of parameters in the computer control system, including the maximum allowed flow set value, the minimum allowed flow set value, the maximum flow rate of change, the flow step response time of the intake flow controller, and the maximum safe opening, the minimum safe opening, the maximum opening rate of change, and the valve action response time of the exhaust pressure control valve.

[0091] According to the characteristic parameters and the preset response parameters, the first ideal compensation curve of the intake flow controller and the second ideal compensation curve of the exhaust pressure control valve are respectively generated. Specifically, for the intake flow controller, the compensation amplitude, the action duration, and the change trajectory are determined according to the disturbance amplitude, the response delay, the disturbance duration, and the change characteristics, the intake flow compensation amount of the reference flow of the intake flow controller is calculated, the execution parameter value is calculated based on the intake flow compensation amount through the parameterized control function, and the first ideal compensation curve is constructed with time as the horizontal axis and the execution parameter value as the vertical axis; for the exhaust pressure control valve, the exhaust flow compensation amount of the reference flow is calculated based on the compensation amplitude, the action duration, and the change trajectory, the compensation amount of the valve opening is calculated based on the exhaust flow compensation amount, the opening parameter value of the exhaust pressure control valve is determined through the coupling optimization algorithm, and the second ideal compensation curve is constructed with time as the horizontal axis and the opening parameter value as the vertical axis.

[0092] Based on the first ideal compensation curve and the second ideal compensation curve, an ideal compensation control curve is constructed, and the corresponding ideal compensation control curve is stored in the feedforward strategy library based on the type identifier of the process event signal. In specific implementation, the first ideal compensation curve and the second ideal compensation curve are combined to obtain the ideal compensation control curve, wherein the characteristic of the ideal compensation control curve is opposite to the change trend of the disturbance pressure curve and matches the amplitude. The process event type identifier is associated with the corresponding ideal compensation control curve and stored in the feedforward strategy library.

[0093] S106, based on the feedback adjustment instruction set and the active adjustment instruction set, generate the final adjustment driving instruction and execute, adjust the execution parameter of the intake flow controller and the opening parameter of the exhaust pressure control valve to maintain the pressure stability of the wafer drying cavity.

[0094] Based on a preset weight distribution strategy, the adjustment amounts corresponding to the instructions in the feedback adjustment instruction set and the active adjustment instruction set are synergistically fused to generate the final adjustment driving instruction.

[0095] It should be noted that, in the steady state condition, the feedback adjustment instruction occupies the dominant position to ensure the control accuracy; when the feedforward disturbance event is detected, the weight coefficient of the active adjustment instruction is automatically increased to strengthen the rapid response capability. For the intake flow controller, the feedback instruction amount and the active compensation amount are vector superimposed; for the exhaust pressure control valve, the same fusion principle is adopted to process the opening adjustment instruction.

[0096] In addition, before generating the final adjustment driving instruction, the fused instruction value is subjected to multi-level safety check. First, the instruction value is compared with the preset safety working interval of each actuator, and when it exceeds the reasonable range, the instruction value is limited within the allowed boundary instruction value. Secondly, considering the coupling characteristics between the two actuators, the coordination of the intake and exhaust instructions is checked to avoid mutually contradictory control actions. Finally, the change amplitude and rate of the instruction are evaluated to prevent the actuator from being damaged due to overaction.

[0097] The final adjustment driving instruction subjected to safety check is respectively packaged as a device-specific communication protocol. For the intake flow controller, a digital instruction containing the accurate flow set value is generated and sent to the control unit of the intake flow controller through the industrial network interface; for the exhaust pressure control valve, a standardized opening control signal is output and sent to the corresponding drive unit.

[0098] The present application realizes non-contact pressure sensing by constructing a cavity pressure model, and combines a feedforward feedback dual-engine control mechanism, which not only can accurately adjust based on pressure deviation, but also can pre-compensate predictable disturbances such as wafer transmission and process parameter adjustment, significantly improving the dynamic response speed of control, so that when a process disturbance occurs, the pressure can be quickly stabilized, avoiding the continuous fluctuation of pressure caused by response delay in traditional pure feedback control.

[0099] By calculating the internal pressure value based on external parameters, the potential pollution risk caused by installing physical sensors inside the cavity and the fatigue drift problem of sensors after long-term use are completely avoided. By outputting the confidence interval and establishing a linkage mechanism between the confidence interval and the control parameter, the control strategy can be automatically adjusted according to the predicted reliability, thereby greatly improving the long-term stability and reliability of the entire pressure control while ensuring the control accuracy.

[0100] As Figure 4 shown in the above embodiments, the embodiments of the present application further provide a pressure adjusting device of a wafer baking cavity, comprising:

[0101] at least one processor; and

[0102] a memory in communication with the at least one processor; wherein

[0103] the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the pressure adjusting method of the wafer baking cavity according to any one of the above embodiments.

[0104] The embodiments of the present application further provide a non-volatile computer storage medium, which stores computer executable instructions, and the computer executable instructions are configured to perform the pressure adjusting method of the wafer baking cavity according to any one of the above embodiments.

[0105] Each of the embodiments of the present application is described in a progressive manner, and the same or similar parts of each of the embodiments can be referred to each other. Each of the embodiments mainly describes the difference from other embodiments. In particular, the device and medium embodiments are basically similar to the method embodiments, and thus the description is relatively simple, and the relevant parts can be referred to the description of the method embodiments.

[0106] The device and medium provided by the embodiments of the present application are one-to-one corresponding to the method, and thus the device and medium also have the similar beneficial technical effects as the method. Since the beneficial technical effects of the method have been described in detail above, the beneficial technical effects of the device and medium will not be described here.

[0107] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can be in the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can be in the form of a computer program product implemented on one or more computer usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer usable program code.

[0108] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 one or more flowcharts and / or blocks in the flowcharts and / or combination thereof. Figure 1 one or more flowcharts and / or blocks in the flowcharts and / or combination thereof.

[0109] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 one or more flowcharts and / or blocks in the flowcharts and / or combination thereof. Figure 1 one or more flowcharts and / or blocks in the flowcharts and / or combination thereof.

[0110] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 one or more flowcharts and / or blocks in the flowcharts and / or combination thereof. Figure 1 one or more flowcharts and / or blocks in the flowcharts and / or combination thereof.

[0111] In one typical configuration, the computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0112] The memory can include non-persistent memory and / or volatile memory, such as random access memory (RAM) 620b. The memory can also include non-volatile memory, such as read only memory (ROM) 620a. The memory can be a memory cache, a buffer, a RAM, or other types of memory. The memory is an example of computer readable media.

[0113] Computer-readable media includes permanent and non-permanent, movable and non-movable media that can implement information storage by any method or technology. The information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassette, magnetic tape, disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information accessible by a computing device. According to the definition herein, computer-readable media does not include transitory media such as modulated data signals and carriers.

[0114] It should also be noted that the terms "comprising", "containing", or any other variant thereof are intended to cover a non-exclusive inclusion, such that a process, method, article or apparatus that comprises a list of elements does not only include those elements, but can also include other elements not expressly listed or inherent to such process, method, article or apparatus. Without more limitations, the element defined by the statement "comprising a" does not exclude the presence of additional identical elements in the process, method, article or apparatus that includes the element.

[0115] The above only describes the embodiments of the present application and is not intended to limit the present application. For those skilled in the art, the present application can have various modifications and changes. Any modification, equivalent replacement, improvement, etc. within the spirit and principle of the present application shall be included in the scope of claims of the present application.

Claims

1. A pressure adjustment method for a wafer baking cavity, applied to a wafer baking cavity with an air inlet system and an air outlet system, characterized in that, The method comprises the following steps: real-time acquisition of process data of the wafer drying cavity; the process data comprises gas flow data and gas pipeline pressure data in the inlet gas pipeline, exhaust gas flow data and exhaust gas pipeline pressure data in the exhaust gas pipeline, and current execution parameters of the inlet gas flow controller and current opening degree parameters of the exhaust gas pressure control valve; based on the process data, the internal pressure value of the wafer drying cavity and the corresponding confidence interval are calculated by a pre-constructed cavity pressure model, specifically comprising: pre-processing the process data, extracting the cavity pressure features of the pre-processed process data, and obtaining the cavity pressure feature set; input the cavity pressure feature set into the pre-trained cavity pressure model, and sequentially match and calculate the cavity pressure features with the real sample features in the preset calibration data set to obtain the feature similarity set; according to the feature similarity set, the real cavity pressure value corresponding to the real sample feature is weighted and fused to obtain the internal pressure value and the corresponding confidence interval; according to the pressure difference value between the internal pressure value and the preset pressure standard value, a feedback control engine is used to generate a feedback adjustment instruction set, specifically comprising: real-time comparison of the width value corresponding to the confidence interval with the preset width threshold value; when the width value exceeds the preset width threshold value, the gain coefficient of the feedback control engine is adjusted based on a preset attenuation factor; the gain coefficient includes proportional gain coefficient, integral gain coefficient and differential gain coefficient; calculate the pressure difference value between the preset pressure standard value and the internal pressure value of the wafer drying cavity, and calculate the corresponding adjustment components of the pressure difference value according to the adjusted gain coefficient; the adjustment components include proportional adjustment component, integral adjustment component and differential adjustment component; based on the adjustment components, calculate the basic feedback control value, and generate the feedback adjustment instruction set; the feedback adjustment instruction set includes feedback parameter adjustment instruction of the inlet gas flow controller and feedback opening degree adjustment instruction of the exhaust gas pressure control valve; At the same time, a plurality of preset process event signals of the wafer drying cavity are collected in parallel to determine whether a disturbance event is triggered; the process event signals include wafer transmission instruction signal, heating base power setting signal and process gas flow setting signal; when the disturbance event is triggered, a feedforward control engine is used to generate an active adjustment instruction set based on the triggered disturbance type, specifically comprising: determine the type identifier of the process event signal triggering the disturbance; based on the type identifier, query in the feedforward strategy library to determine the ideal compensation control curve corresponding to the current disturbance event; set the triggering time of the current disturbance event as the time starting point, and read the recommended control parameter value at the cycle starting time in the ideal compensation control curve based on the preset control period; the recommended control parameter value includes the recommended execution parameter value of the inlet gas flow controller and the recommended opening degree parameter value of the exhaust gas pressure control valve; generate a feedforward parameter adjustment instruction of the intake flow controller based on the recommended execution parameter value, and generate a feedforward opening adjustment instruction of the exhaust pressure control valve based on the recommended opening parameter value; generate final adjustment driving instructions based on the feedback adjustment instruction set and the active adjustment instruction set, and perform adjustment on the execution parameter of the intake flow controller and the opening parameter of the exhaust pressure control valve to maintain the pressure stability of the wafer drying cavity.

2. The method of claim 1, wherein the pressure is adjusted by adjusting the flow rate of the gas. The internal pressure value and the corresponding confidence interval are obtained by weighting and fusing the real cavity pressure values corresponding to the real sample features according to the feature similarity set, and specifically comprising: For each cavity pressure feature in the cavity pressure feature set, the feature similarity is set as the initial weight of the corresponding real sample feature, and the initial weights are summed to obtain an initial weight sum; Based on the initial weight sum, the initial weights are normalized to obtain the normalized weights corresponding to the real sample features; Based on the normalized weights, the real cavity pressure values corresponding to the real sample features in the pre-set calibration data set are weighted and fused to generate the internal pressure value component of the cavity pressure feature; Based on the internal pressure value components corresponding to all cavity pressure features in the cavity pressure feature set, an internal pressure value is generated. The total feature similarity of the cavity pressure feature set is calculated, and the confidence interval corresponding to the internal pressure value is calculated based on the total feature similarity.

3. The method of claim 1, wherein the pressure is regulated by a pressure regulator. The parallel acquisition of multiple preset process event signals of the wafer drying cavity is performed to determine whether a disturbance event is triggered, specifically comprising: Through the device control network, subscribe to and receive state instruction messages from the wafer transfer robot control system to obtain wafer transfer instruction signals; Through the device control network, read the power setting value of the heating base controller based on a preset period to acquire a heating base power setting signal; Through the device control network, read the flow setting value of the intake flow controller based on the preset period to acquire a process gas flow setting signal; The process event signals are analyzed, and when any process event signal meets the corresponding trigger condition, it is determined that a disturbance event of the corresponding disturbance type is triggered.

4. The method of claim 3, wherein the pressure is adjusted by adjusting the flow rate of the gas. The process event signals are analyzed, and when any process event signal meets the corresponding trigger condition, it is determined that a disturbance event of the corresponding disturbance type is triggered, specifically comprising: The message content of the wafer transfer instruction signal is analyzed, and when the message content contains a wafer loading start instruction, it is determined that a wafer loading event is triggered, and when the message content contains a wafer unloading start instruction, it is determined that a wafer unloading event is triggered; The change amount and the change rate of the heating base power setting signal in a unit time are calculated, and when the change amount is greater than a first preset threshold and the change rate is greater than a second preset threshold, it is determined that a heating power step event is triggered; The method further comprises:

5. The method of claim 1, wherein the pressure is regulated by a pressure regulator. Before the generating the active adjustment instruction set based on the triggered disturbance type by the feedforward control engine, the method further comprises: For each process event signal, a corresponding dedicated preset test action is performed, and the current execution parameter of the gas inlet flow controller and the current opening degree parameter of the exhaust pressure control valve are kept unchanged; The method further comprises: The method further comprises: The method further comprises:

6. A pressure regulating apparatus for a wafer bake chamber, comprising: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises:

7. A non-transitory computer storage medium storing computer-executable instructions that, when executed, cause a computer to perform: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises: The method further comprises

Images