A temperature and humidity prediction compensation method and system for suppressing transient disturbance of a developing chamber

Abstract

This application provides a method and system for predicting and compensating temperature and humidity disturbances in a developing chamber, relating to the field of intelligent prediction technology. The method includes: obtaining target multi-source information of the target developing chamber through dynamic continuous monitoring; inputting the target process timing and target state parameters as input information into a transient disturbance prediction model to obtain output results; extracting any predicted transient disturbance from one or more predicted transient disturbances and matching it with any predetermined compensation action in a predetermined compensation list; activating a collaborative compensation mechanism to execute the arbitrary predetermined compensation action, thereby achieving temperature and humidity prediction compensation for the target developing chamber. This application can solve the technical problem of low yield in lithography processes in the prior art, achieving the goal of accurately predicting and executing collaborative feedforward compensation control before the actual occurrence of transient disturbances, thus improving the yield of the developing process.

Description

Technical Field

[0001] This application relates to the field of intelligent prediction technology, and in particular to a method and system for predicting and compensating for temperature and humidity to suppress transient disturbances in the developing chamber. Background Technology

[0002] In semiconductor photolithography, the developing chamber within the coating and developing equipment experiences transient disturbances during the developing process, caused by periodic movements such as the movement of the robotic arm, the opening and closing of the chamber cover, and the reciprocating motion of the developer nozzles. These disturbances lead to drastic but brief fluctuations in the temperature and humidity field within the chamber. Simultaneously, droplets from the developer nozzles, resulting from insufficient backflow, may fall onto the wafer surface during the process, creating edge defects. Current technologies often employ feedback control or passive compensation after disturbances occur, lacking the ability to predict disturbances in advance and implement multi-level collaborative feedforward suppression. This makes it difficult to meet the stringent requirements for developing environment stability at process nodes of 55nm and below. The key challenge lies in accurately predicting transient disturbances in the developing process before they actually occur and implementing collaborative pre-compensation accordingly, thereby fundamentally suppressing wafer temperature and humidity deviations and edge defects caused by transient disturbances in the chamber and droplet buildup from the nozzles.

[0003] In summary, existing technologies suffer from the inability to accurately predict various transient disturbances during the development process, thus failing to provide timely adaptive disturbance pre-compensation, ultimately leading to low lithography yield and poor development results. Summary of the Invention

[0004] The purpose of this application is to provide a method and system for predicting and compensating temperature and humidity to suppress transient disturbances in the developing chamber, in order to solve the technical problem in the prior art that unpredictable transient temperature and humidity disturbances in the developing chamber are caused by various instantaneous operations, and that the compensation action lags behind the occurrence of the disturbance and cannot suppress the fluctuations in advance, which ultimately affects the environmental stability of the developing chamber and the wafer edge defect rate, thereby reducing the yield of the photolithography process.

[0005] In view of the above problems, this application provides a method and system for predicting and compensating temperature and humidity to suppress transient disturbances in the developing chamber.

[0006] In a first aspect, this application provides a method for predicting and compensating temperature and humidity to suppress transient disturbances in a developing chamber. The method is implemented through a system for predicting and compensating temperature and humidity to suppress transient disturbances in a developing chamber. The method includes: dynamically and continuously monitoring and obtaining target multi-source information of the target developing chamber, wherein the target multi-source information includes target process timing and target state parameters; inputting the target process timing and target state parameters as input information into a transient disturbance prediction model to obtain an output result, wherein the output result includes one or more predicted transient disturbances; extracting any predicted transient disturbance from the one or more predicted transient disturbances and matching it with any predetermined compensation action in a predetermined compensation sheet; activating a collaborative compensation mechanism to execute the arbitrary predetermined compensation action, thereby achieving temperature and humidity prediction compensation for the target developing chamber.

[0007] Secondly, this application also provides a temperature and humidity prediction and compensation system for suppressing transient disturbances in a developing chamber, used to execute a temperature and humidity prediction and compensation method for suppressing transient disturbances in a developing chamber as described in the first aspect. The temperature and humidity prediction and compensation system for suppressing transient disturbances in a developing chamber includes: a monitoring module, used to dynamically and continuously monitor and obtain target multi-source information of the target developing chamber, wherein the target multi-source information includes target process timing and target state parameters; a prediction module, used to input the target process timing and the target state parameters as input information into a transient disturbance prediction model to obtain an output result, wherein the output result includes one or more predicted transient disturbances; a compensation module, used to extract any predicted transient disturbance from the one or more predicted transient disturbances and match any predetermined compensation action of the arbitrary predicted transient disturbance in a predetermined compensation sheet; and an execution module, used to activate a collaborative compensation mechanism to execute the arbitrary predetermined compensation action, thereby realizing temperature and humidity prediction and compensation for the target developing chamber.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages: Target multi-source information of the target developing chamber is obtained through dynamic continuous monitoring. This target multi-source information includes target process timing and target state parameters. The target process timing and target state parameters are used as input information and fed into a transient disturbance prediction model to obtain output results, which include one or more predicted transient disturbances. Any predicted transient disturbance from the one or more predicted transient disturbances is extracted and matched with any predetermined compensation action in a predetermined compensation sheet. A collaborative compensation mechanism is activated to execute the arbitrary predetermined compensation action, thereby achieving temperature and humidity prediction compensation for the target developing chamber. In other words, multi-source dynamic information during the developing process is first collected in real time, and transient disturbances caused by robotic arm movement, chamber cover opening and closing, and developer nozzle movement are predicted in advance. Then, a feedforward compensation control signal is generated and executed before the actual disturbance occurs. Finally, closed-loop self-tuning is performed, and the model is periodically calibrated through offline testing, thereby achieving precise temperature and humidity control. It achieves integrated control of transient disturbances in the developing chamber, significantly reducing the amplitude and recovery time of temperature and humidity fluctuations caused by mechanical actions, effectively reducing the incidence of wafer edge defects, thereby improving the yield of the developing process and the utilization rate of equipment.

[0009] The above description is merely an overview of the technical solution of this application. To better understand the technical means of this application and to facilitate its implementation according to the description, and to make the above and other objects, features, and advantages of this application more apparent, specific embodiments of this application are described below. It should be understood that the content described in this section is not intended to identify key or important features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent through the following description. Attached Figure Description

[0010] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0011] Figure 1 This is a schematic flowchart of a temperature and humidity prediction and compensation method for suppressing transient disturbances in a developing chamber, as described in this application. Figure 2 This is a schematic diagram of the steps following the implementation of temperature and humidity prediction compensation for the target developing chamber in the temperature and humidity prediction compensation method for suppressing transient disturbances in the developing chamber according to this application. Figure 3 This is a schematic diagram of the temperature and humidity prediction and compensation system for suppressing transient disturbances in the developing chamber, as described in this application.

[0012] Explanation of reference numerals in the attached diagram: Monitoring module 11, Prediction module 12, Compensation module 13, Execution module 14. Detailed Implementation

[0013] This application provides a method and system for predicting and compensating for transient disturbances in the developing chamber, solving the problem of accurately predicting the type, intensity, and affected area of ​​transient disturbances in the developing process before they actually occur, and performing collaborative pre-compensation accordingly. Simultaneously, a closed-loop self-tuning mechanism continuously optimizes the accuracy of the prediction model, thereby fundamentally suppressing wafer temperature and humidity deviations and edge defects caused by transient disturbances in the chamber and nozzle dripping. The technical goal is to accurately predict the type, intensity, and affected area of ​​transient disturbances before they actually occur, and to perform collaborative feedforward compensation control at multiple physical levels, including airflow barriers, local heating, developer temperature control, and nozzle backflow. This achieves the technical effects of significantly suppressing the amplitude and recovery time of transient temperature and humidity fluctuations in the developing chamber, effectively reducing wafer edge defects, improving developing process yield, and enhancing the long-term robustness and control accuracy of the equipment.

[0014] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. It should be understood that this application is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application. It should also be noted that, for ease of description, only the parts related to this application are shown in the accompanying drawings, not all of them.

[0015] Example 1, please refer to the appendix. Figure 1 This application provides a temperature and humidity prediction and compensation method for suppressing transient disturbances in a developing chamber. The method is applied to a temperature and humidity prediction and compensation system for suppressing transient disturbances in a developing chamber. The specific steps of the temperature and humidity prediction and compensation method for suppressing transient disturbances in a developing chamber are as follows: Dynamic continuous monitoring obtains target multi-source information of the target developing chamber, wherein the target multi-source information includes target process timing and target state parameters; the target process timing and target state parameters are used as input information and input to the transient disturbance prediction model to obtain output results, wherein the output results include one or more predicted transient disturbances; any predicted transient disturbance is extracted from the one or more predicted transient disturbances, and any predetermined compensation action of the any predicted transient disturbance is matched in a predetermined compensation sheet; the collaborative compensation mechanism is activated to execute the any predetermined compensation action to realize temperature and humidity prediction compensation for the target developing chamber.

[0016] Specifically, the temperature and humidity prediction and compensation method for suppressing transient disturbances in the developing chamber is applied to a temperature and humidity prediction and compensation system for suppressing transient disturbances in the developing chamber. It can achieve the technical goal of accurately predicting and executing collaborative feedforward compensation control before the transient disturbance actually occurs.

[0017] First, during the developing process, various sensors or control system interfaces deployed inside and outside the developing chamber are used to dynamically acquire target process timing and overall state parameters that reflect the current developing process node of the target developing chamber, forming target multi-source information. The target developing chamber refers to the developing chamber for which the environmental temperature and humidity prediction and compensation will be performed using the technical solution in this application. An example is continuously acquiring data with millisecond or sub-second time resolution throughout the entire developing process, such as recording the temperature of the developing chamber every 10 milliseconds. The target process timing refers to a list of developing process sequences based on real-time monitoring moments and their corresponding developing process actions. An example is when the developer nozzle moves from its starting position to the wafer center at moment a, and the chamber cover begins to close at moment b. Target state parameters refer to state parameters formed based on the environmental parameters of each chamber at the real-time monitoring moment. An example is the temperature, humidity, and airflow velocity at various locations within the developing chamber. Dynamic continuous monitoring provides accurate and real-time perception of the current state and upcoming actions of the developing chamber, providing a complete, synchronous, and high-resolution input data foundation for subsequent transient disturbance prediction, thus avoiding prediction distortion caused by missing or delayed information.

[0018] Then, the monitored target process time sequence is converted into a code that the model can understand. Simultaneously, the state parameters (temperature, humidity, pressure, etc.) collected by multiple sensors are concatenated into a fixed-length state vector. For sensors with multiple locations, their spatial distribution information is preserved by arranging the corresponding location information according to a predetermined positional order. The formatted time sequence code and state vector are then input into a transient disturbance prediction model with pre-loaded training weights. The model performs a forward propagation calculation, outputting the prediction result within milliseconds. The model's prediction result is then inversely normalized to obtain the specific disturbance. The transient disturbance prediction model is an intelligent model trained on a deep learning-based mathematical model using historical data sets containing past process time sequences, corresponding state parameters, and actual disturbance results. In other words, this transient disturbance prediction model internally learns a nonlinear mapping relationship where "a specific action in a specific environment will trigger a specific disturbance," and the model can simultaneously predict multiple upcoming disturbances. For example, opening the cover in 0.5 seconds will cause a cold airflow, while moving the nozzle in 1 second will cause a localized increase in humidity. By utilizing a trained transient disturbance prediction model, intelligent prediction can quickly determine what transient temperature, humidity, or airflow disturbances will occur in the developing chamber after these timing actions are performed in the current state. For example, the model's input information includes a process timing of "cover plate opens after 0.5 seconds" and state parameters of "current chamber top temperature 23.2℃, bottom temperature 22.9℃, wafer edge airflow velocity 0.3m / s". The transient disturbance prediction model calculates and outputs two predicted transient disturbances: the first disturbance is cold airflow intrusion, with a predicted chamber temperature drop of 0.4℃, affecting a 3050mm annular zone at the wafer edge; the second disturbance is turbulent disturbance, with a predicted chamber airflow pressure fluctuation of ±0.8Pa, affecting the wafer surface from the center outwards.

[0019] Next, multiple transient disturbances predicted by the transient disturbance prediction model are sequentially compensated and matched in the predetermined compensation list to obtain the corresponding compensation actions, i.e., any predetermined compensation action corresponding to any predicted transient disturbance. The predetermined compensation list is a rule base or lookup table pre-defined through process experiments or expert knowledge, containing multiple "conditional action" pairs. The condition part of each "conditional action" pair describes the type, intensity level, and regional characteristics of the predicted disturbance, while the action part describes the actuator and specific parameters to be activated. In other words, the specific compensation action to be performed is pre-defined for each type, intensity range, and affected area of ​​the predicted disturbance. For example, extracting the first disturbance mentioned above and querying the predetermined compensation list reveals that for the disturbance "cold airflow intrusion, intensity ≥0.3℃ and <0.6℃, affecting the wafer edge region," the corresponding pre-defined compensation action is to close the passive exhaust vents on both sides of the bottom of the chamber, while simultaneously increasing the opening of the main air supply valve from 40% to 75% and maintaining it for 2 seconds. By using pre-defined compensation sheets, complex process experience is transformed into standardized control strategies, ensuring that each predicted disturbance can be quickly and accurately matched with the most appropriate compensation method.

[0020] Finally, the system sends execution commands to the collaborative compensation mechanism to execute any predetermined compensation action. The collaborative compensation mechanism includes multiple actuators such as a proportional control valve, heating array, temperature control component, backflow control valve, and electric proportional valve. These actuators work collaboratively, simultaneously or sequentially, to perform their respective actions, jointly resisting upcoming transient disturbances and achieving proactive, multi-level compensation for the temperature and humidity of the developing chamber. The system activates the collaborative compensation mechanism by sending execution commands; the specific activation time is determined by the predicted disturbance time and the time advance required for the compensation action. For example, 0.3 seconds before the cover opens, a command is sent to the proportional control valve to adjust the opening from 40% to 70%, and 0.1 seconds before the cover opens, a command is sent to the heating arrays (numbered 3 and 4) located below the wafer edge to increase the heating power by 8% for 1.2 seconds. After executing these compensation actions, when the cover actually opens and cold air attempts to enter, it is blocked by the air curtain, and the temperature in the edge area is neutralized by the pre-increased heating power, controlling the actual temperature fluctuation within ±0.1℃.

[0021] Further, the target process timing and the target state parameters are used as input information and input into the transient disturbance prediction model to obtain the output result. This includes: acquiring historical development records of the developing chamber; extracting a first record from the historical development records and constructing a first dataset based on the first process timing, first state parameters, and first transient disturbance in the first record; and performing supervised training on the first dataset to obtain the transient disturbance prediction model. The first process timing includes robotic arm movement, chamber cover opening and closing, developer nozzle movement, and wafer stage rotation. The first state parameters include temperature field distribution, humidity distribution, and wafer surface airflow pressure. The first transient disturbance includes fluctuations in wafer surface airflow pressure caused by robotic arm movement and / or wafer stage rotation, abrupt changes in temperature field distribution caused by chamber cover opening and closing, and abrupt changes in humidity distribution caused by developer nozzle movement.

[0022] Specifically, by connecting to the historical data server or local log database of the coating and developing equipment, a historical operational database of the equipment is obtained. A large number of complete and traceable developing process records are exported and organized, and then "sampled" to transform them into training samples that can be directly used for supervised learning. The training samples are then fed into a deep learning network using supervised learning. The network repeatedly compares its predicted output with the actual historical disturbances, ultimately ensuring that for any given process timing and state parameter input, it outputs a transient disturbance that highly approximates the actual occurrence. The trained network then becomes the transient disturbance prediction model. To ensure sufficient generalization ability, continuous operational data of at least three months to one year is selected, covering different seasons, maintenance cycles, and batches of wafer products. The first record refers to any record in the historical developing records; in data engineering, it is a sample instance. The term "first" is used generically, not specifically referring to the first record in chronological order, and this first record has undergone sampled processing. Next, the first process sequence, first state parameters, and first transient disturbance are extracted from the first record and formed into a data triplet, denoted as the first dataset. The first process sequence refers to the list of developing process actions extracted from the first record, arranged chronologically. Exemplary examples include the movement of the robotic arm, the opening and closing of the chamber cover, the movement and spraying of the developer nozzle, and the rotation acceleration and deceleration of the wafer stage. The first state parameters refer to the environmental quantification data extracted from the first record for the time period before and after the occurrence of the time-series actions. Exemplary examples include temperature sensor readings at multiple different spatial locations within the developing chamber. The first transient disturbance refers to the actual observed disturbances such as abnormal changes in temperature, humidity, or airflow caused by the aforementioned actions, extracted from the first record. An exemplary example is a local temperature drop of more than 0.3°C within 0.5 seconds due to the opening and closing of the chamber cover. Next, the first dataset is input into a deep learning network, such as a Long Short-Term Memory network or a gated recurrent unit, using supervised learning. The transient disturbance prediction model is obtained by repeatedly comparing the difference between the output predicted disturbance and the actual disturbance. The technical objective is to obtain a transient disturbance prediction model by training it using historical development records, thereby achieving real-time and accurate prediction of future transient disturbances in the developing chamber and providing a reliable decision-making basis for subsequent feedforward compensation.

[0023] Furthermore, the transient disturbance prediction model is obtained by supervised training on the first dataset, including: encoding the first process timing and the first state parameters to obtain a first encoding vector; and learning the first encoding vector and the first transient disturbance based on a gated recurrent unit or a long short-term memory network to obtain the transient disturbance prediction model.

[0024] Specifically, by state encoding the first process timing and the first state parameters, numerical information that the machine learning model can understand and process is obtained, namely the first encoded vector. The first process timing includes discrete action categories such as cover opening, and the first state parameters include sensor readings such as temperature, wind speed, and humidity. Since the data types are different and cannot be directly used for model training, state encoding is performed to map all heterogeneous data into a fixed-length, continuous numerical vector space. Next, a gated recurrent unit or long short-term memory network with memory capabilities is used to learn the causal relationship between the encoded vector and transient disturbances. After training, the transient disturbance prediction model has the ability to predict future transient disturbances based on the encoded vectors from the current and several past moments.

[0025] Furthermore, the predetermined compensation order includes an airflow compensation action for compensating for fluctuations in airflow pressure on the wafer surface, a temperature compensation action for compensating for abrupt changes in temperature field distribution, and a humidity compensation action for compensating for abrupt changes in humidity distribution. The airflow compensation action refers to forming a target air curtain through a proportional regulating valve in the collaborative compensation mechanism, wherein the target air curtain is used to resist fluctuations in airflow pressure on the wafer surface caused by the movement of the robotic arm and / or the rotation of the wafer stage. The temperature compensation action refers to adjusting the heating power through the heating array in the collaborative compensation mechanism to form a target power, wherein the target power refers to the compensation heating power at the corresponding point where the temperature field distribution changes due to the opening and closing of the chamber cover. The humidity compensation action refers to adjusting the developer temperature through the temperature control component in the collaborative compensation mechanism to form a target temperature, wherein the target temperature refers to the compensation supply temperature for abrupt changes in temperature distribution caused by the movement of the developer nozzle.

[0026] Specifically, the predetermined compensation order includes three core compensation actions for compensating for airflow pressure, temperature field, and humidity: airflow compensation for the proportional control valve in the collaborative compensation mechanism, temperature compensation for the heating array in the collaborative compensation mechanism, and humidity compensation for the temperature control component in the collaborative compensation mechanism. The collaborative compensation mechanism refers to a collection of multiple actuators pre-installed within the coating and developing equipment.

[0027] The airflow compensation action is a dedicated strategy in the pre-defined compensation plan to address airflow pressure fluctuations on the wafer surface. Its core objective is to proactively create a directional, controllable target air curtain within the chamber by adjusting the proportional control valve on the air supply duct when airflow pressure fluctuations due to robotic arm movement or wafer stage rotation are anticipated. This target air curtain acts like an air barrier, offsetting or weakening the impending pressure fluctuations and maintaining a stable airflow environment on the wafer surface. The temperature compensation action is another core strategy in the pre-defined compensation plan to address abrupt temperature field changes caused by the opening and closing of the chamber cover. When it is predicted that external cold or warm air will intrude into a localized area of ​​the chamber when the cover opens or closes, causing a sudden drop or rise in temperature in that area, the temperature compensation action precisely adjusts the heating power of corresponding points in the multi-point zoned heating array located below the affected area before the disturbance occurs. This proactively heats or reduces heating to offset the anticipated temperature change and maintain transient local temperature stability. Humidity compensation is a dedicated strategy in the pre-defined compensation plan to address abrupt changes in humidity distribution caused by the movement of the developer nozzle. When the developer nozzle moves within the chamber and sprays developer, the evaporation of water or organic solvents in the developer causes a sharp increase in local humidity. Furthermore, humidity changes indirectly affect the development quality by influencing heat exchange on the wafer surface and the chemical reaction rate of the photoresist. By pre-adjusting the temperature of the developer supplied to the nozzle, the evaporation rate is actively controlled using the relationship between temperature and saturated vapor pressure, thereby suppressing the magnitude of humidity fluctuations.

[0028] Furthermore, the first state parameter also includes the droplet back-suction height at the end of the developer nozzle, which is monitored in real time.

[0029] Furthermore, the first transient disturbance also includes the droplet ...

[0030] Furthermore, when the droplet ...

[0031] Furthermore, the collaborative compensation mechanism also includes an electric proportional valve, which adjusts the instantaneous exhaust volume at the starting position of the developing arm to form a target instantaneous exhaust volume. The target instantaneous exhaust volume is used to synchronously extract suspended microdroplets excited around the nozzle by the target instantaneous pressurization pulse signal.

[0032] Specifically, in addition to the existing monitoring of temperature, humidity, and airflow pressure, real-time monitoring of the droplet back-suction height at the tip of the developer nozzle has been added. By continuously monitoring the droplet back-suction height at the tip of the developer nozzle, the risk of dripping can be detected in advance, thus providing data for subsequent active anti-drip compensation. The tip of the developer nozzle refers to the outlet tip of the developer spraying device. After the spraying action is completed, the developer residue inside the nozzle will form a visible liquid column or surface at the tip due to surface tension. The droplet back-suction height refers to the distance that the liquid surface at the tip of the nozzle retracts relative to the nozzle orifice plane after the spraying action is completed, caused by the back-suction mechanism. For example, if the liquid surface is level with the nozzle orifice, the back-suction height is 0; if the liquid surface retracts 1mm into the nozzle orifice, the back-suction height is 1mm. The preset standard value for the back-suction height is usually 1-2mm, that is, the liquid surface is lower than the nozzle orifice. By installing a non-contact optical sensor in the developer nozzle's supply line or on the side of the nozzle tip, or by indirectly calculating the backflow height using a pressure / flow sensor, and combining this with parameters such as nozzle movement speed, movement distance, and terminal acceleration, the probability of a suspended droplet being generated at the nozzle's endpoint due to incomplete backflow and falling onto the wafer surface is calculated; this is known as the droplet droplet drop probability. The droplet drop probability is used to trigger subsequent anti-drip compensation actions.

[0033] When the predicted droplet droplet droplet droplet droplet droplet droplet probability exceeds a predetermined limit (e.g., greater than 70%), the nozzle back-suction control valve is activated, sending a short-duration, high-pressure instantaneous boost pulse signal. This pulse signal instantly increases the negative pressure of the back-suction mechanism, further drawing back the previously insufficiently back-suctioned liquid surface, rapidly raising the droplet droplet back-suction height from its current value to a preset safe distance inside the nozzle orifice. The back-suction control valve is a proportional valve or on / off valve located in the developer supply line, typically connected to a nitrogen or compressed air source. The magnitude and duration of the back-suction negative pressure can be adjusted by controlling the valve opening or pulse width. Furthermore, when the back-suction control valve sends the instantaneous boost pulse, the powerful negative pressure impact not only draws back the main droplet droplet but may also generate extremely small suspended microdroplets around the nozzle. If these microdroplets are not removed in time, they may slowly settle onto the wafer surface during subsequent development arm movement, similarly causing fine particle defects. Therefore, by simultaneously adjusting the electric proportional valve on the exhaust pipe at the starting position of the developing arm, the instantaneous exhaust volume at that location is increased from the baseline value to a higher target instantaneous exhaust volume, creating a localized negative pressure suction zone. This rapidly removes the newly generated suspended micro-droplets, achieving thorough cleaning. The electric proportional valve is a valve whose opening can be continuously controlled by an electrical signal, used to regulate the flow rate of the exhaust system, enabling stepless adjustment of the exhaust volume.

[0034] By leveraging the combined action of the backflow control valve and the electric proportional valve, suspended microdroplets accidentally generated by the boost pulse are completely eliminated, achieving comprehensive and clean treatment from suspended droplets to suspended droplets, ensuring that the wafer surface is not contaminated by any form of droplet.

[0035] Furthermore, such as Figure 2 As shown, the activated collaborative compensation mechanism executes any predetermined compensation action to achieve temperature and humidity prediction compensation for the target developing chamber, and then includes: Activate the high-precision temperature and humidity sensor array to acquire temperature and humidity timing data; The residual error sequence is obtained by comparing the temperature and humidity time series with the preset steady-state value; The real-time residual error and the rate of change of residual error obtained from analyzing the residual error sequence are input into the self-tuning controller to obtain the real-time weights; The output of the transient disturbance prediction model is adjusted based on the real-time weights.

[0036] Specifically, a high-precision temperature and humidity sensor array deployed at different spatial locations within the developing chamber is activated within a very short time after a transient disturbance occurs. Temperature and humidity values ​​are continuously collected to form a temperature and humidity time series. The high-precision temperature and humidity sensor array refers to a network composed of multiple independent high-precision temperature and humidity sensors arranged in a specific spatial layout. This temperature and humidity time series is then compared point-by-point with the pre-set steady-state temperature and humidity values ​​that the developing chamber should maintain under ideal conditions, i.e., the preset steady-state values. The deviation at the corresponding sampling time is obtained by subtracting the steady-state value from the actual value. The deviations at all times are arranged chronologically to form the residual error sequence. This precisely quantifies the gap between feedforward compensation and the actual effect, identifying the regions and time points of insufficient compensation corresponding to negative residual errors and overcompensation corresponding to positive residual errors. Next, the residual error sequence is analyzed to obtain the real-time residual error and the rate of change of residual error. A self-tuning controller is then used to calculate the real-time weights for correcting the output of the prediction model, making the next prediction more accurate. The self-tuning controller is a fuzzy PID controller optimized using a genetic algorithm. It first optimizes the fuzzy rule base offline using a genetic algorithm, and then, during runtime, queries the rule table online and calculates the output based on the input error and its rate of change. The self-tuning controller dynamically calculates the optimal adjustment amount based on the magnitude and trend of the current compensation deviation. This provides a quantitative and precise control quantity for identifying deficiencies in feedforward compensation, generating targeted correction weights, and adjusting the predictive model output. The real-time weight refers to the correction coefficient output by the self-tuning controller, with the sign indicating whether the original predicted value needs to be increased or decreased, and the absolute value of the weight representing the magnitude of the correction. The adjusted predicted value is stored in the model's memory or configuration file. In subsequent predictive inference, for the same or similar input conditions, the model will output the corrected value instead of the original trained value.

[0037] In summary, the temperature and humidity prediction and compensation method for suppressing transient disturbances in the developing chamber provided in this application has the following technical effects: Target multi-source information of the target developing chamber is obtained through dynamic continuous monitoring. This target multi-source information includes target process timing and target state parameters. The target process timing and target state parameters are used as input information and fed into a transient disturbance prediction model to obtain output results, which include one or more predicted transient disturbances. Any predicted transient disturbance from the one or more predicted transient disturbances is extracted and matched with any predetermined compensation action in a predetermined compensation sheet. A collaborative compensation mechanism is activated to execute the arbitrary predetermined compensation action, thereby achieving temperature and humidity prediction compensation for the target developing chamber. In other words, multi-source dynamic information during the developing process is first collected in real time, and transient disturbances caused by robotic arm movement, chamber cover opening and closing, and developer nozzle movement are predicted in advance. Then, a feedforward compensation control signal is generated and executed before the actual disturbance occurs. Finally, closed-loop self-tuning is performed, and the model is periodically calibrated through offline testing, thereby achieving precise temperature and humidity control. It achieves integrated control of transient disturbances in the developing chamber, significantly reducing the amplitude and recovery time of temperature and humidity fluctuations caused by mechanical actions, effectively reducing the incidence of wafer edge defects, thereby improving the yield of the developing process and the utilization rate of equipment.

[0038] Example 2: Based on the same inventive concept as the temperature and humidity prediction and compensation method for suppressing transient disturbances in the developing chamber described in the foregoing examples, this application also provides a temperature and humidity prediction and compensation system for suppressing transient disturbances in the developing chamber. Please refer to the appendix. Figure 3 The temperature and humidity prediction and compensation system for suppressing transient disturbances in a developing chamber includes: a monitoring module 11, which dynamically and continuously monitors and obtains target multi-source information of the target developing chamber, wherein the target multi-source information includes target process timing and target state parameters; a prediction module 12, which takes the target process timing and the target state parameters as input information and inputs them into a transient disturbance prediction model to obtain an output result, wherein the output result includes one or more predicted transient disturbances; a compensation module 13, which extracts any predicted transient disturbance from the one or more predicted transient disturbances and matches it with any predetermined compensation action in a predetermined compensation sheet; and an execution module 14, which activates a collaborative compensation mechanism to execute the arbitrary predetermined compensation action to achieve temperature and humidity prediction and compensation for the target developing chamber.

[0039] Furthermore, the prediction module 12 in the temperature and humidity prediction compensation system for suppressing transient disturbances in the developing chamber is further configured to: acquire historical developing records of the developing chamber; extract a first record from the historical developing records, and construct a first dataset based on the first process sequence, first state parameters, and first transient disturbance in the first record; perform supervised training on the first dataset to obtain the transient disturbance prediction model; wherein, the first process sequence includes robotic arm movement, chamber cover opening and closing, developer nozzle movement, and wafer stage rotation; wherein, the first state parameters include temperature field distribution, humidity distribution, and wafer surface airflow pressure; wherein, the first transient disturbance includes fluctuations in wafer surface airflow pressure caused by robotic arm movement and / or wafer stage rotation, abrupt changes in temperature field distribution caused by chamber cover opening and closing, and abrupt changes in humidity distribution caused by developer nozzle movement.

[0040] Furthermore, the prediction module 12 in the temperature and humidity prediction compensation system for suppressing transient disturbances in the developing chamber is also used to: perform state encoding on the first process timing and the first state parameters to obtain a first encoding vector; and learn the first encoding vector and the first transient disturbance based on a gated recurrent unit or a long short-term memory network to obtain the transient disturbance prediction model.

[0041] Furthermore, the compensation module 13 in the temperature and humidity prediction compensation system for suppressing transient disturbances in the developing chamber is also used for: the predetermined compensation order includes an airflow compensation action for compensating for fluctuations in airflow pressure on the wafer surface, a temperature compensation action for compensating for sudden changes in temperature field distribution, and a humidity compensation action for compensating for sudden changes in humidity distribution; wherein, the airflow compensation action refers to forming a target air curtain through a proportional regulating valve in the collaborative compensation mechanism, wherein the target air curtain is used to resist fluctuations in airflow pressure on the wafer surface caused by the movement of the robotic arm and / or the rotation of the wafer stage; wherein, the temperature compensation action refers to adjusting the heating power through the heating array in the collaborative compensation mechanism to form a target power, wherein the target power refers to the compensation heating power at the corresponding point of sudden changes in temperature field distribution caused by the opening and closing of the chamber cover; wherein, the humidity compensation action refers to adjusting the developer temperature through the temperature control component in the collaborative compensation mechanism to form a target temperature, wherein the target temperature refers to the compensation supply temperature for sudden changes in temperature distribution caused by the movement of the developer nozzle.

[0042] Furthermore, the prediction module 12 in the temperature and humidity prediction compensation system for suppressing transient disturbances in the developing chamber is also used to: the first state parameter further includes the droplet back suction height at the end of the developing solution nozzle, which is monitored in real time.

[0043] Furthermore, the prediction module 12 in the temperature and humidity prediction compensation system for suppressing transient disturbances in the developing chamber is also used for: the first transient disturbance further includes the droplet ...

[0044] Furthermore, the prediction module 12 in the temperature and humidity prediction compensation system for suppressing transient disturbances in the developing chamber is also used to: when the droplet drop probability reaches a predetermined limit, activate the back suction control valve in the collaborative compensation mechanism to send a target instantaneous pressure boosting pulse signal, wherein the target instantaneous pressure boosting pulse signal is used to raise the droplet back suction height to a preset safe distance inside the nozzle orifice, thereby physically eliminating the suspended droplets.

[0045] Furthermore, the prediction module 12 in the temperature and humidity prediction compensation system for suppressing transient disturbances in the developing chamber is also used for: the collaborative compensation mechanism further includes an electric proportional valve, which adjusts the instantaneous exhaust volume at the starting position of the developing arm to form a target instantaneous exhaust volume, wherein the target instantaneous exhaust volume is used to synchronously extract suspended microdroplets excited around the nozzle by the target instantaneous pressurization pulse signal.

[0046] Furthermore, the execution module 14 in the temperature and humidity prediction compensation system for suppressing transient disturbances in the developing chamber is also used to: activate a high-precision temperature and humidity sensor array to acquire temperature and humidity time series; compare the temperature and humidity time series with a preset steady-state value to obtain a residual error sequence; input the real-time residual error and residual error change rate obtained by analyzing the residual error sequence to a self-tuning controller to obtain real-time weights; and adjust the output result of the transient disturbance prediction model based on the real-time weights.

[0047] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Figure 1 The temperature and humidity prediction and compensation method and specific examples for suppressing transient disturbances in the developing chamber in Example 1 are also applicable to the temperature and humidity prediction and compensation system for suppressing transient disturbances in the developing chamber in this embodiment. Through the foregoing detailed description of the temperature and humidity prediction and compensation method for suppressing transient disturbances in the developing chamber, those skilled in the art can clearly understand the temperature and humidity prediction and compensation system for suppressing transient disturbances in the developing chamber in this embodiment. Therefore, for the sake of brevity, it will not be described in detail here. For the system disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant details can be found in the method section.

[0048] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0049] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application also intends to include such modifications and variations.

Claims

1. A temperature and humidity prediction compensation method for suppressing transient disturbance in a developing chamber, characterized by, The method comprises the following steps: obtaining target multi-source information of a target developing chamber through dynamic continuity monitoring, wherein the target multi-source information comprises target process timing and target state parameters; inputting the target process timing and the target state parameters as input information into a transient disturbance prediction model to obtain an output result, wherein the output result comprises one or more predicted transient disturbances; extracting any predicted transient disturbance from the one or more predicted transient disturbances and matching any predetermined compensation action of the any predicted transient disturbance in a predetermined compensation table; activating a cooperative compensation mechanism to perform the any predetermined compensation action, so as to realize temperature and humidity compensation prediction of the target developing chamber.

2. The temperature and humidity prediction compensation method for suppressing transient disturbance in a developing chamber according to claim 1, wherein, The method further comprises the following steps before inputting the target process timing and the target state parameters as input information into the transient disturbance prediction model to obtain an output result: obtaining historical developing records of the developing chamber; extracting a first record from the historical developing records and establishing a first data set based on a first process timing, a first state parameter and a first transient disturbance in the first record; performing supervised training on the first data set to obtain the transient disturbance prediction model; wherein the first process timing comprises robot arm movement, chamber cover opening and closing, developing liquid nozzle movement and wafer table rotation; wherein the first state parameter comprises temperature field distribution, humidity distribution and wafer surface airflow pressure; wherein the first transient disturbance comprises fluctuation of wafer surface airflow pressure caused by robot arm movement and / or wafer table rotation, mutation of temperature field distribution caused by chamber cover opening and closing, and mutation of humidity distribution caused by developing liquid nozzle movement.

3. The temperature and humidity prediction compensation method for suppressing transient disturbance in a developing chamber according to claim 2, wherein, The method further comprises the following steps of performing supervised training on the first data set to obtain the transient disturbance prediction model: performing state coding on the first process timing and the first state parameter to obtain a first coding vector; learning the first coding vector and the first transient disturbance based on a gated recurrent unit or a long short-term memory network to obtain the transient disturbance prediction model.

4. The temperature and humidity prediction and compensation method for suppressing transient disturbances in the developing chamber as described in claim 1, characterized in that, The predetermined compensation table comprises airflow compensation action for compensating wafer surface airflow pressure fluctuation, temperature compensation action for compensating temperature field distribution mutation, and humidity compensation action for compensating humidity distribution mutation; wherein the airflow compensation action refers to forming a target air curtain through a proportional adjusting valve in the cooperative compensation mechanism, wherein the target air curtain is used to resist wafer surface airflow pressure fluctuation caused by robot arm movement and / or wafer table rotation; wherein the temperature compensation action refers to adjusting heating power through a heating array in the cooperative compensation mechanism to form a target power, wherein the target power refers to a compensation heating power of a corresponding point caused by mutation of temperature field distribution due to chamber cover opening and closing; wherein the humidity compensation action refers to adjusting developing liquid temperature through a temperature control component in the cooperative compensation mechanism to form a target temperature, wherein the target temperature refers to a compensation supply temperature caused by temperature distribution mutation due to developing liquid nozzle movement.

5. The method of claim 2, wherein the temperature and humidity prediction compensation method for suppressing transient disturbance in a developing chamber is characterized by, The first state parameter further comprises a droplet back absorption height of an end portion of the developing liquid nozzle monitored in real time.

6. The method of claim 5, wherein the temperature and humidity prediction compensation method for suppressing transient disturbance in a developing chamber is characterized by, The first transient disturbance further comprises a liquid drop shedding probability of the end of the developer nozzle, wherein the liquid drop shedding probability refers to a shedding probability of a hanging liquid drop generated at the moving end of the developer nozzle due to incomplete back suction.

7. The temperature and humidity prediction and compensation method for suppressing transient disturbances in the developing chamber as described in claim 6, characterized in that, When the liquid drop shedding probability reaches a predetermined limit value, a back suction control valve in the cooperative compensation mechanism is activated to issue a target transient pressure pulse signal, wherein the target transient pressure pulse signal is used to lift the liquid drop back suction height to a preset safety distance inside the nozzle port to physically eliminate the hanging liquid drop.

8. The method of claim 7, wherein the temperature and humidity prediction compensation is performed to suppress transient disturbances in the development chamber. The cooperative compensation mechanism further comprises an electric proportional valve, through which the transient exhaust amount of the developer arm starting position is adjusted to form a target transient exhaust amount, wherein the target transient exhaust amount is used to synchronously extract the suspended micro-liquid drops caused by the target transient pressure pulse signal around the nozzle.

9. The temperature and humidity prediction and compensation method for suppressing transient disturbances in the developing chamber as described in claim 1, characterized in that, The cooperative compensation mechanism is activated to perform the arbitrary predetermined compensation action to achieve the temperature and humidity prediction compensation of the target developing chamber, and then further comprises: activating a high-precision temperature and humidity sensor array to obtain a temperature and humidity time sequence; comparing the temperature and humidity time sequence with a preset steady-state value to obtain a residual error sequence; inputting real-time residual error and residual error change rate obtained by analyzing the residual error sequence into a self-tuning controller to obtain a real-time weight; adjusting the output result of the transient disturbance prediction model based on the real-time weight.

10. A temperature and humidity prediction compensation system for suppressing transient disturbances in a developing chamber, characterized by, A system for implementing the steps of the temperature and humidity prediction compensation method for suppressing transient disturbance of a developing chamber according to any one of claims 1 to 9, the system comprising: a monitoring module for dynamically and continuously monitoring target multi-source information of a target developing chamber, wherein the target multi-source information comprises target process time sequence and target state parameters; a prediction module for inputting the target process time sequence and the target state parameters as input information into a transient disturbance prediction model to obtain an output result, wherein the output result comprises one or more predicted transient disturbances; a compensation module for extracting any predicted transient disturbance from the one or more predicted transient disturbances and matching any predetermined compensation action of the any predicted transient disturbance in a predetermined compensation unit; an execution module for activating a cooperative compensation mechanism to perform the arbitrary predetermined compensation action to achieve the temperature and humidity prediction compensation of the target developing chamber.

Images