A method for adjusting the vacuum level of ITO coating on a chip

Abstract

This invention relates to the field of semiconductor thin film deposition process control, specifically a method for adjusting the vacuum level of ITO chip deposition. The method includes: constructing a multi-source sensing input layer to collect and integrate macroscopic vacuum level data, target discharge electrical parameters, plasma spectral characteristic parameters, and real-time wafer temperature; constructing a microenvironment digital twin layer to calculate transient oxygen consumption rate and predict the local true oxygen partial pressure on the wafer surface; and constructing a collaborative decision-making and execution layer to adjust the local reactive gas flow rate of the gas inlet device and the opening of the main extraction valve of the chamber based on the deviation between the local true oxygen partial pressure and the target oxygen partial pressure using a feedforward feedback collaborative control mechanism, and performing independent mathematical calculations through action decoupling. This invention achieves direct constraint on the true oxygen partial pressure of the microenvironment on the wafer surface, improving the stability and batch consistency of the ITO thin film's optoelectronic performance.

Description

Technical Field

[0001] This invention relates to the field of semiconductor thin film deposition process control, specifically a method for adjusting the vacuum degree of ITO chip deposition. Background Technology

[0002] In the process of chip ITO coating, vacuum adjustment is usually a traditional control method that revolves around the total pressure of the chamber. Generally, the overall pressure of the chamber is detected by a vacuum gauge, and the process atmosphere is adjusted in conjunction with the air intake device and the air extraction valve to maintain the stability of the coating process. This method can ensure that the macroscopic pressure of the chamber is within the set range to a certain extent and meet the basic requirements of conventional reactive sputtering processes for the vacuum environment.

[0003] As chip devices increasingly demand higher requirements for ITO thin film sheet resistance, transmittance, and batch consistency, adjusting solely based on the total chamber pressure is no longer sufficient to accurately reflect the true oxygen partial pressure state in the microenvironment near the wafer surface. Especially during reactive sputtering, target oxygen consumption, plasma state changes, and wafer temperature disturbances all affect local oxygen supply. Traditional adjustment methods are prone to local oxygen deficiency or enrichment due to sensing lag and limited information, which in turn leads to an imbalance in the film stoichiometry and affects the stability of film performance. Summary of the Invention

[0004] The purpose of this invention is to provide a method for adjusting the vacuum degree of ITO coating on a chip, which solves the following technical problems: avoiding film quality fluctuations caused by simple adjustment and control lag around the total pressure of the chamber, and directly constraining the stoichiometry of the ITO film, thereby improving the stability of the photoelectric performance and batch consistency of the film.

[0005] The objective of this invention can be achieved through the following technical solutions:

[0006] A method for adjusting the vacuum level in ITO chip coating is applied to a coating system including a chamber, a main chamber exhaust valve, an air inlet device with a local gas distributor, a chamber pressure detection unit, a target power supply, a plasma spectral acquisition unit, and a wafer temperature acquisition unit, comprising:

[0007] A multi-source sensing input layer is constructed to collect and integrate macroscopic vacuum data, target discharge electrical parameters, plasma spectral characteristic parameters, and real-time wafer temperature.

[0008] A microenvironment digital twin layer is constructed to calculate the microenvironment state of the wafer surface based on the integrated data. The microenvironment digital twin layer includes a consumption rate calculation module and a partial pressure prediction module. The consumption rate calculation module calculates the transient oxygen consumption rate, and the partial pressure prediction module predicts the local true oxygen partial pressure.

[0009] Construct a collaborative decision-making and execution layer, set up a feedforward feedback collaborative control mechanism, and adjust the local reaction gas flow rate of the air intake device and the opening degree of the main exhaust valve of the chamber according to the feedforward feedback collaborative control mechanism.

[0010] The feedforward feedback coordinated control mechanism includes deviation calculation rules, command generation rules, and action decoupling rules. The deviation calculation rules calculate the deviation between the local actual oxygen partial pressure and the preset target oxygen partial pressure. The command generation rules generate feedforward control commands based on the deviation values, including the intake compensation amount and the chamber main exhaust valve opening adjustment amount. The action decoupling rules perform independent mathematical calculations on the adjustment of the local reaction gas flow rate and the adjustment of the chamber main exhaust valve opening.

[0011] Preferably, before inputting the integrated data into the microenvironment digital twin layer, the process further includes:

[0012] The integrated data is preprocessed to obtain preprocessed data. The preprocessing includes hysteresis compensation, outlier removal, and time alignment. Time alignment involves synchronizing the timestamps of macroscopic vacuum data, target discharge electrical parameters, plasma spectral characteristic parameters, and wafer real-time temperature to eliminate acquisition delay differences. The preprocessed data is then used as the integrated data and input into the microenvironment digital twin layer for subsequent calculations.

[0013] Preferably, the consumption rate calculation module is used to calculate the transient oxygen consumption rate based on the microenvironment gas dynamics model combined with the electrical parameters of the target material discharge;

[0014] The consumption rate calculation module is specifically used for:

[0015] The electrical parameters of the target discharge are mapped to the surface state of the target by a microenvironment gas dynamics model, and the initial transient oxygen consumption rate of the target surface is calculated by combining the surface state of the target and the plasma spectral characteristic parameters.

[0016] Voltage baseline drift, short-cycle voltage fluctuation amplitude, and fluctuation occurrence density are extracted from the electrical parameters of the target discharge as voltage fluctuation characteristics. After classifying the voltage fluctuation characteristics into the corresponding target state labels based on the microenvironment gas dynamics model, the initial transient oxygen consumption rate is quantified using a preset nonlinear mapping function to obtain the transient oxygen consumption rate.

[0017] Preferably, the partial pressure prediction module is used to divide the wafer surface into tiny regions and decompose the transient oxygen consumption rate to obtain the local true oxygen partial pressure of the tiny regions;

[0018] The voltage divider prediction module is specifically used for:

[0019] By combining transient oxygen consumption rate, macroscopic vacuum level data, and real-time wafer temperature, the microenvironment gas distribution trend is assessed. The assessment process includes:

[0020] Global gas distribution is predicted using a computational fluid dynamics model. Combined with the transient oxygen consumption rate after decomposition, the local true oxygen partial pressure of each small region is calculated. Based on the deviation direction between the local true oxygen partial pressure and the target oxygen partial pressure, the intensity of change in the corresponding direction is extracted as the rate of change of the local true oxygen partial pressure. The configuration is as follows: if the rate of change is greater than the preset rate of change threshold, it is determined that there is a risk of oxygen vacancy imbalance within the preset time period; otherwise, it is determined that there is no risk of oxygen vacancy imbalance.

[0021] Preferably, the method further includes:

[0022] By combining the collaborative decision-making and execution layer with the micro-environment digital twin layer, dynamic adjustment of vacuum level based on micro-environment state is achieved. This dynamic adjustment of vacuum level based on micro-environment state includes:

[0023] The deviation between the local true oxygen partial pressure and the target oxygen partial pressure is calculated according to the deviation calculation rules, and feedforward control commands are generated based on the deviation value using the collaborative decision-making and execution layer.

[0024] The integrated data is used as dynamic input to dynamically adjust the control commands. When the absolute value of the detected deviation is less than or equal to the preset deviation threshold, the control mode with smooth fine-tuning is maintained. When the absolute value of the detected deviation is greater than the preset deviation threshold, the action decoupling rule is switched to in real time. The collaborative decision and execution layer is used to synchronously and independently adjust the local reaction gas flow rate and the opening of the main exhaust valve of the chamber through real-time calculation.

[0025] The adjustment parameters are automatically updated based on a closed-loop feedback mechanism that combines feedforward control commands with integrated data.

[0026] Preferably, if the local actual oxygen partial pressure is lower than the target oxygen partial pressure, the vacuum level is dynamically adjusted, including:

[0027] The system is set to increase the local reaction gas flow rate and simultaneously increase the opening of the main evacuation valve in the chamber to absorb the macroscopic pressure fluctuations caused by the increase in reaction gas and maintain the macroscopic vacuum data within the preset pressure fluctuation threshold range.

[0028] If the local actual oxygen partial pressure is higher than the target oxygen partial pressure, the vacuum level is dynamically adjusted, including:

[0029] The system is designed to reduce the flow rate of local reactive gases and simultaneously reduce the opening of the main exhaust valve in the chamber to maintain a constant stoichiometry in the microenvironment on the wafer surface.

[0030] If the local actual oxygen partial pressure equals the target oxygen partial pressure, the vacuum level is dynamically adjusted, including:

[0031] Maintain the local reaction gas flow rate and the opening of the main evacuation valve in the chamber at the current state, and continuously monitor the macroscopic vacuum data, target discharge electrical parameters, plasma spectral characteristic parameters, and real-time wafer temperature.

[0032] Preferably, the intake compensation amount is used to pulse gas injection or cut-off via a local gas distributor near the wafer to regulate the local reactive gas flow rate;

[0033] The chamber main exhaust valve opening adjustment amount is used to control the dynamic opening adjustment of the chamber main exhaust valve to form a basic maintenance control of the background pressure.

[0034] The beneficial effects of this invention are:

[0035] 1) By constructing a multi-source sensing input layer and a microenvironment digital twin layer, this invention elevates the controlled object from a single macroscopic chamber total pressure to a dynamic characterization of the real oxygen partial pressure of the microenvironment on the wafer surface; it solves the problem of local hypoxia or oxygen enrichment caused by sensing lag in traditional methods, realizes direct constraint on the stoichiometry of the thin film, and significantly improves the consistency of the photoelectric properties of the film layer.

[0036] 2) This invention introduces a preprocessing mechanism in the data flow that includes hysteresis compensation, outlier removal and timing alignment; it restores and maps multi-source process signals with different delay characteristics onto a unified time axis, eliminates the acquisition delay differences of various sensors, ensures the authenticity and synchronization of the input digital twin layer data, and improves the accuracy of microenvironment state judgment from the source.

[0037] 3) This invention maps the target discharge parameters to the surface state through a kinetic model, calculates the transient oxygen consumption rate by combining spectral characteristics, and uses voltage fluctuation characteristics for nonlinear correction. This mechanism can capture local oxygen consumption changes before macroscopic pressure is manifested, retaining the sensitivity to target state migration and avoiding false compensation caused by single power supply noise or isolated anomalies.

[0038] 4) This invention divides the wafer surface into tiny regions and combines global prediction and consumption rate decomposition using a fluid dynamics model to accurately calculate the local true oxygen partial pressure and rate of change. This mechanism further refines the total pressure stability into the oxygen supply state of each region, which can identify and intervene in film formation deviations in specific regions of the wafer before structural misalignment, effectively suppressing local film degradation.

[0039] 5) This invention sets up a feedforward feedback collaborative control mechanism and action decoupling rules, which independently calculates and synchronously executes the local reaction gas flow regulation and the main exhaust valve opening regulation; while correcting the microenvironment oxygen partial pressure deviation, the pressure disturbance caused by local air intake is dynamically absorbed through the exhaust valve, thus solving the technical contradiction of sacrificing local oxygen for pressure stabilization in traditional control. Attached Figure Description

[0040] The invention will now be further described with reference to the accompanying drawings.

[0041] Figure 1 This is a schematic flowchart of a method for adjusting the vacuum level of ITO coating on a chip, provided in an embodiment of this application. Detailed Implementation

[0042] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0043] Please see Figure 1 A method for adjusting the vacuum level of ITO (Indium Tin Oxide) chip coating is disclosed, applied to a coating system including a chamber, a main chamber exhaust valve, an air intake device with a local gas distributor, a chamber pressure detection unit, a target power supply, a plasma spectral acquisition unit, and a wafer temperature acquisition unit. The method includes: constructing a multi-source sensing input layer to acquire and integrate macroscopic vacuum level data, target discharge electrical parameters, plasma spectral characteristic parameters, and real-time wafer temperature; and constructing a microenvironment digital twin layer to calculate the microenvironment state of the wafer surface based on the integrated data.

[0044] The microenvironment digital twin layer includes a consumption rate calculation module and a partial pressure prediction module. The consumption rate calculation module calculates the transient oxygen consumption rate, and the partial pressure prediction module predicts the local true oxygen partial pressure. A collaborative decision-making and execution layer is constructed, a feedforward feedback collaborative control mechanism is set, and the local reaction gas flow rate of the air intake device and the opening of the main exhaust valve of the chamber are adjusted according to the feedforward feedback collaborative control mechanism.

[0045] The feedforward feedback coordinated control mechanism includes deviation calculation rules, command generation rules, and action decoupling rules. The deviation calculation rules calculate the deviation between the local actual oxygen partial pressure and the preset target oxygen partial pressure. The command generation rules generate feedforward control commands based on the deviation values, including the intake compensation amount and the chamber main exhaust valve opening adjustment amount. The action decoupling rules perform independent mathematical calculations on the adjustment of the local reaction gas flow rate and the adjustment of the chamber main exhaust valve opening.

[0046] This embodiment provides a mechanism for adjusting the vacuum level of ITO coating on a chip. Specifically, this embodiment uses an ITO reactive magnetron sputtering production line for a 12-inch display driver chip wafer as the main scenario for illustration. The coating system of this production line includes at least a vacuum chamber, a main pumping valve on the main pumping channel, an inlet device for introducing argon and oxygen, a target power supply, a plasma spectral acquisition unit, a wafer temperature acquisition unit, and a controller.

[0047] The intake device structurally includes a global background intake channel for providing the basic process atmosphere, and a local gas distributor located near the wafer surface for independently adjusting the microenvironment atmosphere; the local reactive gas flow rate in this article specifically refers to the reactive gas flow rate output by the local gas distributor alone.

[0048] The controller is equipped with a multi-source sensing input layer, a microenvironment digital twin layer, and a collaborative decision-making and execution layer. The entire regulation process is not simply based on the total pressure displayed by the vacuum gauge at the edge of the chamber, but rather on the actual oxygen partial pressure state in the microenvironment near the wafer surface, thereby keeping the oxygen vacancies within a controllable range during the formation of the ITO thin film.

[0049] Specifically, the multi-source sensing input layer is used to collect and integrate four types of information: the first is macroscopic vacuum data, which is provided by the chamber pressure detection unit and is used to reflect the overall background pressure of the chamber; the second is the electrical parameters of the target material discharge, which may include at least the discharge voltage, discharge current and its fluctuation over time.

[0050] When the target material is in a metallic, transitional, or oxidized state, the electrical response of the discharge will change accordingly. Therefore, these parameters can characterize the reactivity of the target surface. The third is the plasma spectral characteristic parameters, which can be collected by the spectral unit set at the observation window. The oxygen-related emission intensity, argon-related emission intensity, and spectral line ratio in the spectrum can reflect the degree of participation of the reactive gas.

[0051] Fourthly, the real-time temperature of the wafer. The wafer temperature not only affects the kinetics of thin film growth, but also the desorption behavior of the wafer surface and clamping parts. The controller integrates the above four types of information according to the same process cycle to form a joint input that reflects the overall state of the chamber, the state of the target material, the state of the plasma, and the thermal state of the wafer.

[0052] Furthermore, the microenvironment digital twin layer does not provide a static description of the entire chamber, but rather a dynamic characterization of the local reaction space above the wafer surface; this layer includes at least a consumption rate calculation module and a partial pressure prediction module; the consumption rate calculation module is used to estimate the transient oxygen consumption rate;

[0053] Oxygen consumption includes both the absorption of oxygen by chemical reactions on the target surface and the local oxygen depletion effect caused by active particles in the plasma of the deposition region.

[0054] The partial pressure prediction module predicts the local true oxygen partial pressure at different locations on the wafer surface based on the chamber background pressure, gas transmission trend, local disturbances caused by wafer temperature, and the aforementioned oxygen consumption state.

[0055] Since the actual controlled object of ITO film formation is not the total pressure at any point in the chamber, but the oxygen supply level near the wafer surface, the local real oxygen partial pressure is closer to the real controlled object in the thin film stoichiometry formation process.

[0056] Furthermore, a feedforward and feedback collaborative control mechanism is established at the collaborative decision-making and execution layers. The feedforward part is used to generate action commands in advance based on the deviation between the local real oxygen partial pressure and the target oxygen partial pressure before the macro pressure changes significantly. The feedback part is used to continue monitoring the system status after execution and correct control deviations caused by gas hysteresis, thermal inertia and target surface state migration.

[0057] The deviation calculation rule uses the difference between the local true oxygen partial pressure and the target oxygen partial pressure as the control basis. When the deviation is manifested as local oxygen deficiency, it indicates that there is a risk of oxygen deficiency on the wafer surface and that there may be too many oxygen vacancies in the thin film. When the deviation is manifested as local oxygen enrichment, it indicates that the risk of increased resistance is increased.

[0058] The instruction generation rule is used to generate the intake compensation amount and the chamber main exhaust valve opening adjustment amount; the action decoupling rule solves the local reaction gas flow rate adjustment and the main exhaust valve opening adjustment separately. The mechanism is that the two have different objects of action: the former mainly affects the chemical environment near the wafer, while the latter mainly affects the chamber background pressure and the exhaust rhythm.

[0059] If the two are coupled as a single control variable, it is easy to cause a control interference phenomenon where intake compensation leads to an increase in cavity pressure, while maintaining cavity pressure stability leads to local microenvironment hypoxia.

[0060] Specifically, the logic of the independent mathematical solution is as follows: First, establish the objective function for adjusting the local reaction gas flow rate, and the calculation formula is:

[0061] ;

[0062] in, This is the amount of local reaction gas flow rate adjustment. For gaseous feedback coefficient, The deviation between the local true oxygen partial pressure and the target oxygen partial pressure is given; subsequently, a maintenance function for the opening of the main extraction valve of the chamber is established, and the calculation formula is as follows:

[0063] ;

[0064] in, This refers to the adjustment amount of the main exhaust valve opening in the chamber. For valve response coefficient, The deviation between the current macroeconomic pressure and the permissible pressure boundary, This is the estimated disturbance function of the current intake adjustment amount on the overall chamber pressure; through the above independent solution channels, the decoupling of intake and extraction actions is achieved;

[0065] Furthermore, to avoid confusion in the terminology of the same technical object in subsequent implementations, this embodiment makes a unified convention for several core terms: macroscopic vacuum data, total chamber pressure, total pressure, and chamber pressure all refer to data obtained by the chamber pressure detection unit and used to characterize the overall pressure level of the chamber. When the object being collected is involved, macroscopic vacuum data is used, and when describing its physical meaning, it can be expressed as total chamber pressure or background pressure.

[0066] In this article, the main extraction valve of the chamber and the main extraction valve both refer to the same valve actuator installed on the main extraction channel; the local reaction gas flow rate, local oxygen supply, oxygen supplementation, and gas inlet compensation all refer to the adjustment amount implemented by the gas inlet device for the reaction gas in the near field region of the wafer. Among them, the gas inlet compensation is the output expression of the control command layer, and the local reaction gas flow rate is the physical flow state after execution.

[0067] In this paper, the local true oxygen partial pressure is the predicted control quantity output by the partial pressure prediction module, the target oxygen partial pressure is the reference control target corresponding to the process formulation at the current stage, and the deviation value is the result obtained by comparing the two.

[0068] According to the above agreement, if expressions such as maintaining total pressure stability, maintaining chamber pressure stability, or maintaining background pressure appear in subsequent implementation methods, their corresponding control objects and physical meanings will not be changed.

[0069] Furthermore, to maintain a clear hierarchical relationship, in this embodiment, the controller is the overall hardware and software that carries the algorithm and execution logic, while the multi-source sensing input layer, the micro-environment digital twin layer, and the collaborative decision-making and execution layer are functional layers deployed in the controller.

[0070] Therefore, when the following implementation methods include descriptions such as controller determination, collaborative decision-making and execution layer instruction generation, and system execution adjustment, they all refer to the decomposition of actions of the same control system at different functional layers, rather than the addition of new independent control entities.

[0071] In a specific data flow control scenario, the wafer surface can be divided into three micro-regions: the central region, the intermediate ring region, and the edge region.

[0072] If the multi-source sensing input layer detects that the macroscopic pressure is basically stable, but the target voltage shows a slow drift consistent with the target surface oxidation trend, and the oxygen-related spectral intensity decreases in the edge observation direction, and the wafer temperature increases compared to the previous cycle, then the micro-environment digital twin layer will interpret this set of inputs as the overall cavity pressure not being unstable, but the local oxygen supply being weakened by the target oxygen consumption and thermal disturbance.

[0073] At this point, the collaborative decision-making and execution layer anticipates the macro pressure feedback, directly compensating for the oxygen input near the wafer, and simultaneously fine-tuning the main exhaust valve to absorb the total pressure disturbance caused by the gas replenishment.

[0074] Furthermore, in abnormal situations, when a certain type of sensing information is temporarily missing, the controller does not directly terminate the coating process, but maintains degraded operation based on the remaining valid information. For example, if the spectral window is temporarily affected by deposition contamination, resulting in a decrease in the reliability of the spectral signal, the controller reduces the weight of the spectral features and instead uses the electrical parameters of the target material discharge and the wafer temperature as the main basis to maintain conservative feedback regulation.

[0075] If the wafer temperature acquisition is abnormal, the thermal state estimate of the previous stable cycle will be maintained, and the upper limit of the gas inlet compensation amount will be limited to prevent excessive gas inlet.

[0076] If the macroscopic vacuum data shows a significant jump and is inconsistent with other states, it is first judged as a disturbance at the detection end, and the rapid action is frozen in a short time, only the valve adjustment within the safety boundary is retained; if multiple key data fail at the same time, the system switches to safety mode, maintains the conservative flow rate and valve position of the current process formula, and stops the deposition of this film if necessary to avoid the formation of an irreversible abnormal film layer.

[0077] In a certain batch of this 12-inch production line, after the wafers entered the ITO main deposition stage, the chamber edge vacuum gauge showed that the total pressure remained within the predetermined range, but the system simultaneously detected that the target discharge voltage fluctuations were aggravated, the oxygen-related spectrum emission was weakened, the wafer edge temperature was higher than the central region and the temperature difference reached the preset temperature difference threshold.

[0078] Based on this, the controller determined that it was not a simple voltage regulation problem, but rather a local oxygen supply shortage trend near the wafer edge. Therefore, it issued a small oxygen replenishment command to the gas inlet near the wafer and simultaneously adjusted the opening of the main exhaust valve to a state that could absorb the new gas. After this round of actions, the local real oxygen partial pressure in the edge area and the center area tended to be consistent again, and the ITO film formed subsequently remained within the process window in terms of sheet resistance and transmittance.

[0079] The purpose of this step is to shift the control target from the total pressure of the chamber to the actual oxygen partial pressure of the wafer surface microenvironment, and to achieve direct constraint on the stoichiometry of the ITO thin film through multi-source sensing, microenvironment calculation and feedforward feedback, thereby improving the stability and batch consistency of the film's photoelectric properties.

[0080] In a preferred embodiment of the present invention, before inputting the integrated data into the microenvironment digital twin layer, the method further includes: preprocessing the integrated data to obtain preprocessed data, wherein the preprocessing includes hysteresis compensation, outlier removal, and timing alignment, wherein timing alignment involves synchronizing the macroscopic vacuum level data, target discharge electrical parameters, plasma spectral characteristic parameters, and wafer real-time temperature with timestamps to eliminate acquisition delay differences; and the preprocessed data is then input into the microenvironment digital twin layer as integrated data for subsequent calculations.

[0081] This embodiment provides a preprocessing mechanism for multi-source process signals; specifically, in the previous embodiment, the controller was already able to estimate the microenvironment state of the wafer surface based on multi-source data, but in actual production lines, the response speed and installation position of different sensing channels are not consistent.

[0082] If unprocessed data is directly input into the microenvironment digital twin layer, it can easily lead to timing deviations in the process states at the same time, thereby causing distortion in the determination of the microenvironment state. Therefore, this embodiment adds three steps before the data enters the microenvironment digital twin layer: hysteresis compensation, outlier removal, and timing alignment.

[0083] Specifically, hysteresis compensation mainly targets macroscopic vacuum data and some temperature data. Macroscopic vacuum gauges are usually installed at the edge of the chamber or near the pumping channel. They sense pressure changes that have been smoothed by the chamber volume and balancing process, rather than instantaneous gas disturbances near the wafer. If local oxygen replenishment has occurred, but the macroscopic vacuum signal has not yet been reflected, then the original vacuum curve will have an inherent hysteresis of the system.

[0084] This embodiment establishes a correspondence between sampling point location, gas propagation path, and response delay to perform time rollback correction on this type of data, making it closer to the actual occurrence time. Temperature data may also have thermal conduction delays. For example, the temperature measurement on the back of the wafer is not synchronized with the surface deposition reaction. Therefore, the temperature curve can be compensated for the delay based on preset time parameters according to the thermal structure of the equipment.

[0085] Furthermore, outlier removal is used to remove sudden isolated points that do not represent the true process state; due to the electromagnetic interference generated by plasma, short-term spikes may appear in the target discharge parameters; local contamination of the observation window or optical path flicker may cause occasional deviations in the spectral signal; mechanical vibration may also cause temperature sampling jitter; if such outliers are directly interpreted as process mutations, the controller may incorrectly issue large compensation commands.

[0086] This embodiment prioritizes retaining changes that can persist in adjacent sampling periods and are correlated with other signals. For points that only appear in a single period and are inconsistent with other physical quantities, they are marked as abnormal and removed or weakened.

[0087] Furthermore, timing alignment is used to map data from different sources onto a unified process timeline; macroscopic vacuum, target discharge parameters, plasma spectral parameters, and wafer temperature can be labeled with their sampling times and reassembled in the controller according to a unified time base;

[0088] Specifically, if a target voltage change, a spectral change, and a temperature change are recorded within the same control cycle, but they originate from different moments, the system will re-map them to the same process event window to ensure that the target surface reaction change, plasma response, and wafer thermal state change are interpreted as the same physical process, rather than being mistaken for three unrelated events.

[0089] In a specific preprocessing application scenario, four sets of raw data are generated within a certain control cycle: the vacuum degree data has a transmission delay greater than the preset sampling cycle, the target voltage is almost real-time, the spectral data has an isolated spike, and the temperature data lags by one sampling cycle.

[0090] After preprocessing, the system will correct the vacuum data forward to be closer to the moment when local gas injection occurs, regard isolated peaks in the spectrum as anomalies caused by window flickering instead of as evidence of process deterioration, and align the temperature data to the actual heating stage of the wafer. After processing, the input to the microenvironment digital twin layer is a set of synchronously aligned process datasets at the same time, rather than fragmented signals that are misaligned with each other.

[0091] Furthermore, if a channel is distorted for several consecutive cycles, it is not advisable to continue using the simple elimination method, otherwise the channel will not participate in effective calculations for a long time.

[0092] In this embodiment, if an abnormal value persists and gradually forms a stable new baseline, the system will change it from an abnormal state to a state migration and re-participate in the calculation; if a channel has no valid value for a long time, the controller will select a conservative strategy according to the equipment operation stage. For example, in the process stabilization stage, the remaining channel is allowed to maintain the adjustment, while in the formula switching stage, the automatic compensation intensity is reduced. If necessary, the system will prompt the maintenance observation window, calibrate the vacuum gauge, or check the thermocouple connection status.

[0093] During the aforementioned main deposition process of the 12-inch wafer, the target voltage immediately responded after the equipment completed a small oxygen replenishment operation, but the vacuum gauge at the edge of the chamber only showed a pressure change later; at the same time, the spectrometer showed a sudden abnormally high value due to particles attached to the observation window, and the wafer temperature rose later due to thermal inertia.

[0094] After preprocessing, the controller recognizes that the data essentially corresponds to the same microenvironment adjustment process after oxygen supplementation. Therefore, it will not misjudge the spectral peak as oxygen-enriched runaway, nor will it mistakenly believe that oxygen supplementation has not yet taken effect due to the lag of the vacuum gauge.

[0095] The purpose of this step is to restore process signals from different sources, with different delays and different noise characteristics into a consistent input that can jointly characterize the real microenvironment state, thereby improving the reliability of subsequent oxygen consumption estimation and local oxygen partial pressure prediction.

[0096] In a preferred embodiment of the present invention, the consumption rate calculation module is used to calculate the transient oxygen consumption rate based on the microenvironment gas dynamics model combined with the electrical parameters of the target discharge. Specifically, the consumption rate calculation module is used to: map the electrical parameters of the target discharge to the target surface state through the microenvironment gas dynamics model, and calculate the initial transient oxygen consumption rate of the target surface by combining the target surface state and plasma spectral characteristic parameters; extract the voltage baseline drift, voltage short-cycle fluctuation amplitude and fluctuation occurrence density from the electrical parameters of the target discharge as voltage fluctuation characteristics, and classify the voltage fluctuation characteristics into the corresponding target surface state labels according to the microenvironment gas dynamics model, and then quantify the initial transient oxygen consumption rate using a preset nonlinear mapping function to obtain the transient oxygen consumption rate.

[0097] This embodiment provides a mechanism for estimating transient oxygen consumption rate. Specifically, in the previous embodiment, the multi-source inputs have been organized into reliable data on the same time axis. However, it is still difficult to identify the actual oxygen consumption behavior of the target surface in reactive sputtering based solely on the macroscopic vacuum degree and static gas flow rate.

[0098] Specifically, the expression for the nonlinear mapping function is:

[0099] ;

[0100] in, This is the corrected transient oxygen consumption rate. The initial transient oxygen consumption rate on the target surface. This refers to the extracted short-cycle voltage fluctuation amplitude. The density of the extracted fluctuations and The nonlinear weighting coefficients are preset based on the target surface state; when the target surface is determined to be in a transition-sensitive state based on the target discharge electrical parameters, and The value of is increased exponentially to reflect a strong correction to the initial transient oxygen consumption rate;

[0101] Especially in ITO deposition, the oxidation level of the target surface will migrate between metallic, transitional and strongly oxidized states, and the oxygen absorption capacity of different states is significantly different. Therefore, this embodiment uses a microenvironment gas dynamics model to map the electrical parameters of the target material discharge to the target surface state, and combines plasma spectral information to infer the transient oxygen consumption rate.

[0102] Specifically, the electrical parameters of target discharge can reflect the surface state of the target because the conductivity, secondary electron emission characteristics, and local reaction coverage of the target surface directly affect the discharge maintenance conditions.

[0103] When the proportion of metallic state on the target surface reaches the preset ratio, the discharge voltage fluctuation amplitude is lower than the preset voltage fluctuation threshold, and the net consumption of oxygen is mainly reflected in the formation of reaction products; when the target surface migrates to the oxidized state, the discharge characteristics will change, indicating that the adsorption and reaction mode of oxygen on the target surface has changed.

[0104] This embodiment does not require in-situ direct measurement of the chemical composition of the target surface. Instead, it uses the target voltage, target current and their stability characteristics to establish a mapping relationship with the target surface state. Then, combined with the relative changes of oxygen-related spectral lines and inert gas spectral lines in the plasma spectrum, the initial transient oxygen consumption rate of the target surface is estimated.

[0105] Specifically, the electrical parameters of the target discharge are used to characterize the real-time state of the target surface reaction, and the plasma spectral characteristic parameters are used to characterize the consumption rate of reactive oxygen in the cavity under the state. The two are combined to characterize the transient oxygen consumption intensity of the target surface.

[0106] Furthermore, the initial transient oxygen consumption rate alone is insufficient to fully reflect rapid disturbances, because voltage fluctuations common during discharge contain dynamic information such as target surface coating rupture, local reactivation, and reaction interface migration.

[0107] These fluctuations do not correspond linearly to the intensity of oxygen consumption: some small fluctuations represent normal maintenance, while some fluctuations of the same magnitude correspond to the unstable state at the edge of target poisoning; therefore, in this embodiment, voltage fluctuation characteristics are extracted from the electrical parameters of the target discharge, and the initial transient oxygen consumption rate is corrected through a preset nonlinear mapping relationship; the physical mechanism of nonlinear processing is that when the target surface migrates from the stable region to the transition region, its oxygen consumption characteristics exhibit nonlinear step change characteristics.

[0108] Furthermore, to avoid the nonlinear mapping being interpreted as an unstructured process lacking clear input, output, and correction rules, this embodiment can be executed according to the following structured process: First, extract the voltage baseline drift, short-cycle voltage fluctuation amplitude, and fluctuation density from the electrical parameters of the target discharge within a control cycle; then, based on the microenvironment gas dynamics model, classify the above features into three target state labels: stable maintenance, transition enhancement, or transition sensitivity.

[0109] Using the initial transient oxygen consumption rate as the basic quantity, different correction intensities are assigned to different state labels, with weak correction corresponding to stable maintenance, moderate correction corresponding to transition enhancement, and strong correction corresponding to change sensitivity. In this way, the input object, correction basis, and output direction of the nonlinear mapping function all have clear physical meanings, rather than simply amplifying the control result based on a single noise spike.

[0110] Furthermore, in engineering implementation, the preset nonlinear mapping function preferably satisfies two constraints: monotonic segmentation and boundary limiting. Monotonic segmentation means that when the voltage fluctuation characteristics indicate that the target surface is gradually approaching the transition region from the stable region, the transient oxygen consumption rate correction is increased step by step according to the preset segmentation.

[0111] The so-called boundary limit refers to the fact that when the power supply transient disturbance is too large but the spectral characteristics are not coordinated, the correction amount is not allowed to be increased indefinitely, but is limited to the upper limit of the oxygen consumption estimate allowed at the current process stage. In this way, the high sensitivity to the transition region and the non-over-response to occasional noise can be included in the same calculation framework.

[0112] Furthermore, in order to ensure that the aforementioned preset nonlinear mapping function remains unique throughout the text, in this embodiment, the function is specifically used to map voltage fluctuation characteristics to the correction intensity of the initial transient oxygen consumption rate. In subsequent embodiments, if nonlinear mapping, nonlinear correction, or correction function is mentioned again, it all refers to the same processing logic, and no other nonlinear relationships in the airflow model, temperature model, or valve control model will be separately indicated.

[0113] Furthermore, the execution of the nonlinear mapping function in the controller can be refined into the following rule chain: determine whether the voltage baseline drift direction is consistent with the oxygen-related spectrum change direction; if both point to oxygen consumption enhancement, then the enhancement correction branch is allowed.

[0114] If the two are inconsistent, maintain the basic correction or postpone the confirmation by one cycle; determine whether the short-cycle voltage fluctuation amplitude exceeds the fluctuation threshold of the current process stage, and whether the fluctuation density increases for several consecutive short cycles; only when the amplitude and density meet at least one of them and are consistent with the target surface status label, will the correction intensity be increased; furthermore, when the system has classified the target surface status as transition sensitive, the correction increment corresponding to the voltage fluctuation of the same amplitude is greater than the correction increment under the stable maintenance state.

[0115] When the system has classified the target surface state as stable, it places greater emphasis on spectral confirmation and does not directly increase the transient oxygen consumption rate due to a single voltage spike. Through the above rules, the correction process of the initial transient oxygen consumption rate is consistent with both the target surface state and the spectral evidence.

[0116] Furthermore, to avoid the misuse of different feature names in the subsequent description, this embodiment stipulates that: voltage baseline drift is used to characterize a slower state transition trend, voltage short-cycle fluctuation amplitude is used to characterize the strength of short-term disturbances, and fluctuation occurrence density is used to characterize the frequency of disturbance occurrences.

[0117] These three factors together constitute the voltage fluctuation characteristics. Subsequent mentions of enhanced discharge fluctuations, increased voltage micro-fluctuations, or intensified fluctuation characteristics are all simplified descriptions of this set of voltage fluctuation characteristics, rather than introducing new independent parameter objects.

[0118] In a specific control application scenario, the surface state of the target material can be roughly divided into three states: the first state, the second state, and the third state. The first state corresponds to an oxygen supply greater than a preset oxygen supply threshold and a target surface reaction parameter fluctuation rate lower than a preset fluctuation threshold. The second state corresponds to a transition zone where the target surface coating layer switches back and forth between formation and peeling. The third state corresponds to a target surface reaction parameter fluctuation rate greater than or equal to a preset fluctuation threshold and an oxygen supply lower than a preset oxygen supply threshold.

[0119] If, at the same moment, the target voltage baseline slowly shifts upward while the oxygen-related spectrum weakens, the system can interpret this as an increase in the actual oxygen occupancy of the target surface, leading to an upward adjustment of the initial transient oxygen consumption rate. If, on this basis, voltage fluctuations are observed to change from gradual to dense and exhibit an unstable distribution, it indicates that the target surface has entered a more sensitive transition phase, and the system further improves the transient oxygen consumption rate estimate to anticipate the risk of local hypoxia.

[0120] Furthermore, the aforementioned process can be understood as a three-level calculation chain of estimation, correction, and limiting: first, the target state and spectrum jointly provide the initial transient oxygen consumption rate; then, the voltage fluctuation characteristics trigger nonlinear correction; and finally, the correction result is limited and output in combination with the safety boundary of the current process stage.

[0121] Therefore, the transient oxygen consumption rate sent to the partial pressure prediction module retains the sensitivity to target migration and will not be distorted by a single abnormal fluctuation.

[0122] Furthermore, if the electrical parameters of the target material change significantly but spectral information is missing, the system can still give a conservative estimate of oxygen consumption based on the target surface state mapping, but with reduced sensitivity to avoid misinterpreting power supply noise as target poisoning migration.

[0123] If the spectrum is abnormal but the target electrical parameters are stable, the system tends to judge it as optical path contamination or a change in the observation angle, and will not easily increase the transient oxygen consumption rate. If both indicate abnormality but the direction of change is contradictory, the controller will not perform aggressive compensation for the time being, but will wait for confirmation in the next short cycle to prevent the introduction of over-control during the short-term self-recovery phase of the target surface.

[0124] In the same batch of the aforementioned production line, after the wafer entered the middle of the coating stage, the total pressure in the chamber remained stable, but the controller continuously detected that the baseline of the target discharge voltage was raised and the emission intensity of the oxygen-related spectrum decreased. At the same time, the voltage fluctuations were more frequent than in the previous stage.

[0125] Based on this, the system determines that the target surface has migrated from a more stable transition state to a region with higher oxygen consumption. Therefore, it assesses the transient oxygen consumption rate as an increasing state and sends this information to the partial pressure prediction module.

[0126] The purpose of this step is to utilize the physical property of the target material as the reaction core, which is most sensitive to oxygen, to capture local oxygen consumption changes in advance before macroscopic pressure is manifested, thereby improving the ability to perceive the oxygen-deficient or oxygen-enriched trend of the microenvironment on the wafer surface.

[0127] In a preferred embodiment of the present invention, the partial pressure prediction module is used to divide the wafer surface into micro-regions and decompose the transient oxygen consumption rate to obtain the local true oxygen partial pressure of the micro-regions. Specifically, the partial pressure prediction module is used to: combine the transient oxygen consumption rate, macroscopic vacuum data, and real-time wafer temperature to assess the gas distribution trend in the microenvironment. The assessment process includes: using a computational fluid dynamics model to predict the global gas distribution; combining the decomposed transient oxygen consumption rate to calculate the local true oxygen partial pressure of each micro-region; and extracting the change intensity in the corresponding direction as the change rate of the local true oxygen partial pressure based on the deviation direction between the local true oxygen partial pressure and the target oxygen partial pressure. The configuration is as follows: if the change rate is greater than a preset change rate threshold, it is determined that there is a risk of oxygen vacancy imbalance within a preset time period; otherwise, it is determined that there is no risk of oxygen vacancy imbalance.

[0128] This embodiment provides a mechanism for predicting local true oxygen partial pressure and assessing risk. Specifically, the previous embodiment can provide the system-level transient oxygen consumption rate, but the film quality problem of ITO thin films is often not a simultaneous loss of control across the entire wafer, but rather is primarily manifested as a deviation in the stoichiometry of the wafer edge, center, or specific ring regions.

[0129] Therefore, in this embodiment, the wafer surface is divided into multiple micro-regions, and the local true oxygen partial pressure of each region is evaluated in combination with the gas distribution trend, thereby identifying the risk of oxygen vacancy imbalance.

[0130] Specifically, the wafer surface can be divided into multiple concentric ring regions, sectors, or grid regions according to process requirements; the finer the region division, the better it is to capture edge effects and local thermal disturbances; the coarser the region division, the better it is to control the amount of computation; in the reactive magnetron sputtering process, the gas is not statically distributed;

[0131] After oxygen enters from the inlet, it will form spatial distribution differences above the wafer due to the influence of chamber structure, airflow organization, extraction direction, target oxygen consumption, wafer temperature rise, and plasma volume effect; therefore, this embodiment introduces a computational fluid dynamics model in the partial pressure prediction module to predict the global gas distribution.

[0132] The global prediction here is not to obtain the absolute precise value of every point in the chamber, but to provide a background field that can reflect the main airflow path, edge backflow, center coverage and suction traction trend; then the transient oxygen consumption rate is decomposed according to the correspondence between the target action area and the deposition area, and superimposed on each micro-region to obtain a more accurate local oxygen partial pressure that is closer to the actual film formation on the wafer surface.

[0133] Furthermore, the rate of change of local true oxygen partial pressure is an important basis for risk assessment. This is because the anomaly of oxygen vacancies in the film depends not only on whether the oxygen partial pressure is too low or too high at a certain moment, but also on whether this shift changes rapidly in a short period of time. If the local oxygen partial pressure remains within a stable window for a long time, even if the absolute value fluctuates slightly, the film still has the opportunity to form a relatively uniform structure.

[0134] Conversely, if the local oxygen partial pressure drops or rises rapidly in a short period of time, the film growth front will not have enough time to self-balance and will easily solidify in the lattice into a state of too many or too few oxygen vacancies. Therefore, in this embodiment, the rate of change is compared with a preset rate of change threshold. When the rate of change exceeds the threshold, it is determined that there is a risk of oxygen vacancy imbalance within a preset time period. If it does not exceed the threshold, the current risk is considered to be controllable.

[0135] Furthermore, in order to ensure that the aforementioned rate of change is greater than the threshold and is consistent with both hypoxia risk and hyperxia risk scenarios, the rate of change in this embodiment adopts a normalized definition oriented towards the risk direction, rather than simply retaining the original algebraic symbols;

[0136] Specifically, when the current micro-region shows that the local true oxygen partial pressure is lower than the target oxygen partial pressure, the rate of decrease of the micro-region within a preset time period is used as the risk change rate; when the current micro-region shows that the local true oxygen partial pressure is higher than the target oxygen partial pressure, the rate of increase of the micro-region within a preset time period is used as the risk change rate.

[0137] The parameter used for threshold comparison is the intensity of change after taking the absolute value in the risk direction; so that both the shift characteristics towards hypoxia and the shift characteristics towards oxygen enrichment can be identified through a unified judgment logic.

[0138] Furthermore, in engineering calculations, a series of consecutive short-period local real oxygen partial pressure sequences can be formed for each micro-region. Then, combined with the deviation direction of the micro-region relative to the target oxygen partial pressure, the change intensity in the corresponding direction can be selected as the change rate input for the micro-region.

[0139] If a micro-region is currently below the target oxygen partial pressure, we first examine whether it continues to decline rapidly; if a micro-region is currently above the target oxygen partial pressure, we first examine whether it continues to rise rapidly; if a micro-region experiences short-term reverse fluctuations but is generally returning to the target value, this change is not considered an amplification of risk, but rather part of the recovery process; through such directional normalization, risk assessment and subsequent control branches can be naturally connected.

[0140] In a specific predictive application scenario, the wafer can be divided into a central region R1, an intermediate region R2, and an edge region R3. If the transient oxygen consumption rate increases and the global gas distribution prediction shows that the newly entering oxygen is mainly carried to the center by the mainstream gas channel and then diffuses to the edge, while the edge region has stronger desorption disturbance due to slightly higher temperature, the local true oxygen partial pressure of R3 will decrease faster than that of R1 and R2.

[0141] If the downward trend of R3 continues and reaches the risk threshold within several consecutive short periods, the system determines that the edge region is more likely to form anoxic ITO structures in the following deposition period, thereby triggering subsequent control actions; conversely, if R2 fluctuates slightly at a certain moment but recovers quickly, it can be regarded as a normal process disturbance and is not considered as a risk of oxygen vacancy imbalance.

[0142] Furthermore, if certain regions simultaneously exhibit absolute value deviation and an increase in the rate of change within the same preset time period, this embodiment prioritizes identifying them as high-priority risk areas; if the absolute value has deviated but the rate of change is slowing down, they tend to be identified as controllable deviation areas; if the absolute value has not yet deviated significantly but the rate of change continues to increase, they can be identified as forward-looking warning areas.

[0143] This allows the pressure prediction module to not only output whether there is a risk, but also to provide the execution layer with information on the risk evolution stage, so as to distinguish whether it is necessary to perform the first preset interval of flow compensation in advance or to immediately enter the secondary control mode containing action decoupling rules.

[0144] Under abnormal operating conditions, if the boundary conditions of the airflow model are incomplete for a short period of time, such as the lack of local intake unit state feedback, the system uses the flow field template under the most recent stable operating condition for approximate prediction and limits the sensitivity of regional risk assessment to avoid false alarms due to incomplete model boundaries.

[0145] If an abnormally high wafer temperature occurs but only at a few sampling points, the system first checks the adjacent cycles to determine if it is an abnormal sampling. Only when the temperature rise trend continues and is consistent with the downward direction of the partial pressure is the risk level increased. If multiple regions are simultaneously determined to be at increased risk, the system distinguishes between overall insufficient oxygen supply or local flow field bias according to the direction of risk diffusion, thus providing a basis for the subsequent execution layer to select different actions.

[0146] In the mid-to-late stage deposition of the aforementioned batches, the partial pressure prediction module divides the wafer surface into a central ring region, an intermediate ring region, and an edge ring region;

[0147] After comprehensively considering transient oxygen consumption rate, macroscopic vacuum level and wafer temperature, it was found that although the total pressure in the chamber was not significantly abnormal, the local real oxygen partial pressure in the edge ring region decreased much faster than that in the central ring region, and it did not recover for several consecutive process cycles.

[0148] Therefore, the system determines that there is a risk of oxygen vacancy imbalance in the edge region during the next deposition time window, and sends this risk information to the collaborative decision-making and execution layer;

[0149] The purpose of this step is to further refine the total pressure stability into whether the oxygen supply in each region is stable, thereby identifying local membrane risks that will ultimately be reflected in the sheet resistance and permeability distributions.

[0150] In a preferred embodiment of the present invention, the method further includes: realizing dynamic adjustment of vacuum degree based on microenvironment state by combining a collaborative decision-making and execution layer with a microenvironment digital twin layer, wherein the dynamic adjustment of vacuum degree based on microenvironment state includes: calculating the deviation value between the local real oxygen partial pressure and the target oxygen partial pressure according to the deviation calculation rules, and generating feedforward control commands based on the deviation value using the collaborative decision-making and execution layer.

[0151] The integrated data is used as dynamic input to dynamically adjust control commands. When the absolute value of the detected deviation is less than or equal to the preset deviation threshold, the control mode, which is mainly based on smooth fine-tuning, is maintained. When the absolute value of the detected deviation is greater than the preset deviation threshold, the system switches to the action decoupling rule in real time and uses the collaborative decision-making and execution layer to synchronously and independently adjust the local reaction gas flow rate and the opening of the main exhaust valve of the chamber through real-time calculation. Based on the closed-loop feedback mechanism of feedforward control commands and integrated data, the adjustment parameters are automatically updated.

[0152] This embodiment provides a dynamic vacuum adjustment mechanism based on microenvironmental conditions; specifically, the aforementioned implementation method has been able to identify local real oxygen partial pressure and its risk changes, but if it only stays at the prediction level, it still cannot solve the technical defects of control lag in actual production process.

[0153] Especially during ITO reactive sputtering, the target state changes rapidly, the local atmosphere changes rapidly, while the macroscopic pressure feedback is slow. If a fixed formula with low-frequency correction is still used, it is easy to discover the anomaly only after the thin film structure has been formed. Therefore, this embodiment achieves dynamic adjustment by combining a collaborative decision-making and execution layer with a microenvironment digital twin layer.

[0154] Specifically, the system compares the local true oxygen partial pressure of each micro-region or key monitoring area with the target oxygen partial pressure according to the deviation calculation rules. The target oxygen partial pressure here is not a single empirical value, but a set value that matches the current process stage. For example, different target windows can correspond to the nucleation section, the main deposition section, and the final stabilization section.

[0155] The system generates feedforward control commands based on the magnitude and direction of the deviation. The significance of feedforward is that it compensates for the local oxygen supply or pumping background before the macroscopic cavity pressure fluctuates significantly, thereby avoiding passive correction after the deviation expands.

[0156] Furthermore, when the deviation is within a small range, the system can maintain a control mode that is mainly based on smooth fine-tuning to avoid over-response; when the absolute value of the deviation is detected to exceed the preset threshold, it indicates that the local microenvironment has deviated from the stable process window, and continuing to use coupled small corrections will easily cause response lag.

[0157] Therefore, this embodiment switches to the action decoupling rule in real time to synchronously and independently adjust the local reaction gas flow rate and the opening of the main exhaust valve of the chamber;

[0158] The reason for emphasizing synchronous independence is that although the two affect each other physically, their control objectives are different: the local reactive gas flow rate prioritizes restoring the stoichiometry near the wafer surface, while the main extraction valve prioritizes maintaining the chamber background pressure and discharge stability; only by decoupling the two can we avoid sacrificing the local oxygen partial pressure for voltage stabilization or disrupting discharge stability for oxygen replenishment.

[0159] Furthermore, the closed-loop feedback mechanism is used to automatically update the adjustment parameters; that is, after the feedforward control command is issued, the system continues to observe the new macroscopic vacuum degree, target discharge parameters, spectral characteristics and wafer temperature, and uses the execution results in reverse to correct the control intensity in the next round.

[0160] If the recovery rate of the local actual oxygen partial pressure is slower than expected after a certain oxygen replenishment, it indicates that the current oxygen consumption of the target surface or the flow field bias is stronger than the model estimate. The system will appropriately increase the local compensation intensity in the next cycle. If a local oxygen enrichment trend appears after compensation, the system will automatically reduce the compensation amount and adjust the pumping rhythm simultaneously. This forms a closed-loop control that converges continuously with the evolution of the process.

[0161] In a specific regulatory application scenario, the control state can be divided into three levels: slight deviation, moderate deviation, and severe deviation. When there is a slight deviation, the local reaction gas flow rate is corrected stepwise within the first preset range, and the main extraction valve is adjusted according to the first preset opening change rate. When there is a moderate deviation, the local intake valve and the main extraction valve are finely adjusted at the same time, but the smoothing strategy is still the main approach.

[0162] In the event of severe deviation, the system enters decoupled control. One path rapidly adjusts the reaction gas for local oxygen deficiency or enrichment, while the other path independently adjusts the valve opening based on background pressure and discharge stability. The closed-loop feedback then determines whether the action of this round is insufficient recovery, moderate recovery, or excessive compensation.

[0163] Under abnormal operating conditions, if the deviation is close to the threshold boundary and the reliability of each sensor is not completely consistent, the system will first maintain a conservative action and increase the sampling confirmation frequency to avoid valve shaking caused by switching back and forth near the threshold.

[0164] If the discharge stability is found to deteriorate significantly after the decoupling control is started, the plasma maintenance conditions will be prioritized and the local compensation will be limited to the discharge safety window. If the closed-loop feedback shows that the model prediction and the execution result deviate continuously, the system will automatically reduce the feedforward weight, increase the feedback weight, and prompt the user to re-verify the target material state model used in this batch.

[0165] In the aforementioned main process segment of wafer deposition, the edge ring region is continuously identified as having a risk of oxygen deficiency; the system generates feedforward control commands to compensate for the oxygen supply near the wafer.

[0166] The deviation between the edge ring region and the target oxygen partial pressure was found to be continuing to widen and had exceeded the set threshold. Therefore, the controller immediately switched to decoupled control mode: one path continued to increase the local oxygen supply near the edge, while the other path increased the opening of the main exhaust valve to suppress the rise in total pressure, and the action amplitude was continuously corrected through subsequent spectral and discharge parameter feedback.

[0167] After several short cycles, the oxygen partial pressure deviation in the edge ring region returned to the stable window, and the film deposition returned to a consistent state.

[0168] The purpose of this step is to truly transform predictions into executable dynamic controls, thereby achieving synergistic management of the local chemical environment and the chamber background environment, and reducing membrane fluctuations caused by control lag.

[0169] In a preferred embodiment of the present invention, if the local actual oxygen partial pressure is lower than the target oxygen partial pressure, the vacuum degree is dynamically adjusted, including: setting to increase the local reaction gas flow rate and simultaneously increasing the opening of the main pump valve of the chamber to absorb the macroscopic pressure fluctuation caused by the increase of reaction gas and maintain the macroscopic vacuum degree data within the preset pressure fluctuation threshold range.

[0170] If the local actual oxygen partial pressure is higher than the target oxygen partial pressure, the vacuum level is dynamically adjusted, including: setting a reduction in the local reactive gas flow rate and simultaneously reducing the opening of the main pump valve in the chamber to maintain the stoichiometry of the microenvironment on the wafer surface at a constant level.

[0171] If the local actual oxygen partial pressure is equal to the target oxygen partial pressure, the vacuum level is dynamically adjusted, including: maintaining the local reaction gas flow rate and the opening of the main pump valve in the chamber at the current state, and continuously monitoring the macroscopic vacuum level data, the electrical parameters of the target material discharge, the plasma spectral characteristic parameters, and the real-time temperature of the wafer.

[0172] This embodiment provides a branch adjustment mechanism based on the deviation direction; specifically, the previous embodiment has established a dynamic adjustment framework, but in the actual process, the film formation consequences corresponding to local low and high true oxygen partial pressure are not the same, and the control action cannot be simply mirrored.

[0173] If the direction of deviation is not distinguished, new membrane quality problems can easily be introduced during the correction process. Therefore, this embodiment specifies different adjustment strategies for three states: local true oxygen partial pressure is lower than the target value, higher than the target value, and equal to the target value.

[0174] Specifically, when the actual oxygen partial pressure in a given area is lower than the target oxygen partial pressure, it indicates that the wafer surface is in a relatively oxygen-deficient environment.

[0175] For ITO thin films, this often means an increased tendency for oxygen vacancies, and the film may shift towards a direction of low resistance but low transmittance or dark color. In this case, this embodiment sets up an increase in local reactive gas flow rate, so that more oxygen can directly enter the microenvironment near the wafer, rather than relying solely on uniform gas supply throughout the cavity.

[0176] At the same time, the opening of the main pump valve in the chamber is increased synchronously; the action coordination mechanism is used to absorb the background pressure fluctuations caused by the new reactive gas, so as to maintain the macroscopic vacuum data within the allowable pressure fluctuation threshold and avoid changes in discharge state or abnormal deposition rate due to the rise in chamber pressure; in other words, local oxygen replenishment and main pump valve opening are coordinated actions in the same direction: one restores the local stoichiometry and the other stabilizes the global process boundary.

[0177] Furthermore, when the local actual oxygen partial pressure is higher than the target oxygen partial pressure, it indicates that the wafer surface is shifting to a relatively oxygen-rich environment. ITO films are more prone to problems such as decreased carrier concentration and increased sheet resistance under this condition. In this case, this embodiment sets the local reactive gas flow rate to be reduced and the opening of the main evacuation valve of the chamber to be reduced simultaneously.

[0178] Reducing local oxygen supply helps prevent oxygen enrichment from worsening, while appropriately reducing the opening of the evacuation valve helps stabilize the inert gas in the chamber and the overall discharge environment, avoiding the local gas supply organization from becoming unbalanced again due to excessive evacuation. The key control here is not simply to reduce the amount of oxygen supplied, but to restore the local stoichiometry without damaging the established deposition stability.

[0179] Furthermore, to avoid the above-mentioned oxygen-enriched branch being misunderstood as reducing the evacuation rate to retain more oxygen, it should be noted that the direct control object for determining oxygen enrichment in this embodiment is the local reactive gas flow rate near the wafer.

[0180] Specifically, when the local actual oxygen partial pressure is higher than the target oxygen partial pressure, the concentration of reactive oxygen entering the wafer surface microenvironment can be directly reduced by reducing or cutting off the supply of near-field reactive gas.

[0181] The simultaneous slight reduction in the opening of the main extraction valve is not to increase the oxygen partial pressure, but to maintain the chamber background pressure, inert gas density and discharge maintenance conditions relatively stable after reducing local oxygen supply. This is to prevent a sudden drop in total pressure, plasma contraction or reversal of the main airflow path caused by maintaining a high level of extraction, which could induce new local inhomogeneities.

[0182] In other words, in the oxygen-enriched branch, reducing local oxygen supply is responsible for correcting the chemical environment, while appropriately reducing the opening of the main extraction valve is responsible for maintaining the background process boundary. The two are still synchronous coordination under different objectives.

[0183] Furthermore, the reduction of the main extraction valve opening in the oxygen-enriched branch is preferably a limited reduction, rather than an unconditional and large reduction; the controller can first execute the action of reducing the local reaction gas flow rate, and then determine whether it is necessary to reduce the main extraction valve opening synchronously, and to what extent, based on the macroscopic vacuum degree, discharge electrical parameters and spectral feedback.

[0184] If the macroscopic pressure has reached a stable window after the local oxygen supply decreases, the main extraction valve will only make a small follow-up adjustment; if the local oxygen supply decreases and the total pressure tends to drop, the main extraction valve will then make a compensatory reduction to maintain the background pressure stability; this can avoid turning the oxygen enrichment correction operation into a new source of pressure disturbance.

[0185] Furthermore, when the actual partial pressure of oxygen in a given area equals the target partial pressure of oxygen, the system maintains the current local reaction gas flow rate and the opening of the main extraction valve unchanged, and continuously monitors the multi-source sensing input;

[0186] The reason for not continuing frequent fine-tuning in this state is that ITO deposition has intrinsic fluctuations. If adjustments are made continuously when the target window has already been met, small disturbances may be amplified into control oscillations. Therefore, in this embodiment, the value equal to the target value is interpreted as being within the allowable window and observation is maintained, rather than stopping monitoring.

[0187] In a specific control application scenario, if the key monitoring area is currently determined to be oxygen-deficient, the control action will be to increase the local oxygen supply and simultaneously increase the opening of the main exhaust valve; if it is determined to be oxygen-rich, the control action will be to decrease the local oxygen supply and moderately reduce the opening of the main exhaust valve while maintaining the background pressure.

[0188] If the target voltage, spectrum, and temperature are within a stable window, both remain unchanged, and the target voltage, spectrum, and temperature are continuously monitored for new migration signs. This forms a branch control logic with a clear direction and target.

[0189] Under abnormal operating conditions, if the local actual oxygen partial pressure fluctuates slightly around the target value, the system will not immediately switch frequently between the oxygen-deficient and oxygen-rich branches. Instead, it will set a stable range and a minimum holding time to avoid repeated operation of valves and flow controllers, which would lead to mechanical wear and process fluctuations.

[0190] If the macroscopic pressure approaches the upper limit after the execution of the oxygen-deficient branch, the absorption capacity of the main exhaust valve to the background pressure will be increased first, and the local oxygen supply increment will be limited. If necessary, it will be restored gradually in multiple steps. If the fluctuation amplitude of the discharge parameters exceeds the preset stability threshold after the execution of the oxygen-enriched branch, the reduction of the exhaust valve will be limited. The local oxygen supply will be reduced first according to the first preset flow rate reduction step size. After the discharge returns to stability, the correction will continue.

[0191] After the risk of oxygen deficiency appeared in the edge ring area of ​​the aforementioned batch, the system implemented a strategy of increasing the local reaction gas flow rate and simultaneously increasing the opening of the main extraction valve, so that the edge area could receive more direct oxygen compensation, while keeping the total pressure change within the allowable window of the equipment.

[0192] After several cycles, the oxygen partial pressure in the edge region returned to near the target. Due to a slight lag in compensation, the edge region briefly became slightly oxygen-rich within a short period. The system then switched to a strategy of reducing the local reactive gas flow and simultaneously reducing the opening of the main extraction valve. The local oxygen supply was reduced first, and the main extraction valve was then adjusted downwards to maintain the background pressure, so that the local stoichiometry returned to stability. Once all monitored areas fell within the target window, the system maintained its current state, continuing to monitor without further intervention.

[0193] The purpose of this step is to select an action path that conforms to the film formation mechanism based on the direction of the local oxygen partial pressure deviation, thereby inhibiting hypoxic and oxygen-rich membrane degradation respectively, and reducing additional disturbances caused by repeated control.

[0194] In a preferred embodiment of the present invention, the intake compensation amount is used to pulse gas injection or cut-off via a local gas distributor near the wafer to regulate the local reaction gas flow rate; the chamber main exhaust valve opening adjustment amount is used to control the chamber main exhaust valve to dynamically adjust the opening to form a basic maintenance control of the background pressure.

[0195] This embodiment provides an execution layer hardware linkage mechanism; specifically, in the aforementioned dynamic adjustment framework, simply describing increasing or decreasing gas flow and adjusting valve opening is insufficient to reflect the actual engineering implementation method.

[0196] Especially in the rapid disturbance scenario of ITO reactive sputtering, conventional uniform gas intake throughout the chamber and low-frequency valve action often cannot simultaneously achieve local chemical compensation and background pressure stability. Therefore, this embodiment further specifies that: the gas intake compensation amount is implemented by pulsed gas injection or cut-off through a local gas distributor close to the wafer, and the opening adjustment amount of the main extraction valve of the chamber is used to implement dynamic opening adjustment to form the basic maintenance control of the background pressure.

[0197] Specifically, the local gas distributor is preferably located in the gas supply path close to the wafer surface, and can be a spray head structure close to the wafer, an annular distribution channel, or a zoned nozzle structure; pulsed gas injection or cut-off is used instead of gas regulation at a fixed continuous flow rate, because changes in the local microenvironment are often faster than the averaging process of the entire chamber.

[0198] Pulsed gas supply can deliver the reaction gas to the vicinity of the wafer within a preset pulse time period, making the compensation effect more direct and spatially targeted; the cut-off action is beneficial to quickly stop the local oversupply when the oxygen enrichment trend appears; its physical essence is to take advantage of the time and space of near-field gas supply to bypass the buffering effect of large volume chamber on signal and gas changes.

[0199] Furthermore, the main exhaust valve of the chamber does not undertake the fine control of local chemistry, but rather the basic maintenance control of background pressure; specifically, the local gas distributor is used to precisely supply oxygen to the target film-forming area, and the main exhaust valve of the chamber is used to maintain the stability of the overall environmental state of the chamber.

[0200] Dynamic opening adjustment can be a continuous small change or a segmented change. Its goal is to absorb the total pressure disturbance caused by pulsed air intake and maintain discharge stability and deposition rate stability. After distinguishing the two types of actuators into near-field chemical regulation and global pressure maintenance, the system control objective is clearer and easier to tune in engineering.

[0201] Furthermore, in order to keep the terminology consistent with the execution layer hardware relationship in the aforementioned embodiments, this embodiment clarifies that: the local gas distributor belongs to the near-wafer distribution part of the gas inlet device, and the descriptions such as the near-wafer gas inlet end local gas inlet end near-field gas supply path appearing in the aforementioned embodiments all correspond to the local gas distributor or its equivalent gas supply component.

[0202] In the aforementioned embodiments, the intake compensation amount is the target adjustment amount output by the controller to the local gas distributor, and its execution result is manifested as the change in the local reactive gas flow rate in the near field region of the wafer; thus, a one-to-one correspondence is formed between the control command layer, the actuator layer and the actual flow layer, without introducing new intake hardware objects.

[0203] Furthermore, in this embodiment, the pulsed gas injection or cutoff is preferably performed superimposed on the basic gas intake of the process formulation; in other words, the basic gas intake is used to maintain the nominal atmosphere window of this process stage, and the pulsed gas intake compensation is used to quickly correct the local real oxygen partial pressure deviation.

[0204] When performing a cut-off action, it is preferable to cut off the local compensation branch or reduce the local compensation branch to zero, rather than necessarily meaning that the basic oxygen supply of the entire chamber is completely shut off at the same time. In this way, the local compensation action and the basic formula action can be performed in layers, thereby avoiding the misunderstanding of near-field rapid correction as a complete replacement of the entire process formula.

[0205] Furthermore, to ensure a clear boundary between the local gas supply and the main extraction gas regulation, the controller preferably generates pulse parameters for the local gas distributor first, and then generates the following pressure stabilization parameters for the main extraction gas valve of the chamber within the same control cycle; wherein, the pulse parameters may include at least one or more of the following: pulse opening duration, pulse interval, and pulse count;

[0206] Valve following parameters can include at least one or more of the following: the direction of target opening change, the magnitude of opening change, and the duration of holding. When the former is used to prioritize the correction of the local chemical environment, the latter is used to limit the macroscopic vacuum fluctuations caused by it. In this way, synchronization in the execution layer is reflected in the coordinated execution under the same round of control decisions, while decoupling is reflected in the independent setting of the two in mathematical solutions and target constraints.

[0207] In a specific linkage control scenario, if the edge area is determined to be oxygen-deficient, the controller does not significantly increase the oxygen flow rate of the entire cavity. Instead, it prioritizes commanding the local gas distributor near the wafer to execute one or more short-term oxygen replenishment pulses, so that oxygen can quickly enter the microenvironment near the edge.

[0208] At the same time, the main exhaust valve is opened appropriately to absorb the background pressure rise caused by pulse gas supply; if it is determined that the local area tends to be oxygen-rich, the local gas distributor reduces the pulse frequency or directly cuts off the local gas supply, and the main exhaust valve is adjusted to a new maintenance opening at the same time; in this way, the local oxygen partial pressure change is mainly controlled by the near-field gas supply, while the total pressure change in the chamber is mainly buffered by the main exhaust valve.

[0209] Under abnormal operating conditions, if the local gas distributor responds slowly or a nozzle becomes blocked, the system can revert to the conservative global air intake combined with the main exhaust valve control mode, but the adjustment frequency will be reduced and the allowable compensation range will be narrowed to prevent the local gas supply unevenness from being further amplified.

[0210] If the main exhaust valve is restricted in its operation, the system limits the pulse air supply intensity to prevent the total pressure from exceeding the limit. If the actual feedback of the two types of actuators is inconsistent with the target action, the controller prioritizes ensuring safe operating conditions and, if necessary, suspends the high-precision dynamic compensation of the current chip to maintain only basic deposition stability.

[0211] In the aforementioned 12-inch wafer coating process, when the edge ring region is determined to have a continuous oxygen-deficient trend, the local gas distributor arranged near the wafer edge performs a short-term oxygen pulse injection, so that the reactive gas can preferentially reach the edge film formation front without significantly disturbing the average state of the entire cavity.

[0212] At the same time, the main extraction valve is adjusted to increase the preset opening increment according to the pulse injection amount, so that the macroscopic vacuum degree is still kept within the process allowable range. When the risk is eliminated, the local gas distributor stops pulse gas supply, the main extraction valve returns to the new maintenance opening, and the entire wafer continues to complete the subsequent ITO deposition according to the stable window.

[0213] The purpose of this step is to achieve a clear decoupling of control actions in terms of space and function by having the local gas distributor be responsible for near-field compensation and the main extraction valve be responsible for background pressure stabilization. This will improve the speed of microenvironment adjustment and reduce interference with the overall cavity stability.

[0214] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. A method for adjusting the vacuum level of ITO chip coating, applied to a coating system including a chamber, a main chamber exhaust valve, an air inlet device with a local gas distributor, a chamber pressure detection unit, a target power supply, a plasma spectral acquisition unit, and a wafer temperature acquisition unit, characterized in that, include: A multi-source sensing input layer is constructed to collect and integrate macroscopic vacuum data, target discharge electrical parameters, plasma spectral characteristic parameters, and real-time wafer temperature. A microenvironment digital twin layer is constructed to calculate the microenvironment state of the wafer surface based on the integrated data. The microenvironment digital twin layer includes a consumption rate calculation module and a partial pressure prediction module. The consumption rate calculation module calculates the transient oxygen consumption rate, and the partial pressure prediction module predicts the local true oxygen partial pressure. A collaborative decision-making and execution layer is constructed, a feedforward feedback collaborative control mechanism is set, and the local reaction gas flow rate of the air intake device and the opening degree of the main exhaust valve of the chamber are adjusted according to the feedforward feedback collaborative control mechanism. The feedforward feedback collaborative control mechanism includes deviation calculation rules, command generation rules, and action decoupling rules. The deviation calculation rules calculate the deviation between the local true oxygen partial pressure and the preset target oxygen partial pressure. The command generation rules generate feedforward control commands based on the deviation values, including the intake compensation amount and the chamber main exhaust valve opening adjustment amount. The action decoupling rules perform independent mathematical calculations on the adjustment of the local reaction gas flow rate and the adjustment of the chamber main exhaust valve opening.

2. The method for adjusting the vacuum degree of ITO coating on a chip according to claim 1, characterized in that, Before inputting the integrated data into the microenvironment digital twin layer, the process also includes: The integrated data is preprocessed to obtain preprocessed data. The preprocessing includes hysteresis compensation, outlier removal, and time alignment. The time alignment involves synchronizing the timestamps of the macroscopic vacuum level data, the target discharge electrical parameters, the plasma spectral characteristic parameters, and the real-time wafer temperature to eliminate acquisition delay differences. The preprocessed data is then input into the microenvironment digital twin layer as the integrated data for subsequent calculations.

3. The method for adjusting the vacuum degree of ITO coating on a chip according to claim 1, characterized in that, The consumption rate calculation module is used to calculate the transient oxygen consumption rate based on the microenvironment gas dynamics model and the discharge electrical parameters of the target material. The consumption rate calculation module is specifically used for: The electrical parameters of the target material discharge are mapped to the surface state of the target material using a microenvironment gas dynamics model. The initial transient oxygen consumption rate of the target material surface is then calculated by combining the surface state of the target material with the plasma spectral characteristic parameters. The voltage baseline drift, short-cycle voltage fluctuation amplitude, and fluctuation occurrence density are extracted from the electrical parameters of the target discharge as voltage fluctuation characteristics. After the voltage fluctuation characteristics are assigned to the corresponding target state labels according to the microenvironment gas dynamics model, the initial transient oxygen consumption rate is quantified using a preset nonlinear mapping function to obtain the transient oxygen consumption rate.

4. The method for adjusting the vacuum degree of ITO coating on a chip according to claim 1, characterized in that, The partial pressure prediction module is used to divide the wafer surface into tiny regions and decompose the transient oxygen consumption rate to obtain the local true oxygen partial pressure of the tiny regions. The voltage divider prediction module is specifically used for: By combining the transient oxygen consumption rate, the macroscopic vacuum data, and the real-time wafer temperature, the microenvironment gas distribution trend is evaluated. The evaluation process includes: Global gas distribution is predicted using a computational fluid dynamics model. Combined with the transient oxygen consumption rate after decomposition, the local true oxygen partial pressure of each small region is calculated. Based on the deviation direction between the local true oxygen partial pressure and the target oxygen partial pressure, the intensity of change in the corresponding direction is extracted as the rate of change of the local true oxygen partial pressure. The configuration is as follows: if the rate of change is greater than a preset rate of change threshold, it is determined that there is a risk of oxygen vacancy imbalance within a preset time period; otherwise, it is determined that there is no risk of oxygen vacancy imbalance.

5. A method for adjusting the vacuum degree of ITO coating on a chip according to any one of claims 1-4, characterized in that, The method further includes: By combining the collaborative decision-making and execution layer with the micro-environment digital twin layer, dynamic adjustment of vacuum level based on micro-environment state is achieved. This dynamic adjustment of vacuum level based on micro-environment state includes: The deviation between the local true oxygen partial pressure and the target oxygen partial pressure is calculated according to the deviation calculation rule, and the feedforward control command is generated based on the deviation value using the collaborative decision and execution layer. The integrated data is used as dynamic input to dynamically adjust the control commands. When the absolute value of the deviation is less than or equal to the preset deviation threshold, the control mode is maintained with smooth fine-tuning as the main function. When the absolute value of the deviation is greater than the preset deviation threshold, the action decoupling rule is switched to in real time. The collaborative decision-making and execution layer is used to synchronously and independently adjust the local reaction gas flow rate and the opening of the main exhaust valve of the chamber through real-time calculation. Based on the closed-loop feedback mechanism of the feedforward control command and the integrated data, the adjustment parameters are automatically updated.

6. The method for adjusting the vacuum degree of ITO coating on a chip according to claim 5, characterized in that, If the local actual oxygen partial pressure is lower than the target oxygen partial pressure, the vacuum level is dynamically adjusted, including: The local reaction gas flow rate is increased, and the opening of the main evacuation valve of the chamber is increased simultaneously to absorb the macroscopic pressure fluctuations caused by the increase in reaction gas and maintain the macroscopic vacuum data within the preset pressure fluctuation threshold range. If the local actual oxygen partial pressure is higher than the target oxygen partial pressure, the vacuum level is dynamically adjusted, including: The flow rate of the local reaction gas is reduced, and the opening of the main exhaust valve of the chamber is reduced simultaneously to maintain the stoichiometry of the microenvironment on the wafer surface at a constant level. If the local true oxygen partial pressure equals the target oxygen partial pressure, dynamic adjustment of the vacuum degree is performed, including: Maintain the local reaction gas flow rate and the opening of the main extraction valve of the chamber at the current state, and continuously monitor the macroscopic vacuum data, the electrical parameters of the target discharge, the plasma spectral characteristic parameters, and the real-time temperature of the wafer.

7. The method for adjusting the vacuum degree of ITO coating on a chip according to claim 1, characterized in that, The intake compensation amount is used to adjust the local reactive gas flow rate by pulsed gas injection or cutoff through the local gas distributor near the wafer. The opening adjustment amount of the main exhaust valve of the chamber is used to control the dynamic opening adjustment of the main exhaust valve of the chamber to form a basic maintenance control of the background pressure.

Images