Method and system for dynamic pressure balance control of mocvd vacuum chamber

Abstract

The present application belongs to the technical field of semiconductor vapor deposition, and particularly relates to a MOCVD vacuum chamber pressure dynamic balance control method and system, which comprises the following steps: collecting and preprocessing multi-dimensional state data; calculating equivalent flow conductance based on real-time data, and constructing flow conductance attenuation index to represent pipeline blockage degree in combination with a reference benchmark; constructing a gain-scheduled extended state observer, correcting model parameters by using the flow conductance attenuation index, and estimating pressure and disturbance of the system in real time; and generating control instructions with reverse gain compensation based on pressure estimation value, disturbance estimation value and flow conductance attenuation index. The present application can perceive physical aging of pipeline flow conductance caused by deposition of by-products in the MOCVD growth process, compensate attenuation of physical gain by adjusting the controller gain in real time, maintain dynamic response characteristics of the closed-loop control system consistent throughout the growth cycle, and thus improve the yield and consistency of the epitaxial wafer growth.

Description

Technical Field

[0001] This invention relates to the field of semiconductor vapor deposition technology. More specifically, this invention relates to a method and system for dynamic pressure balance control in an MOCVD vacuum chamber. Background Technology

[0002] Metal-organic chemical vapor deposition (MOCVD) is a core process technology for fabricating compound semiconductor devices such as GaAs-based heterojunction bipolar transistors and InP-based edge-emitting lasers. During the epitaxial growth of these devices, the pressure in the reaction chamber is a crucial thermodynamic parameter that directly determines the partial pressure of the precursor source on the substrate surface, the thickness of the vapor boundary layer, and the growth rate of the material.

[0003] For high-end epitaxial wafers, their structure is usually composed of dozens or even hundreds of layers of thin films with different compositions and doping concentrations, and the growth cycle often lasts for several hours. During this process, MOCVD equipment faces severe pressure control challenges: on the one hand, in order to form a steep heterojunction interface, the process requires frequent switching between high-flow-rate metal-organic sources and hydrides in the operating pipeline and the venting pipeline, and the valve action will cause transient and violent fluctuations in the chamber pressure; on the other hand, during long-cycle growth, reaction byproducts will continuously deposit on the inner wall of the exhaust gas pipeline and the valve plate of the pressure regulating valve, causing the effective conductance of the pipeline to gradually decrease and the valve opening-flow nonlinear characteristic curve to drift significantly.

[0004] However, existing pressure control technologies, whether classic PID control or conventional active disturbance rejection control, are usually based on the assumption that the system model parameters are constant or slowly time-varying. Existing technologies cannot detect physical aging such as conductance decay, and the controller still uses the initially set fixed gain parameters for calculation. As growth progresses, the controller's adjustment strength gradually becomes mismatched with the actual physical system, which may lead to response hysteresis or oscillation in the later stages of growth, and thus reduce the consistency of the device's electrical performance. Summary of the Invention

[0005] To address the technical problem that existing pressure control technologies cannot detect physical aging such as conductance decay, leading to mismatch between controller parameters and the actual system, resulting in response hysteresis or oscillation and reduced consistency of device electrical performance, this invention provides solutions in the following aspects.

[0006] In a first aspect, the present invention provides a method for dynamic pressure balance control of an MOCVD vacuum chamber, comprising: acquiring multi-dimensional state data during the operation of the MOCVD equipment, the data including real-time pressure of the reaction chamber, total gas flow rate entering the chamber, valve opening degree, and exhaust gas temperature, and performing time-series alignment and filtering on the acquired data; calculating the equivalent conductance of the pipeline system based on the total gas flow rate, real-time pressure, and valve opening degree, and constructing a conductance attenuation index in combination with exhaust gas temperature, reference temperature, and reference conductance to characterize the physical blockage state of the pipeline; constructing a gain-scheduled conductance adaptive expansion state observer, introducing a gain compensation function calculated based on the conductance attenuation index, correcting the model parameters of the observer, estimating the system pressure and disturbance in real time, and obtaining pressure estimates and disturbance estimates; in the process of calculating the final control command based on the system disturbance estimate, pressure estimate, and pressure setpoint, introducing inverse gain compensation based on the gain compensation function; and outputting the final control command to a throttle valve to maintain the dynamic response characteristics of the closed-loop control system.

[0007] This invention calculates equivalent conductance based on fluid mechanics principles and constructs a conductance attenuation index, enabling real-time quantitative characterization of the physical blockage degree of pipelines. Building upon this, a gain-scheduled adaptive conductance expansion state observer is constructed, and a gain compensation function is introduced to correct model parameters, accurately estimating system pressure and disturbances including unmodeled dynamics. Finally, inverse gain compensation is introduced into the control command calculation, offsetting the valve physical gain attenuation caused by pipeline blockage. This maintains the dynamic response characteristics of the closed-loop control system consistently throughout the entire growth cycle, effectively suppressing pressure fluctuations and control lag caused by pipeline blockage, and improving the yield and consistency of epitaxial wafer growth.

[0008] Preferably, the formula for calculating the equivalent conductance of the pipeline system is: In the formula: for Equivalent conductance of the pipeline system at any given time; for The total gas flow rate entering the chamber at any given time; for It constantly reflects the real-time pressure of the chamber; for The valve opening at any given moment; This is a pre-calibrated nonlinear mapping function between valve opening and geometric flow area under clean pipeline conditions. Its input is valve opening, and its output is geometric flow area.

[0009] This invention utilizes a pre-calibrated nonlinear mapping function between valve opening and geometric flow area to effectively eliminate the interference of the valve's own geometric nonlinear characteristics on the conductance calculation. This ensures that the calculated equivalent conductance only reflects the flow capacity of the pipeline system, thereby enabling a more accurate separation of the changing characteristics of the pipeline's physical state from complex operating conditions and providing a reliable physical basis for assessing conductance attenuation.

[0010] Preferably, the formula for calculating the conductance attenuation index is: In the formula: for The conductance attenuation index at time t; The width of the sliding time window; For integration variables Equivalent conductance at any given moment; for The exhaust gas temperature at any given moment; For reference temperature; Used as the reference flow guide.

[0011] This invention constructs a conductivity attenuation index that includes a sliding time window integral and a viscosity correction term. By using a sliding window to integrate and smooth the data, it can filter out non-stationary noise caused by changes in process formulation or fluctuations in exhaust gas temperature. At the same time, by introducing a temperature-based viscosity correction term, it can eliminate the apparent conductivity decrease caused by the increase in gas viscosity due to the increase in exhaust gas temperature. This ensures that the calculated conductivity attenuation index only reflects physical blockage caused by the deposition of reaction byproducts, thus improving the robustness and accuracy of blockage identification.

[0012] Preferably, the discrete state update calculation formula for the gain-scheduled flow-guided adaptive extended state observer is as follows: In the formula: for Pressure estimates for the system at any given time; for Pressure estimates for the system at any given time; The sampling period of the control system; for The disturbance estimate of the system at any given time; for Gain compensation function at time step; for Control commands output by the time controller; for The time estimation error; for The disturbance estimate of the system at any given time; , The observer's correction gain; This is the system's nominal control gain.

[0013] This invention constructs an extended state observer discrete state update mechanism that includes a gain compensation function. By introducing a coefficient that varies with conductance attenuation into the observer model, the internal model parameters of the observer can be adjusted in real time according to the pipeline blockage state. This mechanism enables the observer to perceive the actual physical attenuation of valve control capability, thereby avoiding the erroneous attribution of slow response caused by physical aging to external disturbances. This ensures the accuracy and speed of the observer's pressure tracking values ​​and system disturbance estimation, and achieves real-time matching between the model and the physical system.

[0014] Preferably, the formula for calculating the estimation error is: In the formula: for The time estimation error; for Pressure estimates for the system at any given time; for It constantly reflects the real-time pressure of the chamber.

[0015] Preferably, the gain compensation function is equal to the reciprocal of the difference between 1 and the conductance attenuation index.

[0016] This invention establishes a functional relationship between the gain compensation function and the conductance attenuation index, which can quantitatively convert the degree of pipeline blockage into a correction coefficient of the control gain. When pipeline blockage causes the conductance attenuation index to increase, the function can generate a compensation value significantly greater than 1, thereby accurately reflecting the degree of weakening of the valve's actual control capability relative to the nominal state, and providing a quantitative basis for subsequent observer parameter correction and inverse compensation of the control law.

[0017] Preferably, the formula for calculating the control command is: In the formula: for Control commands output by the time controller; For the controller's bandwidth; for Pressure setting value at any time; for Pressure estimates for the system at any given time; for The disturbance estimate of the system at any given time; The nominal control gain of the system; for Gain compensation function at time step.

[0018] This invention introduces a gain compensation function into the calculation of control commands for inverse compensation. The gain compensation function amplifies the basic control quantity calculated based on error and disturbance estimation, which can actively offset the decrease in valve physical gain caused by pipeline blockage. This inverse compensation mechanism ensures that the total loop gain of the system does not change with pipeline aging, so that the controller can always output control force matching the current physical state during long-term growth, maintaining the constancy of control performance.

[0019] Preferably, the method for obtaining the reference temperature and baseline conductance is as follows: after the MOCVD equipment has completed maintenance and cleaning and the exhaust gas pipeline is confirmed to be in the cleanest state without deposits, the equipment is started and enters a thermal equilibrium steady state, and equivalent conductance data and exhaust gas temperature data are continuously collected within a set time period; the arithmetic mean of the equivalent conductance within this time period is calculated as the baseline conductance; and the arithmetic mean of the exhaust gas temperature within this time period is calculated as the reference temperature.

[0020] Preferably, the method further includes limiting protection: if the calculated control command exceeds the physical limit of the valve, the control command output by the controller is adjusted to the maximum physical opening, and an alarm is triggered.

[0021] In a second aspect, the present invention provides a dynamic pressure balance control system for an MOCVD vacuum chamber, comprising a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned dynamic pressure balance control method for an MOCVD vacuum chamber is implemented.

[0022] By adopting the above technical solution, the above-mentioned MOCVD vacuum chamber pressure dynamic balance control method is generated into a computer program and stored in a memory so that it can be loaded and executed by a processor. Based on the memory and processor, a terminal device can be made for convenient use.

[0023] The beneficial effects of this invention are as follows:

[0024] This invention calculates equivalent conductance based on fluid mechanics principles and constructs a conductance attenuation index, enabling real-time quantitative characterization of the physical blockage degree of pipelines. Building upon this, a gain-scheduled adaptive conductance expansion state observer is constructed, and a gain compensation function is introduced to correct model parameters, accurately estimating system pressure and disturbances including unmodeled dynamics. Finally, inverse gain compensation is introduced into the control command calculation, offsetting the valve physical gain attenuation caused by pipeline blockage. This maintains the dynamic response characteristics of the closed-loop control system consistently throughout the entire growth cycle, effectively suppressing pressure fluctuations and control lag caused by pipeline blockage, and improving the yield and consistency of epitaxial wafer growth. Attached Figure Description

[0025] Figure 1This is a flowchart illustrating the dynamic balance control method for MOCVD vacuum chamber pressure in this invention;

[0026] Figure 2 This is a schematic diagram showing the comparison of the dynamic balance control effect of MOCVD vacuum chamber pressure;

[0027] Figure 3 This is a schematic diagram showing the comparison of controller output commands. Detailed Implementation

[0028] 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, not all, of the embodiments of the present invention. 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.

[0029] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0030] This invention discloses a method for dynamic pressure balance control in MOCVD vacuum chambers, referring to... Figure 1 This includes steps S1-S4:

[0031] S1: Acquire multidimensional status data during the operation of the MOCVD equipment, and perform time-series alignment and filtering on the acquired data.

[0032] It should be noted that in order to achieve accurate modeling of the pressure control system and real-time identification of the flow conduction state, it is first necessary to acquire multi-dimensional physical parameters during the operation of the equipment. Due to the physical delay in the transmission of gas from the flow controller to the chamber, and the different response times of different sensors, the directly acquired data is misaligned on the time axis. If used directly, it will lead to model identification deviation. Therefore, this invention performs strict time alignment and filtering on the acquired data.

[0033] Specifically, high-frequency data acquisition is performed using the existing sensor network of the MOCVD equipment. The acquired multi-dimensional state data includes: real-time pressure of the reaction chamber acquired by a high-precision capacitive thin-film gauge. The total gas flow rate entering the chamber is obtained by summing the feedback values ​​from all mass flow controllers. The total gas flow rate includes the sum of carrier gas, metal-organic source, hydride, and dopant; the valve opening is acquired by the position encoder of the throttle valve. ; and the exhaust gas temperature collected by thermocouples installed on the exhaust pipe. .

[0034] Furthermore, considering the physical transmission delay of gas from the mass flow controller to the chamber, as well as the differences in the response time of each sensor, the total gas flow data is time-compensated to align it with the real-time pressure on the time axis.

[0035] Furthermore, a sliding weighted average filtering algorithm is used to process the acquired raw data, filtering out high-frequency measurement noise caused by pump vibration or circuit interference, and obtaining a smooth system state sequence.

[0036] S2: Calculate the equivalent conductance of the pipeline system based on the total gas flow rate, real-time pressure and valve opening, and construct the conductance attenuation index by combining the exhaust gas temperature, reference temperature and reference conductance.

[0037] It should be noted that the deposition of reaction byproducts will reduce the effective cross-sectional area of ​​the pipeline, which physically manifests as a decrease in conductance. However, the calculation of physical conductance alone is greatly affected by the fluctuation of exhaust gas temperature and changes in process formulation, and has significant non-stationary noise, so it cannot be directly used as a control basis. Therefore, this invention introduces temperature correction and time-domain integral smoothing to extract stable features from real-time data that are only related to the physical blockage state of the pipeline.

[0038] Specifically, based on fluid mechanics principles, and taking into account the total gas flow rate, real-time pressure, and valve opening, the equivalent conductance of the pipeline system is calculated. The specific calculation formula is as follows:

[0039]

[0040] In the formula: for The equivalent conductance of the pipeline system at any given moment, which reflects the pipeline's ability to carry gas at the current moment; for The total gas flow rate entering the chamber at any given time; the higher the flow rate, the higher the conductivity under the same pressure. for It constantly reflects the real-time pressure of the chamber; the higher the pressure, the lower the conductance at the same flow rate. for The valve opening at any given moment; This is a pre-calibrated nonlinear mapping function between valve opening and geometric flow area under clean pipeline conditions. Its input is the valve opening, and its output is the geometric flow area. This function is used to eliminate the influence of the valve's geometric nonlinear characteristics on conductance calculations, ensuring the equivalent conductance of the pipeline system. It only reflects the condition of the pipeline.

[0041] It should be noted that the valve opening degree When defining the equivalent flow conductance of the piping system .

[0042] Furthermore, a conductance attenuation index was constructed. The specific calculation formula is as follows:

[0043]

[0044] In the formula: for The conductance attenuation exponent at time t, its range is . The larger this value, the more severe the pipe blockage. The width of the sliding time window is determined by the response settling time of the mass flow controller. To set, usually take ; For integration variables Equivalent conductance at any given moment; for The exhaust gas temperature at any given moment; For reference temperature; The reference flow guide is used; where This is a viscosity correction term based on the Sutherland formula, used to eliminate the decrease in apparent conductivity caused by the increase in gas viscosity due to temperature rise, ensuring... It only reflects physical deposition.

[0045] Among them, the reference temperature is acquired synchronously. and reference flow guide The specific process is as follows: After the MOCVD equipment has completed maintenance and cleaning and the exhaust gas pipeline is confirmed to be in its cleanest state without deposits, the equipment is started and typical process flow and pressure conditions are set; after the system enters a thermal equilibrium steady state, equivalent conductance data and exhaust gas temperature data are continuously collected for at least 5 minutes; the arithmetic mean of the equivalent conductance during this period is calculated as the baseline conductance. Simultaneously, the arithmetic mean of the exhaust gas temperature during this period is calculated and directly fixed as the reference temperature. This method ensures and It strictly corresponds to the same physical reference state, eliminating the calculation error caused by reference mismatch.

[0046] S3: Construct a gain-scheduled adaptive expansion state observer for the flow guide, introduce a gain compensation function calculated based on the flow guide attenuation exponent, correct the model parameters of the observer, estimate the system pressure and disturbance in real time, and obtain pressure and disturbance estimates.

[0047] It should be noted that traditional extended state observers (ESOs) typically assume that the system's control gain is constant and attribute all unmodeled dynamics to the disturbance term. However, in the long-growth scenario of MOCVD, pipe blockage causes a deterministic unidirectional decay in the physical control gain of the valve. If the observer does not perceive this change, it will incorrectly estimate the gain decay as an external disturbance, resulting in a lag in the observation of the actual pressure change. Therefore, this invention introduces a gain compensation function to correct the observer's model parameters, thereby achieving real-time matching between the model and the physical system.

[0048] Specifically, a gain-scheduled flow-guide adaptive extended state observer is constructed, and its discrete state update is calculated as follows:

[0049]

[0050] In the formula: for The time estimation error; for The pressure estimate of the system at any given time, i.e., for Real-time pressure of the reaction chamber The tracking estimate; for It constantly reflects the real-time pressure of the chamber; for Pressure estimates for the system at any given time; for Pressure estimates for the system at any given time; The sampling period of the control system; for The disturbance estimate of the time system includes external flow disturbances and model-unmodeled errors; for Gain compensation function at time step; for The control command output by the time controller, that is, the opening command sent to the throttle valve; for The disturbance estimate of the system at any given time; , This is the correction gain of the observer, used to ensure the convergence of observation errors. , , For the bandwidth of the observer, The range of values ​​is , For the controller's bandwidth; The nominal control gain of the system, i.e., the rate of pressure change caused by a unit change in control quantity under clean pipeline conditions, is obtained through the step response calibration method.

[0051] in, Gain compensation function at time step The formula for calculation is:

[0052]

[0053] In the formula: for The conductance attenuation index at time t.

[0054] When the pipeline becomes blocked, Increasing the gain compensation function Significantly greater than 1; in the observer equation, the coefficient of the control input term becomes This means that the observer has sensed the current valve action. stress The actual impact has weakened, thus avoiding the erroneous attribution of this slowdown in response due to physical aging to negative external disturbances. This ensured It can accurately and quickly reflect real external shocks such as gas path switching.

[0055] S4: In the process of calculating the final control command based on the system disturbance estimate, pressure estimate and pressure setpoint, inverse gain compensation is introduced based on the gain compensation function; the final control command is output to the throttle valve to maintain the dynamic response characteristics of the closed-loop control system.

[0056] It should be noted that, due to the decrease in the physical control gain of the valve caused by pipeline blockage, in order to maintain the dynamic response characteristics of the closed-loop control system consistently throughout the entire growth cycle, the controller must output a larger control quantity than the nominal state to offset the attenuation of the physical gain; therefore, this invention introduces inverse gain compensation into the control law to ensure that the total loop gain remains constant.

[0057] Specifically, based on the system's disturbance estimate Pressure estimates and pressure setpoint Calculate the control commands output by the controller. The specific calculation formula is as follows:

[0058]

[0059] In the formula: for The control command output by the time controller, that is, the opening command sent to the throttle valve; The bandwidth of the controller determines the system's response speed in tracking the setpoint; for Before the equipment is put into operation, the process engineer will set the pressure set value at any time on the main control computer of the equipment through the recipe editing software, based on the structural design requirements of the specific semiconductor device. for Pressure estimates for the system at any given time; for The disturbance estimate of the system at any given time; The nominal control gain of the system; for Gain compensation function at time step.

[0060] As the pipeline becomes blocked, the physical control gain of the valve decreases from its nominal value. The gradual attenuation is a shrinking process, which weakens the valve's ability to regulate pressure. To counteract this physical attenuation, the standard control value is amplified after calculation. This is an inverse gain compensation operation that is opposite to the physical attenuation direction.

[0061] Among them, the numerator term The basic control force required to eliminate the current tracking error was calculated. The term utilizes the disturbance estimate provided by the observer for feedforward compensation, which can quickly offset external airflow impacts; the overall control quantity is multiplied by the gain compensation function. ,because This makes the final output control commands The amplification effect precisely offsets the physical gain attenuation caused by pipeline blockage, ensuring the consistency of control performance.

[0062] Furthermore, after calculating the control commands... Then, limit the amplitude protection: if the calculated control command If the valve's physical limit, i.e., the maximum physical opening, is exceeded, the controller will adjust the control command output to the maximum physical opening and trigger an alarm.

[0063] For example, Figure 2The graph shows a comparison of the dynamic balance control effect of MOCVD vacuum chamber pressure. In the initial stage, due to the relatively clean pipeline, the attenuation coefficient is close to 1.0, and the curves corresponding to the existing technology and the present invention can track the set value well. However, as time goes on, physical blockage intensifies, and the attenuation coefficient approaches 0.4. The curves corresponding to the existing technology begin to show obvious response hysteresis and steady-state error, especially at the pressure switching points of 1500 seconds and 2500 seconds. Since the controller still uses the initial parameters, it cannot provide sufficient valve opening to offset the decrease in conductance, resulting in very slow pressure regulation, or even failure to reach the target value. In contrast, the curve corresponding to the present invention closely follows the set value throughout the entire life cycle. Even in the later stages of growth, when the pipeline is severely blocked, the control response remains rapid and accurate. This confirms that the gain compensation function in the present invention effectively corrects the controller model.

[0064] For example, Figure 3 The comparison chart shows the output commands of the controller. The valve opening of the curve corresponding to the present invention is significantly higher than that of the curve corresponding to the prior art, and the gap gradually widens over time. This is because as the conductance attenuation index increases, the gain compensation function increases. The present invention actively amplifies the control command by increasing the gain compensation function, thereby making the controller aware that the valve has become less sensitive. Therefore, it outputs a larger action amplitude to compensate for the decrease in physical capability, thereby maintaining the total gain of the closed-loop system constant.

[0065] In summary, compared with traditional methods that cannot detect physical aging, this solution ensures that MOCVD equipment can maintain rapid and consistent pressure control even when facing severe pipeline deposit blockage during long-cycle epitaxial growth by identifying the conductance attenuation index in real time and introducing a gain compensation function.

[0066] It should be noted that by limiting the calculated adaptive control command, when the opening requested by the controller exceeds the maximum physical opening of the valve, the command can be forcibly limited to the physically feasible range and an alarm can be triggered. This mechanism not only prevents abnormalities caused by integral saturation or calculation overflow of the controller, but also promptly warns the operator that the pipeline blockage has reached the equipment limit, avoiding safety accidents caused by overload or loss of control of the actuator, and ensuring the safe operation of the equipment.

[0067] This invention also discloses a dynamic pressure balance control system for MOCVD vacuum chambers, including a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement the MOCVD vacuum chamber pressure dynamic balance control method according to the present invention.

[0068] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and will not be described in detail here.

Claims

1. A method for dynamic pressure balancing control of a MOCVD vacuum chamber, characterized in that, include: Multidimensional status data during the operation of the MOCVD equipment is acquired, including real-time pressure of the reaction chamber, total gas flow rate into the chamber, valve opening degree, and exhaust gas temperature. The acquired data is then time-aligned and filtered. The equivalent conductance of the pipeline system is calculated based on the total gas flow rate, real-time pressure and valve opening, and a conductance attenuation index is constructed by combining the exhaust gas temperature, reference temperature and benchmark conductance to characterize the physical blockage state of the pipeline. A gain-scheduled adaptive expansion state observer for flow guidance is constructed. A gain compensation function based on the flow guidance attenuation exponent is introduced to correct the model parameters of the observer. The system pressure and disturbance are estimated in real time to obtain pressure and disturbance estimates. In the process of calculating the final control command based on the system disturbance estimate, pressure estimate, and pressure setpoint, inverse gain compensation is introduced based on the gain compensation function; the final control command is output to the throttle valve to maintain the dynamic response characteristics of the closed-loop control system. The formula for calculating the equivalent conductance of a piping system is: ; for Equivalent conductance of the pipeline system at any given time; for The total gas flow rate entering the chamber at any given time; for It constantly reflects the real-time pressure of the chamber; for The valve opening at any given moment; This is a pre-calibrated nonlinear mapping function between valve opening and geometric flow area under clean pipeline conditions. Its input is valve opening and its output is geometric flow area. The formula for calculating the conductivity attenuation index is: ; for The conductance attenuation index at time t; The width of the sliding time window; For integration variables Equivalent conductance at any given moment; for The exhaust gas temperature at any given moment; For reference temperature; As the reference flow guide; The formula for calculating the discrete state update of the gain-scheduled flow-guided adaptive extended state observer is as follows: ； for Pressure estimates for the system at any given time; for Pressure estimates for the system at any given time; The sampling period of the control system; for The disturbance estimate of the system at any given time; for Gain compensation function at time step; for Control commands output by the time controller; for The time estimation error; for The disturbance estimate of the system at any given time; , The observer's correction gain; The nominal control gain of the system; The gain compensation function is equal to the reciprocal of the difference between 1 and the conductance attenuation index; The formula for calculating control commands is: ; for Control commands output by the time controller; For the controller's bandwidth; for Pressure setting value at any time; for Gain compensation function at time step.

2. The MOCVD vacuum chamber pressure dynamic balance control method of claim 1, wherein, The formula for calculating the estimation error is: ； In the formula: for The time estimation error; for Pressure estimates for the system at any given time; for It constantly reflects the real-time pressure of the chamber.

3. The MOCVD vacuum chamber pressure dynamic balance control method of claim 1, wherein, The method for obtaining the reference temperature and the reference conductance is as follows: After the MOCVD equipment has completed maintenance and cleaning and the exhaust gas pipeline is confirmed to be in its cleanest state with no deposits, the equipment is started and enters a thermal equilibrium steady state. The equivalent conductance data and exhaust gas temperature data are continuously collected within a set time period. The arithmetic mean of the equivalent conductance within this time period is calculated as the reference conductance. Calculate the arithmetic mean of the exhaust gas temperature over this period, and use it as a reference temperature.

4. The MOCVD vacuum chamber pressure dynamic balance control method of claim 1, wherein, The method also includes amplitude limiting protection: if the calculated control command exceeds the physical limit of the valve, the control command output by the controller is adjusted to the maximum physical opening and an alarm is triggered.

5. A MOCVD vacuum chamber pressure dynamic balance control system, characterized in that, include: A processor and a memory, wherein the memory stores computer program instructions that, when executed by the processor, implement the MOCVD vacuum chamber pressure dynamic balance control method according to any one of claims 1-4.

Images