Pressure fault-tolerant control method, device and equipment of semiconductor process cavity and medium

Abstract

The application provides a pressure fault-tolerant control method, device, equipment and medium for a semiconductor process cavity, comprising: collecting process parameters of the semiconductor process cavity, applying low-pass filtering to a physical pressure sensor signal to output filtered pressure; determining a cavity virtual pressure in each fixed period based on a virtual pressure model, continuously running a shadow pressure online alignment algorithm during a period when the physical pressure sensor is determined to be in a healthy state to generate a shadow pressure; using a double-criterion joint diagnosis to diagnose whether the physical pressure sensor has failed, and switching the feedback source of the pressure control loop to the shadow pressure when all four gating conditions are met; performing PID disturbance compensation in the current period when the switching occurs, executing a differentiated degradation strategy according to the sensitivity of the current process step, and continuously monitoring the operating conditions. The application realizes online fault-tolerant stable control of the semiconductor process cavity pressure sensor when it fails, reduces process interruption and yield loss caused by sensor failure.

Description

Technical Field

[0001] This application relates to the technical field of intelligent control of semiconductor manufacturing equipment, and in particular to pressure fault-tolerant control methods, devices, equipment and media for semiconductor process cavities. Background Technology

[0002] In semiconductor etching, thin film deposition, cleaning, or precision gas mixing processes, the mixed gas input into the process chamber typically needs to maintain a stable pressure under low pressure or vacuum. A typical configuration of existing systems includes: an inlet unit: a mass flow controller (MFC) to control the gas flow rate into the chamber; an exhaust unit: a high-vacuum control valve + pump assembly, which controls the exhaust resistance and effective pumping speed by adjusting the valve opening; and a gas detection unit: a pressure sensor + a temperature sensor, with the pressure sensor serving as the feedback source for the pressure closed-loop control system. The temperature sensor primarily serves as the feedback source for wafer fabrication process parameters. The process chamber has a fixed volume. Existing control systems typically only offer the following responses to pressure sensor failures, all of which have significant drawbacks: 1. In common PID pressure closed-loop control systems, the feedback source is the sole sensor signal channel. If the sensor fails, the equipment immediately stops, leading to process interruption and scrapping of the current batch of wafers. 2. Common virtual sensor solutions lack a process for aligning virtual sensor values ​​with real values. 3. In traditional shadow pressure control, the feedback source switching condition is a fixed threshold fault judgment. When a switch occurs, the PID control source experiences a jump, causing a sudden change in the opening of the exhaust proportional valve (air hammer effect), which leads to disturbances in the flow field within the cavity and reduces the wafer's process quality. Therefore, improving the stability of pressure fault-tolerant control in semiconductor process cavities has become a significant technical challenge. Summary of the Invention

[0003] In view of this, the purpose of this application is to provide a method, apparatus, equipment and medium for pressure fault-tolerant control of semiconductor process cavities. By online alignment of shadow pressure during health period, dual-criteria fault diagnosis, gating feedback source switching and PID disturbance-free compensation, stable online fault-tolerant control is achieved when the pressure sensor of the semiconductor process cavity fails. This maximizes production line uptime under the boundary conditions of ensuring process safety and wafer quality, and reduces process interruption and yield loss caused by sensor failure.

[0004] This application provides a pressure fault-tolerant control method for a semiconductor process cavity, the pressure fault-tolerant control method comprising: The process parameters of the semiconductor process cavity are synchronously acquired in each fixed cycle, and the physical pressure sensor signal is subjected to a first-order low-pass filter to output the filtered pressure. The virtual pressure model based on the law of conservation of mass determines the virtual pressure of the cavity in each fixed period. During the period when the physical pressure sensor is determined to be in a healthy state, the shadow pressure online alignment algorithm is continuously run based on the filtered pressure and the virtual pressure of the cavity to generate shadow pressure. A dual-criteria joint diagnosis is used to determine whether the physical pressure sensor is faulty. If a fault occurs and all gating conditions are met, the feedback source of the pressure control loop is switched from the filtered pressure to the shadow pressure. During the current cycle in which the feedback source switch occurs, PID non-disruptive compensation is performed. Differentiated degradation strategies are implemented based on the sensitivity of the current process step. Operating conditions are continuously monitored. If any condition exceeds the limit, degradation is terminated and controlled shutdown is performed.

[0005] In one possible implementation, the virtual pressure model based on the law of conservation of mass determines the virtual pressure of the cavity within each fixed period, including: Determine the total intake molar flow rate into the cavity during the current cycle; Based on the equivalent volume pumping speed of the exhaust system under the current state, the virtual pressure of the cavity in the previous cycle, the gas temperature in the cavity, and the gas constant, the total exhaust molar flow rate of the cavity in the current cycle is determined. The total intake molar flow rate, the total exhaust molar flow rate, the pre-calibrated cavity leakage molar flow rate, the effective cavity volume, the gas constant, the gas temperature inside the cavity, and the cavity virtual pressure of the previous cycle are used to determine the cavity virtual pressure of the current cycle.

[0006] In one possible implementation, the step of continuously running a shadow pressure online alignment algorithm based on the filtered pressure and the cavity virtual pressure during the period when the physical pressure sensor is determined to be in a healthy state to generate shadow pressure includes: The deviation of the current cycle is determined based on the filtered pressure, the cavity virtual pressure, the low-pass filter coefficient, and the deviation of the previous cycle. The virtual pressure of the cavity is biased and corrected based on the deviation of the current cycle to generate the shadow pressure, and the update of the deviation is frozen after the fault is confirmed.

[0007] In one possible implementation, the method of using dual criteria to jointly diagnose whether a physical pressure sensor has malfunctioned includes: Based on the virtual pressure of the cavity in the current cycle and the virtual pressure of the cavity in the previous cycle, the physical pressure change rate of the cavity pressure in the current cycle is determined. If the physical pressure change rate is greater than the theoretical maximum pressure change rate in the current cycle, then a sudden fault is determined to have occurred. Furthermore, the consistency of the residual between the shadow pressure and the cavity virtual pressure determines whether a persistent drift fault has occurred; if either condition is met, the physical pressure sensor is confirmed to be faulty.

[0008] In one possible implementation, performing PID disturbance-free compensation in the current cycle in which the feedback source switch occurs includes: The integral state of the PID controller is reconstructed by back-calculating the integral term, while the historical input state of the derivative controller is replaced with the current shadow pressure, so that the change in the output of the PID controller to the valve before and after the switch is controlled within the preset allowable range.

[0009] In one possible implementation, the reconstructing of the integral state of the PID controller through inverse calculation of the integral term includes: I_new=CV_old-Kp×e_new(k)-D_hold-FF–Bias; Where I_new is the reconstructed integral term, CV_old is the PID control output of the previous cycle before switching, Kp is the proportional gain, e_new(k) is the control error after switching, D_hold is the historical hold value of the derivative term, FF is the feedforward compensation term, and Bias is the fixed bias value.

[0010] In one possible implementation, the pressure fault-tolerant control method further includes superimposing a rate-of-change limit on the output of the PID controller: CV_out(k) = CV_last(k-1) + Clamp(CV_raw(k) - CV_last(k-1),-CV_rate_max, +CV_rate_max); Wherein, CV_out(k) is the valve opening output in the current cycle, CV_last(k-1) is the valve opening output in the previous cycle, CV_raw(k) is the unrestricted raw control target, CV_rate_max is the allowed rate of change of physical pressure for each cycle, and Clamp is the limiting function.

[0011] This application embodiment also provides a pressure fault-tolerant control device for a semiconductor process cavity, the pressure fault-tolerant control device comprising: The data processing module is used to synchronously acquire the process parameters of the semiconductor process cavity in each fixed cycle, apply a first-order low-pass filter to the physical pressure sensor signal, and output the filtered pressure. The pressure processing module is used to determine the virtual pressure of the cavity in each fixed period based on the virtual pressure model based on the law of conservation of mass. During the period when the physical pressure sensor is determined to be in a healthy state, the shadow pressure online alignment algorithm is continuously run based on the filtered pressure and the virtual pressure of the cavity to generate shadow pressure. The dual-criteria fault diagnosis module is used to jointly diagnose whether the physical pressure sensor has failed using dual criteria. If a failure occurs and all gating conditions are met, the feedback source of the pressure control loop is switched from the filtered pressure to the shadow pressure. The compensation management module is used to perform PID non-disruptive compensation in the current cycle when the feedback source switch occurs, execute differentiated degradation strategies according to the sensitivity of the current process step, continuously monitor operating conditions, and terminate degradation and execute controlled shutdown if any condition exceeds the limit.

[0012] This application also provides an electronic device, including: a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the electronic device is running, the processor communicates with the memory via the bus. When the machine-readable instructions are executed by the processor, the steps of the pressure fault-tolerant control method for semiconductor process cavities described above are performed.

[0013] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the steps of the pressure fault-tolerant control method for semiconductor process cavities described above.

[0014] The embodiments of this application provide a pressure fault-tolerant control method, apparatus, device, and medium for semiconductor process cavities. The pressure fault-tolerant control method includes: synchronously acquiring process parameters of the semiconductor process cavity in each fixed cycle, applying a first-order low-pass filter to the physical pressure sensor signal, and outputting the filtered pressure; determining the cavity virtual pressure in each fixed cycle based on a virtual pressure model using the law of mass conservation; continuously running a shadow pressure online alignment algorithm based on the filtered pressure and the cavity virtual pressure during the period when the physical pressure sensor is determined to be in a healthy state to generate shadow pressure; using dual criteria to jointly diagnose whether the physical pressure sensor has failed; if a failure occurs and all gating conditions are met, switching the feedback source of the pressure control loop from the filtered pressure to the shadow pressure; performing PID disturbance-free compensation in the current cycle in which the feedback source switch occurs; implementing a differentiated degradation strategy according to the sensitivity of the current process step; continuously monitoring operating conditions; terminating degradation and performing controlled shutdown if any condition exceeds the limit. By implementing online alignment of shadow pressure during the health period, dual-criteria fault diagnosis, gating feedback source switching, and PID non-disruptive compensation, stable online fault-tolerant control is achieved when the pressure sensor in the semiconductor process cavity fails. This maximizes production line uptime while ensuring process safety and wafer quality, and reduces process interruptions and yield losses caused by sensor failures.

[0015] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 A flowchart of a pressure fault-tolerant control method for a semiconductor process cavity provided in an embodiment of this application; Figure 2 A schematic diagram of a pressure-tolerant control method for a semiconductor process cavity provided in this application embodiment; Figure 3 A schematic diagram of a pressure fault-tolerant control device for a semiconductor process cavity provided in an embodiment of this application; Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. Based on the embodiments of this application, every other embodiment obtained by those skilled in the art without inventive effort falls within the scope of protection of this application.

[0019] First, the applicable application scenarios of this application will be introduced. This application can be applied to the technical field of intelligent control of semiconductor manufacturing equipment.

[0020] Research has revealed that existing control systems typically employ only the following countermeasures when pressure sensors malfunction as described above, all of which have significant drawbacks: 1. The feedback source in common PID pressure closed-loop control systems is the sole sensor signal channel. If the sensor fails, the equipment immediately shuts down, leading to process interruption and scrapping of the current batch of wafers. 2. Common virtual sensor solutions lack a process for aligning virtual sensor values ​​with real values. 3. In traditional shadow pressure control, the feedback source switching condition is a fixed threshold fault detection. When the switching occurs, the PID control source jumps, causing a sudden change in the opening of the exhaust proportional valve (air hammer effect), resulting in flow field disturbance within the cavity and reducing wafer process quality. Therefore, improving the stability of pressure fault-tolerant control in semiconductor process cavities has become a critical technical challenge.

[0021] Based on this, this application provides a pressure fault-tolerant control method for semiconductor process cavities. By online alignment of shadow pressure during the health period, dual-criteria fault diagnosis, gating feedback source switching, and PID disturbance-free compensation, stable online fault-tolerant control is achieved when the pressure sensor in the semiconductor process cavity fails. This maximizes production line uptime under the boundary conditions of ensuring process safety and wafer quality, and reduces process interruptions and yield losses caused by sensor failures.

[0022] Please see Figure 1 , Figure 1 This is a flowchart illustrating a pressure-tolerant control method for a semiconductor process cavity, provided as an embodiment of this application. Figure 1 As shown in the figure, the stress-tolerant control method provided in this application embodiment includes: S101: Synchronously acquires process parameters of the semiconductor process cavity in each fixed cycle, applies a first-order low-pass filter to the physical pressure sensor signal, and outputs the filtered pressure.

[0023] It should be noted that the process parameters include the actual flow feedback values ​​of each mass flow controller, the opening feedback value of the high vacuum control valve, the operating status of the vacuum pump group, the physical pressure sensor signal of the pressure sensor, and the current process recipe step information. A first-order low-pass filter is applied to the physical pressure sensor signal to obtain the filtered pressure P_sensor_LPF(k).

[0024] Here, the filtering formula is: P_sensor_LPF(k) = β × P_sensor_LPF(k-1) + (1-β) × P_sensor(k), where β = Tau_f / (Tau_f + Δt), and Δt is the fixed interrupt period. Δt is the PLC fixed interrupt period (typically 10ms), and Tau_f is the desired filtering time constant (recommended range: 20ms~100ms). The larger the β value, the smoother the filtering and the slower the response; the smaller the β value, the faster the response and the weaker the noise reduction effect. P_sensor_LPF(k) is the pressure after first-order low-pass filtering at the current moment, and P_sensor is the physical pressure sensor signal.

[0025] S102: The virtual pressure model based on the law of conservation of mass determines the virtual pressure of the cavity in each fixed period. During the period when the physical pressure sensor is determined to be in a healthy state, the shadow pressure online alignment algorithm is continuously run based on the filtered pressure and the virtual pressure of the cavity to generate shadow pressure.

[0026] In this step, a virtual pressure model based on the law of conservation of mass is constructed, and the virtual pressure P_VIR(k) of the cavity is recursively calculated in each fixed period. During the period when the physical pressure sensor is determined to be in a healthy state, the shadow pressure online alignment algorithm is continuously run, and the virtual pressure is biased and corrected by using the deviation Offset(k) between the filtered pressure and the virtual pressure of the cavity to generate the shadow pressure P_shadow(k).

[0027] In one possible implementation, the virtual pressure model based on the law of conservation of mass determines the virtual pressure of the cavity within each fixed period, including: Determine the total intake molar flow rate into the cavity during the current cycle; based on the equivalent volume pumping speed of the exhaust system under the current state, the cavity virtual pressure of the previous cycle, the cavity gas temperature, and the gas constant, determine the total exhaust molar flow rate out of the cavity during the current cycle; use the total intake molar flow rate, the total exhaust molar flow rate, the pre-calibrated cavity leakage molar flow rate, the cavity effective volume, the gas constant, the cavity gas temperature, and the cavity virtual pressure of the previous cycle to determine the cavity virtual pressure of the current cycle.

[0028] Here, based on the physical model of the cavity, the expected value of the cavity pressure is calculated in each PLC cycle using the law of conservation of mass, forming the virtual pressure P_VIR. The dynamic equation of the cavity pressure (discrete form) is a variation derived from the known ideal gas law PV=nRT, as follows: P_VIR(k)=P_VIR(k-1)+[R×T / V_chamber]×(n_in(k)-n_out(k)-n_leak(k))×Δt; Where V_chamber is the effective volume of the cavity (m³), n_in(k) is the current inlet molar flow rate of the cavity, n_out(k) is the current outlet molar flow rate of the cavity, and n_leak(k) is the current leakage molar flow rate of the cavity. The current leakage molar flow rate needs to be calibrated in advance: close the inlet and outlet, measure the pressure rise rate, and obtain a fixed leakage rate. Δt is the fixed period time, T is the gas temperature inside the cavity, and R is the gas constant.

[0029] Here, the formula for calculating the current inlet molar flow rate of the cavity is as follows: n_in(k) = n_in,mfc_1(k) + n_in,mfc_2(k) + ... + n_in,mfc_n(k), where: n_in,mfc_1(k) is the inlet molar flow rate of the first branch, n_in,mfc_2(k) is the inlet molar flow rate of the second branch, and so on.

[0030] The formula for calculating the intake molar flow rate of a single branch at the current moment is as follows, taking the first branch as an example: n_in,mfc_1(k)=Q_mfc_1×(10 -6 / 60)×[P_std / (R×T_std)]×K_gas,mfc_1, where: Q_mfc_1 is the flow monitoring value of MFC_1 in the first branch, P_std=101325Pa, T_std=273.15K, R=8.314J / (mol·K), and K_gas,mfc_1 is the gas correction coefficient (for gases other than N2, such as CF4, Ar, O2, etc., to correct for differences in thermal conductivity).

[0031] It should be noted that the current cavity exhaust molar flow rate is affected by the error in the number of molecules caused by the chemical reaction of the process gas. The calculation formula is also based on the shadow pressure of the previous healthy period, as follows: n_out(k)=[LUT(θ(k),N_pump(k))]·P_VIR(k-1) / (R·T), where LUT() is an effective pumping capacity lookup table obtained by calibrating the valve opening θ and the vacuum pump operating status, and its output corresponds to the equivalent volume pumping speed (m³ / s). P_VIR(k-1) is the virtual pressure value of the previous moment.

[0032] In this application, when constructing the virtual pressure model, the recursive calculation of the exhaust molar flow rate is based entirely on the virtual pressure of the cavity at the previous moment, without relying on the real-time readings of the physical pressure sensor. This design fundamentally severs the computational coupling chain between the virtual model and the faulty sensor. Even if the physical sensor experiences a serious fault such as a jump, jamming, or disconnection, its erroneous reading will not be injected back into the recursive process of virtual pressure through exhaust flow rate calculation. This ensures that the virtual pressure can still independently, continuously, and without contamination calculate the real pressure of the cavity during sensor failure. This independence is the model foundation for all subsequent steps such as shadow pressure alignment, dual-criteria diagnosis, and fault-tolerant switching, solving the technical problem in traditional solutions where "the coupling between the virtual model and the sensor leads to the loss of both if one is damaged."

[0033] In one possible implementation, the step of continuously running a shadow pressure online alignment algorithm based on the filtered pressure and the cavity virtual pressure during the period when the physical pressure sensor is determined to be in a healthy state to generate shadow pressure includes: A: Based on the filtered pressure, the cavity virtual pressure, the low-pass filter coefficient, and the deviation of the previous cycle, the deviation of the current cycle is determined.

[0034] Here, the deviation Offset(k) for the current period is determined using the following formula: Offset(k)=α×(P_sensor_LPF(k)-P_VIR(k))+(1-α)×Offset(k-1); Where α is the low-pass filter coefficient (0 < α < 1). P_shadow possesses the dynamic response characteristics of the virtual model and absorbs temperature drift, calibration error, and long-term model bias through continuous alignment during the health period, minimizing the feedback source difference (P_sensor_LPF - P_shadow) at the time of failure, thus providing a basis for subsequent disturbance-free handover. Offset continuously absorbs comprehensive errors such as unmodeled dynamics, parameter drift, temperature changes, long-term leakage errors, and system biases caused by gas reactions.

[0035] B: Based on the deviation of the current cycle, the virtual pressure of the cavity is biased and corrected to generate the shadow pressure, and the update of the deviation is frozen after the fault is confirmed.

[0036] Here, the deviation of the current cycle is added to the virtual pressure of the cavity to obtain the shadow pressure P_shadow of the current cycle.

[0037] It should be noted that when the fault flag is triggered, the Offset freezes, stops updating, and retains the last estimated value before the fault. Short alignment time requirement: The system must complete at least 60 seconds of continuous alignment while the sensor is in a healthy state before P_shadow is allowed to take over the control loop.

[0038] To ensure that the shadow pressure P_shadow(k) has fully absorbed the system's systematic deviations and converged to the set high-precision steady-state error band before taking over the control loop, the system runtime must satisfy T_align≥n. τ_α. The convergence factor n is selected as 5, corresponding to a response convergence of 99.33% for the first-order system (i.e., entering the ±1% residual transition region); the estimated time constant τ_α is tuned to 12s based on the temperature drift and slow leakage characteristics in micro-pressure gas dynamics. Under the fixed interruption period Δt = 10"ms" of the controller, the corresponding discrete filter coefficient α is preferably 0.000833. Through this quantitative constraint, the shortest alignment time T_align = 5 × 12 = 60s is derived.

[0039] This application continuously runs the shadow pressure online alignment algorithm during sensor health. Through deviation estimation using a first-order low-pass filter, the shadow pressure retains the rapid dynamic response characteristics of the virtual model while continuously absorbing system deviations caused by temperature drift, calibration errors, long-term leaks, and gas reactions from the health period data. Upon fault confirmation, the deviation is immediately frozen, maintaining the last health period estimate before the fault unchanged, ensuring that the control reference is not secondary contaminated by erroneous sensor readings during the fault period.

[0040] S103: Use dual criteria to jointly diagnose whether the physical pressure sensor has failed. If a failure occurs and all gating conditions are met, switch the feedback source of the pressure control loop from the filtered pressure to the shadow pressure.

[0041] In this step, dual criteria are used to jointly diagnose whether the physical pressure sensor has malfunctioned. If a malfunction occurs and all gating conditions are met, the feedback source of the pressure control loop is switched from the filtered pressure to the shadow pressure.

[0042] In one possible implementation, the method of using dual criteria to jointly diagnose whether a physical pressure sensor has malfunctioned includes: a: Based on the virtual pressure of the cavity in the current cycle and the virtual pressure of the cavity in the previous cycle, the physical pressure change rate of the cavity pressure in the current cycle is determined. If the physical pressure change rate is greater than the theoretical maximum pressure change rate in the current cycle, then a sudden fault is determined to have occurred.

[0043] Here, the physical pressure change rate dP_phys(k) = [P_sensor_LPF(k) - P_sensor_LPF(k-1)] / Δt within the current cycle is calculated, and the theoretical maximum pressure change rate that the cavity pressure can physically reach under the current process conditions is R_max(k) = |R×T / V×(n_in(k) - n_out(k))| + M_noise + M_model, where M_noise is the sensor noise margin and M_model is the model error margin; when |dP_phys(k)| > R_max(k), a sudden fault is identified.

[0044] b: And, based on the consistency of the residual between the shadow pressure and the cavity virtual pressure, determine whether a persistent drift fault has occurred; if either condition is met, confirm that the physical pressure sensor is faulty.

[0045] Here, the residual Residual(k) = |P_sensor_LPF(k) - P_shadow(k)| is calculated. When Residual(k) is greater than the residual limit and the duration exceeds the delay threshold, it is determined that a persistent drift fault has occurred.

[0046] The gating conditions include: virtual model confidence condition, shortest shadow pressure alignment time condition, and recipe allowable degradation condition. Specifically, the fault and gating conditions are: FaultConfirmed=TRUE, meaning dual-criteria confirmation of sensor fault; ModelConfidence=TRUE, meaning the virtual pressure model is ready for takeover, determined by a comprehensive assessment of mass flow controller feedback status, valve feedback status, pump operation status, effective pumping capacity lookup table validity, and health period residual statistical error; HealthAlignTime≥60s, meaning the shortest shadow pressure alignment has been completed during the health period; and RecipeAllowDegrade=TRUE, meaning the current process recipe step allows degraded operation. Feedback source switching is executed when all four conditions are met; a controlled safety shutdown is executed if any condition is not met.

[0047] In this application, criterion one dynamically calculates the physical upper limit of the pressure change rate based on the current actual intake and exhaust conditions, enabling real-time capture of instantaneous faults such as EMI interference, open circuits, and short circuits, with response speed limited only by a single control cycle. Criterion two, through long-term residual consistency monitoring of shadow pressure and physical sensor filter values, can identify slowly degrading faults such as diaphragm aging and temperature drift. The two criteria target two fault modes with distinctly different physical characteristics. Criterion one addresses the "fast, urgent, and large" end of the fault timescale spectrum, while criterion two addresses the "slow, stable, and small" end. When combined using an OR logic, they maintain diagnostic sensitivity across the entire fault mode spectrum. This avoids missing sudden faults due to overly lenient fixed thresholds under low flow conditions, and also avoids frequently reporting normal process transients due to overly stringent fixed thresholds under high flow conditions, overcoming the fundamental deficiency of single-criterion diagnostic schemes that are incomplete in addressing all aspects.

[0048] S104: Execute PID non-disruptive compensation in the current cycle when the feedback source switch occurs, implement differentiated degradation strategy according to the sensitivity of the current process step, continuously monitor operating conditions, and terminate degradation and execute controlled shutdown if any condition exceeds the limit.

[0049] In this step, PID disturbance-free compensation is performed in the current control cycle when the feedback source switch occurs. During the degraded operation, a differentiated degraded strategy is executed according to the sensitivity of the current process formulation step. The operating conditions are continuously monitored, and if any condition exceeds the limit, the degrade is terminated and a controlled shutdown is executed.

[0050] In one possible implementation, performing PID disturbance-free compensation in the current cycle in which the feedback source switch occurs includes: The integral state of the PID controller is reconstructed by back-calculating the integral term, while the historical input state of the derivative controller is replaced with the current shadow pressure, so that the change in the output of the PID controller to the valve before and after the switch is controlled within the preset allowable range.

[0051] Here, the integral state of the PID controller is reconstructed by back-calculation using the integral term, while the historical input state of the derivative controller is replaced with the current shadow pressure, so that the change in the output of the PID controller to the valve before and after the switch is controlled within the preset allowable range.

[0052] In one possible implementation, the reconstructing of the integral state of the PID controller through inverse calculation of the integral term includes: I_new=CV_old-Kp×e_new(k)-D_hold-FF–Bias; Where I_new is the reconstructed integral term, CV_old is the PID control output of the previous cycle before the switch, Kp is the proportional gain, e_new(k) is the control error after the switch, D_hold is the historical hold value of the derivative term, FF is the feedforward compensation term, and Bias is the fixed bias value. The I_new calculated above is assigned as the latest integral term to the PID controller, so that the first control output after the switch, CV_new ≈ CV_old, that is, the difference between the control outputs before and after the switch satisfies: |CV_new - CV_old| < ΔCV_max (preset maximum jump threshold).

[0053] Here, to prevent the PID output from jumping due to the slight deviation between P_sensor_LPF and P_shadow at the moment of feedback source switching, which would lead to a sudden change in the opening of the exhaust high vacuum control valve and pressure shock in the cavity, this application performs integral back-calculation compensation in the current PLC cycle where the switching occurs: the error calculation formula is: e_new(k)=SP-P_shadow, where SP is the set value.

[0054] In one possible implementation, the pressure fault-tolerant control method further includes superimposing a rate-of-change limit on the output of the PID controller: CV_out(k) = CV_last(k-1) + Clamp(CV_raw(k) - CV_last(k-1),-CV_rate_max, +CV_rate_max); Wherein, CV_out(k) is the valve opening output in the current cycle, CV_last(k-1) is the valve opening output in the previous cycle, CV_raw(k) is the unrestricted raw control target, CV_rate_max is the allowed rate of change of physical pressure for each cycle, and Clamp is the limiting function.

[0055] In this application, within a single PLC cycle in which the switching occurs, the system reconstructs the integral term in reverse based on the controller output of the previous cycle and the new error environment; at the same time, the historical input state of the differential controller is forcibly replaced with the current shadow pressure value, thus avoiding false high rate of change feedback caused by sensor switching from a mathematical mechanism perspective.

[0056] In one possible implementation, the differentiated degradation strategy includes: the controller reads the current process recipe step information and executes the corresponding degradation operation mode according to the sensitivity of the step; during the degradation operation, the following conditions are continuously monitored, and if any one of them exceeds the limit, the degradation is immediately terminated and a controlled shutdown is executed: the absolute value of the difference between the shadow pressure P_shadow and the set value SP exceeds the pressure tolerance window specified in the recipe; the virtual model confidence condition ModelConfidence is reduced to FALSE (any feedback abnormality of MFC / valve / pump); the degradation operation time exceeds the preset maximum tolerance time T_degrade_max; a manual forced shutdown command is received.

[0057] It should be noted that implementing a differentiated degradation strategy based on the sensitivity of the current process step means: If the current process step is a highly sensitive step (such as etching or deposition critical steps), the degradation condition is: after the shadow pressure take-off completes the current wafer and the cavity is locked, the behavior after degradation is: requesting manual intervention and generating an alarm record. If the current process step is a low-sensitivity step (such as cavity pre-venting or exhaust gas treatment), the degradation condition is: the pressure tolerance window is met and the remaining step time is controllable, the behavior after degradation is: allowed to continue degradation until the entire batch of wafers is processed. If the current process step does not allow degradation (formula locked), then an immediate controlled safety shutdown is implemented, and shadow pressure take-off is not performed.

[0058] The pressure fault-tolerant control method for a semiconductor process cavity of this application is applied to the pressure control system of a semiconductor process cavity. The system includes a process cavity, a temperature controller, a pressure sensor, a high-vacuum control valve, a pump assembly, and a controller. The process cavity is used to contain semiconductor process gases and provide the process reaction environment; the temperature controller is used to control the gas temperature within the process cavity; the pressure sensor is installed on the process cavity to detect the pressure within the cavity in real time and output a physical pressure signal; the high-vacuum control valve is located on the exhaust pipe of the process cavity to regulate the cavity exhaust flow rate; the pump assembly is connected downstream of the high-vacuum control valve to provide pumping power; and the controller is signal-connected to the temperature controller, pressure sensor, high-vacuum control valve, and pump assembly.

[0059] For further details, please refer to Figure 2 , Figure 2 This application provides a schematic diagram of a pressure-tolerant control method for a semiconductor process cavity, as illustrated in an embodiment. Figure 2As shown, S1: Acquire process parameters of the semiconductor process cavity at a fixed period, apply a first-order low-pass filter to the physical pressure sensor signal, and output the filtered pressure. S2: Determine the virtual pressure of the cavity based on the virtual pressure model. S3: Continuously run the shadow pressure online alignment algorithm to generate shadow pressure during the health state. S4: Use dual criteria to jointly diagnose whether the physical pressure sensor has malfunctioned. S5: Switch the PID control data source from the filtered pressure to the shadow pressure. S6: Perform noise-free compensation for the PID output. S7: Degrade the process program according to the user-defined process recipe.

[0060] This application provides a pressure fault-tolerant control method for a semiconductor process cavity. The pressure fault-tolerant control method includes: synchronously acquiring process parameters of the semiconductor process cavity in each fixed cycle, applying a first-order low-pass filter to the physical pressure sensor signal, and outputting the filtered pressure; determining the cavity virtual pressure in each fixed cycle using a virtual pressure model based on the law of mass conservation; continuously running a shadow pressure online alignment algorithm based on the filtered pressure and the cavity virtual pressure during the period when the physical pressure sensor is determined to be in a healthy state to generate shadow pressure; using dual criteria to jointly diagnose whether the physical pressure sensor has failed; if a failure occurs and all gating conditions are met, switching the feedback source of the pressure control loop from the filtered pressure to the shadow pressure; performing PID disturbance-free compensation in the current cycle in which the feedback source switch occurs, implementing a differentiated degradation strategy according to the sensitivity of the current process step, continuously monitoring operating conditions, and terminating degradation and performing controlled shutdown if any condition exceeds the limit. By implementing online alignment of shadow pressure during the health period, dual-criteria fault diagnosis, gating feedback source switching, and PID non-disruptive compensation, stable online fault-tolerant control is achieved when the pressure sensor in the semiconductor process cavity fails. This maximizes production line uptime while ensuring process safety and wafer quality, and reduces process interruptions and yield losses caused by sensor failures.

[0061] Please see Figure 3 , Figure 3 This is a schematic diagram of a pressure fault-tolerant control device for a semiconductor process cavity provided in an embodiment of this application. Figure 3 As shown, the pressure fault-tolerant control device 300 of the semiconductor process cavity includes: The data processing module 310 is used to synchronously acquire the process parameters of the semiconductor process cavity in each fixed cycle, apply a first-order low-pass filter to the physical pressure sensor signal, and output the filtered pressure. The pressure processing module 320 is used to determine the virtual pressure of the cavity in each fixed period based on the virtual pressure model based on the law of conservation of mass. During the period when the physical pressure sensor is determined to be in a healthy state, the shadow pressure online alignment algorithm is continuously run based on the filtered pressure and the virtual pressure of the cavity to generate shadow pressure. The dual-criteria fault diagnosis module 330 is used to jointly diagnose whether the physical pressure sensor has failed using dual criteria. If a failure occurs and all gating conditions are met, the feedback source of the pressure control loop is switched from the filtered pressure to the shadow pressure. The compensation management module 340 is used to perform PID non-disruptive compensation in the current cycle when the feedback source switch occurs, execute differentiated degradation strategies according to the sensitivity of the current process step, continuously monitor operating conditions, and terminate degradation and execute controlled shutdown if any condition exceeds the limit.

[0062] Furthermore, the pressure processing module 320 is used to determine the virtual pressure of the cavity in each fixed period using the virtual pressure model based on the law of conservation of mass. Determine the total intake molar flow rate into the cavity during the current cycle; Based on the equivalent volume pumping speed of the exhaust system under the current state, the virtual pressure of the cavity in the previous cycle, the gas temperature in the cavity, and the gas constant, the total exhaust molar flow rate of the cavity in the current cycle is determined. The total intake molar flow rate, the total exhaust molar flow rate, the pre-calibrated cavity leakage molar flow rate, the effective cavity volume, the gas constant, the gas temperature inside the cavity, and the cavity virtual pressure of the previous cycle are used to determine the cavity virtual pressure of the current cycle.

[0063] Furthermore, the pressure processing module 320 is used to continuously run the shadow pressure online alignment algorithm based on the filtered pressure and the cavity virtual pressure during the period when the physical pressure sensor is determined to be in a healthy state, to generate shadow pressure: The deviation of the current cycle is determined based on the filtered pressure, the cavity virtual pressure, the low-pass filter coefficient, and the deviation of the previous cycle. The virtual pressure of the cavity is biased and corrected based on the deviation of the current cycle to generate the shadow pressure, and the update of the deviation is frozen after the fault is confirmed.

[0064] Furthermore, the dual-criteria fault diagnosis module 330 is used to diagnose whether the physical pressure sensor has malfunctioned using a dual-criteria joint diagnosis: Based on the virtual pressure of the cavity in the current cycle and the virtual pressure of the cavity in the previous cycle, the physical pressure change rate of the cavity pressure in the current cycle is determined. If the physical pressure change rate is greater than the theoretical maximum pressure change rate in the current cycle, then a sudden fault is determined to have occurred. Furthermore, the consistency of the residual between the shadow pressure and the cavity virtual pressure determines whether a persistent drift fault has occurred; if either condition is met, the physical pressure sensor is confirmed to be faulty.

[0065] Furthermore, the compensation management module 340 is used to perform PID disturbance-free compensation in the current cycle during which the feedback source switch occurs: The integral state of the PID controller is reconstructed by back-calculating the integral term, while the historical input state of the derivative controller is replaced with the current shadow pressure, so that the change in the output of the PID controller to the valve before and after the switch is controlled within the preset allowable range.

[0066] Furthermore, the compensation management module 340 is used to reconstruct the integral state of the PID controller through inverse calculation of the integral term: I_new=CV_old-Kp×e_new(k)-D_hold-FF–Bias; Where I_new is the reconstructed integral term, CV_old is the PID control output of the previous cycle before switching, Kp is the proportional gain, e_new(k) is the control error after switching, D_hold is the historical hold value of the derivative term, FF is the feedforward compensation term, and Bias is the fixed bias value.

[0067] Furthermore, the compensation management module 340 for the pressure fault-tolerant control method also includes superimposing a rate-of-change limit on the output side of the PID controller: CV_out(k) = CV_last(k-1) + Clamp(CV_raw(k) - CV_last(k-1),-CV_rate_max, +CV_rate_max); Wherein, CV_out(k) is the valve opening output in the current cycle, CV_last(k-1) is the valve opening output in the previous cycle, CV_raw(k) is the unrestricted raw control target, CV_rate_max is the allowed rate of change of physical pressure for each cycle, and Clamp is the limiting function.

[0068] This application provides a pressure fault-tolerant control device for a semiconductor process cavity. The pressure fault-tolerant control device includes: a data processing module, used to synchronously acquire process parameters of the semiconductor process cavity in each fixed cycle, apply a first-order low-pass filter to the physical pressure sensor signal, and output the filtered pressure; a pressure processing module, used to determine the cavity virtual pressure in each fixed cycle based on a virtual pressure model of the mass conservation law, and continuously run a shadow pressure online alignment algorithm based on the filtered pressure and the cavity virtual pressure during the period when the physical pressure sensor is determined to be in a healthy state, to generate shadow pressure; a dual-criteria fault diagnosis module, used to jointly diagnose whether the physical pressure sensor has failed using dual criteria, and if a failure occurs and all gating conditions are met, switch the feedback source of the pressure control loop from the filtered pressure to the shadow pressure; and a compensation management module, used to perform PID disturbance-free compensation in the current cycle in which the feedback source switch occurs, perform differentiated degradation strategies according to the sensitivity of the current process step, continuously monitor operating conditions, and terminate degradation and perform controlled shutdown if any condition exceeds the limit. By implementing online alignment of shadow pressure during the health period, dual-criteria fault diagnosis, gating feedback source switching, and PID non-disruptive compensation, stable online fault-tolerant control is achieved when the pressure sensor in the semiconductor process cavity fails. This maximizes production line uptime while ensuring process safety and wafer quality, and reduces process interruptions and yield losses caused by sensor failures.

[0069] Please see Figure 4 , Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 4 As shown, the electronic device 400 includes a processor 410, a memory 420, and a bus 430.

[0070] The memory 420 stores machine-readable instructions executable by the processor 410. When the electronic device 400 is running, the processor 410 communicates with the memory 420 via the bus 430. When the machine-readable instructions are executed by the processor 410, they can perform the operations described above. Figure 1 The steps of the pressure fault-tolerant control method for the semiconductor process cavity in the method embodiment shown are described in detail in the method embodiment, and will not be repeated here.

[0071] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, can perform the above-described actions. Figure 1 The steps of the pressure fault-tolerant control method for the semiconductor process cavity in the method embodiment shown are described in detail in the method embodiment, and will not be repeated here.

[0072] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0073] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the shown or discussed mutual couplings, direct couplings, or communication connections may be through some communication interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0074] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0075] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0076] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a processor-executable, non-volatile, computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0077] Finally, it should be noted that the above-described embodiments are merely specific implementations of this application, used to illustrate the technical solutions of this application, and not to limit them. The scope of protection of this application is not limited thereto. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features, within the scope of the technology disclosed in this application. Such modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A pressure-tolerant fault control method for a semiconductor process cavity, characterized in that, The pressure fault-tolerant control method includes: The process parameters of the semiconductor process cavity are synchronously acquired in each fixed cycle, and the physical pressure sensor signal is subjected to a first-order low-pass filter to output the filtered pressure. The virtual pressure model based on the law of conservation of mass determines the virtual pressure of the cavity in each fixed period. During the period when the physical pressure sensor is determined to be in a healthy state, the shadow pressure online alignment algorithm is continuously run based on the filtered pressure and the virtual pressure of the cavity to generate shadow pressure. A dual-criteria joint diagnosis is used to determine whether the physical pressure sensor is faulty. If a fault occurs and all gating conditions are met, the feedback source of the pressure control loop is switched from the filtered pressure to the shadow pressure. During the current cycle in which the feedback source switch occurs, PID non-disruptive compensation is performed. Differentiated degradation strategies are implemented based on the sensitivity of the current process step. Operating conditions are continuously monitored. If any condition exceeds the limit, degradation is terminated and controlled shutdown is performed.

2. The pressure fault-tolerant control method according to claim 1, characterized in that, The virtual pressure model based on the law of conservation of mass determines the virtual pressure of the cavity in each fixed period, including: Determine the total intake molar flow rate into the cavity during the current cycle; Based on the equivalent volume pumping speed of the exhaust system under the current state, the virtual pressure of the cavity in the previous cycle, the gas temperature in the cavity, and the gas constant, the total exhaust molar flow rate of the cavity in the current cycle is determined. The total intake molar flow rate, the total exhaust molar flow rate, the pre-calibrated cavity leakage molar flow rate, the effective cavity volume, the gas constant, the gas temperature inside the cavity, and the cavity virtual pressure of the previous cycle are used to determine the cavity virtual pressure of the current cycle.

3. The pressure fault-tolerant control method according to claim 1, characterized in that, The process of continuously running an online alignment algorithm for shadow pressure based on filtered pressure and the virtual pressure of the cavity during the period when the physical pressure sensor is determined to be in a healthy state, to generate shadow pressure, includes: The deviation of the current cycle is determined based on the filtered pressure, the cavity virtual pressure, the low-pass filter coefficient, and the deviation of the previous cycle. The virtual pressure of the cavity is biased and corrected based on the deviation of the current cycle to generate the shadow pressure, and the update of the deviation is frozen after the fault is confirmed.

4. The pressure fault-tolerant control method according to claim 1, characterized in that, The method of using dual criteria to jointly diagnose whether a physical pressure sensor is malfunctioning includes: Based on the virtual pressure of the cavity in the current cycle and the virtual pressure of the cavity in the previous cycle, the physical pressure change rate of the cavity pressure in the current cycle is determined. If the physical pressure change rate is greater than the theoretical maximum pressure change rate in the current cycle, then a sudden fault is determined to have occurred. Furthermore, the consistency of the residual between the shadow pressure and the cavity virtual pressure determines whether a persistent drift fault has occurred; if either condition is met, the physical pressure sensor is confirmed to be faulty.

5. The pressure fault-tolerant control method according to claim 1, characterized in that, The step of performing PID disturbance-free compensation in the current cycle in which the feedback source switch occurs includes: The integral state of the PID controller is reconstructed by back-calculating the integral term, while the historical input state of the derivative controller is replaced with the current shadow pressure, so that the change in the output of the PID controller to the valve before and after the switch is controlled within the preset allowable range.

6. The pressure fault-tolerant control method according to claim 5, characterized in that, The method of reconstructing the integral state of the PID controller through inverse calculation of the integral term includes: I_new=CV_old-Kp×e_new(k)-D_hold-FF–Bias; Where I_new is the reconstructed integral term, CV_old is the PID control output of the previous cycle before switching, Kp is the proportional gain, e_new(k) is the control error after switching, D_hold is the historical hold value of the derivative term, FF is the feedforward compensation term, and Bias is the fixed bias value.

7. The pressure fault-tolerant control method according to claim 5, characterized in that, The pressure-tolerant control method also includes superimposing a rate-of-change limit on the output side of the PID controller: CV_out(k) = CV_last(k-1) + Clamp(CV_raw(k) - CV_last(k-1),-CV_rate_max, +CV_rate_max); Wherein, CV_out(k) is the valve opening output in the current cycle, CV_last(k-1) is the valve opening output in the previous cycle, CV_raw(k) is the unrestricted raw control target, CV_rate_max is the allowed rate of change of physical pressure for each cycle, and Clamp is the limiting function.

8. A pressure-tolerant control device for a semiconductor process cavity, characterized in that, The pressure fault-tolerant control device includes: The data processing module is used to synchronously acquire the process parameters of the semiconductor process cavity in each fixed cycle, apply a first-order low-pass filter to the physical pressure sensor signal, and output the filtered pressure. The pressure processing module is used to determine the virtual pressure of the cavity in each fixed period based on the virtual pressure model based on the law of conservation of mass. During the period when the physical pressure sensor is determined to be in a healthy state, the shadow pressure online alignment algorithm is continuously run based on the filtered pressure and the virtual pressure of the cavity to generate shadow pressure. The dual-criteria fault diagnosis module is used to jointly diagnose whether the physical pressure sensor has failed using dual criteria. If a failure occurs and all gating conditions are met, the feedback source of the pressure control loop is switched from the filtered pressure to the shadow pressure. The compensation management module is used to perform PID non-disruptive compensation in the current cycle when the feedback source switch occurs, execute differentiated degradation strategies according to the sensitivity of the current process step, continuously monitor operating conditions, and terminate degradation and execute controlled shutdown if any condition exceeds the limit.

9. An electronic device, characterized in that, include: The device includes a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the electronic device is running, the processor communicates with the memory via the bus. The machine-readable instructions are executed by the processor to perform the steps of the pressure fault-tolerant control method for a semiconductor process cavity as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, performs the steps of the pressure fault-tolerant control method for a semiconductor process cavity as described in any one of claims 1 to 7.

Images