An ignition control method for a remote plasma source

Abstract

The present invention discloses a remote plasma source ignition control method within the technical field of remote plasma sources, comprising the following steps: S1: setting a gas current excitation threshold and introducing excitation gas; S2: determining whether an ignition signal is received, and if so, closing a relay and entering a constant current phase; S3: determining whether the system ignition is normal; if not, adjusting the ignition voltage using a self-learning ignition threshold calibration method; S4: increasing the output power, and adjusting the ignition voltage during this period using an ignition threshold calibration method; S5: entering a constant power phase when the output power reaches full load power; wherein the self-learning ignition threshold calibration method collects gas flow, gas pressure, and cavity temperature in real time, uses the LSF-BP model to predict the current load impedance value, and updates the ignition voltage threshold accordingly. This method dynamically adjusts the ignition threshold according to the operating conditions, effectively improving ignition stability and reducing equipment damage and maintenance costs.

Description

Technical Field

[0001] The present invention relates to the technical field of remote plasma sources, and in particular to an ignition control method of a remote plasma source. Background Art

[0002] Remote Plasma Source (RPS), as an efficient and controllable plasma generation technology, has broad applications in semiconductor manufacturing, material surface treatment, thin film deposition, and pollutant degradation. Its core advantage lies in separating the plasma generation zone from the process zone, thereby preventing direct bombardment of sensitive substrates by high-energy particles, thereby significantly improving process uniformity and device reliability. However, reliable ignition and stable maintenance of the plasma are key prerequisites for achieving these advantages and are also core challenges in technology research and development. Excessively low ignition voltages can easily result in ignition failures, while excessively high ignition voltages can easily damage the plasma chamber.

[0003] The remote plasma source generally starts by introducing argon or helium and applying high pressure for pre-ignition. Then, a clean gas is introduced into the chamber to ionize NF3 to generate atoms for chamber cleaning. The pressure in the chamber of existing CVD / PVD tools is relatively high. In a vacuum, air molecules, water, fluoride gas, and precursors diffused from the chamber can be adsorbed on the surface of the remote plasma source, resulting in poor ignition of the remote plasma source.

[0004] The ignition voltage of traditional remote plasma sources is typically based on a preset threshold. This can vary under different environmental conditions, such as gas pressure, gas flow, and chamber temperature. This can cause corresponding changes in load impedance and sudden changes in the required maintenance voltage for the ionization process, leading to low ignition success rates and even damage to the device. Existing technologies lack the ability to adaptively adjust the ignition voltage. In the event of poor ignition, the remote plasma source can only be replaced with a new reaction chamber, which increases costs. Summary of the Invention

[0005] In order to solve the problem that the ignition voltage of the traditional ignition method cannot be adaptively adjusted, the present application provides an ignition control method for a remote plasma source. Through a self-learning ignition threshold calibration method, the system can predict the load impedance value under different gas flow rates, gas types and cavity temperature changes, automatically adjust the ignition voltage threshold according to the impedance value, and accurately adjust the ignition voltage, thereby reducing the risk of equipment damage, greatly improving the ignition success rate and system stability, and significantly improving the adaptability and reliability of the equipment.

[0006] The present invention provides a method for controlling the ignition of a remote plasma source, comprising the following steps:

[0007] S1: Set the gas current excitation thresholdi min , pass the excitation gas, and calculate the default ignition voltage V based on the preset formula break ;

[0008] S2: Determine whether the ignition signal is received. If so, the relay is closed and the constant current stage is entered. The PWM drive unit is modulated by the current error.

[0009] S3: Determine whether the system ignition is normal. If it is not normal, adjust the ignition voltage through the self-learning ignition threshold calibration method and re-determine. If it is normal, proceed to step S4;

[0010] S4: increasing the output power, during which the ignition voltage is adjusted by the self-learning ignition threshold calibration method;

[0011] S5: When the output power reaches the full load power, it enters the constant power stage, and the output power is controlled by modulating the PWM drive unit through the power error;

[0012] Among them, the self-learning ignition threshold calibration method collects gas flow Q, gas pressure P, and cavity temperature T in real time, uses the LSF-BP model combined with least squares polynomial fitting and BP neural network to predict the load impedance value F, and updates the ignition voltage threshold accordingly.

[0013] The beneficial effect of the above embodiment is that the self-learning ignition threshold self-calibration technology monitors the ignition status in real time and utilizes the LSF-BP load impedance prediction model to iteratively determine the optimal weight λ and output the predicted load impedance value F, enabling the device to dynamically adjust the ignition threshold based on operating conditions. This method can effectively improve ignition stability, reduce equipment damage, and reduce equipment maintenance costs.

[0014] Based on the above embodiments, the present application can be further improved as follows:

[0015] In one embodiment of the present application, in step S1, the resonant output current reference value is set before the ignition operation. Instantly drops to the current reference value By actively reducing the resonant output current reference value 50ms before ignition, the impact of inrush current on relay S is effectively suppressed, avoiding relay contact erosion caused by excessive instantaneous current, significantly extending the relay life, and ensuring the stability of the ignition circuit.

[0016] In one embodiment of the present application, in step S2, after the relay is energized, the resonant output current is sampled and determined to be within a preset range and maintained for a set time. If so, the system proceeds to step S3. If not, the system deems the ignition to have failed, reports an error, and disconnects the relay. By monitoring the current range and duration in real time, ignition anomalies (such as arc instability or gas ionization failure) can be quickly identified, allowing the relay to be disconnected and an error reported promptly. This prevents damage to internal components (such as electrodes or insulation materials) caused by sustained abnormal current flow, thereby reducing equipment maintenance costs.

[0017] In one embodiment of the present application, the constant current stage control is: the resonant output current reference value The current value sampled by the power circuit i pri Output current error after comparison , after current PI modulation, a duty cycle signal is generated. Using current closed-loop feedback control, the duty cycle is dynamically adjusted to maintain constant current output, ensuring current accuracy and stability during the ignition phase, avoiding current fluctuations caused by sudden changes in load impedance, and improving the success rate of plasma breakdown.

[0018] In one embodiment of the present application, the constant power stage control is: the output power reference value P ref With actual power P o Output power error after comparison e 0, after power PI modulation, the resonant output current reference value is obtained i ref , the current reference value i ref The current value sampled by the power circuit i pri Output current error after comparison e 1. Output duty cycle through current PI modulator d ; The current error e 1 The switching frequency is obtained by passing the PI modulator and inverting it f s After the plasma stabilizes, power closed-loop control is introduced to adaptively adjust the output power and switching frequency to effectively respond to real-time changes in load impedance (such as sudden changes in gas flow or pressure), ensure the stability of continuous plasma discharge, and avoid arc extinction problems caused by insufficient power.

[0019] In one embodiment of the present application, the LSF-BP model includes:

[0020] Input layer, receiving normalized data of Q, P, and T;

[0021] In the hidden layer, ReLU activation function is used for nonlinear transformation;

[0022] The output layer outputs the load impedance value F by linearly combining the least-squares fitting results with the neural network predictions. By combining the linear fitting capabilities of the least-squares method with the nonlinear feature learning capabilities of the BP neural network, the model can simultaneously capture the complex relationship between load impedance and gas parameters (Q / P / T), significantly improving the accuracy of impedance prediction and thus optimizing the real-time and accuracy of ignition voltage adjustment.

[0023] In one embodiment of the present application, the LSF-BP model adjusts the weights of the least squares fitting result and the neural network prediction value through the back propagation algorithm until the total error E is no greater than the set threshold, so that the fitting result is close to t The system dynamically optimizes model weights based on the actual value at each moment, enabling adaptive adjustment of control parameters. This allows predictions to quickly adapt to long-term drift in environmental parameters (such as cavity aging or changes in gas composition), ensuring high-precision predictions throughout the system's lifecycle and reducing the need for manual calibration.

[0024] In one embodiment of the present application, the ignition control method is implemented based on a remote plasma source power circuit, and the remote plasma source power circuit includes:

[0025] An adjustable turns ratio coupling transformer, the secondary coils of which are respectively connected to the ignition circuit and the holding circuit;

[0026] A sampling unit, collecting the output voltage and current of the resonant converter and the ignition circuit current;

[0027] The control unit dynamically adjusts the turns ratio of the adjustable turns ratio coupling transformer based on the predicted load impedance. By adjusting the turns ratio of the transformer's secondary winding, the voltage / current ratio between the ignition circuit and the holding circuit is directly changed, achieving impedance matching at the hardware level. This solves the problem of traditional fixed-parameter transformers being unable to adapt to sudden load changes and reduces energy loss.

[0028] In one embodiment of the present application, the sustaining circuit maintains a stable bus voltage through constant power control after the relay is disconnected, ensuring continuous operation during the plasma ionization phase. After the plasma enters a steady state, the sustaining circuit stabilizes the bus voltage through closed-loop control, preventing voltage drops caused by power outages or sudden load changes. This ensures the energy supply required for continuous plasma ionization, improving the uniformity and consistency of processes such as thin film deposition. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly describes the drawings required for the specific embodiments or the description of the prior art. Similar elements or parts are generally identified by similar reference numerals throughout the drawings. Elements or parts in the drawings are not necessarily drawn to scale.

[0030] Figure 1 This is a flowchart of the steps of a remote plasma source ignition control method in an embodiment of the present application;

[0031] Figure 2 This is a control block diagram of an ignition control method for a remote plasma source in an embodiment of the present application;

[0032] Figure 3 This is a control block diagram of the self-learning ignition threshold calibration method in an embodiment of the present application;

[0033] Figure 4 This is a flow chart of least squares polynomial fitting of load impedance in an embodiment of the present application;

[0034] Figure 5 This is a flow chart of the LSF-BP load impedance prediction model algorithm in the embodiment of the present application;

[0035] Figure 6 This is a network structure diagram of each layer of the LSF-BP load impedance prediction model in an embodiment of the present application. DETAILED DESCRIPTION

[0036] The present invention will be further explained below in conjunction with specific embodiments. It should be understood that these embodiments are only used to illustrate the present invention and are not used to limit the scope of the present invention. After reading the present invention, modifications of various equivalent forms of the present invention made by those skilled in the art all fall within the scope defined by the claims attached to this application.

[0037] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, it does not need to be further defined or explained in subsequent drawings.

[0038] In the description of the present invention, it should be noted that, unless otherwise expressly specified or limited, the terms "disposed," "installed," "connected," and "connected" should be understood broadly. For example, they may refer to mechanical or electrical connections; direct or indirect connections through an intermediary; and internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention.

[0039] In the description of the present invention, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. In addition, those skilled in the art may combine and combine different embodiments or examples described in the present invention, as well as features of different embodiments or examples, without any contradiction.

[0040] The embodiments of the present application provide an ignition control method for a remote plasma source to solve the problem that the ignition voltage of the traditional ignition method cannot be adaptively adjusted. It realizes automatic adjustment of the ignition voltage threshold according to the impedance value, accurately adjusts the ignition voltage, reduces the risk of equipment damage, and greatly improves the ignition success rate and system stability.

[0041] The technical solution in the embodiments of the present application is to solve the above problems, and the overall idea is as follows:

[0042] Example:

[0043] like Figure 1 As shown, a remote plasma source ignition control method includes the following steps:

[0044] Step S1:

[0045] After the remote plasma source is turned on, the control system first sets the gas current excitation threshold i min , and introduce the excitation gas, and calculate the default ignition voltage according to the following formula (14) V break The phase adjustment unit is used to adjust the resonant output current reference value 50ms before the ignition operation. Instantly drops to the current reference value , which can effectively avoid surge current and protect relay S.

[0046] Step S2:

[0047] Determine whether the ignition signal is received. If the ignition signal is received, the drive output is closed by relay S. Otherwise, the system continues to pass the excitation gas. After relay S is closed, it enters the constant current stage, that is, the resonant output current reference value The current value sampled by the power circuit i pri Output current error after comparison After the current PI modulation, the PWM drive unit is started. Then it is determined whether i min ≤ i pri ≤ i maxAnd it is maintained for more than 500ms. If it is not satisfied, it is considered that the ignition has failed, the system reports an error and disconnects the relay S. If it is satisfied, jump to step S3.

[0048] Step S3:

[0049] Determine whether the system ignition is normal; if the ignition is abnormal, adjust the ignition voltage through the self-learning ignition threshold calibration method, and use the LSF-BP load impedance prediction model to iterate the optimal weight , and output the predicted value of the load impedance , and then update the self-calibrated ignition voltage according to the following formula (17): V break , ignite, and then judge whether the ignition is normal. If it is not normal, perform cyclic ignition processing. The upper limit of the number of cycles is N times to improve the ignition success rate; if the ignition is still not successful after exceeding the cycle upper limit, it is determined that the ignition failed and the system reports an error and disconnects the relay S; during step S3, when it is determined that the ignition is normal, it is considered that the ignition is successful, the relay is disconnected and jumps to step S4.

[0050] To judge whether the ignition is normal, mainly by judging the resonant current value i pri Is the actual value greater than or equal to the set ignition current threshold, such as 30A. The set ignition current threshold is generally located at i min and i max between.

[0051] Step S4:

[0052] To increase the output power, the ignition voltage is adjusted by the self-learning ignition threshold calibration method. In actual working conditions, after the plasma is successfully ignited, the gas flow rate needs to be increased or the gas pressure needs to be changed to increase the load impedance to increase the output power of the power supply. This results in a sudden increase in the load and a sudden change in the plasma impedance. V o Also increases accordingly to maintain discharge, that is, the ignition voltage V break It also needs to be improved, otherwise the plasma ignition will be extinguished; this embodiment uses a self-learning ignition threshold calibration method to predict the load impedance in real time, and adjusts the plasma ignition voltage according to the load impedance, thereby ensuring the reliability of the remote plasma source operation.

[0053] Step S5:

[0054] The current system is still in the constant current stage, and the output power of the power supply is continuously increased. When the full load power is reached, the remote plasma source enters the constant power stage control. The constant power stage control includes: setting the output power reference valueP ref With resonant output voltage V loop and current i pri The actual power obtained after multiplication P o Output power error after comparison e 0, after power PI modulation, the resonant output current reference value is obtained i ref , the current reference value i ref The current value sampled by the power circuit i pri Output current error after comparison e 1. Output duty cycle through current PI modulator d ; The current error e 1 The switching frequency is obtained by passing the PI modulator and inverting it f s Finally, the PWM drive unit outputs the PWM drive signal of the inverter switch tube D 1~ D 4; thus achieving reliable operation of the remote plasma source.

[0055] The self-learning ignition control method of the remote plasma source is implemented based on the remote plasma source power circuit and the corresponding control unit.

[0056] Figure 2 The control block diagram of the above remote plasma source learning ignition voltage control method is shown, including a remote plasma source power circuit, a constant current stage control unit, and a constant power stage control unit;

[0057] The remote plasma source power circuit includes a control unit (not shown in the figure), a PWM drive unit (not shown in the figure), a DC / AC resonant converter, an ignition circuit, a maintenance circuit, a sampling unit, and a reaction chamber.

[0058] The control unit (including the constant current stage control unit and the constant power stage control unit) is used to control the PWM drive unit to output a pulse signal; the PWM drive unit controls the output current of the resonant converter; the output current of the resonant converter affects the input current of the ignition circuit; the resonant converter converts the pulse signal output by the PWM drive unit into an AC source, and is used to provide an input AC source for the ignition circuit; the ignition circuit generates a high-voltage ignition signal to cause the plasma load to oscillate and ionize at high frequency, and the resonant converter is used to provide an input AC source. The resonant converter includes an adjustable turns ratio coupling transformer T1, and the ignition circuit is connected to the secondary coil N3 of the adjustable turns ratio coupling transformer T1; the maintenance circuit is connected in series with the secondary coil N2 of the adjustable turns ratio coupling transformer T, and the vacuum reaction chamber is used as the secondary side of the high-frequency transformer to only provide the chamber. i o To meet the low-voltage energy input in the ionization maintenance stage; the secondary coils N2 and N3 can both modulate the turns ratio;

[0059] The holding circuit is used to maintain the bus voltage stability after the constant current stage (ignition stage) enters the constant power stage (ionization stage) when the relay is disconnected;

[0060] The sampling unit is used to collect the ignition current of the ignition circuit, and the output voltage and output current of the resonant converter; the relay S is controlled by the control unit to be closed or opened.

[0061] Figure 3 The control block diagram of the self-learning ignition threshold calibration method in the remote plasma source self-learning ignition control method is shown, including: the aforementioned remote plasma source power circuit, multi-sensor module, LSF-BP load impedance prediction model, and ignition voltage adaptive update module;

[0062] The multi-sensor module includes a gas flow meter, a pressure sensor and a temperature sensor to monitor the gas flow in the reaction chamber respectively. Q , gas pressure P and reaction chamber temperature T ;

[0063] The LSF-BP load impedance prediction model is used to iterate the optimal weights , and output the predicted value of the load impedance The ignition voltage adaptive update module calculates the ignition voltage required by the system at the current moment using the following equation (17). The adjustable turns ratio coupling transformer T1 performs hardware adaptive adjustment based on the LSF-BP load impedance prediction model and the ignition voltage adaptive update module to calculate the ignition voltage and holding voltage required by the system at the current moment.

[0064] The LSF-BP load impedance prediction model is further explained as follows:

[0065] First, the least squares method is used to fit the output load impedance. Figure 4 A flowchart showing the least squares polynomial fitting of load impedance is shown; combined with Figure 4 , the least squares polynomial fitting LSF algorithm is as follows:

[0066] In remote plasma systems, gas flow Q , gas pressure P and reaction chamber temperature T The effect on load impedance can be described by the following ideal gas model:

[0067] When the pumping speed S When fixed, pressure P Proportional to traffic P = Q / S , the load impedance Z is proportional to the flow rate and the temperature T Inversely proportional to gas flow Q Load impedance Z The influence of can be expressed by formula (1):

[0068] (1);

[0069] in k Q is a proportional constant related to system parameters such as electrode area, collision cross section, etc.

[0070] At constant temperature, the load impedance is proportional to the pressure: (2);

[0071] in k P is a proportionality constant related to the plasma density and collision frequency.

[0072] At constant pressure, the load impedance decreases inversely with the reaction chamber temperature due to the decrease in gas density: (3);

[0073] in k T is the proportionality constant.

[0074] P1: Define the linear model;

[0075] Combining equations (1), (2) and (3), the load impedance can be fitted using a linear regression model: Z and Q 、 P 、 T The approximate linear relationship is:

[0076] (4);

[0077] in, k 1, k 2, k 3 is the variable coefficient, k 0 is a constant term; assuming there is n Group observation data ( Q i , P i , t i , Z i ),in ; Use the least squares method to fit the variable coefficients and constant terms.

[0078] P2: Construct data matrix and observation vector;

[0079] Construction Design Matrix X And the observation vector Z:

[0080] (5);

[0081] P3: Establishing the standardized equation;

[0082] The goal of the least squares method is to minimize the residual sum of squares: (6);

[0083] The variable vector to be sought Taking the partial derivative and setting it to zero gives the normalized equation: (7);

[0084] P4: solve the fitting coefficient;

[0085] The optimal solution for the parameters is: (8);

[0086] P5: Output the current load impedance prediction value y 1;

[0087] Substituting the optimal solution of the parameters into equation (4) we can get the current load impedance prediction value, which is expressed as y 1.

[0088] The BP neural network is a multi-layer feedforward network. It consists of three layers: input layer, hidden layer, and output layer. During implementation, data is transferred from the input layer to the output layer. The hidden layer primarily performs a nonlinear transformation of the input data into a feature matrix. The output layer then calculates the error between the input and expected data and propagates the error forward from the output layer. The connection weights between each layer and the neuron biases in each layer are adjusted through backpropagation, optimizing the entire neural network and achieving the goal of learning data features. Its workflow is divided into two parts: the forward propagation of the system and the backward propagation of the error. The forward propagation process mainly involves receiving information, processing information, and outputting information. When the output differs significantly from the expected value, the system begins to propagate the error backwards, and the weights are changed and adjusted based on the actual error. Training of the system ends when the output meets the expected value or the set number of iterations is reached.

[0089] An adaptive algorithm was constructed: A prediction model was constructed using a least squares polynomial fitting (LSF) and a BP neural network to adjust the ignition threshold. The final ignition voltage threshold prediction was a weighted combination of the two prediction results. The BP neural network's self-learning capabilities were used to adjust the weights of the value function online, preprocessing to obtain the most recent weight control parameters. To address the overfitting and underfitting errors inherent in the least squares method, an LSF-BP load impedance prediction model was used for optimization.

[0090] Figure 5 The figure shows an algorithm flow chart of the LSF-BP load impedance prediction model of the self-learning ignition threshold calibration method based on load impedance prediction provided by this embodiment; Figure 6 The network structure diagram of each layer of the LSF-BP load impedance prediction model provided by this embodiment is shown; Figure 5 、 Figure 6 , a self-learning ignition threshold calibration method, comprising the following steps:

[0091] Step N1: First determine the structure of the BP network, which includes an input layer, a hidden layer and an output layer; i represents the input layer node, j represents the hidden layer nodes, l Represents the output layer node and determines the number of input nodes m and the number of hidden layer nodes q , the number of output layer nodes is 3; and the initial values of the weighted coefficients from the input layer to the hidden layer and from the hidden layer to the output layer are initialized 、 and and the offset;

[0092] The calculation formula for each layer is: (9);

[0093] in, T ( x ) represents the activation function, represents the weight coefficient, b represents the activation threshold;

[0094] Step N2: Construct a sample data set based on the test data: Use multiple sensors to measure the gas flow rate at the remote plasma source. Q , gas pressure P and cavity temperature T Construct a sample dataset {( x , y , z )},in, x Indicates gas flow Q One-dimensional set vector of , y Indicates gas pressure P One-dimensional set vector of , z Indicates cavity temperature T One-dimensional set vector of ;

[0095] Step N3: Submit the sample data set {( x , y , z )} x 、 y and z Form a two-dimensional input matrix, then perform data normalization operation on the two-dimensional input matrix and update it; at the same time, establish an output sample data set f {( x , y , z )} f A label vector corresponding to the network estimate of the load impedance;

[0096] For the sample data set {( x , y , z )} Normalization of all eigenvalues: Normalization is to convert the eigenvalue sequence of each resonant current into In the interval, the calculation formula is formula (10);

[0097] (10);

[0098] Among them, in an input eigenvalue sequence, M i is the normalized input feature value, N i is the input feature value before normalization, N max is the maximum input eigenvalue, Nmin is the minimum input eigenvalue.

[0099] Step N4: Use the two-dimensional input matrix updated by the normalization operation to train the BP model and output the load impedance f Network Estimation , and initialize the weight set vector on the two-dimensional output node of the BP neural network model , the bias of each layer of neurons in the BP neural network model, and then the label vector f and the network's estimated output Perform error calculations;

[0100] For multi-layer networks, the feedforward propagation method is used for calculation, that is, each layer is calculated according to formula (9) until the last output layer;

[0101] In the forward propagation process, the input data is calculated by the perceptron node and then activated by the activation function. T ( x ) to obtain the output result; the activation function of the hidden layer is selected T ( x ) is the ReLU function, the activation function of the BP neural network l :

[0102] (11);

[0103] The forward propagation output is for: (12);

[0104] Among them, W ij represents the weight coefficient connecting the i-th neuron in the previous layer and the j-th neuron in the current layer, and b represents the bias of the j-th neuron in the current layer.

[0105] Step N5: Construct an LSF-BP load impedance prediction model based on weight update combination, which is expressed as follows:

[0106] (13);

[0107] In formula (13), 、 λ 1 and λ 2 are the predicted load impedance value, LSF weight and BP neural network weight of the combined model respectively. y 2 is the forward propagation output value ;

[0108] Step N6: Calculate the overall error for the entire sample set E , and judge the error E Whether it meets the design requirements, that is, the error , N is the error threshold, where E Calculated by formula (13); if satisfied, the neural network training is terminated; if not satisfied, the neural network learning is performed using back propagation, the output result is compared with the expected structure, and the weights are updated online through multiple iterations. λ 1 and λ 2, so that the fitting results are close to t Moment f ( k ) value, thereby achieving adaptive adjustment of control parameters;

[0109] (13);

[0110] Step N7: Use the LSF-BP load impedance prediction model to iterate and obtain the optimal weight , and output the predicted load impedance value .

[0111] The ignition voltage adaptive update module is further described as follows:

[0112] During remote plasma ignition, the load impedance Z With ignition voltage V ignition The relationship between the breakdown and the plasma can be divided into two stages: the pre-breakdown stage and the post-breakdown stage. The pre-breakdown stage is the ignition critical stage, and the post-breakdown stage is the plasma steady-state stage. The specific formula is as follows:

[0113] 1. The gas is not ionized, in the pre-breakdown stage;

[0114] When the gas is not ionized, the system is approximately open circuit and exhibits high impedance. A sufficient breakdown voltage must be applied to trigger ionization. The breakdown voltage is described by Paschen's law:

[0115] (14);

[0116] p is the gas pressure; d Between electrodes; A , B is a constant related to the gas type (for example, argon: A ≈13.6cm − 1 Torr −1 , B ≈176V·cm −1 Torr −1 ); γ is the secondary electron emission coefficient.

[0117] At this time, the load impedance is extremely large, which is approximately an open circuit state: (15);

[0118] The voltage needs to reach V break To trigger discharge, impedance and voltage have no direct proportional relationship, only the breakdown condition needs to be met.

[0119] 2. Plasma formation, post-breakdown stage;

[0120] Once the gas is ionized to form plasma, the impedance drops sharply and enters a steady state. At this time, the relationship between the load impedance and the voltage follows Ohm's law: (16);

[0121] Among them exists The relationship between, K is the empirical coefficient, , V o To maintain voltage; I o To maintain the current; Z o It is the plasma maintenance resistance; when the load increases suddenly, such as changes in gas flow, pressure or temperature, the plasma impedance increases and the maintenance voltage needs to be increased. V o To maintain discharge, that is, ignition voltage V break It also needs to be improved;

[0122] The relationship between plasma impedance and voltage is affected by the electron density n e , conductivity σ e , electron mobility μ e The impact can be further expressed as: (17);

[0123] Where, the load impedance Z 0 is the predicted load impedance value output by the aforementioned LSF-BP load impedance prediction model .

[0124] For low-pressure plasma, the conductivity σ e It can be approximated as: (18);

[0125] in ν en is the electron-neutral particle collision frequency, e is the electron charge, m e The mass of the electron.

[0126] The above-mentioned self-learning ignition threshold calibration method based on load impedance prediction uses self-learning ignition threshold self-calibration technology. The system can predict the load impedance value under different gas flow rates, gas types and cavity temperature changes, automatically adjust the ignition voltage threshold according to the impedance value, and accurately adjust the ignition voltage, thereby reducing the risk of equipment damage, greatly improving the ignition success rate and system stability, and significantly improving the adaptability and reliability of the equipment.

[0127] Although the embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and are not to be construed as limitations on the present invention. A person skilled in the art may change, modify, replace and modify the above embodiments within the scope of the present invention.

Claims

1. A remote plasma source ignition control method, characterized in that: The following steps are involved: S1: Set the gas current excitation threshold, introduce excitation gas, and calculate the default ignition voltage based on the preset formula; S2: Determine whether the ignition signal is received. If so, the relay is closed and the constant current stage is entered. The PWM drive unit is modulated by the current error. S3: Determine whether the system ignition is normal. If it is not normal, adjust the ignition voltage through the self-learning ignition threshold calibration method and re-determine. If it is normal, proceed to step S4; S4: increasing the output power, during which the ignition voltage is adjusted by the self-learning ignition threshold calibration method; S5: When the output power reaches the full load power, it enters the constant power stage, and the output power is controlled by modulating the PWM drive unit through the power error; Among them, the self-learning ignition threshold calibration method collects gas flow Q, gas pressure P, and cavity temperature T in real time, uses the LSF-BP model combined with least squares polynomial fitting and BP neural network to predict the load impedance value F, and updates the ignition voltage threshold accordingly.

2. The ignition control method according to claim 1, wherein: In step S1 , the resonant output current reference value is actively reduced 50 ms before the ignition operation.

3. The ignition control method according to claim 1, wherein: In step S2, after the relay is energized, the resonant output current needs to be collected and judged whether it is within the preset range and maintained for the set time. If so, step S3 is entered; if not, it is considered that the ignition has failed, the system reports an error and disconnects the relay.

4. The ignition control method according to claim 1, wherein: The constant current stage control is: the resonant output current reference value The current value sampled by the power circuit i pri Output current error after comparison , after current PI modulation, a duty cycle signal is generated.

5. The ignition control method according to claim 1, wherein: The constant power stage control is: the output power reference value P ref With actual power P o Output power error after comparison e 0, after power PI modulation, the resonant output current reference value is obtained i ref , the current reference value i ref The current value sampled by the power circuit i pri Output current error after comparison e 1. Output duty cycle through current PI modulator d ; The current error e 1 The switching frequency is obtained by passing the PI modulator and inverting it f s .

6. The ignition control method according to claim 1, wherein: The LSF-BP model includes: Input layer, receiving normalized data of Q, P, and T; In the hidden layer, ReLU activation function is used for nonlinear transformation; The output layer outputs the load impedance prediction value F by linearly combining the least squares fitting result and the neural network prediction value.

7. The ignition control method according to claim 6, characterized in that: The LSF-BP model adjusts the weights of the least squares fitting result and the neural network prediction value through the back propagation algorithm until the total error E is no greater than the set threshold.

8. The ignition control method according to claim 1, wherein: The ignition control method is implemented based on a remote plasma source power circuit, and the remote plasma source power circuit includes: An adjustable turns ratio coupling transformer, the secondary coils of which are respectively connected to the ignition circuit and the holding circuit; A sampling unit, collecting the output voltage and current of the resonant converter and the ignition circuit current; A control unit dynamically adjusts the turns ratio of the adjustable turns ratio coupling transformer according to the load impedance value F.

9. The ignition control method according to claim 8, characterized in that: The maintenance circuit maintains bus voltage stability through constant power control after the relay is disconnected.

Images