Remote plasma density homogenization control method based on microwave excitation

Abstract

The present invention relates to the field of plasma density control technology, and discloses a remote plasma density uniformity control method based on microwave excitation. The method uses microwave excitation equipment to obtain target area field strength detection signal data and target interval plasma density detection signal data to determine the target interval density distribution. A three-dimensional field strength model is constructed based on the field strength data, and the operation strategy of the microwave control device is determined in combination with the density distribution. The microwave control device obtains the target interval plasma state image data according to the strategy, analyzes the density uniformity, and obtains density control data. The method solves the problem of remote plasma density uniformity control through multi-dimensional data acquisition, three-dimensional model construction and image analysis-driven control logic. It is suitable for scenarios such as semiconductor manufacturing, material surface treatment, chemical vapor deposition, etc., and can improve the quality problems caused by density unevenness in related processes.

Description

Technical Field

[0001] The present invention relates to the technical field of plasma density control, and in particular to a remote plasma density homogenization control method based on microwave excitation. Background Art

[0002] In modern industrial production and scientific research, the application of plasma technology is becoming increasingly widespread. Many key links, such as semiconductor manufacturing, material surface treatment, and chemical vapor deposition, all rely on plasma to achieve specific process goals. In these application scenarios, the uniformity of plasma density plays a decisive role in the process effect. Taking semiconductor manufacturing as an example, if the plasma density is uneven, during the etching process, it may lead to inconsistent etching rates in different areas of the chip, thereby affecting the performance and yield of the chip; during material surface treatment, uneven plasma density will cause uneven surface modification of the material, reducing the overall quality of the material.

[0003] Currently commonly used plasma generation and control technologies have many limitations. Traditional radio frequency plasma systems, while capable of generating plasma to a certain extent, struggle to achieve high-precision density uniformity control. The plasma density distribution they produce is susceptible to interference from factors such as equipment structure, working gas type, and flow rate, resulting in poor plasma density uniformity over large processing areas. Early microwave plasma systems, while offering the advantage of generating high free radical densities, were plagued by frequency and power instabilities. This resulted in extremely unstable plasma excitation states during the production process, making uniformity control extremely difficult and severely limiting their application in areas where plasma uniformity is critical.

[0004] In the existing technology, some methods attempt to improve the uniformity of plasma density by adjusting the device structure or optimizing the operating parameters. For example, the shape and position of the microwave applicator are changed in the hope of obtaining a more uniform field distribution, thereby improving the uniformity of plasma density. However, due to the complexity of the microwave field distribution and the variable nature of the plasma itself, this method can often only achieve limited results and cannot fundamentally solve the uniformity problem. There are also some technologies that regulate plasma by introducing additional magnetic fields or electric fields, but these methods not only increase the complexity and cost of the equipment, but also in actual operation, the precise control of the magnetic field or electric field is extremely difficult, which can easily lead to new unstable factors, making it difficult to achieve the ideal uniformity of plasma density.

[0005] Against this backdrop, the development of a method for precise and efficient remote plasma density homogenization is urgently needed. The microwave-based remote plasma density homogenization method proposed in this paper addresses the shortcomings of existing technologies and meets the stringent requirements for plasma density uniformity in industrial production and scientific research. Summary of the Invention

[0006] The object of the present invention is to provide a remote plasma density homogenization control method based on microwave excitation to solve the problems raised in the above background technology.

[0007] To achieve the above object, the present invention provides a remote plasma density homogenization control method based on microwave excitation, the method comprising:

[0008] Acquire field intensity detection signal data of a target area and plasma density detection signal data of a target interval based on a microwave excitation device, and determine the density distribution of the target interval according to the plasma density detection signal data;

[0009] constructing a three-dimensional field intensity model of the target area based on the field intensity detection signal data, and determining an operating strategy of the microwave control device based on the three-dimensional field intensity model and density distribution;

[0010] The microwave control device obtains plasma state image data of the target interval according to the operation strategy, analyzes the plasma density uniformity of the target interval according to the image data, and obtains density control data.

[0011] Preferably, the microwave excitation device is used to obtain field intensity detection signal data of the target area and plasma density detection signal data of the target interval, and the density distribution of the target interval is determined according to the plasma density detection signal data, specifically:

[0012] Acquiring field intensity detection data of a target area and plasma density detection signal data of a target interval in the target area based on a microwave excitation device;

[0013] performing a denoising and enhancement operation on the plasma density detection signal data based on a sliding window filtering algorithm to obtain interference-suppressed plasma density detection signal data, introducing an empirical mode decomposition algorithm, and setting initial values ​​of the empirical mode decomposition algorithm parameters;

[0014] performing signal decomposition on the interference-suppressed plasma density detection signal data according to the empirical mode decomposition algorithm to obtain m eigenmode components;

[0015] Performing Fourier transform on the intrinsic mode components to construct a frequency domain signal of each intrinsic mode component, calculating the amplitude mean and frequency bandwidth of each intrinsic mode component based on the frequency domain signal, and constructing a frequency domain distribution matrix with the amplitude mean and frequency bandwidth;

[0016] Calculate the energy concentration, amplitude variance, total energy, and amplitude extreme value of each frequency band emitted by the microwave excitation device according to the frequency domain distribution matrix to obtain a density detection characteristic spectrum;

[0017] Acquiring sampling point position information of the density detection characteristic spectrum according to the plasma density detection signal data, and spatially mapping the density detection characteristic spectrum of each sampling point according to the sampling point position information to construct a three-dimensional density detection characteristic spectrum;

[0018] The density gradient of each sampling point is determined according to the three-dimensional density detection characteristic spectrum, the density distribution contour map and density gradient vector field of the target interval are constructed according to the density gradient of each sampling point, and the density distribution of the target interval is determined according to the density distribution contour map and density gradient vector field.

[0019] Preferably, the three-dimensional field intensity model of the target area is constructed according to the field intensity detection signal data, and the operation strategy of the microwave control device is determined according to the three-dimensional field intensity model and density distribution, specifically:

[0020] Performing multi-resolution edge detection on the field strength detection signal data, constructing a field strength feature tensor through grid processing, and performing medium type identification on the echo signal of the field strength detection signal data based on a decision tree classifier to generate a three-dimensional field strength model of the target area including field strength amplitude, edge features, and medium type labels;

[0021] Performing a spatial mapping operation on the density distribution and the three-dimensional field strength model to construct a three-dimensional spatial distribution model of plasma density, and dividing the three-dimensional spatial distribution model of plasma density according to a preset grid size to construct P subspace regions;

[0022] Obtaining density uniformity information of each subspace region according to the density distribution, and clustering subspace regions with similar density uniformity in the three-dimensional spatial distribution model of plasma density based on a hierarchical clustering algorithm to obtain a clustering result;

[0023] Determining the distribution probability of plasma in each subspace region according to the clustering result, obtaining corresponding position information of each subspace region in the target region, and performing density control importance evaluation on each position in the target interval of the target region according to the distribution probability and the corresponding position information to obtain a control importance score for each position in the target interval;

[0024] Determining the regulation requirement information of each position in the target interval according to the regulation importance score of each position, wherein the regulation requirement information includes whether to perform regulation and regulation intensity requirement information;

[0025] Acquiring parameter adjustment sensitivity data of the microwave control device for plasmas at different distances, determining parameter adjustment distance information of the microwave control device for each position in the target interval based on the parameter adjustment sensitivity data and control requirement information, and determining a parameter adjustment point set of the microwave control device based on the parameter adjustment distance information;

[0026] An operating strategy of the microwave control device is determined according to the parameter adjustment point set.

[0027] Preferably, the operation strategy of the microwave control device is determined according to the parameter adjustment point set, specifically:

[0028] Obtaining the three-dimensional coordinate information of each parameter adjustment point and the initial position information of the microwave control device, and determining the field intensity abnormality area in the target area according to the three-dimensional field intensity model;

[0029] Performing path planning on the initial position information and the three-dimensional coordinate information of each parameter adjustment point based on the Dijkstra algorithm, and using the abnormal field strength area as a path planning restriction area to output the shortest initial operation path of the microwave control device;

[0030] acquiring in real time state change data of the microwave control device during operation of the microwave control device according to the shortest initial operation path, and determining the field intensity fluctuation direction and field intensity fluctuation amplitude at the real-time operating position of the microwave control device based on the state change data;

[0031] Acquiring parameter stability data of the microwave control device, wherein the parameter stability data includes data on the adaptability of the microwave control device to field intensity fluctuations of different directions and amplitudes;

[0032] Acquiring recognition response time data of the microwave control device to the direction and amplitude of the field intensity fluctuation, and determining a running path adjustment hysteresis of the microwave control device according to the recognition response time data;

[0033] Analyzing the field intensity fluctuation direction and field intensity fluctuation amplitude of the real-time operating position of the microwave control device according to the adaptability data and the operating path adjustment hysteresis, and determining the cumulative amount of operating path deviation of the microwave control device within the identification response time;

[0034] If the cumulative running path deviation is less than a preset threshold, determining a running deviation direction and a running deviation distance of the microwave control device according to the shortest initial running path based on the cumulative running path deviation, and determining an adjustment direction and an adjustment distance of the microwave control device based on the running deviation direction and the running deviation distance to obtain adjustment data;

[0035] Adjusting the shortest initial operation path of the microwave control device during real-time operation according to the adjustment data;

[0036] If the cumulative deviation of the running path is greater than a preset threshold, obtaining the field intensity fluctuation amplitude and field intensity fluctuation direction data of the real-time running path of the microwave control device, constructing a field intensity change graph based on the field intensity fluctuation amplitude and field intensity fluctuation direction data of the real-time running path, and determining the field intensity change trend in the target interval of the target area based on the field intensity change graph;

[0037] An interpolation operation is performed on the field intensity variation trend based on bilinear interpolation to determine field intensity information within a preset range of the shortest initial operating path. The operational stability of the microwave control device within the preset range of the shortest initial operating path is determined based on the field intensity information and adaptability data. Based on the operational stability, sections of the shortest initial operating path where the cumulative amount of operating path deviation is greater than a preset threshold are optimized to obtain an updated operating path. The microwave control device operates according to the updated operating path.

[0038] Preferably, the microwave control device obtains plasma state image data of the target interval according to the operation strategy, analyzes the plasma density uniformity of the target interval according to the image data, and obtains density control data, specifically:

[0039] Acquiring standard image data of different types of plasma in a target area, and annotating the standard image data for density uniformity to obtain annotated image data;

[0040] Building a density analysis model based on a recurrent neural network, and importing the labeled image data into the density analysis model for training;

[0041] The microwave control device obtains plasma state image data of the target interval according to the operation strategy, imports the plasma state image data into the trained density analysis model to perform density uniformity identification, and performs statistics on the density deviation value of each area to obtain density control data.

[0042] Preferably, the density control data is transmitted to the ground control terminal by constructing a layered data transmission protocol stack based on wireless communication technology, specifically:

[0043] Building a layered data transmission protocol stack based on wireless communication technology, encoding the density control data, and building a density control data transmission signal;

[0044] The density control data transmission signal is sent to a ground control terminal according to the layered data transmission protocol stack, and the density control data transmission signal is decoded to obtain plasma density control result data of the target interval.

[0045] Preferably, the denoising and enhancement operation is performed on the plasma density detection signal data based on a sliding window filtering algorithm, specifically:

[0046] The window size and sliding step of the sliding window are set, the plasma density detection signal data is traversed, the signal data in each window is mean filtered, the average value of the signal data in the window is calculated, and the average value is used as the denoised signal value at the center position of the window to obtain interference-suppressed plasma density detection signal data.

[0047] Preferably, the energy concentration, amplitude variance, total energy, and amplitude extreme value of each frequency band emitted by the microwave excitation device are calculated according to the frequency domain distribution matrix, specifically:

[0048] Traverse each frequency band interval of the frequency domain distribution matrix, calculate the sum of the energy values ​​of all frequency points in the interval as the total energy, calculate the square mean of the amplitude in the interval as the energy concentration, calculate the arithmetic mean of the amplitude in the interval as the amplitude mean, calculate the difference between the maximum and minimum values ​​of the amplitude in the interval as the amplitude extreme value, and obtain the density detection characteristic spectrum.

[0049] Preferably, the field intensity characteristic tensor is constructed by gridding processing, specifically:

[0050] The target area is divided into equally spaced grid points, the field strength detection signal data at each grid point is normalized, the field strength amplitude, gradient direction and phase information of each grid point are extracted, and the information is arranged according to the grid point position to construct a field strength feature tensor.

[0051] Preferably, the density analysis model is constructed based on a recurrent neural network, specifically:

[0052] The input layer, hidden layer and output layer of the recurrent neural network are set. The input layer receives the pixel features of the labeled image data, the hidden layer processes the time series features through long short-term memory units, and the output layer outputs the density uniformity classification results to build a density analysis model.

[0053] Compared with the prior art, the present invention has the following beneficial effects:

[0054] In the data acquisition and analysis phase, the field intensity detection signal data of the target area and the plasma density detection signal data of the target interval are obtained based on the microwave excitation equipment, and the density distribution of the target interval is determined based on the plasma density detection signal data. This process can accurately and comprehensively grasp the real-time status information of the plasma in the target area. Compared with traditional methods, traditional technologies may only be able to obtain partial data in a limited area, or the accuracy and real-time performance of the data are poor. This method uses advanced microwave excitation equipment to achieve all-round and high-precision data collection of the target area, providing a solid data foundation for subsequent control operations. For example, in the semiconductor etching process, the plasma density of different parts of the chip can be accurately known to avoid uneven etching problems caused by insufficient understanding of the plasma state.

[0055] In terms of model building and strategy determination, a three-dimensional field intensity model of the target area is constructed based on the field intensity detection signal data, and the operation strategy of the microwave control device is determined in combination with the density distribution. This three-dimensional field intensity model can intuitively and accurately present the distribution characteristics of the microwave field in the target area. Compared with the previous simple two-dimensional model or empirical field intensity judgment method, this model takes into account more spatial dimensions and actual influencing factors, greatly improving the accuracy of the description of the microwave field. By combining the three-dimensional field intensity model with the plasma density distribution, a more scientific and reasonable operation strategy for the microwave control device can be formulated. Taking the material surface treatment process as an example, the parameters of the microwave control device can be accurately adjusted according to the field intensity and plasma density in different areas, so that the plasma acts evenly on the material surface, effectively improving the consistency and quality of the material surface modification.

[0056] From the perspective of plasma density uniformity analysis and control, the microwave control device obtains plasma state image data of the target interval according to the operation strategy, and then conducts in-depth analysis of the plasma density uniformity in the target interval to obtain density control data. Compared with the traditional method of relying solely on numerical data for analysis, this image data-based analysis method can provide a more intuitive and comprehensive insight into the distribution details of the plasma and promptly discover areas and features with uneven density. Through precise density control data, the plasma density can be adjusted in a targeted manner, significantly improving the uniformity of the plasma density. In the chemical vapor deposition process, it can ensure that the plasma density is uniform within the deposition area, making the thin film deposition more uniform and dense, and improving the quality and performance of the film. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 This is a working principle diagram of the remote plasma density homogenization control method based on microwave excitation according to the present invention;

[0058] Figure 2 is a flow chart of the plasma density uniformity analysis method;

[0059] Figure 3 Flowchart of the sliding window filtering denoising method;

[0060] Figure 4 It is a flow chart of the frequency domain feature calculation method;

[0061] Figure 5 Flowchart of the method for building a density analysis model. DETAILED DESCRIPTION

[0062] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0063] See also Figure 1-Figure 5 The present invention provides a remote plasma density homogenization control method based on microwave excitation, the method comprising:

[0064] Acquire field intensity detection signal data of a target area and plasma density detection signal data of a target interval based on a microwave excitation device, and determine the density distribution of the target interval according to the plasma density detection signal data;

[0065] constructing a three-dimensional field intensity model of the target area based on the field intensity detection signal data, and determining an operating strategy of the microwave control device based on the three-dimensional field intensity model and density distribution;

[0066] The microwave control device obtains plasma state image data of the target interval according to the operation strategy, analyzes the plasma density uniformity of the target interval according to the image data, and obtains density control data.

[0067] Example 1:

[0068] The field intensity detection signal data of the target area and the plasma density detection signal data of the target interval are obtained based on the microwave excitation device, and the density distribution of the target interval is determined according to the plasma density detection signal data. The specific implementation process is as follows:

[0069] The microwave excitation device acquires field intensity data for the target area and plasma density data for the target interval within the target area. The microwave excitation device contains multiple detection modules that simultaneously collect signals for both the field intensity and plasma density within the target interval, ensuring that the acquired data fully reflects the relevant characteristics of the target area and interval.

[0070] The plasma density detection signal data was denoised and enhanced based on a sliding window filtering algorithm. The sliding window size was set to 5, with a sliding step size of 2. The plasma density detection signal data was traversed, and the signal data within each window was mean filtered. The average value of the signal data within the window was calculated, and the average value was used as the denoised signal value at the center of the window to obtain interference-suppressed plasma density detection signal data. An empirical mode decomposition algorithm was introduced, and the initial values ​​of the empirical mode decomposition algorithm parameters were set, with the maximum number of iterations set to 100 and the standard deviation threshold set to 0.2.

[0071] The interference-suppressed plasma density detection signal data is decomposed according to the empirical mode decomposition algorithm to obtain m eigenmode components. These eigenmode components can respectively reflect the characteristics of different frequency ranges in the original signal, which is helpful for subsequent in-depth analysis of the signal.

[0072] Perform Fourier transform on the intrinsic mode components to construct the frequency domain signal of each intrinsic mode component. Fourier transform can convert the time domain signal into a frequency domain signal, so that we can clearly observe the distribution of the signal at different frequencies. According to the frequency domain signal, the amplitude mean and frequency bandwidth of each intrinsic mode component are calculated, and the amplitude mean and frequency bandwidth are used to construct a frequency domain distribution matrix. The frequency domain distribution matrix is ​​a two-dimensional matrix, in which the rows represent different intrinsic mode components and the columns correspond to the amplitude mean and frequency bandwidth, respectively.

[0073] The energy concentration, amplitude variance, total energy, and amplitude extreme value of each frequency band emitted by the microwave excitation device are calculated based on the frequency domain distribution matrix. Each frequency band interval of the frequency domain distribution matrix is ​​traversed, and the sum of the energy values ​​of all frequency points in the interval is calculated as the total energy, the square mean of the amplitude in the interval is calculated as the energy concentration, the arithmetic mean of the amplitude in the interval is calculated as the amplitude mean, and the difference between the maximum and minimum amplitudes in the interval is calculated as the amplitude extreme value to obtain the density detection characteristic spectrum. The density detection characteristic spectrum integrates the key characteristic parameters of each frequency band and can provide rich information for subsequent density distribution analysis.

[0074] The sampling point location information of the density detection characteristic spectrum is obtained based on the plasma density detection signal data. This location information is acquired by the positioning module of the microwave excitation device and is accurate to the centimeter level. The density detection characteristic spectrum of each sampling point is spatially mapped based on the sampling point location information to construct a three-dimensional density detection characteristic spectrum. During the spatial mapping process, a three-dimensional rectangular coordinate system is established with a fixed point in the target interval as the coordinate origin, and the coordinates of each sampling point are associated with the corresponding density detection characteristic spectrum.

[0075] The density gradient of each sampling point is determined based on the three-dimensional density detection characteristic spectrum, and the density gradient is calculated by taking the partial derivative of the three-dimensional density detection characteristic spectrum in three directions of space. The density distribution contour map and density gradient vector field of the target interval are constructed based on the density gradient of each sampling point. In the density distribution contour map, the density difference between adjacent contour lines is equal, which can intuitively show the distribution of density in the target interval; the density gradient vector field represents the direction and speed of density change at each position through the direction and size of the vector. The density distribution of the target interval is determined based on the density distribution contour map and the density gradient vector field. For example, it can be judged which areas have higher density, which areas have lower density, and the locations of areas with more drastic density changes.

[0076] Through the above series of steps, starting from the raw detection signal data, through denoising, decomposition, transformation, calculation, and other processing, the plasma density distribution in the target range can be accurately determined, providing a solid foundation for the subsequent development of the microwave control device operation strategy. Throughout the entire process, each step is closely linked, and the results of the previous step provide the necessary data support for the subsequent step, ensuring the scientific and accurate determination of the entire density distribution process. At the same time, the various algorithms and processing methods used are selected and optimized based on the characteristics of the plasma density detection signal, effectively extracting useful information from the signal, eliminating interference, and improving the accuracy of the density distribution determination.

[0077] Example 2:

[0078] In a microwave plasma processing system, the specific implementation process of determining the operation strategy of the microwave control device based on the plasma density distribution and controlling the operation of the microwave control device based on the operation strategy is as follows:

[0079] 1. Determine the operation strategy of the microwave control device based on the plasma density distribution. The specific steps include:

[0080] Obtain the plasma density distribution by analyzing the density detection characteristic spectrum of each frequency band emitted by the microwave excitation device. The density detection characteristic spectrum contains multiple characteristic parameters, such as energy concentration, amplitude variance, total energy, and amplitude extremes.

[0081] An objective function is constructed based on the density detection characteristic spectrum, and the objective function is used to evaluate the uniformity of plasma density. The objective function is constructed based on characteristic parameters in the density detection characteristic spectrum, and these parameters are combined into a value that can reflect the uniformity of plasma density through a specific mathematical model.

[0082] The objective function was optimized to determine the optimal operating parameters of the microwave control device. This optimization process employed a genetic algorithm with the following parameters: population size of 100, number of iterations of 50, crossover probability of 0.8, and mutation probability of 0.1. In each iteration, individuals in the population were evaluated and selected based on the value of the objective function. New individuals were generated through crossover and mutation operations until the optimal solution was found or the maximum number of iterations was reached.

[0083] An operating strategy for the microwave control device is determined based on the optimal operating parameters. The operating strategy includes at least one of microwave power distribution, microwave frequency configuration, and microwave phase control. For example, based on the optimization results, the power output of the microwave generator is adjusted to achieve a more reasonable microwave power distribution in different regions; the microwave frequency configuration is changed to adapt to the characteristics of the plasma in different regions; and the microwave phase can also be controlled to achieve precise control of the plasma distribution.

[0084] 2. Controlling the operation of the microwave control device based on the operation strategy, specifically the following steps:

[0085] The operation strategy is converted into control instructions that can be executed by the microwave control device. This conversion process involves parsing and encoding the various parameters in the operation strategy so that they can be recognized and executed by the control system of the microwave control device.

[0086] The control instructions are sent to the control system of the microwave control device via a communication interface that uses the industrial standard Modbus protocol to ensure that the control instructions can be accurately and reliably transmitted to the microwave control device.

[0087] The microwave control device's control system receives the control instructions and adjusts the device's operating state based on them. The control system decodes and verifies the control instructions, then adjusts the operating parameters of components like the microwave generator, power splitter, and phase controller based on the instructions.

[0088] During the operation of the microwave control device, the plasma density distribution is monitored in real time. Multiple plasma density sensors arranged in the processing chamber continuously collect plasma density data and transmit this data to the control system.

[0089] The microwave control device's operating strategy is dynamically adjusted based on the real-time monitoring of the plasma density distribution. The control system analyzes the real-time density data and compares it with the target density distribution. If any deviation is found, the optimal operating parameters are recalculated and new control instructions are generated, achieving closed-loop control of the microwave control device.

[0090] By determining the microwave control device's operating strategy based on the plasma density distribution and controlling its operation based on this strategy, precise control of the microwave plasma processing system can be achieved. In practical applications, this process can automatically adjust the microwave control device's operating parameters based on varying processing requirements and plasma conditions, achieving a more uniform plasma density distribution and thus improving the effectiveness and quality of microwave plasma processing. The entire control process requires no human intervention and is highly automated and intelligent, adapting to complex and changing working environments and processing requirements.

[0091] Furthermore, this control process exhibits excellent adaptability and scalability. By adjusting the genetic algorithm's parameters and the construction of the objective function, it can adapt to different types of microwave plasma treatment systems and varying process requirements. Furthermore, with the continuous advancement of sensor and data analysis technologies, the accuracy and efficiency of real-time monitoring and dynamic adjustment will continue to improve, further enhancing the performance and stability of microwave plasma treatment systems. In actual operation, the microwave control device can rapidly adjust its operating state based on the real-time monitored plasma density distribution, ensuring that the plasma density distribution always remains within the ideal range. This dynamic adjustment mechanism not only improves treatment effectiveness but also extends the service life of the microwave control device and reduces operating costs.

[0092] In microwave plasma processing systems, determining the microwave control device's operating strategy based on the plasma density distribution and controlling its operation based on this strategy is a complex and sophisticated system engineering process. Through reasonable parameter settings, optimized algorithm design, and precise control execution, effective microwave plasma control can be achieved, providing reliable technical support for fields such as material surface treatment, semiconductor manufacturing, and plasma etching.

[0093] Example 3:

[0094] Obtain the 3D coordinates of each parameter adjustment point and the initial position of the microwave control device. The 3D coordinates of the parameter adjustment points are collected using a laser positioning system with an accuracy of ±0.5 mm. The initial position is preset to the origin of the 3D coordinate system (0,0,0). Using the 3D field intensity model, extract areas within the target area where the field intensity deviates by ±15% from the preset threshold and mark these areas as abnormal field intensity areas.

[0095] Path planning is performed based on the Dijkstra algorithm using the initial position information and the three-dimensional coordinates of each parameter adjustment point. The algorithm sets the node spacing to 10 mm, and defines the path cost function as the product of the spatial distance between adjacent nodes and the field strength stability coefficient. The field strength stability coefficient ranges from 0.8 to 1.2, with the coefficient increasing with greater field strength fluctuations. Areas with abnormal field strength are designated as no-go zones. The algorithm generates a shortest path tree by traversing all reachable nodes, then selects the shortest initial path from the tree, connecting the initial position with each parameter adjustment point.

[0096] Real-time data on the microwave control device's state changes during operation along the shortest initial path is collected, including movement speed, steering angle, and output power fluctuations. A triaxial accelerometer and gyroscope installed on the device collect state data every 10 milliseconds. The direction and amplitude of the field intensity fluctuation at the real-time operating position are calculated based on the changes in five consecutive sets of data. The direction of the field intensity fluctuation is determined using a vector synthesis algorithm, and the amplitude is calculated using the root mean square value.

[0097] Obtain parameter stability data for the microwave control device, obtained through preliminary testing. During the test, the device output parameter deviations were recorded under field intensity fluctuations ranging from 0-30° in different directions and 0-20% in amplitude. The ratio of the deviation to the fluctuation parameter was used as the adaptability coefficient, forming an adaptability data matrix. The time interval between the device identifying the field intensity fluctuation and initiating the adjustment mechanism was also recorded. The test was repeated 20 times, and the average value was taken as the recognition response time data, which was typically within the range of 50-150ms.

[0098] The hysteresis of the microwave control device's operation path adjustment is determined based on the identification response time data. The hysteresis calculation formula is:

[0099]

[0100] Wherein, L is the path adjustment hysteresis (unit: mm), t is the recognition response time (unit: s), v is the device operating speed (unit: mm / s), and θ is the angle between the field intensity fluctuation direction and the device operating direction (unit: rad).

[0101] Based on the adaptability data and the hysteresis for the operational path, the direction and amplitude of field intensity fluctuations at the real-time operational location are analyzed. The amplitude of field intensity fluctuations is compared with the adaptability coefficient to calculate the maximum allowable offset of the device under the current fluctuation conditions. This is then combined with the hysteresis to calculate the cumulative offset of the operational path. The cumulative offset is the sum of the hysteresis over three consecutive sampling periods.

[0102] If the cumulative path deviation is less than a preset threshold (set at 5mm), the direction and distance of the microwave control device's deviation are determined based on the cumulative path deviation. The deviation direction aligns with the direction of the field intensity fluctuation, and the deviation distance is 80% of the cumulative deviation. The servo control system drives the device in the opposite direction for compensatory movement. Adjustment data includes the stepper motor's rotation angle and operating time to ensure that the adjusted path deviation is within ±1mm.

[0103] If the cumulative deviation from the operating path exceeds a preset threshold, the amplitude and direction of the field intensity fluctuation along the microwave control device's real-time operating path are obtained. A 50 mm area on each side of the real-time path is divided into two sections using a 5 mm x 5 mm grid. The field intensity at each grid point is calculated using an interpolation algorithm, and a three-dimensional map of the field intensity variation is constructed based on this data. Flat areas with a field intensity gradient change rate of less than 0.5% / mm are extracted from the map and selected as candidate areas for path optimization.

[0104] The field intensity trend is interpolated using bilinear interpolation, with interpolation node spacing set to 2 mm, to generate a continuous field intensity distribution surface. Based on this surface data, the field intensity uniformity is calculated for each point within a preset range of the shortest initial operating path (typically 30 mm on either side of the path). Combined with the adaptability data, the operational stability score for each location is determined on a scale of 0-10, with higher scores indicating greater stability.

[0105] Based on the operational stability score, the shortest initial path segments with cumulative path deviations exceeding a preset threshold are optimized. Regions with scores above 8 are prioritized during the optimization process. New path segments are generated through piecewise curve fitting, with the curvature change rate between adjacent segments controlled to within 15° / mm. The optimized path segments are smoothly connected to the remaining path segments to form an updated path. The microwave control device performs operations by reading the coordinate point sequence of the new path.

[0106] Throughout the entire operation process, after each parameter adjustment point is completed, the field intensity and plasma density data at that point are recollected to verify the effectiveness of the path planning and serve as the basis for the next round of path adjustment. This dynamic adjustment mechanism ensures that the microwave control device can maintain stable operation in complex field intensity environments and accurately reach each parameter adjustment point to perform control operations.

[0107] Example 4:

[0108] The microwave control device obtains plasma state image data of the target interval according to the operation strategy, analyzes the plasma density uniformity of the target interval based on the image data, and obtains density control data. The specific implementation process is as follows:

[0109] Obtain standard image data for different types of plasma in the target area. A high-resolution charge-coupled device camera was used to capture images of three typical plasmas, argon, oxygen, and nitrogen, at different density states. 500 sets of images were collected for each type of plasma, each containing 1024×1024 pixels of grayscale information. Parameters such as light source intensity and exposure time were kept consistent during shooting to ensure uniform image brightness. Density uniformity was annotated for the standard image data. The annotation process used a combination of manual and automatic methods. First, a densitometer was used to measure the actual density value of each 10×10 pixel block in the image. The density value was then converted to a grayscale range of 0-255 to generate annotated image data. Areas with a density uniformity above 90% were marked green, areas between 60% and 90% were marked yellow, and areas below 60% were marked red.

[0110] A density analysis model was constructed based on a recurrent neural network. The model consists of an input layer, hidden layers, and an output layer. The input layer has 1024 neurons and receives pixel feature vectors from the annotated image data. These feature vectors are composed of the grayscale value of each pixel and the grayscale difference between the eight adjacent pixels. The hidden layers contain three layers of long-short-term memory units, with 512, 256, and 128 neurons, respectively. The first layer uses the tanh activation function, while the next two use the relu activation function. Dropout is used between layers to prevent overfitting, with a dropout rate of 0.3. The output layer has three neurons corresponding to three levels of density uniformity, and uses a softmax activation function to output classification probabilities. During model training, the annotated image data is split into training and validation sets in a 7:3 ratio. The Adam optimizer is used, with an initial learning rate of 0.001 that decays by 0.8 every 10 epochs. The cross-entropy loss is used as the loss function. Training is repeated for 100 epochs, with 32 images processed in each iteration.

[0111] The microwave control device acquires plasma state image data within the target interval according to its operating strategy. This strategy sets an image acquisition frequency of two frames per second, covering the entire target interval. Dynamic capture is achieved using a camera mounted on the microwave control device's robotic arm. The arm's motion trajectory is synchronized with the microwave transmission path, ensuring a one-to-one correspondence between the image acquisition position and the field intensity control position. The captured image data is transmitted in real time to the edge computing module, which performs image preprocessing. This includes removing blur caused by device vibration, correcting pixel offsets due to temperature changes, and enhancing image contrast through histogram equalization to preserve the edge features of the plasma plume.

[0112] Plasma state image data is imported into a trained density analysis model for density uniformity identification. The model analyzes the input image block by block using a 50×50 pixel sliding window. Each window generates a density uniformity prediction. Regions with a predicted probability above 0.8 are assigned a grade directly, while regions with a predicted probability below 0.8 are voted on based on the predictions of adjacent windows. After identification, the density deviation value for each region is calculated. The deviation value is calculated as the absolute difference between the actual density value and the target density value for that region, divided by the target density value. The target density value is set to 1.2×10^18m^-3 based on process requirements.

[0113] The density control data is obtained and stored in the form of a two-dimensional array. The array dimension is consistent with the image pixel size, and each element contains two parts of information: the density uniformity level and the density deviation value at that location. For areas with a green density uniformity level, if the deviation value is less than 5%, it is marked as no control is required; a deviation value of 5%-10% is marked as slight control. Areas with a yellow level are marked as moderate control regardless of the deviation value. Areas with a red level are marked as heavy control. At the same time, the data also includes the center coordinates, area size, and distance from the nearest microwave transmitting antenna of each area. This information is converted through the mapping relationship between image coordinates and space coordinates, and the mapping error is controlled within ±2mm.

[0114] After density control data is generated, it is transmitted via an internal bus to the microwave control unit's control unit. The control unit then adjusts the corresponding microwave emission parameters based on the control level and location information contained in the data. For example, for areas of severe control, the transmission power of the corresponding antenna is increased, with the power adjustment amplitude proportional to the density deviation value. For areas of moderate control, the antenna's phase angle is fine-tuned within a range of ±15°. For areas of slight control, only the antenna's operating frequency is optimized, with frequency fluctuations controlled within ±5 MHz. The entire analysis process, from image acquisition to application of control data, has a delay of less than 1 second, ensuring real-time response to changes in plasma density.

[0115] During continuous operation, the density analysis model is updated online every hour, using actual control effect data from the previous hour to retrain the model's output layer weights. The update process does not change the hidden layer parameters to maintain model stability. When the plasma type in the target interval changes, the corresponding standard image dataset is automatically switched by identifying the color characteristics of the plume in the image to ensure adaptability of the density analysis. For example, argon plasma appears bluish-purple, while oxygen plasma appears light green. The model distinguishes the type based on the grayscale ratio of the color channel.

[0116] Example 5:

[0117] A layered data transmission protocol stack is constructed based on wireless communication technology. The protocol stack comprises the physical layer, data link layer, network layer, and application layer. The physical layer utilizes ultra-wideband technology, operating in the 3.1-10.6 GHz frequency band. Quadrature Amplitude Modulation (QAM) is used for modulation, with 16QAM corresponding to a data rate of 200 Mbps and 64QAM to 400 Mbps, with dynamic switching based on channel quality. The data link layer employs time division multiple access, dividing one second into 50 time slots, each allocated to a different data source. In the event of a time slot conflict, a backoff algorithm is used to reselect a time slot, with a randomly generated backoff period between 10 and 100 ms. The network layer utilizes a distance vector routing protocol. Each node periodically sends routing update messages to its neighbors. These messages contain the node ID, hop count, and link quality metrics, which are calculated based on the packet reception rate. The application layer defines the data frame format. The frame header includes a frame sequence number, data type identifier, and a checksum. The maximum length of the frame data portion is 1024 bytes, and a stop flag is set at the end of the frame.

[0118] The density control data is encoded. The encoding process consists of two steps: data segmentation and error correction coding. The density control data is first split according to the application layer frame data length, and each data block is assigned a block number and length information. Low-density parity-check code is then used for error correction coding, with a code rate of 1 / 2 and a generator matrix dimension of 2048 × 1024. Real-time encoding is achieved using a hardware encoder, with encoding latency controlled within 5ms. The encoded data blocks are combined with the frame header and trailer to construct the density control data transmission signal.

[0119] Density-controlled data transmission signals are sent to the ground control terminal based on a layered data transmission protocol stack. The physical layer converts the transmission signal into a radio frequency signal and adjusts the transmit power through adaptive power control. The transmit power is set at 10dBm in an unobstructed environment and increases to 20dBm in obstructed conditions, with a power adjustment step of 2dBm. The data link layer performs frame synchronization and collision detection on the signal. The synchronization sequence uses a 16-bit pseudo-random code. Collision detection is implemented through a carrier sense multiple access mechanism. After a collision is detected, retransmission is performed after a random time interval. The network layer selects the optimal transmission path based on the routing protocol and automatically switches to a backup path when the link quality indicator of the primary path falls below 80%. Signals are transmitted via a directional antenna with an antenna gain of 15dBi and a beamwidth of 30°. A mechanical steering mechanism tracks the position of the ground control terminal with tracking accuracy within ±1°.

[0120] The ground control terminal receives the density control data transmission signal, which is first amplified and filtered by the RF front-end using a bandpass filter with a center frequency of 5.8 GHz and a bandwidth of 200 MHz. The RF signal is then converted to baseband, and the synchronization circuit extracts the synchronization sequence to achieve frame synchronization. The data link layer handles collision resolution and retransmission requests for the received signal. If reception fails three times in a row, a retransmission instruction containing the specific frame sequence number is sent to the transmitter. The network layer performs routing decapsulation on the data packet, removing routing information and reassembling the data blocks.

[0121] Decode the density-controlled data transmission signal. The decoding process first corrects transmission errors using a low-density parity-check code decoder, which uses a belief propagation algorithm with 50 iterations. The block number and length information are then removed from the data blocks, and the blocks are reassembled in their original order to recover the complete density-controlled data. The decoded data is verified for integrity using a checksum. Data that fails the checksum is marked as invalid, and the error location is recorded.

[0122] The plasma density control results for the target range are obtained. These data are stored as a two-dimensional array, with the array elements containing the sampling point coordinates, density uniformity level, and control parameter values. The control terminal imports this data into a visualization module, generating a density distribution heat map and a time-series curve of the control effect. Different colors in the heat map correspond to different density uniformity levels, and the time-series curve shows time on the horizontal axis and average density deviation on the vertical axis. Simultaneously, this data is transmitted to the central control system via an Ethernet interface for subsequent optimization of the control strategy.

[0123] During data transmission, transmission performance indicators, including transmission rate, packet loss rate, and latency, are monitored in real time. The transmission rate is calculated as the amount of data successfully received per unit time. The packet loss rate is the ratio of lost frames to the total number of frames sent. Latency is the time interval between signal transmission and reception. When the packet loss rate exceeds 5% or the latency exceeds 100ms, protocol stack parameter adjustments are triggered. The network layer recalculates routing, the data link layer shortens time slots, and the physical layer increases transmit power until performance indicators return to normal.

[0124] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0125] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A remote plasma density homogenization control method based on microwave excitation, characterized in that: The following steps are involved: Acquire field intensity detection signal data of a target area and plasma density detection signal data of a target interval based on a microwave excitation device, and determine the density distribution of the target interval according to the plasma density detection signal data; constructing a three-dimensional field intensity model of the target area based on the field intensity detection signal data, and determining an operating strategy of the microwave control device based on the three-dimensional field intensity model and density distribution; The microwave control device obtains plasma state image data of the target interval according to the operation strategy, analyzes the plasma density uniformity of the target interval according to the image data, and obtains density control data; The microwave excitation device is used to obtain field intensity detection signal data of the target area and plasma density detection signal data of the target interval, and the density distribution of the target interval is determined according to the plasma density detection signal data, specifically: Acquiring field intensity detection data of a target area and plasma density detection signal data of a target interval in the target area based on a microwave excitation device; performing a denoising and enhancement operation on the plasma density detection signal data based on a sliding window filtering algorithm to obtain interference-suppressed plasma density detection signal data, introducing an empirical mode decomposition algorithm, and setting initial values ​​of the empirical mode decomposition algorithm parameters; performing signal decomposition on the interference-suppressed plasma density detection signal data according to the empirical mode decomposition algorithm to obtain m eigenmode components; Performing Fourier transform on the intrinsic mode components to construct a frequency domain signal of each intrinsic mode component, calculating the amplitude mean and frequency bandwidth of each intrinsic mode component based on the frequency domain signal, and constructing a frequency domain distribution matrix with the amplitude mean and frequency bandwidth; Calculate the energy concentration, amplitude variance, total energy, and amplitude extreme value of each frequency band emitted by the microwave excitation device according to the frequency domain distribution matrix to obtain a density detection characteristic spectrum; Acquiring sampling point position information of the density detection characteristic spectrum according to the plasma density detection signal data, and spatially mapping the density detection characteristic spectrum of each sampling point according to the sampling point position information to construct a three-dimensional density detection characteristic spectrum; The density gradient of each sampling point is determined according to the three-dimensional density detection characteristic spectrum, the density distribution contour map and density gradient vector field of the target interval are constructed according to the density gradient of each sampling point, and the density distribution of the target interval is determined according to the density distribution contour map and density gradient vector field.

2. The method for remote plasma density homogenization based on microwave excitation according to claim 1, characterized in that: The three-dimensional field intensity model of the target area is constructed based on the field intensity detection signal data, and the operation strategy of the microwave control device is determined based on the three-dimensional field intensity model and density distribution. Specifically, Performing multi-resolution edge detection on the field strength detection signal data, constructing a field strength feature tensor through grid processing, and performing medium type identification on the echo signal of the field strength detection signal data based on a decision tree classifier to generate a three-dimensional field strength model of the target area including field strength amplitude, edge features, and medium type labels; Performing a spatial mapping operation on the density distribution and the three-dimensional field strength model to construct a three-dimensional spatial distribution model of plasma density, and dividing the three-dimensional spatial distribution model of plasma density according to a preset grid size to construct P subspace regions; Obtaining density uniformity information of each subspace region according to the density distribution, and clustering subspace regions with similar density uniformity in the three-dimensional spatial distribution model of plasma density based on a hierarchical clustering algorithm to obtain a clustering result; Determining the distribution probability of plasma in each subspace region according to the clustering result, obtaining corresponding position information of each subspace region in the target region, and performing density control importance evaluation on each position in the target interval of the target region according to the distribution probability and the corresponding position information to obtain a control importance score for each position in the target interval; Determining the regulation requirement information of each position in the target interval according to the regulation importance score of each position, wherein the regulation requirement information includes whether to perform regulation and regulation intensity requirement information; Acquiring parameter adjustment sensitivity data of the microwave control device for plasmas at different distances, determining parameter adjustment distance information of the microwave control device for each position in the target interval based on the parameter adjustment sensitivity data and control requirement information, and determining a parameter adjustment point set of the microwave control device based on the parameter adjustment distance information; An operating strategy of the microwave control device is determined according to the parameter adjustment point set.

3. The method for remote plasma density homogenization control based on microwave excitation according to claim 2, characterized in that: The operation strategy of the microwave control device is determined according to the parameter adjustment point set, specifically: Obtaining the three-dimensional coordinate information of each parameter adjustment point and the initial position information of the microwave control device, and determining the field intensity abnormality area in the target area according to the three-dimensional field intensity model; Performing path planning on the initial position information and the three-dimensional coordinate information of each parameter adjustment point based on the Dijkstra algorithm, and using the abnormal field strength area as a path planning restriction area to output the shortest initial operation path of the microwave control device; acquiring in real time state change data of the microwave control device during operation of the microwave control device according to the shortest initial operation path, and determining the field intensity fluctuation direction and field intensity fluctuation amplitude at the real-time operating position of the microwave control device based on the state change data; Acquiring parameter stability data of the microwave control device, wherein the parameter stability data includes data on the adaptability of the microwave control device to field intensity fluctuations of different directions and amplitudes; Acquiring recognition response time data of the microwave control device to the direction and amplitude of the field intensity fluctuation, and determining a running path adjustment hysteresis of the microwave control device according to the recognition response time data; Analyzing the field intensity fluctuation direction and field intensity fluctuation amplitude of the real-time operating position of the microwave control device according to the adaptability data and the operating path adjustment hysteresis, and determining the cumulative amount of operating path deviation of the microwave control device within the identification response time; If the cumulative running path deviation is less than a preset threshold, determining a running deviation direction and a running deviation distance of the microwave control device according to the shortest initial running path based on the cumulative running path deviation, and determining an adjustment direction and an adjustment distance of the microwave control device based on the running deviation direction and the running deviation distance to obtain adjustment data; Adjusting the shortest initial operation path of the microwave control device during real-time operation according to the adjustment data; If the cumulative deviation of the running path is greater than a preset threshold, obtaining the field intensity fluctuation amplitude and field intensity fluctuation direction data of the real-time running path of the microwave control device, constructing a field intensity change graph based on the field intensity fluctuation amplitude and field intensity fluctuation direction data of the real-time running path, and determining the field intensity change trend in the target interval of the target area based on the field intensity change graph; An interpolation operation is performed on the field intensity variation trend based on bilinear interpolation to determine field intensity information within a preset range of the shortest initial operating path. The operational stability of the microwave control device within the preset range of the shortest initial operating path is determined based on the field intensity information and adaptability data. Based on the operational stability, sections of the shortest initial operating path where the cumulative amount of operating path deviation is greater than a preset threshold are optimized to obtain an updated operating path. The microwave control device operates according to the updated operating path.

4. The method for remote plasma density homogenization control based on microwave excitation according to claim 1, characterized in that: The microwave control device obtains plasma state image data of the target interval according to the operation strategy, analyzes the plasma density uniformity of the target interval according to the image data, and obtains density control data, specifically: Acquiring standard image data of different types of plasma in a target area, and annotating the standard image data for density uniformity to obtain annotated image data; Building a density analysis model based on a recurrent neural network, and importing the labeled image data into the density analysis model for training; The microwave control device obtains plasma state image data of the target interval according to the operation strategy, imports the plasma state image data into the trained density analysis model to perform density uniformity identification, and performs statistics on the density deviation value of each area to obtain density control data.

5. The method for remote plasma density homogenization control based on microwave excitation according to claim 1, characterized in that: The density control data is transmitted to the ground control terminal by constructing a layered data transmission protocol stack based on wireless communication technology, specifically: Building a layered data transmission protocol stack based on wireless communication technology, encoding the density control data, and building a density control data transmission signal; The density control data transmission signal is sent to a ground control terminal according to the layered data transmission protocol stack, and the density control data transmission signal is decoded to obtain plasma density control result data of the target interval.

6. The method for remote plasma density homogenization control based on microwave excitation according to claim 1, characterized in that: The denoising and enhancement operation is performed on the plasma density detection signal data based on the sliding window filtering algorithm, specifically: The window size and sliding step of the sliding window are set, the plasma density detection signal data is traversed, the signal data in each window is mean filtered, the average value of the signal data in the window is calculated, and the average value is used as the denoised signal value at the center position of the window to obtain interference-suppressed plasma density detection signal data.

7. The method for remote plasma density homogenization control based on microwave excitation according to claim 1, characterized in that: The energy concentration, amplitude variance, total energy, and amplitude extreme value of each frequency band emitted by the microwave excitation device are calculated according to the frequency domain distribution matrix, specifically: Traverse each frequency band interval of the frequency domain distribution matrix, calculate the sum of the energy values ​​of all frequency points in the interval as the total energy, calculate the square mean of the amplitude in the interval as the energy concentration, calculate the arithmetic mean of the amplitude in the interval as the amplitude mean, calculate the difference between the maximum and minimum values ​​of the amplitude in the interval as the amplitude extreme value, and obtain the density detection characteristic spectrum.

8. The method for remote plasma density homogenization control based on microwave excitation according to claim 2, characterized in that: The field intensity characteristic tensor is constructed by gridding processing, specifically: The target area is divided into equally spaced grid points, the field strength detection signal data at each grid point is normalized, the field strength amplitude, gradient direction and phase information of each grid point are extracted, and the information is arranged according to the grid point position to construct a field strength feature tensor.

9. The method for remote plasma density homogenization control based on microwave excitation according to claim 4, characterized in that: The density analysis model is constructed based on the recurrent neural network, specifically: The input layer, hidden layer and output layer of the recurrent neural network are set. The input layer receives the pixel features of the labeled image data, the hidden layer processes the time series features through long short-term memory units, and the output layer outputs the density uniformity classification results to build a density analysis model.

Images