Plasma density detection method and system of remote plasma source

Abstract

The invention discloses a plasma density detection method and system of a remote plasma source, and relates to the technical field of plasma density detection.The method comprises the steps that a laser emitting device is controlled to emit a detection laser beam to penetrate through a to-be-detected plasma area, and scattering signals passing through plasma are collected; carrying out phase change analysis identification, extracting an interference light signal, and carrying out digital signal processing; performing density detection on the plasma, and constructing a plasma density space distribution matrix; and performing decomposition identification according to the plasma density spatial distribution matrix, determining a density abnormal region, and triggering a power partition regulation and control instruction of the remote plasma source. According to the invention, the technical problems in the prior art that the plasma density of the remote plasma source is difficult to monitor in real time and the accuracy and efficiency of detection and regulation are insufficient are solved, and the real-time monitoring of the plasma density of the remote plasma source is realized; and the accuracy and efficiency of plasma density detection and regulation are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of plasma density detection, and in particular to a plasma density detection method and system for a remote plasma source. Background Art

[0002] In the application of remote plasma technology, precise detection and control of plasma density is crucial for ensuring process stability and product quality. Traditional detection methods often suffer from insufficient accuracy and poor real-time performance, making it difficult to capture subtle changes in plasma density and its spatial distribution. Furthermore, in complex plasma transmission environments, they are susceptible to interference from self-luminescence and stray light, which reduces the signal-to-noise ratio of the detection signal and affects the accuracy of density calculations. Furthermore, existing control methods, which mostly rely on overall power adjustment, are unable to precisely control areas with abnormal density, making it difficult to meet the requirements of high-precision plasma processes.

[0003] The existing technology has technical problems such as difficulty in real-time monitoring of the plasma density of a remote plasma source, and insufficient accuracy and efficiency in detection and control. Summary of the Invention

[0004] The present application provides a method and system for detecting the plasma density of a remote plasma source, which is used to solve the technical problems in the prior art that it is difficult to monitor the plasma density of a remote plasma source in real time, and the accuracy and efficiency of detection and control are insufficient.

[0005] In view of the above problems, the present application provides a plasma density detection method and system for a remote plasma source.

[0006] A first aspect of the present application provides a method for detecting plasma density of a remote plasma source, the method comprising: A laser emitting device is controlled to emit a detection laser beam through a plasma region to be measured in a plasma transmission channel, and a scattered signal of the plasma is collected by a laser receiving device; a phase change analysis and identification is performed based on the scattered signal, an interference light signal is extracted, and digital signal processing is performed on the interference light signal to extract a phase change amount; density detection is performed on the plasma according to the phase change amount, and a plasma density spatial distribution matrix is ​​constructed; decomposition and identification are performed according to the plasma density spatial distribution matrix to determine density abnormality areas, and power partition control instructions of the remote plasma source are triggered according to the density abnormality areas.

[0007] A second aspect of the present application provides a plasma density detection system for a remote plasma source, the system comprising: The scattered signal acquisition module is used to control the laser emitting device to emit a detection laser beam through the plasma area to be measured in the plasma transmission channel, and collect the scattered signal of the plasma through the laser receiving device; the phase change extraction module is used to perform phase change analysis and identification based on the scattered signal, extract the interference light signal, perform digital signal processing on the interference light signal, and extract the phase change; the density space distribution matrix construction module is used to detect the density of the plasma according to the phase change and construct the plasma density space distribution matrix; the power control instruction triggering module is used to decompose and identify according to the plasma density space distribution matrix, determine the density abnormality area, and trigger the power partition control instruction of the remote plasma source according to the density abnormality area.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages: The laser emitting device is controlled to emit a detection laser beam through the plasma region to be measured in the plasma transmission channel. The scattered signal from the plasma is collected by the laser receiving device. Phase change analysis and identification are performed to extract the interference light signal, and the interference light signal is digitally processed. The plasma density is detected according to the phase change amount, and a plasma density spatial distribution matrix is ​​constructed. The plasma density spatial distribution matrix is ​​then decomposed and identified to determine the density abnormality area, triggering the power zoning control command of the remote plasma source. This achieves the technical effect of achieving real-time monitoring of the plasma density of the remote plasma source, improving the accuracy and efficiency of plasma density detection and control. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0010] Figure 1 A schematic flow chart of a method for detecting plasma density of a remote plasma source provided in an embodiment of the present application; Figure 2 A schematic structural diagram of a plasma density detection system for a remote plasma source provided in an embodiment of the present application; Description of the reference numerals: scattered signal acquisition module 10 , phase variation extraction module 20 , density space distribution matrix construction module 30 , power control instruction triggering module 40 . DETAILED DESCRIPTION

[0011] The present application provides a method and system for detecting the plasma density of a remote plasma source, which is used to solve the technical problems in the prior art that it is difficult to monitor the plasma density of a remote plasma source in real time, and the accuracy and efficiency of detection and control are insufficient.

[0012] The following will be combined with the accompanying drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making any creative work are within the scope of protection of this application.

[0013] Example 1, as Figure 1 As shown, the present application provides a method for detecting plasma density of a remote plasma source, the method comprising: Step S100: controlling the laser emitting device to emit a detection laser beam to pass through the plasma region to be measured in the plasma transmission channel, and collecting the scattered signal of the plasma through the laser receiving device.

[0014] Specifically, a laser emitting device and a laser receiving device are respectively arranged at both ends of a plasma transmission channel to form a laser interference optical path (the transmission channel contains the plasma region to be measured). The laser emitting device is controlled to emit a single-frequency continuous laser beam to penetrate the plasma region to be measured along the axial direction of the transmission channel, and the laser receiving device collects the initial scattering signal modulated by the plasma. The initial scattering signal is then wavelength filtered to suppress the plasma self-luminescence to obtain the first filtered data, and spatial filtering is performed to block off-axis stray light to obtain the second filtered data. After integrating the two parts of the filtered data, the scattered light intensity and signal phase are analyzed respectively to obtain scattered light intensity data and signal phase data. Finally, a joint analysis is performed based on these two types of data to obtain the required scattering signal.

[0015] Step S200: performing phase change analysis and identification based on the scattered signal, extracting an interference light signal, performing digital signal processing on the interference light signal, and extracting a phase change amount.

[0016] Specifically, the scattered signal is first separated and processed by a polarization beam splitter to obtain the interference component parameters, and then a heterodyne interference optical path is constructed. The scattered signal is synchronized to the optical path and optically mixed with the local oscillator light based on the interference component parameters to generate a difference-frequency interference signal. Subsequently, four groups of local oscillator reference electrical signals with the same phase difference are introduced and mixed and filtered with the difference-frequency interference signal in turn to obtain four groups of baseband light intensity values. The phase angle is calculated based on these four groups of baseband light intensity values, and the phase change is extracted, completing the digital signal processing of the interference light signal and the extraction of the phase change.

[0017] Step S300: performing density detection on the plasma according to the phase variation, and constructing a plasma density spatial distribution matrix.

[0018] Specifically, N detection planes (N is an integer greater than 1) are set along the axial direction of the plasma transmission channel, and M laser interference nodes (M is an integer greater than or equal to N) are evenly arranged circumferentially on each plane. These nodes are traversed and the phase changes are matched to construct the original phase matrix; based on the original phase matrix, the detection area is sparsely constrained and optimized to obtain multiple spatial grids, and the spatial grids are mapped and reconstructed with the original phase matrix to obtain a density distribution matrix; the density distribution matrix is ​​axially layered to obtain multiple equal-interval levels, and multiple interpolation points are obtained through radial interpolation processing. The equal-interval levels and interpolation points are combined to construct a three-dimensional density field, and a three-dimensional density distribution cloud map is generated through visual mapping rendering; after calculating the density gradient field, it is reversely migrated and mapped to the three-dimensional density distribution cloud map, and finally the plasma density spatial distribution matrix is ​​constructed.

[0019] Step S400: Decomposing and identifying the plasma density spatial distribution matrix, determining density abnormality areas, and triggering power partitioning control instructions for the remote plasma source based on the density abnormality areas.

[0020] Specifically, a tensor decomposition is first performed based on the plasma density spatial distribution matrix, which is then expanded into multi-layer two-dimensional slices. The row-direction eigenvectors and column-direction eigenvectors are traversed through the slices and then merged to construct a core tensor. The abnormal areas are marked based on this and the abnormal spatial distribution map is reconstructed. The map is scanned for aggregation analysis to extract abnormal characteristic components. A density fluctuation quantitative analysis is performed on the abnormal characteristic components to set a density fluctuation index, and the density abnormality area containing the abnormal pattern is determined based on the index. Multi-objective optimization is performed based on the abnormal pattern to generate a power partition control instruction, which drives the remote plasma source to perform partition power adjustment.

[0021] In one possible implementation, step S100 further includes: Step S110: arranging a laser emitting device and a laser receiving device at both ends of a plasma transmission channel to form a laser interference optical path, wherein the plasma transmission channel includes a plasma region to be measured.

[0022] Step S120: controlling the laser emitting device to emit a single-frequency continuous laser beam to penetrate the plasma region to be measured along the axial direction of the transmission channel.

[0023] Step S130: collecting the initial scattered signal modulated by the plasma through the laser receiving device.

[0024] Step S140: performing a dual-parameter joint analysis based on the initial scattering signal to obtain the scattering signal.

[0025] Specifically, a laser emitting device and a laser receiving device are installed at both ends of the plasma transmission channel containing the plasma area to be measured, respectively, so that the center lines of the optical paths of the two coincide with the axial direction of the transmission channel, forming a laser interference optical path that runs through the entire plasma area to be measured, ensuring that the subsequent detection laser beam emitted by the laser emitting device can completely penetrate the area to be measured along the axial direction and be accurately captured by the laser receiving device, laying the foundation for the subsequent collection of scattered signals.

[0026] The laser emission device is activated by the control module within the plasma transmission channel (which contains the plasma region to be measured), where a laser interference optical path has been formed. This laser beam propagates strictly along the axial path of the transmission channel, precisely penetrating the plasma region to be measured within the channel. During this penetration, it interacts with the plasma particles, forming a modulated beam that carries characteristic information about the plasma region to be measured, providing the fundamental light source for the subsequent collection of scattered signals.

[0027] After a single-frequency continuous laser beam emitted by a laser transmitter penetrates the plasma region along the axial direction of the transmission channel, a laser receiver is activated to accurately capture the plasma-modulated beam signal, known as the initial scattered signal. This initial scattered signal contains various information generated by the interaction between the laser and plasma, as well as interference from plasma self-luminescence and off-axis stray light, providing raw data for subsequent signal processing.

[0028] Based on the initial plasma-modulated scattering signal collected by the laser receiving device, wavelength filtering is first performed to suppress the interference of plasma self-luminescence to obtain the first filtered data; at the same time, spatial filtering is performed to block off-axis stray light to obtain the second filtered data; then the first filtered data and the second filtered data are integrated, and according to the integration results, scattered light intensity analysis is carried out to obtain scattered light intensity data, and phase analysis is carried out to obtain signal phase data. Finally, a dual-parameter joint analysis is performed based on the scattered light intensity data and the signal phase data to obtain the required scattering signal.

[0029] In one possible implementation, step S140 further includes: Step S141: performing wavelength filtering on the initial scattered signal to suppress plasma self-luminescence and obtain first filtered data.

[0030] Step S142: performing spatial filtering on the initial scattered signal to block off-axis stray light, and obtaining second filtered data.

[0031] Step S143: Integrate the first filtered data and the second filtered data, perform scattered light intensity analysis based on the integration result to obtain scattered light intensity data, and perform phase analysis based on the integration result to obtain signal phase data.

[0032] Step S144: performing a joint analysis based on the scattered light intensity data and the signal phase data to obtain the scattered signal.

[0033] Specifically, the initial scattered signal modulated by the plasma and collected by the laser receiving device is processed by wavelength filtering. This processing process is specifically used to suppress the interference light emitted by the plasma itself (i.e., plasma self-luminescence). After the filtering operation, the first filtered data is obtained without the influence of the plasma self-luminescence.

[0034] When spatially filtering the initial scattered signal collected by the laser receiving device, a spatial filter is used. This spatial filter sets a light-clearing aperture that matches the main axis of the laser interference optical path, allowing only effective scattered light propagating along the axial direction of the optical path to pass through, while forming a physical block for off-axis stray light that deviates from the main axis, thereby filtering out interfering light in the off-axis direction, and finally obtaining the second filtered data without the influence of off-axis stray light, providing a purer signal basis for subsequent data integration.

[0035] The first filtered data (wavelength filtering to suppress plasma self-luminescence) and the second filtered data (spatial filtering to block off-axis stray light) are integrated using a data stitching algorithm, and the two sets of data are associated according to the signal acquisition time sequence to form a complete data set; based on the integration results, the light intensity amplitude in the data is quantified and calculated using the light intensity feature extraction algorithm to obtain the scattered light intensity data; at the same time, the phase feature analysis algorithm is used to extract the phase offset information of the optical signal in the data to obtain the signal phase data, which provides a basis for subsequent dual-parameter joint analysis.

[0036] When jointly analyzing scattered light intensity data and signal phase data to obtain the scattered signal, the two are incorporated into a unified signal analysis framework, and the digital signal processing module is used to collaboratively calculate the energy change characteristics reflected by the intensity data and the fluctuation offset characteristics reflected by the phase data. By establishing a light intensity-phase correlation model, the coupling relationship between the two under the action of plasma is quantified, and the interference caused by the fluctuation of a single parameter is eliminated. Finally, a scattered signal that can accurately reflect the plasma modulation characteristics is synthesized and output, providing effective input for subsequent phase change analysis.

[0037] In one possible implementation, step S200 further includes: Step S210: Separate and process the scattered signal through a polarization beam splitter to obtain interference component parameters.

[0038] Step S220: construct a heterodyne interference optical path, synchronize the scattered signal to the heterodyne interference optical path according to the interference component parameters for analysis, generate an interference optical signal, the interference optical signal is a difference frequency interference signal, perform four-part phase shift demodulation on the difference frequency interference signal, and obtain the phase change.

[0039] Specifically, the acquired scattered signals are separated and processed using the optical properties of a polarization beam splitter. Based on the differences in the polarization states of light, the polarization beam splitter can separate the polarization components contained in the scattered signals that are relevant to interference analysis, while eliminating irrelevant polarization state signals. This allows the extraction of interference component parameters that can be used for subsequent heterodyne interferometry optical path analysis, laying the foundation for generating interference light signals.

[0040] First, a heterodyne interference optical path consisting of a laser transmitter, a spectrometer, a reflector, etc. is built to provide a hardware foundation for signal analysis; according to the interference component parameters, the scattered signal is accurately introduced into the optical path using a synchronous controller, so that the scattered signal and the local oscillator light are mixed through an optical mixer in the optical path to generate a difference-frequency interference signal; then, a four-part phase-shift demodulation device is used to introduce four groups of local oscillator reference electrical signals with the same phase difference, which are mixed with the difference-frequency interference signal in turn through a mixer, and four groups of baseband light intensity values ​​are obtained after filtering by a filter. The phase angle is then calculated based on these four groups of baseband light intensity values ​​through a phase calculator to finally obtain the phase change.

[0041] In one possible implementation, step S220 further includes: Step S221: performing optical mixing with the local oscillator light based on the interference component parameters to generate the difference frequency interference signal.

[0042] Step S222: Introduce four groups of local oscillator reference electrical signals with the same phase difference, and perform mixing and filtering with the difference frequency interference signal in sequence to obtain four groups of baseband light intensity values.

[0043] Step S223: performing phase angle calculation based on the four groups of baseband light intensity values ​​to obtain the phase change.

[0044] Specifically, the interference component parameters separated by the polarization beam splitter are input into an optical mixer along with the local oscillator light, where the frequency difference between the two is exploited for optical mixing. During the mixing process, the plasma modulation information carried by the interference component parameters interacts with the stable frequency of the local oscillator light, generating a beat frequency signal with a specific frequency difference, known as a difference frequency interference signal. This signal more intuitively reflects the phase change characteristics of the laser after passing through the plasma, providing the core signal source for subsequent phase demodulation processing.

[0045] Four sets of local oscillator reference electrical signals with specific phase relationships are prepared. The phase difference between each set of signals is strictly maintained at 90 degrees, namely 0 degrees, 90 degrees, 180 degrees, and 270 degrees, to ensure a stable phase progression. In the specific operation, the first step is to generate a local oscillator reference electrical signal with a 0-degree phase. This signal is input into a mixer along with the obtained difference frequency interference signal for mixing. The two signals interact within the mixer to produce a mixed signal containing phase information. The mixed signal is then fed into an integrator for integration to smooth signal fluctuations and accumulate effective information. The integrated signal is then accurately sampled by a sampler to obtain the first set of baseband optical intensity values ​​corresponding to the 0-degree phase reference signal. The second step is to generate a local oscillator reference electrical signal with a 90-degree phase. Using the same processing flow as the first step, this 90-degree phase reference signal is first mixed with the difference frequency interference signal in a mixer to generate a new mixed signal. This is then integrated by an integrator to eliminate high-frequency noise and transient interference, and then sampled by a sampler to obtain the second set of baseband optical intensity values ​​corresponding to the 90-degree phase reference signal. The third step is to generate a local oscillator reference signal with a 180-degree phase. The mixing, integration, and sampling steps described above are repeated: the 180-degree phase reference signal is first mixed with the difference frequency interference signal to generate a mixed signal containing the phase information. The mixed signal is then integrated using an integrator to enhance signal stability. Finally, it is sampled by a sampler to obtain a third set of baseband intensity values ​​corresponding to the 180-degree phase reference signal. The fourth step is to generate a local oscillator reference signal with a 270-degree phase. The same mixing, integration, and sampling steps are repeated: the 270-degree phase reference signal and the difference frequency interference signal are mixed in a mixer, integrated by an integrator, and sampled by a sampler to obtain a fourth set of baseband intensity values ​​corresponding to the 270-degree phase reference signal. These four sets of baseband intensity values ​​fully record the intensity characteristics of the difference frequency interference signal at different phase references, providing comprehensive and accurate raw data for subsequent phase angle calculation to determine the phase change.

[0046] Phase angle calculations are performed based on the four sets of baseband intensity values ​​obtained (recorded as the first-step intensity value I1, the second-step intensity value I2, the third-step intensity value I3, and the fourth-step intensity value I4). First, the numerator is calculated: the fourth-step intensity value I4 minus the second-step intensity value I2. Next, the denominator is calculated: the first-step intensity value I1 minus the third-step intensity value I3. The numerator and denominator are then divided to obtain a ratio, which is then subjected to an inverse tangent operation. The angle obtained through this series of calculations is the real-time phase change that reflects the phase shift of the laser as it passes through the plasma.

[0047] In one possible implementation, step S300 further includes: Step S310: setting N detection planes along the axial direction of the plasma transmission channel, and evenly arranging M laser interference nodes circumferentially on each of the N detection planes, where N is an integer greater than 1, and M is an integer greater than or equal to N.

[0048] Step S320: traverse the M laser interference nodes and perform matching in combination with the phase variation, obtain the phase variation of the M laser interference nodes, and construct an original phase matrix.

[0049] Step S330: performing compressed sensing reconstruction based on the original phase matrix to obtain a density distribution matrix, performing three-dimensional analysis based on the density distribution matrix, and generating a three-dimensional density distribution cloud map.

[0050] Step S340: performing density calculation based on the three-dimensional density distribution cloud map to obtain a density gradient field.

[0051] Step S350: Mapping the reverse migration of the density gradient field to the three-dimensional density distribution cloud map to construct the plasma density spatial distribution matrix.

[0052] Specifically, N mutually parallel detection planes (N is an integer greater than 1) are arranged at intervals in the axial direction of the plasma transmission channel to achieve coverage of different axial positions of the channel; at the same time, in each detection plane, M laser interference nodes (M is an integer greater than or equal to N) are evenly distributed along the circumference to ensure that each plane can capture signals from multiple angles. Through this multi-plane, multi-node arrangement, a three-dimensional detection network is formed, laying the foundation for comprehensive acquisition of the spatial distribution information of the plasma.

[0053] The M laser interference nodes arranged circumferentially along the N detection planes are traversed one by one, and the spatial position information of each node is accurately matched with the previously extracted phase change to determine the specific value of the phase change corresponding to each laser interference node; then, these M phase changes are arranged in order according to the spatial distribution order of the nodes to construct the original phase matrix, which can intuitively reflect the phase change of each detection node and provide initial data support for the subsequent reconstruction of the density distribution matrix through compressed sensing.

[0054] Based on the original phase matrix, with the help of the sparse constrained optimization method in the compressed sensing theory, the detection area is divided into spatial grids, and these spatial grids are mapped and reconstructed with the original phase matrix to obtain a density distribution matrix that can reflect the plasma density distribution; then, axial stratification is performed according to the density distribution matrix to obtain multiple equal-interval levels, and radial interpolation processing is performed based on these equal-interval levels to obtain multiple interpolation points, and then three-dimensional spatial processing is performed according to the equal-interval levels and interpolation points to construct a three-dimensional density field. Finally, visual mapping and rendering are performed based on the three-dimensional density field to generate a three-dimensional density distribution cloud map that can intuitively display the spatial distribution of plasma density.

[0055] When performing density calculations based on a three-dimensional density distribution cloud map, a spatial sampling algorithm is used to extract density values ​​from discrete points in the cloud map to obtain specific density data for each sampling point. Subsequently, a three-dimensional gradient operator (a three-dimensional extension of the Sobel operator) is used to process these density data. By calculating the density difference and distance ratio of adjacent sampling points in the x, y, and z coordinate axes, the density change rate of each point along the three directions is obtained. These change rates are combined into gradient vectors to construct a complete density gradient field, which can accurately reflect the changing trend and rate of plasma density in space.

[0056] Based on the density gradient field and the three-dimensional density distribution cloud map, the gradient information of the density gradient field is mapped to the spatial position of the three-dimensional density distribution cloud map through the reverse migration algorithm, realizing the mapping of gradient information to the cloud map. In this process, the density gradient vector of each spatial point is associated and fused with the density value of that point to ensure that the two are accurately matched in spatial coordinates. Ultimately, the integration forms a plasma density spatial distribution matrix that comprehensively reflects the plasma density size and its spatial variation trend, providing comprehensive data support for the subsequent identification of density anomaly areas.

[0057] In one possible implementation, step S330 further includes: Step S331: performing sparse constrained optimization partitioning on the area to be detected based on the original phase matrix to obtain multiple spatial grids.

[0058] Step S332: mapping and reconstructing the multiple spatial grids and the original phase matrix to obtain a density distribution matrix.

[0059] Step S333: performing axial stratification according to the density distribution matrix to obtain a plurality of equal-interval levels.

[0060] Step S334: performing radial difference processing based on the multiple equally spaced levels to obtain multiple interpolation points.

[0061] Step S335: performing three-dimensional space processing according to the multiple equal-interval levels and the multiple interpolation points to construct a three-dimensional density field.

[0062] Step S336: Perform visual mapping rendering according to the three-dimensional density field to generate the three-dimensional density distribution cloud map.

[0063] Specifically, a sparse constrained optimization algorithm based on the L1 norm is adopted. The phase data in the original phase matrix is ​​used as the constraint condition. By setting the grid size parameter and the sparsity threshold, an optimization objective function is constructed. The objective function is then solved using the gradient descent method, and multiple spatial grids are automatically divided in the area to be tested. The distribution of these grids matches the valid data points in the original phase matrix, ensuring that the grid division can not only cover the entire area to be tested, but also maintain a high division accuracy in areas with rich data information, and finally obtain multiple spatial grids that meet the sparse constraint conditions.

[0064] A mapping algorithm based on grid index matching is adopted. First, a unique index value is assigned to each spatial grid, and the phase data in the original phase matrix is ​​indexed and labeled according to the spatial position. The spatial grid is associated with the corresponding phase data through index matching, and then the orthogonal matching pursuit algorithm in compressed sensing is used to reconstruct the associated phase data, converting the phase information into the density value corresponding to the grid. Finally, the density values ​​are arranged into a matrix form according to the spatial distribution order of the grid to obtain the density distribution matrix.

[0065] An axial equidistant partitioning algorithm is adopted. The spatial range parameters of the plasma transmission direction are first extracted from the density distribution matrix to determine the total axial length. Then, the spacing between adjacent levels is calculated according to the preset number of levels. By setting equally spaced dividing lines along the axial direction in the density distribution matrix, the entire matrix is ​​divided into multiple continuous and equally spaced sub-matrix areas. Each sub-matrix area corresponds to an axial level, and finally multiple equally spaced levels are obtained, realizing the structured division of the density distribution matrix in the axial direction.

[0066] When performing radial difference processing based on multiple equal-interval levels, a linear interpolation algorithm is used. First, the existing density data points in the radial direction within each equal-interval level are determined. Then, the interpolation interval is set according to the preset interpolation accuracy requirements. Finally, the density value of each interpolation position between adjacent known data points is calculated using the linear interpolation formula, thereby obtaining multiple evenly distributed radial interpolation points to fill the radial data gaps and improve the continuity of the density distribution data.

[0067] Using three-dimensional spatial interpolation and gridding technology, the axial coordinates of multiple equally spaced levels are first associated with the radial interpolation point coordinates within the corresponding levels to establish a three-dimensional spatial coordinate system; then the Kriging interpolation algorithm is used to spatially interpolate the discrete interpolation point data to fill the data gaps in the three-dimensional space and make the density data continuously distributed in space; finally, through three-dimensional gridding processing, the interpolated continuous data is divided into a regular spatial grid, forming a three-dimensional density field that can accurately characterize the density distribution of plasma in three-dimensional space.

[0068] Based on the constructed three-dimensional density field, the numerical information in the density field is converted into intuitive image information through visual mapping rendering technology, and the plasma density values ​​at different positions in the three-dimensional density field are mapped and associated with specific colors or transparencies. Areas with higher density values ​​are displayed with specific highlight colors, and areas with lower density values ​​correspond to darker colors. Rendering algorithms such as ray tracing are then used to enhance the three-dimensionality and layering of the image, and finally a three-dimensional density distribution cloud map is generated that can clearly show the distribution differences and changing trends of plasma density in three-dimensional space.

[0069] In one possible implementation, step S400 further includes: Step S410: performing tensor decomposition based on the plasma density spatial distribution matrix, marking abnormalities according to the decomposition results, and extracting abnormal feature components.

[0070] Step S420: performing density fluctuation quantitative analysis based on the abnormal characteristic components and setting a density fluctuation index.

[0071] Step S430: determining a density abnormality region according to the density fluctuation index, wherein the density abnormality region includes an abnormal pattern.

[0072] Step S440: performing multi-objective optimization according to the abnormal pattern and generating power partition control instructions.

[0073] Step S450: driving the remote plasma source to perform partition power adjustment through the power partition control instruction.

[0074] Specifically, the plasma density spatial distribution matrix is ​​first expanded into multiple layers of two-dimensional slices with the three-dimensional density distribution cloud map as a reference. By traversing these two-dimensional slices, the eigenvectors in the row and column directions are calculated respectively, and then the eigenvectors in these two directions are merged to construct the core tensor. Subsequently, the plasma density spatial distribution matrix is ​​comprehensively traversed based on the core tensor, and the abnormal areas are marked. The three-dimensional density distribution cloud map is reconstructed according to the abnormal areas to obtain the abnormal spatial distribution map. Finally, the distribution map is scanned and aggregated for analysis to extract the components that can characterize the abnormal characteristics.

[0075] First, the density data in the abnormal characteristic component is decomposed into time series to extract trend terms, periodic terms and random fluctuation terms; the density fluctuation amplitude is quantified by calculating the root mean square error (RMSE) and peak factor of the fluctuation term, and the fluctuation frequency per unit time is counted using the sliding window algorithm; then, according to the stability requirements of the plasma process, the quantified results are compared and fitted with the preset threshold range, and the density fluctuation index including the upper limit of the fluctuation amplitude, frequency threshold and duration threshold is set to achieve accurate definition of abnormal fluctuations.

[0076] Using the set density fluctuation index as the judgment standard, the plasma density spatial distribution matrix is ​​scanned point by point, and the areas in the matrix where the density fluctuation amplitude exceeds the upper limit of the index, the fluctuation frequency exceeds the threshold or the fluctuation duration is too long are screened out and determined to be density anomaly areas; the spatial distribution characteristics of these abnormal areas are analyzed, and abnormal patterns such as local high-density aggregation, periodic fluctuations along the transmission direction, and radial asymmetric distribution are summarized. These patterns can reflect the specific forms and laws of density anomalies and provide a basis for subsequent targeted regulation.

[0077] An improved non-dominated sorting genetic algorithm is adopted, and the density adjustment requirements, power control range, energy consumption limit and other requirements corresponding to the abnormal mode are taken as optimization objectives and constraints. The initial population is generated by encoding, the objective function value of each individual is calculated, and non-dominated sorting and congestion calculation are performed. Selection, crossover and mutation operations are performed based on the sorting results, and the Pareto optimal solution set is obtained through iterative optimization. The solution with the best comprehensive performance is selected from the solution set and converted into the power adjustment parameters of each partition, and finally the power partition control instructions are generated.

[0078] The generated power zone control instructions are converted into electrical signals or control codes recognizable by the remote plasma source and sent to the plasma source's power control unit via a signal transmission module. Based on the power adjustment parameters for each zone in the instructions, the control unit precisely adjusts the power output components, such as the RF power supply and microwave generator, in the corresponding area. For example, it increases power output in high-density anomalies to reduce local density, or decreases power output in low-density anomalies to increase local density. This achieves zone-by-zone correction of plasma density, making the overall density distribution uniform and stable.

[0079] In one possible implementation, step S410 further includes: Step S411: Expanding the plasma density spatial distribution matrix into multiple layers of two-dimensional slices based on the three-dimensional density distribution cloud map.

[0080] Step S412: traverse the multiple layers of two-dimensional slices and perform calculations in the row direction to obtain row-direction feature vectors.

[0081] Step S413: traverse the multiple layers of two-dimensional slices and perform calculations in the column direction to obtain column-direction feature vectors.

[0082] Step S414: Merge the row-direction feature vector and the column-direction feature vector to construct a core tensor.

[0083] Step S415: traversing the plasma density spatial distribution matrix according to the core tensor, marking abnormal areas, and reconstructing the three-dimensional density distribution cloud map according to the abnormal areas to obtain an abnormal spatial distribution map.

[0084] Step S416: Scanning is performed based on the abnormal spatial distribution map, and the scanning results are aggregated and analyzed to extract the abnormal characteristic components.

[0085] Specifically, with the spatial structure presented by the three-dimensional density distribution cloud map as a reference, the cutting interval of the plasma density spatial distribution matrix in the axial direction (plasma transmission direction) is determined, and the matrix is ​​unfolded layer by layer according to the interval to obtain multi-layer two-dimensional slices corresponding to each axial position of the three-dimensional density distribution cloud map. Each layer of slice presents the radial and circumferential distribution of the plasma density at the corresponding axial position in the form of a two-dimensional matrix, thereby realizing the layered decomposition of the three-dimensional density data.

[0086] For each slice, a sliding window of fixed size is set along the row direction, and the mean, variance and energy value of the density data in the window are calculated as local features; after traversing all rows, the features of each window are concatenated in row order to form a row feature vector of a single slice; finally, the row feature vectors of all slices are dimensionally aligned and stacked, and the dimensionality reduction processing is performed through principal component analysis to obtain a row feature vector that can characterize the overall row distribution law.

[0087] For each slice, a sliding window of fixed size is set along the column direction, and the mean, standard deviation and peak factor of the density data in the window are calculated as local features. After traversing all columns, the features of each window are concatenated in column order to form a column feature vector of a single slice. The column feature vectors of all slices are dimensionalized and stacked, and feature dimensionality reduction and fusion are performed through linear discriminant analysis (LDA), finally obtaining a column-wise feature vector that can comprehensively reflect the column-wise distribution characteristics of multi-layer slices.

[0088] The row-direction eigenvector and column-direction eigenvector are dimensionally normalized to ensure that the two match each other in feature dimensions; the two vectors are then fused at a high order through the Kronecker product operation to generate a three-dimensional tensor prototype containing row and column cross features; the prototype is then optimized through tensor rank decomposition to retain the core feature dimensions and eliminate redundant information, ultimately constructing a core tensor that can accurately reflect the correlation between row and column features.

[0089] Taking the core tensor as the reference template, a fixed-size three-dimensional sliding window is set to traverse the plasma density spatial distribution matrix point by point. By calculating the cosine similarity between the data in the window and the corresponding area of ​​the core tensor, the window with a similarity lower than the set threshold is marked as an abnormal area; then the spatial coordinates and density values ​​of all abnormal areas are extracted, and these data are rendered separately using a volume rendering algorithm, retaining the three-dimensional morphology and density gradient characteristics of the abnormal area, shielding the data of the normal area, and finally generating an abnormal spatial distribution map that focuses on displaying the spatial distribution of the abnormal area.

[0090] A combination of three-dimensional spatial scanning and cluster analysis is adopted. The abnormal spatial distribution map is discretized into a three-dimensional voxel grid through voxelization processing. The grid is scanned in full space using a ray casting algorithm, and the density value and spatial coordinates of each voxel are recorded. The scanned voxel data are then aggregated using the DBSCAN clustering algorithm, and the voxels are divided into multiple abnormal clusters based on spatial distance and density similarity. The characteristic parameters such as volume proportion, average density, and density gradient extreme value of each abnormal cluster are calculated, and the most representative parameter combination is screened out through a feature selection algorithm, ultimately forming an abnormal feature component that can accurately describe the characteristics of the abnormal area.

[0091] Embodiment 2 is based on the same inventive concept as the plasma density detection method of a remote plasma source in the above embodiment. Figure 2 As shown, the present application provides a plasma density detection system for a remote plasma source. The system and method embodiments in the present application are based on the same inventive concept. The system includes: The scattered signal acquisition module 10 is used to control the laser emitting device to emit a detection laser beam to pass through the plasma area to be measured in the plasma transmission channel, and to collect the scattered signal of the plasma through the laser receiving device.

[0092] The phase variation extraction module 20 is configured to perform phase variation analysis and identification based on the scattered signal, extract the interference light signal, perform digital signal processing on the interference light signal, and extract the phase variation.

[0093] The density spatial distribution matrix construction module 30 is used to detect the density of the plasma according to the phase change amount and construct the plasma density spatial distribution matrix.

[0094] The power control instruction triggering module 40 is used to decompose and identify the plasma density spatial distribution matrix, determine the density abnormality area, and trigger the power partition control instruction of the remote plasma source according to the density abnormality area.

[0095] Furthermore, the system is also used to implement the following functions: A laser emitting device and a laser receiving device are respectively arranged at both ends of a plasma transmission channel to form a laser interference optical path. The plasma transmission channel contains a plasma region to be measured. The laser emitting device is controlled to emit a single-frequency continuous laser beam to penetrate the plasma region to be measured along the axial direction of the transmission channel. The laser receiving device collects an initial scattering signal modulated by the plasma. A dual-parameter joint analysis is performed based on the initial scattering signal to obtain the scattering signal.

[0096] Furthermore, the system is also used to implement the following functions: The initial scattered signal is wavelength filtered to suppress plasma self-luminescence to obtain first filtered data; the initial scattered signal is spatially filtered to block off-axis stray light to obtain second filtered data; the first filtered data and the second filtered data are integrated, scattered light intensity analysis is performed based on the integration result to obtain scattered light intensity data, phase analysis is performed based on the integration result to obtain signal phase data; and the scattered light intensity data and the signal phase data are jointly analyzed to obtain the scattered signal.

[0097] Furthermore, the system is also used to implement the following functions: The scattered signal is separated and processed by a polarization beam splitter prism to obtain interference component parameters; a heterodyne interference optical path is constructed, and the scattered signal is synchronized to the heterodyne interference optical path according to the interference component parameters for analysis to generate an interference light signal, which is a difference frequency interference signal. The difference frequency interference signal is subjected to four-part phase shift demodulation to obtain the phase change amount.

[0098] Furthermore, the system is also used to implement the following functions: Based on the interference component parameters, optical mixing is performed with the local oscillator light to generate the difference frequency interference signal; four groups of local oscillator reference electrical signals with the same phase difference are introduced, and mixed and filtered with the difference frequency interference signal in sequence to obtain four groups of baseband light intensity values; and phase angle calculation is performed based on the four groups of baseband light intensity values ​​to obtain the phase change amount.

[0099] Furthermore, the system is also used to implement the following functions: N detection planes are set along the axial direction of the plasma transmission channel, and M laser interference nodes are evenly arranged circumferentially on each of the N detection planes, where N is an integer greater than 1 and M is an integer greater than or equal to N; the M laser interference nodes are traversed and matched in combination with the phase change amount to obtain the phase change amount of the M laser interference nodes, and an original phase matrix is ​​constructed; compressed sensing reconstruction is performed based on the original phase matrix to obtain a density distribution matrix, and a three-dimensional analysis is performed based on the density distribution matrix to generate a three-dimensional density distribution cloud map; density calculation is performed based on the three-dimensional density distribution cloud map to obtain a density gradient field; the density gradient field is reversely migrated and mapped to the three-dimensional density distribution cloud map to construct the plasma density spatial distribution matrix.

[0100] Furthermore, the system is also used to implement the following functions: Based on the original phase matrix, the detection area is divided into sparse constrained optimization to obtain multiple spatial grids; the multiple spatial grids are mapped and reconstructed with the original phase matrix to obtain a density distribution matrix; axial stratification is performed according to the density distribution matrix to obtain multiple equal-interval levels; radial difference processing is performed based on the multiple equal-interval levels to obtain multiple interpolation points; three-dimensional space processing is performed according to the multiple equal-interval levels and the multiple interpolation points to construct a three-dimensional density field; visual mapping rendering is performed according to the three-dimensional density field to generate the three-dimensional density distribution cloud map.

[0101] Furthermore, the system is also used to implement the following functions: Tensor decomposition is performed based on the plasma density spatial distribution matrix, anomalies are marked according to the decomposition results, and abnormal characteristic components are extracted; density fluctuations are quantitatively analyzed according to the abnormal characteristic components, and density fluctuation indicators are set; density abnormality areas are determined according to the density fluctuation indicators, and the density abnormality areas include abnormal patterns; multi-objective optimization is performed according to the abnormal patterns to generate power partition control instructions; and the power partition control instructions are used to drive a remote plasma source to perform partition power adjustment.

[0102] Furthermore, the system is also used to implement the following functions: The plasma density spatial distribution matrix is ​​expanded into multiple layers of two-dimensional slices based on the three-dimensional density distribution cloud map; the multiple layers of two-dimensional slices are traversed to perform calculations in the row direction to obtain row-direction feature vectors; the multiple layers of two-dimensional slices are traversed to perform calculations in the column direction to obtain column-direction feature vectors; the row-direction feature vectors and the column-direction feature vectors are merged to construct a core tensor; the plasma density spatial distribution matrix is ​​traversed according to the core tensor to mark abnormal areas, and the three-dimensional density distribution cloud map is reconstructed according to the abnormal areas to obtain an abnormal spatial distribution map; scanning is performed based on the abnormal spatial distribution map, the scanning results are aggregated and analyzed, and the abnormal feature components are extracted.

[0103] It should be noted that the order in which the embodiments of the present application are presented is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. Furthermore, the foregoing descriptions of specific embodiments of this specification are provided. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order or sequential sequence shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0104] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should be included in the scope of protection of the present application.

[0105] This specification and drawings are merely illustrative of the present application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Obviously, those skilled in the art may make various modifications and variations to this application without departing from the scope of this application. Thus, this application is intended to include such modifications and variations as fall within the scope of this application and its equivalents.

Claims

1. A method for detecting plasma density of a remote plasma source, characterized in that: The method comprises: Controlling the laser emitting device to emit a detection laser beam through the plasma region to be measured in the plasma transmission channel, and collecting the scattered signal of the plasma through the laser receiving device; Performing phase change analysis and identification based on the scattered signal, extracting an interference light signal, performing digital signal processing on the interference light signal, and extracting a phase change amount; Performing density detection on the plasma according to the phase change amount, and constructing a plasma density spatial distribution matrix; Decomposition and identification are performed according to the plasma density spatial distribution matrix to determine density abnormality areas, and power partition control instructions of the remote plasma source are triggered according to the density abnormality areas.

2. The plasma density detection method of a remote plasma source according to claim 1, wherein: The method includes controlling a laser emitting device to emit a detection laser beam to pass through a plasma region to be measured in a plasma transmission channel, and collecting a scattered signal of the plasma through a laser receiving device. The method includes: A laser emitting device and a laser receiving device are respectively arranged at both ends of a plasma transmission channel to form a laser interference optical path, wherein the plasma transmission channel includes a plasma region to be measured; Controlling the laser emitting device to emit a single-frequency continuous laser beam to penetrate the plasma region to be measured along the axial direction of the transmission channel; collecting the initial scattered signal modulated by the plasma through the laser receiving device; A dual-parameter joint analysis is performed based on the initial scattering signal to obtain the scattering signal.

3. The plasma density detection method of a remote plasma source according to claim 2, wherein: Performing a dual-parameter joint analysis based on the initial scattering signal to obtain the scattering signal, the method comprising: Performing wavelength filtering on the initial scattered signal to suppress plasma self-luminescence, and obtaining first filtered data; Performing spatial filtering on the initial scattered signal to block off-axis stray light, to obtain second filtered data; Integrate the first filtered data and the second filtered data, perform scattered light intensity analysis based on the integration result to obtain scattered light intensity data, and perform phase analysis based on the integration result to obtain signal phase data; The scattered light signal is obtained by performing a joint analysis based on the scattered light intensity data and the signal phase data.

4. The method for detecting plasma density of a remote plasma source according to claim 1, wherein: Performing phase change analysis and identification based on the scattered signal, extracting the interference light signal, performing digital signal processing on the interference light signal, and extracting the phase change amount, the method includes: Separating and processing the scattered signal by a polarization beam splitter prism to obtain interference component parameters; Construct a heterodyne interference optical path, synchronize the scattered signal to the heterodyne interference optical path according to the interference component parameters for analysis, generate an interference optical signal, the interference optical signal is a difference frequency interference signal, and perform four-part phase shift demodulation on the difference frequency interference signal to obtain the phase change amount.

5. The method for detecting plasma density of a remote plasma source according to claim 4, wherein: Synchronizing the scattered signal to the heterodyne interference optical path for analysis according to the interference component parameters to generate an interference optical signal, wherein the interference optical signal is a difference frequency interference signal, and performing four-part phase shift demodulation on the difference frequency interference signal to obtain the phase change amount, the method comprising: Performing optical mixing with the local oscillator light based on the interference component parameters to generate the difference frequency interference signal; Four sets of local oscillator reference electrical signals with the same phase difference are introduced and mixed and filtered with the difference frequency interference signal in sequence to obtain four sets of baseband light intensity values; The phase angle is calculated according to the four groups of baseband light intensity values ​​to obtain the phase change.

6. The method for detecting plasma density of a remote plasma source according to claim 1, wherein: The plasma density is detected according to the phase change amount, and a plasma density spatial distribution matrix is ​​constructed, the method comprising: N detection planes are set along the axial direction of the plasma transmission channel, and M laser interference nodes are evenly arranged circumferentially on each of the N detection planes, where N is an integer greater than 1 and M is an integer greater than or equal to N; Traversing the M laser interference nodes and matching them with the phase change amount, obtaining the phase change amount of the M laser interference nodes, and constructing an original phase matrix; Performing compressed sensing reconstruction based on the original phase matrix to obtain a density distribution matrix, performing three-dimensional analysis based on the density distribution matrix to generate a three-dimensional density distribution cloud map; Performing density calculation based on the three-dimensional density distribution cloud map to obtain a density gradient field; The density gradient field is reversely migrated and mapped to the three-dimensional density distribution cloud map to construct the plasma density spatial distribution matrix.

7. The method for detecting plasma density of a remote plasma source according to claim 6, wherein: Compressed sensing reconstruction is performed based on the original phase matrix to obtain a density distribution matrix, and three-dimensional analysis is performed based on the density distribution matrix to generate a three-dimensional density distribution cloud map. The method includes: Performing sparse constraint optimization partitioning on the area to be detected based on the original phase matrix to obtain multiple spatial grids; Mapping and reconstructing the multiple spatial grids and the original phase matrix to obtain a density distribution matrix; Performing axial stratification according to the density distribution matrix to obtain a plurality of equal-interval levels; Performing radial difference processing based on the multiple equally spaced levels to obtain multiple interpolation points; Performing three-dimensional spatial processing according to the multiple equal-interval levels and the multiple interpolation points to construct a three-dimensional density field; Visual mapping rendering is performed according to the three-dimensional density field to generate the three-dimensional density distribution cloud map.

8. The method for detecting plasma density of a remote plasma source according to claim 6, wherein: Decomposing and identifying the plasma density spatial distribution matrix to determine an abnormal density region, and triggering a power partition control instruction of a remote plasma source according to the abnormal density region, the method includes: Performing tensor decomposition based on the plasma density spatial distribution matrix, marking anomalies according to the decomposition results, and extracting abnormal feature components; Performing a quantitative analysis of density fluctuations based on the abnormal characteristic components and setting a density fluctuation index; determining a density anomaly region according to the density fluctuation index, wherein the density anomaly region includes an abnormal pattern; Perform multi-objective optimization according to the abnormal pattern to generate power partition control instructions; The power partition control instruction is used to drive the remote plasma source to perform partition power adjustment.

9. The method for detecting plasma density of a remote plasma source according to claim 8, wherein: Performing tensor decomposition based on the plasma density spatial distribution matrix, marking anomalies according to the decomposition results, and extracting abnormal feature components, the method comprising: Expanding the plasma density spatial distribution matrix into multi-layer two-dimensional slices based on the three-dimensional density distribution cloud map; Traversing the multiple layers of two-dimensional slices and performing calculations in a row direction to obtain a row-direction feature vector; Traversing the multiple layers of two-dimensional slices and performing calculations in a column direction to obtain column-direction feature vectors; Merging the row-direction eigenvector and the column-direction eigenvector to construct a core tensor; Traversing the plasma density spatial distribution matrix according to the core tensor, marking abnormal areas, and reconstructing the three-dimensional density distribution cloud map based on the abnormal areas to obtain an abnormal spatial distribution map; Scanning is performed based on the abnormal spatial distribution map, and the scanning results are aggregated and analyzed to extract the abnormal characteristic components.

10. A plasma density detection system for a remote plasma source, characterized in that: The system is used to implement the plasma density detection method of a remote plasma source according to any one of claims 1 to 9, and the system comprises: The scattered signal acquisition module is used to control the laser emitting device to emit a detection laser beam through the plasma area to be measured in the plasma transmission channel, and collect the scattered signal of the plasma through the laser receiving device; a phase variation extraction module, configured to perform phase variation analysis and identification based on the scattered signal, extract an interference light signal, perform digital signal processing on the interference light signal, and extract a phase variation; A density spatial distribution matrix construction module is used to detect the density of the plasma according to the phase change amount and construct a plasma density spatial distribution matrix; The power control instruction triggering module is used to decompose and identify the plasma density spatial distribution matrix, determine the density abnormality area, and trigger the power partition control instruction of the remote plasma source according to the density abnormality area.

Images