A method and system for detecting silicon carbide micro-powder impurities

Abstract

This invention belongs to the field of materials physicochemical analysis and testing technology, and relates to a method and system for detecting impurities in silicon carbide micropowder. The method includes: simultaneously acquiring the backscattered electron intensity of the scanning point and the characteristic ray intensity of the matrix element and the impurity element to be tested; determining the matrix absorption distortion ratio based on the ratio of the characteristic ray intensity of the matrix silicon element to that of the matrix carbon element; constructing an anti-absorption physical compensation multiplier and a geometric shielding reciprocal multiplier based on the matrix absorption distortion ratio and the backscattered electron intensity, and compensating for the characteristic ray intensity of the impurity element to be tested to obtain the morphology compensation excitation intensity; constructing a dynamic judgment threshold using the matrix absorption distortion ratio, marking the scanning points with the morphology compensation excitation intensity greater than the threshold as impurity feature points, and outputting a two-dimensional distribution map of microscopic impurities; it can eliminate the interference of microscopic physical morphology fluctuations on the detection of chemical components, overcome the missed detection of deep pits and the misjudgment of edge scattering, and output the true impurity distribution state.

Description

Technical Field

[0001] This invention belongs to the field of materials physicochemical analysis technology, and relates to a method and system for detecting impurities in silicon carbide micro powder. It is used for purity evaluation of silicon carbide micro powder under complex micromorphology and can overcome the missed detection and misjudgment of impurities caused by physical topographic undulations. Background Technology

[0002] With the rapid development of wide-bandgap semiconductors and high-end grinding manufacturing, silicon carbide micropowder, as a core raw material, directly determines the electrical performance and mechanical strength of the final device. During the preparation and purification process of silicon carbide micropowder, metal wear debris such as iron and aluminum is easily introduced. If these microscale inclusions remain in the product, they will become serious sources of electrical breakdown or stress concentration points. Therefore, accurate detection and evaluation of impurity distribution in silicon carbide micropowder has become a crucial step in ensuring product quality.

[0003] Chinese patent application CN101718720A discloses a method for detecting impurity content in silicon carbide using X-ray fluorescence spectroscopy. This method involves grinding and pressing a silicon carbide sample, then measuring the intensity of characteristic spectral lines using an X-ray fluorescence spectrometer and performing matrix correction to achieve semi-quantitative analysis of impurity element content. However, the method described in this patent application falls under the typical category of macroscopic statistical analysis, obtaining only the average content within the detection area. It cannot provide information on the physical morphology and spatial distribution of individual impurity particles at the microscale. Furthermore, when processing non-uniform powder surfaces, the global matrix correction model struggles to account for complex morphological distortions in micro-regions, resulting in insufficient detection limits for trace impurities.

[0004] Chinese patent document CN103018143B discloses a method for detecting the particle size of silicon carbide micropowder. This method is mainly based on the principle of laser diffraction, and evaluates the physical size distribution of micropowder by establishing a functional model of light-blocking degree and particle size distribution. However, this patent document only focuses on the statistical analysis of the physical geometric parameters of the powder, ignoring the chemical composition properties of the particles themselves. It cannot distinguish between silicon carbide matrix particles and foreign metal impurity particles that show similarity on the particle size distribution curve, resulting in a significant technical blind spot in application scenarios involving chemical purity assessment.

[0005] Chinese patent application CN121830440A discloses a method for detecting the roundness of silicon carbide micropowder. This method utilizes microscopic imaging technology to obtain the contours of powder particles and calculates the ratio of area to perimeter to assess their geometric regularity. However, the preprocessing and recognition logic of this patent application is mainly based on the geometric features of two-dimensional contours. When processing silicon carbide micropowder with high reflectivity or complex shadow effects, it is prone to misjudging the shadows generated by deep pits as morphological defects. Furthermore, it lacks the ability to analyze the energy gradient of characteristic rays and cannot distinguish between matrix particles and metal impurities with similar contour features, resulting in limited recognition accuracy in complex backgrounds.

[0006] In existing technologies, although some solutions attempt to detect powder quality through X-ray spectroscopy, physical particle size measurement, or machine vision morphology assessment, these solutions generally face the technical bottleneck of not being able to effectively eliminate the interference of physical morphology on the detection of chemical components. Because the surface of silicon carbide microparticles is covered with extremely irregular cleavage planes and micro-pits, when an electron beam bombards the undulating three-dimensional surface, the excited characteristic rays encounter extremely uneven matrix self-absorption effects as they escape towards the detector. Lower-energy light element rays are severely absorbed and attenuated at deep pits due to the excessively long escape path. Most existing analytical methods use fixed signal thresholds or global static correction models, ignoring the strong interference of micro-region geometry on the excitation signal intensity. This leads to the system easily missing trace impurities at deep pits and depressions, while generating a large number of false positive signals due to geometric enhancement effects at sharp edges. This extremely high false positive and false negative rates prevent existing methods from objectively mapping the true impurity distribution, resulting in an inability to accurately assess the actual purity level of silicon carbide microparticles. Summary of the Invention

[0007] The purpose of this invention is to overcome the shortcomings of the prior art and solve the technical problems that the prior art ignores the interference of the micro-morphological undulations of the surface of silicon carbide micro powder on the characteristic X-ray signal, and that the use of a fixed judgment threshold is prone to causing impurities to be missed in deep pits and depressions and misjudged at sharp edges, thus failing to accurately map the true purity level. The invention provides a method and system for detecting impurities in silicon carbide micro powder.

[0008] To achieve the above-mentioned objective, the present invention provides a method for detecting impurities in silicon carbide micro powder, the main process of which includes the following steps: The test area of ​​the silicon carbide micropowder sample is scanned point by point, and the backscattered electron intensity, matrix silicon element characteristic X-ray intensity, matrix carbon element characteristic X-ray intensity, and the characteristic X-ray intensity of the impurity element to be tested are acquired simultaneously at each scan point. Based on the ratio of the matrix silicon element characteristic X-ray intensity to the matrix carbon element characteristic X-ray intensity, a matrix absorption distortion ratio reflecting the degree of microstructure undulation is constructed. Based on the matrix absorption distortion ratio, an anti-absorption physical compensation multiplier is constructed, and a geometric occlusion reciprocal multiplier is constructed using the backscattered electron intensity. The intensity of the characteristic X-ray of the impurity element to be tested is amplified and compensated by the anti-absorption physical compensation multiplier and the geometric occlusion reciprocal multiplier to obtain the morphology compensation excitation intensity that removes microstructure interference. Based on the distribution characteristics of the morphology compensation excitation intensity of all scan points in the test area and the matrix absorption distortion ratio, a dynamic judgment threshold is constructed. Scan points with morphology compensation excitation intensity greater than the dynamic judgment threshold are marked as impurity feature points to output a two-dimensional distribution map of microstructure impurities.

[0009] This invention synchronously acquires multidimensional excitation signals and constructs a matrix absorption distortion ratio that reflects the degree of micro-morphological undulation, transforming the hidden three-dimensional morphological distortion into a measurable physical index. Based on this, it constructs an anti-absorption physical compensation multiplier and a geometric occlusion reciprocal multiplier for inverse amplification compensation, stripping away the masking interference of matrix self-absorption and physical occlusion on the original signal and restoring the morphological compensation excitation intensity. Furthermore, it utilizes morphological distortion features to construct a dynamic judgment threshold, achieving adaptive interception of edge geometric enhancement noise and identification and protection of weak signals from deep pits. This solves the problem of impurity omission and false alarms that are easily generated by traditional static detection methods under complex morphologies, and can objectively and accurately map the true two-dimensional distribution spectrum of micro-impurities in silicon carbide micropowder.

[0010] The present invention describes the acquisition of the backscattered electron intensity, the characteristic X-ray intensity of the matrix silicon element, the characteristic X-ray intensity of the matrix carbon element, and the characteristic X-ray intensity of the impurity element to be tested at each scanning point, including: driving the high-energy electron beam of the scanning electron microscope to perform point-by-point gridding scanning of the area to be tested; and synchronously recording the backscattered electron intensity received by the backscattered electron detector, as well as the characteristic X-ray intensity of the matrix silicon element, the characteristic X-ray intensity of the matrix carbon element, and the characteristic X-ray intensity of the impurity element to be tested captured by the X-ray energy spectrometer within a preset residence time window at each scanning point.

[0011] This invention ensures strict alignment of heterogeneous physical signals captured by the backscattered electron detector and the X-ray energy spectrometer in both spatial and temporal dimensions through point-by-point gridding scanning of the high-energy electron beam and synchronous recording within a preset dwell time window. This effectively avoids signal characterization deviations that are easily generated by a single physical detection source in complex terrain, and provides complete and high-fidelity multidimensional physical data support for the subsequent accurate identification of the true chemical composition inside the material and the external geometric morphology.

[0012] The matrix absorption distortion ratio described in this invention satisfies the following expression: ;in, Indicates the matrix absorption distortion ratio at the target scan point; This indicates the intensity of the characteristic rays of the matrix silicon element at the target scanning point; Indicates the intensity of the characteristic rays of the matrix carbon element at the target scanning point; This represents the detector's background noise count.

[0013] This invention accurately constructs the matrix absorption distortion ratio by fusing the ratio of the intensity of characteristic rays of silicon to the intensity of characteristic rays of carbon in the matrix with the detector's background noise count. It effectively counteracts the common fluctuation interference of the excitation source by utilizing the physical attenuation differences of characteristic rays of the matrix at different energy levels. At the same time, it introduces the detector's background noise count as the physical cutoff line for the minimum signal, preventing the risk of algebraic overflow caused by the denominator approaching zero when the target scanning point is in an extremely deep pit and the characteristic ray signal of carbon in the matrix is ​​almost completely attenuated. This ensures the computational stability and robustness of the morphology distortion assessment model under extremely complex micro-topography.

[0014] The morphology-compensated excitation intensity described in this invention satisfies the following expression: ; Indicates the excitation intensity for morphological compensation of the target scanning point; This represents the maximum value filtering function; This indicates the intensity of the characteristic rays of the impurity element to be measured at the target scanning point; Indicates the matrix absorption distortion ratio at the target scan point; Indicates the backscattered electron intensity at the target scanning point; This indicates the detector's minimum resolution intensity; Indicates the reference backscatter intensity; Indicates the intrinsic absorption attenuation index, satisfying , This represents the physical linear absorption coefficient of the impurity element being measured. Represents the physical linear absorption coefficient of the matrix carbon element. Indicates the physical linear absorption coefficient of the matrix silicon element; Indicates the physical compensation multiplier against absorption; This represents the reciprocal multiplier of geometric occlusion.

[0015] This invention combines the physical linear absorption coefficients of the impurity element and the matrix element to deduce the anti-absorption physical compensation multiplier, and coordinates with the backscattered electron intensity after being safely truncated by the detector's minimum resolution intensity to construct the geometric shielding reciprocal multiplier. It can accurately reconstruct and multiply amplify the weak impurity characteristic X-ray signal at the bottom of the pit from the dual dimensions of the material's inherent X-ray self-absorption mechanism and spatial geometric escape shielding, while avoiding the computational collapse caused by the complete loss of signals in extreme dark areas, thereby restoring the chemical excitation enrichment level of the impurity.

[0016] The reference backscattering intensity described in this invention is the 95th percentile of the backscattered electron intensity distribution at all scan points in the entire scan data matrix.

[0017] The dynamic determination threshold described in this invention satisfies the expression: ;in, This represents the dynamic threshold for determining the target scan point; This represents the global mean of the morphology-compensated excitation intensity; Indicates the matrix absorption distortion ratio at the target scan point; This indicates the extreme value absorption distortion ratio.

[0018] This invention adaptively reconstructs the dynamic judgment threshold by introducing the extreme absorption distortion ratio and the global mean of the morphology compensation excitation intensity. By utilizing the inverse proportional mapping relationship between the local matrix absorption distortion degree and the extreme state of the deepest pit, the judgment threshold can be adaptively raised by multiples in the edge protrusion region where electron surges are easily generated, so as to strictly intercept the false component peaks of edge geometric enhancement excitation. In the deep pit occlusion region, it automatically and smoothly falls back to the global average threshold. Under the premise of ensuring the effective detection of real weak signals after deep pit compensation, it achieves efficient dynamic filtering of spatial complex background scattering noise.

[0019] The extreme absorption distortion ratio described in this invention is the largest matrix absorption distortion ratio in the entire scan data matrix.

[0020] The energy spectrum count of the metal inclusions containing iron or aluminum is described in this invention, based on the characteristic X-ray intensity of the impurity element to be tested.

[0021] The two-dimensional distribution map of microscopic impurities described in this invention includes: summarizing all scan points marked as impurity feature points in the test area, and generating and outputting a two-dimensional distribution map of microscopic impurities in silicon carbide micropowder based on the spatial coordinates of each impurity feature point.

[0022] This invention also provides a silicon carbide micro powder impurity detection system, the main structure of which includes a processor and a memory. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, the aforementioned silicon carbide micro powder impurity detection method is implemented. By adopting the above technical solution, the aforementioned silicon carbide micro powder impurity detection method is generated into a computer program and stored in the memory for loading and execution by the processor. Thus, a terminal device is made based on the memory and the processor, which is convenient to use.

[0023] Compared with existing technologies, this invention has at least the following beneficial effects: By simultaneously acquiring the backscattered electron intensity and multi-element characteristic ray intensity of silicon carbide micropowder samples, and combining the physical attenuation differences of different element rays in the matrix to construct the matrix absorption distortion ratio, the microscopic three-dimensional morphological features are transformed into measurable physical indicators. Then, based on this indicator and the attenuation law of backscattered electron intensity, an anti-absorption physical compensation multiplier and a geometric occlusion reciprocal multiplier are constructed to amplify and compensate the characteristic ray intensity of the impurity element to be tested, thereby obtaining a morphological compensation excitation intensity that excludes physical terrain interference. The dynamic judgment threshold is adaptively reconstructed using the degree of morphological distortion. While ensuring the effective detection of weak real impurity signals in the deep pit and depression areas, false scattering signals caused by the geometric enhancement effect of sharp edges are filtered out. This effectively overcomes the defects of impurity omission and misjudgment caused by surface undulation interference of fixed detection thresholds, thereby enabling the objective and high-precision output of the true two-dimensional distribution map of microscopic impurities in silicon carbide micropowder. Attached Figure Description

[0024] Figure 1 This is a schematic flowchart of the method for detecting impurities in silicon carbide micropowder involved in this invention.

[0025] Figure 2 This is a schematic diagram of the backscattered electron intensity and matrix characteristic ray intensity distribution involved in the present invention.

[0026] Figure 3 This is a schematic diagram of the matrix absorption distortion ratio distribution involved in the present invention.

[0027] Figure 4 This is a schematic diagram comparing the intensity of characteristic rays of the impurity element to be tested with the intensity of morphology-compensated excitation, as per the present invention.

[0028] Figure 5 This is a schematic diagram illustrating the changes in morphology compensation excitation intensity and dynamic determination threshold involved in the present invention. Detailed Implementation

[0029] The technical solution of the present invention will now be clearly and completely described in conjunction with the embodiments and accompanying drawings.

[0030] Example 1: This embodiment discloses a method for detecting impurities in silicon carbide micro powder, referring to... Figure 1 This includes steps S1-S4: S1. Obtain the backscattered electron intensity and characteristic ray intensity at the scanning point.

[0031] It should be noted that due to the extremely irregular cleavage surfaces and micro-pits on the surface of silicon carbide microparticles, the physical escape paths faced by the rays excited by the electron beam at different stagnation points vary greatly. A single characteristic ray signal cannot distinguish whether the decrease in intensity is due to a reduction in the actual impurity content or physical attenuation caused by morphological geometric occlusion, resulting in a very high false positive rate in impurity detection. Therefore, this invention simultaneously acquires backscattered electron sequences and characteristic ray intensity sequences. The yield of backscattered electrons is highly sensitive to surface tilt angle and geometric solid angle, while the characteristic rays directly reflect the elemental chemical composition of the micro-region. Simultaneously acquiring these two signal sequences with distinctly different excitation mechanisms can provide complete multi-dimensional physical data support for subsequent decoupling of the material's internal chemical composition and external geometric morphology. Through high-frequency and stable electron beam stationary scanning, it can be ensured that the excitation signal at each micro-stagnation point has a true physical correspondence, avoiding the uncertainty error of a single physical detection source in complex terrain.

[0032] Specifically, the high-energy electron beam driving the scanning electron microscope performs point-by-point gridding scanning on the test area of ​​the silicon carbide micropowder sample to obtain a scanning data matrix containing several scanning points.

[0033] Within a preset residence time window of milliseconds at each scanning point, the intensity of backscattered electrons received by the backscattered electron detector and the intensity of characteristic rays of various elements captured by the X-ray spectrometer are recorded simultaneously.

[0034] The data from all scanning points are compiled and integrated to construct a backscattered electron sequence, a matrix silicon element characteristic ray intensity sequence, a matrix carbon element characteristic ray intensity sequence, and a test impurity element characteristic ray intensity sequence covering the entire test area. The test impurity element characteristic ray intensity sequence corresponds to the energy spectrum count of metal inclusions introduced by machining, such as iron or aluminum elements.

[0035] For example, Figure 2 The distribution of matrix silicon characteristic rays, matrix carbon characteristic rays, and backscattered electron intensity obtained by scanning is shown in three typical physical terrain regions: flat normal region, deep pit shading region, and edge protrusion region. In the deep pit shading region, the matrix carbon characteristic rays show attenuation due to their low energy, while the backscattered electron intensity also shows a significant low position due to the geometric shading effect.

[0036] S2. Determine the matrix absorption distortion ratio based on the characteristic ray intensity of the matrix.

[0037] It should be noted that, based on a simplified one-dimensional X-ray absorption model and according to the Beer-Lambert Law, the transmission attenuation of X-rays in a homogeneous medium strictly depends on the photon energy and the equivalent absorption path length. Silicon carbide, as a compound with a strictly fixed stoichiometric ratio, has an absolutely stable macroscopic composition. When an electron beam excites characteristic rays in a deep pit region, the rays must traverse a longer matrix path to escape to the surface. During this absorption attenuation process, the lower-energy characteristic rays of the matrix carbon element experience a much greater physical attenuation than the higher-energy characteristic rays of the matrix silicon element, resulting in severe optical distortion in the ratio of light to heavy elements received by the detector. Therefore, this invention constructs a matrix absorption distortion ratio by comparing the attenuation differences of different element rays in the matrix. The aim is to transform the macroscopically invisible three-dimensional microscopic morphological distortion into a mathematical index that can map physical depth without relying on external auxiliary mapping.

[0038] Specifically, the energy attenuation of X-rays as they propagate through matter strictly follows Beer-Lambert's law: ,in, The apparent intensity of radiation received by the detector. The initial excitation intensity generated by high-energy electron beam excitation. This represents the linear absorption coefficient of the element's radiation in the matrix. Let be the physical depth at which the ray escapes. Then, within the silicon carbide microparticle matrix, the intensity of the characteristic ray from silicon in the matrix and the intensity of the characteristic ray from carbon in the matrix, received by the detector, respectively satisfy... and ,in, and These represent the intensity of the characteristic rays of silicon and carbon elements in the matrix received by the detector at the target scanning point, respectively. and These represent the actual initial excitation intensities of matrix silicon and matrix carbon generated by high-energy electron beam excitation, respectively. and These represent the physical linear absorption coefficients of the matrix silicon and the matrix carbon, respectively. This indicates the actual physical depth at which the characteristic ray escapes from the detector; This represents an exponential function with the natural constant as its base.

[0039] Because carbon is a light element, its characteristic rays have low energy and are easily absorbed by the matrix itself; its linear absorption coefficient is much greater than that of silicon. The expression for the characteristic radiation intensity of silicon in the matrix is ​​divided by the expression for the characteristic radiation intensity of carbon in the matrix to compensate for the common fluctuation interference of the excitation source, thus constructing the matrix radiation ratio formula: In order to extract the true physical depth hidden by complex terrain. By performing the inverse natural logarithm operation on both sides of the matrix ray proportionality equation to eliminate the exponential term constraint, we obtain: Therefore, the true physical depth of the microscopic pit can be deduced. satisfy: .

[0040] From true physical depth As can be seen from the expression, the core variable determining the depth characteristics of micro-pits is the intensity ratio of the characteristic rays of light and heavy elements. Due to the physical constant term and initial strength ratio term In the same batch of material scans, these are all constants; they only perform global linear scaling and translation of the true depth and do not affect the relative spatial differences in terrain distortion between individual scan points. Therefore, this invention directly removes the global constant terms. and Extract the core proportional term driving depth changes As a benchmark for measuring local distortion, and to address the risk of algebraic overflow where the denominator approaches zero due to extreme carbon signal attenuation caused by extremely deep pits, detector background noise counting is introduced. As a physical cutoff line, a matrix absorption distortion ratio for measuring topographic relief is constructed, which satisfies the expression:

[0041] in, Indicates the matrix absorption distortion ratio at the target scan point; and These represent the intensity of the characteristic rays of silicon and carbon elements in the matrix received by the detector at the target scanning point, respectively. This represents the detector's background noise count.

[0042] Based on the attenuation law revealed by Beer-Lambert's law, when the target scanning point is in a deep-seated terrain, the intensity of the characteristic rays of the carbon element in the matrix increases due to the increased penetration thickness. The drop rate is much higher than the characteristic ray intensity of the matrix silicon. , making Spontaneous increase. When the characteristic ray intensity of the matrix carbon element... The smaller the value, the higher the calculated matrix absorption distortion ratio. The larger the matrix absorption distortion ratio, the higher the matrix absorption distortion ratio. It can reflect the overall distortion of physical signals when a local terrain evolves from flat to deep depression.

[0043] Counting the detector's background noise Because the energy spectrum detector inevitably generates a weak thermionic emission background signal during operation due to environmental thermal excitation, in order to prevent the target scanning point from being in an extremely deep pit and thus affecting the intensity of the characteristic rays of the carbon element in the matrix, When the noise level decays to 0, it causes a mathematical error where the denominator is 0. This invention introduces detector background noise counting. This serves as the physical cutoff line for extremely small signals. By performing pre-scan statistics on the scanning area without samples, the background dead time pulse count of the X-ray spectrometer in the absence of electron beam bombardment is obtained and used as the detector background noise count. The empirical range is 5 to 15 counts per second. In other embodiments, the implementer can also directly consult the baseline calibration test report of the microscope equipment and count the detector background noise based on the equipment background noise level recorded in the report. Perform numerical calibration.

[0044] For example, Figure 3 The diagram shows the distribution of matrix absorption distortion ratio, reflecting the variation of matrix absorption distortion ratio with scanning position. It can be seen that the matrix absorption distortion ratio produces a significant numerical peak in the deep pit shading area, which can reflect the degree of micro-topographic undulation.

[0045] S3. Determine the morphology compensation excitation intensity.

[0046] It should be noted that the impurity elements to be tested, distributed at the bottom of the microscopic pit, also follow the physical attenuation law of X-ray self-absorption. The characteristic rays they excite will be severely intercepted and absorbed by the surrounding matrix. Directly using the apparent residual intensity received by the detector to determine the composition will inevitably lead to the omission of trace impurities. In order to reproduce the true chemical enrichment level of the impurities to be tested, this invention uses the matrix self-absorption attenuation law of X-rays under complex microscopic morphology to compensate for the original signal. Based on the signal intensity restoration principle, the intrinsic electromagnetic radiation energy intercepted by the three-dimensional morphology is reversed at the data level to restore the absolute chemical excitation intensity after removing the morphological interference.

[0047] Specifically, the intensity of the characteristic rays of the impurity to be tested also follows the Beer-Lambert law: ,in, This indicates the intensity of the characteristic rays of the impurity element to be measured at the target scanning point; This indicates the true initial excitation intensity of the impurity being tested; Indicates the physical linear absorption coefficient of impurity elements; This represents the true physical depth at which the characteristic ray escapes towards the detector. (The true physical depth is then...) expression Substituting into the relation, we get: ,in, and These represent the intensity of the characteristic rays of silicon and carbon elements in the matrix received by the detector at the target scanning point, respectively. and These represent the actual initial excitation intensities of matrix silicon and matrix carbon generated by high-energy electron beam excitation, respectively. and These represent the physical linear absorption coefficients of the matrix silicon and the matrix carbon, respectively. This represents a logarithmic function with the natural constant as its base.

[0048] To characterize the inherent absorption differences of materials to different ray energies, an intrinsic absorption attenuation index is defined. Due to the constant product term The product term, consisting solely of the initial excitation intensity and intrinsic absorption attenuation index of the matrix elements, is an absolute global constant within the same scanning field of view. In the subsequent feature determination stage, the dynamic determination threshold is derived from the global mean of the excitation intensity compensated for by the topography of all scan points. This means that this global constant product term simultaneously and proportionally scales both the target signal and the determination threshold. According to the principle of proportional scaling propagation of inequalities, multiplying both ends by this positive constant will never change the final physical enrichment determination result. Therefore, to optimize the underlying computational load and improve real-time compensation efficiency, this constant product term is directly stripped and ignored in micro-pixel-level operations. Ultimately based on The physical compensation multiplier for the resistance to absorption of impurities is derived as follows: In conjunction with step S2, the detector's background noise count is introduced to prevent algebraic overflow caused by the denominator approaching zero. Matrix absorption distortion ratio definition Regarding this core proportion item Physical safety truncation replacement was performed, and the physical compensation multiplier for the impurity's resistance to absorption was ultimately determined to be... .

[0049] Furthermore, regarding the geometric obstruction effect caused by the deep pit terrain, based on the detector solid angle collection model, the escape reception rate of the physical signal has a strictly proportional linear relationship with the effective escape solid angle: ,in, This indicates the intensity of backscattered electrons at the deep pit of the target scanning point; This represents the finite escape solid angle of a deep pit after it is geometrically and physically obstructed. This represents the instrument response constant inside the microscope. The 95th percentile of the backscattered electron intensity distribution at all scan points in the entire scan data matrix is ​​extracted and used as the reference backscattered intensity in an unobstructed flat region. Its satisfaction ,in, This represents the reference effective escape solid angle in a flat area without microscopic topographical obstruction. The geometric obstruction fraction at the bottom of the pit, i.e., the proportion of physical signal successfully transmitted, is strictly equal to... To reconstruct the true impurity excitation signal in reverse, the damaged signal is divided by the physical occlusion fraction, thus deriving the inverse multiplier of the geometric occlusion. .

[0050] In practical calculations, to prevent the complete loss of backscattered electrons in extremely dark areas from causing a computational crash with a denominator of 0, a minimum resolution intensity for the hardware detection capability is set. Perform a safety cutoff. Integrate anti-absorption physical compensation multipliers. With geometric occlusion reciprocal multiplier The morphology compensation excitation intensity after removing morphology interference is reconstructed, which satisfies the expression:

[0051] in, Indicates the excitation intensity for morphological compensation of the target scanning point; This represents the maximum value filtering function; This indicates the intensity of the characteristic rays of the impurity element to be measured at the target scanning point; Indicates the matrix absorption distortion ratio at the target scan point; Indicates the backscattered electron intensity at the target scanning point; This indicates the detector's minimum resolution intensity; Indicates the reference backscatter intensity; Indicates the intrinsic absorption attenuation index, satisfying ,in, This represents the physical linear absorption coefficient of the impurity element being measured. Represents the physical linear absorption coefficient of the matrix carbon element. The physical linear absorption coefficient of the matrix silicon element is used to represent the probability of energy attenuation when a photon of a specific energy penetrates a specific material. It is an objectively inherent physical constant, and its specific value is obtained by the implementer by strictly referring to a standard X-ray mass attenuation coefficient physical database, corresponding to the atomic number of the impurity and the matrix element, as well as the energy level of the excited characteristic ray. In other embodiments, the implementer can also obtain the physical linear absorption coefficients of the above elements by performing offline calibration experiments on a standard silicon carbide micropowder reference sample of known fixed thickness to measure and calculate the X-ray transmission attenuation rate.

[0052] When the matrix absorbs distortion ratio High and backscattered electron intensity A low value indicates severe physical geometric occlusion in the region, and at this point, the matrix absorption distortion ratio increases. The increase in the overall resistance to absorption physical compensation multiplier Increase; at the same time, the backscattered electron intensity The reduction makes Significantly increased, In coordination, based on the physical compensation principle, the intensity of the characteristic rays of the weak impurity elements to be measured is determined. Perform appropriate multiplicative amplification. When the target scanning point is on flat terrain, the matrix absorption distortion ratio... Approaching 0, backscattered electron intensity Approaching the reference backscatter intensity The compensation coefficient automatically approaches 1, which conforms to the physical norm that there is no abnormal attenuation of rays in flat areas, and prevents the original signal from being over-amplified.

[0053] For detectors with extremely low resolution Because backscatter probes may completely lose backscattered electron signals when facing the microscopic dark areas of deep pits, measures are needed to prevent backscattered electron intensity from being completely lost. A value of 0 causes a multiplication and division calculation crash; therefore, this invention sets a physical lower limit for the hardware's detection capability. The detector's minimum resolution intensity The empirical range is 0.01 to 0.05 counts per second. In this embodiment, 0.01 counts per second is selected. This value is based on the limiting resolution background detection capability of the high-voltage electron beam under the lowest signal-to-noise ratio condition, which can be obtained by pre-scanning a blank sample to statistically extract the system background fluctuations. In other embodiments, the implementer can set the detector's minimum resolution intensity according to the factory settings of the signal-to-noise ratio parameters of the electron optical system. The settings.

[0054] For example, Figure 4 This is a schematic diagram comparing the intensity of the characteristic rays of the impurity element to be tested with the excitation intensity of the morphology-compensated excitation intensity. It shows the reconstruction effect of the morphology-compensated excitation intensity on the intensity of the characteristic rays of the impurity element to be tested. In the deep pit shielding area, the original weak impurity signal is restored to the excitation intensity close to the true enrichment level after inverse compensation, thus eliminating the interception of the true impurity response by physical masking.

[0055] S4. Construct a dynamic judgment threshold to screen and mark impurity feature points.

[0056] It should be noted that, based on the edge geometry enhancement effect in solid-state physics, irregular scattering is easily generated when the electron beam scans the edge or sharp protrusion of the silicon carbide micropowder surface. This induces chemically meaningless physical spikes in the local X-ray count, causing a large number of false positives in such areas by the traditional static threshold. Since the true element enrichment distribution is spatially independent, while the false signal spikes are strictly accompanied by local morphological features, this invention introduces the degree of local topographic distortion into the digital signal filtering mechanism. By constructing an adaptive threshold function based on spatial physical distortion characteristics, the judgment threshold is forcibly raised in geometric protrusion areas that are prone to electron surges to intercept scattering noise, while the threshold is allowed to fall back in deep pit areas to ensure that the weak true signal after compensation is not mistakenly rejected, thereby achieving adaptive spatial filtering of complex background noise.

[0057] Specifically, based on the geometric scattering model in solid-state physics, the intensity of anomalous scattering signals from electron beams at sharp edges or protrusions shows a significant negative correlation with the effective self-absorption depth of the substrate at that location. This means that the smaller the absorption distortion of the substrate, i.e., the closer it is to an unobstructed sharp edge, the stronger the false scattering signal it generates. To intercept this false signal, the required dynamic decision threshold must be increased proportionally according to this inverse relationship. Conversely, at extremely deep pits without edge scattering interference, the threshold should revert to the true global average background level. Therefore, the largest substrate absorption distortion ratio in the entire scan data matrix is ​​extracted and taken as the extreme absorption distortion ratio. This is used to characterize the state of the deepest pit, and extracts the arithmetic mean of the morphology compensation excitation intensity of all scan points in the entire scan data matrix. This serves as the baseline average threshold. Based on the strictly inverse proportional mapping relationship between the above-mentioned anomalous scattering signal intensity and the effective self-absorption depth, a dynamic judgment threshold expression for the target scanning point is constructed:

[0058] in, This represents the dynamic threshold for determining the target scan point; This represents the global mean of the morphology-compensated excitation intensity; Indicates the matrix absorption distortion ratio at the target scan point; This represents the extreme absorption distortion ratio. When the matrix absorption distortion ratio... The smaller the value, that is, when the target scanning point is located in an edge protrusion region that is prone to electron scattering. This enables dynamic determination of the threshold. This is magnified many times over, thus filtering out spurious component spikes excited by edge geometry enhancement effects; when the matrix absorption distortion ratio is... The matrix absorption distortion ratio is relatively large, meaning that when the target scanning point is located in a deep pit area of ​​the terrain without edge scattering effects. Approaching the extreme value absorption distortion ratio ,at this time Approaching 1, dynamically determining the threshold. Automatically returns to the global mean of morphology-compensated excitation intensity This ensures that the real impurity signal from the deep pit, obtained through reverse physical reconstruction, can successfully pass the threshold test.

[0059] Furthermore, the excitation intensity of the morphology compensation at each target scanning point is compared sequentially. With dynamic judgment threshold The numerical relationship is as follows: the excitation intensity responds to the topography compensation of the target scanning point. Greater than the corresponding dynamic judgment threshold The target scan point is marked as an impurity feature point.

[0060] For example, Figure 5 The diagram illustrates the changes in excitation intensity and dynamic judgment threshold related to topographic compensation, demonstrating the distribution pattern of the dynamic judgment threshold as the degree of topographic distortion changes. In the edge protrusion area, influenced by the edge geometric enhancement effect, the dynamic judgment threshold adaptively shifts upward to form a high-energy barrier, rigorously filtering out false signals. In the deep pit occlusion area, the dynamic judgment threshold adaptively falls back, providing sufficient detection space for the weak real signals after compensation.

[0061] Finally, all marked impurity feature points in the test area are summarized, and a two-dimensional distribution map of microscopic impurities in silicon carbide micro powder is output, thus completing high-precision impurity detection.

[0062] This embodiment also discloses a silicon carbide micro powder impurity detection system, the main structure of which includes a processor and a memory. The memory stores computer program instructions. When the computer program instructions are executed by the processor, a silicon carbide micro powder impurity detection method according to the present invention is implemented. The system also includes other components well known to those skilled in the art, such as a communication bus and a communication interface. Their settings and functions are known in the art.

Claims

1. A method for detecting impurities in silicon carbide micro powder, characterized in that, include: The test area of ​​the silicon carbide micro powder sample is scanned point by point, and the backscattered electron intensity, the characteristic radiation intensity of the matrix silicon element, the characteristic radiation intensity of the matrix carbon element, and the characteristic radiation intensity of the impurity element to be tested are acquired simultaneously at each scan point. Based on the ratio of the characteristic ray intensity of silicon to the characteristic ray intensity of carbon in the matrix, a matrix absorption distortion ratio reflecting the degree of micromorphological undulation is constructed. Based on the matrix absorption distortion ratio, an anti-absorption physical compensation multiplier is constructed, and a geometric shielding reciprocal multiplier is constructed using the backscattered electron intensity. The intensity of the characteristic rays of the impurity element to be tested is amplified and compensated by the anti-absorption physical compensation multiplier and the geometric shielding reciprocal multiplier, so as to obtain the morphology compensation excitation intensity that removes micromorphological interference. Based on the distribution characteristics of the morphology compensation excitation intensity of all scanning points in the test area and the matrix absorption distortion ratio, a dynamic judgment threshold is constructed. Scanning points with morphology compensation excitation intensity greater than the dynamic judgment threshold are marked as impurity feature points to output a two-dimensional distribution map of micro-impurities. The matrix absorption distortion ratio satisfies the expression: ； wherein, represents a base absorption distortion ratio of the target scanning point; represents a base silicon element characteristic ray intensity of the target scanning point; represents a base carbon element characteristic ray intensity of the target scanning point; represents a detector background noise count; The morphology-compensated excitation intensity satisfies the expression: ； a profile compensated excitation intensity representing a target scanning point; a maximum screening function representing; a characteristic X-ray intensity of a target scanning point representing a target scanning point; a matrix absorption distortion ratio of a target scanning point representing; a backscattered electron intensity of a target scanning point representing; a detector minimum resolution intensity representing; a reference backscattered intensity representing; an intrinsic absorption attenuation index representing, satisfying , a physical linear absorption coefficient of a target impurity element representing, a physical linear absorption coefficient of a matrix carbon element representing, a physical linear absorption coefficient of a matrix silicon element representing; an anti-absorption physical compensation multiplier representing; a geometric blocking inverse multiplier representing; The dynamic threshold satisfies the expression: ； wherein, represents a dynamic decision threshold for a target scan point; represents a topography-compensated excitation intensity global mean; represents a bulk absorption distortion ratio for a target scan point; represents an extreme value absorption distortion ratio.

2. The method of claim 1, wherein the method is characterized by: The acquisition of the backscattered electron intensity, the characteristic X-ray intensity of the matrix silicon element, the characteristic X-ray intensity of the matrix carbon element, and the characteristic X-ray intensity of the impurity element to be measured at each scanning point includes: The high-energy electron beam driving the scanning electron microscope performs point-by-point gridding scanning of the area to be tested; within the preset residence time window of the electron beam at each scanning point, the intensity of backscattered electrons received by the backscattered electron detector, as well as the intensity of characteristic rays of matrix silicon, matrix carbon, and impurity elements captured by the X-ray spectrometer are recorded simultaneously.

3. The method of claim 1, wherein the step of detecting the impurities of the silicon carbide micropowder is characterized by, The reference backscatter intensity is the 95th percentile of the backscattered electron intensity distribution at all scan points in the entire scan data matrix.

4. The method of claim 1, wherein the SiC micropowder is a SiC micropowder for a semiconductor. The extreme value absorption distortion ratio is the largest matrix absorption distortion ratio in the entire scan data matrix.

5. The method for detecting impurities in silicon carbide micro powder according to claim 1, characterized in that, The intensity of the characteristic rays of the impurity element to be tested corresponds to the energy spectrum count of the metal inclusions of iron or aluminum.

6. The method for detecting impurities in silicon carbide micro powder according to claim 1, characterized in that, The output two-dimensional distribution map of microscopic impurities includes: The scan points marked as impurity feature points within the test area are summarized, and a two-dimensional distribution map of microscopic impurities in silicon carbide micro powder is generated and output based on the spatial coordinates of each impurity feature point.

7. A silicon carbide micro powder impurity detection system, characterized in that, include: A processor and a memory, the memory storing computer program instructions that, when executed by the processor, implement a method for detecting impurities in silicon carbide micropowder according to any one of claims 1-6.

Images