Gallium oxide substrate surface grinding morphology prediction method and system

Abstract

The invention discloses a gallium oxide substrate surface grinding morphology prediction method and system, and relates to the technical field of nondestructive testing, and the method comprises the steps: collecting a surface fluorescence distribution image, matching the surface fluorescence distribution image with a pre-constructed fluorescence-crack mapping database, and generating a two-dimensional crack depth distribution cloud picture; marking a risk area coordinate set in the two-dimensional crack depth distribution cloud picture based on a preset crack depth threshold value; inputting the risk area coordinate set into a Bessel beam chromatography scanner for multi-angle transmission type scanning, synchronously collecting scattering images generated by scanning, and fusing the scattering images to generate a subsurface damage three-dimensional model; and performing data registration and weight superposition on the two-dimensional crack depth distribution cloud picture and the subsurface damage three-dimensional model, and outputting a morphology prediction report and grinding process optimization parameters. The subsurface damage three-dimensional model is constructed through the Bessel beam tomography scanning and multi-angle scattering image fusion technology, and high-resolution visual reconstruction of microcracks and damage distribution in the gallium oxide substrate is achieved.

Description

Technical Field

[0001] This invention relates to the field of nondestructive testing technology, and in particular to a method and system for predicting the surface morphology of gallium oxide substrates. Background Technology

[0002] In the field of semiconductor material processing, gallium oxide (Ga2O3) exhibits broad application prospects in high-power electronic devices and deep-ultraviolet detectors due to its excellent physicochemical stability and ultra-wide bandgap characteristics. With the continuous improvement of device performance requirements, higher standards are being set for controlling the surface quality of gallium oxide substrates. Currently, the industry commonly uses mechanical polishing combined with optical or electron microscopy for surface morphology evaluation. However, traditional detection methods such as optical microscopy and scanning electron microscopy (SEM) mainly provide two-dimensional surface image information, making it difficult to effectively identify crack depth and subsurface damage. To improve detection efficiency and accuracy, some studies have attempted to introduce image processing-based automatic recognition technology to extract surface defect features for morphology classification and trend prediction.

[0003] Although previous studies have attempted to use laser confocal microscopy to obtain structural information of material surfaces within a certain depth range, its application in wide-bandgap materials such as gallium oxide still faces challenges: existing techniques have not effectively solved the problem of modeling the correlation between surface crack depth and subsurface damage; fluorescence imaging and laser confocal scanning techniques are mainly used to address the difficulty of obtaining surface and subsurface structural information of wide-bandgap materials such as gallium oxide. This method attempts to use a light source of a specific wavelength to excite the sample surface to capture fluorescence signals and construct images of the surface and shallow internal structure. However, due to the special optical properties of gallium oxide, these techniques face challenges in practical applications, particularly in achieving high-resolution, accurate characterization of deep internal damage. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a method for predicting the surface morphology of gallium oxide substrates to solve the problem of modeling the correlation between surface crack depth and subsurface damage.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides a method for predicting the surface morphology of a gallium oxide substrate, comprising,

[0008] A gallium oxide substrate with europium-based fluorescent nanoparticles embedded on its surface was obtained;

[0009] A gallium oxide substrate is irradiated with ultraviolet light, and surface fluorescence distribution images are acquired and matched with a pre-constructed fluorescence-crack mapping database to generate a two-dimensional crack depth distribution cloud map; based on a preset crack depth threshold, the coordinate set of risk areas is marked in the two-dimensional crack depth distribution cloud map;

[0010] The coordinate set of the risk area is input into the Bessel beam tomography scanner for multi-angle penetration scanning, and the scattered images generated by the scanning are acquired simultaneously. The scattered images are then fused to generate a three-dimensional model of subsurface damage.

[0011] The two-dimensional crack depth distribution cloud map and the three-dimensional subsurface damage model are registered and weighted to output a morphology prediction report and grinding process optimization parameters.

[0012] As a preferred embodiment of the gallium oxide substrate surface grinding morphology prediction method of the present invention, the step of acquiring surface fluorescence distribution images and matching them with a pre-constructed fluorescence-crack mapping database to generate a two-dimensional crack depth distribution cloud map is as follows.

[0013] A gallium oxide substrate was illuminated with an ultraviolet light source using a fluorescence microscope to obtain an image of the surface fluorescence distribution;

[0014] Background subtraction and grayscale normalization are performed on the surface fluorescence distribution image to generate a grayscale normalized surface fluorescence distribution image;

[0015] The region of interest of the gallium oxide substrate was selected from the gray-scale normalized surface fluorescence distribution image and divided into multiple uniformly distributed detection cells;

[0016] The average fluorescence intensity of each detection cell is extracted and summarized to form a fluorescence intensity dataset;

[0017] The fluorescence intensity dataset is compared one by one with the fluorescence intensity values ​​in the pre-constructed fluorescence-crack mapping database. The fluorescence intensity value that matches the average fluorescence intensity of each detection cell is found, and the corresponding crack depth value is extracted.

[0018] Record the average fluorescence intensity of each detection cell and the corresponding crack depth value to create a fluorescence crack correspondence table;

[0019] Based on the fluorescence crack correspondence table, a fluorescence intensity visualization algorithm is applied to the region of interest in the gallium oxide substrate to draw color-coded points and generate a color-coded point distribution.

[0020] Interpolation and smoothing are performed on the color-coded point distribution to generate a continuous two-dimensional crack depth distribution cloud map.

[0021] As a preferred embodiment of the gallium oxide substrate surface grinding morphology prediction method of the present invention, wherein: the fluorescence-crack mapping database is a database of the correspondence between fluorescence intensity and crack depth obtained by experimental measurement.

[0022] As a preferred embodiment of the gallium oxide substrate surface grinding morphology prediction method of the present invention, the step of marking the risk region coordinate set in the two-dimensional crack depth distribution cloud map based on a preset crack depth threshold is as follows:

[0023] The crack depth threshold is set according to the reliability requirements of the gallium oxide substrate processing technology;

[0024] Image enhancement techniques based on region segmentation are applied to process continuous two-dimensional crack depth distribution cloud maps, and the crack depth value of each pixel in the continuous two-dimensional crack depth distribution cloud map is compared with the crack depth threshold.

[0025] If the crack depth value of a pixel exceeds the crack depth threshold, the pixel in the continuous two-dimensional crack depth distribution cloud map is marked as a high-risk marker.

[0026] All pixels marked with high risk are aggregated to form a high-risk area;

[0027] Extract the coordinate information of all pixels in the high-risk area, and summarize the coordinate information of all pixels to generate a risk area coordinate set.

[0028] As a preferred embodiment of the gallium oxide substrate surface grinding morphology prediction method of the present invention, the step of inputting the risk region coordinate set into a Bessel beam tomography scanner for multi-angle penetration scanning is as follows.

[0029] Input the risk area coordinate set into the control terminal of the Bessel beam tomography scanner;

[0030] Based on the pixel coordinates in the risk area coordinate set, the control terminal of the Bessel beam tomography scanner adjusts the position of the transmission scanning head to align with the corresponding high-risk area on the gallium oxide substrate.

[0031] Set the initial scanning angle of the Bessel beam tomography scanner, perform a transmission scan on the high-risk area, and record the scattered light signal generated during the transmission scan.

[0032] The control terminal of the Bessel beam tomography scanner issues a rotation command to position the high-risk area at a new scanning angle.

[0033] Repeatedly perform transmissive scanning, record scattered light signals, and rotate to complete multi-angle transmissive scanning.

[0034] As a preferred embodiment of the method for predicting the surface morphology of gallium oxide substrates described in this invention, the synchronous acquisition of the scattering image refers to activating a high-precision image acquisition component to acquire the scattering image generated during each transmission scan when the Bessel beam tomography scanner performs multi-angle transmission scan.

[0035] As a preferred embodiment of the gallium oxide substrate surface grinding morphology prediction method of the present invention, wherein the fused scattering image generates a three-dimensional model of subsurface damage, as detailed below.

[0036] The acquired scattering images are paired with the corresponding penetration scanning angle values ​​to generate a scattering image and scanning angle pairing table, and the scattering image and scanning angle pairing table is converted into a scattering image angle pairing dataset.

[0037] The median filter is used to denoise all scattered images in the scattered image angle pairing dataset, histogram equalization is used to enhance contrast, and Laplacian filter is used to sharpen edges to generate a two-dimensional scattered image set.

[0038] The pixel brightness values ​​of the two-dimensional scattering image corresponding to each penetration scanning angle are extracted from the two-dimensional scattering image set and organized into a projection image data matrix;

[0039] The projected image data matrix is ​​loaded into the image reconstruction algorithm, and stereoscopic data covering high-risk areas is generated by processing with a Ram-Lak filter, back-projection operation, and voxel merging.

[0040] Based on the distribution of voxel brightness values ​​in the stereo data, a brightness threshold is set, and voxels with brightness values ​​higher than the brightness threshold in the stereo data are retained as subsurface damage areas.

[0041] Identify the boundary features of the subsurface damage region, construct the boundary surface, and generate a three-dimensional model of the subsurface damage covering the high-risk region.

[0042] Secondly, the present invention provides a system for predicting the surface morphology of gallium oxide substrates, comprising,

[0043] The acquisition module is used to acquire gallium oxide substrates with europium-based fluorescent nanoparticles embedded on their surface;

[0044] The marking module is used to irradiate the gallium oxide substrate with ultraviolet light, acquire surface fluorescence distribution images and match them with a pre-constructed fluorescence-crack mapping database to generate a two-dimensional crack depth distribution cloud map; and mark the risk area coordinate set in the two-dimensional crack depth distribution cloud map based on a preset crack depth threshold.

[0045] The scanning module is used to input the coordinate set of the risk area into the Bessel beam tomography scanner for multi-angle penetration scanning, and simultaneously acquire the scattered images generated by the scanning, and fuse the scattered images to generate a three-dimensional model of subsurface damage.

[0046] The prediction module is used to register and weight the two-dimensional crack depth distribution cloud map with the three-dimensional subsurface damage model, and output the morphology prediction report and grinding process optimization parameters.

[0047] Thirdly, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, it implements any step of the gallium oxide substrate surface grinding morphology prediction method as described in the first aspect of the present invention.

[0048] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the gallium oxide substrate surface polishing morphology prediction method as described in the first aspect of the present invention.

[0049] The beneficial effects of this invention are as follows: By constructing a three-dimensional model of subsurface damage using Bessel beam tomography and multi-angle scattering image fusion technology, high-resolution visualization and reconstruction of the microcracks and damage distribution inside gallium oxide substrates are achieved. The three-dimensional subsurface damage model not only accurately captures the crack propagation path and density characteristics in the material's depth direction, but also provides crucial geometric basis for subsequent crack propagation trend prediction and grinding process parameter optimization. Compared with traditional shallow detection methods, this significantly improves the depth and accuracy of subsurface structural integrity assessment, fills the gap in deep damage modeling in existing technologies, and enhances the comprehensiveness and reliability of morphology prediction. Attached Figure Description

[0050] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0051] Figure 1 A flowchart for a method to predict the surface morphology of gallium oxide substrates.

[0052] Figure 2 This is a schematic diagram of a gallium oxide substrate surface grinding morphology prediction system.

[0053] Figure 3 This is a flowchart for multi-angle scanning reconstruction of a Bessel beam.

[0054] Figure 4This is a flowchart for data fusion and output. Detailed Implementation

[0055] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0056] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0057] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0058] Reference Figures 1-4 As one embodiment of the present invention, this embodiment provides a method for predicting the surface morphology of a gallium oxide substrate, comprising the following steps:

[0059] S1. Obtain a gallium oxide substrate with europium-based fluorescent nanoparticles embedded on its surface.

[0060] Furthermore, europium-based fluorescent nanoparticles were added to the alumina-based polishing slurry, and the europium-based fluorescent nanoparticles were dispersed using an ultrasonic oscillator to obtain an alumina-based polishing slurry containing europium-based fluorescent nanoparticles.

[0061] An alumina-based polishing slurry containing europium-based fluorescent nanoparticles is loaded into the slurry tank of a chemical mechanical polishing (CMP) grinder, and a polyurethane polishing pad is installed to obtain a ready-to-use CMP grinder.

[0062] The gallium oxide substrate is fixed in the sample holder of a ready chemical mechanical polishing machine. The surface of the gallium oxide substrate is polished with an alumina-based polishing slurry containing europium-based fluorescent nanoparticles, so that the europium-based fluorescent nanoparticles are embedded in the surface cracks and subsurface damage layer of the gallium oxide substrate, and an intermediate sample of gallium oxide substrate with europium-based fluorescent nanoparticles embedded on the surface is obtained.

[0063] The surface of the intermediate gallium oxide substrate with europium-based fluorescent nanoparticles embedded in it was cleaned with deionized water and dried with a high-purity nitrogen spray gun to obtain a gallium oxide substrate with europium-based fluorescent nanoparticles embedded in it.

[0064] S2. Irradiate the gallium oxide substrate with ultraviolet light, collect surface fluorescence distribution images and match them with a pre-constructed fluorescence-crack mapping database to generate a two-dimensional crack depth distribution cloud map; mark the risk area coordinate set in the two-dimensional crack depth distribution cloud map based on a preset crack depth threshold.

[0065] Furthermore, the gallium oxide substrate is placed on the sample stage of a fluorescence microscope, and the fluorescence microscope is configured with an ultraviolet light source and a high-resolution charge-coupled device camera to obtain a fully configured fluorescence microscope.

[0066] Images of surface fluorescence distribution were acquired by illuminating a gallium oxide substrate with ultraviolet light using a configured fluorescence microscope.

[0067] Background subtraction is performed on the surface fluorescence distribution image to remove environmental noise interference. Specifically, a medium-range filter is applied to smooth the image to eliminate environmental noise interference, such as stray light or electronic noise, while preserving fluorescence signals from surface cracks and subsurface damage layers in the gallium oxide substrate. A region in the surface fluorescence distribution image that does not contain europium-based fluorescent nanoparticles, such as the area outside the region of interest in the gallium oxide substrate, is selected as a background reference region. The average pixel brightness value within the background reference region is obtained and used as an estimate of the background fluorescence intensity. This background fluorescence intensity estimate is then applied to adjust the surface fluorescence distribution image, generating the background-subtracted image.

[0068] The grayscale normalization process is performed on the surface fluorescence distribution image after background subtraction so that the image pixel values ​​of the surface fluorescence distribution image are linearly distributed in the range of 0 to 255.

[0069] In the grayscale-normalized surface fluorescence distribution image, the region of interest of the gallium oxide substrate is selected. A grid division tool is applied to divide the region of interest of the gallium oxide substrate into multiple uniformly distributed detection cells, such as a square grid. Each detection cell has the same size to ensure that the entire region of interest of the gallium oxide substrate is covered.

[0070] For each detection cell, extract the brightness values ​​of all pixels in that cell (brightness values ​​range from 0 to 255, derived from grayscale normalization); record the brightness values ​​of all pixels in each detection cell to form a brightness value set; count the total number of pixels in the brightness value set for each detection cell; sum the brightness values ​​of all pixels in the brightness value set; distribute the sum of brightness values ​​evenly according to the total number of pixels to generate the average fluorescence intensity of each detection cell; the expression for the average fluorescence intensity is:

[0071]

[0072] Among them, Iavg The average fluorescence intensity of each detection cell, P i To detect the brightness value of the i-th pixel in a cell, the range is 0 to 255, derived from the grayscale normalized surface fluorescence distribution image. N is the total number of pixels in the cell to be detected, and i is the index of the pixel in the cell to be detected.

[0073] The average fluorescence intensity of all detected cells is aggregated to generate a fluorescence intensity dataset.

[0074] Convert the fluorescence intensity dataset into a readable format;

[0075] Load a readable fluorescence intensity dataset and a pre-built fluorescence-crack mapping database into the image processing software; the fluorescence-crack mapping database is the correspondence data between fluorescence intensity and crack depth obtained through experimental measurements.

[0076] For each detection cell, the average fluorescence intensity is compared sequentially with the fluorescence intensity values ​​in the fluorescence-crack mapping database; the fluorescence intensity value in the fluorescence-crack mapping database that is closest to the average fluorescence intensity of each detection cell is found, and the corresponding crack depth value is extracted; the average fluorescence intensity of each detection cell and the crack depth value corresponding to the average fluorescence intensity of each detection cell are recorded to form a set containing the correspondence between the average fluorescence intensity and crack depth values ​​of all detection cells.

[0077] Create a fluorescence crack mapping table, which contains two columns: the first column records the average fluorescence intensity of each detection cell, and the second column records the crack depth value corresponding to the average fluorescence intensity of each detection cell. Iterate through the set containing the average fluorescence intensity of all detection cells and the corresponding crack depth value for each detection cell, and sequentially fill the corresponding row of the fluorescence crack mapping table with the average fluorescence intensity and corresponding crack depth value of each detection cell. Ensure that the number of rows in the fluorescence crack mapping table matches the total number of detection cells, with each detection cell occupying one row in the fluorescence crack mapping table, generating a fluorescence crack mapping table containing the correspondence between the average fluorescence intensity and crack depth values ​​of all detection cells.

[0078] A fluorescence intensity visualization algorithm is applied to plot color-coded points within the region of interest (ROI) of a gallium oxide (GaO) substrate based on a fluorescence crack mapping table. The color depth represents the different crack depths within each detection cell. Specifically: the crack depth value corresponding to the average fluorescence intensity of each detection cell is extracted from the fluorescence crack mapping table; a color mapping rule is defined to map the crack depth value range to a preset color gradient, for example, using a color gradient from light blue to dark red to represent crack depths from shallow to deep; each detection cell in the fluorescence crack mapping table is traversed, and the corresponding color value is determined according to the crack depth value of each detection cell within the color mapping rule; in the grid coordinate system of the GROI region of the GaO substrate, the determined color value is assigned as a color-coded point according to the grid position of each detection cell, generating a color-coded point distribution covering the GROI region of the GaO substrate; the color mapping rule is defined based on the range of crack depth values ​​in the fluorescence crack mapping table, mapping the crack depth values ​​to a color gradient from light blue to dark red, representing crack depths from shallow to deep.

[0079] Interpolation and smoothing operations are performed on all color-coded points to generate a continuous two-dimensional crack depth distribution cloud map covering the entire region of interest (ROI) of the gallium oxide substrate. Specifically, in the grid coordinate system of the ROI of the gallium oxide substrate, the coordinates of the center point of each detection cell and the corresponding crack depth value of each detection cell are determined. For points in the grid coordinate system without assigned crack depth values, the center points of the four nearest surrounding detection cells are selected, and the crack depth values ​​and coordinates of the four nearest surrounding detection cells are obtained. Based on the crack depth values ​​and coordinates of the four nearest surrounding detection cells, the crack depth value of the horizontal midpoint is generated by linearly interpolating the crack depth values ​​of two adjacent detection cells in the horizontal direction, and then the crack depth value of the unassigned point is generated by linearly interpolating the crack depth values ​​of two adjacent horizontal midpoints in the vertical direction. All points without assigned crack depth values ​​in the ROI of the gallium oxide substrate are processed one by one to generate a continuous two-dimensional crack depth distribution cloud map containing all coordinate points. The crack depth value of each coordinate point is obtained from the continuous two-dimensional crack depth distribution cloud map. The parameters of the Gaussian smoothing filter are defined, for example, using a two-dimensional Gaussian distribution with a standard deviation of 1, to generate a weight matrix based on a normal distribution. For each coordinate point in the continuous two-dimensional crack depth distribution cloud map, a neighborhood range centered on that coordinate point is selected, for example, covering a 5x5 pixel area. The corresponding weights in the weight matrix are applied to perform a weighted average of the crack depth values ​​of each coordinate point within the neighborhood range, generating a smoothed crack depth value for each coordinate point in the continuous two-dimensional crack depth distribution cloud map. The crack depth values ​​of all coordinate points in the continuous two-dimensional crack depth distribution cloud map are processed sequentially to generate a smoothed continuous two-dimensional crack depth distribution cloud map, reducing abrupt changes between color-coded points and enhancing the smooth transition of crack depth values.

[0080] Based on the reliability requirements of the gallium oxide substrate processing technology, a crack depth threshold (e.g., 50 to 200 nanometers) is set.

[0081] A region-segmentation-based image enhancement technique is applied to process a continuous two-dimensional crack depth distribution cloud map covering the entire region of interest in a gallium oxide substrate, generating a risk region coordinate set. Specifically, the crack depth value of each pixel in the two-dimensional crack depth distribution cloud map is compared with a crack depth threshold. If the crack depth value of a pixel exceeds the crack depth threshold, the pixel in the continuous two-dimensional crack depth distribution cloud map is marked as a high-risk marker, and all high-risk marker pixels are aggregated to form a high-risk region. The coordinate information of all pixels in the high-risk region is extracted. The coordinate information of all pixels is then aggregated to generate the risk region coordinate set.

[0082] S3. Input the risk area coordinate set into the Bessel beam tomography scanner for multi-angle penetration scanning, and simultaneously acquire the scattering images generated by the scan. Fuse the scattering images to generate a three-dimensional model of subsurface damage.

[0083] Furthermore, the risk region coordinate set is imported into the control terminal of the Bessel beam tomography scanner to perform automatic positioning operations and identify high-risk regions on the gallium oxide substrate. Specifically: the risk region coordinate set is input into the control terminal of the Bessel beam tomography scanner, which reads the pixel coordinates in the risk region coordinate set; the scanning platform position of the Bessel beam tomography scanner is adjusted to align the physical coordinate system of the gallium oxide substrate with the grid coordinate system of the risk region coordinate set, ensuring accurate correspondence between the physical coordinate system of the gallium oxide substrate and the grid coordinate system of the risk region coordinate set; based on the pixel coordinates in the risk region coordinate set, the control terminal of the Bessel beam tomography scanner adjusts the position of the transmission scanning head to precisely align with the corresponding high-risk region on the gallium oxide substrate, confirming the boundary of the high-risk region.

[0084] Set the initial scanning angle of the Bessel beam tomography scanner to 0 degrees and perform a pass-through scan on the high-risk area;

[0085] Record the scattered light signal generated during the transmissive scanning process;

[0086] The control terminal of the Bessel beam tomography scanner issues a rotation command, causing the gallium oxide substrate to rotate 30 degrees along a predetermined rotation axis, thereby positioning the high-risk area at a new scanning angle.

[0087] Repeatedly perform penetration scanning, record scattered light signals, and rotate to complete penetration scanning at multiple (e.g., 12) scanning angles.

[0088] While the Bessel beam tomography scanner performs multi-angle penetration scanning, a high-precision image acquisition component is activated to acquire the scattered images generated during each scan.

[0089] A pairing operation is performed on the acquired scattering images and their corresponding scanning angle information to establish the interconnection between the scattering images and the scanning angles. Specifically, a scattering image-scanning angle pairing table is created, which contains two columns: the first column records the unique identifier of the scattering image generated by each penetration scan, and the second column records the corresponding penetration scan angle value. All scattering images are traversed, and the unique identifier of each scattering image and its corresponding penetration scan angle value are sequentially filled into the corresponding row of the scattering image-scanning angle pairing table. It is ensured that the number of rows in the scattering image-scanning angle pairing table is consistent with the number of penetration scans, and each scattering image occupies one row in the scattering image-scanning angle pairing table, thus generating the scattering image-scanning angle pairing table (containing the association between the scattering images generated by each penetration scan and their corresponding penetration scan angle values).

[0090] The scattering image and scan angle pairing table is converted into a scattering image angle pairing dataset and stored in a structured data format;

[0091] Denoising, contrast enhancement, and edge sharpening operations are sequentially performed on all scattered images in the scattered image angle pairing dataset to obtain a clear and usable set of two-dimensional scattered images; specifically:

[0092] The denoising process is as follows: For each scattering image, determine the median filter parameters for denoising, for example, using a 3x3 pixel region as the neighborhood window; traverse each pixel in the scattering image and select a 3x3 pixel neighborhood centered on each pixel; collect the brightness values ​​of all pixels within the 3x3 pixel neighborhood to form a brightness value set; sort the pixel brightness values ​​in the brightness value set and select the brightness value in the middle position after sorting as the median; replace the original brightness value of each pixel in the scattering image with the median to eliminate random noise, such as electronic noise or light scattering interference; process the brightness values ​​of all pixels in the scattering image one by one to generate the denoised scattering image.

[0093] The contrast enhancement process involves: obtaining the brightness value of each pixel from the denoised scattering image; statistically analyzing the distribution of brightness values ​​across all pixels in the denoised scattering image to generate a brightness histogram, which records the frequency of each brightness value; constructing a cumulative distribution function based on the brightness histogram, which represents the cumulative probability corresponding to each brightness value; iterating through each pixel in the denoised scattering image, mapping the original brightness value of each pixel to the cumulative distribution function to generate a new brightness value, with the new brightness value range extended to 0 to 255, making the brightness value distribution more uniform; replacing the original brightness value of each pixel in the denoised scattering image with the new brightness value to generate a contrast-enhanced scattering image, making the brightness difference between the subsurface damage area and the background area more obvious, and enhancing the visual clarity of the scattering image.

[0094] Edge sharpening specifically involves: obtaining the brightness value of each pixel from the contrast-enhanced scattering image; defining the convolution kernel of a Laplacian filter, for example, a Laplacian operator for a 3x3 pixel region, with a center weight of -4 and weights of 1 for the eight surrounding neighborhoods; traversing each pixel in the contrast-enhanced scattering image and selecting a 3x3 pixel neighborhood centered on each pixel; applying the Laplacian filter convolution kernel to perform a weighted summation of the brightness values ​​of each pixel within the 3x3 pixel neighborhood to generate an edge enhancement value for each pixel in the contrast-enhanced scattering image; superimposing the edge enhancement value of each pixel in the contrast-enhanced scattering image onto the original brightness value of each pixel in the contrast-enhanced scattering image to generate a new brightness value, highlighting the pixel value gradient of the subsurface damage edge; processing the brightness values ​​of all pixels in the contrast-enhanced scattering image one by one to generate an edge-sharpened scattering image, enhancing the boundary features of cracks or defects; and sequentially processing all scattering images in the scattering image angle pairing dataset to generate a clear and usable two-dimensional scattering image set.

[0095] Based on the two-dimensional scattering image corresponding to each penetration scanning angle, a projection image data matrix is ​​generated. Specifically, this involves: extracting the brightness value of each pixel in the two-dimensional scattering image corresponding to each penetration scanning angle from the two-dimensional scattering image set; the brightness value reflects the characteristics of subsurface damage; accessing the metadata of the two-dimensional scattering image, which contains image resolution information recorded by the high-precision image acquisition component when acquiring the scattering image; extracting the resolution parameter from the metadata, for example, 512x512 pixels represents the number of rows and columns of pixels in the two-dimensional scattering image; verifying that the resolution parameter is consistent with the configuration of the high-precision image acquisition component to ensure that the pixel grid of the two-dimensional scattering image completely covers the scanning range of the high-risk area; recording the resolution parameter as the size of the two-dimensional scattering image; organizing the brightness value of each pixel in the extracted two-dimensional scattering image into a two-dimensional array, where each element of the two-dimensional array corresponds to the brightness value of a pixel in the two-dimensional scattering image; associating the two-dimensional array with the corresponding penetration scanning angle value to generate a projection image data matrix, which contains the pixel brightness value of the two-dimensional scattering image and the corresponding penetration scanning angle value; and sequentially processing the two-dimensional scattering images corresponding to all penetration scanning angles in the two-dimensional scattering image set to generate a projection image data matrix corresponding to each penetration scanning angle.

[0096] The projection image data matrix from all penetration scanning angles is input into the image reconstruction algorithm. A filtered back-projection method is used for 3D spatial reconstruction, outputting a 3D model of subsurface damage covering high-risk areas; specifically:

[0097] First, initialize the processing environment for the image reconstruction algorithm to ensure support for the execution of the filtered back projection method; then, traverse the set of projection image data matrices in the order of the penetration scanning angle values, extract the pixel brightness value and the corresponding penetration scanning angle value of each projection image data matrix, organize them into a record containing the pixel brightness value and the penetration scanning angle value, and generate the input dataset for the image reconstruction algorithm.

[0098] Secondly, for each projected image data matrix in the input dataset, the parameters of the Ram-Lak filter are configured, for example, setting the frequency response range of the Ram-Lak filter with a cutoff frequency of 0.5. A two-dimensional array of the projected image data matrix is ​​extracted, where each element of the two-dimensional array corresponds to the brightness value of a pixel in the two-dimensional scattering image. A Fast Fourier Transform (FFT) is applied to convert the pixel brightness values ​​of the two-dimensional array into a frequency domain representation, generating frequency domain data. Combining the frequency response characteristics of the Ram-Lak filter, the frequency domain data is adjusted to enhance the high-frequency components of the subsurface damage features and suppress low-frequency noise, generating adjusted frequency domain data. An Inverse Fast Fourier Transform (IFT) is applied to restore the adjusted frequency domain data to pixel brightness values ​​in the spatial domain, generating a filtered two-dimensional array of pixel brightness values. The filtered two-dimensional array of pixel brightness values ​​is correlated with the corresponding transmission scanning angle value to generate a filtered projected image data matrix. Here, the FFT and IFT are existing technologies used to convert the pixel brightness values ​​of the projected image data matrix from the spatial domain to the frequency domain or from the frequency domain to the spatial domain, in order to enhance the high-frequency components of the subsurface damage features and suppress low-frequency noise.

[0099] Next, based on each filtered projection image data matrix, a three-dimensional grid coordinate system is defined to cover the spatial range of the high-risk area of ​​the gallium oxide substrate. According to the penetration scanning angle value corresponding to the filtered projection image data matrix, a back-projection operation is performed. The two-dimensional array of filtered pixel brightness values ​​is traversed, and the projection path of the voxel is determined based on the penetration scanning angle value. The pixel brightness values ​​of the filtered projection image data matrix are mapped to voxels in the three-dimensional grid coordinate system, serving as the back-projection result for a single penetration scanning angle. The back-projection result represents the spatial distribution of subsurface damage features along the corresponding penetration scanning angle. All filtered projection image data matrices are processed sequentially to generate the back-projection result corresponding to each penetration scanning angle. The back-projection results corresponding to all penetration scanning angles are traversed, and the voxel values ​​of each back-projection result are merged into the three-dimensional grid coordinate system to generate stereoscopic data covering the high-risk area.

[0100] Finally, subsurface damage features are extracted from the stereo data, the distribution of voxel brightness values ​​in the stereo data is analyzed, and a brightness threshold is set for the stereo data, for example, 100 is used as the brightness threshold. Voxels with brightness values ​​higher than the brightness threshold in the stereo data are retained as subsurface damage areas, and background areas with voxel brightness values ​​lower than the brightness threshold in the stereo data are removed. Based on the retained voxels, the boundary features of subsurface damage are identified, boundary surfaces are constructed, and a three-dimensional model of subsurface damage covering high-risk areas is generated.

[0101] S4. Perform data registration and weight superposition between the two-dimensional crack depth distribution cloud map and the three-dimensional subsurface damage model, and output a morphology prediction report and grinding process optimization parameters.

[0102] Furthermore, spatial coordinate alignment is performed between the two-dimensional crack depth distribution cloud map and the three-dimensional subsurface damage model to complete the initial data registration. Specifically, based on the physical dimensions of the gallium oxide substrate, a mapping relationship is established between the grid coordinate system of the two-dimensional crack depth distribution cloud map and the three-dimensional grid coordinate system of the subsurface damage model. By matching the boundary coordinates of high-risk areas, the pixel coordinates of the two-dimensional crack depth distribution cloud map are projected onto the voxel coordinate plane of the subsurface damage model to achieve coordinate system alignment.

[0103] Using the two-dimensional crack depth distribution cloud map as a reference, local rotation and translation corrections are performed on the corresponding regions in the three-dimensional model of subsurface damage.

[0104] Weighting coefficients were set for the registered two-dimensional crack depth distribution cloud map and the three-dimensional subsurface damage model respectively; the two-dimensional crack depth distribution cloud map was given a lower weight (e.g., 0.3), and the three-dimensional subsurface damage model was given a higher weight (e.g., 0.7).

[0105] A pixel-by-pixel weighted fusion was performed on the two-dimensional crack depth distribution cloud map and the three-dimensional subsurface damage model to generate a fused comprehensive damage assessment dataset.

[0106] Based on the fused comprehensive damage assessment dataset, potential crack propagation paths in the three-dimensional model of subsurface damage are identified, and crack propagation path prediction results are generated.

[0107] The surface morphology evolution trend of gallium oxide substrates is predicted based on the fused comprehensive damage assessment dataset, forming a surface morphology prediction result. Specifically, the brightness value of each voxel in the fused comprehensive damage assessment dataset is extracted, as the brightness value reflects the intensity of subsurface damage features. Voxel connectivity analysis is applied to traverse all voxels in the fused comprehensive damage assessment dataset, identifying voxels with brightness values ​​higher than the stereo data brightness threshold (e.g., 100) and marking them as subsurface damage regions. Based on the subsurface damage regions, the connectivity between adjacent voxels is detected to construct connected regions, and voxel paths with continuous high brightness values ​​are identified as potential crack propagation paths. The spatial orientation of potential crack propagation paths is analyzed based on the voxel coordinate set of potential crack propagation paths, recording the start, end, and intermediate voxel coordinates of each path to form a spatial coordinate set containing all potential crack propagation paths. The spatial coordinate set is organized into structured data to generate crack propagation path prediction results.

[0108] Based on crack propagation path prediction results, the impact of potential crack propagation paths on the gallium oxide substrate surface is analyzed, and the surface morphology evolution trend is predicted to form surface morphology prediction results. Specifically, the spatial coordinate set of potential crack propagation paths is extracted from the crack propagation path prediction results, including the start point, end point, and intermediate voxel coordinates of each potential crack propagation path. For each potential crack propagation path, the length, direction, and coordinates of the intersection point of the potential crack propagation path projected onto the gallium oxide substrate surface are determined. Based on the spatial orientation of the potential crack propagation path, it is determined whether the potential crack propagation path extends to the gallium oxide substrate surface or propagates in the subsurface layer. For potential crack propagation paths extending to the surface, the crack width and depth at the intersection of the potential crack propagation paths are analyzed to assess the degree of damage to the surface morphology. For example, by measuring whether the crack width exceeds 0.5 μm or the depth reaches more than 1 μm, it is confirmed whether significant surface depressions (manifested as local surface depressions of more than 0.2 μm) or cracks (manifested as continuous surface cracks) are formed. For subsurface propagation crack paths, determine the distance between the potential crack propagation path and the surface, and the influence of the potential crack propagation path on the surface stress distribution. Determine whether it may cause indirect changes in surface morphology. For example, by measuring whether the minimum distance between the potential crack propagation path and the surface is less than 0.3 micrometers, confirm whether stress concentration causes surface bulging (manifested as a local surface bulge height exceeding 0.1 micrometers) or microcracks (manifested as fine cracks with a length less than 0.5 micrometers and a depth less than 0.2 micrometers). Combining the analysis results of all potential crack propagation paths, organize the crack distribution and surface influence data to generate a morphology change prediction description of the gallium oxide substrate surface, describing the surface morphology change trend caused by the potential crack propagation path, including the direction of the potential crack propagation path, the distribution of surface defects, and morphology inhomogeneity. Organize the morphology change prediction description into structured data as the surface morphology prediction result, including the spatial coordinates of the potential crack propagation path, the description of the surface morphology change, and the predicted morphology feature distribution.

[0109] Based on the crack propagation direction and material removal rate in the three-dimensional model of subsurface damage, the recommended grinding process optimization parameters are calculated, including grinding depth, grinding time, grinding rate and grinding direction.

[0110] The expression for grinding depth is:

[0111] D g =max(D c );

[0112] Among them, D g The recommended grinding depth (in micrometers) represents the thickness of material to be removed from the gallium oxide substrate, D. cThe set of voxel depths (in micrometers) of potential crack propagation paths in the three-dimensional model of subsurface damage is extracted from the set of spatial coordinates of crack propagation paths through voxel connectivity analysis.

[0113] The expression for grinding time is:

[0114]

[0115] Among them, T g Grinding time (unit: minutes), representing the total time required to complete the grinding process, R. r Material removal rate (unit: micrometers / minute), based on experimental data of the grinding process on gallium oxide substrates, for example, 0.1 micrometers / minute;

[0116] The expression for the grinding rate is:

[0117]

[0118] Among them, V g T represents the grinding rate (unit: micrometers per second), indicating the operating speed of the grinding equipment. t The total actual operating time of the grinding equipment (in seconds), usually based on the grinding time T. g Adjustments to the grinding equipment settings;

[0119] The expression for the grinding direction is:

[0120] θ g =θ c +90°

[0121] Where, θ g The recommended grinding direction (unit: degrees) represents the angle of motion of the grinding equipment relative to the surface of the gallium oxide substrate, θ. c The crack propagation direction (unit: degrees) is determined by analyzing the set of spatial coordinates of potential crack propagation paths in the three-dimensional model of subsurface damage.

[0122] The crack propagation path prediction results, surface morphology prediction results, and grinding process optimization parameters are integrated into a structured text format to output a morphology prediction report and grinding process optimization parameters.

[0123] This embodiment also provides a system for predicting the surface morphology of gallium oxide substrates, including:

[0124] The acquisition module is used to acquire gallium oxide substrates with europium-based fluorescent nanoparticles embedded on their surface;

[0125] The marking module is used to irradiate the gallium oxide substrate with ultraviolet light, acquire surface fluorescence distribution images and match them with a pre-constructed fluorescence-crack mapping database to generate a two-dimensional crack depth distribution cloud map; and mark the risk area coordinate set in the two-dimensional crack depth distribution cloud map based on a preset crack depth threshold.

[0126] The scanning module is used to input the coordinate set of the risk area into the Bessel beam tomography scanner for multi-angle penetration scanning, and simultaneously acquire the scattered images generated by the scanning, and fuse the scattered images to generate a three-dimensional model of subsurface damage.

[0127] The prediction module is used to register and weight the two-dimensional crack depth distribution cloud map with the three-dimensional subsurface damage model, and output the morphology prediction report and grinding process optimization parameters.

[0128] This embodiment also provides a computer device applicable to the method for predicting the surface morphology of gallium oxide substrates, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the method for predicting the surface morphology of gallium oxide substrates as proposed in the above embodiment.

[0129] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0130] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements the method for predicting the surface morphology of a gallium oxide substrate as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0131] In summary, this invention utilizes Bessel beam tomography and multi-angle scattering image fusion technology to construct a three-dimensional model of subsurface damage, achieving high-resolution visualization and reconstruction of the microcracks and damage distribution within gallium oxide substrates. The three-dimensional subsurface damage model not only accurately captures the crack propagation path and density characteristics along the material's depth, but also provides crucial geometric basis for subsequent crack propagation trend prediction and grinding process parameter optimization. Compared to traditional shallow detection methods, this significantly improves the depth and accuracy of subsurface structural integrity assessment, fills the gap in deep damage modeling in existing technologies, and enhances the comprehensiveness and reliability of morphology prediction.

[0132] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for predicting the surface morphology of gallium oxide substrates after grinding, characterized in that: include, A gallium oxide substrate with europium-based fluorescent nanoparticles embedded on its surface was obtained; A gallium oxide substrate is irradiated with ultraviolet light, and surface fluorescence distribution images are acquired and matched with a pre-constructed fluorescence-crack mapping database to generate a two-dimensional crack depth distribution cloud map. Based on a preset crack depth threshold, mark the coordinate set of the risk area in the two-dimensional crack depth distribution cloud map; The coordinate set of the risk area is input into the Bessel beam tomography scanner for multi-angle penetration scanning, and the scattered images generated by the scanning are acquired simultaneously. The scattered images are then fused to generate a three-dimensional model of subsurface damage. The two-dimensional crack depth distribution cloud map and the three-dimensional subsurface damage model are registered and weighted to output a morphology prediction report and grinding process optimization parameters.

2. The method for predicting the surface morphology of gallium oxide substrates as described in claim 1, characterized in that: The process involves acquiring surface fluorescence distribution images and matching them with a pre-constructed fluorescence-crack mapping database to generate a two-dimensional crack depth distribution cloud map, as detailed below. A gallium oxide substrate was illuminated with an ultraviolet light source using a fluorescence microscope to obtain an image of the surface fluorescence distribution; Background subtraction and grayscale normalization are performed on the surface fluorescence distribution image to generate a grayscale normalized surface fluorescence distribution image; The region of interest of the gallium oxide substrate was selected from the gray-scale normalized surface fluorescence distribution image and divided into multiple uniformly distributed detection cells; The average fluorescence intensity of each detection cell is extracted and summarized to form a fluorescence intensity dataset; The fluorescence intensity dataset is compared one by one with the fluorescence intensity values ​​in the pre-constructed fluorescence-crack mapping database. The fluorescence intensity value that matches the average fluorescence intensity of each detection cell is found, and the corresponding crack depth value is extracted. Record the average fluorescence intensity of each detection cell and the corresponding crack depth value to create a fluorescence crack correspondence table; Based on the fluorescence crack correspondence table, a fluorescence intensity visualization algorithm is applied to the region of interest in the gallium oxide substrate to draw color-coded points and generate a color-coded point distribution. Interpolation and smoothing are performed on the color-coded point distribution to generate a continuous two-dimensional crack depth distribution cloud map.

3. The method for predicting the surface morphology of gallium oxide substrates as described in claim 2, characterized in that: The fluorescence-crack mapping database is a database of the correspondence between fluorescence intensity and crack depth obtained through experimental measurements.

4. The method for predicting the surface morphology of gallium oxide substrates as described in claim 1, characterized in that: The process of marking the coordinate set of risk areas in the two-dimensional crack depth distribution cloud map based on a preset crack depth threshold is as follows: The crack depth threshold is set according to the reliability requirements of the gallium oxide substrate processing technology; Image enhancement techniques based on region segmentation are applied to process continuous two-dimensional crack depth distribution cloud maps, and the crack depth value of each pixel in the continuous two-dimensional crack depth distribution cloud map is compared with the crack depth threshold. If the crack depth value of a pixel exceeds the crack depth threshold, the pixel in the continuous two-dimensional crack depth distribution cloud map is marked as a high-risk marker. All pixels marked with high risk are aggregated to form a high-risk area; Extract the coordinate information of all pixels in the high-risk area, and summarize the coordinate information of all pixels to generate a risk area coordinate set.

5. The method for predicting the surface morphology of gallium oxide substrates as described in claim 1, characterized in that: The process of inputting the risk area coordinate set into a Bessel beam tomography scanner for multi-angle penetration scanning is as follows: Input the risk area coordinate set into the control terminal of the Bessel beam tomography scanner; Based on the pixel coordinates in the risk area coordinate set, the control terminal of the Bessel beam tomography scanner adjusts the position of the transmission scanning head to align with the corresponding high-risk area on the gallium oxide substrate. Set the initial scanning angle of the Bessel beam tomography scanner, perform a transmission scan on the high-risk area, and record the scattered light signal generated during the transmission scan. The control terminal of the Bessel beam tomography scanner issues a rotation command to position the high-risk area at a new scanning angle. Repeatedly perform transmissive scanning, record scattered light signals, and rotate to complete multi-angle transmissive scanning.

6. The method for predicting the surface morphology of gallium oxide substrates as described in claim 1, characterized in that: The synchronous acquisition of the scattering image refers to the process of activating the high-precision image acquisition component to acquire the scattering image generated during each transmission scan when the Bessel beam tomography scanner performs multi-angle transmission scans.

7. The method for predicting the surface morphology of gallium oxide substrates as described in claim 1, characterized in that: The fused scattering image generates a three-dimensional model of subsurface damage, as detailed below. The acquired scattering images are paired with the corresponding penetration scanning angle values ​​to generate a scattering image and scanning angle pairing table, and the scattering image and scanning angle pairing table is converted into a scattering image angle pairing dataset. The median filter is used to denoise all scattered images in the scattered image angle pairing dataset, histogram equalization is used to enhance contrast, and Laplacian filter is used to sharpen edges to generate a two-dimensional scattered image set. The pixel brightness values ​​of the two-dimensional scattering image corresponding to each penetration scanning angle are extracted from the two-dimensional scattering image set and organized into a projection image data matrix; The projected image data matrix is ​​loaded into the image reconstruction algorithm, and stereoscopic data covering high-risk areas is generated by processing with a Ram-Lak filter, back-projection operation, and voxel merging. Based on the distribution of voxel brightness values ​​in the stereo data, a brightness threshold is set, and voxels with brightness values ​​higher than the brightness threshold in the stereo data are retained as subsurface damage areas. Identify the boundary features of the subsurface damage region, construct the boundary surface, and generate a three-dimensional model of the subsurface damage covering the high-risk region.

8. A system for predicting the surface morphology of gallium oxide substrates after grinding, based on the method for predicting the surface morphology of gallium oxide substrates according to any one of claims 1 to 7, characterized in that: include, The acquisition module is used to acquire gallium oxide substrates with europium-based fluorescent nanoparticles embedded on their surface; The labeling module is used to irradiate the gallium oxide substrate with ultraviolet light, acquire surface fluorescence distribution images and match them with a pre-constructed fluorescence-crack mapping database to generate a two-dimensional crack depth distribution cloud map. Based on a preset crack depth threshold, mark the coordinate set of the risk area in the two-dimensional crack depth distribution cloud map; The scanning module is used to input the coordinate set of the risk area into the Bessel beam tomography scanner for multi-angle penetration scanning, and simultaneously acquire the scattered images generated by the scanning, and fuse the scattered images to generate a three-dimensional model of subsurface damage. The prediction module is used to register and weight the two-dimensional crack depth distribution cloud map with the three-dimensional subsurface damage model, and output the morphology prediction report and grinding process optimization parameters.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the method for predicting the surface grinding morphology of gallium oxide substrates according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the method for predicting the surface grinding morphology of gallium oxide substrates according to any one of claims 1 to 7.

Images