Method for rapidly positioning high-magnesium substances in lunar soil sample

Abstract

The invention provides a method for rapidly positioning a high-magnesium substance in a lunar soil sample, and belongs to the technical field of lunar substance analysis. Embedding the lunar soil powder sample in an epoxy resin target, and carrying out grinding, polishing and carbon coating; placing a resin target sample of lunar soil powder, a metal cobalt standard sample and a quantitative standard substance in a scanning electron microscope sample chamber; collecting a backscattered electron image (BSE) of the sample to be detected; carrying out quantitative analysis setting; performing automatic back scattering image puzzle; processing data in real time in a scanning process through an edge computing technology, automatically generating an element distribution diagram, and completing energy spectrum surface scanning and jigsaw; performing automatic phase separation on an energy spectrum surface scanning result; carrying out Mg # value calculation, and carrying out classified coloring on chemical components with different Mg contents; randomly selecting part of high-magnesium substance particles, and carrying out secondary energy spectrum analysis and scanning electron microscope image acquisition; and algorithm parameters are optimized through a feedback mechanism. According to the method, the analysis efficiency and accuracy of the high-magnesium substance are remarkably improved.

Description

Technical Field

[0001] The present invention provides a method for quickly locating high-magnesium substances in lunar soil samples, belonging to the technical field of lunar material analysis. Background Art

[0002] Lunar materials are diverse, and many of them have high-magnesium characteristics, including lunar mantle materials, extra-lunar meteorite fragments, magnesian rock suites, and magnesian anorthosites. The study of these high-magnesium substances is of great significance for understanding the composition and structure of lunar surface rocks, exploring the lunar crust composition, early geological evolution process, and internal thermal history. However, there are many challenges in studying these high-magnesium substances. First, the number of particles in lunar soil samples is huge, and each target may contain hundreds of thousands of particles. Second, the structures and compositions of these high-magnesium substances are complex, and the particles often exhibit sharp and irregular shapes, mixed with other rock fragments, making it difficult to quickly distinguish them. In addition, the mineral particles in high-magnesium substances are usually small, posing higher requirements for the resolution of analysis techniques. Through the systematic study of high-magnesium substances in Chang'e-6 lunar soil, scientists can not only answer key scientific questions related to the origin of the moon, internal structure, and magmatic activities, but also lay a scientific foundation for future human deep space exploration and lunar resource development. However, the prerequisite for studying these scientific questions is to accurately answer the following two core questions: where are these high-magnesium substances distributed? What are their specific characteristics? Therefore, establishing a set of efficient and accurate methods for locating high-magnesium substances in lunar soil particles has become a necessary condition for carrying out relevant research.

[0003] Traditional analysis methods, such as Raman spectroscopy, electron probe, and conventional electron microscopy techniques, although able to provide certain composition and structure information, have deficiencies in terms of analysis efficiency and comprehensiveness, and are difficult to meet the requirements for rapid location and identification of complex samples. In recent years, with the integration of automated image processing technology and machine learning, some emerging automated analysis techniques have been applied, such as micro-area X-ray fluorescence technology and automated mineral analysis systems. These techniques have shown significant advantages in improving analysis efficiency and accuracy. In the field of geology, commonly used automated mineral analysis systems include Maps, QuemScan, TIMA, and AMICS, etc. These systems are based on scanning electron microscope (SEM) technology, combined with X-ray energy spectrum analysis, and achieve automatic mineral location through powerful post-processing software. However, due to the relatively large beam spot of micro-area X-ray fluorescence technology, it is difficult to meet the requirements for high-resolution analysis of samples. Although the automated mineral analysis system performs well in mineral classification and location, it relies on a preset database for classification, thus having certain limitations. These systems can only mark the positions of magnesium-containing substances and cannot perform fine grading of samples according to different magnesium contents, making it difficult to achieve precise location of high-magnesium substances. These limitations indicate the need to further develop and optimize analysis techniques to meet the accuracy and efficiency requirements for the study of high-magnesium substances. Summary of the Invention

[0004] In view of the above problems, the present invention adopts energy spectrum surface scanning and mineral phase element similarity analysis technology to classify substances with different magnesium contents.

[0005] The specific technical solution is as follows: A method for quickly locating high-magnesium substances in lunar soil samples, comprising the following steps: Step 1: Embed the lunar soil powder sample in an epoxy resin target and polish it with gradually refined sandpaper. Subsequently, the sample is polished using diamond abrasives of 9 µm, 3 µm, and 1 µm, and then carbon-coated with an 8-20 nm thickness using a Leica EM ACE600 coating system. Experiments have proven that carbon coating of this thickness has no significant impact on the quantitative analysis of most geological samples, and at the same time can effectively improve the conductivity and electron beam damage resistance of the sample.

[0006] Step 2: Place the resin target sample of the lunar soil powder, a metal cobalt standard sample, and a quantitative standard substance (such as albite, diopside, etc.) together in the scanning electron microscope sample chamber. Evacuate to below 10 -6 Pa to ensure that the experimental process is always in a high vacuum state, minimizing the interference of gas molecules on the electron beam and ensuring the accuracy of imaging and energy spectrum analysis. Through the beam current calibration of the cobalt metal standard sample, the reliability of the non-normalized quantitative analysis results is ensured.

[0007] Step 3: Select appropriate acceleration voltage (usually 15-20 kV), beam current intensity (1-20 nA), brightness, and contrast to collect high-quality backscattered electron images (BSE) of the sample to be measured. Use a standard substance containing the same elements to set up the quantitative analysis of the relevant elements of the sample to be measured. Combine multiple standard substances (albite, diopside, forsterite, almandine, rutile, orthoclase, chromite, calcium rhodonite, and apatite) to set up the quantitative analysis of the key elements (Si, Na, Al, Ca, Mg, Fe, Ti, K, Cr, Mn, P, F) in the sample to ensure the comprehensiveness and accuracy of element analysis.

[0008] Step 4: Since only a 2-mm sample area can be observed in a single field of view, in order to locate the high-magnesium substances in the entire sample target, automated backscattered image mosaicking is required. To ensure the quality of a single image, select an image resolution of 1024~8192. Control the movement of the sample stage and image acquisition through automated software to ensure that the spliced image is seamlessly connected and covers the entire sample surface.

[0009] Step 5: Select appropriate resolution (512 - 2048), acquisition mode (fixed number of frames), processing time (3 - 5), energy range (selected according to the X-ray energy of the element to be measured), number of channels, and pixel dwell time (2000 - 12000) to ensure that each single mineral sample has sufficient counts and guarantee the accuracy of quantitative analysis. Through edge computing technology, data is processed in real time during scanning, an elemental distribution map is automatically generated, and an energy spectrum surface scan mosaic is completed.

[0010] Step 6: Use the AutophaseMap technology to perform automatic phase separation on the energy spectrum surface scan results. First, denoise and background correct the energy spectrum data of each pixel point, strip the influence of coating elements, and ensure data quality. Dimension reduction and feature extraction are performed on the energy spectrum data through chemometric methods (such as principal component analysis PCA or non-negative matrix factorization), and pixel points with similar chemical compositions are classified into the same phase by combining clustering algorithms. To avoid minor phases being merged and losing high-magnesium substances with low content of concern, set the boundary tolerance (8 - 10) and grouping level (0 - 2) to accurately distinguish the boundaries of each phase and chemical compositions, and ensure the accurate identification of high-magnesium substances.

[0011] The boundary tolerance controls the behavior at the boundaries of each phase, and its calculation formula is: T boundary =

[0012] where T boundary is the boundary tolerance, representing the average value of pixel errors at the phase boundary, which is used to overall measure the accuracy of phase boundary recognition. The smaller the value, the more accurate the boundary classification; is the pixel error at the phase boundary, which is used to measure the accuracy of phase boundary recognition and reflects the accuracy of pixel classification at the phase boundary, mainly determined by the signal-to-noise ratio of the energy spectrum data; n represents the total number of pixels at the phase boundary, that is, the number of boundary pixel points participating in the calculation.

[0013] Low boundary tolerance: Each pixel represents a pure spectrum, suitable for high-purity phase analysis, but may cause unidentifiable pixels of mixed phases.

[0014] High boundary tolerance: All pixels are fitted to the closest phase, suitable for rapid classification of complex samples.

[0015] Grouping level: Similar phases are combined through intelligent grouping to create smaller and more manageable numbers of phases. A low grouping level shows more minor phases and is suitable for trace compound analysis; a high grouping level shows fewer phases and is suitable for large-area analysis.

[0016] Step 7: Calculate the Mg# value (Mg / (Mg + Fe)×100) for the automatically separated chemical components, and classify them according to the Mg# value. Map the classification information onto the spatial distribution of the samples to generate a high-precision Mg content classification map, and classify and color the chemical components with different Mg contents. Combine the mineral composition characteristics, intelligently determine the mineral type, and adjust the classification criteria of the Mg# value in real time to optimize the classification results.

[0017] Step 8: Randomly select some high-magnesium substance particles for secondary energy spectrum analysis and scanning electron microscope imaging to verify the accuracy of the classification results. Optimize the algorithm parameters through the feedback mechanism to further improve the reliability and accuracy of the analysis.

[0018] The purpose of the present invention is to establish a method that can efficiently and accurately identify and classify high-magnesium substances in lunar soil, reveal the lunar geological evolution history, and provide important technical support and data guarantee for future lunar exploration, resource development, and scientific research.

[0019] The present invention realizes the positioning and accurate classification of high-magnesium substances in lunar soil, overcomes the disadvantages of the prior art, such as low analysis efficiency, insufficient resolution, and difficulty in achieving fine classification of high-magnesium substances due to the large number, irregular shape, complex composition of lunar soil particles and their mixing with other rock fragments. The present invention significantly improves the analysis efficiency and accuracy of high-magnesium substances, and provides strong technical support for lunar geological evolution research and resource development.

[0020] The main technical effects of the present invention are as follows: 1. Intelligent phase separation technology: Combine AutophaseMap and chemometrics methods to achieve high-precision and automated mineral phase separation and chemical composition analysis.

[0021] 2. Real-time edge computing: Process data in real time during scanning, significantly improving the analysis efficiency.

[0022] 3. Dynamic boundary tolerance and grouping level optimization: Dynamically adjust the boundary tolerance and grouping level through intelligent algorithms to ensure the accurate identification of trace phases and complex phases.

[0023] 4. High-resolution full-coverage scanning: Adopt high-resolution automated image stitching technology to achieve seamless coverage and high-precision analysis of the sample surface.

[0024] 5. Multi-dimensional verification mechanism: Ensure the reliability and accuracy of the classification results through secondary energy spectrum analysis and scanning electron microscope imaging. Description of the Drawings

[0025] Figure 1 is the flow chart of the present invention; Figure 2 is the lunar soil powder target diagram of this embodiment; Figure 3 is the forsterite Fo value classification diagram of this embodiment; Figure 4 is the low-calcium pyroxene Mg# classification diagram of this embodiment; Figure 5 is the high-calcium pyroxene Mg# classification diagram of this embodiment; Figure 6 is the X-ray energy spectrum diagram of low-magnesium olivine in this embodiment; Figure 7 is the X-ray energy spectrum diagram of medium-magnesium olivine in this embodiment; Figure 8 is the X-ray energy spectrum diagram of high-magnesium olivine in this embodiment; Figure 9 is the X-ray energy spectrum diagram of low-magnesium and low-calcium pyroxene in this embodiment; Figure 10 is the X-ray energy spectrum diagram of medium-magnesium and low-calcium pyroxene in this embodiment; Figure 11 is the X-ray energy spectrum diagram of high-magnesium and low-calcium pyroxene in this embodiment; Figure 12 is the X-ray energy spectrum diagram of low-magnesium and high-calcium pyroxene in this embodiment; Figure 13 is the X-ray energy spectrum diagram of medium-magnesium and high-calcium pyroxene in this embodiment; Figure 14 is the X-ray energy spectrum diagram of high-magnesium and high-calcium pyroxene in this embodiment; Figure 15 is the detailed scanning electron microscope image of high-magnesium olivine in this embodiment; Figure 16 is the detailed scanning electron microscope image of high-magnesium and low-calcium pyroxene in this embodiment; Figure 17 is the detailed scanning electron microscope image of high-magnesium and high-calcium pyroxene in this embodiment. Detailed implementation mode

[0026] The following further describes in combination with specific samples and the technical solutions involved in the present invention, but it does not limit the content of the present invention. As described above, the present invention provides a method for locating and accurately classifying high-magnesium substances in lunar soil, as Figure 1 shown, the method includes the following steps: (1), Embed the lunar soil powder sample of Chang'e 6 applied in an epoxy resin target, as Figure 2 shown, polish to expose the sample, and use it after cutting, polishing, cleaning and drying.

[0027] (2) Place the sample target prepared in (1), the metal cobalt standard sample, and quantitative standard substances (such as albite, diopside, forsterite, almandine, rutile, orthoclase, chromite, calcium rhodonite, and apatite) together in the sample chamber of a high-vacuum scanning electron microscope.

[0028] (3) Select an acceleration voltage of 20 kV, a beam current intensity of 5 nA, a brightness of 34%, and a contrast of 57% to collect high-quality backscattered electron images of the sample to be measured.

[0029] (4) Since only a 2-mm sample can be observed in a single field of view, in order to locate the high-Mg substances in the entire sample target, automated backscattered image mosaicking is required. To ensure the quality of each single image and the acquisition speed, select an image resolution of 2048 and a dwell time of 7 μs. The acquisition magnification is 60×. According to the size of the sample target, the number of acquisition blocks is 168. Control the movement of the sample stage and image acquisition through automated software to ensure that the spliced image is seamlessly connected and covers the entire sample surface.

[0030] Step 5: Select a resolution of 1024, an acquisition mode of 1 frame, a processing time of 4, an energy range of 15 keV, the number of channels, and a pixel dwell time of 5000 to ensure that each single mineral sample has sufficient counts and guarantee the accuracy of quantitative analysis. Through edge computing technology, process the data in real time during the scanning process, automatically generate element distribution maps, and perform automatic energy spectrum surface scanning mosaicking. After the scanning is completed, arrange automatically and perform clipping processing.

[0031] Step 6: Perform automatic separation of chemical components on the results of the surface scan using AutophaseMap. Select a boundary tolerance of 10 and a grouping level of 0 to accurately distinguish each phase boundary and chemical composition. The results are shown in Table 1.

[0032] Step 7: Analyze each automatically separated chemical component, classify according to the Fo / Mg# value (Mg / (Mg + Fe) × 100), and color the olivine, low-calcium pyroxene, and high-calcium pyroxene with different Fo and Mg# values respectively, as shown in Figure 3 and Figure 5 respectively.

[0033] Step 8: Randomly select some high-Mg substance particles for secondary energy spectrum analysis to verify the accuracy of the classification results. The corresponding energy spectrum results are shown in Figures 6 to 14 . Take detailed SEM images of high-Mg olivine, as shown in Figures 15 to 17 .

Claims

1. A method for quickly locating high-magnesium substances in lunar soil samples, characterized in that: The following steps are involved: Step 1: embed the lunar soil powder sample in an epoxy resin target, and then grind, polish, and carbon-coat it; Step 2: Place the lunar soil powder resin target sample, the metal cobalt standard sample and the quantitative standard substance together in the scanning electron microscope sample chamber; evacuate the chamber, and calibrate the beam current of the cobalt metal standard sample; Step 3: Set the acceleration voltage, beam intensity, brightness and contrast to collect high-quality backscattered electron images (BSE) of the sample to be tested; use standard substances containing the same elements to perform quantitative analysis of related elements of the sample to be tested; combine multiple standard substances to perform quantitative analysis of key elements in the sample; Step 4: Perform automated backscatter image mosaic and select image resolution 1024~8192; control the movement of the sample stage and image acquisition through automated software; Step 5: Set the resolution, acquisition mode, processing time, energy range, number of channels, and pixel dwell time to ensure that each single mineral sample has sufficient counts and the accuracy of quantitative analysis; use edge computing technology to process data in real time during the scanning process, automatically generate element distribution maps, and complete the energy spectrum scanning puzzle; Step 6, using AutophaseMap technology to automatically phase separate the energy spectrum scanning results; Step 7, calculate the Mg# value of the automatically separated chemical components, and classify them according to the Mg# value; map the classification information to the spatial distribution of the sample, generate a high-precision Mg content classification map, and classify and color the chemical components with different Mg contents; combine the mineral composition characteristics, intelligently determine the mineral type, adjust the Mg# value classification standard in real time, and optimize the classification results; Step 8: Randomly select some high-magnesium material particles, conduct secondary energy spectrum analysis and scanning electron microscope imaging, and verify the accuracy of the classification results; Optimize algorithm parameters through feedback mechanism.

2. A method for rapidly locating high-magnesium substances in lunar soil samples according to claim 1, characterized in that: In step 1, the samples were first ground with sandpaper of progressively finer grains; subsequently, the samples were polished using 9 µm, 3 µm, and 1 µm diamond abrasives.

3. The method for rapidly locating high-magnesium substances in lunar soil samples according to claim 1, characterized in that: In step 2, vacuum to 10 -6 Pa to ensure that the experiment is always in a high vacuum state.

4. The method for rapidly locating high-magnesium substances in lunar soil samples according to claim 1, characterized in that: The multiple standard substances described in step 3 include one or more of albite, diopside, forsterite, almandine, rutile, orthoclase, chromite, calcite and apatite.

5. The method for rapidly locating high-magnesium substances in lunar soil samples according to claim 1, characterized in that: The key elements described in step 3 include Si, Na, Al, Ca, Mg, Fe, Ti, K, Cr, Mn, P, and F.

6. The method for rapidly locating high-magnesium substances in lunar soil samples according to claim 1, characterized in that: Step 6: First, the energy spectrum data of each pixel point is denoised and background corrected to remove the influence of coating elements and ensure data quality. The energy spectrum data is reduced in dimension and feature extracted by chemometric methods, and pixels with similar chemical compositions are classified into the same phase in combination with clustering algorithms. In order to avoid the merging of trace phases and the loss of high-magnesium substances with low content of concern, boundary tolerance and grouping level are set to accurately distinguish the boundaries and chemical compositions of each phase to ensure accurate identification of high-magnesium substances. The boundary tolerance controls the behavior at the phase boundaries and is calculated as: T boundary = ， in, T boundary is the boundary tolerance, which represents the average value of pixel errors at the phase boundary and is used to measure the accuracy of phase boundary identification as a whole. The smaller the value, the more accurate the boundary classification. is the phase boundary pixel error, which is used to measure the phase boundary recognition accuracy and reflects the accuracy of pixel classification at the phase boundary, and is determined by the signal-to-noise ratio of the energy spectrum data; n Represents the total number of phase boundary pixels, that is, the number of boundary pixels involved in the calculation; Low boundary tolerance: each pixel represents a pure spectrum, suitable for high-purity phase analysis; High boundary tolerance: All pixels are fitted to the closest phase, suitable for fast classification of complex samples; Grouping level: Combine similar phases through intelligent grouping to create a small and manageable number of phases; low grouping level is suitable for trace compound analysis; high grouping level is suitable for large area analysis.

Images