A method and system for semi-quantitative detection of crystal phase of silicon carbonitride product

Abstract

The application provides a method and system for semi-quantitative detection of crystal phases of silicon carbonitride products. The method comprises the following steps: collecting X-ray diffraction full spectrum raw data of the silicon carbonitride product, performing background deduction and noise smoothing processing on the raw data to generate net diffraction intensity spectrum data; performing diffraction peak position identification and peak shape decomposition processing on the net diffraction intensity spectrum data to generate multi-phase characteristic spectrum parameter data; and combining an improved RIR value correction model and a full spectrum fitting algorithm, so that the mass fraction of each crystal phase and amorphous phase can be accurately inverted even if the silicon carbonitride product has columnar crystal preferred orientation and solid solution component fluctuation, the system deviation caused by the fact that the random orientation assumption of the traditional reference intensity method is not established is effectively avoided, and the detection accuracy of samples with high amorphous phase content is greatly improved.

Description

Technical Field

[0001] This invention proposes a semi-quantitative detection method and system for the crystalline phase of silicon carbonitride products, belonging to the field of inorganic non-metallic material detection technology. Background Technology

[0002] Silicon carbonitride composite materials, with their high high-temperature strength, good thermal shock resistance, and outstanding oxidation resistance, have been widely used in ceramic bearings, high-temperature kiln furniture, and various wear-resistant parts. The final performance of these materials is directly related to the types and proportions of their internal crystalline phases. The ratio of silicon carbide whisker phase, silicon nitride columnar phase, and amorphous phase plays a crucial role in the mechanical properties and service life of the product. Therefore, accurate semi-quantitative detection of each crystalline phase is an indispensable part of production quality control.

[0003] Currently, the industry commonly uses X-ray diffraction combined with the reference intensity method for quantitative phase analysis of ceramic products. This method is based on the premise that the sample is a powder with completely random orientation, and the distribution of grains in space has no directional preference. However, during the sintering process of silicon carbonitride materials, silicon nitride grains tend to grow along specific directions to form columnar crystal structures, resulting in a significant preferred orientation phenomenon. Simultaneously, the composition of the carbonitride solid solution also fluctuates. These factors lead to a significant deviation between the actual diffraction intensity and the theoretical intensity under ideal random orientation conditions. Especially when the proportion of amorphous phase in the sample is high, the calculation deviation of the traditional reference intensity method is further amplified, resulting in insufficient reliability of the test results and making it difficult to meet the practical needs of refined quality control. Summary of the Invention

[0004] This invention provides a semi-quantitative detection method and system for the crystalline phase of silicon carbonitride products, to solve the problems mentioned in the background section above: This invention proposes a semi-quantitative method for detecting the crystalline phase of silicon carbonitride products, the method comprising: S1. Collect raw X-ray diffraction full spectrum data of silicon carbonitride products, perform background subtraction and noise smoothing on the raw data to generate net diffraction intensity spectrum data; perform diffraction peak position identification and peak shape decomposition on the net diffraction intensity spectrum data to generate multiphase characteristic spectrum parameter data. S2. Perform optimal orientation factor quantization calculation on the multiphase characteristic spectrum parameter data, apply the improved RIR value correction model to correct the crystal orientation weight of the traditional reference intensity ratio, and generate orientation-corrected RIR parameter data; perform solid solution composition fluctuation compensation calculation on the orientation-corrected RIR parameter data to generate compensated RIR parameter data. S3. Perform full-spectrum fitting model construction on the compensated RIR parameter data, and weightedly correlate the multiphase characteristic spectrum parameter data with the standard spectrum library of each crystalline phase and the reference spectrum of the amorphous phase to generate the initial constraint model data for full-spectrum fitting; perform least squares iterative optimization on the initial constraint model data for full-spectrum fitting to generate converged fitting spectrum data. S4. Perform mass fraction inversion calculations for each phase on the converged fitted spectrum data, and solve the fitting results together with the corrected RIR parameters to generate semi-quantitative mass fraction data for each crystalline and amorphous phase; perform total normalization and cross-validation on the semi-quantitative mass fraction data to generate corrected mass fraction data. S5. Perform quality control threshold comparison processing on the corrected mass fraction data, compare the mass fraction of each phase with the preset raw material and finished product quality control standards item by item, and generate quality control judgment data; perform batch correlation and trend analysis processing on the quality control judgment data to generate full-link crystal phase quality control data from raw materials to finished products.

[0005] This invention proposes a semi-quantitative detection system for the crystalline phase of silicon carbonitride products, the system comprising: One or more processors; Memory, used to store one or more programs; Wherein, when the one or more programs are executed by the one or more processors, the one or more processors are made to implement the method described in any one of the above.

[0006] The beneficial effects of this invention are as follows: By combining an improved RIR value correction model with a full-spectrum fitting algorithm, it can accurately invert the mass fraction of each crystalline and amorphous phase even when silicon carbonitride products exhibit preferred columnar crystal orientation and solid solution composition fluctuations. This effectively avoids the systematic bias caused by the invalidation of the random orientation assumption in the traditional reference intensity method, significantly improving the detection accuracy of samples with high amorphous phase content. Simultaneously, the dual processing of orientation correction and solid solution compensation significantly reduces calculation errors caused by grain orientation and composition fluctuations, enhancing the comparability and stability of detection results across different batches of products. This method can rapidly screen the crystalline phase ratio at the raw material powder stage and accurately determine the quality of sintered finished products, reducing the risk of batch quality accidents due to detection deviations, reducing the time and cost of repeated testing and manual interpretation, and avoiding misjudgments of product performance due to distorted quality control data. Attached Figure Description

[0007] Figure 1 This is a diagram illustrating the steps of the method described in this invention; Figure 2 This is a detailed flowchart of step S3 in this invention. Detailed Implementation

[0008] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0009] One embodiment of the present invention, such as Figure 1 As shown, a semi-quantitative method for detecting the crystalline phase of silicon carbonitride products is provided, the method comprising: S1. Collect raw X-ray diffraction full spectrum data of silicon carbide products, perform background subtraction and noise smoothing on the raw data to generate net diffraction intensity spectrum data; perform diffraction peak position identification and peak shape decomposition on the net diffraction intensity spectrum data, extract characteristic diffraction peak parameters and diffuse scattering envelope parameters of silicon carbide whisker phase, silicon nitride columnar phase and amorphous phase, and generate multiphase characteristic spectrum parameter data. S2. Perform optimal orientation factor quantization calculation on the multiphase characteristic spectrum parameter data, apply the improved RIR value correction model to correct the crystal orientation weight of the traditional reference intensity ratio, and generate orientation-corrected RIR parameter data; perform solid solution composition fluctuation compensation calculation on the orientation-corrected RIR parameter data to generate compensated RIR parameter data. S3. Perform full-spectrum fitting model construction on the compensated RIR parameter data, and weightedly correlate the multiphase characteristic spectrum parameter data with the standard spectrum library of each crystalline phase and the reference spectrum of the amorphous phase to generate the initial constraint model data for full-spectrum fitting; perform least squares iterative optimization on the initial constraint model data for full-spectrum fitting to generate converged fitting spectrum data. S4. Perform mass fraction inversion calculations for each phase on the converged fitted spectrum data, and solve the fitting results together with the corrected RIR parameters to generate semi-quantitative mass fraction data for each crystalline and amorphous phase; perform total normalization and cross-validation on the semi-quantitative mass fraction data to generate corrected mass fraction data. S5. Perform quality control threshold comparison processing on the corrected mass fraction data, compare the mass fraction of each phase with the preset raw material and finished product quality control standards item by item, and generate quality control judgment data; perform batch correlation and trend analysis processing on the quality control judgment data to generate full-link crystal phase quality control data from raw materials to finished products.

[0010] The working principle and effects of the above technical solution are as follows: By conducting silicon carbonitride composition analysis through a complete crystal phase detection process, the accuracy of multi-phase content measurement can be improved, reducing numerical deviations caused by grain-oriented growth and compositional fluctuations. It can enhance the analytical depth of diffraction pattern data, reducing judgment errors caused by signal noise and peak overlap. It can avoid result distortion caused by parameter mismatch in traditional calculation methods, preventing biased conclusions from a single reference standard. It can accurately distinguish the actual proportions of different crystalline and amorphous phases, and can stabilize the detection output results through multi-round iterative calculations. Furthermore, it can reduce the probability of unqualified materials flowing into subsequent processes, extend the coverage of product quality control, fully present the phase structure change patterns throughout the production process, and effectively improve the overall product quality control level.

[0011] In one embodiment of the present invention, S1 includes: S11. Set up the X-ray diffraction detection equipment and complete the baseline stability calibration of the equipment. Place the silicon carbonitride test sample and carry out full-range diffraction scanning. Collect the complete diffraction signal information corresponding to the sample and summarize the raw data of the full X-ray diffraction spectrum. S12. Perform a substrate signal stripping operation on the acquired full-spectrum raw data to filter out invalid signals caused by environmental clutter and instrument inherent disturbances, complete the signal level regularization and noise reduction operation, and obtain the net diffraction intensity spectrum data after processing. S13. Based on the peak discrimination logic of the diffraction signal, traverse the net diffraction intensity spectrum data, lock the diffraction peak positions corresponding to different objects, split the overlapping and coupled diffraction contour morphology, and obtain independent peak structure information after splitting. S14. Extract the diffraction characteristics of silicon carbide whisker phase, the diffraction characteristics of silicon nitride columnar phase, and the diffuse scattering information of amorphous phase respectively, and count the corresponding peak values, peak width values, and scattering interval values ​​of each type of material. S15. Integrate all statistically obtained phase diffraction-related values, classify and summarize the diffraction characterization information specific to different phases, and integrate and summarize to form multiphase characteristic spectrum parameter data.

[0012] The working principle and effects of the above technical solution are as follows: By standardizing equipment calibration and full-domain scanning acquisition methods, the integrity and stability of the original diffraction data of silicon carbonitride samples can be guaranteed, reducing anomalies in the original data caused by equipment fluctuations and environmental interference. By layer-by-layer stripping of the substrate signal and noise reduction processing, the purity of the diffraction signal can be significantly improved, reducing the interference of invalid noise on subsequent analysis. By accurately identifying peak positions and splitting overlapping peak shapes, the problem of insufficient resolution in complex diffraction patterns can be improved, avoiding the problem of inaccurate feature extraction caused by mutual interference between diffraction signals of adjacent phases. It can comprehensively capture the diffraction feature information of various crystalline and amorphous phases, and accurately extract the core spectral parameters of various phases, effectively improving the precision and reliability of the preliminary feature data, providing high-quality basic data support for subsequent quantitative analysis of crystalline phases, and reducing the generation of detection errors from the source.

[0013] In one embodiment of the present invention, S13 includes: The net diffraction intensity spectrum data that has been normalized is traversed throughout the entire domain. The signal fluctuation and change patterns are compared segment by segment to distinguish the inherent diffraction signal of the phase from the superimposed interference waveform and to divide the independent and effective diffraction signal segments. Based on the diffraction energy distribution variation law, locate the extreme points within the segment, mark the peak positions corresponding to different crystal structures in sequence, and obtain the distribution information of diffraction peak points in the whole domain in batches. The contour shape of multiple overlapping and fused peaks is separated, and the composite peak is separated according to the energy decay change law. After separation, the complete diffraction contour shape corresponding to a single object is obtained. The height, span, and offset of each of the various contour shapes that have been separated are calculated one by one, and the quantitative indicators of each shape are collected in batches. All morphological quantitative indicators and point distribution information are collected, and the information is classified and collected according to the phase category, and combined to form independent peak structure information with complete structural characteristics.

[0014] The working principle and effects of the above technical solution are as follows: By segmentally identifying the signal differences in the net diffraction intensity spectrum data, the effective diffraction signal and interference waveform can be completely separated, reducing the negative impact of redundant signals on peak identification. Based on the diffraction energy distribution law, peak location and composite peak separation can be completed, solving the industry problem of difficulty in distinguishing overlapping and fused multi-phase diffraction peaks, and avoiding the problem of phase identification distortion caused by direct statistical analysis of mixed peaks. By uniformly calculating the morphological indicators of various diffraction profiles, the uniformity and accuracy of peak parameter measurement can be improved. It can completely preserve the diffraction structure characteristics of different crystalline and amorphous phases, and achieve refined segmentation and quantitative statistics of complex spectrum peaks, significantly improving the precision of diffraction peak structure analysis. This reduces the deviation in crystalline phase feature extraction during data analysis, providing accurate and reliable core data support for subsequent crystalline phase parameter calculation and content detection.

[0015] In one embodiment of the present invention, S2 includes: S21. Retrieve the diffraction distribution information of each crystal phase within the multiphase characteristic spectrum parameter data, statistically analyze the signal proportion value corresponding to the grain arrangement direction, convert the bias value corresponding to the grain growth arrangement, and obtain the preferred orientation factor value. S22. Introduce the influence coefficient of grain spatial arrangement to adjust the conventional reference strength calculation value, offset the numerical deviation caused by grain directional growth, adjust the corresponding calculation value of reference strength, and calculate the orientation-corrected RIR parameter data. S23. Statistically analyze the distribution changes of mutual doping and penetration of elements within the silicon carbonitride system, determine the fluctuation range of the component proportion within the lattice, and define the parameter offset caused by component fluctuations. S24. Combine the component offset amplitude to perform deviation compensation calculation on the orientation-corrected RIR parameters to make up for the calculation deviation caused by the change of solid solution components. After completing the deviation compensation process, the compensated RIR parameter data is generated.

[0016] The working principle and effects of the above technical solution are as follows: By quantifying the degree of bias in grain arrangement to complete the orientation factor calculation, the grain growth state of silicon carbonitride samples can be accurately restored, reducing detection errors caused by grain orientation structure. By performing crystal orientation correction on conventional reference intensity values, the shortcomings of traditional calculation methods that ignore the influence of grain orientation can be improved, avoiding the problem of overall shift in detection results caused by preferential crystal growth. By investigating the internal elemental doping and lattice composition fluctuations of the material, parameter deviations caused by changes in solid solution structure can be accurately captured. This not only eliminates systematic errors caused by grain orientation but also offsets calculation interference caused by composition fluctuations, significantly improving the matching accuracy and adaptability of RIR parameters, making subsequent quantitative calculations of crystal phases more consistent with the actual structural state of the sample, and effectively enhancing the stability and data accuracy of the entire crystal phase detection process.

[0017] In one embodiment of the present invention, step S22 includes: Collect the grain spatial distribution data corresponding to the preferred orientation factor, include the basic calculation values ​​of the traditional reference strength calculation process, and summarize to generate the original reference strength calculation data; The degree of interference of the grain orientation growth state on the diffraction response intensity is measured, the calculation deviation caused by different crystal orientation structures is statistically analyzed, and grain orientation deviation statistics are generated. The influence coefficient of embedded grain spatial arrangement is used in the reference strength calculation to perform point-by-point fine-tuning of the original reference strength value and generate preliminary corrected reference strength data; Eliminate the overall calculation offset caused by grain orientation growth, balance the calculation uniformity of polycrystalline phase diffraction intensity, and complete the deviation correction processing of reference intensity data; The corrected reference intensity data are standardized and integrated, and the data operation dimensions and measurement standards are unified to generate orientation-corrected RIR parameter data.

[0018] The working principle and effects of the above technical solution are as follows: Completely capturing the original calculation-related values ​​preserves the overall picture of the initial calculation, reducing correction deviations caused by missing basic data. Accurately determining the interference range caused by grain growth allows for a clear understanding of the impact of orientation factors on intensity values, preventing inaccurate control of the deviation range from affecting subsequent calibration results. Step-by-step adjustment of the reference intensity values ​​can gradually offset adverse effects, weakening data offset problems caused by directional growth. Balancing the calculation scale of diffraction intensities for various crystals can reduce the calculation gap between different phases, avoiding imbalanced results caused by single-dimensional evaluation. Unifying data measurement specifications and completing integration and summarization ensures that the corrected parameters closely match the actual crystallization state of the material, improves the adaptability of subsequent parameter calculations, effectively enhances the reliability of reference intensity-related values, and reduces the negative impact of orientation problems on crystal phase detection results.

[0019] One embodiment of the present invention, such as Figure 2 As shown, S3 includes: S31. Retrieve the basic operational parameters of the RIR parameter data after compensation, retrieve the standard diffraction files of various crystalline phases and the reference diffraction files of amorphous phases, and build a correlation operation system for the mutual correspondence of multiple diffraction patterns. S32. Assign computational weight coefficients to different phase spectra, bind measured characteristic spectrum parameters and standard archive spectrum information, and combine them to form full-spectrum fitting initial constraint model data with constraint conditions. S33. Set the error convergence interval and loop termination condition for the graph fitting operation, and continuously adjust the internal parameter ratio of the model with the residual minimization operation logic to carry out the cyclic calculation operation without interruption. S34. Continuously reduce the numerical difference between the measured spectrum and the fitted spectrum until the parameter variation falls within the set range, then terminate the cyclic calculation process and generate converged fitted spectrum data after stabilization.

[0020] The working principle and effects of the above technical solution are as follows: Integrating various parameters and standard spectra to construct a computational system opens up a comparison channel between measured data and reference samples, reducing data inconsistencies during spectra comparison. Differentiating phase differences and configuring computational weights, along with constraints to build a fitting model, allows for accurate representation of diffraction patterns across different structures, avoiding insufficient fit caused by standardized calculations. Defining error ranges and stopping limits for iterative calculations controls the pace of parameter adjustments, reducing processing time wasted on blind calculations. Continuously narrowing the numerical differences in spectra to achieve model convergence not only brings the fitted results closer to the actual sample diffraction state but also improves model adaptability, significantly reducing the deviation values ​​generated by fitting calculations, enhancing the matching accuracy of the full-spectrum analysis, and providing reliable spectral data that closely matches the characteristics of the actual sample for subsequent phase content inversion calculations.

[0021] In one embodiment of the present invention, S32 includes: Read the measured diffraction characteristic parameters and various standard spectrum archive data in the full spectrum fitting correlation operation system, collect the spectrum response characteristic information of different crystalline phases and amorphous phases, and generate the fitting basic dataset; Based on the differences in the reliability of the characterization of various phase diffraction signals, differentiated operation weight coefficients are assigned to the matching operation of various phase spectra to generate phase weight allocation data. The weighted standard spectral information is linked and bound with the measured feature spectral parameters to complete the fusion and docking of multi-source spectral data and generate spectral fusion and association data. Numerical constraints and matching boundary rules are added to the spectrum fitting operation, and the fused and correlated data are conditionally constrained and regularized to generate initial constraint model data for full spectrum fitting with constraint constraints.

[0022] The working principle and effects of the above technical solution are as follows: By aggregating feature information from multiple types of spectra to form complete basic data, the data dimensions of fitting operations can be supplemented, reducing the analytical bias caused by a single data source. By setting differentiated operational weights for each phase, the true characterization capability of diffraction signals from different phases can be matched, avoiding the problem of weakening weak signal features by equal weighting operations. The fusion and integration of multi-source spectra data enables deep matching between measured data and standard data, improving the tightness of data correlation. Adding fitting operation constraints and boundary rules can standardize the model's operational range and prevent model distortion caused by disordered data fitting. It can balance the fitting accuracy of strong signal phases with the analytical effectiveness of weak phase signals, effectively improving the fit and rationality of the initial fitting model, reducing subsequent iteration errors caused by early model defects, and making the full-spectrum fitting operation process more rigorous and standardized.

[0023] In one embodiment of the present invention, S33 includes: Read the parameter ratio information and spectrum matching rules inside the initial constraint model data of the full spectrum fitting, statistically analyze the distribution of the initial fitting residuals of the model, and generate the initial calculation error data of the model. The allowable error fluctuation range of the fitting operation is calibrated, the stopping criteria of the iterative operation are defined, and the error convergence interval and loop termination condition data corresponding to the spectral fitting are solidified. The model parameter iterative adjustment operation is initiated based on the residual minimization operation logic, and the ratio values ​​of phase weights, peak shape parameters and baseline parameters inside the model are finely adjusted one by one to generate multiple rounds of parameter adjustment data; After each round of parameter adjustment, the system automatically performs a graph fitting recalculation operation, compares the residual change of the fitting results between the two rounds, continuously promotes the optimization and iteration process of model parameters, and generates intermediate data for iterative calculations.

[0024] The working principle and effects of the above technical solution are as follows: By statistically analyzing the initial residual distribution of the model, the error basis of the initial fitting can be accurately grasped, reducing the problem of deviation in the direction of iterative optimization caused by an unclear error base. By fixing the error convergence range and the iteration termination criterion, the evaluation scale of model optimization can be unified, avoiding over-computation or premature termination during the iteration process. By fine-tuning the ratio of various parameters within the model in multiple dimensions, fitting deviations can be specifically corrected, improving the model's adaptability to complex diffraction patterns. By recalculating and comparing residual changes round by round, the optimization effect can be continuously controlled, and the iteration progress rhythm can be stabilized. This not only gradually reduces the overall error of the fitting operation but also ensures the rationality of parameter adjustments in each round, effectively enhancing the accuracy and stability of iterative optimization, reducing the fitting distortion problem caused by disordered parameter adjustments, and providing reliable iterative optimization support for obtaining high-precision converged fitting spectrum data.

[0025] In one embodiment of the present invention, step S4 includes: S41. Extract the diffraction fitting results of various phases contained in the convergent fitting spectrum data, perform simultaneous numerical solutions on the RIR parameter data after linkage compensation, and calculate the mass percentage of different phases layer by layer. S42. Calculate the mass percentage of silicon carbide whisker phase, silicon nitride columnar phase and amorphous phase in sequence, summarize all phase calculation values ​​and generate multi-phase semi-quantitative mass fraction data. S43. Calculate the total mass fraction of all phases, adjust the individual percentage values ​​according to the unified conversion rules of the total, ensure that the overall percentage values ​​conform to the metrological balance law, and complete the value normalization process. S44. Cross-compare the fitting and extrapolation results with diffraction feature characterization information, identify abnormal points where the values ​​deviate and complete the correction. After the correction is completed, generate the corrected quality score data.

[0026] The working principle and effects of the above technical solution are as follows: By combining the fitted spectrum results with correction parameters for joint calculation, the mass proportion of each phase can be accurately decomposed layer by layer, reducing the calculation bias caused by single data calculation. By separately statistically analyzing the proportion values ​​of different crystalline and amorphous phases, the entire internal phase structure of the sample can be fully covered, avoiding the omission of local phases that could lead to biased detection results. Through global total amount normalization, the numerical imbalance caused by multi-phase accumulation can be corrected, ensuring that the proportion values ​​meet the basic metrological specifications and preventing the accumulation of total amount deviations. By cross-checking the fitted data and diffraction characteristics, abnormal values ​​can be captured and corrected in a timely manner. This not only ensures the integrity of multi-phase detection results but also improves the accuracy and balance of mass fraction data, effectively reducing the small errors left over from iterative calculations and spectrum fitting, and significantly improving the reliability and accuracy of semi-quantitative detection results of the crystalline phase of silicon carbonitride products.

[0027] In one embodiment of the present invention, S43 includes: The single-calculation percentage values ​​of various crystalline and amorphous phases in the multi-phase semi-quantitative mass fraction data are collected, and the mass fraction values ​​of all phases are summed to generate the total total mass fraction data of all phases. By comparing the deviation difference between the total aggregate data and the standard measurement balance value, the total offset error caused by multi-parameter fitting calculation is captured, and total deviation statistics are generated. The total amount of the whole domain is converted into equal distribution. Based on the original proportion of each phase, the deviation value is allocated, and the mass fraction of each phase is finely adjusted item by item to generate preliminary normalized fraction data. Verify the overall metrological balance after the summation of the fractions of each phase after fine-tuning, eliminate the numerical accumulation error caused by multiple rounds of iterative calculations, and complete the standardization and normalization of the mass fractions of all phases.

[0028] The working principle and effects of the above technical solution are as follows: By summarizing the mass fraction values ​​of all phases, the overall metrological situation after multiphase accounting can be fully grasped, and the total deviation caused by the cumulative fitting calculation can be accurately located. By quantifying the difference between the total and the standard metrological value, the source of error caused by the superposition of various parameter calculations can be accurately identified, reducing the problem of hidden and undetected errors. By allocating the deviation according to the original proportion and fine-tuning the individual values, the overall metrological deviation can be uniformly resolved, avoiding the distortion of results caused by concentrated correction of single phase data. By verifying the overall metrological balance state a second time, the cumulative error left by multiple iterations can be completely eliminated. It can not only retain the structural characteristics of the original proportion of each phase, but also ensure that the mass fraction of all phases meets the metrological balance standard, effectively improving the regularity and consistency of semi-quantitative detection data, reducing the interference of numerical deviation on the final crystal phase detection results, and making the mass proportion data more consistent with the actual phase composition of the sample.

[0029] In one embodiment of the present invention, step S5 includes: S51. Retrieve the phase ratio results of the corrected mass fraction data record, retrieve the critical judgment values ​​corresponding to the raw material control specifications and the finished product outgoing control specifications, and complete the numerical comparison work in sequence. S52. By comparing the critical judgment values, divide the phase ratio into qualified and abnormal ranges, distinguish the grade range of each index of the sample, and generate quality control judgment data after the division is completed. S53. Collect quality control judgment data of raw material samples, intermediate processing samples and final finished product samples under the same production batch, and establish the corresponding correlation between sample batches and test values. S54. Continuously observe the changes in the proportion of phases in multiple batches of samples, sort out the changes in phase structure throughout the production process, and integrate the detection information throughout the process to generate full-chain crystal phase quality control data.

[0030] The working principle and effects of the above technical solution are as follows: By comparing the measured phase ratio with the control standard values ​​one by one, the actual level of the sample indicators can be intuitively determined, reducing the judgment deviation caused by subjective evaluation. Dividing the grade range according to the critical values ​​can clearly define the qualified and abnormal ranges, avoiding the problem of inaccurate quality judgment caused by vague divisions. Connecting the sample testing information of each stage within a batch can streamline the data correlation of each production link, reducing data fragmentation caused by segmented control. Tracking the trend of phase ratio changes in multiple batches can clarify the structural evolution characteristics of the processing process. It can not only complete the quality identification of a single sample, but also control the structural stability of the entire production line as a whole, effectively improving the rigor of product quality screening, reducing the probability of unqualified products flowing to subsequent processes, and providing a real phase change reference for process parameter adjustment, thus consolidating the overall production quality control level.

[0031] One embodiment of the present invention provides a semi-quantitative detection system for the crystalline phase of silicon carbonitride products, the system comprising: One or more processors; Memory, used to store one or more programs; Wherein, when the one or more programs are executed by the one or more processors, the one or more processors are made to implement the method described in any one of the above.

[0032] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A semi-quantitative method for detecting the crystalline phase of silicon carbonitride products, characterized in that, The method includes: S1. Collect raw X-ray diffraction full spectrum data of silicon carbonitride products, perform background subtraction and noise smoothing on the raw data to generate net diffraction intensity spectrum data; perform diffraction peak position identification and peak shape decomposition on the net diffraction intensity spectrum data to generate multiphase characteristic spectrum parameter data. S2. Perform optimal orientation factor quantization calculation on the multiphase characteristic spectrum parameter data, apply the improved RIR value correction model to correct the crystal orientation weight of the traditional reference intensity ratio, and generate orientation-corrected RIR parameter data; perform solid solution composition fluctuation compensation calculation on the orientation-corrected RIR parameter data to generate compensated RIR parameter data. S3. Perform full-spectrum fitting model construction on the compensated RIR parameter data, and weightedly correlate the multiphase characteristic spectrum parameter data with the standard spectrum library of each crystalline phase and the reference spectrum of the amorphous phase to generate the initial constraint model data for full-spectrum fitting; perform least squares iterative optimization on the initial constraint model data for full-spectrum fitting to generate converged fitting spectrum data. S4. Perform mass fraction inversion calculations for each phase on the converged fitted spectrum data, and solve the fitting results together with the corrected RIR parameters to generate semi-quantitative mass fraction data for each crystalline and amorphous phase; perform total normalization and cross-validation on the semi-quantitative mass fraction data to generate corrected mass fraction data. S5. Perform quality control threshold comparison processing on the corrected mass fraction data, compare the mass fraction of each phase with the preset raw material and finished product quality control standards item by item, and generate quality control judgment data; perform batch correlation and trend analysis processing on the quality control judgment data to generate full-link crystal phase quality control data from raw materials to finished products.

2. The semi-quantitative detection method for the crystalline phase of silicon carbonitride products according to claim 1, characterized in that, S1 includes: S11. Set up the X-ray diffraction detection equipment and complete the baseline stability calibration of the equipment. Place the silicon carbonitride test sample and carry out full-range diffraction scanning. Collect the complete diffraction signal information corresponding to the sample and summarize the raw data of the full X-ray diffraction spectrum. S12. Perform a substrate signal stripping operation on the acquired full-spectrum raw data to filter out invalid signals caused by environmental clutter and instrument inherent disturbances, complete the signal level regularization and noise reduction operation, and obtain the net diffraction intensity spectrum data after processing. S13. Based on the peak discrimination logic of the diffraction signal, traverse the net diffraction intensity spectrum data, lock the diffraction peak positions corresponding to different objects, split the overlapping and coupled diffraction contour morphology, and obtain independent peak structure information after splitting. S14. Extract the diffraction characteristics of silicon carbide whisker phase, the diffraction characteristics of silicon nitride columnar phase, and the diffuse scattering information of amorphous phase respectively, and count the corresponding peak values, peak width values, and scattering interval values ​​of each type of material. S15. Integrate all statistically obtained phase diffraction-related values, classify and summarize the diffraction characterization information specific to different phases, and integrate and summarize to form multiphase characteristic spectrum parameter data.

3. The semi-quantitative detection method for the crystalline phase of silicon carbonitride products according to claim 1, characterized in that, S2 includes: S21. Retrieve the diffraction distribution information of each crystal phase within the multiphase characteristic spectrum parameter data, statistically analyze the signal proportion value corresponding to the grain arrangement direction, convert the bias value corresponding to the grain growth arrangement, and obtain the preferred orientation factor value. S22. Introduce the influence coefficient of grain spatial arrangement to adjust the conventional reference strength calculation value, offset the numerical deviation caused by grain directional growth, adjust the corresponding calculation value of reference strength, and calculate the orientation-corrected RIR parameter data. S23. Statistically analyze the distribution changes of mutual doping and penetration of elements within the silicon carbonitride system, determine the fluctuation range of the component proportion within the lattice, and define the parameter offset caused by component fluctuations. S24. Combine the component offset amplitude to perform deviation compensation calculation on the orientation-corrected RIR parameters to make up for the calculation deviation caused by the change of solid solution components. After completing the deviation compensation process, the compensated RIR parameter data is generated.

4. The semi-quantitative detection method for the crystalline phase of silicon carbonitride products according to claim 3, characterized in that, S22 includes: Collect the grain spatial distribution data corresponding to the preferred orientation factor, include the basic calculation values ​​of the traditional reference strength calculation process, and summarize to generate the original reference strength calculation data; The degree of interference of the grain orientation growth state on the diffraction response intensity is measured, the calculation deviation caused by different crystal orientation structures is statistically analyzed, and grain orientation deviation statistics are generated. The influence coefficient of embedded grain spatial arrangement is used in the reference strength calculation to perform point-by-point fine-tuning of the original reference strength value and generate preliminary corrected reference strength data; Eliminate the overall calculation offset caused by grain orientation growth, balance the calculation uniformity of polycrystalline phase diffraction intensity, and complete the deviation correction processing of reference intensity data; The corrected reference intensity data are standardized and integrated, and the data operation dimensions and measurement standards are unified to generate orientation-corrected RIR parameter data.

5. The semi-quantitative detection method for the crystalline phase of silicon carbonitride products according to claim 1, characterized in that, The S3 includes: S31. Retrieve the basic operational parameters of the RIR parameter data after compensation, retrieve the standard diffraction files of various crystalline phases and the reference diffraction files of amorphous phases, and build a correlation operation system for the mutual correspondence of multiple diffraction patterns. S32. Assign computational weight coefficients to different phase spectra, bind measured characteristic spectrum parameters and standard archive spectrum information, and combine them to form full-spectrum fitting initial constraint model data with constraint conditions. S33. Set the error convergence interval and loop termination condition for the graph fitting operation, and continuously adjust the internal parameter ratio of the model with the residual minimization operation logic to carry out the cyclic calculation operation without interruption. S34. Continuously reduce the numerical difference between the measured spectrum and the fitted spectrum until the parameter variation falls within the set range, then terminate the cyclic calculation process and generate converged fitted spectrum data after stabilization.

6. The semi-quantitative detection method for the crystalline phase of silicon carbonitride products according to claim 5, characterized in that, S32 includes: Read the measured diffraction characteristic parameters and various standard spectrum archive data in the full spectrum fitting correlation operation system, collect the spectrum response characteristic information of different crystalline phases and amorphous phases, and generate the fitting basic dataset; Based on the differences in the reliability of the characterization of various phase diffraction signals, differentiated operation weight coefficients are assigned to the matching operation of various phase spectra to generate phase weight allocation data. The weighted standard spectral information is linked and bound with the measured feature spectral parameters to complete the fusion and docking of multi-source spectral data and generate spectral fusion and association data. Numerical constraints and matching boundary rules are added to the spectrum fitting operation, and the fused and correlated data are conditionally constrained and regularized to generate initial constraint model data for full spectrum fitting with constraint constraints.

7. The semi-quantitative detection method for the crystalline phase of silicon carbonitride products according to claim 5, characterized in that, S33 includes: Read the parameter ratio information and spectrum matching rules inside the initial constraint model data of the full spectrum fitting, statistically analyze the distribution of the initial fitting residuals of the model, and generate the initial calculation error data of the model. The allowable error fluctuation range of the fitting operation is calibrated, the stopping criteria of the iterative operation are defined, and the error convergence interval and loop termination condition data corresponding to the spectral fitting are solidified. The model parameter iterative adjustment operation is initiated based on the residual minimization operation logic, and the ratio values ​​of phase weights, peak shape parameters and baseline parameters inside the model are finely adjusted one by one to generate multiple rounds of parameter adjustment data; After each round of parameter adjustment, the system automatically performs a graph fitting recalculation operation, compares the residual change of the fitting results between the two rounds, continuously promotes the optimization and iteration process of model parameters, and generates intermediate data for iterative calculations.

8. The semi-quantitative detection method for the crystalline phase of silicon carbonitride products according to claim 1, characterized in that, The S4 includes: S41. Extract the diffraction fitting results of various phases contained in the convergent fitting spectrum data, perform simultaneous numerical solutions on the RIR parameter data after linkage compensation, and calculate the mass percentage of different phases layer by layer. S42. Calculate the mass percentage of silicon carbide whisker phase, silicon nitride columnar phase and amorphous phase in sequence, summarize all phase calculation values ​​and generate multi-phase semi-quantitative mass fraction data. S43. Calculate the total mass fraction of all phases, adjust the individual percentage values ​​according to the unified conversion rules of the total, ensure that the overall percentage values ​​conform to the metrological balance law, and complete the value normalization process. S44. Cross-compare the fitting and extrapolation results with diffraction feature characterization information, identify abnormal points where the values ​​deviate and complete the correction. After the correction is completed, generate the corrected quality score data.

9. The semi-quantitative detection method for the crystalline phase of silicon carbonitride products according to claim 1, characterized in that, The S5 includes: S51. Retrieve the phase ratio results of the corrected mass fraction data record, retrieve the critical judgment values ​​corresponding to the raw material control specifications and the finished product outgoing control specifications, and complete the numerical comparison work in sequence. S52. By comparing the critical judgment values, divide the phase ratio into qualified and abnormal ranges, distinguish the grade range of each index of the sample, and generate quality control judgment data after the division is completed. S53. Collect quality control judgment data of raw material samples, intermediate processing samples and final finished product samples under the same production batch, and establish the corresponding correlation between sample batches and test values. S54. Continuously observe the changes in the proportion of phases in multiple batches of samples, sort out the changes in phase structure throughout the production process, and integrate the detection information throughout the process to generate full-chain crystal phase quality control data.

10. A semi-quantitative detection system for the crystalline phase of silicon carbonitride products, characterized in that, The system includes: One or more processors; Memory, used to store one or more programs; Wherein, when the one or more programs are executed by the one or more processors, the one or more processors implement the method of any one of claims 1 to 9.

Images