New chemical material impurity content detection system based on product analysis and inversion reaction path

Abstract

The invention relates to the technical field of chemical material detection, in particular to a new chemical material impurity content detection system based on product analysis and reaction path inversion, which comprises a data processing module, an impurity identification module, a reaction path inversion module, an impurity content correction module and an information tracing module. According to the method, a production process model is constructed by integrating data of intermediate products and byproducts in each process stage, a polymerization reaction kinetic model is constructed by combining a polymerization reaction mechanism and kinetic data, and a detected impurity chemical structure is associated and compared with the data in the production process model; and reverse deduction is carried out by utilizing a polymerization reaction kinetic model, so that not only can a generation path of the impurity be reconstructed, a specific process of impurity generation be accurately determined and effective source tracing of an impurity source be realized, but also an initial impurity concentration measurement value is reversely corrected by utilizing the source tracing information, and detection deviation caused by different generation paths is eliminated.

Description

Technical Field

[0001] This invention relates to the field of chemical material testing technology, and in particular to a system for detecting the impurity content of new chemical materials based on product analysis and reaction pathway inversion. Background Technology

[0002] The field of chemical materials testing technology involves the detection and analysis of the properties, composition, and impurity content of various materials used in chemical engineering. This field encompasses all aspects, including material preparation, performance evaluation, quality control, and contaminant detection.

[0003] Among them, the chemical new material impurity content detection system refers to a system that determines the impurity content in new chemical materials through methods such as chemical analysis, spectroscopic analysis, and chromatographic analysis. This system typically identifies and quantifies various impurities in materials through sampling, separation, and detection steps, thus solving the problem of the potential impact of impurities in new chemical materials on their performance.

[0004] Existing chemical new material impurity content detection systems mainly rely on chemical analysis and spectral analysis for impurity separation and quantitative detection. This approach focuses on determining the impurity content in the final product, lacking in-depth investigation of the impurity source. When a specific impurity is detected, the system cannot automatically link it to the specific process or reaction stage in the production process, making it difficult to trace the impurity source. At the same time, the quantification of impurities in existing technologies is usually based on standard detection methods, without considering the potential measurement deviations caused by different formation paths of impurities in complex polymerization reactions. This makes it difficult for the detection results to accurately reflect the actual production, and it is difficult to provide direct and effective guidance for process optimization, thus delaying the response of quality control. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of existing technologies and propose a new chemical material impurity content detection system based on product analysis and reaction path inversion.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a chemical new material impurity content detection system based on product analysis and reaction pathway inversion includes:

[0007] The data processing module acquires the original spectra and chromatograms of impurities in the chemical new material samples, and extracts the impurity spectral data and impurity chromatographic data of the chemical new material from the original spectra and chromatograms.

[0008] The impurity identification module compares the impurity spectral data and impurity chromatographic data with a preset standard material spectral library to identify the chemical structure, impurity concentration and type of impurities in new chemical materials and generate impurity characteristic information.

[0009] The reaction path inversion module constructs a chemical new material production process model and a polymerization reaction kinetic model, associates the impurity characteristic information with the chemical new material production process model, and inputs it into the polymerization reaction kinetic model for reverse deduction to determine the process in which impurities are generated and generate impurity cause inversion information.

[0010] The impurity content correction module corrects the impurity characteristic information based on the impurity cause inversion information, generates the corrected impurity content, and locates the impurity source information.

[0011] The information traceability module summarizes the corrected impurity content and the impurity source information and performs formatting processing to obtain the material impurity content detection results.

[0012] As a further aspect of the present invention, the impurity spectral data includes absorption peak position and absorption peak intensity; the impurity chromatographic data includes retention time and chromatographic peak area; the impurity characteristic information includes chemical structure, impurity concentration, and impurity type; the impurity origin inversion information includes impurity generation process and key polymerization parameters; the corrected impurity content is specifically a calibration concentration value; the impurity source information is specifically a source process code; and the material impurity content detection results include visualization charts and data traceability indexes.

[0013] As a further aspect of the present invention, the data processing module includes:

[0014] The original spectrum acquisition submodule acquires the original spectra and chromatograms of impurities in chemical new material samples, identifies the signal response segments and noise baseline segments within the original spectra and chromatograms, quantifies the fluctuation amplitude of the noise baseline segments to determine the noise level, obtains the signal peak intensity of the signal response segments, calculates the ratio of the signal peak intensity to the noise level to obtain the signal-to-noise ratio, and compares the signal-to-noise ratio with the set signal-to-noise ratio benchmark value to filter the spectrum signal response intervals;

[0015] The spectral feature extraction submodule identifies the spectral signal response interval in the original spectral image, identifies all peak points within the interval, calculates the peak curvature and peak width of each peak point, removes noise peaks and baseline drift peaks based on the peak curvature reference and peak width range, extracts the position and intensity of the remaining absorption peaks, and generates spectral absorption peak features.

[0016] The chromatographic data combination submodule identifies the spectral signal response intervals in the original chromatogram and integrates the profile of each chromatographic peak to obtain the retention time and peak area of ​​the chromatographic peak. It combines the position and peak intensity of the absorption peaks in the spectral absorption peak characteristics and combines the retention time and peak area of ​​the chromatographic peaks to establish impurity spectral data and impurity chromatographic data of new chemical materials.

[0017] As a further aspect of the present invention, the impurity identification module includes:

[0018] The spectral matching and identification submodule obtains the position and retention time of the absorption peak from the impurity spectral data and impurity chromatographic data, searches for the position and retention time of the corresponding absorption peak in the preset standard substance spectral library, calculates the relative offset between the position of the absorption peak to be matched and the position of the absorption peak in the spectral library, and calculates the time difference between the retention time to be matched and the retention time in the spectral library. The relative offset and time difference are compared with the set position offset threshold and time difference threshold, respectively, to screen potential substance matching codes.

[0019] The structure type determination submodule retrieves the material structure information, standard spectral features, and standard chromatographic features corresponding to the potential substance matching code from the standard material spectral library, calls the peak intensity and peak area in the impurity spectral data, calculates the peak area ratio and peak intensity ratio of each item, classifies and determines the type of impurity based on the peak area ratio, and identifies the chemical structure and impurity type of the impurity.

[0020] The concentration information generation submodule obtains the quantitative calibration coefficient of the corresponding chemical structure from the standard material spectral library based on the chemical structure and impurity type of the impurity. It calculates the impurity concentration using the calibration coefficient based on the peak intensity recorded in the impurity spectral data and the peak area recorded in the impurity chromatographic data. It then combines the impurity chemical structure and impurity type with the impurity concentration to generate impurity characteristic information.

[0021] As a further aspect of the present invention, the reaction path inversion module includes:

[0022] The process model construction submodule collects intermediate product data and by-product data for each process stage in the production of new chemical materials, classifies the intermediate product data and by-product data to the corresponding process stage, and constructs a process model for the production of new chemical materials.

[0023] The kinetic model construction submodule collects polymerization reaction mechanism and kinetic data of new chemical materials, including reaction rate constant and activation energy parameters, and fits the reaction steps in the reaction mechanism with the kinetic data to construct a polymerization reaction kinetic model;

[0024] The impurity reverse deduction submodule calls the impurity chemical structure in the impurity characteristic information and calls the intermediate product and by-product data of each process stage in the chemical new material production process model to establish the correlation results. The correlation results are input into the polymerization reaction kinetic model for reverse deduction. Based on the reaction path and key parameter changes output by the deduction, impurity origin reverse deduction information is generated.

[0025] As a further aspect of the present invention, the process of calling the impurity chemical structure in the impurity feature information and calling the intermediate and by-product data of each process stage in the chemical new material production process model to establish the association result is as follows: the chemical structure of the impurity is compared with the chemical structure of the intermediate and by-products for structural similarity, and intermediate or by-products with structural similarity higher than a preset similarity threshold are selected, and the selected intermediate or by-products are used as the association result.

[0026] As a further aspect of the present invention, the process of inputting the correlation results into the polymerization reaction kinetic model for reverse deduction is specifically as follows: taking the intermediate products or by-products contained in the correlation results as the endpoint of the reverse reaction path, solving the differential equation of reactant concentration changing with time based on the polymerization reaction mechanism recorded in the polymerization reaction kinetic model, reconstructing the reaction path that generates intermediate products or by-products, and identifying the process containing the reaction path as the source of impurities.

[0027] As a further aspect of the present invention, the impurity content correction module includes:

[0028] The content correction calculation submodule obtains the preset correction parameters for the standard sample test and the impurity concentration in the impurity characteristic information, corrects the impurity concentration using the correction parameters, and obtains the corrected impurity content.

[0029] The source process extraction submodule analyzes the changes in key parameters corresponding to the impurity source process from the impurity origin inversion information, compares the impurity concentration in the corrected impurity content with the range of key parameter changes, verifies the correspondence between process information and impurity content, screens key processes, and obtains key information of the source process.

[0030] The impurity source definition submodule defines the process content in the key information of the source process as the impurity source and generates impurity source information.

[0031] As a further aspect of the present invention, the information traceability module includes:

[0032] The report generation submodule summarizes the corrected impurity content and the impurity source information, performs data filling and formatting processing, and generates a test report.

[0033] The quality early warning determination submodule compares the corrected impurity content with a preset quality standard threshold to determine whether a quality early warning is triggered, and integrates the early warning determination result with the test report to generate the material impurity content test result.

[0034] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0035] In this invention, a production process model is constructed by integrating intermediate and by-product data from each process stage, and a polymerization reaction kinetic model is constructed by combining polymerization reaction mechanism and kinetic data. The chemical structure of detected impurities is correlated and compared with the data in the production process model, and the polymerization reaction kinetic model is used for reverse deduction. This not only reconstructs the impurity generation path and accurately determines the specific process in which impurities are generated, thus achieving effective traceability of the impurity source, but also uses this traceability information to reverse correct the initial impurity concentration measurement value, eliminating detection deviations caused by different generation paths. This makes the impurity content results more consistent with the actual process and provides an accurate decision-making basis for production quality control. Attached Figure Description

[0036] Figure 1 This is a system flowchart of the present invention;

[0037] Figure 2 This is a flowchart of the data processing module of the present invention;

[0038] Figure 3 This is a flowchart of the impurity identification module of the present invention;

[0039] Figure 4 This is a flowchart of the reaction path inversion module of the present invention;

[0040] Figure 5 This is a flowchart of the impurity content correction module of the present invention;

[0041] Figure 6 This is a flowchart of the information traceability module of the present invention. Detailed Implementation

[0042] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0043] Please see Figure 1 The chemical new material impurity content detection system based on product analysis and reaction pathway inversion includes:

[0044] The data processing module acquires the original spectra and chromatograms of impurities in the chemical new material samples, and extracts the impurity spectral data and impurity chromatographic data of the chemical new material from the original spectra and chromatograms.

[0045] The impurity identification module compares impurity spectral data and impurity chromatographic data with a preset standard material spectral library to identify the chemical structure, concentration, and type of impurities in new chemical materials and generate impurity characteristic information.

[0046] The reaction path inversion module constructs a chemical new material production process model and a polymerization reaction kinetic model, associates impurity characteristic information with the chemical new material production process model, and inputs it into the polymerization reaction kinetic model for reverse deduction to determine the process in which impurities are generated and generate impurity cause inversion information.

[0047] The impurity content correction module corrects the impurity characteristic information based on the impurity cause inversion information, generates the corrected impurity content, and locates the impurity source information.

[0048] The information traceability module summarizes and formats the corrected impurity content and impurity source information to obtain the material impurity content test results.

[0049] Impurity spectral data includes absorption peak position and intensity; impurity chromatographic data includes retention time and peak area; impurity characteristic information includes chemical structure, concentration, and type; impurity origin inversion information includes impurity generation process and key polymerization parameters; corrected impurity content is the calibration concentration value; impurity source information is the source process code; and material impurity content detection results include visualization charts and data traceability indexes.

[0050] Please see Figure 2 The data processing module includes:

[0051] The original spectrum acquisition submodule acquires the original spectra and chromatograms of impurities in chemical new material samples, identifies the signal response segments and noise baseline segments within the original spectra and chromatograms, quantifies the fluctuation amplitude of the noise baseline segments to determine the noise level, obtains the signal peak intensity of the signal response segments, calculates the ratio of the signal peak intensity to the noise level to obtain the signal-to-noise ratio, and compares the signal-to-noise ratio with the set signal-to-noise ratio benchmark value to filter the spectrum signal response intervals;

[0052] Samples of novel chemical materials to be tested, such as a batch of polyethylene terephthalate (PET), were sliced ​​and analyzed using Fourier transform infrared spectroscopy (FTIR) and high-performance liquid chromatography (HPLC), respectively. The FTIR spectrometer scanned the sample in the range of 4000 to 400 wavelengths per minute (WPM) to obtain a raw spectrum containing absorption peaks and a baseline. The HPLC system operated under a preset mobile phase gradient to obtain a raw chromatogram containing chromatographic peaks and a baseline. For the acquired raw spectra, signal response regions were identified, such as the carbonyl stretching vibration absorption peak appearing in the 1800 to 1700 WPM range. Simultaneously, noise baseline regions without obvious absorption peaks were identified, such as the flat region from 3000 to 2800 WPM. Within the noise baseline region, absorbance values ​​of 100 data points were selected at equal intervals. The standard deviation of these 100 absorbance values ​​was calculated, and the result, for example, 0.0015 absorbance units, was quantified as the noise level. Subsequently, the intensity of the highest absorption peak within the signal response range is obtained; for example, the peak signal intensity at 1725 wavenumber centimeters is 0.08 absorbance units. The ratio of the peak signal intensity to the noise level is calculated, i.e., 0.08 divided by 0.0015, yielding a signal-to-noise ratio (SNR) of 53.3. The process of setting the SNR benchmark is as follows: repeated measurements are performed on a pure PET blank sample free of the analyte. The maximum signal fluctuation within the noise baseline range is statistically analyzed in each measurement. This benchmark is ultimately set to three times the statistical average of these maximum fluctuation values ​​to ensure effective identification of the true signal. For example, statistical analysis yields a benchmark value of 10. The calculated SNR of 53.3 is compared with the benchmark value of 10. Since 53.3 is greater than 10, the spectral signal response range from 1800 to 1700 wavenumber centimeters is determined to be the effective range and is retained for subsequent spectral feature extraction. For the original chromatogram, the same processing procedure is used to identify signal response segments, for example, within the retention time range of 8 to 10 minutes, and to screen out the effective chromatogram signal response intervals based on the calculated signal-to-noise ratio.

[0053] The spectral feature extraction submodule identifies the spectral signal response interval in the original spectrum, identifies all peaks within the interval, calculates the peak curvature and peak width of each peak, removes noise peaks and baseline drift peaks based on the peak curvature reference and peak width range, extracts the position and intensity of the remaining absorption peaks, and generates spectral absorption peak features.

[0054] Within the spectral signal response range of 1800 to 1700 wavenumber centimeters already selected in the original spectrum, data points within this range are scanned point-by-point to identify all local maxima as peaks. For each identified peak, such as the peak at 1725 wavenumber centimeters, its apex curvature is calculated. The apex curvature is calculated as follows: .in, Peak The peak curvature at a certain point, with the dimension of absorbance per square centimeter, is used to quantify the sharpness of the peak. Peak The absorbance value at a given point, measured in absorbance units, represents the signal intensity at that point. and These are the distances from the peak to the apex of the wave. front and back The absorbance values ​​of each data collection point are all measured in absorbance units. The span parameter, used for calculation, is a dimensionless positive integer. It is set based on the average peak width of the target impurities in the standard database. For absorption peaks with broad peak shapes in the spectrum, a larger value is selected. To achieve robust curvature by smoothing noise, a smaller value is selected for absorption peaks with narrow peak shapes. To obtain a more sensitive curvature response; The wavenumber resolution of the spectrometer is the wavenumber interval between adjacent data points, measured in wavenumber centimeters. Taking the peak at 1725 wavenumber centimeters as an example, the absorbance value at that point is calculated as follows: The absorbance is 0.08 units. The wavenumber resolution of the spectrometer. The wavenumber was set to 0.5 cm. Based on the shape of this absorption peak, the calculation span parameter was set. The value is 2. It retrieves the two data points before and after the vertex, i.e., at... wavenumber centimeters and The absorbance value at wavenumber centimeters was collected. It is 0.075 absorbance units. The absorbance is 0.076 units. Substitute the above parameter values ​​into the calculation: The absorbance is measured in centimeters per square centimeter. Simultaneously, the peak width at half the height of the peak intensity is calculated, i.e., the peak width, for example, 1.2 wavenumber centimeters. The peak curvature reference and peak width range are set based on a comprehensive spectral analysis of standards with known structural impurities, forming a standard database. The reference and range are determined based on the statistical distribution of the peak curvature and peak width of the true absorption peaks in this database, for example, using their 95% confidence interval. For example, the peak curvature reference is set to -1.5 to -0.2; any sharp peak with an excessively large absolute curvature value (e.g., -2.5) is considered a noise peak, and a flat peak with an excessively small absolute curvature value (e.g., -0.05) is considered a baseline drift peak. The peak width range is also set as the 95% confidence interval of the peak width distribution of the true absorption peaks of the standards, for example, 0.5 to 2.0 wavenumber centimeters. The calculated peak curvature of -0.8 is compared with the peak curvature reference [-1.5, -0.2], and -0.8 falls within this range. The peak width of 1.2 wavenumber centimeters is compared with the peak width range [0.5, 2.0 wavenumber centimeters], and 1.2 also falls within this range. Therefore, this peak is identified as a valid absorption peak and retained. Conversely, if another peak with a width of 3.5 wavenumber centimeters is detected, exceeding the peak width range, this peak is determined to be a baseline drift peak and is discarded. For all retained valid absorption peaks, the abscissa value corresponding to the peak is extracted as the vertex position (1725 wavenumber centimeters), and the ordinate value corresponding to the peak is extracted as the peak intensity (0.08 absorbance units). Combining all extracted vertex positions and peak intensities finally generates the spectral absorption peak characteristics.

[0055] The chromatographic data combination submodule identifies the spectral signal response intervals in the original chromatogram and integrates the profile of each chromatographic peak to obtain the retention time and peak area of ​​the chromatographic peak. It combines the position and peak intensity of the absorption peaks in the spectral absorption peak characteristics and combines the retention time and peak area of ​​the chromatographic peaks to establish impurity spectral data and impurity chromatographic data of new chemical materials.

[0056] Identify the spectral signal response intervals in the original chromatogram, determined through signal-to-noise ratio screening, such as the retention time range of 8 to 10 minutes. Within this interval, a numerical integration method is used to calculate the peak area of ​​each chromatographic peak by integrating the data points between its start, peak, and end points. For example, for a chromatographic peak appearing at 8.5 minutes, the peak area is calculated to be 150 mAbsorbance Units per second. Simultaneously, the x-coordinate value corresponding to the peak apex is recorded as the retention time, i.e., 8.5 minutes. Each spectral absorption peak feature obtained from the spectral feature extraction submodule, such as the peak position of 1725 wavenumber cm⁻¹ and the peak intensity of 0.08 absorbance units, is combined. Similarly, the retention time of 8.5 minutes and the peak area of ​​150 mAbsorbance Units per second for each chromatographic peak are combined. By processing all the valid signals detected in the sample as described above, a complete set of impurity spectral data for the new chemical material (containing a series of correspondences between absorption peak positions and intensities) and a set of impurity chromatographic data (containing a series of correspondences between retention times and peak areas) were finally established.

[0057] Please see Figure 3 The impurity identification module includes:

[0058] The spectral matching and identification submodule obtains the position and retention time of the absorption peak from the impurity spectral data and impurity chromatographic data, searches for the position and retention time of the corresponding absorption peak in the preset standard substance spectral library, calculates the relative offset between the position of the absorption peak to be matched and the position of the absorption peak in the spectral library, and calculates the time difference between the retention time to be matched and the retention time in the spectral library. The relative offset and time difference are compared with the set position offset threshold and time difference threshold, respectively, to screen potential substance matching codes.

[0059] The absorption peak position, for example, 1725 wavenumber cm, is obtained from the established impurity spectral data, and the retention time, for example, 8.5 minutes, is obtained from the impurity chromatographic data. A search is performed in a pre-defined standard substance library, constructed by systematically collecting and measuring standard infrared and chromatographic data of all known potential impurities, byproducts, and degradation products in PET production. The search process involves finding entries in the library whose recorded absorption peak position and retention time are close to the value to be matched. For example, the standard data for the substance "diethylene terephthalate (DEG)" in the library is found to be: absorption peak position 1728 wavenumber cm, retention time 8.45 minutes. Next, the relative offset between the absorption peak position to be matched and the absorption peak position in the library is calculated by dividing the absolute value of the difference by the absorption peak position value in the library, i.e., (absolute value of the difference between 1725 and 1728) divided by 1728, resulting in 0.00173. Simultaneously, the time difference between the retention time to be matched and the retention time in the spectral library is calculated. This is done by taking the absolute value of the difference, specifically the absolute value of the difference between 8.5 and 8.45, which yields a result of 0.05 minutes. The setting of the position offset threshold and the time difference threshold is based on statistical analysis of data obtained from repeated measurements of standard samples by the analytical instrument during long-term operation. Specifically, the position offset threshold is calculated based on the standard deviation of the measurement position data and the instrument resolution, while the time difference threshold is calculated from the standard deviation of the measured retention time data, ensuring that the thresholds scientifically reflect the normal fluctuation range of the instrument. For example, the position offset threshold is set to 0.0020, and the time difference threshold to 0.1 minutes. Comparing the calculated relative offset of 0.00173 with the position offset threshold of 0.0020, 0.00173 is less than 0.0020; comparing the time difference value of 0.05 minutes with the time difference threshold of 0.1 minutes, 0.05 is less than 0.1. Since both calculated values ​​are within their respective threshold ranges, the unique substance code for "diethylene terephthalate (DEG)" is selected as the potential substance matching code.

[0060] The structure type determination submodule retrieves the material structure information corresponding to the potential substance matching code, as well as the standard spectral features and standard chromatographic features from the standard substance spectral library. It calls the peak intensity and peak area in the impurity spectral data and the peak area in the impurity chromatographic data, calculates the peak area ratio and peak intensity ratio of each item, and classifies and determines the type of impurity based on the peak area ratio, thus identifying the chemical structure and impurity type of the impurity.

[0061] Based on the matched codes of the screened potential substances, the complete structural information, standard spectral characteristics, and standard chromatographic characteristics corresponding to "diethylene terephthalate (DEG)" are retrieved from the standard substance spectral library. Simultaneously, the peak intensity (0.08 absorbance units) matching DEG recorded in the impurity spectral data and the peak area (150 mAbsorbance units per second) recorded in the impurity chromatographic data are retrieved. To determine the impurity type, the peak area ratio needs to be calculated. First, the peak areas of all impurity peaks recorded in the impurity chromatographic data are summed to obtain the total impurity peak area. For example, if other impurities are present in the sample, the total peak area is 600 mAbsorbance units per second. Then, the proportion of the DEG peak area to the total impurity peak area is calculated, i.e., 150 divided by 600, yielding a peak area ratio of 0.25. The classification of impurity types is based on preset rules established according to industry production quality control standards: impurities with a peak area ratio greater than 0.05 are classified as "major impurities," those between 0.001 and 0.05 are classified as "minor impurities," and those less than 0.001 are classified as "trace impurities." Since the calculated peak area ratio of 0.25 is greater than 0.05, DEG is classified as a "major impurity." Finally, the identified chemical structure "diethylene terephthalate (DEG)" and the impurity type "major impurity" are combined to complete the impurity identification.

[0062] The concentration information generation submodule obtains the quantitative calibration coefficient of the corresponding chemical structure from the standard material spectral library based on the chemical structure and impurity type of the impurity. It calculates the impurity concentration using the calibration coefficient based on the peak intensity recorded in the impurity spectral data and the peak area recorded in the impurity chromatographic data. It combines the impurity chemical structure and impurity type with the impurity concentration to generate impurity characteristic information.

[0063] For the chemical structure and "major impurity" of "diethylene terephthalate (DEG)," a quantitative calibration coefficient corresponding to DEG was obtained from a standard material library. The process for obtaining this calibration coefficient was as follows: a series of DEG standard solutions of known concentrations were prepared in advance and measured using the same chromatographic conditions as the sample analysis, recording the peak area for each concentration. A standard curve was plotted with peak area on the ordinate and concentration on the abscissa, and linear regression analysis was performed. The reciprocal of the slope of the resulting regression curve was the quantitative calibration coefficient. For example, the quantitative calibration coefficient obtained by fitting experimental data was 0.004. Based on the peak area (150 mAbsorbance·s) of DEG recorded in the impurity chromatographic data, the impurity concentration was calculated using the obtained calibration coefficient. The calculation process was to multiply the peak area by the quantitative calibration coefficient, i.e., 150 multiplied by 0.004, resulting in an impurity concentration of 0.6 mg / L. Finally, the chemical structure of the impurity, "diethylene terephthalate (DEG)", the impurity type, "major impurity", and the calculated impurity concentration, "0.6 mg / L", are combined to generate the final impurity characteristic information.

[0064] Please see Figure 4 The reaction pathway inversion module includes:

[0065] The process model construction submodule collects intermediate product data and by-product data for each process stage in the production of new chemical materials, classifies the intermediate product data and by-product data to the corresponding process stage, and constructs a process model for the production of new chemical materials.

[0066] First, the complete production process of the new chemical material (PET) is broken down into independent process stages, specifically including: raw material mixing stage, esterification reaction stage, pre-condensation reaction stage, and final condensation reaction stage. For each process stage, data on intermediate products and by-products generated at the end of each stage are systematically collected through online or offline sampling. For example, after the esterification reaction stage, samples are collected and analyzed using gas chromatography-mass spectrometry (GC-MS) to identify and quantify the content of intermediate products such as monoethylene terephthalate (MHET) and diethylene terephthalate (BHET), while simultaneously detecting the concentration of by-products such as acetaldehyde and moisture. In this process, gas chromatography (GC) is first used to separate the components in the sample, and their properties are determined by recording the retention time of each component. Then, mass spectrometry (MS) is used for further qualitative analysis of each component, identifying the mass contrast peaks of specific molecules through mass spectra, thereby determining the molecular structure and its relative content. For example, the retention time provided by gas chromatography and the characteristic mass ratio (m / z) of the ion peaks obtained by mass spectrometry analysis are used for qualitative identification, and then the concentrations of corresponding intermediates and byproducts are calculated by quantitative analysis of the peak areas. In addition, mass spectrometry data need to be acquired separately, and the mass spectra and corresponding peaks are used for qualitative and quantitative analysis of impurities. This mass spectrometry data helps to more accurately confirm the molecular structure and concentration of each component, especially in complex reaction systems, effectively distinguishing substances with similar structures. The acquired data are categorized; for example, MHET and BHET data are categorized into the intermediate product dataset of the esterification reaction stage, and acetaldehyde data are categorized into the byproduct dataset of the esterification reaction stage. The same acquisition and categorization operation is performed on the subsequent pre-condensation and final condensation stages; for example, data on the byproduct diethylene terephthalate (DEG) is acquired and identified in the final condensation stage. By rigorously mapping the intermediate and byproduct data of all process stages to their corresponding process stages, a structured chemical new material production process model is constructed. The model is structured as a directed graph with process stages as nodes and material flow as edges. Each node is associated with a specific material dataset generated at that stage. The model's execution process involves querying a specific material to trace back to which process stages(s) it was recorded at.

[0067] The kinetic model construction submodule collects polymerization reaction mechanism and kinetic data of new chemical materials, including reaction rate constant and activation energy parameters, and fits the reaction steps in the reaction mechanism with the kinetic data to construct a polymerization reaction kinetic model;

[0068] By consulting chemical engineering handbooks, academic literature, and internal experimental records, a detailed mechanism of the PET polymerization reaction was collected, including the elementary reaction steps of esterification, alcoholysis, acidolysis, polycondensation, and the generation of acetaldehyde and DEG as side reactions. For each elementary reaction step, corresponding kinetic data was collected. Specifically, this kinetic data refers to the reaction rate constants and activation energy parameters measured under different temperature, pressure, and catalyst concentration conditions. For example, experimentally measured values ​​of the reaction rate constant for the side reaction generating DEG in the final polycondensation stage (generated by intramolecular or intermolecular reactions of BHET) were collected at different temperatures. The stoichiometric relationships of reactants and products described in the collected reaction mechanism were fitted with the corresponding kinetic data using a nonlinear least squares method. The model structure is based on a set of ordinary differential equations established according to the chemical reaction mechanism, used to describe the changes in the concentration of each species over time. The model execution process involves solving this set of equations using numerical integration methods. By inputting specific process conditions (such as temperature, pressure, and initial material ratio), the model can output dynamic curves showing the concentration changes of all key species over time throughout the entire reaction process. By fitting and validating all key reaction steps, a polymerization kinetic model that can accurately describe the dynamic changes in the concentration of various species during PET polymerization was finally constructed.

[0069] The impurity reverse deduction submodule calls the impurity chemical structure in the impurity characteristic information and calls the intermediate product and by-product data of each process stage in the chemical new material production process model to establish the correlation results. The correlation results are input into the polymerization reaction kinetic model for reverse deduction. Based on the reaction path and key parameter changes output by the deduction, impurity cause reverse deduction information is generated.

[0070] The process of calling the chemical structure of impurities in the impurity feature information and calling the intermediate and by-product data of each process stage in the chemical new material production process model to establish the association result is as follows: the chemical structure of impurities is compared with the chemical structure of intermediate and by-products, intermediate or by-products with structural similarity higher than the preset similarity threshold are selected, and the selected intermediate or by-products are used as the association result.

[0071] The process of inputting the correlation results into the polymerization reaction kinetic model for reverse deduction is as follows: the intermediate products or by-products contained in the correlation results are taken as the endpoint of the reverse reaction path. Based on the polymerization reaction mechanism recorded in the polymerization reaction kinetic model, the differential equation of reactant concentration changing with time is solved in reverse to reconstruct the reaction path that generates intermediate products or by-products, and the process containing the reaction path is identified as the process of impurity source.

[0072] The impurity chemical structure, specifically the chemical structure of "diethylene terephthalate (DEG)", is retrieved from the impurity feature information. Simultaneously, the intermediate product (e.g., BHET) and byproduct data for each process stage recorded in the constructed chemical new material production process model are retrieved. The chemical structure of the impurity molecule to be tested is compared with the chemical structures of all intermediate products and byproducts recorded in the production process model. This comparison is achieved by calculating the similarity coefficient of their chemical substructure feature vectors. The specific calculation method is as follows: .in, impurity molecules to be tested The intermediate or by-product molecules recorded in the production process model The structural similarity coefficient is a dimensionless value. Refers to the impurity molecule to be tested, i.e., DEG; It refers to intermediate or by-product molecules recorded in the production process model, in this case, BHET; The total number of bits in the chemical substructure feature vector is a fixed positive integer, which is 8 in this example; The bit index of the chemical substructure feature vector is traversed from 1 to... ; and The impurity molecules to be tested are respectively Intermediate or by-product molecules recorded in the production process model Its chemical substructure eigenvectors The value of the bit is a binary value of 0 or 1, where 1 indicates that the molecule contains the first bit. A chemical substructure, where 0 indicates its absence; For logical AND operator, The result is 1 if and only if and All are 1; For logical OR operator, The result is 1 when or At least one of them is 1; For the first The weighting coefficients of each chemical substructure are dimensionless positive numbers. The weighting is based on the importance of the substructure in the PET polymerization and degradation reactions. Chemical functional groups that directly participate in the reaction or are easily changed (such as ester bonds and hydroxyl groups) are considered to better reveal the transformation relationships between substances; therefore, the higher their importance, the larger the weighting coefficient. Conversely, chemically stable skeletal structures that are not easily changed in the reaction (such as benzene rings) are less important and are assigned smaller weighting coefficients. Structural similarity calculations are performed between the analyte impurity molecule DEG and the intermediate or by-product molecule BHET recorded in the production process model: the total number of digits in the chemical substructure eigenvector is set. The value is 8. In this example, the chemical substructure eigenvector of the analyte impurity molecule DEG is... The binary vector is [1,1,0,0,0,1,1,0], while the chemical substructure feature vector of the intermediate or by-product molecule BHET recorded in the production process model is... The binary vector is [1,1,1,1,0,1,1,0]. Weight vector. Set to [1.5, 1.5, 1.0, 1.0, 1.0, 1.5, 1.5, 1.0]. Calculate the weighted intersection of the analyte impurity molecule and the intermediate or by-product molecules recorded in the production process model: Calculate the weighted union of the analyte molecule in the denominator and the intermediate or by-product molecules recorded in the production process model: Finally, the similarity coefficient is calculated: The preset similarity threshold was determined by calculating the structural similarity of reactants and products in a large number of known chemical reactions and analyzing the statistical distribution of the calculation results. This threshold, set at 0.7, aims to effectively screen substance pairs with a high probability of transformation. The calculated similarity coefficient of 0.75 was compared with the threshold of 0.7. Since 0.75 is higher than 0.7, BHET was selected as the associated result. The intermediate product BHET included in the associated result was used as the endpoint of the reverse reaction path and input into the constructed polymerization kinetic model. The reverse execution of the model starts from the endpoint substance (DEG) and its precursor (BHET), and solves the differential equations in reverse based on the model's built-in reaction mechanism to reconstruct the reaction conditions and path that can generate the target impurity concentration. The output reaction path and key parameter changes are deduced to determine the process from which the impurities originate, generating impurity origin inversion information.

[0073] Please see Figure 5 The impurity content correction module includes:

[0074] The content correction calculation submodule obtains the impurity concentration from the preset correction parameters and impurity characteristic information of the standard sample test, corrects the impurity concentration using the correction parameters, and obtains the corrected impurity content.

[0075] Obtain the preset correction parameter for the standard sample test. The process for setting this correction parameter is as follows: Select a certified reference material (CRM) containing a precisely known concentration of DEG impurities. Repeat the measurement of this CRM using the same instruments and methods as the analysis of the test sample. The correction parameter is the ratio of the true concentration value of the CRM to the statistical average of the concentrations obtained from multiple measurements, used to correct for inherent biases in the system. For example, if the true CRM concentration is 0.58 mg / L and the average measurement is 0.608 mg / L, the correction parameter is 0.58 divided by 0.608, resulting in 0.954. This parameter reflects the systematic bias of the current analytical system. Next, obtain the impurity concentration calculated from the aforementioned impurity characteristic information, which is 0.6 mg / L. Use the obtained correction parameter 0.954 to correct the impurity concentration. The calculation process is to multiply the impurity concentration to be measured by the correction parameter, i.e., 0.6 multiplied by 0.954, resulting in a corrected impurity content of 0.5724 mg / L.

[0076] The source process extraction submodule analyzes the changes in key parameters corresponding to the impurity source process from the impurity origin inversion information, compares the impurity concentration in the corrected impurity content with the range of key parameter changes, verifies the correspondence between process information and impurity content, screens key processes, and obtains key information of the source process.

[0077] The impurity originating process, namely the "final polycondensation reaction process," was analyzed from the impurity origin inversion information, along with the change in the key parameter directly related to impurity formation, namely, "the reaction temperature increased from 280 degrees Celsius to 285 degrees Celsius." The back-calculation results of the polymerization reaction kinetic model also provided the corresponding ranges of different key parameter values ​​and impurity formation concentrations. For example, when the final polycondensation temperature was in the range of 284 to 286 degrees Celsius, the generated DEG concentration ranged from 0.55 to 0.60 mg / L. The corrected impurity content of 0.5724 mg / L calculated in the previous step was compared with the concentration range [0.55, 0.60] mg / L predicted by the kinetic model. Since 0.5724 mg / L falls within the predicted concentration range, this verifies that there is a precise correspondence between the temperature anomaly information of the "final polycondensation reaction process" and the detected impurity content. This verification identified "temperature control in the final polycondensation reaction process" as the key process leading to impurity generation. The key information of the source process was obtained, which was "abnormal temperature control in the final polycondensation stage, with the actual temperature higher than the set value".

[0078] The impurity source definition submodule defines the process content in the key information of the source process as the impurity source and generates impurity source information.

[0079] The process details explicitly stated in the key information of the source process, namely "abnormal temperature control in the final polycondensation stage, with the actual temperature exceeding the set value," are directly defined as the source of the DEG impurity detected this time. Through this clear definition, the analysis results are linked to specific production operations, ultimately generating a structured impurity source information, which includes the impurity name (DEG), impurity concentration (0.5724 mg / L), and a precise source description (abnormal temperature control in the final polycondensation stage).

[0080] Please see Figure 6 The information traceability module includes:

[0081] The report generation submodule summarizes the corrected impurity content and impurity source information, performs data filling and formatting, and generates a test report.

[0082] The summary module outputs the corrected impurity content, specifically "diethylene terephthalate (DEG)" at 0.5724 mg / L, along with the impurity source information generated by the impurity source definition submodule: "abnormal temperature control during the final polycondensation stage, with the actual temperature exceeding the set value." These two core pieces of information, along with basic information such as the sample batch number, testing date, and analytical method, are then used to populate a pre-defined test report template. The template formatting process includes filling data into specified table cells, automatically generating a summary, and selecting appropriate charts for visualization based on the data type. For example, a bar chart comparing the impurity content with the standard limit is used. After data population and formatting, a complete and properly formatted test report is generated.

[0083] The quality early warning judgment submodule compares the corrected impurity content with the preset quality standard threshold to determine whether a quality early warning is triggered, and integrates the early warning judgment result with the test report to generate the material impurity content test result.

[0084] The calculated corrected impurity content, 0.5724 mg / L, is compared with the preset quality standard threshold for this grade of PET material. This quality standard threshold is set based on relevant national standards or specific customer requirements for product performance. For example, for food contact grade PET material, the DEG content must not exceed 0.5 mg / L, therefore the preset quality standard threshold is 0.5 mg / L. The corrected impurity content of 0.5724 mg / L is compared with the quality standard threshold of 0.5 mg / L. Since 0.5724 is greater than 0.5, this comparison triggers a quality warning. The system generates a "exceeds the standard" warning result. Finally, this "exceeds the standard" warning result is integrated with the detailed test report generated in the previous step. For example, a prominent red "Quality Warning" label is added to the first page of the test report, and the conclusion section clearly states "DEG content exceeds the standard," ultimately generating a material impurity content test result containing complete data, source traceability, and a clear quality assessment.

[0085] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A system for detecting impurity content in new chemical materials based on product analysis and reaction pathway inversion, characterized in that, The system includes: The data processing module acquires the original spectra and chromatograms of impurities in the chemical new material samples, and extracts the impurity spectral data and impurity chromatographic data of the chemical new material from the original spectra and chromatograms. The impurity identification module compares the impurity spectral data and impurity chromatographic data with a preset standard material spectral library to identify the chemical structure, impurity concentration and type of impurities in new chemical materials and generate impurity characteristic information. The reaction path inversion module constructs a chemical new material production process model and a polymerization reaction kinetic model, associates the impurity characteristic information with the chemical new material production process model, and inputs it into the polymerization reaction kinetic model for reverse deduction to determine the process in which impurities are generated and generate impurity cause inversion information. The impurity content correction module corrects the impurity characteristic information based on the impurity cause inversion information, generates the corrected impurity content, and locates the impurity source information. The information traceability module summarizes the corrected impurity content and the impurity source information and performs formatting processing to obtain the material impurity content detection results.

2. The chemical new material impurity content detection system based on product analysis and reaction path inversion according to claim 1, characterized in that, The impurity spectral data includes absorption peak position and absorption peak intensity; the impurity chromatographic data includes retention time and chromatographic peak area; the impurity characteristic information includes chemical structure, impurity concentration, and impurity type; the impurity origin inversion information includes impurity generation process and key polymerization parameters; the corrected impurity content is specifically a calibration concentration value; the impurity source information is specifically the source process code; and the material impurity content detection results include visualization charts and data traceability indexes.

3. The chemical new material impurity content detection system based on product analysis and reaction pathway inversion according to claim 1, characterized in that, The data processing module includes: The original spectrum acquisition submodule acquires the original spectra and chromatograms of impurities in chemical new material samples, identifies the signal response segments and noise baseline segments within the original spectra and chromatograms, quantifies the fluctuation amplitude of the noise baseline segments to determine the noise level, obtains the signal peak intensity of the signal response segments, calculates the ratio of the signal peak intensity to the noise level to obtain the signal-to-noise ratio, and compares the signal-to-noise ratio with the set signal-to-noise ratio benchmark value to filter the spectrum signal response intervals; The spectral feature extraction submodule identifies the spectral signal response interval in the original spectral image, identifies all peak points within the interval, calculates the peak curvature and peak width of each peak point, removes noise peaks and baseline drift peaks based on the peak curvature reference and peak width range, extracts the position and intensity of the remaining absorption peaks, and generates spectral absorption peak features. The chromatographic data combination submodule identifies the spectral signal response intervals in the original chromatogram and integrates the profile of each chromatographic peak to obtain the retention time and peak area of ​​the chromatographic peak. It combines the position and peak intensity of the absorption peaks in the spectral absorption peak characteristics and combines the retention time and peak area of ​​the chromatographic peaks to establish impurity spectral data and impurity chromatographic data of new chemical materials.

4. The chemical new material impurity content detection system based on product analysis and reaction path inversion according to claim 3, characterized in that, The impurity identification module includes: The spectral matching and identification submodule obtains the position and retention time of the absorption peak from the impurity spectral data and impurity chromatographic data, searches for the position and retention time of the corresponding absorption peak in the preset standard substance spectral library, calculates the relative offset between the position of the absorption peak to be matched and the position of the absorption peak in the spectral library, and calculates the time difference between the retention time to be matched and the retention time in the spectral library. The relative offset and time difference are compared with the set position offset threshold and time difference threshold, respectively, to screen potential substance matching codes. The structure type determination submodule retrieves the material structure information, standard spectral features, and standard chromatographic features corresponding to the potential substance matching code from the standard material spectral library, calls the peak intensity and peak area in the impurity spectral data, calculates the peak area ratio and peak intensity ratio of each item, classifies and determines the type of impurity based on the peak area ratio, and identifies the chemical structure and impurity type of the impurity. The concentration information generation submodule obtains the quantitative calibration coefficient of the corresponding chemical structure from the standard material spectral library based on the chemical structure and impurity type of the impurity. It calculates the impurity concentration using the calibration coefficient based on the peak intensity recorded in the impurity spectral data and the peak area recorded in the impurity chromatographic data. It then combines the impurity chemical structure and impurity type with the impurity concentration to generate impurity characteristic information.

5. The chemical new material impurity content detection system based on product analysis and reaction pathway inversion according to claim 4, characterized in that, The reaction pathway inversion module includes: The process model construction submodule collects intermediate product data and by-product data for each process stage in the production of new chemical materials, classifies the intermediate product data and by-product data to the corresponding process stage, and constructs a process model for the production of new chemical materials. The kinetic model construction submodule collects polymerization reaction mechanism and kinetic data of new chemical materials, including reaction rate constant and activation energy parameters, and fits the reaction steps in the reaction mechanism with the kinetic data to construct a polymerization reaction kinetic model; The impurity reverse deduction submodule calls the impurity chemical structure in the impurity characteristic information and calls the intermediate product and by-product data of each process stage in the chemical new material production process model to establish the correlation results. The correlation results are input into the polymerization reaction kinetic model for reverse deduction. Based on the reaction path and key parameter changes output by the deduction, impurity origin reverse deduction information is generated.

6. The chemical new material impurity content detection system based on product analysis and reaction path inversion according to claim 5, characterized in that, The process of calling the impurity chemical structure in the impurity feature information and calling the intermediate and by-product data of each process stage in the chemical new material production process model to establish the association result is as follows: the chemical structure of the impurity is compared with the chemical structure of the intermediate and by-products, the intermediate or by-products with structural similarity higher than the preset similarity threshold are screened out, and the screened intermediate or by-products are used as the association result.

7. The chemical new material impurity content detection system based on product analysis and reaction pathway inversion according to claim 5, characterized in that, The process of inputting the correlation results into the polymerization reaction kinetic model for reverse deduction is as follows: the intermediate products or by-products contained in the correlation results are taken as the endpoint of the reverse reaction path. Based on the polymerization reaction mechanism recorded in the polymerization reaction kinetic model, the differential equation of reactant concentration changing with time is solved in reverse to reconstruct the reaction path that generates intermediate products or by-products. The process containing the reaction path is identified as the process that is the source of impurities.

8. The chemical new material impurity content detection system based on product analysis and reaction pathway inversion according to claim 6, characterized in that, The chemical structures of impurities are compared with those of intermediates and by-products based on structural similarity using the following formula: ； in, impurity molecules to be tested The intermediate or by-product molecules recorded in the production process model The structural similarity coefficient, Refers to the impurity molecule to be tested. Refers to intermediate or by-product molecules recorded in the production process model. The total number of bits in the chemical substructure eigenvectors The bit index of the chemical substructure feature vector. and The impurity molecules to be tested are respectively Intermediate or by-product molecules recorded in the production process model Its chemical substructure eigenvectors The value at each bit, For the first Weighting coefficients for each chemical substructure.

9. The chemical new material impurity content detection system based on product analysis and reaction pathway inversion according to claim 5, characterized in that, The impurity content correction module includes: The content correction calculation submodule obtains the preset correction parameters for the standard sample test and the impurity concentration in the impurity characteristic information, corrects the impurity concentration using the correction parameters, and obtains the corrected impurity content. The source process extraction submodule analyzes the changes in key parameters corresponding to the impurity source process from the impurity origin inversion information, compares the impurity concentration in the corrected impurity content with the range of key parameter changes, verifies the correspondence between process information and impurity content, screens key processes, and obtains key information of the source process. The impurity source definition submodule defines the process content in the key information of the source process as the impurity source and generates impurity source information.

10. The chemical new material impurity content detection system based on product analysis and reaction pathway inversion according to claim 9, characterized in that, The information traceability module includes: The report generation submodule summarizes the corrected impurity content and the impurity source information, performs data filling and formatting processing, and generates a test report. The quality early warning determination submodule compares the corrected impurity content with a preset quality standard threshold to determine whether a quality early warning is triggered, and integrates the early warning determination result with the test report to generate the material impurity content test result.

Images