Quantitative method for benzotriol isomers based on multi-peak consistency correction

Abstract

The application discloses a benzotriol isomer quantitative method based on multi-feature peak consistency correction, and comprises the following steps: collecting a pretreatment spectrum for a mixture to be measured containing two components belonging to a benzotriol isomer group, calculating the signal-to-noise ratio of each exclusive feature peak in the pretreatment spectrum, and completing the qualitative analysis of the two components by multi-feature peak consistency voting according to the signal-to-noise ratio; selecting a main feature peak pair, and sequentially performing conversion and multi-feature peak consistency correction on the peak area of each feature peak to obtain the effective peak area of the two components, and then obtaining the competitive peak area ratio of the two components, and finally calculating the mass fraction of the two components according to a linear self-calibration model to complete the quantitative analysis. 2 The calibration correlation coefficient R is improved from about 0.88 to above 0.99, and fast, lossless and distinguishable quantitative analysis is realized.

Description

Technical Field

[0001] This invention relates to the field of quantitative materials analysis technology, specifically to a quantitative method for pyrogallol isomers based on multi-characteristic peak consistency correction. Background Technology

[0002] Phloroglucinol is an important class of aromatic trihydroxy compounds, all with the molecular formula C6H6O3. Depending on the substitution positions of the three hydroxyl groups on the benzene ring, it exists as three isomers: 1,2,3-pyroglucinol (ortho-pyroglucinol, also known as pyrogallic acid), 1,2,4-pyroglucinol (hydroxyhydroquinone), and 1,3,5-pyroglucinol (re-pyroglucinol). Due to the different hydroxyl substitution positions, the three pyroglucinol isomers exhibit significant differences in chemical reactivity, redox properties, coordination and complexation abilities, environmental migration and transformation behavior, and biotoxicological effects. In chemical synthesis, product quality control, reaction process monitoring, and environmental risk assessment, changes in the types and contents of pyroglucinol isomers can reflect reaction pathways, transformation degrees, and product distribution. Therefore, rapid and accurate differentiation and quantitative analysis of pyroglucinol isomers are of great significance. Currently, the detection of pyroglucinol and other phenolic isomers mainly relies on high-performance liquid chromatography (HPLC), gas chromatography (GC), chromatography-mass spectrometry (GC-MS), and electrochemical methods. While the aforementioned methods offer high sensitivity and quantitative capabilities, they typically require complex sample pretreatment, component separation, or derivatization steps, resulting in lengthy analytical procedures, high costs, and destructive testing. This makes them unsuitable for rapid, in-situ, and non-destructive detection. Furthermore, they present significant challenges in distinguishing isomers. In contrast, Raman spectroscopy offers advantages such as requiring no complex sample pretreatment, rapid detection speed, non-destructive analysis, and the ability to perform in-situ or quasi-in-situ detection.

[0003] However, existing quantitative methods based on Raman spectroscopy still have the following shortcomings when dealing with mixed systems of pyrogallol isomers: First, traditional methods mostly rely on a single characteristic peak for qualitative and quantitative analysis. Single-peak information is easily affected by multiple factors, resulting in insufficient reliability of qualitative differentiation, susceptibility to misjudgment, and the results are prone to drift with overall fluctuations in acquisition conditions, leading to poor linearity of the calibration curve. Second, some methods directly use the ratio of the peak areas of the two component characteristic peaks as a quantitative variable. This variable has no upper bound and increases non-linearly with the component ratio, making it sensitive to peak area extraction errors in the high-content range. Third, due to factors such as peak overlap, local baseline fluctuations, and noise, the extracted peak area of ​​a certain characteristic peak at individual concentration points may deviate abnormally, causing the peak area ratio or proportion to deviate from the true value, resulting in significant dispersion of the calibration curve at these points. Existing technologies often involve correcting single-peak shape parameters (fitting residuals, peak position shifts, peak width variations), adjusting the fitting residuals of the final prediction results, or constructing multiple sets of features into high-dimensional variables before performing dimensionality reduction regression. These methods struggle to identify and correct abnormal peak areas at the source. Therefore, current technologies lack a simple Raman detection method that can identify and correct abnormal peak areas at the feature level, reliably distinguish between and accurately quantify the three isomers of pyrogallol. Summary of the Invention

[0004] Purpose of the Invention: The purpose of this invention is to provide a quantitative method for pyrogallol isomers based on multi-characteristic peak consistency correction. This method differs from traditional methods that only correct single-peak shape parameters or residuals of prediction results. It solves the technical problems of existing Raman quantitative detection methods, such as insufficient reliability in qualitative differentiation in mixed systems of pyrogallol isomers, poor linearity of single-peak absolute peak area or ordinary peak area ratio, and difficulty in identifying and correcting abnormal peak areas at the characteristic level at individual concentration points, which leads to discrete calibration curves and insufficient quantitative accuracy.

[0005] Technical solution: The quantitative method for pyrogallol isomers based on multi-characteristic peak consistency correction described in this invention includes the following steps: For the mixture containing two pyrogallol isomers to be tested, the two pyrogallol isomers are respectively designated as components. and components The Raman spectrum of the mixture was collected and preprocessed to obtain the preprocessed spectrum; Based on the standard spectra of the three pyrogallol isomers, the net peak heights of each specific characteristic peak in the preprocessed spectrum were read and the signal-to-noise ratio was calculated. According to the signal-to-noise ratio A consensus vote was performed on the individual characteristic peaks in the preprocessed spectrum to determine the composition. and components The qualitative analysis was completed by comparing it with two of the three isomers of phloroglucinol. Based on the separability index, the components and components Orthogonal optimization is performed on the exclusive characteristic peaks to determine the main characteristic peak pairs; using the main characteristic peak pairs as the conversion scale, the components are... and components The peak areas of each characteristic peak are calculated separately to obtain the calculated peak areas; multi-characteristic peak consistency correction is then performed on the calculated peak areas to obtain the component. effective peak area and components effective peak area According to the components effective peak area and components effective peak area The components were calculated. competitive peak area ratio ; According to components A linear self-calibrating quantitative model relating competitive peak area percentage to mass fraction was used to calculate the composition. mass fraction and according to Components The mass fraction was determined to complete the quantitative assessment.

[0006] Explanation: The core of this invention lies in the combination of multi-peak consistency correction and competitive normalization. On one hand, this invention utilizes the characteristic that each isomer of pyrogallol has multiple exclusive characteristic peaks, and that there is an inherent area ratio relationship between the exclusive characteristic peaks of the same component determined by the molecular structure. After converting each exclusive characteristic peak to the scale of the main characteristic peak, cross-comparison is performed. When the main characteristic peak at a certain concentration point exhibits an abnormal peak area due to peak overlap, baseline fluctuation, or noise, this anomaly will manifest as the main characteristic peak's converted value deviating from the consensus value (i.e., the median) of the other exclusive characteristic peaks. Based on this, the weight of the abnormal peak is automatically reduced, and it is corrected using multi-peak consensus, thereby correcting the abnormal peak area at the characteristic level (rather than at the prediction end). This correction mechanism differs from the correction of single-peak shape parameters and the correction of residuals in prediction results. It utilizes the consistency between multiple independent exclusive peaks of the same component for cross-validation, with clear physical basis and simple calculation. On the other hand, the competitive proportion of the corrected effective peak area is used... As a quantitative main variable, this variable is bounded (0~1), self-referenced, and insensitive to overall fluctuations in laser power, focusing state, and sample amount. The combination of the two can significantly improve calibration linearity and quantitative stability, enabling rapid differentiation and efficient quantification of any pairwise combinations of the three isomers of pyrogallol.

[0007] In one implementation, components mass fraction The competitive peak area ratio can be used to determine the proportion of peak areas. With mass fraction The linear self-calibration quantitative model between them is calculated according to the following formula: ; In the formula, Components The proportion of competitive peak area. and Components With components The fitting slope and fitting intercept of the mixed system; components mass fraction according to calculate.

[0008] When the two pyrogallol isomers in the mixture are 1,2,3-pyrogallol and 1,2,4-pyrogallol, It is 0.3491. It is 0.5229; When the two pyrogallol isomers in the mixture are 1,2,3-pyrogallol and 1,3,5-pyrogallol, It is 0.8993. It is -0.0105; When the two pyrogallol isomers in the mixture are 1,2,4-pyrogallol and 1,3,5-pyrogallol, It is 0.6814. It is 0.1351.

[0009] In one implementation, the step of basing the composition on the composition... effective peak area and components effective peak area Calculate components competitive peak area ratio ,include: ; In the formula, Components The effective peak area, Components Effective peak area, in units of Counts·cm -1 , Components The area ratio of the competitive peak is a dimensionless quantity.

[0010] Note: The proportion of competitive peak area characterizes the composition. The percentage of the corrected effective peak area in the sum of the corrected effective peak areas of the two components. Compared to the unbounded, nonlinear ordinary peak area. Compared to the absolute peak area of ​​a single peak that drifts with sampling conditions. Strictly limited to the 0~1 range, approximately linear with the mass fraction, and serving as a self-reference quantity for the two-component signals; when the laser power, acquisition time, focusing state, or sample amount fluctuates overall, causing the areas of the two peaks to scale proportionally, It remains unchanged, thereby inherently eliminating such systemic disturbances.

[0011] In one implementation, the net peak height of each specific characteristic peak in the preprocessed spectrum is read and the signal-to-noise ratio is calculated based on the standard spectra of the three pyrogallol isomers. The process includes: determining the specific characteristic peak sets for each of the three pyrogallol isomers based on their standard spectra, each set comprising one main characteristic peak and at least two secondary characteristic peaks; using the center wavenumber of each specific characteristic peak in the set as the search position; reading the net peak height after local baseline subtraction at the corresponding wavenumber of the preprocessed spectrum; and calculating the signal-to-noise ratio of each specific characteristic peak in the preprocessed spectrum based on the net peak height. : ; In the formula, For the first Signal-to-noise ratio of each characteristic peak This represents the net peak height of the characteristic peak after baseline correction, expressed in counts. This represents the standard deviation of the intensity within the peakless region of the spectrum, expressed in Counts.

[0012] In one implementation, the statement based on the signal-to-noise ratio... A consensus vote was performed on the individual characteristic peaks in the preprocessed spectrum to determine the composition. and components Qualitative analysis was performed by comparing the results with two of the three pyrogallol isomers, including: The signal-to-noise ratio (SNR) of each specific characteristic peak in the preprocessed spectrum is calculated. The SNR threshold is set to 3, corresponding to a detection criterion of 3 times the noise standard deviation, i.e., the 3σ detection criterion. The specific characteristic peaks (corresponding to the 3σ detection criterion) are recorded as valid votes. When there are no fewer than 2 valid votes for the same pyrogallol isomer, the pyrogallol isomer is determined to be a component in the mixture. In this embodiment, a signal-to-noise ratio of not less than 3 indicates that the corresponding peak height reaches at least 3 times the noise standard deviation, which is used to reduce the probability that random noise is misidentified as a specific characteristic peak.

[0013] Note: A single characteristic peak is prone to misjudgment due to accidental noise, baseline fluctuations, or interference from adjacent peaks. This invention requires at least two exclusive characteristic peaks of the same isomer to simultaneously reach the signal-to-noise ratio threshold before its existence is determined, significantly improving the reliability of qualitative differentiation through "multi-peak synergy and mutual corroboration".

[0014] In one implementation, the component is analyzed based on the separability index. and components Orthogonal optimization is performed on the exclusive characteristic peaks to determine the main characteristic peak pairs, including: First, Gaussian-Lorentz mixture function is applied to fit each characteristic peak to obtain the peak shape fitting curve. ,include: ; In the formula, The curve is a peak-shaped fitting curve. Raman displacement, in cm -1 , The peak height is the characteristic peak, expressed in Counts. The location of the characteristic peak center is shown in cm. -1 , The half-peak-half-width of the characteristic peak represents the wavenumber distance from the center of the characteristic peak to one of the intersection points at half the peak height, in cm. -1 , This is the Gaussian-Lorentz mixing scaling factor, with a value ranging from 0 to 1; Then, based on the peak shape fitting curve , to obtain components Candidate characteristic peaks With components Candidate characteristic peaks spectral overlap between ,include: ; In the formula, Components Candidate characteristic peaks With components Candidate characteristic peaks The spectral overlap between the candidate characteristic peaks and Corresponding components Components All characteristic peaks in their respective characteristic peak sets , Components Candidate characteristic peaks Components Candidate characteristic peaks The fitted curves obtained by fitting the peak shape use the center position of the characteristic peak calibrated with pure standard samples for each fitting window. Expand 2 to each side from the center. ,in, The half-peak width of the characteristic peak in the pure standard sample is taken as the half-peak width; the integration interval is the union of the fitting windows of the two candidate characteristic peaks. Then based on the components Candidate characteristic peaks With components Candidate characteristic peaks spectral overlap between The separability index is obtained. ,include: ; In the formula, Components Candidate characteristic peaks With components Candidate characteristic peaks The separability index between them; , Candidate feature peaks , The signal-to-noise ratios are all dimensionless quantities; The overlap penalty coefficient is fixed at 1; When the separability index considers both the signal-to-noise ratio and the degree of peak overlap of the two candidate feature peaks, it avoids inconsistencies in the selection of the main feature peak pair due to manual adjustment of the overlap penalty coefficient. The largest separability index The corresponding pair of components and components The exclusive characteristic peaks are used as the main characteristic peak pairs.

[0015] Note: The separability index requires that both selected peaks be sufficiently strong. Large) and with small overlap, sufficiently orthogonal ( Small). The overlap integral (value 0-1, where 0 indicates no overlap) of the fitted curves of the two peaks is used. This index automatically selects the characteristic peak pairs with the strongest independence between the two components, reducing the interference of peak overlap on peak area extraction from the source. The separability index considers both the signal intensity of the candidate characteristic peaks and the degree of overlap between peaks. The greater the spectral overlap, the smaller the separability index; fixing λ=1 avoids additional parameter optimization, making the determination rule of the main characteristic peak pair unique and repeatable.

[0016] In one implementation, the main characteristic peak pair is used as the reduction scale for the components. and components The peak areas of each specific characteristic peak are calculated separately to obtain the converted peak areas, including: Conversion Components No. The calculated peak area of ​​each unique characteristic peak ,include: ; In the formula, Components No. The peak area of ​​each unique characteristic peak, in Counts·cm² -1 , Components No. The conversion factor for each characteristic peak; The components No. Peak area of ​​each unique characteristic peak ,include: ; In the formula, Components No. The peak height of each unique characteristic peak, expressed in counts. Components No. Half-peak width of each unique characteristic peak, in cm. -1 , Components No. Gaussian-Lorentz mixing ratio factor for each unique characteristic peak; The components No. The reduction factor of each characteristic peak ,include: ; In the formula, , Components The main characteristic peak in pure standard samples, the first The peak area of ​​each characteristic peak, in Counts·cm -1 .

[0017] Note: Conversion factor Measured from pure standard samples, the area ratios of the specific characteristic peaks of the same component, determined by the molecular structure, are reflected. After conversion, all specific characteristic peaks are unified to the scale of the main characteristic peak and should theoretically be consistent with each other. When the main characteristic peak and the auxiliary characteristic peak are consistent, the weight of the main characteristic peak is close to 1, and the effective peak area after correction is approximately equal to the area of ​​the main characteristic peak, without introducing additional bias. This mechanism utilizes the consistency among multiple independent specific peaks of the same component for cross-validation, and can automatically identify and correct abnormal peak areas at individual concentration points at the characteristic level. This is the key to the significant improvement in calibration linearity of this invention.

[0018] In one implementation, multi-characteristic peak consistency correction is performed on the areas of each calculated peak to obtain the components. effective peak area and components effective peak area ,include: ; In the formula, These are the normalized weights, and the components The sum of all weights is 1. The peak area is calculated to the scale of the main characteristic peak, in Counts·cm. -1 ; The normalized weights ,include: ; In the formula, Components No. A reliability measure of each characteristic peak; The components No. Reliability measurement of each characteristic peak ,include: ; In the formula, Components No. Signal-to-noise ratio of each characteristic peak Components No. Characteristic peaks and components The maximum value of the spectral overlap of all characteristic peaks, calculated using the same method as... same; ; In the formula, Components No. A measure of the consistency of each unique characteristic peak; Components No. The calculated peak area of ​​each individual characteristic peak after being scaled down to the main characteristic peak is expressed in Counts·cm. -1 ; Components Total converted peak area The median, in Counts·cm -1 ; The consistency tolerance coefficient is set to 0.1 in this embodiment.

[0019] Consistency tolerance coefficient This is used to control the rate of decline in the consensus metric when the calculated peak area deviates from the median consensus value. When, the calculated peak area of ​​a specific characteristic peak is relative to the median When the relative deviation is 10%, its consistency measure The value is 0.5; when the relative deviation is less than 10%, this specific characteristic peak maintains a high consistency measure; when the relative deviation is greater than 10%, its consistency measure decreases as the deviation increases. Therefore, we take... It can maintain a high weight for characteristic peaks that are close to the multi-peak consensus value and reduce the impact of characteristic peaks that deviate significantly from the multi-peak consensus value on the effective peak area calculation results.

[0020] Components effective peak area With components effective peak area The method for obtaining them is the same.

[0021] The quantitative analysis system for pyrogallol isomers of the present invention includes a Raman spectrometer and a data processing device; The Raman spectrometer is used to collect the Raman spectra of the test mixture containing two pyrogallol isomers; The data processing device includes a memory and a processor. The memory stores a computer program. When the processor executes the computer program, it performs preprocessing, qualitative analysis, peak area conversion, multi-characteristic peak consistency correction, competing peak area ratio calculation, and mass fraction calculation on the Raman spectrum according to the above-described method. The data processing device obtains the composition... mass fraction and according to Components mass fraction .

[0022] The computer device described in this invention includes a memory, a processor, and a computer program stored in the memory and executable on the processor; when the processor executes the computer program, it implements the Raman spectroscopy preprocessing, qualitative analysis, peak area conversion, multi-characteristic peak consistency correction, competitive peak area ratio calculation, and mass fraction calculation steps in the above method.

[0023] Beneficial effects: Compared with the prior art, the present invention has the following significant advantages: (1) The present invention proposes a multi-characteristic peak consistency correction mechanism, which uses the inherent area ratio relationship between multiple exclusive characteristic peaks of the same component determined by the molecular structure for cross-verification. It can automatically identify and correct abnormal peak areas at individual concentration points caused by peak overlap, baseline fluctuation or noise at the characteristic level. It is different from the correction of single peak shape parameters and the correction of residuals of prediction results. The physical basis is clear and the calculation is simple.

[0024] (2) The present invention uses the competitive proportion R of the corrected effective peak area as the quantitative main variable. This variable is bounded, self-referenced, and resistant to overall disturbances. Compared with the absolute peak area of ​​a single peak and the unbounded, nonlinear ordinary peak area ratio, it can obtain significantly higher calibration linearity and prediction stability.

[0025] (3) The present invention utilizes multiple exclusive characteristic peaks of each isomer for consistency voting, and replaces single-peak discrimination with multi-peak synergy and mutual verification, which significantly improves the reliability of distinguishing 1,2,3-phenylpyrogallol, 1,2,4-phenylpyrogallol and 1,3,5-phenylpyrogallol.

[0026] (4) This invention requires no sample separation or complex pretreatment, and the detection is rapid and non-destructive; taking the 1,2,3-phenylpyrogallol / 1,2,4-phenylpyrogallol system as an example, the correlation coefficient of the calibration curve constructed from the original data without consistency correction is only R. 2 =0.8833, which is improved to R after multi-characteristic peak consistency correction according to the present invention. 2 =0.9943, the quantitative absolute error for the same sample was reduced from about 2.0 percentage points in Comparative Example 1 (uncorrected competitive ratio) to about 1.3 percentage points, which is significantly better than the error level of about 13.9 percentage points in Comparative Example 2 (single peak absolute peak area method). It can be widely used in fields such as chemical synthesis process monitoring, product quality control, phenolic pollutant degradation process tracking and environmental risk assessment.

[0027] (5) This invention achieves its results through the interdependence of competitive normalization and multi-characteristic peak consistency correction: the former only eliminates overall disturbances such as laser power, focusing and sample amount, while the latter only corrects the local abnormal peak area of ​​individual concentration points. Neither of them alone is sufficient to obtain high linearity and high quantitative accuracy at the same time (see Comparative Example 1 and Comparative Example 2); only by combining the two can the calibration correlation coefficient be increased to above 0.99 and the quantitative absolute error be reduced to about 1 percentage point, demonstrating significant synergistic effect. Attached Figure Description

[0028] Figure 1 These are molecular structure diagrams of three pyrogallol isomers; from left to right, they are 1,2,3-pyrogallol, 1,2,4-pyrogallol, and 1,3,5-pyrogallol. Figure 2 This is the standard spectrum of the mixed sample in Example 1 of the present invention; Figure 3 This is the Raman spectrum of the mixed sample of 1,2,3-phenylpyrogallol and 1,2,4-phenylpyrogallol in Example 1 of the present invention; Figure 4 This is a standard curve of the competitive peak area ratio of 1,2,3-phenylpyrogallol and 1,2,4-phenylpyrogallol after consistency correction in Example 1 of the present invention. Figure 5 This is the Raman spectrum of the mixed sample of 1,2,3-phenylpyrogallol and 1,3,5-phenylpyrogallol in Example 1 of the present invention; Figure 6 This is a standard curve of the competitive peak area ratio of 1,2,3-phenylpyrogallol and 1,3,5-phenylpyrogallol in Example 1 of the present invention. Figure 7 This is the Raman spectrum of the mixed sample of 1,2,4-phenylpyrogallol and 1,3,5-phenylpyrogallol in Example 1 of the present invention; Figure 8 This is a standard curve of the competitive peak area ratio of 1,2,4-phenylpyrogallol and 1,3,5-phenylpyrogallol in Example 1 of the present invention. Figure 9 This is the Raman spectrum of the sample to be tested in Example 1 of this invention; Figure 10 This is a standard curve graph established using the original peak area without consistency correction in Comparative Example 1 of the present invention. Figure 11 This is a standard curve graph established using the absolute peak area of ​​a single peak in Comparative Example 2 of the present invention. Detailed Implementation

[0029] The technical solutions of the present invention will be further described below with reference to the accompanying drawings and embodiments. In the embodiments of this application, the terms "exemplary" or "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design scheme described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or more advantageous than other embodiments or design schemes. Specifically, the use of the terms "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0030] Raman spectroscopy is a method for studying the microstructure of molecules or substances by utilizing the scattering spectra obtained from inelastic collisions between photons and atoms (molecules) in a medium. Its physical basis lies in the Raman scattering effect discovered in 1928 by Indian physicist C.V. Raman: when a sample is irradiated by a laser of a certain frequency, a very small number of photons undergo inelastic collisions with molecules, causing changes in the molecular vibrational and rotational energy levels. The frequency of the scattered light thus shifts relative to the incident light, and the amount of this shift corresponds to the difference in the molecular vibrational / rotational energy levels. Different atomic groups possess characteristic shift "fingerprints," thus allowing the chemical composition and molecular structure of substances to be inferred from the measured Raman scattering spectra. Raman spectroscopy complements infrared spectroscopy, transferring molecular energy spectrum information from the infrared region to the visible light region for observation, making it an important tool for characterizing microstructures in materials science, chemistry, biology, and other fields.

[0031] A specific characteristic peak set refers to a set of signals in a sample fingerprint spectrum obtained using analytical techniques such as chromatography, Raman spectroscopy, or X-ray diffraction. This set consists of multiple characteristic peaks that belong exclusively to a specific substance (or component) and are not present in other coexisting substances. The core concept is "specific," meaning that each characteristic peak or combination thereof in the set serves as a unique identifier distinguishing the substance from others. In practical applications, this set is typically constructed by comparing the fingerprint spectra of the target substance with those of various related substances, selecting peaks unique to the target substance. This set aims to establish a reliable "molecular identity card" or "spectral fingerprint" for complex analytes, used in fields such as traditional Chinese medicine quality control and chemical substance identification to accurately distinguish genuine products from counterfeit products or to determine the presence of specific components.

[0032] Peak area refers to the area enclosed by the outline of a characteristic peak and the baseline in a Raman spectrum, i.e., the area of ​​that peak at the Raman shift (horizontal axis, unit: cm). -1The peak area is the integral value of the intensity (vertical axis, count or cps) within the specified interval. Physically, the peak area is proportional to the product of the number of molecules excited in that vibrational mode and the Raman scattering cross section. Therefore, in quantitative Raman analysis, the peak area is more resistant to instrument resolution fluctuations, slit width variations, and slight band broadening than the peak height, and is often used as the dependent variable in the internal standard method or standard addition method to establish a linear relationship with concentration. In practice, baseline correction (subtracting fluorescence background) should be performed on the spectrum first, and then the integral should be performed between the start and end shifts of the characteristic peaks. In multi-component systems, characteristic peaks that do not interfere with each other in their respective fingerprint regions are selected, and their area ratios are substituted into the calibration curve to determine the concentration. It is important to distinguish that the Raman peak area itself is a relative quantity. The absolute values ​​cannot be compared horizontally under different instruments, laser powers, and integration times. It only has quantitative significance under the same measurement conditions or after normalization with an internal standard.

[0033] The Gaussian-Lorentz mixture function, often referred to as the pseudo-Voigt function in spectroscopy, is a mathematical approximation proposed to approximate the exact Voigt line shape. The exact Voigt function is the convolution of a Gaussian function and a Lorentz function, and its calculation is quite complex. Therefore, the pseudo-Voigt function is often used as an alternative in practical spectral fitting. This function is defined as a linear combination of the Gaussian and Lorentz functions, and the contribution ratio of the two line shapes is controlled by a mixture parameter η (ranging from 0 to 1). When η=1, the function exhibits a pure Lorentz line shape; when η=0, it is a pure Gaussian line shape; and when η takes intermediate values, the function can describe various peak shapes in between. This mixture function is widely used in analytical techniques such as Raman spectroscopy, X-ray photoelectron spectroscopy (XPS), and powder diffraction, especially for actual spectral peaks that cannot be accurately described by pure Gaussian or pure Lorentz functions.

[0034] Raman shift, also known as Raman frequency shift, refers to the change in the frequency of scattered light relative to the frequency of incident light. It originates from inelastic scattering between light and molecules—when monochromatic light strikes a sample, most photons only change direction without changing energy (Rayleigh scattering), but a small percentage of photons exchange energy with molecules, causing a change in their frequency; this phenomenon is Raman scattering. The Raman shift is usually expressed in wavenumbers (cm²). -1 In a Raman spectrum, the horizontal axis represents the Raman shift. Its most crucial characteristic is that the magnitude of the Raman shift is independent of the frequency of the incident light, depending solely on the vibrational or rotational energy level structure of the scattering molecule. This property gives different chemical bonds or groups their own characteristic Raman shifts, much like the "fingerprints" of molecules, thus becoming the fundamental basis for qualitative analysis of material structure using Raman spectroscopy.

[0035] In Raman spectroscopy peak shape fitting analysis, H, W and η are four core parameters describing the profile of a single characteristic peak. Among them, H is the net peak height of the characteristic peak, which refers to the maximum intensity value at the peak apex after local baseline correction. Its value is directly related to the concentration of this vibrational mode and the Raman scattering cross section in the sample, and is the core molecule term for subsequent calculation of signal-to-noise ratio (SNR). The center position of the characteristic peak, i.e., the Raman shift (unit: cm) corresponding to the peak apex. -1 This value is determined by the vibrational energy level difference of specific chemical bonds or functional groups in the molecule, and is independent of the incident laser frequency. It is the most fundamental basis for qualitative identification of substances by Raman spectroscopy. The half-peak half-width (HWHM) of the characteristic peak represents the wavenumber distance from the center of the characteristic peak to one of the intersection points at half the peak height, measured in cm. -1 The corresponding full width at half maximum (FWHM) is: η is the Gaussian-Lorentz mixing factor, a dimensionless parameter ranging from 0 to 1, used to adjust the weight of the peak shape between pure Gaussian (η=0) and pure Lorentz (η=1). When η approaches 0, the peak shape exhibits a symmetrical bell-shaped Gaussian distribution, usually corresponding to a non-uniform broadening mechanism dominated by instrument broadening or Doppler broadening; when η approaches 1, the peak shape exhibits a Lorentz distribution with a "long tail," usually corresponding to a uniform broadening mechanism dominated by molecular natural lifetime broadening; when η takes an intermediate value, the peak shape combines both broadening characteristics.

[0036] In Raman spectroscopy (or chromatography, XRD, etc.) analysis, spectral overlap refers to the degree to which two adjacent characteristic peaks overlap at their bases due to their wavenumber positions being too close, making them difficult to completely separate or quantify independently. This overlap is typically quantitatively evaluated using separation degree. The overlap level can be determined as follows: when Rs ≥ 1.5, the two peaks are baseline separated, indicating low overlap, and peak height or area can be measured independently; when 1.0 ≤ Rs < 1.5, it is a moderate overlap, with distinguishable peak tips but interfering with each other's integrated areas, requiring separation through fitting algorithms; when Rs < 1.0, it is a severe overlap, with the two peaks visually fused into an asymmetric wide packet, necessitating mathematical separation using multi-peak fitting techniques such as Gaussian-Lorentz mixture functions, otherwise the true net peak height or area of ​​each characteristic peak cannot be obtained for subsequent discrimination and quantitative analysis. In practical applications, the degree of overlap directly affects the reliability of characteristic peak signal-to-noise ratio calculations and is one of the key parameters that must be prioritized for evaluation in high-precision spectral analysis and multi-component identification.

[0037] Among the existing methods for detecting phenolic isomers such as pyrogallol, Raman spectroscopy has advantages such as no need for complex sample pretreatment, fast detection speed, non-destructive analysis, and the ability to achieve in-situ or quasi-in-situ detection. However, it still has problems such as poor linearity of calibration curves, sensitivity to peak area extraction errors in high content ranges, and deviations of peak area ratios or proportions from the true values, resulting in significant dispersion of calibration curves at these points.

[0038] Therefore, this application provides a quantitative method for pyrogallol isomers based on multi-characteristic peak consistency correction, solving the technical problems of insufficient reliability in qualitative differentiation of pyrogallol isomer mixtures, poor linearity of single-peak absolute peak area or ordinary peak area ratio, and discrete calibration curves and insufficient quantitative accuracy in existing Raman quantitative detection methods. Specifically: Example 1: This example describes a quantitative method for pyrogallol isomers based on multi-characteristic peak consistency correction, including S1~S4: S1. For the mixture containing two pyrogallol isomers to be tested, the two pyrogallol isomers are respectively designated as components. and components The Raman spectrum of the mixture was collected and preprocessed to obtain the preprocessed spectrum.

[0039] The Raman spectrum acquisition of the mixture includes: placing an appropriate amount of the mixture on a clean sample carrier surface, ensuring the surface is flat and uniform, and then placing it on the sample stage of a Raman spectrometer for acquisition. For example, a 638 nm laser with a laser power of 30 mW, an integration time of 10 s, and a scanning range of 200–1500 cm⁻¹ is used. -1 Each sample was tested three times. The preprocessing described above includes smoothing, baseline correction, and wavenumber calibration. For example, smoothing can employ the Savitzky-Golay method (window of 5-11 points, polynomial order 2-3), baseline correction can use adaptive iterative weighted penalized least squares or polynomial subtraction, and wavenumber calibration can use a standard material (such as silicon 520.7 cm⁻¹). -1 The peaks (or mercury-neon lamp spectral lines) are corrected to obtain the preprocessed spectrum.

[0040] It is understood that the mixture to which this method is applicable is preferably a mixture of two pyrogallol isomers with a mass fraction of 10-90%.

[0041] S2. Based on the standard spectra of the three pyrogallol isomers, read the net peak height of each specific characteristic peak in the preprocessed spectrum and calculate the signal-to-noise ratio. According to the signal-to-noise ratio A consensus vote was performed on the individual characteristic peaks in the preprocessed spectrum to determine the composition. and components Qualitative analysis is completed by comparing the results with two of the three isomers of phloroglucinol. Specifically, this includes the following steps S201-S202: S201. Based on the standard spectra of the three pyrogallol isomers, determine the specific characteristic peak sets for each of the three pyrogallol isomers. Each specific characteristic peak set includes one main characteristic peak and at least two auxiliary characteristic peaks. For example, 1,2,3-pyrogallol has a main characteristic peak at 711 cm⁻¹. -1 Main characteristic peak, 270 cm⁻¹ -1 1065 cm -1 As a secondary characteristic peak; 1,2,4-pyrogallol at 787 cm⁻¹ -1 Main characteristic peak, 964 cm -1 526 cm -1 As a secondary characteristic peak; 1,3,5-phenylpyrogallol at 996 cm⁻¹ -1 The main characteristic peak is at 243 cm⁻¹. -1 602 cm -1 These are auxiliary characteristic peaks. The corresponding isomer at each characteristic peak position has a significant Raman response, while the other two isomers do not have a significant response. Therefore, they are called exclusive characteristic peaks. The center position of each exclusive characteristic peak is pre-calibrated and fixed by pure standard samples. It can be understood that the standard spectra of the three pyrogallol isomers can be obtained by consulting spectral databases or existing literature, or by collecting samples made by oneself (for example, steps (1) to (2) in S4 below). Using the center wavenumber of each characteristic peak in the characteristic peak set as the search position, the net peak height after local baseline subtraction is read at the corresponding wavenumber of the preprocessed spectrum. Calculate the signal-to-noise ratio of each specific characteristic peak in the preprocessed spectrum. : ; In the formula, For the first Signal-to-noise ratio of each characteristic peak The net peak height of this characteristic peak after baseline correction (peak intensity minus local baseline intensity), in counts. The peakless region in the spectrum (e.g., 1400~1450 cm⁻¹) -1 The standard deviation of the noise intensity in a flat baseline segment is the root mean square of the noise. S202, Based on signal-to-noise ratio A consensus vote was performed on the individual characteristic peaks in the preprocessed spectrum to determine the composition. and components The qualitative analysis was completed by comparing it with two of the three isomers of phloroglucinol.

[0042] Signal-to-noise ratio The specific characteristic peaks are recorded as valid votes. In this embodiment, the signal-to-noise ratio (SNR) threshold is set to 3. A SNR of 3 or higher indicates that the corresponding peak height reaches at least 3 times the noise standard deviation, which is used to reduce the probability of random noise being misjudged as a specific characteristic peak. Phloroglucinol isomers with ≥2 valid votes are the components in the mixture. For example, when 711 cm⁻¹... -1 With 270 cm -1 787 cm -1 With 964 cm -1 If all peaks reach the threshold, but the 1,3,5-phenylpyrogallol-specific peak does not, the sample is determined to be a mixture of 1,2,3-phenylpyrogallol and 1,2,4-phenylpyrogallol, with 1,2,3-phenylpyrogallol as the component. 1,2,4-Phoroglossol as a component .

[0043] S3. Based on the separability index, the components... and components Orthogonal optimization is performed on the exclusive characteristic peaks to determine the main characteristic peak pairs. Using the main characteristic peak pairs as the conversion scale, the components are then... and components The peak areas of each characteristic peak are calculated separately to obtain the calculated peak areas; multi-characteristic peak consistency correction is then performed on the calculated peak areas to obtain the component. effective peak area and components effective peak area According to the components effective peak area and components effective peak area The components were calculated. competitive peak area ratio Specifically, this includes the following S301~S304: S301. First, fit each characteristic peak with a Gaussian-Lorentz mixture function to obtain the peak shape fitting curve. ,include: ; In the formula, The curve is a peak-shaped fitting curve. Raman displacement, in cm -1 , The peak height is the characteristic peak, expressed in Counts. The location of the characteristic peak center is shown in cm. -1 , The half-peak-half-width of the characteristic peak represents the wavenumber distance from the center of the characteristic peak to one of the intersection points at half the peak height, in cm. -1 , This is the Gaussian-Lorentz mixing scaling factor, with a value ranging from 0 to 1; Then, based on the peak shape fitting curve , to obtain components Candidate characteristic peaks With components Candidate characteristic peaks spectral overlap between ,include: ; In the formula, Components Candidate characteristic peaks With components Candidate characteristic peaks The spectral overlap between the candidate characteristic peaks and Corresponding components Components All characteristic peaks in their respective characteristic peak sets , Components Candidate characteristic peaks Components Candidate characteristic peaks The peak-shaped fitting curve obtained through fitting, For Raman shift, the integration interval is the fitting window for the two candidate characteristic peaks; Then based on the components Candidate characteristic peaks With components Candidate characteristic peaks spectral overlap between The separability index is obtained. ,include: ; In the formula, Components Candidate characteristic peaks With components Candidate characteristic peaks The separability index between them; , It is a dimensionless quantity; It is fixed at 1 and does not participate in data fitting.

[0044] The largest separability index The corresponding pair of components and components The specific characteristic peaks are used as the main characteristic peak pairs; for the 1,2,3-phenylpyrogallol / 1,2,4-phenylpyrogallol system, 711 cm⁻¹ is preferred. -1 With 787 cm-1 The preferred 711 cm⁻¹ pyrogallol / 1,3,5-phenylpyrogallol system is preferred. -1 With 996 cm -1 The preferred 787 cm⁻¹ chromatin is for the 1,2,4-phenylpyrogallol / 1,3,5-phenylpyrogallol system. -1 With 996 cm -1 The fitting window for each candidate characteristic peak is centered at the peak's location in the pure standard sample. Centered on the low wavenumber side, it extends towards the high wavenumber side respectively. ,in, The half-peak width of the characteristic peak in the pure standard sample is taken as half-peak width; the integral interval of the spectral overlap of the two candidate characteristic peaks is taken as the union of the two fitting windows.

[0045] S302. Using the main characteristic peak pair as the conversion scale, for components and components The peak areas of each specific characteristic peak are calculated separately to obtain the converted peak areas, including: Conversion Components No. The calculated peak area of ​​each unique characteristic peak ,include: ; In the formula, Components No. The peak area of ​​each unique characteristic peak, in Counts·cm² -1 , Components No. The conversion factor for each characteristic peak; The components No. Peak area of ​​each unique characteristic peak ,include: ; In the formula, Components No. The peak height of each unique characteristic peak, expressed in counts. Components No. Half-peak width of each unique characteristic peak, in cm. -1 , Components No. Gaussian-Lorentz mixing ratio factor for each unique characteristic peak; The components No. The reduction factor of each characteristic peak ,include: ; In the formula, , Components The main characteristic peak in pure standard samples, the first The peak area of ​​each characteristic peak, in Counts·cm -1 .

[0046] S303. Perform multi-characteristic peak consistency correction on the calculated peak areas to obtain the components. effective peak area and components effective peak area ,include: ; In the formula, These are the normalized weights, and the components The sum of all weights is 1. The peak area is calculated to the scale of the main characteristic peak, in Counts·cm. -1 ; The normalized weights ,include: ; In the formula, Components No. A reliability measure of each characteristic peak; The components No. Reliability measurement of each characteristic peak ,include: ; In the formula, Components No. Signal-to-noise ratio of each characteristic peak Components No. Characteristic peaks and components The maximum value of the spectral overlap of all characteristic peaks, calculated using the same method as... same; ; In the formula, Components No. A measure of the consistency of each unique characteristic peak; Components No. The calculated peak area of ​​each individual characteristic peak after being scaled down to the main characteristic peak is expressed in Counts·cm. -1 ; Components Total converted peak area The median, in Counts·cm -1 ; The consistency tolerance coefficient is set to 0.1 in this embodiment.

[0047] when When the calculated peak area of ​​the k-th specific characteristic peak is equal to the median If they are the same, then the consistency measure of this specific characteristic peak is 1; if it is relative to the median... If the relative deviation is 10%, the consistency metric is 0.5; as the relative deviation continues to increase, the consistency metric decreases further. Therefore, normal characteristic peaks close to the median consensus value retain a higher weight, while characteristic peaks that significantly deviate from the median consensus value have a lower weight.

[0048] In this embodiment, the effective peak area, competitive peak area ratio, standard curve, and validation sample results mentioned in Tables 1 to 7 are all within the consistency tolerance coefficient. It was calculated under the given conditions.

[0049] When the characteristic peak at a certain concentration point exhibits abnormal peak area due to peak overlap, baseline fluctuations, or noise, its calculated peak area will deviate from the median consensus value. This reduces the consistency measure and normalization weight of the characteristic peak. The area of ​​a calculated peak relative to the median... The greater the relative deviation, the lower the consistency measure and weight. When the relative deviation is 10%, the consistency measure is 0.5, and the weight decreases further as the deviation continues to increase. The corrected effective peak area is mainly obtained by weighting specific characteristic peaks that are close to the median consensus value, thereby reducing the impact of abnormal peak areas on the quantitative results.

[0050] Components effective peak area With components effective peak area The method for obtaining them is the same.

[0051] S304, according to components effective peak area and components effective peak area The components were calculated. Competitive peak area ratio ; ; In the formula, Components The effective peak area, Components Effective peak area, in units of Counts·cm -1 ; Components The proportion of the competitive peak area is a dimensionless quantity. Taking the 1,2,3-phenylpyrogallol / 1,2,4-phenylpyrogallol system as an example: For 1,2,3-pyrogallol (e.g., in 711 cm⁻¹) -1 The corrected effective peak area (of the main characteristic peak) For 1,2,4-pyrogallol (e.g., at 787 cm⁻¹) -1 The effective peak area after correction (main characteristic peak).

[0052] S4, according to components competitive peak area ratio With components quality score Linear self-calibration quantitative model between: ; Calculate components mass fraction: ; Based on: ; Components The quality score.

[0053] In the formula, Components The mass fraction, Components The proportion of competitive peak area. and Components With components The fitting slope and fitting intercept corresponding to the mixed system; When the two pyrogallol isomers in the mixture are 1,2,3-pyrogallol and 1,2,4-pyrogallol, It is 0.3491. It is 0.5229; When the two pyrogallol isomers in the mixture are 1,2,3-pyrogallol and 1,3,5-pyrogallol, It is 0.8993. It is -0.0105; When the two pyrogallol isomers in the mixture are 1,2,4-pyrogallol and 1,3,5-pyrogallol, It is 0.6814. It is 0.1351; Calculate components mass fraction and according to Components The mass fraction was determined to complete the quantitative assessment.

[0054] The method for constructing the aforementioned linear self-calibration quantitative model is as follows: (1) Prepare mixed samples of pyrogallol isomers with different mass ratios. Weigh out the standards of 1,2,3-phenylpyrogallol, 1,2,4-phenylpyrogallol, and 1,3,5-phenylpyrogallol, respectively. The molecular structures of the three phenylpyrogallols are as follows: Figure 1 As shown, mixed samples of pyrogallol isomers with different mass ratios were prepared in pairs, including three groups: 1,2,3-pyrogallol / 1,2,4-pyrogallol, 1,2,3-pyrogallol / 1,3,5-pyrogallol, and 1,2,4-pyrogallol / 1,3,5-pyrogallol. Taking the 1,2,3-pyrogallol / 1,2,4-pyrogallol system as an example, mixed samples with a mass fraction of 1,2,3-pyrogallol of 10% to 90% (interval of 10%) were prepared as examples, and the mass fraction of 1,2,4-pyrogallol was 90% to 10% respectively; the other two groups were prepared in the same way.

[0055] (2) Collect Raman spectra of mixed samples A suitable amount of the mixed sample is placed on a clean sample carrier surface, ensuring the surface is smooth and uniform. The sample is then placed on the sample stage of a Raman spectrometer for acquisition. For example, a 638 nm laser is used, with a laser power of 30 mW, an integration time of 10 s, and a scanning range of 200–1500 cm⁻¹. -1 Each sample was tested three times. The acquired spectra were preprocessed, including smoothing, baseline correction, and wavenumber calibration. The standard Raman spectra of the three phloroglucins are shown below. Figure 2 As shown. The smoothing, baseline correction, and wavenumber calibration all employ conventional Raman spectroscopy preprocessing methods. This invention does not impose specific limitations on the particular algorithms and parameters, nor is it targeted at any specific analyte. For example, smoothing can employ the Savitzky-Golay method (window 5-11 points, polynomial order 2-3), baseline correction can employ adaptive iterative weighted penalized least squares or polynomial subtraction, and wavenumber calibration can use a standard substance (such as silicon 520.7 cm⁻¹). -1 (Peak or mercury-neon lamp spectral line) correction. Because this invention subsequently employs multi-characteristic peak consistency correction and competitive peak area ratio of self-reference, the final quantitative result is not sensitive to the specific values ​​of the above preprocessing parameters.

[0056] (3) Determine the specific characteristic peak sets of the three pyrogallol isomers Standard Raman spectra were acquired using three pyrogallol standards, and their respective characteristic peak sets were extracted: 1,2,3-pyrogallol at 711 cm⁻¹ -1 Main characteristic peak, 270 cm⁻¹ -1 1065 cm -1 As a secondary characteristic peak; 1,2,4-pyrogallol at 787 cm⁻¹ -1 Main characteristic peak, 964 cm -1 526 cm -1 As a secondary characteristic peak; 1,3,5-phenylpyrogallol at 996 cm⁻¹ -1 The main characteristic peak is at 243 cm⁻¹. -1 602 cm -1 It is a secondary characteristic peak.

[0057] (4) Obtain the fitting model The components in each mixed sample were calculated according to the method in Example 1. effective peak area and components effective peak area According to the components effective peak area and components effective peak area The components were calculated. competitive peak area ratio Then, based on the components of each mixed sample mass fraction x-axis, components competitive peak area ratio A linear fit was performed on the ordinate, and the fitting model was determined as follows: ; In the formula, The target component in the mixture of two pyrogallol isomers to be detected The mass fraction, This represents the area ratio of competitive peaks. Components The corrected effective peak area, Components The corrected effective peak area, and Components With components The fitting slope and fitting intercept of the mixed system are shown in Table 4. Among them, a and b are determined according to the coexistence system of pyrogallol isomers.

[0058] Specifically: The Raman spectrum of the 1,2,3-phenylpyrogallol / 1,2,4-phenylpyrogallol mixed sample is as follows: Figure 3 As shown in Table 1, the effective peak area and competing peak area ratios of the main characteristic peaks of the two components at each mass ratio were obtained after processing S2~S3. The established standard curve is shown in Table 1. Figure 4 As shown, the fitted model is R = 0.3491w + 0.5229, and the correlation coefficient R0 is... 2 =0.9943, indicating good linearity. The correlation coefficient R0 is... 2 The coefficient of determination for linear fitting of the standard curve is given by... calculate, This is the sum of squares of the differences between the measured competitive peak area percentage and the fitted value for each standard sample. R is the sum of the squares of the differences between each measured value and its mean. 2 The closer to 1, the better the linearity.

[0059] Table 1: Effective peak area and percentage of competing peak area of ​​mixed samples of 1,2,3-phenylpyrogallol and 1,2,4-phenylpyrogallol after consistency correction

[0060] Correspondingly, the Raman spectra of the 1,2,3-phenylpyrogallol / 1,3,5-phenylpyrogallol mixed sample are as follows: Figure 5 As shown in Table 2, the effective peak area and competing peak area ratios of the main characteristic peaks of the two components at each mass ratio were obtained after processing S2~S3. The standard curve is shown in Table 2. Figure 6 As shown, the fitted model is R = 0.8993w − 0.0105, and the correlation coefficient R0 is... 2 =0.9924.

[0061] Table 2: Effective peak area and percentage of competing peak area of ​​mixed samples of 1,2,3-phenylpyrogallol and 1,3,5-phenylpyrogallol after consistency correction

[0062] Accordingly, the Raman spectra of the 1,2,4-phenylpyrogallol / 1,3,5-phenylpyrogallol mixed sample are as follows: Figure 7 As shown in Table 3, the effective peak area and competing peak area ratios of the main characteristic peaks of the two components at each mass ratio were obtained after processing S2~S3. The standard curve is shown in Table 3. Figure 8 As shown, the fitted model is R = 0.6814w + 0.1351, and the correlation coefficient R0 is... 2 =0.9940.

[0063] Table 3: Effective peak area and percentage of competing peak area of ​​mixed 1,2,4-phenylpyrogallol and 1,3,5-phenylpyrogallol after consistency correction

[0064] All three systems showed good linearity after correction by the present invention. Based on the component correspondence in each mixed sample, a and b were obtained for various pairwise component mixing cases, and the results are shown in Table 4.

[0065] Table 4: Fitting models and correlation coefficients of the three pyrogallol isomer systems of this invention

[0066] (5) Then verify the fitting model in Table 4: For example, a mixed sample of 1,2,3-phenylpyrogallol and 1,2,4-phenylpyrogallol was taken as a verification sample and subjected to discrimination and quantification according to S1~S4. The Raman spectrum of the verification sample is as follows: Figure 9 As shown, the spectrum simultaneously exhibits the characteristic peak of 1,2,3-phenylpyrogallol (711 cm⁻¹). -1 and 269 cm -1 Corresponding to its 270 cm -1 The auxiliary characteristic peak) and the specific characteristic peak of 1,2,4-phenylpyrogallol (787 cm⁻¹) -1 and 965 cm -1 Each isomer had at least two characteristic peaks with a signal-to-noise ratio exceeding the threshold, but the characteristic peak for 1,3,5-phenylpyrogallol did not appear. Therefore, based on the consensus of multiple characteristic peaks, the verification sample was determined to be a mixture of 1,2,3-phenylpyrogallol and 1,2,4-phenylpyrogallol, with 1,2,3-phenylpyrogallol as the component. 1,2,4-Phenylopyrrolidone is a component Regarding this component 711 cm -1 With components 787 cm -1 The peak area of ​​the main characteristic peak is extracted and then subjected to consistency correction (the main and auxiliary characteristic peaks are consistent, and the effective peak area after correction is consistent with the original peak area), resulting in... =66874.11、 =27279.61.

[0067] Calculate the area ratio of competing peaks =66874.11(66874.11+27279.61)=0.7103, substituting into the quantitative model of the 1,2,3-phenylpyrogallol / 1,2,4-phenylpyrogallol system R=0.3491w+0.5229, we get w=(0.7103−0.5229) / 0.3491=0.537, that is, the mass fraction of 1,2,3-phenylpyrogallol in the verification sample is 53.7%, which is close to the actual mass fraction of 55%, with an absolute error of only about 1.3 percentage points, indicating good quantitative accuracy.

[0068] The test results of the verification samples are shown in Table 5.

[0069] Table 5: Effective peak area, competitive peak area ratio, and predicted quantitative results of the verification samples detected by this invention.

[0070] That is, the proportion of competitive peak area is confirmed. With mass fraction Linear self-calibration quantitative model between: ; In the formula, Components The mass fraction, Components The proportion of competitive peak area. and Components With components The fitting slope and fitting intercept of the mixed system, and the components mass fraction by get; When the two pyrogallol isomers in the mixture are 1,2,3-pyrogallol and 1,2,4-pyrogallol, It is 0.3491. It is 0.5229; When the two pyrogallol isomers in the mixture are 1,2,3-pyrogallol and 1,3,5-pyrogallol, It is 0.8993. It is -0.0105; When the two pyrogallol isomers in the mixture are 1,2,4-pyrogallol and 1,3,5-pyrogallol, It is 0.6814. It is 0.1351.

[0071] Experimental example: (1) Detect the impact of consistency correction of multiple characteristic peaks on quantitative results.

[0072] Comparative Example 1: Calculating the proportion of competing peak areas based on the original (direct integration) peak area of ​​a single main characteristic peak. , Components The original peak area of ​​the main characteristic peak. Components The original peak area of ​​the main characteristic peak is used to establish a fitting model; The difference between the linear self-calibrating quantitative model construction method in Comparative Example 1 and Example 1 is that: no multi-characteristic peak consistency correction is performed; Taking the mixed sample of 1,2,3-phenylpyrogallol / 1,2,4-phenylpyrogallol (actual mass fraction of 1,2,3-phenylpyrogallol 55%) in S4 of Example 1 as an example, the uncorrected original 711 cm⁻¹ -1 787 cm -1 The peak area and the proportion of the original competing peak area R0 are shown in Table 6, and the constructed standard curve is shown in Table 6. Figure 10 As shown, the fitted model is R0 = 0.3445w + 0.5140, and the correlation coefficient R0 is... 2 =0.8833.

[0073] Table 6: Comparison of R0 in Comparative Example 1 and R in this invention

[0074] Comparing Table 6 and Table 1, it can be seen that at the five concentration points of 10%, 20%, 30%, 60%, and 70%, neither Comparative Example 1 nor the main characteristic peak of the present invention showed any abnormalities, and the original peak area of ​​Comparative Example 1 was consistent with the effective peak area after correction by the present invention. However, at the four concentration points of 40%, 50%, 80%, and 90%, the original main characteristic peak in Comparative Example 1 showed abnormal peak areas due to peak overlap or baseline fluctuations, causing the proportion of the original competing peak area to deviate significantly from the calibration line (e.g., R0=0.5698 at 40% point and R0=0.7324 at 50% point). After the multi-characteristic peak consistency correction of the present invention, the above-mentioned abnormal points were corrected to near the straight line (R=0.6629 at 40% point and R=0.7100 at 50% point), thereby reducing the correlation coefficient from R0 to R0. 2 =0.8833 increased to R 2 =0.9943.

[0075] The comparison shows that, without multi-characteristic peak consistency correction, the abnormal peak areas at individual concentration points in Comparative Example 1 cause significant dispersion in the calibration curve (R0). 2 =0.8833); while the present invention identifies and corrects these outliers through multi-feature peak consistency correction, thereby increasing the correlation coefficient to 0.9943.

[0076] Further verification was performed by substituting the sample to be tested into the standard curve of Comparative Example 1: The sample to be tested had R0=0.7103, and substituting it into R0=0.3445w+0.5140 yielded w=57.0%, which differed from the actual value of 55% by about 2.0 percentage points, lower than that of the present invention; although the main peak of the sample to be tested itself was not abnormal and the prediction was acceptable, the overall reliability of the calibration curve was lower than that of the fitting model established in the present invention because it was obtained by fitting uncorrected data.

[0077] (2) The combined effect of the proportion of competitive peak area and the consistency correction of multiple characteristic peaks on the quantitative results.

[0078] Comparative Example 2: A fitting model is established using the absolute peak area of ​​a single peak, that is, without constructing any ratio (i.e., the competitive peak area ratio R of this invention), the original absolute peak area of ​​the single main characteristic peak of the target component is directly used as a quantitative variable to establish the fitting model.

[0079] The difference between the linear self-calibration quantitative model construction method in Comparative Example 2 and Example 1 is that: it does not construct competitive peak area ratios, nor does it perform multi-characteristic peak consistency correction.

[0080] Taking the mixed sample of 1,2,3-phenylpyrogallol / 1,2,4-phenylpyrogallol (actual mass fraction of 1,2,3-phenylpyrogallol 55%) in S4 of Example 1 as an example, with 711 cm of 1,2,3-phenylpyrogallol... -1 A standard curve was constructed using the original absolute peak area A(711) of the main characteristic peak (data shown in the second column of Table 6) as the ordinate and the true mass fraction of 1,2,3-phenylpyrogallol as the abscissa. The results are as follows: Figure 11 As shown: the fitted model is A(711)=48414w+33510, and the correlation coefficient is only R. 2 =0.6041, the linearity is significantly lower than that of the present invention.

[0081] Substituting sample A(711)=66874.11 into A(711)=48414w+33510, we get w=68.9%, which differs from the actual value of 55% by about 13.9 percentage points. The error is relatively large because the absolute peak area of ​​a single characteristic peak is affected by the overall acquisition conditions such as component content, laser power, focusing state, and sample amount. It does not only reflect the change in component content, thus resulting in poor linearity and low prediction stability.

[0082] In summary, the quantitative results of the present invention and Comparative Examples 1 and 2 for the same mixed sample are compared in Table 7.

[0083] Table 7: Comparison of quantitative results of the present invention with those of Comparative Examples 1 and 2 for the same sample (1,2,3-phenylpyrogallol / 1,2,4-phenylpyrogallol system)

[0084] As shown in Tables 6 and 7, under the premise of using the same characteristic peaks and the same collected data: Comparative Example 2 directly uses the absolute peak area of ​​a single peak, and due to the influence of fluctuations in the overall collection conditions, the linearity is the worst (R0). 2 =0.6041), with the largest prediction error (13.9 percentage points); Comparative Example 1 uses competitive proportions but does not perform consistency correction, which eliminates overall intensity fluctuations and improves linearity (R). 2The correlation coefficient was 0.8833, but the abnormal peak areas at individual concentration points still caused the curve to become discrete. In comparison, this invention further introduces multi-characteristic peak consistency correction on top of competitive normalization to identify and correct abnormal peak areas, increasing the correlation coefficient to 0.9943 and reducing the absolute quantitative error of the sample to approximately 1.3 percentage points. This verifies the beneficial effect of combining multi-characteristic peak consistency correction and competitive normalization in the efficient and stable quantitative detection of pyrogallol isomers.

Claims

1. A quantitative method for pyrogallol isomers based on multi-characteristic peak consistency correction, characterized in that, Includes the following steps: For the mixture containing two pyrogallol isomers to be tested, the two pyrogallol isomers are respectively designated as components. and components The Raman spectrum of the mixture was collected and preprocessed to obtain the preprocessed spectrum; Based on the standard spectra of the three pyrogallol isomers, the net peak heights of each specific characteristic peak in the preprocessed spectrum were read and the signal-to-noise ratio was calculated. According to the signal-to-noise ratio A consensus vote was performed on the individual characteristic peaks in the preprocessed spectrum to determine the composition. and components The qualitative analysis was completed by comparing it with two of the three isomers of phloroglucinol. Based on the separability index, the components and components Orthogonal optimization is performed on the exclusive characteristic peaks to determine the main characteristic peak pairs; using the main characteristic peak pairs as the conversion scale, the components are... and components The peak areas of each characteristic peak are calculated separately to obtain the calculated peak areas; multi-characteristic peak consistency correction is then performed on the calculated peak areas to obtain the component. effective peak area and components effective peak area According to the components effective peak area and components effective peak area The components were calculated. competitive peak area ratio ; According to components A linear self-calibrating quantitative model relating competitive peak area percentage to mass fraction was used to calculate the composition. mass fraction and according to Components The mass fraction; The components are classified according to the separability index. and components Orthogonal optimization is performed on the exclusive characteristic peaks to determine the main characteristic peak pairs, including: First, Gaussian-Lorentz mixture function is applied to fit each characteristic peak to obtain the peak shape fitting curve. ,include: ； In the formula, The curve is a peak-shaped fitting curve. Raman displacement, in cm -1 , The peak height is the characteristic peak, expressed in Counts. The location of the characteristic peak center is shown in cm. -1 , The half-peak-half-width of the characteristic peak represents the wavenumber distance from the center of the characteristic peak to one of the intersection points at half the peak height, in cm. -1 , This is the Gaussian-Lorentz mixing scaling factor, with a value ranging from 0 to 1; Then, based on the peak shape fitting curve , to obtain components Candidate characteristic peaks With components Candidate characteristic peaks spectral overlap between ,include: ； In the formula, Components Candidate characteristic peaks With components Candidate characteristic peaks The spectral overlap between the candidate characteristic peaks and Corresponding components Components All characteristic peaks in their respective characteristic peak sets , Components Candidate characteristic peaks Components Candidate characteristic peaks The fitted curves obtained by fitting the peak shape use the center position of the characteristic peak calibrated with pure standard samples for each fitting window. Expand 2 to each side from the center. ,in, The half-peak width of the characteristic peak in the pure standard sample is taken as the half-peak width; the integration interval is the union of the fitting windows of the two candidate characteristic peaks. Then based on the components Candidate characteristic peaks With components Candidate characteristic peaks spectral overlap between The separability index is obtained. ,include: ； In the formula, Components Candidate characteristic peaks With components Candidate characteristic peaks The separability index between them; , Candidate feature peaks , The signal-to-noise ratios are all dimensionless quantities; The overlap penalty coefficient is fixed at 1. The largest separability index The corresponding pair of components and components The exclusive characteristic peaks are used as the main characteristic peak pairs.

2. The method according to claim 1, characterized in that, Components mass fraction The competitive peak area ratio can be used to determine the proportion of peak areas. With mass fraction The linear self-calibration quantitative model between them is calculated according to the following formula: ； In the formula, Components The proportion of competitive peak area. and Components With components The fitting slope and fitting intercept of the mixed system; components mass fraction according to calculate.

3. The method according to claim 1, characterized in that, According to the components Effective peak area and composition Calculation of effective peak area of ​​components The competitive peak area percentage includes: ； In the formula, Components The effective peak area, Components Effective peak area, in units of Counts·cm -1 , Components The area ratio of the competitive peak is a dimensionless quantity.

4. The method according to claim 1, characterized in that, The net peak height of each specific characteristic peak in the preprocessed spectrum is read based on the standard spectra of the three pyrogallol isomers, and the signal-to-noise ratio is calculated. The process includes: determining the specific characteristic peak sets for each of the three pyrogallol isomers based on their standard spectra, each set comprising one main characteristic peak and at least two secondary characteristic peaks; using the center wavenumber of each specific characteristic peak in the set as the search position; reading the net peak height after local baseline subtraction at the corresponding wavenumber of the preprocessed spectrum; and calculating the signal-to-noise ratio of each specific characteristic peak in the preprocessed spectrum based on the net peak height. : ； In the formula, For the first Signal-to-noise ratio of each characteristic peak This represents the net peak height of the characteristic peak after baseline correction, expressed in counts. This represents the standard deviation of the intensity within the peakless region of the spectrum, expressed in Counts.

5. The method according to claim 1, characterized in that, The multi-feature peak consensus voting includes: [the following is a process]: signal-to-noise ratio... The exclusive characteristic peak is recorded as a valid vote; when there are no less than 2 valid votes for the same pyrogallol isomer, the pyrogallol isomer is determined to exist.

6. The method according to claim 1, characterized in that, The method of using the main characteristic peak pair as the conversion scale for components and components The peak areas of each specific characteristic peak are calculated separately to obtain the converted peak areas, including: Conversion Components No. The calculated peak area of ​​each unique characteristic peak ,include: ； In the formula, Components No. The peak area of ​​each unique characteristic peak, in Counts·cm² -1 , Components No. The conversion factor for each characteristic peak; The components No. Peak area of ​​each unique characteristic peak ,include: ； In the formula, Components No. The peak height of each unique characteristic peak, expressed in counts. Components No. Half-peak width of each unique characteristic peak, in cm. -1 , Components No. Gaussian-Lorentz mixing ratio factor for each unique characteristic peak; The components No. The reduction factor of each characteristic peak ,include: ； In the formula, , Components The main characteristic peak in pure standard samples, the first The peak area of ​​each characteristic peak, in Counts·cm -1 .

7. The method according to claim 6, characterized in that, Multi-characteristic peak consistency correction is performed on the calculated peak areas to obtain the composition. effective peak area and components effective peak area ,include: ； In the formula, These are the normalized weights, and the components The sum of all weights is 1. The peak area is calculated to the scale of the main characteristic peak, in Counts·cm. -1 ; The normalized weights ,include: ； In the formula, Components No. A reliability measure of each characteristic peak; The components No. Reliability measurement of each characteristic peak ,include: ； In the formula, Components No. Signal-to-noise ratio of each characteristic peak Components No. Characteristic peaks and components The maximum value of the spectral overlap of all characteristic peaks, calculated using the same method as... same; ； In the formula, Components No. A measure of the consistency of each unique characteristic peak; Components No. The calculated peak area of ​​each individual characteristic peak after being scaled down to the main characteristic peak is expressed in Counts·cm. -1 ; Components Total converted peak area The median, in Counts·cm -1 ; The consistency tolerance coefficient is set to 0.

1. Components effective peak area With components effective peak area The method for obtaining them is the same.

8. A quantitative analysis system for pyrogallol isomers for implementing the method according to any one of claims 1-7, characterized in that, Includes a Raman spectrometer and data processing equipment; The Raman spectrometer is used to collect the Raman spectra of the test mixture containing two pyrogallol isomers; The data processing device includes a memory and a processor. The memory stores a computer program. When the processor executes the computer program, it performs preprocessing, qualitative analysis based on the consensus voting of multiple characteristic peaks, peak area conversion, consensus correction of multiple characteristic peaks, calculation of the area ratio of competing peaks, and calculation of mass fraction of the Raman spectrum. The data processing device obtains the components according to any one of claims 1-7. mass fraction and according to Components mass fraction .

9. A computer device, characterized in that, It includes an instruction memory and an operation processor for executing the quantitative method for pyrogallol isomers based on multi-characteristic peak consistency correction as described in any of claims 1-7.

Images